EPA-650/2-73-021
September 1973
ENVIRONMENTAL PROTECTION TECHNOLOGY SERIES
^^^^^x^^^•x»^^^^^^^>x^&*^^^x!&
iisg^iis^w^jSist
-------
EPA-650/2-73-021 PROCEEDINGS, COAL COMBUSTION SEMINAR, JUNE 19-20, 1973
RESEARCH TRIANGLE PARK, N.C. 27711
-------
EPA-650/2-73-021
PROCEEDINGS,
COAL COMBUSTION SEMINAR,
JUNE 19-20, 1973
RESEARCH TRIANGLE PARK, N.C. 27711
Robert E. Hall, Chairman
§
David W. Pershing, Vice Chairman
Environmental Protection Agency
National Environmental Research Center - Research Triangle Park,
Control Systems Laboratory,
Combustion Research Section
Program Element No. 1A2014
Prepared for
NATIONAL ENVIRONMENTAL RESEARCH CENTER
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, N.C. 27711
September 1973
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
ii
-------
PREFACE
The Coal Combustion Seminar was held June 19-20, 1973 in the auditorium
of the National Environmental Research Center, Research Triangle Park, North
Carolina and was sponsored by the U. S. Environmental Protection Agency,
Office of Research and Development, Control Systems Laboratory, Combustion
Research Section.
The Seminar, under the chairmanship and vice-chairmanship of Messrs.
Robert E. Hall and David W. Pershing, began Tuesday morning. The official
welcome and introduction were given by Dr. E. E. Berkau, Chief, Combustion
Research Section.
The Seminar consisted of four sessions divided into two main areas:
fundamental research, and pilot and full scale tests.
Sessions 1 and 2, chaired by Robert E. Hall and G. Blair Martin, respec-
tively, were concerned with fundamental research. David W. Pershing and
David G. Lachapelle were chairmen for Sessions 3 and 4, respectively, which
dealt with pilot and full scale tests.
A tour of the Combustion Research Section's laboratory was given on
Wednesday afternoon by John H. Wasser, G. Blair Martin, David W. Pershing,
and David G. Lachapelle.
All papers presented during the Seminar are included in these proceedings,
Except where noted all ppm values are given corrected to zero percent CL, dry
basis (i.e.,at stoichiometric conditions). To convert ppm values from 0% 02
to 3% 02 multiply the ppm value at 0% 02 by 0.857.
iii
-------
CONTENTS
Title
Preface in
Introduction - E. E. Berkau i
FUNDAMENTAL RESEARCH - PART I
A. E. Axworthy and M. Schuman
Investigation of the Mechanism and Chemistry of Fuel
Nitrogen Conversion to Nitrogen Oxides in
Combustion 9
V. Quan, J. R. Kliegel, N. Bayard de Volo, and D. P. Teixeira
Analytical Scaling of Flowfield and Nitric Oxide in
Combustors 43
FUNDAMENTAL RESEARCH - PART II
D. W. Pershing, J. W. Brown, and E. E. Berkau
Relationship of Burner Design to the Control of
NOX Emissions through Combustion Modification 87
M. P. Heap, T. M. Lowes, R. Walmsley, and H. Bartelds
Burner Design Principles for Minimum NOX Emissions 141
C. England and J. Houseman
NOX Reduction Techniques in Pulverized Coal
Combustion 173
PILOT AND FULL SCALE TESTS - PART I
W. J. Armento and W. L. Sage
The Effect of Design and Operation Variables on NOX
Formation in Coal Fired Furnaces 193
C. R. McCann, J. J. Demeter, and D. Bienstock
Preliminary Evaluation of Combustion Modifications for
Control of Pollutant Emissions from Multi-Burner
Coal-Fired Combustion Systems 205
-------
A. R. Crawford, E. H. Manny, and W. Bartok
NOX Emission Control for Coal-Fired Utility ,,,,-
X Boilers 215
PILOT AND FULL SCALE TESTS - PART II
C. E. Blakeslee and A. P. Selker
Pilot Field Test Program to Study Methods
for Reduction of NOX Formation in
Tangentially Coal Fired Steam
Generating Units 287
G. A. Hoi linden and S. S. Ray
Control of NOX Formation in Wall, Coal-Fired
Utility Boilers: TVA-EPA Interagency
Agreement 305
-------
INTRODUCTION TO
PULVERIZED COAL COMBUSTION SEMINAR
Presented at
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina
June 1973
By
E. E. Berkau
U. S. Environmental Protection Agency
Office of Research and Development
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
-------
COAL COMBUSTION SEMINAR
Good morning, I would like to thank you for taking the time
to attend our meeting and hope that you find it of value. The subject,
as you know, relates to EPA's research and development activities for
control of air pollutant emissions from the combustion of pulverized
coal. To review and discuss the work we have invited representatives
from industry and other government agencies who are intimately involved
with the burning of coal. A list of attendees, as well as copies of the
papers to be presented, will be made available to each of you.
The meeting is being sponsored by the Combustion Research Section
CCRS) of the Control Systems Laboratory. Our responsibilities are to
research and develop economical and efficient combustion modification
techniques for the control of air pollutant emissions from burning of
conventional and waste fuels in all stationary combustion systems. EPA's
Combustion Control Program was officially formulated about 2 years ago
although EPA and formerly NAPCA have been involved with combustion studies
for many years. To date our efforts have concentrated on control of
nitrogen oxides and combustible emissions such as carbon, carbon monoxide,
and unburned hydrocarbons from pulverized coal combustion.
Our program consists of coordinated in-house and contracted studies.
We have selected this approach to allow us in the Combustion Research Section
to become technically involved with the direction and development of
combustion control technology. We feel that this approach is essential
(1) if we are to understand the practical problems involved in the develop-
ment and application of technology to conventional combustion systems and
processes, and (2) to establish private industry's confidence in our
abilities and thereby encourage industry to participate in and accept the
results of our R&D efforts to control air pollutant emissions.
While this meeting is concerned only with coal combustion, we would
like to use the seminar as a tool for disseminating technical information
and for obtaining guidance and participation of industry and other government
-------
agencies in our studies. We feel this approach will provide several benefits:
0) industry will receive the technical results of our work before the final
reports are distributed? (2) industry can have input to, and thereby affect the
direction of, our studies at critical stages; and (3) industry can coordinate
or integrate their own activities with ours. It is anticipated that the
format for future meetings will be similar to this one; but the subject
could relate to either specific combustion hardware or other fuels and will
be largely dependent upon the progress in our R&D efforts. However, future
meetings will always be designed to emphasize applications of technology
for industrial utilization. Your comments as to the suitability of this
approach and the adequacy of the present meeting for these purposes will
be appreciated.
The specific purpose of the present meeting is for us to review
the Combustion Research Section's in-house and contract studies designed
to develop combustion modification technology for control of NOx and
combustible emissions from pulverized coal fired boilers. The studies which
will be presented for your review and discussion encompass very fundamental
research through field testing of practical applications of combustion
control technology. The former are designed to provide quantified under-
standing of the conditions leading to the formation of pollutants and will
establish the basis or foundation for the ultimate in combustion control
technology. For example, Rocketdyne of Rockwell International will present
the results of their work to decipher the combustion chemistry and kinetics
of fuel nitrogen conversion to NOx. On the other extreme, Combustion
Engineering and TVA will discuss their plans to establish installation
and operating costs and develop design guidelines through applications of
combustion control methods to actual field utility boilers. Emphasis will
be placed on techniques which have been established from our field testing
program to be effective for NOX control of pulverized coal fired utility
boilers and which will be described by ESSO Research & Engineering. In
support of the TVA and CE long terra development studies the U. S. Bureau
of Mines will present their preliminary data on the design and NOx control
limitations of various combustion modification techniques for a four burner,
5001/hr experimental boiler.
-------
A number of other studies lying between the very fundamental and the
field testing have also been included in the agenda. The International
Flame Research Foundation's ClFRF] Research Station in Umuiden, Holland,
will review their work to develop single burner design criteria to control
pollutant emissions from coal and oil fired boilers, and Babcock & Wilcox
will discuss their studies to establish the more promising individual or
combination NQx combustion control techniques for use with current burner/
boiler design practice. EPA's in-house study to coordinate the IFRF and
B&W work with coal and other related fuels (gas, oil) R&D contracts will
also be summarized.
Since technology derived from subscale studies with single burner
experimental equipment must ultimately be tested and applied to field
units of much greater size (100 times larger) and involving banks of
burners (16 or more),two other studies are apropos to the meeting.
The Jet Propulsion Laboratory (JPL) in Pasadena, California,will inves-
tigate for EPA on a laboratory scale the effects of multiburner arrays
on single burner NOx control techniques. Since this study has just been
initiated, there are no results to present. However, to introduce you to
JPL, we have asked them to present the results of their company sponsored
bench scale coal combustion studies. Finally, KVB Engineering, who will
be conducting our industrial boiler field testing program, has been asked
to present the results of their company sponsored studies to arrive at
criteria for scaling up combustion control technology from subscale test data.
I have briefly summarized the objectives of the meeting and
introduced the topics to be discussed. I would like to reiterate, however,
that the overall purpose of this meeting is to make you aware of, and
hopefully a part of, EPA's combustion research and development program.
Through your involvement we can be assured that air pollutant emissions
will be controlled through the development and application of technically
and economically sound technology. The initial step toward this rapport
can be accomplished through your active participation in the meeting.
Therefore, we invite your comments, discussion and recommendations for the
technical activities as well as for future meetings. To provide you with
-------
guaranteed attentive ears for your comments, I would like to introduce
the members of the Combustion Research. Section and mention briefly their
primary responsibilities in the Combustion Control Program. They are:
R. E. Hall - Industrial equipment survey and field testing
of conventional stationary combustion systems-
J. H. Wasser - In-house testing of conventional commercial
combustion systems.
D. G. Lachapelle - Applications of combustion control technology
to conventional stationary combustion systems-
D. U. Pershing - Research & Development of combustion techniques
for control of air pollutant emissions from
combustion fuels.
G. B. Martin - Characterization of the emission types and levels
from the combustion of fuels and the identification
of potential combustion control techniques.
To proceed with the program, I will now turn the meeting over to Bob
Hall. Bob has been responsible for the planning and organization of the
Seminar and for making all arrangements. He has a few announcements to
make before introducing the first speaker. Thank you.
-------
FUNDAMENTAL RESEARCH
PART I
-------
INVESTIGATION OF THE MECHANISM'"AND CHEMISTRY OF FUEL NITROGEN
CONVERSION TO NITROGEN OXIDES IN COMBUSTION
BY
A, E, AXWORTHY AND M, SCHUMAN
ROCKETDYNE DIVISION/ROCKWELL INTERNATIONAL
CANOGA PARK, CALIFORNIA
PRESENTED AT THE
COAL COMBUSTION SYMPOSIUM
ENVIRONMENTAL PROTECTION AGENCY
TRIANGLE PARK, NORTH CAROLINA
19-20 JUNE 1973
-------
INVESTIGATION OF THE MECHANISM AND CHEMISTRY OF FUEL NITROGEN
CONVERSION TO NITROGEN OXIDES IN COMBUSTION
by
A. E. Axworthy and M. Schuman
Rocketdyne Division/Rockwell International
Canoga Park, California
INTRODUCTION
This presentation is a progress report covering approximately the first half
of an analytical and experimental program being conducted to determine and
model the kinetics and mechanism of the decomposition of fuel nitrogen com-
pounds,* the fate of the nitrogen-containing pyrolysis products, and the im-
portant physical and chemical processes in the formation of nitrogen oxides
from these species in flames. Much of the background information pertinent
to this study is presented in an excellent review by Sternling and Wendt
(Ref. 1). The general objective of the program is listed in Table 1 and a
program outline is presented in Table 2.
Figure 1 shows the probable chemical path for the formation of "thermal NO"
and hypothetical example of potential chemical paths for the formation of
"fuel NO". Thermal NO is that which forms from the conversion of molecular
nitrogen in air to NO during combustion and fuel NO is that which forms from
nitrogen compounds present in fossil fuels. The chemical mechanisms are
undoubtedly not independent because intermediates formed from fuel nitrogen
have the potential to react with NO or with N atoms formed in the Zeldovitch
mechanism to form N2« The paths for the formation of thermal NO from N«
*The chemically bound nitrogen compounds present in fuel oils and coals,
11
-------
TABLE 1. PROGRAM OBJECTIVE
THE OBJECTIVE OF THIS PROGRAM IS TO DEVELOP A REALISTIC
MATHEMATICAL COMBUSTION MODEL FOR THE FORMATION OF NITROGEN
OXIDES FROM THE CHEMICALLY BOUND NITROGEN COMPOUNDS PRESENT
IN FOSSIL FUELS AND TO INVESTIGATE EXPERIMENTALLY THE IM-
PORTANT PHYSICAL AND CHEMICAL PROCESSES INVOLVED IN THE
FORMATION OF "FUEL NO",
12
-------
TABLE 2. PROGRAM OUTLINE
PHASE IA - THEORETICAL ANALYSIS
•PYROLYSIS OF FUEL NITROGEN COMPOUNDS (PREFLAME REACTIONS)
•FORMATION OF THERMAL AND FUEL NO
•MODELING OF OIL DROPLET/COAL PARTICLE BURNING
PHASE IB - FUEL DECOMPOSITION EXPERIMENTS
•PYROLYSIS OF MODEL FUEL NITROGEN COMPOUNDS
(DECOMPOSITION KINETICS AND PRODUCT DISTRIBUTION)
•PYROLYSIS OF FUEL OILS AND COALS (DETERMINE INITIAL
PRODUCTS OF FUEL NITROGEN COMPOUNDS)
PHASE IIA - NITROGEN COMPOUND COMBUSTION EXPERIMENTS
•BURNER STUDIES WITH INTERMEDIATES FORMED IN PREFLAME
REACTIONS OF FUEL NITROGEN COMPOUNDS
•EFFECTS OF INTERMEDIATE CHEMICAL TYPE AND COMBUSTION
CONDITION ON CONVERSION TO NOX
PHASE I IB - MATHEMATICAL CORRELATIONS
•MODEL FOR FUEL NO FORMATION
•EXTEND DROPLET/PARTICLE-MODELS TO COMBUSTION OF
PARTICLE ENSEMBLES UNDER REALISTIC CONDITIONS
13
-------
CO, H20, C02, 02,
AIR
OH, H
HYDROCARBON
SPECIES
NITROGEN
SPECIES
FUEL
PARTICLE
i
SOLID
PARTICLE
w
- — ^
CO + OH - C02 + H
H + 02 - OH + 0
CO + 02 - C02 + 0
fcr
0-ATOM FORMATION
0X1 DATIVE PYROLYSIS j
1
1
2 w, HETROGENEOUS 1 w .... .
(C02, H^ FUEL
NO ^ "n
N + NO - N + 0
NH •»• NO - N2 + OH
CN + NO • N2 + CO
NH + N - N2 + H
Ik.
0 + N2 - NO + N
N •*• 02 - NO + 0
N2 •«• 02 - 2 NO
THERMAL NO FORMATION
VIA ZELDOVICH MECHANISM
HCN + 0 - NCO -l- H
NCO + 02 - CO + NO + 0
HCN + H - H2 + CN
CN + 02 - CO + NO
FUEL NO FROM HCN
NH + OH - N + H20
N + OH - NO + H
FUEL NO FROM NH
INTERACTION OF
FUEL NO AND THERMAL NO
MECHANISMS
Figure 1. Potential Paths for NO Formation
-------
under fuel-rich conditions by reactions not involving 0 atoms, as proposed
by Fenimore and Iverach, are not included in Fig. 1.
The mechanisms shown for the formation of fuel NO in Fig. 1, which are highly
speculative, indicate the possible complexity of the formation of NO from
fossil fuels. The HCN path is included because experiments conducted during
this program have shown that under certain conditions the nitrogen contained
in coal, fuel oils, and model compounds can be converted nearly quantitatively
to HCN. The HCN reactions listed are those presented in Ref. 1. The NH mechan-
ism in Fig. 1 is of the type proposed by Fenimore (Ref. 2) to account for the
similar behavior of various nitrogen compounds including ammonia. However,
if soot particles are present, the thermodynamically favored reaction
NH3 + Cs = HCN + H2 makes an HCN mechanism plausible even with NHg. A reac-
tion between HCN and OH should be added as a likely initial reaction in the
formation of fuel NO.
The heterogeneous formation of fuel NO is a very likely path not only in the
case of coal particles but with fuel oils, also. Experimental results ob-
tained during this program indicate that even volatile heterocyclic nitrogen
compounds have a strong tendency to form carbonaceous residues during pyroly-
sis and these residues contain considerable nitrogen. Thus, nitrogen-
containing soot particles could form in oil droplet combustion leading to
heterogeneous fuel NO formation in the flame front.
COMBUSTION MODELS
As shown in Fig. 2, the droplet particle combustion models being developed are
of three types: (1) droplet vaporization model, (2) droplet/particle flame-
front model, and (3) heterogeneous coal combustion model. These have been com-
bined with the necessary chemical reaction rate constants to give an average
film kinetic/diffusion model that includes the rate of formation of thermal NO.
Chemical reactions are being added for the formation of fuel NO but the cal-
culated fuel NO formation rates will be only qualitative at present because of
the uncertainties in the reaction mechanisms and rate constants.
15
-------
COMBUSTION MODELING
ANALYSIS
DROPLET MODEL
CONCEPT
DROPLET VAPORIZATION
MODEL
DROPLET 6 COAL
FLAME-FRONT
MODEL
COAL MODEL
CONCEPT
HETEROGENEOUS COAL
COMBUSTION MODEL
DESCRIBE:
COMPOSITION HISTORY, TEMPERATURE HISTORY,
FILM PHENOMENA, RATE OF EVOLUTION OF N
COMPOUNDS, ETC.
PARAMETERS:
FUEL TYPE, DROPLET OR PARTICLE SIZE,
SURROUNDING ENVIRONMENT, DEGREE OF
CONVECTION, ETC.
AVERAGE FILM KINETIC/
DIFFUSION MODEL
r
FUEL N* REACTIONS
FUEL 'NO
^%
1
r
OTHER NITROGEN
PRODUCTS
1
THERMAL N2/
REACTIONS
THERMAL NO
Figure 2. Combustion Modeling Analysis
-------
It can be seen from Fig. 3 through 5 that the heterogeneous coal combustion
model developed on this program gives results that agree with the coal
particle composition histories measured by Howard and Essenhigh (Ref. 3).
Inspection of the initial time region of the volatile matter and fixed car-
bon contents shown in Fig. 3 and 4 suggests that during the first 0.05
seconds, volatile matter is evolved without heterogeneous combustion. The
heterogeneous combustion then begins at a sufficiently high rate that volatile
matter loss due to heterogeneous combustion is as fast or faster than the
volatile matter loss due to gaseous evolution. After about 0.2 seconds, the
volatile reactions become very slow and the remainder of the volatile mate-
rial is lost due to heterogeneous combustion (Fig. 5). The reactions of C02
and HgO at the solid surface (Table 3) have been added to the coal combustion
model. The importance of these reactions on the result obtained is being
investigated.
The chemical reactions presently included in the kinetic/diffusion model are
listed in Table 4. The results of a preliminary calculation with the kinetic/
diffusion model for a No. 2 fuel oil are shown in Fig. 6. Under the conditions
of this calculation (relatively low flame temperature), the diffusion of
species to the droplet surface is quite rapid. The temperature at the flame
zone (and, therefore, the rate of formation of thermal NO) is strongly depend-
ent on the free-stream temperature with this model. Therefore, the use of this
model to predict the rate of formation of thermal NO must await extention of
the model during Phase II.
The concentrations of CO and 0 in Fig. 6 are very much greater than the pre-
dicted equilibrium values. This appears to result mainly from allowing all of
the fuel to react to CO rather than COg. High oxygen atom concentrations then
form by the reactions shown at the top of Fig. 1. This overshoot will be con-
trolled by reducing the fraction of reaction that goes to CO.
17
-------
o
ui
z _i
ui <
o —
o z
a:
ui u.
UI (_}
_i a:
100
• EXPERIMENTAL
POINTS (REF.3)
PARTICLE SIZE
15 ^RADIUS
1*0 A* RAD I US
CALCULATED
0.8
TIME, SECONDS
Volatile Matter Content
100
80
S 2 60
lo
ne i_
20 r
• EXPERIMENTAL
POINTS (REF. 3)
REF. RATES
REF. 5 RATES
0.4
TIME, SECONDS
Figure 4. Fixed Carbon Content
DC
UI
0.
o
CO
ce
o
o
o
o
i
100
80
60
20
1
1
1
• EXPERIMENTAL
POINTS (REF. 3)
PARTICLE SIZE
i 15/U RADIUS
RADIUS
_L
100 80 60 40 20
FIXED CARBON CONTENT, WEIGHT PERCENT OF ORIGINAL
Figure 5. Variation of Composition of Solid Material
With Degree of Burnout
18
-------
CHEMISORPTION REACTIONS:
Kl
CF + 02 -*co + 0
K2
CF + 0-C0
K3
CF + C02 ^ CO + CQ
K5
TABLE 3. COAL COMBUSTION MECHANISM
COAL REACTION RATE:
(C)
+ Kc (HoO)
K1(02) + K2 (0)
(i)
(C02)
GASIFICATION REACTION:
K7
C0 — CO + CF
WATER-GAS SHIFT REACTION:
K8
CO + H20 C02 + H2
-------
TABLE 4. REACTIONS FOR THE KINETIC/DIFFUSION MODEL
(EXCLUDING FUEL NITROGEN REACTIONS)
2. CO + OH = C02 + H -
3. 02 + H2 = 20H 11- H
4. OH + H2 = H20 + H 12- N2 + 0 = NO +
5. 02 + H = OH + 0 13.' M + 02-MO +
6. 0 + H2 = OH + H W. H + 0 + H-HO
7. 0 + H20 = 20H 15. N
8, 2H + H = H2 + H
20
-------
NO. 2 FUEL OIL
VELOCITY - 10 FPS
N(l) NITROGEN-CONTAINED FUEL
.001
RADIUS/DROPLET RADIUS
Figure 6. Preliminary Results With Flame-Front Model
21
-------
PYRQLYSIS EXPERIMENTS
One of the processes involved in the formation of fuel NO that has received
little attention is the pyrolysis of fuel nitrogen compounds (Fig. 1).
Hurd and Simon (Ref. 6) pyrolyzed pyridine and picolines at 850 C but
did not establish the kinetic parameters or the fate of most of the nitrogen.
An activation energy and reaction order are required to permit the pyrolysis
rates to be extrapolated to combustion temperatures for the mathematical
models.
Although it is difficult to determine the exact structures of the nitrogen com-
pounds present in fossil fuels (particularly in coals), it has been established
that they are mainly aromatic compounds—mostly heterocyclics. The structures
listed in Table 5 are believed to include most of the classes of fuel nitrogen
compounds. Because of the large number of nitrogen compounds believed to be
present in fuels, most of the experimental effort during this program has been
with the pyrolysis of model nitrogen compounds. A number of fuel oils have
also been pyrolyzed and one coal sample. Additional coal experiments are
planned using a rapid heating technique that was developed to investigate the
pyrolysis of solid propellant ingredients and is being modified for use with
coal.
APPARATUS
A schematic of the experimental setup is shown in Fig. 7. Most of the model
compound experiments involved the vaporization of a 0.2 microliter sample
into a helium stream that flows through a quartz reactor. The reactor is
2.2 mm ID with a volume of 1.2 cc and a nominal residence time of 0.75
seconds. Experiments were also conducted with a vapor injector in which the
sample vapor was premixed with He or He/02 and a 1 cc slug introduced into the
He stream before it entered the reactor. The organic products were identified
by temperature-programmed gas chromatography and mass spectrometry. HCN and
NH3 were trapped in neutralizing solutions and determined, respectively, by a
22
-------
TABLE 5. NITROGEN COMPOUNDS IN FOSSIL FUELS
S3
OJ
PYRIDINES
QUINCLINES
I SO-QUINCLINES
PYRROLES
INDOLES
CARBAZOLES
PHENAZINES
BENZONITRILES
CN
H
t
QUINOLONES
-------
The unadjusted pH is quite low and the acid requirement is also low.
Considering acid requirements, the need of soda ash and lime and the
importance of proper dosage is illustrated in the following tabulation:
Table 1. ACID REQUIREMENT BASED ON CHEMICAL TYPES USED.
Chemicals used
Soda ash only
Soda ash and lime
Optimum dosage,
soda ash and
lime
Average acid requirement,
per 1,000 gal. brine
0.71 gal.
0.18 gal.
0.04 gal.
Complete tabulation summary of unadjusted brine pH and acid usage is to
be found in Appendix B,
Laboratory bench tests were run to verify empirically established
"optimum" dosages. Since the reactions are equilibria, any of the
reactions can be reversed with a change in conditions. The equilibria
move in a given direction because of relative solubilities. These
solubilities are recorded in Table 2.
Table 2. SOLUBILITY OF COMPOUNDS1-
(in grams/100 grams of water at 20 degrees C.)
CaC03
Ca(OH)2
Mg(C03)
Mg(OH)^
NaCl
Na2C03
Mgci2
CaCl
0.0012
0.165
0.0106
0.0009
36.0
21.5
54.5
59.5
24
-------
cyanide specific electrode and Nessler's reagent. N2 was measured on a molecu-
lar sieve column GC. Between experiments, the chromasorb 103 column was back
flushed and the residue in the reactor burned out. The oil pyrolysis experi-
ments were conducted in a similar manner except that the reactor had a 2 cc
bubble blown near the center and the sample was introduced into the heated reac-
tor by quickly moving a small quartz boat containing 1 to 2 milligrams of oil
into the heated zone. The initial coal experiments were conducted in this
oil pyrolysis apparatus.
MODEL COMPOUND EXPERIMENTS
The model compounds chosen for study were pyridine, pyrrole, quinoline, and
benzonitrile. These represent many of the nitrogen structures shown in
Table 5. The nitrogen compounds present in the fossil fuels are highly sub-
stituted and of much higher molecular weights but it is expected that experi-
ments with these parent compounds will be indicative of the types of high-
temperature reactions that can occur with chemically bound nitrogen.
Decomposition Rates as a Function of Temperature
Pyrolysis experiments were conducted at temperature intervals of about 25
degrees. The experimental decomposition curves obtained for the model compounds
(in helium) are shown in Fig. 8. Pyridine and pyrrole gave similarly shaped
curves with slopes that remained steep until beyond 95 percent decomposi-
tion. The pyrrole is less stable than pyridine, the curves being separated by
about 60 degrees. Quinoline gave a decomposition curve that is nearly linear
with temperature. Quinoline is less stable than all of the other compounds
below 910 C but is more stable than pyrrole above that temperature. Quinoline
is unusual in that its decomposition curve remains steep up to about 960 C and
then tails out to high temperatures. In fact, 4 percent remained undecomposed
even at a temperature of 1100 C. Thus below 1000 C, pyridine is more stable
than benzonitrile, while above 1000 C, the reverse is true.
25
-------
Q
Ul
O
o
100
90
80
70
60
50
30
20
10
PYRROLE
PYRIDINE
QU INCLINE
BENZONITRILE
®
O
^5>»
»N'
•CN
850
900
Figure 8.
950 1000
TEMPERATURE, °C
1050
Model Compound Decomposition Curves in Quartz
(at a nominal residence time of 0.75 seconds,
helium carrier gas)
1100
-------
The experimental decomposition curves were fitted to rate expressions for use
in the combustion models. Figure 9 shows the pyridine pyrolysis curve* plotted
in a semi-log form along with three theoretical first-order rate expressions.
It is apparent that the data fit a first-order expression with an activation
energy of 70 kcal/mole. Similar first-order fits were obtained for the pyrrole
and quinoline data giving the rate parameters listed in Table 6. The benzo-
nitrile data are still being analyzed and it appears that a complex rate expres-
sion may be involved.
The rate parameters obtained for pyridine and pyrrole indicate that the rate-
determining step is a unimolecular reaction. Such reactions typically have
pre-exponential factors in the range of 10 to 10 sec" (Ref. 7). The sur-
prising feature is that pyrrole, which is less aromatic than pyridine, has a
higher activation energy for decomposition. The low quinoline pre-exponential
factor suggests a heterogeneous reaction.
Shown in Fig. 10 is an Arrhenius plot of the decomposition half-life of pyri-
dine as a function of temperature. The results of this study and the rates
measured by Hurd and Simon fall on the solid line. Figure 10 shows that the
(extrapolated) half-life for pyridine is about 0.1 milliseconds at 1800 K and
that this extrapolated half-life will be in error about 15 percent for each
1 kcal/mole error in the activation energy.
Pyridine experiments using the vapor injector gave decomposition rates that
were slower by about a factor of three than when liquid sample injection was
employed. The pyridine concentration in the reactor is estimated to be lower
by about a factor of 100 with the vapor injector. This indicates that the
reaction is of an order somewhat smaller than one but some other experimental
parameters may be changing slightly instead.
Pyridine experiments were conducted with the vapor injector using a mixture of
5 mole percent oxygen in helium as the diluent gas. The data are not yet
*This curve was from an early experiment that gave a smaller pre-exponential
factor than did the curve in Fig. 8.
27
-------
CO
2.0
1.5
o
Ul
«/>
o
a.
o
o
Ul
o
1.0
o
o
0.5
Oh-
900
k - 2.1* X 10UEXP(-65,000/RT)
k - 1.8 X 1012 EXP (-70,000/RT)
k - 1.0 X 1011* EXP (-80,000/RT)-
950
1000
1050
TEMPERATURE, C
Figure 9. Pyridine Pyrolysis Curve (Log Scale)
-------
10.0
6.5
30 60
TIME , minutes
Figure 15. Treated Brine pH, as Functions of Soda Ash
Dosage and Reaction Time.
29
-------
1000
1800 1600
TEMPERATURE, K
JitOO 1200 1100
1000
1000/T, K
Figure 10. Half-Life for Pyridine Decomposition as a Function
of Temperature (Arrhenius Plot)
30
-------
reduced but the oxidatlve pyrolysls temperatures are lower by about 100 to
200 degrees for a given extent of decomposition. The oxidative pyrolysis
rate parameters will also be included in the combustion models.
Decomposition Products
Only minor amounts of ammonia were detected in the products from the pyroly-
sis experiments. Additional tests for ammonia are being made using the
ammonia converter (Fig. 7) to decompose the NH3 to N£ giving a more sensitive
test. The mass balances obtained in the model compound experiments (with
helium carrier gas) are summarized in Table 7 as a function of temperature.
Depending upon the temperature, from 4 to 100 percent of the carbon is present
in the observed products and 0 to 90 percent of the nitrogen is found.
Benzonitrile forms almost no methane, quinoline up to 4 percent methane,
pyridine about 10 percent, and pyrrole about 25 percent methane. The other
observed organic products, HCN, and a few (as yet unidentified) GC peaks
account for from 4 to 74 percent of the carbon under the various conditions.
Thus, the unrecovered carbon (believed to be in the carbonaceous residue
that forms on the reactor wall) amounts to from 0 to 96 percent of the sam-
ples (Table 7).
The amount of HCN formed was strongly temperature dependent ranging from 24
to 89 percent of the nitrogen in the sample at about 1100 C to less than a
few percent (near the detection limit) at temperatures of 1000 C and lower.
The amount of nitrogen found in the identified organic products ranged from
none from benzonitrile to 36 percent for pyrrole. The amount of nitrogen
present in the unknown peaks must be small except for pyrrole at 900 C where
it could amount to as much as 50 percent of the total nitrogen.
It appears that the unrecovered nitrogen in these experiments is contained
in the carbonaceous residue present in the reactor after each experiment.
The residue from two experiments with pyridine at 970 C was analyzed.for
nitrogen by the Dumas method and about two-thirds of the missing nitrogen was
31
-------
reaction time. The graphs clearly show the increased pH values due to
the solubility of calcium hydroxide.
The laboratory data were reviewed with the determination that the
optimum dosages should be less than stoichiometric: 85$ for soda ash
and 68% for hydrated lime. With these dosages, then, a third series
of laboratory tests were made with waste brine to be reclaimed. The
waste brine was treated with soda ash (85% of stoichiometric), stirred
45 minutes, then treated with hydrated lime (68% stoichiometric) and
stirred an additional 45 minutes. The reactants were sampled
periodically and analyzed for calcium, magnesium and pH. Triplicate
tests were performed: the values were averaged for preparation of
Table 19 in Appendix C. The tabular data was then used to prepare the
graphs of Figures 18, 19 and 20.
Figure 18 shows the remaining calcium hardness as a function of time.
The figure shows that the soda ash addition was sufficient to reduce
the calcium hardness to zero but that the subsequent hydrated lime
addition increased the calcium hardness.
Figure 19 plots the remaining magnesium hardness and clearly shows
that the magnesium hardness was unaffected by the soda ash addition
and that the hydrated lime significantly reduced the magnesium.
Figure 20 shows the treated brine pH value. It clearly indicates that
the pH increases with soda ash addition, but that subsequent addition
of hydrated lime reduces the pH. A review of this data indicates that
the optimum dosages will yield a reclaimed brine of suitable quality
except that the pH of about 9 must be reduced with subsequent addition
of acid.
With the established dosages and with the addition of soda ash first
and lime second, minimal amounts of acid were required for pH adjust-
ment of the effluent. This adjustment was made by adding a
predetermined amount (usually 100-150 ml) of hydrochloric acid
(20° Be') as the decantation was occurring. This provided sufficient
agitation for mixing.
The brine as originally drawn off had a turbidity of 20 JTU, due to
unsettled small particles of precipitate. The acid added for pH
adjustment dissolved the precipitate to produce a product brine of
about 1.0 JTU. The slight increase in hardness that resulted was not
sufficient to cause failure to meet specifications.
The established procedure is outlined in detail in Appendix D, which
includes plant operation, lab testing, etc. This procedure was used
throughout the subsequent demonstration runs.
32
-------
recovered. Another experiment was conducted in which the residue from the
pyrolysis of pyridine at 970 C was heated to 1100 C for 15 minutes. No HCN
was evolved indicating that the HCN formed in the high-temperature pyroly-
sis experiments forms directly in the initial pyrolysis reaction.
The individual organic products that were present in the model compound
decomposition products are shown'-in Fig. 11 through 14. The products ob-
tained from pyridine at the lower temperatures (Fig. 11) are the same as
those reprorted by Hurd and Simon but are recovered at much higher concentra-
tions. This probably results from the shorter residence time and the im-
proved experimental procedure.* At higher temperatures, the less thermally
stable products decrease in concentration and only the stable products are
observed. The products from quinoline (Fig. 12) are similar to those from
pyridine (benzene and benzonitrile) except that methane, acetonitrile, and
acrylonitrile are at much lower concentrations or absent. It is reasonable
that quinoline forms mostly residue since it represents the first step in
the condensation of pyridine to residue.
The major products from benzonitrile (Fig. 13) are benzene and biphenyl indi-
cating that the first step is a C-C bond rupture followed by hydrogen abstrac-
tion to form benzene and some association of phenyl radicals. The surprising
feature is that most of the nitrogen apparently goes into the residue even
though the initial formation of CN radicals is indicated.
The experiments with pyrrole, conducted very recently, are interesting in that
all of the carbon is recovered at low temperature and possibly most of the
nitrogen, i.e., little residue is formed. The unknown that accounts for 48
percent of the carbon at 875 C (Fig. 14) has a retention time on the GC column
about the same as do pyridine ard the picolines. It would be quite unexpected
*It was established during this program that the fraction of pyridine going
to residue increases if the residue is allowed to build up (i.e., the resi-
due catalyzes its own formation).
33
-------
10
-p-
o
§
o.
z
Ul
UJ
O
CC
UJ
Q.
100
80
60
30
20
10
8
6
3
2
1
0.8
0.6
0.4
0.3
0.2 -
950
1000 1050
TEMPERATURE °C
1100
Figure 11. Pyrolysis Products of Pyridine
-------
LO
too
80
60
50
IfO
30
20
o
o 10
1 8
z 6
§ 5
(9 > J|
e 3
o
oc
43
111
1
0.8
0.6
0.5
0.4
0.3
0.2
850
900
TEMPERATURE,
950
1000
Figure 12. Products of Quinoline Decomposition
-------
OJ
100
80
60
30
o
a
o
cc
a.
19
o
X
C9
UJ
o
CC
20
10
8
J»
3
1
0.8
0.6
0.4
0.3
0.2 -
HCN
— METHANE
950
1000 1050
TEMPERATURE, °C
UOO
Figure 13. Pyrolysis Products of Benzonitrile
-------
100
80
60 1— UNKNOWN (C5-C6)
50
30
20
I 10
z 8
> 6
3 5
0.8
0.6
0.5
0.4
0.3
0.2
\
UNKNOWN (~C2)
\
\
\
I
I
850
900 950
TEMPERATURE, C
1000
Figure 14. Products of Pyrrole Decomposition
-------
if the first step in pyrolysis of pyrrole turns out to be the formation of the
aromatic pyridyl ring.
To summarize the important results of the model compound experiments to date,
virtually no ammonia or HCN are formed at lower temperatures but large amounts
of HCN are formed at the higher temperatures. It is quite possible that in
the combustion process, where heating rates are high, HCN is the only important
fuel nitrogen intermediate. The other important observation is that even
volatile nitrogen compounds in the vapor phase have a strong tendency to form
a solid residue which contains a major fraction of the nitrogen (at lower tem-
peratures). Thus, the heterogeneous combustion of soot particles could be a
source of fuel NO in a diffusion flame.
FUEL PYROLYSIS EXPERIMENTS
Six samples of No. 6 fuel oil were pyrolyzed in helium in the oil apparatus
at 1100 and 950 C. The amounts of HCN formed are listed in Table 8. Each of
these values is the average of two runs that gave moderately good reproduci-
bility. It can be seen that, as with the model compounds, much more HCN is
formed at the higher temperature than at the lower. Of the fuel oils, only
the first (Table 8) formed appreciable HCN at 950 C. The Wilmington crude
also gave considerable HCN at 950 but gave twice as much at 1100 C.
The one coal sample that has been run in this apparatus gave (reproducibly)
large quantities of HCN at both temperatures. Calculation of the percent
nitrogen that went to HCN is less certain with coals because of the small sam-
ple size and the possibility of a nonhomogeneous nitrogen distribution. In
addition, this coal sample was determined to have about twice the listed nitro-
gen content a few months earlier (see below).
It was found that the presence of sulfide ion can cause an interference in the
cyanide electrode method used to measure the HCN. Calibration experiments
revealed that sulfide ion alone will not be detected as HCN but, in the presence
of HCN, sufficient sulfide ion will increase the response factor by about 40
38
-------
TABLE 8. FUEL PYROLYSIS EXPERIMENTS
GULF NO. 6 FUEL OIL
(VENEZUALIAN CRUDE)
GULF NO. 6 FUEL OIL
(VARIOUS CRUDES)
GULF NO. 6 FUEL OIL
(MAINLY CALIFORNIA CRUDE)
CONOCO NO. 6 FUEL OIL
NO. 6 FUEL OIL
(EPA IN-HOUSE)
NO. 6 FUEL OIL
(EX-ULTRASYSTEMS)
WILMINGTON CRUDE
COAL
(EPA IN-HOUSE)
%N
0.43
0.44
1.41
0.3
0.5
0.38
0.63
(0.59)
JS
2.31
0.73
1.63
0.66
0.9
0.33
1.59
XN AS HCN
1100 C
69
49
33
<6
24
7
127
120
950 C
47
5
6
4
—
57
101
TABLE 9. COAL ANALYSES AND VOLATILE NITROGEN RESULTS
COAL
IFRF-A
IFRF-N
EPA
EPA
(DUPLICATE)
N, XW
JAN '73
1.16
1.47
1.17
MAY '73
0.54, 0.64
0.91, 1.16
0.60
0.58
% VOLATILES
27.3
41.5
38.0
38.9
N, %U IN
RESIDUE
0.16
0.18
0.27
0.25
XN IN
RESIDUE
21.5
11.5
27.8
26.5
39
-------
percent. It may be necessary to reduce the HCN values in Table 8 by about
one-third. This sulfide interference will have no effect on the model
compound results because sulfur was not present. A colorimetric method has
been found that will determine HCN accurately in the presence of sulfide
ion.
Some of these oils and coals were rerun to investigate the amounts of ammonia
and nitrogen that are formed, if any. The data are being reduced but it does
appear that these will be important products.
The fuel pyrolysis results were encouraging in that they indicate that real
fuels behave similarly to the model compounds. That is, they form large
amounts of HCN at the higher temperatures. It is again possible that under
combustion conditions these fuels may form HCN quantitatively.
COAL AND RESIDUE ANALYSIS
Three coal samples were analyzed for nitrogen in January 1973 by the Dumas
method. The results obtained are shown in the first column of Table 9.
These analyses were repeated four months later giving the much lower results
shown. The reason for these lower results is not known but sample in-
homogeneity must be suspected. After the second series of analyses, the per-
cent volatiles were determined for these coals as well as the percent nitro-
gen in the residues. It can be seen from Table 9 that only about one-fourth
of the nitrogen remained in the residue (or less if the higher nitrogen
values are correct). These are smaller amounts of nonvolatile nitrogen than
have been estimated previously (Ref. 1).
40
-------
REFERENCES
1. Sternling and Wendt, Shell Development Company Report No. S-14129,
August 1972.
2. Fenimore, C. P., Combustion and Flame. T9_» 289-296 (1972).
3. Howard, J. B. and R. H. Essenhigh, "Pyrolysis of Coal Particles in Pul-
verized Fuel Flames," I&EC Process Design and Development. Vol. 6,
No. 1, January 1967, pp 74-84.
4. Essenhigh, R. H., R. Froberg, and J. B. Howard, "Combustion Behavior
of Small Particles," Industrial and Engineering Chemistry, Vol. 57,
No. 9, September 1965, pp 33-43.
5. Smith, I. W., "Kinetics of Combustion of Size-Graded Pulverized Fuels
in the Temperature Range 1200-2270 K," Combustion and Flame. Vol. 17,
(1971), pp 303-314.
6. Hurd, C. D. and J. I. Simon, J. Amer. Chem. Soc., 84, 4519 (1962).
7. Benson, S. W., "Thermochemical Kinetics," John Wiley and Sons, New York,
1968.
ACKNOWLEDGMENT
This program is sponsored by the Environmental Protection Agency under Con-
tract 68-02-0635. The EPA program monitor is G. Blair Martin. Other
Rocketdyne personnel who have contributed to this program include: V. H.
Dayan, G. Lindberg, E. Welz, R. I. Wagner, A. Miles, R. Kessler, W. Nurick,
P. Combs, and I. Lysyj.
41
-------
ANALYTIC SCALING OF FLOWFIELD
AND NITRIC OXIDE IN COMBUSTORS*
BY
VICTOR QUAN, JAMES R, KLIEGEL, NICK BAYARD DE VOLO
KVB ENGINEERING, INC,, TUSTIN, CALIF, 92680
AND
DONALD P, TEIXEIRA
SOUTHERN CALIFORNIA EDISON COMPANY, ROSEMEAD, CALIF, 91770
"PRESENTED AT THE EPA PULVERIZED COAL COMBUSTION SEMINAR,
RESEARCH TRIANGLE PARK, NORTH CAROLINA, JUNE 19 AND 20,
1973,
43
-------
CONTENTS
Page
ABSTRACT 46
1. INTRODUCTION 47
2. SCALING CRITERIA AND SIMILARITY LAWS 49
3. DERIVATION OF SCALING RELATIONS 52
3.1 Governing Equations 52
3.2 Scaling of Flow Properties and Nitric Oxide 55
3.3 Turbulent Transport 60
3.4 Boundary Conditions 63
4, SOURCE TERMS AND MOLECULAR TRANSPORT 64
4.1 Oil and Coal Combustion 64
4.2 Molecular Transport 70
4.3 Thermal Radiation 71
4.4 Gravity and Finite-Rate Chemistry 73
5. DISCUSSION AND SUMMARY 75
NOMENCLATURE 78
REFERENCES 81
FIGURES 1 and 2 82
45
-------
ABSTRACT
The criteria for flow and chemical similarity including
nitric oxide formation in turbulent flows are derived from the
conservation equations. It is shown that the flowfield and
primary combustion product concentrations in fullscale com-
bustors can be practically simulated in laboratory subscale
combustors, but that the nitric oxide concentration is pro-
portional to the combustor characteristic dimension if the non-
linear effect of radiation heat loss is neglected. For gas
fired combustors, the similarity conditions require only that
the geometries and boundary conditions be similar. For oil
fired units, only one additional particle size scaling relation
must be satisfied. For coal fired units, however, additional
burning rate scaling relations are imposed.
46
-------
1. INTRODUCTION
The similarity conditions for chemically reactive systems
has been investigated by Penner (1955), Spalding (1963), and
others. Because of the large number of parameters in combustion
processes/ exact combustion system scaling is impossible. Thus,
only partial modeling can be successful; and in typical problems
(liquid fuel rocket engines treated by Penner, flame propagation
in spark-ignition engines discussed by Spalding, etc.) experience
has shown that only a few significant dimensionless groups are
important and need be considered in practical modeling and
scaling.
The present study considers turbulent diffusion flames in
industrial combustors, and special attention is directed at the
formation of nitric oxide therein. The scaling approach taken
is basically pragmatic although its derivation is mathematical.
The objective is to be able to perform simple laboratory experi-
ments in geometrically scaled combustors without pressure or
gravity scaling and be able to correctly interpret the measured
results in terms of fullscale combustor performance with little
error. For this purpose, one must determine those dominant effects
which must be scaled correctly and to determine the scaling
correction factors for small effects and perturbations.
It is a physically known fact that turbulent transport
mechanisms dominate molecular transport mechanisms in industrial
combustors and that combustion kinetics are extremely rapid ex-
cept for kinetically limited contaminant formation. It can be shown
that if a combustion medium is optically thick, the turbulent
transport process dominates the radiant energy transport
process. Gravitational effects are also known to be small in
industrial combustors. To good first approximation, the flow in
an industrial combustor can be treated as an optically thick
47
-------
turbulent flow in chemical equilibrium. Such flows scale
exactly for similar geometries and wall conditions. Main flow
departures from this scaling are small and can generally be
treated as either scaling corrections or accepted as experimental
errors. Wall effects do not scale as directly but may be com-
pensated for by wall temperature changes if important. This
approach results in the simple scaling laws, given in the next
section, which allow realistic combustor scaling. These scaling
laws provide for the practical scaling of the dominant combustor
flow features for gas, oil, and coal fired units and for simple
extrapolation, which requires further correction only for thin
gas radiation effects, of contaminant formation to full size
units.
48
-------
2. SCALING CRITERIA AND SIMILARITY LAWS
A list of the conditions for similarity is given below.
These are sufficient conditions of which some may not be neces-
sary and some may be altered to achieve certain desired scaling
results. Items 1 to 9 are operating variables to be kept equal
for subscale model and fullscale prototype. Item 10 concerns
geometric scaling, and item 11 concerns particle size scaling
in oil and coal fired units. Items 12 to 15 are idealized
assumptions of the physical processes.
1. Fuel composition
2. Oxidizer composition
3. Fuel temperature
4. Oxidizer temperature
5. Pressure
6. Equivalence ratio
7. Inlet velocities
8. Wall temperatures
9. Inlet turbulence levels
10. Geometries are similar between model and prototype
11. With particles, their size distribution varies with
one-half power of combustor dimension
12. Fuel-oxidizer combustion is limited only by diffusion
13. Molecular processes are unimportant compared to
turbulent transport processes
14. Radiation effects are negligible
15. Body forces are negligible
The above conditions for similarity are surprisingly
few. In fact, similarity in combustors for laminar flow is
much more difficult to achieve as noted in a later section. Of
the conditions listed above, the operating variables 1 to 9 can
easily be kept the same between model and prototype. Geometric
similarity, item 10, can be achieved to a large extent, at least
in the important characteristics. Particle size, item 11,
-------
certainly can be chosen at inlet. It will be shown in a later
section that the particle size will remain scaled throughout
the flowfield for oil burning, but an additional scaling con-
dition between burning rate and combustor size is required for
coal burning. Only items 12 to 15, which are physical processes
occurring within the combustor, may be difficult to control in
certain circumstances. Corrections for their effects are
discussed in a subsequent section.
From the scaling study, the following results of
similarity relations are obtained. Items 1 to 7 correspond
to results obtained under the idealized conditions listed above,
and items 8 to 10 concern relaxation of the idealized conditions.
1. The velocity components, temperature, pressure,
density, and major chemical species are equal at corresponding
positions between subscale and prototype.
2. The velocity components, temperature, and density
of the particle cloud are equal at corresponding positions.
3. The mass fraction of nitric oxide is directly pro-
portional to the combustor length at corresponding positions.
(This scaling rule is affected by the non-ideal effects of
radiation, molecular dissipation of turbulent eddies, and
non-equilibrium chemistry.)
4. The heat fluxes, shear stresses, and mass diffusion
rates are equal at corresponding positions.
5. The effective turbulent viscosity is proportional
to the characteristic density PQ, velocity UQ, and combustor
length L.
6. The source term effect is inversely proportional
to the product of PQ and UQ in the mass and energy conservation
equations and to the product of PQ uQ2 in the momentum equations,
50
-------
and these effects are all directly proportional to L. Hence
the effects of these source terms (gravitational force, finite-
rate chemical reaction, and thin gas radiation, etc.) on the
flowfield can be simulated in subscale models by employing smaller
PQ and UQ and by taking the dependencies, if any, of the source
terms on p and u into account.
o o
7. In oil or coal fired units, the rule of varying
the particle size with the square root of combustor length
provides for similarity in gas-particle mass, momentum, and
energy transfers.
8. The reference velocity u need not be maintained
equal for subscale and fullscale, as long as the velocity ratios
at corresponding boundaries are kept equal and the kinetic energy
dissipation and pressure variation are small. The velocity com-
ponents normalized by u are then still similar in the combustors.
9. The reference density p or the reference pressure
p need not be maintained equal for subscale and fullscale, as
long as the density or pressure ratios at corresponding boundaries
and the reference temperature T (and hence the ratio p /p )
are kept equal. In this case, the local density and pressure
normalized by p and p , respectively, remain similar between
subscale and fullscale.
10. Radiation, in optically thick conditions, has
negligible influence on the scaling of turbulent flowfield and
nitric oxide. Under thin gas conditions, however, ratiation
exerts greater effect on larger combustors and affects the
scaling of nitric oxide in a nonlinear manner.
51
-------
3. DERIVATION OF SCALING RELATIONS
The scaling criteria will be derived from the conservation
equations for two-dimensional turbulent flow for simplicity.
The results obtained are applicable to three-dimensional flows
as well. In this section, attention will be focused on gaseous
turbulent diffusion flames. Accounts for two-phase flow,
radiation, finite-rate combustion, molecular transports, etc.
will be pursued in a subsequent section.
3.1 Governing Equations
The conservation equations governing the flow of a
compressible reacting gas can be found in textbooks (e.g., Gosman
et al. (1969)). For steady plane or axisymmetric two-dimensional
flows, these equations can be written in the form
mass:
x-momentum:
f-1^
f n
y-momentum:
6-momentum:
I ^
energy:
- r,,/
species:
»r
52
fu
-------
where a = 0 for planar flow and a = 1 for axisymmetric flow.
The velocity components, u, v, and w are in the directions of
x, y, and 6, respectively, which denote -the axial, vertical
or radial, and azimuthal (for a = 1) coordinates, respectively.
For rectangular geometry or in the absence of swirl for cylin-
drical geometry, the 6-momentum equation is not needed since
'b
w is then zero everywhere. The symbols p, p, h, and m. denote,
respectively, density, pressure, specific stagnation enthalpy,
and mass fraction of chemical species i. Also, q and j.
where a = x, y, or 0 denote heat flux and diffusion flux of
species i, respectively, in the direction a; and T where m, n =
x, y, or 8 denotes shear stress in the plane perpendicular to
the m axis and in the direction parallel to n. The R represents
a mass source due to particle vaporization. The P and F repre-
x y
sent momentum sources due to body forces, particle drags, etc.;
Q is an energy source accounting for thermal radiation, particle
heat transfer, etc.; and R. is a mass source for species i due
to chemical reaction, particle vaporization, etc. These "source
terms will be left unspecified at this point.
<\j
The relations between stagnation enthalpy h, enthalpy
h, and temperature T are given by
h r h H-i^"2*"2) fa
h = f wt- He ft)
where h. is the reference enthalpy of species i at temperature
T , and c . is the constant-pressure specific heat of species i
The equation of state is taken as
P = R T
53
-------
where R is the gas constant of the gaseous mixture. In addition,
the following relations may be used:
03)
f
aw
57 05)
66)
67)
where y, k, and D± denote viscosity, conductivity, and diffusion
coefficient of species i, respectively.
-------
The conservation equations given above are strictly
valid only for laminar flows. At present, no rigorous and
generally successful theory governing turbulent recirculating
flows is available. The simplest approach is the one taken by
Gosman et al. (1969) which merely replaces the molecular transport
coefficients u, k, and D^ by effective values. One may note that
the differential equations for this approach are slightly dif-
ferent from those obtained by taking time-mean values of the
conservation equations. For example, in the Gosman approach,
the first term in equation (1) is simple 3/3x(pu) where the bar
indicates time-mean values. On the other hand, if one takes
time-mean of the term 3/3x(pu), one obtains 3/3x(pu +p'u' )
where the prime indicates fluxtuating quantities. The differences,
however, are generally small and Gosman *s approach is employed
here for simplicity. Furthermore, the discussions here apply
to either approach. The expressions for the effective transport
coefficients will be specified later.
3.2 Scaling of Flow Properties and Nitric Oxide
The following non-dimensional variables are defined:
(J =
H = JiAo £ =
*
P " ?/?o . f=f/fo (24)
(25)
where the subscript o refers to a reference value and L is a
55
-------
characteristic length. The governing equations can then be
written in the form
-
, r/- r«) +
f i
-/*«o
where
p* r f*R
56
-------
-------
means that, under the stated conditions, the flowfields are
completely similar for combustors of various sizes; i.e., the
properties p*, p*, U, V, W, H, and m. are equal at corresponding
positions (?,n) of various geometrically similar combustors.
For combustors containing turbulent diffusion flames, R^ (i = fuel,
oxidant, and combustion products) is zero except at the flame
front where it becomes infinite. The position of the flame front
is similar at corresponding £ and n for equal stoichiometric
fuel-oxidant ratio, and hence R. is independent of L. Thus,
the local mass fractions of combustion reactants and products
can be simulated. This occurs because the combustion chemistry
is extremely rapid and the primary reaction products which dominati
the fluid dynamics are in essential equilibrium. This is not true
of contaminants whose rate of formation (NO) or destruction (CO)
are kinetically controlled. Nitric oxide (NO), being a trace
species, has negligible influence on the flowfield and is con-
sidered separately from the flow variables. If the NO concen-
tration m^0 is far below the equilibrium value as is generally
the case, its formation rate R-j0 is independent of ICL,Q and
equation (31) shows m^ to vary linearly with the geometric size
of the combustor.
The third important observation is that under certain
circumstances, even some boundary conditions need not be
similar in order to produce similar non-dimensionalized properties,
For example, in many types of combustors such as power plant
boilers, the pressure is practically uniform and the kinetic
energy is small compared to thermal and chemical energies. The
pressure gradient terms in equation (27) and (28) then disappear.
Provided that Re is independent of UQ in addition to L, a con-
dition which will be shown to be valid, and provided that the
effects of the source terms on the flowfield are small, the
non-dimensionalized flow properties are then independent of u .
That is, if the reference velocity UQ is different between
58
-------
subscale and prototype, the normalized values of U, V, W, H,
and m.^ (i = fuel, oxidant, and combustion products) are still
similar at corresponding non-dimensionalized positions (€,n)
between subscale and prototype. It is interesting to observe
from equation (31) that nitric oxide concentration m^o/ however,
is inversely proportional to UQ as well as being directly pro-
portional to L. As an illustration, consider applying the
scaling rule to the study of Quan et al. (1972) on nitric oxide
formation in the turbulent diffusion flame formed between semi-
infinite plane streams of fuel and oxidant. There, the scaling
length is the distance downstream, x, and the scaling velocity
may be taken as the fuel velocity, u,. Then, for fixed air-
fuel velocity ratio, the scaled nitric oxide mass fraction
profile, itL,Q u,/x, across the mixing layer is independent of
u, or x. This is shown in Fig. 1. Thus even nitric oxide
concentration profiles may be simulated by changing the injec-
tion velocities in proportion to the scale of the combustor,
providing that one remains within flame stability limits,- etc.
As another example of changing the boundary condition
in scaling, consider changing p or p between model and proto-
type (but keeping h , and hence the ratio P0/PO/ the same),
then the non-dimensionalized equations without the source terms
are still similar. The advantage of this maneuver is that, by
changing the pressure in a model combustor, one may counteract
the effect of changing size so that finite-rate chemical reactions,
for instance, may be scaled as well as the flow properties.
In view of the large number of parameters associated
with combustion problems, it is somewhat surprising that a two-
dimensional combustor can be simulated or modeled, at least
ideally, with so few restrictive conditions. The success can be
attributed mainly to the fact that, in the regions of turbulent
flow, the Reynolds number Re does turn out to be independent of
59
-------
p , u , and L. That is, the effective turbulent viscosity is
Ko' o
determined by the flow and is proportional to the density, the
convective velocity, and the characteristic dimension. This is
the reason for the well known experimental observation on
turbulent diffusion flames that the flame length to jet orifice
diameter ratio is independent of the jet velocity, density,
and orifice diameter. It should be noted, however, that al-
though this length ratio is independent of either the fuel or
the oxidant velocity and density, the fuel/oxidant velocity and
density ratios themselves must be kept constant in order to
achieve similarity in non-dimensionalized boundary conditions.
It may also be interesting to note that for laminar flow, the
Reynolds number Re is, in contrast to turbulent flow, directly
proportional to p , u , and L. Thus, for example, if a model
of one-tenths of the prototype size is used, then the velocity
must be increased by a factor of ten in order to keep Re the same!
At least in this sense, then, a turbulent combustor is easier
to model than a laminar combustor.
Of course, even a turbulent combustor contains regions
of laminar flow near the walls and the dissipation of turbulent
eddies, which has been shown by Quan, et al. (1972) to have a
dominant influence on the amount of nitric oxide formed, involve
molecular processes. Corrections and additions to the above
scaling relations in order to account for the molecular processes,
as well as for the source terms due to two-phase flow, etc., are
discussed in a later section. In the remaining parts of the
present section, the scaling of the effective turbulent viscosity
and the application of boundary conditions will be discussed.
3.3 Turbulent Transport
The scaling of effective turbulent viscosity will be
analyzed for three commonly employed and representative models
of turbulent flows. It will be shown that the viscosity is
-------
proportional to the characteristic density p , velocity u ,
and length L; and that the proportionality coefficient, and
hence the Reynolds number Re as defined by equation (38), is
a function of only the normalized boundary conditions and
the normalized positions (£,n) and is thus independent of p ,
UQ, and L. In regions of turbulent flow, the effective turbu-
lent Prandtl and Schmidt numbers, Pr and Sc., are essentially
invarient between model and prototype and hence will not be
discussed further.
Consider the expression for the effective turbulent vis
cosity given by Gosman et al. (1969) for turbulent diffusion
flames in recirculating flows:
*/3 -1/3 2/3 ,. 2 - 2v'/3
/t = KD W p (nfVfZ+ "M )
where K is a constant, D the combustion chamber diameter, W the
chamber length, m a mass flow rate, and V the velocity; the
subscripts F and A denote conditions at the fuel and oxidant
inlets, respectively. Taking pp, and VF/ and D to be the
characteristic density p velocity u , and length L, respectively,
one obtains .
>« - KAu.L $*)* tiff (I* %$)*
where d is the diameter of the fuel inlet orifice. For combus-
tors of similar geometry, the ratios W/D and d/D are constant;
and for similar boundary conditions, the mass flow ratio m,/fti-.
and the velocity ratio V /V_ are also constant. Hence the
Reynolds number, pouQL/y, is independent of pQ, UQ, and L as
postulated and varies only with p* which is similar at corres-
: ponding locations of model and prototype.
Consider another common model of effective turbulent
viscosity y, namely, the mixing-length theory for parabolic
61
-------
or boundary layer flow. Here,
/*. - fJLZ ?"
where Jt is the mixing length given by
/ r CX (46)
where c is a mixing constant. Equation (45) can be written as
Since £2 and 3U/3n are similar between model and prototype,
the similarity requirement that y/pouQL be independent of
p , u , and L is again satisfied.
As a third illustration, consider the approach of
Gosman et al. (1969) in which y is given by
where c approaches a constant for highly turbulent flow, and
k and j£ denote the turbulence kinetic energy and the turbulence
length scale, respectively. In this approach, k and j£ are
assumed to be governed by differential equations containing
convection, diffusion, and source terms. Examination of their
o
equations show that the ratios k/uQ and j0/L are independent of
PO,UQ, and L for highly turbulent flow. Hence equation (48),
written in the form
c* to)
/
O ^£.
where k* = k/uQ and ji = jf/L, shows that y/p u L is also
independent of pQ, UQ, and L for this method of approach.
It is believed that any rigorous turbulence model will yield
this scaling relation and the above scaling is thus universal.
62
-------
3.4 Boundary Conditions
In the elliptic differential equations governing recir-
culating flow, the values of the gradients of U, V, W, H, mi/ and
p* or p* must be prescribed at all boundaries. To achieve
similarity, these boundary conditions must be equal at corres-
ponding positions of ?b and nb between subscale and fullsccale,
where the subscript b denotes boundary positions.
For diffusion flames where the fuel and oxidant are
injected separately into the combustion chamber, similar boundary
conditions imply that at the inlet the fuel-oxidant ratios of
velocities, temperature or enthalpy, and density or pressure
must be maintained invarient between subscale and fullscale.
The normalization values of u , h , and p or p may be chosen
to be the velocity, enthalpy, and density or pressure, respectively,
of either the fuel or oxidant stream. As indicated earlier,
under the conditions that the pressure variation is small and
the kinetic energy dissipation is negligible, u need not be
equal between subscale and fullscale to achieve similarity.
Also, p and p need not be kept the same as long as h is.
If a k-and-/ type of turbulence model is used to
evaluate y, then the differential equations for k and JL require
additional boundary conditions. Similarity requirement shows
9
that k/u and JL/l> must be similar between subscale and full-
scale at corresponding boundary positions.
63
-------
4. SOURCE TERMS AND MOLECULAR TRANSPORT
The source terms in the non-dimensionalized momentum,
energy, and species equations are all proportional to L, i.e.,
the source term effects on the flow properties are greater for
larger combustors. The source terms due to particle-gas inter-
action, gravitational forces, finite-rate chemistry, and
thermal radiation are discussed in this section. Accounts for
laminar or molecular processes and other factors that may
influence nitric oxide scaling are also discussed.
4.1 Oil and Coal Combustion
For the particle phase, the conservation equations have
the following form
-------
Letting
one may write equations (50) to (54) in the form
- - f
"r V + r n WtW - - -
where
Here again, if the source terms can be made inversely propor-
tional to L, then the particle properties, like the fluid pro-
perties, are seen to be independent of L. This scaling possi-
bility is investigated below.
To consider the particle-gas interaction terms of R ,
F , F , and Q , the effects of turbulent fluctuations on
particles will be neglected. First, consider oil particles
in Stokes flow regime (small particle Reynolds number). Since
R is proportional to the particle number density, which is
proportional to Pp/rp where r is the particle radius, and to the
vaporization rate of a single particle, which is proportional
65
-------
to r , one obtains
Thus, in order to have the right-hand-side of equation (59)
independent of pQ, UQ, and L, one may choose the particle
size such that
The particle momentum sources, from the work of Marble
(1969) , may be shown to be
Since R has the form of equation (65) , one obtains
to;
Thus, in order to have the right-hand-side of equations (60)
2
and (61) independent of PQ, UQ, and L, one needs (L/pQuo ) ^
(r /p U ) and this yields the identical relation as equation
(66).
Similarly, the particle energy source has the form
, U0L/fFpx +
Since R , F , and F have the forms of equations (65) and
(69), one obtains
<3 - f. /* , A *?/*? . f^o / (71 )
Equation (71) shows that Q * 1/r 2. Although u and h must
remain invarient in order to have a strictly valid scaling law,
the contribution to the energy source by the particle drag
forces are small in many instances. In these cases, the second
-------
term on the right-hand-side of equation (71) need not be
considered and, keeping hQ invariant, one obtains from equa-
tion (63) the scaling relation of (L/p u ) ^ (r 2/P ) which
yields again the identical relation as equation (66).
Thus, perhaps fortuitously, the scaling law of r ^ L
satisfies similarity in vaporization, momentum transfer,
and energy transfer as well. In addition, if kinetic energy
dissipation is negligible, u can also be employed for scaling
2
and the rule becomes r ^ L/u . It may be noted that scaling
p o
for two-phase flow in laminar boundary layer is not feasible
because the convection and diffusion terms there have different
scaling lengths.
Two more aspects must be considered for particle scaling.
One is that -the particle radius must remain scaled as vaporization
occurs, i.e., r /r must be independent of L and u at given
position of £, n. The particle radius is governed by
where p denotes the particle bulk density, and n is the number
of particles per unit volume given by
nr *
Equations (72) and (73) become
Up
where r = r /r Since r n is chosen such that R *> p u /L,
p p po po p o o
equation (74) shows that r * is indeed independent of u and L.
The second aspect to be considered is that of non-Stokes
flow. In most regions of oil fired combustors, the particles
move in the vapor stream that originates from the droplet surfaces
and thus the relative velocities between particles and vapor
are small. Consequently, the particle size scaling rule of
67
-------
r 'u L/U derived for Stokes flow is a good approximation
in the dominant regions. However, if one is primarily interested
in simulating the initial region of a combustor where the
particles are injected into the air stream and where the relative
velocities between particles and air are large, then the source
terms for particle vaporization, drag, and heat transfer. are all
multiplied by a factor of approximately (1 4- 0.276 Pr^ ^' Rep ' )
where Pr, is the Prandtl number based on molecular properties
of the gas and Re is the particle Reynolds number defined by
Re - 2 C^-Wp)% (v-^ i
= 2 Koo L(U- Uf
where y^ denotes molecular viscosity of gas. Thus for moderate
particle Reynolds numbers, the scaling relation is given by
o. Z76 ?
pz (\
32
which may be approximated by r ^ L p /u for high particle
Reynolds numbers .
Having investigated the modeling of oil droplet combus-
tion by scaling the droplet size, it may be interesting to
consider solid particles such as coal. Here, the gas-particle
momentum and energy transfers are similar to those for oil and
hence the particle size scaling still applies. The question
concerns the scaling of the combustion rate.
The combustion of coal may be separated into two modes.
One is the gasification of volatiles. This is an internal decom-
position process which, from Field et al. (1967), may be
described by
Rr = ^"^/-/p c ' (77)
68
-------
where GI is the mass fraction of volatile matter in the coal,
C2 is a characteristic decomposition time, and C, is a charac-
teristic temperature. Equation (77) shows that R is inde-
pendent of particle size. To scale the right-hand-side of
equation (59), one needs
Thus, to simulate gasification of volatile matter in subscale
models, one needs to substitute a different volatile matter of
shorter decomposition time and/or to decrease the flow velocity
u . The scaling requirement given by equation (78) arises,
because decomposition is rate-limited and is thus in contrast
to the assumption of diffusion-limited combustion which is
listed as assumption 12 in Section 2.
The other combustion mechanism of coal is the burning
of the char structure. From Davis et al. (1969), one may take
the burning rate for a cloud of particles as
where K, , K0, and K_ may be considered as fixed constants, Xn
\- £. 3 ")
is the mole fraction of 0, at the edge of the particle boundary
layer, and f is the steric factor. The first and second terms
in the denominator of equation (79 ) account for resistances on
surface reaction rate and on counter-diffusion rate of COj and
O9 near the surface, respectively. To satisfy scaling of mass
transfer, equation (59) requires R *> PQUQ/L; and to satisfy
momentum and energy transfers, the scaling rule of r ^ L/uQ
as given by equation (66) is required. The combination requires
R i> o /r . Examination of equation (79) shows that this con-
P ° P
dition is satisfied if the combustion of char is diffusion-
controlled, which is a valid approximation for large particles
69
-------
and/or high pressure. If the surface reaction rate controls,
however, then one must either accept the result of R ^ r
given by equation (79) as an approximation to the scaling rule
-2 "*1
of R ^ r or increase p0or pQ such that pQor pQ ^ r
in order to obtain R ^ r .
For high particle Reynolds numbers, correction for
their effects on the scaling rules may be made in a manner
similar to that given for oil.
4.2 Molecular Transport
Molecular processes affect the scaling of turbulent
flow in three major respects. One is that heat transfer to
the walls are governed by laminar diffusion in the wall regions.
Another is that the dissipation of turbulent eddies, which has
been shown by Quan et al. (1972) to have a dominant influence
on the amount of nitric oxide formed, is governed by molecular
processes. The third is that the gas-particle interaction in
two-phase flow, which has already been considered, is dependent
on molecular properties.
To correct for the heat transfer in the wall regions of
a model, one may balance the heat flux through the laminar
boundary layer by the turbulent heat flux at the edge of this
layer at corresponding positions of £,n. That is, h(T - T )
6 W
= (k9T/3y) where h and k are the convective heat transfer
coefficient and turbulent conductivity, respectively; and the
subscripts e and w denote the edge of the laminar layer and
wall, respectively. Since k3T/3y is proportional to p u and
independent of L and since h may be represented, for example,
by h ^ (P0uo)°'8L~°'2, one obtains
Since Tg is similar (i.e., independent of pQ, u , and L) , equa-
tion (80) provides a scaling relation for T and shows that a
inf
70
-------
subscale model requires higher wall temperatures to keep the
temperatures in the turbulent bulk gas region similar to
those of the prototype. The 0.2 power in equation (80) may
become significant when the model and prototype length ratio
is large.
On the molecular dissipation of turbulent eddies, the
mechanism is not well established at present. Here, one may
merely indicate that no additional requirement for similarity
appears for both limiting cases of no molecular mixing and
complete mixing of the eddies. The situation is analogous to
chemical reaction where no time or length scale is introduced
if the flow is either frozen or at equilibrium, but finite-rate
process would introduce a time scale.
4.3 Thermal Radiation
Radiation affects the formation of NO in a nonlinear
manner, and will be discussed here for the optically thick and
optically thin limits, equations for which are given by Vincenti
and Kruger (1965).
The thick gas approximation is valid when the radiation
mean free path is much smaller than the flame dimensions. In
this case, radiation is treated as heat conduction with the
conductivity given by
"J
where a is the Stefan-Boltzmann constant, and M. and KR are
the mole fraction and Rosseland mean absorption coefficient,
respectively, of species j. The gaseous species from hydrocarbon
combustion that participate in radiation are mainly H2O, C02,
and CO for which the values of KR are given by Abu-Romia and
Tien (1967) as functions of pressure and temperature. The
ratio of radiation conductivity to turbulent conductivity shows
71
-------
that k A ^ (P u L KR) and is generally small except for
small combustors. In typical combustprs, then radiation in the
thick-gas regime need not be considered because radiation con-
duction is small compared to turbulent conduction.
The thin gas approximation may be applied when the
radiation mean free path is much greater than the flame dimen-
sions. Here, radiation is treated as an energy sink in the
energy conservation equation with
QA = - 4 T T* ? M
where K is the Planck mean absorption coefficient of species j.
Values o? K are also given by Abu-Romia and Tien (1967), and
vary with temperature and are proportional to pressure. To
have the source term in equation (30) independent of p , u , and
L, one needs K ^ p u h /L. Since < is proportional to p and
hence p , scaling of thin gas radiation is achievable by
employing u ^ L.
The potential effect of radiation on NO formation has
been assessed, and the differences in flame temperature history
are shown in Fig. 2. This configuration corresponds to the
sample problem investigated by Quan et al. (1973) for nitric
oxide formation in recirculating flows.without radiation. In
the absence of radiation, the calculated results show that the
combustor of 32 ft in length forms 327 ppm of NO, that a subscale
combustor of 3.2 ft in length forms 32 ppm of NO, and that the
turbulent flowfields of these two combustors are similar. Thus,
the scaling relations for both the flowfield and NO production
are observed. With thick-gas approximation, radiation is found
to have negligible effect on the flowfield and NO production.
With thin-gas approximation and considering each gas element
only to emit radiation and not to absorb any, however, radiation
is found to have a strong influence on the temperature field and,
consequently, NO production; although its effect on the velocity
field is still small. For the larger combustor, NO is reduced
72
-------
from 327 ppm to 71 ppm; whereas for the smaller combustor for
which the effect of thin-gas radiation is smaller, NO is reduced
from 32 ppm to 23 ppm. The effect of thin-gas radiation on the
flame temperature distribution is illustrated in Fig. 2. Thus
if a combustor is optically thin or if the subscale combustor
is optically thin while the fullscale combustor is optically
thick, radiation differences between model and prototype will
cause nonlinear NO scaling.
4.4 Gravity and Finite-Rate Chemistry
Gravitational forces are proportional to p g. For
O 2
absolute similarity, equations (28) and (29) require Lg * u
In industrial combustors, gravitational effects are generally
negligible and gravitational scaling is generally unimportant.
Finite-rate kinetics may be simulated, according to
the source term in equation (31), by requiring that R. ^ p u /L.
Thus if the species production rate R. is independent of density
p , one may simply take p u ^ L. If R. is dependent on pressure
p or density p , then one may change the pressure such that
R./P ^ u /L. In the case of NO formation, if the NO concen-
tration is low so that R. is independent of NO, no change in
operating condition is necessary and equation (31) shows that one
may simply scale the NO concentration in direct proportion to L
except for radiation effects.
Besides radiation, there are other factors which may
render the scaling of NO with combustor length nonlinear. If
NO is near equilibrium or if there is significant conversion
of NO to N02, then R..Q is a function of m^, and thus mNO does
not vary linearly with L. In two-phase flow, fuel nitrogen
effects may become important, e.g., there may be a certain
fixed amount of NO formed due to fuel nitrogen regardless of
combustor size. The kinetics of fuel nitrogen conversion is
not well known; but if it is found that a fixed amount of fuel
73
-------
NO is formed, then one may simply sub tract this amount and
scale the remainder of the NO. Keeping the total NO level
high also tends to minimize the fuel nitrogen effect. The
question of "prompt NO" is associated with non-equilibrium oxygen
concentration near the flame front. As indicated in the pre-
ceding paragraph, one way of scaling this finite-rate chemistry
effect is to set u ^ L. However, for small L in subscale, one
must be certain that u is not so low that the flow becomes
o
laminar or unstable.
74
-------
5. DISCUSSION AND SUMMARY
The reason that turbulent flowfields can be scaled in
geometrically similar combustors is that the turbulent transport
coefficients are proportional to density pQ, velocity UQ, and
geometric dimension L. As a result, the ratio of diffusive
transport to convective transport becomes independent of p ,
UQ, and L. This is the reason that, for example, the flame
length to jet diameter ratio in turbulent diffusion flames does
not vary with jet diameter, density, or velocity as long as the
air-fuel density and velocity ratios are kept constant.
The effects of the source terms in the conservation
equations are all proportional to L and inversely proportional
to p u . Hence one may scale the source terms by simply
requiring p u ^ L. In modeling, it may not be practical to
change p , and the range of u that may be modified may also
be limited by the considerations of flame stabilization, flow
laminar iz at ion, etc. In principle, however, the fact tha't equal
absolute velocity is not a necessary condition for similar flow-
field is due to two main reasons. One is that kinetic heating is
negligible so that enthalpy is practically independent of velocity.
The other, again, is that the diffusive transport coefficients
are proportional to velocity. The velocity ratios at corresponding
boundary positions of subscale and fullscale must be kept equal,
however, in order to achieve similarity in non-dimensionalized
boundary conditions. This rule applies to the density ratios
as well.
Another interesting aspect which contributes to similarity
in turbulent flow is that the shear stress, heat conduction flux,
and mass diffusion flux are independent of combustor size. For
example, the temperature gradients are much steeper for subscale,
but the heat conduction coefficient is much smaller by the same
proportion and the net result is that the heat flux at corres-
ponding positions of subscale and fullscale are the same. If
75
-------
the absolute velocities are not the same, however, the fluxes
will differ accordingly.
For oil and coal combustion, one may scale the particle-
gas momentum and energy transfers by scaling the particle radius
as r 2 ^ L for fixed POUQ. For oil, this scaling also insures
proper scaling of vaporization rate. For coal, additional scaling
2
of the burning rate is required. The r ^ L is applicable
\~
for Stokes flow regime. For moderately high Reynolds numbers,
it becomes r * ^ L.
P
The particle size scaling for oil provides for equal
ratio of all three rates: fuel vaporization, vapor and oxidizer
convection, and vapor-oxidizer diffusion. The vaporization-
diffusion ratio is important in ensuring a similar combustion
mode. For example, if vaporization rate is too slow, oxidizer
may diffuse into the cloud of droplets and combustion can occur
mainly around the drops instead of mainly near the vapor-oxidizer
interface as experimentally observed. In combustors that are
only partially similar in geometry, the proper characteristic
length L should be chosen as the diameter of the vapor cloud where
a diffusion flame occurs.
It is interesting to point out here that for laminar flows,
the vaporization-diffusion ratio requires equal r /L while the
vaporization-convection ratio still requires equal r /L. Thus, a
consistent scaling law does not exist for two-phase laminar
boundary layers, whereas it does exist for turbulent boundary
layers as well as for inviscid flows since both vaporization-
diffusion (for turbulent flow) and vaporization-convection ratios
2
are then satisfied by equal r /L. It should be mentioned however,that
the effects of turbulent fluctuations on gas-particle interactions
have not been considered here.
For nitric oxide, since the source term in its continuity
equation is scaled with L/PQUO/ the NO concentration varies
76
-------
linearly with geometric size for given density and velocity.
This conclusion holds for all two or three dimensional flows.
This linear relation, however, is modified if radiation and
other nonlinear effects become important.
In terms of load or power input to a combustor, NO
concentration varies with the square root of load since load
varies with the square of the inlet dimension for axisymmetric
combustors. Load by itself is not a good scaling parameter,
however, because it is a function of two parameters: geometric
size and velocity. If the velocity is maintained constant, then
nitric oxide increases linearly with one-half power of load. If
the geometry is fixed (for same size combustors), however, NO
may increase, decrease, or does not change with varying load.
These conflicting trends are substantiated by experimental data,
and can be accounted for by the trade-off between the effect
of residence time and the effect of increasing power output as
velocity is increased. In general, the effect of increase in
power output, which increases wall temperatures and hence flow-
field temperatures, dominates and NO increases with loading.
However, this trend certainly should not be attributed to
flame length which is independent of velocity for given fuel-
oxidizer velocity ratio.
77
-------
NOMENCLATURE
c specific heat at constant pressure
D effective diffusion coefficient
E (u2 + v2 + w2)/2 hQ
F momentum source
g gravitational acceleration
H h/no
h specific enthalpy
h specific stagnation enthalpy
j mass diffusion flux
j* j/P u
k effective heat conductivity
L characteristic combustor geometric dimension
M molecular weight
m mass fraction
Pr Prandtl number, yc /k
P
p pressure
P* P /P0
Q energy source
q heat diffusion flux
q* <3/P0uoho
R mass source
Re Reynolds number, p u L/JJ
R gas constant
r radius of particle
78
-------
Sc Schmidt number/ y/pD
T temperature
U U/UQ
u axial velocity
V v/uo
v vertical or radial velocity
W W/W0
w swirl velocity in axisymmetric flow
x axial distance
y vertical or radial distance
Greek Letters
n y/L
8 azimuthal coordinate
< Planck mean absorption coefficient
KR Rosseland mean absorption coefficient
y effective viscosity
€ x/L
p density
P* P/P0
p particle bulk density
T shear stress
o =0 for plane flow, = 1 for axisymmetric flow;
Stefan-Boltzmann constant
79
-------
Subscripts
o reference point
i index for chemical species
m index for x, y, 6
n index for x, y, 6
NO nitric oxide
p particle
r reference state
x in x-direction
y in y-direction
a index for x, y, 9
0 in 6-direction
80
-------
REFERENCES
Abu-Romia, M. M. and Tien, C. L., 1967, Appropriate Mean
Absorption Coefficients for Infrared Radiation of Gases,
J. Heat Transfer, 89, 321.
Davies, T. W., Beer, J. M. and Siddall, R. G. , 1969, The Use of
a Mathematical Model for the Prediction of the Burn
Out of Char Suspensions, Chem. Eng.Sci. 24, 1553.
'Field, M. W., Gill, D. W., Morgan, B. B., and Hawksley, P. G. W.,
1967, Combustion of Pulverized Coal, Brit. Coal Utilization
Res. Assoc.
Gosman, A. D., Pun, W. M., Runchal, A. K., Spalding, D. B., and
Wolfshtein, M., 1969, Heat and Mass Transfer in Recir-
culating Flows, Academic Press.
Marble, F. E., 1969, Some Gasdynamic Problems in the Flow of
Condensing Vapors, Astronautica Acta, 14, 585.
Penner, S. S., 1955, Similarity Analysis for Chemical Reactors
and the Scaling of Liquid Fuel Rocket Engines, Combustion
Researches and Reviews, Butterworths Sci.Pub., pp 140-162.
Quan, V., Marble, F. E., and Kliegel, J. R., 1972, Nitric Oxide
Formation in Turbulent Diffusion Flames, presented at
the Fourteenth Symposium (International) on Combustion,
and to be published in Symposium Volume.
Quan., V., Bodeen, C. A., and Teixeira, D. P., 1973, Nitric Oxide
Formation in Recirculating Flows, to be published in
Combustion Science and Technology.
Spalding, D. B., 1963, The Art of Partial Modeling, Ninth
Symposium (International) on Combustion, Academic
Press, pp 833-843.
Vincenti, W. G. and Kruger, C. H. Jr., 1965, Introduction to
Physical Gas Dynamics, John Wiley and Sons.
81
-------
0.06
P - 1.0 ATM
u . /u, , =0.25
air' fuel
T . - 1210 °R
air
537 °R
fuel
Fuel: Methane
FLAME FRONT
BASED ON NO MIXING
OF TURBULENT EDDIES
BASED ON COMPLETE MIXING
TURBULENT EDDIES
Fig. 1.
0.2 0.4 0.6 0.08 1.0
DISTANCE FROM AIR BOUNDARY/MIXING ZONE WIDTH
Scaled Nitric Oxide Concentration Profile in
Turbulent Diffusion Flame Between Plane Streams
of Methane and Air.
82
-------
TEMPERATURES (°R) ALONG STOICHIOMETRIC LINE
POINT:
NO RADIATION:
RADIATION, SUBSCALE:
RADIATION, FULLSCALE:
PEAK TEMPERATURE CONTOUR
(NEAR STOICHIOMETRIC LINE)
32.0 FT FOR FULLSCALE,3.2 FT FOR SUBSCALE.
Fig. 2. Effect of Thin Gas Radiation on Flame Temperatures.
-------
FUNDAMENTAL RESEARCH
PART II
85
-------
RELATIONSHIP OF BURNER DESIGN TO
THE CONTROL OF NO EMISSIONS
THROUGH COMBUSTIONXMODIFICATION
by
D. W. Pershing
J. W. Brown
E. E. Berkau
U. S. Environmental Protection Agency
Office of Research and Development
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
87
-------
ABSTRACT
The combustion of propane, distillate oil, a 0.3 percent nitrogen
oil, and pulverized bituminous coal has been examined in a versatile
laboratory furnace. In each case testing was conducted to determine the
relationship of burner (and process) parameters to the control of NO
A
emissions. The results show that in general propane and distillate oil
give about the same NO . the 0.3 percent N oil about twice the NO, and
X A
coal at least 2-1/2 times as much. Increased burner throat velocity
and flue gas recirculation were shown to be extremely effective in
reducing thermal NO , but neither worked very satisfactorily with the
s\
high nitrogen oil or coal. Increasing air preheat substantially
increased the NO emissions from propane and distillate oil and caused
X
lesser increases with the high nitrogen distillate and pulverized
coal. Future work will extend the program to natural gas and No. 6 oil
and will consider two stage combustion.
This paper summarizes results obtained under ROAP 21ADG-Task 42
(in-house) under the sponsorship of the U. S. Environmental Protection
Agency. The work reported herein was completed June 13, 1973.
88
-------
INTRODUCTION
BACKGROUND
Several groups have recently noted the importance of burner
parameters in NO formation. Some of the earliest work was done by
/\
Wasser et al. in a 3 GPH refractory lined test furnace. Their data
indicated that with distillate oil NOV emissions increase with firing
A
rate, decrease when the excess air is increased from 25 to 45 percent, and
are most strongly a function of burner swirl. With the burners at full
load, Wasser was able to vary NO from 130 ppm to 480 ppm (no air preheat)
by changing only the amount of swirl in the combustion air stream. The
data were correlated using cold flow residence time distribution data. From
this, it was shown that the large changes in NO were due to changes in
primary zone combustion intensity induced by the effect of swirl on the
combustion fluid dynamics.
The most comprehensive study of burner parameters to date was conducted
p
by Heap et al. at the International Flame Research Foundation in a 2 x 2 x
6.25 m refractory wall furnace. The variables considered included fuel
injector design and position, secondary air velocity, quarl type and angle,
burner swirl, and fuel types (gas and coal). Heap found that for natural
gas flames when the fuel jet remains on the flame axis, NO emissions increased
with increasing swirl. Increasing the combustion air velocity from 25 m/sec
(82 fps) to 50 m/sec (164 fps) reduced NO emissions. In natural gas flames
where the internally recirculating gases form a closed zone on the flame
axis (as with a radial hole injector) NO emissions generally decreased with
increasing swirl; however, increasing the combustion air velocity again
89
-------
reduced NO. In all cases the decrease is probably due to the entrainment
of cooled product gases.
In coal flames with strong axial fuel jets, Heap found that increasing
swirl usually decreased the emission of NO . Increased secondary air
A
velocity usually increased emissions. In flames from radial coal injectors
(where there is reverse flow on the centerline) the NO emissions were
/\
essentially independent of swirl. In general, the coal data suggested that
any change which spread the pulverized coal jet or increased the 0,,/fuel
4
ratio near the injector increased NO emissions.
A
Hemsath et al.5 investigated the importance of proper burner design
for low emissions from large industrial natural gas furnaces and found the
key to a low emission burner was reducing the combustion temperatures
followed by rapid mixing of the combustion products with the surrounding
furnace gases. Seven commercial burners were tested in a refractory lined
chamber and based on the results a new "low emission" burner was designed.
The NO emissions from the commercial burners ranged from near 100 ppm at
A
fuel rich or large (> 50 percent) excess air conditions to over 400 ppm
at 7 percent excess air. Emissions from the new low emission burner were
50 to 75 percent lower than those from the conventional burners. While
the exact design of the new burner is proprietary, available information
indicates that emission reduction is achieved through the use of a very
small burner block for ignition stability and very high velocity combustion
air (small burner throat). The latter results in significant entrainment
of cooler flue products and in effect provides aerodynamic flue gas
recirculation.
90
-------
Shoffstall et al. examined the importance of burner design by
testing five different types of natural gas burners. In each case, the
effect of excess air, load, fuel injector design, and air preheat temperature
on NO emissions was determined. Burner/injector pairs, where the fuel jet
A
remained on the axis, gave long luminous flames with low NO emissions.
A
Increasing the excess air in these cases increased emissions up to at least
20 percent excess air (4 percent Og).
Burners with high swirl or radial gas injection had internal reverse
flow (of combusted products) on the centerline and short, very intense flames.
In these cases NO emissions were generally higher than the axial
A
flames and, based on in-flame measurements, nearly all the NO was formed
A
very near the burner, many times within the burner block itself. These
burners also increased in NO output as excess air was increased but peaked
X
much earlier,^generally before 15 percent excess air (3 percent 02). The
earlier peak and higher emissions may be due to better fuel/air mixing which
results in more rapid (and thereby more intense) combustion.
In the 10-20 percent excess air range increasing air preheat increased
NO in all cases tested, sometimes by as much as 1 ppm per degree F. As
A
excess air was reduced toward zero the effect lessened. Shoffstall concluded
that the most obvious means of reducing NO emissions (without lowering air
A
preheat and losing efficiency) was to use either low excess air firing or
change to an axial fuel injector.
While the above groups were examining the control of NO through burner
A
design, many other organizations were conducting laboratory and full scale
testing on the more classical NO control techniques: low excess air (LEA)
91
-------
firing, staged combustion (SC), and flue gas recirculation (FGR). Since
the status of this work has recently been reviewed in considerable depth
it will not be discussed here, except to say that available data indicate
that LEA and SC (with its variations) are effective in reducing NOX from
gas, oil, and coal systems. FGR is extremely effective with natural gas
flames and to a lesser extent with oil. Unfortunately, none of the
techniques are without their problems: FGR is economically unattractive
and both LEA and SC could cause flame instability, slagging, and corrosion
difficulties in coal-fired systems.
PURPOSE
The purpose of this program is to utilize a versatile laboratory combustor
to examine the control of NO and other emissions through both burner and
J\
combustion modifications. The work is directed at present day fossil fuels;
specific goals are to:
(1) Determine what effect burner parameters have on the effectiveness
of known NO control techniques.
/\
(2) Compare the NO reduction possible through proper burner design to
that possible through combustion modifications.
(3) Establish the effect of fuel type on NOV control.
X
(4) Investigate novel design ideas for further development under
contract funding.
This paper is a status report on the work completed to date on
Goals 1, 2,and 3.
92
-------
Environmental Protection Agency policy is to express all measurements
in Agency documents in metric units. When implementing this practice will
result in undue costs or lack of clarity, conversion factors are provided
for the non-metric units used in a report. Generally, this report uses
British units of measure. For conversion to the metric system, use the
following conversions:
To convert from
°F
in.
gal.
GPH
BTU/hr
ft/sec
To
°C
cm
1
1/min
Cal/hr
m/sec
Multiply by
5/9 (°F-32)
2.54
3.79
0.0632
252
0.304
93
-------
EXPERIMENTAL APPROACH
The program is designed to consider present day fossil fuels including
the following:
Natural gas
*Propane
*Distill ate oil
*Distill ate oil doped to 0.3 percent N
No. 6 oil (residual)
*Bituminous coal
The distillate oils were -included in the program because together they provide
a mechanism for examining the conversion of fuel nitrogen. In each case the
oil is identical except that in the latter the nitrogen content has been
artificially increased to 0.3 percent (by weight). It has been postulated
by Martin and others that any difference in NO emissions between the two
A
cases can be directly related to conversion of the fuel nitrogen to NO .
/\
In each case the fuel is first chemically characterized and the effect
of the following burner (and process) variables on NO emissions determined:
/\
*Burner swirl
*Fuel injector design
Wall cooling
*Fuel injector position (in the burner throat or
at the quarl exit)
*Air preheat (ambient to 600°F)
Firing rate (up to 300,000 BTU/hr)
*Combustion air velocity (45, 100, and 200 ft/sec)
*Work either completed or already in progress.
94
-------
Because of time limitations all possible combinations are not
investigated. The 280,000 BTU/hr (input), 5 percent excess air, 530°F
preheat, 100 ft/sec air case has been selected as a base line. The
conditions are then varied parametrically around this.
Once the effect of the burner (and process) parameters on NO emissions
/\
is established for the normal combustor mode, the following control modes
are considered and the process repeated:
*Low excess air (down to 1 percent)
Staged combustion
*Flue gas recirculation
As the asterisks indicate, the experimental work is approximately 50
percent complete at the present time; the mathematical analysis and correlation
of the results is just beginning.
This report considers the most significant results to date; namely,
the effect of:
(1) Fuel type
(2) Burner swirl
(3) Air preheat
(4) Burner throat velocity
(5) Flue gas recirculation
(6) Throat velocity on the effectiveness of flue gas
recirculation
*Work either completed or already in progress.
95
-------
EXPERIMENTAL FACILITY
FURNACE DESIGN
The experimental furnace is illustrated in Figure 1. The combustion
chamber is vertical with the burner mounted on top. The walls are made of a
high temperature plastic refractory and normally run about 2500°F with 530°F
air preheat. A 1.5 inch water positive pressure is maintained within the chamber.
The chamber is cylindrical with a diameter of « 16 inches and is * 55 inches
long. The six observation ports provide good access in two planes and at
varying heights for flame observation, photography, and insertion of either
a water-cooled gaseous sampling or temperature probe. The burner is designed
to accept a variety of injectors for gas, oil, or pulverized coal as fuel and
is water cooled. Staged air addition is accomplished by means of water-cooled
air injectors inserted around the burner through the furnace top. The point
of staged air addition is varied by changing the insertion depth of the injectors.
To cool the combustion gases prior to the flue, a forced air heat exchanger
is attached at the bottom of the furnace perpendicular to its center line.
It consists of concentric steel cylinders - 16 and 20 inches in diameter and
= 75 inches long.
SUPPORT EQUIPMENT
The combustor supporting devices are shown in Figure 2 and are designed
to provide a wide choice of operating conditions with optimum individual process
parameter control. Under normal operating conditions combustion air is supplied
by an ambient temperature main blower through a manifold to the swirl, axial,
and primary (coal only) air lines. A high-temperature (600°F) blower forces
96
-------
BURNER
ACCESS
WINDOW
PORTS
EXHAUST
STACK
Figure 1. Furnace design.
-------
FUEL
INJECTOR
PRIMARY AIR
by
BURNER /
.y V
AXIAL AIR
SWIRL AIR
EXHAUST
SAMPLE PROBE
t
o
00
HEAT
EXCHANGER
occ
=
_ __ |
I
LAMINAR FLOW ELEMENTS
AIR FLOW CONTROL PANEL
-------
flue gas (from the exhaust stack) through a second part of the distribution
manifold to the proper air line. The third main section of the manifold is
connected to a bottled gas supply and provides an Ar/02 atomosphere to any
of the three air lines. The flow in each of the three air lines (swirl, axial,
and primary) is controlled by the manifold valves and is filtered for dust
and particulate removal before going through laminar flow elements which are
connected to inclined water manometers for measuring the quantities delivered.
Each line also has an electric air preheater with proportional controller to
provide controlled air temperatures from ambient to 700°F.
FURNACE BURNER
A specially designed water-cooled burner as illustrated in Figure 3 is
provided with separate axial air inlet and swirl chamber. The axial air enters
through a port angled at = 45 degrees into the center pipe and then passes
through straightening vanes. Swirl air enters a vaned swirl chamber via two
tangential ports 180 degrees opposed and passes through six 3/4-inch curved
swirl vanes as shown in Figure 4. The ID of the burner itself is 2.067 inches;
however, nine burner sleeves are provided so that axial momentum can be main-
tained at air velocities between 45 and 200 ft/sec for a variety of mass
flows, air preheats, etc. The burner is fitted with a 35-degree refractory
quarl and has an adjustable collar at the top (inlet) to allow for varying the
position of the fuel injector relative to the quarl exit.
99
-------
FUEL
INJECTOR
Figure 3. Burner.
100
-------
Figure 4. Swirl vanes.
101
-------
FUEL INJECTORS
A variety of injectors are provided for each type of fuel and are
generally characterized as either rapid or slow mixing as shown in Figures
5 and 6. Short bulbous flames are produced by the rapid mixing injectors.
The propane radial injector has six equally spaced holes (0.05996 inches in
diameter) perpendicular to the axis. The oil injector is a commercial 2.25
GPH nozzle with an 80 degree solid spray angle. The divergent coal injector
has three equally spaced (0.2656 inches in diameter) holes angled to distribute
the coal away from the axis.
The slow mixing injectors give long predominantly axial flames. The
propane axial injector has a single hole (0.1094 inches in diameter) on the
axis. The zero degree air atomizing oil nozzle has one hole (0.5937 inches
in diameter) angled on the top to prevent clogging. While the rapid mixing
injectors produce noticeably more stable flames, all injectors produced
stable flames from about 20 to 95 percent swirl under most conditions. The
work reported here was all conducted with the rapid mixing injectors since
these are more typical of current industrial practice.
STANDARD FUELS
Compositions of the fuels used to date are given in Table 1. The
propane is commercial grade and supplied from 100-gal. pressure-regulated
cylinders. Flow is controlled by a regulating needle valve and measured on
a calibrated gas rotameter. Distillate (#2) oil is delivered by a constant-
volume displacement metering pump which is electronically controlled to maintain
constant speed. The 0.3-percent N distillate oil is supplied by adding the
102
-------
PROPANE
RADIAL
(6-HOLE)
I
o o o
FUEL OIL
80 - DEGREE
SOLID
NOZZLE
r
u
COAL
DIVERGENT
(3 - HOLE)
I
i
•rt
r 1 \
Figure 5. Rapid-mixing injectors.
-------
PROPANE
AXIAL
(1 - HOLE)
I I
FUEL OIL
ZERO-DEGREE
AIR NOZZLE
,*,
AIR
\ I
\l
AIR
K
11
,y
\ • /
1 1
ii
COAL
AXIAL
(1 - HOLE)
k
!\
i !
\
\
Figure 6. Slow-mixing injectors.
-------
Table 1. FUEL ANALYSIS
Component
C
H
S
N
0
Ash
Wt Percent
Distillate
Oil
87.0
12.9
0.22
<.05
0.15
0.004
High Nitrogen
Distillate Oil3
86.7
12.7
0.21
0.31
0.15
0.004
Coal
69.6
5.3
3.0
1.17
9.6
10.4
Propane: Commercial Grade with > 90% CgHg and < 5% propylene, 2% ethane,
1% isobutane, < 0.5% N-butane.
aDoped to 0.31 percent N by the addition of quinoline
105
-------
appropriate amount of quinoline to the oil supply prior to the combustor.
The quinoline is supplied at a constant rate from a special pressurized
feed apparatus which is controlled by a micro-needle valve and measiured by a
calibratedrotameter. (At any given test condition the normal distillate
and 0.3 percent N distillate cases are run consecutively to minimize possible
errors due to slightly different combustor conditions, preheat temperatures,
etc.) Pulverized coal is delivered to the injector along with the primary
air by a vibrating hopper screw feeder. Feed rate is controlled by a
variable-speed gear motor drive.
ANALYTICAL PROCEDURES
The flue gas sampling system used in this work is shown in Figure 7. It
consists of: paramagnetic oxygen analysis; nondispersive infrared analysis for
carbon monoxide, carbon dioxide and nitric oxide; flame ionization analysis
for unburned hydrocarbons; and chemiluminescent analysis for NO and N02 (NOX).
Sample conditioners consist of a dryer (water condenser) and two particulate
filters in the main sample line. When burning coal a glass wool trap is
placed upstream of the dryer.
The Q£, CO, C02, and HC analyzers are further moisure protected by a
Drierite (CaSO^) dessicant cannister and molecular sieve traps (Grace SMR 4-635).
The NDIR NO analyzer is also Drierite protected. The chemiluminescent unit
requires no additional moisture removal. The Drierite cannisters are changed
daily and the sieves are replaced as dictated by moisture indicators. The
particulate filters are inspected daily and replaced as needed. The glass wool
trap is replaced before each coal test. All instruments are calibrated with
106
-------
PARTICULATE
FILTER
GLASS WOOL
TRAP
PARTICULATE
FILTER
DRIERITE
O
HANKISON
DRYER
STACK
m//A
MOLE-
CULAR
SIEVE
O
O
DRIERITE,
PARA-
MAGNETIC
ANALYZER
>?
YZ
GLASS WOOL
TRAP
NDIR
CO
ANALYZER
NDIR
CO?
ANALYZER
FLAME
IONIZING
HC
ANALYZER
O
NDIR
NO
ANALYZER
CHEMI LUMI-
NESCENT
NOX
ANALYZER
Figure 7. Analytical system.
-------
zero and span gas twice daily or before each test if required.
The sample probe is a 3/8-inch diameter quartz tube placed inside the
exhaust stack. All sample lines are either stainless steel or Teflon tubing.
SAFETY FACILITY
Flame failure safe operation is assured by a Honeywell R4150 flame
safeguard detection system. The flame signal for both gas ignition pilot
and main flame is produced by an ultraviolet flame detector. An automatic
power-off cutout is provided for both air flow loss and burner cooling water
flow loss. In addition a thermal limit switch is provided for any unusual
temperature rise at the burner.
108
-------
DISCUSSION OF RESULTS
DEFINITION OF TERMS
Before considering the results it is important to explicitly define
the terminology used. All emissions data are presented as ppm NO, dry reduced
to stoichiometric (zero percent excess air). To obtain the mass/heat input
the following conversions can be used.
Fuel
Propane
Distillate Oil
Coal
To convert to Ibs N02/106 BTU
multiply by
0.00108
0.00108
0.00147
To convert to gms
multiply by
0.00194
0.00194
0.00265
N02/106 cal
All data are reported in terms of a swirl index for common reference.
n
(This is not exactly the same as the swirl number defined by Beer and Chigier
because the swirl vanes used in this study were curved to maximize the efficiency
of swirl generation.) The swirl index S, is defined as:
S = G^
, R
where
Ge = P
and
= P
R = burner throat radius
R = radius at swirl vanes
p = oxidizer density
109
-------
V = volumetric flow rate through swirl vanes
V = total volumetric flow rate
A = minimum open area in swirler*
o
A = burner throat area
(No claim is made for the swirl index as a universal scaling parameter;
analysis of this type is just being initiated. It is used here only as a
reference basis.)
Flue gas recirculation is defined as follows:
Percent FGR = wt FGR x 100^
wt air + wt fuel
where
wt FGR = weight of flue gas recirculated
wt air = weight of the combustion air used
wt fuel = weight of fuel burned
FUEL TYPE AND SWIRL
Figure 8 shows the data taken for propane, distillate oil (< 0.05 percent N),
high-N (0.31 percent N) distillate, and pulverized coal as a function of swirl.
In these and all other tests reported here the propane entered the burner through
a six-hole radial injector at a velocity of 300 ft/sec; distillate oil through
an 80-degree solid-cone pressure-atomizing nozzle; and coal through a divergent
three-hole injector. All the data in Figure 8 were taken at a firing rate of
300,000 BTU/hr, 5 percent excess air, 530°F air preheat, and 100 ft/sec burner
throat velocity (combustion air). As the data indicate, the propane and
*Based on the axial depth of the swirl vanes times the perpendicular
distance between vanes at the point of entry into the axial flow.
110
-------
Figure 8. Effect of fuel type and swirl.
Ill
-------
distillate oil give nearly identical average emissions; however, the oil
is more strongly a function of swirl. The 0.3 percent N distillate has
nearly the same form as the previous pair but is about 220 ppm higher over
the range, corresponding to about 45 percent conversion of the fuel N
(assuming the thermal component can be subtracted directly as proposed by
Martin et al.8). The coal data are generally the highest of all, as would be
expected. The average emission is about 500 ppm of which a large part is
almost certainly due to the fuel N in the coal.
At this point no absolute evidence is available regarding the mechanism
behind the effect swirl has on each of the curves; however, the following
is proposed based on observation of the flames, experience, and work by
others.3'4
Gas and Distillate Oil: At very low swirl, combustion is delayed
farther downstream due to poor fuel/air mixing. This spreads the flame
zone over a larger area and reduces the average local flame temperature
through both the added bulk of entrained products and increased
radiative heat transfer. As the swirl is increased, the fuel/air
mixing increases, the combustion zone shrinks (causing increased
local temperatures), and the NO increases. As the swirl is further
A
increased, the internal reverse-flow zone on the burner axis becomes
substantial and begins forcing significant amounts of burned products
into the base of the flame. This dilutes the fuel/air mixture and
lowers local temperatures by acting as a type of flue gas
recirculation.
112
-------
0.3-Percent N Distillate: The emission data for the high nitrogen
distillate follow the normal distillate curve quite closely over
the entire swirl range. This indicates that the conversion of the
fuel nitrogen is essentially constant (at about 45 percent) and
therefore is not a function of swirl in this case.
Coal: The coal data are most easily explained starting at the high
swirl setting since even the divergent injector tends to give a fairly
axial coal flame.
As the swirl is decreased from its maximum, the NO emissions drop
A
slightly then begin a definite rise. Flame photographs reveal that
this increase begins at the point where the flame lifts off the
injector. At the peak shown in Figure 8, ignition is occurring about
6 inches from the point of injection. Thus, as the emissions are
increasing the point of ignition is moving steadily away from the injector.
The increase in NO here is almost certainly associated with better mixing
A
of coal jets and the combustion air prior to ignition. This, in turn,
increases the availability of oxygen and hence the conversion of fuel
nitrogen. The decrease after the peak is of little consequence since
at this point it is not an acceptable flame for industrial use.
AIR PREHEAT
Figures 9 through 12 show the effect of increasing air preheat from ambient
(100°F) through 300°F to 530°F at 100 fps throat velocity and 5 percent excess air
for propane, No. 2 oil, 0.3 percent N distillate oil, and coal. As the data in
Figures 9 and 10 indicate with gas and distillate oil, increasing air preheat
113
-------
0.4
0.8 1.2 1.6
SWIRL INDEX, S
2.0 2.4
Figure 9. Effect of air preheat using propane.
114
-------
3:
0 0.4
1.2
SWIRL INDEX, S
1.6 2.0 2.4
Figure 10. Effect of air preheat using No.-2 oil.
115
-------
1.2
SWIRL INDEX, S
1.6
2.0
2.4
Figure 11. Effect of air preheat using No. 2 oil (0.3% N).
116
-------
700]
GOO
500
400
300
200
100
530" F
3000 F
0.4
0.8 1.2
SWIRL INDEX, S
1.6
Figure 12. Effect of air preheat using coal.
117
2.0
-------
markedly affects NO . This is almost certainly due to an increase in local flame
/\
temperature. With both gas and No. 2 oil the preheated cases show a definite
peak. (It is certainly possible that the ambient (100°F) cases would have
also peaked had higher swirl been available.) Figure 13 shows the peak NO
for the distillate oil and propane runs plotted against theoretical (adiabatic)
flame temperature for the given case. This figure demonstrates that, for a
given air preheat, the peak emissions from No. 2 oil are only slightly higher
than from the corresponding propane case. Since all the data lie on the
same line (within the experimental error) this suggests that the slightly
higher emissions from oil may be due to the intrinsically higher flame
temperature (because of a higher C/H ratio) rather than to any type of droplet
burning process. There iss therefore, some question as to the importance of
droplet burning in the thermal fixation mechanism.
Figures 11 and 12 show that air preheat has a lesser effect on the emission
from the 0.3 percent N distillate oil and pulverized coal flames. This tends
o
to support the postulate of Martin et al. that fuel nitrogen conversion is
not as temperature sensitive as fixation.
BURNER THROAT VELOCITY
Figures 14 through 17 show the effect of increasing the burner throat
velocity from 100 fps to 200 fps (at 5 percent excess air and 530°F air
preheat) for propane, No. 2 oil, 0.3 percent N distillate oil, and coal. As
the data with gas indicate increased velocity decreases the NO by about
A
60 percent. A similar, but slightly more dramatic, effect is observed with
distillate oil. In both cases the increased velocity increases entrainment
118
-------
600
400
O NO. 2 OIL
D PROPANE
200
100
80
60
40
30*—
23.6
36000
37000 F
f
23.8
24.0 24.2 24.4 24.6
RECIPROCAL ADIABATIC FLAME TEMPERATURE (T-l), 105 OR-1
Figure 13. Peak NO emissions versus T~1.
119
24.8
25.0
-------
280
240
200
160
120
80
40
200 fps
100 fps
0.4
1.6
0.8 1.2
SWIRL INDEX, S
Figure 14. Effect of burner throat velocity using propane.
120
2.0
-------
0.4
0.8 1.2 1.6
SWIRL INDEX, S
2.0
2.4
Figure 15. Effect of burner throat velocity using No. 2 oil.
121
-------
700
600
500
400
a
o"
300
200
200 fps
100
0.4 0.8 1.2 1.6
SWIRL INDEX, S
2.0
2.4
Figure 16. Effect of burner throat velocity using No. 2 oil (0.3% N).
122
-------
1000
900
200 fps
700
a
o
600
500
100 fps
400
300
0.4
0.8 1.2
SWIRL INDEX, S
1.6
2.0
Figure 17. Effect of burner throat velocity using coal.
123
-------
of "cooler" combustion products and thereby decreases the local combustion
temperatures which in turn reduces NO. With the high nitrogen distillate
A
the percent reduction is less but the absolute magnitude is greater, indicating
a decrease in the fuel nitrogen conversion. With coal, however, the emissions
actually increase with velocity. Thus, increased axial velocity appears
to decrease NO emissions where thermal fixation dominates, but gives mixed
A
results in systems giving both thermal and fuel NO .
A
FLUE GAS RECIRCULATION
The effect of approximately 25 percent flue gas recirculation on the NO
emissions from each of the four fuels is presented in Figures 18 through 21.
As the data indicate, this caused about an 80 percent reduction in emissions
with propane and about a 65 percent reduction with distillate oil. In both
cases the reduction was almost certainly due to reduced local flame temperature.
In these cases, as with all previous work to date, the effect of swirl is
essentially negligible with high FGR levels. Hence, in designing a burner for
high FGR running it should be possible to set the swirl to minimize operating
problems (e.g., flame instabilities), fan power, etc. without substantially
affecting emissions.
As the data in Figures 20 and 21 show, FGR is not nearly as effective
in reducing NO emissions from the 0.3 percent N distillate oil and coal
A
flames. Again, however, there is a marked difference in the behavior of the
two systems; the oil shows very little effect while the coal did experience
a' 38 percent reduction at 24 percent FGR. These data suggest that flue gas
recirculation is going to be of limited value in systems with large quantities
of fuel nitrogen (e.g., residual oils and coals).
124
-------
280
240
200
160
120
80
40
0.4 0.8
1.2
SWIRL INDEX, S
Figure 18. Effect of FGR using propane
125
1.6 2.0
2.4
-------
360
_;i 200
0.4 0.8 1.2
SWIRL INDEX, S
1.6
2.0
Figure 19. Effect of FGR using No. 2 oil.
126
-------
700
600
500
400
300
200
100
0.4
0.8
1.2
SWIRL INDEX, S
1.6
2.0
Figure 20. Effect of FGR using No. 2 oil (0.3% N).
127
2.4
-------
0.4
0.8 1.2
SWIRL INDEX, S
1.6
2.0
Figure 21. Effect of FGR using coal.
128
-------
EFFECT OF THROAT VELOCITY ON FGR
In the previous sections we have shown that both burner throat
velocity and flue gas recirculation drastically reduce thermal NO . In
/\
the last test series, the two were combined to investigate possible additive
effects; Figures 22 through 25 show these results. As the data in Figure 22
indicate with propane, high air velocity and 25 percent FGR together reduced
the uncontrolled emissions from about 240 ppm to about 30 ppm. With
distillate oil, a similar level was achieved; however, the addition of FGR
produced no added reduction. With the 0.3 percent N oil, the addition of
FGR has no added effect over just increasing the air velocity from 100 to
200 ft/sec; with 25 percent FGR and high air velocity, the emissions are still
over 200 ppm. With coal, increasing the velocity increased the NO as
A
previously discussed; however, the addition of 24 percent FGR reduced the
emissions to essentially the uncontrolled level.
BURNER PRESSURE DROP
Since burner pressure drop is directly related to required fan power
and hence operating costs, burner AP must be considered in any analysis of
NO control through burner design. The burner used in this work normally
X
runs with a wind box pressure of 3 inches water gauge under axial conditions.
At high swirl the swirl cage pressure is also about 3 inches water. (The
axial would normally be less than the swirl; however, this burner has axial
straightening vanes to ensure proper flow of the axial stream). Increasing
the throat velocity from 100 fps to 200 fps caused a corresponding increase
129
-------
280
240
2001
0% FGR (100 fps)
1601
120)
0% FGR (200 fps)
401
25% FGR (200 fps)
0.4
0.8 1.2
SWIRL INDEX, S
1.6 2.0
Figure 22. Effect of throat velocity on FGR using propane.
130
-------
0% FGR (200 fps)
25% FGR (200 fps)
SWIRL INDEX, S
Figure 23. Effect of throat velocity on FGR using distillate oil.
131
-------
700
600
500
300
200
25% FGR (200 fps)
FGR (200 fps)
0% FGR (100 fps)
100
0.4
0.8
1.2
SWIRL INDEX, S
1.6
2.0
Figure 24. Effect of throat velocity on FGR using No. 2 oil (0.3% N).
132
2.4
-------
1000
900
0% FGR (200 (ps)
700'
600
500
400.
24% FGR (200 fps)
0% FGR (100 fps)
300
0.4
0.8 1.2
SWIRL INDEX, S
1.6 2.0
Figure 25. Effect of throat velocity on FGR using coal.
133
-------
in both the axial and swirl streams to 5 inches water gauge. FGR had
no noticeable effect on burner pressure drop but of course it did require
additional fan power and ducting.
OTHER EMISSIONS
CO and HC emissions were continuously measured during all tests and
were essentially zero with propane and distillate. With pulverized coal
combustion, HC emissions were always less than 10 ppm. Spot checks for
NO,, were also made. The data confirm that the primary NO emission is NO;
^ f\
in all cases examined, NO^ was less than 5 ppm.
134
-------
CONCLUSIONS
RESULTS
1. Under as nearly identical conditions as possible average NOX
emissions are as follows: propane * distillate oil < 0.3% N oil
< coal.
2. Increasing air preheat substantially increases the emissions
from propane and distillate oil and causes a lesser increase with the
high nitrogen distillate and pulverized coal.
3. Under peak NOX conditions propane and distillate oil are
essentially identical with respect to NOX emissions. There is little
evidence that the difference in the phase of the fuel (gas vs. liquid)
has any significant effect on emissions.
4. Increasing burner throat velocities substantially reduces
emissions with propane and distillate oil, has little effect with the
0.3 percent N oil, and increases emissions with coal.
5. Flue gas recirculation is very effective with propane and distil-
late oil, has almost no effect with the 0.3 percent N oil, and is only
moderately effective with pulverized coal. When effective FGR is
essentially independent of burner swirl.
6. Doped distillate oil (0.3 percent N) and coal behave very
differently even though at least half the NOV emissions from both are almost
A
certainly the result of fuel nitrogen conversion.
IMPLICATIONS
1. Conversion of a unit from a low nitrogen oil to gas or vice versa
135
-------
should not result in major increases (or decreases) in NOV emissions.
A
2. Increasing combustion air preheat can be expected to increase
Hi
FGR).
thermal NO unless some type of counteracting measures are taken (e.g.,
A
3. Increasing burner throat velocities (and hence entrainment of cooled
combustion gases) can be as effective a control technique for thermal NO
A
as application of substantial FGR.
4. Flue gas recirculation cannot be expected to provide large emission
reductions with high nitrogen oils or coals.
NOTE
The implications cited above (1 through 4) are based
on the results of this research investigations.
136
-------
FUTURE EFFORTS
The experimental work on this program during the next few months
will be directed at finishing the test matrix and at specialized analysis
experiments. The present coal data will be extended to include a wider
variety of injector types at higher excess airs. Natural gas and No.
6 oil will be examined. Two stage combustion testing will also be con-
ducted with each of the fuels.
In one of the limited, specialized test series coal, No. 6 oil,
and the 0.3 percent N distillate will be burned in an Ar/02 atmosphere.
Finally, some of the anomalies which have surfaced so far will be examined
in greater detail. For example, increasing the burner throat velocity
has been proposed as a means of achieving flue gas recirculation inside
the furnace (through the entrainment cooled product gases). Further, with
gas and distillate oil increased throat velocity and F6R do indeed cause
similar, significant NOX reductions. However, with the 0.3 percent N oil
doubling the velocity halves the NOV while 25 percent F6R does nothing.
A
With coal, doubling the velocity more than doubles the NOX while 24 percent
F6R reduces it by 40 percent.
137
-------
BIBLIOGRAPHY
1. Wasser, J. H. and Berkau, E. E., "Combustion Intensity Relation-
ship to Air Pollution Emissions from a Model Combustion System", Air
Pollution and Its Control, 126, Volume 68, 1972 A.I.Ch.E. Symposium
Series.
2. Heap, M. P., Lowes, T. M., and Walmsley, R., "The Effect of Burner
Parameters on Nitric Oxide Formation in Natural Gas and Pulverized Fuel
Flames," presented at the ARC/EPA "American Flame Days," Chicago,
September 1972.
3. Heap, M. P. and Lowes, T. M., "Nitric Oxide Production in Large
Scale Natural Gas Flames," PR 10, EPA Contract No. 68-02-0202, Inter-
national Flame Research Foundation, December 1972.
4. Heap, M. P. and Lowes, T. M., "Nitric Oxide Formation in Pulverized
Coal Flames," PR # 11, EPA Contract No. 68-02-0202, International Flame
Research Foundation, December 1972.
5. Hemsath, K. H., Schultz, T. J., and Chojnacki, D. A., "Investigation
of NOX Emissions from Industrial Burners," presented at the ARC/EPA
"American Flame Days," Chicago, September 1972.
6. Shoffstall, D. R. and Larson, D. H., "Aerodynamic Influences on Com-
bustion Process Pollution Emissions," presented at Central States Section
Combustion Institute, Urbana, Illinois, March 1973.
138
-------
7. Jain, L. K., Calvin, E. L. and Looper, R. L., "State of the Art for
Controlling NOV Emissions in Utility Boilers," Final Report, EPA
A
Contract No. 68-02-0241, Catalytic, Inc., September 1972.
8. Martin, G. B. and Berkau, E. E., "An Investigation of the Conversion
of Various Fuel Nitrogen Compounds to Nitrogen Oxides in Oil Combustion,"
Air Pollution and Its Control. 126, Volume 68, 1972, A.I.Ch.E. Symposium
Series.
9. Beer, J. M. and Chigier, N. A. Combustion Aerodynamics, Applied
Science Publishers, Ltd., London, 1972.
139
-------
BURNER DESIGN PRINCIPLES FOR MIHTM01I NO EMISSIONS
X
M.P. Heap, T.JI. Lowes, R. Valmsley
and H. Bartelds
International Flame Research Foundation
IJrauiden, Holland
Paper presented at the E.P.A. Coal Combustion Seminar,
19-^0 Juno, 197:5 lies ear ch Triangle Park, North Carolina
141
-------
1. INTRODUCTION
Various control techniques are available to reduce nitrogen oxide
(NO ) emis
include:-
(NO ) emissions from large steam raising plant. These techniques
- operating modifications ie. reduced load, excess air or preheat;
- combustion modifications i.e. flue gas recirculation or staged
combustion;
- burner modifications.
All these techniques will necessitate variations in accepted plant
operating conditions and may also increase the unit cost of the
power produced. This paper discusses the principles upon which
burners with minimum emission characteristics can be designed.
The paper is mainly concerned with pulverised coal (p.f.) burners
for utility boilers, however reference will be made to other fuels.
In the long term burner modifications may well provide the most
efficient method of controlling NO emissions from all forms of
fossil fuel fired furnaces and boilers.
Minimum NO emission characteristics are not the only desirable
design feature of burners. Consideration must also be given to:
- ignition stability;
- fuel burnout;
- noise production;
- the generation of other pollutants;
- burner production and running costs;
- peak temperatures;
- flame shape.
The design principles for minimum NO emissions will have an effect
Jt
upon many of the characteristics listed above and wherever.
possible these effects will be discussed in the present paper.
143
-------
2. NO FORMATION IN TURBULENT DIFFUSION FLAMES
•y Ljjin—m—|—I «:r»-n
The NO emitted from fossil fuel fired combustors is the result
of two processes:-
- the oxidation of molecular nitrogen producing thermal N0x;
- the conversion of nitrogen compounds contained in the fuel,
fuel NO ;
In order to understand the influence of various burner parameters
on the formation of NO in flames it is necessary to briefly
summarise the controlling influences on the formation of thermal
and fuel NO
x.
2.1. The Formation of Thermal NO
The formation of thermal NO in combustion processes has been
.X
studied extensively in recent years. Although the precise details
of the interaction between hydrocarbon combustion and thermal NO
formation remain unknown, the controlling influences of time,
temperature and combustion stoichiometry are generally recognised
Pi, 2^. Virtually all attempts to control the formation thermal
NO involve the reduction of peak temperatures.
Jl.
In the majority of combustors thermal NO formation can be
J^.
considered as a flame phenomenon since residence times at bulk
gas temperatures are normally too short to allow the formation
of significant quantities of NO within the bulk gases. The amount
jC
of N0x produced within the flame region depends upon:-
- the initial temperature of the freshly formed combustion products
within the flame;
- the rate of temperature decay of these freshly formed products.
144
-------
Both the initial temperature and the rate of temperature decay
can "be controlled by burner parameters since these parameters
dictate the mixing pattern of the fuel, air and recirculating
gases.
The temperature of the freshly formed products of combustion within
the flame will obviously depend upon the composition and enthalpy
of the reactants. Combustion in diffusion flames is complicated
because the reactants must be mixed on a molecular scale before
combustion can take place. The location and mixture strength of
the reaction zone in diffusion flames are matters for conjectures;
reaction will proceed wherever the mixture strength lies within
the limits of flammability and there is a source of ignition.
Thus maximum temperatures are attained when the fuel reacts in
proportions close to stoichiometric before either the fuel or air
have been diluted with recirculating products. The high temperature
combustion products are cooled subsequent to formation by mixing
with the bulk gases and their rate of temperature decay depends
upon the rate of mixing with and temperature of the bulk gases.
2.2. The Formation of Fuel NO
x
Both residual fuel oils and coal contain nitrogen compounds.
Although it is almost universally accepted that the oxidation of
these nitrogen compounds contributes significantly to the total
NO emission from combustion processes,very little is known concerning
the oxidation process. The oxidation of the fuel nitrogen compounds
in flames can be considered in two stages:-
- the evolution of nitrogen compounds XN from liquid droplets or
coal particles;
- subsequent reactions of these nitrogen compounds.
Mo positive identification of the intermediate nitrogen compounds
XN has been made under flame conditions. However it is probable
that both the type of compound evolved and the rate of the
evolution will depend upon the heating rate of the fuel. At the
present time theories concerning fuel NO formation must be based
A
upon information gained from experiments with doped fuels or from
the combustion of simple nitrogen compounds.
145
-------
Sternling and Wendt Pjl have recently summarised the available
information concerning the ultimate fate of fuel nitrogen
compounds in combustion processes :-
- with doped fuel oils it has been shown that the fraction of
fuel nitrogen converted to nitric oxide increases with in-
creasing excess air and decreases with fuel nitrogen concen-
tration £4»5[]s
- it is possible to reduce HO to Np under fuel rich conditions;
- the conversion of fuel nitrogen appears to be a strong function
of burner/combustion chamber combination;
- flue gas recirculation does not appear to be effective in
reducing emissions of fuel NO £5j«
In order to explain the above experimental information it is
necessary to postulate some kind of kinetic mechanism. Fenimore £61
suggests that all the fuel nitrogen goes through an intermediate
compound I which reacts either to produce HO or N:-
I + R - ^ if o +
I + NO - '-> N2 +
Fenimore considers that R is probably OH and that I could either
be NH2 or N. In an attempt to model the reduction of NO under
fuel rich conditions Sternling and Vendt [3] account for the
formation of the nitrogen-nitrogen bond by the reaction:-
H + NO' =»N2 + 0 £
which is faster under fuel rich conditions than:-
N + On 5>NO + 0 A
146
-------
2.J NO Formation in Pulverised Coal Flames
The combustion of p.f. is a complex process involving:-
- particle heating by convection and radiation;
- rapid evolution of the volatile fraction;
- combustion of the volatile fraction;
- char burnout.
The fuel is normally injected into the furnace as a coal/air
suspension. The proportion of air in the primary jet varies but is
typically between 15 and 20$ of the total air flow. The additional
air required for combustion is supplied in an .annular preheated
secondary stream surrounding the primary jet. Upon injection into
the furnace the particles receive heat either by radiation
from the surroundings (which may include furnace walls, bulk gases
and the ignition front) or by convection from the preheated
secondary stream and entrained recirculating gases. When the coal
particles reach a sufficiently high temperature they begin to
decompose producing tars and gases usually referred to as
volatiles. The composition of the volatiles and the particle
weight loss depends upon the time/temperature history of the
particle. Thus the rate of heating of the particle and its final
temperature influence the quantity and composition of the evolved
volatiles.
"The volatile fractions begin to combust when the temperature
is sufficiently high and the fuel/air mixture lies within the
flammable limits. The time required for the ignition of the
volatile fractions depends upon the rate of mixing of the primary/
secondary and.recirculation gases and the bulk gas temperature.
The distance from the point of injection and the visible ignition
front will be referred to as the ignition distance. The solid
particles remaining after devolatilisation, soot, cenospheres and
cellular particles are collectively referred to as char. There is
a fundamental difference between the combustion of the char and
volatile.fractions. The reactions involved in the combustion of
147
-------
the volatiles have finite rates but these are usually high
compared with the surface reactions involved in the combustion of
char. Also the combustion of the char particles involves several
steps in sequence:-
- transport of oxygen or other reactant gas to the particle
surface;
- reaction with the surface;
- transport of the reactants away from the surface.
The overall reaction rate is,of course, dependent upon the slowest
of these steps.
NO formation in p.f. flames may be considered in four stages:-
- fuel NO formation during volatile combustion;
A
- thermal NO formation during volatile combustion;
2.
- fuel NO formation during char burnout;
ji.
- thermal NO formation during char burnout.
In the absence of definite information it is necessary to make
several simplifying assumptions concerning NO formation in p.f.
flames in order to explain the effect of burner parameters on
total emissions.-These assumptions are:-
the .most significant fraction of the total emission is fuel
NO . Thermal NO will be produced during the combustion of the
jt JL
volatile fractions but peak temperatures will be low since the
volatiles will be diluted with entrained combustion products
prior to ignition. During char burnout high surface temperatures
could produce thermal NO but this possibility is ignored.
JC.
the majority of the fuel NO is produced during the combustion
jC
of the volatile fraction. Sternling and Wendt [5] consider that
the total nitrogen content of the initial coal will be divided
between the char and volatile fractions and therefore char
combustion could be a source of fuel NO . However, measurements
Jx '
148
-------
tend to suggest that this will be a negligible fraction of
the total emission.
Accepting these assumptions, then the conversion of volatile
nitrogen compounds to NO or N_ is explained by the following
sequence of events :-
a) the coal particles are heated and the volatile fractions
evolved containing nitrogen compounds XN, which may react
directly or undergo pyrolysis prior to reaction in combustion
zones;
b) in the combustion zone two overall competing reactions1 can
take place:
XN + Y - ^ NO +
XN 4- Z - ^ -N2 +
The identity of reactants Y and Z needs to be specified. The
ultimate conversion of XN to either N ' or NO is dependent upon
the quantity of oxygen associated with the combusting volatiles.
In oxygen rich regions reaction 5 will predominate and the
formation of fuel NO is promoted. In oxygen deficient regions
.X.
reaction 6 is dominant and the conversion of fuel nitrogen to
NO is limited. Burner parameters can be used to vary the emission
of NO from p.f. flames because they control the mixing history
JC
of the fuel particles, the combustion air and the recirculating
gases which will dictate the oxygen available during the
combustion of the volatile fractions.
5. BURNER DESIGN PARAMETERS FOR THE CONTROL OF NOY EMISSIONS
PROM FOSSIL FUEL FLAIiES
Investigations at IJmuiden have shown that NO emissions from
wd
natural gas, fuel oil- and p.f. flames can be varied over a wide
range by a suitable choice of burner parameters. The parameters
which have been investigated include:-
HP
-------
- the method of fuel injection;
- the position of the fuel injector;
- the degree of swirl in the combustion air;
- the velocity of the combustion air;
- the angle of the burner exit;
- the presence of swirl impellers on the oil gun;
- the division of the total air between primary/secondary and
tertiary streams;
- the velocity of the tertiary stream.
The investigations were carried out in an almost uncooled
refractory tunnel furnace (2 m x 2m x 6.25 m) and relate to single
burner emissions. The burner used during the investigations(see
fig. 1) had the facility to vary the swirl intensity of the
secondary stream continuously from zero to a maximum value deter-
mined by the burner geometry.
In the results presented in this paper the swirl intensity of the
secondary stream will be expressed as a relative swirl index, Rg,
defined by:
_ Actual opening of the swirl blocks
s ~ Maximum opening of the swirl blocks
Figure 2 shows the relationship between E and S, the swirl number
S
for several combinations of burner throat diameter and outside
diameter of the primary pipe. Swirl number is a dimensionless cri-
terion that has been used to characterise swirling flows and is
defined by:
o flux of angular momentum
D —
flux of axial momentum x burner radius
The two parameters with the most influence on NO emissions are
Jt
the method of fuel injection and the swirl intensity of the
combustion air. Figures 3, 4 and 5 illustrate how flue gas nitric
oxide concentrations depend upon 'these parameters for natural gas,
fuel oil and pulverised coal flames. The combination of these
parameters also controls such important characteristics as flame
150
-------
stability, smoke production and heat release rate. The influence
of the method of fuel injection and the swirl 'intensity of flame
characteristics can be more easily understood by reference to the
flame classification scheme presented in fig. 6. This scheme
refers to gaseous, liquid and solid flames.
a) Lifted Flames (fiff. 6a)
The ignition front is stable some distance downstream from the
primary injector. The stabilisation of lifted natural gas flames
is helped by high external recirculation temperatures. In
increase in' throughput can blow-off the lifted flame completely.
b) Injector Stabilised Flames (fig. 6b)
This type of flame is normally produced by single hole injectors.
Stability is achieved either by a "bluff body effect" or by an
auxiliary pilot. Although the fuel jet is entirely surrounded by
flame, the flame does not completely fill the burner exit.
c) Primary Jet Penetrating; a Region of Reverse Flow (fig. 6c)
With single hole injectors of high primary velocity and "medium"
swirl or low primary velocities and low swirl it is possible that
the primary jet will penetrate the swirl induced internal reverse
flow zone. The internal reverse flow zone then takes the form of
an annulus surrounding the central fuel jet. The flame may be
divided into two sections, a short bulbous zone close to the burner
and a long tail. The two sections are connected by a neck which under
particular circumstances may break and only the bulbous base remains.
d) Divided Fuel Jets (fig. 6d)
This type of flame with a closed internal recirculation zone on
the flame axis is characteristic of the type of flame used in
utility boilers. It is short with a high heat release rate per
unit volume of "flame". It can be produced in p.f. flames with
intermediate swirl values by using radial injectors, coal spreaders,
annular injectors or low velocity single hole injectors. This type
of flame is produced with fuel oil by the use of pressure jets or
steam atomised injectors.
151
-------
Although the flames of different fuels can be classified according
to the simple scheme shown in fig. 6, their emission characteris-
tics are dependent upon the fuel type. Thus although p.f. flames
with divided fuel jets always give maximum emissions, the emission
levels of all four types of natural gas flame are similar under
particular conditions.
In section 2 the conditions necessary to reduce the formation of
"both thermal and fuel HO were discussed. Ample experimental
jt
evidence has been obtained to show that burner parameters affect
NO emissions from natural gas, fuel oil and p.f. flames. Emissions
A
from p.f. flames can be reduced to the same level as those of
natural gas and fuel oil flames with comparable input conditions
(i.e. preheat and thermal load). However, to achieve this
emission level radically different flame characteristics must be
tolerated. Consequently design parameters for minimum NO emissions
^C
can be judged from two viewpoints:-
- what are the burner parameters necessary to give minimum
emissions regardless of flame characteristics?
- which parameters can be used to reduce emissions without
seriously affecting flame characteristics?
Flame characteristics are not the only consideration, the design
parameters necessary to reduce HO may also increase the burner
«*h
pressure drop or increase burner maintenance costs.
If the assumptions suggested in section 2 are correct then the
minimum 110 emissions are achieved by restricting the available
A.
oxygen during the combustion of the volatile fractions. Ideally
the volatile fractions containing the nitrogen compounds XH should
burn in a diffusion flame because this would give the minimum
conversion to HO £3 j. However this is not possible since the coal
is supplied with air and the volatiles are mixed with air as they
are evolved from the coal particles. Also mixing of the primary
and secondary streams prior to ignition will increase the oxygen-
coal ratio from the input condition. '
152
-------
The emission curve presented in fig; 7 illustrates the reduced
emission resulting from restricting the amount of primary/secondary
mixing. These results were obtained accidentally. Initially the
fuel injector was uncooled and at high swirl levels with the
ignition front stable on the injector a coating of red hot char
was deposited on the thick interface. This char acted as an
ignition source so that when the swirl was reduced to zero the
ignition zone remained stable on the injector. Pig. 7 shows
measured flue gas HO concentrations as the swirl level is increased
to a maximum and then decreased. The coal jet was always enclosed
by a visible ignition zone as the swirl was reduced to zero.
The only significant difference between the emission levels is at
low swirl levels. This difference is attributed to the variation
in the coal-oxygen ratio during the combustion of the volatile
fractions for the lifted flame and the injector stabilised flame.
Prior to injection the coal oxygen ratio of the two flames is
identical. However the lifted flame has an ignition distance of
approximately 0.75 m and mixing between the primary and secondary
stream increases the amount of oxygen associated with the volatile
fractions. With the injector stabilised flame the oxygen available
for mixing with the evolved volatile fractions is limited to that
of the input primary stream since the primary and secondary
streams are separated by a region of combusting volatiles.
Provided sufficient combustible gases are available and the ignition
front is complete, oxygen from the secondary stream will not be
able to penetrate into the fuel jet.
Thus the conditions necessary for minimum 1TO emissions from p.f.
flames are:-
minimum primary air supply;
minimum primary/secondary mixing prior to completion of the
combustion of the volatile fraction;
ignition stability at the injector;
dilution of the secondary air with recirculating combustion
products prior to contact with the fuel.
153
-------
These conditions are satisfied by using a single hole high velocity
injector positioned at the exit of the burner divergent. The coal
is supplied with the minimum amount of primary air and swirl is
used to stabilise the ignition at the injector. The consequence
of variation from these conditions can be seen from the following
examples:-
- Fig. 8 shows the effect of primary velocity; the higher the
primary velocity, the lower the emission level. The variations
in emission level are caused by variation in mixing pattern
produced by the interaction of the internal reverse flow regions
and the primary jet. Lifted flames -were observed with all injec-
tors for swirl levels less than R =0.4. With the low primary
s
velocity of injectors B and C the swirl induced reverse flow
zone is sufficient to split the fuel jet as it emerges from the
injector producing the flame type shown in fig. 6c. Thus primary/
secondary mixing is enhanced and emissions increase. As the
swirl level is increased, combustion within the burner divergent
intensifies causing an increase in axial momentum which enables
the primary jet to penetrate the internal reverse flow region.
The primary/secondary mixing is reduced and consequently the
emission decreases. Injector B, which has the lowest primary
momentum, produces a flame with a divided fuel jet which then
changes to a flame where the internal reverse flow zone is
penetrated by the fuel jet and then reverts to a divided fuel
jet flame as the swirl is increased?
- The effect of increasing the primary air supply whilst maintaining
the primary velocity can be seen in fig. 9. Minimum emissions
are obtained with the lowest primary air percentage;
- Figs.lOa and 10b show that emissions are less when the injector
is placed at the exit of the divergent and emissions are less
with a divergent angle of 25° rather than a parallel exit.
154
-------
Earlier it was stated that flames with divided fuel jets are
normally used in utility boilers. It is possible to produce this
type of flame without swirling the combustion air by injecting the
p.f. normal to the burner axis. Provided some swirl is used
divided fuel jet flames can also be produced with an annular or a
low velocity single hole injector or some other device to spread
the fuel. Due to the rapid mixing divided fuel jet flames have
high heat release rates and wide ignition stability limits.
However, this flame type has the maximum ITO emission characteris-
tics with p.f. It is believed that this is because the coal
particles are intimately mixed with all the available air, thus
providing ideal conditions for fuel 110 formation.
JK.
The results presented in figs. 11. and 12 show that with radial or
annular injection varying swirl or primary air percentage has
very little influence on the emission level. Using a coal "spreading
injector" three possibilities exist to vary emission levels:-
- vary the swirl level of the secondary air. In fig. 13 it can
be seen that ITO emissions increase as the swirl intensity of
the secondary air increases. This type of injector would not
normally be used at swirl levels less than R = 4 since ignition
s
is not stable within the burner exit. At high swirl intensities
the emission is reduced after the flame form passes through an
instability condition. The difference between the high and low
emission conditions is visually apparent: at high swirls the
flame forms a closed ball;
- increase the primary air flow. Emissions are reduced because the
increased axial momentum reduces the effectiveness of the spreading.
device and thus less fuel/air mixing takes place (see fig. 14a).
- change the position of the point of fuel injection. In fig. 14b
it can be seen that emissions are reduced when the point of
injection is changed. Emissions are less when the injector is
moved towards the exit plane of the burner.
155
-------
Although emissions from divided fuel jet p.f. flames can be
reduced by burner parameters, the reductions are bought at the
expense of increased burner pressure drop (to produce swirl) or
by a lengthening of the flame. Recent work at IJmuiden has shown
that triple concentric burner systems have the potential to reduce
emissions from flames with divided fuel jets with gaseous liquid
and solid fuels. However, the tertiary velocity is critical and
incomplete combustion may result from inadequate burner design.
4. CONCLUSIONS
Jk.
NO emissions from fossil fuel fired furnaces and combustors can
be reduced by the optimisation of burner design parameters.
it
NO emissions from p.f. flames can be reduced to the same level as
those from comparable gas flames. However, the reduced emission is
achieved by a radical change in flame form. Adequate ignition
stability and burnout are possible but the flame becomes longer
and thinner.
Limited reductions are possible with flame forms which are in use
at present.
Burners with tertiary air supplies can be designed to reduce NO
emissions without changing the flame form. However, care must be
exercised in design, otherwise CO and solid emissions can be
increased.
ACKNOWLEDGEMENTS
The work reported in this paper was carried out under contract
number 68-02-0202 for the Environmental Protection Agency.
156
-------
REFERENCES
~ BREEN, B.P.
Emissions from Continuous Combustions System. Ed. ny
¥. Cornelius and V.G. Agnew, Plenum Publishing Corp.,
New York p. 325.
2~ WESTENBERG, A.A.
Comb. Sci. Techn. ± 59 (1971).
3~ STERKLING, C.Y. and WENDT, J.O.L.
Kinetic Mechanisms governing the Pate of Chemically Bound
Sulfur and Nitrogen in Combustion. Final Report EHS-0-71-45
Task 14 Shell- Development Company, Emeryville, California
(1972).
4" MARTIN, G.B. and BERKAU, E.E.
An Investigation of the Conversion of Various Fuel Nitrogen
Compounds to Nitrogen Oxides in Oil Combustion. Paper
presented at A.I.Ch.E. National Meeting Atlantic City
1971.
5~ TURNER, P.V. et-al.
Influence of Combustion Modifications and Fuel Nitrogen
Content on 1TO Emissions from Fuel Oil Combustion.
Paper presented at Annual Meeting of A.I.Ch.E.,San Francisco
1971.
6~ FENIMORE, C.P.
Combustion and Flame J2 289 (1972).
157
-------
APPEHDIX I.
FUEL COMPOSITIONS
Natural Gas
CH4 81,3 #
C2H6 2,9 °/o
C3H8 0,4 $
C4H1Q 0,4 *
CnHm 0,1 %
C02 0,8 fo
N2 14,4 #
Fuel Oil
C 86,05fo
H2 11,54?*
N2 0,24fb
S C
Ash C
Coal
Volatile Content • 32
ish £
C 78,487o
Hp 4,77?£
0,75?^
158
-------
APPENDIX II.
COAL INJECTOR CHARACTERISTICS
_ t
Injector Outside Diameter Mean Primary Velocity m sec.
cm. 10?o 20$ 30$
Primary air 'Primary air Primary air
A 11,5 19 38 57
3 6,0 19 38 57
C 6,0 26 52
H 6,6 52
159
-------
JEUtt
fe?
block adjustment
mechanism
fuel gas nozzle
(throat DOS.)
cooling
water
fixed blocks
movable blocks B-j
Fig -1: Moving block swiri burner
-------
2.0-
1,5
0.5
0
Curve
2
3
A
Throat
diam.
Injector
o.d
17,6cm
13,1 cm
17,6cm
13.1 cm
0,2
0.4
Q8
1.0 R«
Fig. 2 ;The relationship between relative swirl
index Rg and swirl number S
161
-------
NO
ppm
100-
A
80-
60
40"
Injector Type_
A Multihole Divergent
x Multihole Radial
o Single Hole Axial
20-
Lifted flame
0
0,2
OA
06
0,8
1,0 R,
Fig -3 -.The effect of swirl and injector type on NO
emissions from gas flames 5% excess air
throat 176cm diam injector in throat
injector od 6,0 cm
162
-------
Atomiz er Type
x Single hole air
• Pressure Jet 30° Spray
a Steam 100° Spray
x- -x
100
0.4
0,6 0,8
1,0 R<
Fig. I*: The effect of oil injector type and swirl on
NO emissions
(divergent 35° excess air 5% injector 6,0cm o.d
in throat secondary air 30 °C )
163
-------
400?
200-
Injector Type
x Single hole A
• „ „ C
o H
A Radial hole
0 L
0
0,2
OA
0,6
0.8
10 R,
Fig - 5: Effect of swirl and injector type on the emission
characteristics of PF flames
(30Q°C preheat - 5% excess air )
164
-------
N Ignition
oC?distance
a) Lifted flames
Ignition zone
b) Injector stabilised flame
Reverse flow
c) Primary jet Penetrating Internal Reverse
Flow
Reverse flow
d) Divided fuel jet
Fig. 6 : Simple Flame Classification Scheme
165
-------
NO
ppm
8001
o Swirl Adjustment Increasing
\ a ., ,, Decreasing
1
700 X I
I
1 Lifted Flame
I
I
I
I
1
600
500
I
1
Injector Stabilised
Flame
0,4 0,6 0,8 1,0 Rs
Fig-7:Effect of stabilisation of ignition front on the
injector ( 5% excess air 300 °C preheat
injector o.d. 6,0 cm throat 17.6cm diam)
-------
0
Fig• 8 :The effect of primary velocity (throat 17.6cm diam.-
5% excess air-300 °C preheat - primary air 10%
of stoichiometric)
167
-------
NO
ppm
Proportion of Primary air
x 30% of stoichiometric (B)
• 20% of stoichiometric (C)
o 10% of stoichiometric (H)
600-
400-
200-
X
0
02
0.4
0.6
0.8
1.0 Re
Fig• 9: Effect of primary air supply (throat 176cm di
inj'ector at throat - 5% excess air-300°C preheat]
168
-------
Injection position
o throat
• exit
300
200-
0 0,2 0,4 0,6 0,8 1,0 Rs
Fig.lOa; Effect of injection position (injector H - throat
17.6cm diam - 5% excess air-300°C preheat -
10% primary air
Burner exit geometry
• parellel exit
o 25° divergent
200
0 0,2 0,4 0,6 0,8 1,0 Rs
Fig. 10b:Effect of burner exit geometry (injector H at exit-
throat 176cm diam-5% excess air-300°Cpreheat
10% primary air
169
-------
700-
600-
Proportion of primary air
x 10°/o of stoichiometric
o 20% of stoichiometric
0.8
1.0 R«
o 02 OA 0,6
Rg-11 Emission characteristics of divided fuel jet flames
.produced with a radial injector (injector 6,0 cm o.d-
throat 176cm diam.- 5% excess air-300°C preheat I
NO J
ppm
700
600-
500-
/ Proportion of primary air
o 20°/o of stoichiometric
B 30% of stoichiometric
0,2
OA
0.6
0,8
1P Rs
Fig-12:Emission characteristics of divided fuel jet
produced with an annular injector (injector 11.5cm
throat 17.6cm diam-5%excess air-30Q°C preheat)
170
-------
NO
ppm
900
800
700
600
x
0 Q2 0,4 0,6 0,8 Rs 1,0
Fig. 13 : The Affect of swirl on NO emissions with
device
J_30 (LfCL preheat, - 15% excess air
25° diver g ent 17,6 cm id. throat
injector 6,0 cm o.d. in throat )
-------
NO REDUCTION TECHNIQUES
X.
IN
PULVERIZED COAL COMBUSTION
by
Christopher England and John Houseman
Jet Propulsion Laboratory, Pasadena, California
(Presentation to the Pulverized Coal Combustion Seminar,
June 19-20, 1973, National Environmental Research Center,
Research Triangle Park, North Carolina)
173
-------
NO REDUCTION TECHNIQUES IN
x
PULVERIZED COAL COMBUSTION*
by
Christopher England and John Houseman
Jet Propulsion Laboratory, Pasadena, California
INTRODUCTION
The technology of the combustion of coal in utility boilers has been
developed to optimize combustion efficiency, plant efficiency, and boiler life-
time. Increasing restrictions on emissions, however, require new procedures
which balance overall efficiencies with emission control. Because flue gas
treatment to remove low-level pollutants is generally uneconomical, other
means are being sought to prevent the formation of undesirable compounds
during the combustion process. These efforts center on fuel-processing to re-
move sulfur, ash and nitrogen compounds in the coal, and on combustion pro-
cess modifications to prevent the formation of oxides of nitrogen.
Several studies have been made of the effect of low excess air operation
(_!_, 2, 3) and staged combustion (4) on NO emissions from pulverized coal
combustion, and all showed reduced emission levels with reduced excess air.
Data are available, however, only over limited ranges of coal-air stoichio-
metry, primarily because of experimental difficulties in operating at rich or
very lean conditions. Other combustion modifications designed to lower peak
flame temperatures, such as reduced air preheat and product gas recirculation,
have not been studied widely, and their effects are not fully understood.
The present paper describes a study in which coal was burned in a pre-
fired tubular furnace in which air-fuel ratio and air preheat (both primary and
secondary) could be varied routinely and independently to determine their effects
on NO production. The purpose of the study was to obtain parametric data over
*This paper presents the results of one phase of research carried out at the Jet
Propulsion Laboratory, California Institute of Technology, under Contract No.
NASA 7-100, sponsored by the National Aeronautics and Space Administration.
175
-------
the widest possible range of operating conditions to evaluate the effects of the
various combustion process modifications on NO emissions.
Js.
EXPERIMENTAL APPARATUS
Furnace
Combustion was carried out in a horizontally-fired Mullite furnace tube
with an inside diameter of 7. 6 cm (3. 0 in.) and an overall length of 154 cm
(60 in.). The furnace was insulated with approximately 0.6 cm (0.25 in.) of
zirconia fiber wrapping, and with approximately 5 cm (2. 0 in. ) of vermiculite
packing (see Fig. l). Pulverized coal was fed from a conventional vibrating
screw feeder at a variable rate, usually from about 1. 3 to 5. 5 kg per hour
(3 to 12 lb/hr). Coal was mixed with primary air by means of a jet ejector which
used primary air as the working fluid. The amount of outside air inducted with
the coal was determined by calibration without coal flow and by neglecting the
influence of the solids on the induction rate. The coal-air mixture entered the
furnace through a 2. 5 cm (1. 0 in.) diameter tube, the flow from which was
interrupted by a bluff-body stabilizer as shown in Fig. 1. The stabilizer was
used to increase combustion efficiency over that of coaxial injection. Secondary
air was added from a concentric ring which introduced air uniformly into the
furnace. Natural gas, used for prefiring, was introduced with the primary air.
Both primary and secondary air streams could be preheated independently,
with maximum capabilities of 66° C (150°F) and 400° C (750° F), respectively,
at the 3.2 kg/hr (7 lb/hr) nominal feed rate. Coal and air feed rates were con-
trolled remotely, the former by D.C. motor control on the screw feeder, and
the latter by remote flow regulators which controlled the pressure behind criti-
cal flow metering orifices (see Fig. 2).
Sampling System
Chemical samples were taken at the furnace exit and analyzed by a non-
dispersive infrared analyzer (NDIR) specifically for nitric oxide. A water-cooled
quartz probe was used which separated particulates from the gas to be analyzed
by inertial techniques. Figure 3 shows a schematic drawing of the probe. Dirty
176
-------
NITROGEN BLANKET
LIVE COAL BIN
PREHEATED PRIMARY AIR
NATURAL GAS
ttWVWttmttZ ZIRCONIA rax:;*:*:*:*:.:*:-:*?
PREHEATED SECONDARY AIR
6 cm
v///////.
W////////////A
Fig. 1. Schematic Drawing of Coal Feeding Apparatus and Coal Burner
-------
-J
oo
•M-
V
Y AIR
QUADRUPOLE MASS
SPECTROMETER
NON-DISPERSIVE
INFRARED ANALYZER
MULLITE
FURNACE
3" I.D.
60"
SECONDARY AIR
X
X
O_
COLD
TRAP
COLD
TRAP
\
r
1
]
1
-W\r
-78°C
0°C
1
PROBE
COOLING
WATER
1
i
^
^/
//Y/
yy,
///:
QMS
AMPLIFIER
O/-W
NDIR
_Q AMPLIFIER
TO COMP
9 9
TELE
INPUT FROM
PRESSURE TRANSDUCERS
AMPLIFIER
-I r
MULTIPLEXER
1
DIGITAL
VOLTMETER
• THERMOCOUPL
CONTROL ROOM
COMPUTER
Fig. 2- Schematic Drawing of Coal C ombustion Apparatus and Ins trvzmentatiora
-------
3.
5"
— »»
TO WATER ASPIRATOR
• ^ a/i6" i.u.
It
_x
•i
•••
_x
«
COOLING WATER OUT<~
-^
•••
V.
^_
T
^^ QUARTZ WALL AND TUBING
4 1 «rii r>l*
^ 1 ,ZO UIA.
^ SAMPLE TO NDIR
Fig. 3. Schematic Drawing of a Water-Cooled Dirty Gas Sampler
179
-------
furnace gases were aspirated into the probe at a rate of about 40 */min, corre-
sponding to a sampling velocity of about 30 m/sec. A particulate-free gas
sample was withdrawn from this stream by pumping at right angles to the
accelerated dirty sample stream. The sample rate of the clean gas stream was
approximately 1.5 A/min. Cooling water at 88° C (190° F) was circulated to pre-
vent condensation of water in the probe. Clean gas samples were then dried by
successive wet ice and dry ice cold traps, and analyzed by the NDIR.
Coal
The coal used in the study was from the Mojave field, and was supplied
as a wet powder. The material was air-dried and sifted through a 160-mesh
screen. The resulting powder was such that 70% passed through a 200-mesh
screen. The ultimate analysis for this coal is given in Table 1.
Table 1. Ultimate Analysis of Mojave Coal
% c
% H
% S
% N
% O
% Ash ....
Btu/lb ....
68.3
5. 16
1.0
1. 21
23.4
10.9
9, 520
Temperature Measurements
Furnace temperature was measured with platinum-platinum/10% rhodium
thermocouples placed on the outside of the Mullite furnace tube. Thermocouples
were placed at 2.5 cm (1 in.), 50 cm (20 in.) and 140 cm (54 in.) from the burner.
In addition, a bare-wire thermocouple was placed in the exit plane gases to mea-
sure the temperature of the sample. The furnace temperature was taken as the
middle (50 cm) thermocouple temperature. Air temperature measurements,
both for air preheat and flow metering, were made with chromel-alumel thermo-
couples. All thermocouple readings were referenced to 65.56°C (150° F).
180
-------
Experimental Procedure
The furnace was prefired with natural gas and preheated secondary air
before each experimental point was taken. Upon achieving proper furnace
temperatures, primary and secondary air flows were adjusted, the natural gas
flow was stopped, and coal was introduced. Peak furnace temperatures from
prefiring ranged from 1400° C (2550° F) to 1300°C (2370° F), depending on the
level of secondary air preheat. The NO emission level was read when the
ji.
furnace temperature dropped to the desired point.
EXPERIMENTAL RESULTS
Effect of Equivalence Ratio on NO
J\.
Both constant air flow and constant fuel flow tests were made to determine
the influence of equivalence ratio (defined as fuel-to-air) on the formation of NO .
In the case of constant air flow, the primary air rate was 1.37 kg/hr (30.2 Ib/hr)
and the secondary air rate was 28. 2 kg/hr (64. 3 Ib/hr). At stoichiometric con-
ditions (air-fuel ratio of 13. 8), the coal feed rate was 3.11 kg/hr (6. 85 Ib/hr).
The primary air preheat level was 65. 6° C (150° F) while the secondary air was
unheated and entered at 21° C (70° F). For tests with constant fuel flow, the coal
feed rate was maintained at 3.52 kg/hr (7.75 Ib/hr), the primary air remained
constant at 13.7 kg/hr, and the secondary air was variable. In each case, data
were taken when the furnace temperature fell to 2100°F.
Figure 4 compares the results of each method of operation on an as mea-
sured basis. For constant air flow, the level of NO increased steadily from
160 ppm at an equivalence ratio of 0. 5 (200% theoretical air) to a maximum of
1160 ppm at an equivalence ratio of 1. 8 (56% theoretical air). The NO -equiva-
lence ratio profile was nearly linear and it appeared that the data may have
reflected a counting of coal particles, with each particle contributing equally to
NOx production. The results with constant fuel flow, however, showed quite
similar results, indicating that the effect was truly attributable to mixture ratio.
Figure 5 shows the same data but reduces the NO levels to an equivalence
ratio of unity. This was done by dividing the NO value as measured by its
181
-------
oo
to
i
CL
X
O
z
(J
Z
O
u
1000
800
u_ 600
400
200
0
FURNACE TEMPERATURE = 2100 F
_ COAL ANALYSIS:
C
H
ASH
S
N
CONSTANT AIR
FEED RATE (94.5 Ib/hr)
CONSTANT COAL
FEED RATE (7.75 Ib/hr)
FUEL-RICH
68.31%
5.16%
10.99%
1.00%
1.21%
0.2 0.4 0.6 0.8 1.0 1.2
EQUIVALENCE RATIO
1.4
1.6
Fig. 4. Influence of Burner Equivalence Ratio and Furnace Operating Procedures
on Nitric Oxide Emissions from Pulverized Coal Combustion
-------
II
-e-
1000
z
Q
z
LLJ
U
Z
o
u
800
o
UJ
ss
I
o.
V
LU
X 600
O
Z 400
200
FURNACE TEMPERATURE = 2100°F
CONSTANT AIR FEED
RATE (94.5 IbAr)
CONSTANT COAL FEED
RATE (7.75
I
0.2 0.4 0.6 0.8 1.0
EQUIVALENCE RATIO
1.2
1.4
1.6
Fig. 5. Influence of Burner Equivalence Ratio and Furnace Operating Procedures
on Nitric Oxide Emissions from Pulverized Coal Combustion
-------
equivalence ratio. This procedure is accurate in air-rich flames where the
oxidation product is primarily CO , but is not accurate in fuel-rich flames where
the composition of the combustion gases is uncertain.
On a reduced basis, the level of NO was relatively low in the very lean
!X
flames, but approached a steady value of about 650 ppm between equivalence
ratios of 0. 9 and 1. 8. The fuel-rich behavior was quite unexpected since one
expects a reduction in both fuel nitrogen conversion and air nitrogen fixation in
the overall reducing atmosphere of the fuel-rich flames. Since the same be-
havior was observed in both constant air and constant fuel operating modes,
however, the effects must be attributed to those of air-fuel stoichiometry. For
the Mojave coal, fuel nitrogen amounted to 1.21%, corresponding to nitric
oxide levels of about 1550 ppm on a dry basis. It is possible that the constant
650 ppm level represents the volatile part of the fuel nitrogen, i.e., the fraction
'that burns first in an overall oxidizing atmosphere.
The Effect of Secondary Air Preheat
To test the influence of combustion air temperature on NO , the tempera-
5t
ture of the secondary air was varied while holding the primary air preheat tem-
perature constant at 65. 6° C (150°F). The tests were run with a constant air
feed rate with variable coal feed. Primary air represented 31% (13. 7 kg/hr)
of the total air, and the coal feed rate at stoichiometry was 3. 1 kg/hr (6. 85 Ib/hr),
Figure 6 shows the results of tests over a range of equivalence ratios,
and secondary air temperatures, but a constant furnace temperature of 2100°F.
Secondary air preheat appears to have a strong influence on NO in very air-rich
3C
flames, with NO levels at an equivalence ratio of 0. 5 (200% theoretical air)
X.
ranging from 450 ppm without preheat to 1250 with 343° C (650° F> preheat. All
NO -equivalence ratio profiles, however, converged on the 650 ppm level in
fuel-rich flames. At an equivalence ratio of 0. 8 (125% theoretical air) where
industry prefers to operate, the reduction in NO with secondary air preheat
ji.
amounted to 38% as the secondary air preheat dropped from 343° C (650° F) to
21° C (70° F). The primary air was preheated only to 66° C (150° F) and repre-
sented almost a third of the total air. Thus, the average air preheat at 343° C
(650° F) secondary air preheat was approximately 257° C (495° F).
184
-------
00
Ul
II
-6-
O
Q 1000
UJ
Q
LLJ
800
O
X 600
U
400
UJ
U
Z
O
U
200
FURNACE TEMPERATURE = 2100 F
(70°F)
I
FUEL-RICH -
I I
0.2 0.4 0.6 0.8 1.0
EQUIVALENCE RATIO
1.2
1.4
1.6
Fig. 6. Influence of Burner Equivalence Ratio and Secondary Air Preheat
on Nitric Oxide Emissions from Pulverized Coal Combustion
-------
The Effect of Furnace Temperature
The prefired furnace was designed to operate over a very wide range of
stoichiometry, and, as a result, had a high surface-to-volume ratio (0. 526 cm" )
relative to other furnaces of similar throughput. Thus, the effect of furnace wall
temperature would be magnified in this furnace if, indeed, the wall temperature
influenced NO formation. Figure 7 shows the results of tests with 343° C
secondary air preheat at three air-coal mixtures. As expected, lower furnace
temperatures lower NO emissions, but the reduction from 1200°C (2200° F) to
.X
870° C (1600° F) was only 22%. Bienstock et al. (1) showed similar results
correlating the effect of primary zone flame temperature on NO formation.
While their measured NO levels were lower, a 20% reduction in NO was ob-
x x
served for the same-(600° F) reduction in measured flame temperature. They
observed, however, that the effect of temperature from 1200° C (2200° F) to
1425° C (2600° F) was quite large.
The effect of furnace wall temperature was greater at the lean conditions
than at stoichiometric conditions. This was probably due to the lesser heat re-
lease rate at the lean conditions. The nominal heat release rate at stoichio-
92 3
metric conditions was 2.66 x 10 cal/hr m (300, 000 Btu/hr ft ).
DISCUSSION
The Influence of Furnace Design
The results shown were obtained in a coal furnace of fixed burner design
and with a short residence time for combustion gases. The average residence
time in the furnace at stoichiometry and with a coal feed rate of 3. 1 kg/hr was
about 0. 6 sec, which was not nearly sufficient for char burnout. Thus, the
results pertain primarily to the NO^. formed near the primary flame zone. The
absolute NO levels were relatively high in this study, and this can be attributed
Ji
to the lack of sufficient residence time for the NO to dissociate at lower furnace
.X.
temperatures. Bienstock, et al. (_!) noted a rather large relaxation in NO
X.
levels as the combustion gases traveled through their more conventional furnace,
and attributed these to dissociation of NO. It is also possible that reactions
during char burnout lower the NO substantially.
186
-------
oo
-4
1000
UJ
o
X
o
y aoo
z
IS
u
o
600
400
200
D
AIR FEED RATE = 94.5 Ib/hr
(31% PRIMARY)
SECONDARY A« PREHEAT = 650°F
PRIMARY AW PREHEAT = I50°F
I
I
1600
1800 2000 2200
FURNACE WALL TEMPERATURE, °F
Fig. 7. Influence of Furnace Wall Temperature and Burner Equivalence Ratio
on Nitric Oxide Emissions from Pulverized Coal Combustion
-------
The configuration of the burner is known to be an important factor in NO
X
formation (5). The purpose of the present work, however, was to make a
parametric study of combustion modifications only, with possible burner design
studies to be made later. In the present case, primary and secondary air in-
jection occurred in close proximity, and it is likely that both air streams were
well-mixed before combustion commenced. Thus, in the present design, one
probably would not expect a large influence on NO of the relative flow rates
of the two air streams. In addition, air preheat could probably be considered
as an average primary and secondary air temperature.
The Influence of Combustion Modifications^
Low excess air firing generally reduces emissions of nitric oxide in both
boiler operation (6) and in research furnaces (l_, 2, 3_, 4). Data in highly fuel-rich
flames has not been available, however, and only a limited amount of data have
been presented on the effects of air preheat. The results shown in Fig. 6 indi-
cate that, while the air-rich NO characteristics of the coal-air flame are highly
.X
dependent on secondary air preheat, NO from stoichiometric and fuel-rich
J\.
flames is relatively independent of both preheat and air-coal ratio. These re-
sults are quite different from those from similar tests on gas and oil firing
where moderately fuel-rich operation results in sharp reduction in NO . The
j£
level of NO in fuel-rich coal flames corresponds to about 42% of the NO that
X X
would be found with 100% fuel nitrogen conversion. Coincidentally, it has been
reported that the volatile fraction of fuel nitrogen is about 40% of the total in
some pulverized fuels (3).
It is suggested that NO emissions in near-stoichiometric and fuel-rich
X.
flames are due almost exclusively to conversion of the volatile fuel nitrogen
to nitric oxide. The volatile fraction of the coal is burned in an overall oxidizing
atmosphere as long as the overall burner stoichiometry is below about 2. 5
(corresponding to stoichiometry with 40% volatiles). In these flames, conversion
of gaseous fuel nitrogen is high, and NO emissions, reduced to stoichiometric,
jfi
are relatively independent of coal-air ratio and air preheat. At overall burner
stoichiometries in the air-rich region, emissions of NO are strongly dependent
X,
on secondary air preheat, indicating that the flame temperature in these regions
dominates both fuel nitrogen conversion and molecular nitrogen conversion.
188
-------
This view of NO formation in pulverized coal flames indicates two alter-
3C
natives to NO control by combustion modifications. First, NO can be reduced
X. 3i
by reducing combustion temperature of the volatiles. This can be accomplished
by product gas recirculation, water injection, or similar methods designed to
vitiate the initial fuel-air mixture. Second, NO can be reduced by burning the
jC
volatiles at overall burner equivalence ratios greater than 2. 5. This can be
accomplished by limiting primary air to substantially less than 25% of the total
required for combustion.
ACKNOWLEDGEMENT
This work was supported by the Director's Discretionary Fund at the Jet
Propulsion Laboratory. Coal was supplied by Dr. J. Shapiro of the Bechtel
Corporation.
189
-------
REFERENCES
1. Daniel Bienstock, Robert L. Amsler and Edgar R. Bauer, Jr., "Forma-
tion of Oxides of Nitrogen in Pulverized Coal Combustion, " Journal of the
Air Pollution Control Association, 1_6 (8), 442(1966).
2. C. R. McCann, J. J. Demeter, A. A. Orning, D. Bienstock,
Emissions at Low Excess Air Levels in Pulverized Coal Combustion, "
Presentation at ASME Winter Meeting, New York, N.Y., November 29-
December 3, 1970.
3. M. P. Heap and T. M. Lowes, "Development of Combustion System De-
sign Criteria for the Control of Nitrogen Oxide Emission from Heavy Oil
and Coal Furnaces, " Progress Report No. 11, Contract No. 68-02-0202,
U.S. Environmental Protection Agency, Durham, N. C., December 15,
1972.
4. C. R. McCann, J. J. Demeter, J. Dzubay, and D. Bienstock, "NO
Emissions from Two-Stage Combustion of Pulverized Coal, " Presenta-
tion at 65th Annual Meeting of the Air Pollution Control Association,
Miami Beach, Fla. , June 18-22, 1972.
5. M. P. Heam, T. M. Lowes and R. Walmsley, "The Effect of Burner
Parameters on Nitric Oxide Formation in Natural Gas and Pulverized
Fuel Flames, " Presentation at "American Flame Days, " Air Pollution
Meeting, Chicago, 111., September 6-7, 1972.
6. Dr. J. Shapiro, Bechtel Corporation, Personal Communication, June 1972.
190
-------
PILOT AND FULL SCALE TESTS
PART I
191
-------
The Effect of Design
and Operation
Variables on NOX
Formation in
Coal Fired Furnaces
W. J. Armento
Research Specialist
Chemistry and Combustion Section
Alliance Research Center
Alliance, Ohio
W. L. Sage
Group Leader, Combustion Systems
Chemistry and Combustion Section
Alliance Research Center
Alliance, Ohio
Presented to
Pulverized Coal Combustion Seminar
National Environmental Research Center
Research Triangle Park, North Carolina
June 19-20, 1973
Sponsored by
Environmental Protection Agency
Research Triangle Park, North Carolina
Contract No. 68-02-0634
DTP 73-4 193
-------
THE EFFECT OF DESIGN AND OPERATION VARIABLES
ON NOX FORMATION IN COAL FIRED FURNACES
W. J. Armento, Research Specialist, Chemistry and Combustion Section,
Alliance Research Center, Alliance, Ohio
W. L. Sage, Group Leader, Combustion Systems, Chemistry and Combustion Section,
Alliance Research Center, Alliance, Ohio
Presented to Pulverized Coal Combustion Seminar, National Environmental Research Center,
Research Triangle Park, North Carolina, June 19-20, 1973
Sponsored by Environmental Protection Agency,
Research Triangle Park, North Carolina
Contract No. 68-02-0634
I. INTRODUCTION
The purpose of this EPA contract is to
determine the effectiveness of methods
of NOX control which can be used on coal
fired utility boilers, present and
future. In addition, we wish to iden-
tify potential problems in boiler oper-
ation and thermal efficiency. A com-
parison of the relative effectiveness
of these combustion control methods is
also to be made for gas and oil.
The contract is divided into three
phases which are defined below:
• Phase I - Identification of impor-
tant variables in NOX control and
qualitative determination of change
in NOX emission levels with change
in each method of combustion control
separately (6-month effort, now fin-
ished using a single burner unit).
• Phase II - Quantitative correlation
of the trends found in Phase I and
determination of the extent of
interdependence of variables (6-
month effort which will include con-
struction of a multiburner unit for
use in Phase III).
• Phase III - Verification and expan-
sion of the quantitative correla-
tions found in Phase II using the
new multiburner unit (12-month
effort due to start about January,
1974).
The basic combustion tunnel used for
testing in Phase I is a single burner
unit which fires ~ 227 kg hr-1 of coal
(~500 Ib/hr) at normal load. The coal
used in Phase I had a combustion enthal-
py of 6.67 kcal g-1 ( 12,000 Btu/lb)
dry and this corresponded to a heat
release rate in the active furnace vol-
ume of 390,000 kcal m"3 hr'1 (-44,000
Btu/ft3/hr). The total heat release
rate is ~ 1.4 x 10° kcal hr'1 (~ 5 6 x
10° Btu/hr).
There are two types of control for the
reduction of NOX which can be used:
operational control techniques and
design methods. Operational control
techniques are more easily applied to
existing units; the physical change on
the unit is minimized and the cost of
the change will be lower. Design
methods would be applied to new units
yet to be constructed; thus a minimum
of redesign/reconstruction costs would
be entailed. The operational variables
that have been studied in Phase I are:
(1) excess air, (2) firing rate of the
fuel (or unit load), (3) preheat of the
air, and (4) swirl (for natural gas
firing only). The design variables
included in Phase I studies were: (1)
flue gas recirculation, (2) two-stage
combustion, (3) fuel type (gas, coal, or
oil), and (4) heat removal rate from the
combustion gases (quench).
194
-------
II. UNIT DESIGN
The basic combustion unit used for the
Phase I tests (Figure 1) is a cylindri-
cal tunnel 1.4 meters (4.5 feet) in
diameter and 2.4 meters (8 feet) in
length with an overall effective furnace
volume of 3.60 m3 (127 ft3). The fuel
and air normally enter into the furnace
together through the burner area. For
coal transport, about 15% of the rated
load total air carries the coal into the
furnace through the central fuel pipe;
this air is defined as the primary air.
The balance of the combustion air nor-
mally enters the furnace around the cen-
ter pipe and is defined as the secondary
air. With gas, there is no primary air
and oil is steam atomized so that again
there is no primary air.
perpendicular to the burner axis and
are offset circumferentially to create
a swirling action. The combined inlet
area of the side slots is the same as
the front slots.
The coal burner (Figure 4) consists of
a pipe with concentric pieces of flared
welded sheet metal at the end to spread
the coal into the secondary air. The
burner arrangement used during most of
the tests consisted of six fixed
(non-moveable), curved vanes in the
burner throat to add some swirl to the
secondary air. The coal spreader is
normally just at the inside entrance to
the throat during firing. At other
positions further into the throat,
the flame leaves the burner and travels
partway down the furnace.
NATURAL GAS
LIGHTER
FIGURE 1. BASIC COMBUSTION UNIT
(SINGLE BURNER)
FIGURE 2. FRONT SLOT POSITIONS
There are two slot positions for two-
stage combustion in the front and side
of the furnace (Figures 2 and 3). The
two front slots are cut 3" by 12"; the
1" on either side of the center 1" x 12"
opening is bricked in. The front slots
are arranged so that the second stage
air enters parallel to the burner axis.
The combined inlet area of the front
slots is one half of the burner inlet
area.
The two side slots for two-stage com-
bustion are cut 6" by 6" and bricked 2"
on either side of the 2" x 6" opening.
The side slots add the second stage air
FIGURE 3. SIDE SLOT POSITIONS
195
-------
FIGURE 4. COAL BURNER
The natural gas igniter is left on
during all coal firing tests to ensure
that ignition is maintained in the
furnace, especially during unstable
combustion conditions. The flow is set
at 20% full gas flow for the igniter,
but this flow of gas supplies less than
1% of the total heat release at normal
loads. Under reduced loads, it is less
than 2% of the total heat release in
the furnace.
The gas burner (Figure 5) is a supply
ring with eight equally spaced spuds
for gas inlet into the throat. There
is a split washer about 1" from the tip
of the spud which serves the purpose of
holding the flame. Next to the washer
are two sets of holes drilled diamet-
rically opposed. The two sets are side
by side on the burner axis. The holes
are oriented for all eight spuds so
that the gas exits tangent to an imag-
inary circle which is approximately one
half of the burner diameter.
FIGURE 5. GAS BURNER
The secondary air swirler used for gas
and oil firing is a set of 16 moveable
vanes which can be set from 0°, or no
swirl, to about 30°, or maximum swirl.
Only the gas flame remained ignited in
a stable manner at no swirl; the coal
flame left the burner at lower swirl
angles (less than 20° to 30°) and went
out or was half-way down the furnace
under no swirl (0° to 10°).
The dual fluid oil burner (Figure 6)
employs steam as the atomizing medium
and consists of three concentric pipes
mounted on the axis of the burner.
The outermost pipe holds the impeller
which is a conical slotted disk of
sheet metal with the vanes bent to
create a slight swirl. The innermost
pipe carries the steam and the middle
pipe carries the oil. There is an
atomizer tip on the end of the two
inner pipes (Figure 7). The center
hole in the atomizer (Y-jet) carries
the steam and the No. 6 oil, which is
preheated to retain fluidity, is in
the branched opening which enters the
steam jet.
FIGURE 6. OIL BURNER
FIGURE 7. OIL ATOMIZER
196
-------
III. INSTRUMENTATION
The flue gas is sampled in the stack
where the average temperature is down
to less than 500°K (450°F). A stain-
less steel probe extends half-way
across the stack. The gas sample is
passed through a condenser maintained
at 283°K (50°F) and then is split and
monitored by the following instrumen-
tation (Figure 8):
(1) An MSA LIRA CO analyzer (measur-
ing from 0 to 5000 ppm)
(2) A Whittaker S02 analyzer of the
chemical cell type (0 to 5000
ppm)
(3) An MSA paramagnetic 02 analyzer
(0 to 25%)
(4) A Bailey 02 Meter (0 to 10%)
and total combustibles (0 to 5%)
hot wire analyzer.
(5) A TECo (Thermo Electron Corp.)
chemiluminescence NOX monitor
(0 to 10,000 ppm)
(6) Whittaker NOX chemical cell
analyzer (0 to 5000 ppm) with a
Mallcosorb column
(7) A Beckman NDIR for NO (0 to
1500 ppm)
The Beckman NDIR requires the use of
another ice bath maintained at a con-
stant 274°K (34°F exactly, ±0.5°F) to
remove a maximum amount of water. Even
so, enough water remains in the gas to
be measured so that a 60 ppm correction
must be made to the final measurement
as a residual H20 correction.
The gases used for the NOX calibrations
are certified calibration gases from
Matheson containing about 200, 500,
800, and 1200 ppm NO (less than 5 - 20
ppm N02) in N2- The gases, when pur-
chased, are checked against the previous
standards before being put into reg-
ular use.
MALLCOSORB COLUMN
60 F ICE BATH
WHITTAKER
S02
FIGURE 8. INSTRUMENTATION
IV. TEST PROGRAM
The test program for Phase I called for
identification of the major variables
for NOX control. The major variables to
be investigated were divided into design
and operational variables. Design vari-
ables were expected to be more easily
implemented in new units not yet built.
In order to change existing units, the
expense would be very great when com-
pared to modifying operational condi-
tions. It might also be physically or
economically impractical to modify
existing units. The design variables,
therefore, suggest that physical and
mechanical changes could be made in the
unit and would include flue gas recir-
culation, staged combustion, quench
rate, and basic fuel type. On the
other hand, it would prove less expen-
sive to modify the operational variables
for existing units such as excess air,
fuel firing rate, air preheat, and
swirl. (These groups of variables are
not meant to be mutually exclusive).
Since the basic contract fuel require-
ment was for coal firing, limited tests
with gas and oil were used for compar-
ison with the coal results under the
same furnace conditions. The burner
configuration for the coal was changed
from the fixed 6-vane burner to the
variable angle 16-vane burner only
once in an attempt to study the effect
of swirl on coal combustion. Base
tests with coal through either burner
showed less than 10% variation in NOX
emission levels.
197
-------
V. RESULTS
The objective of the initial phase of
this investigation was to establish
relative effects of changes in oper-
ational and design variables. Hence,
an attempt was made to hold other con-
ditions constant and to change the
variable under study over a wide range.
However, many of these are interrelated,
hence as load, excess air, preheat,
staged combustion, flue gas recircula-
tion and the other variables were
changed, air velocities changed leading
to a variation in mix rate, turbulence
and combustion intensity. To date, no
attempt has been made to interrelate
these variables, but an attempt to do
so is projected as a part of Phase III.
During operations, continuous mea-
surements were made of C>2 (two instru-
ments), CO, combustibles, SCL, and NO
(three instruments). The measurements
were all recorded on strip chart
recorders. On random occasions, about
201 of the time, ash loadings in the
stack gases were obtained for deter-
mination of unburned carbon.
When surveying a single variable,
generally the extremes and the middle
of the range were tested. The trends
were determined in this way. However,
it should be emphasized that in many
cases, the extremes on a single vari-
able are beyond the range of what is
considered acceptable operating prac-
tice. Furthermore, it should be
emphasized that these results apply to
a small single burner test unit.
Although it is believed that the trends
shown will probably hold for a large
multiburner furnace, the magnitude of
the changes may be appreciably dif-
ferent; also, some of the test vari-
ables may not be operationally accept-
able on large units.
The effect of each single variable
will be discussed below. In the case
where absolute measurements of NO
emission levels are shown, a single
test (or two) to verify proper instru-
mentation operation was made and the
test runs for data collection were made.
But, when relative measurements of NO
emission levels are shown, the tests
were run in pairs. The base line test
and the data test were run in the same
day to eliminate possible day-to-day
variations. The NO reduction was then
calculated by comparing the change in
NO emission level to the emission level
in the base line test.
A. Excess Air
Figure 9 shows the effect of excess
air at rated load with 600 - 650 °K
(600 - 700°F) air preheat on NO emis-
sion for natural gas (no fuel-bound
nitrogen), #6 fuel oil (0.23% fuel-
bound nitrogen), and coal (1.1% fuel-
bound nitrogen). The relative
positions of the curves with coal
highest, gas intermediate, and oil
slightly below gas agrees in general
with data from field units. However,
the natural gas and oil NO emission
levels fall in a low range and one
explanation for this is the fairly
conservative rating (furnace heat lib-
eration rate) for these fuels in our
test unit.
20
EXCESS AIR, %
FIGURE 9. EFFECT OF EXCESS AIR ON NOX
The NO emission from gas is a result
of thermal fixation of atmospheric
nitrogen. The degree of thermal fix-
ation of N2 into NO depends on tem-
perature and excess oxygen level.
At lower excess air levels, the effect
of increasing the oxygen level more
198
-------
than offsets the decrease in temper-
ature, hence the rising curve. In
addition, at low excess air, the flame
can be more luminous and this may lead
to a more rapid quench rate. At higher
excess air, eventually the decreasing
temperature becomes the overriding
effect and the curve drops again after
reaching a maximum.
In contrast, the coal contains fuel-
bound nitrogen and the conversion of
this represents a second source of NO.
It is expected that the differences in
the shape of the curve and peak posi-
tion when compared to the gas curve is
a result of the fuel-bound nitrogen
conversion. The trend shows a cor-
relation of higher fuel-bound nitrogen
conversion to NO with higher excess
air levels. The evidence for this is
shown in a steeper slope to the NO
curve at lower excess air levels and
a peak position at higher excess air
level. It is not felt at this stage
in the program that the fuel-bound
nitrogen conversion can yet be quanti-
tatively evaluated.
The initial thought for oil is that
this curve should be higher than gas
due to the fuel-bound nitrogen con-
tent. In our unit, oil burns with a
more luminous flame than gas and has
a larger visible flame envelope.
Hence, the lower NO levels from oil
are attributed to better radiating
properties and therefore a lower bulk
gas temperature.
B. Load
The dependence of coal NO emissions
on load is illustrated in Figure 10.
There is no indication of a peak level
being reached under these conditions.
Gas and oil NO levels show only a
slight dependence on firing rate.
Again it should be pointed out that
with the burner arrangement used in
these tests, the air velocity through
the 'burner and the turbulence change
with load.
Also to be considered is the refractory
shielding in the front of the fur-
nace. This probably means that there
is less response of NO to load than
would be obtained with a higher
quench rate.
C. Preheat
The dependence of NO emission level
on air preheat can be seen in Fig-
ures 11, 12, and 13. Figure 11
results from plotting all points run
at 15% excess air, Figure 12 from all
points run at normal load and low
excess air, and Figure 13 from all
points run at normal load and 151
excess air. Figure 11 indicates a
high dependence of coal and gas on
air preheat (about 50 ppm NO increase
per 100°F increase in preheat) with
oil showing a smaller effect. How-
ever, if the percentage increase in
NO is used instead of absolute change
in ppm NO from 400 to 650°K (300 to
700°F) preheat, coal shows a 35%
increase, oil 33%, and gas 71%. Fig-
ure 12, the curves for low excess air,
shows a significant dependence on pre-
heat for gas only. There is no effect
for coal and oil. The change for gas
is still about 50 ppm NO increase per
100°F of preheat increase. Figure 13
at normal load and 15% excess air shows
about the same effects as Figure 11.
A comparison of these three figures
leads to the tentative conclusion that
conversion of fuel-bound nitrogen
increases with decreasing air pre-
heat.
200
FIRING RATE, Kg. COAL/HR.
FIGURE 10. EFFECT OF FIRING RATE ON NO
199
-------
500
PREHEAT. F
FIGURE 11. EFFECT OF PREHEAT ON NOX
(CONSTANT EXCESS AIR)
500
PREHEAT, F
FIGURE 13.
EFFECT OF PREHEAT ON NOX
(CONSTANT LOAD AT NORMAL
EXCESS AIR)
D. Flue Gas Recirculation
The coal burner arrangement which has
been used requires transport air for
the coal which is defined as primary
air (see Figure 14). The balance of
the burner air is called secondary air.
The primary air is about 15% of the
total air under normal load and excess
air. At low loads, the percentage of
primary air increases to over 20%. The
oil is steam atomized and, therefore,
for both gas and oil, there is no
primary air in the burner.
I PREHEAT UP TO 400 C
• SECONDARY \ 900-2300 Kg./HR
r* PRIMARY
NO PREHEAT
180-270 Kg./HR
300 500 700
PREHEAT. P
FIGURE 12. EFFECT OF PREHEAT ON NOX
(CONSTANT LOAD AT LOW EXCESS
AIR)
Primary Flue Gas Recirculation
(Coal Only)
Figure 15 shows the result of primary
flue gas recirculation for coal com-
bustion. The flue gas was substituted
for air so that the total transport
gas weight remained constant. The air
displaced by the flue gas was added
to the secondary air to maintain a
constant air to fuel ratio. Since the
primary air represents about 15% of
the total air for combustion, a high
level of flue gas substitution repre-
sents only a small amount of flue gas
recirculation. Blending the flue gas
with the primary air does tend to
affect the primary ignition zone and
hence burner performance. As a result,
data obtained under these test con-
ditions showed considerable scatter.
However, it is felt that the trend
shown in Figure 15 is probably real-
istic and that any reduction in NO by
this approach is relatively insignif-
icant.
FIGURE 14. AIR INPUT FOR COAL COMBUSTION
02468
FLUE GAS RECYCLED. %
FIGURE 15. PRIMARY FLUE GAS RECIRCULATION
(COAL ONLY)
200
-------
Secondary Flue Gas Recirculation
Figure 16 shows the percentage of NO
reduction versus the percentage of
flue gas recycled. The flue gas was
added to the secondary flow and the
air to fuel ratio was held constant.
FLUE GAS RECYCLED, '
FIGURE 16. FLUE GAS RECIRCULATION
The curve for coal indicates for the
application of 10% to 15% flue gas
recirculation that a maximum reduction
of only 10% to 15% can be expected in
NO emission levels. The curve for gas
shows a very great reduction in NO
levels even with relatively low flue
gas recirculation. The dotted portion
of the line indicates that no tests
have been run below 101 flue gas recir-
culation for gas and the curve is
simply interpolated from the origin to
the data available. The oil curve
shows very little reduction of NO
emissions with flue gas recirculation.
At present, it is felt that these
data indicate flue gas recirculation
is effective in reducing thermal fix-
ation of NO but may increase the
amount of NO formed from conversion
of fuel-bound nitrogen.
E. Staged Combustion
As seen in Figure 17, there are two
areas in a normal utility unit where
second stage air can be added for
final combustion. The air can be
added in the burner area by running
some burners rich and some lean, or
by putting air only through separate
burners, etc. Or, the air can be
added at a distance away from the
burner area. Our two sets of slots
are spatially arranged to simulate
either method of second stage air
addition. The front slots simulate
addition of air in the burner area,
or very close to it, whereas the side
slots simulate addition of the air
further away from the burner area
with rapid mixing for the second
stage combustion.
#2
(SIMILAR TO SIDE SLOTSI
#1
• (SIMILAR TO
FRONT SLOTS)
FIGURE 17. STAGED COMBUSTION OPTIONS
A comparison of results obtained with
the two port locations is shown in
Figure 18. At the higher burner to
total air ratios results are similar.
It should be noted that these data
were obtained with fixed burner and
slot openings. Thus as the burner
to total air ratio decreases and
staging increases, the air velocity
through the ports increases and the
burner velocity decreases. This
effect on combustion performance is
unsatisfactory at lower burner to
total air ratios when using the side
slots for staging. The curves do
indicate that a 50% reduction in NO
emission levels by staging is possible
for coal firing.
201
-------
STOICHIOMETRY AT BURNERS,
FIGURE 18. STAGED COMBUSTION FOR COAL
PORT POSITION VARIABLE
Figures 19, 20, and 21 represent the
results from observation of staged
combustion for coal, oil, and gas,
respectively. For coal and oil, flue
gas recirculation showed no further
reduction in NO when used in com-
bination with staged combustion.
There are two curves for staged com-
bustion with coal (Figure 19). The
"S" shaped curve represents normal
two-stage combustion for the front
slot firing at 15% total excess air
and normal load. The straighter line
represents substoichiometric firing
in which the second stage air was
simply shut off and the overall air to
fuel ratio is the same as the burner
ratio in the first stage. The straight
line intercept indicates that below
75% burner air to fuel ratio, no NO
would be expected to appear in the
first stage combustion. The two-stage
combustion curve indicates that a min-
imum in the NO occurs at a burner air
to fuel ratio of about 50%. This min-
imum is not unexpected because it
represents the diminishing return on
the reduction of conversion of the
fuel-bound nitrogen in the first stage
as contrasted to increasing thermal NO
formation due to increased Btu content
in the second stage. There are two
major conclusions to be drawn from
this information. First, there is a
lower level of burner stoichiometry
below which the NO formation in the
first stage appears to drop to zero and
it is of no further benefit to lower
the air to fuel ratio at the burner.
Secondly, with the latent Btu content
of the gas increasing rapidly in the
residual gas to be burned in the sec-
ond stage, the second stage flame tem-
perature and therefore the thermal fix-
ation of NO is expected to rise again.
The position of maximum NO reduction
is influenced by port position and
heat removal rate as well as other
factors such as air velocity and mix-
ing rate. It is therefore certain
that such a turnover in the NO reduc-
tion curve should and probably does
exist.
The oil curve (Figure 20) is the same
shape as the coal curve. The degree
of reduction in NO levels for staging
.oil is less than coal because of the
much lower base levels of NO.
STOICHIOMETRY AT BURNERS, %
FIGURE 19. STAGED COMBUSTION (COAL)
I
I
I
100
STOICHIOMETRY AT BURNERS, %
FIGURE 20. STAGED COMBUSTION (OIL)
202
-------
For two-stage combustion of gas
(Figure 21), the addition of flue
gas recirculation becomes very impor-
tant. The greater effect for the
flue gas recirculation is found in
the burner. This indicates that it
prevents or reduces the formation of
the precursors in the first stage
which leads to thermal NO as well as
preventing formation of NO in the
second stage. If the only function
of flue gas recirculation in a natural
gas flame were temperature drop and
initially lowered 02 concentration,
it would be just as effective in the
second stage air as it is in the
burner air since combustion taking
place in the first stage is usually
substoichiometric. An increase in
velocity and mixing in the first
stage is also expected when flue gas
is added through the burner.
STOICHIOMETHY AT BURNERS, %
FIGURE 21. STAGED COMBUSTION (GAS)
F. Swirl (gas only) and Quench (coal only)
Decreasing the swirl with our burner,
as shown in Figure 5, for coal firing
led to combustion instability; hence
this variable was not investigated
for coal during this phase.
Only maximum swirl (30° vane setting)
and no swirl (0° vane setting) tests
were run for gas (Table 1). In gen-
eral, the effect of increased velocity
on mixing is shown. Higher load and
higher excess air led to greater
reductions in NO. The effect at low
air regardless of load (the point at
low load, low air was 10, the average
of 13 and 6 although the value of 6
is thought to be the correct one)
indicates that the 02 concentration
is never high enough to directly
influence thermal NO formation, or in
other words, the NO may be kinetically
limited rather than diffusion limited.
TABLE 1. EFFECT OF SWIRL (GAS ONLY) NO
REDUCTION, % (MAXIMUM TO NO SWIRL)
LOW
AIR
LOW
LOAD
10
MID
LOAD
HIGH
LOAD
4
MID
AIR
8
18
18
HIGH
AIR
12
18
21
The change in quench in the basic com-
bustion unit was made by removal of
part of the refractory brick lining
in the furnace. All tests except the
quench tests were run with a 1-inch
thick refractory brick lining on the
inside of the furnace from the burner
to half way down the furnace, or to
a distance of 4 feet. For a change
in heat removal rate for the quench
test, half of the brick was stripped
out so that it covered the interior
from the burner down a 2-foot length
of the furnace. The results of these
tests are shown in Table 2. Although
this alteration changed the quench
rate, there is not enough information
to make a quantitative calculation of
the change. All conditions except
low excess air showed about the same
reduction in NO emission levels.
TABLE 2. EFFECT OF QUENCH ON COAL
(PERCENT REDUCTION IN NO)
LOW
AIR
LOW
LOAD
MID
LOAD 0
HIGH
LOAD
MID
AIR
20
(20 STAGED)
20
HIGH
AIR
—
20
203
-------
VI. SUMMARY
The most effective means of controlling
NO emissions from an operational point
of view as found in the basic combustion
unit would appear to be to control
excess air, preheat, and load together
since they are heavily interdependent
(ref. Table 3). If load cannot be
varied, control of the combination of
lower preheat and excess air does not
appear to be more effective than con-
trol of the excess air alone.
To date, pilot plant results indicate
the concept of staged combustion or
perhaps delayed mixing appears to be
the most effective means of NOx reduc-
tion. If this can be accomplished by
progressive mixing in the individual
burner zone, then operator acceptance
seems assured. However, if effective
control requires a wide physical sep-
aration of the two sources of air,
there is a real concern that unaccept-
able slagging and corrosion problems
may be encountered. Therefore, field
testing of a unit under controlled
conditions to establish the long term
effect of operating in this manner
would be required before universal
acceptance can be assured.
TABLE 3. SUMMARY OF QUALITATIVE EVALUATION
INCREASING:
EXCESS AIR
LOAD - LOW PREHEAT
- HIGH PREHEAT
PREHEAT - LOW AIR
- HIGH AIR
FLUE GAS
RECIRCULATION
STAGED COMBUSTION
- POSITION
- WITH FGR
QUENCH (DECREASING)
SWIRL
COAL
•H-f
-H-
0
0
++
+
0
+
ND
GAS
+
0
+
++
-H-
0
ND
+
OIL
+
•f
•H
0
0
-
0
0
ND
ND
204
-------
PRELIMINARY EVALUATION OF COMBUSTION MODIFICATIONS FOR CONTROL OF
POLLUTANT EMISSIONS FROM MULTI-BURNER COAL-FIRED COMBUSTION SYSTEMS
By
C. R. McCann, J. J. Demeter and D. Bienstock
U. S. Department of the Interior, Bureau of Mines
Pittsburgh Energy Research Center, Pittsburgh,Pa. 15213
The Bureau of Mines (through an interagency agreement with EPA)
has extended a program to evaluate the effects of combustion modifications
on control of emissions from multi-burner coal-fired systems. Experimenta-
tion was conducted in a 500 Ib. per hour pulverized-coal fired unit whose
operation closely simulates industrial practice. A photograph of the furnace
is shown in figure 1. The unit is 12 ft. high, 7 ft. wide, and 5 ft. deep,
with water cooled walls. Four burners are located on the front wall. A
half section of the combustor is shown in figure 2. Combustion gases
leave the furnace at about 2000° F, cool to about 1000° F in the convective
heat transfer zone, then exchange heat with secondary air in the recupera-
tive air heater. The effects of several operating techniques have been
investigated single stage combustion with reduced excess air, two stage
combustion, bias-firing, and flue gas recirculation.
Coal feed rate was maintained at 500 Ib per hour, fuel particle
size at 75 percent through 200 mesh and secondary air temperature at 600° F
in all tests. Except for the series of reduced excess air tests, excess
air was maintained at 20 percent.
205
-------
Variations in Excess Air
By decreasing the amount of excess air from the conventional
levels of 20-25 percent, a substantial lowering of NOX emissions can be
achieved. As shown in figure 5, NOx decreased from 1.45 gm/10 cal at
20 percent excess air, to .5 gm/106 cal at about 2 percent excess air.
However, as shown in figure 4, this reduction in NOX emission was
accompanied by a decrease in carbon combustion efficiency.
Two-Stage Combustion
In two-stage combustion, the first stage was supplied with
air ranging from 80 to 105 percent of stoichiometric. Sufficient air
was introduced at the furnace outlet to produce 20 percent overall excess
air. Figure 5 shows a plot of NOX emissions as a function of air supply
to the first stage. Nitrogen oxides emission decreased from about
1.1 gm NO2/10 cal when 105 percent of stoichiometric air was supplied
to the first stage to .77 gm NO2/106 cal when 80 percent of stoichiometric
air was supplied to the first stage. This reduction was accomplished while
maintaining carbon combustion efficiency greater than 98 percent. Also
shown in this figure are the results of bias-firing experiments. This
is a variation of staged combustion, where stoichiometric air or less than
stoichiometric is supplied to the lower burners, with sufficient air
supplied to the upper burners to provide 20 percent excess air overall.
It is evident that little reduction occurred when 105 percent of
stoichiometric air was supplied to the lower burners. Some reduction did
occur when 90 percent of stoichiometric air was supplied to the lower
burners. It wasn't possible to reduce the air to the lower burners below
206
-------
90 percent, because of the increased demand on the upper burners.
A reduction to 80 percent of stoichiometric to the lower burners would
require 160 percent of stoichiometric to the upper burners. Stable
flames could not be maintained with air supplied in this amount.
To further investigate the effect of two-stage combustion, the second
stage air probe was relocated to a point nearer to the flame zone. A
sketch of the probe location is shown in figure 6. with air introduced
at this point, a survey was made of NOX emissions when the probe was rotated
through an angle of 180° from a point normal to the front wall to a point
normal to the rear wall. Also shown are the NOX values obtained when second
stage air was introduced at the various angles. Most significant reduction
occurred with air introduced normal to the front and rear walls. Highest
NOX values were obtained when air was introduced at an angle approximately
30° from normal to the front wall. At this point the second stage air
penetrated the primary combustion zone, increasing the intensity of
combustion. The flames were forced sown along the front sloping wall
resulting in overheating of the lower furnace section. Since the coal
feed rate was constant during the test, the oxygen values are an indication
of carbon combustion efficiency. Although a relatively large change in
NOV emissions occurred as the angle of introduction of second stage air
A
was varied, the oxygen values indicate that carbon combustion efficiency
was relatively constant. Furnace outlet temperatures were 1750° F, 1920° F,
and 1850° F when air was introduced at points 1, 2, and 7, respectively.
When the second stage air was introduced at an angle of about 45° at the
original probe location near the exit of the furnace, the NOX value was
about 300 ppm and the furnace outlet temperature was
207
-------
Flue Gas Circulation to Secondary Air
Figure 7 shows the results of tests in which various amounts
of flue gas were recycled to the furnace through the secondary air streams.
Temperature of the recycled gas was about 300° F. The curve indicates
that a significant reduction in NOV emission occurs when flue gas is
A
recycled to the furnace through the secondary air streams. Furnace outlet
temperature decreased with recycle from 2000° F at zero recycle to 1890° F
at 24 percent recycle. The curve appears to flatten out between 24 and
30 percent recycle. The 30 percent point and an intermediate point will
be checked to further define the curve in this region. In addition,
several tests are planned to determine the effect of flue gas recycle
in the primary air stream.
In addition to monitoring Oj and NOX, the following components
were also measured during the tests - NC^, SC>2, CO, CC^, hydrocarbons,
particulate, and furnace outlet temperature. The NO2 was monitored with
a chemiluminescent analyzer; the NO2 generally ranged from 3-7 percent
of the NOX value. An NDIR analyzer was also used to monitor NO.
The CO emissions, as measured with an NDIR analyzer, generally
ranged from 30 to 60 ppm for the standard and other tests except those
operated at low excess air levels. In the test operated at 5 percent
excess air, the CO increased to 1,000 ppm, at 2 percent excess air the
CO further increased to 5,000 ppm.
208
-------
Total hydrocarbon emissions were monitored during the tests
with a flame ionization analyzer. The analyzer indicated that ambient
and combustion air contained about 2-3 ppm hydrocarbons. Flue gas concentra-
tions were on the order of 0.5 to 0.8 ppm in tests conducted at 20 percent
excess air levels. The hydrocarbon emissions increased at lower
excess air levels to 2 to 5 ppm at 5 percent excess air.
The amount of slagging experienced at a given operating condition
is difficult to ascertain because of the relatively short period of opera-
tion at a given test condition. As far as could be noted visually, the
degree of slagging was minor, and no difference could be noted between
tests.
After the flue gas recycle tests are completed, the survey of
the effect of the point of introduction of second stage air will be
completed to include rotation of the second stage air probe through the
remaining two quadrants. Thereafter the combustion studies will be
continued to include combinations of the various combustion modifications.
209
-------
Figure 1. View of 500 Ib/hr pulverized coal-fired combustor.
210
-------
R«cup«rotiv«
0" preheoter,,
Convtclive heol tfontfer section
Figure 2. 500 Ib/hr coal combustor.
EPA regulations, coat fired plants
5 10
EXCESS AIR f percent
Figure 3. Nitric oxide formation as a function of excess air.
211
-------
100
1
i
98
96
94
(fl
S
I
8
o
00
ft 9O
3 O
\
\
\
\
IO IS 2O 25
EXCESS AIR, percent
3O
Figure 4. Carbon combustion efficiency as a function of excess air.
1.6
i
8
1.2
a
8 i-o
*
"5
<•
2 O.8
O* 0.6
«
0.*
EPA regulations , coat find plants
A
A Bias firing
x Two-stage combustion
I
7O 8O 9O IOO
Alt? TO FIRST STAGE, % of stoicAiomettic
110
Figure 5. Nitric oxide formation with off-stoichiometric firing.
212
-------
BURNERS
f. 3'-V 4^1' tin
' ^¥
: ^
^5* L * 7
«3^
P 3 ; .
l_
1
^^=^
^
>
i
•—
POINT
1
2
3
4
5
6
\ / ?
3.5
3.5
3.6
3.5
3.7
3£>
3JS
2SO 265
375 39O
3OO 310
32O 33O
345 355
295 3IO
270 29O
Figure 6. Effect of air injection ang1? upon NO emissions.
X
s~
o
8
£
o
3
o
o
M
O
s
Dl
2.O
EPA regulations , coal fired plants
O .05 .10 .15 .20 .25 .30 .35
FLUE GAS RECIRCULATION , mass flu* gas/ mass inlet air (fuel
Figure 7. Flue gas recirculated to secondary air, percent.
213
-------
EMISSION CONTROL FOR
COAL-FIRED UTILITY BOILERS
by A. R. Crawford, E. H. Manny and W. Bartok
Government Research Laboratory
Esso Research and Engineering Company
Linden, New Jersey
Prepared for "Coal Combustion Seminar"
Organized by the Combustion Research Section,
Control Systems Laboratory,
Environmental Protection Agency
Research Triangle Park, North Carolina
June 19-20, 1973
SUMMARY
Esso Research and Engineering Company is conducting field studies
on utility boilers under EPA sponsorship, to develop NOX and other pollutant
control technology, by modifying combustion operating conditions. In the
current phase of continuing work on this problem, Esso's mobile sampling-
analytical system has been used to test eight pulverized coal fired boilers
of cooperating electric utilities. These boilers, including wall, tangentially,
and turbo-furnace fired units, had been recommended by major utility boiler
manufacturers as representative of their current design practices.
In addition to gaseous emission measurements, particulate emissions
and accelerated furnace corrosion rates have been also determined in a number
of cases. Esso's test design consisted of three phases. First, statistically
designed short term runs were made, to define the optimum "low NOX" conditions
within the constraints imposed by boiler operability and safety, slagging, unburned
combustible emissions and other undesirable side effects. Second, the boilers
were usually operated for about two days under the "low NOX" conditions defined
in the first phase, to check operability on a sustained basis. Third, several
boilers were operated under both baseline and "low NOX" conditions for about
300 hours, with carbon steel corrosion coupons mounted on air-cooled probes
exposed near the water walls of the furnaces, to obtain relative corrosion
tendencies with accelerated rates.
Analysis of the gaseous emission data obtained shows that combustion
operating modifications, chiefly low excess air firing, coupled with staged
burner patterns, can reduce NOX emissions from the coal fired boilers tested
by 25 to 60%, depending on the unit and its flexibility for modifications.
The NOX emissions measured have been successfully correlated for both normal
and modified firing conditions with the per cent stoichiometric air supplied
to the burners.
215
-------
There are no major differences between particulate loadings under
baseline and "low NOX" operating conditions. However, unburned carbon in
the fly-ash increases considerably with "low NOX" firing for front wall and
horizontally opposed fired boilers but, decreases substantially for tangentially
fired units. The potential debits in overall performance based on these
limited data for front wall and horizontally opposed fired boilers should
be offset by improved efficiencies realized by lower excess air operation
in "low NOX" firing.
The accelerated corrosion tests have not revealed major
differences in corrosion rates between normal and staged firing operations.
More tests and long term runs, with particular emphasis on corrosion and
slagging problems are needed to demonstrate the promising leads uncovered
to date in this study.
216
-------
1. INTRODUCTION
In continuing studies sponsored by EPA, Esso Research and Engineering
Company (Esso) is involved in the development of nitrogen oxides (NOx) emission
control techniques for stationary sources. Our "Systems Study of Nitrogen Oxide
Control Methods for Stationary Sources" (1-3) characterized the nature and
magnitude of the stationary NOX emission problem, assessed existing and potential
control technology based on technical feasibility and cost-effectiveness,
developed a first-generation model of NOX formation in combustion processes,
and prepared a set of comprehensive 5-year R&D plan recommendations for the
Government with priority rankings.
Fossil fuel fired electric utility boilers were identified by the
above study as the largest single stationary NOX emission sector, responsible
for about 40% of all stationary NOX. Consequently, as part of Phase II of our
"Systems Study of Nitrogen Oxide Control Methods for Stationary Sources", we
conducted a systematic field study of NOX control methods for utility boilers
(4-6). xhe objectives of this field study were to determine new or improved
NOX emission factors according to fossil fuel type and boiler design type,
and to explore the application of combustion modification techniques to control
NOX emissions from such installations.
Esso provided a specially designed mobile sampling-analytical van
for the above field testing. Our van was equipped with gas sample, thermocouple
and velocity probes, with associated sample treating equipment, and continuous
monitoring instrumentation for measuring NO, N02, CO, C02, 02, S02, and
hydrocarbons.
Gas, oil, and coal fired utility boiler representative of the U.S.
boiler population were tested, with gas, oil, and coal fuels, respectively.
Combustion modifications were implemented in cooperation with utility
owner-operators (and with major boiler manufacturer subcontractors for three
of the coal fired boilers tested), and emission data were obtained in a
statistically designed field program. The 17 boilers (25 boiler-fuel
combinations) tested included wall-fired, tangentially-fired, cyclone-fired,
and vertically-fired units ranging in size between 66 and 820 MW generating
capacity.
Major combustion operating parameters investigated consisted of
the variation of gross boiler load, excess air level, staged firing patterns,
flue gas recirculation, burner tilt, primary/secondary air ratio, and air
preheat temperature. Operation under reduced load conditions reduced the NOX
emissions, but only for gas firing was the percent NOX reduction greater than
the percent load reduction. Base-line emissions were correlated in a
statistically significant manner with the MW generated per "equivalent" furnace
firing wall. In general, unburned combustible emissions, i.e., CO and
hydrocarbons were found to be negligibly small under base-line conditions,
and acceptably low even with NOX control combustion modifications. The N02
portion of the flue gas was always five percent or less of the total NOX emitted.
217
-------
The effectiveness of combustion modifications was found to vary
with individual boiler characteristics for each fuel. For gas fired boilers,
NOX emissions could be reduced on the average by about 60% at full load, even
though in large, gas fired boilers limited by heat transfer surface, NOX
emission levels as high as 1000 ppm prevailed in the absence of combustion
modifications. Uncontrolled emissions from fuel-oil fired boilers averaged
lower values than for>gas firing, but combustion modifications could be less
readily implemented. With coal firing, only two of the seven boilers tested
(one a tangential unit, the other a front wall fired boiler) could be operated
in a manner conducive to reducing NOX emissions. This operation consisted of
firing the operating burners in the lower burner rows or levels with
substoichiometric quantities of air, and supplying the additional air required
for the burn-out of combustibles (keeping overall excess air as low as possible)
through the air registers of the uppermost row or level. In these short-term,
exploratory tests, NOX emissions were reduced by over 50%, compared with the
standard firing mode, without decreasing thermal efficiency or increasing the
amount of unburned carbon in the fly-ash. Due to deactivating the pulverizer
mill to the top level of burners, the amount of fuel that could be fired was
reduced, resulting in a decrease of about 15% from maximum rated capacity.
The NOX reductions achieved were not affected by these reductions in load,
as normal and modified combustion operations were compared at the same boiler
load.
While the exploratory data obtained in the above study on controlling
NOX and other pollutant emissions from utility boilers by combustion modifications
showed good potential, a number of critical questions have remained to be
answered. Thus, for coal fired utility boilers, problems of slagging, corrosion,
flame instability and impingement, increased carbon in the fly-ash, the actual
particulate loadings and potential decreases in boiler efficiency which may
result from the modified combustion operations need to be assessed in sustained
test runs.
The purpose of Esso's present field testing program, sponsored by
EPA under Contract No. 68-02-0227, is to obtain the necessary data on the
application of combustion modification techniques to coal fired utility boilers,
in cooperation with boiler operators and manufacturers coordinated by EPA.
Major U.S. utility boiler manufacturers (Babcock and Wilcox, Combustion
Engineering, Foster Wheeler, and Riley-Stoker) have recommended boilers
characteristic of their current design practices. They have provided their
help in making arrangements for testing with the cooperating boiler
owner/operators, and in a number of cases assigned representatives to
participate in Esso's field tests.
In addition to the continuous monitoring instrumentation described
above, four EPA-type particulate sampling trains have been added to Esso's
system. These trains and other equipment are transported to the testing site
in an auxiliary van.
218
-------
The approach used for field testing coal-fired boilers is first,
to define the optimum operating conditions for NOX emission control in
short-term, statistically design test programs, without apparent unfavorable
side effects. Second, the boiler is operated for 1-3 days under the "low NOX"
determined during the optimization phase, for assessing boiler operability
problems. Finally, where possible, sustained 300-hour runs are made under
both baseline and modified combustion ("low NOX") operating conditions.
During this period, air-cooled carbon steel coupons are exposed on corrosion
probes in the vicinity of furnace water tubes, to determine through accelerated
corrosion tests whether operating the boiler under the reducing conditions
associated with staged firing results in increased fire-side water tube
corrosion rates. Particulate samples are obtained under both base-line and
"low NOX" conditions, and engineering information on boiler operability, e.g.,
on slagging problems, data boiler performance are also obtained.
So far, two front-wall, two horizontally opposed, three tangential,
and one turbo-furnace coal fired boilers have been tested in the present study.
The results obtained on these coal-fired boilers are discussed in this paper.
219
-------
2. TEST PROGRAM APPROACH
This section of the present paper discusses the approaches used
for representative boiler selection (in conjunction with EPA and utility
boiler manufacturers); the various phases of the statistical test program
designs; and the test methods employed. Methods of gaseous emission testing
were quite similar to those used in Esso's "systematic field study" vi~JL'•
In addition, particulate loadings of the flue gas stream, and the carbon
content of the particulates were determined, and corrosion probing tests
were conducted.
2.1 Test Program Design
This cooperative program of field testing utility boilers was
conducted by Esso Research with the cooperation of utility boiler manufacturers
and operators under the coordination of EPA. The proper selection of boilers
representing current design practices for this program was the result of a
cooperative planning effort. Esso Research developed a comprehensive list
of selection criteria (see Appendix), to assist EPA and boiler manufacturers
in preparing a list of potential boiler candidates. Each boiler manufacturer
submitted a list of suggested boilers to EPA for review and screening. After
consideration of such factors as design variables, operating flexibility,
fuel type, geographic location and logistics, a tentative list of boilers
was selected by EPA and Esso. Field meetings were then held at power stations
to confirm the validity of the boilers selected and to obtain necessary boiler
operating and design data.
The field meetings were attended by representatives of EPA, Esso
Research, boiler manufacturers and utility boiler operating management.
EPA described the background and need for the program and how it fits into
the overall EPA program. Esso Research presented a broad summary of our
previous findings, an outline of the three-phase program to be run at each
boiler, and led the discussion aimed at developing the information necessary
to construct a detailed program plan. These discussions produced an agreed
upon list of combustion operating variables, the specific levels to be tested,
estimated ease and length of time to change from one level to another, how
the variables were interrelated, and what operating limitations or restrictions
might be encountered. In addition, the proper number and specific location of
sampling ports for gaseous, particulate, and corrosion probe insertion were
also agreed upon. Tentative testing dates were scheduled with provision
made for possible segregation of coal types, scheduling of pretest boiler
inspection, calibration of measuring instruments and controls, scheduled
maintenance, and other preparatory steps.
The up-to-date, comprehensive information obtained in these field
meetings provided the necessary data for Esso to develop detailed, run-by-run
test program plans for review by all interested parties. Each test program,
tailored to take full advantage of the particular combustion control flexibility
of each boiler, was comprised of three phases: (1) short test-period runs
to determine NOX emission reduction capability of the boiler; (2) a 1-3 day
sustained "low NOX" run to determine if slagging or other operating problems
exist, and (3) 100-hour sustained "low N0x" and normal operation runs, to
determine quantitative measures of accelerated furnace side wall corrosion
rates.
220
-------
Statistical principles (as explained in more detail in our
"Systematic Field Study" (4) provide paractical guidance in planning test
programs, i.e., how many, and which test runs to conduct, as well as the
proper order in which they should be run. These procedures allow valid
conclusions to be drawn from analysis of data on only a small fraction of
the total possible number of different test runs that could have been made.
Table 1 will be used to illustrate briefly these principles applied to a
front-wall fired boiler, TVA's Widows Creek Boiler No. 6. (Tangentially
fired boilers present a more complex problem in experimental planning, since
there are additional operating variables such as burner tilt, and secondary
air register settings, that should be included in the experimental design.
However, the same statistical principles apply,) There are four operating
variables: (1) load, (2) excess air load, (3) secondary air register settings,
and (4) burner firing pattern. Assuming three levels of each of the first
three variables and eight different firing patterns available at each load,
there are 216 different operating modes. However, only the 33 test runs
shown, i.e., 15% of the potential maximum, provided the required information
on this boiler to define practical "low NOX" operating conditons.
Test run 10 operating conditions were chosen for the second phase
of the experimental program, while test run 26 operating conditions are
recommended for "low NOX" operation under reduced load conditions. Test
run 10 conditions can be selected with considerable confidence, since examination
of the data indicates that each of the So firing pattern runs produced lower
NOjj levels than the corresponding 82 firing pattern. The effects of day-to-day
variables, such as coal type variability, etc. not under study were balanced
between the two firing patterns, since runs 5,6, 7 and 8 were run on one day,
and 9, 10, 11 and 12 were run on another day. It should also be noted that
each day's runs completed a one-half replicate of the complete factorial
accomplished by two days of testing. Thus, the main effects of each factor
and interactions between factors can be estimated independently of each other,
with maximum precision. Repeat test runs under test run 10 conditions, during
a two-day sustained period, were used to validate these results and to obtain
an independent estimate of experimental error.
The same principles were applied to planning the 16 runs (13 through 28)
Four test runs were completed each day, one run on each of the four staged
firing patterns, and one run on each of the four excess air and secondary
air register setting combinations. Thus, the conclusion that the S^ firing
pattern is the best of those tested rests on the combined results of 16 runs
over a total of four days, and is generally true over all excess air and
secondary air register settings. Note also that during staged firing, low
excess air with secondary air registers set at 20% open, always gave the
lowest NQjj levels over each of the four days of testing at 110 MW, and also,
over the two days of testing at 125 MW. Because of the consistency of results
such as these, the number of test runs required on similar type boilers tested
later could be substantially reduced by judicious selection of run conditions
for validation purposes.
Table 2 contains a summary of the eight coal fired boilers tested
to date. Four are wall fired (two front-wall and two horizontally opposed);
three are tangentially fired; and one is turbo-furnace fired. Boilers 1, 3
and 4 are Babcock and Wilcox units, boilers 5, 6 and 7 are Combustion Engineering
boilers, while Crist No. 6 is a Foster Wheeler boiler, and Big Bend No. 2 was
designed and constructed by Riley-Stoker. Full load ratings, the number of
burners, and number of burner levels are shown for each boiler in Table 2,
as well as the number of combustion operating test variables, and the number
of test runs completed on each of these boilers.
221
-------
TABLE 1
TEST PROGRAM DESIGN FOR WIDOWS CREEK NO. 6 UNIT
(Run No., Average % 00 and Average ppm NO Emissions (0% 0 , Dry))
£. X fc
\. 2nd
v . >s. Air
Firings.
Pattern\.
Si - 16 Coal
0 Air Only
S2 - 14 Coal
DI 04 Air
83 - 14 Coal
AI A^ Air
84 - 12 Coal
AI A2 A3 A4
S5 - 12 Coal
AI A4 B2 B3
S6 - 12 Coal
Al A4 Bl B4
87 - 12 Coal
AI A4 DI D4
Sg - 12 Coal
BI B2 83 84
L - Full Load (125 MW)
A - Normal Air
20%
Open
(3) 2.8%
706
(11) 3.8%
724
(7) 4.5%
645
60%
Open
(1) 3.2%
693
(5) 4.0%
639
(9) 4.1%
622
A - Low Air
20%
Open
(4) 1.9%
581
(6) 2.0%
451
(10)* „
-L • / /o
397
60%
Open
(2) "2.0%
593
(12) 1.5%
498
(8) 2.7%
458
/
L£ - Reduced Load (110 MW)
AI - Normal Air
20%
Open
(31) 4.9%
794
(24> 4.5%
465
(27) 4.9%
579
(15) 5.2%
549
(18) 4.3%
488
60%
Open
(29) 4.8%
734
(13) 4.5%
537
(17) 4.4%
560
(21) 6.1%
641
(25) 4.5%
577
A - Low Air
20%
Open
(32) 2.8%
541
(26)**
*• ; 2.7%
346
(22) 3.4%
357
(19) 3.1%
351
(16) 3.0%
384
60%
Open
(30) 2.7*
525
(20) 3.0^
402
(14) .
l.bh
399
(28) 4.5Z
511
(23) q,
J • 7/0
511
(20A) ,
£. • £,m
454
* "Low NOX" condition
selected for sustained
run.
** "Low NOx" condition
with further load
reduction.
Pulverizer
A - Top Row
B - 2nd Row
C - 3rd Row
D - Bot. Row
Burner No.
1
0
0
0
0
234
000
000
000
000
222
-------
TABLE 2
SUMMARY OF COAL FIRED BOILERS TESTED
OJ
* FURNACE DIVISION WALL
** TWIN FURNACE
— — — — ___^___^_^^^^____
STATION AND
BOILER NO.
1. WIDOWS CREEK 6
2 . CRIST 6
3. HARLLEE BRANCH 3
4. FOUR CORNERS 4
j
i
| 5. NAUGHTON 3
|
| 6. BARRY 4
|
! 7. BARRY 3
8. BIG BEND 2
j
TYPE OF
FIRING
FRONT WALL*
FRONT WALL
HOR. OPPOSED
HOR. OPPOSED*
TANGENTIAL
TANGENTIAL
TANGENTIAL**
TURBO
FULL
LOAD (MW)
125
320
480
800
330
350
250
350
NO. OF
BURNERS
16
16
40
54
20
20
48
24
NO. OF
LEVELS
4
4
4
6
5
5
6
x
TEST
VARIABLES
4
4
4
5
6
!
7
4
* i
NO. OF
RUNS
43
22
45
26
26
35
8 j
14 i
219 1
$
-------
2.2 Test Methods
In this section the gaseous and particulate sampling and analytical
methods are described. Furnace corrosion probing techniques and equipment
used are also discussed.
2.2.1 Gaseous Sampling and Analysis
The objective of obtaining reliable gaseous emission data in field
testing boilers requires a sophisticated sampling system. The sampling and
analytical system used in this program has already been described in detail
in the Esso Research and Engineering Company Report, "Systematic Field Study
of NOX Emission Control Methods for Utility Boilers" (4).
For the present study, further capabilities were added to the
analytical instrument train by installing a Thermo-Electron chemiluminescent
analyzer to provide measurements of NO and NOX in addition to those obtained
with the Beckman NO and N0£ spectroscopic monitors. Figure 1 is a
schematic diagram of the present configuration of Esso's sampling and
analytical system.
Since samples are taken from zones of "equal areas" in the flue
gas ducts, gas sampling probes are "tailor-made" for each individual boiler
tested. Three stainless steel sampling tubes (short, medium, and long) are
fabricated on the job site, and installed in quick-disconnect mounting probe
assemblies, along with a thermocouple located at the mid-point of the duct
for gas temperature measurement. At least two probes of this type are installed
in each flue gas duct, or a minimum of four are used when there is only one
large flue duct on the boiler. Thus, a minimum of 6 sample points per duct,
or 12 per boiler are provided, thus assuring representative gas samples.
All connections between the Esso Analytical Van and the probes are of the
quick-disconnect type for ease of assembly and assurance of leak-proof joints.
In running field tests, the gas samples are withdrawn from the
boiler under vacuum through the stainless steel probes to heated paper
filters where the particulate matter is removed. These paper filters are
maintained at 300-500°F. The gases then pass through rotameters, which are
followed by a packed glass wool column for SOg removal. Initially, gas
temperatures are kept as high as possible to minimize condensation in the
particulate filters. After leaving the packed column at 250-300°F, the
gas samples pass at temperatures above the dew-point through heated Teflon
lines to the vacuum/pressure pumps. The sample is then refrigerated to a
35°F dew-point before being sent to the van for analysis. Usually, the van
is located 100 to 200 feet from this point and the gas stream flows through
Teflon lines throughout this distance.
As in our previous studies (4-6), our analytical van was equipped
with Beckman non-dispersive infrared analyzers to measure NO, CO, C02 and S02»
a non-dispersive ultraviolet analyzer for N0£ measurement, a polarographic 02
analyzer and a flame ionization detector for hydrocarbon analysis. The
Thermo-Electron chemiluminescent instrument, as indicated above, was added
to provide improved capabilities for NO and NOX measurements. The measuring
ranges of these continuous monitors are listed in Table 3.
224
-------
PROBE (4 EACH)
ESSO RESEARCH TRANSPORTABLE SAMPLING
AND ANALYTICAL SYSTEM
THERMOCOUPLE
BOILER
DUCT
800°F
PITOT TUBE
PARTICULATE FILTERS (HEATED)
ROTAMETERS
HEATED LINES
CO
CO,,
NO
SO,
CU
HYDROCARBONS
NO & NO
x
REFRIGERATOR
•fr
SAMPLING
VAN
J
SOLENOID
VALVE
if
if
VENT
5 PSI RELIEF VALVE
-------
TABLE 3
CONTINUOUS ANALYTICAL
INSTRUMENTS IN ESSO VAN
Beckman
Instruments
NO
Technique
°2
co2
CO
so2
Hydrocarbons
Thermo Electron
NO/NO
Non-dispersive Infrared
Non-dispersive ultraviolet
Polarographic
Non-dispersive infrared
Non-dispersive infrared
Non^dispersive infrared
Flame ionization detection
Chemiluminescent
Measuring
Range
0-400 ppm
0-2000 ppm
0-100 ppm
0-400 ppm
0-5%
0-25%
0-20%
0-200 ppm
0-1000 ppm
0-23,600 ppm
0-600 ppm
0-3000 ppm
0-10 ppm
0-100 ppm
0-1000 ppm
0-2.5 ppm
0-10.0 ppm
0-25 ppm
0-100 ppm
0-250 ppm
0-1000 ppm
0-2500 ppm
0-10,000 ppm
226
-------
A complete range of calibration gas cylinders in appropriate
concentrations with N2 carrier gas for each analyzer are installed in the
system. Instruments are calibrated daily before each test, and in-between
tests, if necessary, assuring reliable, accurate analyses.
Boiler flue gas samples are pumped continuously to the analytical
van through four composite probes. While one sample is being analyzed, the other
three are being vented. Switching to a new sample requires only the flushing
of a very short section of sample line before reliable readings may be obtained.
Four duplicate sets of analyses from each probe can be obtained in less than
32 minutes, thus speeding up the task of obtaining reliable gaseous emissions,
and/or avoiding the need to hold the boiler too long at steady state conditions.
The validity of using the Thermo-Electron chemiluminescent NO/NOX
analyzer as the primary NOX monitoring instrument was checked during the
first series of tests conducted in this program, on TVA's Widows Creek Boiler
No. 6. As shown in Figure 2, the NOX data measured with the chemi luminescent
analyzer were correlated with the sum of NO plus NC>2 data measured with the
Beckman non-dispersive infrared NO and non-dispersive ultraviolet N02
instruments. As seen from the regression in Figure 2, excellent agreement
was obtained between the chemiluminescent and spectroscopic instrumental
methods. Thus, the chemiluminescent monitor was validated against the
spectroscopic instruments, which in turn had been validated against a variety
of other techniques, including the wet chemical phenoldisulfonic acid method,
in previous Esso field studies (4-6).
Our instrumental measurement technique for flue gas Q£ and C02
determinations were checked by comparing the measured 02 vs. C02 relationship
to that calculated from fuel analysis, firing rate, and known excess air level.
In our previous studies (4-6) we validated the instrumental 02 and C02
measurements against Orsat analyses of grab samples.
The comparison of measured to calculated 02 vs. C02 relationships
is shown in Figure 3, based on data obtained in testing TVA's. Widows, Creek
No. 6 Boiler.
As can be seen from Figure 3, the agreement between the regressions
based on measurements and calculations is very good over the range of actual
measurements.
2.2.2 Particulate Sampling
Modifications in the combustion process to minimize NOX emissions
tend to result in slower, less intense combustion conditions. Lowering excess
air increases flame temperature which: aids combustion, but limits the amount
of oxygen available for the combustion process, directionally increasing
the probability of burnout problems. Similarly, staging of burners,
where some burners are operated at sub-stoichiometric conditions, and the
remaining burners (or ports) are used as "air-ports" to complete combustion
of the fuel, drastically limits available oxygen in the initial combustion
phase, lengthens out flames and, with the slower, less intimate mixing of
air and fuel, potentially increases unburned combustibles. Therefore, it
was necessary to consider that combustion modifications implemented to
minimize NOX emissions could potentially increase particulate emissions
from pulverized coal-fired boilers.
227
-------
FIGURE 2
ro
ro
oo
NO REGRESSION - BECKMAN NO + NO,, VS CHEMILUMINESCENCE NO MEASUREMENTS
700
600
ii
w 500
3
CN
O
* 400
o
S5
u 300
H
PQ
O
g 200
P-I
P-I
100
A ^ ^
' ' ' ~/?
- QO Oo/
nO
— rv
ocSr
o cf
0%'
O .^^
c^Xo
yT O
XQ
0^0
— >n
^ ° y = 0.42 +
/ r = 0.985
y Sy(est) =
1 ' 1
—
~™
~
-
1.0172 X
-
L9 ppm NO
X
/ALL READINGS EXPRESSED AS
PPM NOX, CORRECTED TO 3%
^ 02, DRY BASIS.
' /
/ . 1 . 1 i 1 . 1 i 1 i 1 .
—
1 . 1
0 100 200 300 400 500 600 700 800
-------
FIGURE 3
RELATIONSHIP BETWEEN % C00 AND % 00 FLUE GAS MEASUREMENTS
fa ' ^
(WIDOWS CREEK BOILER NO. 6 - RUN IB)
18
16
CALCULATED FROM
/COAL ANALYSIS
(Y = 18.5-0.
14
CALCULATED FROM FLUE GAS ANALYSIS
(Y = 18.4-0.95% 02)
O
I
0 IN FLUE GAS (DRY BASIS)
229
-------
In view of the above, an important phase of our field test program
on coal fired boilers was directed at particulate emissions. The objective
of this effort was to obtain sufficient dust loading information to determine
the potential adverse side effects of "low NOX" combustion modifications
on particulate emissions, with respect to total quantities and per cent
unburned carbon, vs. similar data obtained under normal or baseline operating
conditions. Specifically, such data are needed to evaluate the changes,
if any, that might occur in total dry filterable solids passing through the
boilers, and on unburned combustibles in the fly-ash resulting from "low NOX"
emission modifications. Other information, such as changes in particle size
distribution or in electrical conductivity which could affect electrostatic
precipitator collection efficiency, would also be of interest, but this was
beyond the scope of our program.
Four Research Appliance Company EPA-type particulate sampling
trains, including four sample boxes, probes, and two isokinetic pumping sets
were used in obtaining dust loading data on six pulverized coal fired utility
boilers to date. The names of the utilities and details of the boilers tested,
including size, type of firing, numbers of burners, etc. are given in Table 2.
Except for tests at Utah Power & Light Company's Naughton Station, Boiler No. 3,
all dust loading data were obtained in the flues at convenient locations
downstream of the air-heaters. At the Naughton Station testing was done
ahead of the air-heaters, due to inaccessibility of locations downstream of
the air heaters. Also, on Alabama Power Company's Boiler No. 4 at Barry
Station, testing was carried out downstream of the precipitator (with the
precipitator shut-off), immediately before entering the stack. In all cases
two traverses were made in each flue with one probe assembly, in accordance
with prescribed procedures. However, strict adherence to EPA-recommended
test methods was not always possible due to the limited availability of sample
port locations, interferences with building and boiler appurtenances, and
the limited time and manpower available for these tests. However, it should
be remembered that the objective of these tests was to determine relative
changes between normal and modified firing operations, not to measure accurate
dust loadings.
2.2.3 Furnace Corrosion Probe Testing
Pulverized coal fired boilers do, on occasion, experience wastage
of the furnace wall tubes (corrosion). Normally, this type of corrosion is
experienced in areas of localized reducing atmospheres adjacent to the
midpoint of furnace sidewalls near burner elevations where flame impingement
could occur. Remedies have been to increase the excess air level so that
an oxidizing atmosphere prevails at these locations, and to increase the
fineness of pulverization, so that complete oxidation of the pyrites in the
coal is accomplished before these species have a chance to reach the furnace
wall tubes. For new boilers, another remedy has been to increase distances
between the burners and the sidewalls, to minimize potential impingement.
Several mechanisms have been postulated for this type of corrosion which
appears to be associated with the formation of pyrosulfates from the coal
ash (at 600-900°F) and iron sulfide, or S03 from the pyrites.
230
-------
Combustion modifications for NOX emission control are generally
most effective at low excess air or substoichiometric air conditions in
the flame zone, i.e., at conditions that may contribute to furnace tube
wall corrosion. Our prior field tests of coal-fired boilers have been of
short duration, allowing no time to assess such side-effects. However,
it has been recognized that the effects of modified firing operations on
furnace tube wall corrosion need to be evaluated (4_-6). Discussions with
boiler manufacturers and operators indicated that this potential problem
was one of their greatest concerns. It was also evident that acclerated
corrosion probe tests would be necessary to establish that "low NOX"
operation with coal could be carried out, since there was a general
unwillingness to operate on a long-term basis using the boiler as a test
medium.
Accordingly, one of the important aspects of our field test
program was to design and construct controllable corrosion probes, and to
define the extent of the potential corrosion problem. The objective of
our furnace corrosion probing runs was to obtain "measurable" corrosion
data on potential side effects of "low NOX" firing conditions on furnace
wall tubes. We received excellent advice and help in evaluating the problem
and defining our approach to corrosion studies by Combustion Engineering
Company research representatives, and by Esso's corrosion experts.
Our approach to obtaining data was to expose corrosion probes
inserted into available openings located at the "vulnerable" areas of the
furnace (see Figure 4), under both baseline and "low NOX" firing conditions.
Based on prior corrosion testing, it was concluded that exposure for
approximately 300 hours at elevated coupon metal temperatures (above normal
furnace tube metal temperatures) to accelerate corrosion, would produce
"measurable" corrosion on SA-192 carbon steel (furnace tube type) coupon
material. Since our objective was to show relative differences in corrosion,
if any, between baseline and "low NOX" firing, exposure temperatures at both
conditions were set at approximately 875°F; i.e., high enough to accelerate
corrosion, and just below the 900°F limit above which pyrosulfates apparently
are not formed.
Figure 5 and 6 show details of the corrosion probes, based
on Combustion Engineering's design. Essentially, this design consists of
a "pipe within a pipe", where the cooling air (plant air supply) is admitted
to the ring-coupons exposed to furnace atmospheres at one end of the probe
through a %-inch stainless steel tube roughly centered inside of the coupons.
The amount of cooling air is automatically controlled to maintain the desired
set-point temperature (875°F) on the coupons. The cooling air supply tube
is axially adjustable with respect to the corrosion coupons, so that temperatures
of both coupons may be balanced. To simplify the presentation, thermocouples
installed in each coupon are not shown in Figures 5 and 6. Normally,
one thermocouple is used for control, and the other one for recording. The
cooling air travels backwards along the 2%-inch extension pipe and discharge
outside of the furnace. Thus, there is no interference with the cooling air
and the furnace atmosphere at the coupon location.
231
-------
FIGURE 4
GEORGIA POWER
HARLLEE BRANCH STATION
BOILERS NO. 3&4
FURNACE CORROSION
PROBE LOCATIONS
UTAH POWER AND LIGHT COMPANY
/
F.W.
BURNE
RS
SI
SLAG BLC
BLOWER NO,
NO. 9 1
.\ I
/ I
PROBE
3B PROBE!
3 A, 4 A, 41
V
AC
)WE
3
~1
B
RS
&8
SLAG
• BLOWER INSPECTION
= 8' TOP
i x TJTTDKTT7T?
X ELEV.
1 TOP BURNER
ELEV.
V
R.W.
URNERS
—
-PROBE PROBE
NO. 2 NO. 1
\ /
-D Q D 00
INSP.
DOORS .
PROBE PROBE
NO. 4 NO. 3
^••K
SIDE ELEV.
FRONT ELEVATION
(CORNER BURNERS)
SLAG
BLOWER
ARIZONA PUBLIC SERVICE COMPANY
FOUR CORNERS STATION - BLRS. NO. 4&5
SLAG .BLOWERS
TOP
BURNER-
ELEV.
BURNERS
A
- • —
-V
PROBE
LOCATIONS
(BOTH SIDES)
v
\
BURNERS
SIDE ELEV.
ALABAMA POWER COMPANY
BARRY STATION - BOILER HO. 4
SLAG BLOWERS
N0.18&26
LOWER
BURNER
ELEV.
SLAG
ELEV.
I
11'
t
r
t
PROBES NOS. 3
PROBES I
• - t
_ j
IOS. 1
»-
232
SIDE ELEV.
(CORNER BURNERS)
-------
FIGURE 5
CORROSION PROBE
DETAIL OF 2%" IPS EXTENSION PIPE AND END PLATE
(OUTSIDE OF BOILER)
DRILLED AND TAPPED FOR 1/8" IPT THREAD.
(SWAGELOCK FITTINGS - FOR THERMOCOUPLES)
13
DRILLED
ACCEPT
" SS AIR
SUPPLY
TUBING
HOLE FOR 1/4" SS
GAS SAMPLING TUBE
d
END PLATE
END PLATE
WELD
2%" I.P.S. PIPE
EXTENSION
(
(
A
u
^
H~
4:
^^^^
^
i4
=HJ
A
/
X^
^
XX
'//
'//
//,
%
P^$J$$JJ^^
,1/16" THERMOCOUPLES (2)
X /
T)
AIR SUPPLY y
** (%" SS TUBING) \
- )
1/4" GAS SAMPLING TUBING (SsT
SEAL
SWAGELOCK FITTING DRILLED FOR %" SS AIR SUPPLY
TUBE (THREADS CUT OFF AND FITTING WELDED OR SILVER
SOLDERED TO END PLATE)
WELD
AIR DISCHARGE
1-1/4" COUPLING
-------
FIGURE 6
CORROSION PROBE
DETAIL OF CORROSION COUPON ASSEMBLY
(INSIDE OF FURNACE)
2%" PIPE EXTENSION
\
NJ
OJ
1/4"
CORROSION
COUPONS
S.S. COOLING AIR SUPPLY TUBE
S.S. GAS SAMPLING
TUBE
THERMOCOUPLE SOCKETS
1
1
^
?
1-1/4'I »
-5/8*V
END CAP
FACE OF FURNACE WALL TUBES
-------
Sustained, 300-hour corrosion probe tests were run on boilers of
four utility companies, as shown in Table 4 below:
TABLE 4
SUMMARY OF CORROSION PROBING TESTS
Utility
Station
Georgia Power Co.
Utah Power & Light Co.
Arizona Public Service Co.
Alabama Power Co.
Harllee Branch
Naughton
Four Corners
Barry
Boiler Number
Base "Low NO "
x—
4 3
3
5 4
4 4
Type of Firing
Horizontally Opposed
Tangential
Horizontally Opposed
Tangential
235
-------
3. FIELD TEST RESULTS
The field test results obtained on individual boilers under diverse
operating conditions are presented in three sections. These sections consist
of gaseous emission measurements, flue gas particulate loading measured upstream
of particulate collector equipment, and corrosion probing data obtained in
accelerated furnace fire-side water-tube corrosion tests. Gaseous emission
and most of the particulate emission data were obtained under normal, as well
as staged firing conditions. As discussed before, particulate loadings of
the flue gas were determined only under conditions corresponding to baseline
and "low NOX" operation, for purposes of comparison on the relative effect of
staged firing patterns on flue gas particulate loadings in coal combustion.
Similar considerations apply to the sustained, 300-hour corrosion tests, which
had as their objective the determination whether staged firing of coal accelerates
furnace water tube corrosion rates.
The gaseous emission data obtained under baseline and staged firing
conditions, at various load levels, are presented first. Throughout this paper,
pollutant concentrations are expressed as ppm, adjusted to zero per cent 02 in
the flue gas, on a dry basis.
3.1 Gaseous Emission Results for
Individual Coal-Fired Boilers
The data obtained are grouped according to boiler design type,
i.e., wall-fired (front wall or horizontally opposed), tantentially fired,
and turbo-furnace boilers.
3.1.1 Widows Creek Boiler No. 6
The Tennessee Valley Authority's Boiler No. 6 at their Widows Creek
Station was the first boiler tested in our present study. Thirty-two short-term
test runs were made in a statistically design optimization program, to minimize
NOX emissions. These tests were followed by two sustained runs, one at full load,
the other one at reduced load, with the optimum staging patterns. The sustained
corrosion probing run was deferred at TVA's request, until high sulfur coal
could be fired, and other data would be available to show that staged firing
would not cause abnormally high furnace corrosion rates.
Widows Creek Unit No. 6 is a 125 MW, 16-burner, front-wall,
pulverized coal fired Babcock and Wilcox boiler. It has a single dry-bottom
furnace with a division wall, and the 16 burners are arranged with four burners
in each of four rows. Each row is fed with coal by a separate pulverizer.
The statistical test design shown in Table 1 for this boiler has
been discussed earlier. The NOX emission data, expressed as ppm NOX corrected
to zero per cent 02 in the flue gas (dry basis) obtained with the various firing
patterns tested are presented in Figures 7 and 8. In Figure 7, the measured
emissions are plotted vs. per cent of stoichiometric air to the active burners.
Figure 8 shows the same emission data, but plotted as a function of the overall
per cent stoichiometric air. Corresponding to each firing pattern (designated
by "S"), least squares regression lines have been fitted to the data points.
236
-------
FIGURE 7
PPM NOX (0% 02, DRY) VS % STOICHIOMETRIC AIR
TO ACTIVE BURNERS
(WIDOWS CREEK - NO. 6 BOILER)
900
800
700
CO
H
3
o
&•?
o
o
a
600
500
400
300
200
S4_7 (80-110 MW)
I
(110 MW)
/ A *
/ A
/ s\
D •/ /A S2-3 (125 m
a*
x°°
/
/
f
FIRING
PATTERN:
SYMBOL (ACTIVE /AIR)
0 S1 (16/0)
• S^^ (16/0)
A S2_3 (14/2)
D S4_? (12/4)
1 1 1 1 1
GROSS
LOAD
(MW)
125
110
120-125
80-112
-
-
I
80
90
100
110
120
130
140
% STOICHIOMETRIC AIR TO ACTIVE BURNERS
237
-------
FIGURE 8
PPM NOX (0% 02, DRY) VS OVERALL
STOICHIOMETRIC AIR
(WIDOWS CREEK - NO. 6 BOILER)
900
800 -
700 -
(=>
(N
O
6-S
O
O
a
PH
PLI
600 -
500 -
400 -
300,
'80
90
100
110
120
130
140
% OVERALL STOICHIOMETRIC AIR
238
-------
The data show the strong influence of reduced oxygen supply on
decreasing NOX emissions. This effect is further enhanced by staged firing
patterns, as shown in Figures 7 and 8.
Low excess air operations consistently reduced NOX emission levels.
Average reductions were 23% under normal firing, 34% at full load staged
operation and 27% at reduced load staged operation. Staged firing at full
load resulted in an average of 14% NOX reduction at full load, and an average
of 27% at reduced load. The lowest practical level of excess air was dictated
by acceptable CO emissions and stack appearance. The combination of low
excess air and staged firing reduced NOX emissions by 40% at full load, and
from 33% to 50% at reduced load. The optimum combination of operating
variables reduced NOX by 47% at full load and by 54% at reduced load.
Of the firing patterns tested, operating the top row of burners on
air only, with low overall excess air, gave the highest reductions in NOX.
Operating with the top wing burners on air only resulted in slightly lower
NOX emissions than with the bottom wing burners on air only.
Opening or closing down the secondary air registers was found to
have small, but statistically significant effects on NOX emission levels.
The data shown in Figure 7 call attention to an apparent anomaly.
A cursory inspection of the data would indicate, that while as expected, NOX
levels decrease with decreasing air supply to the active burners, staging the
burners could result in higher NOX emissions than normal operation at the
same burner air/fuel air ratio. The true interpretation of these data is that
for staged burner configurations, it is generally necessary to operate at a higher
overall level of air supply than for normal firing. Thus, for a given air/fuel
ratio to the burners, the overall level of excess air is higher in staged firing
than in normal operation. Furthermore, the desirable effect of heat removal
between "first" and "second" stages is not as effective in staging the burners
as when special "NO-ports" or "over-fire" air ports are available for secondary
air admission. Thus, it is reasonable to expect that in new installations,
designed for staged combustion, even higher NOX reductions may be accomplished
than in these tests which deliberately "de-activated" burners to achieve staging.
The foregoing remarks on comparing NOX emissions with staging to
those with normal firing, as a function of the per cent stoichiometric air
to the active burners, are not unique to Widows Creek No. 6, but to all boilers
tested, as will be shown later.
3.1.2 Crist No. 6 Boiler
Gulf Power Company's Boiler No. 6 at their Crist Station is also
a front-wall fired unit. This Foster Wheeler boiler has a single furnace,
with a maximum continuous rating of 345 MW gross load. Its 16 burners are
arranged in four rows of four burners each.
A cooperative test program by Gulf Power, Foster Wheeler and Esso,
coordinated by EPA, was planned for this unit. Plans included short-term
firing pattern optimization runs for minimizing NOX emissions, accompanied
by boiler performance tests by Foster Wheeler, followed by boiler operability
check-out at "low NOX", than a sustained 300-hour test under "low NOX" and
baseline operating conditions for assessing corrosion problems, and an optional
long-term test period of about 6 months for determining actual furnace water
tube wastage.
239
-------
Because of load demands on this boiler, to date it has been possible
only to explore firing patterns in short-term runs, without performance tests,
for minimizing NOX from this boiler.
As shown by the data in Figure 9, the general trend of decreasing
NOX emissions with reduction of per cent stoichiometric air followed trends
similar to those observed in the Widows Creek Boiler No. 6 tests. Again, NOX
levels decrease sharply with decreasing the % stoichiometric air to the active
burners, including normal firing patterns.
The data in Figure 9 are subdivided into the "A" and "B" sides
of the boiler. The flue gas stream leaving the furnace is split into these two
ducting faths, and although the boiler operator and manufacturer could at times
achieve 02 balance in the two sides, the NOX levels measured were clearly
higher for the A side than the B side, with all firing patterns tested. The
reason for this behavior is unexplained at present, although it may be related
to differences in air flow, and uncertainties of the accuracy of air damper
settings on the two sides of the boiler.
To simplify the presentation, Figure 9 shows only the least square
regression lines fitted to the data points. Si denotes normal firing, while
S2> S3> and S4 are staged firing patterns, where 82 denotes top row using
burners on air only, 83, top row middle burners on air only, and 84, all top
row burners on air only. Combining all data points for the A and B sides of
the boiler, respectively, results in the two similar, although not quite
parallel least squares regressions shown in heavy lines, indicating similar
trends.
As expected, the 83 pattern results in somewhat lower NOX emission
levels than the 82 pattern, because of the larger spacing between active
burners in the former configuration. The most effective combustion modification
is, of course, operating the top row burners on air only (84) with low overall
excess air, but for this boiler, such a firing pattern entails a load reduction
of about 15%.
It is hoped that eventually an opportunity may arise for completing
the planned program on this unit.
3.1.3 Harlee Branch Boiler No. 3
Georgia Power Company's No. 3 Boiler at their Harlee Branch Station
was tested through all three phases of our test program procedure. In this
section of the present paper, the gaseous emission results are summarized while
particulate measurements and corrosion probing results will be presented later.
Harlee Branch unit No. 3, with a maximum rated capacity of 480 MW
gross load, is a single furnace, pulverized coal fired Babcock and Wilcox boiler.
It has 40 burners arranged in twenty burner cells of two burners each, with two
rows of five burner cells located in both the front and rear walls of the furnace.
The burner configuration and pulverizer layout are shown in Figure 10 for this
boiler.
240
-------
FIGURE 9
1200
1100 H
1000 h
H
CO
o
6*!
O
g
g
PJ
CM
PPM NOX (0% 02, DRY) VS % STOICHIOMETRIC
AIR TO ACTIVE BURNERS
(CRIST NO. 6 BOILER)
I T
S -S. - "A" DUCT
1 4
s s - "B" DUCT
1 4
GROSS
FIRING PATTERN LOAD
(ACTIVE/AIR) MW
(16/0)
(14/2)
(14/2)
(12/4)
320-270
320
320
270
100 110 120 130
% STOICHIOMETRIC AIR TO ACTIVE BURNERS
140
241
-------
FIGURE 10
HARLLEE BRANCH NO. 3 BOILER
PULVERIZER AND COAL PIPE LAYOUT
TWO-BURNER CELL
PULVERIZER LETTER
BURNER NUMBER
©
©
D
3
^-^
J
,•*->
4
©
B
^—«
4
©
©
4
*+-~*
E
f—•
3
4
—/
G
^—•«
3
4
1^—
C
<-
,3
2
>^_
A
K
©
2
.—'
D
•—v
3
3
* '
K
f—•.
4
A
T
FACING REAR FACE
H
©
©
©
4
*>—.
F
FACING FRONT FACE
[4
H
(3*
242
-------
Because of the arrangement of pulverizer mills for this boiler, it
was possible to shut off the coal supply through individual pipes, and therefore,
there was added flexibility for exploring staging patterns during the short-term
testing phase.
Baseline NOX emission levels averaged about 870 ppm. Lowering the
level of excess air was possible both under normal and staged operating conditions
down to flue gas 02 concentrations of about 1.5% or even lower, without apparent
undesirable side effects. The steep effect of reducing the per cent of
stoichiometric air to the active burners on decreasing NOX emissions is shown
by the least squares regressions of the data in Figure 11. This figure combines
the data obtained under normal firing using 40 burners, with those measured using
a large variety of burner staging patterns.
Interestingly, by operating four to six top burner cell row burners
on air only, it was possible to maintain boiler load at 480 MW, and reducing
the NOX emission levels to about 580 ppm. This level corresponds to a reduction
in NOx of about one-third, compared with the baseline level. Usually,
using burners of the top rows of front and rear walls were operated on air
only, but the NOX emission levels were not particularly sensitive to the exact
location of the inactive burners in the top row.
With only 30 active burners, i.e., 10 top row burners on air only,
it was possible to reduce NOX emissions to about 390 ppm at low levels of
excess air, or a reduction of over 50% from the baseline level. However,
load was also reduced by 17% from 480 MW to 400 MW using this staging patterns.
For the 300-hour sustained run, a full load, low excess air condition
(overall 107% stoichiometric air, with three wing burners on both front and
rear faces of the boiler on air only) was selected.
3.1.4 Four Corners Boiler No. 4
Arizona Public Service's No. 4 Boiler at their Four Corners Station
was also tested according to our test program design, except that continuous
electricity demand on the station prevented testing at low loads, and the
currently inoperative flue gas recirculation system could not be utilized.
This unit, with a maximum rated capacity of 800 MW gross load, is a single
furnace (with division wall), pulverized coal fired Babcock and Wilcox boiler.
It is fired with low sulfur, high ash Western coal. Boiler No. 5 at the
Four Corners Station is a "sister"-unit of similar size and design. The latter
was used for determining accelerated furnace water tube corrosion rates under
baseline operating conditions.
In each of these two boilers, nine pulverizers feed 54 burners,
arranged in 18 cells of three burners each, as shown in Figure 12. The front
wall has ten burner cells, while eight burner cells are located in the rear
wall of the furnace. Each boiler can maintain the full load capacity of
800 MW with eight or nine pulverizers in operation, when good quality coal
is fired, and all equipment is in good operating conditions.
243
-------
- 30 -
FIGURE 11
PPM NO (0% 02, DRY) VS % STOICHIOMETRIC
AIR TO ACTIVE BURNERS
(HARLLEE BRANCH - NO. 3 BOILER)
900
800 -
w 700 -
600 ~
6-S
O
O
a
s
500 -
400-
300
NORMAL FIRING - 480 MW
(40 BURNERS FIRING)
STAGED FIRING - 480 MW
(4 TO 6 BURNERS AIR ONLY)
STAGED FIRING - 400 MW
(10 BURNERS - AIR ONLYl
80 90 100 110
% STOICHIOMETRIC AIR TO ACTIVE BURNERS
244
-------
FIGURE 12
FOUR CORNERS STATION - BOILER NO. 4
PULVERIZER-BURNER CONFIGURATION
REAR WALL (EAST)
NORTH
WALL
42S.X"
SOUTH WALL
FRONT WALL (WEST)
9 PULVERIZERS NUMBERED 41 THROUGH 49.
18 BURNER CELLS NUMBERED WITH PULV. NO. "N" OR "S" FOR NORTH OR SOUTH OF DIVISION WALL.
54 BURNERS DESIGNATED "T", "M" OR "B" FOR TOP, MIDDLE OR BOTTOM OF EACH CELL.
E.G., 45NT IS TOP LEFT BURNER IN FRONT WALL OF NO. 4 BOILER
•TOP BURNER OF CELL
'NORTH SIDE OF DIVISION WALL
'NO. 5 PULVERIZER
'NO. 4 BOILER
245
-------
Operating variables during the short-term optimization phase of
the tests were boiler load, burner firing pattern, excess air level and
secondary air register setting. Our gaseoug sampling system was modified
to allow sampling from 18, instead of the usual 12 duct positions, with two
three-probe assembly each in the north, middle, and south ducts between the
economizer and the air heaters.
The NOX emission data measured are summarized in Figure 13. Baseline
NOX emissions under normal operating conditions averaged a high level of about
1070 ppm, which is consistent with that expected from a large, horizontally
opposed, coal-fired boiler. Reducing the per cent stoichiometric air to the
active burners sharply reduced NOX emissions for both normal and staged firing.
Staged firing with 46 active burners (eight top burners on air only)
resulted in further reductions in NOX, particularly when the per cent
stoichiometric air to the active burners was decreased below 100%. As shown
by the least squares regressions of the data in Figure 13, NOX emissions could
be reduced by about 45% to 570 ppm, with 95% of the stoichiometric air supplied
to the active burners. Even further reductions in NOX could be achieved at
the full load of 800 MW by operating 12 burners on air only, to 530 ppm, or a
reduction of 50%.
Wide open secondary air register settings could reduce NOX emissions
by a small amount compared with closed settings (presumably because of reduced
combustion intensity), but only in combination with low excess air firing.
As before, the effect of damper settings on NOX emissions was significant but
second-order with respect to the main effects of reduced excess air and staging.
As for the boilers discussed before, the plot of ppm NOX vs. per cent
stoichiometric air to the active burners shows the effect of staging. Again,
the apparent dichotomy of higher NOX emissions measured under staged firing
conditions than for baseline operation at the same levels of per cent
stoichiometric air is due to the fact that the overall excess air is higher
at the same burner air supply for staging than for normal firing.
The solid triangle data points obtained with staged firing
(46 active burners, eight burners on air) were measured while the boiler
operator used water injection to help improve precipitator efficiency for
particulate removal. The rather impressive reduction in NOX of over 100 ppm
from about 700 ppm is not altogether surprising, based on our estimate of
0.2 Ib. HO injected/lb. coal fired. This quantity of water injection should
reduce flame temperatures sufficiently to allow for the above degree in NOX
emission reduction.
3.1.5 Naughton Boiler No. 3
Utah Power and Light's No. 3 boiler at their Naughton Station was
one of two modern, 350 MW maximum rated single furnace, pulverized coal fired,
Combustion Engineering boilers tested. The other one was Alabama Power's No. 4
Boiler at their Barry Station. Both boilers have five levels of four corner
burners each. Gaseous emission results obtained in testing the latter will
be presented in the next section of this paper.
246
-------
FIGURE 13
PPM N0x (0% 02, DRY) VS % STOICHIOMETRIC
AIR TO ACTIVE BURNERS
(FOUR CORNERS NO. 4 BOILER)
1100
1000
900
800
700
600
STAGED FIRING
(12 AIR, 42 COAL)
STAGED FIRING
(8 AIR, 46 COAL)
NORMAL FIRING
_L
500,
90
100
110 120 130 140
% STOICHIOMETRIC AIR TO BURNERS
150
247
-------
Naughton unit No, 3 was designed to fire a sub-bituminous, low
heat content (9,500 Btu/lb. HHV), low sulfur, high moisture content Western
coal. The boiler was designed for a larger turbine-generator than the one
actually installed. This factor, in combination with the lack of "seasoning"
of the superheat and reheat surfaces, and the type of coal fired in this new
unit has resulted in a steam temperature control problem. The use of tilting
burners and attemperation water are the means available for controlling steam
temperatures. To the date of our test, it has been necessary at load levels
exceeding 280 MW to tilt the burners down, add attemperation water, lower
excess air and use furnace soot blowers almost continuously. It may be
necessary, according to Combustion Engineering representatives, to reduce
the reheat surface area to overcome this control problem.
Other operating problems encountered in this test program were
furnace slagging (particularly at high loads, with low excess air and tilting
burners down) even under normal operating conditions, and the high silica content
of the boiler feed-water, causing pin-hole leaks in the condenser tubing.
The above problems were taken into account for the design of the
statistical test program. Our short-term, NOX optimization phase was conducted
at less than full load levels, to avoid the limited flexibility associated
with operating problems. The six operating variables studied in the short term
optimization tests were gross boiler load, burner firing pattern, excess air
level, burner tilt, secondary air damper setting, and coal pulverizer fineness
setting. Because of the above-mentioned operating problems with this new
boiler, the 300-hour accelerated test was performed only under normal operating
conditions, as will be discussed later.
Normal classifier fineness, horizontal or down-tilt burner position,
and low coal-air register settings resulted in the lowest NOX levels. Based
on these initial findings, most of the NOX optimization test were run under
these conditions, to explore the effectiveness of the major variables, staging
and excess air, on NOX emissions.
The emission data obtained are in testing this boiler are shown by
the least square regressions of Figure 14. Significant reductions in NOX
emissions were achieved from the baseline level of about 650 ppm (which is
relatively low for a coal fired boiler of this size, but typical of tangential
fired units, from the standpoint of NOX emissions). With normal firing, quite
a steep decrease was found by reducing the per cent stoichiometric air to the
active burners to 110%, resulting in a reduction by about 30% to 450 ppm.
Staged firing in combination with low overall excess air (less than stoichiometric
air/fuel ratio in the active burners) resulted in NOX levels as low as 250 ppm,
or a reduction of about 60% from the baseline NOX level. The highest reductions
in NOX were achieved with "abnormal" air register settings (coal-air 30% open,
and auxiliary air 70% open), as opposed to the normal settings of coal-air 80%
open, and auxiliary air 20% open. Additional small reductions in NOX emissions
could be obtained through the use of optimum burner tilt positions, and
pulverizer mill fineness, each contributing about 10% to the NOx emission
reductions achieved.
248
-------
FIGURE 14
PPM NOX (0% 02, DRY) VS % STOICHIOMETRIC
AIR TO ACTIVE BURNERS
(NAUGHTON STATION, NO. 3 BOILER)
700
600
«
B 500|
8
NORMAL FIRING
(260-330 MW)
DOWN-TILT
/
• H
HORIZONTAL TILT
400
300
200
STAGED FIRING
(250-260 MW)
HORIZONTAL AND UP-TILT
100
_L
J.
_L
_L
60
70
80 90 100 110
I STOICHIOMETRIC AIR TO ACTIVE BURNERS
120
130
249
-------
3.1.6 Barry Boiler No. 4
Alabama Power's Boiler No. 4 at their Barry Station was tested
successfully through all three phases of our test program design. Representatives
of Combustion Engineering actively participated in this series of tests. As
mentioned before, this new 350 MW maximum rated capacity, single furnace,
pulverized coal fired Combustion Engineering boiler is similar to Naughton
unit No. 3. Both are representative of that manufacturer's current design
practices. In Barry No. 4, five pulverizers feed 200 burners that are
corner-mounted at five levels of the furnace. This boiler is designed for
firing Eastern bituminous coal, having a HHV of 12,000 Btu/lb.
Altogether, 35 statistically designed short term period
optimization runs were made during the first phase of the test program. The
gaseous emission data obtained in this phase are presented in the least squares
correlations of Figure 15.
Among the minor operating variables from the standpoint of
emission control, burner tilt position had the most promised effect, as shown
in Figure 15. Horizontal and 30° up-tilt burner positions produced lower NOx
emissions than the 30° down-tilt position, particularly with all burners firing
coal. Varying burner tilt position, of course, has as its primary purpose to
control steam temperatures. With the burners tilted down, a large proportion
of the combustion process occurs in the direction of the bottom of the furnace
(hence increased slagging there) , and higher peak flame temperatures at longer
gas residence times produce more NOX. These increased NOX levels can be
partially offset by the ability to operate at lower excess air levels with the
down-tilt position, as more residence time is available for burn-out of the fuel.
With horizontal or up-tilt burner positions, there may be less than one complete
rotation of the swirling gases before reaching the boiler arch, which can lead
to some stratification (about one per cent difference in 62 between "A" and "B"
ducts.
The other minor variables studied, air register settings and coal
pulverizer fineness, did not produce significant changes in N0x« As expected,
the most pronounced effect on NOX was that of reducing the per cent stoichiometric
air to the burners under normal firing conditions, as shown by the data of
Figure 15 . From baseline levels of about 510 ppm (these remarkably low levels
are likely to be due in part to the ability to operate this boiler with only
10% excess air, and in part to the slightly lower bound N-content of Alabama
and Midwestern coals than that of the Western coal fired at Naughton) , reducing
the per cent stoichiometric air to 104% resulted in NOX emissions of 390 ppm,
or a reduction of about 23%.
Further significant reductions in NOX were achieved by staged firing,
admitting air only to the top level burners , and operating four or three
pulverizers, depending on load conditions. With only one pulverizer inactive,
and supplying 90% of the stoichiometric air to the active burners, the NOX
level was as low as 290 ppm, or a reduction of about 43% from the baseline
level with a reduction in load of about 15% due to staging. Interestingly,
the NO level was slightly higher, and the per cent reduction somewhat lower
than those found in testing Naughton Boiler No. 3, presumably because of
differences in boiler operation and coal type.
250
-------
FIGURE 15
PPM NOX (0% 02, DRY) VS % STOICHIOMETRIC
AIR TO ACTIVE BURNERS
(BARRY STATION NO. 4 BOILER)
7QO|- NORMAL FIRING
(325-350 MW)
DOWN-TILT
6001
(HORIZONTAL
|AND UP-TILT
500[ ' u
0N I STAGED FIRING
K 4001- A~ 280-300 MW / O
O - 180-210 MW D ' «A i HORIZONTAL
') AND UP-TILT
o* I A m' A^& "^ ^ (280-300 MW)
^ I DOWN-TILT
300r (280-300
7m, , -HORIZONTAL AND UP-TILT
P (180-210 MW)
100
70 80 90 100 110 120 130
% STOICHIOMETRIC AIR TO ACTIVE BURNERS
251
-------
Significantly, no slagging problems were encountered in testing
Barry unit No. 4 even at the lowest levels of excess air used (as opposed to
Naughton) and consequently, the 300-hour sustained "low NOX" run could be made
under excellent conditions.
3.1.7 Barry Boiler No. 3
Alabama Power Company's Boiler No. 3 at their Barry Station was
tested at the boiler operator's request for gaseous emissions only in a
short-term optimization program.
This unit is a 250 MW maximum continuous rating, twin furnace,
tangential, pulverized coal fired Combustion Engineering boiler. It has a
separated furnace arrangement, with radiant and horizontal superheater surfaces
in both furnaces. The pendant and platen sections constitute the superheat
surface in one furnace, and reheat surface in the other one. Six pulverizers feed
24 tangential burners (six levels of four burners) in each of the two furnaces.
This boiler was of special interest, because of the small value of
31.25 MW per "equivalent furnace firing wall". Our correlation based on
previously obtained data for coal fired boilers (4) would predict a baseline
NOX emission level of 475 ppm for this parameter. Actual measurements gave
baseline NO value of 479 ppm, in good agreement with the correlation.
Operating variables included in the actual test program were excess
air level, air damper settings, and mill pulverizer fineness setting. Planned
reduced load and staged firing tests could not be implemented, because mechanical
problems with a condenser water valve prevented such operation, despite all
the efforts of the plant personnel to correct the problem.
As expected, excess air level exerted a major effect on NOX emissions,
while that of damper settings was very small, and that of mill fineness was
negligible.
These results are shown in the least squares regressions of Figure 16.
From a baseline level of about 530 ppm at 122% stoichiometric air to the burners,
NOX emissions were reduced by about 32% (in line with our previous experience)
to 360 ppm at 106% stoichiometric air.
3.1.8 Big Bend Boiler No. 2
Tampa Electric Company's Boiler No. 2 at their Big Bend Station has
been the only Riley-Stoker turbo-furnace unit tested by us to date. This
pulverized coal fired, 450 MW maximum continuous rating, single furnace boiler
is fed by three pulverizer mills. Altogether, 24 Riley directional flame
burners are fired normally, with one row of 12 burners in the front wall, and
another row of 12 burners in the rear wall.
Maximum load was limited to 375 MW, due to steam temperature,
potential slagging, and other operating problems. (It is our understanding
that gross load on this unit has never exceeded 400 MW.) Excess air was set
at normal operating levels, or at the minimum level dictated by maximum acceptable
CO levels measured in the flue gas, and in the slag catcher at the bottom of the
furnace. Other operating variables included in the statistically design
short-term phase (this was the only phase of our overall program design performed
252
-------
FIGURE 16
PPM NOX (0% 02, DRY) VS % STOICHIOMETRIC
AIR TO ACTIVE BURNERS
(BARRY NO. 3 BOILER)
7001
6001
500!
400
100 AUX/30 COAL
I 2ND AIR REG.
40 AUX/100 COAL
2ND AIR REG.
300
200
J_
_L
80
90
100 110 120 130
% STOICHIOMETRIC AIR TO BURNERS
253
-------
at Big Bend) were operating with fly-ash reinjection (practiced to improve
carbon burn-out efficiency and slagging characteristics) or without it, and
positioning of the directional air vanes. Normal position is 15° below the
horizontal for the air vanes. During our tests, they were aligned either 15
below the normal position, in both front and rear burners, or the front
directional vanes were set at 15° below the normal position, and the rear
directional vanes 15° above it. Simulated "staged" firing, at reduced load
levels, was attempted by opening up the secondary air registers on selected
burners, so that the active burners were supplied with 80% of stoichiometric air.
The NOX emission results obtained are shown in the least squares
regression of Figure 17. Reducing the air to the burners from the normal
level of 115% of stoichiometric to 107%, decreased NOX emissions from about
675 ppm at 370 MW to 470 ppm with 107% of stoichiometric air, or a reduction
of about 45%. This decrease in NOX with reducing excess air is steeper than
that generally observed in wall and tangentially fired units. On the other
hand, it should be noted that the "baseline" NOX emission was determined at a
load reduction of 18%, compared with maximum continuous rating. Further load
reduction produced, as expected, further decreases in NOX.
"Staged" firing, which in this instance was quite different from the
normal pattern of staging burners, produced only a 10% reduction in NO at the
low load of 230 MW, as shown in Figure 17.
The best NO,, reductions were obtained with front wall directional
air vanes tilted 15°
-------
FIGURE 17
PPM NOX (0% 02, DRY) VS % STOICHIOMETRIC
AIR TO ACTIVE BURNERS
(BIG BEND NO. 2 BOILER)
80C
700
600
500
400
•NORMAL FIRING (370 MW)
NORMAL FIRING (300 MW
"STAGED" FIRING (230 MW)
I I L
300
70
80
90 100 110 120
% STOICHIOMETRIC AIR TO ACTIVE BURNERS
130
255
-------
TABLE 5
PARTICULATE
Utility
TVA
Georgia
Power
Company
Arizona
Public
Service Co.
Alabama
Power
Company
Utah Power
& Light Co.
Gulf Power Co.
Test No.
1A
IB
10-C-l
10-C-3
26-A-l
1C
ID
IE
1G
1H
52D
52E
IE
IF
12A
12B
42A
42B
19A
19B
23
23
25
26
1
26B
Firing
Condition
Base
Base
Low NOx
Low NOX
Low NOX
Base
Base
Base
Low NOX
Low NOX
Low NOX
Low NOX
Base
Base
Low NOX
Low NOX
Base
Base
Low NOX
Low NOX
Base
Base
Base
Base
Base
Low NOX
02
Before
A.M.
3.9
3.6
3.05
3.23
2.73
3.0
3.7
3.5
1.2
1.3
1.9
2.0
3.44
3.12
4.28
3.67
—
—
—
— —
3.56
3.56
4.31
4.46
3.6
3.4
Calc.
°2
After
A.H.
5.17
4.89
4.38
4.55
4i09;
4.34
4.99
4.80
2.67
2.76
3.32
3.41
4.75
4.45
5.52
4.96
5.03*
4.49*
4.64*
4.34*
4.86
4.86
5.55
5.69
4.89
4.71
Av.
Gr/SCF
@ Std.
Cond.
2.68
4.62
2.32
3.36
3.13
1.83
1.86
2.26
2.47
2.60
2.00
2.65
4.52
5.36
4.87
3.26
1.17
3.08
3.31
3.32
0.448
0.301
0.752
0.800
2.54
3.82
DATA
Reqd.
Efficiency
Gr/SCF
@
0% 0?
3.55
6.02
2.93
4.29
3.89
2.31
2.44
2.93
2.83
2.99
2.38
3.16
5.84
16.80
6.40
4.27
1.53
3.92
4.25
4,19
0.58
0.38
0.34
1.10
3.31
4.92
lb./106
BTU
4.65
7.89
3.84
5.62
5.10
3.03
3.20
3.84
3.71
3.92
3.12
4.14
7.65
8.91
8.38
5.59
2.00
5.14
5.57
5.49
0.76
0.51
0.44
1.48
4.34
6.45
Grams /
106 cal.
8.37
14.20
6.91
10.12
9.18
5.45
5.76
6.91
6.68
7.06
5.62
7.45
13.77
16.04
15.08
10.06
3.60
9.25
10.03
9.88
1.37
0.92
0.81
2.59
7.81
11.61
To Meet
0.1 lb/
106 BTU
97.85
98.73
97.40
98.22
98.04
96.70
96.88
97.40
97.30
97.45
86.79
97.58
98.69
98.88
98.81
98.21
95.00
98.05
98.20
98.18
86.91
80.55
77.73
93.04
97.70
98.45
%
Carbon :cm
Particulate
6.29
5.90
10.55
8.46
12.40
5.50
3.17
2.80
6.73
11.82
9.98
7.41
0.69
0.53
0.18
0.46
24.23
25.83
14.75
18.77
22.62
22.62
4.44
1.80
5.08
8.15
Coal HHV
Ash BTU/lb.
Wet, % Wet
15.87 11,452
18.39 11,477
11.5 11,918
14.38 11,231
15.39 10,961
12.05 12,310
9.72 12,589
8.58 12,121
11.28 12,200
8.43 12,574
10.3 11,178
11.86 11,887
21.92 8,821
21.96 8,811
23.13 8,913
21.12 8,915
4.89 12,706
4.86 12,641
10.68' 11,918
8.82 12,720
8.16 10,293
8.16 10,293
6.78 10,273
8.10 9,992
Not Yet .
Available
-------
Average emission data in grains per cubic foot at standard conditions
were calculated from the particulate data obtained in the tests and are listed
in Table 5. These data are also presented in terms of grains per SCF at
zero percent oxygen, pounds per million BTU and grains per million calories.
For comparison purposes, a calculation of the required precipitator collection
efficiency to meet current federal standards of 0.1 pounds per million BTU
fired is included in Table 5.
Tennessee VajJ-ey_Authority
As indicated in Table 5, five particulate tests were made at TVA.
The unit tested was boiler No. 6 at the Widows Creek Station. Two tests were
run at baseline conditions and three at optimized "low NOX" firing. Comparing
results of tests 1A with 10-C-3 and 26-A-l, all conducted while firing
approximately the same ash content coal, it can be seen that emissions increase
from a baseline of 4.65 lb/106 BTU to 5.62 and 5.10 lb/106 BTU at "low
operation, respectively. The carbon content of the particulate in these
tests increases from 6.29% at baseline to 8.46 and 12.40%, respectively,
under "low NOX" operation. Relatively small increases in electrostatic
precipitator efficiency are required, however, to meet present standards,
i.e., 97.85% to 98.22% and 98.04%, respectively.
Based on the data obtained on the No. 6, 125 MW, front wall fired
boiler at TVA's Widows Creek Station, it appears that particulate emissions
increase directionally, but not significantly, when "low NQX" firing
configurations are employed. Particulate carbon content which initially is
relatively low on the TVA boiler, increases substantially (doubling in one case)
under "low NOX" firing conditions but the increases do not appear to be in
direct relationship with the emissions or coal ash data. Only small increases
in electrostatic precipitator collection efficiency would be required to
accommodate the higher dust loadings produced with "low NOX" firing. However,
if real, the incremental increases at these levels of performance may be
difficult and costly to achieve.
Georgia Power
A total of seven particulate tests were conducted on the No. 3,
480 MW, horizontally opposed fired boiler at Georgia Power Company's Harllee
Branch Station, three at baseline or normal operating conditions and four
while firing the boiler using "low NOX" modifications. Particulate emission
data for all tests were relatively consistent (see Table 5), ranging from
3.03 to 4.14 pounds per million BTU's fired. Average emissions at normal
or "base" conditions were 3.36 lb/106 BTU compared to 3.72 lb/106 BTU for
"low NO " producing conditions. Particulate carbon content was variable
and again was inconsistent with other factors. Under normal firing conditions,
percent carbon was reasonably low, varying from a low of 2.8 to 5.5 percent.
At "low NO " operation these values ranged from 6.73. to 11.83 percent.
Increases in the required precipitator efficiency to meet present standards
again would be small, and probably of minor importance.
257
-------
Arizona Public Service Company
Particulate test results obtained on the No. 4, 800 MW, horizontally
opposed fired boiler at Arizona Public Service Company's Four Corners plant
are of particular interest because of the low sulfur, high ash Western coal
fired. Four tests were conducted, two under baseline and two under "low NOx"
operation. Particulate emissions, as expected, due to the high ash (^ 23%)
coal fired were high in all tests. Referring to Table 5, the results are
somewhat confused by the high emissions obtained in one base run and the
low emissions obtained in one of the "low NOX" tests. However, if results
of test IE are compared with 12A, a more normal pattern is apparent.
Particulate emissions of 7.65 lb/106 BTU for baseline operation (test IE)
increase to 8.38 lb/106 BTU for "low NOX" test 12A. The increase, as in
previous tests, is not substantial but is in line with what might be expected.
Data on the percent carbon on particulates for all tests are startling. In
the first place, the values are very low, averaging about 0.47%, confirming
the easy burning qualities of Western coals. Secondly, particulate carbon
content decreases with "low NOX" firing as indicated in Table 5, from
0.695 and 0.528% in "base" tests IE and IF, respectively, down to 0.182 and
0.461% for "low NOX" tests 12A and 12B, respectively. Thus, according to
these data a benefit accrues to "low NO " operation with respect to unburned
combustibles for this horizontally opposed fired boiler. With respect to
precipitator performance, efficiency should only increase from 98.7 percent
(baseline operation, test IE) to 98.8 percent (low NOX firing, test 12A) to
accommodate the increased particulate emission produced under "low NOX"
firing conditions.
Alabama Power Company
Four dust loading tests were conducted on the 350 MW, tangential
fired No. 4 boiler at Alabama Power Company's Barry Station, two at normal
operation and two while using "low NOX" emission reduction techniques. Referring
to Table 5, the data obtained under baseline operation in test 42A appear
to be unreliable. Comparing the results obtained in the other 3 tests
(tests 42B, 19A & 19B) it can be seen that particulate emissions for this
tangentially fired boiler increase from 5.14 lb/10" BTU under normal operation,
to 5.57 and 5.49 lb/106 BTU when "low NOX" emission techniques are used.
Particulate carbon content for baseline operation (test 42B) of 25.83 percent
(see Table 5) is very high, and considerably higher than for other types of
firing. Surprisingly, substantial reductions in these carbon losses appear
to occur with "low NOX" operation. This behavior is shown by tests 19A and 19B
with decreases in particulate carbon content down to 14.75 and 18.77 percent,
respectively. Here again, "low NOX" firing techniques apparently have
beneficial results. Only nominal increases, probably of no major importance,
are required in precipitator collection efficiency when employing "low NOX"
techniques.
Utah Power & Light Company
Tests for particulates on the 330 MW, tangentially fired, No. 3
boiler at the Naughton Station were made in the ducts leading into the air
heaters, since downstream test locations were poor, and accessibility was
limited. Four tests were run, all of which were conducted under normal or
baseline firing conditions. Referring to Table 5 it can be seen that the
emission values are not consistent with other data obtained in this study,
especially in view of the consistency and levels of the ash content of the
coal fired.
258
-------
Since the tests were conducted in accordance with prescribed
procedures, it is difficult to understand the reasons for these inconsistencies.
One possible explanation is that the superheater and reheater surfaces on
the No. 3 boiler were overdesigned necessitating operation with the burners
at a horizontal position or tilted downwards. As a result, the lower furnace
surfaces including the ash hopper slopes were slagged, while the superheater
reheater and convection section surfaces were extremely clean on this relatively
new unit. It is possible that a major portion of the ash was impinging on the
sticky slag particles and remained in the boiler, thus accounting for the low
dust loadings.
» u - vflues reP°rted for particulate carbon content are of interest.
As shown in Table 5, it may be noted that carbon content in test No. 23 was
22.62 percent. This value is consistent with the high values reported for
the Alabama Power Company tests (Tests 42A and 42B) . Particulate carbon
content reported for the latter, tangentially fired boiler was 24.23 and
25.83 percent, respectively, at baseline operating conditions. However, the
values of 4.44 and 1.80% carbon content reported for tests 25 and 26 for normal
operation are surprising since they are low and at variance with the Alabama
Power Company test results for baseline conditions. More data on other
tangentially fired boilers are required to resolve this apparent anomaly.
Gulf Power Company
Due to the limited scope of testing, particulate data were obtained
in only two tests on the 320 MW, front wall fired, No. 6 boiler of Gulf Power
Company's Crist Station. One was at normal operating conditions, and the other
under "low NOX" operating conditions. As shown in Table 5, the results
appear to be in line with other emission data measured. Particulate emissions
for baseline operation increases, as might be expected from 4.35 percent to
6.45 percent under "low NOX" operating conditions. Similarly, carbon content
of the particulate increases from 5.08 percent at normal operation to 8.15
percent when "low NOX" burner configurations are used. Required precipitator
efficiency to meet present standards would increase from 97.7% for base
conditions to 98.4% to accommodate the higher emissions produced with "low NOX"
firing.
3.3 Corrosion Probing Results
As mentioned in section 2.2.3, corrosion probes were installed in
the furnaces of the boilers tested, by inserting them through available
openings closest to the areas of the furnace susceptible to corrosion, as
indicated in Figure 4. Prior to installing the probes in the test furnace,
the probes were prepared by mild acid pickling and pre-weighing the coupons ,
and screwing them onto the probes along with the necessary thermocouples.
Each probe was then exposed to the furnace atmosphere prevailing for the
particular type of operation desired for approximately 300 hours at coupon
temperatures of about 875°F in order to accelerate corrosion. After exposure,
furnace slag was cleaned off and saved for future analyses, and the coupons
were carefully removed from the probes. In the laboratory the coupons were
cleaned ultrasonically with fine glass beads to the base metal, and re-weighed
to determine the weight loss. To date in our tests, corrosion rates have
been determined for 40 coupons installed on 20 probes (2 coupons /probe) ,
in boilers at four different generating stations as listed in section 2.2.3.
Corrosion data obtained are tabulated in Tables 6 through. 10.
259
-------
TABLE 6
GEORGIA POWER COMPANY
HARLLEE BRANCH STATION
CORROSION PROBE DATA
Boiler
No.
3
3
4
4
Firing
Condition
Low NO
X
Low NO
X
Base
Base
Exposure
Hrs
297
297
304
304
Probe
No.
3A
3B
4A
4B
Coupon
No.
0
0
f"
111
f12
I 13
Corrosion Rate
Mils/Yr
27.5
122.0
75.9
155.0
75.3
72.2
25.7
47.9
260
-------
TABLE 7
UTAH POWER & LIGHT COMPANY
NAUGHTON STATION
CORROSION DATA
Boiler Firing Exposure Probe Coupon Corrosion
No. Cond. Hrs . No. No. MPY
Base
Base
Base
Base
287.0
287.5
283.5
283.75
124
65
43
47
16
24
25
25
261
-------
TABLE 8
ARIZONA PUBLIC SERVICE COMPANY
FOUR CORNERS STATION
CORROSION DATA
Boiler Firing Exposure .Probe Coupon Corrosion
No. Cond. Hrs No. No. MPY
4 Low NO 255.25
4 Low NO 255.5
Base
Base
273.5
273.75
C
R
P
Q
N
0
61
160
25
24
157
59
45
59
262
-------
TABLE 9
ALABAMA POWER COMPANY
BARRY STATION
CORROSION DATA
BASE OPERATION
Boiler Firing Exposure Probe Coupon
No. Cond. Hrs No. No.
( W
4 Base 295.5 1 /
I W
4 Base 295.5 2 /
I5
4 Base 295.75 3 <
I"
/17
4 Base 295.25 4 (
! 18
Corrosion
MPY
34
24
17
18
11
13
16
17
263
-------
TABLE 10
ALABAMA POWER COMPANY
BARRY STATION
."LOW ,NO^" OPERATION
X
Boiler Firing Exposure Probe Coupon Corrosion
No. Cond. Hrs No. No. MPY
u
4 Low NO 282.75 I* (
\B
f K
4 Low NO 282 2* \
1 M
32
26
41
52
77
13
4 Low NO.. 281.75 4**
18
* Eleven feet below lower burners in side walls (slag blowers
No. 3 and 11).
** Eleven feet above top burners in side walls (slag blowers
No. 18 and 26).
264
-------
Total weight loss data were converted to corrosion rates on a mils
per year basis, using the combined inner and outer coupon areas, coupon
material density, and exposure time. Wastage was found to have'occurred
on the internal surfaces of some of the coupons, possibly because of the
oxidation of the hot metal by the cooling air. Attempts were made to
determine "internal" and "external" corrosion rates by selective cleaning
and weight loss determinations, but the results were found to be more
consistent and reliable on an overall basis.
3.3.1 Georgia Power Company
The first furnace corrosion probe tests were conducted on boilers
numbers 3 and 4 at the Harllee Branch Station of the Georgia Power Company.
Boiler No. 3, as indicated in Table 6, was fired at "low NOX" conditions
while boiler No. 4, a sister-unit was used to obtain baseline data under
normal operation. All probes were inserted through slag blower ports
extending through the windbox at an elevation about 8 ft. above the top
burners. Exposure of the probes was maintained for approximately 300 hours
at these firing conditions, after which the corrosion coupons were removed
and processed in the laboratory.
Corrosion rate determinations for the Georgia Power Company tests
are tabulated in Table 6. It may be noted that the corrosion rates on
coupons No. 6 and 8 exposed to "low NOX" firing conditions are the same as
those obtained on coupons No. 10 and 12, exposed to normal firing. Rates on
"low NOjj" coupons Nos. 7 & 9, however, are considerably higher than those
obtained under normal operation. Analyses of these data, however, indicate
that the differences are not statistically significant. Since this one was
our first furnace corrosion tests conducted, a possible explanation for the
higher corrosion rates obtained on the same probe, i.e., coupons No. 6, 7,
and 8, 9, is that metal temperatures on adjacent coupons may not have been
balanced, and potentially could have been higher on those coupons showing
higher corrosion rates. This could also explain some of the differences in
corrosion rates between "low NOX" and baseline operation.
It is concluded that there are no significant differences between
the corrosion rates under "low NOX" firing conditions in Harllee Branch
boiler No. 3, and those in No. 4 operated under normal firing conditions,
even though somewhat higher corrosion rates were measured on two coupons on
probes exposed to "low NOX" conditions.
3.3.2 Utah Power & Light Company
Four corrosion probes were installed in inspection doors on the
front wall of boiler No. 3 at the Naughton Station of the Utah Power & Light
Company, as indicated in Figure 4. The objective was to obtain both
"baseline" and "low NOX" corrosion data at the same time, under "low NOX"
firing conditions. This attempt was based on the high 02 levels expected
to prevail at the-upper inspection doors, where probes No. 1 & 2 were located.
Reducing atmospheres were expected to prevail in the vicinity of probes
No 3 & 4 located in the middle of the burner array. However, potential
furnace slagging conditions and critical system load conditions prevented
sustained operation at "low NOX" conditions on this unit. Accordingly,
the corrosion probes were exposed to normal operating conditions for a
period of 300 hours. Although the data obtained on these probes do not
permit comparison of corrosion rates at "low NOX" and baseline conditions
for this unit they do provide information on baseline corrosion rates of
value for comparison with similar data obtained on other boilers.
265
-------
Corrosion rates obtained at the Naughton Station are listed in
Table 7. It may be seen that the corrosion rates on all coupons are
reasonably consistent, ranging between 16 and 65 mils/yr., with the exception
of probe No. 1 which had a higher rate and is out of line with the others.
Corrosion rates on probes Nos. 1 & 2, ranging between 43 and 124 mils/yr.,
are higher than those on probes No. 3 & 4 (16-25 mils/yr.). Measured
oxygen levels at the latter probes were lower at this location than that
prevailing at probes No. 1 & 2. These data, therefore, provide some
indication that corrosion rates at lower oxygen levels may be less than
that of coupons exposed to higher oxygen atmospheres.
It is concluded that the Naughton data are indicative of accelerated
corrosion rates prevailing under normal firing conditions on tangentially
fired boilers. These data also indicate that corrosion rates may be lower
on probes exposed to lower oxygen level environments.
3.3.3 Arizona. Public Service Company
Two corrosion probes each were installed in Boiler No. 4, at the
Four Corners Station, the "low NOX" test unit, and Boiler No. 5, the "base"
operation unit, which is a duplicate of Boiler No. 4. The probes were
inserted through the slag blower ports which extend through the windbox
of these horizontally opposed fired units. Corrosion data obtained are
shown in Table 8.
Probes No. 1 & 2 in boiler No. 4 were exposed for about 255 hours,
essentially at "low NOX" firing conditions. However, there were periods
during this time span when the unit was not operated entirely at the
prescribed "low NOX" conditions, due to mill losses, upsets in plant
operation, and other problems. For the major part of the time, though,
the probes were exposed under "low NOX" firing conditions. It can be seen
from Table 8 that corrosion rates range between 24 and 160 mils/yr. on
the coupons mounted on probes No. 1 & 2 exposed to "low NOX" conditions.
These values compare to a range of 45 to 157 mils/yr. obtained on the
"baseline" probes (No. 3 & 4) which were exposed to normal firing conditions
for about 274 hours. The lowest corrosion rates, 24 & 25 mils/yr., were
experienced on coupons R & S mounted on probe No. 2 exposed to "low NOX"
firing. The lowest rate on the base operation probes was 45 mils/yr.,
on coupon "N". Comparison of the corrosion rates on the remaining coupons
shows that there are no significant differences between the coupons exposed
under "low NOX" firing and those exposed under normal firing conditions.
In fact, the rates are remarkably consistent, and practically equal.
From the corrosion data obtained at the Four Corners Station, it
is concluded that corrosion rates are essentially the same under "low NOX"
and baseline firing conditions. Also, based on these limited data, there
is an indication that corrosion rates under "low NO " conditions may even
be somewhat lower than under "baseline" operating conditions.
266
-------
Four corrosion probes were installed on boiler No. 4 at the
Barry Station of the Alabama Power Company and exposed for about 282 hours
under "low NOX" operating conditions. The probes were removed at the conclusion
of this test, refitted with new coupons and re-inserted in the same boiler
and locations and exposed for about 282 hours under normal firing conditions.
Locations of the probes in the boiler are detailed in Figure 4 and corrosion
data are tabulated in Table 9 for baseline operation, and Table 10 for
"low NOX" firing.
Comparing Tables 9 and 10 it will be noted that coupon corrosion
rates for "low NOX" operation are higher than for a baseline operation. The
"low NOX" corrosion rates are significantly higher at about the one per cent
probability level. The corrosion test in boiler No. 4 was probably the most
reliable test made under "low NOX" conditions to date. Load and "low NOX"
firing conditions during the test period were maintained steadier than in
other tests. Also, except for some minor variations; the coupon temperatures
were maintained more consistently at the 875°F set point than in other tests.
In contrast, even though coupon temperatures were maintained at a reasonably.
consistent level, boiler load conditions under the base operation test period
varied widely. This problem was primarily due to pulverizer failures which
required removing pulverizers from service. When the test was first started
four pulverizers were in service. Mills were progressively dropped off until
only two were in service for a good portion of the test period. This factor
could have a significant effect on the corrosion rates shown in Table 9
for baseline operation.
Referring to Table 9, it should be noted that coupon corrosion
rates on probes 1 & 2 installed below the lower burners averaged 24 mils
per yr. compared to an average of 14 mils per yr. for probes Nos. 3 & 4,
located above the top burners. Under "low NOX" firing conditions (Table 10)
the reverse occurred; average corrosion of coupons on probes 1 & 2 was
38 mils per yr. compared to 49 mils per yr. for probes No. 3 & 4. This
difference could be explained by the less intense firing conditions prevailing
in the area of probes No. 3 & 4 during the baseline test due to firing the
lower burners only. However, it is doubtful that this reversal in corrosion
rates between the upper and lower probes under the different firing conditions
is truly significant.
Based on the data of Tables 9 and 10 for the Alabama Power Company
tests, it is concluded that corrosion rates under "low NOX" firing conditions
are significantly (but not catastrophically) higher than those measured under
baseline operating conditions. Furthermore, corrosion rates on probes at
different locations in the boiler may be different, depending on whether the
boiler is fired normally, or with "low NOX" firing modifications, and also,
with the location of the flame zone in the furnace.
267
-------
4. DISCUSSION
In this section of the paper, the overall correlations of the NO^
emission data, and the significance of particulate emission measurements and
accelerated corrosion tests are discussed.
4.1 Gaseous Emission Measurements
Tables 11 and 12 summarize the NO emission levels measured from
wall-fired and tangentially-f ired (plus a turbo -furnace) boilers, respectively.
Inspection of Table 11 reveals that all of the wall-fired boilers have baseline
NOX emission levels greater than the current federal standard of 0,7 pounds
NOX/106 BTU or 1.26 grams NOX/106 cal for new units. "Low NOX" operation
at full load reduced NOX emission levels by 25 to 47% from baseline levels,
and only Crist No. 6 boiler was unable to meet the federal NOX standard for
new boilers. "Low NOX" operation at reduced load resulted in 40 to 54% NOX
emission reductions from baseline operation. The high CO emission levels shown
under "low NOX" operations were generally reduced to acceptable levels during
the sustained test periods.
Examination of Table 12 reveals that baseline NOx emission levels
from tangentially fired boilers are lower than baseline NOx emission levels
from wall fired boilers. (The turbo-furnace boiler was tested at 370 MW
compared to design full load of 450 MW, and hence, additional testing is needed
to measure baseline, full load NOX emission levels.) "Low NOx" staged firins
operation with 15-20% load reduction enabled these boilers to decrease NOX
below the federal NOx emission standard for new coal fired boilers by a large
margin, while each of these boilers demonstrated the capability of meeting
such standards at full load with low excess air operation during the short-
period tests. "Low NOx" operation with further load reduction resulted in
NOx emission reductions of 55 to 64% compared to full load, baseline emission
rates.
As noted in Section 3, it should be recognized that these results
were obtained during short-term test periods and that long-term testing is
needed to study slagging, corrosion and other operating conditions. It is
expected that slagging problems in some boilers can be largely overcome by
increasing slag blower steam pressures, increasing the use of slag blowers
and perhaps the addition of slag blowers at troublesome locations. Lower NOX
emissions would also be expected in many boilers from improved furnace maintenance,
so that air-to-fuel ratios are as uniform as practical across the furnace.
Research at extremely low % stoichiometric air to the active burners (less
than 75%) with staged firing may yield significantly improved NOX emission
rates with decreased slagging, because of lower temperatures. Also, the
addition of "NO-ports" would probably allow most boilers to reduce NOX emissions
significantly during full-load operation with all burners firing coal.
268
-------
TABLE 11
SUMMARY OF NOX EMISSION LEVELS FROM
WALL FIRED BOILERS
BOILER
1. WIDOWS
OPERATING MODE
(GROSS LOAD-MW)
BASE (125)
CREEK No. 6 | "LOW NOX"* (125)
(FW)
2. CRIST
NO. 6
(FW)
3. HARLLEE
BRANCH NO. 3
(HO)
4. FOUR
CORNERS NO. 4
(HO)
i
i "LOW NOX"** (100)
BASE (320)
"LOW NOX"* (320)
"LOW NOX"** (272)
BASE (480)
"LOW NOX"* (478)
"LOW NOX"** (400)
BASE (800)
"LOW NOX"* (800)
I "LOW NOX"** (600)
°2
3.0
1.7
2.7
3.6
2.6
3.1
NOY EMISSIONS
PPM
(0% 00)
750
395
346
990
740
600
3.5 } 870
LBS/
106 BTU
0.84
0.44
0.39
:
1.10
0.82 i
0.67 [
!
0.97
1.7 575 1 0.64
1.3
5.0
390 | 0.43
1070
3.2 570
1.19
0.64
3.0 ] 525 I 0.59
GRAMS/
106 BTU
1.50
0.79
0.69
.
1.98
1.48
1.20
PPM
CO***
(0% 0,)
300
1140
980
26
580
310
1.75 24
1.15 61
0.78 | 1080
si
2.15 ( 23
1.15 ( 200
1.05 } 33
* "LOW NO " CONDITIONS SELECTED FOR SUSTAINED RUN, AT FULL LOAD.
** "LOW NO " CONDITIONS AT REDUCED LOAD.
X
*** LOWER CO LEVELS MEASURED UNDER SUSTAINED STEADY-STATE CONDITION IN REPEAT RUNS.
-------
TABLE 12
N)
•^J
O
BOILER -
5. NAUGHTON
NO. 3
6. BARRY
NO. 4
7 . BARRY
NO. 3
SUMMARY OF NOX EMISSION .LEVELS FROM
TANGENT lALLY FIRED BOILERS
OPERATING. MODE
(GROSS LOAD-MW)
BASE (328)
"LOW NOX"* (256)
"LOW NOX"** (200)
BASE (348)
"LOW NOx"* (285)
"LOW NOX"** (185)
BASE (250)
"LOW NOx"* (250)
°2
3.9
2.9
3.2
4.6
3.3
3.7
3.1
1.3
NOX EMISSIONS
PPM
(0% 00)
600
230
214
485
285
220
480
360
LBS/
106 BTU
0.67
0.26
0.24
0.54
0.32
0.24
0.54
0.40
GRAMS/
106 CAL.
1.20
0.46
0.43
0.97
0.57
0.44
0.96
0.72
PPM
CO
(0% OJ
35
440-
65
1
28
68
328
71
116 l
TURBO-FURNACE BOILER |
8. BIG BEND
NO. 2
BASE (370)
"LOW NOX" * (300)
"LOW NOX"** (230)
2.8
1.8
3.5
715
400
365
0.79
0.44
0.41
;
1.43
0.80
0.73
30
80 \
260 I
* "LOW NO " CONDITIONS SELECTED FOR SUSTAINED RUN.
x
** "LOW NO " CONDITIONS WITH FURTHER LOAD REDUCTION.
x
-------
Figures 18, 19 and 20 have been prepared to show the overall
correlations of NOX emissions vs % stoichiometric air, and gross load
per furnace firing wall for the eight coal fired boilers tested to date in
this program.
Figure 18 is a plot of "normalized" NO emissions, expressed as
a % of baseline NOX emissions (full load and 20% excess air) vs. % overall
stoichiometric air (or % stoichiometric air to active burners) under normal
firing conditions. The solid lines shown for each boiler are based on
least-squares, linear regression analysis of all test runs made under normal
(all burners firing coal) full load firing conditions. With the exception
of the turbo-furnace boiler, all of these regression show very good agreement
with about a 20% reduction in N0x at 110% vs. 120% stoichiometric air. The
three tangentially fired boilers show especially good agreement in this
significant correlation of NOX emission levels with excess air levels.
Figure 19 is a plot of normalized NOx emissions expressed as a %
of baseline NOX emissions (full load and 20% overall excess air) vs. %
stoichiometric air to the active burners under modified firing conditions.
Thus, the ordinates are identical in Figures 18 and 19. However, the least
squares regression lines of Figure C2 do not necessarily pass through the
100% normalized NOX point at 120% stoichiometric air to the active burners,
as they must, by definition, in Figure 18. Regression lines for Barry No. 3
and Big Bend No. 7 do pass through the 100%/120% point, since staged firing
was not employed for those boilers.
Figure 19 indicates the importance of low excess air firing on
emissions, as well as the further benefits of staged firing and additional
firing modifications. The opposed wall fired boilers Harllee Branch No. 3
and Four Corners No. 4 boilers showed excellent agreement, as would be expected,
since both of them represent modern design practices of Babcock and Wilcox
with their cell-type burners. The tangentially fired boilers, Barry No. 4
and Naughton No. 3, that employed staged firing showed similar trends, with
Naughton No. 4 giving the lower NOX emissions because it was tested at lower
% stoichiometric air levels. Widows Creek No. 6 boiler showed consistently
larger reductions with normalized NOX than Crist No. 6 boiler at the same
stoichiometric levels. Boiler parameters such as size, coal type fired,
pulverizer conditions, and other design and operating variables undoubtedly
contribute to the differences found. We are in the course of examining
whether the correlation methods of Figures 18 and 19 are applicable to oil
and gas fired boiler NOX emission data obtained in earlier work.
Figure 20 is a plot of baseline NOX emission levels (ppm at 0% 02,
dry basis) vs. gross load per furnace firing wall for the 8 boilers under
baseline operation. The dashed line is calculated from the 1971 "Systematic
Field Study" (4). There appears to be a good correlation on this basis.
However, we expect to find an improvement by combining the results of all
15 coal'fired boilers tested to date. The regression intercept of 478 ppm
NO at zero load corresponds to about 25% conversion of the average fuel
nitrogen content of 1.3 wt% of the coals fired in this study. This
observation is a strong indication of the significant contribution of bound
fuel nitrogen to NO emissions from coal fired boilers. Substoichiometric
air supply to the active burners is expected to reduce both the fixation
of molecular N2, and the oxidation of fuel nitrogen, based on independent
laboratory data.
271
-------
-------
FIGURE 19
EFFECT OF EXCESS AIR ON NOX
UNDER MODIFIED EJRING CONDITIONS
100
w
TYPE OF
FIRING
HARLLEE BRANCH NO. 3 OPPOSED W
FOUR CORNERS NO.
CRIST NO.
BARRY NO.
NAUGHTON NO.
WIDOWS CREEK NO.
BARRY NO.
BIG BEND NO.
OPPOSED W
FRONT W.
TANG.
TANG.
FRONT W.
TANG.
TURBO
90
100
110
120
STOICHIOMETRIC AIR TO ACTIVE BURNERS
-------
FIGURE 20
COAL FIRED BOILERS
UNCONTROLLED N0x EMISSIONS VS GROSS LOAD
PER FURNACE FIRING WALL
1200T
1000-
800
M
600
O
53
400
200
1971 "SYSTEMATIC
;;LD STUDY" (4.)
PPM NO = 451 + 1.622 MW/FFW
CODE TYPE OF FIRING
O
A
O
FRONT WALL
OPPOSED WALL
TANGENTIAL
TURBO
LETTERS INSIDE SYMBOLS
DENOTE BOILER ABBREVIATIONS
50
100
150
200
250
300
350
400
GROSS LOAD PER FURNACE FIRING WALL - MW
274
-------
4.2 Particulate Measurements
Obtaining good particulate data is a difficult, time consuming
task. One must be involved in the actual test work to appreciate the
difficulties encountered. C. A. Gallaer discusses this matter in detail
in his paper on the testing of large precipitators (7).
In this program, four Research Appliance Company EPA-type particulate
sampling trains were used. The design of this equipment is very good, but
many difficulties occur in operation, as is inherent to particulate testing.
Care must be taken to assure that the probes and tests boxes are at specified
temperatures. Even then, especially in cold weather, moisture in the flue
gases condensing in the apparatus can quickly plug filters and abort the
test. Tests for leaks in each train prior to testing is also needed if
meaningful data are to be obtained. Plugging of sampling probes on occasion
also occurs, and can present difficulties in boilers with high dust loadings.
The facilities to be tested are another source of numerous problems.
Rarely is a boiler encountered with convenient testing facilities. Sample
test ports are usually located too close to bends in the flue ducts where
particulate concentrations, due to centrifugal action, cannot possibly be
uniform. Interferences of the probes with supports inside the flue ducts
and of the test apparatus with other obstructions outside the boiler, near
test locations, all contribute to the difficulty in running particulate
loading tests. Last but not least, the EPA-type test train is built
for horizontal probing, while most boiler test locations require vertical
probing. Our equipment has been modified for7 vertical probing, so that
usually the construction of scaffolding is necessary for access to the equipment.
Despite the problems of conducting particulate tests, the results
obtained on this program, summarized in Table 5, are internally consistent
and appear to be reliable within the limitations of this type of testing.
The objective of our work was to develop information on potential "side effects"
of "low NOX" firing techniques on total quantities and the carbon content of
the particulates generated. It is recognized that strict adherence to EPA
procedures was not always possible especially with regard to the number of
sampling ports and traverse points, but the same procedures were used for
under both baseline and "low NOX" conditions. Therefore, the differences
shown by the results on particulate emissions and particulate carbon content
in Table 5 should be quite reliable.
Not unexpectedly, some "side effects" did develop with "low NOX"
firing. Total quantities of particulate tend to increase but not signicantly
and the consequences appear to be relatively minor. This trend would have
an adverse effect on the required collection efficiency of electrostatic
precipitators to meet present Federal emission standards, but the increases
in efficiency indicated by these limited tests appear to be quite small.
275
-------
Another "side effect" of "low NOX" operation is that on carbon
losses. Carbon content of the particulates with "low NOx" operation,
according to the data, increase significantly for front wall and horizontally
opposed fired boilers. The data are quite scattered, and these increases
do not appear to be directly related to the change in emissions with "low NOX"
firing techniques, or other boiler operating variables. Comparative performance
calculations have not yet been completed for assessing the magnitude of such
adverse effects of staged firing with coal. It is possible, however, that
this debit may be offset by improved boiler efficiency due to the lower excess
air operation at "low NQx" conditions. Surprisingly, there is some evidence
that "low NOx" firing techniques for tangentially fired boilers decrease
carbon losses significantly. If this finding can be substantiated for other
tangentially fired boilers, a net credit may be applied to "low NOX" operation
of these units. It also appears that "low NOX" firing may decrease carbon
losses for boilers fired with western coals. Such improvements, however,
would not be substantial since unburned combustible losses with the easy-to-burn
Western coals are already low.
More data are needed on all types of boilers to substantiate these
findings. It is important to note, however, that no major adverse "side effects"
appear to result from "low NOX" firing with regard to particulate emissions.
4.3 Furnace Corrosion Testing
Corrosion of boiler furnace sidewall tubes was experienced in the
early days of the development of pulverized coal firing. Considerable effort
was expended at the time in the field, to find solutions to the problem; and
in the laboratory, to determine the corrosion mechanism. Eventually, simple
solutions were found by increasing the level of excess air and taking steps
to avoid impingement of ash particles on sidewall tubes. Apparently, not
much information had been published, probably because a practical solution
to the problem was available.
Recent regulations requiring reduction of nitrogen oxide emissions
have led to the reduction of excess air levels in firing boilers, as one of the
techniques to achieve lower emission levels. This approach has resulted in
considerable speculation and apprehension that furnace sidewall tube corrosion
problems will again be encountered. Quite naturally, boiler owners are
reluctant to subject their units to long term tests to determine potential
corrosion problems without some assurance that risks are not grave.
For the above reasons, part of the current program was devoted to
obtaining "measurable" corrosion rates on probes exposed to actual furnace
conditions. The objective of this effort was to obtain data on potential
effects of "low NOX" firing conditions on furnace wall tube corrosion rates.
The approach used in obtaining these data was to deliberately accelerate
corrosion on coupons exposed to temperatures in excess of normal tube metal
temperatures. It was decided that exposure for 300 hours at 875°F in
susceptible furnace areas would be sufficient to show major differences in
corrosion rates between coupons exposed to "low NOX" firing conditions and
those'exposed under normal conditions.
276
-------
Aj.chougn there was some scatter in the data obtained, most of the
information was quite consistent. A major finding was that no major
differences in corrosion rates were observed for coupons exposed to "low NOX"
conditions compared to those subjected to normal operation. In fact, for
some probes, the corrosion rates were found to be even lower than for "low NOX"
exposure.
Since corrosion was deliberately accelerated in the corrosion probe
test work in order to develop "measurable" corrosion rates in a short time
period, measured rates, as expected, are much higher than normal wastage
experienced on actual furnace wall tubes. In future tests, coupons will not
be pickled to remove oxide coatings, and coupon temperatures will be reduced
to bring corrosion rates more nearly in line with actual tube wastage.
Much more data are obviously required to resolve the question of
furnace tube corrosion under "low NOX" firing conditions. The limited data
obtained in this program should be helpful in providing evidence that furnace
tube corrosion may not necessarily be a severe "side effect" of low NOX
firing. Long term "low NOX" tests using corrosion probes and the simultaneous
development of actual furnace wall tube corrosion rates by "before" and "after"
ultrasonic thickness determinations are recommended for future studies.
277
-------
5. CONCLUSIONS
The results obtained in this study to date show that modifications
of the combustion operation have a good potential for reducing NOX emissions
from coal-fired utility "boilers without undesirable side-effects. Lowering
the level of excess air and staging the firing of burners resulted in
significant reductions in NOX, averaging about 40-50% for the boilers tested
in the short-term phase of our test programs. The degree of reduction, as
well as the baseline NOX level varied with the type and size of the coal-fired
boiler tested, and presumably also with coal type. In general, tangentially
fired boilers were found to produce the lowest NOX emissions, both under
baseline and modified firing conditions. However, the burner firing patterns
could be changed for wall-fired units without load reduction, resulting in
decreases in NOX emissions of as much as 50% under full load conditions.
There is too little information on the one turbo-furnace unit tested to draw
firm conclusions on the effect of this type of boiler design on NOX emissions.
The NOX emission data were successfully correlated with per cent
stoichiometric air supplied to the burners, for both normal and staged firing
patterns. These correlations show the strong effect of firing coal burners
under net reducing conditions on decreasing NOX emissions. One anticipates
even further improvements if boiler operability problems, particularly slagging
and corrosion can be overcome. In combination with the correlation of baseline
NOX emissions per megawatts generated (or firing rate) per "equivalent firing
wall", these correlations of the present study should be useful for predicting
the level of NOX emitted under normal and staged firing conditions. (Similar
correlations may be developed for gas and oil fired boilers, based on our
data obtained in previous studies.) The correlations suggest that on the
average, about 25% of the chemically bound nitrogen in coal is converted
to NOX. Thus, fuel nitrogen is an important factor in NOX emissions from
coal-fired utility boilers.
Particulate loading and carbon in fly-ash measurements made under
baseline and staged firing, "low NO " conditions, appear to show some increase
for both of these parameters under low NOX" conditions, but in some cases
the opposite behavior has been observed. No extreme differences in flue gas
particulate loadings and in the carbon content of the fly-ash have been found
during our boiler tests.
Under the "low NOX" firing conditions defined during the short-term
optimization tests, 300-hour accelerated corrosion tests have been conducted
on several boilers in this program. Comparison of the accelerated corrosion
rates measured under "low NO " and those measured under normal firing conditions
does not reveal major differences. Therefore, it is recommended that
long-term corrosion tests should be conducted with staged firing of coal on
carefully selected, representative boilers.
Further test work is needed to optimize and demonstrate the promising
NOX control technology based on the results of this work. As mentioned above,
particular attention should be paid to long term corrosion testing. In addition,
boilers fired with different coal types should be tested to define the limitations
imposed at present by slagging problems on the level of substoichiometric air
that can be supplied to the active burners. Based on such information, techniques
for minimizing slagging problems should be developed. Careful control of boiler
operation in additional tests should allow the optimization of NOX emission
control for coal-fired utility boilers.
278
-------
6. REFERENCES
1. W. Bartok, A. R. Crawford, A. R. Cunningham, H. J. Hall, E. H. Manny
and A. Skopp, "Systems Study of Nitrogen Oxide Control Methods for
Stationary Sources," Esso Research and Engineering Company Final Report
GR-2-NOS-69, Contract No. PH 22-68-55 (PB 192 789), November, 1969.
2. Idem, in "Proceedings of the Second International Clean Air Congress"
H. M. England and W. T. Beery, editors, pp. 801-818, Academic Press,
New York, 1971.
3. W. Bartok, A. R. Crawford and A. Skopp, "Control of NO Emissions from
Stationary Sources," Chem. Eng. Prog. 67, 64 (1971). X
4. W. Bartok, A. R. Crawford and G. J. Piegari, "Systematic Field Study of
N0x Emission Control Methods for Utility Boilers," Esso Research and
Engineering Company Final Report No. GRU.4G No. 71, Contract No. CPA 70-90
(PB 210 739), December 1971.
5. Idem, "Systematic Investigation of Nitrogen Oxide Emissions and Combustion
Control Methods for Power Plant Boilers, in "Air Pollution and its Control,"
AIChE Symposium Series, 6J5 (126), pp. 66-74, 1972.
6. W. Bartok, A. R. Crawford, E. H. Manny and G. J. Piegari, "Reduction of
Nitrogen Oxide Emissions from Electric Utility Boilers by Modified Combustion
Operation," presented at American Flame Days," American Flame Research
Committee, Chicago, September, 1972.
7. Gallaer, C. A., "Practical Problems in Efficiency Testing of Large Fly Ash
Precipitators," presented at ASME Winter Annual Meeting, Washington, D.C.,
November, 1971.
279
-------
7. ACKNOWLEDGMENTS
This study was conducted under the sponsorship of the Environmental
Protection Agency, pursuant to Contract No. 68-02-0227. We wish to acknowledge
the active participation of Mr. R. E. Hall, the EPA Project Officer, in planning
the test program and providing coordination with boiler manufacturers and
operators. The cooperation and advice of major U.S. utility boiler manufacturers,
Babcock & Wilcox, Combustion Engineering, Inc., Foster Wheeler Corp., and
Riley-Stoker were essential to planning, and scheduling these tests.
Our thanks are due to the electric utility concerns for their voluntary
participation in making their boilers available for testing. These boiler
operators were the Alabama Power Company, Arizona Public Service, Georgia
Power Company, Gulf Power Company, Tampa Electric Company, Tennessee Valley
Authority, and Utah Power and Light Company. Also, the able assistance of
Messrs. L. W. Blanken, R. W. Schroeder and A. J. Smith in performing these
test programs is acknowledged.
280
-------
APPENDIX
EMISSION FIELD PROGRAM
EPA/OAP Contract No. 68-02-022?
Esso Research and Engineering Company
Recommendations for Selection Criteria
for Field Testing Coal Fired Boilers
I. Design Factors
1. Size: 150 to 1200 MW max. cont. rating, representative of
current and future design practices of boiler manufacturers.
2. Type of firing: tangential, horizontally opposed, front wall,
and cyclone*.
3. Furnace loading.
*i. Furnace design: number of furnaces and/or division walls.
5- Furnace bottom design: wet or dry.
6. Burner configuration: size, number, and spacing.
7. Draft system: pressurized or balanced.
8. Special features available for NO control: NO-ports, flue gas
recirculation into flame zone**. Control of air flows (e.g.,
primary/secondary).
* Cyclone boiler to be tested only if combustion modification flexibility
available.
** Or possibility of diverting existing FGR (now used for steam temperature
control) into flame zone.
II. Boiler Operating Flexibility
1. Excess air: 5$ to 30%. LEA operation (< 15* desirable for
sustained operation).
2. Furnace load with all burners firing: 60* to 100* of MCR.
3. Staged firing: individual burners or rows of burners on air
only; or biased firing of individual burners.
231
-------
h. Flue gas recirculation: location of injection point and
amount recirculated.
5. Windbox pressure: control from low to high, over full range
of furnace load and excess air levels.
6. Combustion air preheat temperature variation.
7. Air register settings.
8. Fuels available: coal types, characteristic of major U.S.
regions*.
* Potential of mixed oil or gas/coal firing for stable staged combustion
may be a desirable feature.
III. Boiler Measurement and Control Capability
1. Fuel rate: by furnace and pulverizer.
2. Air flow rate.
3. Steam temperature control: attemperation water, tilting
burners*, secondary to primary air ratios.
Ij.. Flue gas components monitored by operator: CU, CO, combustibles,
smoke.
5. Steam temperature and pressure, air and flue gas temperatures
for steam side efficiency analysis.
6. Availability of furnace viewing ports for burner flames, slag
build-up, etc. observation.
7. Availability of adequate fuel and flue gas sampling ports.
* For tangential boilers only.
282
-------
IV. Management Operating and Research Policy
1. Management support: make available necessary supervisory,
technical and operating personnel for planning and testing.
2. Research-mindedness: willingness to exploit full boiler
operating flexibility in test program.
3. Willingness to schedule load changes, calibrate boiler
instruments, and bring boiler into proper operating condition
for test program. Cooperation in coal sampling and analysis
desirable.
U. Prior experience in emission test programs.
V. Logistics and Efficiency
Other factors being equal:
1. Select utility and/or station with more than one boiler
meeting criteria.
2, Scheduled annual outage to suit test program schedule.
3. Increased program efficiency by minimizing travel costs.
283
-------
PILOT AND FULL SCALE TESTS
PART II
285
-------
PILOT FIELD TEST PROGRAM TO
STUDY METHODS FOR REDUCTION OF
NOX FORMATION IN TANGENTIALLY COAL
AFIRED STEAM GENERATING UNITS
BY C. E. BLAKESLEE
A. P. SELKER
COMBUSTION ENGINEERING, INC.
FOR PRESENTATION AT
PULVERIZED COAL COMBUSTION SEMINAR
JUNE 19 & 20,1973
SPONSORED BY THE COMBUSTION RESEARCH SECTION OF
THE ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK
NORTH CAROLINA 27711
287
-------
ABSTRACT
This paper describes the work completed on Phase I
of a "Pilot Field Test Program to Study Methods for
Reduction of NOx Formation in Tangentially Coal Fired
Steam Generating Units" performed under the sponsor-
ship of the Office of Air Programs of the Environmental
Protection Agency (Contract No. 62-02-0264). Phase I
of the program consisted of selecting a suitable
utility field steam generator to be modified for experi-
mental studies to evaluate NOx emissions control. This
effort included the preparation of engineering drawings,
a detailed preliminary test program, a cost estimate
and detailed time schedule of the following program
phases and a preliminary application economic study
indicating the cost range of each combustion technique
as applied to existing and new steam generators.
288
-------
INTRODUCTION
The purpose of this program is to investigate various means for
emission control as applied to coal fired utility steam generators.
While current coal firing combustion and control technology have
minimized smoke, CO, hydrocarbon and solid combustible emissions,
proven techniques for the control of NOX have not been fully developed
and evaluated. Review of combustion process modifications which had
been found effective in reducing NOX emissions from oil and gas fired
steam generators and recent staged combustion simulations with coal
firing indicated that gas recirculation to the firing zone and/or
staged combustion should be evaluated as commercially feasible methods
of NOx reduction. For these reasons a program was developed to eval-
uate the feasibility of these as well as other methods of NOX control
on a commercially sized pilot plant unit. This unit would be modified
to incorporate the systems to be studied for evaluation of potential
operating and control problems and the establishment of optimum methods
for both transient and long term operation.
Phase I was conducted as part of a projected five phase program to
identify, develop and recommend the most promising combustion modi-
fication techniques for control of N0xs without objectionable increases
in related pollutants, from tangentially coal fired utility steam gener-
ators. Phase I comprises the following tasks.
Task I - Selection of a suitable tangentially coal fired
unit for emission control modification and testing.
Task II - Preparation of a detailed preliminary test program.
Task III - Preparation of engineering drawings, modification
costs and time schedule.
Task IV - Estimate modification cost ranges for each combustion
modification technique as applied to existing and
new boilers.
DISCUSSION
TaSK I - UNIT SELECTION
To select a test unit, Combustion Engineering conducted a survey of utility
companies using tangentially coal fired steam generators to determine their
interest in participating in the NOX control program. As a result of this
survey seven (7) utility companies expressed a desire to cooperate with CE
in the oroaram These companies were subsequently reviewed to determine if
they had within their generating systems units meeting the remaining criteria
specified for the test unit.
nf <=PVPral units found to be generally acceptable for the test program,
Alabama Power Co.[ Barry Station Unit No. 1 was finally selected.
239
-------
This unit is a natural circulation, balanced draft, steam generator,
firing coal through four elevations of tilting tangential fuel nozzles.
The superheat steam capacity at maximum continuous rating is 900,000 IBS/
Hr main steam flow with a superheat outlet temperature and pressure of
1000 F and 1875 PSI6. Superheat and reheat temperatures are controlled
by fuel nozzle tilt and spray desuperheating. A side elevation of this
unit is shown in Figure 1.
The criteria upon which the selection was based are as follows.
1. The unit is representative of the tangentially coal fired
steam generators currently designed by CE which facilitates
the transfer of technology to existing and new boiler designs.
2. The unit, while representative of current utility boiler
design, is small enough (125 MW) to minimize modification
costs and permit a versatile experimental program. The
control system installation can be coordinated with a
planned outage for installation of a hot electrostatic pre-
cipitator. This precipitator would eliminate the need for
additional dust removal equipment to protect the gas re-
circulation system fan.
3. The unit location permits testing of various coals without
incurring additional coal transportation costs. Coals cur-
rently being burned at the station include both local Ala-
bama and Illinois varieties. The station has existing
facilities for receiving and handling of both rail and barge
coal deliveries.
4. Alabama Power Company had expressed their willingness to
cooperate and participate in this program by making the
unit available for the required modifications and tests.
5. The results of a unit operating survey indicated that Barry 1
is acceptable for the planned experimental NOv control study
modifications. Briefly, unit operating flexibility, ash
handling systems, fan capacities and normal operation NOv
levels were found to be acceptable for the purposes of this
program. A plot of NOx values versus excess air at various
unit loadings is shown in Figure 2.
Task II - DETAILED TEST PROGRAMS
The detailed test programs were developed using a statistical program
design approach. In this manner maximum program efficiency can be
attained by obtaining the maximum informational output from each test.
Using this approach the individual variables considered for evaluation
were first identified and then the minimum number of variable combina-
tions which must be tested to properly evaluate each variable was
established.
The individual variables identified for evaluation in one case were as
follows:
290
-------
Excess Air
Unit Loading
Air Preheat Temperature
Biased Firing
Gas Recirculation to:
a. Secondary Air Ducts
b. Coal Pulverizers
c. Combination of the above.
Overfire Air
Water Injection to the Firing Zone
For the second case, the variables to be evaluated were:
Excess Air
Unit Loading
Biased Firing
Overfire Air
The degree to which each process variable or modification would be applied
and the process measurements necessary to evaluate unit performance follow,
Process Modifications
A. Overfire Air System
The overfire air system was designed to introduce a maximum of
20 percent of full load combustion air above the fuel admission
nozzles through two additional compartments in each furnace cor-
ner located approximately eight feet above the fuel admission zone.
Overfire air can also be supplied to the furnace through the top
two compartments of the existing windbox when the upper elevation
of fuel nozzles is not in use. The overfire air nozzles will
tilt +30° in the vertical plane independently of the main fuel
and air nozzles. Independent dampers for each overfire air com-
partment will be provided as a means to study the influence of
location and velocity of overfire air introduction.
B. Gas Recirculation System
The gas recirculation system was designed to recirculate flue
gas to the secondary air duct and coal pulverizers either
separately or in combination. The system would provide for a
maximum of 40 percent recirculation at 80 percent unit loading
and permit substituting gas recirculation for hot air to the coal
291
-------
pulverizers while introducing tempering air in the conventional
manner. A gas recirculation temperature range from 300 to 650F
would be possible by varying the weight ratio of flue gas taken
from the air preheater gas inlet and outlet.
C. Air Preheat System
The preheated air temperature entering the secondary air duct
can be varied by bypassing the air and/or gas side of the air
preheaters to provide the maximum system flexibility.
D. Hater Injection System
Water injection can be admitted into the furnace through two eleva-
tions of atomizing spray nozzles located between the top two and
bottom two fuel nozzle elevations. A maximum injection rate of 50
pounds per million BID fired can be used.
Process Variables
Excess air, unit load, and fuel and air distribution will be varied within
the current limitations of the existing equipment. These limits were evalu-
ated in the unit operating survey conducted in Task I.
Process Measurements
Operation of the unit as proposed in the experimental study will produce
variations in unit operation and thermal performance. The following process
measurements are required to properly assess the impact of these changes on
new unit design and the retrofitting of existing units.
A. Furnace Absorption
Recirculating gases to the secondary air compartments and
staging of combustion air will effect changes in both peak
and average furnace waterwall temperatures and absorption
rates. The waterwall crown temperatures and absorption rates
must therefore be determined to evaluate the impact of
variations in average and peak rates and absorption profiles
on unit design.
B. Furnace Corrosion Probes
Unit operation with staged combustion air may result in local
reducing atmospheres within the furnace envelope, resulting
in accelerated waterwall corrosion rates. To assess the
impact of this type of operation on waterwall wastage, furnace
corrosion probes will be utilized.
292
-------
C. Sensible Heat Leaving the Furnace
Variations in furnace heat absorption rates due to modifying
the combustion process will result in increasing or decreasing
the sensible heat leaving the furnace envelope and entering
the superheat and reheat sections of the unit. To determine
the sensible heat leaving the furnace, the exit gas temperature
will be measured at the vertical furnace outlet plane using
water cooled probes with radiation shielded thermocouples.
D. Superheat, Reheat and Economizer Section Absorptions
Variations in the gas temperature and gas flow leaving the
furnace envelope and entering the convective sections of the
unit will affect the total heat pickup of each section. To
assess the impact of modified operation on superheat, reheat
and economizer performance, the absorption rates for each sec-
tion will be determined.
Variation in heat absorption rates may require resurfacing when
retrofitting existing units for modified operation.
E. Air Heater Performance
Air and gas temperatures and gas side oxygen concentrations
entering and leaving the air heater are required to calculate
air heater performance, unit efficiency, heat losses and air
and gas flow rates.
F. Fuel and Ash Analysis
During each test, a representative fuel sample must be obtained
for later analysis. The fuel analyses are required to perform
combustion calculations necessary to determine excess air levels
and unit gas and air flow rates. Pulverized coal fineness samp-
les will be obtained to determine the effect, if any, on furnace
wall deposit characteristics, solid combustibles losses, NOx
levels and related emissions.
In addition, coal ash analyses are required to determine ash
properties such as base/acid ratios and ash deformation, soft-
ening and fluid temperatures necessary for evaluating the furnace
wall deposit characteristics of coal fuels. Furnace bottom ash,
fly ash and coal pulverizer rejects analyses are also required
to determine heat losses and material balances. Specific instru-
mentation and methods to be used in measuring these process vari-
ables and the flue gas emission constituents are defined in the
detailed test plan.
293
-------
Task III - ENGINEERING DRAWINGS. COST ESTIMATES AND DETAILED TlMh sCntDUL.E
Engineering Drawings
Arrangement drawings were completed showing necessary duct arrangements
for the overfire air and gas recirculation systems, the overfire air
register arrangements and control system interfaces with the existing
unit. The general arrangement drawings for the ductwork indicate that
the proposed control systems can be physically installed within the ex-
isting station without serious structural interferences.
The modification ductwork final locations were determined by an extensive
design review and engineering field check of actual existing equipment
configurations and locations.
Cost Estimates
The cost of fabricating, installing and testing the overfire air and gas
recirculation systems were estimated both as a complete system and as
individually installed systems. These estimates do not include additional
fuel costs incurred during the test program as Alabama Power Company has
agreed to assume these costs.
Detailed Time Schedules
Due to difficulties encountered in establishing when authorization to pro-
ceed with follow-on program phases would be received, it was not possible
to finalize a detailed time schedule for installation of the control sys-
tems. Schedules based on elapsed time from start of contract were pre-
pared and are shown in Figures 3 and 4. These schedules must be coordina-
ted with a unit outage occurring in the tenth to twelfth program month.
Such an outage is currently available in the spring of 1974.
Task IV - COMBUSTION TECHNIQUE APPLICATION COSTS
Application Study Results
Based on the cost estimates developed under Task III and Combustion Engineer-
ing, Inc. 's current knowledge, cost ranges were developed for applying the
NOx control techniques proposed in this program to new and existing unit
designs. These cost ranges are illustrated in Figures 5 and 6.
Specifically, four possible methods of reducing NOx emission levels from
tangentially coal fired steam generators were evaluated and the cost trends
for each method estimated for both new and existing units. The reduction
methods considered included overfire air, gas recirculation to the secondary
air ducts, gas recirculation to the coal pulverizer/primary air system and
furnace water injection. The cost trends for these methods were projected
over a unit size range of 125 to 750 MW.
294
-------
The results of the study indicate that for any given unit size (450 MW
chosen for an^example comparison) the lowest cost method is found to
be overfire air which results in a .14 to .50 $/KW additional unit cost
for a new or existing unit respectively.
This method incurs no loss in unit efficiency or increased operating
expenses. J K 3
Gas recirculation introduced either through the secondary air ducts or
the coal pulverizers and primary transport air system results in higher
equipment costs than overfire air and requires additional power for fan
operation.
Water injection introduced into the fuel firing zone of the unit is at-
tractive from the standpoint of low initial equipment costs, however,
losses in unit efficiency resulting in increased fuel costs and signi-
ficant water consumption make it the most expensive system to operate.
The use of either gas recirculation or water injection in existing
units could result in a 10 to 20 percent decrease in load capability
due to increased gas flow weights.
The following conclusions were drawn from this study.
1. The lowest cost method for reducing NOX emission levels on
new and existing units is the incorporation of an overfire
air system. No additional operating costs are involved.
2. Gas recirculation either to the windbox or coal pulverizers
is a promising control system but is significantly more costly
than overfire air and requires additional fan power. In ex-
isting units, the necessity to reduce unit capacity to main-
tain acceptable gas velocities imposes an additional penalty.
3. Gas recirculation to the coal pulverizers would cost approxi-
mately 15 percent less than windbox gas recirculation, however,
this method may require increased excess air to maintain ade-
quate combustion.
4. Water injection has initially low equipment costs, but due to
high operating costs resulting from losses in unit efficiency,
is the least desirable of the systems evaluated. This system
may also require reduced unit capacity.
5. In general, the cost of applying any of the control methods
studied to an existing unit is approximately twice that of a
new unit design.
Application Study Design
For the purpose of this study the following five modes of unit operation
were chosen as potentially effective means for the reduction of NOX emis-
sions.
295
-------
2.
The quantities of overfire air, gas recirculation and water injection
selected for the economic evaluation, while reasonable, do not neces-
sarily represent commercially feasible operation or control methods
which would be recommended by Combustion Engineering, Inc.
1. Introducing 20 percent of the total combustion air over the
fuel firing zone as overfire air.
Introducing 30 percent flue gas recirculation through the
secondary air ducts and windbox compartments.
3. Combining the 20 percent overfire air and 30 percent flue
gas recirculation of 1 and 2.
4. Introducing 17 percent flue gas recirculation through the
transport air/coal pulverizer system.
5. Introducing water injection into the fuel firing zone at a
rate of 5 percent of total evaporation.
The economic comparisons of the five NO emission control methods were
based on 1973 delivered and erected costs for the steam generators and
associated equipment.
The cost estimates presented for the revision of existing units were
based on studies performed on units within the 125 to 750 MW size range
including those costs generated under Phase I, Task 3, for the Barry
No. 1 unit. The cost estimates presented for incorporating control
methods in new unit designs were based on Combustion Engineering expe-
rience and current practice for overfire air and gas recirculation
systems.
As can be seen from Figures 5 and 6 the cost ranges for existing units
vary more widely than new units. This is due mainly to variations in
unit design and construction which either hinder or. aid the installation
of a given control system. For example, an overfire air system may be
designed as a windbox extension unless existing structural requirements
and obstructions necessitate installation of a more costly system in-
cluding extensive ductwork and individual air injection ports. The
same condition exists for water injection systems when the need to
maintain unit capacity dictates changes in unit ducting. Except where
noted, all system costs are estimated on a +JO percent basis. The
cost range of the combined overfire air and windbox gas recirculation
system was arrived at as the sum of the cost ranges of the individual
systems. The cost ranges presented for existing units do not include
any changes to heating surface as these changes must be calculated on
an individual unit basis. Due to variations in existing designs,
heating surfaces may increase, decrease or remain unchanged for a
given control method.
At approximately 600 MW, single cell fired furnaces reach a practical
size limit and divided furnace designs are employed. Since a divided
tangentially fired furnace has double the firing corners of a single
296
-------
cell furnace, the costs of windboxes and ducts increase significantly
as shown on Figures 5 and 6. As shown, the costs of overfire air,
windbox gas recirculation and windbox water injection increase from 30
to 50 percent.
In addition to the increased capital costs resulting from including an
NOx control system in a unit design, the increased unit operating costs
must be considered. The increased annual operating costs were determined
for a 100, 450 and 750 MW unit of new design and are shown in Table I.
The equipment costs shown are determined from Figure 5. Using the 450
MW unit as an example at a rate of .14 $/KW results in an increase in unit
capital cost of $63,000. The additional annual fixed charges, fuel and
fan power costs for each of the five NOx control methods studied and the
criteria on which these costs are based are also listed on Table 1.
Again using the 450 MW unit as an example the study indicates that water
injection is the most expensive system to operate at .332 mills/KWHR due
primarily to increased fuel costs resulting from losses in unit efficiency.
The least expensive control system to operate was overfire air at .004 mills/
KWHR with gas recirculation either alone or in combination with overfire
air ranging from .108 to .121 mills/KWHR.
To put these operating costs in perspective, they can be compared to
"average" generating costs presented in Table 1 for various sizes of un-
modified units.
Operating costs were developed only for a new unit design as it is possible
to assume that design parameters would remain unchanged from a unit designed
without NOx controls. However for existing units, gas and air flow rate
changes, increased draft losses and changes in unit load capabilities would
vary to such a degree that each unit would have to be treated individually
regardless of rating and costs would vary to such a degree that they would
not be useful to a general study.
297
-------
ALABAMA POWER COMPANY - BARRY NO. 1
298
-------
ALABAMA POWER COMPANY - BARRY NO. 1
NOX VS. PERCENT EXCESS AIR
CO
o
cc.
X
o
0.6-
0.5-
0.4--
O.3.-
0.2--
0.1
4 Mill Operation
3 Mil! Operation
Overfire Air Operation
LEGEND
Unit Load
A 142 MW
£) 127 MW
0 113 MW
500
400
300
-200
--100
CVJ
o
ro
O
•-D
Q
Q.
Q.
X
O
PERCENT EXCESS AIR
FIGURE 2
29!?
-------
PROGRAM SCHEDULE
FOR EVALUATION OF OVERFIRE AIR, GAS RECIRCULATION, AIR
PREHEAT AND WATER INJECTION SYSTEMS AND EXISTING PROCESS VARIABLES
Program Month
Phase
2
3
4
5
Task
1
2
3
4
5
1
2
1
2
1
Task Description
Prepare Design Drawings for
Fabrication & Erection of NOv
Control Systems
Purchase Equipment & Fabricate
Equipment
Install Test Instrumentation
Perform Baseline Tests
Perform Bias Firing Tests
Deliver Eauipment and Modify
Unit
Final Test Preparation
Conduct Tests
Evaluate Results & Prepare
Final Report
Prepare Application Guidelines
for Minimizing NO*, •
O i— CMfO^t-LoiDr^OOCTtOr— CVJCO^t-inVQ
1
jPurch Fabricate
1
Test RPT
| Test | RPT |
1 1
I
1 Test 1
| Evaluate | Report !
!
FIGURE 3
-------
PROGRAM SCHEDULE.
Phase
FOR EVALUATION OF BIASED AND .OVERFIRE AIR FIRING
AND EXISTING PROCESS VARIABLES
Task
Task Description
Program Month
r— CM o <• to to rx co
Oi— CM w <•
CMCM CM CM CM
Prepare Design Drawings for
Fabrication & Erection of NOX
Control Systems
Purchase Equipment & Fabricate
Equipment
3 Install Test Instrumentation
4 Perform Baseline Tests
5 Perform Bias Firing Tests
Deliver Equipment and Modify
Unit
Final Test Preparation
1 Conduct Tests
2 Evaluate Results & Prepare
Final Report
Prepare Application Guidelines
for Minimizing NOX
|Purch (Fabricate
(Test I RPTJ
| Test! RPT
j jest
\-Evaluate- -f Report—
FIGURE 4
-------
COSTS OF NOX CONTROL METHODS
NEW COAL FIRED UNITS
(INCLUDED IN INITIAL DESIGN)
WIN DBOX GAS RECIRCULATION
azfOVERFIRE AIR
COMBINED
OV£RFIRE AIR AND WINDBOX
GAS RECIRCULATION
p~ RECIRCULATION THRU
MILLS
1SUMDBOX WATER INJECTION
200
300 400 500
UNIT SIZE
(MW)
600
700
800
FIGURE 5
302
-------
COSTS OF NOX CONTROL METHODS
COAL FIRED UNITS
100
(HEATING SURFACE CHANGES NOT
WINDBOX GAS RECIRCULAT10N
OVKRFIRE AIR
200
300
400 500 600
UNIT SIZE
(MW)
700
CONFINED
OVEJRFIRE AIR AND WINDBOX
RECIRCULATION
RECIRCULATION THRU MILLS
ER INJECTION INCLUDING FAN
6 DUCT CHANGES
IR INJECTION WITHOUT FAN
1DUCT CHANGES
800
FIGURE 6
303
-------
TABLE I
1973 OPERATING COSTS OF NOX CONTROL METHODS FOR
NEW COAL FIRED UNITS
SINGLE FURNACE
CONTROL METHOD
MW RATING
EQUIPMENT COSTS 10 $
LJ ANNUAL FIXED CHARGE 10 $
O
ADDITIONAL ANNUAL FUEL
•COST 10J$
ADDITIONAL ANNUAL FAN
POWER COST 10 $
*5
TOTAL ANNUAL COST 10 $
OPERATING COST MILLS/KWHR
OVERFIRE
AIR (20$)
100
31
5
.
5
0.009
450
63
10
...
10
0.004
750
90
14
-__
14
0.003
WlNDBOX
FLUE GAs
RECIRC. (30$)
100
350
56
21
77
0.143
450
1185
190
95
285
0.117
750
1650
264
158
422
0.104
COMBINATION
OF 1 AND 2
100
375
60
21
81
0.150
450
1248
200
95
295
0.121
750
1800
288
158
446
0.110
COAL MILL
FLUE GAS
RECIRC. (17$)
100
300
48
22
70
0.130
450
1015
162
100
262
0.108
750
1425
228
166
394
0.097
WATER
INJECTION
100
160
26
147
13
186
0.344
450
560
90
660
58
808
0.332
750
825
132
1099
97
1328
0.3271
BASED ON: A. DELIVERED AND ERECTED EQUIPMENT COSTS (+ 10$ ACCURACY). EXCLUDING CONTINGENCY AND INTEREST DURING CONSTRUCTION.
B. 5400 HR/YR AT RATED MW AND NET PLANT HEAT RATE OF 9400 BTU/KWHR.
C. 50$i/106BTU COAL COST.
D. $250/HP FAN POWER COST, OR $40/HP PER YEAR.
E. ANNUAL FIXED CHARGE RATE OF 16$.
F. OPERATING COSTS ARE + 10$.
G. DOES NOT INCLUDE COST OF WATER PIPING IN PLANT OR COST OF MAKEUP WATER.
BASE UNIT OPERATING COSTS* FOR COAL FIRED POWER PLANTS EXCLUDING 50% REMOVAL SYSTEMS.
UNIT SIZE MW 100 450 750
OPERATING COST MILLS/KWHR 16.2 13.5 12,6
•INCLUDES 1973 CAPITAL COSTS, LABOR, MAINTENANCE, FUEL COSTS +20$ CONTINGENCY +17$ INTEREST DURING CONSTRUCTION.
-------
CONTROL OF NOX FORMATION IN WALL, COAL-FIRED UTILITY BOILERS:
TVA-EPA INTERAGENCY AGREEMENT
By
Gerald A. Hollinden and Shirley S. Ray
Power Research Staff
Tennessee Valley Authority
Chattanooga, Tennessee
Prepared for Presentation at
Pulverized Coal Combustion Seminar
Sponsored by the Environmental Protection Agency
Research Triangle Park, Worth Carolina
June 19-20, 1973
305
-------
CONTROL OF NOX FORMATION IN WALL, COAL-FIRED UTILITY BOILERS:
TVA-EPA INTERAGENCY AGREEMENT
By
Gerald A. Hollinden and Shirley S. Ray
Power Research. Staff
Tennessee Valley Authority
Chattanooga, Tennessee
ABSTRACT
An agreement has been formed between the Tennessee Valley Authority and
the Environmental Protection Agency to study, on a field utility boiler,
combustion modification techniques to control NOX and related pollutant
emissions from wall, coal-fired utility boilers. This agreement will
provide more accurate and detailed engineering design information on the
application of specific combustion modification techniques and their
effects on NOX and other pollutant emissions, slagging, fouling, corrosion,
and general boiler operation and performance over a longer period of time
than has been possible in other field tests.
306
-------
specific combustion modifications to a wall, coal-fired utility boiler have
on the formation of NOX and related pollutants and to assess their effects
on corrosion, flame stability, slagging, and general boiler performance.
The project will result in a guide for operational control using this
technology to reduce NOX and other emissions under a variety of conditions.
The agreement was drawn under Section lOU of the Clean Air Act as amended.
Duration of the project will be the 12-month period from June 1, 1973, to
June 1,
The tasks that TVA will perform under this agreement include selection of
a suitable unit, preparation of test program, management of the baseline
emissions study and testing program, and evaluation of results. TVA will
also prepare cost estimates for studies relating to the practicality of
utilizing "NO" ports, flue gas recirculation system, alternate fuels, and
combinations of these techniques for reduction of NOX emissions.
Unit Selection
A wall-fired pulverized coal utility boiler will be selected to satisfy
the following criteria:
1. Representative of current design and use.
2. Have boiler size such as to minimize modification costs, probably from
100 to 250 MW.
3. Permit testing of various coals at reasonable costs.
h. Have capability for flue gas recirculation.
5. Have demonstrated operational flexibility to permit evaluation of the
combustion modifications to be studied (e.g., pulverizer capacity).
307
-------
Test Program
A test program will be designed to characterize normal boiler performance
and to evaluate specific process variables. Baseline emissions and thermal
performance will be determined, as well as the operating conditions within
which the unit can be operated reliably. Baseline corrosion rates will be
established also as a necessary reference to which corrosion data from
experimental studies can be compared. Among the process variables to be
evaluated through the test program are excess air level, load, effect of
furnace wall deposits, and firing patterns for staged firing.
Staged firing will be particularly tested as a control technique for
reducing WOX and related pollutant emissions. The number and duration of
these tests will be sufficient to permit characterization of the effects
that staged firing has on emissions as well as its long-term effects on
corrosion, flame stability, and thermal performance.
In optimizing this technique, TVA will evaluate such factors as:
1. Maximum emissions control throughout the normal load range.
2. Maximum emissions control at full load only.
3. Control of emissions to meet and maintain emissions standards through-
out the normal load range. This may require varying levels of control
for different loads to maintain a fixed emissions level.
The results of these tests should provide guidelines as to when combustion
modifications must be employed to meet promulgated standards. For example,
at reduced loads, the degree to which staged firing is necessary may be
less than at higher loads.
308
-------
The objective of these tests is to develop guidelines as to the levels of
pollutant reductions which are possible with staged firing, the load
conditions under which staged firing must be employed to meet emissions
standards, ard the long-term effects that staged firing has on corrosion
and boiler performance.
TVA will provide for use in the test procedures continuous monitoring
instruments to measure NOX, 02, CO, CC>2, and hydrocarbons, as well as
particulate monitoring equipment, corrosion probes, and the instrumentation
required to characterize unit performance.
Cost Estimates
In addition to the preparation, conduct, and evaluation of the test program
to determine the effects of combustion modifications on pollutant emissions
and on boiler performance, TVA will provide cost analyses for studies to
determine the effects of "NO" ports, flue gas recirculation, combinations
of techniques, and alternative fuels on WOX emissions and boiler performance.
A. Effects of "WO" Forts
Costs and a timetable will be determined for the following tasks:
1. Design, construct, and install tilting "WO" ports above the top row
of burners. These ports will be sized to allow introduction of up
to 25 percent of the total combustion air and will permit adjust-
ment of the discharging air velocity and temperature (through air
preheater control).
2. Conduct a study to optimize the use of "HO" ports for control of
NOX emissions consistent with reliable boiler performance, consider-
ing rate, velocity, temperature, and tilt of overfire air.
309
-------
3. Operate the unit under the optimum conditions of overfire air
firing for at least 300 hours in order to assess the effects of
overfire air firing on fireside corrosion. In determining costs,
consideration should be given to operating the unit at other than
the optimum condition (in terms of pollutant emissions), since
the optimum condition for emissions control may not be best with
regard to corrosion, slagging, stability, and boiler performance.
k. Operate the unit under the optimum conditions of overfire air
firing for four to six months in order to study long-term corrosion
effects.
For each testing phase, the costs associated with emissions testing,
fuel and corrosion probe analysis, boiler derating, outages, and data
reduction will be included in the required cost estimates.
B. Effects of Flue Gas Recirculation
Costs and time needed to complete the following tasks will be determined.
1. Design, construct, and install a flue gas recirculation (FGR)
system. This system will be sized to allow up to l±0-percent
recirculation at 80-percent load, will permit the use of recircu-
lated gas as both transport air to the coal mills and as secondary
air, and will provide for mixing hot air with flue gas to supply
the coal mills and secondary air compartments.
2. Conduct a study to determine the optimum use of recirculated flue
gas for control of NOX emissions consistent with reliable boiler
performance. This study will evaluate the location where flue gas
is introduced and provide for at least two rates of recirculation
and three flue gas temperatures.
310
-------
3. Operate the unit under the optimum conditions of flue gas recircu-
lation for at least 300, hours in order to assess the effects of
flue gas recirculation on fireside corrosion.
k. Operate the unit under the optimum conditions of flue gas recircu-
lation which are consistent with reliable.boiler performance for
four to six months in order to determine long-term corrosion
effects.
Costs for each testing phase will include the costs of emissions testing,
fuel and corrosion probe analysis, boiler derating, outages, differential
operating costs, and data analysis.
C. Combination of Techniques
Costs and time will be determined for an optimization study, an operation
phase of at least 300 hours, and a long-term corrosion study of two to
three months. The optimization study will evaluate the most effective
combination of overfire air, staged firing, and flue gas recirculation
with regard to HOX emissions and boiler performance. It will also
determine the minimum level of excess air which can be achieved, the
effects of low air preheat, and the effects of load variation consider-
ing flue gas recirculation, staged firing, and overfire air.
D. Evaluation of Alternate Fuels
The costs and time needed to evaluate the effects of coal type on NOX
emissions when the most effective combination of combustion techniques
is employed will be determined. At least two different types of coal
supplies will be evaluated. These tests will define the operating
limits within which these fuels may be burned to achieve reduced NOX
311
-------
emissions and reliable boiler performance. Costs data will include
differential fuel and operating costs, corrosion probe analysis, emissions
testing, derating, outages, and data reduction and correlation.
312
-------
BIBLIOGRAPHIC DATA
SHEET
1. Report Not
EPA-650 72-73-021
4. Title and Subtitle
Proceedings Coal Combustion Seminar, June 19-20, 1973
3. Recipient's Accession No.
Research Triangle Park, N.C. 27711
5. Report Date
September 1973
6.
7. Author(s)
R.E
D W. Pershing (Chairman and Vice Chairman)
,. • _ M*i mo *» n/^ Afinrf*ss
8. Performing Organization Kept.
No.
• --.".*-.*. u-*"rT "- '* - ....
9. Performing Organization Name and Address
Miscellaneous
10. Project/Task/Work Unit No.
Pgm Element 1A2014
11. Contract/Grant No.
12. Sponsoring Organization Name and Address
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, N.C. 27711
13. Type of Report & Period
Covered
Proceedings
14.
5. Supplementary Notes
6. Abstracts Tne proceedings document the 10 presentations made during the Seminar,
which dealt with subjects related to EPA's research and development activities for
control of air pollutant emissions from the combustion of pulverized coal. The
Seminar was divided in two parts: participating in the portion on fundamental
research were Rockwell Inter national's Rocketdyne Division, KVB Engineering, Inc.
and Southern California Edison Co. , EPA, Holland's International Flame Research
Foundation, and Jet Propulsion Laboratory; and taking part in the portion on pilot-
and full-scale tests were Babcock and Wilcbx"(Alliance Research Center), U.S.
Bureau of Mines, Esso Research and Engineering Co. , Combustion Engineering, Inc.
and Tennessee Valley Authority. Purpose of the Seminar was to provide contractors
and industrial representatives with the latest information on coal combustion
research.
7. Key Words and Document Analysis. 17a.
Air Pollution
Combustion
Combustion Control
Combustion Chambers
Coal
Nitrogen Oxides
Carbon Monoxide
Carbon
Hydrocarbons
7b. Identifiers/Open-Ended Terms
Air Pollution Control
Stationary Sources
Unburned Hydrocarbons
Fuel Nitrogen
Fundamental Research
c. COSATI Field/Group 13A , 13B , 21B
Descriptors
Pulverized Fuels
Boilers
Utilities
Pilot-Scale Tests
Full-Scale Tests
Availability Statement
Unlimited
19..Security Class (This
Report)
UNCLASSIFIED
20. Security Class (This
Page
UNCLASSIFIED
21. No. of Pages
319
22. Price
'RM NTIS-35
- 3'72)
313
USCOMM-DC M952-P72
-------
INSTRUCTIONS FOR COMPLETING FORM NTIS-35 (10-70) (Bibliographic Data Sheet based on COSATI
Guidelines to Format Standards for Scientific and Technical Reports Prepared by or for' the Federal Government,
PB-180 600).
1. Report Number. Each individually bound report shall carry a unique alphanumeric designation selected by the performing
organization or provided by the sponsoring organization. Use uppercase letters and Arabic numerals only. Examples
FASEB-NS-87 and FAA-RD-68-09.
2. Leave blank.
1 Recipient's Accession Number. Reserved for use by each report recipient.
4. Title and Subtitle. Title should indicate clearly and briefly the subject coverage of the report, and be displayed promi-
nently. Set subtitle, if used, in smaller type or otherwise subordinate it to main title. When a report is prepared in more
than one volume, repeat the primary title, add volume number and include subtitle for the specific volume.
5. Report Date, l-.ach report shall carry a date indicating at least month and year. Indicate the basis on which it was selected
(e.g., date of issue, date of approval, date of preparation.
6. Performing Organization Code. Leave blank.
7. Author(s). Give name(s) in conventional order (e.g., John R. Doe, or J.Robert Doe). List author's affiliation if it differs
from the performing organization.
8. Performing Organization Report Number. Insert if performing organization wishes to assign this number.
9. Performing Organization Name and Address. Give name, street, city, s.tate, and zip code. List no more than two levels of
an organizational hierarchy. Display the name of the organization exactly as it should appear in Government indexes such
as USGRDR-I.
10. Project/Task/Work Unit Number. Use the project, task and work unit numbers under which the report was prepared.
II. Controct/Grant Number. Insert contract or grant number under which report was prepared.
12- Sponsoring Agency Name and Address. Include- zip code.
13. Type of Report and Period Covered. Indicate interim, final, etc., and, if applicable, dates covered.
14. Sponsoring Agency Code. Leave blank.
15. Suoplementory Notes. Enter information not included elsewhere but useful, such a.c: Prepared in cooperation with . . .
Translation of ... Presented at conference of ... To be published in ... Supersedes . . . Supplements . . .
16. Abstract. Include a brief (200 words or less) factual .summary of the most significant information contained in the report.
If the report contains a significant bibliography or literature survey, mention it here.
17. Key Words and Document Analysis, (a). Descriptors. Select from the Thesaurus of Engineering and Scientific Terms the
proper authorized terms that identify the major concept of the research and are sufficiently specific and precise to be used
as index entries for cataloging.
(b). Identifiers and Open-Ended Terms. Use identifiers for project names, code names, equipment designators, etc. Use
open-ended terms written in descriptor form for those subjects for which no descriptor exists.
(c). COSATI Field/Group. Field and Group assignments are to be taken from the 1965 COSATI Subject Category List.
Since the majority of documents are multidisctplinary in nature, the primary Field/Group assignment(s) will be the specific
discipline, area of human endeavor, or type- of physical object. The applications) will be cross-referenced with secondary
Field/Group assignments that will follow the primary posting(s).
18. Distribution Statement. Denote rc-leasability to the public or limitation for reasons other than security for example "Re-
lease unlimited". Cite any availability to the public, with address and price.
19 & 20. Security Classification. Do not submit classified reports to the National Technical
21. Number of Pages. Insert the total number of pages, including this one and unnumbered pages, but excluding distribution
list, if any.
22 Price. Insert the price set by the National Technical Information Service ot the Government Printing Office, if known.
FORM NTIS-35 (REV. 3-72) USCOMM-DC 14952-P72
------- |