You are on page 1of 11

C O M B U S T I O N A N D F L A M E 26, 311-321 (1976)

311

High Intensity Oxy-Methane Combustion in a Jet-Mixing Burner


R. J. HARRIS*, W. HORNE and A. WILLIAMS
Department of Fuel and Combustion Science, The University of Leeds, Leeds LS2 9JT.

An investigation has been conducted into the combustion of methane with oxygen in an 0.6 MW jet-mixing burner. Stable flame species concentrations, noise levels, and combustion intensities have been measured over a range of equivalence ratios. A number of isothermal studies have been undertaken and the primary combustion region of the flame is discussed in terms of a simple entrainment model.

Introduction The use of oxygen-natural gas burners in large scale industrial applications where high temperatures, high combustion intensities and high rates of heat transfer are required present a number of design problems. Firstly, as a consequence of the danger of flashback in large burners, some form of jet mixing of the reactants must be employed. The mode and rate of mixing of the two streams of reactants thus determines the combustion intensity and the flame shape, both of which are important practical factors. Secondly, because of the high gas velocities and rapid combustion rates, burner noise levels can be extremely high and the reduction of these to an acceptable value often imposes an upper limit on the designed rate of mixing of oxygen and methane at the burner face. Finally, the high gas velocities may also create flame instability problems due to the relatively low flame speed of natural gas. A number of different burner designs have been proposed of which the principal are the Shell Toroidal burner [ 1,2] which uses internal jet mixing and recirculation for flame stability, the British Gas Corporation M.R.S. Mark II burner which uses impinging jet face mixing [3,4] and the British Gas Corporation M.R.S. Mark III burner [5,6] which uses coaxial jet mixing. Few investigations have been made involving oxy-fuel burners in which jet mixing
*Present address: British Gas Corporation Midlands Research Station, Solihull, Warwickshire.

occurs because of the practical difficulties involved, and the object of this paper is to describe some recent studies concerning the mixing properties and flame characteristics of the British Gas Corporation Mark III burner. The Mark III burner is manufactured under licence by B.O.C. Ltd., and is available in a range of sizes. The present investigation was conducted using the smallest burner which was rated at 570 kW (20 therm/hr). The burner, shown schematically in Fig. 1., consists of a cluster of six oxygen gas supply nozzles of frustro-conical end section arranged annularly within a water-cooled fuel supply duct. In operation the oxygen supply issues from the nozzles at high velocity relative to the fuel supply. As the fuel passes the end of the common supply tube it expands and fills the space between the nozzles such that upon reaching the end section of the nozzles large interfacial areas are created between the respective fluid streams. Entrainment of the fuel stream by the oxygen supply takes place, resulting in mixing and subsequent combustion. The divergent endsection of the oxygen nozzles enables the burner to operate at high gas exit velocities despite the low burning velocity of natural gas [5].
Experimental Oxygen (99.8%) and methane (99%) were obtained from cylinders without further purification and a dried and filtered compressed air supply was available in the laboratory. All gas

Copyright 1976 by the Combustion Institute Published by American Elsevier Publishing Company, Inc.

312

R . J . HARRIS, W. H O R N E AND A. WILLIAMS Table 1. Squire and Trouncer [7] in a study using a single jet exhausting into a co-current stream, have shown, however, that at such high values of h the entrainment characteristics are similar to those exhibited by a jet exhausting into stagnant surroundings. In view of this, the dissimilarity of the injection momentum ratios in the present instance was neglected. Figure 2 shows radial velocity profiles at a series of axial stations from the burner face. Initially the jets of oxygen simulant remain separate and exhibit mean velocity profiles with maxima centred on the individual axes of the nozzles comprising the oxygen delivery cluster. As these jets expand and entrain surrounding fluid, the initial high exit velocities progressively decay, and by a distance of 10.5 cm from the nozzle cluster, the jets have coalesced to form a single stream whose mean velocity profile is centred on the burner axis. In Figure 3, the decay in oxygen simulant concentration close to the burner mouth is shown along the axis of an oxygen delivery nozzle. The axial distance has been normalised in terms of the diameter (d) of the oxygen nozzle and it is apparent that there is no measurable potential core region, and that the entrainment efficiency is slightly dependent on throughput. In addition, the nonlinear dependence of decay rate on axial distance is indicative of a transition region in which the similarity criterion for fully developed turbulent jets cannot be strictly applied. The absence of potential core is unusual, but agrees with the results of experiments in which

flows were m e t e r e d using p r e c a l i b r a t e d rotameters. The burner was fired horizontally into the open laboratory and the flames obtained were examined by the use of direct colour (Kodak Ektachrome) and Schlieren photography. Flame gas samples were removed by means of a single orifice water-cooled probe, dried, and analysed by gas chromatography. Isothermal experiments were carried out using air to represent both fuel and oxidant flows, and gas velocities under these conditions were measured with a single wire anemometer (DISA Ltd.) and with pitostatic tubes. The mixing pattern of the isothermal fluid streams was followed by monitoring with an infra-red analyser propane tracer gas added to the air supply. Sound frequency spectra were obtained with a Bruell and Kjaer type 411 standard condenser microphone positioned 2 m fron the burner face at a forward angle of 30 to the burner axis. This was coupled to a frequency analyser which measured sound pressure levels at 1/3 octave intervals.
Results and Discussion
(A) Mixing Studies

The mixing characteristics of the burner were studied isothermally on the basis of a Reynolds number similarity criterion. The calculated values of pipe Reynolds number, together with other relevant exit parameters are listed in Table 1 for one quarter, and for fully rated stoichiometric flames. Under isothermal conditions, the injection momentum ratio (h) was approximately one half the value listed in

I 42ram

lOre ~ ' ~ ~ z m.~ 'O

~'~.~41~ZZ

/ f
. ._

--

~ FUEL I FUEL I

"~,

FUEL

Fig. 1. Two views of the burner head assembly.

OXYMETHANE COMBUSTION

313

TABLE 1 Exit Parameters for Representative Stoichiometric Flames Nominal rating (kW) Vol. flow rate (m a rain- 1) Fuel Oxygen 0.25 1.00 0.5 2.00 Calc. mean exit velocity (cold) (m sec -1) Oxygen 43 173 Reynolds No. (cold) ( 10-4) Oxygen Injection momentum ratio (pluj/p2u~)

Fuel 4.6 18.2

Fuel

143 570

~.0
3.8

1.8
7.3

) )
)

18.7

a tracer gas was injected into the air representing the fuel supply. In this case, finite tracer concentrations were measured within the conical sectionof the oxygen nozzles. Further, if magnesium oxide smoke was incorporated in the fuel simulant upstream of the burner exit, white oxide deposits were clearly visible within the exit cones of the oxygen nozzles. These observations, together with a slight dependence of decay rate on nozzle throughput are consistent with a rapid wake-like mixing phenomenon localised within the divergent section of each oxygen nozzle tip. The variation in velocity along the burner axis is plotted in Fig. 4. In this case the axial distance is nondimensionalised using the diameter (D) of the fuel supply-duct. The axial velocity reaches a maximum as X/D-2.3 and then decays with increasing distance from the burner face. Whilst the mixing behaviour exhibited in this isothermal study provides an interesting qualitative insight to the flame behaviour, attempts at quantitative application should be treated with caution. This is due to two factors: firstly, uncertainty concerning the mechanism of interaction within an array of jet flames; and secondly, flame buoyancy, which was not modelled in the cold-flow study. In order to estimate the importance of buoyancy effects in the flame, Froude numbers (Fr) were calculated using an equation proposed by Ricou and Spalding [8] for combustion systems:

where the subscripts (j) and (a) refer to the properties of the jet and the surroundings respectively, Ysz is the mass fraction of fuel in the jet, Ah is the heat of combustion per unit mass of fuel, C the specific heat, T the absolute temperature, and dj the jet diameter. Two Froude numbers were considered for each flame. The first of these (F~) corresponded to the region close to the burner mouth where oxygen jet interaction effects were considered to be negligible and buoyancy forces were therefore estimated on the basis of the diameter (d) of an oxygen delivery nozzle and the mean oxygen exit velocity. The second Froude number ( F ; ' ) was applicable when jet interaction was complete and the burner fluid behaved as a single stream. In this case the characteristic parameters were taken to be the fuel duct diameter (D) and the mean total fluid exit velocity. The minimum values of F; and F ~' (corresponding to the burner operating at one quarter full nominal rating) were 500 and 3 respectively. The data of Ricou and Spalding show that for values of(x/dj) (p2/pj) 1/2Fr-1/2 exceeding approximately 2.1, entrainment into a flame becomes increasingly dominated by buoyancy considerations. In the present case, therefore, only entrainment into the mixing region of the flame close to the burner mouth, can be regarded as primarily thrust-controlled.
(B) The Visible Flame Structure

r ~kg.d ] j

PJ I [Yf,jAh+Cj(T/-Ta)]

The flames consisted of six intense, blue combustion jets stabilised within the exit cones of the oxygen delivery nozzles, and surrounded by a less intense blue, afterburning mantle. In addition, a relatively small yellow

314
, ,

R.J.

HARRIS,
,

W. HORNE

AND A. WILLIAMS
|

(e)

40 ~o
30

20

,ot
,O{-o.~
o "~.2~1

td)

t
t

,
I

~ I

/.,0

ot
2O 10 0

,,

,"iv

(.b)

t
IM

100 80 60 /..0 20 .30

IA:
.20

,,10

-fl-i

AXIS -10 -20 RADIAL POSITION (ram)

-30

Fig. 2. Isothermal results: radial mean velocity profiles (a) at the burner face, and(b) 2.1 cm;(c)4.2cm;(d) 10.5 cm; and (e) 21 cm. from the burner face.

OXYMETHANE COMBUSTION
lO

315 flame was observed outside the main oxymethane flame and this was assumed to be due to a small proportion of the methane burning in air as a fuel-rich diffusion flame. Measurements from colour photographs of the stoichiometric flames showed that the length of the inner combustion jets (0.20 m) was apparently independent of throughput and that the length to the tip of the visible afterburning mantle increased from 0.70 m at one quarter maximum nominal burner rating to 1.30 m at full nominal rating. In addition, the afterburning mantle was deflected upwards from the burner horizontal axis to an extent which decreased with increasing throughput. A typical flame is shown in Fig. 5.
(C) Flame Composition

06
z

06
z

53

~_
m l

~o~
I I I I I I [

4 6 AXIAL DISTANCE

10

12

14

Fig. 3. Isothermal results: axial decay of oxygen jet fluid along a jet axis. A, full flow rating; I-1, % flow rating; O , flow rating; , V4 flow rating.
>-

Species concentration profiles were used to demonstrate that the visible tip of the combustion jet corresponded to 99% complete combustion on the basis of residual methane levels in the flame gases. Some representative results are shown in Fig. 6 for a stoichiometric flame on the burner operating at full rating (570 kW) and demonstrating the following points:
1 r 1

I 2O

k-

O ~J1:75 > LLI 7" .<

1.5
x

Ld ~'~'1.25
M

< < ~075 Ld #r"


bO

<

I 2.5

I 5

I 75

__

10 ; ~

AXIAL

DISTANCE FROM

0 2 CLUSTER

Fig. 4. Isothermal results: variation of mean velocity along the burner axis. A, full flow rating; I--],~ flow rating; O , flow rating; , 1,4 flow rating.

316

R . J . HARRIS, W. H O R N E AND A. WILLIAMS

Fig. 5. Oxy-methane flame; tb = 1, 190 kW.

(i) Oxygen leaving the nozzles is consumed rapidly until a low concentration results at about 0. l m from the 02 nozzle plane, beyond this region the oxygen concentration at all radial positions increases as a result of air entrainment, (ii) The natural gas concentration shows a steady decline up to 0.2 m from the nozzle plane. This fact was also demonstrated independently by use of the probe in conjunction with the infra-red hydrocarbon analyser, (iii) Nitrogen is entrained extensively into the flame gases from the surroundings, this being very marked by about 0.1 m from the nozzles, and (iv) The concentrations of both CO and COs are small at all radial positions at positions close to the burner face. The concentrations then rise sharply in the region of 0.0775 m to 0.1 m from the nozzles and beyond the position of the flare, carbon monoxide profiles show higher concentrations at the flame centre than at the flame edge, whilst carbon dioxide profiles exhibit the reverse. It is clear that 0.1 m from the nozzles the combustion process is

dominated by the afterburning of carbon monoxide.


(D) Combustion Intensity Measurements

The volumetric heat release rate or combustion intensity (CO is commonly determined from experimental data by use of the relation:CI = F ' H / V c ' P ,

where F is the fuel firing rate, H is the heat of combustion to give completely oxidised products at S.T.P., V, is the volume of the combustion zone and P is the pressure. In the case of jet mixed flames, such as those considered here, the parameter Vc presents some difficulty in definition. It is often common practice for Vc to be taken as the volume of the luminous flame zone. However, with high temperature flames exhibiting a large degree of dissociation simple visual estimates of V, based on the luminous zone may be misleading, since the burnt-gases are highly radiating. In the present study the combustion volume was considered to occupy a cylinder of length determined by the position of measured fuel burn-out, and with a width obtained from flame photographs. In addition, because the flame gases at the position of fuel burn-out are at temperatures at which there is considerable dissociation, the

OXYMETHANE COMBUSTION
O.0375m FROM Oz NOZZLE PLANE I I , , w

317

100
9O

O.OZrSm FROMOzNOZZLE PLANE 100 , , , , , x CH. 90 02

50rz

O.lm FROM 02 NOZZLE PLANE

.o
,o o

80 ~
t

Q N~

4O

i
RADIALDISTANCE(cm)

60 5O 20 3O 15 10

~J

,-

,a t ~ ty/',
b

Fig. 6. Radial distribution of stable flame species at various axial positions from the burner face. (For clarity the position of an oxygen nozzle is marked on the axis).

use of a value of H based on reaction products at S.T.P. is erroneous. Because of this, the effective heat liberated in the flame was calculated from heats of formation based on computed species mean equilibrium concentrations corresponding to the actual flame temperatures. Mean combustion intensities calculated from these reaction volumes and effective heats of reaction are plotted in Fig.7 for various stoichiometries. Additionally, the mean volumetric heat release rates in the luminous afterburning region of the flames are shown in the same figure. These were calculated from the volume of the luminous afterburning zone and the effective heat released by the equilibrium flame species to give completely oxidised products at S.T.P. Thus it can be seen that the heat release rate in the afterburning zone is an order of magnitude less than in the primary reaction zone, although the quantity of heat released in each region is of the same order. The combustion intensities obtained here are high, the maximum value obtained 1.23 x 103 MWm -~ being achieved at ~b = 1 and full rating (570 kW). Overall, the values are similar to those obtained in other high intensity oxy-fuel

burners [2,4]. They are however, considerably lower than theoretical values for a well-stirred reactor situation calculated either by flame strength considerations, or on the basis of laminar flame properties. Thus Bittker and Brokaw [9] have shown that the upper limit for the chemical kinetic controlled heat release rates for a gaseous hydrocarbon with oxygen may be as high as 4.7 x 106 MWm -8. It may be concluded therefore, that even in these high temperature, highly turbulent flames, the rate determining step must be the mixing process.
(E) Primary Combustion Length

In oxy-fuel jet mixing flames, the overall visible flame length and the length to fuel burn-out are not normally analogous. In the present instance the visible flame was composed predominantly of a buoyancy-controlled afterburning zone consisting of radical recombination reactions and combustion of carbon monoxide with entrained air. The primary fuel-oxygen reaction was localised within a relatively small volume close to the burner face, where the reactants were mixed by thrustcontrolled entrainment of methane into the oxygen jets. Because of the complexity of the

318
1500

R . J . HARRIS, W. HORNE AND A. WILLIAMS diameters, (0.038 m) of the burner face. This is considerably less than the experimental burnout length, but it is instructive to compare this result with the calculated burn-out distance for the fuel and oxidant interchanged; that is, with six high velocity methane jets entraining low momentum oxygen gas. In this case, primarily due to stoichiometric considerations, the burn-out length increases by a factor of about 3. Generally therefore, it may be stated that in jet mixing burners, combustion intensity will be enhanced by entraining the fuel into high velocity oxidant streams, rather than the reverse case, providing of course, that flame stability, etc., are not adversely affected.
(F) Flame blow-off

lOOO

500

-,
06 0"8 10 1.2 1'4 1"6

-o
1"8

--,,.

Fig. 7. Variation of heat release rate with equivalence ratio aA, full rating; o, estimated from the volume of the primary combustion zone; 0, estimated from the volume of the afterburning region.

afterburning region, the ensuing discussion will be confined to the primary reaction zone. Ricou and Spalding [8] have shown that entrainment into a turbulent thrust-controlled jet of fuel burning in still air can be described by the relationship:

M/Mo = 0.32 x do (fla / PX)1/2


where Mo and M are the mass flow rates at the nozzle and at a distance x from the nozzle respectively, and the other symbols have been defined previously. The validity of this relationship for oxy-fuel combustion has not been evaluated, and a further complication arises in the present case, as cold-flow studies have indicated that stoichiometric entrainment is essentially complete within the transition region of the high velocity jets. Nevertheless, the application of this equation provides an interesting insight into the factors determining mixing length, and hence combustion intensity in oxyfuel systems. Thus, if the burner-face mixing region is regarded as six independent, high Froude number oxygen jets, then each of these will have entrained the stoichiometric proportion of methane within approximately six jet

The flame stabilises within the conical exit sections of the oxygen nozzles. Previous workers [ 1] have demonstrated that the oxygen flow rate, at which flame blow-off occurs is critically dependent on the included angle of this conical section, reaching a maximum at 36 . Further, flame stability is relatively unaffected by the magnitude of the fuel flow rate and by the thickness of the nozzle wall separating the fuel and oxidant streams. These facts point ot the existence of a recirculating region within the cone, caused by the separation of the oxygen fluid stream from the nozzle walls at the commencement of the conical section. In order to obtain an indication of the intensity of the recirculation zone and its dependence on the magnitude of the nozzle exit angle, a series of perspex models of a single oxygen nozzle were constructed in the present work, with bore 12 mm and exit angles varying from 20 to 60 o (Fig. 8(b)). Air at room temperature was exhausted through each model at a pipe Reynolds number of 8000 and the pressure distribution within the exit sections was investigated by means of a pitot microprobe positioned at right angles to the major flow axis and coupled to a sensitive flexible-membrane manometer (Furness Controls Ltd.). Figure 8a shows typical pressure profiles for each of the models studied. It is evident that a total pressure minimum exists close to the nozzle face, corresponding to a region ofrecirculation, and becoming most pronounced at nozzle exit angles within the range 30 to 40 o.

OXYMETHANE

COMBUSTION

319

(a)

(b)

EXIT A N G L E

I00

( VAI~IED )

V---7
50

12.5 m m

F'I
37.5 mm

I
6 7

I
8

I
9

I
I0

I ~6 I
II
mm

AXIAL DISTANCE /

Fig. 8(a) Radial total pressure profiles at an axial station 3.0 him upstream of nozzle exit plane for various nozzle exit angles. Nozzles operating on air at Re = 8000 and exhausting into ambient surroundings. (b) Section through nozzles used.

(G)

(b)

.-------- O

"

O--

36

300

>ZOO J
w >

LO

g o
uJ

100

/
' J 16 rnm

t
I

I
2

t
3

LENGTH OF CONICAL SECTION / /

L/%

Fig. 9(a) Variation of mean oxygen exit velocity at flame blow-off with L Mo for a single oxygen nozzle. (Mean methane exit velocity = 1.1 m sec "1) (b) Section through nozzle arrangement.

320

R. J. HARRIS, W. HORNE AND A. WILLIAMS


TABLE 2 Overall Noise Levels

Nominal rating (kW) 143 285 428 570 528 560 614 336 370

<~ Burning 1 1 1 1 1.25 1.5 1.75 0.7 0.8 77 88 91 96 91 89.5 88.5 90.5 91

Noise level (db (A)) Non-burning 74 83 87.5 91.5 86 85.5 85.5 89 88

In view of the high resistance to flame blowoff which is achieved by means of promoting recirculation of hot gases in this manner, it might be expected that a further improvement in flame stability could be effected by adjustment of the length of the conical section, in order to optimise the extent of the recirculation zone. This possibility was tested by the use of a single oxygen nozzle surrounded concentrically by a fuel gas delivery tube (Fig. 9(b)). The mean oxygen exit velocity at flame blow-offfor this nozzle is shown in Fig. 9(a) as a function of the ratio L/do, where (L) is the length of the conical section and (do) is the bore of the oxygen nozzle. It can be seen that the blow-off velocity increases with increasing L/do and approaches an assymptote at a value of L/do equal to ca. 3. For comparison, the predicted blow-off performance of the nozzle geometry adopted in the Mk III burner is arrowed in the same figure.
(G) Noise Measurements

90

T8O ~'70 u460

50

100 200 500 1000 2000 FREQUENCY (Hz)

5000 10000

Fig. 10. Noise frequency spectra ~ lean flame, (ok = 0.8) 370 kW nominal rating; . . . . . burner operated isothermally at similar Reynolds numbers.

The noise levels obtained for flames together with those produced by the corresponding isothermal gas flows are shown in Table 2. Figure I0 shows frequency spectra obtained under typical burning and nonburning conditions. The increase in the low frequency component of noise in the presence of combustion is typical [I0] and is due to the rapid expansion of individual elements of fuel. The apparent

damping of high frequency noise under burning conditions has previously been observed with swirling flows [11] and is thought to be due to an alteration in the directional characteristics of the noise source. Spectra for all flames exhibited a similar frequency characteristic with a maximum centred around 500 Hz, and the maximum S.P.L. was almost independent of stoichiometry, but varied exponentially with oxygen exit Reynolds number. Overall S.P.L. in the present investigation under flame conditions were found to be considerably lower than those obtained with oxyoil combustion [2, 12] where a large proportion of the noise is due to the small diameter passages which are necessary to ensure efficient blast atomisation of the oil.

OXYMETHANE COMBUSTION We would like to thank the Science Research Council for a research grant and a research fellowship (W.H.) and the British Gas Corporation for a research studentship (R.J.H.).

321
6. Harris, R.J., Home,W. and Williams, A.,in Combustion Institute European Symposium (F.J. Weinberg, Ed.) Academic Press 1973, p.589. 7, Squire, H.B. and Trouncer, J,.A.R.C.Rept. No. 1974 (1944). 8. Ricou, F.P.,and Spalding, D.B., J. FluidMech, II 1 21 (1961). 9. Bittker, D.A. and Brokaw, R.S., A.R.S. Journal 30, 179 (1960). 10. Kilham, J. and Smith, T.J.B. , J. Accoust. Soc. Am. 35,715 (1963). 11. Putnam, A.A. and Giammer, R.D., Summary Rept. Am. Gas Assoc. Inc. (1971). 12. Durrant, A.W. and Blunt, D.H., J. Iron & 207, 452 (1969).

References

1. Bagge,L.P., OPD Rept. No. 228/64 MShellInt. Pet. Co. Ltd. (1964). 2. Alfred, J.W., Baxendale, D.N., Chamberlanin, C.T., Roberts, A.L. and Williams,A..J. Inst. Fuel 44, 99 (1971) 3. Arnold, G.D., J. Inst. Fuel 40, 117 (1967). 4. Baxendale, D.N., Horne,W. and Williams, A.,J. lnst Fuel 47, 139 (1974). 5. Pomfret, K.F. Gas Council M.R.S. Rept. No. 120 (1968).

Received 23 July 1974; revised 20 August 1975

You might also like