950R80043
U.S. Environmental Electric Power	IERL-RTP-1086
Protection Agency Research Institute October 1980
Proceedings of the Joint
Symposium on Stationary
Combustion NOx Control
Volume IV
Fundamental Combustion Research
and Advanced Processes
*'.* 4: ¦

SEPAtlEPRI
SEPA IS EPFS

-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1.	Environmental Health Effects Research
2.	Environmental Protection Technology
3.	Ecological Research
4.	Environmental Monitoring
5.	Socioeconomic Environmental Studies
6.	Scientific and Technical Assessment Reports (STAR)
7.	Interagency Energy-Environment Research and Development
8.	"Special" Reports
9.	Miscellaneous Reports
This report has been assigned to the MISCELLANEOUS
REPORTS series. This series is reserved for reports whose
content does not fit into one of the other specific series.
Conference proceedings, annual reports, and bibliographies
are examples of miscellaneous reports.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.

-------
IERL-RTP-1086
October 1980
Proceedings of the Joint
Symposium on Stationary
Combustion NOx Control
Volume IV
Fundamental Combustion Research
and Advanced Processes
Symposium Cochairmen
Robert E. Hall, EPA
and
J. Edward Ciehanowicz, EPRI
Program Element No, N130
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20400
and
ELECTRIC POWER RESEARCH INSTITUTE
3412 Hillview Avenue
Palo Alto, California 94303

-------
PREFACE
These proceedings document more than 50 presentations given at the
Joint Symposium on Stationary Combustion N0"x Control held October 6-9,
1980 at the Stouffer's Denver Inn in Denver, Colorado. The symposium was
sponsored by the Combustion Research Branch o£ the Environmental
Protection Agency's (EPA) Industrial Environmental Research
Laboratory-Research Triangle Park and the Electric Power Research
Institute (EPRI). The presentations emphasized recent developments in
NOx control technology. Cochairmen of the symposium were Robert E.
Hall, EPA, and J. Edward Cichanowicz, EPRI. Introductory remarks were
made by Kurt E. Yeager, Director, Coal Combustion Systems Division, EPRI,
and the welcoming address was given by Roger L. Williams, Regional
Administrator, EPA Region VIII. Stephen J. Gage, Assistant Administrator
for Research and Development, EPA, was the keynote Bpeaker. The symposium
had 11 sessions:
I:	N0X Emissions Issues
Michael J. Miller, EPRI, Session Chairman
II: Manufacturers Update of Commercially Available Combustion
Technology
Joshua S. Bowen, EPA, Session Chairman
III: N0X Emissions Characterization of Full Scale Utility
Powerplants
David G. Lachapelle, EPA, Session Chairman
IV: Low N0X Combustion Development
Michael W. McElroy, EPRI, Session Chairman
Va: Postcombustion N0X Control
George P. Green, Public Service Company of Colorado,
Session Chairman
Vb: Fundamental Combustion Research
Tom V. Lester, EPA, Session Chairman
VI: Status of Flue Gas Treatment for Coal-Fired Boilers
Dan V. Giovanni, EPRI, Session Chairman
VII: Small Industrial, Commercial, and Residential Systems
Robert E. Hall, EPA, Session Chairman
VIII: Large Industrial Boilers
J. David Mobley, EPA, Session Chairman
IX: Environmental Assessment
Robert P. Hangebrauck, EPA, Session Chairman
X:	Stationary Engines and Industrial Process Combustion Systems
John H. Wasser, EPA, Session Chairman
XI: Advanced Processes
G. Blair Martin, EPA, Session Chairman
ii

-------
VOLUME XV
TABLE OF CONTENTS
Session Vb: Fundamental Combustion Research
Page
Session Vb: Fundamental Combustion Research
"The Fundamental Combustion Research Program,11 T. J. Tyson,
C.	J. Kau, T. L. Corley, W. R. Seeker, W. Clark,
J. Kramlich, M. P. Heap, and W. S. Lanier		1
"Two Phase Processes Involved in the Control of Nitrogen
Oxide Formation in Fossil Fuel Flames," J. M. Beer,
A. F. Sarofim, L. D. Timothy, S. P. Hanson, A. Gupta, and
J. M. Levy		43
"Gas Phase Processes Involved in the Control of Nitrogen
Oxide Formation in Fossil Fuel Flames," J. M. Levy		84
Session XI: Advanced Processes
"Low N0X Combustors for High Nitrogen Liquid Fuels,"
G. C. England, M. P. Heap, D. W. Pershing,
J. H. Tomlinson, and T. L. Corley	 115
"Fate of Coal Nitrogen During Combustion," S. L. Chen,
M. P. Heap, D. W. Pershing, R. K. Nihart, and
D.	P. Rees	 154
"System Applications of Catalytic Combustion,"
J. P. Kesselring, W. V. Krill, S. J. Anderson, and
M. J. Friedman	 181
"Fixed-Bed and Suspension Firing of Coal,"
S. P. Pureell, D. M. Slaughter, J. M. Munro, G. P. Starley,
S. L. Manis, and D. W. Pershing	 216
"Pressurized Bench Scale Testing of Low N0X LBG
Combustors," W. D. Clark, B. A. Folsom, W. R. Seeker,
C. W. Courtney, and M. P. Heap	 248
"Control of N0X and Particulates Emission from SRC-II
Spray Flames," J. M. Beer, M. T. Jacques, S. Hanson,
A. K. Gupta, and W. Rovesti			 279
iii

-------
THE FUNDAMENTAL COMBUSTION RESEARCH PROGRAM
By:
T. J. Tyson, C. J. Kau, T. L. Corley, W. R. Seeker,
W. Clark, J. Kramlich, and M. P. Heap
Energy and Environmental Research Corporation
8001 Irvine Boulevard
Santa Ana, California 92705
W. S. Lanier
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
1

-------
I. FCR-I STRUCTURE AND STATUS
The first Fundamental Combustion Research Applied to Pollution Control
Program (FCR-I) had three major objectives. These were:
•	To generate the understanding of combustor behavior necessary
to aid the Combustion Research Branch in the development of
control strategies to minimize N0X emissions from stationary
sources.
•	To develop engineering models which would allow effective
utilization of a large "body of fundamental information in
the development of new N0X control techniques.
•	To identify critical information necessary for low N0x
combustor development and to generate it in a time frame
which was consistent with the needs of the CRB technology
development programs.
The overall goal of the CRB, and hence, for the FCR-I program is to provide
the technology for the maximum control of N0X emissions from stationary
sources. The intent of FCR-I was to develop a focused program directed at
the important issues of relevance to the CRB in order that the results
could be applied. The program plan emphasized well-defined priority target
areas, and the ensuing research effort was always guided towards engineer-
ing solutions to specific problems. There was a conscious effort to avoid
projects which did not provide information critical to the needs of the
CRB. Consequently, it was necessary to isolate the relevant issues, and
the program was directed away from studied of the physics and chemistry of
combustion which were not relevant to the goal of the CRB.
Fifty-six percent of the annual emission of nitrogen oxides from
stationary sources in this country emanate from one general combustor
category, boilers firing coal or oil. The dominant characteristics in the
combustors can be characterized as follows:
The flames are large, with energy release zones whose dimensions
are on the order of tens of feet.
2

-------
The time mean motion is in steady state.
The pressure is atmospheric.
Radiation heat transfer has a dominant impact upon exchange
from the heat release zone to the cold confining walls.
Mass and thermal transport is turbulent.
Fuel/air contacting occurs by diffusion and particle penetration.
Fuels are injected as solids or liquids giving rise to two-phase
transport and homogeneous and heterogeneous reactions.
The high fuel-bound nitrogen content generally requires that NO
emissions are minimized by staged combustion with the associated
generation of copious levels of carbonaceous particulates, and
problems associated with the burnout of this material.
Thus, the primary initial objective of the FCR-I program was to immediately
establish a subcontractor-oriented program of fundamental investigations
focused on the simultaneous control of N0x and particulates from large, con-
fined, one-atmosphere turbulent diffusion flames burning heavy residual oil
and pulverized coal.
This section of the paper will describe the initial program plan, and
\
summarize the status of the various programs initiated by EER during FCR-I.
II. Program Structure
The combustion of pulverized coal and oil in boilers is accomplished
through the use of a burner of which there are many different designs. In
fact, Combustion Engineering claims that they do not use burners, just fuel
and air injectors, and the whole boiler acts as the burner. For the pur-
poses of this discussion, a burner will be considered as a device which
allows fuel and air to be injected in such a way as to provide a stable
flame whose characteristics are suited to a given combustion chamber.
Pulverized coal is injected either through annular or axial nozzles with
spreading devices. Liquid fuels are normally broken into liquids by
atomizers of many different designs which produce a variety of fuel spray
characteristics (spray angle, spray type, drop size). 'The combustion air
3

-------
is supplied axially without tangential momentum, or with tangential momentum
which may vary acorss the burner throat which might be composed of more than
one annular passage. Depending upon the fuel and air injection parameters,
two different types of flames can be readily distinguished, both of which
are shown in Figure 1. The top half of the sketch shows a near-field dom-
inated burner stabilized flame. These are relatively short, high-intensity
flames typical of wall-fired boilers produced with rapid fuel/air mixing and
each flame is stabilized independently, although there is some flame/flame
interaction. The second class of flames are typified by those used in cement
kilns or corner-fired boilers where the air is injected completely axially,
producing a long, simple jet flame (see the bottom half sketch in Figure 1).
In a corner-fired boiler these jet flames intersect on a firing circle and
mix in the fireball which occupies a significant fraction of a combustion
chamber. Regardless of the type of flame, the liquid droplets or pulverized
coal particles exhibit the same phenomena as their energy is converted from
potential chemical energy in the fuel to thermal energy in the combustion
products.
A simplified description of the phenomena occurring to solid particles
and liquid droplets of fuels in turbulent diffusion flames is illustrated
in Figure 1. The fuels first decompose giving a vapor and a solid product.
The vapor may be ejected from the solid as a single jet from a blowhole or
may vaporize uniformly from the particle or droplets. This vapor will then
either undergo pyrolysis producing soot particles, or undergo gas phase
oxidation. The solid remaining after devolatilization of coal is normally
referred to as char. The char and soot may then undergo heterogeneous
reaction which include oxidation (burnout) and possible reaction with gaseous
nitrogen specie. Throughout this sequence of events the fuel droplets/
particles and their products must be brought into contact with oxidant and
the high temperature combustion products in order to cause decomposition
and subsequent reaction. This contacting is important on a macroscale because
it dictates the major flame characteristics (i.e., near-field dominated burner
stabilized, or far-field dominated jet flame), and on a microscale because it
ultimately dictates the production of pollutant specie. FCR-I has been
4

-------
primarily concerned with the phenomena associated with the conversion of
the nitrogen in the fuel droplets/particles to the ultimate nitrogen oxide
released from combustion chamber.
FCR-I was divided into three program areas. These were concerned with
(1) transport processes in reacting systems; (2) gas phase chemistry; and
(3) the physics and chemistry of two-phase systems. Figure 2 presents an
overall view of the program structure divided into these major program
areas, specific program elements, and support areas. This program structure
was planned to lead to two major program outputs. These were:
•	A description of the chemical limits of NO production in order
to ascertain the lower bounds of both fuel and thermal NO pro-
duction under a series of process constraints which were not
limited in any way by fuel/air contacting.
•	A description of fuel NO formation in turbulent diffusion
flames for gas, liquid and pulverized coal systems.
This overall program was planned as a combined in-house and subcontract
project effort. EER took responsibility for program planning management and
synthesis of the overall program, and subcontracted to various organizations
separate projects included within the various program elements. Subcontracts
accounted for 70 percent of the program and were used to ensure that FCR-I
had the benefit of the best scientific talent available. Every effort was
made to use subcontractors who had the necessary experience and, in many
cases, equipment. The main synthesis role was accomplished by EER using the
analytical tools developed in one of the program areas.
The reasoning behind the division of the total program into the three
major program areas can most easily be described by consideration of the
phenomena occurring to solid and liquid fuels in turbulent diffusion flames.
A pulverized coal particle is heated, decomposes and the fuel-bound nitrogen
component is divided between that which stays with the solid (char) and that
which is evolved with the volatile gases. The processes that lead to the
division of fuel nitrogen are part of the physics and chemistry of two-phase
systems, as are the subsequent reactions of the solid phase. Thus, two of
5

-------
the program elemen.cs in the two-phase program area were concerned with
the thermal decomposition of the fuel and heterogeneous NO/char reactions.
The fate of the volatiles depends upon the chemistry of nitrogenous specie
during combustion of the gas phase pyrolysate composed primarily of hydro-
carbons. This whole area was treated in the gas phase chemistry program
area. The primary goal of the gas phase chemistry program was to produce
a kinetic mechanism which was adequate for the description of the fate of
fuel nitrogen in gaseous mixtures which were likely to be found in pulver-
sized coal and residual fuel oil flames. Thus, the gas phase chemistry
area was divided into two program elements: one experimental and one aimed
at mechanism development using computer simulations. A considerable effort
was expended in the development of two-phase reactor experiments. These
provided coupled experiments in which both gas phase chemistry and fuel
decomposition were involved.
All of the above neglected the physics of turbulent transport. In
flames the particles must be mixed with high-temperature gases before they
can decompose, and oxidant must be mixed with the volatile fuel fractions
before they can react. These fuel/air contacting processes were treated
in the program area defined as transport processes in reacting systems.
Three projects were carried out in this major program area. These were
concerned with turbulence:kinetics coupling, fuel injection systems, and
turbulent diffusion flames.
Although each of the projects in the various program areas were planned
to provide information which was of value in itself and was of direct use
for data interpretation, the overall program flow was directed towards the
development and verification of engineering tools. These tools could be
used in the attainment of the two major goals which defined fuel NO forma-
tion in turbulent diffusion flames. It is not possible to show the overall
interaction of the various program areas and program elements because of the
additional complexity which would be added to Figure 2. However, it does
demonstrate the importance placed upon the use of analytical tools to pro-
vide the program synthesis. Indeed, all the program elements can be con-
sidered as providing input to allow analytical tools to be developed, and
the application of these tools then provides the output of the FCR program.
6

-------
It was recognized from the outset that within the time frame of FCR-I
the development of a unified mathematical model for N0X production in pul-
verized coal flames was unattainable. The major modeling effort in FCR-I
concentrated upon the development of a framework, of semiempirical modular
models. Thus, a complex system could be modeled using a collection of
limit-case elements linked together by means of empirical knowledge of the
exchange of heat and mass transfer between these elements. In its simplest
application, these modular elements could be used to analyze the data gen-
erated in several of the program elements involving reactor experiments.
The modules describing fuel decomposition could be applied to both reactors
and turbulent diffusion flames. Recognizing how complex a pulverized-coal
system is, the limit-case approach was deemed to be the most appropriate in
analyzing fundamental data for application to real systems. In limit-case
studies certain phenomena are assumed to dominate the process while other
phenomena become suppressed, thus allowing the implications of specific
phenomenological behavior to be examined. Limit-case studies could be used
to define the times and temperatures needed to reach a certain total fixed-
nitrogen content in a fuel-rich combustion system, or in an attempt to define
to what extent heterogeneous NO reduction took place in pulverized coal
flames.
Although FCR-I placed considerable emphasis on the provision of informa-
tion which was necessary to provide near-term solutions, long-term research
was not entirely neglected. FCR-I pioneered the use of holography and two-
color pyrometry as tools to study the thermal decomposition and combustion
of fuels. Activities in turbulence kinetics coupling are beginning to show
considerable promise for the future. Effort was initiated to evaluate
numerical techniques which could be used in the ultimate models of furnace
performance. Projects such as these ensured the CRB with a balanced directed
fundamental combustion research program.
The following subsections will discuss the projects in the three major
program areas. Another section will describe the products of several proj-
ects which were defined as part of the Measurement Systems Program Support
area. Finally, this section will conclude with a review of FCR-I's achieve-
ments in the areas of modeling of fuel NO formation in reactors and flames.
7

-------
m. Gas Phase Chemistry and Heterogeneous NO Reduction
The projects initiated by FCR-I in this program area were:
•	Kinetic Mechanism Development, T. L. Corley, J. C. Kramlich, and
W. R. Seeker, EER.
•	The Modeling of Fuel Nitrogen Chemistry in Combustion, J. Levy
and A. F. Sarofim, Massachusetts Institute of Technology.
•	The Formation and Destruction of Nitrogenous Specie During
Hydrocarbon/Air Combustion, D. W. Blair and A. Myerson.
•	N0X Formation in a Flat Opposed-Jet Diffusion Flame, W. A. Hahn
and J.O.L. Wendt, University of Arizona.
•	Mechanisms of NO Reduction on Solid Particles, G. G. de Soete, IFP.
•	Reactor Studies, W. Clark., EER.
Kinetic Mechanism Development. T. L. Corley, J. C. Kramlich, W. R. Seeker,
EER—
This project was focused on the development of a kinetic mechanism which
could give an adequate prediction of the conversion of fuel nitrogen com-
pounds to NO in gaseous systems. The major objective was to provide a vali-
dated reaction model which could then be used as one component of more com-
plex models describing the conversion of fuel nitrogen during the combustion
of coal and liquid fuels.
The total chemical kinetic mechanism describing hydrocarbon combustion
and nitrogen conversion can be divided into a number of natural subsets.
Starting with the most simple and proceeding to subsets which require the
addition of more reactions, these subsets are:
-	H2/02;
C0/H2/02;
NH3/02;
HCH/02;
CH20/02;
8

-------
-	CH4/02;
-	
-------
of the mechanism can be based upon both its ability to predict burning
velocities and to reproduce species profiles.
Step 4 of the methodology refers to the manner in which particular
subelements are incorporated into more complex elements of the reaction
model. After an element of the mechanism has successfully passed the
appropriate tests it cannot be modified when used in the analysis of more
complex elements of the reaction model unless it is shown that the dis-
crepancies in the new system can only be corrected by changing features
of the subelement. If this step is required, the modified subelement
must again be checked against the relevant data to ensure consistency with
the old data. This feedback cycling ensures the internal consistency of
the final mechanism.
Table I summarizes the ability of the mechanism subelements to pre-
dict shock tube and flame data. Application of the procedure to elements
of the reaction model indicates:
•	CH^O - CO - Hg - O2 system is complete
HCO is important for chain-branching in rich, wet CO
flames
Decomposition and oxidation of CH^O is slower than pre-
vious predictions
•	NH^ - O2 mechanism requires additional analysis
Chain-branching path required: NH^ + NO ¦ + H + OH
Ignition delay controlled by: NH^ +	+ ^
species may be important for rich processing of NH^
Overall oxidation of NH^ too slow with present mechanism
•	HCN - 0^ reaction model must be compared with other data
Most of the Tecent work has been directed towards a description of
the CH^-O^ mechanism which was divided into two components:
•	C1 " chemistry
17 species: CH^, CH^, CH^, CH, CH^O + the CH^O - 0^ set
61 reactions
10
-------
C^ and - chemistry
23 species: ^^i (i = 1 to 6) + the set
101 reactions
The mechanism was first examined for correlation with shock tube
ignition delays and, in particular, the simulations examined the impact
of various products for the reaction CH^ + and the impact of the C£
chemistry. The principal result was that CH^O was a necessary component
in the mechanism to correctly predict ignition delays. It was not possible
to conclude that C£ chemistry was necessary.
Flat flame data of Biordi et al were adequately modeled without the
necessity of changing the mechanism. The flame velocities were in excel-
lent agreement (model ¦ 76 cm/sec; data » 80 cm/sec). The species profiles
were also in excellent agreement, with the exception of CO and C0£ (see
Figure 4). This divergence is consistent with the observation that the
rate of CO + OH = CO2 + H is too slow at high temperatures.
The data Harvey and MacColl (1979) provide a test of the - chemistry
portion of the CH^ mechanism; a number of	profiles are reported. Since
the flat flame was attached (i.e., burner stabilized) no flame velocity
check was possible. As shown in Figure 5, the species profiles indicated
that hydrocarbon intermediates were over-predicted. The inclusion of the
C^ chemistry set improved, but did not eliminate the divergence.
The Modeling of Fuel Nitrogen Chemistry in Combustion: The Influence of
Hydrocarbons. J. M. Levy and A. F. Sarofim, MIT—
There are uncertainties in the currently available mechanistic and
kinetic data base suitable for modeling of fuel nitrogen conversion in the
presence of "simple" fuels; however, an attempt must be made to attack the
problems of real fossil fuels. Real fuels introduce mechanistic complexi-
ties due to the presence of hydrocarbon specie which do not decouple easily
from solution of the fuel nitrogen problem. This project was concerned
with an assessment of these complexities and to the development of a meth-
odology for dealing with them which would allow fuel nitrogen processing
in real fuels to be modeled.
11
-------
The ability to assess the factors affecting bound nitrogen conversion
in a real-fuel, fuel-rich environment is dependent upon an understanding
of hydrocarbon decay kinetics because:
1.	Interactions of fuel nitrogen and hydrocarbon fragements
are significant, causing recycling of NO back to HCN.
2.	The radical pool (H, 0, OH) which drives the fuel nitrogen
chemistry is determined primarily by the hydrocarbon chemistry
which dominates the system at early times.
Several approaches to the problem of treating hydrocarbons were
considered:
•	Detailed kinetics rejected because of time, cost and lack of
detailed information.
•	Global and semiglobal methods may not adequately describe
the free radical pool.
Having exhausted, as impractical or inapplicable, existing methods of
treating hydrocarbons, a new quasi-global method was proposed with the
following general form:
Parent Hydrocarbon
Global
^ Distribution of Hydrocarbon Fragments
Detailed Kinetics
	co2 + h2o
The method assumes: (1) that the initial degration of the parent
hydrocarbon in the preflame zone via pyrolysis and oxidative pyrolysis is
very rapid; (2) that the initial decomposition products may be represented
by a relatively small set of low molecular weight, stable, and/or unstable,
hydrocarbon species; and (3) that the decomposition mechanism can be
justifiably represented by a global step.
12
-------
The potential utility of this proposed method, referred to as the
Quasi-Global Method Applied to Initial Reaction Extents (QUAGMIRE), lies
in the following:
•	The advantages of modeling free radical chemistry by detailed
kinetics are retained, but the magnitude of the elementary
reaction set required is vastly reduced;
•	Unlike the quasi-global proposed by others, hydrocarbons are
retained in the elementary reaction scheme.
QUAGMIRE is currently being used to model data generated in the post-
flame region of flat flames doped with nitrogen specie. Initial results
are encouraging and point to the need for more detailed measurement of the
relevant free radicals in hydrocarbon flames in order to provide a basis
for the initial global step.
The Formation and Destruction of Nitrogenous Specie During Hydrocarbon/
Air Combustion. D. W. Blair, Exxon—
Earlier work at Exxon has been funded by the CRB to develop two reactors
for the investigation of NO formation: a strongly backmixed jet-stirred
combustor (JSC) and an adiabatic laminar quasi-plug flow combustor - the
multiburner (MB). This project was continued by FCR-I to utilize this
expertise and equipment in several experiments:
•	Extend the measurements to include nitrogen specie other than
NO.
•	Investigate the impact of temperature on fuel NO formation .
•	Indicate experiments on fuel nitrogen conversion with hydrocarbon
fuels other than CH^.
This information was necessary if a kinetic mechanism for fuel nitrogen
oxidation was to be developed.
The following conclusions were drawn from Exxon studies:
•	With both equivalence ratio and flame temperature held constant,
specific hydrocarbon chemistry influences both thermal nitrogen
fixation and fuel nitrogen conversion.
13
-------
•	Fuel chemistry is more important in the multiburner than it
is in the jet-stirred combustor. The contributions of com-
bustion environment to this difference appear to exceed those
of reaction time.
•	While the conversion of fuel nitrogen is dependent upon fuel
nitrogen species, the dependency in the MB considerably exceeds
that in the JSC. This argues that differences in combustion
environment between the two combustors is important.
While there is a need for considerable additional work in this area
(extending range of variables, verifying results, expanding range and areas
of overlap between the several independent variables), this project has
been terminated with EER and further work is not contemplated at this time.
N0X Formation in Flat Laminar Opposed-Jet Methane Diffusion Flames. W. A.
Hahn and J.O.L. Wendt, University of Arizona—
The oxidation and pyrolysis of fuel nitrogen, as ammonia, was investi-
gated by first injecting ammonia in the fuel and then in the air, and then
comparing the predicted versus measured NO profile. The good agreement
between model and experiment when the ammonia was injected with the fuel
indicates that the pyrolysis reactions of ammonia in the presence of fuel
hydrocarbon fragments can be described by the mechanism used. When ammonia
was injected with the oxidant, the poor agreement with theory indicates that
the mechanism may not accurately describe ammonia pyrolysis in the presence
of CO, CO2, H^O and oxidant.
The effect of rate of stretching on the axial NO profile is well-predicted
by theory. NO formation and NO destruction mechanisms are of equal importance
in these types of flames, and the experimental and theoretical results are in
qualitative agreement with observed effects of increased turbulent intensity
on NO formed in turbulent diffusion flames.
Mechanisms of Nitric Oxide Reduction on Solid Particles. G. G. de Soete, IFP—
The scope of the work is to provide information on:
•	The mechanism of heterogeneous NO reduction under typical flame
and flue gas conditions.
14
-------
• Determine the reaction rates of reactions controlling the
disappearance of NO, the formation of intermediates, and
their transformation to molecular nitrogen.
The study performed on packed bed reactors was restricted to a general
investigation of the heterogeneous NO reducing mechanism.
It has been shown that the heterogeneous reduction of NO by solids
takes place via a complex series of reactions which are dependent upon the
nature of the solid and the composition of the gas phase. In particular:
When the solid contains neither hydrogen nor carbon, reduction
rates are negligible at temperatures below 1100°C unless gas
phase reducing agents are present.
NH3 has been identified as an intermediate if there is a source
of hydrogen.
HCN is produced with carbon-condensing solids.
- Oxygen can act either as an inhibitor or promoter of the NO
reduction reactions.
These studies were carried out under equilibrium conditions, and any extrap-
olation to flame conditions should be carried out only with the same
restriction.
Reactor Studies. W. D. Clark, EER—
A series of short reactor studies were carried out in-house to provide
Information to aid program planning and management. A high temperature plug
flow reactor was built originally to test model predictions which suggested
that NO decayed relatively rapidly in the absence of hydrocarbons in fuel-
rich systems. Experiments indicated that this decay did not take place,
and the reaction rate of the reaction CO + NO « NCO + 0 was found to be too
high.
The reactor was modified (see Figure 6) to allow an investigation of
heterogeneous effects (NO reduction and char nitrogen conversion) in a dis-
posed phase system.
15
-------
Particles were injected into NO-doped post-flame combustion gases to
investigate their NO reducing potential under realistic conditions. Effects
of temperature and particle concentration on the NO reducing properties of
alumina, graphite, char, and anthracite coal were examined. Alumina and
graphite had little effect on NO; however, char and anthracite coal caused
a decrease in NO with increasing particle concentration. Alumina particles
were injected into slightly lean and into lightly rich NO-doped methane
combustion gases at injection point temperatures up to 1100°C. Particle
3
concentrations up to 200 g/standard m had little effect on NO. Injection
of graphite particles also had little effect on NO at temperatures up to
1100°C and particle concentrations up to 70 g/sm^.
Figure 7 shows the effects of injecting low volatile char particles
into slightly rich NO-doped methane combustion gases. With the injection
of char particles, NO concentration dropped sharply, CO rose due to char
oxidation and volatile release, and CO^ and temperatures fell primarily due
to dilution. NO decreased with increasing feed rate (particle concentra-
tion) , and returned to its initial concentration when the particle injection
stopped. At an injection temperature of 900°C and a particle mass loading
3
of 180 g/sm , 30 percent of the NO was reduced (after correction for
reduction).
Figure 8 shows the effects of anthracite coal particles on NO. Reduc-
tions of up to 20 percent were caused by injection of anthracite into
slightly rich NO-doped methane combustion products. Further increases in
particle concentration had little effect on NO. Injection of anthracite
particles into slightly lean undoped methane combustion gases caused a
30 PPM rise in NO. This cast doubt as to whether the observed NO reduction
in the fuel-rich case was due to heterogeneous or homogeneous reactions.
However, other investigators have measured negligible amounts of NH^ and
HCN in anthracite combustion, even under very fuel-rich conditions. There-
fore, the observed reduction of NO by anthracite in fuel-rich combustion
gases was probably due to heterogeneous reactions.
16
-------
Physics and Chemistry of Two-Phase Systems
FCR-I projects which were included in this program area were oriented
towards providing data on the fate of fuel nitrogen during thermal decom-
position. Two investigations were initiated to study the fate of fuel
nitrogen from real fuels using two-phase backmixed reactors.
Volatility of Fuel Nitrogen. R. L. Gay, A. E. Axworthy, V. H. Payne,
and H. L. Recht, Rockwell International—
An experimental program measured fuel nitrogen volatility using a
quartz two-stage pyrolysis reactor developed under previous CRB programs.
Solid and liquid fossil fuels were heated in helium in the first stage of
the reactor at selected temperatures up to 1100°C. The volatile nitrogen
compounds released were measured by converting them to HON in the second
stage of the reactor at 1100°C. The HCN was collected in.dilute sodium
carbonate solutions and measured colorimetrically.
Samples of 20 fossil fuels which were used on other CRB programs were
supplied by EER. Petroleum-based fuel oils, synthetic oils from coal and
shale, and coals were studied. All of the samples tested produced the
greatest yields of volatile nitrogen compounds at a first-stage temperature
of 1100°C. The maximum yield of volatile nitrogen from petroleum fuel oils
was 35-48 percent of the nitrogen in the sample. The yield curves increased
continuously from 300°C with a slight dip at 600-75Q°C. The maximum volatile
nitrogen yields from shale oils were 70-79 percent, while the coal liquids
yielded 50 percent. The nitrogen species in shale oils were very volatile
at low temperature, giving HCN yields of about 65 percent at a pyrolysis
temperature of only 200°C.
The data generated by this experiment was used to correlate N0X emissions
from pulverized coal and liquid fuels. The results of this attempt were pre-
sented in Figure 9 which shows percent NO in combustion as a function of
percent HCN in pyrolysis at 110
-------
which allowed the prediction of the time and temperature dependent evolution
of the major products of pyrolysis from a knowledge of the coal's functional
group distribution. The model assumes that large molecular fragments
("monomers") are released from the coal "polymer" with only minor alterna-
tion to form tar, while simultaneous cracking of the chemical structure
forms the light molecules of the gas. Any chemical component of the coal
can, therefore, evolve as part of the tar or as an independent species.
The model may be used to predict the amount of volatile fuel nitrogen
evolved in pyrolysis which is essential if the formation of fuel nitrogen
is to be accurately assessed.
A study was initiated to apply these concepts to the study of several
coals which were being investigated in other programs. Particular attention
was paid to the rate and forms of nitrogen evolution. Pyrolysis experiments
were performed on Beulah, a North Dakota lignite, a Pittsburgh bituminous,
Savage, a Montana lignite, and Rosa, a bituminous coal, between 400°C and
1800°C.
A rank dependence in the retention of nitrogen in the char was observed
as indicated in Figure 10. The higher rank bituminous coals appear to
retain much more nitrogen at high temperature than do the lignites. Possi-
ble explanations for this could be the higher oxygen content or the smaller
aromatic cluster size in the lignite. Figure 11 shows some interesting
aspects of the evolution of nitrogen during pyrolysis. At temperatures
below 1200°C, nitrogen balance is obtained by measuring the tar and char
nitrogen, HCN and NH^. At higher temperatures, there appears to be some
missing material indicating some other nitrogen form evolved which is not
determined. Of the gaseous species, HCN yields are substantially higher
than NH^ and the ease of evolution of these two gases at low temperature
appears to be rank-dependent. As an example of this, efforts have already
begun to utilize the kinetic parameters derived in this experiment in the
cool reactor models being developed as part of the Analytical Tool Develop-
ment Support Area.
18
-------
The Physical and Chemical Effects Occurring During the Thermal Decomposition
of Coal Particles and Oil Droplets. W. R. Seeker, G. S. Samuelsen, J. D.
Trolinger, and C. F. Hess, EER/SDL—
The events occurring during the thermal decomposition of liquid droplets
and coal particles in the initial stages of heat release in turbulent dif-
fusion flames are of considerable significance to the designers of low pol-
lutant emission combustors. In pulverized coal flames the partition of fuel
nitrogen between the volatile and char fractions, for example, will impact
the production of fuel NO. Moreover, since the fate of fuel nitrogen in the
volatiles is dictated by the environment in which they react, their conver-
sion to NO or N2 will depend upon the degree to which they mix with the bulk
flow before reacting. In liquid fuel flames the minimum achievable N0X emis-
sions is quite often related to the trade-off between N0X and smoke emissions.
Therefore, the factors controlling the formation of particulates under fuel-
rich conditions are germane to the problem of low N0X burner design for
liquid fuels.
Earlier studies supported by FCR-I demonstrated that holography could
be used to observe the behavior of coal particles during combustion. The
approaih taken in the present study was to observe particles/droplets under
well-controlled conditions which simulated those encountered in real systems.
This was accomplished by the construction of a reactor which allowed the
particles/droplets to be injected into high temperature gases whose tempera-
ture and composition could be varied. Diagnostics were designed to allow
visualization of the particles/droplets during combustion, both by holygraphy
and high-speed photography. Two-color particle pyrometry was employed to
measure particle temperatures, and solid and gaseous samples were extracted
to investigate the variation of composition with time.
The major goals were to explore the effects of controllable parameters
on combustion phenomena and pollution formation. The specific parameters
investigated included the following:
•	The effect of coal type. The coals selected were those used
in other CRB programs.
•	The effect of particle/droplet size.
19
-------
•	The effect of particle number density.
•	The effect of environment, i.e., either reducing or oxidizing.
•	The effect of relative velocity between the particles and
background gas,
•	Exploratory investigation using liquid droplets.
Initial evaluations of the results obtained in this study to date on
the physical phenomena which occur during the thermal decomposition of pul-
verized coal particles indicates that:
' • Volatile coal fractions are ejected from coal particles in
jets, and that for large bituminous coal particles these jets
produce a trail of small particles.
•	Coal composition and volatile evolution rate influences
particle temperature.
•	Large soot structures can be formed from the bulk, gases pro-
duced when bituminous coal particles decompose.
Pollutant Formation from Combusting P-ulverizad Coal Clouds. P. H. Goldberg,
H. long, and R. Kendall, Acurex—
The objective of this experimental study was to provide information on
the mechanisms of coal combustion and subsequent oxides of nitrogen forma-
tion under conditions similar to those found in the near-burner region of
a high-intensity pulverized coal flame. A jet-stirred reactor system is
utilized in this study to simulate this type of combustion environment.
The conditions in the stirred combustor approximate those found in flames
when a high degree of recirculation is present. In addition, a stirred
reactor is capable of operating over a wide range of coal concentrations at
high temperatures and residence times ranging from about 20 to 100 msec.
An overview of the experimental program plan by tasks is presented in
Table II. Referring to the table, the first goal of the program was to
design a well-mixed coal reactor capable of operating over a residence time
range of 20 to 200 milliseconds. After completing the design and fabrica-
tion, noncombusting mixing studies were to be conducted to verify that a
20
-------
high level of gas phase mixing exists in the reactor. Once gas phase mixing
performance was verified, subsequent experiments were to be directed at
quantifying particle mixing behavior. Combustion experiments were to begin
after the completion of this task.
The original experimental plan was modified somewhat during the course
of this program. Instead of one reactor, two similar reactors were designed
to cover the residence time ranges of about 20 to 60 and 50 to 100 msec,
respectively, during combustion testing. The gas phase mixing efficiency
of the longer time scale reactor (SRII) was examined using a "cold" model
reactor in conjunction with an He tracer method. Attempts have been made
to quantify particle mixing in the SRII cold model by using a light attenua-
tion technique. Experiments indicate that good mixing is achieved in SRII.
However, the particle mixing study results indicate that the particle resi-
dence time might be longer than the gas residence time. The results also
indicate that the particle residence time may vary significantly less than
that of the gas.
The combustion results show that devolatilization may be essentially
completed in the SRII residence time range. The exhaust concentrations and
temperature dependence on equivalence ratio is similar to those expected in
larger time scale systems. This might lead to the conclusion that enhanced
volatile yields play a major role in combustion.
It has been observed that fuel ni.trogen conversion is correlatable
with equivalence ratio, reaction efficiency, and temperature. The trends
in these correlations show similarities with other experimental work.. It
was also found that the nitrogen depletion of the char corresponds roughly
to the weight loss exhibited by the char, although a limited range of
weight loss has been examined to date (75-95 percent).
Pollutant Formation During the Combustion of Residual Fuel Oils in Backmixed
Reactors. M. J. Murphy and A. Levy, Battelle/Columbus—
This program was included in FCR-I in an attempt to relate the produc-
tion of N0X from burning residual fuel oils to fuel nitrogen in these fuels
and to the combustion variables such as spray droplet size, fuel properties
and time/temperature history. This objective was addressed through the use
21
-------
of a well-stirred reactor fired on fuel oil having the capability of
variable residence time, stoichiometry and droplet parameters (size and
fuel composition). The program was organized into three tasks:
1.	Design, Construct and Test a Reactor Model,
2.	Construct and Test the Actual Reactor,
3.	Demonstrate Reactor Operation and Data Acquisition.
The performance of the well-stirred reactor design (Task 1) has been
verified by extensive cold flow testing. Criteria for well-mixedness were
found to be met satisfactorily with the selected design. Measurements of
gas residence times were performed on the cold flow model using laser
extinction techniques applied at the exhaust duct.
Technical problems associated with the prototype combustor have led to
the design of the current combustor configuration. This combustor has eight
fuel jets utilizing Sonicore nozzles and eight air jets fired into a cylin-
drical chamber. The chamber dimensions were selected to allow a residence
time range between 20 and 200 msec at a firing rate variable between 100,000
to 1.5 x 10^ Btu/hr. Emissions data have been collected for natural gas
burning over a stoichiometry range of 0.6 to 1.60 as a function of residence
time. The combustor has been successfully fired for extended periods on a
No. 6 fuel oil to equivalence ratios greater than 2.0.
Transport Processes in Reacting Systems
The major program areas discussed in the two previous sections were
concerned with types of reactants and their probable products which would
be found in heavy liquid or pulverized coal flames. The projects included
in the program area defined as Transport Processes in Reacting Systems were
concerned with problems associated with reactant mixing.
An Experimental Approach to the Study of Heavy Oil Spray Combustion in
Shear Layers. A. Vranos, UTRC—
This program was included during FCR-I and consisted of two phases:
(1) design and fabrication of an apparatus, and (2) preliminary combustion
experiments. A unique shear-layer mixing and combustion apparatus was
22
-------
developed (see Figure 12) which simulates high-shear diffusive combustion
with heavy oil droplet injection as found in the near-field combustion zone
of boilers and furnaces. The apparatus provides: (1) two-dimensional mixing
and combustion of uniform streams of air and rich combustion products fed
from opposite sides of a splitter plate; (2) uniform injection of a mono-
disperse heavy oil spray into the shear layer by means of a. linear array of
fuel injectors located beneath the combustor wall; (3) control of droplet
path, residence time, and extent of vaporization; (4) control of hot- and
cold-stream inlet properties; and (5) three-dimensional probing of the flow
field.
The experimental phase of the program consisted of: (1) verification
of uniform properties of the hot and cold streams at the splitter plate
trailing edge; (2) demonstration of high injection velocities and uniform
droplet size and penetration with Nos. 2, 5, and 6 oils in vibrating multi-
capillary injectors; (3) determination of stable flow regimes as a function
of primary and secondary stream velocities; (4) flame stability and flash-
back studies; (5) high-speed cinematography of the shear layer with injection
of Nos. 2 and 6 fuel oil; and (6) preliminary flame probing.
The Application of Droplet-Sizing Interferometry and Holography to the
Measurement of Spray Droplet Size. C. F. Hess, W. D. Bacha.lo, SDL—
Two optical techniques were used to characterize the droplet field pro-
duced by a Sonicore nozzle under laboratory conditions. Droplet-sizing
interferometry (DSI) and holography were used to obtain the size and velocity
of fuel droplets in the presence of a surrounding airstream. The experi-
ments were performed under cold conditions to provide data on droplet size
and velocity correlations, size stratification, and droplet evaporation.
Due to the complex nature of the flow (two phases, small droplets, high
numher densities, recirculation, etc.) nonintrusive measuring methods must
be used to avoid disturbing the flow. The methods must also have very high
resolution since the droplets can be very small and move at high speeds.
The techniques must be flexible enough to access the measuring volume from
any direction since physical obstacles will be present in most combustion
applications. Droplet sizing interferometry proved to be a very strong
23
-------
technique with which to study these fuel sprays since it meets the criteria
listed above. Holography was also used since it provided a quick profile
of the shape of the plume and its resolution was adequate to confirm the
results obtained with the DSI.
It was found that:
•	The DSI instrument proved to be very powerful for the study of
fuel sprays in that:
it measures the size and velocity of individual droplets;
it provides a point measurement;
it can be used to distinguish air velocity from droplet
velocity.
•	Holography is adequate for sizing droplets, but the analysis
of holograms is very time-consuming.
Spray Characterization. W. R. Seeker and G. S. Samuelsen, EER—
A program was established to help explain data obtained on a CRB-funded
development program which showed the dependency of N0X emission in fuel oil
combustion on atomizing nozzle type, and nozzle operating conditions. Toward
this end, a cold chamber spray rig, patterned after the combustion system,
was built for the purpose, of characterizing nozzle spray behavior. As the
initial step in the program, nonintrusive optical techniques are being
applied and evaluated for (1) consistency of data measured, and (2) appli-
cability to studying the spray behavior most likely important to N0X emis-
sion. The optical techniques include diffraction, visibility, and holography.
At this juncture in the program the visiblity measurements have been
concluded. Significant differences were found to exist between these data
and data acquired in an independent program conducted at the IFRF. It
appears that these differences can be attributed to:
1.	Difference between the spray rig configuration and properties
of the fuel used on the two studies;
2.	Differences associated with the actual parameters being measured
•such as size distributions averaged across the spray (line
24
-------
integrated) as opposed to point measurements, and size
distribution measured in terms of volume versus number which
must be transposed.
3. Differences associated with the resolution limits of the vari-
ous instruments. For example, when taking holograms through
dense sprays, the resolution limit is known to increase to near
20 ym. There is uncertainty associated with the defined resolu-
tion limits of some of the newer techniques which must be well-
defined in order to directly compare with other instruments.
Development of a Coherent Flame Model for Turbulent Chemically-Reacting
Flame. F. E. Marble, and J. E. Broadwell, CIT/TRW—
A coherent flame model was applied to a methane air turbulent diffusion
flame with the objective of describing the production of nitric oxide. The
example of a circular jet of methane discharging into a stationary air atmos-
phere was used to illustrate application of the model. In the model, the
chemical reactions take place in laminar flame elements which are lengthened
by the turbulent fluid motion and shortened when adjacent flame segments
consume intervening reactant. The rates with which methane and air are con-
sumed and nitric oxide generated in the strained laminar flame are computed
numerically in an independent calculation.
Detailed Measurements of Long Pulverized Coal Flames for the Characteriza-
tion of Pollutant Formation. R. Payne, IFRF—
This program was initiated to provide experimental data which could be
used both for model development and to provide a mechanistic understanding
of nitrogen processing in flames of this type.
Six "long" coal flames were produced with the same burner under similar
boundary conditions. The differences between these flames were mainly due
to air and fuel input velocities and temperatures which have a direct effect
on the mixing characteristics and the coal heating rates. A comprehensive
set of flame data was generated. Despite the many difficulties encountered
in making measurements in such flames, particularly for sampling in regions
of high temperature and solid concentrations, the consistency between the
various measurements is generally good. Indications are that globally, the
25
-------
combustion is controlled by mixing and that the influence of fuel kinetics
is basically the same in all cases. In particular, NH^ concentrations were
always very low, while HCN was found to form very rapidly and in large
amounts.
Measurement Systems Support
Three efforts were carried out in this Program Support Area:
a study to assess the accuracy of chemiluminescent analyses
when measuring NO in combustion products after calibration
with NO in nitrogen;
tests to standardize the use of ion-specific electrodes for
HCN and NH3 analysis; and
a study to assess the problem of sampling sulfur and nitrogen
specie in "dirty" flames.
Analytical Tool Development
The major integration function in the FCR-I program was to be carried
out through the use of mathematical models. The models were developed in
close accord with the experimental investigations. These models include a
description of the following phenomena:
particle thermal2 history, including the existence of particle
supertemperatures,
finite rate thermal decomposition,
heterogeneous chemical behavior,
finite rate gas phase chemistry,
fluid mechanics and ballistics of turbulent transport in two-
phase systems,
complex flow fields associated with the near-field of highly
swirl-stability combustors,
the interaction between turbulent unmixedness and chemistry.
26
-------
These modeling efforts have resulted in a hierarchy of analysis codes.
Treatment of the fully-coupled finite rate processes associated with "particle
heat up, thermal decomposition, heterogeneous reactions, and gas phase chemis-
try was emphasized. The codes can handle with ease and efficiency as many as
200 such coupled processes, although often such complexity is not warranted.
In the course of developing an understanding of the dominant mechanisms
governing complex and poorly understood behavior, it is often necessary to
start with a large coupled system and, through sensitivity analysis, reduce
it in such a manner that it contains first-order phenomena only. A good
example of a situation requiring such an approach is the development of an
understanding of N0X generation in p.c. flames. The complexities of HCN,
NH_^, and C£ chemistry, in conjunction with finite rate devolatilization,
defies a simple intuitive understanding and requires complex analysis in
order to generate such intuition.
The decomposition of coal is described assuming that the coal is com-
posed of an arbitrary number of functional groups; the quantification and
characterization of these is provided by utilization of independent experi-
mental data (the work of Solomon; The Characterization of Coals During
Thermal Decomposition). These functional groups are envisioned to decompose
by one or more parallel paths with Arrhenius-like rate behavior. A given
species may evolve from more than one functional group, and the functional
groups can exchange between themselves with an equilibrium composition that
is temperature-dependent. This generalized model lends itself easily to
the incorporation of observations from many different investigators.
The hierarchy of models developed during FCR-I allow a description of
the following systems:
•	Simple Reactors
•	Simple Diffusion Flames
•	Long Turbulent Diffusion Flames
Simple Reactor Models—
These operational codes are routinely used for the analysis of well-
stirred (SR) and plug flow (PF) reactors for both gas and coal. The SR
27
-------
module assumes either a perfectly stirred exponential distribution of
particle residence times, or it will accept as input an arbitrary distribu-
tion obtained from experimental evidence, as will the PF module. The heat
loss or heat-loss distribution can be specified as appropriate. The modules
can be coupled, and the PF can receive additional combustion air and fuel in
a staged fashion. These coupled reactors are ideally suited for simple
modular modeling of combustion system performance (ignition stability, energy
release distribution, heat transfer) and pollutant formation as a function of
fuel, burner and furnace characteristics.
Simple Laboratory Diffusion Flames—
These codes provide for the analysis of fundamental experiments on
laminar one- and two-dimensional diffusion flames. In addition to simple
coaxial flames, this code will describe behavior in opposed-jet flames.
The mathematical solution for such a system proves to be self-similar with
all properties independent of the radial coordinate with the exception of
the radial velocity.
Long Turbulent Diffusion Flames—
This code represents the most ambitious, and perhaps the most satisfying,
output of FCR-I. It is directly applicable to a large class of industrial
furnaces. This operational code treats long coaxial turbulent p.c. diffusion
flames typified by tunnel2 furnace flames stabilized through convective back-
flow along the walls, as well as by radiation from the flame to the incoming
particles. The model and its free shear layer equivalent can also be used
as modular elements in conjunction with the idealized stirred and plug flow
reactors in the modular modeling of staged systems and high swirl-stabilized
burners.
28
-------
Burner
Fuel
=pf
Irculatljfu products
TUTTTi
Combustion
Air
External Recirculating
Products
Near Field Dominated Burner
Stabilized Flame.
Far Field Dominant
Jet Flame
Contacting
Nixing
Pyrolysis
Fuel
Parti cl es/l)rO|il e ts
°o
o00°
O o
Thermal
Decomposition
Soot
Char
Gas Phase Reaction
Oxidant
Contacting
Heterogeneous Reaction
Heat Transfer
Products
Figure 1. Solid Liquid Fuel Processing in Turbulent Diffusion Flames.
-------
F C * - 1 MOSHM STRUCT J HE
Major ?rt^nm A/tts
Trwiwort 9w*u ta
XM^igreSiwtaiii:
phytic* «r» c>Mt»ry
S*Jt4«
Ui nwt ifafciftry
Turbul«nc*/
Klmtlc*
Fi*l TTwrmtl
Oacaaoofftion
RMCtar Fl
£xMHMfitS
FimJ [nJtcflOfl
Tm »tv«s«
Tur*il««
Offfujien fl
rtdt»rog«n«rt
>10 Reduction
SffWlMMdt
n«Mnu
I'raqna Suepprt Arwts
iM rtMM
tmctor
fedulM
01ffua1o»
n«i .
-------
u>
SIMPLE
FUELS
1
I	
I
	I
FUF.L
NITROGEN
fill
r
HYDROCARBONS
I
I
I
r
i
1



¦—J—




CH,,

c2n2

C2],'l

c2H6
T
I HYDROCARBON
| COMBUSTION
FUEL NITROGEN
PROCESSING IN
n
Figure 3. Overview of Kinetic Mechanism.
-------
Distance from Burner Face (cm)
Comparison of Gas Phase Mechanism to CH^/Air
Figure 4.	Flame Profiles of J. C. Blordi et al
(15th International Symposium on Combustion, p,917, 1974)
-------
2400
2000 ^
1600
1200
CL
800
400
0.8
Solid =» data
Dashed = model
0.6
0.4
0.2
Distance, nvn
Figure 5. (17th International Symposium on Combustion* p. 857 (1979))
Stoichiometric CH^-^-Ar Flame Profiles From R. Harvey and A. MacColl
-------
PftEMIXED, DOPED RfACTANTS
	 MATER
ori'ASs.
NITROGEN
ftAT FLAME BURNER
CRITICAL flOM
CAS METERING
MIXING PANEL
PRESSURE
CONTROL
VALVE
GRADED
INSULATION
STEEL SHELL
n in
DILUTION
AIR
VACUUM —•
EXHAUST
DILUTION HAMPER
PARTICLE INJECTOR
VACUUM DAMPER
HEATED SAMPLE
LINES
•y THERMOCOUPLES
MATER
PARTICLE STREAM
SAMPLE WATER FILTER
PUMP TRAP
Figure 6. Heterogeneous Reactor System.
-------
TIME (min)
300
950
900
850
600
550
500
Temperature at Injector
Temperature at Probe
Figure 7. Heterogeneous Effects of Coed Char.
35
-------
NO Formation from Anthracite Coal
150
-o

i-
a
a.
o.
50
1

A
£
Feed Rate (g/min)
I
Methane Lean Combustion Gas
Undoped
A - Anthracite Coal
Injection Point Temperature - J070°C
Corrected for Dilution
100
-o

c
a
90
80
A
&
50
i
100
I
NO Conversion vs Anthracite Coal Feed Rate
Methane Rich Combustion Gas
500 ppm NO Doped
A - Anthracite Coal
Injection Point Temperature
Corrected for Dilution
70
Particle Mass Loading(g/standardm^)
1
1
0	2	4
Feed Rate (g/min)
1070°C
Figure 8. Heterogeneous Effects of Anthracite Coal.
36
-------
CO
50
z
o
fc
m 40
S
o
o
z
o
30
w 20
H
z
ut
o
cc
UJ
a.
10
BC19
• MS11
• TRW2
% TO NO = -0.013 + 1.99 (% TO HCN)i
0	5	10	15	20	25
PERCENT FUEL -N TO HCN IN PYROLYSIS AT 1100°C
Figure 9. Percent NO in Combustion vs Percent HCN in
Pyrolysis at 1100°C.
-------
hosa
l' | CIIAtl N 10 set
¦ CIIAH N — 4 sue
MUOLfttll WMIOM Ilr»rt«»Il»i 101U t>
PITTSBURGH BITUMINOUS
¥	1 w
a CMMl N

M UOlIt IIII I0M IfMTfMIUaf (ML C»
BEULAH
(3 CHM M
MMH.nl IL 17*1 IOM TCNPtRftTMt 
SAVAGE
Q (HM N
m
m
~
a
~
aam
m
ma
MUOtMllimiMI ItHPtMTUftf (ML C)
Figure 10. Temperature Dependence of Nitrogen Retention in Chars.
-------
nosA
¦f HM3 M
X * HCM M
¦f ~ CNMI M
sue I It >1 I) Al II Mil HAIUIU ¦
JM
WWLA1ILUAII0N 1tl*(««TUti (OCC O
PITTSBURGH BITUMINOUS
+ ~ CMM M
MUOlftf 1LI7MI0M HNttlfAltMC (OIL C>
!*'i gure 1J .
Evolution of Nitrogen
BEULAH
-f NMl N
X • MCH M
+ ~ IM N
+ ~ CHAfl M

 x
X
1-.+
X
4
Xf ^
3;
r
KMM.AIIU2AT ION HMrflAllMC 

SAVAGE
4 HH] N
X ~ MCH M
+ ~ (Alt H
~ « CMft* M


X
~ X
X *
-K-4-—
DfVOt AIIMZAIIOM HftPitrtlURf ( Dl L C>
i
*
4 1110'
Lignites and Bituminous Coals.
-------
nW'S
OMHHOW

i h )ta ^
SAMinWOH^*
s

O® *®
NtOVM"
OU/VtW ?
t
v in n <

1)1 0

1)1
»N-«0U,n
Figure
e 12.
Heavy
Oil Spray
Combustor.
-------
TABLE I. MECHANISTIC DEVELOPMENT PROGRAM RESULT SUMMARY
Experimental
System
Reactants
Measured
Quality of
Comparison (a)
Shock Tube
Flat Flame
Burner
Shock Tube
Flat Flame
Shock Tube
Shock Tube
Flat Flame
Burner
Shock Tube
Shock Tube
V°2
H2-°2
C0-H2-02-Ar
CO-air
CH 0-0 -CO-Ar
^ L
CH.-O.-Ar
4 I
CV°2
CH.-O-Ar
4
CH.-air
4
NH -O^Ar
NH„
Ignition Delay
Exponential Growth
Constants for OH
Freely Propagating
Burning Velocity
Flame Temperature
and Species Profiles
(1)	Exponential Growth
Constant
(2)	Exponential Growth
Constant for CO
(3)	Time for CO to
Reach a Given Value
Freely Propagating
Burning Velocity
Ignition Delay
Ignition Delay
(1)	Freely Propagating
Burning Velocities
(2)	Flame Temperature
and Species Profiles
Ignition Delays
Pyrolysis Profiles
***
ickji
***


***
*	(b)
*	(c)
***
**

"kit
***
(d)
(a)
(b)
Quality of comparison: *** = excellent; ** = good; * = fair.
Cause by rate of CO + OH CO+ H being too low above 2000 K.
^ Difficult to estimate water content of reactants from most papers.
Modeled burning velocity is a strong function of water content when
water is a trace specie.
^ Suggested that additional species (i.e.,	were necessary to
correctly model NH3 pyrolysis.
41
-------
TABLE II. ACCUREX STIRRED REACTOR PROGRAM PLAN
Task	Description
1	Design stirred reactor to be well-mixed on gas over
the residence time range between 20-200 msec
2	Verify (optimize) gas mixing intensity using a
"cold" model of the reactor
3	Qualtify coal particle mixing vehavior in the
stirred reactor
4	Proceed with coal combustion tests
•	Covering the reactor residence time range at
equivalence ratios up to A
•	Performing wet sieving tests and chemical
analysis on selected char samples
•	Using conventional gas analysis, supplemented
with NH3 and HCN measurements during selected
tests
42
-------
TWO PHASE PROCESSES INVOLVED IN THE CONTROL OF
NITROGEN OXIDE FORMATION IN FOSSIL FUEL FLAMES
By:
J. M. Beer, A. F. Sarofim, L. D. Timothy, S. P. Hanson,
A. Gupta, and J. M. Levy
Department of Chemical Engineering and the Energy Laboratory
Massachusetts Institute of Technology
Cambridge, Massachusetts 02139
43
-------
ABSTRACT
The conversion of fuel-nitrogen to nitric oxide in flames is dependent
upon a number of physical and chemical factors, three of which are discussed
in this paper: the rate of evolution of fuel nitrogen by heavy fuel oils,
the temperature-time history of burning coal particles, and the kinetics of
the reduction of NO by char.
Measurements of the nitrogen evolution of a stream of 150 ym fuel drop-
let injected into a heated helium stream were made for a Raw Paraho Shale
oil and an Indo-Malaysian residual fuel oil. The nitrogen evolution during
vaporization of the dispersed oil droplets is found to depart significantly
from that obtained under equilibrium distillation. For a Paraho shale oil
the rate of nitrogen evolution under the rapid heating experienced by the
droplets is retarded relative to that observed under equilibrium conditions.
By contrast preferential vaporization of the nitrogen was observed for an
Indo-Malaysian residual fuel oil. The temperature-time history and burning
times of coal particles burning singly were determined by two-color optical
pyrometry in order to provide insights on the role of volatile combustion on
nitric oxide formation. The burning times and intensity traces showed
that 100 ym particles of a bituminous coal produced a detached volatile
flame which was not evident during the combustion of smaller 40 ym particles.
The last part of the paper summarized data on the kinetics of NO reduction
by char, the enhancement of the rate of reduction in the presence of CO and
the inhibition of the reduction reaction by 1^0.
44
-------
ACKNOWLEDGEMENT
We are grateful for the proficient laboratory assistance of Mr. Anthony
Modestino. Ihe two-color pyrometer studies were ably performed by Diana
Altrichter. Support for this work from the U.S. Environmental Protection
Agency is gratefully acknowledged. The authors gratefully acknowledge
guidance and advice from Steve Lanier and Blair Martin.
45
-------
SECTION 1
INTRODUCTION
Control of fuel nitrogen conversion to NO in the process of combustion
X
of high-nitrogen content fuels requires understanding the chemical transfor-
mations of fuel nitrogen during the inital pyrolysis and subsequent com-
bustion of coal particles and fuel droplets which can lead to the forma-
tion of molecular nitrogen and nitrogen oxides in variable proportions.
Schematic pathways of fuel nitrogen transformations during pyrolysis and
oxidation in combustion processes are illustrated in Fig 1. In coal com-
bustion the fuel nitrogen evolves during the devolatilization of the
coal and the subsequent combustion of the individual char particle. The
fraction of fuel nitrogen evolved with the volatiles varies with coal
type, particle size and temperature. In the case of fuel oil droplets
the fuel nitrogen becomes available for reaction in the gas phase through
vaporization; in multicomponent fuel droplets the nitrogen evolution
will therefore depend on the distribution of fuel nitrogen over the
various boiling fractions, the droplet size and the rate of droplet
vaporization. Understanding the details of the non-equilibrium
distillation process is necessary for determining the nitrogenous pro-
duct distribution and hence the rate at which nitrogen containing
compounds become availabe for gas phase reactions. The first two parts
of this paper are concerned with the presentation and discussion of re-
sults of experimental studies on the evolution of fuel nitrogen from
coal particles and fuel oil droplets under conditions similar to those
in practical flames.
46
-------
In heavy fuel oil droplets the nitrogen concentration distribution
is normally biased towards the higher molecular weight hydrocarbons. Heavy
fuel oils are not completely vaporizable and the amount and nature of the
coke residue depends, for a given oil type, on the drop size and the rate
of vaporization. It can therefore be expected that the evolution of
nitrogen from a heavy fuel oil droplet during its vaporization in a flame
will be significantly different from the results of equilibrium distillation
obtained under slow heating conditions. For the study of fuel nitrogen
evolution in small fuel droplets representative of those in fuel sprays of
industrial flames, a laminar flow reactor was developed with provision for
controlling the droplet size and the thermal and chemical environment of
freely suspended monosize droplet streams. Results are presented in this
paper of the time resolved nitrogen evolution from a shale oil and residual
petroleum fuel oil. These results illustrate the effect of rapid heating
of multicomponent fuel droplets upon the fuel nitrogen evolution and the
significant departure from results of equilibrium distillation under high
rates of droplet vaporization.
The second part of the paper presents measurements of the temperature-
time history of pulverized coal particles during combustion. In earlier
work carried out by Pohl and Sarofim (1) and by Cheng (2) the rate of
evolution of fuel nitrogen during pyrolysis and combustion of pulverized
coal particles was determined. While the thermal environment of the
devolatilizing particles in an inert atmosphere is well defined, the combus-
tion of volatiles in oxidizing atmospheres makes this less well determined.
The existence of a volatile flame, the chemical composition of the volatiles,
the position of the volatile flame in relation to the particle surface all
influence the temperature-time history of the thermally decomposing,;
devolatilizing coal particle. The development of a two color optical
temperature measurement technique capable of following the variation of
the particle temperature during combustion made it possible to obtain
additional information of the interaction between the volatile flame and
the devolatization and combustion of coal particles.
47
-------
The third part of the paper deals with the heterogeneous reduc-
tion of NO on a char surfare. This reaction can play an important role
in reducing the formed NO concentration in coal combustion processes.
In an earlier study Pereira and Beer (3) have obtained data on the rate
of reduction of NO in fluidized coal combustors; Chan (4) and Sprouse(5)
have determined kinetic parameters of this reaction in the fluid-
ized coal combustion temperature range and Song (6) obtained kinetic
data relevant to pulverized coal combustion. In the following,results
of laminar flow reactor experiments are presented including time resolved
reaction product concentration variation	CO) and NO-Char reactors
rates as affected by CO and water vapor in the feed gas.
The three studies presented in this paper are self contained
separate investigations which, however, are closely interrelated through
their contribution to the better understanding of the details of the
transformations of fuel nitrogen in pulvferized coal and liquid fuel
spray flames.
48
-------
SECTION 2
FUEL-NITROGEN EVOLUTION FROM VAPORIZING HEAVY FUEL OIL DROPLETS
For the study of fuel-nitrogen evolution from liquid fuel droplets
during pyrolysis and oxidation a laminar flow reactor was designed with
provision for, feeding a monosize droplet array into a laminar gas stream
and the capability of intercepting the droplets after a chosen residence
time by a fast quenching sampling probe. Results of experiments with the
pyrolysis of 150 ym droplet streams of Raw Paraho Shale and Indo-Malaysian
residual petroleum fuel oil are presented in Figs 2 and 3. Equilibrium
distillation data for these two fuels are shown in Figs 4 and 5 respectively.
The data points in Figs 2 and 3 are calculated from the weight loss due to
vaporization between the point of the droplet injection and their capture
by the sampling probe, and from the nitrogen concentration of the residue
collected at the sampling position. The equilibrium distillation curves
(Figs 4 and 5) were obtained by slowly heating the fuel oil at atmospheric
pressure. The distillation curves show that in the shale oil the nitrogen
is quite uniformly distibuted over the different boiling fractions in contrast
to the heavy petroleum fuel oil where the nitrogen is concentrated in the
high temperature boiling fractions. Under conditions of rapid heating,
however, the above trend is reversed. The shale oil droplet at the beginning
of its vaporization yield little nitrogen and the distillation curve shifts
more and more to the left as the furnace temperature and hence the vaporiza-
tion rate is increased. Once thelow nitrogen content, low boiling fraction
compounds are vaporized the slope of the distillation curve is maintained
at nearly 45° indicating that the fraction of nitrogen evolved relative to
the fraction of mass vaporized is constant. The heavy petroleum fuel shows,
however, a different picture; here the nitrogenous compounds are concentrated
49
-------
in the high boiling fractions and after an initial period in which little
nitrogen is released, rapid nitrogen evolution follows resulting in a much
higher total nitrogen loss per vaporized mass then under equilibrium distil-
lation conditions. While the details of the non equilibrium distillation-
pyrolysis process are far from being completely understood it is thought
that the nature of the nitrogen evolution in heavy fuel oil droplets can
be explained by diffusional limitation of the mass transfer within the
liquid droplet. At high rates of vaporization the low nitrogen content, low
boiling fraction is depleted at the particle surface because the transport
of the low boiling compounds from the interior of the drop to the surface
is slow compared to its removal from the surface by vaporization. As a
consequence, the nitrogen concentration at the surface increases and, as
the temperature increases to the boiling point of this surface layer, nitro-
gen evolves at a strongly increased rate. The low boiling fraction trapped
in the droplet is superheated; this can increase the drop size by internal
vaporization and in the limit can disrupt the droplet by microexplosions.
The unexpected trend in the nitrogen evolution from heavy fuel oil droplets
and shale oils under rapid heating conditions has important implications
for staged combustion applications. It is clear that equilibrium distilla-
tion may provide a poor measure of the time over which the majority of the
fuel nitrogen is released, a parameter of importance in locating the point
of injection of the secondary air. It is emphasized, that the above data
are based on a limited number of experiments and cover limited ranges of
the variables. Further investigations are in process to better understand
the effects of high rates of vaporization upon the departure from equilibrium
distillation.
50
-------
SECTION 3
MEASUREMENT OF TEMPERATURES AND BURNING TIMES OF COAL PARTICLES
In as much as the release of the bound nitrogen in coal is kinetically
controlled (1), it is important to determine the temperature time histories
of burning coal particles and how they are influenced by coal composition.
A major uncertainty in predicting the temperature is the influence of vola-
tiles the composition of which varies greatly from coal to coal and which
may burn either heterogeneously at the particle surface or in detached
flames. The following sections provide some preliminary data on the
influence of coal composition of the temperature and burning time.
EXPERIMENTAL METHOD
A two-color pyrometer has been developed for measuring the temperatures
and burning times of individual coal particles (7). A dilute stream of
coal pafticles is produced by elutriating coal into a hypodermic needle
at a controlled rate from a vial in which the coal level is maintained
constant. The resulting coal stream is discharged as a jet and particles
withdrawn from an appropriate position in the jet are injected along
the axis of a laminar flow furnace (Fig 6). The radiation from the burning
particles is focused on to an optical fiber which is sighted along the
axis of the furnace. The optical fiber splits the radiation and feeds it
throught two narrow band pass filters with band centers at 450 and 550
nanometers. The intensity of the burning particle at the two wavelengths
is then measured using photomultiplier tubes. The particle temperatures
are obtained from the ratio of the intensities at the two wavelengths
assuming that the particle emissivity is not a function of wavelength.
The signals from the photomultiplier tubes were measured with a
Bascom-Turner model 8110 digital recorder that can sample simultaneously
-------
on two channels. Reference calibration signals were produced by a chopped
signal from a tungsten strip lamp viewed throught the furnace.
RESULTS
Experiments were conducted with five coals (Illinois No.6, Utah No.l,
Alabama Rose, North Dakota Beulah, and Montana Savage), two size fractions
(38-45 y an{i 90-105 y), furnace temperatures of 1250 and 1700 K, and oxygen
mole fractions of 0.15 to 1.0. For each condition the burning profiles for
a large number of particles were obtained. Selected results will be presen-
ted mostly for the Illinois #6 bituminous coal and the Montana Savage
lignite since the behavior of these coals bracket most of the others.
An illustration of the photomultipller output is shown in Fig 7 for
lignite particles burning in 15 and 20 percent oxygen. Each trace shows
a steep rise following ignition, followed by a slower decay. The traces
provide information on burning time, temperature, and changes in surface
area with time during combustion. The burning times are well approximated
by the duration of each signal, the temperatures are obtained by two-color
pyrometry assuming the particles are gray emitters, and the decrease in area
with combustion can be inferred from the decay in intensity with time during
combustion.
Temperatures traces for combustion of Illinois #6 coal particles in
streams containing from 15 to 100 percent oxygen is shown in Fig 8. The
irregulatities in the curves are believed to be experimental errors
involved in taking the ratio of the two signals and the background noise.
The peaking in temperature in 100 percent oxygen concentrations is seen in
many particle traces and is believed to be real. The average butning temper-
ature of Illinois #6 coal was consistently higher than that of the Montana
lignite (Fig 9). Although this is expected, on the basis of the higher heat-
ing value of the bituminous coal the story is more complicated as will be
discussed below.
Burning times for the 90-105 y Montana lignite and Illinois #6 are
shown as function of oxygen partial pressure in Fig 10. At the high
52
-------
temperatures of the study, the burning times are expected to be diffusion
limited and given by
p RT d2
t = c
8a D PXn
2
where p is the density of the coal,
c
DP is the product of diffusitivity and total pressure,
X-	is the oxygen mole fraction,
°2
d	is the particle diameter,
a	is the weight of coal that reacts with a mole of oxygen.
Calculated burning times are found to be in good agreement with measure-
ments for the lignite when it is assumed that the products of combustion are
CO and	The burning times for the bituminous coal (Illinois #6) are,
however, considerably lower than the predicted values. This could be explain-
ed either by particle swelling (the product Pc
-------
this would be expected since the soot which provides the luminous signal
is expected to be formed in significant amounts only for the bituminous
coals.
The formation of a volatile flame is expected when the rate of volatile
evolution exceeds the rate of the diffusion to the particle surface of the
oxygen needed for its combustion. As the oxygen diffusion rate is increased^
by increasing oxygen concentration the volatiles burn heterogeneously at the
particle surface as soon as they are evolved and no distinct volatile flame
is seen. Consistent with this hypothesis is the failure to detect volatile
flames when 40 u particles were burned, since the diffusion rate will
increase as the particle size is decreased. For the smaller bituminous
particles the burning times were in good agreement with the diffusion limited
estimates, when it was assumed that the oxygen diffusing to the surface was
that required to burn the char and volatiles; this provide further support
for the postulate that the volatiles burn heterogeneously when the oxygen
diffusion rate is sufficiently high. The effect of particle size on burning
time is shown in Fig 12, The data suggest that the volatile combustion
undergoes transition from detached flame to heterogeneous surface oxidation
in the size range found in pulverized coal flames. This is consistent with
the predictions of Howard and Essenhigh (10).
The above interpretation of the dkta are preliminary but provide
interesting insights on the variation that can be expected in burning temp-
eratures and burning times, as the coal type, size, and ambient conditions
are varied. These changes in particle temperature and mode of volatile
combustion will influence fuel nitrogen release rates and fche efficiency
of cbnversion.
54
-------
SECTION 4
NO/CHAR REACTIONS AT PULVERIZED COAL FLAME CONDITIONS
Nitric oxide produced early in pulverized coal flames may be subsequent-
ly reduced by char generated by the partial combustion of coal. Indirect
evidence for the importance of NO reduction reactions in coal combustors is
provided by the observation in pilot-scale combustors that the NO concentra-
tion passes through a maximum and undergoes substantial reduction in the
latter stages of combustion, particularly under fuel-rich conditions (11).
That the rate of reduction of NO observed in coal flames Is much higher
than that in the combustion products of gaseous or liquid fuels suggests
that coal products, such as ash or char, may be responsible for the reduction.
This section of the paper provides data on the NO/char reaction under
conditions of interest to pulverized coal flames for possible use in assess-
ing the role of char in reducing NO. It covers research under the EPA grant
which has also been presented at the 18th Symposium (International) on
Combustion. Extensive studies of the reduction of NO by carbonaceous solids
have been carried out in the temperature ranges of interest for auto-
mobile exhaust-reactors (12—14) and fluidized bed combustors (15—18). These
data show a strong influence of secondary reactants such as CO, Hg, and O2
on the rate of the NO/char reaction. The results of these studies are not
directly pertinent to the temperatures in combustors since the extent of
surface coverage by adsorbed species varies markedly with temperature.
Kinetics of the NO/char reaction under conditions of interest to pulve-
rized coal combustion were obtained in a laminar flow furnace for the tempe-
rature range of 1250 to 1750 K.
55
-------
Size graded, pulverized char particles are fed into the furnace in a
helium (carrier) stream, where they are rapidly heated to the furnace temper-
ature by conduction from the surrounding main-gases and by radiation from
the furnace walls. The main gases (consisting of an NO/He mixture with
varying amounts of CO and fi^) are heated by passage through an alumina
honeycomb/flow-straightener prior to entering the furnace. Furnace tempe-
rature of up to 1750 K can be achieved. The reaction is quenched by passage
of the products from the heated zone through a water-cooled section at the
base of the furnace. The char is collected on a cold, sintered bronze disk
and the composition of the effluent gases are measured by a chemi luminescent
analyzer (NO) and non-dispersive infrared (NDIR) analyzers (CO, CO2). The
composition of the gases may also be monitored in cold-flow prior to entry
to the furnace.
The experiments were performed in the laminar flow furnace as follows.
Nitric oxide of a certified concentration in helium was flowed through a
calibrated mass flow controller and mixed with an additional helium stream
to yield the desired NO (typically about 950 ppm) in cold flow. The total
flow rates were determined from the NO dilution factor (i.e., by using NO
itself as a tracer) and the known NO flow rate or, in some cases, by
measurement of the additive flow rates by passage through calibrated mass
flow meters. Upon subsequent additions of CO or H^O, the helium flow was
reduced until the initial NO concentration was restored. Char feeds were
timed, and the weight of char fed was measured (by difference) to determine
the char feed rate. The duration of each char feed was typically 60 to 210
seconds at approximately 0.1 gm/min.
During the char feed, a steady state is achieved in the furnace. Thus,
from measurement of the entrance and exit NO concentrations, knowledge of
the furnace volume gas flow rate and char feed rate, it is possible to derive
the effective rate constant for the reduction of NO on the char surface.
The char was produced by heating a pulverized Montana lignite in a crucible
to its asymptotic weight loss under an inert atmosphere at 1750 K, followed
by size-grading to 44-53 ym. The elemental analysis of the char is 84.4%
carbon by weight, 0.16% hydrogen, 0.215% nitrogen, and 0.76% sulfur.
56
-------
RESULTS AND DISCUSSION
Rate of the NO/Char Reaction and Influence of Additives.
A trace of the temporal variation of NO in the effluent gas is shown
in Fig. 13. The experiment was designed to obtain the rate of the NO/char
reaction and also to show the effect fo CO upon the net reduction of NO.
956 ppm NO in helium was fed at a rate of 3.0 1/min into the laminar flow
furnace which was maintained at 1250 K. The first drop in NO concentration
follows the introduction of the char transport gas (carried gas) at a rate
of 115 ml/min. After the NO concentration stabilized, the char was fed at
a rate of 0.16 gm/min. for a period of about 60 seconds (the interval shown
between the char feed on and off designations in Fig 13). The resultant
dip in the NO trace provides a measure of the NO/char reaction. The dis-
placement in time between the response in the NO concentration trace and
the step changes in the char feed reflect the holdup of gases in the furnace
and sampling train. The total amount of NO consumed was determined from
an integration of the measured deficit in NO concentration and the gas flow
rate.
The effect of CO on the NO/char reaction rate was determined by intro-
ducing CO (1.4%). into the NO /He stream (a corresponding amount of helium
was removed to avoid altering the NO level by dilution)» waiting for the NO
concentration to stabilize, and then feeding char into the heated NO/He/CO
mixture. This sequence of operations provided a means of separating the
NO reduction by CO at the walls from that occuring at the char surface.
Examination of the concentration trace in Fig 13 shows that the wall cata-
lyzed reaction causes the NO concentration to decrease from 922 ppm NO to
about 860 ppm. The extent of this reduction, varies with wall conditioning.
Once the NO trace had stabilized, the char was again introduced for a short
interval and the incremental NO reduction due to ractions at the char surface
measured. The remainder of the trace in Fig 13 shows a two-step increase
in NO concentration to its original concentration as a consequence of
sequentially stopping the carrier gas stream and then eliminating the wall
reactions by turning off the CO stream. For the conditions of the experiment
57
-------
in Fig 13 (T = 1250 K, NOin^et = 1^5 ppm) , the influence of CO on the
NO/char reaction appears to be that of a slight enhancement; the extent of
the reaction is small and the value is uncertain. Repeated measurements
showed a consistent slight enhancement of the NO/C reaction in the presence
of 1.4% CO. Fig 14 shows the reproducibility of the NO/char reaction as the
char feed was cycled on and off.
A previous study in this laboratory (6 ) has shown the NO/char reaction
to be first order in NO in the concentration range of these experiments. The
effective rate for the additive free NO/char reaction was determined, and is
reported in Fig 15. The rate coefficient for the NO/char reaction based on
the external surface area of the particle is given by
« 4.8 x 10^ exp	K °al Ag P^Q moles/sec
2
where Ag is the external surface area of the char in nl /gm and is in
atmospheres. The internal surface area as measured by ^ BET is 20.3 m^/gm.
Because it is known that may give a poor measure of the microporous
structure, the rate coefficient was not reduced to intrinsic rate.
The enhancement of the NO/char reaction rate by CO is slight. The
results of the increase in rate constant from the addition of 1.4% CO to the
gas stream are shown in Table I. The increase in rate is a consequence of
an increase in active sites by the reaction of CO with any surface oxides.
The rate of the NO/char reaction decreases as increasing amounts of
water vapor are added to the reactant gases as shown in Fig 16. Water was
added by flowing a metered stream of helium through a thermostated bubbler
connected to the furnace through heated gas-transfer lines. Flow rates were
corrected accordingly. The effect of H20 on the NO/char reaction is seen to
become insignificant as the temperature is raised.
Mechanistic Considerations.
The above observations of the NO/char reactions can be rationalized by
the following kinetic scheme (4). The reduction of NO by carbon is probably
through dissociation of the NO on the surface with a rapid surface diffusion
58
-------
of the dissociated atoms to form N2» The oxygen produced by the dissociation
is strongly chemisorbed and will inhibit further reaction, i.e.,
NO + 2C 	> C(N) + C(0)	(1)
C(N) + C(N) 	» (N2) + 2C	(2)
where C represents a surface carbon, C(N) and C(0) adsorbed nitrogen oxygen
atoms. The chemisorbed oxygen can either desorb to produce CO or react with
CO to form CO2, i.e.,
C(0)		> (C0)g	(3)
C(0) + CO 	=> (C02) + C	(4)
S
The inhibition of the NO/C reaction by H^O, or any other oxidant, is then
explained by the formation of the chemisorbed oxygen layer. As the tempera-
ture is increased, enhancement of the rate of activated decomposition of
the chemisorbed oxygen by reaction such as (3) is expected to decrease the
steady-state coverage of the surface by the oxide layer; this is consistent
with the decrease in the inhibitory effect of E^O at the higher temperatures.
The role of CO in enhancing the NO/char reaction is then by reaction with
the chemisorbed oxygen (Reaction 4) to increase the available active sites.
The decrease of enhancement by CO of the NO/char reaction with increasing
temperature is consistent with the expected change with temperature of the
chemisorbed oxygen layer.
Product Distribution in the High-Temperature NO/Char Reaction.
To supplement the understanding of the high-temperature NO/char reaction
mechanism a careful attempt was made to identify the distribution of reaction
products (CO and CO^) in the high temperature furnace, and to check for
closure of the oxygen material balance. A minor experimental difficulty
resulted from a residual emission of a small amount of CO from the char,
necessitating measurement of a CO "baseline" by feeding the char through the
furnace under an inert atmosphere but the same total gas flow rate as that
used in the experiments with NO, for each set of experimental conditions.
59
-------
As was to be discovered later, however, this CO could react with NO to form
CO2« Therefore, in order to minimize the relative contribution of this
effect, as well as in order to minimize the instrumental error associated
with measurement of very small CO and CO2 concentrations, the concentration
of the NO introduced into the furnace was increased ten-fold from about
1,000 to about 10,000 ppm. Because of the first-order nature of the NO/char
reaction, this did not influence the NO conversion. A representative data
trace, showing NO depletion and CO and CO^ formation as a function of time
is shown in Fig 6. NO depletion begins when the char feed is turned on,
and ends when the feed is turned off. Quantities of NO, CO and CO^ destroyed
or formed are determined either from the integrated areas under the peaks,
or from direct reference to the NDIR instrument calibration curves for the
"flat", steady-state signals attained at the peak maxima. Great care was
taken to repeatedly calibrate the analytical instruments. For this purpose,
multiple manufacturer "calibration" gas tanks were used. Instrument calibra-
tions are believed accurate to within a few percent.
The results of experiments run in triplicate with the freshly devolatil-
ized char and high entrance NO concentrations (y 9500 ppm, typically) are
shown in Table II. Repeated efforts were made to close the oxygen balance,
but results converged to around 75%. Repetition of the 1750 K experiments
several weeks later yielded results within 0.1% of those shown in the table.
Insofar as the measurements are believed to be accurate to within 10%, given
the care in calibration of flow rates and instrument responses, the source
of the missing oxygen is currently unknown. Thus, it is possible, for
example, that the missing oxygen is re-adsorbed on the char surface in the
cooling section of the furnace. A rough calculation for one data point
2
(D-178A), using a measured B.E.T. surface area of 20m /gm and an area of
°2
10 A per surface site reveals that the quantity of missing oxygen is just
marginally more than that required to form monolayer on the char surface.
The CO2 surface area is expected to be an order of magnitude larger so that
the results are consistent with the expected fractional occupancy of total
sites.
60
-------
Further reference to Table II reveals a systematic variation in the CO
to CO2 ratio detected, the ratio increasing with temperature. In fact, we
speculate that the data support the idea that the primary product of the
NO/char reaction at these temperatures is CO, the CO2 being formed in a
secondary reaction of the CO with the excess NO. As has been observed in
past experiments with added CO, a reaction, apparently catalyzed by the walls
of the reactor, does occur between CO and NO, forming CO 2 as a product.
To illustrate the feasability of this premise, experiments were performed
at each furnace temperature, in which CO and NO, at the levels detected in
the effluent stream, were fed through the hot furnace in the absence of a
char feed, and the CO^ formed was measured. An example, corresponding to
the data of Fig 17 left, is shown in Fig 17 right. From the known flow rates
and char feed times, a lower limit to the CO 2 formed in the char experiments
could, thereby, be determined. The results are shown in Table III. Note
that these represent only approximate estimates because only exit concentrations
of CO and NO were reacted. More accurate assissment of the magnitude of the
CO/NO reaction extent would require knowledge of the CO and NO axial profiles,
knowledge of the (apparatus dependent) reaction rate, and integration over
the furnace hot zone. Since the reaction is surface catalyzed, it may will
proceed further during the char feed when a dispersed catalyst is present.
In any case, carbon monoxide formation accounts for of the order of 90 percent
of the total carbon oxides formed (Table II with corrections from Table III);
the small amount of CO2 observed may be due to secondary surface catalyzed
NO/CO reactions which are difficult to eliminate in the experiment.
CONCLUDING COMMENTS
The rate of reduction of NO by char has been determined over the temperature
ranges of interest in pulverized coal flames. The NO/char reaction at com-
bustion temperatures is found to be enhanced slightly by the presence of CO
and inhibited slightly by water vapor, with both effects decreasing with
increasing temperature. The observations are consistent with a hypothesis
that the inhibition of the reaction is due to the tying-up of active sites
by chemisorbed oxygen and that the enhancement by CO is due to reduction of
61
-------
the chemisorbed oxygen. The data also suggest that the primary product of
the NO/char reaction at elevated temperatures is CO, CO^ being formed by
surface-catalyzed secondary reaction with NO.
62
-------
REFERENCES
1.	Pohl, J.H., and A.F. Sarofim. Devolatilization and Oxidation of Coal
Nitrogen. 16th Symposium (International) on Combustion, The Combustion
Institute, 1976. p.491.
2.	Cheng, I. Coal Nitrogen Conversion to NO During Simultaneous Oxidation
X
and Pyrolysis. S.M. Thesis, M.I.T., Cambridge, Mass., 1978.
3.	Pereira, F.J., J.M. Beer, J.M. Gibbs, and A.B. Headley. NO^ Emission
from Fluidized-Bed Coal Combustors. 15th Symposium (International) on
Combustion, The Combustion Institute, 19740 p.1149.
4.	Chan, L.K. Nitric Oxide Reduction by Char in a Fixed Bed Reactor. Sc.D.
Thesis, M.I.T., Cambridge, Mass., 1980.
5.	Sprouse, A.M. Reduction of Nitric Oxide by Coal Char in a Fluidized Bed.
S.M. Thesis, M.I.T., Cambridge, Mass., 1974,
6.	Song, I. Fate of Fuel Nitrogen During Pulverized Coal Combustion. Sc.D.
Thesis, M.I.T., Cambridge, Mass., 1974.
7.	Altrichter, D.M. Optical Determination of Time Temperature Profiles
for Single Particle Coal Combustion, S.M. Thesis, M.I.T., Cambridge,
Mass., 1980.
8.	Jungten, H., and K.H. Van Heek. Fuel Processing Technology, 2:261-293,
1979.
9.	Pereira, F.J., and J.M. Beer. NO Formation from Coal Combustion in a
Small Experimental Fluidized Bed. Second European Symposium on Combus-
tion, French Section, Orleans, France, 1975.
10. Howard, J.B., and R.H. Essenhigh. Mechanism of Solid Particle Combustion
with Simultaneous Gas—Phase Volatiles Combustion. 11th Symposium (Inter-
national) on Combustion, The Combustion Institute, 1967. p.399.
63
-------
11.	Wendt, J.O.L., D.W. Pershing, J.W. Lee, and J.W. Glass. Pulverized Coal
Combustion: N0_ Formation Mechanisms under Fuel Rich and Stage Combus-
tion Conditions. 17th Symposium (International) on Combustion, The
Combustion Institute, 1979. p.77.
12.	Shelef, M., and K. Otto. Simultaneous Catalytic Reaction of 0^ and NO
with CO and Solid Carbon. Journal of Colloid and Interface Science,
31;73, 1969.
13.	Edwards, H.W. Interaction of Nitric Oxide with Graphite. A.I.Ch.E.
Symposium Series No.126, 68:91, 1972.
14.	Shepard, W.M. A Kinetic Study of the Reaction of Nitric Oxide and Activ-
ated Carbon. Ph.D. Thesis, Clemson University, S. Carolina, 1974.
15.	Pereira, F.J. N0x Emissions from Fluidized Coal Combustion. Ph.D.
Thesis. University of Sheffield, Sheffield, England.
1975.
16.	Furusawa, T., and K. Kunii. Kinetic Study of Nitric Oxide Reduction by
Carbonaceous Materials. Society of Chemical Engineering, Japan, 1977.
17.	Beer, J.M., A.F. Sarofim, L.K. Chan, and A.M. Sprouse. NO Reduction by
Char in Fluidized Combustion. Fifth International Conference of Fluidi-
zed Bed Combustion, Washington, D.C., 1977.
18.	DeSoete, G.G.. Mechanism of Nitric Oxide Reduction of Solid Particles.
Fifth EPA Fundamental Combustion Research Workshop, Newport Beach, Cal.,
1980.
64
-------
HON
Product
Process
Char - N
Heterogeneous
Oxidation
Fig 1. Schematic of Process Occur
ing During the Formation of Nitric
Oxide from Coal Nitrogen. An
Oxidative PJtrolysis Route is not
Shown
65
-------
.00
00
BO
70
60
50
40
30
20
10
0
•	975 K
~	1050 K
¦	1160 K
*	1270 K
~	1300 K
0 10 20 30 m 50 60 70 EO 90 100
HEIGHT LOSS, I
iF«& Nitrogen Evolution from
Shale 01?	°f Raw Paraho
66
-------
1000 K
1200 K
1400 K
i
i	i
80 90 100
WEIGHT LOSS, %
Fig 3. Nitrogen Evolutfon from
150 yim Droplet Array of Indo-
Malay #6 Oil
67
-------
100
90
30
70
- 60
^ 50
UJ
* 40
^ 30
ID
O
as 20
10
0
/•

J	I	L
1
1
10 20 30 40 50 60 70 80
WEIGHT LOSS, %
I
90 100
Fig 4. Nitrogen Evolution from
the Equilibrium Distillation of
Raw Paraho Shale 011 at Atmospheric
Pressure
68
-------

100

90
•>«

•\
80
CO

CO
o
70
-J


60
t—
sc

CD
50
•—«
111

LLI
40
Z

UJ
30
tr>

o

ac
20
1—

Hi
10

0
10 20 30 40 50 60 70 80 90 100
WEIGHT LOSS, %
Fig 5. Nitrogen Evolution from
the Equilibrium Distillation of
Indo-Malay #6 Oil at Atmospheric
Pressure
69
-------
FIBER
OPTICS
o
n>
n> to
CL
o>
t/> •
«<
in
c+ —I
is
CO
(+
0>
(O
CD
go
-J.
3

PURGE
GAS
FEEDER
PROBE'
DILUTION
STAGE
LENS
\^//////z
20-GAUGE
FEED TUBE
FINE
JET
COOLING
WATER

VIBRATOR
PARTICLE
FLOW
CARRIER
GAS
, VACUUM
i-«" SEAL
20-GAUGE
TUBE
WV*
u
w
VIAL OF
COAL
• MOTOR-DRIVEN
i PLATFORM
-------
50 ms
0.4
0.3
0.2
0.1
0.0
y 0.4
~j
2 0.3
0.1
0.0
TIME
Fig 7. PMT Outputs for Montana
Lignite in 15# and 20% Oxygen
-------
3500
3000
U!
cc
z>
£
cc
LlI
Q.
{±! 2500
V_20% 0.
2000
20
30
25
TIME (msec)
Fig 8. Typical Temperature Traces
for Illinois #6 at Five Oxidizing
Conditions
72
-------
UJ
cc
H
<
cc
lu
CL
2
UJ
f-
UJ
—1
O
CC
<
CL
UJ
o
<
cc
UJ
>
<
3200
3000
2800
2600
2400
2200
2000 -
A 1LL1N01 $& 6
O MONTANA LIGNITE
1800
1
1
0.0 Q2 0,4 0.6
X,
0.8
1.0
Fig 9. Average Particle Temp-
erature Traces for Montana
Lignite and Illinois #6 at 1700 K
73
-------
120
110
100
o MONTANA LIGNITE 90-105/im
° ILLINOIS #6	90-l05/xm_
90
% 80
6
uj 70
S
3
O
Z
0c
3
0
60
50
40
30
20
10
0
\
I
A

\
\
\
v.
1

r
1
f
—--1
		1
_L
.20	.40	.60	.80
OXYGEN PARTIAL PRESSURE
I.00
Fig 10. Average Burning Times
for Montana Lignite and Illinois
#6 at 1700 K
74
-------
10
H
.3
0.
I-
3
O
(K
UJ
0-
=>
2
O
h-
o
x
0.
35% 0;
UJ
>
20% 02
I-
<
_i
LJ
o:
50ms r—
Ti ME

F1g 11. PMT Outputs for Illinois
#6 at Six Oxidizing Conditions
75
-------
50
40
90-105
30
H 20
38-45 p
0.8
0.4
0.6
1.0
0.2
0D
OXYGEN PARTIAL PRESSURE,
ATM
Fig 12. The Effect of Oxygen
Partial Pressure and Particle size
on the Burning Times of a North
Dakota Beulafa Coal* Furnace Temp-
erature 1250 K
76
-------
9 56
_ppn
Montana Lignite Char
(I 750 K , Cr uc ibl
-------
a oo
700
Feed Feed
On
CO
On
Off
600
E 500
Montono Lignite Chor
( 1750 K. Crucible, 44-53 Jim)
T Furnoce ¦1500 K
Feed Rate « 0.15 gm /roin.
O 400
300
200
150
Seconds
100
NO vs TIME
Fig 14. Poisoning of the Surface
Catalysis of the NO/CO Reaction at
1500 K In the Laminar Flow Furnace
78
-------
= 34.70kcal/
mole
act
- 15
- 18
- 19
4
Fig 15. Arrhenius Plot of the NO/
Char Reaction Rate Constant
79
-------
E.ff
-------
Feed off
(9060)
= 177 sec
NO
Feed
(2000)
Feed
on
UJ
<
o
to
(5400)
CO
_i
Z)
u.
(475)
0
O
CO.
T =	1750 K r
NO = 10,000ppm ts.1
CO =	3,000ppm ts.1
CO? = 2,000ppm f.s.
TIME
NO = 10,000ppm fa,
CO = 3,000ppmfs
COp = 2000ppmfS.
CO
NO
CO
Fig 17. (a) CO and CO, Product
Distribution from the Nu/Char
Reaction at 1750 K. Montana Lig-
nite Char (Produced in a Crucible
at 1750 K). 44-52 y;m Feed
(b) CO2 Formation by the Surface
Catalyzed Reaction of NO with CO at
1750 K.
Rate » 0.10 gm/minute. Total
Flow Rate =3.13 1/min.
81
-------
TABLE I. ENHANCEMENT OF NO/CHAR REACTION RATE BY ADDED CO (1.4%).
T Furnace
kCO^
1300 K
1600 K
1.28
1.27
1.28
1.09
1.05
1.28
1.14
TABLE II. OXYGEN MATERIAL BALANCES IN THE NO/CHAR REACTION.
Run
Furnace
Temp.
Oxygen A
Balance
CO/CO
2**
D-211 A
D-208 A
D-205 A
D-178 A
D-175 A
D-172 A
D-193 A
D-190 A
D-187 A
1500 K
1500 K
1500 K
1625 K
1625 K
1625 K
1750 K
1750 K
1750 K
72.5%
64.1%
64.3%
77.2%
76.1%
73.1%
78.2% )
77.3%
76.5%
67.0%
75.5%
77.3%
0.78
0.78
0.78
2.53
2.62
2.64
4.27
3.80
3,90
Notes:	.
CO + 2CCOJ - (CO)
* Oxygen balance is computed as •\.•
inlet " observed
** CO to COj ratio is corrected by substractlcm of baseline CO
from measured CO.
82
-------
TABLE III. C02 FORMED BY REACTION OF NO WITH CO.
Run
Furnace
Temp.
% CO2 Accounted
for by CO/NO
Reaction
D-211 A
D-208 A
D-205 A
D-178 A
D-175 A
D-172 A
D-193 A
D-190 A
D-187 A
1500 K
1500 K
1500 K
1625 K
1625 K
1625 K
1750 K
1750 K
1750 K
73.7%
79.2%
78.9 %
55.6%
61.3%
67.4%
74.8%
73.8%
71.4%
77.3%
60.4%
73.3%
83
-------
GAS PHASE PROCESSES INVOLVED IN THE CONTROL OF
NITROGEN OXIDE FORMATION IN FOSSIL FUEL FLAMES
By:
J. M. Levy
M.I.T. Energy Laboratory
Cambridge, Massachusetts 02139
84
-------
ABSTRACT
Optimization of a control strategy for NO emissions from fossil fuel com-
bustion requires an understanding of the mechanistic chemistry of fuel-nitrogen
conversion. Computational capabilities are demonstrated to be quite accurate
in the presence of simple fuels, but break down somewhat in the presence of
hydrocarbons. A quasi-global method for computing fuel-nitrogen conversion in
a higher hydrocarbon environment is described, and the current status of mod-
eling bound nitrogen profiles in a C^/C2 environment is presented.
85
-------
A. Introduction
With the depletion of natural gas and petroleum reserves (which are lack-
ing or low in nitrogen content - and, hence, relatively low in the emission of
NO formed from the nitrogen bound in the fuel), and the increasing substitu-
tion of coal, coal-derived liquids, residual fuel-oils, and shale oils (which
are high in nitrogen content), NO emissions, which are already unacceptably
high in several urban areas, are bound to increase unless an abatement strat-
egy is developed and implemented. Insofar as the mechanism by which fuel-
bound nitrogen is converted to N0x is complex, and insofar as early attempts
(1,2) to reduce N0x emissions by empirical adjustment of combustion parameters
(e.g., temperature, fuel/air ratio, degree of mixedness, etc.) utilizing con-
ventional combustors have achieved only a fraction of their full potential, a
more fundamental elucidation of the NO formation mechanism has become essen-
tial in order to achieve the desired engineering goal of modifying the com-
bustion process in such a way as to favor the conversion of fuel-bound nitro-
gen to the desired (combustion) end product, rather than to the undesired
product, NO.
It is well understood that for any condensed phase fuel, combustion may
proceed through parallel homogeneous and heterogeneous processes wherein the
fuel is initially heated and undergoes devolatilization followed by rapid
homogeneous gas phase oxidation of the volatile species and, if a devolat.il-
ized residue remains, slower heterogeneous burnout of the char. This is
apparently the case for the combustion of pulverized coal and residual fuel
oils. Distillate oils, however, may well burn entirely in the gas phase as
a diffusion flame surrounding a shrinking, evaporating droplet. Certain
fuels, of course, are gaseous from the outset.
Of primary interest here is the chemistry of volatile fue1-nitrogen con-
version. Traditional gaseous fuels (natural gas) contain little or no bound
nitrogen - although it is projected that the products of coal gasification
which undergo a high temperature desulfurization may well contain significant
quantities of ammonia. In addition, light distillate oils which for coal and
shale-derived fuels may have nitrogen contents in excess of 1% by weight burn
entirely in the volatile phase.
86
-------
For fuels which leave a devolatilized residue, the relative significance
of gas phase (versus heterogeneous.) processes on NO formation is, less obvious.
The distribution of nitrogeneous species between volatiles and char has recent-
ly been studied (3,4,5,6,7) for pulverized coal combustion for temperatures
ranging from 1250 to 1750° K. The results (for a Montana Lignite) demonstrate
that at higher temperatures under fuel lean conditions, as much as 80.% of the
NO formed may originate in the volatile phase. This reflects the fact that at
high enough temperatures, nearly all of the nitrogen in coal can be driven into
the gas phase (3). Indeed, even under conditions (1750° K) where 70% or more of
the fuel nitrogen is volatilized during the course of the experiments (5), for
equivalence ratios above about 1.5, over 50% of the N0x originates from the
char! These results reflect the long established facts that increases in fuel
nitrogen concentration (8) and equivalence ratio (8,9,10) in the gas phase all
tend to lower the conversion efficiency to NO^. The strategy for controlling
N0x emission, therefore, is to drive as much fuel nitrogen as possible into the
gas phase in a primary fuel rich zone (7,8,9,10,11) where the. reaction kinetics,
can be "engineered" to minimize conversion to NO , as opposed to the alternative
of allowing the nitrogen to remain in the char where subsequently it will be
partially converted to NO^ in the fuel lean secondary combustion zone (12).
Clearly, optimizing this reaction engineering (minimizing NO emissions) is best
achieved through a true understanding of the gas phase reaction mechanism. In-
sofar as the residence times in typical combustors are too short to achieve
chemical equilibrium in the burnt gas^s, it is necessary to consider the time
dependent flow of nitrogeneous species through the many sequential and compet-
ing reaction pathways leading to N0x and Ng formation, in order to develop a
methodology for maximizing or minimizing certain product yields - i.e., both
the elementary reaction pathways and reaction rates must be considered.
Unfortunately, trace emissions from rich stationary source combustors are
the product of several, lumped chemical/physical processes, and even if the
problems of coupling finite-rate chemistry with complex flow-fields are ne~
glected for the moment, the necessity of computing NO^ levels to within the ppm
range places rather stringent requirements on the mechanistic accuracy of what-
ever model is invoked.
Despite residual uncertainties in the currently available mechanistic and
kinetic data base, adequate if imperfect, modeling of fuel-nitrogen conver-
-------
sion An the presence of "simple" fuels can now be achieved (11,13,14). Unfor-
tunately, however, real combustion systems, i.e. real fossil fuels, present
incremental mechanistic complexities which do not decouple easily from solu-
tion of the fuel-nitrogen problem. It is to an assessment of these complexi-
ties and to the development of a methodology for dealing with them that this
paper is addressed.
B. The Fuel-Nitrogen Mechanism
A large literature has appeared recently directed towards the experimental
elucidation of the mechanistic details of fuel-nitrogen conversion under flame
conditions (8,15). Synthesis of these observations into a master reaction set
has been the subject of other studies reported elsewhere (16,17,18). Valida-
tion and refinement of the reaction set, based upon efforts to model indepen-
dent data under as broad a range of conditions as possible, is an ongoing
effort (11,13).
Initial attempts (18) by this author to construct a comprehensive fuel
nitrogen mechanism utilized the method of Engleman (19) in which all species
expected to be present are permitted to inter-react via elementary reactions.
Rate constants were selected from tabulated estimates (made, generally, by the
method of Johnston (20)). The method, however, suffers from the limitation of
often necessitating the use of an impractically large reaction set (especially
when combined with hydrocarbon mechanisms. See below.), and, additionally,
is limited in its accuracy by the capability to select, a priori, all species,
including intermediates, which are present. Recent suggestions by Branch
et al. (21) of mechanistic refinements to the NO/NH^ reaction set underline
this limitation. Current efforts have, therefore, tended back towards the use
of a much smaller, conceptually clearer, fuel nitrogen reaction set.
To this end, equal success has been achieved in modeling the post-flame
zones of several nitrogen-doped hydrocarbon flames using the "minimum subset"
of fuel nitrogen reactions shown in Table I, as with the larger master set.
It is cautioned, however, that the full range of applicability of this minimum
subset is not yet established. Thus, for example, though significant contri-
bution to the destruction of NO by reaction with N»2 is precluded in modeling
post-flame data in the range of 2000 K, recent indications (22) of the very
88
-------
strong inverse temperature coefficient q£ the effective NO -f- rate constant
suggest that that mechanistic addition may still prove appropriate at lower
temperatures.
An example of the modeling success achievable is indicated in Figure 1,
showing the data of Haynes (23) in the post-flame gases of a rich (
 = 1.66) doped with 500 ppm NO. The major
distinction in this flame is the "breakthrough" of hydrocarbons (roughly 0.2%
of the total flow) into the post-flame zone. The results of Figure 2 were gen-
erated by ignoring the hydrocarbons. Although the fits to the concentration
profiles are not really bad, better success in fitting the HCN and NO profiles
has been achieved (see below) through consideration of the pertinent hydrocar-
bon chemistry. (Ammonia profiles remain systematically in error). It should
be clearly noted, also, that the influence of hydrocarbons upon fuel-nitrogen
chemistry is potentially much greater in the highly backmixed environment of a
practical combustor than may be suggested by Figure 2. This is demonstrated by
a series of computations performed (using a detailed hydrocarbon mechanism) on
a methane/air system doped with 1% ammonia. In these calculations, a well-
stirred reactor (WSR) section of varying residence time fed a plug-flow reactor
(PFR) section, also of varying residence time, such that the total residence
time remained approximately constant. Bound nitrogen concentrations are shown
in Fig. 3 for an adiabatic calculation at a fuel/air equivalence ratio of 1.33,
and demonstrate nicely that for a fixed total reaction time, HCN exit concentra-
tions (the major product of fuel-nitrogen/hydrocarbon interactions) are in-
creased with increased stirred reactor residence time. This conclusion follows
logically, of course, from the recognition that strong backmixing necessarily
89
-------
increases the possibility for reaction between hydrocarbon and nitrogeneous
fragments.
C. The Hydrocarbon Problem
The results of Figure 2 serve in miniature to demonstrate that additional
mechanistic complexities arise in the computation of fuel-nitrogen chemistry
when hydrocarbons are present. Indeed, predictive modeling of bound-nitrogen
emissions in a real-fuel, fuel-rich environment can be seen to be dependent
upon the capability to model hydrocarbon decay kinetics for the following
reasons:
(1)	Interactions of fuel-nitrogen and hydrocarbon fragments may be
quite significant, especially in recycling NO back to HCN. Omission
of this class of reactions is consistent with the enhanced NO profile
and diminished HCN profile calculated in Figure 2. Also,
(2)	the radical pool (H,0,0H) which drives the fuel-nitrogen chem-
istry is determined primarily by the hydrocarbon chemistry which dom-
inates the system at early times.
Insofar as fuel-rich operation contributes to NO^ abatement strategies,
it seems necessary to model hydrocarbon burnout if ppm-level accuracy is to be
demanded of a fuel-nitrogen mechanism. Unfortunately, however, aside from hav-
ing only the most phenomenological knowledge of the specifics of the interac-
tive reactions between hydrocarbons and nitrogenous species, even modeling the
combustion of methane - the simplest of hydrocarbons - currently comes close to
straining the state-of-the-art. In addition, the validity of generalizing con-
clusions applicable to design considerations of real fuel systems based upon
computations using methane as a model hydrocarbon fuel is still not, a priori,
clear. Indeed, methane is an atypical fuel in that
(1)	its H/C ratio of 4 greatly exceeds the H/C ratios of fuels of
interest (H/C might equal 1.3, for example, for a coal-derived
liquid),
(2)	its combustion details might simulate paraffin-type species, but
many fuels of practical nature are highly aromatic in character, and
90
-------
(3) even as a paraffin, methane may be atypical in that methyl re-
combinations may be of unusual significance in its combustion
mechanism.
Two questions of significance have, therefore, been identified.
(1)	To what extent can the results of NO modeling in the presence
X
of methane be taken as representative of fuels of practical inter-
est? And, in order to answer that question,
(2)	how can the combustion kinetics of higher hydrocarbons, more rep-
resentative of fuels of practical interest, be modeled for the pur-
pose of predicting fuel-nitrogen conversion to NO ?
X
D. Methodology for Dealing with Higher Hydrocarbons
1. Detailed Kinetics
In this method, all species of significance and all essential elementary
reactions of each specie are identified, resulting in an extensive set of
coupled rate equations which describe the kinetic behavior of the system. This,
of course, is the most correct way to characterise the system. The method suf-
fers, however, from the limitations that:
(a)	For higher hydrocarbons, even straight chain paraffins, neither a
complete set of species nor, certainly, of reactions is known - i.e.,
with few exceptions, the detailed combustion mechanisms are simply
not available.
(b)	Theoretical synthesis of combustion mechanisms for higher hydrocar-
bons (as performed for CH^, for example, by the method of Engleman
(19,24) is impractical because the magnitude of the problem grows
exponentially (25) with the number of carbon atoms, resulting in
reaction sets impractically large for computational studies, and
because, in any event, virtually none of the elementary reaction
rate constants are known. Additionally,
(c)	Even if such an approach were viable, the time and cost necessary to
screen such reaction sets down to reasonable sizes, as well as the
current dearth of experimental data necessary for model validation
91
-------
would render the method impractical.
Detailed kinetics are therefore rejected as a viable method of -modeling
the combustion chemistry of fuels of practical interest.
2. The Global and Semi-Global Methods
In this method, hydrocarbon burnout is modeled by a single "global" (or
overall) relation of the form
CnHm + (n + |) 02 nCC>2 + y H20
for vhich a rate expression must be empirically determined. Even as a first
estimate, this method really has little basis in theory, and has been shown to
be applicable only where a single major heat release step is involved, thereby
excluding its applicability to hydrocarbon oxidation. In any event, the method
completely overlooks elementary reaction kinetics and intermediate radical for-*
roation and is of no utility in computing NO levels.
Hydrocarbon oxidation can be conceived of occuring, however, in two over-
lapping heat release steps: a rapid oxidation of the hydrocarbon to carbon
monoxide and hydrogen (or water1, followed by subsequent oxidation of the CO
(and H2)• It is possible, therefore, to develop a semi-global method utilizing
two (or more) sequential global steps and one(or more) stable intermediate
species to describe the overall combustion process. Thus, for example
CnHm + (f + f) °2 nC0 + (m/2)H20
followed by
CO + 1/2 02 + C02
This method has been used with some success (26) for the relatively simple
case of methane oxidation, but again, in any case, suffers from the limitation
of completely neglecting intermediate, radical species essential to prediction
of trace NO formation levels.
A conceptually similar, but somewhat more detailed semi-global method has
been under development by Dryer (27) for application to hydrocarbon burnout.
In this method, paraffin oxidation has been separable into three global steps
o< the form
92
-------
Allcane -> Alkene -> CO ->• CO2
Although this may be regarded as a significant improvement over two-step
semi-global modeling for purposes of modeling heat release, it offers little
practical improvement for purposes of modeling 150 levels. Additionally, em-
pirical rate expressions characterizing each global step are currently avail-
able only for paraffins up to butane. No rates are yet available for aromatic
systems.
3. The. Quasi-Global Method
In this method, the simple global step
C H + f 0 - nCO + f H
n m / I	L 2
is followed by an elementary reaction mechanism (which is well known) for the
l^/COy^ system. Because the time scale for CO burnout is long compared with
the global conversion of the parent hydrocarbon to CO, a certain amount of mod-
eling success may be achieved by assuming the global step to occur at an infin*-
ite rate - especially if attention is focussed primarily upon post-flame reac-
tions - although, following the work of Edelman and Fortune (28), a finite
rate expression for the global step may also be used to enhance modeling accu-
racy. Utilization of detailed kinetics for the CO/H^ oxidation admits free
radical chemistry to the mechanism, thus opening the door to computations cou-
pling the detailed kinetics of fuel-nitrogen conversion with hydrocarbon
burnout.
Several serious limitations to this method, however, must be noted:
(a)	Although a degree of "elementary" reaction chemistry is preserved
in modeling the CO/H2 oxidation step via detailed kinetics, the actual
temporal history of the free radical pool may be seriously miscalcu-
lated. Thus, attempts to model thermal-NO formation (29) in a
methane flame by means of the quasi-global method have proven inad-
equate largely due to errors in the computed oxygen atom levels.
(b)	The nature of the global step is such as to remove all consideration
of hydrocarbons from the model. However, as suggested by Figure 3
above, "recycling" of NO to form HCN in fuel-rich systems via reac-
93
-------
tions with hydrocarbons (e.g., CH + NO -*¦ 11CN ->¦ 0) is a limiting
factor in bound nitrogen decay in the presence of hydrocarbons, and
must be considered in developing staging strategies to eliminate
bound nitrogen from the corabustor exhaust stream. Omission of all
hydrocarbons from consideration is tantamount to neglect of many of
the processes which require modeling.
(c) Distinction among different hydrocarbon fuels is not readily apparent
in any other manner other than purely stoichiometric in the quasi-
global model, although differences in molar enthalpies of fuels
should lead to somewhat different computed time/temperature his-
tories. Of course, the degree to which mechanistic chemistry differ-
ences (rather than temperature differences) distinguish NO^ emission
levels from the combustion of different hydrocarbons is currently un-
clear (no definitive experimental data are currently available), and
the hope is that mechanistic differences (for example, the distribu-
tion of intermediate hydrocarbon fragments) will prove to have only
a second order influence upon N0^ levels. However, constructioi\ of
a computational model which a priori neglects such considerations
does seem less desirable than one which permits them.
It is concluded, therefore, that the quasi-global method, as currently
utilized, is, if not inapplicable to the modeling of NO emissions from higher
hydrocarbons, at least undesirable as a method with adequate predictive capa-
bilities.
A. Development of a New Quasi-Global Method - QUAGMIRE
Having exhausted, as impractical or inapplicable, all existing methods of
treating hydrocarbons of whirh we are aware, we have proposed a new quasi-
global method of the following, general form:
Parent Hydrocarbon
Global
^ Distribution of Hydrocarbon Fragments
Detailed Kinetics
I		> co2 + h20
94
-------
The basis underlying this hypothesized method rests upon the assumptions
that the initial degradation of the parent hydrocarbon in the preflame zone
via pyrolysis and oxidative pyrolysis routes is very rapid; that the initial
decomposition products may be represented by a relatively small set of low
molecular weight, stable and/or unstable hydrocarbon species; and, of course,
that the decomposition mechanism can be justifiably represented by a global
step.
The potential utility of the method lies in the following:
(a)	The advantages of modeling free radical chemistry by detailed kinet-
ics are retained, but the magnitude of the elementary reaction set
required is vastly reduced. Again, this reflects the exponential-
like growth in reaction set magnitude with number of carbon atoms in
the parent specie. If only hydrocarbon fragments with relatively few
carbon atoms are present following the global step, then the magni-
tude of the reaction set - given by the union of the combustion reac-
tion sets of the fragment species - will be small relative to the
(hypothetical) reaction set of the parent hydrocarbon.
(b)	Whereas, even if the magnitude of the problem could be handled compu-
tationally, the detailed combustion mechanism of the parent hydrocar-
bon is generally unknown, detailed combustion mechanisms of many of
the hydrocarbon fragments, which may be assumed to predominate at
flame conditions, are known or are in varying stages of development.
(c)	Unlike the quasi-global method of Edelman and Fortune, hydrocarbons
are retained in the elementary reaction scheme.
(d)	To the extent that an initial global step is applicable according to
the method of Edelman and Fortune, the approximation should be more
accurate here. This reflects a reduction in the mechanistic size of
the global step, or, more significantly, a reduction in the magnitude
of the reaction time or extent encompassed by the global step. That
is, whereas the method of Edelman and Fortune takes a giant global
step nearly to the postflame zone (where hydrocarbons have disap-
peared and CO has formed), it is arguable that the global step pro-
posed here is completed in the preflame zone or early into the pri-
mary reaction zone (30) - which is where much of the essential fuel-
95
-------
nitrogen chemistry occurs J Thus, the global step may be regarded
as faster on a time scale pertinent to fuel-nitrogen conversion.
The probability of miscalculating the concentration/time history of
the radical pool, for purposes of modeling NO^ emissions, is concur-
rently reduced.
In distinction to the quasi-global method of Edelman and Fortune, the
technique proposed here may be referred to as the Quasi-Global Method applied
to Initial Reaction Extents (QUAGMIRE). Details of QUAGMIRE, including justif-
ication of its computational applicability and discussion of associated re-
search questions are presented elsewhere (31)~
E. Modeling of Fuel-Nitrogen Conversion in a Hydrocarbon Environment
It can be shown (31) that the detailed kinetic mechanisms of combustion of
the and 0,^ species under fuel-rich conditions are the basic building blocks
of QUAGMIRE. Recent developments in this area have been rapid and substantial,
and have laid an adequate foundation upon which it is possible to develop QUAG-
MIRE. Olson & Gardiner (32) have published a methane mechanism which includes
chemistry through the species, although the formaldehyde reaction rates
should, probably, be updated by Dean's (33) more recent shock tube measure-
ments. Through extensive modeling of shock tube data, Gardiner (54) has con-
cluded that his is the best existing methane mechanism, although Gardiner (35)
has also identified inadequacies in it in modeling chemistry. Levy (31),
however, has rectified at least one of the residual problems in Gardiner's
mechanism by demonstrating that the products of hydroxy1 attack upon acetylene
do not feed strongly into the CH^ set, but are better modeled as running
through a ketene intermediate to formaldehyde. Accordingly, in an attempt to
model Haynes' rich, fuel-nitrogen doped ethylene/air post-flame hydrocarbon
decay data (Figure 2), Gardiner's modified reaction set was combined with the
minimum subset fuel-nitrogen mechanism of Table I, a small subset of CH-gener-
ating reactions borrowed from the Engleman (24) methane reaction set compila-
tion shown in Table II (Gardiner's set does not include CH), and a small set
of postulated hydrocarbon/fuel-nitrogen interactions, shown in Table III, of
which all but one are gleaned from Engleman's compilation (24), the remaining
dominant (global) reaction of CH^ with NO being taken from Haynes' (17,23)
96
-------
analysis. The resulting reaction mechanism of 102 reactions in 36 species is
integrated to yield the results shown in Figure 4, which demonstrate nicely the
desired hydrocarbon decay data, as well as coupled fuel-nitrogen profiles im-
proved over those of Figure 2,
The prognosis for establishment of an adequate C^ modeling capability in
the near future does, therefore, look quite bright, although the reader should
not be misled into believing that either or C£ mechanisms can yet be con-
sidered to be adequately characterized. Thus, free radical profiles computed
for the hydrocarbon "breakthrough" case fall far below the measured values.
This demonstrates a residual problem with Gardiner's C^/C^ mechanism, and in~
dicates the need for further developments of the elementary hydrocarbon mechan-
isms. It should be noted that excessive radical destruction is also predicted
for this system (36) by the newly published mechanism of Westbrook (37). In
this regard, it is vital that additional data, similar to those of Haynes (23),
be obtained for validation or refinement, under post-flame conditions pertinent
to practical fuel-nitrogen chemistry, of the mechanistic hydrocarbon chemistry
developments which are generally derived from shock tube studies.
F, Summary
Capabilites to model fuel-nitrogen conversion in the absence of hydrocar-
bons under aerodynamically clean conditions have been demonstrated, although
the mechanistic chemistry of the amines still requires refinement. Modeling
capabilities under conditions of hydrocarbon "breakthrough", however, are lim-
ited by the necessity to model higher hydrocarbon oxidation. This problem,
however, may be reduced in computational magnitude through a quasi-global
method which limits necessary hydrocarbon mechanisms to and C2 species. Al-
though not yet fully developed, available and mechanisms are improving.
Additional experimental data are required, however, for refinement and valida-
tion of and C2 reaction sets; elucidation of hydrocarbon fragment distri-
butions in the post-flame zone as a function of parent higher hydrocarbon,
equivalence ratio and temperature; and provision of a data base for the model-
ing of the coupled chemistry of fuel-nitrogen conversion in a higher hydrocar-
bon environment.
97
-------
REFERENCES
la. Turner, D.W., and C.W.Siegiuund, "Staged Combustion and Flue Gas Recycle:
Potential for Minimizing NO from Fuel Oil Combustion", American Flames
Research Committee Flame Days, Chicago, September, 1972.
b.	Turner, D.W., R.L.Andrews, and C.W.Siegiuund, "Influence of Combustion
Modifications and Fuel Nitrogen Content on Nitrogen Oxide Emissions from
Fuel Oil Combustion", AIChE Symposium Series 68, 55 (1972).
c.	Siegmund, C.W. and D.W.Turner, "NO Emissions from Industrial Boilers:
Potential Control Methods", J. Eng. Power, p.1-6, January, 1974.
2a. Bartok, W., A.R.Crawford, and G.J.Piegari, "Systematic Field Study of
NO Emission Control Methods for Utility Boilers", Esso Research and
Engineering Co., Report No. GRU. 4GNOS.71 to the Office of Air Programs,
Research Triangle Park, North Carolina, December 1971.
b.	Martin, G.B. and E.E.Berkau, "An Investigation of the Conversion of
Various Fuel-Nitrogen Compounds to Nitrogen Oxides in Oil Combustion",
AIChE Symp. Ser. 6^, 126, 45 (1972).
c.	Heap, M.P., T.M.Lowes, R.Walmsley, and H. Bartelds, Burner Design
Principles for Minimum NO Emissions, Proceedings Coal Combustion Seminar,
Research Triangle Park, N*C., EPA Report No. EPA 650/2-73-021, Sept. 1973
3.	Pohl, J.H., "Fate of Fuel Nitrogen", SC.D. Thesis, M.I.T. (1976).
4.	Pohl, J.H., and A.F.Sarofim, "Devolatilization and Oxidation of Coal
Nitrogen", Sixteenth Symposium (International) on Combustion, The
Combustion Institute, p. 491, Pittsburgh (1977).
5.	Song, Y.H., "Fate of Fuel Nitrogen During Pulverized Coal Combustion",
Sc.D. Thesis, M.I.T. (1978).
6.	Song, Y.H., J.M. Beer and A.F. Sarofim, "Fate of Fuel Nitrogen During
Pyrolysis and Oxidation", Second Annual Symposium on Stationary Source
Combustion, New Orleans, LA (1977).
7.	Levy, J.M., J.H.Pohl, A.F.Sarofim, and Y.H.Song, "Combustion Research on
The Fate of Fuel-Nitrogen Under Conditions of Pulverized Coal Combustion"
EPA Report EPA-600/7-78-165 (August, 1978).
8.	Fenimore, C.P., "Formation of Nitric Oxide from Fuel Nitrogen in Ethylene
Flames", Comb, and Flame, 19^, 289 (1972).
9a. DeSoete, G.G.,"Le Mecanisme de Formation D'Oxyde Azotique a Partir
D'Ammoniac et D"amines Dans Les Flammes D'Hydrocarbures", Revue de
L1Institute Francaia Du Petrole 28, 95 (1973).
b. DeSoete, G.G. and A. Queraud, "Formation D'Oxyde D'Azote Dans LesFlammes:
Formation D'Oxyde D'Azote A Partir De L'Azote Du Combustible. C-Addition
de NO Dans Les Melanges Inflammables", Report No.I.F.P. No. 21.681,
Institute Francais Du Petrole, Rueil Malmaison, France (1973).
98
-------
10.	Sarofim, A.F., G. C. Williams, M. Modell.. and S. H. Slater, "Conversion
of Fuel Nitrogen to Nitric Oxide in Preraixed and Diffusion Flames",
AIChE Symposium Series.148, Volume 7, 51 (1975)
11.	Levy, J.M., J.P.Longwell, A.F.Sarofim, T.L.Corley, M.P.Heap and T.J.
Tyson, "NO Abatement in Fossil Fuel Combustion: Chemical Kinetic Consi-
derations,^ Proceedings of the Third Symposium on Stationary Source
Combustion, Vol. IV, EPA-600/7-79-050, p.3 (March, 1979).
12.	See, also, Pershing, D.W. and J.O.L.Wendt, "Pulverized Coal Combustion:
The Influence of Flame Temperature and Coal Combustion on Thermal and
Fuel NO Sixteenth Symposium (International) on Combustion, The Combu-
stion Institute, p. 389, Pittsburgh (1977).
13.	Corley, T.L., "Development of a Kinetic Mechanism to Describe the Fate
of Fuel Nitrogen in Gaseous Systems," Fifth E.P.A. Fundamental Combus-
tion Research Workshop, Newport Beach, California, January, 1980.
14.	Heap, M.P., T.J.Tyson, J.E.Cichanwicz, R.Gersham, C.J.Kau, G.B.Martin
and W.S.Lanier, "Environmental Aspects of low BTU Gas Combustion,"
Sixteenth Symposium (International) on Combustion, The Combustion Insti-
tute, p.535 (1977).
15.	Recent examples include:
a.	Fenimore, C.P., "Reactions of Fuel-Nitrogen in Rich Flame Gases,"
Comb, and Flame 26, 249 (1976).
b.	Morley, C., "The Formation and Destruction of Hydrogen Cyanide
from Atmospheric and Fuel-Nitrogen in Rich, Atmospheric-Pressure
Flames," Comb, and Flame 27_, 189 (1976).
c.	Haynes, B.S., "Reactions of Ammonia and Nitric Oxide in the Burnt
Gases of Fuel-Rich Hydrocarbon-Air Flames", Comb, and Flame 28,
81 (1977).
d.	Haynes, B.S., "The Oxidation of Hydrogen Cyanide in Fuel-Rich
Flames," Comb, and Flame 28, 113 (1977).
16.	Sarofim, A.F., J.H.Pohl, and B.R.Taylor, "Strategies for Controlling
Nitragen Oxide Emissions During Combustion of Nitrogen Bearing Fuels,"
AIChE 69th Annual Meeting, Chicago (1976).
17.	Haynes, B.S., "Kinetics of Nitrogen Oxide Formation in Combustion," in
Progress in Astronautics and Aeronautics, Volume 62 (C.T.Bowman and
J.Birkeland, eds.), Alternative Hydrocarbon Fuels: Combustion and Chemi-
cal Kinetics, American Institute of Aeronautics and Astronautics (1978).
13. Levy, J.M., J.P.Longwell and A.F.Sarofim, "Conversion of Fuel-Nitrogen
to Nitrogen Oxides in Fossil Fuel Combustion: Mechanistic Considera1-
tions," Report to Energy and Environmental Research Corp. by the MIT
Energy Laboratory. EPa FCR Program (1978).
15. Engleman, V.S., "Detailed Approach to Kinetic Mechanisms in Complex
Systems," J. Phys. Chem., 81, 2320 (1977).
99
-------
20a. Mayer, S.W., L.Schieler, and H.S.Johnston, "Computation of High-
Temperature Rate Constants for Bimolecular Reactions of Combustion
Products," Eleventh Symposium (International) on Combustion, The
Combustion Institute, p. 837, Pittsburgh (1966).
b. Tunder, R., S.Mayer, E.Cook, and L.Schieler, Aerospace Corporation
Report TR-1001 (9210-02)-l (1967).
21. Branch, M.C., J.A.Miller, and R.J.Kee, "Chemical Kinetics of the
NH^/NO/CL System," 1979 Fall (Western States Section) Combustion Meeting
The Combustion Institute, WSS Paper No. 79-38, Berkeley, California,
October, 1979.
22a. Hack, W., H.Schacke, M.Schroter and H.G.Wagner, "Reaction Rates of
NH2 Radicals with NO, NO2, ^H;?, ^2^4 anc* ®t^ier Hydrocarbons," Seven-
teenth Symposium (International) on Combustion, The Combustion Insti-
tute, p. 505 (1979).
b. Silver, J.A., C.M.Gozewski, and C.E.Kolb,"Chemical Kinetics of the
NHj/NO Reaction System Under Combustion Exhaust Flow Conditions," to be
presented at the Eighteenth Syposium (International) on Combustion,
Waterloo, Ontario, August, 1980.
23.	Haynes, B.S., "The Formation and Behavior of Nitrogen Species in Fuel
Rich Hydrocarbon Flames," Ph.D. Thesis, University Of New South Wales,
Sydney, Australia (1975).
24.	Engleman, V.S., "Survey and Evaluation of Kinetic Data on Reactions in
Methane/Air Combustion," EPA Report, EPA-600/2-76-003 (1976).
25.	Cohen, R.C., "The High Temperature Oxidation and Pyrolysis of Ethane,"
Ph.D. Thesis, Princeton University, Department of Aerospace and
Mechanical Sciences (1972).
26a. Dryer, F.L., "High Temperature Oxidation of Carbon Monoxide and Methane
in a Turbulent Flow Reactor," Ph.D. Thesis, Princeton University, Depart-
ment of Aerospace and Mechanical Sciences (1972).
b. Dryer, F.L., and I. Glassman, "High Temperature Oxidation of CO and CH^"
Fourteenth Symposium (International) on Combustion, The Combustion
Institute, Pittsburgh, PA p.987 (1973).
27.	Dryer, F.L., and I. Glassman, "Combustion Chemistry of Chain Hydrocar-
bons," in Progress in Astronautics and Aeronautics, Volume 62, (Eowman,
C.T. and J. Birkeland eds.) Alternative Hydrocarbon Fuels: Combustion
and Chemical Kinetics, American Institute of Aeronautics and Astronau-
tics (1978).
28.	Edelroan, R.B. and O.F.Fortune, "A Quasi-Global Chemical Kinetic Model
for the Finite Rate Combustion of Hydrocarbon Fuels with Application
to Turbulent Burning and Mixing in Hypersonic Engines and Nozzle,"
AIAA Paper No. 69-86 (1969).
29.	Bowman, C.T. and A.S.Kesten, "Kinetic Modeling of Nitric Oxide Formation
in Combustion Processes," presented at the Fall Meeting, Western States
Sect-ion of the Combustion Institute, Paper No. 71-28, Irvine, California
(October, 1971).
100
-------
30.	Millikan, 11.C., "Non-equilibrium Soot Formation in Premixed Flames,"
J. Phys. Chem. _66, 794 (1962).
31.	Levy, J.M., "Modelitig of Fuel-Nitrogen Chemistry in Combustion: The In-
fluence of Hydrocarbons," Fifth E.P.A. Fundamental Combustion Research
Workshop, Newport Beach, California, January, 1980.
32.	Olson, D.B. and W.C.Gardiner, Jr., "Combustion of Methane in Fuel-Rich
Mixtures," Comb, and Flame 32_, 151 (1978).
33a. Dean, A.M., B.L.Craig, R.L.Johnson, M.C.Schultz, and E.E.Wang,"Shock
Tube Studies of Formaldehyde Pyrolysis," Seventeenth Symposium (Interna-
tional) on Combustion, The Combustion Institute, p.577 (1979).
b. Dean, A.M., R.L.Johnson, and D.C.Steiner, "Schock Tube Studies of
Formaldehyde Oxidation," in press 1979.
34.	Oj.son, D.B. arid W.C.Gardiner, Jr. , "An Evaluation of Methane Combustion
Mechanisms," J. Phys. Chem. _81, 2514 (1977).
35.	White, J.N. and W.C.Gardiner, Jr., "An Evaluation of Methane Combustion
Mechanisms: 2 - Comparison of Model Predictions with Experimental Data
from Shock-initiated Combustion of C0H9, C„H,, and C,H,," J. Phys. Chem.
JJ3, 562 (1979). '	Z z H	* 0
36.	Levy, J.M., unpublished results.
37.	Westbrook, C.K., "An Analytical Study of the Shock Tube Ignition of
Mixtures of Methane and Ethane," Comb. Sci. and Technology 20, 5 (1979).
101
-------
II I I II II I
300
200
>-
2
CL
Q.
100
Exp. Comp.
O • "NO
A ~ 3>NHi
i i i i i
I , i T
1 23456789 10
COMPUTED TIME (msec)
Figure la. Post-flame zone of a CjHjVO^/Nj flame doped with
700 PPM NH3 ($« 1.52, T « 2010 K) modeled with the Reaction
set of Table I. Data from Haynes [23].
102
-------
u
5
CL
Q-
400
I 300 -|
200
100
Comp.
1	I I I	I	I	I
1 23456789 10
COMPUTED TIME ( msec )
Figure lb. Post-flame zone of a C-H,/0„/N. flame dooed with
2 4 2 2
70J PPM (;f> = 1.52, T =* 2010 K) modeled with the Reaction
set of Table I. Data from Haynes [23].
103
-------
2000
in
O
2
CL
CL
1500
1000
I I I I I
Expt. Comp
500 —
12 3 4
COMPUTED
5 6 7 8 9 10
TIME ( msec)
Figure lc. Post-flame zone of a C^/O,,/]^ flame doped with
700 PPM NH^ (<{> = 1.52, T » 2010 K) modeled with the Reaction
set of Table I . Data from Haynes [23],
104
-------
400
300
2
£L
CL
200
100
Comp
~ < NH
1 2 34 5 6789 10
COMPUTED TIME (msec)
Figure 2a. Post-flame zone of a C0H./air flame dooed with
2 4
500 PPM NO ($ = 1.66, T «* 2000 JO modeled with the hydro-
carbon Tree reaction set of Table 1. Data from Havnes £233•
105
-------
500
I 4 00
i—m—r
CL
CL
300
Exp, Comp.
O	# H
200 1 I 1 i 1 I	I	I	I	I	
1 23456789 10
COMPUTED TIME (msec)
Figure 2b. Post-flame zone of a C2H^/air flame dooed with
500 PPM NO ( - 1.66, T = 2000 K) modeled with the hydro-
carbon-free reaction set of Table I. Data from Haynes [23],
106
-------
CH4 / AIR . 1 % NH3
ADIABATIC
-flow sections. Thus,
ECN » 4.38 x 10
-10
for T * 2 msec, but falls to 3.08 x 10 ^ after an addi-
lu '\
ticmal 20 casec in piug-iiow (total reaction time - 22 msec), whereas	-
2.51 x 10~5 for = 20 msec, but falls to 1.51 x 10"^ after an additional
2 msec in nlurr-flow*
107
-------
1500
1000 -
2
CL
CL
500
Comp
1 23456789 10
COMPUTED TIME ( msec )
Figure 4a. Post-flame zone of a C2H^/air flame doped with
500 PPM NO (<}> » 1.66, T * 2000 K) including hydrocarbon
"breakthrough', modeled with Gardiner's reaction set [32]
altered to include a ketene intermediate, suoplemented by the
reactions of Tables I-11I, Data from Haynes [23].
108
-------
400
Expt. Comp.
HCN
NO
300
2 200
£L
CL
100
6 7 8
10
COMPUTED TIME (msec)
Figure 4b. Post-flame zone of a CgH^/air flame dooed with
500 PT»M NO (
-------
TABLE I. SIMPLIFIED FUEL-NITROGEN REACTION MECHANISM*
Rate constant, = ATnexp(-E/RT)
, A	E
REACTION	(cm /mole-sec) n	(cal/mole) REFERENCE
1 HCN + OH ->-
CN + H20
2.00X1011
0.60
5000.
b
2 CN + H- •*
HCN + H
12
3.16x10
0
5000.
c
z
3 CN + OH-*
NCO + H
5.60xl013
0
0.
d
4 HCN + OH->
HNCO + H
4.00X1011
0
0.
e
5 NCO + H-
HNCO + H
3.20xl0U
0
0
8
z
6 NCO + H -*¦
NH + CO
3.00xl014
0
0
a
7 HNCO + H -+
NH- + CO
l.OOxlO13
0
0
a
8 NHg + H
NH2 + H2
5.00X1011
1.39X1011
3.50xl010
3.OOxlO10
l.OOxlO12
0.50
2000.
f
9	NH3 + OH-*-
10	NH2 + H ¦*
NH2 + H20
NH + H2
0
0.79
1600.
4460.
m,n,o
g
11	NH2 + OH
12	NH + H -+
NH + H20
N + H_
0.68
0.68
1300.
1900.
f
f
13 NH + OH
z
N + Ho0
S.OOxlO11
0.50
2000.
f
14 NO + H +
Jm
N + OH
1.34xl014
0
49200.
h
15 N + NO -»•
n2 + o
3.10xl013
0
344.
1
16 NH + NO ¦*
N-0 + H
9.00xl09
0.75
0.
j
17	N20 + H +
18	N20 + M +
19	NH + NH +
JL
N2 + OH
N2 + 0 + M
N + NH„
7.60xl013
1.42xl0W
3.60X1011
0
0
0.55
15101.
51280.
1900.
k
1
f
20	N + NH^-+-
21	NH + N +
z
NH2 + NH2
N- + H
2.lOxlO11
6.30X1011
0.50
0.50
23160.
0.
£
*
22 NH2 + NH^
JL
NH + NH3
1.70xl0U
0.63
3600.
f »P
110
-------
REFERENCES - TABLE I
For a discussion of reactions 1 - 18, see Reference [a],
[a]	Foster, D.E., J.B.Heywood, J.C.Keck, J.P.Longwell, A.F.Sarofim, and B.
R.Taylor, "Control of NO Emissions from Combustion of Fuels Derived
from Shale and Coal," Summary Technical Progress Report, October 23,
1978. Prepared for the Department of Energy, under Contract EX-76-A-
01-2295, The Energy Laboratory, MIT,
[b]	Tunder, R., S.Mayer, E.Cook, and L.Schieler, Aerospace Corp., Thermo-
themistry Research Dept., Aerospace Report No. TR-1001 (9210-02)-l,
AD 813 485 (1967).
[c]	Benson, S.W., D.M.Golden, and R.Shai*, "Estimating the Kinetics of Combu-
stion," EPA Report EPA-600/2-75-019 (1975).
[d]	Haynes, R.S., "The Formation and Behavior of Nitrogen Species in Fuel-
Rich Hydrocarbon Flames," PhD. Thesis, The University of New South Wales
CL975).
[e]	Estimate, this work.
[f]	Bahn, G.S., Reaction Rate Compilation for the N-O-H System, Gordon
and Breach (1968).
[g]	Mayer, S.W., L.Schieler, and H.S.Johnston, "Computations of High-Temper-
ature Rate Constants for Bimolecular Reactions of Combustion Products,"
Eleventh Symposium (International) on Combustion, The Combustion Insti-
tute, Pittsburgh, PA (1967).
[h]	Flower, W.L., R.K.Hanson, and C.H.Kruger, "Kinetics of the Reaction of
Nitric Oxide with Hydrogen," Fifteenth Symposium (International) on
Combustion, The Combustion Institute, p.823 (1974).
[ij Baulch, D.L., D.D.Drysdale, and A.C.Lloyd, "Critical Evaluation of Rate
Data for Homogeneous Gas-Phase Reactions of Interest in High Temperature
Systems, Vol. 4, "The University of Leeds 2, England (1969).
[j] Roose, T.R., R.K.Hanson, and C.H.Kruger, "Decomposition of NO in the
Presence of Nil-," presented at the 11th Annual International Shock Tube
Symposium, Seattle Washington, (1974).
tk] Baulch, D.L. et al., Evaluated Kinetic Data for High Temperature Reac-
tions, Volume 2, CRC Press (1973).
Ill
-------
Monat, J.P., R.K.Hanson, and C.H.Kruger, "Kinetics of Nitrous Oxide
Decomposition," Combustion Science and Technology, 16, 21-28 (1977).
Dove, J.E. £nd S.N.Wing, "Shock Tube Studies of the Reactions of
Hydrogen Atoms: I - The Reaction of H+NH_-»H?+NH ", Can. J. Chem., 52,
1171 (1974).
Zellner, R. and I.W.Smith, "Rate Constants for the Reactions of OH with
NH3 and HN03>" Chemical Physics Letters, 26 (1) (1974).
Perry, R.A., R.Atkinson and J.H.Pitts, Jr., "Rate Constants for the
Reactions of OH+K^S-^H^O+SH and OIl+NHo^-H~O+NH2Over the Temperature Range
297-427°K," The Journal of Chemical Physics, 64 (8) 3237 (1976).
Mayer, S.W. and L.Schieler, Aerospace Corp. Report TR-66 (8210-02)-3
(June, 1966).
Engleman, V.S., "Survey and Evaluation of Kinetic Data on Reactions
in Methane/Air Combustion, EPA Report, EPA-600/2-76-003 (1976).
Benson, S.W., D.M.Golden, R.W. Lawrence and R.Shaw, Quarterly Progress
Report No. 2, EPA Grant itR-800798 (Feb. 1973).
-------
TABLE II. REACTIONS INVOLVING CH TO BE APPENDED TO THE OLSON AND GARDINER
MECHANISM
kf « ATNexp {-E/RT}.

Reaction
3
A(cm /mole-sec)
N
E(Cal/mole)
1
CH2 + H « CH + H2
11
3.16 x 10
0.70
5,000
2
CH2 + OH * CII + E20
5.00 x 1011
0.50
6,000
3
CH + C02 = CHO + CO
1.00 x 1010
0.50
6,000
4
CH + OH - CHO + H
5.00 x 1011
0.50
10,000
5
CH + 02 » CHO + 0
5.00 x 1011
0.50
6,000
6
CH + 0 - CO + H'
5.00 x 1011
0.50
0
All rates from Englecan. V«S., "Survey and Evaluation of Kinetic Data on
Reactions tin Methane/Air Combustion," EPA Report EPA-600/2-76-003 (1976)
113
-------
TABLE III. HYDROCARBON FRAGMENT/FUEL-NITROGEN INTERACTIONS
kf = ATN exp {~E/rt}
3
Reaction	A(cm /mole-sec) N (Cal/mole) Ref.
1
CHA + CN
ss
CH3
+
HCN
3.16x1011
0.
70
5,000
a
2
ch2 + n2
ss
HCN
+
NH '
l.OOxlO14
0

60,000
b
3
CH2 + NO
85
ch2o
+
N
1.60X1012
0

7,000
a
4
CHO + N
W
CH
+
NO
l.OOxlO14
0

48,600
c
5
CHO + N
8
HCN
+
0
l.OOxlO1*
0

0
a
6
CHO + N

CO
+
NH
2.00X1011
0.
50
2,000
a
7
HCN + N
St
CH
+
N2
2.50X1011
0

16,000
b
8
HCN + 0
8
CH
+
NO
l.OOxlO14
0

72,000
b
9
CN + NH
-
CH

N2
l.OOxlO14
0

40,000
b
10
CH3 + NO
SZ
... - hcn+r2o
2.2Oxl011
0

0
d
References
[a]	Engleman, V.S. , "Survey and Evaluation of Kinetic Data on Reactions
in Methane/Air Combustion," EPA Report, EPA-600/2-76-003 (1976).
[b]	Benson, S.W., D.M. Golden, R.W. Lawrence, R. Shew^and R.W. Woolfolk,
"Estimating the Kinetics of Combustion: Including Oxides of Nitrogen
and Sulfur," EPA Report, EPA-600/2-75-019 (1975), and estimates based
thereon.
[c]	Estimate based upon Engleman (Reference [a]).
fd] Haynes, B.S., "Kinetics of Nitrogen Oxide Formation in Combustion,"
in Progress in Astronautics and Aeronautics* Vol. 62: Alternative
Hydrocarbon Fuels: Combustion and Chemical Kinetics, (C.T. Bowman
knd J. Birkeland, eds.) American Institute of Aeronautics and
Astronautics (1978).
114
-------
LOW NOx COMBUSTORS FOR
HIGH NITROGEN LIQUID FUELS
By:
G, C. England, M, P. Heap, D. W. Pershing,
J. H. Tomlinson, and T. L. Corely
Energy and Environmental Research Corporation
8001 Irvine Boulevard
Santa Ana, California 92705
115
-------
ABSTRACT
The results of bench-scale experiments In a 21-kW tunnel furnace show
that under unstaged combustion conditions total and fuel NO^ emissions from
twenty-six petroleum and alternative liquid fuels correlate well with fuel
nitrogen content. The optimization of staged combustion parameters in the
fuel-rich primary zone was studied in order to provide direct guidance for
advanced low-NOx burner designs for evaluation in a 900-kW pilot-scale com-
bustor. Detailed in-flame measurements were made in addition to exhaust
measurements to quantify the influence of first-stage stoichiometry and tem-
perature on the fate of fuel nitrogen species. Exhaust NO emissions were
found to be directly related to the amount of total fixed-nitrogen species
(TFN « NO+HCN+NH^) leaving the first stage. Increasing the temperature of
the primary zone decreased TFN concentration resulting in lower exhaust NO^
emission.
116
-------
ACKNOWLEDGMENTS
These investigations were carried out under United States Environmental
Protection Agency (EPA) Contracts 68-02-3125 and 68-02-2624. The support
and assistance of Mr. W. S. Lanier and Mr. G. B. Martin, the Project Officers,
has been appreciated. The authors also wish to acknowledge the assistance of
their colleagues, R. K. Nihart, J. A. Naccarato, and J. G. Llanos, in conduct-
ing the experiments; and they wish to express their gratitude to W. C. Rovesti
of EPRI, J. E. Haebig of Gulf Research, and L. Lukins of the U. S. Navy, for
their help in obtaining several alternative fuels.
117
-------
SECTION 1
INTRODUCTION
Combustion of liquid fuels derived from petroleum sources accounts for a
significant fraction of fossil fuel consumption in stationary combustors. As
petroleum reserves grow smaller, the United States is projected to place
heavy reliance on coal, the most abundant fossil fuel available, in the search
for new energy supplies. Coal can be burned directly or converted into either
a liquid or a gaseous fuel. A balanced fuel economy necessitates that in the
future many industrial users will burn petroleum- and coal- or shale-derived
liquid fuels. Since these liquid fuels have relatively high nitrogen content
and low hydrogen-to-carbon ratios, there will be the potential for adverse
environmental impact due to the increased emission of combustion-generated
pollutants unless preventative measures are taken (1-2). The pollutant of
major concern in this paper is nitrogen oxides (NO ). The paper addresses
X
the impact of switching from a petroleum fuel to a shale- or coal-derived
liquid, and on the mechanisms of combustion modification techniques used to
control NO^ emissions from all liquid fuels.
Alternative liquid fuels can be broadly classified as those synthesized
from the products of coal gasification, and those derived directly as liquids.
The fuels in the first category tend to be clean, low-boiling-point fuels
such as alcohols, and are essentially free from nitrogen and sulfur; thus,
their impact upon pollutant emissions is minimal. The liquids in the second
category may be compared to crude petroleum oils containing a wide range of
hydrocarbon compounds with boiling points from 300°K to greater than 900°K.
The bound nitrogen content of crude synfuels is generally higher than petro-
leum crudes, and for many applications it might be necessary to upgrade the
fuel by removing the nitrogen. Recognizing that alternative liquid fuels
contain more bound nitrogen than the petroleum fuels that they would be
replacing, one key factor in their production is to what extent combustion
118
-------
modification will allow control of N0x emissions and reduce the necessity
for substantial denitrification, thereby reducing the cost of synfuels.
Nitrogen oxides produced during combustion emanate from two sources.
Thermal NO is formed by the fixation of molecular nitrogen and its forma-
tion rate is strongly dependent upon temperature (3). Fuel NO is formed by
the oxidation of chemically-bound nitrogen in the fuel by reactions with a
weak temperature dependence, but a strong dependence upon oxygen avail-
ability (4-5-6-7). Thus, those emission control techniques which minimize
peak flame temperature by the addition of inert diluents (e.g., cooled recycled
combustion products or water addition) minimize thermal NO formation, but
have a minor impact upon fuel NO production. Staged heat release (staged com-
bustion) provides the most effective NO^ control technique for nitrogen-
containing fuels because fuel NO formation is mainly dependent upon local
stoichiometry. It can be accomplished either by separating the combustion
chamber into two zones and dividing the total combustion air into two streams,
or by appropriate burner design which promotes localized fuel-rich conditions.
Minimizing fuel NO^ formation requires the existence of a fuel-rich
primary combustion zone to maximize the conversion of fuel nitrogen to
molecular nitrogen since the fate of fuel-bound nitrogen is strongly con-
trolled by the reactant stoichiometry. Many studies (8-12) have shown that
under fuel-rich conditions the efficiency of conversion to N2 increases
significantly. Thus, there are two fuel nitrogen reaction paths leading to
the production of ^ or NO, namely:
Path A.	Fuel-lean
XN + Oxidant NO + ....
Path B.	Fuel-rich
XN +	N2 +
The objective of staged combustion emission control techniques is the pro-
vision of conditions which maximize N^ production via Path B. Two factors of
Practical importance are the residence time and the stoichiometry required to
maximize production in the fuel-rich primary zone.
119
-------
If the residence time is insufficient, then the original fuel nitrogen
species will exist in the gaseous state as some XN compound which can be
converted to NO in the second-stage heat release zone. The stoichiometry
required to achieve minimum XN concentrations at the exit of the primary
stage will be determined by (1) the rate of evolution of nitrogen species from
the fuel; (2) the inevitable distribution of stoichiometrics from fuel-rich
to fuel-lean which occurs because the primary zone is supplied by a diffusion
flame; and (3) the overall temperature of the primary zone. From equilib-
rium considerations the total fixed nitrogen (TFN given by NO + HCN + NH^)
is a in-fn-fmimi at approximately 65 percent theoretical air with levels less
than 10 ppm depending upon temperature and fuel C/H ratio. Exhaust NO emis-
3C
sions are considerably greater than levels predicted by equilibrium, suggest-
ing the existence of kinetic limitations in the fuel-rich primary stage.
NO formation during combustion of alternate fuels is not well-understood;
x
however, recent test results have Indicated that replacing a petroleum oil
with a coal- or shale-derived liquid may result in a major increase in NO
emissions. Bench-scale experiments (13) have shown that the smoke and com-
bustion characteristics of the SRC II coal liquids axe equivalent to light oil,
but uncontrolled NO emissions axe high due to the 0.8 to 1.2 percent N in the
fuel. Pilot-scale SRC II studies (14-16) have demonstrated that both fuel
blending and staged combustion are effective in reducing NO^ emissions and
that improved atomlzation^ increased preheat, and Increased excess O2 increase
NO^. Full-scale testing (17} has confirmed the need for optimized combustion
modifications. Similar results have also been achieved during bench-scale
(18) and field tests (19) with shale-derived liquids.
The results presented in this paper represent a portion of a study to
investigate the influence of spray properties, fuel composition, stoichiometry
and thermal environment on pollutant formation in liquid fuel flames. The
technical approach is experimental, involving parallel programs in two corn-
bus tors of different scales:
—	a downfired 21 kW (70,000 Btu/hr) refractory-lined tunnel furnace,
and
—	a 900 kW (3,000,000 Btu/hr) cold wall axisymmetrlc combustor
which simulates the firetube of a package boiler.
120
-------
This paper focuses on recent small-scale results on the relationship between
fuel chemistry and exhaust emissions under both excess air and staged com-
bustion conditions. In particular, the optimization of staged combustor
parameters in the fuel-rich primary zone were studied to provide direct guid-
ance for advanced low-NO^ burner design. Twenty-six liquid fuels (3 distill-
ate oils, 14 petroleum residual oils, 5 shale- and 4 coal-derived liquids)
were burned in the tunnel furnace and, in addition to exhaust concentrations,
in-flame measurements were made to quantify the influence of first-stage
temperature and stoichiometry on the fate of fuel nitrogen species.
121
-------
SECTION 2
EXPERIMENTAL SYSTEMS
TUNNEL FURNACE
The downfired tunnel furnace illustrated in Figure 1 was designed to
allow utilization of commercially-available spray nozzles, and yet be capable
of testing with artificial atmospheres. This connbustor, which has been
described in detail elsewhere (6), was 2.1 m long and 20 cm In inside diameter.
The walls consisted of insulating and high temperature castable refractories
and the full-load firing rate was 0.53 cc/sec, which corresponds to a nominal
heat release of 20 kW. All airstreams were metered with precision rotameters.
The main combustion air was preheated with an electric circulation heater; the
atomization air was not preheated. In certain tests the "air" was enriched
or replaced with varying amounts of carbon dioxide, argon, and oxygen, all of
which were supplied from high-pressure cylinders.
Fuel and air entered the combustion chamber through the burner illus-
trated in Figure 1. The combustion air was introduced axially (without swirl)
and a long refractory burner exit was used (L/"D » 5.0). The interchangeable
fuel injection system consisted of a 19 mm diameter stainless steel tube which
contained the atomizing air supply tube, the fuel oil supply tube, a cartridge
heater for final oil temperature control, and a chromel/alumel thermocouple
positioned in the Injector tip, prior to the nozzle for accurate oil tempera-
ture measurement. A commercial air-assisted ultrasonic oil atomizer (capac-
ity 0.55 cc/sec) was used in this investigation because it provided adequate
atomization of the heavy liquid fuels at relatively low flow rates. The drop-
size distribution was narrow and centered at 20 microns (6). The tip of the
fuel nozzle was positioned at the beginning of the burner divergent section
(as shown in Figure 1); and, in general, the visible flame front was displaced
one to four nozzle diameters from the nozzle tip.
122
-------
ANALYTICAL SYSTEM
Exhaust concentrations were monitored continuously using a chemilumi-
nescent analyzer for NO and NO , a NDIR analyzer for CO and C0_, and a para-
magnetic analyzer for 02» The flue gas was withdrawn from the stack through
a water-cooled, stainless steel probe using a stainless steel/Teflon sampling
pump. Sample conditioning prior to the instrumentation consisted of an ice
bath water condenser and glass wool and Teflon fiber filters. All sample
lines were 6.3 mm Teflon and all fittings 316 stainless steel.
In-flame temperature measurements were made with a standard suction
pyrometer containing a platinum/rhodium thermocouple. In-flame gas samples
were withdrawn with a long, stainless steel water-quench probe. HCN and NH^
were absorbed in a series of wet impingers and concentrations determined
using specific ion electrodes. Sulfide ion interference was minimized by
the addition of lead carbonate (20). Hydrocarbons were measured using a water
cooled probe, heated sample line, and an FID analyzer.
FUELS
Figure 2 illustrates the wide spectrum of composition for the distillate
oils (half-filled symbols), heavy petroleum liquids (open symbols), and alter-
native liquid fuels (solid symbols) investigated to date. The petroleum-
derived fuels had sulfur contents ranging from 0.2 to 2.22 percent with a
maximum nitrogen content of 0.86. The nitrogen content of the alternative
fuels range from 0.24 to 2.5 percent. Table 1 lists the complete chemical
analysis and physical properties of each fuel as determined by an independent
laboratory. The shale liquids included crude shale from the Paraho process
(A3) and four refined products: diesel fuel marine (DEM, Al) residual fuel
oil (A5), a 520-to-850°F distillation cut (A7), and a 5.75/1 medium/heavy
SRC II blend (A6), a heavy SRC II distillate (A9), and an SRC II blended with
the donor solvent (A4).
123
-------
SECTION 3
EXCESS AIR RESULTS
PETROLEUM LIQUIDS
To define the influence of fuel composition on total and fuel NO^ emis-
sions, each oil was tested under similar conditions in the tunnel furnace.
Fuel NO formation was determined by substitution of the combustion air with
x
a mixture of argon, oxygen, and carbon dioxide. The argon replaced the nitro-
gen, thereby eliminating thermal NO formation and the CO. provided the proper
X	«
heat capacity so that flame temperatures were matched. Total and fuel NO
emissions were measured with an air preheat level of 405 +5 K and an atomlza-
tion pressure of 15 pslg. Figure 3 presents a composite plot for total and
fuel NO^ (defined by argon substitution) as a function of weight percent
nitrogen in the fuel for a wide range of petroleum and blended distillate
fuels. In Figure 3 the various symbols represent different base fuels (see
Table I for symbol key). Those symbols shown with a line refer to distillate
or residual fuels doped with pyridine or thiophene. It can be seen that both
total and fuel NO Increase with increasing fuel nitrogen content, and that
total fuel nitrogen level is the dominant factor controlling fuel NO forma-
tlon in this system. The form of the nitrogen does not appear to signifi-
cantly influence fuel NO formation under excess air conditions, as doping
with a volatile nitrogen compound (pyridine) resulted in NO emission similar
<2C
to that from a less volatile residual oil of the same nitrogen content. Since
the data is for a system where very fine oil droplets (approximately 25 micron
mean diameter) are well-dispersed in the oxidizer under hot fuel-lean condi-
tions, it is not surprising that fuel NO^ emissions are somewhat higher than
those achieved In practical systems.
124
-------
ALTERNATIVE FUELS
Figure 4 presents a composite plot of total and fuel NO for the range
3C
of petroleum fuels together with alternative fuels and mixtures. The Paraho
shale was mixed with the same low sulfur oil used by Mansour (19). Synthoil
could not be pumped without blending and the results presented in Figure 4
refer to 80 and 90 percent Synthoil blends with distillate oil. The SRC II
blend refers to a mixture of SRC II and the donor solvent. Under the con-
ditions tested, fuel NO^ emissions increase approximately linearly with
increasing fuel nitrogen and it can be seen that the fate of fuel nitrogen
in alternative fuels is similar to that in petroleum-derived fuels. Figure 5
presents the fuel N0^ data plotted as a percentage of the fuel nitrogen con-
verted to fuel NO^. For low fuel nitrogen contents, the conversion decreases
rapidly (from greater than 90 percent) as fuel N increases. Eventually,
however, the conversion becomes almost independent of fuel nitrogen content;
hence, the linear dependence shown in Figure 4.
The absolute level of the fuel N conversion can be influenced by alter-
ing the fuel/air contacting and/or the fuel atomization (2), but the results
obtained in this study suggest that fuel nitrogen is the only first-order
fuel composition parameter controlling NO formation in fuel-lean flames. This
conclusion applies to petroleum-, coal-, and shale-derived liquid fuels. How-
ever, there appear to be second-order effects where the volatility of the fuel
nitrogen compound does have an influence upon fuel NO formation. Comparison
¦X
of the data for the fuels with fuel nitrogen content of approximately 0.24 per-
cent indicates that the highest conversion is achieved with a shale-derived
distillate fuel with a large volatile nitrogen fraction.
125
-------
SECTION 4
STAGED COMBUSTION RESULTS
POTENTIAL FOR NOx CONTROL
Staged combustion, i.e., the operation of a combustion system in which
the fuel originally burns under oxygen-deficient conditions, provides the
most cost-effective control techniques established to date for reducing fuel
NO^. Figure 6 shows the influence of primary zone stoichiometric ratio on
total NO^ emissions for two coal-derived and two shale-derived liquids under
staged combustion conditions and 3 percent overall excess 0^. All the data
in Figure 6 were obtained in the tunnel furnace with ultrasonic atomization
and with a first-stage residence time of approximately 800 ms. As the primary
zone becomes more fuel-rich, NO^ emissions decrease dramatically to a minimum
and then increase again. This trend is in agreement with previously-reported
data on petroleum fuels (21).
FUEL CHEMISTRY
First Stage Stoichiometry
In an effort to better understand the mechanisms of NO formation under
staged combustion conditions, the original furnace was modified to allow in-
flame sampling of the XN (NO, HCN, NH^) species and cooling of the first-stage
and/or second-stage combustion products, as illustrated in Figure 7. A "radia-
tion shield" (choke) was installed near the top of the furnace to minimize
the effects of downstream changes on the fuel vaporization zone. A secondary
air injection ring and cast refractory choke were installed at 41 in. to insure
isolation of the first stage. Variable cooling was achieved by insertion of
multiple stainless steel water-cooling colls.
Figures 8, 9, and 10 show typical results of the detailed in-flame measure-
ments made at the exit of the first stage for a distillate oil (Dl-Alaskan
126
-------
diesel), a high nitrogen residual oil (R-14-Kem County, California) and an alter-
native liquid fuel (A8-+850°F shale fraction). These measurements were made
on the centerline of the furnace at a distance of 104 cm (approximately
630 msec) from the oil nozzle. Detailed radial measurements indicated that
the concentration profile was essentially uniform at this location. All of
the in-flame data are reported on a dry, as-measured basis. After each
in-flame measurement, second-stage air was added at 107 cm and exhaust NO^
measurements were also made (shown on a dry, 0% O2 basis). In general,
decreasing the first-stage stoichiometric ratio reduced the NO concentration
leaving the first stage. However, below a stoichiometric ratio of approxi-
mately 0.8 significant amounts of NH^ and HCN were measured. Thus, there
exists a minimum in exhaust N0^ concentrations because of a competition
between decreased first-stage NO and increased oxidizable nitrogen species
such as HCN. Figures 9 and 10 indicate that the petroleum-derived oil (0.83
percent N) and the heavy shale liquid (2.49 percent N) produce large amounts
of HCN. In addition, both fuels exhibited a minimum in TFN at a first-stage
stoichiometry of approximately 0.8.
Data for the Alaskan dlesel oil (Figure 8) also show the presence of
much smaller but significant concentrations of HCN and NH^, although this
fuel is essentially nitrogen-free. Total conversion of the fuel nitrogen
would produce 21 ppm TFN at SR^«0.7. This confirms previous work (10-12)
which demonstrated that reactions involving hydrocarbon fragments and Nj or
NO can produce HCN.
Hydrocarbons
The rapid increase in HCN concentration below SR^-0.8 was accompanied by
an increase in hydrocarbon content of the partially oxidized combustion prod-
ucts. Figure 11 summarizes the in-flame hydrocarbon measurements for the
Alaskan Diesel (Dl), three petroleum-derived residual oils (Indoneslan-R4,
Alaskan-R9, Kern County-R14), three alternative liquids (SRC-II heavy dis-
tillate-A9, crude shale-A3, heavy shale fraction-A8) and methane containing
0.75 weight percent nitrogen as NH^C^"). Hydrocarbon concentrations correlate
well on the basis of first-stage stoichiometry. At very low stoichiometric
ratios the distillate oil (^) and CH^/NH3 produced slightly higher hydrocarbon
concentrations than the heavier liquid fuels.
127
-------
XN Distribution
Figure 12 shows typical results on the percentage of the original fuel
nitrogen existing as either NO, NH^ or HCN at various stoichiometric ratios
for four fuels. Above SR^«0.8, NO was the dominant TFN species} at lower
stoichiometrics HCN dominated with all fuels tested except the CH^/NH^.
Axial profiles with the liquid fuels indicate that near SR^-O.8, signifi-
cant amounts of NH^ may be formed early in the rich zone but they decay
rapidly. These data are in strong contrast to similar results obtained with
pulverized coal (20) which indicate that the preferred TFN species is a
strong function of coal composition.
In general, both the alternate and petroleum-derived liquid fuels
behaved very similarly with the exception of the Kern County, California
crude (R14). It produced less HCN under rich, conditions* and this tendency
cannot be readily associated with common fuel properties. Hydrocarbon and
nitrogen distillation data indicated that in terms of equilibrium volatile
evolution the Kern County fuel Is intermediate among the liquids tested.
The Indonesian oil was the lightest of the liquid fuels and it produced the
highest TFN concentration at the minimum (SR^-0.8).
SECOND-STAGE NO^ FORMATION
Exhaust NO^ emissions in a staged cambustor result from conversion of
TFN exiting the first stage and any thermal NO^ production during burnout.
Thermal NO^ production was not considered to be significant in this study
because changes in heat extraction in the burnout region had almost no effect
on final emissions. Figure 13 shows exhaust NO^ emissions as a function of
total fixed nitrogen in the first stage at stoichiometrics between 0.5 and
0.8 for all fuels. The form of this correlation can be compared with that
presented in Figure 3 for excess air conditions since the second stage burnout
can be considered an excess air flame. Exhaust emissions increase with
increasing oxidizable nitrogen content, but the conversion efficiency
decreases as the TFN concentration increases. There are three possible
explanations for the data scatter shown in Figure 13: (1) TFN is not indi-
cative of the oxidizable nitrogen compounds that are leaving the first stage;
(2) TFN conversion in the burnout zone is dependent upon the form of the TFN;
128
-------
and (3) TFN conversion is also dependent upon the oxidation of the partial
products of combustion at the exit of the fuel-rich zone.
IMPACT OF THERMAL ENVIRONMENT
The TFN concentrations shown in Figure 12 are in excess of equilibrium
levels and Sarofim and co-workers (25) have suggested that increasing the
temperature of the primary zone would prove beneficial. The results presented
in Figure 14 were obtained with the shale crude (A3) to demonstrate the impact
of first- and second-stage heat removal on the fate of fuel nitrogen. Fig-
ure 14a indicates that adding the radiation shield with cooling coils in both
the first- and second-stage (hence, increasing the temperature of the vapori-
zation zone) reduced the minimum NO emissions and shifted the optimum stoi-
chiometry more fuel-rich. Figure 14b shows that removing the water cooling
coils from tlie first stage reduced the exhaust emissions. Removing the second
stage coils did not alter the minimum level; however, it did shift the minimum
SR more fuel-rich.. Thus, the optimum thermal environment has a high tempera-
ture vaporization zone, a hot, rich hold-up zone, and a cooled second stage
(Figure 14c).
The axial profiles (22) provide an explanation for this shift in the
minimum emission levels. Heat extraction in the first stage impacts the
rate of decay of TFN. Under cold conditions, both NO and HCN essentially
freeze, whereas without heat extraction the initial rate of decay for all
three species is much faster leading to low TFN concentrations at the exit
of the fuel-rich first stage. It should be noted that heat extraction also
affects the rate of CO oxidation.
129
-------
SECTION 5
DISCUSSION
NO control via staged combustion involves operating the first stage of
the combustor at some optimum stoichiometry before adding air to complete
combustion. Measurements within the rich first stage show that at this
optimum stoichiometry NO, HCN and NH^ co-exist. For more fuel-rich condi-
tions TFN exists mainly as HCN and NH^, and for conditions less fuel-rich
than the optimum, NO is the primary fixed nitrogen species. Two factors
influence the stoichiometry at which minimum exhaust NO^ emissions occur:
(1) the fuel type, and (2) the temperature of the primary zone. The data
presented in Figure 14 clearly Indicate that an increase in the first-stage
temperature results in a decrease in NO^ emissions. The implication of this
data with respect to practical low N0x burner designs utilizing the staged
combustion concept appears to be that heat loss from the fuel-rich first stage
should be minimized in order to optimize NO control for liquid fuels.
Figure 15 presents the results of calculations made with an idealized
plug flow well-stirred reactor sequence and a kinetic mechanism for methane-
air-fuel nitrogen mixtures. The latter stages of the rich zone in the tunnel
furnace may approximate to plug flow conditions, but this reactor sequence
does not attempt to model the details of the initial mixing and heat release.
The kinetic mechanism (23) contains 36 species and 184 reactions and represents
an attempt to develop an adequate set which has been validated against a
diverse body of experimental data including ignition delay times, flame speeds
and species profiles. A methane plus ammonia mixture was used as the fuel to
simulate an experimental system without liquid drops. The model results are
in qualitative agreement with the experimental data. As the first stage
becomes more fuel-rich the MO concentration decays, and the HCN and NH^ con-
centrations increase (Figure 15a). The predicted distribution of nitrogenous
130
-------
species is in agreement with the experimental data. The three isothermal con-
ditions shown in Figure 15b were chosen because they include the range of
temperatures measured in the furnace. The model results reveal two of the
same trends which were observed experimentally: (1) minimum TEN decreases
with increasing temperature, and (2) the stoichiometry at which the minimum
occurs becomes more fuel-rich as the temperature increases. Thus, the data
and the calculations both indicate that a high temperature fuel-rich zone
favors lower NO emissions,
x
The information presented in Figure 13 on the conversion of TFN to NO
during burnout suggests that there might be a combination of certain fuel
and nitrogen species which would minimize the second-stage conversion to
NO. The presence of hydrocarbons leaving the first stage could affect
second-stage NO formation. Folsom, et al (24) have shown that in a simple
diffusion flame the conversion (or retention) of NH^, HCN or NO is affected
by the CH^ content of the fuel. NH^ conversion is low when hydrocarbons are
present, but hydrocarbons are necessary for the reduction of NO or HCN. Thus,
the minimum NO emission may depend upon the production of an optimum TFN/fuel
mixture for burnout. The presence of HCN in the rich zone is strongly linked
to the existence of hydrocarbons as shown in Figure 16. The line corresponds
to results obtained from the model calculations at 1650°K, If HCN is undesir-
able for the ultimate reduction of NO it appears that the temperature should
be sufficient to oxidize all of the hydrocarbon species within the rich zone.
131
-------
SECTION 6
CONCLUSIONS
The results of the bench-scale studies on the Influence of fuel properties
and thermal environment indicate that:
•	With liquid fuels, fuel nitrogen content is the primary composi-
tion variable affecting fuel NO formation. N0x emissions Increase
with increasing fuel nitrogen. Alternative liquid fuels correlate
with the high-nitrogen petroleum oils.
•	Staged combustion dramatically reduces both fuel and thermal
NO^ formation. Minimum emissions occur at a primary zone
stoichiometric ratio between 0.75 and 0.85 depending on the
combustion conditions.
•	First-stage stoichiometry determines the dominant TFN species.
Below SR^"0.8 HCN is the dominant species, and above SR^-0.8, NO
is the dominant species. NH^ concentrations at the first-stage
exit generally accounted for less than 20 percent of the fuel nitrogen.
•	Exhaust NO^ emissions are directly related to the TFN concentra-
tion at the first-stage exit. NO^ control for high-nitrogen fuels
is most effective when a rich primary zone is held at an optimum
stoichiometry to minimize the TFN concentration. This concentration
is further minimized by increasing the temperature of the fuel-rich
zone.
132
-------
REFERENCES
1.	Mansour, M. N. and M. Gerstein. Correlation of Fuel Nitrogen Conversion
to NOx During Combustion of Shale Oil Blends in a Utility Boiler. In:
Proceedings of Symposium on Combustion of Coal and Synthetic Fuels,
American Chemical Society, March 1978.
2.	Heap, M. P. et al. Control of Pollutant Emissions from Oil-Fired Package
Boilers. In: Proceedings of the Stationary Source Combustion Symposium,
EPA-600/2-76-1526, NTIS, Springfield, Virginia, 1976.
3.	Bowman, C. T. Kinetics of Nitric Oxide Formation in Combustion Processes.
In: Proceedings of Fourteenth Symposium (International) on Combustion,
The Combustion Institute, Pittsburgh, Pennsylvania 1973.
4.	Martin, G. B. and E. E. Berkau. An Investigation of the Conversion of
Various Fuel Nitrogen Compounds to NO in Oil Combustion. In: Proceed-
ings of AIChE Symposium Series No. 126, 68, 1972.
5.	Turner, D. W., R. L. Andrews, and C. W. Siegmund. Influence of Combus-
tion Modification and Fuel Nitrogen Content on Nitrogen Oxide Emissions
from Fuel Oil Combustion. In: Proceedings of AIChE Symposium Series
No. 126, 68, 1972.
6.	Pershing, D. W., J. E. Cichanowicz, G. C. England, M. P. Heap, and
G. B. Martin. The Influence of Fuel Composition and Flame Temperature
on the Formation of Thermal and Fuel N0X in Residual Oil FLames. In:
Proceedings of Seventeenth Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, Pennsylvania, 1979.
7.	Heap, M. P. NQx Emissions from Heavy Oil Combustion. International
Flame Research Foundation Report for Contract 68-02-0202, IJmuiden,
Holland, 1977.
8'. Malte, P. C. The Behavior of NH and CN in Nitrogen-Doped High Intensity
Recirculatlve Combustion. Paper presented at the WSS Combustion Institute,
Berkeley, California, October 1979.
9. Takagi, T., T. Tatsumi, and M. Ogasawara. Nitric Oxide Formation from
Fuel Nitrogen in Staged Combustion: Roles of HCN and NHi. Combustion and
Flame, 35, 17, 1979.
133
-------
10.	Fenimore, C. P. Studies of Fuel Nitrogen, species in. Rich Flame Cases.
In: Proceedings of Seventeenth Symposium (International) on Combustion,
The Combustion Institute, Pittsburgh, Pennsylvania, 1979.
11.	Haynes, B. S. Reactions of NH3 and NO in the Burnt Gases of Fuel-Rich
Hydrocarbon-Air Flames. Combustion and Flame, 28, 81, 1977.
12.	Gerhold, B. W., C. P. Fenimore, and P. K. Dederick. Two Stage Combustion
of Plain and N Doped Oil. In: Proceedings of Seventeenth Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh,
Pennsylvania,1979.
13.	Haebig, J. E., B. D. Davis, E. R. Dzuna. Preliminary Small-Scale Com-
bustion Tests of Coal Liquids. Environmental Science Technology, 10:3,
243, 1976.
14.	Muzio, L. J., and J. K. Arand. Small-Scale Evaluation of the Combustion
and Emissions Characteristics of SRC Oil. Paper presented at the ACS
Fuel Chemistry Symposium, Anaheim, California, March 1978.
15.	Downs, W., and A. J. Kubasco. Characterization and Combustion of SRC-II
Fuel Oil. EPRI Report No. FP-1028, Palo Alto, California, 1979.
16.	Hansour, M. N. Factors Influencing NOx Production During the Combustion
of SRC-II Fuel Oil. Paper presented at the WSS Combustion Institute,
Berkeley, California, October 1979.
17.	Hersch, S., B. F. Piper, D. J. Mormile, G. Stegman, E. G. Alfonsin, and
W. C. Rovesti. Combustion Demonstration of SRC-II Fuel Oil in a Utility
Boiler. Paper presented at the ASHE Winter Annual Meeting, New York,
New York, December 1979.
18.	Dzuna, E. R. Combustion Test of Shale Oils. Paper presented at the CSS
Combustion Institute, Columbus, Ohio, April 1976.
19.	Mansour, M. N., and M. Gerstein. Correlation of Fuel Nitrogen Conversion
to NOx During Combustion of Shale Oil Blends in a Utility Boiler. Paper
presented at the ACS Symposium on Combustion, Anaheim, California,
March 1978.
20.	Chen, S. L., M. P. Heap, R. K. Nihart, D. W. Pershing, and D. P. Rees.
The Influence of Fuel Composition on the Formation and Control of N0X in
Pulverized Coal Flames. Paper presented at the WSS Combustion Institute,
Irvine, California, 1980.
21.	England, G. C., D. W. Pershing, J. H. Tomlinson, and M. P. Heap. Emis-
sion Characteristics of Petroleum and Alternate Liquid Fuels. Paper
presented at the AFRC N0X Symposium, Houston, Texas, October 1979.
22.	England, G. C., M. P. Heap, D. W. Pershing, R. K. Nihart, and G. B. Martin.
Mechanisms of N0X Formation and Control: Alternative and Petroleum-Derived
Fuels. Paper presented at the Eighteenth Symposium (International) on Com-
bustion, The Combustion Institute, Waterloo, Canada, August 1980.
134
-------
23.	Corley, T. L. Development of a Kinetic Mechanism to Describe the Fate
of Fuel Nitrogen in Gaseous Systems. Paper presented at the Fifth
E.P.A. Fundamental Combustion Research Workshop, Newport Beach, California,
1980.
24.	Folsom, B. A., C. W. Courtney, and M. P. Heap. The Effects of LBG Com-
position and Combustor Characteristics on Fuel N0X Formation. Paper
presented at the ASME Gas Turbine Conference and Exhibit and Solar
Energy Conference, San Diego, California, 1979.
25.	Sarofim, A.F., H. Pohl and B. R. Taylor. Mechanisms and Kinetics of
N0X Formation: Recent Developments. Paper presented at the 69th Annual
Meeting AICHE, Chicago, Illinois, 1976.
135
-------
Ultrasonic Twin-fluid Atomizer
Burner
Viewing Port Section
I

Thermocouple,
Connection
Oil Heater
Connecti on
Combustion
Air	X
Atomizing Air
-—Oil Pressure
Tap
Oil Inlet
Viewing, Port

Itrasonic
Nozzle
Burner Detail
Insulating
Block
Insulating
Refractory
High Temperature
Refractory
Tunnel Furnace
Furnace Cross-Section
Figure 1. Details of the Tunnel Furnace System.
136
-------
a
1 O
o *
•-V
JL
e>
p
io
Al.
).0	2.0
Wt. Percent Sulfur
3.0
	- i 		—r-
1
t
-
%
¦
8
o
o
0
1 1
t ~

-
k Q
h ao
<
^ «•
o ^
-
®! * 1
A
l I l I 1 | II »» I » i i i i i
oO
6.0	7-Q	H.O	$.q
Carbon Hydrogen Ratio
O
*0K
O	*
I I 1 I 1 l-i I..I I > i I I I 1,1.
2 4 $ a 10 12 14 J6 ia
Conradson Carbon Residue (%)
Figure 2. Properties of Fuels Tested.
-------
1000
£
"O
CM
o
£
o
a.
a.
/ Indicates addition of pyridine
and/or thiophene to base fuel.
Fuel Nitrogen (vt.Z)
Figure 3a Total and Fuel N0x Emissions from Pure and Doped (pyridine
thiophene Petroleum Fuels-Tunnel Furnace.
138
-------
TOTAL NOL
- &

/
/

FUEL NOv
O PETROLEUM DERIVED
^ DFM
A SRC II BLEND
^ SHALE RESIDUAL
4SRC II
BSYNTHOIL/BLENDS
+ PAiymO SHALE/BLENDS
U2
1.6
2,0
Fuel Nitrogen (Wt. %)
Figure 4. The Effect of Fuel Nitrogen Content on Total
and Fuel NOx (5% Excess Oxygen).
139
-------
90
T
O Petrol euro Derived
^Coal Derived
• Shale Derived
50 .
X
X
X
J-
X
X
_L
.2 .4 .6 .8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6
Fuel Nitrogen (Wt. X)
Figure 5. Fuel Nitrogen Conversion - Comparison of Alternate and Petroleum Derived Fuels.
-------
800 _
600 _
IP
XI
IN
'8
§
i
P-*
400 _
200 _
0.60
O.70
0.80
0.90
SR.
Figure 6. Minimum NOx Levels Achieved with Alternate Fuels.
(Tunnel Furnace Primary Zone Residence Time .83 Sec.)
141
-------
Atomizing A1r
011
Combustion Air
Cast Refractory Choke Section
Nozzle
Position
Secondary A1r
Injector
Cooling Colls
Figure 7. Modified Tunnel Furnace.
-------
220 _
azxn
Figure 8. Detailed In-Flame Species Measurements
with Alaskan Diesel Oil (Dl).
143
-------
T3
2
3
3 600L.
T)
Figure 9. Detailed In-Flame Measurements
with Kem County Crude Oil (R14).
144
-------
5000
4600
4200
3800
3400
3000
2600
2200
1800
1200<
800
400(
0(
0.
i	1	1	r
O N0-|
~ HCN
O nh3
A ZXN]
• (N0X)E*
0.6
0.7
0.8
0.9
¦Q
Z02
SRi
Figure 10. Detailed In-Floae Measurements
With +850°F Shale Fraction (A8).
145
-------
10,000 _
T000 _
SR
1
Figure 11. First Stage Hydrocarbon Productioa.
146
-------
INDO/MALAYSIAN (R4)
SRC-IJ HEAVY DISTILLATE (A9)
KERN COUNTY (R14)
Figure 12. Distribution of First-Stage XN Species for Alternative
and Petroleum Derived Fuels.
-------
400
/•"N
CN
m 300
o
• 2.
*
*o
w 200
01a
100
1200
1400
800
1000
200
600
400
PPM EXN^ (dry, 0X0,,)
Figure 13. Exhaust N0^ Versus TFN at the Exit of the First Stage.
-------
v©
2
a.
a-
300
2?
a
o* 200
o
100
T	r
^ Without Radiation Shield
r~i With Radiation Shield -
*—1 First State
J	I—
0B7
j	L
0«8
t	1	r
t	r
O Primary * Secondary Cooling
Mq Cooling
A Secondary Cooling
t	r
o Prlaary ~ Secondary Cooling
O Radiation Shield ~
Secondary Cooling
-I	1-
0,9 Q»7
,0,8
0.9
-0®7
0*8
0.9
SR.
(a)
(b)
Cc)
Figure 14. Influence of Heat Extraction Profile in the First and
Second Stage upon Exhaust NOx Emissions.
-------
1400
1200
1000
800
600
400
200
AIR v^WSR
ch4+nh3
(0.86 WT % N)
400 MS
ISOTHERMAL
O N0
A HCN
O NH3
0.5
0.6
0.7
(a)
1
1 XN AT;
0 1350K
O .16SQK
1900K
1
0.8 0«9 SRj 0.5 0.6
1
0,7
(b)
1
0.8 0.9
Figure IS. Calculated XN Species Distributions,
-------

« 1000
a
a
at
to
at
•d
g 100
ss
£
(X4
10
r—r
i i m
T	1 I I I
T	1 I II	1	1 I U
I?A
10
CALCULATION
I I I I	I	I II I	» » » I
100	1000	10000
PPM HC (wet, as measured)
Figure 16. HCN Production Versus Hydrocarbons in the First Stage.
-------



Table
1. Detailed Fuel Analyses






1)1
02
03
«1
*2
R3
R4
K5
«6
#7
RB
R9
RiO

1 1
Alaskan
Diesel
W. Texas
OUsel
California
No. 2 Oil
East
Coast
Mi
A
~
V
Ultiaute Analysis:













Carbon, i
86.99
aa.os
86.B
86.64
86.78
86.57
86.53
85.92
84.82
84.62
85.24
86.04
85.75
Hydrogen, 1
12.07
9.76
12.52
12.31
11.95
12.52
11.93
12.05
11.21
10.77
10.96
11.18
11.83
Nitrogen, 1
0.008
0.026
0.063
0.16
0.18
0.22
0.24
0.24
0.34
0.36
0.40
0.51
0.62
Sulfur, t
0.31
i.iia
U.2?
0.36
0.67
0.21
0.22
0.93
2.26
2.44
2.22
1.63
1.05
Hill, t
<.001
<•001
<.001
0.023
0.0)2
0.02
0.036
0.031
0.067
0.027
0.081
0.034
0.038
Onjrueo, t
0.62
0.24
0.36
0.61
0.41
0.46
1.04
0.83
1.3
1.78
1.10
0.61
0.71
ConndsMi Carbon Residue, X



2.1
6.0
4.4
3.98
5.1
12.4
14.8
6.8
12.9

Aspbaltcne, *



0.34
3.24
0.94
0-?4
2,59
4.04
7.02
8.4
5.6

Hash Point. "F



206
360
326
210
176
*75
155
210
215

Pour Point, "f



SO
48
105
61
48
66
40
58
3a
19.5
API Gravity at 60°F
33.1
ia.3
32.6
24.9
19.8
25.1
21.8
23.3
15.4
13.2
14.1
15.6
246.1
Viscosity. SSU. at 140*F
33.0
32.0
30.8
131.Z
490
222.4
199
113.2
1049
835
742
1.071
70.00
at ZIO'F
29.5
2a.a
29.5
45
131.S
£9.6
65
50,5
240
181
19§.7
194

Heat of Combustion:













Gross Btu/lb


19.330
19.260
19,070
19.110
19.070
18.400
18,520
18.240
IA.240
18,470

Net btu/lb



18.140
17,980
17.970
,17.980
17,300
17.500
17.260
17.400
17.580

Calciuai. plan



7.1
1.2
9.52
14
8.7
9.2
4.4
M
6.9

Iron, ppw



16
2.6
123.6
16
6.5
13.2
19
11
24

Nayauese. wm



0.09
0.02
0.46
0.13
0.09
0.10
0.13
0.09
0.06

Kayitesiu*. ppa



3.7
0.08
2.23
3.6
3.6
3.3
0.4
3.8
1.4

Nickel, p|M



6.7
13
14.10
19

32.7
29
52
50

Sodium. ppa



37
0.98
3.74
15

'&4.5
3.6
32
37

VjnaJluw. pp«



14
25
3.11
101

81.5
45
226
67

-------


Table 1.
Detailed Fuel Analyses
(Continued)





Rll
R12
R13
R14
A1
A2
A3
A4
A5
A6
A7
A8
A9

California
California
California
California
(Kern County)
Shale-
Oertved
DFM
Syntholl
Crude
Shale
SRC 11
Blend
Shale
Residua)
SRC 11
Shale
Fraction
(520-850'F)
Shale
Fraction
(+850*F)
SRC II
Heavy
Distillate

o
0
n
0

¦
•
~
L
~
fc
m
A
Ultimate Analysts:













Carbon, X
OS.4
85.33
86.66
86.6)
86.18
86.30
84.6
89.91
86.71
85.9)
85.39
85.92
88.98
Hydrogen, X
11.44
11.23
10.44
10.9}
13.00
7.44
11.3
9.27
12.76
8.74
11.53
10.41
7.64
Mflroyeii. 1
0.77
0.79
0.86
0.63
0.24
1.36
2.08
0.45
0.46
0.96
1.92
2.49
1.03
Sulfur, X
1.63
1.60
0.99
1.16
0.51
0.80
0.63
0.065
0.28
0.30
0.72
0.63
0.39
Ash. X
0.043
0.032
0.20
0.030
0.003
1.56
.026
0.004
0.009
0.04
0.002
0.24
0.058
Oxyyen, X
0.71
1.02
0.85
0.44
1.07
2.54
1.36
0.30
0.03
4.08
0.44
0.11
1.90
Conradson Carbon Residue, X
8.72
9.22
15.2
8.3
4.1
23.9
2.9
6.18
0.19
0.51
0.07
9.3

Asfha 1 tene, X
5.18
5.18
8.62
3.92
0.036
16.56
1.33
4.10
0.083

0.12
4.24

Flash Point, *F

150
180
255
lt2
210
250
70
235
1.73
255
370
265
Pour Point, *F
ia
30
42
65
40
80
80
<-72
90
-55
70
95
8
API Gravity at 60°F
15.4
15.1
12.6
12.3
33.1
S-1.14
20.3
10.0
29.0

22.3
12.0
1.3
Viscosity, SSU, at 140°F
854
748.0
720
4630
36.1
10.880
97
40.6
54.3
39
62.9
3050
67.2
at 210°F
129
131.6
200
352
30.7
575
44.1
32.5
37.3

41.8
490
41.3
Heat of Combustion:













Gross Btu/lb
18.470
18,460
18.230
18,430
19.430
16,480
18.290
17,980
19.350
17.100
18.520
18.000
17,120
Net Btu/lb
17.430
17.440
17.280
17.430
18,240
15.BOO
17.260
17.130
18.190

17.470
17.030
16,240
Calciun, ppw
21
14
90.6
4.4
0.13
1.670
1.5
0.33
4.20

<,05
238

Iron, ppiu
n
53
77.2
15
6.3
109
47.9
3.9
<0.5

2.9
86

Hanyanese, pi**
0.8
0.1
0.87
0.15
0.06
6.2
0.17
<0.5
<0.5

0.033
1.3

Maynesluw, ppw
5.1
3.8
31.4
1.1

170
5.40
0.17
0.15

0.021
51

Nickel, ppw
65
82
88.0
68
0.43
2.6
5.00
<0.5
<0.5

<0.5
7.4

Sodium, ppw
21
2.6
22.3
3.4
0.09
148
11.71
0.31
2.51

<0.1
U

Vanadium, ppia
44
53
66.2
39
<•1
6.5
<.3
<1.0
<1.0

<0.2
1.1

-------
FATE OF COAL NITROGEN DURING COMBUSTION
By:
S. L. Chen, M. P. Heap, D. W. Pershing, R. K. Nihart,
and D. P. Rees
Energy and Environmental Research Corporation
8001 Irvine Boulevard
Santa Ana, California 92705
154
-------
ABSTRACT
Twenty-six coals covering all ranks have been burned under a wide
variety of conditions to ascertain the impact of coal properties on the fate
of fuel nitrogen. Three burner systems were used to vary the rate of fuel
air mixing and fuel NO was Identified by using a nitrogen free oxidant. It
was found that fuel nitrogen content is not the only property controlling
fuel NO formation. It appears that nitrogen volatility as well as total
nitrogen content is Important, particularly under well-mixed conditions.
Detailed specie concentration measurements were made under fuel rich
conditions and it was found that:
the partition of nitrogen between NO, NH^ and HCN was
dependent upon coal type;
total gas-phase nitrogen specie (fuel rich) correlated with
exhaust NO (fuel lean);
reducing the temperature of the first stage increased gas-phase
nitrogen specie concentrations but reduced fuel emissions.
This work will help in the generalization of low NO burner technology to a
wide range of fuels.	x
155
-------
ACKNOWLEDGMENTS
This work was supported under EPA Contract 68-02-2667. The authors
are happy to acknowledge the contribution of their colleagues in the conduct
of the experiments, particularly W. R. Seeker for his help with particle
temperature measurement and also the encouragement and ideas resulting from
discussions with A. Axworthy, A. F. Sarofim and P. Soloman.
156
-------
SECTION 1
INTRODUCTION
The development of NO control strategies for coal-fired boilers has
received considerable attention over the past decade. Pilot-scale studies*
established that fuel/air mixing controlled by burner design dictated the
formation of NO in pulverized coal flames, and that rapid mixing of coal
volatiles and air promoted NO formation. Bench-scale and research studies
2
further established the significance of fuel nitrogen oxidation, the factors
3 4
controlling fuel nitrogen evolution, ' and the conditions that affect NO
5 6
formation from pulverized coal under fuel-rich conditions. ' Burner
7 8 9
designs have been demonstrated at various scales ' ' which are capable of
minimizing emissions from field-operating boilers, and more advanced combustion
control techniques have been proposed which will radically change current
state-of-the-art boiler designs.*^ This study is directed toward an assessment
of the impact of coal properties on the fate of fuel bound nitrogen under both
fuel lean and fuel rich conbustion conditions.
The impact of coal properties on NO emissions from practical boilers is .
of considerable importance at a time when boiler operators are being faced
with more stringent pollution control requirements, and the need to burn a
wide range of fuels. Twenty^ six coals from three continents have been burned
under well-defined conditions using air and nitrogen-free mixtures as the
oxidant. Experimental data has been obtained on the fate of fuel nitrogen
specie under fuel-rich conditions as a means of establishing optimum time/
temperature stoichiometry histories for minimum NO production under staged
combustion conditions.
157
-------
SECTION 2
EXPERIMENTAL
Figure 1 presents a schematic of the down fired tunnel furnace and the
three burner systems used in this study. Complete details of both the furnace
and the associated measurement systems have been presented elsewhere^.
Pulverized coal was supplied pneumatically from a hopper-fed screw feeder to
one of two burners which allowed the coal to be burned entirely premixed or
as a diffusion flame. Premixing was achieved by direct impingement of the
coal jets with the main combustion air supply in a premixing chamber which
was separated from the combustion zone by a series of watercooled tubes which
prevented flashback (see Fig. 1). The coal/air mixture burned in a plug flow
mode with the ignition zone situated in the refractory divergent. The coal-
plus-transport air could also be supplied to a variable swirl burner and
injected either axially or radially, thus allowing experiments to be conducted
with very different fuel/air mixing characteristics. Variable swirl was
achieved by dividing the total oxidant flow into two streams, one of which
flowed axially around the fuel injector and the other through swirl vanes.
Swirling all the combustion air and injecting the coal/air mixture radially
into the combustion airstream produced a flame with backmixed characteristics.
An axial diffusion flame was obtained with a single-hole axial fuel injector
and sufficient swirl in the combustion air to provide a stable ignition zone
at the fuel injector. The characteristics of the solid fuels tested are
presented in Table 1.
158
-------
SECTION 3
FUEL-LEAN CONDITIONS
Figure 2 shows fuel NO emissions as a function of percent nitrogen
in the original coal (dry, ash free) for 26 different coals ranging from
anthracite through lignite. These data were obtained in the unmodified,*-*
tunnel furnace at a nominal firing rate of 21 Kw (approximately 5 lbs/hr coal)
with 5 percent excess O2 and 550°K secondary-air preheat. The particle-size
distribution was 65 percent through 200 mesh and the unpreheated primary air
was maintained at 15 percent of stoichiometric air. Fuel NO was defined as
the emission when molecular nitrogen was excluded from the oxidant and the
fuel was burned in a mixture of argon, carbon dioxide, and oxygen.
The data obtained with the premixed burner (upper plot) indicate that
in general fuel NO increases with increasing fuel nitrogen; however, within
any rank there are other properties which have a significant impact on fuel
nitrogen oxidation. For example, the Savage (A) and Beulah (Q) lignites
both contain approximately 1.05 weight percent nitrogen, yet their emissions
differ by more than 60 percent. The German bituminous coal (Saar<$) correlates
well with the American bituminous coals; however, a predried Australian lignite
(p* ) gives uncharacteristically high emissions. As expected, the low
volatile anthracite ( Q) produces significantly less fuel NO than the high
volatile bituminous coals.
Radial swirling and axial diffusion flames were also used to investigate
the coupling between fuel properties and fuel/air contacting. These data
(lower plots, Figure 2) indicate that fuel effects are also significant in
diffusion flames. Fuel NO emissions produced with the radial diffusions burner
also tend to increase with increasing fuel nitrogen content. The axial flame
data show no obvious dependence on nitrogen content, but other composition
parameters are obviously of major importance. The Australian lignite ( )
produced an extremely stable, well attached axial flame and uncharacteristically
low NO emissions.
Comparison of data for the three burner types indicates that initial
premixing of the fuel and air leads to the highest conversion of the fuel nitrogen
159
-------
to fuel NO, particularly with high volatile, bituminous coals. This can be
attributed to the fact that as the coal/air mixing varies from premixed to
axial diffusion there is less oxygen in contact with the volatile fuel nitrogen
fractions, and, therefore, the formation from the volatile nitrogen specie
is maximized. The anthracite data ( {J ) indicate that with the low volatile
coal, initial fuel/air contacting has relatively little effect on the exhaust
emissions.
Attempts to correlate the data shown in Figure 2 on the basis of the nitrogen
lost in the ASTM volatile determination or with such basic fuel properties
12
as % nitrogen, % volatiles or coal rank proved unsuccessful. Axworthy et al
carried out a series of inert pyrolysis experiments and measured the volatile
nitrogen which was converted to HCN at 1373°K using six of the coals from this
study and Figure 3 shows that this pyrolysis data agrees qualitatively with the
experimental results obtained in the premixed and radial diffusion flames.
The Savage lignite (A) has the highest percentage conversion of fuel nitrogen
to NO in the combustion experiment and also has the highest fractional volatile
nitrogen yield in the inert pyrolysis study. The slope of the correlation
decreases with decreasing mixing rate (premixed to radial diffusion) as would
be expected.
160
-------
SECTION 4
FUEL-RICH CONDITIONS
The control of NO emissions from coal fired combustors involves operations
under fuel rich conditions initially, therefore the optimization of the residence
time, temperature and stoichiometry of this initial zone and the impact of fuel
properties are of considerable practical significance.
PRIMARY ZONE STOICHIOMETRY
Figure 4 summarizes staged combustion results for a variety of fuels
with both the radial diffusion and premixed burners. The second stage air was
injected radially inward from two sides at a distance of 0.86m from the burner.
All of the data were obtained at 5 percent overall excess 02*
Figure 4a shows typical results for two bituminous coals (Utah, Gauley Eagle),
two lignite coals (Savage, Beulah) and a Pennsylvania anthracite. For most
of the coals tested, as the first stage stoichiometry is reduced the NO decreases,
reaches a m-tn-tmiini before increasing as the stoichiometry is further reduced.
Anthracite (and a Utah coal char, not shown) do not exhibit this minimum.
Figure 4b summarizes this data in the form of minimum NO emissions as a function
of the stoichiometric ratio at which the minimum occurs. Decreasing the initial
fuel/air contacting rate decreases the minimum achievable NO emissions. Further,
the optimum stoichiometric ratio depends both upon the fuel composition and
the initial mixing rate. Detailed in-flame measurements of gas concentrations,
solids composition and concentration, and temperature were made for nine coals
to define the relationship between coal composition and first stage composition.
Figure 5 shows a set of typical results for the Beulah, North Dakota lignite
with the premixed burner. All gas-phase concentration measurements are shown
on an as-measured (dry) basis, except the exhaust NO concentrations which have
been reduced to 0 percent 0£ (SR ¦ 1.0). As the first-stage stoichiometry
is reduced, the NO exiting the first stage decreases, but ultimately this is
compensated for by an increase in other oxidizable nitrogen specie (e.g., HCN
and NHO. The particle temperatures (measured at 0.53 m using two-color
13
pyrometry) also decrease and a significant amount of both the nitrogen and
carbon remain in the solid phase and are carried into the second-stage flame.
161
-------
Thus, there exists a minimum in exhaust NO concentrations because of the com-
petition between decreased first-stage NO and increased oxidizable gas- and
solid-phase nitrogen specie.
FUEL PROPERTIES
The relationship between fuel composition and the fraction of the fuel
nitrogen retained in the solid phase under a given set of combustion conditions
must be known if the optimum first-stage stoichiometry is to be predicted for
a given fuel. Figure 6a shows the fraction of the initial nitrogen which
remained in the char at 1.17m as a function of the initial nitrogen content for
the premixed burner at a first-stage stoichiometry of 0.5. As expected, there
is no obvious correlation with either total, fuel nitrogen or rank. A standard
ASTM proximate analysis of the original coal was made and the resulting solid
was analyzed for nitrogen to determine the fraction of the initial nitrogen
volatized under standard conditions. Figure 6b shows these results and indicates
that there is at least qualitative agreement between the in-flame solids
sampling results and the modified ASTM coal analysis. The Savage, North Dakota
lignite (£ ) retained only approximately 33 percent of its original nitrogen
in the char after the ASTM volatile determination, and a similar, small fraction
of the initial nitrogen was found in the solid material exiting the first stage
of the experimental furnace. In contrast, the Alabama, bituminous coal
(Rosa, 0) retained over 60 percent of its nitrogen in the ASTM test, and also
had a much higher fraction of its initial nitrogen still in the solid phase
at the end of the first stage of the furnace.
Figure 7 summarizes the total gas-phase fixed nitrogen (TFN » NO +
NH^ + HCN) measurements for the nine coals tested. These data were obtaine
at a distance of 1.17m and a primary-zone stoichiometry of 0.75 with the
premixed burner. In general, TFN increases with increasing fuel nitrogen
content (Figure 7a) and correlates well with the exhaust NO emissions measured
under unstaged (excess air) conditions (Figure 7b). Those coals which have
high emissions under excess-air conditions, produce large amounts of gas-phase
nitrogen specie under fuel rich conditions (e.g., the Utah, bituminous coal, Q )
The low-volatile anthracite (Q ) gave very low emissions under excess-air
operation and produced little TFN under fuel rich conditions.
162
-------
Figure 8 shows the influence of coal composition on the partition of
the gas-phase nitrogen specie. The upper curves are for bituminous coals from
the West (Utah), the South (Alabama), and Germany (Saar). The fraction of the
TFN present as NH^ was small for all of the bituminous-coal studies (including
the Rosa, Alabama coal, not shown). The Utah coal produced a significant
amount of HCN compared to the other bituminous coals.
The Pennsylvania anthracite (lower left) formed almost no HCN or ammonia;
at all stoichiometries the dominant TFN specie was NO. The Western, sub-
bituminous coals exhibited much the same behavior as the Western lignite coals.
In each of these cases, the dominant TFN specie was NH^ below a stoichiometric
ratio of approximately 0.7. HCN concentrations at 1.17m were small for all of
the sub-bituminous and lignite coals. The fraction of the TFN leaving the
first stage as NO was also substantially smaller than in the case of the bitu-
minous and anthracite coals.
TIME/TEMPERATURE HISTORY
Figure 9 shows detailed axial profiles for 3 coals at a common stoichiometric
ratio (SR^ ¦ 0.60) without the addition of second-stage air. The burnout
results (based on the CO and the C0^ measurements) indicate that the anthracite
bums much slower initially than the bituminous and lignite coals. The
first-stage NO from the anthracite forms slowly and then subsequently decays.
Similar decays were observed with the Utah, bituminous and the North Dakota
lignite; however, the initial NO formation was too rapid to be measured.
Significant amounts of HCN or ammonia were not observed with the Pennsyl-
vania anthracite. Large quantities of HCN were formed and subsequently decayed
with the Utah coal; small, relatively constant amounts of HCN were measured
with the lignite coal. Significant amounts of NH^ were measured at all axial
Positions with the lignite and to a lesser extent with the bituminous coals.
From these data it is clear that increasing the residence time in the first
stage of a pulverized-coal combustor decreases the exhaust NO emissions because
it allows time for the TFN concentrations to decrease from- their early high
values. It should be noted that although the input mixture is overall fuel rich,
the gas-phase will be oxidizing in the early stages.
163
-------
Figure 10 shows similar results for the Utah coal with the radial-diffusion
burner at a first-stage stoichiometry of 0.8. The uncooled data (open symbols)
are analygous to the premixed results presented in Figure 9. As noted previously,
HCN concentrations exceed the NH^ concentrations and the NO decays rapidly with
axial distance. Figure 10 also shows data for a case in which the first stage
was heavily cooled by inserting a water-cooled coil (see Figure 1). This coil
reduced the gas temperature approximately 350K and appears to have frozen
the NO concentration. However, the HCN and NH^ concentration continued to decay
with time. Overall reduction of the first-stage temperature increased the TFN
concentration at 1.17m substantially, because of the decreased rate of NO
reduction. However, when second-stage air was added at 1.17m the exhaust
emissions were lower with the lower first stage temperature which gave the higher
TFN concentration.
164
-------
SECTION 5
CONCLUDING REMARKS
Results have been presented ongoing Investigations to assess the impact
of coal properties on nitrogen oxide formation in pulverized coal flames.
The effort to date has concentrated upon providing information to answer the
following questions:
•	What are the fuel characteristics that dominate fuel NO formation?
Are they entirely chemical in nature? What is the impact of
physical properties such as the particle size of the thermal
environment on fuel NO formation?
•	To what extent does fuel/air contacting interact with fuel
characteristics in the formation of fuel NO?
Is it possible to assess the emission characteristics of fuels
in practical systems without prior field tests?
Conclusions to date can be summarized as follows:
fuel-lean
Tests with 26 different coals indicate that fuel nitrogen content is
not the only fuel property controlling the fate of fuel bound nitrogen during
combustion. The volatility of fuel nitrogen compounds appears to affect NO
formation particularly under well mixed conditions. Fuel NO emissions are
reduced as the rate of fuel/air mixing is reduced and become even less dependent
upon total fuel nitrogen content.
FUEL-RICH
Detailed studies on the optimization of the fuel-rich stage suggest that
the optimum stoichiometry for minimum emissions is a function of both fuel
composition and primary-zone mixing. As first-stage stoichiometry is decreased,
the NO formed in the first stage decreases, but the other oxidizable gas nitrogen
specie increase as does nitrogen retention in the char. Nitrogen retention
in the solid phase may be correlated with appropriate coal analyses, but the
production of gas-phase nitrogen specie appears to correlate only with excess
air emissions.
165
-------
The distribution of the TFN specie is strongly dependent upon the coal
composition. With bituminous coals NH^ appears to be a substantially less
fraction of the tota than with sub-bituminous and lignite coals. Of the nine
coals tested in detail, only one (Utah bituminous) formed substantial amounts
of HCN under well mixed conditions. Pennsylvania anthracite formed almost no
HCN or ammonia, even under extremely fuel rich conditions.
Johnson, et al,*^ showed that heat extraction in the primary stage of low
NO combustor decreased exhaust NO emissions. The results from the present
xx
study indicate that reducing the first stage temperature slowed the rate of
decay of both NO and other XN specie. Although the lower temperature fuel rich
stage produced a higher TFN concentration, this did not result in higher final
NO emissions.
166
-------
REFERENCES
1.	Heap, M. P., T. M. Lowes and R. Walmsley. Emission of Nitric
Oxide from Large Turbulent Diffusion Flames. In: Proceedings of
the Fourteenth Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, PA 1973. p. 883.
2.	Pershing, D. W. and J.O.L. Wendt. Pulverized Coal Combustion,
The Influence of Flame Temperature and Coal Composition on Thermal
and Fuel NO . In: Proceedings of the Sixteenth Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh,
PA, 1977. p. 389.
3.	Pohl, J. H. and A. F. Sarofim. Devolatilization and Oxidation of
Coal Nitrogen. In: Proceedings of the Sixteenth Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh,
PA, 1977. p. 491.
4.	Solomon, P. R. and M. B. Colket. Coal Devolatilization. In: Proceedings
of the Seventeenth Symposium (International) on Combustion, The
"Combustion Institute, Pittsburgh, PA, 1979. p. 131.
5.	Wendt, J.O.L., D. W. Pershing, J. W. Lee and J. W. Glass. Pulverized
Coal Combustion, NO Formation Mechanisms under Fuel-Rich and
Staged Combustion Conditions. In: Proceedings of the Seventeenth
Symposium (International) on Combustion, The Combustion Institute,
Pittsburgh, PA, 1979. p. 77.
6.	Rees, D. P., L. D. Smoot and P. 0. Hedman. Nitrogen Oxide Formation
Inside a Laboratory Pulverized Coal Combustor. Submitted for
Consideration to the Eighteenth Symposium (International) on Combustion,
The Combustion Institute, Pittsburgh, PA, August, 1980.
7.	Barsin, J. A. Pulverized Coal Firing N0X Control. In: Proceedings of
the Second N0„ Control Technology Seminar, Electric Power Research
Institute, 1978.
Vatsky, J. Experience in Reducing N0^ Emissions in Operating Steam
Generators. In: Proceedings of the Second NO.. Control Technology
Seminar, Electric Power Research Institute, July, 1979.
Zallen, D. M., R. Gershman, R., M. P. Heap and W. H. Nurick. The
Generalization of Low Emission Coal Burner Technology. In: Proceedings
of the Third Stationary Source Combustion Symposium, Vol. II,
San Francisco, CA, 1979. EPA-600/7-79-0506. p. 73.
167
-------
10. Johnson, S. A., P. L. Cioffi, T. M. Sommer and M. W. McElroy.
The Primary Combustion Furnace System, An Advanced Low NO Concept
for Pulverized Coal Combustion. In: Proceedings of the Sscond
NO Control Technology Seminar, Electric Power Research Institute,
11.	Chen, S. L., et al. The Influence of Fuel Composition of the Formation
and Control of N0X in Pulverized Coal Flames. Western States
Section, The Combustion Institute, Spring Meeting, Irvine, CA,
April 1980.
12.	Axworthy, A. Pyrolysis Products of Various Coals under Inert
Conditions. EPA Fundamental Combustion Research Meeting,
Newport Beach, CA, January 1980.
13.	Seeker, W. R., G. S. Samuelsen, M. P. Heap and J. D. Trolinger.
The Thermal Decomposition of Pulverized Coal Particles. Submitted
for Consideration to the Eighteenth Symposium (International) on
Combustion, Pittsburgh, PA, August, 1980.
168
-------
, COOLING WATER .
Removable coils)
staging air
, COOLING WATER .
(removable coils)
INSULBL0CK
INSULATING
REFRACTORY
HIGH TEMPERATURE
REFRACTORY
METERS
-1.0
0.5
1.17
k 1.5
-2.0
2.5
COAL & TRANSPORT
COMBUSTION
AIR
COOLING
WATER
AXIAL
AIR
SWIRL AIR
AXIAL
AIR
SWIRL AIR
i
PREMIXED
COAL & TRANSPORT AIR
COOLING WATER RADIAL DIFFUSION
COAL & TRANSPORT AIR
COOLING WATER AXIAL DIFFUSION
Figure 1. Schematic of Down Fired Tunnel Furnace and Burner Systems.
169
-------
woo
1300
1200
UOO
1000
900
300
700
600
500
400
300
a..
900
800
700
600
500
400
sod
700
600
500
400
300
200
0.1
A	CP s voQ
r	3 <3 cP
0 ~

¦¦
i—[——i—i—i—i—i—i—i—i—
PREFIXED (SI O2 IN STACK)
c
D
1 1 1 1 1 1 1 1 1
1—i—1—1—1—1—r
=US[Q
a
RAOlAC DIFFUSION (51 Oj IN STACK)
<0
o
OF
0
A	<3	_
~ o
o
(7
\ t 11 1 1 1 1 . i
1 1 1 1 1 1 1 1
AXIAL DIFFUSION (51 IN STACK) H
A ^ ^	„ O17
O	
-------
- PREMIXED
RADIAL DIFFUSION
AXIAL DIFFUSION "~
0.2
0.1
0.1
0.1
0.2
0.2
FRACTION OF FUEL N CONVERTED TO HCH AT 1373°K
Figure 3. Fractional Conversion of Fuel Nitrogen to NO as a Function of
HCN Formed During Inert Pyrolysia.
-------
I—ft
PSEMIXED - OPEN SYMBOLS
RADIAL DIFF. - SOLID SYMBOLS
STAGED 8 0.86 n
SR - 1.25
RADIAL DIFFUSION
STAGED 8 0.86 m
SR - 1.25
600
600
500 £
500
£ 100
C*

o
300 g
K>
300
o
200
200
0.9
0.8
0.5
0.7
0.6
SR2
SR, OF MINIMUM NO
Figure 4. Staged NO Emissions: a) Effect of Coal Type, b) Effect of Mixing
Characteristics.
-------
600
500
100
300
200
100
0
50
10
30
20
10
0
50
HQ
30
20
10
0
ss-
•&>
J	_1
BEULAH
PREMIXED
STACK NO (OZ 02)
STAGED a 1.17m
SR - 1.25
ONO
A uru [ 9 1»17m
V I(AS MEASURED)
PARTICLE TEMPERATURE (a 0.53m)
57-CO2 (a 1.17m)
CO (a 1,17m)
t	r
1	r
1
j	1
i	P—i
O c
A n
~ H
a 1,17m
0,4 0.5 0.6 0,7
sr2
0,8 0,9 1.0
Figure 5. First Stage Exit Conditions.
173
-------
-4h
s* a
•—«
ae
^ 10 -

O
X
T	T
PR0UXE0
SR - 0.5
X
X
X
Q
V O
1.0 1.1 1.2 1.3 1.* 1.5
X MlltOGU IN FUEL (IMF)
1.6
1.7
-fh
PREFIXED
SR - 0.5
20	30	40	50	00
I OF INITIAL NITROGEN RETAINED 1H ASIM CliAR
(a)
(b)
Figure 6. Nitrogen Remaining in Coal Char Under Staged Conditions.
-------
SR - 0.75
cT1
8
, tooh
>-
ac
a
g 300
*
w
U 0?9 L0 171 O 17! "174 175 176 177 ITS 179
X NITROGEN IN FUEL (DAF)
(a)
rfh
2-500
100
w
300
200
100
0 L/f.
SR - 0.75
PREFIXED
I
I
J,
I
I
o -
1000 1100 1200 1300
300 400 500 600 700 800 900
UNSTA6ED NO, PPM CORY, OX 02)
(b)
Figure 7. TFN Content as a Function of Coal Nitrogen Content and
Unstaged NO Emissions.
175
-------
JUO
ttWWI BlTlitUNOUS
alabaaa biiurinous
5 to
HC*
° 10-
211 —
WESItW SUBBMUMNOOS
MUHMCIIE
80-
Ui-
o SU -
20-
0.8
0.5
0.6
0.1
0.6
0.5
0.6
o.;
o.s
0.7
0.8
0.7
Figure 8. The Effect of Coal Composition on the Partition of Gas
Phase Nitrogen Specie.
-------
100
¦0
M
40
20
500
400
300
200
too
e
400
100
200
100
SQO
400
300
100
100
0
KM MIIMACUB SK » «.i
UTAH IIJWIMOUS u - 0.U
MOUTH DAKOTA LIGMtTt SA - 0.(4
o-
' »//'
i.n o.si
i.w
i.i?
DttflMCE	¦
1.11
	I—
0.91
1
X
x
I.S5
AmOIIMAH MSIDCKCf TIME, S
Decay of Nitrogen Specie and Fixed Carbon with Time under
L Rich Conditions.
-------
15
14
12
14
320
300
280
260
240
220
200
180
160
WO
120
100
80
100
30
60
40
20
140
120
100
80
60
W
20
0
STAGED a 1.17 m
SRi « 0.8
SR - 1.2S
¦"ita	ut lt$ ff rrfe
,	, DISTANCE FROM BURNER, m
Vf I ¦	I		 1 ,,	ft I,
^ 0728	0773 L25
APPROXIMATE TIME.	S
Figure 10. The ImpacC of First Stage BeaC Removal.
178
-------
TABLE I. FUEL ANALYSIS
fuel Syabol
u
o
Q
o
o
A
o
O
O

0

a
Fuel Source
Hazel Ion
PA.
Upper Cliff
Alabaaa
Rosa
Alabaaa
Black Creek
Alabaaa
W. ICY.
U. VA.
£lkay
V. VA.
Gauley
Eagle
H. VA.
Price
Utah
Price
Utah
Utah
(Coal)
Utah
(Char)
Cadiz
Ohio
Proxiaate Analysis
(X as received)













Moisture
5.13
3.00
8.02
2.2S
5.43
7.57
0.60
3.72
6.39
7.41
4.20
2.70
4.29
Ash
5.74
9.49
6.79
4.45
7.53
12.05
11.61
26.89
7.40
8.83
9.62
16.83
16.05
Volatile Hatter
4.39
20.44
21.81
28.28
37.79
29.12
33.35
26.65
38.89
38.84
42.97
10.29
34.87
Fixed Carbon
84.74
67.07
63.38
65.02
49.25
51.26
54.44
42.74
47.32
44.92
43.21
70.18
42.79
Ultiaate Analysis
it Dry)













C
88.4S
79.32
81.23
82.12
71.58
74.68
. 73.96
59.43
73.17
72.24
69.36
75.39
62.08
H
2.14
4.47
4.73
5.21
5.43
4.96
4.93
4.11
5.55
5.75
5.32
1.28
4.44
N
0.79
1.47
1.54
1.79
1.55
1.38
1.39
1.35
1.54
1.55
1.50
1.22
1.07
S
0.47
1.30
1.04
0.76
3.21
0.86
2.47
0.B4
0.66
0.76
1.04
1.12
7.40
Ash
6.OS
9.78
7.38
4.56
7.96
13.04
11.69
27.94
7.90
9.54
10.05
17.30
18.86
0
2.10
3.66
4.08
5.56
10.27
5.08
5.56
6.33
11.18
10.16
12.73
3.69
6.15
Heatinq Value
(Btu/lb. Met)
13.124
13,254
13.394
14.284
12.082
12.228
13.115
10.110
12.340
11.877
11.718
11.185
11.038
CLASSIFICATION
(ASTH 0388)
Anthracite
Mediua
Volatile
Bituainous
Mediiai
Volatile
Bituainous
Hediua
Volatile
Bituainous
High
Volatile
A Bitu-
ainous
High High High High
Volatile Volatile Volatile Volatile
A Bitu- A Bitu- A Bitu- 6 Bitu-
ainous ainous a)nous ainous
High
Volatile
B Bitu-
ainous
High
Volatile
B Bitu-
ainous

High
Volatile
B Bitu-
ainous
-------
TABLE I. FUEL ANALYSIS (Continued)
syboi	C7 0
20.49
25.23
34.96
33.10
34.63
36.36
9.07

Ash
17.96
24.06
20.72
9.56
11.26
8.62
10.28
7.50
7.12
4.97
4.61
3.3a

Volatile Hatter
34.85
35.94
31.66
30.82
31.26
33.24
35.31
28.85
28.65
27.02
28.48
48.79

Fixed Carbon
39.33
34.76
36.60
38.33
34.88
37.65
29.18
28.69
31.13
33.38
30.55
38.76

lilt 1mte Analysis
It bryj













C
63.06
55.99
58.71
67.52
65.54
67.88
60.99
64.61
65.29
66.15
64.99
66.25
78.06
H
4.65
4.71
4.21
4.36
4.15
4.65
4.49
4.17
3.96
4.20
4.04
5.01
4.86
N
1.40
1.23
1.30
1.38
0.95
0.99
1.13
0.83
0.99
0.96
1.00
0.65
1.25
S
0.B1
1.03
0.93
0.63
0.79
1.07
1.02
1.52
1.14
0.37
0.42
0.28
o.ao
Ash
19.47
25.39
23.29
12.17
14.57
10.84
13.74
11.53
10.64
7.60
7.25
3.72
6.46
0
10.61
11.65
11.56
13.94
14.00
14.57
18.63
17.34
17.98
20.72
22.30
24.09
8.57
Heat 1 imi Value
(Btu/lb, Met)
10.391
9.425
9,344
9,169
8.603
9.229
8.131
6.446
7.245
7.254
6.995
10,051
13,657
(dry)
CLASSIFICATION
(ASTN D388)
High
Volatile C
Bituainous
High
Volatile C
BituRinous
High Sub-
Volatile C Bituainous
Bituainous B
Sub-
Bitualnous
B
Sub-
Bituainous
8
Sub-
Bitualnous
C
Lignite
A
Lignite
A
Lignite
A
Lignite
A


-------
SYSTEM APPLICATIONS OF CATALYTIC COMBUSTION
By:
J. P. Kesselring, W. V. Krill, S. J. Anderson,
and M. J. Friedman
Acurex Corporation
485 Clyde Avenue
Mountain View, California 94042
181
-------
ABSTRACT
The development of catalytic combustion systems is continuing toward
the prototype demonstration phase. Improved catalyst materials have shown
higher maximum throughput capability and uniform axial temperature
profiles. Special auxiliary components required for fuel injection,
ignition, and catalyst temperature measurement have been developed and
incorporated into system concept designs. The three combustor concepts
developed to the system integration phase include a small gas turbine
combustor, a watertube boiler concept, and a firetube boiler burner.
The model gas turbine combustor shows continued promise for low NO^
emissions with gaseous and distillate fuels. Greatest development
difficulties are associated with introduction of the premixed fuel/air
mixture and its interaction with catalyst lightoff systems. An integrated
system has been developed, including a multiple nozzle, atomizing injector
and an opposed jet lightoff burner. Testing of the concept is nearing
completion to show its transient and steady-state capabilities.
The watertube boiler concept uses direct radiative transfer to
watertubes in the combustion region. Structural problems of the radiative
zone are currently being addressed, and final integration of the concept
will follow. Thermal emissions are typically less than 2 ppm.
The firetube boiler burner also uses radiative heat transfer from a
fiber matrix burner to the wall of the firetube. The matrix burner operates
at a surface temperature below 1644K at low excess air levels to control the
formation of thermal N0X*	material screening tests have been
conducted, and a mockup burner test is in preparation. Continuing program
work will focus on prototype development of the watertube and firetube
boiler concepts.
182
-------
INTRODUCTION
The advancement of catalytic combustor concepts toward the system
demonstration phase is continuing. Through a series of contracts with the
Environmental Protection Agency (EPA), Acurex Corporation has continued the
development of several system concepts. These concepts are being further
developed under the current EPA contract 68-02-3122: (1) a model gas
turbine combustor, (2) a watertube boiler system, and (3) a firetube boiler
burner. This paper presents a brief history of each concept and progress
since the Third Stationary Source Combustion Symposium (Reference 1).
The development of each of the three combustor types has been
supported by other activities of the program. Catalyst development has
continued to improve performance and operability of each of the combustors.
The separate development of auxiliary systems has allowed fuel injection,
lightoff, and temperature measurement problems to be solved. Data analysis
and fundamental studies have provided greater ability to predict reactor
performance. Finally, technology transfer through report publication,
technical panel meetings, and workshops has allowed valuable interchange
with the user industries. This multilevel approach is continuing, focusing
on the prototype demonstration of watertube and firetube boiler systems.
183
-------
CATALYST DEVELOPMENT
Catalytic combustors typically use ceramic or metallic honeycomb
monolithic supports with many small diameter cells. Premixed fuel and air
flow through the cells, where a catalyst coating on the interior surface
promotes conversion of fuel and air to combustion products. Near the cell
entrance, where most of the gas is at low temperature, gas-phase chemical
reactions are unimportant. In this region, catalytic wall chemical
reactions control heat release. Further down the channel, where wall
reactions have preheated the gas to higih temperature, gas-phase reactions
become active. In this region, fuel is rapidly consumed by a "flame type"
phenomenon that controls the amount of unburned hydrocarbon emissions that
escape from the system.
The high throughput performance of graded cell catalysts had
previously been documented (Reference 2), but a direct comparison between
identical graded cell and single cell catalyst configurations had not been
conducted. To compare the performance of the two systems, a series of
comparative tests were run. Two catalyst configurations were supplied by
UOP, Inc., on DuPont Torvex alumina honeycomb substrates. The graded cell
system consisted of three 2.54-cm segments with cell sizes of 6.4 mm,
4.8 mm, and 3.2 nm. The single cell system used three 2.54-cm segments with
a 3.2-mn cell size. Both systems were washcoated and catalyzed by UOP with
a proprietary catalyst.
Pretest analysis of the single cell model indicated a surface area
2
and dispersion of 4.88 m /g and 1.22 micromoles H./g, respectively,
2
while the graded cell indicated 5.36 m /g and 17.63 micromoles/g
catalyst. Although initially the graded cell had a much higher dispersion
than the single cell, following 5 hours of aging, dispersion for both
configurations should be essentially the same. This was verified by
184
-------
posttest analyses that indicated greatly reduced dispersions of 0.39 and 1.3
for the single and graded cell monoliths, respectively.
A comparison of the graded cell and single cell catalyst
configurations, characterizations, and combustion performance on natural gas
fuel is given in Table I. Both systems were operated in an identical manner
during testing, which included 5 hours of steady-state operation under lean
conditions at 1560K temperature, followed by maximum throughput tests. As a
result of this testing, the following was noted:
•	At a standard heat release rate of 105.5 MJ/hr, emissions from
the graded cell and the single cell monoliths were approximately
the same
•	Volumetric heat release at maximum throughout for the graded cell
was almost three times higher than for the single cell system
(7.4 x 10^ J/hr-Pa-m^ versus 2.6 x 10^ J/hr-Pa-m^)
•	Under fuel-rich conditions, both reactor types maintained
combustion at preheat temperatures as low as 320K. However,
hydrocarbon emissions from the single cell configuration were an
order of magnitude higher than for the graded cell catalyst.
•	At the standard heat release rate of 105.5 MJ/hr, the graded cell
catalyst had a much more uniform axial temperature profile than
the single cell system.
The HET code (Reference 3) was used to predict surface temperature
profiles in the graded cell and single cell reactors, and the results are
shown in Figure 1. Excellent agreement betyeen code prediction and surface
temperature as measured by in-depth thermocouples is noted. The more
uniform temperature profile for the graded cell system is important, since
to avoid poisoning effects, the catalyst should operate at as high a
temperature as possible over its entire length. As shown in Figure 1, the
single cell reactor has a high, relatively uniform temperature profile. In
addition to running the risk of poisoning, the single cell reactor is also
operating in an unstable mode. A slight change in inlet conditions could
move the reaction zone downstream in the combustor, eventually resulting in
blowout.
185
-------
These comparative tests verify that combustors with high volumetric
heat release rates can best be achieved with catalysts in the graded cell
configuration with no penalties in combustible or N0x emissions. In
addition, the high uniform temperature of the catalyst walls throughout the
length of the graded cell model may have important implications in avoiding
catalyst poisoning or liquid fuel deposition on the walls.
186
-------
AUXILIARY COMPONENTS
The auxiliary components developed under this program include fuel
injection, ignition, and temperature measurement subsystems. Due to unique
features of the catalytic combustor, special ignition techniques and
premixed fuel and air are required. For catalyst lightoff, an ignitor must
heat the catalyst to over 700K in a controlled manner. Once lightoff is
achieved, the ignitor must be easily shut off. To effectively use all
catalyst material and prevent excessive temperatures due to locally enriched
reactant mixtures, air and fuel should be evenly distributed prior to
entering the catalyst. Fuel injectors must provide thorough and rapid
mixing to minimize mixing length, prevent recirculation that may lead to
flameholding, and promote rapid prevaporization of liquid fuels. Finally,
catalyst temperature must be carefully monitored during startup and
steady-state operation to maintain catalyst materials within allowable
operating limits. Correspondingly, a reliable system must be developed to
continuously measure catalyst surface temperature.
Fuel Injection Concepts
Several fuel injection techniques exist that can provide rapid
fuel/air mixing and fine liquid fuel atomization. Aerodynamic fuel
injectors for both gaseous and liquid fuels and atomizing spray nozzles for
liquid fuels were selected as candidate injectors. While the aerodynamic
fuel injectors showed excellent performance with gaseous fuels, performance
with distillate fuels has resulted in operational problems that require
great care to ensure uniform injection in all aerodynamic zones under
varying fuel flowrates. To avoid these problems, atomizing spray nozzle
injectors were developed for liquid fuels.
A multiple point atomizing spray injection system, shown in Figure 2,
has been developed for the gas turbine combustor system. The three
187
-------
air-atomizing nozzles produce mean drop sizes of approximately 60 to
100 micrometers with distillate fuel. The spray nozzles are mounted on
extensions approximately 5 cm away from the injector plate. The injector
plate is mounted between flanges of the test facility. Vaporization or
mixing are nearly complete in the allowed mixing length without autoignition
difficulties.
During initial testing of the multiple injectors, the nozzles' wide
spray angle caused fuel to impinge directly on the wall of the combustor
can. Some thermal NO^ was produced, apparently as a result of upstream
burning of the distillate fuel on the wall. This was confirmed by cold
water model tests using a clear acrylic combustor can. Fuel contact with
the wall was suppressed by reducing the nozzle spray angle from 77 to
15°. Subsequent combustion testing with a low spray angle was quite
successful. The injection system provided a uniform fuel mixture without
upstream burning.
In sunmary, aerodynamic fuel injectors provide rapid and complete
fuel-air mixing of gaseous fuels over a wide range of flowrates. If system
pressure drop is not important, a single injector is sufficient. To
minimize pressure drop, multiple point injection may be necessary.
Aerodynamic injectors do not adequately atomize liquid fuel, however.
Consequently, spray nozzles are recommended for these applications. In
addition, for large systems, multiple point injection should be employed to
reduce mixing length while maintaining uniform fuel-air distribution.
Combustor Ignition Concepts
Catalyst lightoff occurs when a minimum catalyst temperature is
achieved. The lightoff temperature varies primarily with catalyst and fuel
type and secondarily with stoichiometry and mass throughput. In general,
temperatures in excess of 700K (800°F) are necessary for lightoff. For
example, noble metal catalysts such as platinum and palladium require 617 to
784K (650° to 950°F) lightoff temperatures, while metal oxide catalysts
such as nickel and cobalt require 866 to 1367K (1100° to 2000°F)
temperatures. In early EPA testing, lightoff temperature was achieved using
electric air preheaters. However, air preheat temperatures at this level
188
-------
are not available in boiler and gas turbine systems. Consequently)
alternate techniques for lightoff were investigated.
The preferred technique is the opposed jet system where a jet of
premixed fuel and air (approximately 1 percent of the total mainstream mass
flow) is injected counterflow to the main stream of reactants, producing a
stagnation point flow condition. An upstream spark plug ignites the jet and
a bow flame front stabilizes in the jet region. By consuming the mainstream
reactants, the opposed jet acts as an aerodynamic flameholder, providing a
preheated fuel-air mixture to the catalyst located downstream. When the
desired ignition temperature is achieved, the jet fuel and air are turned
off, thereby extinguishing the flame. The opposed jet ignitor was evaluated
using both gaseous and prevaporized liquid fuel in the mainstream.
Optimum performance of the jet was identified initially without a
catalyst by varying mainstream and jet stoichiometry. Catalyst lightoff
capability was then evaluated with a catalyst. Although tests were
conducted mainly with natural gas in the jet and mainstream, similar results
are achievable for other gases with similar heating values (e.g., propane).
Opposed jet stability depended primarily on the relative magnitude of gas
momentum in the mainstream duct and the jet. Based on visual observation,
the most stable jet operating condition occurred at a jet-to-mainstream
velocity ratio of 7.4 and a jet-to-mainstream mass flowrate ratio of 0.01 to
0.02. Preheat temperatures delivered by the opposed jet show small
variations with jet stoichiometry and large variations with mainstream
stoichiometry. At a nominal 105 MJ/hr (100,000 Btu/hr) heat release rate
with natural gas under lean mainstream conditions, the jet operated at
stoichiometrics between 5.6 and 17.0 percent theoretical air. Exit
temperature varied from 28 to 56K (50° to 100°F) with changes in jet
stoichiometry. Smaller temperature variations occurred at decreasing
mainstream stoichiometries.
The jet operated effectively over a range of mainstream
stoichiometries from 180 to 270 percent theoretical air. This temperature
range appears suitable for ignition of both noble metal and metal oxide
catalysts. Figure 3 plots exit temperature as a function of mainstream
stoichiometry for 10.9 percent theoretical air in the jet. The opposed jet
demonstrated excellent catalyst lightoff capability under lean conditions
189
-------
with natural gas. A comparison of an opposed jet lightoff sequence with a
typical lightoff using preheated air and fuel is given in Figure 4. As
shown, the preheated air technique with a noble metal catalyst required a
bed temperature near 750K (900°F) for startup under typical flowrate
conditions with the lightoff rate controlled by the stoichiometry of the
reactant stream (two are shown). In contrast, the opposed jet can operate
with a reactant stream at ambient temperature while the heating rate is
controlled by jet and mainstream stoichiometry and the thermal mass of the
particular combustion system. For a specific application, the desired
heating rate is dictated by both catalyst material and other system startup
constraints.
Catalyst Temperature Measurement Techniques
To maintain efficient, stable combustion and not exceed its material
temperature limitations, a catalyst must be operated within a relatively
narrow range of steady-state temperatures. Thus, measurement of the
catalyst surface temperature is a key parameter in system control. The
instrument used for temperature measurement must be accurate within 28 to
56K (50° to 100°F), maintain adequate life at low cost, and respond
quickly to temperature transients. Furthermore, it should be simple to
install and readily accessible for easy maintenance.
Techniques used to date include in-depth thermocouples attached
directly to the substrate wall, thermocouple probes that measure gas-phase
temperature, and optical pyrometers.
Based on the testing done with each of these techniques, it was found
that all systems require further development for system applications.
In-depth thermocouples are too fragile and may adversely affect the
structural integrity of the honeycomb. Pyrometers can accurately measure
catalyst temperature (within 30K) if surface emissivity is known as a
function of temperature. Because the sensor is external to the combustion
zone, it does not suffer reliability problems due to direct exposure to high
temperatures (as do thermocouple probes). Temperature measurement by
thermocouple probes located downstream of the catalyst bed depend in a
complicated manner on gas velocity and system geometry. Consequently, they
are not an accurate method of temperature measurement. In contrast, at the
190
-------
gas velocities investigated, interbed thermocouple probes do appear to
accurately measure catalyst temperature.
The integration of catalyst and auxiliary system components into the
three combustion concepts has resulted in further system testing, as
described in the following section.
191
-------
MODEL GAS TURBINE COMBUSTOR
The demand for low NO^ gas turbine combustors is increasing.
Potential applications include utility power systems, pipeline pumping,
transportation vehicles, and future combined cycle processes. As a result,
government and private development activities have greatly increased. Among
selective catalytic reduction, thermal DeNO^, and NO^ scrubbing,
catalytic combustion may prove to be the most cost-effective NO^ control
technique.
Under early work with fuel-lean combustors, Acurex developed a model
gas turbine can combustor (Reference 4). The combustor consisted of a
stainless steel can, multiple cone fuel injector (based on a concept
developed at NASA, Reference 5), and the catalyst element. The concept was
tested under steady-state operatng conditions in both Acurex and Pratt and
Whitney test facilities. Fuels used were natural gas, propane, and No. 2
oil at test pressures from 0.101 to 0.808 MPa (1 to 8 atmospheres). Very
low thermal NO^ emissions (less than 5 ppm) were achieved throughout the
test series. The combustor was also tested with nitrogen-doped fuels to
investigate fuel nitrogen conversion to NO^. Lean conversion rates were
generally high (greater than 50 percent) at all test conditions.
The greatest system operational problem was injection of fuel to
obtain a well-mixed, prevaporized reactant stream at the catalyst. The
injector design resulted in flameholding under some operating conditions
that would be encountered in machine applications. Therefore, the range of
operability required improvement.
A second distillate fuel injector was designed and constructed, as
described previously. The injector incorporated three air-assist atomizing
spray nozzles spraying through a perforated air distribution plate. Cold
flow and combustion tests were run on the configuration to verify the
192
-------
injector performance. Flameholding caused by the fuel introduction to the
air stream was alleviated following optimization of the spray pattern into
the combustor.
To further advance the combustor design, an opposed jet lightoff
system was incorporated to simulate typical turbine ignition and startup
requirements. The advanced concept is shown in Figure 5. In application,
the jet would be ignited during low air flow conditions and produce energy
to accelerate the machine and preheat the catalyst to its lightoff
temperature. As the machine approaches a loaded condition, the jet is
extinguished by depletion of its flow, and combustion occurs totally in the
catalyst. To date, the overall concept has been tested to simulate startup
without the catalyst element to investigate operating characteristics of the
fuel injector and opposed jet.
Tests were conducted with premixed air and natural gas in the opposed
jet and atomized diesel fuel in the mainstream air. The lightoff
temperature that could be achieved was determined by varying the amount of
diesel flow and hence the mixture adiabatic flame temperature. Tests were
run at atmospheric pressure with a mainstream reference velocity of
4.6 m/sec (15 ft/sec) and a preheat temperature of 547K (525°F).
The test data is shown in Figure 6. As the mixture stoichiometry
(including jet and mainstream fuels) is decreased toward stoichiometric, the
lightoff temperature increases. The increasing adiabatic flame temperature
of the mixture is shown for comparison. The actual temperature achieved at
a given stoichiometry is dependent on the degree of mainstream fuel
consumed. Less fuel is consumed near stoichiometric conditions due to
increasing difficulty of burning liquid fuel droplets at 547K preheat.
Increasing the preheat would result in greater fuel consumption and higher
lightoff temperatures at a given mixture stoichiometry.
At 400 percent theoretical air, the combustor is operating with only
jet fuel at a temperature of 1005 K (1350°F). Above 400 percent, flame
cannot be stabilized in the combustor. At 100 to 150 percent theoretical
air, the majority of the fuel is diesel. Below 150 percent theoretical air
at the preheat and velocity conditions described, the stabilized flame could
not be extinguished upon depletion of the jet flow.
193
-------
The wide range of jet operability should allow smooth turbine startup
in applications. At low air flow, the jet fuel would be introduced and
ignited. During acceleration (and increasing compressor discharge
temperature), mainstream fuel would be increased to the catalyst lightoff
temperature and the jet extinguished. Final acceleration and loading could
then be achieved catalytically by further fuel increases. Testing of the
startup and steady-state operation of the combustor with a catalyst is in
progress.
194
-------
WATERTUBE BOILER SYSTEM
The effect of control on boiler systems is even more
significant than for gas turbines. Whereas stationary gas turbine
combustors contribute approximately 3 percent of the total U.S. NO^
inventory, packaged and utility boilers are estimated to contribute
20 percent and 48 percent, respectively (Reference 6). Clearly, controls
developed for watertube and firetube boilers are an important factor in
reducing this
country's NO^ emissions.
A catalytic watertube boiler system has been devised and extensively
tested. Combustion occurs on the exterior surface of catalyzed cylinders in
crossflow to the premixed reactant stream. Energy is transferred to
watertubes by radiation from the combustion surfaces.
The concept is shown in Figure 7. A stoichiometric fuel/air mixture
is fed to the radiative section which contains a close-packed array of
catalyst elements and watertubes. The mixture is partially combusted by the
catalyst which is kept at a low surface temperature by radiation heat loss
to the watertubes. The combustion products and remaining unburned fuel and
air are then passed to a downstream catalytic adiabatic combustor to
complete combustion reactions. A final convective section extracts energy
from the fully combusted gases. Both catalyst sections operate well below
the maximum use temperature of the catalyst supports — the radiative
section by radiative cooling and the adiabatic section by dilution of the
fuel/air mixture with exhaust products from the radiative section.
The radiative section has been constructed and tested independently
of the downstream adiabatic combustor and convective sections. In early
testing, the staggered tube arrangement (separate catalyst tubes and
watertubes) was used. Later tests were conducted with a concentric tube
design. The results for the staggered tube configuration were reported in
195
-------
Reference 2. The radiative section was run with natural gas at varying
stoichiometry (40 to 220 percent theoretical air), heat release rate (1.09 x
8	8
10 to 3.44 x 10 J/hr), and preheat to establish its combustion
efficiency, thermal and fuel NO^ emissions, and heat transfer
characteristics at varying conditions. The catalyst was a platinum
preparation on alumina tubes.
At stoichiometric conditions, the combustion efficiency of the
staggered tube combustor was approximately 37 percent. This result was
slightly lower than the anticipated level of 50 percent necessary to avoid
flame temperatures above the temperature capability of the adiabatic
section. The thermal NDx emissions were very low, however, never
exceeding 2 ppm at these conditions. To evaluate the conversion of fuel
nitrogen to	natural gas was doped with ammonia (NH^)• A minimum
conversion level of 20 percent was measured at approximately 55 to
60 percent theoretical air.
These early combustion test results suggested several design
modifications for improvement of the radiative section. Closer spacing of
the catalyst tubes would improve the combustion efficiency. It would also
be desirable to isolate the cold surfaces of the watertubes from the
combustion zone to avoid fouling when firing with fuel oils. To meet these
requirements, a concentric tube design was completed where the catalyst tube
surrounds the water tube with an air gap between the two. Therefore,
combustion energy generated on the catalyst tube outside surface would be
transferred by conduction to its inner surface and by radiation across the
air gap to the watertube.
The concentric configuration was first tested under conditions
similar to those for the staggered arrangement with a platinum catalyst on
alumina tubes. Combustion efficiency was markedly improved, reaching a high
of 93 percent fuel conversion at stoichiometric conditions as shown in
Figure 8. Thermal MO emissions were again less than 2 ppm, and fuel
X
nitrogen conversion was below 30 percent under fuel-rich conditions. Tests
were not performed with fuel oil to verify that capability, however.
-Despite the improved combustion performance of the concentric
watertube design, structural problems were encountered. The configuration
develops a substantial temperature gradient across the walls of the ceramic
196
-------
tubes. The temperature difference results in nonuniform radial expansion
and, ultimately, fracturing of the material. The stresses that would be
encountered in application were calculated as 198 x 10 Pa (28,700 psia)
to 257 x 10^ Pa (37,300 psia), clearly greater than the 190 x 10^
tensile strength of alumina tubes at 1367K (2000°F).
Only one other ceramic can be made with adequate strength and is
appropriate for combustion applications. Silicon carbide not only has a
high strength of over 400 x 10^ Pa at the above conditions, but it has a
higher thermal conductivity than alumina and therefore results in lower
temperature gradients and developed stresses. As a result, tubes of silicon
carbide were purchased from the Norton Company in the proper dimensions and
coated with a chromia (C^Oj) and platinum catalyst preparation for an
additional test series. Testing of the concept is in progress.
197
-------
FIRETUBE BOILER BURNER
A large percentage (approximately 50 to 65 percent) of the commercial
size boiler market is composed of firetube boilers. These smaller units
account for over 30,000 gas-fired boilers and an equal number of distillate
oil-fired boilers in the 3.2 x 10^ J/hr (3 x 10^ Btu/hr) size alone
(Reference 7). In fact, of all commercial and industrial size boilers in
the field, the majority are firetubes. These applications are principally
gas- and distillate oil-fired.
Thus, firetube boilers represent another area for significant NO^
reduction. A catalytic burner would be especially attractive if it were
retrofitable into the large number of boilers already installed or at least
presented only minor changes to current designs.
A number of catalytic burner designs were conceptualized and reviewed
with major firetube boiler manufacturers. The simplest and most acceptable
to the manufacturing community utilizes a fiber matrix material that is
molded into a cylindrical surface burner for insertion into the firetube.
The burner concept is shown schematically in Figure 9. A fuel/air
mixture is passed into the central region of the cylindrical fiber matrix.
The cylinder construction is best described by Figure 10. The matrix is
approximately 2.5 cm (2 inches) in thickness, is closed with similar
material at the downstream flow end, and is sealed at the upstream end by a
combustion chamber flange. The matrix material is rigid, is supported at
the upstream end, and has internal flow passages as shown in Figure 10,
maintaining concentricity with the boiler firetube.
In the described configuration, premixed fuel and air are forced to
flow into an annulus between the fiber matrix and an internal cylinder. The
fuel may be gaseous or vaporized liquid. The mixture then flows through the
196
-------
matrix and is ignited at the outer cylinder surface to start the burner, k
heterogeneous surface reaction zone is established on the cylinder outer
surface, within a depth of only a few millimeters of the matrix thickness.
The majority of the matrix thickness remains at the temperature of the
incoming reactants and promotes no additional reaction.
Heat is transferred directly from the matrix outer surface to the
firetube as shown in Figure 9. The transfer is primarily radiative although
convective transfer also contributes. This heat transfer limits the matrix
surface temperature to less than 1644K (2500°F), the temperature
limitation of the material, although the adiabatic flame temperature of the
reactive mixture exceeds 2200K. The heat is transferred through the metal
firetube wall to the boiler water to heat the water or raise steam. Flue
gases are forced to the end of the firetube and out to
the additional firetube passes.
Catalytic fiber matrix materials have demonstrated low NO^
emissions and high combustion efficiency with gaseous fuels (Reference 8).
However, the optimum matrix material, appropriate heat transfer
configuration, and ability to operate on distillate fuel required additional
study. A series of tests on fiber matrix disks was conducted to evaluate
these parameters.
Figure 11 shows the matrix surface temperature of the combustion zone
as a function of theoretical air for three values of face velocity. Lower
face velocities result in lower surface temperatures at low excess air
levels (100 to 130 percent theoretical air) where firetube boilers normally
operate. This low surface temperature is advantageous in reducing NO^
emissions as shown in Figure 12. Emissions of less than 10 ppm at
115 percent theoretical air are possible. Tests also indicate high
combustion efficiency (low CO and HC) can be achieved simultaneously with
low NO emissions,
x
The low surface temperatures of Figure 11 also demonstrate the
effective radiative heat transfer away from the combustion zone. The
indicated temperatures of 1478K to 1644K (2200°F to 2500°F) at 100
percent theoretical air are within the use temperature capabilities of the
alumina silicate fibers and far below the adiabatic flame temperature of
199
-------
2228K. From these tests, the optimum flow velocity for the selected matrix
material appears to be 10 to 15 cm/sec (1/3 to 1/2 ft/sec).
The catalytic burner was also operated on diesel fuel. To achieve
stable combustion with similar results to natural gas, it was necessary to
partially vaporize the oil. This was accomplished by preheating the air
stream to 492K (425°P) prior to admitting the fuel. Current firetube
boiler designs operate without air preheating systems. Therefore, operation
of the matrix burner with fuel oil in a firetube boiler will require
additional development of an air preheating technique.
These flat disk burner tests identified the operating requirements of
the matrix material. To further develop the concept, a cylindrical burner
as in Figure 10 was constructed for testing in a simulated firetube.
Results of the tests are pending. The fiber matrix burner does appear
attractive for further development and should be applicable to many firetube
designs with only minor modifications.
200
-------
CONCLUSIONS
The development of catalytic gas turbine, watertube boiler, and
firetube boiler combustors is rapidly progressing. The potential for
near-term reduction of thermal NO^ emissions with clean fuels is promising.
An integrated model gas turbine combustor has been tested to simulate
both startup and steady state operating conditions, following refinement of
fuel injection and lightoff techniques. Only further refinement of the
concept is required for small machine applications. The concept may also be
scalable to larger combustor sizes.
The watertube boiler concept shows promise for further development
despite being a radically different design from present systems. Production
units would probably differ further from those tested. Stress analyses on
the catalyst supports determine silicon carbide as the most appropriate
ceramic in the preferred concentric tube design. Ongoing testing will
verify the material recommendations and lead to testing of the integrated
concept.
The firetube boiler burner is a retrofitable design for gas-fired
units. For distillate-fired firetubes, the fiber matrix concept requires
preheated air for acceptable operation. Detailed tests on fiber materials
have shown low NO emissions due to surface cooling with concurrent low CO
x
and hydrocarbon emissions. A cylindrical burner of the most acceptable
material is being fabricated for testing.
The development of burners for these three classes of equipment
addresses a large portion of the NO^ emissions in this country. Further
EPA program development will focus on prototype demonstration of catalytic
boiler systems.
201
-------
REFERENCES
1.	Kesselring, J. P., et al. Catalytic Combustion System Development
for Stationary Source Applications. In: Proceedings of the Third
Stationary Source Combustion Symposium; Volume IX. Advanced
Processes and Special Topics. EPA-600/7-79-050b, February 1979,
pp. 207-240.
2.	Krill, W. V. and Kesselring, J. P. The Development of Catalytic
Combustors for Stationary Source Applications. In: Proceedings:
Third Workshop on Catalytic Combustion. EPA-600/7-79-038, February
1979, pp. 239-268.
3.	Kendall, R. M., et al. An Analysis of Catalytic Combustion in
Monolithic Honeycomb Beds. In: Proceedings: Third Workshop on
Catalytic Combustion. EPA-600/7-79-038, February 1979, pp. 197-238.
4.	Kesselring, J. P., et al. Design Criteria for Stationary Source
Catalytic Combustion Systems. EPA-600/7-79-181, August 1979.
5.	Tacina, Robert R. Degree of Vaporization Using an Airblast Type
Injector for a Premixed-Prevaporized Combustor. NASA TM-7883b,
August 1978.
6.	Mason, H. B. and Waterland, L. R. Environmental Assessment of
Stationary Source N0X Combustion Modification Techniques. In:
Proceedings of the Second Stationary Source Combustion Symposium.
EPA-600/7-77-073a, July 1977.
7.	Evaluation of National Boiler Inventory. EPA-600/2-75-067, 1975.
8.	Martin, G. Blair. Prototype Surface Combustion Furnace Evaluation.
EPA-600/7-79-038, February 1979, pp. 175-196.
202
-------
1600
1500
1400
1300
1200
1100
1000
900
800
700
600
Single cell reactor
# 140,000 hr"1 data
	 140,000 hr"1 prediction
Graded cell reactor
~ 184,000 hr"1 data
—184,000 hr"1 prediction
/
Preheat
temp
-2
2	4
Bed depth (cm)
8
Figure 1. Comparison of graded cell and single cell reactor
temperature profiles with HET code predictions.
203
-------
b
D
a
a
o a
Figure 2.
Injector
plate
Propane
Injection
tube
Nozzle
support
Air
Spray nozzle
Multiple spray fuel injector.
-------
1500
1400
1300
1200
1100
1000
900
300
700
600
500
I
0
©>
Preheat temperature = 478K (400°F)
Jet stoichiometry = 10.9 percent T.A.
©
J	L
J-

_L
I
100 120 140 160 180 200 220 240 260 280 300
Main stream stoichiometry (percent T.A.)
Figure 3. Opposed jet ignitor — exit temperature versus
mainstream stoichiometry.
205
-------
2500
1500
260% TA
Typical catalyst
lightoffs with
prheated air
305% TA
2000
Opposed jet
1ightoff
1500 —
3 1000
u_
o
1000
500
500
Jet on
2
7
8
0
3
4
5
6
Time (min)
Figure 4. Lightoff time comparison.
-------

"njE^MOcoupve
SUPPORT	WIRES
C^ETZklL. A
X
Mwsl
reNPe
ezzz-
X
CAT^M-VST
BEP>
Figure 5. Model gas turbine combustor with opposed jet
startup system.
-------
3600
3400
3200
3000
2800
2600
2600
2400
2200
2000
1800
1600
1400
1200
Face velocity
O 10.5 ft/sec
A 13.4 ft/sec
~ 15.5 ft/sec
Adiabatlc
flame
temperature
Jet could
not be
extinguished
100
200	300
Mixture stolcManetry (percent theoretical air)
400
Figure 6. Gas turbine opposed jet results.
-------
Steam
drum
Refractory lining
Convectlve heat exchanger
Monolith bed
Adlabatlc
combustor
Radiative heat
transfer section
Catalyst
coated cylinder
Watertube
Figure 7. Radiative watertube boiler concept.
-------
100
N>
»—1
o
o
u
s~
tt)
CL
O
c
QJ
•r-
u
' I—
4-
4—
(V
a
o
jQ
E
O
o
90
80
70
60
50
1
1
70
80
90	100	110
Theoretical air (percent)
120
130
Figure 8. Concentric watertube boiler combustion efficiency.
-------
Flue gas
Boiler water
Fiber matrix
Firetube surface
Figure 9. Firetube boiler catalytic burner concept.
211
-------
Fire matrix
radiative
burner
Firetube
combustion
chamber
Flow annul us
Fuel/air
mixture
Figure 10.
Catalytic firetube boiler burner.
-------
1600
1400
•o
!_
OJ
a.
E
0)
1200
1000
800
2500
2400
2300
2200
2100
2000
1900
1800
1700
1600
1500
1400
1300
1200
1100
1000
900
Face velocity
O 0.61 m/sec
O 0.30 m/sec
A 0.15 m/sec
t
J	I I I	i I t
J	L
100 110 120 130 140 150 160 170 180 190 200 210
Theiretlcal air (%)
Figure 11. Fiber matrix operating temperature.
213
-------
140
130
120
110
100
90
80
70
60
50
40
30
20
10
0
Face velocity
O 0.61 m/sec
O 0-30 m/sec
A 0.15 m/sec
110 120 130 140 150 160 170 180 190
Theoretical air {%)
Figure 12. Fiber matrix MO emissions.
x
-------
TABLE I. COMPARISON OF SINGLE CELL AND GRADED CELL CATALYST PERFORMANCE
Parameter
Graded Cell
Catalyst
Single Cell
Catalyst
Catalyst
manufacturer
UOP
UOP
Cell geometry
(3) 25.4-tnm long
segments of 6.4,
4.8, and 3.2-mm
cell size
(3) 25.4-mm long
segments of 3.2-mm
cell size
Substrate
DuPont alumina
DuPont alumina
Surface area
(m2/g)


Pre-test
Post-test
5.36
0.986
4.88
0.05
Dispersion
(moles H2/g)


Pre-test
Post-test
17.63
1.30
1.22
0.39
Operating
temperature (K)
1561
1561
Maximum heat
release rate
(J/hr-Pa-m^)
7.4 x 106
2.6 x 106
Typical emis-
sions (ppm)


N0X
CO
HC
1
30
12
2
60
35
215
-------
FIXED-BED AND SUSPENSION FIRING OF COAL
Byi
S. P. Purcell, D. M. Slaughter, J. H. Munro, G. P. Starley,
S. L. Mania, and D. W. Pershing
Department of Chemical Engineering
University of Utah
Salt Lake City, Utah 84112
216
-------
ABSTRACT
This paper summarizes the progress made during the second year of a grant
to study the formation of pollutant species, particularly nitrogen and sulfur
oxides in fixed-bed and suspension combustion of coal. During this period,
the suspension furnace was completed and preliminary experiments conducted.
The results suggest that the large particles are essentially unreacted when
they reach the stoker bed. The burning rate of the small particles increased
with increased local oxygen concentration and increased heating rate.
The fixed-bed furnace was also completed and initial results were obtained.
These data suggest that the nitrogen volatiles evolve from the bed early in
the combustion process and form significant amounts of nitrogen oxides. Staged
combustion appears to be a potentially effective means of controlling NOx
emissions from a fixed-bed system. Increased clinkering problems were not
observed under staged combustion conditions; however, this is a potential
Problem in the application of this technology to larger scale units. Both
overall excess-air level and overfire-air height were studied, but neither
appeared to be of first-order importance at the conditions investigated.
The model spreader stoker was designed and the fabrication drawings pre-
pared. Construction should be completed in late 1980.
217
-------
ACKNOWLEDGMENTS
This research was supported by the United States Environmental Protection
Agency under Grant No. R-805899. The considerable help and advice of G. Blair
Martin, EPA Project Officer, is sincerely appreciated. In addition, input from
the engineering staff at Delco-Remy Division, General Motors: Dr. David Junge,
Oregon State University; Robert Giammar, Battelle; Mr. Niel Johnson and the
engineering staff at Detroit Stoker; and, Dr. M. P. Heap, Energy and Environ-
mental Research, is gratefullly acknowledged. Thanks are also due to the
Vibra Screw Corporation for providing the very accurate solids-metering system
used in this work. Last, but certainly not least, the authors gratefully
acknowledge Mrs. Colleen Anderson for her help in managing the project, and
preparing the required manuscripts, and Blake Beckstrom and Robert Brodbeck
for their assistance in building the equipment and gathering the data.
218
-------
SECTION 1
INTRODUCTION
Clean, liquid and gaseous fuels have substantially increased in price
and have become more difficult to obtain during the past decade. These
trends, along with a heightened environmental awareness, have resulted in
a renewed interest in coal as a domestically available alternative fuel.
Considerable research has been focused on utility coal combustion; however,
mid-range commercial and industrial coal-fired systems have been largely
ignored.
Industrial users have three potential alternatives available for burn-
ing coal to raise steam for industrial process utilization or facility space
heating. These coal-firing methods include:
. pulverized coal-fired boilers;
. fluid-bed coal-fired boilers; and,
. stoker coal-fired boilers.
In general, pulverized-coal (PC) firing is restricted to very large
industrial installations (above 500,000 pph steam), because of the high
capital cost associated with the pulverization equipment, and the large
inherent maintenance costs of this equipment. In addition, many industrial
and commercial users desire coal-firing systems which can utilize the fuel
as it is delivered without additional preparation.
Fluid-bed combustion (FBC) is a promising new technology that
utilizes large coal particles (relative to PC) and provides very high heat
transfer rates, FBC may eventually dominate the industrial sector; however,
to date it has experienced very limited acceptance in the United States.
219
-------
Stoker coal-fired systems are recognized as the current field
technology for medium capacity steam generation facilities. Stoker coal-
fired furnaces are most often located in densely populated, urban centers
because of size classification and method of utilization (i.e., factory
space heating, hospital heating, etc.). Hie presently operating installa-
tions represent major stationary polluting sources and increased future
utilization of stoker-fired boilers poses a serious threat to ambient
air quality in metropolitan areas. It is important, therefore, that
further knowledge be obtained on the formation and control of nitrogen
and sulfur oxides under conditions typical of stoker firing.
Recent large-scale pilot and field tests have at least partially
demonstrated the potential of combustion modifications for N0X control in
stoker-fired boilers. Giammar, et al (1) have conducted tests on a
25,000 pph steam spreader stoker with a water-cooled vibrating grate and
found that decreased overall excess air resulted in lower NO* emissions
(at a constant load). Gabrielson and Langsjoen (2, 3) have tested several,
large-scale industrial stokers and reported little effect of overfire air
percentage on NQx emissions and an apparent increase in NO* with increasing
load. However, the fundamental mechanisms controlling these effects are
not well defined.
220
-------
SECTION 2
PROJECT OVERVIEW
PURPOSE
The overall objective of this program is to study the formation of
nitrogen and sulfur oxides under carefully controlled experimental condi-
tions typical of stoker-fired boilers. In particular, this program is
considering the following major research areas:
1.	the evolution and oxidation of fuel
nitrogen and sulfur;
2.	the retention of sulfur oxides by ash
and/or solid-chemical sorbents in both
suspension and fixed-bed burning;
3.	the effectiveness of combustion modifications
for N0X control in stoker-fired coal systems.
In addition, the study will attempt to quantify the combustion process in
a stoker environment and consider possible detrimental effects of control
technology on boiler operation.
APPROACH
This project is considering both spreader and mass-burning (thick
bed) stoker systems. Conceptually, the combustion process in a spreader
stoker can be divided into two major parts:
. suspension burning and
. fixed-bed burning.
As the particles pass through the overthrow zone, they are heated,
partially devolatilized, and may even burn completely if they are small
221
-------
enough. The large particles form a thin (2 to 4 inches) active fuel bed
and burn out heterogeneously. In a mass burning stoker (e.g., underfeed)
the particles may be partially devolatilized by radiant heating prior to
actual ignition.
The approach in this study has been to divide the processes into the
separate pieces for individual study before attempting research on the
coupled system. In particular, the program consists of three major tasks:
Task 1. An experimental investigation of
suspension burning of bituminous
coal;
Task 2. An experimental investigation of
fixed-bed burning of bituminous
coal; and,
Task 3. An experimental optimization of
combustion modification technology
using a model stoker system.
Although the primary thrust of the study is experimental, each of the
furnace studies will be supported by analytical modeling.
222
-------
SECTION 3
SUSPENSION BURNING
SUSPENSION FURNACE
The purpose of this task is to investigate the formation of pollutant
emissions during the suspension-burning phase of the combustion process.
Figure 1 illustrates the furnace system which was fabricated for these
studies. The inner combustion chamber is six inches in diameter and the
outer shell is 28-inches square. The walls of the main furnace sections
consist of an outer steel shell, five inches of 1900°F insulating block,
four inches of 2500°F insulating refractory, and two inches of 3400°F high-
temperature castable refractory. Each of the mid-sections contain five,
two-inch diameter ports for second-stage air injection and/or insertion
of species and temperature probes.
The side walls have provision for auxiliary heating (or cooling) by
firing natural gas or passing cooling air through outer channels. These
channels are two inches in width and cover 50 percent of the circumference.
This allows partial independent control of the radiant heat transfer to
the particles and more complete control of wall temperature.
A high-intensity gas burner is attached to the horizontal extension
at the furnace bottom. This burner produces hot combustion gases which
flow vertically upward and simulate the combustion products leaving the
stoker bed. Large, stoker-size particles are fed into a water-cooled collar
at the furnace top via an auger/vibrating tray system (Figure la). The
partially burned particles are collected at the bottom of the furnace in
a water-spray trap (Figure lb). The trap contains six commercial spray
223
-------
nozzles which create a water screen through which the particles must fall.
A water bath at the bottom of the trap insures complete quenching.
Exhaust gas samples are continuously withdrawn through a water-cooled,
stainless steel probe positioned in the flue. NO and NOx are measured with
a chemiluminescent analyzer while CO, CO2, and O2 are determined by NDIR's
and paramagnetic analyzers. Coal and char compositions are determined with
a Perkin-Elmer 240 B Elemental Analyzer.
EXPERIMENTAL RESULTS
The experimental results from the suspension phase experiments are
both direct and indirect. During the actual experiment, real-time data on
the nature and extent of combustion are obtained indirectly by measurement
of the composition of the flue gases leaving the furnace. Subsequent to
the actual experiments, the coal and partially burned particles are
analyzed to directly establish the carbon, hydrogen and nitrogen loss.
Table I shows the ultimate and proximate analyses of the coal being
used throughout this program. It is a high-volatile, bituminous coal from
southern Utah, and it contains 1.68 percent N (DAP). Figure 2 shows the
particle-size-distribution data for this coal plotted on the standard,
spreader stoker size chart.
In the first experimental tests, the normal size distribution coal
was fed at a rate of 12 lbs/hr, and the gas burner (bottom) was fired at
150,000 Btu/hr. The upper curve in Figure 3 summarizes these results. As
the mean free-stream-oxygen concentration was increased from two percent
to 10 percent, the coal burning rate increased significantly, indicating
a strong dependence on oxygen partial pressure. However, at all conditions,
the burning rate was low; less than 10 percent of the 12 lbs/hr being fed
were actually consumed.
Analysis of the solid particles collected at the bottom of the furnace
indicated that:
. nitrogen was being preferentially evolved; and,
. the larger particles were essentially unreacted.
224
-------
These results suggested that the fines (particles less than 0.1
inch) were very important in the suspension-phase burning. To investi-
gate this, the standard coal was sieved with an 0.096-inch screen. This
removed approximately 15 percent of the coal (by weight). The remaining
large particles were then fed into the furnace, and the experiments
repeated. The lower curve of Figure 3 shows these results which
confirmed that under the conditions of these tests (1700°F walls,
6 ft vertical drop), only the smallest particles reacted.
To investigate the influence of particle-heating rate, similar
tests were conducted with the auxiliary burners each firing at 40,000
Btu/hr. Figure 4 shows the influence of these burners on the wall-
temperature profile. The auxiliary heating both raises the overall
temperatures and makes the profile more uniform. Figure 5 shows the
influence of the increased heating rate on the large particles (greater
than 0.1 inch). In these tests, the coal-feed rate was 10 lbs/hr; hence,
less than three precent of the coal was burning.
PRELIMINARY CONCLUSIONS
Based on the limited data obtained to date, it appears that the
overthrow phase of a spreader stoker may have a significant impact on the
fines, but it probably has little effect on the bulk of the coal feed.
Further, nitrogen evolved from the small particles may be converted to
NOx in much higher proportions than nitrogen evolved in the bed; hence,
the fines may be of exaggerated importance in terms of NO* formation.
225
-------
SECTION 4
FIXED-BED COMBUSTION
FURNACE DESIGN
The fixed-bed furnace was designed to provide a Lagrangian simulation
of the time/temperature/environmental history seen by a small section of
the stoker bed as it moves through a large, mass-burning chain-grate or
vibrating-grate stoker. In this way, the horizontal, planer dimensions
were reduced dramatically compared to the full-size unit; however, the
vertical dimension was scaled 1:1. Figure 6 shows a side view of the fixed-
bed furnace which was designed to have a maximum superficial firing rate
of approximately 600,000 Btu/hr-ft2. The furnace itself is 27 inches by
23 inches by eight feet outside, and the inner combustion chamber is 11
inches by 15 inches.
The furnace is composed of four, separate, flanged sections. Section
A, the upper most section, contains the six-inch circular flue, and a three-
inch, expanding quartz window to provide a view of the combustion chamber
from above. Sections B and C are the central furnace pieces and contain
overfire-air slots, large observation ports, and probe ports. All three
of the upper furnace sections have composite walls consisting of one-eight-
inch carbon steel, four inches of block insulation, and two inches of high
temperature refractory.
The lower furnace section, labeled D, contains the experimental coal
bed. The bed proper is 10 inches wide and six inches deep with the total
surface area of 0.42 ft2. Immediately adjacent and surrounding the bed
is a one-inch, high temperature, refractory wall, which is used to confine
the coal. Just outside the wall, there are porous stainless steel, methane
burners with an equivalent surface area (0.42 ft2). These gas burners are
226
-------
used to provide uniform heating of the combustion chamber when coal is not
being burned. The coal bed and gas burners are mounted in a "drawer-like"
assembly which rolls out of Section D to permit easy loading of the coal
particles.
The combustion air is distributed between the underfire-air plenum
and two levels of overfire-air ports. In this study, the air was metered
with 600 mm rotameters, but was not preheated.
Exhaust concentration measurements were made in the flue above Section
A with a water-cooled, stainless steel probe. The instrumentation described
in the previous section was also used for the fixed-bed experiments.
TYPICAL EXPERIMENTAL RESULTS
Figure 7 shows typical experimental results obtained for coal-burning
rates as a function of run time. The run was conducted with the Utah coal
(Table I), sized at one-half inch through one inch. The data are presented
in terms of superficial burning rate; the rate of heat release per unit
area of bed surface. The superficial burning rate was computed from exhaust
oxygen concentrations and air-supply flow rates. The data show that as
the run proceeded the burning rate increased, reached a maximum, and then
decreased. It is believed that the burning rate increase is attributable
to an increase in bed temperature. Limited temperature measurements indicated
that the coal bed burned in excess of 2400°F. The average coal-burning
rate over the control interval (six minutes to 25 minutes in this case) was
311,000 Btu/hr-ft2 for this particular run.
In order to separate parametric effects, it was necessary to keep both
total and bed-region stoichiometrics constant during a coal run, even though
the burning rate changed significantly. This required precise control of
the air supply, which was achieved by continuously monitoring the carbon
dioxide concentrations in the exhaust gases.
Figure 8 gives the overall and bed-region stoichiometrics as a function
of run time. The stoichiometric ratios were computed from effluent gas
oxygen corcentrations and air-supply-flow rates at one-minute intervals.
Both the overall and bed-region stoichiometrics were controlled to within
approximately ± 0.1 of their respective average stoichiometrics for the
227
-------
run, except in the early stages of the experiment. A control interval was
defined for each run on the basis of these narrowly bounded, average
stoichiometrics. Average values for NO emissions and burning rates were
computed for this interval only.
Figure 9 shows the effluent-gas concentration of NO converted to
zero percent exhaust O2 as a function of run time. The exhaust gas
concentration of NO increased rapidly in the initial stages of the run,
passed through a maximum and then decayed toward the termination of the
run. Comparison of Figures 7 and 9 indicates that NO formation peaks prior
to achieving the maximum combustion rate. This "early" NO generation is
typical of all the coal runs.
INFLUENCE OF BED STOICHIOMETRY
Figure 10 shows NO emission results (plotted as lbs NO2/IO6 Btu) for
three different bed stoichiometrics. In each test, the overall stoichiometry
was held constant at 1.35 ± 0.03, and the firing rate was approximately
500,000 Btu/hr-ft2. The overfire air was added in two volumetrically
equivalent portions through opposing jets at 15j inches and 44 inches above
the coal bed. Neither the overfire nor underfire air was preheated in these
experiments.
As the combustion was staged, with the total air supply held constant,
the NO emissions decreased substantially. Based on these limited data the
dependence of N0X on primary zone stoichiometry appears to be approximately
linear, and this is in marked contrast to pulverized-coal combustion. No
problems with clinkering or incomplete combustion were encountered; CO
missions were always less than 0.03 percent.
BURNIN& RATE
Figure 11 shows the integral superficial firing rates for two cases
with nearly identical stoichiometric distributions (S.R.g » 0.83 ± 0.02,
S.R>t « 1.37 ± 0.05), yet widely different burning rates. The lower burn-
ing rate experiment was conducted by significantly cooling the walls of
the b'ed prior to charging the reactor with coal. For each experiment, the
data are presented for the combustion of 4.5 pounds of coal (hence, in
Figure 11, the high burning rate case ends first at 16 minutes). An average
228
-------
burning rate of 546,000 Btu/hr-ft2 was achieved for the "hot" run, while
a corresponding burning rate of only 398,000 Btu/hr-ft2 was attained for
the "cold" run.
Figure 12 shows NO emissions for both the "hot" and "cold" experi-
ments . The rate of NO formation was substantially greater for the "hot"
case, in spite of the fact that in both experiments the bed was fired with-
out excess air. The disparity in rates of NO formation cannot be attributed
simply to a difference in the amount of coal consumption since the data
were normalized to the same basis (lbm NO2/IO6 Btu) independent of coal-
feed rate. The integrated emission factor over the control interval for
the high temperature run was 0.273 lbm NO2/IO6 Btu as compared with the
respective average emission factor for the low temperature experiment of
0.180 lbm NO2/IO6 Btu. The marked difference in NO emissions between high
and low load firing was observed during oxidant-rich combustion as well,
indicating a strong dependence of NO formation upon bed heat-release rates.
The dependence of NO emissions upon bed heat-release rates suggests
that as the burning rate increases, local bed temperatures increase. This
may result in higher volatile nitrogen yields and, hence, higher NOx
emissions.
PRELIMINARY CONCLUSIONS
The following tentative conclusions have been reached regarding the
formation of nitrogen oxides in a fixed-bed combustion system:
1.	NO formation is not uniform throughout
the burning time of the bed. The
formation maximizes early in the
process, prior to the peak combustion
rate and then gradually decreases.
2.	Staged air addition is a potentially
effective means of NO control. Exhaust
concentrations decrease approximately
linearly with decreasing bed stoichiometry.
229
-------
3. Increasing the bed heat-release rate
increases the N0X emissions with
staged-air conditions.
230
-------
SECTION 5
MODEL SPREADER STOKER
The purpose of the third task is to apply the pollutant reduction
concepts developed in Tasks 1 and 2 to an actual stoker system in which
the suspension and fixed-bed burning phases are coupled. A model spreader
stoker has been designed based on the following parameters which were
scaled 1:1 with current commercial practice:
. bed heat-release (approximately
750,000 Btu/hr-ft2);
. spreader-to-grate distance
(approximately 32 inches); and,
. grate-to-convective passage distance.
The bed heat-release rate is a direct reflection of the local burning rate
and, hence, the particle temperatures, local oxygen availability, etc. The
spreader-to-grate distance determines the residence time of the particles
in the suspension phase and, therefore, the amount of devolatilization that
occurs prior to the bed. The grate-to-convective passage distance controls
the gas-phase residence time prior to strong quenching.
The overall furnace design is illustrated in Figure 13. As shown,
the furnace will stand 10.8-ft high, which corresponds approximately to
the height of the radiant section in a spreader stoker furnace. Section 1
is the ash storage secton and will be equipped with an ash drawer capable
of holding ash from a 14-hour run.
Section 2 is the grate-burning bed section. The 1 ft2 coal grate
will actually be a type of dumping grate design in which the ash is allowed
to build up to some predetermined depth and then dumped. After dumping,
231
-------
two inches of ash will remain on the grate to partially insulate the metal
grate from excessive heat. Section 2 will contain ports for flame and solids
sampling, a large removable section for access to the bed, and a 4-inch
window for observation of the bed.
Section 3 will contain two horizontally opposed gas burners (300,000
Btu/hr each) for use between experimental runs and overnight when the coal-
feed system will be shut down. Overfire-air ports will be located in both
front and back walls and there will be a large slot port for insertion of
probes or cooling coils. The side walls will contain 2-inch diameter ports
for a pilot burner and ultraviolet flame detectors.
Section 4 will include the coal distribution system. Coal from a sealed
storage hopper will be metered into the furnace at a rate of up to 60 lbs/hr
via an auger. The coal will then drop onto a vibrating tray which will
serve to dampen any pulsations from the auger, and also to distribute the
coal uniformly to a width of eight inches. At the end of the tray, the
coal will drop into the furnace shell and onto the rotating spreader. An
adjustable plate will be used to vary the exact position at which the rotatinc
spreader strikes the coal particles. The variable-speed spreader will be
air-cooled.
Sections 5 and 6 will be equipped with slot ports for overfire air,
gas-sampling probes, or cooling coils. Although it is anticipated that
the furnace will be operated in an uncooled mode of operation, the capa-
bility of inserting cooling coils will allow simulation of the bed adjacent
to a water-tube wall.
Section 7 will contain probe ports for temperature, particle mass
and size distribution, and gas-species concentration measurements.
The model spreader stoker is currently being fabricated and should
be ready for initial shakedown testing in October, 1980. Initial experiments
will be conducted to define the interaction between the suspension burning
and the fixed-bed combustion since each of these will have been considered
separately in the previous tasks. Later, testing will be directed toward
defining and optimizing new combustion control technology.
232
-------
REFERENCES
1.	Giammar, R. D. Evaluation of Emissions and Control Technology for
Industrial Stoker Boilers. In: Proceedings of the Third Stationary
Source Combustion Symposium, EPA-600/7-79-050, NTIS, Springfield,
Virginia, 1979.
2.	Gabrielson, J. E. Field Tests of Industrial Stoker-Fired Boilers for
Emission Control. In: Proceedings of the Third Stationary Source
Combustion Symposium, EPA-600/7-79-050a, NTIS, Springfield, Virginia,
1979.
3.	Langsjoen, P. L. Characterization of Emissions from Industrial
Stoker Boilers. Presented: APCA Conference on Industrial Boilers,
Research Triangle Park, North Carolina, December, 1979.
233
-------
ir///////

Coal
i
Auger
Feeder
Window
Ports
|Exhaust
Burner
Water —-
—•	n ^Vibrating Tray
Water-Cooled Tube
Figure la. Feeding System Detail
Particle Collection System
mm
mm
mm
— Water
Noz2les
20 Mesh Screen
Water
Bath
3 Inch
Gate Valve
Figure lb. Particle Collection System
Figure 1. Suspension Furnace
234
-------
95
90
80
70
60
50
40
Recommended
Upper Limit"
This Study
10
8
6
4
3
Recommended
Lower Limit
2
1
200 100
30 20 16 12 8
i 3/81 3/4 1
2
U.S. Std. Sieve Designation	Screen Opening
Figure 2. Particle-Size Distribution
-------
1.0
0.8
in
JQ
(O
u
o*
-------
2200 -
Hot - Auxiliary Burners On
2000
u_
« 1800
a.
E 1600
Normal
1400
Heated Zone
EZZZZZZZZZ
1200
Distance from Sample Quench (inches)
Figure 4. Wall-Temperature Distributions
-------
Oxygen Concentration (%)
Figure 5. Influence of Particle Heating Rate
-------
Flue
Figure 6.
Side View of the
Fixed-Bed Furnace
Overfire air port
Observation Port
6" x 10" Coal Bed
Underfire air plenum
-------
in
w a
oc 4
CT>
Q.
Time (min)
Figure 7. Superficial Burning Rate: UBS - 1/9/80-4
-------
• S.R. Total
O S.R. Bed
I	•
I	I
I	I
	I	I	I	I	
10 15 20	25 30
Time (min)
Figure 8. Overall and Bed-Region Stoichiometry:	UBS - 1/9/80-4
-------
400
T
T
300
£
o
CM
o
K3
s 200
E
Q.
Q.
100
10
15
Figure 9.
T
T
o —
20
Time (min)
NO Emissions:
25
30
UBS - 1/9/80-4
-------
0.6
o
SRBED 1,13
3
•M
OQ
 z
0.2
SRT0TAL 1'33
srbed = 0,83
Time (min)
25
Figure 10. NO Emission Results for Three Different Bed Stoichiometrics
-------
L 600,000
CO
-------
B.R. = 600,000 Btu/hr-ftz
400•000 "Kr^ft2
srtotal * 1,35
srbed = 0,82
10	15
Time (min)
20
25
Figure 12. Influence of Burning Rate on NO Emissions
-------

Gas
Burners

Cooling
Coil
Slots
Removable'
Section
Gas
Saxnpling
Port
Underfirer
Air		
Ash
Drawer

Window
Solid
Sampling 2
Port
Figure 13.
Model Spreader Stoker
246
-------
Table I
Chemical Analysis - Utah B Coal
Proximate Analysis
Weight % As Received	Dry Basis
Moisture	6.22
Ash	9.59	10.23
Volatile Matter	37.84	40.35
Fixed Carbon	46.35	49.42
Total	100.00	100.00
Calorific Value (Btu/lb)
Proximate Value	11,658	12,432
Mineral Matter Free	- -	14,004
Ultimate Analysis
Weight % As Received	Dry Basis
Moisture 6.22
Carbon 66.44	70.85
Hydrogen 4.86	5.19
Nitrogen 1.42	1.51
Sulfur 0.88	0.94
Ash 9.59	10.23
Oxygen (difference) 10.59	11.28
Total	100.00	100.00
247
-------
PRESSURIZED BENCH SCALE TESTING OF LOW N0X
LBG COMBUSTORS
By:
H. D. Clark, B. A. Folsom, W. R. Seeker,
C. W. Courtney, and M. P. Heap
Energy and Environmental Research Corporation
8001 Irvine Boulevard
Santa Ana, California 92705
248
-------
ABSTRACT
The high efficiencies obtained in a combined gas turbine-steam turbine
power cycle burning low Btu gas (LBG) make it a potentially attractive alter-
native to the high sulfur emitting direct coal-fired steam cycle. In the
gasification process, much of the bound nitrogen in coal is converted to
ammonia in the LBG. This ammonia is largely converted to nitrogen oxides
(N0X) in conventional combustors. This paper examines the bench scale per-
formance of reactors previously demonstrated to produce low NQX emissions in
laboratory scale experiments. Low Btu gas was synthesized in a catalytic
reformer and fired in two primary combustors: a diffusion flame and a
platinum/nickel oxide catalytic combustor. Effects of scale, primary
stoichiometry, pressure, throughput, and primary residence time were examined,
Lowest N0X emissions were produced in rich/lean combustion, utilizing either
the diffusion flame or the catalyst in the fuel rich primary stage.
249
-------
ACKNOWLEDGMENT
The investigations described in this paper were carried out under EPA
contract 68-02-2196. The authors wish to express their appreciation to
Mr. G. B. Martin, the EPA Project Officer, to Dr. J. Kesselring of the Acurex
Corporation for his help in obtaining the catalysts, and to Mister M. Maser
of Energy and Environmental Research Corporation for his assistance in conduct-
ing the experiments.
250
-------
SECTION 1
INTRODUCTION
The limited availability and the increasing cost of petroleum based fuels
have generated an increased interest in coal as a primary power source. The
potential for low sulfur emissions makes combustion of gasified coal an environ-
mentally attractive alternative to direct fired coal combustion. However, low
Btu coal gases can contain ammonia concentrations a high as 0.38 percent (1).
In a conventional combustor, much of this ammonia may be converted to nitrogen
oxides resulting in significant pollutant emission: up to 1370 ng/J (3.2 lbm/
106 Btu) for full conversion of NH3 to NO2 •
The gas turbine - steam turbine combined cycle is an efficient method of
low Btu gas combustion. The current New Source Performance Standard (NSPS) for
steam generators firing coal-derived gaseous fuels with greater than 73.3 MW
(250 X 106 Btu/hr) heat input is 210 ng/J (0.5 lbm NO2/IO6 Btu)(2). For a
combined cycle firing an LBG with 3800 ppm NH3 to NO would have to be less than
15.7 percent. With the advent of new combustor and cleanup technology it is
reasonable to expect that future NSPS will be more stringent; therefore, in
the development of energy-efficient LBG combustion systems it is important to
achieve the lowest N0X emissions possible.
Equilibrium and kinetic calculations on fuel nitrogen processing in LBG
combustion have been tested at atmospheric pressure in laboratory scale experi-
ments (3-4) resulting in the development of several low NOjj combustor concepts.
This paper describes experiments performed on scaled-up LBG combustors, utili-
zing these low NOx concepts, operated at pressures more nearly approximating gas
turbine combustors, Sections are included to discuss background, experimental
systems, results and conclusions.
251
-------
SECTION 2
BACKGROUND
Coal-derived low Btu gas comprises a family of fuels of varying composition,
ranging in heating value from 3.0 X 10s to 7.5 X 10s J/m3 (80 to 200 Btu/ft3),
produced by reforming coal with air and steam. ' Its principle components are CO,
Rz and light hydrocarbons as combustibles; and N2, H2O and CO2 as inerts. Signi-
ficant amounts of ammonia, sulfur compounds and particulate matter may also be
present. These impurities may be partially removed in several alternative gas
cleaning systems prior to combustion (5). The most energy efficient cleanup
systems operate hot, removing a majority of sulfur products and particulates,
but some systems do not remove much of the ammonia from the LBG.
Figure 1 is a simplified schematic of an LBG-fired gas turbine-steam turbine
combined cycle power plant with integrated gasifier. In the high pressure gas
turbine topping cycle, LBG is produced in a gasifier and burned in an adiabatic
gas generator to produce the hot pressurized gas that turns the gas turbine.
Heat in the exhaust gas is recovered in a waste heat boiler and utilized to
drive a steam turbine bottoming cycle. Under optimum operating conditions
utilizing current technology including high-temperature sulfur removal, an
overall efficiency of 40% can be achieved in the combined cycle. Future genera-
tion systems, operating at higher pressures and temperatures, could achieve
efficiencies up to 48%. Variations of the combined cycle which can improve
the performance of the bottoming cycle and increase the overall efficiency
include utilizing the gas generator as a super-charged boiler and reheating
the gas turbine exhaust. A more detailed discussion of combined cycle optimi-
zation Is Included in reference 6. A low NOg LBG combustor would function as
the gas generator of the combined cycle. It must operate at high pressure and
its effluent gas should be low in NOx, CO and hydrocarbons.
252
-------
Figure 2 shows that minimum equilibrium fixed nitrogen concentrations of
less than 1 pptn occur in the adiabatic combustion of a 3.35 X 106 J/m3 (90 Btu/
ft3) LBG between 60 and 80% theoretical air. The location and value of this
minimum change with adiabatic flame temperature, depending upon the composition
and preheat temperature of the LBG. Lower temperatures push the minimum to
leaner stoichiometrics. An ideal low N0X combustor, shown in Figure 3, would
operate in two stages. In the fuel rich-primary stage, sufficient holdup time
would allow EXN (2XN - NH3 + HCN + NO + NO2) to approach an N2 favored equilib-
rium. In the fuel-lean secondary stage, the remnants of the fuel would be
burned out, meeting the low CO and HC requirements.
In a practical combustor, the mechanics of primary fuel/air contacting
affect the fuel nitrogen processing. Laboratory scale flame reactor experiments,
discussed in detail in references 3, 4 and 7, to investigate and optimize these
effects were performed in a previous part of this program. Promising results
were achieved utilizing either a catalytic reactor or a diffusion flame reactor.
Use of a catalyst in the rich stage could shorten the holdup time required
to reach low ZXN levels, reducing the size of the combustor. Studies of NO
reduction in automotive exhaust have shown that platinum effectively promotes
the reduction of NO to NH3 and in combination with an NH3 to N2 catalyst, such
as nickel, can speed the complete reduction of NO to N2 (8). Catalytic combus-
tion studies have indicated the potential of several catalysts for XN reduction
in rich combustion, including platinum and nickel oxide (9). For gas turbine
combustion application, pressure-drop must be kept to a minimum. The graded
cell catalyst concept has been developed to maximize throughput while keeping
emissions and pressure-drop low (10). Ceramic honeycomb monoliths are coated
with the catalyst and stacked in the order of decreasing cell size. The coarse
upstream cells tend to hold the flame while the fine downstream cells tend to
complete the reaction.
In a lean-diffusion flame, a single fuel nitrogen molecule passes through
the entire range of stoichiometrics: from an initially very rich zone, through
the stoichiometric flame front, into the lean post flame zone. Thus a lean
253
-------
diffusion, flame is, in effect, a multistage combustor. Low conversions of fuel
nitrogen to NOx ahve been observed in fuel-lean laminar diffusion flames.
Turbulence tends to increase NOx emissions (11).
In a fuel-rich laminar diffusion flame, part of the fuel never passes
through the flame front; thus, depending on downstream temperatures and mixing
conditions, it is possible for some of the fuel nitrogen to pass through unpro-
cessed. High NH3 concentrations have been measured in the products of an NH3-
doped fuel-rich laminar diffusion flame (12). Turbulence in a rich diffusion
flame would be expected to enhance mixing in the hot region within the flame
and immediately downstream, increasing the reduction of fuel nitrogen.
254
-------
SECTION 3
EXPERIMENTAL SYSTEMS
The apparatus for the bench scale experiments can be divided into four
subsystems: LfiG supply, modular combustors, sample train and control systems.
A simplified schematic of the facility can be seen in Figure 4,
LBG is synthetically produced from hot air premixed with vaporized water
and heptane passed through a catalytic reformer. The reformer is operated at
pressures between 6.4 and 11.9 atmospheres at a stoichiometry of 45% theoreti-
cal air* the richest stoichiometry attainable without excessive sooting. The
water acts as a diluent to maintain the maximum catalyst bed temperature at
around 1370° K. Table I shows measured and equilibrium LBG compositions at
the standard operating point. The reformer product gas passes through a varia-
ble heat exchanger, cooling it to the desired preheat temperature. Ammonia
and methane are added to trim the gas to the desired fuel nitrogen and hydro-
carbon content. The LBG passes through a soot filter and into a valve system,
controlling the fraction of the LBG which goes to the combustors and the frac-
tion which is bypassed. If none of the gas is bypassed, maximum combustor
capacity is 60,000 J/s (200,000 Btu/hr).
The combustors consist of a series of modules with 5 cm (2 in) ID reaction/
flow chambers enclosed in 15 cm (6 in) OD low-density insulation and housed in
flanged steel pipe. Primary ignition modules include the catalyst and the
diffusion flame. Secondary burnout is achieved in the jet-stirred secondary
air injector. Plug flow modules of various lengths allow control of primary
and secondary residence times. Platinum/rhodium thermocouples in each module
allow primary and secondary temperature measurements. The interchangeable
modules can be combined to model and optimize a variety of combustor concepts.
The primary ignition modules are shown in Figure 5. In the catalyst module,
premixed LBG and primary air pass through a stainless steel flow straighltener/
flame arrestor and into the graded cell catalyst. The catalyst, supplied by
Acurex, consists of three zironia honeycomb monoliths of decreasing cell size,
255
-------
coated with nickel oxide. Platinum has been added to the coating of the up-
stream monolith to promote ignition. Thermocouples detect flashback and
measure catalyst bed and wall temperatures. In the concentric diffusion flame
module, LBG is introduced through a removeable fuel tube of variable diameter.
Straightened primary air passes annularly around the fuel tube in the direction
of the fuel flow.
Samples are taken in the secondary region, through a water-cooled stainless
steel probe situated on the centerline of the flow chamber. The cooling water
is preheated and the stainless steel sample lines are wrapped with heat tape
to maintain the sample system above the dewpoint of the exhaust gases. The
sample stream is throttled to nearly atmospheric pressure. Batch wet samples can
be taken by two bubblers in series and later analyzed by specific ion techniques
to determine the ammonia and hydrogen cyanide content of the gas. Low NH3 and
HCN concentrations have been measured (less than 10% of ZXN), indicating that
the secondary zone is long enough for nearly complete processing to N0X.
Continuous gas samples are passed through a water trap maintained at 0°C,
through a teflon-lined sample pump and into the emissions console which measures
CO and CO2 by infrared analysis, NO and NO* by chemiluminescence, O2 by para-
magnetic analysis, and hydrocarbons by flame ionization. LBG samples are taken
downstream of the soot filter in a similar system.
Air, water, heptane, methane and ammonia streams are measured by rotameter.
LBG flow to the combustors and primary and overall stoichiometry are calculated
by oxygen balance, utilizing the secondary O2 concentration and the primary and
secondary air flows. Reformer pressure is controlled by a valve on the LBG
bypass line. Combustor pressure is maintained by passing the exhaust gases
through a small orifice downstream of the combustors. The pressure upstream
of the orifice is a function of the mass flow rate and the temperature of the
exhaust gases. Constant pressure tests are performed by keeping the LBG flow
and the total air flow constant. The primary stoichiometry can be changed by
varying the ratio of primary to secondary air, but the overall stoichiometry
is maintained constant. The exhaust gas temperature in the nearly adiabatic
system is a function primarily of the overall stoichiometry. Variable pressure
256
-------
tests are performed by varying the total flow. The opposing effects of in-
creased pressure and increased flow tend to keep residence tines constant
for variable pressure tests. Variable pressure tests at constant flow (variable
residence time) can be performed by changing the size of the flow-restricting
orifice.
257
-------
SECTION 4
RESULTS
Encouragingly low N0X levels have been achieved on the bench scale
utilizing a catalytic reactor and a diffusion flame reactor. An effective
fuel nitrogen-reducing catalyst was identified in laboratory scale experi-
ments and the effects of scale and stoichiometry were examined in the bench
scale experiments. A fuel-lean diffusion flame was identified as an attrac-
tive low N0X combustor concept in laboratory scale experiments and effects of
scale, stoichiometry, hydrocarbon content of the fuel, fuel tube size, pressure,
and primary residence time were examined in the bench scale experiments.
Effects of catalyst type on fuel nitrogen processing in LBG combustion
were examined on the laboratory scale in an unstaged catalytic reactor operated
at a constant adiabatic flame temperature of 1473°K. Figure 6 shows the variable
stoichiometry results for two catalysts. The alumina supported platinum cata-
lyst converted almost all fuel nitrogen to N0X in fuel-lean combustion and had
a minimum conversion of 40% in fuel-rich combustion. At stoichiometrics richer
than 60% theoretical air, decreasing NO concentrations were overwhelmed by
increasing NH3, and HCN concentrations, causing a sharp rise in EXN. The
zirconia supported platinum/nickel oxide catalyst converted 80% of the fuel
nitrogen to N0X in lean combustion, but had very low conversions in rich
combustion. For a 500 ppm NH3 in LBG dopant level, less than 10 ppm EXN were
measured at stoichiometrics as rich as 40% theoretical air. Tests of the
platinum/nickel oxide catalyst over a range of adiabatic flame temperatures
(1273-1673°K) and with CHit as the fuel yielded similar results.
A rich/lean series staged platinum/nickel oxide primary catalytic reactor
was selected as a potential low NOx concept for bench scale testing. The
scale-up results were in general agreement with the laboratory scale results.
Figure 7 compares the results of staged combustion of a 500 ppm NH3 doped LBG
at two scales: 1200 and 20,000 J/sec (4000 and 70,000 Btu/hr). Each had high
conversions of NH3 to NOx in fuel-lean combustion. Minimum conversions occurred
258
-------
in rich/lean staged combustion when the primary was operated close to-stoichio-
metric. At a primary stoichiometry of 90% theoretical air, the laboratory
scale catalytic reactor converted 8% of the input NH3 to N0X while the bench
scale combustor had an average conversion of 14%. Conversions in the bench
scale combustor remained low (less than 18%) over all rich primary stoichio-
metries under normal operation; but breakthrough occurred if the primary was
operated richer than 75% theoretical air: the temperatures on the walls of
the catalyst monoliths dropped and the conversion rose sharply. Breakthrough
was not observed in the laboratory scale experiment where the adiabatic flame
temperature was maintained at a constant 1473°K by varying the amount of
nitrogen diluent in the reactants. The undiluted flame reactor LBG had a
higher heating value (HHV) of 6.7 X 10s J/m3 (180 Btu/ft3) while the HHV of
the bench scale LBG was only 3.0 X 10® J/m3 (80 Btu/ft3). This indicates that
raising the heating value of the gas could extend the operating range of the
Pt/NiO catalyst, and that catalyst effectiveness is limited by a threshold
flame temperature below which breakthrough occurs.
It is difficult to compare the laboratory and the bench scale diffusion
flame combustors. Figure 8 shows laboratory and bench scale results for diffu-
sion flame combustion of LBG containing about 500 ppm NH3 and varying amounts
of methane. In the unstaged laboratory scale experiment, performed at atmos-
pheric pressure in a cold wall reactor under attached laminar flow conditions,
the hydrocarbon content of the LBG had the most significant effect on XN con-
version. Conversions as low as 10% were observed for combustion of hydrocarbon-
free LBG under nearly stoichiometric conditions. Under richer conditions,
conversion increased due primarily to increasing ammonia concentrations. However,
under leaner conditions conversions remained quite low. Similar trends were
observed in combustion of LBG containing 5% methane, but XN conversion was much
higher. In the staged bench scale experiment, performed at 8 atmospheres in a
nearly adiabatic combustor under turbulent flow conditions, effects of hydro-
carbon content and stoichiometry were not so pronounced.
259
-------
The bench scale flame was not visible and there was no reliable indicator
as to whether the flame was attached or lifted. However, throughput and tube
size ranged from conditions where the flame should definitely be attached to
conditions where the flame should definitely be lifted. No sharp changes were
observed in N0X emissions or in other measured parameters, indicating that the
attached/lifted transition was not an important factor. This agreed with pre-
vious variable throughput laboratory scale tests of a hydrocarbon containing
diffusion flame, where a smooth N0X transition was observed as the flame became
detached (3).
Figure 9 shows the effect of fuel tube size on XN conversion in the bench
scale diffusion flame operated at eight atmospheres. In constant pressure
operation at a fixed stoichiometry, fuel flow and primary residence time were
independent of fuel tube size, while Reynolds number was inversely proportional
to the fuel tube I.D., fuel tube size had little effect on XN conversion in fuel
rich combustion. However, in lean combustion, increasing tube size (decreasing
Reynolds number) decreased NH3 conversion to N0X. Increased tube size also
decreased the N0X noise level (high frequency concentration fluctuation shown
by the error bars in the figure), perhaps an indication of flame stability.
Figure 10 shows the effect of pressure on rich/lean and lean diffusion
flames. In the bench scale system, pressure is maintained by passing the
exhaust gases through a critical flow orifice. For a fixed stoichiometry, fuel
flow and Reynolds number are proportional to pressure while primary residence
time is independent of pressure. The staged tests were performed at a primary
stoichiometry of about 95% theoretical air. For low hydrocarbon LBG, NH3
conversion to NOx remained constant at 33% over pressures ranging from 4 to 8
atmospheres. For LBG containing 2.1% CH^, conversions remained constant around
40% with changing pressure. The lean tests were performed at a stoichiometry
of about 150% theoretical air. Noise levels were higher than in the staged case.
Conversions increased slightly with increasing pressure in low hydrocarbon com-
bustion. Little change in conversion was seen with changing pressure for the
2.1% CHu LBG.
260
-------
Primary residence time appeared to have the most pronounced effect on XN
conversion in a staged diffusion flame. Residence time was varied at constant
pressure by changing the pressure control orifice size. In constant pressure
operation at a fixed stoichiometry, fuel flow and Reynolds number were inversely
proportional to primary residence time. Figure 11 shows XN conversion with
primary stoichiometry for two different primary residence times. Using a large
pressure control orifice, a pressure of 8 atm was achieved at fuel tube Reynolds
numbers around 40,000 and primary residence times around 120 msec. A minimum
XN conversion of 34% was observed at a primary stoichiometry around 90% theoreti-
cal air. Using a smaller pressure control orifice, a pressure of 8 atm was
achieved at fuel tube Reynolds numbers around 20,000 and primary residence times
around 250 msec. A minimum XN conversion of about 22% was observed at a primary
stoichiometry around 90% theoretical air. For a 553 ppm doped LBG burned out to
150% theoretical air, this XN minimum corresponded to a N0X concentration of
100 ppm.
Figure 12 shows N0X concentration as a function of NH3 in the LBG for rich/
lean staged combustion in a diffusion flame and in a platinum/nickel oxide
catalytic combustor. For both the catalytic and the diffusion flame combustors,
N0X emissions increased with increasing fuel nitrogen content, but the increase
in N0X was much less than proportional to the increase in fuel nitrogen content.
The 3/8 0D tube diffusion flame, operated at a primary stoichiometry of 76%
theoretical air and a pressure of 4.4 atmospheres, converted 40% of its fuel
nitrogen to N0X at 553 ppm NH3 in the LBG and had conversions of only 11% at a
3800 ppm doping level. The catalyst, operated at a primary stoichiometry of
80% theoretical air and a pressure of 2.4 atmospheres, had XN conversions of
16% at the low NH3 doping level and 6% at the high doping level. Similar
trends, but higher N0X concentrations, were observed for both combustors in fuel
lean combustion.
261
-------
SECTION 5
DISCUSSION OF RESULTS
A rich/lean series staged combustor with a platinum/nickel oxide primary-
was the most promising low N0X combustor investigated. It had low conversions
of fuel nitrogen to NOx over a wide range of fuel rich primary stoichiometrics.
Thus, it could be operated rich enough to maintain the adiabatic flame tempera-
ture relatively cool, prolonging the life of the catalyst. However, catalyst
coated ceramics are often short-lived due to loss of activity of the coating
and structural problems of the support caused by thermal shock. During the
course of the bench scale experiments there was a great change in the appearance
of the Pt/NiO catalyst. A green coating formed on the surface. Also, the zirconia
honeycombs became quite fragile after repeated thermal cycling, especially the
fine-cell downstream monolith which was almost completely destroyed in the final
experiments. Further investigation is necessary of catalyst aging and of pressure
and throughput effects under optimized combustor conditions before a catalytic
combustor could be considered a serious candidate for a gas turbine combustor.
A rich/lean series staged combustor with a diffusion flame primary also
had low conversions of fuel nitrogen to N0X. Primary stoichiometry and residence
time had the most significant effects on fuel nitrogen conversion. Minimum N0X
emissions were achieved at primary stoichiometries around 90% theoretical air
for long primary residence times (250 msec or longer). Pressure and Reynolds
number had little effect on NOx in a staged diffusion flame, while an increase
in the hydrocarbon content of the LBG caused a slight increase in NOx emissions.
Combustion of a hydrocarbon-free LBG was not tested on the bench scale, but
laboratory scale tests indicated that the absence of hydrocarbons in the fuel
could cause a significant reduction in NOx emissions.
A lean unstaged diffusion flame produced higher NOx emissions than the
rich/lean staged diffusion flame. However, because of its simplicity, it
remains an attractive low N0X combustor concept. The influence of Reynolds
number on N0X levels in the lean flame suggests that NOx emissions could be
lowered by utilizing larger fuel tubes, perhaps approaching the levels achieved
by the staged diffusion flame.
262
-------
It is planned to investigate other combustor configurations including a
premixed backmixed simulated stirred reactor and a combination diffusion flame/
catalyst hybrid combustor. The zero dimensional stirred reactor is easy to
model. It will provide experimental feedback, for tha fuel nitrogen processing
kinetics code to be used in future prototype combustor design. The hybrid
system will input low £XN containing fuel rich diffusion flame exhaust into a
Pt/NiO cleanup catalyst prior to secondary burnout.
263
-------
REFERENCES
1.	Robson, F. L., W. A. Blecher and A. J. Giramonti. Combined - Cycle
Power Systems. EPA-600/2-76-149, U. S. Environmental Protection
Agency, Washington, D.C., 1976. p.359.
2.	Environmental Protection Agency Title 40, Chapter 1, Subcharter C,
Part 60 - Standards of Performance for New Stationary Sources,
Federal Register, Vol. 44, No. 113, June 11, 1979.
3.	Folsom, B. A., C. W. Courtney, T. L. Corley and W. D. Clark.
Advanced Combustion Concepts for Low Btu Gas Combustion. In:
Proceedings of the Third Stationary Source Combustion Symposium,
Vol. 2, Advanced Processes and Special Topics, EPA-600-17-79-050b,
February 1979.
4.	Folsom, B. A., W. D. Clark, C. W. Courtney, M. P. Heap, and W. R.
Seeker. Fuel Nitrogen Conversion - The Impact of Catalyst Type.
In: Proceedings - Fourth Workshop on Catalytic Combustion, EPA
Publication not yet in print.
5.	Dravo Corporation, Handbook of Gasifiers and Gas Treatment Systems,
NTIS Report No. FE-1772-11, February 1976.
6.	Folsom, B. A., T. A. Corley, M. H. Lobell, C. J. Kau, M. P. Heap,
and T. J. Tyson. Evaluation of Combustor Design Concepts for
Advanced Energy Conversion Systems. In: Proceedings of the Second
Stationary Source Combustion Symposium, Vol. 5, Addendum, EPA-60017-
77-073, July 1977.
7.	Folsom, B. A., C. Courtney and M. P. Heap. Environmental Aspects
of Low Btu Gas-Fired Catalytic Combustion. In: Proceedings -
Third Workshop on Catalytic Combustion, EPA-60017-79-038, February
1979.
8.	Klimisch, R. L. and K. C. Taylor, Ammonia Intermediacy as a Basis
for Catalyst Selection for Nitric Oxide Reduction. Environmental
Science & Technology 7 (2): 127-131, February 1973.
9.	Chu, E. K. and J. P. Kesselring. Fuel NOy Control by Catalytic
Combustion. In: Proceedings - Third Workshop on Catalytic
Combustion, EPA-60017-79-038, February 1979.
10.	Krill, W. V. and J. P Kesselring. The Development of Catalytic
Combustors for Stationary Source Applications. In: Proceedings -
Third Workshop on Catalytic Combustion, EPA-60017-79-038, February
1979.
264
-------
REFERENCES (cont.)
Fenimore, C. P. Effects of Diluents and Mixing on Nitric Oxide
from Fuel-Nitrogen Species in Diffusion Flames. In: Proceedings-
of the Sixteenth Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, Pa., 1976.
Folsom, B. A., C. W. Courtney and M. P. Heap. The Effects of LBG
Composition and Combustor Characteristics on Fuel NOy Formation.
In: Proceedings of the Twenty-Fourth International Gas Turbine
Conference, ASME, San Diego, Ca., 1979.
265
-------
Coal Input
Steam
Pressurized
Gaslfler And
Cleanup System
LOG
Turbine
Compressor
Air
Inlet
Gas Generator
Steam Turbine
Generator
Haste Heat
Boiler
Condenser
Feed-
Hater
Cooling Hater
Generator
Figure 1. Combines Cycle Power Plant
-------
looo
100
u
1000
¦O— txn
	 ADlABATlc FLAME
TEMPERATURE
LESS THAN t
300
100
ZOO
STOlCMIOMETRT (* THEORETICAL AIR)
Figure 2. Equilibrium ZXN and Adiabatic Flame Temperature for LBG.
-------
LBG
Fuel
Air
Stirred

Plug Flow




Plug Flow
- To Turbir
Rich Primary
Stage
Secondary Burnout
Stage
Figure 3. Rich/Lean Series Staged Combustor.
-------
REFORMER AIR
PRIMARY AIR
SECONDARY AIR
BYPASS
SAMPLE
TRAIN
AIR
FLOW
RESTRICTOR
TRIM
GASES
EMISSIONS
CONSOLE
WATER
HEPTANE
MODULAR
COMBUSTORS
CATALYTIC
REFORMER
Figure 4. Bench Scale Pressurized Test Facility.
-------
BENCH SCALE CATALYST
BENCH SCALE DIFFUSION FLAME
To Second Stage
M
o
Catalyst Wall
Thermocouples
Flash Back
Thermocouple
To Secondary
,1
Plug Flow
Module
' Catalyst Bed
Thermocouples
Graded Catalyst
Cells
Low Density
Insulation
Injector/Flow Stratghtener

Low Density
Insulation
Primary
Air
Preralxed LBG + Primary Air
LBG
Primary
Air
Figure 5. Primary Ignition Modules.
-------
O	NO
~	NH,
A	HCN
O	EXN or N0„
NHj In LBG = 500 ppm
Pt Catalyst
200 y
Approximate Full
Conversion
o
T.
O.
Q.
z
X
Ul
fl—8
lOfl
200
Percent Theoretical Air, Primary
100
Percent
O NO
300
A HCN
O EXN or N0„
NH, in LBG = 500 ppm
Pt/NjO Catalyst
200
Approximate
Full Conversion
o.
Q.
100
120
Primary
80	100
Percent Theoretical Air, Primary
Figure 6. Laboratory Scale Catalyst Comparison.
-------
BENCH SCALE
LABORATORY SCALE
O - "Ox- normal OPERATIC*!
# - N0X, BREAKTHROUGH
NHj in t.BG - 553 PPM
VARIABLE ADIABATIC FLAME
TEMPERATURE
O- N0„
MHj in LBR - 500 PPH
ADIABATIC FLAME TEMPERATURF. -
1473"K
M
•¦J
to
100
- 80
g
I
C
60 —
S
jj 40
ru
20 —

— '<00
3
H
300 §
K
§
3
200 «
a
o
0
S
— 100
Percent Theoretical Air - Primary
Figure 7. Pt/NiO Catalyst Scale-up.
-------
LABORATORY SCALE
BENCH SCALE
UNSTAGED
NH, In LOG - 471 PPM
O EXN, CIU = 0
Q EXN, CH„ = 5*
STAGED
Nil, In LBG = 553 PPM
% N0X, CIU = .6*
¦ N0X, CHh = Z.u
~ ~ ~
o
o
o
o
~
:
o
~
~
~
o
A	A	si-
~
~

o
"25	*5	EC	lo	Too	tA	do	rio"
~
180
PERCENT THEORETICAL AIR, PRIMARY
ZOO
Figure 8. Diffusion Flame Scale-up.
-------
•P*
X
o
z
80
60
£ 40
Ui
o 20
a
# N0X, 3/8 FUEL TUBE, Re = 90,000
O N0X, 3/4 FUEL TUBE, Re = 40,000
HH, In LBG = 553 PPM
100* CONVERSION = 430 PPM
400
300:
a.
IX
1:8
K
.O
ZOO
x
o
oc
100&
<
20
40
60	80	100	120
PERCENT THEORETICAL AIR - PRIMARY
140
160
Figure 9. Bench Scale Diffusion Flame: Effect of Reynolds Number.
-------
REYNOLDS NUMBER
I
16000
T
T
40000
24000
K3
in
100 _
o
z:
(J
K
UJ
80 —
60
40 _
20 _
NM, in IBG = 553 PPM
3/4 O.D. FUEL TUBE
T Primary * 120 msec
95* T.A, PRIMARY
Q N0X, CH„ = .fit
41 N0X, CH* - 2.IX
RICH/LEAH STAGED
- 400
300«
h*
35
>-
tt
O
- 200?
X
o
_ 100
40	60	80	100
C0MBUST0R PRESSURE (PSIG)
Figure 10. Bench Scale Diffusion Flame
REYNOLDS NUMBER
10(1 -
80
X
O
80
zc
z
az
UJ 40h-
5 _
£ W
ui
,	,	n
18000 25000	40000
NII, In IBG = 553 PPM
3/4 O.D. FUEL TUBE
T Primary = 90 msec
1501 T.A.
O N0X» CH» = .61
# N0X, CIU - 2. IX
LEAN, UNSTAGED
I
] -
o
UJ
5
&
I
X"
o
a:
a.
a_
<
40	60	80	100
C0MBUST0R PRESSURE, PSIG
Variable pressure/Reynolds Number,
-------
—I	,	1	1	1	1	1	r
NH, in LBG - 553 PPM	Q NO,. T Prfmry - 120 msec
PRESSURE = 100 PSIG
3/4 0.0. FUEL TUBE	^ T PHlMry = 250 msec
o
:
i1
20	40	60	80	100	120	140 160
PERCENT THEORETICAL AIR - PRIMARY
Figure 11. Bench Scale Diffusion Flame: Effect of Residence Time.
-------
RICH/LEAN STAGED
80* T.A. PRIMARY
600
O	N0X, Pt/NlO Catalyst
^	N0X, 3/8 0.0. Tube Diffusion Flame
—	—• — Full Conversion - Catalyst
¦ Full Conversion - Diffusion Flame
400
>-
oc
o
o.
d
><
200
1000
2000
NH, In LBG PPMV MET
4000
3000
Figure 12. Bench Scale Reactor Comparison: Effect of Dopant Level.
-------
Table I. Reformer-produced Low STU Gas Composition.
Measured	-	Equilibrium
Product 80 psig	120 psig	160 psig
NO (ppm) 2	1	0
NH^ (ppm) 44	50	66
HCN (ppm) 11	0
Q2 (2) 0.0	0.0	0.0
CO (2) 7A	7.1	7.3
C02 (5) a.a	3.7	7.0
(2) 0,2	0.3	0.0
H,0 (2) 27.0	27.1	23.4*
H2 (2) IS.9*	19.5*	17.7
C(S) (2) 0.0*	0.0*	0
(2) 36.6*	37.3*	39.2
Heating Value(Btu/ft3) 89.4	87.2	80.7
* By difference between measured output and input
reformer feed stock - heptane.
278
-------
CONTROL OF NOx AND PARTICULATES EMISSION
FROM SRC-II SPRAY FLAMES
By:
J. M. Beer, M. T. Jacques, S. Hanson, A. K. Gupta
Massachusetts Institute of Technology
Department of Chemical Engineering and the Energy Laboratory
Cambridge, Massachusetts 02139
W. Rovesti
Electric Power Research Institute
Palo Alto, California 94303
279
-------
ABSTRACT
Experiments were carried out with SRC-II fuels in a laboratory scale
laminar flow reactor and in the 4ft x 4ft MIT Combustion Research Facility.
In the laminar flow reactor monosize droplet arrays are pyrolyzed in an inert
atmosphere at variable temperatures and the time resolved evolution of the
fuel nitrogen is determined. This information is needed for the development
of an NC>x control strategy by staged combustion. Parallel with the laboratory
studies,experiments are carried out with SRC-II liquid fuel sprays in unstaged
and staged turbulent diffusion flames in a thermal environment similar to that
in a utility boiler. The N0x and particulates emission is determined in un-
staged flames for the effects of the rate of fuel/air mixing, air preheat and
the quality of atomization. The results show that low overall excess air in a
long slowly mixing turbulent flame with a low degree of swirl in the air for
flame stabilization, can reduce the NC>x emission level to about 250 ppm from
550 ppm obtained for high intensity fast mixing flames. A significant further
reduction of N0^ can be achieved with staged combustion by physically separated
stages. In these latter experiments a computer analysis of the fuel nitrogen
conversion is used to guide the experiments carried out using the MIT Combustion
Research Facility. The distributions of N0^ and particulates in the flames are
determined for the effects of the primary stage fuel/air ratio, temperature and
atomization quality. The mixing in both the fuel rich and lean stages and the
heat extraction along the flame is closely controlled in these experiments. Re-
sults show that N0^ (3% 0^) levels below 100 ppm can be achieved without exces-
sive emission of particulates. The experimental data show the same general
trends for NO emission as a function of fuel equivalence ratio, as that pre-
dieted by the computer model. While the NO^ and particulate levels achieved in
these studies are most encouraging, it is the development of a strategy of N0x
and particulate emission control, that is the main objective of the present
study. The emission levels achieved in this study can be considered as lower
280
-------
bounds for conditions in which the mixing in the flame is controlled more
closely than can be achieved in practical utility boiler combustion chambers.
281
-------
SECTION 1
INTRODUCTION
It is generally accepted that staged combustion techniques currently offer
the most attractive strategy for the control of NO^ emission from the combustion
of high nitrogen content fuels. Several experimental studies [1-3] have demon-
strated the potential of these techniques for achieving environmentally accept-
able emissions from the combustion of the lighter coal derived liquid fuels.
However, there is still some concern over the potential for soot emission from
the heavier, highly aromatic, high C/H ratio coal derived liquids when fired
under staged combustion conditions. The high temperature, fuel-rich flame con-
ditions required to suppress conversions of fuel-nitrogen to	result in the
formation of a significant amount of soot. Complete burn-out of this soot must
be achieved in the fuel-lean flame zone by ensuring good mixing and sufficient
residence time at elevated temperatures.
Consequently, it is of considerable interest to establish both theoretical
and practical lower limits of NO and soot emissions which can be achieved by
x	J
employing staged combustion techniques for the utilization of coal derived
liquid fuels.
A research program is underway at MIT under EPRI sponsorship. The program
objectives are to identify and quantitatively describe the critical steps in-
volved in the processes of fuel-nitrogen evolution and conversion in order to
develop a design strategy for minimizing NO^ and particulate emissions from the
combustion of coal derived liquid fuels. The research complements a fundamental
study on the kinetics of fuel nitrogen evolution from pyrolysing fuel particles
and droplets sponsored by EPA [4] and on fuel nitrogen conversion in fuel rich
flames, under the sponsorship of DOE [5].
The research presented in this paper involves a combination of laboratory
282
-------
scale experiments on the pyrolysis of single droplet streams, computer modeling
of the fuel-nitrogen conversion reactions, and an experimental study using the
pilot-scale MIT Combustion Research Facility on the effect of staged combustion
on the NO^ and particulate emissions from a blended SRC-II fuel oil (2.9 parts
by weight middle distillate to 1 part heavy distillate).
283
-------
SECTION 2
DROPLET PYROLYSIS STUDIES
The important fuel-nitrogen conversion reactions, which form the basis of
the staged combustion N0x control strategy, i.e., those leading to the forma-
tion of R , are dominated by gas phase reactions involving the nitrogen con-
taining fuel vapors. Consequently the rate of vaporization and yield of fuel-
nitrogen species from the parent droplet represent a critical first step in
the fuel-nitrogen conversion process. Fuel-nitrogen present in the form of
high molecular weight heterocyclic compounds may concentrate in the heavy
liquid fractions or become locked-up in solid residues formed during the pyrol-
ysis of the fuel [1]. Any fuel-nitrogen which escapes unreacted from the fuel-
rich flame zone will be readily oxidized to N0X in the fuel-lean flame zone.
It is, therefore, of considerable interest to understand the factors that
determine the rate at which fuel-nitrogen evolves from liquid fuel droplets
during pyrolysis under conditions of rapid heating.
2.1 The Experimental Apparatus
The experimental apparatus and the measurement techniques were developed
within the scope of an EPA grant to MIT entitled: "Reduction of Pollutant
Formation in Coal Particle and Liquid Fuel Flames;" brief description is given
in the following.
For the study of fuel-nitrogen evolution from liquid fuel droplets during
pyrolysis and oxidation, a laminar flow reactor (Fig. 1) was designed with pro-
vision for feeding a monosize droplet array into a laminar gas stream and the
capability of intercepting thp droplets, after a chosen residence time, by a
fast quenching sampling probe. Optical access to the pyrolyzing, burning oil
droplet array through quartz windows along the full length of the uniformly
heated section of the reactor enables photographic and schlieren-interferometric
284
-------
measurements to be made of droplet size variation, droplet velocity and the tem-
perature distribution in the droplet boundary layer.
The objective of carrying out experiments with droplet arrays of heavy re-
sidual fuel oils, posed significant experimental problems concerning the fuel
preparation and droplet generation and also the sampling.
A vibrating orifice-type droplet generator was chosen for the introduction
of fuel droplets into the reactor. A monosized droplet array is injected ver-
tically downward into the furnace through a small (50 or 100 pm) orifice mounted
in a piezo-electric vibrator. The vibrator is driven by a variable frequency
signal generator. The variation of oil flow rate, vibrator frequency and ori-
fice size permits the control of droplet size and interdrop distance. The
droplets are injected into a laminar co-flowing gas stream; in recent studies
helium was chosen to give a chemically inert environment for the vaporizing-
pyrolyzing drops.
The sampling probe inserted through the base of the furnace is aligned with
the drop generator by means of a small laser.
The sampling probe takes the total gas flow, the residual droplet array and
the vapor.
A transpiring type of probe following the conceptual design by Mims [7],
combined with a cascade impactor was used.
2.2 Experimental Results
Results of experiments carried out with the pyrolysis of 150 ym SRC-II
droplet streams in helium at furnace temperatures of 975, 1050, 1160 and 1270 K
are presented in the form of the percentage of fuel bound nitrogen evolved
versus mass loss by vaporization, in Fig. 2. The data points are calculated
from the weight loss due to vaporization between the droplet generator and the
sampling position and from the nitrogen concentration of the liquid residue
collected at that sampling position. Equilibrium distillation data are also
given in Fig. 2; these were determined at atmospheric pressure by slow heat-
ing of the SRC-II fuel. The equilibrium distillation curve is close to a 45°
line showing that under slow heating conditions the nitrogen is nearly uni-
formly distributed amongst the different boiling fractions of the fuel. Under
conditions of rapid heating, however, the rate at which nitrogen is released
285
-------
relative to that of vaporization becomes dependent on the rate of heating
of the fuel droplet.
In the case of equilibrium distillation under well mixed conditions,
the evolution of a particular component is given by the following
thermodynamic expression,
Xf rAHvfl
*i ,AHvfl 1
bi
where y^ is the vapor phase mole fraction of i
X^ is the liquid phase mole fraction of i
Pa is the ambient pressure
AHv is the enthalpy of vaporization
R is the ideal gas constant
T is the liquid temperature
is the normal boiling point for pure i
The vapor phase contribution of the component in question is
influenced by its liquid phase mole fraction and temperature. The
overall composition determines the temperature and competition with
other species of differing relative volatilities can shift the
contribution of any particular component in its attempt to satisfy the
constraints Exi=l and Zy-£=1. Thus, a fuel containing a single
nitrogenous compound may, due to the influence of the other species,
produce a nitrogen distillation curve similar to that presented for the
SRC-II middle fraction. For the SRC-II middle fraction, it is known
that there are many nitrogen containing species and what is observed is
286
-------
the summation of the nitrogen evolved by all of them. The utility of the
distillation curve is not that it identifies what nitrogenous species
are present, but that it represents one extreme of vaporization from
droplets where diffusive mixing is rapid relative to weight loss by
vaporization.
When a 150ym droplet of SRC-II middle fraction is exposed to a hot
inert environment, the light boiling fractions, which contain little
nitrogen, are vaporized from the surface. This depletion causes a
concentration gradient to be established which permits more light
material to travel to the surface. The more rapidly one vaporizes the
droplet, the steeper are the gradients and the greater the light fraction
contribution in the volatile fraction. Thus one observes an initial
retention of fuel nitrogen with increasing mass vaporization rate. What
has been said so far is true up to the point where mass vaporization is
so rapid that diffusion cannot keep up with the removal of low boiling
fractions. Nitrogen containing species concentrated at the surface of
the droplet now contribute to the mass evolved and the more rapid the
vaporization thereafter the earlier the nitrogen will evolve. Needless
to say, the description just given has been greatly simplified. The
effects of changing surface temperature and relative volatility with
changing surface composition are implicit in the discussion. Liquid
phase reactions, superheating within the droplet and other effects are
recognized to be present but are not considered.
287
-------
SECTION 3
KINETICS OF FUEL-NITROGEN CONVERSION-MODELING STUDIES
Nitrogen oxides emission from combustion processes originates from two
main sources: the fixation of atmospheric nitrogen at high combustion tem-
peratures ("thermal-NOx") and the oxidation of fuel bound nitrogen ("fuel
NO^"). When burning high nitrogen content fuels (N > 0.2%), an increasing
proportion of the N0x emitted will be of "fuel NO" type. In contrast to
"thermal NO" the rate of formation of which is a strong function of tempera-
ture, the formation rate of "fuel NO" is only slightly dependent upon tempera-
ture but increases markedly with increased oxygen concentrations in the flame.
This is because the conversion efficiency of fuel nitrogen to NO is thought to
depend on the rate of competing reactions involving nitrogen fragments to pro-
duce NO or	The exact mechanism of fuel nitrogen conversion is not known,
but there is now a significant body of knowledge for general conclusions to be
drawn with respect to the development of a design strategy for reducing NO
emissions from combustion processes when burning high nitrogen content fuels.
Details of the chemistry of the formation and destruction of "fuel-NO" are
discussed in a recent survey by Levy et al. [8], Figure 4 illustrates the
various pathways of fuel-nitrogen during pyrolysis and oxidation reactions in
a flame. In liquid fuels the fuel nitrogen evolved during vaporization con-
verts in fast reactions to HCN, which will react further via oxycyanide inter-
mediates to form amines. At every step of the reaction sequence, if oxidants
such as OH or 0 are available NO will be formed. However, in the absence of
other oxidants the nitrogenous species react either with NO, or with each
other to form N-. These latter reactions form the basis of NO control by
£>	X
staged combustion techniques in which the conditions in the fuel-rich zone
favor the conversion of the fuel-nitrogen species to ^ rather than NO^. In-
formation on the thermodynamics and chemical kinetics of the above mentioned
288
-------
reaction sequence suggests that such a control strategy requires the pyr'olysis
of the fuel vapor under fuel-rich conditions at sufficiently high temperature
to ensure rapid conversion of the fuel-nitrogen to N2- To achieve complete
combustion and to minimize soot emission, the fuel-rich zone has to be followed
by the introduction of additional combustion air. In this fuel lean flame
zone, any fuel-nitrogen fragments not converted to N2 will be rapidly oxidized
to form NO^. Consequently an important parameter is the sum of bound nitrogen
species present in the combustion gases leaving the fuel-rich flame zone.
The bound nitrogen species considered include all nitrogen species present
in the combustion products, i.e., N, NH, NO, ^0, HCN, CN, NCO and HNCO with
the notable exception of N2« Hence a decrease in total bound nitrogen species
represent conversion of fuel-nitrogen to N2-
A simplified computer model, incorporating the kinetics of gas phase nitro-
gen reactions under plug-flow conditions developed by Taylor [9] was used by
Farmayan [10] to identify the influence of important fuel-rich stage parameters
on the sum of bound nitrogen species within the flame. The kinetic mechanism
contained in the combustion model consists of reactions (shown in Table I)
which are believed to dominate the conversion of bound nitrogen species under
high temperature fuel-rich combustion conditions. The initialization of the
kinetic calculations was obtained by assuming that the hydrocarbon/air rections
are partially equilibrated, yielding a mixture of CO, C02, H2, H20, 0, OH, and
0^. In addition, the fuel nitrogen was assumed to be instantaneously pyrolyzed
to HCN and CN which are also in partial equilibrium with the hydrocarbon com-
bustion products.
This model was employed by Farmayan [10] first in an extensive computer
study to determine the influence of fuel-rich zone parameters, namely, fuel
equivalence ratio, adiabatic flame temperature and residence time, on the rate
of fuel-nitrogen conversion during the combustion of a high-nitrogen content #6
fuel oil (0.73 wt% N).
Figure 5 shows results of this computer study on the high-nitrogen content
J1/6 fuel oil in terms of the predicted sum of bound nitrogen species mole frac-
tion as a function of the fuel-rich zone residence time. Data are presented
for three fuel equivalence ratios, 4> = 1.4, 1.6 and 1.8 and air inlet tempera-
tures 298K, 500K and 700K. This latter variable influences the adiabatic flame
289
-------
temperature which is shown in parenthesis for each condition. These data
clearly show that the bound nitrogen species mole fraction can be reduced by
at least an order of magnitude below the initial value within residence times
of practical significance (approx. one second). Furthermore they show that
the inlet air temperature, and hence the adiabatic flame temperature, has a
significant influence on the extent of fuel-nitrogen conversion. At a fuel
equivalence ratio of 1.4 increasing the inlet air temperature from 298K to
700K decreases the initial rate of bound nitrogen decay and would result in
higher NC>x emissions. This behavior can be explained by the fact that at
 = 1.6 and 1.8 the equilibrium mole fraction of
bound nitrogen species is much lower than that in the initial mixture1 and
increasing flame temperature increases the rate at which bound nitrogen spec-
ies are converted to N^. At  = 1.6 where the
bound nitrogen species decay rate and final values after 1 second residence
time are almost identical. The slightly higher final value at <}> = 1.4, of
bound nitrogen species mole fraction for the SRC-II is attributed to the
higher initial value resulting from the higher adiabatic flame temperature.
Figure 7 further illustrates the sensitivity of the rate of decay of
290
-------
bound nitrogen species to adiabatic flame temperature. At <|> = 1.6 and an
inlet air temperature of 298K the SRC-II exhibits a much faster rate of decay
than the #6 fuel oil, indicating that reduced residence times are possible at
the lower air preheat levels for the same extent of fuel-nitrogen conversion.
The general conclusion from these comparative studies are that both fuels be-
have in a very similar manner, particularly with respect to the extent of
fuel-nitrogen conversion at residence times of = 1 sec.Consequently it was felt
that the computer model results contained for #6 fuel oil could be used to
guide the choice of experimental parameter values for SRC-II studies.
Of particular interest is the influence of fuel equivalence ratio on the
extent fuel nitrogen conversion. Figure 8 shows the results obtained for the
#6 fuel oil and it can be clearly seen that the total bound nitrogen mole
fraction achieves a minimum with a fuel equivalence ratio within the range of
1.6 - 1.8 depending upon air preheat level and residence time. Furthermore
the minimum values of total bound nitrogen which are attainable are, in terms
of equivalent NO of the order of lOppm (at 3% 0 ). The minima of the curves
X	/
representing the sum of bound nitrogen species is seen to shift towards higher
fuel/air ratios with increasing air preheat temperature. The air preheat tem-
perature is given for characterization of the flame temperatures because these
latter are varying along the curves as a result of the variation of the fuel/
air ratio. The absolute value of the minimum of bound nitrogen species de-
creases with increased residence time indicating the controlling influence of
chemical kinetics.
291
-------
SECTION 4
PILOT SCALE FLAME STUDIES
The experimental combustion program involved the use of the MIT Combustion
Research Facility (CRF) which was designed specifically to facilitate detailed
experimental investigations of large turbulent flames. The MIT-CRF consists of
a 4 ft x 4 ft cross-section, 30 ft long combustion tunnel equipped with a
single burner having a 3MW, multi-fuel firing capability. Figure 9 shows the
general arrangement of the CRF as used in the present SRC-II flame studies.
The burner consists of a variable swirl generator, of the 'movable - block'
design developed by the International Flame Research Foundation at Ijmuiden.
The fuel gun and atomizer nozzle are inserted through the swirl generator for
location at the burner throat. The furnace chamber comprises a number of in-
dividual, 1 ft wide, water-cooled, refractory lined or bare metal interchange-
able sections, an arrangement of which permits variable heat sink for control
of heat extraction along the length of the flame. This variable heat sink per-
mits simulation of the thermal environment of a wide range of industrial
flames. In the present study, in which interest is in utilization of SRC-II
fuels in utility boilers, the condition needed to ensure chemical similarity
between utility boiler flames and MIT-CRF flames required the use of hot wall
conditions.
Experiments were performed using two SRC-II fuel types. The blend of 2.9
parts of middle distillate to one part of heavy distillate (2.9/1) and the un-
blended heavy distillate. The properties of these fuels are given in Table II.
Both unstaged and staged combustion conditions were investigated. The unstaged
experiments were designed to provide baseline data on NO^ and particulate emis-
sions from these two SRC-II fuel types.
For the discussion of the arrangement used for staged combustion studies
292
-------
we refer to Fig. 9. Primary zone air was admitted through the swirl generator/
burner assembly and secondary zone air was introduced perpendicular to the
burner/furnace axis at an axial distance of 8.5 ft from the burner, through
two sets of injector nozzles arranged directly opposite each other and mounted
in the furnace side walls as shown in Fig. 10. Each nozzle injection assembly
consists of five interchangeable stainless steel pipes which extend to the
inner surface of the refractory lined furnace wall, thus permitting air injec-
tion through single of multiple nozzles to provide a wide range of air injec-
tion velocities.
Measurements of temperature and gaseous species concentrations (N0x,
CO, CO^) were made along the axis of the unstaged flames by the use of water
cooled probes which are inserted through rectangular slots centrally located
in the side walls of each furnace section. Solids and NO concentrations
x
and Baccarach smoke number were measured in the exhaust gas at the exit of the
experimental section.
4.1 Unstaged Flame Studies
Baseline data were obtained for two types of unstaged flames: a long,
slow mixing low swirl flame, typical of corner fired boiler units; and a short,
fast mixing, high swirl flame characteristic of front or side wall fired units.
The effects of design and operating variables on N0x and particulate emissions
were investigated for two types of atomizer, a twin fluid, steam assisted Y-
jet atomizer and a steam assisted Sonicore atomizer; two levels of air preheat
500 and 850°F and two burner swirl numbers 0.53 (low swirl) and 2.70 (high
swirl). (The swirl number is the value of a nondimensional group characteriz-
ing the degree of rotation in the jet).
A total of twelve flames were studied, six of which were obtained using
the SRC-1I 2.9/1 blend of middle to heavy distillate and six using the SRC-II
heavy distillate. The input conditions for these twelve flames are summarized
in Table III. The MIT-CRF was operated with hot (refractory lined) walls, a
total of 15 wall sections were used giving a combustor length of 15 ft. The
thermal input was maintained constant at 1MW and excess air was controlled at
2-10% to give 0.5-1.9% 0^ in the flue gas. Furnace wall temperatures were
typically 2200°F and the flue gas exit temperature from the combustion chamber
was =2400°F.
293
-------
The SRC-II 2.9/1 blend was fired without heating the fuel, the tempera-
ture at the atomizer being 100°F, with a corresponding viscosity of AO SSU.
The SRC-II heavy distillate was heated using in-line electric heaters, to
maintain a temperature of 210°F at the atomizing nozzle to maintain the vis-
cosity at 40 SSU. The axial NO profiles shown in Fig. 11 are characteris-
tic of the two flame types studied. Both profiles are relatively flat in-
dicating that most of the NO^ is formed very early in the flames. There is
a significant increase in NO^ emission for the high swirl flame, in this
case from 300 ppm, in flame #3, to 550 ppm in flame #4, (both at 3% O2).
This increase in N0x emission is due entirely to the change in flame aero-
dynamics resulting from the increase in swirl of the combustion air. It is
not possible to distinguish the fraction of NO^ produced from fuel-nitrogen
in these flames from that produced as thermal NO . In addition to the ex-
pected increase in fuel-nitrogen conversion to N0x due to faster mixing of
fuel and air near the burner there is possibly an increase in thermal NO
formation due to an increase in peak flame temperature in the high swirl
flame.
Solids emissions were measured by two separate techniques. Firstly,
the flue-gas was sampled intermittently using a Baccarach Smoke Tester to
obtain the smoke number. Samples were taken at 5 minute intervals during
steady-state conditions for each flame. Secondly, a high-volume flue^-gas
sampling system was employed which consisted of a 1" diameter quartz glass
tube placed in the exhaust gas, through which the flue gas was aspirated.
A glass fibre filter was used to capture all particulates in the flue gas
above 0.1 ym.
Solids emissions data are presented in Table 4 together with those of
N0x emission. It can be seen that the concentration of solids in the flue
gas is well below the current solids emission standards for all the flames
studied, and that in general the SRC-II heavy distillate gave higher solids
concentrations than the 2.9/1 blend.
The variability in Smoke // does not correlate well with the measured
solids concentrations, and it is felt that this is due to the fact that most
flames'were run at close to minimum possible excess air levels, where periodic
fluctuations in fuel or air flow rate resulted in puffs of smoke being emitted.
Smoke #'s measured during these periods gave higher values than the mean,
294
-------
whereas the continuous sampling for solid concentration averaged these' inter-
mittent smoke emissions over the entire sampling period.
An attempt was made to characterize the onset of smoke formation for the
two types of SRC-II fuel. Test conditions were chosen which would tend to
minimize smoke formation, i.e., high swirl and high air preheat using the
Sonicore atomizer. Smoke numbers were measured as the 0^ concentration in
the flue-gas was reduced, until CO appeared in the flue-gas. The results of
these brief tests are shown in Fig. 12 where it is clear that the heavy dis-
tillate has a considerably higher tendency to form smoke. Under the condi-
tions of these experiments excess air levels greater than 5% are required to
reduce the smoke number below a value of 2. The 2.9/1 blend exhibited less
of a tendency to form smoke under these high intensity combustion conditions,
and excess air could be reduced to zero without the smoke number exceeding a
value of 2.
4.2 Staged Combustion Flame Studies
The results obtained from computer modeling of the phase reactions, to-
gether with experimental data obtained from a previous study on a high-
nitrogen content #6 fuel oil [11], indicated that the design and operating
variables which influence N0x emissions under staged combustion conditions
are:
o fuel-rich zone fuel equivalence ratio  .
b
o fuel-rich zone mixing pattern as influenced by burner air swirl
number, S.
o flame temperature as influenced by combustion air preheat.
Throughout the staged combustion experiments the following operating condi-
tions were maintained constant:
o
thermal input - 1MW (3.4 x 10^ Btu/hr)
o overall excess air -10% (2% O2 in the flue-gas)
o furnace wall temperature -2200°F (exhaust gas temperature -2400°F)
o atomization conditions - a twin fluid, Y-jet steam atomizer.
The effects of varying 
-------
was found to be the minimum swirl number required to maintain good
flame stability over the range of fuel equivalence ratios being studied.
Five staged combustion conditions corresponding to = 1.20, 1.30, 1,44,
1.62, and 1.76 were examined. The results obtained are shown in Fig. 13
where it can be seen that the concentration of N0x in the flue gas changes
from the unstaged value of =; 290 ppm @ 3% 0^ to a minimum of = 95 ppm @ 0^
as f}>k is increased to between 1.3 and 1.4. Increasing ^ beyond a value
of 1.4 resulted in a slight increase in NC>x emissions.
Smoke //, as measured with the Baccarach smoke tester, is seen to in-
crease significantly as the burner fuel equivalence ratio is increased
beyond , » 1.3. However, it is felt that smoke number measurements should
b
be used only to indicate trends in solids emissions, since individual mea-
surements made under constant furnace operating conditions can exhibit
considerable scatter.
The influence of swirl on N0^ and smoke emissions was investigated by
maintaining staged combustion conditions with = 1-3 (minimum observed
N0x) and varying the swirl number from 0 to 2.7. Air preheat was maintained
at 100°F. Flue-gas NO concentration and Baccarach smoke number were mea-
x
sured at six swirl settings.
The results are shown in Fig. 14, and it is seen that a significant
change in NO emissions occurs as the swirl number is changed from 0.25 to
x
0.45. NOx emissions changed from 260 ppm to 90 ppm at 3% 0^, and smoke num-
ber increased. The high N0x values at low swirl settings are very similar
to the unstaged values, and it appears as through fuel/air mixing is suf-
ficiently influenced in the primary zone under these conditions as to cancel
out the effects of staging at swirl numbers less than 0.45.
Visual observation of the flame clearly showed that the change in swirl
from 0.25 to 0.45 corresponded to a change from a "lifted" to a "non-lifted"
flame. The "lifted" flame was stabilized 2-3 ft from the spray nozzle while
the "non-lifted" flame was stabilized close to the spray nozzle. It is not
clearly understood why the position of the flame has such a strong influence
on NO emission under staged combustion conditions,
x
The effect of air preheat temperature, upon N0x and smoke emissions was
measured for constant values of , =1.3 and S = 0.65 which were identified
b
296
-------
as giving minimum NO emissions and low smoke levels. Increasing air pre-
heat resulted in a decrease in both NO^ and Smoke # as shown in Fig. 15.
NO^ emission was reduced by-20%, when air preheat was increased from 100°F
to 800°F. The fact that increasing air preheat has resulted in the reduc-
tion of flue gas N0x concentration is an indication that higher temperatures
in the fuel rich zone were instrumental to lowering the sum of bound nitro-
gen species entering the lean stage, and that little or no incremental
thermal NO was formed as a result of higher air preheat. This is likely
due to the fast mixing at the introduction of the high velocity secondary
air stream resulting in a lowering of the mixture temperature before the
lean stage combustion takes place.
The trend of decreasing Smoke # with air preheat was verified by the
measurement of flue-gas particulate concentration using a high-volume sam-
pling probe/filter arrangement. Particulate concentration was measured at
only two air preheat levels; and the results are given below:
Air Preheat	Particulate Concentration
°F	lb/106 BTU
260	.32 x 10~2
800	.28 x 10"2
It is encouraging to note that these particulate concentration levels ob-
tained with staged combustion conditions are well below the emission stan-
dard and are of the same order as those for single stage combustion.
297
-------
SECTION 5
CONCLUSIONS
5.1	Drop Tube Studies
o The extent of nitrogen evolution depends upon the rate of vapor-
ization.
o Under typical droplet heating conditions, the initial products
of vaporization are deficient of nitrogen,
o The nitrogen poor product, as a fraction of the total mass of the
fuel droplet, is increasing with increasing heating rate up to a
limit above which this trend is reversed,
o No carbonaceous residue is formed by the rapid heating of 150 ym
droplets.
5.2	Computer Modeling Studies
o A model to predict total bound nitrogen species as a function of
fuel rich stage fuel equivalence ratio, temperature and residence
time was developed,
o The model shows that the fuel equivalence ratio of the fuel rich
stage is the primary variable controlling fuel-nitrogen conver-
sion and that the process is chemical kinetically limited,
o The model exhibited the ability to obtain significant reductions
in bound nitrogen species for residence times (less than one sec-
ond) typical of many utility boilers and industrial furnaces.
5.3	Combustion Research Facility Experiments
5.3.1 . Single Stage Combustion Studies
o NO emission changed from 500 ppm (3% 0.) for a highly swirling jet
X	•
flame to about 250 ppm (3% O2) for a long turbulent jet diffusion
298
-------
flame. Solids emissions were well below existing standards for
both the blend and heavy distillate fuel.
Staged Combustion Studies
NO levels were reduced to below 100 ppm (at 3% 0 ) by staged
X	z
combustion.
NO reduction was most strongly affected by fuel rich stage fuel
equivalence ratio. Minimum N0x emissions occured in the range of
k = 1.3-1.5 in agreement with computation.
Swirl affected N0x emissions only so far as it modified the flow
pattern. For values of swirl number near the critical (onset of
recirculation) N0x emission was sensitive to flame stability.
Both staged and unstaged flame studies demonstrated stepwise in-
crease in NO^ emission when the flame "lifted" upon reduction of
swirl degree.
Air preheat, increasing air inlet temperature from 100°F to 800°F,
reduced NO emission by approximately 20% and is in agreement with
computation.
The Bacearach Smoke number increased with increasing fuel equiv-
alence ratio and increasing burner swirl number. Nevertheless,
at optimum staged combustion conditions (
-------
REFERENCES
Downs, W., and A. J. Kubasco, "Characterization and Combustion of SRC-II
Fuel Oil," EPRI Report FP-1028, June 1979.
Piper, B. F., et al., "Combustion Demonstration of SRC-II Fuel Oil in a
Tangentially Fired Boiler," EPRI Report FP-1029, May 1979.
Muzio, L. J., J. K. Avand and W. C. Rovesti, "Combustion and Emission
Characteristics of SRC-II Fuels," Western States Section, The Combustion
Institute 1980 Spring Meeting, April 1980.
Be^r, J. M., A. F. Sarofim, S. Hanson and A. K. Gupta, "High Temperature
Pyrolysis of Oil Droplet and Coal Particle Streams," EPA Fundamental
Combustion Workshop, Newport Beach, CA., January 1980. Grant No. R805552-01.
"Control of N0x Emissions from Combustion of Fuels Derived from Shale and
Coal." DOE Contract EX76-A-A-01-2205 T036 at M.I.T.
Martin, G. B., and E. E. Berkau, "An Investigation of the Conversion of
Various Fuel Nitrogen Compounds to Nitrogen Oxides in Oil Combustion,"
AIChE Symp. Ser. Air Pollution and Its Control, 68 (1972).
Mims et al., "Laboratory Studies of Trace Element Transformations during
Coal Combustion." Paper presented at the National 87th AIChE Meeting,
August 1979.
Levy, J. M., J. P. Longwell and A. F. Sarofim, "Conversion of Fuel-
Nitrogen to Nitrogen Oxides in Fossil Fuel Combustion: Mechanistic Con-
siderations." Report submitted to the Energy and Environmental Research
Corporation by the M.I.T. Energy Laboratory, April 1978.
Taylor, B. R., Sc.D. Thesis (to be submitted) Massachusetts Institute of
Technology, Cambridge, Mass., (1980).
300
-------
10.	Farmavan, W. F., M.Sc. Thesis, "The Control of Nitrogen Oxides Emissions
by Staged Combustion," Massachusetts Institute of Technology, Cambridge,
Mass., April 1980.
11.	Jacques, M. T., and J. M. Beer, "Experimental Studies on the Conversion
of Fuel-Nitrogen in Staged Combustion." Paper presented at the Italian
Flame Day Meeting, Naples, June 1980.
301
-------
FUEL
I'TgfgyHio*
CASING
OBSERVATI ON
WINDOW
INSULATION'
DUCT GAS *
WITHDRAWAL
.DUCT GAS
INTRODUCTION
GRAPHITE
-HEATING
ELEMENT
¦QUARTZ
DUCT
¦pTi»?8fEL°
~
S AMP LING
PROBE
FIGURE 1: THE EXPERIMiNTAL SYSTEM
302
-------
100
975 K
1050 K
1160, K
1270 K
90
30
CO
2 70
60
UJ
/~
^ 30
to
o
^ 20
0
0 10 20 30 40 50 60 70 80 90 100
WEIGHT LOSS, %
FIGURE 2 : NITROGEN EVOLUTION FROM 150m DROPLET ARRAY OF
SRC II (MIDDLE DISTILLATE)
303
-------
100
90
&>«
80
CO
M 70
O
	I
60
h-
ai
0 50
XlI
40
^ 30
CD
CD
^ 20
i—
i—i
10
0
0 10 20 30 40 50 60 70 80 90 100
WEIGHT LOSS, %
FIGURE 3 : NITROGEN EVOLUTION FROM THE EQUILIBRIUM
DISTILLATION OF SRC II (MIDDLE DISTILLATE)
AT ATMOSPHERIC PRESSURE
304
-------
HETEROGENEOUS
HOMOGENEOUS GAS PHASE REACTIONS	REACTIONS
_A.
NO - CO
REACTION
(CATALYZED)
OXIDATION
FIXATION
OF N2 BY
HYDROCARBON
FRAGMENTS
NO -
CARBON
REACTION
CYANOGENS
(HCN, CN)
ZELDOVICH
MECHANISM
NOx
NOx
AIR
OXYCYANOGENS
(OCN, HNCO)
FUEL
DROPLET
OR
PARTICLE
AMMONIA SPECIES
VOLATILES
(HETEROCYCLIC
NITROGEN COMPOUNDS)
RESIDUE
(CARBON,
NITROGEN COMPOUNDS)
Figure $. The formation of nitrogen oxides in fossil fuel
combustion: mechanistic pathways.
305
-------
inlet Air Icmp
 = 1.4
U
_l
o
5
CO
LJ
U
Li
CL
m
X
o
z
2
Ql
CL
Z H-
LI 2
O H
° <
a §
»- —
— z>
2 a
LJ
O
2	Z
3	—
o
dJ
L.
O
2
3
I/)
l/>
Z
o ~
. <\j
o°
<
DC ^
Li- CO
o
6*
o
3 r
2
I -
.298 K ( 2065K )
/ ,500 K < 2188 K)
J j 7 00 K ( 2 3 0 8 K )
29 8 K ( 1915 K )
500 K ( 2036 K )
700 K ( 21 59 K )
298K (1771 K )
500 K (I892K)
700 K ( 20I5K )
12	3	4
RESIDENCE TIME ( SEC )
Figure 5 Bound Nitrog
-------
UJ
o
2
t/>
U
(\J
o
.0
o
ro
x
O
2
U
UJ
0- 2
l/> CL
_ CL
bJ
o
O
cc
J-
z
3
o
CD
L.
O
2
3
CO
O
O
2
U
_J
3
o
u
2
o
»-
u
<
en
2 -
3 r~
0=1.4 Tair=500 K
SRC # 2.9/1
^High - N * 6
1 	I	I	I	L
0.5
1.0
4> = 1.6 T0ir = 500 K
SRC -I 2.9 / 1
High-N * 6
1
0	0.5	1.0
RESIDENCE TIME (see)
Figure 6 Com'porison of SRC # (2.9/1 ) and High
Nitrogen Content Computer Model
Predictions for Effect of 0
307
-------
(M
o
(f> .1.6 T oir = 700 K
— CO
o
Z
w
c#	x
u	O
O	Z
£ £
a
c a
o
D> -H
O C
t- a
.t: o
2	>
"5
x> cr
c Id
D
o	c
CD.	-
_	c
o .2
+¦»
p u
t o
3	i_
to u.
C7V
o
Figure 7
SRC-H 2.9/1
«^High-N 4t 6
J	I	I	I	I	L
7
0.5
1.0
0 = 1.6 T oir = 298 K
SRC II 2.9/1
High - N St 6
/
0.5	1.0
RESIDENCE TIME (SECS)
Comparison of SRC-H ( 2.9/1) and
High Nitrogen Content *6 Computer
Model Predictions for Effect of Air
Preheat
308
-------
z

O

Y-

u

<

CE

L.

Ld

-J

o

2



i/i

LJ
o
U

Ld

CL
5*
i/>
CO
z

u
o
O

O
X
cc
o
1—
z
z

Q
z
z>
PPM
o

CD
1-
THE
z
LJ
-J
<
L.
>
o
Z>

O
2
u
z>
i/)
z
w

o

o

o

-J

Residence Time =0.5 Sec
3 -
inlet air temp.
I
1.3	1.5	1.7
3 !_ Residence Time =1.0 Sec
1.9
-inlet air temp.
700
Residence Time-3.0
inlet air temp
700
1.3	1.5	1.7	1.9
FUEL EQUIVALENCE RATIO
Figure 8 Bound Nitrogen Species Mole Fraction
vs Fuel Equivalence Ratio
309
-------
45-
Combustlon
Air inlet
179
65
49
179
Burner
txper imentol
Sect ion
Af terburnc
Sectlon
Cxhaust
Sec tIon
Figure 9. The MIT-Coni>ustion Research Facility
-------
SECONDARY ZONE
AIR SUPPLY
PRIMARY ZONE
AIR SUPPLY
u>
	 u
NOZZLE
ASSEMBLY
IB



Figure 10. Sectional View of the MIT Combustion Research Facility showing
a Schematic of the Secondary Zone Air Supply to the Furnace.
-------
Combustion of SRC -II
Single Stage Baseline?
High Air Preheat
T -- 730 K
Tw i n Fluid At o m i z e r
Fuel Oil 2.9/1 Blend
Study
( y - Jet )
-------
q:
lj
£D
Z
3
Z
10
8
~ H
-------
300 ~
A NOx
~ Smoke Number
200
100

q:

u

rf)

Z

ID

Z

u



O

2

LO
10


X
8
u

<
6
q:

<
4
U

U
2
<

CO
0

0.9 1.1	1.3 1.5 1.7 1.9
PRIMARY ZONE EQUIVALENCE
RATIO 0b
Figure 13 Measured NOx and Smoke Number
vs Primary Zone Fuel Equi valence
Ratio,  fa ( Sw irl Number : 0.6 5 )
314
-------
300
?00
10 0
ANOx
~ Smoke Number
~
~
O
1
1
I
J	L
10

0
or
u
d)
2
z>
ZZL
Ld
o
z
X
U
<
o:
<
U
U
<
CD
5	1.0 l.b 2.0 2,5
SWIRL NUMBER
Figure? M Mcosurcd NO* and Smok
-------
150
100
50
A NOx
[j] Smoke Number
I
I
i
10
8
C£
U
CD
3
U
0
2
co
1
u
<
cc
<
o
u
<
CD
400
500 600 700
800
AIR TEMPERATURE K
Figure? 15 Measured NOx and Smoke Number
vs Air Preheat ( 0b= 1.3 , S=0.65)
316
-------
TABLE I *
REACTIONS AND ACCOMPANYING RATE DATA
INCORPORATED INTO THE PLUG FLOW KINETIC MODEL
FOR MODELLING OF NOx FORMATION RATES
kf = ATN exp(-E/RT)
Reaction
cm3/mole s
N
E
cal/mole
CN + H2 = HCN + H
3.16
E
+
12
0.0
5000
HCN + OH = CN + H2O
2.00
E
+
11
0.6
5000
CN + O = CO + N
6.31
E
+
11
0.5
0
CN + OH = NCO + H
5.60
E
+
13
0.0
0
CN + O2 = NCO + O
3.20
E
+
13
0.0
1000
HCN + OH = HNCO + H
2.00
E
+
11
0.0
0
HCN + O = NCO + H
5.20
E
+
12
0.0
8100
NCO + H2 = HNCO + H
3.20
E
+
11
0.0
0
NCO + H = NH +CO
3.00
E
+
14
0.0
0
HNCO + H = NH2 + CO
1.40
E
+
13
0.0
0
NO + 0 = N + 02
1.59
E
+
09
1.00
3864Q
NO + H = N + OH
1.34
E
+
14
0.0
49200
NH3 + OH = NH2 + H20
1.26
E
+
10
0.68
1100
NH3 + 0 = NH2 + 0H
1.50
E
+
12
0.0
6000
NH2 + O = NH + OH
9.20
E
+
11
0.50
0
NH + OH = N + H2O
5.00
E
+
11
0.50
2000
NH + O = N + OH
6.31
E
+
11
0.50
8000
NH + H = N + H2
1.00
E
+
12
0.68
1900
NH2 + OH = NH + H2O
3.00
E
+
10
0.68
1300
NH2 + H = NH + H2
3.50
E
+
10
0.79
4460
NH3 + H = NH2+H2
5.00
E
+
11
0.50
2000
NH + NH2 = NH3 + N
1.00
E
+
11
0.50
4107
317
-------
Table I continued
Reaction

A


N
E
NH + NO = N2O + H
9.00
E
+
09
0.75
0
N2O + H = N2 + OH
7.60
E
+
13
0.0
15101
N2O + M = N2 + O + M
5.00
E
+
14
0.0
58000
NH + N = N2 + H
6.30
E
+
11
0.50
0
NO + N = N 2 + 0
3.10
E
+
13
0.0
334
NH + O = NO + H
6.30
E
+
11
0.50
0
NCO + O = NO + CO
2.00
E
+
13
0.0
0
H2 + 0 = H + OH
1.82
E
+
10
1.00
8900
H2O + O = OH + OH
6.76
E
+
13
0.0
18350
H2 + OH = H2O + H
2.51
E
+
13
0.0
5200
O2 + H = O + OH
2.24
E
+
14
0.0
16800
H + OH+M = H2O+M
4.70
E
+
16
0.0
0
H + H+M = H2 + M
4.70
E
+
16
0.0
0
H + O + M = OH + M
7.94
E
+
15
0.0
0
O2 + M = O + O + M
2.51
E
+
18
-1.00
118700
CO2+H = CO + OH
1.48
E
+
14
0.0
26440
* Rate data was compiled by B. R. Taylor from the literature.
318
-------
TABLE II. SRC-II FUEL OIL ANALYSES
SRC-II
(Middle)
API Gravity	13.A
Viscosity ssu @ 70°F	-
ssu @ 100°F	-
ssu @ 140°F	36.4
ssu @ 210°F	31.8
Flash Point °F	147
Pour Point °F	<-20
Normal Boiling Range °F	350-550
Heat of Combustion
Gross BTU/Lb	17,000
Net BTU/Lb	16,190
Elemental Analysis
Carbon %	83.79
Hydrogen %	9.04
Nitrogen %	0.94
Sulfur %	0.23
Ash %	<0.01
Oxygen %	4.84
SRC-II	SRC-II	Source of
(2.9/1) (Heavy)	Analysis
10.4	1.3
36.8	67.2
33.0	41.3
150	265 V-Truesdail Labs
<-20	+8
350-850 550-850
17,110	17,120
16,340	16,420
86.28
8.83
0.96
0.28
<0.01
3.68







319
-------
TABLE III
INPUT AND OPERATING CONDITIONS FOR THE 12 FLAMES STUDIED
SRC-II Thermal
Flame // Fuel Type Input
1	2.9/1	1MW
2	2.9/1	1MW
3	2.9/1	1MW
4	2.9/1	1MW
5	2.9/1	1MW
6	2.9/1	1MW
7	Heavy	1MW
8	Heavy	1MW
9	Heavy	1MW
10	Heavy	1MW
11	Heavy	1MW
12	Heavy	1MW
Atomizer	Air
Type	Preheat	Swirl #
y-jet	500°F	0.53
y-jet	500°F	2.70
y-jet	850°F	0.53
y-jet	850°F	2.70
Sonicore	500°F	0.53
Sonicore	500°F	2.70
y-jet	500°F	0.53
y-jet	500°F	2.70
y-jet	850°F	0.53
y-jet	850°F	2.70
Sonicore	500°F	0.53
Sonicore	500°F	2.70
320
-------


TABLE IV. SOLIDS
EMISSIONS DATA


Flame //
SRC-II
Fuel Type
Solids Cone.
Lb/M BTU
Smoke //
(Baccarach)
NOx
ppm @ 3% O2
% Excess
Air
1
2.9/1
0.006
0
240
3-8%
2
2.9/1
0.006
0
525
5-8%
3
2.9/1
0.008
0-4
300
2-10%
4
2.9/1
0.004
0
575
2-5%
5
2.9/1
0.005
0-3
250
3-7%
6
2.9/1
0.005
0
500
4-7%
7
Heavy
0.006
0
250
2-5%
8
Heavy
0.010
0
425
2-5%
9
Heavy
0.008
0-2
275
4-8%
10
Heavy
0.008
0
600
5-10%
11
Heavy
0.008
0-2
250
2-8%
12
Heavy
0.008
0-1
515
4-7%
321
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTfTLE
Proceedings of the Joint Symposium on Stationary
Combustion NOx Control. Vol. 4. Fundamental
Combustion Research and Advanced Processes
6. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Symposium Cochairmen: Robert E. Hall (EPA) and
J.E. Cichanowicz (EPRI)
8. PERFORMING ORGANIZATION REPORT NO.
IERL-RTP-1086
9. PERFORMING ORGANIZATION NAME AND ADDRESS
See Block 12.
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
NA (Inhouse)
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings; 10/6-9/80
14. SPONSORING AGENCY CODE
EPA/600/13
15. supplementary notes EPA-600/7-79-050a through -050e describe the previous sympo-
sium.
is. abstract The procee£jingS document the approximately 50 presentations made during
the symposium, October 6-9, 1980, in Denver, CO. The symposium was sponsored
by the Combustion Research Branch of EPA's Industrial Environmental Research
Laboratory, Research Triangle Park, NC, and the Electric Power Research Institute
(EPRI), Palo Alto, CA. Main topics included utility boiler field tests; NOx flue gas
treatment; advanced combustion processes; environmental assessments; industrial,
commercial, and residential combustion sources; and fundamental combustion re-
search. This volume relates to NOx control as applicable to both fundamental combus-
tion research and advanced processes.
17.	KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b. IDENTIFIERS/OPEN ENDEDTERMS
c. cosati Field/Group
Pollution Flue Gases
Combustion Engines
Nitrogen Oxides
Boilers
Tests
Assessments
Pollution Control
Stationary Sources
Environmental Assess-
ment
13B
2 IB 2 IK
07B
13A
14B
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
325
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (t>73)	322
-------