-------
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-HXJr>
-------
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
-------