SEPA
         United States
         Environmental Protection
         Agency
          Industrial Environmental Research
          Laboratory
          Research Triangle Park NC 27711
EPA-600 7-80-038
March 1980
Photochemical Study of
IMOx Removal from
Stack Gases
          nteragency
         Energy/Environment
         R&D Program Report

-------
                  RESEARCH REPORTING SERIES


 Research reports of the Office of Research and Development, U.S. Environmental
 Protection Agency, have been grouped into nine series. These nine broad cate-
 gories were established to facilitate further development and application of en-
 vironmental technology.  Elimination  of  traditional grouping was consciously
 planned to foster technology transfer and a maximum interface in related fields.
 The nine series are:

     1. Environmental Health Effects Research

     2. Environmental Protection Technology

     3. Ecological Research

     4. Environmental Monitoring

     5. Socioeconomic Environmental Studies

    6. Scientific and Technical Assessment  Reports (STAR)

    7. Interagency Energy-Environment Research and Development

    8. "Special" Reports

    9. Miscellaneous Reports

 This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
 RESEARCH AND DEVELOPMENT series. Reports in this series result from the
 effort funded under the 17-agency Federal Energy/Environment Research and
 Development Program. These studies relate to EPA's mission to protect the public
 health and welfare from adverse effects of pollutants associated with energy sys-
 tems. The goal of the Program is to assure the  rapid  development of domestic
 energy supplies in an environmentally-compatible manner by providing the nec-
 essary environmental data and control technology. Investigations include analy-
 ses of the transport of energy-related pollutants  and their health and ecological
 effects; assessments  of, and development of, control technologies for  energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.



                        EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for  publication. Approval does not signify that the contents necessarily reflect
the  views and policies of the Government, 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-600/7-80-038
                                            March 1980
Photochemical Study of  NOx
  Removal from  Stack  Gases
                     by

           John R. Richards and Donald L. Fox

             University of North Carolina
              School of Public Health
           Chapel Hill, North Carolina 27514
                Grant No R804740
             Program Element No. INE623
         EPA Project Officer: Joseph A. McSorley

        Industrial Environmental Research Laboratory
     Office of Environmental Engineering and Technology
           Research Triangle Park, NC 27711
                  Prepared for

       U.S. ENVIRONMENTAL PROTECTION AGENCY
          Office of Research and Development
               Washington, DC 20460

-------
                              ABSTRACT
      The technical feasibility of a photochemical pretreatment system
 for NO  control at coal-fired boilers was evaluated.  This approach uti-
       n
 lizes reaction mechanism similar to those responsible for photochemical
 oxidant incidents.  The reactions are initiated under controlled condi-
 tions while the pollutants are at high concentration and while the reac-
 tion products can be removed.  Results indicated that under time limited
 and light limited conditions, it is possible to quench the photochemical
 reactions at the NOo peak and prior to the formation of ozone, aerosols,
 and other secondary products.
      Energy and hydrocarbon requirements  were estimated in a  series  of
 159 experimental  runs.   The optimal  operating conditions were identified
 as a NO  /C3H6 stoichiometric ratio  of 1 and an N02  photolysis rate of
 1  to 1.5 reciprocal  minutes which is  equivalent to  approximately  3 to 5%
 station  power.   These  conditions allowed  for an &3% oxidation of  NO  with
 an exit  concentration  of NO less than  100 ppm.   Variation  of  the  energy
 input levels  or the  propylene injection rates  resulted  in  NO  oxidation
 efficiencies  of 10 to  99%.   Photochemical  oxidation of  NO  was  insensitive
 to SOp concentration and C02  concentration.   Conversion  was enhanced at
 low  initial NO   levels,  high  temperatures,  and  high CO  concentrations.
              J\
 Oxygen concentrations above 2% were necessary.   The photochemical system
 appears compatible with  conditions resulting  from combustion modifica-
 tions to suppress  NO  generation.
                    A
     Reactor products consisted primarily of NOg. formaldehyde, acetalde-
 hyde, and acetic acid all of which should be removed in  the downstream
 flue gas desulfurization system.  Mass spectrometer analyses  indicated
 the presence of a large number of organic compounds; nevertheless, all
of these were present in only trace quantities.  A major fraction of the
 injected propylene was also emitted.  It was concluded that propylene is
not sufficiently reactive for the photochemical system.  Other hydrocar-
                                ii

-------
bons are recommended for further study.
     Capital and annuallzed cost estimates have been prepared for a
model coal-fired power plant of 500 MW generating capacity.  Depending
on the extent of NO oxidation desired, the capital costs v*ry from $37/
KW to $50/KW.  These estimates include the photochemical reactor with
mercury lamps, and the propylene storage and distribution system.  The
annualized costs estimates range from 3.8 to 5.6 mills/kWh.  The upper
value corresponds to &500 ppm NO while the lower value is for A200 ppm
NO.
     The cost estimates are considered high due to the poor reactivity of
propylene and the low efficiency of presently available lamps.  Further
bench scale testing could conceivably yield a photochemical system re-
quiring only 1/2 the annualized cost of the present system.  Even with
the present limitations, the costs for the photochemical system are com-
petitive with other control techniques involving NO gas phase oxidation
followed by N02 absorption.  Costs for alternative approaches such as
selective noncatalytic reduction, however, are substantially lower.  It
is concluded that the photochemical system will  not be commercially com-
petitive unless serious operational problems limit the-alternative tech-
niques or unless the photochemical system can be refined to reduce cost.
     This report was submitted in fulfillment of Grant No.  R804740
by the Department of Environmental  Sciences and Engineering of the Uni-
versity of North Carolina.   The study was performed under the sponsor-
ship of the Industrial  Environmental  Research Laboratory of the U.S.
Environmental  Protection Agency.   This report covers the period of
December 27,  1976,  to December 27, 1978,  and work was completed as of
February 28,  1979.
                               ill

-------
                           ACKNOWLEDGMENTS
      This  report  was  prepared  for the Industrial  Environmental  Research
 Laboratory of the U.S.  Environmental  Protection Agency.   The  EPA  Project
 Officer was Mr. Joseph  McSorley.   UNC appreciates  the  direction and as-
 sistance provided by  Mr. McSorley.
      The authors  gratefully acknowledge  the assistance of Mr. Frank
 Malcolm in  the fabrication of  the  reactor assembly and the gas blending
 equipment.  X-ray fluorescence analyses  and liquid chromotography tests
 of the Millipore  filter samples were  performed by the U.S.E.P.A. Environ-
 mental Research Laboratory.  The GC mass  spectrometer tests were done
 under contract at  Research Triangle Institute by Dr. E. Pellizzari.
     The principal investigator was Dr.  Donald L.  Fox.  The co-principal
 investigator and  principal  author of  the  report was Mr. John Richards.
 Consulting support was provided by Dr. Harvey Jeffries, Dr.  E. Stahel,
and Dr.  Parker Reist.
                                iv

-------
                             CONTENTS


                                                                Page
Abstract	  ii
Acknowledgments 	  iv
Figures 	 vii
Tables 	   x
    1.   Introduction	   1

    2.   Summary and  Conclusions	   4

    3.   Recommendations	   8

    4.   Background	   9

           NO   Control  Techniques	   9
            /\
           Photochemical NO  Control Techniques	  26
                          J\
    5.   Experimental	  43

           General  Design Objectives	  43

           Reactor  Design and  Performance	  45

           Acti nometry	  L5

           Reactant and  Product Measurement	  f8

           Gas  Handl ing  Apparatus	  62

           Experimental  Design	  ("6

           Test  Procedure	  £7

    6.   Results and Discussion	  71

           Spectral Effects	  ^

           Hydrocarbon Requirements	79

-------
          Variables Influencing NO Oxidation	91
          Reaction Products	102
          Cost Estimates	117
References	126
Appendices
                               vi

-------
                               FIGURES


 Number                                                            Page


 1.  Trends in the emissions of NOX from stationary sources 	   10

 2.  Possible formation mechanism for fuel  NOV 	   14
                                             /\

 3.  Effect of excess air on NOo emissions   	   16

 4.  Effect of secondary air addition on NOX emissions  	   17

 5.  Classification of NOX flue gas treatment processes  	   20

 6.  Effect of temperature on NOX removal in the selective
        noncatalytic reduction process  	   23

 7.  Photochemical system conceptual  design  	   29

 8.  Absorption coefficients of flue  gas components  	   33

 9.  Power requirements as a function of net product quantum
        yield  	   35

IQa.  Photostationary state reactions   	   37

lOb.  Simplified olefin/NOx photochemical reactions  	   37

lOc  Simplified aldehyde photochemical reactions  	   37

11.  Typical pollutant concentration-time profile for
        photochemical reactions in batch reactor  	   40

12a.  View of quartz reactor on support rod  	   46

12b.  View of outer CSTR  	   46

13.  Mercury doped xenon arc output spectra  	   47

14.  Light distribution in quartz reactor  	   49

15.  N(>2 photolysis rate as a function of lamp current  	   51

16.  Initial inlet configuration  	   54
                                 vil

-------
 Number                                                         pdge

 16b.  Reactor  inlet configuration   	• 54

 17.   Gas analyzers for NOV, SO?, and condensation nuclei  	 59
                         A    ^-

 18.   Gas analyzers for hydrocarbons  	 59

 19.   Influence of N02 on pulsed fluorescent S02 readings  	 61

 20.   Side view of reactor and spectrometer  	 63

 21.   Front view of blending box  	 63

 22.   Gas flow system  	  65

 23.   Influence of middle ultraviolet light on NO oxidation  	  73

 24.   Typical comparative spectra;  during and after
        experimental  run  	  75

 25.   Influence of NOX on absorption spectra 	  76

 26.   Influence of $62 on NO oxidation  	  78

 27.   NO oxidation,  influence of initial propylene concentration
        and light intensity.  Low  initial  NOX concentration	  81

 28.   NO oxidation,  influence of initial propylene concentration
        and light intensity.  High  initial  NOX concentration.  ...  82

29.  NO oxidation,  influence of initial propylene concentration
        and light intensity.  Low  initial  NOx concentration	  83

30.  NO oxidation,  influence of initial propylene concentration
        and light intensity.  High  initial  NOx concentration.  ...  84

31.  Comparason  of  propylene and ethylene   	  86

J2.  Unreacted propylene concentration  at  various inlet NOX
        levels and  light intensities   	  88
33.   NO oxidation and propylene consumption as a function of N02
        photolysis rate  	  89

34.   Effect of oxygen content on NO oxidation  	  92

35.   Effect of CO concentration on NO oxidation  	  96

36.   Effect of initial  NOX concentration on NO oxidation  	  99
                                viii

-------
Number                                                          Page





37.  Effect of C02 concentration on NO oxidation  	101



38.  Influence of temperature on NO oxidation  	 103



39.  Nitrate deposits  	 iny



40.  Sulfate deposits  	 Ill



41.  Typical gas chromatograph output  	Ill



42.  Aldehyde/propylene decay in sampling bags  	 11^
                                 lx

-------
                               TABLES






 Number                                                            Page






 1  Typical Flue Gas Composition of Coal -Fired Boilers .............  11



 2  Effect of Combustion Modification Techniques on the Emission

       of Polycycl i c Organi c Matter ................................  21



 3  Photochemi cal Reactor Desi gn Assumptions .......................  30



 4  Effluent Gas Composition Assumptions ...........................  30



 5  Energy Del ivery in Model System ................................  34



 6  Comparison of Experimental System and Full Scale Systems .......  44



 7  Tracer Results .................................................  53



 8  Experimental Design ............................................  68



 9  Influence of Oxygen Content on Propylene Consumption ...........  94



lOa Nitrogen Material  Balance, [NO ]0 ^600 PPM ..................... 105
                                  /\


lOb Nitrogen Material  Balances, [NO ]0 ^1200 PPM (v/v) ............. 105
                                   A


11   Nitrate Aerosol  Quantities Compared with NO Oxidation .......... 108



12  Sulfur Material  Balance ........................................ 108



13  X-Ray Fluorescent Sulfur Measurement Expressed as APPM  S02 ..... 112
14a Carbon Material  Balances, [N0v]0 ^QQ PPM (v/v) ................  112
                                 A



14b Carbon Material  Balances, [NO ]0 V1200 PPM (v/v) ...............  112
                                 A



1 5  Carbon Bal ance Correct! ons for Al dehydes .......................  114




16  Relationship Between Aldehyde Formation and NO Oxidation .......  lie




17  Major Organic Compounds Determined in GC/MS Tests, PPM (v/v) ---- -113
                                  X

-------
 Number                                                         Page
 18.  Capital  Cost Estimates 	  122
 19.  Annualized Cost Estimates 	  123
 20.  Comparison of Costs for Flue Gas  Treatment Processes  	  124
B-l.  Experimental Data, Influence of Middle UV  Filter
          on  NO Oxidation  	  136
B-2.  Experimental Data,Influence of Sulfur Dioxide  on NO
          Oxidation 	  138
B-3.  Experimental Data, Influence of Propylene  Concentration
          and  UV Light Intensity on NO  Oxidation 	  141
B-4.  Experimental Data, Comparison of  Ethylene  and  Propylene  ..  166
B-5.  Experimental Data, Influence of Oxygen Content on
          NO  Oxidation 	168
B-6.  Experimental Data, Influence of CO on NO Oxidation 	170
B-7.  ExDerimental Data, Influence of Initial NOX Concentration
          on  NO Oxidation 	171
B-8.  Experimental Data, Influence of C02 Concentration
          on  NO Oxidation 	173
B-9.  Experimental Data, Influence of Gas Temperature
          on  NO Oxidation 	  174
                                  xi

-------
                            CHAPTER 1
                           INTRODUCTION

GENERAL
     Both utility and industrial  boilers are major sources  of  anthropo-
genic nitrogen oxide emissions, contributing over 40%  of total yearly
emissions in the United States.  These primary nitrogen  oxides (NOX)
emissions react chemically with hydrocarbons in the ambient air  through
a set of partially understood photochemical  reaction mechanisms.   Conse-
quences of these reactions include elevated  ozoria levels and potentially
toxic secondary pollutants.  Previous  photochemical  oxidant control  stra-
tegies directed only at hydrocarbon reduction may be set aside in
favor of more balanced NOX and hydrocarbon control.   In  fact,  there  is
growing awareness that elevated photochemical oxidant  levels in  some non-
urban areas are sensitive primarily to NOX concentration.
     Strict NO  control will be required in order to reduce the impact of
              J{
photochemically generated secondary pollutants such as ozone.   Coal-fired
boilers will receive a disproportionate share of the control burden since
centralized stationary sources are administratively easier to regulate
than motor vehicles and since the shift in national energy  policy toward
greater reliance on domestic coal supplies will result in increased NO
                                                                      rt
emissions from this source category.
     Unfortunately, combustion modifications which substantially reduce
NO  generation in oil- and gas-fired boilers are not nearly as success-
  /\
ful when applied to coal-fired boilers.  Such  techniques are limited to
30 to  50% NO  reduction and  sometimes  result  in reduced boiler perfor-
            X
mance.   It  is  probable that  combustion modification alone will not be
adequate.   There  is a  need  for  "add-on" control technology  which supple-
ments  combustion modifications  so  that much  greater N0x reductions can
be achieved.

-------
      A photochemical  process  is  a  possible  solution.  This approach
 takes advantage  of chemical reactions analogous to those culpable ambi-
 ent  reactions  leading  to  photochemical oxidants.  The reactions are ini-
 tiated while the pollutants are  still highly concentrated within the flue
 gas.   In  this  manner,  control  is maintained over the reaction conditions
 and  collection of reaction products  is possible.
      There are a  number of advantages inherent in photochemical  control
 of NO  from coal-fired boilers.  Since the photochemical system is oper-
      /\
 ated  downstream  of  the boiler, there is no adverse impact on the boiler
 energy efficiency or on boiler maintenance costs.   The stoichiometric  re-
 quirements of  hydrocarbons are much lower than equivalent requirements
 for  ozone or other  dark phase oxidizing agents.  As a concentration de-
 pendent approach, energy requirements are limited  to that increment nec-
 essary to maintain  compliance with applicable standards.  By using NO
                                                                    7\
 outlet  concentration monitor feedback control  of lamp power, it is pos-
 sible  to follow rapid fluctuations in boiler NO  emissions, thereby in-
                                               A
 suring continuous compliance and avoiding expensive over-control.   Under
 light-limited and time-limited conditions, it is possible to selectively
 oxidize NO to N0? and then quench the reactions before most of the unde-
 sirable secondary products form.

 PURPOSE
     The purpose of this  study is to evaluate the  technical  and  economic
 feasibility of photochemical  pretreatment for NO  control.   It is  a bench
                                                J\
 scale analysis  emphasizing those basic  chemical  reaction mechanism para-
meters which influence costs  and performance.   It  is  recognized  that a
certain amount  of engineering  development is also  necessary with respect
to lamp optimization and  reaction product collection.   These subjects  are
necessarily reserved for  later work on  a  pilot scale.
     Specifically, the study  is designed  to  evaluate the efficiency of
the reaction mechanisms in the high concentration  pollutant gas  mixtures
characteristic  of coal-fired  boilers.  Relationships  between nitric oxide
oxidation and energy consumption  under  various  physical  conditions are
obtained in order to define the limits  of the  control  technique.   Molar
ratios of SO^,  NO and hydrocarbons are  varied  to consider the effect of
                                 2

-------
initial concentrations on t.ie type and extent  of  the  secondary chemical
reactions.  Major reaction products are identified.

APPROACH
     A flow reactor system was designed to  simulate stack concMiions and
to allow rapid turnaround for maximum experimental runs.  Simulated flue
gas was blended externally and then irradiated inside a quartz reactor
using a high pressure xenon arc.   This UV  lamp was a  compromise choice
                                                             o
having moderate energy output in  the 2400-3000 and 3000-4000  A spectral
bands of interest, and also having reasonable  temperature stability.
                                 o
Light intensity in the 3000-4000  A spectral  band  was  routinely measured
using N02 photolysis.  The outer  chamber was used as  a mounting frame,
light shield and heater for the inner reactor.  In this manner, the walls
of the quartz reactor could be maintained  slightly above the  internal
gas temperature to minimize wall  deposition  of aerosols.  The outer cham-
ber was constantly purged with N2 to reduce the possibility of 0~  leak-
age.
     Gas analyzers operating off  a common  outlet  manifold with a short
residence time were used to measure reactant and  product concentrations
on a continuous basis.  Millipore filter samples  were taken for subse-
quent analysis of possible aerosols.  After dark  phase steady state con-
ditions matched planned conditions, the xenon  arc was turned  on.   The
reactor was allowed to achieve steady state conditions at which
time all appropriate data was recorded.  The difference  between dark and
light phase steady state conditions was used as a measure of  NO oxida-
tion at the various run conditions.

-------
                             CHAPTER 2

                       SUMMARY AND CONCLUSIONS

     This  study was  performed to determine the technical and economical
 feasibility of a  photochemical technique for the oxidation of NO.  This
 approach takes advantage of  chemical reactions similar to those responsi-
 ble for ambient photochemical smog incidents.  The reactions are initiat-
 ed while the pollutants are  still highly concentrated within the flue
 gas and while the reaction products can be safely collected.  With care-
 fully controlled light limited and time limited conditions, it is possi-
 ble to maintain good control on the extent of NO oxidation and the reac-
 tion products.  Photochemical pretreatment for NO oxidation was proposed
 as a near term solution for the need for supplemental NO  control tech-
                                                        /\
 nology for coal-fired boilers.
     The experimental program has been designed to address five specific
 questions concerning photochemical  system operating characteristics and
cost.   One of the principal issues  is  the total energy requirement.   The
 present analyses are necessarily restricted to the commercially available,
 low efficiency, laboratory-scale, ultraviolet lamps.  Energy requirements
using  these lamps were anticipated  to  be high.  A second issue concerns
the possible need for injected hydrocarbons to sustain the photochemical-
 ly initiated free radical  chain reactions.   Propylene has been arbitrar-
ily chosen for study since it is a  readily available hydrocarbon of low
cost.   The fraction of hydrocarbon  consumed is the third issue of impor-
tance  since the hydrocarbon is itself  a criteria pollutant.   A number of
chemical  and physical conditions could conceivably influence the extent
of NO  oxidation and the study of these variables constitutes the fourth
major  work area.  The quantities and chemical characteristics of reaction
 products are the remaining issues addressed in the experimental  program.
     A continuous stirred tank reacfor was  designed to simulate coal-

-------
fired power plant stack conditions and to allow rapid  turnaround  for
maximum experimental runs.  Simulated flue gas  was  blended  externally
and then irradiated inside a quartz reactor using a high  pressure mercury-
doped xenon arc.   Light intensity in the 3000 to 4000  A spectral  band was
routinely measured using NOo photolysis.  Gas analyzers optraiing off a
common outlet manifold with a short residence time  were urc.d  to measure
reactant and product concentrations on a continuous basis.  Millipore
filter samples were taken for subsequent analysis of possible aerosols.
     Strong spectral effects were identified.   Ultraviolet  light  in the
                                            o
middle ultraviolet spectrum of 2400 to 3000 A appeared to be  particularly
effective in NO oxidation.  It was determined that  SO^ was  not responsi-
ble for this phenomenon, despite the fact that  it was  the strongest mid-
dle UV absorber in the system.  NO conversion actually improved as the
S02 concentration decreased.  The enhancing effect  was probably due to
the photolysis of aldehydes, nitrous acid and hydrogen peroxide;  however,
definitive conclusions could not be reached on  the  basis  of available
data.
     The effect of UV light intensity and propylene concentration were
tested in a large series of tests.  NO oxidation in the absence of UV
light was essentially zero in all runs tested.   There was measurable NO
oxidation in the absence of hydrocarbon; however, the efficiency  of con-
version was limited to approximately 10% of the inlet NO  .  A series of
                                                       /\
NO conversion isopleths were defined for the various levels of the two
principal variables.  The relationships were characterized  by distinct
light sensitive zones, hydrocarbon sensitive zones  and a  ridge between
the two in which both variables were important.  Generally, it appeared
necessary to use propylene at a stoichiometric  ratio of  1 to  2,  depending
on the light intensity.  The required light energies appeared to  be  equi-
valent to 3 to 5% of station power for tests done with inlet  NO  concen-
                                                               y\
trations of 600 ppr.i  (v/v).  System performance at higher  NO  levels  of
                                                           A
1200 ppm (v/v) was considerably  poorer than those at the  low  concentra-
tions.
     A comparison of system performance using propylene and ethylene in-
dicated that the reactivity of the hydrocarbon  is an  important limitation
on the amount of NO  oxidized.  Except at  high concentrations, the NO con-

-------
 version was substantially higher with propylene.   This  work  was  consis-
 tent with results  of the S02  sensitivity  tests  which  indicated that  com-
 petition for hydroxv radicals suppresses  system performance.
      A major problem involved in the  use  of  the photochemical system with
 propylene appears  to be  the fraction  of injected  hydrocarbon which is not
 consumed in the  reactor.   In  the experimental runs, over 50% of  the  mate-
 rial  was released  without reacting.
      Due to the  favorable cost,  it  is  assumed that combustion modifica-
 tions  will  be  used to the maximum extent  possible before the use of  a
 flue  gas  treatment system will be considered.   Therefore, it is  important
 that  the system  be compatible with  the conditions resulting from such
 modifications.   Tests with various  excess air levels confirms that the
 photochemical  systems performs adequately down  to oxygen concentrations
 of 2  to  3%.  Also, the efficiency of the  reaction mechanisms appears to
 increase  at  lower  initial NOX concentrations.   It is proposed that the
 latter phenomenon  is due  to the  consumption of  hydroxy radicals by N02>
 nitrous  acid, and  nitric  acid, which would otherwise react with propy-
 lene in  radical  branching reactions.  Since combustion modifications
 could conceivably  lead to increased carbon monoxide concentrations, a
 small test series was performed  to test the influence on the system.   As
 expected, the hydroxy radical  reaction with CO  leading to a hydroperoxy
 radical appears to enhance conversion of NO.   Carbon dioxide at concen-
 trations of up to 20% did not have any measurable influence on the sys-
 tem.
     Aerosols were not generated in measurable quantities in any of the
runs.  Condensation nuclei counter readings were consistently at back-
ground levels.  X-ray fluorescence tests of the Mi 11ipore filter samples
 indicated that less than  1% of the sulfur was present as aerosol.  Minor
quantities of deposited nitrate and sulfate material were occasionally
 visible on the inner surfaces of the reactor and in the effluent lines.
     Generally, the sulfur material  balance could account for 97 to 99%
of the input quantities.   Unaccountable nitrogen quantities reached 15%
of input values  for the very reactive conditions, presumable because the
 system had proceeded beyond the N02 peak and substantial quantities of
 nitric acid and  organic nitrate  compounds had begun to form.   Between

-------
25 and 50% of the input carbon cculd not be accounted for as propylene.
The principal reaction products included formaldehyde and acetaldehyde.
Mass spectrometer analyses of a selected group of runs suggest that an
extremely large group of organic coumpounds can form at trace levels.   The
najor compounds identified in the GC-mass spectrometer tests included
acetic acid and acetaldehyde.  Other compounds present in the range of
1 to 2 ppm included nitromethane, methyl formate, methyl nitrate, butyl
formate, toluene, and acetophenone.
     Capital and annualized costs were estimated for a model plant.   Re-
sults indicate that capital costs could be in the range of $37 to $50  per
kW, and that the annualized costs would be 3.8 to 5.6 mills/kWh.  The
range in values reflects the variation in cost as a function of the de-
sired NO  conversion efficiency.  These costs are lower than for some  costs
of comparable systems.  Catalytic reductions systems and NH3 injection
systems, however, have reported costs approximately % those for the
photochemical system.  It is concluded that the photochemical system
based on propylene will be used only if the alternative processes cannot
be adapted for coal-fired boilers, or if there are operational problems
with such systems when used in full-scale systems.
     The principal conclusion of the study is that photochemical control
techniques, as presently conceived, are not competitive.  Additional
bench-scale study of different hydrocarbons and improved lamps, however,
could conceivably yield a system considerably better than the alterna-
tive flue gas denitrification systems.

-------
                             CHAPTER 3
                          RECOMMENDATIONS

      1.   It  is  recommended  that  the present experimental system be uti-
 lized to  evaluate  the  system performance with acetone and formaldehyde.
 Both  compounds  participate  in  primary photochemical processes and should,
 therefore, be less susceptible to  the reactivity problem identified with
 propylene.

      2.   Further studies should also  be done to test system efficiency at
 lower  initial NOX levels.  The experimental tests in this project sugges-
 ted improved NO oxidation efficiency  at lower NOX concentrations.   The
 600 ppm level used in most of the  present work is higher than would be
 expected  for coal-fired boilers modified for reduced NOX generation.
 Accordingly, the cost estimates could be high.

      3.  The analytical capability of the system should be improved so
 that better nitrogen balances could be prepared.  The response of the
 chemiluminescent analyzer to nitrous acid and to nitric acid should be
 determined.
     4.  The performance of the photochemical system at temperatures of
700 to 800°F should be analyzed to determine if the irradiation could be
one downstream from a hot side electrostatic precipitator.
     5.   A kinetic simulation model  should be used as an aid to under-
standing the reaction system.

-------
                            CHAPTER 4
                            BACKGROUND
NO  CONTROL TECHNIQUES
  /\
     The photochemical process  for NO  control  is being developed in an-
                                     A
ticipation of a need for control  of coal-fired  utility boiler emissions.
In this section, the growing importance of coal-fired boiler NO  emis-
                                                              /\
sions is discussed.   NO  formation mechanisms  in  the combustion process
                       A
are briefly addressed to indicate the potential benefits of modified
operating conditions for NO  suppression.   It  is  concluded that such
                           A
"combustion modification techniques" are economical and effective;  how-
ever, these may not  be adequate to achieve the  very stringent NO  emis-
                                                               /\
sion levels being considered by EPA.  Accordingly,  some type of supple-
mental control technology may be necessary. Competitive add-on tech-
niques are briefly described so that the photochemical  process studied
in this project can be critically evaluated.
Coal-Fired Boilers as Sources of Nitrogen Oxides
     Constantly growing energy demands, coupled with  the  shift toward
greater national dependence on available coal  supplies, will result in  a
major increase in NO  emissions from coal-fired boilers  in  the next 20
                    A
years (Acurex, 1977).  As shown in Figure 1,  the NO  emissions from
                                                   A
stationary sources  (point and area) have increased  from 3.9 Tg/Yr   in
1950 to a  present rate of 12 Tg/Yr  and will  probably reach  17 Tg/Yr  by
1990.   Coal-fired boilers are responsible for approximately 30% of  these
emissions, and thus,  represent the largest source category ( Salvesenejt
§1., 1978).
     The  typical flue gas composition  of coal-fired boilers is  provided
in  Table  1.   This is  based  on 3.5%  sulfur, 16% ash coal,  10,500  Btu/Lb.
and an  excess air rate  of approximately 25%.  Note that there is  a wide

-------
  17.
  15.O
112.5
iio.o
  7.5
  2.5L
1   Acurex. 1978
?  NAS, 1977
3  Kemp «nd Dykema,  1977
«   Rlccl. 1977
5   Acurex, 1977
6   Salvesen. 1978
                                             Stationary Sources
                                             Combined Total
                                              Coal-Fired
                                              Utility Boilers
     rear:
             1960
                      1970
                                 1980
                                           1990
                                                     2000
    Figure 1.    Trends  in  the  emissions  of NOX
                   from stationary  sources.
                                10

-------
       TABLE 1.  TYPICAL FLUE GAS COMPOSITION
	OF COAL-FIRED BOILERS	
Contaminants
        Horizontally Opposed Firing3     400- 900 ppm
        Tangential Firing3               300- 600 ppm
        Front  Wall9                     400- 850 ppm
        Cyclone9                         500-1200 ppm

   S02                                  1000-3000 ppm

   S03                                     5-  30 ppm

   HC                                      1- 100 ppm

   CO                                     10-  50 ppm

Major Components

   N2                                        74%

   02                                         4%

   C02                                       13%

   H20                                        9%


Reference - Acurex, 1978.
                          11

-------
variation in the initial NOX levels for common firing modes.  Tangential
units presently comprise 35% of the utility boiler population and will
represent a greater share in the future due, partially, to the favorable
NOX emission rates.  Cyclone boilers represent only 10-30% of present
coal-fired generating capacity but are responsible for over 19-22% of
the NOX emissions from this source category (Ctvrtnecek and Rusek, 1977).
These boilers connot achieve the required 0.7 IBs./MMbtu NOX emission
rate3 due to the high heat release rates characteristic of the cyclone
boilers.  Because of this problem very few cyclone boilers have been
ordered since 1974 (Batayko, 1979) despite the slight energy efficiency
advantage over other types of boilers, and the ability to burn high ash-
low fusion temperature coals.
     The NO  control  requirements will probably be set as  low as possible
           /\
considering available technology due to the increasing concern over NO
                                                                      X
participation in oxidant formation and the health implications of  NOp.
EPA has informally stated control technology goals of 200 ppm (v/v) in
1980 and 100 ppm (v/v) in 1985 for pulverized coal-fired  boilers (McCut-
chen, 1977).
     Control will be achieved by use of combustion modification tech-
niques, add-on control processes or a combination of the  two.  To  date,
most of the research has been devoted to the development of combustion
control due to low inherent costs (Ricci, 1977).   There is growing reali-
zation, however, that such techniques are limited with respect to NO
                                                                    J\
reduction in coal-fired boilers and can have some adverse impact on
boiler performance when applied zealously.
Combusion Modifications for Control  of Nitrogen Oxides
     There are two general  formation mechanisms leading to the generation
of NO  during the combustion of coal.  "Thermal" NO  results from the
     x                                             x
high temperature fixation of molecular nitrogen.   "Fuel" NO  results
                                                           />
from the oxidation of organically bound nitrogen.  The relative impor-
tance of each mechanism cannot be determined with presently available
     aStandard of Performance for New Sources (NSPS), a regulation prom-
ulgated by the U.S. Environmental Protection Agency.
                                12

-------
information (Armento,  1975);  however,  it is  believed  that  fuel  NO   com-
                                                                X
prises 50 to 100% of the total  NO  (Dykema,  1978).  Nitric oxide (NO)
                                 J\
is the predominant species in each case, with  less  than  5  to  10% nitrogen
dioxide (N02).   While N0£ is  the thermodynamically  favored form at  stack
temperatures, the formation mechanisms are quenched before significant
conversion of NO can occur in the boiler.
     Fuel nitrogen oxidation  is a complex physiochemical process governed
primarily by the local conditions surrounding  the initial  combustion  zone
(Acurex, 1977).  As shown in  Figure 2, the nitrogen in the coal is  parti-
tioned into a vapor fraction  and a char fraction upon injection into  the
boiler.  A portion of the volatile fraction  is oxidized to NO,  some of
which escapes the fuel rich reducing conditions of the burner flame (Heap
et al., 1976).   The nitrogen  compounds remaining on the char  particle are
oxidized at a slower rate.  The NO formed on the particle  surface must
penetrate the boundary layer  before it is reduced to  N~.   The NO  yield
                                                      £        X
of this set of simultaneous oxidation mechanisms is fortuitously only 10
to 40% with the lowest yields at the higher fuel nitrogen  levels  (Faucett
et al., 1977).
     Thermal NO  results from a partially understood  set of free  radical
               J\
reactions operative throughout the high temperature zone of the boiler.
During the early combustion stage, a super equilibrium level  of free  rad-
icals leads to the rapid formation of NO in a process which is sometimes
termed "prompt" NO .  In the  later stages of fuel oxidation,  a slower
                  A
equilibrium moderated set of reactions represented by the Zeldovich mech-
anism (Reactions 1 to 3).
                          N2 + 0 +  NO + N        (1, a & b)
                          M  + 02 + M + 20        (2, a & b)
                          N  + 02 - NO + 0        (3, a & b)
      These reactions  have a strong temperature dependence due  to  the high
activation energy of  reaction  1.  The role of oxygen in a free radical
transfer step  explains  the strong effect of available oxygen on the ex-
tent  of  thermal NOX formation.
      A number  of combustion modification techniques have  been  developed
in order to  interfere with the fuel and  thermal  NOV formation  processes.
                                                  J\
                                13

-------
                                                                     ash
                                                                   virtual
                                                                  nitrogen
                                                                     free
                                                      Reduction
                                                      in
                                                          ayer
                                                         Escape
                                                          from
                                                        Boundary
                                                          Layer
      Oil
    Droplet
       or
oal  Particle
Volatile
  raction
                                            Reduced in Hea
                                               Releasp
                                                 Zone
  Source: Acurex 1977
        Figure 2.  Possible formation mechanism for fuel NO .
                                  14

-------
These include: (1) low excels air, (2) off-stoichiometric combustion,
(3) reduced air preheat, and (4) advanced burner design.  Flue gas recir-
culation, a technique proven useful for gas- and oil-fired systems is
not considered acceptable for coal-fired boilers (Armento, 1975) due pos-
sibly to the importance of fuel nitrogen oxidation and the insensitivity
of fuel  nitrogen  to temperature reduction schemes.
     The most successful techniques appear to be low excess air combus-
tion and off-stoichiometric combustion (Armento, 1975; Selker, 1975).
The positive effect of these two approaches is partially because the
restricted oxygen  level in the high temperature fuel oxidation zone inhi-
bits both fuel and thermal NOX  formation ( Armento, 1975).
     Low excess air combustion is sometimes considered just good oper-
ating practice which should be used at all boilers regardless of NOX re-
duction requirements.  When the excess air is maintained  in the 15 to
25% range, the flue gas volume is minimized and sensible heat loss is
minimized.  Boiler energy efficiency can be improved more than 1% in this
manner.  At lower  levels of excess air, localized reducing conditions can
lead to increased  slagging.  The potential benefit of low excess air is
illustrated in Figure 3.  Generally, the effectiveness  of low excess air
alone is limited to a maximum 20% NO  reduction (Acurex, 1978).
                                    /N
     Off-stoichiometric combustion may be accomplished  in several ways
depending on whether the boiler is new or is an existing unit which re-
quires retrofit.   For the latter case, selected burners can be operated
with air only so that others can be operated fuel rich.  This minimizes
thermal and fuel NOX formation mechanisms in the high temperature zones.
With biased firing in this manner, boiler derating is necessary.  For a
new boiler, the same result can be accomplished without derating penal-
ties by including  a set of overfire air  ports.  The potential NOX reduc-
tion available by  means of off-stoichiometric  combustion  is limited to
less than 40%  (Acurex,  1978) as shown  in  Figure 4.  New unit  design
equipped with overfire  air ports  show typical  NOX reductions  of about
 50%  (Lachapelle,  1979).
     A  third  approach  for NOV  suppression which appears useful  is  the
                             ^
 use  of  advanced  burners.   Pilot-scale tests  have  suggested that 200 ppm
 levels  are  achievable  with low NOX burner designs.   Demonstration of
 this  equipment on full-scale boilers  is  in  progress (Lachapelle,  1979).
                                 15

-------
                                  Normal  Firing
                                  178-227 MW
   600
   500
                                                           Biased  Firing
                                                           166-210 MW
   400
Q.
D-
   300
 X
o
                    Coal-Fired, Front Wall
                    Boiler
   200
                                    Source:  Crawford etal., 1977
   100
                  1
                         J_
              23
     Average % Oxygen  in Flue Gas
Figure 3.   Effect of excess air on
                                                      emissions
                                 16

-------
   800  -
CO
CO
 CM
O
ro
   600
    500
    400
    300
    200
    100
                                Full Load,  High  Excess Air
                Reduced  Load
                Normal Excess Air
                              1 Load, Normal
                                Excess Air
                        Coal-Fired  Utility
                        Boiler,  Crawford  et al .."
                        1977
                   I
                 I
  Figure 4.
    20          40         60
     Ov erf ire Air Dampers - % Open
                                                    80
Effect of secondary air'addition on NO  emissions,
                                      1\
                               17

-------
     A  number of other combustion modifications have been tested and
 found less desirable than  low excess air, off-stoichiometric combustion,
 and advanced burner designs.  These techniques include: flue gas recirc-
 ulation,  reduced air preheat, load reduction, and use of low nitrogen
 coal.   All except the last involve energy penalties.  These approaches
 generally provide much less than 20% NOX reduction and probably will not
 be used substantially in the near future.
     A  coal-fired boiler utilizing the low excess air firing and/or off-
 stoichiometric combustion will probably have flue gases with 2 to 3% oxy-
 gen and 300 to 500 ppm (v/v) NOX.  The predominant chemical composition
 of the  NOX will continue to be NO.  It is probable that all new boilers
 and most existing units less than 10 to 20 years old will  include ad-
 vanced  combustion control techniques due to the inherent low cost.   The
 use of  low excess air involves no significant capital cost and provides
 a slight benefit in energy efficiency.   The use of off-stoichiometric
 combustion does require additional burner ports and other equipment;
 however, retrofit capital costs remain  at a very modest $1  to $2/kW
 (Mason et al., 1977).
     There are several  practical  limitations to the use of combustion
modification techniques due simply to the fact that boiler equipment was
designed to operate under oxidizing conditions.   As excess  air and  burner
conditions are increasingly optimized for low NOX boilers,  operational
problems appear.   One  of the most common of these is  the reduction  of ash
fusion temperature near the base  of the boiler which  can lead to in-
creased slagging (Crawford  et al, 1977).   Other potential  problems  in-
clude generation of carbon  monoxide and high carbon carryover.   The lat-
ter condition  is important  primarily because it represents  a loss  of
boiler energy  efficiency.    An optimized level  on NOX control  can minimize
these side effects.
     Potential  aggravation  of polycyclic organic matter concentration (POM)
is an important disadvantage of combustion modifications.   In one study
done by Thompson and McElroy (1976), the use of staged  combustion and
flue gas recirculation  resulted in a 40% increase in  boiler POM emissions.
 It is apparent in Table 2 that some of  the individual hydrocarbons  such
as pyrene and  methyl anthracene exhibited dramatic increases when  these
                                 18

-------
combustion modifications were used.   Considerably more  experimental  evi-
dence is necessary to document the relationship between POM generation
and NO  combustion controls.
      /\
Flue Gas Treatment for Nitrogen Oxides Control
    While the need for flue gas denitrication cannot be corcl»
-------
                                            Selective Catalytic  Reduction -
                                           Selective Noncatalytic Reduction - 1
                                           Nonselectlve Catalytic  Reduction - 1
                                                     Radiation - 1
                                                     Adsorption - 1
                                              Liquid Phase Oxidation - 4
                                              Liquid Phase Reduction - 5
                                              Liquid Phase Oxidation - 3
                                              Liquid Phase Reduction - 5
Figure 5.   Classification of  NOX  flue  gas treatment  processes,

-------
                    TABLE 2.  EFFECT OF COMBUSTION MODIFICATION TECHNIQUES
                              ON THE EMISSION OF POLYCYCLIC ORGANIC MATTER
Compound
Anthracene/ PI enanthrene
Methyl Anthracenes
Fluoranthracenes
Pyrene
Total
% of Baseline
Baseline
115.0
34.4
33.2
9.7
192.3
100
Biased Firing
119.5
32.3
73.5
34.7
260.0
135
Concentration
[yg/dscm]
15« Recirculation
92.8
57.1
15.8
14.9
180.6
94
Biased Firing &
Recirculation
157.3
65.0
28.9
31.0
277.2
144



From:  Thompson and McElroy, 1976.

-------
 Within a very narrow flue gas temperature range, reaction 4 is favored
 over reaction 5, thereby providing an opportunity for a net reduction in
 NO.  Applied improperly, however, there is the potential  for an increase
 in total NO  levels.  Unless injection is done within 150°F temperature
           /\
 range (Muzio and Arand, 1976), a number of undesirable conditions  can re-
 sult, including:   (1) incomplete reduction of NO, (2) emission of excess
 NH-, and (3) increase in total NO  emissions due to NH, oxidation.  This
  tj                              A                    O
 narrow optimum temperature range is illustrated in Figure 6.  While it is
 possible to modify this temperature zone by combined use of NH_ and H»,
                                                              3      t-
 the safety problems involved in the use of a highly explosive gas  in a
 boiler may hamper H2 use (Faucett et a! . , 1977).  Thus, there Is a seri-
 ous unresolved problem.  Further, gas temperature at any one injection
 port in the convective region of a boiler experiences swings of more
 than 200°F as boiler load varies in diurnal  and seasonal  cycles (Muzio
 and Arand, 1976).
     Exxon, a developer of one such process, has estimated annual ized
 costs for a 1000 MW coal-fired boiler with an initial  NO   level  of 300
                                                       ^
 ppm (v/v) and a final  desired level of 100 ppm (v/v) at 4.76 to 10.2
million dollars per year (Lyon as reported by Muzio  and Arand, 1976).
This is equivalent to 0.75 to 1.6 mills per  kWh assuming  an average ca-
 pacity factor of 75% for a base loaded unit.  Capital  costs would  be low
since equipment requirements are limited to  an NH,  storage and distribu-
                                                 O
tion system.   The principal  operating cost is for NH-,  which is  dependent
on the rapidly rising  cost of natural  gas.   The estimates of annual ized
costs provided above are for an ideal  NO  inlet concentration.   Costs
                                        /\
should rise proportionally to the inlet level of NO  .
                                                  /\
     While this  approach is  the most  straightforward,  it  is apparent
that some serious operational problems and high operating costs  discour-
age use in this  country.  Since NO reduction is limited  to 50 to  70%,
                                  /\
even at very  high NH?/NOy  ratios of 3 to 4 (Faucett  et  al . . 1977),  it is
possible that this approach  will  serve primarily oil-fired boilers.
Selective Catalytic (Dry.  Gas Phase)  Reduction
     This category includes  the majority of  processes  under development,
as demonstrated in Figure 5.   Reduction of NOX occurs  in  a catalytic re-
                                22

-------
      350
      300
      250
      200
   §.  150
   CL
    ro
   :r
       100
        50
                                     I
                                           NH
     Source:   Faucett et al

              1977
Figure 6.
  _J	 I  ^
    12001400      1600       1800

                  Gas Temperature,  °F

Effect of temperature on NO  removal  in the selective

noncatalytic reduction process.
                               23

-------
 actor  downstream  of  the boiler.  By using a catalyst, it is possible to
 avoid  the  potentially crippling temperature sensitivity problems and
 fouling  problems  which plague the selective noncatalytic process per-
 formed in  the  boiler superheater.  These catalytic reactions require only
 a  0.9  to 1.2 NH3/NO  stoichiometric ratio, thus minimizing sensitivity to
 rising NH., costs.
     The two sets of composite reactions shown below generally illustrate
 the  type of overall  reactions responsible for NO reduction in the various
 catalytic  processes  (Faucett et al . , 1977).  The catalyst and design char-
 acteristics affect the actual overall stoichiometry of the processes.
4NH3
4 NH-
3
(g)

(g)
6 N0(g)
6 NO-
2
4 NH,
3

(g)

(g)
+
+

+
+

+
2
4

4
8

3
N02
NO,

NH3
NH,
3
0.
(g)
\
g)
(g)

(g)

+ 0 na^.ja,.
2(g)
+ Catalyst
2(g)
Catalyst
Catalyst

Catalyst
'(g)
3N2
4 N0
2
5 N2
/ 1 » o
2
2 N0
2
(g)

(g)
(g)

(g) "
H
(g)
f 6
* 6

^ 6
h i;

H 6

H2
U
2
H2
? H

H-
2
Q
Q
(g)
°(9)
-0,
2 (g
Q
(g)
                                                                (1Q)
     The catalytic process generally requires a high efficiency electro-
static precipitator on the "hot" side (before air preheater) to protect
the catalyst from potential poisons and flyash coating.  Malfunctions of
these precipitators represent a serious threat to the catalyst life and
long term NO  reduction performance.  In addition to the precipitator, a
            A
catalyst-containing reactor operating in a temperature range of 300 to
500°C is necessary.  For these reasons, the capital  costs for catalytic
systems will probably be much higher than the previously discussed non-
catalytic approach.  A recent compilation of these processes include
capital costs estimates ranging from $10 to $36 per kW (Mason et al . .
1977).  To put these numbers in perspective, consider that the boiler-
turbine assembly costs $400 to $600 per kW and nonregenerative SO,, con-
trol systems cost $50 to $150 per kW.  The NO  costs are low but cer-
                                             ^\
tainly not insignificant.  Annualized cost estimates range from a very
low 0.2 mills per kWh to a moderate level of 1.8 mills per kWh.  The
lower NH-, requirements offsets the higher capital costs involved in
                                24

-------
catalytic approach.   There is very little meaningful  data concerning the
possible  useful  life of the catalyst under conditions similar to coal-
fired boilers, since almost all experimental work has concerned oil-fired
boilers.
Radiation Oxidation
     The EBARA-JAERI process is of interest in this report primarily be-
cause of the superficial similarity to the photochemical  process.  Acutal-
ly, the two radiation techniques are quite different with respect to chemi-
cal mechanisms used  and to reaction products formed.   The flue gas from
the air preheater is treated in a cold side (after air preheater) elec-
trostatic precipitator and then irradiated by electron beams to yield a
highly corrosive particulate.  This material is removed from the gas
stream in a second  precipitator (Acurex, 1977; Machi et al, 1975).  Costs
for this highly complex process range from a moderately high $100 per kW
to a phenomenal $1000 per kW.  The high value is probably most realistic
for coal-fired boilers.  Power plant derating of approximately 3.3% is
required to satisfy energy demand.  Due to limited size of presently
available electron beam accelerators, large numbers (100 to 500  ) would
be required for average sized coal-fired power plant applications.
     It is readily apparent that this process is not a prime candidate
for control of coal-fired boilers.  The only advantage to the approach
is the freedom from chemical additives and the simultaneous S02 and NOX
control without the use of wet scrubbers.  It is doubtful that these
advantages will outweigh the high costs and potential operational diff-
iculties of electron beam accelerators.
Absorption of NO
     There are at least eight  processes under development which  first
absorb NO and then attempt  to  oxidize or  reduce  the nitrites and  nitrates
in solution.  Very large absorbers are required  to overcome the  poor
solubilities of NO.  Performance appears  partially sensitive to  the
highly  Variable S02  concentrations and to 02 concentrations (Faucett
et al,  1977).  To date,  this   group  of processes has  only  been studied
on the  bench  and pilot  scales.   Development work has  been  directed
toward  oil-fired boilers.
                                25

-------
      Preliminary cost estimates suggest capital costs of $65 to $130 per
 kW and annualized costs ranging from 4.8 to 7.4 mills per kWh (Faucett
 et al, 1977).  These represent ideal costs based on an absolute minimum
 NOX reduction requirement of 125 ppm.  All of these costs are for oil-
 fired boiler applications.

 Oxidation  of NO with Liquid  Phase Oxidation or  Reduction
      This  category of control techniques  is most similar to the  photo-
 chemical technique being  studied.   In each process, an additive  is in-
 jected into  the flue gas  directly downstream of the air  preheater.  Addi-
 tives such as ozone and chlorine dioxide are used to oxidize the NO to
 N02 while minimizing formation of N,>05 and other nitrogen compounds.  By
 converting the very insoluble NO to  the relatively soluble form of NO,,,
 it is possible to use much smaller absorbers than necessary for the di-
 rect NO absorption processes.  This  cost saving is partially offset by
 the much higher cost of the gas pnase oxidant.   Harrison (1976) has
estimated the capital  and operating costs of an ozone plant for a 500 MW
coal-fired power plant with 600 ppm inlet NOX concentration at $57.60/kW
and 4.7mills/kWh respectively.  Due to the cooling requirements of the
ozone plant,  the flue  gas from the processes  is increased by as much  as
10%.   The high  concentration ozone stream presents  serious  safety problems
which may not be adequately reflected in these  costs.
     Absorption processes range from $84/kW to  $134/kW (Acurex, 1977).
Corresponding annualized costs vary  from 6.7 mills/kWh to 8.9 mills/kWh
 (Acurex, 1977).  All of these costs are for oil-fired applications which
require considerably less NO  reduction than coal-fired boilers.  Most
                            A
of these processes have an optimal control efficiency of 60 to 70%.  The
potential for toxic secondary product formation due to ozone or chlorine
dioxide side reactions has not yet been addressed.
PHOTOCHEMICAL NO  CONTROL TECHNIQUES
                A
     A conceptual design of a  photoc
trol  at a coal-fired utility boiler is analyzed  in  this  section.   Based
     A conceptual design of a photochemical control system for NO  con-
                                                                 A
                                 26

-------
on anticipated light absorption characteristics and commercial  lamp emis-
sion spectra, the potential  energy requirements are calculated  as  a func-
tion of the net product quantum yield of the various photochemical  reac-
tions.
     The reaction products and N0? quantum yields of four possible sets
of reaction mechanisms are reviewed in this section.  It is concluded
that photochemical systems could be economically competitive.  Major is-
sues which need to be resolved in the experimental phase of the project
are identified.  Previous studies of photochemical techniques are dis-
cussed briefly to clarify possible reasons why this control approach has
not been studied earlier.
Previous Studies
     General Electric (Browne and Stone, 1965) tested the SO^ control ef-
ficiency of ultraviolet light and corona discharges.  Unfortunately, the
energy calculations were not corrected for the 2% UV light absorption
which occurred in the small diameter tubular reactor (path length of 1
inch).  Because of this error, it was concluded that costs for a photo-
chemical system were not competitive.  Westinghouse  (Hamm, 1965)
performed a screening study of possible physiochemical means to collect
sulfur dioxide, and photochemistry was only one of a large number of ap-
proaches tested.  The researchers restricted the study to the infrared
spectrum and not surprisingly reached the conclusion that "photochemis-
try" was not sufficiently energy efficient for use  in S0? control.  This
study is of interest only because it may have limited research interest
in the further evaluation of more reasonable photochemical techniques.
Neither of the above studies considered NOX control since at that time
it was assumed that combustion modification techniques would be adequate
for the modest NOX  reductions considered necessary.
      Dols  and  Boknoven  (1974) described a  process  for the oxidation of NO
 in  coke oven  gas  to prevent  pipeline gumming.  This  appears  to be  the
 first  process developed for the photochemical oxidation  of  NO for
 commercial  purposes.   Oxidation  is  accomplished  using a  butadiene/NO
 ratio of  two.   The  initial  NO concentration  is reported  to  be  only
 o.25  ppm  (v/v),  three orders of  magnitude less than levels  in  coal-fired

                                 27

-------
 boilers.   The  gas  stream was  irradiated  in the 2050 to 2200 A absorption
 band  of  the  butadiene.  For these reasons, the system appears to be quite
 different  from the photochemical system  studied in this project.
      Machi et  al.  (1975) developed a  process for the simultaneous control
 of  S0? and NO   using either ionizing  radiation or ultraviolet light.
      C,       A
 This  process is  the basis for the EBARA-JAERI system described earlier.
 This  is  probably the first photochemical-type system proposed for control
 of  NO from a  power plant.  Irradiation  is done with an electron beam
      /\
 which presumably results in hydroxyl  and hydroperoxy radical formation
 by  the dissociation of water vapor.
      It  is apparent that photochemical techniques have not been studied
 extensively.   The  work to date has been  limited to several specialized
 or  exotic  applications.  Considering  the widespread nature of photoche-
 mical smog incidents involving the ambient level  oxidation of nitrogen
 oxides, it is  surprising that a system based on conventional  smog-type
 reaction mechanisms has not been proposed.
 Photochemical  System Conceptual  Design
     A flowsheet for a photochemical  treatment  system  on  a coal-fired
boiler is shown  in Figure 7.   This  includes the existing  equipment'for
 particulate and sulfur dioxide collection plus  the equipment needed  to
photochemically treat  the nitrogen  oxides.   The photochemical  reactor has
been located downstream of the electrostatic  precipitator  in  order to  re-
duce erosion of lamp optical  surfaces.  Using a cylindrical  design with
the lamps centrally located,  it  is  possible to  achieve  reasonable light
absorption.  The dimensions  of the  reactor  are  selected to obtain ade-
quate path length and  gas  retention  time.  Irradiated  effluent  is then
sent to  the flue gas desulfurization system where  the  SOo  and N0? are ab-
sorbed in an alkaline  slurry.   The  lamps  are  mounted on vertical  support
rods running parallel  to the reactor centerline.   Due  to  the limited size
of presently available ultraviolet  lamps, over  700 units  (length, b  feet;
diameter, 6 inches) would  be required to  get  the intensities  considered
necessary.  The average superficial  gas velocity has been  set at  5 feet
 per second so that the necessary retention  time can be achieved without
 excessive reactor heights.  Other assumed design parameters are  summa-
 rized in Tables  3  and 4.
                                 28

-------
ro
                                                                                                       Stack
                Liquor/Propyl ene stream

                Flue Gas  Streams

                                    Reactor
       Flue Gas from
       Air Preheater
        Electrostatic
        Preci pita tor
        Participate Control System
Photochemical  System
               Sludge Treatment Equip.


Flue Gas Desulfurization System -»-»•
                               Figure 7.   Photochemical  system conceptual design.

-------
TABLE 3.  PHOTOCHEMICAL REACTOR DESIGN ASSUMPTIONS
Boiler Rating
Capacity Factor
Gas Flow Rate at Full Load
Gas Temperature
Gas Retention Time
Gas Flow Characteristics
Lamp Type
Lamp Efficiency
Light Path Length
500 MW
60% ,
1x10° scfm
275°F
6 seconds
Well mixed
Mercury doped xenon arc
20%
10 feet
   TABLE 4.   EFFLUENT GAS COMPOSITION ASSUMPTIONS^-
Pollutant
NO
N02
HNO
RCHOb
°3
so2
Inlet
150"
50
0
0
10
1500
Outlet
TGCf
400
50
100
50
1500
  — All  concentrations  in  ppm.
  ^ Includes  formaldehyde  and acetaldehyde
                        30

-------
Photochemical  System Operating Characteristics
     The feasibility of photochemical  systems is  dependent  on  the  effi-
ciency of energy delivery/utilization  and on the  characteristics of  the
reaction products.   These two subjects are reviewed  using the  conceptual
design in the preceding section.
Energy Delivery Efficiency--
     The energy cost of a photochemical  system  is the product  of the en-
ergy delivery efficiency and the absorbed energy  utilization  efficiency.
This is expressed in Equation 1.
                              \
                   f*  •*»
Mnox
V
                                P_          (Equation 1)
                                 c
Where:  C    = the total electrical power cost
        MNOX = moles of NOX treated per unit time
        $    = net product quantum yield of the combined reaction mecha-
         P     nisms
        E    = energy delivery efficiency for the specific reactor system
               and effluent gas composition
        P    = energy cost

Equation 1 indicates that energy cost is an inverse function of the net
product quantum yield and the energy efficiency of generation/absorption.
If either of these factors approach zero, the energy cost of a photoche-
mical system would become prohibitive.  Obviously, this is the case when
deactivation of excited species occurs without chemical reaction and when
there is a poor match between lamp emission spectrum and absorption bands
of effluent compounds.  Equation 1 also indicates that the energy is a
linear  function of the  pollutant concentration.  This is somewhat mis-
leading since  the energy delivery  efficiency, E, should also be a func-
tion  of the pollutant concentration  in  some cases.  For example,  if NO
                                                                      /\
compounds are  involved  in  an  important  primary  photochemical process,
then  a  reduction of  the NO   inlet  concentration  could reduce the  overall
                           J\
energy  delivery  effectiveness.  This would  partially offset the  cost  sav-
 ings  of the  initial  NO   inlet reduction.
                      y\
                                31

-------
      In equation 2,  the energy delivery efficiency factor has been ex-
 panded to specifically include the  lamp generation efficiency across the
 middle and near UV spectrum,  the average reactor pathlength, and the pol-
 lutant extinction coefficients.  The net product quantum yield factor has
 been  kept outside the  summation sign since it is not strictly wavelength
 dependent.
              A  =  12500 A
              A  =   1800 A                                (Equation  2)
where  S  = fraction of lamp input power emitted at wavelength A
       a. = extinction coefficient of species i
       c. = concentration of species i
       L  = average pathlength for reactor
     The parameter 4> -E has been estimated for the reactor  configuration
and effluent composition presented earlier.   The data  provided  in Tables
3 and  4 has been combined with typical  mercury-doped xenon  arc  emission
spectra and the absorption spectra of the absorbing species (Figure 8).
The results of the energy delivery calculations are shown in  Table 5.
The energy delivery to each of the absorbing species has  been provided
along with the total  energy input to the gas stream.  Almost  complete
light absorption is expected in the middle UV band of  2000  to 3000 A,
while only 60% absorption is achieved in the near UV band of  3000 to
3650   A.
     It is  apparent that at the assumed pollutant concentrations, the
aldehydes cannot compete effectively for the limited light  available
within its absorption spectrum.  The middle ultraviolet band  will be
dominated by sulfur dioxide and ozone photolysis while the  near ultra-
violet band will be dominated by nitrogen dioxide photolysis.
     The energy delivery estimates have been substituted  back into equa-
tion 2 to derive a preliminary estimate of the net product  quantum yield
necessary for an economical control system.   The desired  energy demand  is
in the range of 0.5 to 2% of total station.   It is apparent that the net
product quantum yield must be greater than 0.5 to ensure  reasonable

                                 32

-------
           I        I       I        I
                                                             I       I        I       I
OJ
co
           1000
            100
       o>

       o
       o
       o
       c
       o
       Q.
       s-
       o
       to
       _O
       
-------
                                      TABLE  5.   ENERGY DELIVERY IN MODEL SYSTEM3
uv
B^nd
rim

185-210

210-240
240-260
260-280

280-300

300-320
320-340
340-380
s
A
Quanta/
Min

0

26
46
39

29

74
9
85

c
N09
c
-
_3
1.5x10
3.9xlO"4
5.2xlO"4
_4
7.7x10
_3
1.0x10
1.8xlO"3
2.6xlO"3

.a. (moles/1
HN09
d.
-
-4
6.4x10
2.9xlO"4
6.5xlO"4 1
-4
3.2x10 2
.5
4.8x10 2
6.7xlO"6
Neg.

iter x 1
RCHOb

-

Neg.
Neg.
.4xlO"5
c
.6x10"°
_5
.0x10
Neg .
Neg .

iters/mol e
0,
3
-
_3
3.2x10 J
9.6xlO"3
6.4xlO"3
-4
9.6x10
_5
3.2x10
Neg .
Neg .

)
SO
2
_
_T
7.7x10
2.9xlO"3
9.6xlO"3
T
9.6x10"
_•>
1.9x10 J
l.OxlO"4
Neg.

E
Quanta/
Eff. Min

-

0.99 26
0.99 46
0.99 39

0.99 29

0.65 48
0.47 4
0.80 68




E , Quanta/Min
i
NO,
2


3
1.4
1.2

2.0

16.4
4.0
51 .6
HNCL
2


1 .3
1.0
0.1

0.1

0.8
0
17.0
RHCOb 00
3


0 6.4
0 33.5
0 15.0

0.1 2.4

0.3 0.5
0 0
0 0
S00
2


15.3
10.1
22.6

24.4

30.2
0.2
0
Applicable to 500 MW station,  2% derating  for  lamp  power and 20% lamp efficiency in spectral band <4000 A
blncludes formaldehyde  and  acetaldehyde

-------
                10
00
 r
              OJ
              I
              
              OO

              :
              01
              CT
              01
              01

Near UV Photolysis only
                              Optimum
                              Operating
                              Band
                               Near and Middle
                                 V .Photolysis,,
                                                 I
                            i
                                          3456
                                       Net Product Quantum  Yield
                 Figure  9.   Power requirements as a function  of net  product quantum yield.

-------
 energy demands.  If N02 photolysis is the only important reaction mech-
 anism, then the product quantum yield requirements are considerably more
 severe as indicated by the upper curve.  Without any contribution from
 reactions initiated in the middle UV, the quantum yield of NC^ photolysis
 reactions must be greater than 5.

 Energy Utilization--
     The preceding section concerned the problems in the delivery of en-
 ergy to an effluent gas stream.  Results of this analysis indicate that
 the economical feasibility of photochemical systems depends directly on
 the net product quantum yield which can be achieved.  The reaction mech-
 anisms which will  be initiated in such a system are now reviewed with
 respect to the possible product quantum yields and with respect to the
 possible types of reaction products.
     It should be noted that the term reaction mechanism is used in a
 very broad sense.   It applied to an entire assembly of reactions initi-
 ated by the absorption of ultraviolet light in a specific spectral  band
 by a specific compound.  Some of the  possible mechanisms are identified
 in the list below:
                                                        Absorbing
              Mechanists                Spectral  Band      Species
   1.   Vacuum UV Reactions             1700 -  2000 A   02, H20
   2.   Excited State Chemistry         2400 - 12500 A   S02, N02, 02
   3.   Molecular Complex  Photolysis    2400 -  3400 A   EDA Complexes
   4.   N02 Photolysis                  2400 -  4000 A   N02> HCHO,  CH3CHO

tach of these "mechanisms" could consist of thousands of individual  chem-
ical  reactions involving  large numbers of distinct species.  N02 photo-
lysis  1s  emphasized since it is possibly the dominant reaction mechanism
 in the photochemical  system.
     NCL  Photolysis—The  irradiation  of coal-fired boiler effluent in the
 near UV spectral band can promote several simultaneous cyclic free radi-
 cal processes.  Simplified illustrations of the three major sets of reac-
 tions  are provided in Figures lOa, b, and c.
                                36

-------
                                        A\ denotes reaction  number
                                                      i
       Figure  lOa.  Photostationary state reactions.
                                                RCHO
                                                HCHO
                                           RH
Figure IQb.   Simplified  olefin/NO  photochemical reactions,
                                          OH    RCHO
                                                 HCHO
 Figure lOc.  Simplified aldehyde photochemical  reactions.
                            37

-------
     In the absence of appreciable olefin concentration, the reactions
are limited to the photostationary state conditions of Figure lOa.   Reac-
tions 11  to 13 are involved in this cycle.
           N00   + hv(3000-4000 A) -> NO  + 0(3P)    (11)
                   00  + M         -> 0,  + M        (12)
                                           0,       (13)
0(3P)
0.
                    2
                 + NO
                    N0
Unless there is sink for the ozone formed in reaction 12, there is little
change in the N0/N0? ratio.  The N0? product quantum yield of this set of
reactions is zero.
     When olefins are added to the flue gas, several additional free rad-
ical cycles are initiated.  Basic elements of these processes are the hy-
droxy radical attack on virgin olefin and the peroxy radical  oxidation of
NO to yield N02 plus additional hydroxy and peroxy radicals.   Three gen-
eral types of reactions are listed below:
                 •OH   + C0Hr
R-

RQ2,
RQi
                       + NO
*• + H20 (14)
R02- (15)
OH-
RO-
ROS
RCHO
HCHOJ

f N02 (16)



Reaction 14 involves addition of the OH- group to the olefin.   The
resulting radical will react with oxygen to yield the peroxy radical.
The oxidation of NO in reaction 16 can result in the formation of one  or
more of the products listed in the brackets.  The organic radicals can
undergo further reaction with oxygen to give additional  peroxy radicals.
While hydroxy attack of propylene accounts for most of the olefin
consumption, the reaction with ozone also could yield many similar organ-
ic radicals.
     The aldehydes formed in reaction 16 can photolysis  directly if there
is sufficient energy available in the 2400 to 3400 A band.  As shown in
reactions 17 and 18 ,  the products include the formyl radical, or simply
ethane and CO.  The latter reaction becomes important only at  the high
energy side of the absorption spectrum.  The aldehydes may also react
                                38

-------
with hydroxy radicals by means of hydrogen abstraction to give products
analogous to those formed in reaction 17.   Due to the limited light
available in the middle UV, reaction 19 will probably be the principal
reaction path for the aldehydes.

            RCHO + hv(2400-3400 A) + R«  + HCO-    (17)
            RCHO + hv(2400-3400 A) + R-  + R. + CO 08)
            RCHO + OH-             .> RQ. + H20     (19)

The aldehyde reactions listed above could be important free radical
propagation reactions.
     The N02 product quantum yield of the N0x-olefin reaction system can
conceivably be high.  As shown  in the propylene and aldehyde degradation
steps, the attack of one ozone  molecule or one hydroxy radical can result
in  the direct oxidation of NO and can result in the formation of numerous
hydroperoxy radicals.  The latter react almost exclusively with NO to
yield N02-  While the yield is  a function of olefin and  total NOX
concentration, the value at high concentrations could be in the range of
2 to 10.
     The combined result of these free  radical cycles is a gradual
depletion of the NO  and the formation of N02-  Once the  NO is exhausted,
ozone is no longer consumed in  reaction 13,  it builds up to appreciable
concentrations.  Other stable secondary products  such as peroxyacyl
nitrate, nitric acid and organic aerosols also accumulate in  the latter
stages  of the reactions.   An  example concentration profile during  the
course  of the photochemical reactions  is  provided in  Figure  11.
     With time  limited  and light  limited  conditions,  it  should  be
possible to avoid the  latter  stages  of the  overall process during  which
most of the  toxic secondary products are  formed.   By  designing  the
reactor to  reach  the N02 peak,  an  optimum reaction product  composition
should  be obtained.   At this  stage  of  the reaction,  however,  there
potentially is  incomplete consumption  of  the olefin.   Some  olefin  could
 be  emitted  from the process.
 Vacuum ultraviolet  reactions  --  Both  water vapor and oxygen compounds
                                     o
 absorb strongly in  the 1700 to 1900 A band and undergo  dissociation to

                                 39

-------
 c
 o
 tO
 i-
O
C
o
CJ

-M
C

-------
yield free radical species.   The applicable reactions are presented below.

          H20  +  h, (1700 - I960 A)-^H'    +  OH'        (20)
          H-    +  02                — ^H02-               (21)
          02   +  h  (1700 - 1900 A)-*-2 0(3P)           (22)
          0(3P) + 02   +  M         -MJa   + M          (23)

These reactions could result in the oxidation of NO to N02 with or without
the addition of olefins.  In the absence of hydrocarbons, the NO? product
quantum yield would be in the range of 1 to 2 assuming that radical
recombination is not a serious problem.  With olefins, the product
quantum yield and the reaction products would be very similar to those
discussed in the previous section.
Molecular complex photolysis —  The formation of molecular complexes
between S02 and either NO or NOp would create the possibility of a
primary photochemical process involving absorption within the charge
transfer band.  Based on a review of the ionization potentials of the NOX
compounds, it has been suggested  (Richards, Fox, and Reist, 1976) that
this absorption could be in the middle UV band.  The reactins involved
are entirely speculctive at this  t;i:e since ci.ly one S02 gas phase
complex has been studied (Christian and Grundnes, 1967).  The importance
of this reaction mechanism depends on the equilibrium constant of the
complex and the extinction coefficient of the complex, neither of which
are presently known.  The product quantum yields of these reactions are
probably less than 1 since non-volatile reaction products are formed.
There is no reason to believe that a free radical chain  reaction process
can be initiated by molecular complex photolysis.

     Excited state reactions— The absorption  of long wavelength UV
and visible light could  result  in the  formation of various excited
state species.   Generally,  these  species revert rapidly  to the  ground
state and  do  not participate  in chemical reactions.   The product  quan-
tum yields for the  possible reactions would be quite  low unless at high
concentrations the  excited species  are temporarily stabilized within a
cluster  or a  complex.

-------
Study Requirements
     The preliminary analyses suggest that photochemical systems are
an economically viable technique for NOV control assuming that the net
                                       A
product quantum yield of the combined reaction mechanisms is in the range
of 1  to 5.   Factors which influence quantum yields, such as olefin
concentration and pollutant concentrations are potentially important.
The extent of hydrocarbon consumption at the point of optimal  N02
concentration is a third possible issue.
                               42

-------
                             CHAPTER 5
                           EXPERIMENTAL

GENERAL DESIGN OBJECTIVES
     The experimental  apparatus was designed to simulate anticipated
field conditions.   An  artificial flue gas and components more sophisti-
cated than necessary for a full system were used to gain maximum control
of the independent variables.
     A continuous  stirred tank reactor (CSTR) was chosed as the irradia-
tion chamber since this was most compatible with the geometric configura-
tion of the high quality ultraviolet (UV) lamps.  Light absorption path
length and light distribution are more favorable in a CSTR than in a plug
flow reactor (PFR).  Due to the offsetting factors of holding time and
light intensity in the CSTR, it was possible to match light energy per
gram mole levels in this study with levels considered achievable in
commercial units.   Thus, the CSTR results can be related to conditions
to be expected in larger systems.  Risks inherent in scale-up are also
minimized since non-zero order reactions proceed more rapidly in real
gas streams represented by dispersed plug flow  conditions than in the
well-mixed conditions of this  study.
     Conditions within the experimental  unit and those  expected on a
commercial scale are compared  in Table 6.  Two  typical  utility boilers
have been listed:  one without NO  controls and one with state-of-the-art
                                 A
combustion controls.  With two  exceptions, all  variables are adequately
represented in the quartz CSTR system.   Despite its high concentration
in  flue  gases, C02 was not considered important either  as  a  third body
or  as  a  direct photochemical  reactant; and  therefore, it was omitted.
Humidity was  purposely minimized since the  reactants  would have  been
partially removed  in  the impingers.   Preliminary test work indicated
 that the reactions  are  not  highly  sensitive to humidity.

                                 43

-------
TABLE 6.  COMPARISON OF EXPERIMENTAL SYSTEM AND FULL SCALE SYSTEMS
Parameter
Dominant Flow
Characteristics
Gas Retention Time,
Seconds
Gas Flow Rate, SCFM
Flue Gas Composition
co2, %
o2, %
H20, %
NOX, PPM
S02, PPM
HC, PPM
CO, PPM
Particulate, qr/dscf
Gas Temperature, °F
UV Absorption, % Inp
2400-300C A
3000-4000 A
Watts UV Absorbed pe
Gram Mole NO
/^
N02, Photolysis Rate
Kr min"1
Quartz Reactor
Bench Scale System
Completely Mixed
45
0.04
0
0-8
1-3
0-2000
0-3000
0-2000
0-3000
0
250-350
Jt
20
6
r
0-100
»
0-5.0
Util
Full
Uncontrolled
Dispersed
Plug Flow
6-12
2xl05-2xl06
11-15
4-8
3-8
400-1200
1000-3000
10-100
10-200
0.5-2
250-325
99
70
50-100
0.2-2.0
ity Boiler
Scale System
State of the Art
Combustion Controls
Dispersed
Plug Flow
6-12
2xl05-2xl06
11-15
2-5
3-8
300-600
1000-3000
10-100
10-2000
0.5-2
250-325
99
70
50-100
0.2-2.0
                           44

-------
REACTOR DESIGN AND PERFORMANCE
      The dual  CSTR system  shown  in  Figure  12 was used in this study.  The
outer reactor is a 22"  diameter,  1/2" thick titanium sphere originally
built for use as a helium storage tank  for  a NASA Lunar Rover.  Access
holes were cut in the sphere and  brass  flanges were permanently mounted
using Armstrong 612 Epoxy resin.   A  front brass  plate was built to serve
as the lamp house and reactor carrier.   A 1/2" diameter SS 304 support rod
attached to this front plate was  used to mount and align the  inner reactor.
Whenever cleaning of the inner  reactor  was  necessary, it was  possible to
remove the entire assembly from the  outer sphere without disconnecting
any gas lines, lamp power cords or thermocouple  leads.
      The inner reactor had an  internal volume of  1.39  liters and was com-
posed of commercial grade quartz. The  front and back windows were made of
                                                             o
Suprasil 2 quartz with UV transmission  extending down to 1800 A.  A  ground
glass fitting was mounted on the back extension  of the  reactor  so that the
inner surfaces of the reactor could  be  flushed  with distilled water  for
cleaning.

Light Source
      An  Illumination Industries model  XL-2009-150 mercury-doped xenon arc
was used as a light source.  This was selected after a  preliminary  survey
of various high  pressure mercury arcs,  compact xenon arcs  and mercury-
doped xenon arcs.  The latter group was found to be highly temperature
dependent with  strong shifts in output spectrum with only 5-10°F changes
in lamp house temperature.   Standard xenon  arcs without mercury-doping  had
insufficient  energy output  despite  superior temperature stability and
start-up characteristics.   The mercury-doped xenon lamp was a compromise
choice  that allowed reasonable energy  output with only minor temperature
dependence.   None of the lamps had  any significant output in the 1700 to
      O
1850 A  band;  therefore,  study of  vacuum ultraviolet reaction mechanisms
was  impossible.   The  spectrum of  the mercury-doped xenon arc as measured
by an Optronic  Model  400 Radiometer is  illustrated in Figure 13.  It ex-
 hibits  the  peaks characteristic  of  mercury, specifically 2537, 2860, 2960,
 3030, 3130  and  3660  A.   Small  quantities of 1930 A light are also emitted
                                  45

-------
Figure 12a.   View of quartz reactor on support rod.
         Figure  12b.   View  of  outer  CSTR.
                         46

-------
0.9
0.8
0.7
0.6
0.5

0.4
0.3

0.2

0.1
22C
i i
-
-
A.
- -f
CNi
1
o
- X
O)
QJ
^J
c
0}
c
^_
O)
Q.
CO
_ +J
4_>
IT3

)0 2400 2600










|_j~n J I—
i
2800







•^



i l









J-
t
1MB















1







•




3000 3200
1 J
-









1 ^
-








-
-

-


_

	 —
3400 3600 380(
            Wavelength Nanometers
Figure "\2.   Mercury doped xenon  arc  output  spectra.

-------
due to the choice of Suprasil quartz lamp envelopes.   This  type  of  lamp
achieves 90% full output in 2 minutes and 98% in 3 minutes  based on  radio-
meter and spectrometer readings.  An Electronic Measurements  Model 308
Power Supply was used to operate the lamp.   Starting  voltages were appro-
ximately 35 kv and operating conditions were 20vdc and 6-8  amps.  The
lower value was used in most of the tests so that the 150-watt lamp  was
actually operating at 120-130 watts.  This  minimized  temperatures at the
base of the xenon arc and at the nylon insulators surrounding the anode.
Lamp House
     The xenon arc lamp is  mounted  with the anode 10° off vertical which
is within the manufacturer's specifications.  The lamp bulb is 1-3/8"
from the front window of the inner  CSTR.
     A polished aluminum conical reflector  is located directly behind the
xenon arc to improve energy delivery to the reactor.  N02 photolysis tests
indicated that 40 to 60% of the total  near  UV energy absorbed  in the re-
actor was reflected from the aluminum cone. Aluminum was chosen due to
its high reflectivity in the middle UV  band.  The lamp bulb was  never
touched or cleaned with solvents.  A dry piece  of lens tissue was used
daily to clean the bulb.
     The distribution of light within the quartz reactor was  measured
qualitatively  using thermocouples  mounted  on a  flexible, movable rod.
The millivolt difference between a  black thermocouple directly in the
light path and a white thermocouple shielded from direct light was tak-
en as a measure of light intensity.  Due to the mounting equipment neces-
sary to support the thermocouple rod inside the  quartz reactor,  it was
impossible to perform the light distribution tests  inside the  large  sphere.
Reflected light from the titanium surfaces  may moderate the light inten-
sity gradients shown in Figure 14.   It  is apparent  that light  intensity
is low near the walls of the quartz reactor.
     Lamp operating currents were monitored with a  Simpson Microampmeter
connected parallel to a 0.25 ohm shunt.  Meter  values represent  1.7xlO~
of actual currents.
     A set of 3 aperture plates was constructed  of multiple layers of
decorative metal foil of 1/16" thickness.   The  conventional diaphragyi
aperture was considered inappropriate since the intense lamp  house tem-
                                48

-------
VO
                                                                1.1   REAR
                            2.2
                            1.
1.0

i
             1.0
           SHOWN ACTUAL SCALE

Light distribution expressed as difference in
exposed thermocouple and shielded thermocouple
at locations shown
                      Figure  14.  Light distribution in quartz reactor.

-------
 peratures would  have  vaporized the  lubricating oil and warped the shutter
 blades.  NO-  photolysis tests on various occasions indicated the follow-
 ing  intensity ratios:
                    Plate 1       0.16
                    Plate 2       0.30
                    Plate 3       0.57
                    Open (Basis)  1  .00
 These plates were used to select the general light intensity range de-
 sired.
     The lamp electrodes extended through the interior walls of the fil-
 ter  house.  Cylindrical nylon was used to prevent sparkover of the high
 voltage starting pulse.  A constant stream of commercial  grade N? at a
 rate of 15-20 liters per minute aided lamp house heat dissipation and
 precluded ozone formation.   The N? flowed axially around  the lamp,  per
 manufacturer's recommendations.  During certain  tests when heat dissipa-
 tion was inadequate, slight vaporization of the  nylon insulators occurred.
 This was readily apparent on the relatively "cold" surfaces within the
 filter house such as the aluminum reflector and  the interior brass  walls.
 No discoloration was ever noted on the lamp or reactor face.
     For certain runs, it was necessary to filter out a portion of the UV
 emission spectrum.   A 3" O.D. Suprasil  quartz window  was  used for full
                  o
 spectrum   (>1800 A) runs and a commercial  glass  window was used to iso-
 late the near UV (>3200 A).
     Lamp current was  the principal  independent  variable  affecting  ultra-
 violet energy output.   The  k, data for both lamps used in the experimen-
 tal run indicated a consistent relationship with  lamp current.   The
 linear regression equation  representing this data of  Figure 15  is shown
 below.  This  equation  was calculated after throwing out the March 16th
and March 18th data points.   These low values were contributed  to visi-
                         y  = 0.96 + 2.52  (x-10)
where:  y = k,,  min
        x = lamp current, amps
            (as  read by Simpson
            Microampmeter.)
                                50

-------
 c

 i
 X 3
0)
to  0
>> 2

'o
•V
o
£
D.
               I
              I
L
                                     <>• March  16




                                        March  18
             10.0




        Figure 15.
   10.2       10.4      10.6

        Lamp  Current,  Amps
        10.8
11 .0
N02 photolysis rate  as  a  function of lamp current.

-------
 ble deposits  on  the  inner  surfaces of the reactor during high S0? con-
 centration  runs  (>3000  ppm SCL).  The slope of the k, plot of Figure 15
 is  slightly higher than corresponding plot of total UV volts versus cur-
 rent as determined by the  Radiometer.
 Reactor Gas Flow Characteristics
      The gaseous mixing characteristics determine the average local re-
 actant concentrations in the reactor zones where secondary thermal reac-
 tions are proceeding.  The resulting influence on competitive reaction
 mechanisms  is prejudiced against second and third order reactions when
 non-plug flow conditions occur (Levenspiel, 1972).  The inner reactor
 was  purposely designed with minimum possible plug flow conditions since
 this minimizes several non-photochemical NC^ formation reactions and
 since it represents the slowest possible reaction times for all  mecha-
 nisms.
     A set of tracer tests were performed in order to evaluate flow pat-
 tern characteristics as a  function of flow rate.   The procedures of Wolf
 and Resnick (1963) were used.  After a steady state NOV concentration was
                                                      A
 established in the reactor for a period of five bed volumes, the NO  flow
                                                                   A
was abruptly stopped.  The response to this step change in inlet NO  flow
                                                                   A
was measured using the TECO chemiluminescent monitor.   The basic equation
 (3) of Wolf and Resnick was used to analyze the data.
  F(t) = 1 - exp
(Equation  3)
where:  F(t) = fraction of material  which remains in the reactor
           f = fraction of gas flow short circuiting
           d = fraction of reactor volume with only diffusion mass
               transfer (dead space)
           r = residence time correction factor
           a = fraction of non-dead space which is well  mixed
           p = fraction of non-dead space which is in plug flow
           e = average residence time
           t = elapsed time
           L = lag
           m = lag factor
                                52

-------
By assuming lag  is  negligible  and  that  the  average residence time is
known, Equation  3 can  be rearranged  to  give Equation 4.  This was used
to match results from  the tracer tests.
     [~T
[l^jl
                                            +       (Equation 4)

Since one equation with three independent variables cannot  yield a unique
solution, some judgment is necessary to derive a  reasonable solution.
The reactor configuration studied in these tracer runs  is  shown  in Fig-
ure 16.  The small diameter Teflon tube on the inlet side  was  added  af-
ter preliminary tracer work indicated serious  short circuiting.  This
tube plus the outlet tubing leading to the NO   instrument  added  a  plug
                                            /\
flow section to the "reactor".  The long neck  near the outlet  tube intro-
duces a  potential dead space of up to 5 percent.   These factors  were con-
sidered  in the trial and error solutions.
     Results are  summarized in Table 2.  Corresponding graphs  illustra-
ting the match between Equation 4 values and the observed  l-F(t) versus
t/9 curves are provided  in  Appendix A.
                     TABLE  7.  TRACER RESULTS
Flow Rate
(1pm)
1.0
1.5
2.0
2.5
3.0
% CSTR
92
91
92
85
72
% Plug
Flow
8
9
8
15
28
% Dead
Space
2
2
2
4
8
% Short
Circuiting
6
6
6
1
1
      Mixing characteristics  within  the  1  to  2  Iiters-per-m1nute range
 closely approach design objectives.  Dead zones  are  somewhat smaller than
 expected while shortcircuiting is slightly greater.  At the higher flow
 rates, it is apparent that some plug flow character  can be maintained
 across the quartz sphere.   An increase  in dead space at 3.0 liters per
 minute is probably due to isolation of  gases in the  outlet  neck.  Over
 the range of flow rates used in the study, CSTR conditions  can be as-
 sumed to prevail with only slight error.
                                 53

-------
  Suprasil Window
                                                           Suprasil  Window
   1/4"  O.D.  Inlet-
   Extension
 •1/4" O.D.  Outlet
Extension
              Figure  16a.   Initial  inlet  configuration.
 1/8" O.D.
 Teflon Tubing
1/4" O.D.
Inlet
Extension
1/8" O.D. Teflon Tube
                                            NOTE:  Drawings not to scale
              Figure 16b.  Reactor inlet configuration.

                                54

-------
ACTINOMETRY
     The experimental  apparatus included actinometer systems  for  both
the near UV (3000 to 4000 A) and the middle UV (2400 to 3000  A) spectral
bands.  The function of these actinometry systems was to provide  a mea-
sure of light absorption so that the N0? net product quantum  yield de-
fined in equation 1 could be calculated.  Quantum yield can be related
            _
        N°?   TIN - +fi - JTET       (Equation 5)
              LV NEAR  MIDDLE" J
where:     NNQ    =  Gram moles of N02 formed in CSTR
           NNEAR  =  Quanta of uv ]l"9ht absorbed in the 3000 to 4000
                     A band.
           NMfnm r - Quanta of UV light absorbed in the 2400 to 3000
            pIIUULL — o
                     A  band.
           t      =  Unit of time
 to  the overall  energy requirements using the lower curve in Figure 9.
 The rate of N02 formation in the CSTR  (numerator of Equation 5) can be
 calculated using the steady state N02  concentration in the reactor efflu-
 ent and the effluent flow rate.  This  is shown  in Equation 6.

                                                    n     /I gram mol e
                                                    N02
iu   /+\  - (r  1 iters^ (7Q°F\ fcram  Mol es   .  .  ,\
(NNO?  }  " 1G  ""Tnj \T~j \22.4  liters  totalj
    *                                         l
                                                      C   -~
                                                      SS/ 10b  ppm N02
                              (Equation  6)
 where:   G      =  the CSTR effluent  flow rate
         T      =  the gas  temperature at the reactor  outlet
         ^  2'ss = steady state N02 concentration  in the  reactor
      Acetone photolysis was used to  measure middle UV energy penetrating
 through the quartz sphere.  The difference between runs  with and without
 reactive gases in the quartz sphere served as a measure  of the N..Trjrj, r
 term of Equation 5.
      The acetone pressure was chosen to maximize UV absorption within the
                                    o
 outer CSTR.  Over the 2400 to 3000 A spectral range, UV  absorption ranged
                                  55

-------
 from  75  to  94%.  The  results of  Howe and Noyes  (1936) indicate that the
 CO  quantum  yield of the overall  set of reactions is approximately equal
 to  one at the acetone vapor pressure, gas temperatures and UV spectral
 bands used  in the actinometry  system.  Thus, with only minor corrections
 the CO yield provided a direct measure of middle UV energy input to the
 quartz sphere.  The CO was measured in a Miran Dispersive Infrared Analy-
 zer.  A  complex system of dry  ice and liquid nitrogen baths were used to
 remove interfering vapors, particularly acetone.  Numerous cold trap de-
 signs were  evaluated, however, none could operate for longer than 45 min-
 utes without plugging due to acetone freezing.  Because of this problem
 the middle UV actinometry system was abandoned.
     The inability to measure light absorption in the middle UV forced a
 redefinition of the NOp product quantum yield term to be measured in the
 study.  This new definition is given in Equation 7.  Since the contribu-
 tion of middle UV light is ignored in the revised definition, the calcu-
 lated product quantum yield should be greater than or equal  to the actual
                                                         o
 quantum yield for the entire UV spectrum of 2400 to 4000 A.   The values
would be equal  only in the case that UV absorption in the middle UV band
does not enhance or reduce N02 formation.
                      *          -
                       N02(calc)     IMU2         (Equation 7)
                            H(ca1c)   >}       (Equations)
     The revised definition of N02 product quantum yield can be related
to the overall energy requirements of the system using the upper curve of
Figure 9.  This approach overestimates the actual  energy requirements by
the factor represented by the ratio of quantum yields (Equation 8).
     The NNEAR term was determined using the N02 photolysis procedures
of Sickles and Jeffries (1975).  The N0£ photolysis is measured by
irradiating a N02-N2 gas stream and observing the steady state
                                56

-------
 concentrations  of NO and  N02  in  the  reactor effluent.  The photolysis
 rate,  k-,  is  calculated using  Equation  9  provided in their paper.

               A[NOJ  f          Ro([NO]o  + A[NOJ)]
      Klss=^^A7L1  +Rl     	[N02]ss      J   Aquation 9)

where:   A(N02)  =  the drop in N02 concentration  between  inlet and outlet
                   streams
        (NO.,),. =  the steady state N00 concentration  in  the  reactor
           i. SS                       £.
                   average residence time
        (NO)Q   =  concentration of NO in the  reactor  influent
        R1      =  0.27
        R2      =  0.25
The k-j  tests were done in the actual  CSTR to account for  internal
light reflection and the non-ideal light distribution  conditions discussed
earlier.  Tests were done at least once per day.  An attempt  was made to
perform the k, determination at the temperatures expected in  the runs,
however,  some deviation generally existed.  Results of three  successive
tests indicate a drop  in observed k, values of 20% over a temperature
range of  275 to 125°F.  Based on  these results, corrections were consi-
dered unnecessary  for  the relatively small temperature variation between
the runs  and the applicable k, test.
      The k.j value was used in Equation  10 to calculate the NN™R term
of the  product  quantum yield.  Substitution of the right side of this
equality  and  the  right side of Equation  6  leads to a very simple defini-
tion of the  N02  product quantum  yield.
                  /-mo A A-     i< i   -r *. t \       /Gram Mole NO,
 /N    /tx_,,       /70°FVGram  Moles Total
 ^EAR71;"MitersI  T  A     22.4 Liters
                       «e"N02)    Cation  ,0)
 where:  k,  = the NO^ photolysis rate
         T  = the gas temperature in the CSTR
         V  * the volume of the CSTR
                                  57

-------
             *Mn        = G/V-k  = 1/rk,     (Equation  11)
              NU2(calc)        '        '
       Equation 11  suggests  that  the N02  product  quantum yield  is a depen-
 dent variable of the N02 photolysis rate,  k,.  This unusual  condition  re-
 sults because the  photochemical  system utilizes  NCL photolysis  to produce
 N02.  Equation 11  provides  a  very convenient means to  relate the observed
 steady state NC^ concentration with the  energy requirements  of  the system
 as indicated on  the  upper curve  of Figure  9.

 REACTANT AND PRODUCT MEASUREMENT
       The diluted  effluent  stream in  the outlet  manifold was analyzed with
 a  set of gas analyzers  shown  in  Figures  17 and 18.  All of these instru-
 ments were calibrated by preparing  a  known gas concentration in a 40-liter
 Teflon  bag.   In  initial  tests, all  the components in the system used were
 mixed together during calibration.  There appeared to be essentially no
 interferences at the  undiluted concentrations of initial  components.
 Nitrogen  Oxides Analysis
      A Thermo Electron Model 10A Chemiluminescent NO-NO  Analyzer was
                                                        /\
 used.  This  instrument  utilizes  the reaction between  NO and 00  which  re-
                                 o                            o
 suits  in  light emission at 3100 A.  The light intensity measured by  the
 photomultiplier assembly is linearly related to the concentration  of  NO.
The concentration of combined NO  species is determined by reduction  to
                                A
NO in a metal coil  at 650°C-700°C.  The difference between the  NO  and
N0x values is representative of the N02 concentration  plus an unknown
number of other species such as nitrous acid,  nitric acid  and peroxyacyl-
nitrate.
      The Chemiluminescent  instrument was calibrated routinely  using  Tef-
lon bag mixtures containing  known quantities of NO, CL, S02  and propylene.
The instrument was  operated  using laboratory air  instead  of  02  since  the
diluted outlet concentrations rarely exceeded  500 ppm.
Sulfur Dioxide Analysis
      It was necessary to use a pulse fluorescent technique  in  order  to
ensure S02 specificity.  A Thermo Electron  Model  40 Analyzer  was chosen.
                                 58

-------
Figure 17.   Gas analyzers  for NOX,  S02»  and
            condensation nuclei.

Figure 18.  Gas analyzer for hydrocarbons.

                   59

-------
 It operates by excitation  of SO,  to  the  singlet  state  by  irradiation at
           o                    £
 2200-2300 A using a  pulsed UV source.  The  excited molecules decay to the
                                                            o
 ground state by two  possible pathways  -  fluorescent at 3400 A and quench-
 ing.   Light from the fluorescence reaction  is linearly related to the SOp
 concentration.
                S02   -i- hv(2200-2300 A)     S02*        (24)
                S02*                   -*-  S02 + hv    (25)
                S02*  + m               +-r  S02 + m     (26
       Using  the Teflon bag mixtures with typical  (diluted) reactor efflu-
 ent,  it  was  observed that the fluorescence is sensitive to N02 concentra-
 tion,  a  product of the photochemical reactions.  The significance of this
 interference is illustrated in Figure 19.  It is proposed that this pre-
 viously  unreported negative interference is due to increased quenching ef-
 ficiency via a contact molecular complex, as indicated in reactions 27
 and 28.
             S02*          +N02- [(S02)(N02)]*    (27)
             [(S02)(N02)]* + m   + S02 + N02 + m1   (28)
 The molecular complex would provide additional vibrational degrees of
 freedom  which would facilitate radiationless decay.  It was necessary to
 adjust observed S02 values according to the prevailing N02 concentration
 due to the negative interference.
 Hydrocarbon Analysis
      Gas chromatography analyses were performed on a routine basis using
a Varian Series 2440 Analyzer equipped with a fame ionization detector.
A 10 cc  sample loop was purged continuous with outlet manifold gas and in-
jections were then done twice during each irradiation period.  A  polypax  3
meter column at 80°C was used for separation.  Calibration was again done
using Teflon bag samples of known concentrations.
Condensation Nuclei  Analysis
      An Environment/One Rich 100 condensation nuclei monitor was  used to
evaluate possible aerosol  formation.   For a  portion of the irradiation
period, the Millipore filter  was  taken  off line to permit  unfiltered gases
                                  60

-------
      -40
 (VI
I
c
0)
o

0)
Q.
      -30
-20
      -10
                                        I
                                                          I	I
              100   200   300   400   500   600   700   800     900   1000    1100   TJOO   1300 1400

          [N02] PPM

                               Figure  ID.   Influence of NOo on  pulsed fluorescent  SO^ Readings

-------
 to  enter  the  outlet manifold.  This  instrument exposes the sample gas to
 a  300%  supersaturated condition which causes condensation of water vapor
 on  condensation  nuclei.  A  portion of these grow into the light scatter-
 ing  size  region  causing a reduction  in light transmission in a tubular
 cell.   This instrument qualitatively indicated the presence or absence of
 large quantities of condensation nuclei.
 Spectral  Analysis
     A  number of spectral changes were anticipated due to aging of quartz
 windows,  lamp deterioration and reactor cleanliness.   The k,  responds to
 a relatively broad spectral band and would not serve  as a useful  indica-
 tor  of  such developing problems.  For this reason,  a  Heath Model  703  IR/UV
 spectrometer was used.  The PM tube detector module was placed on  the cen-
 terline defined by the xenon arc and reactor (see Figure ?0).   A  Hewlett
                                                                 o
 Packard x-y plotter was used to record  spectra from 2400 to 3700  A for N?
 and  during actual test situations.   A set of spectra  was generally taken
 once per  run.  Relative peak height values on  the N?  spectra  revealed the
 deterioration effects listed earlier.  The differences in the  two  spectra
 indicated major absorbing species.

 GAS  HANDLING APPARATUS
     The  experimental  apparatus was designed to maximize flexibility  in
 and to facilitate routine checks  for nonoptional  performance  caused by
 leaks, surface mediated catalytic  reaction or  other factors.   The  system
was composed of a number of separate units including:
                            (1)  gas  blending
                            (2)  heaters
                            (3)  temperature measurement
                            (4)  ventilation
                            (5)  humidification
Gas Blending
     The simulated flue gas  was prepared  by blending  of pollutant
gases and oxygen  into  a nitrogen carrier  gas stream.  This permitted
complete flexibility  in the concentrations of  all components.   The
 blending box is  shown in Figure 21.  All  gas lines  from the cylinders

                                 G2

-------
Figure 20.   Side view of reactor and spectrometer.
     Figure 21.  Front view of blending box.

                        63

-------
 to the blending  box and within the blending box were Teflon so that hy-
 drocarbon  contamination could be minimized.  As shown in Figure 22, the
 N2 and 02  were mixed  first  followed by humidification in Greenberg-Smith
 impingers  filled with distilled water.  Part of the humidified gas stream
 was drawn  off to a Dew Point Hygrometer for measurement of humidity.   The
 N,-0,>  stream was mixed with S09, NO  and hydrocarbon lines to complete
  tL  c.                         £    X
 the blending process.  All flows were monitored with calibrated rotameters
 shown  in Figure  21.
     The simulated flue gas stream was split in half with one portion
 going  to the reactor  via a tube furnace and the other half proceeding di-
 rectly to  the outlet  manifold solenoids.  A pressure gauge on the reactor
 side of the line served as a very sensitive indicator of flow variations.
     The gas stream entering the outlet manifold was diluted a factor of
 2 to 5 by  dried  laboratory air.  This was necessary to support the large
 flow requirements of  the gas analyzers feeding off of the outlet monitor.
 The complete gas flow system is illustrated in Figure 22.
 Heaters
     A tube furnace shown in the center of Figure 21 served as the ini-
 tial reactor gas heating system.  The temperature of the heated gas gene-
 rally  was  125-150° Fat the front flange of the large reactor.
     A series of seven 100-watt power resistors were mounted in the outer
 CSTR.   By  varying the voltage from 0 to 85 V.A.C.  across the parallel
 circuit, it was  possible to control  gas temperatures inside the quartz
 CSTR from  125°F  to >300°F.  By heating from the outside  walls  of the
 quartz CSTR, it was possible to keep the reactor walls at least 10°F
 higher than the  inner gas  temperature.   This minimized deposition of
aerosols on the  reactor's  inner walls.
Temperature Measurement
     The gas temperature inside the quartz reactor was measured using  a
Chromel-Constantin thermocouple (Type E thermocouple).   The zero to 10
multivolt signal  generated were measured  on a  Model  10A  Fluke  digital
Multimeter (see top right, Figure 21).
     The thermocouple probe was  located on the quartz  reactor  center!ine,
                                64

-------
CJ1
9                                                                                                                        Pressure
                                                                                                                        Gauge
        uosi
                    to Hooc!
1
ik
1


'
Monitor



, 1 '
Monitor



Mon1 tor

r

1
h
&
                                                                           To *od
                  To Hood F        To Hood r        To Hood^
                                                                 Rotaneter
                                                                               Ur1er1te
                                                                               Column
                                                      Figure  22.    Gas  flow  system.

-------
 directly above the  gas  exit  port.  The  probe was loosely wrapped with
 several  layers of white  Teflon  tape  so  that radiant energy would not
 bias  the results.
 Humidification
      The combined N2 and 02  streams were humidified in a set of 2 500-ml
 impingers filled  with distilled water.  Desired humidities were achieved
 by  bypassing some of the gas around the impingers.  Treatment of the simu-
 lated  flue gas would have allowed higher humidity; however, the reser-
 voirs  would have  collected soluble gases such as S02 and N02.

 EXPERIMENTAL DESIGN
     A preliminary test series was performed to aid in the preparation
 of  the experiments.  Considering that over 10 major variables were con-
 sidered  of possible importance, a complete factorial  design was not con-
 sidered  feasible.  Even with the rapid turnaround capabilities of CSTR
 systems,  the maximum number of tests which could reasonably be done with
 available  resources was estimated to range from 150 to 200.  It was nec-
 essary to  screen  the variables and subdivide the test program so that
 this limitation was not exceeded.
     The  very rough initial  tests indicated that the variables of most
 importance with respect to NO oxidation  were the UV light intensity and
 the hydrocarbon concentration.  Accordingly, a 5 x 5 x 2 factorial  design
was used  to study these variables at NO   levels of 600 and 1000 ppm (v/v).
                                      /\
The ranges of the variables  roughly approximate the levels considered
achievable in commercial systems.  Hydrocarbon concentrations well-be-
yond stoichiometric requirements were used in order to identify the
 point, if any,  that the system becomes hydrocarbon  insensitive.   Speci-
 fics concerning this major test series are provided in Table 8.   The
 test series was repeated once so that interactions  among these three
 variables  could be identified.
     A number of small  factorial test series were devoted to secondary
variables  also observed to influence NO  oxidation in the preliminary
 test work.  These runs were done at equal  light intensities and hydro-
 carbon levels so that the sensitivity of the system to each variable

-------
could be qualitatively compared.   Specifics  concerning  these  studies are
also covered in Table 8.
     Most of the series in Table  8 provided  information concerning varia-
tions expected in commercial  systems.   The  range  in  SC^ levels, 02 levels
and gas temperatures were selected to  correspond  to  the anticipated levels
at utility boilers.   The  spectral  characteristics  study (series 5) was in-
tended to aid in the selection of lamps for  subsequent  pilot  plant stu-
ies.  The hydrocarbon comparison  study (series  6)  served to evaluate
relative advantages  of each compound in terms  of  NO  conversion
efficiency.  This information would be of use  in  cost estimation  and
design of subsequent work.
     Despite the high concentrations of C02  in  boiler effluent  it was
assumed that there would  be no relationship  between  the NO oxidation ef-
ficiency and the level of C02-  Series 7 was included  in the  study to
check this assumption.  The possible beneficial impact  of carbon  monoxide
was studied in Series 8.

TEST  PROCEDURE
      Each  run consisted  of three  distinct phases:   (1)  establishment of
dark  phase steady state  conditions with respect to gas  flow rates, re-
actor  temperature and pollutant concentrations, (2) a  15- to 20-r,iinute
irradiation  period,  and  (3) a recheck  of dark  phase steady state condi-
tions.   In this manner,  it was possible to  check  for drift of pollutant
concentrations, temperature,  and  other factors which would have  influ-
enced  the  reactions  during the irradiation  period.  During the initial
dark  phase period the rotameter readings, temperatures, and gas  analy-
zer values were recorded along with information concerning the run.   Af-
ter this data was recorded in the project notebook, the lamp was started.
Immediately after lamp start-up,  the  reactor effluent  was diverted to the
Millipore  filter.   After a warm-up period of 4.0  minutes, the Heath Spec-
                                                  o
trometer was started on  a scan beginning at 2450  A  and extending to 3700
o
A.   The complete  scan required 11  minutes.  At the  5.0-minute point after
lamp start-up,  a  GC injection was made and  the NOX  and S02 analyzer data
was recorded.   Rotameter settings, temperatures,  lamp  currents,  and other

                                 67

-------
                                             TABLE 8.   EXPERIMENTAL DESIGN
O>
Series
No.
1




2




3




4



5



Variable
1
Light
Intensity



[S0~]o
c.



[02]
£.



Tempera-
ture


Spectral
Charac-
teristics


Variable
Levels 2
0 (min"1)
0.25
0.50
1.00
2.00
0 ppm [NOjo
500 x
1000
1500
3000
0 % [NOjo
1
3
5
10 %
125°F [NOjo
175 x
225
275
2400A Light
3200A ^tensity


Total
Number
Variable of
Levels 3 Levels REPS Tests
0 [NO ]0 600 2 100
250 x 1000 ppm
500
1000
2000 ppm (v/v)
600 - 1 10
1000



600 - 1 10
1000



600 - - 1 8
100


0 (min"1) - 1 1Q
0.25
0.50
1.00
2.00

-------
                                                  TABLE  8 (continued)
01
UD



Series
No.
6
7


8




Variable
1
Hydro-
carbon
Tvpe
[co2]


CO




Variable
Levels 2 Levels
Propylene Hydrocarbon 250
Ethyl ene Concentration 500
1000
2000
0%
8
12
20
0 -
1000 ppm
1500
2500
Total
Number
Variable of
3 Levels REPS Tests
1 8
1 4


1 4



-------
data of interest was then checked.  At the 10.0 minute point, another
GC injection was made and the NO  and S0? analyzer data was recorded.
                                /\       L-
Rotameter settings, temperatures, lamp currents and other data of
interest was then checked.  Near the completion of the irradiation period,
the Millipore filter was taken off line to prevent exposure to dark phase
conditions.   The CN counter was observed for several  minutes to determine
if aerosol formation was detectable.   When all  necessary data had been
recorded,  the lamp was shut down and time was allowed to reestablish dark
phase steady state conditions.  All  flow rates, and pollutant concentra-
tions, and gas temperatures were then rechecked and recorded.
     After each run, the lamp filters and/or aperture plates were changed.
At the beginning of each day, the lamp condition was checked using the  k,
level.  After each day's run, cumulative lamp operating hours and the
number of starts was recorded.  The xenon-mercury arc was replaced when
operating  time exceeded 100 hours or when the number of starts exceeded
200.
                                70

-------
                            CHAPTER 6
                      RESULTS AND DISCUSSION

     The experimental  program has been  designed  to address five basic
factors concerning photochemical  system operating characteristics and
cost.  These principal  issues are listed below:
     1.  What is  the relative importance of UV absorption  in the middle
     UV spectrum?

     2.  What olefin concentrations are necessary to achieve adequate
     NO oxidation?
     3.  At the point of optimal  NOp formation,  what fraction  of the ole-
     fin has not reacted?
     4.  To what extent do the flue gas chemical and physical  characteris-
     tics affect the photochemical  system performance?
     5.  What secondary products  are emitted  from the reactor?
     Data from the 159 runs performed have been  divided  into five sec-
tions, each one of which is devoted to one of the  issues listed above.
These analyses culminate in an assessment of  order-of-magnitude annua-
lized costs.
     While the primary purpose of this project is simply to  evaluate the
competitive potential for photochemical NOV control , some attention is
                                          A
also given to the identification  of possible  reaction mechanisms.  This
is useful in transferring information gained  in  this study to  consider
other  industrial applications.

SPECTRAL EFFECTS
NO Oxidation Effectiveness
      Filters were used to test the  relative effectiveness of near UV  and
middle UV  light.  A  standard  piece  of window glass was used to eliminate
light of wavelengths   3000 A  and a  piece  of SuprasiT^or quartz was used
                                 71

-------
 for  full spectrum tests.  Use of a Suprasil quartz filter was not strict-
 ly necessary since the front window of the CSTR is composed of similar
 material.  The filter was added to equalize the effects of light reflec-
 tion.  Each of these filters were 1/4" thick, 2" diameter ?nd were
 placed between the lamp and the CSTR front window.
     Heath Spectrometer scans indicated that the window glass filter re-
 duced UV light of <2960 A by 99%.  It also significantly reduced radia-
             o          o
 tion of 3030 A and 3130 A (two major peaks in the lamp  output spectra) and
 it moderately attenuated the entire near UV spectrum entering the CSTR.
 This unintended impact on the near UV band was quantitatively accounted
 for  by performing k,  tests with each of the filters.  The ratio between
 the  observed k] values was 2.04 (>1800 A/>3000 A) ±0.40 based on 3  sepa-
 rate determinations done over a 4-week period.
 Differences in NO Oxidation
     The extent of NO oxidation at equivalent k,  levels is the criteria
 used to evaluate spectral  effects.  The aperature plates were used  to
 vary the incident light intensities  to give 4 distinct  levels.   Results
 of eight tests, four  with each filter, are presented in Figure 23.   The
 upper curve is the NO conversion in ppm (v/v) using the Suprasir^
 quartz filter while the lower curve is the NO conversion using the  win-
 dow glass filter.  Other variables such as propylene concentration,  oxy-
 gen concentration and gas  temperature were held relatively constant  as
 indicated by the data summary table included within Figure 23.   The
eight tests were performed in random order.
     The shaded region between the two curves represents the  contribu-
tion of middle UV light.  At the UV  light intensities which will  be  ty-
 pical of commercial  installation, namely 0.2 to 0.4 mirr'  adjustrd  k-, the
middle UV appears responsible for 65 to 85% of the NO converted.   This is
 remarKaole considering that the middle UV output  from a mercury doped
 xenon arc is only 15% of the output in the near UV band (see  Section 4,
 Figure 13).
     The $.,-. /  -,  \/$wn  factor introduced earlier appears to be a  func-
          iiUn^caicj  NUp
 tion of the total  light intensity entering the reactor.  At low levels of

                                 72

-------
500
400
 i


 s.
 a

r—i
 O
 z
l__l
 <
300
200
100
Variable
NOn, ppm
rt
C3Hg, ppm
S02, ppm
02,. %
Temp., °F
T , ' m i n .
Min
596
1302
1312
3.8
251
0.737
Mean
627
1377
1428
4.4
273
0.760
Max
665
1441
1500
4.7
295
0.783
                0.20         0.40         0.60

                NO, Photolysis Rate, Min";
                                                           0.80
1.00
                    Figure  23.   Influence  of middle  ultraviolet  light on NO oxidation,

-------
    light,  this ratio is  1  to 3.   The ratio approaches  1  at very high  light
 intensities presumably because reactions initiated in  the near UV are
 sufficient to convert the  available NO.   This revised  data indicates  that
 the k-j  must be in the range of 0.2 to 0.4 min   in order to keep energy
 demand  less than  2% of station power.
 Possible Causes of Spectral  Effects
      The Heath Spectrometer was  used to  identify possible absorbing
 species in the middle ultraviolet  band of 2400  to 3000 A which  could  be
 responsible for the enhanced NO  conversion.   It was  necessary  to analyze
 experimental  runs  performed  without the  filters.
      Figure 24 includes  two  separate spectra; one taken  during  the
 actual  run and one taken after all  gas streams  except the dry  N2 stream
 have  been  turned  off.  Strong  UV absorption  is  apparent  in  the  band from
              o
 2600  to 3030  A.   This  set  of  comparative spectra is characteristic of
 virtually  all  spectra  obtained during  the experimental program.   For  a
 small set  of  the  runs, however, attempts  were made to determine what
 fraction of the middle UV  absorption  could be assigned   to  S02  and what
 fraction was  the result  of  other primary  absorbers such  HCHO,  CH3CHO,
 HN02, and  EDA  Complexes.   This was  accomplished  by modifying the  procedure
 used  in taking the  second, "baseline"  spectra.   In this  set of  runs,  the
 S02 injection  rate was kept  constant and  only the NOX inlet stream was
 turned off.  Since  the latter  volumetric  flow rate was quite small
 relative to the total, the S02 concentration in  the "baseline"  spectra
was almost equal to that used  in the run.  The result of  these  comparative
spectra is  shown in Figure 25.  S02 obviously accounts for essentially
all middle  UV  absorption measurable by means of  the Heath Spectrometer.
     Considering the apparent  dominance of S02 absorption,  it is  logical
 to propose  that S02 participates in one or more  primary  photochemical
 processes which contribute to  MO oxidation.  Possible reaction  mechanisms
 include EDA comples reactions  and singlet  S02 reactions  (excited  state
 possibly stabilized within a complex or cluster).  A series of  10
 experimental  runs  were performed to  check for the role of S02  in  the
 overall reaction  scheme.   The  S02  content in  the reactor influent was
 varied  from 0 to  3000 ppm  to correspond  to the  general range  in coal-
 fired boilers  of  500  to  3000 ppm.   The tests  were divided into  five low
                                 74

-------
   100 -
    80 -
01
    60
    40..
    20,-
                  2550
        Wavelength A
2675
2800
2925
3050
3175
                                                      Spectra, N2  Purge
                                                      Spectra, Reactive
                                                      Gas  Mixture  (S02
                                                                                  NOX,  C3H6)
                                                                                 NOTE:  Spectra,,
                                                                                 merge at 3150 A
3300
                                                                                   3425
3550
                          Figure 24.   Typical  comparative spectra; during and after experimental run.

-------
   100
    80
    60
en
    40
    20
         C
         
-------
NOX inlet runs and five high NOX inlet runs.   The results of interest
were the extent of NO oxidation across this wide range of S0£ concentra-
tions.  These are presented in Figure 26 along with a summary of the
variability of other factors.
     The upper line of Figure 26 is the observed SOg effect on NO oxida-
tion with the inlet NOX concentrations averaging 544 ppm.   While there
may be a discernable association, the influence of S0£ on this type of
system is insignificant.   In the case of the high NOX inlet tests
( average of 1165 ppm), there is a definite suppression of NO oxidation
at the higher S02 concentrations.  This is contrary to the trend  which
could be inferred from the spectral band tests.  The probable reason  for
the negative effect of S02 on the overall system is the hydroxv radical
reaction listed below:

                    •OH + S02 + OHS02    (29)


This reaction would compete with hydroxy attack on the propylene which is
important in maintaining the free radical chain process resulting in  NO
oxidation.  The HOS02 radical formed in the above reaction would react
further to form probably sulfuric acid, peroxysulfuric acid, and related
compounds according to the reactions proposed by Calvert et a!. (1977).
At the high S02 concentrations tests, aerosol deposits containing con-
siderable sulfur were observed.  This confirms that S0«  is being oxidized
and the above reaction is the only known reaction fast enough to account
for the conversion to aerosol precursors.
     The S02 test series demonstrate that S02 absorption is not respon-
sible for the strong effect of middle UV light.  Based on these results,
it appears very unlikely that EDA complexes composed partially of S02 or
that excited S02 reactions have a significant  impact on  NO conversion.
It is necessary to identify other possible middle UV absorbers to account
for the spectral effects.  Three possible  species are 03,  HN02 and H20,>.
All of these absorb  strongly  in  portions of  the middle UV  spectrum, and
the photolysis reaction yield free radicals which could  enhance  the
oxidation process.   The reactions are provided below:
                                 77

-------
          90

          80

          70

          60
                                         Low Inlet NOX Tests
                                   High  Inlet NOX Tests.
oo
      o
      o
      o
40

30

20

10
                        500
Variable
NO -low, ppm
NO -high
k, , min
C3H   ppm
                                                          3
                                                         (v/9)
mm    max  mean
 506   577   544
1068  1248  1165
 4.2   4.7  4.32
 4.1  4.63  4.49
 543   990   808
                          1000
   2000
 2500
              1500
              SO?, ppm
Figure 26.   Influence of S02 on NO oxidation
3000

-------
                 hv(2400-2800 A)-^20n> 1 •** 203    (30)
                 hv(2400-3600)  — «-OH- + NO         (31)
                 hv(1900-4000)  -*-20H-             (32)
     The presence of these absorbing species cannot be dismissed on  the
basis of the comparative spectra discussed earlier.  A check of the  Heath
Spectrometer using Equation 12 indicates that the minimum detectable
concentrations are quite high.  For this analysis, minimum detectable
was arbitrary (MDC) defined as the concentration which would result  in
2% transmittance change at a specific wavelength.  The results indicates
that the MDC's for 03 is 17 ppm, for HN02 is 145 ppm and for H2<32 is
740 ppm.

           Ln(0.98)    =    C        (Equation  12)
           a2660  L

It is  conceivable that one or  more of the above  species  does  exist at
steady state concentrations  below those detectable, and  that  these pho-
tolyze  to produce free  radicals  which support NO oxidation.
     In summary,  the  analyses  of  possible causes of the  middle UV en-
hancement did not conclusively indicate what species were responsible.
It is  clear, however,  that S02 does not participate in any primary pho-
tochemical  processes  which significantly effect  the overall  system.

HYDROCARBON REQUIREMENTS
     The energy requirements and  the hydrogen requirements are considered
two major components  of  the photochemical system operating costs,  A
large test matrix was performed to determine the extent  of NO oxidation
over wide ranges of both variables.  These tests were done using propy-
lene.   The relative effectiveness of ethyl ene and propylene was tested
in a separate set of tests.
Propylene Tests
     The efficiency of NO  oxidation at 5 levels of hydrocarbon
concentration and 5 levels of UV light intensity were tested in two sets
                                79

-------
 of matrices.   One  set was  done  at  an  approximate inlet NOX concentration
 of 600 ppm which  is  representative of many coal-fired boilers without
 combustion modifications for  NOX suppression.  The second set of tests
 was  done  at an  inlet  NOV concentration of approximately 1200 ppm which
                       A
 is  an  upper limit  concentration probably representative of cyclone
 fired  boilers  or units firing coal of very high fuel nitrogen content.
 The  entire test series was repeated.
     Propylene was a  convenient hydrocarbon to use since the reactions
 are  fairly well understood and since  it does not degrade into light
 scattering aerosol which could deposit on the optical surfaces within
 the  CSTR.   Propylene  is also an attractive candidate for commerical
 systems,  because it is a readily available bulk chemical of reasonable
 cost.
     Results are presented in a series of 4 figures (27 to 30).   The
 first  and  the third are for the low NOX tests and the second and the
 fourth  are for the high NOX tests.   In each case, the observed NO is
 plotted with respect  to the prevailing k-,  level and the inlet propylene
 concentration.  Isopleths have been drawn to indicate conditions of
 equal  NO  oxidation.
     There are several outlying  points in  each of the four figures.  It
 is believed that this is  due to  errors inherent in  the k,  measurement and
 in the propylene measurement.   Light deposits on  the inside window  of the
CSTR could reduce k1  values by as much as  50% based on N02 photolysis
tests before and after reactor cleaning.   There were also  random fluctua-
tions in the lamp current caused by variation of  lamp operating  tempera-
ture.  After the first 40 runs,  a  series of electrical  problems  caused
by construction in adjacent labs resulted  in  the  premature failure  of
the xenon-mercury arc.  It  was replaced  by a  supposedly identical unit;
however, operating characteristics  were  different.   Due to a higher oper-
ating temperature, it was necessary to reduce the lamp current so that
volatilization of the nylon insulator around  the  lamp electrodes would
not significantly affect  the light  intensity  levels.
     Gradual deterioration  of the capillary used  to inject the gas  sam-
 ple to the Varian Gas Chromatograph resulted  in some error in the propy-
 lene measurement.  Calculated propylene concentration based on the  cali-

                                80

-------
CO
         re
         a:
            1.5
            1.0
          o
          o
         £ 0.5
          CM
         O
                   CO
                   o
                   o
• 37
          \
                                               o
                                                o
         • 342
\
                             • 480
                     • 505
                Runs  19-43

                Numbers next to data  points
                indicate APPM at specified
                experimental  conditions
                                                                                    400
                  •23  .155.0,
                       200   400   600   800   1000 1200  1400  1600  1800   2000   2200  2400
                                                [C3H ], PPM
          Figure 27.   NO oxidation, influence of initial  propylene concentration  and light intensity,
                       low initial NO  concentration (-v600 ppm (v/v))
                                     A

-------
             1.5
00
PO
        10
        cc
        in
o
4->
O
.c
O_

 CNJ
O
            i.o  L
            0.5  I
                                                 Numbers  next  to  data  points
                                                 indicate  ANO  at  specified
                                                        *532      conditions
                                              • 153
                                                 • 103
                 Figure 28.
                                ~80U~
                                   [C3Hg] ppm (v/v)
                     NO oxidation, influence of initial  propylene concentration
                     and light intensity, high initial  NOX concentration  (  1200 ppm  v/v)

-------
           1.5  -
00
u>
ce

42  1.0
M
        O
        a.
         CM
        o
           0.5
Runs 69-93
Numbers next to data points
[ndicate Appm NO at speci-~
     ied experimental condi
         tions.
       1600  1800  2000
                                                  I200  1400
                                                 ppm, v/v
                Figure 29.   NO oxidation,  influence  of initial  propylene concentration
                            and light intensity,  low initial  NOX  concentration ( 600 ppm v/v)

-------
          1.5
co
       c
       •r-
       s:

        *
       CU
       4J
       CO
       CU
       o
       •M
       o
        CM
       o
                                                                             594
                                                                              •
                                          Runs 94-118
                                          Numbers next to data  points
                                          indicate Appro'  NO  at  specified
                                           xperimental conditions.
1.0  -
          0.5  ~
                                                                             •477
                         400    600
                            800   1000   1200   1400
                               [C3H6J  ppm  v/v
600  1800  2000
              Figure 30, NO oxidation, influence of initial  propylene  concentration
                         and light intensity,  high initial  NOY  concentration  (  1200 ppm v/v)
                                                             A

-------
brated rotameter values indicated up to a 25% error due to these problems.
     The low N0x inlet ^-olefin plots have three distinct regions,  (1)
an area of apparent UV light insensitivity below 500 ppm (v/v)  of olefin,
(2) an area of olefin insensitivity at UV intensities below 0.5 k, ,  and
(3) a central region where both variables exert an influence.
     The lignt insensitive region is obviously unattractive due to the
high cost of electrical power.   Selection of an optimum operating point
in the remaining two regions depends on the necessary control  efficiency
and the relative costs of propylene and energy.  To achieve an  outlet NO
concentration of 100 ppm it appears necessary to use propylene  concentra-
tions greater than 500 ppm with k-j values of approximately 1 to 2.
     A comparison of the  high  and low NO  test series shows the efficien-
                                         /\
cy of NO conversion appears to  be 10 to 25% better in the  low NOX tests.
Because of this unexpected result, an additional  series of tests were
performed to study the relationship between inlet NO  concentration  and
                                                    J\
the efficiency of NO oxidation.  These runs are described  in a  later sec-
tion.
     In the high NO  series, there is a light insensitive  region; however,
                   /\
there does not appear to be a propylene insensitive region.  At the  same
stoichiometric NO/C3Hg ratio; however, the UV light requirements are
close to those for the low NO  test program.
                             /\
Comparison of Propylene and Ethylene Effectiveness
     One of the principal  design requirements is the selection  of the
most effective hydrocarbon on a cost per unit NO converted basis. A
screening of possible hydrocarbons is beyond the scope of this  project;
nevertheless, tests using ethylene were considered useful  to evaluate
the differences in performance.
     A set of eight runs were  performed,  four with propylene and four
with ethylene.  The extent of  NO conversion  is presented  in Figure 31
along with a summary of variability  in the  run conditions.  There is a
clear difference between the two  compounds  on  a  Appm NO basis.   This
could be due to  the reactivity of  the alkyl  radical  in  the propylene
reactions or due simply to a difference  in  the reaction rate constants
                                 85

-------
        800
        600
00
CTl
    Q.
    Q-
     X
     o
400
        200
                                                1
                                                    JL
                       600
                          800
  1000        1200         1400
Olefin Concentration, ppm  (v/v)
600
1800
                                  Figure 31.   Comparison  of propylene and ethylene

-------
of the two compounds with hydroxy radicals.   The convergence  of  the  two
curves of Figure 31  supports the reaction rate explanation.   At  the  high
olefin concentrations, there should be more  effective competition  for  the
reactive species.
Utureactod Hydrocarbon Emissions
     The environmental acceptability of the  photochemical  process  is
partially dependent on the fraction of injected olefin which  does  not
react and is emitted from the control system.   The downstream flue gas
desulfurization system would not be effective in removing these  hydro-
carbons.
     The fraction of propylene consumed at various light intensity levels
and various inlet propylene concentrations was evaluated using the data
of runs 74 to 93.  The purpose was to determine if there was  an  optimal
operating range in which the emission of unreacted hydrocarbon is  not  a
problem.  The data are displayed in Figures  32a-d, each of which presents
results at a distinct k-j level.  The top line of these figures is  the
inlet propylene concentration and the bottom line is the reactor outlet
concentration.  Unreacted olefin is represented as the shaded portion.
The run number adjacent to each data point corresponds to the run
summaries of Appendix B.
     Unreacted propylene emissions are a significant problem  across  the
range of propylene inlet concentrations used in these runs.   Even  at
very low concentrations, a major fraction of the olefin remains  uncon-
sumed.  Light intensity does have a favorable impact as demonstrated by
the shrinking unreacted hydrocarbon concentrations evident by comparing
Figures 32a-d.  Based on these results,  it appears that control  of  the
hydrocarbon/NO  stoichiometric ratio is  not effective  in preventing the
              J\
emission of unreacted hydrocarbon.  There are only three options  remain-
ing:   (1) increase light  intensity,  (2)  alter propylene reactivity, and
(3) select more reactive  hydrocarbon.  The first option can  be  evaluated
using  the same  data  set  presented  in  Figure  32.
     The  percent  propylene  consumption  derived  from  the data of Figure 32
has been  combined with  the  observed percent  NO  oxidation  for the  same runs
to derive a  second  set  of four Figure  33a-d.   The  purpose of these  is to
determine operating  point of the system with respect to  the  NO^ peak  (re-
                                 87

-------
    2000
    1500
    1000
            average k, - 0.25 Mi
                     Inlet
 h
  ,
r^   500
                                     2000
                                     1500-
                                         100C -
                                          50C
                                                                          83
                                            average k]  =  0.48  Min~y/ s*
            500   1000   1500    2000
         Initial  C$\6  ,  ppm v/v
                Figure 31a
                                                   500     1000   1500    2000
                                                 Initia'i   03^ ,  ppm  v/v
                                                        Figure 31b
                                      88
>
>
 ID
    2000
1500
    1000
     500
            average k   : 0.66 Min
                                 -1
                          87.
                  Tnlet
                                       •2000
                                    1500
                                        1000
                                         500
                              I
                                  I
average k,
                                                        1 .57  Min
          500     1000   1500   2000
         Initial  C3H6  , ppm v/v
                Figure 31c
                                                   500    1000   1500    2000
                                                 Initial   C3Hg ,  ppm v/v
                                                        Figure 31d
         Figure 32.     Unreacted propylene concentration at various inlet
                       NOX levels and light intensities


-------
1 UU
o
1 75
C
o
^ 50
0~>
°a
I "
+->
ro
T3
X
o
o
s* Figu
— _^l UU
o..
E
average [C^] = 316 g 75
m
" 50
C
0
| 25

— .
— average [C^H^] = 640
/i ill
0.5 1 .0 1.5 2.0 0.5 1 .u ] _5 2.
re 33a. N0? photolysis Figure 33b. N0? photolysis
                 rate,
                                                     rate, mirr'
o

4->
CL
c
O
 X
o
    100  r-
     75
     50  -
25  -
          0.5     1.0   1.5
Figure 33c.  N09 photolysis
                                                                        = 2026
                                                            C3H6
                                                                           I
                                      2.0
                                               0.5     1.0    1.5    2.0
                                         Figure 33d.  NCL photolysis
                                                        C-       1
                                                      rate, .inn~ '
     Figure  33.  NO oxidation and propylene consumption as a function of
                 NO,, photolysis rate.

                                  89

-------
 fer to Figure 11),  and  to  extrapolate  the  possible light energy require-
 ments  to  achieve  low  olefin  emissions.  On these graphs, the point of
 NO-NOo crossover  is shown  as a  horizontal  line at the 5Q% conversion le-
 vel  and the  point of  the NOp peak  is arbitrary shown as a horizontal
 line at the  95% conversion level.
     At the  low propylene  concentration test series, Figure 33a, it is
 apparent  that the reaction system  does not begin to even approach the
 NO-NO/, crossover  point.  The extrapolated  light requirements approach
 positive  infinity since tne operating conditions are within the nght
 insensitive region described in the previous section.  As the propylene
 concentration is  increased stepwise in the set of four figures, it is ap-
 parent  that the operating point approaches and then goes beyond the point
 of the  N02 peak.  The extrapolated light energy requirements gradually
 decrease  to a level  of k1  =1.5 mlrf^this can be converted to  station
 power  retirements using Figure 9).  In the case of propylene^control  of
 unreacted emissions  using increased light energy requires undesirably
 high energy levels.
     Assuming that the energy costs were tolerable, use of increased
 light energy to control  hydrocarbon emissions  entails a  second  risk.  As
 illustrated in Figure 11 of Chapter 4,  secondary reaction products form
 readily when ambient-type photochemical systems pass  the N02 peak.  Some
 of the compounds which could  be expected include peroxyacylnitrate (PAN),
ozone  (Oo)» nitric acid  (HNO.J  and hydrogen peroxide  (H^Oo).   Fine par-
 ticulate composed of sulfates,  nitrates,  and organic  particulate could
also form.  All  of these reaction products would strain the control  capa-
bility of the scrubbers  used  the downstream flue gas  desulfurization sys-
 tems.  The fine particulate would be especially difficult to remove.   In
the case of propylene, the  data presented  in Figures  33a-d.  demonstrates
 that control  of unreacted  emissions using  light energy is done  at the
 risk of secondary pollutant generation, assuming that reaction  schemes
 at stack concentrations  generally resemble those at ambient concentra-
 tions.
     Solutions more  fundamental than just control  of the C,H,/NOv stoi-
                                                          •JO   A
 chiometric ratio or  the increased use of light appear necessary to pre-
 vent emissions of unreacted hydrocarbons.   The alteration of effluent
                                 90

-------
conditions to improve the propylene competition for reactive species
such as 63 and OH will be analyzed in a subsequent section.   Generally,
the reaction rates with OH radicals, ozone or atomic oxygen  increase
with the molecular weight of the hydrocarbon; however the potential  for
aerosol formation also increases.  Presumably, a compromise  choice will
be required.

VARIABLES  INFLUENCING NO OXIDATION

     The effect  of various  physical  and  chemical  variables are analyzed
to determine the operating  limitations  of  the photochemical  system and
to evaluate the  extent to which propylene  reactions  can  be optimized.
Possibly the most important variable is  the oxygen  concentration  since
this is essential to the  free radical chain reactions  and since NO  sup-
                                                                  J\
pression techniques can lead to reduced flue gas  oxygen  content.   Such
modifications can also lead to increased CO levels;  therefore, this is
examined over the concentration range anticipated.   Due  to the unexpected
difference between the high and low NO   test series discussed in  a pre-
                                     A
vious section, a set of tests have been devoted to  this  subject.   Other
variables considered are effluent gas temperature and C02 concentration.
All of the experiments described in this section  were done at reaction
conditions resembling as  close as possible those used in the large test
matrices of runs 19 to 118.  Each test  program is discussed  with  respect
to the influence of the NO conversion efficiency and with respect to  the
possible underlying chemical reasons for the observed effect.

Influence of Oxygen Content
     The principal questions of interest are:  (1) whether the  reduced
oxygen content of NO  suppressed boilers interferes with photochemical
oxidation, and  (2) whether the  increase of oxygen would reduce the unre-
acted  propylene  concentration.  These were evaluated  in a set of 10 runs
 (numbers 127 to  136),  five of which were done at high NO  levels and five
                                                        A
of which were done  at moderate  NO   levels.   The  oxygen  content of the
                                 /\
 reactor  feed  stream  was  modified by adjusting  the  balance between the

                                91

-------
    1200
    1000
    800
0.
Q.
    600
Variable
N0v0
X
C^H,
3 6
so2
kl
Temp
T
Min
1020

569

1252
4.5
256
0.738
Max
1101

702

1347
4.5
295
0.784
Mean
1080

636

1304
4.5
271
0.763
    400
    200
Variable
NV
C3H6
so2
k!
Temp.
-T
Min
515
569
1260
4.5
261
0,738
Max
604
695
1465
4.5
295
0.772
Mean
563
621
1348
4.5
272
0.7R1
                        % Op  (volume)
                                                         8
                Figure  34.   Effect of oxygen content on NO oxidation

                                   92

-------
dry nitrogen and dry air streams  in the blending  box.  The  prevailing
oxygen concentrations were calculated based  on  the  rotameter values as
calibrated at the operating pressure of the  particular run.  Other varia-
bles were held as constant as possible in the set of runs.
     The results are presented in Figure 34.   For each set  of  five, the
average inlet NO  concentration is shown as  a horizontal  line.  The typi-
                A
cal oxygen contents for high NOX  emitting boilers are shown as a range of
3 to 5% by volume.  The corresponding values  for  more typical  boilers
with and without NOX suppression  capability  are shown as  a  range of "\h
to 4%.  Those units with low excess air capability  would  be in the low
end of the range and unmodified boilers would be  close to the  3 to 4%
range.  With respect to the lower curve, it  is  apparent that below an
oxygen concentration of 2% there  is rather severe oxygen  starvation of
the free radical cycles.  However, within the region typical of most
installations, this loss in NO oxidation capability is limited to 5 to
15%.  This is a very important result in that it  appears  that  a photo-
chemical system is compatible with low excess air combustion modifications,
It is desirable, but not necessary, to add oxygen through flue gas
dilution.
     At high inlet N0tf levels, the extent of oxygen starvation is severe
                     ]\
throughout the range studied.  NO oxidation  effectiveness is impaired to
the extent of 10 to 90% with the  latter value occurring at  extremely low
oxygen content.  For such applications, it may be advantageous to either
reduce inlet NO  concentrations and/or dilute the effluent  stream prior
               4\
to the photochemical reactor.
     The influence pf oxygen content on the utilization of  propylene  is
unclear.  The data presented in Table 9 indicate   that  the  untreated  ole-
fin concentration varies from 67 to 95% of the inlet concentration  over
the oxygen range of 0 to 7%.  High oxygen content such as in  run  128  did
not significantly  improve the propylene emissions problem.   In fact,  the
extent of  propylene conversion at  the  k, level of 2.84 min'1  appears  to
be slightly  lower  than  comparable  values discussed  in the  previous sec-
tion  (refer  to  Figures  32a-d).   Conversely,  low  oxygen content did not
drive propylene emissions  any  higher  than would  be  expected for the quan-
tity  of NO oxidized.

                                 93

-------
TABLE 9.  INFLUENCE OF OXYGEN  CONTENT  ON  PROPYLENE CONSUMPTION

              Oxygen                  Unreacted C-H,
              Content         C~Hfi        Emissions       ANO
              (X by Vol.)      (fern)     (% of  [C3H6]0)    (ppm)

     High       0              102            82          126
     [NOjo      0.86           108            83          404
     Tests      2.54           234            67          626
                3.88           N/D            N/D         781
                6.14           210            70          990

     Low        0              26            95          120
     [NO ]0      0.81            117            79          392
     Tests      2.45           190            67          558
                4.43           157            78          493
                7.42  	N/D	N/D	559
                          94

-------
     The NO oxidation in the absence of 02  (runs  133 and 134) is inter-
esting because it indicates  the extent  of conversion possible in the ab-
sence of the free radical  chain reaction mechanisms and other mechanisms
dependent on oxygen.   One possible  reaction mechanism  is the oxidation of
NO by atomic oxygen as shown in reactions 33  to 38,

         N02   + hv(3000-4000 A) •>  NO + 0(3P)               (33)
         0(3P) + NO             %  N02                      (34)
         0(3P) + C3H6           +  C"H2CH =  CH2  +  OH         (35)
                                +  -C3H60-                  (36)
         C"3H60 + NO             +  C3H6ONO                  (37)
         0(3P) + N02            +  NO + 02                  (38)
     The N02 product quantum yield  for  such a mechanism must be  small  due
to the rate of the reverse reaction, 38.
Influence of Carbon Monoxide Concentration
     The typical levels of carbon  monoxide  in the effluent of coal-fired
boilers  is 10 to 50 ppm.  With certain  combustion modifications  for NOX
suppression, however, increased CO levels can result.   The maximum expect-
ed concentration is 2000 ppm since above this point there  is probably  un-
acceptable carbon carryover and there is a  risk of explosive conditions
(Acurex, 1978).  A set of four runs  (numbers  137 to 140)  were  performed
to examine the  possible effect of CO within the 0 to 3000  ppm  range.   The
purposes of this work was to determine the extent to which CO  enhances
the  NO conversion  efficiency and to  determine if unreacted hydrocarbons
are  reduced at  high CO  levels.
      CO  was  injected  into the  influent  gas stream using the S02 line, and
the  prevailing  CO  concentration was  calculated based on the rotameter cal-
ibration.   Except  for the absences  of  SCL, the effluent conditions were
 similar  to those used in  the large  test series  (runs  19 to 118).  The
variability of reaction conditions  is  presented  in the table included
with Figure 35.  This figure illustrates the NO  oxidation effectiveness
 in terms of ppm NO converted at the corresponding  CO  inlet concentration.
 The run numbers shown adjacent to  each of  the  data  points corresponds to
 the data compilation in Appendix   B.

                                 95

-------
    800
    600
Cu
a.
    400
    200
Variable
[NOX]0
[C3H]
Cso2]
-Co2]%
kl
Temp.
T
Min
842
1137
0
4.5
1.55
266
0.764
Max
896
1180
0
4.7
1.95
269
0.765
Mean
865
1152
0
4.6
1.65
267
0. 767
                      1000
                      2000

                    [CO], PPM +

Figure 35.   Effect of CO on  NO  oxidation
"OTT
                                96

-------
      The results  displayed in Figure 35  show that CO enhances  the oxida-
 tion of NO.   There is  a  22% increase in  the ppm of NO converted  at 3000
 ppm CO relative to 0 ppm CO.   This  effect  is attributed  to  the reaction
 of CO with hydroxy radicals which ultimately results in  the hydroperoxy
 radical  attack of NO.
                       CO + OH-  -> C0? +  H.                  (39)
                       02 + H.  SHOg                      (40)
                       NO + H02-+ N02 +  OH-                (41)
      The effectiveness of the chain process is limited by the  competition
for the hydroxy radicals.  This competition  primarily comes from the  reac-
tions listed below:
                                                            (42)
                                                            (43)
                                                            (44)
      The fraction of hydroxy radicals retained in the CO chain reaction
can be approximated using the ratio  of Equation 13.  The  summation term
in the denominator accounts for OH sinks  not expressly considered in the
above set of reactions.
OH- +C3H6+
OH- -i- N00 +
2
OH- + NO ->
CH3CHOH=CH2
HNOo
3
HN02
c   __     _    _
•"OH   k[CO] + k[C3H6] + k[N02] + k[NO] +  " k^]
                                         1=1
where:  Fn,, = the fraction of hydroxy radicals reacting according to
         un   reaction  39
        k.  = the rate constant for the reaction number i
Assuming that the summation term is neglible, and using the reaction  39
rate  constant of 2.50 x 102 ppm"1 min"1 and constants of Demerjian,   Kerr
and Calvert  (1975)  for  the other hydroxy reactions,  indicates  that  CO uti-
lizes 5% of  the hydroxy radicals at 3000 ppm.  This  small  fraction  appar-
ently accounts  for  the  22% increased  in NO  oxidation.   These  results  con-
firm  previous conclusions that  hydroxy radical utilization is  an impor-
tant  factor  in  the  overall reaction scheme  active  in the  photochemical
process.   CO has  a  moderate  influence on  NO conversion due to its OH
 reaction  and resulting free  radical  chain  reactions.  This positive ef-
                                  97

-------
 feet again demonstrates that the photochemical  system  is  compatible with
 conditions expected at boilers  modified to suppress  NOX generation.
      The high concentrations of NO and  N02 in  addition to the  large rate
 constants for hydroxy  radical reaction  may hinder  the optimal  utilization
 of hydroxy radicals.   In  time-limited and  light-limited conditions, both
 reactions 43 and  44 may represent free  radical  termination steps.  The
 adverse  impact of high NOx  levels on system performance has already been
 discussed  earlier concerning the large test series  and is evaluated
 specifically in the next  section.

 Influence  of  Initial NO  Concentration
                       /\
     During the course of other experimental subsets, it was  observed that
 there was  a moderate negative correlation between the extent of NO conver-
 sion and the  initial NO concentration.   This was studied in a special  ser-
 ies of runs with  initial concentrations varying from 400 to 2000 ppm.   Ob-
 viously, the  high  levels are not representative of any commonly encoun-
 tered fossil-fuel  fired boiler.   Unlike most of the experimental work,
 this set of tests  has been done specifically to evaluate mechanistic  ques-
 tions and not to evaluate technical/economic feasibility of the technique.
     A series of 7 runs (numbers 141 to  147) were performed under nearly
 identical conditions.  The influence of  initial  NOX concentration was  in-
dicated by the difference  in ppm NO converted.   Results  are presented  in
Figure 36.  The probable reasons for the negative slope are the reactions
1 isted below:
             OH-  + N02         +HN03                      (43)
             OH-  + NO           2HNO-                      (44a,b)
             OH-   + HN02        +N02-  + H30               (45)
             OH   + HN03        +N03  + H20               (46)
             HN02 + hv(<4000 A)  +NO   + OH-               (47)
             HN03 + hv(<3400 A)  +N02  + OH-               (48)
Reactions 43 and  44 should be important sinks for the hydroxy radicals.
With the time and light limited  conditions, it is unlikely that the pho-
tolysis of nitrous acid and/or nitric  acid would significantly offset the
overall dampening effect of increased  NO .

                                  98

-------
 1000  -
  800
      /

   /
Hi  I
  600
Q.
a.
O
z
<3
  400
   200
Min
904
1582
4.5
1 .55
261
0.758
Mean
959
1709
4.55
1.55
269
0.764
Max
1183
2086
4.7
1.55
272
0.771
            200   400  600    800   1000   1200  1400   1600  1800  2000  2200

                                    [NOXL,  PPM


                   Figure 36.   Effect of  initial NOX concentration
                               on  NO oxidation
                                  99

-------
      It should be noted  that  this  is  not a problem for the photochemical
 system since  both nitrous  acid  and nitric acid should be effectively
 removed downstream in  the  flue  gas desulfurization system.  These results
 could  suggest that the difference between the high concentration runs
 and  the low concentration  runs  is partially due to the increased presence
 of nitrous acid and  nitric acid  in the high concentration tests.
 Influence of Carbon Dioxjde
     The simulated flue gas streams used in  this  project  did  not include
C0? since photolysis was negligible in the  lamp emission  spectra and since
C0? does not actively participate in known  dark phase reactions.  Sus-
pected chemical activity is limited to enhancement of the atomic oxygen-
nitric oxide reaction  (Demerjian et al., 1975).  This relatively minor
third body activity did not warrant the additional complexity involved
with adding another component to the reactor influent stream.  This  as-
sumption was tested in a set of four runs  (numbers 148 to 151)  done  at
C02 concentrations ranging from 0 to 20%.  Typical C02 levels are in the
range of 10 to 15%.
     The influence of C02 was measured with  respect to the change in NO
concentration at various CO- levels.   The  results of the  four tests  are
presented in Figure  37.  Variability of experimental  conditions is re-
ported  in the table enclosed within this Figure.
     The observed NO concentration drop varied 3% for the tests and
there is no apparent trend.  It is concluded that CO,, does not  exert
any appreciable influence on the overall photochemical  process  and that
the system can in fact be operated successfully in the high C02 environ-
ment of coal-firea boiler effluent.
 Influence of Gas Temperature
     There are three possible locations where hydrocarbons could be  in-
jected and the irradiation performed:  following the FGD  system presatu-
 rator,  preceding  the presaturator, and preceding the air  preheater.   A
 series  of tests were done to evaluate system performance  at the various
 temperature levels characteristic of these possible locations.  Due to
 physical limitations of the reactor, it was  impossible to approach the
 500-700°F range representative of the air preheater inlet.   In  these runs,
                                100

-------
  500  i i_.
  400
   300
OJ
TJ
*¥—
X
O
   200
   100
Variable
N0x
C3H6
so2
ki
Temp.
  Min
  846
  855
   0
 1 .55
  261
0.768
  Max
  898
 1018
   0
 1.55
  264
0.771
 Mean
  866
  960
   0
 1.55
  262
0.770
                    R^
                                                     15
                                       20
                                        % CO,
             Figure  37.  Effect of CC>2 concentration on NO oxidation
                             101

-------
the gas  temperature was varied from a low of 115 F to a high of almost
300 F.   The latter value is characteristic of older boilers with minimal
air preheater surface areas.  Newer boiler designs have maximized thermal
efficiency by reducing air preheater exit temperature from 300 to  275 F.
This is  possible due to the common use of low sulfur western coals with
low acid dewpoint.  After the presaturator, which sometimes accompanies a
flue gas desulfurization system, the gas temperature is reduced to
adiabatic saturation levels of approximately 125 F.   These two common
temperature ranges are illustrated on Figure 38.   Effluent temperature
between these two regions is uncommon.  Data points  have been taken in
this temperature range only to obtain a clearer picture of the relation-
ship between the extent of NO oxidation and the gas  temperature.
     The test series was divided in high N0tf and low NO  subsets.  The
                                           A           X
flow rate of the blended gas stream to the reactor was adjusted to ac-
count for the density changes occurring upon heating in the tube furnace
and CSTR.  The purpose of this was to ensure that the energy per gram
mole of  gas was approximately equal for the eight tests.  The runs (num-
bers 152-159) were performed in random order.
     The results are presented in terms of percent NO oxidation.  This
was chosen instead of change in NO concentration since there was consi-
derable unintentional  variation 1n the Initial  NO  concentrations as indi-
                                                 A
cated in the tables provided with Figure 38.   Conversion of NO at high
temperatures is obviously favored.  There is a considerable loss of sys-
tem effectiveness as the temperature if reduced; therefore, the preferred
location for the system is prior to the presaturator.  It is conceivable
that performance could be improved with the reactor on the "hot side"  of
the air preheater if the observed trend continues.  This should be exa-
mined in subsequent bench scale tests.

REACTION PRODUCTS
     The quantities and characteristics of the photochemical system reac-
tion products will partially determine the commercial acceptability of
the technique.  Ideally, the system would operate at the point of the NOp
peak and there would be a minimum of secondary reaction products such as
ozone, PAN, and aerosols.  The formation of submicron aerosols would be
                               102

-------
             Runs  152  to  159
   1000
   800
Q.


 *

O
C
o
X
o
   600
Variable
[NOX].
r H
L3H6
so2
°2
T2
Min
1190

541
1232
3.8
0.742
Mean
1294

677
1396
4.3
0.753
Max
1543

849
1637
5.0
0.772
   400
    200
Variable
[N0x],
CoHc
3 6
so2
°2
kl
Min
547
541

1284
3.8
0.8
Mean
579
573

1380
4.0
1 .15
Max
651
625

146E
4.4
1.5!
                         I
I
I
                  120   140   160   180   200   220   240   260   280

                            Gas Temperature,  F

              Figure 38.   Influence of temperature on NO oxidation
                                 103

-------
particularly unfortunate since the collection of most  scrubbers used  in
FGD systems are at a minimum in the 0.2 to 0.7 microns.  A  net  increase
in the levels of ROM's and other potentially toxic  organic  compounds
would also reduce the appeal  of a  photochemical  approach.
      The generation of secondary  reacuon products  has been evaluated by
 performing nitrogen, sulfur, and  carbon material balance calculations a-
 round the quartz CSTR and determining  the fraction  unaccountable.  This
 represented the maximum quantity  of material  which  could be in the form
 of toxic secondary products.  Millipore filter  samples were analyzed with
 x-ray fluorescence and ion chromotoaraphy to  evaluate the  fraction of
 missing sulfur and nitrogen  which were in the form  of sultate and nitrate
 aerosols.  The sulfate and nitrate content  of wash  waters of the reactor
 and effluent gas lines was also added  to  complete the material balances
 of sulfur and nitrogen.
      For selected  runs, samples of the reactor  effluent were taken in 40-
 liter Teflon bags  and were analyzed at the  UNC  Pittsboro research site
 where it was possible to convert  the formaldehyde to CO for analysis in
 the gas chromotography.  For the  same  set of  runs,  gas samples were taken
 in Tenax columns for GC-mass spectrometry.  The  purpose of these runs was
 to identify possible toxic organic products present at low concentrations,
 Nitrogen Balances
      The fraction  of the nitrogen  compounds measurable with the chemilumi-
 nescent analyzer are reported  in  Table 10.  Low  light and/or low propy-
 lene conditions  provide only limited photochemical   activity; therefore,
most of the initial  nitrogen can  be accounted for as NO or measurable
 NO  compounds.   As  reactivity  increases,  it is apparent that the portion
   A
 of unaccountable nitrogen  increases to 10 to  20% of the initial NO  con-
                                                                  A
 centration.  One possible  reason  for this is  that the system has the
 capability to  proceed beyond the  peak  where some of the NO  has been con-
                                                          A
 verted to gaseous  or aerosol  forms  which  cannot be measured by the TECO
 instrument.  This  unaccountable nitrogen  amounts to a maximum of 120 ppm.
      To determine  the fraction of the  nitrogen  present as aerosol, some
 of the Millipore filters (see  Chapter  4,  Figure  22} were analyzed using
 ion chromotography.  The oxygen test series  (runs 127 to 136) which

                                104

-------
    TABLE lOa.   NITROGEN MATERIAL BALANCE, [NO ]0 ^600 PPM
Relative
Light Level
0
1/4
1/2
3/4
1
0 250
Hydrocarbon
1 .00
1.00
1.00
0.99
1.00
1.00
1 .06
0.96
0.92
0.94
500
Target
PPM (v
1.00
0.95
0.97
0.92
0.87
1000 1500
Concentration,
/v)
1.00
0.94
0.93
0.98
0.82
1.00
0.97
0.93
0.86
0.87
Runs
69 to 73
74 to 78
79 to 83
84 to 88
89 to 93
TABLE lOb.  NITROGEN MATERIAL BALANCES, [NOV]0 =1200 PPM (v/v)
                                          A
Relative
Liqht Level
0
1/4
1/2
3/4
1
0 250
Hydrocarbon
1.00
1.00
1.02
1.00
1.01
1 .00
0.99
0.98
0.95
0.99
500
Target
PPM (v
1 .00
1.00
0.95
0.97
0.97
1000 1500
Concentration,
/v)
1.00
0.97
0.95
0.96
0.90
1 .00
0.99
0.94
0.95
0.90
Runs
94 to 98
99 to 103
104 to 108
109 to 113
114 to 118
                                105

-------
exhibited substantial variability In NO conversion were used to  identify
the possible range in quantities of nitrate aerosol.   It should  be  noted
that the gas temperature at the filter point had generally cooled to
ambient levels; therefore, it is likely that most of  the nitrate had con-
densed prior to collection.  Observed nitrate levels  expressed in terms
of ppm of NO are in the range of 1 to 5, trivial when compared with the
total NO unaccountable.  The variability of the nitrate concentrations
in Table 11 does not appear correlated with the extent of NO conversion.
     At regular intervals, the reactor was flushed with distilled water
to measure quantities of deposited material and to prevent major changes
in the light transmission properties of the CSTR.  Deposits were highly
water soluble.  Wet chemical analyses of the samples  indicated very small
quantities of nitrates as indicated on Figure 39.   Considering that ap-
proximately 30 runs were performed between each washing, the quantities
of nitrate represent less than 1% of the unaccountable fraction. Analy-
sis of the line rinse water at the end of the experimental  program  also
indicated that very little of the nitrates materials  were deposited.
     It is concluded that most of the unaccountable nitrogen is  present
as gaseous material  which is not adequately reduced in the converter of
the chemiluminescent analyzer.  Possible compounds include, but  are not
limited to, nitric acid, nitrous acid, and organic nitrates.   Since the
chemical  composition is uncertain and the concentrations may exceed 100
ppm,  it must be concluded that some nitrogen compounds may represent
toxic reaction products.
Sulfur Balances
     Very little change in $02 concentrations was observed  during the
change from dark phase to light phase conditions in each of the  runs.
As indicated in Table 12, the maximum drop was 5% and the normal drop
was in the range of 1 to 3%.  These relatively small  fractions represent
the maximum amounts of potentially toxic sulfur compounds formed in the
reactor.   These quantities could be even smaller than suggested  by  the
material  balance since NO,, formed during irradiation  suppresses  the S02
signal in the pulse fluorescent analyzer (refer to Chapter 4).  Correct-
ing the light phase S02 concentrations according to the steady state N02

                                106

-------
  Line Deposits (4-28-78)
   Gram Moles N00"
Line 1
  Sample 1
  Sample 2

Line 2
  Sample 1
  Sample 2

Line 3
  Sample
0
0


5.8x10
1.1x10
-5
-5
                                   Quartz Reactor
                                               L-Line 1
                                                 Teflon,  1/4" O.D
Reactor
Gram Mol
Date of Wash
3-15-78
3-23-78
4-5-78
4-14-78
4-27-78
Deposits
es N03"

2.9xlO~j?
5.0x10";?
7.0x10"^
1.4x10"°
3.0x10"°
                                 Line 3,  Teflon,  1/4
                                Figure  39.   Nitrate deposits

-------
TABLE 11.   NITRATE AEROSOL  QUANTITIES COMPARED WITH NO OXIDATION

Run
127
128
129
130
131
132
133
134
135
136
90
91
114
115
116
117
118
ANO as
N03-, ppm
2.39
3.00
2.12
1.42
2.15
1.87
1 .96
1.98
2.52
2.50
3.05
4.78
2.91
2.08
1.44
1.78
2.22
ANO
ppm
404
990
493
626
392
558
126
120
559
781
285
464
27
247
339
517
594
ANO
%
37
95
98
61
69
95
12
24
96
69
44
76
3
25
34
51
60
                TABLE 12.  SULFUR MATERIAL BALANCE
Relative
Light Level
0
1/4
1/2
3/4
1
SULFUR MATERIAL BALANCE, CORRECTED
AND UNCORRECTED
Hydrocarbon Target Concentration,
PPM (v/v)
0 250 500 1000 1500
N.D. N.D. N.D. 1.00 1.00
N.D. N.D. N.D. 1.00 1.00
1.00 0.99 0.98 0.99 1.00
1.00 1.00 1.00 1.01 1.03
0.98 0.98 0.99 0.97 0.98
1.00 1.00 1.01 1.00 1.02
0.97 0.97 0.99 0.95 1.00
0.98 0.99 1.00 1.00 1.04
0.97 0.97 0.95 0.97 0.98
0.97 1.00 0.99 1.02 1.00
Runs
69 to 73
74 to 78
79 to 83
84 to 88
89 to 93
                           108

-------
levels and the relationship  of  Figure  18 accounts for most of the observed
SO,,.  These corrected  sulfur material  balances are reported as the second
entry for each of the  runs covered  in  Table 12.  The amount of sulfur un-
accountable is very small and approaches the  inherent errors involved in
reading the S0~ analyzer analog scale.
      X-ray fluorescent tests for sulfur have been  performed  for  all  of
the  runs of the experimental program.   Results for  runs 68  to  93 are pre-
sented in Table 13 in terms of the Appro of SCU.   The very low  quantities
of  sulfate aerosol confirm  expectations of insignificant S0? conversion.
This  conclusion was checked further by performing ion  chromatography
analyses on a  selected group of the Millipore filters.   Results  present-
ed  in Appendix B  for each of the runs indicate that the sulfate levels
are conceivably 50% lower than those  suggested by x-ray fluorescence.
No  attempt was made to determine the  most accurate analytical  approach
due to the  small  quantities of sulfur in the aerosol form.
      There was an order of  magnitude  more sulfate material than nitrate
material  present  in the reactor deposits and  in the effluent handling
 lines.  The measured  quantities have  been expressed in terms of gram moles
of  sulfate in Figure 40.  These quantities are equivalent to approximately
 3%  of the sulfur  unaccountable over the combined set of experimental runs.
 It  is concluded  that  sulfate aerosol  or toxic sulfur compounds  are  not
 formed in significant quantities  in the photochemical  system.
 Carbon Material  Balances
      The extent  of propylene consumption was  a strong  function  of the
 light intensity  and the initial  propylene concentration.  With  reactive
 conditions 25 to 50% of the propylene reacted to form  one or more degra-
 dation products.  The fraction accountable  as propylene for runs 68 to
 118  is presented in Table  14.   This  set of runs  was chosen  since the
 quality of the Varian GC data was considered the best.
       Possible reaction products accounting for the missing carbon  include,
 but are not  limited to, formaldehyde, acetaldehyde, carbon dioxide, car-
 bon monoxide, and organic  aerosols.  The Varian gas chromotograph  res-
 ults (a  sample run output  is shown in Figure 41) suggests that only one
 or two  major reaction  products were  formed.  The large peak following

                                109

-------
Line Deposits
(4-28-78)
Line 1
Wash 1
Wash 2
Line 2
Mash 1
Wash 2
Line 3
Wash 1
Gram Moles
so4=
2.0x10"?
7.2X10"5
4.1xlO~a
1.0x10
l.SxlO"4
     Line 2
     SS, V O.D.
                    Line  1
                    Teflon, V  O.D.
Reactor
Date of Wash

3-15
3-23
4-5
4-14
4-27
Deposits
Gram Moles
so4 =
4x1 0"J
2.7xlO"J
3.3x10": -
4.4xlO"r
1.7xlO~3
   Line 3-
   Teflon,  V O.D.
Figure 40.  Sulfate deposits

-------
    60
    50
•§, 40
•f~
0)
2:
eo
   30
   20
   10
           1         I         i         I          I
                       Initial  Dark Phase Injections
            Propylene          Propylene
                                         Final  Dark  Phase
                                              Injection
                Light  Phase  Injection
             Propylene
                       :etalde-
                                lend Acetaldehyde
                Attenuation 32
           ATT-32    Att-2   Att-32  Att-2

               Time -*-*•

Figure *1.  Typical  gas chromatograph  output

-------
TABLE 13.  X-RAY FLUORESCENT SULFUR MEASUREMENT EXPRESSED AS  APPM  SO
Hydrocarbon Target Concentrations,
Relative PPM (v/v)
Light Level 0 250 500 1000 1500
0
1/4
1/2
3/4
1
0.80
N.D.
0.21
0.25
0.40
0.33
0.45
1.53
1.59
1.48
0.11
0.10
0.09
0.22
1,03
N.D.
0.18
0.12
0.37
0.80
N.D.
0.13
0.73
0.25
0.97
Runs
69 to 73
74 to 78
79 to 83
84 to 88
89 to 93
    TABLE 14a.   CARBON MATERIAL  BALANCES.  [NQ..]0 ^600  PPM  (v/v)
Hydrocarbon
Relative
Light Level 0 250
0
1/4
1/2
3/4
1
N/A
N/A
N/A
N/A
N//L
1
0
0
0
0
.CO
.83
.73
.68
.51
Target Concentration,
PPM (v/v)
500 1000 1500
1
0
0
0
0
.00
nc
.OD
.83
.76
.50
1
0
0
0
0
.00
.91
.85
.73
.58
1.00
0.91
0.85
0.78
0.76
C9
74
79
84
89
Runs
to
to
to
to
to
73
78
83
88
93
   TABLE 14b.   CARBON  MATERIAL  BALANCES,  [NOV]0 ^1200  PPM  (v/v)
                                            A
Hydrocarbon Target Concentration,
Relative PPM (v/v)
Light Level 0 250 500 1000 1500 Runs
0
1/4
1/2
3/4
1
N/A
N/A
N/A
N/A
N/A
N.D.
0.85
0.80
0.70
0.58
1.00
0.90
N.D.
0.76
0.56
0.99
0.89
0.86
0.76
0.60
1.00
0.93
0.89
0.82
0.65
94 to 98
99 to 103
104 to 108
109 to 113
114 to 118
                              112

-------
the propylene was  identified  as  acetaldehyde and the very small peak di-
rectly proceeding  it  was  presumed  to  be  formaldehyde.  Due to the limited
ability of the Varian to  detect  formaldehyde,  bag  samples were taken for
a limited group of runs so that  the effluent could be  analyzed at the
UNC Pittsboro research site.   The  formaldehyde was converted to carbon
monoxide and the treated  effluent  was analyzed on  a Carle  II gas chroma-
to graph.  The result of  the special  GC  tests  are  presented  in Table 15.
The total aldehyde concentration (formaldehyde and acetaldehyde)  account-
ed for 20 to  50% of the missing carbon.   The major product was  the  for-r
maldehyde which was generally present at concentrations  of 1  to  3  times
the levels of the acetaldehyde.  The observed aldehyde concentrations
displayed a  strong correlation with the extent of NO conversion  indica-
 ting  that these are  principal reaction  products in propylene degradation
 and that  the  subsequent  reaction of the aldehydes  is  incomplete.   This
 is consistent with the observations in  previous sections that aldehydes
would compete only weakly for hydroxy radicals and would not photolyze
 rapidly  under light  limited  conditions.
      It  is  possible  that the actual aldehyde  concentrations are somewhat
 higher than  indicated in Table  15.   The observed  propvlene concentrations
 were  approximately 50% lower than the corresponding values measured dur-
 ing  the  run  using the Varian GC.  Part  of  this difference could be due to
 decay by further  reaction of the  gas mixture  during transport to the
 Pittsboro site.   Elapsed times  of 2  to  4 hours occurred between the time
 of sample acquisition and analysis due  to  the 18  mile distance between
 the labs and the  difficulty in  coordinating the runs  with the transpor-
 tation.   The importance  of this delay was  evaluated by  taking repeated
 readings on a bag sample which  was not  transported.   The  rate of propy-
 lene and acetaldehyde decay are apparent  in  Figure 42.  Both compounds
 exhibit a first order rate of decay  of  5% which  is represented  by  the
 dotted line.  If this is characteristic of the other  bag  samples,  the
 aldehyde concentrations  reported in Table 15 could be 10  to 15% higher.
      While decay of  the  samples accounts for some of the  difference in
 observed propylene concentrations there remains  some difference between
 the  two  sets of  GC data.  The Varian data is considered more reliable
                                  113

-------
TABLE 15.  CARBON BALANCE CORRECTIONS  FOR  ALDEHYDES
Run
%C Unaccounted for as C^Hg
APPM C as C3Hg
Measured [C3Hg]ss vs. Actual [C3Hg]ss ,
Measured HCHO, ppmC
Measured CH3CHO, ppmC
Total RCHO, ppmC
%C Unaccounted for as RCHO
%C Unaccounted, Corrected for RCHO
102
10.9
600
% 54
250
58
308
51
5.4
107
13.6
540
46
N.D.
32
32
6
12.8
112
24.4
1035
62
325
114
439
42
14.1
115
42.1
372
100
195
60
255
69
13.2
116
44.4
840
51
186
104
290
35
29
117
40.5
1602
N.D.
315
42
357
22
32
118
35.0
1941
N.D.
348
212
610
31
24

-------
20
10
          NOTE:   Curves  represent calculated first
                 order decay rate of 0.05/nT1
                                I
I
I
 8  PM          12  PM          4 AM           8 AM          12 Noon

    Figure 42.   Aldehyde/propylene decay in  sampling bags
                          115

-------
TABLE 16.   RELATIONSHIP BETWEEN ALDEHYDE FORMATION AND NO OXIDATION
                          HCHO   CH.CHO   ANO
                    Run    ppm	ppm   ppm
                    102    250     58     156
                    107    N.D.    32     219
                    112    325    114     334
                    115    195     60     247
                    116    186    315     339
                    117    315     42     517
                    118    398    212     594
                           116

-------
since the absolute values  agree  within  10%  of  the  propylene  tank  concen-
tration and the observed gas  stream flow rates.
     Following inclusion of the  aldehydes,  the carbon  material  balances
approach 80 to 95%.   It is conceivable  that the  remaining  fraction  is
partially composed of polycyclic organic matter  and/or other potentially
toxic organic  compounds.  To  evaluate this  possibility, a  selected  series
of runs were performed with Tenax sampling  columns.  These were analyzed
using a GC mass spectrometer at  the Research Triangle  Institute.  The prin-
cipal compounds identified in the 10 test samples  are  reported  in Table
IT.  In all of these tests, acetaldehyde and acetic  acid were present in
small but nevertheless significant quantities.  The  other  compounds were
present on an irregular basis and generally in much  smaller  quantities.
      It is very unlikely that some of  the observed  compounds actually
originated in the photochemical  system.  For example,  methylene chloride
is used in the extraction step of the Tenax column treatment.  Also, there
are not halogenated precursors which could explain the presence of  hexa-
fluorobenzene and trichloroethylene.  It must be assumed that these com-
pounds and possibly some of the others  in the upper group actually  entered
by way of the dilution  air stream  (refer to Figure 22) or were contami-
nants in the GC mass  spectrometry  laboratory.
      The main conclusion  drawn  from the mass spectrometer results  is  that
a  large number of organic  compounds are formed; however, the total  concen-
tration of such compounds  measurable by the GC mass spectrometer is very
small.

COST ESTIMATES
       Photochemical  system capital  and  annualized cost estimates have been
 prepared  based  on the experimental results discussed  earlier.  The basis
 of these calculations is  the model system  presented in Chapter 4.  The
 cost estimates are indicative only of  a propylene system  as studied in
 this project.  A more reactive  hydrocarbon could  conceivably allow a more
 economical system without the problem  of unreacted  hydrocarbon emissions.
 Development of improved ultraviolet lamps and reflectors  could also  have
 a major impact on the capital and annualized  costs.  For these reasons,
 the propylene system costs are considered a maximum cost for a photo-
                                  117

-------
                  TABLE 17.  MAJOR ORGANIC COMPOUNDS DETERMINED IN GC/MS TESTS. PPM (y/y)
CO


acetoni trite
acetaldehyde
C^HgO isomer
pfopylene oxide
nitromethane
4-methyl-l ,3-dioxolane
acetic acid
acetone
methyl formate
methyl nitrate
butyl formate
ethyl ene
pro penal -Hn ethyl ether
toluene
xylene isomer
alkyl dioxolane
acetophenone
methyl ene chloride
hexafluorobenzene
m-hexane
trichloroethylene
RUN
117 115 118 117- 107

3.1 1.4 4.8 1.2 4.7
1.1 0.5 2.3 0.7
0
1.3 0.44 2.1

24.8 13.2 1.6 1.6 3.2
1.1 1.1
2.0
1.1
0.4







107 101 116

4.8 2.4
1.6


2.8 3.6 2.3
1.0





1.7
1.8


0.18
        >3000 A filter in place.

-------
chemical  system for  NO  oxidation.
Model Plant Characteristics
      The cost  estimates have been prepared for a 500-megawatt base
 loaded coal-fired boiler.  It is assumed that no combustion modifications
 have  been made  to suppress NO  formation end that the average inlet concen-
                             A
 tration  is 600  ppm  (v/v).  Cost estimates were not prepared for the 1200
 ppm  (v/v) case  since the results indicated maximum NO oxidation efficien-
 cies  of  only  50%.
       Irradiation of the effluent gas stream is accomplished in a set of
 cylindrical reactors (stainless steel) with medium pressure mercury arc
 lamps mounted on  vertical  supports along the centerline of the reactor.
 Baffles  are  included to  improve mixing of  the  gas stream.  The reactors
 are  connected in  parallel  directly following the electrostatic preci pita-
 tor  so  that  flyash  abrasion  of the optical surfaces  is minimized.  A sys-
 tem  of  propylene  distribution pipes  is located in the breeching leading
 to the  reactor.  Liquified propylene is  stored in spherical  tanks  (stain-
 less steel).   Sufficient supply  for  2 months continuous operation  is kept
 on site.
       Absorption  of the  N02  formed  in the  photochemical reactor would  be
 accomplished in conventional flue gas desulfurization  system.   It  is as-
 sumed that absorption  occurs according  to  reactions  49  and 50.  Unlike
 nitric acid absorption towers,  there does  not  appear to be any NO  regene-
 ration.
      2N02(g)  + 350-^, £* ZNOj^j + 2HSO;|,j  + SO^, ,  +  2H+     (49)

      4S03~fl)  + 2N02(1)~ %)   *«»i?l)                     (50)
 Approximately 10 to 50% of the absorbed N02 is ultimately corrected to
 N03" salts (Faucett et al . , 1977).  Presumably, the remainder is absorbed
 according  to reaction 50.  Laboratory scale work done by Chappel (1973)
  indicates  that  sulfite scrubbing liquors such as used in some FGD  systems
  provide high efficiency  NOg  removal.  Based on this work complete  N02 ab-
  sorption will  be assumed.  No credit will  be taken for  possible NO absorp-
  tion which in  some cases appears to reach  10 to  20%  (Chappel, 1973).  Due
  to the very high  solubilities of nitrites  and nitrates, it  is unlikely

                                   119

-------
that any scaling will occur in the FGD system as a result of the  photo-
chemical system.  Disposal of the nitrites and nitrates  will  be done at
the same time that the calcium sulfite and calcium sulfate sludges are
treated.
     The system costs are limited to only those items needed in  addition
to the  conventional FGD system.  These include the propylene storage and
distribution system, the photochemical reactor and lamps, and the elec-
trical  costs.
Model Plant Costs
     The dominant cost components are the propylene cost and the  electri-
cal power costs.  The latter were calculated from the k, level using
equation 1 developed in Section 5.  Based on the spectral effects runs
(numbers 1 to 8), it was apparent that the $Ng ca-|c/*Ng   factor  is  in  the
range of 3 to 5.  A value of 3 was used in conjunction #ith Figure  9,
which relates the product quantum yield with the required electrical
power.  This biases the results high with respect to electrical costs.
     Results of the propylene test series (runs 19 to 118) indicate  that
the desired NO reduction could be achieved at various combinations  of
propylene and UV light intensities.  The possible operating conditions
are indicated as isopleths on Figures 28 to 31.  For several  levels of
NO control, the optimum operating point with respect to  combined  energy
and propylene cost was determined by trial and error methods.  The opti-
mal operating conditions are the high UV light/low propylene regions of
Figures 27 to 30.   This is desirable in another sense since unreacted
hydrocarbon emissions are also minimized in these regions.  For these
reasons, operating points shown in Table 18 have been selected for the
cost evaluations.
     The cost of propylene is based on reports in the Chemical Marketing
Reporter (1979).  The quoted price has been stable for over a year.  An
excess  supply is forecast for at least the next 2 years  (Ponder,  1978);
therefore, it is unlikely that substantial price increases will occur.
The quantities for the model plant represent 0.00075 of  the national  pro-
duction rate.
     Electrical costs have been assumed to be 30 mills per kWh.   This  is
                                 120

-------
equivalent to the purchase of power  for  some utilities.  The  relatively
high costs have been used to account for use of  gas  turbines  to  supple-
ment generating capacity during  peak periods.
     The installed capital costs of the  reactors and the storage tanks  have
bee based on the costs reported in Peters and Timmerhaus  (1968).   These
have been updated to January 1979 using  the CE factors as  reported in
Chemical Engineering (1979).  The lamp and power supply costs are based
on Canrad Hanovia lamp HST.  These costs have also been updated
     The calculated costs have been reported in  Tables 18  and 19.   Each
of these includes four separate columns  for various levels of NO oxida-
tion.  It is assumed that in most cases  it would be desirable to reduce
the NO by 500 PPM in order to achieve the stated EPA objectives.  The
other columns are presented to indicate  the sensitivity of cost  to NO
control level and to facilitate comparisons with cost estimates  of other
systems most of which are applicable to  300 pprn NO oxidation.  The capi-
tal cost estimates range  from 37 to $50 per KW, depending on the number
of UV lamps required.  This  is approximately 1/2 the cost for a  limestone
based flue gas desulfurization system (PEDCo.,  1975).  The annualized
costs range from 3.8 to  5.6 mills per kWh.  This is higher than origi-
nally anticipated due partially to  poor  propylene reactivity.  A compari-
son of the reported costs  for various flue  gas  treatment  processes is
presented in Table  20.   The photochemical system is obviously much less
costly than the  various  NO oxidation-N02  absorption processes.  The
reason that  photochemistry is more  attractive than  the  latter is the high
cost  of  the  ozone  generating facility ($40/kW and  2.7 mills/kWh).  The
NH-  injection  technique is considerably more attractive based on  the
 present  cost  estimates.  These  two  systems, however,  are  subject  to
 serious  operational  problems which have been discussed  in earlier sec-
 tions.
      Based on the photochemical  system  cost estimates, it appears that a
 system based on propylene  is only marginally competitive.  Commercial
 utilization of the process would be unlikely unless there were serious
 operational problems in the selective noncatalytic and selective cataly-
 tic reduction techniques.   Several improvements in the photochemical  sys-
 tem could substantially  improve the marketability of the technique.  For
                                  121

-------
                                     TABLE 16.   CAPITAL COST ESTIMATES
ro
ro
Cost Component
Direct Costs
Lamp and Power Supplies
Propylene Distribution
Subtotal Direct
Indirect Costs (42% Direct
Cost)
Subtotal Direct and
Indirect Costs
Contingency and Fees (25%
Direct and Indirect Costs)
Subtotal (lamps and
distribution system)
Subtotal (reactors)
Subtotal (propylene
stage)
TOTAL
$/kW
COST, $
500
5,300,000
1,000,000
6,300,000
2,646,000
8,946,000
2,236,500
11,182,500
12,380,000
1,343,283
24,905,783
50
A PPM NO
400 300
4,321,540
1 ,000,000
5,321,540
2,235,046
7,556,586
1,889,146
9,445,733
12,380,000
1,343,283
23,169,017
46
3,261 ,540
1 ,000,000
4,261,540
1,789,847
6,051,387
1,512,846
7,564,234
12,380,000
1,343,283
21,287,518
43
200
1,793,846
1,000,000
2,793,846
1,173,415
3,967,261
991,815
4,959,076
12,380,000
1,343,283
18,682,359
37

-------
                                           TABLE 19.  ANNUALIZED COSTS
ro
CO
ANO, ppm A500-
A.
B.
C.



D.



E.



F.


Raw Materials
Utilities 5,031,000
Operating Labor
Direct
Supervision
Subtotal
Maintenance
Labor and
Materials
Supplies
Subtotal
Overhead
Plant
Payrol 1
Subtotal
Fixed 5,058,205
TOTAL 14,579,370
MILLS/kWh 5.65
A400
1 ,800,000
4,102,200

175,200
26,280
201,480

1,389,900
208,489
1,598,389

900,000
40,300
940,300
4,705,627
13,347,966
5.15
A30(£ A20(£

3,096,000 1,548,000












4,323,495 3,794,307
11,914,630 9,832,387
4.62 3.81
                  ^Detailed cost components not shown.   These can be calculated  as  a  ratio of
                   A400 values.

-------
  TABLE 20.  COMPARISON OF COSTS FOR FLUE GAS TREATMENT PROCESSES
Technique
Photochemical System
Selective Non-Catalytic
Reduction
Selective Catalytic
Reduction
Radiation Oxidation
NO Oxidation/NO,, Absorption
NO Absorption, liquid
phase condition
£ ppm NO
500
400
300
200
200
200-300
200-300
200-300
125
Capital
Cost
$/kW
50
46
43
37
N.D.
10-36
100-1000
84-134
65-130
Annual i zed
Cost
Mills/kWh
5.65
5.15
4.62
3.81
0.75-1.60
0.20-1.80
N.D.
6.7-8.9
4.8-7.4 ~
N.D. - NO Data Available
                            124

-------
example, development of more efficient lamps or lamps with effective re-
flectors could allow installation of the lamps in the electrostatic pre-
' )f»i »'iM/f •..  UH-M-K/ 'illowlri'j «•! irnj nation of the very expensive reactors.
This would have a secondary benefit since the reaction products would
serve to reduce resistivity.  Use of a more reactive hydrocarbon would
reduce hydrocarbon demand and possible reduce the light energy require-
ments.  A photochemical system based on hexene or acetone could concei-
vably operate at capital and annualized costs less than 1/2 those pre-
sented here for the propylene system.
                                  125

-------
                             REFERENCES


 1.  Acurex Corporation (1977).  EPA Stationary Source Combustion Control
     Technology Program - FY1976.  EPA Publication EPA-600/7-77-077,
     July 1977.

 2.  Acurex Corporation (1978).  Control Techniques for Nitrogen Oxides
     Emissions from Stationary Sources - Second Edition.  EPA Publication
     EPA-650/2-76-001, January 1978.

 3.  Armento, W. J. (1975).   Effects of Design and Operating Variables on
     NOX from Coal-Fired Furnaces - Phase II.  EPA Publication EPA-650/2-
     74-002b, February 1975.

 4.  Browne, W. R. and Stone, E.  E. (1965).   Sulfur Dioxide Conversion
     Under Corona Discharge  Catalysis.  Final Report, U. S. Public Health
     Service Contract No.  PH 86-67-2, March  1965.

 5.  Batayko, R. (1979).   Personal  Communication,  March 1979.

 6.  Calvert, J. G.,  Su,  F., Bottenheim, J.  W., and Strauss, 0.  P. (1977).
     Mechanism of the Homogeneous Oxidation  of Sulfur Dioxide  in the
     Troposphere.  Atmospheric Environment,  Volume 12, Pages 197-226,
     February  1977.

 7.  Canrad Hanovia (1973).   Publications B0238 and L0238,  November 1973.

 8.  Chappell, 6. A.  (1973).  Development of Aqueous Processes for Removing
     NOX from Flue Gases.   EPA Publication EPA-R2-73-051,  June 1973.

 9.  Anonymous (1979).   Economic  Indicators.   Chemical Engineering, Volume
     86, No. 2, Page  7, January 15, 1979.

10.  Anonymous (1979).   Current Prices.   Chemical  Marketing Reporter,
     Volume 215, No.  4, Page 45,  January 22,  1979.

11.  Christian, S. D.  and  Grundnes, J. (1967).  Charge Transfer  Complex
     Between Sulfur Dioxide  and Trimethyl amine in  the Gas  Phase  and in
     Heptane.  Nature,  Volume 214,  Pages 1111-1112. 1967.

12.  Corbett, W. E. Jones, G. D., Micheletti, W.  C., Wells, R. M., and
     Wilkens, G. F. (1978).   Assessment of the Need for NOX Flue Gas
     Treatment Technology.  EPA Publication  EPA-600/7-78-215,  November, 1978.
                                126

-------
13.   Crawford,  A.  R.,  Manny,  E.  H.,  and  Bartok, W.  (1977). "Application
     of Combustion Modifications to  Power  Generating Combustion Sources"
     in Proceedings of the  Second Stationary  Source Combustion Symposium;
     Volume II,  Utility and Large Industrial  Boilers.   EPA Publication
     EPA-600/7-77-073b,  Pages  3-40,  July 1977.

14.   Ctvrtnicek, T.  E.,  and Rusek, S.  N. (1977).   "Cyclone Boilers - Their
     Emissions  and Population"  in Proceedings of  the Second  Stationary
     Source Combustion Symposium-, Volume II,  Utility and  Large Boilers.
     EPA Publication EPA_600/7-77-073b,  Pages 123-142,  July  1977.

15.   Demerjian,  K. L., Kerr,  J.  A.,  and  Calvert,  J. 6.  (1975).   "The
     Mechanism of  Photochemical  Smog Formation"  in Advances  in Environmental
     Sciences and  Technology,  Volume 4,  Pages 1-213,  1975.

16.   Dols, M. 0. and Boknoven, C. (1974).   Process for the  Photochemical
     Oxidation of  NO to N02.   U. S.  Patent 3,120,479,  Issued February  4.
     1979.

17.   Dykema, 0. W. (1978).   Combustion Modification Effects  on NOX Emissions
     from Gas-, Oil-, and Coal-Fired Utility Boilers.   EPA Publication
     EPA-600/2-78-217, December 1978.

18.   Faucett, H.  L., Maxwell, J. D., and Burnett, T.  A.  (1977).   Technical
     Assessment of NOX Removal  Processes for Utility Applications.  EPA
     Publication  EPA-600/7-77-127, November  1977.

19.  Hamm, J. R.  (1965).   Study of Physiochemical Techniques for Removal
     of S02  from  Stack Gases.   Final Report, U.S  Department of Health,
     Education and Welfare Contract, November 1978.

20.  Harrison, L.  J.  (1976).  Technology  and Economics  of  Flue Gas NOX
     Oxidation by Ozone.   Final Report, EPA  Contract 68-02-1325, November
     1976.

21.  Heap, M.  P., Tyson, T.  J., Carver, G.  P., Martin, G. B., and Lowes,
     T. M.  (1976).   "The Optimization of  Burner  Design Parameters to
     Control NOX  Formation in Pulverized  Coal and Heavy  Oil Flames" in
     Proceedings  of the Stationary  Source Combustion  Symposium  EPA
     Publication  EPA-600/2-76-152b, June  1976.

 22.  Howe ,  J.  P., and Noyes, W. A. Jr. (1936).   Photochemical  Studies
      XXIII,  The Photochemical Decomposition of  Acetone Vapor  Near 1900  A.
      A Comparison with the Decomposition  at Longer Wave Lengths.   Journal
      of The American Chemical Society,  Volume 58, Pages 1404-1412,  1936.

 23.    Jeffries j n.E.  Photochemical Kinetic Simulation  System,  Version 2a.
      Unpublished  Report. June  1977.

 24.    Kemp,  V. E., and Dykema,  0. W. (1977).  "Inventory of Atmospheric
       Emissions from Stationary  Point Sources" in Proceedings of the Second
       Stationary  Source Combustion Symposium, Volume II, Utility and Large
       Boilers.  EPA Publication  EPA-600/7-77-073b, Pages 257- 278, July  1977.

                                  127

-------
25.  Lachapelle, D. (1979).  Personal Communication, May 1979.
26.   Levenspeil,  0.  (1972).   Chemical Reaction Engineering.  John Wiley
      and  Sons  Publishing Co, New York, 1972.

27.   Machi, S., Kawamura, K., Dawaakami, W., Aoki, S., Hashimoto,S.,
      Yotamoto, K., and Sunaga, H. (1975).   Process for the Photochemical
      Oxidation of NO to N02«  U. S. Patent  3,869,362, Issued February  4
      1975.

28.   Mason, H., Shimiza, A.,  Ferrell, J., Poe, G., Waterland, L., and
      Evans, R. (1977).  Preliminary Environmental Assessment of Combustion
      Modification Techniques; Volume II, Technical Results.  EPA Publica-
      tion EPA-600/7-77-119b, October 1977.

29.   McCutchen,  G. P. (1977).  NOx Emission Trends and Federal Regulation.
      Chemical Engineering Progress, Page 38, August 1977.

30.   Muzio, L. J., and Arand J.  K. (1976).  Homogeneous Gas Phase De-
      composition of Oxides of Nitrogen.   Electric Power Research Institute
      Publication FP-253, August 1976.

31.   National Academy of Sciences, Subcommittee on Nitrogen Oxides  (1977).
      Ozone and Other Photochemical Oxidants. EPA Publication EPA-600/ 1-
      77-013, February 1977.

32.   PEDCo Environmental (1975).   Flue Gas Desulfurization Process Cost
      Assessment.   Final Report,  Environmental Protection Agency Contract
      No. 68-01-3150, May 1975.

33.   Peters M. S., and Timmerhaus, K.  D. (1968).   Plant Design and
      Economics for Chemical  Engineers. Second Edition.McGraw Hill Book
      Company, 1968.

34.   Ponder T.  (1978).  U.  S. Propylene:  Demand vs.  Supply.  Hydrocarbon
      Processing,  Volume 57,  Number 7,  Pages 186-188,  July 1978.

35.   Ricci, L. J.  (1977).   Nixing NOX Emissions.   Chemical Engineering,
      Volume 84, Number 4,  Pages  84-89, April 11,  1977.

36.   Richards, J., Fox, D.,  and Reist, P.    (1976).  The Influence of
     Molecular Complexes on  the Photooxidation of Sulfur Dioxide.
      Atmospheric Environment, Volume 10, Pages 211-217, 1976.

37.   Salvesen, K.  G., Wolfe K. J., Chu,  E., and Herther, M. A. (1978).
      Emission Characterization of Stationary NOX Sources; Volume I, Results,
      EPA Publication EPA-650/7-78-120a,  June 1978.

38    Selker, A P.   (1975).   Program for Reduction of NOX from Tangential
      Coal-Fired Boilers, Phase II.  EPA Publication EPA-650/2-73-005a,
      June 1975.

                                128

-------
39.   Sickles,  J.  E.  II,  and Jeffries  ,  H.  E.   (1975).   Development  and
     Operation of a  Device for the Continuous  Measurement  of  ka  for
     Nitrogen  Dioxide.   University of North  Carolina  at Chapel  Hill,
     Department of Environmental  Sciences  and  Engineering  Publication
     ESE-396,  June,  1976.

40.   Thompson, R.  E.,and McElroy, M.  W.  (1976).   Effectiveness of Gas
     Recirculation and  Staged Combustion in  Reducing  NOX on a 560 MW
     Coal-Fired Boiler.   Electric Power Research Institute Publication
     FP-257, September  1976.

41.   Wolf, D., and Resnick, W. (1963).   Residence Time Distribution in
     Real Systems.  Chemcial Engineering Progress, Volume 2, Pages 283-
     293, 1963.

42.   Yeager, K E., and Teixeira,  D. P.  (1977).  "The EPRI Program on NOX
     Control Using Combustion Modification Techniques" in Proceedings of
     the Second Stationary Source Combustion Symposium; Volume II, Utility
     and Large Boilers.  EPA Publication EPA-600/7-77-073b, July 1977.
                                  129

-------
         APPENDIX A





Reactor Flow Characterization
            130

-------
                                                     I      I      I      I
 0.1
0.01
                      Tracer Curve for 3.0 LPM
                             Observed  Data  Points
                             Calculated  Curve
                I      I      I      I
I
                                                                        Calculated
                                                                          Results
                                           I      I
I      I  	I      tilll
i
         0.2   0.4   0.6   0.8   1.0   1.2
                             t/t •*-*
     1.4   1.6   1.8   2.0   2.2   2.4    2.6   2.8    3.0
                                               Appendix A Curve 1

-------
CO
INJ
      0.0
                         Tracer  Curve for  2.0 LPM


                             	  Observed  data points
                             	  Calculated curve
                                 I
J_
I
_L
I
                                                              Calculated Results

                                                                 a = 0.92

                                                                 p = 0.08

                                                                 d = 0.02

                                                                      .f = 0.06
I
I
I
I
         0    0.2   0.4   0.6   0.8   1.0   1.2   1.4    1.6    1.8    2.0    2.2   2.4   2.6
                             t/t-^
                                                Appendix A Curve 2
                                                     2.8   3.0

-------
      1.0
    -l->
    u_
     I
      0.1
CO
     0.01 —
                            I      I
                              Tracer Curve for 1.5 LPM
                                	 uoserved Data Points
                                	 calculated Curve
J
I
1
                                             L
1
1
                                                                     Calculated Results
                                                                       a = 0.91
                                                                       p = 0.09
                                                                       d = 0.02
                                                                       f = 0.06-
i.
J
L
         0    0.2   0.4   0.6   0.8   1.0   1.2   1-4    ]  6    ] >8    2.0    2.2   2.4   2.6   28   30
                       t/t
                                                Appendix A  Curve  3

-------
1.0
                                                                     I
I
I
                                                                       Calculated  Results
                                                                           a  =  0.92
                                                                           p  =  0.08
                                                                           d  =  0.02
                                                                           f  =  0.06
 0.1
                      Tracer Curve for  1.0 LPM
                       	 Observed Data Points
                       	 Calculated Curve
0.01
                                              I   	I
    0   0.2    0.4   0.6   P.8   1.0   1.2   1.4    1.6    1.8   2.0   2.2   2.4   2.6   2.8   3.0

                                     Appendix A Curve 4

-------
    APPENDIX B
Experimental  Results
          135

-------
TABLE B-l.  EXPERIMENTAL DATA, INFLUENCE OF MIDDLE UV
            FILTER ON NO OXIDATION.
Run Number
Chronological No.
Conditions
[N0]0» ppm
[C3Hg]0, ppm
[S02J0> ppm
[02L %
k, , min
T, min
Temp., °F
Filter >
Gas Analyses
ANO, ppm
NO Oxid., %
[N02J, ppm
[NOX]/[NOXL
AC-H.., ppm
6 0
ANO/A C3H6
ASOp, ppm
S0? Oxid., %
Filter Analysis
XRF, ppm S02
1
81
596
N.D.
1500
4.5
0.124
0.783
251
o
1800, A
223
37.5
202
0.929
N.D.
N.D.
64
4.3
0.01
2
84
665
1441
1478
4.7
0.248
0.758
274
0
>1800 A
367
55.2
283
0.929
168
2.18
37
2.5
0.01
3
79
626
1355
1339
4.5
0.34
0.760
272
o
>3000 A
108
17.3
92
0.957
194
0.56
27
2.0
0.01
4
77
637
1390
1485
4.0
1.20
0.755
277
o
>1800 A
599
95.6
535
0.88
256
2.34
86
5.8
0.07
                       136

-------
TABLE B-l  (continued)
Run Number
Chronological No
Conditions
[N0]0, Ppm
[C3Hg]0, ppm
[S02L ppm
[o2], %
k, , min"
T, min
Temp., °F

Filter
Gas Analyses
ANO, ppm
NO Oxid., %
[NOJ, ppm
[NO 1/[NO 0]
A A
AC3H6, ppm
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
5
. 82

608
N.D.
1457
4.5
0.08
0.764
269
o
>3000 A

16
2.6
31
1.00
N.D.
N.D.
16
1.1

N.D.
6
83

655
1302
1460
4.7
0.158
0.760
272
o
>3000 A

36
5.5
41
0.982
77
0.47
26
1.8

N.D.
7
80

610
1415
1393
4.4
0.68
0.758
274
0
>1800 A

567
93.0
481
0.832
281
2.02
108
7.8

0.04
8
78

616
1360
1312
3.8
0.60
0.737
295
0
>3000 A

448
72.7
394
0.895
161
2.78
59
4.5

0.03
          137

-------
   TABLE B-2.  EXPERIMENTAL DATA,  INFLUENCE OF SULFUR DIOXIDE
               ON NO OXIDATION.                       uiuAiut
Run Number
Chronological No.
Conditions
[NOX]0, ppm
[C3H,]0, ppm
[S02]0, ppm
[02L %
k, , min"
, min
Temp., °F
Gas Analyses
NO, ppm
NO Oxid., %
[N02L ppm
[NOx]o/[NOx]0
C_H, , ppm
NO/ C,HR
•3 D
S02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
9
49
1068
837
510
4.3
4.1
0.768
266
902
88.8
807
0.861
300
3.0
48
9.4
0.08
10
50
550
990
540
4.4
4.52
0.758
274
541
99.2
462
0.836
376
1.44
92
16.9
0.08
11
51
550
819
1066
4.3
4.52
0.753
279
526
98.8
464
0.855
N.D.
N.D.
44
4.2
0.07
12
52
577
790
0
4.2
4.52
0.753
279
555
98.4
493
0.869
N.D.
N.D.
0
0
N.D.
(continued)
                             138

-------
Run Number
Chronological No.
Conditions
[NOjo. PPm
X
[CoH/-]o ' PP"1
[S02]0, ppm
[02], I
k, , min~
T, min
Temp. , °F
Gas Analyses
AND, ppm
NO Oxid., %
[N02], ppm
[NOx]/[NOx]0
AC3H6, ppm
ANO/AC3H6
AS02, ppm
S02 Oxid. , %
Filter Analyses
XRF, ppm S02
13
53
1203
790
1119
4.5
4.52
0.753
279
928
77.4
804
0.893
359
2.58
62
5.6

0.15
14
1248
818
3241
4.0
4.51
0.750
282
857
69.2
675
0.847
321
2.67
133
4.1

0.04
15
55
506
877
1332
4.3
4.52
0.767
266
479
98.2
426
0.860
189
2.54
67
5.0

0.08
16
56
528
819
3197
4.3
4.63
0.767
266
497
97.4
448
0.874
N.D.
N.D.
133
4.2

N.D.
(continued)
                               139

-------
TABLE B-2 (continued)
Run Number
Chronological No.
Conditions
[N0x]0, ppm
[C3H6L, ppm
[S02]0, ppm
[0?], %
*- _i
k, , min
, min
Temp. , °F
Gas Analyses
NO, ppm
NO Oxid., %
[N02L ppm
[NOj0/[NOjo
A A
C-jHg, ppm
NO/ C.,H_
•j b
S02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
17
57

1199
643
1368
4.7
4.52
0.758
274

972
87.6
861
0.889
N.D.
N.D.
71
5.2

N.D.
18
58

1106
702
0
4.2
4.52
0.764
269

941
86.9
826
0.276
219
4.30
0
0

0.02
                  140

-------
TABLE B-3.  EXPERIMENTAL DATA, INFLUENCE OF PROPYLENE CONCENTRATION
            AND UV LIGHT INTENSITY Oil ON OXIDATION.
Run Number
Chronological No.
Conditions
[NOX]0 ppm
[C3H6]0 ppm
[so2]0
[02] %
k, , min"
1
T, min
Temp., °F
Gas Analyses
AND, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC3H6, ppm
ANO/AC-H,
«J O
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
19
39

592
0
1492
4.3
0
0.794
241

0
0
14
1.00
0
_
N.D.
N.D.

0.01
20
47

569
331
1374
4.5
0
0.786
248

0
0
9
1.00
16
0
N.D.
N.D.

N.D.
21
28

548
475
1772
4.6
0
0.788
246

0
0
14
1.02
N.D.
N.D.
0
0

N.D.
22
18

N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.

N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.

N.D.
                                   141

-------
                      TABLE  B-3  (Continued)
Run Number
Chronological No.
Conditions
[N0x]0 ppm
[C~H ]0 ppm
[so2]0
c.
[02] I
k, , min
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid. , %
[N02], ppm
[NOx]/[NOx]0
AC^H, , ppm
ANO/AC-Hg
A Qft n run
n ou n > p pin
SO, Oxid., %
<:
Filter Analyses
XRF, ppm S02
23
34

557
1856
1446

4.5
0
0.813
225

0
0
14
1.00
N.D.
N.D.
N.D.

N.D.


0.04
24
40

597
0
1492

4.6
0.10
0.781
253

23
3.8
27
0.992
0
-
9

0.6


0.02
25
7

542
246
1487

4.2
0.10
0.750
282

155
0.289
115
0.918
0
-
44

3.0


N.D.
26
23

512
394
1386

4.3
0.38
0.758
274

155
31.4
155
0.967
N.D.
N.D.
0

0


0.01
(continued)
                            142

-------
                     TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[N0x]o» ppm
[C3H5L, ppm
[S02]0, PPf"
[02L °/°
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, min
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC3H6> ppm
ANO/AC3H6
AS02, ppm
so2, %
Filter Analyses
XRF, ppm S09
L.
27
17

497
905
1439
4.5
0.38
0.794
241

244
50.5
222
0.929
0
-
44
3.0

0.02

28
6

529
1788
1615
4.8
0.38
0.764
269

356
67.2
300
0.942
80
4.45
94
5.8

0.02

29
25

529
0
1431
4.3
0.96
0.758
274

37
7.0
32
0.982
0
-
37
2.6

N.D.

27

566
247
1558
4.2
0.96
0.767
266

183
33.1
159
0.933
N.D.
N.D.
5
0.3

0.03

(continued)
                             143

-------
                     TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[NOX]0, ppm
[C3H6L, ppm
[S02]oi Ppm
[02L %
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, min
NO Oxid., %
[N02], ppm
[NOx]/[NOx]0
AC3H6, ppm
ANO/AC-H,.
AS02, ppm
so2, %
Filter Analyses
XRF, ppm S02
31
7
566
519
1700
4.0
0.72
0.764
269
269
47.5
222
0.958
74
3.61
71
4.2
0.02
32
36
561
979
1333
4.3
0.33
0.767
266
363
66.4
331
0.918
280
1.30
68
5.1
0.03
33
44
565
1850
1275
4.3
0.47
0.774
259
454
81.5
398
0.886
162
2.81
34
2.7
N.D.
34
45
565
0
1425
4.2
0.41
0.764
269
39
6.9
34
0.977
0
-
47
3.3
N.D.
(continued)
                             144

-------
                     TABLE  B-3  (Continued)
Run Number
Chronological No.
Conditions
[NO ]0» PPm
X
[C-jHglo, ppm
[S02]0> PPm
[o2], %
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02L ppm
[NOX]/[NOX]0
AC,H,-> ppm
ANO/ AC^ H£
j b
ASO~, ppm
L.
S0? Oxid. , %
Filter Analyses
XRF, ppm S02
35
41

555
331
1381
4.5
1.37
0.768
264

262
48.0
226
0.920
126
2.08
4

0.3

0.05
36
33

524
475
1550
4.6
1.83
0.758
274

317
61.6
262
0.877
N.D.
N.D.
46

3.0

0.03
37
46

586
475
1395
4.5
1.11
0.760
272

471
81.5
411
0.883
153
3.07
51

3.7

N.D.
38
15

606
2108
1761
5.3
1.37
0.783
251

585
98.4
528
0.889
0
-
71

4.1

0.16
(continued)
                              145

-------
                        TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[NO lo, ppm
X
[C3H6L, pom
[S02]o, ppm
[02L %
k , min ,
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOXL
AC-jHg, ppm
ANO/AC3H6
ASOp, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm SO 2
39
31

538
475
1498
4.9
3.68
0.750
282

61
11.5
61
0.983
N.D.
N.D.
37
2.5

0.44
40
38

505
344
1465
4.6
3.21
0.745
287

342
57.9
275
0.880
200
1.70
71
4.8

0.07
41
35

524
475
1460
4.5
2.40
0.764
269

389
75.4
344
0.897
283
1.38
18
1.2

0.02
42
22

502
979
1442
4.2
3.21
0.774
259

481
98.3
418
0.850
N.D.
N.D.
64
4.3

N.D.
(continued)
                              146

-------
                    TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[N0x]o, PP
[C3H6]0, ppm
[S02]0» Ppm
[o2], %
k, , min
T, min
Gas Analyses
ANO, ppm
NO Oxid., %
[N02L ppm
[NO l/[NOy]0
A A
AC3Hg, ppm
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
43
12

534
1738
1685
4.0
3.21
0.772

505
0.973
440
0.851
217
2.33
47
2.8

0.11
44
32

1150
0
1481
4.5
0
0.788

0
0
9
1.00
N.D.
N.D.
0
0

0.04
45 46-
30

955
265
1488
4.6
0
0.797

0
0
14
N.D.
N.D.
N.D.
N.D.
N.D.

N.D.
     missng
(continued)
                             147

-------
                     TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[NOX]0, Ppm
[C3H6]0, ppm
[S02]0, ppm
[OJ, %
_]
k,min
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02L ppm
[NOX]/[NOX]0
AC3H6, ppm
ANO/AC3H&
ASOp, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
47
42

1055
1072
1386
4.3
0
0.810


9
0.8
17
1.00
0
0
f!.r;.
N.D.

N.D.
48
48

1327
2066
1340
4.7
0
0.792


9
0.6
21
1 .00
N.D.
N.D.
N.D.
N.D.

N.D.
49
3

1217
0
1615
4.3
0.32
0.758
274

0
0
23 ,
1.00
0
0
19
1.2

0.01
50
13

1030
243
1615
4.8
0.18
0.768
264

103
10.1
103
0.990
0
-
23
1.4

0.01
(continued)
                            148

-------
                     TABLE  B-3  (Continued)
Run Number
Chronological No.
Conditions
[NOX]0, ppm
[C3H6]0, ppm
[S02]0» PP1"
[02], %
k, , min"
T, min.
Temp., °F
Gas Analyses
AND, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC3Hg, ppm
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
51
14

1095
562
1714
4.8
0.18
0.794
241

138
12.8
133
0.983
0
•-
0
0

0.01
52
21

1018
937
1576
4.5
0.25
0.783
251

158
16.6
200
0.975
0
-
64
4.1

0.02
53
16

1028
1917
1547
4.3
0.38
0.768
264

271
26.8
262
0.977
0
-
24
1.5

0.02
54
5

1123
1886
1451
4.
0.
0.
269

460
41
398
0
98
4
70
4

0





8
72
,764



.0

.917

.69

.8

.05
(continued)
                              149

-------
                       TABLE  B-3  (Continued)
Run Number
Chronological No.
Conditions
[NOjo, ppm
A
[C,H,]0, ppm
O 0
[S00]0, ppm
L.
[o2L %
k, , min
T, min
Temp., °F
Gas Analyses
ANO, min
NO Oxid., %
[N02J, pom
[NOX]/[NOX]0
AC3Hg, ppm
ANO/AC0H,
o b
ASO?, ppm
so2, %
Filter Analyses
XRF, ppm S07
55
20

980
265
1555

4.5
0.72
0.768
264

145
15.4
153
0.970
0

21
1.4

0.03
56
29

1025
475
1460

4.6
0.96
0.781
253

187
18.5
178
0.977
N.D.
N.D.
42
2.9

0.02
57 58
9

1040
1016
1627

4.6
1.14
0.760
272

312
30
249
0.972
157
1.99
45
2.8

0.03
(continued)
                            150

-------
                    TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[NOX]0. ppm
[C3H63o. ppm
[S02]0, PF
[023. *
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N023, PPm
[NOX3/[NOX]0
AC3Hg, ppm
ANO/AC3H&
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
59 6(
24

949
0
1344
4.5
1.83
0.755
277

29
3.3
67
0.991
0
-
34
2.5

0.02
£ 61
19

963
380
1541
4.2
1.37
0.764
269

201
21.7
214
0.978
0
-
43
2.8

0.05
62
2

1035
740
1541
4.3
1.85
0.745
287

391
37.8
345
0.933
73
5.36
69
4.5

0.06
     missng
(continued)
                             151

-------
                     TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[NOjo, Ppm
X
[C-jHglo, ppm
[SOplo. ppm
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02L ppm
*[NOX]/[NOX]0
ACgHg, ppm
ANO/AC-jH,.
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S0?
63
1
1150
1587
1472
4.6
2.34
0.767
272
552
48.0
460
0.920
253
2.18
46
3.1
0.33
64
26
1063
0
1509
4.5
3.21
0.747
285
41
4.0
60
0.996
0
-
28
1.8
0.08
65
43
976
255
1361
4.3
3.21
0.760
272
295
30.8
218
0.904
142
2.08
86
6.3
4.13
66
4
1147
543
1755
4.7
3.21
0.750
282
445
38.8
328
0.878
136
3.28
140
8.0
0.33
(continued)
                            152

-------
                    TABLE B-3  (Continued)
Run Number
Chronological No.
Conditions
[NOjo, ppm
X
[C3Hg]0, ppm
[so23<,, ppm
[02]. J
k, , min
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N023, ppm
[NOX]/[NOX]0
AC3Hg, ppm
ANO/AC3H6
ASO?, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
67
37

1175
979
1424
4.3
2.80
0.750
282

601
53.2
511
0.885
519
1.16
86
6.0

0.06
68
10

977
1856
1554
4.5
5.00
0,740
292

932
96.7
755
0.805
257
3.62
111
7.1

0.40
69
121

658
0
1984
4.64
0
0.772
261

0
0
16
l.QO
0
N.A.
0
0

0.08
70
157

460
480
1770
4.17
0
0.774
259

0
0
5
N.D.
0
N.A.
N.D.
N.D.

N.D.
(continued)
                              153

-------
                     TABLE B-3  (Continued)
Run Number
Chronological No.
Conditions
[NOL, ppm
[C3H6L, ppm
[S02]0, ppm
[Q2], %
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Ox id., %
[N02L ppm
[NOX]/[NOX]0
AC-jHg, ppm
ANO/AC-.H..
o o
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm SO.
c
71
156

443
540
1701
4.64
0
0.774
259

0
0
5
N.D.
0
-

N.D.
N.D.

N.D.

72
144

612
1415
1835
4.64
0
0.774
259

0
0
11
0.994
0
-

10
0.5

0.18

73
128

625
1970
1792
4.64
0
0.774
259

0
0
11
1.00
0
™

0
0

0.11

74
120

614
0
1934
4.56
0.27
0.752
280

11
1.78
27.4
1 .00
0
N.A.

0
0

N.D.

(continued)
                             154

-------
                    TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[NOX]0, ppm
[C3H6]0, ppm
[S02]0, PP
-------
                      TABLE B-3  (Continued)
Run Number
Chronological No.
Conditions
[N0]o, ppm
[C3Hg]0, ppm
[S02]0, ppm
[02], %
k, , min~
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
([NO] + [N02]/
[NOD
AC-)H,-, ppm
36
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses

XRF, ppm S02
79
133

570
0
1907
4.48
0.49
0.777
256

44
7.7
44

0.983
0
N.A.
33
1.7

0.27
80
108

626
303
2035
4.70
0.48
0.76
269

192
30.9
184

0.980
81
2.39
38
1.8

1.08
81
124

652
620
1912
4.64
0.47
0.764
269

252
38.6
225

0.991
106
1.80
16
0.9
(Both sides)
0.15/0.19
82
139

583
1300
1825
4.47
0.47
0.760
272

367
63.0
324

0.973
196
2.04
9
2.7

N.D.
(continued)
                            156

-------
                     TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[NOJo. Ppm
X
[C3H6]0, ppm
[S02]0» ppm
[02], *
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NO 1/LNOlo
A A
ACoH,-, ppm
36
ANO/AC3Hg
AS02, ppm
S02 Oxid. , %
Filter Analyses
XRF, S02 ppm
83
113

619
2005
1911
4.56
0.48
0.745
287

416
67.3
384
0.977
305
1.36
44
2.3

3.80
84
135

616
0
1809
4.47
0.58
0.760
272

54
8.8
49
0.970
0
N.A.
54
3.0

0.82
85
145

617
370
1863
4.64
0.63
0.769
264

183
29.7
133
0.902
120
1.31
44
2.4

1.15
86
116

641
630
1902
4.64
0.91
0.755
277

323
50.4
285
0.940
161
2.48
22
1.15

0.35
(continued)
                             157

-------
                     TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[N0]0, ppm
[C0HC]0, ppm
o u
[S02]0, ppm
[02], %
k, , min
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC0HC, ppm
36
ANO/AC3H6
AS02, ppm
SQp Oxid. , %
Filter Analyses
XRF, S02 ppm
87
118

614
1270
1896
4.56
0.63
0.755
277

449
73.1
449
0.954
343
1.04
87
4.6

0.25
88
148

548
2040
1836
4.56
0.63
0.750
282

482
88.0
405
0.845
452
0.89
0
0

0.81
89
154

592
0
1808
4.46
1.55
0.747
285

38
6.5
38
0.982
0
N.A.
54
3.0

0.28
90
126

647
350
1819
4.64
1.55
0.760
272

285
44.0
247
0.910
170
1.67
55
3.0

2.19
(continued)
                            158

-------
                     TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[N0]0, ppm
[C.jHx.]o» ppm
[S02]0» ppm
[02L %
k, , min"
T, min
Temp., °F
Gas Analyses
AND, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC3H6, ppm
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
91
136

610
680
1809
4.47
1.55
0.758
274

464
76.0
383
C.949
342
1.29
92
5.1

5.67
92
112

635
1320
1880
4.64
1.59
0.75
279

603
94.9
504
0.976
550
1.10
60
3.2

N.D.
93
109

606
2120
1995
4.64
1.59
0.75
282

596
98.3
532
0.894
N.D.
N.D.
37
1.8

1.07
94
127

992
0
2000
4.64
0
0.777
256

0
0
16
N.D.
N.D.
N.A.
N.D.
N.D.

0.70
(continued)
                             159

-------
                     TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[NO ]0, ppm
X
[C.,Hfi]0, ppm
[S02L, ppm
[02L %
k, , min~
T, min
Temp., °F
Gas Analyses
AND, ppm
NO Oxid., %
[N02L ppm
[NOX]/[NOX]0
AC3Hg, ppm
ANO/AC,HC
J D
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S0?
95
129

943
320
1748
4.64
0
0.781
253

0
0
16
1 .00
0
-
0
0

0.41
96
125

1008
670
1852
4.64
0
0.788
246

0
0
22
0.991
0
-
16
0.9

0.75
97
115

1162
1320
1857
4.64
0
0.771
262

0
0
22
1.00
17
0
0
0

N.D.
98
in

1090
1927
1852
4.64
0
0.77
259

6
0.55
29
1.00
0
-
0
0

N.D.
(continued)
                          160

-------
                     TABLE B-3  (Continued)
Run Number
Chronological No.
Conditions
[N0]0, PPm
[CgHgL, ppm
[S02L, Ppm
[02L %
k, , min
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02L ppm
[NOX]/[NOX]0
AC3H6, ppm
ANO/AC3Hg
AS09, ppm
u
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
99
117

1178
0
1956
4.56
0.27
0.769
264

6
0.51
33
0.994
0
N.A.
11

0.6

N.D.
100
131

992
310
1765
4.48
0.31
0.764
269

93
9.4
88
0.991
45
2.1
16

0.9

0.95
101
157

1014
720
1874
4.64
0.20
0.753
279

82
8.1
88
0.984
72
1.0
22

1.2

0.25
102
143

1006
1415
1835
4.64
0.18
0.769
264

156
15.5
128
0.957
200
0.78
11

0.6

0.59
(continued)
                              161

-------
                      TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[N0]0» ppm
[C3Hg]0, ppm
[S02]0> ppm
[02], %
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
AC3H6, ppm
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
103
149

1014
2040
1863
4.56
0.18
0.755
277

170
16.7
137
147
0.74
33
1.8

0.31
104
134

1026
0
1874
4.47
0.47
0.764
269

19
1.8
38
0
N.A.
22
1.2

0.42
105
130

937
320
1754
4.48
0.47
0.767
266

148
15.8
121
64

11
0.6

3.20
106
no

1138
^670
1942
4.64
0.48
0.75
282

250
22.0
218
N.D.
N.D.
48
2.5

0.25
(continued)
                             162

-------
                   TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[N0]0, ppm
[C3H6]0, ppm
[S02]0, Ppm
[OJ, %
k. , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC-H/-, ppm
3 o
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
107
152

1008
1320
1841
4.64
0.33
0.772
261

219
21.7
164

180
1 .22
71
3.9

0.83
108
138

1010
2030
1809
4.47
0.47
0.758
274

367
36.3
308

235
0.94
54
3.0

1.78
109
142

1006
0
1846
4.64

0.764
269

39
3.9
39
0.984
0
N.A.
0
0

0.42
no
137

1021
370
1793
4.47
0.88
0.755
277

200
19.6
146
0.927
110
1.54
32
1.8

2.65
(continued)
                            163

-------
TABLE B-3 (Continued)
Run Number
Chronological No.
Conditions
[N0]o, ppm
[C3H6J0, ppm
[S02]0, ppm
[02L %
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC3Hg, ppm
ANO/AC3H6
AS02> ppm
S02 Ox id., %
Filter Analyses
XRF, S02 ppm
111
150

1014
720
1874
4.64
0.63
0.750
282

197
19.5
164
0.942
170
1.0
38
2.0

0.76
112
141

1001
1415
1910
4.64
0.63
0.786
248

334
33.3
295
0.930
345
0.97
73
3.8

1.00
113
119

1222
2035
1869
4.56
0.88
0.747
285

477
39.0
444
0.962
363
0.94
71
3.8

0.54
114
153

948
0
1814
4.46
1.30
0.747
285

27
2.9
33
0.988
0
N.A.
60
3.3

0.53
      164

-------
                     TABLE  B-3  (Continued)
Run Number
Chronological No.
Conditions
[N0v]0, ppm
X
[CoH,- Jo » ppm
[S02]0» ppm
[02], %
k, , min ,
1 ' -1
T, min
Temp., °F
Gas Analyses
AND, ppm
NO Oxid., %
[NO ], ppm
[NOX]/[NOX]0
AC^H,, ppm
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, ppm S02
115
123

1200
270
1923
4.56
0.47

0.750
282

236
19.6
219
0.980
124
1.82
38
2.0

0.57
116
122

1174
630
1918
4.64
1.55

0.758
274

378
32.2
340
0.977
280
1.35
60
3.1

1.43
117
146

1012
1320
1812
4.64
1.10

0.755
277

517
51.0
417
0.882
534
0.89
78
4.3

2.94
118
147

983
1848
1728
4.56
1.75

0.747
285

594
60.4
497
0.886
647
2.59
76
4.4

2.91
(continued)
                             165

-------
   TABLE B-4  EXPERIMENTAL DATA, COMPARISON OF ETHYLENE
              AND PROPYLENE
Run Number
Chronological No.
Conditions
[NO Jo, ppm
X
[OlefinL, ppm
[S00]0» ppm
L -I
k, , min"
T, min
Temp., °F
Olefin
Gas Analyses
AND, ppm
NO Oxid., %
[N02L ppm
AOlefin, ppm
ANO/A Olefin
AS09, ppm
c.
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
119
92

913
1326
1600
1.8
0.764
269
C3H6

593
65.0
511
516
1.15
46

2.9

0.23
120
90

960
759
1553
1.8
0.769
264
C2H4

243
25.3
206
242
0.76
83

5.3

0.08
121
86

917
1048
1582
1 .8
0.769
264
C2H4

330
36.0
262
335
0.68
131

8.3

0.12
122
87

901
1886
1651
1.8
0.767
266
C2H4

744
82.6
561
581
1.09
162

9.8

0.11
(continued)
                            166

-------
TABLE B-4 (Continued)
Run Number
Chronological No.
Conditions
[N0x]o, ppm
[Olefin], ppm
[S02]0, ppm
[02]0» PPm
k, , min"
T, min
Temp., °F
Olefin
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AOlefin, ppm
ANO/AOlefin
ASOo> ppm
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
123
91
898
1393
1610

1.8
0.764
269
C2H4
521
48.0
439
412
1.26
88
5.4
0.06
124
85
865
482
1609
4.6
1.8
0.774
259
C3H6
314
36.3
241
0.958
194
1.62
68
4.2
0.26
125
89
954
980
1598
4.5
1.8
0.767
266
C3H6
487
51.1
409
0.961
371
1.31
62
3.9
0.06
126
88
927
1624
1614
4.7
1.8
0.769
264
C3H6
686
74.1
561
0.916
566
1.21
136
8.4
0.27
        167

-------
TABLE B-5  EXPERIMENTAL DATA, INFLUENCE OF OXYGEN CONTENT
           ON NO OXIDATION.
Run Number
Chronological No.
Conditions
[NOX]0, ppm
[CgHgL, ppm
[S02]0, ppm
[02], J
k, , min
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC3H6, ppm
ANO/AC3H6
ASOp, ppm
S02 Oxid. , %
Filter Analyses
XRF, S02 ppm
127
66

515
569
1371
0
4.5
0.738
295

120
23.5
44
0.845
26
4.64
9
0.6

0.06
128
63

580
569
1260
0.81
4.5
0.768
264

392
68.5
300
0.827
117
3.34
52
4.1

0.05
T29
64

597
569
1317
2.45
4.5
0.758
274

558
94.8
462
0.825
190
2.94
118
8.9

0.13
130
61

519
702
1328
4.43
4.5
0.772
261

493
98.2
444
0.872
157
3.14
40
3.0

0.12
                            168

-------
TABLE B-5 (Continued)
Run Number
Chronological No.
Conditions
[N0xlo. ppm
[C3H6]0, ppm
[S02]0, ppm
[Op], %
-1
k, , min
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC-jHg, ppm
ANO/AC-H,
o o
ASO^, ppm
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
131
67

604
695
1465
7.42
4.5
0.767
266

559
95.5
506
0.882
N.D.
N.D.
49
3.3

0.09
132
65

1020
569
1347
0
4.5
0.738
295

126
12.4
44
0.914
102
1.23
9
0.6

0.08
133
59

1101
637
1314
0.86
4.5
0.764
269

404
37.1
2GO
0.875
109
3.70
49
3.7

0.06
134
62

1043
702
1296
2.54
4.5
0.784
256

626
61.3
484
0.842
231
2.71
N.D.
N.D.

0.08
            169

-------
TABLE B-6  EXPERIMENTAL DATA, INFLUENCE OF CO
           ON NO OXIDATION.
Run Number
Chronological No.
Conditions
[NOXJ0, ppm
[C3H6]OJ ppm
[S02L, ppm
[02L %
k, , min"
T, min
Temp., °F
[CO], ppm
Gas Analyses
ANO, ppm
NO Oxid., %
[N02L ppm
[NOX]/[NOX]0
AC3Hg, ppm
ANO/AC^H,.
o b
AS09, ppm
C.
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
137
106

875
1137
0
4.5
1 .55
0.764
269
0

680
81.3
653
0.926
N.A.
N.A.
N.A.

N.A.

N.D.
138
105

848
1180
0
4.7
1 .45
0.764
269
1124

718
86.4
621
0.866
N.A.
N.A.
N.A.

N.A.

N.D.
139
107

842
1137
0
4.5
1.65
0.767
266
1985

783
95.4
707
0.885
N.A.
N.A.
N.A.

N.A.

N.D.
140
104

896
1180
0
4.7
1 .95
0.764
269
3000

832
96.2
718
0.837
N.A.
N.A.
N.A.

N.A.

N.D.
                       170

-------
TABLE B-7  EXPERIMENTAL DATA,  INFLUENCE OF INITIAL NOX
           CONCENTRATION ON NO OXIDATION.
Run Number
Chronological No.
Conditions
[NO ]0, ppm
A
[C3H6], ppm
[S02]0. ppm
[02], I
— 1
k , min
1
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02L ppm
[NOX]/[NOX]0
AC3Hg, ppm
ANO/ACgHg
ASOp. ppm
S02 Oxid., %
Filter Analyses
XRF, S02 pom
141
97

445
1183
1582
4.5

N.D.

0.771
261

430
100.0
372
0.835
N.D.
N.D.
73
4.6

N.D.
142
98

770
917
2086
4.5

N.D.

0.764
269

519
68.8
445
0.884
N.D.
N.D.
73
3.5

N.D.
143
100

1231
917
1614
4.5

N.D.

0.758
274

445
37.0
377
0.923
N.D.
N.D.
63
3.9

N.D.
144
102

1001
904
1719
4.7

N.D.

0.764
269

482
49.5
419
0.911
N.D.
N.D.
37
2.1

N.D.
                               171

-------
TABLE B-7 (Continued)
Run Number
Chronological No.
Conditions
[NOX]0, ppm
[C3Hg]0, ppm
[S02]OJ ppm
[02L %
I
k, , min
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[N02L ppm
[NOX]/[NOX]0
AC^Hg, ppm
ANO/AC3Hg
AS02, ppm
S02, Oxid., %
Filter Analyses
XRF, S0? ppm
145
103

1467
904
1729
4.7

N.D.
0.764
269

445
30.9
367
0.929
N.D.
N.D.
47
2.7

N.D.
146
101

1886
872
1651
4.5

N.D.
0.767
266

367
20.0
341
0.958
N.D.
N.D.
52
3.2

M r>
n . L/ .
i in
147
99

2017
1019
1582
4.5

N.D.
0.760
272

367
18.7
341
0.961
N.D.
N.D.
37
2.3

N.D.
••»•— •H^B— -••—
           172

-------
TABLE B-8  EXPERIMENTAL DATA,  INFLUENCE OF
           CONCENTRATION ON NO OXIDATION.
Run Number
Chronological No.
Conditions
[NO¥L, ppm
A
[C3H6]0, ppm
[SO^loi ppm
[02], %
k, , min"
T, min
Temp., °F
[C02]0. %
Gas Analyses
ANO, ppm
NO Oxid., %
[N02], ppm
[NOX]/[NOX]0
AC3H6' ppm
ANO/AC0H,
3 0
ASOp, ppm
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
148
94
862
1018
0
4.2
1 .55
0.768
264
0
501
59.1
428
0.898
N.D.
N.D.
N.A.
N.A.

149
93
859
966
0
4.4
1 .55
0.771
261
6.3
477
57.2
409
0.890
N.D.
N.D.
N.A.
N.A.

150
96
898
1000
0
4.5
K55
0.771
261
12,6
506
58.0
439
0.897
N.D,
N.D.
N.A.
N.A.

151
95
846
855
0
4.2
1 .55
0.768
264
18.5
470
56.9
413
0.908
N.D.
N.D.
N.A.
N.A.

                           173

-------
TABLE B-9  EXPERIMENTAL DATA, INFLUENCE OF GAS TEMPERATURE
           ON NO OXIDATION.
Run Number
Chronoloqical No.
Conditions
[NO ]0, ppm
[C.Hg]0, ppm
[S02]0» ppm
[02L %
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[NOX]/[NOX]0
AC3Hg, ppm
ANO/AC0HC
J D
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
152
73

547
548
1387
3.7
1.55
0.772
115

453
84.9
0.844
199
2.28
100
7.2

0.33
153
74

572
541
1385
4.2
1.45
0.760
187

542
97.3
0.845
90
6.09
44
3.2

0.53
154
71

547
625
1/165
4.4
0.8
0.747
214

498
93.2
0.851
74
6.75
63
4.3

N.D.
155
76

651
579
]?.M
4.2
1.45
0.753
285

600
94.7
0.829
77
7.80
N.D.
N.D.

0.16
                             174

-------
TABLE B-9 (Continued)
Run Number
Chronological No.
Conditions
[NOXL, ppm
[C-,Hg]0» PP111
[S02]0, Ppm
[02], %
k, , min"
T, min
Temp., °F
Gas Analyses
ANO, ppm
NO Oxid., %
[NOX]/[NOX]0
AC3H6, ppm
ANO/AC3H6
AS02, ppm
S02 Oxid., %
Filter Analyses
XRF, S02 ppm
156
70

1543
849
1359
5.0
1.45
0.742
132

703
46.8
0.911
100
-7.04
98
7.2

0.25
157
75

1203
584
1232
3.8
1.55
0.763
198

631
C3.5
0.881
N.D.
N.D.
59
. 4.8

0.08
158
69

1190
734
1357
4.3
1.45
0.753
214

790
68.0
0.912
145
5.34
143
10.5

0.17
159
72

1238
541
1637
4.0
1.23
0.753
295

823
67.8
0.885
127
6.70
100
6.1

0.15
          175

-------
  APPENDIX C





Kinetic Model
    176

-------
                   APPENDIX C - KINETIC MODELING

     A preliminary attempt was made to adapt a propylene/NOx kinetic
mechanism developed by Jeffries (1977) for use in analyzing experimental
results with the Photochemical Kinetics Simulation System (PKSS) also
developed by Jeffries.  This kinetic mechanism had successfully been used
to simulate smog chamber results at the UNC research facility.  Since
these tests were conducted in a batch reactor and at ambient conditions,
some mechanism modifications were necessary.
     One change was to operate the program so as to simulate a CSTR using
dilution and continuous mass input.  By balancing these two factors it
was possible to simulate the CSTR conditions at an average reactor residence
time of 0.75 minutes.  Model runs under dark phase conditions confirmed
that the concentrations of NO, S02, and 03^ remained constant.
     Initial simulation runs indicated that certain reactions were too
rapid for the time scale chosen.  Rescaling of  the variables was necessary
to simulate the short time scale events;  the units used in the  rescaling
were one-thousandth of a minute.  These units required no  change in  the
second order rate constants.   It was necessary  to divide the  first order
rate constants  (including  k-|)  by a  factor of 1000 and multiply  the  third
order rate  constants  by a  factor of 1000.   The  resulting mechanism  is
shown in the attached sample computer  printout.   The basic PKSS program
was not changed,  therefore,  the  units  continued to be printed in units of
 "minutes" while the  actual  time  was "minutes  x  10~3".
      There  are  several  obvious deficiencies in  this mechanism:  (1)  ozone
 -propylene  reactions were inadvertently omitted, (2) no  attempt was made
 to adjust  photolysis rates of aldehydes due to  middle UV light, and (3)
 photolysis  rates were not adjusted to reflect absorbance changes  as a
 function of temperature.
      The NO conversion estimates of the sample run are approximately 15%
 of the actual  results based on runs 44 to 68.  The primary reason for the
                                   177

-------
difference is probably the first error identified above.  Other possible
contributing factors are listed below:
     (1)  Aldehyde photolysis in the middle UV
     (2)  Photolysis of species other than aldehydes in the middle UV
     (3)  Improper selection of composite reactions due to high gas
          temperature and/or low oxygen concentration.
     (4)  Improper selection of photolysis rates due to temperature
          effects and/or middle UV light.
     (5)  Improper use of kinetic model
     (6)  Improper mechanism for high concentration, high temperature
          system
     Although, initial attempts at modeling the experimental results with
PKSS were not successful, there is no apparent reason why this mechanism
can not be improved so that PKSS could be used to simulate a system such
as that used in the project.   It is recommended that modeling efforts be
emphasized in future work.
                                178

-------
                                ~P" H u T~0~~C" H C  M~~I~C~J~TK I N E TIC ^   SIMULATION   S Y S T E~~M
                                            VERSION  003A         CREATED 07/12/77
                                 ............   USE*	GOCUxCNTATION
                                 IHIS JS A  TEST RUN
                                 U5I»»;-Ti~ROT HCCFWT5M  BA5LD  UN KLC£N' FINDINGS INCLUDING!
                                     II NEW RATES  FOR  H02tOH WITH NO » N02 GIVING HN02.  HN03.  I
                                           TTO2W07T~iHICH  HAS HIGR'
                                     ?) NEW 03 I OLEFIN  CHEF;  CREGIEE "ECH. STABLE BIRAOICALS,  HOT
                                      ftLCTK 8C1U.	
                                     si NEW PAN CHEH: EITHER  HCNDRY«S RATES OR cox«s RATES.
                                         » HAS UNntHGONE  SUCEb^IVL HLVIS10N AS> L'OXPARISON  HKOCEED—
                                     I    ADDED R02 + NOOKETONE+HN02 TO PROPY CHEM	

                                      VLRSION 6B
         THL WnXJMUM  NU"HLH  OF:
             SPECIES  IS   "»0l

             REACTIONS  I!>   75T
             TABLE VALUES  ISIT!

             ITEMS IU BL' SAVED 15  ZOo-NO SELCCTIvC pUNCHlMC  OF  OUTPUTS
        DILUTION HATE PROFILE"  FRUHlgnNSTANT DILUTION OF 1.5E-3  USED  TO SIMULATE TAU OF 0.75

       "INITIAL CONCLNTHA1I9N  OF  NU
        INITIAL CdNCtNTRATfON  OF  NO2=0.1
        "IW1TIAL eftNCLNTRAtlON  OF  PflOPTLsl.O
        'INITIAL CJivCtNTRATION  OF  03    =0.05
        INITIAL CONtLNTRTTiON  OF  SO?=2.5
         INITIAL CONCENTRATION  OF  HCHT=0.1
                        RATION  or  02     =<«o.
        INITIAL CONCLNTRATION  OF  N20n=i.oc-5
        INITIAL CONC(.NTRATION  OF  NDS   =i.OE-5

        INITIAL CONCINTRATION  OF  HNO?=i.oE-"r

        INITIAL CONCtMTRATlON  of  CCwi   =TT."5

        INITIAL coNCtNTRATiorj  OF  coo   =l.oC-7
                ToNcENTRTfjiiV o? fcbi "=i.ot^T

^   	INITIAL cm
-------
	 INITIAL" cnNCtNTRaTioN
INIHAL CONCENTRATION
INITIAL CONCENTRftTION
OPTIONS SELECTED:
CNSTOIL
TtlAX
CNSTCAT
CNSTTf"
CNSTH2
CUSUI
C«">TFI
Hh4X
»RTST£P 50
PLTSTEP so
END
THIS RUS u«;cs CONSTANT
Of H02 =l.?E-6
OF o =i.:->io 	 	
OF PAN =l.:E-3

0.0015
no oo.
1.
itOO.
10.
0.0017 '- —
20.

LIGHT INTEMSITY EOU4L TO 1.70000E-03
THIS RUN USES CONSTANT  TEHPEHiTURE  EQUAL TO t.OOOOOE*OZ
THIS RON USES CONSTANT  yjitER  cONC  E8U»L TO 1.00000E*01
THIS RuN USCS CONSTANT  CATALYSIS  CO'JC EQUAL TQ 1.00000E*00
THIS RUN USES  CONSTANT  DILUTI'lM RATE COUOL TO 1.50000E-03

-------
»•  t  c"
                                                                                                         "A  N  i " s  *
00

1)
( 2)
31

« 31

7)
( 6)
91
<
10t


13 1

15 (











( 261
t


THIS is- » ™ ST "UN
N02 --HV-->
03 "HV-->
o 	 >





N03 * NO 	 >
N03 + N02 	 >
N205 	 >
rj205 	 >
03 --HV-->
010 	 >
010 ...... >
n* + — --^L
05 + OH 	 >
OH » NO 	 >
HNO2 --HV-O
OH « N02 	 >
OH » HN03 	 ->
03 « HO? 	 	 	 >

H02 + NO? 	 >
H02U02 	 >
HO2 * N03 -._-.->
Mn? » H02 -•-».->
H20? --HV-O



1.00 NO * 1.00 0
1.00 010
1.00 03
1.00 N02
1.00 NO?
1.00 NO
1.00 N03
1.00 N03
2.00 N02
1.00 N205
1.00 NO 3 « 1.00 NO?
2.00 WHN03
0.11 01D + 0.89 0
1.00 0
2.00 OH

1.00 HO?
1.00 HN02
1.00 NO * 1.00 OH
1.00 HN03
1.00 N03
1.00 OH
1.00 N02 * 1>00 OH
1.00 H02N02
1.00 N02 •> 1.00 H02
1.00 HN03
1.00 H202
2,00 OH
2.00 NO?


1.000E«00 <
l.OOOE+00 <
. .«.OOE*02^E<°( 5.10E»02/TX
1.3iOF*03»Ei(«l-1.20E«03/TX
«.!i«nr»os <
1.360E»OH <
3.610E*03 <
1.77nE + 02»E«P(-2.i»5E + 03/TX
2.808E+OM <
2.187E»02»E*P( 8.61E*02/T)<
7.<-1.03E*Oi»/TX
(H20I* 8.000E-07 <
(E)» 5.380C-02 <
8.5tOE-01 <
(H2CI. 9.110E.06 <
H.620E-07 <
2.217E«03«EXP(-1.00E«03/T)<
3.859E + 02»E*P( 1.11E+03/TK
5.000E-01 <
1.UOOC*04 <
8.868E*02«EXP(-'t.OOE*02/TX
1.079E«02»EXP(-1.28E + 03/TX
1.182E + 01* <
9.000E»03 <
H.500E«m»E«°(-1.17E + OH/TX
2.500E*03 <
lt.B9'»E»Ot»CX<>(-5.mE*02/TX
l.OOOE-01 <
2.5?ffE-05»EX»( 5.30E«02/T)<




e.85«*02>
2.371E«Ol>
3.500t«03>
1 .SSOEtOO
3.6
 i
2.808E«0«>
I
3.933E«03>
6.85tE»00> <
a.OOOE-03>
^.382E-02>
8,5tOE-01> . r
_5jltOE-02>
it.620C.07>
7.73HE*01>
1.600C«0
2.
l.«OOE*0»>
2.317E*02>
1 .H96E+00>
1.182E*0
•J.OOOEt-OS)
..20fcE-OS>
2.500E403>
7.980E«03>

l.»9?E.O»> ,


-------
00
30T
31 1
32)
33)
' 311
351
(
ib)
' r§7>
38)
(
vr\
t ••„,
"It
f
";?)
*- U3'
r
"it
' "61
NO V NOZ 	 --> 2
HNOZ + HlMUZ ------> J
OH » CO 	 } 1
HCMU --HV > -1

CCHO" --HV--> 1


CrOS » M02 	 > 1

	 > 1
	 > 1
	 > 1
	 > 1
N03 * PROPYL 	 > 1
63 + PROPYL 	 > I

7THJ
vmr
.08
.00
rmr
.00
.00
.OS
.00
.00
.06
.00
.08
.80
.00
.66
.00
HNOZ " IH20)
NO « I.Ofl-NOS ' .
H02
CO + 0.5B— HO?
Hos + i.oo-cn
CO « 1.00"H05 + 0.99 COD
CC03
N02 + 1.00 COO
PAN
PROCTL
0?
NO
N02
S02
N02
HCHO * 6.20 CCH02 »' O.«0 MOCMO,
CCHO + 0.20 HO2 + 0.70 CO
?,
1
u
(CJ» Z
T
(c>» u
.300E-I
;TTBOE-I

)3
< ?.?DOE-05>
< 1.300E-"D5>
)} ( H.DUnL*U^^
3?
[>«-
;onoE-02
< K.buoi--o»^

C I.9ZOE OZ>

"
1
1
6
1
1
3
7
U
14
**
03
;560C«"OS
73DOT-ir3
^OITOC.
.500E-
,4601-
.700C-
,820E«
.siof*
02
03
05
03
00
00»E«°<-1
.510E«00«Ex°(-l

< 1

f^S
< i
* J
< 3
< 7
,90E+03/T)< 7
;?f6rnrr5

. OOOE-OZ>
.500E-dS>
.500E-05>
.700E-03>
,8ZOE»00>
,677E-03>
(


r

*•
r

.90E»03/T)< 7.677E-OS>
U7|
«8|
a,,
50)
511
521
531
C3OHO?
C3OHO?
MO
»0
OH
TJOH02
C30H02
+
+
•
+
+
•*
+
H02 	
C30H02 	
N02 	
M02 	

NO 	
NO 	
— >
---> 2
.--•> 1
---> i
— > i
---> i
• --> i

.00
.00
.00
.00
.00
.00

CCHO *
MON02
HCHO *
C30H02
N02 »
N02 »

2.00

1.00

2.00
2.00

'HCHO » 270FT?05"

HM02

HCHO + 2.00 H02
CCHO
8.0CCE»03 < 8.000E»03>
5,00:E«02 < 5.000E«OZ>
lf5CCE*0« < 1.500E«OH>
*.uc:E»o3 < •..»noi.oS>
3.7<-;E»0'» < 3.7»OE + 0*>
fe.OCCC«03 < 6.000E+03>
6,c::E«03 < 6.000E»03>
(
(
c
vn 	
55)
561
^71
58)
59)
CC03
coo
coo
coo
coo
02
» H02
+ NO
+ NO
+ H02
* con
+ MO

	 > 1
	 > 1

	 > 1
	 > l

.00
.00

.50
.00

N02 » 1.00 MO
HCHO » 1.00 HN02

MO
HCHO + 1.00 H02
6«C-CC+03 ^6»
9«fct"E*03 ^9»
2.*::r4os < 2.
8.~t~C*G3 ^8.
e.:trr»03 < 8,
i.c:cE»oo < i.
OOOE+03>
600E*03>
tOOE*
Hoof*
03>
03>
OOOE«03>
OOOE*
oo>

- s\
O


-------
                ->  1.00 NO?    * 1.00 CC6T"
                ^>  1.30 R03
                                                                                                         1.190E*lB.E«»l-1.33t*0«»/T)<
          621
                       50?
 * r<03
'.• H8?
-> 1.00 S03    * 1.00 N02
-> 1.06 OhT
                                                             »  i.flo  SOS
Z.300C-03
1.3SOC»00
2.500t-OS>
                       COO
                               *  so?
                                                     SOS
                                                             »  1.00  MO
                                                                                                                                      i.S38C«flO>~
                                                                           7.50'OC-01
00
00
                                                                                                                                                     c
                                                                                                                                                     r

-------
                                                                              13 C C  N T ITKTTT) "N~s~
                                        is  »
    (

2  (
    <
N02
	 RTTO7' '
PROPTL
	 n
NO
«r
' "NN03
02
STHSIZ NVTSTP " OBCUP
r> 03 01C
h/TISNOJ KS03 TO
S',?i CCH02 HOC*O»
N03
C30H02
TtPP
N20S
MO
	 VJTER
MHM03
MON02
OH H02
503

1 .noooit -01
	 r7TTrnri)t-u«
1.00000E«00
— -----
	 i..'7TE,7r.rtrr
U.B610')E-0*
,.^S,1L-U1
1 u 0
1 .52017E-01
9.9977HE-01
150
llsisSStlos
•,.00'->ii--oi
200
1.03232E.01
	 'S.BlSibL-OS
9.78371E-01
250
ilvmnlos
.,.8,,«,oncf-1 0
2:-POOCE«DO
n.noof.
3.5* f« 7L-IIP
•
3:^iMiE-p8
2.1<>987E*00
'n.OOOO
8.70868E-P8
?. »oo-»t oo
20.0000
2.«f>770E*00
211.0000
^iltO^ElSs
2.l6607l>5ir
20^.0000
3.70379E-08
'llftwzc+So
20.0000
j ,W7»02E-09
2.*f>635E*00
•i.onnoor-05
C.OOOOOE»On
o.oiou
«tl37896E-ll

— 2U.OOOO 	
3.60070E-07
6.03096E-06
20.0000
2.573"*3E-10
3.8003 E Ob
20.0000
3.01285E-07
1.30780E-10
2.11U32E-06
20.0000
2.67302E-07
<».<»6937E-11
1.57861E-06
20.0000
2.522»6E-07
.06579E-11
20.0000
M.99967E-12
-*:!!!^--.!
	 n- o
\ .oopoor-ns
o.oroooEfOO
2 33
l!2VeiSElo5
*
	 1 	 36
2.32189E-07
2.M1239E-05
2 79
9.73990E-03
I. Z01 0
2 89
fc.02KXOE.10
8.'<5728E.06
2 90
3.23165F.03
fc.Jlfc'.tT-et
2 90
5.0»311E.10
5.52972C-06
2 90
1.97037E.10
7.9SS7ME-01*
5.3<»56<*E-06
1.7DOOOT-03
1 .OOOOOE-0!
O.OOOCOE»00
1.70000E-03
1.06037E-01
*
1.7DOOUC-03
8.229S5E-10
5.03536E-09
1.70000E-03
6.91<*60E.02
E 0
1.70000E-03
i*.67518E-10
2.2357'»E-09
1.70000E-03
3.3967SE-10
1.502<«OE-02
B. 75Z1-JE-10
1.70000E-03
3.17958E-10
3.2U731E-10
1.70000E-03
3.0B381E-10
3.87917E-03
1.96867E-10
**.OOOOoE + D2
1 .OOOOOE-03
0-OOOOOE400
<*.OOOOOE«02
5.00026E-01

5.03068E-12
1.S8812E-08
02t3.
1.

1.
2.

1.
1 .
1.

1.
3.
2 .
1.
2.
7.

5000BE"-U3 	
OPOOOE-06

50000E-03
3?3B3l-08 ~
10720E.09

•500UUE-03 	
5506, "»r- 09
TTjeir^rir—
50000E-03
60053E-10

5 000 0 c-03 ~
78301E-09
865 E""
50000E-03
U525Et.fl9
7915HE.12

50000E-05
86050E.10
0 0 E 1
SDOOOE-03
'"""S
	 	
(

c
     t

     c
o
c

-------
           "FINAL  i~fiTE6M*TCD R E A c T I ON unit's
          :  1.78TP-02To:   s.egeE-QH
  TH
                            IT:
                                   _ ,
                                    (jr^T

                                    TTT
                                                              ; '3.99-iC^Ol   ~^i   l.Bl^f-TVi    7 J"~%". T72E-09   8;  g.t3
-------




'



f

(
(
J
00
CTl
f

'
,




'


1


(
THIS IS A TtST RUN
4° * +l 4 ? 3 M * ^ 7
i.oo *
.
4
*
. *
A A A
O.*0 *
•
4
^
m
O.fcO +
•
4
•
•
o.to *
•
»
•
'
o.zo *
•
+
•
;
0.00 +
0.00 TIME, HINS
»=so2 LIN s.oooo: o.oonoi
DATE: 78072?
6910
»
t
•
+
•
A
+
,
4
•
*
4
• - 	 	 	
4
•

4
*
'
4
f
*
t
4-
• _ 	 	

4
•
•
+
<* oo o.oo








•

(
r

c
c

1
t











C

-------
                                                                                                                 O»TC::

                                                                                                                     » -      id
                  "1,00 +•__                                         _             _
                       .«            c.                         c                         c
                 "B.PO *"
   (
—      	irrwr
00 ,   __	
   •

   f
                  o.tn «
                  0.20 «
                       »B
                  d.oo
                                                                          . "TINS"   '     —- --    .   . ..      _..-    _    "nooo.bo     "  "       ~ <"*>
                                                                                            -P                                                    '
                        4-NO     L1M   1."000:    Q.OOOOtB=N02     LIN    1.0000;     0 . 0000 1 C-PROPTL LIN   1.0000:    0.00001
   c	

-------
                            -TOir~i5~Tr~TF^T-uuig—        -   -     -                  -          DATC:7eo72?

                                   1         !          3         It	' — -5	6          78          5         10
                T.10  *
  (    	     •	                                  .               	  r
—i           •     0.60  +	'   ~~~~  ~	+	
8g    	;	E	,	
                                                                                                                                                         c
                  O.«*0  *
                 0^20  +
                       • *
                 0.00  *C                                                              "            "  ~"  -   -    -     -   -       -  ^

 C                       6'. 6"6                                           TiMYi"M"i"NS    ~~               ~   - ---  -•-             " <40~ao~.'6o" ~ "              0-
     	a=HCHp    AUT   ii.iooit;     o.DOOoiBiCCHQ    am   o.sooo:	o. oooo i c =H2o;    AUT  ?.^7r-ip:	o.oooo i	

 C   	                                     	                                                                                  	 <

-------
                               TECHNICAL REPORT DATA
                         (Please read Instructions on the reverse before completing)
. REPORT NO.
 EPA-600/7-80-038
                                                     1. RECIPIENT'S ACCESSION NO.
. TITLE AND SUBTITLE
Photochemical Study of NOx Removal from
 Stack Gases
            . REPORT DATE
            March 1980
           6. PERFORMING ORGANIZATION CODE
. AUTHOR(S)

John R. Richards and Donald L.  Fox
                                                     8. PERFORMING ORGANIZATION REPORT NO.
. PERFORMING ORGANIZATION NAME AND ADDRESS
University of North Carolina
School of Public Health
Chapel Hill, North Carolina 27514
                                                     10. PROGRAM ELEMENT NO.
           INE623
           11. CONTRACT/GRANT NO.
           Grant No. R804740
 2. 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
                                                     Final; 1/77 - 3/79
           14. SPONSORING AGENCY CODE
             EPA/600/13
 5. SUPPLEMENTARY NOTES JERL-RTP project officer is Joseph A. McSorley, Mail Drop 63,
919/541-2745.
 . ABSTRACT The report gives results of an evaluation of the technical feasibility of a
photochemical pretreatment system for NOx control at coal-fired boilers. The ap-
proach utilizes reaction mechanisms similar to those responsible for photochemical
oxidant incidents. The reactions are initiated under  controlled conditions while
the pollutants are at high concentration and while the reaction products can be remo-
ved. Results indicate that,  under time- and light-limited conditions, it is possible to
quench the photochemical reactions at  the NO2  peak  and prior to the formation of
ozone, aerosols,  and other secondary  products. Energy and hydrocarbon require-
ments were estimated in a series of 159 experimental runs.  The optimal operating
conditions were identified as a NOx/C3H6 stoichiometric ratio of 1 and an NO2 photo-
lysis rate of 1  to 1. 5 reciprocal minutes (equivalent  to 3 to 5% station power).  These
conditions allowed for an 83% oxidation of NO with an exit concentration of NO less
than 100 ppm.  Variation of the energy input levels or the propylene injection rates
resulted in NO oxidation efficiencies of 10 to 99%.  Photochemical oxidation of  NO
was insensitive to SO2 concentration and CO2 concentration. The photochemical sys-
tem appears compatible with conditions resulting from combustion modifications
to suppress NOx generation".
 7.
                             KEY WORDS AND DOCUMENT ANALYSIS
                DESCRIPTORS
 Pollution            Combustion
 Photochemical Reactions
 Oxidation            Boilers
 Flue Gases          Propylene
 Nitrogen Oxides
 Coal
                                          b. IDENTIFIERS/OPEN ENDED TERMS
Pollution Control
Stationary Sources
                           COSATI Field/Group
13B
07E
07B,07C
21B

21D
                                    13A
13. DISTRIBUTION STATEMENT

 Release to Public
19. SECURITY CLASS (ThtsReport)
Unclassified
21. NO. OF PAGES
    200
20. SECURITY CLASS (Tkit pagej
Unclassified
23. PRICE
EPA Form 22ZO-1 (*-73)
                                        189

-------