EPA-600-2-85-106 Evaluation of the Efficiency of Industrial Flares: Flare Head Design and Gas...

8/14/2019 EPA-600-2-85-106 Evaluation of the Efficiency of Industrial Flares: Flare Head Design and Gas Composition

1/140


2/140

RESEARCH REPORTING SERIESResearch reports of the Office of Research and Development, U.S. EnvironmentalProtection Agency, have been grouped into nine series. These nine broad categories were established to facilitate further development and application of environmental technology. Elimination of traditional grouping was consciouslyplanned to foster technology transfer and a maximum interface in related fields.The nine series are:

1. Environmental Health Effects Research2. Environmental Protection Technology3. Ecological Research4. Environmental Monitoring5. Socioeconomic Environmental Studies6. Scientific and Technical Assessment Reports (STAR)7. Interagency Energy-Environment Research and Development8. "Special" Reports9. Miscellaneous Reports

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

EPA REVIEW NOTICEThis report has been reviewed by the U.S. Environmental Protection Agency, andapproved for publication. Approval does not signify that the contents necessarilyref lect the views and pol icy of the Agency, nor does mention of trade names orcommercial products constitute endorsement or recommendation for use.This document is available to the public through the National Technicallnformation service. Springfield, Virginia 22161.


3/140

EPA- 600/2- 85-106September 1985

EVALUATION OF THE EFFICIENCYOF INDUSTRIAL FLARES: FLARE HEAD DESIGN

AND GAS COMPOSITION

byJ . H. Pohl and N. R. Soelberg

ENERGY AND E N V I R O N ~ E N T A L RESEARCH CORPORATION18 MASONIRVINE, CALIFORNIA 92718

EPA Contract No. 68-02-366!

EPA Project Officer:

Air and Energy Engineering Research LaboratoryHazardous Air Technology Branch

Research Triangle Park, North Carolina 27711

Prepared for:u.S. ENVIRONMENTAL PROTECTION AGENCYOffice of Research and Development

Washington, D.C. 20460


4/140

ABSTRACT

The U.S. Environmental Protection Agency has contracted with Energy andEnvi ronmental Research Corporati on to conduct a research program whi ch wi 11result in quantification of emissions from, and eff iciencies o f, industrialflares. The program is divided into four phases. Phase I - Experimental Designand Phase II - Des i gn of Test Fac il i t i es have been reported in EPA Report No.600/2-83-070. Phase III - Development of Test Facilities and the initial workin Phase IV - Data Collection has been reported in EPA Report No. 600/2-84-095.Further data collection during Phase IV is reported herein.

Initi a1 results were 1imited to tests conducted burni ng propane-n i trogenmi xtures on pi pe fl ames without pil ot 1ight stabil i zati on. The work reportedhere extends the previ ous results to other gases and fl are head desi gns, andincludes a limited investigation of the influence of pilot flames on flareperformance. The following results were obtained:

o Flames from nozzles less than 1-1/2 inches* in diameter general ly arenot 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 steami nj ected head.

o For the limited conditions tested, the flame stability and combustionefficiency of the air-assisted head correlated with the momentum

*English units are generally used throughout this report. Appendix E providesconversion factors for English to Metric units.ii


5/140

ratio of air to fuel; the heating value of the gas had only a minori nfl uence.

o Limited data on an air-ass is ted flare shows that use of a pilotlight improves flame stability.

o The destruction efficiency of different compounds can be correlatedwi th the fl ame stabi 1i ty curve for each compound, for the compoundstested in this program.

o Stable flare flames and high (>98-99 percent) combustion and destruction efficiencies are attained when the flares are operated withinoperating envelopes specific to each flare head and gas mixturetested. Operation beyond the edge of the operating envelope resultsin rapi d fl ame de-stabil i zati on and a decrease in combustion anddestruction efficiencies.

iii


6/140

TABLE OF CONTENTSSection

LIST OF FIGURES LI ST OF TABLES vviii1.0 I NTRODUCTI ON 1-1

FLARE AERODYNAMICSSUMMARY AND CONCLUSIONS

1.1 Review of Previous Work 1.2 Current EER Flare Program.1.3 References2.03.0

3.13.23.33.43.5

Reynolds Number and Richardson NumberFlame Length Fl arne Stabil ity Pseudo Adiabatic Flame Temperatures and StabilityReferences

1-11-61-112-13-13-13-43-163-163-23

FLARE NOX AND HYDROCARBON EMISSIONS.

Coanda Steam-Injected Head DPressure-Assisted Head E Pressure-Assisted Head FAir-Assisted Head G Commercial Head Summary

4-14-34-34-84-84-195-15-15-15-35-115-176-1

Compound SelectionCompound Screening Tests Gas Mixture Flare Tests.Flame Stability CorrelationsReferences .

GAS COMPOSITION TEST RESULTS5.15.25.35.45.5

COMMERCIAL FLARE HEAD TEST RESULTS4.14.24.34.44.5

5.0

4.0

6.0APPENDIX A.APPENDIX B.APPENDI XC.

EPA FLARE TEST FACILITY AND TEST PROCEDURES FLARE SCREENING FACILITY AND TEST PROCEDURESDATA ANALYSIS

A-IB-1C-1

APPENDIX D. QUALITY ASSURANCE. 0-1APPENDIX E. CONVERSION FACTORS E-1

iv


7/140

Figure2-12-22-32-42-52-62-7

2-82-93-13-23-33-43-53-6

3-73-83-93-103-113-12

LI ST OF FI GURES

EPA Flare Test Facility (FTF) at EER Flare Screening Facility (FSF) Region of flame stability for steam-injected andpressure-assisted heads D, E, and F Maximum gas exit velocity for stability versus airassist to gas momentum ratio for the air-assistedhead G, with and without pilot Combustion efficiency vs. flame stability forsteam-injected and pressure-assisted flare heads.Combustion efficiency vs. air-assist to gasmomentum ratio for commercial air-assisted head GFlame stability curves for pipe flare heads, 1/16 inchthrough 2 1/2 inch flares are without flame retentiondevices. The 3, 6, and 12 inch heads were tested withand without flame retention devices Region of flame stability for the 3 inch open pipeflare head burning selected relief gas mixturesDestruction efficiency of different gases Aerodynamic conditions of f la res tes ted at EERcompared to typical commercial flare conditionsComparison of flame lengths predicted by availablestudies with those observed in this study Correlation of flame length with heat release for1/2 inch to 31 inch flares .Correlation of flame length to heat release for1/16 inch through 2 1/2 inch flares Empirical correlation of radiant heat loss frompropane-nitrogen flames. Richardson number=2.0 Predicted flame length from modified Richardsonnumber correlation compared to observed flamelength for previously tested 3, 6, and 12 inchfl are heads . .Predicted flame length from modified Richardsonnumber correlation compared to observed flamelength for 1/2 through 2 1/2 inch flare heads . .Predicted flame length from modified Richardsonnumber correlation compared to observed flamelength for 1/16 through 1/4 inch flare heads Predicted flame length from modified Richardsonnumber correlation compared to observed flamelength for 1 1/2 through 12 inch commercial headsFlame stability curves for pipe flare headsMaximum relief gas exit velocity vs. flame temperaturefor previously tested 3 through 12 inch flare heads Maximum relief gas exit velocity vs. flame temperaturefor commercial heads D through G .

v

2-22-32-142-152-162-17

2-182-202-213-23-53-63-93-10

3-123-133-14

3-153-173-193-20


8/140

LIST OF FIGURES (continued)Figure3-13 Relation between flame st ab i l i t y and pseudoadiabatic flame temperature 3-214-1 Flame st ab i l i t y curve fo r a commercial 12 inch coandasteam-injected head D. Steam flowrate at 140 lb/hr 4-44-2 Relation of flame st ab i l i t y to combustion efficiencyfor the commercial 12 inch coanda steam-injectedflare head D. Steam flowrate at 140 lb/hr 4-54-3 Region of flame st ab i l i t y for 1.5 inch commercialpressure-assisted head E . . 4-64-4 Relation of flame st ab i l i t y to combustion efficiencyfo r commercial 1. 5 inch pressure-assisted head E 4-74-5 Region of flame st ab i l i t y for 3.8 inch commercialpressure head F . . 4-94-6 Relation of flame st ab i l i t y to combustion efficiencyfo r a commercial 3.8 inch pressure-assisted head F 4-104-7 Region of flame st ab i l i t y fo r a 1.5 inch commerciala ir -a ss is te d f la re G without a pilot flame 4-114-8 Region of flame st ab i l i t y for 1.5 inch commercialair-assisted flare G with pilot flame. . 4-124-9 Region of flame st ab i l i t y for 1.5 inch commercialair-assisted head G, with no air-assis t . 4-134-10 Maximum gas exit v el oc it y f or st ab i l i t y versus airassist to gas momentum ratio for th e air-assistedhead G, with and without pilot . 4-154-11 Combustion efficiency vs. ai r-assi st to gas momentumratio for commercial air-assisted head G 4-174-12 Destruction efficiency of incompletely combustedproducts from air-assisted head G 4-184-13 Region of flame stability for steam and pressure-assi sted heads D, E, and F 4-204-14 Combustion efficiency vs . flame st ab i l i t y fo r steam-i nj ected and pressure-assi sted fl are heads .. 4-215-1 Region of flame stability fo r th e 3 inch open pipeflare head burning selected rel i ef gas m ixtur e s. 5-55-2 Destruction efficiency of different g a s e s . 5-85-3 Hydrocarbon combustion efficiency of gas mixtures 5-95-4 Pilot-scale FTF destruction efficiency resultscompared to lab-scale FSF destruction efficiencyresults. All results at operating conditions nearth e st ab i l i t y lim it. FTF nozzle size = 3 inches.

FSF nozzle size = 0.042 inches . . . 5-105-5 Calculated adiabatic flame temperature vs . limitingstable gas exit velocity fo r different gas m ixtures . . 5-13

v i


9/140

Figure5-6 (a)5-6(b)6-16-2

LIST OF FIGURES (continued)

Nozzle velocity vs Experimental Index for data fromNoble, et al ...................Nozzle exit velocity vs Experimental Index for gasmixture tests.NO x concentration (air-free basis) at the plumecenterline from pilot-scale f lares N02 emissions from pilot-scale flares

vii

5-155-156-26-4


10/140

LIST OF TABLES .Table Page1-1 S U ~ t ~ R Y OF PREVIOUS FLARE COMBUSTION EFFICIENCY

2-12-22-3

STUDIES . . . . . . . . . . . . .COMMERCIAL 12 INCH DIAMETER COANDA STEAM-INJECTEDHEAD D TEST RESULTS. STEAM FLOWRATE = 140 LB/HR.COMMERCIAL 1.5 INCH DIAMETER PRESSURE-ASSISTEDHEAD E TEST RESULTS . . . . . . . . . . .COMr'lERCIAL 3.8 INCH DIAMETER PRESSURE-ASSISTEDHEAD F TEST RESULTS .

1-22-42-52-6

2-4

2-7

COMMERCIAL 1.5 INCH DIAMETER AIR-ASSISTED HEAD GTEST R ESULTS . . . . . . . . . . . . . . .2-5 3 INCH DIAMETER OPEN PIPE FLARE (WITH PILOT) TESTRESULTS .2-6 RESULTS OF SCREENING TESTS ON FLARE SCREENING

FAC ILITY. . . . . . . . . . . . . . . .GAS MIXTURE COMBUSTION AND DESTRUCTION EFFICIENCYTEST RESULTS: 3 INCH FLARE, NO PILOT. . . . . . .

4-1 COMMERCIAL HEADS SELECTED FOR COMBUSTION EFFICIENCYTESTS .

5-1 RESULTS OF SCREENING TESTS ON FLARE SCREENINGFAC ILITY. . . . . . . . . . . . . . .

2-7

2-9

2-102-124-25-2

5-2 PHYSICAL PROPERTIES (1 ATM, 600F) OF THE COMPOUNDSTESTED U SIN G THE FLARE TEST FACILITY . . . . . . 5-66-1 GC-MS ANALYSIS OF PLUME SAMPLE FROM LABORATORY-SCALE

TESTS . . . . . . . . . . . . . . . . . . . . . . . 6-66-2 GC-MS ANALYSIS OF PLUME CENTERLINE SAMPLES FROM GASMIXTURE TESTS, USING A 3 INCH OPEN PIPE FLARE . . .

viii

6-7


11/140

1.0 INTRODUCTIONIndustrial flares are commonly used to safely and economically destroy

large amounts of industrial waste gases. Since most of the gas flared in theUnited States is from leaks, purges, and emergency vents, the amounts andcompositions of flared gases vary widely and are difficult to measure. Flareemissions are also difficult to measure. Most flares are elevated todecrease noise radiation and combustion products at ground level. Probecollection of plume material in such situations is impractical. Remotesensing of flare emissions is an alternative to direct sampling , butinstrumentation and techniques for this purpose are still undeveloped.

Evaluation and control of industrial flare emissions requires pi1otscale research with direct sampling of flare emissions. Flare research hasbeen conducted at Energy and Environmental Research Corporation (EER) since1980. A pilot-scale flare te st fa cility was constructed for the U.S.Environmental Protection Agency in 1982. This research has been sponsored bythe U.S. E.P.A. as part of an effort to provide data upon which to base newregulations for industrial flaring pract ices (1.1, 1.2).1.1 Review of Previous Work

Several previous studies have evaluated flare emissions by testingcombustion efficiency of small-scale or pilot-scale flames (1.3). A primaryfinding of these studies was that flare flame combustion efficiencies can bevery high, exceeding 98 percent, but that under certain operating conditions,such as excessive steam injection, low efficiency can result (1.4). Poh1, etal. (1.3) and Keller and Noble (1.5) have also reported that when a flame isstable (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 flarestudies are summarized in Table 1-1. A range of flare sizes , designs, andoperating conditions have been examined. The results show combustionefficiencies ranging between 55 - 100 percent. The measured combustion

1-1


12/140

IN

Table 1-1.SUMMARY OF PREVIOUS FLARE COMBUSTION EFFICIENCY STUDIES

FLARESIZE VELOCITYSTUDY DATE (in.) DESIGN (ft/see) GAS FLAR

Palmer (1.6) 1972 0.5 Experimental Nozzle 50 - 250 EthyleneLee &Whipple (1. 7) 1981 2.0 Holes in 2" Cap 1.8 PropanSiegel (1.8) 1980 27. Commercial Flaregas 0.7 - 16 RefineryCoanda FS-6Howes, et al (1. 9) 1981 6 ~ c ) COllfllercia 1 Ai r 40 - 60 PropanAssist. link LHHowes, et a1 (1. 9) 1981 3at 4(b) Commercia1 I t P. Zink Near Sonic Natural

LRGO (estimate)McDaniel (1.10) 1983 8 COl1ll1lercia1 Zi nk 0.03 - 62 PrOPY1ene/NKeller and Noble (1.5) STF-S-8McDaniel (1.10) 1983 6(c) COl1ll1lercia1 Ai r 1.4 - 218 Propv1ene/NiKeller and Noble (1.5) Assist.link STF-lH-457-5Poh1, et a1 (1.3) 1984 3-12 Open pipe and 0.2 - 420 Propane/Nitrocommercial.

(a'50% hydrogen plus light hydrocarbons(b)Three sp1ders, each with an open area of 1.3 1n2(e'Supp1ied through spiders; high Btu gas through area of 5.30 in2 and low Btu gas(d) .Heating value varled from 209 to 2183 Btu/scf( e ~ e a t i n g value varied from 83 to 2183 Btu/scft f ~ e a t i n g value varied from 291-2350 Btu/sef


13/140

efficiencies show that flare flame stability and combustion efficiency mayvary, depending on flare head size, design, gas composition, and operatingconditions.

Accurate measurement of flare emissions and combustion efficiency isdifficult even on a pilot-scale facility. Problems encountered by previousresearchers include:

Inability to close mass balances Inability to measure soot emissions Sampling only on the plume centerline Flare flame fluctuations due to turbulence and/or windThe Flare Test Facility at EER has been designed and built in order to

minimize such problems. The following procedures have been developed toverify the accuracy of combustion efficiency measurements:

Material balance closure was verified using a hood to capture theen tire flare plume for small flames, and by using S02 as a tracerfor 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 theentire plume by (1) using a hood to completely capture smallflames, and (2) simultaneous sampling at five radial positions inthe plume for large flames. These local values were combined withvelocities calculated from jet theory and integrated to calculateoverall combustion efficiency.

1-3


14/140

The effects of flare flame fluctuations were limited by mlxlng asample taken over a 20 minute time period. This time span wasexperimentally determined to be sufficient to average flamefluctuations.

Using the Flare Test Faci l i ty, EER has developed a data base ofcombustion 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, includinglift-off and flame length, were also studied.

Due to industrial interest in higher flare exit velocities, testing wasconducted at exit velocities up to 420 ft /sec. The same parameters ofscaling, combustion efficiency, and aerodynamics were evaluated at these highvelocities, using a 3 inch open pipe flare both with and without flameretention devices.

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

Flame length can be estimated using Richardson number and anestimate of the flame temperature.

Flare burn gases with combus tion efficiencies greater than98 percent unless operated within 30 percent of the blow-off limit.

Soot con trib ute s le ss than 0.5 percent to the unburnedhydrocarbons.

A single probe can yield good estimates of overall combustionefficiency.

1-4


15/140

o For the range of f lares and conditions studied, flame structure andcombustion efficiency did no t depend on f lare size or design.

In order to extend the basic open-pipe f lare tes t resul ts to industrialapplication, three commercial 12 inch heads were also tested. The purpose ofthese tes ts was to determine the dependence of combustion efficiency on f larehead type and design. Combustion efficiency can vary, depending on commercialf lare head type and design, because f lare heads are often tai lored to achievespeci fi c r esu lt s for specific gas mixtures and operating conditions, and maynot operate well under o ther cond i tions .

The EER f lare t es t program has established a data base for evaluation offuture f lare tes t resul ts . In general, i t has been demonstrated that when af lare flame is operated under stable conditions, combustion e ff ic iency grea te rthan 98 percent is at tained. For propane-ni trogen mixtures f lare using pipeflares without pi lots , a relationship has been demonstrated between the gasheati ng va l ue and the stabil i ty 1imit , and between the stabil i ty 1imit andcombustion efficiency. A s im il ar r el at ion sh ip between heating value, flamestabil i ty , and combusti on effici ency was observed for the three commerci al12 inch pipe f lares tes ted.

Throughout the EER fl are research program, advi ce and consultati on wasprovi ded by a Techni ca 1 Advi sory Commi t tee . Commi t tee members i ncl uded representat i ves from EER, EPA, Cali forni a Ai r Resources Board, fl are manufacturers(Peabody Engineering, McGill, Inc. , John Zink, and Flaregas Corpora tion) andindustrial f lare users (Exxon Chemical Company, Exxon R&E, Union Carbide, GettyRefining and Marketing Co., Chevron USA, and Dow Chemical Company). The indust r ia l users included representatives of the Chemical Manufacturers Association(CMA) and the American Petroleum Inst i tute (API). This committee met throughout the program to review and cr i t ique tes t plans, ensure relevance of thestudy, and faci l i ta te efficient information t ransfer .

1-5

.


16/140

1.2 Current EER Flare ProgramThe previous EER studies have established accurate flare combustion

efficiency test methodology, developed a pilot-scale test facility, and established a data base of combustion efficiency test results for 3 through 12 inchdiameter fl are heads burni ng propane-nitrogen mi xtures on pi pe fl ares withoutpilot light stabilization (1.3, 1.11, and 1.12).

The current fl are test program is an extension of the previ ous work.Objectives were based on recommendations of the Fl are Advisory Panel and weredesigned to extend the data base to cover a wider range of flare operatingand design conditions.1. 2.1 Objectives

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

1. 2. 2

o

o

Evaluate the effects of flare head type on flare combustionefficiency.

Evaluate effects of relief gas composition on flare combustionand destruction efficiency.Approach

The overall objective of this program is to assess pollutant emissionsfrom i ndustri al fl ares. Di rect measurements on full-scale operati ng f1 aresare difficult. Therefore, direct measurements were made on pilot-scaleflares in order to measure pollutant emissions, combustion and destructionefficiency, and scaling criteria.

1-6


17/140

The program was divided into four major tasks: Task 1 - Evaluation of combustion efficiency from d if fe rent f la re

head types Task 2 - Identification of representative, potentially difficult to

destroy gas compounds Task 3 - Evaluat ion of combustion and destruction efficiency of

selected relief gas mixtures. Task 4 - Data analysis and reportingIn Task 1, EER obtained from flare suppliers the following fourcommercial heads: 1 air-assisted flare head 2 pressure-assisted flare heads 1 coanda steam-injected flare headEach of these heads was tested on the EER pilot-scale Flare Test

Facility (FTF). Flame stability and combustion efficiency were measured asfunctions of the following operating conditions:

Relief gas and exit velocity 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


18/140

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 industrialcompounds are frequently flared in the United States. Most often, they areflared in mixtures containing several compounds. Each mixture may exhibitsomewhat different flaring characteristics. Pilot or large-scale testing ofevery conceivable relief gas mixture would be expensive and unending.

A sensible approach is to test compounds in a laboratory-scale facilitywhich are representatives of classes of compounds industrially flared. Thesmall scale laboratory facility produces flames which are not aerodynamicallysimilar 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 acommercial 12 inch diameter flare head, Reynolds numbers may range from 4,000to 40,000, Richardson numbers range from 0.1 to 1,000, and buoyant forcesoften 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 orsmaller, and inertial forces dominate. Even though lab-scale tests using a1/16 inch diameter nozzle are aerodynamically dissimilar to large-scaleflares, such tests can be used to economically screen compounds to determinecomparative potential for successful destruction of these compounds byflaring. Compounds which demonstrate flaring difficulties in the FlareScreening Facility (FSF) are candidates for testing on the FTF.

Upon recommendations of the Flare Advisory Committee, twenty-one of themost commonly flared, potentially hazardous, or difficult to flare compoundswere selected for laboratory-scale testing on the (FSF). These compounds arerepresentative of:

1-8


19/140

Sulfur compounds Nitrogen compounds Chlorinated compounds Oxygenated compounds Aliphatic hydrocarbons

Aromatic hydrocarbons Compounds with low heating valueOf the twenty-one compounds screened, six were selected as candidates

for test ing on the FTF. Selection criteria included low destructionefficiency, poor ignitabili ty, and high soot production. In addition,hydrogen sulfide, although not tested in the screening facility, was alsoselected due to its industrial importance and toxicity.

Four of the seven compounds selected during the screening tests weretested on the FTF. Hydrogen sulfide and ammonia were each tested in mixtureswith propane and nitrogen. Ethylene oxide and 1,3-butadiene were each testeddiluted with nitrogen to vary the heating value. Flame stability, combustionand destruction efficiency, soot production, and by-product formation fromincomplete combustion were measured for each compound. All tests wereconduc ted u sing the 3 inch open pipe f lare, without pilot flamestabilization.

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

1-9


20/140

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

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

Sample analysis was conducted during tests on both the FSF and FTF toevaluate air dilution, mass balances, combustion efficiency, and destructionefficiency. Sulfur dioxide was injected during some of the pilo t tes ts andused as a tracer for mass balances. Mass balances on the FTF were moredifficult because of product loss, air dilution in the large exposed flame,and plume concentration gradients. Local mass balances were used toaccurately evaluate local mass fluxes, local combustion efficiency, anddestruction efficiency. Local mass fluxes were radially integrated tocalculate overall combustion and destruction efficiencies.

In Task 4, flare combustion and destruction efficiencies were correlatedto flame stabil i ty. Flame stabili ty was correlated to relief gas heatingvalue and exit velocity for the gas mixture tests and for all the commercialheads except for the air-assisted head. Flame stability for the air-assistedhead was related to air-fuel stoichiometry and momentum ratios. Results forthe air-assisted head showed that the effects of relief gas heating value andexit velocity on flame stability were minimal, compared to the effect of theair-assist stream to fuel stream momentum ratio.

In order to evaluate the dependence of flare flame combustion efficiencyon flame stability and operating conditions, i t was necessary to investigateflare operating conditions which result in both efficient and inefficientcombustion in order to define the region of efficient operation and todetermine parameters which are critical to flare performance. Thus, many of

1-10


21/140

th e tes t cases represent conditions outside the normal range of commercialflare operations.

1.3 References

1.1 Davis, B. C., "U.S. EPA's Flare Policy: Update and Review", ChemicalEngineering 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 Pbhl, J. H., R. Payne, and J. Lee, "Evaluation of the Efficiency ofI ndus tr ia l F la re s: Test Results", EPA Report No. 600/2-84-095, t ~ a y1984.1.4 Dubno\'/ski, J. J. and B. C. Davis, "Flaring Combustion Efficiency: AReview of the Sta te of Current Knowledge, "The 76th Annual AirPollution Control Association Meeting, Atlanta, GA, 1983.1.5 Keller, N. and R. Noble, "RACT for VOC - A Burning I ssue", Polluti onEngineering, July 1983.1.6 Palmer, P. A., "A Tracer Technique for Determining Efficiency of anElevated Flare," E. 1. du Ponte de Nemours and Co., Wilmington, DE,1972 .1.7 Lee, K. C. and G. M. Whipple, "Waste Gas Hydrocarbon Combusiton in aFlare, "Union Carbide Corporation, South Charleston, WV, -1981.1.8 Siegel, K. D., "Degree of Conversion of Flare Gas in Refinery HighFlares, "Ph.D. Dissertation, Univers ity of Karlsruhe (Federal Republicof Germany), February, 1980.1.9 Howes, J. L , T. L Hill , R. N. Smith, G. R. Ward, and W. F. Herget,"Development of Fl are Emi ssi on r ~ e a s u r e m e n t Methodology, Draft Report,"EPA Contract No. 68-02-2682, U.S. Environmental Protection Agency,1981 (Draft Report) .1.10 McDaniel, M., "Flare Effiiciency Study, II EPA Report No. 600/2-83-052,July 1983.1.11 Joseph, D., J. Lee, C. McKinnon, R. Payne, and J. Pohl, IIEvaluation ofthe Efficiency of Industr ia l Flares: Background-Experimental DesignFacilityll, EPA Report No. 600/2-83-070, August 1983.1.12 Pohl, J . H., J . Lee, R. Payne and B. Tichenor, "The CombustionEfficiency of Flare Flames", 77th Annual t ~ e e t i n g and Exhib it ion of theAir Pollution Control Association, San Francisco, CA., June 1984.

1-11


22/140

2.0 S U ~ 1 t 4 A R Y AND CONCLUSIONSThe objective of this phase of "Evaluation of the Efficiency of

Industrial Flares" was to determine the influence of additional flare headdesigns and gas composition on the stabili ty and structure, destructionefficiency (DE), and combustion efficiency (CE) of flare flames. This phaseextended the previous programs which reviewed the literature available onflare performance, designed and built a flare test facility, and determinedthe combustion efficiency of pipe flares firing mixtures of propane andnitrogen in the absence of a pilot flame.

The current program extends previous work to include different classesof commercial flare heads, and types of relief gas mixtures burned on smallpipe 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 inFigure 2-1.

Twenty-one different gases were selected and screened for potentialcombustion problems on the Flare Screening Facility (FSF) shown in Figure2-2, with a 1/16 inch nozzle (10 = 0.042 inches). The combustion anddestruction efficiencies of four of these gases were measured on the FlareTest Facility (FTF).

The data collected on the FTF for the d if fe rent f la re heads issummarized in Tables 2-1 through 2-4. The air-assisted head was tested bothwith and vdthout a pilot flame, as shown in Table 2-4. Limited tests toevaluate the effects of pilot flames were also conducted on a 3 inch openpipe flare, without flame retention devices. These test results are shown inTable 2-5. The results of the FSF gas mixture screening tests are shown inTable 2-6, and the results from the FTF gas mixture tests are shown in

2-1


23/140

Adj us ta b 1e Rake forSample Probes, F i lt e rs , J ri e rs ,Tenax Traps, and Subblersfo r HZS, S02 & ~ H

WInd Sen...

Sallll'le lines

COCOZHydrocarbonSOZHZSNO-NOxNH3

Air Supply Fan

Fi gure 2-1. EPA flare test facility (FTF) at EER.

2-2


24/140

MIXING CHAMBE:R' __ Fr=-__~ ~ = = CONTINUOUS SAMPLING===== TENAX SAMPLING

1/16-1/8 IN. NOZZLE

GLASS BEADSHEATED LINE

'*===GASSYRINGE DRIVE 8

AIR FAN

Figure 2-2. Flare screening facility (FSF).

2-3


25/140

NI

Table 2-1COMMERCIAL 12 INCH DIAMETER1 COANDA STEAM-INJECTED HEAD 0

TEST RESULTS. STEAM FLOWRATE = 140 LB./HRObservationsActual SteamExit 'lPropane Low Ratio Probe Wind Flame U f tTest Velocity1 in Htg Val (lb steam/ Ht 3 Speed Length OffNo. (ft/sec) Nitrogen (Btu/ft3) lb fuel ) ( f t ) (mph) (ft ) (i n) Color Smoke

200A 0.17 14.43 338 3.7200B 0.20 13.0 305 3.3200C 4.0 12.1 284 0.642000 4.3 11.1 261 0.15 Stability Curve Tests200E 9.3 13.1 308 0.068200F 9.3 13.0 305 0.068201 0.20 13.0 306 3.3 5 8 1.5 0 dim orange none202 4.38 11.7 275 0.15 5 7 6 0 dim orange none203 9.19 14.0 329 0.069 15 6 12 0 dim orange none204 0.20 17 .9 421 3.1 4 3 3 0 dim orange none205 3.95 18.1 425 0.16 10 8 7 0 dim orange none206 9.90 18.1 425 0.081 22 5 22 0 dim orange none

Based on d e s i ~ n capacity for a 12 inch flare head. Open area at base ofcone = 113 in .2 Based on light diffraction through flame envelope for invisible flames.3 Height above flare tip.


26/140

N,U1

Table 2-2COMMERCIAL 1.5 INCH DIAMETER l PRESSURE-ASSISTED HEAD E. TEST RE

Actual .1PObservations

Exit "Propane Low Across P r ~ b e \lind Flame LiftTest Veloci tyl in Htg Val Head Ht Speed L e n ~ t h OffNo. (ft/sec) Nitrogen (Btu/ft3) (psig) ( ft) (mph) (ft (in) Color Smoke

207 14.3 15.8 371 )08A 112 20.2 474208B 94.7 23.9 562 Stabil it y Curve Tests209 472 23.9 562210 78.5 26.0 612 0211 12.4 18.1 426 0 6 6 2 2 dim orange none212 95.9 21.9 514 0 7 7 2.5 2 ye 11 ow-purp1e none

base214 238 48.1 1130 NR2 12 3.5 12 1 ye 11 ow-b1ue none216 384 29.6 696 NR 16 6 12 4 ye 11 ow-b1ue none217 14.2 23.7 557 0 6 2.5 3 0 yellow none218 470 35.2 828 3 16 5 14 3 blue base none219 109 28.4 668 0 7 5 5 0 orange-blue none220 761 28.1 661 7 20 5 18 4 ye 11 ow-b1ue none221 907 36.6 870 10 30 6 20 3 ye 11 ow-b1ue none

Based on total open area of exit ports of 1.77 in2.2 NR = Not recorded.3 Above flare tip.


27/140

!" )Im

Table 2-3COMMERCIAL 3.8 INCH DIAMETER! PRESSURE-ASSISTED HEAD F TEST RE

Actual l1P ObservationsExit '.tPropane Low Across Probe Wind Flame LiftTest Vel oci tyl in Htg Val . Head Ht 3 Speed Length2 OffNo. 1ft/sec) Nitrogen IBtu/ft3) ! (psig) (ft) (mph) (ft ) (in) Color Smoke ====== ===========F========== ==========, = = = = ~ c C . ~ = = = = = = = = = = = = = = = ========== ======= ==============b=======269 33.9 5.93 139 NR I70 119 7.46 175 NR271 112 7.49 176

NR )Stability Curve Tests

272 158 7.20 169 NR273 11.1 5.18 122 NR274 42.3 7.50 176 0 13 4.5 4.5 0 1ight bl ue none275 52.2 9.08 213 0 15 3 4.5 0 dim blue none276 5.55 9.10 214 0 8 52 2 0 transparent none277 4.39 9.92 233 0 5 6 2 0 transparent none278 157 13.B 323 4 25 2 17 . 0 transparent none279 139 18.5 453 4 25 7 18 0 ye 11 ow-b1ue none280 25.3 10.7 251 1 15 0.5 8.5 0 yellow none

Based on total open area of exit ports of 11.3 in2.2 Based on light diffraction through flame envelope for invisible flame.3 Height above f lare tip4 Not Recorded.


28/140

r-vI

'- I

Table 2-4COMMERCIAL 1.5 INCH DIAMETER1 AIR-ASSISTED HEAD GTEST RESULT

Actual Propane Air-Assist ObservationsFuel EXif in Low 2 Air-Assist Exit Air- (pv)air4 Wind IFlame li ftTest Veloci ty Nitrogen IItg Val Flovlrate Velocity Assist - - - - - - - - Speed Length OffNo. ( f t/se c ) (%) (Btu/ft3) (SCFM) ( f tlse c ) 5.Q,3 (pv) fuel (mph) (ft ) (in) Color Smoke S222 0.73 100 2350 1030 112 79.0 99.4224 2B.7 1l.03 189 0 0 0 0225 46.B 13 .0 305 0 0 0 0226 263 40.3 947 0 0 0 0227 12.2 9.40 221 0 0 0 0227A 13.8 8.57 201 0 0 0 0228 1.15 100 2350 705 76.6 34.3 43.5229 4.21 27.7 650 412 44.8 19.7 9.38230 90.0 21.4 505 0 0 0 0231A 60.1 32.2 756 784 85.1 2.26 1.22232A 64.0 30.0 706 1110 121 3.21 1.65231B 44.0 43.9 1032 780 84.7 2.25 1.57 Stability Curve Tests23211 62.7 30.7 722 1100 119 3.18 1.66 I233 380 38.8 912 0 0 0 0234 03 48.7 1144 780 84.7 0.294 0.224235 368 38.9 913 1110 120 0.434 0.274236 591 29.3 690 0 0 0 0237 469 54.2 1273 1100 120 0.242 0.199238 464 55.8 1311 544 59.1 0.117 0.0987239 301 47.8 1123 542 58.9 0.210 0.157240 4.59 20.1 473 529 57.4 31.9 11.6241 57.7 20.2 475 535 58.1 2.57 0.924

30.114 l b / ft , vfuel

Based on total area of fuel e xit ports = 1.77 in2. Area o f c o- ax ia l a ir chann!'l = 22.1 i1l 2.2 Includes contribution of pilot3 S.H. = Stoichiometric ratio4 . 3For example, Tes t No. 228: "air = 0.0744 l b / f t , vai r = 76.6 ftls ec, "fuel I. 15 ft/s ec, (pv)a i rl (0


29/140

NI(X )

Table 2-4 (Continued)COMMERCIAL 1.5 INCH DIAMETER l AIR-ASSISTED HEAD G TEST RESULT

Actual PI-opane Air-Assist ObservationsFuel Exi t in Low2 Air-Assist Exit Air- (pv)air4 Wind Flame Li ftTest Veloci tyl Nitrogen IItg. va1 Flourate Velocity Ass i s t - - - - - - - - Speed l.ength OffNo. (ft/sec) ('t) (Btu/ft ) (SCFM) (ft/sec) SIl} (pv)fue'l (mph) (ft) (in) Color Smoke- _.242 90.4 30.3 712 515 56.0 0.90 0.494 9 transparent 4 transparent none243 0.47 23.1 547 56 56.0 14.7 5.97 4.5 1 I blue base none244 428 65.0 1528 511 55.6 0.10 0.01)6 6 12.5 10 ye 11 o ~ l - b 1ue none245 69. I 43.9 1032 1050 114 1.93 1.14 5.5 1 3 blue base none246 425 60.5 1425 1050 114 0.23 0.203 5 10 0 blue base none247 10.9 100 2350 1040 113 5.42 6.75 4 transparent 1 blue base none240 504 15.0 376 0 0 0 0 "249 514 17.3 413 515 56.0 0.320 0.101250 509 16.5 391 1040 113 0.694 0.208251 395 20.l 476 0 0 0 1252 327 20.4 474 513 55.7 0.156 1.464253 431 14.0 334 1040 113 0.249 0.960 > Stability Curve Tests254 207 9.35 229 0 0 0 0255 223 '0.48 210 513 55.7 0.244 1.50256 126 15.6 300 1030 112 0.836 2.95257 07.1 8.69 229 0 0 0 0258 67.8 11.1 290 513 55.7 0.788 3.787259 41.7 18.0 460 1030 112 2.49 7.64260 394 20.9 496 1050 114 0.73 0.265 2.5 1.5 4 blue base none261 10J 18.2 441 1040 113 3.06 1.01 4 I 2 blue base none262 357 37.3 877 1050 114 0.45 0.267 5 1.5 6 blue base none

Oased on total area of fuel exit ports = 1.77 in2. Area of co-axial ai r channel = 22.1 in2.2 Includes contribution of pilot

S.R. = Stoichiometric ratioFor example, Tes t No. 228: . = 0.0744 Ib/ft3, v . = 76.6 f t /sec, "f 1= 0.114 lbfft3, vf 1 = 1.15 f t /sec, (,v) , / ( ' va1 r a1 r ue ue a1 r


30/140

I 1"-- - - - -

TestNo.ActualExit 1Velocity(ft/sec)

Tabl e 2-53 INCH DIAMETER OPEN PIPE FLARE (WITH PILOT) TEST RESULTS

----- ; - i ----...--J--J -J":r--J% Propane Low 2 Probe Wind Flame Li f t~ ; t c ~ ! ~ ~ l . ; ! ! ~ ! ! ; ~ l . J . ; ! ~ .. . . ~ ; l : : . - . ! ; ; l ; ~ ~ : ~ ~ ~ ~ ~ ~263 203 9.96 237 }N I 264 68.8 2.89 76.1I 265 9.82 9.07 260 Flame Stability Tests\0 266 96.7 7.61 184267 1. 93 7.51 375 11 6 1 2 faint yellow Inone I low h268 209 21.0 495 32 5 30 10 blue, yellow none med.

Based upon pipe inside diameter of 3.125 inches.2 Including pilot contribution.


31/140

Table 2-6RESULTS OF SCREENING TESTS ON FLARE SCREENING FACILITY


32/140

Table 2-7. In order to determine the limits of stable operation for theseflares and gas mixtures, and the key operating conditions that a ffect flames tabi l i ty and efficiency, some conditions with poor stabil i ty and lowcombustion efficiencies were measured. Such results merely indicated flareoperation at or beyond the edge of the operating envelope, and are notindicative 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 pressureassisted heads. The stability of the air-assisted head flame was controlledby the momentum ratio of air-assist to fuel streams as shown in Figure 2-4;heating value of the gas had l i t t le influence on flame stability, exceptunder conditions with no air-assist. Combustion efficiency for the pressureand coanda steam-injected heads correlated with the individual flamestabili ty limit fo r each head as shown in Figure 2-5. Figure 2-6 shows therelationship 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 flamestructure, flame stability, and combustion and destruction efficiencies willbe different than measured on pilot and industrial scale flares. Suchdifferences in stability are shown in Figure 2-7, comparing flame stabilityfor different sized open-pipe nozzles. Comparison of ethylene oxidedestruction efficiency on the FSF (1/16 inch nozzle) and the pilot-scale FTF(3 inch open pipe flare) shows the difference in destruction efficiencymeasured in the two facili t ies. Ethylene oxide DE on the FSF was 96.95percent, compared to 98.4 and 99.5 percent on the FTF. Also, industrialflares typically employ pilots and/or flame retention devices to stabilizethe flame and enhance combustion which were absent in these tests.

Tests identified six compounds which were difficult to destroy on theFSF, using a 1/16 inch nozzle (10 = 0.042 inches). These were:

2-11


33/140

NI......

N

Table 2-7GAS MIXTURE COMBUSTION AND DESTRUCTION EFFICIENCYTEST RESULTS: 3 INCH FLARE, NO PILOT

ObservationsActual Gas C o m p o s i t i o l ~ Gas P r o ~ e Wind Flame Lift-Exit Htg Val Ht. Speed Length OffTest ID Velocity (X) (Btu/ft3) (ft) (mph) ( f t ) (f n. ) Color(ft/sec)

281 NH 3 8.79 NH 3 1.5 561 12 5 6 3 Blue base,orange282 NH 3 9.47 NH 3 1.7 877 12 5 7 0 Bright OrangePropane 36.9 L283 NH 3 0.15 NH 3 4.43 336 4 2 0.5 0 OrangePropane 13.2

284 NH 3 138 NH 3 4.02 658 33 3 20 36 Blue-orangePropane 27.0285 NH 3 9.56 tlH3 1.98 416 12 1 8 24 Blue base,Propane 17.2 orange286 H2S 8.89 H2S 1.5 556 12 5 7 24 OrangePropane 23.3287 H2S 139.1 H2S 4.14 646 36 3 22 24 Blue base,Propane 26:5 orange288 H2S 9.11 H2S 4.29 757 14 6 9 0 Bright orangePropane 31. 2 - - - - - - - - -

Rema inde r N22 Destruction ef fic iency data for H2S is not available due to analyticalerrors in the measurernen ts .3 Height above flare tip


34/140

NI....W

Table 2-7 (Continued)GAS MIXTURE COMBUSTION AND DESTRUCTION EFFICIENCYTEST RESULTS: 3 INCH FLARE. NO PILOT

ObservationsActual 1 GasExit Gas Composition Htg Val Profe Wind flame l i ft~ e l o c i t y (%1) (Btu/ft3) Ht. Speed L e n ~ t h OffTes ID (ft/sec) (ftl (mph) (ft ( in. l Color Smok289 Hi:' S 9.11 H2S 7.61 539 14 5 8 12 Orange V.liPropane 21.1290 H2S 0.669 H2S 4.74 370 4 4 1.5 0 Dim-orange Non291 1,3 0.296 1,3- B.2 305 4 3 1 2 Orange Sli~ u t a d i e m But. Dar292 1,3 6.87 1,3- 5.32 145 13 3 6 6 Orange Sliautadiene But.293 1,3- 7.03 1,3- 7.77 212 15 7 5 0 Orange Gre~ u t a d i e n e But. Bla294 thylene 0.320 Eth. 13.2 178 4 2 1 0 Dim-orange Non~ x i d e Oxide295 thylene 4.63 Eth. 13.0 175 7 7 4 NR Transparent Nonbxide Oxide

Remainder N2Z Destruction efficiency data for HZS is not available due to analyticalerrors in the measurements.3 Height above flare tip


35/140

o Pressure-Ass isted Head F

100.0. Q

10.0Exit Velocity (ft/sec)

. " " ~\.,',..l:;e{\cfl.

u:p"~ ' J " t >V : , ~ ~ e (

1.0

:::"v


36/140

1000 r - - - - . . ., - - - - - , -- -"T" "-- - - , r - - - - . . ,. . - - - - ,- - - . . . . . , . - ~

100 i l0

u 0lIVI........+ ''+-

>, 8...- 10.-.Q.... ',o/ithout Pilot..VII- 180-400 Btu/ft30....>, 0 401-900...- 0 901-1400::> .6 1401-2350lJ:>x With Pilot...::::

i ) 209-400 Btu/ft3' I


37/140

--

-

---

I

o

12 in. Coanda Steam-InjectedHead 0[j 1.5 in. Pressure-Assisted Head Eo 3.8 in. Pressure-Ass.i sted Head F

3 in. Open Pipe Head with Pilot

& ~ ~ - : . : . g = .t:cr---fr 0I DorI'TII!IIIIIIo

I100 -

>,us:::C1J

90-='. 0Eou

86_

96 -

8 8

98 -

. 92f-4 -4 -U..Is:::o....

....--0)s:::C1Jus.. 94-

2.584 ' - I l . . . . . .__012

as Heating Value/Minimum Heating Value for StabilityFigure 2-5. Combustion efficiency vs. flame stability forsteam-injected and pressure-assisted flareheads.

2...16


38/140

99.9 0.1a Without Pilot0.5 0.2 S. R.Q With Pilot a uS.R. Air-Assist to Fuel -J:JStoichiometric Ratio ..on:::I- .J:Je0.. '- 'C


39/140

. /rt. 7/;' J ,

w R,,a> 1300'" 1200


40/140

1,3-butadiene has the potential without steam or air assist toyield 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 HCN yielded low destruction efficiency of 85.00 percent NH3 was difficult to ignite when pure without a pilot

Destruction efficiencies (DE1s) for NH3, 1,3-butadiene, ethylene oxide,propane, and H2S were measured on the FTF. The flame stabili ty curvedepended on the compound as shown in Figure 2-8. (H2S and NH3 were tested asminor constituents in propane-nitrogen mixtures). The DE of the individualcompounds depended on compound type but correlated with the respectivestabili ty curve for NH3, 1,3-butadiene, ethylene oxide, and propane as shownin Figure 2-9. The primary influence on destruction and combustionefficiency is flame stabil i ty, not gas heating value or exit gas velocity.HZS destruction efficiency results are unavailab le because of analyticalproblems. The destruction efficiency of propane in mixtures with smallamounts of NH3 or H2S are reported in Figure 2-9 as propane in ammonia testsand propane in hydrogen sulfide tests.

The following conclusions were reached based on the results: Flames from nozzles less than 1 1/2 inches in diameter generally

are not similar to flames from larger nozzles. 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 steaminjected head.

For the l im it ed condit ions tes ted, the flame stabili ty andcombustion efficiency of the air-assisted head correlated with the

2-19


41/140

1200 ~ CJ Ammonia. 1.5-4.4 Percent. in Propane-Nitrogen Mixturel100L 0 1.3-Butadiene Mixed with Nitrogen

1O00l- D Ethylene Oxide Mixed with Nitrogen\l Hydrogen Sulfide. 1.5-7.6 Percent in Propane-Nitrogen MixtureI900800M

...-.-.......

+.J 7001Il

600'">(J )C"r -

N +.J 500'"1IlN ::J:0 1/1'"!) 400300

1 0 0 ~ 0iJ -01001 ID. 1 1.0 10.0 100.0

Exit Velocity (ft/sec)

Figure 2-8. Region of flame stability for the 3 inch open pipflare head burning selected relief gas mixtures.


42/140

99.99

99.9

oWc:

o

\J Ammonia, 1.5-4.5 Percent in Propane-Nitrogen() l,3-8utadiene in NitrogenCI Ethylene Oxide in Nitrogen17 Propane in Anmonia Tests" Propane in Hydrogen Su 1fide Tests

1.- ....&- - -1 . . . 100 . 0o 1.0 2.0Heating Value/Minimum Heating Value for Stability

Figure 2-9. Destruction efficiency of different gases.1 Scale is log(lOO-DE)

2-21


43/140

momentum ratio of ai r to fuel; the heating value of the gas hadonly a minor influence.

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

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

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

Stable flare flames and high (>98-99 percent) combustion anddestruction efficiencies are attained when the flares are operatedwithin operating envelopes specific to each flare head and gasmixture tested. Operation beyond the edge of the operatingenvelope results in rapid flame de-stabilization and a decrease incombustion and destruction efficiencies.

2-22


44/140

3.0 FLARE AERODYNAMICSThe aerodynamics of different sized flare heads was studied prior to

test ing the influence of flare head types or gas composition on flareefficiency. The aerodynamic studies were conducted to determine the minimumhead size which would yield results similar to full scale commercial flareheads. Flare flame length prediction techniques were also investigatedduring this study. The results were used to select flare head sizes for thecommercial head and gas composition tests.

The aerodynamics study included a review of previous flare studies aswell as experimental work testing 1/2 inch to 2 1/2 inch open pipe flareheads, without flame retention devices. The results were used to developrelations 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 thenozzle exi t determine the aerodynamic structure of the flare flames.Reynolds Number (Re) is a measure of the inertial to viscous forces of theflame, and Ri is a measure of buoyant forces to inertial forces of the flame.A flame with Ri greater than one is dominated by buoyant forces and one withRi less than one is dominated by inertial forces. Figure 3-1 compares Re andRi for 1/16 inch through 12 inch flare heads tested at EER, using simulatedrelief gas mixtures of propane-nitrogen. The shaded area indicates theregion of head size and exit velocities for typical industrial sized flares.Since Re depends on the rel ief gas composition, the data points arefrequently not exactly on the approximate grid line drawn for flares ofdifferent sizes. The general trend, however, is evident. As the flare sizedecreases, the nozzle Re and Ri decrease. Flames from smaller nozzles aredominated by iner t ial and turbulent forces, and are aerodynamicallydissimilar to larger, buoyancy dominated flames from industrial flare heads.

3-1


45/140

43"I104

103 ~ " ~ ..102 [ Co I t6" ~ V l q ~ ~ I > _ _ _ 1.0 t ' ~ / o ( ' i tJ--N 101 '-.'' " V - ~ L ( ' ~ \ (1 /> .11, 2 .... /'\. l>'-. t:fh S)

......-0 100n10-1).aE::Jz 10-2c0lI"I

-0 10-3'"cu

. 10-4VJ a:IN -510 r10-610-7

I I I I101 102 103 104 105 106 1Reynolds Number ( ~ d v / ~ )

Figure 3-1. Aerodynamic conditions of flares tested at EER cotypical commercial flare conditions.


46/140

Legend for Figure 3-1

New EER Data 1/16 in. (0.042 in . 10) Nozzle0 1/8 in. (0.085 in. 10) Nozzle1/4 in. (0.180 in. ID ) NozzleLl 1/2 in. Nozzle

'V 1 in. Nozzle1 1/2 in. Nozzle

D 1 1/2 in. Commercial Pressure-Assisted Head E(J 1 1/2 in. Commerci al Air-Assisted Head GQ 1 1/2 in. Commercial Air-Assisted Head G with pil otLJ 2 in. Open Pipe Flare Head0 2 1/2 in. Open Pipe Flare Head0 3.8 in. Commercial Pressure-Assisted Head F

12 in. Commercial Coanda Steam-Injected Head 0Previous EER Data (1.3 )0 3 in. Open Pipe Flare Head

6 6 in. Open Pipe Flare Heado 12 in. Open Pipe Flare Head12 in. Commercial Head Ao 12 in. Commercial Head Bo 12 in. Commercial Head C

3-3


47/140

3.2 Flame LengthCurrent and previous test results of measured flame lengths from flares

were compared to predicted flame lengths using information and techniquesavailable in the literature (3.2-3.4). Neither empirical (3.5-3.7), complexintegral (3.8, 3.9), nor differential (3.10-3.13) techniques could predictthe flame lengths or trends in the flame length of this or other studies oflarge flare flames. Figure 3-2 shows the comparison of flame length measuredin a previous EER flare study (1.3) with predictions using techniquessuggested in the literature. For example, the technique of Brzustowski, etal. (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 35feet, when the observed flame length was 30 feet.

Much better agreement between observed and predicted flame lengths wasachieved by fitting the data with a cruder approach (3.1, 3.2). The approachused by API (3.2) correlates the flame length with the heat release. Thiscorrelation, suggested by Hottel and Hawthorne (3.17), is based on data fromsmall flames, where combustion was limited by air dif fusion into the flame.Although this prediction has been criticized for i ts inaccuracy (3.18),Figure 3-3 shows relatively good agreement with the data of this study, datafrom larger flares, and the original data of the API study. For flame lengthpredictions of large flares, the coefficients in this correlation wereadjusted. Using the same API prediction with coefficients adjusted for largeflares of this study, poorer agreement was found for the flame lengthsproduced from some smaller flares. Flame lengths not accurately predicted bythe API correlation were from small nozzles (Ll-1/2 inch, \7-1 inch, ~ 11/2 inch, LJ-2 inch), the coanda steam-injected head ( ~ ) , and the airassi sted head (C) and Q ). The flame length from nozzles smaller than 1/2inch agree with previous predictions. However, use of the relationshipderived from nozzles less than 1 1/2 inch in diameter would severely underpredict the length of the flames from large flare heads.

Figure 3-3 also shows the least-squares f i t of flame length vs heatrelease for 3 through 12 inch heads tested previously at EER. Data from the

3-4


48/140

6G

55 6 ~ ~ 50 6 6645 0040 0 Hottel &Harthorne

t::. Brzustowski..... Becker and Liang... 85.s::: .... >.:Q,l...J 30

6'" 66.... 2"5-.:l


49/140

2

~ A \ ~ _ _0.0388

API 521 E x t ~ o l a t i o n

4 108 2Heat Release (Btu/hr)

o

}O3

..,4-'t -4-a..,4-.... 102-'>,


50/140

legend for Figure 3-3

AP IDa ta (3.2) at Fuel Gas--20 in . Stack (in line)__ Algerian Gas Well

Catalytic Reformer--Recycle Gas--24 in. StackCatalytic Reformer--Reactor Effluent Gas--24 in . Stack

~ D e h y d r o g e n a t i o n Unit--12 in . StackS3 Hydrogen--31 in. Stack Hydrogen--30 in. Stack

Straitz Data(3.3, 3.4)

() Natural Gas, 2, 3, 6 in. Heado Propane, 2, 3 in.

Nozzle'Nozzle

in. Open Pipe Flare Headin. Commercial Pressure-Assisted Head Ein. Commercial Air-Assisted Head Gwithout Pilotin. Commercial Air-Assisted Head Gwith Pilot

Open Pipe Flare HeadOpen Pipe Flare HeadCommercial Pressure-Assisted Head FCommercial Coanda Steam-Injected Head D

LJ 1/2 in.\] 1 in.

1 1/2D 1 1/2Q 1 1/201 1/2o 2 in .o 2 1/2 in.o 3.8 in.

12 in.

New EER Data

Previous EER Data o 3 in . Open Pipe Flare Head(1.3 ) 6 6 in . Open Pipe Flare Head

0 12 in. Open Pipe Flare Head12 in. Commercial Head Ao12 in. Commercial Head Bo 12 in. Commercial Head C

3-7


51/140

present test program shows surprisingly gpod correlation with this leastsquares f i t for flame lengths from 1/16 through 1/4 inch nozzles (Figure3-4), but somewhat poorer correlation for some of the larger, commercialheads as discussed above.

A correlation was developed in previous work (1.3) relating flame lengthto the Richardson number, corrected for the volume change caused by increasein the gas temperature from ambient to the flame temperature:

L/d = 7.41 q x FCpoo Too (1+26x) 0.60 Ri- 0.2163-1

where L = flame length, ftd = flare head diameter, ftq = 2,350 Btu/ft3, propane gas lower heating valuex =mole fraction of propane in the fuelF = fraction of heat lost by radiation from the flame

Cpoo = constant pressure hea t capaci ty of the air, Btu/ft3RToo = ambient temperature, R26 = 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 wasobtained from a single set of data with different mole fraction of propaneand a constant Richardson number of 2.0 as shown in Figure 3-5. Thisrelation was then applied to all other flames. The absolute value of F doesnot matter for a correlation with a single fuel. However, F factors areavailable in the literature and will be required for correlations of flamesburning different gases.

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

3-8


52/140

102 0


53/140

a,1"=GQ .Qs-=-....= 1.0a. Q.Is-a.=-Q .8-..II I


54/140

correlated flame length to heat release to the 0.4 power. This is close tothe exponent of 0.388 determined by a least squares fit of 3,6 and 12 inchhead data as shown in Figure 3-3. Such good agreement also verified thesuccess in correlat ing large flames using Hottel and Hawthorne'scalculations. In addition, convers ion of Stewart's (3.25) formula forcombustion number, predicts flame length based upon density to the 0.4 powerand Richardson number to the -0.2 power. Flame lengths correla ted withequivalent expressions by Thomas (3.19) predicted that the flame lengthshould be proportional to the density change to the 0.61 power and theRichardson Number to the -0.3 power.

Using Equation 3-1, good correlat ion is seen in Figure 3-6 forpreviously tested 3, 6 and 12 inch flare heads (3.1). Good agreement is alsoseen for 2 and 2 1/2 inch f lare heads shown in Figure 3-7, but thecorrelation 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 theRichardson number and hence equation 3-1 tends to overestimate the flamelength. In addition the fraction of heat lost by radiation may be differentfo r small flames than large ones. Inspection of Figure 3-7 shows thatcorrelations for each individual small nozzle could be developed by alteringthe constants in Equation 3-1. However, the correlation would not beuniversal, as i t apparently is for open pipe flares greater than 2 inches indiameter. Similar discrepancies between the correlation and even smallernozzles are shown in Figure 3-8.

The observed flame lengths of the air-assisted heads shown in Figure 3-9are considerably shorter than predicted. This may be the result of thesmaller indiv idual por ts in the fuel spider and/or the increased mixingcaused by the co-flowing air stream.

Flame length predictions shown in Figure 3-9 for commercial flare heads0, E, F, and G also deviate from measured lengths. Poor agreement betweenpredicted and measured flame lengths for these heads is due to the effect of

3-11


55/140

50

45

40

35

30

25

20u

.

15

1U

5

o o 5 10 15 20 25 30 35Observed Flame Length (ft)

Figure 3-6. Predicted flame length from modified Richardsonnumber correlation compared to observed flamelength f or previou sly tested 3,6 and 12 inchflare heads. (From 3.1)

3-12


56/140

1/2 in.

284

2 in.

L] 1/2 in. Nozzl e\} 1 in. Nozzle


57/140

1/8 in.

3

.r:

....g' 2QJ

.....

QJato....

QJ....

U

QJl-e..

1

oo

Linear regression cf1/16-1/4 in. data

1/16 in. Nozzleo 1/8 in. Nozzle

" ' l 1/4 in. Nozzle

2Observed Flame Length ( f t l

3

Figure 3-8. Predicted flame length from modified Richardson numbercorrelation compared to observed flame length for 1/16through 1/4 inch flare heads.

3-14


58/140

D

328

pressure-assisted

24

1/2 in Air-Assisted Head GD 1/2 in. Pressure-Assisted Head E

j 12 in. Coanda Head D

Q 1 1/2 in. Air-Assis ted Head Gwlth Pi joto 3.8 in. Pressure-Assisted Head F

a

12 16 20Observed Flame Length (ftl

8

1 1/2 in. air-assisted head G

4 ' ---''''-''.J

8

36

32

12

40

28.....

....

J:: 24....a>c::


59/140

steam, pressure, and design on flare flames, combined with the previouslymentioned aerodynamic differences between small and large flares.3.3 Flame Stability

The flame stability limit for a given propane-nitrogen gas exit velocityis the minimum gas heating value that maintains a flare flame. For eachf lare head, the minimum heating value to maintain a stable flame wasevaluated at several velocities. Limiting heating values for 1/16 through 21/2 inch heads are compared with previous values for 3 through 12 inch flareheads in Figure 3-10. At heating values on or above the stability curve fora given flare heads, a flame is maintained. The flame is extinguished whenthe gas heating value drops below the minimum value defined by the stabilitycurve.

The broad stabili ty band on Figure 3-10 is for stabili ty curves ofpreviously 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. Thisindicates that flare stabili ty is similar for 1 1/2 inch through 12 inchflare heads. The stability of 1 inch and 1/2 inch nozzles, however, is quitedi fferent. As shown by compari son of the Reynol ds number and Richardsonnumber and by the incompatibility of large and small flame lengthpredictions, the characteristics of flames burned on large flare heads aredifferent from characteristics of flames burned on small nozzles. Thisconclusion is verified by the flame stability results, which show tha t flareheads less than 1 1/2 inches in diameter have different stabili tycharacteristics 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 veloci ty is equal to or greaterthan the relief gas velocity in all directions. The flame will be unstablewhen the gas velocity remains greater than the flame veloci ty until the gasis diluted beyond its lower flammability limit. The flame velocity, however,is controlled by the Arrhenius kinetic parameters of the flame reactions.

3-16


60/140

1000.000.00.0.0


61/140

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

Different flame temperature and stability results are shown in Figure3-12 for the coanda steam-injected, pressure-assisted, and air-assistedcommercial heads tested in the current study. Correlations of the limitingvelocity with pseudo adiabatic flame temperature are good for each head, buthead size and design have a large influence on the correlation. The designcriteria for these commercial pressure and air-assisted flare heads were:

Relief gas flow capacity of 3,440 lb/hr (36,000 SCFH) for gas with1,200 Btu/ft3 heating value.

Approximately 0.01 ft 2 open area, for an equivalent diameter of 1.5inches.

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

For the air-assisted head, two air delivery rates, with a maximumrate of 200,000 ft 3/hr at 20 inches water pressure.

Pressure-assisted head E and air-assisted head G met these criteria, butpressure-assisted head F had an open area of 0.0785 ft2, and an equivalentdiameter of 3.8 inches. For the same calculated flame temperature, each ofthe commercial heads can be operated at a higher stable velocity than theopen pipe flares. Pressure-assisted head E produces stable flames at muchhigher 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

3-18


62/140

.-.

100 PREVIOUS EER DATA (1.3)0 3 in. Open Pipe Flare Head\ 6, 6 in. Open Pfpe Flare Head0 12 in. Open Pipe Flare Heade::. 12 in. Conmte rcia1 Head A0 12 in. COl1lllercial Head B

0 0 12 in. COlTIllercial Head C100

P o l y n O ~ i a l least-squaresfit (R = 0.88)

10

o

Low exit velocity. not representativeof typical operation.

4.0.8.60.1 ! .r__ - ' -__ .......I"--__-'- L..-__ .....__ .......IL..-__-'-__- - - l2.

Adiabatic Flame Temperature (l/R) x 104

Figure 3- 11. Maximum relief gas exit velocity vs flame temperaturefor previously tested 3 through 12 inch flare heads.

3-19


63/140

1000

100

uC1J'".........>,.... 10

>,....

uoC1J::-0


64/140

1 1/2 in. commercialpressure-assistedhead F

1 1/2 in. commercial pressureassisted head E

coanda steam-injected head 0

1 1/2 in. commercial airassisted head G. with noair-assist or pilot

(0.085 in. ID) Nozzle1/8 in.

Low Flowrateso 1/16 in. (0.042 in. ID) ~ - 3-12 in. flare heads, withand without flame retentiondevices

1.0 1/4 in. (0.180 in. 10\ Nozzle1/2 in. Nozzle

'V 1 in. Nozzle1 1/2 in. Open Pipe Flare Head

CJ 2 in. Open Pipe Flare Head0 2 1/2 in. Open Pipe Flare Head0.12.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0

Adiabatic Flame Temperature (l/R) X 104

1000

>, 10...

>,...uoO'lc:

E.-...J.-...

Figure 3-13. Relation between flame stability and pseudoadiabatic flame temperature.

3-21


65/140

exit velocity decreases with decreasing diameter. The limiting veloci ty forflares 1 through 2 1/2 inches in diameter is similar to the limiting velocityobtained for flare heads 3 through 12 inches in diameter, but lower than forthe commercial heads D through G. Figures 3-11 through 3-13 show that formany of the flare heads there is a minimum gas flowrate and correspondingexit velocity, below which the flame becomes less stable. This may be due toflame operation below the velocity where the flame will become unstable fromminor fluctuations in flow or wind conditions, and/or where there is tool i t t le heat input to overcome heat losses.

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

3-22


66/140

3.5 References3.1 Pohl, J . H., N. R. Soelberg, E. Poncelet, and B. A. Tichenor, liTheStructure of Large Buoyant Fl ames", Ameri can Fl arne Research CommitteeFall 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, I I I , J . F. and J . A. O'Leary, "Flare Radiation", presented atth e 2nd International Symposium on Loss Prevention and Safety Promotionin the Process Industries, Heidelberg, Germany, September 1977.3.4 Straitz, I I I , J . F., "Flaring for Gaseous Control in th e PetroleumIndustry", presented at 1978 Annual Meeting of the Air Pollution ControlAssociation, 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 5 2 , 9 1 , 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 In st i t u t e, 1979.3.12 Galant, S., "Un Modele Simplifie de Rayonnement Emis par les Flames deDiffusion Libres,1I 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, CombustionScience 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) onCombustion, The Williams and Wilkins Company, p. 254, 1949.

3-23


67/140

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) onCombustion, p. 905, The Combustion Institute, 1982.3.25 Stewart, F. B., Combustion Science and Technology 2, 203, 1970.

3-24


68/140

4.0 COMMERCIAL FLARE HEAD TEST RESULTSThe primary objective of these tests was to determine the influence of

flare head design on flare combustion efficiency. The flare combustionefficiency was measured for propane-nitrogen mixtures flared using differentcommercial heads. Results show that the combustion efficiency is greaterthan 98 percent for flare heads, except for the air-assisted flare head, whenoperated within their stable flame regimes. The air-assisted head testresults were less conclusive.

The Flare Advisory Committee recommended in January 1984 that additionalcombustion efficiency tests be conducted on other types of commerciallyavailable flare heads. Table 4-1 lists the four different head types thatwere tested. The design of the commercial flare heads is proprietary, andlimited design information can be disclosed. Due to the proprietary natureof 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 exitveloci t ies ranging from 0.2 to 9.9 ft/sec based on the 12 inch designopening. The actual imposed velocities vary in this head because ofinduction of steam and air and the variable cross section. Pressure-assistedhead E had an equivalent diameter of 1.5 inches, and was tested at 14.2 - 907ft /sec. Pressure-assisted head F had an equivalent diameter of 3.8 inches.This head ~ a tested at 4.4-157 ft /sec. The air-assisted head G had anequivalent 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, combustionefficiencies measured on the heads could not be directly compared. Flareheads are designed for specific conditions, such as relief gas composition,heating value and exi t velocity. The operating regime was definedexper imentally for each flare head by determining the minimum heating valuerequired to maintain the flame at a given gas exit velocity. Aminimumstability curve was generated in this manner for each flare head. Combustionefficiency of each head was measured at conditions on this curve of minimumstabili ty and at heating values 50 percent greater than required to maintaina stable flame. Therefore, combustion efficiency was measured at flare

4-1


69/140

Table 4-1.COMMERCIAL HEADS SELECTED FOR COMBUSTION EFFICIENCY TESTS

EquippedEER Open Equivalent With FlameDesig- Area Diameter Retentionnation Flare Head Geometry (i n. 2) (in.) DeviceD Coanda Steam- Upward opening 113 12 noInjected Head cone with steam-injection portsE Pressure- Hori zonta1 bar 1. 77 1.5 yes

Assisted HeadF Pressure- Open 11.3 3.8 yesAssisted HeadG Air-Assisted Cross spi der 1.77 1.5 noHead

4-2


70/140

operating conditions both within and at the lJmits of the operating envelopefor the flare. Low combustion efficiency results for tests at the limits ofthe operating envelope are expected, and are not indicative of normalcommercial operation.

4.1 Coanda Steam-Injected Head 0Flame stabili ty and combustion efficiency were measured on the coanda

steam-injected flare head. This head was designed to have the capacity of a12 inch open pipe flare. Figure 4-1 shows the limits of flame stability forthis head. For a given exit velocity indicated by each data point, reductionof the heating value resulted in blow-out of the flame. All measurements onthis head were made with the minimum steam flow (140 lb/hr) to prevent therelief gas from exiting the injection ports. The shaded area on Figure 4-1represents the region of flame stability previously reported for 3, 6, and 12inch diameter flare heads, both with and without flame retention devices.

Combustion efficiency was measured at relief gas exit velocities between0.2 and 9.9 ft /sec, and at gas heating values near and above the minimumheating value required fo r flame stabil i ty. In Figure 4-2, combustionefficiency is correlated to flame stability using the ratio of gas heatingvalue to that minimum heating value required to maintain flame stability.High combustion efficiency (>99 percent) is obtained where the gas heatingvalue is greater than 10 percent above the heating value required for astable flame. Combustion efficiency decreases as the heating value for astable flame limit is approached.4.2 Pressure-Assisted Head E

Similar measurements were made for pressure-assisted head E. Figure 4-3shows the flame stability curve for pressure-assisted head E. Combustionefficiency 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 showsthe correlation between combustion efficiency and flame stability. Thecorrelation is the same as for other heads tested. Combustion efficiencies

4-3


71/140

1200

llOO

1000

900

(V ) 800+ ''+-'-::>+ 'ro 700--'" 600>enc:.....p- '" 500


72/140

99

98

...c:ClluSoCll 97..>,Uc:Cll. U

.... 96c:0

...V'I..cE0u

95

94

293 ' - --I. ......Jo

Heating Value/Minimum Heating Value for Stability

Figure 4-2. Relation of flame stability to combustionefficiency for commercial 12 inch coandasteam-injected flare head D. Steam flowrateat 140 lb /hr.

4-5


73/140


74/140


75/140

greater than 98 percent are achieved under stable operating conditions.Stable operating conditions are those for which the ratio of heating value ofthe gas to the minimum heating value required for stability is greater than1.3. Combustion efficiency rapidly decreases as the limit of stability isapproached.4.3 Pressure-Assisted Head F

Figure 4-5 shows the stability curve for pressure head F. The stabilitylimit curve is significantly lower than those curves for previously testedheads. Figure 4-6 shows that combustion efficiency can be correlated toflame stabil i ty. Combustion efficiency is greater than 98 percent when thegas heating value ratio is greater than 1.3, but decreases rapidly as theheating value ratio approaches 1.0.4.4 Air-Assisted Head G

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

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

4-8


76/140

~ o _ -0 - - - - -

1200

1100

1000

900(V)

.....

...........:> 800....coQI:> 700'"O lc:

. 600

.....

'"I:t :Vl 500'"L':J

\ 0400

300

200

1000.1 1. 0 10.0

Exit Velocity (ft/sec)100.0

Figure 4-5. Region of flame stability for 3.8 inch commercial pressu


77/140

98

100 ~ - - - - - , . . . . . - - - - - - - , r - - - ~

96.... 0:C1uC1 94:..>,uc:C1.- 92.-UJc:0.- 90...II>::s.a=au

88

86

84 - ! - ~ _ 012eating Value/Minimum Heating Value for StabilttyFigure 4-6. Relation of flame stability to combustionefficiency for commercial 3.8 inch pressure-assisted head F.

4-10


78/140

1300 0 8 1.9547 3.91200

04.71100

0 lb airllb35.9' /1000....'+- 900......::>....coQJ 800:>'">C' l 700 0::::.... 0 0'" 314J 600VI'"'"' 500 Stabil it y Curvewith No Air-Assist

400

300

200

100 0.1 1.0 10.0 100.0 1000.0Exit Velocity (ft/sec)

Figure 4-7. Region of flame stability for 1.5 inch commercialair-assisted flare Gwithout pilot flame.

4-11


79/140

I - ~ - I

Swith

lb ai/711'. ) 46.8 O ~

100.0

12.2

10.0.0

1200noo1000900

,-(Y ) 800'.....--::l+ 'coQI 700:::l 600'"-0>C

. + ''" 500po QI, :I :.... '"to

8/14/2019 EPA-600-2-85-106 Evaluation of the Efficiency of Industrial Flare

EPA-600-2-85-106 Evaluation of the Efficiency of Industrial Flares: Flare Head Design and Gas...

Documents

Transcript of EPA-600-2-85-106 Evaluation of the Efficiency of Industrial Flares: Flare Head Design and Gas...