Conical Grid Plate Flame Stabilizers for Combustor Primary Zones

Transcription

1 THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS 345 E. 47 St., New York, N.Y The Society shall not be responsible for statements or opinions advanced in papers or in discussion at meetings of the Society or of its Divisions or Sections, or printed in its publications. Discussion is printed only if the paper is published in an ASME Journal. Released for general publication upon presentation. Full credit should be given to ASME,, the Technical Division, and the author(si. Papers are available from ASME for nine months after the meeting. Printed in USA. Copyright 1985 by ASME 85-GT-53 Conical Grid Plate Flame Stabilizers for Combustor Primary Zones A. F. ALI and G. E. ANDREWS, Department of Fuel and Energy University of Leeds Leeds, LS2.9JT, U.K. ABSTRACT Emission results are presented for a jet shear layer flame stabiliser design consisting of a 90 0 conical flame stabiliser with an array of holes and a central annular vaporiser fuel injection system. This design was tested with premixed propane and air and with direct propane injection into the vaporiser at two blockages and approach velocities, The results showed that an array of jet shear layers could be fuelled by a single fuel injector without incurring excessive NO emissions. An increase in the primary zone residence time was found to result in an improved combustion efficiency, with no increase in NO x, provided that the stabiliser blockage was increased to maintain the pressure loss. INTRODUCTION Improvements in fuel and air mixing within future gas turbine combustors are necessary if the problems of emission, future fuels and higher operating temperatures are to be overcome. Fully premixed prevaporised combustion systems, although offering a possible solution to some emission problems have many practical problems (1,2). The authors are involved in an intensive investigation of rapid fuel and air mixing using fuel injection into jet shear layers (1,3-5) and swirling flow shear zones (6-8). This work has shown that it is possible to achieve the advantages of premixed systems without the disadvantages. The present work represents an extension of previous work (5) into multihole jet stabiliser systems. The major change was to investigate single point fuel injection, which is more easily applied to liquid fuels than the previous approach of separate fuel injection into each shear layer (4,5). Hakluytt and Tilston (9,10) have developed a technique of central fuel injection into an array of air jets of a specific design in a conical flow stabiliser. Their design, which has become known as the 'pepperpot combustor has been tested in Spey and Marine Olympus configurations at engine conditions. However, this work involved the replacement of the head section of these combustors whilst retaining the existing wall cooling and radial air jets of the conventional can. In the design of the primary zone air flow, a mean equivalence ratio of 0.6 was assumed. The present work was aimed at assessing whether this design primary zone equivalence ratio was the optimum and to investigate the performance of the conical stabiliser alone in the absence of the downstream conventional combustor aerodynamics. FLAME STABILISER DESIGN AND TESL CONDITIONS Previous work by the authors (1,4,5) used grid plates with arrays of equal sized holes. For these systems both premixed (1) and direct shear layer fuel injection has been used (4,5). In the latter case a simple 4 hole design has been investigated over a range of blockage, 66, 78 and 82% (4) and at the 78 blockage a 16 hole system was also investigated (5). This type of design, although having the advantage of simplicity, does have two problems in the present application. Firstly, the stabiliser surface area is a minimum and thus the holes are close together. Secondly, with a central fuel injector the fuel must he injected radially outward into the base of each shear layer which is difficult to achieve with a flat grid plate. The design of Hakluytt and Tilston (9,10) overcame the first of these problems by using a 90 0 cone angle conical flame stabiliser. Their design aimed to overcome the second problem by progressively increasing the size of each succeeding jet shear layer to entrain the fuel and air mixture of the upstream shear layer. The increase in hole area from one row to the adjacent inner row was approximately 20/ and this resulted in approximately 50% of the air flow passing through the outer two of the nine holes per radial row. Combustion would occur once the shear layer was at a composition in the flammable range. Thus at high power the combustion would be mainly in the outer shear layer and at low power in the inner shear layers. Consequently, the design was aimed at achieving a wide flame stability by this jet shear layer staging process. To prevent liquid fuel, in droplets, sheets and vapour penetrating between the cone of holes and reaching the combustor walls, each row of holes was arranged in a curve. This resulted in all possible radial paths for a fuel droplet Presented at the Gas Turbine Conference and Exhibit Houston, Texas March 18-21, 1985

2 FUEL r' 2.0H H FUEL--- 60^ A AIR-- 32mm WEB FIG. 1. Grid Conical Flame Stabiliser being blocked by a jet shear layer. In the present work one of the 140 mm combustor designs of Hakluytt and Tilston for the Spey combustor was scaled down to fit the 76 mm combustor diameter used in previous work by the authors (1,3-8) as shown in Fig. 1. The design was investigated in its basic form without any interference from wall cooling or radial combustor wall air jets. The resultant simple design will be called a 'grid conical stabiliser with the abbreviation GC followed by a design type number. Two designs were studied with the same 10 row, 9 holes per row configuration and fuel injector. The first, GC5, was a simple scale of the Hakluytt and Tilston design. As this was for a combustor application with approximately only half the combustor air through the stabiliser, the blockage of 84.1% was higher than previously used by the authors (1,3-8). The primary zone thus operated with a higher residence time and lower mean Mach number. In the present work a Mach number of 0.03 was used and the GC5 stabiliser gave a pressure loss of 2.8%. To compare directly with previous work by the authors a lower blockage stabiliser, GC7, was designed with each hole increased in size by the same amount. This was designed with a blockage of 77.7% to give a pressure loss of 4.0% at the usual reference Mach number of and was also tested at the 0.03 Mach number where a pressure loss of 1.8i; was achieved. Both stabilisers were tested at 400K and 600K inlet temperature at the above Mach numbers. These inlet temperatures simulated at atmospheric pressure low power and high power conditions respectively. Previous work by the authors (4,5) at the reference Mach number of for the Grid Mix design with a 600K inlet temperature has shown an optimum equivalence ratio of 0.4 for minimum NO x with a high combustion efficiency. This is substantially weaker than envisaged in the work of Hakluytt and Tilston (9,10). For an overall air/fuel of 60/1, typical of an industrial gas turbine high power condition, the primary zone air flow for an equivalence ratio of 0.4 (37/1 A/F) results in the present mean primary zone Mach number of Clearly, if a different optimum primary zone equivalence ratio is required then this will result in a change in primary zone Mach number. Similarly, if a design for a richer overall operating condition was required, this would influence the primary zone Mach number and residence time, The fuel injector was a type of annular air assisted atomiser. It has some features in common with vaporiser designs (11) and has been referred to as an annular vaporiser (9). In the present work the fuel injector is used as a partial premixing tube with propane injection as shown in Fig. 2. In an associated investigation (12) it was used as an air assisted FIG. 2. Fuel Injection System. atomiser with liquid fuel injection. COMBUSTOR TEST EQUIPMENT AND GAS SAMPLING SYSTEM The electrically preheated 76 mm diameter combustor test facility has been described previously (1, 3-8). The conical flame stabiliser was located at the head of a 330 mm long 76 mm diameter uncooled combustor instrumentated with an axial row of wall thermocouples and static pressure tappings. Mean gas samples were obtained at the combustor exit plane using a water cooled ''X' probe with twenty holes on centres of equal area. The air to fuel ratio computed from this gas sample agreed to within 10% of the rig metered air to fuel ratio indicating that a reasonably representative mean gas sample was collected by the probe. The sample lines were electrically heated to 150 C and passed through a sample pump and filter mounted in an oven controlled at 150 C. A heated sample line connected this to the unburned hydrocarbons and NO x analyser which were internally heated. The sample was dried prior to passing to the CO and CO2 infra red analysers and the computer programme used to analyse the gas analysis results calculated the equivalent wet CO and CO2 concentration. This programme also calculated the air to fuel ratio, NO x emissions corrected to 15% oxygen at a standard humidity and the combustion inefficiency. The latter was based on the potential heat release in the CO and UHC in the exhaust gas. It took no account of equilibrium CO. An equivalent inefficiency due to equilibrium CO is shown in the present results. The difference between the computed inefficiency and the equivalent equilibrium CO inefficiency represents the true inefficiency. WEAK EXTINCTION CHARACTERISTICS At a constant mean velocity and temperature the fuel was gradually reduced until visual observation showed the flame to go out. The extinction process was also detectable from the gas analysis by a sudden increase in unburnt hydrocarbon (UHC) emissions. The results are summarised in Table 1. The weak extinction for premixed combustion were similar to previous measurements for flat grid plates (13) and exhibit only a small improvement due to the lower Mach number as shown in Table 1. The weak extinction mechanism has been shown to be explained by a turbulent flame blow off, with the flame stabilised at the tip of the jet shear layers just prior to blow off (13). At richer premixed mixtures, combustion occurs within the shear layers. For direct propane injection, rich zones within the shear layer permit combustion to be stabilised at much weaker overall mixture compositions (4,5). However, for a Mach number of 0.047

3 L -- BEE. 'I. Gt12h,REF. 1, 1 RE11KED,RED fl LDL 0 -- K'i,REII. 11. El2B,REF. I. S T 67 EF. fl, sru XFD ED 11 fldref S^Y,PP,U11KEE,RE I LOG10 Gr'iREO.7. 6t19.4FF. I. FJ E, cr- PREn -- + Yi,KF P n M w U L- w 0 0I_ EQUILIBRIUM CO 2 "INEFFICIENCY" Lb LJ JJ 't 9J A JJ b'd bj /li /J a y^ F0UIVALENCF RATIO FIG. 3. Variation of Inefficiency with Equivalence Ratio 600K. TABLE 1. Weak Extinction Equivalence Ratio Stabiliser Fuel M 400K 600K X1e 2 GC7 Premixed Propane Propane GC5 Premixed Propane Hole Premixed GM2B Propane Hole GM2A Propane Hole (Ref.5) (REF. M) the results in Table 1 show the present central fuel injection system to have an inferior flame stability to the previous work with direct fuel injection into each shear layer (4,5). This is a somewhat unexpected result as the greater variation of local air/fuel ratios in the shear layers were expected to improve the flame stability. Only at the Mach number (RED. M) at 600K did both the GC7 and GC5 stabilisers have a superior weak extinction. It is considered that the large number of relatively small holes may be a contributory factor to the poor flame stability. Previous work with premixed combustion (14) has shown that a very large number of small holes was detrimental to the achievement of a good flame stability. The reason for this was argued to be associated with the small shear layer size, the resultant small turbulent lengths and the consequent rapid decay of the generated turbulence. In the present work most of the inner shear layers were possibly too small to sustain significant combustion within the shear layer. This problem resulted from the scaling of the original 140 mm design (9,10) 3 EQUILI3RIUM Go "TNEF7IC,I N0 '0.; FO E ,fir 30 xie -2 FOUIVAL NCE RATIO FIG, 4. Variation of Inefficiency with Equivalence Ratio 400K. keeping the number and relative size of the conical grid plate holes the same. It is considered that a better performance would result if the hole sizes were kept the same as in the 140 mm design but the number of holes were reduced. Recent work with a much smaller number of larger holes with the same blockage has shown that an improved stability can be achieved. PREMIXED COMBUSTION The two flame stabilisers were tested with premixed propane and air at the 0.03 Mach number condition. This was achieved, as in previous work (1,5), by injecting propane, through twenty holes in an 'X injector, 1.5 m upstream of the stabiliser. Fig. 3 shows the combustion inefficiency of the two stabilisers at a 600K inlet temperature and 0.03 Mach number. Both stabilisers have a low inefficiency and the higher pressure loss of the GC5 stabiliser results in a reduction in the inefficiency compared with the GC7. Fig. 4 shows that at 400K and 0.03 Mach number the GC7 stabiliser has a very good efficiency, comparable with previous flat grid plate results at this pressure loss (1,5). This indicates that the stabiliser aerodynamics are good. Previous work has shown that premixed combustion with large recirculation zone systems such as baffles (1) or swirlers (6) result in a high combustion inefficiency and hence an inadequate basis for the design of weak primary zone systems. Figs. 5-8 present the unburned hydrocarbons (UHC) and CO emissions data at the two inlet temperatures. These show that differences in UHC between the two stabilisers account for the main differences in combustion inefficiency shown in Fig. 3. At 600K, the CO emissions are at the equilibrium level, except close to weak extinction. At 400K the CO levels are somewhat above the equilibrium levels. These indicate that for both stabilisers there is sufficient residence time and turbulent mixing available for optimum

4 a -+ GL`,REO. 1. Rfl,EF. h. -- GPRE 1YBEF fl G^:i,RH. h. Gi2E.4EF.. LOG1U I { 1 z o. G^,^,PER. 1. L2,4EE 1. -a- L5 RE%XEI,FD!1 L F7REI t12f EF. 1. w m, X18- EOUIVRCENCE RA? X10-7 EOUIV,ALENCE RATIO FIG. 5. Variation of UHC with Equivalence Ratio 600K combustion efficiencies to be achieved. Deviations from these results with central propane injection will be due to fuel and air mixing effects. The N0 x emissions for the two stabilisers at 600K inlet temperature and 0.03 Mach number are shown in Fig. 9 as a function of the mean gas temperature. These show no significant influence of stabiliser blockage or pressure loss. The levels are slightly lower than previous results for flat grid plates (1,5,14), in spite of the higher residence time. This indicates that the NO x may not be principally generated by the Zeldovich mechanism which predicts a linear dependence on residence time. Internal traverses of flat grid plate premixed flames have shown the NOx to be principally formed by a 'prompt mechanism with NO x generation in the shear layer region (14). CENTRAL PROPANE INJECTION THE INFLUENCE OF MACH NUMBER The 1' er blockage stabiliser, GC7, was tested at two Mach. ers, (REF. M) and 0.03 (RED. M), where the pressure loss was approximately 4.0 and 1.8% respectively. There was a 57% increase in mean residence time at the lower Mach number. However, the observed difference in performance may also be contributed to by the lower pressure loss and hence lower mixing forces and this may result in the influence of residence time being overestimated. Combustion Inefficiency Figs. 3 and 4 show that a significant reduction in combustion inefficiency occurs at the lower Mach number and this occurs at both inlet temperatures. Although the inefficiencies are higher than for premixed combustion the differences at the 0.03 Mach number are relatively small. This indicates that in spite of the central injection, the fuel was entrained rapidly into the multi shear layers giving a relatively uniform combustion close to premixed. Visual observations of the 4 FIG. 6. Variation of UHC with Equivalence Ratio 400K. flame show it to have a similar shape to a premixed flame and to fill the width of the combustor. There was no central core flame associated with the central fuel injection. UHC Emissions Fig. 5 shows that at equivalence ratios above 0,50 the high Mach number test gave lower hydrocarbon emissions than at the higher residence time. Thus in this region the higher turbulence levels are probably more important than residence time. For mixtures weaker than 0.5 the reverse occurred and in this region the longer residence time becomes more important than turbulence. At the low Mach number the UHC emissions were almost independant of equivalence ratio except close to weak extinction. Similar effects are shown in Fig.5 at 400K, although there is a stronger dependence of UHC on equivalence ratio at the low Mach number. CO Emissions Figs. 7 and 8 show that the effect of reducing the residence time on CO emissions was very significant, especially at 600K. Global CO oxidation mechanisms (15) show that for the same equivalence ratio and operating temperature the rate of CO oxidation would be dependant on the CO, 02 and H2O concentrations at the start of the uniform temperature region. The present shear layer combustion system may be split into a shear layer region where initial heat release and flame stabilisation takes place and a downstream CO and UHC burn-out region at more uniform temperature conditions. In the present work the ratio of the residence time in this post shear layer zone is the same as the Mach number ratio of However, the exit plane CO emissions are only reduced by this type of magnitude at the 400K inlet temperature. At 600K the reduction is of the order of a factor of 4 over a wide range of equivalence ratios. This can only be accounted for by an improvement in CO levels at the shear layer zone exit due to better mixing at the lower Mach number. At the

5 RI ED. I.. m2ef. I. -- GC7REt11YEEEI I GG^,RE^. 612A,REF. I, 5L7fEIUEIE5 I [ GF'6RED. fl, -- rm EF.! SURE. I. 6fl1P.,REF. 11. ^ 6Cl REE. 1 h. --+ Gds ar {FD.- e- H- 0 * a H- z 0 L EQUILIBRIUM CO ^ 1 EQUILIBRIUM CO 75 ' IO Ii e X18 ^ X1U FOU],ALEN.F R.A?lU FOUIVALENCE RA 5 1U FIG. 7. Variation of CO with Equivalence Ratio 600K. 400K condition mean velocities are reduced by nearly 20% due to the constant Mach number operation at a fixed stabiliser blockage. Thus the residence time in the shear layer and on a mean basis are increased by 20% relative to the 600K condition, This may partially explain the lower influence of Mach number at 400K. The CO levels are higher than for premixed combustion, except in the weak region where the premixed situation is close to weak extinction. These higher p - emissions reflect the local variation in equivalence ratio in the shear layers. Local rich zones generate high CO emissions which do not have time to burn out in the post flame gases. Clearly, this effect will be strongly influenced by pressure loss as this governs the turbulence energy input into the shear layer mixing process (4). NO Emissions The NO x emissions at the 600K condition are shown as a function of flame temperature in Fig. 9 and in Fig. 10 as NO x corrected to 15% oxygen as a function of equivalence ratio. The influence of the reduction in Mach number at constant blockage was significant and varied with the flame temperature. At high temperature and equivalence ratios richer than 0.45, the NO x increased by approximately 30% compared with an increase in mean residence time of 57%. This increase may be explained if the NOx was generated by the Zeldovich mechanism in the post flame gases. In this region the flame was burning in the shear layer with an early heat release, as shown by the wall temperature profiles in Fig.11. However, at equivalence ratios below 0.45 the flame did not burn to the same extent in the shear layers at the Mach number, as shown by the low wall temperature in Fig. 11. At the reduced Mach number, shear layer burning continued to be very effective, as shown in Fig. 11, and the NO x emissions were increased by a factor of two, much larger than the residence time. The burning in local rich zones in the shear layer at the lower Mach number FIG. 8. Variation of CO with Equivalence Ratio 400K. produced this large increase in NOx. INFLUENCE OF PRESSURE LOSS AT CONSTANT MACH NUMBER The variation of Mach number, M, for a fixed stabiliser, simultaneously varies the pressure loss, LW/P, as shown by Equation 1, A 2 y (^ ^) (1) D 2 where y is the ratio of specific heats. is the Discharge Coefficient, Al is the approach pipe area. A2 is the stabiliser total open area. Thus the Mach number or residence time effects discussed above may also be contributed to by the pressure loss and consequent changes in turbulent mixing (4).The GC5 stabiliser was designed to achieve a similar pressure loss at.03m as the GC7 stabiliser produced at 0.047M (40%). However, manufacturing differences for the large number of holes resulted in a somewhat lower pressure loss, but one that was significantly higher than the GC7 stabiliser gave at the.03 Mach number (1.8%). Thus a more realistic influence of residence time may be obtained by comparing the GC7 stabiliser at 0.047M with the GC5 stabiliser at.03m. However, comparison of GC7 and GC5 at.03m also gives the influence of an increased pressure loss at constant residence time. Figs. 3 and 4 show that the increased pressure loss of the GC5 stabiliser at 0.03M produced a significant fall in the combustion inefficiency with levels close to that for premixed combustion at the optimum equivalence ratio. This occurred at both inlet temperatures and Figs. 5-8 show that it was mainly due to a reduction in CO emissions, with little change in UHC. This may be explained by the improvement in shear layer mixing at the higher pressure loss giving fewer and shorter duration locally rich zones. This is also supported by the 5

6 0 - GC5;RF^ 9. G521EF. h ,PRBllE,ED M X19 16rJ,4E. 9. Gn2R EF. I. ---e.-- 6f5 RFD h. 6!12h,EE a- irejieeree h 6fi. p-r M _-- F1^A.P1- M X71 Ln U U. U 2, O X ' IS 50 c ?. FLR^1E TEMPERATURE!DEC,. K KlU ' tit 50 FDUIVALENLF R,A`]D X10 - ^ FIG. 9. Variation of NO with Flame Temperature 600K. FIG. 10, Variation of Corrected NO x with Equation Ratio 600K. NO x emissions in Figs. 9 and 10, where there is a reduction almost back to the level of the GC7 stabiliser at M= Thus residence time effects on NO x are not very significant provided that the pressure loss is maintained the same. COMPARISON WITH PREVIOUS SHEAR LAYER MIXING COMBUSTION SYSTEMS The authors are involved in an extensive systematic study of shear layer stabiliser combustion systems for gas turbines using a range of stabiliser geometries (1,3-8). Although several investigators are using swirl systems (6-8) no previous work apart from Hakluytt and Tilston (9,10) has investigated fuel injection into simple jet shear layers. Previous work by the authors (4,5) has investigated a flat stabiliser with each jet hole separately fed with propane using two injection techniques. The most appropriate results to compare with the present work are the Grid Mix 2 results in Ref. 6. In this work the propane was fed into the outer part of the shear layer by an annular fuel supply around each hole. Tests were carried out in a 16 and 4 hole configuration at the 0.047M condition with a similar pressure loss to GC7. Both the GM and GC system feed fuel into the outer part of the sheer layer giving local rich zones in the low velocity intenet regions. The GM2 results are compared with the present in Figs and are directly comparable with the GC7 at 0.047M. The weak extinction results have already been discussed and shown to be superior to the present results. Fig. 3 shows a major reduction in the combustion inefficiency at 600K with results comparable with GC5 at.03m. However, at 400K the two sets of data are very similar except at very weak equivalence ratios, where the better stability of the GM design gives it an advantage. At 600K both UHC and CO emissions are well below the GC7 design. NO x emission for the GM2B 16 hole designs are very similar to the GC7, except at very weak mixtures where Fig. 11 shows the flame does not burn effectively in the shear layer region, However, the GM2A 4 hole results have lower NO x emissions than the GC designs without any deterioration in combustion inefficiency. Consequently, it may be concluded that the present large number of holes is not a major requirement of the system and advantages could be obtained by operating with a reduced number of holes. Recent work has shown this to be true and improvements in all performance and emissions parameters have been achieved. Although, there are clearly penalties in CO and UHC emission in using a central fuel injection system, the advantages in mechanical simplicity and compatability with existing combustion chamber fuelling arrangements make the GC design attractive, As optimum primary zone equivalence ratios, of approximately 0.4 are richer than current overall equivalence ratios a reduced primary zone Mach number is required. With this the GC5 design has similar CO, UHC and NO x emissions to the GM2B design. Future work will investigate possible improvements in the GM design at low Mach number, but these will be difficult to achieve in view of the low emission levels at.047m. The success of central fuel injection in injecting fuel into a large number of shear zones is the most important result from the present work. Although the technique is not as effective as direct shear layer injection it is a much simpler technique. The reason it is successful is due to the high entrainment forces at the base of a jet system. Jets entrain their own mass flow approximately every two jet diameters. Thus the fuel will be rapidly entrained at the base of each jet. Comparison of the present work with the directly fed shear layers of the GM design (5) shows that a jet shear layer can stabilise burning over a wide range of equivalence ratios and hence the graded jet hole size concept of the GC design (9,10) may not be necessary. This may result in much simpler stabiliser designs with fewer number of holes. 6

7 6SRED. PHI. 422,REf I -- -PH1,fii,RED tu U. HU21,RFF. U. 6C7,PRFIIIXED,RFD tu X10 1 $PH1.16":,REFn (OG10 H121 RFD. I. 012B,REF. I. f ^ r w O_. w _z e LL LL. cn C2 m } ^ o oy ` / o 4 6 y H z B :,0 52 X10 1 RXJRL POSITION /MM FIG. 11. Axial Wall Temperature Profiles; GC7, 600K.. OPTIMUM PRIMARY ZONE CONDITIONS For low NO x emissions the weakest primary zone is required that is compatible with achieving a low combustion inefficiency. The correlation of combustion inefficiency with NO x emissions corrected to 15% oxygen is shown in Fig. 12 for the 600K condition. This shows that reducing the Mach number at constant blockage produces an increase in the optimum NOx which is eliminated once the blockage is increased to achieve the original pressure loss. At one atmosphere pressure the EPA 75 ppm corrected NO x regulation is equivalent to approximately 24 ppm assuming a 10 bar engine condition and square root pressure dependence of NO x. Fig. 12 shows that the GC5 at 0.03M meets this requirement with a very low combustion inefficiency. Fig.10 shows that for the GC5 stabiliser at 0,03 M the minimum corrected NO x emission compatable with a low combustion inefficiency is approximately 13 ppm. This level of NOx is achieved at an equivalence ratio of 0.37, as shown in Fig. 10. This optimum equivalence ratio is close to that for the four hole grid plate GM2A and the 16 hole grid plate GM2B in Ref. 5 at 0.4 and 0.78 respectively. However, the optimum NO x for GM2A was 7.5 ppm compared with 17 ppm for the GM2B and 13 ppm for the present Gc5 results. Thus the central fuel injection technique of the present design has achieved comparable NOx and combustion inefficiency emissions to the more complex system of directly fuelling each jet shear layer. For the GC7 stabiliser at 0.03 M the equivalent optimum condition is 24 ppm NO x at an equivalence ratio of At the reference Mach number this stabiliser had an optimum performance of 14 ppm NOx at the same 0.45 equivalence ratio, although the combustion inefficiency was unsatisfactory at this condition. The present results demonstrate that NO x emissions and combustion inefficiency can be simulataneously reduced by reducing the Mach number whilst increasing the blockage. An optimum primary zone equivalence ratio of 0.37 has been identified for the design, ?S NOX CORRECTED /PPNI FIG. 12. Variation of Inefficiency with Corrected NO x ; 600K. substantially lower than originally postulated by Hakluytt and Tilston (9,10). CONCLUSIONS 1. Central fuel injection in the Grid Conical design achieves effective fuel entrainment into a large number of jets. This is a much simpler design concept than direct individual shear layer fuelling. 2. Primary zone Mach number has a very strong influence on combustion inefficiency, especially if the stabiliser blockage is increased to achieve the same pressure loss at the lower Mach number. 3. NOx emissions are not strongly influenced by primary zone Mach numbers, for the same stabiliser pressure loss. This indicates that the NOx is mainly generated in the shear layer region, where the residence time is constant, and is not strongly influenced by the increase in residence time in the post flame gases. 4. A reduction in the number of jet shear layers could bring about substantial improvements in flame stability and combustion inefficiency. ACKNOWLEDGEMENTS We would like to thank the U.K. Science and Engineering Research Council for a research grant in support of this work (GR/B/92812). Technical discussions with Mr. J.B. Jamieson, J.P.D. Hakluytt and J.R. Tilston at the U.K. Royal Aircraft Establishment, Pyestock are gratefully acknowledged. The test rig was operated by Mr. R. Boreham.

8 U REFERENCES 1. Al-Dabbagh, N.A. and Andrews, G.E., The Influence of Premixed Combustion Flame Stabiliser Geometry on Flame Stability and Emissions. Trans. ASME J. Eng. Power, Vol. 103, 1981, pp Sotheran, A., Pearce, D.E. and Overton, C.L., Some Practical Aspects of Staged Premixed, Low Emissions Combustion, ASME Paper 84-GT Andrews, G.E., Al-Dabbagh, N.A. and Abdul-Aziz, M.M. Mixing and Fuel Atomisation Effects on Premixed Combustion Performance, ASME Paper 83-GT Al-Dabbagh, N.A. and Andrews, G.E., The Influence of Flame Stabiliser Pressure Loss on Mixing, Combustion Performance and Flame Stability. Sixth International Air Breathing Engine Symposium, Paris, AIAA Paper , Al-Dabbagh, N.A., Andrews, G.E. and Manoharan, R., Shear Layer Mixing for Low Emission Gas Turbine Primary Zones, ASME Paper 84-GT-13, Andrews, G.E. and Ahmad, N.T., Emissions from Enclosed Swirl Stabilised Premixed Flames, ASME Paper 83- GT-192, Ahmad, N.T. and Andrews, G.E., Gas and Liquid Fuel Injection Into an Enclosed Swirling Flow, ASME Paper 84-GT-98, Ahmad, N.T., Andrews, G.E., Kowkabi, M. and Sharif, S.F., Centrifugal Mixing Forces in Enclosed Swirl Flames, Twentieth Symposium (International) on Combustion, The Combustion Institute, Hakluytt, J.P.D. and Tilston, J.E.,, On the Prospects for Close Control of Combustion by Shear Layer Mechanics, I. Mech. E. Conference, Combustion in Engineering, Oxford, Paper C87/83, Tilston, J.R. and Hakluytt, J.P.D., The Design and Performance of a Combustor with a Multiple Jet Primary Zone, AGARD CP-353, Combustion Problems in Turbine Engines, Paper 17, Sotheran, A., The Rolls Royce Annular Vaporiser Combustor, ASME Paper 83-GT-49, Ali, A.F. and Andrews, G.E. Paper in preparation. 13. Al-Dabbagh, N.A. and Andrews, G.E., Weak Extinction and Turbulent Burning Velocity for Grid Plate Stabilised Premixed Flames, Combustion and Flame, Vol. 55, p.31-52, Al-Dabbagh, N.A., Emissions and Stability of Gas Turbine Combustors with Rapid Fuel and Air Mixing, Ph.D. Thesis, University of Leeds, Howard, J.B., Williams, G.C. and Fine, D.H., Kine= tics of Carbon Monoxide Oxidation of Post Flame Gases, Fourteenth Symposium (International) on Combustion, pp , 1973.