Environmental Protection Technology Series Proceedings of the Stationary Source Combustion Symposium Volume I--Fundamental Research


EPA-600/2-76-152a
June 1976
Environmental Protection Technology Series
       PROCEEDINGS OF THE STATIONARY  SOURCE
                            COMBUSTION SYMPOSIUM
                                               Volume I
                                Fundamental Research
                                 Industrial Environmental Research Laboratory
                                      Office of Research and Development
                                     U.S. Environmental Protection Agency
                               Research Triangle Park, North Carolina 27711

-------
               RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
     1.   Environmental Health Effects Research
     2.   Environmental Protection Technology
     3.   Ecological Research
     4.   Environmental Monitoring'
     5.   Socioeconomic Environmental Studies

This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and norj-point sources of pollution. This
work provides the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
                    EPA REVIEW NOTICE

This report has been reviewed by the U. 8. Environmental
Protection Agency, and approved for publication.  Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                                        EPA-600/2-76-152a
                                                        June 1976
                         PROCEEDINGS OF  THE

          STATIONARY SOURCE  COMBUSTION  SYMPOSIUM

                    VOLUME I--FUNDAMENTAL RESEARCH
                          JoshuaS. Bowen, Chairman
                        Robert E. Hall, Vice-Chairman

                  Industrial Environmental Research Laboratory
                    Office of Energy, Minerals, and Industry
                       Research Triangle Park, NC 27711
vSi
     ROAPNo.  21BCC
Program Element No. 1AB014
X
to
                                Prepared for

                 U.S. ENVIRONMENTAL PROTECTION AGENCY
                       Office of Research and Development
                            Washington, DC 20460
                                          . .;;;K7;GM AGENCY
                                ii. j>

-------

-------
                               PREFACE
     The Stationary Source Combustion Symposium was held on September




24-26, 1975, at the Fairmont Colony Square Hotel in Atlanta, Georgia.




The symposium was sponsored by the Combustion Research Branch of E.P.A.'s




Industrial Environmental Research Laboratory (IERL).   The Combustion




Research Branch has been involved in developing improved combustion




technology for the reduction of air pollutant emissions from stationary




sources, and improving equipment efficiency.









     Dr. Joshua S. Bowen, Chief, Combustion Research Branch, was Symposium




Chairman; Robert E. Hall, Research Mechanical Engineer, Combustion Research




Branch, was Symposium Vice Chairman and Project Officer.  The Welcome




Address was delivered by Dr. John K. Burchard,  Director of the Industrial




Environmental Research Laboratory.  Frank Princiotta, Acting Director of




the Energy Processes Division of E.P.A.'s Office of Energy, Minerals,




and Industry, was the Keynote Speaker.









     The Symposium consisted of four Sessions:




     Session I;  Fundamental Research




     Co-chairmen:  Dr. Joshua A. Bowen




                   W. Steven Lanier, Research Mechanical Engineer, E.P.A.,




                       IERL, Combustion Research Branch
                                   iii

-------
     Session II:  Fuels Research and Development




     Chairman:  G. Blair Martin, Chemical Engineer, E.P.A., IERL,




                   Combustion Research Branch








     Session III;  Process Research and Development




     Chairman:  David G. Lachapelle, Research Chemical Engineer,




                  E.P.A., IERL, Combustion Research Branch








     Session IV;  Field Testing and Surveys




     Co-chairmen:  Robert E. Hall




                   John H. Wasser, Research Chemical Engineer, E.P.A.,




                      Combustion Research Branch








     These Session Chairmen have reviewed the transcriptions of the




question and answer sessions, and, in addition, have worked with authors




to clarify and revise presentations, where appropriate, and to make them




clear and meaningful for these printed proceedings.








      We are grateful for the cooperation of Marjorie Maws,  Project Leader;




 Anita Lord, Symposium Administrator;  and Margaret Kilburn,  Program Director,




 of Arthur D.  Little, Inc.,  who coordinated the symposium for E.P.A.
                                  IV

-------
                                  CONTENTS
Preface ....

Welcome Address

Keynote Address
                                                                        ill
                  J.K. Burchard	1-1

                  F.T. Princiotta  	  ..... 1-3


                      SESSION  I  - FUNDAMENTAL RESEARCH

Formation of Soot and Polycyc.lic Aromatic Hydrocarbons  (PCAH)
in Combustion Systems 	 1-18

     J.D. Bittner, G.P. Prado, J.B. Howard, R.A. Kites

Questions and Answers	1-32


Effects of Fuel Sulfur on Nitrogen Oxide Emissions	1-35

     J.O.L. Wendt, J.M. Ekmann

Questions and Answers 	 1-88


Two-Dimensional Combustor Modeling  	 1-91

     R.C. Buggeln, H. McDonald

Questions and Answers 	 (n/a)


Effects of Interaction Between Fluid Dynamics and Chemistry on
Pollutant Formation in Combustion 	 1-109

     C.T. Bowman, L.S. Cohen, L.J. Spadaccini, F.K.  Owen

Questions and Answers 	 1-119


Fate of Coal Nitrogen During Pyrolysis and Oxidation  	 1-125

     J.H. Pohl,  A.F.  Sarofim

Questions and Answers	1-147

-------
A Detailed Approach to the Chemistry of Methane/Air Combustion:
Critical Survey of Rates and Applications  	 1-153

     V.S. Engleman

Questions and Answers	I-182


Chemical Reactions in the Conversion of Fuel Nitrogen to NOx  	 1-185

     A.E. Axworthy, G.R. Schneider, V.H. Dayan

Questions and Answers 	 1-211


Prediction of Premixed Laminar Flat Flame Kinetics, Including
the Effects of Diffusion	1-217

     R.M. Kendall, J.T. Kelly, W.S. Lanier

Questions and Answers 	 1-266


Estimation of Rate Constants	1-267

     S.W. Benson, R. Shaw, R.W. Woolfolk

Questions and Answers	(n/a)


Production of Oxides of Nitrogen in Interacting Flames  	 1-291

     C. England

Questions and Answers 	 1-316


Concurrent Panel Discussions:

1. Combustion Chemistry and Modeling: An Overview

   A.F. Sarofim - Combustion Chemistry and Modeling	1-325

   Questions and Answers	I- 345


   T.J. Tyson - The Mathematical Modeling of Combustion Devices .... I- 347

   Questions and Answers	1-410
                                    vi

-------
2. Federal, Regional, State, and Local Air Pollution
   Regulations: An Overview

   S. Cuffe - Federal	I-4I3
   G.T. Helms - Regional	1^429
   R.H. Collum - State	1-443
   H.W. Poston - Local	1-447
               SESSION II - FUELS RESEARCH AND DEVELOPMENT


Assessment of Combustion and Emission Characteristics of
Methanol and Other Alternate Fuels  	 II-3

     G.B. Martin

Questions and Answers 	 ....... II-3C


Burner Design Criteria for Control of Pollutant Emissions
from Natural Gas Flames	11-31

     D.F. Shoffstall

Questions and Answers 	 (n/a)


Integrated Low Emission Residential Furnace 	 11-81

     L.P. Combs, W.H. Nurick, A.S. Okuda

Questions and Answers 	 . 	 11-101


The Control of Pollutant Emissions from Oil Fired
Package Boilers 	 11-109

     M.P. Heap, T.J. Tyson, E. Cichanowicz, R.E. McMillan, F.D. Zoldak

Questions and Answers 	 ... 11-160


Pilot Scale Investigation of Catalytic Combustion Concepts for
Industrial and Residential Applications 	 .  	 II-163

     J.P. Kesselring, R.M. Kendall,  C.B. Moyer, G.B. Martin

Questions and Answers	'11-196
                                      vii

-------
The Optimization of Burner Design Parameters to Reduce NOx
Formation in Pulverized Coal and Heavy Oil Flames 	 11-197

     M.P. Heap, T.J. Tyson, G.P. Carver, G.B. Martin, T.M. Lowes

Questions and Answers 	 11-239


Pilot 'Scale Investigation of Combustion Modification Techniques
for NOx Control in Industrial and Utility Boilers	11-241

     R.E. Brown, C.B. Moyer, H.B. Mason, D.G. Lachapelle

Questions and Answers	II- 267
             SESSION III - PROCESS RESEARCH AND DEVELOPMENT


Overfire Air as an NOx Control Technique for Tangential
Coal-Fired Boilers	Ill-3

     A.P. Selker

Questions and Answers 	 Ill-26


Control of NOx Formation in Wall Coal-Fired Boilers 	 Ill-31

     G.A. Hollinden, J.R. Crooks, N.D. Moore, R.L. Zielke,
     C. Gottschalk

Questions and Answers 	 Ill-77


The Effect of Additives in Reducing Particulate Emissions
from Residual Oil Combustion  	 Ill-33

     R.D. Giammar, H.H. Krause, A.E. Weller, D.W. Locklin

Questions and Answers	 •	Ill-115


System Design for Power Generation from Low Btu Gas Boilers 	 III-119

     M.P. Heap, T.J. Tyson, N.D. Brown

Questions and Answers 	 (n/a)
                                       viii

-------
                  SESSION IV - FIELD TESTING AND SURVEYS


The Effect of Combustion Modification on Pollutants and
Equipment Performance of Power Generation Equipment	I.V-3

     A.R. Crawford, E.H. Manny, M.W. Gregory, W. Bartok

Questions and Answers 	 IV-109


Analysis of Gas-, Oil-, and Coal-Fired Utility Boiler
Test Data	tV-115

     O.W. Dykema, R.E. Hall

Questions and Answers 	 ... IV-161


Influence of Combustion Modifications on Pollutant Emissions
from Industrial Boilers 	 IV-163

     G.A. Cato, L.J. Muzio, R.E. Hall

Questions and Answers	IV-219


Emission Characteristics of Small Gas Turbine Engines 	 IV-227

     J.H. Wasser

Questions and Answers 	 "V-252


Systems Evaluation of the Use of Low-Sulfur Western Coal in
Existing Small- and Intermediate-Sized Boilers	 ::v-255

     K.L. Maloney

Questions and Answers	".V- 316


A Survey of Emissions Control and Combustion Equipment Data in
Industrial Process Heating  	 IV-321

     P.A. Ketels, J.D. Nesbitt, D.R. Shoffstall, M.E.  Fejer

Questions and Answers 	 (n/a)
                                        ix

-------
POM and Particulate Emissions from Small Commercial
Stoker-Fired Boilers 	
IV-411
     R.D. Giammar, R.B. Engdahl, R.E. Barrett

Questions and Answers  	
IV-4 39
Concluding Remarks - J.S. Bowen	IV-441
Appendix
      List of Speakers 	
      List of Participants
              Alphabetically by Name ....
              Alphabetically by Organization
A-l

A-2
A-6

-------
                 WELCOME ADDRESS

U.S. E.P.A. STATIONARY SOURCE COMBUSTION SYMPOSIl
                September 24, 1975

                 Atlanta, Georgia
               Dr.  John K.  Burchard
           Director, E.P.A. Industrial
        Environmental Research Laboratory
                        1-1

-------
     I see by the program that I have just given the welcoming speech.
I think one of the chief virtues of welcoming speeches is that they be
brief, so 1 will be.
     On behalf of the Environmental Protection Agency and particularly
the Industrial Environmental Research Laboratory, I would like to
welcome you to this symposium.  1 still have trouble saying 1ERL,
because up until our recent reorganization we were known as the Control
Systems Laboratory, but we are basically the same bunch of people.  Our
mission now is enlarged slightly in that whereas we used to be strictly
confined to air pollution activities, we will now be involved in certain
water pollution and solid waste activities, so the scope of our activities
has broadened some.  I would like to emphasize that we want to engage in
technology transfer in these kind of symposia.  We'd like to tell you
what we're doing, what people that we are sponsoring are doing, and we'd
like to get feedback from you on questions you have and things that you
would like to know more about.  There is a special questionnaire in the
program booklet for this kind of thing.
     As is shown in the program, Day One will focus primarily on funda-
mental research and development.  The second day will be on fuels research
and development and process R&D, and the last day will focus on field
testing surveys.  At the end of today there will be a concurrent panel
discussion, and 1 hope you can decide on which one to attend.  One is
on air pollution regulations and the other on combustion chemistry.
     Well, as I said, I'd like to keep it brief.  Again, it's a pleasure
to have you here.  We, after all, feel that the major purpose of all our
work is to try to spread the results of our work amongst the technical
and user community and those are you people; so, I hope this will be a
worthwhile and productive few days.
     Thank you.
                                   1-2

-------
         Keynote Address at EPA's
  Stationary Source Combustion Symposium
             Atlanta, Georgia
            September 24, 1975
            Frank T. Princiotta
Acting Director, Energy Processes Division
  Office of Energy, Minerals & Industry
     Office of Research & Development
     Environmental Protection Agency
             Washington, D.C.

                    1-3

-------
    I welcome you to the Industrial Environmental Research Laboratory's




 (IERL) Stationary Source Combustion Symposium.  I am pleased that I have




been asked to present the Keynote address, and trust that ray remarks will be




relevant and informative.  I believe the most fruitful approach for my




10-minute address is to briefly explain the EPA research and development




organization and show where EPA-sponsored stationary source combustion efforts




fit in.  Also, even more importantly, I will touch upon the status of




EPA's NOX emission control strategy, since this represents the driving




forces for utilization of much of the technology we will hear about at this




Symposium.




     Recently, the EPA's Office of Research and Development has been




reorganized under the direction of the newly appointed Assistant Administrator




(AA), Dr. Wilson Talley, and his Associate Assistant Administrator,




Carl Gerbcr.  Briefly, the new organization calls for each of the 15




field laboratories to report directly to one of the four headquarters




Office Directors, who in turn report directly to Dr. Talley.  Figure 1




shows each office and its performing laboratories.




     Generally, the headquarters staff performs strategic planning with




emphasis on resource allocation by major technical category; the laboratory




implements the research and development program through extramural and




intramural efforts.  As you can see, the IERL-RTP reports to Dr. Stephen Gage's




Office of Energy, Minerals and Industry.  This office has two headquarters




staff Divisions:  the Energy Processes Division of which I am Acting




Director, and the Industrial and Extractive Processes Division under the






                                 1-4

-------
directorship of Dr. Peter Ledennan.  As you may have guessed by




now, all control technology development work for energy and industrial




sources falls within the responsibility of OEMI and its two 1ERL




laboratories.  Generally, the assignment of R,D&D implementing responsi-




bilities between the two IERL laboratories is based on assigning specific




industries (and their pollutants) to each laboratory.  Since 1ERL-RTP




has been assigned the utility and industrial boiler "industries,"




essentially the total stationary source combustion control technology




program is performed under their auspices.  As many of you know, Josh Bowen's




Combustion Research Branch reporting to Bob Hangebrouck's Energy Assessment




and Control Division is the group within IERL-RTP responsible for the




stationary source combustion program.




     It is important to note that the funding for the NOX control technology




part of EPA's program is based upon a special Congressional energy




appropriation which is supplemental to the Office of Research and




Development's Base Program.  This supplementary appropriation,which was




initiated in fiscal year 1975, includes funding for a comprehensive




energy/environmental research and development program involving control




technology development and process and effects research related to




conventional and second-generation energy sources.  The energy appropria-




tions was $134 million in FY 1975 of which $81 million was for control




technology development.  In fiscal year 1976, the funding leveal was




reduced to $100 million of which $59 million is for control technology




development.  We are hopeful that the FY 77 appropriation will be close to
                                    1-5

-------
the FY 76 funding levels.  Despite the reduction in control technology

funding from FY 75 to FY 76, energy funding for the NOx control

technology program increased from $3.8 million in FY 75 to $7.7 million

in FY 76.  Further increases in funding are considered likely.  Funding

will continue to increase in this area since NOX control from stationary

sources is growing in importance.  This is an appropriate lead-in to the

next subject 1 would like to briefly discuss—the overall EPA strategy—

attaining acceptable ambient air quality with regard to N02 concentrations.

     As many of you know, there has not been substantial progress made toward

attainment of the national ambient air quality standard for N(>2 since

it was discovered in 1972 that the Federal reference method (Jacobs-Hochheiser

technique) for measuring ambient N02 was unreliable.  As a result of this

discovery, 43 of the 47 Air Quality Control Regions which were originally

thought to have a serious N02 problem (Priority I) were reclassified to

Priority III, since alternative* NC>2 measurement techniques indicated that

standards were not exceeded in these regions.  This left Los Angeles,

New York, Chicago and Salt Lake City AQCR's which are presently classified

as Priority I.  However, recent data which the EPA Administrator presented

at his May 1975 news conference indicate that now at least 16 AQCR's are

presently exceeding the NC-2 standard and that several other regions may
*  EPA has activated a schedule which will designate reference N02 measure-
ment methods by June 1976; this should help clarify the N02 situation and
allow a more aggressive NOx control strategy to be undertaken.
                                    1-6

-------
have difficulty in maintaining the NC>2 standard.  It is apparent that the




N02 problem is getting more serious, and more stringent stationary and mobile




source NOx control will be required.




     At the present time, only modest KOX control emission standards




have been applied to both new mobile and stationary sources on a national




basis.  It appears  that the existing 3.1 gin/mile emission standard for




automobiles and New Source Performance Standards (NSPS) for coal, oil




and gas-fired steam generators and nitric acid plants will not be




adequate to meet and maintain ambient air quality standards in many parts




of the country.  Presently, additional NSPS for additional sources are




being considered and appear likely for:  Stationary gas turbines, lignite




steam generators, stationary internal combustion engines and intermediate-size




coal, oil and gas-fired boilers.  But again, these additional NSPS, even




if combined with a more stringent (2.0 gm/mile) auto standard, may not




provide an adequate level of control for many regions.




     The most definitive document evaluating various NOx mobile and




stationary control strategies which I am familiar with, is the Air Quality,




Noise and Health Panel Report of the Department of Transportation Study on




Mobile Source Goals Beyond 1980.  This report is only in a draft version and




is not yet available for publication.  The basic analytical approach utilized




was to estimate the incremental cost of various levels of emission controls




for a variety of mobile and stationary sources needed to decrease emissions




from assumed 1980 levels.  The incremental costs are then used to determine




the cost-effectiveness of each level of control.  It was assumed that




baseline emissions in 1980 vould be as follows:






                                    1-7

-------
Source

light-duty vehicles

heavy-duty vehicles

stationary sources
   /


       Total
 1980 Emissions
1980 Control Technology Assumed
 3.8 x 10& tons N0x/yr   2.0 gin/mile (catalytic converter)

 4.8 x 106               10.0 gm/mile
25.5 x 106
  Based on NSPS constraints, an
  average of 25% emission control
  from all sources
34.1 x 106
     Based on these baseline emissions, and the best available information

regarding the effectiveness and cost for various second-generation mobile

and stationary control technologies, these technologies were evaluated for

their cost-effectiveness and impact on NOX emissions.  Table 1 summarizes

the results of this exercise.

     The implications of this study are as follows:

     *  The most cost-effective NOx control approaches, which can control

        up to 27% of baseline emissions, involve second-generation combustion

        modification techniques for utility and industrial boilers and

        stationary 1C engines.  Such control can be achieved for an invest-

        ment of only $100-225/ton of NO  removed.
                                       X

     •  The cumulative control of baseline emissions can be increased

        from 27 to 44% with heavy-duty vehicular control tightened from

        10 gm/mile to 2 gm/mile and for automobile control tightened to

        1.0 gm/mile.  Such incremental control will cost about$450/ton.

     *  The next major reduction of NOX emissions can be attained by

        more stringent control of utility boilers using the expensive
                                    1-8

-------
         flue gas cleaning techniques estimated at $1200/ton.  Such




         technology used in conjunction with the aforementioned




         technique can yield overall cumulative control of baseline




         emissions to 63%.




      •  Further emission control beybnd the utilization of flue




         gas cleaning may be prohibitively expensive and will only




         increase overall control from 63 to 69%.  This would require




         increasing control from stationary 1C engines from 75 to 90%




         control ($1700/ton) and further control of auto emissions to




         0.4 gm/mile.






     The implications of this study are displayed graphically in Figure 2




using cost as a function of tons of NOX removed.  Three categories are




shown:  (1) a mobile source approach; (2) a stationary source approach;




and  (3) most cost-effective approach for combined stationary and mobile




source control.




     Although such a study is only as good as the many assumptions made




regarding effectiveness and cost of second-generation control technologies,




I believe it clearly indicates the major role that stationary source




control will play in the years ahead.  I have no doubt that many of the




control techniques described at this Symposium will help us achieve acceptable




N02 ambient levels.
                                  1-9

-------
o
u
§M
c
W O
a- M
Sf&
Z «•«.
V-*
J
S
H
en
85
j e
< o
Z °IM
§3
1 X)
j C.
8
R5G
o tu
-SI
S65
Z •**
§ tu ,0
5 o —
W »— s
O o
UJ -M
U. «9
(U S^r
H
8
0
U. CO g; SO
Z O Z
05 o *•*
ill .5 «• J
rH »} Ul
u co o J tg
ru I-H o P <
P- S o u eo
HC w CM

1-1 t- >«i:
Ko" w
« Z !U Z
•1 BU S- >t *
.O > U. «J •
a » O U U3 CO
H&-42 2§
ui O os on »-•
tt. eg S tfl
(X P Q U. «
iu 2 05
i o £
f 	 ?% "v* U3
r* ^,j «^c •**
0
Q
cu *-^
is
52
fe


>• /-^»
j G a
u u 3
£§:
%^-

^6
O CZJ
u cn



rt 0 <* « M M * ^. "°.
« ^*^22SSS



g Rsaassss*
o oo-«-*2t32




ss s 2 a s a a. * s.
o 0'd«a«or-rHfM


a a a s 5 s a g 8





^«s^?sss
«-H r^ «~* ^^





0 ^CMOOrH^WM"
rH "^






j^t r^ CM rH CM vO m
S $S£*S-o»-
„; ^^N^^*O-
^-^ r^
0£ Psm» O •> » O
oo o gg Oc. eeo~^
s-1* "• t 15 'HI «s -«
«-.g < S «t 51 -S§5 55 S*. ^§2
3«£,fe S S z« uu  * «
s n ^s ^
Is -a -2 IS ^o 2 «a is ^
O01" a * .j, .5 0 G S 0 rl -I *•
>vo ^o -g «2 sis «5 5S -?2
-** -- S" &« -§as B-g -s- w^. o
-H*» .M** 36 t)fS— so 0-^ 03O— om-4
-m— r-m^i g-r HS^ Sd«> °B So K r- o o
^^S tt^S ^^2 2 UO U.M • _ r. >. *•
O *H I/} t« *I. .nil dJ00O rH6*J '6*^
ill !!! IE! s!j li. ss sii ill i


















V
t-4
'^
r=
M
O
r^
?
C3
T5
C
o •
4J (A
CO C
O
rs
j «
•1
c
00
= 1
^i
41 en

ec co
c o
(3 o^.
O-H
rH H
E --0
5
«rH
4> CO
TJ Ofcl
3 t-«J
0 §
C 0!_1
rH rH
••
4^ xin
5-M.
W 1 «1
O «r-
S 1. •
•*v.
^ tflg
£§"
«•" "HO
e «-<
o •»
O.s«
u i3--
K £s
10 li
U -^
KH — • k
U U
< '•£
H « f
S ao
• #
«
1-10

-------
"O r-
<0 C -f
> '.- .O
•r- X> O
•M t: £1
O O
*t~ CO
«*- i- (0 S-
Ul O -M
I <*- >, C
-4-> I- O
U5 -C (O <_>
O U C
<_> TO O O
O -f- O
+J J- 4-> J-
i/> Q. rg 3
O OL +-> O
i
C C
o o
I- H-
o
O \D UD
O O O
CM «— r—
X X

to ir>

coin
C
O

o
o
w
tr
O
Ol
•»!—
Q> QJ r3
O U -l->
S- $- O

o b
*/> oo

CO
CVJ
CM
CM
O
CM
CO
OJ
Of
C
«3
CO

O
o
CD
O
CM

i-
HJ
(3
C*
J-
ro
Q)
C
o
10
CM O
T3
o
oo
CU
an
O irt
E OJ
QJ O
xo
>
<«- !-. 0
O rj -l->
C «

O 4-> I/)
t— <0
•*-> ai
t/> to >
to cu
-------
I .
I'M
ill!!
j'ji
ill
51!
O *
I
J
00
1-12
-------
8:35 a.m.
Keynote Address
Frank Princiotta
Q: On the column where it said percent of control, say ±t
was 10%, did that mean that you could control 10% of
the emissions or reduce them to 10%?
A: No, that is 10% control of the base line emissions.
In other words, I gave you the number of 34,000,000
tons per year assumed in 1980. That very first level
of control would mean that roughly 3.4 million tons
could be controlled utilizing the technology, then as
you add these other strategies you are able to increase
that percentage to, at the very most, roughly 2/3 of all
the emissions. I might point out that obviously this
is a somewhat simplified analysis and one has got to be
worried about ambient air quality in specific locations;
one has got to be worried about nitrates as probably
being an important pollutant, perhaps as important as NO.-
There are many complications through an analysis of this
type. I don't want to imply that, by any means, this
will be the sole basis of EPA policies; it certainly will
not.
Q: You use a dollar per ton basis; what's the unit?
that tons of coal used?
A: It's tons of NO removed by the control process.
Is
Q: Would you care to comment on these 1980's standards
that you are hinting at, in terms of pounds per million
Btu's fired for stationary power plants of the 500 mega-
watt variety - what the implications were? For example,
1-13
-------
at the moment it is .7 pounds of NOX per million Btu's
fired for coal. Do you contemplate a tighter restriction
than that?
A: It's a very complex issue. 1 can just give you a personal
opinion and that is that we will, within the next decade,
see tighter standards on the big sources. In my own
opinion, the only question is will it be a tighter
standard than a combustion modification technology
would allow. As you may know, in Japan, which is primarily an
oil burning country, there are standards so tight that
combustion modifications are not sufficient. Controls
beyond 50-60% are required in some areas and there they
have to go to flue gas cleaning. Whether or not this
will happen in this country I don't know, but I personally
think it's likely that within the foreseeable future,
within the next 10 years, we may see a tightening up
on coal-fired steam generators. It's apparent that
our present strategy is not good enough.
Q: I don't want to totally quote Dr. Bowen out of context,
but in the April 14 American Flame Research Committee
meeting he quoted coal standards for 1980 around 200 ppm
and for 1985 around 100 ppm; do you know if these are
still good numbers?
[Dr. Bowen]
A: I don't believe I quoted these as standards that we would
be achieving. These were targets that we would be aiming
for in our technology development effort.
In noting the numbers you have on the chart there, it
was emission tons controlled and certainly the thing
we're all concerned with is ambient concentrations.
Do you intend to expand your study to relate the
control to tons at an ambient level rather than a
1-14
-------

A:
Q:
A:
ton emission level? We see in our area that a certain
amount of control on the mobile sources can take care
of a great deal more of the ambient level concentra-
tions of NO than control on the industrial sources.
Yes, as I mentioned before I think this is right.
Analysis is continuing in this area and our Office
of Air Quality, Planning and Standards is doing
these kinds of analyses. Clearly they'll have to
do it on a regional basis and will have to take into
account transport phenomena. My presentation was
based on a superficial study indicating the relative
priorities of control for stationary versus mobile
sources.
I wonder if you were able to look at the energy impact
of the various options you had listed.
No, I'm afraid this particular study from the Depart-
ment of Transportation, to my knowledge, did not look
at the energy impact. I think we will probably be
hearing at this symposium a little bit about what the
energy situation might be with combustion modifica-
tion technology associated with so many strategies.
As far as flue gas cleaning is concerned, that is a
fairly energy-intensive operation, and some of the
systems in Japan would probably require several
percentage points of a total power plant output to
run their flue gas cleaning units; probably energy
usage is not quite as high but in the same ball park
as FCD scrubbers.
1-15
-------
-------
SESSION I

FUNDAMENTAL RESEARCH
Dr. J. S. Bowen; W. S. Lanier
Co-Chairmen
1-17
-------
THE FORMATION OF SOOT AND POLYCYCLIC AROMATIC

HYDROCARBONS (FCAR) IN COMBUSTION SYSTEMS
By:

J.D. Bittner, G.P. Prado, J.B. Howard, R.A. Hites

Fuels Research Laboratory

Department of Chemical Engineering

Massachusetts Institute of Technology

Cambridge, Massachusetts
1-18
-------
Introduction

Soot formation In combustion systems has been studied for many

years with the long range objective of attaining a working knowledge

of the process to aid in predicting flame emlsslvities for radiative

heat transfer problems. However, recently concern has been expressed

that soot and some of the organic compounds associated with incom-

plete combustion may become pollution problems that will have to be

controlled. It is known that some polycyclic aromatic hydrocarbons

(FCAH) produced in flames and adsorbed on soot particles are carcin-

ogenic. Soot and other organic particulate matter may not be the

most abundant particulates emitted, but because the small particles

(< 0.2 pm) are easily invested deep into the respiratory system,

they may be one of the types of particulates most hazardous to human

health. The NOx control strategies such as lower temperature com-

bustion and staged combustion presently being considered work to

Increase PCAH and soot formation. Another change in combustion

practices that could Increase the emissions of particulate organic

matter is the predicted increased use of coal as a solid or heavy

liquid since highly aromatic fuels are known to be troublesome in

this respect.

The mechanism of soot formation in combustion systems is not

well understood. Some of the mechanisms proposed over the years and

the confusion surrounding the role of PCAH in the process are illus-
I- 19
-------
trated in Figure 1. Since soot particles contain many more carbon

atoms and a much lower H/C ratio than the fuel molecules, soot

formation must involve processes of aggregration (top to bottom) and

dehydrogenation (left to right). The extreme routes, the C_ route

and the saturated polymer route, are unlikely to occur under typical

combustion conditions. The C. route may be important at high

temperatures around 3000°K, and below 700*K and at long reaction

times the saturated polymer route may occur. But, around 2000°K

most mechanisms can be represented by the central portion of Figure

1. From this figure it is seen that PCAH species have been postu-

lated to serve as nuclei for soot growth by surface decomposition of

any number of hydrocarbon species (including themselves). They have

been postulated to be of sufficiently low vapor pressure to physically

condense to liquid droplets, which then form soot. They have also

been postulated to be stable byproducts of the soot formation process.

Project Objectives

In light of the concern over particulace organic matter as a

pollutant and the confusion surrounding soot formation mechanisms

and the role of PCAH, the overall objective of the project is to

assess qualitatively and quantitatively the production of particulate

organic matter (soot and organic compounds) in well-defined yet

relevant combustion systems. To that end two types of information

are of interest: the identities of PCAH and concentrations of soot
I- 20
-------
and PCAH i) emitted from the flame and 11) at different stages of

the combustion process.

To obtain the information on PCAH and soot emissions measure-

ments on both laminar and turbulent atmospheric pressure diffusion

flames are being made. Extensive measurements on the turbulent

diffusion flame are being made possible by Interaction with the

Mechanical Engineering Department at M.I.T. on soot and other

emissions research. One benefit of this type of collaboration is

the opportunity to study the role of turbulent mixing on soot and

PCAH emissions and the resulting interdependencles with NOx and CO

emissions. In prior research efforts measurements of CO and NOx

emissions have been made on this burner system and a model to

9-11
characterize the mixing process has been developed and tested.

To obtain the information on factors that govern PCAH and soot

emissions a low pressure laminar flat premixed flame was chosen.

In such a flame the kinetic processes are easily studied because the

different stages are well resolved in time and space. The low

pressure system allows the use of on-line mass spectrometry which is

difficult with high pressure particle-laden gases because of clogging

of the sampling probe orifice.

Low Pressure Flat Flame Study

A schematic of the apparatus to be used in the low pressure

flat flame study is shown in Figure 2. A modulated molecular beam
1-21
-------
mass spectrometer system will be used for on-line analysis of gas

phase species over the atomic mass range of 1 - 1000 a.m.u. A major

advantage of this sampling-analysis technique over other techniques

is that the excellent quenching possible with molecular beam sampling

allows detection of reactive free radical species. Other advantages

are the extension of the mass range beyond that usually studied with

gas chromatography (< 400 a.m.u., limited by volatility) and the

quick analysis of species of all molecular weights within the

extended mass range. One disadvantage of mass spectrometry Is that

it cannot distinguish between isomers. To help sort out the isomer

problem a cryogenic sample collector has been designed to trap the

beam of gas for combined gas chromatography - mass spectrometry

(G.C.-M.S.) analysis.

Samples of soot particles will be collected by placing electron

microscope grids into the beam. From the electron microscope analysis

particle number concentrations and size distributions will be obtained.

Data to be taken are concentration profiles of low and high

molecular weight species, temperature profiles and particle number

concentrations and size distributions as a function of position in

the flame. Acetylene-oxygen flames of several fuel equivalence

ratios and burner exit velocities will be investigated over the

2
pressure range 2 to 5.33 kN/m (15 to 40 torr). The particle data

will be reduced to rates of particle nucleation, growth and coagu-
1-22
-------
2 3
lation with an existing model. ' These rates will then be used

with the organic species gas phase concentration data to Investigate

' the factors governing PCAH and soot formation.

Turbulent Diffusion Flame Studjr

The burner being used for the atmospheric pressure turbulent

diffusion flame study is shown in Figure 3. This burner can be

operated with either liquid or gaseous fuel, and a selection of

nozzles with and without atomizing air. Gas and soot samples are

collected with the water-flushed sampling probe shown in Figure 4.

The water injection design has several advantages over conventional

water-cooled probes. The injected water prevents soot deposition and

clogging thereby allowing quantitative collection of the soot with-

out having to perform the tedious task of scraping the probe after

each sample is taken. After passing through the probe,the water,

gas and particulates pass through a tightly packed glass fiber filter

where the soot is deposited. The water and gas then pass through a

condenser for cooling. The water is collected in a large flask and

the gas flow is measured by a wet test meter after passing through

an ice bath condensate trap and vacuum pump. The amount of soot

collected is determined by weighing the dried filter before and after

collection. The filter with the soot deposit is then extracted with

methylene dichloride. The filter is again dried and weighed. The

soot is thus classified as carbonaceous residue and methylene
1-23
-------
dichloride soluble material. The methylene dichloride soluble

portion is then concentrated by evaporation and analyzed by G.C.-M.S.

for PCAH compounds. The carbonaceous residue is then removed from

the filter and dispersed in an ultrasonic bath and deposited on

electron microscope grids for particle size distribution analysis.

The G.C.-M.S. systems used in this study have both high and low

resolution mass spectrometry interfaced with an IBM 1800 computer

for data storage and processing. The details of the system and its
* A
use for Identification of PCAH are described elsewhere.

The data being collected on this burner system are the mass

loadings of soot, the number concentration and size distribution of

soot particles and concentrations of individual PCAH species. This

information is being obtained at different positions along the axis

of the burner as well as in the exhaust gases. Formation of particu-

late organic matter is being studied for several fuel equivalence

ratios and fuel injection modes. Kerosene is currently being used

as the fuel. Future plans are to study different liquid and gaseous

fuels.

Preliminary Measurements of Spot Mass Loadings

Preliminary scoping runs to measure soot loadings with kerosene

and air have yielded some interesting results. The soot loadings in

units of mg of soot per standard cubic meter of gas for a fuel equi-

valence ratio, 4 - 1.0 and three atomizing air pressures are shown
1-24
-------
in Figure 5. Profiles along the axis of the burner for atomizing

air pressures of P - 205 kN/a2 (15 psig) and P « 239 kN/m2 (20 psig)

show decreasing soot loadings beyond 10 en from the nozzle. The

Interesting point, however, is the strong dependence that the soot

loading at the exhaust (53 cm) has on the atomizing air pressure.

The soot emission increases four orders of magnitude for a decrease

in atomizing air pressure from P - 239 kN/m2 (20 psig) to 205 kN/m2

o
(10 psig). Flagan and Appleton measured NOx emissions from this

burner under similar conditions (
-------
References
1. Palmer, H.B. and C.F. Cull Is, "The Formation of Carbon from
Gases" in "Chemistry and Physics of Carbon," Vol. 1, P.L.
Walker, ed., pp. 265-325, Marcel Dekker, Inc., N.Y. (1965).

2. Wersborg, B.L., J.B. Howard and G.C. Williams, Fifteenth
Symposium (International) on Combustion. The Combustion
Institute, Pittsburgh, p. 929, (1973).

3. Wersborg, B.L., J.B. Howard and G.C. Williams, Faraday
Symposia of the Chemical Society. No. 7. 109 (1973).

4. Hites, R.A. and K. Biemann, Anal. Chem., 39, 965 (1967).

5. Hites, R.A. and K. Biemann, Anal. Chem.. 40, 1217 (1968).

6. Hites, R.A. and K. Biemann, Anal. Chen.. 42, 855 (1970).

7. Hertz, H.S., R.A. Hites and K. Biemann, Anal. Chem.. 43,
681 (1971).

8. Hites, R.A., A.C.S. Division of Petroleum Chemistry Preprints
20, 824 (1975).

9. Flagan, R.C. and J.P. Appleton, Combustion and Flame. 23.
249 (1974).

10. Appleton. J.P. and J.B. Heyvood, The Fourteenth Symposium
(International) on Combustion, The Combustion Institute,
Pittsburgh, p. 777 (1973).

11. Pompei, F. and Heywood, J.B., Combustion and Flame, 19,
407 (1972).
1-26
-------
Dehydrogenation
c
o
•fj
o
O)
O)
O)
C2 molecules
Paraffinic
Fuel
Polymerization
ond
Dehydrogcnotion
Stable
PCAH
By-Products
Hydrocarbon
Ions
j I Condensation
Soot
Nucleus
Polyacetylenes
- Surface
Growth by
Decomposition
Physical
Condensation
Dehydrogenation
and
Polymerization
Polymerization
Saturated or Liquid Droplets Carbon Particles
Unsaturated —"" Containing —•" With Graphite -
Polymers Polycylic Compounds Like Structure
Figure 1 Summary of proposed mechanisms of soot formation
1-27
-------
quadrapoie mass

spectrometer
collimator

orifice
chopper
skimmer
quartz nozzle
-4
4x10 torr
i
burner
8x106 torn
2 x 10 5 torn
20 torr
02
= +- CoH
2n2
Figure 2 Modulated beam - wss spectrometer sampling apparatus
I-2S
-------
o>
c
CP
' C.
N <\
*• A\
.-. fl^^N
rr> ITTl
i
SET
v_
o ^-— —
§?
**~ "c.
tr IJ
i
Tir
c.
"o.
§ =J
o
si
••
(
L
"^

1
i
i
\
i

^m
D
3
L
1
- —

f
^

\ L
^
-s

t m

N
' 5
.if

vv^>. .
— > \
7
r — *

••

L.
<

\
WA
\
!
i

>
^_
CL

im
mr ,,
"oo
im y,
J HIT
m
nof
i — ,

d>
I
O
•H
W
Q

5
r-l

g
HI
M

00
a.
1-29
-------
0)
ja
o
M
e.

eo
C

-------
10
O
2
*»
O
Z
5
<
o
o
o
I/)
10
10
10
T
P=190 kN/rr/
(10 psig)
P= 205 kN/m (15psig)
P= 239 kN /m ( 20psig)
O 10 20 30 4O 5O 60
AXIAL DISTANCE FROM NOZZLE , CM
Figure 5 Preliminary results of soot loading as a function of
axial position and atomizing air pressure. For
conversion from mg/sra3 to the mass fraction of soot
in the combustion gases multiply by 7.81 x 10~?
g Poducts
/

/
mg/sm3-
1-31
-------
8:45 a.m.
Formation of Soot and Polycyclic Aromatic
Hydrocarbons in Combustion Systems
James D. Bittner, Massachusetts Institute of Technology

'Is there any correlation to experience between the
laboratory condition of low atmospheric and sub-
atmospheric pressure where you obtain your radicals
and your small, semi-solid particles, and the
reality of actual combustion and atmospheric pressure?
Your question is, are conditions the same. Peak
temperatures for acetylene oxygen flame are
around 2000° K. I have no evidence on production of
radicals' effect on temperature. We really haven't
gotten to the study of hydrocarbon free radicals; that
system is still in the construction stages.
What type of liquid fuels are you using in the
atmospheric furnace and are you studying the effects
of varying asphaltene concentrations or various
sulphur concentrations in the fuel?
Currently we are using aviation kerosene. There
are some plans of investigating the effect of the
degree of varomaticity of the fuel on hydrocarbon
and polycyclic aromatic formation; we haven't looked
into sulphur effects.
Q: You mentioned that the carbonaceous residue, which
is insoluble, is dispersed in an ultrasonic bath and
then particle size distribution is measured with the
electro-microscope. I was wondering if you would
comment on the potential irreversible agglomeration
that might occur and affect your results. If agglom-
eration occurs which can be reversed with the ultra-
sonic bath, then the distribution you're measuring
mav not be the field distribution.
1-32
-------
A: Yes, I have to agree with that. We've had some problems
in getting good dispersions, and rather than a bath or
a tub, we've gone to an ultrasonic probe which seems to
provide much more intense agitation; but you're right,
we have no guarantee that we can completely re-disperse
the material.
1-33
-------
1-34
-------
EFFECTS OF FUEL SULFUR ON

NITROGEN OXIDE EMISSIONS
By:

J.O.L. Wendt and J.M. Ekmann

Department of Chemical Engineering

University of Arizona

Tucson, Arizona 85721

Presented at EPA Symposium on Stationary Source Combustion

Atlanta, Georgia, September 24-26, 1975
1-35
-------
ABSTRACT

The effect of some fuel sulfur compounds on NOX emissions

arising from thermal fixation was investigated. Labora-

tory experiments, using a methane-air flat flame doped

with either S02 or H2S, showed that fuel sulfur can in-

hibit the formation of nitrogen oxides, and that this in-

hibition is significant at all air fuel ratios and

especially at high air preheats. A mathematical simula-

tion of a flat flame was developed and this showed that

the observed effect could be explained by a kinetic

mechanism involving the catalysis of atom recombination

reactions by S02-

The experimental and theoretical results obtained are

especially significant from a practical point of view,

since they indicate that fuel desulfurization may lead to

increased NOX.
1-36
-------
INTRODUCTION

One of the roles of fundamental combustion research is to

help identify new problem areas of concern to EPA. A po-

tential area of concern is the interrelationship between

sulfur oxide and nitrogen oxide abatement. Sulfur oxide

pollution may be abated by either fuel desulfurization or

stack gas scrubbing. The choice of abatement method is

usually dictated by economic considerations. However, in

calculating the cost effectiveness of various sulfur oxide

abatement strategies, it is important to determine the ex-

tent to which the technology used has an adverse effect on

other pollutants such as NOX. Should sulfur compounds in

the flame front have an inhibitory influence on the forma-

tion of NOX, this would indicate that fuel desulfurization

might require additional NOX abatement methods to be im-

plemented, and that this would Involve additional costs

which would not occur with stack gas scrubbing, where the

sulfur species are removed after the combustion process.

That sulfur and nitrogen oxides interact at low tempera-

tures is not new and has been studied exhaustively C1*2).

This interaction results in the catalysis, by NO, of the

oxidation of SOa to S03. At higher temperatures, under

combustion conditions, the situation is quite different

and a clear distinction should be drawn between low
1-37
-------
temperature Interactions. At high temperatures under com-

bustion conditions, free radicals are produced in super-

equilibrium amounts and this fact has been shown to be im-

portant in explaining high NO production rates'3 »**) where

the NO is formed by atmospheric fixation (Thermal NO).

Since it has been shown(5»6) that sulfur dioxide is an

effective catalyst in reducing superequilibrium free

radical concentrations, it is reasonable to expect that

sulfur dioxide and possibly other fuel sulfur compounds

inhibit the formation of NO in flames. In order to ex-

plore this possibility further, it is necessary to focus

on certain fundamental aspects of NO formation mechanisms

In spite of much research in the area, there is even now

no general agreement on the kinetic mechanisms of thermal

NO production. Most widely recognized is the mechanism

proposed by Zeldovich^7)
N2 + 0 -»• NO + N

N + 02 -»• NO + 0

with the modification

N + OH -* NO + H
(I)
(ID
(III)
and with the free radicals necessary for these reactions

being produced through the combustion process. It should
1-38
-------
be noted that the free radicals so produced can have con-

centrations many fold In excess of those determined by

equilibrium and that the decay of these radicals towards

equilibrium is relatively slow and occurs downstream from

the flame front. It is in this region of free radical de-

cay that the bulk of the NO is formed through reactions

(1) through (3). Thompson and Beer(3) have shown that, in-

deed, superequilibrium concentrations of oxygen atoms are

responsible for high rates of NO formed in their apparatus

and their conclusion was corroborated by other workers,

("*>8) especially as regards NO formation in the fuel lean

regime. In the fuel rich regime, however, it appears that

for hydrocarbon flames, an NO formation mechanism involv-

ing cyanide compounds as intermediates may be applicable

(**t9), and under these conditions the role of superequi-

librium atom and radical concentrations is unclear. It is

generally recognized, however, that high rates of NO for-

mation can result from superequilibrium atom concentra-

tions, and it would therefore appear that catalysts and

other impurities, such as S02, that have been shown to de-

crease radical concentrations, should tend to lower

thermal NO formation rates.

In order to determine the effect of fuel sulfur on thermal

NO it is necessary to devise a well defined laboratory ex-
1-39
-------
periment to answer the following questions:

• Does SOz have an inhibitory effect on thermal

NO emissions?

• Under which conditions is any inhibition of

NO formation by SOz most significant?

• Does S02 affect the formation of "prompt NO"

and if so what conclusions can be drawn about

the role of superequilibrium atom concentrations

and "prompt NO"?

• With H2S in the fuel, is conversion of the

fuel sulfur to S02 sufficiently rapid to allow

the SO2 formed to have the same effect as when

added directly to the fuel?

In order for laboratory results to be extrapolated to

other conditions, it is also necessary to develop theo-

retical mathematical models that describe the appropriate

kinetic mechanisms, and that can be used to determine the

significance of the results in other, more real world

combustion environments. In particular, mathematical

models will give insight into fundamental questions such as

• Can the observed effect be explained by catalysis

of atom and radical recombination rates by SOj?

• What can be expected under different time

temperature histories?
1-40
_
-------
• What can be expected in real furnace flames?

EXPERIMENTAL APPARATUS

Combustion Rig

A schematic of the premixed combustion rig and supporting

equipment used is shown in Figure 1 and a diagram of the

burner itself in Figure 2. Metered amounts of methane

(Matheson, C.P.) preheated house air, and when appli-

cable, S02 or HaS (Matheson, C.P.)» were allowed to mix,

then preheated further before being fed into a modified

Meker burner. The temperature of the gas mixture entering

the burner was controlled. The Meker burner was modified

so that an approximately flat flame could be supported

above the burner grid. The burner was at atmospheric

pressure and enclosed in a pyrex glass chimney.

The combustion rig was designed primarily for a large

number of input/output measurements rather than for de-

tailed in-flame probing. However, some detailed probing

was successfully attempted, and this showed that the flame

could be considered flat to within our experimental error.

Partial temperature profiles showed that the flame had

temperatures in excess of 2000°K, even with no air pre-

heat. This means that heat loss to the surroundings was

Figures 1 and 2 follow
1-41
-------
TO ANALYSIS
TRAIN
BURNER
ASSEMBLY
MIXING
LENGTH
VARIAC
PREHEAT ASSEMBLY '.'•
TO EXTERNAL
POWER SOURCE
^
LtriJ
INSULATION
a
HEATING TAPES
'L" - £1 ______ '
------
REGULATOR /
LINE FILTER
ROTAMETER
/VALVE
ASSEMBLY
METHANE
SOZ/H2S
AMMONIA
LOW PRESSURE AIR
£Q£Q
FIGURE 1, SCHEMATIC OF APPARATUS
1-42
-------
QUARTZ
GLASS
CHIMNEY
GRID

GLASS
BEADS

POROUS
METAL
PLATE
MIXTURE
INLET

COOLING
'WATER
COIL
PREMIXED
"BURNER
THERMO-
COUPLE
FUEL and
S02/H2S+AIR
SWAGELOK
FITTING
COOLING
WATER
FIGURE 2. FLAT FLAME COMBUSTOR
1-43
-------
not great, and might distinguish this flat flame from

othersO0,11).

Sampling and Analysis Train

A schematic of the sampling and analysis train is shown in

Figure 3. The sample was drawn through a 7mm diameter

orifice into a 6mm OD, uncooled, quartz sampling tube.

Preliminary experiments in which the sampling rate was

varied by a factor of four showed that our data did not

depend on sampling rate and this, together with data

from other experimentalists(l2\ indicates that all reac-

tions were effectively rapidly quenched and no further

reaction took place in the tube. The height of the sam-

pling tube could be positioned accurately to within

0. 03mm.

Quartz and teflon tubing were used throughout the sampling

and analysis train since stainless steel tubing has been

shown to interfere with NO analysis under rich conditions.

A cooled knockout pot removed moisture in the burned gas

sample. The analysis train had the following features:

• NO/NOX analysis by Thermoelectron Chemilumin-

escence Analyzer with stainless steel converter

• 02 analysis by Beckman Model 715 (Electrochemical)

Oa Monitor

Figure 3 follows

1-44
-------
AIR BYPASS
FROM BURNER
ASSEMBLY
,FROM CALIBRA-
*TION GAS
OXYGEN
MONITOR
CHEMILUM1N-
ESCENT t
DETECTOR]

PUMPM
GAS CHROM-
ATOGRAPH
sop
DETECTOR
TO EXHAUST HOOD
FIGURE 3. SCHEMATIC OF ANALYSIS SYSTEM
1-45
-------
* CO analysis by chromatograph with Porapak Q

column

• S02 analysis by Theta Sensors S02 (Electro-

chemical) Monitor.

The chemiluminescence analyzer worked perfectly and showed

no interference by SOz, 02 or CO. This confirmed pre-

vious results^13) which showed that SOa does not interfere

with (Thermoelectron) chemiluminescence measurements of NO,

RESULTS

Pr emjlxed Comb us t or P er f or man ce

Figure 4 and Figure 5 show results obtained with no S02

or HzS added to the fuel. These are the base cases show-

ing exhaust NO emissions as functions of air fuel ratio,

air preheat (Figure 4) and NO concentration within the

flame as a function of residence time from the burner

(Figure 5). Figure 4 shows that with no preheat a maximum

of 152 ppm (dry, reduced to 100% stoichiometric air) NO

was obtained at 104% stoichiometric air while with 240°C

air preheat the maximum was 232 ppm. There is also a

strong dependence on air fuel ratio. Figure 5 shows that

formation of NO was complete at 6 cm above the burner grid

or after a residence time of approximately 20 milliseconds
Figures 4 and 5 follow
1-46
-------
Q RUN I
£ RUN 2
ORUN 3
ARUN 4
ORUN 5
250 -•
200-
A-"xa
o*P
240°
PREHEAT
NO
PREHEAT
80
FIGURE 4
90
100
110
120
% STOICHIOMETRIC AIR
THERMAL NO EXHAUST EMISSIONS - BASE CASE
1-47
-------
CHOIOJ.S) ON Wdd
1-48
-------
and that sampling at that point was truly representative

of exhaust NO emissions. Figure 5 also shows that under

fuel rich conditions all the NO is formed very early in

the flame and that this "prompt NO" was not a strong

function of air preheat and that more "prompt NO" was

formed under fuel rich conditions than under fuel lean

conditions. These results agree somewhat with those of

Fenimore(9 ) .

The ppm NO measured, under no preheat conditions, is sub-

stantially greater than that measured by other workers in

flat flames^11). This is probably due to the low heat

loss rate in our system, and by the resulting high tem-

peratures. The existence of temperatures well above

2000°K was confirmed by (incomplete) temperature measure-

ments^1") .

At each point (Figure 5) under fuel lean conditions both

NO and NOX were measured by the chemiluminescence ana-

lyzer. At most 3 ppm N02 were observed, and then only

under very fuel lean conditions. We thus did not

observe any early N0?formation as reported by Merryman and

Levy(l5).

Additional runs were also made to investigate whether the

addition of a fuel additive, such as SOa, would lower NO
1-49
-------
emissions significantly by virtue of dilution alone. With

molecular Nj as the fuel diluent at zero preheat, 104%

stoichiometric air, it was found that 10% N2 in the fuel

led to a reduction of less than 7 ppm NO in the exhaust.

This means that any effect (larger than this) due to

addition of up to 5% S02 to the fuel is due to kinetic
t.

interactions and not just simple dilution and temperature

reduction.

Effect of SOg on NOX Emissions

The effect of S02 as an additive in the fuel on the ex-

haust emissions of nitrogen oxide is shown in Figures 6

and 7. In Figure 6 the ppm NO (dry, reduced to

stoichiometric) in the exhaust is shown in the absence of

air preheat with and without 4.9 percent by volume S02

in the fuel. 4.9 percent by volume S02 in methane leads

to approximately 6800 ppm S02 in the exhaust. It can be

seen that at approximately 101% stoichiometric air, 4.9%

S02 in the fuel lowers NO exhaust emissions by 50 ppm or

by about 36%. At other air/fuel ratios the percent re-

duction is somewhat less as shown on Table 1. At a pre-

heat of 240°C (Figure 7) and at 101.% stoichiometric air

the addition of 4.9% of S02 in the fuel lowers NOX emis-

sions by about 60 ppm or 30%. Conversely, looking at the

Table 1, Figures 6 and 7

1-50
-------
TABLE 1. REDUCTION IN NOX EMISSIONS BY S0a ADDITION TO FUEL
Percent
S toichiometric
Air
Reduction in NOX Emissions
2.5% S0a in Fuel
4.9% S02 in Fuel
no preheat 240°C preheat no preheat 240°C preheat
PPro
ppm
ppm
ppm
80.1
90.0
101.3
103.0
110.0
117.4
7
—
22.0
—
--
12.0
12.6
__
14.0
—
--
11.7
6.0
__
21.0
—
—
37.0
10.7
—
10.7
—
—
27.4
6.0
9.0
50.0
40.0
27.0
16.0
16.0
17.3
35.7
26.0
24.5
13.9
13.0
15.0
60.0
68.0
60.0
22.0
24.6
23.1
30.9
30.4
26.4
21.6
1-51
-------
NO PREHEAT
0 RUN 2
A RUN 6
Q RUN 7
90 IOO 110 120

% STOICHIOMETRIC AIR
FIGURE 6. S02 INHIBITS NO FORMATION AT ZERO PREHEAT
1-52
-------
3
4
ORUNS
ORUN e
D RUN
240°C PREHEAT
90
100
no
120
FIGURE 7
% STOICHIOMETRIC AIR
S02 INHIBITS NO FORMATION AT HIGH PREHEAT
1-53
-------
effect of removal of S02 from the fuel one can say in

this case, fuel desulfurization caused increases in NOX

emissions of up to 55%.

Further details are shown in Table 1 in which results from

two different flames are presented. (Flame 1 has slightly

lower base case NO emissionst) These results clearly show

that as the S02 level in the fuel decreases so does reduc-

tion in NOX emissions. With less than 1% S02 in the fuel

any inhibition effect was not significant.

Figure 8 shows the results of probing within the flame

(118.5% stoichiometric air, 4.9% S02 in Fuel) and clearly

demonstrates that at both preheats the effect of SOg is to

quench the formation of NO fairly early in the flame.

This data gives insight into a probable kinetic mechanism

as described later.

In Figure 9 the effect of 2.5% and 4.9% S02 on the forma-

tion of "prompt NO" is shown, where "prompt NO" is defined

in this case as that formed within 0.3 cm of the burner

grid. It can be seen that at both preheats and at all air

fuel ratios, the effect of increasing 862 is to decrease

"prompt NO" formation. This indicates that superequilib-

rium concentrations of atoms and free radicals might be

Figures 8 and 9 follow

1-54
-------
q
6
P
10
o
z
o

-------
O 0%
Q 2.5% S02
0 4.9%S02
75 •
25- -
240°C PREHEAT
NO PREHEAT
75-
GL
OL
80
FIGURE 9.
90
100
110
120
% STOICHIOMETR1C AIR
EARLY-FORMED NO INHIBITED BY SO2
1-56
-------
Important under all air/fuel ratio conditions.
Effect of H2S on
Emissions
In fossil fuels, sulfur is normally present in the reduced
»

state. Thus, some experiments were completed with H2S as

the fuel additive, in order to determine whether fuel

sulfur in this form has an effect on thermal NOX for-

mation.

Figures 10 and 11 show the effect of H2S addition at two

levels of mixture preheat. Since H2S is a fuel, addition

of this compound changes the air fuel ratio, and this has

been taken into account in labeling the abscissa axis.

In Figure 10 it is clear that both 2.6% and 5.0% H2S in

the fuel, with no mixture preheat, inhibit the formation

of NOX. NOX emissions were reduced up to 31.6% under fuel

lean conditions as shown on Table 2. With 250°C mixture

preheat, H2S inhibits NOX formation on the average by an

even greater extent as shown on Figure 11 and Table 2.

It is instructive to compare the inhibition of NOx forma-

tion by S02 and by H2S. Comparison of Table 1 and Table 2

show that without preheat S02 had a greater effect while

with 240°C mixture preheat the effects of H2S and S02

addition are roughly the same. This is probably due to

Table 2 and

Figures 10 and 11 follow

1-57
-------
TABLE 2. REDUCTION IN NOX EMISSIONS BY H2S ADDITION TO FUEL
Percent
Stoichiometric
Air
80
90
100
103
110
115
Reduction in NQX Emissions
2.6% HaS.in Fuel
5.0Z H2S in Fuel
No Preheat 240°C Preheat No Preheat 240°C Preheat
ppm
ppm
ppm
5.0
8.0
28.0
17.0
15.0
15.0
11.4
14.6
23.3
11.4
11.1
15.8
7.0
10.0
26.0
50.0
44.0
23.0
13.2
15.4
17.1
22.8
19.1
14.1
5.0
8.0
28.0
28.0
26.0
30.0
11.4
14.6
23.2
18.9
19.4
31.6
13.0
20.0
42.0
75.0
63.0
48.0
24.6
30.8
27.7
34.0
28.0
29.4
1-58
-------
ORUN i
DRUN 2
14
ARUN 15
NO PREHEAT
FIGURE 10,
90 100 110 120

% STOICHIOMETRIC AIR

H2S INHIBITS NO FORMATION AT ZERO PREHEAT
1-59
-------
QRUN 3
RUN 4
ORUN 5
QRUN is
QRUN is
RUN 20
0 RUN 21
0% HoS
240°C PREHEAT
2.6% HoS
5.0% H2S
9O 100 MO
% STOICHIOMETR1C AIR
FIGURE 11. H2S INHIBITS NO FORMATION AT HIGH PREHEAT
1-60
-------
the effect of temperature on conversion rates of H2S to

S02.

Figure 12 shows the effect of H2S on "prompt NO", and

indicates that under fuel lean conditions the presence of

H2S does lower "prompt NO" formation rates. Under fuel

rich conditions, where conversion of H2S to S02 is not

complete, there is little effect of H2S on "prompt NO".

The foregoing indicates that H2S must be converted to S02

before inhibition of NO is important, and that this occurs

rapidly under fuel lean conditions. This is confirmed in

Figures 13 and 14 in which S02, NO, and 02 concentrations

are plotted as functions of time for zero preheat. At

98% stoichiometric air (Figure 13) H2S conversion to S02

is essentially complete when NO has attained 70% of its

final value; at 113.5% stoichiometric air the conversion

of H2S to S02 is essentially complete when the NO has

attained only 34% of its final value. Since H2S has a

greater inhibiting effect in the fuel lean case, it would

appear that inhibition of NO formation occurs through the

rapid conversion of H2S to S02 and by the subsequent in-

hibiting effect of S02. Thus, under fuel lean conditions,

a kinetic model simulating fuel sulfur as S02, rather than

as H2S, would be adequate.

Figures 12, 13 and 14
follow

1-61
-------
O 0% H2S
0 2.6% H2S
Q 5.0 % H2S
240° PREHEAT
••25
•o
NO PREHEAT
-•75
80 9O 110
% ST01CHIOMETRIC AIR
FIGURE 12. EARLY-FORMED NO INHIBITED BY H2S
120
1-62
-------
(utdd)2()S
T
O
o
in
O
O
O
00
(%)|OI
o
0)
o
CNJ
O
x ON
1-63
-------
(ujdd)
I
o
o
o
w
I
52
O
O
CM
O S

10
5 ^
— x
O
iq
csj
o
o
o
00
o
<0
o
CM
(%) |OI «zO/(tudd)0oi x ON
O
O Q
M 7Z
w o
a: u
W
o w
O A
1-64
-------
MATHEMATICAL MODELING

In order to model the kinetic mechanisms of the sulfur-

nitrogen oxide interactions experimentally observed it is

first necessary to model the physical environment of a

flat flame in which the reaction chemistry occurs. In a

flat flame the physical processes of convection and

diffusion are important and simple plug flow models are

inadequate. Indeed a substantial amount of back diffusion

into the unburned gases is crucial in allowing a stable

flame front to be maintained, and in allowing ignition to

occur. Simple models that impose a specified time tem-

perature-history on the kinetics environment can be mis-

leading, especially in the case examined here, where tem-

perature, free radical concentrations and nitrogen oxide

kinetics are intimately coupled. For example, high

superequilibrium concentrations of atoms and free radicals

necessarily lead to significantly lower temperatures be-

cause of the enthalpy of disassociation of oxygen, hy-

drogen and water molecules. Thus a substance that

catalyzes atom recombination rates and lowers free radical

concentrations will also raise the flame temperature at

that point and this rise in temperature may offset, in

some degree, the effect of lower oxygen atom concentra-

tions as regards NO formation. Thus any reasonable model
1-65
-------
describing the observed effects must

a) calculate the resulting temperature from a

heat balance and

b) properly take account of diffusion in the

flame front.

Unfortunately, no model of a flat flame is generally

available and so it was necessary to develop a very

approximate simulation to be used in this study. It is

recognized that substantial improvements in such a simu-

lation are desirable. However, our model is an improve-

ment on others that take no account of diffusion in the

flame front. The approach used here was to develop a

simplified premixed flat flame model that takes account

of diffusion, then to calibrate this model against

a) literature data on free radical concentrations

and

b) our own base case of NO formation without sulfur

present.

It should be noted that the model therefore used only two

unknown parameters, one of which was obtained from the

open literature, the other of which uses our own base case

data. This model was then used to test kinetic mechanisms

of sulfur oxide - nitrogen oxide interactions, and the
1-66
-------
resulting mechanism was then used to determine the effect

of different environments and different heat loss rates

corresponding to those likely in a furnace.

Premixed Flat Flame Model

The salient features of our preliminary flat flame model

is that the diffusion in the ignition zone is assumed to

be such that it can be simulated by a (hypothetical) well

stirred stage or pointwise calculation. The ignition zone

is then followed by a plug flow calculation. This model

can be shown to be exact(16) only when the true concen-

tration profiles are parabolic. Since this is seldom

the case, the model should be regarded as only an approxi-

mate simulation of our flat flame. The volume of the

hypothetical stirred stage is determined by that which

allows a certain fraction of a species (designated "fuel")

to be converted.

The basic tool used was computer program REKIKET which in-

tegrates the conservation equations for well stirred and

plug flow reactors both with and without a heat balance.

In addition, this program allows the volume of the well

stirred stage to vary until a specified fraction of a

specified species is converted.
1-67
-------
The basic approach used to model the flat flame investi-

gated here consisted of the following steps:

* choose a kinetic mechanism for CH^/air combustion

• determine which value of percent CHi» consumed (de-

noted by x) during pointwise calculation led to

measured atom concentrations of Peeters and
• use this value of X to simulate our flat flame and

calibrate the heat loss parameters in the model

until the predicted and measured base case NO pro-

files matched

• investigate changes due to S02 addition, assuming

atom recombination catalysis with mechanisms and

rates proposed by Halstead and Jenkins'5) and

Merryman and LevyC18). No other parameters should

be altered in this phase.

Kinetic Mechanism

The methane air reactions used were those suggested by

Waldman it a£.,("*) in an EPA sponsored investigation of

kinetic mechanisms of methane/air combustion with pollu-

tant formation. A list of reactions is shown on Table 3.

The reaction named ULT36 through ULT143 denote the re-

actions numbered 36 through 143 in Table 7.5 of Reference

(4).
1-68
-------
In addition, for the runs simulating sulfur addition, the

atom recombination catalysis is described by reactions

shown in Table 4. Those reactions named MERLf and MERL3A

are reactions numbered 7 and 3A by Merryman and Levy(18'

while those labeled JENKI are from Halstead and Jenkinst5'.

Reaction JOHN11 is from Johnston's review of 0 atom

kinetics'l9'. These reactions demonstrate the catalysis of

0 atom recombination by S02 via S03 as an intermediate as

well as the catalysis of H atom and OH radical recombina-

tion to form HaO, with HSOa as an intermediate. No adjust-

ment of rate coefficient values from those suggested by

the original authors was made and it was assumed throughout

that:

k£/kr =

An important result obtained from this kinetic model is to

determine whether this atom recombination catalysis is

sufficient to account for the drop in NO emissions caused

by S02 addition. In addition, a kinetic calculation of

this type allows the separate effects of temperature pro-

file changes and radical concentration changes to be in-

vestigated. This should lead to greater insight into the

salient features involved.

Tables 3 and 4 follow
1-69
-------
TABLE 3. METHANE COMBUSTION MECHANISM
ULT36
ULT77
ULT84
ULT99
ULT101
ULT140
ULT44
ULT46
ULT47
ULT52
ULT59
ULT63
ULT65
ULT66
ULT70
ULT83
ULT85
ULT88
ULT91
ULT98
ULT100
ULT117
ULT125
ULT133
ULT135
ULT143
JOHN11
CHO
CO 2
H20
H
H
N2O
CHO
CH20
CHO
CHO
CH3
CH4
CH4
CH4
CO
H
H
OH
OH
H
OH
OH
N
N
N20
CHO
02
+M
+M
+M
+O
+02
+M
+H
+0
+OH
+0
+0
+O
+H
+OH
+OH
+OH
+H02
+H2
+N
+N20
+0
+OH
+NO
+02
+0
+02
+M
=CO
=CO
=OH
+M =OH
+M =HO2
=N2
=CO
=CHO
=CO
=CO
=CH20
=CH3
=CH3
=CH3
=CO2
=H2
=OH
=H
=H
=OH
=H
=H20
=N2
=NO
=NO
=CO
=O
+H
+0
+H
+M
+M
+0
+H2
+OH
+H20
+OH
+H
+OH
+H2
+H20
+H
+O
+OH
+H20
+NO
+N2
+02
+O
+0
+O
+NO
+H02
+0
+M
+M
+M

TABLE 4. S02 CATALYZED RECOMBINATION OF ATOMS AND RADICALS
MSRLf
MERL3A
JENKlJ
JENKI2
S02
S03
H
HS02
+0 +M
+0
+S02 +M
+OH
-S03
-S02
-HS02
-H20
+M
+02
+M
+S02
1-70
-------
Calibration of FlatFlame Simulation with Data of Peeters

In our simulation of a flat flame the Ignition zone is

simulated by a well stirred stage or pointwise calcula-

tion where the hypothetical volume is determined by that

volume which will convert a certain fraction X °f the

primary fuel. This is followed by a plug flow heat

balanced calculation. We settled on a value of x ^7

calibrating our simulation with the data of Feeters and

Mahnen^17'. Heat loss in the ignition zone was assumed

to be negligible. A value of

X = 0.98

was chosen because, as shown in Table 5:

• the maximum CH4 consumption rate was then

similar to that measured

• the peak 0 atom mole fraction (0.048) was of the

same order as that measured (0.025) compared to

the equilibrium 02 mole fraction which was two

orders of magnitude lower

• the temperature of the hypothetical well stirred

stage matched that measured at the maximum CHt,

consumption rate.
Table 5 follows
1- 71
-------
TABLE 5, SIMULATION OF FLAME OF PEETERS et al. (1973)
Simulation
Experiment
Temperature at
Ignition Zone Exit
Max. Rate of CHi+
Consumption
Ignition Zone Exit,
Mole Fractions
CO
0
OH
CH20
Max.O Atom
Mole Fraction
1569'K
1550°K
5.53 x 10
-s moles
cc sec
8.4 x 10
- 5 moles
cc sec
0.0336
0.0343
0.0173
0.00126

0.0474
0.042
0.011
0.015
0.001

0.025
1-72
-------
The simulation did over predict the atom concentration by

a factor of two and also tended to under predict the rate

of temperature increase. Obviously the simulation does

not give a true picture of the flat flame at this stage,

and the discrepancies are probably due to inaccuracies in

both the model and the kinetic mechanism. Nevertheless,

the simulation was considered sufficiently adequate to in-

vestigate the kinetic mechanism appropriate to S02 inhibi-

tion of nitrogen oxides.

This calculation also demonstrated that the kinetic

mechanism of methane combustion proposed by Waldman' '

did contain the salient features observed by Peeters and

Mahnen'17'. For example, the predicted formaldehyde,

hydroxyl and carbon monoxide profiles were reasonably

close to those measured. This gives both the kinetic

mechanism and the simulation some credence.

Calibration with Base Ca8e__N_0_.._Mea_au_r_e_d

The base cases used to test the kinetic model were those

with 104% stoichiometric air at both zero and 240°C

mixture preheat. We restricted our investigation to the

fuel lean regime because the dominant NO formation ki-

netics are there better understood. Using the value of

X = 0.98 determined previously and the kinetic mechanism
1-73
-------
for methane combustion shown on Table 3 it was found that

the NO measurements in our flat flame could be matched by

the simulation with a radiative heat loss coefficient

a « 3.45 x 10-1* cal/sec cm3 eK"

for the case with no preheat, and with

0 » 8.0 x 10"1* cal/sec cm3 "K1*

for the case with 24Q°C mixture preheat.

The discrepancy between these two values Indicate short-

comings in our model. However, since the purpose of our

model is to predict the change in NO due to S02 addition,

it is reasonable to calibrate against both the zero and

high preheat base cases individually. Obviously an im-

proved model should be able to predict the effect of

mixture preheat, without additional calibration.

Effect of SOg in Fuel

With 4.9% S02 in the fuel the simulation showed a drop of

49 ppm NO in the exhaust for the case with no preheat.

This compares with a measured drop of 40 ppm as shown in

Table 6. At 240°C preheat the simulation predicted a drop

of 51 ppm NO compared to a measured reduction of 70 ppm.

Given the Inaccuracies of the physical model, the kinetic

Table 6 follows
1-74
-------

en
H
13

•^3
>
Q
M
:=>
en
M
Q
55
25
O
H
«d
,j
!=»
53
H
»*
0
w
H
H
l"""4
x
(4

X
O
S5
C
O

C-J
O
Cn

fn
O

H
U
a
fn
ft,
W

US
TABLE
•o
woo
M U"> «H
3 H H
m
a
(3
0
4J GO <*
cC *^ O^
«-• i-\
3
B
•H
en

S
a.
ex

o
55
ca
«
^13
0)
n
(X
o
u
0)
N

M
•f-t
A a
• P,
x: o.
u
•H O
0 &
4J
0) •>
I-l
M 0
sr 3
O MJ
H
d
- *H
OJ
0}
o
ft 3
•H -0
It
6 H
O
•H 4->
Reduct
Percen
cs|
CO
O CM
O
PI
CM i-»
m co
e>i
S

•%
4J

01
A
V
O
o
O
•*
CM
n
1-1
B>

U
•H
O
8
a
o
o

c
0
• p.
• a. c
-H o
at
3
u-i
O -H
Z -M
O
C 3
(U
(0
(0

PQ
a>
U
3
•a
41
M
at
o
VI
0
Put
1-75
-------
rate coefficients, the simulation predicts the correct

effect of SOa addition with remarkable accuracy,

especially for the no preheat case. The discrepancy be-

tween theory and experiment in the high preheat case may

be due to an inaccurate simulation of the heat loss under

that condition. It is clear, therefore, that the S02

catalysis of atoms and free radicals as described by

Reactions MERL1 through JENKI2 on Table 4 can explain the

observed inhibition NO formation by S02.

Analysis of Simulation and Application

Calculated profiles of oxygen atom concentration, NO con-

centration and temperature are shown on Figure 15 for the

zero preheat case. It is clear that the addition of S02

to the fuel changes both the oxygen atom and the tempera-

ture profiles (assuming that the radiative heat transfer

coefficient remains unchanged). In the presence of S02

the higher temperature at early times is intimately

coupled with the drop in atom concentration, which is sig-

nificant. This indicates that before 10""* seconds the NO

formation rate is actually slightly higher with SOz than

with no SOz- However, during the time when most of the NO

is being formed and when radiative heat loss is important,

the drop in 0 atom concentration dominates, and the re-

sultant NO formed is significantly lower. It is clear,

Figure 15 follows
1-76
-------
1000
100
5
Q.
CL
to
Oxygen Atom x 10
10-4
0% S02 in fuel
4.9%S0£ in fuel
0 Measured NO
No preheat, radiation
heat loss
\
TIME (SECONDS)
10
r2
2100
2000
1900
1800
UJ
or
a:
UJ
0_
FIGURE is. S02 ADDITION AFFECTS BOTH OXYGEN ATOM

AND TEMPERATURE PROFILES, BUT LOWERS NO
1-77
-------
therefore, that the reason behind the observed effect is
that the presence of SOz catalyzes the recombination of
oxygen atoms, and that the drop in oxygen atom concentra-
tion is sufficient to lower the NO formation rate.

There is, however, a qualitative discrepancy between the
predicted profile of NO shown on Figure 15 and that
measured (at a different air/fuel ratio) and shown on
Figure 8. In general, the measured profile showed a more
rapid formation of prompt NO than that predicted. The
qualitative discrepancy is probably due to unknown
features in the mechanism of prompt NO formation. It is
felt, however, that this discrepancy is not serious and
does not detract from the point that 0 atom recombination
catalysis by 502 can explain the drop in exhaust NO
measured, with no adjustment to known rate coefficients
being necessary. Further details of the results from the
model are shown on Table 7 for the no preheat case. These
tables show the early formation of superequilibrium con-
centrations of SOa, which is an intermediate species in
the recombination catalysis scheme, followed by a decline
to relatively low values, corresponding to approximately
1% conversion of SOa to S03. Surprisingly, the calcula-
tions also indicate that the addition of SOa also appear
to hasten the CO burnout rate, although low CO levels were
Table 7 follows
1-78
-------
TABLE 7. SIMULATION DETAILS - NO PREHEAT
Time (sec)
0% S02 in Fuel
Ignition
Exit
0.11
0.125
0.482
0.329
0.133
0.780
0.140
0.365
0.996
0.280
0.425
4.9%
Zone
x 10~5
X
X
X
X
X
X
X
X
X
X
S02
Ignition
Exit
0.104
0.118
0.347
0.298
0.723
0.123
0.303
0.835
0.123
0.256
0.372
X
X
X
X
X
X
X
X
X
X
X
10-*
10-"
io-3
io-3
io-3
io-2
io-2
io-2
io-1
io-1
in Fuel
Zone
io-5
10-"
10~"
io-3
io-3
io-2
io-2
io-2
io-1
io-1
io-1
Mole
NO
0.479E-7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.488E-7
.720E-
.253E-
.404E-
.112E-
.118E-
.223E-
.537E-
.103E-
.140E-
.145E-
.622E-
.632E-
.883E-
.204E-
.388E-
.106E-
.178E-
7
6
5
5
4
4
4
3
3
3
7
7
7
6
5
4
4
.377E-4
.701E-
4
.819E-4
.956E-
.978E-
4
4
Fractions
0
0.458E-2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.484E-2
.555E-2
.465E-2
.236E-2
.343E-2
.158E-2
.119E-2
.749E-3
.454E-3
.197E-3
.119E-3
.396E-2
.415E-2
.466E-2
.386E-2
.158E-2
.994E-3
.755E-3
,480E-3
.282E-3
.212E-3
.903E-4
.457E-4
S03
--
—
--
--
—
--
__
--
--
__
—
0.384E-3
0.387E-3
0.325E-3
0.221E-3
0.321E-4
0.136E-4
0.887E-5
0.519E-5
0.425E-5
0.438E-5
0.534E-5
0.650E-5
CO
0.408E-1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.414E-1
.364E-1
.267E-1
.183E-1
.216E-1
.151E-1
-129E-1
.979E-2
.686E-2
.338E-2
.208E-2
.400E-1
.406E-1
.362E-1
.286E-1
.154E-1
.118E-1
.100E-1
.740E-2
.494E-2
.389E-2
.180E-2
.916E-3
Temp .
T°K
1659
1678
1754
1850
2003
1935
2055
2081
2099
2064
1939
1852
1686
1704
1783
1864
2046
2094
2113
2123
2090
2058
1959
1882
1-79
-------
obtained in all cases.

It is instructive to examine the effect of SOz in the fuel
on NO formation rates when there is zero radiative heat
loss, i.e., under adiabatic conditions. This is shown in
Figure 16 where it is apparent that under adiabatic con-
ditions the presence of SOz causes NO to reach its equi-
librium value more slowly. This simulation also indicated
that S02 causes a change in the time temperature history,
but that the primary effect was due to lower 0 atom con-
centrations .

In order to separate out kinetic and temperature effects
it is useful to determine the role of SOa under a speci-
fied time-temperature history. In this case there is no
attempt made to satisfy the heat balance, but rather, it
is assumed that the temperature and heat transfer are con-
trolled by the furnace configuration. A realistic tem-
perature history is one with an exponential temperature
drop from 2100°K to 1050°K in one second. This time-
temperature history is roughly representative of that felt
by a labeled volume of premixed gas and fuel as it combusts
and moves through the convection section of a furnace.
Thus this simulation can give some indication of what
might happen in a utility boiler, under conditions where
Figure 16 follows
1-30
-------
1000
0.
Q.
100
10
10
0%S02 in fuel
7 4.9% SOa in fuel
.3
IO-2 IO-1

TIME (SECONDS)
FIGURE 16. UNDER ADIABATIC CONDITIONS S02 DELAYS NO FORMATION
1-81
-------
fuel and air mixing is very rapid. Results are shown on

Table 8, and indicate that under such conditions fuel sul-

fur is likely to inhibit the formation of NOX.

CONCLUSIONS

It was found that both sulfur dioxide and hydrogen sulfide,

when added to a gaseous fuel, had a significant inhibition

effect on thermal NOX emissions. The presence of sulfur to

make about 6800 ppm S02 in the exhaust lowered NOX emis-

sions by up to 36%. Although it should be noted that these

results are valid for premixed gaseous flames , they do

imply that fuel desulfurization may lead to increased

(thermal) NOX emissions from combustion processes.

It appears to make very little difference whether the fuel

sulfur is introduced into the fuel as S02 or H2S, except

under fuel rich conditions where conversion of H2S to S02

is not rapid. This indicates that the inhibition of NO

formation by fuel sulfur occurs through mechanisms in-

volving SOj.

A preliminary model of a premixed flat flame showed that

this inhibition effect could be explained by the homoge-

neous catalysis of free radical recombination rates by

sulfur dioxide. This mechanism lowers the oxygen atom

Table 8 follows
1-82
-------
TABLE 8. FURNACE SIMULATION-PREMIXED MIXTURE, SPECIFIED

TEMPERATURE FALLING FROM 21QO'iK to 105Q°K IN ONE

SECOND. APPROXIMATELY 4% EXCESS AIR
NO ppm
Approximate Time

0.4 tn sec

1 m sec

2.5 m sec

Exhaust
Without sulfur

33.6

63.0

218.0

339.0
With sulfur
(4.9% S02 Fuel)

24.5

38.8

125.0

189.0
1-83
-------
concentration and, for a given radiative heat loss, lowers

NO formation rates. The effect of lower atom concentra-

tions is greater than the ensuing (coupled) temperature

Increase. Under conditions where the temperature was

fixed as an independent variable the predicted inhibitory

effect of SC>2 on NOx emissions was even larger.

Observations involving inhibition of NO formation by S02

can therefore be used to arrive at conclusions concerning

the role of superequilibrium oxygen atoms. Since inhibi-

tion was observed under fuel rich and fuel lean conditions

it appears that superequilibrium oxygen atoms play an

important role in both regimes. The effect of S02 was

especially pronounced on "prompt" NO formation, and this

supports theories that superequilibrium concentrations

oxygen atoms are a factor in the rapid formation of NO

early in the flame.

In addition, the experimental results showed that "prompt"

NO accounted for essentially all the NOX emissions under

fuel rich conditions and that it was not a strong function

of mixture preheat. This implies that NOX formation mech-

anisms other than those of Zeldovich are controlling early

in the flame.
1-84
-------
Current efforts at the University of Arizona are focused on

the effect of fuel sulfur on Fuel NO emissions and on

sulfur oxide nitrogen oxide Interactions In more practical

oil diffusion flames. Results are expected shortly. Then

it will be possible to gauge the practical importance of

adverse effects of fuel desulfurization.

ACKNOWEDGMENTS

This research was supported by the U.S. Environmental

Protection Agency under Grant R-802204. However, the

opinions and results presented do not necessarily reflect

the views of the Environmental Protection Agency. The

help and advice of W. Steven Lanler is gratefully

acknowledged.
1-85
-------
REFERENCES

1. Cullis, G.F., R.M. Benson, and D.L. Trimm, Proc. Roy.
Soc. (London) A295, 72 (1966).

2. Wendt, J.O.L., and C.V. Sternling, Comb. & Flame, 21,
387 (1973)

3. Thompson, D., T.D. Brown and J.M. Beer, Fourteenth
Symposium (International) on Combustion, p.787, The
Combustion Institute, (1973).

4. Waldman, C.H., R.P. Wilson, Jr., and K.L. Maloney,
"Kinetic Mechanism of Methane/Air Combustion with
Pollutant Formation". Environmental Protection
Technology Series Report EPA-650/2-74-045 (1974).

5. Halstead, D.J., and D.R. Jenkins, Trans. Faraday Soc.,
65., 3013 (1969).

6. Durie, R.A., G.M. Johnson, and M.V. Smith, Comb. &
Flame, 17. 197 (1971) .

7. Zeldovich, Y., Acta Physiochim. URSS 21, 577 (1946).

8. Bowman, C.T., Fourteenth Symposium (International) on
Combustion, p.729, The Combustion Institute, (1973).

9. Fenimore, C.P., Thirteenth Symposium (International)
on Combustion, p.373, The Combustion Institute,
(1971).

10. Wendt, J.O.L., C.V. Sternling, and M.A. Matovich,
Fourteenth Symposium (International) on Combustion,
p.897, The Combustion Institute (1973).

11. Sarofim, A.F., and J.H. Pohl, Fourteenth Symposium
(International) on Combustion, p.739, The Combustion
Institute (1973).

12. Fristrom, R., and A. Westenberg, "Flame Structure".

13. Brown, J.W., D.W. Pershing, J,H. Wasser, and
E.E. Berkau, "Interactions of Stack Gas Sulfur and
Nitrogen Oxides on Dry Sorbents", U.S. Environmental
Protection Series No. EPA-650/2-73-029 (1973).
1-86
-------
14. Ektnann, J.J., M.S. Thesis, Department of Chemical
Engineering, University of Arizona (1974).

15. Merryman, E.L., and A. Levy, "Nitrogen Oxide Forma-
tion in Flames: The Roles of N02 and Fuel Nitrogen"
Presented at Fifteenth Symposium (International) on
Combustion, Tokyo, (August 1974).

16. Sternling, C.V., and J.O.L. Wendt, "Kinetic Mechanisms
Governing the Fate of Chemically Bound Sulfur and
Nitrogen in Combustion". Environmental Protection
Technology Series, Report EPA-650/2-74-017 (1972).

17. Peeters, J., and G. Mahnen, Fourteenth Symposium
(International) on Combustion, p.133, The Combustion
Institute, (1971).

19. Johnston, H.S., "Gas Phase Reaction Kinetics of
Neutral Oxygen Species", National Stand. Ref. Data
Ser. , NBS 2Q_ (1968) .
1-87
-------
9:25 a.m.
Effects of Fuel Sulfur on
Nitrogen Oxide Emissions
Dr. Jost O.L. Wendt, University of Arizona
Jost, are you implying that your reduction is also
going to follow the same mechanism for the prompt
NO?
Well, what I am implying is that superequilibrium
atom and free radical concentrations are important
in the formation of prompt NO, but I am not implying
that, for example) the Zeldovich mechanism is the
mechanism to form prompt NO.
You're right, because I wonder if for your prompt NO,
there may be more likelihood of an OH type mechanism
accounting for the oxidation.
Absolutely. In fact, I should point out that when
you lower the oxygen atom concentration by whatever
mechanism, you tend to lower the free radical
concentrations of all other species as well, and so
there are secondary effects that may be important.
Q: In all your reported measurements you report NOX as
NO; are those actually just simple NO measurements
or measurements of total NOX reported as NO?
A: Those measurements are NO. It turns out we did look
for NO ; unfortunately we only had a stainless steel
converter on our chemaluminescent analyzer so we
could only do this in the fuel lean regime, and at
most we only found 3% of the NO was NO.. So we
found very little N00.
Q: Jost, did you say that the effect of added sulfur
was to reduce the oxygen atom concentration but to
1-88
-------
increase the temperature, so that the overall effect
was almost none?
A: No. I didn't say that. It turus out that your temperature
profile is a function of both the composition of a mixture
at a point and of the radiative heat loss that your flame
sees as a function of time. Now as you lower the oxygen
atom concentration, the initial temperature early on in
the profile will increase, but then the heat loss more
than compensates for that, so the drop in the 0-atom
concentration more than offsets any increase in tempera-
ture that you get. That's what the calculations showed
as well.
Q: I didn't quite get how you measure your oxygen atom
concentration; or do you just assume a level for them?
A: No, we did not measure that.
Q: Basically then, what you're saying is that if you don't
really know that, you also don't know whether or not you're
really catalyzing the recombination; I mean that's an
assumption.
A: Well, I think you're thinking about that second to last
slide, and that is the result of a calculation that we
made in which we modelled the chemical and physical
processes occurring in the flat flame. We used a fairly
detailed set of chemical reactions, actually those
recommended by Ultrasystems Corporation. So those curves
that you saw were the results of a calculation. What we
did was to calibrate a model with our base case and we
left it alone and used Merryman's and Levy's and Halstad
and Jenkin's data on S0_ atom recombination rates and wanted
to see what the effect of that would be. The effect
predicted is to lower the atom concentrations —
theoretically, that is.
Q: I want to just establish that all you really know for
fact is that addition of S0« lowered your NO,, or NO.
A: As an experimental fact that is all we know.
1-89
-------
Q: In reading some of those charts, although I couldn't
see them too clearly, the fifth or fourth one before
the end looked to me like the NO concentration peaked
at about 5% excess air.
A: Right.
Q: And that reading out to about 20% excess air, it looked
to me like the concentration was approximately half.
It looked like the addition of the 20% excess air
dropped the concentration much more radically.
This would imply a lot more free oxygen and yet the
results you foundwere that the reduction of the free
oxygen reduced the NO; I do not quite put those two
facts together.
A: The first thing you have to realize is that these
results are for pre-mixed laminar flames; in other
words, you can't compare our results with results
from diffusion flames or typical burners. The base
case results are in line with what other people have
gotten. It's quite normal to see the nitrogen oxides
be decreased rapidly as you increase the excess air.
This is a thermal temperature effect. In terms of
your other question about free oxygen, I think we
will have to discuss that later.
1-90
-------
TWO-DIMENSIONAL COMBUSTOR MODELING*
By:

R. C. Buggeln and H. McDonald

United Technologies Research Center

East Hartford, Connecticut 06108
*Work sponsored in part by the Environmental Protection Agency,
Research Triangle Park, North Carolina under Contracts 68-02-1092
and 68-02-1873
1-91
-------
ABSTRACT

The equations that describe the conservation of mass, momentum,
and energy in the flow fields of industrial furnaces are presented
and the concepts of stream function and vorticity are introduced.
The stream function-vorticity formulation reduces the number of
governing equations to be solved by one, and at the same time
removes the troublesome pressure terms. In the present formulation
each equation has the same general form, a feature which is
convenient for ease of solution. A numerical procedure is then
described which transforms each equation into a series of linear
algebraic equations suitable for solution on a high-speed computer.
Coupling and nonlinear effects are accounted for by using an
iterative technique. Once the solution of the flow field is
completed, the same numerical procedure is utilized to solve for the
NO levels.
1-92
-------
INTRODUCTION

Events of the last decade, especially in the areas of pollution
control and fuel consumption, have placed a considerable burden on
the present-day designers of industrial furnaces. Prediction of
industrial furnace performance and pollutant emission characteristics
require modeling procedures possessing a high degree of sophistica-
tion in the areas of chemistry and fluid dynamics. Methods of
modeling furnaces have generally been less than completely satis-
factory largely due to a lack of understanding of the fundamental
flow processes which, through heat, mass, and momentum exchange,
directly effect pollutant emission and combustion efficiency. For
example, it has been shown (Refs. 1 and 2) that stability and
combustion intensity of flames as well as residence time in furnaces
are strongly influenced by swirling flow. Residence times and flame
stability and intensity are, in turn, related to furnace performance
and efficiency as well as to pollutant emission formation (Refs. 3
and k). Thus, because of the strong influence of flow processes on
combustion characteristics, it is particularly important that the
fluid mechanics be properly taken into account in the development of
furnace modeling techniques.

Most previous procedures employed for modeling combustion
processes have generally been highly simplified, particularly in
regard to flow modeling where stirred reactor concepts and one-
dimensional assumptions are employed. As a general criticism, it
seems that these previous methods are lacking in their ability to
properly account for mixing and reverse flow phenomena occurring
in the recirculation zone of an industrial furnace. In contrast to
these one-dimensional, stirred reactor models, the recently developed
computational method of Gosman, et al., (Refs. 5 and 6) has demon-
strated the feasibility of making computations in combustors that
have recirculation zones. However, the Gosman procedure utilizes
a relatively inefficient point-by-point (Ref. 5) or line-by-line
(Ref. 6) relation technique rather than a field relaxation procedure.
In order to achieve a more efficient computational scheme a field
relaxation technique, the field relaxation elliptic procedure (FREP),
was developed at United Technologies Research Center (UTRC).
1-93
-------
GOVERNING EQUATIONS

The set of elliptic partial differential equations that describe
the combustion system under consideration are based on the conserva-
tion laws of mass, chemical species, momentum and energy (Ref. 7).
For simplicity and generality, these equations are expressed in
vector notation below; however, since the present study is concerned
in particular with axisymmetric flow, future discussion will concen-
trate on the axisymmetric form of these equations. In the present
effort, effects of thermal diffusion, pressure diffusion, and
forced diffusion upon mass transport as well as the effects of bulk
viscosity in the momentum equations are assumed to be negligible.
Fick's law is presumed to be valid throughout, implying equal binary
diffusion coefficients. With these assumptions in mind, the
governing equations can be written as:
Continuity
V • pv = 0
(1)
Conservation of Gas Phase Species

/ ^eff \
V- pm.V- V- ( —-— Vm. - r. = 0
I \ 5C I / 1

Conservation of Liquid Phase Species
(2)
(3)
Conservation of Momentum
[Vv
VP = 0
00
1-94
I
-------
Conservation of Energy
(5)
In addition to the above conservation equations, it is necessary to
specify a gas law which in the present procedure is given by
P =
where the mixture molecular weight, M, is defined by
(6)
M =
m.
M.
Finally, the stagnation enthalpy, H, is defined by
(7)
H =
Ivl2
I m. [ fTC
-------
based upon an empirical knowledge of these parameters in turbulent
flows of gases.

When the primitive flow variables, p , V, and P are employed,
experience has shown the explicit appearance of the pressure terms
in the momenta equations can lead to numerical difficulties. The
problems associated with pressure have often been eliminated (for
two-dimensional and axisymmetric flows) by the formation of the
vorticity transport equation, an equation obtained by taking the
derivative of the first momentum equation with respect to the second
coordinate direction and taking the derivative of the second momentum
equation with respect to the first coordinate direction, and then
subtracting the first equation from the second. In this manner the
explicit appearance of the pressure terms is eliminated as the
momenta conservation equations are replaced by a transport equation
for the vorticity, cj, where, by definition, W= V X V. For axi-
symmetric flow the vorticity vector is in the azimuthal direction and
is given by
Of =
av
~dx~
(9)
After much rearrangement, the vorticity equation emerges in the form
do)
2 ..2
Jf^<^>£]-'=»
where Sw is defined by
(11)
1-96
-------
and where lx and lr are unit vectors in the axial and radial direc-
tions. It may appear that little has been achieved by introducing
the concept of vorticity since Eq. (10) still has the two velocity
components as unknowns. However, it should be noted that the incon-
venient pressure terms have been eliminated through the introduction
of an additional unknown, the vorticity. Further advantage of this
formulation is obtained by introducing the stream function relation-
ships to close the system. The stream function, \fj, is related to the
velocity components u and v by
I
(12)
dr
and
(13)
Substitution of Eqs. (12) and (13) into Eq. (l) shows that the stream
function by definition satisfies the continuity equation, while the
combination of Eq. (9) with Eqs. (12) and (13) yields the stream
function equation
(pr dr
(cl
4- —- ( — ) = —
dr pr d* T
(lU)
When Eqs. (10) and (ik) are used to replace the continuity and the
axial and radial momentum equations there is thus a dual beneficial
effect: (i) the awkward pressure terms are eliminated and (ii) the
number of equations is reduced by one, i.e., three equations are
equivalently replaced by two. After solving Eqs. (10) and (1*0 (and
Eqs. (2), (3), (5), and Eq. (U) for the swirl component of velocity),
the variables that were eliminated by the introduction of stream
function and vorticity, u, v, and P can be reconstructed. The axial
and radial velocity components are computed from Eqs. (12) and (13).
The pressure gradients are calculated from Eq. (U), the conservation
form of the momentum equation, and then integrated to yield the
pressure distribution.
1-97
-------
It is also convenient to replace the axial and radial velocity
components in the species conservation equations, Eqs. (2) and (3),
the swirl equation, Eq. (U), and the energy equation, Eq. (5), by
the stream function formulation of velocity. When this is done, it
is seen that all of the partial differential equations solved in this
study are of the general form
dr
= 0
(15)
where the nonlinear coefficients a^, b$, b^ , and c$ and the source
term S^ are tabulated in TABLE I for each of the dependent variables,
^, under consideration. The generalized form of all the equations to
be solved is a very convenient feature for ease of application to any
numerical technique. The general form given above was first
suggested by Gosman, et al., (Ref. 5) and for convenience was adopted
for use in the present work.

COMPUTATIONAL ANALYSIS

In the present study, solution of the set of coupled elliptic
partial differential equations describing the flow inside a furnace
is achieved through a field residual relaxation procedure. The
procedure is discussed in detail in a soon to be published EPA report
and is reviewed briefly here. In the procedure a finite difference
analogy is first constructed at interior grid points for each solu-
tion variable by replacing the derivatives of Eq. (15) by second-
order central difference formulae and rearranging the difference
equations into the form
(16)
1-98
-------
TABLE I

COEFFICIENTS OF GENERAL ELLIPTIC EQUATION
<£
ll/
T
U>
r

rw
m.
i
f.

°*
o

.2
r

1.0
t.O
1.0

1.0

b*.
1
Jr*
,2
T

M.ff''
^eff
Sc
^eff
Sc
A*eff
Pr

b*2
I
^
r2
r

^effrZ
^eff
Sc
^eff
Sc
^eff
Pr

C*
1 O

Meff

i
r«
t.O
1 .0

1.0

s*
OJ
r
d (£>w2)-r[ d (^*V* }df>
dx ^W } rUx l 2 ]dr
d u2*!/2 a/> I ,
ar ( 2 Ja» J •' "w
0
0
-r.
i
o
1
' d \ u. r((l ' i ^ (V* )
r ax L^effrU'-pr 'ax ( 2 }
+ ?<^>
^L|] _ i d r r
i ax JJ r ar L^-eff
/( , ' x d , V% .
V ' Pr ' dr ( 2 } *
V, I 1 v ^ drniH
t(ss-pr)hiV/J + 0«
1-99
-------
The c's represent a collection of terms that account for the convec-
tion and diffusion of the variable 0 throughout the mesh system. In
general, the c's may be functions of the 0's as well as other
dependent variables. Thus, Eq. (l6) represents a system of nonlinear
coupled equations. Linearization of Eq. (l6) is accomplished by the
utilization of a Newton-Raphson procedure, as follows. First a
guess, 0n.H, is made for each solution variable at each grid point.
Using this guess, a residual of the form
n
R.
n . n
= c.*.
(17)
is defined. The Newton-Raphson procedure seeks to drive the
residual at the n+lst level to zero by expanding the residual in a
Taylor series about the residual level n. This results in the
equation
n n+i n n

(18)
Derivatives of the residual with respect to the solution variables
are replaced with the appropriate c's, e.g.,
i-'.j
for which n-level values are available. Substitution and
rearranging leads to a linear form of Eq. (l8), viz.
.n+i
1 + 1
. .
i + l,
(19)
I-100
-------
Equation (19) is constructed at interior grid points for each
selection variable. If the grid mesh were of size MxN, this would
result in the need to solve MxN simultaneous linear equations to
update each solution variable, $, to the n+1 st iteration level.
For small values of M and N this could readily be accomplished;
however, for larger values of M and N a more efficient means is the
Alternating Direction Implicit (ADl) procedure (Ref. 9)- Once each
solution variable is updated to the n+1 st level, new mesh values of
the c's can be calculated for each equation and the overall solution
procedure can be performed to update the variables to the n+2 nd
iteration level. This overall solution procedure is repeated until
the desired degree of convergence is achieved.

CALCULATION OF WO LEVELS

Once a converged flow field solution is available, it is now
possible to solve the NO mass fraction equation for the nitric oxide
levels. The form of this mass fraction equation is the same as the
form of Eq. (15), hence the same general solution technique can be
utilized. The source term of this mass fraction equation, rWQ, re-
quires an expression for the rate of creation of nitric oxide. A
generally accepted reaction model for nitric oxide formation and
decomposition in post-flame gases is that proposed by Lavoie, at al.,
(Ref. 10) which consists of the following six reactions:
N2 + 0 ^ NO + N

N + 02 ^ NO + 0

N + OH =i WO + H

H + N20 ^ N2 + OH

0 + N20 ~ N2 + 02

0 + N20 — NO + NO
(20)

(21)

(22)

(23)
(25)
1-101
-------
The first two reactions, Eqs. (20) and (2l), form the Zeldovitch
mechanism (Ref. 11) and are considered to be the principal nitric
oxide formation reaction mechanism. The two reactions together with
the third reaction, Eq. (22), which assumes minor importance under
fuel-rich conditions, form the extended Zeldovitch mechanism which is
employed in the present study. At low temperatures, when nitric
oxide concentrations are much greater than equilibrium values, the
fourth, fifth, and sixth reactions, Eqs. (23) through (25), involving
N20 as an intermediary, may become important; however, because overall
reaction rates are so low, the net effects of these reactions are
probably negligible. Therefore, these last three reactions are
neglected in the present study. The results of Bowman and Seery
(Ref. 12), as well as other investigators (Ref. 13) in investigation
of nitric oxide formation kinetics in combustion processes, lend
support to this nitric oxide formation model.

Use of the above information or the 'Law of Mass Action1 enables
one to derive an expression for the term r™. r^Q is a function of
the temperature (which at this point is assumed to be unaffected by
the level of NO) and the concentrations of N2, 0, N, and (%> (which
are assumed to be the equilibrium values). Since the velocity,
temperature, diffusion coefficient, etc. do not vary during the
solution of this equation, there is no coupling; however, nonlinear
effects enter the equation through the source term and hence an
iterative procedure must also be used to solve this equation.

Solution of the NO equation along with the previously solved
flow field thus gives the designer the capability of being able to
predict both the efficiency and pollutant levels from industrial
furnaces. For typical cases, a converged solution can be achieved
in approximately one hour of UNIVAC 1110 computer time. Wide ranges
of flow conditions can be considered at a relatively low cost,
certainly much more inexpensively than by performing an experiment.
The numerical procedure is very general and additional equations,
geometrical considerations, etc. can be included in a relatively
short period of time.
I- 102
-------
LIST OF SYMBOLS
CP

OR
0
r
Coefficient in general elliptic equation, Eq. (15)

Coefficient in general elliptic equation, Eq. (15)

Coefficient in general elliptic equation, Eq. (15);

collection of convection and diffusion terms

Species specific heat

Mixture specific heat

Liquid particle mass fraction

Heat of formation

Stagnation enthalpy

Unit vector in axial direction

Unit vector in radial direction

Gaseous mass fraction

Mixture molecular weight; axial grid locations

species molecular weight

Radial grid locations

Pressure

Prandtl number

Radiation flux

Radial coordinate

Production term in species conservation equation
1-103
-------
R

u
Universal gas constant; residual

Source term

Schmidt number

Group of terms in vorticity equation (see Eq. (ll))

Absolute temperature

Axial velocity

Radial velocity

Time-averaged velocity

Swirl velocity

Axial coordinate
Greek Symbols
6

0
Kronecker delta

Normalized vorticity (cu/r)

Viscosity

Density

General function

Stream function

u) Vorticity

7 Nabla operator
1-104
-------
Subscripts

eff Imparts molecular (laminar) and turbulent characteristics

i Gas phase component; axial grid node

j Liquid phase component; radial grid node

o Reference

0 General function

01 General function axial component

$2 General function radial component

Superscripts
n
Iteration level
Reference
Transpose of a tensor
1-105
-------
REFERENCES

1. Beer, J. M. and N. A. Chigier: Stability and Combustion
Intensity of Pulverized Coal Flames - Effect of Swirl and
Impingement. Journal of the Institute of Fuel, December 1969.

2. Beer, J. M. and W. Leucker: Turbulent Flames in Rotating Flow
Systems. Paper No. Inst. F-NAFTC-7, Ncrth American Fuel
Technology Conference, Ottawa, Canada, 1970.

3. Beer, J. M. and J. B. Lee: The Effects of Residence Time
Distribution on the Performance and Efficiency of Combustors.
The Combustion Institute, 1965, pp. 1187-1202.

U. Marteney, P. J.: Analytical Study of the Kinetics of Formation
of Nitrogen Oxide in Hydrocarbon - Air Combustion. Combustion
Science and Technology, Vol. 1, 1970, pp. 37-^5-

5. Gosman, A. D., W. M. Pun, A. K. Runchal, D. B. Spalding, and
M. Wolfshtein: Heat and Mass Transfer in Recirculating Flows.
Academic Press, New York, New York, 1969.

6. Gosman, A. D. and W. M. Pun: Calculation of Recirculating
Flows. Lecture notes, Imperical College of Science and
Technology, London, England, 1973.

7. Bird, R. B., W. E. Stewart, and E. N. Lightfoot: Transport
Phenomena. John Wiley & Sons, Inc., New York, New York, 1960.

8. Launder, B. E. and D. B. Spalding: Turbulence Models and Their
Application to the Prediction of Internal Flows. Imperial
College of London, Technical Engineering Department Report No.
TM/TN/A/18, 1971.

9. Smith, G. D.: Numerical Solution of Partial Differential
Equations. Oxford University Press, London, England, 1965.

10. Lavoie, G. A., J. B. Heywood, and J. C. Keck: Experimental and
Theoretical Study of Nitric Oxide Formation in Internal
Combustion Engines. Combustion Science and Technology, Vol. 1,
1970, pp. 313-326.
1-106
-------
11. Zeldovitch, Ya. B., P. Ya Sadounikov, and D. A. Frank-Kamenetskll:
Oxidation of Nitrogen in Combustion. Academy of Sciences of
USSR, Institute of Chemical Physics, Moscow-Leningrad, 19^7.

12. Bowman, C. T. and D. J. Seery: Investigation of NO Formation
Kinetics in Combustion Process: The Methane-Oxygen-Nitrogen
Reaction, Emissions from Continuous Combustion Systems.
Plenum Publishing Company, New York, New York, 1972.

13. Caretto, L. S., L. H. Muzio, R. T. Sawyer, and E. S. Starkman:
The Role of Kinetics in Engine Emission of Nitric Oxide.
Combustion Sciences and Technology, Vol. 3, 1971.
I- 107
-------
1-108
-------
EFFECTS OF INTERACTION BETWEEN FLUID DYNAMICS AND
CHEMISTRY ON POLLUTANT FORMATION IN COMBUSTION
C. T. Bowman, L. S. Cohen
L. J. Spadaccini and F. K. Owen
United Technologies Research Center
East Hartford, Connecticut 06108
*Work sponsored in part "by the Environmental Protection Agency
Research Triangle Park, North Carolina under Contracts 68-02-1092
and 68-02-1873
1-109
-------
ABSTRACT

Average concentration levels of the pollutants NO, HOg, CO and
unhurried hydrocarbons (THC) were measured at the exhaust of an
axisymmetric combustor over a significant range of operating condi-
tions. In addition, detailed species concentration, temperature
and velocity maps were obtained throughout the combustor for repre-
sentative operating conditions. Major combustor input parameters
were varied over the following ranges: overall fuel/air equivalence
ratio 0.5 - 1.3, air-fuel velocity ratio 0.1 - kQ, inlet air swirl
number 0 - 0.6, air flow rate 0.09 - O.l4 kg/sec, inlet air tempera-
ture 730 - 860 K and combustor pressure 1-7 atm. Water-cooled
probes were used to remove samples from the flow for on-line concen-
tration analysis and to measure temperature. Velocity and flow
direction were measured using both an impact-static pressure probe
and a laser velocimeter.

Elevated pressure and introduction of swirl to the extent
considered in the experiments creates "unmixedness" in the com-
bustor flow field, producing high local temperatures, which in
turn result in enhanced NO formation and consumption of hydrocarbons.
Aerodynamic flame stabilization, achieved without benefit of swirl
or physical flandholders in systems having large air-fuel momentum
flux ratios, produces strong stirring which results in reduced
temperatures and in relatively low NO formation and hydrocarbon
consumption rates.
I-UO
-------
INTRODUCTION

Recent experimental investigations of factors affecting pollu-
tant emissions from various continuous combustion devices (Refs. 1-
have shown that changes in operating conditions, which alter the
flow pattern in the combustion chamber, can have a substantial
effect on pollutant emissions. These observations suggest that
coupling between fluid dynamic and chemical processes inside the
combustion chamber is a major factor in governing pollutant emis-
sions from many combustion devices. Analytical studies of reacting
flow fields (Refs. 5-8) have indicated in a qualitative manner how
mixing and turbulence can influence pollutant formation. However,
at the present time our understanding of the nature of the coupling
between fluid dynamic and chemical processes is insufficient to
'permit extrapolation of results obtained from one combustion device
to other devices or to allow a_ priori determination of the effects
of changes in operating conditions on pollutant emissions.

An experimental investigation of the interaction between fluid
dynamics and chemistry in a combustor and the subsequent effects
on pollutant formation and destruction has been carried out. The
overall objectives of this investigation are:

(i) To examine experimentally the interaction between
fluid dynamic and chemical processes inside a model
combustor as operating conditions are varied, and
to correlate changes in pollutant emissions with
these variations.

(ii) To compare the experimental results with results
from a combustor flow analysis to assist in inter-
pretation of the experimental results and to
evaluate and calibrate the theoretical model.
EXPERIMENTAL APPARATUS

The experimental configuration consists of an axisymmetric
combustor in which a central gaseous fuel stream mixes with a
coaxial annular air stream. In this configuration, which resembles
the primary combustion zone in many practical combustion devices,
the resulting flame will not be stabilized in general, and blow-off
I- 111
-------
will occur. In order to stabilize a flame in such a system,it is
necessary to provide a continuous ignition source either by means
of a pilot flame or through creation of zones of hot recirculating
combustion gases. These stabilizing recirculation zones may be
created by the introduction of physical flameholders into the flow
or by imparting a swirl component to the flow. Swirl stabilization
techniques have been used extensively in continuous flow combustion
devices such as furnaces and gas turbines. Hence, a significant
portion of the current effort focuses on swirl-stabilized flames.

A new approach to flame stabilization also was employed. A
review of recent literature on mixing of coaxial non-reacting jets
indicated that flame stabilization might be achieved in gaseous-
fueled systems by judicious selection of the momentum flux ratio of
the fuel and air streams. This flameholding approach relies neither
on a physical flameholder nor on swirl. Rather, it makes use of
the observation that a recirculation zone is formed in the initial
region of coaxial streams when the outer, annular stream (air
stream) is moving very rapidly relative to the central (fuel) jet.
The aerodynamic flameholding technique, in principle, should result
in a simplified determination of inlet conditions and also should
provide a unique opportunity to contrast swirling and non-swirling
flow fields without the disruptive presence of a physical flame-
holder.

Tests were conducted in an instrumented, water-cooled combus-
tion system, Fig. 1. The facility design placed particular emphasis
on acquisition of species concentration, temperature and velocity
distributions throughout the reacting flow field. Air from an
electrical heater section flows around a replaceable fuel injector.
Swirl may be imparted to the air stream by inserting various swirl
vanes into the annular air passage. Observation of the combusting
flow is made through window ports in the entry and combustor
sections, Fig. 1. The combustor probes used to make these measure-
ments are compatible with all ports and may be replaced by a quartz
window for laser velocimetry measurements. The 12.23 cm diameter,
100 cm long instrumented combustor is divided into three water-
cooling zones of approximately equal length. Flow exhausts from
the coribustor to the facility exhaust stack. Chokes can be installed
downstream of the extender to raise the pressure in the combustor.

An exhaust probe rake is placed at the exit of the combustor
1-112
-------
to determine concentrations of pollutant species in the exhaust
flow. Gas composition is determined on-line using redundant sampling
instrumentation. Temperature distributions in the combustor were
obtained using both a traversing calibrated-heat-loss thermocouple
probe and a double-sonic-orifice probe. Flow velocity and direc-
tion in the combustor were determined using a five-hole impact-
static pressure probe. This probe performed satisfactorily through-
out the combustor flow field except in those regions, near the
injector, with high turbulence levels or reverse flow. In these
regions, flow velocity and direction were measured using a laser
velocimeter.
EXPERIMENTAL RESULTS

The experimental program was comprised of input-output tests
and flow-field mapping tests. In the input-output tests, the
influence of various combustor input parameters on pollutant con-
centration levels in the exhaust flow was determined. Generally,
for a particular fuel injector and swirl vane insert, the overall
fuel-air equivalence ratio was varied from fuel-lean to fuel-rich
conditions at two or more levels of inlet air temperature, air flow
rate and combustor pressure, and the composition of the exhaust
flow was measured for each operating condition.

Results from the input-output tests, discussed in detail in
Ref. 9j indicate that the pollutant emission levels are particularly
sensitive to certain combustor input parameters. NO and THC emis-
sions are strongly dependent on swirl and pressure and also are
dependent on air-fuel velocity ratio. Over most of the ranges of
combustor input parameters investigated, exhaust CO concentrations
are the equilibrium levels; however, for non-swirling flows and for
swirling flows with low air-fuel velocity ratios, exhaust CO levels
can significantly exceed equilibrium levels. To investigate how
changes in these significant input parameters influence the combustor
flow field and subsequent pollutant formation, detailed maps of
the flow field properties were obtained.

In these mapping experiments, radial distributions of flow
velocity and direction, temperature and species concentration were
acquired at a number of axial locations in the combustor. Based
on results from these experiments, several general observations
1-113
-------
were made concerning the interaction "between the flow field and
the pollutant formation and destruction processes, Ref. 9- For
the conditions investigated, the primary hydrocarbon combustion
zone, where most of the hydrocarbon fuel reacts to form CO, is
located relatively close to the injector. In this zone, which
contains the stabilizing recirculation zone, a significant fraction
of the total available chemical energy is released, and high levels
of CO and NO are established. Peak CO levels in the primary zone
are significantly larger than final equilibrium levels. Measured
N02 concentrations in the primary combustion zone are relatively
large. Near the injection plane, N02 concentrations in excess of
the NO concentrations were measured in the shear region separating
the inner and outer flows. Downstream from the primary combustion
zone, the remaining hydrocarbon fuel and the CO oxidize. There are
no significant increases in temperature in this post-combustion
zone due to heat loss to the combustor walls; however, because of
the relatively high temperatures, significant additional NO forma-
tion occurs. In the post-combustion zone, the N02/N0 ratio de-
creases, primarily because of the increasing NO concentration.

In the present experiments, the reacting flows can be character-
ized roughly as chemistry-limited or mixing-limited, depending on
the relative rates of chemical reaction and mixing. In non-swirling,
atmospheric pressure flames, large scale fluctuations of the stabili-
zing recirculation zone enhance mixing of the fuel and air streams and
result in a primary combustion zone which resembles a stirred reactor.
Downstream from the primary zone, the flow resembles plug flow. These
flows appear to be chemistry-limited in that mixing rates are rapid
relative to chemical reaction rates. As a result of the well-mixed
nature of the flow, temperatures are moderated so that hydrocarbon
and CO oxidation rates and NO formation rates are relatively slow.
Hence, chemistry-limited flows are characterized by relatively high
THC and CO emissions and relatively low NO emissions.

Increasing combustor pressure or imparting swirl to the inlet
air stream results in a shift from chemistry-limited to mixing-
limited behavior. In mixing-limited flows, the time-mean flow field
exhibits the characteristics of a diffusion flame, with separation
of the fuel and air streams and with chemical reaction occurring
principally in an annular region surrounding the combustor centerline.
Maximum temperatures in the region are considerably larger than found
1-114
-------
in the chemistry-limited flows, resulting in increased hydrocarbon
and CO oxidation rates and HO formation rates. Hence, mixing-
limited flows are characterized by relatively low TEC and CO emis-
sions and relatively high NO emissions.

Laser velocimeter data and high-speed motion picture films
of the reacting flow field in the vicinity of the injection plane
reveal significant fluctuations in the flame structure. Recent
analytical studies of turbulent flames (Ref. 5) suggest that these
fluctuations will significantly influence pollutant formation in
the cotnbustor.

Initial comparisons of measured distributions of mean velocity
and mean temperature with those obtained from the combustor flow
analysis show relatively good agreement. However, the agreement
of measured and calculated pollutant species concentration profiles
in the combustor is less satisfactory. The poorer agreement may
result from the effect of turbulent fluctuations on pollutant
formation and destruction, discussed above, or from the use of a
relatively simple hydrocarbon combustion model in the combustor
flow analysis.
1-115
-------
REFERENCES

1. a) M. P. Heap, T. M. Lowes and R. Valmsley: The Effect of
Burner Parameters on Nitric Oxide Formation in Natural Gas
and Pulverized Fuel Flames, Paper presented at First Ameri-
can Flame Days Meeting, Chicago, Illinois, September 1972.

"b) M. P. Heap, T. M. Lowes and R. Walmsley: Emission of Nitric
Oxide From Large Turbulent Diffusion Flames, Fourteenth
Symposium (international) on Combustion (The Combustion
Institute, Pittsburgh, 1973), P. 883-

2. D. R. Shoffstall and D. H. Larson: Aerodynamic Control
of Nitrogen Oxides and Other Pollutants from Fossil Fuel
Combustion, EPA Report 650/2-T3-033a, October 1973.

3. a) A. M. Mellor, R. D. Anderson, R. A. Altenkirch and J.
H. Tuttle: Emissions from and within an Allison J-33
Combustor, Combust. Sci. and Tech. 6, 169 (1972).

b) J. H. Tuttle, R. A. Altenkirch and A. M. Mellor:
Emissions from and within an Allison J-33 Combustor II:
The Effect of Inlet Air Temperature, Combust. Sci. and
Tech. J_, 125 (1973).

^-. R. E. Jones and J. Grobman: Design and Evaluation of
Corabustors for Reducing Aircraft Engine Pollution, AGARD
Document CP-125, April 1973-

5. F. C. Gouldin: Role of Turbulent Fluctuations in NO Forma-
tions, Combust. Sci. and Tech. 9, 17 (197*0 •

6. R. F. Anasoulis and H. McDonald: A Study of Combustor
Flow Computations and Comparison with Experiment, EPA
Report 650/2-73-014-5, December 1973-

7. L. S. Caretto: Modeling Pollutant Formation in Combustion
Processes, Fourteenth Symposium (international) on Com-
bustion (The Combustion Institute, Pittsburgh, 1973)*
p. 803.

8. D. B. Spalding: Mathematical Models on Continuous Com-
1-116
-------
9.
busticm, Proceedings of the Symposium on Emissions from
Continuous Combustion Systems (Plenum Press, New York, 1972)
P. 3.

C. T. Bowman and L. S. Cohen: Influence of Aerodynamic
Phenomena on Pollutant Formation in Combustion, EPA Report
650/2-75-06la, July 1975.
I- 117
-------
Fig. 1
0)
o

Off
g
DC
91*

O
a a
-------
10:20 a.m.
Effects of Interaction Between
Fluid Dynamics and Chemistry
on Pollutant Formation in
Combus t ion
Dr. Craig T. Bowman, United Technology Research Center
Q: The question is related to the interpretation of thpse
fine experimental results, in a sense that, if we
take the swirl for instance, you show for the highest
swirl the NO concentrations are reduced.
A: No, increased.

Q: Then, the question is what kind of heat transfer did
you have on your combustor? Was it a hot refractory
wall or was it a r-:ater-cooled wall?
A: It is a water-cooled segmented combustor; we have
three segments along the combustor, we measure the
temperature in, the termperature out, and we measure
the wall temperature as well. The amount of transport
that is out of this particular combustion is rather
small. We maintain a hot wall but not an adiabatic
wall, and we feel that we are taking out something on
the order of 5-10% of the total enthalpy of the
flow over the entire length of the combustor for
these conditions.
There have been a number of cases in the literature
of periodicity of flow patterns such as precessing
vortices. Tom, you mentioned how important it is
to allow for turbulence; I wonder if you see any
evidence of periodicity of flow and if there are
periodic flows, how do you see the model handling
these?
Yes, we have seen the precessing vortex core which
has been reported by Syrec, Beer and others. What
1-119
-------
Q:

A:
Adel is referring to here is that with the swirling flow
there is a vortex which processes around the fuel injector,
goes around and around at some well-defined frequencies, and
in fact the flame itself stabilizes on that precessed
vortexing core. The high speed motion picture films which we
have taken show, in fact, that in swirling flow the flame is
helical, and that it is processing at a rate which is equal
to the precession rate of the core. In most of the flows
which we have examined, this rate of precessing has been
between 20 and 150 hertz. We have seen it very clearly on
the high speed films. It means that what is normally a time
mean-steady flow is really not that at all. It's actually
three-dimensional and time dependent. So this is going to
have a very significant impact on analytical modeling. I
am going to beg the issue by saying that what one would do is
probably best answered by Harry McDonald. Do you want to
comment on that, Harry?

[Mr. McDonald]
The only observation I can make at this point is that the
turbulent fluctuations can be represented in terms of corre-
lation coefficients in the various chemical reaction
equations and in the various equations of fluid mechanics.
The central problem of most of the active research in this
area at the present time is how to adequately model these
correlations and that particular subject is one that has to
be measured at great length.

Tom, you mentioned that the measured C0« concentrations were
those given by chemical equilibrium.

Very nearly.

Could you comment on how this could be and what significance
it might have?

Under certain conditions we find that the CO concentration is
its equilibrium value, and specifically we find this under
the fuel rich combustion conditions. In the case of fuel
lean combustion, we find that the CO concentrations are
greater than equilibrium; however, the actual levels of CO
1-120
-------
Q:
A:
Q:
A:
within the flow under lean conditions are very, very much
lower than they are under rich conditions. So whereas the
CO concentration itself is not in equilibrium, the CC>2 is
very nearly its equilibrium. Obviously it's not exactly
equilibrium because you have carbon tied up in both species,
so I was trying to deal with it in a more simplistic sense.
Obviously, there has to be a carbon balance within the flow.
Could you tell us briefly what this Laser Raman technique of
measuring temperature is?
This work is going on in our laboratory and elsewhere. In
our lab, Dr. Alan Eckbreth is heading up this effort. This
is a technique which is based on the Raman principle of
scattering of radiation off of a characteristic molecule.
*
In our case, the characteristic molecule is nitrogen, since
that is the most plentiful species. This involves irradiating
the gas with a well-collimated, well-defined laser beam of
a given energy and a given wave length, and observing the
scattered radiation that comes off of the molecule. There
are two types of scattering in the Raman technique. One is
Stokes and one is anti-Stokes. It really means you are
scattering off of a higher and a lower energy state, and
by ratioing those two signals and assuming the Boltzman
distribution within the nitrogen molecule, one can get a measure
of temperature. This is highly valuable in turbulent flows.
The technique we are using involves a pulsed di-laser,
microsecond pulse in, and by running hundreds of pulses in
and tracing out the probability distribution of the temperature
measurements you can get an awful lot of information about
the turbulent fluctuations in temperature as well, which is
very important.
I have a question on velocity measurements. Have you observed
isotropic behavior in the flow or not?
I do not think we have had a chance to do the level of
comparison with the laser velocimeter data to allow us to

1-121
-------
answer that question. The slides which I showed were
generated yesterday from data which has been taken over the
past several weeks. This is one question which is important,
and I think we will be able to address it based on the data
which we have taken.
Tom, you alluded to the difficulty of measuring total oxides
of nitrogen and mentioned that the resolution of the actual
ratio of NO to N0£ remains to be solved. You have referred
in your paper to your findings as NO rather than oxides of
nitrogen. Is that a prejudgment on your part that you are
seeing NO or is it for the sake of simplicity that you are
presenting your data in this form?

I would say it is for the sake of simplicity and accuracy.
We have a great deal of confidence in being able to measure
NO. We have a certain lack of confidence in our ability to
measure total oxides of nitrogen using chemaluminescent
detectors and the sampling trains which exist today. I
think there is a substantial uncertainty in the types of
measurement I am talking about. NO is one of the species
for which we know the analytical measurement is good. We
are less certain of the analytical measurement of NO£;
however, we don't know whether we've lost some of that NO
in coming down our sampling probe and into our device. It
could best be viewed as a lower limit; if you will, of the
amount of NO that exists in the system or lower limit to
the nitrogen oxides. This quantifying of nitrogen oxides
within reacting flow fields, I think, is a topic area that
really requires attention.
Tom, did you say why increasing the total pressure increases
NO?
I didn't say, but I think we know why. There are
actually two effects. If we keep the total mass input
into the combustor constant and increase the pressure,
the velocity levels within the combustor will decrease
I- 122
-------
because of the increasing density; so, increasing pressure
first of all increases combustor residence time, but it
does more than that. I didn't have a chance to show it
today, but it also changes the structure of the flow field
considerably and results in higher peak temperatures than
in low pressure cases. So it's really a combined effect;
higher temperatures coupling with Increased combustor
residence times, both.of which serve to provide high NO
emissions.
1-123
-------
-------
PATE OF COAL NITROGEN

DURING PYROLYSIS AND OXIDATION
By:

J. H. Pohl and A. F. Sarofim

Fuels Research Laboratory

Massachusetts Institute of Technology

Cambridge, Massachusetts 02139
1-125
-------
ABSTRACT

U.S. coals contain 0.5 to 2.0% nitrogen by weight, of
which about 10 to 50% is converted to nitric oxide in com-
bustors. Little information is available, however, on
whether the conversion occurs by oxidation of nitrogen com-
pounds devolatilized from the coal or by oxidation of nitro-
gen in the char. In order to provide a better understanding
of the factors influencing fuel nitrogen conversion, and to
obtain guidance for control strategies, a laboratory furnace
has been designed for studies of devolatilization and oxida-
tion of coal under simulated combustor conditions. Pre-
liminary results show a decrease in nitrogen retention in
the char with increasing exposure temperature and time, the
conversion of part of the nitrogen devolatilized in inert
atmospheres to N2/ HCN, NH3, and the decrease in NOX emis-
sions with increases in the fuel equivalence ratio.

Introduction

A major obstacle preventing the widespread utilization
of the extensive U.S. coal reserves is that the emissions of
sulfur and nitrogen oxides from coal-burning facilities of
current design are often unacceptable. Examination of the
composition of U.S. coals shows that if their nitrogen and
sulfur contents were quantitatively converted to their cor-
responding oxides the resulting emissions would be in excess
of the EPA New Source Performance standards for NOX for most
U.S. coals and in excess of the SOX standards for approxi-
mately 70 percent of U.S. reserves (see Fig. 1, prepared
from a statistical sampling of U.S. coal reserves).

Fortunately, only a fraction of the fuel nitrogen is
converted to NOX as shown in Fig. 2. The limited data on
the conversion of fuel nitrogen in coal is supplemented in
the figure with some of the more extensive measurements on
oils and synthetic mixtures. The fuel nitrogen conversion
to NOX in general decreases with increases in both fuel/air
ratio and fuel nitrogen content. The data in Fig. 2 for
coal represent uncontrolled emissions and are seen often to
lie above the EPA standards, even without the incremental
NOX contribution from atmospheric nitrogen fixation.

The potential exists for reducing the fuel nitrogen con-
version by combustion process modification but the develop-
ment of an optimum strategy must depend on a better under-
standing of the mechanism of fuel nitrogen behavior in
flames. Some of the processes that occur during coal com-
1-126
-------
5
5
:i
15 1
"» * fa
u 2 °*
> On
c w>
GE .r en
U » 3 | \
^ £ o «* \
UJ u c ~ .» >
o: PC™"
~ t i —
-i -2*5,
•*— -• p* *^ ^ _« n
Ld E^^"**-*
O 2 • - -S jc L 5 c
O U-^wcJ:^*
o- fl3 > o 5 • o
f 0^<^Io.Q. ^
— o z>
Z 0

0) ^
bO

O
01

3
t^L

J™\

o
on

-------
D
CO
Z
O
u
y

a
a.

_i
UJ
b.
O

Z
z
O
U
OJ
"ON Ol NOIS83ANO3 N 130J JO e/o
1-128
-------
bustion are depicted schematically in Fig. 3. Coal may un-
dergo rapid devolatization by heating in the products of
combustion recirculated to the burner, the volatiles may
subsequently burn either in the boundary layers of individ-
ual particles or in the flames engulfing clouds of parti-
cles, and the char v/ill ultimately burn out in the tail of
the flame. Alternatively, ignition may precede complete
devolatization and char and volatiles may burn contemporane-
ously. The extent of volatilization prior to ignition and
the distribution of nitrogen between char and volatiles will
undoubtedly influence NOX emissions. For example, if the
nitrogen can be retained in the char and allowed to burn in
the diffusion limited regime, the fuel nitrogen will be
subjected to locally reducing conditions which should sup-
press NO formation.

The present study has as its long-term objective the
determination of the distribution of fuel nitrogen between
char and volatiles and the fate of the char and volatiles
under simulated combustion conditions. Preliminary findings
are reported below.

Apparatus

The apparatus consists of an essentially isothermal
furnace with a controlled atmosphere into which coal can be
injected through a narrow water-cooled feeder. The pyroly-
sis or oxidation products can be withdrawn through a water-
cooled probe, and the samples quenched by water injection at
the inlet of the probe. The solids are collected on a
bronze filter and the gases withdrawn for analysis. For
studies at short residence times a high gas flow rate pre-
heated to furnace temperatures by a plasma-gun is utilized.
Data at longer residence times can be obtained by allowing
the particles to free-fall through the furnace or by heating
the coal in crucibles raised into the furnace. A schematic
of the furnace set up for use in the short residence time
studies is shown in Fig. 4.

Coals Studied

A Pittsburgh seam bituminous coal and a Montana lignite
were chosen as representative of swelling and non-swelling
coals. The proximate analyses (moisture, ASTM volatiles,
and ASTM ash), ultimate analyses (carbon, hydrogen, nitrogen,
sulfur), and sulfur forms (pyritic, organic and sulfate) are
reported for each coal in Table I. The coals were sieved to
yield 38-45y and 75-90y fractions. Representative scanning
1-129
-------
COAL DEVOLATILIZATION AND COMBUSTION

HEATING DEVOLATILIZATION VOLATILE BURNING METROGENOUS COMBUSTION

SECONDARY AIR
PRIMARY AIR

COAL ""
COAL
PARTICLE
COj.HjO
DIFFUSION
S FLAME
SECONDARY AIR
TIME SCALE 2 MSEC
EXPERIMENTS INERT
DEVOLATILIZATION
1OO MSEC

COAL OXIDATION
30O MSEC
CHAR OXIDATION
Fig. 3 Schematic of processes occurring in
pulverized coal flame
1-130
.
-------
Coal and Carrier Gas
Radiation
Shield
Graphite
Honeycomb
Flow Straightener
Plasma Gun
c
Argon
Quartz
Observation
Ports
Water Out -—==
Graphite
Heating
Element
Graphite
Muffle
Tube
Bronze
Filter

Water
Cooled
Collector

Exhaust
Cooling
Coil
= Water
Cooling Water __

Quenching Gas —
Vacuum -—
Fig.
Schematic of short-residence-time furnace
1-131
-------
O
H
EH

N
H
EC
pa
E-t
o
<
K
<
X
o
O
o
3
EH
a;
£
K
&
PL
*§
CO
PL
&H
1-1
CO P
M O
CO 5M
>H PQ
. 1
<
< CO
w
<
s
M S5
EH
>J
K
O
CO
CO
SH ffi
•J CO
< <
**
|B 5
S •
M a: •
X< 0
O OJ
K W
P-.
[X]
S,Hg
X.CO fe
& M *
EH 0 S
MS; 2
w
&
H
£
^
<
0
O
C\J
t-
*
o
0
t—
•
H
rH
\D
•
CM
ro
t~-
»
00
en
o
*
in
.=r
o
•
fH
O
O
*
tf\
O
CO
•
c^
VD
t~-
•
O
H
t>-
•
O
•sf
O
t—
•
rH
O
O
a\
M
m
rH
M
S5
<
J
g
H
PITTSBURG
SEAM #8
hvA-b
VD
O
*
O
on
vo
*
o
VJD
H
*
O
O
in
•
\D
rH
in
CO
•
o
iH
VD
•
O
rn
CT>
•
-^•
rH
w
•
in
in
on
•
CO
OJ
•
VQ
on
o
\0
•
m
rH
O
O
CO
«\
CO
w
o
^
<
CO
MONTANA
LIGNITE-A
O
SI
O
h4

B
1-132
-------
electronmicrographs of the coals are presented in Fig. 5.

Results

Physical Behavior

Figure 6 shows a bituminous coal particle after devola-
tization (approx. 1 second residence time) in an inert
(argon) atmosphere at 1250°K. The particle has swelled from
an original size of about 40y to 400pr an approximately ten-
fold expansion in linear dimensions. Examination of the
electromicrographs in Fig. 6 show holes in the surface
through which the volatiles escaped; high magnification of
the surface suggests that the coal has passed through a
plastic stage as a consequence of which some of the open
pore structure is lost (this is confirmed by a decrease in
adsorption surface areas for coals that have been heated
above about 1250'K). Focussing on the inside of one of the
holes shows the shell-like structure of the coal particle,
further evidence of this is provided by the photomicrograph
of a polished section (see Fig. 8). Electronmicrographs in
Fig. 7 of lignite particles devolatilized under inert con-
ditions at 1750°K show evidence of little swelling.

The marked contrast in behavior of the bituminous and
lignite coals is shown by the photomicrographs in Fig. 8 of
particles devolatilized in an inert atmosphere at 1750°K.
The particles were originally about 40 y in size. The top
photomicrograph in Fig. 8 shows that for a devolatilized
bituminous coal the organic matter is distributed in a thin
circumferential shell whereas the lower photomicrograph
shows that for devolatilized lignite particles the organic
matter is distributed throughout the particles. Under
oxidation conditions the organic matter is burned away leav-
ing fused spherical ash particles for both coals (Fig. 9).

Kinetics of Pyrolysis

The selective loss of different species during the
rapid pyrolysis step is shown in Fig. 10. The data on the
figure were obtained on the char samples collected after the
coals had been devolatized for periods of 10 to 100 milli-
seconds in the flow furnace, for approximately a second in
the free fall furnace, and for periods up to ten minutes in
crucibles. The furnace temperatures were set at 1000, 1500,
and 1750°K. The results show the expected progressive loss
in the different elemental components of the coal. The
carbon results at long times show anomolously low losses.
1-133
-------
E
^
in
o
if)
{/>
3
^
3
c_>
w
1
C
o
•H
-P
0)
O
•H

0)
0)
t3 .H
EH CO
M
pq g
5«.
•o
cu o
IS1 O\
Q)
O,
p
0)
-p
K)
cti
o
o

GO
CO
•a
£1
m
PH

«H
O

n
J3

fe
O
(H
O
g
O
V
rH
W
bO
»H
PH
1-134
-------
2:
o
o
M
fc,

O
1
o
M
EH
o>
-!->
rt
0)
4J
•H

6
§

•P

§
w
X!
If
En
CC
o

W
E-l
O
Q e
w =.
N
M O
K ON
cd
^
W)
o
w
60
O

£
C
O

-P
o
0)

w
in
1-135
-------
DEVOLAT1LIZED BITUMINOUS COAL 1250 K
100X
400 urn
3COOX
300X
20um
200
Pig. 6 Electronmicrographs of a devolatilized
bituminous coal particle
1-136
-------
DEVOLATILIZED LIGNITE COAL 1750K
100X
500 urn
300X
2 00
500OX
'Oum
Pig. 7 Electronmlcrographs of devolatilized
lignite particles
1-137
-------
POLISHED DEVOLATILIZED COAL 1750 K
BITUMINOUS
LIGNITE
450X
50 um
Pig. 8 Photomicrographs of a devolatilized coal:
a single bituminous coal particle (top),
a collection of lignite particles (bottom)

1-138
.
-------
ASH FROM COAL COMBUSTION
2000X
20 urn
BITUMINOUS
20OOX
20um
LIGNITE
Pig. 9 Electronmicrograph of ash obtained on the
complete combustion of bituminous coal (top),
lignite (bottom)

1-139
-------
u
i
o
8
101-
ELEMENTS RETAINED IN PYROLYZED CHAR

Montana Lignite
Pyrolysis T«mp.
O 1OOO K
A 150O K
D 1750K

z
a
o
-
*"
0
O
i/>
_o
o
D
1UVJ
O
1OO
50
0
4 /NA
50
0
50
O
10O
50
0
1OO
50
0
1OO
SO
0
I fc^I^-^-L _ Jj ° o

— ~"~""^*^) D ~~
o _
U.^ _

^-Wi A °
-o _
— A Ax^
~~ D 0 A^*ft— . J

- ° A4j A J D -
D

— t-Flo« Furnace — — j " Frelfari*4^ -CruciOleJJ —
101
10
10
10"
RESIDENCE TIME , MILLI SECONDS
Fig. 10 Percent retention of elemental components
of lignite in char produced by pyrolysis
at temperatures of 1000 to 1750°K
1-140
-------
This may be explained in part by the low heating rates ex-
perienced in the crucible experiments and by the deposition
of pyrolysis products on the crucible walls and the char
over which the volatiles flowed. The major element of
interest to this study is nitrogen, which is found to be
retained by the char in amounts that depend on both tempera-
ture and time (see Fig. 11, for details). At 1000°K less
than 8 percent of the nitrogen in lignite (20% in bituminous
coal) is devolatilized for exposure times less than a second.
At 1500PK the amounts of nitrogen devolatized increase to
about 20% for lignite and 45% for bituminous coal, and at
1700°K, measurements were obtained only for the lignite coal;
these showed a 70% nitrogen volatile yield for an exposure
time of a second. Prolonged heating in a crucible at 1750°K
yielded 90% devolatilization of the nitrogen content of the
coals. From these results it is apparent that the distribu-
tion of nitrogen between char and volatiles is determined
by kinetics and that the rate of devolatilization of nitro-
gen compounds is a function of coal type. The studies are
being extended to obtain the kinetic parameters needed for
the modelling of fuel nitrogen behavior.

Only exploratory data are available at this stage of
the present study on the composition of the nitrogen-
containing volatile species. To place the problem in per-
spective, data reported in the literature on the nitrogen
distribution between products obtained during coking are
summarized in Fig. 12. For coking conditions the nitrogen
retained in the coke falls from about 80% at 600°C to about
40% at 1100°C. The nitrogen evolved appears predominantly
as ammonia at the lower temperatures but molecular nitrogen
becomes the dominant gaseous product at hither temperatures. Small
amounts of nitrogen are evolved as HCN, heterocyclic com-
pounds, and in tars. The nitrogen retained in the chars
from the present crucible experiments follow the extrapola-
tion to higher temperatures of the coking data. Analysis
of gas phase constituents in the present study show that
relatively little of the coal nitrogen was converted to
ammonia (under 5 percent), up to 15 percent was evolved as
molecular nitrogen, and up to 10 percent as hydrogen cyanide.
More quantitative information on the distribution of nitrogen
in the gaseous pyrolysis products will be obtained in the
continuation of the study.

The behavior of the fuel nitrogen when the coal was in-
jected into an air stream (or a helium oxygen mixture simu-
lating air) is shown in Fig. 13 as a function of the over-all
fuel/air equivalence ratio. Under very lean conditions about
1-141
-------
•
o
o
O*:
*~O
X O
o m
o r-
32
DC
D
0)
o
D
£_
(J
Q
O
»^-

0)
O
o
o
o
ao
o
(0
o
Nl Q3NIV13d N %
bO
1-142
-------
DISTRIBUTION OF N IN COAL PYROLYSIS PRODUCTS
•/. N as •/. N as •/. N in •/, N as N2 •/. N as NH3 «/„ N Retained
Hetrocyclics HCN.(CNJZ Tor m coke
1OO
80
60
40
20
0
2O
10
0
50
4O
30
20
TO
0
1O
5
0
2.0
1.0
0
1.0
0.5
0
4
_ ilNk'/'N RetQined 'nchar DATA: "
,3**gjs$gj|^i3%%»>. * i-ielder &Davisd934)
^^^llllili^ • Kirner(1945)
~~ 1^^^^ ° ,« -Lowry et ol (1942)
_ ^^ ° Empirical Fit to Fielder &
g Davis M934),Avg Properties
•U N Evolved aNH, ,, „ ... . , _
Q ' ° Open Symbols, Coking
— 0/8° 4 o ° — ° * Half Shaded Symbols
"~ Jg 8 • Crucible t-4 MIN
o-^o
O 0 o O
o VoN Retained in Tar
'%MM^$$$$P °
•/. N Evolved as HCN end (CN)2
*/o N Retained in Hetrocyclic Form A Total
O Gas
o Condensed
1 1^1 1 1 1 ! 1
00 800 1200 1600 2000
T °C
Fig. 12 Pate of fuel nitrogen during coal
pyrolysis. Data below 1200°C correspond
to coking conditions
i- 143
-------
O
rH

LJ
o
o
o:
O
u
b.
O

Z
O
Cn
a:
UJ
O
U
ID'Q. moo
t— P^ FT\
Ift
Oc
(D (U C F
-»-' •»-•
<\) c'c
o
00000 OOOOOOO O O O
S; i^ssk: b^ lOin^riCNj'- O J> oo
O c*co<- £> ^jw^ K|| M e._ , -S1Q|XO N390yilN 01
inoNyne %
O
a:
ui
U
O
ui
UJ

1-144
-------
60 percent of the fuel nitrogen was converted to nitric oxide.
As the fuel/air equivalence ratio increased the conversion ef-
ficiency decreased to values under 10 ten percent for equiva-
lence ratios greater than about 1.5. At fuel/air equivalence
ratios of 0.7 to 0.8, the range of interest for many utility
boilers, the fraction of the nitrogen converted to NOX was
found to be about 25 to 30 percent, a value in agreement with
estimates derived from field data. It should be noted that
as the fuel/air equivalence ratio increased beyond 0.8 the
fraction of the carbon unreacted increased reaching an asymp-
totic value of about 30 percent at a fuel/air equivalence
ratio of 1.5. This unreacted carbon was found to contain up
to 40 percent of the nitrogen content of the original coal.
The fate of the nitrogen retained in the char during oxida-
tion will determine the maximum potential for reducing fuel
NO formation by use of staged combustion.
X

The above results are preliminary and were obtained with
a coal flame stabilized on the coal feeder. Subsequent tests
will examine the effect on nitric oxide yields of operating
under conditions favoring combustion of coals in a dispersed
phase and will also attempt to close a nitrogen material
balance by measuring the molecular nitrogen formed during
combustion in helium/oxygen mixtures.

Concluding Comments

Preliminary experiments on the behavior of fuel nitrogen
under simulated combustion conditions indicate that (1) fuel
nitrogen devolatization is kinetically controlled; (2) sig-
nificant amounts of nitrogen (about 70% at 1750°K) may remain
in the char after devolatilization. Since the nitrogen re-
tained in the char will be oxidized heterogenously, optimum
strategies for control of fuel nitrogen in coals may differ
from those for oils; (3) nitric oxide yields decrease mono-
tically with increasing fuel/air equivalence ratios; (4)
at fuel/air equivalence ratios greater than 1.5 as much as
forty percent of the nitrogen content of the coal may be
retained in the char. From these findings, the current data
indicate oxidation of nitrogen in the char may contribute sig-
nificantly to NOX emissions at temperatures below 1750°K but
less at higher temperatures. Complete models of nitric oxide
formation from coal flames should include the kinetics of nitro-
gen devolatilization from char.

Ack n owle d ge me n t s

The results presented in this paper were obtained under
EPA Grant No. R-803242-01-0. The data were obtained on equip-
ment developed jointly with a parallel OCR study on coal
devolatization and ash behavior. The authors wish to
I- 145
-------
acknowledge that the results reported here are part of a
larger collaborative activity involving A. Padia, H.
Kobayashi, G. Mandel, and Professor J. B. Howard. The authors
also wish to acknowledge early recognition by Dr. J. Wendt of
the University of Arizona that oxidation of the nitrogen re-
tained in the char could contribute significantly to nitric
oxide emissions.
- 146
-------
11:00 a.m.
Fate of Coal Nitrogen During
Pyrolysis and Oxidation
John H. Pohl, Massachusetts
Institute of Technology
I noticed on one of your early charts you made mention
of crude oil and residual fuels' conversion rates of
bound nitrogen. Have you made any comparisons as
to how effective the conversion is of these liquid
fuels compared to solids?
That data was not mine. It was gathered from a number
of different sources, some of it was in fact from the
Environmental Protection Agency data - Martin's, for
instance - and, they appeared to lie on the same curve
is my best answer to that. There is another para-
meter in that curve in terms of fuel equivalence ratio
and if you looked at it, there is a .6 at the top
and a 1.4 at the bottom and so there was a variation,
but they seem to go about in the same way. If we
wanted to look at the slide again it shows about
40-50% conversion. It is a little bit higher as we
go down on the lower nitrogen concentrations.
Q: At this time, do you have any idea of how much of the
volatilized nitrogen will be converted to NO and how
A
much of the nitrogen remaining in the char will be
burned to NO ?
x
A: Are you asking in terms of how much nitric oxide is
included in the devolatilized char or in the oxidized?
Q: No. What percent of the volatalized nitrogen?

A: We don't have direct measurements on that but by
going to the very rich fuel equivalent ratios you
1-147
-------
can see that there is less than 5% from strict de-
volatilization. There is some oxygen particularly
in the lignite — it's up to about 16%.
Q: Have you passed HCN through your furnace to see if
the N that you observed could be the HCN de-
composition?
A: Not as yet.
Q: We .run into that in our system all the time
and we have to be sure that doesn't happen if we
run an experiment.
A: I was aware that you had that problem.
Q: You indicated that as you increased your heating rates --
A: The heating rates are fixed by the geometry of the
furnace; the maximum temperature we can vary.
Q: Okay, but you did seem to indicate that as you
changed your derivative of temperature with respect
to time for 2very little particle as you increased
that, that more nitrogen went into the char. How do
you interpret your results as a function of heating
rate of your particle?
A: We really haven't investigated the heating rate at
all; I can't answer that but if you want me to
speculate on what I would think, that is, if I
increased the heating rate, I would expect that I
would remove more nitrogen from the char.
Q: But as you lowered your residence time, you increased
the percent of nitrogen retained in the char, and as
you lower your residence time ...
I- 148
-------
That might also have an effect on the heating rate,
certainly. As we lower the residence time what we
do is lower the quenching apparatus up towards
the probe so that the heating times are smaller
but the heating rates should be approximately the
same.
You mentioned that the bituminous coal particles
swell while the lignite particles do not; can you
comment on whether you expect any gross changes
in the ability of a bituminous particle to burn,
compared to the lignite as a result of the ex-
pansion of the particle and the bursting of it, and
increase the surface area?

We haven't seen any bursting yetjthere has been,
apparently, some speculation that they do, but my
conclusion at this time would be they are molten
enough so that they don't burst. They just form
a plastic sphere which then when we cool down goes that
way. The answer is that, apparently, people have been
relatively successful in modeling oxidation of coal as
a diffusion model so that it doesn't appear to make
too much difference.
Q: Did I read one of your charts correctly - that under
s toichiometric conditions that 50% of the N involved
would transform into N?
A:
I think you are referring to the pyrolysis chart.
That was the Bureau of Mines data; we have not achieved
that high. We do put a lot of nitrogen into the N
but we have not gotten to 50% yet, and yes, you did
read it right.
1-149
-------
A:
Would this indicate a trend for design in which the
fuel bed should be pyrolyzed before the admission
of combustion air?
Well, I would wonder what you would do with the
nitrogen which is retained in the char which would
still be a relatively large fraction.
...to allow for the N to N conversion, just an
opinion. . .

I think that the N to N_ conversion is of course
favored by rich conditions and long residence time,
and I think there's some indication that it is also
favored by possibly an increase in temperature.
Q: There's some difficulty in measuring fuel nitrogen
when the nitrogen is partially oxidized in the fuel.
Could you just tell me how you measured fuel nitrogen
and how you are sure that you got it all?

A: We did it two ways and it didn't seem to make much
difference. First, I didn't mention the furnace wall
temperatures were about 1500 K. Optical pyrometer
measurements of the flame temperature showed them to
be about 100-150 higher. We ran with air as an
oxidizer; we ran the air through the furnace and
saw no nitric oxide; we then went back and repeated
the experiments using helium oxygen mixtures which
eliminates, of course, the problems of thermal fixation-
and we saw essentially the same results.
Q: I'd like to make one cautionary remark and solicit
1-150
-------
your response to it; I think you are going to agree
with me. I think you have to be very careful in
correlating nitric oxide yields from fuel nitrogen
in terms of overall equivalence ratios because the
actual conversion which takes place is strongly
dependent upon the local fuel air equivalence ratio,
and so we don't want to be misled into thinking that
you are going to get a fixed number for a fixed
overall fuel-air equivalence ratio.

Tom is certainly right and we have thought about this
problem a bit. He's approaching the problem from the
turbulence standpoint where he would have a mixture
of different fuel equivalence ratios as he looked
across his system. Our system is essentially a
laminar diffusion flame when we burn the coal and
we get a gross fuel equivalence ratio. What it amounts
to is how much of the nitric oxide passes through a
region where they can be reduced by other reactive
species such as ammonia, hydrogen cyanide and hydro-
carbon species.

I'd just like to comment on the question that was
asked about the conversion of N to N». It seems to
me that it makes sense that it would be favored by
high temperatures because the formation of N^ from
nitrogen in the coal is most likely to be single atoms
of nitrogen and you have to get them to go to K?.
That process is bound to be favored by high con-
centrations of nitrogen-containing species. The
high temperatures would, I think, cause the high
concentrations. My own guess, and there is very
little information on the reactions of these free
radicals and species containing a nitrogen atom,
1-151
-------
is that the effect of temperature on the kinetics
of these individual reactions would be minimal. I
should imagine that they are mostly low activation
energy processes; so, the effect of temperature
would be to increase the concentration of the
species.
1-152
-------
A DETAILED APPROACH TO THE

CHEMISTRY OF METHANE/AIR COMBUSTION:

CRITICAL SURVEY OF RATES AND APPLICATIONS
By:

Victor S. Engleman

Exxon Research and Engineering Company

Government Research Laboratories

Linden, New Jersey
1-153
-------
A DETAILED APPROACH TO THE
CHEMISTRY OF METHANE/AIR COMBUSTION
Modeling of full-scale combustion systems will require
considerations of chemical kinetics, fluid mechanics and heat transfer.
Progress is being made on all three fronts to develop fundamental
understanding which may ultimately lead to a practical model. In the
area of chemical kinetics, much progress has been made in recent years
in rate measurements, critical evaluations, and estimation techniques.
One of the potential difficulties in the establishment of chemical
mechanisms is in oversimplification or neglect of important reactions
in such a fashion that cancelling errors result in apparent agreement-
with "theory". In the past such simplifications were justified on the
grounds that rate data were not available. However, the situation is
improving sufficiently so that an investigation of a detailed approach
to the chemistry of methane/air combustion is fully warranted.

The objective of this paper, which discusses a detailed ap-
proach to the chemistry of methane/air combustion, is threefold:

(1) To describe the approach itself
(2) To disseminate the results of the kinetics evaluation
(3) To discuss applications and examples

The work described here was part of a larger effort, under EPA Contract
68-02-0224, to study the coupling between NOX formation and combustion
reactions under normal and combustion modification conditions. Experi-
mental studies with flat flames, diffusion flames, and stirred reactors
under both adiabatic and non-adiabatic conditions, as well as theoreti-
cal calculations using the approach described here were accomplished
in this program. Detailed results of calculations using this approach
and the comparison with experimental data which were obtained in the
overall study will be discussed elsewhere.
scribed:
There are basically five steps in the approach to be de-
(1) Selection of species
(2) Construction of reactions
(3) Evaluation of probable relative importance
(4) Survey of kinetics data
(5) Evaluation and selection of rates

When the above steps have been accomplished the results can be applied
to the problem with appropriate knowledge of both the utility of the
approach and the hazards in its application.
1-154
-------
Selection of Species

The selection of species determines the order of magnitude
of the problem. The fewer species, the simpler the problem; however,
it is clearly undesirable to eliminate important species. Therefore,
in the methane/air system, one should seek the carbon- hydrogen- nitrogen-
and oxygen-containing species of potential importance. In this case,
such an evaluation was made for species of potential importance for the
interaction between combustion and NOx formation processes for methene/air
at one atmosphere between 80 and 125 percent stoichiometric air and
between 1500 and 2500 K. Fewer species could be considered if NOX
formation were not of importance, but additional species might be
needed if one were to venture outside the specified limits (to more
fuel-rich conditions, for example). Giving consideration to thermo-
chemical parameters, the nature of the problem (hydrocarbon combustion
with NOX formation), and critical species proposed in the literature
to be of importance, a list of 25 primary species was chosen (Table 1).*
As will be seen in the next section, this list of species results in
a very large number of potential reactions. While these species were
considered to be of primary importance, 14 additional species were
considered to be of secondary importance; that is, unlikely to be of
importance within the range of conditions considered but possibly
important outside that range. These species are listed in Table 2.*
Charged species and excited species were not considered for the purpose
of this study.

Constructipn of Reactions and Filing Ojrder

A computer program was developed to construct all mathemati-
cally possible unimolecular and bimolecular reactions among a selected
list of species. The program is dimensioned to accept up to 99 species
with each species containing up to four different kinds of atoms (for
the purpose of this study they were carbon, hydrogen, nitrogen and
oxygen) although the dimensioning is easily expanded. The program pro-
duces a master list containing all of the reactions for the selected
species in a filing order consistent with the alphanumeric sequence
used in the JAMNAF Thermochemical Tables. That is, each side of the
reaction is listed in alphanumeric sequence and the side of the react-
ion with earliest alphanumeric species is listed first.
The alphanumeric name for a given species does not necessarily
represent structure. Consideration should be given to the
likelihood of alternative structures. In most cases one structure
will be the most likely.
I- 155
-------
Table 1
Primary Species for
Methane-Air Combustion
Species
CH
CHN
CHO
mZ
CH20
CH3
CH30
CH
*T
CH
CO
CO,
H
HN
HNO
HO
H02
H2
HO
im
N
NO
N02
N2
NO
£t
0
o.
Log KP
at 2000K
-10
-2
+1
-7
+1
+5
-6
-3
-6
+7
+10
-3
-9
-5
0
•~3
0
+4
-9
-2
-4
0
-6
-3
0
Considerations
hydrocarbon radical
possible role in prompt NO
stable radical
hydrocarbon radical
combustion intermediate
hydrocarbon radical
possible role in ignition
starting material
possible role in prompt NO
combustion product
combustion product
combustion intermediate
possible role in prompt NO
possible role in prompt NO
combustion intermediate
combustion intermediate
combustion product
combustion product
important role NO formation
of prime interest
oxidation of NO
starting material
possible role NO formation
combustion intermediate
starting material
JANNAF Reference
12/67
12/69
12/70
12/72
3/61
6/69
constructed from
CH2F 12/63
3/61
6/69
3/61
9/65
9/65
7/72
3/63
12/70
3/64
3/61
3/61
3/61
6/63
9/64
3/61
12/64
6/62
3/61
I- 156
-------
Table 2

Secondary Species for
Methane-Air Combustion
Species

CNO

Co
C2H

C2H3
C2H4
C2H6

H2N

H2°2

H3N

NO.
Log KP
at 2000K

-10
-12

-6

-3

-5
Considerations
-6

-5

-10

-7
possible role in soot formation

possible role NO/HC interaction

possible role in soot formation

C2 intermediates

possibly important for

fuel rich methane or

higher hydrocarbons

possible intermediate

possible role in ignition

possible role fuel rich

higher oxidation of NO

possible role in ignition
I- 157
-------
As an example, the reaction N + OH = NO + H is frequently of
interest for NOx formation. To find its location in the master list
three steps would be required:

(1) Make sure each of the species is written according to
the alphanumeric sequence

N + HO = NO + H

(2) Sort each side into alphanumeric order

HO + N = H + NO

(3) Switch left and right side of the reaction if necessary
to put the earliest sorted species on the left

H + NO » HO + N

In the master list of reactions every reaction is listed,
but each reaction is listed only once. That is, the reverse of a
given reaction does not appear elsewhere in the master list. The
program also produces an auxiliary list in which each species in
each reaction is listed first in turn<, Thus each reaction may appear
up to four times in the auxiliary listo This auxiliary list is
useful for cross-comparisons of relative reaction rates and for
evaluation of probable relative importance of reactions.

The 25 primary species selected for consideration resulted
in the construction of 322 unimolecular and bimolecular reactions
(double that number if one considers both the forward and reverse
reactions). These reactions are listed in the Appendix along with
additional information which will be discussed below. To give an
appreciation of the rapid growth in the number of reactions with
additional species, when the 14 secondary species are added to the
list of 25 primary species, the number of reactions increases to 1078.
1-158
-------
Evaluation of Probable Relative Importance

As discouraging and formidable as the task may seem with
322 reactions (worse yet with 1078) an evaluation can be made of the
probable relative importance of the reactions based on thermochemical
and stereochemical considerations to drastically reduce the number
of reactions. It should be emphasized that the reaction lists con-
structed by the computer are mathematical constructions and many of
them are not likely to be elementary reactions. In addition many of
them are likely to be slow compared to competing reactions among the
same species.

Such an evaluation was made for the methane/air
system and the results are shown in the Appendix. The evaluation
divides the reactions into three groups (1) those that are probably
important (2) those that are possibly important and (3) those that art
probably unimportant. Those in the first group are reactions that are
likely to be important for combustion or NOX formation across the
range of conditions considered. Those in the second group consist
of those reactions that may become important under specific con-
ditions, or those reactions that may be important for species of
marginal importance, or those reactions that have been proposed in
the literature as important but appear to be marginal. Those in
the third group are reactions that can be screened out as not being
elementary or as being too slow to be of importance under the conditions
of interest. As a result of the evaluation, 28 reactions were judged
to be of probable importance, 73 reactions of possible importance,
and 221 reactions probably unimportant. This evaluation aids in
determining potentially important reactions that have not been reported
in the literature.
Survey of Kinetic Data

The literature was surveyed for data on reactions of the 25
primary species*. The data in the literature and prior evaluations of
those data were considered in establishing recommendations for rates
to be used for the combustion calculations. In a number of cases,
data for reactions of potential importance was unavailable or was
clearly unreliable. In those case estimates were either made by the
author or estimates were requested through EPA to be made by Stanford
Research Institute under EPA Grant R 800798. The results of the
recommendations are presented in the Appendix. Further details on
the survey and evaluation will be found in a forthcoming EPA report
by the author.
* only a cursory survey was made for reactions of CH^O
1-159
-------
Applications and Examples

The approach described in this paper can be applied for
several purposes related to reaction mechanisms. There is, of course,
the ultimate purpose of screening reactions for importance under
specific conditions,but this goal has potential pitfalls which will
be discussed shortly. The approach is useful for the evaluation
of competitive reactions. By listing competitive reactions on a
common basis, the thermochemistry and stereochemistry can be compared
with experimental or theoretical rate constants in order to judge
whether their relative values are consistent. By knowing that all
competitive reactions among considered species have been included,
the chances of oversight are minimized. The approach can be used to
identify important reactions that require further study. In many
cases, a reaction that is found to be of key importance will have a
large uncertainty or may not have been studied at all. In that case,
further study of such reactions is warranted. As a corrollary to
establishing reaction mechanisms, the approach can be useful to
establish "global" behavior. If it can be established that a sequence
of elementary reactions leads from the same reactants to the same
products over the entire range of conditions of interest, then that
sequence would behave as if it were a single reaction over that range.
Establishing such global behavior is desirable to simplify the com-
plexity of reaction models. However, the more detailed the predictions
one wishes to make, the less likely one is to find such global behavior.

There are a number of dangers in applying this detailed approach
if one is lulled into a false sense of security. Although all of
the reactions for the species selected are constructed by the computer
program, the decision to include enough species must be made first.
While including too many species makes the job more laborious and time
consuming, the unimportant species will be weeded out eventually. In-
cluding too few species, on the other hand, will result in neglecting
important reactions. Even if all the important reactions are constructed,
care must be taken not to eliminate too many too soon. A good example
of such a temptation might be one of the Zeldovich reactions:

N2 + 0 —? N + NO

With an activation energy of 75 kcal, one might be tempted to eliminate
this reaction because it is quite endothermic. But it is still one
of only two reactions between N2 and 0, which are important species in hydro-
carbon/air combustion. The other reaction is

N2 + 0 + M - N20 + M

and N20 can decompose rapidly at high temperatures.
1-160
-------
Two other pitfalls are very closely related. In view of
the complicated nature of the calculations using a large number of
rate constants there is a temptation to overlook the uncertainties
in the individual rates. Placing too heavy a reliance on uncertain
rates can lead to a serious hazard in the construction of minimum
reaction sets, a hazard which I shall call the quasi-detailed mechanism.
Chemical kinetics has made substantial progress in recent years
and with the development of improved diagnostic techniques, im-
provements should continue. But in the kinetics of methane/air
combustion, a number of key reactions still have large uncertainties
(some of these will be discussed below). While the detailed approach
can point out critical reactions, and an assessment of the sensitivity
of the calculations to the uncertainty of the rates can be made,
minimum sets can be considered only quasi-detailed until sufficient
accuracy in the key rates is obtained. Another hazard to be avoided
in the construction of minimum sets is extrapolation beyond the
range studied.

Two examples of the application of this detailed approach
to the chemical kinetics of combustion problems of current interest
are (1) NO/NC>2 conversion and (2) reactions between N2 and carbon-
containing species. In the case of NO/NC^ conversion there are twelve
reactions from the 25 species list:
43.
161.
203.
227.
268.
248.
295.
300.
307.
317.
318.
319.
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
+ CHO
+ CH20
+ CH30
+ C02
+ HNO
+ HO
+ H02
+ H20
+ NO
+ N20
+ 0
+ 02
M
CH
CH2
CH3
CO
HN
H
HO
H2
N
N02
N02
N02
4- N02
+ N02
+ N02
+ N02
+ N02
+ N02
+ N02
+ N02
+ N02
+ N2

+ 0
+ M
For a we11-stirred reactor burning methane/air between 80 and 125
percent stoichiometric air, calculations indicate that the main reactions
producing N02 are
and
NO + 0 + M-
NO + HOi
>N02 + HO
and the main reaction destroying N02 is

NOO + H > NO + HO
1-161
-------
For this case the rate constants are such that no substantial amount
of N02 (< 1 ppm) is calculated even though the concentrations of NO
(•x, 100 ppm) and 0 atom (^ 2,000 ppm) are substantial. Other species
with NO bonds (such as HNO, ^0 and possibly CNO) could conceivably
lead to formation of "prompt" N02 without going through the NO
molecule itself but these do not appear to be of importance for
stirred reactor calculations.
Twelve reactions are also found between
containing species:
and carbon-
47.
48.
50.
82.
132.
65.
67.
162.
204.
216.
218.
228.
N2
N2
N2
N2
N2
N2
N2
N2
N2
N2
N2
N2
+ CH
+ CH
+ CHO
+ CHO
+ CHO
+ CH2
-1- CH20
+ CH20
+ CH30
+ CO
+ C02
+ C02
CHN
CN
CH
CHN
CN
CHN
CHN
CH2
CH3
CN
CN
CO
+ N
+ HN
+ N20
+ NO
+ HNO
+ HN
+ HNO
+ N20
+ N20
+ NO
+ N02
+ N20
For the same stirred reactor case the reaction that appears to play
a. major role is

CH -h N2 = CHN + N

Further details on the complete calculations will be provided else-
where.

All of the reactions mentioned as important in the above
examples have a fairly high degree of uncertainty in the 1500-2500K
temperature range and bear further study. It is recommended that
these reactions receive attention in studies of elementary rate
constants.

Conclusions and Recommendations

The detailed approach to the chemistry of methane-air com-
bustion is useful for

• Guidance in establishing reaction mechanisms
• Evaluation of competitive reactions
• Establishing global behavior
» Identifying critical reactions
1-162
-------
Caution should be exercised when applying the approach to
avoid the pitfalls of

• Insufficient species
• Overelimination of reactions
• Overreliance on uncertain rate constants
* Construction of minimum sets which are actually quasi-
detailed mechanisms

The following reactions or their reverse bear further study
in the temperature range 1500-2500K to help answer questions of
current interest
NO + 0 -f M

H02 + NO -
H + NO.
CH + N,
>N02 + M

HO + N02

HO + NO

CHN + N
1-163
-------
SUMKAJW OF INFOmATlO
OK REACTIOHS IS MSETICS
RECQMHEKDED RATES FOR RBACTKW5 FOUND IN THE
Tl'RE AyO APOITtONAL, ftEACTlOSS Of POTEMTTAl IW
This Appendix contains * Buiwiaty of th* probable relative
Importance atut recoMaended rates for reaction* in a* th sne/ai r
coftbuation. The fonsac of the table* in (hie Appendix and the rang*
of Applicability intended far the** data 1« explained In Che not**
below and Co Che right. The not** below refer to the left hand Hide
of the table* and the notes on the right mfir to the right hand aide
of the table*. Itor* detailed inforwation will be found In • forthcoming
EfA report entitled "Survey and Evaluation of Kinetic B*ta PI» Reaction*
In Methane-Air Coe*uatio«" by Victor S. Eaglemin.
> aa not to inpose an add!1
made of reliable experimental data In the appropriate temperature
range. Where data In the appropriate range were roc available,

range of Interest v£th a rea&onable tenperature dependence. Where
such data weie j vail able, literature estlmaten chat seemed ri-jsonahl
vece uned. And finally, where no literature estinuitc*, wer« .ivailabl
Into line, uiill
there may *tllL
slailai

where older estimates have activation energies significantly below the

made.

The heading* in Lh($ section are 48 fcllovs:

Heading* Description _____
EVALUATION OF P ROB ABU
RELATIVE REACTIOH 1KPORTANCE
Rever** rate mav be obtained bv using the expression
for Kc tn Section f> at the full survey
Thi» section contain* a reaetioti-by-reactlun evaluation of
the potential Importance of each of the 322 reaction* conaldered In
Chi* itud?. It is extrcnely Important ta undervtand the crowndrule*
for the evaluation and acope af It* applicability. The ton*Mnt» found
90-125= atoichioHtrlc air. fro* 1SOO-2500K. This h
inf
val
rel
spe
re*
roailon in thia section outside the 1
d, but should be approached with ceut
tlvc importance of the given reaction
tflc conditions but other species and
tionS wliht hav* to be considered out
as been stated
lAiits mentioned. My be
lati. Not only would tti
» require evaluation to
, therefore, additional
side those Halt*. And
e
r
of
ir*e, other reaction tystena such a* higher hydrocarbon* or fueL-
cAfitainirti >r*tnss would bear their ovn evalua

The heading* In chit section are MS follows:
Wttttan In both direction*
Reaction nuaber Indicated
F • Forward
R • Xeverse

&Hr for both direction* given at 2000K
(«ld-ranCe of 1SOO-J5WK)
for log KC in Section & of the. foil

A Probably lapartant
R Possibly Important
C Probably unlajportarvt
* F1*B indicating reaction vaa rated
either A or B

Note* explaining reasons (or ranking,
Each reaction if numbered and direction lndii.dC.ed
F - Forward of »a»t*r reaction
R - Reverse of flia*t«r T*4ctlon
In Initial screening for combusti
at one Jt BO sphere between l^QO-^^
fiColchlotaecrlc air.
of
4nd 90-1^5"

preenporser-i ial Cera.
Uncertainty in parenthesis.

Exponent of T

Acttv«tlcn energy (kcal/moLe)

, , w- j . f • arr
• 1C)
References Llit*d In Section <> at tl,c fgl)
Explanation of notes:
Note B - !Io other estli
Note C - S'o literature
as indicated.
.
(J-P) - Johnston-Parr method.
(EST) - Order of aagnltude estimate.
(EVAL) - Evaluation.
(XPT) - Based on e«peilfr.ental di-Cermi n
1-164
-------
£g «
js :
„ ,.
»|s;
i :

i *
X
• 1
a

c
* ii
s :

Q <.
4
5

z
x

s .
a a
X
' i
i
, '
2

*••
~

&
"t

! f"
a
I O

' « o
1 « Z

..
i O
i *i O
.

S" o •"* ^
IS. 01 »
;i 3?
i ?i

ft U IV
»
fN

•a
M
•ft
»
(k
_
"
s.
O

s!
;-;
-
o
= i
o
»
r
Q
x

s
a

5
x
1
3
Z
Z
X
z

7 =
J?
ii
•- X
o
v y

m
Z

5
a
s

0
,
X
o
r
"
*

S
X U
«s
2 x1
z
0.

o c
32
Si
J
•3
Z •.'
•1
in. or
* 4
F- *«
% s

* I*
IV ISt
* *
« *
— &
tfi i
w ~>
«l
* * S * S
C
J
: 1 i
V) »- —
u, « u
-i o >-> o a
J 52 | 5 S
o o * a x - 1
£ x a ^— u! u
^i *z 5* u *
I- *• i- z
a ^ •* j * *- o x
zw « r « a " ^-1
* 3 -0 * Q 0
Si II !i ! i
^ ^

** T S3 * * ? ? " T
a a
•M« ?S rr oo —*
« !i ^ ii «
«" "N "o "» •"
r, 3 -n o *u -3 -1
tS *i 15 zS J5

*. tt u.a M.DC >bK u-r
!ri^ a« -*- ««0 » »
°. £g *: i
M > o Ml «T
tU «
* S
~. iS^m. °.
a wtfi ** •>
*• a us o
« >n « «
«t
«D
at
S O t
I I 1 2 I
*- JC t- « X
g 5 * * $
a S° £ o"

:i 5s § ii
•on o o^
4 9 • f

Ofi t 0 iA am
OO U C^ 0 - O
— I- O — - M «r
X
X Z
* r * T
? tJ
s s 1 i •

£ I
X u
« S - 1 2
u tj ui O
i. is : :
! s 1 s
1 ill
i 1 1 !
•n
0 Z
2 * ss ;; s;
* T "" 7 ** '
K
" * !: :!
*5 ^ " §5 55
§3 h i| ^
• 0 i xxxi
oo "" ™" --*

«
5

^
.
5
i

(J
S

ENtK'IIONS FORHAKO
HICHLV ENOOTHEKHIC
lEH «E>Clim
Bsv
-1 OC U.

r5
r x
ii
.1 •*
1 1
X Z
« «
tfd
«

*
a
|A
« <
s 2 ;

« « o

o a 4i
o
*;
0
-•

z z z
I ? Z

T X ±

1.1 lJ LJ

I I
I i
— Q a
1 ! i I
si2, £
U, » O >" ™ W
.32 SJ S
Ns ii i
O-— > K I
O — w Z «f Z
a Z VI Q<* O
ill if §

m n -o -a f
O 3 WO O <
x z x z z
x x
* * Q * *»
ix x »: X
ij ij i
if it i_
»S SS f.

. 2
o o
« ..

i i s
n

f>
i

.
z .

I
X

t
3
i
tit
E**t*»flOh(S FORWARD
ENDOTHERN[C
ENTRATtONS fDRWAND
HIGHLY I-NOOTHERNIC
TER REiCTIOM FOR MOM
|S §2v
is sjg

" '" 7-
i :i 11
C I? Z Z X
I i: fi
: x x! c z
j a tj oo
: 55 *S
«
M

w
«t
M
"
«

S
O

si
=:
•:
o
f. ^

o
z
z

FORWARD
EMDQTMiRNlC
a
•

v *
Is
ii
=i
i!
ss
I- 165
-------
3 =2
50 O 5 U U 3
K W

« *V •«< p* r* IM

S=!iii?iS
is i s
Is g =
S H
ss s-s
"= sis ^S
(

It ls= sis iss o ^
zi uf* =ii Iifl : "
.. s-s s.- s»- ss s
99 ""'o SeJ -~J »- e -*i <
^* «§: s" srl ii : s^i i
» • OX<^^IZtfX W -< »- X
1 i!!
5*"
?!;
W
E H
KEi
9° =S2 sis sss gs ;
sss
* IK
S ¥ £
.• ES- i<: ~.» «s - «.s «
x; <£s l.i ?«i si 5 l<: «
tn# ^ K * «.u« TUB htit * ««.ij« c u
"T S
S 2i
- V«J
• t
JO

s3
ss :i 1? !? ,3
*x o* *+ ** »
S* S* o« 3, «•
ox oa >j x
ss 5s SS
5 2

ii
it!
i
vV
f
"
g

3
i
•c

s
I
0

s
i

i
«

o
8

X
*
I
o
*

£
S

I**1
* •*

Z£
is
QIN
•w X
55
a. *
#
*
VI
Ml
g^
Ii
a
- O
S

X
0
^
s

,
m

tL
S
5
•r
0
X
s
ac
i
..

~~
"r

r
5*
u x
0 •
X?
« *
X X
as

x:
Q
S
0
*
,:

y
|

« «
ss

» X
f*
X
o u
a x
2.
ii
sis

i
8

D
*
(A
X
a
X
u.'

S
s
s
1
i
Ul
Si
o
^
*
§

oe
"?

"8
* X
|H *
o '
x°
*
X X
u o
u! *

1ET
O
1*1
i

o
X

*
o
T
,j!

§
«
*

--
•»*
m x
as
X x
is
O X
s.
* o
J5
iS

I
?

X
1

.
z
s
s
i
w
*
i-
i|
55
5*"
.-5
a ui
— »

-;
5:

X
§
c

i
2§
is
*" ".
M a>
2 "
o
2 1 I!
s s H5
o o
0 0
£ t
0 X

; i §
. • •
! ! !
£ S ?

a u x x
if f !!
"5 S- So
i£ of gf
=1 II II
^ is is
» > » > z >
O W Q tu O ui
.

»» ^~ .«
?" 7° ?*

o a * o
z z r x
* r oo oo
• • " " * '
" °° °*
ii is i§
is is is

o
w
f

o
X.
5
•
-
o

S
^
S
t
I

--
T"

??
X X
* •
ss
+ *
X *
ii
I- 166
-------
X *

X «

X O
i s
«O *•D «S O
• z «uz a 2 2
p in m *t
i !
S S 2
O U-

55
si

¥«

sl i
5* "*"
er & ij D LU

J— or h- *

-. U *
*- Q <• w

J > 1*J I- E
JJ |S
a i
O
*•> O
32

!i
¥-
i
a z:
?
I> IS I: IJ
-
S, z
T «
•- 5
**" D

»•-
-
1-167
-------
U) *

;-,s,
uitft >«
•J.
«0 i
*•?

-i
i :i ; 2 i KS
u 60 O w - - O
S ? ?
s s
a s s g
* O O D O D
DZ z z z r
555

Ct X It-
: i n
\ i i
. D ~ S

« , £ 5 S
;???S7?sj

#»«#4-«. + fc

yZ'Z'^JT^Tz1
il i » i
gs si
*• * i*3! —51 "" S *"
1* ":> £> S i
Q _j -i -. o «
9 >- M. >• r- a
r- j* ^« i u.
§| 25 2* S S
•3 •" « C Ma a Q C
5> a: * « * 2 r *
* 00 00 D 0 0
<_i a * ^ u- _« w -, ^ u. _.
:; ;:• :; °° 3: :
»~* ^^ ^T "» T 'T •
• • > I • 4
I, =, ' * . • -.r ?
°3 i1, ;f ,i J.:.
-I '^ •• •• "Ti ?
s;. ^,, *, r-. ,, ,
Si I?. ?.* !; «A ;
II 55 ;5 Ji ss i

-
•

s s
H i
= 1

is is1 i s
35 S* ° *
f S3 °
1 Is |
If! S
« *•>. - i
SS SS5 S s
v»
>U • *
'" °" w °° s •• •
r>*s O O ** *•* — _. ,
J; ss i; ;s s ~~ .
"1 »0 =0 »D 00
. X * ' * . r °°
^ **•* * *»
??^°_ » (
'• " ?r ?: s s! '
J¥=ItOO-B Sn^
. r« II I. S SS
;l i? is ?| ii
ii il 51 ii -^
i

IN
C* C ACT EON
MOM NCC
E« EACTIOt
NOO HE UK |C
NfB TEONS KEVER
LV TQ fit ELEMENT
*T a
KB X
oo) ma, 4
n i** tsi ft n
4 « *
?(%( !>• («• O
O OO 2
C E 7 X
O
•3 X
li. M BC
« 09 «
09 0 U
* *
Z rv j
T Z

f» * •
55 .
' »
i r
i^
u. or u
tn in ^
p- f- f
i § §
*•
*
z
1C
si
> z
M
a*- a
E z 1/1 zr
-• -< o a « «
on z 3
z •* r -v
' «. I E E
i +
* T
t z r I IT
X 00 O.I
o o o r
i ii i ^
r* ^S
i T r IT
! « P- •- a> w
1 1

1 D J
D n z
• **
r. *
• :
c z T
J 1 u
3 3
c x r
» m z
c T r
- K. 1
1-168
-------
I I
I

^
5 „.

t 1 13 I
I S? 55 SX

i 15 if ii
I- 169
-------
O D
j

*
b
«•*
3
*
§ ?
Z Q
« Z
*
**
*.
§
s
Cf
I
i 2
a
>
1
3
2
* rt m
w T
o z
~ 3

I;
s*
2
a
z

o
X
X
IJ
*
z
a

u
*
M
Z
«
»•
*
§

QO
0 *
xo
Z u
a

z z
££
o

a
a ji
i li
.
1 p
E X
]; ?j
JO E
X
EC Z
* O
Z I Z
I ££
Si 3
•* 3 9
a w
r£ O

i
Z »
Z Z
* *
0 0
s s*

1 i
* o
! !
• s1
« Z uJ ^
» >- * «
ac o m
°l SI
o " ^ o
II Is
*
£« - —

x * »
* ? ^?
wo * •
z o
Z M Z Z
X
z a IT
-" *s
5
o

z
a
Z
I
?
o
s

1
iv
o
X
i
u*
X
ti
«
Z
i
s

o o

e *
S!
, ,
y y
I I
U >J
S*
at

O
^
1
•
?

^
i

9 »
^ ^
-f
»
, .
y c
1 z
3s
1!
ac
_

i
z
a
i

O
=s
i
0
§0*
i!|
v!i
its

« «
ss
si
» •
t ?
• •
1?
?ni
z
SJ
D «
sr •
zs ?

e
a — -
o
z
PW
5 * X
i i ?
i ?
.
« u
I i|
z
S @e
£ SQ
V * >
SK _i
y. SS
SS £*
SS SI
. .
-— o 0
!? is
!i is
« 0 . .
S? !S
XX I O
u u u vi

: I
a w
-
Q
-f

a
IM
Z »
ftj
S 5
s 1
i i

: -

o
I i
£18 s
* * "* a
« O -J S
• o o
?r o *
v!s s
3 D > Z
» *
O O ^ r+
CO W OJ
~- SI
Z Z •
~o is
C X i-fO
• • " *
X Z* Z
?S JS
LJ 0 00

: : sss :
a o •» «"O o
2
"2

r z a
X* Z I
: x * ±
a
D o* O
: x z z

i ! §
! s 1
. * :
1 i g
• « K
*- O
Z O JT O S •- "~
? |!f
f ls°SssI :
i* *5*o«^So
3D D « f* O * » *^
OE O9X«3u>X
..
^ 22 ^ 5
Z I 3
5 i 5o I
or r t x
* • • « «
XX X I
C ^ o 3 3
X O Z Z Z Z-

1-170
-------
fs
5 :
1:
".ssr
tut - -
isSi.
• ~*~"«
i i
a o
B i
f s
5IS233-3
M (A * £! I? £f • £?
s f
s ; ;
2 S
S
1
i
I
3

£
Z
1 «4

* X

T
s :«
•j
I i
X X
s § g
o S s
XX£
i § i
li I
o •-
ii
rf « X
Ji * + * *
SI- 3

oSJ r* g
.

~^ m« ,A*
-- i i

«J 3 -33
*•=
IT ?r XQ
LJ
o«* 20 a "
I Z I IS
3
IX IX It
2
0
s
XfACriQN
IB.
*
1U
O

* *
"*7

00
* r
x a
O
*.
«
•
?
i
§
i
1
i
V
o

9 9

or
Is
, ,
Z O
*-.
a o
? ? s
s s »
"• *i M
§
•r w
i S
I If
=i : 55
5i = x'
§i : ii

O 2j o * «
# * *

; »• « ^ ^ «
• 1 1

** S* S'
:? 5* =5
. . • • • •
'i '? Sf
a •s a
X 0 T « TO
o
S
M
S
S
f
So
If
*.
• Qg
V Z
in • —

It!

>c-^
•

i3
s?
• •
ii
10
!
•

*
i
u

£
?
I

1
j
•

1
2
5

.
si
Z X

»^
•

I?
*• *
3S

°i
SS
J
?
K
I
i
i
5
S
e
if

,v
--

f •
55
• •
*|
*5
•
O
?
s
I
1
al
5r

SS
•

Ji
I

« .
00
53
O
f
Ilk

00
'

ia
2!
. «
=?
Z 3
S S
X Z
5 S
i
I §
i i
<
s!z
ssl
s.i

Is:

~~ ^«
• 7"

a o
1=1
At D o 1r
D Z f
0 .
3° 53
?s ;;
! S
z
i
I 5
I Is
s I !>
"'Ii *s
ss» z:
o a« oo
W Z MA
3""£ *-

BE > M Z >•
* *

•^ ^ < *
1
-*•»
"^ 5 XT
0*
" 5-
?S J?
5^ i3
55 55
D
IV
Z
1
O
s
IV
D
||
!i

4«
•

i r
~ =
""
O
SJ SS
* S * 5 S* i *
I- 171
-------
. *
s s s

i S i
O OO
3 S l~f

>— >n K- — *•» u
V)O •< O O O * « 1
u £ d Z Z Z vinti
i R s
X O
! s ? s 1

l 3? z x r
J S
•COMotXte-to-N OO
juxxxxSzzzzS

»V1iM«is»'"rM'>j«(fvi*«
xrxxrxxxxz

555J5S5SSS
5 g
M — O tu O
c ss i s
i il s L K
s p i
u « *
1 ?1
O w.
II
£ £
! SE „* ! 3 s si
z Xt^^OZ 2 V ««

* 5« PS S I 5 |S
£ Si
3 S SJ
r o z >
qjrtj 3 o
5; S
r ? a o
-1 'I
Z Z II.
rf\ r* 55 S ^ o-o •» ** "*«
«•«• -* -* i J J J- •* •* •*•<•

s i «i i *
I!
I- 172
-------
S i
S S S

•!•««
« « C

DOD
Z Z *
Q^^^uiOO « u *

3T • * of « ^ Q a O
* ** vl«If VI r
S i

s
0
nr

z-
X

;
°

«j
^

*
a
i
•V
r
*
i

^ o
Is
O X
11
Ii

o «
CM

•r *
IP ,n
i? a
3*
r x

^ 0
I X
-* «

ll
ax
P
51

!l
5 i

**»
O O
=

c-
'r
0
• -0
X
••
S
•
!
3
*
X
~6

X
f
a a
|!
§.
s«
OK
I u.

-_.
D 0
X 1
o o
r *
"y
* "~J

~ .
* o
55
« -0

I
o
t
O

I
|
i

=
5

->-
0-" q-'
1
I
= 7

"*
*
T n
•0 -C

•

X
X w
||

II
J.^j
-

^-
IB C

C
C'

* 3
x x
£5

X
5

o
•
I
f
1
i

j
^

« •»
aj «
'
=? 3
01

-11
°l
••
x a
i <
J> -O
!
*
±. »M
r
o a
•
s *
o
; J
f~ tO
•a -o
s s §
! 'rt 5
5 ;
X X
Al IK
c c

J _l _J
3 — 5

0- o fl ^ c,
J * (SO
1
••
K
•LJ
S
LJ
?s
a
LJ

J
3

^^
2 ^
P
Z a
^^

;|
. ^
^ *
^ >j
! ss
i f
2

r
VI

«
1
I
1
I
0 «
I ;
0

Ei
•£
a
X a
a *-
a
-. at
f > '
,
I

! £
a
X
il
Q > Z
5 * *
ot a «
* n •«
s ill

j
s

s
f
z
§
i

I |

z
X V-
O™*
S*£
|2V
ill
,
i

£
2

o
—1 IL.
K
O >
ts
gl
Ii
§
|

o«
52
X X
« u
^|
Is

£>
J«
!f

D
Z
5s*
Out
JS.
JoS
fs.
fix
«« Kltfi «4 4^ 44 nirw «« ww 99 r-P-

*h« •!*! nh |4>A UN^ r-P^ ~« OO 4« C9

«^ r»~ t tf<^ ^« Nfv * * ** nl(M 44
gl ?S

Ts "s Is
„ si sf
-1 s? °s -1 s;
f ?
;„ ss s^ ss g.
KZ II T T XT XZ
52 2.
T T IT

1-173
-------

*1J1
s s. s
00
II
o o o
g -?
s s
0*1
* 1 X X
S S 2 2 S
Z X Z X X
^ °
j; ? i
I
t
*
o
S 5
I ^
X *
i!
*"£

X j« f%
*c *
!?
So
* 5?
C *J
i -*
0 0
r s
4- «
22
O
*
at
S
I
£
o
1

•«•*
0?
53
I O
T
a
f J
ss
«.

X
a
SS
*:
So
§«
..s

• •
l\»pj
II
«« o
I IV
Q
rwiM
r a
r
¥ X
o o
ss

sllels 5S"
O O ij u _|i > uj
"'•'* "3 -j *

WKtK * O tt, 0 «. O
2 : : - - s s „ ,i s? ^5
!;l : z * §!-*!« -•
MUlw — °*-5Slra: KU* H*U
SjJ *" " ™ *" * Z — — " i*> Qj^MttiKO.,
*•* o a D S3 2 5 S 52 £«££•£ *2
O»x->~»- aitl OCKKiCO £*- ttOk.*.
ow o *» a ». i- K Q o— ••OQO--
si s 5 = H •; ; * t is ss *: Sa !:

gOO 3 O O <4« « O £i QiuD SoXXoioOO
-.0^-3-25 _,_ 2 S 3 2- 3 3« 55 SS SS ¥
* IU«* ••*» *«•
*

SO
M "«0 31^, 0 n| X I JM
* lOZCJO'nOZ -J »# 11^ EOC3
* * iw «i r •• xx riaa* xx**i
a • T * I * * * » * £ ± * X** T
"3-* *== I» ••*1**
i*»ro+fl»o ca««ivi""i ij» + o 4 •> n «> +
o3Cij*j A> i-* 1^1 r. o I T X j1 *i^%»i^ Xtl""-*
Q O" »• •• •• *" "" ^I
-------
. S
- S

s ?
2 :
i i
S S S
s

NO -> a
'
i f

o o
" o o

;:siigis4s
3 O * * ******

5555555355

isSiSSSSSS
! 5
O "i
So o

sj :
i S
*x ZX
Oa

i *
\ i
I a
K
at
*
1
i
&
• o
o
It

1
i
I

y •
•A Z *• *
UU 4 O K « O
ar Si ^ s
D«r z * c B.
Z X X E ••
OO O U O J w 5
SJS ^ £ || S
s!l I! s si I
S>5"? •- z> 5
UOEU.U.* z u« •>•
i 5
~~
X
o
I
•±
X
J
4 »»
f
3 T Z
r * *
i "»
;s *• =
:i ^:
Z O Z O
or a z
z c. * a
22 ^^
O N
o z
r f r r
Z O Z O
00
1
r a
o z
3 *•
y
f a
£ £

z o
•3 .e
a z
z o
ss

o z
I O
1-175
-------
o *•
. g
O — am

5
-

|« 2|

*
s*

=-
s«
«i

on DO
i ar * 7 7 aD o c
S I • I I -* 3
» * * I

* s IE ii ^ r "j
-c^

•-•
1-176
-------
rt» c 5 O * I * «i

S * S< Sfl ZQ *
I I =r ?
= 21 f:
S i* Is
i
1-177
-------
g *.
S8=M n>
« „,
?t ^j
XX II
i; jj
il is
is ii
8.
!!
o
-o
3

f
i
i
•

i
o
X
3
S
*
i
s
ft m
?s
*
• •
?s
il
A -d
£
!

f
i
i
;
\
i
z

ii
3lUtt»HlOQN
MM)! IMUlVkLM
S *10IOV3M Hi
5S*1
ass
* u«

4 *
li
?i
* •
ii
* *
IS
ii

i
IS*
1
S
§
E

*
5
3
IM
9
*
§
OO
»3
li

i*
ii
ii

ii1
i
IM
*
I
3

u
•
•r
s

§
5
Ul

M
B«

3~
i
|A

-
5.
0*
H
"!
0
0»
-2

?
•
f
s
*
i
s

s
oNCiMTMrtMs FHM
1C EHOOrHEMK
»£

ss
ii

*
*
£

~
S
a!
s!
°f
a
Out
-
K

0
-
*
O
X
u-

0
04 D£* FORM MO
SE H1CHLV (N&QTHER*
a m
*- 41

2 =

* m

•>
2
i
* w
*s
a^
II
• m
2
>«
= 5

Z
i

O
i
s

is
31
Ik. «
Hi IK
||
X X
• •
O *
* *
ii

P-- H.

* *

\
2
*

"
S
o-
^<
O
ol
*• o

o
X
^
f
I
•r
•4
rw

WHIM AH
EM>mj»10 HI01 «90
ii
Si
* •

^
a
o
::

Q
^
i!
•* ^
o " «

s :!i
0^
o
ol
— o

S a
i f
!•* Z
O
i I
i s

S |
! 1.
CENTER 4E*Cn
s's •«
2?S 12

SS SS

«•
^ S * o
3: Sz

3
^
^

-
«.
i!
0
a
o
-
X

s
X
X

z
s
*

0
• 0 HICHLV ENOOTHMI
MOE* KEYEME
tlUCTUIE
ffi

a o
•a o

* K

C •

«
S

o
*
o

o
^
i
IL
Hf
IK
3T
X
I
CENTER HCAcnoN* si
i

ss
oo

* *
ss
z f

ft
3
tw
?

3
*
ft
r
s
*
I
i
s
1
>
IH
•n
V*
CENTEH H£*CT|OM
HO • H20 • ID NOR)
ONCCHnUTLQMS *EVtl
!ii

•r •*

Si
a
ii

£
Wl
i
«
•
s
s
°:
o
*»!
-S

S
*
a
i

i
i

M* fflAtMRO
SE fHQOTHERMlC
I|
»

ii
« *
0°
X X
ri-V C "•* 3 "O *M * X XX I a
Z O OZO DVD Z X
• * * +»**+»»

*a *o * *d ** **• o* ao o* o o

if if is i? f? is is i? ?5 is

It * it tt ILK tkic u. K u.W «.« n- *r kk« «.«
...<• M*SI MM « « 1/1 *V «« »» "- * * »• » OO
1-178
-------
5

u
g

g
o

vt
f
o

Wl
?
*

§
i
0
S

O
o
X
•*
o
Z
I
;
Ul
|
x

£
|
1
i

ii
II

ii
11
5
S

2
S
i
I
3

II
gj
w ^
I- O
£ x
X•

X X
* 5

•« £ S*
* « IW
*':
M m

*
£
S
I
;

|
x"
O
5
i
2
o
1

ii
ss

il
si
a
2

i
s
i
u.
*

S
5
V*
ii
E* HE1CIIM,
HTMMOHS F0«
MQDIHEMNIC
SS"
?s:
|§>

«
ZS

11
s?
0
s

IM
X
'
f
i
1
I

1
X
0
K
Ml
M
i
i
VI
I
i

v
is

?!
"»O
s?
s

o »
o
-« o

o
X
s
1
I
V,

II
Ul
— *
Xs
SS
*

mm
oo

il
a o
*
«
s

X
f
"
I
J

*
I
tU Ml
HO « NO FAST
NTRATIONS «(V
oV
ii

sj
« o

ii
»
s
I

IN
•
s
s
a
;

1
X

£
'
I
o
i

?•
33

ii
si
5
S

i
-
i
;

1
5
a
S
iR REACTION.
*
\j
i

"7
S3

if
3S
go
3!
9> a
»

s
i
•
:
s
i
oi
S*
|S
•> X
£5
i:|
Ml
S3
iii

£d

0 X
* *
Jl
is
S3 o O Q *v O** a at o 3 i is ii j? ?? i? <:

53 25 *5 SI ii ii £:' ii il ii

B
|

U
•
2

!
s
s

;
4J
IU
S"

?
2
s
o

a 3
t >
M W
• a
o
*n o
*
O
S
•
s
s

1
u
HI
X
a
o
I
oo
if
00
|
s
*2 °
e£ 141
S3 s
:: s
i-
Q
i?
*
0 I
f I
m Q
S £
S i
IV *l

1
*
i
: FORMAftb F0R.
ENOOtHERNIC
ITEH RE ACT (ON
E«DOTH€RhIC
II il
^ ^
:* !f
0* ~*
?s ??
33 SS

3
f

^
3
•
s
S
u.

•f
2
fl
P
!§
2?
"
if
S'
^
i
-•c

2S
SS
"!
o
om
*
i
i
^
*
i
*

s
£
82
Di.
:«;
is2
X di
e 1*1
*•*
ii
p

-s
»- lit
|3
2?
s
«
;
•
s
IM
o

f
i

i
BOTH DtRECTI
NlfM
i!
•u a
*4«>
*
ii
«!
^S

s;
Is
0
e
e
*
*
S
*
o
s
£

0
si
is
I!
il
X t
i:
0*

S=
I:
--
o
MO
*
ft*
S
I
°
Ml

1
«
S
EWOTHflWIC
«ev««it FO*
fOKUARD 1
i • COHSIOEI
X5
a
*
o a
3 =

2
8

x-
*
a
X
•
s
Z
»
1
i

•mm
i?
s*

J
S

i
1
*
i
x
g
s
5
i
*
I
8
O
i
3
i
jj
!S
*
38
si s~ ii si ?I si si si $i

ii ii ii ii ii ii
•Mf* AIM fv«t M A(A|
1-179
-------
O «»
_*"»
S "
Is"1

V «C
IT S
1 I
m
i
w
•9
,J
X

1 5
< 0
* z
•>*
g

Z
i :i

zS
s 1?
5 s*
si

, .
s s
at «
i S

O
D

O
X
K
o 3

S S

i i
5
•f
i
i
o
s
55 •=?
„
f- ?3
x r i
" * ° •
a o
XX r r
i

*
1

•w
(V
*
K
||
o a
:E
si
7

,
B

|
i
PM
V
*
O
1
i
LOW CONCEMRA
V ENOOlH€RH]C
£5
tx
s
at u>
t- ai
^ *
ni ft
f» IV

f X
* *
Z
X
z ^
Z X
_

o
o
o
o
sr
03
s3
c
•_
• "

it.
i^
o

X
HCCHANl
DIRECTION^
OF mooviCH
X
S Bt
a
^ • CONSIDER
L • REVERSE
32

0 0
fs,
X Z
DO
z
z z

o
O O
£"*
S£

X
o

i
D
\
u.
a
ION CONUNfRA
Y CNOaiHERHK
si
o z
'. THIRD OR
; R€VfRSE
T-
• K
o rs
^ *
K
Z •
ii
— _

O O
:l ll
ll° s«=

23S sss
a .A om
0 0
• *

z o

z z
S 5
z z

"A- u.
r~ •}
D O

1 S
s i
S
i
"" *
O BC
SJ
O a
u. K
II i
:•; 22

?IV PH JM
O OO
* * * *
* *
a "•
z z * z
3? 3S
* O * "i
_

O
p
u
'1=
25°
ssl
o ««
O

a
i

O
i
s
w
L
ii
» X

*l • NQZ
I REVERIE
-* «•
oo
* *•

!3
Z X
i!
+ o
X Z

.
*
g^
a
Is
o«<
o

z
a

I!
a
i!
P
o *
Si
N *V

"4 O
Z
0*
»

"""^ '
"* «
c o •

o *
e»

i i
I £s.l
X UNO
a * x
n « s
•»z, 3
0 ?i
* -l
0 °~
X

5 °
r z
• c
a
2

D
§
? 11
s -j

gs
S|
iS
o o u
z
X. * *

X X
2 !'
s °!
is
-

1
i!
ii
y> 4
s-
4 0
:-
•

x
I
OUECriOHS
OF tELDOVECH 1
X
O at
• CONSIDER
• FORWARD 1
f a
ni i«j

si
3 ^
D
X 7

:
0
V
g

^
a
z
K
3

•
2
z
3?
a
i
'.
K
„
•
o
ii
—

' '

X " >. " Z § -
IH A ^ O O *W ^
.^stSjy
ft°2ii!5
sj
o
O "
•

o z

Z X
O 0
D r*
ii IX

SS2
§!»
"IS
•T « vi
"- O
«0
• t
Eo
0 Z
is
* «
*. N ft ».
* * n vi

MO O -3
0 Z Z
a
* z * *
?3 J°
Z Z Z Z

1
w-
JP M
no
* o
ill
o
o

a
a
°
u.
4

a
V
X
UJ
0
^
1
•a v

S3
O
19
33
85

JU
O
Z

z
s

z
o
o
s

<
z
z
a. a
ii
ss
v a-
•» -f

*» a
i?
O "•
is

•
o
o
• a
o -r
s~
Si
•) K
!-
O
X

a
X
K
i

z
a BC
H
ac
SS
i:
•r -f
x z
rvO
Z *
Z
z
o
si
—

i
O 4
a <*a

O 3«
:Si
10
™ a
0
*

s
a
i

is
ij
Jo
ii
a) A

k o
i>
3!
S?

;
*
0" is
g|
D
s:
S"
o
X

z
z
(V
O
ni
O
0
o
K
DlREtUQNS FE
o
• CON^rOER
•
j;
x x
go
^ *
**,
i*
r z
ss
L B U.K U-K It « UC
a-e I-*- « B) ff-» ec
1-180
-------
I « Z >•

w* £ r

si ss

II I!
u. u u. «

u • »
I Z-.
-
1-181
-------
11:40 a.m.
A Detailed Approach to the
Chemistry of Methane/Air Combustion:
Critical Survey of Rates and Applications
Dr. Victor S. Engleman, Exxon Research and Engineering
Q: With respect to that NO formation right in the flame
zone, we have data that shows that NO- is formed right
after the flame zone, and it is mostly by OH radicals
rather than by any other species. We have quantitative
profiles both for NO formation as well as NO,, formation
right through the flame zone.

A: Are those probe or optical?
Q: These are obtained by a supersonic mass spectrometry
probe. The analysis has data for all the hydro-
carbon radicals as well as NO species like HO, HO-,
hydrogen peroxide, formyl radical, etc.
A: I would really like to see some optical evidence, some
non-interfering evidence. Whenever you stick a probe
into a flame there's always a chance of interference.
I know how good molecular beam mass spec sampling is
as well. I have a lot of faith in it, but I would
really like to see some unequivocal optical evidence
for early N02 formation.
[Questioner:]
I would argue against the skepticism against such
probing. We have studied probing extensively in
various systems just to establish this as a technique.
Let's put it this way, there are a number of ways that
you can test where in fact you are perturbing your
flame; the obvious ones are to you look at your flame,
and see if you perturb it, if it's lifting or attaching
1-182
-------
Q:
A:
itself to the probe. We don't see that; but also,
then, it may perturb the temperature and thus the
reaction rates;we don't see that either. Of
course ultimately the probe perturbation, indeed,
I agree with you in that respect and for the inter-
mediate species it should be independently checked.
We have done that for our data, but if you compare
the estimates that have been made by other methods,
they come embarrassingly close, so there is cor-
respondence, and if perturbation is present it is
certainly not very major.
Vic, we've known which rate constants we would like
measurements of for some time. As modelers we haven't
really convinced anybody to go out and measure them.
I wonder if you had any recommendations of what might
be done in terms of supplying some stimulus to this.

Well, I am sure EPA is always responsive to unsolicited
proposals from anyone in the audience who would be
anxious to measure these rates. I think there is
more and more evidence from some of the
basic studies; especially when you want to get to
low N0x concentrations, some of the flame phenomena are
quite important in NO formation.
I'd like to add to your comment on the uncertainty
of the rate coefficients. When we were doing the
simple Zeldovitch mechanism without the OH reaction
and doing the literature search for the complete sets
of coefficients, we found that, depending on the source,
you get variations of a factor of 1000 in the rate co-
efficient between one source and another. I am not
going to point any fingers. The results in calculating
1-183
-------
your thermal NO are reasonably spectacular.
X
Q: Vic, I wonder if you would comment about the
Zeldovich rate constant. Leeds has readjusted the
values down; what is your own particular opinion
on that?

A: Well, I'm glad you asked that question because it
just so happens that Bob Shaw will be talking about
estimating rate constants a little later on.
SRI has come up with a very neat way of estimating
reactions between unimolecular and bimolecular
species, and it turns out, without scooping Bob too
much, that the estimates come very, very close to
the rates that are measured in the literature, and
in this particular case, for that particular class
of reactions (and the Zeldovich mechanism does fall
into that class), that technique is applicable. I'll
leave it at that and I'll let Bob tell you what the
real answer is.
1-184
-------
CHEMICAL REACTIONS IN THE CONVERSION

OF FUEL NITROGEN TO NO,,
By:

A. E. Axworthy, G. R. Schneider, and V. H. Dayan

Rockwell International, Rocketdyne Division

Canoga Park, California
1-185
-------
INTRODUCTION

An experimental and analytical program was conducted to investigate
the chemical and physical processes involved in the conversion of fuel-
bound nitrogen to NOX in combustion.* To study the types of reactions
volatile fuel nitrogen species can undergo near the surface of a burn-
ing coal particle or oil droplet, model fuel nitrogen compounds were
pyrolyzed in helium at elevated temperatures. Fuel oils and coals
were then pyrolyzed under similar conditions and the fate of the nitro-
gen species determined. This study demonstrated that HCN and, to a
much smaller extent, NH3 are likely to form from nitrogen compounds in
the preflame stages of combustion. A premlxed, flat-flame burner
study was then carried out to investigate the combustion kinetics in-
volved in the conversion of HCN and NH3 to NO and N2 in rich and lean
2~Ar flames.

INERT PYROLYSIS OF MODEL NITROGEN COMPOUNDS

Pyrolysis rates for pyridine and picolines calculated from available
data (Ref. 1) are shown in Fig. 1. The picolines are less thermally
stable than pyridine at lower temperatures, suggesting that hydrogen
abstraction from the substituted methyl group is rate determining.
However, the decomposition rate of pyridine approaches that of the
picolines at higher temperatures. It appears, therefore, that a di-
rect attack on the N-heterocyclic ring will be rate-determining at
the high heating rates involved in combustion. This is the rationale
for employing low-molecular-weight nitrogen compounds, such as pyri-
dine, to study the higher-molecular-weight analogs present in fossil
fuels.

Previous studies of the pyrolysis of organic nitrogen compounds have
not emphasized determination of the fate of the nitrogen (Ref. 1). HCN
has been reported as a product, but there has been no indication as to
whether it is a major or minor product, although the amount of HCN
formed is stated to increase at higher pyrolysis temperatures.

The model compound pyrolysis experiments were carried out in a quartz
flow reactor at the nominal conditions listed in Table 1. The unre-
acted model compound and the organic products were measured by temper-
ature-programmed gas chromatography. The inorganic nitrogen products
*The details of the work described here can be found in the final re-
port for EPA Contract 68-02-0635. The effort on modeling particle
and droplet combustion is not included in this presentation.
1-186
-------
HCN, NH3, and N2 were determined by sensitive GC and colorimetric chem-
ical analysis techniques.

PYROLYSIS RATES

The results obtained on the thermal stabilities of the model compounds
pyridine, quinoline, benzonitrile, and pyrrole are compared in Fig. 2,
which shows percent undecomposed as a function of reaction temperature
at constant pressure and mass flowrate. The relative thermal stabili-
ties of these compounds vary with temperature. For example, benzoni-
trile decomposed to a greater extent than did pyridine below 1020 C,
but appreciable benzonitrile remained unreacted even at 1100 C.

The decomposition curves shown in Fig. 2 were fitted to empirical rate
expressions by integrating the rate expression through the reactor and
varying the kinetic parameters (reaction order, pre-exponential factor,
and activation energy) until the predicted percent-undecomposed curve
followed the experimental curve. The integration procedure took into
account the reactor temperature profile and the change in residence
time with temperature as well as the change in rate with reactant con-
centration and temperature. The data obtained for pyridine, pyrrole,
and quinoline were found to fit the first-order rate expressions listed
in Table 2. The pre-exponential factors for pyridine and pyrrole sug-
gest that the rate-determining step may be a homogeneous unimolecular
reaction. However, the pre-exponential factor for pyridine was found
to vary with the initial pyridine concentration (decreasing with de-
creasing concentration).* The rate expression for quinoline indicates
a possible heterogeneous or chain reaction, and the benzonitrile data
do not appear to fit any simple-order rate expression.

The inert pyrolysis rate expression assigned to pyridine in Table 2
predicts the decomposition half-life denoted by the solid line in Fig.
3. It can be seen that when an N-heterocyclic fuel nitrogen compound
approaches the flame front around a droplet or particle, its decomposi-
tion rate will become very rapid and it should decompose or react be-
fore entering the flame zone. Oxidative pyrolysis and oxidation may
become competitive with thermal decomposition as the flame front is
approached. Preliminary oxidative pyrolysis experiments with pyridine
and benzonitrile have shown that oxygen markedly increases the reac-
tion rate under the conditions of this study. It can be seen from Fig.
3, however, that if the activation energy for oxidative pyrolysis were
only one-half of that of the thermal decomposition process, thermal
*Pyridine was the only compound for which the initial reactant
concentration was varied.
I- 187
-------
decomposition could be the predominate initial preflame reaction of
organic nitrogen compounds at sufficiently high heating rates even
though oxidative pyrolysis is faster at lower temperatures.

PYROLYSIS PRODUCTS

The organic products from the inert pyrolysis of pyridine are shown in
Fig. 4. The more thermally stable products predominate at the higher
temperatures. This suggests that the primary product distribution may
approximate that obtained at 950 C, but the less stable products react
at the higher temperatures before exiting the reactor. More data on
product decomposition rates would be required before the product dis-
tributions could be modeled in enough detail to substantiate this hypo-
thesis. The nitrogen-containing organic products recovered from pyri-
dine pyrolysis (solid curves in Fig. 4) contain a maximum of about 10
percent of the nitrogen from the decomposed pyridine.

HCN was found to be the major nitrogen-containing product from the
inert pyrolysis of pyridine and benzonitrile (Table 3). The amount
of HCN formed increased with temperature, and the HCN yield from pyri-
dine was about 100 percent at 1100 C. No detectable N£ (i.e., less
than 1 percent) was formed in the pyrolysis of these two model com-
pounds. The extent of conversion at 960 C of model compound-N to NH3
was 2 to 8 percent for pyridine, and 3 to 5 percent for benzonitrile.
The carbonaceous residue that formed on the reactor wall was analyzed
for nitrogen in a single experiment with pyridine at 960 C. It con-
tained 49 percent of the nitrogen from the portion of the pyridine
that had decomposed. As shown in Table 4, these data permit the nitro-
gen mass balance for pyridine to be closed at both 960 and 1100 C.

INERT PYROLYSIS OF FOSSIL FUELS

Samples of fuel oils and coals were pyrolyzed in helium under condi-
tions similar to those employed with the model compounds and the inor-
ganic products measured. The fuel sample (about 1.5 mg) was placed in
a small quartz boat that was rapidly moved into a preheated quartz re-
actor, and the volatile products were swept from the reactor in a flow-
ing carrier gas stream. HCN was collected in Na2C03 solution as before
and measured colorimetrieally. Because the fuel reactor promoted de-
composition of NH3 to N2, the sum of any NH3 and N£ formed in fuel py-
rolysis was determined by measuring by GC the N£ present in the car-
rier gas stream. A converter was placed upstream of the GC to convert
any NH3 that escaped the reactor to N£, and an ascarite trap preceded
this to remove HCN, which would also have been converted to N2-
I- 188
-------
Listed in Table 5 are the HCN yields from six No. 6 fuel oils and from
Wilmington crude oil. From 23 to 50 percent of the nitrogen in the
samples formed HCN at 1100 C. The amount of HCN formed at 950 C for
each sample was from one-half to two-thirds of the amount formed at
1100 C. The sums of NH3 + N2 formed from four of these oils are listed
in Table 6 as percent of fuel-N. The NH3 + N2 measured at 1100 C repre-
sented 28 percent of the fuel nitrogen contained in the lowest N-content
oil but only 6 percent in the oil with the highest N content. Table 7
shows that the amount NHj + N£ formed was relatively independent of the
N-content of the fuel when expressed as micrograms of N2 per milligram
of oil suggesting that part of the measured N2 might be dissolved nitro-
gen actually present as molecular nitrogen in the sample.

The HCN and NH3 + N2 yields obtained from two coals are summarized in
Table 8. The results were similar to those obtained with the fuel oils.
Since the residues formed in the fuel pyrolysis experiments were not
analyzed, it is not known if the remainder of the fuel-N was contained
in the residue or if some unmeasured volatile products were formed.

FLAT-FLAME BURNER EXPERIMENTS

HCN and NH3 were added to premixed CH4-02~Ar flames to investigate the
combustion kinetics involved in the formation of NO from these combus-
tion intermediates. NO was added initially in some experiments to de-
termine its fate once formed. The design of the water-cooled burner
is shown in Fig. 5. An uncooled quartz probe and an Al2(>3-coated ther-
mocouple were used to measure species concentrations and temperature
along the centerline of the burner. The burner was enclosed in a glass
envelope so that a pressure of 0.1 atm could be maintained to spread
the flame and permit more detailed probing. The mole fractions of the
major stable species were measured by mass spectrometry. A chemilumi-
nescent analyzer (CA) was used to measure NO directly and N02, HCN,
and NH3 by conversion of these species to NO. A molybdenum catalyst
converter was used at 400 C to convert N02 and at 800 C to convert HCN
(60- to 90-percent conversion to NO, depending on the concentration)
and NH3 (45- to 60-percent conversion). Air was added to the gas sam-
ple upstream of the converter. The nominal burner conditions employed
in this study are listed in Table 9. The detailed probing experiments
were conducted at equivalence ratios, , of 0.8 (fuel-lean) and 1.5.

SCREENING EXPERIMENTS

A series of screening experiments were conducted with the probes posi-
tioned well downstream (80 mm above the burner) to study the effects
of various parameters on the overall NO yield. The overall yields of
NO were nearly the same from HCN and NH3 when the conditions were iden-
tical (Fig. 6).
1-189
-------
High NO yields on the order of 80 percent* were obtained in fuel-lean
flames, but the yield dropped off continually when the flame was made
more and more fuel rich. Varying the pressure, burner feed rate, or
Ax/02 ratio usually affected the flame temperature apparently by chang-
ing the position of the flame front and thereby the rate of heat loss
to the burner. (Argon dilution also decreased the adiabatic flame tem-
perature, and the effects were partially compensating because the lower
temperature and reaction rates moved the flame further from the burner
decreasing the heat loss.) The NO yield appeared to be related to but
not directly proportional to flame temperature, i.e., increasing some-
what with increasing temperature. Changing the reactant concentrations
by varying the pressure or Ar/02 ratio had little effect on the final
NO yield if the effect of the attendant temperature change was taken
into account. This indicates that the reactions that form NO and N2
ar«* of the same order--undoubtedly second. The rate of NO formation
is presumably a strong function of species concentration even though
the yield apparently is not.

DETAILED PROBING EXPERIMENTS

The measured profile of species mole fraction vs distance from the
burner is affected by both the species reaction rate at various posi-
tions and the axial diffusion of the species. The probe data were
analyzed by the procedure of Fristrom and Westenberg (Ref. 2) to sepa-
rate the chemical reaction rates from the effects of diffusion. For
each species of interest, the species diffusion velocity, V±, relative
to the velocity of the total gas stream was calculated at various
points from an approximation of the equation:
i>:
dx
The species flux was then obtained at each point by adding the species
diffusion velocity to the main gas velocity and multiplying by the
species concentration. Unlike the mole fraction profiles, the flux
profiles only change as the result of chemical reaction, and the slope
of the flux curve (on a distance plot) gives directly the species re-
action rate at that point (in moles/cm3-sec). Flux plots will, there-
fore, be employed in the following discussion. The flux profiles are
*The results of other studies of fuel NO formation suggest that even
higher NO yields would have been obtained at lower additive
concentrations.
1-190
-------
plotted on a time scale (obtained by integrating the gas velocity as
a function of distance) to indicate the relative times required for the
various chemical processes to occur. The zeros on the time scales
were defined as the point where the temperature reached 1250 K in fuel-
rich flames (4. =1.5) and 1550 K (<{> =0.8).

The importance of the diffusion correction is demonstrated in Fig. 7
(NHs additive, = 0.8). The NO mole fraction profile suggests that
most of the NO forms in and below the luminous zone. The NO flux pro-
file, however, demonstrates that nearly all of the NO exiting the
burner was formed just above the luminous zone. Most of the NO pre-
sent in the luminous zone was formed above the flame zone and diffused
upstream. A small amount of NO does form near the bottom of the lumi-
nous zone and then reacts farther up in the zone. Likewise, the Gfy
mole fraction is about one-half of its initial value at the bottom of
the luminous zone, but the flux curve demonstrates that CH4 reaction
does not begin until just at that point. The decrease in CH^ mole
fraction below the luminous zone results solely from downstream dif-
fusion of CH. toward the flame front.

The typical temperature profile shown in Fig. 8 increases rapidly
through the luminous zone, where the CH4 reacts, continues to increase
above the luminous zone as CO oxidation occurs, reaches a maximum
about 8 mm above the burner, and then decays slowly.

Before discussing each of the four selected conditions in more detail,
the NO flux yields are compared on a single plot in Fig. 9. The time
scales were adjusted so that the tops of the luminous zones coincide
in this figure. In the fuel-rich flames ( = 0.8), the NO forms rap-
idly and in high yield just above the luminous zone. With NH^ addi-
tive, the NO forms more rapidly than with HCN (reaching its maximum
rate of formation just before C02 formation reaches its maximum in the
case of NH3 and just after with HCN). In fuel-rich flames, the NO
forms in low yield from either additive, and most of the NO forms far
above the luminous zone. An apparent surge of NO formation occurs
just above the top of the luminous zone. The final NO yields obtained
in the screening experiments (Fig. 6) were about 82 percent at = 0.8
and 32 percent at = 0.8 leveled off
at 77 percent in Fig. 9, which is slightly less than was obtained in
the screening experiments. The other flux yield curves were still in-
creasing at the last measurement point and could have reached their
expected values farther above the flame front.
1-191
-------
More detailed flux plots are presented In Fig. 10 through 13 for these
four conditions. Shown on each plot are (1) the initial fluxes of CH4
and additive (calculated from the metered f lowrates to the burner) ;
(2) CH4 and NO flux profiles; (3) "NH3" or "HCN" flux profile, which
is put in quotes because the converter-CA system could not distinguish
directly between NH3, HCN, and NC>2;* (4) an N-BAL flux profile that
represents a nitrogen mass balance and was calculated at each point by
subtracting the measured NO flux and the "NH3" or "HCN" flux from the
initial additive flux, and (5) temperatures at selected positions. The
N-BAL curve includes the N£ flux (multiplied by 2) and the fluxes of
any other nitrogen-containing species that do not form NO upon passing
through the quartz probe and the molybdenum catalyst converter (at 800
C with added air) before entering the CA reactor.

NH3. 4> - 0.8 (Fie. 10)

In the fuel-lean flame, NH^ reacts in and possibly below the luminous
zone. At the top of the luminous zone, the "NH-j" flux, as well as the
measured "NH3" mole fraction, is nearly zero, and the NO flux is zero
(although the mole fraction of NO is large at this point, (Fig. 7)).
These results indicate that, within the accuracy of the rather large
NO diffusion correction, nearly all of the added NH3 is converted in
the luminous zone to nitrogen species that are not measurable in the
probe-converter-CA system. Only a small fraction of these "N-BAL"
species can be N2, because the final yield of N2 at = 0.8 is only 18
percent, and N2 is not expected to react in these moderate- temperature
flames. Small amounts of N02 were observed under these conditions, but
the maximum M>2 flux, which occurred at the center of the luminous zone,
was only 0.3 percent of the initial NH3 flux. Thus, all of the NO forms
from an unidentified reaction intermediate.
HCN.
0.8 (Fig. 11)
In contrast to NH3, most of the added HCN apparently survived the lumi-
nous zone in the fuel-lean flame. The "HCN" reacts rapidly just above
the luminous zone, and NO forms in high yield, reaching its maximum
rate about 0.1 msec after the reaction of HCN reaches its maximum.
(The rate of C02 formation reaches a maximum between these two maxima.)
The relatively slow reaction of HCN appears to be the cause of the NO
forming more slowly from HCN than from NH3 (Fig. 9) .
and "HCN" denote the amount of NO formed in the converter mul-
tiplied by the converter calibration factor for either NH3 or HCN.
I- 192
-------
In the fuel-lean flame with HCN addition, it can be shown that much
and possibly all of the NO is formed from an intermediate (N-BAL spe-
cies) that builds up above the luminous zone. In particular, the max-
imum rate of NO formation is more than twice the rate of HCN consump-
tion at the point where the NO maximum rate occurs.

HCN. - 1.5 (Fig. 12)

In the fuel-rich flame, part of the HCN reacts in the luminous zone
(in two surges), but most of the HCN apparently again survives the
luminous zone but then reacts slowly, forming NO in low yield far
above the flame front. A small surge in NO formation occurs just at
the top of the luminous zone. Because of the low yield of NO, it is
not possible to determine how directly the NO forms from HCN in the
rich flame (i.e., if a time delay is involved with the NO forming from
a relatively long-lived intermediate as was the case in the fuel-lean
flame). In the rich flame, the NO formation rate never exceeds the
HCN consumption rate except just above the luminous zone.

MH. 4> = 1.5 (Fig. 13)

The results obtained with NH3 addition are quite interesting. As dis-
cussed, all of the NHg reacts below the top of the luminous in the
lean flame forming only N-BAL nitrogen species (Fig. 10). In the fuel-
rich flame, the NH3 apparently reacts in or below the luminous zone
with about one-half forming HCN and one-half forming N-BAL species.
Some indirect reasoning was employed in concluding that the "HCN" flux
curve represents mainly HCN, because the converter-CA system could not
distinguish directly between HCN and NH3. However, the differences in
conversion efficiencies for HCN and NH3 did permit this tentative con-
clusion to be drawn from the experimental flux curves for "NH3" (not
shown in Fig. 13) and "HCN."

De Soete (Ref, 3) has reported that NH3 is partially converted to HCN
in a fuel-rich C2H4-02~Ar flame at atmospheric pressure. Another strong
argument for the conversion NH3 to HCN in our study is the similarity
of the flux profiles above the luminous zone in Fig. 12 and 13. Thus
the mechanism of NO formation in the fuel-rich flame appears to be the
same whether the additive is NH3 or HCN, and involves the formation of
NO from HCN.

Comparison of Results

The conclusions drawn from the addition of NH3 and HCN to lean and rich
methane flames are compared in Table 10. NH3 reacts much more rapidly
than does HCN in both flames. The similarity of the overall NO yields
from NH3 and HCN in the rich flame (Fig. 6) apparently results from the
I- 193
-------
fact that the NO forms from HCN with either additive. The reason for
the similarity of the NO yields in the lean flame is not as apparent,
since a common stable intermediate is not formed. However, a common
free radical intermediate could be involved.

Effect of Argon Dilution

An experiment (not shown) was conducted with HCN addition at <|> = 0.8
in which the Ar/02 ratio was increased by 40 percent, and the total
burner feed rate was reduced to increase the heat loss to the burner.
Although these conditions reduced the maximum flame temperature by 150
degrees and the initial reactant concentrations by 23 percent, the max-
imum reaction rates for NO, HCN, CH4, and C02 only decreased by 30, 8,
42, and 29 percent, respectively. An analysis of these results, pre-
sented in the final report, indicates that certain compensating effects
control the maximum reaction rates, but that the measured rates do have a
normal dependency on temperature and species concentration.

Effect of NO Addition
When NO was added initially (at 625 ppm) to the fuel-rich flame, to-
gether with HCN (Fig. 14) or alone (Fig. 15), the rapid consumption of
NO occurred just below the top of luminous zone. NO then formed rap-
idly just above the luminous zone and decayed rapidly again farther
downstream. The magnitude of the NO surge was greater with HCN pre-
sent, but the shape of the NO flux profile in this region is highly
dependent on the diffusion correction (i.e., on the exact shape of the
NO mole fraction curve). In the absence of added HCN, the consumed NO
was partially converted (40 percent) to a measurable nitrogen species
that was presumably HCN. When NO was added to the fuel-lean flame,
together with HCN (not shown), a small consumption of NO occurred in
the luminous zone before the formation of NO from HCN obscured any fur-
ther reaction of the added NO. These results, with added NO, are in
general agreement with the results of De Soete (Ref. 4) that are sum-
marized in Table 2.

Reaction Mechanism and Conclusions
A possible reaction mechanism for the formation of NO and N2 from NHj
is presented in Table 12. The experimental results indicate that a
rather long-lived nitrogen intermediate forms in high concentration in
the fuel-lean flame. A likely candidate that could have the required
characteristics of (1) low reactivity in the luminous zone and (2) not
forming NO in the probe-converter system is the NCO radical.* Bosco
*NCO could form N2 in the probe via 2 NCO - N2 + 2 CO.
1-194
-------
(Ref. 5) has shown that CN radicals react with Q£ to form NCO and 0
atom. The competing reaction to form CO and NO is slower by a factor
of more than 7 at room temperature. Davies and Thrush (Ref. 6) have
proposed that NCO is an intermediate in the formation of NO from the
reaction of oxygen atom with HCN. More recently, Mulvihill and Phil-
lips have proposed NCO as the precursor to NO formation from cyanogen
added to a low-temperature H2-N2-02 flame (Ref. 7).

According to the proposed reaction scheme, CN radicals form in fuel-
lean flames from NH3 and (more slowly) from HCN, and then react to
form NCO:
CN + 0,
NCO + 0
CN + OH = NCO + H.

The NCO radical is postulated to be relatively stable and react rapid
ly with oxygen atoms above the luminous zone to form NO:

NCO + 0 - NO + CO.

The most likely path for N2 formation appears to be the reaction pro-
posed by Mui and Back (Ref. 8):
NCO + NCO
+ 2 CO.
The amount of N2 that can form via the reverse Zeldovich reaction:

N + NO = N2 + 0

is limited by the competing reaction

N + 02 = NO + 0.

The mechanism of NO formation in fuel-rich flames may also involve the
NCO radical, but there is little evidence relating to this point. The
rapid consumption of NO in the fuel-rich flame must result from reac-
tion with hydrocarbon radicals. It can be shown that the reverse
Zeldovich reaction

0 + NO = N + 02

is much too slow to cause the observed consumption of NO. Additional
data and theoretical modeling are required to test this somewhat spec-
ulative overall mechanism for the formation of fuel NO.
I- 195
-------
The conclusions from the burner experiments are outlined in Table 13.
The major conclusion is that the formation of NO from NH3 and HCN is
a relatively slow process even under fuel-lean conditions where the
NO yields are high. The NO forms in lean premixed flames in the same
region that CO oxidation occurs.

CONVERSION TO SI UNITS
To Convert From
atm
inch
kilocalorie (Kcal)
mil
psi
torr
To
pascal (Pa)
metre (m)
joule (J)
metre (m)
pascal (Pa)
pascal (Pa)
Multiply
1.013 x 105
2.54 x 10~2
4184
2.54 x HT5
6894
133.3
and
Pressure (Pa) - 9864* + 6894 x Pressure (psig)
REFERENCES

1. Hurd, C. D., and J. I. Simon, J. Amer. Chem. Soc.. 84, 4519 (1962).

2. Fristrom, R. M., and A. A. Westenberg, Flame ; Structure, McGraw-Hill,
New York, 1965.

3. De Soete, G., Rev. Inst. Fr. duPetr. Ann. Combustion Liquide.
28(1). 96 (1973).
4. De Soete, G., and A. Querand, Preprint, Inst. Fr. du Petrole, 1973.

5. Bosco, N., Proc. Roy. Soc. (London). 283A. 291 (1965).

6. Davies, P. B., and B. A. Thrush, Trans. Far. Soc.. 64, 1836 (1968).

7. Mulvihill, J. N., and L. F. Phillips, 15th Combustion Symposium
(International). Tokyo, 1974.

8. Mui, J.Y.P., and R. A. Back, Can. J. Chem.. 41, 826 (1963).
*At local barometric pressure of 74 torr.
1-196
-------

I-
TEMPERATURE. C
100 775 750 726 700
1.00
looorr.
Figure 1. Arrhenius Plots of Pyrolysis Rates for Picolines
and Pyrldine Obtained by Kurd and Simon
00
rvmioic
rvxiomt
OU.HOL.Mf
•ENZOMTMILI " °*
P-ltPHO
PLOWKATt - «.! CCMHI
EHAMfTtH • U MM 10
M* 1MO
TEMPER ATUHE.C
Figure 2. Comparison of Model Compound Decomposition Rates in Helium
1-197
-------
TEMPERATURE. K
KXniMOfMO 1400 1200 1100 1000
«
i 0.1

i 0.01

I Kf3
10*
k • U X n" EXP ( JO.OOO/RT). MC
0.5 0.1 0.7 0.1 O.f 1.0 1.1
1000/T. K'1
Figure 3. Calculated Decomposition Half-life
for Inert Pyrolysis of Pyridine
TEMPERATURE °C
Figure 4. Organic Products From Inert Pyrolysis of Pyridine
1-193
-------
J INCH IHOMI
9CMEDUU «0 riH
ALUMNA TIMING

tCHIIL luftOKT HIRE
Mill. COATED
THf RMOCOUn.E WIRE
•— t-CAGE
ITAIN1.E»«TEEL (CREEK
COOL INC
MATER
OUT
Figure 5. Flat-Flame Burner
1-199
-------
90
*- 80
5 70
O 60
SO
*>
20
HCN ADDITION
--- NH, ADDITION
(FUEL-LCAN)
(FUCL-RICH)
0.4 0.6 0.8 1.0 1.2 1.4
EQUIVALENCE RATIO. 0
Figure 6. Conversion of Additive to NOX as a Function of
Equivalence Ratio in Screening Experiments
(P = 76 torr, DR « 1, w - 7520 cc/min and
Additive = 2500 ppm)
TIME. MSEC
Figure 7. Effect of Diffusion Correction
on CH4 and NO Profiles
I- 200
-------
2000
MJ
cc
D
1500
in
OL

UJ
1000
TEMPERATURE
PROFILE
DISTANCE, MM
10
Figure 8. Typical Temperature Profile
100

60
c
SO
3
UJ
20

-10

•20
LUMINOUS ZONE
I* - 1.BI
LUMINOUS
1 2

TIME, MSEC
Figure 9. Comparison of NO Flux Yield Profiles
at Various Conditions
I- 201
-------
KOSOZ
»OCOZ
X0i6l
XOSSl
§
u
Tl
«w
o
PK
a 3
V I
1
•s
o

I
fe
(U
h

bO
•H
,OIX
S3ID3dS-N
.01 X Xm J S9I3UTN
•3
a
at
in « m CM ^ e

X Xnid SlIDidS-N
00
•H
1-202
-------
X
X 3
M 2
Ul
u
£ i
2.5

2.0

1.5 "^
^
X
X
1.0 =
u.
•»

0,5 °
-0.5
TIME, MSEC
Figure 14. Flux Profiles (NO and HCN, rich flame)
TIME, MSEC
Figure 15. Flux Profiles (NO alone, rich flame)
1-203
-------
TABLE I. MODEL COMPOUND PYROLYSIS CONDITIONS
• 850 TO 1100 C
• QUARTZ REACTOR
• 2 MM ID
• 1 CC VOLUME
• 0.5 SEC RESIDENCE TIME
• HELIUM OR HE-O2 CARRIER GAS
• SAMPLE INJECTION
• 0.2 fJ-L PR E VAPOR I ZED AT 160 C
OR 1 CC CARRIER GAS SATURATED
WITHPYRIDINE
TABLE II. KINETIC PARAMETERS FOR MODEL COMPOUND PYROLYSIS
•dC/dt = k(C)N = A •
-------
TABLE III. HCN YIELD FROM MODEL COMPOUNDS
1 TEVPERATURE. C
PYRIDINE

PYRIOINE

BENZONITRILE
BENZONITRILE

956
960
959
1102
1106
1107
1107
957
955
1107
1106
PERCENT
DECOMPOSED
162)
65
65
100
100
100
100
171)
69
98.0
98.0
PERCENT N CONVERTED TO
HCN'
35.6
45.7
36.8
106
98
98
105
45
53
82
81
•BASED ON N CONTENT OF DECOMPOSED PORTION OF SAMPLE
TABLE IV. NITROGEN BALANCE IN PYRIDINE PRODUCTS
PRODUCT
RESIDUE
HCN
BENZONITRILE
QUINOLINE
ACRYLONITRILE
NH3
N2

PERCENT N IN PRODUCT*
960 C
49
40
5
3
3
<8
<0.1
ss 100
1100C
NMt
102
0.5
<0.1
<0.1
MM
NM
102
'BASED ON N IN DECOMPOSED PORTION OF
SAMPLE (65 PERCENT AT 960 C AND 99.9
PERCENT AT 1100 Cl
+NM = NOT MEASURED
1-205
-------
TABLE V. HCN YIELDS FROM OIL PYROLYSIS
SAMPLE
GULF NO. 6 FUEL OIL
(VENEZUELAN CRUDE 1
GULF NO. 6 FUEL OIL
VARIOUS CRUDES)
GULF NO. 6 FUEL OIL
(MAINIY CALIFORNIA)
CONOCO NO 6 FUE I 01 L
•EPA" NO oFUEL OIL
(IN HOUSE)
WO. 6 FUEL OIL
(ULTHASYSTEMSI
WILMINGTON CRUDE OIL
X N
0.43

0.44

1.41

03
0.22

036

0.63
*S
2.3

073

0.7
0.9

1.6
% N TO HCN
(1100 Cl
329

388

23.2

42.0
37.0

36.2

498
1950 C!
19.4

23.2

145

20.4
246

17.8

303
HCN (950CI/MCN I1100CI
059

06O

0.63

0.49
067

0.51

0.61
TABLE VI. PERCENTAGE YIELDS OF NH3 + NZ FROM OIL PYROLYSIS
OIL SAMPLE
EPA NO. 6
WILMINGTON CRUDE
GULF NO. 6
ULTRASYSTIM NO. 6
%N
0.22
0.63
1.41
0.38

950 C
23.8 iO-5
10.5 ±0.8
5,2 ±0.8

1100C
28.2 ±1.0
10.9+0.2
5,9+0,4
16.1
TABLE VII. NH3 4- N2 YIELDS FROM OIL PYROLYSIS CALCULATED
AS yg of N2 RECOVERED PER MG OF OIL
OIL SAMPLE
EPA NO 6
WILMINGTON CRUDE
GULF NO. 6
ULTRASYSTEM NO. 6
%N
0.22
0.63
1.41
0.38

950 C
0.52i001
0 66 10 05
0.73 ±010

nooc
0.62 10 02
0.75 iO 01
0 83 10 05
o.ei
1-206
-------
TABLE VIII. INORGANIC PRODUCTS OF COAL PYROLYSIS
PERCENT FUEL-N TO HCN:

EPA COAL (1. 17% N)

IFRF-N (1.8%N)

N2. %OF FUEL-N

EPA COAL (1. 17% N)
EPA COAL (1.17%N)
955 C
20.3, 18.0
8.3, 10.5
0.97, 1.23
1106C
29.8, 30.2

26.2, 22.7
1-207
-------
TABLE IX. BUHNER CONDITIONS (NOMINAL)
• PR£MlXEDCH4-02 AR FLAT FLAME
• DILUENT RATIO - 1
DR « *f. . /^Z.\
O2 \N2 I AIR
• 0.1 ATM
• EQUIVALENCE RATIO IFUEL/AIR), 0 = D.5 TO 1.5
• ADIABATIC FLAME TEMPERATURE CALCULATED
2230 K 10=1.5) 2300 K If' 0.8)
• MEASURED TEMF-ERATURE (CORRECTED FOR RADIATION)
1980 K TO 2080 K
• VELOCITY AT TMAX « 240 TO 340 CM/SEC
• ADDITIVES HCN (2500 PPM), NH3 (250O PPM), NO (625 PPMI
• EQUIVALENT WT %-N IN CH< FOR 2500 PPM ADDITIVE
0= 1.5, 1.6% t = 1.0, 2.3%, 0 = 0.8. 2.8%
TABLE X. COMPARISON OF RESULTS OF HCN AND NH3 ADDITION
ADDITIVE
HCN
NH,
(HIGH NO YIELDS)
65% HCN SURVIVES LZ"
HCN REACTS RAPIDLY ABOVE LZ
ALL NO FORMS RAPIDLY ABOVE LZ
NO FROM INTERMEDIATE
ALL NH3 REACTS BELOW TOP LZ
NO FORMS RAPIDLY ABOVE LZ
ALL NO FORMS FROM INTERMEDIATE
I LOW NO YIELDS)
70% HCN SURVIVES LZ
HCN REACTS SLOWLY ABOVE LZ
ALL NO FORMS SLOWLY ABOVE LZ
NO FROM INTERMEDIATE AND/OR HCN
ALL NH3 REACTS BELOW TOP LZ
(ONE-HALF CONVERTED TO HCN)
"HCN" REACTS SLOWLY ABOVE LZ
MOST NO FORMS SLOWLY ABOVE L2
NO FROM INTERMEDIATE AND/OR HCN
•LZ DENOTES LUMINOUS ZONE
1-208
-------
TABLE XI. NO ADDITION RESULTS OF DE SOETE
O2 - AR, 1 ATM, PREMtXED
= O.7 (23002400 K)
NO = 215 PPM*-160 PPM*-210 PPM
0= 1.5
NO = 200 PPM—50 PPM*-75 PPM
NO = 265 PPM*-60 PPM*-70 PPM
(HCN = 180 PPM WHEN NO = 60 PPM)
TABLE XII. PROPOSED MECHANISM FOR NO FORMATION FROM HCN AND NH,
R CN
HCN + 0—IMCO + H
HCN + OH--CN + H20
CN +02»~NCO+O
CN + OH*-NCO + H
NCO + 0—NO + CO
CN + 0 «-CO + N
N + NO*-N2+0
N +O2»-MO +O
NCO + NCO*-N2 + 2 CO
NCO + N*-!M2 + CO
KCAL/MOLE
-1.7
+4.4
6.3
23.3
-102.5
77.0
75.0
-31.8
129.0
-177.5
1-209
-------
TABLE XIII. CONCLUSIONS FROM BURNER STUDIES
• MOST HCN SURVIVES THE LUMINOUS ZONE OF LEAN OR RICH FLAMES

• LEAN FLAMES: NO FORMS RAPIDLY JUST ABOVE THE LUMINOUS ZONE
FROM A REACTION INTERMEDIATE (PROBABLY NCO). NH3 REACTS
MUCH FASTER THAN HCN. N2 FORMS FROM NCO.

• RICH FLAMES: ADDED NH3 REACTS RAPIDLY IN THE FLAME FORMING
HCN. SOME NO AND N2. HCN (ADDED OR FORMED FROM NH3) REACTS
SLOWLY IN THE POST-FLAME GASES FORMING NO IN LOW YIELD.

• IN LEAN FLAMES, FUEL-NO FORMS IN THE SAME REGION AS DOES CO2-
NO FORMS MUCH LATER IN RICH FLAMES.

» NO ADDED TO RICH FLAMES IS PARTLY CONVERTED TO HCN AND
OTHER SPECIES IN AND ABOVE THE LUMINOUS ZONE. THE HCN
REACTS SLOWLY IN THE POST-FLAME GASES TO FORM NO AND N2.
NO CONSUMPTION IS NOT INITIATED BY REVERSE ZELDOVICH
MECHANISM.
1-210
-------
1:15 p.m.
Chemical Reactions In the Conversion
of Fuel Nitrogen to N0x
Dr. Arthur E. Axworthy - Rocketdyne
Q: I think there is some support for your NCO thinking,
Art, but I can't remember the reference now ....

A: Moba, Hill, and Phillips at the 15th meeting of the
Combustion Institute proposed NCO as an intermediate
when they added cyenogen into what I believe was a
methane flame; at least it was a hydrocarbon flame.
Q: But there is an older reference. Somebody did, I
think, a spectroscopic study on the oxidation of
HCN where 85% of it went to NCO plus OH.

Aj Norish did some work that showed NCO radicals ,then
Davies and Thrush on the 0 atom reaction with HCN where
they proposed NCO as an intermediate. Yes, there ^s
a lot of background for it.

Q: We were trying to justify the possibility of the
NCO to even explain our early NO™ on the basis that
we would not require NO as a precursor. One reaction
you don't have there, and maybe it's quite legitimate
not to have it, is NCO plus 0 where if the 0» came
from the back side you may form N0_ but you'd have to
go through a kind of shifting around. It's probably
something that Bob wouldn't put up with anyway.

A: He doesn't like, at least on the last meeting, the
0 atom plus NCO because of the spin conservation.
On your early slides I missed a particular thing.
Were you using a platinum resistance wire when you
were heating your medium in the quartz, that 1 cc
1-211
-------
quartz reactor?
A: It was just quartz, the wire was on the outside of the
quartz. Just a heated quartz tube. We were passing
the helium carrier gas.
Q: Those ups and downs that you have for your flux
profiles there.•.how do you know that for those very
reactive intermediates those are really not computa-
tional artifacts?

A: I'll try to discuss that. It turns out that the
flux curves are extremely sensitive to the shape that
you drew through your measured mole fractions; so
the method we used, the Tristrum and Westenburg method,
is open to these types of artifacts.
Q: The very computational method is dependent on the fact
that the species diffuses and doesn't react. Yet
those species that you report your profiles for are
highly reactive. It is questionable; can they really
diffuse as far as this sort of a simple diffusion
computation will indicate, and that's where you
draw your flux profiles from?

A: No, the method takes that all into account accurately,
and we think the diffusion coeffients are absent.

Q: The reactive term is absent in diffusion/velocity
computations. All you have there — the capital
V -- does not have any reactive term.

A: O.K., right, we correct for that.

Q: So, therefore, for very reactive species, you are
really undercompensating for any possibility that
the species does not go as far as this indicates.
1-212
-------
A: Obviously, if something reacts and the flame is consumed,
diffusion can affect where it reacts. I'll agree to that,
and the method could be inaccurate in either direction, but
I can't see how you say it undercompensates in one direction,
Q:
If you compute diffusion velocity, the capital V , as
was first formulated by Hirshfielder and Curtis, and
then taken up in Tristrum and Westenburg's book, has no
reactive term in it. In other words, you compute or
assume a diffusion coefficient for the species in question.
For some of the reactive species, you don't even have
anything to take from literature. You just make one up.
Well, that's the least of the errors that you produce
in there. But when you then compute just the diffusive
term there, then, this assumes that the particle is
completely free to go as if it would be going in a completely
inert surrounding. And that is not the case.
A: Well, so what actually happens is that it will diffuse
in and will reach steady state. It'll be reacting, if
it does react; or if it doesn't react, it'll just
diffuse and reach a concentration where diffusion equals
the flow, and it will just level off. But, that's a subtle
thing -- I could be wrong. Maybe you could give me some
insight into it. But the way I look at it is that it could
be off in either direction.
It is a minor point, but on an earlier slide, you
speculated that if you had oxygen presence in pyrolysis
rather than nitrogen, the rates would change and the
activation energies would change. And you show at higher
temperatures that the rate of pyrolysis in oxygen is
lower than in inert.

That was a hypothetical curve assuming that the activation
energy was about half.
1-213
-------
Q: I'd just like to take a quick poke at this if I can
and maybe get a comment from you now or perhaps later.
Your mechanism for ammonia addition had the reaction
N + R going on to other things. Ammonia breaking
down to N, then N + R going on to other things, and
then NO eventually forming from some NCO reaction with
O^- Now, my question is why in these flames do you
feel that the N atom would preferentially react with
some radical R to form these species rather than
hydroxyl radical, which has a very rapid rate constant
at these temperatures -- on the order of 1 X 10 cm cubed
per mole second? I don't understand why you've left that
reaction out of consideration.

A: Well, I didn't show it really -- I showed N atom
before we got more reactions, but I had N* on that
second one. It could have been NH. It may only go
to NH and then react with the CH or C_. It may never
get the N atom.

Q: Well, there are some data on ammonia flames by Kesken
and Hughes which suggest that our whole nitrogen pool
equilibrates very rapidly in these flames, and if
that's the case, then I would question that assumption.
I'm having some trouble accepting conceptually your
method.

A: The reason we did it that way is because our results
showed that. The ammonia decays and NO doesn't form
in our flame, so we have to assume that it went by
that path. And I say this is just a possible route
that has to be modeled.

Q: I would say your experimental results are controversial
because there are other data out with ammonia added
to flames and ammonia reactions in shock tubes which
show the NO coming out much earlier.

A: Yes, I realize that. Sawyer said it can't be, you
must make NO when ammonia reacts. But I just have

1-214
-------
to report what I think we see.
Just a quick comment, Art. Bob Woolfolk and I have
done some thermodynamic calculations on solvent
refined coal and peradine which overlap, particularly
the peradine calculations overlap yours. These are
thermodynamic calculations, not kinetic. But these
show that thermodynamically, at equilibrium,
most of the fuel nitrogen ends up as N«; 98% of it.
There is only about 1 or 2% ammonia and HCN.
1-215
-------
1-216
-------
PREDICTION OF PREMIXED LAMINAR FLAT

FLAME KINETICS, INCLUDING THE

EFFECTS OF DIFFUSION
by
R. M. Kendall
J. T. Kelly

Aerotherm/Acurex Corporation
Mountain View, California
W. Steven Lanier

Environmental Protection Agency
Combustion Research Section
Research Triangle Park, N.C.
1-217
-------
PREDICTION OF PREMIXED LAMINAR FLAT
FLAME KINETICS, INCLUDING THE
EFFECTS OF DIFFUSION

R. M. Kendall and J. T. Kelly
Aerothenn Division/Acurex Corporation
Mountain View, California

W. Steven Lanier
Environmental Protection Agency
Combustion Research Section
Research Triangle Park, N.C.
Pollutant formation in flames depends on the interaction of com-

plex chemical and physical processes. Comprehensive theoretical treat-

ment of these processes requires the application of multivariable boun-

dary value problem solution procedures. The PRemixed One-dimensional

Flame computer code (PROF), described herein, has been developed to

treat this difficult class of boundary value problems in an efficient

manner. The technique is based on a matrix or quasilinearization solu-

tion procedure which employs predictor and linearized corrector steps

to solve the finite difference form of the flame conservation equa-

tions. The method includes a general chemical kinetics formulation and

unequal diffusion of all species.

Several initial flame system calculations, with up to 42 kinetic

reactions and 19 species, are presented to demonstrate the speed, reli-

ability and accuracy of the PROF code in predicting complex combustion

and pollutant formation chemistry. In particular, CH./AIR and H2/AIR

flames are calculated and their laminar flame velocities are favorably
1-218
-------
compared with experimental data. tL/CL/N- and CO/02/N2/H20 flame cal-

culations with NO pollutant formation kinetics are also presented and

species concentration results are favorably compared with experimental

data. The H and CO flame predictions suggest that the rapid formation

of NO in the flame zone is a result of the presence of super-equilibrium

0 atom concentrations which generate NO through the reaction:
N2 + 0 -»• NO + N

Results of a CO/02/N /H 0 flame calculation to which NO is initially

added are also presented. Agreement of predicted and experimental de-

tailed species concentrations is good. The numerical results indicate

that HO- plays a major role in flame zone NO, N02 kinetics through the

reactions:
NO + H02 -»• N02 + OH
and
NO, + 0 + NO + 0,
INTRODUCTION

Experimental probing of flat flame burners has yielded considerable

information on the formation of pollutants in flames. However, in many

cases a clear interpretation of the data is not possible due to (i)

lack of fundamental knowledge of elementary reaction steps,
1- 219
-------
(ii) insufficient amount of specie concentration data, (iii) unknown

experimental probe effects, and (iv) obscuration of chemical events due

to the complex interaction of chemical and physical (i.e., diffusion)

processes. Utilizing elementary kinetic reaction and diffusion data,

comprehensive theoretical predictions of flame combustion and pollutant

formation processes can be a valuable aid in the interpretation of

data previously obscured by effects (ii) to (iv). Further, having es-

tablished the validity of the theoretical flame model through comparison

with data at a single condition, several calculations may be carried

out to determine the effect of variation of pressure, preheat or initial

species concentrations on pollutant formation in flames. Even though

the advantages of theoretical modeling of flames is clear, only recently

has significant progress been made towards a comprehensive treatment of

flames. This delay is partly due to the difficulty in solving the

multivariable boundary value flame problem formed as a result of the

mathematical modeling of coupled chemical kinetic and diffusion pro-

cesses. The PRemixed One-dimensional Flame computer code (PROF), de-

scribed herein, has been developed to treat this difficult class of

boundary value problems in an efficient manner.

Several mathematical approaches have been applied to the solution

of the flame equations. In Reference 1 a "shooting" method is described

where, beginning at the burned gas state (presumed known), the calcula-

tion marches upstream until a match of the unburned state is realized.
1-220
-------
For simple systems, this approach can be successful but is unwieldy for

complex systems. Others (References 2, 3, 4, 5) have utilized time de-

pendent solutions, seeking the steady state solution as the time asymptote.

Although relatively simple to formulate, some of these procedures suf-

fer when complex coupled chemistry is introduced due to the inevitable

"stiffness" of the resulting equations. Also, several hundred time

steps are required to reach the steady state resulting in lengthy com-

puter runs.

In the treatment of "stiff" boundary value problems, a class of

procedures, variously described by "quasilinearization", "Newton-

Raphson" or simply "matrix", have achieved considerable success. Wilde

(Reference 6) presents such a method and utilizes it with considerable

success to calculate several flame velocities. Matrix methods (e.g.,

Reference 7) have been widely used for boundary and shock layer solutions

including coupled field effects, chemical nonequilibrium, and complex

interactive boundary conditions.
of:
The efficient matrix procedures applied in the PROF code consist

1. Reduction of the differential equations to sets of algebraic

equations through finite difference relationships

2. Solution of these algebraic equations by an iterative linear-

ization procedure. This solution is performed simultaneously
1-221
-------
over the entire domain of interest and includes additional

constraints (eigen-functions) which allows the easy deter-

mination of flame speed.

In the next section the PROF code formulation is briefly discussed.

Following this, several initial flame system calculations are presented

which demonstrate the speed, reliability and accuracy of the PROF code

in predicting complex combustion and pollutant formation chemistry.

FORMULATION

The flow model for the PRemixed One-dimensional Flame (PROF) code

is depicted in the schematic below. Flow, m, is presumed to be uniform

across plans perpendicular to the axis. Diffusion of species, J., and

heat, q, occur only in the axial direction.
1-222
-------
Several simplifying assumptions have been utilized in the PROF

code formulation, including steady state one-dimensional flow, perfect

gas behavior, mean Lewis number of unity, negligible pressure drop, ra-

diation and thermal and pressure diffusion. Also, heat and species

losses at the flandholder are neglected. Applying these simplifying

assumptions and standard reduction procedures, the mass, momentum and

energy conservation equations can be reduced to a set of equations for

the conservation of species and enthalpy along the flame axis. Since

the formulation of the enthalpy conservation equation follows a develop-

ment similar to the species conservation equation, only development of

the latter is presented.

The species conservation equation for the flame problem is:
dY^ , f
i -± . AW mL-\
ds ids!
dY-
(1)
where
m = total mass flow in flame, gm/sec
A = cross sectional area of flame, cm2
s = axial distance
X = mole fraction of species i
Y. = mass fraction of species i
W. = kinetic production rate of species i, gm/cia3sec
I- 223
-------
Pm = pressure molecular weight product, atm gm/mole

p = density, gm/cm3

M = molecular weight of i

V = diffusion coefficient for species i into the mixture of other
im .
species
Introducing
r = mole kinetic rate of production of species i, gm-mole/cm3sec
ex. = gin-mole of species i per gm of system

and dividing Equation (1) by M., yields
m
do,

ds
(2)
In the development of the diffusional flux term, the bifurcation ap-

proximation developed in NASA CR-1063 (Reference 8) has been adopted.

Basically, the approximation assumes that the contributions of species

i and j to the binary diffusion coefficient, V.., can be separated via;
U FiFj
(3)
where D is a reference diffusion coefficient and F, and F. might be

termed diffusion factors.* The primary advantage of this approximation
Equation (3) is exact for binary and ternary systems and should be

thought of as a good correlating function of diffusion data for larger

systems.
I- 224
-------
is that it permits an explicit formulation for diffusional mass flux,

j., to be developed from the Stefan-Maxwell relations

y F _ds
J dy'
__i i z£. F i
M \ds ~ 1 ds
where
and
In calculating multicomponent diffusion this formulation results

in a considerable increase of computational efficiency over approaches

which apply the full Stefan-Maxwell implicit equations. For Y. ap-

proaching zero or unity or equal F.'s:
F± ds
im ds
(5)
and is typically a good approximation otherwise. Equation (5) will be

used as the defining relation for V. .
im

In Equation (3), D is typically taken as the self diffusion coeffi-

cient of some reference species, e.g., CL. The F. are then found to be

constants effectively independent of temperature and pressure. From

References 8 and 9 the D value, with ()„ as the reference species, is

found to be
D = 0.172 x ICT1* Tl<659/P (cmVsec)
(6)
,1.659
1-225
-------
Introducing Equations (5) and (6) and the normalized distance

coordinate s, defined by ds ~ ds/ApT1'659, into equation (2)
m dcu - A^rjpT1-659 ' ds • d p-
^1 1

^ J
(7)
The finite difference form of Equation (7) is:
L ,a \_.
(\ \J •»
°
•a.
a.
n
a
n-1
(8)
-a
i I a a .
n\ n n-1
where n denotes axial grid location index and the composite quantities

are defined as
_ A26
m
D
s . - s = a -6
n+1 n n
In this equation, well stirred reactor control volumes are assumed to

Sn + Sn+1 , Sn-1 + Sn
exist between axial stations
and
I- 226
-------
Equation (8) can also be arranged to yield an equation for a.
in terms of a composite convection and diffusion "initial" concentration,

P1.6S9\
n
r T1
a (a , + a )
n n-1 n
(9)
1 + TT^ J a<
O 1 / i i
k n-1/ n-1
a ai
n n+l
n
_ o

" a
i,n
It is this set of relations that is solved in the kinetic sub-

routine to PROF, given values ot. and the necessary state parameters
l,n

or equations, e.g., an energy equation and assigned pressure. The

energy equation can be developed in a similar manner and leads to
h - a
n n
PmT
0-659
a , + a
n-1 n
R * / , fi .4- K /! x * VI
n/ 2 1 -f b ,• [ — + 1
n ! I a a , J
L \ n n-l/J
' bi \ bi
1 + —^ h . + -^
i a . / n-1 O
V n-1 / n
n
(10)
n
where Q is the heat loss to the environment per unit volume of the re-

acting gases. Implicit in this equation is the use of a mean Lewis

number that leads to a heat flux relation of the form
1-227
-------
_k_ dh
Cp ds
D dh
ds
where F is the Lewis number for 0,.,.

The procedure whereby Equations (9) and (10) are solved with the

appropriate kinetic relations is the key to the success of the technique.

This procedure is discussed in detail in References 10 and 11 and only

a brief description of the approach will be presented herein.

The kinetic production term, r , in Equation (9) can be written
as:
ri =
.in P. - Jin K
m
YVTP
^-^ j j
where for reaction m

k = forward rate constant in moles — atm units
m
T
p. = third body efficiency of species j
hi

K = equilibrium constant for reaction m
pm

In the Newton-Raphson solution procedure for the species conser-

vation Equation (9), the errors in all equations are calculated and a

matrix indicating the variation of these errors with respect to all

variables is calculated. This matrix is inverted and multiplied by

the errors to give a set of variable corrections. When the initial
1-228
-------
guess is close to the answer, or, the equations are fairly linear, the

procedure is rarely defeated. However, as discussed at some length in

References 10 and 11 the solution of these equations is subject to

several pitfalls which need to be carefully circumvented if efficient

and reliable coupled kinetics/diffusion boundary value problem solutions

are to be achieved. The most discussed problem is associated with k
m
becoming very large, forcing the bracketed term in Equation (11) to zero

to maintain a finite net contribution of the term to the species conser-

vation Equation (9). The forcing of the bracketed term toward zero is

equivalent to imposing equilibirum. Through experience, it has been

found that a combination of gradual recharacterization of equations and

damping has proved highly successful at overcoming the difficulties as-

sociated with chemical systems near equilibrium.

A side benefit of the Newton-Raphson procedure is the immediate

accessibility of a linearized representation of the chemical equations.

Referring to Equation (9), one can obtain its linearized replacement
3d.
- mp)
(12)
where the subscripts C and P refer to the corrector and predictor steps,

respectively, and the repeated j index implies summation.
1-229
-------
Thus, the solution procedure becomes:

1. Initial values are selected for all a.. (To date a crude lin-

ear interpolation between the initial and equilibrium com-

positions are used at all n.)

2. At each n:

a. Evaluate a. from Equation (9) using only the guessed value

of a
3.
4.

5.
b. Solve Equations (9) and (10) in the chemistry routine and

evaluate the (9a,/3a.) matrix and the (9a,/3m) vector.

Formulate the block tri-diagonal matrix resulting from the sub-

stitution of Equation (9) into Equation (12), and solve it

•

simultaneously for all a. r in terms of mp.
JL 9 L» u

Introduce constraint to determine nu and thus all a, .
v» X , t*

Using these a. r as new guesses, repeat the predictor step
1,L

(Step 2) .
Although solutions have been achieved by simply repeating the pre-

dictor step, the addition of the matrix corrector step has reduced the

number of required passes by factors of from 2 to 10. The constraint

introduced at Step 4 to calculate the mass velocity can be an assigned

temperature or an a, at some grid point. By requiring some value inter-

mediate between the inlet and equilibrium state to exist at a grid
1-230
-------
point well removed from the inlet, flame speed calculations can be

achieved.

The proof of any procedure is in its application to problems.

This application is presented in the next section.

DISCUSSION OF RESULTS

Several preliminary calculations are presented which demonstrate

the speed, reliability and accuracy of the PROF code in predicting com-

plex combustion and pollutant formation chemistry. Comparison with

data is shown to be good for all the results presented.

Even though the calculation of flame speed is not an extremely

sensitive function of most reaction rates, correct predictions of

flame speed provide a means of indirectly validating the combined

diffusive/chemical kinetic procedures which are applied in the code.

Of course, a more complete test of the technique is the comparison, with

experimental data, of predictions of individual specie concentration

distributions through the flame. In the following discussion compari-

sons of experimental and predicted H2/02/N_ and CH,/02/N2 flame speeds

and some CO/Oj/N^/H^O flame specie concentrations are presented.

Figures 1 and 2 present the results of a hydrogen rich (50

percent H2> 10.5 percent 0?, 39.5 percent N2) premixed laminar flat

flame calculation. This flame was studied by Wilde (Reference 6) and

provides a convenient point of comparison between flame calculational
I- 231
-------
§
•u O

e c
o o
•H -H
W 4J
3 O
A C
•H 3
a)
S-i

4J
td
M
01
CO
<0
H
(U
»J
1-232
-------
o

4J
O

Figure 2. Specie concentrations through a K2/°2
fiat flame as a function of normalized
distance.
1-233
-------
techniques. The reactants are initially at 298°K and the combustion

process is assumed to take place without any heat loss to the environ-

ment. The calculational grid spans the zone of significant reaction

(i.e., flame front) with the last grid point located in a zone of

negligible diffusion. To determine the laminar flame speed a constraint

condition on ()„ concentration (ex.. - 0.002) is applied at the 10th

point of the 21 specified grid points. Nine species, H, H,,, H^O, N«, 0

HO-, H_0?, 0, OH), and 13 kinetically controlled reactions are consi-

dered in the calculation. Reaction rate constants were obtained from

Wilde (Reference 6) and are given along with their associated reactions

in Table 1. Of the 15 reactions specified in Reference 6, one has

been deleted* and two combined to yield the 13 reactions used in the

present computation. Diffusion parameters 1/F., for the species H_,

02, H, OH, 0, H02, H202, H20 and N2 are, respectively, 3.0, 1.0, 5.0,

1.5, 1.5, 1.0, 0.8, 1.2, and 1.0. These values were specified by Wilde

(Reference 6) and are consistent with the bifurcation approximation

values given in Reference 8.

Figure 1 shows the characteristic temperature distribution

through the laminar flat flame. From this figure and Figure 2 it can

be seen that the length of the rapid change region, i.e., flame front,
t.
A duplicate reaction cast in the reverse manner

Present technique can include individual third bodies and their ef-
ficiencies for a general reaction.
1-234
-------
r-
r-
to >>
3 -0
-O O
T3 J-
C PO
o
rv»
X.
u
CO
*
tr>
fn
Pn
8
CJ
to
M
»^
\^,
M r,l
+J UJ *"
5 3

0.
X
QJ
1
-^c- —• t i r\i
o
O
c- •**
Q H
v£>
^as
w S
z »s g
§3^
§ss
I
Q. e
X O
1- O
0. E
Ol/IOCOh-OOO
04
O P
M >H
H £
W
Z 04
M O
w
j
PQ
i O z:

O
I
a
1 X X O f\|
x o a o x rr»
C r\i o COO
x: r fv T. rvi C3 'u x x nj »\i ru
rr z~ 5; s:

(V fV
c ru fv c: c rv
Mfxify rvirni^ orufxio
XOIXXOCT xxxx

rvi
oxoxaxxxirrojr
1-235
-------
is confined to roughly one-fourth of the specified total grid length.

This result substantiates the application of the no diffusion boundary

condition at the last grid point. Besides the species concentrations

illustrated in Figure 2, detailed information on the contributions of

kinetic reactions to the individual species mass balances are output

for each grid point. Examining this output it appears that, for the

reaction mechanism specified:

1. In the early part of the flame (T < 1000°K) significant heat

release occurs through the following chain reaction
H + 02 + M •* H02 + M
H + H02 •* 20H
OH +
+ H
where the initial H diffued from the hot flame zone. Also,

H_0? plays an insignificant role compared to the above H0_

mechanism. These observations are in line with other investi-

gators (References 1 and 5).

2. In the latter part of the flame (T > 1000°K) water is produced

via the branching mechanism
OH +
H20 + H
H + 0 -*• OH + 0
0 + H2 -»• OH + H
1-236
-------
followed primarily by the recombination
H + OH + M
+ M
This observation is also consistent with other investigators

(Reference 5) .

These brief observations are an example of the value of the PROF

code where, for a given reaction mechanism, an in-depth, study of the

detailed chemical events can be carried out to elucidate the important

reaction paths in the mechanism. A comparable experimental program

would require the measurement of all major and minor chemical species

through the flame. This latter approach is rarely achieved and then

only at great expense.

It should be mentioned that, for the H_ flame, each sweep of

the 21 grid points, including the corrector step, required 6 seconds of

computer time on a UNIVAC 1108. If a good first guess is utilized, un-

der five sweeps are required to achieve a fully converged solution,

resulting in a total computer run time of about 30 seconds.

The calculated flame speed for this flame was 268 cm/sec which is

close to the 241 cm/ sec predicted by Wilde (Reference 6) . As indicated

in his paper, Wilde, besides making several simplifying assumptions,

roughly equivalent to those made in the present analysis, also assumes

a linear enthalpy variation with temperature. This difference combined
I- 237
-------
with the extraction of a duplicate reaction from his reaction set could

easily account for the difference in flame speed noted.

The reaction and species sets for this problem were applied in

several calculations to determine flame speed as a function of percent

hydrogen and preheat in the unburnt gas. The predictions of 1 atm zero

preheat (T. * 298°K) flame speeds as a function of percent hydrogen are

compared in Figure 3 with experimental results from several sources

(References 12, 13, 14, 15, 16, 17). Though scatter in the data is

wide, the trends of the predicted flame speeds are consistent with

those found experimentally. On an absolute level comparison, the

predictions fall intermediate to some recent (Reference 13, 17) and

past (Reference 15) data. Most of the data shown suffer from inaccura-

cies due to heat and reactive specie losses to the burner, curvature of

the flame surface, smallness of the flame being measured, etc. Gunther

and Janisch (Reference 13) suggest that their data might be 10 percent

high due to the smallness of the flame measured. Even the recent "dou-

ble kernel" flame data of Andrews and Bradley (Reference 17) is not

completely satisfactory since only 8 of 30 H2 flame experiments were

usable. As indicated by them the data is particularly suspect near the

maximum velocity point where they measured a velocity which falls at

the high end of their recent compilation of flame speed data (Reference

16). Recognizing the limitations of the experimental data, the present

inherent inadequacies of the code (e.g., mean Lewis number of unity,
1-238
-------
K-l

M
•H
cfl

CM
CD •
4-) 09
C «j
00
I
0)
13 C
C 0)
Rt O
M
a. c
o
<4-l -H
O JJ
09
13 -H
O M
<0 CO
•H >
K
efl *O
a
-------
thermal diffusion neglected) and the limitations of the kinetic models

and associated rates input, the agreement between predictions and ex-

perimental data is certainly acceptable.

For this problem and all those subsequently discussed, it should

be noted that no attempt was made to adjust the reaction rate coeffi-

cients or diffusion data in order to bring the calculations into line

with the data. The calculations were made to simply demonstrate the

capability of the code in calculating complex flame combustion and

pollutant formation chemistry.

Several years ago Padley and Sugden (Reference 18) noted that

flame speed correlated quite precisely as a linear function of the max-

imum H concentration found in H /02/N2 flames. This correlation of

flame speed with radical concentration was first suggested by Tanford

and Pease (Reference 19) who on much simplified theoretical grounds

argued that the flame speed should be proportional to the square root

of the radical concentration. Whatever the true form of the flame

speed variation with maximum H concentration, it was felt that the com-

parison of PROF predicted flame speed as a function of maximum H con-

centration with the data of Reference 18 would provide an interesting

further test of codes capability in calculating premixed flame proper-

ties.
1-240
-------
Figure 4 shows the good correspondence achieved between the meas-

ured values and PROF predictions as well as the linear variation* of

flame speeds with maximum H atom concentration. The various predicted

maximum H atom concentrations were obtained either by adding more H

to the initial gas than stoichiometrically needed (equivalence ratios

from 2.0 to 3.0) or by preheating the reactants (T initial varied from

298°K to 735°K) for an initial composition equivalence ratio of 2.38.

Having established some confidence in the combustion mechanism

of H./O^/N- flames through favorable comparison of predicted flame

speed and maximum H atoms concentration with data, NO chemistry was

added to the reaction set to determine pollutant formation in flames.

Previously, predictions of pollutant formation could only be carried

out on post-flame gases where diffusion effects are negligible and

available plug flow reactor codes are applicable. These calculations

could provide estimates of rates of production of NO in the post-flame

gases. However, in many cases the absolute level of NO predicted fell

short of the measured value. This difference, which remained constant

all along the flame axis, was termed "prompt" NO by Fenimore (Reference

21). Various mechanisms for the rapid formation of NO in the flame
Only stoichiometric or fuel rich flame velocities vary linearly with
H atom concentration.
I- 241
-------
sd-

ll)

00
•r(
U*
(oas/uo) paads
I- 242
-------
zone have been postulated (e.g., see Reference 21). A likely mechanism

has been the presence of superequllibrium 0 atom concentrations in the

flame zone which through the reaction
N_ + 0 -»• NO + N
results in a rapid formation of NO. Homer and Sutton (Reference 22)

attempted to establish a proof for this conjecture through measurement

and calculations of NO concentration through H-/02/N7 flames. They

employed a plug flow reactor calculational technique to the post-flame

gases. The reaction scheme made use of the burnt gas equilibrium

composition plus an assumed 0 atom concentration.* It was hoped that

the specification of 0 atom concentration would model the effect of

superequilibrium 0 atom concentration on NO formation in the flame zone.

The results achieved indicated that the level of NO in H7/0 /H flames

could be accounted for utilizing the 0 atom overshoot mechanism. How-

ever, the 0 atom specification was arbitrary and the growth rates of

NO did not correspond to data very well. Exercising the PROF code with

N0x formation chemistry removes the arbitrariness of the above approach

and allows the simultaneous prediction of all species concentrations

through the flame zone as well as the post-flame gases. Results of a
The other radical concentrations were determined through exchange
reactions.
1-243
-------
PROF calculation which utilized the same initial conditions as Homer

and Buttons flame 1 are presented in Figure 5 (H2/02/N2 initial vol-

ume ratios were 2/1.4/4.6, at 1 atm and 298°K. Reactions and rates

applied are given in Table 2). Correspondence with the data is reason

able with the rapid production of NO (i.e., "prompt" NO) clearly shown

followed by a more gradual rise indicative of the post-flame NO pro-

duction. The rapid rise in NO formation follows the peaking of the 0

atom concentrations and from the detailed PROF output it is evident

that the primary production of flame region NO is through the presence

of superequilibrium 0 atom concentration and the reaction
+ 0
NO + N
As a further test of the code's capability in predicting complex

flame chemistry, an atmospheric pressure stoichiometric methane/air

flame calculation was carried out. Fifteen species (CH,, 0^, CO,,, H20,

H2, CO, CH3, CH20, CHO, H02, H, 0, HO, CH2> and N2> were considered

along with 30 reactions which model approximately the processes of

methane decomposition and partial oxidation, methyl conversion to for-

maldehyde, formation and depletion of formyl and finally carbon monoxide,

hydrogen, oxygen combustion. The reactions and their associated rates,

extracted from Reference 23, are given in Table 3. Results of the

calculation are given in Figures 6 and 7 where some major specie

concentrations are plotted as a function of distance through the flame.

The calculated flame speed was 48.7 cm/sec which, considering the
I- 244
-------
03 0)
w B
B n]

-------
V4
-H
n)

>
cj

f
o
(0
c
o
•H

-------
o
o
Ol
I/I
-H

IB
00

o
M
(0
c
o
•H
C
0)
O

o •
o o
•H
U H
0)
D.-G-
bfl
•H
[03]
'[203]
1-249
-------
uncertainties of the code input data, compares favorably with the value

of 42 cm/sec found in Reference 18.

In addition to this calculation, a one-tenth atmosphere stolchio-

metric methane/air flame calculation was carried out. Nineteen species,

including those related to NO formation (CH,, 02, CO-, H20, H-, CO,

CH3, CH20, CHO, H02, H, 0, HO, CH2, NZ, N02> N20, N, NO), were consid-

ered along with 42 reactions, extracted from Reference 23. The reac-

tions and their associated rates are given in Table 4. The calculated

flame speed was 62.6 cm/sec which, considering the scatter of experi-

mental data (Reference 24), is a reasonable value. Predicted results

from these calculations are presented in Figures 8 and 9. Before

proceeding to discuss these results it is necessary to reiterate that

the purpose of the calculations was to assess the numerical capabili-

ties of the PROF code. Accordingly, no rigorous attempt has been made

to select a definitive set of reaction mechanisms, reaction rates or

diffusion coefficients. Indeed, there are a number of reactions not

included in the set presented in Table 4, which the authors view as

essential to describing NO kinetics. In spite of the above disclaimers,
X

this particular PROF calculation predicts several phenomena of signifi-

cant interest. Shown in Figure 9 are axial profiles of the hydroxyl,

oxygen hydrogen, and hydroperoxyl radicals in addition to the NO pro-

file. What is most significant about these profiles are the relative

axial positions of appearance of the various radicals. That signifi-

cant quantities of HO were predicted prior to the appearance of the
1-250
-------
TABLE 4. KINETIC REACTIONS AND ARRHENIUS COEFFICIENTS FOR
METHANE/AIR FLAMES INCLUDING NOX FORMATION KINETICS
(Reference 23)
CM4
CH
-------
J 31-JW
o «±- eg
^ o o o o

/
1

•i
•-
o
0.1
0
o
0
od
o

O
i-l
a)
M
0)

t •
4) O
TJ
C II
CO
-e-
cn
4-1 CO
rt
J-l I-l
•U •
C O
0
-H
1-252
-------
I
1-1
•rl

-------
hydrogen, oxygen or hydroxyl radicals was an unexpected result. This

general behavior is, however, in agreement with the data of Peeters

and Mahnen (Reference 25). Another phenomena predicted by PROF and

in agreement with the data from Reference 25 is that the appearance

of substantial quantities of 0, H, and OH occur almost simultaneously

and that the position of this appearance coincides closely with the

point of maximum H0_ concentration. It should also be noted that

these predictions qualitatively agree with the data of Reference 26.

Examination of the kinetic production rates yields no clear-cut expla-

nation to the above phenomena. It can be said that the primary path

for production of H02 is through the third body reaction
H + 02 + M = H02 + M
Hydrogen atoms are, of course, available through upstream diffusion

from the flame front, and from the minute concentrations of H prior to

the 0.8 cm location it would appear that the H atoms are rapidly con-

sumed by 0_ as they are stripped from the CH^ molecule. This would

tend to indicate the possibility of describing CH, pyrolysis by some

type of quasi-global step. This calculation required approximately

47 seconds of UNIVAC 1110 computer time to sweep 21 grid points includ-

ing the corrector step. A total of four sweeps was required to achieve

a fully converged solution resulting in a total computer run time of

two and one half minutes.

As mentioned previously, this calculation was performed simply to

demonstrate the code's capabilities. The comments made about this
1-254
-------
particular prediction are, therefore, tentative at best. It should be

obvious that the possibilities of the code are many and that phenomena

such as those described above can only be predicted by a code which

includes diffusional effects in the computations.

To this point in the discussion comparisons of detailed specie con-

centration predictions with experimental data have been limited to NO

concentrations in H./O^/N,, flames. To further substantiate the capa-

bility of the code in predicting detailed combustion and pollution

chemistry events some predictions of CO/02/N2/H20 flame specie concen-

trations are compared with the data of Merryman and Levy (Reference 27).

The specie concentrations presented in Figures 10 and 11 correspond

to run number 1 of Reference 27. For this case the initial CO/02/N2/H20

volume ratios were 0.190/0.155/0.638/0.017, respectively. The pressure

was 1 atmosphere and the reactants entered the burner at 298°K. For

this calculation, 14 species (CO, H20, 02, N2, C02> H, 0, HO, H2, H02,

NO,, NO, N, NO) and 30 reactions were specified. Table 5 lists the

reactions and their associated rates which were extracted from Refer-

ence 23. Predictions of CO, CO , 02 and NO species concentrations

through the flame are favorably compared with data in Figures 10 and

11. The good agreement between the predicted and measured rates of

change of CO, CO- and 0? indicates that, at the very least, gross com-

bustion characteristics are adequately modeled. As in the H_/02/N2 flame,

the detailed output indicates that in the early part of the flame H02
1-255
-------

Sd
CN
CO S3
4J — .
C C
0) O
M Q
(U
O* ctf

4) 4-

•O 3

§£

13 -M

-------
o
c
o •
3 0>
cfl iH
4J M-l

0) O
a CM
•5 a:
j-i —

-------
I
§ »
? S?g
III
-o o ir o ec «c in -<
•^« f» < ••« ^ o . «-• IP * in «xi (v *-•
•-• *« »»

QOOOOOOOOOOOOOOOOOOOOOOOOOOOO o
oocoooezooocoooooooooooooooooc a

i e

oooooe-oooocooooeercooooeoc-cooc o
OCOGOOOOGOOOOOOCOClTtOOaOCOGOOO O

^ O O
Ofuoo Aim
-------
plays a dominant role leading to the formation of CCL through the

chain reaction
H + 02 + M •* H02 + M
H + H02 -»• 20H
CO + OH -»• C02 + H
In the latter part of the flame the 0 atom concentration is large

enough to produce CO, through the reaction
CO + 0 + M •* C02 + M
as well as through the reaction
CO + OH -f C02 + M
Figure 11 shows the good agreement between the predicted and measured

NO concentrations as well as the prediction of superequilibrium 0 atom

concentrations in the flame zone. It is primarily the large concentra-

tion of 0 atoms which drives the production of NO in the flame zone

through the reaction

N2 + 0 -> NO 4- N

Merryman and Levy were interested in NO production through fuel

nitrogen oxidation as well as through thermal fixation of air nitrogen.

To model the effect of fuel nitrogen on NO production they injected
1-259
-------
various additives into CO/0-/N2/H20 flames. They postulated that the

fuel nitrogen compounds would yield NH and CN type species which sub-

sequently oxidize to NO in the early part of the flame. This NO will

then be scavenged by H0? radicals producing "early NO " which, in the

latter part of the flame, is converted back into NO via reactions with

excess 0 atoms. The mechanism for this process can be written
NH (or CN) + 02 -*• NO + OH (or CO)
NO + H02 •* N02 + OH
N02 + 0 + NO +
To test steps 2 and 3 of this mechanism Merryman and Levy in ex-

perimental run three injected NO into a CO/02/N2/H_0 flame. The

changes in initial conditions from those of run 1 were CO/0?/N?/H?0

volume ratios of 0.187/0.154/0.642/0.017 and an initial NO concentra-

tion of 34 ppm. Utilizing the above initial conditions and the species

and reactions of run 1, flame species concentrations were predicted

and are compared with experimental results in Figure 12. Considering

the experimental uncertainties, the inherent inadequacies of the code

and the limitations of the kinetic model the agreement is adequate.

The detailed output indicates that the proposed kinetic mechanism which

involves NO scavenging by H02 and oxidation of the resulting N02 via

superequilibrium 0 concentration is indeed operable and yields the

characteristic concentration profile shapes shown in Figure 12.
1-260
-------
OJ
u
fl
r-t
M-l

CMO
O o
53 SB
C c-
0) O

•So
n u
0)
P< cd

(U 4:

•a 3
c o
CO M
cu
i-i in
u c
•H o
(U
M-l C
O
-------
Before leaving these computations, it should be pointed out that

they will be repeated in the near future in a series of computer runs

which concentrate more attention on the kinetic mechanisms, reaction

rates, and diffusional coefficients. It is fully expected that these

computations will shed light on the prompt NO and early NO- questions

as well as to help unravel the complex reaction mechanisms of the

hydrocarbon ignition and burnout.

In summary, several preliminary calculations have been carried out

to demonstrate the speed, reliability and accuracy of the PROF code in

predicting complex combustion and pollutant formation chemistry in

flames. In general, the good agreement of predicted quantities and

those measured has been encouraging. Further improvement in agreement

might be easily accomplished through code modifications to include (i)

thermal diffusion (ii) heat losses to the environment (iii) heat and

reactive specie losses to the flameholder.

The brief experiences of this program have brought out the power

and usefulness of the PROF calculational technique in elucidating the

important steps in a proposed flame reaction mechanism. Also, it has

been demonstrated that PROF predictions can be an important complement

to experimental complex chemistry flame studies.
1-262
-------
REFERENCES
1. Dixon-Lewis, G., "Flame Structure and Flame Reaction Kinetics, V.
Investigation of Reaction Mechanism in a Rich Hydrogen 4- Nitrogen +
Oxygen Flame by Solution of Conservation Equations," Proc. Roy Soc.
Lond., A, Volume 317, 1970, pp. 235-263.

2. Bledjian, L., "Computation of Time-Dependent Laminar Flame Structure,"
Combustion and Flame, Volume 20, 1973, pp. 5-17.

3. Spalding, D. B., et al., "A Calculation Procedure for the Prediction
of Laminar Flame Speeds," Combustion and Flame, Volume 17, 1971,
pp. 55-64.

4. Stephenson, P. L. and Taylor, R. G., "Laminar Flame Propagation in
Hydrogen, Oxygen, Nitrogen Mixtures," Combustion and Flame, Volume
20, 1973, pp. 231-244.

5. Eberius, K. H., et al., "Experimental and Mathematical Study of a
Hydrogen-Oxygen Flame," Thirteenth Symposium (International) on Com-
bustion, The Combustion Institute, 1971, pp. 713-721.

6. Wilde, K. A., "Boundary-Value Solutions of the One-Dimensional Lam-
inar Flame Propagation Equations," Combustion and Flame, Volume 18,
1972, pp. 43-52.

7. Kendall, R. M. and Bartlett, E. P., "Nonsimilar Solution of the
Multicomponent Laminar Boundary Layer by an Integral Matrix Method,"
AIAA Journal, Volume 6, No. 6, June 1968.

8. Bartlett, E. P,, Kendall, R. M., and Rindal, R. A., "A Unified Ap-
proximation for Mixture Transport Properties for Multicomponent
Boundary Layer Applications," Aerotherm Corporation Final Report
66-7, Part IV (also NASA CR-1063), March 14, 1967.

9. Hirschfelder, J. 0., Curtiss, C- F., and Bird, R. B., Molecular
Theory of Gases and Liquids, Second Printing, Corrected, with notes
added, John Wiley and Sons, Inc., New York, 1964.

10. Kendall, R. M., "An Analysis of the Coupled Chemically Reacting
Boundary Layer and Charring Ablator, Part V, A General Approach to
the Thermochemical Solution of Mixed Equilibrium-Nonequilibrium,
Homogeneous or Heterogeneous Systems," NASA CR-1064, June 1968.
1-263
-------
11. Kendall, R. M. and Kelly, J. T., "Premixed One-Dimensional Flame
Code — Its Formulation Manipulation and Evaluation," Aerotherm
Corporation Final Report TR-75-158, July 1975.

12. Scholte, T. G. and Vaags, P. B., "Combustion and Flame," Volume 3,
1959, p. 495.

13. Gunther, R. and Janisch, G., "Measurements of Burning Velocity in a
Flat Flame Front," Combustion and Flame, Volume 19, 1972, pp. 49-53.

14. Edmondson, H. and Heap, M. P., "The Burning Velocity of Hydrogen-
Air Flame," Combustion and Flame, Volume 16, 1971, pp. 161-165.

15. "Flame Velocities of Various Gas-Air Mixtures," Bureau of Standards
J. Research, 17, 7-43, 1936.

16. Andrews, G. E. and Bradley, D., "Determination of Burning Velocities:
A Critical Review," Combustion and Flame, Volume 18, 1972, pp. 133-
153.

17. Andrews, G. E. and Bradley, D., "Determination of Burning Velocity
by Double Ignition in a Closed Vessel," Combustion and Flame, Volume
20, 1973, pp. 77-89.

18. Padley, P. J. and Sugden, T. M., "Chemiluminescence and Radical
Recombinations in Hydrogen Flames," Seventh Symposium (International)
on Combustion, Butterworths, London, 1959, p. 235.

19. Tanford, C. and Pease, R. N., Journal of Chemical Physics, 15, 1947,
p. 861.

20. Brown, N. J., Fristrom, R. M., and Sawyer, R. F., "A Simple Premixed
Flame Model Including Application of H2 + Air Flames," Combustion
and Flame, Volume 23, Number 2, October 1974.

21. Fenimore, C. P., "Formation of Nitric Oxide in Premixed Hydrocarbon
Flames," Thirteenth Symposium (International) on Combustion, The
Combustion Institute, 1971, p. 373.

22. Homer, J. B. and Sutton, M. M. , "Nitric Oxide Formation and Radical
Overshoot in Premixed Hydrogen Flames," Combustion and Flame, Volume
20, 1973, pp. 71-76.

23. Waldman, C. H., Wilson, R. P., Jr., and Maloney, K. L., "Kinetic
Mechanism of Methane/Air Combustion with Pollutant Formation," EPA-
650/2-74-045, June 1974.
1-264
-------
24. Andrews, C. E. and Bradley, D., "The Burning Velocity of Methane-
Air Mixtures," Combustion and Flame, Volume 19, 1972, pp. 275-288.

25. Peters, J. and Mahnen, G., "Reaction Mechanisms and Rate Constants
of Elementary Steps in Methane-Oxygen Flames," Fourteenth Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh,
1973, pp. 133-U6.

26. Lazzaru, C. P., Biordi, J. C., and Papp, J. F., "Concentration Pro-
files for Radical Species in a Methane-Oxygen-Argon Flame," Combus-
tion and Flame, Volume 21, 1973, pp. 371-382.

27. Merryman, E. L. and Levy, A., "NOX Formation in CO Flames," in prepa-
ration, 1975.
1-265
-------
2:10 p.m.
Prediction of Premixed Laminar Flat
Flame Kinetics Including the
Effects of Diffusion
Dr. Robert M. Kendall, Aerotherm Acurex
Q: Do you have any feeling, Bob, for how important
your diffusion correction is? If you were to make
the same run, setting all your diffusion coefficients
equal to 0, would your predicted curves change very
much?

A: Well, actually, in the problems that we were doing,
we were usually looking for a flame speed, and there's
no way you can solve these problems without diffusion.
If diffusion did not exist, the reactions would never
get started. So, if there was no diffusion, the
thing would never light. Still, there would be no
transfer of energy upstream. You have to have the
diffusion in there. You could, perhaps, go in and
take a measured temperature, and impose that, and
leave out the energy equation. And, in that case,
you would predict things I would think would be
radically different, because the hydrogen atom dif-
fusion upstream is a very important ignition mechanism,
Without that, I think quite a few things would be dif-
ferent .
1-266
-------
June 1975
ESTIMATION OF RATE CONSTANTS AS A FUNCTION
OF TEMPERATURE FOR THE REACTIONS X + YZ r* XY + Z
WHERE X, Y, AND Z ARE ATOMS OF THE ELEMENTS
CARBON, HYDROGEN, NITROGEN, OXYGEN, AND SULFUR

Running Title: ESTIMATION OF RATE CONSTANTS
S. W. Benson, Robert Shaw, and R. W. Woolfolk
Physical Sciences Division
Stanford Research Institute
Menlo Park, California 94025
1-267
-------
ABSTRACT
Estimated rate constants for some reactions of the type
X + YZ -» XY + Z where X, vf and Z are atoms of the elements carbon,
hydrogen, nitrogen, oxygen, and sulfur have been compared with rate
constants measured by others. This work completes earlier comparisons
for the above reaction where X, Y, and Z were atoms of the elements
hydrogen, nitrogen, and oxygen. In all cases except a single experi-
mental value for the reaction H + HS -» H2 + S, there is good agreement
between measured and estimated rate constants. The estimates are empiri-
cal, but are based on transition state theory. The technique gives the
following expression for the rate constant of the above reactions in the
exothermic direction, k/(cm3 mol'1 s'1) = 1011-8 T°-5 exp[-(AH°*0 + 1)/RT],
where AH300 has the value of -1 kcal/mol for reactions that have a stable
XYZ intermediate, or 7 kcal/mol for reactions that are concerted or have
an unstable XYZ intermediate.
I- 268
-------
INTRODUCTION

We have previously shown [l] that all the critically evaluated

rate constants for the reactions
X + YZ & XY + Z
(1)
where X, Y, and Z are atoms of the elements hydrogen, nitrogen, and

oxygen can be fitted in the temperature range 200 to 4000 K by the

following equation derived from transition state theory,
k/(cm3 molds'1) = AT exp(-C/RT)
(2)
where log10[A/(cro3 mol^g-1) ] = 16.6 + [AS5*0/(2.3R>] , B = 0.5,

C/(kcal/mol) = AH°*0 + 1, R = gas constant, T = absolute temperature,

Assoo = entropy of activation at 300 K, and AH^0 = enthalpy of

activation at 300 K.
In the exothermic direction, AS°*0 has a value of -22 cal/(mol K) ,

and AH^oo nas a value of either -1 kcal/mol for reactions that have a

transition state preceded by an intermediate that is stable with

respect to the reactants (see Figure 1) or 7 kcal/mol for reactions

that are concerted or that have a transition state preceded by an inter-

mediate that is unstable with respect to the reactants (see Figure 2).
I-269
-------
We have recently [3] completed estimates of the heat of formation
of the intermediate XYZ where X, Y, and Z are atoms of the elements
carbon, hydrogen, nitrogen, oxygen, and sulfur, which have enabled us
to assign values of AHg*0 to all reactions like Reaction (1) where X,
Y, and Z are C, H, N, 0 and S. The assignments are in Appendices A and B.
The purpose of the present paper is to compare the estimated rate
constants with experimental data for reactions involving carbon and
sulfur atoms that were not covered previously, and with more recent data
for the reaction of hydrogen atoms with nitric oxide and oxygen atoms
with nitric oxide. The reactions of interest and the estimated rate
parameters are given in Table 1.
I- 270
-------
RESULTS

For each reaction in Table 1, the experimental and estimated

results have been compared. Boden and Thrush [4} have measured the

rate of Reaction (3)
CN
OC + N
(3)
between 570 and 687 K. The only rate constant reported by Boden and

Thrush was 1013-° cm3 rool""1s~1 at 687 K. Albens, Schmatjko, Wagner,

and Wolfrum [5] recently reported a rate constant of 1013'1 cm3 mol~1s~1

independent of temperature at 298, 400, and 500 K. The experimental

and estimated values are compared in Figure 3.

Figure 3

Hancock and Smith [61 have studied reaction (4).
O + CS
OC + S
(4)
They measured a rate constant of 1012'8 cm3 mol"1s~1 at 298 K and

reported a previous measurement of 1014*1 cm3 mol~1s~1 at 1100 K by

Homann, Krome, and Wagner [7~J. The experimental results are compared

with estimated values in Figure 3,

Flower, Hanson, and Kruger [8] recently completed a study of

Reaction (5) in a shock tube between 2400 and 4500 K.
H + ON
HO + N
(5)
1-271
-------
The experimental and estimated results are compared in Figure 4.

Figure 4

Cupitt and and Glass [9] have studied the reaction of atomic oxygen

with hydrogen sulfide by esr spectroscopy. The variation in the con-

centration of free radicals was measured, and the results were inter-

preted in terms of a mechanism consisting of seven elementary reactions.

One of the the reactions was
0 + SH
OS + H
(6)
The experimental results were consistent with a rate constant of 1014'5

cm3 mol~1s~1 at 295 K for Reaction (6). This rate constant is compared

with the estimated one in Figure 3.

Mihelcic and Schindler [10"] have studied the sulfur analog of

Reaction (6)
S + SH
H
(7)
by the same method as Cupitt and Glass. The rate constant of Reaction

(7) at 300 K was found to be 1013*4 cm3 mol~1s"1. This rate constant

is compared with the estimated one in Figure 5.

Figure 5

Using the same system as was used to study Reaction (6), Cupitt and

Glass measured the rate constant at 295 K for Reaction (8)
H + HS
H2 + S
(8)
to be 1013*9 cm3 mol~1s~1. This rate constant is compared with the
1-272
-------
estimated one in Figure 6.
Figure 6
Hanson, Flower, and Kruger [ll] recently investigated the

decomposition of nitric oxide in a shock tube in the range of 2500 to

4100 K. From their rate measurements, they derived the rate constants

for Reaction (9)
0 + ON
N
(9)
shown in Figure 7 with rate constants measured by other workers [12 to

15"j and with the estimated rate constants.
Figure 7
Reaction (10)
SO + 0
(10)
has been studied by Cupitt and Glass [9], who found k10 = 1011*
cm'
molds'1 at 295 K, and by Fair and Thrush [16], who found k10 = 1012'

cm3 raol 1s~1 at 298 K. The experimentally determined rate constants

are compared with the estimated ones in Figure 5.
I- 273
-------
DISCUSSION

The agreement between measured and estimated rate constants for

H, N, O reactions is excellent, as shown in Figures 4 and 7. For

reactions involving carbon and sulfur atoms, there is enough experi-

mental evidence to suggest that the method of estimation is at least

a good first approximation.

Consider reactions of the type
O + YZ
OY + Z
(11)
shown in Figure 3. At first sight it appears that the worst agreement

between measured and estimated values is for Reaction (6)
0 + SH
— OS + H
(6)
However, if all the difference between measured and estimated rate

constants is attributed to activation energy, as is likely, then the

difference between observed and calculated rate constants for Reaction

(6) corresponds to less than 2 kcal/mol, whereas the difference is

3 kcal/mol for Reaction (4)
O + CS
CO + S
(4)
These differences are not large for combustion reactions,
I- 274
-------
Similarly, for reactions of the type
S + YZ
SY + Z
(12)
where Y is a sulfur or an oxygen atom, the difference between estimated

and observed rate constants at 300 K corresponds to a difference of

about 2 kcal/mol, as shown in Figure 5. However, when Y is a hydrogen

atom, as in Reaction (-8)
S + H2 , - SH + H
H + HS H2 + S
(-8)

(8)
there is a big discrepancy (about 7 powers of 10) between the rate

constants at 300 K. This discrepancy in the rate constants for Reaction

(8) corresponds to about 10 kcal/mol in the activation energies. There

is only one experimental value for Reaction (8), but the other two rate

constants in the same experiment, namely, for Reactions (6) and (10)
O + SH
S + 0,
OS + H
SO + 0
(6)

(10)
are in good agreement with other experimental work and with the estimates.

On the other hand, this one value for Reaction (8) is the only example,

out of all available experimental data for the CHNOS system, where the

estimated and experimental data are not in agreenent. There is no room for

adjusting the estimated value, because the thermochemistry is too well

known for Reaction (8) to be anything other than exothermic. Furthermore,

since the atom being transferred is a hydrogen atom, there is no need to

estimate the heat of formation of the transition state. Clearly, more
1-275
-------
work is required on this reaction.

Acknowledgements

We are pleased to acknowledge the support and encouragement of

W. Steve Lanier and the very helpful discussions with R. M. Kendall

and R. K. Hanson. We thank W. L. Flower, R. K. Hanson, and C. H. Kruger

for permission to quote their unpublished results, M. K. Stein and

J. J. Hayes for typing and C. J. Reeds for editing the manuscript. The

research was made possible by a grant from the U.S. Environmental

Protection Agency, with financial assistance from Stanford Research

Institute.
1-276
-------
REFERENCES

Benson, S. W., Golden, D. M., Lawrence, R. W., Shaw, R., and
Woolfolk, R, W., Estimation of rate constants as a function of
temperature for reactions X + YZ^XY + Z, X + Y + MssXY+M, and
X + YZ + M ^ XYZ + M, where X, Y, and Z are atoms, H, N, and O,
presented at the Symposium on Chemical Kinetics Data for the Lower
and Upper Atmosphere, September 16-18, 1974, Airlie House, Warrenton,
Virginia, U.S.A., to be published in a special issue of the
International Journal of Chemical Kinetics.

Baulch, D. L., Drysdale, D. D., and Home, D. G., Evaluated Kinetic
Data for High Temperature Reactions, Butterworths, London, 1973.
Benson, S. W., Golden, D. M., Lawrence, R. W. , Shaw, R., and Woolfolk,
R. W. , Estimates of kinetics of combustion, especially reactions
involving NO and SO , U.S. Environmental Protection Agency Report
300-00-0000,XJune
4. Boden, J. C., and Thrush, B. A., Proc. Roy. Soc. (London) 305A,
107 (1968) .

5. Albers, E. A., Schroatjko, K. J., Wagner, H. Gg., and Wolf rum, J.,
15th Symposium (International) on Combustion, The Combustion
Institute, Pittsburgh, Pa., Paper No. 73.

6. Hancock, G., and Smith, I.W.M., Trans. Faraday Soc. 67, 2586 (1971).

7. Homann, K. H., Krone , G., and Wagner, H. Gg., Ber. Bunsenges, Phys.
Chem. 72, 998 (1968) .

8. Flower, W. L., Hanson, R. K., and Kruger, C. H., 15th Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh,
Pa. , Paper No. 78.

9. Cupitt, L. T., and Glass, G. P., Trans. Faraday Soc. 66, 3007 (1970)

10. Mihelcic, D., and Schindler, R. N., Be r . Bun s e n ge s . Phy s . Chem . 74,
1280 (1970) .
1-277
-------
REFERENCES, (Concluded)

11. Hanson, R. K,, Flower, W. L., and Krugcr, C. H., Combustion Sci.
Tech., in press.

12. Kaufman, F., and Kelso, J. R., J. Chcm. Phys. 23, 1702 (1955).

13. Kaufman, F., and Decker, L. J., 7th Symposium (International) on
Combustion, the Combustion Institute, Pittsburgh, Pa., 1959, p. 57.

14. Wray, K. L., and Teare, J. U., J. Chetn. Phys. 36, 2582 (1962).

15. Clark, T. C., Garnett, S. H., and Kistiakowsky, G. B., J. Ghent.
Phys. 51, 2885 (1969) .

16. Fair, R. W., and Thrush, B. A., Trans. Faraday Soc. 65, 1557 (1969)
1-278
-------
w

H
O C
O Ci>
O E
,H (U
0>
£
C
CO
C »
O E
U O
4-J
Cu ra

ns (U

co
•O
C N

•a
^ c
u >*

^
(U
"M
O
a;
+ c
(U
X bfi

CD !?
a o
c
a>
be
0
S-i
0) -i-> Wl
•HOC
(Q cd
h CD

73
OJ
E in o
-H -Q
+J ^ in
w o ca
U 4-i CJ
t<
T3

r-l
a
o
•H
O.
w
1 — 1
r*
T-t
1
o
E
n
£
X.
1 I
bJG
O
r-l

tH
C

d
rH
3
O
4-1

O
0

-p
n!

r-t
O
e
H
0
O
r-i
H
I
in
r-i
o

n
e
o
x^
L_J
bo
0
rH

d
0
•H
CS

T
CO
I-l

o
o

00
IH

*
+
8
u
+
o

o
°

TJ< O> "S1 -^ CM 0) •«<
CO (C CO CO OJ CO rt
r-l »-( i-l r-l r-l

1-279
-------
§
a n
I
REACTANTS
X + YZ
TRANSITION
STATE
(X ••• Y -•• Zl*
1 kcal/mol
STABLE
INTERMEDIATE
XYZ
PRODUCTS
XY + Z
REACTION
SA-2OO8-3
FIGURE 1
HEAT CHANGES FOR REACTIONS HAVING TRANSITION STATES
PRECEDED BY AN INTERMEDIATE THAT IS STABLE WITH RESPECT
TO THE REACTANTS
In such cases AH = -1 kcal/mol.
1-280
-------
o n
I

TRANSITION
STATE
7 kcal/mol
REACTANTS
X + Y2
UNSTABLE
INTERMEDIATE
PRODUCTS
XY + Z
REACTION
S A-2008-4
FIGURE 2 HEAT CHANGES FOR REACTIONS THAT ARE CONCERTED
OR THAT HAVE A TRANSITION STATE THAT IS PRECEDED
BY AN INTERMEDIATE THAT IS UNSTABLE WITH RESPECT
TO THE REACTANTS
In such cases
= 7 kcal/mol.
1-281
-------
13.0
12.0
| 114,
10.0
2000
~I~
1000
~T~
O * CN -» OC + N, BT |4|

O + CN - OC + N, ASWW [5]

/\ O + CS - OC + S, HKW [7]

O + CS -» OC + S, HS [61

O * SH - OS + H, CG [9]

Ettimnid. Thit Work and [3J
A
T/K
600
300
1.0
2.0
3.0
4.0

SA-2O08-5
FIGURE 3 COMPARISON OF MEASURED AND ESTIMATED RATE CONSTANTS
FOR REACTIONS OF OXYGEN ATOMS
1-282
-------
10.00
7 8.00
j>
V
6.00
4.00
10,000
O
5000
[8]
Estimated This
Work and (3]

L_
T/K
3000
2000
1500
0.2
0.4
103KAT
0.6

SA-2008-6
FIGURE 4 COMPARISON OF MEASURED AND ESTIMATED RATE CONSTANTS
FOR THE REACTION H + ON -* HO + N
1-283
-------
12.0
11.0
I
1
10.0
9.0
2000
1000
T/K

500
O S + 02 - SO + O, FT [161

| S + 02 - SO + O, CG [91

A. S + SH - S2 + H. MS |101

Estimated. This Work and {31
300
A
o
1.0
20

103K/T
3.0
4.0
SA-2O08-?
FIGURE 5 COMPARISON OF MEASURED AND ESTIMATED RATE CONSTANTS
FOR REACTIONS OF SULFUR ATOMS
1-234
-------
12.0
T/K
2000 1000
500
300
11.0
o
6.0
J»
5.0
4.0
O CG [9]

Estimated This Work and [3]
1.0
2.0
103K/I
3.0
SA-2008-8
FIGURE 6 COMPARISON OF MEASURED AND ESTIMATED RATE CONSTANTS
FOR THE REACTION H + HS -* H2 + S
1-285
-------
T/K
8.00
-~ 6.00
o
jj 4.00
2.00
10,000
5000
3000
2000
1500
HFK {111
WT [14]

/\ KD [13]
KK [12]
CGK {15]
Estimated This Work and [3]
0.2
0.4
103K/T
0.6
0.8

SA-2008-9
FIGURE 7 COMPARISON OF MEASURED AND ESTIMATED RATE CONSTANTS
FOR THE REACTION 0 + ON
N
1-286
-------
Appendix A

HYDROGEN TRANSFER REACTIONS*
Formula of Triatomic Transition State XYZ
Empirical
CHN
CHO
CHS
CH2
C2H
HNO
HNS
HN2
HOS
HO2
HS2
H2N
H20
H8S
H3
Structural
CHN
OHN
SHC
HHC
CHC
OHN
SHN
NHN
OHS
OHO
SHS
HHN
HHO
HHS
HHH
Reaction

C + NH -> CH + N
0 + HC -> OH + C
S + HC -» SH + C
H + HC -> H2 + C
C + HC -4 CH + C
0 + HN -4 OH + N
S + HN -» SH + N
N + HN -» NH + N
0 + HS -* OH 4- S
0 + HO -» OH + 0
S + HS -+ SH + S
H + HN -4 H2 + N
H + HO -» H2 + O
H + HS -> H2 + 0
H + H2 -» H2 + H
Reactions X + HZ -» XH + Z where X and Z are atoms of the elements
carbon, hydrogen, nitrogen, oxygen, and sulfur. The rate constant
for every reaction is: k/(cm3 mol"1 s"1) = 1011>8T0-5 exp(-8/RT).
Reactions are in the exothermic direction.
1-287
-------
Appendix B
ATOM TRANSFER REACTIONS
a
Formula of Triatomic
Intermediate XYZ
Empirical
CHN
CHN
CHO
CHO
CHS
CHS
CH2
CNO
CNO
CNO
CNS
CNS
CNS
CN,
CN2
Structural
CNH
NCH
OCH
COH
CSH
SCH
HCH
CNO
CON
OCN
CNS
CSN
SCN
NCN
NNC
Reaction
C + NH -* NC + H
N + CH — ' NC + H
0 + CH - OC + H
C + OH - CO + H
C + SH - CS + H
S + CH -* SC + H
H + CH -> HC + H
C + NO - CN + O
C + ON - CO + N
O + CN - OC + N
C •+• NS - CN + S
C + SN -* CS + N
S + CN - SC + N
N + CN - NC + N
N + NC - N2 + C
[AH«(X) + AH°(YZ)]
155 ± 5
255 ± 1
201.6 ±0.1
180 ± 1
203 ± 1
208.3 ± 0.1
194.1 ±0.1
192.5 ± 0.5
192.5 ± 0.5
164 ± 3
234 ± 10
234 ± 10
170 ± 3
217 ± 4
217 ± 4
AH°(XYZ)
110 ± 3
32.3 ± 2°
I 7.2*
< c
110.4 ± 2
81 ± 15b
123 ± 15
b
92 ± 8
92.4 ± 1°
b
145 ± 17
b
215 ± 29
38 ± 3°
b
186 ± 25
b
250 ± 29
73 ± 3
113 ± 5°
140 ± 30°
AH?*0
-1
-1
-1
-1
-1
-1
-1
-1
7
-1
-1
7
-1
-1
-1
Reactions X + YZ -> XY + Z where Y is not a hydrogen atom and where X
and Z are atoms of the elements carbon, hydrogen, nitrogen, oxygen
and sulfur. If the heat of formation of the intermediate XYZ is less
than the sum of the heats of formation of X and YZ, then AH500 = -1
kcal/tnol in the equation for the rate constant: k/(cm3 mol"1 s-1) =
loii.8To.s exp [_(£H°*0 + D/RT], If the heat of formation of the
intermediate XYZ is greater than the sum of the heats of formation
of X and YZ, then AH°00 = 7 kcal/mol. All units are kcal/mol. Reactions
are in the exothermic direction.

Estimated value for AH°(XYZ).

"Reference 9.
Reference 4.
1-288
-------
Appendix B

ATOM TRANSFER REACTIONS (Continued)
Formula of Triatomic
Intermediate XYZ
Empirical

COS
COS
COS

C°2

co2

CS2
cs2

C2H

C2N
C2N

C20
c2o

C2S

c2s

C3
HNO

HNO

HNS

HN2

HOS
Structural

COS
OCS
CSO

COO

OCO

CSS
scs

CCH

NCC
CNC

occ
coc

sec

CSC

ccc
ONH

NOH

SNH

NSH

HHH

SOH
Reaction

C + OS - CO + S
0 + CS - OC + S
C + SO -• CS + O

C + O2 -» CO + 0

0 + CO - OC + 0

C + S2 - CS + S
S + CS - SC + S

C + CH - C2 + H

N + Cz -> NC + C
C + NC - CN + C

0 + C2 -. OC + C
C + OC - CO + C

S + CC _ SC + C

C + SC -. CS + C

C + C2 -. C2 + C
0 + NH _ ON -H H

N + OH -. NO + H

S + NH - SN + H

N + SH - NS + H

N + NH ^ N2 + H

S + OH -» SO + H
[AH°(X) + AH°(YZ)]

172.1 ± 0.8
115 ± 5
172.1 ± 0.8

170.9 ± 0.5

33.2 ± 0.6

201.7 ± 0.7
121 ± 5

312.9 ± 0.6

313 ± 2
275 ± 4

259.7 ± 0 .9
144.5 ± 1.1

266.5 ±0.9

266 ± 5

371 ± 1
150 ± 4

122 ± 1

156 ± 4

146 ± 2

203 ± 5

75.6 ± 0.3
AH° (XYZ)

b
161 ± 36
-33.1 ± 0.3
b
164 ± 19
b
135 ± 22
c
-94.05 ± 0.01
b
164 ± 19
C
28.0 ± 0.2
c
114 ± 7
b
146 ± 16
133 ± 30°
c
69 ± 15
213 ± 30
b
130 ± 15
b
248 ± 30
b
196 ± 4
23. 8b
b
77 ± 11
b
65 ± 11
b
112 ± 11
b
152 ± 16
b
11 ± 13
AH°*o

-1
-1
-1

-1

-1
-1

-1

-1
-1

-1
7

-1

-1
-1

-1

-1
1-289
-------
Appendix B

ATOM TRANSFER REACTIONS (Concluded)
Formula of Triatomic
Intermediate XYZ
Empirical
HOS
H02
HS2
H2N
«2°
H2S
NOS
NOS
NOS
NO,
N02
NS2
NS2
N20
N20
N2S
N2S
s2o
soa
so2
°3
Structural
OSH
OOH
SSH
HNH
HOH
HSH
NOS
OSN
ONS
NOO
ONO
NSS
SNS
NNO
NON
NNS
NSN
NNN
OSS
SOS
soo
oso
ooo
sss
Reaction
0 + SH -, OS + H
0 + OH - O2 + H
S •*• SH - S2 + H
H + NH _> HN + H
H + OH _ HO + H
H + SH _ HS + H
N + OS _ NO + S
O + SN _ OS + N
O + NS _ ON + S
N + O2 _ NO + O
O + NO _, ON + O
N + S2 -* NS + S
S + NS - SN + S
N + NO _ N2 + O
N + ON - NO + N
N + NS _ N2 + S
N + SN -. NS + N
N + N2 - N2 + N
O + S2 -, OS + S
S + OS -. SO + S
s + oa _ so + o
O + SO _ OS + 0
0 + 02 _ 02 +0
S + S2 - S2 +0
[AH°(X) + AH°(YZ)]
93 ±1
69.9 ± 0.3
99.6 ± 1.2
142 ± 4
61.4 ± 0.3
85.4 ± 1.2
114 ± 1
123 ± 10
123 ± 10
113 i 1
81.1 ± 0.4
144 ± 1
129 ± 10
135 ± 1
135 ± 1
176 ± 11
176 ± 11
113 ± 1
90.4 ± Q.2
67.5 ± 0.3
66.29± 0.01
60.8 ± 0.3
59.55± 0-02
97.1 ± 0.2
AH»(XYZ)
23 ± 10
c
5 ± 2
33 ± 5
46 ± 2°
-57.8 ± 0.001°
-4.9 ± 0.2°
149 ± 21
166 ± 39
b
49 ± 10
136 ± 17b
7.9 ± 0.2°
164 ± 32
90 ± 20
19.6 ± 0.1°
209 ± 27
b
61 ± 10
244 ± 31
b, c.e
171 ± 13 ' '
-14 ± 8°
b
36.5 ± 1.0
b
35.3 ± 0.7
-70.95 ± 0.05°
34.1 ± 0.4
b
65 ± 1
A«§*0
-1
-1
-1
-1
-1
-1
7
7
-1
7
-1
7
-1
-1
7
-1
7
7
-1
-1
-1
-1
-1
-1
BThe AH°(XYZ) includes an additional 60 ± 6 kcal/mol for spin forbidden
reaction.
1-290
-------
PRODUCTION OF OXIDES OF NITROGEN

IN INTERACTING FLAMES
Christopher England
Jet Propulsion Laboratory
Pasadena, California 91106
Paper Presented at the
Symposium on Stationary Source Combustion
Fairmont Colony Square Hotel
Atlanta, Georgia
September 24-26, 1975
Sponsored by
The Combustion Research Section
U.S. Environmental Protection Agency
1-291
-------
-------
PRODUCTION OF OXIDES OF NITROGEN

IN INTERACTING FLAMES
by
Christopher England
ABSTRACT

Measurements of oxides of nitrogen (NO ) were made in a 250,000 BTU/h
laboratory furnace fired with multiple ourners to determine the effects
of burner interactions on the effectiveness of NO control techniques.
Measurements were made with both gas and oil firings. When all burners
were operated at equal air-fuel ratios, the behavior of the furnace closely
matched that of full-scale boilers with furnace gas temperature and air-fuel
ratio the major controlling factors. When burners were operated at widely
different air-fuel ratios, NO levels were generally higher than at uniform
operation, a result which conflicted greatly with predicted values. Tracer
measurements indicated a very high degree of mixing in the furnace, a feature
which is believed responsible for the anomalous results.
1-293
-------
Introduction

Techniques to reduce emissions of oxides of nitrogen

in boilers are well known and generally center on either lowering

peak temperatures in the firebox, or lowering the available oxygen

in high temperature zones. In-plant testing has demonstrated the

effectiveness of these techniques, and laboratory testing has at-

tempted to quantify the reductions that are possible. Laboratory

testing, however, has been confined to the testing of single burner

systems, in contrast to commercial practice in which furnaces are

often fired with arrays of burners. The purpose of this work is

to evaluate NO reduction methods on a laboratory scale but in a

furnace with arrayed burners that can approximate the interburner

interactions that occur in commercial furnaces.
Apparatus
Furnace
Experiments were conducted in a 34.3 cm x 34.3 cm x 81.3 cm

(13.5 in x 13.5 in x 32 in) refractory furnace shown in Figures 1

and 2. Burners were placed at the bottom of the furnace in a machined

"M"-type refractory block. The walls of the furnace were firebrick

with a thickness of 11.4 cm (4.5 in). Covering the brick were sheets

of 2.5 cm "M" type refractory which in turn was surrounded by 0.64 cm

thick transite board. The ceiling of the furnace was a refractory
1-294
-------
cast around water-cooled pipes. The celling contained exit ports

in line with the gas burners followed by a refractory mixing section

from which samples of average furnace composition were taken.

Gas Burners

Three types of gas burners were used in the study including a

ring-type, a coaxial spud type and a radial spud type as shown

schematically in Figure 3. Swirl vanes were also used in conjunction

with both spud-type burners to determine the effects of air rotation

on the burner interactions.

Technical grade methane 92% CH4>[30.5 kj/m3 (1022 btu/ft3)]

from cylinders was throttled through critical flow orifices to meter

the flow of fuel to the burners. Air was supplied from the laboratory

supply and was also throttled through critical flow devices. Elec-

trical heaters were used to preheat air up to 315 C (600 F).

Two independent controls were maintained on fuel supplies so

that the air-fuel ratio of individual or manifolded burners could

be operated independently from the others. A single air supply was

used, and all tests were conducted with equal air flows in all burners.

Oil Burners

Oil burners were similar to the concentric spud gas burners

with 60 solid-cone spray nozzles mounted on spuds which were

fitted with swirl vanes as shown in Figure 3. A symmetrical pattern

of 5 burners was used including a central burner and a square array

of 4 burners, each spaced 14.0 cm (5.5 in) from the burner walls.
I- 295
-------
The spray nozzles were supplied by Delavan, and had capacities

of 1.9 1/min (0.5 gpm) each.

Instrumentation an4 Sampling

A chemiluminescent analyzer (Thermoelectron) was used for NO

analysis. Anomalous results due to peculiarities in the instrument

were found when measurements were made of total NO in fuel-rich
x

flames. This was believed to be caused by the reduction of NO by
A

carbon monoxide in the stainless steel converter which was used to

thermally decompose NO- to NO. Catalytic effects of metals in fuel-

rich flames are well known, and in this case, reduction of NO was
X

nearly complete. At equivalence ratios of 1.2 where CO concentrations

are on the order of 4%, NO was reduced to about 5 ppm when NO was
X

measured at concentrations above 400 ppra. As a result, all measure-

ments reported are for NO on a dry basis. Sample gases were dried in

successive ice and dry ice baths before measurement. This assured

that the same was free of all easily condensable gases.

Sampling was by a water-cooled quartz probe. Problems of poor

reproducibility were noted if this probe was exposed to fuel-rich flames

prior to use. This apparently was due to chemical reduction of NO by
X

carbon near the tip of the probe. As a result, it was necessary to

clean the probe after any exposure to flames containing free carbon.

Probing the interior of the flames was very difficult because of this

phenomena and, while several tests were made to measure centerline

NO concentrations, the results were not reproducible.
1-296
-------
The probe which was used had an orifice diameter of 1.6 mm

(0.063 in) and sampled at a rate of 9 1/min giving a sampling velocity

of from about 75-100 m/sec depending on the temperature. This apparently

was not sufficiently fast to prevent reduction. A better design might

have been a thinner quartz interface near the tip where heat transfer

and quenching would be faster. This design, however, would not be

compatible with sampling of coal flames that the probe was designed

to handle.

Operating Procedures

Operating procedures varied with the type of data desired and

the type of fuel used. For baseline parametric tests used to charac-

terize the furnace of the burner arrav, NO was measured as a
x

function of equivalence ratio (defined as the actual fuel-air ratio

divided by the stoichiometric fuel-air ratio) and temperature by

simply firing the furnace and scanning over various air—fuel mixtures.

Temperature readings were taken at each data point as the furnace

was heated; the resulting data were plotted as NO vs temperature
X

at constant equivalence ratio, and cross-plotted to give NO vs
X

equivalence ratio.

The temperature of the furnace was taken as the temperature

given by a thermocouple imbedded in the furnace wall. This tempera-

ture followed closely the temperature of the furnace gases when taken

approximately 2.5 cm above the furnace floor between burners, or
1-297
-------
when taken 25 cm above the furnace floor. Typically, the tempera-

tures were within 8 C of each other, an indication of the high

surface-to-volume ratio of the furnace.

When data were desired at a constant temperature, the furnace

was heated by firing at an air-fuel ratio at which this temperature

was self-sustaining. The furnace was then set at a desired air-fuel

setting and an NO reading was taken.
X
Results
Gas Burners
A variety of tests was made with an array of ring burners in

which the general characteristics of the furnace with respect to

NO were measured. Later tests were made with spud-type burners,
X

with and without swirl to determine the sensitivity of emissions

to different burner aerodynamics and flame shape.

Baseline Tests

For the ring burners, NO -equivalence ratio profiles were taken
X

as a function of air preheat temperature and furnace temperature

when all burners in the 9-burner array were held at a uniform equiva-

lence ratio. Figure 4 shows that air preheat temperature had a sub-

stantial effect on NO with emissions roughly doubling as the
X

temperature of the incoming combustion air was raised from 90 to

315°C (200° to 600°F). Emissions of NO were also greatly in-
X
1-298
-------
fluenced by furnace temperature as shown in Figure 5. A relatively

strong temperature dependence was observed above 1100 C (2000 F) in

air-rich flames while a weaker dependence was observed in fuel-rich

flames. Figure 6 is a crossplot of Figure 5 which shows more clearly

the relative effects of temperature in which lean flames display

marked increases in NO , especially above 1200 C (2200 F).
X

Ring Burner Interactions

Tests of the effects of flame interactions were run with the ring

burners in which the air-fuel ratios of the individual burners were

changed to determine the differences between single burner and arrayed

burner systems. "Interactions" were defined simply as deviations from

the calculated NO -equivalence ratio profiles based on the sum of
X

individual NO contributions as determined by plots such as shown in

Figure 6.

For example, Figure 7 shows data for a nine-burners array with 3

burners off. The predicted NO concentration when six burners are

operated at (J> = 1.0 and 3 burners are off is:
NO
x
6 3

9 N0..//t , rt,+ 9 NO
x(4 = 1.0)"
x«J) = 0.0)
or, from Figure 7
NO
9 (150) + 9 (0) - 100 ppm
1-299
-------
where the overall furnace equivalence ratio is

i 1
= 9 (1.0) + 9 (0.0) « 0.67.

It is apparent from these calculations that for the particular case
of a 6 on/3 off array, the maximum "predicted" NO should be two-thirds
X
of the maximum "identical" NO , a suggestion which is invalidated by
2C
measurement. Either by coincidence or by physics, "identical" burner
data corresponds closely with single burner data in smaller systems and
with tests with a large (7.62 cm, 3 in) burner in the multiburner furnace.
Concentric Spud Interactions
Figure 8 shows the interactions of concentric spud burners
with and without swirl. Again, surprising results were found in that
interactions were dominant in NO production with near-predicted
X
levels occurring in only very lean systems. The levels of NO were
substantially higher with the spud-type burner relative to the ring
burner, indicating possibly a better mixing pattern in the latter type.
The addition of swirl (a swirl number of approximately 0.5) raised
the overall NO levels of the concentric spud to peak levels above
1200 ppm although it did not change the interaction effects appre-
ciably. The swirling flames burn with a much greater intensity than
non-swirling ones, probably resulting in the higher levels.
Radial Spud Interactions
Figure 9 shows the interactions of radial spud burners with and
without swirl. The results were much like those of the concentric
1-300
-------
spud, although the observed flame intensities were greatest for these

burners. Reductions of NO below those of uniform burners were not
3t

found, and very high levels of NO in swirling flames were observed
X

in near-stoichiometric flames. As in the case of ring and concentric

spud burners, modest deviations from uniform burner operations, say,

operating three burners at an equivalence ratio of 0.4 and six at 1.4,

gave very little measurable change from NO levels with uniform opera-
X

tion. For this particular case, NO levels of 83 ppm would be expected
A

at the overall equivalence ratio of 1.07, while levels of about 1000 ppm

were measured.

Oil Burner Interactions

A five-burner array of burners was used in the oil tests, a feature

which allowed a greater burner spacing in the vertically-fired furnace.

Figure 10 shows the typical NO -equivalence characteristic when all

burners were operating at equal air-fuel mixtures. With the 60 solid-

cone nozzles and swirl vanes (swirl no. 0.5), the NO was only influ-
A

enced by air-fuel mixture ratio in fuel-rich flames, with steady levels

of about 400 ppm occurring from equivalence ratios from 0.4 to 1.0.

These levels were lower than for gas, a somewhat unexpected result.

Interactions for oil were much more predictable than for gas as

shown in Figure 10. Each burner appeared to act relatively independently

of the other with respect to NO , and, as a result, less interactions

were found. Why this should be the case is not clear, but may be
I- 301
-------
related to the insensitivlty of the oil burners to stoichlometry.

Increased air preheat, however, was capable of NO increase com-

parable to gas, indicating that the burners were not insensitive to

incoming enthalpy.

Discussion
The results with gas burners showed that, with the particular

geometry selected, the furnace was "resistant" to NO control by
X

burner patterning. Some probing of the furnace was done to determine

the cause of this resistance and the results are shown in Figure 11.

In these tests, done without combustion, helium was added in trace

quantities to determine the entrainment of gases between burners.

When helium was added to the center burner, it was entrained in the

outer burners in a reasonable manner, and the rate of dilution of

helium in the center jet was predictable from jet entrainment theory.

When helium was added to burners near the wall, extremely high rates

of entrainment of gases occurred, and after about 5 diameters, the

center jet actually exhibited more helium than the jet that was

introducing helium to the furnace.

What these results indicate is the very high degree of mixing in

the furnace, a feature that exceeds in importance the actual fuel-

air mixtures of the burners. Only a few diameters from the bur-

ners, the furnace appears to be almost homogeneous in composition.
1-302
-------
That the NO production levels tend to follow this composition
X

indicates that NO occurs away from the flame and in the plume,
A

possibly greater than 5 diameters from the burner.

By contrast, the oil burners were well-behaved, and

relatively predictable NO levels occurred when the burners were
X

patterned for low NO . The major differences between the gas and
X

oil systems were the burner spacing and, of course, the nature of

the fuel. Interburners distances were 11.4 cm (4.5 in) and 14.0 cm

(5.5 in.) for gas and oil, respectively. It is possible that a

critical burner spacing exists between these two values. A more

probable explanation, however, lies in the widely different NO
Jv

characteristics of the two fuels even when all burners are operated

with identical fuel-air mixtures. While multiple gas burner data

show a sharp peak in NO concentration in near-stoichiometric
X.

flames, multiple oil burners show no sharp peak. This may indicate that

NO from the latter burners is formed nearer the burner and is not
J\

as influenced by furnace mixing. It is expected that the jet entrain-

ment rates in both oil and gas burners are similar so that bulk mixing

phenomena should not be widely different.

The purpose of the tests is to allow the prediction of NO

levels from commercial furnaces by means of inexpensive laboratory

testing and suitable modelling to take into account factors of scale.

In the present tests, the laboratory furnace was built to match the
1-303
-------
volumetric heat release rate of commercial equipment, about 75,000

•j

btu/ft . It is not, however, possible in small furnaces to match

another important furnace variable, the surface-to-volume ratio.

An example of its influence on NO lies in the fact that the sub-
•. Jt

scale furnace walls take on the temperature of the furnace gases, say,

2200 F, while commercial equipment operates at cooled wall temperatures

of 1000°F. This ratio for the laboratory furnace 0.3 cm1 (0.89 ft1)

is more than 7 times higher than a commercial system.

Further tests have been performed on nitrogen-doped oil and are

now being made with a 5-burner array on coal. These results will be

reported at a later date.

Acknowledgment

This paper represents the results of one phase of research carried

out at the Jet Propulsion Laboratory, California Institute of Technology,

under Contract No. NAS 7-100, National Aeronautics and Space Adminis-

tration, and was sponsored by the Environmental Protection Agency.

The author wishes to acknowledge the valuable support by G. Blair Martin

of the EPA, and by Robert Silver and Raymond Devlin at JPL.
I-304
-------
Figure 1. Photograph of the Furnace Used for
Flame-Interaction Studies
I- 305
-------
Figure 2. Interior Photograph of the Furnace Installed
with an Array of Gas Burners
1-306
-------
UJ
CQ -— •>•-<
= CM
000

O O
o

(X
<
I-H
Q
to tj

o LU
ce
0
Studies
eract
s_
o
in
at
Q.
D __ 1
CO O
O
1
IS

LU O
£d
L
CO LU
CO
r
tn
s-
^
CT
1-307
-------
600
tn
500
400
X
o

S300
o
o
200
UJ
O

I
100
T
T
I
Burner Type: Ring
Furnace Load: 233,000 BTU/H
No. of Burners: 9
Air Preheat: 600°F
Furnace Temperature: 2100°F
I
I
600°F
500° F
400°F
300°F
200°F

I
0.6 0.8 1.0 1.2
OVERALL FURNACE EQUIVALENCE RATIO
Figure 4. Influence of Air Preheat on NOX Emissions
from the Multiburner Furnace
1-308
-------
600
ac.
CO
500
400
a.
a.
§300
o
z
o
200
o
o
100
I I I ' I
Burner Type: Ring
Furnace Load: 233,000 BTU/H
No. of Burners: 9
Air Preheat: 600°F
Furnace Temperature: Variable
2400°F
2000
1800
I
I
I
I
0.6 0.8 1.0 1.2
OVERALL FURNACE EQUIVALENCE RATIO
Figure 5. Influence of Equivalence Ratio and Temperature
on NOX Emissions from the Multiburner Furnace
1-309
-------
600
500
a:
a
s:
o.
a.
LU
o
1—4
X
o
400
o
* 300
o
z
o
as
LU
O

I
200
100
Burner Type: Ring
Furnace Load: 233,000 BTU/H
No. of Burners: 9
Air Preheat: 600°F
Furnace Temperature: Variable
EQUIV.
RATIO
0.8

0.9
I
t
I
1.0 -
1.1

1.2

1.3
1.4
1800 2000 2200 2400
FURNACE WALL TEMPERATURE, °F
Figure 6. Influence of Furnace Temperature on NOX Emissions
1-310
-------
600
a
LU
a:
500
a:
a
a.
Q-
400
x
o
shoo
o
o
200
o
o
100
| I | I T
Burner Type: Ring
Furnace Load: 233,000 BTU/H
No. of Burners: 9
Air Preheat: 600°F
Furnace Temperature: 2100°F
Interacting
I
I
air only-
O o
O O
O O
Identical Burners
I i I
0.6 0.8 1.0 1.2
OVERALL FURNACE EQUIVALENCE RATIO
Figure 7. Measured and Predicted Levels of NOX from
an Array of Ring Burners
1-311
-------
QJ
(U
c
o

1-
•f—
(0
T3
3 Z LU
CX »s, O
CO ID O
I- O
O CQ i—
•r- CSJ
J- O
+J O
CO o 3
C CO CD •*->
O CM •• O rd
(_> (/I ^O i_
$- (U
«• -O
>> _J 3 OJ I—
(— CQ -C

i. (J M- i- O
a> ro o a- *
c c c
V I- . i- fc.
330-^3
CQ U. 2: cC U-
V
•
o
o
•
0
0
•
0
o
o
-I—
•M

CO
•
o
o x
c-i O
O
a- -z.
• LU
> _i a»
i-i C
~^ t3 W
o- a» 3

" t5m

B S!

^S £5-
,— a: o.
= -, °
u_ ^^

CM _l « -^

.-1 rf T3 30
O O ««<->
^ 8*.
z o
00
•
o
CO
01
i.
o
o
o
o
CVJ
8
O
o
CO
o
o
IO
o
o
o
o
CM
(03ansv3w sv
Ndd '301x0 DIHIIN do NOiivyiN30Noo
1-312

-------
3 CO
O.
«/> O
CD
.— O
as

TJ CO
to CM
°0
o
CM
CM
o
O
o
O)
O.
-a
S- U 4- t- O
CO <0 O d. (O
c c c
5- i- . I- t.
3 3 O ••- 3
CO Li- Z «^ 11
CM
CO

O
S-
£
(O

O
S-
• LlJ
O _J
o
•a:
CM
O O
CO
VO
*
O
(U

CO
_J

o
•f—

-a

s_
«/»
s.
QJ
C
s-
3
QQ
OC
01
S~
3
O)
o
o
o
o
CM
o
o
o
o
o
00
o
o
10
o
o
o
o
CM
(G3ynsv3W sv
wdd '301x0 OIUIIN do NoiivyiN3DNOD
1-313
-------
600
=3
V)
500
400
Q.
Q.
X
o
^ 300
oc
u_
o
200
LU
CJ
100
Burner Type: Oil-45° Swirl
Furnace Load: 289,000 BTU/H
No. of Burners: 5
Air Preheat: 600°F
Furnace Temperature: 2100°F
dentical
air only
i I i I I i i I f
0.6 0.8 1.0 1-2
OVERALL FURNACE EQUIVALENCE RATIO
Figure 10. Measured and Predicted Levels of NO
from an Array of Oil Burners
1-314
-------
O
CM
O)

uj co
n:
L-
u_
S3H3NI '
woyj IDNVISIQ
1-315
-------
3:45 p.m.
Production of Oxides of Nitrogen
In Interacting Flames
Dr. Christopher England, Jet Propulsion Laboratory
Q: When you measured your NO , I didn't see where you
-n»
established that at the point of measuring, you were
sure that you didn't have any stratification. I didn't
see where you established that you really had no strati-
fication at the point of measuring, which can happen
when you have a portion of your burners off.

A: If you probe in the furnace after 10 diameters, every-
thing is mixed completely.

Q: That is a general assumption that is made in orifice
measurements. However, I don't think you can just
take it without establishing it.

A: Oh, no, no -- we have the tracer measurement.

Q: O.K., that was the question I asked. When you probed
for NO under combustion conditions, not when you had
X
a cold flow, did you probe for NO at different points,
A
and show uniformity of your data?

A: Oh, yes. And uniformity of temperature also.

Q: O.K., that was the question.
Q: If I heard you correctly, I believe you said that you
had correlated the results of your small-scale unit
with commercial units. I would like to ask what type
of commercial units you correlated the results with,
and secondly, what was the basis of the correlation - -
what measurements were made in the small unit -- and
what measurements were made in the commercial units •
A: The data I correlated with are data that were
taken by Southern California Edison in support of a
separate single burner project I had some years ago.
The single burner data, at that time, didn't agree —
1-316
-------
This is basically the air pre-heat data — that's
che best data. But, it was an old gas-fired Peabody
ring burner system, similar in many, many respects to
the furnace we operate.
Q: What type of fuel were you firing -- liquid fuel --
and what type of atomization was it?
A: This was with dieselene, using a 60° solid cone, Delzvan
nozzle that we operate at full capacity. I don't
know what the particles are.
Q: That's mechanical or air atomized?
A: Air atomized.
Q: And how were you determining your equivalence ratio?
A: The equivalence ratio was -- well, for oil?
[Yes.]
See, this is a small system. We had the oil on
rotometers and had the air on orifices.
Q: So you were measuring the air flow. You weren't
taking 0. readings.
A: That's correct.
Q: And your probing in the flame — this was done with
probes -- water-cooled probes?
A: The hot stuff was all done with -- well, NO was
A
done with a water-cooled quartz probe; the oxygen
stuff -- that's just done with a stainless steel
probe.
Q: The last question was what type of helium tracing
technique did you use?
A: Well, we used just a straight leak detector.
Q: Could you explain the rationale behind taking 2/3
I- 317
-------
of the emissions for estimating a proportion for
your projected or estimated emissions in dealing with
concentrations rather than quantities? If I understand
correctly, you were expressing your findings in terms
of PPM, not mass of NO Could you explain the rationale
X «
behind-you know, just taking — if you take three
burners out of action, then you're saying you're dealing
with only 6/9 of the total emissions.

A: O.K., we're turning the fuel off - the air is already
on -- so the total mass flow through the furnace is
comparable when compared on overall equivalence ratio.
The mass flow, total fuel flow and the air flow
are the same for the two curves for identical equi-
valence ratio and what we call interacting. Now,
we're just trying to define some handle to look at
just what the interactions are. That's a convenient
way to do it. If you look at the jets as though they're
completely independent and you put cooled walls between
them or adiabatic walls, you would expect 2/3 of the
NO .
If we operate six at 1.4 and three at air only,
expect to get the NO atom equiva.
of 1.4, 6 x 1.4 and 3x0, divide by 9.
you'd expect to get the NO atom equivalence ratio
Q: I think I'm bothered by the same thing Dr. Bartok is.
If you shut off the fuel, and you don't shut off the
air, when you're measuring the concentrations, then,
actually, you've changed the equivalence ratio on
the whole system. You're only putting in 2/3 as much
fuel, and if you haven't changed the air, you've changed
your equivalence ratio for the whole 9 configuration
unit.
1-318
-------
A: The abscissa is overall furnace equivalence ratio.

Q: Yes, I think maybe it ought to be plotted against
something else.

A: We're trying to compare two different cases, and the
overall furnace equivalence ratio is what we're
interested in. We're interested in overall stack
oxygen levels of 37, or whatever. So that's how they're
plotted.

Q: But when you plot on equivalence ratio, then actually
you're plotting not against the furnace with nine
burners on and the furnace with six burners on;
you're plotting against a situation where, in effect,
you've transferred that fuel from those three burners
to the other six burners. So, the question is whether
you release the same amount of heat through nine burners
or through six burners. If you release it through
six burners, you might have locally higher heat release
rates, which could locally up your NO generation and
X
explain your results. In that side of the furnace,
you're releasing more heat because you're putting all
the fuel in only 2/3 of the furnace. You may get
higher local temperatures and up your NO .

A: Yes, on the first data slide, I showed the
effect of load was essentially the effect of heat
release. And that is constant in our measurements
over a factor of 5 or 6, or whatever -- that's about
our turndown ratio. So that probably is not a factor.
Q: Would you care to reidentify the term "identical"
with respect to interacting?
1-319
-------
A:
Q:
O.K., the identical is when we took the furnace and
we ran all the burners at the same equivalence ratio.
This kind of gives us a baseline feeling for what the
furnace does as far as NO is concerned. Then, what
A
we define as interacting is what happens when we pattern
in some pattern. We alternate or otherwise change the
equivalence ratios on the different burners, and then
we measure the overall furnace NO . Originally, what
we wanted to do is determine optimum pattern. What
we found with gas was that there was no pattern that would
help us.
There is evidence from your charts and graphs that
high swirl and high NO were correlated. Could you
identify your observations on the shape of the flame
at high swirl numbers? Was it much reduced in volume,
for example?
It was much reduced in volume, yes.
Getting back to that previous question, I don't know —
I seem to be missing the boat here, but you justify
the conclusion by saying, well, you proved it by changing
the furnace load. However, if I understood it right
when you presented it, when you changed the load, you
maintain the equivalence ratio, you just vary the
total throughput. O.K., now, the point that the
gentleman was making, and the point I was trying to
make, is when you shut off three of your burners, and
you maintain your overall equivalence ratio, the equi-
valence ratio on the other burners has, therefore,
changed drastically. And I don't know how you can
justify it by saying, well, — I changed the load. . .
1-320
-------
with the same equivalence ratio on all burners.
I just don't see how to relate it; I think the
gentleman's point is very well taken, that when you
looked at your total equivalence ratio, and maintain
that constant or turning off the gas to three burners,
you therefore had to change the other burners, their
individual equivalence ratios, and therefore their
proportion of NO that they released had to be changed
X.
for the very data you presented.

A: Yes, but I don't think there is any question that load
is a second order factor in NO . I don't think that
-- our effects aren't second order effects. The other
thing is that in an equivalence ratio of 1.4, N0x should
be low, regardless of the heat release.
Q: You know, I am one of those semi-bad hardware guys.
There is no question that I can go air-only burners,
you know, on a large furnace and drop the NO. And
if I saw your photograph correctly, it looks like you
have got an adiabatic combustion system. As soon
as you do that, the staging doesn't mean a damn thing,

A: It's not adiabatic. The small systems have a small
burner

Q: Unless you lose heat between the staging, you've got
to get the same temperature. And I know I can pull
air-only burners on any large size furnace and I know
I could change the NO.

A: Well, that is true. That is why I said wall firing
helps you; burner spacing helps you. We have a very
well mixed system, and in the paper that will be in
1-321
-------
the proceedings, I point out that the difference of
our furnace and the furnaces that you work with is
that, in any small system, you have a very high surface
to volume ratio. The surface to volume ratio controls the
furnace temperature. To keep our furnace temperatures, our
gas temperatures up and our NO up to a level that is
A
comparable to a utility boiler, we must run wall
temperatures of 2200° F. Well, that is substantially
lower than adiabatic flame temperature, so there's no
way you can contend that it is adiabatic, if there's lots
of heat released through the walls. But the furnace's
dominant method of heat transfer is convection from the
walls. So if you measure the wall temperature, if you
measure the gas temperature, if you measure the plume
temperature, it's ten diameters away from the plume.
Everything is exactly the same. It's a very well mixed
system. It's different from a large system in which the
wall temperatures may be 1000 or 1500 degrees -- many,
many hundred degrees below the temperatures of the gases
that are entrained in the plume. That is the difference
between small furnaces and large furnaces; it is a tough
problem to model.
Q: I think you just responded to the question I was going
to ask. I recall from talking to you before that this was
an unsteady operation that you were working with, so that
when you had a fixed preheat temperature, and a fixed OF
ratio, you were able to get variable temperature by
variable conditioning of the furnace.
A. That's right . . .
I- 322
-------
Q: And therefore, locally right at the injectors, or at the
burners, you would have conditions that would be perhaps
typical of a large system, whereas in the bulk of the flow,
you would have conditions that were unique to your own
system. Therefore, you had two variables really -- if you
were looking at the near burner effects, preheat would be
a primary variable and the OF ratio to that burner; whereas
if you're looking at post flame, or bulk effects, it
became the wall temperature that was controlling. And
therefore, your curves almost need an extra parameter
in them. It just compounds the confusion, is all I was
saying.

A: One of the things our data implies is that the NO
is formed away from the burners, though.
I- 323
-------
-------
Combustion Chemistry
and Modeling: An Overview
by
Adel F. Sarofim
Department of Chemical Engineering
Massachusetts Institute of Technology
Cambridge, Mass. 02139
I-325
-------
ABSTRACT

The current understanding of mechanisms and kinetics
of reactions pertinent to the combustion of fossil fuels
is reviewed. Special emphasis is placed on reactions
leading to pollutant formation.

Int roduction

This overview presents a brief and necessarily incom-
plete coverage of developments in combustion chemistry and
modeling that pertain to emissions from stationary sources.
Prior to the passage of the Clean Air Act of 1967r in most
practical combustion problems the rate of reaction was just-
ifiably treated as being limited by the rate of mixing of
fuel and air. In a few cases, such as the blow out of high
output combustors or the velocity of flame propagation, the
importance of chemical kinetics was recognised but allowance
for kinetics was usually made using global rate expressions
(_1,2^ . The study of the details of combustion kinetics was
mainly the preserve of the scientists interested in mechan-
ism. It is only with the relatively recent recognition of
the dominant contribution to air pollution of trace noxious
species in combustion products that engineers have become
actively interested in modeling combustion kinetics.

As an illustration of the importance of knowledge of
flame kinetics on the development of pollution control strat-
egy, the emissions from the first stage of a two-stage com-
bustor burning a nitrogen-containing fuel are reported as a
function of fuel/air equivalence ratio in Fig. 1. The nitric
oxide emissions decrease, as expected, with increasing fuel/
air ratio. Reyond a critical fuel/air equivalence ratio,
however, an increasing amount of bound nitrogen escapes from
the first staqe of the combustor, and is available for oxida-
tion to nitric oxide in the second stage. The maximum reduc-
tion in NOX emissions and the fuel/air equivalence ratio at
which it will be attained (1.7 for the example in Fig. 1)
will depend on fuel composition, burner type, and air preheat
in a manner that depends upon the governing chemical kinetics.
Herein lies one practical motivation for studying mechanism
and kinetics.

It is of interest to note that the data in Fig. 1 were
generated by Smithells and Dent in 1894 (_3) using a cyanogen/
air mixture in a Smithells burner. Their objective was to
demonstrate that the combustion of cyanogen occurred in two
1-326
-------
o
O)
o
1/1
L.
- O
-«-•
C m
C9 =3
cn -Q

2 E
h o
— u
L-
CD
r

O
3QIXO Old-UN Ol N01Sd3ANCO %>

JO
cn
L.
c
O

(J
cr
•H
fr.
-------
stages, first to CO then to CC>2.
oxide were incidental.
Measurements of nitric
During the recent surge of interest in the modeling of
pollutant formation in flames, a majority of the effort has
been directed towards the understanding of the mechanism of
formation of nitrogen oxides. The complications that arise
in the modeling of the formation in flames of even this
simple pollutant are summarized in the following sections.

Formation and Destruction of Nitrogen^xides in Flames

A. Zeldovich Mechanism

The Zeldovich mechanism (4) has been firmly established
as the dominant reaction leading to the direct oxidation of
nitrogen in flames. The Zeldovich reactions are augmented
bv the N + OH reaction under fuel rich conditions (5). The
extended Zeldovich mechanism is given by the following three
reactions:
O = NO + N
N + O = WO + O
N + OH = NO + II
(1)

(2)

(3)
In the post flame zone, in which the O and OH radical
concentrations can be approximated by their equilibrium
values, good agreement is obtained between predictions based
on the Zeldovich mechanism and experimentally determined NO
formation rates. 7\n illustration of validity of the Zeldovich
mechanism is provided by the data and calculations of Lange
(([} shown in Fig. 2. Although Lange adjusted the rate limit-
ing constant (the forward reaction of Eq. (1)) to a value
slightly higher than the value reported in the Leeds com-
pilation (7) in order to improve the match of the calculated
NO profiles with his measurements, his adjusted constant is
within the range of uncertainty of the recommended value.

The NO profiles in Fig. 2 extrapolate to finite values
at zero time. This anomalous behavior has become known as
Fenimore or prompt NO (8) and is discussed next.
1-323
-------
1400
1200
IOOO
EXP.
CALC.
RUN 6 • 1012 % TA 403*
PREHEAT __
RUNS • 114.6 X T.A. 98O*
PREHEAT
RUN 4 • 128 5 % T A 350*
PREHEAT
0 02 04 06 .08 01
RESIDENCE TIME ABOVE BOTTOM PORT , SEC.

Fig. 2 Comparison of measured NOX profiles
with values calculated using the
Zeldovich mechanism (from Ref. 6)
I- 329
-------
B. Fenimore or Prompt NO

Two mechanisms have been proposed to explain the very
rapid build-up of MO in the early portions of premixed
hydrocarbon air flames.

(1) The contribution of superequilibrium free radical con-
centrations to pronpt NO

As a consequence of the increase in the number of moles
during the early stages of the combustion of hydrocarbon
fuels and the relatively slov; rate of the termolecular re-
combination reactions, there exists Tegiraes in flames
containing oxygen and hydroxyl radicals in concentrations
exceeding their equilibrium values by an order of magnitude
or more. This excess radical concentration has been found
to be sufficient to explain the prompt NO observed under fuel
lean conditions (5^1^,1^) . £n example of the effect of the
excess radical concentrations is shown in Fig. 3. The lower
of the two NO profiles shown was calculated using equilibrium
oxvgen concentrations and the upper was obtained with allow-
ance for superequilibrium oxygen concentrations (12_) ; the
shaded area represents the contribution to prompt NO of
superequilibrium radical concentrations. The magnitude of
this prompt NO is of the order of 50 ppm, modest by compari-
son to uncontrolled levels of emissions from large boilers,
but significant relative to controlled emissions or emis-
sions from smaller units. The results in Fig. 4 are used to
illustrate that the phenomenon of superequilibrium oxidant
concentrations is also encountered in laminar diffusion
flames (1^). Agents such as SO2 which can scavenge the ex-
cess oxidant radicals can be effective in reducing the mag-
nitude of this source of prompt NO as has been shown by
Wendt and Ekmann (this Symposium).

(2) The contribution of the formation of fuel nitrogen to
prompt NO

Fenimore (8) proposed originally that prompt NO is
formed indirectly by the oxidation of an organic nitrogen
compound, produced by a reaction such as
CH
= HCN
N
(4)
Such reactions appear to account for most of the prompt NO
encountered under fuel rich conditions. Conclusive evi-
dence for this mechanism is provided by the results of
Haynes et al. (14) summarized in Fig. 5 which show the
build-up and decay of HCN in a fuel-rich ethylene flame.
I- 330
-------
u
E
a
a
•I
O
d NO/dt ( equilibrium O )
4 -
2 -
O
O
Fig. 3
8
12 16 2O 24
TIME ( m Sec )

Rate of nitric oxide formation calculated

using equilibrium (bottom curve) and non-

equilibrium (top curve) radical concentra-
tions (from Ref. 12)
T- 331
-------
o
I—
(J
o:

UJ
O

z
llj
u
z
o
u
-3
10
-4
1O
2 = 1.2 cm
jFlame -
= 2.4cm
Z = 5.0cm
I
.65 .75 .65 .75 .85 25 .35 .45 .so

DISTANCE FROM SYMMETRIC AXIS .cm.

Fig 4 Comparison of equilibrium and nonequilibrium
oxygen atom concentrations at three elevations
in a methane/air laminar diffusion flame (from
Ref. 14)
1-332
-------
1000
NO
• HCN
• HYDROCARBONS
TIME (MSEC )
Fie;, s Measured profiles of XO, HCN, and
hydrocarbons in an ethylene-air flame, 0=1.64,
7'-2(MO°K. (from Ref. 14)
1- 333
-------
The NO formed by this route can be considerably in excess
of the equilibrium NO concentration for fuel rich condi-
tions.

C. Fuel Nitrogen Conversion to Nitric Oxide

Although the oxidation of organic nitrogen to nitric
oxide was recognized as early as the 19th century (15), Shaw
and Thomas (16), first recognized the importance of th"e con-
tribution oFThese reactions to NOX emission from fossil-
fuel combustion. In the past decade extensive data has
been generated on the conversion in flames of nitrogen
naturally occurring in oils and coals and of fuels doped
with a variety of model nitrogen compounds. The conversion
efficiency of the fuel nitrogen to NO in such systems is
determined by the competition of the reactions of a fuel
nitrogen derivative with an oxidant to form NO or with NO
to form N2. This competition can explain the observed de-
crease in conversion efficiency with increasing fuel nitro-
gen concentration and fuel/air ratio (for example, see Fig.
6). If the two competing reactions leading to the end
nitrogen products of N2 and NO are represented by
and
RN + OX
NO
NO + RX
RO
(5)
(6)
it can be shown (B) that the maximum NO concentration is
given by k5(OX)/k6. This simple relation was used by
Penimore to correlate his results. Although numerous
postulates have been offered for the identity of the key
intermediate RN, additional mechanistic studies are needed
in order to resolve present conflicting statements.

D. Reduction of Nitric Oxide with Hydrocarbons

The reduction of nitric oxide by hydrocarbons has been
demonstrated by Stern ling and Wendt (17) , Myerson (18) ,
De Soete (19) , amongst others. The probable mechanism is
the formation of a bound nitrogen intermediate by a re-
action such as
HC
NO
HCN
(7)
followed by the partial reduction of the fuel nitrogen
intermediate to N2- The results by De Soete shown in Fig.
7 demonstrate quite clearly the occurrence of such a
1-334
-------
r- in in co
C 01 00 N tp ID ID
d 6 6 d d d
on o
-------
proftl de

de NO (en suppogont

que HCN, HO et

aont lea seulee
+ 20,5 02 + 60,7 A
(X?IC)0 = 0,000290
d
0)
o
(0
v o
O
•H

O
-P
(N

O
c
-H

0)
0)
l-l
•H

-------
sequence of reactions. The NO injected into an ethylene/
^xygen/argon flame is converted to HCN, which in turn is
partially reduced to N2. The HCN intermediate may also
be converted to nitric oxide in the tail of the flame,
particularly under fuel lean conditions (19).

E. NC>2 Formation in Flames

Until recently periodic reports of significant NO2
formation in flames was discounted as being due to probe
errors. It now appears probable that NO^ is formed in
high concentrations by reactions of NO with H02 and/or
OH, and that the N02 may be subsequently reduced by re-
actions with 0 atoms. Evidence for this is provided
by Herryman and Levy (20), one of whose profiles is re-
produced in Fig. 8, and by the numerical simulations by
Kendall et al. (This Symposium). In most practical sys-
tems NO2 accounts for less than 5 percent of the NOX in
the exhaust but higher values can be encountered when
flames are rapidly quenched.

F. Catalytic Reduction of NO

Probe effects in NO sampling provide a good example
of the catalytic reduction of NO (21). The combustion
system most amenable to the exploitation of catalytic re-
duction is the fluidized bed combustor because the bed
substrate provides a high specific surface area. Indeed,
heterogeneous reduction of NO in fluidized beds has been
reported by Hammond and Skopp (22) , Jonke (23) , and
Pereira et al. (24).

G. Complete Kinetic Models for NO Formation

Detailed kinetic models of the reactions in flames
leading to NO formation have been developed for a few
well-defined systems. An example of such modelling is
the study of Engleman et al. (15) of NO formation in a
well-stirred reactor. They obtained excellent agreement
between calculations and measurements for CO/air and H2/
air mixtures for all the stoichiometries studied but
found that, for rich hydrocarbon/air mixtures, the mea-
sured NOX levels were higher than predicted. Later cal-
culations by Tyson and Heap (26) suggest that the dis-
crepancies for hydrocarbon/air mixtures are attributable
to the HC + N2 reaction.
1-337
-------
I
I
o
o »

I
I
i>
I
I
I
I
o
I
I i

i I
o «
CO

t
X
8 * I
41
i
-i
N-
xb o o
x "- ?
< c
« -?

« I •?
-S I i
b «s
e;
c

y.
\

p
M
p
6
p
«
p
*
p
N
06
uidd *ZON'ON
00
1-338
-------
H. Developing Combustion Process Modification for
Control
The understanding of the mechanism for NO formation,
although incomplete, is sufficient to permit the develop-
ment of optimized control strategies for aerodynamically
simple systems. Unfortunately, in many practical com-
hustors the complexities introduced by the turbulent mix-
ing processes and recirculation leads to an enormous num-
ber of possible paths for NO formation and destruction
(shown schematically in Fig. 9) and optimization is dif-
ficult. It is for this reason that the full potential
for NOX reduction by combustion process modification has
not been achieved.

Pollutants Other Than NOX

Of the emissions from stationary sources, the nitrogen
oxides have justifiably received the greatest attention
since they both pose a critical environmental problem and
are amenable to control by combustion process modifications
A listing of other combustion-generated pollutants is pro-
vided in Table I derived from the study by Locklin et al.
(27) . Many of these pollutants involve complex molecules
whose formation chemistry is poorly understood. With the
probable trend toward the burning of more coal, coal-
derived fuels, and possibly shale oils it is apparent that
the future problems of modelling combustion kinetics will
be associated with the chemistry of larger molecules, hav-
ing lower H/C ratios, and possibly higher nitrogen and
sulfur concentrations. In addition the chemistry of ash
behavior may become increasingly important. The gaps in
the chemical kinetics pertinent to the combustion of coal
and heavy oils that we should strive to fill include:

• solid and liquid phase pyrolysis: the rates and
mechanisms of formation of volatiles, carbonaceous
cenospheres and tar; and the distribution of critical
trace elements

• gas phase pyrolysis: the competitive and sequential
reactions leading to the formation of soot, tars,
polycyclic aromatic hydrocarbons, and light con-
stituents

• catalytic effects on the above: the modification
of the rates of the pyrolysis reactions by free
radicals or solid catalysts
1-339
-------
W
r}

*-"•.
£l
6?
o
rH
O
c
*cn -g
M (1)
r^l _i_l
g **
fa C
•rl
EH
Z 01
§ 8
E S
fa >
H -O
D X)
X O
CQ C
cn tn

O *H
M g
(0 3
in 01
M (0
S At
M
O
Q m
B 3
S i
i^ t*-
S o^
W -H
0
1 tJ
z o
0 -H
M r4
E-t 0)
CO P*
D
i 5
8 *
rl
0 0
frt
•H
4J
•H
•a
•o
<
rH
<1)
O
p.
MM

CO
4J
c
n)

•H
E
10
rH 4J
V C
a o
fa CJ

S
•H
+J
W
P
•8
0
CJ
«M
O 0)
41
CO 0)
4J--I
^&
•0 o
o o
M C
0< M

4J
(0
+>
d
rH
rH
O
(X
g
H
4J
CQ
n o)
i-2
L O O
U 2

*
01
§o
e
O rl 4J
Z HH <

r?
^

(0
CQ

J3
PU

01
>irH
? *
(04J
0) Q)
ac s

§r-4
Q)
QMS
2 «H fa
X
O
cn

J3
S

,3
J3 in
•H CD
-P «H
01 U
3-rl
.Q -P
^ W
O <0
U CU
tH
€
0

£<
O Z
a, ix

o
X

o
u

l-l
OJ
S
fa

^-•» <*•%,
H CN
X X
1 1
rH «H
^w^ ^-»
•*»• f^ OV
• • •
VO VO VO
oj m H

o
o
rH

0
o
rH

0
o
^J
^1

i-i fN
X X
** r-
• i
VO VO
CN in
m r*.
• • i
OS rH
00 rH

00
r- H

P- rH C
X 0
rn O
M
cn
g
*T -rl
«n »
CO
•r4
e
V

X
o
2
VM
O

in CQ
g
•rl
4J
U
A
IH
>M

<•
ro
X
^
.
I -x
*O in w
0) -H C
-P e o
-------
„ * «
z
1/1 o 7_
o (J "

o_ u o
^ 9

z £
o
n
O

t. i/j

u -r ^

~ D ""
I -t
L.

cn
1-341
-------
9 solid reactions: reactions such as the reduction of
metal and mineral oxides, e.g., SiC>2 + C > SiO + CO,
which can lead to the augmentation of the vaporiza-
tion of the oxides and subsequent recondensation as
submicron particles.

Literature Citations^

1. Longwell, J. P., and Weiss, M. A., Ind. Eng. Chem.,
4_7_, 1634 (1955).

2. Spalding, D. B. , Combustion and Flame, 1, 287 (1957).

3. Smithells, A., and F. Dant, Chemical Society Journal
Transactions, 65, 603 (1894).

4. Sarofim, A. F., and J. H. Pohl, Fourteenth Symposium
(International) Combustion, p739, The Combustion
Institute, Pittsburgh (1973).

5. Lavoie, G. B., Heywood, J. B., and Keck, J. C. ,
Combustion Science and Technology, 3L, 313-326 (1970) .

6. Lange, II. B. , Jr., "NOX Formation in Premixed Combus-
tion: A Kinetic r*odel and Experimental Data," Air
Pollution and Its Control, AIChE Symposium Series
126, Vol. 68, p!7 (1972).

7. Baulch, D. L. , D. D. Drysdale, D. G. Home, and L. C.
Lloyd, "Critical Evaluation of Rate Constants for
Homogeneous, Gas Phase Reactions of Interest in High
Temperature Systems," Dept. of Physical Chemistry,
The University of Leeds, England, Report No. 4,
December, 1969.

8. Fenimore, 0. P., Thirteenth Symposium (International)
on Combustion, p373, The Combustion Institute, 1971.

9. Thompson, P., T. D. Brown, and J. M. Beer, Combustion
and Flame, 19_, 69 (1972) .

10. Livesey, J. B., A. L. Roberts, and A. Williams, Com-
bustion Sci. Tech., 4_, 9(1971).

11. Williams, G. C., A. F. Sarofim, and N. Lambert, Proc.
General Motors Symposium on Emissions from Continuous
Combustion Sources, p!41, Plenun Press, 1972.
I- 342
-------
12. Sarofim, A. F., G. C. Williams, A. Padia, "Control
of NOX Emission by Partial Quenching," Presented
at Central States Section Meeting of the Combustion
Institute, Champaign, 111., March, 1973.

13. Mitchell, R. E., and Sarofim, A. E., "Nitrogen Oxide
Formation in Laminar Methane Air Diffusion Flames,"
Paper 75-3, Western Sections Meeting, Combustion
Institute, October, 1975.

14. Haynes, B. S., Iverach, D., and Krov, N. Y., Fif-
teenth Symposium (International) on Combustion,
Combustion Institute, Pittsburgh, 1975, p!03.

15. Berthelot, M. , Ann. Ohm. Phys. , 5_, 433 (1875), and
Later works.

16. Shaw, J. T., and Thomas, A. C. , 7th International
Conference on Coal Science, Prague, June 10-14, 1965.

17. Wendt, J.O.L., Sternling, C. V., and Matovich, M. A.,
(1973), Fourteenth Symposium (International) on Com-
bustion, Combustion Institute, Pittsburgh (1973).

18. Myerson, A. L., Fifteenth Symposium (International)
on Combustion, Combustion Institute, Pittsburgh (1975),

19. De Soete, G., and A. Muerand, Institute Francois du
Petrol, Report No. 21.381, May, 1973.

20. Merryman, E. L., and Levy, A., Fifteenth Symposium
(International) Combustion, p!073, The Combustion
Institute, Pittsburgh (1975).

21. Halstead, C. J., and A.V.E. Munro, "Sampling, Analysis
and study of the Nitrogen Oxides Formed in Natural
Gas/Air Flames," Presented at 1st AGA-IGT Conference
on Gas Research and Technology, Chicago, 111.,(1971).

22. Hammons, G. A., and A. Skopp, "NOX Formation and
Control in Fluidized Bed Coal Combustion Processes,"
ASME Paper No. 71-WA/APC-3, Presented at ASME Winter
Annual Meeting, Washington, D. C., November -
December (1971) .

23. Jonke, A. A., G. J. Vogel, L. J. Anastasia, R. L.
Jarry, D. Ramaswami, M. Haas, C. B. Schoffstou, J. R.
Pavlik, G. N. Vargo, and R. Green, "Reduction of
1-343
-------
Atmospheric Pollution by The Application of Fluidized-
Bed Combustion," Report No. ANL/ES-CFN-1004, Annual
Report July 1970 - June 1971 by Chemical Engineering
Division Argonne National Laboratory, 111.

24. Pereira, F. J., J. M. Beer, B. Gibbs, and A. B.
Hedlye, Fifteenth Symposium (International) Combus-
tion, pl!49, The Combustion Institute, Pittsburgh,
1975.

25. Engleman, V., S., W. Bartok, and J. P. Longwell,
Fourteenth Symposium (International) Combustion,
p755, The Combustion Institute, Pittsburgh, (1973).

26. Heap, M. P., and Tyson, T. J., Reaction Screening
for Methane/Air Combustion," EPA Contractors Meeting,
SRI, Palo Alto, Calif., June, 1975.

27. Locklin, D. W., et al., Air Pollution and Its Control,
AIChE Symposium Series, 126, Vol. 68, (1972).
1-344
-------
4:25 p.m.
Combustion Chemistry and Modeling: An Overview
Adel Sarofim, Massachusetts Institute of Technology
Q: I do not think that NO is strictly necessary in order
to form N,. from fuel nitrogen. Reactions that will
produce N» additional to that of NO plus NR are those
of two NR's or NR plus N. Four center reactions pro-
posed to explain the formation of N» would be ex-
pected to be slow, i.e., have a high activation energy,

A: The primary reactions that form N2 from fuel nitrogen
in flames are not well established. That the fuel
nitrogen conversion is determined by competition
between reactions involving NO to form N and an
oxidant to form NO is supported by the successful
correlation of data using such simplified models.
There is, however, need for more detailed studies of
mechanisms and rates of the dominant reactions.

Q: The formation of HCN by the reaction of a CH radical
with N_ should not be labeled fuel nitrogen. Although
such reactions can explain the formation of prompt
NO under fuel-rich conditions, their kinetics are not
as well understood as the reactions leading to NO
formation under fuel-lean conditions.

A: Agreed.
In a boiler furnace with water walls at, say, 450° F,
flame temperatures of 2000 to 2200° F, and a luminous
wrinkled flame, there is evidence of a black space
between the burned gases and the walls. Would you
hazard a comment on the nature of the soot or soot-
like products in the black space?

The most likely constituents are soot and unburned
I- 345
-------
char for coal-fired boilers and soot and carbonaceous
cenospheres for boilers fired vith residual fuel oils.
At flame temperatures the soot and carbonaceous particles
would be expected to contribute to the flame luminosity.
Dark spaces are usually confined to cooler regions,
close to the burner nozzle or where non-uniform
mixing has led to quenching of partially burned
combustion products.
1-346
-------
THE MATHEMATICAL MODELING

COMBUSTION DEVICES
By:

T. J. Tyson

Ultrasysterns, Inc.
2400 Michelson Drive
Irvine, California 92715
1-347
-------
1.0 INTRODUCTION

The mathematical modeling of combustion devices to predict

pollutant formation involves the detailed description of molecular

transport and chemical kinetics in regions which are small in compari-

son to the total device volume and where reactant concentrations are

determined by macroscale mixing. This contrasts with the prediction

of heat transfer in these same devices where the fine detail of heat

release can often be ignored for practical design purposes. This

paper has three objectives:

• To review the status of mathematical modeling of
combustion physics and chemistry, and the engineering
analysis of practical combustion devices.

• To discuss the potential for the development of analytical
tools which could be of value to the combustion engineer
faced with the development of advanced environmentally
acceptable combustors for fossil and alternate fuels.

• To suggest areas in which future efforts should be
concentrated in order to ensure that the maximum use
can be made of mathematical models for the solution
of engineering problems.

The subject area is exceedingly broad and the paper cannot be considered

as an overview but rather as a discussion of what might reasonably be

expected in the near and long term from mathematical models. In

assessing the present status and prospects for the future, modeling

will be examined both from its mathematical (computational) aspects

and from the adequacy of the imbedded physics and chemistry.
1-348
-------
There exists a broad spectrum of combustion models with

regard to their mathematical and physical sophistication, complexity,

accuracy, cost both to develop and use, and applicability. Such

models may, on the one hand, be suitable only for the detailed

analysis of a specific phenomena occurring within a combustion device

or under idealized laboratory conditions such as droplet burning,

coal particle devolatilization, or the kinetics of complex chemical

systems. On the other hand, the engineer also needs in his inventory

of modeling tools, techniques which are suitable for the analysis of

complete practical combustion devices which represent the arena for

a multitude of complex physical and chemical phenomena. Mathematical

models are not necessarily restricted to those which are totally

predictive in nature, that is, models which allow the prediction of

the entire behavior of a given combustion device and hence allow the

synthesis of new designs through parametric variations. There is

another important, and often mathematically simpler, class of models

for use in the analysis of experimental observations. Such models

can be used to deduce the nature of physical and chemical processes

which have not been, or cannot be, directly measured.
1-349
-------
2.0
BACKGROUND
2.1 Role of Mathematical Modeling

Before addressing the three objectives stated above it is

necessary to recognize how an engineer utilizes mathematical modeling

techniques in the design and development process and to realize the

limitations of these techniques. As in the case of many other

engineering endeavors the utility of mathematical analysis in the

development of combustion devices can be very substantial. However,

the maximization of this contribution will only come about through

a realistic appraisal of what can and cannot be accomplished with

present-day and near-future analytical tools. It is wishful thinking,

for example, to suppose that three-dimensional elliptic codes will be

available in the near future for the satisfactory detailed analysis of

complex turbulent flame behavior. To suggest that this is possible

provides a disservice to the combustion engineering community since

failure to achieve anticipated results will only lead to a lack of

enthusiasm for the application of techniques which do have potential

for success. On the other hand the academic community should not be

discouraged from developing such codes since only then can their

limitations be exposed and corrected.

Seldom, if ever, is one faced with an engineering task that can

be solved in its entirety through the straightforward application of

a mathematical model to synthesize an optimum design by the systematic
1-350
-------
variation of operational and design parameters. This situation is

especially apparent in the field of combustor design. The reasons are

three-fold: (1) the fundamental phenomena associated with the dominant

physical and chemical processes are not sufficiently well understood;

(2) the mathematics of numerical solution techniques and the capabili-

ties of modern computational machines are not sufficiently advanced;

and (3) the design parameter definitions and variations are either

unknown or too numerous at the outset of the synthesis process to

allow a systematic treatment. These factors are, of course, intimately

coupled and advances in any one area relate strongly to the require-

ments imposed upon the others. For example, as is so often the case,

physical behavior may be well understood on the molecular level while

gross physical phenomena which are a consequence of this behavior are

difficult to correlate with easily identified physical observables.

Taking turbulent transport as an example, the time dependent Navier-

Stokes equations adequately describe the continuum physics of fluid

motion except for the unusual situation of highly rarified gases.

These equations are based upon an understanding of behavior at the

molecular level. However, because of computational limitations it is

impossible today, and maybe for the next several decades, to solve

these equations of fluid physics. Thus the model is forced to adopt

cruder "phenomenological" descriptions of turbulent transport which

are based upon examination of time-averaged motion over periods

which are long compared to the unsteady turbulent fluctuation periods
1-351
-------
which characterize fluid motion at high Reynolds numbers. Satisfactory

phenomenological descriptions of turbulence have defied the very sub-

stantial efforts of physicists and engineers for all but the very

simplest of flows. Thus we have sacrificed understanding in one area

in order to accommodate limitations in another.

Perhaps the most useful contribution of mathematical modeling

to the design of combustors in the near-future will come from the

analysis of so-called "limit-case" situations in which some single

aspect of the physics or chemistry dominates and the remaining pheno-

mena can either be ignored or modeled by simple idealized processes.

Fundamental laboratory experiments are generally designed to fit this

situation and the availability of mathematical models with which to

analyze the results are of considerable value. "Limit-case" analysis

serves another useful function in that it often allows an estimation

of the upper bounds of performance of a combustion device as limited,

for example, only by fundamental chemical constraints. Such an analy-

sis provides a significant yardstick by which to measure the perform-

ance of various design concepts which may be limited in performance

by fluid mechanical and heat transfer constraints.
1-352
-------
2.2
Classification of Mathematical Models
In order to appreciate the breadth of the domain referred to

as "mathematical modeling of combustion behavior" it is useful to

classify the various model types according to their engineering

functions as follows:

• Data Analysis. Procedures which allow the extraction
of the maximum amount of information from experimental
data. For example, mean (time averaged) flow field
and concentration measurements contain implicit
information about the nature of turbulent transport
and chemical production which can be obtained through
rather straightforward analysis.

• Prediction of Behavior of Fundamental Laboratory
Experiments. Here the models and experiments focus in
considerable depth on some limited aspects of the
physical or chemical behavior of flames with the compli-
cations of other phenomena being eliminated by careful
design. Examples are the modeling of well stirred
and plug flow chemical reactors, flat molecular diffusion
flames, spherically symmetric particle flames, and cold
flow experiments.

* Empirical and Order-of-Magnitude Analysis of Engineering
Devices. This important category covers analytical
techniques ranging from "back-of-the-envelope"
estimates to extensive correlations of empirical
evidence cast into relatively simple models for
predicting energy release, heat transfer, internal
flow behavior, fuel burn-out, pollutant formation,
flame stability, ignition characteristics, etc.

• Development of Engineering Intuition, Guidance of
Experiments, and the Synthesis of Design Concepts.
Many types of engineering analysis tools fall into this
category. "Limit-case" models which treat certain limiting
situations in considerable detail can establish bounds on
the performance of a device. Such analysis can lead,
through numerical experimentation, to an intuitive
feel for the influence of significant design and
operational parameters on performance and scaling.
1-353
-------
2.3 Some Physical Aspects of Combustion Modeling

The modeling of combustion devices is amongst the most

difficult of all the classical engineering problems in continuum

mechanics. This is primarily due to our lack of basic understanding

of the phenomena of turbulence. Virtually all combustion devices

involve turbulent flow. In those instances where fuel and air are

admitted separately to the combustion chamber the rate of heat

release is controlled by turbulent mixing. It is difficult to give

a precise definition of turbulence, texts rather list the character-

istics of turbulent flows . Turbulent flows are irregular and

dissipative, their diffusivity causes rapid mixing. Turbulence is

rotational and three-dimensional and is a continuum phenomenon

governed by the equations of fluid mechanics. Even the smallest

scales occurring in turbulent flows are ordinarily far larger than

any molecular length scale. From the viewpoint of combustor modeling

the most important characteristic of turbulent flows is that they are

flows.

Turbulence is a property of the flow and not a property of the

fluid. It has been well characterized for only a few classical

situations which can be divided into two groups. One group consists

of boundary influenced turbulence and the other group consists of

boundary-free or "free" shear layers. The latter includes planar

and axisymmetric jet mixing and free mixing layers in which two

fluids of different velocity and large lateral extent merge. To a
1-354
-------
much lesser degree turbulence measurements have been made in other

types of flows such as boundary layer separation and reattachment

zones and in internal recirculation regions.

Provided that the mean flow gradients are not too large

the turbulent behavior is dependent on the local velocity distribu-

tion over a region of the scale of the largest turbulent eddies.

Such a condition is referred to as equilibrium turbulence and

represents a condition where turbulence production and dissipation

are in balance. In flows where the mean flow gradients are very large

this equilibrium condition is violated and turbulence generated up-

stream may be convected to the point under consideration without

experiencing equilibration. Similarly local turbulence production

may lag far behind in a rapidly accelerating flow leaving the region

"laminarized." Under these circumstances of rapid mean flow variations

a nonequilibrium (or so-called "history dependent")turbulence model

is required to predict the local turbulence energy and length scale.

Such a model must contain a description of equilibrium turbulence

in such a way that when the mean flow gradients decay the turbulence

proceeds asymptotically to the equilibrium value.

Figure 1 presents a schematic of a typical combustor flow

field with swirl induced recirculation. For the purposes of

discussion the mean flow gradients are assumed to be sufficiently

small that the turbulence has an equilibrium distribution. The
1-355
-------
\\\\\\\\
c
o
1-1
4J
03
r-1

U
M
"O
V
U
4)
tJ
o
4J

§ 8
u o
•H
•H 00
(6 (U
u erf

IX 01
tN 4J
tH 0)
u
•S
00
1-356
-------
circled numbers indicate regions where the equilibrium turbulence

levels probably take on a distinctly different dependence on the

local conditions due to the different mean flow character. The

regions can be distinguished as follows:

Region 1 - Internal recirculation

Region 2 - Typical of confined planar jet mixing

Region 3 - Typical of planar free shear layer

Region 4 - Boundary layer reattachment

Region 5 - Recirculation zone boundary layer

Region 6 - Conventional wall boundary layer

Region 7 - Axisymmetric jet mixing

Hopefully measurements taken in flow environments typical of these

will yield the appropriate relationship between turbulent viscosity

and the local flow kinematics.

The physics of turbulent transport is described by the time

dependent Navier-Stokes equations for which no solutions have yet been

obtained and these equations will probably continue to defy mathe-

matical solution for several decades. In an attempt to overcome this

dilemma the modeler wishing to predict such flow fields as shown

in Figure 1 is forced to solve the time averaged Navier-Stokes

equations. This approach entails the "phenomenological" or "empi-

rical" modeling of turbulent transport which can be undertaken at

various levels of sophistication and mathematical complexity.

Unfortunately very little data from detailed physical observations
1-357
-------
exists to guide these efforts. The result, all too often, is the

substitution of complexity for reality in order to hide our ignorance

of turbulent transport. Thus "history dependent" turbulence models

may be substituted when classical mixing length theories prove

inadequate because of a lack of understanding of the relationship

between turbulent transport and the local kinematics of the flow in

slowly varying mean flow fields.

Another aspect of turbulent behavior which adds greatly to

the complexity of combustion phenomena is the unmixedness on a mole-

cular scale which exists throughout the combustion field. This un-

mixedness is a consequence of the marble-cake like structure

associated with the large scale turbulent eddies which entwine fluid

originating from different parts of a mixing zone and which dominate

the mass trasport process. These eddy structures are in themselves

unstable and a cascading process ensues with breakdown to smaller

and smaller eddies. This process is eventually terminated by mole-

cular diffusion eliminating the last traces of unmixedness. It is

only in this final stage of the mixing process that chemical action

and energy exchange can take place between the originally separate

reactants. This delay time between "macromixing" on the turbulent

scale and "micromixing" on the molecular scale must be compared with

the characteristic times for chemical reaction in order to quali-

tatively assess the significance of this segregation of material.
1-358
-------
For large low velocity combustion systems the eddy break-up time

tends to be long and the significance of this "turbulence-kinetics"

interaction can be substantial. The phenomena effects the rates of

energy release, pollution formation, and combustion product

quenching which in turn can effect the ultimate levels of nitrogen

oxides and incomplete products of combustion. No adequate treat-

ment of turbulence-kinetics coupling exists at present for even the

simplest flow and chemical systems. However, it is not unreasonable

to expect progress in the near future in this area based on relatively

simple models allowing for turbulent macromixing in conjunction with

segregation on the microscale which decays with time. Adequate

modeling here awaits the generation of a definitive set of experiments.

Two-phase flow and spray combustion behavior adds enormously

to the complexity of any analysis approach which attempts to solve

for the detail structure of the combustion field. Vaporization and

burning histories of droplet clouds are very different from the

simple idealized case of single drops burning in a quiescent non-

vitiated and unconfined environment. Droplet and solid fuel behavior

can be strongly coupled to the local relative Reynolds number between

the particle and the gas if the particles are large and the gas phase

undergoes substantial acceleration. Any model which attempts a

detailed treatment of two-phase combustion must integrate the particle

ballistics equations in order to establish gas-particle velocity

lags and to locate the particle trajectories within the combustion
1-359
-------
flow field. This is particularly important during the initial

injection period even for the smallest particles. The degree to which

particles participate in turbulent motion is poorly understood and

yet may be crucial to their combustion behavior. In general, for

large combustion systems particles tend to follow the turbulent

motion. That is, an eddy is likely to carry its particulate content

with it as it moves. For small systems, on the other hand, particles

tend to follow the mean motion. The velocity lags associated with

turbulence tend to be higher in small systems giving rise to a larger

dependence of vaporization and burning rates on turbulence intensity.

(2)
Figure 2, taken from Beer and Chigier and McCreath and

(3)
Chigier , illustrates the complexities of spray combustion in

swirl and bluff body stabilized flames.

2.4 Some Jtethematical Implications of Various Modeling Approaches

Before examining the current status and future potential for

modeling it is appropriate to anticipate some of the mathematical

problems which must be adequately handled by the numerical techniques

employed. To place these mathematical problems in perspective it

should be noted that the two areas which will undoubtedly pace

advances in mathematical modeling for some time to come are improve-

ments in numerical computational capability and advances in the

understanding of turbulent behavior.
1-360
-------
C 60
O -H
T( J3
*J U
W
g ^

§
M M
CL U
(A O
O T3

M B)

-------
2.4.1 Near Equi1ibrium Behayior

Several phenomena occurring in combustors are often of a "near

equilibrium" nature in that an equilibrium condition is rapidly

approached at a rate which is strongly coupled to the degree to

which the system deviates from equilibrium. The equilibrium state

is determined by the local kinematics and thermodynamics of the

system. Chemical kinetics, turbulence, and particle dynamics can all

exhibit this behavior. The differential equations governing these

approaches to equilibrium are said to be "stiff" in that any distur-

bance from equilibrium is met by a strong restoring "force." The

numerical integration of coupled stiff equations requires special

(4 5)
attention in order to avoid numerical instabilities ' . Implicit

integration techniques are required which can contribute to very long

computational times.

2.4.2 Characteristics of Full Equations of Fluid Motion

In general the time-averaged low-speed Navier-Stokes equations

have the following features in addition to the nonequilibrium behavior

discussed above:

• Nonlinearity

• "Singular" limit solutions as the effective
Reynolds number becomes large

• Thin flame (energy release) zones in the limit of rapid
chemical kinetics

• Elliptic behavior

* Small pressure differences, i.e., AP/P «1
1-362
-------
These features have implications regarding the numerical solution

treatment and provide motivation to attempt simplifications. Non-

linearity rules out "general" solutions and superposition of solutions.

It also introduces further complexity into numerical convergence schemes.

The singular nature of high effective Reynolds number solutions refers

to the fact that the limiting solutions, as the effective viscosity

approaches zero, are not the same as zero "viscosity" solutions. This

is reflected in the existence of thin regions of high turbulent

transport such as boundary layers and boundary-free mixing layers.

The presence of these thin zones has significant numerical implica-

tions. They suggest the need for a fine numerical mesh in certain

regions of the flow where the gradients are large whereas a coarse

mesh would suffice elsewhere. The possible existence of thin flame

zones imbedded within mixing layers carries similar implications

with regard to numerical mesh size control.

The elliptic nature of the problem refers to the short-range

turbulent-transport (diffusion) coupling and pressure wave coupling

between an element of fluid and its surrounding neighbors. One

consequence of the turbulent diffusion of momentum is the possible

presence of large scale recirculation zones. These vortex structures

provide strong convective "upstream influence." Such a situation

is properly set as a "boundary-value" problem because all boundaries

can influence the entire flow and downstream behavior can effect
1-363
-------
upstream processes through convective signalling. Again this has

strong implications for the type of numerical procedures to be

employed.

All numerical procedures replace the differential equations

with fully coupled algebraic equations linking the unknowns at

every node point in the numerical mesh. It is the numerical solution

of this coupled set of nonlinear algebraic equations that presents

the difficulties. The iterative method of solution amounts to

obtaining successive approximations through a procedure of uncoupling

the node points to a greater or lesser degree depending upon the

method. After each iteration an error component or "residual" is

calculated for each subset of uncoupled nodes which represents a

correction to the coupling terms. This residual is then used to

adjust the solution and the cycle is then repeated. As can be in-

ferred from the "singular" nature of the problem it is not surprising

that the physics of the element-to-element coupling plays an

important role in the convergence of any scheme based on partial

uncoupling. As high effective Reynolds numbers are approached

the diffusive coupling allows for solutions with very large gradients

over very short distances (thin layers). Such solutions are needed

to accommodate wall boundary conditions and mixing layers disconti-

nuities. The analogue to this behavior in the finite difference

numerical iteration scheme is that the reaction to a small error
1-364
-------
at a node point is a large gradient to eliminate or diffuse away

that error. However, if the grid size is too large the analogy of a

"thin" high gradient region does not hold. The numerical scheme

applies the large gradient over a large distance and introduces an

error at the adjacent grid point which is larger than the original

error. This leads to exponential error growth. Hence for numerical

stability we require a small mesh size when the turbulent viscosity

becomes small. In fact, a careful analysis shows that the "cell

Reynolds number" must be small to ensure stability. The numerical

values of this critical cell Reynolds number depends upon the degree

to which the nodes are uncoupled in the iterative solution process.

The greater the coupling the greater the numerical stability. This

is the motivation behind going to so-called "block", "line", and

"alternating direction" relaxation procedures.

One approach to achieve numerical stability without the need

to use an excessively small grid mesh is to introduce an artificial

viscosity which satisfies the critical cell Reynolds number criteria.

Meteorologists originally introduced the concept of "weather"

differencing which provides such an artificial viscosity. More

commonly called "upwind" differencing this technique biases the

numerical estimate of the convective terms to the values existing

on the upstream side of the node point in question. The numerical

error introduced by this procedure is equivalent to an additive
1-365
-------
diffusion term with an artificial Reynolds number on the order of 2.0.

This in turn is equivalent to introducing additional "turbulent

eddies" into flow whose size is of the order of the grid spacing.

This poses no serious problem when the real eddy scale is substantially

larger than the grid spacing necessary for accuracy in regions of

steep gradients such as those which exist in mixing layers. However,

outside the thin boundary and mixing layer zones in the so-called

"outer" regions of the flow the turbulence scale is very small and

the gradients are gentle. In these regions it is desirable to use a

coarse grid. In doing so, however, one introduces a substantial error

through the artificial viscosity.

The numerical procedure that is typically followed at present

is a combination of the above approaches. The alternating direction

line relaxation or so-called ADI (Alternating Direction Implicit) method

( 6 )
first introduced by Peaceman and Rachford is employed to maximize

the value of the critical Reynolds number. The cell size is then

reduced to avoid violation of this criteria. In addition, some codes

still invoke a partial upwind differencing to gain stability. Finally,

the method of iteration to achieve convergence of the ADI procedure

is related to cell Reynolds number and stability. Briley at

UARL has developed a method of maximizing the convergence rate for a

given cell Reynolds number.
1-366
-------
2.4.3 Characteristics of the Parabolic Equations of Motion

Great simplification of the mathematical and numerical

treatment can be achieved if the classical Prandtl "thin shear layer"

assumptions can be invoked. The necessary conditions are that:

• There is no reverse flow.

• The flow has a single dominant direction along one
set of coordinate lines with the velocities in the
lateral plane small compared to the streaming velocity.

• The laminar or effective turbulent Reynolds number
be large, say greater than 10 (this condition is
satisfied by typical turbulent free shear layers
which have effective Reynolds numbers in the range
15-35 based on the length of the shear layer).

The significant implications of these assumptions are:

• Shear layers, boundary or free, will be thin
compared to the streamwise length) and lie close to
the principal coordinate lines.

* Derivatives in the streamwise direction (x) are much
smaller than in the lateral direction (y,z) and as
a consequence d/dx. (T , T , T ) are small compared
with other stress terms ancrcan oe set equal to zero.

• The pressure field can be represented by

P = P~(x) + eP'(y,z)

where e « 1. This implies that the pressure gradient
in the streamwise direction is approximately independent
of the position in the lateral plane.

As a consequence of these implications regarding the flow behavior

the overall governing equations reduce to parabolic form which means

they are properly posed as an "initial value" problem. That is,

the solutions exhibit no upstream influence and the equations can
1-367
-------
be integrated by "marching" downstream from known conditions at an

upstream starting point. This is an enormous simplification which

allows two-dimensional flows to be treated almost with the simplicity

of one-dimensional problems and three-dimensional flows to be solved

with machine times comparable to the iterative solution of two-

dimensional elliptic problems. Hence it is important to try to

recognize those regions in combustion devices which may satisfy the

necessary conditions stated above. Figure 3 suggests two possibi-

lities. The lower schematic represents the combustion field of a fuel-

staged device where fuel is injected in such a way that it penetrates

far downstream from the initial heat release zone. If the ignition

region can be solved in an approximate fashion, then such a solution

could be used as the initial condition for a parabolic two-dimensional

(axisymmetric) procedure for the remainder of the reaction zone.

The upper schematic in Figure 3 represents one possible model of

flow in the secondary (S) and tertiary (T) zones of a gas turbine

combustor. Here it is assumed that the (S) or (T) air jets can be

modeled by aerodynamically clean "source bodies" immersed in the

flow which circumvents the problem of calculating the jet penetration

behavior (which is done in this case as a preliminary to the calcula-

tion by simple jet penetration theory). Under these assumptions the

three-dimensional jet interaction problem can be solved with a

three-dimensional parabolic procedure.
1-368
-------
CO
Ml
•H
1-369
-------
3.0 STATUS OF CURRENT MATHEMATICAL MODELING PROGRAMS

This section presents a brief review of the current status of

several combustor device modeling programs which have been supported

by the EPA either in whole or in part. The comments are limited to

modeling of continuous combustion devices. None of the EPA-sponsored

work on reciprocating internal combustion engines is discussed. The

review is not comprehensive either in depth or in breadth and is only

intended to provide some general insight into the type of investiga-

tions which are being pursued. In addition to -these efforts brief

mention is made of other pertinent non-EPA-sponsored work.
3.1
One-Dimensional Diffusion Flames
3.1.1 The Aerotherm Flat FlameCode
/ a \
Kendall and Kelly have developed the PROF code at

Aerotherm for the analysis of premixed flat flames. The code couples

an arbitrarily large set of finite rate chemical reactions with

molecular diffusion. Only solutions for low-speed combustion waves

are sought and hence the assumption of constant pressure can be

invoked. The problem is properly set as a boundary value problem and

utilizes matrix inversion techniques to solve the fully coupled set

of finite difference equations. For the constant area duct computa-

tion the flame speed becomes the eigenvalue for which a steady state
1-370
-------
solution exists. An alternative to the eigenvalue formulation is to

perform the calculations in a duct of slowly varying cross section

which diverges in the direction of the flow, A stable flame will

stand at the station corresponding to the appropriate flame speed.

Such a flame deviates slightly from one in a constant area channel due

to the small divergence. The primary purpose of the code is to

examine flat flame data in order to extract information regarding

the kinetic mechanism of NO formation in flames. Of particular
X

interest is to establish to what degree there is a deviation from

the simple Zeldovich theory, and to elucidate the mechanisms of this

"prompt" NO formation in hydrocarbon flames.

Figure 4 presents a comparison between the carbon monoxide

(9)
flat flame data of Merryman and Levy and the prediction of the

species distributions. The agreement is excellent. Figure 5 shows

a comparison between predicted and measured H_/air flame speeds.

3.1-2 The Rocketdyne Spherically Symmetric Particle Diffusion Code

Nurick and co-workers at Rocketdyne have developed a set

of spherically symmetric diffusion codes for the prediction of:

• Droplet fractional distillation history

• Coal devolatilzation and surface reaction history

• Pollutant formation in the near field of particles

with emphasis placed on the fate of fuel nitrogen

compounds
1-371
-------
ca oo

T3 O
V H
•H
"O W

-------
CO
c«
O

c
(S
0)

(Xi
"S
•D
0)
•o o
0) -H
M 4-1
PM a
•H
4-1 M
o cd
o -o
co a)
•H
-------
Four diffusion models were developed. They are:

* Droplet Vaporization

• Droplet Thin Flame

• Coal Combustion

• Finite Rate Kinetics-Diffusion

Figure 6 illustrates the Finite Rate Kinetics Model. Preliminary

numerical experiments have been made with the coal combustion model.

Comparison of predicted fixed carbon and volatile matter loss with

time against measured values are given in Figure 7 . The two calcu-

lated curves were obtained with the upper and lower bounds of reaction

rate data available in the literature. The agreement appears to be

good.

3.1.3 One-Dimensional Limit-Case Solution from Two-Dimensional Codes

Both two-dimensional elliptic and parabolic formulations

contain the one-dimensional premixed flat flame as limiting cases.

The elliptic programs can be set up as one-dimensional by using only

one line of node points and specifying the lateral boundary conditions

as zero-shear (rather than zero velocity) and zero heat-flux. It

would be a worthwhile exercise to run the UARL CRISTY program in this

fashion as a numerical check-case and as an investigation of the

limitations of CRISTY imposed by the use of completely uncoupled
1-374
-------
\
M
4-1
DO
•H

I
O

0)
4-1
01
4J

'c
•H
N

•H
C
O
•H
M

tW
•H
n

4-1
P.
O
M
Q

&
•a
O
O
1-375
-------
V/////77///y/777//7,
00

o
^J-

o
o
o
o
oo
o

?o
c
o
o
01
to
•H

H
'5U31UOO
c
o
O
to

I
o
'/X///////////////////////
en
'O

o
u
01
01
e
o
o
•o
.u
cu
J«
o
00
•H
o
o
CO
o
c
JO
1-376
-------
chemical kinetics. As discussed below CRISTY relies on diffusion-

coupling to stabilize the stiff uncoupled chemical equations. If

this procedure works adequately (which seems unlikely for situations

involving complex chemistry) then the one-dimensional CRISTY calcula-

tions will be much faster than the PROF code.

A parabolic two-dimensional code can be set up to handle

the flat flame calculation by using the initial condition option

for planar free-shear-layer mixing. In this case the initial

conditions would be unburned pre-mixed air/fuel on one side and burned

pre-mixed on the other. The flame structure will develop out of this

discontinuity as the asymptotic solution for fully developed flow.

A small error is introduced since diffusion is allowed only on the

lateral-coordinate direction. This can be made arbitrarily small by

increasing the streaming velocity such that the flame lies flat

alongside the axis of flow. A fully implicit parabolic code will

yield flat flame solutions in the same computational time as a

one-dimensional code by rapid convergence on the asymptotic solution

through the use of large streamwise steps. Again such cross-check

calculations between different codes are useful as a verification

of sound numerical procedures.
1-377
-------
3.2 Chemical Kinetics and Modular Modeling Codes

3.2.1 The Ultrasys terns Kinetics Code

The IKAP kinetics program is capable of treating large

chemical systems in stirred reactors or generalized plug flow

situations. It is based on a chemical kinetics analysis procedure

developed at TRU Systems by Tyson which, through fully implicit

integration, allows the rapid solution of complex chemical systems.

A principal feature of the program is its ability to determine which

reactions dominate in the production or destruction of a particular

species under various reactor conditions dictated, for example, by

residence times, heat transfer and mixing histories. This procedure

known as "reaction screening" has been used in the course of esta-

blishing a definitive reaction set for the prediction of NO production

and destruction in combustion devices. Figure 8 shows a comparison

between prediction from this set and the experimental stirred reactor

results obtained by Bartok, et al at the Exxon Laboratories.

3.2.2 Modular Modeling Programs

The modeling of chemical processes by a collection of well -

stirred and plug-flow chemical reactors has been a standard chemical
engineering approach for many years
(12)
During the last five years
the practice has been extended to the analysis of combustion devices

by a number of investigators ' ' . Ultrasystems has, to a

limited degree, automated the coupling of elemental kinetics modules
1-378
-------
cc
<
o
t—1
05
H
W

O
i—i
W
o

o
E-i
CO
H
s
w
o

s
O
H
CO
D
OQ
2
o
o
p
w
CO

2;
t—»
CO
C/3
CO
w
00
00
•H
(aannsvaw sv
1-379
-------
to form an integrated engineering analysis code referred to as

MKAP.

The basic coupling arrangements are shown in Figures 9 and 10 .

As indicated these are represented by series, feedback, and parallel

coupling arrays. The most time consuming are the feedback elements

which require an iterative solution procedure. The estimate of

exchange rates between elements requires an independent fluid

mechanical analysis or direct experimental evidence from flow field

measurements or stimulus-response tracer studies.

MKAP has been applied to conventional gas turbine combustors

as illustrated in Figure 11 . Here the recirculation zone which

stabilizes the primary is modeled with a stirred reactor. This in

turn is coupled in a feedback fashion to a plug flow simulation of the

streaming flow around the recirculation zone. The plug flow element

receives combustion air from the swirler as well as from the primary

stabilization ports. Fuel enters both the SR and the PFR. The

remainder of the system is simulated as shown by a PFR with distributed

air addition.

MKAP has also been used in a study of gas turbine combustors

utilizing low-Btu coal derived gas. In this instance the secondary

combustion stage diffusion flame was modeled as shown in the upper

sketch of Figure 10 . The thin flame zone was represented by short

residence time well stirred zones with near stoichiometric equivalence

ratios. The residence times were determined by estimates of the flame
1-380
-------
a
E-t
U
w
o
o
w
H
w
CO
to
•H
W
I

(I)
£

M
O
U-l
01

o
•H
10
a
«
m
at
00

•H
1-381
-------
FUEL —
PFR.
SR:
AIR
PFR,,
1
1
ft

I
TL.
1
J7~
ft
i

•1
i i
i
PFl

1
PI

1 '
PI

1
I
B
*1
t
"R«
2
' I '
TR
R3
> I. i
i i
!

;
'

i i
i i

i
1

,
9
^

u
i •-

V *
^

Fig.UO Examples of Parallel Coupled Elements for Modular Analysis
1-382
-------
X
w
CO
•H

1
H

W
:> .

\
at
CO
1

PS
CO
IX

§
u
oo
•H
1-383
-------
zone volume and diffusion flame mixing rate. The flame was divided

into three axial zones (only two are shown in the sketch). The products

of combustion from the flame zone were returned to both the fuel/

products PFR and the air/products PFR for further chemical reaction.
3.3
Two-Dimensional Codes
3.3.1 Parabolic

Numerous boundary-layer and free-shear-layer parabolic

two-dimensional programs have been developed for chemically reacting

flows. Aerotherm, GASL, AeroChem, Ultrasystems, Sandia, TRW Systems,

Avco, General Electric, The Rocket Propulsion Establishment in

England, and Spalding's group at Imperial College have all, for

example, developed such computational methods. These programs have

resulted from the investigation of: reentry vehicle flow fields,

rocket plume afterburning, ramjet and afterburner propulsion, and

chemical lasers. In general the existing programs suffer from a variety

of problems which may include: excessive computation times, inability

to handle thin high energy release rate flame zones, and inadequate

description of turbulent transport. The best available programs for

free-shear layer computations, known to the author, are those

developed at AeroChem by Mikatarian, Kau, and Pergament and at

the Rocket Propulsion Establishment in England by Jensen and Wilson

Partially implicit integration techniques are employed and considerable

attention is paid to the description of turbulent transport. Jensen
1-384
-------
and Wilson, however, make a common error in assuming that because

simple mixing-length concepts are inadequate to describe the distri-

bution of equilibrium turbulence the situation necessarily calls for

a "two-equation" nonequilibrium model.

3.3.2 Two-pimensional El1iptic Programs

In the area of combustion device modeling the principal contri-

butions to the two-dimensional numerical solutions of the full elliptic

Navier-Stokes equations have come from the Imperial College

group(18~22) under Spalding and the United Aircraft group under

McDonald^23~25\ in addition several investigators have used

the Gosman, et al code in a variety of numerical calcula-

te,27)
tions

In many ways the two groups, having proceeded down separate

paths in the course of developing their respective methodologies,

are now at a point where the programs share many common features.

Both programs are applicable to furnace situations which are:

• Roughly cylindrical in shape as, for example, would
be the case for a typical industrial boiler.

• Aerodynatnically controlled by the burner design as
distinct from furnace designs which have discrete
staging (air or fuel) along the length of the
combustor which in turn introduce three-dimensional
effects.

Both programs have essentially the same embedded physics and chemistry.

Turbulent transport is predicted through the use of a two-equation non-

equilibrium model. Radiant heat transfer is predicted by means of a
1-385
-------
four-flux formulation. Two-phase flow is modeled on the assumption

of zero velocity and temperature lag and the particles are assumed

to participate in turbulent transfer to the same extent the gas phase

does. Finite rate chemical kinetics is allowed in both models.

Various versions of the Gosman-type programs have appeared with

different dependent variables, i.e., vorticity/stream function

variables or primitive velocity and pressure variables. Contrary to

the arguments that are generated by this issue there appears to be

little sound basis for the choice of one formulation over the other.

3.3.2.1 Numerical Procedures

It is in the numerical methods where the two programs

show the greatest difference in Character. Yet even here there are

definite signs of eventual convergence to the same general approach.

Both programs have gone to an ADI type of alternating

direction line relaxation procedure in order to increase the value of

the critical cell Reynolds number. At this point, however, the

programs differ in the iteration technique whereby the ADI procedure

is brought to convergence. The UARL code uses a modification of the

Peaceman and Rachford acceleration parameter to achieve optimum

convergence for a given cell Reynolds number and in this regard has

a definite advantage over the Gosman code. This advantage is reflected

in two related ways. First, the Gosman procedure apparently still
1-386
-------
requires the use of partial upwind differencing in order to maintain

stability. Secondly, the optimum ADI convergence method used in

CRISTY will yield shorter computation times.

Both the Gosman and UARL programs uncouple all of chemical

kinetics equations in the ADI iteration procedure. This relies on

the stabilizing effect of the diffusion coupling to maintain stability

of the kinetics integration. It is unlikely that this procedure

will work for complex chemistry and the equations will necessarily

have to be coupled thus increasing the computation times by a large

factor. The ratio of computational times between uncoupled and

coupled systems of equations goes roughly as the square of the number

of equations.

Both codes are forced by computation time consideration to use

a relatively coarse node point mesh as illustrated in Figure 12. For

the ADI procedure the computation time is proportional to the number

of node points provided the number of iterations required for conver-

gence remain constant,

3.3.2.2 Computational Results

Unfortunately these codes have had little opportunity

to be adequately tested against reliable furnace data. Nor has there

been the opportunity to perform the necessary numerical experiment to

establish the solution dependence of various physical and numerical

parameters. Some comparisons have been made between the UARL FREP
r-387
-------
O H
CM
60
1-388
-------
Code (predecessor to CRISTY) and cold swirling turbulent flow

(25)
data . For low to moderate degrees of swirl the comparisons

showed good agreement based on calculations using a sample Prandtl

mixing length hypothesis. For higher swirl levels which yielded

internal recirculation the simple turbulence model still appeared

to give reasonable agreement with the data.

(22)
Gosman and Lockwood have reported some comparisons between

calculated behavior and the gas fired "Ml Trials" carried out at

the International Flame Research Foundation at Ijrauiden. A partially

burned gaseous fuel-air mixture enters a square cross section furnace

along the centerline. Additional air enters in a concentric duct

located around the centerline duct. Figure 13 shows comparisons

between predicted and measured temperature profiles and streamline

patterns. The agreement can only be considered fair.

3.4 Three-Dimensional Program

3.4.1 Elliptic

( 2.8)
Both the United Aircraft Research Laboratories and

Imperial College groups are developing three-dimensional elliptic

programs for the stated purpose of analyzing combustion devices. In

the author's opinion, the practical application of this type of

program will be at least a decade off and will require dramatic

advances in computer capability.
1-389
-------
Fig.13a. Comparison of the Predicted and Measured
Streamline Patterns - from Gosman and

Lockwood^22'
1800
1 700
1600

3 1 SOD
a
M
(U

I1 moo
0)
H

1300
— Measured

Predicted
0 0.2 O.it 0.6 0.6 1.0

r (m)
Wall
Fig. .13b. Comparison of Predicted and Measured Radial

Temperature Profiles - from Gosman and
Lockwood'22)
1-390
-------
3.4.2 Parabolic

As mentioned earlier this type of program holds promise for

those three-dimensional situations in which no upstream influence

exists and in which the mixing regions lie along the coordinate

lines. The turbulent transport laws for two-dimensional boundary

and free shear layers can probably be generalized with some degree of

confidence to this type of situation. The numerical problems are

comparable to those encountered in two-dimensional elliptic problems.

Hence if a satisfactory resolution is found in the two-dimensional

case it should carry over to the three-dimensional situations.

Several three-dimensional laminar boundary layer programs

have been developed over the last ten years. With straightforward

modifications to the boundary conditions these codes can be immediately

generalized to the three-dimensional combustor containing free-mixing

layers. Patankar and Spaulding J have made such a modification and

illustrate its use by applying the method to the developing laminar

flow and heat transfer in a duct of square cross section with a

laterally-moving wall as shown in Figure 14' . The asymptotic cross

flow solution obtained when the flow becomes fully developed is

identical to the two-dimensional flow in a square cavity with a

moving wall. Patankar and Spalding compare their fully developed

solution with those obtained by others for the two-dimensional

case, (see Figure 14 )
1-391
-------
§ -0

««s
(-10°
00-H i
00-0 "S
fc 0> «
O M M
« PL, fe
CM

CM
•

1
O
I
O o\ oo
• • .
«-• o o
^o "^
• •
o o
r- ^
N-o.
-*
•
o
CM
•

o
o

o
4-1
o
o
a §
efl -H
0) 00
M 4)
4J oi
CO
I -O
M 0)
10 a
o o
1-1 r-l
O V

CO 01
o
y-l I
O Ps
•H

-------
3.5 Data Analysis Codes

3.5.1 Deduction of Turbulent Transport and Chemical Source

Distributions From Mean Flow Measurements

In recent years a considerable body of experimental data has

been obtained on the time averaged temperature, velocity and concen-

tration fields in various combustion devices. This experimental data

does not generally lead to an intuitive understanding of the pheno-

mena of turbulent mixing, heat release and the sources and sinks

of trace species such as pollutants. Latent in the measured maps is

a great deal of information regarding time-averaged turbulent trans-

port of species, momentum, and energy. This information along with

a description of the species source strength distributions can be

extracted through the integration of the conservation principles

of mass momentum and energy using the measured mean flow distributions.

(49)
Sadakata and Be4r have recently utilized a program of this

type to calculate local NO formation rates in swirling turbulent

methane-air flames. The NO formation rates in the reaction zone were

found to be 3 to 5 times higher than those in the post flame region.

Work is presently under way at Ultrasysterns to develop such a program

to evaluate flame data generated by the IFRF, IGT and others.
3.6
Radiative Heat Transfer
In any enclosure where the temperature and distribution of

radiating species is known the prediction of radiation exchange to all
1-393
-------
points within that enclosure can be considered in two parts:

(1) Evaluation of radiative exchange between all points
within the enclosure if the absorption, emission and
scattering characteristics of the material within the
enclosure and its bounding surfaces are known.

(2) Evaluation of the absorption, emission and scattering
characteristics from a knowledge of the temperature
and concentration of radiating species.

Theoretically an accurate solution of radiative exchange within any

enclosure, at equilibrium, can be obtained by summation over the

spectrum of the multiple reflection, absorption and scattering of an

infinite number of emitted monochromatic beams. However, spectral

information is not readily available for real solid surfaces '

and only limited information is available concerning emission and

absorption characteristics of flame gases which may contain solid
(31 32)
particles * . Consequently these limitations plus those asso-

ciated with computational capabilities necessitated the development

of different techniques for the evaluation of radiative exchange.

The various methods available for the evaluation of radiant

heat exchange in combustors have recently been reviewed by Bartelds,

(32)
et al amongst which are:

• The Monte Carlo Method. This method includes a statistical
approach to radiative exchange processes in which
Monte Carlo sampling techniques are used to examine the
action of small bundles of energy, rather than attempting
to simultaneously solve for the behavior of all the
energy involved in the process . If other numerical
techniques are available to solve the problem under
consideration HowellW does not recommend the use
of Monte Carlo methods.
1-394
-------
Zone Method ' . In this method the enclosure is divided
into several gas and surface zones and exchange between
each zone pair is evaluated from the radiative transfer
equation. The method has been successfully applied to
furnaces(37,38,39) where acceptable predictions of heat
flux to the walls and heat sinks have been reported.
However, predictions of gas temperature profiles within
the enclosure are not good. This can be attributed to
coarse zoning or incorrect allowance for the emission/
attenuation characteristics of the flame gases. In
principle, the zone method is not limited to coarse
zoning, computation times normally limit the number of
of zones making it unacceptable for application to
problems requiring definition of steep temperature
gradients.

Electrical Analogue Methods. Oppenheim developed
a technique in which the radiative quantities (emission,
heat flux, etc.) are replaced by analogeous electrical
quantities (potential current, etc.) in a network. The
method has not been used extensively although advances
in hybrid computing techniques could encourage further
development and use.
Flux Methods
(39,41,42)
In flux methods, the solid angle
of 4ir steradians surrounding a point is divided into a
number of solid angles in which the intensity of radiation
is assumed to be a specific function. The exact integro-
differential equations of radiative transfer are replaced
by approximate differential ones and can be easily solved
by standard finite difference techniques. Consequently
flux methods are ideally suited for coupling with^he
elliptic or parabolic flow field analysis programs
discussed earlier. Results of this type of coupling
have been presented by Gosman and Lockwood^-* and
Lowes, et al^"-^. Bartelds(32) has recently developed
two and four flux models which can be conveniently
coupled with either physical or mathematical models
to allow furnace performance to be predicted. Bartelds
has increased the accuracy of these flux models by
considering the intensity distribution around a point
to be a continuous function rather than being constant
over a given solid angle.
1-395
-------
3.7 Spray Combustion Programs

3.7.1 Qne-Dimens ional

Rocketdyne and Ultrasystems have developed one-dimensional

spray combustion programs under sponsorship by the Air Force and NASA.

These programs consider the dynamics, evaporation, and flame-mantle

burning of liquid fuels. The gas phase is assumed to be in chemical

equilibrium. Whether the droplet is burning with a flame mantle or

evaporating with the vapor burning in the wake is determined by a

criteria which is a function of the droplet relative Reynolds number

and the thermodynamic state of the gas phase. The process rates are,

of course, also dependent on these same variables. The droplet

velocity lag and hence Reynolds number is dictated by the acceleration

of the gas phase. The programs accept, as initial conditions, the

droplet size distribution and initial injection velocities.

The codes have been applied in combustion systems analyses to

determine approximate droplet lifetimes and the effective fuel source

distribution for use in more complex nonequilibrium gas phase chemistry

calculations. In those cases where the droplet spray initially

crosses streamlines an independent estimate must be made of the

degree of droplet consumption before they become "locked" into stream-

tubes. The initial distribution of drops by size and quantity as a

function of streamtube location must then be specified.
1-396
-------
3.7.2 Two-Dimensional Parabolic Codes

Considerable work has been done, under Department of Defense

(44-46)
support, by Smoot and his co-workers at Brigham Young University

in the development of parabolic two-dimensional, two-phase combustion

programs. The particle dynamics have been treated for a distribution

of particle sizes without the assumption of zero velocity or zero

temperature lag. For small particles these equations have the same

"stiff" behavior as near equilibrium chemistry. In general it has been

assumed that the particles respond to the mean motion only. Some

consideration has been given to the turbulent particle diffusion in

those cases for which the particles are small and eddy scale is large.

To date there has been no application of this type of two-dimensional

analysis to practical industrial furnaces.

3.7.3 Two-Dimensional Elliptic Codes

Spalding and McDonald, et al have employed a relatively

simple model to predict the behavior of droplet clouds in two-phase

combustion systems governed by two-dimensional elliptic behavior.

These models have been incorporated into the Imperial College and UARL

codes. The basic assumption is, that from a transport point of view,

each size class of particles can be considered to behave as

another gas phase species. These particle "species" are taken to have

infinite "molecular weight" and hence do not contribute to the thermo-

dynamic pressure. The particles are assumed to enter into convective

motion with zero velocity lag and exhibit the same turbulent transport
1-397
-------
behavior as the gas phase. The net source (or sink) strength of each

droplet size group is determined by a droplet evaporation or burning

law. This mass loss behavior depends on local conditions as described

in 3.7.1 above.

The shortcomings of such a model are certainly obvious and yet

the numerical complexities of the model described in 3.7.2 combined with

the elliptic nature of the flow make the resolution of this problem

formidable. In the limiting situation of large particles and small

eddy scale the assumptions made above break down completely. That is,

the velocity lags are substantial and relative Reynolds number effects

dominate the evaporation or burning rates. Also under these circum-

stances the particle paths will not necessarily coincide with mean flow

lines and the particles will tend not to participate in the fluctuating

eddy motion.

Perhaps the greatest drawback of this type of analysis is its

inability to handle the initial behavior of a liquid fuel spray as it

penetrates across stream lines. In this sense the model is similar to

the one-dimensional methods described in 3.7.1 and requires similar

auxiliary treatment.

3.8 Engineering Models o_f__Furnace_ Behavior

The ultimate test of the value of any model of a combustion

device is the degree to which it aids in the construction of such a

device. Since furnaces have been built without the aid of models for

some time the accuracy of any useful model must be at least equivalent
1-398
-------
to the empirical based design standards it was developed to supplant.

Due to the complex nature of the problem useful engineering models

will almost certainly contain adjustable empirical constants which

restrict the model to particular furnace types. An excellent example

of this type of model is that developed by Beuters, et al at

Combustion Engineering.

The analysis of the heat flux distribution and nitric oxide

formation in a tangential fired boiler represents a problem of not

trivial proportions. Figure 15 presents a schematic of a tangential

fired boiler. Beuters has described in detail the development of

a program to predict the performance of the lower furnace. The furnace

is divided into a series of horizontal strips with an arbitrarily

assigned heat release distribution in the firing gone. The products

of combustion "created" in each slice are directly proportional to

the fraction of heat release and to the recirculation or backmix flow.

Thus the flow originates in the heat release zone, circulates through

the hopper region and "peels-off" to rejoin the exit stream (see

Figure 15 ). The empirical factors included in the model have been

established by extensive field testing. The gas emissivity factor

(F ), for example, takes account of the emission/absorption character-

istics of the combustion gases and is fuel dependent. It also

contains a correction for convective heat transfer. The lower
1-399
-------
turn otra
— <••
~ ^
— -*

— -••
— -*
— -*

— -»•
— * _^_
^ •^
zy

3K02 aSV3T3H XV3H
— --^
— ^
-^-
I

— ^*-
— -^
•^
1

— -*-
— -^
— ^*-
^»-
1

— •"••
— -^
— -^
-^-

3N07

— <^-
- -*
— -«-
1

NOUV

— -^«-
— -^r-
1

inoai
r
-------
furnace model has been modified to allow it to predict NO formation

based upon the Zeldovich mechanism. The success of this empirical

engineering model can be judged from the comparisons between

measured and predicted wall absorption rates and NO emissions for

oil fired utility boilers shown in Figure 16 . The Bueters model

has been included as one example of several empirical models which

can be successfully used to predict furnace behavior over restricted

ranges.
4.0 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE ACTIVITIES
RELATED TO MATHEMATICAL MODELING

From the "catalogue" of Section 3 it is certainly clear

that a great deal of thoughtful work has gone into the development

of mathematical models in recent years. It would be premature to

judge the product of these efforts completely on the basis of what

has been accomplished to date in the way of successful engineering

application. Such a judgement would necessarily yield a low grade

since most of the tools are still in the development stage and there

has not been the opportunity to adequately test them against

reliable measurements.

Nevertheless a few general conclusions can be reached regarding

the achievements. The advancement in numerical analysis techniques

should be applauded. The work at United Aircraft on the improvement
1-401
-------
40 80 120 160 200 240 280
NO Measured
x
Nitric Oxide Emissions - Predicted vs Measured
Values on Oil-Fired Units

-'o
9) ^
(3 *
G fsj
O 4J
4J
O •~-
M 9
t§ M
H
tH
-------
in numerical convergence techniques for the elliptic Navier-Stokes

equations without the need to artificially introduce significant

errors is a good example. The large increase in the number and type

of models available today for both numerical experimentation and those

with which to attempt engineering analysis tasks represent a signifi-

cant contribution. The success of numerical manipulations of large

sets of chemical reactions in helping to sort out the significant

chemistry of pollutant formation is a notable accomplishment. Related

to this is the ability, in certain cases, to establish the lower bounds

on pollutant formation that are set by chemical considerations.

The single most important area that needs further investigation

in order to dramatically improve our ability to predict combustor

behavior is the physics of turbulent transport. Without a knowledge

of this fuel/air contacting behavior on the "macro" scale there is

little that can be accomplished with a more refined knowlege of the

other physical and chemical aspects of combustion. In this regard

several well designed experimental programs could be of great help.

These would be designed to obtain the following information:
Detailed mean flow and fluctuation measurements
in types of flow regimes which are similar in
character to the distinguishable regions of
practical combustion devices. This would provide
an empirical coupling between the turbulence level
and the kinematics of the mean flow.
1-403
-------
* Stimulus-response measurements with inert tracers
introduced into various regions of typical
combustion devices - this would provide gross
coupling parameters for use in modular modeling.

• Determination of nonequilibrium turbulence parameters
by stimulus-response measurements using a distur-
bance of the inflow turbulence level as the stimulus
and measuring the decay of this disturbance within
the system.

The second category of experiments could lead to a significant

body of empirical data upon which to build a sound modular modeling

capability. In light of the difficulties in obtaining a more

detailed understanding of turbulent transport this empirical approach

may prove to be the most promising of all. Possibly second only to

maeromixing in importance is the attainment of a better understanding of

turbulence/kinetics coupling. An experiment is called for in which

the eddy breakdown time can be varied for a well defined nonpremixed

flow using reactants which have a simple and well understood

chemistry.

Since the two-dimensional elliptic modeling techniques have

significant potential for assisting in the design process, if the

physics can be improved, there is considerable motivation to improve

the numerical techniques. Some areas which deserve further investiga-

tion are:

• Automatic mesh control where grid size is linked to
the gradients in the system

• Combined ADI iterations and iterations to "update"
the nonlinear coefficients
1-404
-------
• Further research to improve the ADI convergence rate
for a given cell Reynolds number

There is a need for the smallest possible chemical reaction sets

for use in complex fluid mechanical codes. Numerical experiments

using simple stirred and plug flow reactors should be conducted to

sort out these "reduced" sets. Improvements in "reaction screening"

procedures are needed to accomplish this and to identify reaction

rates which need further definition.
1-405
-------
REFERENCES
1. Tennekes, H. and Lumley, J.L., A First Course in Turbulence.
The MIT Press, 1972.

2. Beer, J.M., and Chigier, N.A., Combustion Aerodynamics,
John Wiley & Sons, Inc., 1972.

3. McCreath, C.G., and Chigier, N.A., Fourteenth Symposium
(International) on Combustion, p. 1355, The Combustion
Institute, 1972.

4. Curtiss, C.F., and Hirschfelder, J.O., "Integration of Stiff
Equations," Proceedings of the National Aeronautical Society,
Vol. 38, pp. 235-243, 1952.

5. Tyson, T.J., "An Implicit Integration Method for Chemical
Kinetics," TRW Systems Report 9840-6002-RUOOO, Sept. 1964.

6. Peaceman, D.W., and Rachford, H.H., Jr., Journal of the Society
of Industrial and Applied Mathematics, Vol. 3, 1965.

7. Briley, W.R., "A Numerical Study of Laminar Separation Bubbles
Using the Navier-Stokes Equations," Report J110614-1, United
Aircraft Research Laboratories, East Hartford, Conn., 1970.

8. Kendall, R.M., and Kelly, J.T., "Premixed One-Dimensional Flame
Code (PROF) - Its Formulation, Manipulation, and Evaluation,"
Aerotherm TR-75-158, July 1975.

9. Merryman, E.L., and Levy, A., Fifteenth Symposium (International)
on Combustion, p. 1073, The Combustion Institute, 1974.

10. Nurick, W.H., Private communication.

11. Bartok, W., Engleman, V.S., and del Valle, E.G., "Laboratory
Studies and Mathematical Modeling of NOX Formation in Combustion
Processes," ESSO Research and Engineering Co., Report GRU.3GNOS.71,

12. Levenspiel, 0., Chemical Reaction Engineering, Second Edition,
John Wiley & Sons, Inc.

13. Mellor, A.M., "Current Kinetic Modeling Techniques for Continuous
Flow Combustors," presented at the Symposium on Emissions from
Continuous Combustion Systems, pp. 23^55, September 1971.
1-406
-------
14. Hammond, B.C., Jr., and Mellor, A.M., Combustion Science and
Technology, Vol. 4, pp. 101-112, 1971.

15. Swithenbank, J., Poll, I., Vincent, M.W., and Wright, D.D.,
Fourteenth Symposium (International) on Combustion, p. 627,
The Combustion Institute, 1972.

16. Mikatarian, R.R., Kau, C.J., and Pergament, H.S., "A Fast
Computer Program for Nonequilibrium Rocket Plume Predictions,"
AFRPL-TR-72-94, 1972.

17. Jensen, D.E., and Wilson, A.S., Combustion and Flame, 25, pp. 43-
55, 1975.

18. Runchal, A.K., and Wolfshtein, M., Journal of Mechanical
Engineering Science, Vol. 11, No. 5, 1969.

19. Gosman, A.D., Pun, W.M., Runchal, A.K., Spalding, D.B., and
Wolfshtein, M., "Heat and Mass Transfer in Recirculating Flows,"
Academic Press, New York, New York, 1969.

20. Spalding, D.B., "Mathematical Models of Continuous Combustion,"
p. 23, Emissions From Continuous Combustion Systems, Proceedings
of the Symposium on Emissions from Continuous Combustion Systems,
September 1971, Plenum Press, New York, London.

21. Khalil, E.E., and Whitelaw, J.H., "The Calculation of Local-Flow
Properties in Two-Dimensional Furnaces," Imperial College of
Science and Technology, HTS/74/38, November 1974,

22. Gosman, A.D., and Lockwood, F.C., Fourteenth Symposium (Inter-
national) on Combustion, p. 661, The Combustion Institute, 1972.

23. McDonald, H., et al., Draft Final Report on Two-Dimensional
Elliptic Program for Combustor Flow Field Analysis (CRISTY) -
Developed under EPA Contract - August 1975.

24. Anasoulis, R.F., McDonald, H., and.Buggein, R.C., "Development
of a Combustor Flow Analysis, Part I: Theoretical Studies,"
AFAPL-TR-73-98, Technical Report, January 1974.

25. Anasoulis, R.F., and McDonald, H., "A Study of Combustor Flow
Computations and Comparison with Experiment," EPA Report
EPA-650/2-73-045, December 1973.

26. Quan, V., Bodeen, C.A., and Teixeira, D.P., Combustion Science
and Technology, _7, pp. 65-75, 1973.
1-407
-------
27. Samuelsen, G.S., and Peck, R.E., "Pollutant Formation in
Reacting Flows with Recirculation," Paper presented at Western
States Section, The Combustion Institute, October 1972.

28. Gibeling, H.J., McDonald, H., and Briley, W.R., "Development
of a Three-Dimensional Combustor Flow Analysis," Volume I:
Technical Report, United Aircraft Research Laboratories,
R75-911850-18, July 1975.

29. Patankar, S.V., and Spalding, D.B., Fourteenth Symposium
(International) on Combustion, p. 605, The Combustion
Institute, 1972.

30. Hottel, B.C., and Sarofim, A.F., Radiative Transfer, McGraw-
Hill, New York, 1967.

31. Sparrow, E.M., and Cess, R.D., Radiation Heat Transfer, Brooks/
Cole, Beltnont, USA, 1970.

32. Lowes, T.M. and Heap, M.P., Emission/Attenuation Coefficients
of Luminous Radiation, Second Members Conference, I.F.R.F.,
Umuiden, 1971.

33. Bartelds, H., Heap, M.P., and Lowes, T.M., Prediction of
Radiative Exchange Within Furnace Enclosures, Vol. II, Final
Report, EPA Contract 68-02-0202.

34. Howell, J.R., Application of Monte Carlo to Heat Transfer
Problems, Advances in Heat Transfer, ^, p. 1-54, Academic
Press, New York, 1968.

35. Hottel, H.C., and Cohen, E.S., Radiant Heat Exchange in a Gas-
Filled Enclosure: Allowance for Non-uniformity of Gastempera-
ture, AIChE Journal, 4^ p. 3-24, 1958.

36. Hotell, H.C., and Sarofim, A.F., Theory and Fundamental Research
in Heat Transfer (ed. J.A. Clark), p. 139-160, Pergamon Press,
New York, 1963.

37. Johnson, T.R., and Beer, J.M., Fourteenth Symposium (International)
on Combustion, p. 639, The Combustion Institute, 1973.

38. Latsch, R., Mathematisches Modell fur eine turbulente Diffusions-
flamme und deren zylindrischen Brennraum, Dissertation,
University Karlsruhe, 1972.
I- 408
-------
39. Lowes, T.M., Bartelds, H., Heap, M.P., Michelfelder, S., and
Pal, B.R., Prediction of Furance Heat Transfer, I.F.R.F.
Document K20/a/72, August 1974.

40. Oppenheim, A.K., Trans. A.S.M.E., 78, p. 725-735, 1956.

41. Viskanta, R., Radiative Transfer and Interaction of Convection
with Radiation Heat Transfer, Advances in Heat Transfer (eds.
Irvine, T.F. and Hartnett, J.P.), ^, p. 175-252, Academic Press,
New York, 1966.

42. Siddall, R.G., Flux Methods for the Analysis of Radiant Heat
Transfer, Fourth Symposium on Flames and Industry, British
Flame Research Committee I.F.R.F. and Inst. of Fuel, London,
1972.

43. Gosman, A.D., Lockwood, F.C., Fourteenth Symposium (International)
on Combustion, p. 661, The Combustion Institute, 1973.

44. Smoot, D.L., and Allred, L.D., "Particle-Gas Mixing in Air-
Breathing Ducts with Nonparallel Air Injection," AIAA Paper
No. 74-1158, AIAA/SAE 10th Propulsion Conference, October 1974.

45. Smoot, L.D., Douglas, R.A., Simonsen, J.M., and Tufts, L.N.,
"A Model for Mixing and Combustion of Compressible, Particle-
Laden Ducted Flows," Paper 70-736, AIAA Conference, 1970.

46. Hedman, P.O., "Particle-Gas Dispersion Effects," Ph.D.
Dissertation, Brigham Young University, Provo, Utah, August
1973.

47. Bueters, K.A., and Habelt, W.W., "NOX Emissions from Tangentially
Fired Utility Boilers - A Two Part Paper," Part I - Theory,
Presented at 66th Annual AIChE Meeting, November 1973.

48. Bueters, K.A., Cogoli, J.G., and Habelt, W.W., Fifteenth
Symposium (International) on Combustion, p. 1245, The Com-
bustion Institute, 1974.

49. Sadakata, M., and Be6r, J.M., "Spatial Distribution of Nitric
Oxide Formation Rates in a Swirling Turbulent Methane-Air
Flames," The Combustion Institute European Combustion Symposium,
1975.
1-409
-------
4:25 p.m.
The Mathematical Modeling of Combustion Devices
Thomas J. Tyson, Ultrasystems, Inc.
Q: I should like to elicit a comment from Dr. Tyson concerning
the use of time-smoothed equations of motion in large
furnaces containing thin flame zones.

A: The question raised is a major area requiring further
clarification, and is related to the questions of macroscale
turbulent transport; particle ballistics, evaporation, and
combustion in turbulent flows; and turbulence/chemistry
coupling. Large furnaces have less confined flames where the
gross motion of the mixing zones can be influenced by the
large scale low frequency turbulent motion of the furnace as
a whole. Time averaging on this scale undoubtedly is question-
able. Nevertheless, I think the macromizing associated with
fuel/air contacting on the scale of the active flame zone can
be predicted by time-averaged behavior even though the precise
location of the flame is uncertain. With regard to the thin
flame zones, I don't believe this is a more critical issue
than for smaller-sized combustors. During the time that
fuel and air are initially entwined in a large eddy the
opportunity for molecular diffusion is small even in small
devices. Chemical action awaits the cascading breakdown
process to smaller eddies. Similarly, the "hold up" period
after hot products become entwined with cooler zones but
still continue to be chemically active is dictated by the
eddy breakdown rate. Since this breakdown time is longer
in large furnaces and the quenching is more severe following
breakdown, the issue of turbulence kinetics coupling is
indeed more important in large furnaces; hence the time-
averaged equations should come under greater scrutiny.
1-410
-------
I believe the resolution to this, however, will not be to
abandon the time-averaged formulation, but rather to develop
a phenomenological eddy "hold-up" model.
The situation of modeling flames often involves attention
to the local scale — reactions are occurring locally —
either in the shear layers or around droplets. In many
instances the modeler is trying to define the conditions
under which this local action is occurring -- be it NO
formation or soot formation. One interesting way to do
this and one which has met with some success, is to feign
ignorance of the total flow field and the exact distribution
of temperature and equivalence ratio. In many instances they
are too hard to define. Heywood and others have met with
some success by specifying a distribution of reaction zone
equivalence ratio.

I sympathize with this approach, which was suggested some
twenty years ago by Corrsin in describing the statistical
behavior of unmixedness decay in a turbulent field and
developed by Heywood for combustion devices. It stems from
a desire to simplify the task of predicting furnace
performance. However, I find it very difficult to know how
to do this. To employ this approach one needs to describe,
in some statistical fashion, the environment in which the
local chemical action is taking place. For example, the
statistical quantities defining the "background"
stoichiometry and temperature fields (say mean and
deviation from the mean parameters) must be modeled in some
1-411
-------
manner. How, for example, does the deviation from the mean
of overall fuel/air ratio decay with distance down a complex
combustion device? I would rather gamble on the development
of a mechanistic approach, based on a fluid mechanical
description of the device, to describe this behavior
rather than on an ad hoc statistical description. I
agree with the implication of your question that attention
may need to be focused on local scale high temperature
thin molecular diffusion layers within eddies or around
drops or droplet clouds. However, in considering the
importance of such zones in comparison with adjacent locally
well mixed regions, one must consider relative scales,
residence times, compositions, and temperatures. The
lifetime and contribution from thin molecular diffusion
zones within large scale eddies is, for example, dictated
by the eddy breakdown rate which is such that the diffusion
layer becomes of less significance as the combustor size
increases.
I- 412
-------
DEVELOPMENT OF FEDERAL STANDARDS OF

PERFORMANCE
By:

Stanley T. Cuffe

Environmental Protection Agency

Research Triangle Park, N.C.
1-413
-------
DEVELOPMENT OF FEDERAL STANDARDS OF

PERFORMANCE
I am pleased to have been invited to participate in this
Stationary Source Combustion Symposium. My comments will include a
summary of the procedures used to develop standards of performance,
typical problem areas we have encountered in the past, sources we
are presently working on, and new sources we will survey.

As many of you know, the authority of the U.S. Environmental
Protection Agency (EPA) to control the discharge of pollutants into
the atmosphere is provided by the Clean Air Act, as amended in 1970.

Section 112 of the Act directs the Administrator to issue
national emission standards for air pollutants which cause an increase
in mortality or an increase in a serious, irreversible, or incapacitating
reversible, illness. Standards for hazardous pollutant emissions are
designed to provide an ample margin of safety to protect public health.
Hazardous pollutant standards have been issued for 13 sources of
asbestos, beryllium and mercury, as noted in Table 1.

Most of our efforts in developing federal standards, however, have
been for sources covered under Section 111 of the Act. This section
requires the establishment of standards of performance for new
stationary sources of air pollution which "... may contribute signifi-
cantly to air pollution which causes or contributes to the endangerment
of public health or welfare." The Act requires that standards of
performance for such sources reflect "... the degree of emission
limitation achievable through the application of the best system of
emission reduction which (taking into account the cost of achieving
such reduction) the Administrator determines has been adequately
demonstrated."

During the past several years of developing new source performance
standards, EPA staff have generally used the following guidelines:

1. Source test data on existing well-controlled plants are the
most desirable basis for setting emission limits for new
plants. Most of these data are obtained from EPA source
tests on best controlled plants.

2. Interpretation of test results from any single best-controlled
plant must consider (a) representativeness of the plant tested
(feedstock, operation, size, age, etc.); (b) age and
maintenance schedule for the control equipment tested; and
(c) design uncertainties for the type of control equipment
being considered and the "safety factor" that must be used
to ensure meeting the emission standard.
1-414
-------
3. For sources where emerging technology is significant,
consideration is given to (a) test data from pilot and
prototype installations and application of reasonable
engineering judgment to these data; (b) vendor guarantees;
(c) existing design contracts; and (d) foreign technology.

4. Cost of control is related to the cost of the new industrial
installations and the economic impact on the industry, not
in reference to air quality improvements or the economic
benefits of such improvement.

5. Where possible, standards should be able to be met through
the use of more than one control technique or licensed
process. (Electrostatic precipitators or fabric filters can
meet standards for cement plants.)

6. Uhere possible, standards should encourage, or at least permit,
the use of process modifications or new processes in place of
add-on air pollution control systems, I.e. use of low sulfur
fuel.

7. Where possible, standards should allow the use of control
systems capable of controlling other pollutants (i.e., scrubbers
versus electrostatic precipitators on steam electric generators).

8. ^There possible standards should allow the use of control
systems that can minimize the impact on other aspects of the
environment.

9. Where appropriate, visibility standards are established which
are compatible with mass emission standards. A prime purpose
of this type of standard is to facilitate surveillance and
enforcement.

The timetable in the Clean Air Act requires that EPA first publish
a list of source categories for which it intends to establish performance
standards. Within 120 days after the list is published in the Federal
Register, the Agency must propose standards for those categories.
Within 90 days following their proposal, EPA must promulgate the
standards. As required by the Act, time is provided for interested
parties to comment on the proposed standards and for the Agency to
review and react to the comments prior to promulgation. The standards
are effective upon promulgation; however, a source is subject to the
standards if construction or modification is begun after the date the
applicable standards are proposed.

The procedures which are followed by EPA project officers or by
their contractors in preparing a standards support and environmental
impact document (SSEID) and in having the document reviewed prior to
1-415
-------
proposal of standards of performance in the Federal Register
include:

1. Available background data (particularly emission control
technology) is gathered by conducting literature searches,
and short term screening studies, and by meeting with
representatives of industry trade associations, industrial
plants, air pollution control agencies, and with vendors
of air pollution control equipment.

2. Field inspections are made of candidate plants which have
more efficient processes and/or emission control systems.

3. After an evaluation of the inspected facilities, specific
processes and/or emission control systems are selected for
source testing pollutants of concern. Emission measurements
are then made at selected facilities by EPA staff or their
contractors.

4. Letters are forwarded to owners or operators of specific
plants requesting information on performance of their
emission control systems and on the capital and annual
costs of these systems. In sending the letters, EPA
normally invokes Section 114 of the Clean Air Act which
requires that the owners or operators provide this
information to the Administrator of EPA. The letters are
usually a primary source of data on emission control costs
and a secondary source of emission data.

5. After the gathering of these data is completed, a standards
support and environmental impact document (SSEID) is written.
This report details the performance and costs of alternate
emission control systems and the rationale for the proposed
standard of performance. Also discussed are the environmental
effects of water pollution, solid waste, noise and particularly
the energy requirements if the proposed performance standard
were adopted. Up to this point, the gathering of data and
the writing of the SSEID will require between 12 and 16
months.

6. The SSEID then undergoes a series of reviews and proposal of
standards for a period of about one year before a final
standard of performance is promulgated.

Initial review is provided within EPA by a working group composed of
representatives from Enforcement, Research, Water Pollution, Solid
Waste, Economic Analysis, and the Office of General Counsel. Outside
of EPA, the proposed standards of performance are reviewed in detail
1-416
-------
by two advisory committees to EPA. The first of these, The National
Air Pollution Control Techniques Advisory Committee, has 16 members
representing industry, control equipment manufacturers, air pollution
control agencies, consultants specializing in air pollution control
and environmental groups. To assist the committee, several individuals
with specialized expertise related to the specific source categories
being considered are asked to attend the meetings as technical
consultants. Next a steering committee composed of Assistant
Administrators of each office within EPA reviews the SSEID. Prior
to proposal of performance standards in the Federal Register, the
document is then reviewed by other Federal agencies such as the
Departments of Commerce, Interior & Defense. This review is
coordinated by the President's Office of Management and Budget. After
proposal in the Federal Register, usually several hundred comments on
the proposed standards are received from interested industries, citizens
groups, control agencies, universities and research institutes. These
comments are made known to the Advisory Committee to obtain their
comments before the proposed standards become law.

During the data gathering phase of developing a standard, several
main types of problems are encountered which have resulted in extended
time schedules. One expensive and time consuming area is the locating
of candidate "best systems of emission reduction." This has been a
particular problem in the category of metallurgical processes. During
the past three years our project engineers have inspected ferroalloy
and by-product coke plants in Europe and Japan to gather information on
efficient particulate control systems. Inspections of copper smelters
in Europe were also made to assess efficient systems for control of
sulfur dioxide. Municipal incinerators equipped with efficient par-
ticulate collectors were also visited in Europe. For most other types
of industrial processes, however, we have been able to locate candidate
"best controlled plants" in the United States.

Although many well controlled sources are located in the United
States, the control systems are often not well maintained and operated.
Dense plumes of particulate emissions have been observed from fabric
filters and electrostatic precipitators of efficient design. Screens
on centrifugal pumps serving packed scrubbers have been partially
plugged as have spray heads within the scrubber. Proper maintenance
must be performed before emission measurements are made or else these
specific installations are not tested.

A second problem area is the lack of demonstrated control
technology for oxides of nitrogen and for hydrocarbons. As noted in
this symposium, control techniques for nitrogen oxides have been
applied to fossil fuel burning steam generators. However, these
techniques have apparently not been applied to fossil fuel fired kilns,
dryers, process heaters and incinerators. For the control of
hydrocarbons, particularly solvents, the most commonly used method of
I- 417
-------
control is an afterburner or incinerator where recovery of heat is
not practiced. Other types of control systems, which are not energy
intensive, have had limited use in controlling hydrocarbons.

The third problem area is a lack of standard methods of
measurement for emissions of atmospheric pollutants. Fugitive
emissions of gaseous and particulate pollutants from charging coal
into and pushing coke from by-produce coke ovens are not amenable
to conventional methods of emission measurement. Emissions from
coke oven charging holes, standpipes and covers, chuck and end doors
often occur over a period of several minutes and cannot reasonably
be captured to determine the mass rates of pollutant emissions. For
other industrial sources, particulate emissions from centrifugal
scrubbers and from screens serving grain dryers are usually not
amenable to isokinetic, traverse type sampling procedures. Also the
measurement of condensible hydrocarbons to determine particulate and
gaseous portions have been a problem in the measurement of emissions
from asphalt roofing plants. In another area, standard techniques
for measuring asbestos emissions from taconite beneficiation plants
are not available.

To date, standards of performance for new stationary sources have
been set for pollutants from 17 source categories (see Table 2).
The preparation of SSEID has been completed for 14 other sources
(see Table 3). Engineers in our Industrial Studies Branch are
presently gathering information for development of standards of
performance for 17 other sources (See Table 4). Stationary fossil
fuel combustion sources which are covered in this data gathering
phase include 1) combined coal-municipal refuse fired steam generators,
2} stationary internal combustion engines, 3) stationary gas turbines
and 4) intermediate size boilers.

For the combined firing of coal and municipal refuse, emission
measurement results from the Union Electric Power Plant (the only
commercial size operating unit) show some increase in emissions of
particulate when firing coal and refuse over that from burning coal
alone. As expected, however, no increase in emissions of nitrogen or
sulfur oxides was noted. A performance standard for particulate
emissions will therefore be proposed. The demonstrated control
technology for particulate is an electrostatic precipitator which
in order to maintain the same particulate emission limit will generally
be slightly larger than that for firing coal alone. A draft SSEID
has been written and has undergone review by the NAPCTAC. A standard
of performance for this source should become law by next summer.

For stationary internal combustion engines, emission measurement
results for nitrogen oxides (and also carbon monoxide and hydrocarbons)
are being obtained from natural gas and diesel oil-fired engines
tested in manufacturer's laboratories. The various control techniques
for nitrogen oxide which are used alone or in combination include:
1-418
-------
1. Turbo charging and reduced compression ratio
2. Spark retard
3. Changed air to fuel ratio
4. Increased engine speed
5. Modified injectors
6. Exhaust gas recirculation
7. Engine derating
8. Water injection

Many of these control techniques cause some increase in fuel consumption,
particularly retard and engine derating. A compromise between reduction
of NOX and increased fuel consumption will therefore be considered. Also,
because promulgated standards could not be implemented immediately by
engine manufacturers, time will be provided to incorporate control
techniques into their production engines. A draft SSEID should be
available by this coming December or January. Performance standards
for this source category should be forthcoming in about a year.

Regarding stationary gas turbines, the primary emissions when firing
oil or gas are nitrogen oxides and carbon monoxide. Sulfur dioxide can
also be a problem when firing oil. The nitrogen oxides can be controlled
by a) dry techniques such as external combustion, variable geometry
combustors, catalytic combustors and exhaust gas recirculation and b)
water or steam injection into the combustion chamber. Although sulfur
oxides can be controlled by using low sulfur fuels, in some cases,
desulfurization of oil will be required. Some of the problems which
have been encountered to date in developing proposed performance standards
are:

1. Lack of treated water at remote locations for treatment
of nitrogen oxides.

2. Availability of low sulfur fuel.

3. High content of bound nitrogen in the fuel.

4. Increased emissions of nitrogen oxides with increased
turbine efficiency.

A draft SSEID should be available by April 1976 while performance
standards for turbines should be promulgated about 10 months later.

A study of intermediate size boilers (smaller than 250 x 10
Btu/hr heat input) will be started by a contractor in about 3 months.
Information will be developed on the performance of control techniques
for particulates, nitrogen oxides, sulfur oxides, hydrocarbons, and
carbon monoxide. After a survey has been completed in about a year,
suitable types and sizes of boilers and specific pollutants will be
selected for development of standards of performance.
1-419
-------
Source categories involving the processing of fossil fuels
which are presently being surveyed include a) sulfur recovery plants
in petroleum refineries and in natural gas fields and b) the
gasification of coal. Pollutants of concern from the sulfur recovery
plants include sulfur dioxide (SCL) , hydrogen sulfide (H.~S) , carbon
disulfide (CS?) and carbonyl sulfide (CDS). Efficient control systems
for these pollutants include the Wellman Lord, Scott and Beavon. The
Wellman Lord system consists of incineration followed by scrubbing of
S0«. For the other two systems, sulfur compounds are converted to
H S and removed by scrubbing. Several of these processes have been
tested by EPA staff or contractors at petroleum refineries in the
United States. A Stretford system is scheduled for testing at a natural
gas field plant in Texas. A SSEID has been completed and is undergoing
review for control of sulfur plants in petroleum refineries. A draft
SSEID for control of sulfur compounds from natural gas fields should
be complete by April, 1976.

Conversely, there are no commercial size coal gasification plants
operating in the United States. To define basic components of exit gas
streams from lurgi coal gasification plants and the performance of
systems to recover sulfur compounds, the EPA project officer inspected
and obtained data from coal gasification plants in South Africa,
Germany and Scotland. The pollutants of primary concern are sulfur
oxides, sulfides and mercaptans, hydrocarbons and particulates. A
draft SSEID for coal gasification plants should be complete by
February, 1976.

Finally, the source categories which will be surveyed during the
next two years to determine the magnitude of emissions and the
availability of demonstrated control technology for the probable
development of standards of performance are:
1. Power plant boilers
>250 x 106 Btu/hr

2. Cement plants

3. Explosives

4. Steel foundries

5. Industrial surface
coatings

6. Charcoal manufacture
7. Ethylene oxide

8. Coal liquefaction

9. Industrial & Commercial
incineration

10. Fibers (Nylon & Acetate)

11. Plywood manufacture

12. Beer & whiskey processing
1-420
-------
Standards of performance were promulgated in December of 1971 for
pollutants from power plant boilers and cement plants. However, recent
information indicates that revised standards are needed for particulates
and nitrogen oxides from power plants and that standards should be
developed for nitrogen and sulfur oxides from cement kilns. Most of
the other industries, e.g. industrial surface coatings, charcoal
manufacture, ethylene oxide, coal liquefaction, nylon and acetate
fibers and beer and whiskey processing, are significant sources of
hydrocarbon emissions. Because of the widespread problem of photochemical
oxidants, which may extend into rural as well as urban areas, it is
apparent that maximum control of all new hydrocarbon sources nationwide
will be an important factor in controlling the oxidant problem.
1-421
-------
TABLE I

SUMMARY OF HAZARDOUS POLLUTANT STANDARDS
Pollutant
Affected Facility
Limitation
Mercury Mercury ore processing facilities
Mercury cell chlor-alkali plants

Beryllium Extraction plants, foundries
ceramic manufacturing plants
beryllium waste disposal
incinerators propellant plants,
machine shops processing alloys
with >5% Be.

Rocket testing facilities

Asbestos Asbestos mills, manufacturing
operations
Spraying of asbestos flreproofing
and insulation that contains
more than 1% asbestos on
buildings, structures, pipes and
conduits.

Spraying of asbestos fireproofing
and insulation that contains more
than 1% asbestos on equipment and
machinery.

Use of asbestos mill tailings on
roadways

Demolition operations
Not more than 2300 gm/day
for the entire facility.

Not more than 10 gm/day
(Option of meeting ambient
level of 0.01 ugm/m^ if
three years of ambient
data available) .
Limited to 75 ugm-min/m

No visible emissions or
use control equipment
meeting specific
performance characteristics

Banned
No visible emissions
Banned except on asbestos
ore deposits

Good control practices are
required
1-422
-------
TAjiLE 2

SUMMARY OF STANDARDS FOR NEW OR SUBSTANTIALLY MODIFIED SOURCES

1. Steam Generators [ >250 million Btu/hour heat input]

(a) Particulate Matter:
(1) 0.1 Ibs. per million Btu heat input (0.18 grams per
million calorie)
(2) No more than 20% opacity visible emissions, except for
two minutes in any hour visible emissions may be as
great as 40% opacity.

(b) Sulfur Dioxide:
(1) 0.8 Ibs. per million Btu heat input (1.4 grams per million
calorie) when oil is fired.
(2) 1.2 Ibs. per million Btu heat input (2.2 grams per million
calorie) when coal is fired.

(c) Nitrogen Oxides (as NO,.,) :
(1) 0.20 Ibs. per million Btu heat input (0.36 grams per
million calorie) when gas is fired.
(2) 0.30 Ibs. per million Btu heat input (0.54 grams per million
calorie) when oil is fired.
(3) 0.70 Ibs. per million Btu heat input (1.26 grams per million
calorie) when coal is fired.

2. Incinerators [>50 tons per day charging rate]

Particulate Matter:
0.08 grains per standard cubic feet corrected to 12% C0?
(0.18 grams/NM3) l

3. Portland Cement Plants

Particulate Matter:

(1) 0.30 Ibs. from the kiln per ton of feed to the kiln (0.15 kg
per metric ton of feed).
(2) 0.10 Ibs. from the clinker cooler per ton of feed to the kiln
(0.05 kg per metric ton of feed).
(3) No more than 10% opacity visible emission from kiln and cooler.
(4) Less than 10% opacity visible emission from all other
sources in the plant.
1-423
-------
4. Nitric Acid Plants

Nitrogen Oxide (as NO-):
(1) 3 Ibs. per ton 01 acid produced (1.5 kg per metric ton).
(2) Less than 10% opacity visible emissions.

5. Sulfuric Acid Plants

(a) Sulfur Oxide:
4 Ibs. per ton of acid produced (2 kg per metric ton).
(b) Acid Mist:
(1) 0.15 Ibs. per ton of acid produced (0.075 kg per metric ton)
(2) Less than 10% opacity visible emission.

6. Asphalt Concrete Plants

Particulate Matter:
(1) 0.04 grain per dry standard cubic foot (90 mg/dscm).
(2) Less than 20% opacity visible emissions.

7. Petroleum Refineries

(a) Particulate Matter
(1) 1.0 pound per 1000 pounds of coke burnoff in a FCC
catalyst regenerator.
(2) Less than 30% opacity except for 3 minutes in any one hour.
(b) Carbon Monoxide
0.05% by volume in gases from a FCC catalyst regenerator
(c) Sulfur Dioxide
No fuel gas shall be burned which contains H2S in excess of
0.10 gram per dry standard cubic foot (230 mg/dscm).

8. Storage Vessles for Petroleum Liquids

Hydrocarbons:
(1) If the true vapor pressue of the petroleum liquid, as stored,
is equal to or greater than 78 mm Hg (1.5 psia) but not
greater than 570 mm Hg (11.1 psia), the storage vessel shall
be equipped with a floating roof, a vapor recovery system,
or their equivalents.
(2) If the true vapor pressure of the petroleum liquid as stored
is greater than 570 mm Hg (11.1 psia), the storage vessel
shall be equipped with a vapor recovery system or its
equivalent.
1-424
-------
9. Secondary Lead Smelters

Particulate Matter:
(1) 0.022 grain per dry standard cubic foot (50 mg/dscm).
(2) Less than 20% opacity visible emissions.

10. Secondary Brass and Bronze Plants

Particulate Matter:
(1) 0.022 grain per dry standard cubic foot (50 mg/dscm).
(2) Less than 20% opacity visible emissions.

11. Iron and Steel Plants
Particulate Matter:
0.022 grain per dry standard cubic foot (50 mg/dscm) from a
basic oxygen furnace.

12. Sewage Treatment Plants

Particulate Matter:
(1) 1.30 pounds per ton of dry sludge input (0.65 g/kg) .
(2) Less than 20% opacity visible emissions.
feed (10.0 g/metric ton).
13. Wet Process Phosphoric Acid Plants

Total Fluorides:
0.020 pound per ton of equivalent ?

14. Super-Phosphoric Acid Plant

Total Fluorides;
0.010 pound per ton of equivalent P-,0 feed (5.0 g/metric ton).

15. Diammonium Phosphate Plants

Total Fluorides:
0.060 pound per ton of equivalent P~0 feed (30 g/metric ton).
£ J

16, Triple Superphosphate Plants
Total Fluorides:
0.20 pound per ton of equivalent P

17. Granular Triple Superphosphate Storage
feed (100 g/metric ton).
Total Fluorides :
5.0 x 10 pound per hour per ton of equivalent
g per hour per metric ton) .
stored (0.25
1-425
-------
TABLE 3
SOURCES FOR WHICH DRAFT SSEID HAVE

BEEN COMPLETED AND REVIEW INITIATED
Source
1. Copper smelters
2. Lead smelters
3, Zinc smelters
4. Aluminum reduction
plants

5. Ferroalloy plants

6. Iron and steel
mills

7. Coal cleaning plants

8. Polyvinyl chloride
plants

9. Lignite-fired steam
generators

10. Grain elevators
11. Sulfur recovery
plants in
petroleum
refineries

12. Kraft pulp mills
Affected Facility

Roasters, reverberatory
furnaces, converters, or
metallurgically equivalent

Sintering machines, roasters
or metallurgically equivalent

Pot lines
Furnaces

Electric furnaces

Air tables, thermal dryers

Entire process

Boiler
Truck, railcar, barge and
ship unloading and loading
operations, and conveyors,
cleaners and dryers.

Sulfur recovery plant
Recovery furnace, lime kiln,
smelt dissolving tank, digester
and multiple effect evaporators.
pulp washer, black liquor
oxidation tanks, and condensate
stripping system.
Pollutant^

Particulates
Sulfur Dioxide
Particulates
Sulfur Dioxide

Particulates
Sulfur Dioxide

Fluorides
Particulates

Particulates
Carbon Monoxide

Particulates

Vinyl chloride

Nitrogen oxides

Particulates
Total reduce
sulfur and sulfur
dioxide
Total reduced
sulfur and
particulates
1-426
-------
Source

13. By-product coke ovens
charging operation
1A. Crushed stone plants
Affected Facility

Truck, railcar, barge and
ship unloading and loading
operations, and conveyors,
cleaners and dryers

Crushers, screens, conveyor
transfer points, surge and
storage bins, and drilling
operations
Pollutant
Particulates
Particulates
1-427
-------
TABLE 4

SOURCE CATEGORIES BEING SURVEYED
Source
1. Primary aluminum - 111-d

2. Kraft pulp mill - 111-d
3. Chlor-alkali plants

4. Electric furnaces in grey iron
foundries

5. Carbon black plants
6. By-product coke ovens - pushing

7. Lime plants

8. Phosphate rock processing

9. Drycleaning

10. Sintering plants in iron and steel
mills

11. Solvent degreasing

12. Sulfur recovery plants in natural gas
fields 111-d

13. Asphalt roofing plants

14. Low grade iron ore beneficiation

15. Lead battery manufacture

16. Gasoline additive manufacture

17. Lead typesetting plants
Pollutants

Fluorides

Total reduced
sulfur

Mercury

Particulates

Hydrocarbons, carbon
monoxide &
particulates

Particulates &
hydrocarbons

Particulates & sulfur
dioxide

Particulate

Hydrocarbons

Particulate

Hydrocarbons

Sulfur dioxide

Particulates &
hydrocarbons

Asbestos

Lead

Lead
1-428
-------
AN OVERVIEW OF REGIONAL FUEL COMBUSTION REGULATIONS:

PARTICULATE MATTER AND SULFUR OXIDES
By:

G. T. Helms

Deputy Director, Air and Hazardous Materials Division

United States Environmental Protection Agency

Region IV

Atlanta, Georgia
1-429
-------
INTRODUCTION

The Clean Air Amendments of 1970 provide the basis for current national
efforts to control air pollution. This legislation requires the
Administrator of the Environmental Protection Agency (EPA): 1) to
designate air pollutants that have an adverse effect on public health
and welfare; 2) to publish and periodically update monographs, known
as air quality criteria documents, which accurately summarize current
scientific knowledge of the effects of designated pollutants; 3) to
issue documents summarizing control technology applicable to des-
ignated pollutants, and 4) to propose and promulgate national ambient
air quality standards for designated (criteria) pollutants. Primary
standards are to be sufficiently stringent to protect public health,
allowing an adequate margin of safety, while secondary standards are
to protect public welfare. Thus far, six criteria pollutants have been
designated: particulate matter, sulfur dioxide (S02), nitrogen oxides,
carbon monoxide, hydrocarbons, and oxidants. This paper deals principally
with two of these pollutants, SO, and particulate matter.

Under Section 110 of the Clean Air Act, states are required to adopt
and submit for EPA's approval plans which provide for the implementation,
maintenance, and enforcement of the national ambient air quality standards.
Such plans are to be approved by EPA if they provide for the attainment
of primary standards by mid-1975. The plans must include "emissions
limitations, schedules, and timetables for compliance with such limita-
tions, and such other measures as may be necessary to insure attainment
and maintenance of such primary or secondary standard." The emission
limitations developed under Section 110 apply to existing air pollution
sources and once approved by EPA become by that fact Federally enforceable.
New air pollution sources are governed by emission limits set pursuant
to Section 111 of the Act. These "New Source Performance Standards"
are set independently of air quality considerations and represent the
best available control technology considering economics.

CONTROL STRATEGY TESTING

EPA's regulations governing the preparation of Section 110 State Implemen-
tation Plans were promulgated in the August 14, 1971, Federal Register.
These regulations, now known as 40 CFR Part 51, provided a rather rough
and ready guidance by which states could develop emission limits adequate
to assure attainment of the national standards. Section 51.13(d), Control
Strategy for Particulates and S02, allowed the use of the so-called
1-430
-------
"example region" approach to control strategy testing. In practice,
the use of this approach meant that s'tates could develop regulations
for general application on the basis of the degree of control needed
only in the most polluted areas of the state. The resulting emission
limits in some cases were unnecessarily stringent when applied to
sources outside these heavily polluted areas. In Florida, for example,
fuel combustion SO^ limits needed to meet standards in Hillsborough
County were applied to sources in all other parts of the State as well.
States like Kentucky, where limits varied according to an Air Quality
Control Region's (AQCR) priority classification, were the exception.

40 CFR 51.13(e) provided for the adequacy of states' regulations to be
tested either by a proportional model, generally referred to as the
rollback technique, or by diffusion modeling. Since the rollback method
was easier and faster to use, many states developed their regulations in
this way. The major flaw in the rollback method, aside from the crude-
ness of the formula itself, is that rollback neither considers the
special distribution of air pollution sources nor the meteorological
conditions influencing atmospheric dispersion.

As a result of improved expertise and less restrictive time frames, a
shift has taken place toward the use of diffusion models to estimate
the effects of emissions and emission reductions. For fuel combustion
sources, the most commonly used diffusion models are EPA's Air Quality
Display Model (AQDM) and 24-hour power plant model. As its name implies,
the latter is concerned with predicting short-term concentrations in the
vicinity of power plants under varying meteorological conditions and
terrain configurations. The AQDM is a multi-source, long-term urban
dispersion model which is used to determine the impact of a wide
variety of stationary source categories on annual average concentrations
of S02 and particulate matter. Basic inputs are a comprehensive emission
inventory, including both point and area sources, a joint frequency
distribution of wind speed, wind direction and stability classes, and
an annual average mixing height.

REGULATORY REVISIONS
The increasing sophistication of mathematical models has been in part
spurred by the desire to remove some of the excessive stringency in
S02 emission limiting regulations adopted by the states in 1972. With
the energy crisis of 1973, and the low-sulfur fuels shortage, EPA
responded with its Clean Fuels Policy, which sought to reserve the use
of low-sulfur fuels for areas where they were needed to meet health
standards. Also, the Energy Supply and Environmental Coordination Act
1-431
-------
(ESECA) of 1974 required that state implementation plans be examined
to determine if changes could be made in emission limits applicable to
fuel-burning sources without jeopardizing the attainment of ambient
air quality standards. Before the passage of ESECA, EPA Region IV
had already received S0? revisions from Alabama, South Carolina, and
Tennessee. These have'since been approved, except for the immediate
vicinity of three TVA plants.

Tennessee is currently preparing a further revision of its SC^ limits
for fuel-burning sources. EPA has just announced receipt of Florida's
ESECA revision and called for public comment on it by means of a notice
in the Federal Register. Kentucky has developed an ESECA SG^ revision
with the assistance of EPA Region IV. In the Kentucky revision, virtually
all the power plants in the State were subjected to diffusion modeling.
The resulting emission limits are contained in a five-tiered structure
coupled with a county classification system. The limits are in a few
cases more stringent than the old ones, but in general far less re-
strictive than those the State originally developed in 1972.

SUMMARY OF STATE REGULATIONS

With the help of visual aids, I would like now to give you a summary of
Region IV state regulations applicable to particulate and SC^ emissions
from existing fuel combustion sources. Particulate regulations have
remained essentially unchanged since their development in early 1972.
Except for those of one state, all take the general form shown in
Figure 1. They provide for varying degrees of emission control as a
function of the heat input of a boiler or boilers. States vary in
the application of this regulation, some employing the rated capacity of
each individual boiler to arrive at an allowable emission limit, while
others use the total heat input for all boilers at a plant or facility.

Table 1 contains a summary compilation of all the applicable Region IV
particulate emission limits for existing stationary fuel combustion
sources. As can be seen, the limits range from 0.8 Ibs. per million BTU
of heat input for small sources in Class II counties in Alabama to
0.10 Ibs. per million BTU for larger sources in North Carolina, Tennessee,
and Florida.

Particulate emissions from new sources are governed by the Federal New
Source Performance Standards contained in 40 CFR 60.42 which contain a
limit of 0.10 Ibs. per million BTU.
1-432
-------
Regarding SO- emissions, EPA's clean fuels policy and ESECA have
caused emission limiting regulations to undergo numerous and continu-
ing modifications. The SC>2 regulations currently federally enforceable
for existing fuel combustion sources are presented in Table 2. As can
be seen, many states have gone beyond the example region approach and
tailored their limits to specific air quality considerations.

Table 3 presents additional SC>2 regulation changes currently under
review by EPA. These proposed changes represent a further refinement
and tailoring to reflect the air quality and economic needs of Individual
states.

New fuel combustion sources are again governed by the Federal New Source
Performance Standards as contained in 40 CFR 60.43. This required
compliance with S02 limits of 1.2 Ibs. per million BTU and 0.8 Ibs. per
million BTU respectively for coal and oil burning facilities with heat
inputs of 250 million BTUs per hour or greater.

SUMMARY

I have attempted to provide an insight into the Clean Air Act and the
methods and techniques employed to establish fuel combustion limiting
regulations for partlculate matter and S02. Currently applicable parti-
culate and SOj emission limiting regulations (Tables 1 and 2) were
discussed along with changes
presently considering.
(Table 3) that Region IV states are
For more detail on the regulations which were presented today, I
have appended a listing of the State Air Pollution Control Directors.
I suggest that you contact them about the applicability of any
regulation.
1-433
-------
UJ
o
cc

o « -
to
3
CO
UJ
UJ

U-
o
o:
CO
to
LU

LU
o
ex
s
OQ

O
11
I I I
I I I 1 I I I 1 I I I I
o
o
o
o
OQ
o
UJ
^H O O O O

POUNDS PER MILLION RTU-HEAT INPUT
1-434
-------

prf
S3
I
§
§

o
o
o
1
A
a
B
M CO
H

S W §
|J 0 O
M 0
£5 3 —V 0
O 0 p
W CO H O
OT pq rH
w O
rH 0 £ O
S -H 2 4J
£3 4-1 <1
-* m J <3 S «H
rt rJ o 3
<5 »3 CJ CJ
H CJ iH
H H 4J
H O Ol
in co m
CO CO rH
CM CN CM
O O O
1 1 1
rrj prj 23
«?• m CM
co CN m
\0 00 rH
O\ CM CO
O rH rH
II II II
MUM

^O ^3 ^^
in co m
CO CO rH
CM CN CN
O O O
EC tC EC
<± m CM
co CM in
VD OO rH
^J^ ^^ C^
O rH i-H
II I! II
jy Id fT^

vo m o
m i-- oo
o o o

w
O H rH
HUM
««
O r*** r*"l r*%
4J 4J 4J
>, -H -rl -H
^ |-l ^1 ^l
o o o o
P T( -H -rl
4J id H i-l
(3 fu Pn PH
tS

O\
rH
CD

in
x^
^o
rH
O
EC
CO
0
00
00
o
II
u

m
vO
VO
rH
0
1
CO
o
oo
00
o
u

VD
O

t)
•rl
(X
CX
•H
CO
10
•H
CO
CO
iH
S

o
rH
O

^^
ON
m
CN
o
EC
CO
0
01
o
rH
II
Pd

•^
0\
m
CN
0
t
CO
o
o
rH
A

VO
O

TJ
cd
rj
•H
rH
0

(fl
CJ

,r{
4J
M
O
5Z
C
O
4-1
TJ

o
rH
w

\O
d

*
T3

CU
CU
CO
CO
CU
C
a
H
4-1
CX
4-1
cd
S
a
o
•H
H
rH
iH
6
M
CX
CU
4J
rH
3
U
•H
4J
(-1
cd
(X
CO
42
^
(U

cd
M
e
o
•H
CO
CO
T-t
£3
CU

CU
rH
"cd
I?
O
rH
rH
^

logy (LRACT) as defined by the state.
(old Regulation 7) for sources that installed control equipment

rl
o
r<
S.

uj
B
PQ

S
S

(3
•H

4-1
3
cuo
c o
•H rH

•M Jxj
(0
d) i-H

II
II

BB S

O
M
CJ
CU
4J
rH
2
*^
o
CJ

cu
rH
43
cd
rH
•H

J>
cd

^*»
^
^f>
cd
(3
o
CO
cd
cu
rl

4-1
CO
cu
4J
3
t
cd

CO
4J
•p-f
g
•H
4-1
C

I
^4
4J
CO
CO
CO
cu
rH

T)
CU
s
0
rH
rH
cd
CO
C
0
•rl
4-1
cd
rH
3
00
0)
1
41

.
o\
VO
*-^
Is**
CM
^s^.
rH
rH
O

i-i
0
•H
rl
CX

CO
•H
CO
cd
43
4J
•H
?>*
K*>
43
4J
•H
§
cd

d
0

TJ
cu
M
3
oo
•rl
MH
4-1
3
CX
C
•H

4-1
cd
cu
i
o

4-1
rj
CX
•H
4-1
cd
r!
r**f
4-1
a
cd
rH
fX
rH
cd
4J
o
4J

4-1

CO
•H
4J
3
CX
C
•H

4-1
td
cu
1
•a
Knoxville, and Memphis.
ur, then a constant limit of 0.24 # per MM BTU.

cu
rH
rH
-H
43
^
cd
etc
0
o
C
4J
fl!
rdt
CJ

0
4-1

CU
rH
o
1 43
«|
OJ

•i
>o
o

Q
^J

ft
3

CM
o
CN

•H
T-l
O-
D.
cd

4-1
S
1
cu
EC

in
•
-------
CN
CO 13
H 01
M 4-1
S co
M CO CJ
i-J 01 tH
CJ -O
M O
CO CO CU
CO CO
M a -H

S-SB
•M C)

s g*
•i
. _ o co
PCI M CJ M
^J Q qj
H . tH rH
60 H
CO
< X CM
O W O
M CO
i-l
PH CO
5!

***w
P
^
PQ

i
o
m
CN
A

S
^^^
B
PQ

^Tj
°

s
CM

O
rH
V
1

£
£

0
rH

4J
9
(X
CJ
w

4J
cd
CD
JU

cd
4-1
CO

jj
r-
o

<
o
CJ

,j
H
O

rJ
•«rf
O
o

t-J
H
o

J
5]
P

B
m

vO
o
rH
^"- "J
OO O tH O
* * » •
rH sr iH en
g

sD
0
tH
^ "2
oo o ^n o
• • • •
rH ST H CO

o
o
tH »CO
Cd 6s?
o 3 o
4J • — .
p co
H CO H
CJ B*? PQ CJ
oo o -rf m cu
* * 05 • S^ _C
rH •* ,-q CM g 4->

o
0
rH ** CO
O P3 CO
4J ~^ •
p cn
H CO H

oo o <2 LO cu
* * PQ • ^i .*"!
rH ^T ,J CM g 4-1

H CO
O fr*
oo o ^« **
IH H cd U
O O
cd oo oo cd -, >.
«d 4J 4J
^s^ <|«4 *rH
^ W VI
O O O
3 -H •> CU 0
4-1 O. 1-1
•H >H cd
H CO U
O CO
•H tH JC
t-l (04-1
P.| {fl J^

co
•
CM

CO
•
CM

m
,
CO

co
CU
•H
4-1
ct) C
C 3

rH O
o
I-l M

cj co
CO
-C cd
4-1 rH
3 U
O
CO

y*
s
O 0)
0 A
O 4-1
rH

O 03
4J ^
p cn
>n H * m
• PQ CM •
co co
O cu
O JC
O 4J

• * •
i-i »

CO
(0 01
CO 01 -H
0) -H 4-1
•rl 4J (3
4-J C 3

jmj Q cj
o o
CJ IH
Cd MM

«J
•50

3
IH o

-•B
O
U Q)
M Q)
Q)
01 P4
O
4-> CO
10 rH

-------
CO
H
M
g
B
n
o
m
CM
A

a
b
o
m
CM
o
4J
O
i-l
S»
«
O
>J
•4->
9
o.
a
M
4J
* x
w/
a)
u
CO
4J
CO
f!
O
g
o
I-J
H
O
3
0
CJ
rt
o
d
0
u

•^
VO
«
i-H
*••» 10
rH in i-^
rH CM CM
cfl
p~ r^. r~
rH rH rH
vC vO VD
^
VD
•
i-l
»«•. in
•H in r^
• • *
rH CM CM
r«. r-» r*»
•H H rH
• • •
vO VO VD
43
in
\o
rH
•^ m
fH m r~-
i-H CM CM
i-» r-» r^
rH rH rH
vO vO \o
Florida
Hillsborough Co.
Duval Co.
All Other Counties
oo CM H m o
o tH CM ro "O *O
w w co en w
Q) cu CD QJ cu
T-l r-l r-l i-l i-l
cO cO cfl cd tO
U U U U U
Eft CO W CO W
M M 60 GO 60
C C C C C
•H -H i-t -H iH
T3 *O *O *X3 ^
iH rH rH r-l iH
en w to w tn
%£) vt O
n
H M M H >
^ w to * m
u u
CM CM \O O O
T-I I-H i-i -a- m
vD O \D O O
1-1 m i-i
(U M H M H >
(1)
CO W CO CO M W
CO CO CO CO CO 01
CU CO CO cO CO cO
gr-1 r-l iH r-H i-H
CJ U U U U
0)
H
vO
O
X

c
o
c
e
CO
O

4J
C
cO
rH
Pu
4-1
CO
vO

m
CO
•H
Ci
4J
O.
0)
CJ
X
cu

CO
CU
CI
IH
9
O
CO

cu
^4
•H
111 ^^
vn
rH "
CO vO
0

o
H
rH 4-1

O -H
4-1 rH
co
o> s
•H 3
rH «
a
CX 60
Cfl 1-1
CO
p">
F^ *M
p^ flj

3 "^
§ £j
<. cfl
CO
ft /— V
rH <»
•
r>- CM
rH
• *' '
• *HW
vO O

ttH 4-1
O -H
W iH
•H rH
*H

1
cO
V
0
M
9
O
CO

h
cu

4_)
0

rH
CO
M
0
ii i
*W
P
P3
s
^
CO
5

in
\0
*
rH

•o

M
*
-------
APPENDIX
1-438
-------
REGION IV STATE AIR POLLUTION CONTROL DIRECTORS
Mr. James W. Cooper, Director
Division of Air Pollution Control
State of Alabama Department of Public Health
645 S. McDonough Street
Montgomery, Alabama 36104

Mr. Joseph W. Landers, Jr., Secretary
Florida Department of Environmental Regulations
2562 Executive Center Circle East
Tallahassee, Florida 32301

Mr. Robert H. Collom, Jr., Chief
Air Protection Branch
Environmental Protection Division
Georgia Department of Natural Resources
270 Washington Street
Atlanta, Georgia 30334

Mr. John T. Smither, Director
Division of Air Pollution
Kentucky Department for Natural Resources
and Environmental Protection
West Frankfort Office Complex
US 127 South
Frankfort, Kentucky 40601

Mr. Jerry M. Stubberfield, Chief
Division of Air Pollution Control
Mississippi Air & Water Pollution Control Commission
P.O. Box 827
Jackson, Mississippi 39205

Mr. J. A. McColman, Chief
Air Quality Section
Division of Environmental Management
North Carolina Department of Natural & Economic Resources
P.O. Box 27687
Raleigh, North Carolina 27611

Mr. William G. Crosby, Director
Bureau of Air Quality Control
Department of Health and Environmental Control
2600 Bull Street
Columbia, South Carolina 29201
1-439
-------
Mr. Harold Hodges, Director
Division of Air Pollution Control
Tennessee Department of Public Health
256 Capitol Hill Building
Nashville, Tennessee 37219
1-440
-------
COMMON METRIC CONVERSION FACTORS

1#/106 BTU = 1.8g/106 Calories

BTU X 0.252 = Kilogram - Calorie
1-441
-------
-------
THE PAST, PRESENT AND FUTURE OF

STATE AIR POLLUTION REGULATIONS

FOR COMBUSTION SOURCES
By:

Robert H. Collom, Jr.

Air Protection Branch
Georgia Department of Natural Resources
Atlanta, Georgia
1-443
-------
As has already been discussed by EPA representatives, the 1970
amendments to the Clean Air Act required a good deal of short-term
and specific action from the federal, state and local governments.
The end result of all this hurried activity was the State Implemen-
tation Plans which, at a minimum, had to be adequate to attain and
maintain national ambient air quality standards for the six criteria
pollutants. Almost all States felt it feasible to meet the primary
and secondary air quality levels at the same time (July 1975).

To achieve the ambient sulfur dioxide standards many States
elected to utilize sulfur limitations in fuel as their emission regu-
lation of choice. Others adopted the approach of specifying an allow-
able sulfur emission in pounds per hour or pounds per million BTU's
from a fuel-burning boiler. There was almost no attention given to
sulfur dioxide emissions from any sources other than fuel burning.

Due to the fuels crisis and the economic crunch we are now facing,
EPA has deemed it necessary to suggest to the States that if possible,
they relax their emission standards and tailor them more specifically
to those limits actually needed to achieve national sulfur dioxide
ambient standards. The Energy Supply and Environmental Coordination
Act of 1974 required that EPA review all State Implementation Plans
with this goal in mind.

Control of particulate matter from fuel-burning sources was more
uniformly approached, and based on proven control systems such as
electrostatic precipitators. Technical understanding in this area
was much more available to the sources, and the particulate regulations
have been more readily complied with throughout the country.

During the development of the Georgia Implementation Plan it was
determined that the State's air quality levels for sulfur dioxide were
generally much better than the national standards required. This
coupled with a need for a margin of safety and a desire to maintain
good air quality brought about adoption of a State ambient regulation
lower than the national standard. It was determined through computer
modelling of two "example regions" in the State that the use of readily
available 2.5 or 3.0 percent sulfur in fuel (depending on boiler size),
coupled with minimum stack height requirements, that the State ambient
standard could be achieved. This has been proven out by subsequent
ambient air sampling throughout the State.

Since the State plans were submitted to EPA and approved by them
in early 1972, a number of significant developments have transpired.
Most significant of these are the fuels crisis, the supreme court de-
cision (or lack thereof) in relation to significant deterioration,
1-444
-------
and a suit brought by NRDC against EPA for having approved certain
portions of the Georgia Plan (specifically those utilizing stack
height and dispersion).

In February of 1974 the Fifth Circuit of Appeals made a partial
ruling on the case involving use of stack heights. It was made clear
that dispersion techniques are not allowed under the Clean Air Act as
a means to achieve national ambient air standards. Emission reducing
regulations are the only acceptable means of meeting the national
standards. However, stack heights or any other techniques are accept-
able to achieve a better air quality than national air standards re
quire. A final decision is pending from the Fifth Circuit concerning
this point. Computer modelling has shown that the emission limiting
regulations of the Georgia Plan are adequate by themselves to meet
national standards. Therefore, it should be permissible for the State
to use dispersion techniques to achieve the State ambient air standards.

At the present time there is no legally or nationally uniform
policy as to acceptable stack heights. It is necessary that a techni-
cally and legally acceptable approach to stack heights be developed
as soon as possible since the States and EPA are now facing the need
to project into the future, through the use of modelling, to determine
any necessary amendments to regulations needed to attain and maintain
air standards. Not necessarily accepting the need for it, the require-
ment to not exceed a specified air quality increment for non-significant
deterioration purposes, will be next to impossible unless acceptable
stack heights are established.

A newly-resurrected concern has appeared on the scene, namely that
of suspended sulphates and their effects upon human health and vegeta-
tion. The various reports and publications in this area tend to agree
that suspended sulphates can produce such effects, but there is a
total lack of agreement on the levels of concern and the sources and/or
mechanisms of formation of the sulphates themselves. A more thorough
understanding and grasp of these issues is necessary before any defini-
tive action of a significant nature is initiated.

It is likely that a future area of concern which will require
additional knowledge is that of the potential harmful effects from
oxides of nitrogen from fuel combustion. There have been some sugges-
tions about cancerous properties of nitrogen dioxide in the atmosphere.
Our knowledge about photochemical oxidant formation and the role which
oxides of nitrogen play does not appear to be as well defined as we
once thought. I notice that the agenda for this seminar includes many
technical papers on formation of oxides of nitrogen and their propaga-
tion and control in a boiler. Hopefully, the attention being addressed
1-445
-------
to this matter will provide industry and government experts with enough
technical understanding to take logical control action, if needed at a
future time. We have too many times in the past had to act to meet
legal deadlines on very complicated problems, with only a limited amount
of knowledge.

Thus far I think it is reasonable to say that the preparation of
State Implementation Flans and enforcement of emission regulations
has generally produced the hoped for results indicated in the Clean
Air Act. However, it would be imprudent not to take into account for
future actions and legislation the knowledge which we have gained over
the recent past. I would hope that the emotional basis for action, the
so-called "Chicken Little" approach, is behind us. The air quality im-
provements which present programs have produced should allow us to take
more deliberate and knowledgeable action in the future in dealing with
more refined air pollution issues.
1-446
-------
Statement by H. W. Poston, Commissioner
Chicago Department of Environmental Control
before the EPA Combustion Symposium
September 24, 1975, Atlanta, Georgia
MEETING AND MAINTAINING ENVIRONMENTAL
STANDARDS THROUGH PRODUCTCONTROL

When in the late '60's we became aware of the status of our environment, we

sought the advice of engineers and chemists. This was natural enough. They were the

ones who had designed the automobiles, steel mills, furnaces and boilers, air conditioners

and phosphate detergents - - all the symbols of an affluent society. Our task was to re-

verse a situation that, if not restrained, could destroy the very progress these things had

brought about. A balance was clearly needed which continued to encourage growth and

at the same time encouraged conservation and preservation. It was not feasible in the be-

ginning of our endeavor to merely close all sources of emissions and build "clean" facilities.

So we hired engineers to design controls for every conceivable emission their products had

created. This pattern of attacking the problem still prevails for the most part. The major-

ity of beverage containers used continue to be throw away bottles and aluminum cans; the

internal combustion engine remains our predominant form of transportation. Recycling

centers and catalytic converters were developed to cure the problems. The problems re-

mained largely unsolved.

Obviously new approaches were needed to solve age old problems. Environmental

standards had to be attained and maintained. Controls had to be implemented but in a

manner that would not require costly installations, additional bond issues and increased

taxes to our citizens. The controls had to be placed on the products that were causing the

environmental assault.
1-447
-------
The experiences of the Chicago Department of Environmental Control can well

serve as an example of product control to attain standards. When, in 1970, the Clean

Air Act mandated air quality standards for certain pollutants, the sulfur dioxide levels

in the city were greater than .03 parts per million (the 1975 federal annual average to

be achieved for that pollutant) at almost half of the department's monitoring sites. A

survey taken by the department determined that the major source of coal burning came

from older apartment buildings in residential areas. The Chicago City Council enacted an

ordinance in 1970, which limited the sulfur content of fuels to one per cent, for all processes,

by 1972. In addition to the use of low sulfur coal, owners switched to low sulfur fuel

oil and applied for natural gas hookups. Negotiations between the Department of Environ-

mental Control and Peoples Gas Company produced an agreement whereby residential

users were given first priority as more gas became available. Industrial emitters were

required to employ control devices in addition to the use of low sulfur fuels and the

largest cumulative source of sulfur dioxide emitters was eliminated. The result was dramatic.

By the end of 1973, every monitoring site had an annual arithmetic mean of .03 parts per

million or below and only two of the 23 sites had an annual average above .022 — the

secondary standard. In 1974, progress continued and only one monitoring site exceeded

.022 parts per million.

It might seem as if all this were accomplished without a hitch. But shortly after

the ordinance for low sulfur fuel took effect in 1970, an organization representing some

45 coal dealers filed an appeal for a variance from the limitations of the ordinance. In

its appeal for a variance, the association claimed, among other things, that solid fuel
1-448
-------
containing less than two per cent sulfur by weight was very difficult to obtain in any quantity

in the City of Chicago. Hie association listed such factors as railroad car shortages, foreign

shipments, union problems, the demand for low sulfur fuels by utility companies, the reluct-

ance of mine owners to open new mines and the threat of conversions of coal burning equip*1

ment to competing fuels.

While some of the factors may have been valid, the department brought in expert

testimony to state that low sulfur coal was indeed available for Chicago and that adequate

railroad facilities were available to transport low sulfur fuel from the west to Chicago.

The corporation counsel for the city filed a motion with the Appeal Board to

dismiss the association's petition for a variance on the grounds that the Appeal Board had

jurisdiction only to grant "individual variances." The petition as submitted by the

Chicago Coal Merchants Association did not enumerate any individual petitioner or coal

dealer a party to the proceeding.

Other interested parties, including the Attorney General of the State of Illinois and

the Environmental Law Society of the University of Chicago,also filed motions to dismiss

the petition by the association.

The association itself filed a motion to withdraw its petition. Its motion to

dismiss was based on the ground that the !'... Appeal Board did not have jurisdiction to

entertain its said petition because petitioner and its members are not individual users,

owners or operators of coal burning equipment in the City of Chicago, but are suppliers

to those individuals ..."
1-449
-------
On October 28, 1970 the Appeal Board ordered that the motion be allowed and

the petition was withdrawn.

Subsequent to this action, no individual company sought a variance before the

Appeai Board and the availability of low sulfur fuel did not pose a problem for the '70 - '71

heating season nor in the years that followed.

Control at the source has been successful in another of the department's con-

cerns with the pollution of our waterways. At its treatment plants the Metropolitan

Sanitary District removes 90 per cent of the effluent pollution. The district serves the

metropolitan Chicago area and would have to spend over $27 million each year to remove

excess phosphorus and to reduce the biochemical oxygen demand (BOD) and suspended

solids to acceptable levels. In October of 1970, the Chicago City Council passed an

ordinance limiting the amount of phosphates in household laundry detergents to 8.7 per

cent by February 1971 and after June 1972 the phosphorus content was reduced to zero

or trace amounts of phosphorus.

The reduction to 8.7 per cent had marked results. At one plant the Metropolitan

Sanitary District recorded phosphorus concentrations of its effluent ranging from 1.6 to

3.5 milligrams per liter in 1971. For the period of July through December 1972, the

average effluent was .65 milligrams per titer. Therefore, without additional treatment

the facility was in compliance with the State of Illinois water quality standard for phosphate

concentrations.
1-450
-------
Not only was this standard met, but additional benefits were realized when

BOD levels dropped to less than 10 milligrams per liter and suspended solids to less

than 12 milligrams per liter and at a savings in excess of $27 million.

Once again, the ordinance did not go unchallenged.

The constitutionality of the ordinance was challenged in the federal district

court on the grounds it imposed an impermissible burden on interstate commerce. On

March 6, 1973, the court ruled ". . . the entire ordinance, including the labeling

and enforcement provisions, is of no effect."

The local U.S. Court of Appeals upheld the city's ordinance classifying it as

"... a slight burden on interstate commerce" on January 15, 1975.

Following this reversal, Procter & Gamble and the FMC Corporation petitioned

the U.S. Supreme Court to review the earlier court proceedings. This court's refusal

in effect upheld the city ordinance.

Penalties range from $100 to $300 for the first violation and from $300 to $500

for additional offenses within a six-month period.

Enforcement of Chicago's phosphate ban in household laundry detergents was

•esumed in Chicago stores on August 19. The enforcement action came 90 days after

I'he U.S. Supreme Court refused to hear an appeal of a lower court ruling which upheld

the city law.

Zero phosphate limitations apply to household cleaning compounds except those

used in automatic dishwashers. Other cleaning products exempted until December 31,

1975 include food processing and dairy equipment cleansers; phosphoric acid products such

as sanitlaers, acid cleaners and metal conditioners; and laundry detergents used

exclusively by the food and dairy industries and institutional health care facilities,

1-451
-------
These exemptions are similar to those granted by other Jurisdictions in the

country with phosphate controls on cleaning products. During the grace period, depart-

ment engineers will assess the technological developments by industry in developing

phosphate substitutes for these products.

Product control such as this eliminates the need for tertiary treatment, elim-

inates the problem at the source and saves the citizen tax dollars.

Suspended particulate levels were also above the ambient air quality standard

of 75 micrograms per cubic meter. We attacked this situation in much the same way

as sulfur dioxide. An ordinance enacted in 1970 prohibited burning refuse in boilers

and leaf burning. Single chamber incinerators were prohibited in 1971 which encouraged

the use of compactors. Industrial pollution control programs were made mandatory for

all major emitters. Our results have been substantial. In 1971, only 10 per cent of

the land area of Chicago had a concentration below 75 micrograms per cubic meter.

By May of this year, 51 per cent of the land area met the standard. Likewise, con-

centrations were reduced substantially in areas not presently meeting the standard.

Forty-one per cent of the land area had concentrations over 100 micrograms per cubic

meter in 1971. Today this has been reduced to five per cent.

In the area of noise pollution, the city's ordinance prohibits the sale of

motor vehicles which emit excessive noise. The determination of noise is made by

acoustical instruments. It is the manufacturer's obligation to certify that vehicles sold

in the city meet specific standards. Because motor vehicles constitute the major urban
1-452
-------
noise source, the ordinance attempts to solve noise problems at the source without

building costly sound barriers or tunnels to mask the source of noise. Beginning in

1980, no motor vehicle emitting more than 75 dB(A) as it comes off the assembly line

can be sold in the city of Chicago.

Thus, by product control which controls pollution at the source, the city has

been successful in meeting standards that a decade ago seemed impossible to attain.
1-453
-------
1-454
-------
JuOCJOOUOOC
CITY OF CHICAGO

Department Of Environmental Control

SULFUR DIOXIDE
... ojoJQjjjuuoooooooonoooooooooooai
.. JOUUUQJjjouccoooooo'jooooooooooooooooo
.. UOUJJU'JJJOOOOOMOOOOUOOOOOOOJOOOUOOOOO
. OOOOiJOOOJJOPOOOOOOOOOOOOOOOOOOOOOOOC—
. OOOOOUOOOiJiJunoooOOOOOOOOOOOOOOOQOOQaC
OOJOJ3JOOJJUUOOOOOOOQQOQO 000000 OOQQOiJ f
jooooJooooiX>jooon)aooi»B8o Booo;
JUNE 1970 THRU MAY 1971
. JOJOUJJjaUvJJUOOOO .
. joaodJJOOOujuonon ti
. JOliOOUUOOOUuOOCT! Ml
riJOOUUJJOJOJJOOOC "•
. JOJLJJJJUJJJUQOOO
. JOuO-jDJOuJJbGOG •
.. OJOJJJOGCJOUOO
OOUJUUOOOO
. JJOUOOOOOO
. JOUJuODOO \
.. 'juuuoooao
... jjuoooon __
. .. uUUOOCOOO II
.. oouooooooo a
. .. JOUQOOOOOO
.... JJ0000300Q
.... uuoooonooo
.... Dunonauaooi
.... auoooooo"
. ... JOOTOOM
.... jaoooooo
. ... 000000031
. .. JOUOOOOOOOOl
... jyuoaoaoo-
*
. .100 _
JOOOOOOO
gooooooouao ooooooooooa
_OOOQOOOOQOOOOOOOOOOOOOUOUOODDOOOODOCK)
jjonnonoooocunoaoootwaooooooooooczoooQQ
joooonocio: ooooonooooooooaooooaooooogoooooooo
DjujoooooaooooooooooooooaoaoooooonooooaooTOoo
jjODonooomooonnooooouoooooooooooooooooooo DO
ooooooooooooooaaoooooouoootiaaanooooooooooQ oo
.... ouojuooonoooouaooooooooooaououooaooooaoaaao oo
... jooooouooooooooooaoaoooooooooaoaooaooooooooo oo
... jaOOUOOOOOOOOOOOOOaOOOOnuQOOOOOOOOOOOOOOOOOO
. . iJJDOOrjOOOOOOOOOOfXJOOOOOOOOQOOOaiJOOOOOOOOOOOOOO
. oguououooooonnorooooogooooooooodoagooggpogogQOo
V) 5 5 oa ooci o Cinn T) oooaaoocoooo 30 ooaoooooooooooooao
jojjooaaujooocci oDooooooooooocooo'joo oooo ooooo!
jouooouojuoooaooooooooooaooooooouooooooooooa
oooojuoaoouyoooooooooaooooooooooonjooooaooo
j oooo J J u-jo o o ocoor 'j'joo oaoo oooooooo ooooooooaoo
jooaojuuuuuL.uc!oncoa oo noouoooooo 0000300
jopoppbpojjujjgocgooogpooooooooogopoqg

10000
0000
OCO 00 ju 000 JLflJ 0000000
•guooooooiooonooooo
oaujuuOuuuuoooa?
•jOOOuuoooDuoconrj
ouoauoouuuuu
jouo-
10 OUOUO u 000000 000000 0
1.1 on ojuvncoo ooooooooo
'JOOJ .10 000 00 00 00 00 000
oooaoooooooooooooooooooao
u.juouuuoooocMOijoooooGaooooaooauoooaoooooo
JOOOUOI/UOOOOOOOOOOUOQ0000030!
. JOOOOOUOO .
ooooaooooooogooaooa i
10000 uaooooaooBooODOO
oooooaooooooooo i
nOOOOOOOOOOQOOQ i
I 000000000000 J 0
-- 00000000000000
oaooooouoaaooo
............. oaooooooooi
ooaojaooo'
0000000
oaoojo loooooo^oooooooo
. oouooooooooooooooonoo
ouooooouooooaoooo
00.10
CODE:

In Parts Per Million

Percent Of Land Area For Each Level
.022 AND UNDER
nnononimr.
no"onnnon
mnrv^oooc'

nor nc^rn^

.013 TO .030
22%

FEDERAL AIR QUALITY STANOARO - O3O
00.
OOUI
uooool
oooaoaoi
ooooooooooi
- oooaooooooooaj
OD|OQoaoooooaoaoi
ooooooooooaooaooaj
oooooooooooooooooaoI
ooooooooaooooaoauooaj
ooooooooooaoooooooooo*
oaoooooooaoooooooaooooo!
aoooooonooooooooooooocxinr
oooooooooocooooooooo
ooooooaooooocooooooooo
oaooooanooooacooaaoo—
ouoooouooooa oocoooooo
DUOOOJOOOOOOOOOOOOOOO
.1 ooauaooooaoooooooooo -
ooaoooouaaoaooaoooo r
ouoouoouujaaoooao'Xi ~
oocooauoiAJLiooooaoao^
oououuooaojogooooooo
oojOuMOOooauoaooooao
OOOOOJOOOQOOOOOOOOOOOO
oojooouajoooooooooooooo
JOoijo-jooooooojouoooaoooo
ooooooooi ooooooouaoooaooo
,.., uooooooooooocooouooQoaoooao
xiijooooooooggoogodouogoogogoopgo

'•W-10°TO
aooooo
OUOOOOI
L ,
-------
0, CITY OF CHICAGO

Deportment Of Environmental Control
JOOOOOO
uugooogo
ooaooooao
aouoanoaoao
oooa
Ukigoooooooo
oaoo
OOJOU
0
SULFUR DIOXIDE

JUNE 1971 THRU MAY 1972
ou oaaooj
Oiiuoooa
OJOOO
OOQO
oiooaaQuoDuouoooa
dOOOUDUQOOOO •
000030000030 1
ooaoouuaoooD
JUOOODOOiiOOOUOU
0
a
3
CODE:

In Port* P*r Million

Porcont Of land Ar«a For Each level
-------
CITY OF CHICAGO
Of Invlronmwiml Control
SULFUR DIOXIDE
JUNi 1972 THtU MAY 1973
CODE:
In Parti P»r Million
Percent Of Land Area For E«ch L*v«l
.092 AND UNDM «» TO .030 OVI1 .030

10%
HMtAL All QUAUTY STAN DAK D - .030
•»*»•••••*••••**••*••••*****»«
1-457
-------
CITY OF CHICAGO

; Department Of Environmental Control

SULFUR DIOXIDE
JUNE 1973 THRU MAY 1974
.1.
CODE:

hi Parts Per Million

Percent Of land Area For loch Level

ntin'mooac
oofioonnoo
.032 AND UNDCR
aooi
aaoaai

:::"o0o,oo8388grw'
;. aoooooor —
. oooaooaoL-_
. ooooooooJOG
. oooaoooooa <
. ooaaaooaao
. OOOUQOOOJ i
aoooooaoaao
. OOOQOUOJJO
1 rtMRAL AIR QUALITV STANDARD - .030

I
1-458
-------

f :::: ::";:::::::: ":::::: ;:::::." "":;;;•.:

I ••••::::::::!:::::::::::::::::::::::::::::::::::::

'• ::::::::::::::::::::::::::::::::::::::::::::::

t :::::::::::::::::::::::::::::::::::

I *

.-,..

........1...

J
! i:::::::::::::!::::::::::::

.......... 1.

4
{ COM:
j ft»rc*nt Of Land Area For Each Level
t .023 AND UNDf i .023 TO .030 OVEI .030 '.'.:
riNBAl AIR QUAIITT STANDARD -.030
f

---7* 1 » 2 »—-- 3 »- 4 « 1 » t » —

::::::::::::!! CITY OF CHICAGO
Deportment Of Environmental
:::::::::::::: SULFUR DIOXIDE
:::::::::::::: JUNE i«74 THRU MAY 1975

QOOOOOOOUOOOO

ooooouooooooo

OOOOOTOOOBOOO
• i poooooooooooa

ooaoooooo
-
-------
CVOOOP
nocooonooooonnooooo

10003000000900000900
CITY OF CHICAGO

Department Of Environmental Control

SUSPENDED PARTICULATE

JUNE 1970 THRU MAY 1971
OOPOOOOOnOWfJOPOCO
rwooonnooorioTo'iaef'oo
nooooooonooooonc'X'ono
roooo"nnoopooponr"">oonano''Ooo''ono'70oni
rdooooiooooononoooiyyinoooaooooDooaooio'
otioooooooomooaporu
Dononoonoonprocog'
oooofoooononoo
oooonnoooooononoo
oaooooo onooonooo
oonnnpo
ooooooooooooo
•jpconocncoon
pionooooaoo
oooooooaooo
aooooreooo •
p
0
0000000
sorortn
CODE:

hi Mkrograms Per Cubic Meter And

I Percent Of Land Area For Coeh Level
ODD
oonoaooao

aonooongoDO
oonooooooononon
on
-------
OOOPOOOCOOOOn

OOPOOOOOOOOT
onooono
nooopnn
nocoior
nooooonn
ooooooonnponoo
CITY OF CHICAGO

Department Of Environmental Control
000001100100000000
OOO
•o
•00000
opionppiir
ne
in
nPO
POOOPPOOOTOntJOO
SUSPENDED PARTICULATE

JUNE 1971 THRU MAY 1972
mooooooooonooooiotrnposor
nnooonoooorjrid oppPocoioTOOoooaoaoDOCO
oooiooaooooopoooncpoooooooocooooooop
nonponoTooooooonooDOpooooonoooooooori
oio
OBOononorooorionoooooc
onnooooonaoo n nooooo oc nomo
(loooonooomnaoonocono^
o no oo onoponooonoopo
ooonooonnonoofloooo
o nioooponooo ooooor
30000POOO
n oooo noonoooo OOP
nno one oooooooo
aooooooooooonop
ooorinopnoioooo
aooooooopoioo
oooooorocnooi
oioooocooooo
oonooopooonoo
onooonooooonoooo
proooooooooooooooooonc
ODOoopciopoooonooooooooocc-
oooooonnooooooooooooooop
POOOOPOOOOO nooooo oooonooocipo
oaooipo • Hiii o
ooooooaao o o ooon
ooooopoooonopnonoooooooo^
aofloopopoioopoonrioooooocooPTOo
OIPPPOOOOPOOO OOPOOO 000000 000000
ororooooocor'Oopooo
POPPOCOOOOOO
nopoooroopnop
00 P^n?PPnPOPOOOP"?n 00 Pn 0000
cooooonooo
OOOCOOPPOC
opoooo
poooopcoopoccppnonooooo
ooooocoooonnoo'jn
ononoooojooooooP
ic-oorcfloao'iooooo
oa oponomooo oo cio oo ooooo 00001000000000000000000000001000000000000 ic OOOOOP
OOP nooooaoooopoonnoonoooopoop''oooooo no oooooo onofinonoopopppooppoocp
oooo aooofpzaooooooPCoMr oo oooo oc oriPooocnopnoorrjooQionoooppppo OP opon
OOOOOOOODPP OOOOOOOOO
01000 0010003 011 OOOn
. . __.
OOOOPPPOOPP"OOOi
OOOOOOnpnOODOOOOOOOOOOOnOPOIPOOOPOnoO"
OOOOOPOOO
OOOOCOPOP
pcopnoocr
PPOOOOOPP
OOOOOOOCOL'O
0000000000000

ooocopr
pnpnnnoppp
oboopoooopop
CODE:

In Mkrograms Per Cubic Meter And

Percent Of Land Area For Each Level
orcoonpnnnopOOOOO
oooooo OOOCOOPCOO
onoonponpcopo
ooooocpoooor
OPOOOODPOOOOO
Pniooomoropooon
oooooropocoaoooo
ooooonoooooooooo
OPpooooooroPOPOP
ooooCTPooocpOfmo
Onooooncpccooopp
ouooonoooooooootionoo
ooooio TOOOOOOOOO
onpooonoooponoooo
noooonoooopciopono I
OPPorrHOOP fmoooP»
ucnooooooooi
noon
OOPf)
oooo
coorpooooooat
aonooonpooa
pononnoooon
rnnooocooc
POfJflOOOUOO 001
OOOOPOOOOO pri

POOPOOO on ooooooooo
o oooopr
OOOOOOOOOOPOOOOOOO
pn
—
_____ ooonoorpooonoi
oooooonoooopopooooi
P.OOPO nppponopoo oo
ooooponoooropoooooo
oooonooooi ---------
1000000 poo-
01000 ________
looooooooooonooooi
cnoooooooooooooi
lonooroopooo ----
OOOO OOOOOO
FtCIRAL AIR QUALITY STANDARD- 75
00000000000060
oaacQoopoaoi —
ooooiinno
-------
CITY OF CHICAGO

Department Of Environmental Control

SUSPENDED PARTICULATE

JUNE 1972 THRU MAY 1973
CODE:
In Mkrooranu Per Cubic Meter And

Percent Off Und Area for Each Level
......... IP- W 1CU
.... 1 ..... i inn?nro
.........
.........

7SAUNOIR
HOItAL All QUALITY STANDARD -75
-------
CITY OF CHICAGO
Department Of Environmental Control
SUSPENDED PARTICULATE
JUNE 1973 THRU MAY 1974
CODE:
hi Mfcrograms For Cubic Motor Ami
Percent Of Land Area For Each Level
TSAUNOfl »-WO OVMIOO
»*% S7%
MM RAL All QUALITT STANDARD- 75
Tt»f
0.0
-------
CITY OF CHICAGO
Department Of Environmental Control
SUSPENDED PARTICIPATE
JUNE 1974 THRU MAY 1975
ccctccccccctccccccc
EEKKKfflirHHKE
rccaooo
88S
8S8
0
tec
ccccccc
cttcccctcc
IHHH
CODE:
In MUcrogranu Por Cubk Motor And
Percent Of Land Area For Each Level
»*UNDI«

44% S%

HDIRAl All QUALirr STANDARD-75
-------
SPEAKERS LIST
(NOTE: To facilitate their Identification, speakers are listed alphabetically
together with the name of the organization they represent. The complete address
of- each organization represented at the conference appears at the end of the list
ol: attendees.)
LIST OF SPEAKERS

Ncime

Axworthy, Dr. Arthur E.
Bittner, James D.
Bowen, Dr. Joshua A.
Bowman, Dr. Craig T.
Brown, Richard A.
Burchard, Dr. John K.
Cato, Glenn A.
Co Horn, Jr., Robert H.
Conbs, L. Paul
Crawford, Allen R.
Cuffe, Stanley T.
Dycema, Owen W.
England, Dr. Christopher
En;jleman, Dr. Victor S.
Giammar, Robert P.
Hall, Robert E.
Heap, Dr. Michael P.
Hei.ms, G. Thomas
Hol.linden, Dr. Gerald A.
Kasion, T.T.
Kendall, Dr. Robert M
Kesiselring, Dr. John P.
Ketels, Peter
Lac.hapelle, David G.
Lanier, W. Steven
McDonald, Henry
Maloney, Dr. Kenneth L.
Manny, Erwin H.
Martin, G. Blair
Muzio, Dr. Lawrence J.
Pohl, John H.
Poston, H. Wallace
Priticiotta, Frank
Rosenberg, Dr. Robert B.
Sarafim, Dr. Adel F.
Sel'cer, Ambrose P.
Shav, Dr. Robert
Shoffstall, Dr. Donald R.
Tyson, Dr. Thomas J.
Wasser, John H.
Wendt, Dr. Jost O.L.
Repres en ting

Rockwell International, Rocketdyne Division
Massachusetts Institute of Technology
EPA, IERL, Combustion Research Branch
United Technology Research Center
Acurex, Aerotherm Division
EPA, IERL
KVB, Inc.
State of Georgia, Department of Natural Resources
Rockwell International, Rocketdyne Division
Exxon Research and Engineering
EPA, Office of Air Quality Planning and Standards
The Aerospace Corporation
Jet Propulsion Laboratory
Exxon Research and Engineering
Battelle-Columbus Laboratories
EPA, IERL, Combustion Research Branch
Ultrasysterns
EPA, Region IV, Air and Hazardous Materials Division
Tennessee Valley Authority
City of Chicago, Department of Environmental Control
Acurex, Aerotherm Division
Acurex, Aerotherm Division
Institute of Gas Technology
EPA, IERL, Combustion Research Branch
EPA, IERL, Combustion Research Branch
United Technology Research Center
KVB, Inc.
Exxon Research and Engineering
EPA, IERL, Combustion Research Branch
KVB, Inc.
Massachusetts Institute of Technology
City of Chicago, Department of Environmental Control
EPA, Energy Processes Division
Institute of Gas Technology
Massachusetts Institute of Technology
Combustion Engineering
'Stanford Research Institute
Institute of Gas Technology
Ultrasystems
EPA, IERL, Combustion Research Branch
University of Arizona
A-l
-------
PARTICIPANTS LIST
(NOTE: To facilitate their identification, participants are listed alphabetically
together with the name of the organization they represent. The complete address
of each organization represented at the conference appears at the end of the list
of attendees.)
LIST OF PARTICIPANTS

Name

Alvey, Courtney D.
Anderson, Dr. Larry W.
Axtman, William H.
Bagwell, Fred A.
Baker, Burke
Ban, Stephen D.
Barrett, Richard E.
Barsin, Joseph
Bartok, William
Batra, Sushil K.
Bauman, Robert D.
Beals, Rixford A.
Beatty, James D.
Bennett, Dr. Robert
Blandford, Jr., J.B.
Blythe, R. Allen
Buechler, Lester
Bonne, Ulrich
Booth, Michael R.
Bowman, Barry R.
Bueters, K.A.
Carpenter, Ronald C.
Cernansky, Dr. Nicholas P.
Christiano, John P.
Chu, Richard R.
Clark, Norman D.
Cleverdon, R.F.
Cotton, Ernest
Creekmore, Andrew T.
Daughtridge, Jimmy T.
Degler, Gerald H.
Demetri, E.P.
DeWerth, D.W.
Dingo, T.T.
Donaldson, Thomas M.
Dowling, Daniel J.
Downey, Thomas A.
Dyer, T. Michael
Dygert, J.C.
Representing

Baltimore Gas & Electric
Acurex, Aerotherm Division
American Boiler Manufacturers Association
South California Edison
Shell Development Company
Battelle-Columbus Laboratories
Battelle-Columbus Laboratories
Babcock & Wilcox
Exxon Research & Engineering
New England Electric Systems
EPA, Office of Air Quality, Planning & Standards
NOFI
Procter & Gamble
Apollo Chemical
Englehard Industries
International Boiler Works
Systems Research Labs
Honeywell
Ontario Hydro
Lawrence Livermore Laboratories
Combustion Engineering
Armstrong Cork
Drexel University
EPA, Office of Air Quality, Planning & Standards
EBASCO
C-E Air Preheater
Chevron Research Company
American Petroleum Institute
EPA, Control Programs Development Division
Pratt & Whitney Aircraft
Systems Research Labs
Northern Research and Engineering Corporation
American Gas Association Labs
General Motors
EPA, Office of Air Quality, Planning & Standards
Union Carbide
Gamlen Chemical Company
Sandia Laboratories
Shell Development Company
A-2
-------
LIST OF PARTICIPANTS (CONT'D)
Name

Dzuna, Eugene R.
Erskine, George
Feng, C.L.
Fennelly, Paul F.
Fletcher, James
Fletcher, Roy J.
Freelain, Kenneth
Frisch, Dr. N.W.
Fuhrman, Jr., Theodore C.
Glbbs, Thomas
Goetz, Gary
Goodley, Allan R.
Graham, David J.
Greene, Jack H.
Grimshaw, Vincent C.
Grossman, Ralph
Hangebrauck, Robert P.
Heck, Ronald
Hensel, Thomas E.
Holden, Edward A.
Honea, Dr. Franklin I.
Howard, Jack B.
Hudson, Jr., James L.
Jackson, Dr. A.W.
Jepson, Dr. A.F.
Karas, Dennis T.
Kemmerer, Jeffrey
Khan, M. Ali
Khoo, Dr. S.W.
Kloecker, J.F.
Kykendal, William
Lahre, Thomas
Lange, Dr. Howard
Lavoie, Raymond C.
Lee, James E.
Lenney, Ronald J.
Levy, Arthur
Lewis, F. Michael
Lii, Dr. C.C.
Lin, Donald J.L.
Locklin, David W.
Lord, Harry C.
Loweth, Carl
Marshall, David
Marshall, John H.
Representing

Gulf Research & Development Company
Mitre Corporation
Selas Corporation of America
GCA/Technology Division
Industrial Combustion
Peabody Engineering Corporation
Federal Energy Administration
Research-Cottrell
Erie City Energy Division
EPA, Region IV
Combustion Engineering
California Air Resources Board
EPA, Office of Research and Development
EPA, Administrative Office
Process Combustion
Ralph Grossman, Ltd.
EPA, Energy Assessment and Control Division
Englehard Industries
Turbo Power and Marine Systems
General Foods Corporation
Midwest Research Institute
Massachusetts Institute of Technology
Tampa Electric Company
Ontario Hydro
Environmental Measurements, Inc.
East Chicago Air Quality Control
Fuller Company
East Chicago Air Quality Control
Canadian Gas Research Institute
Erie City Energy Division
EPA, Process Measurement Branch
EPA, Office of Air Quality, Planning & Standards
Babcock & Wilcox
Rohm & Haas Company
Facilities Engineering Command (U.S. Navy)
Ronald J. Lenney Associates
Battelle-Columbus Laboratories
Stanford Research Institute
EPA, Combustion Research Branch
Forney Engineering Company
Battelle-Columbus Laboratories
Environmental Data Corporation
The Trane Company
Babcock & Wilcox
Combustion Engineering
A-3
-------
LIST OF PARTICIPANTS (CONT'B)
Name

Marton, Miklos B.
Mayfield, D. Randell
Meier, John G.
Moore, Douglas S.
Moore, Edward E.
Morton, William J.
Moscowitz, Charles
Mosier, Stanley A.
Newton, Charles L.
Nurick, W.H.
Pantzer, Karl
Pershing, David W.
Pertel, Dr. Richard
Renner, Ted
Riley, Joseph
Robert, J.
Roberts, Dr. George
Robertson, J.F.
Roffe, Gerald
Rosen, Meyer
Ross, Marvin
Rulseh, Roy
Sadowski, R.S.
Samples, J.R.
Scott, Donald R.
Sheffield, E.W.
Slack, A.V.
Smith, Lowell L.
Spadaccini, L.J.
Sterman, Sam
Sullivan, Robert E.
Swearingen, W.E.
Takacs, Dr. L.
Taylor, Barry R.
Utterback, Paul M.
Van Grouw, Sam J.
Vatsky, Joel
Vershaw, Jim
Watson, Raymond A.
Webb, R.
Weiland, J.H.
Weinberger, Dr. Lawrence
White, David J.
White, James H.
White, Phil
Representing

IBM
EPA, Region IV
International Harvester, Solar Division
Chevron Research Company
Eclipse, Inc.
E. Keeler Company
Monsanto Research Corporation
Pratt & Whitney
Colt Industries
Rockwell International, Rocketdyne Division
Babcock & Wilcox
University of Arizona
Institute of Gas Technology
Fuel Merchants Association of New Jersey
EPA, Region IV
Canadian Department of Environment
Englehard Industries
Crystal Petroleum Company
General Applied Science Laboratories
Union Carbide
Lawrence Livermore Laboratory
Cleaver-Brooks
Riley Stoker Corporation
Union Carbide
Columbia Gas System Service Corporation
TRW
SAS Corporation
KVB, Inc.
United Technology Research Center
Union Carbide
General Motors
Koppers Company
General Motors
Massachusetts Institute of Technology
Babcock & Wilcox
KVB, Inc.
Foster-Wheeler Energy Corporation
The Trane Company
Florida Power & Light Company
The Trane Company
Texaco, Inc.
Mitre Corporation
International Harvester, Solar Division
Coen Company
Ventura Company
A-4
-------
Name

Wiedersum, George C.
Wilhelm, Ronald
Wilson, Jr., R.P.
Winters, Harry K.
Wittig, Dr. Sigmar L.K.
WooIfoik, Dr. Robert
Wright, Richard
Young, Dexter E.
Ziarkowski, Stanley
Zielke, Robert L.
Zirkel, Eric C.
LIST OF PARTICIPANTS (CONT'D)

Representing

Philadelphia Electric Company
Aqua-Chem, Inc.
Arthur D. Little, Inc.
Ray Burner Company
Purdue University
Stanford Research Institute
Industrial Combustion
EPA, Control Programs Development Division
Garnien Chemical Company
Tennessee Valley Authority
Armstrong Cork
A-5
-------
LIST OF ORGANIZATIONS REPRESENTED
Name (Represented By)

Acurex Corporation, Aerotherm Division
(Mr. Anderson)

American Boiler Manufacturers Association
(Mr. Axtman)

American Gas Association Labs
(Mr. De Werth)

American Petroleum Institute
(Mr. Cotton)

Apollo Chemical
(Dr. Bennett)

Aqua-Chem, Inc.
(Mr. Wilhelm)

Armstrong Cork Company
(Mr. Carpenter, Mr. Zirkel)

Babcock & Wilcox
(Mr. Barsin, Mr. Lange, Mr. I-core)

(Mr. Utterback, Mr. Pantzer,
Mr. Marshall)

Baltimore Gas and Electric Company
(Mr. Alvey)

Battelle-Columbus Labs
(Mr. Ban, Mr. Barrett,
Mr. Levy, Mr. Locklin)

C-E Air Preheater
(Mr. Clark)

California Air Resources Board
(Mr. Goodley)

Canadian Department of Environment
(Mr. Robert)

Canadian Gas Research Institute
(Dr. Khoo)

Chevron Research Company
(Mr. Cleverdon, Mr. D. Moore)
Address

485 Clyde Avenue
Mountain View, California 94042

1500 Wilson Boulevard, Suite 317
Arlington, Virginia 22209

8501 East Pleasant Valley Road
Cleveland, Ohio 44131

1801 K Street, N.W.
Washington, D.C. 20006

35 South Jefferson Road
Whippany, New Jersey 07981

P.O. Box 421
Milwaukee, Wisconsin 53201

Liberty & Charlotte Streets
Lancaster, Pennsylvania 17604

20 South Van Buren Avenue
Barberton, Ohio 44203

P.O. Box 2423
North Canton, Ohio 44720

2012 Gas and Electric Building
Baltimore, Maryland 21203

505 King Avenue
Columbus, Ohio 43201
Andover Road
Wellsville, New York 14895

1709 llth Street
Sacramento, California 95814

351 St. Joseph Boulevard
Houll, Quebec, Canada

55 Scarsdale Road, Don Mills
Ontario, M3B2R3, Canada

P.O. Box 1627
Richmond, California 94802
A-6
-------
LIST OF ORGANIZATIONS (CONT'D)
Name (Represented By)

Cleaver-Brooks
(Mr. Rulseh)

Coen Company
(Mr. J.H. White)

Colt Industries
(Mr. Newton)

Columbia Gas System Service Corporation
(Mr. Scott)

Combustion Engineering
(Mr. Bueters, Mr. Goetz, Mr. Marshall)

Crystal Petroleum Company
(Mr. Robertson)

Drexel University
(Dr. Cernansky)

EBASCO
(Mr. Chu)

East Chicago Air Quality Control
(Mr. Karas, Mr. Khan)

Eclipse, Inc.
(Mr. E. Moore)

Englehard Industries
(Mr. Blandford, Dr. Heck, Dr. Roberts)

Environmental Data Corporation
(Mr. Lord)

Environmental Measurements, Inc.
(Dr. Jepsen)

Environmental Protection Agency
EPA - Administrative Office
(Mr. Greene)
Address

3707 North Richards Street
Milwaukee, Wisconsin 53201

1510 Rollins Road
Burlingame, California 94010

701 Lawton Avenue
Beloit, Wisconsin 53511

1600 Dublin Road
Columbus, Ohio 43215

1000 Prospect Hill Road
Windsor, Connecticut 06095

P.O. Box 4180
Corpus Christi, Texas 78408

Philadelphia
Pennsylvania 19104

145 Technology Park
Norcross, Georgia 30071

900 East Chicago Avenue
East Chicago, Indiana 45312

1100 Buchanan
Rockford, Illinois 61101

Middlesex Turnpike, Wood Avenue
Edison, New Jersey 08876

608 Fig Avenue
Monrovia, California 91016

2 Lincoln Court
Annapolis, Maryland 21401
Research Triangle Park
North Carolina 27711
A-7
-------
LIST OF ORGANIZATIONS (CONT'D)
Name (Represented By)

EPA - Combustion Research Branch
(Dr. Bowen, Mr. Hall, Mr. Lachapelle
Mr. Lanier, Dr. Lii, Mr. Martin, Mr. Wasser)

EPA - Control Programs Development Division
(Mr. Creekmore, Mr. Young)

EPA - Energy Assessment & Control Division
(Mr. Hangebrauck)

EPA - Office of Air Quality, Planning,
and Standards
(Mr. Bauman, Mr. Christiano,
Mr. Donaldson, Mr. Lahre)

EPA - Office of Research and Development
(Mr. Graham)

EPA - Process Measurement Branch
(Mr. Kuykendal)

EPA - Region IV
(Mr. Biggs, Mr. Mayfield, Mr. Riley)

Erie City Energy Division
(Mr. Fuhraan, Mr. Kloecher)

Exxon Research & Engineering Company
(Mr. Bartok)

Federal Energy Administration
(Mr. Freelain)

Florida Power & Light Company
(Mr. Watson)

Forney Engineering Company
(Mr. Lin)

Foster Wheeler Energy Corporation
(Mr. Vatsky)

Fuel Merchants Association of New Jersey
(Mr. Renner)
Address

Research Triangle Park
North Carolina 27711
Research Triangle Park
North Carolina 27711

Research Triangle Park
North Carolina 27711
Research Triangle Park
North Carolina 27711
Washington, D.C. 20460
Research Triangle Park
North Carolina 27711

1421 Peachtree Street, N.E.
Atlanta, Georgia 30309

1422 East Avenue
Erie, Pennsylvania 16502

P.O. Box 8
Linden, New Jersey 07036

1200 Pennsylvania Avenue, N.W.
Washington, D.C. 20461

P.O. Box 013100
Miami, Florida 33101

P.O. Box 189
Addison, Texas 75001

10 South Orange Avenue
Livingston, New Jersey 07039

66 Morris Avenue
Springfield, New Jersey 07081
A-8
-------
LIST OF ORGANIZATIONS (CONT'Dl
Name_(Represented By)

Fuller Company
(Mr. Kemmerer)

GCA/Technology Division
(Dr. Fennelly)

Gamlen Chemical Company
(Mr. Downey, Mr. Ziarkowski)

General Applied Science Laboratories
(Mr. Roffe)

General Foods - Technical Center
(Mr. Holden)

General Motors Corporation
(Mr. Sullivan)

(Mr. Dingo, Mr. Takacs)
Gulf Research and Development Company
(Mr. Dzuna)

Hcneywell, Inc.
(Mr. Bonne)

IBM
(Mr. Marton)

Industrial Combustion
(Mr. Wright, Mr. Fletcher)

Institute of Gas Technology
(Dr. Pertel)

In:ernational Boiler Works
(Mr. Blythe)

International Harvester, Solar Division
(Mr. Meier, Mr. D.G. White)
Address

124 Bridge Street
Catasauqua, Pennsylvania 18032

Burlington Road
Bedford, Massachusetts 01730

299 Market Street
Saddle Brook, New Jersey 07662

Merrick & Stewart Avenues
Westbury, New York 11790

250 North Street
White Plains, New York 10625

5735 West 25th Street
Indianapolis, Indiana 46224

Technical Center
Warren, Michigan 48090

P.O. Box 2038
Pittsburgh, Pennsylvania 15230

Bloomington
Minnesota 55420

1000 Westchester Avenue
White Plains, New York 10604

4465 North Oakland
Milwaukee, Wisconsin 53211

3424 South State Street
Chicago, Illinois 60616

P.O. Box 498
East Stroudsburg, Pennsylvania 18301

2200 Pacific Highway
San Diego, California 92119
A-9
-------
LIST OF ORGANIZATIONS (CONT'D)
Name (Represented By)

KVB Engineering, Inc.
(Mr. L. Smith, Mr. Van Grouw)

E. Keeler Company
(Mr. Morton)

Koppers Company, Inc.
(Mr. Swearingen)

Ronald J. Lenney Associates
(Mr. Lenney)

Arthur D. Little, Inc.
(Mr. R.D. Wilson)

Lawrence Livermore Laboratories
(Mr. Bowman, Mr. Ross)

Massachusetts Institute of Technology
(Mr. Howard, Mr. Taylor)

Midwest Research Institute
(Dr. Honea)

Mitre Corporation
(Dr. Weinberger, Mr. Erskine)
Monsanto Research Corporation
(Mr. Moscowitz)

NOFI
(Mr. Seals)

Naval Facilities Engineering Command,
Southern Division
(Mr. J. Lee)

New England Electric Systems
(Mr. Batra)

Northern Research & Engineering Corporation
(Mr. Demetri)
Address

6624 Hornwood Drive
Houston, Texas 77036

Williamsport
Pennsylvania 17701

Koppers Building
Pittsburgh, Pennsylvania 15219

2001 Palmer Avenue
Larchmont, New York 10538

Acorn Park
Cambridge, Massachusetts 02140

P.O. Box 808
Livermore, California 94550

Massachusetts Avenue
Cambridge, Massachusetts 02139

425 Volker Boulevard
Kansas City, Missouri 64110

Westgate Research Park
1820 Dolly Madison Boulevard
McLean, Virginia 22101

Station B. Box 8
Dayton, Ohio 45407

New York, New York
2144 Melbourne Street
P.O. Box 10068
Charleston, South Carolina 29411

20 Turnpike Road
Weston, Massachusetts 01581

219 Vassar Street
Cambridge, Massachusetts 02139
A-10
-------
LIST OF ORGANIZATIONS (CONT'D)
Name (Represented By)

Ontario Hydro Corporation
(Mr. Booth, Dr. Jackson)

Peabody Engineering Corporation
(Mr. R. Fletcher)

Philadelphia Electric Company
(Mr. Wiedersom)

Pratt & Whitney Aircraft
(Mr. Daughtridge, Mr. Mosier)

Process Combustion Corporation
(Mr. Gritnshaw)

Procter & Gamble Company
(Mr. Beatty)

Purdue University
(Mr. Wittig)

Ralph Grossman, Ltd.
(Mr. Grossman)

Ray Burner Company
(Mr. Winters)

Res^arch-Cottrell, Inc.
(Dr. Frisch)

Riley Stoker Corporation
(Mr. Sadowski)

Rockwell International Corporation,
Rocketdyne Division
(Mr. Nurick)

Rohm & Haas Company
(Mr. Lavoie)

SAS Corporation
(Mr. Slack)
Address

620 Union Avenue
Toronto, Ontario, Canada M561X6

835 Hope Street
Stamford, Connecticut 06907

2301 Market Street, S10-1
Philadelphia, Pennsylvania 19101

P.O. Box 2691
West Palm Beach, Florida 33402

1675 Washington Road
Pittsburgh, Pennsylvania 15228

610 South Center Hill Road
Cincinnati, Ohio 45224

West Lafayette
Indiana 47907

P.O. Box 70, Town of Mt. Royal
Montreal, Canada H3P 3B8

1301 San Jose Avenue
San Francisco, California 94112

P.O. Box 750
Boundbrook, New Jersey 08805

9 Neponset Street
Worcester, Massachusetts 01613

6633 Canoga Avenue
Canoga Park, California 91304
P.O. Box 584
Bristol, Pennsylvania 19007

RFD #1
Sheffield, Alabama 35660
A-H
-------
LIST OF ORGANIZATIONS (CONT'D)
Naiue (nfr iresentetl By)

Sandia Laboratories
(Mr. Dyer)

Selas Corporation of America
(Mr. Feng)

Shell Development Company
(Mr. Dygert, Mr. Baker)

South California Edison
(Mr. Bagwell)

Stanford Research Institute
(Dr. Woolfolk, Mr. Lewis)

System-! Re^arch Labs
(Mr. Buechler, Mr. Degler)

TRW, Inc.
(Mr. Sheffield)

Tampa Electric Company
(Mr. Hudson)

Tennessee Valley Authority
(Mr. Zielke)

Texaco, Inc.
(Mr. Weiland)

The Trane Cor-^-my
(Mr. Loweth, Mr. Vershaw, Mr. Webb)

Turbo Power and Marine Systems
(Mr. Hensel)

Union Carbide Corporation
(Mr. Rosen, Mr. Sterman)

(Mr. Bowling)
Address

Livermore
California 94550

Dresher
Pennsylvania 19025

P.O. Box 481
Houston, Texas 77001

P.O. Box 800
Rosemead, California 91770

1611 North Kent Street
Arlington, Virginia 22209

2800 Indian Ripple Road
Dayton, Ohio 45440

1 Space Park - R4/2020
Redondo Beach, California 90278

P.O. Box 111
Tampa, Florida 33601

524 Power Building
Chattanooga, Tennessee 37401

P.O. Box 509
Beacon, New York 12508

3600 Pammel Creek Road
La Crosse, Wisconsin 54601

1690 New Britain Avenue
Farmington, Connecticut 06032

Tarrytown Technical Center
Tarrytown, New York 10591

Box 180
Sistersville, West Virginia 26175
A-12
-------
LIST OF ORGANIZATIONS (CONT'D)
Nane (Represented By)
Address
Union Carbide Corporation
(Mr. Samples)

United Technologies Research Center
(Mr. Spadaccini)

University of Arizona
(Mr. Pershing)

Ventura County A.P.C.D.
(Mr. P. White)
Box 4361
South Charleston, West Virginia 25353

400 Main Street
East Hartford, Connecticut 06040

Tucson
Arizona 85721

740 East Main Street
Ventura, California 93001
•U.S GOVERNMENT PR1NTIKC OFFICE: ly?f,-MJ.Jl'/SSH Bejion So. 4
A-13
-------
TECHNICAL REPORT DATA
(Please read Instruction* on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-152a
2.
3. RECIPIENT'S ACCESSION-NO,
4 TITL6ANOSUBTITI-E Proceedings of the Stationary Source
Combustion Symposium; Volume I—Fundamental
Research
S. REPORT DATE
June 1976
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)

Miscellaneous
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
NA
10. PROGRAM ELEMENT NO.
1AB014; ROAP 21BCC
11. CONTRACT/GRANT NO.

NA (In-house)
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings; 9/24-26/75
14. SPONSORING AGENCY CODE
EPA-ORD
5. SUPPLEMENTARY NOTEsgymposium Chairman J.S. Bowen, Vice-Chair man R.E. Hall,
Mail Drop 65, Ext. 2470/2477.
ABSTRACT The proceedings document the 37 presentations made during the Stationary
Source Combustion Symposium held in Atlanta, Ga. , September 24-26, 1975. Spon-
sored by the Combustion Research Branch of EPA's Industrial Environmental Resea-
rch Laboratory—RTP, the symposium dealt with subjects related both to developing
improved combustion technology for the reduction of air pollutant emissions from
stationary sources, and to improving equipment efficiency. The symposium was
divided into four parts and the proceedings were issued in three volumes: Volume I--
Fundamental Research, Volume n—Fuels and Process Research and Development,
and Volume m--Field Testing and Surveys. The symposium was intended to provide
contractor, industrial, and Government representatives with the latest information
on EPA in-house and contract combustion research projects related to pollution
control, with emphasis on reducing nitrogen oxides.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C, COSATI Field/Group
Air Pollution, Combustion, Field Tests
ombustion Control, Coal, Oils
Natural Gas, Nitrogen Oxides, Carbon
arbon Monoxide, Hydrocarbons, Boilers
Pulverized Fuels, Fossil Fuels, Utilities
Gas Turbines, Efficiency
Air Pollution Control
Stationary Sources
Combustion Modification
Unburned Hydrocarbons
Fundamental Research
Fuel Nitrogen
Burner Tests
13B
2 IB 14B
2 ID 11H
07B
07C 13A
13G 14A
8. DISTRIBUTION STATEMENT

Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
488
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
A-14
-------