Flare Efficiency Study
Engineering-Science, Inc., Austin, TX
Prepared for
Industrial Environmental Research Lab,
Research Triangle Park, NC
PB83-261644
Jul 83
-------
-------
KJ'A-600/ 2-83-052
July 1983
FLARE EFFICIENCY STUDY
by
Marc McDaniel
Engineering-Science, Inc.
2901 North Interregional
Austin, Texas 73722
EPA Contract 68-02-3541-6
EPA Task Officer: Bruce A. Tichenor
Industrial Processes Branch
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
Prepared for:
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
TECHNICAL (UI'OIU DATA
I l;iiril: /|."l> ."I /'«•/, ../•. .'., /.•/, , .
KPA-600/2-83-052
: ' ' . • >-..- -.-. :'' •• i •. i
Klarc T.'fficicnc v Study
Marc' Me Daniel
'A I ,..>••, PJAMI AND AOl.lHI i;S
ciciict.', Inc.
3109 -North Intofre.qional
Austin, Texas 78722
<: V:-NSi)iH( fiBAM.f'Lf MENl" NO."""
ii CON f HACT/GHAN T NO "
68-02-3541, Task 6
13. TYPt
-------
. 'NOTICE .
THIS DOCUMENT HAS BEEN REPRODUCED
FROM THE BEST COPY FURNISHED U-S BY
THE SPONSORING AGENCY. ALTHOUGH IT
IS RECOGNIZED THAT CERTAIN PORTIONS
ARE ILLEGIBLE, IT IS BEING RELEASED
IN THE INTEREST OF. MAKING AVAILABLE
AS MUCH INFORMATION AS POSSIBLE.
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
ii
-------
ABSTRACT
A full-scale experimental study was performed to determine the effi-
ciencies of flare burners as devices for the disposal of hydrocarbon emissions
from refinery and petrochemical processes. The primary objectives of the study
were to determine the combustion efficiency and hydrocarbon destruction
efficiency for both air- and steam-assisted flares under a wide range of
operating conditions. Test results indicate that flaring is generally an
efficient hydrocarbon disposal method for the conditions as evaluated. The
study provides a data base for defining the air quality impact of flaring
operations.
The test methodology utilized during the study employed a specially
constructed 27-foot sample probe suspended by a crane over the flare flame. The
sample extracted by the probe was analyzed by contiguous emission monitors to
determine concentrations of carbon dioxide (CO;?), carbon monoxide (CO), total
hydrocarbons (THC), sulfur dioxide (SO;?), oxides of nitrogen (N0x)»'and oxygen
(0;?). In additton, the probe tip temperature, ambient-air temperature, and wind
speed and direction were measured. Integrated samples of the relief gas were
collected for hydrocarbon species analysis by gas chromatograph. Particulate
matter samples were also collected during the smoking flare tests.
The .rigorous test program included flare testing under thirty-four dif-
ferent operating conditions during .a three-week period in June 198?. Test
variables included Btu content of the relief gas (propylene diluted with
nitrogen), relief gas flow rates, steam flow rates, and air flow rates. When
flares were operated under conditions representative of good industrial opera-
ting practices, the combustion efficiencies at the sampling probe were
determined to be greater than 98 percent. Combustion efficiencies were observed
to decline under conditions of excessive steam (steam quenching) and high exit
velocities o.f low Btu gases.
-------
CONTENTS
Abstract .............. ....... ..... ..... ill
Figures ....... . . ...... . ..... . ..... ..... viii
Tables .......... ...... . ..... .......... ix
Abbreviations and Symbols . . . ......... ......',... x
Section 1 Introduction ..... ..... .... ..... .... 1
Section 2 Conclu-sions. ..... . . ........ ........ 2
Technical Summary ... ..... ............ 2
Conclusions and Observations ....... ..... 5
Section 3 Testing Methodology. ........... ..... ... 6
Experiment Design and Flare Operation ........ 6
Sampling and Analysis . ... . ...-. ......... 8
Types of Flare Burners Tested .... ..... . . . 8
Flare Test Procedures .......... . ..... 13
Background Measurements ..... '-.........-..-.. 14
Continuous Emission Analyzers . ........... 15
Hydrocarbon Species Analysis ..... ..... . . 15
Temperature Measurements ..... ..... ....... 18
Participate' Analyses. ...... ...... .... 18
Moisture Determinations . . ..... ..... . . . . 18
Meteorological Measurements . . . ..... ..... 19
Audio and Video Recordings . . ..... ...... 19
Section 4 Data Collection and Calculations . .-." ....... ... 20
Continuous Analyzers' Data Acquisition ....... 20
-------
Page
Hydrocarbon Species Data .............. 21
Documentation . . . .............. . . , 21
Calculations ............... 22
Section 5 Review of Flare Test Results ...... . . . .... .-.-. 25
Steam-Assisted Flare Tests'-..; ........... 25
High Btu Content Relief Gases . . ....... 27
Low Flow Rate, High Btu Relief Gases . . . . .-.'. 28
Low Btu Content Relief Gases '....„ 29
Purge Rote Relief Gas Flows . . . . . . .... 31
Air-Assisted Flare Tests .31
High Btu Content Relief Gases ......... 31
Low Btu Content Relief Gases .......... 33
Purge Rate Relief Gas Flows .......... 34
Sensitivity of'Combustion Efficiency to
Probe Height . . . .... . . . . . ... . ... . . 34
Effect of Steam-to-Relief Gas Ratio on
Combustion Efficiency • .... 34
Flare NOX Emissions .'.. . 37
Hydrocarbon Analyses ... 39
Particulate Material Analyses ...... . . . . . . 43
Dilution Ratio and Destruction Efficiency
Determinations ................... 43
Moisture Determinations . ...... . 48
Other Flare Test Analyses .............. 48
Section 6 Quality Assurance and Quality Control Activities . . .'. .50
Multipoint Calibrations . . . . . . . . . . . . . . . 50
Zero and Span Checks ................ 50
Instrument Response Times and Through-Probe
Calibration Checks . . . ... ... 53
vi :
-------
Page
Background Measurements . . ... . . . . . .. . , . .56
Combustion Efficiency Error Analysis . , . 56
Appendices
A. Graphical Review of Selected Tests ......... 59
B. Statistical Summaries . . ... .78
C. Calculation of Destruction Efficiency (OE). ... . . 125
D. Soot Composition. . ............ 129
vtl
-------
FIGURES
Number Page
1 Flare efficiency test systems ............. 3
2 Flare flow control system ..... ... 4
3 Flow control and nitrogen cylinder manifolds 10
4 Flare sampling and analysis system ........... 11
5 Flare emission sampling probe 12
6 Sensitivity of combustion efficiency to
probe height (Test 28). . . 35
7 Sensitivity of combustion efficiency to
probe height (Test 57). ......... 36
8 Effect of steam-to-relief gas ratios on flare
combustion efficiency (High Btu content relief
gases) . 38
9 Example gas chromatogram hydrocarbon analysis
(Test 50) ........ 44
vlii
-------
TABLES
Number Page
1 Flare efficiency test results 4
2 Flare emission analyzers and instrumentation ...... 16
3 Gas chromatograph operating conditions 17
4 Steam-assisted flare summary. . . . . ..... . . .... . 26
5 Air-assisted flare summary . ........ . '. . . . . 32
6 Flare NOX results . .'-. . . .... . , ......... 40
7 Hydrocarbon analysis summary
Steam-assisted flare tests .... ..... 41
8 Hydrocarbon analysis summary.
Air-assisted flare tests . . . .... . . 42
9 Particulate analysis .................. 45
10 Smoking flare combustion efficiencies . . . . . . .... 46
11 Flare efficiency test
Moisture content of samples (EPA Test Method 4). . . . 49
1.2 Multipoint calibration checks . . . . .. .- . . . . .... 51
13 Zero/span check summary . . . . 52
14 Instrument response times ....... . ...... .. . . 54
15 Sampling system leak checks . ..'..- . . . . 55
16 Error estimates . . ... . . . ... . .57
ix
-------
LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
AGL
Btu
Btu/hr
Btu/min
Btu/SCF
FEP
ft/min
HP
•ID,
in2
Ibs/hr
1/9
mg/1 .
OD
PPM
psia
psig
SCFM
Above ground level
British thermal unit
British thermal unit per hour
British thermal unit per minute
British thermal unit per standard cubic foot
fluro elastic polymer
feet per minute
horsepower
inside diameter
square inch
pounds per hour
1iters per gram
milligrams per liter ; .
outside diameter
parts per million by volume
pounds per square inch absolute
pounds per square inch gauge pressure
standard cubic feet per minute P 14.7 psia and 70°F
SYMBOLS
CO
C02
N2
NOX
P2
SF6
S02
THC
carbon monoxide
carbon dioxide
nitrogen
nitrogen oxides
oxygen
sulfur hexafluoride
sulfur dioxide
total hydrocarbon
-------
SECTION 1
INTRODUCTION
This document is a, report on an experimental study to determine the
efficiencies of flare burners as devices for the control of continuous
hydrocarbon emissions. The primary objectives of this study were to determine
the combustion efficiency and hydrocarbon destruction efficiency for both air-
and steam-assisted flares over a wide range of operating conditions that might
be encountered in continuous low flow industrial applications. The study
excluded abnormal flaring conditions which might represent large hydrocarbon
releases during process upsets, start-ups and shutdowns.
Both government and industry environmental officials are concerned with
the effects of flaring hydrocarbons on the air quality. However, since flares
do not lend themselves to conventional emission testing techniques, few
attempts have been made to characterize flare emissions. Flare emission
measurement problems include: the effects of high temperatures and radiant heat
on test equipment, the meandering and irregular nature of flare flames due to
external winds and intrinsic turbulence, ihe undefined dilution of flare
emission plume with ambient air, and the lack of suitable sampling locations due
to flare and/or flame heights, especially during process upsets when safety
problems would predominate.
Previous flare efficiency studies did not encompass the range of variables
encountered in the industrial setting. Limited test conditions of flare types,
relief gas types, Btu content, relief gas flow rate, and steam-to-relief gas
ratios were explored. This study was intended to add to the available
literature on the subject by testing the flaring of an olefin (propylene) in
both air- and steam-assisted flares with test variables of relief gas flow rate,
relief gas Btu content, and steam-to-relief gas ratio.
Separate elements of this flare efficiency study were sponsored by-the U.S.
Environmental Protection Agency (EPA) and the Chemical Manufacturers Associa-
tion (CMA). Other project participants included John 2ink Company who provided
flares, test facility and flare operation, and Optimetrics, Inc. who operated
the EPA's Remote Optical Sensing of Emissions (ROSE) system. Engineering-
Science, Inc. (ES) operated the extractive flare sampling and analysis systems
and prepared this report.
-------
SECTION 2
CONCLUSIONS
TECHNICAL SUMMARY
Figure 1 is an overview of the equipment used to operate and test the
flares. The test methodology utlized during the study employed a specially
constructed 27-foot sample probe suspended by a crane over the flare flame. The
sample extracted by the probe was analyzed by continuous emission monitors to
determine concentrations of carbon dioxide (C02), carbon monoxide (CO), total
hydrocarbons (THC), sulfur dioxide (S02), oxides of nitrogen (NOX) and oxygen
(Op). In addition, the probe tip temperature, ambient air temperature and wind
speed and direction were measured. Integrated samples of the flare plume were
collected for hydrocarbon species analysis by gas chromatograph. Particulate
matter samples were collected during the smoking flare tests. Sulfur use was
attempted as a tracer material in an effort to determine the dilution of the
relief gas between the flare burner and the sampling probe location. However,
the implementation of this unproven sulfur balance method for determining
dilution ratios was unsuccessful.
The term "combustion efficiency" was used during this study as the primary
measure of the flares' performance. Conceptually, this term defines the
percentage of flare emissions that are completely oxidized to COp. Mathe-
matically the combustion efficiency is defined as:
C02
" C02 + CO + THC + Soot °
Where:
C02 = parts per million by volume of carbon dioxide
CO = parts per million by volume of carbon monoxide
THC - parts per million by volume of total hydrocarbon as methane
Soot = parts per million by volume of soot as carbon*.
Table 1 summarizes the results of the flare efficiency tests. The rigorous
test program included flare testing under thirty-four different operating
conditions during a three-week period in June 1982. Test variables included Btu
content of the relief gas (propylene diluted with nitrogen), relief gas flow
rates, steam flow rates and air flow rates. Five of the thirty-four tests were
divided into thirteen subtests for purposes of d*ta analysis because the flare
operation did not represent steady-state conditions. The Btu content of the
relief gas was varied from 2,133 to 192 Btu/SCF for the steam-assisted flare,
and from 2,183 to 83 Btu/SCF for the air-assisted flare. The relief gas flow
* In most cases, the "soot" term was zero.
• • ' ' ' : 2 '• ' '. . '
-------
CRUDE PROPYLENE
PROPYLENE
FLOW
ROTOMETER
NITROGEN
FLOW
ROTOMETER
ENGWEERWG SOENCE ANALYSIS TRALERS
STEAM
MEASURING STATION
METEOROLOGICAL
STATION
STEAM BOILER
^"""wiiSw^ifciiiJB
Figure 1. Flare efficiency test systems.
-------
TABLE 1. FLARE EFFICIENCY TEST RESULTS
. . Relief Gas .
lest
Number
F low
(SDM)
Heating
Value
(Btu/SCF)
Steani-io-Relief
Gas Ratio
(Lb/Lb)
. Combustion
Efficiency
(*)
Comments
STEAM-ASSISTED FLARE TES^S
1
2
3
4
8
7
5
67
17
50
66
61
?5
57
J . J
i* .
1U
59a
59D
60
51
16d
16D
16C
16d
54
23
52
53
AIR-ASSISTED
26
65
28
31
66
?9a
29b
64
6?
63
33
32a
32D
473
464
456
283
157
154
149
148
?4.5
?4.4
24.5
25.0
24.7
703
660
B99
656
591
496
3J4
325
320
212
194
169
, 0.356
0.494
0.556
0.356
FLARE TESTS
481. 6
Ii9
157
22.7
639
510
39?
249
217
1?]
0.714
0.656
0.537
2183
21B3
?1U3
?163
2183
?1B3
?1H3
2183
?183
?183
. 21tl3
21B3
2183
294
305
342
364 .
19?
?32
298
309
339
408
519
634
209
267
268
209
2183
21B3
2183
2183
158
16B
146
282
153
289
83
294
228
0.688 ••
O.SOB
0.448
0.
0
0.757
1.66
0.725
0.9^6
3.0/
3.45
_• 6.67
6.86
0.150
.0 .
0
0
0
0
0
0.168
0
0
0
0
0
0
77.5
123
Air Flow, Hi,
Low, Off
Hi
Off
Hi
low
Off
' LOW
Low
Low
Low
Low
Low
Low
Low
99.96
9«.82
99.82
99.80"
98.81«
99.84
9H.94
..
99.84
99. 4i>
99.70
82.18
68.95
99. 9U
99.79
99.86
99.82
97.95
99.33
98.92
98.66
99.73
99.75
99.74
99.78
• 99.90
100.01
98.82
99.40
99.97
99.57* •
99 . 94
99.17
61.94
54.13
64.03
99.74
94.18
99.37
98. ?4
98.94
98.6?
Incipient wiokintj flan-
Smok ing flare
Smoking flare
Incipient smoking flare
Sampling probe in fljre flame
Steam-quenched flame
Steam-quenched flare
No smoke
No snioke
Incipient smoking flare
Smoking flare
Smoking flare; no air o-.MilJnce
Detached flame observed
Detached flanie; no air assistance
Detached flame; with air assistance
Flame slightly detached
— •
* Not accounting for carbon present as soot (see Table 10).
-------
rates ranged from 703 SCFM to 0.35 SCFM (purge flow rate) for the steam-assisted
flare, and from 639 SCFM to 0.54 SCFM (purge flow rate) for the air-assisted
flare.
CONCLUSIONS AND OBSERVATIONS
• When flares are operated under conditions which are representative of
industrial practices, the combustion efficiencies in the flare plume are
greater than 98%. ,
• Steam- and air-assisted flares are generally . an efficient means of
hydrocarbon disposal over the range of operating conditions evaluated.
• Varying flow rates of relief gas have no effect on steam-assisted flare
combustion efficiencies below an exit velocity of 62.5 ft/sec.
• Varying Btu content of relief gases have no observed effect on steam-
assisted flare combustion efficiencies for relief gases above 300 Btu/SCF.
A slight dec'Hne in combustion efficiency was noted for relief gases below
300 Btu/SCF.
» Flaring low Btu content gases at high exit velocities may result in lower
combustion efficiencies for air-assisted flares.
. Smoking flares achieve high gaseous hydrocarbon destruction efficiencies.
• .In- many Cases, where high combustion efficiencies were observed, the
carbon monoxide and hydrocarbon concentrations observed in the flare plume
were approximately equal to those found in ambient air.
• Concentrations of NOX emissions in the flare plume were observed to range
from 0.5 to 8.16 ppm.
• The combustion efficiency data were insensitive to sampling probe height
within the normal operating heights of the probe.
• Further development of a technique to use sulfur or another material as a
tracer material to determine the flare dilution ratios is required.
. Steam-assisted flares burning relief gases with less than 450 Btu/SCF
lower heating value did not smoke, even with zero steam assistance.
• The meandering of the flame's position relative to the sampling probe with
varying wind conditions affected the continuous measurements but had no
apparent effect on the combustion efficiency values.
• Higher concentrations of THC and CO were not observed during the purge
rate flare tests.
-------
SECTION 3
TESTING METHODOLOGY
EXPERIMENT DESIGN AND FLARE OPERATION
The flare tests were designed to determine the combustion efficiency and
hydrocarbon destruction efficiency of flares under a variety of operating
conditions. The tests were devised to investigate routine industrial flaring
operations. Conditions representative of emergency flaring operations were not
investigated. The primary flare operating variables were:
• Flow rate of relief gas; .
• Heating value of relief gases; and
• Steam-to-relief gai ratio (steam flare only).
The preliminary,test plan called for twenty-seven tests, with each test having
a different combination of flare operating variables. The operating variables
were defined as follows:
Relief Gas Flow
High - 25 foot flame length.
Intermediate - 1/6 of high flow.
Low -1/20 of high flow.
The maximum practical flame length that could be tested was approximately 25
feet due to height limitations of the crane boom holding the sampling probe.
This was the limiting factor for setting the maximum relief gas flow rate.
Heating Value •-....;
High - Heating value of the undiluted relief gas (zero nitrogen flow)
(2,200 Btu/ft*).
Intermediate - Twice the low heating value condition (300-600
Btu/ft3).
Low— Lowest heating value that will maintain combustion (high
nitrogen flow) (less than 200 Btu/ft3).
Steam Flow
High- Steam-to-relicf gas mass ratio of 1.0.
Intermediate - Steam-to-relief gas mass ratio of 0.5.
Low- Steam flow at incipient smoking.
Zero
-------
The preliminary test plan called for determination of the vertical profile
of the plume by sampling at least four different heights above the flame. As
discussed on page 34, this was not done due to the insensitivity of combustion
efficiency to probe height. Following the vertical profile measurements, the
flare's efficiency was to be determined at the vertical point where the
combustion reactions are complete but prior to further dilution with ambient
air.
A technical prete.st meeting was held on May 6, 1982 at the John Zink Company
flare demonstration facility in Tulsa, Oklahoma to allow the project partici-
pants to finalize the test plans. During this meeting, six (6) smoking flare
test designs were adopted in addition to the 2.7 tests previously mentioned, for
a total of 33 planned tests. Other items discussed during this meeting
included: the division of responsibilities, lines of communication, quality
assurance procedures, safety considerations, schedules and testing sequence.
During the early stages of the test program, the participants learned more
about the characteristics of flares, and it became apparent that several of the
planned tests were not practical and/or did not represent the intended flare
operating conditions. Therefore, sixteen of the thirty-three planned tests
(numbered 1 through 33) were cancelled and a substitute group of tests
(numbered 50 through 67) were formulated in the field and executed in their
place. The most common reason for abandoning tests was that many of the planned
incipient smoking tests and smoking steam-assisted flare tests would not smoke,
even with zero steam flow.
During each test the flows of the flare feed gases were monitored and
.maintained as close as practical to the target levels. For several tests it was
not possible, due to physical constraints, to maintain all the flow rates at
constant level. This was particularly true for those tests that called for high
nitrogen flow. As the pressure in the N? cylinder banks declined during a test,
the nitrogen flow.would tend to decrease, resulting in higher relief gas heating
values. i
Sulfur was selected as a tracer material to allow estimation of the
dilution of the relief gas from the flare burner tip to the sampling probe.
Sulfur was chosen primarily because of the availability of monitoring instru-
mentation to measure part-per-billion levels of sulfur using flame photometry.
Helium was considered as a tracer material. However, this material is
difficult to quantify at levels less than several tens of parts per million and
thus, would require large quantities of gas. Additionally, helium cannot be
detected on a continuous basis as can S02- Sulfurhexafluoride (SFg) was also
considered as a tracer material. However, SFs is net stable at the elevated
temperatures found in a flare flame.
The sulfur in the relief gas originated from three primary sources: 1)
naturally occurring reduced sulfur in the crude propylene, 2) sulfur added to
the propylene in the form of butyl mercaptan (approximately 1 gallon butyl
mercaptan/6,800 gallons crude propylene), and 3) sulfur dioxide gas added to the
relief gas stream. All three sources and forms of sulfur are presumably
oxidized to S02 as the relief gas is burned. The flare emissions were then
analyzed for total sulfur as SO? using flame photometry.
-------
Crude propylene was selected for the relief gas because it is relativly
difficult to burn smokelessly, as compared to paraffins. The availability of
propylene and safety considerations also influenced its selection as the relief
.gas. Lower Btu. content relief gases were obtained by diluting the crude
propylene with inert nitrogen. Flow rates for both the propylene and the
' nitrogen, were eont» oiled by appropriately sized metering valves and rotameters.
Steam flow to the steam-assisted flares was controlled by a metering valve and
monitored by an orifice meter. Sulfur dioxide was added to the relief gas during
some of the tests to increase the levels of tracer material. The flow of SO^ was
monitored and regulated by a rotameter and metering valve assembly. Figure 2 is
a schematic of the flow controls and plumbing used to operate the test flares.
Figure 3 presents photographs of the flow control manifold assembly and the
nitrogen cylinder manifold arrangement. (
SAMPLING. AND ANALYSIS
An extractive sampling system was use to collect the flare emission samples
and transport these samples to two mobile analytical laboratories. Figure 4 is
a diagram of the sampling and analysis system. The extractive sampling system
consisted of a specially designed 27-foot sampling probe which was suspended
over the flare flame by support cables and a hydraulic crane. This probe
consisted of a 5-foot unheated section of I" stain,less steel pipe coupled with
a 22-foot heated section of 5/8" stainless steel tubing. The heated section was
insulated and housed in a 3" pipe which provided support for the entire probe
assembly. Guy wires were attached to both ends of the 3" pipe support to
position and secure the probe from ground level. Figure 5 contains photographs
of the flare emission sampling probe.
Gaseous flare emission samples entered the sampling system via the probe
tip, passed through the particulate filter, through the heated probe section and
then were carried to ground level by a 3/8" heated FEP teflon tube sample line.
The sampling system temperature was maintained above 100 C to prevent the
condensation of water vapor. The flare emission sample was divided into three
possible paths. A fraction of the heated sample was passed through an EPA
Reference Method 4 sampling train to determine the moisture content of the
sample. A second fraction was directed through a moisture removal cold trap and
thence, into a sampling manifold in one of the mobile laboratories. Sample gas
in this manifold was analyzed by continuous monitors for 0?, CO, C02, NOX and THC
on a dry-sample basis. A third fraction of the sample was directed into a heated
sampling manifold in the other mobile laboratory. Sample gas in this manifold
was analyzed for S02 and hydrocarbon species on a wet basis.
TYPES OF F ,di BURNERS TESTED
The steam-assisted flare used for the test series was a John Zink Standard
STF-S-8 flare tip with two constant ignition silots. Overall length was 12'-3
1/2" with the upper 7'-3" constructed of stainless steel and the lower 5V-1/2"
made from carbon steel. The maximum capacity of the tip is rated by John Zink
Company at approximately 53,300 Ibs/hr for crude propylene at 0.8 Mach exit
velocity. However, the STK-S-8 would not burn this volume of gas and remain
totally smokeless. The capacity of steam flow through the flare steam manifold
: ' . . ' • • - 8 • . ' , ' . •:••..'•••'
-------
•ITKKC*
f.
•*•
iKomn '
COHTHM.
MITMCEll
-95— 1 JTM
-------
Flow.Co
Nitrogen Cy
Figure 3. Flow control and
10
-------
CRANE SUPPORT
/PARTICIPATE
I ^THERMOCOUPLE
KEAHP PROBE-1 UNHEATED PROBE
SECTION SECTION
H_PJ-
) PROBE-'
A
1
•'
•
"
1
1 CO |
1 ««Al"ZE» |
•
-
,
r
EXHAUST"
MOBILE LAB II
THC 1 I CO*
M*LYZE» 1 1 ANALYZER
rrr
/^
777
\
HEAT "TRACE -*
SAHPU LINt
I MOrSTURE I
• j nrrFRM i NAT inn ^, , '
1 APPARATUS J
•
•
I CO 1 HO, I
j ANALYZER | [ ANALYZER |
^
'
—
,
.
HEATED HAH 1 FOLD
H TEMPERATURE
HONITOR
LE
| DATA I
1
I
/
'>
/
H
^
•y
+
:<
.•.
'••::
f=i
•:
T
V
\
v
Jt
METEOROLOG 1 CAL
TOWER
MOBILE LAB t2
' Jt
<2
&
INTEGRATED
BAG SAMPLE
TAPE
HYDROCARBON
SPECIE
-4 SAMPLE 1
SO? TRACER I
ANALYZER |
t
r
,
s
s,
/
s
/
/
\
/
s.
/
\
/
\
/
\
•EXHAUST
Figure 4. Flare sampling and analysis system.
-------
Flare emission sampling probe.
12
-------
is 10,080 Ibs/hr based on steam conditions of 100 psig and 338°F. The
recommended steam flow for this flare is approximately 0.4 pounds of steam per
pound of crude propylene. The steam jet total flow area is 1.92 in? and the
unobstructed flow area at the exit of the 8 5/8" ID steam-assisted flare tip is
27.0 In*.-
The air-assisted1 flare was the John Zink STF-LH-457-5 flare with two
constant ignition pilots. The overall length of this flare was 13'-2". The
upper portion of this flare's air plenum and burner are constructed of stainless
steel and the lower portion is of carbon steel. The maximum capacity of the air-
assisted flare is approximately 23,500 Ibs/hr of crude propylene, which can be
burned smokelessly through use of an air blower. The blower used for this test
series was 7 1/2 HP vane axial fan located in the base of the 18 1/4" ID air
riser. The relief gas is delivered to the tip by an 4" OD internal riser with*
the air supplied around the outside through the air riser and plenum. The relief
gas is discharged via a specially designed "spider" on the end of the internal
riser. The total area of the relief gas holes in the spider burner was 5.30 in2
for the tests on high Btu content relief gas and 11.24 in^ for the tests on low
Btu content gases. The air flow and air flow velocity are proprietary
information and are not included in this report.
Typical field installations of air-assisted flares-utilize two-speed forced
draft fans. The blower normally runs at low speed with automatic advancement to
high speed upon an increased relief gas flow signal. The blower is also
automatically returned to low speed when the increased relief gas condition
subsides. Some deadband is normally provided to avoid excessive speed cycling
of the blower with oscillating flows. Normal low speed operation handles
approximately one-third of the maximum smokeless duty. The air-assisted flare
used in these tests employed an adjustable air inlet vane assembly instead of a
two-speed fan. Adjustment of the vane assembly allowed duplication of the high
and low speed air flow rates without the two-speed fan.
Two different "spider" burner tips were employed during the air-assisted
flare tests. The LH burner tip, designated at "A", was used for tests 26, 65,
28 and 31 for high Btu content gases, and the burner tip designated as "B" was
used for the low Btu content gas tests 66, 29, 64, 62, 63, 33 and 32.
John Zink Standard STF-6-2 pilots were used for both flare tips. At 15
psig, the pilots were designed to burn 300 SCFH of natural gas. The natural gas
burned in the pilots had a lower heating value of 921 Btu/SCF. Two pilots were
used on both the air- and steam-assisted flares, resulting in 552,600 Btu/hr
being supplied to the flare by the pilots.
FLARE TEST PROCEDURES
All key personnel involved in the execution of the flare tests were in
communication with one another via a hard wire intercom system. This
communications system included the following: J2 test coordinator, ES
instrument operator, CMA test observer, EPA ROSE operator, steam flow operator,
rotameter operator, vaporizer operator, crane operator, propylene truck/nitro-
gen bank operator and video camera operator. All conversations between these
persons during tests were recorded on the video tape and on a portable tape
13
-------
recorder. In addition, the two ES mobile laboratories were in communication via
a separate intercom system.
Flaring was not begun until all key personnel were at their stations and
verified that they were prepared to initiate a formal test. Then the test
coordinator would call for flare ignition and the gas flows to the flare would
be adjusted to the previously agreed nominal values. Once the flows were
stabilized the probe would be brought into position by manipulating the
hydraulic crane and guy wires.
t , — ' - :
The probe positioning objective was to place the probe tip as close as
possible to the flare flame without the probe being in the flame. The intent was
to sample the flame emission plume as close as possible to the combustion zone
to minimize the of dilution of the plume by ambient air. The prohe tip was kept
out of the flame so as not to bias the data with gases that were still undergoing
combustion reactions.
Probe positioning was directed by the JZ test coordinator. The test
coordinator's visual probe positioning was aided by observers located in
different quadrants surrounding the flare and the CMA observer who was situated
on an elevated platform. Additionally, the ES instrument operator monitored the
probe tip temperature, CO, C02, and THC.
When the project participants agreed that the probe was positioned as well
as was feasible, the test coordinator announced the initiation of the test and
data collection ensued. The probe position was adjusted as required during th?
test to compensate for changes in wind conditions causing movement of the flame
and the plume away from the probe tip. These adjustments were both vertical and
lateral. The primary criteria for determining the need to adjust the probe
position was a decline in probe tip temperature. Short-term declines in
temperature (i.e. less than one minute) were common as the flare flame and plume
moved with intermittent changes in the wind. However, extended temperature
declines (i.e., greater than two minutes) were regarded as a significant shift
of the wind and signaled • ie need to adjust the probe position.
Data collecton continued for each test for a target period of 20 minutes.
The actual test duration was dependent on a number of factor, which influenced
decision of when to terminate the tests. These factors included;
1. The effects of the flare's radiant heat on buildings, personnel and
test equipment in the area;
2. The representativeness of the data from the standpoint of being able
to maintain good probe positioning during the majoritv of **" i< .t-
and .-.'-.
3. The consumption rates of propylene and nitrogen.
BACKGROUND MEASUREMENTS
Ambient alt concentrations of the compounds of interest were r,?dsurev" in
the test area before and after each test or series of tests. The*e
14 •-'..•' "'.;'
-------
measurements were collected for . a 'minimum period of five minutes. The
background measurements collected before the tests were typically initiated
fifteen to thirty minutes before the anticipated start of the next test.
Background measurements collected after the tests were initiated as soon as all
the instruments indicated a complete return to base!ine concentrations (typi-
cally five to ten minutes after test completion). On occasions when several
tests were executed in a relatively short time period (less than four hours),
the same pair of before and after background measurements were applied to more
than one test. On othe" occasions a set of background measurements collected
after a test would also suffice as the background data set collected before the
next test.
CONTINUOUS EMISSION ANALYZERS
Flare emission measurements of carbon monoxide (CO), carbon dioxide (CO?),
oxygen (0;?), oxides of nitrogen (NOX)» total hydrocarbons (THC) and sulfur
dioxide (SO?) were measured by continuous analyzers that responded to real time
changes in concentrations. These analyzers obtained their samples from the
sample manifolds in the two mobile laboratories. Table 2 is a summary of the
instrumentation used during the tests. The operating principles of these
instruments are well known and are not discussed in detail in this report.
The instruments were operated according to the manufacturers' recom-
mendations, utilizing the primary measurement ranges listed in Table 2. The
only exceptions to this were the operation of the THC and SO? analyzers. During
some tests it was necessary to change the operating range of the THC analyzer to
higher scales due to elevated levels of these compounds. The Meloy SA 285 SO?
analyzer was modified to incorporate a 1:5 sample dilution system. This was
necessary in order to minimize the effect of variable 0? content in the flare
emissions on the instrument response.
All instruments were housed in, air conditioned mobile laboratories to
minimize the effects of temperature on instrument response. However, given the
high radiant heat effects of some of the flare tests, it was not always possible
to maintain a constant temperature within the mobile labs. This factor had the
greatest effect on the NOX and SO? analyzers which employed photomultiplier
tubes in their detection systems. The effect of rising ambient temperature was
noted as a slight shift in the instrument baseline.
HYDROCARBON SPECIES ANALYSIS
Flare emission samples were collected during each test for gas chromato-
graphic analysis for hydrocarbon species. These samples were of two forms:
instantaneous samples and time integrated samples. The instantaneous samples
were periodically withdrawn directly from the sample manifold during each test
and injected into the chromatograph via a gas sample loop. The time integrated
samples were transferred from the manifold into a six liter Tedlar® bag over a
period of five to ten minutes. Subsequently, the integrated samples were
analyzed by gas chromatograph. The analysis techniques for the integrated and
Instantaneous samples were the same. Only the sampling differed. Table 3
outlines the operating conditions of gas chromatograph.
15 " .
-------
TABLE 2. FLARE EMISSION ANALYZERS AND INSTRUMENTATION
Make and Model
Parameter
Primary
Operating Range
Operating Principle
-------
TABLE 3. GAS CHROMATOGRAPH OPERATING CONDITIONS
Gas Chromatograph:
Column:
Packing:
Oven Temperature:
Sample Loop Temperature:
Carrier Gas:
Carrier Flow:
Sample Loop Size:
Sample Valve:
Detector Type:
Calibration Basis:
Lower Detection Limit:
Valving Scheme:
Carle 211
4.9 foot x 1/8 in. stainless steel
n-octane/porasil C, 100/120 mesh
35 °C
35°C
Nitrogen
35 cc/min.
1 cc
Carle 6 port, selonoid activation
Flame ionization detector
Methane equivalents (parts per million)
0.05 ppmas Cfy
Direct injection, no backflush
Elution Times: Minutes
Methane 1.27
Ethane/Ethylene 1.62
Propane 2.44
Propylene 2.91
Butane 4.68
17
-------
TEMPERATURE MEASUREMENTS
The temperature at the sampling probe tip was continuously monitored
during the tests with a chromel-alumel thermocouple in conjunction with a
digital thermometer. The thermocouple selected was an exposed bead type so as
to. minimize the response time. An open end stainless steel shield protected the
thermocouple from the flame's radiant heat and still allowed free circulation of
the flare emission around the thermocouple. Thermocouples were also installed
in the heat trace line, the heated manifold and the heated probe assembly to
allow monitoring of these tempertures during the tests.
PARTICULATE ANALYSES
The probe assembly included an in-line particulate filter housed inside
the heated section of the probe about six feet from tne probe tip. This in-line
particulate filter assembly served two purposes: 1) collection of particulate
samples from smoking flares for subsequent analysis, and 2) maintaining the
cleanliness of the sampling system. The preweighed filter elements used were of
the thimble configuration and constructed of 0.3 micrometer glass fiber.
The filters were changed before and after each of the smoking flare tests.
Following the tests the filters were reweighed to determine the mass of
particulate collected. This information, combined with the measured flow rate
of sample through the probe assembly, allowed the calculation of the gross
particulate concentration of the flare emission at the sampling location. It
should be noted however, that these particulate samples were not collected
isokinetically and thus, represent only gross estimates of the particulate
concentration. The flare particulate emissions were not isokinetically sampled
because it was not practical to directly measure the plume velocity. Due to
small particle sizes, the lack of isokinetic sampling conditions is probably
insignificant.
MOISTURE DETERMINATIONS
The moisture content of the sampled flare emissions were determined by the
procedures set forth in EPA's Reference Method 4 (40 CFR 60 Appendix A). A gas
sample was extracted from the heated sample line and passed through a series of
four impingers immersed in an ice bath. The impingers removed the water from the
sample stream by condensation and by adsorption on silica gel. The weight gain
of the impingers was measured to determine the moisture content of the sample.
The only deviation from the published method required by this appication w
-------
for steam-assisted flare tests and 3.0% average for air-assisted flare tests).
Therefore, moisture corrections were not applied to the data because of their
low levels and questionable accuracy. It is not believed that moisture
corrections would enhance the value of the data.
The results of the moisture determinations may be found in Section 4 of
this report,
METEOROLOGICAL MEASUREMENTS
The ambient wind speed, wind direction and temperature was monitored at the
flare test facility concurrently with the collection of flare emission data.
The meteorological sensors were situated as close as was practical to the test
flares at an elevation approximately the same as the flare tip (12 feet, 8 inches
AGL).
Due to the numerous air flow obstructions in the test area the wind data are
not expected to correspond with the prevailing Tulsa area winds. Rather, the
wind data were intended to represent the wind encountered by the subject flare
flames.
Testing of the flares was found to be infeasible when wind velocities
exceeded 5 miles per hour. Elevated wind velocities prevented sustained and
consistent positioning of the probe in the flare plume.
AUDIO AND VIDEO-RECORDINGS
Audio and video recordings were made during the flare tests. Video
recordings were made to document the flame behavior and the probe position
relative to the flame. The video camera was positioned to have an unobstructed
view of the flame by placing it on a platform approximately 20 feet above ground
level. The distance from the flare to the camera was approximately 50 feet.
Audio recordings were made of the verbal observations of the participants
during the tests. The audio recordings were made on the same magnetic tape used
for the video recordings. The intercom system served as the source of all audio
recordings.
The audio and video recordings were made primarily as means of documenta-
tion of the tests and to allow possible future more detailed analyses of the data
with respect to flame behavior. These recordings were not generally used in the
data analysis contained in this report. The one exception to this is the use of
the recordings to identify the point at which smoking began during Test 11
relative to tfce increasing Btu content of the relief gas.
19
-------
SECTION 4
DATA COLLECTION AND CALCULATIONS
CONTINUOUS ANALYZERS' DATA ACQUISITION
The outputs of the continuous monitoring instruments used for this study
were analog signals that were proportional to the magnitude of the parameter
being monitored. These output signals were recorded on both a strip chart
recorder and on a data logging system. The strip chart records provided a
permanent, continuous record of the analyzer output and a graphical display that
aided in the data interpretation. The electronic data logger system provided a
convenient means to record and process a large quantity of data. Although the
data logger served as the primary means of data acquisition, the strip chart
records provided a back-up data acquisition system :.nd documentation for the
data logger.
The data logger employed for this project was a Monitor Labs Model 9300.
This instrument was coupled with a 9-track magnetic tape recorder (Kennedy Model
9800) and a ten-digit manual data entry system. The functions of the data logger
were as follows:
• Scan each instrument output (approximately every 12 seconds);
• Convert the analyzer's analog output to a digital value;
. Scale the digital value to a useful unit of measure (ppm, mph, etc.);
. Record the scaled instantaneous value on the 9-track magnetic tape.
• Average the instantaneous Values to one-minute averages;
. Print the one-minute, averages on paper tape for on-site review; and
• Label each set of data with the time and the appropriate manually entered
status data.
The original test plan called for the data logger to scan each channel once
every six seconds. However, this was not possible given the number of input
channels to the data logger (10), the required functions, and the speed of the
instrument. Each input channel was scanned for instantaneous data approxi-
mately once every twelve seconds.
The data logger's internal clock was set as closely as possible to Central
Daylight Savings Time (COST). This clock was used as the standard time for all
data acquisition relating to the flare tests.
The printed paper tape output of the data logger provided means to review
the data being recorded on-site by the data logger. This data was compared with
the strip chart data to ensure integrity in the entire data acquisition system.
Likewise, the paper tape output was used to indicate the combined instrument and
data logger responses to the routine zero and span calibration gas inputs. An
example paper tape output may be found in this report's Appendix.
?0
-------
A 10-digit manual data entry system allowed the labeling of each set of
data as it was collected. This system was used to record the following:
Record
Digits Parameter
Test number designation
Sampling probe height (feet and
• inches)
6, 7, 8 Spare
9 Status of data
0 = Calibration data •
1 = Acceptable test data
2 - Change in test conditions
3 = Questionable data
4 = Ambient background data
5 = Trial test burn
6 = Probe positioned in fire
7,8= Spare :
9 = Disregard data ^
HYDROCARBON SPECIES DATA
The gas chromatography data for hydrocarbon species was recorded by a
Hewlett-Packard Model 3390 Integrator. This device accepted the analog signal
from the gas chromatograph and plotted the peaks which correspond to the
hydrocarbon species. The integrator also determined the retention time for each
peak and the peak areas which are proportional to the hydrocarbon concentration.
This data was recorded by a printer/plotter on a paper tape. The peak area values
recorded on this tape were subsequently reduced to units of parts per million by
volume of methane equivalents.
DOCUMENTATION
The performance of these tests was documented by the following:
. Logbooks maintained by CMA project participants. These records contain a
test chronology, records of field observations, records of flow rates of
gases feeding the flare and preliminary field data records copied from the
data logger paper tape. These logbooks are stored at CMA's headquarters in
Washington, D.C.
• Logbooks maintained by ES test personnel. These records contain a
chronology of all events associated with the flare tests that are related
to thf' analysis of flare plume gases. This recorded data includes records
of Calibrations, zero and span checks, samp liny probe heights, test
j observations, moisture determination data, particulate mass loading data,
difficulties encountered and solutions offered. These logbooks are stored
at the ES Austin, Texas office.
21
-------
• Strip chart records. This includes continuous recordings of CO, CO?, NOX,
THC, SO?, 0;?, probe temperature, wind speed, wind direction, ambient
temperature and the gas chromatqgraph integrator. These records are
maintained at the ES Austin, Texas office.
. Video and audio recordings. These magnetic tapes, include the: audio
recordings of the participants' comments and observations made during each
test through the intercom system. The video tapes ?.lso include a visual
. record of the flare flame during the tests. Copies of these tapes are
stored at the JZ Tulsa, OK facility; the CPA, IERL Office, Research
Triangle Park, North Carolina; and the CMA headquarters in Washington,
D.C. •• = .- ...-.•
• Data tapes. These paper tapes and magnetic computer tape contain all the
validated data logged by the data logger during the tests. The paper tapes
are stored at the ES Austin, Texas office. Copies of magnetic tape are
stored at CMA headquarters, Washington, D.C., and at the ES Austin, Texas
office.
CALCULATIONS ; ; •
The following calculation formulas and constants were employed to reduce
the data presented in this report.
Combustion Efficiency
CO?
% CE =
CO + C02 + THC
.100
Where: CO, CO? and THC are flare plume concentrations (corrected for
background) of these parameters in parts per million by volume. (The
THC term is expressed in terms of.methane equivalents corrected for
background.) Note that soot corrections were made only for Tests 4,
8, and 65 (see pages 43-46).
Gaseous Flows
Ft
Ft
% MR
T00~ . MF
•
ffiw .
V29
V29 "
14.7 .
P+14.7
MR
prro "
T + 460
530
~ "53CT
Where:
Standard Rotameters (SCFM)
r±' Direct Reading Rotameters (SCFH)
F^ - gas flow at time t
% MR = percent of full-scale meter reading for standard flow
MR = meter reading for direct reading rotameters
MF - flowmeter calibration factor (SCFM)
-------
MW - molecular weight of gas in flow meter
P = flowmeter back pressure (PSIG)
T = temperature of gas (°F.)
Flow Meter Calibration Factors
Meter Designation
699 MT
391 MT
R13M-25-3
R10M-25-3
R8M-26-2
Flow Meter Calibration Factor (SCFM)
745.4
409.5
128.8
26.04
2.13
Gas Constants
Gas
Crude Propylene
Nitrogen
Sulfur Dioxide
Density (Ibs/ftlj
0.1106
0.1700
0.0730
Lower Heating
Value (Btu/ft3)
2183
0
0
Molecular
Weight
Steam Orifice Flows
3" orifice maximum flow = 2,250 Ibs/hr
1-1/2" orifice maximum flow =600 Ibs/hr
1/2" orifice maximum flow = 200 Ibs/hr
Ft =
Chart - 20
~80~
Where: F. = steam flow at time t (Ibs/hr)
\r
FMJIX = maximum steam flow through orifice (Ibs/hr)
% Chart = response of recorder coupled with flow transducer
20 = zero offset of recorder
80 - full-scale recorder response
Pg - steam pressure, (psia)
P0 = base pressure, (psia) ,
23
-------
Average Flows
.
~
Where: FA = average flow rate for each test
z = number of flow rate readings during each test
'"t? = flow reading numbered l...z at time t
-------
SECTION 5
REVIEW OF FLARE TEST RESULTS
The test reviews contained in this section are grouped by the experimental
variables of flow rate and Btu content of the relief gas. The test reviews
consist of a narrative description of the test conditions and the results by
test group. The measured combustion efficiency of the flare is the term which
is used in these discussions to evaluate the flare performance.
Statistical data summaries are presented on a test-by-test basis in the
appendix to this report (Appendix B). The data presented in these summaries
were calculated from the instantaneous data values (collected at approximately
12-second intervals) which have been corrected by subtracting background
concentrations. No adjustment was made for moisture. These summaries include
average values, standard deviations, number of observations and combustion
efficiency calculations for each test. The combustion efficiencies for each
test were calculated by two methods: (1) the "average combustion efficiency"
values listed in the summaries are the average values of all of the instan-
taneous combustion efficiency calculations performed on the instantaneous data
values; (2) the "overall combustion efficiency" term was calcuelated from the
average concentration values of CO, C0£ and THC for the entire test. The
differences between these two calculation methods are not regarded as signifi-
cant. The "overall combustion efficiency" term is used in this report for
comparison between tests.
The chronological order in which the tests were performed minimized piping
and equipment changes in the field, and this order is substantially different
from the groupings listed here. The test numbering system is not sequential
since many tests were added or deleted from the planned sequence. Tests
numbered 1 through 33 represent tests which were completed in accordance with
the planned test series. Tests numbered 50 through 67 represent test conditions
which were planned and implemented in the field in place of the deleted tests.
The results for tests numbered 11, 16, 59, 29 and 32 have been divided into
subtests (designated ll(a), ll(b), etc.). These test data were divided into
subtests for data analysis because the flare operating conditions significantly
changed during the tests. The division into subtests allowed the data to more
closely represent steady-state flare operation. All other tests were judged to
represent steady-state flare operation. The criteria for steady-state flare
operation were that all individual flow readings must be within +10% of the
average flow.
STEAM-ASSISTED FLARE TESTS
Twenty-three tests were completed on the John Zink Company (STF-S-8)
steam-assisted flare. The flare operating conditions and the results of these
tests are summarized in Table 4.
25
-------
TABLE 4. STEAM-ASSISTED FLARE SUMMARY
T*St Conditions
RELIEF ens'
Test
(SCFM)
Exit
Velocity
(ft/inn)
Lower
Heating
Value
(Btu/SCF)
Prop>lene Flop
(Ibs/hr)
Hitrogen Flow
(Ibs/hr)
Steam
Flow
(SCFM)(Ibs/hr)
Steam-to-Rplief
Gas Ratio
(Ib/lb)
Combust if>"
Efficiency
Percent
i!
1
2
3
4
8
7
5
67
473
AHA
456
263
157
154
149
148
2,523
2.475
2.432
1.509
837
821
79b
789
2.183
2.183
2.183
2.183
2.183
2.183
2.1B3
2.1E3
3.138
3,078
3.027
1.875
1.044
1.019
991
980
473
464
465
283
157
154
149
148
2,lb9
1,564
1,355
72?
1,513
711
0.688
0.508
0.448
0.757
1.56
0.725
99.96
99.82
99^82
9!..BO"
9E.81"
99.84
99.94
See pg. 28
= *E
xcessive,
Stew Flo
Rates
24.5
24.4
24.5
25.0
24.7
131
130
131
133
132
2,183
2.183
2,183
2.183
2,183
162
162
163
166
164
24.5
24.4
24.5;
25.0
24.7
150
498
652
941
1,125
0.926
3.07
3.45
5.67
6.86
99.84
99.45
99.70
82.18
66.95
OV
57
1Kb)
59
-------
High Btu Content Relief Gases
This test group addresses the steam flare's combustion efficiency while
burning high Btu content relief gases at variable flow rates and various steam-
to-relief gas ratios. Tests numbered 1, 2 and 3 examined the burning of the
crude propylene at the normal (high) rate of approximately 3,100 Ibs/hr with
steam-to-relief gas ratios ranging from 0.688 to 0.448. Test 3 was run at the
steam flow which yielded incipient smoking of the flare. No significant change
in the combustion efficiency values was noted between these three high flow rate
tests. What little hydrocarbons were present were predominately methane.
Likewise, the average corrected CO concentrations for these tests wer:- low,
ranging from 3.8 to 13.8 ppm.
The CO? values reported during the first minute of test 3 are lower than
those which were prevalent during the remainder of the test. The combustion
efficiency data does not appear to be influenced by this unexplained anomoly.
The background data file applied to test 3 has a negative average value for
CO (-0.4 ppm). This was caused by the physical limitations of the CO analyzer.
This instrument was operated on the lowest available range (0-1000 ppm). At
this range, the practical limit for accurately adjusting the analyzer's zero
response was +3.0 ppm. Therefore, it is not surprising that siightly negative
background CO values could be recorded during conditions of low ambient CO
concentrations*
The average corrected total hydrocarbon ,/alue reported for test 1 is -0.7
ppm. This negative value results from thr measured ambient background THC
concentration being higher than the THC corjentration measured above the flare
flame. '
Tests numbered 4 and 8 were perforred with crude propylene relief gas flows
of 1,875 and 1,044 Ibs/hr without an> steam assistance to the flare. These
conditions purposely resulted in a heavily smoking flare condition. The relief
gas flow rates for test number 4 wp.-e reduced from those used in tests 1,2 and
3 in order to keep the flame lerjth within the probe height constraints.
The combustion efficiencies for these two tests are reported as 98.80% and
98.81%. It should be noted fiat these combustion efficiency calculations do not
account for the carbon los* as soot; only carbon present as gaseous species are
considered (CO, COp, and THC). Higher levels of CO were observed during these
tests (61 to 75 ppm corrected) in comparison to tests 1, ? and 3. However
elevated levels of iMburned gaseous hydrocarbons were not detected. Tr,e
hydrocarbon species f4ata shows the predominant species present to be methane and
acetylene.
Particulate samples of the spot were collected during these two tests.
This data is addressed later in this section.
Tests nuMbered 7 and 5 were designed to represent flaring of a hi"h Btu
content gas at an intermediate flow rate. The steam-to-relief gus rat « was
1.56 for t'.-st 5 and represented the high steam flow case, while the ratio of
0.757 for test 7 yielded incipient smoking. Both of these tests at Intermediate
' 27 . . .' . ,
-------
flow rates yielded similar combustion efficiency results to the high flow rate
tests (1, 2 and 3). The observed combustion efficiencies were 99.94% and 99.84%
for tests 5 and 7 respectively. Methane accounted for the major fraction of the
total hydrocarbons present in the flare emissions. The corrected CO levels for
both tests were fairly low at 4.1 ppm for test 5 and 7.9 ppm for test 7.
During test 67 the sampling probe was deliberately placed in the flare
flame. This is in contrast to the other tests which sought to sample in the
flare plume above the flame. The purpose of this short test was to demonstrate
the upscale instrument responses to the partially combusted gases in the flame.
Concentrations of CO and THC were observed to rise sharply and off scale as the
probe was placed in the flame. THC concentrations were observed to be greater
than 100 ppm and CO concentrations were observed to be greater than 2,280 ppm.
The data collected dunng this test does not represent the combustion efficiency
of the flare since the sample was collected within the flame. The average values
for THC and CO reported in the statistical summary are disregarded since these
numbers excluded the overrange observations.
Low Flow Rate. High Btu Relief Gases
Tests numbered 17, 50, 56, 61 and 55 examined the effects of increasing
steam flows on the flaring of a high Btu content relief gas at a low flow rate
(approximately 164 Ibs/hr). Test 17 yielded resilts similar to the high and
intermediate flow rate tests. The overall combustion efficiency was calculated
to be 99.84% and the corrected average concentrations of THC and CO were low at
-0.5 and 6.1 ppm, respectively. (The negative THC value resulted from the
measured concentration being lower than the background concentration.) It was
determined during this test that a steam-to-relief gas ratio of 0.926 was
required for smokeless operation at the designated flow rate.
Tests 50, 56, 61 and 55 were performed at increasing steam flow rates. The
steam-to-relief gas ratios used for these tests are regarded as being higher
than those that would represent good engineering practice. Steam-to-relief gas
ratios for tests 50 and 56 were 3.07 and 3.45 and yielded combustion
efficiencies of 99.45% and 99.70%. By contrast the steam-to-relief gas ratios
for tests 61 and 55 were 5.67 and 6.86 and resulted in lower observed combustion
efficiencies of 82.18% and 68.95$. This data suggests that steam-to-relief gas
ratios above 3.5 may cause inefficient combustion.
The total hydrocarbon and CO concentrations for tests 50 and 56 were fairly
low in keeping with the high observed combustion efficiencies. However, the
hydrocarbon specie data for these two tests show that a larger fraction of the
total hydrocarbon was present as unburned propylene (approximately 1/4 of the
total hydrocarbon for test 56 and 1/2 of the total hydrocarbon tor test 50) in
comparison to the previously discussed tests. In tests 61 and 55, with the lower
observed combustion efficiencies, the CO and THC concentrations were elevated
and propylene represented approximately 3/4 fraction of the total hydrocarbon.
Test ol was a repeat of Test 55. this repeat test was performed because of
uncertainties regarding probe placement during test 55. The flaring of the high
heating value relief gas at a low flow with a very high steam rate yielded a low
luminosity flame that prevented accurate visual placement of the probe.
' . : . 28 ".-;• ' .
-------
Additionally, test 55 was conducted during variable wind conditions. Test 61
was performed at night to aid visual probe positioning and to take advantage of
stable wind conditions. The only significant difference between test 61 and
test 55 was that the steam-to-relief gas ratio for test 61 was somewhat lower
(5.7 versus 6.9). This ratio is still regarded as being very high and not
representative of typical industrial operating practices. The effect of steam
quenching on the flare combustion efficiency is evidenced in the test data.
Low Btu Content Relief Gases
The flaring of low Btu content relief gases was simulated by diluting the
high Btu crude propylene with inert nitrogen. Thus, by changing the relative
flow rates of nitrogen and crude propylene to the flare, the heating value of the
relief gas could be varied. For this series of tests, the Btu content of the
relief gases ranged from 634 Btu/SCF to 192 Btu/SCF, and the relief gas flow
rates ranged from 3,292 Ibs/hr to 803 Ibs/hr.
The original test plan called for the series of tests involving low Btu
content relief gases to include variations in the steam flow to achieve
incipient smoking and smoking conditions. However, for most of these tests
smoking was not observed, even with zero steam flow. Only when the lower heating
value rose above 450 Btu/SCF during test 16 was smoking observed.
Tests numbered 11, 59 and 16 in this series were divided into subtests
because the flare operation was not steady-state during these tests. Due to
physical limitations in the nitrogen flow control system, the flow of nitrogen
decreased with respect to time causing a corresponding increase in the Tower
heating value of the relief gas. The division into subtests allowed the data to
more closely represent steady-state flare operation.
: m .
Test 57 represented the highest flow rate of a low Btu content gas that was
tested. The flare was supplied with 3292 Ibs/hr of relief gas with a lower
heating value of 294 Btu/SCF and a steam-to-relief gas ratio of 0.150 steam/lb
relief gas. Test 51, by comparison, represented flaring of a similar heating
vaiue gas (309 Btu/SCF) with a similar steam-to-relief gas ratio (0.168), but at
a lower flow rate of 1,527 Ibs/hr. Tests 57 and 51 achieved combustion
efficiencies of 99.90% and • 98.66%, respectively. Corrected hydrocarbon
concentrations of 2.0 ppm and 11.5 ppm and CO concentrations of 5.0 ppm and 34.1
ppm were obtained for tests 57 and 51, respectively. The slightly lower
combustion efticiency of test 51 is also observed in the hydrocarbon species
data. The observed hydrocarbons in test 57 were approximately 20% non-methane
species, while the hydrocarbons in test 51 were comprised of 58% non-methane
species.
TI rlarnes for tests 57 and 51 were of low luminosity and visual
positioning of the probe was difficult. These two tests were the only low Btu
flare tests where steam was supplied to the flare. The background file as
applied to test 51 (and tests 23 and 52) lists probe tip temperatures that are
higher than ambient levels. This is believed to be caused by the probe acting
as a heat reservoir from the test event that immediately preceded. This anomaly
does not effect the combustion efficiency data.
29
-------
Tests 11, 59, 60 and 16 examined the flaring of relief gases with heating
values of 19? Btu/SCF to 634 Btu/SCF at flow rates ranging from 3,101 Ibs/hr to
803 Ibs/hr with zero steam flow to the f'are. The variations in observed
combustion efficiencies for this set of tests was fairly narrow, ranging from
99.93* to 98.11%.
Test 59 demonstrated the flaring of a low Btu content gas at a high flow
rate with no steam. The nitrogen flow decreased during this test from 2,453
Ibs/hr to 1,726 Ibs/hr due to declining pressure in the nitrogen cylinders.
This resulted in an increase in the. Btu content of the relief gas from 182
Btu/SCF to 257 Btu/SCF from the beginning to the end of the test. This
corresponds to a slightly lower combustion efficiency for test 59(a) than for
test 59(b). .
Tests 59(a) and 59(b) had the lowest Btu content relief gases of the group.
Likewise, these tests exhibited slightly lower combustion efficiencies. This
observation is confirmed in the hydrocarbon species data which shows test 59 to
have elevated total hydrocarbon concentrations (as compared with tests 11, 60
and 16), and non-methane hydrocarbons representing 92% of the total. These
results indicate that some unburned hydrocarbons :were sampled during this test.
Test 11 was to demonstrate the flaring of low heating value gas at a flow
rate of approximately 3,100 Ibs/hr. No steam was supplied to the flare. The
flow rate of nitrogen to the flare declined somewhat during the test, thus,
Musing a corresponding increase in the heating value of the flare gas.
rnerefore, this test has been divided into three subtests [ll(a), ll(b) and
ll(c)] for purposes of data analysis. The data does not indicate any change in
the flare combustion efficiency with the change in nitrogen flow.
Test 16 was designed to be a smoking flare test utilizing an intermediate
flare gas flow with a low heating value gas. No steam was supplied to the flare.
As was the case with Test 11, the nitrogen flow declined during the test and
hence, the test was divided into subtests for aata analysis [tests 16(a), 16(b),
During the initial period of the test, when the heat content of the flare
gas was the lowest, the flare did not emit smoke. However, as the nitrogen flow
declined and the heat content of the flare gas increased, the flare began to
smoke. The smoking began approximately nine and cne-half minutes from the start
of the test [during subtest 16(b)] when the heating value of the flare gas
reached approximately 450 Btu/SCF. The smoking increased with increasing Btu
content of the relief gas. The onset of smoking and the change in heating value
did not have any obvious effects on the gaseous combustion efficiency data (if
carbon lost as smoke is excluded from tne combustion efficiency calculations).
Test 60 was similar to Test 16( a) except the Btu content was slightly lower
at 298 Btu/SCF instead of 339 Btu/SCF. The flow rates for the two tests were
similar with exit velocities of 1781 and 1707 ft/min. The observed combustion
efficiency for test 60 was 98.92% as compared with 99.74 for test 16(a). As was
the case for test 59, this slightly lower combustion efficiency is believed to
be a result of the lower Btu content of the relief gas.
30
-------
Purge Rate Relief Gas Flows
Tests 54, 23, 52 and 53 examined purge gas flare operations. Purge flows
are sometimes used in flare operations to prevent oxygen encroachment into the
flare system during the time that no relief gas is provided the flare. It should
be noted for these tests that the flow of natural gas from the flare pilots was
significantly greater than the flow of the purge gases. The two pilots burned
a total of 10 SCFM (9210 Btu/min) of natural gas as compared with purge flows of
0.56 to 0.36 SCFM (149 to 74 Btu/min). Thus, the overall combustion efficiency
measurements for these tests were primarily a measure of the flare pilots.
During these tests only an occasional flicker of flame could be observed at the
flare header.
Tests 54 and 23 were performed without the addition of steam to the flare.
Thest tests yielded high observed overall combustion efficiencies of 99.90% and
ilOO.01%. The calculated combustion efficiency greater than 100% for test 23
resulted from the observed hydrocarbon level above the flare being slightly
lower than the measured ambient background hydrocarbon concentrations. The
corrected total hydrocarbon concentration for tests 54 and 23 were 0.0 and -5.0
'
.Tests 52 and 53 were similar to tests 54 and 23 except 210 Ibs/hr steam was
supplied for the former. The calculated combustion efficiencies for tests 52
and 53 were 98.82% and 99.40%. This slight decline in the combustion efficiency
is believed to be due to steam quenching of the combustion process. Corrected
total hydrocarbon values observed for tests 52 and 53 are 15.2 and 10.9 ppm.
Correspondingly, the CO concentrations for tests 54 and 23 were lower than for
tests 52 and 53 (6.8 and 4.5 ppm versus 16.0 and 23.9 ppm). Likewise, non-
methane species represented a larger percentage of the total hydrocarbon for
tests 52 and 53 than for tests 54 and 23,
The probe tip temperatures during the first three and one-half minutes of
test 53 were not recorded by the data logger. This temperature data was
recovered from the strip chart record. The low, steady wind speeds that
prevailed during test 54 allowed the collection of twenty minutes of relatively
consistent data. However, during the latter part of the test the wind speed was
observed to increase with a corresponding decrease in probe tip temperature, C02
concentration, and S02 concentration.
AIR-ASSISTED FLARE TESTS
Eleven tests were completed on the John Zink Company STF-LH-457-5 air-
assisted flare. The flare operating conditions and results are summarized in
Table 5.
High Btu Content Relief Gases
Four tests numbered 26, 65, 28 and 31 were conducted on undiluted crude
propylene burned in the air-assisted flare. The flow of relief gas for these
tests ranged from 3,196 Ibs/hr to 150.8 Ibs/hr. All these tests achieved
observed combustion efficiencies greater than 99. OX.
''•""• 31 ' . ' .
-------
TABLE 5. AIR-ASSISTED FLARE SUMMARY
Test
Test Conditions Hunter
c
w
**
s0
3
00
«J
_ c
5 u
o»
o>
c
— » o»
5 ° **
L
u
&
0>
S82
J-Hcr
. '
*JM
O.U.OC
26
65
28
31
66
29(a)
29(b)
64
62
63
33
32(a)
32(b)
Flow
(SCFM)
481.6
159
157
22.7
639
510
392
249
217
121
0.714
0.556
0.537
Velocity
(ft/mln)
13087
4320
4266
617
8192
6538
5025
3192
2782
1551
9.1
7.1
6.9
Heating
Value
(Btu/SCF)
2183
2183
2183
2183
158
168
146
282
153
289
83
294
228
RELIEF GAS*
Propvlene
(Ibs/nr)
3196
1056
1043
151.8
308
261.9
173
214
101
106
0,181
0.498
0.374
Flow
(StFM)
481.6
159
157
22.7
46.4
39.3
26.2
32.2
15.3
16
0.0272
0.0750
0.0563
Nitrogen
(Ibs/hr)
-
*
2598
2062
1602
949
884
464
3.01
2.10
2.10
Flow
(SCFMJ
-
593
471
366
217
202 '
105
0.687
0.481
0.481
Air Flow
High
Off
High
Low
Off
Low
Low
Low
Low
Low
Low
Low
Low
Combustion
Efficiency
99. »7
99.57**
99.94
99.17
61.94
55.14
65.65
99.74
94.18
99.37
98.24
98.91
96,86
* A!) values at standard conditions of 70*F and 29.92 in Hg.
** Itot accounting for carbon present as soot (see Table 10).
-------
The hydrocarbon species data for the higher flow rate tests 26, 65 and 28
show the bulk of the total hydrocarbon present as methane. Test 31 integrated
hydrocarbon species data shows only 14% of the total hydrocarbon present as
methane. Correspondingly test 31 has the lowest flow rate and combustion
efficiency of the group. The data collected during test 28 exhibits more
variation than usual due to the unstable wind conditions that were present.
. One of the ambient background files that is applied to this data (file 32)
shows slightly higher concentrations of CO and CO? and lower concentrations of
THC during the first minute of data than are prevalent during the majority of the
background period. The probable explanation for this is that the probe
temporarily was in the plume of another combustion source in the area. This
abberation does not significantly effect the test results.
Test 65 represents the combustion of a high Btu content hydrocarbon at an
intermediate flow rate and no air assistance. This test essentially represents
a repeat of test 28 without the air blower switched on. During the test, the
flame was observed to smoke.
Low Btu Conjent jelief Cases .
Five tests were performed on low Btu content relief gases with the air-
assisted flare. The relief gas flows for these tests ranged from 2,906 to 570
Ibs/hr and the lower heating values varied from 146 to 289 Btu/SCF.
Tests 66, 29 and 62 of this group yielded the lowest combustion effi-
ciencies observed for the air-assisted flare tests. Correspondingly, these
tests involved the lowest Btu content relief gases (146 to 158 Btu/SCF) that
were tested on the air-assisted flare. The flare flames for these tests were of
low luminosity and were observed to be detached from the flare tip. This
detached flame condition is not regarded as good engineering practice.
Predictably, the major portions of the unburned hydrocarbons present in the
flare plume were in the form of propane and propylene. Likewise, elevated CO
concentrations were observed during tests 66, 29 and 62.
' . ' • • '
In contrast to the above low efficiency tests, the air-assisted flaring of
282 and 289 Btu/SCF relief gases during tests 64 and 63 proved to be much more
efficient. These higher Btu content relief gases were flared at lower flow
rates (1,163 and 590 Ibs/hr) than the previously discussed tests and yielded
good combustion efficiencies of 99.74% and 99.37%. Methane comprised 61% of the
total hydrocarbon for test 64 and only 29% of the total hydrocarbon for the less
efficient test,63.
The CO? data from test 63 shows a three minute period in the middle of the
test with CO^ concentrations observed near ambient levels. This is believed to
have been caused by the flare plume shifting away from thi? sampling probe due to
a wind shift. This is evidenced by shifts in wind speed and direction and a
decline in probe tip temperature that corresponds to the decline in CO?
concentrations. This shift in CO? concentrations caused a corresponding
decline in the combustion efficiency data. Therefore, the average combustion
effir<«ncy data presented for this test is regarded as conservative.
33
-------
The first thirteen minutes of data collected during test 29 was designated
as test 66. The difference between these two tests was that the air-assisted
flare's axial fan was turned off for test 66 and turned on for test 29. Both the
propylene and the nitrogen flows were observed to decrease during test 29, thus
resulting in unsteady flare operation. Therefore, the test was divided into two
subtests [29(a) 29(b)J in an effort to make the data within each subtest more
closely approximate steady-state flare operation.
Purge Rate Relief Gas Flows
Tests 33 and 32 evaluated the performance of the air-aisisted flare in
burning purge rate flows of low Btu content gases. As was the case for the steam
flare purge gas tests, the overall efficiency of the purge gas combustion is
masked by the flare pilots.
The purge flows for tests' 33 and 32 are ranged from 0.714 SCFM to 0.537 SCFM
as compared with the 10 SGFM flow of natural gas from the pilots. The lower
heating values of the purge gases for these tests ranged from 83 Btu/SCF to 294
Btu/SCF. The observed combustion efficiencies for these tests were 98.2458 for
test 33 and 98.87% for test 32. These values are slightly lower than those
observed for the steam-assisted flare purge gas tests. However, the majority of
hydrocarbon measured in the flare plume 'was found to be methane, thus,
suggesting that incomplete combustion of the natural gas.from the flare pilots
may have caused the lower combustion efficiencies.
The flow of crude propylene to the flare did not remain constant throughout
test 32. Hence, the test data was divided into two subtests [32(a) and 32(b)J
appropriate for data analysis.
SENSITJVITY OF COMBUSTION EFFICIENCY TO PROBE HEIGHT
During the course of the test series the position of the flare sampling
probe was frequently adjusted to keep the probe tip as near as possible to the
middle of the flare plume and as close to the flame as possible without being in
the flame. These changes were necessary to compensate for changes in the wind
that occurred during the tests and resulted in changes in the flare flame
pattern and location. Not infrequently, the probe was situated at several
different locations and heights during a test.
The vertical position of the probe did not have a definable effect on the
combustion efficiency data. Figures 6 and 7 are graphs of combustion efficiency
versus probe height that demonstrate-the insensitivity of the vertical probe
position to the combustion efficiency measured at the probe tip.
EFFECT OF STEAM-TO-RELIEF GAS RATIO
ON COMBUSTION EFFICIENCY
Steam injection is a techrique commonly used in flare operations to enhance
the combustion process. The steam-assisted flare tests performed in this
project included a wide range of steam flows and steam-to-relief gas ratios.
. 34 .".':'
-------
3J
32
31
30
o
or
a.
6 28
£
27
26
ii.tt) 95.7f' -95.80 99.90 100.00 I'OO.Ip 100.20 100.30 100.40
COMBUITION
figure 6. Sensitivity of combustion efficiency to probe height.
Test 28
35
-------
u
< U2
m
o
cc
a
u
UJ
X
39
38
37
95.80
99.90
COMBUSTION
100.00
100.10
Figure 7. Sensitivity of combustion efficiency to probe height.
Test 57
36
-------
Figure 8 is a graph of the ?ffeet of steam-to-relief gas ratios on the
measured combustion efficiencies of high Btu content relief gases. This plot
shows general tendencies for combustion efficiencies to decline at higher or
lower than normal steam flows. This data suggests that steam-to-relief gas
ratios ranging from 0.4 to 1.5 yield the best combustion efficiencies. The
smoking flare tests at zero steam flow were observed to have slightly lower
combustion efficiencies than the other comparable tests at normal steam flows.
Presumably this is due to the lack of steam-induced turbulence and reaction in
the combustion process. It should be noted that these combustion efficiency
values do not account for carbon lost as smoke.
The steam flows during the low flow rate tests were at too low a velocity
to promote good combustion. Likewise, because of the low relief gas flows the
steam to hydrocarbon ratios were greater than for the higher flow rate tests. In
the case of tests 61 and 55, the excessive steam-to-relief gas ratios are
believed to have caused steam quenching of the flame.
FLARE NOX EMISSIONS
Emissions of NOX from both steam- and air- ^sisted flare plumes were
measured during this test program. The NOX conrsiv- jns observed during these
tests were fairly low in comparison to other type--, of combustion sources.
However, the NOX concentrations were subject to undefined dilutions of ambient
air and steam not normally encountered in other sources. Corrected NO
concentrations ranged from 0.50 to 8.16 ppm.
'X
The NOX mass emission rates were estimated from the NOX and CO? data
suggested by EPA:
Ef o = Mo1es NOy . 46 Ibs/mole N0y . 132 Ibs CO? produced
'* Moles CO? 44 Ibs/mole C02 42 Ibs propylene burned
. 47.2 Ibs propylene burned
~ 106 Btu
- "5-0 • )b NIO Btu
Where:
Moles NO* PPM NO* Measured
Mole CO? " PPM CO? Measured
Assumptions: .
1. 100 % combustion of propylene (fuel assumed to be 1005K propylene);
2. Equal dilution of NUX and COp between flare plume and sampling probe;
3. Neglect Btu content of flare pilots (612,600 Btu/hr, gross);*
4. 47.2 lbs/106 Btu higher heating value for propylene.
* For purge tests, this assumption Is Invalid.
37
-------
100
99
98
97
96
95
9<-
93
92
91
90
89
88
87
86
5 85
3 gi,
o
E 83
UJ
5 BJ
2 m
1 ou
79
78
77
76
it
71*
71
7?
it
70
68
67
0
CD^O )
KEY
• - SMOKING flMr
• • HIGH riOW HAT
X • LOW now RAH
rrsrs
! TfSTS
< X
V
x
— r
) I * S
STfAM TO MLIir MS «ATIO(lb/lh)
Figure 8. Effect of steam-to-rel1ef gas ratios on flare combustion efficiency.
(High Btu content relief gases)
38
-------
Table 6 summarizes the NOX results of these calculations. This treatment
of the NOX data yields NOX emission rates ranging from 0..0.18 to 0.208 lbs/10&
Btu. Examining this data shows no clear patterns of high or low emissions
between test groups. One possible exception to this is the high Btu content air-
assisted flare tests which yielded the highest calculated NOX emission rates.
HYDROCARBON ANALYSES
Hydrocarbon analyses were performed both by continuous total hydrocarbon
monitor and by gas chromatograph for hydrocarbon species. The samples for the
gas chromatograph were taken from the heated sample manifold (wet basis) and
either directly injected into the instrument (an instantaneous sample) or
collected in a Tedlar® bag over a period of time (integrated bag sample), and
subsequently, analyzed by the same gas chromatograph. The continuous hydro-
carbon analyzer withdrew its sample from an unheated sample manifold (dry basis)
and measured total hydrocarbon (THG) directly. Both the chromatograph and the
continuous hydrocarbon analyzer utilized flame ionization detectors. Thus,
three sets of hydrocarbon data are available for each test.
Tables 7 and 8 prfsent a summary of the hydrocarbon data collected during
the steam- and air-assisted flare tests. All three sets of hydrocarbon data
show good agreement between their total hydrocarbon values for those tests with
lower THC concentrations (high combustion efficiency tests). In addition, the
instantaneous anu bag sample values show good agreement (considering the
different sampling techniques) throughout the range of values. However, some
discrepancies are noted between the continuous THC values and the gas
Chromatograph THC analyses at the higher concentrations encountered during the
lower combustion efficiency tests. These discrepancies at higher THC concen-
trations are believed due primarily to the absorption of unburned propylene in
the cold trap associated with the dry basis sampling system utilized by the
continuous THC analyzer. It is believed that the propylene was subject to loss
by virtue of its solubility in the water in the cold trap. This may have been
the situation despite the precaution of using a minimum-contact design cold trap
condenser.
The sample concentrating effect of the cold trap i? believed to be
negligible due to the low moisture content of the gaseous samples. Variations
between the response characteristics of the gas chromatograph1? and the
continuous THC analyzer's detectors are not thought to be significant. Both
instruments were calibrated in terms of parts per million by volume of methane
equivalents.
The continuous total hydrocarbon analyzer's data is believed to be the most
useful for evaluating the higher combustion efficiency tests where methane was
the major fraction of the total hydrocarbon. However, in the case of the lower
combustion efficiency tests where water soluble propylene could have be^n lost
in the continuous analyzer's sampling system, the integrated bag samples
provide the most representative total hydrocarbon data. Likewise, since the
reported combustion efficiency values were based on the continuous total
hydrocarbon data, these values may be biased high for the lower combustion
efficiency tests due to thp potential loss of propylene In the sampling system.
39
-------
TABLE 6. FLARE NO* RESULTS
Test
No.
0>
L.
<0
U_
•o
o»
*J
*n
*A
I/)
§
o>
*».
i/>
QJ
U
«
u.
1
2
1/t
<
1
L»
•^
< . •
*J
=
V
*•»
§
%••*
3
*•*
O3
f
X
•M
C
OJ
4-»
§
u
3
4J
co
E
_j
a w
•^J ~
CD a
.wl
£ C
' Oi ' O
.£•3
«. *J
£ £
CO ^j
_ C
|«s
^J
1
2
3
4
8
7
5
67
17
50
56
61
55
57
11
59
, 60
- 51
16
54
23
52
53.
26
65
28
31
66
29
64
62
63
33
32
•Wii*
Concentration
(PPMy)
3.09
2.16 ;
1.54
1.96
1.45
1.62
2.09
3.77
1.00
0.50
0.58
1.32
0.38
2.68
3.69
1.41
0.99
0.57
1.87
5.00
5.90
0.68
2.83
5.34
2.40
8.16
4.02
0.97
1.06
1.24
0.60
1.57
0.74
1.75
C02*
Concentration
(PPMV)
7,052
4.719
2,496
6,61'
5,400
5.224
7,052
. H/A
3.499
4,220
3,120
6,273
2,012
6,945
5.269
5,413
3,685
3,347
4.059
7,115
8,465
2,622
5,741
6,270
4,878
6.076
4,568
2,432
2,179
3,282
3,076
4,164
1,857
3,702
NOX
Mass Emission
(Ibs/lO* BTU)
0.068
0.071
0.095
0.046
•0.042
0.048
0.046
H/A
0.044
0.018
0.029
0.033
0.029
0.060
0.108
0.040
0.042
0.026
0.071
0.109
0.108
0.040
0.076
0.132
0.076
0.208
0.136
0.062
0.075
0.058
0.030
0.058
0.061
0.073
• Corrected for background.
-------
TABLE 7. HYDROCARBON ANALYSIS SUMMARY
STEAM-ASSISTED FLARE TESTS
Test
NO.
1
2
3
c 4
•V
1 ;
£ 5
•g. 17
* 50
56
61
55
57
n
I»
Jsi
£ U
§ **
""•23'.
5?
53
Continuous THC
Analyzer
»*g. Cone.*
3.7
6.5
4.3
10.5
9.5
9.9
4.9
4.7
1Z.O
6.2
964
742
7.8
6.6
36.1
14.3
" (.-. Z2.8
6.4
6.5
6.3
?4.6
15.6
Instantaneous Samples, Average Values*
t\ Cj/C?* C2* £3 C3= C< THC
2.2 - - - - - 2.2"
3.0 - - - - - 3.0**
2.5 . - ... - J.5"
4.73 0.68 2.51 0.25 ?.66 - 10.8
5.70 0.63 3.00 0.15 1.8? - 11.3
1.79 O.?0 0.30 0.1? 0.35 - 2.8
2.4 - - - - . - 2.4**
2.59 0.25 - 0.09 - - 2.9
3.59 1.08 0.49 1.34 5.65 - 12.1
4.06 0.12 0.10 0.19 1.52 - 6.0
45.6 143 69.2 635 2,675 - 3.568
14.07 3.35 0.20 268 1,075 - 1,36)
6.72 0.37 0.03 0.29 O.SO - 7.9
4.29 0.24 0.52 0.31 0.71 - 6.1
4.00 2.96 1.05 3.30 9.63 - 20.9
3.75 1.28 0.55 1.47 6.22 - 13.3
5.82 0.56 0.53 0.50 1.78 - 9.2
6.06 0.49 0.18 0.64 2.16 - 9.5
3.15 0.64 - 1.76 2.42 - 8.0
3.47 0.02 - 0.06 - - 3.6
7.66 1.74 0.03 5.21 21.24 - 35.9
3.39 0.90 0.31 2.04 6.10 - 12.7
Bag Sanoles, Average Values*
Cj Cy/Cp* C?s C3 C3" tt THC
3.0 - 1 - - - - 3.0**
2.8 - ... - 2.8**
3.5 • -' - •-'•'-• .: : 3.5**
6.19 0.76 2.95 0.34 0.77 - 11.0
5.87 0.63 1.46 0.18 0.34 - 8.5
2.9? - G.22 0.23 0.56 - - - 3.9
3.08 - 0.32 0.30 0.29 0.03 4.0
3.12 0.59 0.35 0.15 0.06 - 4.3
4.31 1.90 0.95 2.2) 7.65 - 12.6
4.91 1.46 1.06 0.88 2.22 - 10.5
24.2 67.4 33.5 282 1,182 1.08 1603
16.5 36.6 21.0 259 1,061 - 1402
7.54 0.53 0.16 0.36 O.B2 - 9.4
5.70 1.33 0.35 0.3? 0.79 0.02 8.5
4.31 5.17 1.7? 13.0 31.4. - 55.6
4.45 3.05 1.31 2.63 14.? 0.08 J5.6
8.19 3.73 3.83 1.38 5.92 - 23.1
6.55 0.48 0.25 0.55 1.15 - 9.0
3.46 0.64 0.33 0.86 1.57 0.02 6,9
3.95 0.05 - 0.09 0.11 - <.2
6.65 2.49 4.23 6.70 27.8 - 47.9
3.97 1.53 0.55 3.22 13.6 - 2?. 9
* ATI ir*lu«s are pp* by volume methane tquivalents; untc-rected for background THC.
••The gK clwo«atogr«ph electrometer attenuation was se' to a less sensitive scale for these analyses.
K£T: Kcthjne. C|
Cthane, Cj Propjlene, (.3* Total Hydrocarbon, THC
Ethylene, Cj*
Fropane, C3
Propjlene, (.3*
Acetylene, Cj*
-------
TABLE b; HYDROCARBON ANALYSIS C'JMMARY
A)S-ASSiS7EO FLARE TESTS
Test
No.
i-»
S £ 65
»J28
31
^ 66/29
£ 64
a 63
«»
CO -
5 »
Continuous TKC
Analyzer
Avg. Cone.*
11.3
4.8
6.0
15.7
1.238
8.7
109
15.3
32.1
34.1
Instantaneous Samples. Average
Cl
3.36
4.63
4.15
4.85
25.5
7.59
13.1
5.97
25.1
15.6
C2/C2*
0.86
0.07
0.19
4.93
69.1
C.54
14.7
1.57
2.71
1.71
C2= C3 C3S
0.15 0.67 2.02
0.18 0.07 0.06
0.03 0.18 0.54
1.20 3.62 29.0
27.4 513 1,992
0.06 0.36 0.82
5.64 32.8 249
0.78 1.67 6.45
0.10 1.44 3.38
0.28 0.93 2.19
values*
C4 THC
7.06
5.0
- '5-1
-
-------
Figure 9 is an example of a gas chromatograph analysis of the flare
emissions.
PARTICULATE MATERIAL ANALYSES
Samples of theparticipate material emitted from the flare flame were
collected during the flare test series. An in-line fiberglass filter collected
these samples for determination of particulate mass loading and'subsequent
analysis for polynuclear aromatic compounds (PNA's) by gas chromatography/mass
spectroscopy. The PNA data is reported in Appendix D.
Table 9 is a summary of the mass particulate concentration data collected
during the test series. The data show distinct differences between particulate
loadings of nonsmoking and smoking flare tests.
The combustion efficiency calculations used in this report as a measure of
the flares' performance did not account for the carbon lost as particulate
material in the smoke. Only terms for CO, COz^and THC concentrations are used
in these combustion efficiency calculations. Therefore, the gaseous combustion
efficiency values reported for the smoking flare tests would be expected to be
higher than the real combustion efficiency.
The following equation was used to include the carbon lost as particulate
material for smoking flare tests 4, 8 and 65.
CE% = C0? „ x 100
C02 + CO + THC + Cp
Where:
CO? = carbon dioxide concentration (PPMV)
CO = carbon monoxide concentration (PPMV)
THC = total hydrocarbon concentration (PPMV as methane)
Cp = particulate concentration (PPMV assuming smoke particulate as gaseous
elemental carbon and ideal gas, 2.03 1/g).
Table 10 outlines the results of calculating the combustion efficiency
using this particulate'corrected method.
DILUTION RATIO AND DESTRUCTION EFFICIENCY DETERMINATIONS
The attempt to use sulfur as a tracer material for the flare tests yielded
disappointing results. The tracer technique was employed in Heu of measuring
the volumetric flow rate of the flare plume. Volumetric flow rate determina-
tions in an open combustion system such as a flare are not feasible. The intent
was to complete a sulfur balance between the flare burner and the samp 1 ing probe
in order to calculate the effective dilution of the flare gas due to combustion,
steam, forced air and ambient air. By knowing the dilution ratio, estimates of
flare destruction efficiency and emission rates can be calculated. Unfortu-
. . . ' ' 43 . - - '• .
-------
1.62 ETHANE
ACETYLENE
?.<*'. PROPANE
1.23 METHANE
2.89 PROPYLENE
STOP-
I 194
TD 26128 68
ESTD
RT
1.2S
1.6?:
1,91
? .44
2.851
3,5£
JUN/2 1/8.2- 14-11 88
ise
AREA TW- CAL.
-PB
PV
W
vv
VP
PH
PV
46859
.23?8i
87374
AMOUNT
e.eee
3.581
.6.988
« eee
e.eee
8 '.886
6.688
TOTAL AREA-
HUL FrtfTOR- 1.8e88E-»98
Figure 9. Example gas chromatogram hydrocarbon analysis.
(Test 50)
-------
TABLE 9. PARTICIPATE ANALYSIS
Test No.
2, 3, 1, 5, 7
7, 17, 50, 51
23, 52, 53, 54
4 (Smoking)
8 (Smoking)
55, 56, 11, 57
16, 59, 60, 61
28, 31, 26, 29
33, 32, 62, 63
64, (80. 81, 82
83, 84)2
65 (Smoking)
Filter No,
A-l
F-l
F-2
F-3
F-4
F-5
Wt. gain,
(grams)
0.0063
0.0071
0.0810
0.0819
0.0179
0.0183
Sample Time *
(minutes)
_ —
16
25
—
25
Sample Rate Concentration
(&/min) (jjg/A)
18.51
18.51
18.51 274
18.51 177
18.51
18.51 40
Includes time for,probe positioning while probe was in the plume but
before formal initiation of tfie test.
2- Separate tests conducted for John Zink Company not reported herein.
-------
TABLE 10. SMOKING FLARE COMBUSTION EFFICIENCIES
Combustion
Participate Efficiency
Concentration (percent)
Test
4
8
65
(yg/i)
274
177
40
(pl/1) Particul ate Corrected
557 91.21
360 92.72
81 97.95
Reported
99.80
98.81
99.57
-------
nately, considerable difficulty was encountered in quantifying and maintaining
constant levels of sulfur in the relief gas.
The primary source of flare gas for this series of tests was a tank truck
of 6,800 gallons of crude propylene to which approximately 4 liters of butyl
mercaptan were added. This addition of mercaptan should have yielded a sulfur
concentration of approximately 89 ppm (by weight) in addition to the naturally
occurring sulfur in propylene (estimated at 2 - 10 ppm). An extremely wide range
of results were obtained from these analyses. Thus, there is considerable doubt
as to the actual concentration of sulfur in the propylene. Additionally, this
data cast doubt on whether or not the concentration of sulfur was stable or if
a fraction of the sulfur could have combined with the steel tank, iron pipe or
other materials which were in contact with the propylene.
The sulfur content of the nitrogen used for dilution of the propylene and
the sulfur content of the r-team condensate were found to be very low and did not
represent a significant contribution to the overall sulfur balance.
The levels of S02 measured during the initial tests were lower than
expected, and in some cases, challenged the practical detection limit-of the
analyzer. After completion of ten of the thirty-four tests, provisions were
made to add S02 gas into the relief gas stream in order to elevate the sulfur
levels in the flare emission into a midrange level. This modification was
plagued with mechanical falures that prevented the addition of SO? to many of
the subsequent tests and was not accomplished at a uniform rate.
The sulfur balance data yielded results that were much higher than would
reasonably be expected to occur. In addition, the ratio of SO? to CO?
concentrations in the flare emissions were not constant as was expected, either
between tests or within tests. The SO? and CO? ratio should be constant given
that both these gases are subject to the same dilution effects and assuming a
constant level of sulfur in the relief gas and no effect due to background
concentrations.
Because of the doubts surrounding the sulfur balance dilution ratios, an
alternate means of calculating dilution ratios was formulated using the CO?
data. This technique assumes stoichiometric combustion and is frequently used
in evaluating envisions from other combustion sources. However, the CO? dilution
technique was not well received because: (1) it did not provide a measure of
dilution independent of the combustion measurement, and (2) the flare plume did
not maintain steady-state conditions relative to the sampling probe.
By making simplifying assumptions, destruction efficiency calculations can
be made and are provided in Appendix C. With regard to flare efficiencies, it
should be noted that in a smokeless flare, the total hydrocarbon destruction
efficiency will always be greater than the combustion efficiency. This is
because, by definition, the percentage of hydrocarbon that is converted to CO?
(combustion efficiency) is equal to or less than the percentage of relief gas
that is converted to CO and CO? (destruction efficiency).
-------
MOISTURE DETERMINATIONS
The moisture content of the flare emissions was gravimetrically determined
using an ice-bath condenser type of moisture trap. Table 11 lists the results
of these moisture analyses. The moisture data were lower than expected with an
average of 3.0% moisture measured for the air-assisted flares and 3.8% moisture
measured in the steam-assisted flare samples. A description of the methodology
used for moisture determinations may be found page 18 of this report.
OTHER FLARE TEST ANALYSES
A composite sample of the steam condensate was collected and analyzed for
sulfur and hydrocarbon content. The results of these analyses are as follows:
Sulfate less than 1 mg/1
Sulfide 0.03 mg/1
Hydrocarbon 0.13 mg/1
The crude prqpylene that was used during the test series as the basis for
flare fuel was analyzed by both the propylene vendor and by ES. The results of
these analyses are as follows:
Analyst Propylene Propane Ethane/Ethylene
Vendor 80.2 19.8
ES 79.4 18.5 1.9%
For the purposes of this test series, the differences between these two analyses
is not significant. .
The nitrogen used to dilute the crude propylene to a lower heating value
gas was analyzed for hydrocarbon content. The nitrogen was found to contain
0.33 ppm by volume of hydrocarbon methane equivalents.
4B
-------
TABLE 11. FLARE EFFICIENCY TES1
MOISTURE CONTENT OF SAMPLES
(EPA METHOD 4)
Test Number
7
3
]
5
7
1 1
50
51
23
b?
53
54
4
g
55
56
1 1
57
16
59
60
61
?8
31
26
66
?9
33
32
62
6^
63
65
Mci sture
(Volume %}
<1
N/A
7. 1
-------
SECTION 6
QUALITY ASSURANCE AND QUALITY CONTROL ACTIVITIES
MULTIPOINT CALIBRATIONS . '
Before the collection of data was initiated, the continuous analyzers were
checked for proper operation. A key element of the operational checks was the
multipoint calibrations. These multipoint calibrations consisted of chal-
lenuing the analyzers with a zero gas and several upscale concentrations of the
gaseous compounds of interest. Calibrations were conducted at the concentra-
tions anticipated to be in the flare emission plume. The purpose of these
multipoint calibrations was to demonstrate instrument response linearity both
within instrument ranges and between instrument ranges. Table 12 lists the
analyzer ranges and gas concentration ranges used for these checks. The
criteria for calibration curve acceptability was no more than 5% deviation from
the input value within a range and no more than a 10% deviation from the
reference standard between ranges.
The source of the calibration gases was aluminum compressed gas cylinders
certified by their manufacturers to be accurate within + 2% of their listed
concentrations. The gas standards for NOX and CO were traceable to'the National
Bureau of Standards' standard reference materials. .The compressed gas standard
for S02 calibrations was contained in a Teflon® lined aluminum cyl inder and was
certified on-site against a graviametrial ly calibrated S02 permeation device.
In addition, the efficiency of the N02 to NO converter in the NOX analyzer
was checked using the procedure recommended in EPA Reference Method 20 (40 CFR
60 Appendix A). The converter efficiency was found to be greater than 99%
efficiency in the conversion of N0£ to NO.
The response of the probe tip thermocouple was verified at ambient
temperature against the calibrated meteorological station thermistor and was
checked for 0°C response in an ice water bath.
ZERO AND SPAN CHECKS
Before and after each test or series of tests, all the continuous analyzers
were challenged with zero gas and a single upscale concentration of the compound
of interest. The purpose of these calibration checks was two-fold. First the
col lection 'Of the zero/span data provided a data base to allow estimation of the
analyzer's precision of measurement. Secondly, the gas standards used for the
zero and span checks provided points of reference for instrument calibration
adjustment between tests. Thus, the zero/span checks provided a mechanism to
both monitor and maintain the precision of the d'atii.
Table 13 is a summary of the zero span check data.
50
-------
TABLE 12. MULTIPOINT CALIBRATION CHECKS
Analyzer Instrument Ranges Checked Calibration Gas Levels V
CO
CO?
THC4
NOx
02
S02
0-5,000 ppm
0-3,000 ppm
0-1,000 ppm2
0-5%2
0-10%
0-15%
0- 100 ppm2
0-1000 ppm
0- 500 ppm
0-25 ppm2
0-100 ppm
0-250 ppm
9-1000 ppm
0-25%2
0-1 ppm range
(Instrument modified with
1:5 dilution --system to
yield an effective 0-5 ppm
range)
3,490 ppm
1,003 ppm
252 ppm3
5.01%3
10.0%
50 ppm3
159 ppm
500 ppm
3.7 ppm
17 ppm3
92 ppm
21. 0%3
12.1%
2.0%
.934 ppm3
.404 ppm
.103 ppm
1 All instruments' zero responses were verified with zero nitrogen.
2 Primary operating range used during tests.
3 Concentrations used for routine span checks and calibration adjustments,
4 THC analyzer calibrated on methane.
51
-------
TABLE 13. ZERO/SPAN CHECK SUMMARY
Parameter
CO
C02
THC
HOX
02
S02
HC Species
No. of
Checks
29
30
31
29
29
34
33
Average Deviation
of Instrument Response
to Span Value
-0.91%
-1.16%
-0.13%
+2.94%
-0.07%
-3.25%
-0.80%
Standard
Deviation
11.88%
13.09%
±2.65%
+4.37
±1.32%
±4.98%
±2.20%
Upper 95%
Probability Limit
+2.77%
+4.90%
+5.06%
+11.51%
+2.52%
-t-6.51%
+5.11%
Lower 95%
Probabil ity Limit
-3.59%
-7.16%
-5.32%
-5.63%
-2.66%
-13.01%
-3.51%
-------
During the initial tests, zero/ipan checks were conducted between each
test. As the project progressed, confidence i,i the reliability of the
instruments increased and the frequency of these calibration checks was reduced
by allowing two or three tests to be completed between zero/span checks. As a
minimum, zero/span checks were conducted at the beginning and end of each test
day.
INSTRUMENT RESPONSE TIMES AND
THROUGH-PROBE CALIBRATION CHECKS
The response time for each instrument was adjusted to allow comparision of
individual data values. The goal was to ensure that all the instruments were
analyzing a fraction of the same sample at the same time. The synchronization of
the instrument responses permitted data calculations to be performed on the
instantaneous data as well as on the one-minute and test averages. For example,
the combustion efficiency calculations, which included data from three separate
instruments, could be performed on each set of instantaneous data because the
three analyzers were simultaneously sampling the same flare emission gas.
The instrument synchronization was accomplished by adjusting the sample
tubing length and volume between the sample manifold and the instrument and
adjusting the sample flow rate to the analyzer. In this manner, the transit time
of the sample between the manifold and the instrument could be adjusted. These
adjustments were made in reference to the instruments' initial upscale response
time.
The upscale and downscale response times of the instruments were checked by
introducing calibration gases through the probe and timing the subsequent
response of the instrument. Table 14 summarizes the results of these response
time checks.
An adaptor fitting was attached to the probe tip to allow the introduction
of the calibration gases. This 1/4" tube to 1" pipe adapter fitting presented
a flow restriction on the entire sampling system that was not present during the
i.ormal sampling. Thus, the response times reported in Table 14 may be somewhat
slower than the actual response times encountered during the sampling.
The introduction of calibration gases through the probe tip also served to
verify the integrity of the sampling system. There were no significant
differences observed in instrument responses to calibration gases introduced
directly into the analyzers as compared to calibration gases which were passed
through the probe. This evidence suggests that the sampling assembly did not
alter the composition of the flare emissions during the sampling.
Leak checks of the sampling system were performed on several occasions
during the test series. The procedure followed was to introduce zero nitrogen
in the probe tip and monitor the oxygen content of the sampling system. If leaks
were present in the sample system, they would be evidenced by elevated oxygen
levels in the sampling system. Table 15 presents the results of these leak
checks. This leak check data is believed to be conservative since the
installation of the required adapter fitting on the probe tip during these
53
-------
TABLE-.14. 'INSTRUMENT RESPONSE TIMES
(Up Scale (Seconds)
CO
CO?
THC
NOX
02
S02
Initial
32
30
35
32
30
29
90% Response
115
123
147
77
123
90
Down Scale (Seconds)
Initial
30
25
,34
25
25
32
90% Response
115
120
. 125
69
120
80
-------
TABLE 15. SAMPLING SYSTEM LEAK CHECKS
Date
6/17/82
6/17/82
6/18/82
6/19/82
71 1/82
02 Analyzer
Input Response
N? 0.80%
N2 0.25%
NZ 0.31%
Ng 0.24%
N? 0.30%
Percent *
Leakage
3.8%
1.2%
1.5%
1.1%
1.4*.
Assuming 20.95% oxygen in ambient air.
55
-------
Checks caused an increase in the vacuum present in the sampling system as
compared to during normal flare emission sampling.
The zero response of the other continuous analyzers to zero nitrogen passed
through the sampling system was also verified during the leak checks.
BACKGROUND MEASUREMENTS
Ambient air component concentrations of CO, CO?, THC, NOX, 0^ and SO? were
collected immediately before and after each of the tests or series of tests. The
collection of this background data documented the influence of local ambient air
concentrations on the flare emission measurements. The background concentra-
tions of these compounds of interest could have had a significant effect on the
data given the dilution effects of ambient air in both steam- and air-assisted
flare emissions. This is especially true in this situation since other
combustion sources were present and operating in the test area. A complete
listing of the background.concentration measurements is contained in the data
volume (Volume 11)'of this reort.
The background data was applied to the flare emission data by subtraction.
The average background values were subtracted from each raw flare emission
concentration values to determine the corresponding corrected value. This
subtraction removed any data bias between tests that was caused by variable
ambient concentrations. Additionally, the subtraction of the average of the
background concentrations before and after each test helped to compensate the
data for. instrument calibration drift that may have occurred between these
periods. It is recognized that to properly account for background concen-
trations, a dilution factor correction is required. As discussed previously,
dilution factors were not obtained in this study. Calculations indicate that an
error of less than 0.1% in combustion efficiency (for CE's greater than 98%) is
caused by the background correction method employed herein. As shown below,
this error is less than the sampling and analysis error. Therefore, the
background correction described above is suitable for this study.
•COMBUSTION EFFICIENCY ERROR ANALYSIS-'• .
It is assumed that accuracy of the combustion efficiency calculations is
dependent on two primary sources of error: (1) the accuracy of the listed
concentrations of the calibration gases, and (2) the accuracy of the instru-
ments' measurements of the gaseous samplei (instrument drift, interference,
repeatability, etc.). Other systemic errors due to sampling, data acquisition,
etc. are assumed to be negligible.
Table 16 lists estimates of the magnitude of these two pr -nary types of
error for each of the continuous analyzers. The instrument error data in this
table was calculated from the instrument responses to the routine span
calibration chocks. The calibration gas error data are nominal values supplied
by the vendors of the calibration gases.
Although these two types of errors are independent of each other, it is
assumed for this worst-case error analysis that they reinforce and are additive.
Likewise, it should be noted that the measurement errors of variables in the
56
-------
TABLE 16. ERROR ESTIMATES
Parameter
CO
C0?
HC
NOX
o?
SO?
Instrument Precision
Standard Deviation of Span
l.RB%
3.09*
2.65%
4.37%
1.32%
4.98%
Instrument
Errcr
.1.4 . 7 ppm
+1545 ppm
+1.3 ppm
±0.7 ppm
±0.27%
±0.046 ppm
Calibration Gas
Accuracy Error
±5.0 ppm
±1000 ppm
±1.0 ppm
±0.3 ppm
±0.42%
±0.019 ppm
Total Worst Case
Accuracy Error
±9.7 ppm
±2545 ppm
±2.3 ppm
±•1.0 ppm
±0.69%
0.065 ppm
u»
-------
combustion efficiency calculations are independent of one another. However,
for this worst-case analysis it is assumed that the errors in the measurements
of CO, C02 and THC concentrations reinforce.
For this worst-case analysis two sets of concentration values for CO,
and THC were selected: (]} for the high combustion efficiency case, and (2) for
the low combustion efficiency case. The total worst-case accuracy error values
were applied to these two sets of concentration values to determine the effects
on the combustion efficiency calculations.
• • . • ' ! •
High Efficiency Test Low Efficiency Test
CO Observation (ppm) 8 1000
CO? Observation (ppm) 7000 5000
THC Observation (ppm) 4 90
% CE Best Estimate 99.83 82.10
% CE Highest Estimate 100 • 87.49
% CE Lowest Estimate 99.46 69.02
This exercise shows that the high combustion efficiency test data are less
sensitive to accuracy errors than are the low combustion efficiency test data.
This analysis also provides worst-case estimates of the magnitudes of the
combustion efficiency accuracy errors. It should be noted that the real
accuracy errors (which were not directly measured) are expected to be less than
those listed above since the sources of error are independent and do not
necessarily reinforce.
Another mechanism for examining the' quality of the combustion efficiency
determinations is to examine the variance of repetitive determinations based on
consecutive measurements of CO, C02 and THC within each test. The data listing
and analysis program used to compile the data from these tests incorporated
routines to calculate the combustion efficiency from each set of data collected
every twelve seconds. The variance (and standard deviation) of the average
combustion efficiency value was determined from this data set as an indicator of
the precision of the data. The standard deviations of the consecutive
measurements of combustion efficiency ranged from + 0.15» to +11.1%. As in the
worst-case error analysis, the largest variances occurred with the low
combustion efficiency tests and the smallest variances were obseved with the
high combustion efficiency tests.
58
-------
APPENDIX A
GRAPHICAL-REVIEW OF -SELECTED TESTS
Five of the thirty-four flare efficiency tests completed during this
project were selected for more detailed data analysis via graphical aids. The
selected tests are as follows:
Test
3
55
57
28
33
Flare Type
Steam- assisted
Steam-assisted
Steam-assisted
Air-assisted
Air-assisted
Flare Gas Flow
(SCFM)
456
24.7
703
157
0.714
Heating Value
(Btu/SCF)
2183
2183
294
/•
2183
83
Steam-to-Flare
Gas Ratio
(Ib/lb)
0.45
6.9
0.15
. --
Combustion
Efficiency
Percent
99.8
68.9
99.9
99.4
98.2
These tests were selected because they represented a fairly wide range of
flow rates, heating values and combustion efficiencies. Figures Arl and A-2 are
photographs of the flare taken during these tests.
Three types of graphical test data plots were compiled from the digitized
instantaneous data collected on magnetic tape via a data logger. A detailed
listing of these data may be found in a separate data volume to this report
(Volume II). The plots chosen for this exercise are as follows:
SO;? and Probe Temperature versus Time;
THC and CO versus Time; and
CO and Combustion Efficiency versus time. ' • .
Figures A-3 through A-17 are the graphical plots of this data.
The plots of $02 and probe temperature show a positive correlation between
these two parameters. The temperature values are observed to lead the SO?
values by about 0.3 minutes. This time lead indicates the difference in
response times between the thermocouple/digital thermometer assembly and the
sampling system/SO? analyzer. The positive correlation between these two
parameters is due to both parameters being indicators of probe position relative
to the flame.
The graphical presentations of THC and CO? versus time show variable
amounts of positive correlation between these two species. Theoretically,
' 59 •'
-------
given a flare burning with constant combustion efficiency, the ratio of THC to
C02 should be constant regardless of flame post ion relative to the probe. This
is shown to be the case in the plot, for Test 55 and to a lesser extent Test 57.
Tests 3, 28 and 33 show little or no correlation bstween concentrations of THC
and CO?.
The graphs of combustion efficiency and CO versus time show an inverse
relationship between CO and concentrations on the combustion efficiency. This
demonstrates the importance of .'the CO term in the combustion efficiency
calculations and he usefulness of CO measurements as a primary indicator of
relative combustion efficiency.
.These plots of the test data also serve to graphically show the typical
range and variation of concentrations and temperature observed during the flare
tests. •
60
-------
Test 3 '•:,. Test 33
Figure A-l. Photographs of flare taken during tests.
-------
-------
O
O
ru
0.5
OJ
oo
o
o
o
o
o
o
ru
o
o
TEST 03 : JUNE 18 : HOUR .05:54
-1 — --1
<:,%'..<•> f;?.0
S02
TEMP
O
c
. o>
m
o
o
m
(M
(M
(M
o
oo
o
•»
o
. o
70.0 70,0
o
V
TIME:
Figure A-3. SO? and temperature vs. time.
63
-------
TEST 03 : JUNE 18 : HOUR 09:54
(f>
66.0 - 70.0
T IME: . (MIN)
74,0
78.0
Figure. A-4. THC and CO? Vs, time.
64
-------
o
00
o-,
Ch
TEST 03 : JUNE 18 : HOUR 09:54
o
o
Ch J
oo
o\
o
CO
o
o
a. '
, o
EFF
CO
58,0
f?.0
- 66.0 70.0
TIME (MI N)
74.0
78.0
Figure A-5. Efficiency and'CO vs. time.
65
io
K
oo
in
0
in"
05:
ICM O.
Q.
o
I*
ion
Jvo
Irvi
r
Joo
p.
Jo
j°
rvi
-------
TEST 55 : JUNE 23 :.H&UR 14:03
£02
TEMP
r>.0 19.0 23.0
11% (M1N)
Figure A-6. SO? and temperature vs. time.
66
o
o
-Jo
0
o o
\ ^
0.
51
LU
O
-------
o
o
o
o
o
o
in
li
0.2-
o
o
o
o
o .
o
o
o
o
CO
TEST 55 : JUNE 23 : HOUR 14:03
THC -
C02 . .
3.0
7,0
1.1 .0.
15.0 19,0
TIME (MIN)
23,0
27.0
Figure A-7-. THC and CO? vs. time.
6?
°
-------
o
f-
o
AJ
Li.
00 _
vo ,
vo
o
(>j _
a
u
TEST 55 : JUNE 23 : HOUR 14:03
3.0
7.0 11.0 1S.O 19.0 23.0 ,
TIME (MIN)
Figure A-8. Efficiency and CO vs. time.
68
27.0
ro
o
-------
TEST 57 : JUNE 24 : HOUR 09:13
•sr
ro
O
O
m
O
S02
TEMP
o
o
T
o
^o
m
vo
-------
TEST 57; JUNE 24 : HOUR 09:13
13.0
17.0
21.0 -25.0 29.0
TIME (MIN)
33.0
37.0
Figure A-10. THC and CO? vs. time.
70
° n
CM
-------
TEST 57 :. JUNE 24 : HOUR 09:13
1.7.0
?! .,0
25.0 29.0
TIME (MINI)
33.0
37.0
Figure A-II, Efficiency and CO vs. time.
71
(M
m
oo
(M
o
"3-
(M
O
ro
O
O
o
vo
o
nJ
o
00
o
-------
TEST 28 .: JUNE 28;: HOUR 17:14
SO 2
TEMP
o
o
' m
<=>
1°
Jr--
LJJ-
o
Jm
OJ
o
JON
P
,'tr>
o
o
.0 30.0
TIME (M1N)
Figure.A-1?. SO? and temperature vs. time.
7?
36.0
-------
TEST 28 : JUNE 28 : HOUR 17:14
1 4.0
TIME CM IN
Figure A-13. Efficiency and CO vs. time.
73
38.0
-------
TEST 28 : JUNE 28 •••. HOUR 17:14
THC
C52
TIME (M1N)
Figure A-14. THC and CO? vs. time.
74
38,0
§°
o
00
-------
TEST 33 : JUNE 29 : HOUR 22:49
fcl .0
TIME
er.. o
MIN)
69.0
73.0
Figure A-1B. SO? and temperature vs. time.
75
oO
Q_
z:
UJ
o f—
OD
OD
-------
TEST 33 : JUNE 29 : HOUR 22r49
O ,-^
6 s:
0 A
V> U_
m Q_
fc*^
Ol
§°
49.0
53.0
61.0 6S.O
TIME..(MIN)
69.0
73,0
Figure A-i6. THC and COp vs. time.
' >. 76
-------
TEST 33 : JUNE 29 : HOUR 22:49
ru
O
EFF -
CO ..
\o
m
ON
ON
AJ
f>
ON .
ON
LL
03
ON
ON
00
0s'
•*
0s
o
a,
—r—.
S3.0
—-r
S7.0 61.0 65,0
TIME (-M.IN)
69.0
73.0
Figure A-17. Efficiency and CO vs. time.
77
0)
rvj
o 21
aj 0_
\^
O
o
oj
o
VO
o
rvj
o
CO
o
*»
-------
*V£.IUGE*
OF OBS^RVtTIO'iS
03-
APPENDIX B
TEST 1
STATISTICAL. SUH1ART
PHOSE
lEMfMCl
175.3
73.8
117
28.2
S02 wax
tPPKI (PPNI
0.053 3.09
0.047 2.12
117 117
0.073 0.45
BACKGROUND
BACKGROUND
FILE
BACKGROUND 4
BACKGROUND 5
02
fPCTJ
19. 46
0.62
117
20.43
AMBICNT
CO C02
(PPH> IPPH)
3.8 7052.
1.4 3418.
117 117
4.7 373.
NEAS'J«fMf NTS
TIME
BEGIN
18/13
16/14
:i7:2«
:i4:22
THt WS UD AMBIENT COHHUSTION
(PPHI (HPHI TEHP(C» EFFICJCNCT
-0.7 3.5 267. 30.1 99.9
0.2 1.2 24. 0.6 0.1
117 117 117 117 117
4.4 2.9 265. 26.0
TIME
END
18^13:32:10
i«/i4:22:o7
OVERALL COMBUSTION EFFICIENCY = 99.96%
AH concentrations here and throughout Appendix B have been corrected
via subtraction. Thus, the actual measured value (i.e., uncorrected)
is the "average" plus the "average background" (last row).
-------
TEST 2
STATISTICAL SUMMARY
AVERAGE
STMOARO DEVIATION
NUMBER or OBSERVATIONS
AVERAGE BACKGROUND
PROBE
TEMP CO
158.7
49.6
84
38.2
SO2
CPPM)
0. 197
0.152
84
-0.008
NOX
(PPM)
2.16
1.31
84
1.41
02
19.73
0.42
84
20.26
CO
(PPM)
8.5
12.9
84
6.3
CO2
(PPM)
4719.
1684.
84
see.
THC
(PPM)
0.2
1.3
84
5.2
WS
(MPH)
2.8
0.9
84
2.7
UD
(DEC)
199.
33.
84
198.
AMBIENT
TEMMO
33.6
0.6
84
31.3
COMBUSTION
EFFICIENCY
99.8
O.2
84
BACKGROUND AMBIENT MEASUREMENTS
vO
BACKGROUND
FILE
BACKGROUND 1
BACKGROUND 2
TIME
BEGIN
17/16!32t27
17/17t25l51
TIME
END
17/16«55«38
17/17130147
OVERALL COMBUSTION EFFICIENCY = 99.823J
-------
IfSt 3
STATISTICAL SUMMAUT
STANDARD
OF
AVERAGE BACKGROUND
as
o
PRtRC S02 W» 0? CO C02
TEHPICl tPPPI CPPK) (PPM> «DEG) TCHP(C) EFFICIENCT
1.2 3.0 221. 30.5 99,8
1.3 0.9 27. 0.4 0.4
98 98 98 98 98
3.1 3.2 218. 27.2
END .
BACKGROUND 3 18/10:26:06
OVERALL COMBUSTION EFFICIENCY = 99.82%
-------
TEST 4
STAT1ST1CAL SUMMARY
AVERAGE
STANDARD DEVIATION
NUMBER OF OBSERVATIONS
BACKGROUND
PROBE
TEMPtC)
153.4
38.6
57
24.?
S02
(PPM)
O. 153
0.080
57
0.003
NOX
(PPM)
1.94,
0.96
57
0. 1 1
02
(PCT)
20.05
0.52
57
21.09
CO
(PPMT
75.3
35.5
57
$.1
C02
(PPM)
6616.
3004.
57
497.
THC
(PPM)
5.3
2.9
57
5.2
US
(MPH)
1.4
0.2
57
0.8
WD
(DEO)
164.
8.
57
186.
AMBIENT
TEMF(Q)
32.8
0.2
57
26.3
COMBUST I ON
EFFICIENCY
98. &
0.2
57
BACKGROUND AMBIENT MEASUREMENTS
C3
BACKGROUND
FILE
BACKGROUND 18
BACKGROUND 19
TIME
BEGIN
22/21:21:17
22/22S48t09
TIME
END
22/21:27:29
22/22:54:34
OVERALL COMBUSTION EFFICIENCY = 98.80%
-------
TCST H
STATISTICAL SUHfUBT
»mw>i
1*0 DEVI4TJOH
OHSf RVftTIOSS
,i SiCKSBOuHJ
FROBt SOi
12T.C O.CfaS
33.2 O.B40
84 H4
31.4 O.COS
OlOX 0?
(PPH1
1.45 20.30
0.9E. 0.52
8* 64
0.12 21. M
CO CO?
tPPH» IPP^J
61.1 5400.
36.1 3094.
84 84
3.8 495.
7HC
(PP«»
3.7
3.9
64
5.8
WS
«HPH)
0.9
0.3
84
0.6
UD
(OEG>
165.
11.
64
153.
AKBIENT CO«aUS'»10N
TE«P
-------
TEST 7
STATISTICAL SUHHART
AvCRAGC
PBOBE S02 NOK 02 CO C02
Tt«tP
AfCRAGC 138.4 0.044 1.62 .20.13 7.9 5224.
OCVIATION 39. T 0.026 O.q6 0.41 3.4 2254.
EPV4TIONS 103 103 103 103 103 103
AC^G»0'Jf»0 24.0 0.002 O.Q6 20.B7 6.1 396.
BACKGROUND TIHC
FILE BEGIN
BACKGROUND R 21/10tb5:50
BACKGROUSO 9 21/11:55:06
THC US
tPPN) f»IPH)
0.3 3.1
1.0 0.9
103 103
9.6 2.4
TIME
2i/n:o4:2l»
21/12:01:30
UC ANHieNT CONRUSTIOK
»DES) TENP«C) EFFICIENCY
262. 26.7 99.8
24. 0.5 0.1
103 103 103
280. 24.5
OVERALL COMBUSTION EFFICIENCY - 99.84%
-------
TEST 5
STATISTICAL SUMMARY
OVERAGE
DEVI AT ICM
c* ofr-ERvATioK?
«>RO8ex
TEMPtC*
172.8
34.6
1*5
3;. e
SO2 NO*
O2 CO CO2
(PCT) (PPM) (PPM)
0.070 2.O9 19.42 4.1 6113.
0.04-0 1.25 0.49 1.6 2908.
165 165 165 165 165
0.022 l.OO '20.28 3.4 368.
BACKGROUND AMP7ENT MEASUREMENTS
BACKGROUND
TILE
JiftCKPROUND 6
BACK GROUND ~>
TIME
BEGIN
"18/16!23!46
18/17145133
THC MS
(PPM) (MPH)
-0. 1 2. 9
0.2 1.0
165 Ifc^
—^5.0 2. 7
TIME
END
18/161 39s 05.
18/17:49:39
WD AMBIENT COMBUSTION
(DEG) TEMP(C> EFFICIENCY
294. 30.8 99.9
59. 0.6 O.I
165 165 165
289. 29.9
/
OVERALL COMBUSTION EFFICIENCY = 99.94%
-------
TEST 6T
STATISTICAL SIWKAHT
OF
AWtRA&C B*C*GROvlM5
00
on
PH03E S02 HO" 0?
Tt«P«C» IPPM> «PP»> IPCT)
«70.9 0.01R 3.77 20.19
l&O.O 0.001 *>-?2 0.69
21 21 ?1 21
22.2 -0.002 0.1<» ?0.9?
BACKGROUND AfflU
hACKGRO'JMJ
FILE
CO C02
tPPNI (PP><)
N/A 3758.
N/A 22*1.
N/A 21
5.4 362.
NT MEASUREMENTS
BtGIl
THC
N/A
N/A
N/A
3.3
TIME
fND
WS UD AMBIENT COWPUSTIOH
t«PH) TENPCC) EFFICIENCT
6.6 77. 25. * N/A
2.0 13. 0.3 N/A
21 21 21 N/A
3.1 131. 22.1
HACKGROUnlO SI
19/10:32:5?
-------
TEST 17
STATISTICAL SUN«URT
AVERAGE
STAHOMD DEVIATION'
1UMBE* OF OBSERVATIONS
AVERAGE
O3
CT>
PROBE
TE*P«C»
124.8
35.1
135
29.4
-
SQ2 NO*
tPPH) (PPK)
0.023 1.00
0.011 O.R4
135 135
0.005 -0.05
BACKGROUND
BACKGROUND
FILE
BACKGROUND 9
BACKGROUND 10
02 CO
CPCTI (PPM)
20.38 6.1
0.40 3.9
135 135
20.86 0.2
AMBIEM WTASi/Pt
TIHC
BEGIN
21/11:55:06
21/13:25:34
C02 . THC US VD AHetEsT CP«"JUSTl"N
iPt*l JPP") («10H) JOEG) TtHPJCJ EFflCITHCY
349?. -0.5 1.8'. 255. 26.9 99.8
2274. 0.6 0.9 77. 0.4 0.2
135 135 ]J5 135 135 135
434. 5.2 1.8 276. 26.3
TIHC
two
21/12:01:30
21/13:38:56
OVERALL COMBUSTION EFFICIENCY = 99.84%
-------
OF OB5iRV4II
20.19
0.38
103
20.
6.6 ?.2 269.
?.9 C.8 m.
103 10? 103
5.4 ?.0 2?<3.
UNHICNT
27.7
3.5
103
?3.n
CO»«U?T10N
"»9»%
0.3
153
00
FILE
10
11
TUT
21/13:;
21/14:1
21/13:36:53
OVERALL COMBUSTION EFFICIENCY = 99,45%
-------
TEST 56
STATISTICAL SUMMARY
•VEH4SE
STANDARD DEVIATION
NimdEft OF DBSCRVAT101S
AVERAGE b
OS
CO
pPOBt
TEMPtCI
120.4
134
32.9
S02 NOX 02 CO CC2
(PPK) IPCT) (PPM) CPPHJ
0.252 0.58 20.67 7.3 3120.
0.221 0.43 0.44 6.3 2190.
134 13» 134 154 134
0.016 0.78 21 .23 0.? 515.
BACKGROUND tHBHMT «C ftSUP.CHt«TS
BACKGSOUNO 1IHC
FILE BEGIN
BaCKCROUNP 21 23/lj:0
-------
TEST 61
STATISTICAL StMWAHY
AVERAGE
STANDARD OEWHTION
NUMBER OF 09SCRVATIONS
03
tO
pnngc
TEWMO
lf.6.8
32.6
58
21*2
SO?
IPPM)
N3X
tPPH)
0.033 1.32
0.012 0.47
58 58
0.003 0.11
BACKGROUND
BACKGROUND
FILC
BACKGROUND
BACKGROUND
0?
IPCT)
19.47
0.34
59
20.46
AH81CM7
CO C02 THC US
tPP«> IPPN> IPPH) «MPH>
393. 4 £273. 961.7 1.7
1U.Q 1623. 251.4 0.6
58 58 58 58
-0.6 357. -N 3.ej 0.1
MtASUREKENIS
VO AUBICST COMBUSTION
IDCG1 TE»P»C> EFFtCICNCjr
68. 21.1 81.6
12. 0.1 3. &
58 58 - 5»
70. ?!.?
V
- -
'' TIHE TI«E
BfGIW CND
30
31
24^??
25/00
:ie:40 ?«/Z2:3«:oj
:oa:59 25/00:17:09
OVERALL COMBUSTION EFFICIENCY =
-------
TfST 55
STATISTICAL SUNHART
AVERAGE
STANDARD OCVIATlo\
NUMBER OF OEEIBVAT1ONS
AVIRA6E BACKGROUND
PROSt
85.6
16.0
93
32.9
SO?
«PPH»
NOX
tPPMI
0.112 0.18
0.030 0.32
93 93
0.016 0.76
BACKGROUND
BACKGROUND
TILE
BACKGROUND
BACKGROUND
21
22
02 CO CO? TMC WS
20.83 171.0 2012. 735.5 2.2.
0.21 83.9 686. 212.1 O.H
93 93 S3 93 13
21 .23 9.2 515. 6.6 2.3
AflDHNT HE ASURfM£.N1S
TIKE IIMC
BEGIN t MO
23/li:o TEMPfCt EFF IC1EMCT
187. 30.9 68.9
59. 0.4 J.I
93 93 93
177. 30. 8
OVERALL COMBUSTION EFFICIENCY = 68.95%
-------
TEST 57
STATISTIC*!. SU«»URY
STANDARD DEVIATION
NUMBER OF 06SERVATIOHS
AVCCA6E BACKGROUND
IPPWI
0.113 2.68 19.80 5.0
0.067 2.05 0-89 7.6
100 100 100 100
-0.004 0.19 21.11 5.3
BACKGROUND AHHICNT HCASURI
BACKGROUND
FILC
BACKGROUND
BACKGROUND
TJBE
BEGIN
" ;»":""'
C02 THC US
(PPftl (PPM) tHPHI
6945. 2.0 1.5
4163. 2.3 0.5
100 100 100
42*. 5,8 1.0
IHEWtS
TIME
END
2^^09:50:57
UD AMBIENT COMBUSTION
«DCG» TEHPCO E^PiciEircir
168. 23.7 99.9
25. 0.3 0.1
100 100 100
156. 22.7
OVERALL COMBUSTION EFFICIENCY = 99.90%
-------
TEST 11
STATISTICAL SUMMARY
AVERAGE
STANDARD DEVIATION
NUMBER OF OBSERVATIONS
AVERAGE BACKGROUND
vO
PROBE
TEMP CO
159.7
31.6
63
21.4
S02
(PPM)
NOX
(PPM)
02
(PCT)
0.163 3.69 20.03
0.037 1.18 0.33
63 63 63
-0.003 O.22 21.34
BACKGROUND AMBIENT
BACKGROUND
FILE
BACKGROUND
BACKGROUND
CO C02
IPPtt)
7.1 5269.
5.2 3156.
63 63
3.6 512.
MEASUREMENTS
TIME
BEGIN
23
24
24/07
24/07
••O7:O6
:55:07
THC WS WD
(PPM) IMPH) (DEC)
1.6 1.2 135.
1.9 0.4 24.
63 63 63
5.0 1.1 B8.
TIME
END
24/07: 1 4:22
24/08:04:46
AMBIENT COMBUSTION
TEMP(C) EFFICIENCY
23.3 97.8
0.3 0.2
63 4,3
21. 0
OVERALL COMBUSTION EFFICIENCY = 99.83%
-------
TEST 11A
STATISTICAL SUHHART
tJEPtGC
sTMDMtD ce»!«nov
IL» OF OBSCMUKONS.
AVCRAGC B*C
3.31
1.02
33
0.22
02
IPCT»
20.10
0.35
33
21.34
CO
tPPHI
4.7
4.2
33
3.6
C02
IPPH>
6fe77.
1446.
33
512.
THC
IPPH)
0.2
1.4
33
$.0
US
UJPMI
^•2
0.4
33
1.1
WO
IDEG)
130.
24.
33
88.
AUBIENT
Tt«P«C)
23.2
0.3
33
2.1.0
COMBUSTION
EFFICIENCT
99.9
0.1
33
BACKGROUND ANBICNT HEASUREHENTS
BACK5BOUNO
FIlE
eeciN
TlflE
t«0
nACXGWOUHO 23
BACKGROUND 24
ir:o6 24/07:14:22
iS:oT ?4/os:o4:46
OVERALL COI^IBUSTION EFFICIENCY = 99.93%
-------
TEST 11B
STATISTICAL SUMMARY
AVERAGE
STANDARD DEVIATION
NUMBER OF OBSERVATIONS
AVERAGE BACKGROUND
PROBE
TEHP(C)
181.9
7.7
18
21.4
SO2 NOX
(PPM)
02
(PCT)
0.198 4.17 19.92
0.028 O.74 6.21
18 18 18
-O.OO3 O.22 21.34
BACKGROUND AMBIENT
BACKGROUND
FILE
BACKGROUND
BACKGROUND
23
24
CO CO2
-------
TEST I1C
STATISTICAL St/fHAftr
1IWK*
095r*V*TIC*>
iJSS.
181.4
19.2
12
21.4
SO?
CPPHI
0.189
0.081
1?
-0.003
MO*
tPPWl
4.00
1.T3
12
0.22
02
CPCTI
19.-J8
0.3*
12
21.34
CO
CPPfO
11.6
4.6
12
3.6
C02
IPPHI
8210.
1108.
12
512.
THC
3.3
1.5
12
5.0
IIS
«MPH1
1.4
0.4
12
1.1
WO
tt)t6»
125.
25.
12
88.
AMBIENT
TEHPfO
23. *>
0.1
12
21.0
COMRUSTION
EFFICIEUCT
99.8
0.1
1?
VD
PILE
23
BftCKGi-CUND 2»
24/^7:07:06
END
J5:OT 24/08:04:46
OVERALL COMBUSTION EFFICIENCY = 99.82%
-------
TEST 39
STATISTICAL SUMMARY
AVERAGE
STANDARD 3EV1ATIOK
NUMBER OF OBSERVATIONS
AVERAGE BACKGROUND
PRPDC
TEMP tPPH)
19.53 49.9 5413.
0.39 26.4 1481.
57 37 57
20.61 0.0 421.
AMBIENT MEASUREMENTS
TIME
BEGIN
24/2U 16112
24/22118:40
THC US
tPPM) (MPH)
33.3 0.9
24.7 0.6
57 57
2.«> 0.5
TIKE
END
24/2J:3lsll
24/22:30:09
UD AMBIENT COMBUSTION
(DEC) TEMP*C) ItFFIClENCY
94. 22.2 98.4
67. 0.: 1.0
57 57 37
93. 21.8
OVERALL COMBUSTION EFFICIENCY - 98.49%
-------
TPST 59'.
MEIUG-;
STA4DMO P^f
«V*BE* e^ fl
qiCKt»*oim
PSOBE
1*2.*
29.9
J8
20.*
$02
0.052
0.025
3S
0.002
n-n 02
tPP«i (PCTI
1.30 ' 19.M
0.53 0.^0
!•!» 38
0.14 20. (t\
CO
«PP«»
62.1
?3.B
31
0.0
C02
5575.
1659.
3R
421*
THC
«PPM
»5.3 .
21.7
38
2,9
US
fHPHI
i»o
0.6
38
P.?
UD
«DtS»
1?..
46.
38
93.
AMBIENT
TCHP«CI
22.1
0.1
38
21. P
EVf JClENCt
97.9
3.9
5«
BACKGROUNH
FILE
RUCKGOOUftD 29
nUCKGROUVO 39
END
n:*o 34/22:30:0^
OVERALL COMBUSTION EFFICIENCY = 98.11%
-------
TfST 59<»
STATISTICAL
SIAftOAND DEVI*TI3N
•MMBC* OF OBSERVATIONS
r.f BACKS*OUND
«o
O9
PROBE
TE*°«C>
133.1
15.9
20.4
S02 X0» 0? CO C02
iPPM CPPM) tPCTI tPPN) IPPH)
0.076 1.62 19.71 25.4 5090.
0.014 0.35 0.30 7.9 1002.
19 19 19 19 19
0.002 0.18 20.61 0.0 421.
BACKGROUND AHBU.Nt *£ ASURtfKNTS
FILE BE6I«
HACKG^OU^O 30 24/22:13:4?
TMC
CPPH)
9.4
3.1
19
2.9
TIKE
EMD
24/21:31:11
24/22:30:Oa
US HO AMBIfNT COH8USTIOM
ii'fH* EFFICIEXCT
0.7 140. 22.3 9«.3
0.6 79. 0.0 8.1
19 19 19 1«
0.5 ^S. 21.8
OVERALL COMBUSTION EFFICIENCY = 99.32%
-------
AVEHA6E
STANDARD DEVIATION
ER OP OBSERVATIONS
AVERAGE BACKGROUKO
TEST 60
STATISTICAL SUHHARY
PROBE
TEHPIC1
«.7
20.7
97
21.?
S02
tPPH>
0.052
0.014
97
0.003
NOX
0.99
0.47
97
0.11
02
CPCT)
19.70
0.33
97
20.46
CO
CPPHI
28.3
9.5
97
-0.6
coa
(PPHI
3685.
1290.
97
397.
TMC
(PPH)
11.8
?.9
97
3.5
US
(HPH)
1.3
0.4
97
0.4
UD
(DEC)
&5.
12.
97
TO.
AHDIENT
TiHPCCl
il.7
0.0
97
v 21.2
COMBUST IOW
EFFICIENCY
98.6
0.4
97
vo
0ftC(«GROU«C AHBIENT fieASURCBENTS
BACKGPOUNO
FILE
30
BACKGROUND 31
TIPE
BEGIN
2»/22:in:4«
25/oo:oa:5?
TIME
END
25/oo:i7:-09
OVERALL COMBUSTION EFFICIENCY = 98.92%
-------
TEST 31
STATISTICAL SUMMARY
AVERAGE
STANDARD DEVIATION
OF OBSERVATIONS
AVERAGE BACKGROUND
PROBE
TEHPtC)
119.6
43.4
78
35.1
S02 NOX
0.039 O.57
O.017 0.66
78 78
-0. 003 0. 42
BACKGROUND
BACKGROUND
FILE
02
tPCT)
19.85
O.60
78
2O.45
CO
tPPM)
34.1
32.3
78
0.6
C02
(PPM)
3347.
2564.
78
410.
THC WS
(PPM) (MPH)
11.3 2.3
12.1 0.8
78 78
11.3 2.3
HD AMBIENT COMBUSTION
-------
TEST 16
STATISTICAL SUMMARY
AVERAGE
STANDARC DEVIATION
NUMBER OF OBSERVATIONS
AVERAGE 6ACKGRQUND
PROBE
TEMP(C»
127.5
20.7
103
24.7
S02
(PPM)
NOX
(PPM)
02
CO
(PPM)
C02
(PPM)
0.121 1.87 20.43 7.7 4059.
O.056 1.02 0.38 3.5 1501.
103 103 103 103 103
-O.002 0.26 21.17 O.6 394.
BACKGROUND AMBIENT MEASUREMENTS
BACKGROUND
FILE
TIME
BEGIN
1HC
(PPrl)
2.4
1.5
108
«.o
TIME
END
US WD AMBIENT COMBUSTION
tMPH) (DEC) TEMP(C) EFFICIENCY
1.7 149. 22.7 99.7
0.6 22. 0.2 0. 1
103 103 .103 103
o.a tea. 22.3
BACKGROUND 27 24/11 i2fr: 05 2A<'ll:3'5M4
OVERALL COMBUSTION EFFICIENCY = 99.75?!
-------
TEST 16ft
STATISTICAL SUMMARY
STANDARD DEVIATION
ER OF OBSERVATIONS
O
INJ
PSOUE
TEHPtO
113.0
22.4
34
24. T
SOS NO*
02 CO C02 THC
tPCTI tPPHI CPPHI «PP«>
0.082 1.39 20.55 6.1 3236* 2.3
0.033 1.02 0.44 3.5 1657. 1.8
34 34 34 34 34 34
-O.QC2 0.26 21.17 0.4 3<»4. 6.0
BACK&ROUKO AMBIENT HC *Slǣǣ NTS
BACKGROUND
FUE
TJRE TIHE
BE6IN END
US WD A«BICNT COMBUSTION
t«PH> «OE6> TEHP*C» EFFICIEUCY
1.8 136. 22. fl 99.7
0.9 13. 0.1 0.1
34 34 34 34
0.8 IBS. 22.3
BACKGROUND 27 24/11:26:05 24/11:35144
OVERALL COMBUSTION EFFICIENCY = 99.74?
-------
TTST 16R
ST«TlSTie«L SUHfURT
O
U)
PBHBf
TCPP«C1
HP. 6
11.7
28
*,'
SO?
(PPWI
0.16I>
c.osp
2B
-0.002
Fp.r
unit 02 CO CO?
(PPH1 (t»CTl (PPHI (PPMI
2.4? 20. ?0 9.6 5291.
0.98 0.36 3.7 11P3.
?« 28 28 28
0.26 21. IT 0.4 394.
PACKGROUND aHflt^Nl ^t *SU"»EHL NTS
BEGIt
THC US UO ftHBICNT COMBUST 10*
(PPHI ««PH»
-------
STANDARD 01VIST ION
OF 3BSt«*A110*43
I RICKGftOUNP
rrsT i6c
ST»TISTIC»L SIMK»»T
PHOHt
TtPPtC)
125.0
B.7
2?
24.7
S02
IPPHI
0.107
0.052
22
-0.002
WO*
TEHPtCl
164. 22.4
14. D.I
22 22
19«. 22.3
COMBUST 1 OK
EFTtCltHC*
99.7
0.1
2?
BBCK6POUNO
Fur
2T
TIHC
EWO
OVERALL COMBUSTION EFFICIENCY =99.74%
-------
160
ST»TISTIC«L
STAN3MO OCVTATI3N
CF OBStRVMlOY;
PROnC SO?
T£P»P«C» lPP«n
13*. 3 0.13H
7.4 0.045
19 19
2».7 -0.002
NOX
2.?«
O.»l
19
0.26
0?
tPCT*
20.36
0,?4
19
21.17
CO
IPPB)
7.9
2.5
19
0.4
CO?
«PPB)
4*58.
627.
19
39«.
TMC
(PPKI
2.0
1.3
19
6.0
MS
f«PH*
1.9
0.6
19
0.8
HO
155.
26.
19
1*8.
TCHPCCI
22.4
0.1
19
22.3
COMBUST I OH
9S.?
P.I
1?
?7
tlft
TIME
r NO
OVERALL COMBUSTION EFFICIENCY = 99.78%
-------
TEST 5*
STATISTICAL SUHHART
AVERAGE:
STAMURO DEVIATION
OF OBSERVATIONS
»VE«AC-C BACKliHOUNO
PROBE
TEHPCCI
197.1
46.1
92
34.4
S02
CPPH)
0.55*
0.476
92
0.085
NO*
CPPH)
5.00
2.29
92
0.22
02
(PCT)
19.15
0.51
92
20. .5 "5
CO
(PPM)
6.8
3.8
92
0.8
C02
-------
Tc"St 23
STATISTICAL SUNHART
AVERAGE
STAWDAftD DEVIATION
NUftBEM OF OBSERVATIONS
E RAC«GR3'J10
TEMPUI
235.1
111.5
103
3-=. I
S02
0.018
0.017
103
-0.003
.SI,
5.9D
5.31
103
0.*?
B465.
«,4fe.
5 13
410.
THC
-5.0
0.9
103
11.3
US tfO
1.5 160.
0.5 21.
10? 103
?.3 186.
ABBIENT
TEMP(C»
28.4
a.i
103
29.6
COMBUSTION
.100.2
0.5
103
BACKGROUND
RftCKSROUSO
FltE
BEGI*
21/18:10:47
TlWf
ruin
BACKGROUND 13 21/18:19:47 21/18:37:3*
OVERALL COMBUSTION EFFICIENCY = 100.01%
-------
STANDARD OFVI AT 10*
•WISC"» OF OBSERVATIONS
AVERAGE BACKGROUND
O
Co
TEST 52
STATISTICAL SUH^AfU
PR01E SO? NOK 0? CO C02 T«C US
TE«tP«CI (PPM) (PPHI (PCTI (PPt)
-------
TEST 53
STATISTICAL SUMMARY
AVERAGE
STANDARD DEVIATION
NUMBER OF OBSERVATIONS
AVERAGE BACKGROUND
PROBE
TEHP(C)
160.6
47.1
112
30.5
SO2 NOX
(PPMI (PPM)
0.729 2.83
0.264 1.89
112 112
0.050 0.14
BACKGROUND
BACKGROUND
FILE
BACKGROUND
BACKGROUND
13
16
O2 CO C02
EFFICIENCY
235. 30.4 99.3
33. 0.3 O.4
112 112 112
225. 3O.6
OVERALL COMBUSTION EFFICIENCY = 99.40*
-------
TEST 26
STATISTIC*!. SIWJURT
MERAGE
STANDARD
NWBE* OF
AVERAC-L PkCRSIOUID
PR3BE
TE»"»«C>
194.0
42.5
32. 8
S02 NOX
CPPt» tPP»">
0.356 5.34
0.220 2.93
124 124
0.013 O.PO
BACKGROUND
FILE
BACKGROUND 35
B«CKG«?OUM0 36
02 CO C02
CPCTl «Pi»N» €PPR»
17.62 5.5 6270.
0.44 2.1 2646.
124 124 124
29.63 0.2 375.
ASBIIHT qrasiHt'cwrs
illn
29/13:34:20
29/14:44:11 ;
THC US HO AMBIENT COMBUSTION
IPPH> t«PH»
-------
TCST 65
STATISTICAL SUNHAKV
AVERAGE
DCVIATIOM
•IffBC* Of OBSCftVATIOKS
P«OBE
TEM»tC>
111.3
26.6
83
26.4
S02
0.062
0.024
83
0.091
<"!i
2.40
1.82
83
0.15
02
t»CTI
20.33
0.35
63
21.14
CO
20.3
5.1
83
4.1
C02
. tPPfll
4878*
2168.
83
399.
THC
tPPH»
0.6
1.4
83
4.2
US
2.1
O.S
83
1.8
wo
«DE6»
1T2.
12.
83
179.
.TW"!,
31.0
0.2
83
26.9
COMBUSTION
EFFICIENCY
99. S
0.3
i
83
Flit
BACKGROUV3
BACKG43U13
««BtE«T ^
1/00:10:33 1/00:14:28
OVERALL COMBUSTION EFFICIENCY * 99.57
-------
TEST 28
STATISTICAL SUKRART
STAB9MD DEVI*T10*
HWtBCft Of OBSERVATIONS
AVERAGE
rv>
PROBH S02
TC*PiC> IPPHI
212.1 0.329
71.* 0.224
143 143
30.7 0.009
BACK6POU*
FItE
BACH&BOUN
ROY 02 C3 C02 THC US
CW«1 CPCTI
-------
TEST 31
STATISTICAL SUHNART
AVCRAGE
STAUDBftO
•MIME* OF
AVEHftG£ BACKGROUND
PftOHE S02 HOX 0? CO C02 TMC US
TEKPICI CPP«» I PPM) tPCT) «?PW> fPPH) «PP*> («PH)
159.3 1.226 4.32 19.83 27.9 4568. 10.1 2.S
37.6 0.751 2.3* 0.56 12.1 2258. 3.7 1.1
121 121 121 121 121 121 121 121
39. » 0.022 C.'ig 20.53 0.5 327. 5.6 3.0
BACKGROUND »BBIfNT HC*SURt «ir?iTS
FILt 85IGIN E«IO
HftCRG'3'J««G 33 28/l?:57:H ?P^lfl:04*^0
HRCR&BOUNO 3% 2A/ia:5&:5« 28/19:05:46
WO ft« TEHPtO EfFlCiriCT
315. 31.1 99.1
43. 0.3 0.4
12-1 121 121
317. 30.2
•
OVERALL COMBUSTION EFFICIENCY = 99.17%
-------
ST»HD«RD
OF
PROBE
TEHPICJ
102.2
25.5
69
2*.9
S02
0.935
0.395
6"
0*026
NO*
0.97
0.53
69
0.00
TEST 66
STATISTICAL SIMNARr
02 CO C02 THC
20.05 !2?.4 2*32. 1265.0
0.39 114.2 1177. 510.7
69 69 6-J 6?
21.01 ?'.» 337. 12.7
US WO
«*PH» tOE6)
0.6 144.
0.2 23.
69 69
1.1 174.
AMBIENT
TEHPICI
29.7
0.1
69
29.1
COHDUSTIOM
EFFICltKCt
60.6
ll.l
69
BACK6ROUSO
FltE
3B
two
t 10
OVERALL COMBUSTION EFFICIENCY » 61.94%
-------
AVCftAGt
STANDARD OCHIATION
OF OBStRVAIIO'WS
• VEIUGF BACKGROUND
PHOBC
66.3
14.3
57
?8.9
SO?
IPP«U
1.365
0.502
57
0.025
NOX
1.06
0.40
57
0.00
TCST 29
STATISTICAL SUM1
QZ CO
(PCTI
-------
TEST 29«
ST»TISTJC»L
•UMBER C* OBSi«W»TIO«r
* B«CRCBOU>*3
PWO:IE - 50?
TTHPICI «PP*»
75.1 1.236
P.I 0.509
Z* ?B
2^.9 D.0?6
N0«
«PPW)
1.09
0.40
?R
0.00
0?
«PCT»
20.34
0.22
28
21.01
CO
IPPH1
146.6
2R.6
28
2.«
C02
IPPHI
1529.
4T6.
26
33 T.
THC
(PPMI
1097.1
243.0
2»
12.7
US
<«PHI
0.6
0.3
28
1.1
uo
191.
?1.
2B
17B.
AMBIENT
TEHP«C»
29.5
0.1
2B
29.1
COMBUSTION
54.1
10.1
2*
CUT:
T.TPE
HE6I1
BtCKGROUNO 39 29/21:46:20
OVERALL COMBUSTION EFFICIENCY = 55.14%
-------
TEST 29*
STATISTICAL SUNHART
StANOARD DEVIATION
OF OBSERVATIONS
BACKGROUND
PROBE
TE«PtC»
97.2
9*9
29
28.9
SO?
1.499
0.470
?9
0.0 2f
NIX
1.04
0.41
?9
0.00
02
IPCT)
20.12
0.18
29
21.01
CO
CPPHI
213.9
34.6
29
2*4
CO?
IPPB)
2808.
440.
29
337.
THC
tPPH)
1255.4
325.6
29
12.7
US
0.7
0.3
29
1.1
UO
tOEG)
189.
15.
29
17S.
AMBIENT
TEHPCC)
29.2
0.1
29
29.1
COMBUSTION
rrriciEKCT
65.6
7.7
29
B8CKKROUND AHRIENT HC 6SUP.EHCKT S
TJHC
FILE
BACKGfiCUND 38
HACK6PPUNO 19
TIRE
: 56: S3
29/21:55:10
OVERALL COMBUSTION EFFICIENCY = 65.6*
-------
PROBE SO?
AVEMGC 105.0 0.051
STA*0»«3 DEVIATION 10.1 0.014
•njMBE" OF DBS r«»»T IONS 67 67
•V[*A3£ l«CK,43u-4J 2<*.3 0.032
UST 64
STATISTICAL SUMMARY
W0» 0? CO C02
fPPf! tPCTI M» tOtG>
•0.1 0.9 185.
1.9 ?.3 23.
67 €>7 67
8.8 T.7 1P«.
AHBICKT COWBUSTJ01
TEHPtO CFFICIEWCt
25.3 99.7
0.1 0.2
67 67
25.1
OB
FILE
BACfGROUNO
90
41
PECIH
30/oo:i3:
30/02:01:
EHO
30/00:15:1*
OVERALL COMBUSTION EFFICIENCY = 99.74%
-------
TEST 62
STATISTICAL SUHHART
AVERAGC
S7AHDAHD DEVIATION
MU»BE« OF OeSCRVATIOHS
SJACK6ROUNO
\o
ROBE
WP«C»
108.3
22.1
113
29.0
SO? NOX
IPPH> «PPH)
0.841 0.60
0.667 0.36
113 113
0.032 0.03
BACKGROUND
BACKGROUND
FILE
B4CKGRCUMD AO
BACKGROUND «]
02
90.2
32*0
113
-1.7
HEftSURf
BCGUI
30/00
30/02
:io:oa
:oi:s3
CO? THC US
IPPHI (PPH) INPH)
3076. 99.8 0.9
1206. 30.6 0.2
113 113 113
• '28.' 8.8 9.7
:«EfsTs
TINE
tUD
30/oo:is:i*
•3p/02:oe:-M
UD A1BIENT COHHUSTlO'i
175. 25.2 93.8
IS. 0.1 l.t
113 113 113
16*. 25.1
OVERALL COMBUSTION EFFICIENCY = 94.18%
-------
STAH3ABD DEVIATION
Of
PROBE
TCMPfO
123.%
30*4
106
?o.O
S02
JPPH»
9.057
0.037
156
0.032
TtST 63
STATISTICAL SIM
NOV 02 CO
CPPfO CPCTI tPPXt
1.5T ?0.13 19.9
1.06 0*48 9.8
106 106 106
0.01 71..11 -l.T
H»RT
C02
4184.
2204.
1 06
4?R.
THC
«PPK»
6.5
6.7
106
a.B
us
1.4
0.7
106
0.7
tfO
fOEGI
97.
4«.
106
194.
AMBIENT
TEHPICI
24.8
0.6
106
25.1
COMBUSTION
99.1
1.?
ICf
BiCKGROUNC
FILE
40
30/00:lO:0-!
30/02:01:33
EVD
30/00:15:1*
30/02:00:14
OVERALL COMBUSTION EFFICIENCY = 99.37%
-------
Si
STATISTICAL
JLVIATION
OF
R<»Bt
HPtCt
86.6
12.5
10?
?7.5
502 NOK
•PP«» IPPWJ
3.793 0.74
1.440 0.?
-------
TPST 32
STATISTICAL SUHHAHT
»VERASE
SlftHDARD DEVIATION
NUMBER OF OBSERVATIONS
AVERAGE BACKGftOUNO
PROBf
120.9
36.7
121
27.5
S02 NQX
IPPHI
-------
STATISTIC*!. SU1KIRT
BVCRAGt
OCVIMtO*
.CMV*TI(HIS ,
IACKK*OURQ
P*0*E SO?
TEHPICI (PPf)
80.3 2.204
9.7 0.827
4« 44
27.5^ 0.044
MOk
0.63
0.25
44
-0.03
02
«PCT>
20.90
0.11
44
21.23
CO
CPP«»
12.2
6.3
44
0.1
COS
»PPH»
ITfil.
S20.
44
436.
THC
7.3
5.4
«4
14.6
WS
CfcO
29/21 :55MO
30/00: 15: 14
OVERALL COMBUSTION EFFICIENCY - 98.91%
-------
CF
AVERAGE BiC
TEST T?i
STATISTICAL SU*»A«r
PROBE
TE«P«CI
144.1
23.9
7T".
2T.5
SO?
CPPHI
3.896
0.680
T7
0.044
»n» o?
(pp»r (PCTi
2.39 20. ?7
0.63 0.22
17 77
-O.OT 21.23
CO CO? THC
|PPH» tPPPJ IPPM>
24.8 «RI1. 26.5
8.2 1077. ^.3
T7 77 7t
'.O.I «3&. 1*.6
OS «D
0.7 134.
0.2 11.
77 77
1.9 175.
ABBIEHT COBBOSTtOW
26.3 98.9
0.1 0.2
77 77
26."
FILE
TfPt
two
BACKG'OUVD 40
OVERALL COMBUSTION EFFICIENCY = 98.86*
-------
APPENDIX C.
CALCULATION OF DESTRUCTION EFFICIENCY (DE)
While dilution factors are -required for an accurate determination of
destruction efficiency (DE), estimates of DE can be made for total hydrocarbons
(THC) and for individual hydrocarbon species. Assuming:
(a) All carbon resulting from combustion is accounted for in the
measurements,
/ • . . • . . .
(b) Dilution is neglected, and
(c) The relief gas is 80% propylene and 20% propane.
One can calculate DE's as follows:
= c°2 + CO 4 Soot x 100
C02 + CO 4 Soot 4 THC
DEpropylene = MJCgZ + CO 4 Soot * THC) - Cprnpy1pnp
0.8(C02 + CO 4 Soot 4 THC)
DEPropane = MM2 + CO 4 Soot 4 THC) - Cprnpnne x 10Q
0.2(C02 + CO 4 Soot 4 THC)
where: DEjHC = Total Hydrocarbon DE (%)
DEPropylene = Propylene DE (%)
DEpropanc. " Propane DE (%) .
Cpropyiene ~ Measured Propylene Concentration (ppmv)
^•Propane = Measured Propane Concentration (ppmv)
and all other variables as previously defined.
Tables C-l and C-2 provide calculated values for DEjHC» DEPropvlene»
DEprQpane for each of the tests. Note that a Combustion Efficiency t fit) value
is also provided, am' in some rases, it differs from CE values reported in
previous tables. Thi1- & due to the method of calculating CE:
(a) THC, Cpr9pyiene» and cPropane values from the integrated bag samples
(Tables 7 and 8) were used instead of the continuous THC data used in
previous calculations, As noted in the report, there are differences
between the continuous and integrated bag THC results,
-------
TABLE C-l. DESTRUCTION EFFICIENCY ESTIMATES
STEAM-ASSISTED FLARE TESTS
Test
Number
High Btu Tests
1
2
. 3-
4
8
7
5
17
50
56
61
55
Low Btu Tests
57
11
59
60
51
16
54
23
52
53
CE
99.85
99.66
99.80
91.67
93.15
99.79
99.82
99. 7o
99.37
99.49
76.92
61.63
99.73
99.67
98.22
98.71
98.48
99.61
99.81
99.84
97.93
99.24
D
-------
TABLE C-2. DESTRUCTION EFFICIENCY ESTIMATES
AIR-ASSISTED FLARE TESTS
Test
Number
High Btu Tests
26
65
28
31
Low Btu Tests
66
29
64
62
63
33
32
CE
(X)
99.85
97.95
99.78
98.84
49.98
48.03
99.49
92.05
99. 14
97.31
98.50
i flu
I Of \
99. 94
99.90
99.93
99.42
54.17
•51.53
99.68
94.38.
99.54
97.98
99.04
OEfr^lmi,
99.99
100.00
99.98
99.61
55.66
53.11
99.93
95.44
99.72
99.72
99.87
^Propane
(SO
99.98
99.99
99.98
99.61
57.22
54.76
99.89
95.57
99.72
99.61
99.80
127
-------
(b) Since the integrated bag sample hydrocarbon data were used and no.
background data were available for propane and propylene, the
measured values for hydrocarbons in the flare plume were not
"corrected" by subtracting background concentrations. Thus, the CE
and DE values in the following tables were calculated from a
consistent set of data.
(c) Did not segment tests 11, 16, 29, and 32 due to lack of hydrocarbon
species data.
128
-------
APPENDIX D
SOOT COMPOSITION
ENGINEERING-SCIENCE
CABLE ADDRESS ENGINSCI
TELEX: 77-6442
3109 NORTH INTERREGIONAL • AUSTIN.TEXAS 78722 • 512/477-9901
CABLE ADC
February 1, 1-983
Dr. Bruce Tichenor
Industrial Processes Branch (MD-63)
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Dear Dr. Tichenor:
In response to your letter dated December 9, 1982, and our subsequent
telephone conversation, I ain providing a summary of the procedures used for
the PNA analyses of soot collected during the flare efficiency test.
The sample probe assembly included an in-line particulate filter housed
inside the heated section of the probe about six feet from the probe tip.
This in-line particulate filter assembly served two purposes: 1) collection
of particulate samples from smoking flares for subsequent analysis, and 2)
maintaining the cleanliness of the sampling system. The preweighed filter
element? usrd were of the thimble configuration and constructed of 0.3
micrometer glass fiber.
The filters were changed before and after each of the smoking flare
tests. Following the tests the filters were reweighed to determine the mass
of particulate col lected. This information, combined with the measured flow
rate of sample through the probe assembly, allowed the calculation of the
gross particulate concentration of the flare emission at the sampling
location. It should be noted however, that these paniculate samples were
not collected isokinetical ly and thus, represent only gross estimates of the
particulate concentration. The flare particulate emissions were not
isokinetically sampled because it was not practical to directly measure the
plume velocity.
Table 1 is a summary of the mass particulate concentration data
collected during the test series. Although these samples were not collected
isokinetically, the data shows distinct differences between particulate
loadings of nonsmoking, lightly smoking (Test 65) and heavily smoking (Test
4) flare tests.
Samples were prepared for PKA analysis by Soxhlet extraction of the air
filters as received with 200 ml of pfithylene chloride for 24 (+2) hours. The
condenser water was chilled to 1 - 6°C and no solvent loss (bp 40°C) was
noted. The samples were transferred to bottles and the glassware washed with
additional tnrthylene chloride which was added to sample. Samples were dried
for several days .or anhydrous sodium eulfate which had been kiln-fired at
450°C to remove organic compounds. Sample extract volumes were carefully
reduced to 1 mL using Kuderna-Danish flasks and three-ball Snyder columns.
129
-------
ENGINEERING SCIENCE
Dr. Bruce Tichenor
February I, 1983
Page 2.
Sample extracts were analyzed by f /MS using a 60 meter J & W fused-
silica DB-1 bonded SE-30 capillary column, using d^-Chrysene as an internal
standard. Individual compound response factors were obtained by running a
standard mix prepared from EPA standard solution concentrates, and spiked
with d]2~chrysene. Quantitation was based on integrated peak areas and was
performed by the GC/MS data system.
As a further quality assurance measure, a solvent blank was run, spiked
with di2~chrysene, and was found to be free of interfering peaks. No
recovery efficiency data can be given, since there was only one of each
sample. Ideally, one should consider collecting at least one replicate
sample which could be spiked (directly onto the soot) with a known addition
of a surrogate PNA. By this means, an indication of the efficiency of
recovery of PNA's fron the soot matrix could be obtained. Tables 2 and 3
summarize the results of the PNA analyses.
As I indicated in our telephone conversation, I am unabje to calculate
these results in terms of mass emission rates (e.g., mg/10^ Btu.'rcg/hr, etc.)
because of the lack of isokinetic sampling and a measure of the dilution
between the reaction zone and the sampling probe.
If you have any questions, please call me at 512/444-5830.
! ' ' '
Sincerely,
Marc McDaniel
Attch.
/kg
130
-------
ENGINEERING SCIENCE
Dr. Bruce Tichenor
February 1, 1983
Page 3.
Table 1. Flare Efficiency Study Participate Analysis.
Wt. gain, Sample Time Sample Rate Concentration
Test No. Filter No. (grams) (minutes) U/min) (u9/£)
2, 3, 1, 5, 7 A-l 0.0063
7, 17, 50, 51
23, 52, 53, 54 F-l • 0.0071
4 (Smoking) F-2 0.0810
8 (Smoking), F-3 Q.0819
55, 56, 11, 57
16, 59, 60, 61
28, 31, 26, 29 F-4 0.0179
33, 32, 62, 63
64, 80, 81, 82
83, 84
65 (Smoking) F-S 0.0183
25
25
18.51
18.51
18.51
18.51
18.51
18.51
274
177
40
) - ' - - .'.,-•
Includes time for.probe positioning while probe was in the plume but
before formal initiation of the test.
131
-------
ENGINEERING-SCIENCE
Dr. Bruce Tichenor
February 1, 1983
Page 4.
Table 2. Total Polynuclear Aromatic Hydrocarbons in Filter Samples, in
Micrograms*.
E.S.
Sample *
F F F F F
1 2 3 *4 F5
napthalene (0.05)" 1.9- 1.8 -3 -
acenapthylene - 3.3 6.2 - -
acenapthene - - 0.25 - '>
fluorene ' - (0.02) 0.61 - -
phenanthrene 0.35 4.7 11. (0.02) (0.06)
anthracene - 0.24 1.5 0.32
pyrene 0.77 6.0 17. 0.25 0.38
fluoranthene 0.96 8.3 21. 0.93 0.63
benzanthracene 0.13 .0.18 4.7 1.0 0.°!
chrysene 0.12 0.28 5.6 1.2 0.17
benzo(a)pyrene , — 1.2 4.4 2.6 -
1,2;5,6 dibenzanthracene - - - - -
1,12 benzoperylene - - - 1.2 -
1 - Also equal to concentration in final 1 mL extract, in vg/mL (ppm).
2 - The calculated amount is given in parenthesis if it is below twice the
stated detection limit in the extract (O.OSppm). Note'that some com-
pounds were detected at concentrations below the stated detection limit.
3 - a dash indicates that the compound was not present above the stated
detection limit.
132
-------
ENGINEERING.SCIENCE
Dr. Bruce Tichenor
February 1, 1983
Page 5.
Table 3. Analytical Results for Polynuclear Aromatics in Filter Samples.
E.S. Sample #
SumX Sample #
Sample weight, g
Compound
napthalene
acenapthylene
acenapthene
fluorene
phenanthrene
anthracene
pyrene
fluoranthene
benzanthracene
chrysene
benzo (a) pyrene
Fl
1767
0.0071
F2
1768
0.0810
PNA concentrations
< I'M
nd
nd
nd
49
nd
110
140
18
1-7
nd
1,2; 5, 6 dibenzanthracene nd
1, 12-benzoperylene
Detection Limit1
nd
7
23
41
nd
<1
58
3.0
74
100
. .2.2
3.5
15
nd
nd
0^6
F3
1769
0.0819
, ug/g
22
76
3.1
7.4
130
18
210
260
57
68
54
nd
nd
0.6
F4
1770
0.0179
soot (ppm)
nd 2
nd
nd
nd
<5
18
14
52
56
67
145
nd
67
3.
F5
1771
0.0183
nd
nd
nd
nd
<5
nd
"i
i **
11
9.3
nd
nd
nd
3.
1 _ Dppcndent on sample size. Corresponds to Q.OSppm in 1 ml> extract. Con-
centrations of substances found but less than twice the detection limit
are reported as less than (<) twice the detection limit.
2 - nd - none detected at or above the stated detection limit.
1H ,
Jj
------- |