EPA-600/2-76-061a
March 1976
Environmental Protection Technology Series
BURNER CRITERIA FOR NOX CONTROL
Volume I - Influence of Burner
Variables on NOX in Pulverized Coal Flames
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of pollution. This
work provides the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-76-061a
March 1976
BURNER CRITERIA FOR NOV CONTROL
A
VOLUME I. INFLUENCE OF BURNER VARIABLES
ON NOX IN PULVERIZED COAL FLAMES
by
M. P. Heap, T. M. Lowes,
R. Walmsley, H. Bartelds, and P. LeVaguerese
International Flame Research Foundation
IJmuiden, Holland
Contract No. 68-02-0202
ROAP No. 21ADG-040
Program Element No. 1AB014
EPA Project Officer: G. Blair Martin
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
FOREWORD
This report presents the experimental results and initial evaluations of an
investigation to develop burner design criteria to control the formation of
nitrogen oxides in large scale turbulent diffusion flames. The investigation
was carried out at the IJmuiden Research Station of the International Flame
Research Foundation.
We wish to express our gratitude to former colleagues at the Research
Station who did much of the groundwork for this investigation, particularly
Dr. W. LEUCKEL, Dr. N. FRICKER and Dr. K. HEIN. We would also like
to express our thanks to Professor J. BEER and members of the Programme
Executive for their helpful advice and to the staff members of the Research
Station for help in the preparation and execution of the experimental pro-
gramme. We are also indebted to the many Foundation members who gave
freely of their own experience.
Mr. G. B. MARTIN is the E.P.A. technical project officer whose invaluable
contribution is gratefully acknowledged. Finally, we wish to thank
Dr. E. BERKAU and Mr. D. W. PERSHING for the stimulating discussions.
11
-------
TABLE OF CONTENTS
Page No.
FOREWORD iii
SECTIONS
1. SUMMARY I
2 . PURPOSE AND SCOPE OF THE RESEARCH PROGRAM 7
3. BACKROUND 11
4. TEST EQUIPMENT AND OPERATIONS 45
5. MEASUREMENT SYSTEMS 53
6. RESULTS OF FURNACE INVESTIGATIONS 75
7. DISCUSSION OF RESULTS - NATURAL GAS FLAMES 95
8. DISCUSSION OF RESULTS - PULVERIZED FUEL FLAMES 139
9. CONCLUSIONS 163
iii
-------
LIST OF ILLUSTRATIONS
Figure
1.1 NO Emission Characteristics of the Various Flame Type 3
1.2 Sample of the Influence of Swirl and Injector Type on the
NO Emission Characteristics of Coal Flames 4
3.1 Movable Block-Type Swirl Generator 20
3.2 The Relationship Between Relative Swirl Index R and Swirl
Number S s 2 1
3.3 Simple Flame Classification Scheme 24
3.4 Region of Stable Combustion for Type II Flame (ref. 3.27) 27
3.5 Axial Temperature Distribution Showing the Characteristics
of Different Flame Types (ref. 3.27) 29
3.6 Flow Patterns Measured in Swirling Flames (ref. 3.27) 31
3.7 Radial Gas Concentration Measurements in High Intensity
Flames (3.27) 32
3 . 8 The Effect of Swirl on the Axial Temperature Distribution
of Anthracite Flames (ref. 3.32) 34
3.9 The Effect of Swirl on the Burnout Characteristics of
Anthracite Flames (ref. 3.32) 35
3. 10 The Effect of Swirl on the Radial Temperature Distribution
of Anthracite (ref. 3.32) 37
3.11 The Effect of Swirl on the Axial Temperature Distribution
and the Burnout Characteristics of Medium Volatile Coal
Flames (ref. 3.32) 39
3 . 12 Influence of the Volatiles on the Flow Pattern (Swirl No. 1 . 7) 40
4. la Experimental Furnace No. 1 as Used for the AP Trials 46
4.1b Schematic Arrangement of Cooling Pipes in Furnace No. 1 47
4.2 Double Concentric Burner System 50
5. 1 Suction Pyrometer with Replaceable Refractory Shields 54
5.2 5 -Hole Impact Probe 57
5.3 Water-Cooled Quartz-Lined Probe for Sampling from Gas
Flames 63
5.4 A Heavily Cooled Filter Probe for Use in P.P. Flames 66
5.5 Water Quench Probe for NO Measurement 67
5.6 A Steam-Cooled Quartz Probe for Use in Coal Flames 68
5.7 Comparison of Coal Sampling Probes in a Gas Flame 69
5.8 Solid Concentration Sampling Probe (Low Loadings) 70
5.9 Solid Sampling Probe (High Loadings) 71
5.10 Gaseous Species Sampling Probe (Clean Environment) 73
iv
-------
LIST OF ILLUSTRATIONS (Cont.)
Figure Page
6.1 The Relationship Between Relative Swirl Index R and
Swirl Number S s 76
6.2 Effect of Fuel Injector Type on NO Emission; Fuel Injector
at the Exit of a 25° Divergent 78
6.3 Effect of Fuel Injector Type on NO Emission; Fuel Injector
at the Exit of a 25° Divergent 79
6.4 The Effect of Preheat on the Emission of Nitric Oxide (Radial
Fuel Injector, 5% Excess Air) 81
6.5 The Influence of Excess Air on NO Emission Radial Injector
in the Throat of a 25° Divergent 82
6. 6 The Effect of Excess Air on NO Emission at Two Swirl Levels.
Radial Injector at the Exit of a 25° Divergent 83
6.7 The Effect of Fuel Injector Type and Swirl on NO Formation
in P.F. Flames 89
6.8 The Effect of Primary Air Percentage on NO Formation in
Coal Flames 90
6.9 The Effect of Primary Air Percentage on NO Formation in
Coal Flames 91
7.1 NO Emission Characteristics of the Various Flame Types 96
7.2 The Influence of Swirl on Axial NO Distribution of Jet Flames 97
7.3 Radial Temperature and NO Distribution for Non-Swirling
let Flames 101
7.4 Radial Temperature and NO Distributions for Slightly Swirling
(s=0.5) let Flame 102
7.5 Comparison of the Forward Mass Flow in the Early Regions
of Non-Swirling and Slightly Swirling Flames 104
7.6 Axial Temperature and NO Distribution for 3 MW Propane
Oxygen Nitrogen Flames 105
7.7 Radial Temperature and NO Distributions for Non-Swirling
Propane Flame 106
7.8 Radial Temperature and NO Distributions for Slightly Swirling
Propane Flames (s=0.5) 107
7.9 Radial Gas Concentrations for Slightly Swirling Flames (s^O.S) 108
7.10 The Influence of Swirl Level on NO Formation in High
Intensity Type I Flames 110
7.11 Diagramatic Representation of Conditions at the Base of a
High Intensity Type I Flame, 111
-------
LIST OF ILLUSTRATIONS (Cont.)
Figure Page
7.12 Emission Characteristics of High Intensity Type II Flames
Produced with a Divergent Fuel Injector 114
7.13 Emission Characteristics of High Intensity Type II Flames
Produced with a Radial Fuel Injector 11S
7.14a Comparison of Emission Curves Multihole Injector
(25 msec'1) 116
7.14b Comparison of Emission Curves Multihole Injector
(50msec-1) 116
7.15 Isothermal Tracer Concentration Measurements Comparing
Divergent and Radial Injectors 118
7.16 Emission Curves for Type II Flames Produced with a Radial
Injector at the Burner Exit 120
7.17 Mixing Pattern at the Base of Type II Flames Produced with
a Radial Fuel Injector at the Burner Exit 121
7.18 Primary Concentration and Position of Recirculation Zone
for a Radial Injector Placed at the Exit of the Burner 122
7.19 Effect of Heat Extraction at the Flame Base on NO Emission 124
7.20 Influence of the Heat Extraction on the NO Formation in
Type I and Type II Flames 125
7.21 Flue Gas Concentration of NO, NOX and NO2 as a Function
of Swirl Number for a High Intensity Type I Flame 127
7.22 Flue Gas Concentration of NO, NOX and NO2 as a Function
of Swirl Number for a High Intensity Type II Flame 128
7.23 Radial Distributions of Temperature O2, NO and NOi 10 cm
from the Burner Exit in a Type II Natural Gas Flame 129
7.24 Axial Distribution of NO, NO2 and NOX in Natural Gas Flames 131
8.1 Idealized Sketch of a Lifted Pulverized Coal Flame 140
8.2 Conditions at the Fuel let Boundary with an Injector
Stabilized Flame 143
8.3 The Effect of Primary Air Percentage on NO Emission 146
8.4 The Effect of Primary Air Percentage on NO Emission 146
8.5 The Effect on Nitric Oxide Emission of an Ignition Front
Surrounding the Coal let (Injector A, Throat of 35° Quarl,
20% Excess Air, 20% Primary Air) I48
8.6 The Influence of Primary Air Percentage with Injector in the
Throat of a 25° Divergent 150
vi
-------
LIST OF ILLUSTRATIONS (Cont.)
Figure Page
8.7 The Influence of Primary Air Percentage with Injectors
at the Exit of a 0° Divergent 150
8.8 The Influence of Primary Air Percentage with Injectors
at the Exit of a 25° Divergent 151
8.9 Effect of Injection Position (Injector H, Throat 176 mm
Diameter, 5% Excess Air, 300°C Preheat, 10% Excess Air 153
8.10 Effect of Burner Exit Geometry (Injector H, at Exit Throat
176 mm Diaemter, 5% Excess Air, 300° Preheat, 10%
Primary Air) 153
8.11 Axial Temperature and Species Distribution, Flame 123 154
8.12 Axial Temperature and Species Distribution, Flame 124 155
8.13 The Influence of Primary Velocity on NO Formation in PF
Flames (Injector in the Throat) 158
8.14 The Influence of Primary Velocity on NO Formation in PF
Flames (Injector at the Exit of the Divergent) 158
8.15 The Influence of Coal Nitrogen Content on NO Formation 160
8.16 The Influence of Coal Nitrogen Content on NO Formation- 161
vii
-------
LIST OF TABLES
Table
4.1 Gas Injectors 51
5.1 Analytical Equipment ^9
5.2a Test to Determine Suitability of Materials for Sampling Probes 62
5 . 2b Effect of Temperature 62
5.2c Gas Mixtures Used in Probe Material Tests 62
6. 1 Burner Parameters Investigated with Natural Gas Air Flames
(No Preheat, 5% Excess Air 166.13 kgs hr-1 Fuel Input) 77
6.2a Conditions Investigated during the Pulverized Coal Input/
Output Section of the AP-1 Trials 84
6.2b Coal Burner Characteristics 88
6.3 Input Conditions for Main Flame Measurements 92
7.1 Summary of Experimental Data 132
7.2 Multiple Linear Regression Correlations 132
7.3 Parameters Used in Multi-Regression Analysis 134
vill
-------
1. SUMMARY
Modern society's increasing demand for energy has created two problems
of general importance for combustion engineers: the depletion of the world's
reserves of fossil fuel, and the pollution of the atmosphere by combustion-
generated pollutants. These problems are interlinked and should not be con-
sidered separately. This report includes the results obtained during the first
phase of an investigation to specify burner design criteria to control NOV in
5C
natural gas and pulverized coal flames. Parametric investigations have been
carried out to determine the influence of burner parameters on NO formation in
natural gas and pulverized coal flames.
The influence of the following burner parameters has been investigated:
the method of fuel injection;
the throat velocity;
the geometry of the burner exit;
the position of the fuel injector;
the type of burner exit;
the proportion of primary air; and
the swirl intensity of the combustion air.
The two parameters found to have a major influence on NO formation were
the method of fuel injection and the degree of swirl. Nitric oxide formation
can be controlled by the optimization of burner design parameters because its
rate of formation is determined by the detailed mixing history of the fuel, the
combustion air and recirculating combustion products. The same parameters
also dictate flame characteristics.
The flames investigated can be broadly classified into three groups:
1) Lifted flames - the ignition front is stable in space with time but it
is established downstream from the fuel injector;
2) Simple jet flames - flames produced with axial fuel injection and
without a swirl-induced internal recirculation zone; and
3) High intensity swirling flames - in Type I flames the fuel jet
penetrates the internal recirculation vortex. In Type II flames
the fuel jet is diverted around the recirculation vortex which forms
on the flame axis. (See Section 3.2.1 for explanation of flame types.)
-------
Figure 1.1 shows a composite diagram of the emission characteristics of the
various types of natural gas flames. It can be seen that the emission char-
acteristics of the various flame types overlap, this has important practical
implications since it is possible to optimize the burner design to control pol-
lutant formation and yet provide flame characteristics which satisfy process
requirements. This was not observed with pulverized coal flames^ maximum
emissions were always obtained with high intensity Type II flames. However,
minimum emissions were provided by axial jet flames and the emission level
for these flames was comparable to emissions from natural gas flames with
similar input conditions.
Figure 1.2 gives an indication of the range of emission characteristics
which were observed with coal flames when two parameters, fuel injector and
swirl level were varied. The characteristics shown in Figure 1.2 have several
interesting features:
1) The step change in emission level with increasing swirl observed
with injector A. This was due to the establishment of an ignition
zone at the fuel injector.
2) The high emission level flames produced with the radial fuel
injector (F) and the almost lack of dependence on swirl level.
3) The low emissions measured with high velocity, single hole axial
fuel injection (H).
In order to explain these observations two basic assumptions have been
made. These are that the most significant factor of the total emission is fuel
NO and that the variation in the emission depends upon the fate of the volatile
nitrogen compounds. Fuel NO formation can be reduced by ensuring that the
volatile nitrogen compounds react under oxygen-deficient conditions. Maximum
emissions occur with radial fuel injection because the coal is rapidly mixed with
the total air supply and hot recirculating products. These conditions ensure
early stable ignition which is an important practical requirement. However, the
fuel/air mixing promotes NO formation.
Conversely, NO formation can be restricted by maintaining the fuel in a
coherent axial jet and discouraging primary/secondary mixing by surrounding the
-------
100
80
s
a
a.
50
40
Lifted
Flame
20.
Type II
1
Swirl Number (3)
Figure 1.1 NO Emission Characteristics Of The Various Flame
Types
-------
1000
800
600
S
a
a
400
200
5% Excess Air
10% Primary Air
Figure 1.2 Sample Of The Influence Of Swirl And Injector Type On The NO
Emission Characteristics Of Coal Flames
-------
fuel jet with an ignition front. The fuel must also be delivered with the minimum
amount of primary air.
Unfortunately these conditions produce flame characteristics which are
incompatible with existing boiler designs. Future work will concentrate
upon providing flame characteristics which satisfy process requirements and yet
have minimum emissions.
-------
2. PURPOSE AND SCOPE OF THE RESEARCH PROGRAM
Nitrogen oxides are produced during the combustion of all fossil fuels
using air as the oxidant and the recognition (2.1) of their significance as
atmospheric pollutants has resulted in public pressure to control their emission
into the atmosphere. It has been estimated (2.2) that stationary combustion
sources accounted for 50% of the 23 million tons of NO emitted in the United
X
States in 1972. The objective of this work is to provide criteria to enable
burners to be designed which will allow fossil fuels to be utilized efficiently
with the minimum pollutant emission.
Various control techniques are available to reduce NO emissions from
* j£
large steam raising plants (2.3, and 2.4). These techniques include:
operating modifications e.g., reduced load, excess air or preheat;
combustion modifications e.g., flue gas recirculation or staged com-
bustion;
burner redesign or modification.
All these techniques will necessitate variations in the accepted plant
operating conditions which may reduce the boiler efficiency. In the long term
burner design may well prove to be the most efficient method of controlling
pollutant emission from all forms of fossil fuel fired furnaces and boilers.
This report covers Phase I of a research program whose overall objective
is the specification of burner design criteria for minimum pollutant emissions from
p.f. and heavy oil fired furnaces. The program was divided originally into three
phases and the scope of the work for each phase is summarized below.
Phase I
1. Construct and evaluate a system capable of withdrawing and analyzing
gas smaples from natural gas and p.f. flames for nitrogen oxides.
2. Determine the effect of burner parameters on emissions from gas and p.f.
flames.
"7
-------
3. Make detailed temperature, concentration and velocity maps of several
gas and coal flames.
4. Inititate a program which will allow the amount of nitrogen oxides
emitted by natural gas and p.f. flames to be predicted.
Phase 'II
1. Investigate the effect of burner parameters on nitrogen oxide emissions
from fuel oil flame.
2. Investigate the effect of the following variables on nitric oxide emission
from fossil fuel flames:
- air preheat level;
- firing rate;
- nitrogen content of the fuel.
3. Continue the work begun in Phase I to develop a predictive model.
Phase III
1. Either design and construct a furnace to investigate the effect of rate ~
of heat loss on nitrogen oxide formation in flames or continue promising
lines of research indicated by the earlier work.
2. Continue the development of the predictive model.
This report presents the results of Phase I and attempts to explain the
apparently contradictory information obtained on the effect of burner parameters on
nitric oxide formation.
-------
REFERENCES
2.1 "Air Quality Criteria for Photochemical Oxidants", U.S. Dept. of
Health Education and Welfare Public Health Service, Natl. Air
Pollution Control Administration, Publ. AP 63, March 1970.
2.2 "Air Quality and Stationary Source Emission Control", a Report pre-
pared for the Committee on Public Works, U.S. Senate, March 1975 ,
2.3 BartOk, W. , Crawford, A.R., Cunningham, A.R. , Hall, H.J.,
Manny, E.H. and Skopp, A., "Systems Study of Nitrogen Oxide
Control Methods for Stationary Sources". Final report, Contract
No. PH-22-68-55, PB 192789, 1969.
2.4 Breen, B.P., "Emissions from Continuous Combustion Systems".
Ed. by W. Cornelius and W. G. Agnew, p. 325, Plenum Publishing
Corporation, New York, 1972,
-------
3. BACKROUND
3.1 Nitric Oxide Formation In Fossil Fuel Flames
The parametric studies described below demonstrate the strong influence
of burner variables on nitric oxide formation in natuial gas and pulverized
coal flames. The nitric oxide emitted from fossil fuel flames originates from
two sources:
the oxidation of molecular nitrogen, termed thermal NO;
the conversion of nitrogen compounds contained in the fuel, termed fuel
NO;
In order to understand the effect of burner variables on NO emissions it is
necessary to review the factors which control the formation of both thermal and
fuel NO in flames.
3.1.1 The Formation .of Thermal NO
It is generally agreed that the principal reactions involved in the oxidation of
molecular nitrogen flames are (3.1 -3.5):
N2 + O NO + N (1)
N * °2 NO * O (2)
N + OH NO + H (3)
and that NO formation rates in post flame gases can be successfully modelled
with these reactions assuming equilibration of the oxygen atoms. Lavoie et ai
(3.6) have suggested that in addition reactions (4) to (6) should also be considered
OH + N, N,O H- H (4)
% *•
O + NO NO + NO (5)
0 H- N2O N2 * 02 (6)
However, Bowman (3 .2) and Westenberg (3 .3) consider that in practical cases
reactions involving N?O do not play a significant role in the formation mechanism.
11
-------
It has been reported (3.2, 3.5 and 3.7) that nitric oxide formation rates
measured in the vicinity of flame fronts exceed the rates predicted by reaction
( 1) - ( 3 ). Fenimore (3.5) refers to this rapidly formed nitric oxide as "prompt"
NO and considers that its formation is due to the destruction of the N = N bond
by reactions other than those shown above. Since the formation of "prompt"
NO was restricted to hydrocarbon mixtures, Fenimore suggested that the formation
of HCN or CN by reactions such as
CH2 + N2 -CHN + (7)
C + N «-CN + CN ' (8)
L* £t
followed by oxidation of the CN could easily lead to nitric oxide formation.
Several workers (3.2 and 3.8) consider that the increased rate of formation
of nitric oxide in the vicinity of flame fronts is attributable to non equilibrium
radical concentrations. In shock tube studies Bowman and Seery (3 .4) were able
to predict measured rates of nitric oxide formation with reactions ( 1) - ( 3 )
provided the O and OH concentrations were correctly evaluated De Soete
(3 .8) also considers that the concentration of free radicals will exceed equilibrium
levels near flame fronts. De Soete found that the overall activation energy for the
formation of NO in combustion products was dependent on residence time and
varied from values as low as 87 kcal/mole for short times and increased to 135
kcal/mole as the residence time increased.
The controlling influences of time and temperature on the formation of nitric
oxide in combustion products are aptly demonstrated in the work of Breen et al
(3.10). The rate of thermal NO formation is a maximum for adiabatic,
stoichiometric combustion and rapidly decreases as the fuel air mixture ratio varies
from stoichiometric. Consequently, all the techniques which have been used to
control thermal NO formation (3.11 and 3 .12) attempt to reduce peak flame
temperatures and residence times in high temperature zone.
-------
3.1.2 The Formation of Fuel NO
It is universally accepted that the oxidation of nitrogen compounds in fossil
fuels contributes significantly to the total nitric oxide emission from flames.
However, the conversion of fuel nitrogen compounds to nitric oxide in flames
has not been studied as extensively as the formation of thermal NO. It is
known that residual fuel oils and coal contain bound nitrogen. The nitrogen
content of bitumenous coal varies in the range of 1.0% to 2.1%. Typical nitrogen
containing compounds which have been isolated from coal by solvent extraction
and temperature vacuum distillation include 2 and 4 methylpyridine, trimethylpyridine,
quinoline, 2 methylquinoline, aniline and toluidine (3.13).
At the time of writing almost all the available information concerning fuel
nitrogen conversion has been obtained with laboratory scale experiments using
additives and doped fuels. Only limited information is available from large scale
practical units. Bartok et al (3 .14) estimate that the presence of 1% N in the fuel
oil of a 175 M.W. front walled fired utility boiler increase NO emission from a
baseline 320 to 580 ppm. In a 320 M.W. tangential fired unit 1% nitrogen concen-
tration resulted in an increase in the flue gas nitric oxide content of 140 ppm
from a baseline of 150 ppm. Emission of 400 ppm NO have been reported of a
coal fired fluidized bed combustor using argon oxygen mixtures instead of air
for combustion (3 .15). Essentially all the nitric oxide emitted was derived from
fuel nitrogen, bed temperatures were of the order of 800 C and it was found that
emissions increased as the oxygen content of the "argon air" was increased.
Shaw and Thomas (3 .16) have reported the conversion of gas phase nitrogen
compounds including pyridine, to NO in low temperature premixed rich carbon
monoxide flames. Nitric oxide was produced whether the oxygen and fuel
were mixed with either nitrogen or argon.
13
-------
Bartok, Engelman and del Valle have studied the effect of the addition of
various nitrogen compounds to a Longwell combustor (3.17). This combustor
allows examination of kinetically-limited combustion phenomena where trans-
port effects are minimized. The compounds investigated were NO, NO2, NH3,
CH.J-NH2 and C^N,, and results can be summarized as:
conversion to or retention of NO decreases as the combustion
mixture becomes fuel-rich;
complete conversion of NO and NO2 was noted with fuel-lean
mixtures;
the percent conversion of NH- to NO decreased as the NH^ con-
centration increased; and
complete conversion of cyanogen was never achieved.
One of the most significant results reported by Bartok et al was that nitric
oxide could be reduced to molecular nitrogen (indicated by absence of NO in
combustion products) in fuel-rich combustion conditions. Also a recent patent
of the John Zinc Company (3.18) is concerned with the homogeneous conver-
sion of NO to N, by combustion in a fuel-rich environment.
Martin and Berkau (3.19) and Turner et al (3.20) have investigated the
conversion of fuel nitrogen compounds in combustion systems using doped
distillate fuels. However, both workers used different techniques to separate
the thermal NO from the fuel nitrogen NO . Martin assumes that for any given
X X
condition the thermal NOV is the emission from the undoped fuel. Turner et al
•A,
postulated that flue gas recirculation only reduced thermal NO and that the
elimination of all thermal NO formation could be achieved by 30 percent
recirculation. This postulate was based on the fact that 30 percent recircula-
tion reduced the emission of NO from a No. 2 oil (0.03 percent N«) to 40 ppm
with both high and low atomizing air flow rates. The emission of 40 ppm was
exactly that amount which would be expected from total conversion of the fuel
nitrogen. With doped fuels 30 percent recirculation also reduced emissions
by the same amount.
The assumptions used by both Martin and Turner to allow them to determine
the amount of fuel NO produced with doped fuels are open to criticism. The nitro-
gen additive may influence thermal NO formation by altering the surface tension ortte
14
-------
viscosity of the fuel. Flue gas recirculation of 30% does not entirely eliminate
NO formation from gas fired systems and the basis for Turner's assumptions may
be due to some peculiarity of his particular experimental system. A major
criticism of experiments with doped fuels is associated with the type of nitrogen
bonds within the fuel. However these experiments have given important
indications of the behaviour of nitrogen compounds in real fuels.
Both Martin and Turner found that the fraction of fuel nitrogen converted to
nitric oxide increases with increasing excess air and decreases with fuel
nitrogen concentration. Martin found that quinoline (C_H N) gave lower
conversions than piperdine (C,.H N) and pyridine (CCH N). The latter two
D JL JL 5 w
gave identical results. Turner found little influence of the boiling point of the
nitrogen additive with twenty different compounds in a No. 2 oil. Only when the
b.p. was less than 100 C did the conversion to fuel NO decrease considerably.
The work of Martin and Turner produced similar general conclusions but there
are detailed differences in their results. Turner found greater conversions than
Martin suggesting that the formation of fuel NO is a strong function of the burner/
combustion chamber system. This conclusion is supported by Shell data reported
by Sternling and Wendt {3 .21). Emissions of 543 ppm NO and 300 ppm NO were
X X
measured from two different burners firing the same oil into two similar heaters.
These workers also report that inefficient atomization systems tend to give lower
nitrogen fuel at the same excess air level.
Ammonia is the most simple fuel molecule containing chemically bound nitrogen
and it could be formed by pyrolysis of a more complicated molecule. Approximately
16% of the nitrogen in coal is converted to ammonia during conventional
carbonization processes. Detailed species concentration measurements in low
pressure ammonia flames (3.22) have shown that the nitric oxide concentration
increases rapidly in the flame zone to values significantly greater than the
equilibrium values. The NO concentration then decreases in the post flame zone.
IS
-------
The rate of disappearance of NO is relatively slow in lean flames but can be
quite rapid in rich flames. Fenimore and Tones (3.22) proposed that NO was
reduced by NH0:
£t
NO + NH2 •+ N2 + ... 0)
whereas Drummond and Hiscock (see Ref. 3.21) account for the reduction by the
reaction:
NO + NH -» N- + OH (10)
£*
Bowman (3 .2) has recently proposed a partial equilibrium mechanism to account
for nitric oxide formation in shock-induced combustion of lean hydrogen, oxygen
and ammonia mixtures. Bowman considers that nitric oxide is produced by the
rapid reaction of O or OH with the fuel nitrogen intermediate. Subsequent reaction
of NO will then occur through relatively slow reactions such as:
O + NO -> N + O (11)
£*
NO + NO -» N2O + O (12)
NO + RN -» . ...eventually N- (13)
Consequently the non-equilibrium NO concentrations which are observed in
the combustion zone are the consequence of non-equilibrium radical concentrations
in that zone.
Sternling and Wendt (3.21) have attempted to develop a kinetic model which
will take account of the experimental results obtained with doped fuels. They
consider that the model should explain the following:
the fractional conversion of fuel nitrogen to NO decreases as the quantity
of fuel nitrogen increases;
the conversion of fuel nitrogen to NO increases with increasing excess
air;
2*
under rich flame condition NO can be reduced to N,
t.
Sternling and Wendt consider that the Zeldovich mechanism alone is not
adequate to explain the reduction of NO under fuel rich conditions. The nitrogen
16
-------
bond is formed from the reaction
N + NO - N + O (14)
which is faster under fuel rich conditions than
N + Q2 ~ NO + O (15)
The nitrogen atoms are built up by reactions involving HNO, NH etc.
Fenimore (3*23) has recently reported experiments in which he added nitrogen
compounds to premixed ethylene flames. Fenimore considers that all fuel nitrogen
compounds are converted to an intermediate compound I.which subsequently
reacts to produce either NO or N~:
I + R - NO + .... (16)
I + NO N2 + (17)
Fenimore suggests that I could be either NH, or N.
Experiments both with doped fuels and simple nitrogen compounds suggest that
the formation of fuel NO is strongly dependent on the combustion conditions under
which the nitrogen compounds react. All the studies indicate that the formation
of fuel NO can be reduced by restricting the available oxygen during the
combustion of the fuel nitrogen compounds.
17
-------
3.2 The Characteristics of Swirling Flames
Ignition stability is of paramount importance for all industrial burner
systems'. Two mechanisms of flame stabilization can be distinguished:
stability is achieved because the tendency for the flame to propa-
gate is in equilibrium with the flow of reactants, e.g., a simple
premixed flame; and
stability is achieved by the continuous ignition of a combustible
mixture by mixing with hot gases which have been produced as a
result of an earlier ignition, e.g., a bluff body stabilized flame.
All the flames discussed in this report were stabilized by the latter mechanism.
A combustible mixture is mixed with hot recirculating gases, combustion is
initiated and a flame propagates throughout the mixture. This flame may then
act as an ignition source enabling the combustion of the fuel to take place and
to maintain a stable ignition zone at the burner. Should the feedback of heat
by the recirculating gases be insufficient to maintain a stable ignition source
the flame will move downstream from the burner.
From the above description it can be seen that the presence of a recircula-
tion zone is necessary to ensure adequate flame stability. The size of the
recirculation zone and the temperature and velocity of the recirculating gases
must be such that ignition is assured to maintain the recirculation zone in a
condition necessary to assure subsequent ignition. Two types of recirculation
zones are found in combustors:
external recirculation zones; recirculation takes place between the
boundary of the flame and the combustor walls. Although matter is
entrained close to the burner, it originates further downstream; and
internal recirculation; recirculation takes place within the flame
boundaries and can be induced in the wake of a bluff body, or by
imposing a strong axial gradient of static pressure at the center of
the jet close to the burner. This static pressure gradient can be
obtained by imparting a swirling tangential velocity component
(swirl) to the jet fluid.
18
-------
Swirl Induced internal recirculation was utilized to ensure ignition stability for
the majority of flames discussed in this report. The burner systems used during
the investigation can be described as a double concentric jet burner with axial
fuel injection and an annular variable swirling air flow.
The major effects of swirl are:
reduced flame length;
increased entrainment rates;
improved stability.
The intraction of the axial fuel jet and the swirl induced torroidal vortex
produce characteristic flame types. The swirl level necessary to produce a
particular flame type depends very strongly upon the method of fuel injection.
The condition necessary to produce these flame types and their properties will
be described later.
The parameter most commonly used to characterize swirling flows is the swirl
number S, defined by reference 3.24 as:
G
s- -^r us)
X
where:
G is the axial flux of angular momentum;
o
G is the axial flux of linear momentum;
J\
R is the burner throat radius.
Leuckel (3*25) has compared, measured and calculated swirl parameters for the
moving block swirl burner and these results are reproduced in Fig. 3.1. Figure
3.2 shows the relationship between block setting, R and swirl number for
S
combinations of primary injector and throat diameters.
19
-------
c
CD
§
a
fi
o
o
moving
blocks
Experimental
Values
n 1300 kg. alr/hr
o 2000 kg alr/hr
A 3000 kg air/hr
m
Figure 3,1 Movable-Block Type Swirl Generator
-------
2.0
17.6 cm
13B6 cm
17,6 cm
13,1 cm
6 cm
5 cm
11.5 cm
11.5 cm
Figure 3.2 The Relationship Between Relative Swirl Index R
And Swirl Number S s
21
-------
3.2.1 Classification of Swirling Flames
Swirl increases the mixing rate between the fuel and air close to the burner
and at the boundaries of the recirculation zones (3.24 and 3.26). Consequently,
flame length is reduced and combustion intensity (in terms of flame volume) is
increased. The flame types described in this report can be conveniently
classified in terms of combustion intensity. The flame sketches shown in Fig.
3.3 illustrate this simple classification scheme.
3.2.1.1 Low Intensity Jet Flames (Fig. 3.3a)
The swirl level is zero/ or below that necessary to form an internal
recirculation vortex, i.e. S=Q,6. The flame could be either stabilized by a
flame retention device (e.g. pilot flame) or by the recirculation zone formed at
the interface between the primary and secondary streams. These flames are long
and are characterised by low mean temperatures and slow increases in axial
temperature. Increasing swirl from zero to S = 0,6 shortens the flame and increases
the gradient of axial temperature rise.
3.2.1.2 High Intensity Type I Flames (Fig. 3 .3b)
Increasing the swirl intensity beyond the critical forms an internal recirculation
vortex close to the burner. If the momentum of the fuel jet is sufficient, then
the internal recirculation zone is split by the fuel jet.
The resulting flame is divided into two sections:
a short bulbous zone close to the burner;
a longer tail section.
The two sections are connected by a short neck although under certain circumstant
this connection may consist of only a few fillaments of flame which may be
severed leaving only the bulbous front section. Alternately the two sections may
merge and be almost indistinguishable. These flames are rarely encountered in
practical systems.
22
-------
3.2.1.3 High Intensity Type II Flames (Fig. 3.3c)
This type of flame is characterized by an internal recirculation zone on the
flame axis and can be produced by:
radial or wide angle divergent injection at low or zero swirl.
At zero swirl the recirculation is due to the blockage caused by
the fuel injector;
divergent, annular or low momentum axial fuel injection at
medium swirl levels;
high momentum axial fuel injection at high swirl numbers.
Type II flames have a higher intensity than Type I flames and they can be
produced at low swirl levels with the appropriate fuel injection system.
Consequently, this is the type of flame is generally used in large steam raising
boilers.
3.2.1.4 High Intensity Type III Flames (3.3d)
In addition to the two distinct flame forms referred to as Type I and Type II
an intermediate form can occur. This flame has the bulbous base, characteristic
of Type I flames but the tail is absent. This type of flame is associated with
low annular air velocities and axial or annular fuel injectors. With high air
velocities the change from Type I to Type III flames with increasing swirl is
abrupt. The axial fuel jet enters the internal flow zone, but fails to achieve
complete penetration and stagnates.
3.2,1.5 Lifted Flames (Fig. 3 .3e)
Lifted flames are those where the ignition front is stable some distance down-
stream from the fuel injector. Increasing the swirl intensity normally produces
a burner stabiliaed flame. The stability of lifted flames is enhanced by high
external recirculation temperatures.
23
-------
A.
Low
Intensity
Jet Flame
B.
High
Intensity
Type I
Flame
C.
High
Intensity
Type II
Flame
D.
High
Intensity
Type in
Flame
Figure 3.3 Simple Flame Classification Schema
-24
-------
3.2.2 The Effect of Fuel Type
The four different flame types illustrated in Fig. 3 .3 can be distinguished
with either gaseous, liquid or solid fuels. This report is concerned with two
fuels, natural gas and pulverized coal. In this section the effect of
fuel type on the combustion characteristics of swirling flames will be discussed.
3.2.2.1 The Characteristics of Swirling Natural Gas Flames
The characteristics of swirling natural gas flames produced with a double
concentric burner have been studied extensively by Leuckel and Fricker (3.27) and
the following discussion relies heavily upon their work.
Conditions Necessary to Form Type I or Type II Flames
Several workers have established the importance of the parameter:
primary jet velocity
G =
mean axial velocity of annular flow
in describing flow patterns in double concentric jets. The critical value of the
parameter, G , at which the f
c
function of several variables:
parameter, G , at which the flame form changes from Type I to Type II is a
C
Gc = f (S,VQ (i),«, L/D, ~, N,fg, f) (19)
where S = swirl number
V = distribution of tangential velocity
o
oc = angle of burner divergent
L = length of burner divergent
D = diameter of throat
x = position of fuel injector within the divergent
N = factor representing injector type
f = fuel density
g
f = a factor to take account of the density distribution within
the burner divergent and close to the exit
25
-------
The last factor is an attempt to include a dependency on the "progress of
combustion" . The flow field in the region close to the burner will be strongly
dependent on the static under-pressure developed by the swirling flow.
The distribution of static pressure is given by
v2
where V = tangential velocity at radius r
p = the density at radius r
Consequently, the static pressure field generated by a given distribution of
tangential velocity will vary with the density and therefore the temperature of the
swirling jet.
The relationship expressed in equation (19) contains the factor N to take an
account of the method of fuel injection. With axial fuel injection low values
of G promote the formation of Type II flames whereas high values of G allow the
formation of Type I flames . When G =* G the flame form fluctutates between the
o
two stable forms. With annular fuel injection G is higher than for axial injectors,
c
while with divergent gas injection it may not be possible to produce Type I flames.
The effect of moving the fuel injector towards. the exit of the burner divergent is
to reduce the critical value of GC . This can be explained by considering that the
movement of the fuel injector towards the exit allows the gas jet to overcome part
of the static pressure gradient without the associated loss in momentum of moving
from the burner throat to the exit. Experiments have shown that divergent angles
smaller than 15 encourage the formation of Type II flames (i.e. , G is high) .
o o ^
However, angles between 15 and 35 were not found to have an appreciable effect
on flame type.
The Stability Characteristics of Swirling Natural Gas Flames
One of the benefits associated with the use of swirling flames is the
greatly enhanced stability limits . Fig . 3 . 4 shows both fuel rich and fuel lean
blow-off limit as a function of burner load at three swirl levels, it can be seen
-------
Condition s
a = 25°
L/D = 1
Gas injector, annular (No. 21)
Gas velocity 35 m/s (186 kg A)
Gas injected at burner throat
300
Figure 3.4
1.0 1.5
Equivalence Ratio (fuel/air)
Region of Stable Combustion for Type II Flame
(ref. 3.27)
2.0
27
-------
that increased swirl widens the stability limits, in particular the rich limit. The
improvement in flame stability is probably associated with two factors:
improved fuel air mixing;
an increase in the mass recirculated products to the flame base.
With premixed flames firing in the open stability limits can be reduced at high
swirl levels. If the recirculation zone becomes too long, cold gases are returned
to the base of the flames (3.28). Syred and Beer (3.26) were able to improve the
stability limits of premixed flames by ensuring that short recirculation zones were
formed.
The blow-off limits measured with the moving block swirl burner are dependent
not only on swirl level, but also on the method of fuel inj ection and the burner
geometry. Fricker(3 .27)reports that:
for flame stability an optimum divergent angle of 25 was found for
both Type I and Type II flames. Angles less than 15° seriously
reduced the burner load;
for Type II flames there are little advantage in terms of flame
stability in increasing the length of the burner divergent. However,
when the fuel was injected from the throat the stability of Type I
flames was improved by increasing the length of the divergent exit;
with Type I flames blow-off could be attributed to excessive gas veloc-
ities . With Type II flame blow-off occurred if the annular air veloc-
ity was too high;
for weak swirl the size of the interface between the fuel and air
streams influenced the stability;
divergent multihole or annular fuel injection improved the stability
of Type II flames in comparison to those flames produced with low
velocity single hole injectors.
Detailed Flame Measurements
Measurements reported by Michelfelder(3.3Dillustrate the difference between
three of the flame types descirbed in Section 3.2.1. Figure 3.5 shows the axial
temperature distribution and flame length of four flames, two jet flames and two
high intensity flames. The results illustrate:
-------
1500
O
Q)
B
s
0)
Q.
6
1000
500
Q Jet Flame s=0
A Jet Flame. s=0.5
O Type I
O Type II
Flame Length
JetrPlame I Jet Flame
s=0.5 s=0
Figure 3.5
1.6
Axial Dist. L/D
Axial Temperature Distribution Showing The Characteristics Of Different Flame Types (ref. 3,27)
-------
the application of weak swirl increases the rate of axial temperature
rise in jet flames;
the high initial temperatures in Type II flames are associated with
the axial internal recirculation zone;
Type II flames are shorter than Type I flames although the dis-
tribution of heat flux distributions from both flames is similar.
The flow patterns presented in Fig. 3.6 indicate the difference between the
three high intensity swirling flames. This difference is also reflected in the gas
concentration measurements shown in Fig. 3.7. Although the fuel jet remains on
the axis with the Type III flame the methane concentration decays much more
rapidly than in the Type I flame. This can be contrasted with the rapid reduction
of methane concentration shown with the Type II flame.
3.2.2.2 The Characteristics of Swirling P.P.. Flames
The increased rate of entrainment and mixing rates caused by the application
of swirl to p.f. flames intensifies combustion and improves ignition stability.
Since the rate of decay of axial velocity is increased by the application of swirl
and coal burnout is time dependent, burnout distances are also reduced.
With a medium to high volatile coal and a single hole axial fuel injector at
zero swirl a visible ignition front is established some distance downstream from
the injector. This type of flame can be considered to be analogous to the lifted
gas flame. The ignition distance is dependent on such factors as the:
volatile content of the coal;
primary to secondary velocity rates;
degree of preheat;
amount of primary air;
temperature of the external recirculation;
furnace wall temperature.
As the swirl is applied to the combustion air the ignition front moves upstream
towards the burner. At some critical swirl level the ignition front jumps the
30
-------
0.20
0.10
0
0.10
0.20
0.10
0.20
0 10 20 30 40 50 cm
»•• ' i ' i -1 1 i
stagnation
zone
40
cm
Fig. 3.6a
Type I
Fig. 3.6b
Type II
Fig. 3.6c
Type HI
Figure 3.6 Flow Patterns Measured In Swirling Flames (ref. 3.27)
31
-------
-------
distance to the injector. At this point, depending on the injector type, the four
other flame types discussed in Section 3.1.1 may be formed:
with low velocity axial or annular injectors a type II flame can be
established initially;
with slightly higher velocities or with increased swirl either a Type
III or a Type I flame can be produced;
with high velocity axial injector the flame looks like a simple jet
flame even at high swirl levels when an annular reverse flow zone
is present. This is because very little of the high momentum primary
jet is entrained and returned to the base of the flame. Prior
to the establishment of a stable ignition front the primary jet may
become unstable. Under these conditions the coal jet is no longer
directed along the axis but its direction varies randomly about the
axis. In general the sequence of flame types with increasing swirl
can be seen from the diagram shown in Fig. 3.3. It can be
seen that with certain injectors continued increase in swirl after
the formation of a Type II can allow a Type III flame to be formed.
This phenomenon has not been reported with natural gas. It is believed
that it is due to changes in the static pressure field associated with
increased combustion intensity within the burner divergent.
The Influence of Burner Parameters on Ignition Stability and
Burnout of P.P. Flames
The rate of combustion of p.f. is dependent on the temperature and composition
of the gases surrounding the particles. Consequently, both ignition stability
and bumout are related to the mixing pattern of the fuel jet, the combustion
air and the recirculating gases. The heating rate of the coal particle will be
dependent on the incident radiation both from the p.f. flame itself and from the
furnace walls and upon convection from the surrounding gases. The burner para-
meters which influence the size of the recirculation zone and the mass of recirculating
gases will influence the combustion characteristics of p.f. flames.
If low volatile coals are used then the axial temperature distribution provides
a means of assessing ignition stability since a visible ignition front may not be
present. The optimum condition for ignition is considered to be a steep axial
temperature gradient. Figs. 3.8 and 3.9 indicate the influence of swirl on the
axial temperature distribution and of anthracite burnout of flames.
33
-------
1250,,,
1000
750
O
o
2
GJ
a
S
500
250
Swirl Number
0 = O
0.6-- = A
Axial Distance
Figure 3.8 The Effect Of Swirl On The Axial Temperature
Distribution Of Anthracite Flames (ref. 3.32)
-34
-------
c
o
"§
3 .3
,0
1234 56
Axial Distance (m)
Figure 3'. 9 The'Effect Of'SwirUOn The" Burnout Characteristics Of -Anthracite ~Pla'ih«sc(ref. 3.32).
-------
It can be seen that there exists an optimum degree of swirl both for
ignition stability and burnout. It can be seen from the radial temperature pro-
files (Fig. 3.10) that the critical condition is not a Type II flame. Although the
base of the flame shows the characteristics Type I pattern the critical flame
is more nearly classified as a Type III flame. The coal jet is not swept backwards
by the recirculation zone but stagnates and then spreads some distance from the
injection point. After the critical conditions have been attained any further
increase in swirl causes a deterioration in the combustion characteristics.
Hein(3.29)has shown that the optimum swirl level for early stable ignition of
anthracite flames increases with increases in primary jet momentum and reduced
secondary air velocity. The optimum swirl level is reduced if primary swirl is
used, the burner exit is divergent rather than parallel and the length of the burner
divergent is increased. It was possible to achieve visible ignition if 12% of the
stoichiometric air requirement was used for the primary supply but not if 24% was
used. Although visible igntion meant a steeper axial temperature rise it was not
accompanied by reduced burnout distance. The fuel burnout is associated with
residence time and this is controlled by the jet momentum and the swirl level.
Beer and Lee (3.30)have shown that the burnout of anthracite in swirling
flames can be predicted by considering the furnace to be a combination of two
adjacent reactors, one well-stirred and the other a plug flow reactor. By a
combination of furnace measurements and isothermal experiments these workers
were able to show that swirl influenced the mean residence time in the well
stirred section.
Increasing the volatile content of the coal means that the ignition front
will be visible and stable ignition can be considered to be achieved when this
visible ignition front is stable at the injector. The visible ignition fron cor-
responds to the combustion of the volatile coal fractions which are released
from the coal particles when they are heated. Thus, the conditions for stable
ignition are:
rapid heating of the coal particles to ensure an adequate quantity
of combustible gas;
36.
-------
1000
800
Q
o
2
o
Q.
O
c-i
600
400
200
0.40
0.80
Radial Position (m)
1.20
1.60
Figure 3.10
The Effect Of Swirl On The Radial Temperature Distribution
Of Anthracite (ref. 3.32)
37
-------
the volatile gases must be mixed with air
a stable ignition front must be present.
The satisfaction of these conditions necessitates rapid mixing both with air and
with hot recirculating products (i.e., internal re circulation). Consequently,
the burner parameters which control the size and strength of the recirculation zone
will influence ignition stability. Thus divergent burner exits reduce the optimum
swirl level from that necessary to ensure ignition stability with parallel exits.
As with natural gas flames the method of fuel injection has a profound effect on
stability. In %his program it was possible to obtain stable ignition at zero swirl
by the use of a radial coal injector from which the coal was injected though a
series of holes around the primary pipe perpendicular to the burner axis.
As with low volatile coals burnout distances can be increased if the swirl
level is increased beyond some critical level. This is illustrated by the results
presented in Fig. 3.11.
The Influence of Coal Type on Flame Pattern
In the earlier discussions it was assumed that burner variables (swirl, geometry
etc.) dictated the flow pattern (flame type, size of recirculation zone) in swirling
flames. However, the temperature gradients associated with swirling flows also
influence the flow pattern. Fricker and Leuckel(3.27)compared swirling isothermal
flow with swirling natural gas flames and found:
flow patterns existed in isothermal flow which were not found in
swirling flames;
phenomena (i.e., transformation from Type I to Type II) happended in
isothermal flows at lower swirl numbers than in combusting flows.
The effect of combustion on the characteristics of swirling flows is also evident
from the results presented in Fig 3.12 (ref. 3.32). These figures compare the flow
patterns for identical burner conditions when burning two coals of different volatile
contents. It can be seen that the flow pattern produced at medium swirl is Type
I for the medium volatile coal and Type II for the low volatile fuel. At high swirl
the internal recirculation zone is reduced in size when firing the medium volatile
coal. Although me asurements of static pressure were not made, it is believed
38
-------
CO
1500
O
u)
e
a.
Q.
e
-------
Volatiles 0.06 kg/kg daf
Volatlles 0.21 kg/kg daf
OJ60
U)
o
c
it)
0.50
0.40
0.30
0.20
0.10
0.40 0.60 0.80 1.00
1.20 1,40
Radial Distance (m)
0,60 0.80 1.00 1.20 1.40
-------
that the static pressure field is influenced by the higher temperatures and steep
temperature gradients produced by the combusting volatile coal fractions.
41
-------
REFERENCES
3.1 Livesey, T.B., Roberts, A.L. and Williams, A., Combustion Science
Technique, ± 9 (1971).
3.2 Bowman, C.T., Paper presented at Fourteenth Symp. (International) on
Combustion, Penn State, August 1972.
3.3 Westenberg, A.A., Combustion Science and Technique, ±, 59, 1971.
3.4 Bowman, C.T., Seery, D.J., "Emissions from Continuous Combustion
Systems," Ed W. Cornelius and W.G. Agnew'Plenium Publishing
Corporation, 1972.
3.5 Fenimore, C.P., Thirteenth Symp. (International) on Combustion, p. 373,
The Combustion Institute, 1972.
3.6 Lavoie, G.A., Heywood, J.B. and Keck, J.C., Combustion Science
Technique, 1_, 313, 1970.
3.7 Bowman, C.T., Combustion Science Technique, 3., 37, 1971.
3.8 De Soete, G.G., Paper presented at "American Flame Days" Chicago,
September, 1972.
3.10 Breen, B.P. et al, Thirteenth Symp. (International) on Combustion,
p. 391, The Combustion Institute, 1971.
3.11 Bartok, W., Crawford, A.F. and Piegari, GJ., "Systematic Field Study
of NO Emission Control Methods for Utility Boilers." Contract No.
CPA 7u"-90 Esso Research and Engineering Company, 1971.
3.12 Blokeslee, C.E. and Burback, H.E., "Controlling NO Emissions from
Steam Generators." Combustion Engineering Preprint.
3.13 Chemistry of Coal Utilization, Ed. Lowrey, John Wiley and Sons, New
York, 1945.
3.14 Bartok, W. et al, "Systems Study of Nitrogen Oxides Control Methods
for Stationary Sources." Esso Research and Engineering Co
Final Report ER-2-NO s 69, 1969.
42
-------
3.15 Jonke, A.A., "Reduction of Atmospheric Pollution by the Application
of Fluidized Bed Combustion." Argonne National Laboratory,
Monthly Progress Report 8, 1969.
3.16 Shaw, J.T. and Thomas, A.C., "Oxides of Nitrogen in Relation to
the Combustion of Coal." Paper presented to the 7th International
Conference on Coal Science, Prague, 1968.
3.17 Bartok, W. et al, "Laboratory Studies and Mathematical Modelling
of NO Formation in Combustion Processes." Esso Research and
Development Co. EPA No. CPA 70-90, 1972.
3.18 Patent John Zinc Company, Pat. No. 2040117.
3.19 Martin, G.B. and Berkau, E.E., An Investigation of the Conversion of
Various Fuel Nitrogen Compounds to Nitrogen Oxides in Oil
Combustor." Paper presented to the A.I.Ch.E. National Meeting,
Atlantic City, 1971.
3.20 Turner, D.W., Andrews, R.L. and Sigmund, C.W./'Influence of
Combustion Modifications and Fuel Nitrogen Content on NO Emissions
from Fuel Oil Combustion." Paper presented to the Annual A.I.Ch.E.
Meeting, San Francisco, 1971.
3.21 Sternling, C.V. and Wendt, J.O.L., "Kinetic Mechanisms Governing the
Fate of Chemically Bound Sulfur and Nitrogen in Combustion."
Shell Development Company, Emeryville, California EHS-071-45,
1972.
3.22 Fenimore, C.P., and Jones, G.W., J. of Phys. Chem. .65 298, 1961.
3.23 Fenimore, C.P., Combustion and Flame 19_ 289, 1972.
43
-------
3.24 Beer, J.M. and Chigier, N.A.
"Combustion Aerodynamics", Applied Science Publishers, Ltd., 1972
3.25 Leuckel, W.
I.F.R.F. Doc. Nr. G02/a/16
3.26 Syred, N., Chigier, N.A. and Beer, J.M.
"Flame Stabilization in Recirculation Zones of Jets with Swirl",
13 Symposium (Int.) on Combustion, The Combustion Institute,
p. 563, 1971
3.27 Fricker, N. and Leuckel, W.
I.F.R.F. Doc.nr. F 35/a/4
3.28 Bafuwa, G. G. and MacCallum, N.R.L.
Combustion Institute European Symposium, p. 565
Academic Press, 1973
3.29 Hein, K.
Paper presented to the I.F.R.F., 1st Members Conference, May 1969
3.30 Beer, J. M. and Lee, K.B.
10th Symposium on Combustion, 1. 1187
The Combustion Institute, p. 165
3.31 Michelfelder, S. and Lowes, T.M.
I.F.R.F. Doc.nr. F 36/a/4
3,32 Van Heyden, L., Heap, M.P. and Fricker, N.
Gas Warme International Dec. 1971
44
-------
4- TEST EQUIPMENT AND OPERATIONS
The experimental results presented in this report were obtained in two
separate trial series:
M-2 trials (intermittent measurements during Oct. Nov. Dec. 1971),
evaluation of measuring equipment for gas flames and an investigation
of the effect of furnace cooling load on the emission of nitrogen
oxides from gas flames;
AP-1 trials (continuous furnace operation from March 17th to May 16th, 1972),
an investigation of the effect of burner parameters on nitrogen oxide
emissions from natural gas and p.f. flames.
The AP-1 trials can be subdivided into four separate sections:
natural gas I/O;
p.f. I/O with low nitrogen coal;
flame mapping;
p.f. I/O with high nitrogen coal.
The furnace was operated continuously during the trial period. The
measurements necessary to enable complete velocity, temperature and concentration
maps of the flame to be drawn were made by four, two man teams working 3 shifts
per 24 hours. No fixed shift system was used for the I/O investigations; the
measuring times were dictated by the necessity for allowing furnace conditions
to equilibrate after changing burner parameters. Details of the furnace, input
conditions and burner design used during the investigations are presented in
this section.
4.1 Furnace Conditions and Inputs
The furnace investigations were carried out in the Ijmuiden No. 1 furnace
a plan of which is presented in Fig. 4. la. The furnace is a horizontal,
refractory tunnel of approximately square cross section, with internal dimensions
2m x 2m x 6.25 m. In the M-2 trials the furnace was cooled by a symetric
arrangement of seventeen cooling loops (illustrated in Fig. 4.1b). However,
for the AP-1 trials, the cooling system was removed and the furnace was
operated virtually uncooled. In order to prevent the furnace stack from overheating
45
-------
cn
'/////ty///t.
Dimensions In mm
Figure 4.la Experimental Furnace No. 1 As Used For The AP Trials
-------
185 mms
185 tnms
92 mms
6 loops
8 9 10 11 12 13 14 15 16 17
11 loops
Figure 4.1b Schematic Arrangement of Cooling Pipes In Furnace No. 1
-------
16 cooling pipes were placed in the final 2 m of the furnace length; the pipes
were evenly divided between the furnace roof and hearth. Access to the furnace
for both measurement and observation was provided by the horizontal and
vertical slots (see Fig. 4.1). The horizontal slot enabled continuous probe
movement in an axial direction for a distance of 2m from the furnace wall.
In leakage of air was prevented by operating the furnace at a slight overpressure
and using close fitting doors for the measuring slits.
The Slochteren natural gas used during the investigation had the following
nominal composition:
CH4 - 81.3% v/v
C0HC - 2.9% v/v
t, o
C.HQ - 0.4% v/v
3 o
C4H10~ °'1% V/V
C H - 0.1% v/v
n m
CO - 0.8% v/v
&
N2 - 14.4% v/v
The firing rates used during the investigation were:
AP-1 200 NrnVhr (heat input of 1.76 Mw)
M-2 340 Nm /hr (heat input of 3.00 Mw)
Two coals of different nitrogen contents were used in the AP-1 trials. The
majority of the work was carried out with a low nitrogen coal from Lorraine, France
which had the following properties:
volatile content 32.7% (on a dry basis)
ash content 6.3% (on a dry basis)
carbon content 78.48% (on a dry basis)
hydrogen content 4.77% (on a dry basis)
nitrogen content 1.05% (on a dry basis)
sulphur content .75% (on a dry basis)
A firing rate of 204 kg/hr of coal was used which assumed a moisture content
of 2% to give an equivalent thermal input to 200 Nm3 of natural gas i.e., 1.76 Mw.
48
-------
Experiments were also carried out with a high volatile (35%) coal from North-umber-
land, England with a nitrogen content of 1.8%.
4.2 Burner Design
The burner used for the investigations is shown in Fig. 4.2 and was
positioned to fire along the length of the furnace from the center of the front
wall. The burner can be described as a double concentric jet burner with
variable swirl intensity in the annular air supply. The burner was not designed
to represent any particular practical burner but it is typical of all double concentric
systems. One of the features of the burner was its versatility; it readily allowed
the following parameters to be varied:
the type of fuel injector;
the position of the fuel injector relative to the burner exit. Although
the design allowed continuous axial movement, only the throat and
axial positions were investigated;
the velocity of the combustion air;
the angle of the burner exit;
the type of burner divergent (refractory or water cooled);
the degree of rotation (swirl) in the annular combustion air stream.
The swirl was produced by a moving block swirl generator which is shown
in Fig. 4 .2. The swirl register consists of two annular plates (P^ and PZ) and
two series of interlocking wedge shaped blocks (B. and Bj each attached to one
X C»
of the plates, interlocked the blocks form alternating radial and tangential
air flow channels, such that the air flow splits into an equal number of radial
and tangential streams which combine downstream from the actual swirl generator
into one swirling flow. Rotating the back plate ?2 (the plate P^^ is fixed to the
burner) progressively closes the radial channels and opens the tangential ones
(or vice versa). Thus the resulting flux of angular momentum increases (or
decreases) continuously between zero and a maximum value dependent upon the
dimensions of the system.
Four different types of fuel injector were used during the investigations
and they are classified with reference to the type of injection orifice:
49
-------
Cn
O
Block Adjustment
Mechanism
Fuel Gas Nozzle
(Throat Pos.)
Fixed Blocks
Movable Blocks
Figure 4.2 Double Concentric Burner System
-------
single hole;
multihole divergent;
multihole radial/-
annular.
The dimensions of the various injectors are given in Table 4.1.
Table 4.1 GAS INJECTORS
Nozzle number
1
8
19
32
33
Nozzle type
single hole
single hole
16 divergent
holes at 35°
16 radial
holes
16 radial
holes
Length of hole
mm
99.2
26.0
-
Diameter of hole
mm
46.1
13.0
4.7
9.65
4.7
Gas exit _,
Velocity m. sec
35
383
200
50
200
51/52
-------
5. MEASUREMENT SYSTEMS
\
The amount of nitrogen oxides emitted from fossil fuel fired furnaces depends
upon the detailed time-temperature-concentration history of the input fuel and air
and the products of combustion. During the AP-1 trials the following properties
were measured:
temperature;
velocity;
gas and solid concentrations;
nitrogen oxide concentrations.
In order to determine the spatial distribution of nitrogen oxide within
flames it was necessary to develop new sampling systems. The other properties
were determined by measuring systems developed by the I.F.R.F. and described
by Chedaille and Braud (5.1).
5.1 Temperature Measurement
Suction pyrometers were used to measure temperature since these give an
accurate direct reading of temperature. The temperature is measured by a
thermocouple positioned on the axis of a system of shields and the gas whose
temperature is to be measured is operated through the annular channels formed
by the shields (see fig. 5,1). The shields serve a dual purpose:
they isolate the thermocouple from the surrounding radiation;
they protect the thermocouple from contamination by the furnace matter.
The efficiency of the pyrometer increases as the suction rate increases
up to velocities of the order of 250 msec .
53
-------
Thermocouple [unction
en
>£>•
Refractory Shields
T./C. Wires
Fig. 5.1 Suction Pyrometer with Replaceable Refractory Shields
-------
During the measurement period the suction mass flow rate was
continuously monitored by an orifice plate and suction velocities greater than
200 msec" were maintained. The solids in p.f. flames tend to block the
shields and necessitate frequent shield changes even though the wake
sampling position (Figure 5.1) is used with p.f. flames.
The construction of the pyrometer almost eliminates many of the errors
associated with bare thermocouples. However, the measured value may be
in error because:
the sampled matter may contain solid particles whose temperature
is different from that of the gas and the measured temperature will
be neither gas nor the particle temperature;
within the "flame" the sample will contain fuel and oxygen which
are not mixed on the molecular scale and combustion will not take
place. It is possible that micromixing could be intensified at the
high suction velocities causing excessive after-combustion
within the shields.
the measured temperature is some time averaged value of a fluctua-
ting temperature. This temperature may be biased because of the
effect of viscosity changes on sample flow.
spatial resolution. The zone of high temperature may be small
compared to the sample volume.
5.2 Velocity Measurements
Detailed measurements were made in highly swirling flames during the
AP-2 trials where the three-dimensional nature of the flow requires both the
magnitude and direction of the flow to be known. It is not possible to use
standard yaw and pitch meters because rotation of the probe is inconvenient
under furnace conditions. Consequently, measurements of velocity magnitude
and direction were made with a hemispherical five-hole pitot probe. These
properties are determined from the measured pressure distribution around the
head of the probe when placed in the flowing stream.
55
-------
Lee and Ash (5.2) have demonstrated that with five holes on the circumference
of a sphere it is possible to calibrate an instrument to enable the magnitude
and direction of the flow in three dimensional field to be determined. Details
of the probes used in the AP-1 trials are shown in Fig. 5.2. Before and after
use the probes are calibrated in a wind tunnel. The calibration procedure
developed by Leuckel (5.3) allows the rapid evaluation of the desired properties
from measurements of five pressure differentials: PQ - Pg, PI - Pg, PQ - ?4/ ?2 - P.
and P - Patm.
o
The static pressures were recorded by mutual inductance transducers with a
sensitivity of .01 mm water column and electronically integrated. Measurements
were made with the probe pointing in the direction of mean axial flow. In the
boundary regions measurements were made with the probe directed both towards
and away from the burner and the required velocity profile was obtained from the
combined measurements. Decisions concerning the flow direction (either towards
or away from the burner) were based upon the measurements themselves and
checked by referring to measurements made by a two-hole cylindrical probe.
Errors involved in measurements with impact tubes have been adequately
discussed elsewhere (5.1). Measurements in pulverized fuel flames introduce
further complications due to the presence of solid particles. In gas and oil flames
solid concentrations rarely exceed a few milligrams per litre and mean diameter
is normally less than Ifi. In p.f. flames particle sizes range from Luto 200/*
and concentrations are of the order of several grams per litre. Consequently,
coal particles affect the velocity measurements by virtue of:
deposition upon the probe surface. Large scale deposition upon the
head will affect the calibration of the probe;
blockage of the pressure holes/ which will result in damping of the
fluctuations and a drift in the signal;
variation in fluid density. Both the static and total pressure heads
will depend upon the concentration and size of the particles. The
velocities were calculated assuming the fluid density to be given by
Pfluid = "gas + C
where C is the particle concentration under the measuring conditions.
56
-------
en
Connection for Universal
Probe
Fig. 5.2 5-Hole Impact Probe
-------
The problems associated with probe blockage and particle deposition cannot
be eliminated. However, it has been found that problems can be reduced
if the head is not completely cooled . Thus , the head is heated by radiation
and any deposits burn away. A disadvantage of this system is that the
combustion of particles either on the surface or in the pressure holes will
also produce errors. Pressure hole blockage can also be reduced by frequent
purging with compressed air.
Even when precautions are taken velocity measurements in high volatile
coal flames are extremely difficult. The particles are sticky and measurements
of the complete field are impossible using a five hole probe.
5.3 Species Concentration Measurements
The measurements of species concentration in flame gases involves
two separate operations:
the collection and delivery of a representive sample to the analytical
equipment;
the analyses of the sample.
During the AP-1 trials the concentration of following species were
determined:
- gases (N£, O2/ CO^ CO, HZ, CH4/ C^, C^, C^, CnHm/ NO,
N02);
solids.
The analytical equipment is listed in Table 5 . 1. Both the chromatograph
and chemiluminescent analyzer were constructed from component parts. The
chromatograph allows the concentration of specific hydrocarbons to be identified
and quantified. Automatic operation with a program sequence of three detectors
and five columns allows the analysis of the major components to be completed in thn
5.8"
-------
TABLE 5.1 ANALYTICAL EQUIPMENT
Class
Flue
gases
Plaae
gases
t
Solids
Species
°2
CO
co2
H2
C02
V VCS4
CO
Hydrocarbons
HO
1T02
Ash, fixed
carbon,
volatiles
ITitrogen
Analytical equipment
Paramagnetic
Infrared
Infrared
Gas Chrotaatograph
Carrier gas
rlitrogen
Helium
Helium
Helium
Column material
Molecular sieve
Silica gel
Molecular sieve
Alumina
Cheailuainescent analyser
Chesiluainescent analyser v/ith ^uar-ta'converter
Pozetto
Coleman
59
-------
minutes. The chemiluminescent analyzer was built to E.P.A. specification and
operated satisfactory in the NO mode. NO concentrations were determined by
•K,
diverting the sample through a heated packed quartz tube to convert nitrogen
dioxide to nitric oxide.
Stainless steel was not used because it is known that NO can be reduced
if passed through heated steel tubes in reducing mixtures (5.4). Although the
converter operated satisfactorily with completely combusted products NO
X
concentrations were less than the NO values when the sampled gases contained
H , CO and hydrocarbons. Consequently, NO- concentrations could not be
4rf U
determined in the flame gases.
The function of any sampling system is to collect a representative specimen
and ensure that the sample does not undergo further reaction after collection.
In order to satisfy the former requirement it is advisable to use a continuous
sampling system. In this way it is possible to take account of the concentration
fluctuactions which are inevitable large turbulent difussion flames.
The latter requirement is less easily satisfied when sampling gases containing
nitrogen oxides. The nature and concentration of the oxides of nitrogen necessitated
the development of a separate sampling system for nitrogen oxides from that used
for the other species.
5.3.1 Sampling Systems for Nitrogen Oxides
The sample which is withdrawn from the furnace is at a high temperature
and normally in a highly reactive state. Rapid cooling of the sample (to less
than 300 C in less than 3 x 10 seconds) effectively quenches the reactions
involving permanent gases, p.f. particles or soot (5.1). However, several
workers (5.4 and 5.5) have reported that under certain circumstances normal
sample probe materials, particularly stainless steel are not inert with respect
to nitrogen oxides. Although nitrogen oxide measurements have been made in
p.f. flames prior to this investigation (5.6), these workers did not indicate
60
-------
whether checks were made to ensure that reactions involving char particles
and nitrogen oxides did not occur within the sampling equipment. Consequently
experiments were carried out to establish design criteria for probes capable of
sampling nitrogen oxides from coal flames.
The most comprehensive test of any sampling system is the measurement
of known concentrations under representative operating conditions. Such a
test is difficult to carry out when the measurement concerns the NO content of
flame gases, whose time mean temperature may exceed 1700°C and contain coal
particles in a state of rapid decomposition. Therefore, it was decided to determine under
which conditions certain materials were no longer inert with respect to nitrogen
oxides by passing selected mixtures through heated tubes of the material to
be tested. The tubes were heated in an electric furnace whose temperature
could be controlled up to 1200 C. The nitric oxide content of the gas mixture
was determined by a chemiluminescent analyzer before and after passing through
the tube under test.
Three tube materials were tested:
stainless steel;
gold plated stainless steel;
quartz;
and the results confirmed the conclusions of Halstead (5.4) and Shaw (5.5).
Quartz is the only suitable material for sampling from reducing gases. A summary
of the test results are presented in Table 5.2a. Although a probe constructed
from stainless steel should be satisfactory in oxidizing conditions it was
decided that a probe constructed with a quartz sampling tube would be used in
natural gas flames. A diagram showing the construction of the probe is presented
in Fig. 5.3. The quartz tube was ground to ensure good contact with the water
cooled stainless steel liner. The gases are rapidly cooled to less than 2°C in
a gas cooler at the probe exit ot remove water vapour.
Sampling from p.f. flames involves the separation of the solid and gaseous
phases at some stage. This is normally accomplished by filtration close to the
sample point to prevent blockage of the sample lines. Further experiments were
61
-------
Table 5.2a TEST TO DETERMINE SUITABILITY OF MATERIALS FOR SAMPLING
PROBES
Material
Quartz
Quartz
Gold Plated s.s.
Gold plated s.s.
Stainless steel
Stainless steel
Gas Mixture
A
B
A
A
A
A
Temp. Range °C
20-1000°C
20-1000°C
10- 720°C
725°C
20- 590°C
595°C
Effect
None
None
None
4% loss of NO
None
60% loss of NO
Table 5.2b EFFECT OF TEMPERATURE
Temperature Range
70-170°C
20-240°C
20-840°C
170-250°C
500
620
1000
1100
1170
Mixture B
Dusted Tube
no effect
10% loss of NO
20% loss of NO
25% loss of NO
Mixture B,Coal
Dust Supported
By Quartz Wool
Plug
no effect
8% increase of NO
80% loss of NO
100% loss of NO
Mixture C5Coal
Dust Supported
By Quartz Wool
Plug
no effect
70% loss of NO
Table 5. 2c GAS MIXTURES USED IN PROBE MATERIAL TESTS
Constituent
H2
C02
v^/JTl .
A
CO
C2H6
C2H4
C3H8
N2
NO
Mixture A
36%
64%
80 ppm
Mixture B
8.5%
2 .9%
5 %
6 %
0.2%
0.3%
0.2%
66.9%
80 ppm
Mixture C
100%
121 ppm
^62
-------
or>
to
•quartz liner
Sample Line
Water Injet
XX \\XX\XXXX
X X X XX XXX
Water Outlet
Figure 5.3 Water Cooled Quartz Lined Probe For Sampling From Gas Flames
-------
carried out to assess the effect of char particles on the sampled gases containing
nitrogen oxides. Char was supported on a quartz wool plug held in a quartz tube.
1. The effect of cold coal particles
During this series of tests the gas mixture leaving the heated tube was
passed through a "plug" of cold pulverized coal, supported on a quartz
wool filter. The tests described in Table 5.2a were repeated with the
same results.
2. The effect of heat char particles
The intert nature of quartz had already been demonstrated and these tests
were restricted to the effect of hot coal particles in quartz tubes. In
the first instance the coal particles were dusted into the tube surface
before it was placed in the furnace and in the second test the coal was
supported on a quartz wool plug. Gas mixture C was also passed through
a plug of coal. Results of these tests are shown in Table 5.2b.
The results have important implications for the construction of probes
suitable for sampling from p.f. flames since they show that contact between
the gas and hot char particles cannot be tolerated.
64
-------
Three probes were designed and built for testing in p.f. flames:
a filter probe (Fig. 5.4);
a water quench probe (Fig. 5.5);
a steam quench probe (Fig. 5.6);
The probes were designed to ensure that the following conditions were satisfied:
that any metal surface which was in contact with the sampled gases was
adequately cooled;
that the solid particles were rapidly quenched;
that the probe was simple to operate and capable of withstanding flame
conditions (high temperatures and solid loadings.)
The three probes were tested in both gas and p.f. flames and the steam quench
probe was found to be unsatisfactory; the probe was blocked before a representative
sample could be collected.
Both the water quench probe and the filter probe could be used in p.f. flames
and similar concentrations were determined with both probes. These two probes
were also compared with the quartz probes in"a gas flame.' The recorded concen-
trations are shown in Fig. 5.7. and it can be seen that there is no significant
difference between the values obtained with the different probe types.
The water quench probe had one considerable advantage over the filter
probe. It could be used continuously for long periods without blocking. The
filter probe was cooled so efficiently that condensation occurred in the filter
housing and the filter was rapidly blocked even with low solid loadings.
Therefore the water quench probe was used to sample from p.f. flames.
5.3.2, Sampling Systems for Gaseous and Solid Material
The flame conditions dictated the type of probe which was used to collect
flame gas and solid samples. Two types of probes with centered bronze filters
to collect solid matter were used in p.f. flames (see Figs. 5.8 and 5.9). The
two designs of filter probes were necessary because in regions of high solids
loadings (e.g., the initial stages of p.f. flames) the small filter blocks rapidly
and an accurate determination of the gas volume is not possible. The
65
-------
RADIATION
o SHIELD
en
COOLING
FINS
GOLD LINED TUBE
REPLACEABLE STAINLESS
STEEL FILTER
SAMPLE LINE
WATER INLET
Figure 5.4 A Heavily Cooled Filter Probe For Use In P.F. Flames
WATER OUTLET
-------
en
Quench Water Flow
Main Suction
Cooling Water
Fiq. 5.5 Water Quench Probe for NO Measurement
-------
Quartz Tube
Sample Line
( . '?''*..
\ I
en
CO
Steam Inlet
Figure 5,6 A Steam Cooled Quartz Probe For Use In Coal Flames
-------
Filter Probe Coal
Water Probe Coal
Quartz Probe Gas
s
a
Q.
o
75*
0
-. Time
Figure 5.7 Comparison Of Coal Sampling Probes In A Gas Flame
69
-------
Sealing Nut
Sample Holder
Connection for Universal
Probe
\ X *v \ X X X X X X \ \
\\X\\\\\\X\\ 1
Fig. 5.8 Solid Concentration Sampling Probe (Low Loadings)
70
-------
Fig. 5.97 Solid Sampling Probe ( High Loadings )
-------
filters were weighed dry before and after use thus enabling the concentration
of solids to be calculated if the total volume of gas sampled and the temperature
at the sampling point was known. The concentration of the gaseous species were
evaluated by gas chromatography from a dry clean sample.
The soot content of the natural gas flames measured during the AP-1 trials
was low and was neglected. Consequently, gas samples were taken with a
probe without a filter in the sampling head. Thus the water cooled sampling head
has considerably reduced dimensions (see Fig. 5.10).
The major problem associated with the sampling of solid matter is that the
sample should have the same particle concentration and size distribution as
present in the flame. In completely axial flow isokinetic sampling (suction velocity
is equal to the gas velocity at the sample point) techniques ensure that these
conditions are satisfied. However, in strongly swirling flows a "true sample" is
virtually impossible to obtain due to the inertial forces on the particles.
Consequently the solid concentrations reported in the appendix cannot be
considered as truly representative of the flame conditions.
72
-------
Stainless Steel
Cooling
Water Fllter
SfSS/ZS/S // / / / / S
Connection for
Universal Probe Holder
Fig. S.10 Gaseous Species Sampling Probe (Clean Environment)
-------
REFERENCES
5.1 Chedaille, J., and Braud, Y., Industrial Flames, Vol. 1- "Measurements
in Flames," Edward Arnold, 1972.
5.2 Lee, J.C.andAsh, J. E., Trans, of the ASME 78., p. 603, 1956.
5.3 Leuckel, W., I.F.R.F., Doc.Nr. F., 72/a/12.
5.4 Halstead, C.J., Nation, G.H., and Turner, L., "The Determination of
Nitric Oxide and Nitrogen Oxide in Flue Gases." Part 1: Sampling
and Colorimetric Determination, Shell International Gas Ltd.,
Report S 16,71/1, 1971.
5.5 Shaw, J.T., B.C.U.R.A. magazine, Vol. XXXIV, No. 10, October 1970.
5.6 Dumoutet, P., Gaget, J. and Nomine, M., "Formation des oxydes d'azote
au cours de la combustion du ch rbon." Pollution Atmospherique,
No. 52, Octobre, 1971.
74
-------
6. RESULTS OF FURNACE INVESTIGATIONS
A complete tabulation of the experimental results is presented in Reference
6.1. In the following sections it is intended that the results of the furnace
investigations will be summarized prior to discussing the influence of burner
parameters on nitric oxide formation in pulverized coal and natural gas flame.
The major variable involved in the parametric investigations was the
degree of rotational swirl in the annular air stream. Throughout the report two
methods are used to characterize the degree of swirl:
R - relative swirl which is 0 for axial flow and 1.0 at maximum swirl.
5
The intervals correspond to degrees of swirl block adjustment from
open to closed; and
S - the dimensionless swirl number (see Section 3).
The relationship between R and S can be seen in Fig .6.1. Relative swirl
S
number is used because in certain instances it allows a comparison to be
made more easily between several variables.
6.1 Parametric Investigations Involving Natural Gas
Table 6.1 lists the burner conditions that were investigated with natural
gas at a fixed thermal input, excess air rate and preheat level. Fig. 6.2 and
Fig. 6.3 give an indication of the range of emission characteristics obtained
by varying the fuel injector type, injector position and swirl level. The experi-
mental observations can be summarized as follows:
increasing the swirl intensity of the combustion air decreased nitric
oxide formation in flames produced with multiholed injectors.
maximum emissions were obtained with the radial hole injector when
it was placed at the exit of the burner divergent.
' - in almost all instances an increase in throat velocity caused a
decreased emission.
75
-------
2.0
17.6 cm
13.6 cm
17,6 cm
13,1 cm
6 cm
6 cm
11,5 cm
11,5 cm
Figure 6.1 The Relationship Between Relative Swirl Index R
And Swirl Number S s
76
-------
TABLE 6.1 BURNER PARAMETERS INVESTIGATED WITH NATURAL GAS AIR FLAMES
(No preheat, 5% excess air 166.13 kgs. hr"1 fuel Input)
'lane
no.
1
6
8
10
11
2
5
7
9
12
19
21
22.
24
20
23
25
26
27
28
30
31
•32
33 '
34
35
36
37
38
Semi diver-
gent quarl
angle
25°- -
25°
.
25° >
25°
35°
15°
15°
35°
Combustion
air
velocity
25 msec""1
25 msec.""1
.
"
50 msec~
50 msec""
50 msec""
4
-1
50 msec
-
50 msec"
25 msec""
Quarl
type
water
cooled
water
cooled
water
cooled
vater
cooled
vater
cooled
water
cooled
refrac-
tory
water
cooled
Fuel injector .
type
radial hole
single hole ( 1 )
single hole (2)
swirling1 gas
divergent
radial hole
single hole (l)
single hole (2)
swirling gas
divergent
divergent
single hole fl^
single hole (2)
radial
divergent
radial
radial
radial
single hole (l)
divergent
divergent
single hole (l)
radial
radial
divergent
single hole (l)
single hole (l)
radial
divergent
. Position
of
injector
exit
t
throat
•
throat
.
exit
*
exit
throat
throat
"
throat
throat
a single hole l 383
single hole (2) 35 mae<
77
-------
100
£
Q.
Q.
O
z
Injectors In Throat
O Radial hole
^ Single hole 383 m sec
p Single hole 35 m sec
A Swirling gas
Divergent
-1
-1
Figure 6.2 Effect Of Fuel Injector Type On NO Emission; Fuel Injector In
The Throat Of A 25° Divergent
78
-------
100
80
60
a
a
O
35
40
20
O Radial hole
O Single hole 383 m sec"*
Q Single hole 35m sec"
Swirling gas Injection
V Divergent
Figure 6 3 Effect Of Fuel Injector Type On NO Emission; Fuel Injector
At The Exit Of A 25° Divergent
79
-------
flames stabilized in refractory burner divergents produced more NO than
those stabilized in water cooled divergents.
As can be seen from Fig. 6.4 , it appears that burner parameters will also have
a strong influence on emission at high elevated preheat levels. Figure 6.5 shows
that the general shape of the emission characteristics of radial fuel injectors
are unaffected by the excess air level. However, as can be seen from the results
presented in Fig. 6.6 the emission level versus excess air peaks at different
excess air levels with different swirl intensities. It must be remembered that the
increase in excess air will also increase the burner throat velocity.
6.2. Parametric Investigations Involving Pulverized Fuel
Two additional variables were employed in the input/Output pulverized
coal measurements, the primary air supply and the coal type. The conditions
investigated are presented in Table 6.2 a/b. The influence of fuel injector type
and swirl can be seen in Fig. 6,7,.'. An attempt will be made to explain the reason
for some of the emission curves later in this report. Maximum emissions were
observed with radial injectors and minimum emission levels were obtained
with high velocity single hole injectors. One important burner variable which
was found to influence emissions from pulverized fuel flames was the primary
air supply. In some instances the amount of primary air changed the absolute
level of the emission curve but not the form of the curve (see Fig. 6.8 }.
However, this was not generally applicable (see Fig. 6.9 ). Under most burner .
conditions minimum emissions were observed with 20% primary air .supply.
6.3 Detailed Flame Measurements
Six flames were measured in detail during the furnace investigations and their
input conditions are listed in Table 6.3. Complete measurements details are
presented in Reference 6.1.
-------
200
a
a
i
Radial Fiiel Preheat
Injector Position
Figure 6.4 The Effect Of Preheat On The Emission Of Nitric Oxide (Radial
Fuel Injector,5% excess Air)
81 :
-------
O'.'IS
O 5% excess air
25% excess air
0.03
Figure 6.5 The Influence Of Excess Air On NO Emission Radial Injector In
The Throat Of A 25° Divergent
82
-------
0.15
H
CO
03
0.12
0.09
0.06
0.03
20
% Excess Air
Figure 6.5 The Effect Of Excess Air On NO Emission At
Two Swirl Levels. Radial Injector At The Exit
Of A 25° Divergent
83
-------
TABLE 6.2a CONDITIONS INVESTIGATED DURING THE PULVERIZED COAL INPUT/
OUTPUT SECTION OF THE AP-1 TRIALS
?Uco i
no.
66
^
85'
70
105
88
66
43
85
69
102
104
87
67
44
84
68
106
101
105
75
95
74
93
71
96
75
ioa
109
99
72
97
76
100
burner
i
A
M
R
it
u
IS
K
M
R
K
It
n
n
it
K
U
tt
It
tt
11
B
H
M
tt
tt
tt
H
W
It
R
it
. M
n
"
exo ess
air
5$
t!
tt
R
n
K
tt
R
t!
R
R
n
R
R
tt
tt
R
U
n
H
R
u
n
tt
R
R
R
H
n
n
i)
H
it
n
\
throc.t
176
1!
R
H
151
176
n .
R
II
R
151
•3
. 176
»
n
it
n
151
151
n
176
n
H
151
176
a
«
151
K
N
176
R
n
151
position.
throat
R
exit
H
H
throat
. R
R
exit
ir
R
R
throat
R
fl
ezit
n
throat
exit
R
throat
ezit
R
a
throat
ezit
H
throat
R
exit
throat
ezit
R
U
P
pria
10$
R
tt
H
R
20$
R
R
ft
H
R
R
50$
R
n
n
it
R
tt
IT
10$
R
It
it
20$
tt
R
R
»
R
3C$
R
H
It
Quarl
L
0
25
0
25
0
0
25
55
0
25
0
25
0
25
55
0 .
25
25
0
25
25
0
25 7
0
25
0
25
0
25
0
25
0
25
0
Coal jSsJ
1.1$ |
R j
U 1
H j
K 1
R I
" i
n 1
R I
" I
H I
R I
.R J
n j
n I
tt 1
R I
R I
R j
R |
It 1
R 1
n |
H I
« ' 1
n j
R 1
R I
R j
R j
R 1
R I
R |
"
84
-------
TABLE 6.2a (ContinueH]
N.ame
no.
53
45
94
57
46
93
47
79
80
91
89
90
73
77
82
81
60
92
59
138
134
142
126
146
130
s 150
151
132
153
!»!•.— I II
buiaer
C
it
R
n
n
it
ti
D
R
E
R
R
'Jl
**
G
R
H
n
R
A
w
R
R
R
R
3
it
n
n.
1 ,
eucesa j
air
53*
n
n
R
R
R
R
" " .
R
R
R
H
II
H
".
R
n
n
R
n
R
n
R
n . •
R
H
-
R
R
II
.———.———
throat
176
"
R
R
R
H
R
0 '
It
R I
R
."
H j
n
»
R
»
a
n
n
n
R
R
H
H
II
n
R
R
position
throat
H
exit
H
throat
exit
throat
throat
R
ezit
R
throat
R
throat
R
tbroaf
ezit
R
!
I throat
i
exit
i throat
ezit
throat
R
throat
exit
throat
ex.it
I
prifc
1056
R
II
H
20$
n
3<#
20%
3C#
10#
2C£
3056
10£
to*
2Qfi
3055
1056
R
.. »
1C*
. "
.20J6
R
3^
R
20#
R
50^
II
•MHMMiiW^MHM
O.uarl,
! L
25
35
o
25
35
0
35
25
R I
(
•
0
R I
n i
I
I
25
1
tt 1
i
25
R 1
25
0
2-
25
R
R
II
R
R
25
R
R
II
Coal %$2 1
1.1
R I
R I
H I
It I
II I
R I
H I
R I
If I
II I
"
R I
II I
" I
II J
B I
"» j
R j
1.8
"•
R
It I
•'" i
I M I
n
R I
I It 1
1 R I
JMMW^M^BMHMM^MMM
85
-------
TABLE 6.2a (Continued)
STlaae !
no « I
154
156
160
158
40
41
42
64
65
51
52
.43
49
50
54
55
56
53
127
128
129
131'
132
133
135
136
137
159
140
141
i
baraar
i
!«
7
M
A
n
n
n
»
P
I!
c
n
K
a
n
tt
n
1.
1!
li
If •
n
n
it
w
»
it
n
u
excess
air
*£
*i~
ti
n
n
20$
it
n
n
30$
20$
n
20$
n
n
15$ '
20$
30$
30$
10$
20$
T*~?
}3p
10$
20$
30$
10$
20$
30$
10$
20$
30$
throat
176 '
«
n
n
176
K
n
M
tt
»
n
tt
»
H
It
n
n
n
n
n
n
*n
it
n
• n
R
n
*r
n
n
B
position
threat-
it
n
n
throat
R
n
n
• w
n
it
n
n
n
it
n
n
exit
exit
u
n
»
»
n
it
it
N
throat
M
n
Pprim
20$
30$
10$
20$
30$
10$
20$
10$
10$
20$
10$
20$
10$
30$
10$
w
it
it-
20$
it
n
30$
n
n
10$
9
10$
n
n
it
0
Quarl
L
25
n
n
,,
35
tt
n
25
25
55
n
35
35
'35
25
tt
IT
It
25"
25
tt
„
js
IT
„
it
R
n
it
it
Coal %ll
1.8
n
ft
tt
1.1
«
ft
r?
if
1*1
H
H
n
n
it
n
n
n
1,8
n
«
n
n
„
it
»
it
tt
n
n
86
-------
TABLE 6.2 a (Continued)
Flame ! burner
no » |
143 .
144
145
147
143
149
155
157
159
A
"
n
n
it
B
3)
n
P
excess
air
ICjfc
20$
3<#
10£
2056
5C£.
20$ .
I!
It
throat
176
I!
IJ
II
«
*
IJ
t?
II
1
position
throat
n
n
M
«
if
K
n
w
P .
pria
20?4
e
n
30^
n
tt
O/^fly
fl» Wyw
!><$
20*
Quarl
L
25
n
M
n
n
t!
t»
it
"
Coal Jto2
1.8
n
n
i;
n
K
II
n
a
87
-------
TABLE 6.2b COAL BURNER CHARACTERISTICS
Burner
A
B
C
D
E
F
G
H
Type Of
Injection
Single hole
Single hole
Single hole
Annular
Single hole
Radial
, Single hole
Single hole
Dimensions
•
54.5 mm id
115 mm od
54.4 mm id
60 mm od
46 mm id
60 mm od
44 . 5 mm x
107 mm
115 mm od
70 mm id
115 mm od
16 holes
60 mm od
64 mm id
115 mm od
33 mm id
Mean Injection Velocity m/sec:
10%
Primary Air
19
19
26
—
11
27
14
52
20%
Primary Air
38
38
52
12
22
54
28
-
-30%
Primary Air
57
57
-
18
33
-
42
-
Secondary*
Air Velocity
High
Low
Low
High
High
Low
High
Low
*Actual Velocity Dependent upon primary air flow,
88
-------
1000-
o
600
e
a
a
O
400
200
5% Excess Air
20% Primary Air
1.3% N2 Coal
A Injector A
^ Injector D
O Injector F
Injector E
Figure 6.7 The Effect Of Fuel Injector Type And Swirl On NO Formation
In P.F. Flames
89
-------
1000
800
600
s
Q.
a
O
2
400
200
Primary Air Percentage
10%
Injector A
5% Excess Air
Figure 6.8 The Effect Of Primary Air Percentage On NO Formation
In Coal Flames
90
-------
1000
800
600
6
a
a
400
200
Primary Air Percentage
O 10%
O 20%
D 30%
Injector B
5% Excess Air
1.1% N2 Coal
Figure 6.9 The Effect Of Primary Air Percentage On NO Formation In Coal
Flames
91
-------
TABLE 6.3 INPUT CONDITIONS FOR MAIN FLAME MEASUREMENTS
Flame
No.
119
120
121
122
123
124
Fuel
gas
gas
coal
coal
coal
coal
Injector
Type
33
8
F
B
A
A
Position
throat
throat
throat
throat
exit
exit
Rs
.6
.9
.6
. .9
.9'
.9
Primary
Air
-
-
20%
20%
30%
20%
Excess
Air
5%
5%
25%
25%
10%
10%
For all flames burner divergent 25°, and throat diameter 17.6 cms,300°C air preheat.
92
-------
REFERENCES
6.1 Heap, M.P. , Lowes, T.M. , Walmsley, R. , Bartelds, H., and
Le Vaguerese, P., Appendix to Volume I - Investigations to define
the influence of burner variables upon nitric oxide formation in
natural gas and pulverized coal flames - Experimental Results
(to be published).
93/94
-------
7. DISCUSSION OF RESULTS - NATURAL GAS FLAMES
The influence of burner variables on nitric oxide formation in natural gas
flames has been studied in a series of parametric investigations which were
described in Ssection 6. The range of emission levels measured with the differ-
ent variables can be seen in Fig. 7.1 when plotted as a function of swirl
number. Fig. 7.1 also indicates the range of emission levels for the various
characteristics flame types. At present there is no fundamental model which
is capable of predicting the effect of burner variables on emission levels.
However, some kind of mechanistic understanding must be obtained if the
results are to be used to specify design criteria for low emitting burners.
7.1 A Qualitative Model of Nitric Oxide Formation in Turbulent Gaseous
Flames
The natural gas used in this investigation does not contain chemically-
bound nitrogen, although approximately 10-14 percent of molecular nitrogen
is present in the gas. Consequently, only the formation of thermal NO need
be considered.
The discussions presented in Section 3 emphasized the strong influence
of temperature on the rate of formation of NO. Figure 7.2 presents axial nitric
oxide concentrations for four flames. The corresponding axial temperature
distributions are presented in Fig. 3.5 which also shows the maximum mea-
sured temperatures. It can be seen that for none of the flames does the mea-
sured temperature exceed 1550°C. Also, the maximum temperatures measured
for the two jet flames are very similar and yet the emission of the flame with
weak swirl is about half that of the flame with no swirl. When no preheat is
used external recirculation temperatures do not exceed 1200°C. Thus, for
the flames under discussion in this report, nitric oxide formation can be con-
sidered solely as a flame phenomenon since insignificant amounts of nitric
oxide will be formed in the bulk gases. The nitric oxide distributions shown
in Fig. 7.2 cannot be accounted for on the basis of the flame temperatures
shown in Fig. 3.5. These temperatures are time mean values. Neither the
response time of the suction pyrometer nor the spatial resolution are sufficient
to measure the temperatures which are believed to be of importance for NO
formation in turbulent diffusion flames (7.1).
95
-------
100
Q.
a
80
S 60
40
20
Type II
Figure 7.1 NO Emission Characteristics Of The Various Flame
Types
96
-------
CD
O s=0.5
propane
— — natural gas
12 345
Axial Distance (in)
Figure 7.2 The Influence Of Swirl An Axial NO Distribution Of Jet Flames
-------
In turbulent diffusion flames the total heat release is distributed over
discrete zones. The sum of the zone volumes is less than the "time mean" total
flame volume which consists of:
unmixed fuel air and combustion products;
mixed but unreacted fuel air and combustion products;
reacting fuel and air;
products of combustion.
The amount of nitric oxide produced from the combustion products associated with
one element of fuel will depend upon:
their initial temperature T0;
D
the rate of change of temperature from T to some cutoff temperature T
below which the rate of NO production becomes insignificant;
the true oxygen atom concentration.
Although kinetic studies have indicated the importance of the last factor it will
be neglected in further discussion and it will be assumed that the temperature
dT
T and the function — control NO production in turbulent diffusion flames.
F dt
The total heat released within the flame volume is fixed at the input rate Q
which is the sum of all the heat release rates within the discrete reaction zones,
this
qx (2D
X = 1
where q represents the heat release rate of one reaction zone. Obviously
J\
the number of the individual zones will vary with time depending upon the turbulent
mass transfer process. Two essentially different types of reaction zones can be
envisaged:
mixing between fuel and air takes place before the reactant temperature
is sufficient to ensure that reaction will take place. Thus,
provided the mixture composition lies within the limits of flamability,
if reaction is initiation at any one point a flame can propagate
throughout the entire mixture;
due to mixing with combustion products either of the reactants are at
a sufficient temperature to allow reaction to proceed should
mixing occur.
98
-------
The temperature T of the freshly formed products of combustion will depend
on the composition and enthalpy of the mixture prior to reaction. It is important
to note that T does not depend on the fuel air equivalence ratio alone, but
also on the degree of dilution by and the temperature of any combustion products.
Maximum values of T will occur when the fuel and air are undiluted with
combustion products react in proportions near to stoichiometric. The rate at which
the temperature of the freshly formed products of combustion will be reduced
depends upon:
the bulk gas temperature;
the rate of mixing with the bulk gases;
the amount of radiative heat loss which will be dependent on the volume
and emissivity of the combustion products and the temperature of the
furnace enclosure and the bulk gases.
Heat loss by radiation will play only a minor role. The major influence on the
rate of temperature decrease will be the mixing rate between the bulk gases and
the flame gases. Should the reacting gases come close to cooled surfaces then
convective heat transfer could also be important. Burner variables influence the
amount of nitric oxide produced in natural gas flames because they control the
mixing history of the four components of interest:
the fuel;
the air;
internally recirculating combustion products;
externally recirculating combustion products;
The mixing history of these components will dictate the degree of dilution and
the reaction zone equivalence ratio and therefore the temperature T?. Also the
rate of mixing will influence the rate of temperature decrease of the products of
combustions.
The influence of burner variables on the mixing history of the four separate
components can be summarized by:
the type of fuel injector and its position will affect both the fuel air
mixing pattern and the mass exchange between the fuel and internally
recirculation vortex;
99
-------
the degree of swirl will control the rate of entrainment of external
recirculation and the size/ intensity and position of the internal
recirculation vortex;
the throat velocity will influence the fuel air mixing pattern and the
rate of entrainment of external recirculation;
the geometry of the burner exit will influence the size and intensity
of the recirculating gases.
With a knowledge of the properties of swirling flames it is possible to use
the qualitative model describing the formation of nitric oxide to explain the
influence of burner variables on the emission characteristics of large scale
natural gas flames.
7.2 Burner Variables and Nitric Oxide Formation
7.2.1 Axial Fuel Tet Flames
Axial fuel injectors of sufficient velocity can produce:
lifted flames;
simple jet flames with O
-------
1600
1200
O
o
a
6
800-1
400
0.8
6
a
o.
Axial Distance
O 2.75 m
O 2.035 m
A 1.48 m
0.4 0 "0.8 0.4
Radial Distance (in)
Figure 7.3 Radial Temperature And NO Distribution For Non-Swirling Jet Flames
-------
1400
o
o
(U
tf 13
s
-------
increased entrainment of external recirculation increases the rate of
decrease of temperature from T to T of the combustion products;
increased entrainment will mean that the dilution of the reactants
prior to combustion will increase thus reducing T .
From visual observations it is known that the application of swirl spreads
the flame and the flame length is decreased. The increased rate of entrainment
of the swirling jet can be seen from the results presented in Fig. 7.5 which com-
pares the forward mass flow for the two flames. Similar measurements have been
made in two propane flames. Swirl also reduces the emission of nitric oxide from
propane jet flames although the reduction is not so startling. Fig. 7.6 shows both
the axial temperature and NO distributions for the two flames. As with natural
gas the applications of swirl spreads the flame and the axial temperature gradient
is steeper although the measured temperatures with and without swirl are similar.
The most significant difference, associated with nitric oxide formation with different
fuel gases is their radial NO distributions. Compare the radial profiles presented
in Figs. 7.7 and 7.8 for propane with those in Figs. 7.3 and 7.4. Both propane
curves have a maximum on the axis while with natural gas the curves have a
minimum on the axis. Although the temperature profiles were similar for comparable
flames with different fuels species concentrations were different (See Fig, 7.9).
Larger concentrations of hydrogen, carbon monoxide and less oxygen were measured
on the axis of propane flames than on natural gas flames. Visually the propane
flames were more luminous than their natural gas counterparts.
The observations described above can be attributed to:
measurement errors. Soot deposited at the tip of the water-cooled
quartz sampling probe may react with the sampled gases producing
erroneous results;
the nitric oxide formed within the fuel rich core of the propane flame
may be produced by the "prompt NO" mechanism suggested by Fenimore
(7.1).
The above arguments are speculative and more detailed measurements should
be carried out to ascertain whether or not the effects observed with propane jet
103
-------
4*0
3'.0
2.0
O
i.o
s=Q
O s=0.5
i.o
2.0
3.0
Axial Distance (m)
Figure 7.5
Comparison The Forward Mass Flow In The Early Regions Of
Non-Swirling And Slightly-Swirling Flames
104
-------
1500-
1000"
O
o
G
-------
Axial Distance
O 0.74 m
0,20 0,40 0,60 0.80 1.00
1400
1200
1000
O
o
I 800
S
600
400
200
Axial Distance
O 0.74 m
O 1.665m
~
0.20 0.40 0.60 0.80 1.00
Radial Distance (m)
Figure 7.7 Radial Temperatue And NO Distributions For Non Swirling Propane Flame
-------
1200
1000
BOO
O
o
(0
IH
-------
,20 .40
.60
1.00
Radial Plstance (m)
Figure 7.9 Radial Gas Concentrations For Slightly Swirling Flames (S = 0.5)
-------
flames were real and the reasons for them.
The results presented in Fig. 7 .10 show that for high intensity Type I
flames an increase in swirl level causes an increase in the emission level.
This behaviour is contrary to that of simple jet flames where emissions are reduced
with increasing swirl. The influence of swirl on Type I flames can be explained
by reference to Fig. 7.11 which is a diagramatic representation of the conditions
prevailing at the bulbous root of the flame. The axial fuel jet penetrates the
internal recirculation zone which forms an annulus around the fuel jet* Mass exchange
takes place between the fuel jet and the reverse flow such that:
the fuel jet entrains combustion products
some of the fuel is stripped froir the main jet and is returned to the
root of the flame.
The hot fuel in the internal reverse flow region cannot react until it is mixed
with air. Combustion takes place in the flame region which separates the zone
of internal recirculation from the swirling air flow. The fuel remaining in the main
jet reacts further downstream with air that has been diluted with inerts due to
entrainment of external recirculation. If the swirl level is increased the intensity
of the internal reverse flow increases and more fuel in entrained by the internally
recirculating products. The formation of the internal recirculation marks the
conversion of a simple jet flame to a type I. Measurements suggest that minimum
emissions occur at this point. The increased emission with swirl for Type I flames
is due to a redistribution of the zones of heat release. As the swirl increases more
fuel is entrained by the internal recirculation and reacts at the base of the flame.
Thus less heat release occurs in the main flame body. Heat release at the flame
base will produce more NO because:
T will be higher since dilution of the reactants will be less;
the rate of decrease of T_ to T will be slower because the internally
recirculating gases have a higher temperature than externally recirculating
gases.
Consequently, after initiation of the internal recirculation zone any further
increase in swirl increases the emission of nitric oxide.
109
-------
40
a
a
30
22
Air
Velocity
m
D
A
O 25
O 25
50
50
Quarl Angle
35
25
.o
O 50
Lifted Flames
.20
.40 .60
R
.80
1.0
Figure 7.10 The Influence Of Swirl Level On NO Formation In High
Intensity Type I Flame
110
-------
External Reclrculatlon
Reverse Flow Region
Figure 7.11 Diagramatic Representation Of Conditions At The
Base Of A High Intensity Type I Flame
-------
Three other burner parameters influence the emission of nitric oxide from high
intensity Type I flames; the throat velocity, the angle of the burner divergent
and the position of the fuel injector.
The velocity of the combustion air: the mass of external recirculation entrained
by the flame jet is dependent upon the momentum of the combined primary and
secondary jets. Thus increasing the velocity of the air supply increases the mass
of external recirculation entrained by the flame jet. This will tend to reduce
nitric oxide formation because the combustion air will contain a greater proportion
of inert diluent and the flame products will be quenched more readily. The results
presented in Fig. 7.10 show: that the maximum emissions occur with the lower air
velocity.
The angle of the burner divergent: the angle of the burner divergent has two
effects upon flame characteristics. The smaller the angle the longer and narrower
the flame. Also there is an optimum angle of approximately 25 for flame stability.
The increase in stability is believed to be associated with the strength of the
internal reverse flow region. "Strength" is a nebulous term and may refer to either
the reverse flow velocity or the mass flow rate of recirculating products. Under
the circumstances discussed above it refers to the ability of the reverse flow to
entrain fuel. The results presented in Fig. 7.10 suggest that with Type I flames
the most important factor is the entrianment of external recirculation. At 25m.sec.
the narrower angle gives the minimum emission because the longer flame means that
the combustion is delayed and the air will be more dilute thus reducing T . At high
air velocities the angle of the quarl has a negligible effect because the entrainment
of external recirculation is dominated by the higher jet momentum.
The position of the fuel injector: moving the fuel injector from the throat to
the exit of the quarl means that less fuel is entrained by the internally recirculating
products. Consequently the emission of nitric oxide from Type I is lower when the
injector is at the exit rather than in the throat of the quarl. The emissions from
lifted flames are generally low (See Fig. 7 .1) because reaction takes place remote
from the fuel injector. The total heat release occurs after significant dilution with
entrained products. Thus potential T of each zone of heat release are low and
consequently emissions tend to be low.
112
-------
7.2.2 High Intensity Type II Flames
Three different fuel injector types were used during the investigation to produce
high intensity Type II flames:
a low velocity single hole injector;
a multihole divergent injector;
a multihole radial injector.
The single hole injector was not used extensively as it tended to produce
asymmetric flames and the following discussion will be restricted to the emission
characteristics of multihole injectors. The emission characteristics of the two
injectors are presented separately in Figs. 7.12 and 7.13. Comparisons between the
two injectors are possible by reference to Figs. 7.14 a and b.
The effect of burner variables on the emission characteristics of high intensity
Type II flames can be summarized by:
increasing swirl reduces emissions with divergent fuel injection;
the influence of swirl is less pronounced with radial fuel injection
and trends are not clear cut;
at the lower air velocity emissions from both injector types are
dependent of divergent angle;
optimum burner conditions for minimum emissions with both injectors are
high swirl -35° burner divergent at throat velocity of 50 m.sec. . These
burner conditions also give minimum emissions at zero swirl.
It is believed that the mass transfer within the burner divergent dominates the
production of nitric oxide in Type II flames. Unfortunately it is extremely difficult
to make accurate measurements within the divergent in combusting flow. Consequently,
the discussion of the phenomena summarized above must be based upon a composite
of the information presented in Section 3 , relevant isothermal measurements and in-
tuition. The most significant influence of the fuel injector type upon emissions will
be the variation in the fuel air mixing pattern.
113
-------
80
70
60
a
a
40
30
0
Figure 7.12 Emission Characteristics Of High Intensity Type II Flames
Produced With A Divergent Fuel Injector
114
-------
a
a
O
Air Velocity
m sec"
1.0
Figure 7.13
Emission Characteristics Of High Intensity Type II Flames
Produced With A Radial Fuel Injector
115
-------
a
a
O
S
Divergent, exit 25°
Radial, exit 25°
D Radial, exit 35°
A Divergent, exit 35°
Figure 7.14a Comparison Of Emission Curves Multihole Injectors ;
(25 m sec""-)
Figure 7.14b
Comparison Of Emission Curves Multihole Injectors
(50 m sec'1)
116
-------
From the earlier discussion the fuel air mixing pattern will control the
distribution of the total heat release between many separate zones each with
its own equivalence ratio and degree of dilution. The burner variables will influence
the size and intensity of the recirculation zone and the amount of entrained com-
bustion products.
The radial fuel injector will produce more rapid fuel air mixing than the divergent
injector. This effect has been confirmed by isothermal tracer measurements with
scaled injectors (Fig. 7.15). A tracer was added to the primary stream and the decay
of tracer concentration and flow boundaries were measured within the divergent.
The concentration measurements presented in Fig. 7.15 confirm that at zero swirl
primary secondary mixing is much faster with the radial injector than with the
divergent injector. Increasing swirl improves the mixing with both injector types.
With the radial injector the fuel is injected perpendicularly to the air stream.
Therefore mixing of some of the fuel is most likely to take place without reaction
taking place. Under these circumstances the total heat will be released over a
wide range of equivalence ratios with consequent reduction of T . The divergent
injector directs the fuel parallel to the air stream and mixing will take place at the
boundary between the two, allowing more of the fuel to be burned in zones with
an equivalence ratio close to unity and T will be higher. Thus at zero swirl and
low velocities the emission from the divergent injector will be higher than that from
the radial injector.
The results presented in Fig. 7.14a indicate that swirl has a greater effect
upon emission from divergent rather than from radial injectors. The difference in
the flow patterns within the divergent probably account for this observation. With
the divergent injector the fuel and air can more easily be diluted with recirculating
products prior to reaction than with the radial injector. Fuel air mixing will take place
prior to reaction with the radial injector and significant contact with hot recirculating
products will cause ignition of an undiluted mixture. Increasing swirl will increase
the rate of temperature decrease of the high temperature products for both injectors.
Increasing the throat velocity decreased the emission levels produced with the
117
-------
Radial Injection
Divergent Injection
Rs=1
Concentration Scale
0 20 40 % Cone.
R = 0.5
s
Nl
\
\
\
\
15 10
Figure 7.15
0 cms
15 10
0 cms
Isothermal Tracer Concentration Measurements Comparing
Divergent And Radial Injectors
118
-------
divergent injector markedly at low swirl levels. The emission levels measured
with throat velocities of 50 m. sec. were similar for both multihole injectors.
The formation of nitric oxide will be reduced at high throat velocities because of the
increase rate of entrainment due to the higher jet momentum. It is not readily
apparent why the burner divergent only has a significant influence upon the emission
levels with high throat velocities.
For all the burner conditions investigated similar emission characteristics were
observed when the radial injector was placed at the exit of the divergent (See Fig.
7.16). As the swirl level was increased the emission increased to a maximum and
then decreased. The form of the emission curve is independent of burner divergent
and throat velocity. In an attempt to explain these emission characteristics
conditions at the injector can be represented as in the sketch shown in Fig. 7.17.
At zero swirl the internal reverse flow zone is formed in the wake of the fuel
injector. Mixing between the air and fuel is not so efficient as when the injector is
in the throat due to a decrease of air velocity because of expansion in the divergent.
Combustion is delayed, thus more of the air becomes diluted with inerts and T
is reduced and emissions are low. When the swirl is increased the fuel and air mix
faster before significant dilution takes place, T increases as does the emission.
As the swirl is increased the internal reverse flow zone is stabilized and then it
begins to increase in size. The boundary of the reverse flow zone moves upstream
until it covers the fuel injector (See Fig. 7.17). At this point three factors have
a major influence upon nitric oxide formation:
improved fuel air mixing produce earlier combustion before dilution thus
NO increases;
the fuel jet must now penetrate the reverse flow zone, entrain inert
products thus NO formation decreases;
increased swirl causes an increase in entrainment and NO formation is
reduced.
Considering the above factors it can be seen how the emission can increase and
reach a maximum and then decrease.
Evidence in support of the above explanation has also been obtained by tracer
studies in isothermal flow. The change in position of the reverse flow zone with
increasing swirl can be seen in the results presented in Fig. 7.18. It can also be
119
-------
80
70
,$0
50
40
O
30
20
10
Quarl Angle Air Velocit1
O 25 25
O 35
D 25
50
50
T
m sec
0,2 0.4 0.6
R
0.3
1.0
Figure 7.16 Emission Curves For Type II Flames
Produced With A Radial Injector At
The Burner Exit
120
-------
Flow Boundary
Radial Fuel Injector
Low Swirl
Internal
Reclrculation
77/77
hgure 7.17
High Swirl
Flow Boundary
Mixing Pattern At The Base Of
Type II Flames Produced With
A Radial Fuel Injector At The
Burner Exit
External
Reclrculation
-------
t • I «—»-
0 20 % concentration
sw = 10
-10
sw = 5
-10
sw = 0
\\\\\
-10
Figure 7.18
Primary Concentration And Position Of Recirculation
Zone For A Radial Injector Placed At The Exit Of The
Burner
122
-------
seen that the primary matter is carried upstream by the reverse flow allowing
combustion to take place within the divergent.
7.2.3 The Influence of Flame Quench Rate
Considering the dependence of nitric oxide formation in natural gas flames upon
temperature it is not surprising that quench rate has a significant influence upon
nitric oxide emissions. Two different experiments were carried out which illustrate
this influence:
refractory burner divergents were used instead of water-cooled divergents;
the furnace cooling load was reduced.
For the three injector types tested emissions were lower when a water-cooled
divergent was used rather than a refractory one. In all three instances the form of
the emission curve is similar (See Fig. 7.19). The surprising feature of the results
is the relative influence of the divergent type with the three fuels. The high velocity
single hole injector which produced a high intensity Type I flame showed the most
influence.
When the furnace cooling load is reduced, the bulk gas temperature increases thus
the enthalpy of the reactant mixture is increased and the rate of temperature decrease
from T to T is reduced. The influence of furnace cooling rate on NO formation is
also dependent upon flame type. It can be seen from Fig. 7.20 that the bulk gas
temperature affects emissions from Type I flames more than from Type II flames.
In Type II flames mass exchange between the flame volume and the internal recirculation
zone predominates. The temperature of the internal recirculating gases is not strongly
dependent upon furnace cooling load. Therefore the bulk gas temperature has less
influence upon nitric oxide formation than in Type I flames which are longer and the
volume entrains larger masses of external recirculation.
123
-------
Divergent Multihole Injector
60
50
40
30
O Refractory Exit
Water Cooled Exit
a
a
50
40
O 30
Single Hole Injector
40
30
20
Radial Multihole Injector
Figure 7.19
.2
.4
.8
1.0
Effect Of Heat Extraction At The Flame Base On
NO Emission
124
-------
60
50
E
S 40
O
a
30
20
Type I
0.8
1.0 1.2
Heat Extracted (MW)
1.4
Picture 7 20 Influence Of The Heat Extraction On The NO Formation
' In Type I And Type Flames
125
-------
7.3 Nitrogen Dioxide Formation in Natural Gas Flames
It is generally accepted that the oxide of nitrogen produced initially in
combustion processes is nitric oxide (NO).
During the natural gas investigations significant quantities of nitrogen dioxide
(NO ) were found in the flue gases. Figures 7.21 and 7.22 present flue gas (NO,
NO and NO concentrations at various swirl levels for a Type I flame and a Type
£» X
II flame. It can be seen that although the NO emission varies significantly the
J\
NO_ content remains constant with swirl.
£
The oxides of nitrogen concentrations were determined with a chemiluminescent
analyzer. The NO concentration was determined by diverting the sample through
5C
a heated packed quartz tube to thermally convert NO, to NO prior to analysis.
-------
a
a
1
CD
O
o
O
Fuel Injector Single Hoi
Fuel Injector Pos. Thro
Figure 7.21
Flue Gas Concentration Of NO, NO and NO, As A Function
Of Swirl Number For A High Intensity Type I Flame
127
-------
100
Fuel Injector Type Radia
Fuel Injector Pos. Exit
Figure 7.22
Flue Gas Concentration of NO, NO and NO,, As A Function
Of Swirl Number For A High Intensity Type IFFlame
128
-------
20
h 2000H
15
12
(M
Flama 119
AD 10
O NO
Temperature
0.30
0.20
0.10 0 0.1
Radial Distance (m)
0.2
0.3
Figure 7.23
Radial Distributions Of Temperature O2 NO And NO2 10cm From
The Burner Exit In A Type II Natural Ga% Flame
129
-------
The reduction in NO? concentration to zero in the flame region may not be a real
effect but due to the reduction of NO in the converter by CO, HZ and solid
carbon. However, the increase of NO, concentration is a real effect and cannot
be attributed to the converter. Thus it appears that there may be considerable
concentrations in the flame region. There are three explanations for the
~
observations concerned with nitrogen dioxide:
nitric oxide in the recirculation is oxidized in the air jet;
nitrogen dioxide is produced in the flame region;
the compound is not nitrogen dioxide but is something which can be
converted to nitric oxide within the quartz converter.
Nitrogen dioxide is unstable at high temperatures and it will decompose as the
combustion products flow along the furnace. This effect is illustrated in Fig. 7,24,
Plug flow conditions can be considered to be established 2 m from the firing wall
under highly swirling conditions. After this point it can be seen that although the
NO content of the combustion products remains constant the NO level increases and
jC
the NO concentration decreases. This effect may explain observations reported
by other investigators (7 . 2) and also at the IFRF where combustion is completed ,
the bulk temperature is low and dropping and yet nitric oxide coincentrations are
increasing.
7.4 Correlation of Emission Data
Wendt (7.3) has investigated the use of dimensional analysis to model the
effect of swirl on NO emission levels from a test furnace. He concluded that
X
the conventional dimensionless analysis approach was not very useful but that
considerable insight into the problem could be gained from what Spalding (7.4)
refers to as partial modeling. Wendt suggests that since nitric oxide formation
in natural gas flames is a function of the furnace, the input conditions and the
burner variables should be included. Although the populations are small, an
attempt has been made to correlate the emission data against a number of para-
meters related to the burner variables.
Table 7 . 1 summarizes all the experimental data obtained at a fixed thermal
input, furnace cooling load and excess air.
130
-------
Fuel Injector Radial
Fuel Injector Pos. Exit
Figure 7.24
Axial Distance (m)
Axial Distribution Of NO, NO2 And NOX In Natural Gas Flame
-------
TABLE 7.1 Summary of Experimental Data
Injector Type
Single hole
Divergent multihole
Radial multihole
All types
No. of
Observations
42
53
69
164
Mean
Value ppm
36.2
42 .6
45.6 .
41,6
St. devi-
ation
6.02
11,2
9,93
10.5
Max . value
ppm
50
80
76
80
Min,
Value
23
30
28.5
23
Multiple linear regression analysis was used in an attempt to correlate the data
against the experimental variables. The results are presented in Table 7.2 and
as could be expected from the emission curves presented earlier the correlations
are not very significant.
Table 7 .2 Multiple Linear Regression Correlations
Injector
Type
Single hole
Divergent
multihole
Radial
multihole
All Types
Regression Equation
ppm NO 42- ,231 (Vel.rn.sec~1)
ppm NO= 89-13,24 (swirl number)
- 6.48 (vel.m.sec""1)
- 3.. 85 (quarl angle)
ppm NO 69, 2-8. 8 (swirl number)
-441 (vel.rn.sec~1)
ppm NO 61, 9-8. 21 (swirl number)
-.357 (vel.m.sec"1)
Correl. Coeff.
0.466
0.785
0.664
0.542
Standard
Deviation
5.4
7.1
7.54
8.91
132
-------
Further correlations were attempted with 25 "manufactured " parameters which are
listed in Table 7.3. The correlation coefficient for all tae 25 variables was 0.
-------
TABLE 7.3 PARAMETERS USED IN MULTI-REGRESSION ANALYSIS
Variable
s£
S
s2'
sin
G .G "1
(GpU).GsY~1)
(21 tan o& * 3D)2 . B~2
d2 . (21 tan x + D2)""1
G Usiny. (G T [1 + S21)~1
P S <- 0J 4
opDcos?. (B.Y [t + s8])-1
T
S . D
-1
S . D '
2
S sia f
(L - 1)L-1 cos £ oc S2 sin/3
G « 8 . (1 -f cos y5)
P
OC
[G . (G )-1l1 •+ sia P
*•» I^ pi 9fm
Grt/Gp . S
S G T . (G U)~^
(L - i)L~1 cos % o. . GB(G )"1
(I - i)iT1 cos ^cc. « G T(G u)~1
1 B p
cc
Gfl T (1 + S2)
greater than 1$
significance
X
X
X
X
X
•
X
X
X
X
X
X
X
X
aspect
I
X
I
III
III
IX
II
IT
I
I
y
IT
II
II
T
T.
T
134
-------
NOMENCLATURE
A - recirculation constant
c - specific heat
P
D - throat diameter
F - empirical heat release factor
G - momentum
k - emission, attenuation coefficient
L - length of quarl
1 - distance of injector from the throat
N - exponent on the oxygen concentration
N - degree of oxidation
R - radius
S - swirl number
T - temperature
t - time
U - velocity of the primary jet
V - velocity of the secondary jet
V - volume of reactor
P - density
oc - quarl angle
-------
SUBSCRIPTS
p - primary
pf - plug flow
s - secondary
ws - well stirred
138
-------
REFERENCES
7 .1 Fenimore, C.P.
Thirteenth Symposium (Int.) on Combustion, p. 373,
The Combustion Institute, 1971
7 .2 Shof stall, D.
Private communication
7.3 Wendt
EPA Services Contract E.H. SD-71-45 Task No. 16
7 .4 Spalding, D.B.
Ninth Symposium (Int.) on Combustion, p. 833,
The Combustion Institute, 1962
1.3-7
-------
8. DISCUSSION OF RESULTS.- PULVERIZED FUEL FIAMES
The parametric investigations carried out with coal as a fuel had the
same object as those using natural gas as a fuel - the isoluation of burner
variables which are important to the formation of nitric oxide in coal flames.
8.1 Nitric Oxide Formation in Pulverized Fuel Flames
The combustion of pulverized coal is a complex process involving two-
phase flow, rapid particle heating both by convection and radiation, devolatiliza-
tion of the coal, combustion of volatile fractions and heterogeneous combustion
of the solid char. During this process it is possible that both thermal and
fuel NO can be produced. Consider the sketch of the lifted coal flame shown
in Fig. 8.1 which separates the progress of combustion into three sections:
the fuel preparation zone;
the combustion of the volatile fractions; and
char burnout.
139
-------
>£>
Coal Jet
Char Burnout
Fuel Injector
Fuel Preparation
Volatile Combustion
Char Combustion
Figure 8.1 Idealized Sketch Of A Lifted Pulverized Coal Flame
-------
8.1.1 Fuel Preparation
Pulverized fuel is coal that has been crushed and ground in order that it
can be carried as a suspension in an air stream. The size distribution is
extremely important for rapid burnout. The mean particle size is normally in
the range 30 to 70 microns; however, inevitably there is a wide size range.
Ten percent by weight of the fuel may be less than 10 microns and 10 percent
greater than 100 microns. The fuel is injected into the furnace with air as a
primary jet where the particles are heated and the ignition of the volatile
fractions marks the end of the fuel preparation zone.
The primary jet normally contains 20 percent of the total air flow with a
temperature less than 100°C. The secondary air is normally preheated to
temperatures about 250°C. Some distance from the injection point the primary
and secondary jets combine and the velocity distribution can be described by
a single Gaussean distribution (8.1). The mixing rate between the two jets is
dependent upon the ratio of the primary and secondary velocities and the thick-
ness of the primary/secondary interface. The entrainment of hot external
recirculation is dependent upon the combined mass and momentum fluxes.
In the fuel preparation zone the coal particles receive heat by two
processes:
by radiation from the furnace walls and the ignition front; and
by convection from the preheated secondary stream and the entrained
recirculating gases.
At a certain temperature the coal particles begin to decompose producing
tars and gases collectively referred to as volatiles. The volatile fractions con-
sist of carbon dioxide, water vapor, carbon monoxide, hydrogen, hydrocarbons,
and nitrogen containing species. The composition of the volatiles and the
weight loss of the particle depend upon the temperature/time history of the
particle. Therefore, the rate of heating of the particle and its final tempera-
ture influence the quantity and composition of the evolved volatiles. The heat-
ing rate of p.f. particles normally exceeds 104 °C per second and the evolved
gases are rapidly transported away from the particle surface.
141
-------
8.1.2 Combustion of the Volatile Fractions
The volatile fractions will combust when they are mixed on a microscale with
oxygen, the mixture lies within the flammable limits and a source of ignition is
present. The distance between the visible ignition front and injector will be referred
to as the ignition distance. This distance will be strongly dependent on the mixing
rate of the coal with high temperature gases and will also be dependent upon fur-
nace wall temperature. The combustion of the volatile coal fractions may be
considered to occur in an envelope around the fuel jet or as a combustion zone
associated either with a single particle or a group of a small number of particles.
When the flame envelopes the fuel jet,conditions such as those shown in Fig.
8.2 previal. The combustion region can be considered as a region into which fuel
gases and oxygen pass from the fuel jet side and oxygen passes from the air side.
Combustion products leave the region at both boundaries. The amount of oxygen
mixed with the fuel gases will be dependent upon the initial amount of primary air
and the rate of mixing between the primary and secondary streams prior to the
ignition front. When combustion is associated with a single particle the volatile
fractions can either burn in a diffusion flame around the particle or as a premixed
flame in the wake of the particle.
8.1.3 Char Combustion
Char are those solid particles which remain after the devolatilization of the coal
particles is complete. It may consist of soot particles, cenospheres and cellular
particles. Sternling and Wendt (8.2) consider that the fuel nitrogen is divided between
the volatile and char fractions of the coal. Analysis of the char remaining after
proximate analysis of coal volatile content confirms that all the nitrogen is not
volatile. This experiment can only be considered indicative and it will not give a
quantitative value of the volatile fuel nitrogen for the pulverized coal particles
since the time temperature history of the coal in the two cases is different.
142
-------
gas, O2, N2/ and CO2
Secondary jet
flame boundary
Coal jet
flame boundary
flame region
coal + O2 + CO2 + N2 + evolved volatiles (including XN)
Figure 8.2 Conditions At The Fuel Jet Boundary With An
Injector Stabilized Flame
143
-------
There is a fundamental difference between the combustion of the volatile and
char fractions. The reactions involved in the combustion of the volatiles have
finite rates but these are usually high compared with surface reactions involved
in the combustion of the char. Also the combustion of the char particles involves
several steps in sequence:
transport of oxygen or other reactant gas to the particle surface;
reaction with the surface;
transport of the reactants away from the surface.
The overall reaction rate is of course dependent upon the slowest of these
steps. It is probable that in the combustion of the char the product released at
the surface of the particle is carbon monoxide (8.3). In Section 3 .1 it was concluded
that although the precise mechanism of fuel NO formation was unknown the amount
produced could be attributed to two competing reactions:
+ ... (16)
+ ... (17-)
where I is an unknown nitrogen containing intermediate. Under fuel rich
conditions reaction (17) is faster, but with fuel lean combustion large amounts of
the fuel nitrogen can be converted to NO. Earlier it was stated that of the
nitrogen bound in the original coal a proportion can be considered to be "refractory"
and remain in the char. It has been assumed that this fuel nitrogen does not form
significant quantities of fuel NO. This assumption is based on the fact that the
char particle will be surrounded by a stagnant boundary layer deficient in oxygen
containing carbon monoxide. Thus even though NO were to be formed at the
particle surface it could be reduced on passage through the fuel rich stagnant
boundary layer.
Consequently, if the assumptions stated earlier are valid the influence of
burner variables on nitric oxide formation in pulverized fuel flames can be explained
by variations in the mode of combustion of the volatile fractions. If the kinetic
144
-------
studies are correct then the fuel NO formation can be reduced by reducing the
oxygen content of the volatile flame region. If the volatile gases bum in the wake
of a series of particles as a premixed flame it can be assumed that the conversion of
fuel nitrogen to fuel NO will be high. If, however, the volatile fractions burn in
an envelope surrounding the coal jet it can be assumed that minimum conversions
will occur when the oxygen for combustion approaches the flame envelope from one
side only. This means that the primary fuel jet does not contain oxygen.
8.2 The Influence of Burner Variables on Nitric Oxide Formation in Pulverized
Coal Flames
8.2.1 Burner Conditions Associated with Maximum Emissions
In Section 8.1 the emission of nitric oxide from pulverized fuel flames was
attributed primarily to the formation of fuel NO from the volatile fuel nitrogen
compounds. If this is correct then burner conditions which tend to rapidly mix the
coal with all the available air will produce maximum emissions. This rapid mixing
of coal and air is most efficiently achieved in high intensity Type II flames.
In the parametric investigations described in Section 6 , three fuel injectors produced
Type II flames:
low primary velocity axial fuel injectors at medium swirl levels;
annular fuel injectors at medium swirl levels;
radial fuel injectors at all swirl levels.
Figs. 8.3 and 8 .4 show the emission characteristics of radial and divergent injectors
for both the high nitrogen and the low nitrogen coals. The following features can
be seen in these results:
nitric oxide emissions are always higher with the high nitrogen coal;
swirl has almost no effect on nitric oxide formation in .dames produced
with the radial fuel injector but causes an increase in the emissions
from the annular injectors.
emissions with both injector types are almost independent of the primary
air supply.
The radial fuel injector rapidly mixes the coal and air at all swirl levels, because
mixing is rapid then it is immaterial whether or not the air supplied either with the
primary or with secondary supply. Also rapid mixing means that the volatile nitrogen
145
-------
s
a
a
100&
90C —
80C
70C
60C
50C
40C
Annular Injector
5% Excess Air
Figure 8.3 The Effect Of Primary Air Percentage On NO Emission
1000
S
a.
a
900
800
700
600
10% PA
\— onoiPA
Radial Injector
5% Excess Air
300°C
1.1% N2 Coal
1.8%Nrt Coal
0.4
1.0
1.2
0,6 0.8
S
Figure 8.4 The Effect Of Primary Air Percentage On NO Emission
146
1.4
-------
compounds are well mixed with all the available oxygen and conditions favor
reaction (16), thus fuel NO formation is a maximum. Similar conditions are
achieved with the annular fuel injector at high and medium swirl levels when
a normal Type II flame is produced. Mixing between fuel and air will always
take place rapidly with this type of flame. It is this characteristic which
enables high combustion intensities and short burnout times to be achieved
plus early stable ignition within the burner divergent. Therefore, these charac-
teristics will be required of practical burners for wall-fired boilers and emissions
will tend to be high and of the same order of magnitude as those shown in
Figs. 8.3 and 8.4.
8.2.2 Reductions of Nitric Oxide Emissions Associated with the
Stabilization of the Ignition Front at the Injector
During the parametric investigations with axial fuel injection it was
observed that nitric oxide emissions made a step decrease when the ignition
front was established at the burner when the swirl level was increased. This
corresponded to the conversion of a lifted flame to a jet flame with an ignition
front stable at the injector. The data presented in Fig. 8.5 confirms that nitric
oxide emissions are reduced when a flame zone surrounds the coal jet. These
measurements were obtained acidentally. Initially, fuel injector A was uncooled
and at high swirl levels with the ignition front stable on the injector a coating
of red hot coke was deposited on the thick interface. This char acted as an
ignition source so that when the swirl was reduced to zero the ignition zone
remained stable at the injector. Figure 8.5 shows the measured flue gas nitric
oxide concentration as the swirl level is increased to a maximum and then
decreased. The coal jet was always enclosed by a flame zone as the swirl
was reduced to zero.
A significant difference only exists between the measured concentrations
with increasing and decreasing swirl at low swirl levels. At zero swirl with
the ignition front some distance from the injector the emission was 650 ppm.
This was reduced to 380 ppm when the entire coal jet was surrounded by a region
of combusting volatiles. It was stated earlier that high oxygen concentrations
associated with the volatile combustion will preomote fuel NO formation.
In the lifted flame the ignition distance was approximately 0.75 m. Thus at
the ignition point the primary and secondary jets are well mixed, the axial coal
147
-------
700 .
OSwirl Adjustment Increasing
Adjustment Decreasing
.2
.4
1.0
Figure 8.5 Hie Effect On Nitric Oxide Emission Of An Ignition Front Surrounding
The Coal Jet (Injector A, Throat Of 35° Quarl, 20% Excess Air 20%
Primary Air)
148
-------
concentration reduced and the axial oxygen/coal ratio increased. Consequently,
when the volatile gases begin to burn, not only is oxygen already mixed with
volatile fractions, but oxygen can mix into the ignition front from both boundaries.
When an injector stabilized ignition front is formed only that oxygen originally
supplied as primary air is mixed with the fuel jet. Therefore, the oxygen/coal
ratio will be reduced and less oxygen will be mixed with the volatile fractions
prior to ignition. The presence of the ignition front acts as a semi-permeable
barrier preventing further fuel/air mixing before the volatiles have been burned.
8.2.3 Burner Conditions for Minimum Nitric Oxide Emissions from
Pulverized Coal Flames
If the preceding discussions contain any truth, then the burner conditions
which will give minimum emissions can be specified:
the coal jet should contain the minimum quantity of primary air;
mixing between the primary and secondary streams should be
avoided;
the ignition zone for.the volatile fractions should be formed as
close to the injector as possible.
These conditions are best fulfilled by using a high velocity, single hole
axial injector with the minimum amount of primary air. The design of the
experimental system used during the furnace investigation did not allow the
primary air percentage to be varied independently. Any variation in the primary
air supply to a particular injector also resulted in a change in both primary
and secondary velocities. To overcome this problem three fuel injectors (B,
C and H) were used, which had similar primary velocities with different
quantities of primary air. Thus the effect of the primary air percentage on
nitric oxide formation could be investigated without interference from varia-
tions in other parameters, particularly the primary velocity. The effect of the
primary air percentage on the emission curves for three different quarl-injector
position arrangements can be seen in Figs. 8.6, 8.7 and 8.8.
In general, the emission of NO decreases as the swirl intensity increases,
this is particularly evident for the parallel burner exit. This is attributable to
dilution of the combustion air with external recirculation prior to mixing with the
149
-------
500
500
a
a
§ 400
300
200
Primary Air Percentage
O 30% __
O 10%
0
Figure 8.5 The Influence Of Primary Air Percentage With Injector
In'The Throat Of A 25° -Divergent
800
700
600
s
a 500
400
300
200
Primary Air Percentage
O 10%
D 20% ~
O 30%
0
Figure 8.7 The Influence Of Primary Air Percentage With Injectors
At The Exit Of A 0 Divergent
150
-------
S
a
a
0
2
700
500
Primary Air Percentage
O 10%
O 30%
Figure 8.8 The Influence Of Primary Air Percentage With Injectors At The
Exit Of A 25° Divergent
Burner Conditions Concerning Figures 8.6. 8.7, and 8.3
Primary Air Percentage Fuel Injector Primary Velocity
10%
20%
30%
H
C
B
52 m sec
52 m sec"
57 m sec"
151
-------
fuel. At the high primary velocities heat release occurs at some distance
from the burner throat and primary/secondary mixing is delayed.
The results obtained with the high velocity, single hole injectors are
in agreement with the assumptions concerning nitric oxide formation in
pulverized fuel flames. Minimum emissions occur with minimum primary air
percentage provided primary/secondary mixing is limited to a minimum before
volatile combustion is initiated. The influence of divergent angle and fuel
injector position can be seen in Figs. 8.9 and 8.10. Optimum conditions
appear to occur when the fuel injector is placed at the exit of a 25° quarl.
This may be the optimum condition because:
fuel injection at the exit will mean that combustion air will entrain
more inert products prior to combustion. This will both lower
oxygen concentrations and also reduce thermal NO formation which
may now be significant at these low emission levels;
ignition stability at the injector will require mixing at the jet
boundary between hot products produced upstream. A divergent
exit of 25° is optimum for the strength of the internal recircula-
tion zone.
It must be clearly stated that although these minimum emissions can be obtained,
the flames which produce them are jet flames and not high intensity Type II
flames, and as such, may not be suitable for industrial application. Also,
high swirl levels and primary velocities require the acceptance of larger
burner pressure drops than those used at present.
Two flames measured in detail during the furnace investigation also pro-
vided further evidence to support the hypothesis that an axial jet flame with
minimum oxygen/coal ratio in the primary stream produced minimum NO. Fig-
ures 8.11 and 8.12 allow a comparison to be made between the axial distribu-
tion of temperature and species concentration.
152
-------
400
Injection position
O throat
Figure 8.9 Effect Of Injection Position (Injector H, Throat 176 mm Diam.,
5% Excess Air, 300°C Preheat, 10% Primary Air)
500,
400
c
a
a
Burner ebcit geometry
O parellel exit
100'
Figure 8.10
Effect Of Burner Exit Geometry (Injector H, At Exit Throat
176 mm Diam., 5% Excess Air, 300°C Preheat, 10% Primary
153
-------
G r- 1800
en
•f" '
11
12
10
- 1700
- 1600
-01.500
c
.2
4-»
ID
0)
O
c
O
O
§
g!400
0)
Q.
6
0)
-^1300
- 1200
- 1100
800
700
600
500
!
h ^400
300
200
100
0 L- 1000 -
Flue gas NO 625 ppm
A Temp.
O °2
Figure 8.11
60 80 100 120 140 160
Axial Distance From Burner Exit (cm)
Axial Temperature ,Anct Species Distribution, Flame 123
180
200
220
-------
FLUE GAS NO 284 ppm
14 -i
12-
1700-1
700 -,
1600-
10-
1500-
IT)
8-
o
ge
o
o
o
1400-
a
6
1300-
600-
500-
6
a
a
400-
O
300-
0
1200-
1100-
1000
200-
100-
0 20
Figure 8.12
40
60
80 100 120 140 160
Axial Distance From Burner Exit (cm)
Axial Temperature And Species Distribution, Flame 124
180
CH,
I
200 220
-------
The only difference between the two flames was the amount of primary air
content. It can be seen that in the flame with the lower final emission and
primary air percentage the axial oxygen concentration drops rapidly to zero.
High concentrations of carbon monoxide and hydrogen also occur with the
lower primary air flame and significant quantities of methane were also detected
on the axis. These measurements confirm that low oxygen/coal ratios within
the primary jet reduce NO formation.
8.2.4 The Influence of Primary Velocity on Nitric Oxide Formation
One of the burner variables which were specified as being necessary for
minimum nitric oxide emission characteristics was a high primary velocity.
The influence of the primary velocity on the emission characteristics of
pulverized fuel flames can be judged from the results presented in 8.13 and
8.14. Fuel injectors B, C, and H had primary velocities of 19, 26, and
52 msec"^- respectively, and the secondary velocity was 41 msec .
At zero swirl with the injectors both at the throat and at the exit the
emissions were in the order:
B greater than H greater than C; however, at higher swirl values
the emission increases with decreasing primary velocity. The
anomolous behavior at zero swirl is probably due to a combination
of different primary/secondary mixing rates and ignition distance.
The effects of primary velocity are more pronounced when the fuel is
injected from the throat of the burner divergent (Fig. 8.13). The form of the
emission curyes is associated with the transition from one high intensity
flame type to another. In Section 3 the conditions necessary to form Type I
and Type II flames were discussed, and it was stated that the parameter G,
ratio of the primary to secondary velocities has a strong influence on the
type of flame produced. The results presented in Fig. 8.13 are for a fixed
secondary velocity and variable primary velocity. The emission characteris-
tics are attributable to:
High Primary Velocity (Injector H)
The high primary velocity ensures that a reverse flow zone cannot form on
the axis and an injector stabilized flame is produced. The interaction between
156
-------
the annular reverse flow zone and the primary jet gives ignition stability but
does not entrain sufficient fuel to form the bulbous base characterisics of
Type I flames and the emission curve is almost flat.
Medium Primary Velocities (Injector C)
Even at low swirl intensities, the primary jet velocity is insufficient to
penetrate the axial reverse flow zone. The fuel jet is split, a Type II flame
is produced and the emission increases. The reason for the high emissions
with Type II flames has been explained earlier. The high combustion intensity
within the divergent due to increasing swirl alters the static pressure distri-
bution and the primary velocity is sufficient to allow the fuel jet to penetrate
some way into the reverse flow zone. However, complete penetration does
not occur and a Type III flame is produced. Mixing between primary jet and
secondary air is restricted and the amount of nitric oxide produced is reduced.
Low Primary Velocities (Injector B)
Type II flames are more easily produced and better fuel/air mixing is
possible at lower swirl levels. Thus emissions are higher than with injector
C. Only at high swirl numbers is it possible to produce partial Type III flames
which quickly form Type II flames as the swirl is increased.
The difference between the emission curves presented in Figs. 8.13
and 8.14 is due to the modifying influence of the position of the injector on
swirl. With the injector at the exit of the burner, the primary jet does not
experience the fuE effects of the axial pressure gradients and thus, the
primary/secondary mixing is less intense. Consequently, even with the low
velocity injector B, a Type II flame is not produced.
8.2.5 The Influence of Throat Velocity on Nitric Oxide Formation
In natural gas flames an increase in throat velocity invariably produced a
decreased emission. In Section 7 this effect was attributed to a reduction in
thermal NO formation. If thermal NO formation is discounted in coal flames,
then the primary effect of throat velocity will be to change the rate of primary/
secondary mixing or to reduce the oxygen concentration of the primary jet by
entrapment of combustion products. The results are confused and the effect
of throat velocity is obviously dependent on other burner parameters, e.g.,
157
-------
a
a
800
700
600
500
400
300
200
Fig. 8.13 The Influence of Primary Velocity on NO Formation in
P/F Flames ( Injector in the Throats)
600
a
a
500 _
400
300
200
0
Fig. 8.14 The Influence of Primary Velocity on NO Formation in ?F
Flames (Injector at the Exit of the Divergent)
158
-------
injector type, divergent angle and primary air percentage. The results
obtained during the parametric investigations covered the entire range;
increased throat velocity either increases, decreases or has no effect on the
emission of nitric oxide.
8.3 The The Influence of Coal Type
Two different coals were used during the parametric investigations to
determine whether or not the influence of burner parameters on nitric oxide
formation in pulverized coal flames are dependent on coal type. The results
presented in Figs. 8.3 and 8.4 confirm that the increase of emission is not
directly proportional to the increased nitrogen content. However, the form
of the emission curves is similar for both coals. This type of behavior was
typical of all cases. The results presented in Figs. 8.15 and 8.16 show
cases in which the higher nitrogen coal gave lower values and in which the
difference in the emission levels was greater than that shown in Figs. 8.3
and 8.4
159
-------
750
500
250
Injector A
Position Throat; Throat
o
176 mm id; Quarl 25
water cooled; 5% Excess
a in 20% Primary Air
1.4
Fig. 8.15 The Influence of Coal Nitrogen Content on NO Formation
160
-------
1000
800
600
a
a
400
200
Injector A
Position Exit
Throat 176 mm id
Quart 25 water cooled
5% Excess air
30% Primary Air
O 1.1 % N Coal
1.8 % N Coal
1
0.6 0.8 1.0 1.2
S
0 0.2 0.4
Fig. 8.16 The Influence of Coal Nitrogen Content on NO Formation
1.4
161
-------
REFERENCES
8.1 Chigier, N.A. and Beer, J.M.
I Basic Eng. (Trans ASME) p, 797, 1964
8.2 Sternling, C.V. and Wendt, J.O.C.
Environmental Protection Agency, Office of Air Programs
Contract EHS-D-71-45 Final Report August 1972
8.3 Field, M.A. et al
Combustion of Pulverized Coal
B.C.U.R.A. , 1967
8.4 Mulcahy, M.
Paper presented at the I.F.R.F., 2nd-Members Conference, May 1971
162
-------
9. CONCLUSIONS
The test program demonstrated that burner variables can significantly
affect NO formation in natural gas and pulverized fuel flames and provided
insight into the causative phenomena.
9.1 Natural Gas Flames
The degree of swirl was the most significant burner variable exercised
in the natural gas test program. In simple jet flames, swirl was found to
reduce emission of NO by increasing entrainment, thus, resulting in dilution
of the reactants and a more rapid cooling of the products. For swirl levels
large enough to produce internal recirculation (Type I flames), emissions were
found to increase with increasing swirl, apparently because fuel was entrained
and reacted at the base, thus reducing reactant dilution. In Type II flames
the effect of swirl was dependent on burner geometry. With a divergent
injector, increased swirl resulted in decreased emissions because of increased
cooling rate for the product. The effect of swirl was less pronounced with a
radial injector.
Other burner variables which had significant effects on NO production
were air velocity, burner divergent angle, and fuel injector position. Increased
air velocity tended to decrease emissions in all cases. The other two factors
had varying effects depending on swirl level and exact geometry.
9-2 Pulverized Fuel Flames
The test results supported the theory that the most significant source of
emissions in p.f. flames is the fuel NO produced in combustion of the volatile
fractions of the coal. A review of the test data produced the following
generalizations:
1. Low coal concentrations in the primary stream during the initial
stages of heat release promote nitric oxide formation.
2. High mixing rates between fuel and air promote nitric oxide formation.
3. Any tendency to spread the coal jet promotes nitric oxide formation.
4. High primary velocities tend to reduce nitric oxide formation.
5. The influence of throat velocity is not well-defined.
6. For every burner condition there is an optimum primary air per-
centage. In many instances this optimum appears to be around
20 percent of the stoichiometric requirement.
163
-------
7, Ignition stability at the injector can reduce emissions with axial fuel
injection.
8. Maximum emissions are higher with higher fuel nitrogen contents.
However, minimum emissions appear to be independent of coal type.
Caution must be applied with all these generalizations since there are many excep-
tions .
It can seen that many of the conditions that tend to provide the high in-
tensity combustion and rapid burnout desired for practical systems also produce
high NO emissions. Subsequent work on p.f. burner designs (described in
Volumes II and III of this report) was oriented toward maintaining the desired
flame characteristics while reducing the initial rate of fuel/air mixing.
164
-------
TECHNICAL REPORT DATA
iPIeese read Iituntctions on die reverse before completing)
. REPORT NO.
EPA-600/2-76-061a
2.
3. RECIPIENT'S ACCESSION NO.
4. TITL5 AND SUBTITLE
Burner Criteria for NOx Control; Volume I. Influence
of Burner Variables on NOx in Pulverized _Coal Flames
5. REPORT OATS
March 1976
6. PERFORMING ORGANIZATION CODE
7. AUTHORS
p> Heap f To Me Lowes f R< walmsley ,
H. BartelUs , and P. Le Vaguer ese
3. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND AOORESS
International Flame Research Foundation
IJmuiden, Holland
10. PROGRAM ELEMENT NO.
1AB014; ROAP 21ADG-040
11. CONTRACT/GRANT NO.
68-02-0202
12. SPONSORING AGENCY NAME ANO ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPS OF REPORT ANO PERIOD COVERED
Phase Final:.6/71-6/74
14. SPONSORING AGENCY COOS
EPA-ORD
^.SUPPLEMENTARYNOTES pr0ject officer for this report is G.B.Martin, Mail Drop 65,
Ext 2235.
16. ABSTRACT
The report gives results of the first phase of an investigation to specify
burner design criteria to control NOx in natural gas and pulverized coal flames. The
two parameters found to have major influence on NO formation were the method of
fuel injection and the degree of swirl. NO formation can be controlled by optimizing
burner design parameters because its rate of formation depends on the detailed mix-
ing history of the fuel, combustion air, and recirculating combustion products. The
same parameters also dictate such flame characteristics as stability, length, and
luminosity. An explanation of the influence of burner parameters on pulverized coal
flames is based on two assumptions: the most significant factor of the total emission
is fuel NO, and the emission variation depends on the fate of the volatile nitrogen
compounds. Fuel NO formation can be reduced by ensuring that the volatile nitrogen
compounds react under oxygen deficient conditions. Maximum emissions occur with
radial fuel injects because the coal is rapidly mixed with the total air supply and
hot recirculating products. These conditions ensure early stable ignition. However,
fuel/air mixing promotes NO formation. Conversely, NO formation can be restricted
by maintaining the fuel in a coherent axial jet and discouraging primary/secondary
by surrounding the fuel jet with an ignition front. The coal must also be
JL1 "-.e minimum amount nf primary air _ ^_
" ^ff
mixing by surrou
delivered with th«
17.
KEY WORDS ANO DOCUMENT ANALYSIS
DESCRIPTORS
b.lOENTIFIERS/OPEN ENDED TERMS
c. COSATI Fi-iid/Group
Air Pollution
Combustion
Flames
Nitrogen Oxides
Burners
Design
Coal
Natural Gas
Fuel Injectors
Swirling
Air Pollution Control
Stationary Sources
13B
21B
07B
ISA
14A
2 ID
13H
3. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
171
20 SECURITY CL-\<5.~ tTi-l ir.fr
Unciassmeti
EPA Form 2220-1 (9-73)
155
------- |