-------
§
r«»
3
c
ul
03
d)
L.
o
3 HI
o «->
» t
*-> 2
t- O
1C (/I
4-> 1C
E O
ul o.
o
^-
CM
1_
3
O)
2-19
-------
8
01
*J
!
Ol
.*>
o
0
u
V)
o «
ift
u i
^«
i
(O CO
J. <4-
O O
<»-
OJ
tf) *->
$-
*>
o
u. 10
8
8
2-20
-------
I/I
o
4-1
«J
>
4)
O
§
Ot
s
§
u
0>
4J
O
3
c
o>
r- O
4- V>
*-
U O
I
to
O)
> en
j-
CM
I
CM
01
3
a>
s s s §
V <*» CM ^
s«g
2-21
-------
Adiabatic flame temperature can be used as an Indicator of flame
stability for specific gases. The flame stability limit is approached when
the flame speed approaches the relief gas velocity. Flame speed depends on
mixing and flame reaction rates. Flame reaction rates are functions of flame
temperature, frequency factors, and activation energies for flame reactions.
Figure 2-13 shows that the limiting stable gas exit velocity can be
correlated with an estimated adiabatic flame temperature of the gas mixture
burned in a stoichiometric air mixture. However, the curves for different
gas mixtures and for the pilot assisted flares are different. This indicates
that differences in activation energies, frequency factors, mixture
strengths, and mixing rates for different compounds may affect flame
stability differently for different flare gas mixtures and pilot assisted
flares.
Combustion efficiency for the pilot assisted flares is correlated to
flame stability in Figure 2-14. Combustion efficiency is greater than
98 percent for heating value stability ratios greater than about 1.2.
Combustion efficiency can rapidly decrease when the heating value ratio
decreases below about 1.2.
There are some subtle differences in combustion efficiency performance
of the pilot assisted flames:
Combustion efficiency greater than 98 percent is maintained for'the
3-inch head down to a heating value stability ratio of 1.0. This
could be because the Impact of the pilot on the flame for the 3-
inch flare is greater than on the flame from the 6-inch flare.
Also, tests for the 3-inch head were conducted at slightly higher
gas heating values than for the 6-inch head, because the stability
limit curve for the 3-inch head requires higher heating value than
the curves for the 6-inch head.
Below a heating value stability ratio of 1.0, however, the decrease
in combustion efficiency for the 6-inch head at the high pilot gas
rate and for the double and triple pilot is less rapid than for the
2-22
-------
1000
100
£ 10
a*
1.0
0.1
O
o
O
a
D
O
3-1nch flare with
3-inch flare with
6-inch flare with
6-inch flare with
6-inch flare with
6-inch flare with
single pilot
single pilot
single pilot
single pilot
single pilot
double pilot
Q 6-inch flare with triple pilot
at 2 scfm pilot gas rate '
at 2 scfm pilot gas rate
at 1 scfm pilot gas rate
at 2 scfm pilot gas rate
at 4-5 scfm pilot gas rate
at 2 scfm pilot gas rate
at Z scfm pilot gas rate
Ethylene oxide/N2 gas
mixtures flared front
unassisted 3-inch
open pipe flares
,,y"i»2 9as mixtures flared
from unassisted 3-inch open
pipe flares
Propane/N- dominated gas mixtures flared from
3-inch and 6-inch open pipe flames (unassisted)
2-6 2.8 3.0 3.2 3.4 3.6 3.8
Adiabatic Flame Temperature (1/R x 104)
4.0
Figure 2-13. Calculated adiabatic flame temperature vs. limiting
stable gas exit velocity for propane/N9 gas mixtures
flared using pilot-assisted 3-inch and 6-inch open
pipe flares.
2-23
-------
100
90
0)
"o
1
80
70
3-inch flare with single pilot at 2 scfm pilot gas rate (1984) (2.1)
3-Inch flare with single pilot at 2 scfra pilot gas rate
6-inch flare with single pilot at 1 scfm pilot gas rate
6-inch flare with single pilot at 2 scfm pilot gas rate
6-inch flare with single pilot at 4-5 scfm pilot gas rate
6-inch flare with double pilot at 2 scfm pilot gas rate
6-inch flare with triple pilot at 2 scfm pilot gas rate
Region of combustion efficiency
measurements for previously tested
propane/N- mixtures flared from
3, 6, and 12-inch open pipe flares (2.
0 1.0 2.0
Heating Value/Minimum Heating Value Required for Stability
Figure 2-14. Combustion efficiency of pilot-assisted flares,
2-24
-------
2.2
6-inch head with the single pilot medium and low pilot gas rates,
and the 3-inch pilot assisted head. This may be because, even at
or below the stability conditions, high relative pilot gas rates or
multiple pilots can improve combustion efficiency.
Conclusions
The following conclusions were made based on the results of this report:
There are viable methods for accurate and reliable measurement of
and S02 in flare plume samples.
Flame stability depends not only on flare head design and flare gas
volumetric heating value, but also on compounds present in the
flare gas. Gas mixtures of H2$/N2 can be stably flared at much
lower volumetric gas heating values than can propane/N2 or ~5
percent f^S in propane/N2 gas mixtures.
High H2S destruction efficiency is achieved for H2$/N2 and -5 -
70 percent H2S in propane/^2 gas mixtures when the gas heating
value is at least 1.2 times the level required for flame stability.
For gas mixtures containing both H2$ and propane, H2$ destruction
efficiency was consistently higher than propane combustion
efficiency.
The total gas heating value required for a stable flame, including
pilot contribution, is much lower for pilot assisted 3 and 6-inch
open pipe flares than for the same unassisted flares.
High combustion efficiency is achieved for the pilot assisted tests
when the gas heating value is at least 1.2 times the level required
for flame stability.
2-25
-------
There are subtle differences in the flare combustion efficiency
performance of the different pilot assist configurations tested in
this program. Combustion efficiency greater than 98 percent is
maintained for the 3-inch head even down to a heating value
stability ratio of 1.00. Below a heating value stability ratio of
1.00, however, combustion efficiency decreases more rapidly for the
3-inch and 6-inch head with the single pilot at medium and low
pilot gas rates, than for the 6-inch head with the single pilot at
the high pilot gas rates and with the double and triple pilots.
2.3
References
2.1 Pohl, J. H. and N. R. Soelberg, "Evaluation of the Efficiency of
Industrial Flares: Flare Head Design and Gas Composition," EPA Report
Mo. 600/2-85-106, NTIS No. PB86-100559/AS, 1985.
2-26
-------
3.0
GAS MIXTURE TESTS OF H2S
Unburned hydrogen sulfide emissions from industrial flares are of great
concern because of the large amounts of highly poisonous and noxious hydrogen
sulfide gases flared in the petroleum industry. Work was conducted during
the test program to evaluate the overall combustion efficiency and
destruction efficiency of H£S in gas mixtures. This work repeats limited
testing conducted at EER in 1984 (3.1) on gas mixtures containing propane,
nitrogen, and 5 percent HgS. Measurement errors for f^S prevented evaluation
of H2$ destruction efficiency for those tests.
Testing of HgS gas mixtures proceeded in phases. First, H£S and
measurement techniques were evaluated for accuracy and adaptability to
measurements of plume samples containing both ^S and S0£ at high and low
concentrations. After verifying the accuracy of measurement techniques using
gas standards, tests were conducted using the laboratory-scale Flare
Screening Facility. These tests were conducted to determine at what
concentrations HgS could be burned efficiently. These tests were also used
to verify the usefulness of H2S and S02 sampling and analysis.
Although the flames produced on the 0.042 inch ID nozzle of the Flare
Screening Facility are not similar to pilot scale and full sized flares, use
of this facility has several advantages. Its low capital and operating costs
provided an economical way to quickly evaluate if HgS could be burned
efficiently. The nozzle is enclosed in a reactor shell, isolated from the
environment which allowed undiluted .plume samples to be collected. The small
size and reactor shell enclosure increase safety for flaring toxic gases.
The isolating enclosure also gives the ability to close inlet and exit mass
balances, for verification of sampling techniques.
After the review of measurement techniques and laboratory scale tests
were completed, pilot scale HgS gas tests were conducted using the Flare Test
Facility. These tests repeated the incomplete and partially uncertain H2S
gas mixture tests in EER's 1984 test program, and extended those results to
3-1
-------
include efficiency measurements for flare gases with higher H2S
concentrations.
3.1
Measurement Techniques
Accurate measurement of HgS and $02 levels in plume samples is essential
0
to estimating H£S destruction efficiencies. It is important that
measurements are made before H2S and S02 levels in the samples can
significantly change due to reaction with other sampled species or with the
sample apparatus or containers. It is also important that either the H2$ or
SC>2 species be physically separated (such as by gas chromatography ) before
analysis, or that the presence of $03 in the sample does not interfere with
H2$ measurement, and vice versa. The HgS destruction efficiency measurements
conducted at EER in 1984 were inconclusive because SO? in the samples
interfered with H2$ measurements.
Prior to H2$ gas mixture testing in this test program, a thorough
investigation of H2S and 502 sample techniques was conducted. Analysis
methods for H2$ included:
Concentration of gaseous H2S gas scrubber bottles containing
aqueous cadmium sulfate solution, followed by iodometric titration
Concentration of gaseous H2$ gas scrubber bottles containing
aqueous zinc acetate solution, followed by col on" metric analysis
using the methylene blue method
Reaction of gaseous H2$ with lead acetate, causing darkening of
lead acetate impregnated paper tape
Gas chromatography
t Reaction with color indicators in Draeger tubes
3-2
-------
Methods for S02 sample analysis that were investigated were:
Concentration of gaseous $03 gas scrubber bottles containing
aqueous hydrogen peroxide solutions followed by barium perchlorate
titration
Gas chromatography
t Reaction with color indicators in Draeger tubes
Tests were conducted using standardized gases and gas mixtures. These
tests were conducted to verify applicability of the techniques to measure
plume gas concentration of HgS and SOg by evaluating accuracy, repeatability,
reliability, and measurable ranges. Results of these tests are shown in
Table 3-1. The iodometric titration and lead acetate methods for H2$
measurement gave very inaccurate and unreliable results and were not used for
further analyses. The methylene blue and Draeger tube methods proved to be
most accurate and reliable in the required sample range of 0-100 ppm. Where
possible, both methods were used for H2$ sample analysis. Gas chromatography
was also used when the H2S concentration was within the GC detection range of
25 - 1000 ppm.
For SO 2 measurement, the barium perchl orate titration method was most
accurate. Occasionally, however, low measurements were obtained, possibly
due to sample system leaks or analytical errors. Oraeger tubes and gas
chromatography were used to verify results from the titration measurements.
3.2
Laboratory Scale Tests
Laboratory scale tests were conducted to determine concentrations at
which H2$ could be burned efficiently and to verify the usefulness of H2S and
S02 sample procedures. The Flare Screening Facility, equipped with a 0.042
inch ID nozzle, was used for these tests, because it is less expensive and
safer to operate than the larger FTP. Complete inlet and outlet mass balance
3-3
-------
CM
O
HOLDING TIN
SENSITIVITY
6 M
S
§
^ V* ^*
II-2 5
W) W i
2k B< « I
O X -O-
« X V 0^
-j O> X i
11
* 2
M k
*f i
SJjl
S M CM
...2 fe
8S£
C4*
£*,§
w
+1
CM
in
iSs 5S
Ol
t
+ 1
cv.
Ul
M
in
cw
SS
* ,2
tea
)W LU ^-
sT
sP
?
SSL
3?
tit
5
k
&i I
s
2
e
1
£S
I
3-4
-------
closure could also be determined with this facility to provide verification
of sample procedures.
Figure 3-1 compares flame stability limit results for the HgS gas
mixtures to those for propane/Ng mixtures flared using the 0.042-inch nozzle
and 3, 6, and 12-inch flare heads. The flame stability limit is defined as
the operating condition where an increase in gas exit velocity or decrease in
gas heating value results in flame blowout. The curves in Figure 3-1 were
generated by determining the minimum gas heating value attainable before
flame blowout for different gas exit velocities. Much higher gas heating
values are required to maintain a flame on the 0.042 inch nozzle than for the
3, 6, and 12-inch heads. This is because of the aerodynamic differences
between small and large diameter jets. Reynolds number (Re) indicates the
degree of turbulence in a fluid stream. For jets from the 0.042 inch nozzle,
Re ranged between 10^ to 104, indicating that the flow was in the laminar to
turbulent transition regime. Re for jets from the larger flare heads ranged
from 103 to 106, denoting transition to turbulent flow. Richardson number*
(Ri) indicates relative influence of buoyant forces on aerodynamic flow.
Richardson numbers greater than one indicate dominance by buoyant forces.
The jets from the 0.042-inch nozzle had Ri ranging from 10~6 to 10~3,
signifying inertial force dominance over buoyant forces. For the 3 through
12-inch heads, Ri was between 10~4 and 103. Some of these flames were
dominated by inertial forces and some by buoyant forces. Most commercial
flare flames are dominated by buoyant forces.
The destruction efficiency of HgS and mass balances for these tests are
shown in Table 3-2. The HgS gas content was 4.31 to 69.5 percent for these
tests and the nozzle exit velocities ranged from 8.52 to 27.5 f/s. Greater
than 99 percent H2$ destruction efficiency was measured for all test
conditions. Mass balance closure for carbon, oxygen, and sulfur was less
than 1.00 for all but one case. The best closure was with oxygen. Carbon
and sulfur closure ratios were generally less than the corresponding oxygen
closure ratios, but satisfactory for measurement of flare combustion and
destruction efficiency.
3-5
-------
2400
2200
2000
1800
1600
«£"
u
§ 1400
*-»
oo
0
5 1200
0>
19
= 1000
O 1984 Tests, Propane/«2 Mixtures (3.1]
<£> 1985 Tests, Propane/(^Mixtures
O Tests, 4-5X H-S i Mixtures
^ Tests, 30-41* H2S 1 Mixtures
<£> Tests, 50-70X H2S Mixtures
«0
800
600
400
200
10.0 100.0
Exit Velocity (f/s)
1000.0
Figure 3-1. Flame stability curves for propane/N2 and FLS/propane/
N2 gas mixtures flared using a 0.042 inch ID nozzle.
1. Remainder is propane and N
3-6
-------
VI
(/) UJ
l-l
Csj
Z CM
o
I-* II
o
UJ UJ
2 _l
i f>J
z r-j
on
o »
co
3
O
V
e a> i/i
« u a
^1/T
N "- 0^
0 £ "
c? «*u
I/I p- 3 *
« ^ 0 '
a.'o
o ^ C3
v at CM
* f» in
at at eo
o d o
P» g eo
in P* eo
o o d
at
at at at
at at at
^^ ov ^n
O) 91 ^n
O) OV O>
0) Ol Ol
A A A
m a o
* at at
m CM CM
tn « in
P* PK 10
«O P>> O
eo CM eo
c ^ v
CM O O
in CM in
^ en in
t^ eo at
***) ^3
M CM m
1-4
d
eo
r-t
at
d
CM
m
at
d
at
at
a«
fv.
at
at
A
£
«
O
CM
i
in
m
eo
CM
w
p*.
d
at
eo
d
at
CM
d
VO
eo
at
at
at
at
at
A
s
*
CM
in
eo
in
CM
tn
O
at
vo
in
3-7
-------
3.3
Pilot Scale Tests
Pilot scale tests of H£S gas mixtures were conducted using a non-
assisted 3-inch open pipe flare head at the Flare Test Facility. These tests
were conducted to measure flame stability, combustion efficiency, and
destruction efficiency for HgS gas mixtures. Results were used to verify
incomplete results of the EER 1984 test program and to extend the 'test data
to include gas mixtures with higher ^S concentrations.
Flame stability limit measurements are shown in Figure 3-2. These
curves were generated by determining the minimum gas heating value attainable
before flame blowout for a range of different gas exit velocities. Good
reproducibility was attained between the 1984 and 1985 tests with -5 percent
HgS in prop«ne/N2 mixtures. Since the dominant combustible gas in these
mixtures is propane (14.4 - 20.1 percent), it is not surprising that the
stability curve for these mixtures is within boundaries previously determined
for propane/Ng mixtures flared on the 3, 6, and 12-inch heads.
One other method of-depicting the flame stability limit is to correlate
the minimum gas heating value on a mass basis to the gas exit velocity. The
difference between volumetric heating value and mass heating value for
gaseous compounds depends on the compound molecular weight. Compounds with
high volumetric heating values (1,3-butadiene, for example, having 2,730 Btu/
scf) may have low mass heating values (19,500 Btu/lb for 1,3-butadiene)
compared to the heating values of other compounds (hydrogen, for example,
having only 275 Btu/scf, but 51,000 Btu/lb}. In Figure 3-3, the stability
curve shown in Figure 3-2 is presented showing heating value on a mass basis.
The correlation with exit velocity for different gases is no better than in
Figure 3-2. Since the majority of the flare gas in all these mixtures is
nitrogen (50 - 95 percent), the overall molecular weight and density of these
gas mixtures only changes slightly from mixture to mixture.
The flame stability limit curve for HgS/N2 gas mixtures in either
Figure 3-2 or Figure 3-3 is much lower than for the H2S/propane/N2 or
propane/N2 mixtures, but the volumetric heating value for H2S gas (588 Btu/
3-8
-------
M tn
2! £
CM CM
^
s i
Q.
Q.
O.
2
O.
t- f- U)
(/I l/>
CM CM
in in
t I
VI
CM
I
o
o
8
> 01
4-> I-
1- It)
8
sag
3-9
-------
3
u
o
o
*J
<0
(/I
v>
3
u
>»
CO
CO
8
3-10
-------
scf) Is much lower than that of propane (2350 Btu/scf). There is actually a
higher concentration of H£$ (28.9 - 49.5 percent) in the H2S/N2 mixtures at
the stability limit than the combined propane and H2$ in the 5 percent H2$
mixtures (18.8 - 25.2 percent). The presence of higher H2$ concentrations,
and/or better combustion kinetics for H2S than for propane, may help to
maintain a stable flame at lower heat input.
The flame stability limit is approached when the flame propagation speed
in the gas approaches the relief gas velocity. The determination of flame
speed is very complex, involving reaction kinetics and mixing patterns. The
inherent complexities discourage a direct evaluation of flare flame kinetic
and mixing rates. However, the higher levels of H2$ concentration in a
mixture near the stability limit could increase the flame speed (unless the
flame reactions are kinetically limited) by providing a greater number of H2$
- 02 molecular interactions relative to the total number of molecular
interactions.
However, the flame speed may be limited by reaction kinetics, and not
affected by mixing. Kinetic rates of flame reactions are dependent on flame
temperature and other parameters according to the exponential Arrhenius
equati on :
Ae
-E/RT
(3-1)
where k is the rate constant for a given reaction, A is the frequency factor,
E is the activation energy, R is the ideal gas constant, and T is
temperature. If all parameters except temperature in Equation 3-1 are
maintained constant as the temperature is increased, the reaction rate
constant exponentially increases. If the flame speed is limited by reaction
rates, the flame speed (and limiting maximum gas exit velocity) should
increase exponentially as the flame temperature increases.
The actual flare flame temperature is however difficult to determine
accurately. Since the flame is in the open air, subject to wind and eddy
conditions, both in-flame and optical temperature measurement methods are
3-11
-------
subject to errors. Also, within the flame envelope, thereis a wide
temperature range due to varying degrees of air dilution and reaction
completion. As a basis for comparison, a pseudo flame temperature can be
used for showing the dependence of flame stability on temperature for
different flare gases. This pseudo flame temperature may be calculated
assuming (1) adiabatic, complete combustion of the flare gas with a
stoichiometric amount of air, or (2) adiabatic, complete combustion of a
flare gas-air mixture with stoichiometry between the upper and lower
fl amiability limits of the mixture, or (3) non -adiabatic, complete combustion
of a flare gas-air mixture, assuming a certain amount of heat loss. In
previous studies (3.1 and 3.2), pseudo adiabatic flame temperatures based on
stoichiometric combustion of flare gas with air were .compared to flame
stability. This pseudo adiabatic flame temperature was also combined with
both upper and lower flammability limits and relarted to flame stability.
Within the scope of this report, and in order to make comparisons with
previous EER research (3.1), pseudo adiabatic flame temperatures based on
stoichiometric combustion were related to flame stability in Figure 3-4.
The limiting maximum stable gas exit velocity is shown to correlate well
with adiabatic flame temperature for individual gas mixtures. However, there
is poor agreement between the correlations for the different gas mixtures. A
much higher flame temperature is required to burn propane mixtures at a given
velocity than other gas mixtures. Gas mixtures of H2S exhibit the lowest
flame temperature for a given gas exit velocity. Therefore factors other
than flame temperature also affect the flame stability of different gases.
Such differences in flame stability of different gas mixtures shown in
Figure 3-4 1s expected, considering the structural, physical, and chemical
differences of the different compounds. Each of the compounds shown in
Figure 3-4 has a different structure and properties distinctive of different
classes of materials - alkanes (propane), conjugated dienes (1,3-butadiene),
oxygenated compounds (ethylene oxide), and sulfur compounds (hydrogen
sulflde). Some comparative physical properties for H^S and propane are shown
in Table 3-3. H2$ has a wider flammable range, a lower adiabatic flame
temperature, and a lower minimum ignition temperature than does propane.
3-12
-------
1000
100
10
+J
GO
£ 10
u
o
s
O>
1.0
0.1
Flare Gas Composition
O Propane N2
Q Ammonia/Propane/Np
O 1,3 Butadiene/N? c
O Ethylene Oxide/N-
^ Hydrogen Su1fide7Propane/N? (1984)
(7 Hydrogen Sulfide/Propane/N- (1985)
O Hydrogen Sulfide/N2
1,3-Butadiene
Propane-Dominated'
Gas Mixture
«-rvo.
Hydrogen Sulfide
Ethylene Oxide
2.4 2.6
2.8
3.0 3.2 3.4 3.6
3.8 4.0
Adiabatic Flame Temperature (1/R) x 10
Figure 3-4. Calculated adiabatic flame temperature versus
limiting stable gas exit velocity from an
unassisted 3-inch open pipe flare.
3-13
-------
TABLE 3-3. PHYSICAL PROPERTIES1 OF HS AND PROPANE AT 60°F, 1 ATMOSPHERE
Property
Molecular Weight
Gas Density, (scf/lb)
Lower Heating Value (Btu/scf)
(Btu/lb)
Flammability Limits in A1r (%)
Lower
Upper
Adiabatic Flame Temperature for
Stoich1omet1c Combustion 'With Air (R)
Calculated2
Literature
Ignition Temperature (F)
Flame Velocity (f/s)
H2S
34.076
11.4
588
6700
4.3
45.5
3338
NA3
558
NA3
Propane
44.097
8.77
2350
20600
2.1
10.1
3838
4055
871-898
0.95-1.3
1. Data From: Balzhlser, R. E., M. R. Samuels, and J. D. Eliasson,
Chemical Engineering Thermodynamics, 1972
CRC Handbook or Tables for Applied Engineering Science, 2nd Ed., 1976
CRC Handbook of Chemistry and Physics, 53rd Ed., 1972-73
Howard, H. F. and G. W. Jones, "Limits of Flammabllity of Gases and
Vapors", USBM Bulletin 503, 1952
GPSA Engineering Data Book, 1972
Gas Engineers Handbook, 1st Ed., 1965
North American Combustion Handbook, 1st Ed., 1952
2. Calculated by integration of heat capacity data
3. Not Available
3-14
-------
These differences indicate differences in other factors in Equation 3-1 which
could influence kinetic rates for reactions in 838 and propane flames.
The relationship between flame stability and flare combustion efficiency
(see Appendix D for efficiency definitions) has been developed in previous
studies. Figure 3-5 relates propane combustion efficiency for the H2$/
propane/N£ gas mixtures to the ratio of actual gas heating value and the
minimum heating value required to produce a stable flame at that velocity.
This correlation is within the range previously developed for a variety of
gas mixtures and flare head types and designs. Combustion efficiency is
high, exceeding 98 percent, for stability conditions greater than about 1.2.
When stability conditions decrease below 1.2, combustion efficiency can
rapidly decrease below 98 percent.
A similar correlation for H2S destruction efficiency is seen in
Figure 3-6. All data points except one fall within or above the previously
determined destruction efficiency region for ammonia, 1,3-butadiene, ethylene
oxide, and propane. High destruction efficiency, greater than 99 percent, is
attained for stability conditions greater than about 1.2. Destruction
efficiency rapidly drops below 98 percent when the stability criteria
decreases below 1.2. The presence of the single low efficiency point outside
the shaded area is from a test that may have been influenced by ambient
conditions or an upset or discontinuity in flow control or sample procedures.
Since these tests were predominantly conducted very near the limit of flame
stability, any slight change radically affected flare performance and
efficiency.
Destruction efficiency of H2$ is compared to destruction efficiency of
propane in Figure 3-7. For the HgS/propane/^ mixtures and operating
conditions of these tests, H2S destruction efficiency was consistently higher
than propane combustion efficiency. This indicates that ^S flames may be
more stable and react more quickly than propane flames. Figure 3-2 shows
that H2$ flames are more stable than propane flames.
3-15
-------
99.99
99.9
oi
u
0>
o
Ol
99-0
o
in
O
90.0-
Region for hydrocarbon cgabastlojt
efficiency for previously tested
propane/ll* wfxtures flared fro* 3&
through 12 inch heads, and for 1,3-
butadiene/*-, etngtene oxide/S-, MM
propnt/ll^* and H^S/propatte/M, gas J
vfxtores reared froiR an onasj
0.01
*
O
0.1 £
01
0)
Q.
I
fe
U
V
s.
1.0
1984 5X H9S in >,
Propane/N,^Mixtures (3-1) *
f- **
[7 1985
1985 5X H-S In
Propane /N 2 Mixtures -
HwS in Propane/
Mixtures
0 1.0 2.0
Heating Value/Minimum Heating Value for Stability
0}
10.0
100.0
Figure 3-5. Propane combustion efficiency of H?S/
propane/N, gas mixtures flared from an
uriassistefl 3-inch open pipe flare.
1. Scale is log (100 - Combustion Efficiency)
3-16
-------
99.99
~ 99.9
o>
u
-------
41
O
41
100
98
4>
z «
u
*>
1/1
CM
92
90
45° Line
5X H2S 1n Propane/N2 Mixtures
1n Propane/N2 Mixtures
90 92 94 96 98 100
Propane Combustion Efficiency (Percent)
Figure 3-7. Correlation between H-S destruction efficiency
and propane combustion efficiency for H-S/propane/
N~ mixtures flared from an unassisted 3-inch open
pipe flare.
3-18
-------
3.4
References
Pohl, J. H. and N. R. Soelberg, "Evaluation of the Efficiency of
Industrial Flares: Flare Head Design and Gas Composition," EPA Report
No. 600/2-85-106, NTIS No. PB86-100559/AS, 1985.
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.
3-19
-------
-------
4.0
PILOT ASSISTED TEST RESULTS
Very little testing of pilot assisted flares has previously been
conducted at EER or reported in the literature. Most previous work at EER
has been on unassisted flares, in order to simplify the test programs and to
evaluate operating parameters independent of pilot assist. Industrially,
however, continuous gas pilots are commonly used to ensure safe, continuous
and emergency flaring.
Upon recommendation of the flare Technical Advisory Committee, more
extensive pilot assisted testing has been conducted at EER. For these tests,
the flare gas was a mixture of propane blended with nitrogen to vary the gas
heating value. Utility-supplied natural gas was used for the pilot gas.
Open pipes of 3 and 6-inch diameter were used as the flare heads. Single,
double, and triple pilots were tested. The single pilot was a John link Co.
2-inch gas pilot equipped with a flame front generator. The double and
triple pilots were manufactured by EER and are similar to the Zink pilot.
Orientation of the pilots to the flare as suggested by the Advisory Panel is
shown in Figure 4-1.
Operating parameters evaluated were the number of pilots and the pilot
gas flowrate. Tests were conducted on the 3-inch and 6-inch flares with a
single pilot and a nominal pilot gas flowrate of 2 scfm. Using the 6-inch
flare, tests were also conducted with the single pilot at pilot gas flowrates
ranging from 1-5 scfm. Tests were also conducted on the 6-inch flare with
single, double, and triple pilots, maintaining the total pilot gas rate at
2 scfm.
The flare gas exit velocity range for all these tests was limited to the
range between 8-150 f/s for the 3-inch head and 5-50 f/s for the 6-inch head.
Lower velocities were not used because of the high pilot gas rate relative to
propane flowrate in the flare gas. Higher velocities were limited by the
maximum nitrogen flowrate capacity of the Flare Test Facility.
4-1
-------
A. Single pilot
B. Double pilot . Radial
angle between pilots is
180 degrees.
C. Triple pilot. Radial
angle between pilots is
120 degrees.
Dimensions;
A. 30 degrees
8. 2.5 Inches
C. 1 inch
Figure 4-1. Schematics of single, double, and triple pilots.
4-2
-------
4.1
Flame Stability
For all combinations of pilot assist, flame stability limit curves were
determined. The flame stability limit is defined in Appendix C as the
condition where a decrease in flare gas heating value or an increase in flare
gas exit velocity results in flame blow-out. The previously used flame
stability limit determination procedure was to incrementally decrease the
flare gas heating value at constant gas exit velocity until flame
destabilization and blowout occurs. This definition and procedure could not
be applied in the pilot assisted tests because the presence of a pilot
prevented flame blowout, even at very low flare gas heating values.
Consequently, the definition and determination of the flame stability limit
were more subjective. The gas heating value required for 98 percent
combustion efficiency at a given flare gas velocity was observed to be the
operating condition where the last faint flickers of orange color completely
disappeare and the flame envelope becomes completely transparent, except for
blue-orange cones near the pilot and flare tips. This limit is determined by
gradually decreasing the flare gas heating value, maintaining constant gas
exit velocity and pilot gas flowrate and observing the flame, as orange and
yellow color in the flame decreases and eventually disappears. This
definition applies to specific pilot assisted tests of this study burning
propane/nitrogen mixtures.
Flame stability for the 3-inch pilot assisted head is shown in Figure
4-2. Curves defining stable, stability limit, and unstable flames are shown.
The gas heating value includes the added heating value contribution of the
pilot gas. Some scatter in the data points is due to the subjectivity of the
visual measurements and to slight fluctuations in the ambient conditions,
such as lighting and wind variations. For a given gas exit velocity, the
stability curves occur at significantly lower gas heating values than the
region of stability limit curves for unassisted 3 through 12-inch flares.
Clearly, this pilot-assisted open pipe flare can be operated stably at^much
lower total heating values (pilot gas included) than the same flare without
pilot assist. Again, it should be noted that these data are for very simple
open pipe flares. In commercial practice a variety of aerodynamic
4-3
-------
§
o
o.
O
t
r- at
to ^
M O
0) C
«J tfl
U. O)
I
8
I*.
8
VO
§ 8
§
-------
stabilization devices are commonly used. Flares with such features may
exhibit differences between piloted and non-piloted performance which are
smaller than for the simple system studied here.
Flame stability curves for the 6-inch flare with a single pilot at low,
medium, and high pilot gas rates are shown in Figure 4-3 through 4-5. For
each pilot gas flowrate series, the curves of total gas heating value
(including pilot) vs flare gas exit velocity for the stability limit, as
defined by the disappearance of orange coloring in the flame envelope, are
shown. These stability limit curves depend only slightly on velocity in the
range tested. For low and medium pilot gas flowrate, the limiting total gas
heating value is about 150 Btu/scf for flare gas exit velocities between
5.6-45 f/s. For comparison, Figure 4-4 shows that this limiting heating
value is the same as that for the 3-inch pilot assisted flare with a medium
pilot gas rate, at a flare gas exit velocity of 35 f/s. The limiting heating
value for the tests with high pilot gas rate is also constant, but slightly
lower at about 120 Btu/scf. Tests conducted below this level were in the
unstable regime.
Flame stability curves for the 6-inch flare with multiple pilots are
shown in Figure 4-6. The stability limit for the double and triple pilot
tests is flat at a constant heating value of about 130 Btu/scf, for gas exit
velocities between 10-30 f/s. This is slightly lower than for the single
pilot assisted tests at equal or lower pilot gas rates, but slightly higher
than for the single pilot tests at higher pilot gas rates. At gas exit
velocities above 40 f/s, the stability limit for the multiple pilot tests
gradually increases as it does for the 3-inch pilot assisted flare tests.
Use of pilot assist greatly enhanced flame stability during these tests.
For 3 and 6-inch unassisted open pipe flares, operated with propane-nitrogen
gas exit velocity of 40 f/s, the minimum gas heating value required to
maintain a flame is about 540 Btu/scf. If a 2 scfm natural gas pilot used,
the total heating.value (including pilot contribution) required for a stable
flame can be reduced to 150 Btu/scf. At this condition, the flame envelope
becomes transparent and by definition, the stability limit is reached. For
4-5
-------
S **
*><*
t I
i
I »
3
§
o
3
3
to
3
in
2 «
E i
^ ->
O M
%
§ .
s 6
II
it
I :
**
I
i :
§
u
o
"5.
I
o
g
a.
9)
§
I
o *«
0) ^
S_ «>
IB
-------
§
ex
i- (O
I- *
* J_
0) C
o,*-
u
c
0)
T- X
,UJ
VI
C9
O
.c
O I
c
r" ^~
|
IO CM
o o
4-
O r
10 O
-> f
«/) V
Q.
$-
CD
iZ
4-7
-------
I/)
s
at
*J
a
u
v> '
at
i/i
£ <«
u
C U)
I I
at
i_
o <*>
0)
0)
^ CO
ja ia
ifl O)
3
o
o
CM
o
o
4-8
-------
V)
O
01
ut
o
u
VI
O)
(O
o o
1 g
§ 8
3 3
CM T-
o
3
10
in
O
Q.
0)
3
o o
*-- O
a
o> s-
i. 3
0) E
Q.<*-
C <*)
Q.CM
O
u o
c *
T- CM
VO H-
O
O «
fc. O
0 «f^
O
t-
^ o
01
s_
O)
(J9s/n*a) ante A
4-9
-------
the 6-inch flare, the same heating value reduction can be attained with the
pilot at only 1 scfm. Added pilot assist, however, only marginally increases
flame stability. Increasing the pilot gas rate to 5 scfm only reduces the
limiting heating value to 120 8tu/scf for the 6-inch flare. Increasing the
number of pilots to two or three while keeping the total pilot gas rate
constant at 2 scfm decreases the limiting heating value to 130 Btu/scf.
Another indicator of improved flame stability with pilot assist is the
calculated adiabatic flame temperature of the gas mixtures flared at or near
the stability limit. In Figure 4-7, the maximum stable exit velocity is
compared to the adiabatic flame temperature for pilot assisted and non
assisted flares tested at EER. The pilot assisted flares could stably burn
propane/N2 gas mixtures with much lower adiabatic flame temperatures than non
assisted flares could burn HgS/Ng, ethylene oxide/Nj, and especially propane/
Mg gas mixtures. The scatter in the data for the pilot assisted tests Is
apparently due to the increased difficulty in determining the flame stability
limit due to fluctuations in ambient conditions and because the flame cannot
be extinguished with pilot assist.
4.1.1 Scaling Flame Stability
The quantitative flame stability results discussed above are applicable
to flares of similar size and geometry (open-pipe, no aerodynamic or other
stabilization devices). There is however a question with regard to the
extent that these data are scalable to flares of a larger size. In an
earlier study, Pohl and Soelberg (4.1) have reported aerodynamic and flame
stability similarities for flare heads sized from 2-1/2 to 12 inches in
diameter. Extrapolation of the pi lot-assisted results for the 3 and 6-inch
heads, with one to three pilots at a pilot gas flowrate range between 1-5
scfm, may be used to predict flame stability for pilot-assisted open pipe
flares up to 12 inches in diameter. A conservative prediction would be
further IJmited by (1) using the minimum pilot gas to flare gas volumetric
flowrate ratio of these tests (0.002) as the minimum allowable ratio for
operation, (2) using three pilots on a 12-inch flare (even though little
dependence of stability on number of pilots was observed on the 6-inch head)
4-10
-------
1 1 1 1
O 3-inch flare with single pilot at
O 3-inch flare with single pilot at
O 6-inch flare with single pilot at
Q 6-inch flare with single pilot at
Q 6-inch flare with single pilot at
Q 6-inch flare with double pilot at
Q 6-inch flare with triple pilot at
T
T
1000
100
u
o
10
1.0
0.1
2 scfm pilot gas rate
2 scfm pilot gas rate
1 scfm pilot gas rate
2 scfm pilot gas rate
4-5 scfm pilot gas rate
2 scfm pilot gas rate
2 scfm pilot gas rate
84
Ethylene oxide/N, gas
mixtures flared from
unassisted 3-inch
open pipe flares
H-S/N- gas mixtures flared
from unassisted 3-inch open
pipe flares
Propane/N, dominated gas mixtures flared from
3-inch and 6-inch open pipe flames (unassisted)
2.6 2.8
3.0 3.2
Adiabatic F1
3.4
3.6
3.8
4.0
Temperature (1/R x 10*)
Figure 4-7. Calculated adiabatic flame temperature vs. limiting
stable gas exit velocity for propane/N- gas mixtures
flared using pilot-assisted 3-inch and 6-inch open
pipe flares.
4-11
-------
for improved mixing, and (3) limiting the maximum exit velocities to ranges
tested.
With these limitations, a conservative prediction of minimum heating
value is 150 Btu/scf for a stable flame on a 12-inch open pipe flare with
three natural gas pilots. The maximum flare gas exit velocity for stability
would be about 35 f/s. For velocities above 35 f/s, results from the 3-inch
pilot-assisted head tests show increasing heating value limits with
increasing velocity, up to a minimum heating value of 240 Btu/scf at 200 f/s.
Maintaining the minimum experimental pilot gas/flare gas ratio of 0.002, the
minimum total pilot gas rate would be about 20 scfm when the flare gas exit
velocity is 200 f/s.
Predictions such as these, although conservative, are based on
extrapolations of the pilot scale data. Verification of these predictions by
further, larger scale pilot-assisted stability testing is however required
before these predictions can be considered valid. Only qualitative results
of the test data reported herein can be applied to flares that have
aerodynamic or other stability enhancing devices.
4.2
Combustion Efficiency
Previous EER studies have shown that flare combustion efficiency can be
related to flame stability. This correlation also holds for the pilot
assisted flares with the flame stability limits defined as above, as shown in
Figure 4-8. Measured combustion efficiencies for the pilot assisted flares
are within approximately the same region of efficiency vs stability ratio
already developed for unassisted 3, 6, and 12-inch flares.
Combustion efficiency is greater than 98 percent for stability ratio
conditions greater than about 1.2. Combustion efficiency rapidly decreases
when the ratio of (heating yalue)/(mini mum heating value required for
stability) decreases below about 1.2. The major difference between the pilot
assisted and unassisted flares efficiency results is that measurements for
the assisted flares were made at very low gas heating values, ranging from
4-12
-------
O 3-inch flare with single pilot at 2 scfm pilot gas rate (1984) (4-1)
O 3-inch flare with single pilot at 2 scfm pilot gas rate
O 6-inch flare with single pilot at 1 scfm pilot gas rate
Q 6-inch flare with single pilot at 2 scfm pilot gas rate
Q 6-inch flare with single pilot at 4-5 scfm pilot gas rate
Q 6-inch flare with double pilot at 2 scfm pilot gas rate
Q 6-inch flare with triple pilot at 2 scfm pilot gas rate
100
41
O
g
01
"o
90
f
80
70
O
Region of combustion efficiency
measurements for previously tested
propane/N- mixtures flared from
3, 6, and£12-1nch open pipe flares (4-1')
0 1.0 2.0
Heating Value/Minimum Heating Value Required for Stability
Figure 4-8. Combustion efficiency of pilot-assisted flares,
4-13
-------
100-211 Btu/scf while measurements for the unassisted flares made at higher
gas heating values of 273-2350 Btu/scf. The stability limits for the pilot
assisted flares are much lower than the stability limits at the same gas exit
velocities compared to the unassisted flares. However, high combustion
efficiency is maintained for the pilot assisted flares until the gas heating
value, including the pilot contribution, decreases to within 1.2 times the
stability limit heating value.
There are some subtle differences between combustion efficiency
performance of the pilot assisted heads. Combustion efficiency greater than
98 percent for the 3-inch pilot assisted flare is maintained even down to a
stability ratio of 1.0, demonstrating higher combustion efficiency between
1.0 and 1.2 stability ratio than for the 6-inch pilot assisted head. This
could be due to the fact that, for the 3-inch head, the differences between
the pilot (and pilot gas flow) are less than for the 6-inch head. Therefore,
the impact of the pilot flame is greater on the flame from the 3-inch head
than on the flame from the 6-inch head. Also, tests for the 3-inch head were
conducted at slightly higher gas heating values than for the 6-inch head,
because the stability limit curve for the 3Mnch head was not as low as the
curves for the 6-inch head.
Below a stability ratio of 1.0, however, the decrease in combustion
efficiency for the 6-inch head at the high pilot gas rate and for the double
and triple pilot is less rapid than for the 6-inch head with the single pilot
at medium and low pilot gas rates, and the 3-inch pilot assisted head. This
is because, even at conditions near and below the stability limit, high
relative pilot gas rates or multiple pilots can improve otherwise very low
combustion efficiency.
4.3
References
4.1
Pohl, J. H. and M. R. Soelberg, "Evaluation of the Efficiency of
Industrial Flares: Flare Head Design and Gas Composition," EPA Report
No. 600/2-85-106, NTIS No. PB86-100559/AS, 1985.
4-14
-------
APPENDIX A
EPA PURE TEST FACILITY AND 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. The facility is reviewed briefly below, and is described in more
detail by Joseph, et al. (A.l) and Pohl, et al. (A.2).
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.
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, for
compatability with 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
exchangers for vaporizing sulfur dioxide and flare test gases, and (3) for
A-1
-------
01
J_
2
LU
i.
en
A-2
-------
Bolltr
..
CEI3II-HXJ-
Figure A-2. Fuel flow control and metering system schematic.
A-3
-------
sample probe heating. Sulfur dioxide Is used as a tracer for flare test mass
balances. Air is used during air-assisted flare tests.
Flare and tracer gases (except natural gas) are supplied from cylinders
or tanks. 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 is delivered from
either a liquid nitrogen tank or liquid nitrogen cylinders to banks of
finned-tube atmospheric vaporizers capable of providing a maximum nitrogen
flowrate exceeding 24,000 scfh. Other liquified flare gases such as hydrogen
sulfide are delivered from cylinders and vaporized in a steam heat exchanger.
This system can deliver over 4,000 scfh of gas, depending upon the compound.
Steam is produced using 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, 1s fed from liquid S02 cylinders and vaporized through a steam-
heated vaporizer at a flowrate of up to 7 scfh. A1r 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 usin$ 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. Membrane tube bundles in Permapure dryers are used to
selectively remove water vapor from the sample stream. The dried gas samples
are collected in Tedlar bags for analysis of 02, CO, C02, total hydrocarbons,
H2S, S02, and NO/NOx content. Other species such as S02, H2S or NH3 are
concentrated into aqueous solutions in absorption bubblers.
A-4
-------
_>1 * 2
2
«
si
T T
:::;.;; iz;;;; 12:
i
>>
O)
-------
A.2
FTF Procedures
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 when the
wind speed is 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 H2S or NH3,
if applicable. The flame is then ignited using a hand-held spark igniter or
a Zink flame front propagating 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 envelope. 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.
While the plume 1s 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 Og, CO,
C02, HC, NO/NOx, soot, and other species in the plume samples.
A-6
-------
A. 3
References
A.I Joseph, D., J. Lee, C. McKinnon, R. Payne, and J. Pohl, "evaluation of
the Efficiency of Industrial Flares: Background-Experimental Design-
Facility", EPA Report No. 600/2-83-070, NTIS No. P883-263723, August
1983.
A.2 Pohl, J. H., R. Payne, and J. Lee, "Evaluation of the Efficiency of
Industrial Flares: Test Results", EPA Report No. 600/2-84-095, NTIS No.
PB84-199371, May 1984.
A-7
-------
-------
B.I
APPENDIX B
FLARE SCREENING FACILITY AND TEST PROCEDURES
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. This facility is reviewed briefly below, and is described in
more detail by Pohl and Soelberg (B.I).
Figure B-l shows the FSF schematic. 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 behavior is similar to that of 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. A single sample probe
is located at the reactor shell outlet to enable sampling of a well-mixed
plume sample.
B-1
-------
MIXING CHAMBER
HEATED LINE
CONTINUOUS SAMPLING
TENAX SAMPLING
.1/16-1/8 IN, NOZZLE
SCREEN
GLASS BEADS
SYRINGE DRIVE L-
GAS
AIR FAN
Figure B-l. Flare Screening Facility (FSF),
B-2
-------
8.2
FSF Test Procedures
The FSF test procedures are simpler 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. The flame
is ignited by hand, and flow conditions are set. There is some time lapse,
depending on the gas flowrates, for flare steady state to be achieved after
gas flowrate or mixture changes because of gas residence time in the
flowlines. After steady state is reached, sample collection is initiated and
visual and photographic flame observations are recorded. Plume samples
collected in Tedlar bags are analyzed for 02, CO, C02, total hydrocarbons,
HgS, SOg, NO/NOX content, and other species using the instruments shown in
Figure A-3. Plume samples are also bubbled through gas scrubber bottles for
concentration and subsequent analysis of such species as ^S, SOg, and NH3-
B.3
Reference
B.I Pohl, J. H. and N. R. Soelberg, "Evaluation of the Efficiency of
Industrial Flares: Flare Head Design and Gas Composition", EPA Report
No. EPA-600/2-85-106, NTIS No. PB86-100559/AS, September 1985.
B-3
-------
-------
APPENDIX C
FLARE FLAME STABILITY LIMIT
C.I
Stability Limits
Previous work (C.I through C.4) has related flare combustion and
destruction efficiency to flame stability. A stable flame exhibits high
efficiency. However, as flame conditions approach the stability limit,
efficiency can rapidly decrease.
The flame stability limit is defined as the operating condition where an
increase in flare gas exit velocity or a decrease in flare gas heating value
results in flame blowout. Theoretically, this occurs when the gas velocity
is not reduced to the flame velocity before the gas becomes diluted (by air
sntraintnent and mixing) below its lower flammability limit or when the flame
speed is less than the imposed velocity.
A characteristic exit velocity vs heating value curve, maintaining all
other conditions constant, can be generated for each flare head and gas
mixture combination. This curve is generated by determining the minimum gas
heating value attainable before flame blowout, for a range of different gas
exit velocities. At exit velocity and heating value combinations above this
curve, a flame will be present; at conditions below the curve, there will be
no flame.
C.2
References
C.I Pohl, J. H., R. Payne, and J. Lee, "Evaluation of the Efficiency of
Industrial Flares: Test Results", EPA Report No. 600/2-84-095, NTIS No.
PB84-199371, May 1984.
C.2 Pohl, J. H., J. Lee, R. Payne and B. Tichenor, "The Combustion
Efficiency of Flares", 77th Annual Meeting and Exhibition of the
Air Pollution Control Association, San Francisco, CA, June 1984.
C-1
-------
C.3 Pohl, J. H. and N. R. Soelberg, "Evaulatlon of the Efficiency of
Industrial Flares: Flare Head Design and Gas Composition"; EPA Report
No. EPA-600/2-85-106, NTIS No. PB86-100559/AS, September 1985.
C.4 Pohl, J. H., N. R. Soelberg and E. Poncelet, "The Structure of Large
Buoyant Flames", American Flame Research Committee Fall Meeting,
Llvermore, CA, 16-18 October 1985.
C-2
-------
APPENDIX D
DATA ANALYSIS
0.1
FTP Data Analysis
Data reduction and analysis must be conducted on the flare test results
to determine flare combustion efficiency. Results of FTP tests must be
corrected for background levels of sampled species and air dilution of the
olume. Also, numerical integration must be conducted using the local probe
measurements and velocities calculated from jet theory. The development and
letails of the Flare Test Facility data analysis procedures are already
"eported (D.I, D.2) so only a summary is provided here.
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:
OF
Ym -
D-l
where DF = dilution factor = volume of air in the local sample divided
by the volume of stoichiometric combustion products.
Y = local concentration of 02, C02, or S02 (tracer)
m = measured in plume
af = air-diluent-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
D-1
-------
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 D-2:
Yh.c
/ DF \
'" "(opTIj Yh»c
D-2
where h - plume species
c = corrected
Local combustion efficiency (CE) can then be calculated using equation
0-3:
CE
i -
"i YT,c
*J YJ,c
D-3
where v* stolchiometric coefficient
1 = 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. 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 D-4:
D-2
-------
DE - 1 -
D-4
where k - fuel species
1 » completely and incompletely burned species fcom the fuel
speci es.
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
D-2, and the local plume velocity is calculated from jet theory using
equation's D-5 and D-6:
Vr.x - Vi
max
D-5
'max
= Vf
D-6
where V * velocity
R » radial distance from plume centerline
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 D-7, D-8, and D-9:
D-3
-------
CE 1 -
Ir £f Z/T Yt>c Vr Ar
D-7
DE * 1 -
V A
D-8
- Rr)
D-9
where Ar = radial area sampled by probe r (Figure D-l)
0.2
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:
Vr.a.
D-10
where V = volumetric flowrate, scfh
e.a. = excess air
S.R. 3 stoichlometric ratio
r.a. » required air for 100% combustion
Vp « Vg
D-ll
D-4
-------
Figure 0-1. Schematic of integration geometry.
D-5
-------
where p « stoichiometric products of combustion with air (air-free
basis, 0 percent 03)
g « inlet gas
v * stoichiometric coefficient
1 » combustion product species "i"
"e.a.
D-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, D.2).
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, because species concentrations In the plume sample
are representative of average plume concentrations. Species flowrates in the
plume are calculated using the measured concentrations and the plume flowrate
found from equation 0-13:
Vt Y,
D-13
0-6
-------
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
background levels. The plume species flowrates must then be corrected by
subtracting the background contribution:
vi,c - Vi - Va Y,.b
D-14
where c = corrected
b = background
Combustion and destruction efficiencies are calculated using equations
D-15 and D-16.
CE * 1 -
D-15
,c
where i = incompletely burned species
j = incompletely and completely burned species
DE - 1 -
1.
D-16
where k = fuel species
1 = incompletely and completely burned species that came from
the fuel species
D-7
-------
0.3
References
D.I Pohl, J. H., R. Payne, and J. Lee, "Evaluation of the Efficiency of
Industrial Flares: Test Results", EPA Report Mo. 600/2-84-095, MTIS No.
PB84-199371, May 1984.
0.2 Pohl, J. H. and N. R. Soelberg, "Evaluation of the Efficiency of
Industrial Flares: Flare Head Design and Gas Composition", EPA Report
Mo. EPA-600/2-85-106, NTIS No. PB86-100559/AS, September 1985.
D-8
-------
APPENDIX E
QUALITY ASSURANCE
E.I
Flowrate Measurement
The flare gases used at both the Flare Test Facility and the Flare
Screening Facility were prepared by mixing the pure components. Accurate
measurement and flow control of the pure gases was required to prepare the
desired flare gas composition and operate the flare at the desired gas exit
velocity. Accurate flowrate measurement and control over a very wide range
was achieved by using banks of calibrated parallel-flow square-edged orifice
plates at the Flare Test Facility. For the smaller-scale Flare Screening
Facility, banks of parallel-flow rotameters were used for accurate flowrate
measurements.
Each of the orifice plates used at the Fare Test Facility was calibrated
jsing the working gas for that orifice and a standard pre-callbrated laminar
flowmeter, dry gas meter, or wet test meter to obtain the flow coefficient
for that particular orifice and gas. This orifice coefficient was used in
jquation E-l for flowrate measurement:
1/2
\MW T
E-l
where Y = flowrate, scfra
K * orifice coefficient
P » static orifice pressure, psia
4P * orifice differential pressure, feet 1^0 column
MM * gas molecular weight
T * orifice temperature, R.
E-1
-------
The standard deviation of K for 13 different orifices used in this test
program was less than 5.4 percent, and less than 3.0 percent for the
majority.
A similar procedure was used to calibrate the rotameters used for
flowrate measurement in the Flare Screening Facility. Each rotameter was
calibrated using the working gas and pre-calibrated dry gas meters, wet test
meters, or water displacement columns. Variations in pressure and
temperature between the calibrated and actual operating values were corrected
for using equation E-2:
1/2
E-2
where subscript 1 is for conditions during the calibration, and subscript 2
is for operating conditions.
E.2
Sample Analysis
Accurate sample analysis is critical for determining reliable combustion
and destruction efficiency results. Table E-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 E-l.
Many of the analytical methods of Table E-l were developed in previous
EER flare studies (E.I. E.2), but methods for t^S and SO 2 measurement were
evaluated in this study. Results of these evaluations are reported in
Section 3 of this report. Only the most reliable of these methods were used
for H£S and SO 2 measurements in the typical ranges observed at the Flare test
Facility shown in Table E-l.
E-2
-------
_I
OQ
<
«/>
§
o <
ss
= ae
*/» H-
< at
as
0
CJ
g
at
=9
U
i
4ff£
PRINCIPLE
1
ae
M
vi
Ul
r j>
f . j
a.
(O
H
If
n
n
n
H
H
n »
M «-
n CM
N 1
it ao
II ~4
n
ii
H
n
i"5!
n « 3
n CM vi
II iQ
M
n
n
n
N
II
if
n 0
M O
U 1
n i
H a
H
II
II
II
II
N B
N
3C
ex
ex
^
i
o
a
3
94 Ut
§i
I
?
i/>
a
01
3
^B
Methyl ene b
Reaction
01
o
*o
o s.
3
o IB
tt
ex
Q
in
i
in
O
c
O>
c *^
^ -M
^j U
*>
1 U
e a»
-. Ul
$
E
ex
ex
o
4
o
_!
O 3
l/l
sj
g
o
1
o
0»
5
V
u
tfk
Chemllunlne
Ul
ao
O
u
at
}w
X
§
I
u
x.
A
1
O
I
co
o
»-
<*- ai
0 t-
3
^4 i/l
2S
t^
A
^
«
o
0
Timed
Collection
S
^
u.
01
3
U
a.
E-3
-------
E.3
References
E.I Joseph, D., J. Lee, C. McKinnon, R. Payne, and J. Pohl, "Evaluation of
the Efficiency of Industrial Flares: Background-Experimental Design-
Facility", EPA Report No. 600/2-83-070, NTIS No. PB83-263723, August
1983.
E,2 Pohl, J. H., R. Payne, and J. Lee, "Evaluation of the Efficiency of
Industrial Flares: Test Results", EPA Report No. 600/2-84-095, NTIS No.
PB84-199371, May 1984.
E-4
-------
APPENDIX F
CONVERSION FACTORS
To Convert From
English
Utu
cfm
in
in Hg
psl
"t
cf
b
nph
British Thermal Unit
Cubic Feet per Minute
Inch
0 Inches Water Column
Pounds per Square Inch
Foot
Cubic Foot
Pound
Miles per Hour
To
Metric
kJ
m3/h
m
Pa
Pa
m
m3
kg
km/h
Kilojoule
Cubic Meters per Hour
Meter
Pascal
Pascal
Metef
Cubic Meter
Ki 1 ogram
Kilometers per Hour
Multiply
By
1.055
1.700
0.0254
249
6893
0.3048
0.02832
0.4536
1.609
Degrees Rankine {R) is converted to degrees Celsius {C) via the following
formula:
C 5/9 (R - 492)
Degrees Fahrenheit (F) is converted to degrees Celsius (C) via:
C = 5/9 (F - 32)
F-1
-------
TECHNICAL REPORT DATA
(Please read Inunctions on the reverse before completing)
1. REPORT NO.
EPA-600/2-86-080
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Evaluation of the Efficiency of Industrial Flares:
Gas Mixtures and Pilot Assisted Flares
5. REPORT DATE
September 1986
6. PERFORMING ORGANIZATION CODE
7. AUTHORISI
J. H. Pohl and N. R. Soelberg
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME ANO 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 ANJ3 PERIOD COVERED
Final; 4/85 - 7/86
14. SPONSORING AGENCY CODE
EPA/600/13
is.SUPPLEMENTARY NOTES AEERL project officer is Bruce A. Tichenor, Mail Drop 54. 919/
541-2991. Earlier reports in this series are EPA-600/2-85-106. -84-095. and
-83-070
£
TR
**
is. ABSTRACT -pj^ repOrt j_g ^g fourth in a series on a research program which will re-
sult in quantification of emissions from, and efficiencies of. industrial flares. The
report gives test data on the combustion efficiency and destruction efficiency of (1)
gas mixtures containing H2S, and (2) flare flames with pilot flame stabilization.
(NOTE: Previous reports were limited to tests burning propane/N2 mixtures on pipe
flares without pilot flare stabilization, and discussed the influence of the flared gas
and flare head design on destruction and combustion efficiency without stabilization
by pilot flares.) The latest tests, conducted on 3- and 6-in. open pipe flares without
aerodynamic flame stabilization devices, led to conclusions including: (1) gas mix-
tures of H2S/N2 can be stably flared at much lower volumetric gas heating values
than can propane/N2 mixtures; (2) destruction and combustion efficiencies greater
than 98% are obtained for gas mixtures of H2S/N2 and H2S/propane/N2 when the gas
heating value is at least 1. 2 times the level required to produce a stable flame; (3)
for mixtures containing both H2S and propane, H2S destruction efficiency was consis-
tently higher than propane combustion efficiency; and (4) combustion efficiencies
greater than 98% for pilot assisted flares are achieved when the heating value is
greater than 1.2 times that required to stabilize the flame.
KEY WORDS ANO DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Croup
Pollution
Flames
Hydrogen Sulfide
Nitrogen
Propane
Pollution Control
Stationary Sources
Industrial Flares
Flame Stabilization
13B
21B
07B
07C
3. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
103
20. SECURITY CLASS (Thit page)
Unclassified
22. PRICE
EPA Form 222O-1 (9-73)
F-2
-------
DATE DUE
Demco. Inc. 38-293
U.S. Environmental Protection Agency
Library, Room 2404 PM-211-A
401 M Street, S.W.
Washington, DC 20460
,.
-------
2 TJ
i« ni
C < C
2 > 0>
O u
< £
o
C
£
o'
fi>
o'
m
o
o
o
i
00
a>
o
oo
o
C
01
ffl
<
I s
i S
I S
o o
3 »
O n
^ fi
Pi (fl
O
-------