EPA-600/2-76-274
October 1976
Environmental Protection Technology Series
ANALYSIS OF TEST DATA FOR
NOX CONTROL IN
COAL-FIRED UTILITY BOILERS
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangie Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development,
U.S. EDvironmental 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.
2.
3.
4.
s.
Environmental Health Effects Research
Environmental Protection Technology
Ecological Research
Environmental Monitoring
Socioeconomic Environmental Studies
This r(!port has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed
to dev(~lop and demonstrate instrumentation, equipment and
methodology to repair or prevent environmental degradation from
point Clnd non-point sources of pollution. This work provides the
new or improved technology required for the control and treatment
of pol:.ution 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 policies of the Agency, nor
does mHntion of trade names or commercial products constitute endorse-
ment OJ' recommendation for use.
This dJcument is available to the public through the National
Techni:al Information Service, Springfield, Virginia 22161.
-------
EPA-600/2-76-274
October 1976
ANALYSIS OF TEST DATA
FOR NO CONTROL
x
IN COAL-FIRED UTILITY BOILERS
by
Owen W. Dyke ma
The Aerospace Corporation
Environment and Energy Conservation Division
Los Angeles . California 90009
Grant No. R803283-01
ROAPNo. 21ADG-089
Program Element No. 1AB014
EPA Project Officer: Robert E. Hall
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
-------
ABSTRACT
This report describes The Aerospace Corporation analyses
of a large quantity of emissions, operating conditions" and boiler configura-
tion data from full- scale, multiple-burner electric-generating boilers firing
coal fuel. It is a companion publication to a previous Aerospace report on
similar analyses of data from natural gas- and oil-fired utility boilers con-
ducted for the U. S. Environmental Protection Agency, Research Triangle
Park, North Carolina ("Analysis of Test Data for NO Control in Gas- and
x
Oil:Fired Utility Boilers, II EPA- 650/2-75-012, January 1975). Objectives
of this study, as of the previous study, include (1) evaluation of the effects
of combustion modifications on NO emissions, in fu,ndamental combustion
x ~
terms, and (2) evaluation of techniques for further reductions in NO emis-
x
sions. The report includes the following results pertaining to coal-fired
utility boilers: (1) discussion of the major sources of NO emissions;
x
(2) parametric investigations of the effects on NO emissions of two- stage
x
combustion, burners out of service, combustion air temperature, and excess
air reduction; (3) discussion of probable short- .and long.:.term hardware p.nd
operating condition modifications likely to yield further significant reductions
in NO emission in coal-fired boilers; and (4) general comparisons of NO
x x
reduction techniques in utility boilers firing coal, 'oil, and natural gas fuels.
A total of 186 tests conducted on eight utility boilers firing four significantly
different coal type s was used in the analysis. Boiler firing types included
single-wall, opposed, and tangential configurations.'
This report was submitted in fulfillment of Grant No.
R-803283-01 by The Aerospace Corporation under the sponsorship of the
Environmental Protection Agency.
iii
-------
CONTENTS
ABSTRACT. . . . . . . .
....... .... ..... ......... ... ...
ACKNOWLEDGMENT S
. .............. ........ ........
FOREWORD. . . .
........
..........
........
........
I.
EXECUTIVE SUMMARY
.....
.......
. . . . .
1.1
1.2
1.3
1.4
.....
Conclusions. . .
.........
............
.....
Recommendations
Introduction. . . .
..................
.......
........
. . . . .
.....
Summary. . .
........
. . . . .
......
. . . . .
II.
EFFECTS OF COMBUSTION MODIFICATIONS ON
NO EMISSIONS........... . . . . . . . . . .
x .
.......
2. 1
2. 2
2.3
Data Analysis Approach. . . . . . . . . . . . . . . . .
R e suIt s ............ . . . . . . . . . . . . . . . .
General Discussion of the Study of Coal-, Oil-,
and Natural Gas-Fired Data. . . . . . . . . .
REFERENCES.
APPENDIXES
A.
B.
C.
GLOSSARY
......... ...........
.........
.......
Mixing Zone Lengths in Tangential Boilers
........
Model for NOx from Conversion of Fuel-Bound
Nitrogen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Final Mixing Zone
.....
. . . . . .
.......
.......
.................. .... ... ...... ..... ...
v
iii
ix
Xl
1
1
4
5
8
11
13
34
66
69
71
75
83
91
-------
2.
3.
4.
7.
10.
11.
12.
13.
14.
FIGU RES
1.
Mixing Zone Model-Horizontal Section at Lowest
Bur:.:1er Level. . . . . . . . . . . . . . . . . . . . . . . .
. . . . .
Mixlng Zone Model-Vertical Section.
.........
Mixing Zone Model for Tangential Boilers-
Horlzontal Section at Lowest Burner Level.
. . . .
. . . . . . .
......,.....
Mixlng Zone Model-Definition of Burner
Configurations. . . . . . . . .,. . . . . . . . .
. . . . . . . .
5.
Varlation of Efficiency of Conversion of Fuel-Bound
Nitrogen to NO with Concentration of Nitrogen
. F' 1 x
In u e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.
NO Generated by Conversion of Fuel-Bound
N"tX
1 ragen. . . . . . . . . . . . ", . . . . . . . . . . . . . . .
Major Sources of Total NO Emissions....
x
.......
8.
Re salts of Parametric CalcuJ.ations on Effect of
Locating Air-Only Burners at Various Vertical
Levels in Burner Array. . . . . . . . . . . . . . . .
. . . . . .
......
,..,","
9.
Typical Single- Wall Boiler Firing Nominal Coal
Type with Typical Combustion Air Temperature
I""""
EffE:cts of Low Combustion Air Temperature in
Boner Firing Nom.inal Coal. . . . . . . . . . . . . .
EffE:cts of High Combustion Air Temperature and
High Fuel Nitrogen in Tangential Boiler. . . . . .
EffE:cts of High Moisture, Low Com.bustion
Tenlperature Rise Coal (Lignite) with Nominal
Conlbustion Air Temperature. . . . . . . . . . . .
EffE:cts of High Ash Coal with Moderate Combustion
Air Temperature. . . . . . . . . . . . . . . . . . . . . . . .
Parametric Calculations for Boiler Not in Data
Sarnple U sed to Derive Equation. . . . . . . . . . .
vi
"""'"
........
. . . . . CI
"""
I""""
14
15
16
18
21
23
37
40
43
44
4S
46
47
60
-------
15.
16.
2.
FIGURES (Continued)
Effects of Excess Air and Carbon Monoxide Levels on NO ..
x
Empirical Fit of NOx Generated in Final Mixing
Zone as a Function of Temperature in This Zone.
........
TABLES
1.
Summary of Total Data Sample. . . .
. . . . .
. . . . . .
. . . . . .
Analyses of Coal Types. . . . . .
..........
. . . . .
. . . . .
vii
63
87
31
33
-------
ACKNOWLEDGMENTS
Sincere appreciation is acknowledged for the guidance and
assistance provided by Mr. Robert E. Hall of the Combustion Research
Branch, Industrial Environmental Research Laboratory (Research Triangle
Park, N. C.), who was the U. S. Environmental Protection Agency Project
Officer during the conduct of this study. The study was conducted under
EPA Grant No. R- 803283-0 1.
A special acknowledgment is due Mr. A. R. Crawford of
Exxon Research and Engineering Company for his assistance in supplying
certain additional unreported data related to Exxon's field testing. Sincere
appreciation is expressed to the Tennessee Valley Authority, in particular
Dr. G. A. Hollinden and Mr. J. R. Crooks, for supplying and authorizing
the use of additional data from in-house te sting of the TV A Widows Creek
facility.
Acknowledgment "is also due Mrs. Sandra Barnes of The
Aerospace Corporation for her assistance in computer programming and
operation and in statistical interpretation of the data.
Cf /fAh<; It!
Owen W. Dykema,
Combustion Effects
Office of Stationary Systems
Approved by:
erome Rossoff, Direc
Office of Stationary Sy e
Environment and Energy
Conservation Division
ix
-------
FOREWORD
This study is a continuation of a previous Aerospace effort
reported in "Analysis of Test Data for NO Control in Gas- and Oil-Fired
x
Utility Boilers.' (EPA- 650/2-75-012, January 1975). The data analysis tech-.
nique developed in that program is applied in this study, with some modifi-
cation, to coal-fired utility boiler NO control test data.
x
The study reported herein was conducted for the U. S. Environ-
mental Protection Agency, Combustion Research Branch, Industrial Environ-
mental Research Laboratory, Research Triangle Park, North Carolina,
during the first year of a three-year continuing grant. (The previous study
of gas- and oil-fired utility boiler test data was conducted under a separate
EPA Grant No. R-802366 for this same EPA office.) The current study is
primarily concerned with NO control methods and does not specifically
x
address the possible limitations imposed on NO reduction by other undesir-
x
able side effects. However, the second year of the current grant includes
analysis of combustion and flame instability mechanisms and the effects of
combustion modifications, for the purpose of NO reduction, on plant effi-
x
ciency. The se are three side effects that can represent real, practicallimi-
tations on NO reduction.
x
A brief introduction and a summary of results of this study are
contained in Section I. Section II describes the analysis of NO reduction
x
techniques for coal-fired boilers, and includes a discussion of comparisons
of NO control techniques for coal-, oil-, and natural gas-fired utility boilers.
x .
The appendixes include greater detail on some of the more important aspects
of the study. More complete understanding of the analysis approach requires
reference to the report on the previous study mentioned above. An overall
view of the two years of study conducted by Aerospace on NO reduction in
x
utility boilers may be found in a paper by O. W. Dykema and R. E. Hall in
the proceedings of the EPA Symposium on Stationary Source Combustion,
EPA-600/2-76-152c.
Xl
-------
1.1
SECTION I
EXECUTIVE SUMMARY
CONCLUSIONS
Major conclusions from this study of oxide s of nitrogen (NO)
x
control data from coal-fired utility boilers, which appear well supported by
the data shown and discussed in Section II of this report, are listed below.
a.
Test data available to this study show that NO emissions
x
as low as 200 to 250 parts per million (ppm) have been
reached in full- scale coal-fired utility boilers by using
combustion modification technique s. Re sults of this study
indicate the potential for significant reductions beyond this
point by further application of simple combustion modifications.
However, this further reduction can only be demonstrated
through full- scale testing, during which any limiting side
effects would be evaluated (see Section 1. 2, Recommendations).
b.
Conver sion of fuel-bound nitrogen to NO appear s to be the
x
dominant source of NO emissions in coal-fired utility boilers.
x .
In the normal firing configuration, with all burners active,
this source appears to contribute 60 to 92 percent of the total
measured NO emissions.
x
c.
Reduction of NO derived from the fuel-bound nitrogen, at
x
least with the combustion modifications repre sented in the
data of this study, can only be accomplished by providing a
fuel- rich environment in the furnace regions where the initial
1
-------
hydrocarbon reactions are taking place. As a minimum, this
implies that some of the combustion air must be diverted away
from the active burners, the fuel flow increased in the active
burners, or both.
The main unknown in this combustion modi-
fication, as well as an area where practical, effective solu-
tions to NO from fuel-bound nitrogen conversion may be found,
x
is in the definition of where in the furnace and where relative
to the burning coal particle s the appropriate initial hydrocarbon
reactions are taking place. In this study, the as sumption was
made, and the data tend to indicate, that these appropriate
reactions are occurring over the full distance from the burner
exit to burnout of the coal particles and in the local product
gases rather than exclusively in a stoichiometric flame sur-
rounding the individual coal particle s.
d.
Biased-firing burner array configurations with the air-only
burner s concentrated in the top levels, or elevations, of the
burner array yield lower total NO emis sions (with coal fuels)
x
than configurations with the same number of air-only burners
located elsewhere in the array., Although it is difficult to sub-
stantiate, with available data, the separate effects of the se
configurations on conver sion of fuel- bound nitrogen and on
thermally generated NOx' it appears that the former configura-
tions always yield lower conversion of fuel-bound nitrogen
than the latter and probably yield equal or lower thermally
generated NO .
x
e.
A substantial portion (8 to 20 percent) of the total NO emlS-
x
sions generated in boilers where fairly large numbers of air-
only burners are concentrated in the top levels of the burner
array appears to be thermally generated in the final mixing
zone where the remaining excess air, from the air-only burners,
is mixed with the gases resulting from fuel-rich combustion in
2
-------
the active burner region. Reduction of NO from this final
x
mixing zone requires reduction of the average temperature in
the zone by reduction of combustion air (and fuel) temperature,
by use of a coal with a low combustion temperature rise, or
by increased cooling of the combustion gas enroute to this final
mixing zone. (The remaining 0 to 20 percent of the total NO
x
not discussed in conclusions band e is thermally generated
in the active burner region. )
f.
Reduction in the overall boiler excess air, with most burner
array configurations, results in reduction of NOx emissions.
A significant reduction in NO with all burner array configura-
x
tions, however, appears to result if the overall boiler excess
air is reduced until carbon monoxide emis sions become large.
There appears to be a direct effect of carbon monoxide in the
flue gase s on NO .
x
g.
A general comparison of conclusions of this study of NO con-
x
trol data from coal-fired utility boilers with those reported in
the previous study of data from oil- and natural gas -fired
boilers indicates that the same general combustion modifica-
tion techniques for NO reduction, involving the use of NO
x x
ports, burners-out-of-service and combustion air temperature
control, apply with all three fuels.
Specific hardware and
operating conditions to achieve minimum NO emissions with
x
each of the se fuels depend primarily on the concentration of
chemically bound nitrogen in the fuel and on the physical state
of the fuel.
Section 2.3.
The se specific difference s are discussed in
3
-------
1.2
RECOMMENDA TIONS
Resul~s of this study indicate that full- sca.le testing cL:uld be
conducted, not only to verify the trends in NO reduction indicated by this
x
study but also to evaluate any limiting side effects. Such te sting should in-
clude the following combin.edcombustion modifications:
, b.
a.
Reduce overall boiler excess air until carbon monoxide emis-
sions becorrie large and limits because of excessive carbon
monoxide or smoke are encountered.
Provide a separate path into the boiler for more than 25 per-
cent of the combustion air,' preferably through NO ports
x
located as high in the boiler as possible, but alternatively
through, or supplemented by, air-only burners concentrated
in the top levels (one or more) of the burner array.
c.
Reduce the temperature in the final mixing zone. The only
directly applicable method of reducing this temperature (in a
given boiler firing a fixed coal type) within the data of this
study is to reduce the combustion air temperature. Simply
reducing the heat transfer efficiency of the air preheater, of
course, can result in significant efficiency losses. Some
means (such as an additional economizer stage) must be pro-
vided to transfer more heat from the flue gases to the steam
cycle prior to entering the preheater.
The above recommendations for full- scale te sting addres s
what appear to be the major variable s that affect NO emis sions from coal-
. x
fired boilers in terms of the combustion modifications present in the data
sample ana:.yzed. To the degree that the effects of the se combustion modifi-
cations are properly interpreted, other modifications can be postulated that
would be expected to accomplish the same result but have not been demon-
strated, at least within the data sample of this study.
4
-------
1.3
INTRODUCTION
Requirements for the reduction of NO emissions from
x
large utility boile rs were established at a time when only general guide-
lines concerning the desired combustion conditions for minimum NO genera-
x
tion were available from laboratory research. Methods of operating a boiler
or necessary hardware modifications to provide those combustion conditions
in a full- scale, multiple - burner boiler were not clearly established. Analyti-
cal and experimental research in this area continues today. As is often the
case in rapid technology development, the hardware and operating sides of
the industry were required to achieve certain goals using limited guidelines
supplemented by the conventional, effective method of "cut-and-try." A vital
part of the iterative research and development process is the feedback to re-
search of the results of this full-scale testing, both to provide evaluation of
the initial guideline s developed as well as to provide a new source of informa-
tion to guide further research and development.
The problem of de scribing and controlling the combined aero-
dynamics, reaction, and heat transfer within the reaction section of a full-
scale combustor is highly complex and involves a large number of independent
variables.
A reasonable analysis of full- scale test re sults, then, requires a
fairly large number of tests in which all of the significant variables are varied,
even though the significance of the variables cannot be easily assessed before
the results are analyzed. The large number of variables and the resulting
large number of tests required for analyses dictate the use of a computer, at
least until the major variables can be identified and the analysis simplified.
A widely used method of analysis of large data samples is
called multiple regression analysis. This is very similar to the least squares
fit technique often used to develop simple relations between single dependent
and independent variables from data except that in this case the computer pro-
gram is used to least squares fit a single dependent variable simultaneously
to a large number of independent variables.
5
-------
A powerful and easily used form of this analytical tool for the
study of NC' control results if the independent variables that are thought to
x
significcmtly affect NOx emissions are grouped, in roughly quantitative ways,
to define a new set of independent, composite variables each of which affect
NOx in a linear way. A single, linear equation for NOx can then be written,
and the siIY~pler linear regres sion analysis technique. can be used. This is
particularl:r necessary in studying NOx control because thermally generated
NO is kno'Nn. to be ~ higbly nonlinear function of-local air-fuel ratios and is
x
exponential in. temperatu~e. Th~ advantage of such an approach to NO data
x
analysis is that most of the unknown quantities in the reL~.tions of NO to the
x
large number of fundamental variables can be lumped into the coefficients.of
these grouped, linear terms, and these coefficients then can be determined
empirically in the linear regression analysis of the data.
This approach, then, was the one taken in this analysis of test
data for NO control in utility boiler s. A rough model of mixing in large,
x
multiple- bl.rner utility boilers was constructed to provide estimates of the
air-fuel ratios and temperatures in a large number of mixing zones. NO
x
emissions i~enerated in each type of zone were then calculated, using a
Zeldovich-type of NO formation rate equation for thermally generated NO
x x
and a unique model for conver sion of fuel-bound nitrogen to NO. The re-
x
sulting equa.tion consisted of eight terms describing thermal NO generated
x
in eight types of mixing zones in the boiler, one term to calculate the NO
x
generated from conversion of fuel-bound nitrogen, and a constant (necessary
for the linear regression analysis). Thelinear sum of these ten terms repre-
sents the NO
x
emissions from the boiler.
The calculation of the values of each of these terms (except the
constant) fClr each test condition is complex and requires a corn.puter prograrn.
By means clf input of .relatively detailed information on boiler geornetry,
burner array configurations, and operating conditions, however, any nmnber
of fundamental variables, singly or in combination, can be investigated para-
metrically. In this manner, the entire approach can be evaluated (by cor).1-
parison of data with the parametric calculations), the n1.ajor variables
6
-------
determined, and general approaches to NO reduction investigated. The
x
most useful end results, then, are the general, major considerations, sub-
stantiated by data, which appear to lead to significant reduction of NO emis-
x
sions. Such general conclusions can be used to guide development testing.
7
-------
1.4
SUMMAR Y
Most of the construction of the rough mixing model, d",rivation
of the terr.1.S representing thermally generated NO , and initial regression
x
analyses and parametric analyses were conducted in a previous study of
natural gas- and oil-fired utility boilers.
1
these fuel:; are reported elsewhere.
In order to apply that same technique to coal-fired utility
boilers in the subject study, it was necessary to (1) modify the mixing model
to include tangentially configured boiler s, (2) expand the model of the con-
version of fuel-bound nitrogen to NO , and (3) expand the interpretation of
x
NO thernLally generated in a final mixing zone (for cases where the final
x
excess air is added to the products of fuel-rich combustion in the active
burner region through NO ports or air-only burners concentrated in the top
x
levels of h.e burner array).
In general, results of this study indicate rather specific
methods 0:£ reducing NO. Nothing in this study of NO reduction techniques
. x x
alone indicates fundamental phenomena that would prevent reduction of NO
x
emissions to zero. This latter conclusion in itself, however, is not suffi-
cient to draw the broader conclusions that reduction of NO to zero is pos-
x
sible, pra,::tical, or even feasible. A number of other practical limits such
as flame instability, boiler wall corrosion, boiler operational instability,
plant effic:..ency losses, or excessive emission of other air pollutants can and
probably will set limits on NO reduction.
x
As this study depends on empirical correlations, extrapola-
tions beyo;J.d the existing data involve some degree of risk. In the practical
sense, thi, means that other phenomena that affect NO emissions, but are
x
currently unknown or are not taken into account in the model of this study,
may beconle important in the hardware configurations and operating regimes
beyond existing data and could also limit further NO reduction. Because
x
all analysE s of NO reduction methods depend on empiricism, however, a
x
risk in extrapolation is always present in these analyses.
This work and results related to
8
-------
The advantages of this type of study, however, are that results
are expressed directly in terms of real, full-scale, multiple-burner utility
boilers and that these results are substantiated directly by data from these
boilers. Thus, they can be used directly to guide full- scale developmental
testing leading to maximum reduction of NO emissions and to evaluate re-
x
sults of such te sting relative to previous full- scale te st data. Limits un-
covered by such full- scale testing can then be superimposed on the NO reduc-
x
tion data to identify rapidly the real, practical problems limiting NO and
x
the real, practical NO minima.
x
It seems very important to establish where conversion of fuel-
bound nitrogen occurs spatially in relation to the burning coal particles and
clouds of particles and in relation to the burnout cycle of the particles and
clouds. This information can be used to establish burner design and coal
particle size s and size distributions such that the initial portions of coal com-
bustion involving fuel-bound nitrogen conversion occur in locally fuel-rich
environments, with the remainder occurring under conditions designed to
enhance complete hydrocarbon reactions.
Other modifications, which might substantially reduce thermal
NO in the final mixing zone by reducing the temperature in this zone, could
x
include water spray in the combustion air or recirculation of cooled flue gases
in the combustion air. Because cooling or diluting the combustion air could
have an adverse effect on stability of a very fuel-rich flame, it might be pos-
sible (and necessary) to introduce such cooling or dilution only in the com-
bustion air introduced through the NO ports or the air-only burners. NO
x x
emissions generated in this final mixing zone might also be reduced by de-
creasing the mixing time through increasing the rate of mixing of the final-
stage air with the products of combustion coming from the active burnoer
region. Any or all of these additional combustion modifications may be
nece s sary to achieve significant NO reductions in the light of practicallimi-
x
tations on the simpler approaches, discussed in Section 1.2.
9
-------
This study of NO reduction in coal-fired utility boilers
x
clearly indicate s the significance. of the fuel-bound nitrogen as a source of
NO emis s ions. If the model for conversion of fuel-bound nitrogen developed
x
in this study is reasonably accurate, however, then the total NO emis sion
x
problem can be separated into that as sodated with this source and that asso-
ciated with thermally generated NO . Results of both this study and the
x
previous one indicate that the two mechanisms appear to be relatively inde-
pendent (i. e., thermally generated NO does not appear to be a strong func-
x
tion of the NO generated from conversion of fuel-bound nitrogen and vice
x
versa). R(~sults of this study of coal-fired data, then, can be interpreted in
terms of thermally generated NO alone (as with natural gas fuels) or in
x
terms of fuels with much lower (than coal) concentrations of nitrogen chemi-
cally bound in the fuel (as with low nitrogen oil fuels). When the results of
this study of coal-fired test data are viewed in this light, they sub stantiate
the concluE ions reached in the previous study of natural gas - and oil-fired
utility boil(~r te st data.
10
-------
SECTION II
EFFECTS OF COMB USTION MODIFICATIONS
ON NO EMISSIONS
x
The purpose of conducting this study on large quantities of
NO emis sions data from operational, multiple-burner utility boiler s is to
x
gain some useful insight into the combustion processes occurring in boilers
as they affect NO emissions. Although a single equation containing only ten
x
apparently linear terms is developed and used in parametric calculations in
this study, this equation is not a simple one and is not intended for ready use
. by any interested party. A relatively complex computer program is neces-
sary to calculate the appropriate values of each of these ten terms, for each
te st condition and hardware configuration, from a number of independent
variables, and through a number of highly nonlinear functions. The purpose
of this study, instead, is to use this complex equation and the large sample
of data to define the major variables and processes affecting NO emissions,
x
evaluate the effects of these variables and processes on NO emissions, and
x
use the insight gained in this manner to develop simpler rationale to explain
existing data. This simpler rationale can then be used to guide developmental
testing toward even further reduction of NO emissions.
x
Most of the major assumptions, approaches, and analyses
used to develop the equation used in this study are discussed in the report on
the previous analysis of natural gas- and oil-fired data. 1 Those major modi-
fications and improvements necessary to apply that analysis technique to data
from coal-fixed utility boilers are discussed in this report. These two re-
ports, then, are considered to contain sufficient information on the approach
11
-------
used here that similar models and equations could be developed by any agency
desiring to do so.
The coal-fired utility boiler data used in this study were
selected to represent the major firing types (i. e., single-wall, horizontally
opposed, an:l tangential) and a range of coal types, including a nominal type
and types representing the available extremes of high nitrogen, high ash, and
high moisture content. Throughout this study, an attempt was made to develop
an equation that would adequately correlate all of the coal-fired data in a
single samp'le. With all of the diversity in firing and coal types represented
in the samp] e, it was hoped that the resulting observations and conclusions
might have hroad applicability to other boilers and coal types not in the data
sample. As a single verification case, data from one particularly complex
boiler and coal configuration were held out of the data sample used to develop
the parametric equation.
This section, therefore, contains discussion of (1) the modifi-
cations to the approach necessary to, apply the previous analysis to coal-fired
data, (2) ob:;;ervations from the correlations, (3) results of parametric studies
of the data, (4) re sults of analysis of the boiler not in the data sample used to
develop the equation, and (5) general conclusions.
12
-------
2. 1
DATA ANALYSIS APPROACH
1
The general approach taken in this and the previous study
was the following: (1) assemble a rough model of NO generation in a large
x
multiple- burner utility boiler, using gene rally accepted principles; (2) use
that model to generate a single equation, which, if all of our input knowledge
were correct and accurate, would directly predict NO emissions; (3) use
x
that equation to correlate the available data and correct for inadequacies in
the input assumptions; and (4) analyze the resulting correlation equation and
use that equation to parametrically analyze the data to gain such insight as
is pos sible relative to control of NO emis sions. This procedure is little
x
different from that used in the analyses in the previous study. 1 Exceptions
that required some modifications both to the model and to the approach re-
sulted from (1) the strong dominance of NO generated from conversion of
x
nitrogen chemically bound in coal fuels, l2) the wide range of peak combus-
tion temperatures repre sented by the range of coal types and combustion air
temperatures in the data sample, and (3) the explicit attempt to generate a
single equation applicable to all o! the coal data sample and with as broad
applicability as possible. Two major modifications resulting from these three
major differences are discus sed in Sections 2. 1. 2 and 2. 1. 3 below.
2. 1. 1
NO Model
-x
In the previous study, 1 all of the boilers in the data sample
were of the face-fired type, both single-wall and horizontally opposed. The
coal-fired data sample of this study includes. both of these firing types as well
as tangentially fired boilers. The tangentially fired configuration is particu-
larly important because more pulverized coal is fired in tangential boilers
than in any other utility boiler configuration.
Figure s 1 and 2 show schematics of the mixing zones model
constructed for the boilers of the earlier study1 and used, without change,
for the face-fired coal boiler types of this study. Figure 3 shows a horizontal
section of the model tangential boiler used in this study. Basically, the
13
-------
RECIRCULATION/ I I "
FRONT .(~// I I',. n. REAR
WALL ~ / / SECONDARY)__\ '" \V ALL
/ / I ~ I" '
._L__-, / I Z , r---'-- -
BURNER PRIMARY I / I ~ I , I
-1 / 13:1 ,L_-7--
._"~- // Ixl " ./
'>/ ADJACENT I z II ',,//
" Ie") /,
/ ,MIXING cr I /,
/ , Ie: / ,
. J- - -I' I ~ I /, - - - - -
I , I~ I / I
. -" - - ...J 'I ~ I // L - -: 7-
U''- "-J ~-I/ U/
, I I /A
, I I /
" 10 I /
, /
"
OPPOS ITE MIX I NG
LOWEST BURNER LEVEL
Figure 1.
Mixing zone model - horizontal section
at lowest burner level
14
-------
t
------
~-----
NO PO RT M I X I N G
x
I
NOx ~J~---
PORTS \
\
\
\
- - --~
___1\
--,\
--- \
---I \
---
\
\
\
\
\
\
\
f t
FGR
---I
BURNERS
---
==-1
PR IMARY
ZONES
I
BULK GASES
/
/
/
/
,,- - -
/---
/ 1---
I ---
/ I
/
I
/
/
/
/
/
1--
1--
---
I
Figure 2.
Mixing zone model - vertical section
15
I-~
-------
~.~ -- ---------,,-/ ~"{
~ \~....... (./ ~\~'r'- .//,
) "% \ ' SECONDARY ) ~ // / I
,'\ ~ ' '" ,,-/ 'v/ / I
\ ;t. .> '" , / ,,-
I v\ o~i~,s~~ .....'X ADJACENT / I
I \ ....../ \ MIXING / I
I >. \. ./ ..... ./ \ / I
0:: \,,- \ /
I ~.~ /(. \ / I
I 8 /' :AUS~KS \ / i';: I
L.J.J / \ \ / c::(
~ 0
I / \ />', @ I
I / \ ,,-// \ V) I
I / ADJACENT \././ \ I
/ MIX ING V/ OPPOSITE '>- "\
I //', MIXING (..0 \ I
I / "\ /"- " \ '%- ' I
" -y/ " '/': \
/ ",,,,,, ~~"{ > SECONDARY', , ~
v'" \~ ", . ~ ,
~~ ",/ ........., \
~:~----------~
Figure 3.
Mixing zone model for tangential boilers -
horizontal section at lowest burner level
16
-------
burners of the horizontally opposed model were simply moved into the corners,
with their centerlines aligned slightly off center to induce the swirling bulk
gas flow typical of a tangential boiler. With this swirling flow, it was assumed
that the recirculation flow does not recirculate to the burner directly above
but instead recirculates to mix with the flow from the burner that is located in
the next highest level but in the corner that. is next in the direction of swirl.
Figure 4 shows the assumed recirculation paths of both the face-fired and
tangential boilers. The most significant result of this change is the much
longer cooling time for the recirculation gases in the tangential configuration.
The other significant change in the model considered necessary
to adequately describe tangential boilers results from the vertical distribution
of primary and secondary combustion air in each corner array of burners. In
a face-fired boiler, each of the burners appear to be identical to every other
burner, and each fuel-and-primary air port is surrounded by its proportionate
share of secondary air. With all burners active, each burner unit, consisting
of fuel plus primary air plus secondary air, is operating at the overall boiler
air-fuel ratio. Similary, all of the bulk gases are at the overall boiler air-
fuel ratio. The vertical burner arrangement in a tangential boiler, however,
appears to be such that if there are n fuel-pIus-primary air burner levels, or
vertical fuel levels, there are n + 1 secondary air port levels. Each active
burner level, then, consists of a 1/ n fraction of fuel plus primary air but
only a 1/(n+1) fraction of secondary air. Thus, all of the burner units and
all of the bulk gases in the region of the active burners would be operating at
an air-fuel ratio less than that of the overall boiler until the final (top) second-
ary air port flow is added to the total. This operation is analogous to opera-
tion with NO ports in a face- fired unit. In this model, for tangential boilers,
x .
then, the top level of secondary air ports is treated as a fixed NO port level,
x
admitting the remaining 1/(n+1) fraction of the secondary air. Details of the
mixing zone lengths, cooling times, and NO port flows used in the model for
x
tangential boilers are discussed in Appendix A.
17
-------
TOP REAR':'
tG)@ : 0) 0
~ t;;;\ oppas ITE (;;;\ ~DJACEN! t;;;\
t~ ~.. I ~~J .~
r- . BELOW
t~ 0 : 0 G
t~ I
» I
(./)
o t~ 0 I
RECIRCULATION C f:\::1 ~ f:\ I
FLOWS ~ T 0:J I
(Face-Fired) C 8 ~ 0 i
I
FRONP'
o
o
o
.-
00
':' Vi ewed from the outs ide
o 0
~
(j) 0
~
o RECI RCU- 8
LATION FLOWS
(Tangential)
Figure 4.
Mixing zone model - definition of burner configurations
.
-------
Conversion of Fuel-Bound Nitrogen
The previous study1 dealt with natural gas and low nitrogen
oil fuels. The conversion of fuel- bound nitrogen was of no importance with
the natural gas fuel and of little importance with the oil fuel. As a result,
the majority of the modeling and analysis effort was devoted to the generation
of NO by thermal mechanisms. 'Initially, the equation developed for the
x
regression analyses contained as many as 22 terms describing thermal NO
x
generation and only one term accounting for NO generated from conversion
x
of fuel-bound nitrogen. The latter term represented a simple, constant
fraction of conversion of the fuel- bound nitrogen, regardless of the burner
2.1.2
operating conditions.
Regression analyses of the oil-fired data, however,
established a final equation that behaved as though the efficiency of conversion
of the fuel-bound nitrogen was approximately a linear function of the burner
air-fuel ratio, with the conversion efficiency ranging from about 63 percent
at excess air levels represented by three percent oxygen approximately to
zero under burner air-fuel ratio. conditions involving about 70 percent of
theoretical (stoichiometric) air; For the purposes of understanding the
control of NO from low nitrogen oil-fired boilers, that representation of the
x
conversion of fuel- bound nitrogen appeared adequate.
In coal-fired boilers, however, fuel- b,ound nitrogen concen-
trations are so much larger that NO derived from this source dominates the
x
It was necessary, therefore, to improve the description
total NO emis s ions.
x
or model of fuel- bound nitrogen conversion. Observations from the analyses
of the oil- fired data served as a starting point for this improvement, amplified
by the research literature. Basically, many observers2- 6 have noted that
(1) the conversion efficiency of fuel- bound nitrogen to NO at a fixed fraction
x
of excess air is some inverse function of the weight fraction of nitrogen
chemically bound in the fuel, (2) the efficiency of this conversion with a given
coal is directly proportional to the local air-fuel ratio, (3) the rate of con-
version of this nitrogen is of the order of the hydrocarbon-air reactioI1s
themselves, and (4) the conversion efficiency appears to be a weak function
(if at all) of local gas temperatures. These observations were used to
19
-------
postulate a ;,imple model of the 'conversion process in which (1) the conversion
efficiency h; inversely propo'rtional (empirically to the two-thirds power) to
the weight fraction of nitrogen in the fuel (ash-and-moisture-free); (2),:the
conversion ,~fficiency is linearly proportional to the local air-fuel. ratio, with
zero NO occurring at an air- fuel ratio where there is just sufficient oxygen
x '
present in the air to oxidize the carbon to carbon monoxide and the hydrogen
(other than 'i;hat already bound to oxygen in the moisture) to water; (3) the
appropriate local air':"fuel ra~io is the average of that in the regions where
the initial hydrocarbon reactions are taking place; and (4) the conversion
efficiency in independent of the local' temperature.
Figure 5 shows a plot of data available to this study from full-
scale tests :md some laboratory experiments relating conversion efficiencies
at three percent oxygen to weight fraction of nitrogen in the fuel. As the full-
scale data r.:lay involve some thermally generated NO , the empirical curve
x
shoWn in the figure was established to represent a lower bound of the full-
scale data but represents the actual data in the case of the controlled te sts
conduded in the laboratory. Much more data of this type are available, 2
but not all of the appropriate test conditions were available to this study. In
general, thE~se additional data confirm a curve of the type shown.
As the mechanism( s) involved in the conversion of fuel- bound
nitrogen iS1.ot well known, the functional relationship between conversion
efficiency a'1.d local air-fuel is also not well known. It appears, however,
that the conversion process represents some sort of exothermic oxidation
reactions tbat occur at about the same time and at about the same rate as the
normal hydrocarbon oxidation reactions. If such is the case, nitrogen oxida-
tion l'TlUst compete with carbon and hydrogen oxidation reactions for the avail-
able oxygen. Indications are that the conversion efficiency goes to zero
approximately when the available oxygen is not sufficient to oxidize the carbon
to carbon monoxide and the hydrogen (that which is not already bound to
oxygen in tbe moisture in the coal) to water. 1, 2,3 In the simplest sense, this
might be thought to imply,that the nitrogen oxidation reactions can reasonably
compete with carbon monoxide for the availabl~ oxygen but not at all with
hydrogen and free carbon.
20
-------
100
90
80
..... 70
~ 60
u
a:::
~ 50
G~ 40
z
lJ.J
u
N
-
tt 30
lJ.J
z 25
o
l/)
ffi 20
>
z
o
u 15
Q)~
o
o DYKEMA, Ref 1
o CRAWFORD, Ref 7
6 TURNER, Ref 5
o CATO, Ref 8
3% 02
o
o
6
e = 22.6 (WFBN2) -2/3 /
1~. 1 O. 15 O. 2 O. 3 O. 4 O. 6 O. 8 1. 0 1. 5 2. 0 3. 0
WFBN2, WEIGHT PERCENT OF NITROGEN IN ASH- AND MOISTURE-FREE FUEL
Figure 5.
Variation of efficiency of conversion of fuel-bound nitrogen to
NO with concentration of nitrogen in fuel
x
-------
~:a"cKihg be~n~r'd'efiriitiop;a'simp1e conversion"efficiency model
was initially p.6s~tulated in which the oxygen available for oxidation of fue'l-
. . ~ ' , .
bound nitrogen ':Yas 't~at which is left over after all of the available hydrogen
is oxidized to w~ter ~l1d all of the carbon at least to carbon monoxide. It was
further as~umed th~t, for: a. given fuel, a fixed fraction of this r~rnaitling
. R '. .' . . .
oxygen oxiC.ized the: fuel-hound nitrogen to NO, with the rerriainde:t;further
'I . ," .',
oxidizing H.e ~a~bon monoiide to car?on "dioxide. The fraction of ,oxygen
involved in ()xidizing the fuel':';Pound nitrogen to NO was ~stablished such that
the converE ion effidency at three percent oxygen was that given by the empir-
ical expression (and curve) of Fig}:lre 5 for the specific weight concentration
. .
of nitrogen in the ash- and moisture-free fuel.
The resulting model indicates a linear relation, for a given
fuel, between fu.el- bound nitrogen conversion efficiency and the local air-fuel
ratio and, ::or excess air fixed at three percent oxygen, an inverse relation
(to the two- thirds power) between conversion efficiency and the concentration
of nitrogen chetnically bound in the fuel. The conversion efficiency, in this
model, is 2 ero when the fraction of theoretical air is such that all of the
oxygen in the air (or more) is necessary to oxidize the carbon to carbon
monoxide and the available hydrogen to water. Figure 6 shows calculations
from this nlodel for the four coal types of this study and the low nitrogen oil
of Referenc e 1. While the model is largely empirical and the theoretical
considerations involved are sketchy, it will be seen later in this report that
the available data tend to show good agreement. There has not appeared to
be any significant reason to reexamine this model. The calculations involved
in the modE I are described in more detail in Appendix B.
The problem of calculating the efficiency of conversio~ of
fuel- bound nitrogen to NOx from the above model, as used in this study,.
would now he qu.ite simple and straightforward if the coal and oil gasific~tion
and the gas mixing rates in the active burner region were well known. In that
case, the average 10.cat air-fuel~atio~in th~ regio~s where the initial hydro-
carbon reactions were taking place could be accurately determined and the
. average conversion efficiency; ca~culated.As deyelopment of these rates was
22
-------
500
400
N
W
N
o
~
('t"\
~ 300
-
e::::
o
E
~ 200
x
o
z
100
o
0.5
5TO ICHIOMETR I C
COAL TYPE
4
2
3
1
O. 6 O. 7 O. 8 O. 9 1. 0 1. 1
FRACTION OF THEORETICAL AIR IN THE ACTIVE BURNER REGION
1.2
Figure 6. NO generated by conversion of fuel-bound nitrogen
. x
-------
not within the scope of this study, an approximate method was develop~d at
least to take into c.CCOUrlt the effects of the slow gasification rates of coal
particles on this conversion efficiency.
Observations on coal flames in boilers of various firing types
indicate thc.t luminous, reacting flames appear to extend 15 to 20 feet into
the boiler from the burners. If the firing configuration is such that gross
mixing between burner flows and bulk gases is forced to occur within that
distance, then the average local air-fuel ratio in the region where the initial
hydrocarbon reactions are taking place (including some further distance for
gas mixing) would approximately be the average of all of the burner flows and
the bulk ga:,es at that level. If such mixing does not occur within that distance,
then the appropriate average would be that of the burner flows alone. Both the
opposed and tangential firing configurations are designed to force such mixing
early. The opposed, coal-fired boiler in this study allows a maximum of
about 14 feot for mixing within the burner flows before forced mixing with
opposed bUJ~ner flows begins. Certainly mixing with bulk gases begins well
before that, at least in the higher burner levels. The tangential configuration
is designed to induce swirling bulk gases and more direct and early mixing of
bulk gases with burner flows and between the burner flows.
For the opposed and tangential boiler configurations, then, the
appropriate average air-fuel ratio for calculation of the fuel-bound nitrogen
conversion efficiency at a given burner vertical level was taken to be the
average of all of the burner flows introduced into the boiler up to and including
that burner vertical level. The appropriate air-fuel ratio for all of the coal
introduced :.nto the boiler (the average for the region of the active burners)
was then taken to be the average of all of the levels where fuel is being intro-
duced. This averaging process can be described as follows:
AFRP
AFRB
= t [NBFAi ( t1 NBTi )]
'-1 NBFAT i
1- L NBFA.
i =1' 1 .
( 1)
24
-------
where
AFRB
= air-fuel ratio of the active burners
AFRP
= average air-fuel ratio in the active burner region
NBFA.
1
= number of active burners in the i burner level
NBFAT = total number of active burners in the burner array
NBT.
1
= total number of burners in the i burner level
n
= number of burner levels
If all of the active burners, and none of the air-only burners, are .located
below some vertical level (for example, all burners in the top row air-only
and all burners below the top row active), then the average air-fuel ratio of
the active burner region (AFRP) is equal to that of the active burners (AFRB).
In the case where air-only burners are mixed with active burners, the aver-
age air-fuel ratio is always higher than that of the burners. As the efficiency
of conver sion of fuel- bound nitrogen has been observed to inc rease with the
local, effective air-fuel ratio, these higher average air-fuel ratios always
increase the NO generated from the fuel nitrogen. In general, this accounts
x
for the fact that, when air-only burners are mixed (vertically, horizontally,
or both) with active burners and gross forced mixing is early, some of the
gasified fuel will initially react in local regions of high air-fuel ratio and
generate higher levels of NO from fuel- bound nitrogen conversion. From
x .
the standpoint of NO generated from conversion of fuel- bound nitrogen alone,
x
this latter case implies that minimum NO requires that air-only burners be
x
located in the highest levels of the burner array.
Single wall-fired boilers, however, involve no direct me.chanism
for creating this gros s mixing~
In boilers with small numbers of burners
and limited vertical distribution of the burner array, the array begins to
resemble a single burner.
In a single burner configuration, there is little
or no mixing with bulk gases and, in fact, the transition from tlburner flowstl
to tlbulk gasestl is even difficult to define.
In the single wall-fired coal
25
-------
boilers of this study, the burner flows at the lowest level can proceed, with
little or no forced mixing with bulk gases, for 22 to 32 feet across the boiler.
In this study, therefore, the average air-fuel ratio in the initial reaction
period was taken as that of the active bu;rners, regardless of the locations of
air-only bUJ:ners in the burner array. Treating the conversion of fuel- bound
nitrogen in single-wall burners in this manner implies that any effect of the
verticalloc :l.tion of air-only burners in these boilers must result from effects
on thermally- generated NO .
x
For purposes of regression analysis of the coal data sample, tqe
NO contributed by conversion of fuel- bound nitrogen was calculated from th~
x
model and f:le mixing assumptions discussed above, and subtracted from the
measured v,llue of NO .
x
on that port:Lon of the NO assumed to be due to thermal mechanisms.
x
fication of this total approach will be discussed in Section 2.2.2. 1.
The regression analysis, then, was conducted only
Veri-
2.1.3
Final Mixing Zone
Throughout the gas and oil fuel studie s, it was evident that
there was a source of thermally generated NO within the boiler which was
x
essentially :.ndependent of variations in NO port air flow, burners out of
x
service, and excess air. This source was represented, at least partially, by
the constant in the equation resulting from the linear regression analyses. 1
In the previous study, 1 the two fuels were analyzed separately.
Thus, in each data sample used in the regression analyses, the fuel combus-
tion temperature rise and product species at given air-fuel ratios were con-
. ,
stant. Also, the combustion air (reactant) temperatures measured at rated
load in the boilers in those data samples were not' greatly different (31 K,
56°R difference). As a result, peak combustion product temperatures at full
rated load, in all of the data used for any regression analysis, were constant,
within the scatter of measured combustion air temperatures. The linear
regression analyses, then, were able to fit the data acceptably by assigning
the NOx gen'~rated in that zone to a single constant. Parametric analyses
discus sed el sewhere 1 concluded that the source of this thermal NO was in the
x
NOx port or final mixing zone but did not further identify other characteristics.
26
-------
The coal data sample, however, includes coal types with
widely varying combustion temperature rises and boilers involving widely
different full-load combustion air temperatures, resulting in peak combustion
temperatures varying over a range of more than 370 K (6700R). The linear
regression analyses on the coal data sample still attempted to fit the effects
of this temperature variation between coal and boiler types to a single con-
stant applicable to all of the coal and boiler types. Comparison of parametric
calculations for the individual boilers with the data from the boilers, using
the equation developed from regression analysis of the entire coal sample,
showed that the parametric calculation was clearly not accounting for varia-
tions in this constant with the individual coal and boiler types. Although this
constant, amounting to 146 ppm of NO , was present in all of the parametric
x
calculati<:ms, the constants represented in the data appeared to range from as
low as 33 ppm to well over 200 ppm for the various coal and boiler types. It
was concluded early in this study, then, that the zone where this NO is gen-
. x
erated was not adequately represented by any of the terms in the regression
analysis equation.
If the data from the individual boilers were analyzed separately,
new constants would be obtained appropriate to those firing conditions. Unfor-
tunately, most of the coal-fired boiler data samples are too small to obtain
meaningful results. In any case, such an approach would limit the generality
of the overall study results and ~imit the application of these results to those
boilers in the data sample analyzed, at least to the extent that 100 to 150 ppm
error could result from inadequate knowledge of this constant. It was also
concluded, therefore, that further study was necessary to develop an addi-
tional thermat NO term, or an empirical correction term, to account for the
x
variation of this constant for various coal and boiler types.
Details of the observations from available data, and the devel-
opment of this new term, are described in Appendix C. This study concluded
that this new term, independent of burner air-fuel ratio (constant), probably
represents the NO generated during the final, large-scale mixing of any
x
excess air not entering the boiler through the active burners (i. e., entering
27
-------
through air-only burners, NO ports, or both) with the products of fuel-rich
x .
combustio.:\ in the l.ctiv~ burn'er region.. This i~ a transient mixing region
that was not included in the original model and equatiop.. It does not exist
when the ,b'U.rner air':fuel ratio is greater than stoichiometric (as when all
burners arE! active). The new term is a .function only of the equilibrium
combustion temperature rise and the nitrogen and oxygen concentrations in
the product specie s at the stoichiometric air- fuel ratio, and the gas cooling
enroute to ihe location in the furnace where this mixing takes place.
This new term is the primary one, in the model and in the
equations used for parame,tric calculations, that shows the beneficial effects
of NO poris over air-only burners located in the top row of the burner array.
x .
It also shoVls the beneficial effects of locating such NO ports high in the
. x
boiler to m:!.ximize the cooling time before introducing this final excess air.
General confirmation of this term will be shown in Section 2.2, Results.
2.1.4
Coal- Fired Data Sample
Most of the data on coal-fired boilers used in this study were
obtained from the field testing of the Exxon Research and Engineering Com-
pany.7 Son1e additional, unreported data were obtained from the Tennessee
Valley Authority from tests on the Widows Creek No.5 boiler. Instrumenta-
tion and gaf sample analysis techniques used are discussed elsewhere. 7
A major problem usually encountered in using any test data
after the te 5ting has been completed is that not all of the data necessary for
a particula]' analysis technique or model were measured. In full- scale hard-
ware testin:~, it is also often impractical to measure the desired primary
data. As a result of these problems, much of the primary data necessary to
this study had to be calculated from other, secondary data.
Data on coal feed rates were not available. As a result, coal
feed rates were calculated from the measured boiler load, the lower heating
value of the coal type, and an assumption of constant plant efficiency. This
28
-------
IS not a very accurate calculation, particularly at low boiler loads, but was
considered adequate for this study for the following reasons:
a.
NO emissions do not appear to be a strong function of
x
the total throughput flow. Although it is widely recognized
that NO is, in many cases, a significant function of load,
x
this effect appears to be primarily a result of variations in
combustion air temperature with load rather than of the total
flow rate. Combustion air temperatures were measured.
b.
Although the plant efficiency in some cases decreases
significantly at low load, the primary emphasis of this
study was on reduction of NO at full rated load. Data
x
taken at low load were almost never used for any purpose
in this study. In any case, low load testing often involved
burner configurations where some of the air registers were
fully or partially closed. These data were eliminated from
the data sample for other reasons, as discussed below.
1
As in the previous study, overall boiler excess air data were
calculated from the ratio of the measured oxygen and carbon dioxide levels.
Theoretical stoichiometry was used to relate the exce s s air level to this ratio.
It is recognized that measured oxygen and carbon dioxide levels do not agree
accuratel y with theoretical stoichiometry but, without measured flow rates,
no method was available to calibrate this error.
As this error tends to be
constant for all of the data, trend results should be relatively accurate.
Combustion air temperatures are actually measured at the
1
exit of the air preheater, as in the previous study. These temperatures
were used in that and the subject study to represent temperatures of the
combustion air entering the burners from the windbox. No data were avail-
able in eithe r study with which an estimate could be made of the combustion
air temperature change occurring between these relatively widely separated
locations.
29
-------
Finally, a m9st important parameter to this study is the
distribution of combustion air between separate windboxes (if more than one)
and between the multiple active and air-only burners fed from a common
windbox. In the previous study, 1 it was found that the presence of a signifi-
cant fraction of combustion within a burner (such as a recirculation pilot zone
anchored within the burne.r) can significantly change the resistance to air flow
through the burner, particularly ~elative to this resistance in an air-only
burner. F-Hther, a typical windbox is not normally aerodynamically designed,
. .
and air flow maldistribution among the burners can be inherent. For lack of
appropriah: data, the combustion air flow distribution was assumed to be
equal to multiple windboxes. It was also assumed equal to all burners (active
and air-only) fed from a common windbox as long as all air registers were
in approximately the same position. A great deal of effort was expended,
particularly in the previous study, 1 to account for the changes in air flow
- distribution in configurations where some air registers were fully or partly
closed. These efforts were finally abandoned in this study and all such data
were deletE:d from the sample. It appears that air leakage even through a
so- called closed air register can be appreciable and can lead to significant
errors in calculated air flow distribution.
For comparison purposes, Table 1 shows the entire data
sample used in this study of coal-fired boilers as well as that previously
reported. 1 Data from a total of 186 tests in eight coal-fired boilers were
used in'thif; study. Boiler configurations represented single-wall, opposed,
and tangential firing. In addition, data from 12 tests on a second opposeq-
fired boiler were held out of the data sample used for regression analysis for
a later test of the generality of the resulting prediction equation.
Table 1 shows a total data sample of 575 test conditions with
coal, oil, and natural gas fuels. An additional 70 test conditions were also
reduced and entered into the program, which involved attempts to shut off the
combustion air, or secondary air, to some burners by closing the air regis-
ters. As discussed above, these tests were later deleted from the sample.
30
-------
w
-
Table 1. SUMMARY OF TOTAL DATA SAMPLE
FIR ING No. OF R ATE D No. OF NOx No. OF TESTS
TYPE BOI LERS LOAD, M'vV BURNERS PORTS COAL OIL NA T GAS
SINGLE- 1 105 18 NO 3 - -
WALL 2 125 16 NO 72
- -
2 180 16 NO - '2!} 27
2 240 12 NO - 43 31
1 260 16 NO 18 - -
1 340 16 NO 14 - -
OPPOSED 2 240 12 YES - 6 47
2 350 24 YES - 61 145
1 800 54 NO 17 - -
TANGENTIAL 1 330 20 NO 22 - -
1 360 20 NO 40 - -
TOTAL: 16 TOT ALS: 186 139 250
GRAND TOTAL: 575
-------
Overall, tr..e effect of deleting these tests, representing only 11 percent of
the total, was miLor.
Analyses of the coals burned in the boilers of this study are
reported elsewhere. 7 In order to minimize the work load in ca1c~lating
equilibriunl combustion temperatures and product species,
types fired were reduced to four, more general coal types,
of these synthesized coals are shown in Table 2.
the seven coal
The analyses
32
-------
w
w
Table 2. ANALYSES OF COAL TYPES
HIGHER
HEATI NG
ULTIMATE ANALYSIS, % DRY MOl STURE, % VALUE,
COAL TYPE C H 0 N S ASH (P roxi mate) B t u lIb
1-H I G H .63.6 4.5 19.2 0.90 O. 70 11. 1 32.0 7, 200
MOl STURE
2-H I GH 58.8 . 4.7 9.3 1.3 0.90 25.0 10.5 8,900
ASH.
3- 70.5 4. 7 9.3 1.3 3. 1 11. 1 10.5 11, 700
NOM I NAL
4-H I GH 72.3 4. 7 9.3 1.7 1.7 11.1 10.5 11, 700
NITROGEN
-------
2.2
RESULTS
The. primary observation of all of the data analyses and
parametric calculations involving coal-fired data is the predominance of
the NO geo.erated from conversion of the fuel-bound nitrogen. With the
x
coals of thi s study, with all burners active and operating at excess air
corre sponC.ing to three percent oxygen (116 percent of theoretical air), this
source contributes 390 to 450 ppm of NO (see Figure 6), representing 60
x
to 92 percEnt of the total NO emissions measured under the se operating
x
conditions in the boile rs of this study. Thus, the model de rived for fuel-
bound nitrc,gen conversion is the most significant part of this analysis of
coal-fired data. The model indicates that NO from fuel-bound nitrogen
x
conversion is a function of the weight fraction of nitrogen chemically bound
in the ash- and moisture-free fuel, and the effective fraction of excess air.
Therefore, that model can be confirmed by plotting the data for a given coal
and boiler type as a function of the average effective fraction of. excess
air in the active burner region, as defined in Section 2.2.2 and
Equation (1).
A second significant source of NO in coal-fired boilers
x
appears to be that thermally generated in the final mixing zone, as defined
and discuss ed in Section 2. 1. 3 and Appendix C. Calculations from Figure 16
(Appendix C) show that this zone can be the source of more than 170 ppm of
NO , repre senting from eight to more than 20 percent of the total NO emis-
x x
sions, in tbe boilers of this study. (The remaining zero to 20 percent not
"
resulting from conversion of fuel-bound nitrogen or from this final mixing zone
represents NO thermally generated in the active burner region.) NO from
x x
this zone i~ independent of the fraction of theoretical air in the burne rs, and
a function c,£ peak combustion tempe rature s (at stoichiometric in the appro-
priate zone). It will appea r as a relatively constant level of NO , at burne r
x
air-fuel ratios les s than stoichiometric, for a given coal type and boile r,
added to th(~ NO derived from conversion of fuel-bound nitrogen. This final
x
mixing zonE~ exists only in boiler configurations in which relatively large
fractions of air are added to the products of fuel-rich combustion in the
34
-------
active burne r region through air- only burners concentrated in the top levels
of the burner array or through NO ports (or both). This zone will not exist
x .
in boiler configurations in which (1) all burners are active, (2) the air-fuel
ratio in the actiye burners is greater than stoichiometric (regardless of
the presence or location of air-only burners), and (3) air-only burners are
located low in the active burner array. Data confirmation of the model of
NO from this zone results from the variation of this relatively constant
x
level of NO added to that of the bound nitrogen line.
x
Confirmation of the prediction of thermal NO generated in
x
the complex gasification, mixing, and reaction region of the active burners,
however, is not easy to show. The parametric equation requires eleven
terms involving highly nonlinear Zeldovich NO formation rate terms that
x
represent functions of as much as 11 different air-fuel ratios and as many
more local temperatures to adequately describe the thermal NO generated
x
in the zones in this region. These terms generally reflect the effects of
burner air-fuel ratio and peak combustion temperatures but are also com-
plex functions of a large number of other variables. It was for this reason
that the data in the previous oil and gas study1 were analyzed parametri-
cally, as functions of a number of single variables, with no attempt to show
all of the data as a function of just one or two va riable s.
In this study, the primary concern is with significant reduc-
tion of NO from coal-fired boilers, in which NO emissions are primarily
x x
functions of two major variables: the fraction of excess air in the active
burner region and the peak combustion product temperature downstream
(above) this region. Change s to the se two va riable s to reduce NO from
x
conversion of fuel- bound nitrogen and from the final mixing zone also tend to
reduce thermal NO from the active burner region as well. As thermal
x
NO from the activE: burner region already represents a relatively small
x
fraction of the total in coal-fired boilers and should totally disappear if
significant reductions are made in NO from the major sources, it will
x
generally be neglected in the following data presentations. All of the data
35
-------
from each significant coal and boiler type will be shown as a function of the
average f:'"acti0n l/f "'xce8s ~~r in the active burner regio!1. It must ~'" kept
in mind thclt, although thermal NO generated in the active burner region
. .x
is some fW1ctioI?- of. ~his si~gle parameter, it is also a complex function of
other para}neters. As a result, that po!tion of the total NOx thought to be
generated :.n that source may require further observations with respect to
. . , .
the variati')n of other variables. A rough indication, of the thermal NO
, x,
generated i.n the active bu~ner region will be shown by a parametric calcu-
lation for r..ormal operation (al~ burners activ~).
I . ':, . '.
2.2. 1 ,
Data Interpretation
Figure 7 shows the data from the sister boilers, Widows,
Creek Units No.5 and 6 of the Tennessee Valley Authority, plotted against
. .
the single parameter, fraction of theoretical air in the active burner region
(100 percent of theoretical air represents the stoichiometric air-fuel ratio
for that coal). As the se boilers are of the single-wall configuration, accord-
ing to the discussion in Section 2.1. 2 the fraction of theoretical air in the
active burner region is the same as the fraction of theoretical air in the
burners. Superimposed on this data are three schematic-like lines or
curves showing the three major sources of NO discussed above. The line
x
shown, describing NO from fuel-bound nitrogen conversion, is taken directly
x
from Figure 6. For this particular boiler, because it is so nominal, the
constant determined directly from regression analysis' of the total data
sample is a.ppropriate. This constant, then, repre senting NO from the
x
final mixing zone, is shown added to the fuel-bound nitrogen line. Finally,
thermal NO generated in the active burner region is bounded by adding to
x. ,
. the sum of the first two sources a typical Zeldovich curve, for an arbitrary
constant tin1.e.
It can be seen that a schematic such as shown in Figure 7 is
almost en01:lgn to exptain the ~ajority of the sig~ifica~t variations in NO
, " ,,' .,' " ' " ", ," x
emissions from coal-fired boilers. The lowe r limit of all of the data is
36
-------
800
W
-J
N
o 600
~
('t"\
I-
-
~ 400
E
0.
0.
x
~ 200
DATA
125 MW SINGLE-WALL BOILER
16 BURNERS, NOMI NAL COAL
Q)
o NO (COs 500 ppm) 0 0 ~
6 NOX (CO >500 ppm) 0 0 ,&~~ ~,~
X . 12~~O 0 ~
THERMAL NOx ~o 0 6 ~o~.,
FROM FINAL R 0 ~I
MIXING ZONE 0 6 6~.,
1 - ~"66 ~
~ ~., FUEL-BOUND
~ ., NITROGEN
~,~ CONVERSION
'C-
o
0.6
THERMAL NOx
FROM ACTIVE
BURNER REGION
o
o. 7 O. 8 O. 9 1. 0 1. 1 1. 2 1. 3
FRACTION OF THEORETICAL AIR IN THE ACTIVE BURNER REGION
1.4
Figure 7. Major sources .of total NO emissions
x
-------
reasonably close to, but not below, the fUel-bound nitrogen conversion line.
, '
The slope of the lower limit of the data generally tends to confirm the slope
of the bound nitrogen line.
The second, parallel line generally correlates the lower limit
of NO under full-load and nominal excess air conditions and a variety of
x '
air-only bt.rner configurations. Those configurations with the air-only
, ,
burners in the top rows lie close to this line. Of the data below this line,
80 percent exhibited exc~~sively high. carbon monoxide emissions (greater
than 500 ppm), represent~d ,tests at le'ss than 80 percent of rated load, or
both.
Finally, the data above the second line represent the therm,al
NO generated in the active burner region and are reasonably bounded by a
x I ,
Zeldovich curve, added to the sum of the first two line s. It can be seen that
at least the upper bound lof the data is related to the fraction ~f theoretical
air in the burners but other variables are also effective. For example, all
of the data points shoWn in Figure 7 above 600 ppm repre sent configurations
.
either with all burners active (above 115 percent of theoretical air) or with
two air-only burners located in the bottom row of the burner array.
In the case shown in Figure 7, because it is a single-wall
configuratbn and the fraction of theoretical air in the active burner region
, ,
is the same as that in the 'burners, the Zeldovich curve and the data
describing thermal NO generated in the active burner region are roughly
x
centered aJ.'ound the stoichiometric air-fuel ratio (100 percent theoretical
air), as ex:pected. If these boilers were of the opposed or tangentially fired
configurati:ms, however, the 'average fraction of theoretical air in the
active 1;>urner region, in the cases where air-only burners were located in
the bottom level of the burner array, would be higher than, the fraction in
the burner!! [see Equation (1)). The data shown in Figure 7 representing
thermal NO gene,r,ated in tb.e.a.ctive purner region, and the bounding curve,
x ". , ...,d ,... , ",' .. .
would be sJ;:ewed toward higher levels.of the fraction of theoretical air in
the active burner region, and .the Zeldovich curve would not be centered
. I " . " . ...
3R
-------
about stoichiometric. This illustrates the point that the average fraction
of theoretical air in the active burner region is not the only relevant
parameter with re spect to NO thermally generated in the active burner
. x
region. NOx da.ta plotted against this paramete r alone, particularly data
from opposed and tangentially fired boilers, can be expected to show some
high levels of thermal NO that require further study, if it is desirable to
x
explain these high levels.
The data shown in Figure 8 indicate that total NO emissions
x
are higher when air-only burners are located in the bottom level of the
.
burner array than when they. are located in the top level. Figure 8 shows
the re suits of parametric calculations on the effects on NO of the vertical
x
level of a full horizontal row of air-only burners, for all of the firing types.
This figure confirms that, in all cases, NO emissions are lowest when the
x
air-only burners are locate~ in the top level. Also, the discussions in
Section 2. 1. 2 as well as Equation (1) indicate that when the air-only burners
are concentrated in the top rows the re is no air introduced into the region
of the active burners (below the top rows) other than that normally entering
through the active burne rs. The average fraction of theoretical air in the
active burner region, therefore, is equal to that in each active burner.
Thus, because there is no apparent reason not to locate the air-only burners
in the top row, it is assumed at this point that only that case is of interest.
If this assumption is made early, most of the questions and complications
involved in the thermal NO generated in the active burner region can be
x
dismis sed or avoided.
The remaining data and effects on NO of interest from coal-
x
fired boilers, then, are those related to the two straight lines shown in
Figure 7 and to the methods of reducing the NO from the final mixing zone
x
and from conversion of bo~nd nitrogen, all in boiler configurations in which
air-only burners are located in the top levels of the burner array.
39
-------
2..0
...J ...J
UJ UJ .
~ ~ 1.8
>- C-
Z 0
« l-
I- I-
~ ~ 1.6
e::: e:::
UJ UJ
Z Z
e::: e:::
=> =>
ca ca
~ ~
z z
o 0
I I
e::: e:::
- -
« «
~ x 1. 2
Xo
~ z
1.4
,
\\
\ \~ TANGENTIAL
\ \'
\
\
\
\ --...........
~ , ,
/,' ,
OPPOSED' ,
~......
C-
o
I-
1.0
...J
UJ
>
~
o 0.2 0.4 0.6 0.8 1.0
FRACTION OF VERTICAL HEIGHT OF BURNER ARRAY
Figure 8.
Results of parametric calculations on effect of .
locating air-only burners at various vertical
levels in burner array
40
-------
2.2.2
Parametric Calculations
In line with the discussion in Section 2.2.1, it is considered
that only boiler configurations in which the air-only burners are located in
the top levels of the burner arrays are of interest here. All other burner
configurations in the same boiler will yield the same or higher NO levels.
x
The effects of. the number of air-only burners cap be shown, and confirmed
by data, directly on a plot of NO ve rsus the fra'ction of theoretical air in
x
the active burner region. The effects of peak combustion temperatures on
the thermal NO generated in the final mixing zone can be shown by para-
x
metric variations in this parameter on a single boiler and verified by the
data variations between boi-Iers. Because NO reduction by load reduction
x
is not an acceptable means of controlling NO from utility boilers, only
x
NO emissions under full rated load conditions are of interest here.
x
Thus, the four parametric calculations of interest in each
of the significant combinations of coal and boiler types, under full rated
load conditions are (1) a reference or nominal calculation with all burne rs
active and with variations in overall boiler excess air, (2) increasing num-
bers of air-only burners located in the top levels of the burner array,
(3) the same calculation as in (2) but with the combustion air temperature
reduced (arbitrarily to 422 K, 3000 F), and (4) the line desc ribing the NO
x
generated from the conversion of fuel-bound nitrogen.
In all cases, the parametric calculations and all of the data
available for each boiler are plotted against the fraction of theoretical air
in the active burner region. This paramete.r is appropriate for all of the
parametric calculations and most of the data because the fraction of theoreti-
cal air in the active burner region is the same as that in the burners (a uni-
form air-fuel ratio in the a'ctive burner region equal to that in the burners).
All of the data are plotted because some verification can be obtained from
most of the data. Finally, attempts to select only the data closely represent-
ing one of the parametric calculations can reduce the applicable data sample
nearly to zero.
41
-------
Figures 9 through 13 show such plots 6f parametric
calculation:, and data for the significant combinations of coal and boiler
types in the data sample used for regression analysis. Significant observa-
tions and confirmation by data will be discus sed separately for (1) the con-
version of fuel-bound nitrogen, (2) NO generated in the final mixing zone,
x
and (3) othe r significant observations.
2.2.2.1
Fuel-Bound Nitrogen
According to the discus sion in Section 2.1.2, the model
developed b calculate NO generated from conversion of fuel- bound nitrogen
x
should predict a lower limit of NO at any fraction of theoretical air for
x
each of the coal type s. This semiempirical model was not a part of the
regression a.nalyses because the NO calculated from this model was sub-
x
tracted from. the measured NO before the remainder was analyzed. The
x
data can se:rve to verify the magnitude of the NO calculated from the model
x
by closely a.pproaching the calculated value but not going below it. The
linearity, the slope, .and (by extrapolation) the intercept of the calculated
line as a fWlction of the fraction of theoretical air in the activ~ burne r region
can be verified by comparing the calculated line with the lower limit of the
data.
With the exception of the data de scribing the tests with the
high ash coal, Figure 13, the lower limit of all of the coal-fired data tends
to closely approach the fuel-bound nitrogen conversion curve. In Figures 10
and 12, SOITle data even lie right on or slightly below this line. Within the
,
accuracy of the data it can be concluded that (1) no data are available that are
significantly below the calculated fuel- bound nitrogen conve r sion line, and
(2) the data are sufficiently close to the calculated line that no general change
to the mode:~ (or the empirical curve of Figure 5) to increase the magnitude
of the NO from this source is warranted.
x
Figure 13, showing the results from firings with the high ash
coal (type 2), appears to show considerable deviation from the calculated
fuel-bound nitrogen conversion line. None of this data come closer to this
42
-------
800
ON 600
~
('t"\
I-
«
>-
~ ~ 400
w 0
E
c..
c..
><
0 200
z
125 MW SINGLE-WALL - 16 BURNERS - NOM I NAL COAL
OAD 3 0 DATA:
FULL L . % 2. ALL BURNERS ACTIVE ALL AVA I LABLE DATA
606 K (630°F) COMB AI R TEMP
PARAMETRIC CALCULATIONS: 0 NOx (CO ~ 500 ppm)
CD VARIABLE EXCESS AIR ~ NOx (CO> 500 ppm)
Q) 0-8 AI R-ONLY BURNERS IN TOP LEVELS 0
CD SAME AS (1) EXCEPT 422 K 0 Q) ,3~
(300°F) COMB AIR TEMP ~--_-o-._-~ 0 .....-.....-'.0
\.!J 0 """&- -- 9>--
o ~ 0 0 0-. 0 0 0 0 .,.,.,.,.,
o CD-' 6. _ft'O ,.'
~ ....."'" ~ - .,.,.
o ."" ~ .,..--
6. p"'" .,.,.'
0' ("'j ~..... .'."'.1
\!::..J l:::.A ~)II>"'" . ' . ... . ...
~..... .'
..... ~ .'
..... .'
.....\:Q)"'" ,.,.,., FUEL-BOUND
."."'" ..,.,. NITROGEN
....
,.....,. 3 CONVERS ION
,.
,.
,.
,.
,.
.,.
o
0.6
Figure 9.
O. 7 O. 8 O. 9 1. 0 1. 1 1. 2
FRACTION OF THEORETICAL AIR TO THE ACTIVE BURNERS
1.3
Typical single-wall boiler firing nominal coal type with typical
combustion air temperature
-------
ROO r-
ON 600
~
C'I:\
l-
e:(
>-
~ 400
~.
~
E
c.
c.
oX
~ 200 ~
360 MW TANGENTIAL - 20 BURNERS - NOMINAL COAL
428 K (310VF) COMB AI R TEMP
CD V A R I ABLE EX CE S S A I R
CV 0-8 AI R-ONLY BURNERS I N TOP LEVELS
o
0.6
o
O. 7 O. 8 O. 9 1. 0 1. 1 1. 2
FRACTION OF THEORETICAL AIR IN THE ACTIVE BURNER REGION
.....J
1.3
Figure 10.
Effects of low combustion air temperature in boiler
firing nominal coal. .
-------
ON 600
tfi!.
C'f'
......
«
>-
c::: 400
o
~ E
c.
U1 c.
><
0 200
z
800
o
330 MW TANGENTIAL - 20 BURNERS - HIGH NITROGEN COAL
667 K (740°F) COMB AI R TEMP
CD VARIABLE EXCESS AIR
o 0-8 AIR-ONLY BURNERS IN TOP LEVELS CD
CD SAME AS 0 EXCEPT 422 K 0 <2 -<' /""
(300°F) COMB AIR TEMP O~ 0 ."'"
- ~ ."'"
~ .-'
o NO (CO:S 200 ppm) ~ 0 ..............
x . :;;.....-.- "".. . .... .
,~ .....
I::J. NO (CO> 200 ppm) ~,...... .....t"".""
x .-' ..........
J"t!"~ .....
A~ ,"""u ..........
..f££;GJ ..........,......."" FUEL -BOUND
,.,.,.., ~ ~........... NITROGEN
2 ,......' ................ CONVERS ION
.-' .....
GJ'~ ..........
3 ."'"
.....
.....
.....
....
.....
O. 6 O. 7 O. 8 O. 9 1. 0 1. 1 1. 2
FRACTION OF THEORETICAL AIR IN THE ACTIVE BURNER REGION
Figure 11.
Effects of high combustion air temperature and high
fuel nitrogen in tangential boiler
-------
~
0'
ON 600
~
C't'
I-
«
>-
~. 400
E
c..
c..
x
o
z- 200
onn
OLiU -
105 MW SINGLE-WAll - 18 BURNERS - HIGH MOISTURE COAL (Lignite)
550 K (530°F) (est) COMB AI R TEMP
CD VAR IABLE EXCESS AI R
~
@ 0-8 AIR-ONLY BURNERS IN TOP LEVELS
G) SAME AS CV EXCE PT 422 K
(300°F) COMB A I R TEMP
o
0.6
O. 7 O. 8 O. 9 1. 0 . 1. 1
FRACTION OF THEORETICAL AIR TO THE ACTIVE BURNERS
1.2
Figure 12.
Effects of high moisture, low combustion temperature rise coal
(ligAite) with nominal combustion air temperature
-------
800 MW OPPOSED - 54 BURNERS - HIGH ASH COAL
574 K (574°F) COMB AI R TEMP
CD VARIABLE EXCESS AIR
Q) 0-24 AIR-ONLY BURNERS IN TOP LEVELS
Q) SAME AS (g) EXCEPT 422 K
(300°F) COMB A I R TEM pO............... CD
00 ~
ONOx (CO ~ 200 ppm) ~ 0 ~
b. NO (CO> 200 ppm) 0 -- --- ... ..............
x fii-. -- ... ... ..."""""""
/ -",-' ~
",,-,~,
..,., " -' " ~ I
(g) -' .,., -' , ~ FUEL-BOUND
200 Q)-''''''''' ~ NITROGEN
~" CONVERSION
1000
800
~
-.J
N
a
~
('f'\
I--
«
>-
e:::
o
E
~ 400
600
x
a
z
o
O. 7
o .0
o
@
fJ
0.8 0.9 1.0 1.1 1.2 1.3
FRACTION OF THEORETICAL AI R IN THE ACTIVE BURNER REGION
1.4
Figure 13.
Effects of high ash coal with moderate combustion
air temperature
-------
line than about 160 ppm, even though high ca rbon monoxide levels are shown
in some tc~ts and some data-represent operation at less than 75 percent of
rated load. Unfortunately, data from other boilers firing this high ash
coal were not available to this study. This boiler was also by far the largest
in the data sample and involved the most complex and unsymmetrical burner
array configuration, with a large number of burners (54) arranged such that
some burners are opposed and some are not. Thus, there were insufficient
data to determine if the apparent disagreement with the magnitude of fuel-
bound nitro,~en conversion calculation is real and, if so, whether the dis-
agreement results from the coal type or the large and complex burner
c onfigu ra tiCin .
In gene ral, most of the data also tend to confirm the calcu-
lated slope of the fuel-bound nitrogen conversion line. The lower limit of
the data shclwn in Figure s 9 and 11 shows good agreement in slope.
Figure 10 might be interpreted to indicate c;t somewhat shallower slope, but
Figures 11 a.nd 13 might indicate that a somewhat steeper slope is appro-
priate. Although not shown here, the lower limit of all of the oil-fired data
from the previous. study also tends to confirm the magnitude and slope of
the fuel-bound nitrogen conversion line. There appears to be no reason,
from the data of this study, then, to make any gene ral, empirical change to
the slope oJ the fuel-bound nitrogen conversion line calculated from the
model.
To the extent that this linear relation can be assumed to hold
in extrapolations to zero NO , the general agreement in both magnitude and
x
slope tends also to verify this intercept. The model discussed in Section
2.1.2 and dl~veloped in Appendix B, then, appears to be rather well substan-
tiated by, thE data.
2.2.2.2
Final Mixin~ Zone
Verification of the existence of the final mixing zone, as
discussed in Section 2.1.3, and of the appropriate levels of NO generated
x
48
-------
in the final mixing zone can be generally obtained from four main
observations:
a.
b.
This zone should exist only in firing configurations
where air-only burners or NO ports (or both) are
x
concentrated above the active burners.
In firing configurations such as described in case a above,
where there are few air-only burners, small NO ports,
x
or ove rall boiler exce s s air is ve ry high, the
air-fuel ratio in the burners may be above stoichio-
metric. In this case, while a zone of final mixing of
exce s s air with the products of combustion in the active
burner region would physically exist, these combustion
products would not be fuel-rich. As a result, the air-
fuel ratio in this mixing zone would not pass through
stoichiometric and the "final mixing zone, " as defined
in Section 2.1.3, would not exist. The thermal
NO generated under such conditions could be higher or
x
lower than that generated when the burner ai r-fuel ratios
are Ie s s than stoichiometric, depending primarily on
the temperature in this final mixing zone. With firing
configurations such as described in case a, operated under
conditions whe re the burne r air-fuel ratios are below
and above stoichiometric, the data may show a step
change in thermal NO levels near stoichiometric. This
x
step change, if it is observed in the data, may be the
mo s t di re ct evidence of the exi stence of this final mixing
zone.
c.
Burner configurations without NO ports and with all
x
burner s active should show the rmal NO levels highe r
x
or lower than those involving this final mixing zone,
49
-------
depend~ng on the temperature in this zone. Because the
mixing patte rns in the active burner region in this case
are not the same as those of case b, thermal NO
x
levels may not be the same as those in case b. In either
ca'se b or c, the parametric calculation for the appro-
priate configuration should account for these different
the rmal NO levels.
x
d.
Burner configurations with air-only burners distributed
in the lower levels should show NO levels highe r than
x
those with the air-only burners in the top levels. It
should be recalled, however, that none of the parametric
calculations shown in the subsequent figures are intended
to duplicate this case.
Figure 9 shows data from a boiler firing coal with a relatively
nominal combustion temperature rise and involving a relatively nominal full-
load combu:otion air temperature. This boiler has no NO ports. The
x
parametric curve shown in the figure as no. 1 represents the firing configu-
ration where all burners are active and only the overall boiler excess air
was varied. It is recognized that in the actual boiler testing this excess air
was never I'educed below 105 percent of the theoretical (stoichiometric) air.
The paramE tric calculation below this level of exce ss air is intended only
to approxirr..ate some of the burner configurations with air-only burners
located in lower levels of the burner array. The parametric calculation~
labeled nos. 2 and 3 represent the firing configuration with varying num-
bers of air- only burners located in the top levels of the burner array.
The se latte:~ two calculations do involve a final mixing zone. The calcula-
tion labeled no. 3 represents the NO reduction that would be expected if
x
the combustion air temperature were reduced to 422 K (300°F) in the firing
configuraticn no. 2. As such a temperature reduction was not accomplished
on any sing]e boiler, confirmation of the effect of this change can only be
estimated f:~om the effects of different combustion air temperatures in
diffe rent bo iler s.
50
-------
The data shown in Figure 9 generally tend to confirm the
parametric calculations. More detailed uefinition of the data (than shown
in the figure) indicates that the measured NO levels with air-only burners
x
concentrated in ,the top levels of the burner array tend to group around
curve no~ 2. All of the data grouped around curve no. 1 represent configu-
rations either with air-only burners located in the lowest levels of the
burner array (less than about 112 percent of theoretical air in the burners)
or with all burners active. The difference between the constant amount of
NO thermally generated in the final mixing zone and that thermally
x
generated in the active burner region with all burne rs active (no final mixing
zone) is small. (Curve no. 2 is about the same level above the fuel-bound
nitrogen line as is curve no. 1 for excess air levels above about 115 percent
of theor~tical air.) Reduction in the combustion air temperature by 182 K
(3270 F) is predicted to result in reduction of thermal NO by about 70 ppm
x
over the burner air-fuel ratio range between 85 and 115 percent of theoreti-
cal air. Actual full-load combustion air temperatures (604 K, 6270 F) over
this range show about 125 ppm of thermal NO .
x
Figure 10 shows the effect of reduction in combustion air
temperature of 176 K (317°F) in a boiler firing the same coal type as that
fired in the previous boiler (Figure 9). The relatively constant level of
NO added to that from fuel-bound nitrogen conversion in firing configura-
x
tions involving the final mixing zone is about 65 ppm, a reduction of about
60 ppm over the previous case.
The parametric calculations shown in Figure 10 indicate that
curve no. 1, however, was not significantly affected by the combustion air'
temperature reduction. As discussed in Section 2.1.1, because of the way
the secondary air is introduced in a tangential boiler, this firing type is
treated, in this study, as a case effectively involving fixed NO ports.
x
Also, in all of the burner configurations te sted in the two tangential boiler s
of this study that involved air-only burners, the air-only burners were
located only in the top level of the burner array. Thus, all of the data and
51
-------
the param'~tric calculations shown in Figure 10, and later in Figure 11,
repres~nt c()nfi~ur3.tions where air-only burners, NOx por.ts, or both, arp-
concentrated above the active burners. The final mixing zone, as defined
in Section 2. 1. ~, howev.e r, would exist only.in those case s whe re the frac-
tion of the ::>retical air in the active burner region (the same .as in the burners,
in all of these configurations) was less than stoichiometric. Thus, with.
the se' tang~ntial boiler data, a' step transition between curV.es nos. 1 and 2
couldbe expected to appear in the data around stoichiometric.. Above this
level, the a.greement ispoo;r. All.of the data shown above stoichiometric
excess air resulted from firing configurations with all burners active. As
such, they should be grouped about parametric curve no. 1. A step change
of about 70 ppm would be expected.around stoichi,)metric excess air, in a
transition from curves nos. 1 and 2. Instead, most of the data show no
step and, in fact, indicate that the thermal NO drops off to zero at very
x
high leveln of excess air. This apparent trend in this boiler is unexplained.
It will be !ieen, however, that this trend is not evidenced in the data from
the other tangential boiler in this study.
Comparison of Figures 9 and 10 appears to indicate the
reduction In the NOx from. the final mixing zone expected from the reduction
in combustion air temperature. There is also a difference in firing type
between tbese two boilers, however, which could have caused the observed
reduction. This question can be reasonably resolved by comparison of
Figures 10 and 11. These data both represent tangential boilers of approxi-
mately the same design and size. There is a small difference in the coal
type. The boiler of Figure 10 was fired with. the nominal coal type 3, which
contained 1.3 percent by weight (of the total coal weight) chemically bound
nitrogen, while the coal fired in the tangential boiler of Figure 11 contained
1.7 percent: nitrogen (type 4). This difference should be accounted for by the
calculation of NOx generated from convers.ion of the fuel-bound nitrogen.
The temperature rise due to combustion at stoichiometric conditions is very
nearly the same for these two coal types (only 13 K, 23° F, difference).
52
-------
Curve no. 2 shown in Figure 11 indicates that the thermal
NO , in this case, is between about 150 to 190 ppm. This is commensurate
x
with the higher combustion air temperature of this boiler compared to the
other tangential configuration (240 K, 4320 F, higher). Both of the para-
met,ric calculations shown in Figure 11 show considerably greater thermal
NO generated in the active burner region over the air-fuel ratio range
x
between about 85 and 105 percent of theoretical air. At the normal NO
x
levels, it appears that incorporating just a few air-only burners, located
in the top level, into the burner array actually increases the thermal NOx
and partially offsets the decrease in NO from fuel nitrogen. Incorporating
x
four or more air-only burners in the top row, however, appears to reduce
NO from both the thermal and fuel nitrogen sources, and rather steep re-
x
ductions in overall NO emissions result.
x
The data shown in Figure 11 represent just two main firing
configurations: (1) those representing four air-only burners concentrated
in the top level of the burner array, shown grouped around 81 percent of
theore~ical air; and (2) those with all burners active, grouped around 107
percent of theoretical air. The first data group should be represented by
curve no. 2 and the second by curve no. 1. The data around 81 percent of
theoretical air show NO levels well (50 to 115 ppm) below curve no. 2
x
when excessive carbon monoxide levels are present, as expected, but
the tests with acceptable carbon monoxide levels are also, on the average,
lower than curve no. 2 by 64 ppm.
The data grouped a round 107 percent of theoretical air show
good agreement with parametric curve no. '1. Thus, this tangential boiler
does not show the tendency for the thermal NO to approach zero at high
x
levels of excess air that was indicated in Figure 10.
The data shown in Figure 11, although limited in range,
appear to indicate that parametric curve no. 2 perhaps should be more like
curve no. 3 and that a step transition from curve no. 2 to curve no. 1 may
exist around stoichiometric excess air.
53
-------
The effect of combustion air temperature on NO is
x
postulated bere to result from variations in the peak combustion temperature
in the final mixing zone., These temperatures result from (1) an initial
reactant teD,perature established primarily by the combustion air tempera-
ture; (2) a temperature rise from this level due to, and inherent in, the
combustion of the coal; and (3) cooling of the, combustion products enroute
to the region of intere'st in the-furnace. Thus far, the data shown have
concentrated on the' 'effect of combustion air temperature in boilers firing
coals with essentially the same combustion temperature rise (at' the same
air-fuel raHo).
Coal type 1 (see Table 2) isa high moisture (lignite) coal.
The high moisture content results in a lower heating value and a lower tem-
perat\lre ri:>e due to combustion. The temperature rise, due to combustion
, ,
of coal type 1, at stoichiometric air-fuel ratio, is calculated to be 260 to
270 K (470 1:0 4900 F) lower than those of the coal types fired in the boilers
discus sed thus far (coal types 3 and 4). Parametric calculations for a
single-wall boiler firing this type of coal are shown in Figure 12.
The
parametric calculations show no indication of significant thermal NO gen-
x
erated in ei':her the active burner region or the final mixing zone when any
number of air-only burners are located in the top levels of the burner array
(curves nos. 2 and 3). With all burner s active, however, the parametric
calculations (curve no. 1) indicate significant thermal NO (100 to 130 ppm)
x
The data available for verification of theseca1culations
are very liI':1ited. Of 14 tests for which data were recorded, 7 11 tests we're
deleted frOD'l the data sample because some secondary air registers were
fully or partially closed during these te sts. As di'scus sed in Section 2. 1.4,
it was considered impossible to determine the distribution of combustion
air in the se cas'e s with accuracy sufficient for this study.
The two dat'a points shown in Figure 12 on the line repre-
from cdnversionof fuel-bound nitrogen'resulted from configu-
air-only burners concentrah';d above' the activ'e burri'e'rs. .
senting NO
x
rations with
54
-------
The appropriate parametric calculation shown in the figure is curve no. 2.
These two firing configurations showed excessive carbon monoxide emis-
sions (near 1000 ppm) and NO levels, as expected, slightly (30 to 45 ppm)
. x
below curve no., 2. The remaining useful test involved air-only burners
located iri the lowest levels of the burner array and, therefore, measured
NO levels are not related to any parametric calculation shown in the figure.
x
It is interesting to note, however, that, despite the very low combustion
temperature rise of this coal, large increases in NO levels can result with
x
air-only burne rs located low in the lower levels of the burner array.
This may indicate greater mixing of the burner flows and bulk gases in the
active burner region in this single-wall boiler than anticipated (and the
resulting greater conversion of fuel-bound nitrogen).
Further confirmation of calculations for lignite coal will be
shown later in this section in the analysis of a boiler not in the data sample
used for regression analysis.
A further example of the effect of reduced combustion tem.-
perature rise because of the coal type could be obtained from the boiler in
the data sample firing the high ash coal (coal type 2 of Table 2). The high
weight fraction of inert ash in this coal limits the heat release per unit
weight of this coal and absorbs some of the heat of combustion. Equilibrium
combustion calculations indicate that the temperature rise due to combus-
tion of this coal under a stoichiometric air-fuel ratio is 1697 K (3055° F),
compared to 1566 K (2819°F) for the lignite coal (type 1) and 1836 to 1823 K
(3305 to 3281°F) for coal types 3 and 4, respectively. Thus, the combustion
temperature ris~ of this high ash coal is intermediate between the lignite
coal and the nominal coal represented by coal type 3. The full-load com-
bustion air temperature of this boiler was nominal (574 K, 574°F).
Figure 13 shows the calculations and available data for the
boiler firing this high ash coal. The calculations, as expected, show a
level of thermal NO generation (above the NO levels from fuel- bound
x x
nitrogen conve rsion) inte rrnediate between that calculated for the coals with
55
-------
low and high peak combustion temperature s. The data, howeve r, deviate
considera!-ly froru 'tl1e calculations, particularly at high Jevels of overall
boiler excess air.
As mentioned in the discus sion with re spect to Figure 13 in
Section 2. ;~. 2.1, the burrier corlfiguration in this boiler is so complex and
unsymmet,~ical that it is difficult to estimate the appropriate effective
fraction of theoretical air in the active burner region when air-only burners
are located anywhere but in the top level of the burner array. Nearly half
of the data represent te sts with such mixed burner configurations. Of the
remaining data only one test (that with the lowest level of excess air in the
burners) was conducted with excess air in the burners less than stoichio-
metric. That te st, although it showed excessive, carbon monoxide emis sions
(260 ppm), still showed NO levels 78 ppm higher than the parametric cal-
x
culation (curve no. 2). All of the data shown in Figure 13 at NO levels
x
above 820 :Jpm resulted from testing with all burners active. The disagree-
ment here with the appropriate parametric calculation (curve no. 1) is large,
averaging more than 200 ppm.
At this writing, no satisfactory explanation of this disagree-
ment has been discovered. The data appear to show a large (about 200 ppm)
step change in NO levels when the excess air in the burners is about 105
x
to 110 percent of theoretical. This might imply that much more thermal
NO is being gene rated than calculated. On the other hand, it is conceivable
x
that the fuel-bound nitrogen conversion efficiency is higher with this high
ash coal tha.n with the others in the sample. It might be conjectured that a
larger fraction of the nitrogen normally remaining in the char is converted
to NO und~r the very high levels of excess air required in these boilers
x ' '
to burn out the carbon in the char. Unfortunately, no further data were
available to this study involving the se very high ash coals.
Although the disagreement between the calculations and the
data shown in Figure 13 is large for test conditions where the excess air in
the burner!! is high, the disagreement is much less at the lower levels of
56
-------
excess air and appears to be decreasing further with excess air in the
burners. From the standpoint of a NO reduction program, then, the para-
x
metric calculations (particularly curve no. 2) may still be useful. Perhaps
the general conclusion from the data of Figure 13 should be that very strong
reductions in NO levels can be achieved by reducing the excess air in the
x
burners to less than about 105 percent of theoretical and that further reduc-
tions may be indicated by the parametric calculations.
2.2.2.3
Boiler Not in the Data Sample
If the entire data analysis approach taken in this study of
coal-fired boilers is indeed gene ral, and if the range of coal type sand
boiler configurations included in the data sample analyzed adequately repre-
sent the operational spectrum, then the re sulting parametric equation
should also be sufficiently general to predict ,NO emissions in other boilers
x
not in the data sample.
Use of the parametric equation to accurately predict NO
. x
emissions, however, requires a great deal of information on boiler
geometry, coal analysis, and operating conditions. It also requires a rela-
tively complex computer program to calculate the values of the terms in
the equation as the desired variables are parametrically varied. The
results of this study would be much more widely useful if just a few simple
calculations could be made at least to predict reasonably obtainable lower
limits of NO and to provide direction on how to modify combustion to reach
x
these lower limits. To some extent it is academic to explain why certain
test conditions yield' very high NO emissions if NO reduction is the pur-
x x
pose of the exercise.
In orde r to evaluate such an approach, the data from Leland
Olds Boiler No.1, of the Basin Electric Power Cooperative, 7 were excluded
from the sample used for the regression analyses. This boiler was thought
to represent a complex test of the generality of the results of this study.
A high moisture (lignite) coal (with the low combustion temperature rise)
was. fired in this boiler but full-load combustion air temperatures were very
57
-------
high (783 K, 9500 F).. The burner configuration in this boiler is also very
complex, ctmsisting of two levels of horizontally, opposed burne rs, four
burners in (~ach wall at each of these two levels, topped by four unopposed
burne rs in a single level in only one wall.
Initial considerations regarding the boiler configurations of
interest to a NO reduction program were the following:
x
a.
b.
c.
Only NO emissions under full rated load are of interest.
x
A nominal, refe rence operating condition, with all
burners active, is desirable.
Most of the literature and the results of this study
(see Figure 8) indicate that minimum NO emis sions
x
with air-only burners in the burner array result when
these air-only burners are located in the top level of
the array. Prediction of NO levels with other con-
x .
figurations is not of interest. This also simplifies
prediction because, with all air-only burners located in
the top level, the average, effective fraction of excess
air in the active burne r region is equal to the fraction
of exce ss air in the active burners.
equal to one.]
[Equation (1) is
d.
Minimum NO levels are achieved when the ove raIl
x
boiler excess air is as iowa's possible, consistent with
acceptable carbon monoxide emis sions. For configura-
tion s involving air - only burne r s, then, the ove rall boile r
excess air was set at a level 'that yields three percent
exce s s oxygen (116 pe rcent of theore tical air).
e.
An indication of the effect of reduction in combustion
air temperature is desirable.
58
-------
Under the se considerations, the prediction of NO levels
. . x
for this boiler, not in the data sample used for regression analyses, was
accomplished as follows:
a.
NO emissions generated from conversion of fuel- bound
x
nitrogen were estimated by selecting the appropriate
curve from Figure 6 (in this case, coal type 1). This
curve could also be calculated according to the pro-
cedure outlined in Appendix B.
b.
The thermal NO levels with all burners active were
x
taken directly from the constant resulting from the
regression analysis of the entire coal sample
(i. e., 146 ppm).
c.
Thermal NO levels with air-only burners located in the
x
top levels of the burner array, a constant re suIting from
the final mixing zone, were estimated from the curve
shown in Figure 16 of Appendix C. The appropriate
temperature for that figure was estimated from (1) the
combustion temperature rise at stoichiometric air-
fuel ratio, from equilibrium combustion calculations for
coal type 1; (2) the measured full-load combustion air
temperature or the arbitrarily chosen 422 K (3000 F);
and (3) a rough estimate of 110 K (2000 F) cooling of the
peak combustion temperature enroute to the final
mixing zone.
Figure 14 shows the results of these simple parametric
calculations for Leland Olds No.1. The data show good agreement with the
parametric calculatlons, both curves nos. 1 and 2, including the lower levels of
NO consistently observed when carbon monoxide emissions are high. These.
x
data also show a distinct step transition between curves nos. 1 and 2 at a level
of excess air in the burners of about 105 percent.
This is consistent with
59
-------
218 MW OPPOSED - 20 BURNERS - HIGH MOISTURE COAL (Lignite)
780 K (950°F) COMB AI R TEMP
smn ~ CD V A R I ABLE EX CE S S A I R
,",vV' -
CV 0-8 AIR-ONLY BURNERS IN TOP LEVELS
Q) SAME AS CV EXCEPT 422 K
(300°f) COMB A I R TEMP
o NOx (CO ~ 200 ppm)
6 NOx (CO> 200 ppm)
0' .
o
ON 600
~
("t'\
I- -.
«
~ 400
c . .
E
c..
c..
0>< 200
z
, 0
.CD
o~~
..~
~~'6, .--
~ -~
A~~
.,-~ -
.,-- ~
",..",tY -
-",.--~ .
,-"-- ~
.cg) .,- .,-'" -
","- -~
0), ,- .--"':: ~
O. 6 O. 7 O. 8 O. 9 1. 0 1. 1 1. 2
. .
FRACTION OF THEORETICAL AIR IN THE ACTIVE BURNER REGION
Figure 14.
Parametric calculations for boiler not in data sample
used to derive equation
-------
the level of excess air above which the final mixing zone, as defined in
Section 2.1. 3, would cease to exist. This is because the final mixing would
not force the air-fuel ratio of the products of combustion from the active
burner region to pass through, or even approach, the region of high thermal
NO generation near stoichiometric.
x
In gene ral, then, the agreement between data and the simple
parametric calculations are perhaps better in this case, of a boiler not in
the data sample used for regression analysis, than it is with any of the
others. As this boiler is largely a horizontally opposed configuration, the
agreement here also tends to indicate that the disagreement observed with
the high ash coal (Figpre 13) is probably not due to the opposed-firing
configuration, as such.
2.2.3
Boiler Excess Air
One other observation from the calculations and data of this
study deserve s some attention. In the comparison of data with the para-
metric calculations shown in Figures 9 through 14,' it was consistently
observed that NO levels measured during tests that also exhibited exces-
x
sively high levels of carbon monoxide tended to be lower than the calculated
curve. While not expre s sly stated, closer examination of the data also
shows that NO levels measured during te sts in which the overall boiler
x
excess air was very high (and carbon monoxide levels very low) tended to
be higher than the calculated curve. These observations appeared to be
generally true regardless of the boiler or burner configuration represented
by the te st data.
It is well known that when the overall boiler excess air is
reduced, some level is reached below which carbon monoxide emissi.ons
begin to rapidly increase. Thus, the above observations could simply be
attributed to a strong effect of exce ss air on NO emis sions. It seemed
x
clear, however, that either the model used in this study or the regression
analysis of the data, or both, failed to account for a very strong, step change
61
-------
in NO when overall boiler excess air was in the range where carbon
x
monoxide levels begin to become excessive (about two to three percent
oxygen). This could imply a phenomenon that was not properly modeled.
A separate, more detailed study of the effect of ove raIl
boiler exce:iS air alone on NO and carbon monoxide emissions with avail-
x
able data is usually difficult to interpret because either a number of other
significant variables are also varying or there is too little data. By the
time that all of the datil with differ.ent boiler geometries, burner configu-
rations, and significantly different load levels are eliminated, usually very
little data are left to show this effect. One exception in the .data sample of
this study ii, the data from the sister boilers Widows Creek No.5 and 6.
. .
Data from these sister boilers represent the largest coal-
fired data sample from a single coal and boiler type in this study. It was
possible, ft..en, to select only data from this sample where the firing con-
figuration VI'as held constant (all burners active) and the boiler was operated
near rated :.oad (greater than 80 percent of rated capacity), and still have
enough data left to show meaningful results. These selected data were
averaged oyer each O. 5 unit of exces s oxygen and are shown in Figure 15.
This figure shows the usual trend for NO to decrease and
x
carbon monoxide to increase as the overall excess air is decreased, but it
also shows what might be interpreted as an independent effect of the presence
of high concentrations of carbon monoxide on the NO levels. The effect of
x
boiler excess oxygen (or air) appears relatively small at excess oxygen
levels abov(~ about 3.5 percent and below about 2.7 percent. In between
these excess oxygen levels, decreasing excess oxygen appears to result in
a step-like change in NO levels amounting to about 150 ppm, accompanied
x
by a strong increase in carbon monoxide emissions from negligible levels
to over 1000ppm. While this trend is not as clear in other data samples,
it appears that the step reduction in NOx with the appearance of high levels
of carbon ITlonoxide eriIis sions as overall boiler exces s air is reduced
,
in this exces s oxygen range, occurs in other 'boilers fired in this same
62
-------
700
650
N
0
~
C'i'
......, 600
co
>-
L-
"'C
E
0..
0.. 550
0'
LN X
0
Z
500
450
1
CO, ppm
o 0 - 200
~ 200 - 500
. > 500
P ARAMETR I C
CALCULATION
125 MW - SINGLE-WALL
D AT A:
ALL BURNERS ACT I VE
> 80% RATED LOAD
PARAMETRIC CALCULATIONS:
ALL BURNERS ACT I VE
RATED LOAD
5
6
2
3 4
EXCES S OXYGEN, percent
Figure 15.
Effects of excess air and carbon monoxide levels on NO
x
-------
configuration as well as in this same boiler fired with some burners out of
service (a i.r- or..ly).
The possibility of some sort of competition between carbon
monoxide a.nd NO for the available oxygen (or reduction of NO by carbon
x x
monoxide) ~ras implied in the assumptions in the fuel-bound nitrogen model,
discussed jn Section 2.1.2. It was assumed there that the oxygen remaining
after all of the carbon had been oxidized to carbon monoxide and the free
hydrogen to water was divided in some fixed proportions between oxidation
of the carbon monoxide to carbon dioxide and the fuel-bound nitrogen to
NO . Figure 15 indicate s that, for levels of exces s oxygen greater than
x
about 3.25 percent, there is more than enough oxygen available to oxidize
all of the ca rbon monoxide to carbon 'dioxide. The re sulting NO levels,
, x
under these conditions, may be established by the availability of the fuel-
bound nitrogen (for example, in the volatiles rather than in the ash).
Reducing the exce s s oxygen from 3.25 to about 2.25 percent (only a one
percent reduction) results in a strong increase in carbon monoxide emissions
with measu:red values exceeding 1000 ppm. At 2.25 percent excess oxygen,
then, the efficiency of conversion of the available fuel-bound nitrogen may
now be at least partly established ~y the competition between the carbon
monoxide and fuel- bound nitrogen for the remaining oxygen (oxygen limited).
Although much of the above paragraph is conjecture,
Figure 15 d:)es appear to indicate that such a step change in NO levels with
, x
overall boih~r excess air exists. This observation is somewhat supported
by the data ~3hown in Figure 9 and to a more limited extent by that shown in
Figures 10, 11, and 12. Closer examination of data in these figures shows
that what appears to be unexplained data scatter around some of the para-
metric calculations is, to a large extent, a result of this strong but limited
(step) decrease in NO with the appearance of excessive levels of carbon
x
monoxide em.issions.
It is not clear, at least from this study, what characteristics
of the coal control the magnitude of this step or even if the coal charac-
teristics alone are the determining factors. Perhaps it is not important,
64
-------
to answer this question in this
from any study of NO control
x
reduction of NO emissions requires that the overall boiler excess air be
x
reduced to the point where carbon monoxide emissions begin to become
unacceptable. In the light of the apparent significant effect of carbon
monoxide emission levels on NO , however, it might be desirable to
. x
reevaluate those levels currently considered unacceptable as well as those
operating practice s that tend to maintain carbon monoxide emis sions well
below the se level,s.
NO reduction study smce it is clear
x.
in coal-fired boilers that maximum
65
-------
2.3
GENERAL DISCUSSION OF THE STUDY OF COAL-,
OIL, AND NATURAL GAS-FIRED DATA
Study of techniques for NOx reduction in coal-fired boilers
clearly indicate s the significance of the fuel-bound nitrogen as a source of
NO emissions. If the model for conversion of fuel-bound nitrogen developed
x
in this study is reasonably accurate, however, then the total NOx emissions
problem can be separated into that associated with this source and that
associated with thermally generated NOx' Results from this study of coal-
fired data and thos e from the previous study of oil- and natural gas -fired
data indicate that the two mechanisms do not appear to be inherently related
to each othe]" (1.. e., the rmally generated NO does not appear to be a strong
x
function of the NO gene rated from conversion of fuel- bound nitrogen, and
x
vice versa). All of the results of this study, then, should serve to extend
and corrobo]'ate the results of the previous study.
The observations on NO reduction from the study of oil- and
x
natural gas-::ired data are related to the more basic observation that thermal
NO can be rninimized by operating either on the fuel- rich or on the fue I-lean
x
sides, well c,~way from stoichiometric. Subtraction of the NO calculated for
x
conversion of fuel-bound nitrogen from the test values shown and discussed
in this report shows the same general conclusions of the previous report. 1
General obs€:rvations of combustion modifications for minimum NO emissions
x '
shown to depend on the
then, can be di scussed for all three fuels and can be
amount of ni':rogen chemically bound in the fuel.
The general technique for minimizing NO emissions from the
x
conversion of fuel-bound nitrogen involves maximum reduction of the air
available for combustion in the local regions of the boilers where the solid
or liquid fuel is gasifying and where the initial hydrocarbon reactions are
taking place. Without major modifications to existing boile rs, this implie s
the ope ration of large numbers of burners on air-only, NO Ports or both
x ' ,
all located a3 high as possible in the boiler.
66
-------
Techniques for minimizing thermally generated NO depend
x
on the concentration of nitrogen in the fuel. In high nitrogen fuels such as
coal, the air available for combustion in the active burne r region must be
maintained as loY' as possible, at least until the initial hydrocarbon reactions
are completed. This will also minimize NO thermally generated in the
x
active burne r region. The remaining problem area, then, is in the boiler
mixing zone where the remaining exce s s air must be added to the hot products
of the earlier fuel-rich combustion to bring the total for the overall boiler
up to about three percent excess oxygen.
It is in this region that the local
air-fuel ratio of the bulk gases must pass through the region of high thermal
NO generation rates around stoichiometric as the remaining excess air is
x
added. The data of this study indicate that techniques which reduce the gas
tempe rature in this region (i. e., reduce combustion-air tempe rature or
low temperature' rise due to combustion) are quite effective in minimizing
NO thermally generated in this region.
x
If there is no nitrogen chemically bound in the fuel, as with
natural gas, it is not necessary to maintain fuel-rich conditions in the early
combustion regions. In fact, in order to avoid establishing mixing zones
anywhere in the boiler where the air-fuel ratio must pass through the regions
of high thermal NO generation rates, it appears most desirable to maintain
x
air-fuel ratio conditions in natural gas burners well above that of the overall
boiler, finally approaching the boiler air-fuel ratio at the latest possible
moment through the use of fuel-rich burners in the top level of the burner
array, or through the use of fuel-rich NO ports. This optimized concept
x
has not been tested in full-scale utility boilers. With more standard
modifications, involving air-only burners, air-only NO ports, or both, the
x
remaining active burners can only be fuel- rich. With gaseous fuels, where
the intimate air-fuel mixing in the burner flows is rapid, minimum NO
x
emissions normally result from the same burner and NO port configurations
x
that minimize NO emissions in coal-fired boilers (i. e., large numbers of
x
67
-------
air-only bu]'ners concentrated in the top levels of the burner array, open
NO ports and reduced peak ~ombustion temperatures).
x .
Low nitrogen-bearing oil fuels (less than about 0.25 percent
nitrogen, by weight) represent an intermediate case between that of coal and
natural gas. Clearly, if very low NO emissions are required, the NO
x . x
resulting from conversion of this low concentration of fue,l-bound nitrogen
must be eliminated. This means that the burne r air-fuel ratio must be the
same as for minimum NO in coal-fired boilers. . If, however, the level of
. .. x . .
NO emissions resulting from conversion of the fuel-bound nitrogen at air-
x . . . ., '.
fuel ratios Eomewh~t above that of the overall boiler is acceptable, then the
optimum natural ga~ configuration might be appropriate. The natural gas
configuration might at least minimize thermally generated NO , leaving
x
only the NO. resulting from the conversion of some fraction of the fuel-
, .~
bound nitrogen. As liquid fuels require considerable time to vaporize and,
therefore, intimate air-vapor fuel mixing is slow compared to that of natural
gas, the aVE rage, local air-fuel ratio surrounding the burning oil vapor can
approximatEly be maintained air-rich simply by locating air-only burners in
the lower levels of the burner array (i. e., by maintaining air-rich bulk gases).
All of the oil-fired boiler data analyzed previously1 indicated that, within the,
range of the configurations tested, minimum NO emissions resulted from
x
this latter C3.se.
68
-------
REFERENCES
1.
O. W. Dykema, Analysis of Test Data for NO,c Control in Gas- and Oil-
Fired Utility Boilers, EPA-650/2-75-012 (NTIS No. PB 241-918/AS),
Environmental Protection Agency, Research Triangle Park, N. C.
January 1975.
2.
A. F. Sarofim, "Kinetics of Devolatilization of Nitrogen Compounds
During the High Temperature Pyrolysis of Coals, " Paper presented
at the 24th Symposium (International) on Combustion, The Combustion
Institute, Tokyo 1974.
3.
Rocketdyne Division of Rockwell International, ":Pyrolysis of Model
Fuel Nitrogen Compounds and Fossil Fuels," Paper presented at
EPA/CRS Fundamental Combustion Research Contractors Meeting,
Menlo Park, Calif., June 1975.
4.
Rocketdyne Division of Rockwell International, "Flat. Flame Burner
Studies with HCN, NH.3 and NO Addition, " Paper presented at EPA/CRS
Fundamental CombustIon Research Contractors Meeting, Menlo Park,
Calif. , . June 1975.
5.
D. W. Turner and C. W. Siegmund, "Staged Combustion and Flue Gas
Recycle: Potential for Minimizing NOx from Fuel Oil Combustion, "
Paper presented at American Flame Research Committee Flame Days,
Chicago, 6-7 September 1972.
6.
A. F. Sarofim, "Sources and Control of Combustion Generated
Pollutants," lecture notes for a summer course, Energy, a Unified
View, Massachusetts Institute of Technology, Cambridge 1974.
7.
A. R. Crawford, E. H. Manny, and W. Bartok, Field Testing:
Application of Combustion Modifications to Control NO Emissions
from Utility Boilers, EPA-650/2-74-066 (NTIS No. PB~37-344/AS),
Environmental Protection Agency, Research Triangle Park, N. C.
June 1974.
8.
G. A. Cato et aI., Field Testing: Application of Combustion Modifi-
cations to Control Pollutant Emissions from Industrial Boilers -
Phase I, EPA- 650 /2- 74-078a (NTIS No. PB 2 38-920 / AS), Environ-
mental Protection Agency, Research Triangle Park, N. C. October
1974.
6g
-------
APPENDIX A
MIXING ZONE LENGTHS IN TANGENTIAL BOILERS
In the mixing model for all boiler firing types, the mixing of
various flows entering a zone is assumed to occur instantaneously at the
beginning of the zone. The resulting mixture is then assumed to flow uniformly
through the zone, cooling and generating NO . The average time for cooling
x
enroute to a given zone is taken as the sum of the stay times in all of the
zones upstream of the given zone plus one-half of the stay time in the given
zone. The time for NO generation in that given zone is the stay time in
x
that zone. .
Stay times are all calculated from the length of the zones
divided by the average flow velocity within the zones. Because of the signi-
ficantly different flow and mixing patterns, it was necessary to modify some
of the mixing zone definitions and lengths used with face-fired boilers to
approximate the mixing in tangential boile rs. Figures 1 and 3 of the main
body of the text generally show the modifications in burner flow mixing. The
resulting differences in zone lengths are discus sed briefly here.
The primary zone lengths are considered to be approximate.ly
the same in face and tangentially fired configurations. The s e lengths are
taken as twice the horizontal width (or diameter) of the burners. In the tan-
gential configuration, the horizontal width of Iia burner II was taken as that of
. .
the primary (carrier) air and fuel duct assuming a square geometry.
In the opposed-fired boilers, the sum of the lengths of the
secondary and the opposite zones were taken to be the distance from the
. 7 1
-------
,firing wall to the center of the furnace (midway between the firing faces)
minus the length of the prima ry zone. The lengtr.. of the opposite (and
adjacent) Ir.',ixing zone was taken as twice that of the secondary zone. Thus,
for opposec,-fired boilers,
ZLAO = (HDF /2) - ZLP
1.5
(2)
and
ZLS = ZLAO /2
( 3)
These same assumptions were used for the tangential boilers except instead
of the burner flow going from a firing wall to the center of the furnace (see
Figure 1 of the main body of the report) the flow from a burne r in the tan-
gential furnace goes from a corner of the furnace to a point off the centerline
of the furnace (see Figure 3). This total distance was taken as the average
horizontal ::_ength of a side of the- furnace times the cosine of 30 degrees. If
the same p:roportions as in the opposed-fired boiler are used, the lengths
of the oppo:3ite (and adjacent) and the secondary mixing zones can be cal-
culated fr07ll
ZLAO = [0.866 (HWF + HDF)/2] - ZLP
1.5
(4)
and
ZLS = ZLAO /2
(5)
The greatest difference between the wall-fired and the tan-
gentially configured boiler flow patterns appears to be in the recirculation
flow. The recirculation flow in a face-fired boiler was as sumed to originate
from one huner, recirculate in essentially a 3600 circle and mix with the
72
-------
flow from the burner in the next elevation above.
This case was assumed
because the flows from all otber burners are either parallel or directly
opposed to the flow from a given burner. In the tangential configuration,
however, the flow from a given burner is not parallel to, nor does it directly
\
oppose, the flow from other burners. Instead, a given burner flow is more
or less directed at the next burner in the direction of the general rotation of
the bulk gases. Low-pres sure regions generated by entrainment of gases
near a burner in one corner are much more likely to generate a recirculation
flow from the burner in the adjacent corner.
In the wall-fired boilers, the recirculation flow was taken to
travel a distance of 'iT times the length of the primary zone at the velocity in
the primary zone. In the tangential boilers, the same amount of recircula-
tion flow was taken to travel a distance equal to the average horizontal
length of the furnace walls at one-half of the velocity in the primary zone.
Thus, the stay time of a given quantity of reactants in a recirculation zone
in an opposed-fired furnace is given by
T UR = 'iT Z LP
o VP
(6)
while the equivalent stay time in a recirculation zone in a tangential furnace
is
TUR = HWF HDF 2
t 2 VP
(7)
For typical opposed-fired and tangential boilers where the
horizontal section might be a square 12.2 meters (40 feet) on a side and the
horizontal width of the burners might be 0.914 meters (3 feet), the ratio of
stay times in each of the mixing zones in a tangential boiler compared to these
same zones in an opposed-fired boiler (in this study) are shown below:
73
-------
Mixing Zone
Ratio of Stay Time,
Tangential/Opposed
Primary
Recirculation
1. 00
4.24
2.04
2.05
Secondary
Adjacent/Oppo site
The stay time in the recirculation region in a tangential
boiler, unde r the mixing as sumptions of this study, is more than four time s
that in a face-fired boiler. (In all cases only about one-third of the flow
from a given burner is assumed to pass through the recirculation zone.)
This flow ir. a tangential boiler also passes very close to a water wall along
the entire l.mgth of the recirculation zone. Unfortunately, no allowance
could be me.de in this study for the possibly increased cooling rate in this
zone caused by combined radiative and convective heat transfer.
In general, the total cooling time involved in burner flows in
a tangential boiler, before they enter and mix with the bulk gases, is about
twice that of face-fired boilers. The stay times in the bulk gas mixing zones,
for the same vertical distance between burners, are taken as the same, in
all cases, as are the NO port or final mixing zones. The effect of these
x
longer cooling times is particularly strong in reducing the temperatures in
the final mixing zone, equivalent to configurations with NO ports located
x
very high in face-fired boilers.
74
-------
APPENDIX B
MODEL FOR NO FROM CONVERSION OF FUEL-BOUND NITROGEN
x
The basic assumptions on which this largely empirical model
of conversion of fuel-bound nitrogen to NO is based are discussed in
x
Section 2. 1. 2 of this report. The data for oil fuels shown in Figure 5 of the
main body of the report need no explanation other than the reference sources
cited in the figure. Several attempts were made to include the coal data
with the oil data such that the oil and coal data were internally consistent
in the same fashion that the oil and coal data samples were consistent with
each other (i. e., the curve that separately fit the oil data and the one that
separately fit the coal data were the same curve). It was found that the
measured efficiency of conver sion of the nitrogen bound in the coal was
always lower than that which would be predicted by extrapolating a reason-
able fit of the oil conversion efficiency data unless the appropriate weight
fraction of bound nitrogen in the coal was taken as that fraction of the coal
weight with all of the inert diluents '(ash and moisture) removed. This re-
sult seems reasonable because (1) this is es sentially the same definition of
weight fraction of bound nitrogen in the oil fuels (as they contain negligibl~
ash and moisture), and (2) addition of any amount of inert solids containing
no nitrogen to a coal (which would change the weight fraction of the overall
fuel that is bound nitrogen) would not be expected to affect the amount of NO
x
generated during the reaction.
For the data shown in Figure 5, the fraction of the coal weight
that is ash was obtained directly from the ultimate analyses for the coal
75
-------
fired. 7,8 The moisture was obtained not from the proximate, analyses but
from an ass'.1mption that all of the oxygen reported in the ultimate analyse s
was bound with the appropriate weight of hydrogen in the form of water. On
the average, this calculation results in a weight percent of moisture in the
coal about,2. 4 percent less than that cited in the proximate analysis. This
difference if; as large as ten percent (low) with the lignite coals. No attempt
was made to correct for this difference.
The empirical curve shown in Figure 5 was established to fit
the average of the oil data because these data resulted from te sting either
with large q'.1antities of flue gas dilutionS or with very low combustion air
temperature s. 8 Nearly all 'of the NO measured in these tests should have
x
resulted froD1. conversion of fuel- bound nitrogen. The data for coal-firings,
however, 7, II were obtain~d under test conditions where thermal NO could
x
be appreciahle. The curve was the refore fit to that data as a lower bound.
No serious attempt was made to generate any data, or to fit
the curve, f::>r fuels where the weight fraction of fuel-bound nitrogen was less
than about 0.2 percent. It is meaningless that the curve fit to the data
reaches 100 percent conversion for a fuel with a weight percent of bound
nitrogen slightly greate r than O. 1 percent.
The expression for NO emissions, in ppm (dry), can be
x
written front stoichiometry. If the equivalent coal molecule is written
CHaObNcSd + A (ash)
(8)
the NO emissions, expressed as nitric oxide (NO), ppm (dry), are
x
'( ec) 106
NO = 0.9996 K 1 (AFRCL) + (biZ) - (a/4)
(9)
whe re
e = fraction of fuel-bound nitrogen converted to NO
AFRCL = air-fuel ratio of the overall boiler, by weight
76
-------
K1 = (1 + A)(MW fuel) {(MW air)
(10)
The two main assumptions used in the furthe r development
of this model of fuel- bound nitrogen conversion, as discussed in Section 2. 1. 2
of the text, are (1) the conversion efficiency e is directly proportional to the
availability of oxygen in the boiler region where the initial hydrocarbon re-
actions are taking place, and (2) only the oxygen remaining after all of the
carbon in the fuel has been oxidized to carbon monoxide and all of the
hydrogen (not already bound in moisture in the fuel) to water is available to
form NO. The conversion efficiency e, according to these assumptions, can
be written
e = K (AFRP - AFRCOS)
( 11)
whe re
AFRCOS = air-fuel ratio, by weight, at which assumption (2)
is exactly satisfied [here called II CO stoichiometric II
( COS ) ]
AFRP
= air-fuel ratio, by weight, in the effective local region,
defined in assumption (1)
When AFRP is that value corresponding to three percent excess
oxygen, then the conversion efficiency should be given by the curve or equa-
tion of Figure 5 of the main body of the report. Therefore,
( AFRP - AFR COS )
e = e3 AFRP3 - AFRCOS
(12)
where e3 and AFRP3 are the values of e and AFRP when AFRP corresponds
to three percent exce ss oxygen. In order to generalize Equation (12), all of
the air-fuel ratios can be related to stoichiometric (reaction to carbon
dioxide). If AFRS is defined as this stoichiometric air-fuel ratio, the
77
-------
following ratios-of-ratios can be defined:
AFRP
R AFRP - AFRS
( 13)
RCOS
=
AFRCOS.
AFRS
( 14)
AFRP3
RAFR3 = AFRS
( 15)
and Equatior, (12) can be written
(RAFRP - RCOS\
e = e3 \RAFR3 - RC0S)
(16)
In order to tlse the equation describing the curve fit of Figure 5, it is
convenient tI> write that equation in terms of fractions rather than percents:
e3 = 0.01049 (WFBN2)-2/3
( 17)
where e3 and WFBN2 are now the fraction of conversion of bound nitrogen
and the frac1;1.on of the ash- and moisture-free fuel that is bound nitrogen,
respectively. Substitution of Equation (17) into Equation (16) yields
= 0 01049 (WFBN2)-2/3 fRAFRP - RCOS)
e . \RAFR3 - RCOS
( 18)
The constant.c, defining the atoms of nitrogen in the equivalent
coal molecu:.e in Equations (8) and (9), can be written
78
-------
(1-f -f~
c = O. 8574(WFBN2) r20 A
. Ca
(19)
where fH20' fA' and fCa are the fractions of the total weight of the fuel that
are water, ash, and carbon, respectively; and WFBN2 is again the weight
fraction of nitrogen in the ash- and moisture-free fuel.
For convenience, the denominator of Equation (9) can be
rearranged such that
6
(ec) 10
NO = 0.9996 K1 (AFRCL - K3)
(20)
where
K - (a/4) - (b/2)
3 - O. 9996 K 1
(21 )
Finally, through substitution of Equations (18) and (19) into
Equation (20) and grouping of terms into those which are functions of the fuel
alone and those which are functions of the operating conditions, the NO.
. . x
generated by conversion of fuel-bound nitrogen, in ppm (dry), can be written
NO - FFUEL (RAFRP - RCOS)
. - AFRCL - K3
(2.2)
whe re
(1_fH20_fA\f '(WFBN2)1/3 1
FFUEL = 8',94.4 , . fCa - J lK1 (RAFR3 - RCOS~
(23)
79
-------
For purposes of calculation of the constant K l' accord.ing to
Equation (10), a representative ash was assumed consisting of 50 percent by
weight of Si02' 20 percent each of A1203 and Fe203' and percent CaO.
The molecuOlar weight of this ash is 75.09. The equivalent moles of ash in
the equivalent fuel (8) is then given by
A ~ O. 16~f~)
(24)
Appropriate values of the fuel-related terms in Equation (22)
for the four coal types listed in Table 2 of the main body of the report and
for the oil and gas fuels used in the data of the previous study, 1 plus values
for the stoichiometric air-fuel ratio, by weight, are listed below:
Fud AFRS FFUEL RCOS K3
Coal Type
No. 1 8.048 5626 0.546 O. 150
No.2 8.001 6520 0.578 0.254
No.3 9.437 7338 0.571 0.254
No.4 9.548 7945 0~565 0.254
,I,
Oil" 13.84 5557 0.638 0.798
Natural Gas 15.88 0 0.738 1. 586
In Equation (22), the constant FFUEL is a function only of the
characteristics of fuel fired. It is a function of the one-third power of the
weight fract:':~on of nitrogen chemically bound in the ash- and moisture-free
fuel. For a';.l of the fuels listed above, the fraction of theoretical air
corresponding to three percent excess oxygen (RAFR3) is about 1. 16.
The numerator of Equation (22) essentially represents the
oxygen available in the boiler region where the hydorcarbon reactions are
,I,
"'A low sulfur, low nitrogen oil, with 0.24 percent nitrogen, by weight.
80
-------
taking place after the oxygen necessary to oxidize the carbon to carbon
monoxide and the hydrogen (pot already bound in moisture in the fuel) to
water has been removed.
Because RAFRP is linear, linear averages of
varying conditions in this region are appropriate.
The denominator of Equation (22) es sentially represents the
total flue gas (the dilution factor), of which the NO is one specie. In order
to calculate the NO, in ppm (dry), which would be measured if the overall
boiler excess air were such that three percent oxygen were also measured,
AFRCL must be set to approximately 1. 16 times AFRS. The calculations
of NO in ppm (dry), at three percent excess oxygen from Equation (22) for
the fuels listed above, are plotted in Figure 6 of the main body of the report.
81
-------
APPENDIX C
FINAL MIXING ZONE
It became apparent in the early regression analyses of samples
of coal-fired data containing wide variations: of combustion temperature rise
and full-load combustion air temperatures that errors of as much as 120 ppm
NO in the constant derived by the analyses could result when applying the
x
results of analyses of the full data sample to a particular coal and boiler
type. It was concluded that a new term should be developed to account for
this apparent variation in the constant.
basis for this new term:
Four main observations formed the
b.
a.
The correlation equation used in the early regression analyses
appeared to contain no term or terms that could account for
differences in the constant appropriate to the different coal
and boile r type s. Therefore, the related phenomena must not
be described in the initial equation.
The variation in the constant indicated by the individual boiler
data samples, including all of the coal, oil, and natural gas
samples, appeared to be some exponential function of the peak
combustion tempe rature.
c.
Parametric calculations for all of the coal-, oil-, and natural
gas -fired boilers indicated that the source of this constant
NO is in the NO port or final mixing zone.
x x
There was some evidence that the appropriate value of this
constant was different when all burners were active and no
d.
NO port flow was involved, as compared to tests where some
x
burne rs were operated air-only.
83
-------
As a result of these observations and further parametric
calculationE it was concluded that the most likely source of this NO is the
. x
region (and period of time) where the final excess air is mixed with the
products of fuel- rich combustion corning from the active burner region. In
this region, the fuel- rich gase s are forced to shift relatively slowly through
the air-fuel ratio region of high thermal NO generation rates near stoichio-
x
metric to tr.e fuel-lean air-fuel ratio of the overall boiler. As long as a
burner configuration is involved where the burner air-fuel ratio is less than
stoichiometric; this. region will exist somewhere in the boiler . It will be
independent of the actual burner and overall boiler air-fuel ratios (constant
with respect to these variables) and should be a function only of the com-
bustion product temperature in that mixing zone. This assumption is sup-
ported by the main obse rvations discus sed above in the following ways:
a.
The "tank-and-tube II type of mixing as sumed throughout the
model, including this final mixing zone, assumes instantaneous
mixing and, therefore, no term was present in the initial
equation accounting for NO generated during any transient
x
mixing. In this final zone, large fractions of the total boiler
flow are involved in the mixing process and the assumption
of instantaneous mixing is in the greatest error. This is
also the most critical zone because the rnaximum thermal
b.
NOx generation rates are representl>d in the mixing.
The rate of thermal NO generation in this transient mixing
x
II
zone, always at air-fuel ratios JW~ \. stoichiometric, should be
a function only of the combustion pl"Oduct temperatures in that
zone.
c.
A significant transient mixing zone' of this type, involving
the mixing of relatively large fractions of the total flow,
would exist only where large fractions of the combustion air
are added to the total flow in some concentrated region of the
84
11
-------
boiler. In the data samples of this and the previous study
. .
this case is represented almost totally by configurations
where large numbers of air-only burners are concentrated
in the top levels of the burner array, where NO port air
x
flow is involved, or both.
d.
A concentrated region of mixing of large fractions of the
total flow such that the air-fuel ratio passes through stoichio-
metric during the mixing would not exist if all of the burners
were active and, therefore, already operating at air-fuel
ratios above stoichiometric. Thus, NO data plotted as a
x
function of the burner air-fuel ratio can, in some cases,
exhibit some type of discontinuity near the stoichiometric
air-fuel ratio.
As a result of the above assumptions, the data from each of
the boilers in the entire 575-test coal, oil, and natural gas sample were
examined to derive estimates of the value of this constant. NO levels
x
measured under the following conditions were taken as the appropriate data:
(1) full-load firing conditions, (2) air-only burners concentrated in the top
levels of the burner array, (3) NO port flow involved, and (4) the active
x
burner air-fuel ratio was calculated to be less than stoichiometric. For
the natural gas fuel, these measured levels approximately represent the
constant directly. For the coal and oil fuels, and NO generated from the
x
conversion of fuel-bound nitrogen, calculated from the model described in
the previous subsection, was subtracted from the measured NO levels to
x
derive the estimate of the constant. In many cases involving air-only
burners, the boiler was never tested under full load with the air-only
burners concentrated in the top levels of the burner array. In such cases
it could only be assumed that NO levels and the constant would be lower if
x
such configurations had been tested.
85
-------
, , For e~ch of .,th,e boilers for which estimates of the appropriate
constant cOHl~ b~ der,ived, the eguilibrium combustion product gas tempera-
tures and ft,e concentrations of ,N2 and 02 in the products in the appropriate
zone in the 'boil,e,r (~ONST in Figure 16 is. the product of these concentrations:
[N2][02]1/2), under stoichiometric combustion conditions, were calculated
from the flew and cooling model. Figure 16 shows a plot of the available
data generated in the above manner and an approximate empirical fit to the
data. Of the five coal-fired' boilers where reasonable estimates could be
derived, fonr provide a good empirica) fit. The one that deviates considerably
from the fit (82 ppm higher than the curve) is the high ash coal, which, as
shown in thl~ text of this report, also deviates from the other coal types and
boilers in nlany other ways. Three of the four available estimates for, oil-
and natural gas -fired boilers indicate only that the constant should be lower
than the plotted value while data from one gas-fired boiler indicates an
appropriate constant about 34 ppm higher than the empirical curve. This
latter error is small enough to be considered within the scatter of the
primary da':a used to derive the data shown in the figure.
Thus, the available data, limited as it is, tend to show a
reasonable fit to an equation of the appropriate form; i. e., an exponential
in temperature. From the discussion thus far, one might expect that this
empiri,cal Ht should represent some mixing time, ~t, time s a Zelodvich-type
equation fo]' the rate of formation of NO . The constants in the empirical
x
equation, h)wever, would not be expected to be exactly those of a Zeldovich
rate equation. The entire mixing proces s, during which the local air-fuel
ratios pass through the entire range around stoichiometric and the NO
x
formation rates are always less than or (briefly) equal to the maximum,
was approxlmated in the empirical fit only by the maximum NO formation
x
rate. Thus, the empirical equation would necessarily imply much shorter
mixing times and much less sensitivity to the local temperature in order to
correctly fit the actual, .average process.
86
-------
8.2 FUEL In (SONSn
-
8t 0 NATURAL GAS - 2. 760
~ OIL - 2.538
7.8 T 0 8 COAL TYPE 1 - 3.079
- TYPE 2 - 2.813
I--
V) TY P E 3 - 2.538
z:
0 TY P E 4 - 2.577
~ 7.4
c
-
-
I--
a..
z: '
0 7.0
00 u
--.)
c
- In (CONPn = 14.88
6.6 + I n (SONSn
-16. 344/ Tfm
. 6.'2
4.2
4.4
4.6 4.8
10.000/ Tfm
5.0
5.2
Figure 1.6.
Empirical fit of NO generated in final mixing zone (CONPT)
as a function of temperature in this zone (Tfm)
-------
The Zeldovich NO formation rate equation used throughout
1 x
this and the previous study was
4
~= (2.4X 1018) [N2] [02]1/2 exp(- 6. 79TX 10)
(25)
The rate equ.ation used to fit the data of Figure 16 was
~ "(Z.9A~ IQ6)[NZJ [O/IZ ex{ 1.63.; 104)
(26)
where 6t is the mixing time, in seconds; [NOl is in ppmj [N2] and [02] are
in mole fra:tionsj and t and T are in seconds and degrees Kelvin, respectively.
For peak local gas tempe ratures in the range of 2100 to 2500 K (3780 to
4500° R), tr..ese two equations yield the same NO formation rates if the equiv-
x
alent mixir.g times ~t are in the range of 1 to 55 milliseconds. This is
a reasonable range of mixing times considering that only the maximum forma-
tion rates were used to represent the average mixing process. Similarly,
the derivat:.ves of Equations (25) and (26) with respect to temperature show
that, in th4~ above temperature range, the empirical rate equation is two
to four ord(~rs of magnitude less sensitive to gas temperature variations than
is the theoretical. Again, this observation is a reasonable result of the
approximation used in the empirical fit.
Thus, the 'general observations and the empirical fit to avai.lable
data tend to confirm the existence of a zone of mixing, where the final excess
air is added to the products of fuel-rich combustion in the active burner re-
gion, which is a significant contributor to the overall NO emis sions. At the
x
very least, an empirical correction was developed to account for potential
errors of as much as 120 ppm in the constant derived from regression analy-
ses of data samples containing wide variations in combustion temperature
88
-------
rise and full-load combustion air temperatures.
This new term was used
to correct the constant obtained from regression analysis of the full data
sample for parametric calculations involving a single coal and boiler type.
89
-------
GLOSSARY
AFRCOS
ash
air-fuel ratio, by weight, of the active burners
air-fuel ra tio, by weight, of COS
air-fuel ratio, by weight, of the overall boiler
A
AFRB
AFCRL
AFRP
air-fuel ratio, by weight, in the effective local region
where the initial hydrocarbon reactions are taking place
AFRS
stoichiometric air-fuel ratio
AFR3
air-fuel ratio at three percent excess oxygen
c
atom of nitrogen in the equivalent coal molecule
CONPT
NO gene rated in the final mixing zone, ppm
x
COS
carbon monoxide stoichiometric
e
fraction of fuel- bound nitrogen conve rted to nitric oxide
(conversion efficiency)
f
fraction of total weight of fuel
FFUEL
function of characteristics of fuel fired
HDF
horizontal depth of the furnace from front to back burner
walls in the region of the active burners, meters (feet)
HWF
width between the side walls of the furnace in the region
of the active burners, meters (feet)
K1
constant (Equation 10)
K3
constant (Equation 21)
NBFA
number of active burners
NBFAT
total numbe'r of active burners' in th'e burner array
NBT
total number of burners
91
-------
NO
NO
x
n
ppm
R
SONST
T
TUR
VP
WFBNZ
ZLAO
ZLP
ZLS
~t
Subscripts
fm
i
o
t
3
nitric oxide
oxides of nitrogen
number of burne r levels
\ parts per million
ratio, '
prodl,lct involving the mole fractions of nitrogen
. 1/2
and oxygen; [NZ] [02]
temperature, kelvin
stay time in the recirculation zone, seconds
average flow velocity in the primary zone, meters /
second (feet/ second)
weight fraction of bound nitrogen in ash- and
moisture-free fuel
length in the direction of flow in the adjacent and
opposite mixing zones, meters (feet)
length in the direction of flow in the primary
mixing zone, meters (feet)
length in the direction of flow in the secondary
mixing zone, meters (feet)
mixing time, seconds
final mixing zone
burne r levels
opposed fired boilers
tangentially firE~d boilers
three per<;:ent excess oxygen
92
-------
TECHNICAL REPORT DATA
(Please read !m;JrllctioIlS 011 the rerCTse before completing)
1. REPORT NO. 12. 3. RECIPIENT'S ACCESSION'NO,
EPA-600/2-76-274
4. TITLE AND SUBTITLE ANALYSIS OF TEST DATA FOR NOx 5. REPORT DATE
CONTROL IN COAL- FffiED UTILITY BOILERS October 1976
6. PERFORMING ORGANIZATION CODE
7. AUTHOA(S) 8. PERFORMING ORGANIZATION REPORT NO.
Owen W. Dykema
9. PERFORMING OROANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT NO.
The Aerospace Corporation lAB014; ROAP 2lADG-089
Environment and Energy Conservation Division 1 1. CONTRACT /G RANT NO.
Los Angeles, California 90009 Grant R803283-01
12. SPONSORING AGENCY NAME AND ADDRESS 13, TYPE OF REPORT'I,ND PE~/rD COVERED
EPA, Office of Research and Development Task Final' 7 74-10 75
Industrial Environmental Research Laboratory 14. SPONSORING AGENCY CODE
Research Triangle Park, NC 27711 EPA-ORD
15. SUPPLEMENTARY NOTES Project officer for this report is R. E. Hall, Mail Drop 65, 919/-
549-8411 Ext. 2477. EPA-600/7 -76-012 was earlier report in this series. 16. ABSTRACT The report describes the analyses of a large quantity of emissions, opera-
ting conditions, and boiler configuration data from full-scale, multiple-burner, elec-
tric-generating boilers firing coal fuel. Objectives of the study include: (1) evaluation
of the effects of combustion modifications on NOx emissions, in fundamental combus-
tion terms; and (2) evaluation of techniques for further reductions in NOx emissions.
The report includes the following, pertaining to coal-fired utility boilers: (1) dis cus-
sion of the major sources of NOx emissions; (2) parametric investigations of the
effects on NOx emissions of two-stage combustion, burners out of service, combustion
air temperature, and excess air reduction; (3) discussion of probable short- and long-
term hardware and operating condition modifications likely to yield further significant
reductions in NOx emission in coal-fired boilers; and (4) general comparisons of NOx
reduction techniques in utility boilers firing coal, oil, and natural gas fuels. Boiler
firing types included single-wall, opposed, and tangential configurations.
17. KEY WORDS AND DOCUMENT ANALYSIS
-.
a. DESCRIPTORS b.IDENTIFIERS/OPEN ENDED TERMS c. COSATI Field/Group
Air Pollution Smoke Air Pollution Control 13B
Combustion Coal Stationary Sources 21B 21D
Emission Natural Gas Utility Boilers
Boilers Fue 1 Oil Emission Characteris- 13A
Nitrogen Oxides tics 07B
Carbon Monoxides
18. DISTRIBUTION STATEMENT 19. SECURITY CLASS (This Report) 21. NO. OF PAGES
Unclass ified 96
Unlimited 20. SECURITY CLASS (This page) 22. PRICE
Unclassified
EPA Form 2220-1 (9-73)
93
------- |