UnitecLStlftes                   EPA-600/2-85-106
            Environmental Protection
            Agency                      September 1985
v>EPA     Research and
            Development
            EVALUATION OF THE EFFICIENCY

            OF INDUSTRIAL FLARES: FLARE HEAD

            DESIGN AND GAS COMPOSITION
            Prepared for
            Office of Air Quality Planning and Standards
            Prepared by
            Air and Energy Engineering Research
            Laboratory
            Research Triangle Park NC 27711

-------
                  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 ENVIRONMENTAL PROTECTION TECH-
 NOLOGY series. This series describes research performed to develop and dem-
 onstrate instrumentation, equipment, and methodology to repair or prevent en-
 vironmental degradation from point and non-point sources of pollution. This work
 provides the new or improved technology required for the control and treatment
 of pollution sources to meet environmental quality standards.
                        EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.

This document is available to the public through the National Technical Informa-
tion Service, Springfield. Virginia 22161.

-------
                                     EPA-600/2-85-106
                                     September 1985
         EVALUATION OF THE EFFICIENCY
   OF INDUSTRIAL FLARES:  FLARE HEAD DESIGN
              AND GAS COMPOSITION
                      by

        J. H.  Pohl  and N.  R.  Soelberg
ENERGY AND ENVIRONMENTAL RESEARCH CORPORATION
                   18 MASON
          IRVINE,  CALIFORNIA  92718
         EPA Contract No.  68-02-3661
  EPA Project Officer:      Bruce  A.  Tichenor
Air and Energy Engineering Research  Laboratory
       Hazardous Air Technology Branch
Research Triangle Park,  North Carolina   27711
                Prepared for:

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

-------
                                    ABSTRACT
     The U.S.  Environmental  Protection  Agency has  contracted  with  Energy  and
Environmental Research  Corporation  to  conduct  a research  program which  will
result in  quantification  of  emissions  from, and  efficiencies of,  industrial
flares.  The program is divided into four phases.  Phase I - Experimental  Design
and Phase  II -  Design  of  Test  Facilities  have been  reported in EPA  Report  No.
600/2-83-070.  Phase III  - Development  of  Test  Facilities  and  the  initial  work
in Phase IV - Data Collection has been reported  in EPA Report No.  600/2-84-095.
Further data collection during Phase IV is reported  herein.

     Initial results were  limited  to  tests conducted  burning  propane-nitrogen
mixtures on pipe  flames  without pilot light  stabilization.  The work  reported
here extends the  previous  results  to  other  gases  and flare head  designs,  and
includes a  limited  investigation of  the  influence  of pilot  flames  on  flare
performance.  The following results were obtained:

     o    Flames from nozzles less  than 1-1/2 inches* in  diameter  generally are
          not similar to flames from large nozzles.

     o    Flare head design can influence  the flame  stability curve.

     o    Combustion efficiency can be correlated with flame stability  based
          upon gas heating value for  the  pressure heads  and the coanda  steam-
          injected head.

     o    For the limited conditions tested, the flame stability and  combustion
          efficiency of the  air-assisted  head   correlated  with  the  momentum
*English units are generally  used  throughout  this  report.   Appendix E provides
conversion factors for English to Metric units.

                                        ii

-------
ratio of air to  fuel;  the  heating  value of the gas had only a  minor
influence.

Limited data on an air-assisted flare shows that use of a  pilot
light improves  flame stability.

The destruction efficiency  of different compounds  can  be correlated
with the flame  stability curve for each compound,  for  the  compounds
tested in this  program.

Stable flare flames and high (>98-99 percent)  combustion and destruc-
tion efficiencies  are  attained when the  flares  are operated within
operating envelopes  specific  to  each  flare   head  and  gas mixture
tested.  Operation beyond the  edge  of  the  operating envelope results
in rapid  flame de-stabilization  and  a  decrease  in  combustion  and
destruction efficiencies.
                              111

-------
                               TABLE OF CONTENTS



Section                                                                  Page


1.0



2.0
3.0





4.0





5.0





6.0
APPENDIX
APPENDIX
APPENDIX
APPENDIX
APPENDIX
Lib! Oh HliUKhb 	
LIST OF TABLES 	
INTRODUCTION 	
1.1 Review of Previous Work ... 	
1.2 Current EER Flare Program 	
1.3 References 	
SUMMARY AND CONCLUSIONS 	
FLARE AERODYNAMICS 	
3.1 Reynolds Number and Richardson Number 	
3.2 Flame Length 	
3.3 Flame Stability 	
3.4 Pseudo Adiabatic Flame Temperatures and Stability . .
3.5 References 	
COMMERCIAL FLARE HEAD TEST RESULTS 	
4.1 Coanda Steam-Injected Head D 	
4.2 Pressure-Assisted Head E 	
4.3 Pressure-Assisted Head F 	
4.4 Air-Assisted Head G 	
4.5 Commercial Head Summary 	
GAS COMPOSITION TEST RESULTS 	 	
5.1 Compound Selection 	
5.2 Compound Screening Tests 	
5.3 Gas Mixture Flare Tests 	
5.4 Flame Stability Correlations 	
5.5 References 	
FLARE NOX AND HYDROCARBON EMISSIONS 	
A. EPA FLARE TEST FACILITY AND TEST PROCEDURES 	
B. FLARE SCREENING FACILITY AND TEST PROCEDURES 	
C. DATA ANALYSIS 	
D. QUALITY ASSURANCE 	
E. CONVERSION FACTORS 	
V
viii
1-1
1-1
1-6
1-11
2-1
3-1
3-1
3-4
3-16
3-16
3-23
4-1
4-3
4-3
4-8
4-8
4-19
5-1
5-1
5-1
5-3
5-11
5-17
6-1
A-l
B-l
C-l
D-l
E-l

-------
                                LIST OF FIGURES

Figure                                                                   Page

 2-1      EPA Flare Test Facility (FTF) at EER	    2-2
 2-2      Flare Screening Facility (FSF)  	    2-3
 2-3      Region of flame stability for steam-injected and
          pressure-assisted heads D, E, and F	    2-14
 2-4      Maximum gas exit velocity for stability versus  air-
          assist to gas momentum ratio for the air-assisted
          head G, with and without pilot	    2-15
 2-5      Combustion efficiency vs. flame stability for
          steam-injected and pressure-assisted flare heads 	    2-16
 2-6      Combustion efficiency vs. air-assist to gas
          momentum ratio for commercial air-assisted head G   	    2-17
 2-7      Flame stability curves for pipe flare heads, 1/16  inch
          through 2 1/2 inch flares are without flame retention
          devices.  The 3, 6, and 12 inch heads were tested  with
          and without flame retention devices  	    2-18
 2-8      Region of flame stability for the 3 inch open pipe
          flare head burning selected relief gas mixtures  	    2-20
 2-9      Destruction efficiency of different gases  	    2-21
 3-1      Aerodynamic conditions of flares tested at EER
          compared to typical  commercial  flare conditions  	    3-2
 3-2      Comparison of flame lengths predicted by available
          studies with those observed in  this study	    3-5
 3-3      Correlation of flame length with heat release for
          1/2 inch to 31 inch flares	    3-6
 3-4      Correlation of flame length to  heat release for
          1/16 inch through 2 1/2 inch flares	    3-9
 3-5      Empirical  correlation of radiant heat loss from
          propane-nitrogen flames.   Richardson number=2.0 	    3-10
 3-6      Predicted flame length from modified Richardson
          number correlation compared to  observed flame
          length for previously tested 3, 6, and 12 inch
          flare heads	    3-12
 3-7      Predicted flame length from modified Richardson
          number correlation compared to  observed flame
          length for 1/2 through 2 1/2 inch flare heads	    3-13
 3-8      Predicted flame length from modified Richardson
          number correlation compared to  observed flame
          length for 1/16 through 1/4 inch flare heads	    3-14
 3-9      Predicted flame length from modified Richardson
          number correlation compared to  observed flame
          length for 1 1/2 through 12 inch commercial  heads	    3-15
 3-10     Flame stability curves for pipe flare heads	    3-17
 3-11     Maximum relief gas exit velocity vs. flame temperature
          for previously tested 3 through 12 inch flare heads  ....    3-19
 3-12     Maximum relief gas exit velocity vs. flame temperature
          for commercial  heads D through  G	    3-20

-------
                          LIST OF FIGURES (continued)

Figure                                                                   Page

 3-13     Relation between flame stability and pseudo
          adiabatic flame temperature 	   3-21
 4-1      Flame stability curve for a commercial  12 inch coanda
          steam-injected head D.  Steam flowrate  at 140 Ib/hr . ;  .  .  .   4-4
 4-2      Relation of flame stability to combustion efficiency
          for the commercial  12 inch coanda steam-injected
          flare head D.  Steam flowrate at 140 Ib/hr	4-5
 4-3      Region of flame stability for 1.5 inch  commercial
          pressure-assisted head E	4-6
 4-4      Relation of flame stability to combustion efficiency
          for commercial 1.5 inch pressure-assisted head E	4-7
 4-5      Region of flame stability for 3.8 inch  commercial
          pressure head F	4-9
 4-6      Relation of flame stability to combustion efficiency
          for a commercial 3.8 inch pressure-assisted  head  F  	   4-10
 4-7      Region of flame stability for a 1.5 inch commercial
          air-assisted flare G without a pilot flame  	   4-11
 4-8      Region of flame stability for 1.5 inch  commercial
          air-assisted flare G with pilot flame	4-12
 4-9      Region of flame stability for 1.5 inch  commercial
          air-assisted head G, with no air-assist	4-13
 4-10     Maximum gas exit velocity for stability versus air-
          assist to gas momentum ratio for the air-assisted
          head G, with and without pilot	4-15
 4-11     Combustion efficiency vs. air-assist to gas  momentum
          ratio for commercial air-assisted head  G	4-17
 4-12     Destruction efficiency of incompletely  combusted
          products from air-assisted head G	4-18
 4-13     Region of flame stability for steam and pressure-
          assisted heads D, E, and F  .	4-20
 4-14     Combustion efficiency vs. flame stability for steam-
          injected and pressure-assisted flare heads  	   4-21
 5-1      Region of flame stability for the 3 inch open pipe
          flare head burning selected relief gas  mixtures 	   5-5
 5-2      Destruction efficiency of different gases 	   5-8
 5-3      Hydrocarbon combustion efficiency of gas mixtures  	   5-9
 5-4      Pilot-scale FTF destruction efficiency  results
          compared to lab-scale FSF destruction efficiency
          results.  All results at operating conditions near
          the stability limit.  FTF nozzle size = 3 inches.
          FSF nozzle size = 0.042 inches	5-10
 5-5      Calculated adiabatic flame temperature  vs. limiting
          stable gas exit velocity for different  gas mixtures 	   5-13

-------
                        LIST OF FIGURES  (continued)

Figure                                                                 Page

5-6(a)    Nozzle velocity vs Experimental  Index  for  data  from
          Noble, et al	    5-15
5-6(b)    Nozzle exit velocity vs Experimental Index for  gas
          mixture tests.                                               5-15
6-1       NOX concentration (air-free basis)  at  the  plume
          center!ine from pilot-scale flares	    6-2
6-2       NC>2 emissions from pilot-scale  flares	    6-4
                                      vii

-------
LIST OF TABLES  .
Table
1-1

2-1

2-2

2-3

2-4

2-5

2-6

2-7

4-1

5-1

5-2

6-1

6-2


SUMMARY OF PREVIOUS FLARE COMBUSTION EFFICIENCY
STUDIES 	
COMMERCIAL 12 INCH DIAMETER COANDA STEAM- INJECTED
HEAD D TEST RESULTS. STEAM FLOWRATE = 140 LB/HR. . . .
COMMERCIAL 1.5 INCH DIAMETER PRESSURE-ASSISTED
HEAD E TEST RESULTS 	
COMMERCIAL 3.8 INCH DIAMETER PRESSURE-ASSISTED
HEAD F TEST RESULTS 	
COMMERCIAL 1.5 INCH DIAMETER AIR-ASSISTED HEAD G
TEST RESULTS 	
3 INCH DIAMETER OPEN PIPE FLARE (WITH PILOT) TEST
RESULTS 	
RESULTS OF SCREENING TESTS ON FLARE SCREENING
FACILITY 	
GAS MIXTURE COMBUSTION AND DESTRUCTION EFFICIENCY
TEST RESULTS: 3 INCH FLARE, NO PILOT 	
COMMERCIAL HEADS SELECTED FOR COMBUSTION EFFICIENCY
TESTS 	
RESULTS OF SCREENING TESTS ON FLARE SCREENING
FACILITY 	
PHYSICAL PROPERTIES (1 ATM, 6QOF) OF THE COMPOUNDS
TESTED USING THE FLARE TEST FACILITY 	
GC-MS ANALYSIS OF PLUME SAMPLE FROM LABORATORY-SCALE
TESTS 	
GC-MS ANALYSIS OF PLUME CENTERLINE SAMPLES FROM GAS
MIXTURE TESTS, USING A 3 INCH OPEN PIPE FLARE 	
Page

1-2

2-4

2-5

2-6

2-7

2-9

2-10

2-12

. 4-2

5-2

5-6

6-6

6-7
      Vlll

-------
1.0       INTRODUCTION

     Industrial  flares are commonly  used to safely and economically destroy
large amounts  of  industrial waste gases.  Since most of the gas  flared in the
United  States is from leaks,  purges, and emergency vents,  the amounts and
compositions of flared gases vary widely and are difficult to  measure.  Flare
emissions  are  also difficult to measure.   Most flares are elevated to
decrease  noise  radiation  and  combustion products at ground  level.  Probe
collection  of plume material  in such situations is impractical.  Remote
sensing  of  flare  emissions  is an  alternative  to direct sampling,  but
instrumentation and techniques for this purpose are still  undeveloped.

     Evaluation  and  control of  industrial  flare emissions  requires pilot-
scale research  with direct sampling  of flare emissions.   Flare research has
been conducted  at Energy and Environmental Research Corporation (EER) since
1980.   A pilot-scale flare  test facility was  constructed  for  the  U.S.
Environmental  Protection Agency in 1982.  This research has been sponsored by
the U.S.  E.P.A.  as part of an  effort to provide data upon which to base new
regulations  for industrial flaring practices (1.1, 1.2).

1.1       Review  of Previous Work

     Several  previous  studies  have  evaluated flare  emissions  by testing
combustion efficiency of small-scale  or pilot-scale flames (1.3).  A primary
finding of these studies was that flare flame combustion  efficiencies can be
very high, exceeding 98 percent, but that under certain operating conditions,
such as  excessive steam injection, low efficiency can result (1.4).  Pohl, et
al . (1.3) and Keller and Noble  (1.5)  have also reported that  when a flame is
stable  (i.e.,  not near blow-out conditions) efficient combustion is achieved.
However, flares operating with unstable flames tend to be  inefficient.

     Test  conditions  and measured combustion efficiencies for recent flare
studies are  summarized  in  Table 1-1.  A range of flare  sizes, designs, and
operating conditions have  been examined.   The  results show combustion
efficiencies  ranging between  55 -  100 percent.   The measured combustion
                                    1-1

-------
                                                 Table 1-1.
                            SUMMARY OF PREVIOUS FLARE COMBUSTION EFFICIENCY  STUDIES
STUDY
Palmer (1.6)
Lee & Whipple (1.7)
Siegel (1.8)
Howes, et al (1.9)
Howes, et al (1.9)
McDaniel (1.10)
Keller and Noble (1.5)
McDaniel (1.10)
Keller and Noble (1.5)
Pohl, et al (1.3)
DATE
1972
1981
1980
1981
1981
1983
1983
1984
FLARE
SIZE
(in.)
0.5
2.0
27.
6
3at4
8
6(0
3-12
DESIGN
Experimental Nozzle
Holes In 2" Cap
Commercial Flaregas
Coanda FS-6
Commercial A1r
Assist. Zink LH
Commercial H. P. Zink
LRGO
Commercial Zink
STF-S-8
Commercial Air
Assist.
Zink STF-LH-457-5
Open pipe and
commercial
VELOCITY
(ft/sec)
50 - 250
1.8
0.7 - 16
40 - 60
Near Sonic
(estimate)
0.03 - 62
1.4 - 218
0.2 - 420
GAS FLARED
Ethylene
Propane
Refinery Gas*3'
Propane
Natural Gas
Propylene/Ni trogen
(e)
Propvlene/Ni trogen
Propane/Ni trogen ' '
MEASURED
EFFICIENCY
ft)
> 97.8
96 - 100
97 - > 99
92 - 100
> 99
67 - 100
55 - 100
90 - 99.9
i
IV)
              hydrogen plus light hydrocarbons
       *  'Three spiders, each with an open area of 1.3 1n
       ^Supplied through spiders; high Btu gas through area of 5.30  in2 and  low  Btu  gas  through 11.24 in
         Heating  value  varied from 209 to 2183 Btu/scf
       *e'Heating  value  varied from 83 to 2183 Btu/scf
       Seating  value  varied from 291-2350 Btu/scf

-------
efficiencies  show that  flare  flame stability and combustion efficiency  may
vary,  depending on flare  head size,  design, gas composition, and operating
conditions.

     Accurate measurement of  flare emissions and combustion efficiency is
difficult  even  on a pilot-scale facility.   Problems encountered by previous
researchers  include:

     •    Inability to close mass balances

     •    Inability to measure soot emissions

     •    Sampling only on the plume center!ine

     •    Flare  flame fluctuations due to turbulence and/or wind

     The Flare  Test  Facility at EER has been designed and built in order to
minimize  such problems.   The  following  procedures have been developed to
verify the  accuracy of combustion efficiency measurements:

     t    Material balance closure was verified using a hood  to capture  the
          entire flare plume for small  flames, and by using S02 as a  tracer
          for  large flames.

     •    Soot concentration was measured for all  tests.

     •    The  average concentration of completely and incompletely  burned
          combustion species from  the  flare flame was  determined for  the
          entire  plume  by  (1) using a hood to  completely capture small
          flames, and  (2)  simultaneous  sampling  at five radial positions in
          the  plume for large flames.  These local values were combined with
          velocities calculated from jet theory  and integrated to calculate
          overall  combustion efficiency.
                                    1-3

-------
     •   The effects of  flare flame fluctuations  were limited by mixing a
         sample  taken over a 20 minute  time  period.   This time span was
         experimentally  determined to be sufficient to average flame
         fluctuations.

     Using  the  Flare  Test Facility, EER has  developed a data base of
combustion efficiency for a variety of  flare heads and operating conditions.
Combustion  efficiency and  flare scaling parameters were  studied using 3, 6,
and 12 inch open  pipe flares with and without flame  stabilization devices.
Scaling parameters  investigated included exit  velocity, residence time,
Reynolds number, and Richardson number. Flare flame aerodynamics, including
lift-off and flame length, were also studied.

     Due  to industrial  interest in higher  flare exit velocities, testing was
conducted  at exit velocities up to 420  ft/sec.   The  same parameters of
scaling, combustion efficiency, and aerodynamics were evaluated at these high
velocities, using a 3 inch open pipe flare both  with  and without flame
retention devices.

     The  previous trials were  largely limited to propane/nitrogen mixtures
burned on  open and commercial pipe flares  without pilot flame stabilization.
Results from these trials established:

     •   Flame  length can  be estimated  using Richardson number  and an
         estimate of the flame temperature.

     •   Flare  burn  gases with combustion efficiencies greater than
         98 percent unless  operated within 30 percent  of  the blow-off limit.

     t   Soot  contributes  less than  0.5  percent  to the  unburned
         hydrocarbons.

     «   A single probe can yield good  estimates of  overall combustion
         efficiency.
                                   1-4

-------
     o    For the range of flares and conditions studied,  flame  structure  and
          combustion efficiency  did  not  depend  on   flare   size   or   design.

     In order to  extend  the basic  open-pipe  flare test  results to  industrial
application, three commercial  12  inch  heads  were also tested.  The  purpose of
these tests was to  determine the dependence  of combustion efficiency on  flare
head type and design.  Combustion efficiency can vary, depending on  commercial
flare head type and  design,  because flare heads are often tailored  to  achieve
specific results  for  specific gas  mixtures  and operating conditions,  and may
not operate well under other conditions.

     The EER flare  test  program  has established a data base  for evaluation of
future flare test  results.   In general, it  has been  demonstrated  that when  a
flare flame is operated under  stable conditions,  combustion  efficiency  greater
than 98 percent  is  attained.  For  propane-nitrogen mixtures flare  using  pipe
flares without  pilots,  a relationship  has been demonstrated  between  the gas
heating value  and the  stability limit,  and  between  the  stability limit and
combustion efficiency.   A similar  relationship between  heating  value,  flame
stability, and  combustion efficiency  was observed  for  the three  commercial
12 inch pipe flares tested.
     Throughout the  EER  flare  research  program, advice  and  consultation  was
provided by a Technical Advisory  Committee.   Committee  members  included  repre-
sentatives from EER,  EPA,  California Air Resources  Board,  flare  manufacturers
(Peabody Engineering,  McGill,  Inc.,  John Zink,  and Flaregas  Corporation)  and
industrial flare users (Exxon Chemical  Company, Exxon R&E, Union Carbide, Getty
Refining and Marketing Co., Chevron USA, and Dow Chemical  Company).   The  indus-
trial users included  representatives of  the  Chemical  Manufacturers  Association
(CMA) and the American Petroleum  Institute  (API).   This committee met through-
out the  program  to  review  and critique  test plans, ensure  relevance  of  the
study, and facilitate efficient information  transfer.
                                       1-5

-------
1.2       Current EER Flare Program

     The previous  EER  studies  have  established  accurate  flare  combustion
efficiency test methodology, developed a pilot-scale test facility,  and  estab-
lished a data base of  combustion  efficiency test  results  for  3 through  12  inch
diameter flare heads burning  propane-nitrogen  mixtures  on pipe flares  without
pilot light stabilization (1.3, 1.11, and 1.12).

     The current  flare test  program is  an  extension  of  the previous work.
Objectives were based  on  recommendations of  the  Flare Advisory Panel and  were
designed to extend  the data  base to cover  a  wider  range  of  flare  operating
and design conditions.

1.2.1     Objectives

     The objectives of the flare test program reported here  are to:

     o    Evaluate the effects of flare  head  type on flare combustion
          efficiency.

     o    Evaluate effects of relief gas composition on  flare  combustion
          and destruction efficiency.

1.2.2     Approach

     The overall  objective  of  this  program  is  to  assess  pollutant  emissions
from industrial   flares.   Direct  measurements  on full-scale   operating  flares
are difficult.  Therefore, direct measurements  were  made on  pilot-scale
flares in  order  to  measure  pollutant  emissions,  combustion   and  destruction
efficiency, and scaling criteria.
                                      1-6

-------
     The program was divided into four major  tasks:

     •    Task  1 - Evaluation of combustion efficiency  from different flare
          head  types

     •    Task  2 - Identification of representative, potentially difficult to
          destroy gas compounds

     •    Task  3  - Evaluation of combustion  and destruction efficiency of
          selected relief gas mixtures.

     t    Task  4 - Data analysis and reporting

     In Task 1, EER  obtained  from flare suppliers  the  following four
commercial  heads:

     •    1 air-assisted flare head

     •    2 pressure-assisted flare heads

     •    1 coanda steam-injected flare head

     Each  of these  heads  was tested  on the EER pilot-scale Flare Test
Facility  (FTF).   Flame stability and combustion efficiency were measured as
functions  of the following operating conditions:

     t    Relief gas and exit velocity

     t    Relief gas heating value

     •    Steam injection  flow rate  (for the coanda steam-injected flare
          head)

     •    Air assist velocity (for the air-assisted flare head)
                                    1-7

-------
     0    Relief  gas pressure (for the  pressure-assisted flare head)

     •    With  and without pilot flame  (for  the air-assisted flare head)

     The relief  gas for these tests  was  propane, mixed with nitrogen  to vary
the heating  value.  Natural gas was used  for the pilot flame.

     Tasks  2 and  3  were  designed to  measure  the effects  of  relief gas
composition on flare pollutant emissions.   A  wide variety  of industrial
compounds are  frequently flared in  the  United States.  Most often, they are
flared  in mixtures containing several  compounds.  Each mixture  may  exhibit
somewhat  different flaring characteristics.  Pilot or large-scale testing of
every conceivable relief gas mixture would be expensive and unending.

     A sensible  approach is to test compounds  in a laboratory-scale facility
which are representatives of classes  of compounds industrially flared.  The
small scale  laboratory facility produces  flames which are not aerodynamically
similar  to those  produced on the pilot-scale flame or in industrial  practice,
because a  1/16 inch nozzle was used in the  screening  studies.   For  a
commercial 12 inch diameter flare head, Reynolds numbers may range from 4,000
to 40,000,  Richardson numbers range from 0.1 to 1,000, and buoyant forces
often dominate over  inertial  forces.   For a 1/16 inch  diameter nozzle,
Reynolds numbers  are 102 to 104, Richardson  numbers are typically 1 x  10~4 or
smaller,  and inertial forces dominate.   Even  though lab-scale tests  using  a
1/16  inch diameter nozzle are aerodynamically dissimilar  to  large-scale
flares,  such tests can be used to economically screen compounds to determine
comparative potential for successful  destruction of  these compounds by
flaring.  Compounds  which demonstrate flaring difficulties in the Flare
Screening Facility  (FSF) are candidates for  testing on the FTF.

     Upon recommendations of the Flare Advisory Committee, twenty-one of the
most  commonly  flared, potentially hazardous, or difficult to flare compounds
were selected for laboratory-scale testing on the  (FSF).  These compounds are
representative  of:
                                    1-8

-------
     •    Sulfur compounds

     •    Nitrogen compounds

     •    Chlorinated compounds

     t    Oxygenated compounds

     •    Aliphatic hydrocarbons

     t    Aromatic hydrocarbons

     t    Compounds with low  heating value

     Of the twenty-one  compounds screened,  six were selected as  candidates
for  testing  on  the  FTP.  Selection  criteria included  low destruction
efficiency,  poor  ignitabi1ity,  and  high soot production.   In  addition,
hydrogen  sulfide,  although not  tested in the  screening facility, was also
selected due  to its industrial  importance and toxicity.

     Four of the seven  compounds selected  during the screening  tests were
tested on  the FTF.  Hydrogen  sulfide and ammonia  were each tested in mixtures
with propane  and nitrogen. Ethylene oxide and 1,3-butadiene were each  tested
diluted with  nitrogen to vary the  heating value.  Flame stability,  combustion
and destruction  efficiency, soot production, and by-product formation from
incomplete combustion were measured  for each  compound.   All  tests were
conducted  using  the  3 inch   open  pipe  flare, without  pilot  flame
stabilization.

     Sample procedures used during Tasks 1 through 3 were consistent with the
protocols  of previous EER studies  (1.3, 1.11).   In the  FTF,  flare plume
samples were  taken at five radial  positions above the flame.  These local
samples were  analyzed for 02, CO,  C02,  total  hydrocarbons, NOx, and soot
concentration.   Where applicable,  the samples were also analyzed for H2S,
S02, and  NH3  concentration.  Limited numbers of samples were also adsorbed
                                  1-9

-------
using  charcoal -Tenax traps.   Gases were .desorbed  from these traps  and
analyzed by  GC-MS for combustion  by-products.

     Sampling in the FSF was easier.   In this facility,  the 1/16 inch nozzle
was enclosed  within a reaction chamber, which isolated  the flame from the
external environment.   Tests  verified that  the  flames behaved  as if
discharged  into air because the  flame  was very small  relative to the size of
the chamber.   Sampling of  the  well-mixed products  at the reactor outlet
required only  one probe and ensured complete mass balance closure.

     Sample analysis  was conducted  during tests on  both  the FSF and FTF to
evaluate  air  dilution, mass balances,  combustion efficiency, and destruction
efficiency.   Sulfur dioxide was injected during some  of the pilot tests and
used as  a tracer for mass  balances.  Mass  balances on  the FTF were more
difficult because of product loss, air dilution in the  large exposed flame,
and  plume  concentration  gradients.  Local  mass  balances were  used to
accurately  evaluate local  mass fluxes, local  combustion efficiency, and
destruction efficiency.   Local  mass fluxes were radially integrated to
calculate overall combustion and  destruction efficiencies.

     In Task 4, flare combustion  and destruction efficiencies were correlated
to flame  stability.  Flame stability was correlated to  relief gas heating
value  and exit velocity for the  gas mixture tests and  for  all the commercial
heads except for  the air-assisted head. Flame stability  for  the air-assisted
head was  related to air-fuel stoichiometry and momentum  ratios.  Results for
the air-assisted  head showed that the  effects  of relief gas heating value and
exit velocity on flame stability were minimal, compared  to the effect of the
air-assist stream to fuel stream  momentum ratio.

     In order  to  evaluate the dependence of flare flame combustion efficiency
on flame  stability and operating conditions,  it was necessary to investigate
flare  operating conditions which result in both efficient and inefficient
combustion  in order to  define  the region of  efficient operation  and to
determine parameters which are critical to flare performance.  Thus, many of
                                  1-10

-------
the test  cases represent  conditions  outside the normal range of commercial

flare  operations.



1.3       References
1.1    Davis, B.  C.,  "U.S. EPA's Flare Policy:   Update  and Review", Chemical
       Engineering Progress, April 1983.

1.2    Davis, B.  C.,  "Flares-An Update of Environmental  Regulatory Policy",
       AIChE National  Meeting, Philadelphia, PA, August 19-22, 1984.

1.3    Pohl, J.  H., R.  Payne, and J. Lee, "Evaluation  of the Efficiency of
       Industrial  Flares: Test  Results", EPA  Report  No. 600/2-84-095,  May
       1984.

1.4    Dubnowski, J.  J.  and B. C. Davis, "Flaring Combustion Efficiency:  A
       Review of  the  State of Current  Knowledge,  "The  76th Annual  Air
       Pollution  Control Association Meeting,  Atlanta,  GA, 1983.

1.5    Keller, N. and R. Noble, "RACT for VOC  - A Burning Issue",  Pollution
       Engineering, July 1983.

1.6    Palmer, P. A., "A Tracer Technique for  Determining Efficiency  of an
       Elevated  Flare," E.  I.  du  Ponte de Nemours  and Co., Wilmington,  DE,
       1972.

1.7    Lee, K. C. and G. M. Whipple, "Waste Gas Hydrocarbon  Combusiton in a
       Flare, "Union  Carbide Corporation, South Charleston, WV, 1981.

1.8    Siege! , K.  D., "Degree of  Conversion  of  Flare Gas in Refinery High
       Flares, "Ph.D. Dissertation, University of Karlsruhe  (Federal Republic
       of Germany), February, 1980.

1.9    Howes, J.  E.,  T.  E. Hill, R. N. Smith,  G. R. Ward, and W. F.  Herget,
       "Development of Flare  Emission Measurement Methodology, Draft Report,"
       EPA  Contract  No. 68-02-2682,  U.S. Environmental Protection Agency,
       1981  (Draft Report).

1.10   McDaniel,  M.,  "Flare Efficiency Study," EPA Report No. 600/2-83-052,
       July  1983.

1.11   Joseph, D., J. Lee, C. McKinnon, R.  Payne, and J. Pohl, "Evaluation of
       the   Efficiency  of Industrial Flares:  Background-Experimental Design-
       Facility", EPA Report  No. 600/2-83-070, August 1983.

1.12   Pohl,  J.  H.,  J. Lee, R.  Payne and B.  Tichenor,  "The  Combustion
       Efficiency of  Flare Flames", 77th Annual Meeting and  Exhibition  of the
       Air Pollution  Control  Association, San Francisco, CA., June  1984.
                                  1-11

-------
2.0       SUMMARY AND CONCLUSIONS

     The  objective  of this phase  of "Evaluation  of the Efficiency  of
Industrial  Flares" was to determine the influence  of additional flare head
designs  and gas composition  on  the  stability and structure, destruction
efficiency  (DE), and combustion efficiency (CE)  of flare flames.  This  phase
extended  the  previous programs  which  reviewed the literature available  on
flare performance, designed  and  built a flare test  facility, and determined
the combustion  efficiency of pipe  flares firing mixtures of propane and
nitrogen  in  the  absence of a pilot flame.

     The  current program extends previous work  to include different classes
of commercial  flare heads, and types of relief  gas  mixtures burned on  small
pipe  flares.  The flare head types include:

     •    Coanda steam-injected flare head
     •    Two pressure-assisted flare heads
     •    Air-assisted flare head  without flame retention devices

Each of  these  flare heads was evaluated on the  Flare  Test Facility shown  in
Figure 2-1.

     Twenty-one different gases were  selected and  screened for potential
combustion  problems on the  Flare Screening Facility (FSF) shown in Figure
2-2,  with  a  1/16  inch nozzle  (ID  =  0.042 inches).  The  combustion and
destruction  efficiencies  of  four of these gases  were measured on the  Flare
Test  Facility  (FTF).

     The  data collected  on the  FTF  for  the  different  flare heads  is
summarized  in Tables 2-1  through  2-4.  The air-assisted head was tested both
with and  without  a pilot  flame, as  shown in Table 2-4.    Limited tests  to
evaluate  the  effects of  pilot flames  were also  conducted  on a 3 inch open
pipe  flare,  without flame  retention devices.  These test results are shown  in
Table 2-5.   The results  of the FSF gas mixture  screening tests are shown  in
Table 2-6,  and  the results  from the  FTF gas mixture tests are shown  in
                                  2-1

-------
Adjustable Rake for
Sample Prodes, Filters,  Driers
Tenax Traps,  and Bubblers
for H,S, S02  & NH3
                                                                   Support StruclUTf
                                                                   Sample Lines
                                                                    Sample Collection S Analyj
                                                                     for 02	
CO
C02
Hydrocarbon
soz

^—^ u



Flira <]• Lin. '


S,.«n^

^\J(
Air


,





K








k


X^"




Conlral
And
Ml. Hi



/
Flow
Connol



/n

>
•4-


•4-

*UL
/7


*^
^
J
rf* ^
^)-^ Auxll.Gas TankQ
^X "
^_^ P.OPIM Tmll Q


^M N"""« T«"- 0
-^



Sttim Gtn«r*lor s&\
£ 	 ^
	 	 , ***• ^'Ipply P»« '
^
                      Figure 2-1.    EPA flare test facility  (FTF)  at  EER.
                                                       2-2

-------
     MIXING CHAMBER
CONTINUOUS SAMPLING
TENAX SAMPLING
                                 /16-1/8 IN, NOZZLE
                             SCREEN
                            GLASS BEADS
                        AIR FAN
Figure 2-2.   Flare screening facility  (FSF).
                      2-3

-------
                                                            Table  2-1
                                COMMERCIAL  12 INCH DIAMETER   COANDA  STEAM-INJECTED HEAD D
                                         TEST RESULTS.   STEAM  FLOWRATE = 140 LB./HR
Test
No.
200A
200B
200C
2000
200 E
200F
201
202
203
204
205
206
Actual
Exit .
Velocity'
(ft/sec)
0.17
0.20
4.0
4.3
9.3
9.3
0.20
4.38
9.19
0.20
3.95
9.90
SPropane
In
Nitrogen
14.43
13.0
12.1
11.1
13.1
13.0
13.0
11.7
14.0
17.9
18.1
18.1
Low
Htg Val
(Btu/ft3)
338
305
284
261
308
305
306
275
329
421
425
425
Steam
Ratio
(Ib steam/
1 b fuel )
3.7 '
3.3
0.64
0.15
0.068
0.068
Probe
Ht3
(ft)
Observations
Wind
Speed
(mph)
Flame 2
Length
(ft)
Lift
Off
(1n)
Color
Smoke
Sound
Comb
Eff
(%)
Hydro-
Carbon
Dest
Eff U)


Stability Curve Tests


3.3
0.15
0.069
3.1
0.16
0.081
5
5
15
4
10
22
8
7
6
3
8
5
1.5
6
12
3
7
22
0
0
0
0
0
0
dim orange
dim orange
dim orange
dim orange
dim orange
dim orange
none
none
none
none
none
none
none
low roar
jet
none
rumbl e
roar
99.3
99.4
97.9
99.9
99.7
99.6
99.7
99.8
98.6
100.0
100.0
99-. 9
INi
I
    1    Based on design capacity for a  12 inch flare  head.   Open area at base of
        cone = 113 in2.

    2    Based on light diffraction through flame envelope for invisible flames.

    3    Height above flare tip.

-------
                                                        Table 2-2
                        COMMERCIAL  1.5 INCH  DIAMETER1  PRESSURE-ASSISTED  HEAD E.  TEST RESULTS
Test
No.
207
208A
208B
209
210
211
212

214
216
217
218
219
220
221
Actual
Exit
Velocity1
(ft/ sec)
14.3
112
94.7
472
78.5
12.4
95.9

238
384
14.2
470
109
761
907
^Propane
1n
Nitrogen
15.8
20.2
23.9
23.9
26.0
18.1
21.9

48.1
29.6
23.7
35.2
28.4
28.1
36.6
Low
Htg Val
(Btu/ft3)
371
474
562
562
612
426
514

1130
696
557
828
668
661
870
dP
Across
Head
(psig)
0
0
0
2
0 >
0
0

NR2
NR
0
3
0
7
10
Probe
Ht3
(ft)
Observations
Wind
Speed
(mph)
Flame
Length
(ft)
Lift
Off
(in)
Color
Smoke
Sound
Comb
Eff
(*)
Hydro
Carbon
Dest
Eff (%)


Stability Curve Tests












6
7

12
16
6
16
7
20
30
6
7

3.5
6
2.5
5
5
5
6
2
2.5

12
12
3
14
5
18
20
2
2

1
4
0
3
0
4
3
dim orange
yellow-purple
base
yellow-blue
yellow-blue
yel 1 ow
blue base
orange-blue
yellow-blue
yellow-blue
none
none

none
none
none
none
none
none
none
none
dull rumble-
roar
roar
roar
none
jet
low rumble
loud roar
load roar
97.0
94.2

99.3
98.3
99.1
96.2
98.8
97.7
99.4
98.4
95.1

99.8
98.9.
99.6
98.0
99.2
98.5
99.7
1    Based on total open area of exit ports of 1.77
2    NR = Not recorded.
3    Above flare tip.

-------
                                                                 Table  2-3
                             COMMERCIAL 3.8 INCH  DIAMETER1  PRESSURE-ASSISTED HEAD F TEST RESULTS


Test
No.
269
270
271
272
273
274
275
276
277
278
279
280

Actual
Exit
Velocity1
(ft/sec)
33.9
119
112
158
11.1
42.3
52.2
5.55
4.39
157
139
25.3


%Propane
1n
Nitrogen
5.93
7.46
7.49
7.20
5.18
7.50
9.08
9.10
9.92
13.8
18.5
10.7


Low
Htg Val
(Btu/ft3)
139
175
176
169
122
176
213
214
233
323
453
251

A P
Across
Head
(pslg)
NR \


Probe
Ht3
(ft)

Observations

Wind
Speed
(mph)

Flame 9
Length^
(ft)

Lift
Off
(in)

Color
:=:;::ss:3:s = st;5:33 — :=::
Smoke
sss: s:s :s sc :
Sound



Comb
Eff
(I)


H wirn
lljrUl U
Carbon
Dest
Eff 1%)

NR 1
NR \ Stability Curve Tests
NR I
NR '
0
0
0
0
4
4
1
13
15
8
5
25
25
15
4.5
3
52
6
2
7
0.5
4.5
4.5
2
2
17
18
8.5
0
0
0
0
0
0
0
light blue
dim blue
transparent
transparent
transparent
yellow-blue
yellow
none
none
none
none
none
none
none
faint hiss
faint hiss
none
none
rumbl e
rumble
faint hiss
95.8
96.5
97.6
85.5
95.0
99.5
99.4
97.2
97. b
98.5
86.4
95.4
100.0
99.8
ro
i
cr>
      1    Based on total  open area of exit ports  of 11.3  in^.
      2    Based on light  diffraction through flame envelope for invisible flame.
      3    Height above flare tip
      4    Not Recorded.

-------
                                                                    Table  2-4



                                  COMMERCIAL 1.5  INCH  DIAMETER1 AIR-ASSISTED HEAD G  TEST RESULTS
Test
No.
222
224
225
226
227
227A
228
229
230
231A
232A
231B
23211
233
234
235
236
237
238
239
240
241
Actual
Fuel Exit
Velocity1
(ft/sec)
0.73
28.7
46.8
263
12.2
13.8
1.15
4.21
90.0
60.1
64.0
44.0
62.7
380
03
368
591
469
464
301
4.59
57.7
Propane
in
Nitrogen
m
100
8.03
13.0
40.3
9.40
8.57
100
27.7
21.4
32.2
30.0
43.9
30.7
38.8
48.7
38.9
29.3
54.2
55.8
47.8
20.1
20.2
Low^
lltg Val
(Btu/ft3)
2350
189
305
947
221
201
2350
650
505
756
706
1032
722
912
1144
913
690
1273
1311
1123
473
475
Air-Assist
Flovirate
(SCFM)
1030
0
0
0
0
0
705
412
0
784
1110
780
1100
0
780
1110
0
1100
544
542
529
535
Air-Assist
Exit
Velocity
(ft/sec)
112
0
0
0
0
0
76.6
44.8
0
85.1
121
84.7
119
0
84.7
120
0
120
59.1
58.9
57.4
58.1
Air-
Assist
S.R.3
79.0
0
0
0
0
0
34.3
19.7
0
2.26
3.21
2.25
3.18
0
0.294
0.434
0
0.242
0.117
0.210
31.9
2.57
(pv)air4
(pv)fuel
99.4
0
0
0
0
0
43.5
9.38
0
1.22
1.65
1.57
1.66
0
0.224
0.274
0
0.199
0.0987
0.157
11.6
0.924
Observations Hydro- pilot
Wind Flame Lift 1 Comb. Carbon Flow-
Speed Length Off Eff. nest. rate
(mph) (ft) (in) Color Smoke | Sound (%) Eff. (%) (SCFM)
N




















;
0
0
0
0
0
0
0
0
0
0
0
> Stability Curve Tests g
0
0
0
0
0
0
0
4'
0
0
0
IX)

 I
        1                                           2                                    7
         Based  on total area of fuel  exit ports = 1.77 in .  Area of co-axial  air channel =  22.1 in .

        2
         Includes contribution of pilot



         S.R. = Stoichiometric ratio



         For example,  Test No. 228: r>a1r = 0.0744 lb/ft3, va1r = 76.6 ft/sec,  / fuel  = 0.114  lb/ft3, vfuel  = 1.15  ft/sec,  ("v)ajr/("v)fuel

-------
                                                               Table 2-4  (Continued)



                                     COMMERCIAL  1.5  INCH DIAMETER1  AIR-ASSISTED HEAD  G TEST  RESULTS
ro

oo
Test
No.
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
Actual
Fuel Exit
Velocity1
(ft/sec)
98.4
8.47
428
69.1
425
10.9
504
514
509
395
327
431
207
223
126
87.1
67.8
41.7
394
103
357
Propane
in
Ni trogon
(%)
30.3
23.1
65.0
43.9
60.5
100
15.8
17.3
16.5
20.1
20.4
14.0
9.35
•8.48
15.6
8.69
11.1
18.0
20.9
18.2
37.3
Low2
lltg. Val
(Btu/ft3)
712
547
1528
1032
1425
2350
376
413
391
476
474
334
229
210
300
229
290
460
496
441
877
Air-Assist
Flowrate
(SCFM)
515
56
511
1050
1050
1040
0
515
1040
0
513
1040
0
513
1030
0
513
1030
1050
1040
1050
Air-Assist
Exit
Velocity
(ft/sec)
56.0
56.0
55.6
114
114
113
0
56.0
113
0
55.7
113
0
55.7
112
0
55.7
112
114
113
114
Air-
Assist
*«•?
0.98
14.7
0.10
1.93
0.23
5.42
0
0.320
0.694
0
0.156
0.249
0
0.244
0.836
0
0.788
2.49
0.73
3.06
0.45
(pv)air4
(pv)fuc'l
0.494
5.97
0.0%
1.34
0.203
6.75
0
0.101
0.208
1
1.464
0.960
0
1.50
2.95
0
3.787
7.64
0.265
1.01
0.267
Wind
Speed
(mpl )
9
4.5
6
5.5
5
4
Observations
Flame
Length
(ft)
transparent
1
12.5
1
10
transparent
Lift
Off
(in)
4
1
10
3
8
1
Color
transparent
blue base
yellow-blue
blue base
blue base
blue base
Smoke
none
none
none
none
none
none
Sound
low rumble
low rumble
loud roar
roar
loud roar
drone
Comb.
Eff.
(%)
88.8
32.1
99.7
62.3
99.7
54.5
Carbon
Dest.
Eff. (%)
93.5
37.3
99.9
66.6
99.9
58.4
\














Stability Curve Tests





/
2.5
4
5
1.5
1
1.5
4
2
6
blue base
blue base
blue base
none
none
none
loud rumble
low roar
loud roar
57.7
34.3
99.6
59.1
35.8
99.9
Pilot
Flow-
rate
(SCFM)
0
0
0
0
0
0
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
           1                                          2                                   ?
             Based on total  area of fuel  exit ports = 1.77 in .  Area of co-axial air  channel = 22.1  in.

           2
             Includes contribution of pilot
S.R. = Stoichiometric  ratio



For example, Test No.  228: ,
                                     ajr = 0.0744 lb/ft, va1r = 76.6 ft/sec, i fuel = 0.114 lb/ft, vf(jel = 1.15 ft/sec,  (''")a1r/(' v)f|je1 = 43.5

-------
                                                            Table 2-5


                                3  INCH DIAMETER  OPEN  PIPE FLARE (WITH PILOT) TEST RESULTS
[V)
I


Test
No.
263
264
265
266
267
268
Actual
Exit ,
Velocity1
(ft/sec)
203
68.8
9.82
96.7
1.93
209

% Propane
in
Nitrogen
9.96
2.89
9.07
7.61
7.51
21.0

Low ,,
Htg Val,
(Btu/ftJ)

Probe
Ht.
(ft)

Wind
Speed
(mph)

Flame
Length
(ft)

Lift
Off
(in)



Color
237 \



Smoke




Sound


Comb
Eff.
(%)

Hydro -
Carbon
Oest.
Eff (X)

76.1 1
> Flame Stability Tests
260 1
184 J
375
495
11
32
6
5
1
30
2
10
faint yellow
blue, yellow
none
none
low hiss
med. roar
98.7
98.8
99.2
99.97

Pilot
Flowrate
(SCFM)
2.08
2.11
2.10
2.09
2.07
2.09
           1    Based upon pipe inside diameter of 3.125 inches.


           2    Including pilot contribution.

-------
                                Table 2-6
       RESULTS OF SCREENING TESTS ON FLARE SCREENING  FACILITY
Compound
Acetylene
Ethylene
Propylene
l,3-8utadiene

Sutane
Propane
Propane
Benzene
Toluene
Chlorobenzene
Carbon Monoxide
Carbon Monoxide
Carbon Monoxide
Acetone
Aceta 1 dehyde
Ethylene Oxide
CO^ Diluent
Methyl Chloride
Ethylene Oichloride
Vinyl Chloride
Methyl Mercaptan
Acrylonitrile
Hydrogen Cyanide
Ammonia
Ammonia
Gas
Composition
Compd
100
100
100
100

100
100
75
1.50
1.50
1.15
100
20
17
1.43
2.07
1.42
7.58
9.17
1.43
0.11
10.7
1.47
0.013
100
20
Propane 1 N£
0
0
0 0
0 j 0
0 0
1
0
0
0
0
0
25
98.5 0
98.5 0
98.35
0
30
37
98.57
97.93
98.58
92.42
90.33
98.57
99.39
89.30
98.53
99.99
0
30
0
0
0
46
0-
0
0
0
0
0
0
0
0
0
0
0
Velocity at
Stability
Limit ,
(ft/ sec)1
854
443
134
127

58
143
48
61
51
58

108
30
59
58
58
93
65
58
31
65
58
78

74
Lower
Heating
Value
(Btu/ft3)
1475
1580
2300
2730

3321
2350
1763
2370
2381
-2350
OE2
w
99.99
99.91
99.98
99.93

99.99
99.98
99.97
99.59
99.99
99.49
ICould Not Ignite)
1943
923
2347
2331
2337
2171
2212
2335
-2350
2228
-2350
-2350

99.60
79.727
99.30
99.99
96.95
NA
99.94
99.70
96.79
99.39
99.99
35.00

ICould Not Iqmtel
1967
99.90
8!
99.97
99.92
99.93
99.93

99.96
98.18
NA4
99.95
99.90
99.95

99.38
99.42
99.96
99.97
99.95
99.93
99.96
99.95
NA
99.82
99.96
NA
6
NA
Soot
(mg/nH)
<1.5
<1.5
<1.5
755

<1.5
<1.5
<1.5
<1.Q
<1.0
'

<1.0
<1.0
<1.0
<1.5
<1.0
<1.0
<1.0
<1.0
<1.0
<1.0
<1.Q
<1.0

<1.0
1 = Nozzle ID » Q.Q42 inches
2 = Destruction Efficiency
3 = Combustion Efficiency
4 = Not available
5 = Without steam or air assist
6 = On 1/16 inch nozzle  without pilot flames
7 = DE calculated assuming no  CO originated
    from propane
                                      2-10

-------
Table 2-7.   In  order to determine the limits of stable operation for these
flares and  gas  mixtures,  and  the key operating conditions that affect flame
stability  and  efficiency, some conditions with poor  stability and low
combustion  efficiencies were  measured.  Such results merely indicated flare
operation at  or beyond the edge of  the operating envelope, and are not
indicative of  normal commercial  flare operation.

     The  results showed that flare head design influenced the flame stability
as shown in Figure 2-3 for the  coanda steam-injected head and the pressure-
assisted heads.  The stability of the air-assisted head flame was controlled
by the momentum ratio of  air-assist to fuel  streams as shown in Figure 2-4;
heating  value of  the gas  had  little  influence  on flame stability,  except
under conditions with no air-assist.  Combustion efficiency for the pressure
and  coanda  steam-injected  heads  correlated  with the  individual   flame
stability limit for each  head as shown in Figure 2-5.  Figure 2-6 shows the
relationship  between combustion efficiency and air to fuel momentum  ratio,
(pv)air/(pv)fuel, for the air-assisted head.

     The  relative  flare performance of different gases can be determined  on
the  FSF, although the value  determined on  the  lab-scale  FSF for  flame
structure,  flame stability, and  combustion and destruction efficiencies will
be different  than  measured on  pilot  and industrial  scale  flares.   Such
differences in  stability  are  shown in Figure 2-7, comparing flame stability
for  different sized open-pipe nozzles.  Comparison  of ethylene  oxide
destruction efficiency on  the  FSF (1/16 inch nozzle) and the pilot-scale FTF
(3 inch  open  pipe  flare)  shows the  difference  in destruction efficiency
measured in the two facilities.   Ethylene oxide DE  on  the  FSF was 96.95
percent, compared  to 98.4  and  99.5  percent  on  the FTF.  Also, industrial
flares typically  employ pilots and/or flame retention devices to stabilize
the flame and  enhance combustion  which were absent in these tests.

     Tests   identified six compounds which were difficult to destroy  on the
FSF,  using a 1/16 inch nozzle (ID = 0.042 inches).  These were:
                                   2-11

-------
ro

(—<
ro
                                                             Table 2-7


                                        GAS MIXTURE COMBUSTION AND  DESTRUCTION  EFFICIENCY
                                              TEST RESULTS:   3  INCH FLARE. NO  PILOT
Test
281

282

283

284

285

286

287

288

ID
NH3

NH,
J

NH3

NH3

NH,
•J

H2S

H2S

H2S

Actual
Exit
Velocity
(ft/sec)
8.79

9.47

0.15

138

9.56

8.89

139.1

9.11

Gas Composition
(*)
NH3 1.5

NH3 1.7
Propane 36.9
NH3 4.43
Propane 13.2
NH3 4.02
Propane 27.0
NH3 1.98
Propane 17.2
H2S 1.5
Propane 23.3
H2S 4.14
Propane 26! 5
H2S 4.29
Propane 31.2
Gas
Htg Val,
(Btu/ftJ)
561

877

336

658

416

556

646

757

Probe
Ht.3
(ft)
12

12

4

33

12

12

36

14


Wind
Speed
(mph)
5

5

2

3

1

5

3

6

Observations
Flame
Length
(ft)
6

7

0.5

20

8

7

22

9

Lift-
Off
(in.)
3

0

0

36

24

24

24

0

Color
Blue base,
orange
Bright Orange

Orange

Blue-orange

Blue base,
orange
Orange

Blue base,
orange
Bright orange

Smoke
None

Very
Little
None

None

None

None

None

None

Sound
None

None

None

Rumble

None

None

Rumble

None

Hydro-
Carbon
Comb.
Eff
00
99.3

99.6

99.3

99.7

92.1

93.2

98.1

99.4

)est
Iff
(%)
)2.1

39.7

-95

)9.5

"enax

NA2

NA2

NA2

                   Remainder N?



                   Destruction efficiency data for H-S is not available due to analytical
                   errors  in the measurements.
                  Height above  flare  tip

-------
                                                             Table 2-7 (Continued)



                                       GAS MIXTURE COMBUSTION AND DESTRUCTION EFFICIENCY

                                       TEST RESULTS:   3 INCH  FLARE.  NO  PILOT

Test
289

290

291
292
293
294
295

ID
H2S

H2S

1.3
iutadient
1,3
iutadiene
1,3-
iutadiene
ithylene
txlde
ithylene
Dxide
/\Ctual
Exit
Velocity
(ft/sec)
9.11

0.669

0.296
6.87
7.03
0.320
4.63

Gas Composition
(»)
H2S 7.61
Propane 21.1
H2S 4.74

1,3- 11 2
But.
1,3- 5.32
But.
1,3- 7.77
But.
Eth. 13.2
Oxide
Eth. 13.0
Oxide

Gas
Htg Val
(Btu/ft3)
539

370

305
145
212
178
175

Probe
Ht.J
(ft)
14

4

4
13
15
4
7
Observations
Wind
Speed
(mph)
5

4

3
3
7
2
7
Maine
Length
(ft)
8

1.5

1
6
5
1
4
Lift
Off
(in.)
12

0

2
6
0
0
NR
Color
Orange

Dim-orange

Orange
Orange
Orange
Dim-orange
Transparent
Smoke
V. Little

None

Slight
Dark
Slight
Grey-
Black
None
None
Sound
None

None

None
None
None
None
None
Hydro-
Carbon
Comb.
Eff
CO
99.2

97.7

98.2
95.8
97.7
98.4
99.5

Dest
Eff
(%)
Tenax

Outside
Sample
Limits
99.2
96.8
99.9
99.2
99.7
ro

t—<
co
             Remainder
             Destruction efficiency data for H^S is  not available due to analytical

             errors  in the measurements.
             Height above  flare tip

-------
ro
i
                 
-------
1000
               B   .0
                 S

                 o

                 o
                                             8
                                                 Without Pilot

                                               O 180-400 Btu/fr
                                               n 401-900
                                               O 901-1400
                                               A 1401-2350
                                                  With Pilot

                                                   209-400 Btu/ft3

                                                   <400
 0.1
                   .4
.6
                                     .8
                                    1
1.0
                           1+ (Pv)air/(/>v) gas
                        1.2
                                                             1.4
   Figure 2-4 .    Maximum gas exit velocity for  stability
                   versus  air-assist to  gas momentum ratio
                   for  the air-assisted  head G, with and
                   without pilot.
                                2-15

-------
   100-
c
OJ
O


I   94
O


0)


5   92
q-
o

    90
    88
    86
    84
   I\ 12 in. Coanda Steam-Injected

j           Head  D

.  Di 1.5 in. Pressure-Assisted  Head  E


!  Q 3.8 in. Pressure-Assisted  Head  F


I  © 3 in. Open Pipe  Head with  Pilot
*


I
•

1
 I
 e



O
      0                  1                  2        2.5


      Gas Heating Value/Minimum Heating Value for Stability


      Figure 2-5.   Combustion efficiency vs. flame  stability  for

                    steam-injected and pressure-assisted  flare
                    heads.
                               2-16

-------
   99.9
   99.0
01
   90.0
             (25  Without Pilot

             A  With Pilot

             S.R. - Air-Assist to Fuel
                   Stolchiometric Ratio
         5.0
                 0.2        0.4        0.6

                                 1

                         1 + (pv)air/{pv)gas
                                                                0.1
                                                               1.0
                    01

                   .0
                                                                     
-------
O1
C
2400
2300
2200

2100
2000
1900
1800

1700

1600
1500
1400

1300
1200
1100
1000
 900
 300
 700
 600
 500
 400
 300

 200
 100
                                                               1/2 in.
          ^ 1/16  in.  Nozzle
          Q 1/8 in.   Nozzle
          ^ 1/4 in.   Nozzle
          l\ III in.   Nozzle
          [7 1  in.     Nozzle
          <3 1 1/2  in. Open  Pipe Flare
          Q 2  in. Open  Pipe Flare  [.>'
          G 2 1/2  in. Open  Pipe Flare
                                                                       If
                                                                         r-,-
       o.i
1.0
                               10.0           100.0
                                irr             ,
-------
     •    1,3-butadiene has  the  potential without  steam or air assist to
          yield large amounts of soot  (75 mg/m3)
     •    CO was difficult to ignite when pure
     •    Ethylene oxide yielded destruction efficiency of 96.95 percent
     •    Vinyl chloride yielded destruction efficiency of 96.79 percent
     t    HCN yielded low destruction  efficiency of 85.00  percent
     •    NH3 was difficult to ignite  when pure without a  pilot

Destruction efficiencies  (DE's)  for NH3, 1,3-butadiene, ethylene oxide,
propane, and  H£S  were measured  on  the FTP.  The  flame stability  curve
depended on the compound as shown  in Figure 2-8.  (HgS and NH3 were tested as
minor constituents in propane-nitrogen mixtures).  The DE  of the individual
compounds depended on compound  type but correlated with the respective
stability curve for NH3, 1,3-butadiene, ethylene oxide, and propane as shown
in  Figure 2-9.    The primary influence on  destruction and combustion
efficiency is  flame stability,  not  gas heating value or exit gas velocity.
H2$ destruction  efficiency  results  are unavailable because of analytical
problems.  The  destruction  efficiency of propane  in mixtures with small
amounts  of NH3 or H£S are reported  in  Figure 2-9  as propane in ammonia tests
and propane in hydrogen sulfide tests.

     The following conclusions were reached based on the results:

     0    Flames from nozzles less  than  1 1/2  inches  in diameter  generally
          are not similar to flames from larger nozzles.

     t    Flare head design can influence the flame stability curve.

     •    Combustion efficiency can be correlated with flame stability  based
          upon gas heating value for the pressure heads and  the coanda steam-
          injected head.

     •    For  the  limited  conditions tested,  the flame stability  and
          combustion efficiency of the air-assisted head  correlated with the
                                   2-19

-------
                  1200.
 i
ro
o
                  1100




                  1000





                  900





                  800





                  700




                  600
               a
               
-------
     99.99
    99.9

-------
     momentum ratio  of air to fuel;  the  heating value of  the gas had
     only  a minor influence.

•    Limited data on  an  air-assisted  flare  shows that use of a pilot
     light improves flame stability.

•    The destruction efficiency of compounds depends on the  structure of
     the compounds.

•    The destruction efficiency of different compounds can be correlated
     with  the  flame stability curve for  each compound, for the  compounds
     tested  in this program.

•    Stable  flare flames and high  (>98-99  percent) combustion  and
     destruction efficiencies  are attained when the flares  are operated
     within  operating envelopes  specific  to each  flare  head and gas
     mixture  tested.   Operation  beyond  the edge  of  the operating
     envelope results in rapid flame de-stabilization and a  decrease in
     combustion  and destruction efficiencies.
                               2-22

-------
3.0       FLARE AERODYNAMICS

     The aerodynamics  of different sized  flare  heads was studied prior to
testing the  influence of flare  head types or  gas  composition on flare
efficiency.   The aerodynamic  studies were conducted  to determine the minimum
head size  which  would yield  results similar to full  scale commercial  flare
heads.   Flare flame length  prediction techniques were  also investigated
during  this study.  The results were used to select  flare head sizes for the
commercial  head and gas composition tests.

     The aerodynamics  study  included a review  of previous flare studies as
well as  experimental  work testing  1/2 inch to  2  1/2 inch open pipe  flare
heads,  without  flame retention  devices.   The  results were used to develop
relations  between Reynolds  number  and Richardson  number and flame length,
flame stability and flame temperature.  (3.1)

3.1       Reynolds Number and  Richardson Number

     The relief gas Reynolds number  (Re)  and Richardson number (Ri) at the
nozzle  exit  determine the  aerodynamic  structure  of  the  flare flames.
Reynolds  Number  (Re)  is a measure of the  inertia!  to viscous forces of the
flame, and  Ri  is a measure of  buoyant forces to inertia! forces of the flame.
A flame  with  Ri greater than  one is dominated by buoyant forces and one with
Ri less  than one is dominated  by inertial forces.  Figure 3-1 compares Re and
Ri for  1/16  inch  through 12  inch flare heads tested at EER, using simulated
relief  gas mixtures of propane-nitrogen.  The  shaded  area indicates the
region  of  head size and exit  velocities for typical  industrial sized flares.
Since  Re  depends  on  the relief  gas composition, the  data points are
frequently not  exactly on the approximate grid line drawn for flares of
different  sizes.  The  general  trend, however, is evident.  As the flare size
decreases, the  nozzle Re and Ri decrease.  Flames  from smaller nozzles are
dominated by inertial  and  turbulent forces, and are  aerodynamical ly
dissimilar  to  larger, buoyancy dominated flames from  industrial flare heads.
                                   3-1

-------
to
I
                          10
                                                        Reynolds Number
                          Figure 3-1.  Aerodynamic conditions of flares  tested  at HER compared to
                                       typical  commercial flare conditions.

-------
                           Legend for Figure 3-1
New EER Data          O  1/16 in.  (0.042 in. ID)   Nozzle
                      O  1/8 in.   (0.085 in. ID)   Nozzle
                      ^3  1/4 in.   (0.180 in. ID)   Nozzle
                      A  1/2 in.   Nozzle
                      V  1 in.     Nozzle
                      <\]  1 1/2 in. Nozzle
                       Q>  1 1/2 in. Commercial Pressure-Assisted Head E
                       £)  1 1/2 in. Commercial Air-Assisted Head G
                       £1  1 1/2 in. Commercial Air-Assisted Head G with pilot
                      C\  2 in.     Open Pipe Flare Head
                       Q  2 1/2 in. Open Pipe Flare Head
                       O  3.8 in.   Commercial Pressure-Assisted Head F
                       h  12 in.    Commercial Coanda Steam-Injected Head D

Previous EER Data (1.3)O  3 in.     Open Pipe Flare Head
                       A  6 in.     Open Pipe Flare Head
                      O  12 in.    Open Pipe Flare Head
                       C^  12 in.    Commercial Head A
                       v  12 in.    Commercial Head B
                       D  12 in.    Commercial Head C
                                 3-3

-------
3.2       Flame  Length

     Current and previous test results  of measured flame lengths  from flares
were compared  to predicted  flame lengths using information  and techniques
available  in the literature (3.2-3.4).   Neither empirical  (3.5-3.7), complex
integral  (3.8,  3.9), nor  differential  (3.10-3.13) techniques could predict
the  flame  lengths or trends in the flame length of this or other studies of
large flare flames.  Figure 3-2 shows the comparison of flame length measured
in  a previous  EER flare  study (1.3)  with  predictions  using  techniques
suggested  in  the literature.  For example, the technique of Brzustowski, et
al.  (3.14)  predicted a flame length of 18-28 feet, depending upon  conditions,
and  Becker,  et  al . (3.9,  3.15, 3.16)  predicted a flame length  of 18 to 35
feet, when  the observed flame length was 30 feet.

     Much  better  agreement between observed  and predicted flame  lengths was
achieved by fitting the data with a cruder approach (3.1, 3.2).  The approach
used by  API  (3.2) correlates the flame length with the heat release.   This
correlation, suggested by Hottel and Hawthorne  (3.17), is based on data  from
small  flames,  where combustion was limited by  air diffusion into the flame.
Although this  prediction  has been criticized for its  inaccuracy  (3.18),
Figure  3-3 shows relatively good agreement with the data of this  study,  data
from larger flares, and the original data of the API study.  For flame length
predictions  of  large flares, the coefficients  in this  correlation  were
adjusted.  Using the same API prediction with  coefficients adjusted for large
flares  of  this  study, poorer agreement was found for  the  flame  lengths
produced from  some  smaller flares.  Flame lengths not accurately predicted by
the  API  correlation were  from small  nozzles (Zl-1/2 inch, V-l  inch, ^] - 1
1/2  inch,  £3 -2  inch), the  coanda steam-injected head  ([\ ),  and the  air-
assisted head  ( & and & ).   The flame length  from nozzles smaller than 1/2
inch agree with previous  predictions.  However, use of  the  relationship
derived  from  nozzles less than 1 1/2 inch in diameter would severely under-
predict the length  of the flames from large flare heads.

     Figure 3-3 also  shows  the least-squares fit  of flame length  vs  heat
release  for 3 through 12 inch heads tested previously at EER.   Data from the
                                    3-4

-------
  O)
  c
  cu
u
T3
        60

        55


        50

        45

        40
        35  _
        30  -
        25

        20

        15

        10

         5
             O Hottel & Harthorne
                Brzustowski
                Becker and Liang
                      10    15    20    25   30    35

                        Observed  Flame Length (ft)
Figure  3-2.   Comparison of flame lengths predicted
               by available studies with  those observed
               in this  study.
                              3-5

-------
                10
co

CT)
                10'
              c
              -5
                 10U
                   104  2
                                                                         II     III      I
                                                                                                0.388
                                                                        (1)
                                            o
                                                               ^D


                                                       1\     API 521 Extrapolation

4   105  2     4   106"  2    4  107   2

                            Heat  Release (Btu/hr)
                                                                        108   2     4   109   2     4   1010
                     Figure 3-3.   Correlation  of flame length with heat  release for 1/2 inch  to 31 inch
                                   flares.
                       '  Correlation based on  3  through  12  inch flare  heads tested  previously  (1.3).

-------
                    Legend for Figure 3-3
API Data (3.2)
                   3 Fuel  Gas—20 in.  Stack  (in  line)
                   Q Algerian Gas Well
                   & Catalytic Reformer—Recycle Gas—24  in. Stack
                   B Catalytic Reformer—Reactor Effluent Gas—24 in. Stack
                   $> Dehydrogenation  Unit—12  in.  Stack
                   C3 Hydrogen—31 in.  Stack
                   O Hydrogen—30 in.  Stack
Straitz Data
 (3.3, 3.4)
                   Q Natural Gas, 2, 3, 6 in. Head
                    0 Propane, 2, 3 in.
New EER Data
                    12 in.
                   k 12 in.
                   Q 12 in.
                   Q 12 in.
                                Open  Pipe  Flare  Head
                                Open  Pipe  Flare  Head
                                Open  Pipe  Flare  Head
                                Commercial  Head  A
                                Commercial  Head  B
                                Commercial  Head  C
                                     3-7

-------
present  test program  shows  surprisingly  good correlation with this least
squares  fit  for flame  lengths from 1/16  through 1/4 inch nozzles (Figure
3-4), but  somewhat poorer  correlation  for some of the larger, commercial
heads as  discussed above.

     A correlation was developed in previous  work  (1.3) relating flame length
to the Richardson number, corrected for the  volume change caused by increase
in the gas  temperature from ambient to the flame temperature:
                    q  x  F
     L/d = 7.41	  0.60 R1- °-216
                 C^ (l+26x)
     where  L  =  flame length, ft
           d  =  flare head diameter, ft
           q  =  2,350 Btu/ft3, propane gas lower  heating value
           x  =  mole fraction of propane in the fuel
           F  =  fraction of heat lost by radiation  from the flame
         Cp^  =  constant pressure heat capacity of  the air, Btu/ft^R
         To,  =  ambient temperature, R
         26  =  factor to account for the change  in moles as a result of
               stoichiometric combustion of propane.
         Ri  =  Richardson Number based on conditions at flare head (gd/v2)

     The factor F was not measured in this study, but its relative value was
obtained from  a single  set of data with different mole fraction of propane
and  a  constant Richardson number  of  2.0 as  shown in Figure 3-5.  This
relation was then applied to all  other flames.  The absolute value of F does
not matter for a correlation with  a  single  fuel.  However, F factors are
available  in the literature and will be required for correlations of flames
burning different gases.

     Equation  3-1 was  empirically derived, but is  based upon  results  of
several  previous studies.   Workers in combustion research (3.19-3.25) have
                                   3-8

-------
en
c:

-------
      a.

      £
      31

      U
         1.0
      I   -3
          .6
      c
      a
         .2
                    •a
                  20
40
60
30
100
                    Mole Fraction Propane in N,
Figure 3-5.   Empirical  correlation  of radiant heat loss

              from propane-nitrogen  flames.  Richardson

              number = 2.0.
                              3-10

-------
correlated  flame  length  to heat release  to the 0.4 power.  This is close  to
the exponent  of 0.388 determined by a least squares fit of 3,6 and 12  inch
head data as  shown  in Figure 3-3.  Such good  agreement also verified the
success in  correlating large flames using  Hottel and Hawthorne's
calculations.   In  addition,  conversion of Stewart's (3.25) formula for
combustion number,  predicts flame length  based upon density to the 0.4 power
and Richardson  number to the -0.2 power.  Flame lengths  correlated  with
equivalent  expressions by Thomas (3.19) predicted that  the flame length
should be proportional to the density  change  to the 0.61  power and the
Richardson Number to the -0.3 power.

     Using  Equation  3-1, good correlation  is seen in  Figure 3-6 for
previously tested 3,  6  and 12 inch flare heads (3.1).  Good agreement is  also
seen  for 2 and 2 1/2 inch  flare  heads  shown in Figure 3-7, but the
correlation overestimates the length for 1/2,  1, and 1  1/2 inch flames.
Flames burned on small  nozzles have low Richardson numbers (see Figure 3-1).
In operation, the  length  of these flames  is not completely controlled by the
Richardson  number  and hence  equation 3-1 tends  to overestimate the flame
length.  In addition the  fraction of  heat lost by radiation may be different
for small flames  than large  ones.  Inspection  of Figure  3-7 shows  that
correlations for each  individual small nozzle could be developed by altering
the  constants  in  Equation  3-1.  However, the  correlation would  not  be
universal, as it apparently is for open pipe flares greater than 2 inches  in
diameter.   Similar  discrepancies between the correlation and even smaller
nozzles  are  shown in Figure 3-8.

     The  observed flame lengths of the air-assisted heads shown in Figure 3-9
are considerably  shorter than predicted.  This  may be  the result of the
smaller  individual  ports in  the fuel spider  and/or the  increased mixing
caused by the co-flowing air stream.

     Flame  length  predictions shown in Figure 3-9 for commercial flare heads
D, E, F, and  G  also deviate  from measured lengths.  Poor agreement between
predicted and measured flame lengths  for  these heads is due to the effect  of
                                  3-11

-------
-C
•M
Ol
T3
0)
    50
    45
    40
    35
   30
    25
-a   20
-0)
    15
    10
                                                 o  P   -
                                                20%
                    10      15      20      25

                        Observed Flame Length  (ft)
30
35
  Figure  3-6.   Predicted  flame length from  modified Richardson
                number correlation compared  to observed flame
                length for previously tested 3,6 and 12 inch
                flare heads.   (From 3.1)
                               3-12

-------
-o
0)
     40
    36
     32
     28
    24
5   20
o>
     16
    12
                                                            2 1/2 in.
                                    2 in,






                     1/2 in. Nozzle




                 ^7  1 in.  Nozzle




                 ^3  1 1/2  in.  Open Pipe Flare Head




                 u  2 in.  Open Pipe Flare Head



                 Q  2 1/2  in.  Open Pipe Flare Head



                	I	I	I
                               12       16       20


                          Observed Flame Length  (ft)
                                      24
28
    Figure 3-7.
Predicted flame length  from modified  Richardson number

correlation compared  to observed flame length  for 1/2

through 2 1/2  inch flare heads.
                                        3-13

-------
     3  _
0)
	1


-------
-C
4->
01
c
cu
73
OJ
   40
   36
   32
   28
24
   20
   16
   12
          1 1/2 in.  air-assisted head G
                                                   1  1/2 in.  pressure-assisted
                                                   head E
                                              3.3 in.  pressure head F
                                      Q  1 1/2 in.  Pressure-Assisted Head E


                                      Cl  1 1/2 in Air-Assisted Head G

                                          1 1/2 in.  Air-Assisted Head G
                                           with Pilot

                                      O  3.8 in.  Pressure-Assisted Head F

                                          12 in.  Coanda  Head D
                                I
                               12       16        20
                              Observed Flame  Length (ft)
                                                      24
28
32
Figure 3-9.   Predicted flame length  from modified Richardson  number
                correlation  compared to  observed flame length  for 1  1/2
                through  12 inch commercial heads.
                                         3-15

-------
steam, pressure, and design  on  flare flames, combined with  the previously
mentioned aerodynamic differences  between small and large flares.

3.3        Flame  Stability

     The  flame  stability limit for a  given propane-nitrogen  gas exit velocity
is the minimum gas heating  value that maintains  a flare  flame.  For each
flare head,  the  minimum heating  value to  maintain a stable flame  was
evaluated at several velocities.   Limiting heating values for  1/16 through 2
1/2  inch  heads are compared with  previous values for 3 through 12 inch flare
heads in  Figure  3-10.  At heating values on or above the stability curve for
a given  flare  heads, a flame is  maintained.  The flame is  extinguished when
the  gas  heating  value drops below the  minimum value defined  by the stability
curve.

     The   broad stability band on Figure  3-10 is  for  stability curves of
previously tested 3,6, and 12  inch diameter flare heads (1.3).   The 1 1/2, 2,
and 2 1/2 inch  flare heads exhibit stability curves within this region.  This
indicates that flare stability  is  similar  for  1  1/2 inch through 12 inch
flare heads.  The stability of 1 inch and 1/2  inch nozzles,  however, is quite
different.   As shown by  comparison of the  Reynolds number  and Richardson
number  and  by  the  incompatibility of  large  and  small  flame  length
predictions,  the characteristics of  flames burned on large flare heads are
different from characteristics  of  flames burned  on small  nozzles.  This
conclusion  is  verified by the flame  stability results, which show that flare
heads less  than  1  1/2  inches  in diameter have different stability
characteristics than flare heads 1 1/2  - 12 inches in diameter.

3.4       Pseudo  Adiabatic Flame Temperatures  and Stability

     A flame will  be  stable  when the flame velocity is equal  to or greater
than  the  relief  gas velocity in  all  directions.  The flame  will be unstable
when  the  gas  velocity remains greater than the flame velocity until the gas
is diluted beyond its lower flammability limit.  The flame velocity, however,
is controlled  by the Arrhenius  kinetic parameters of the  flame reactions.
                                   3-16

-------
3
•M
co
2400
2300
2200

2100
2000
1900
1800

1700 j-

1600 -
1500 -
1400 U

1300
1200
1100
1000

 900
 800
 700
 600
 500

 400
300

200
     100
                                                1/16 in./    I 1/8 in.
                                                                    1/2 in.
           o  1/16 in.
           O  1/8 in.
           ^3  1/4 in.
           /J  1/2 in.
           [7  1 in.
 Nozzle
 Nozzle
 Nozzle
 Nozzle
 Nozzle
              1 1/2 in.  Open  Pipe Flare
              2 in.  Open  Pipe Flare
             i 2 1/2 in.  Open  Pipe Flare
                                                     2 1/2 in.
                                        I
       0.1
1.0            10.0   .         100.0
         Exit Velocity (ft/sec)
                                                                   1000.0
      Figure 3-10.   Flame  stability curves for pipe  flare heads.
                                    3-17

-------
Higher  relief gas exit  velocities  can  be attained when the adiabatic flame
temperature is high,  because the flame speed increases with  temperature.
Figure 3-11  shows  the maximum gas exit velocity  versus pseudo adiabatic flame
temperature  for  previously tested 3 through 12 inch  flare   heads.  A
relatively  good correlation  is obtained,  and there is  little  difference
between  the  performance of different  head sizes.

     Different  flame  temperature and stability results are shown  in Figure
3-12  for  the  coanda steam-injected, pressure-assisted,  and  air-assisted
commercial  heads tested in the current  study.  Correlations of  the  limiting
velocity  with  pseudo adiabatic flame temperature are good for each  head, but
head  size and design have a large influence on the correlation.   The design
criteria for these  commercial pressure and air-assisted flare heads were:

     t    Relief  gas flow capacity of 3,440 Ib/hr (36,000 SCFH) for gas with
          1,200  Btu/ft3 heating value.

     •    Approximately 0.01 ft2 open area, for  an equivalent diameter of 1.5
          inches.

     •    For  the  pressure-assisted heads, a  maximum pressure drop  across the
          flare  head of 15 psig.

     t    For  the  air-assisted head,  two  air  delivery rates,  with  a maximum
          rate  of  200,000 ft3/hr at 20 inches water pressure.

Pressure-assisted head  E  and air-assisted head  G met these criteria, but
pressure-assisted head  F  had an open area of 0.0785 ft2, and an  equivalent
diameter  of 3.8 inches.  For the same calculated flame temperature, each of
the commercial  heads can  be  operated  at a higher stable velocity  than the
open  pipe flares.  Pressure-assisted head E produces stable flames at much
higher velocities  than other flare heads  tested.

     The  size of  smaller flare  heads  adversely affects flame stability.
Figure  3-13 shows  that for 1/16 through  2 1/2 inch heads, the maximum stable
                                   3-13

-------
    1000-
    100
o
CD
.0
us
10
O
CD
    1.0
   0.1
                   PREVIOUS  EER DATA (1.3)
                   O   3 in.  Open Pipe  Flare Head
                   A   6 in.  Open Pipe  Flare Head
                       12 in. Open Pipe Flare Head
                       12 in. Commercial Head A
                       12 in. Commercial Head B
                       12 in. Commercial Head C
                        Polynomial least-squares
                        fit  R^ = O.S8)
                                        Low exit velocity, not representative
                                        of typical operation.
              2.6
                   2.8
3.0
3.2
                                                  3.4
                                                      3.6
                                    3.8
                                                                             4.0
                       Adiabatic Flame Temperature (1/R) x 10
 Figure  3-11.   Maximum relief gas exit velocity vs  flame temperature
                  for previously tested  3 through 12  inch  flare heads.
                                         3-19

-------
 1000
  100
   10
s_
a
o
o
£  1.0
   0.1
Polynomial
least-squares
fit of previously
tested 3-12 in.
flare heads, with
and without flame
retention devices
                Low Flowrates
                                         1 1/2  in. Pressure-Assisted  Head E
                                  1  1/2 in.  Air-Assisted  Head 6
                                           with Mo Air-Assist, No Pilot

                                  3.8 in. Pressure-Assisted Head  F

                                  12 in.  Coanda Steam-Injected Head D
                                          I
     2.4
       2.6
                       2.8
                         3.0
3.2
                                                   '', a.
3.6
                                                                              4.0
                              Adiabatic Flame  Temperature  (1/R x 10
      Figure 3-12.   Maximum  relief gas exit  velocity  vs flame
                       temperature  for  commercial  heads  D through  G.
                                         3-20

-------
     1000 -
.a
a
u
o
       10
      1.0
     0.1
       2
                        12  in. coanda steam-injected head D
                              1  1/2 in. commercial air-
                              assisted head G,  with no
                              air-assist or pilot
                                               1  1/2 in. commercial
                                               pressure-assisted
                                               head F
 Low Flowrates

O 1/16  in.  (0.042 in.  ID) NozzTe'

Q  1/8 in.   (0.085 in.  ID) Nozzle

•^  1/4 in.   (0.180 in.  ID; Nozzle

£  1/2 in.  Nozzle

^  1 in.  Nozzle

<-,  1 1/2  in. Open Pipe  Flare Head


/-]  2 in.  Open  Pipe Flare Head

Q  2  1/2 in. Open Pipe  Flare Head
                                               1 1/2 in.  commercial  pressure-
                                               assisted head E
                                3-12 in. flare heads,  with
                                and without flame  retention
                                devices
2.6      2.8       3.0       3.2        3.4

            Adiabatic Flame Temperature  (1/R) X 10
                                                     3.6

                                                     ,4
                                                                         3.8
4.0
       Figure  3-13.   Relation  between  flame  stability and pseudo
                        adiabatic flame  temperature.
                                            3-21

-------
exit velocity decreases  with decreasing diameter.  The  limiting  velocity for
flares 1 through 2  1/2  inches in diameter is similar to  the  limiting velocity
obtained  for flare  heads 3 through 12 inches in diameter, but lower than for
the commercial heads  D through G.  Figures 3-11  through 3-13 show that for
many  of  the flare  heads  there is a  minimum gas  flowrate  and corresponding
exit velocity,  below which the flame becomes less stable.  This may be due to
flame  operation below  the velocity where the flame will become unstable from
minor  fluctuations  in  flow or wind  conditions,  and/or where  there is too
little heat input to overcome heat losses.

     In conclusion, small flares and commercial flares tested were found to
yield  flame length  and stability curves which are different from previously
reported  3 to  12 inch  flares heads.   The characteristics of flames produced
on nozzles less than  1 1/2 inches in diameter depend on the nozzle diameter
and are  less stable than  flames produced by  larger  diameter  pipe flares.
Flames  produced on  commercial pressure-assisted and  coanda steam-injected
heads  are more stable  than flames produced on previously tested  3 to 12 inch
pipe flares.
                                 3-22

-------
3.5       References

3.1    Pohl ,  J. H., N. R. Soelberg, E. Poncelet, and B.  A.  Tichenor,  "The
       Structure of Large  Buoyant  Flames", American  Flame Research  Committee
       Fall  Meeting, Livermore,  CA,  16-18  October  1985.   (Abstract  submitted)

3.2    "Guide for Pressure - Relieving Despresuring Systems",  API  RP  521,
       Washington, DC, 1982.

3.3    Straitz, III, J. F. and 0. A.  O'Leary,  "Flare Radiation",  presented  at
       the 2nd International  Symposium on  Loss  Prevention  and  Safety Promotion
       in the Process  Industries, Heidelberg,  Germany, September  1977.

3.4    Straitz, III, J. F., "Flaring  for  Gaseous Control  in the Petroleum
       Industry", presented at 1978 Annual  Meeting of the Air  Pollution  Control
       Association, Paper 78-58.8,  Austin,  Texas, 1978.

3.5    Kent,  G. R., Hydrocarbon Processing  43(8), 121, 1964.

3.6    Reed,  R. D., Chemical Engineering  Progress 64(6),  53, 1968.

3.7    Kalghatgi, G. T., Combustion and Flame  52, 91, 1983.

3.8    Escudier, M. P., Combustion  Science  and Technology 4, 293,  1972.

3.9    Becker, H. A.,  D. Kiang, and C. I.  Downey, Eighteenth Symposium
       (International) on Combustion, p.  1061, The  Combustion  Institute,  1981.

3.10   Tamanini, F., Combustion and Flame  30,  85, 1977.

3.11   Botros, P. E. and T. A. Brzustowski, Seventeenth  Symposium  (International)
       on Combustion,  p. 389,  The Combustion Institute,  1979.

3.12   Gal ant, S., "Un Modele  Simp!ifie de  Rayonnement Emis par les  Flames  de
       Diffusion Libres," Societe  Bertin  and  Cie,  Report  Number N.  52.81.15,
       Plaisir, France, 1981.

3.13   You,  H. Z. and  G. M. Faeth,  Combustion  and Flame  44, p.  261,  1982.

3.14   Brzustowski, T. A., S.  R. Gollahalli and  H. F. Sullivan, Combustion
       Science and Technology  11, 29, 1975.

3.15   Becker, H. A. and D. Diang,  Combustion  and Flame  32, 115,  1978.

3.16   Becker,H. A. and S. Yamazaki,  Combustion  and Flame 33,  123, 1978.

3.17   Hottle, H. C. and W. R. Hawthorne.   Third Symposium  (International)  on
       Combustion, The Williams and Wilkins Company, p.  254, 1949.
                                      3-23

-------
3.18   Heitner,  I.,  Hydrocarbon Processing, 209, 1970.

3.19   Thomas, P.  H.,  Ninth Symposium (International)  on  Combustion,  p. 844,
       The Combustion  Institute, 1977.

3.20   Morton, B.  R.,  Tenth Symposium (International)  on  Combustion,  p. 973,
       Academic  Press,  1965.

3.21   Putnam,  A. A.  and C. F. Speich,  Ninth  Symposium  (International)
       Combustion, p.  867, Academic Press, 1963.

3.22   Cox, C. and R.  Chitty,  Combustion Flame 39,  191,  1980.

3.23   Zukoski,  E. E.,  T. Kubota and B. Cetegen,  Fire  Safety Journal 3, 107,
       1980/1981.

3.24   Orloff,  L. and J. Deris,   Nineteenth  Symposium (International) on
       Combustion, p.  905,  The Combustion Institute,  1982.

3.25   Stewart,  F. B.,  Combustion Science" and Technology  2,  203,  1970.
                                  3-24

-------
4.0       COMMERCIAL FLARE HEAD  TEST RESULTS .

    The primary  objective of these tests was  to  determine the influence of
flare  head  design on flare combustion  efficiency. The  flare combustion
efficiency  was measured for  propane-nitrogen mixtures flared using  different
commercial  heads.  Results show  that  the combustion efficiency is  greater
than 98 percent for flare heads,  except for the  air-assisted flare head, when
operated  within  their stable  flame  regimes.   The air-assisted  head test
results were less  conclusive.

     The Flare Advisory Committee recommended in January 1984 that additional
combustion  efficiency tests be conducted on  other types  of comrnercially
available flare  heads.  Table  4-1 lists the four different head types that
were tested.   The design of the commercial flare heads is proprietary, and
limited  design  information  can  be disclosed.  Due to the proprietary nature
of the  flare heads, each is  identified only by  type and the letters D, E, F,
and  G.  The  coanda  steam-injected  flare head  was tested at  gas exit
velocities  ranging from 0.2  to  9.9  ft/sec based on the  12 inch  design
opening.   The actual  imposed velocities vary  in this  head because of
induction of steam and air and the  variable cross section.  Pressure-assisted
head E  had an equivalent diameter of 1.5 inches, and was tested at 14.2 - 907
ft/sec.   Pressure-assisted  head F had an equivalent diameter of 3.8  inches.
This head was  tested at 4.4-157  ft/sec. The  air-assisted head  G  had an
equivalent diameter of 1.5 inches,  and was tested between 8.5 and 428  ft/sec.

     Due  to the  difference in  size  and design  of the heads, combustion
efficiencies  measured on the  heads  could not be directly compared. Flare
heads  are designed for specific conditions, such  as relief gas composition,
heating value and  exit velocity.   The  operating regime was defined
experimentally for each flare  head by determining the minimum heating value
required  to maintain the flame at  a  given gas  exit velocity.  A  minimum
stability curve was generated in  this manner for each flare head. Combustion
efficiency  of  each head was measured at conditions on this curve of  minimum
stability and at  heating values  50 percent greater than required to maintain
a stable  flame.   Therefore, combustion efficiency was  measured at flare
                                   4-1

-------
                        Table 4-1.
COMMERCIAL HEADS SELECTED FOR COMBUSTION EFFICIENCY TESTS
EER
Desig-
nation
D
E
F
G
Flare Head
Coanda Steam-
Injected Head
Pressure-
Assisted Head
Pressure-
Assisted Head
Air-Assisted
Head
Geometry
Upward opening
cone with steam-
injection ports
Horizontal bar
Open
Cross spider
Open
Area
(in.2)
113
1.77
11.3
1.77
Equivalent
Diameter
(in.)
12
1.5
3.8
1.5
Equipped
With Flame
Retention
Device
no
yes
yes
no
                            4-2

-------
operating conditions both within and at the  limits of the operating  envelope
for the  flare.  Low combustion efficiency results for tests at the limits of
the  operating envelope  are  expected, and  are not indicative of normal
commercial  operation.

4.1       Coanda Steam-Injected Head D

     Flame stability  and combustion efficiency were measured on the coanda
steam-injected flare head.  This head was designed to have the capacity  of a
12 inch  open  pipe flare.  Figure 4-1 shows  the limits of flame stability for
this  head.  For a given exit velocity indicated by each data point, reduction
of the heating value resulted  in blow-out of  the flame.  All  measurements on
this head  were made with  the minimum steam  flow (140 Ib/hr) to prevent the
relief gas from exiting  the  injection  ports. The shaded area on Figure 4-1
represents  the region of flame stability  previously reported for 3, 6, and 12
inch  diameter  flare heads, both with and  without  flame retention devices.

     Combustion efficiency was  measured at relief gas exit velocities between
0.2 and  9.9  ft/sec, and  at gas heating values near and above the minimum
heating  value required for flame stability.   In  Figure 4-2,  combustion
efficiency is correlated  to  flame stability using the ratio of gas heating
value to  that minimum heating value required to maintain flame stability.
High combustion  efficiency  (>99 percent)  is obtained where  the gas heating
value is  greater than 10  percent above the heating value  required for a
stable flame.  Combustion  efficiency  decreases as the heating value for a
stable  flame  limit is approached.

4.2       Pressure-Assisted Head E

     Similar measurements were  made for pressure-assisted head E.  Figure 4-3
shows the  flame stability  curve for pressure-assisted head E.   Combustion
efficiency was measured at gas exit velocities between 12.4  and 907 ft/sec,
for gas  heating values at  and above the stability curve.  Figure 4-4 shows
the correlation  between  combustion efficiency and flame  stability.  The
correlation  is the same as for other heads tested.   Combustion efficiencies
                                    4-3

-------
1200


1100


1000



 900



 800



 700


 600



 500


 400



 300


 200


 100
       _L
  _L
  JL
                        1.0
      10.0

Exit Velocity  (ft/sec)
100.0
1000.0
    Figure 4-1.   Flame stability curve for commercial  12 inch coanda steam injected
                   head D.   Steam flowrate at  140 Ib /hr.

-------
       99
    u
    i.
    
-------
1200


1100



1000


 900


 800


 700


 600


 500


 400


 300


 200
 100
   0.1
                       1.0
      10.0

Exit Velocity (ft/sec)
                                                               TOD
1000.0
    Figure 4-3.   Region of flame stability  for 1.5  inch commercial pressure-assisted
                   head E.

-------
e
CD
CJ
01

u
  no
   99 _
   98
   97
   96
   95
   94
   93
   92
                      I
                                     I
      0123


           Heating Value/Minimum Heating Value for Stability



Figure 4-4.  Relation of  flame stability to combustion

              efficiency for commercial  1.5 inch

              pressure-assisted head  E.
                            4-7

-------
greater  than 98 percent  are achieved  under stable  operating conditions.
Stable operating conditions are those for which the  ratio  of heating value of
the gas  to  the minimum heating value required for  stability is greater than
1.3.   Combustion efficiency rapidly decreases  as  the limit of stability is
approached.

4.3       Pressure-Assisted Head F

     Figure  4-5 shows  the  stability curve for pressure  head F.  The stability
limit  curve is significantly lower than those curves  for previously  tested
heads.   Figure 4-6 shows that combustion efficiency can be correlated to
flame  stability.  Combustion efficiency is greater than  98 percent when the
gas heating value ratio  is  greater  than  1.3,  but decreases rapidly as the
heating value ratio approaches 1.0.

4.4       Air-Assisted Head G

     The air-assisted head  G  performed differently from the steam-injected
and pressure-assisted  heads.   Figure 4-7  shows a  poor correlation of flame
stability with gas heating  value and  exit velocity  when the flare is air-
assisted.   The air-assist flowrate influences flame stability.  The minimum
gas heating value required  to produce a stable flame increased as the flow
rate  of  assist air increased relative to the propane flowrate.  A reasonable
correlation of limiting  gas heating value  to exit  velocity is obtained only
when there is no air-assist flow.

     Results  of stability tests  with  one  pilot flame  (natural gas at 2.1
SCFM)  also  show a poor correlation of flame stability  with gas heating value
and exit velocity (Figure  4-8).  Also, use of the pilot flame allows a stable
flame  to be maintained at a  lower gas heating value (including the pilot gas
contribution) than without the pilot, at the same gas  exit velocity and air-
assist rate  (Figure 4-9).  As in the tests conducted without the pilot flame,
the air-assist flowrate  influences flame stability.
                                      4-8

-------
.p-
 I
             CO




             OJ
   1200





   1100





   1000





    900





    800





    700





    600
£   500

CD
                 400




                 300




                 200





                 100
                    0.1
                                        1.0
                                                10.0



                                            Exit Velocity (ft/sec)
                                                                                 100.0
                                                                                                     1000.0
                     Figure  4-5.   Region  of flame stability for 3.8  inch commercial  pressure head F.

-------
       100
       98 -
       96
    <£  94
     u

     01

     u   92
     c
     o

     £  90
        88
        86
        84
                        O
                         I
                                        I
          0              1               2


          Heating Value/Minimum Heating Value for Stability
Figure 4-6.   Relation of flame  stability  to combustion
              efficiency for commercial 3.8  inch pressure-

              assisted head F.
                             4-10

-------
1300


1200



1100


1000


 900



 800


 700


 600



 500


 400


 300



 200


 100
                       547
                                                                                   1.9
                                                                                   3.9
                                                                            (2)4.7
                                                             1b  air/lb  propane
                                                                           Stability Curve
                                                                           with No Air-Assist
                                                0
                                            J_
                                                                  I
                                                                                        I
0.1
                          1.0
      10.0

Exit Velocity (ft/sec)
                                                                100.0
1000.0
            Figure 4-7.   Region of flame stability  for 1.5  inch  commercial
                           air-assisted flare G  without pilot flame.
                                           4-11

-------
i-»
ro
                   1200



                   1100



                   1000



                   900



                   800



                   700

                   600
               I  500
                   400


                   300


                   200



                   100
                       0.1
                                             1.0
                                   24<1   Stability curve
                                       with no air assist
      10.0

Exit Velocity (ft/sec)
                                                                                     100.0
1000:0
                                    Figure 4-8.   Region of flame stability for  1.5 inch
                                                   commercial  air-assisted  flare  G  with
                                                   pilot flame.

-------
3
4->
CO
1100


1000



900



800


700



600



500


400



300



200



100
              D Without Pilot

              £) With  Pilot

              1.5-in. Air Assisted Head G
                0 Ib Air/Ib Propane
              10.0
                                            With Pilot
                                100.0

                         Exit Velocity  (ft/sec)
                                                            I
                                                           1000.0
   Figure 4-9.  Region of flame  stability for 1.5  inch
                  commercial air-assisted head G, with no
                  air-assist.
                                     4-13

-------
       A considerable amount  of  data  was collected on the  air-assisted  flare.
However,the available data does  not allow  a systematic evaluation of the  per-
formance of air-assisted flares because factors controlling  this  flare were not
recognized prior to  the test  program  and  the  experimental  and  data analysis
procedures were not properly  designed to establish definite conclusions concern-
ing air-assisted flare  performance.  The hypothesis  on the performance of  the
air-assisted head  that  is presented  below is  based  upon  this  test data  and
requires further data collection  for verification.

       The results show that  the  degree  of air-assist affects the flare  flame
stability.  The flame will  become unstable  when the gas velocity  is not  reduced
to the  flame  velocity before the  gas  is diluted below its lower  flammability
limit.  The gas velocity is  reduced by entrainment with the surrounding  assist
and ambient air.   Large differences between the  co-axial   air-assist  and  relief
gas stream will increase mixing  (entrainment)  of air with the relief gas,  but
will inhibit the reduction of the gas  velocity to the flame  velocity when the
air-assist velocity is higher than the  relief gas velocity.   This causes  rapid
relief gas dilution and narrows the region of flammability.  With  no air-assist,
flame stability at a  given velocity is a function of the gas heating value as
shown in Figure 4-9.  When air-assist  is  applied, the  dominate factor affecting
flame stability is the relative amount  of air  assist to relief gas.

       The momentum  ratio  of the  air-assist   stream  to  the  fuel  gas   stream,
(pv)air/((0v)gas, is  a measure of the  amount of  shear  between the air and  fuel
and controls the entrainment  rate.  Figure  4-10 compares the maximum  relief gas
velocity allowable for stability  to the momentum ratio.  The correlating  param-
eter is chosen  so  that  it equals  1 when the  air momentum is  zero.  For  tests
using no air-assist,  the maximum  velocity  generally increases with  increasing
gas heating value,  as indicated  by the  vertical  line at  1.0  on the  x-axis.
When air-assist is applied in flame operation  the dominate factor affecting the
flame stability is the momentum ratio  of air-assist to relief gas  streams.   In
fact, the  two  tests   (No.  222 and Mo. 228)  with  the  highest momentum  ratios
(99.4 and 43.5) exhibited the lowest maximum exit velocity,  even  though the gas
heating value was high (2,350 Btu/ft3).
                                      4-14

-------
   1000
    100
s_
o
o
o

-------
     The combustion  efficiency  of the  air-assisted head may  be  correlated
directly to the air-to-gas momentum  ratio  as shown in Figure 4-11.   Combus-
tion efficiency is 99 percent or greater when the momentum ratio  is less  than
about 0.39 for tests conducted  without  the pilot.   For tests with  the pilot,
>99 percent combustion efficiency is  attained when the momentum  ratio  is  less
than 0.25.  Figure 4-11 also shows curves indicating air-assist to  fuel stoi-
chiometric ratio  (S.R.)  for  the  air-assisted  head tests.  Combustion effi-
ciency is greater  than  99 percent when S.R. is less than 0.7, but  drops  rapidly
to less than 90 percent (without pilot)  when S.R.  increases to  1.0.   Limited
data for  the  pilot tests  indicates  a decrease  in combustion  efficiency  to
less than 55 percent  when  S.R.  exceeds  1.0.   Combustion  efficiency with  the
pilot flame is lower  than  without,  probably  due to the wide stability range
achieved with  the pilot  flame.  Much  of this  data was  taken  outside  the
normal  operating  range of  commercial   air-assisted  flares  (S.R.  £  0.3)  to
establish the  limits  of  efficient operation  for air-assisted  flare  flames.

     Figure 4-11  is presented as  a mechanism  for discussing the  relationship
between combustion efficiency and the air/relief gas momentum  ratios for  air-
assisted flares.   Only the  data obtained during  this  study  were  used  to
develop this figure.   A full  and more accurate description of this  relation-
ship must await additional  data.

     Industrial air-assisted  flare  heads  are  typically  employed  to   reduce
soot production in flare  flames.  Since  several   of the air-assisted  head
tests exhibited low  combustion  efficiency, analysis was  conducted to evalu-
ate the emission  levels of the  incompletely  combusted products—soot, hydro-
carbons, and  CO.   Figure  4-12  shows the  combustion  inefficiency  caused  by
incomplete combustion of soot, hydrocarbons,  CO, versus  stoichiometric ratio.
Soot combustion efficiency  was   very  high,  greater  than  99.9  percent,  even
when total combustion  efficiency was low  (30-40  percent).  Unburned hydro-
carbons were the  major incomplete burned  products at high and low values  of
combustions efficiency.  Carbon monoxide  was  the  second major contributor
                                      4-16

-------
   99.9
c
O)
o
>,
u
c
o
   99.0
   90.0
             O  Without Pilot


             £>  With  Pilot
             S.R.
  Air-Assist to Fuel

  Stoichiometric Ratio
          5.0
                                                                0.1
         01


         -O
                                                                1.0
                                                                     s_
                                                                     o
                                                                     
-------
   99.99
 Carbon Monoxide Combustion  Eff. w/pilot
                          D Hydrocarbon  Combustion Efficiency
                          H Hydrocarbon  Combustion Eff. w/pilot
                          Osoot Combustion Efficiency
                          ^>Soot Combustion Efficiency  w/pilot
                          A Total  Combustion Efficiency
                          A Total  Combustion Efficiency w/pilot
                      Carbon  Monoxide Combustion  Efficiency
                          O
                                  Hydrocarbon Combustion Efficiency
                                                                             0.01
                                                                             o.i  •-
                                                                             1.0
                                                                             10
                                                                             100
               1.0             10.0           100.0
                   Air-Fuel  Stoichiometric Ratio
                                                                       1000.0
                                                                                  +J
                                                                                   C
                                                                                  -a
                                                                                  
-------
to incomplete combustion, but CO combustion.efficiency was still  relatively
high,  ranging  from  95-99.8 percent.

4.5       Commercial Head Summary

     Flame  stability and  combustion efficiency were measured for four
commercial  flare heads.   For the coanda  steam-injected  head and  the  two
pressure  heads, flame stability was  related to relief gas heating value  and
exit velocity,  according  to  Figure 4-13.  The pressure assisted head F is
capable of stable flaring of propane-nitrogen gas mixtures with lower  heating
values  than  the  other heads.

     Combustion efficiency was measured for the steam-injected and pressure-
assisted  heads  at operating conditions near and above the stability limit as
defined for each head in Figure 4-13.  As Figure 4-14 shows, high  combustion
efficiency, 98  percent or greater, was obtained for each flare head when  the
heating value ratio was  greater than about 1.3.  Combustion efficiency  was
high when  the  flames were  operated within  the limits  defined by their
respective  stability curve.  Operations at  conditions on the edge  or  outside
the stability  envelope resulted in a rapid decrease in combustion efficiency.

     Flame stability  and combustion  efficiency  measurements for the air-
assisted  head required  a  different treatment.  The momentum ratio  of air-
assist  to fuel  streams  was  found to be the  major factor  affecting flame
stability; the relief gas heating value appeared to be less important  for  the
limited tests of this series.  Therefore, flame stability and  combustion
efficiency was correlated with the air-assist to fuel momentum ratio (Figures
4-10 and 4-11),  for tests with and without pilot flame.  Limited tests of  the
air-assisted  head were  conducted using pilot flame to increase  stability.
Tests without  pilot flames indicated high combustion efficiency (>99 percent)
when the  momentum ratio is less than 0.39  (stoichiometric ratio (S.R.) less
than about  0.7).  For tests  with the pilot flames, combustion efficiencies
greater than 99  percent were obtained  at momentum ratios less than  0.25  (S.R.
less than  about  0.6).
                                    4-19

-------
i
ro
O
CO

01
1200

1100


1000

 900

 800

 700

 600


 500


 400

 300

 200

 100
                               k  Coanda Steam-Assisted Head D
                               Q  Pressure-Assisted Head E
                               Q  Pressure-Assisted Head F
                           o.i
                                                             •Pfc	
                             1.0
                                               10.0
                                       Exit Velocity (ft/sec)
                                                                                           100.0
                                                                                             1000.0
                                       Figure 4-13.   Region  of flame stability for steam and
                                                       pressure-assisted  heads  D,E,  and  F.

-------
   100
    98
    96
£
0)
U


I   94
o
c
O)
LU
0
o
    88
    86
    84
                         I


0  92f-                  !

                         I
                              12 in. Coanda Steam- Injected
                         |            Head D
                         .  D> 1.5 in. Pressure-Assisted Head E
    90U                  1
^                        1  O 3.8 in. Pressure-Assisted Head F
                         1  © 3 in. Open Pipe Head with Pilot

                         i

                         i

                         i
      0                  1                  2        2.5

      Gas Heating Value/Minimum Heating Value for Stability


      Figure 4-14.  Combustion efficiency vs. flame stability for
                    steam-injected and pressure-assisted flare
                    heads.
                                  4-21

-------
5.0       GAS  COMPOSITION TEST RESULTS

     The second  major objective of this  program was to evaluate the  effects
of relief  gas composition  on flare combustion and destruction efficiency.
This portion  of the program included Tasks  2 and 3 of the test program.   In
Task 2,  destruction efficiency, soot production, and  ignitability were
measured in  screening  tests conducted  on  the lab-scale  Flare Screening
Facility (FSF) for 21 different compounds.  Of these 21 compounds,  only  six
exhibited  any suggestion  of flaring difficulty.   In Task 3, combustion  and
destruction  efficiencies of three of the six compounds identified in Task 2
as potentially having  combustion problems were tested on the pilot-scale
Flare Test  Facility  (FTF).  Hydrogen sulfide, although not screened,  was also
selected  for testing on  the FTF.  Unique  sampling procedure problems
prevented completion of the H2S tests.  Reliable sample techniques have since
been  developed, and  H2S destruction efficiency tests are planned for  the next
series  of tests.

5.1       Compound Selection

     Compounds screened were selected based on the potential  flaring  problems
of the  compound  and the extent to which the  compound is industrially  flared.
Information  was collected from literature,  industrial sources, and  the EPA.
Final compound selection was accomplished with the advice of the Technology
Advisory Committee and the EPA Project Officer.

5.2       Compound Screening Tests

     Potential flaring difficulties of 21 different compounds were evaluated
on the  FSF.  The  compounds tested v/ere representative of aliphatic, aromatic,
sulfur,  nitrogen,  chlorinated, oxygenated, and low heating value compounds
industrially  flared.   Each  compound was either mixed with propane and/or
nitrogen gas,  or  tested as a pure compound.

     Table 5-1 presents the results of the screening tests.   The nozzle used
for  these  tests  was a 1/16 inch O.D.  (0.042  inch I.D.) stainless steel tube.
                                   5-1

-------
                                Table 5-1
      RESULTS OF  SCREENING TESTS ON FLARE SCREENING  FACILITY
Compound
Acetylene
Ethyl ene
Propylene
l,3-8utadiene
Butane
Propane
Propane
Benzene
Toluene
Chlorobenrene
Carbon Monoxide
Carbon Monoxide
Carbon Monoxide
Acetone
Acetaldehyde
Ethyl ene Oxide
C02 Diluent
Methyl Chloride
Ethylene Dichloride
Vinyl Chloride
Methyl Mercaptan
Acrylonitrile
Hydrogen Cyanide
Ammonia
Ammonia
Gas
Composition
t*\
(a)
Compd
100
100
100
100
100
100
75
1.50
1.50
1.15
100
20
17
1.43
2.07 .
1.42
7.58
9.17
1.43
0.11
10.7
1.47
0.013
100
20
Propane 1 Nj
0
0
0
0
0
0
0
98.5
98.5
98.35
0
30
37
98.57
97.93
98.58
92.42
90.83
98.57
99.89
89.30
98.53
99.99
0
80
0
0
0
0
0
0
25
0
0
0
0
0
46
0-
0
0
0
0
0
0
0
0
0
0
0
Velocity at
Stability
Limit ,
(ft/ sec)1
854
443
184
127
58
143
48
61
61
58
108
30
59
58
58
93
65
58
31
65
58
78
74
Lower
Heating
Value,
(Btu/ft3)
1475
1580
2300
2730
3321
2350
1763
2370
2381
-2350
OE2
(*)
99.99
99.91
99.98
99.93
99.99
99.98
99.97
99.59
99.99
99.49
ICould Not Iqnitel
1943
923
2347
2331
2337
2171
2212
2335
-2350
2228
-2350
-2350
99.60
79.727
99.80
99.99
96.95
NA
99.94
99.70
96.79
99.39
99.99
85.00
ICould Not Iqnitel
1967
99.90
CE3
(%)
99.97
99.92
99.93
99.93
99.96
98.18
NA4
99.95
99.90
99.95
6
99.38
99.42
99.96
99.97
99.95
99.93
99.96
99.95
NA
99.82
99.96
NA
6
NA
Soot
(mg/m3)
<1.5
<1.5
<1.5
755
<1.5
<1.5
<1.5
<1.0
<1.0
<1.0
<1.0
<1.0
<1.0
<1.5
<1.0
<1.0
<1.0
<1.0
<1.0
<1.0
<1.0
<1.0
<1.0
1 = Nozzle ID = 0.042 inches
2 « Destruction Efficiency
3 = Combustion Efficiency
4 = Not available
5 = Without steam or air assist
6 = On 1/16 inch nozzle  without pilot flames
7 = DE calculated assuming no CO originated
    from propane
                                       5-2

-------
Combustion  efficiency (CE), destruction  efficiency (DE), and soot production
were measured at the stability limit conditions for each gas or gas  mixture
on the  1/16 inch nozzle.  Most compounds demonstrated high CE and DE  and  low
soot production.  Six compounds did not.   Two  compounds, pure carbon monoxide
and pure  ammonia,  could not be ignited  when  flared on the 1/16 inch  nozzle.
Carbon monoxide  is  known  to be difficult to ignite  in  the absence of
hydrogen.   Also, ignition,  but not combustion,  of ammonia is known  to be
difficult.   One compound,  1,3-butadiene,  resulted in high soot production.
Four compounds or mixtures of compounds resulted  in low DE.  These gases were
CO-N2  mixture (79.72 percent DE), ethylene oxide (96.92  percent),  vinyl
chloride (96.79 percent), and hydrogen cyanide (85.00 percent).

     Based  upon  poor  igni tabil i ty,  high soot production, or low DE, carbon
monoxide,  ammonia, 1,3-butadiene, ethylene oxide, vinyl  chloride, and
hydrogen  cyanide were selected as candidates for pilot-scale testing on  the
Flare  Test  Facility.   Hydrogen sulfide,  although not  screened, was also
selected for FTF  testing because of its toxicity  and industrial importance in
flaring.

5.3       Gas Mixture Flare Tests

     Pilot-scale destruction efficiency tests were conducted on four of  the
seven  candidate  compounds identified in  the FSF  Tests.  These four compounds
were ammonia, 1,3-butadiene, ethylene oxide, and  hydrogen sulfide.  The  other
three,  carbon monoxide, hydrogen cyanide, and  vinyl chloride, were not tested
in this  program  due to cost and restricted  availability.  The four compounds
were tested on the Flare Test Facility  (FTF) using a 3 inch open pipe  flare
head,  without flame retention devices, steam,  or pilot assist.  Two of  the
gases,  ammonia and hydrogen sulfide, were tested  in mixtures with propane  and
nitrogen.  This was done to increase operational  safety and reduce cost  while
matching low concentration levels of these gases  (~5 percent) sometimes  found
in industrial relief gases.  Ammonia can also be flared as a pure compound.
The propane-nitrogen proportion was varied  to maintain the gas heating  value
near the  limit of flame  stability.  Ethylene  oxide and 1,3-butadiene were
each tested diluted with  nitrogen only.  Propane was not mixed with  these
                                   5-3

-------
gases,  since  the products  of  propane combustion would have interfered with
interpretation of DE for these compounds.  The heating value was  varied for
these tests  by  varying the amount of nitrogen in the relief gas.

     Results from  testing these  gases were similar to the previous  results on
commercial  flare  heads.  By varying the gas exit velocity and heating value,
the  stability  curve was defined for each mixture (Figure 5-1).  The velocity
for  these  stability tests  ranged from 0.15 to 139 ft/sec,  although most of
the  tests  were conducted  at velocities between 0.5 and 10  ft/sec.  The gas
heating  value  was varied between 145-877 Btu/ft3.  The stability  curves for
the ammonia  and hydrogen sulfide mixtures were within the region of stability
for propane-nitrogen mixtures.   This is not  surprising, since the majority of
combustible gas  in those  tests was propane.   Stability  curves for  1,3-
butadiene  and  ethylene oxide  are lower than  for the ammonia-propane and
hydrogen sulfide-propane mixtures.

     Physical  properties of these compounds are shown in Table 5-2, and can
be  used to help explain  differences in  the  stability curves.  A  lower
stability curve is indicative of increased flame stability for that compound,
due to flammability in a wider concentration range and/or higher flame speed.
The  flammability range for ethylene oxide is very wide, from 3  percent to
100  percent,  and the calculated adiabatic flame temperature (4,038 R) is
slightly higher  than that  of  propane (3,838 R).  Typically, the  higher the
flame  temperature, the higher the flame  speed.  Although  the flammability
limits  for  1,3-butadiene  differ only slightly  from propane flammability
limits, the  adiabatic flame temperature (4,105 R) and flame speed are higher.
In  addition,  the high lower flammability limits and lower adiabatic flame
temperature of ammonia can be  used to explain why pure ammonia could not be
ignited in  the  screening tests.

     Destruction  efficiency tests were  conducted  for each gas  mixture at
operating  conditions near the  stability limit.  Measurements were thus made
under  conditions of potentially low combustion and destruction efficiency.
Tests  were  also conducted under more optimum conditions within the operating
envelope to  measure  efficiencies more  representative  of normal  flare
                                    5-4

-------
en
i
en
1200




1100



1000




 900




 800




 700
                                            1	\

                        *3  Ammonia, 1.5-4.4 Percent,  in Propane-Nitrogen Mixture

                         <3  1,3-Butadiene Mixed with Nitrogen


                         V  Ethylene Oxide Mixed with  Nitrogen

                         n  Hydrogen Sulfide, 1.5-7.6  Percent  in Propane-Nitrogen  Mixture
               c

               +J
                  600
                  500
                                                                   10.0

                                                             Exit Velocity (ft/sec)
                                                                       100.0
                                                                                                                1000.0
                                Figure  5-1.   Region of flame stability for the  3 inch open pipe
                                                flare head burning  selected  relief gas  mixtures.

-------
                                                                Table  5-2

                                 PHYSICAL  PROPERTIES  (1  ATM,  60°F) OF THE COMPOUNDS  TESTED
                                                    USING THE FLARE  TEST  FACILITY
en
 i
Name .
Propane
Anwonia
Ethylene Oxide
1,3-Butadiene
Hydrogen Sulfide
Formula
C3H8
NH3
(CH2)20
(C4n6)
H2S
M.W.
44.097
17.031
44.053
54.092
34.076
Spec. Vol.
(ftVlb)
8.606
22.28
8.62
7.016
11.14
Lower Htg
Value (Btu/ft3)
2350
359
1349
2730
588
Flammability
Limits in Air (%)
Lower
2.1
15.50
3.0
2.0
4.3
Upper
10.1
27.00
10.0
11.5
45.5
Adlabatic
Flame
Temu.
(R)1
3838
3505
4105
4038
3338
                                                                                                                  Ignition
                                                                                                                   Temp
                                                                                                                    871

                                                                                                                    1204
                                                                                                                    NA2

                                                                                                                    484

                                                                                                                    558
                      Data From:   Balzhiser,  R. £., M. R.  Samuels, and J.  D. Eliasson,   Chemical Engineering Thermodynamics .  1972
                                  CRC Handbook of Tables for Applied  Engineering Science. 2nd Ed. 1976
                                  CRC Handbook of Chemistry and Physics,  53rd Ed.,  1972-73
                                  Howard, H.  F.  and  G. W. Jones, "Limits  of Flamniability  of Gases and Vapors", USBM Bulletin 503, 1952
                                  GPSA Engineering Data Book, 1972
                                  Gas Engineers Handbook, 1st Ed.,  1965
                                  Chemical Engineers  Handbook, 5th Fd.  1973

                      1.  Calculated by  integration  of heat capacity  data
                      2.  Not Available

-------
operations.   Figure 5-2  shows the destruction efficiency of ammonia,  1,3-
butadiene,  ethylene oxide,  and propane  correlated to flame stability.  The
measured destruction efficiency varied  with  compounds.   Destruction
efficiency  results for hydrogen sulfide are  not reported because they  were
found  to  be  in error, due to collection  and  analysis problems unique  to H£S
and S02 mixtures.  The analytical and  sampling techniques have been improved         i
since  this  testing was completed and further hydrogen sulfide destruction  LiHiH*
efficiency  measurements  are planned for the next series  of tests.  The
results shown  for  H2S in Figure 5-2 are the destruction efficiency of"^^^^
in propane-H2S mixtures.

     Figure 5-3  shows  the hydrocarbon  combustion  efficiency  for the gas
mixture tests.  Since the ammonia and hydrogen  sulfide gas mixtures contained
propane support gas, the hydrocarbon combustion efficiency is the combustion
efficiency of  propane to carbon dioxide and water.  For the 1,3-butadiene and
ethylene oxide test,  no  propane was added, and hydrocarbon combustion
efficiency  is the respective conversion efficiency of  1,3-butadiene and
ethylene oxide to  carbon dioxide and water.

     Comparison  of Figures  5-2  and 5-3   shows that as  the flame stability
limit  is  approached, both destruction  and combustion efficiency decrease for
all the species tested.  This behavior is identical to previous test results
for propane-nitrogen mixtures flared with  a variety of flare heads.  Ethylene
oxide combustion efficiency was quite high, only slightly lower than ethylene
oxide  destruction  efficiency.  Small  amounts  of CO and soot were produced  by
flaring 1,3-butadiene, although both combustion and destruction efficiencies
were high.  Figures 5-2 and 5-3 also show  that when the combustion efficiency
of  the support propane was  high in  combustion  tests with ammonia-propane
mixtures,  ammonia  destruction efficiency was correspondingly high.

     Destruction  efficiencies measured on  the  FSF are not directly comparable
with those  measured on the FTF.  Figure 5-4 shows that for propane, ammonia,
and 1,3-butadiene, the destruction efficiency measured on the FTF were lower
than those  measured on FSF.  The values measured on the FTF are more closely
related  to  industrial  practice since the 3 inch flare head on the  FTF  is
                                   5-7

-------
    99.99
    99.9
OJ
o.
.§    99.0
     90.0
                 Ethylene  oxide
               1,3-
              Butadiene
               Propane
                            ^ Ammorm, 1.5-4.5  Percent in Propane-Nitrogen

                            O 1.3-Butadiene in Nitrogen

                            C7 Ethylene Oxide in Nitrogen

                            |7 Propane  in  Ammonia Tests

                            D Propane  in  Hydrogen Sulfide Tests
                                                       I
                                                                          0.01
                                                                          0.1
                                                                          1.0
                                                                               Ol
                                                                               0
                                                                               s_
                                                                               (U
                                                                               J3

                                                                               O
10.0
                                                                          100.0
          0                     1.0                   2.0


                   Heating Value/Minimum Heating Value  for Stability




     Figure 5-2.   Destruction  efficiency of  different gases,



      1 Scale is  log(lOO-DE)
                                     5-8

-------
99.9 _
u


it-
99.0
90.0
           O  1.3-Butadiene in Nitrogen


           O  Ethylene  Oxide in Nitrogen

            n  Propane in Ammonia-Nitrogen


            n  Propane in Hydrogen Sulfide-Nitrogen
      Ethylene Oxide"
     1,  3-Butadiene
                                                                                 0.1  —
                                                                                     c
                                                                                     O
                                                                                     _o
                                                                                     i.
                                                                                     a
                                                                                     u
                                                 Propane in  ammonia-nitrogen
                                       Propane in hydrogen sulfide-nitrogen
                                            Lower boundary of propane combustion

                                            efficiency for previous 3, 6,  and

                                            12 inch flare heads (1.3).
                                                                             1.0
-a

-------
        10.0
    99.99
c
o
o
i-
OJ
0.
Ol

o
99.9
      99  _
                Unburned Combustible (percent of  initial combustible)

                         1.0                   0.1
                              I
            Propane
                Ammonia
                Propane
                1,3-Butadiene
                No
                          RELIEF GAS  COMPOSITION (PERCENT)
 75,100    11.6 - 48.7
Remainder   Remainder
                           20
                           80
                            0

                           100
                            0
            Ethylene Oxide    1.42
            Propane        Remainder
           1.5 - 4.4
          13.2 - 31.1
           Remainder

           5.3 - 11.1
           Remainder

          13.0 - 13.2
              0
           Remainder
                                                      O
                             99
                                                 99.9
                                                                        0.01
                                                                    o  -
                                                                          0.01
                                                                               3
                                                                               .0
                                                                          0.1
                                                                               e
                                                                               a
                                                                          1.0
                                                                         10.0
                             FSF Destruction Efficiency  (Percent)
                                                             1
                                                                       99.99
    Figure  5-4.   Pilot-scale  FTP destruction efficiency results
                    compared to  lab-scale FSF destruction  efficiency
                    results.  All  results at  operating conditions near
                    the stability  limit.   FTF nozzle size  = 3  inches.
                    FSF nozzle size =  0.042  inches.
      Scale is log(lOO-OE)
                                            5-10

-------
aerodynamically similar to  that  of industrial flares,  while the small 1/16
inch diameter nozzle used in  the FSF is not.   Destruction efficiencies
measured  on the FSF can be  used to judge the relative destruction efficiency
of different  compounds.  Values of absolute destruction efficiency measured
on the FSF  can not be extrapolated  to estimate  emissions from industrial
flares.   Also, although poor  flaring performance for  these gases has been
shown on  the  lab-scale FSF,  in  industry these  compounds are successfully
flared by employing combustion  enhancement techniques such as steam, air, or
pressure-assist, support gas or pilots,  or flame retention devices.

5.4       Flame Stability Correlations

     It has been shown in Section  5.3  that high combustion efficiency can be
expected  for  the  conditions  and  gas  mixtures tested  in this study if the
flame is  above the stability  limit  (Figure  5-3).   Pohl, et al. (5.1, 5.2)
Noble, et al.  (5.3) reached  the  same conclusion.  As shown in Figure 5-1, the
gas heating  value at the flame stability limit can be different for different
gas mixtures.   For example,  the  lower stability limit of propane-nitrogen
mixtures  flared at 10 ft/sec  using  a 3 inch open pipe flare is around 500
Btu/ft^.  At  the  same conditions, the stability limit  heat content of 1,3-
butadiene is  only 180 Btu/ft3.   Clearly, gas heating value is not the only
factor affecting flame stability.

     The  flame stability  limit is  approached  when  the flame velocity
approaches  the relief gas  velocity.   The determination  of flame velocity is
very complex,  involving reaction kinetics and mixing  (5.4).  The inherent
complexities discourage a direct evaluation of flare flame kinetic and mixing
rates.    However,  empirical  relations between  gas parameters and  flame
stability may prove tractable. Besides the heating value of a given gas or
gas mixture,  there are (1)  adiabatic  flame temperature, (2) upper and lower
flammabi 1 ity  limits in air,  (3)  minimum ignition temperature, (4) maximum
flame velocity, and (5) bond energies.  Unfortunately,  much of this data is
unavailable  for many gases and most gas  mixtures flared.
                                   5-11

-------
     A simplistic  approach is to relate flame  stability to flame temperature
as a  surrogate for reaction kinetics,  diffusion,  and flame velocity.  Is it
reasonable  to  assume that the flame velocity  depends upon temperature, since
reaction  kinetic  rates are strongly temperature dependent.  A high calculated
adiabatic  flame temperature for a gas  or gas mixture would theoretically
indicate  a  fast flame velocity and a correspondingly low gas heating value at
the stability  limit.

     The  relationship between adiabatic flame  temperature and limiting nozzle
exit  velocity  for  the relief gas is shown in  Figure 5-5 for the gases tested
in this  study.  The stability correlations  are different for propane, 1,3-
butadiene,  and ethylene  oxide mixtures.  This shows that flame temperature
correlates  the limiting  velocity at  stability for a single gas but cannot
correlate  the  stability limit of different gases.  Both l,3-butadiene-N2 and
ethylene oxide-N2 mixtures  can be stably flared at a lower heating value
(Figure  5-1)  and  flame  temperature than can propane-N2 mixtures, with the
same  exit  velocity.  Evidently, 1,3-butadiene and ethylene oxide are more
reactive  with  air than propane.

     Another measure  of reactivity  of  a  compound with air is the range of
flammability  of that compound in air.   Table 5-2 shows that the range of
fl ammabi1i ty  for  both ethylene oxide  and 1,3-butadiene are greater than for
propane.   A flame stability correlation  has  been  developed by Noble, et al.
(5.3) that  includes both  a  measure of  flame temperature and the range of
flammability limits in air for a gas or gas mixture:
                                                                  5-1
     where   N  =  "Experimental Index"
           LHV  =  lower heating value of the relief  gas  (Btu/ft3)
             Q  =  enthalpy at 2700 R of stoichiometric products of
                 combustion of one cubic foot of the relief gas
                 (Btu/ft3)
                                     5-12

-------
   1000
o
cu
    100
-O
fO
+J
CO


I

>,
4J
o
o    10
0)
    1.0
    0.1
                 Hydrogen
                 Sulfide and
                 Propane
                         I
                                          I         I
                                    3-Inch Open Pipe Flare
                                     Propane - Nitrogen

                                     Ammonia - Propane -  Nitrogen

                                     1,3 - Butadiene - Nitrogen

                                     Ethylene Oxide -Nitrogen

                                     Hydrogen Sulfide - Propane  -Nitrogen
                                   Ammonia and Propane
                                        "o\-
                                                            butadiene
                                                   Ethylene oxide
                                  I
                                          j_
                                                   I
                                                            I
      2.4       2.6       2.8       3.0      3.2      3.4

                              Adiabatic Flame Temperature (1/R) X 10
                                                            3.6       3.8

                                                                ,4
4.0
    Figure  5-5..     Calculated  adiabatic  flame temperature vs limiting
                     stable gas  exit  velocity  for different gas mixtures,
                                           5-13

-------
       UFL,LFL  = upper and lower flammability. limits of the relief
                gas in air, estimated  using calculation method  in
                Gas Engineers Handbook (5.5) for gas mixtures and
                gas flammability limits  from (5.6) and (5.7)

     The  term LHV/Q  is  a ratio  of the flame  temperature  to  a  minimum
temperature  of  2700 R, the calculated  temperature of lower limit hydrocarbon-
air flames.   The ratio UFL/LFL adjusts  the flame temperature ratio  to account
for the reactivity of the relief gas.

     The Experimental  Index, N, is  then  correlated to the limiting nozzle
mach number (the  exit velocity/sonic velocity)  of the  relief  gas.  This
correlation  is  shown in Figure 5-6(a)  for the results of Noble,  et  al. (5.3).
Darkened  points  in Figure 5-6(a)  indicate flame blow-out, and open points
indicate  a  flame is present.  The stability limit is the line separating the
majority  of darkened points  from  open points.   Figure  5-6(b)  shows the
results of  this  work compared to the  correlation line of Noble's  data.  The
closed  points  indicate  combustion efficiency less than 98 percent, and the
open points  indicate combustion efficiency greater than 98 percent.   Only the
results  of tests conducted  near  the stability  limit were  used  in  this
comparison. Even so, a difference  is  seen between the results  of  this  study
and Noble's results.  Although there  is a good deal of scatter in both data
sets,  the location and  slope of the stability  limit lines are  different.
These  differences may be  due to the difference  in head size (2  inch  vs
3 inch),  head  design, and the difference in determination of stability  limit
(point  of incipient flame-out vs point  of 98 percent CE on the two  studies.
It should  also  be noted that the parameter used does not clearly separate the
region of  stability from instability and errors of a  factor  of two  in
velocity  can be  expected.   The separation is  even less distinct when the
relationship is  used to  divide flares  with high combustion efficiency (>98
percent) from those with low combustion efficiency (<98 percent).

     The stability limit line for the  EER data is nearly vertical,  indicating
that nozzle exit velocity depends very little upon N, but that  there is a
limit  of  N  between 4 - 7, below which instability occurs regardless of exit
velocity.   This  boundary  is  more  restrictive than the boundary  defined by
                                    5-14

-------
                                             1.0
o
o
c
o
   1.0
    .9
    .8
    .7
    .6

    .5

    .4

    .3
    .25

    .2
   .15
3  .1
'G  .09
o  .08
5  .07
^  .06
x  .05
UJ

2  .04
N
tM
i  .03
    .025

    .02

    .015
   .01
        .i   i   i   i   i   i  n
        •2-In. Open Pipe Flare
        O NAT. GAS & N
           NAT. GAS &
= BURNING
DARK SYMBOL
=> FLAMEOUT
               D
                FLAME
              MAINTAINED
                REGION
                               i i i
           1.5  2 2.5 3  4  5  6789
             w -
             N -
                  2700
                     \ /UFL\
                     J (Ul)
                                             0.1
                                        c
                                        o
                                        01
                                         o
                                         o
                                            0.01
                                           0.001
                                          0.0001
                                     Flame
                                     Out
                                     Region
                                               1.0
    3-in. Open Pipe Flare

    Open Symbol,  CE > 98X
    Closed Symbol, CE98%
                                                                       (a)
                                                                             10.0
  Figure 5-6(a).
                     Nozzle  velocity
                     vs  Experimental
                     Index for data
                     from Noble,
                     et   al.(S.l).
                                                 Figure 5-6(b).
                                                         Nozzle exit velocity
                                                         vs  Experimental  Index
                                                         for  gas mixture  tests,
                                              5-15

-------
Noble's  data, where  N  can be as  low  as 1.5 at  low exit velocities.  The
region  between the two correlation lines could represent an  unstable flame
region,  where CE may be less than 98 percent even though  the  flame is still
marginally maintained.   Such a wide region of indicated flame instability and
low CE is not surprising, considering the data scatter and  that, when a flame
is near  instability, minor variations in operating or ambient  conditions can
greatly effect stability  and combustion efficiency.
                                     5-16

-------
5.5       References

5.1  Pohl, J. H., R. Payne,  and J. Lee, "Evaluation  of the Efficiency of
     Industrial Flares:  Test Results", EPA Report  No. 600/2-84-095,  May 1984.

5.2  Pohl, J.  H., J.  Lee,  R.  Payne  and B.  Tichenor,  "The  Combustion
     Efficiency of Flare  Flames", 77th Annual Meeting and Exhibition of the
     Air Pollution Control  Association, San Francisco, CA, June 1984.

5.3  Noble, R. K., M.  R.  Keller, and R. E. Schwartz,  "An  Experimental
     Analysis of Flame Stability  of  Open  Air Diffusion  Flames",  AFRC
     International Symposium on Alternative Fuels  and Hazardous Wastes,
     Tulsa, OK, 1984.

5.4  Burgess,  D. and M.  Hertzberg, "The Flammability Limits of Lean  Fuel  Air
     Mixtures: Thermochemical  and Kinetic  Criteria for Explosion  Hazards",
     ISA Transections  14(2), p. 129, 1974.

5.5  Gas Engineers Handbook, First Edition, Industrial Press, p. 2/75, 1965.

5.6  Zabatakis, M. G., "Flammability Characteristics of Combustible  Gases and
     Vapors",  U.S.B.M. Bulletin 627, 1965.

5.7  Coward,  H. F. and  G. W.  Jones, "Limits of  Flammability of  Gases and
     Vapors",  U.S.B.M. Bulletin 503, 1952.
                                  5-17

-------
6.0       FLARE NOX AND HYDROCARBON EMISSIONS.

     Emissions of  NOX  and hydrocarbons were  measured in conjunction with
measurements of combustion and  destruction efficiency.  Figure 6-1 shows NOX
concentration  (on  an air-free basis, 0 percent 02)  at the plume center!ine
for  the tests  of  this  study.   This figure show only  a  vague general
relationship  between NOX  formation and combustion  efficiency; NOX
concentration increases with  increasing combustion efficiency for most  flare
heads and  gas mixtures.  Results for the coanda steam-injected head are
contrary  to  this  trend,  but NOX emissions for this  head were measured only
over a  narrow combustion efficiency range (99.6-99.9 percent).  The poor
correlation between NOX concentrations  in the plume and combustion efficiency
results from  some  variables influencing NOX formation  and combustion
efficiency differently.  Figure 6-2 shows the correlation of total NOX
emissions with  flare heat  release  rate.  This figure  shows a trend of
increased  M02  emissions per  10^ Btu with increasing  heat release for all of
the flare  heads  and gas mixtures.  The values  of NO  and N02 were determined
by radial  integration of local fluxes.  Emissions of NOX were almost entirely
NO except  for tests of the ammonia gas mixtures.  The reported N02 emission
levels  for  the ammonia tests are generally between 2-100 times as high as N02
emissions  of  the  other tests.  The  highest NOX emission level was for the
ammonia mixture  and was  under 1 Ib  N02/10^ Btu.  NOX emissions from  other
tests  were  typically less  than 0.1 Ib N02/10^ Btu for  hydrocarbon gas
mixtures.

     Limited qualitative  and semi-quantitative hydrocarbon emissions were
measured for the gas mixture tested.  Hydrocarbon emissions were measured for
1,3-butadiene using the Flare Screening Facility.  Table 6-1 shows that very
low concentrations of hydrocarbons were detected in  the plume samples.' The
concentration measurements  are accurate to within one  order of magnitude.

     Hydrocarbon emission measurements were  also made for the gas mixture
tests on  the  Flare Test  Facility.   Results are shown in Table 6-2.   These
results have  been  corrected  for background air hydrocarbon levels. Many of
the species detected were  present in very low concentrations.  A few species

                                  6-1

-------
   1000
    100
g.
CL



-------
                    Legend for  Figure  6-1
    Coanda 12 in.  Steam Injected Head  D
Q  1.5 in.  Pressure Assisted Head E
Q  3.8 in.  Pressured Assisted Head F
£)  1.5 in.  Air-Assisted Head G
£  1.5 in.  Air-Assisted Head G with Pilot1
€)  3 in.  Open Pipe Head with Pilot1
^3  3 in.  Open Pipe Head, Ammonia-Propane-N2  Mix
O  3 in.  Open Pipe Head, 1,3 Butadiene-N2 Mix
V  3 in.  Open Pipe Head, Ethylene Oxide-Ng Mix
^  3 in.  Open Pipe Head, Hydrogen Sulfide-Propane N2  Mix
     Heat  release  includes  pilot when operating
                             6-3

-------
 CO

^o
 o
      10  _
      10
      10
        -5
         0.01
0.1            1.0
                          Heat Release  (10  Btu/hr.;
10.0
                                                              100
         Figure  6-2.   N02 emissions from pilot-scale flares,
                                  6-4

-------
                    Legend for Figure 6-2
IX   Coanda 12 in. Steam Injected Head D
Q  1.5 in. Pressure Assisted Head E
O  3.8 in. Pressured Assisted Head F
&  1.5 in. Air-Assisted Head G
£l  1.5 in. Air-Assisted Head G with Pilot1
€>  3 in.  Open Pipe Head with Pilot1
Q  3 in.  Open Pipe Head, Ammonia-Propane-N2 Mix
O  3 in.  Open Pipe Head, 1,3 Butadiene-N2 Mix
O  3 in.  Open Pipe Head, Ethylene Oxide-N2  Mix
\  3 in.  Open Pipe Head, Hydrogen Sulfide-Propane-^2  Mix
     Heat  release  includes pilot when operating
                             6-5

-------
                                   Table 6-1.

        GC-MS  ANALYSIS OF PLUME SAMPLE FROM  LABORATORY-SCALE TEST
Compound
Acenaphthylene
Benzaldehyde
Benzofuran
Biphenyl (or acenaphthene)
Dibromomethane?
Ethenyl benzene
E thy nyl benzene
E thy nyl -me thy! benzene
Methyl naphthalene
Naphthalene
Phenol
Tetrachl oroethene^
Toluene
Estimated Concentration (ppm)
224 Percent
Excess Air
0.0002
0.005
0.001
0.0004
0.0002
0.005
0.01
0.01
0.0006
0.01
0.005
0.0005
0.01
0 Percent
Excess Air
0.0006
0.02
0.001
0.001
0.0006
0.02
0.03
0.03
0.002
0.03
0.02
0.002
0.03
Potentially
Hazardous^




X




X
X
X
X
^Listed  as hazardous under Appendix 8 Regulation,  EPA (Hazardous Waste and
 Consolidated Permit Regulations), Federal  Register 45:98 (May 19, 1980) and
 Federal  Register 45:138 (July 16, 1980).


^Result of contamination.
                                     6-6

-------
                               Table  6-2

         GC-MS ANALYSIS  OF PLUME CENTERLINE SAMPLES  FROM
         GAS  MIXTURE TESTS, USING  A 3 IN.  OPEN  PIPE  FLARE
Compound
Propane
Methyl Chloride1
Methyl Ethyl Ketone
1,1,1-Trichloroe thane
1,2-Dichloropropane
Toluene
Butylcellosol ve
(2-butoxyethanol )
Xylene
Trichloroethylene
Thiophene
Hexane
Tetrachloroethene
Methyl Cyclohexane
Methyl Bromide1
Benzene
1,3-Butadiene
4-Vinyl Cyclohexane
Combustion Efficiency (%)
Gas Htg Value (Btu/ft3)
Gas Comp. Propane (%)
Nitrogen (%)
(Other) (%}

Exit Velocity (ft/sec)
Dilution Factor (DF)
SOg Tracer Used
Approximate Concentration (air-free, 0 percent 02) ppm
Test No. 235
5
0
300
0.3
0
0.4
0

0.4
0
1.8
0.1
TR<0.009
~0
TR<0.009
TR70.02
~0
0
92.1
416
19.2
78.3
2.0 (NH3)

9.56
45.4
Yes
No. 289
200
0
200
4
8
1
0

0
2
5.2
0.2
0.2
0.1
0.02
0.07
0
0
99.2
539
28.7
63.7
7.6 (H2S)

8.04
69.5
No
No. 292
40
100
300
0
6
1
2

1
0
20
1
1
0.6
0.06
0.3
2
5
99.8
145
0
94.68
5.32 (1,3-
butadiene)
6.87
60
Yes
No. 295
0
400
50
0
1
4
0.1

0.4
0.2
10
0.4
0.4
0.4
0.07
0.1
0
0
99.5
175
0
87.0
13.0 (ethylene
oxi de )
4.53
36.4
No
NOTE:  Approximate plume concentrations are  found by dividing the air-free
      value by (DF + 1)

  Probable contaminant

2 Trace detected, but below indicated  minimum measurable concentration level.
                                   6-7

-------
were measured in significant  amounts in one  or more of the  samples.  Of
these,  the  most common were  propane, methyl chloride,  and  thiophene.  The
compounds,  1,3-butadiene,  butyl  cellosolve,  and 4-vinyl cyclohexane were
found only  in the test of  1,3-butadiene.  No nitrogen-bearing  hydrocarbons
were detected  for  any of the samples  including tests of ammonia  doped flames.
Only one sulfur-bearing species  (thiophene) was detected.   It  was detected in
all the  samples.  All but one  test  included either H2$ or S02  tracer in the
relief  gas,  possible sources  for  thiophene.  The source  of thiophene in the
other test  (No. 255) is  unknown, but the thiophene level in  this test was
very low and  may be contamination.  Many chloride-containing  species were
detected in  varying amounts.   These species  are thought  to  result  from
contaminants  in the relief gas or in  the sampling and analytic  procedure as
no chlorine  compounds were intentionally introduced to the flames or samples.
                                   6-3

-------
                                APPENDIX A

               EPA FLARE TEST FACILITY AND TEST PROCEDURES
A.I       Flare Test Facility

     The EPA  Flare Test Facility  (FTF), shown in Figure  A-l, was designed and
built by EER  at their El Toro Test  Site for the U.S.  Environmental Protection
Agency,  under  EPA Contract  No. 68-02-3661.  The facility was completed in
1982.

     For wind protection,  the FTF  is located in a  box canyon surrounded by
70-foot  cliffs.   The facility  includes gas delivery  systems, a flare head
mount enclosed in a framework structure supporting (1)  screens for additional
wind  protection,  and  (2)  plume sample probes,  and  a building containing
delivery  system controls  and  analytical  instruments.   The facility is
designed for relief gas flows  ranging from  10  to over  40,000 SCFH.  The
maximum  flow depends on gas  composition.  The facility is reviewed briefly
below,  and is  described in more detail by Pohl, et  al. (1.3) and Joseph, et
al. (1.11).

     Gases are delivered  to the flare and  auxiliary equipment  through
parallel  manifolds shown in Figure  A-2.  Propane, natural gas, and nitrogen
manifolds each  have three orifice meters and one small rotameter, each with
its  own  control  valve.  These manifolds were designed  to  accurately measure
and  control  a  wide range  of  flowrates.  One  additional manifold  of two
parallel  orifice meters  (not shown) is used  to measure the flow  of one
additional flare gas.  The propane,  natural gas, and  nitrogen manifolds and
flow  lines are  constructed  of  carbon steel.   The flow  line and manifold
system  for the  additional  flare  gas is constructed  of stainless steel, to
allow use of  corrosive gases such as  ammonia  and hydrogen sulfide.

     There are also similar supply  systems for steam, sulfur dioxide (tracer)
and air.  Steam is used (1) for steam-assisted flare tests, (2) in steam heat
                                  A-l

-------
                                                                                    Suppurl Slnietwt
ro
                      Adjustable Rake for
                      Sample Probes, Filters, Driers,
                      Tenax Traps,  and Bubblers
                      for H2S, S02  & NH3
                                            Figure A-l.   EPA flare  test facility  (FTP)  at  EER.

-------
      c
     L i(|U id
>
          S]
   	H>kMx3
             PR 4    V 18
             PR 1    CV 1
  City
Natu
  Gas
      'a i-HXJ—r>
-------
exchangers  for vaporizing sulfur dioxide and flare test gases,  and  (3) for
sample probe heating.   Sulfur dioxide is used as a tracer for flare  test mass
balances.   Air is  used  during air-assisted flare tests.

     These supply  systems can provide and mix propane, natural  gas,  nitrogen,
one additional flare  gas  (such as  ammonia or hydrogen sulfide) and sulfur
dioxide tracer.  Propane  is stored as a liquid in a 2,100  gallon  tank.  At
low flowrates, natural vaporization is sufficient to supply propane  gas for
flaring;  at higher  flowrates, propane-fired vaporizers are  used  to increase
the propane flowrate  up to  15,000 SCFH.  Natural gas is supplied by  the local
utility at a maximum  flowrate of 7,000 SCFH.  Nitrogen gas is used  to vary
the heating value of  the  flared  gas.   Nitrogen  is delivered  from liquid
nitrogen  cylinders to  banks of finned-tube atmospheric vaporizers capable of
providing  a maximum nitrogen flowrate exceeding 24,000 SCFH.

     The  system to  supply an  additional  gas is new  to the FTF.  Portable
cylinders of a liquefied test gas such as ammonia or hydrogen sulfide can be
connected  to this  system.   A steam heat exchanger vaporizes the compound, and
the flow  rate  is controlled and metered using pressure regulators,  control
valves, and orifice meters.  This system is constructed of stainless steel to
resist corrosion,  and can deliver up  to 4,000 SCFH of gas, depending upon the
compound.

     Steam is produced  in a 15 hp gas-fired boiler.  The boiler can  supply up
to 400 Ibs/hr of  100 psig  saturated  steam.  Sulfur dioxide,  used  as an inert
tracer, is fed from liquid  S02 cylinders and vaporized through a steam-heated
vaporizer at 7 SCFH.   Air is supplied by a  forced-draft  fan at  a  maximum
flowrate of 60,000 SCFH, at a static  pressure of 17.6 inches  H20.

     The sample collection  and analysis system  is shown in Figure A-3. Plume
samples are collected using five stainless-steel, steam-heated probes  mounted
on a  movable  rake.  Samples are collected concurrently from five  different
radial  locations in the  plume.   Pumps draw  the  soot-  and  moisture-laden
samples into the probes, where filters collect  the soot for subsequent weight
measurement.  Permapure dryers using  membrane tube bundles selectively remove
                                     A-4

-------
Hake
I)     Dryer
                                                                          Participate
                                                                                  Charcoal
Met  Sample
Bypass
    Bay Fill
                  Flow Meter
                                 Flow Meter
                                           -Bypass
                         Figure  A-3.   Flare test facility  sample system.

-------
water  vapor from the sample stream.   The  dried gas samples are collected in
Tedlar  bags for analysis of 02,  CO, C02, total hydrocarbons, S02,  and NO/NOx
content.

     Other species such as S02, H2S or NH3  are absorbed into liquid solutions
in absorption bubblers.   The concentrations of H2S and S02 are measured by
titration and NH3 concentration is measured using an ion-specific electrode.

A.2       FTP Test Procedure

     The  Flare Test Facility  (FTF) test  procedure includes  measuring
background conditions,  igniting the flame,  establishing test conditions,
sampling,  and analysis.  Tests are not conducted in rainy weather  or at wind
speeds  greater than 5 mph.   Most testing is done in the morning, when the
weather is calm.  A typical test requires  about 4 hours, although  the actual
sample period is only 20 minutes.

     Before each test, the ambient air is  sampled and analyzed for background
levels  of 02, CO, C02, hydrocarbons, S02, NO/NOx, soot, and for H2$ or NHa,
if applicable.  The flame is then ignited  using a hand-held spark  igniter or
a  Zink  igniter,  and the test conditions  are  set by  adjusting the gas
flowrates.  Most of the tests were conducted  near the stability limit of the
flame.   The flame stability limit is determined by adjusting the flowrates
until the flame becomes unstable and is eventually extinguished.

     After test conditions  are set, plume  samples are  collected for 20
minutes in order to time-average perturbations  and collect  sufficient  amounts
of sample for analysis.   Samples are collected from five different radial
locations, at a height above the flame experimentally determined to be beyond
the  flame.   If the probes are too high,  air  dilution of the samples  reduces
combustion product measurement accuracy.   If  the probes are located too low,
inside  the flame envelope,  incompletely burned samples may be  collected,
which would result  in artificially low combustion efficiency measurements.
                                    A-6

-------
     While  the plume  is being sampled,  .the  flame structure and other
characteristics  such as color are recorded visually and  photographically.
After sample  collection and flame observations are complete, the flame is
shut down.   Sample  analysis is then  conducted to measure  levels of 03,  CO,
C02,  HC,  NO/NOx, soot, and other species in  the plume samples.
                                   A-7

-------
                                APPENDIX B

                FLARE  SCREENING FACILITY AND TEST PROCEDURES
8.1       Flare  Screening Facility

     The  laboratory-scale  Flare  Screening Facility (FSF) is  used to
inexpensively,  quickly,  and easily identify potential  difficulties  in flaring
a wide  variety of compounds.  Advantages of the FSF  over the  Flare Test
Facility  are  its  small  size,  low operating cost for gases  and materials, the
ability  to obtain complete,  undiluted samples of  flare  combustion products,
the ability to close  mass  balances,  and the increased  safety for flaring
toxic  gases.

     Figure B-l  shows the FSF schematic.  This facility is an adaptation of
EER's Turbulent Flame  Reactor, originally designed to measure emissions and
combustion efficiency  of hazardous waste compounds. The  facility was adapted
to burn  either liquid  or gaseous compounds supplied  from  pressure cylinders
and metered through calibrated rotameters.  Combustion air  is  injected to the
reactor  co-axially with  the fuel stream, through a  flow straightening screen.
By maintaining  a very  low air  velocity relative to  the fuel  velocity, effects
of the  co-current air  stream  on the fuel  stream are minimized.  Test results
verify that the  flame  behaves  similarly to a jet in a  quiescent atmosphere.

     The  flame  is completely enclosed in a water-cooled reactor shell, with
sample  probes located  at  the reactor outlet.  The shell  isolates the flame
from  the  environment  and prevents air dilution of the flare products.  This
allows complete  mass balance closure over the system.

     Instrumentation  for plume sampling and analysis  is  shown in Figure B-2.
There are two separate  sample systems:  one for continuous monitors and one
for gas  chromatograph  samples.  The sample for the  continuous monitors is run
through  a heated filter and  divided into two streams.   One leads  to a total
                                    B-l

-------
    MIXING CHAMBER
  HEATED LINE
 'SYRINGE DRIVE
= CONTINUOUS  SAMPLING
  TENAX SAMPLING
                                .1/16-1/8  IN,  NOZZLE
                             SCREEN
                            GLASS BEADS
                              GAS
                       AIR FAN
Figure B-l.  Flare Screening Facility (FSF).
                     B-2

-------
       Horiba
     10-Analyzer
                    Seckman
                      02
                   Analyzer
    IVent
      Anarad
    Model AR500RI
    CO Analyzer
    Beckman
   Model 402
Total  Hydrocar-
 bon Analyzer
        Burner
                  Beckman
                 Model 864
                  Methane
                 Analyzer
Z  Rotameter
                                                       SilicaJ
                                                          Gas
                                                        OrierL.
                                                         Tenax Low Absorption
                                                                  Bubbler)
     Probes '


         To

 Electrically^
Heated  Sample
    Lines
      Figure B-2.    Flare  Screening  Facility sample  system.
                                  B-3

-------
hydrocarbon  analyzer  (flame ionization  detector) and the other passes through
a water-bath and  then to CO, C02,  City,  and Q£ analyzers.

     A second sample stream passes through a filter  into  either a solid
sorbent  cartridge or an  absorption bubbler.  The solid sorbent, Tenax and
activated charcoal, is used  to  collect and  concentrate heavy  and  light
hydrocarbons in the gas sample.   Aqueous  solutions in the absorption bubblers
collect  and concentrate  other  species in the sample, such  as HCN and NH3-
The hydrocarbon  species adsorbed in the sorbent cartridges are subsequently
desorbed  by  heating for gas chromatographic analysis and/or mass spectrometry
analysis.   Species collected  in the  absorption  bubblers are subsequently
measured  by  titration, ion-specific electrodes, or colometric  techniques.

B.2       FSF Test Procedures

     The  FSF test  procedures are  more  simple than the FTF test procedures.
Since  the facility is enclosed,  it is not subject to environmental  conditions
such  as  wind  or  rain.   The system is smaller and hence, more easily
monitored.   Probe positioning  is unnecessary, since the sample  probes are
located permanently at the reactor outlet.

     For  each test, all the instruments  are zeroed and calibrated.  After the
flame  is  ignited and the air and  fuel  flowrates are adjusted,  on-line sample
collection  and  analysis is initiated.   The effects of air and fuel flowrate
changes  on  flare emissions  can  be monitored by varying the  flowrates while
operating the on-line sample system. Solid sorbent and bubbler collection of
sampled  species  is time-averaged, however, and can only be  conducted while
air  and  fuel  flowrates  are kept  constant.   Final  on-line emission
measurements are made of plume species  after the on-line instruments indicate
that steady  state has been reached for  the system.

     Solid sorbent samples of plume species can be sealed, cooled,  and stored
for short periods of time.   Quantitative and qualitative analysis of these
samples  is  conducted by  heating the  sample  to  desorbe the concentrated
hydrocarbons  into a gas  chromatograph (GC) and/or mass spectrometer (MS).
                                   B-4

-------
Species  types and  concentration  are  determined by comparing the results to
standards.

     Bubbler samples,  depending  upon  the  sampled species, can be titrated for
HCL,H2S or $03,  or analyzed using  ion-specific electrodes for Nti-$ or HCN.
                                     B-5

-------
                                APPENDIX  C

                              DATA ANALYSIS
C.I       FTP  Data Analysis

     Data  analysis procedures for the  pilot-scale FTP tests were developed in
previous  HER  flare studies  and reported by Pohl, et al. (1.3).  Analysis of
the  pilot-scale tests  is  much more complicated than analysis  of flare
screening tests  done on the laboratory-scale FSF.  Results of FTF tests must
be corrected for  background levels of  sampled species and air dilution of the
plume.  Also, numerical integration  must be conducted using the local probe
measurements  and velocities calculated from jet theory.  These  steps are
unnecessary  in the FSF  analysis procedures, because the flare is isolated
from the  environment, and well-mixed samples  are collected at the  outlet of
the reactor.   Since the  development and details of the Flare Test Facility
data analysis  procedures are already reported (1.3), only a brief summary and
new additions will  be  reported here.  The terminology, however, has been
changed to be  more uniform and compatible with  the additions.

     Data  reduction is conducted on the FTF  plume sample results to  determine
local  air  dilution  of  the combustion  products, local  combustion and
destruction  efficiencies,  and integrated overall  average  combustion and
destruction efficiencies.  The local dilution factor is:
           DF  =  	                                              C-l
     where DF =  dilution factor = volume of  air  in the  local  sample divided
                by  the volume of stoichiometric  combustion products.
            Y =  local concentration of 63, C02,  or S02  (tracer)
                                  C-l

-------
          m = measured in plume
         af = air-free, stoichiometric basis
          b = background

     Combustion efficiency is defined as the degree to which all  fuel  materials
have been  completely  oxidized.   The local  combustion efficiency  is based  upon
local  probe measurements  of  plume  constituents,  whereas  the integrated average
combustion efficiency  is  calculated by  integrating  the  local plume  fluxes  to
obtain average compositions  of plume species.   Since local  plume measurements
are diluted by ambient air,  they must  be  corrected for the  background  levels
of plume species  in the ambient air.  These corrections  are made  using equation
C-2:
                         /OF \
          Yh c = Yh m -  I	   Yn b                                    C-2
           II ,L-    II ,111    I nc.j.1 1   " »u
     where h = plume species
           c = corrected
     Local  combustion  efficiency  (CE)  can  then  be  calculated  using  equation
C-3:
                    Z.   v.  v-
                     i    i  n ,c
          CE = 1	                                          C-3
     where \> = stoichiometric coefficient
           i = incompletely burned species
           j = completely and incompletely burned species

     Local destruction efficiency  is  similar  to local   combustion  efficiency,
but is a measure of the degree  of  destruction  of the  particular  fuel  material.
                                      C-2

-------
It is equal to  the  combustion efficiency for that species only when there  are
no incompletely  burned   intermediates,  such  as  CO  or  soot  for  hydrocarbon
species.  Local  destruction  efficiency  (DE)  for  a  fuel  species is  calculated
using equation C-4:
          DE = 1 -
                      Yk,c
                                                                 C-4
       where k = fuel  species
             1  = completely and incompletely burned species  from fuel  species.

       Integrated average combustion and destruction efficiencies are  computed
by first combining the  local  corrected  plume composition with the local  plume
velocity to  obtain  a  local  corrected  mass  flux  for  each  plume constituent.
The local  corrected plume species concentrations are found using equation  C-2,
and the local plume velocity is calculated  from jet  theory using  equations  C-5^
and C-6:
                   exP
                                                                 C-5
max
            = V0 [o.!6  /JU - 1.5J
                                                                 C-6
•Mhe coefficient preceeding  Rp/X in  equation  C-5  was  reported  as  -90 in  a
 previous EER  report  (1.3),   based  upon  jet  theory.   In   order  to   better
 match radial  profiles   measured in  the  flare  tests,  the   coefficient   was
 changed to -5.
                                      C-3

-------
     where V = velocity
           R = radial  distance from plume center!ine
           X = probe  axial  distance above flare head
           r = radial  position
         max = maximum
           o = flare  head  outlet
     Numerical  integration of the local fluxes is used to calculate average
combustion and destruction  efficiencies  using equations C-7, C-8, and C-9:
                    Er 1^  v-  YT  c  Vr Ar
           CE = 1	                                C-7
                    7  T •  \i •  Y •    V  A
                     y i.-]  v-i  i-j  p  »y. M^
                     '   J  »J  w 3 ^*  '
                       Zr Yk)C  Vr  Ar
           DE = 1	-	                                C-8
                    ^r ^ vl YlfC  Vr  Ar
           A  = IT(R+  -  R)                                           C-9
     where Ar = radial  area  sampled  by  probe  r  (Figure C-l).

 C.2       FSF Data Analysis

     Data  analysis for  the Flare  Screening Facility (FSF) test results is
much  simpler than  for the Flare Test Facility (FTF) test results.  The FSF
flare  flame is completely  enclosed within a steel reactor shell.  The inlet
fuel  and  combustion  air flowrates are metered, so the plume flowrates of
excess  air and air-free  combustion  products can be directly calculated based
upon the combustion stoichiometry  of the  gas:
                                     C-4

-------
Figure C-l.   Schematic of integration geometry.
                   C-5

-------
e.a.
                        r.a.
                                                                    C-10
     where   V = volumetric  flowrate, SCFH
         e.a. = excess air
         S.R. = stoichiometric  ratio
         r.a. = required air  for 100% combustion
                                                                    C-ll
     where
   p = stoichiometric products  of combustion with air (air-free
       basis, 0 percent 03)
   g = inlet gas
   v = stoichiometric coefficient
   i = combustion species "i"
              = V
                 e.a.
                                                          C-12
     where   t  = total  plume

     This approach assumes 100 percent combustion,  in order to determine the
excess  air,  combustion product,  and  total  plume flowrates.   The same
assumption  was  used in data reduction  of the pilot-scale  tests.  Where
combustion  is  only slightly less  than  100 percent, flowrate errors  due to
this assumption are small.   Even where combustion  is significantly less than
100 percent,  the  error in total plume flowrate  is  small, because  a majority
of  the  plume  gas  is  nitrogen,  unaffected  by combustion  efficiency
(discounting  Ng ->• NOx reactions and combustion of nitrogen-containing fuel
species).
                                   C-6

-------
     The plume is sampled  at  the  reactor exit, where the plume is well  mixed.
This eliminates the need for  collection  of  multiple  local plume  samples  across
the plume  radius,  assumptions of local  velocities  at radial locations  in  the
plume, and the  integration of local  species fluxes  to  calculate total   plume
species flowrates.  Species concentrations  in the plume  sample are  representa-
tive of  average  plume  concentrations.   Species  flowrates  in  the  plume  are
calculated using  the  measured  concentrations  and   the  plume  flowrate   found
from equation C-13:
          Vi = Vt Yi                                                      C-13


     where   Y = mole fraction

     In cases  of high  combustion  efficiency, the  plume concentration  levels
of incompletely  combusted  species  such as  CO and  hydrocarbons are near  back-
ground levels.   The  plume species  flowrates must  then  be  corrected  by  sub-
tracting the background contribution:


          VT,c = vi  - va YT jb                                            c-i4


     where   c = corrected
             b = background

     Combustion and   destruction  efficiencies  are   calculated  using equations
C-15 and C-16.
                    M  vi vi ,c
          CE = 1	—                                          C-15
                                      C-7

-------
    where    i = incompletely burned species .
             j = incompletely and completely burned species
                     Vk.c
          DE  =  i	                                          C-16
     where    k  =  fuel  species
             1  =  incompletely  and completely burned species that came from
                 the  fuel  species
C.2       References

C.I    Beer, J.M.  and Chigier,  Combustion  Aerodynamics,  Halsted Press Divi
       sion, John  Wiley and Sons,  Inc.,  New  York,  1982.
                                    C-8

-------
                               APPENDIX D

                            QUALITY ASSURANCE
D.I       Flowrate Measurement

     Accurate measurement  of the  gas flowrates is very  important in
determining  the  relief gas composition and  velocity.  Also,  since the level
of air-assist has  such a strong  impact upon the stability and combustion
efficiency  of the  air-assisted head, accurate air-assist  flowrate
measurements  are also important.

     Square-edged  orifice  plates  were used to measure  the  flare  gas
flowrates.   Each  orifice was calibrated using  air and a laminar flowmeter,
dry gas  meter,  or wet test meter to obtain an  orifice coefficient, to be used
in equation D-l for flowrate measurement:
                     /PAP t
            V = K Fa (	                                            D-l
                   9 \MWT;
     where   V = flowrate,  SCFM
            K = orifice coefficient
           Fg = gas correction factor
            P = static orifice pressure, psia
           AP = orifice differential pressure,  feet H£0 column
           MW = gas molecular weight
            T = orifice temperature, R.
     The standard  deviation of K for 21  different orifices was  less than
7.9%, and  less than 3.0% for the majority.   For air  and nitrogen, Fg  = 1.00.
For the  other  gases, Fg was determined by calibration.
                                  D-l

-------
     The air-assist  flowrate was measured using a venturi meter, which was
calibrated  by using a laminar flowmeter to determine a  flow coefficient for
use in equation  D-l.

D.2       Sample Analysis

     Accurate  sample analysis is critical for determining reliable combustion
and destruction efficiency results.  Table D-l shows the analytical  methods,
instruments, and accuracies used in this test program. The  listed accuracies
are for the  concentration ranges most typically encountered  at the Flare Test
Facility.   Accuracy for a specific method may change if  concentration levels
for the sampled  species are outside the ranges listed in  Table D-l.

     Many of  the  analytical  methods of Table D-l were developed in previous
EER flare studies  (1.3), but several additional methods were developed during
this  study.   New  techniques  were  needed for the  analysis  of previously
untested non-hydrocarbon  gas  species,  H2S and  NH3-   Also,  in order  to
qualitatively measure hydrocarbon  emissions,  a  method was required which
could analyze  very low concentrations  of hydrocarbon species in  plume
samples.

     For measuring H2S emissions, an adaptation of EPA Method 11 (D.I) was
used.  This method involved  (1)  absorption h^S  from  plume samples in gas
bubblers  of  and  (2) iodometric titration of the bubbled solution.  The method
can be very accurate, depending  upon bubbler  collection efficiency, the
aqueous  H2S concentration,  and bubbled  gas flowrate measurement accuracy.
For  this  study,  the  combined  accuracy for  measuring  gaseous H2$
concentrations  of 1-100 ppni is +_ 15 percent.  This  accuracy was not achieved
due to S02 presence in the bubbled solutions, which caused great errors when
the aqueous S02 concentration  was  high  relative  to the H2$ concentration.
Accurate techniques for determining H2S and S02 concentrations in mixtures
have been developed for future H2S-S02 analysis.

     Emission of NH3 in the flare plume was measured by  absorption of NH3  in
gas bubblers, followed by  aqueous  analysis using ion-specific electrodes.
                                   D-2

-------
                                                    TABLE  D-l

                                         FLARE FACILITY ANALYTICAL METHODS
SPECIES
°2
CO
co2
Total
Hydrocarbons
Individual
Hydrocarbons
H2S
so2
NH3
NO/NOV
A
Particulate
INSTRUMENT
Taylor 570A
Beckman 315A
Beckman 315B
Beckman 400
Tenax Cartridge
GC-MS
Titration
Titration
Melloy SA 260

Teco 14B-E
Filter
PRINCIPLE
Paramagnetic
NDIR
NDIR
FID
FID,
Spectroscopy
lodometric
Method
Perchlorate
RXN
FPD
Ion-Specific
Electrode
Chemi lumi-
nescence
Timed
Collection
RANGE
0-100%
0-2%
0-20%
0-5000 ppm
0.0002-1 ppm
1-100 ppm
1-200 ppm
0.5-10 ppm
0.1-10 ppm
0.05-10 ppm
0-10"o
lb/ftj
ACCURACY
±0.2%
±0.1 ppm
±0.02%
±0.5 ppm
One order
of magnitude
±15% of
Measured
±15% of
Measured
±1% of
Measured
±15% of
Measured
±5% of
Measured
±10% of
Measured
MEASURED
CONCENTRATIONS
18-21%
3-200 ppm
0.05-2%
3-300 ppm
0.0002-1 ppm
5-100 ppm
10-200 ppm
0.1-0.6 ppm
0.02-10 ppm
10"8-10"6 lb/ft3
o
co

-------
The accuracy of this method,  +_ 15  percent, depends on the  bubbler collection
efficiency,  the aqueous NH3 level,  and  the accuracy of the  bubbled gas volume
measurement.

     Qualitative  analysis of flare hydrocarbon emissions  is usually very
difficult,  because  of  the typically low concentrations of  specific
hydrocarbon  species in the  flare plume.  These species were collected from
plume samples  by adsorption  using solid sorbent cartridges.   High molecular
weight  molecules  were  adsorbed onto Tenax,  and low molecular  weight
hydrocarbons  were adsorbed  onto  charcoal .   The samples were subsequently
analyzed qualitatively and semi-quantitatively  by gas chromatography and mass
spectrometry.  Quantitative accuracy was low  (one order of.magnitude)  because
expense  prohibited using calibration standards.  Since  measured levels of
these species  were very  low compared to the predominant species CO, C02,
total hydrocarbons, and soot, this error had negligible impact on combustion
and destruction efficiency results.

D.3       Quality  Control Problems  and  Solutions

     Few quality  assurance problems were encountered during this program.  A
key problem, mentioned in Section  D-2, was that for a few  tests, the  sampled
concentration  levels of  various  species were outside the range of accurate
measurement  indicated in Table  D-l.   In these  cases,  the  data for those
species  was  considered invalid.   Usually the  remainder of the test data was
still  reliable.  For example,  an inaccurate soot measurement for a particular
test invalidated the soot values, but negligibly affected  the carbon mass
balance  and  efficiency calculations, when  the  flare was   operated under
smokeless conditions.

     Related to this problem  was the problem of reliable mass balances using
S02 as  a  tracer.  Accurate plume S02 concentration measurements required S02
plume levels  of  10 ppm or greater.  Under many of the flare conditions of
this test program, this required a  higher S02 tracer flowrate  than available.
To remedy this, an additional  S02 vaporizer was designed and constructed, but
                                    D-4

-------
was  not installed until after completion, of the test program.   It is,
however, available for future  test programs.

     There was a distinct problem with the  H2S measurement  technique, which
invalidated  all  of the H2$ destruction efficiency results.   In  sampling the
plume for  H2S  and  S02 levels, absorption bubblers were used in  series, one
for H2$  collection, and the second for 862 collection.  After the completion
of the  H2$ test  series,  it was discovered by  follow-up testing  that the H2S
bubbler  also collected SC>2, and vice  versa.  The S02 titration method is
relatively insensitive to H2S presence, but S02 presence in the  ^S bubbler
greatly  affected the H2S results.   The  accuracy of several methods for
measuring  H2$  levels in H2S-S02 mixtures has recently been verified at EER
for  use in  future  flare test  programs.  These methods include (1)
colorimetry, (2) Draeger tubes, and  (3) gas chromatography using a flame
photometric detector.

D.4      Reference
D.I  "Revision  of Reference  Method  11", Federal Register, 43(6), p.1494,
     1978.
                                    D-5

-------
                                 APPENDIX E

                             CONVERSION FACTORS
To Convert From                        To                       Multiply
    English                          Metri c                        By

   Btu                                KJ                           1.055
   CFM                               m3/h                          1.700
   in.                                 m                           0.0254
   in. H20                            Pa                         249
   psi                                Pa                        6893
   ft                                  m                           0.3048
   ft3                                 m3                          0.02832
   lb                                 kg                           0.4536
   mph                                km/h                         1.609
   °R (Rankine) is converted to °C (Celsius) via the following formula:

                              °C = 5/9 (R - 492)


   °F (Fahrenheit) is converted to °C (Celsius) via:

                              °C = 5/9 (°F - 32)
                                     E-l

-------
                                 TECHNICAL REPORT DATA     .
                          (Please read Instructions on the reverse before completing)
1. REPORT NO
 EPA-600/2-85-106
                            2.
                                                        3. RECIPIENT'S ACCESSION NO.
A. TITLE AND SUBTITLE
 Evaluation of the Efficiency of Industrial Flares:
  Flare Head Design and Gas Composition
                                    5. REPORT DATE
                                    September 1985
                                    6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
 J. H. Pohl and N. R. Soelberg
                                                        8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
 Energy and Environmental Research Corporation
 18 Mason
 Irvine, California  92718
                                                        10. PROGRAM ELEMENT NO.
                                    11. CONTRACT/GRANT NO.
                                    68-02-3661
12. SPONSORING AGENCY NAME AND ADDRESS
 EPA, Office of Research and Development
 Air and Energy Engineering Research Laboratory
 Research Triangle Park, NC 27711
                                                        13. TYPE OF REPORT AND PERIOD COVERED
                                                        Final; 10/83 - 12/84
                                    14. SPONSORING AGENCY CODE
                                      EPA/600/13
is. SUPPLEMENTARY NOTES AEERL project officer is Bruce A.  Tichenor, Mail Drop 54; 919/
 541-2991.  EFA-6GO/2-83-070  and -84-095 are earlier related (Phases 1 through 4)
 reports.	
is. ABSTRACT
              repOrt gives continued Phase 4 results of a research program to quan-
 tify emissions from,  and efficiencies of, industrial flares.  Initial results were limi-
 ted to tests conducted burning propane/nitrogen mixtures in pipe flares without pilot
 light stabilization.  The work reported here extends the previous results to  other
 flare head designs  and other gases and includes a limited investigation of the influ-
 ence of pilot flames on flare performance.  Results included: (1) flare head design in-
 fluences the flame  stability  curve,  (2) combustion efficiency can be correlated with
 flame stability for  pressure heads and  coanda steam injection heads; (3) for the limi-
 ted conditions tested,  flame stability and combustion efficiency of air- assisted heads
 correlated with the momentum ratio of air  to fuel (the heating value of the gas  had
 only minor influence), (4) limited data  on an air- assisted flare  show that a pilot light
 improves flame stability,  (5) the destruction efficiency of compounds depends on the
 structure of the compounds,  and (6) for compounds tested in this program,  the des-
 truction efficiency  of different compounds could be correlated with the flame stability
 curve for each.
17.
                              KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                           b.IDENTIFIERS/OPEN ENDED TERMS
                                                 c. COSATl Field/Group
 Pollution
 Exhaust Gases
 Efficiency
 Flames
 Measurement
 Surveys
Analyzing
Design
Chemical Analysis
Pollution Control
Stationary Sources
Industrial Flares
Flare  Head Design
13B
2 IB
14G      07D
                                                 14B
13. DISTRIBUTION STATEMENT
 Release to Public
                                            19. SECURITY CLASS (This Report)
                                            Unclassified
                                                 21. NO. OF PAGES
                                                    138
                        20. SECURITY CLASS (This page}
                        Unclassified
                                                 22. PRICE
EPA Form 2220-1 (9-73)
                      E-2

-------