EPA-600/2-76-079
March 1976
FLARE
SYSTEMS STUDY
by
M. G. Klett and J. B. Galeski
Lockheed Missiles and Space Co. , Inc.
4800 Bradford Drive
Huntsville, Alabama 35807
Contract No. 68-02-1331, Task 3
ROAP No. 21AXM-030
Program Element No. 1AB015
EPA Task Officer: Max Sam fie Id
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
LMSC-HREC TR D390190
FOREWORD
This study of industrial flare technology was conducted
under Task 3 of Contract EPA-68-02-1331 by personnel of
Lockheed Missiles Sc Space Company, Inc., Huntsville Research
& Engineering Center, Huntsville, Alabama, for the Controls
System Laboratory of the Environmental Protection Agency,
Research Park Triangle, North Carolina. Dr. Max M. Samfield
was the EPA Task Officer. In addition to the authors, Dr.
S.V. Bourgeois participated in the study as Lockheed Project
Manager.
The authors are grateful for the cooperation and time
of the staffs of the equipment manufacturers, flare users and
the Air Pollution Control Districts, who provided much of the
information upon which this study is based.
11
-------
LMSC-HREC TR D390190
TABLE OF CONTENTS
Section Page
FOREWORD ii
1 INTRODUCTION AND SUMMARY 1
1.1 Introduction 1
1.2 Summary 1
1.2.1 Elevated Flares 2
1.2.2 Low-Level Enclosed Flares 2
1.2.3 Auxiliary Equipment 2
1,2.4 Costs 3
1.2.5 Flare Performance and Emissions 3
1.2.6 Proposed Research and Development
Programs 4
II BACKGROUND 5
2,1 Applications of Flaring for Waste Gas
Disposal 5
2.2 Flaring Methods ?
HI COMMERCIALLY AVAILABLE FLARE SYSTEMS 9
3.1 Elevated Flares 9
3.1.1 Flare Tips 9
3.1.2 Gas Traps 12
3.1.3 Pilot and Ignition System 18
3.1.4 The Stack and Its Support 18
3.1.5 Water Seals, Flame Arresters and
Knockout Drums 18
3.2 Ground Flares 22
3.3 Forced Draft Flares 22
3.4 Comparative Costs of Flare Systems 27
IV FLARE DESIGN CRITERIA 29
4.1 Selection of Applicable Flare System 29
4,2 Flamxnability Limits and Flame Stability 30
4.3 Flare Emissions 32
iii
-------
LMSC-HREC TR D390190
TABLE OF CONTENTS (Continued)
Section Page
IV 4,3.1 Thermal Emissions and Luminosity 32
4.3.2 Noise Emission 33
4.3.3 Smoke and Particulate Emissions 36
4.3.4 Chemical Emissions 39
4.3.5 Oxidation 'Products 40
4.3.6 Other Gaseous Emission Sources 41
4.3,7 Dispersion of Chemical Emissions
and Flammable Gases 41
4.3.8 Air Pollution Rules and Regulations
Affecting Flares 51
4.3.9 Flare Emission Factors 53
4.4 Flare Safety 54
4.4.1 Explosion Potential 54
4.4.2 V-apor Purging 55
4.4.3 Molecular Seals 56
4.4.4 Fluidic Seal 56
4.4,5 Explosion Suppression Systems 56
4.4.6 Water Seals and Flame Arresters 56
4.4.7 External Fires and Emissions 58
4.4.8 Knockout Drum Sizing and Design
Criteria 58
4.4.9 Thermal Radiation Hazards 60
V RECOMMENDED DESIGN METHOD 71
5.1 Elevated Flare System 71
5.1.1 Required Design Information 71
5.1.2 Flare Burner Diameter 72
5,1.3 Utility Requirements 75
5.1.4 Flare Height 77
5.1.5 Supporting Structures 79
5.1.6 Auxiliary and Control Components 79
5.1.7 Endothermic Flaring Low Btu Gas
Streams 80
5.2 Low-Level Flare 81
VI SAMPLING AND ANALYSIS TECHNIQUES 82
6.1 Present Sampling Practices and Problems 82
6,2 Analytical Techniques 83
6.2.1 Hydrocarbons 84
6.2.2 Oxidized Hydrocarbons, Carbon
Monoxide, Carbon Dioxide 84
I ₯
-------
LMSC-HREC TR D39019D
TABLE OF CONTENTS (Continued)
Section
VI 6.2.3 Nitrogen Oxides 84
6.2,4 Sulfur Dioxide 87
6.3 Long Path Remote Sensing Techniques 8?
VII FLARE LOADINGS AND EMISSIONS FOR VARIOUS
90
7.1 Questionnaire Format and Circulation 90
1,2 Refinery Questionnaire Results 90
7.3 Impact of Flares on Refinery Emissions 97
7.4 Iron and Steel Mills Questionnaire Results 99
7.5 Impact of Flares on Iron and Steel Mill
Emissions 100
7,6 Manufacturing Chemists Questionnaire
Results 104
7.7 Summary of Flare Loadings 106
VIII RECOMMENDED RESEARCH PROGRAM 107
8.1 Theoretical Analysis of Combustion
Modifications Applicable to Flaring 107
8.1.1 Summary and Objectives 107
8.1.3 Background 107
8,1.3 Validation of the Analytical Model 109
8.1.4 Evaluation of Flare Design Modifications 109
8.1.5 Priority 109
8.2 Evaluation of Remote Sampling Methods 109
8.2.1 Summary and Objectives 109
8.2.2 Background 109
8.2.3 Summary of Remote Sampling
Technology 109
8.2.4 Remote Sampling Field Studies 110
8.2.5 Priority 110
8.3 Application of Flaring to Control of Gaseous
Emissions 110
8.3,1 Summary and Objectives 110
8.3.2 Background 110
8.3.3 Theoretical Analysis 110
8,3.4 Experimental Analysis 113
-------
LMSC-HREC TR D390190
TABLE OF CONTENTS (Continued)
Section Page
VIII 8,3,4 Experimental Analysis 113
8,3,5 Priority 113
8.4 Economic Analysis of Waste Stream Recovery
and Alternate Disposal Methods 113
8.4.1 Summary and Objectives 113
8.4.2 Background 114
8.4.3 Identify Economic Considerations Now
Used to Determine Whether a Given
Flared Stream has Sufficient By-Product
Value for Recovery 114
8.4.4 Identify Alternative Uses of Low Pres-
sure Flammable Hydrocarbon Gases 115
8.4.5 Evaluation of Alternative Disposal
Methods 115
8.4,6 Priority 115
8.5 Emission Factors for Elevated Flare Systems 115
8.5.1 Summary and Objectives 115
8.5.2 Background 115
8.5.3 Site Selection and Evaluation of Sampling
Methods and Hardware 116
8.5.4 Field Testing of Elevated Flare Systems 116
8.5.5 Priority 116
IX REFERENCES 117
VI
-------
LMSC-HREC TR D390190
SECTION I
INTRODUCTION AND SUMMARY
1,1 Introduction
This report presents the results of a study of emissions from flare sys-
tems. Flares are used for the control of gaseous combustible emissions
from stationary sources. The scope of the study includes an evaluation
of existing flare systems, an examination of flare design and sizing cri-
teria, recommended design methods and features, an assessment of pre-
sent emission problems and a recommended research program for flare
emissions control. Information was obtained from the published litera-
ture, equipment manufacturers, equipment users, air pollution control
agencies and universities. Visits were made to many of these sources of
information in order to hold detailed technical discussions about the de-
sign and performance of flare systems.
Flaring is intended primarily as a safety measure for disposing of large
quantities of gases during plant emergencies. Flows are typically inter-
mittent with flow rates of several million cubic feet an hour during major
upsets. Continuous flaring is generally limited to flows not greater than
a few hundred cubic feet an hour. Since flaring is relatively inexpensive,
this technique has been suggested for the control of gaseous combustible
emissions from stationary sources. However, emissions from flares
could also create a potential problem. This study was carried out with
two objectives in mind. One was to determine the potential of flares as
a control system and the second was to assess the emission hazards of
present industrial flares.
Section II of this report explains the different applications of flaring waste
gases. Section III describes the commercially available flare systems and
gives comparative cost data. Section IV discusses flare design criteria
including in some detail the two main problem areas of flare emissions
and safety. Section V presents recommended design methods; Section VII
discusses present flare loadings for various industries and their impact on
emissions; Section VIH contains a recommended flare research program.
1.2 Summary
Commercially available flare systems are of two basic types elevated
and ground flares. Presently, these serve separate functions; elevated
flares are used primarily for disposal of gaseous wastes generated during
plant emergencies such as during power failure, plant fires, component
failure and other overpressure situations in which discharge directly to
the atmosphere could result in explosion hazards. Elevated flares are
therefore used primarily in conjunction with vapor relief collection sys-
tems in large-scale chemical manufacturing or petroleum refining opera-
tions. Other limited applications include venting of storage tanks and
loading platforms.
Although steam, water and air are frequently injected into the elevated
flare burner to reduce smoke and luminosity, expedient vapor disposal
rather than pollution control has been the design emphasis. Recently
-------
LMSC-HREC TR D390190
developed low-level flare systems represent a departure from conven-
tional design. With recent emphasis reducing noise, chemical emissions,
heat and luminosity, low-level flares have become increasing popular
for disposing of routine discharges. These include disposal of flam-
mable gases leaking from process and relief valves, process waste
streams, and excess or off-specification product.
1,2.1 Elevated Flares
Elevated flare systems provide a means for disposal of gaseous waste
streams with an almost unlimited range of flows and a minimal pressure
drop of 0 to 60 inches H^O. As such, elevated flares provide a unique
function which cannot be duplicated by other types of combustion equip-
ment.
Design criteria for elevated flare systems are oriented almost exclusively
toward safe rather than efficient combustion of gaseous wastes. Accord-
ingly, sizing calculations presently available are based on allowable pres-
sure drop (Section 5.1.2) and dispersion of thermal radiation (Section 5.1.4}
or the dispersion of toxic gases when a flare-aut occurs (Section 4.1.7).
Discharge of liquids into the flare system can cause problems, and "knock-
out" or liquid disentratnment drums are required for liquid removal.
1.2.2 Low-Level Enclosed Flares
Low-level flares with enclosed combustion are being used in conjunction
with the elevated flare in response to recent emphasis on pollution. These
are described in detail in Section III. The study indicates that low-level
flares, although relatively expensive to build and maintain, are effective
in reducing noise and thermal emissions,
Relatively little information has been found on sizing and design of low-
level flares. The normal configuration for construction of a low-level
flare involves a steel outer shell, lined with refractory material. The
outer shell serves to conceal the flame and prevent thermal and luminous
radiation. As in other types of combustion equipment, the refractory
also protects the steel shell from direct exposure to the effects of high
temperatures and corrosive materials, and to improve combustion effi-
ciency by minimizing heat losses. Refractory thicknesses typically varies
from about 4 to 8 inches. The refractory used results in a sluggish re-
sponse to abrupt changes in gas flow and adds considerably to the con-
struction and maintenance costs of a low-level flare. Because of the slow
heatup associated with refractory construction, the low-level flare is
normally used only for low or continuous flow rates, with an elevated
flare of conventional design used to accommodate sudden upsets. An ele-
vated flare must be associated with low-level flare applications in most
conventional designs.
1.2.3 Auxiliary Equipment
Auxiliary equipment for the flare system includes igniters, pilots and
safety-oriented equipment described in Sections 3.1, 4,4 and 5.1.6.
-------
LMSC-HREC TR D390190
Knockout drums are normally provided for removal of liquids from the
flared stream. Water seals and, less frequently, flame arresters are
used to isolate the flare stack from the vent collection system. Purge
gas generators and vapor traps serve to prevent the formation of ex-
plosive mixtures within the flare stack. Maintenance of the liquid level
in water seals and disentrainment drums is critical; liquid level control
and alarm systems are available for these systems. Pilot burners are
also frequently equipped with flame detection and alarm systems.
1.2.4 Costs
Capital costs for low-level flares and various types of elevated flares
are given in Section 3.4. This information is based on discussions with
flare vendors and users.
Elevated flare equipment costs vary considerably because of the dis-
proportionate costs for auxiliary and control equipment and the relatively
low cost of the flare stack and burner. As a result, equipment coats are
rarely diameter-dependent. Typical installed costs range from $30,000
to about $100,000. Low-level flares are approximately ten times more
expensive for similar capacity ranges.
Operating costs are determined chiefly by fuel costs for purge gas and
pilot burners, and by steam required for smokeless flaring. Steam and
other requirements are discussed in Sections 5.1.3 and 5.1.7. On the
basis of 30 cents per million Btu's fuel requirements, typical elevated
flare stack operating costs (2-foot-diameter stack) are about $1,500 per
year.
1.2.5 Flare Performance and Emissions
Since flaring has traditionally been used for the safe disposal of gases
discharged under emergency conditions, performance standards relating
to combustion efficiency and gaseous emissions are limited. Probable
air pollutants from elevated flares include CO, unburned hydrocarbons,
aldehydes, and particulates as expected from any combustion process
involving large, turbulent diffusion flames. These emissions result
from flame quenching. Relatively low flame temperatures are typically
observed for both elevated and low-level flares, probably resulting in
low NOX emission factors compared to other types of industrial combus-
tion equipment.
Results of a survey to determine flare loadings and estimated flare
emissions are discussed in Section VII. It was found that the average
yearly emissions from flares constitute just a small fraction, less than
1%, of the average yearly plant* emissions. Total flare emissions over
a year's time therefore probably only have a small impact on total plant
Representative plants include U.S. pertroleurn refineries, iron and steel
mills and chemical manufacturing facilities.
-------
JLMSC-HREC TR D390190
emissions. However, because of the intermittent nature of flaring, the
majority of flare emissions are concentrated into just a few minutes of^
actual flaring. During this time five or more times the normal plant
emissions are released into the atmosphere.
1.2.6 Proposed Research and Development Programs
Programs have been developed to provide technology where deficiencies
exist, to generate the data required to evaluate combustion modifications
and extend the application of flaring to air pollution control.
Since little quantitative performance data were found in this study, field
testing of elevated and enclosed ground level flare systems is recommeneU
Testing should be done to determine the concentration and characteristics
of flare combustion products as well as the mass rate of emissions in
order to evaluate the efficiency of flare systems as a control device.
A combustion research program is recommended to fill the gaps existing
in the technology of large diffusion flames. For this study, construction
of a large scale flare burner and combustion chamber is recommended.
Part of the rationale and incentive for this program is that many industrial
flames are of the turbulent-diffusion-flarne type.
-------
LMSC-HREC TR D390190
SECTION II
BACKGROUND
In many industrial operations, and particularly in chemical plants and
petroleum refineries, large volemes of combustible waste gases are
produced. These gases result from undetected leaks in the operating
equipment, from upset conditions in the normal operation of a plant
where gases must be vented to avoid dangerously high pressures in
operating equipment, from plant start ups and from emergency shut
downs. Large quantities of gases may also result from "off-spec"
product or excess product which cannot be sold. Flows are typically
intermittent with flow rates during major upsets of several million
cubic feet per hour.
The preferred control method for excess gases and vapors is to recover
them in a blowdown recovery system. However, large quantities of gas,
especially those produced during upset and emergency conditions, are
difficult to contain and reprocess. In the past all waste gases were
vented directly into the atmosphere. However, widespread venting
caused safety and environmental problems. In practice, therefore, it
is now customary to collect such gases in a closed flare system and to
burn these gases as they are discharged from an elevated flare stack or
alternately the gases may be discharged and burned at ground level usually
with shielding for the flame.
The flare system is used primarily as a safe method for disposing of
excess waste gases. However, the flare system itself can present addi-
tional safety problems. These include the explosion potential of a flare,
thermal radiation hazards from the flame, and the problem of toxic
asphyxiation during flame-out. Aside from safety there are several other
problems associated with flaring which must be dealt with during the de-
sign and operation of a flare system. These problems fall into the general
area of emissions from flares and include the formation of smoke, the
luminosity of the flame, noise during flaring and the possible emission of
air pollutants during flaring.
2.1 Applications of Flaring for Waste Gas Disposal
There are three main considerations in deciding whether to flare a waste
gas. These are; (1) the variability of the flow of the waste stream, (2) the
expected maximum volume of the stream to be flared, and (3) the heat con-
tent of the waste stream.
A high variability of flow of the waste stream is probably the most im-
portant factor. A flare is designed to operate for practically an infinite
"turndown" range of flows. Alternate waste gas disposal systems such as
incinerators or afterburners need an adequate control on the flow of waste
gases and can only be used for continuous or at least fairly continuous gas
flows.
-------
LMSC-HREC TR D390190
The volume of the waste stream to be disposed is also an important
factor. With very large volumes of gas, direct flame combustion by
incineration or a flame afterburner device becomes impractical due to
the size of equipment needed. However, capacity for an elevated flare
can be increased easily by increasing the diameter of the stack. A
typical small flare with a four-inch diameter stack has a capacity of
30,000 scfh. A normal refinery flare with a capacity of 5,000,000 scfh
would need only a 36-inch diameter flare stack.
The heat content of a waste gas falls into two classes. The gases can
either maintain their own combustion or they cannot maintain their own
combustion. In general, a waste gas with a heating value greater than
200 Btu/ft^ can be flared successfully. The heating value is based on
the lower heating value of the waste gas at the flare. Below 200 Btu/ft
enriching the waste gas by injecting a gas with a high heating value may
be necessary. The addition of such a rich gas is called endothermic
flaring. Gases with a heating value as low as 60 Btu/ft^ have been flared
but at a significant fuel demand (Ref. 1). ItJ.s usually not feasible to flare
a gas with a heating value below 100 Btu/ft (Ref. 2). If the flow of low
BTU gas is continuous, incineration can be used to dispose of the gas.
For intermittent flows, endothermic flaring is the only possibility.
Flares are well suited for disposing of intermittent flows of large and
small volumes of waste gases that have an adequate heat value to sustain
combustion. For intermittent flows of low heating value waste gases,
additional fuel must be added to the waste stream in order to flare. Since
the value of the additional fuel can become considerable and is completely
lost during flaring, endothermic flaring can become expensive. However,
if intermittent flows of low heat waste gases are in large volumes, the
only practical alternative to flaring is to vent the gases directly to the
atmosphere. This is usually unacceptable for environmental reasons.
Most flares are used to dispose of the intermittent flow of waste gases.
There are some continuous flares but they are used generally for small
volumes of gases on the order of 500 cfm or less. The heating value of
larger continuous flows of a high heat waste stream is usually too valuable
to waste in a flare. Vapor recovery or the use of the vapor as fuel in a
process heater is preferred over flaring. For large continuous flows of
a low heating value gas, auxiliary fuel must be added to the gas in order
to flare. It is much more efficient to burn the gas in an enclosed inciner
ator rather than in the flame of a flare. For small continuous flow of
gases, flares are sometimes used even though fuel or heat is either lost
or wasted. In these cases the equipment costs are sometimes more im-
portant than fuel savings and a flare is more economical to use.
Flares are mostly used for the disposal of hydrocarbons. Waste gases
composed of natural gas, propane, ethylene, propylene, butadiene and
butane probably constitute over 95% of the material flared. Flares have
been used successfully to control malodorous gases such as mercaptans
and amines (Ref. 3). However, care must be taken when flaring these
gases. Unless the flare is very efficient and gives good combustion,
obnoxious fumes can escape unburned and cause a nuisance (Ref. 4),
-------
LMSC-HREC TR D390190
Flaring of hydrogen sulfide should be avoided because of its toxicity and
low odor threshold. In addition, burning relatively small amounts of
hydrogen sulfide can create enough sulfur dioxide to cause crop damage
or local nuisance (Ref, 5). In recent years gases whose combustion
products may cause problems, such as those containing hydrogen sulfide
or chlorinated hydrocarbons, have not been recommended for flaring,
2.2 Flaring Methods
The elevated flare is the most common type of flare system in use today.
In this flare, gas is discharged without substantial premixing, and ignited
and burned at the point of discharge. Combustion of the discharged gases
takes place in the ambient atmospheric air by means of a diffusion flame.
This type of combustion often results in an insufficient supply of air and
thus a smoky flame. A smokeless flame can be obtained when an adequate
amount of combustion air is mixed sufficiently with the gas so that it burns
completely. Smokeless burning is usually accomplished by injecting steam
into the flame. The modern elevated flare allows large volumes of waste
gases to be burned safely and inexpensively. However, the elevated flare
can also present other emission problems including the emission of noise,
light and chemical air pollutants into the atmosphere.
A second type of flare often found is the ground flare. A ground flare
consists of a burner and auxiliaries located at or near ground level. The
burner may be with or without shielding but It must allow for the free
escape of the flame and combustion products. Ground flares have the
advantage of being able to have the flame shielded. Compared to elevated
flares they either require more land if unshielded or the burners, controls
and shielding may be more expensive than a stack. Also if the ignition or
pilot system fails, the ground flare cannot disperse the gases as well as
an elevated flare.
A third system which has been recently developed and is being employed
more frequently, particularly where noise luminosity and smoke formation
are severely criticized by local residents, is an enclosed "low-level" ground
flare used in conjunction with an elevated flare. In more than 90% of the
flare occurrences the load to the flare is less than 10% of design, capacity
of a flare stack (Ref. 2). The "low-level" ground flare is designed to handle
most of the flare occurrences, and the remaining large releases use both
systems. This system, called an integrated flare, although expensive can
greatly reduce smoke, noise and light emissions that cause complaints from
local residents.
Forced draft flaring, where combustion air is mechanically blown to pre-
mix with the gas before igniting, is ideal as far as combustion is con-
cerned. This type of flare achieves smokeless burning without the use of
steam injection. However, this method has a limited turndown ratio and
requires a much larger flare stack for the added combustion air. While
this approach has been utilized for some special applications, mostly in
places where smokeless burning is required but steam is not available,
it has generally been found uneconomical for most uses.
-------
LMSC-HREC TR D390190
The use of air-inspirating burners for premixed air has also been
attempted with flares. This type of operation requires the gas to be
supplied at substantially constant rate and pressure of the order of 1 to
4 psig. In many cases such pressure cannot be made available because
limitations of the vent gas collecting system. For air-inspirating instal-
lations it is also generally necessary to provide a numberof burners of
different capacities to handle the wide range of venting rates normally
encountered. Flare systems baaed on this principle have been largely
unsuccessful.
Usually, if there is a continuous flow of gas, a vapor recovery system
is considered. While the collection, storage, and return of gas is ex-
pensive, the continuous wasting of gas may be much more expensive.
The capital expenditures to store and recompress immense volumes
released intermittently and irregularly usually exceeds the operating
expense of flaring the gas. Many plants are now using their flare sys-
tem in conjunction with a vapor recovery system. They have a triad
system for control of waste gases which consists of a vapor recovery
system, a low-level flare for most of the flare occurrences which over-
load the vapor recovery system and an elevated flare for large releases
which overload the low-level flare,
Horton et al., (Ref. 6) have discussed what they feel is the future answer
to reducing the possible load to a flare. The nuclear power industry has
installed highly reliable instrumented systems to eliminate the need for
relief valves and still protect a system from overpressure (Ref. 7). How-
ever, these systems have not achieved wide use in the chemical or
petroleum industry.
The real source of most pressure in gas-liquid systems is heat. Fired
heaters and heat exchangers create large volumes of gas which must be
relieved, A highly reliable means for automatically cutting off heat,
when the pressure reaches a specified value, would decrease or eliminate
the need for a safety relief valve. It would therefore decrease the quantity
of gas sent to the flare. Reliability is usually assured by independent and
redundant instrumentation (Ref. 7).
The high integrity protection system can never totally eliminate all safety
relief valves in a plant and thus the need for a flare. However, the load
to the flare would be greatly reduced with the flare being used only in
major emergency situations.
-------
LMSC-HREC TR D390190
SECTION III
COMMERCIALLY AVAILABLE FLARE SYSTEMS
In general there are three types of flare systems in use today, the ele-
vated, ground and forced draft flare. This section will describe the
equipment available for flaring waste gases by these systems and will
also present relative cost data for the different systems.
3.1 Elevated Flares
The modern elevated flare system is made up of several components
including the flare tip, some type of gas trap directly below the tip, a
pilot and ignition system at the top of the flare tip, and the stack and
its support. When smokeless burning is required, a steam injection
system must also be provided at the top of the flare. Water seals and
knockout drums are also usually required for safety reasons. Figure
3-1 shows a schematic of a typical elevated flare system.
3.1.1 Flare Tips
A flare tip must be capable of operating over a wide range of turndown
ratios. To achieve this, the flare must have excellent flame holding
ability and mixing characteristics. Flameholding is ensured by pro-
viding multiple continuous pilots around the combustion tip and by pro-
viding a flame stabilization ring on the combustion tip. Figure 3-2 shows
the standard flare tips available from John Zink Company. The flare tip
is usually made of stainless steel or some other high temperature and
corrosion-resistant alloy.
Smokeless burning can be achieved with special flare tips which inject
water, natural gas or steam into the flame thereby increasing air-gas
mixing to ensure complete combustion. Water injection has many dis-
advantages including ice formation in the winter, a mist in the summer,
the tremendous pressure head needed for an elevated flare and a turn-
down ratio much less than steam, making control very difficult with the
possibility of quenching the flame. Natural gas has also been used to
inject into the flame for smokeless burning but only in the case where the
gas itself has no value since it is also burned during flaring. For these
reasons steam is the most common utility used for smokeless burning.
There are two basic steam injection techniques used in elevated flares.
In one method steam is injected from nozzles on an external ring around
the top of the tip. In the second method the steam is injected by a single
nozzle located concentrically within the burner tip. Vendors use various
types of nozzles to create a circular, swirl, fan, jet or Coanda effect.
In recent years environmental regulations have required flares to be
smokeless for large turndown ratios. To ensure satisfactory operation
under varied flow conditions, the two types of steam injection have been
combined into one tip. T*he internal nozzle provides steam at low flow
-------
LMSC-HREC TR D390190
Flare Burner and Location
of Fluidic Seal
Gas Trap
Riser Sections
Entry, Disentrainment
or Water Seal
Ladders and Platforms
Fig, 3-1 - Integrated Flare Stack Components
10
-------
LMSC-HREC TRD390190
Utility Field Flare Tip
Flame Retention Ring
'Pilot Assembly
Endothermic Field Flare Tip
Smokeless Field Flare Tip
Endothermic
Assist Gas
Supply
Steam Distribution Ring
Fig. 3-2 - Flare Tips from John Zink Company
1 1
-------
LMSC-HREC TR D390190
rates while the external jets are available at large flow rates. Figure
3-3 shows a schematic of National AirOil flare tips illustrating the
different steam injection methods.
While these are the most common types of tips, there are several other
mainly special purpose tips commercially available. A further modifica-
tion of the steam injection tip is shown in Fig. 3-4. Here, an internal
nozzle is used to inject both steam and air into the tip. The major dis-
advantage of this system is that a larger tip is needed because of the
increased pressure drop. Under some circumstances, the gases may
actually burn inside the tip. Figure 3-5 shows a tip using a Coanda effect
of steam injection to achieve the required air gas mixture. While this
method provides efficient mixing, the burning of the gas takes place
inside the flare tip instead of outside or above as with the other tips.
Burning inside the tip can drastically shorten the life of the tip. Figure
3-6 shows National AirOil's jet mix vortex tip. These can be used with
relatively high pressure waste gases with little or no steam needed for
smokless operations. Figure 3-7 shows the special purpose Indair flare
tip which burns gases smokelessly without steam. It has limited use
since it requires both high pressures and low pressure gas in the ratio
of about three to bne. Also its maximum turndown ratio is only about
two. Other special purpose tips are available including endothermic tips
that inject gas to raise the heat value of the waste stream and tips with
added muffling for quiter flaring.
The rate of steam injection to the flare tip can be controlled manually or
automatically. While automatic control is usually not mandatory, it is
preferred because it reduces steam usage, greatly reduces the amount
of smoking and minimizes noise. Automatic systems use flow measure-
ment devices with ratio control on steam. Since the flow rate measure-
ment cannot include the variables of degree of saturation and molecular
weight, the ratio control is usually set for some average hydrocarbon
composition. It is usually necessary to have a fixed quantity of steam
flowing at all times to cool the distribution nozzles at the tip.
3.1.2 Gas Traps
To prevent air migration into the flare stack as a result of wind effects
or density difference between air and flare gas, a continuous purge gas
flow through the flare system is maintained. The system can be purged
with natural gas, processed gas, inert gas or nitrogen. To reduce the
amount of purge gas requirement and to keep air out of a flare system,
gas trap devices are normally located in the stack directly under the
flare tip. One type of gas trap commercially available is the molecular
seal (Fig, 3-8). This type trap may not prevent air from getting in the
stack as a result of gas cooling in the flare headers. Instrumentation
systems are available to automatically increase the purge rate to prevent
air from entering the stack during rapid gas cooling. A new development
in gas traps is National AirOil's Fluidic Seal (Fig. 3-9). This seal weighs
much less than a molecular seal and thus can be placed much closer to the
flare tip.
12
-------
TRD390130
MMC «nd
IMMK «r |M
Ilk FtoMIc 5
MW nrvj wiM!
i I
f?>
V
^
u
H 5
i it J
^
JAO 48-Inch Ring Center for
Fig. 3-3 - Flare Tips Illustrating Ring and Center Steam Injection Units
(from National AirOil)
13
-------
LMSC-HREC TR D390130
nun
INC
Plan
Elevation
Fig. 3-4 - Detail of Internal Steam Injection System from John Zink
Company
Fig, 3-5 - Coanda-Type Flare Tip from Flargas Engineering, Ltd.
14
-------
LMSC-HREC TR D390130
Fig. 3-6 - Jet Mix Vortex Flare Tip with Steam Assist
(from National AirOil)
Bimt fun}
Gos/eir mixtvrt
17ft
Air
CCI
Fig. 3-7 - The Indair Flare Tip (from Oil and Gas Journal)
15
-------
LMSC-HREC TR D390V)'
Outlet to Flare Burner
Outlet to Flare Burner
|L_
Clean-Out
rr
/
i ._
UF
«E=
f*
1
i
U
BgB
U
i
**-v^
3IB=
r
|i
r
*
as
_ i,
\
1 i
f
L
^
4
r
1
;
^
Inlet from Flare Riser
National Air Oil NDS Double Seal
(Patent applied for)
Inlet from.Flare Riser
John Zink Molecular Seal
(U.S. 3,055,417)
Fig. 3-8 - Air Reentry Seals
16
-------
LMSC-HREC TR D39019G
Upp*r x
Section |/
Lower
Section
N
*
Entering Air
f!
f
f*
\
Purge
Flow
Velocity Gradient
of Waste Gas
3 Flare Riser
Flare Burner with Seal Baffles
Velocity Profiles
Fig, 3-9 - National Air Oil Flmidic Seal
17
-------
LMSC-HREC TR D390U
3.1.3 Pilot and Ignition System
The ignition mechanism for a flare installation usually consists of the
pilot burners and the pilot burner igniters. The pilot burners serve to
ignite the outflowing gases and to keep the gas burning. These pilots
must provide a stable flame to ignite the flare gases, and in many cases
to keep them burning. To accomplish this more than one and usually
three or four pilot burners are always used. The pilot burners are alsc
sometimes provided with separate wind shields as shown in Fig. 3-10.
A separate system must be provided for the ignition of the pilot burner
to safeguard against flare failure. The usual method used is to ignite
a gas/air mixture in an ignition chamber by a spark. The flame front
travels through an igniter tube to the pilot burner at the top of the flare.
This system permits the igniter to be set up at a safe distance from the
flare, up to 100 feet, and still ignite the pilots satisfactorily. Figure 3-
shows one arrangement for the ignition system. The whole device is
mounted on an ignition panel and set up in an accessible spot on the grot
The ignition panel must be explosion proof, have an unlimited life, and
insensitive to all weather conditions. On elevated flares, the pilot flarr
is usually not visible and an alarm system to indicate pilot flame failuri
is desirable. This is usually done by a thermocouple in the pilot burne:
flame. In the event of flame failure, the temperature drops and an alai
sounds.
3.1.4 The Stack and Its Support
Figure 3-12 shows the methods used to support the complete flare towe
These towers must be provided with a climbing ladder with a cage and
landing on top for repair and maintenance purposes. These towers for
refineries can range from 200 to 400 feet high. Flare towers with a
proportion of length-to-diameter ratio less than 30 are usually con-
structed as self-supporting stacks; towers with a proportion L./D < 100
are supported with a set of guys, and when the proportion is L/D > 100,
the towers are made with two or more sets of guys (Ref. 2). Self-
supporting stacks are usually not built over 50 feet high because of the
large and expensive foundation required (Ref. 4).
The guys need a large area for high stacks; that is why it is often pre-
ferred to build steel supports to which the stack is fastened. These ar>
usually steel framework with a square cross section widened at the bas
A triangular cross section, adopted from the modern television antenna
is more economical and has been used in several refineries (Ref. 8). Tl
flare stack will expand because of the hot gas flow, and the supporting
structure must be able to accommodate this expansion.
3.1.5 Water Seals, Flame Arresters and Knockout Drums
Water seals and flame arrestors are used to prevent a flame front fron
entering the flare system. Flame arrestors have a tendency to plug an
obstruct flow and are not capable of stopping a flame front in mixtures
air with hydrogen, acetylene, ethylene oxide and carbon disulfide; thus
they are of little value (Ref. 1).
18
-------
LMSC-HREC TRD390190
Igniter
Inlet
Shielded Pilot Nozzle
2 in. Pilot Tube
I in. Igaiter Tube
Inspirator
Air Adjuster
Thermocouple
Explosion-Proof-
Weather-Proof
Junction Box
Gas Inlet
Air Inlet
Fig, 3-10 - Flare Pilot Burner System
19
-------
LMSC-HREC TR D390190
Description
(T) Mounting Plate - 18 x 36 in.
(2) Air Control Valve (1/2 in.)
(3) Gas Control Valve (1/2 in.)
\4J Gas Pressure Gage
(^) Air Pressure Gage
(6) Spark Sight Port
(7) Spark Plug
(₯) Explosionproof Button (Push)
(9) Transformer in Explosion-Proof-
Weather-Proof Housing
@ Three-Way Valves
NOTE; Quantity of Item 10 will vary with number of pilots on flare.
1/2" Electrlea1 Connection
Pilot Ignitor Outlet
Pi lot Ignj tor Out Iet
Pllo.t Ignttor Outlet
Igniter Outlet ffl
Six (6) 5/8" Dta. Mounting
TT Holes
Fig, 3-11 - Flare Ignition System from National Air Oil
20
-------
LMSC-HREC TR D390190
a. Self-Supporting
b. Flare with Support Tower
c. Flare with Guys
Fig. 3-12 - Flare Stack Supports
21
-------
LMSC-HREC TR D390
Water seals are used to prevent a flame front and air from entering th^
flare gas collection system. The weight of the water seal causes it to 4
be located at or near grade and therefore the seal cannot be used to pre
vent air from entering the stack.
Knockout drums are located at or near the base of elevated flares to sej
arate liquid from gases being burned. If the large liquid droplets are n
removed, they could burn all the way to the ground. Designed for gase
flare lines can contain liquids from liquid expansion reliefs, liquid can
over from gas reliefs, and condensed vapors. The knockout drum is ui
to remove these liquids before the gases are flared. Water seals and
knockout drums are found on most flare systems for safety reasons.
3,2 Ground Flares
A ground flare consists of a burner and auxilaries, such as, a seal, pi.
burner and igniter. Two types are found. One consists of conventional
burners discharging horizontally with no enclosures. This flare must
installed in a large open area for safe operation and fire protection. II
the ignition system fails this is not as capable in dispersing the gases j
an elevated flare. For these reasons this type of ground flare has foun
only limited applications.
Ground flares may also consist of multiple burners enclosed within a r
fractory shell as in the recently developed "low level" flares (Figs. 3-'.
and 3-14). The essential purpose of a low level flare is complete con-
cealment of the flare flame as well as smokeless burning at a low noisi
level. The flared gases are connected by a manifold to a series of bur
heads which discharge the gas into a refractory enclosure. Mixing oft
gas and air is accomplished by a series of multi-jet nozzles. Combust
air is provided by the natural draft of the enclosure. Smokeless burnii
is obtained with little or no steam because of the turbulence and tempe
ture of the burning zone due to the natural draft and the enclosure. Th
size of the enclosure depends upon the capacity of the flare but can be-
come quite large. An enclosed ground flare with a capacity of 25,000
Ib/hr has an enclosure 100 feet high and 20 feet in diameter (Ref. 9). T
same capacity could be handled by an 8-inch diameter elevated flare.
The initial costs of an enclosed ground flare usually limits their capaci
to just a portion of a plants emergency dump rates. However, the groi
flare can be designed to handle most flare occurrences and the remaini
large releases can be diverted to an elevated flare. Figure 3-15 is a
schematic showing how such a system might work. This type of inte-
grated flare system is now becoming common especially in populated
areas.
3.3 Forced Draft Flares
The forced draft flare uses air provided by a blower to supply primary
air and turbulence necessary to provide smokeless burning of relief
gases without the use of steam. Figure 3-16 shows two common desigr
22
-------
LMSC-HREC TRD39Q19Q
(5)
oU/o
131
o o o o
ofo
o o
o|ooj|o
0^0
o o
o||0
Description
Patented Jet Mix Tips
Flare Gas Risers
Flare Gas Header(s)
Flare Gas Connection's)
Combustion Chamber
Refractory Lining and
Anchors
Safety Fence (Collapsible)
NSFP (Pilots with Igniter
Tubes
Sight Ports
Fig. 3-13 - Ground Flare (from National AirOil Burner Company)
23
-------
LMSC-HREC TRD3901*
Thermal Oxidizer Flare
Fig. 3-14 - Ground Flare ZTOF from John Zink Company
24
-------
LMSC-HREC TR D390190
Pilot Gas
Line
\
Flare Gasx
Line
/A
Elevated Flare Burner
Diversion
Seals
Control System
Fig. 3-15 - Ground Flare and Elevated Flare Connected by a Double
Stage Water Seal
25
-------
LMSC-HREC TR D3901
Combustion
Air Inlet
Flare Gas
Inlet
-UJ
ih
1
Cornbu
Air Inl
a. Biaxial Forced Draft Unit
b. Coaxial Forced Draft Unit
Fig. 3-16 - Two Designs for Forced Draft Flare Systems
26
-------
LMSC-HREC TR D390190
of forced draft flares. This type of flare combines smokeless burning with
low operating cost and reliability because only pilot gas and electricity are
required. The flame is also stiffer and, because of the forced draft, is less
affected by the wind.
However, this flare also has a high initial cost. The cost can run two
to three times the cost of a conventional flare, mainly since two stacks
are necessary to keep the air and gas separated until they are mixed and
ignited at the tip. A blower flare should have an automatic air turndown
device to prevent excess air from quenching the flame and creating smoke
if the flare gas rate is reduced. Variable speed blowers or baffles coupled
to flow sensing devices have been used on these flares to extend their turn-
down ratio. Because of costs and turndown ratio limitations, this flare
has been used mostly in special applications. It has been used mainly to
provide smokeless burning where steam is not available. It has also been
used in tankage transfer and venting and in conjunction with a smoking
elevated flare to provide smokeless burning for day-to-day flaring.
3.4 Comparative Costs of Flare Systems
The capital and operating costs for a given flare system depend on many
factors such as the availability of steam, the size of the flare, the com-
position of the waste gas and the frequency of flaring. Each installation
is a special problem, the economics of which must be solved for the spe-
cific case.
Vanderlinde (Ref. 9) estimated the relative cost of equipment used in the
smokeless flare systems. Equipment costs include a guyed stack, ignition
piping, pilot piping, the burner ring and accessories. As shown in Table
3-1 he found that the relative cost of smokeless flare systems was not
stack diameter dependent. On the other hand, relative cost of the equip-
ment for a forced air system is diameter dependent, because a stack
Table 3-1
RELATIVE COSTS OF FLARE SYSTEMS
Type of Flare Equipment Costs
12-in. Diam. 24-in. Diam.
Smoking
Standard Tip 1.00 1.00
Smokeless
Steam Tip 1.25 1.25
Gas Tip 1.30 1.30
Water Tip 1.20 1.20
Forced Draft 2.80 3.38
2?
-------
LMSC-HREC TR D390
within a stack is actually being purchased. Low level enclosed flares
with an equivalent capacity of an elevated flare can be as much as ten
times more costly (Ref. 10). For this reason the enclosed flare is onl
designed to handle the smaller day-to-day flare occurrences.
Typical costs for the flare system of a 350,000 bbl/day refinery would
be of the order of $750,000. This cost includes $500,000 for equipmen
for two elevated and one enclosed low level flare. Of the $500,000 for
equipment, $300,000 would be for the low level flare. Another $250,00
would be needed for the waste gas collection system (Ref. 11).
28
-------
LMSC-HREC TR D390190
SECTION IV
FLARE DESIGN CRITERIA
The complete design specification of a given flare system for use in
safety relief is highly specialized and requires close cooperation be-
tween the buyer and manufacturer. In addition, some factors affecting
design are determined by the type of equipment used; in these cases in
which the equipment is proprietary, design information is not readily
available. Nevertheless, a number of design guidelines have been
published in recent years which serve as general guidelines for equip-
ment sizing and estimation of plant space requirements. These are
given as Refs. 2, 4, 5, 12, 13 and 14.
The objective of this section is to examine the available design and sizing
criteria in order to describe the state of the art of flare design. Emphasis
is placed upon calculations which affect emissions of heat, light, noise,
smoke, particulates and chemicals and the dispersion of gases and par-
ticulates. Auxiliary equipment such as drums, seals and flame arresters
are also discussed in this section.
As noted previously, flaring is intended primarily as a safety measure
for disposing of large quantities of gases primarily during plant emer-
gencies such as fires, electrical failure, failure of cooling water supplies
and other utilities, equipment overpressure, compressor failure, or
problems which may be encountered during start-up. Less frequent
applications during which large quantities of gas may be sent to flare
can include the disposal of "off-spec" product and excess product which
cannot be stored. Flows are typically intermittent with very large flow
rates during major upsets in the range of several hundred thousand pounds
per hour. Flare systems are therefore required to accommodate a very
large "turndown" range of flows. Total capacity and turndown range are
normally the-deciding factors in selecting the applicable flare system.
The type of flare used will depend to a lesser extent upon the type of
materials being sent to flare, the flare location and available utilities.
4.1 Selection of Applicable Flare System
In general, flare systems are divided into two broad categories, ground
flares and elevated flares which discharge the waste stream at some
distance above ground level. Ground flares may consist either of con-
ventional flare burners discharging horizontally at or near ground level
or of distributed burners enclosed within a refractory shell, as in the
more recently developed "low-level" flares. Low-level flares have a
relatively large diameter which reduces discharge velocity and, thus,
sonic emissions. Enclosing the flame reduces light and thermal emissions.
Air for the low-level flare is normally provided by natural draft; for this
reason, and because of the time required to heat the refractory, the low level
flare design has a more sluggish response to sudden upsets than elevated
flares. Low-level flares are normally used for minor upsets or for small,
steady state flows with an elevated flare of conventional design used to
29
-------
LMSC-HREC TR D390190
accommodate full-scale emergency upsets. Horizontal discharge flares
are essentially elevated flare systems discharging at ground level and
have a somewhat limited application because of the large open area of
a minimum of 1500 ft^ required for safe operation. Heat and sound
emissions and other reasons for this requirement will be discussed later
in this section. Flares discharging at ground level are generally con-
sidered unsuitable for flaring gases which may be odorous, noxious, or
toxic in nature or for flaring gases which may produce compounds having
these properties as intermediates or final combustion products.
For general purposes, in which a variety of flow ranges and compositions
may be encountered, the elevated flare is more common. Elevated flares
(and elevated flare burners discharging at ground level) provide air for
combustion either by forced draft or by diffusion of air into the fuel be-
yond the point of ignition and discharge to the atmosphere. Burning the
waste stream by means of natural convection (as in a ground flare) or by
forced convection results in a premixed flame, while burning without
added air results in a diffusion flame.
Typically, elevated flares used for large waste systems are diffusion
burning with steam added to reduce smoking. The application of forced
draft flares is limited to smaller, steady flows such as in tankage transfer
storage tanks, and for use in plant facilities where steam is not available.
Typical maximum flare capacity ranges are
Type Capacity (1000 Ib/hr)
Low Level Flare 80 - 100
Elevated, Diffusion Flame 1000 - 2000
Elevated, Forced Draft 100
The maximum capacity ranges were obtained from conversations with
flare vendors and should be used as a guideline only. Actual capacity
will vary somewhat with the type of gas being flared and other require-
ments .
A number of specialized flare burner designs are also available to
accommodate high pressure side streams. Endothermic flares are also
available to support combustion of gases which are too lean or have too
little heat content to support a flame. Endothermic flaring may be
accomplished using either auxiliary heaters or an "assist" fuel gas.
4.2 Flammability Limits and Flame Stability
Whether or not a given waste stream will support a flame is normally
determined experimentally, but methods are available for estimating
flammability limits (Ref. 1). In some cases, flammable mixtures may
not release sufficient combustion heat to maintain the flame at a stable
temperature. The lower (net) heating value required to support a flame
30
-------
LMSC-HREC TR D390190
varies somewhat with flare burner design; larger flames require a
higher heating value fuel than would be required for combustion in a
distributed burner. A lower heating value of 200 - 250 Btu/scf is
normally considered adequate for flaring in large elevated flares.
Heating values for gases normally flared may be calculated using stan-
dard methods or obtained from furnace handbooks such as Ref. 1. Endo-
thermic flare systems with auxiliary heaters or assist gas addition to
increase heat content may sometimes be used in flaring low heating
value gases.
Flame instability may occur when the discharge velocity exceeds or falls
below the burning velocity. In the case of either premixed or diffusion
flames, an instability may occur when the discharge velocity exceeds the
flame velocity leading to a lifted flame in which mixing of the fuel and
dilution with air must precede the re-ignition of the flame. This condi-
tion is known as "blowoff" (Ref. 12). The flame itself may even blow out
if the discharge velocity greatly exceeds the flame velocity. The opposite
condition in which the gas velocity falls below the burning velocity results
in a condition known as "flashback."
Maximum discharge velocity, and therefore flare burner diameter is fixed
between these upper and lower limits of "blowoff" and "flashback" by the
burning rate of the fuel. In practice, in order to minimize capital costs and
increase the flare throughput, most flares are designed for maximum through-
put based on the maximum allowable pressure drop. Flame holders are
used to maintain flame stability and extend these stability limits. These
are of proprietary design, typically consisting of a perforated ring at
the circumference of the flare tip. The gas flow is divided by the ring
into small streams thereby increasing air-gas mixing in a portion of
the gas stream (Ref. 9). Large pilot flames can also be used to stabilize
the flame. Small amounts of gas having a relatively high burning rate,
such as hydrogen, may be added to the flared stream in order to widen
the stability limits (Ref. 12). The instability at the lower velocity limit
can be avoided by the use of a purge gas which may be either a flammable
or inert gas. The low flow instability is not a problem when vapor purging
is employed, for safety reasons, to prevent the formation of flammable
mixtures in the flare stack at low or no flow. Vapor purging is discussed
further in Section 4.4.2.
Flare diameters are normally sized, within the maximum allowable pres-
sure drop, to provide vapor velocities at maximum throughput of about 20%
of the sonic velocity in the gas (Refs. 12 through 14). There is evidence
that flame stability can be maintained at Mach numbers up to 0.5
(Ref. 12),
Exact analysis of flame stability appears to be beyond the state of the art
for flare design. It is doubtful whether a model exists for turbulent flames
which is satisfactory for estimation of the burning velocity. It has been
determined (Ref. 15) that the burping rate is several orders of magnitude
lower than theoretical even for highly efficient combustion equipment
31
-------
LMSC-HREC TR D390190
such as gas-turbine combustors. It is probable that mixing controls
the burning velocity in flare systems. Recent flare tip designs for
smokeless burning have included tangential discharge of either the flare
stream or steam to stabilize the flame at high discharge velocities, but
such developments appear to be based on empirical observation rather
than analysis.
4.3 Flare Emissions
Flare emissions include chemicals and particulates, thermal and visible
radiation and noise. It is the purpose of this section to discuss the
probable causes of emissions, the state of the art in quantifying and con-
trolling these emissions, and the extent to which flare design has been
affected.
4.3.1 Thermal Emissions and Luminosity
Emission of heat from flares will be discussed in detail in Section 4.4.9.
As in the case of thermal radiation, it is probable that most of the visible
radiation is the result of radiation from hot carbon particles. Electronic
transitions, such as in the formation and recombination of certain radicals;
CH, C_, HCO, NH, and NH£ are also accompanied by emission in the
visible and near ultraviolet, but probably contributes only a small fraction
of the total luminous radiation (Ref, 16). The distribution of radiation fre-
quencies from hot carbon particles is predicted from Planck's radiation
law and requires a knowledge of the flame temperature. For practical
use, a close approximation is given by Wien's law (Ref. 16) for XT < 0.2
cm-deg:
-C.
XT
I. = 2E,AC.X~5e ^ dX (4.1)
A A I
where
A = radiation wavelength, cm
L = radiation intensity between X and X -f dX, W/cm
(per unit surface of the emitter)
Ev = the emissivity at X (for blackbody radiation,
E^ = 1 for all values of X)
A 2
A = the surface area of the emitter, cm
T = absolute temperature, K
C. = first radiation constant
= 0.588 x 10"12 W/cm2
C? = second radiation constant
= 1.438 cm-°K
The radiation maximum calculated from Wien's law allows an estimation
'>>£ the tempesature dependence of the fraction of visible light emitted:
-------
LMSC-HREC TR D390190
X T = 0.289 cm-°K (4.2)
thus, the maximum wavelength depends strongly on temperature. Since
the intensity at this wavelength is directly proportional to area, it follows
that control of the emission of visible light is closely related to the con-
centration and surface area, of participates and the flame temperture.
For hotter flames, the radiation is shifted toward the visible portion of
the spectrum. In flaring practice therefore, injection of steam to reduce
carbon formation decreases both the flame temperature and the area for
emissions and therefore the emission of visible light. Increasing the
steam beyond the amount needed to prevent soot formation causes a further
reduction in luminosity (Ref. 12). Smokeless flaring achieved by pre-mix
burning or multijet burning should result in a higher flame temperature
and a higher luminosity than would be observed during steam injection.
No design modification has been developed which will completely eliminate
luminosity, and in practice the tendency in populated areas has been to
enclose the flame at ground level. This requires a special type of ground
flare and has several disadvantages and limitations. Such flares are
essentially ground level distributed burners (to reduce flame height) en-
closed within a refractory shield to reduce thermal and light emissions.
Air is supplied by a natural draft, therefore turndown is limited and an
initial time lag between initial fuel firing and air supply is inevitable
(Ref. 17). Capital costs for these units are higher than those for con-
ventional flares of the same capacity by about a factor of 10, and main-
tenance costs are also higher. Because of the relatively low discharge
height, such flares are not suitable for flaring toxic or hazardous gases.
Because of the limited turndown and inability to respond to sudden flow
changes, low-level flares are more suitable for flaring when normal flows
are continuous. Elevated flares are recommended for use in addition to
the ground flare whenever protection against sudden upsets is required.
4.3.2 Noise Emission
Sonic emissions from flares consist of contributions from high frequency
jet noise and combustion noise which is of relatively low frequency (Refs.
9 and 18). Jet noise is caused by a fluid passing through a constriction
and is directly proportional to the pressure drop (Ref. 9) or (equivalently)
roughly proportional to the square of the mass flow rate through a nozzle
of fixed diameter (Refs. 18 and 19) according to the behavior expected for
highly turbulent flow. The intensity of jet noise is also a function of the
fluid properties. Combustion noise is a function of flame turbulence and
is directly proportional to the amount of air mixed with the flare gas
(Ref. 9).
Jet noise in flare systems results mostly from high pressure steam in-
jection to achieve smokeless flaring, and this is the major source of the
noise problem. The major steps taken to curb high frequency noise
emission have involved re-designing steam injectors to reduce the steam
exit velocity and the use of peripheral mufflers (shrouds) to prevent both
33
-------
LMSC-HREC TR D390190
the direct sound radiation and reflection from the flare stack (Refs. 19
and 20). A multiport nozzle design reported by Chevron (Ref. 19} re-
sulted in a 96% reduction (by 14 decibels) in the sound power radiated to
the steam-air injection system. The major reduction was in the range
1000 to 2000 Hz with little reduction of low frequency combustion roar
(290 HE <). A Coanda-principle injector developed by Fiaregas (Ref, 21)
reduced high frequency noise by about 7-16 dBA 150 feet from the flare
with steam rates varying from less than 1 to about ZO tons per hour below
the sound levels for center steam and external jet injection. The Coanda
effect injector had little effect on low frequency combustion noise less
than about 250 Hz.
High frequency jet noise resulting from steam injection has been the subject
of most design attention (Ref , 20) because of its higher intensity levels,
while the problem of steady combustion noise has not yet been dealt with
effectively {Ref. 18). Injection of steam for smokeless operation increases
turbulence intensity, causes the flame to become shorter and stiffer and
increases the relatively low frequency combustion noise. The only avail-
able solution to the low frequency combustion noise problem has been the
use of low level enclosed ground flares. The enclosed ground flare both
damps the noise arid allows the burners to be fine tuned in order to reduce
noise. If combustion noise is objectionable, moderate and most frequent
releases can be burned at ground level, with an auxiliary elevated flare for
full-scale venting.
Both combustion and jet noise are functions of the individual flare tip
design. However, flare vendors indicate that scaling of various sizes
and capacities is possible using standard procedures (Ref. 20) and refer-
ence sound level data for most flare burners is typically available from
the manufacturer. An experimental correlation for estimation of jet
noise is also available (Ref. 22),
For design purposes, in order to determine whether jet noise must be
reduced by shrouding or other means, other sound sources in the area
must also be considered (Ref. 23). Variation of sound intensity with
distance from the source is obtained using an equation for hemispherical
spreading (Ref. 12):
Lp = L1QO - ZO Iog10(r/100) (4.3)
where
L = sound pressure at a distance r, dB
L -» = reference sound pressure at a distance
100 of 100 ieet, dB
iiom aovmd. source, tt.
For multiple sources, the following inequality must be satisfied (Ref. 24):
5 --
-------
where
LMSC-HREC TR D390190
C = exposure duration
T = allowable exposure duration
Present and proposed regulations limiting noise exposure are summarized
below (Ref.24)
Exposure per Day
(hr)
8.0
4.0
2.0
1.0
0.5
O.Z5
Current OSHA
Regulation
(dBA)
90
99
100
105
110
119
Proposed NIOSH
Regulation
(dBA)
85
90
95
100
105
110
Very serious low frequency noise problems can result from improperly
designed water seals which may vibrate at frequency levels (Ref. 18):
T = 0.31 D/HA/^ (4.5)
a
T = 0.149 D/H1/2 (4.6)
where
T = period of asymmetric pulsation, sec
T - period of symmetric pulsation, sec
S
D, H = diameter and height of seal drum, respectively, ft.
To alleviate this problem, water seal downcomers are usually terminated
in slotted tips of irregular length, and a perforated baffle plate is installed
at or very slightly below the equilibrium liquid level to increase damping.
A smaller and less submerged auxiliary relief pipe may also be installed
(Ref. 18).
Pulsations from water seals and from low flow instability can be avoided
if a lower linear velocity of 1 to 3 ft/sec is maintained in the flare stack
at all times. This figure is based on observations of low flow instability
in larger flare systems (Ref, 18).
35
-------
LMSC-HREC TR D390190
Low level flares reduce combustion noise by about 10 dB (Ref, 20). High
frequency jet noise is not a problem with this type of flare because of the
distributed burner design and the absence of steam jets. However, these
flares are subject to poorly understood "combustion driven" pulsation
which may cause low frequency vibrations (Ref. 18).
4,3,3 Smoke and Particulate Emissions
Almost all flames of organic gases and vapors are luminous due to incan-
descent carbon particles formed in their flames; exceptions are carbon
monoxide, methyl alcohol, formaldehyde, formic acid, carbon disulphide,
and cyanogen (Ref. 16). Under certain conditions these particles are re-
leased from luminous flames as smoke. The tendency for smoke pro-
duction is related to: (1) the quantity and distribution of oxygen in the
combustion zone, and (2) the type of gas being burned.
The most critical determinant of smoke production is the amount and
distribution of oxygen in the combustion zone. For complete combustion
of a product, a stoichiometric quantity of oxygen is required in the burn-
ing zone. For smokeless combustion to take place a portion of the stoi-
chiometric quantity of air must be evenly distributed in the primary
mixing zone. The remaining air required to complete the combustion
process is induced into the flame through aspiration and thermal draft
effects. This primary air must be well mixed with the gas prior to
flame ignition or soot will escape from the flame due to incomplete oxi-
dation taking place (Ref. 9).
The second factor influencing carbon formation is the molecular struc-
ture of the gases burned. The carbon to hydrogen ratio is one of the
principal factors controlling the tendency to carbon formation. The
structure of the molecule is also important, thus branched chain paraffins
smoke more readily than the corresponding normal isomers. The more
highly branched the paraffin, the greater the tendency to smoke (Ref. 25).
In forming soot or smoke in flames one starts with a small hydrocarbon
or organic molecule and ends with a relatively large particle containing
many thousands of atoms and a much higher carbon-to-hydrogen ratio.
Thus there must be both dehydrogenation and polymerization side re-
actions involved. The exact route from small hydrocarbons to large
soot particles is, however, uncertain (Ref. 25). Once soot is formed, it
is either consumed in the flame or emitted from the flame. Soot particles
emitted from the flame large enough to be visible can be detected as smoke,
Studies of particulate emissions from flames have thus far been concerned
with visible emissions. It is possible that a luminous flame can appear
smokeless and still be emitting particles too small to be visible. Thus
making a flame smokeless will eliminate visible emissions but could cause
an increase in very small particle emissions. These fine particulates would
be incorporated into the plume by viscous drag forces and would disperse
into the atmosphere. Particulates are removed from the atmosphere by ad-
sorption surfaces (i.e., vegetation, pavement buildings, etc.). The mechanisms
which cause adsorption are gravitational settling, diffusion down to and im-
pacting with the surface and precipitation scavenging.
36
-------
LMSC-HREC TRD390190
The carbon formed in flames generally contains at least one percent by
weight of hydrogen. The deposited carbon appears to consist of roughly
spherical particles with diameters varying from 100 to 2000 A and most
commonly between 100 and 500 A. The smallest particles are found
in luminous but nonsooting flames while the largest are obtained in
heavily sooting flames (Ref. 26).
For flares a smokeless flame can be obtained when an adequate amount
of air is mixed sufficiently with the fuel so that it burns completely be-
fore aide reactions can cause smoke. In order to get good mixing of
oxygen or air with a combustible material, an expenditure of energy is
required. This energy may be provided by the flare gas stream itself
through pressure reduction and/or thermal draft; or it may be provided
from an external source, such as steam injection, power gas assist or
a blower fan. The usual method to achieve a smokeless flame in a flare
is to introduce steam into the combustion zone to promote combustion
and retard the carbon forming reactions. This tends to retard smoke
formation in three ways: (1) the injection of the steam can be used to
inspirate air and provide turbulence to aid in the mixing of the fuel and
air; (2) the steam reacts with the fuel to form oxygenated compounds
that burn readily and also lowers the temperature of the unburnt gases
retarding thermal cracking; and (3) the steam dilutes the fuel and re-
duces the partial pressure of the fuel which reduces the carbon form-
ation in diffusion flames.
Steam is injected either through a center spray nozzle or a circumferential
steam ring, or both. Required steam-to-fuel ratios are empirical and vary
with the type of material being flared. For hydrocarbons, a design equa-
tion from Tan (Ref. 13) is available based on experimental data (Ref. 12):
W . = W.,_ (0.68 - 10.8/M) (4.7)
steam Hd
where
Wtl/^ = hydrocarbon flow rate, Ib/hr
rid
W x = steam flow rate, Ib/hr
steam
M = average molecular weight of the flared stream.
The equation arbitrarily assumes a constant steam-to-hydrocarbon ratio.
Experimental studies have not reflected such a simple dependence of
smoke formation on molecular weight. The most significant parameter
for hydrocarbons appears to be the ratio of carbon-to-hydrogen, but other
factors must also be considered such as the type of hydrocarbon being
burned, the temperature of the combustion zone and the quantity and distri-
bution of oxygen in the combustion zone.
In flare systems utilizing steam for reduction of particulate emissions,
steam requirements are based on some fraction of the maximum design
capacity. Discussions with flare vendors indicate a typical basis of
about 50% design capacity. Thus, only moderate relief rates will be
37
-------
LMSC-HREC TR D390190
burned smokelessly. Economics, not design factors, limit the smokeless
capacity of a flare. It is usually not feasible to maintain enough steam
capacity to handle very large flare loadings. The practical limit of smoke-
less burning is about 300,000 Ib/hr (Ref, 27).
The minimum amount of steam required produces a luminous flame with
no smoke. Increasing the amount of steam injection further decreases the
luminosity of the flame. An injection of an excessive amount of steam
causes the flame to disappear completely and be replaced with a steam
plume. Too little steam results in black smoke which is objectionable
and excessive steam produces an invisible emission of unburned hydro-
carbons (Ref. 5). With automatic steam control in flare installations,
the use of an excessive amount of steam and the emission of unburned
hydrocarbons is greatly reduced.
Other methods used to achieve smokeless burning in flares include the
use of water spray rather than steam, blowers to provide a forced air
draft, distributed flame or multijet burners, and low level ground flares
which rely on natural draft to provide air. Water sprayed is 50% less
efficient than steam and is frequently completely ineffective in providing
smokeless operation when the flame is tilted away from the injection ring
by a strong wind. Other problems with water spray injection include a
limited turndown rate of about 0.8 to 1, freezing in cold weather, and
quenching of the flame at low vapor relief rates.
Water spray injection has been applied successfully in special applications
such as with ground flares (Ref. 8) but is not recommended in applications
where air pollution is a problem {Ref. 5),
Smokeless flaring by means of multijet or distributed burner designs pose
safety problems inherent with throttling the discharge stream. Standby
flares of conventional design are usually considered necessary to accom-
modate emergency upsets. Low-level ground flares are essentially dis-
tributed burners at ground level and are sometimes useful in reducing
smoke emissions (Ref. 29). Other distributed burner designs include a
multiflare in which high and low pressure streams are collected in segre-
gated piping systems and discharged at the flare tip through nozzles of
different design. This technique is intended to utilize the energy of the
high pressure waste stream to achieve increased mixing.
Forced draft flares use a blower to provide air for smokeless burning.
For operating safety, air and fuel are injected in concentric stacks and
mixed in a chamber near the flare tip. Premixing limits the turndown
rate unless special blower control is .provided. Forced draft flares are
less popular than conventional elevated flares because of the high initial
and operating costs and limited turndown and are normally used only when
flare location prevents steam or water injection for smoke control.
Because of the intended use of flare systems as emergency control de-
vices, particulates are not monitored. Thus, there is no experimental
-------
LMSC-HREC TR D390190
basis for choosing between steam (or water) and air injection for par-
ticulate control. Kinetic measurements on soot formation and oxidation
in small flames (Ref. 16) indicate that the net rate of formation is kinetic-
ally controlled with a higher activation energy for the oxidation. On this
basis, air addition would appear to be preferable.
4.3.4 Chemical Emissions
Chemical emissions from flares may result from unburned fuel and partial
or complete oxidation products. Flare discharge rates are extremely
high; at the emergency release rate, burning rates on the order of 10^
Btu/hr are typical (Ref. 30), with the reaction distributed over flame
lengths of several hundred feet.
Discharge flows are typically highly turbulent at the tip of the flare burner;
literature design methods such as given by Tan (Ref. 13), Kent (Ref. 14) and
API RP 521 (Ref. 12) recommend the use of Mach 0.2 for sizing calculations.
Conversations with flare vendors suggejt that this figure may be low by
30% or more. Flame "holders" installed at the tip of conventional burners
allow velocities greater than Mach 0.2 before flame lift-off occurs. It has
been suggested that quenching of the combustion by turbulent mixing with
the cold ambient air may result in the formation of partial oxidation pro-
ducts and unburned hydrocarbons. Data of Sussman et al., (Ref. 31)
suggest that combustion of hydrocarbons in the steam-inpirated type of
elevated flares may be incomplete. The results of a field test on an ele-
vated flare with steam inspiration indicated that hydrocarbon and carbon
monoxide emissions from a flare can be much greater than those from a
properly operated refinery boiler or furnace where the CO emission is negli-
gible and hydrocarbon emissions are 0.14 Ib/bbl of fuel burned (Ref. 58)
Sussman et al., reported the results in the form of ratios:
CO: hydrocarbons = 2,100:1
CO2: CO = 243:1
Irritants (Ref. 32) such as aldehydes and other partial oxidation products
are also possible if combustion is incomplete. Aldehydes, CO and other
partially oxidized hydrocarbons have been established (Refs. 5 and 16) as
intermediates in hydrocarbon combustion. The formation of partial oxi-
dation products is enhanced by low temperatures as would occur in steam-
assisted smokeless flaring or flame quenching. Excessive use of steam
produces a white steam plume and an invisible emission of unburned
hydrocarbons (Ref. 5).
In spite of the air pollution associated with emission of unburned hydro-
carbons and partial oxidation products, there is little evidence of flare
design modifications to supress emissions resulting from incomplete
combustion. Little actual experimental data of flare emissions are avail-
able, mostly for reasons of sampling difficulty. DuPont (unpublished
report) has recently begun a development program for measuring flare
emissions by using helium as a tracer to account for sample dilution.
39
-------
LMSC-HREC TR D390190
Low-level enclosed flares should produce fewer unburn^d hydrocarbons
and partial oxidation products because of their lower design velocities.
4.3.5 Oxidation Products
Emissions in the form of complete combustion products may occur in
some cases, such as when flaring NH-, H_S, or organic bases contain-
ing heteroatoms such as sulfur and nitrogen. API Manual on the Disposal
of Refinery Wastes (Ref. 4) recommends other methods for disposal of
these gases or vapors. Flaring is recommended only as a last resort,
when other methods of disposal are not available. Alternate methods
typically involve collection of the streams by vacuum or pressure methods,
or solution (scrubbing), usually followed by a treatment of gases for segre-
gation of contaminating constituents. Incineration in thermal or catalytic
afterburners (Ref. 15) may also be a practical alternative if waste streams
containing contaminants can be isolated and when the waste steam load is
fairly constant.
It is frequently found that flares are used for the disposal of such gases,
however. Hydrogen sulfide streams in particular are frequently included
in flared waste streams. In these cases, the type of gas affects chiefly
the height of the flare stack and materials of construction. Stainless steel
types 304, 309 and 310 have been found to be particularly sensitive to fail-
ure due to intergranular corrosion promoted by sulfur acids. Inconel 800
or 629 are recommended for use in flare tips, stainless steel types 321 or
347, Inconel 625 or Incoloy 800 or 801 are recommended for use in steam
rings when streams containing sulfur compounds are being flared (Ref. 33).
Flare heights are chosen to provide adequate dispersion of either flared
streams (in the event of flame extinguishment) or combustion products
according to methods that will be discussed. Local regulations are used
to establish maximum ground level concentrations; in some cases these
regulations may also dictate which dispersion model must be used and
which gases may not be included in flared streams. In the combustion of
organic bases, some estimate of conversion efficiency to toxic or harmful
compounds must be obtained for the sizing calculation. For example, in
the combustion of nitrogen-containing fuels, NO« may be further decom-
posed to N and NO by secondary reaction. In the oxidation of fuel nitrogen
in a diffusion flame, conversion to NO and NO_ varies with nitrogen content,
combustion efficiency, flame temperature, and residence time in the flame,
but typically ranges from 20 to 100% of theoretical (Ref. 15). A detailed
calculation is available for nitrogen-containing fuels (Ref. 34). Such infor-
mation is not available for oxidation of waste streams containing chlorine,
sulfur, phosphorous or other heteroatoms. In the absence of detailed
information, it is recommended that the equilibrium product distribution
be used for the sizing calculation.
40
-------
LMSC-HREC PR D390190
4,3.6 Other Gaseous Emission Sources
Other sources of air pollution from industrial flaring include emission
of unflared hydrocarbons from vent lines and formation of nitrogen oxides
in the combustion process resulting from the contact of nitrogen in the
air with oxygen at the elevated flame temperature. To reduce flare sys-
tem cost, it has been a practice to vent gases of less than 40 mole weight
directly to the atmosphere with the remainder being sent to the flare
(Ref. 23), and this may contribute to the air pollution problem. Because
of the use of steam in most flare systems, flame temperatures will be
lower and emission factors for NOX will be lower than, for example, gas-
fired burners. Experimental measurement of NOX emissions from ele-
vated flares is difficult, but such measurements from low level flares is
practical and should serve as an upper limit for elevated flares with
steam injection. Experimental studies by a major flare vendor including
NOX measurements of a low-level flare unit are being conducted at this
time.
4.3.7 Dispersion of Chemical Emissions and Flammable Gases
Chemical emissions from the flare may result from flame extinguishment
while flaring toxic gases, as either complete or partial combustion pro-
ducts or as unburned fuel. In many cases the exact type and volume of
chemical pollutants is difficult to predict. However, when flaring streams
containing ammonia or trace amounts of phosgene, hydrogen sulfide, and
hydrogen cyanide, height is frequently determined by possible atmospheric
emissions rather than thermal considerations (Ref. 35). Even when hydro-
carbons are being flared, high concentrations emitted during flame extin-
guishment may cause asphyxiation or exceed the lower flarnmability limit
for the fuel at or near ground level. Since, apart from state, local, and
federal requirements, emissions may represent a safety hazard to plant
operations through loss of efficiency of operating personnel or low-level
fire and explosion hazards, sizing of flare stacks should include design
calculations to ensure that dispersion is sufficient to prevent concentra-
tions of flammable or physiologically harmful emissions from exceeding
safe limits at ground level.
Two distinct cases arise. For the case in which small concentrations of
pollutants are emitted from the flare in the parts-per-million range, rapid
mixing of the pollutants with the air occurs, and net changes in bulk density
do not result in appreciable fallout rates. In this case, the pollutants re-
main suspended in the ambient air and diffuse at the same rate as the air.
The concentration build-up at ground level occurs at a predictable rate.
This build-up can be determined by applying classical atmospheric dis-
persion equations.
Unignited discharges which are heavier than air may occur in the event
of flame extinguishment. Although these discharges do not have appre-
ciable particle settling rates, fallout may occur because of the higher
density of the bulk layer than the surrounding air. Very little published
information is available to describe the safety hazards which may accom-
pany this phenomenon. A recent study at DuPont on safety hazards of
41
-------
LMSC-HREC TR D390190
unflared discharges recommends the following design equation based on
wind tunnel studies (Ref. 36):
hs = 1.33 (E) (D)2 (4.8)
where
h = vent height above exposure level, ft
E = number of dilutions of stack gas with air to
reach lower explosive limit
D = diameter of vent tip, ft.
The design equation is recommended only for short-period emergency
releases of heavier -than-air gases. One additional factor which is known
to be significant is the stack gas velocity.
For the case of small concentrations of contaminants or discharges which
have molecular weights less than or approximately the same as the ambient
air, dispersion from flare stacks and other elevated sources is modeled
using Gaussian probability function to account for spreading and dilution
of pollutants by diffusion. According to the Gaussian model, the emitted
plume from a continuous source diffuses in the two directions perpendicular
to the wind (Fig. 4-1). The maximum ground level concentration occurs at
the distance from emission where the vertical downward diffusion equals
the emissions height and the spreading plume touches the ground. On
level terrain, this condition is expressed by the equation (Ref. 37):
a (x) = H/Z (4.9)
Zt
where
H = the effective height of emission
0 (x) = the vertical diffusion coefficient
2
(corresponds to distance).
The vertical and horizontal diffusion coefficients are standard deviations
for the Gaussian model. These are related to the eddy diffusivity. Dis-
persion of emissions in air occurs more rapidly than would be predicted
from molecular diffusion and all gases and suspended particles diffuse
at approximately the same rate (Ref. 38). Mixing is dominated by atmos-
pheric turbulence which is supported by temperature and velocity gradients.
Therefore, the vertical and horizontal diffusion coefficients depend on
atmospheric conditions and are usually determined from empirical
correlation.
Atmospheric conditions may also lead to conditions in which the continuous
dispersion model is no longer valid. In the presence of a low level inversion
layer, "trapping" of the emitted plume by the inversion layer may result in
an increase in ground level concentrations (Refs.39 and 40). Ground level
42
-------
LMSC-HREC TR D390190
x.-y.Z)
Fig, 4-1 - Coordinate System Showing Gaussian Distributions
in the Horizontal and Vertical (from Turner, Re£, 39}
43
-------
LMSC-HREC TR D390190
concentrations may also increase beyond the level calculated for normal
conditions during inversion breakup or "fumigation." In certain geo-
graphical areas, frequent occurrence of these phenomena may prevent
the safe emission of certain chemicals at any height.
Chemical emissions may be classified as (Ref. 4):
1. Toxic gases and vapors
2. Irritants
3, Malodorous gases and vapors
4. Asphyxiants
5. Aerosols smoke, mists and fumes
6. Dust and ash.
Threshold limits for a number of toxic substances set by the Americal Con-
ference of Governmental Industrial Hygienists are given in Table 4-1. Limits
of detection of selected odorous and noxious gases are given in Table 4-2
(API Manual). The lower flammability limits for gases and vapors being
flared can be obtained from general references such as Bureau of Mines
Bulletin No. 503 (Ref. 41). More recent references which provide flam-
mability limits and toxicity data include the handbook Industrial Hygiene and
Toxicology (Ref.42), the Factory Mutual Engineering Corporation publication,
Handbook of Industrial Loss Prevention (Ref. 43) and the Bulletin of the
Bureau of Mines, "Flammability Characteristics of Gases and Vapors"
(Ref. 44).
Dispersion models derived assuming a continuous plume are based on the
original publication by Sutton (Ref. 45). More recent "coning" models for
continuous dispersion differ chiefly in the method used for estimation of
the dispersion coefficients. API RP 521 (Ref. 12) recommends the use of
a weighted Gaussian distribution function for the calculation of required
stack height. Given the maximum allowable concentration (max) at ground
level, the required stack height may be calculated from the following:
VM Dz 4
uh ^
where
C = parts per million of polluting gas (maximum
ground level concentration)
M = tons of polluting gas emitted per day
V = volume of polluting gas, ft /lb
Dy.Dz = horizontal and vertical dispersion coefficients, ft
u = air velocity, mph
h = height of stack, ft.
44
-------
LMSC-HS.EC TE D390lfC(
Table 4-1
THRESHOLD LIMITS " FOR CERTAIN TOXIC SUBSTANCES
Gas or Vapor PPM'
Acelaldehyde 200
Acetic acid 10
Acetic anhydride 5
Acetone 1,000
Acrolein , 0.5
Acrylonitrile 20
AUyl alcohol . . 5
Allyl propyl disulfide 2
Ammonia 100
Amy! acetate 200
Amyi alcohol (uoamyl alcohol) 100
Aniline . . . 5
Arsine . 0.05
Benzene (benzol) 35
Benzyl chloride I
Bromine 1
Butadiene (1,3-buudiene) 1,000
Butanone (methyl ethyl ketone) 250
Buty! acetaie (nbutyl acetate) . 200
Butyl alcohol (nbutanol) 100
Butylamine 5
Butyl cellosolve (2-butoxyetba-
nol) . . . . , .200
Carbon dioxide . 5,000
Carbon disulfide ... . 20
Carbon monoxide 100
Carbon tetrachloride . . . . 25
Cellosolve (2-ethoxyeihanoI) 200
Cellosolve acetate (2-*thoxy-
elhyl acetate) . . .100
Chlorine . . . I
Chlorine triftuoride O.I
Chlorobcnzene (monochloroben-
zene) . 75
Chloroform (trichloromethane) 100
1-Chloro-i-nitropropane 20
Chloroprene (2-chloro-l,3-buta-
diene) . . .... 25
Cresol (all isomers) 5
Cyclohexane 400
Cyclohexanol 100
Cyclohexanone . . . 100
Cyclohexene . . . 400
Cyclopropane . . . 400
Diaceione alcohol (4-hydroxy-
4-me!hyi-2-pentanonc) 50
Diborane 0,1
oDichlorobenzene 50
Dichlorodifluoromelhane 1,000
l.t-Dichloroethane 100
).2-Dichloroethylene . .. 200
Dichloroethyl ether . . . 15
Dichloromonofluoromethane 1,000
1,1-Dichloro-J-nitroethane . . . 10
Dichlorotetrafluoroeihane . 1,000
Diethylamine . 25
Difluorodibrornomethane !00
DiMobutyl ketone . . . 50
Gases and Vapors
Gas or Vapor PPM"
Dimcthylaniline (N-dimethy!ani-
line) 5
Dimethylsulfate 1
Dioxane (diethylene dioxide) . 100
Ethyl acetate 400
Ethyl alcohol (ethanol) 1,000
Eihylamine 25
Ethylbenzene 200
Ethyl bromide 200
Ethyl chloride 1,000
Ethyl ether 400
Ethyl formate 100
Ethylene chlorohydrin
Hlhylenediamine
Ethylene dibromide (1,2-dibro-
moethane)
Ethylene dichloride (1,2-dichlo-
roethane)
Ethylene imine
Ethylene oxide .
Fluorine
Fiuorotrichlordmethane
Formaldehyde
Gasoline
Heptane (nheptane)
Hexane (nhexane)
Hexanone (methyl butyl ke-
tone)
Hemone (meikyl ootvtjl ke~
tone)
Hydrazine .
Hydrogen bromide
Hydrogen chloride .
Hydrogen cyanide
Hydrogen fluoride
Hydrogen peroxide, 90 per cent
Hydrogen selenide
Hydrogen sulfide .
Iodine . .
/iophorone
/jopropylarnine
Mesityl oxide
Methyl acetate
Methyl acetylene
Methyl alcohol (methanol) .
Methyl bromide ...
5
10
25
100
5
100
0.1
1.000
5
500
500
500
100
100
1
5
5
10
3
1
0.05
20
0.1
25
5
50
200
1,000
200
20
Mcihyi cellosolve (2-melhoxy-
ethanol) 25
Methyl cellosolve acetate (ethyl-
ene gjycol monomethyl ether
acetate) 25
Methyl chloride 100
Methylal (dimethoxymethane) .1,000
Methyl chloroform (1,1,1 -tri-
chloroethane) 500
Msthylcyclohexane 500
Mtthylcyclohexanol 100
Gas or Vapor
Methylcyclohexaoone .
Methyl formate
Methyl ijobutyl carbinol
(methyl amyl alcohol) .
Melhylene chloride (dichloro-
meihane) . , , ,
Naphtha (coal tar)
Naphtha (petroleum)
Nickel carbonyl
pNitroaniliiie ......
Nitrobenzene
Nitroethane
Nitrogen dioxide
Nitroglycerin
Nilromethane
2-Niuropropane
Nitrotoluene
Octane
Ozone
Pentane . .... . 1
Pentanone (methyl propyl ke-
tone)
Perchlorethylene (tetrachloro-
ethylene)
Phenol
Phenylhydrazine
Phosgene (carbonyl chloride) .
If
100
100
25
500
200
500
0001
1
1
100
5
05
100
50
5
500
0.1
,000
200
200
5
5
1
Phosphine 0,05
Fliospfcorn Crkhloriue 0.5
rropyi acetate 200
Propyl alcohol (topropyl alco-
hol) 400
Propyl ether (riopropyl ether) 500
Propylene dichloride (1,2-di-
chloropropane) 75
Propylene imine 25
Pyridine 10
Quinone 0,1
Stibine 0.1
Stoddard solvent 500
Styrene monomer (pheaylcthyi*
ene) 200
Sulfur dioxide 10
Sulfur hexafluoride 1,000
Sulfur monochloride 1
Sulfur pentafluorMe 0.02$
/>TeniarybutyItoIueDe 10
1,1,2,2-Teirachloroethane 5
Tetranitrometnaae I
Toluene (toluol) 200
oToluidine 5
Trichloroethylene 200
Tri/tuoromonobromomethaae.. 1,000
Turpentine 100
Vinyl chloride (chloroethylene) 500
Xylene (xylol) 200
45
-------
LMSC-HREC TR D390190
Table 4-1 (Continued)
Dust, Fume, or Mist
Aldrin (1,2,3,4,10,10-hexachlo-
ro-1,4,4a.5,8.8a-nexahydro-1,
4,5,8-dimcthanonaphthaIcnc).
Ammnie (ammonium sulfamate)
Antimony
Arsenic
Barium (soluble compounds)
Cadmium oxide fume
Chlordanc (1,2,4,5,6,7,8,8-octa-
ch!oro-3a,4,7,7a-tetrahydro-4,
7-mcthanoindane)
Chlorinated dfphtnyl oxide
Chlorodiphcnyl (42 per cent
chlorine)
Chromic acid and chromates
(as CrOi) . . .
Crag herbicide (sodium 2-[2,4-
dichlorophenoxy] ethanol hy-
drogen suifate
Cyanide (as CN)
2,4-D( 2,4 -dichlorophenoxy ace-
tic acid) . . .
Dieldrin (1,2,3.4,10.10-hexachlo-
ro-6.7-cpo>;y-l..-i.-1a.5.£.'',S.Sa-
octahydro -1,4,5,8 - dimcthano-
naphthalene)
Diriirotolucne
Dinitro-o-cresol
EPN (O-eihyl O-/»-nitrophenyl
tbionobtnzenephosphonate) ,
Fc:iov, dust) 0.5
(V.Q. fume) 0.1
Zinc oxide fumes 15
Zirconium compounds (as Zr) 5
Lead
Lmdane (hexachjorocyclohex-
ane, gamma isomer) 0.5
* Parts of gas or vapor per million parts of air by volume.
t Milligrams of dust, fume, or mist per cubic meter of air.
Millions of particles per cubic foot of air.
Source: "Threshold Limit Values for 1956," 18th Annual Meeting.
gtenists, Philadelphia, Apr. (1956).
Radioactivity: For permissible con-
centrations of radioisotopes in air, see
"Maximum Permissible Amounts of Ra-
dioisotopes in the Human Body and Maxi-
mum Permissible Concentrations in Air
and Water," Handbook 52, U.S. Depart-
ment of Commerce. National Bureau of
Standards, March 1953. See also, "Per-
missible Dose from External Sources of
Ionizing Radiation," Handbook 59, U.S.
Department of Commerce, National Bu-
reau of Standards, Sept. 24, 1954.
Mineral Dust
Aluminum oxide
Asbestos
Dust (nuisance, no free silica)
Mica (below 5 per cent free
silics)
Portland cement
Talc
MPPCFt
50
5
50
20
50
20
Silica
High (above 50 per cent free
SiO,) 5
Medium (5 to 50 per cent
free SiO,) 20
Low (below 5 per cent free
SiO,) 50
Silicon carbide 50
Slate (below 5 per cent free
SiO,) 50
Soapstone (below 5 per cent
free SiO,) 20
Total dust (below 5 per cent
free SiO,) 50
American Conference of Governmental Industrial Hy-
"Threshold limit values define the concentration levels of chemical compounds
and physical agents below which the- average healthy worker will suffer no demon-
^trably damaging effects.
46
-------
Table 4-2
PROPERTIES OF ODOROUS AND NOXIOUS GASES
Contaminating
Compound
Acetaltlrbyde
Acetone
Arid*
Arctlr
Butyric
Hy
(Clli)iCIICIIiCHi
cii.ciiciit
C>H»
CiH.
r.H«ru>
C«Hi(Clfi)i
CiltiHII
(ClU)tCHCHtSH
(('ll!)tCHer ft*; 1
ilar weight
Dfnolty
1'olnt
I.owfut Detectable
Odor Concentration*
illllltraroi Partn K.T
i Air -:i| (Fahrenheit) per Liter Million
2 14
2.00
2,07
3.04
1,20
0.00
1 11
1.5S
2.07
2.07
2,58
2.58
3.04
3.04
0.58
2 44
2,58
0.555
1 04fl
1.547
2071
2,0fl7
2.4008
2,4008
0.9084
1.4526
1 0368
1 9308
2.4211
2.70
3.18
3,68
1.60
2.14
2.02
311
311
3.50
i.s»
324
3.73
4.28
4,28
2.73
111
2 14
3.11
4.08
5.05
5.05
6.01
2.21
2.75
ppm at 32 F and
time* 4 4« time*
r.g
1.13
245
324
-121
87
148
173
207
180
244
228
280
270
-28
-30
04
-258.7
-127.5
-43.7
31 i
10.9
90.9
82.1
-1547
-03.9
20.7
19.0
120.7
178
231
291
40
05
153
208
190
259
205
358
375
382
S83
240
-75
07
107
28«
301
340
421
14
112
0.71 030
4.10 1.00
7.00 2.00
9.401 2.40|
590.00 41000
5.10 1,00
0.0039 0.0030
240.601 C3.00I
0.0010 0002«
40.00 53.00
11.00 344
0.76 0.23
Not detectable.
Not detectable
Not detectable
Not detectable
Not detectable
Not detectable
Not detectable
Not detectable
Not detectable
Not detectable
Not detectable
Not detectable
5.20 1.56
0.10 0.48
0.81 017
O.OB10 00410
0.0072 0.0028
0.0050 0.00 Id
0.0037 0.0010
0.0018 0*.60043
1.20 0.29
0.02 019
0.014 OOO27
35.001 10.001
1.50 1.00
0.094 0.0037
0.00023 O.UW0058
0.05.1 0.011
0.009 0.015
l'.50 o".20
8. GO 3.00
> International Critical Tablri, 3
(1933).
Ceefflclent of
Diffusion In Air . ».*., . ., -..*,. ,~ .,..*~.
# pfret t*h w«!nlncrlr Mfl
r«-i
32 Fand 1 Atm from Prolonged Bxponure
(Cm' ficr See)
o.ioot
0.083f
0.1061
o iist
0.14D
0.1 33t
0.004 1
OOR5
0.082J
0.070J
0.0731
o'.ois
0.170f
0.093 r
0.078}
,
....
00771
0.071J
0.002
0.095
0.004
0071
0.084
0004
0059
0.059
O.OOSf
0062f
0.055
0.055
0.122t
0.094
0.0«4t
0.054
0.04B
0.048
0.040
0.102f
0.083 1
271 3, McOraw
In High Concentratlona
Irrilunt
Irritant, anetthetlc
Irritant
Naunentlng
Irritant
Irritant
Antithetic, Winding
Sllulilly Irritant, aneithetlc
Anesthetic
Anesthetic
SIlRhlly Irritant
SH^btly Irritant
Irritant
Silently Irritant
Irritant
Irritftnt
A neat bet le
Anetthetlc
Anealhctle
Anesthetic
Anesthetic
ABenthetlc
Anestbellc
Anesthetic
Anenlhelic
Anetifhetic
Anesthetic
Aneithetic
Anesthetic
Neuro-lrrltnnt, «neatu«!!e
Aneithetic
Anesthetic
NsuseatlfiK
Nauseating
Nauseating
Nauseating
Nauae^tiNg
Naiiaett tlrtg
Nauseating
Irritant
Irritant, (iphyslint
NaunratlnE
Nan seating
Irritant
Irritant
Hill Book Co Inc., New
York
en
n
M
n
H
50
a
U)
I U.S. Bur. Ul*ti TecA. Paper in.
-------
LMSC-HREC TR D390190
The calculation is reported to be valid, on the conservative side, within
a factor of 2 within about six miles and to within a factor of 5 for greater
distances.
The eddy diffusion factors Dz and Dy are dependent upon atmospheric
stability and discharge height. For emissions at heights greater than
about 25 feet and for "neutral" atmospheric conditions (lapse rate equal
to the dry adiabatic rate)* the ratio Dz/Dy is approximately unity. For a
given stack height, the maximum ground level concentration may vary
by as much as a factor of two under varying atmospheric conditions. The
calculation is based on the original report published in 1932 (Ref.45). An
expression reported by Bosanquet and Pearson (1936) is of a similar form.
As applied to the calculation of the required stack height (Ref.46):
W
C (x, y)= 1.55x10* 5-2 -f (4.11)
max 2
where
uH MC
W = emission rate of contaminant, Ib/hr
c
C = maximum concentration of contaminants contained
in the air-gas stream at ground level downwind from
the stack of a point where x = H/2p and y = 0
M = molecular weight of the contaminant
p = vertical diffusion coefficient, dimensionless
q = horizontal diffusion coefficient, dimensionless
u = mean wind speed, mph
x = horizontal distance downwind from point of
emission, ft
y = horizontal distance crosswind from point of
emission, ft
H = effective stack height above ground, ft (physical
stack height plus plume height caused by the
velocity of the stack gases plus rise of plume
caused by difference in density between stack
gases and the surrounding atmospheres).
Bosanquet-Pearson Turbulence Parameters (Ref. 46)
p q p/q
Low turbulence 0.02 0.04 0.50
Average turbulence Q.05 0.08 0.63
Moderate turbulence 0.10 0.16 0.63
F
Lapse rate is defined as the rate of decrease of atmospheric tempera-
ture with increase in height, while the dry adiabatic rate is 5,50°F/1000 ft
48
-------
LMSC-HREC TR D390190
A number of empirical expressions are available for estimating plume
height above the stack (Refs. 47 and 48). The plume rise is sometimes
included in the design safety factor since the case Ah = 0 corresponds to
the upper ground level concentration limit,
Both the dispersion models of Sutton (1932) and Bosanquet-Pearson (1936)
are based on experimental data and are widely used for flare stack sizing
(Ref. 35). Sample "averaging times" differ between the two methods; this is
reflected in the difference between Button's a coefficients and the Bosanquet-
Pearson turbulence parameters, p and q, Button's method as applied
to stack sizing (Ref. 4) is based on a three-minute interval and is usually
considered conservative. In practice, chemical companies have tended
to use the Bosanquet-Pearson equation which is based on a 30-minute
concentration interval (Ref. 49). Sample averaging time affects the de-
sign Stack height since, because of wind fluctuations, instantaneous con-
centrations are much higher than concentrations averaged over time
periods long enough to be physiologically significant. The effect of sample
averaging time on concentration is discussed by Turner (Ref. 39). Com-
parison of stack heights calculated on the basis of outton's and Bosanquet-
Pearson methods is given in Ref. 2,
The dispersion model chosen and required averaging times will frequently
depend not only upon plant safety but also upon local air quality regulations.
Turner's Workbook of Atmospheric Dispersion Estimates (Ref. 39) is
recommended for use by the U.S. Environmental Protection Agency and is
considered typical of recent guidelines. The chief advantage of the more
recent techniques appears to be in the measuring techniques used in esti-
mating the diffusion coefficients and the inclusion of methods applicable
to more severe atmospheric conditions.
Atmospheric or "eddy" diffusion coefficients, o, given by Turner {Ref. 39)
are given as functions of downwind distance and atmospheric conditions.
The most severe conditions correspond to stagnant wind conditions and
strong sunlight. Under these conditions, the maximum ratio of o /cr is
about 2.0 within about 1000 meters. The instantaneous maximum con-
centration at ground level is:
where
H = effective discharge height, m
3
X - concentration of gases or aerosols, g/m
(particles <
Q - emissions rate, g/sec
u = mean wind velocity, m/sec
0 ,a = horizontal and vertical dispersion coefficients.
z' y *
n = 3.14
e = 2.72
49
-------
LMSC-HREC TR D 3 90190
Conditions which may significantly affect the maximum ground level
concentration include fumigation and plume trapping in the presence
of a low level inversion layer. In some cases, terrain effects must also
be considered (Ref. 39).
Fumigation or inversion breakup may occur when a superadiabatic rate
develops in the lowest layer in the atmosphere. This situation occurs
when the surface is warmer than the overlying air such as in conditions
when the ground is being warmed by solar radiation or when air flows
from a cold to a relatively warm surface (Refs. 4 and 39). Impingement
of the resulting vertical air current with the plume may break up the plume
and bring isolated portions of it to the ground almost undiluted (Ref. 4).
The heavy concentration of polluting materials may persist for as long as
30 minutes (Refs. 2 and 40), The equations for estimating maximum ground
level concentrations during fumigation conditions is based on an adaptation
of Eq. (4.10).
W *"* t M t -» *
yf
where
h, = H + 2o-
1 z
cr f = ay + H/8
X, Q, u as previously defined
The presence of a trapping inversion layer at mixing height, L, leads
to higher ground level concentrations, depending on the height above
ground of the inversion layer (Ref. 39):
X =
It has been observed that the surface concentration may be increased by
as much as a factor of 3 for mixing heights ranging from 760 to 1065
meters (Ref. 40). The mixing height limitation will not be significant if
the concentration maximum calculated from Eq. (4.9) occurs within a
relatively short distance from the stack. This distance x (corresponding
to <7Z) is given by the approximate equation (Ref. 39):
a = 0.47L (4.15)
For distances x ^> x the invariance of Eq. (4.14) with stack height may
limit the total rate of certain chemical emissions.
50
-------
LMSC-HREC TR D390190
4.3.8 Air Pollution Rules and Regulations Affecting Flares
The 1970 Clean Air Amendments (Ref. 50) have provided three rule-setting
mechanisms that could directly affect the flaring of waste streams in the
United States: (1) national primary and secondary ambient air quality
standards, (Z) standards of performance for new stationary sources, and
(3) national emission standards for hazardous air pollutants.
National Primary Air Quality Standards are time-based, maximum-
allowable ambient air pollutant concentrations that can be tolerated with-
out adversely affecting public health. Secondary standards are those con-
centrations of a pollutant that can be tolerated on a time basis without
affecting public welfare. Setting and enforcing emission control regula-
tions to ensure that ambient air standards are met is a state function.
Both primary and secondary standards have been published for six classi-
fications of pollutants sulfur oxides, particulate matter, carbon monoxide,
photochemical oxidants, hydrocarbons and nitrogen dioxide (Ref. 51) and
are summarized in Table 4-3. Ambient air standards are most applicable
when applied against pollutants that are emitted by a large number of dif-
ferent sources. Except in special cases, such as flaring a high t^S stream,
it is unlikely that flaring alone will cause ambient air quality standards to
be exceeded. Many states restrict the flaring of H^S, chlorinated hydro-
carbons or other gases whose combustion products they feel may cause an
emissions problem.
Table 4-3
NATIONAL PRIMARY AND SECONDARY STANDARDS
Pollutant
Sulfur oxides
(measured as-SOf)
ParticuUte mallei
Carbon monoxi
-------
LMSC-HREC TR D390190
New stationary' source performance standards are allowable emissions
for new or modified emission sources. The mechanism regulates a
specific industry rather than a pollutant. The Federal Government has
primary responsibility for enforcing new source standards. It can be
delegated however, to qualified State control agencies. When Federal
new source performance standards are set, the law requires that State
governments establish emission standards for the same existing sources.
Although the rule making is a state function, the variation of the standards
across the nation may be relatively small (JRef. 56),
The performance standard for new petroleum refineries is the only one
that specifically mentions flares (Ref. 53). Standards limit only the emis-
sion of sulfur dioxide from flares. These standards limit the emission
of sulfur dioxide from fuel gas combustion systems which include flares.
The regulation prohibits the burning of any fuel gas which contains r^S
in excess of 230 mg/drv scm unless the resulting gases are treated to reduce!
the release of SO2 to the atmosphere. However, the combustion of process
upset gas in a flare, or the combustion in a flare of process gas or fuel
gas which is released to the flare as a result of relief valve leakage, is
exempt from the regulation. Process upset gas means any gas generated
by a petroleum refinery process unit as a result of start-up, shut-down,
upset or malfunction.
Smoke emissions are limited from new sources by opacity standards.
The standard limits the time, two minutes per hour, when the average
opacity can exceed 20%. All hourly periods during which there are
three or more one-minute periods when the average opacity exceeds 20%
are considered periods of excess emission and must be reported. How-
ever, the opacity standards do not apply during periods of start-up, shut-
down and malfunction. The opacity standards in effect require flares to
operate smokeless except for emergency occurrences. All states have
opacity standards for existing sources which require smokeless flaring
at least for the major portion of the time.
National emission standards for hazardous air pollutants provide the
third standard setting tool. A hazardous air pollutant is defined as one
that "will cause or contribute to an increase in mortality or an increase
in serious irreversible or incapacitating reversible illness (Ref. 54)."
Asbestos, beryllium and mercury have been designated hazardous pol-
lutants and allowable emission standards have been set (Ref. 55). The
extent to which this rule-setting tool will be used to set standards for
other hazardous pollutants is unclear at present (Ref. 56).
The Province of Alberta, Canada, has a number of sour gas processing
plants (gas containing hydrogen sulfide). Normal releases of tail gas
from these plants must be incinerated. All plant emergency releases
of sour gas are required to be flared in specially designed flare stacks
with an adequate amount of fuel gas so as to ensure gases with low heating
values are successfully flared (Ref. 57). Their experience has shown
that a minimum of 250 Btu/ft-^ of gas going to the flare should be
52
-------
LMSC-HREC TR D390190
maintained. The additional fuel gas serves two purposes. It ensures
a more complete combustion of the gas and also gives the combustion
products a greater lift off the stack. Depending on the length of the
flaring and the amount of gas to be flared, the height of the flare stack
is designed so that the resulting ground level SO? concentration does not
exceed 0.2 ppm for flaring greater than one hour and 1 ppm for flaring
less than one hour. The computational method used for the expected
ground level pollutant concentration and thus the estimation of stack
height required are the Sutton equation with the Lowry modification for
the case of even ground and the Pasquill method for the case of uneven
ground.
In populated areas the flare's flame has caused a nuisance to people living
nearby. In parts of Germany regulations limit the amount of time that an
elevated flare can be used (Ref.27). Ground level enclosed flares must
be used to hide the flame for 95% of the time the flare operates. The
elevated flare is used only for severe emergencies.
The main thrust of the air pollution regulations on flares has been toward
smokeless operation, at least for most flare occurrences. Many states
further prohibit or restrict flaring of waste streams whose combustion
products may cause an emissions problem. However, restrictions do not
apply during upset conditions when safety is the overriding concern. Very
little is known about the emissions of unburned hydrocarbons, carbon
monoxide and nitrogen oxides from flares and as long as ambient standards
are met there are no regulations or standards affecting these emissions.
4.3.9 Flare Emission Factors
The emission factor is a statistical average or a quantitative estimate of
the rate at which a pollutant is released to the atmosphere as a result of
an activity such as combustion or industrial production, divided by the
level of that activity. The emission factor thus relates the quantity of
pollutants emitted to some indicator of activity such as production capacity,
quantity of fuel burned, or vehicle miles traveled. Emission factors may
be found in a number of literature sources. The most complete collection
of factors has been published by the Environmental Protection Agency in
AP-42, Compilation of Air Pollutant Emission Factors (revised 1973)
(Ref. 58). Emission factor accuracy and reliabilty are dependent upon
many variables. It is generally accepted that emission factors generated
from on-site source-test data are preferred and will give more realistic
estimates than those developed strictly from engineering analysis or
material balances (Ref. 59),
The EPA compilation lists, in the section for petroleum refining, an emis-
sion factor for the "vapor recovery system or flaring." The hydrocarbon
emission factor is given as 5 pounds per 1000 barrels of refining capacity.
Emissions of particulates, sulfur oxides, carbon monoxide, aldehydes,
ammonia and nitrogen oxides are listed as negligible. The basis for this
factor on hydrocarbon emissions is not given, but it is unlikely that it is
based on any on-site source-testing of flares.
53
-------
LMSC-HREC TR D390190
C rose et al., (JRefs. 60, 61) estimated refinery emiss ion sources from hypo -
thetical 100,000 barrel-per-day capacity refinery. The only significant
emission that they considered coming from flares was NO. Their esti-
mates of the NO emission factor for flares was 0.7 pounds per 1000 barrels
of refinery capacity. They did not, however, give the basis for their
estimate. Section VII discusses the results of a user's survey that was
used to develop quantity and quality data on gases now being flared.
4.4 Flare Safety
Since the various flare systems were developed as a means of safely
disposing of refinery and petrochemical wastes, the bulk of the design
calculations involved in sizing the flare system are concerned with plant
and operator safety. Flare tip diameter is sized for flame stabilty and
maximum throughput within a given allowable pressure drop. However,
knockout drum and design sizing, flare height, location and material of
construction are based entirely on safety of operation. Primary con-
siderations include explosion potential, toxicity and plume dispersion,
and thermal radiation. The design bases for these calculations is de-
scribed in detail in the following section.
4.4.1 Explosion Potential
Explosions associated with flaring result from, two major sources the
formation of a flammable air-fuel mixture within the flare system which
can be ignited by the pilot burner, and secondary explosions resulting
from flare emissions of thermal radiation and burning liquid droplets.
With the flare system operating normally during an upset, thermal
radiation from the flame can produce sufficient heat to ignite flammable
gases and liquids stored nearby. Usually resulting from malfunction of
the knockout drum, injection of droplets of heavy liquids into the flare
stack can produce a high velocity spray of burning liquid droplets which
can cause damage to plant equipment and personnel.
Formation of air-fuel explosive mixtures within the flare system may
result from:
1. Back-diffusion of air into the flare system,
2. Leaks in the flare system resulting from improper
design or selection of valves and valve locations
corrosion or low temperature failure of flare piping,
and
3. Accidental injection of air during maintenance or
in start-up following maintenance.
Formation of explosive mixtures within the flare system may result through
design error in the mixing of reactive streams into a. common flare system
or through operator error in mixing reactive streams into a single flare
system during flare maintenance.
54
-------
LMSC-HREC TR D390190
4.4.2 Vapor Purging
Vapor purging has traditionally been used to prevent the formation of ex-
plosive mixtures in flare and vent systems by preventing the admission
of air into the flare system through leaks, back-flow of air at the flare
tip at very low flows and back-diffusion of air into the flare tip. Either
combustible gases such as methane or natural gas, or inert gases such
as nitrogen or CO, are frequently used as for purging flare systems
(Ref. 62). Inert gas generators are commercially available which use
the combustion products of natural gas (CC»2. ^2) as the purge gas. Utili-
zation of the combustion air in these generators allows purge gas (natural
gas) to be reduced by about a factor of 11:1.
Back-diffusion of air becomes a problem at very low flare rates. In
general, the diffusion or mixing process is more rapid and air influx
rates greater when lighter gases are being flared (Refs. 12 and 62).
When gases with molecular weights less than air are being flared (such
as hydrogen or CO) the pressure at the flare tip may frequently be less
than atmospheric, even with large gas flow rates (Ref. 12). A continuous
flow of purge gas is generally required for such systems. Conversely,
the effectiveness of the purge gas increases with molecular weight. Heavy
gases tend to displace the air as a piston while lighter gases mix with the
air and are less effective for displacement. An empirical equation for
estimation of purge gas requirements has been given by Husa (Ref. 62).
This correlation includes terms for diffusion and natural draft but ignores
contributions from the molecular weight of the flared gas. Discussions
with a flare vendor indicate that the correlation is not suitable for scale-up
to flares having diameters larger than about 24 in.
In practice, vapor purging is frequently also used to maintain a stable
flame at low flow rates to provide a flammable mixture or increase the
heat content of flared gases. Purging therefore adds substantially to
normal operating costs. Vapor purging is also required for displacement
of air in vent systems prior to pilot ignition during normal start up, for
example, following shut-down for maintenance.
Safety problems associated with vapor purging have been discussed in
detail by Bluhm (Ref. 63). Internal explosions occur primarily by acci-
dental introduction of air into the flare stack. Introduction of air occurs
not only through back flow and diffusion but also results from improper
selection of valves and other components, failure to purge the flare and
associated lines following maintenance, and through malfunction of com-
pressor controls on flares equipped with vapor recovery systems. Purging
with flammable gases can also present safety problems if oxygen monitors
are used to adjust flow rates since oxygen monitors do not differentiate
between air entry from leaks and from back-flow and diffusion.
55
-------
LMSC-HREC TR D390190
4.4.3 Molecular Seals
Molecular seals are intended to form a second line of defense against the
entry of air into the flare stack. While in a vertical flare stack, gravity
exerts an additional driving force to increase the diffusion of air into the
stack, "molecular" seals create an inverted flow field to reverse the
gravitational effect and further reduce air entry. Use of a lighter-than-
air purge gas further creates a pocket of light gas at the top of the air
entry path. Purge gases such as methane or natural gas are therefore
most effective for flare systems equipped with a molecular seal. Use of
a molecular seal and a lighter-than-air purge gas allows a much lower
purge gas rate. Installation of the John Zink molecular seal (Ref.64) at
the top of the flare and immediately below the burning point (Fig. 3-8)
reduces purge gas required to approximately 10% of the volume required
if the molecular seal is not used (Ref. 65).
4.4.4 Fluidic Seal
A recent development in air reentry seal design uses baffles to redirect
the flow field at a point near the flare tip and prevent air entry (Fig. 3-9).
The fluidic seal (Ref. 66) reportedly requires only 25 to 33% of the purge
gas used in molecular seals and weighs only 10% as much as the inverted
seal. The fluidic seal is designed to be used instead of the molecular seal.
Limited published information (Ref. 67) indicates that it has performed
satisfactorily in its intended application.
4.4.5 Explosion Suppression Systems
Explosion suppression systems have recently been developed which are
suitable for installation at the base of flare stacks (Ref. 68). The purpose
of the system is to detect ignition using light or pressure sensors before
the flame can propagate an appreciable distance. The sensing device
fires an explosive activator discharging an extinguishing agent (typically
a halogenated hydrocarbon) up and down the flow field through a tee. The
elapsed time from the detection of an alarm condition to suppression may
be only a few milliseconds. The first suppression system was installed at
Sun Oil Company's Marcus Hook, Pennsylvania refinery in June 1967 (Ref. 8).
4.4.6 Water Seals and Flame Arresters
Flame arrestors and water seals are intended to prevent a fire once
started from spreading throughout the flare and vent system. Because
of their weight these systems are typically located at or near the base
of the flare stack and offer no protection for the stack itself. Either
flame arrestor or water seals must be used if a flammable mixture is
being flared.
56
-------
LMSC-HREC TR D390190
Flame arresters are typically long narrow passageways within a sponge-
like metal structure designed to prevent flame propagation by chilling the
burning gases below their ignition temperature. Flame arresters have
become somewhat unpopular for use in flare systems because of.their
tendency to plug, and are not recommended for use without an emergency
by-pass arrangement (Ref. 63). A proprietary flame arrestor design is
also available (John Zink Company) which combines the water seal with
the barrier principle.
Water seals are normally provided in the gas inlet line to the flare at or
near the base of the flare stack and are intended to prevent flash back to
the flare header. These are of two types; pipe seals and seal drums.
Pipe seals usually consist either of a loop in the flare inlet line or a trap
built into the base of the flare stack. A pipe seal, therefore, provides
only limited disengagement space for removal of water or condensed
hydrocarbons. Seal drums are larger, usually more expensive, contain
a larger liquid volume and provide both a liquid disengaging space and a
reduction in liquid velocity above the liquid. Seal drums are thus less
susceptible to pulsation at low flow rates, and have small likelihood of
the water seal being blown at high vapor relief rates.
The seal drum is often incorporated into the base of the flare stack as
part of the structural unit. When a separate drum is used, the drum
should be located between the flare stack and the header drums and as
close to the stack as possible (Ref. 12), Special design considerations
include auto-refrigeration cooling of the gas inlet which requires auxiliary
heating, solubility of the gas stream in the seal water, corrosiveness of
the dissolved gas, and condensation of the gas stream in the liquid which
requires continuous water inlet and skimming or removal of the liquid
pha s e.
Under normal operating conditions, the most important design considera-
tion is the stability of the liquid seal. This affects both the flashback pro-
tection and flame stability at low flow rates. Stability of the liquid seal is
affected by the ratio of the inlet or outlet gas areas, dispersion of the gas
into the seal liquid, the temperature of the gas inlet stream and mainte-
nance of the liquid level by means of alarm and control devices.
Sizing requirements which are applicable to both drums and pipe seals
are discussed in API RP 521 (Ref. 12). For seal drums, the recommended
maximum ratio of inlet cross-section to vessel free area should be 1:3.
The vapor space in a vertical seal drum should be 2 to 3 times the diameter
to provide disengaging space for entrained seal liquid. A minimum dis-
engag*ng space of three feet is suggested for a horizontal seal drum. For
a pipe seal, the gas flow area above the seal should be at least as great
as the inlet line area. This requirement is considerably less stringent
than for drums, and pulsations are frequently encountered at low flows.
57
-------
LMSC-HREC TR D390190
It is recommended (Ref. 12) that the seal contain a volume of water below
the normal sealing line equivalent to the volume contained in a 10-foot
length of the inlet line to prevent flashback. To reduce pulsations, the
depth is arbitrarily" limited to a maximum of about 12 inches.
Provisions for establishing and maintaining an adequate seal level are
required. Because of the nature of the materials handled and the condi-
tions under which components must operate, instrumentation for all parts
of the flare and vent collection systems should be reliable, easy to main-
tain and readily available for replacement. For example, tri cocks (valves
may be desirable in addition to, or instead of, level gauges (Ref. 12).
Special design attention should be given to seals which are to be used for
flaring heated gases or which may be subjected to thermal upsets. Cool-
ing of the gas by the seal liquid will create a partial vacuum in the cooler
disengagement portion of the seal drum. The construction of an inlet
vacuum leg is recommended to compensate for this effect provided that
the maximum inlet temperature and flow rate can be estimated. The addi-
tional volume of liquid in the inlet line required to form the vacuum leg
must be contained within the seal drum and this may necessitate an in-
crease in drum size (Ref. 12).
4.4.7 External Fires and Emissions
External fires may result from the discharge of burning liquids from the
flare or from thermal emissions from luminous flames. Principal con-
trol methods include the use of knockout drums to separate flammable
liquids and entrained liquid droplets, insulation and safety relieving of
pressure vessels, remote location of the flare stack, and the injection
of air, water, or steam into the flare tip to reduce luminosity. Knockout
drums are usually located either between the process units and the col-
lection system or at the flare itself. These require fairly precise level
control of the accumulated liquids in order to prevent additional hazards
from accumulated flammable liquids (Ref. 63). Design methods for insu-
lation and relief value sizing of pressure vessels are discussed in the
API Guide for Pressure Relief and Depressuring Systems (Ref. 12). Steam
is widely used to reduce smoking and luminosity, but thermal emissions
from flares have not been well quantified. For this reason, flare stacks
are usually located several hundred feet from process units handling low-
flash point materials (Ref. 69).
4.4.8 Knockout Drum Sizing and Design Criteria
The design method used for sizing knockout drums is based on experi-
mental measurements of terminal velocities of spherical droplets in
gases (Ref. 70). In the application of drum sizing, the maximum allowable
stream velocity for separation of liquid droplets of a specified size occurs
when the droplet imparts a drag force equal to the gravitational force. The
maximum allowable velocity fixes the knockout drum diameter.
58
-------
LMSC-HREC TR D390190
Since the drag force depends on particle diameter, the maximum allowable
vapor velocity depends on the maximum size liquid droplet which can be
burned in the flame. From Ref, 12, the relationship between liquid drop
diameter and its terminal velocity is given by;
4Dg (p^ - pv)
3PVC
where
D = maximum allowable droplet diameter, ft
p. = liquid density, Ib/ft
3
p = vapor density, Ib/ft'"
U. - terminal velocity of liquid droplet, ft/sec
g = gravitational acceleration constant = 32,174 ft/sec
C = drag or friction factor (dimensionless)
The maximum allowable droplet size to be admitted to the flare can be esti-
mated from experimentally determined burning rates (Ref. 71) but is normally
chosen arbitrarily (Refs. 12 and 13). A typical value is 150|im. For removal
of particles larger than 150 /-tm, Eq. (4.16) reduces to;
u
i ' Pv
t
(4.17)
This basic equation is widely accepted for all forms of entrainment separa-
tion (Ref. 12).
The drag coefficient, C, corresponding to a given droplet diameter is obtained
using an empirical correlation developed by Lapple and Shepherd (Ref. 70).
The following equation approximates the experimental data to within +5%.
where
Re = Reynolds number
= p U.D/fi (4.19)
- gas viscosity, Ibm-ft -sec
Since the (unknown velocity appears both in Eqs, (4.16) and (4,18), the velocity
is determined using an iterative solution method. There is no convenient closed
form equation to express the terminal velocity explicitly in terms of system
parameters which is sufficiently valid for the entire range of applicable con-
ditions. The range of validity of the correlation is discussed in Ref. 70.
59
-------
LMSC-HREC TR D390190
Having determined the maximum allowable vapor velocity corresponding to
a given maximum droplet diameter, the knockout drum diameter can be deter-
termined. For a vertical knockout drum with a tangential gas flow inlet, the
drum diameter can be determined using the following equation:
V
4Q
3600
where
Q = vapor relief rate, Ib/hr
D, = knockout drum diameter, ft
k
p , U = as previously defined
Horizontal and vertical knockout drums are available in many designs, the
main differences being in how the path of the vapor is directed (Ref. 12). Ln
sizing horizontal drums, it is necessary to consider the volume of stored
liquid which will obstruct part of the vapor path (depending on liquid level).
A liquid holdup (residence) time of 10-30 min is suggested (Ref. 12).
Safety factors are frequently considered necessary in drum sizing (Ref. 69).
Very large drums on the order of 10 feet in diameter and 40 feet long are
frequently considered necessary (Ref. 69). Other safety considerations in-
clude auto-refrigeration cooling and maintenance of the liquid level (Refs. 12
and 63). Heating coils may be required in cold climates or in applications
where auto-refrigeration may be a problem (Ref. 63). A high level alarm is
required to prevent liquid accumulation with a low level alarm to prevent
vapor entry into the liquid disposal or recovery system (Ref. 12). Since level
control is of vital importance to prevent liquid entry into the flare and to en-
sure an unobstructed vapor path at all times, duplication of alarm and control
devices is frequently recommended (Ref. 63). Knockout drums equipped with
automatic pump out systems should also include a means of manually con-
trolling the liquid level.
4.4.9 Thermal Radiation Hazards
Hazards to people who are working in the vicinity of flares and to process
equipment are normally the principal factors which determine location of
the flare and flare height. Thermal emissions fromthe flare are dependent
upon flame geometry and luminosity and upon ambient conditions such as
relative humidity, wind effects and solar intensity. Some disagreement exists
in the estimation of luminosity from flames and design methods employed
have generally tended to be conservative. Fundamental understanding of the
nature of turbulent flames and even the causes for thermal (infrared) radiation
from flames is currently not available, so that meaningful correlations to allow
luminosity and flame geometry to be predicted have not been established. As
a result considerable variation exists in the estimation of these parameters.
Most methods ignore effects such as absorption of thermal energy by the atmos-
phere and convective aeat transfer between the flame and the ambient air and
therefore serve to predict upper limits for radiation intensity. This is important
60
-------
LMSC-HREC TR D390190
from a manufacturing cost standpoint but not from a safety point of view
as long as current practice of estimating flare emissions is continued.
Methods of current practice are described briefly in this section.
The principal design equation used in estimating thermal emissions
assumes spherical emission of radiant energy from a point source
(Refs. 12, 13, 14 and 30).
K = (4.21)
47T D*
where
K = radiant heat flux, Btu/hr-ft
Q = total heating rate of flame, Btu/hr
F = the fraction of the total heat generated which is
released as radiant energy
D = distance from the point of emission, ft
The emlssivity factor, F, may be estimated from Planck's law (Ref. 71)
or empirical correlation (Ref. 13) but is usually adjusted from literature
values (Refs. 12, and 72). Emmisivity is believed to result from: (1)
hot CO- and H_O which emit in the near infrared, and (2) solid particles
of carbon heated by the flame (Ref. 30). Emission from carbon particles
is probably more important since emissions from CO_ and H_O are atten-
uated by CO? and HLO in the air surrounding the flame. Flames of hydrogen
and H_S (which give H_O only) and of methanol (which yields HLO, CO?)
emit very little visible or infrared radiation (Refs. 16 and 72). Small-
scale tests of hydrogen flames (Ref. 72) indicate that 98% of the radiation
(from hot H«O) would be adsorbed within the first 100 feet from the flame
(Ref. 30) under normal atmospheric conditions.
Other evidence in favor of carbon emissions include (Ref. 16):
1. The intensity distribution is near that of a Planckian
radiator
2, Emitted light obeys the Rayleigh law for small solid
particles, and
3. The scattered light is polarized.
Even with the simplifying assumption of carbon as the predominant emitter,
the emissivity would depend on a number of factors including particle size
distribution, flame geometry and flame temperature. The usual approach
(Refs. 4, 12, 13, 14), is to assume a constant value for K, based on adjusted
experimental or literature values.
Experimental emissivity values for diffusirn flames vary with burner
diameter and carbon formation. Measured emissivity values increase
with diameter to a value which is approximate!'/ constant (Ref. 72). This
61
-------
LMSC-HREC TR D390190
Experimental emissivity values for diffusion flames vary with burner
diameter and carbon formation. Measured emissivity values increase
with diameter to a value which is approximately constant (Ref, 72)*, This
upper limit is used for design purposes. Injection of steam, premixing
with air and multijet designs reduce smoke, luminosity and thermal
emissions (Ref. 12). However, smokeless burning with reduced emis-
sivity is applicable in the use of Eq. (4,21) only for reduced flows. For
full-scale emergency upsets, higher emissivity occurs. For example,
when steam is used for smokeless flaring, common practice is to assume
smokeless operation at 10% of the maximum flare discharge rate (Ref. 73).
Estimations of the flame boundary and wind effects are needed for the
application of Eq. (4.21), in order to locate the origin of the emissions
source (D =0). Estimation of the flame boundary is complicated and
methods exist only for the approximate calculation of the flame length.
Therefore, usual practice (Refs. 12, 13 and 31} is to consider the emis-
sion of thermal energy as a point source near the midpoint of the flame
axis. This method is considered to be adequate except for radiation cal-
culations very close to the flare stack where view factors must be calcu-
lated to account for the approximately cylindrical flame geometry and the
angle of orientation of the flame with respect to the object receiving the
radiation (Ref. 30). Wind effects are normally considered separately,
An empirical correlation (Fig. 4-2) is recommended for estimating flame
length by API RP 521 (Ref. 12). Flame length is considered to be a func-
tion only of the total heat released from the flame. Actual flame lengths
may vary by as much as 50% (Ref. 49), and this variation should be allowed
in the design method. Wind effects are considered separately.
A design equation for estimating the length of turbulent flames is recom-
mended by Craven (Ref. 71) based on the work of Hawthorne et al., (Ref. 74):
L.-S
where
L = flame length, ft
S = the height of the breakpoint from the nozzle, ft
d = nozzle diameter, ft
C. = mole fraction of the nozzle fluid at the flame boundary
T-. = adiabatic combustion temperature, R
1?
a = the ratio of number of moles of reaetants to number
of moles of products for a stoichiometric mixture
T,. = temperature at the nozzle, R
M = molecular weight of the surrounding air
S
M = molecular weight of the fuel.
Also see page 78 of this report for values of this upper limit,
62
-------
LMSC-HREC TR D390190
FLAME LENGTH, FT, INCLUDING ANY LIFTOFF
o 5 c
Ik « 0» ~ N i CB » N M twOOl
X*
x-"
<
X-
X
^x-
^
x*^
A
xX
-X
O
V^"
j»^
0
^
^
a
a
x"
x-
LEGEND:
FUEL GAS -20" STAC
0 ALGERIAN GAS WELL
A CATALYTIC REFORMEF
RECYCLE GAS- 24" S
D CATALYTIC REFORMEF
REACTOR EFFLUENT
O DEHYDROGENATION U
X HYDROGEN -31" STAC
X HYDROGEN -30" STAC
^
^
K
\-
TACK
I .
GAS -2^
NIT - 12'
K
K
^
I" STACK
STACK
107
6 aio»
10"
eio'°
HEAT RELEASE , BTU PER HR
Note: Multiple points signify separate observations or different assumptions of heat content.
Fig. 4-2 - Flame Length vs Heat Release (Industrial Sizes and Releases
(from API RP 521, Ref. 12)
63
-------
LMSC-HREC TR D390190
This equation (4.22) is based partially on empirical correlation (Fig. 4-3).
The correlation includes data for diffusion flames of CO, CO_, city gas,
acetylene, H_ and propane.. ^The validity of the correlation for fuels
other than these is unknown,
An exact analysis of the flame shape is not available because suitable
models are not available for the eddy diffusivity and the burning rate in
turbulent flames. The analysis is complicated by axial changes in
temperature and concentration along the jet. Definition of the flame
boundary is necessarily arbitrary. A number of previous theoretical
treatments formulate the diffusion equation for a single gas. The theoret-
ical equation is then modified by empirical or semi-empirical relation-
ships which compensate for the effect of other factors such as changes
of temperature along the jet (Ref. 75).
Experimental studies have shown fairly simple scale-up behavior for
both laminar and turbulent flames. For a given fuel, laminar flame
lengths become independent of velocity (resulting presumably from the
appearance of a velocity term in the eddy diffusion coefficient) and, for
a given fuel, the ratio L/D is approximately constant (Ref. 75). For
Mach numbers greater than about 0.05 flames are shortest for fully de-
veloped turbulent flow (see Fig. 4-4). Thermal emissions at grade are
correspondingly greater, and flame length for fully developed turbulent
flow are sometimes used as an upper limit for flare stack design purposes
(Ref. 14). Following Kent (Ref. 14), the length-to-diameter ratio approaches
L/D = 118 (4.23)
where
L. = flame length, ft
d = discharge diameter, ft.
During plant upsets, flame lengths may reach several hundred feet and
moderate winds of 20 to 30 mph result in increased hazards to certain
areas occupied by workmen and structures downwind from the flare.
Under these conditions, the assumption that the flame is a point source
at or directly above the flare stack is unrealistic. Most design calcula-
tions described in the literature have allowed for wind effects by dis-
locating the "flame center" and assuming that the flame length is
unaffected by the wind.
The following method is based on experimental data of jets impinging into
still air. The empirical equation for velocity as a function of axial dis-
tance is modified by the constraint that at some distance 1 = L, the
jet velocity is zero (Ref. 12). From API RP 521 (Ref. 12):
T -1)
U = 1.6 d U (4- -4-1 (4.24)
ct \J O
64
-------
LMSC-HREC TR D390190
300
250
200
-ilx. 150
100
SO
f*
P
pfcp«n«
rafiine ft
mm. frm
Acttylen* max. ft a 307,
l
Crtygi
A
^
-54,001
36.000-
t
J
r
**A,
« Fr « 201,000 Y »
/ X!
*
/
/
\_imf '
"^^
'A
-<
CQFr-
' Hy
Fr
:ot +cit
Ff «
2,400
>
y
/
r
ifogsfi m
- J58,0(
rgMmu
STvOOO
~r
n.
0
10 20 30 40 50 60
Fig. 4-3 - Plot of L/d (Flame Length/Nozzle Diameter) vs 2
(from Hawthorne et al., Ref. 74)
010 OK
run ovMnrrm HOMES
030
040
Fig. 4-4 - Diffusion Flames of City Gas in Air (from Ref. 74)
y_. is the distance from the jet axis to the flame boundary. Thus
65
2 y = jet diameter.
-------
LMSC-HREC TR D390190
where
U = average axial velocity at I, ft/sec
3.
U = nozzle exit velocity, ft/sec
d = nozzle diameter, ft
o
t = distance along the jet axis from the nozzle tip, ft
L, = the effective flame length, ft. (The length is
assumed to remain constant under the influence
of the wind.)
Dividing the length L into M equal increments, the vertical, Y, and
horizontal, X, displacements of the flame tip are calculated by the
following equations from API RP 521 (U = wind velocity):
i
'1/2
AX = A* |1 + - (4.Z6)
AY = Al-|l +[*&} \ (4-27)
then
X = ^ and Y = AY (4.28)
Recent studies of thermal emissions include wind tunnel studies of wind
effects on a diffusion flame (Ref. 77) which have been useful in the char-
acterization of the flame boundary (Ref. 78). At least one experimental
measurement of thermal emissions during flaring under controlled condi-
tions has been made (Ref. 79) but is not yet available. Details of the effect
of discharge velocity on flame length under full-scale flaring of hydrogen
has been reported (Ref. 80).
The definitive recent work on the subject of flame boundary calculation
and thermal radiation from flares appears to be that of Brzustowski and
Sommer (Ref. 30). The design method includes the calculation of the heat
flux and temperature rise in surrounding structures with and without wind
cooling. The "point method" for estimation of thermal emissions is shown
to be adequate (compared with the Battelle flare emissions study (Ref. 79)),
given a reliable estimation of flame length and diameter and wind distortion.
66
-------
LMSC-HREC TR D390190
According to this report (Ref. 30), a factor which may seriously affect
operating safety is the uncertainty surrounding the calculation of flame
emissivity. Design methods by Tan (Ref. 13) and a report by Reed (Ref. 35]
give empirical equations from which estimates of the emls0ivity may be
obtained. However, the best available design method is apparently still
based on estimations based on literature values {Ref. 72),
In the flame boundary calculation of Brzustowski (Ref. 78), effects of the
wind and plume velocity are combined. The flame boundary is considered
to be the concentration envelope where the concentration is equal to the
lean flamrnability limit. Flame propagation is assumed to be hydro-
dynamically controlled at high Reynolds numbers, according to the
assumption of Chomiak {Ref. 81), The method of calculating the flame
boundary is based on experimental observations of Hoehne and Luce
(Ref. 77). A scaling parameter is used to account for the relative dy-
namic pressures of the jet and the wind:;
R = (-a.) (-L) (4>29)
where
U, = flare gas discharge velocity, ft/sec
U = wind speed, ft/sec
p. = density of flare gas at discharge, Ibm/ft
^ 2
p - density of ambient air, Ibm/ft .
Using this scaling parameter, the air molecular weight, M , and the fuel
molecular weight, M., scaling of the vertical and horizontal coordinates
of the flame tip and the lean concentration limit permits the use of corre-
lations based on the study of Hoehne and Luce, Two cases arise. For
L
the flame tip coordinates are calculated from the following empirical
relationships:
i n t
ST * E.04/CT (4.30)
1~J JL-i
XT = ST - 1.65 (4.31)
67
-------
LMSC-HREC TR D390190
for CT > 0.5:
_ 0-625
SL = 2.51/CL (4.32)
XL = SL - 1.65 (for S"L> 2.35) (4.33)
= £(S. ) (for ST < 2.35) (4.34)
Jj j_i ~~
Where, in the above expressions:
CT = dimensionless fuel concentration
X-i
S. = dimensionless distance measured from
orifice along the axis of maximum flare
gas concentration
X- = dimensionless horizontal displacement
of the flame tip
f(ST ) = solution for XT of the equation
LJ i-i
S. = 1.04 X2 + 2.05XT°'28 (4.35)
The vertical rise of the flare tip is calculated from a correlation which
applies along the axis of maximum concentration. The relationship used
by Brzustowski is very close to that observed for a non-buoyant jet plume
at high Reynolds number:
7-2 05 X°-28
where _^L ~ 'Ob XL (4.36)
Z = dimensionless rise of the flame tip
L above the flare
The computational procedure is described in detail by Brzustowski (Ref. 78)
and a numerical example based on the procedure is given which includes
sizing calculations for the flare stack (Ref. 30).
Absorption of thermal radiation by the atmosphere depends on the chemical
nature of the emitting species. Thus, emissions of a given species in the
flame, such as CC>2 and r^O, are absorbed selectively by the same species
in the atmosphere. Assuming blackbody radiation, attenuation by species
in the atmosphere can be calculated. Hottel (Ref. 82) gives experimental
emissivity values from which emissivity of the gas can be calculated given
the temperature of the gas, the path length, JL, and the partial pressure,
Pw, of the absorbing species. Emissivity curves are given for CO^,
water vapor, SO2, CO, and NH3. Curves for reduction in emissivity of
T and H?O mixtures resulting from spectral overlap are also given.
68
-------
LMSC-HREC TR D390190
Brzustowski and Sommer (Ref. 30) give the following formula, based on
the Hottel charts, from which atmospheric attenuation may be estimated:
1/16 1/6
t - 0.79-^ -^ (4.37)
where
t = fraction of K, Eq. (4.21), transmitted through the
atmosphere
r = relative humidity, %
D = distance from flame to illuminated area, ft.
The above equation is strictly applicable only to a luminous hydrocarbon
flame emitting at Z240°F, 80°F dry bulb ambient temperature, relative
humidity more than 80%, and a distance from the flame of between 100
and 500 feet, but can be used to estimate the atmospheric attenuation
under a wider range of conditions. In the case of flares, atmospheric
absorption attenuates K by about 10 to 20% over distances of 500 feet.
Except when flaring gas streams which may contain toxic, odorous or
noxious components, the stack height is determined by the height required
to prevent thermal radiation intensity at ground level from reaching
dangerous levels. Effects on operating personnel and plant and local
processing and storage facilities are considered. For personnel, maxi-
mum intensity levels and exposure times given in API RP 521 (Ref. 12)
are widely accepted. The maximum intensity level for continuous ex-
posure without burns or blistering is 440 Btu/hr-ft^.
Maximum intensity levels at grade (ground level nearest the flare stack)
are normally calculated allowing a reasonable length of time for affected
personnel to react and move to safety. A widely accepted (Ref. 36) in-
tensity level at grade is 1000 Btu/hr-ft^, which allows 30 seconds escape
time to avoid pain, but numerous other standards are used or recommended
in the processing literature. API RP 521 (Ref. 12) uses an intensity of 2000
Btu/hr-ft at a distance of 150 feet from the base of the flare as the design
criterion, but this level may be high. The various standards may result
from the inexactness of the calculations for flame emissivity and total radia
tion. Maximum solar intensities are not usually considered but are signif-
icant (Ref. 35). At the latitude of Boston, solar radiation is on the order of
260 Btu/ft2-hr and may approach 300 Btu/ft -hr in the Gulf Coast area.
Maximum allowable thermal radiation may vary with the proximity of
equipment and storage facilities, the extent to which these can be pro-
tected, and the need for attendant personnel. Insulation and pressure
relieving requirements of process equipment and storage facilities are
described in API RP 521 (Ref. 12), Geometry and orientation of tanks
or other equipment with respect to the flame and distance from the flare
stack define the "view factor." This factor, the emissivity of the material
of construction, and ambient conditions such as wind effects determine the
maximum design temperature of surrounding objects. An upper limit which
69
-------
LMSC-HREC TR D390190
may be useful for design purposes is the temperature reached by the
object in free convection in the absence of wind cooling. This tempera-
ture may be calculated from the equation of Bruztowski and Sommer
(Ref. 30):
T4
K = 0.1713-j^- +MiTs-T)4/3 (4.38)
9
where
K = average incident heat flux over the surface,
Btu-hr-l.ft"2
T = surface temperature, R
s
T = ambient temperature, R
1! = emissivity of the surface.
S
At a radiant density of 1000 Btu/hr-ft , and assuming a ground emissivity
of 0.8, ground temperatures at the end of one minute can be as high as
195°F, reaching 315°F in about 20 minutes (Ref. 3S). In operation under
these conditions, a barren radial area about the flare is formed having
a radius approximately equal to the flame length (Ref. 35).
70
-------
LMSC-HREC TR D390190
SECTION V
RECOMMENDED DESIGN METHOD
General design criteria applicable to flare selection and sizing were
discussed in detail in Section IV in order to establish the state-of-
the-art foundation of existing design calculations. The purpose of
this section is to summarize the sizing methods that are considered
representative of methods being used at present. Along with Section
IV, Refs. Z, 4, 5, 12, 13 and 14 are recommended for detailed explana-
tion of the sizing calculations. A recent review article by Horton
et al. (Ref. 17) provides a detailed qualitative description of auxiliary
flare components. A review of Vanderlinde (Ref. 9) is recommended
along with Section in of this report for a discussion of the flare stack
components including methods of steam injection available for smoke-
less flaring.
The basis of the following discussion is an integrated disposal system
including both an elevated and an enclosed ground (low level) flare. In
the elevated flare, either air or steam injection is recommended as
the preferred control method for smoke and particulates. Water injec-
tion is a less desirable means of control (Ref. 5).
In the integrated system, the low level flare is used for handling routine
discharges to reduce noise, smoke, luminosity and thermal emissions.
The elevated flare, with air or steam for particulate control is provided
for handling full-scale emergency upsets. A "double seal" or "diversion
seal," common to both flares is used to divert the waste gas stream from
the elevated flare toward the low level flare (Fig. 15). Only during
major upsets do gases pass through the upper level of the liquid seal
and burn at the elevated flare.
Toxic, noxious or odorous gases or gases which yield hazardous com-
bustion products should be collected in segregated piping systems and
preferably disposed of by some means other than flaring. If flaring of
such streams is required, these should be discharged directly into the
elevated flare burner rather than into the diversion seal. Depending
on stream volume, these streams can often be treated according to
usual methods for segregation and disposal (Refs. 4 and 15).
5.1 Elevated Flare System
5.1.1 Required Design Information
The following information is recommended (Refs. 2, 4, 5, 12, 13 and 14).
Much of this information can normally be obtained from relief valve
sizing calculations.
1. Type of Material to be Flared
2. Average Molecular Weight, M
71
-------
LMSC-HREC TR D390190
3. Percent Unsaturation
4. Lower (Net) Heating Value, Q Btu/scf
5. Specific Heat Ratio, k = C /C
6. Mass Flow Rate at Maximum Discharge, W Ib/hr
7. Average Vapor Temperature, T °F
8. Flowing Pressure, p psig
9. Percent Toxic, Odorous or Noxious Gases
5.1.2 Flare Burner Diameter
Design parameters applicable to the calculation of the flare burner
diameter include mass flow rate and discharge conditions, the type of
flame holder used by the manufacturer, and density and heat capacity
ratios which serve to define the sonic velocity in the fluid stream.
Principal design considerations include pressure drop and flame
stability considerations. Either may serve to define the diameter.
Pressure drop rather than flow velocity is usually the controlling factor.
The flare burner is usually limited to a pressure drop of 2 psig (60 in.
H20).
To facilitate the sizing calculations, the maximum discharge rate is
converted to the equivalent volume of air in scfh. The following orifice
equation is used to calculate the burner diameter.
Ve = 1656 (K)(A)^Ap (5.1)
where
V = volume equivalent flow rate, scfh
K = orifice factor, dimensionless
A = area of flare burner tip, ft^
Ap = allowable pressure drop at tip, in. H^O
The orifice factor K is normally about 0.9. The orifice factor and the
effective internal tip area are somewhat variable depending on the flare
design and manufacturer.
Maximum Discharge Velocity: The sonic velocity is determined by
assuming perfect gas behavior (Refs. 2, 12, 13, 14 and 83) according to
the following equation;
= Jg KRT/M (5.2)
cl » C
72
-------
LMSC-HREC TR D390190
where
V = the acoustic velocity, or velocity of sound in the fluid,
a ft/
sec
2
g = dimensional constant, 32.17 Ib-ft/lbf-sec
K = ratio of specific heats, C /C
P ^
R = gas constant, 1546 ft-lbf/°R-lb-mole
M = average molecular weight
The maximum discharge velocity for flame stability depends somewhat
on the design of the flame holder and the height of the flare pilots above
the flare tip. A maximum vapor velocity of Mach 0.2 is frequently
recommended (Refs. 2, 12, 13 and 14). Discussions with flare vendors
indicate that higher design velocities may be acceptable in some cases,
but this figure seems to be a safe design basis. If this velocity is
exceeded in Eq. (5.1), the pressure drop of the tip should be reduced
accordingly.
Pressure losses in the flare seal, riser, entry and drums may also
require a reduction in vapor velocity depending upon the available
flowing pressure of the vent stream. For convenience, the pressure
losses from these components is commonly expressed in terms of cor-
responding loss in velocity head (Refs. 83 and 84). For air under turbulent
flow conditions, the equivalent length of pipe (in pipe diameters) is
approximately equal to the velocity head loss multiplied by a factor of
about 55. Pressure losses from the various flare components are sum-
marized in Table 5-1.
Water Seal and Flame Arresters: Water seals and flame arresters are
discussed in detail in Section 4.4.6 dealing with flare safety. For purpose
of the present discussion, selection rather than sizing criteria are
significant in determining the pressure drop. It is the result of this
study that flame arresters of conventional design are unacceptable because
of their tendency to fill with solids and become plugged. Either water seals
or flame arrestors are required when flammable mixtures are being flared,
but a water seal is normally included in all flare systems as a precautionary
measure.
Disentrainment or "Knockout" Drum: A disentrainment system is required
whenever the dewpoint of the flared stream is higher than ambient tempera-
ture. A variety of designs have been employed for liquid disentrainment.
These may be either vertical or horizontal and may be combined with the
water seal. In some cases, a sloped piping arrangement with a drain at
the lower end has been found adequate for removal of small amounts of liquid.
Typically, liquid removal requirements are difficult to determine and a
separate drum is installed between the flare stack and header, located
external to the flare stack.
73
-------
LMSC-HREC TR D390190
Table 5.1
PRESSURE LOSSES IN ELEVATED FLARE SYSTEM
Component
Flare Tip
Fluidic Seal
Flare Riser
Molecular Seal
Knockout Drum
Water Seal
Equivalent Pipe Configuration
Orifice
Orifice
Equivalent Length of Straight Pipe
Z to 3 Close Return 180° Bends
1 Close Return 180° Bend
1 Close Return 180° Bend and
Depth of Diptube Immersion
Velocity
7 *%
Head Loss
Equation 5. 1
Equation 5.1
3.0-4.5
1.5
1.5
If the temperature of the vent gas is significantly higher than ambient,
a material and energy balance on the water seal is required to determine
the final temperature and the mole fraction of water in the flared stream.
' Units in equivalent pipe diameters. See Refs.83 and 84 for details.
-------
LMSC-HREC TR D390190
Knockout drum design and sizing criteria are given in Section 4.4.8.
Reference 63 is recommended for a qualitative discussion of the knockout
drum as it affects operating safety. Drum location, liquid holding capacity,
pump capacity, other equipment requirements and a discussion of sizing
methods are given in No. 10 of Ref. 2. For purpose of the pressure drop
calculation, the drum is approximated by a close return 180-degree bend
(Table 5-1).
5.1.3 Utility Requirements
Steam and Air Requirements: Either steam or air is required for flaring
most gaseous hydrocarbons other than components such as methanol, carbon
monoxide, hydrogen sulfide and methane or natural gas which burn smokelessly
Discussions with flare manufacturing firms indicate that steam requirements
commonly cited (Refs, 2, 12 and 13) may not be suitable for design. The
following empirical guideline is suggested for use:
WSt/WHc = tt - 10/Mw (5'3)
where
WC,/WTJ. = steam-to-hydrocarbon mass ratio
ot rlc
M = average molecular weight of flared gases
a = empirically determined parameter dependent
on the type of material being flared
For paraffins heavier than propane, a = 0.50. For olefins, a = 0.60. Other
variables include nozzle design and point of injection. Because of these
variables, steam utilization predicted by Eq. (5.3) may vary by +25%. Re-
quired steam pressure is at least 10 psig at the point of discharge.
Air requirements for smokeless burning are somewhat higher. For paraffins,
the recommended value for a is 0.55, increasing to 0.69 for olefins. For
forced draft flares, blower requirements are about 0.8 hp for each 1000 Ib/hr
of gas flared.
As mentioned previously (Section 4.3.3), use of water to control particulates
is not recommended because of flame quenching, limited turndown, wind
effects and other problems. Water requirements are also fairly high;
1.0 to 1.2 Ib water are required for each pound of hydrocarbon gas.
High Pressure Hydrocarbon Streams: Hydrocarbon streams having pressures
greater than about 2 psi may be burned smokelessly by means of special
flare tip designs which increase air-fuel mixing. This is a rather specialized
application and design criteria are not generally available except as they
may apply to specific flare tip designs.
75
-------
LMSC-HREC TR D390190
High pressure hydrocarbon jets of oil field or production gas may also be
used instead of steam to achieve smokeless burning. Discussions with
flare vendors indicate that gas requirements are extremely high. For
example, approximately twice as much natural gas as propane is required
to burn the propane stream smokelessly. The gas requirement for propylene
increases by about another factor of two. The application of gas -assisted
smokeless flaring is therefore limited to those applications in which the
gas stream has no recovery value.
Vapor Purging Requirements: Safety considerations relating to vapor purging
were discussed previously in Section 4.4.2. Purge gas requirements are
normally based on the amount of gas flow required to prevent explosive
mixtures in the flare stack, but another consideration is maintaining a
stable flame (Ref. 2). The approach used will vary with the requirements
of the system.
Husa (Ref. 62) has given the following empirical correlation to estimate the
velocity required for prevention of explosive mixtures within the flare stack:
r6]°-64 f28ln f 0.16D 0.16(D-M)| ., A.
Ixl Isr I ie -°-96e I (5-4)
..
v =
where
V = purge gas velocity, ft/sec
H = height of flare stack, ft
h = protected length of stack, ft (typically H-h ~25 ft)
X = concentration of oxygen at the explosive limit, mole %
M = molecular weight of purge gas
D = flare tip diameter, in.
n = dimensionless constant characteristic of the stack diameter
The exponent n is approximately unity for most stack diameters:
Diameter Exponent Diameter Exponent
(in.) n (in.) n
4 1.00 16 1.18
6 1.30 18 1.10
8 1.40 20 1.00
10 1.40 22 0.91
12 1.34 24 0.82
14 1.27
Discussions with flare vendors indicate that the preceding equation should
be used cautiously for flare diameters larger than about 24 inches.
76
-------
LMSC-HREC TR D390190
In general, vapor purging requirements will vary with the type of gas being
flared and the nature of the purge gas. Auxiliaries such as molecular and
fluidic seals (Figs. 3-8 and 3-9) allow purge gas requirements to be reduced
by 90% or more. Inert gas generators available from flare manufacturers
can further reduce fuel required for purging by combustion of fuel gas at
stoicheometric air-to-fuel ratio to produce a mixture of N^. CO, CCs and
at approximately 11 times the original fuel gas volume.
5.1,4 Flare Height
Design considerations applicable to the calculation of flare height include
thermal emissions and dispersion of gaseous emissions. Separate calcula-
tions are required for heat and dispersion. The larger result is used as
the flare height. Noise and visible emissions calculations are performed
as a check according to procedures described in Section 4.3.2. In some
cases, these calculations may require an upward adjustment of the flare
height.
Thermal Emissions: Factors affecting thermal emissions include flame
length, wind effects and available unoccupied radius from the base of the
flare stack. The basic design equation is that given as Eq. (4.21) of this
report. The applicable design equation assumes spherical spreading of
thermal emissions from an assumed point source. Experimental measure-
ments (Ref. 30) show that the assumption of a point source or "flame center"
is valid for distances greater than about three flame lengths from the flare
stack. Closer to the stack, the point source model is too conservative, but
serves as an upper limit for design purposes. Rearranging Eq. (4.21), the
distance, D, required for reducing thermal intensity below the safe limit
defines the stack height:
(5.5)
^ "X // XY
where
D = safe distance from the point of emission, ft
K = maximum allowable radiant heat flux, Btu/hr-ft
F = the emissivity factor
Q = total heating rate of the flame, Btu/hr
Commonly accepted values for K are 1500 Btu/hr-ft for human exposure
and 3000 for equipment exposure including view factors but ignoring atmos-
pheric attenuation (see Section 4.3.1).
Emissivity factors are chosen which correspond to the flame under condi-
tions in which smoke suppression is not used. Typical sources are Refs.
12, 13, 30, 67 and 68. Emissivity values from the sources are summarized
on the following page:
77
-------
LMSC-HREC TR D390190
Component Emissivity
CO or low Btu gas 0.05
H2 0.075
CH4 0.10
C^Hg and most paraffins 0.13
C-jH^ and most olefins 0.15
C^Hk and hydrocarbons
with M > 100 0.20
Factors which are somewhat variable and which determine the center of
emissions for use in Eq. (5.5) are flame length and wind effects. Wind
speed is frequently arbitrarily chosen as 30 mph for the design calculation.
Given this wind speed and assuming highly turbulent flow under maximum
discharge conditions, a number of semi-empirical expressions are avail-
able for locating the flame center. These are given in Refs. 12, 13, 14, 30,
70 and 71).
If justified by potential economic savings, comparison of several methods
may serve to allow a basis for selecting the least expensive alternative,
but the approximate nature of the calculations prevents an estimate of
accuracy of any of these. Thermal radiation calculations used by flare
manufacturing firms and chemical manufacturers contacted differ considerably
and are frequently proprietary. Thus, it is not possible to summarize these
in a single method. The following adaptation (Ref. 2) of a method described
by Kent (Ref. 14) is representative in terms of the complexity of the calcula-
tions involved and the factors considered:
Lf ~ 120 D (5.6)
%vhere
L, = flame length, ft
D = flare tip diameter, ft
For wind effects, tilting of the flame through an angle 6 displaces the
flame center:
l V
9 - tan" ~ (5.7)
e
where
V = velocity of wind, ft/sec
w
V = flare discharge velocity, ft/sec
0 = the angle of inclination of the flame, in degrees from
the normal
78
-------
LMSC-HREC TR D390190
The axial distance along the flame from the top of the flare stack to the
flame center is chosen somewhat arbitrarily in the various design methods
described. Some fraction of the flame length is typically chosen. In the
absence of wind effects, the difference between the effective emission
height, Y, and the stack height, H, is approximately (Ref, 12):
Y -H = LF/2 (5.8)
Wind increases the required stack height by reducing the effective emissions
height and shifting the flame horizontally (Ref. 2):
Xj = R - [H + (Y - H) cos0] -MY - H) sm8 (5.9)
where
R =
JH
X, = the required safe distance from the base of the
flare stack, ft
L_, H, Y, 0 as previously defined
If the available space at the flare location is known, the stack height re-
quired from thermal emission considerations can be calculated directly
from Eqs. (5.6 through 5.9). A minimum safety boundary of at least 100
feet (Ref. 2) or approximately equal to the flame length is recommended.
Flare Height Dispersion Calculations: Flare height calculations for dis-
persion of gaseous emissions were discussed in detail in Section 4.3.7.
In most cases, methods of calculation will be specified by local air pollu-
tion control agencies. The dispersion calculations given in Section 4.3.7
therefore serve only to illustrate some typical methods. Numerical ex-
amples illustrating the application of several of these methods are given
in Refs. 2 and 12.
5.1-5 Supporting Structures
Elevated flare systems are usually supported by guy wires or derricks
(Ref. 5). The type of support required affects foundation and piping costs
and must be specified in the design. In general, self-supporting flares
are feasible when stack heights are less than about 40 feet. Flare stacks
between 50 and 100 feet high may be supported by guy wires, while stacks
taller than 100 feet usually require a supporting derrick (Ref. 5). More
complete guidelines are given in Ref. 2, as discussed in Section 3.1.4.
5.1.6 Auxiliary and Control Components
Flare auxiliaries typically include steam ratio control, flame front
generators and pilot ignition systems, purge gas generators and control
systems. Other controls may be required for forced draft and other more
specialized disposal systems including blower and damper controls for
79
-------
LMSC-HREC TR D390190
example, fluidic control of steam for smokeless buying is n^vaxlable
from one flare vendor. A new ignition system developed by F^re^s
^
and n ^
tion
y a spark. The flame generated isthen "shot" by presure
to the pilot burner(s) over distances up to about 1000 feet (Ref. 2). Igniti
systems presently require manual (push button) ignition, although automatic
thermocouple alarm systems are available which have the capability to
detect pilot flame extinguishment. Either instrument air or venturi inspira-
tion are useful for air supply. Air and fuel premixing is controlled manually.
Filtration and drying systems for air and fuel are normally recommended
(Ref. 2). The ignition and pilot system should be fully specified in the sizing
stages of design because provisions for locating the ignition system and
controls must be made. Location of the ignition panel depends on flare
stack height and distance with which pilot ignition can be accomplished
reliably with a specified system.
Controls are also recommended (Ref. 63) for major auxiliary components
such as the water seal or diversion seal and knockout drum as discussed
in Section 4.4.3 and Ref. 12.
5.1.7 Endothermic Flaring Low Btu Gas Streams
Flammability limits and heating value requirements are actually somewhat
variable, depending on factors such as burner diameter and gas velocity
(Ref. 1). At present, regulations for flaring gas streams seldom include
guidelines to prevent the practice of flaring streams which will not burn.
This requirement must therefore be self-enforced.
A lower (net) heating value of between 200 to 250 Btu/scf is normally con-
sidered adequate for flaring (Ref. 1) without additional heat inputs. Gas
streams having heating values between about 100 and 200 Btu/scf can be
flared provided that additional fuel is added to increase the heating value
to the required minimum (Ref. 2). Gas streams with heating values less
than about 100 Btu/scf are probably better suited to disposal by direct
incineration.
Heating values may be calculated by standard methods found in furnace
handbooks such as (Ref. 1). Flammability limits for individual components
can be found in references such as Refs. 41, 42 and 43. Methods of esti-
mating flammability limits of gas mixtures are given in Refs. 71 and 74.
80
-------
LMSC-HREC TR D390190
5,2 Low-Level Flare
Low-level flare systems are a relatively new development. As a
result, much of the information relating to design performance is
proprietary.
The diameter of a low level flare will be approximately 18 times larger
than an elevated flare of conventional design. This observation is based
on discussions with flare vendors (Ref, 10) and is based on performance
comparisons between an eight inch elevated flare burner and a 12 foot
diameter low-level flare, both designed for capacities of 25,000 Ib/hr.
Low-level flares are nominally sized for combustion rates in Btu/hr.
The height of low-level flare systems is determined by the height of re-
fractory required to enclose the flame. The flare height is therefore
strongly dependent on the type and number of discharge nozzles and their
elevation within the flare.
81
-------
LMSC-HREC TR D39Q190
SECTION VI
SAMPLING AND ANALYSIS TECHNIQUES
The sampling and analysis of flare combustion products are necessary to
determine both the nature and amount of flare emissions and the efficiency
of a flare as a combustion device. However, flare systems especially
elevated flares present some very difficult sampling problems. As a
result, very little emission data are available from flares. This section
will discuss some of the problems of flare sampling, techniques used to
sample flares, and some of the methods of gas analysis. Also discussed
will be the measurement of air pollutants by remote sensing devices,
While these methods are still generally in the development stage, measuring
emissions from flares would be ideal.by this technique since it eliminates
the need for sample extraction.
6.1 Present Sampling Practices and Problems
An elevated flare presents almost impossible sampling conditions. Sampling
of a stationary source is done to determine the concentration and character-
istics of the contaminants, the mass rates of emissions as well as the effi-
ciency of the device for reducing emissions. Flares present problems not
only in just physically obtaining a sample but also in determining the mass
rates of emissions and thus the efficiency.
To obtain a sample from an elevated flare, a probe must be inserted into
the plume of a flare above the flame. Since the flame of just a medium
sized flare can easily reach 300 feet, the logistics of obtaining a sample
becomes formidable. Moreover, because of the heat and radiation of the
flame, equipment and personnel must be located at a safe distance from
the flare. To further complicate matters, the flame is never still, moving
continuously because of wind and convection effects. This makes locating
the plume of the flare difficult at best. Also since even small flares have
very large capacities, any field tests will of necessity require very short
sampling times, less than a minute, and a limited number of tests. These
time constraints, added to the difficulty of obtaining a sample, makes it
hard to obtain good reproducible data.
An additional problem of sampling elevated flares results from the fact that
these flares discharge to the atmosphere before igniting. Combustion air
is provided by the ambient atmosphere. The concentration of combustion
products in the plume; of a flare cannot be related to the mass rate of emis-
sions without estimating or measuring the dilution of the plume with com-
bustion air. At present there is really no good way to test for flare
emissions. Until remote sensing methods are developed which require
no sample extraction, flare emission testing will remain troublesome and
expensive.
Because of these sampling problems and the intermittent nature of most
flaring, only a few tests of flare emissions have been attempted and these
82
-------
LMSC-HREC TR D390190
tests were mainly on small flares. Published results were found for only
one elevated flare emission test (Ref. 31). Generally the data on the few
flare tests completed are not published and are considered proprietary.
The purpose of most testa was to determine the type and concentration
of emissions. Little if any work has been done to determine the mass
rate of emissions or the efficiency of the combustion process.
Discussions with vendors and flare users indicate that the sampling of
elevated flares has usually been done using either cranes, derricks or
scaffolding to reach the plume. However, because of sudden movements
of the flame, care must be taken to protect both equipment and personnel.
Samples have been taken either with stainless steel probes or evacuated
grab samplers. The entire sampling train should be heated in the case of
a probe to prevent condensation of water or heavy hydrocarbons while
sampling.
In one test helium was injected into the gas before flaring and used as a
tracer to measure plume dilution. Helium is inert during the flare com-
bustion and its background atmospheric concentration is essentially con-
stant. If one assumes that the diffusion and turbulent mixing of helium is
the same as the other combustion products, the dilution of the plume by
atmospheric air is linked to the concentration of helium in the plume. The
concentration of the combustion products in the plume can then be related
to the mass rate of emissions. In this test the helium concentration of the
sample was analyzed by mass spectroscopy.
Sampling from a forced draft flare, while presenting many of the same
problems as an elevated flare, is somewhat easier. The forced draft
flare provides a stiffer flame and good outlet velocity which reduces the
movement of the flame in the air and makes sampling easier. Also gen-
erally the forced draft flare is not nearly as tall as an elevated flare making
obtaining the sample simpler. Low-level ground flares present much less
of a sampling problem. The enclosure forms, in effect, a stack in which
all the exhaust gases are directed. Thus standard stack sampling methods
can be used. In some of the larger ground level flares, sampling ports
have been built into the enclosure.
6.2 Analytical Techniques
The main emissions of interest in flare combustion are unburned hydro-
carbons, partially oxidized hydrocarbons, carbon monoxide, nitrogen
oxides and if sulfur is present in the waste gases sulfur dioxide. The
analysis of carbon dioxide, while not a pollutant, is necessary to deter-
mine the efficiency of the combustion process. Information is given below
on the existing analytical methods for these emissions and the commercial
instruments which have been developed. No detailed procedures are in-
cluded but references are given to sources of methods. Several reviews
are available covering the subject of air pollutant analysis (Refs.85 and 86).
A particularly good review on general gas sampling and analysis techniques
in combustion phenomena is presented by Lengelle and Verdier (Ref. 87).
Much of the information presented below was obtained from Ref. 15.
83
-------
LMSC-HREC TR D390190
6.2.1 Hydrocarbons
The gas chromatograph with a flame ionization detector (FID) is used tn
m6 3. Sure tot ail h.VCll*r>r*ai* hrm« TK * " i r * *-'/ *CI*.*wtr\4Hj
linear relationship of ion formation toThT concentration o^aViten organic
compound m a flame. The gaseous mixture is burned at a smalHet and
the change m electrical conductivity is measured. ResponseTsa function
eral theTe !» Tl *"* ofldi?*?* carb°<> atom, in a molecule £Tg
hvdror^h * ^ i Variatl°n in the "sponse of widely differing
hydrocarbon types However, oxyginated organic compounds have a lower
relative response, decreasing with increasing oxygen content. CO and CO,
and inorganic gases show essentially no response. Instruments are avail- '
able which use this technique to measure total hydrocarbons and methane
separately and also carbon monoxide separately following its catalytic
hydrogeneration tc methane (Ref. 88). Individual hydrocarbons are deter-
mined by gas chromatography using a flame ionization detector. ASTM D
2820 describes a method for determining C to C hydrocarbons.
6.2.2 Oxidized Hydrocarbons, Carbon Monoxide, Carbon Dioxide
The partially oxidized hydrocarbons respond to the FID and can be meas-
ured the same way as hydrocarbons. However, in most instances these
compounds can be determined specifically by virtue of their functional
group. Table 6-1 lists a number of approaches for several oxidized species
which may be encountered. Carbon dioxide is most easily measured by the
Orsat technique. Gas chromatographic and nondispersive infrared analyzers
can also be used. In general gas chromatography is used for carbon mon-
oxide measurements. For very low CO levels (less than 50 ppm) the CO
must be converted to methane since the flame ionization detector does not
respond to CO. Infrared spectrophotometry is also often used for spot CO
analysis. Continuous monitoring for CO is usually performed by nondispersive
infrared (NDIR) analyzers, NDIR analyzers have the advantage of rapid re-
sponse and good sensitivity over a wide range of concentrations.
6.2.3 Nitrogen Oxides
The applicability and limitations of the principal methods for determining
NO are shown in Table 6-2. The two chemical methods are suitable for
x
NO concentrations between 5 and 1000 ppm. When oxides can be deter-
x
mined without differentiation as NO the phenoldisulfonic acid method of
analysis is usually used. This is one of the few air pollution methods
generally recognized to be accurate and reliable (Ref. 88). The instru-
mentation methods for determining NO include ultraviolet and infrared
absorption, electrochemical sensor and chemiluminescence. The chem-
iluminescent is fairly new and is based on the reaction of NO with ozone
which results in the emission of light. This method is very sensitive and
can be used to determine low levels of nitrogen oxides.
84
-------
LMSC-HREC TR D390190
Table 6-1
METHODS FOR DETERMINATION OF OXIDIZED
HYDROCARBONS AND OTHER COMPONENTS (From Ref. 15)
Compound
Determined
Aldehydes
Aldehydes and
Ketones
Acrolein
Formaldehyde
Carboxylic Acids
Esters
Carbon Monoxide
Carbon Dioxide
Method
Methylbenzothiazolone
Hydr ozone
Spectrophotornetric (650 nm)
Dinitrophenylhydrazine
Spectrophotornetric
4-Hexylresorcinol
Spectrophotornetric (605 nm)
Chromotropic Acid
Spectrophotornetric (570 nm)
Absorption-titration
Hydroxamic Acid
Spectrophotornetric (530 nm)
Non-dispersive Infrared
(NDIR)
NDIR
Or sat.
Lower
Limit
(ppb)
20
40
20
20
100
1000
2000
0-5%
Absorptivity
(L/mol cm)
50,000
27,000
17,000
19,000
1,100
85
-------
Table D-Z
METHODS FOR NO (From Ref. 15)
oo
MtTliOL'
Phwnol uisulfonic
acid ovjthod
ASTM D 1608
(Method 4 EPA)
Sulfariilic Acid
method
(Saltzman)
ASTM 0 2012
SAE J177
Ultraviolet ab-
sorption
method
Infrared absorp-
tion method
Electrochemical
Sensor
Chemi luminescence
PRINCIPLE OF PROCcUU;;:
N°X flgfc* HN°3
Phenol disulfonic acid
nitrated to produce
yellow color.
N02 converts sulfanilic
acid to diazonium salt.
Salt couples with amine
to produce deep violet
color
N(>2 has maximum
at 400 ray with
absorptivity « 170
liters/mole-cra.
NO is transparent above
230 mti.
NO has band at 5.4 u
with absorptivity
« 2 liters/mole-cm.
NO*N02 permeate
merit) r an e on sensor
and are electrochem-
ical ly oxidized. Re-
sulting current is
proportional to NO
concentration. *
The light resulting from
the reaction of NO with
ozone is measured with
a photonultiplier.
NO 2 must be converted
to NO to be Measured.
INSTRUMENTATION
Laboratory spcctio-
photorccter at
400 my
Laboratory spectro-
photometers it
550 my.
Continuous analyzers:
Beckman Acralyier.
Technicon Aiito-
analyzer.
Continuous Analyzers:
Becknan NDUV Model
255 plus oxida-
tion system.
Continuous Analyzers:
Beckman NOIR Model
315A.
Mine Safety Appli-
ance Model LIRA 200,
Continuous Analyzers:
Dynasciences NX-110
and NX-130
EnviroJtetrics Ftodel
N-122
Theta Sensors Ftodel
LS-800-ANX
Continuous Analyzers:
Thenao-Electron Corp.
Bendix, Environmental
Science Division
REM. Inc. Model 642
APPLICABILITY
Range: S to 1000 ppm
for all nitrogen oxides
except N20.
Range: 0.01 to 4000 ppm.
Specific for N02
NO determined by
prior oxidation.
Useful for air
and exhaust analysis.
Faster than disulfonic
acid method
Range: 10 to 6000 ppm.
Determines N02 directly.
NO determined by prior
oxidation.
Range: 10 to 4000 pp».
Determines NO directly.
Range: 2 to 10000 ppre
Models available for NO
or N02 *
Range: 0.01 to 10000 ppm
LIMITATIONS
Not sensitive bt-low
5 ppm.
Equipment somewhat
more complex than
for phenol disul-
fonic acid method
NO is a reactive
gas and can be
partially lost un-
less precautions
are taken.
Water vapor inter-
feres and must
either be con-
stant, or pre-
ferably removed.
S02 interferes but
can be eliminated
or compensated for.
No known interferences
Cfl
o
W
n
xO
o
-------
LMSC-HREC TR D390190
6.2.4 Sulfur Dioxide
Table 6-3 summarizes some of the important features of the principal
methods in current use for determining sulfur dioxide. In general,
sulfur dioxide measurement involves two problems; obtaining a valid
sample, and eliminating interferences. Because of its reactivity, SO_
is best captured by using bubblers. The method generally considered
most specific for determining SC>2 is the pararosaniline method (Ref. 89).
Since the method requires close attention, it is not as widely used as
the conductivity or the coulometric techniques.
6.3 Long Path Remote Sensing Techniques
The absorption techniques for remote sensing are promising new methods
now being developed which would be ideally suited for measuring flare
emissions.
Resonance absorption by molecules and atoms is the basis of a well-
established method for determining the concentrations of such species.
Usually, a continuum background from a flash lamp or other light source
is viewed by a spectorgraph through the sample of interest. Absorption
of the continuum occurs at wavelengths corresponding to transitions be-
tween specific energy levels of the species, the degree of absorption being
a function of the specie concentration in the energy state corresponding to
the lower energy level of the transition.
An equivalent technique consists of monitoring the absorption of the beam.
of a tunable laser after propagation through the sample under study, as it
is tuned over the spectral region of interest. The latter method has the
advantage of a considerably increased sensitivity because of the increased
photon concentration per unit wavelength interval of such a light source,
while its small beam divergence is also an advantage for long path applica-
tions. Also, their narrow linewidth admits to better discrimination against
background radiation and an increased spectral resolution. Such tunable
light sources can be considered to have revolutionized spectroscopy and
especially the application of spectroscopic methods to pollution monitoring.
For example, tunable sources have made possible consideration of such
techniques as resonance backscattering and resonance Raman backscattering
monitoring as the basis for potential monostatic remote sensors of ambient
atmospheric pollutants. However, it is variations of the resonance absorp-
tion method that admit to the greatest probability of success as monostatic
remote pollutant sensors (Refs, 90 and 91). Several such variations are
presently being given considerable attention. One method, termed the
Differential Absorption and Scattering (DAS) method, consists of monitoring
the radiation backscattered elastically by ambient particulate matter and
molecules in the atmosphere. By tuning the laser both on and off an ab-
sorption line of the pollutant of interest one can directly obtain the specie
concentration. Another scheme, a direct absorption method, involves
monitoring the beam absorption after transmission of the beam, through
87
-------
Table 6- 1
METHODS FOR SO2 (From Ref.
Method
Principle
Inat r umentation
Applicability
Limitations
Hydrogen Peroxide
Tttration Method
(EPA method J)
Hydrogen Peroxide
Conductometrie
Method
Electrolytic
(Boulometric I
Method
WEST-GAEKE
Colorimetric
Method
ASTM D 29H
E! petrochemical
Sensor
S02 + H20 - H2S04 ^
lltration with ba«e or Ba
Laboratory Equipment:
Absorber and titratioh units
Range: 0.01 to 100 ppm
Require* reagent addition..
S0a + H202 - H2S04
Measure Conductivity
Monitors:
Leeds 5r Northrup, AEROSCAN
Wonlhoff IBS ULTRAGAS ANALYZER
Instruments Development, IDC 902-1
Scientific Instruments, Sl-67
Range: 0.01 to 5 ppm
Interferences by salt aerosols
and acidic and ba*ic ga*es
which may be eliminated by
filter*.
2HBr
SO + 1, 4 H?O ZH1 +
Br? and 1^ generated
elect rolytic ally
Formation of dyestuH by
reaction with bleached
pararosaniline
Oxidation in a membrane-
covered efcii
Monitor a:
Consolidated, TITRILOG
Beckman Inatnimenta, Model 906
Barton 286 SULFUR TITRATOR
Phillips Instruments, Model PW 700
Atlas Electric Devices, Model 1200
Laboratory Equipment;
Spectrophotometer
Monitors:
Atlas Electric Devices, Model 1500
Technicon Corp.. AUTO ANALYZER
Mpnttpri;
Dynaictences, SS-JJO
Envirometrics NS-200
Theta Sensors LS-800-AS
Range- 0.01 to 5 ppm
Monitors simple to
operate and reliable
for unattended iervice.
Interference by oxidizing
materials, aldehyden, oteflns
and hydrogen lulfide. Some
interference can be alienated
by filters.
Rang**: 0.01 to S ppm
Mast nearly specific
Procedure cumbersome.
Continuous analyzer needs
close attention
method for SO,
Range: 0 to 5000 pprn
Simple to operate
NO and N02 Interfere slightly.
r
Z
en
O
50
PI
n
V
CJ
-------
LMSC-HREC TR D390190
the medium of interest, to a remote retroreflector. Again readings are
obtained while the laser is both tuned to and detuned from an absorption
line of the specie of interest, which leads to a direct measurement of the
pollutant concentration.
Comparison of DAS and direct absorption methods show that DAS provides
ranging capability by time-of-fllght measurement, spatial resolution and
three-dimensional, single ended measurement capability. The direct
absorption method is simpler in that many of the low-power laser and
broad-band sources presently available can be used (Ref. 92).
A common limitation inherent in all the absorption techniques is a practical
limit on the detection sensitivity caused by atmospheric turbulence. Turbu-
lent transfer of heat from the earth to the atmosphere causes localized
variations in the index of refraction of air. Collimated light passed through
the atmosphere is subject to distortion by the attendant focusing-defocusing
effect (Ref, 93). Beam spreading, destructive interference within the beam
cross section, and beam deflection can result. In remote measurements
turbulence can cause the beam to overfill the receiver and can cause the
energy received to vary as a function of time. One way to avoid these
problems is to complete a measurement in less than a millisecond (Ref. 94).
An alternative is signal-averaging over an appropriate time interval.
Long path techniques have many challenges to offer researchers over the
next few years. Among the more important are the development of tunable
sources and methods of tuning, the measurement of absorption coefficients
with sources actually used in the remote-sens ing system, and the thorough
evaluation of systems to establish their sensitivity and accuracy under real
measurement conditions. Once these challenges are met, the remote sensing
of air pollutants should become a useful tool (Ref, 92).
89
-------
LMSC-HREC TR D390190
SECTION VII
FLARE LOADINGS AND EMISSIONS FOR VARIOUS INDUSTRIES
To determine the impact of flaring on industrial emissions it was necessary to
develop data on the quantity and composition of material being flared in order
to estimate emissions. During this study of flare systems we have found al-
most no published data on the amount of flaring for a particular process. In
talking to flare manufacturers and users, we have also found that usually users
do not keep detailed data on what or how much they are flaring. However, it
was generally agreed that the individual plant production people have a fairly
good idea of the quantity and quality of gases being flared. It was decided that
the best way to obtain this type of information on an industry -wide basis was
through a questionnaire survey of a number of different users in each of the
major industries that utilize flares. From the results of the survey, estimates
were made of total flaring rates of various industries and also of the impact of
flaring on total emissions. This section discusses the results of this survey
including the calculation of flare loadings and emissions.
7.1 Questionnaire Format and Circulation
The primary purpose of the questionnaire was to determine the quantity and
composition of waste streams now being flared. In addition general information
on the type and operation of the flare unit was also sought. A copy of the questioi
naire, together with the cover letter, is included in this section.
The questionnaire was submitted for approval to the Office of Management
and Budget (OMB) in April 1974. After some modifications of the questionnaire,
final approval was received from OMB in September 1974.
The flare survey was circulated to the following industries: petroleum refining,
chemical manufacturing and iron and steel making. Except for petroleum and
'*s production, these three industries are the main users of flares. The
actual circulation was done by the industry's trade association: The American
Petroleum Institute (API) for petroleum refining, the Manufacturing Chemists
Association (MCA) for chemical manufacturers and the American Iron & Steel
Institute (AISI) for iron and steel making. Working through the trade associ-
ations not only made distribution of the survey simpler, since their mailing
lists were used, but also helped the response. Response was excellent from
all three industry groups with about 75% of the surveys being returned.
7,2 Refinery Questionnaire Results
Through cooperation of the American Petroleum Institute (API), a
task force consisting of 10 representatives of the petroleum industry was
90
-------
LMSC-HREC TR D390190
MISSILES HUNTsmn RESEARCH t ENGINEERING CENTER P o. BOX 1103 - MUNTSVIUE, ALABAMA sseo?
6c SPACE
COMPANY,
INC.
USER SURVEY - EPA FLARE SYSTEMS STUDY
Contract EPA 68-02-1331
We are currently engaged in an Environmental Protection Agency (EPA)
sponsored engineering study of flare systems for control of gaseous emis-
sions from stationary sources. The objective of this study is to evaluate
the potential of flaring for hazardous emission control. Our final report,
which will be publicly available, will include an evaluation of present flaring
practice and design methods, general cost data, and data on any air pollution
problem that flares themselves may cause. The EPA plans to use this report
as a guide for potential utilization of flares and as a basis for future research
and development programs in flare technology. We are obtaining our infor-
mation from the literature by contacting flare manufacturers and from this user
survey we are circulating.
We believe that industrial users of flare equipment comprise an important
source of information for this study. Of particular interest is determining
what waste streams are now being flared and the amount of flaring that is
occurring. We ask that you participate in this survey by supplying the infor-
mation requested on the enclosed questionnaire. Your participation will be
valuable even if you can only supply part of the information requested. The
information you supply will be held confidential by JLMSC through the report
writing stage, then destroyed. Some of it may appear in tabular or statistial
form in our report but without identifying your company.
We would appreciate your completing a separate copy of Sections H-IV for
each flare unit. We can supply additional copies if needed. If you have any
questions, please call us at (205) 837-1800 and ask for M. G. Klett, J. B.
Galeski or S. V, Bourgeois. Please return this questionnaire to Lockheed
Missiles & Space Company, P.O. Box 1103, Huntsville, Alabama 35807,
Attention: M.G, Klett.
Your cooperation in participating in this survey will be greatly appreciated.
Sincerely yours,
S. V. Bourgeois
Project Manager
Enclosure: (1) Survey Questionnaire
91
U B S I O I A * Y or LOCKHEED AIWCHAPT CORPORATION
-------
O. M. B. No. 158-S74008
Appr, Exp: 30 April 1975
SURVEY OF USERS OF INDUSTRIAL FLARE SYSTEMS
Contract EPA 68-02-1331
Date
Section I - Plant Identification
1, Name and Location:
a. Name of Company:
b. Plant/Division:
c. No., Street:
d. City: State: Zip:
2. Person to contact regarding information contained in this report:
a. Name:
b. Department/Division _______________________________
c. Telephone: (Area Code)^
3, Principal product(s) of this plant:
4. How many flare systems (individual stack/burners) do you have at this
location?
(If two or more, please complete Sections II to IV for each system.
Additional blanks are enclosed.)
Section U - General Information
1. Flare identification (if more than one at location): _
2. Name of process(es) generating waste gas stream:
3. Capacity of process(es) (Ib/hr, b/d, etc.):
4. Is the flare operated principally to control (check applicable items):
a. Intermittent flow of excess waste gas
b. Continuous flow of waste gas ______^_________________________
c. Odor nuisance
d. Toxic nuisance
92
-------
-2-
LMSC-HREC TH D390190
e. Emergency or abnormal process venting
f. Other (please specify)
5. Description of waste stream fed to flare (Engineering Estimate Permissible):
a. Waste stream(s) being flared:
b. Average composition of waste stream being flared:
c. Average load to flare for each combustible constituents) (for
intermittent flares average load over a year):
d. Number of major dumps to flare in previous year:
Ib/hr.
e. Amount of gas flared/dump | lb or scf
f. Heating value of waste stream: _____________________ Btu/scf
6. Is the waste stream pretreated prior to flaring? .
If yes, please specify: .
Section in - Flare Information
1. What is the type of flare (check one)?
a. Elevated Height (ft)
b. Ground Level
c. Burning Pit ____________________________
d. Other (Please specify) _____________________
2. Flare Capacity (Ib/hr)
3. Flare Diameter (inches)
4. Does the flare have the following auxiliaries (check applicable items)
a. Knockout Drums
b. Water Seals
c. Flame Arrestor
d. Purging Type of Purge Gas_
Purge Rate Ib/hr.
93
-------
-3-
LMSC-HREC TR D390190
e. Stack Seal
f. Smokeless Burning Water Stream
g. Automatic Control of Smokeless Burning
5. Do you monitor flare emissions? If yes, please specify how:
Are these data available to us?
6. What problems have you had in keeping the system operable?
7. Name of manufacturer of flare:
Did the same company design and install the system?
If not, name of company(s) "which did:
Date of Installation:
Section TV - Follow-Up
Would you be willing to discuss in more detail system performance,
data, and design features through a telephone call or visit to your
plant by one of our representatives? __________________
Tr you have any questions, please call one of the people listed on the cover
_etter at (205) 837-1800. Please return this questionnaire to Lockheed
Missiles & Space Company, P.O. Box 1103, Huntsville, Alabama 35807,
Attention: M. G. Klett. Thank you again for your cooperation.
-------
LMSC-HREC TR D390190
assembled in order to develop the information that was required. The task
force knew of no actual measurements of quantity and quality data for flares.
However, they agreed that personnel at many refineries could make reason-
able engineering estimates of these data. It was decided to obtain this infor-
mation by means of the survey from a relatively small number of represen-
tative refineries. The sample included 18 refineries, three each from six
different geographical locations, operated by 11 different oil companies. The
questionnaire was circulated to the refineries through the API.
Replies were received from 17 of the 18 refineries contacted. All replies
have reiterated that the quantity and quality data were from engineering es-
timates since these data were not measured. Five of the refineries that
replied, supplied information on the number of flares and design specifica-
tions but felt they could not make even engineering estimates on the quantity
and quality of material being flared.
For the remaining 12 refineries that supplied estimates on quantity and quality
11 estimates were reasonably consistent. However, one estimate was so
large, an-order-of-magnitude greater than the previous largest estimate, that
it was not used for estimating flare loading but is included in the tabulated
data for completeness (Refinery 12).
The refineries contacted had previously been selected for study in a joint
API-EPA refinery modeling program because it was felt that they formed a
representative sample of the total United States petroleum industry. The 11
refineries on which flare data are reported include at least one from each
geographical location. These refineries represent 4% of the total number of
refineries in the United States. However, their throughput totaled 14% of the
total United States throughput for the 1973-74 time period. While our sample
included refineries of varying size ranges, refineries greater than 100,000
bbl/cd predominated.
Table 7-1 shows a summary of the reduced data for the 11 refineries. This
table includes the number of flares, the sum of the flare loads for each
refinery broken down by composition, the percent of the refinery throughput
that is sent to the flare and the heat loss for each refinery computed from
the heating value of the streams sent to each flare. Most of the quantity data
were given for both 1973 and 1974. The numbers reported in the summary
table are the two year average value. Normally flare loading is very inter-
mittent with flare occurrences happening on the order of 8 to 10 times a year.
The reported flare loadings are the two year averaged loadings reduced to a
calendar day basis.
The amount of gas flared from each refinery ranged from 0.04 to 0.60%
of the refinery's crude runs with an average of 0.19% for the 11 refineries.
Applying this percentage to the total crude proeessed in the United States of
12,281,000 bbl/cd would indicate an amount of flaring from refineries for
1973 and 1974 of 7.2 x 10° pounds per calendar day or about 24,000 bbl/cd.
95
-------
Table 7-1
SUMMARY OF REFINERY FLARE DATA
Refintry
Number
I
2
J
4
5
6
7
a
t
10
it
Tot»l
No. ol
Flares
3
Z
4
2
S
7
4
8
2
6
I
45
Design
Capacity
(Mlb/hr)
333
240
3,305
1,407
385
2,319
640
2,600
1,600
3,051
684
16.564
C,
(Ib/cd)
-
135,825
2,124
15,481
2.090
3,526
3,409
17,047
42.946
15,922
20,084
258.260
c2
(Ib/cd)
345
39,836
4,691
4.806
4.926
9,683
3.384
21.309
36.602
22,332
10,938
158.852
C,
(Ib/cd)
505
34,400
18,799
4,575
339
8,737
13.788
128,139
71,577
12,584
55,685
350,128
c4
(Ib/cd)
664
25,617
27,987
1,930
531
4,286
16,721
12.JS9
23.588
4,999
5,847
126.529
cs+
(Ib/cd)
-
30.488
36,709
2,878
477
2,021
1,807
20,457
17,569
6,206
3,482
122.094
Aro-
matic «
(Ib/cd)
-
_
10,536
-
_
359
-
-
-
-
10.895
Ql«fln«
(Ib/cd)
-
28,199
27,093
3.J43
2.411
5,282
3,911
23,917
40,480
8,184
-
142,820
Paraffin*
(Ib/cd)
1.514
238,967
52.681
28,334
5.952
22.412
35.198
175.394
151,802
53,859
78,035
844,148
Total
Hydro-
carbon
(Ib/ed)
1,514
267,166
90,310
31,677
8,363
28,053
39,109
199,311
192,282
62,043
78.035
997,863
H2
(Ib/cd)
-
5,233
1,456
343
181
963
577
852
3,254
3,181
}?2
16.412
«2S
(Ib/cd)
2,640
Z.308
32
280
8,278
872
426
4.830
2,766
4,396
390
27.218
NHj
(Ib/cd)
-
-
2
-
244
-
-
-
-
-
_
246
Other
(Ib/cd)
2J.760
-
-
-
1,844
8
-
28.556
6.S30
_
J.90T
64.905
Total
(Ib/cd)
27,914
274.707
91.800
32,300
18.910
29,896
40,112
233.549
205,132
69.620
82.704
1. 106. 644
R ef ine ry
Thru put
(bbt/cd)
54,437
167.658
213.000
73.700
106.064
255.000
239.400
369,500
112,652
162,908
145,060
1.899,419
% to
Flare
0.170
0,554
0.143
0,145
0.059
0.039
0.056
0.210
0.604
0.142
0,189
JLiHj
Heat Lorn*
(Hlu/cd x 10 )
49
6.060
1,896
702
250
653
835
4,177
4,243
1,541
1.910
22.316
12
8
18.701
213,928
167.131
1,244.011
864.345
80,223
- 1 -
2,569.638
2.569,638
I
-
_
-
2,569.638
306. S90 2,781
52.446
M
O
H
X
-------
LMSC-HREC TR D390190
The heat loss that flaring represents was calculated for each refinery and
averaged nearly 20,000 Btu/lb. This would indicate a total heat loss from
refinery flaring in the United States of 1.4 x 10^* Btu/cd, This represents
about 0.6% of the total gas sold for industrial use in the United States for
1973 and 1974 (Ref. 95).
Figure 7-1 is a plot of the crude run versus flare loading for each refinery.
The solid curve represents the simple average flare loading for these 11
refineries. While the flare loading generally increased with refinery through-
put, the scatter of the data indicates that there are other parameters involved
in flare loading other than refinery throughput. However, the average flare
loading of these 11 representative refineries is probably a good indication
of the average flare loading of the petroleum industry.
Ninety percent by weight of the total load to flares consisted of hydrocarbons.
Hydrogen made up 1.6% of the load and hydrogen sulfide 2.6% with the remainder
consisting of mainly water vapor and nitrogen, Much of the hydrogen sulfide
flared was of low concentration in hydrocarbon streams. However, there
were flares mainly in sulfur recovery units where streams containing hydrogen
sulfide concentration of up to 50% were flared.
7.3 Impact of Flares on Refinery Emissions
In order to determine the impact of flares on refinery emissions not only
data on the quantity and quality of gases being flared are necessary but
also information is needed on the efficiency of flares as combustion devices
and the nature and amount of flare emissions. However flare systems-
especially elevated flares-present very difficult sampling problems. As
a result, very little emission data are available from flares.
The only known published report of a field test on a flare unit was by
Sussman et al. (Ref. 31). He reported the results of the test for a steam
inspirated type of elevated flare in the form of volume ratios:
CCL: Hydrocarbon 2100:1
CO2: CO 243:1
-------
l.MSC:-HREC TR
TO
*
o
_
»
X
a
a
3
^*
O
98
-------
LMSC-HREC TR D390190
Calculations based on these data were made using the estimated quantity
and quality data of the previous section in order to obtain an estimate of the
total emissions of carbon monoxide and hydrocarbons caused by flaring.
The calculations assumed a gas with three carbon atoms and a molecular
weight of 42, the average molecular weight of the refinery gas being flared.
NQx emissions were estimated from the data of Chase and George (Ref. 96)
and SO2 emissions were calculated from the total amount of sulfur being
flared. Table 7-2 shows the calculated total emissions of hydrocarbon,
carbon monoxide, nitrogen oxide and sulfur dioxide from refinery flares.
Table 7-2 also shows the percent of the total refinery emission from each
gas due to flaring. The total refinery emissions were estimated from
refinery emission factors (Ref. 58) and base on average refinery runs for
1973 and 1974.
Table 7-2
TOTAL ESTIMATED EMISSIONS FROM REFINERY FLARES
Gas
HC
CO
NO
X
so2
Emissions
(106 Ib/yr)
3.4
6.5
17.1
137.3
Percent of Total
Refinery Emissions
0.2
0.1
0.5
0.9
These numbers, based on engineering estimates of quantity and quality
and a minimum of field testing, should be considered tentative. However,
they do indicate that the average yearly emission from flares constitutes
just a small fraction, less than 1%, of the average yearly refinery emissions.
Total flare emissions over a year's time therefore probably only have a small
impact on total refinery emissions. However, because of the intermittent
nature of flaring, the majority of flare emissions are concentrated into just
a few minutes of actual flaring. During this time four or five times the
normal refinery emissions are released into the atmosphere. While design
modifications for flares to suppress smoke formation has been largely success'
ful, very little if any work has been done to suppress emissions resulting from
unburned hydrocarbons and partial oxidation products.
7.4 Iron and Steel Mills Questionnaire Results
Through the cooperation of the American Iron and Steel Institute (AISI),
the survey was distributed by the AISI to the major manufacturer* of iron and
steel in the United States. There are two types of gases that are flared in
iron and steel mills, excess blast furnace gas and excess coke oven gas.
Flaring is only done on an intermittent basis, usually to control lime pressure,
and generally the gases are scrubbed before flaring.
99
-------
LMSC-HREC TR D390190
Ninety-nine percent by weight of the combustible blast furnace gas con-
sisted of carbon monoxide. The remaining one percent consisted mainly of
hydrogen and methane. Hydrocarbons made up 73% by weight of the coke
oven gas, carbon monoxide 17%, hydrogen 9% and hydrogen sulfide 1%.
Replies were received for 61 blast furnace gas flares and 30 coke oven
gas flares. Several of the replies received supplied information on the
capacity and design specifications of the flares but felt that they could not
make engineering estimates on the quantity and quality of material being
flared. Quantity and quality data were given for 35 blast furnace gas flares
and 20 coke oven gas flares. The replies represent 38% of the raw steel
production capacity in the United States.
Table 7-3 shows a summary of the reduced data for the blast furnace
flares and Table 7-4 for the coke oven flares. The table includes the num-
ber of flares, the sum of the flare loads broken down by composition and
the heat loss associated with this flaring. The reported flare loadings are
averaged yearly loadings reduced to a calendar day basis.
The weight of combustible gas flared from blast furnaces averaged 6.6% of
the furnace's capacity. Applying this percentage to the total 1974 United
States' capacity of 145.5 x 10" tons would indicate an amount of flaring from
blast furnaces in 1974 of 5. 3 x 10' pounds of combustible gases per calendar
day. The heat loss that this flaring represented amounted to 2. 5 x 10 Btu/cd.
The amount of combustible gas flared from coke ovens averaged 0.4% of the
ovens' capacities. Applying this percentage to the total iron and steel industry's
coke capacity of 55 x 10" tons would indicate the amount of flaring from coke
ovens in 1974 of 1.1 x 1Q& Ib/cd. The heat loss that this flaring represented
amounted to 1.9 x lO*0 Btu/cd.
While the lost heating value of blast furnace gas that is flared is comparable
to the heating value of the gas flared from refineries, the iron and steel in-
dustry has little alternative but to flare the excess gas. Blast furnace gas
typically consists of 25% CO and the remaining inert gases. Therefore, the
heating value of the gas is low, around 9C Btu/ft^, making it uneconomic to
recover any that cannot be used immediately.
In addition to blast furnace gas and coke oven gas flares there were a few
other flares reported from the iron and steel industry on miscellaneous
processes including sulfur plants and an annealing plant. Table 7-5 gives
a summary of the reduced data for these plants.
7.5 Impact of Flares on Iron and Steel Mill Emissions
While there have been no published report of field tests on blast furnace gas
flares, the data of Sussman et al. (Ref. 31) for a refinery flare indicates
greater than 99% complete combustion of hydrocarbons. Assuming a 99%
efficiency for blast furnace flares, the emissions of CO from these flares in
1974 was 1.9 x 10^ Ib which is equal to about 1% of CO emmissions from
industrial processes.
100
-------
LMSC-HREC TR D3900190
Table 7-3
SUMMARY OF BLAST FURNACE FLARE DATA
No
1
2
3
4
5
6
7
8
9
10
It
12
13
14
15
16
11
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Total
Process
Capacity
(tons/da/)
6,700
2,200
2,800
3,600
900
4,300
2,500
6,900
500
5,200
2,600
2,500
2,000
3,900
3,200
7,600
16,000
3.400
11,700
5,500
2,700
5,900
800
4,000
6,000
6,600
5,800
13,000
2.600
600
3,500
1,200
1,100
2,500
2,500
152,800
No, of
Flares
3
1
1
1
1
1
1
1
1
2
I
1
1
1
1
1
4
2
1
1
1
1
1
1
3
1
2
4
1
1
1
1
1
1
1
48
CO
(lb/cd)
333,000
3,000
936,000
864,000
197,000
16,000
459,000
1,386,000
92.000
3,229,000
2,228,000
240,000
348,000
254,000
207,000
171,000
164,000
232,000
701,000
179,000
217,000
392,000
177,000
246,000
189,000
753,000
556,000
1,639,000
674,000
65,000
914,000
883,000
34,000
248,000
1,107,000
20,233,000
H2
(lb/cd)
1,800
-
4,800
2,400
1,100
-
7,300
20,400
1,500
50,100
12.300
600
3,300
2,000
1,600
-
500
2,200
7,200
1,400
100
-
500
2,000
1,500
2,000
1,400
12,900
5,900
500
8,500
7.100
9,600
100
18,100
191,300
CH4
(lb/cd)
-
-
-
-
-
-
-
-
-
-
4,700
-
-
-
-
-
-
-
12,000
500
-
-
-
-
-
-
-
22,400
-
-
-
-
-
-
-
39,600
N2and CO2
(lb/cd)
1.048,000
9,000
2,082,000
3,975,000
275,000
43,000
1,638,000
4,585,000
321,000
11,295.000
5,757,000
656,000
1,164,000
867,000
708,000
121,000
493,000
712,000
2,280,000
662,000
781,000
214,000
517,000
712,000
720,000
2,138,000
1,578,000
6,439.000
2,565,000
247,000
3,393,000
2,845.000
48,000
713,000
3,533,000
66,034,000
Total
Combust.
(lb/cd)
335,000
3,000
941,000
866,000
198,000
16,000
466,000
1,306,000
93,000
3.279,000
2,244,000
241,000
351,000
256,000
209,000
171,000
165,000
232,000
691.000
181,000
217,000
392,000
177,000
248,000
190,000
755,000
557,000
1,674,000
680,000
66,000
922,000
890,000
43,000
249,000
1,125,000
20,431,000
Heat Lose
(Btu/cd x 106)
1,600
700
4,200
3,800
800
100
2,300
6,400
500
17,100
9,500
900
1,500
1,100
900
700
700
1,100
3,200
900
1,000
1,600
800
1,000
1,000
3,200
2,600
8,900
3,200
300
4,300
4,000
600
i.too
5,200
96,800
Height
(ft)
160
109
130
201
112
140
109
200
160
150
111
125
167
200
200
240
140
89
113
198
155
230
160
150
137
150
125
200
139
110
202
100
230
150
160
154
(Average
101
-------
Table 7-4
SUMMARY OF COKE OVEN FLARE DATA
No.
I
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Total
Process
Capacity
(ton /day)
500
450
2,170
1,340
3,940
2,010
410
2,340
2,830
1,910
260
510
1,570
4,750
2,480
3,800
3,500
1,000
5,070
2,510
43,350
Hydrocarbon
(Ib/cd)
62,500
2,000
900
900
5,300
21,300
10,400
5,200
70,700
3,100
5,500
600
800
1,100
8,300
6,500
700
12,000
1,700
6,600
226,100
H2
(Ib/cd)
1,800
300
100
100
1,500
1,500
2,000
1,200
11,100
500
1,100
100
100
200
300
1,200
100
1,600
300
1,200
26,300
CO
(Ib/cd)
18,700
600
100
300
2,700
1,400
2,800
1,700
7,500
700
1,500
100
200
400
4,500
3,500
300
1,800
600
2,100
51,500
H2S
(Ib/cd)
850
_
170
210
270
210
200
1,910
N2,C02
H2O
(Ib/cd)
377,000
600
200
400
1,600
4,600
4,000
1,100
36,800
1,400
2,100
100
100
700
3,900
3,000
400
1,100
4,400
443,500
Total
Combust .
(Ib/cd)
84,500
2,900
1,100
1,300
9,600
24,500
15,200
8,100
89,300
4,300
8,000
800
1,100
1,700
13,400
11,400
1,100
15.400
2,600
10,100
306,400
Heat Loss
(Btu/cd x ID6)
1,080
60
10
30
240
280
360
170
1,780
110
190
10
20
20
290
250
30
250
70
230
5,480
Height
(ft)
36
100
115
100
135
121
80
100
100
80
100
100
100
110
151
150
225
125
104
112*
r
S
en
O
i
ffi
JO
ft
O
H
»
O
u>
sO
o
Average".
-------
Table 7-5
SUMMARY OF FLARE DATA FROM MISCELLANEOUS IRON AND STEEL PROCESSES
No,
1
2
3
4
Process
Sulfur Plant
Desulfurization
Anneal Atmos.
Gas
NH, Destruction
J3
Capacity
(ton/day)
15
94
3,000
87
Hydro-
carbon
(lb/d)
1,200
""""*"
2
H2
(lb/d)
Hf*N
A A V^jLlI
2,530
6
NO
Jll V,/
X
29
CO
(lb/d)
1,120
66
fSjH
3
29
H2S
(lb/d)
36,100
16,847
SO
OU2
2,470
Inert
(lb/d)
20,520
9,630
710
171,370
Total
Combust.
37,750
20,500
74
29
Heat Loss
t
(Btu/d x 10b)
328
240
1
Height
(ft)
150
206
8
100
o
OJ
5
co
O
B
O
sO
o
-------
LMSC-HREC TR D390190
For coke oven flares, calculations were made for emissions of HC
CO, NOX and SC>2 in the same manner as the refinery flares. Table 7-6
shows the results of these calculations along with the percent of the total
emission from coking from each gas due to flaring. The total coke plant
emissions were estimated from emission factors (Ref. 58) and based on
1974 coke production.
Table 7-6
TOTAL ESTIMATED EMISSIONS FROM COKE OVEN GAS FLARES
Gas
HC
CO
NO
X
so2
Emissions
(106 Ib/yr)
0.4
1.5
0.7
4.8
Percent of Total
Coke Plant Emissions
0.2
2.1
5.8
0.9
These results, based on engineering estimates and a minimum, of field
testing, are tentative. However, as with refinery flares, they indicate that
the emissions from coke oven flares constitute a small portion of the average
yearly emissions from coke plants.
7.6 Manufacturing Chemists Questionnaire Results
Through the cooperation of the Manufacturing Chemist Association
(MCA), the survey was distributed by the MCA to members likely to make
use of flare systems. Replies were received for 75 different flare units.
However, many of the questionnaires did not give information on the quantity
and quality of gases being flared. Forty replies were received covering the
manufacture of 15 different chemicals which gave data on the quantity and
composition of gases being flared.
Table 7-7 shows a summary of the reduced data for these chemical
process flares. The table includes the identification of the process, the
capacity of the process, the sum of the flare loads broken down by composi-
tion and the heat loss associated with this flaring. The reported flare load-
ings are averaged yearly loadings reduced to a calender day basis.
Most of the different chemicals for which flare loading data were re-
ported included data from only one or two plants. Because of the scatter of
the flare loading data from plant to plant, meaningful estimates of industry
flaring loads can only be made by averaging the loadings for a number of
individual plants. The only chemical in which flare loading data were avail-
able from a number of different plants was ethylene. However, the other
data give a rough idea of the magnitude of flare loadings for these processes.
104
-------
LMSC-HREC TR D390190
Table 7-7
SUMMARY OF CHEMICAL PROCESS INDUSTRIES FLARE DATA
Procem
Olefin*
Ethylene
Elhyiene
Ethylene
Ethylene
Ethyiene
Ethylene
Acetylene
Aromatiei
Petrochemical*
Petrochemical*
Polypropylene
Poiyproplylene
Butyl Rubber
Acetic Acid
Acetic Acid
Acetic Anhydride
Acetic Anhydride
Adipir Acid
Ac ry Inni! r i le
Ac ry loiut rde
Ammonia
Arnmoma
Ammonia
Al cohol *
Carbon Bia, k
Hhoiphor us
C,S and S Recovery
Na HS
Aldicarb
CO For Pho*f ene
Oil Additive
Storage and Loading
Ethylene Loading
Ethylene Storage
Buladine Storage
Ammonia Storage
HCN Storage
Tank Car Loading
Aiodrin
Nudrin
Nudrin
Capaiity
llb/yrS
964 MM
630 MM
500 MM
750 MM
830 MM
775 MM
325 MM
750 MM
2.000 MM
660 MM
260 MM
1 10 MM
200 MM
110 MM
1 10 MM
160 MM
140 MM
380 MM
365 MM
350 MM
550 MM
660 MM
800 MM
1,670
600
39,600
300
13,500
2,900
37
CO,"
(lb/cdl
2,580
700
20,000
11,900
3,600
16,200
430.000
236
120
7-H2C
18
Other
llb/cd)
1300-H2S
9600-NO^
16-HCN
276-HCN
192-NHj
4800-NH2
fc50-H2S
84-C2S
io-c2s
528-HCN
950-NHj
480-HCN
10 -HCN
6-HCN
Total
Combust .
(lb/cdl
10,000
47,420
18.700
10,100
26.900
48,000
1,726
7,900
96,000
300
55.200
2,500
36,000
8,700
67,200
15,600
100,700
16
276
54,100
192
4.800
16,300
37.500
10.400
444
10
3,600
1.440
10,900
12.000
2
1,100
950
480
1,080
96
32
6
Heat Loss
(Btu/cd x 10S
195
660
350
155
562
960
33
157
2 600
5
476
37
650
152
455
1,080
1
3
1,680
1
43
335
421
46
9
1
27
2
216
243
21
9
2
9
1
1
I
105
-------
LMSC-HREC TR D390190
Data were received from six different elhylene plants representing 19%
of the total U.S. ethylene capacity. The weight of the combustible gas flared
by these plants averaged 1.3% of the capacity. Applying this percentage to the
total U.S. 1974 capacity o£ 24 x 10° Ib would indicate an amount of flaring from
ethylene plants in 1974 of 8k7 x 105 lb/cd. The heat loss that this flaring repre-
sented amounted to 1.6 x 1010 Btu/cd.
7.7 Summary of Flare Loadings
From the survey results, flare loadings of combustible gases were calculated
for four process industries; (1) petroleum refining; (2) ethylene production;
(3) blast furnace operation; and (4) coke production. Table 7-8 summarizes
the data for these industries.
Table 7-8
INDUSTRY FLARE LOADINGS AND HEAT LOSS
Industry
Petroleum Refining.
Ethylene Production
Blast Furnace Operation
Coke Production
Industry Flare
Loading
(lb/cd)
7.2 x 106
8.7 x 105
5.3 x 107
1.1 x 106
Flare Loading as
Percent of Capacity
0.19
1.3
6.6
0.37
Heat Loss
(Btu/cd)
1.4 x 10H
1.6 xlO10
2.5 x 1011
1.9 xlO10
To estimate emissions from flares, information is needed on the efficiency
of flares as combustion devices. Estimating emissions from very limited field
test data on flares and using industry flare loadings from the survey results
indicate that the average yearly emission from flares constitutes just a small
fraction, less than 1%, of the average yearly industry emission. Total flare
emissions over a year's time, therefore, probably only have a small impact on
total emissions. However, because of the intermittent nature of flaring, most
of flare emissions are concentrated into just a few minutes of actual flaring.
During this time four or five times the normal industry emission are released
into the atmosphere.
106
-------
LMSC-HREC TR D390190
SECTION VIII
RECOMMENDED RESEARCH PROGRAM
8.1 Theoretical Analysis of Combustion Modifications Applicable
to Flaring
8.1.1 Summary and Objectives
Because of the lack of present sampling capability and emissions data for
elevated flares, other means of estimating gaseous emissions are required
for evaluating proposed pollution control methods and regulations and for
evaluating the applicability of current combustion technology to flare emis-
sion control. In particular, some means of calculating combustion effi-
ciency and partial oxidation products is required.
The objective of this research is to extend previously developed technology
to the analysis of flare systems. The theoretical model developed would be
applied to evaluating combustion modifications applicable to flaring (Section
7.1.4) and to the evaluation of the applicability of flaring to the control of
gaseous emissions (Section 7.3).
8.1.2 Background
Analysis of turbulent combustion depends on combining turbulent mixing
models with kinetic data for elementary reaction steps. Combustion rates
are limited by turbulent mixing rates and are typically several orders of
magnitude lower than theoretical even for highly efficient gas turbine
combustors. No simple analytical methods have been developed.
When analyzing turbulent mixing problems it is customary to use empirical
correlations to describe the transport rates because of the lack of useful
theoretical formulations. Unfortunately, empirical correlations have not
been developed which are suitable for detailed analysis of subsonic reacting
flows because of the dearth of experimental data.
Numerical analysis techniques have recently become available for
the precise analysis of temperature, composition and velocity profiles
in reacting flows. Figure 8-1 illustrates the application of such a model
to the analysis of a hydrogen diffusion flame, comparing theoretical pre-
dictions (Ref. 97) against experimental measurements (Ref. 98). The jet
diameter was 7.62 mm. Jet velocity was 590 ft/sec.
The recommended research program would involve the application of
present analytical capability to the measurement of combustion efficiencies,
partial oxidation products, and nitrogen oxides formed in a diffusion flame
analogous to an elevated flare system. The program would consist of the
following parts:
107
-------
LMSC-HREC TR D390190
Fig. 8-1 - Comparison Between Measured and Calculated Centerline
Distributions (Hydrogen Jet Exhausting into Air). Upper
Figure: Species Distributions. Lower Figure; Tempera-
ture and Velocity Distributions (Ref. 97),
108
-------
LMSC-HREC TR D390190
8.1.3 Validation of the Analytical Model
Sample cases would be run to check the validity of the selected analytical
model for large diffusion flames. Data available in the literature would
be summarized. Comparison would be made between predicted and ex-
perimentally measured flame properties for selected representative
cases.
8.1.4 Evaluation of Flare Design Modifications
Representative cases would be run to evaluate the effect of combustion
modifications applicable to flaring. Variables considered would include
gas discharge velocity, burner diameter, flow distribution through multiple
ports, effect of steam distribution and discharge velocity and substitution
of air and oxygen for steam. Calculations would be made of emission
rates of nitrogen oxides, partial oxidation products and soot or particulates.
Combustion efficiencies would be calculated to estimate unburned hydro-
carbons.
8.1.5 Priority
On a scale of A through E, the priority for research described in Sections
8.1.3 and 8.1.4 is A.
8.2 Evaluation of Remote Sampling Methods
8.2.1 Summary and Objectives
Elevated flare systems have eluded present sampling methods for reasons
of remoteness and non-stoichiometric air-fuel dilution. Evaluation of
remote sampling techniques for typical flare emissions is therefore needed.
8.2.Z Background
The problem of sampling elevated flare emissions is essentially one of
accessibility. Flare stacks typically range from 200 to 400 feet in length
with flames reaching 200 ar 300 feet in emergency flaring. A summary
of conventional sampling techniques and application to flare systems is
presented in Section 6. Recently developed sampling methods which may
be applicable rely on spectroscopic techniques and may include laser
sources.
8.2.3 Summary of Remote Sampling Technology
Remote sampling methods and instrumentation would be summarized
according to cost, performance and availability. For each instrument
selected as applicable to flare emissions monitoring, instrument range,
sensitivity and other operating characteristics such as drift and repro-
ducibility would be included. Complete monitoring systems would be
109
-------
LMSC-HREC TR D390190
chosen based on suitable components and auxiliaries. Instrument manu-
facturers would be contacted for complete instrument specification and
other available performance data based on previous applications.
8.2.4 Remote Sampling Field Studies
A remote sampling unit would be selected or assembled for components
for field testing at selected locations. Emissions measured would include
particulates, hydrocarbon classes, nitrogen and sulfur oxides and hydro-
carbon oxide classes. Resolution of emission classes would be defined.
8.2.5 Priority
On a scale of A through E, the priority for research described in Sections
8.2.3 and 8.2.4 is C.
8.3 Application of Flaring to Control of Gaseous Emissions
8.3.1 Summary and Objectives
The objective of the following research program would be to evaluate the
potential of flaring as a means of pollution control. Guidelines for deter-
mining the suitability of given waste streams for flaring would also be
established.
8.3.2 Background
The application of flaring for controlling gaseous emission promises to
be a relatively inexpensive means of pollution control when compared to
conventional methods such as incineration. Flaring has been applied to
odor control in removal of trace quantities of NH(CH-)-. In this applica-
tion, flaring was reportedly more effective than other methods of control
(Ref. 3). Application of flaring to other streams and components requires
experimental confirmation of effectiveness for reasons discussed pre-
viously, i.e., lack of suitable theoretical and experimental data for large
turbulent diffusion flames.
A list of the types and magnitudes of emissions from petrochemical manu-
facturing is given in Table 8-1. Of these, emission control by flaring is
most promising for those emissions which are themselves combustion
intermediates: organic acids and anhydrides, esters, ethers and oxides.
These constitute a large part of present petrochemical emissions.
8.3.3 Theoretical Analysis
Theoretical analysis of combustion products and efficiencies would be
conducted for selected components and conventional flare systems. The
modeling technique described in Section 8.1 or similar techniques would
be used for the analysis. Maximum concentration limits and other operating
conditions would be defined.
110
-------
LMSC-HREC TR D390190
Table 8-1
EMISSIONS FROM PETROCHEMICAL MANUFACTURE, MM LBS (Ref.
"'' ' ' ""* ' "~" ----- - .... .- . ~ ..... «~ ,.
I'tOir-ftro.'.H'.' Ic ilylrucar' on -
! uc-lylf;."
IH.I i.y i iMir':;
oy h *.'>: u it-
f'lry i i,.,'1
hi,,;h«'r oiefi:;n
1 ; |. ( , '- 'I i C Tl C
To tul
Aro'i'-.'-.i i- Hydrocr:,rbo;i:,
be(v..':'»e
a tyres, f -
toli.e K:
xy.lenoc
Total
Orguni c Ilalidc:;
ally! c hi or iiie
benzy IchloricK-
cli chi orobenzcnes
et'rr,- j chlorides
methyl hrJiden
vinyl chloride
Total
Organic Acids and Anhydrides
ace t ic acid
rr« a 1 e i c an hy d rid e
phenol
xylene base acids
Total
Alconols
butyl alcohols
etliy] alcohol
isooctyl alcohol
i sopropy 1 aJ cohol
methyl alcohol
Total
"<"""' *" *****"
1060
2.3
lli.'t
3-0
33.9
6.9
l't.0
73-9
1
16.1
6.9
6.0
3 ':
32.0
0.6
0.1
U.^
1 3 . 7
0.7
1-3
16.6
0, 1
0-5
3.1
3.2
6.9
0.0
2.6
O.'o
3-0
e.o
15-8
* -
l'J'l'u
1.9
17 . 0
3-1'
9i.6
7-9
i 7 . H
09 . 1
19.1
6.5
6. '4
2.6
, t jf
;^l , u
0.7
0.1
u , '>
". 3 . «
0.9
1.2
36.9
0-2
0.5
3.8
2.6
7-1
1 .0
2.8
0.7
4 . 0
10.lt
1C. 9
""* ' '** "*"
\ *- j 1 ", \ j
1 . 2
~fl .9
S.6
- ' v . -
7 . 9
). i . H
"i 2 . 9
19-9
9 >
14.2
76.2
0.9
0. 1
U- 1
67.'-.
J.8
0.7
0.3
1.3
9.14
»(.2
15.2
1.8
3.8
1 . !.
6.0
2.1 . 1
3U.1
" - - ^*»»-
1'"M'3
0.7
59 7
21 . "
376.5
7-9
JOB. 7
57 ' - 9
95-6
51. it
]Ji . 0
6.8
367-8
1.3
0.2
u.s
191.7
'» 3
0.5
200.5
o.e
3.3
22.7
6.8
33.6
3.14
5-3
2.6
6.9
50.6
70 . 8
-- "" 1 --I-U '" "
2000 :
t
OJ;
7 0 . 0
53^
i ,'"'': f|. 2
i y
?vo.:i !
;
222 . 1 !
126.1 '
19 -9 '
11.2
379-6 i
1.9
0.3
U. i
560.2
10.1
0.3 "'
573.5 :
1.8
6.7
50.7
11.2
78.14
14.5
7 3 ,
3.'* !
13-6 !
118.0
1146.8 j
111
[Continued)
-------
LMSC-HREC TR D390190
Table 8-1 (Concluded)
^v 1 i *"* " y r] c ^
,'!<-roleir>
bul yrf, J c-eliyci"
i ' ;...'.'*:;". 1 dehydi-
To till
I'itro^cn-orpan I. cs
ficrylonl'trl] e.
aniline
methyl f'jnim1:;
toU'r-nt- dii,';oi:yo2U:tc
Total
Ester.-,, Ether;;, Oxidts
tu-ry J nU'.s
cthyl-r-j'!,- oxJdi.
glycol ethers
pro'-y !.«!;( fx i (j..>
To !<»il
ucctone
Totul Pctrocli'.:,;;i cal
J9C"
.1.
0.
1 .
.'* .
;.
,3 *
0.
0.
0.
f'.
1.
7.
0.
£^
10
1.
167.
r]
T>
}
f
J
6
';
7
5
3
5
0
3
1
3
"s
J.
B
2
6
l
0
1
3
6
it
0
0
0
5
J.
8
0
o
-
^3
1
19 H
70
.»
.1
.6
3
.»!
.6
.3
.3
-5
.9
.6
.8
.It
.a
.f>
.5
.0
]0
3
C
3
6
13
9
0
0
0
12
2
20
0
~\ n
?!
1|
W
^J
.3
3
r
.7
.0
.7
.9
-7
.8
.J
. h
.0
.9
. f
.Q
.5
.0
1990
7.7
0.7
7-7
16.1
3?.P
21.8
1.6
1.6
l.l
26.1
5.3
it5.*!
2.1
?8 ?
tt.1.0
12.7
1,199-6
2000
18.1
1.5 '
36.9
37-5
7^ . 0 i
67.1
2.5
3-7
1.6
7M |
12.0
101 . 5
IK 6
70.8 '
1
lyo.i
i
35.2
3,020.9
112
-------
LMSC-HREC TR D390190
8.3.4 Experimental Analysis
This study would define experimental techniques and sampling methods
for evaluating industrial flaring as applied to pollution control. The
main result of the study would be a user guide with recommendations for
determining the suitability of a given waste stream for flaring.
A pilot-scale flare burner and combustion chamber would be constructed.
Suitable sampling techniques would also be developed. Components se-
lected from Table 8-1 and at least one component evaluated as part of the
study outlined in Section 8.3.3 would be tested using the pilot flare burner.
For each component selected for testing, operating conditions would be
varied to determine optimum conditions for pollution control. Effects of
flame stability and turbulence level on the production of pollutants would
be determined.
8.3.5 Priority
On a scale of A through E, the priority of research described in Section
8.3.3 is A. The priority of research described in Section 8.3.4 is B.
An experimental study almost identical to that described in Section 8.3.4
has been recommended as part of the Federal R&D Plan for Air Pollution
Control by Combustion-Process Modification (Ref.99):
The objective is to determine the effect of turbulence and fuel type on the
production of pollutants is turbulent diffusion flames with gaseous fuels.
A large burner is recommended for this study, especially if the level of
effort is minimum. Turbulence scale and intensity should be the major
variables considered. The effect of fuel type should also be investigated.
Special instrumentation might have to be developed for solving problems
related to the effect of "unmixedness" on the production of pollutants.
Attention would be given to the part that flame stability plays in the pro-
duction of pollutants. The rationale and incentive for this proposed re-
search (R&D Opportunity: VIII-22) is that many industrial flames are of
the turbulent-diffusion-flame type. The research would provide guide-
lines for the optimization of turbulent conditions in gaseous-fuel com-
bustion systems to minimize pollutant emission and form a basis for
studies of other fuels burned in like manner. The relative overall priority
rated for this research is 2 on a scale of 1 through 5.
8.4 Economic Analysis of Waste Stream Recovery and Alternate
Disposal Methods
8.4.1 Summary and Objectives
An inventory of waste streams currently being flared is being compiled
by means of a questionaire as part of this Task Order. Waste streams
burned in flares represent a potential loss of profit as well as a source of
113
-------
LMSC-HREC TR D390190
gaseous and particulate emissions. For these reasons, and in order to
define a basis for pollution control purposes, the economic basis for
flaring as opposed to stream recovery or alternate disposal methods is
needed.
8.4.2 Background
There are numerous gaseous plant emissions which are disposed of by
means of flaring which are not associated with outright emergencies.
These include:
1. Low pressure vent gases (Ref. 100) from an absorber.
These gases contain light hydrocarbons, methane, ethane
and propane plus oil droplets. The heating value of these
discharged gases will not vary appreciably.
2. Partial Condenser Vent Gases. These gases may contain
water and oil droplets (Ref. 100).
3. Disposal of off-spec or excess product (Ref. 23). This
disposal problem is most frequent during plant start up
which may last for periods up to about one year.
4. Leakage of gas through safety valves and block valves.
Valve leakage to flare during routine operation of a 550
million-pound-per-year ethylene unit has been estimated
at 4,000 Ib/hr (Ref.23 ).
5. Disposal of by-product streams which are produced in
quantities too small or of insufficient purity for economical
recovery (Ref. 23).
6. Venting of fuel and product storage tanks and loading
platforms.
Gases which are sent to the flare system from the above sources are
produced in quantities which can be estimated and for which storage for
sale, recyling to process units or use as fuel in heaters and incinerators
appear to be practical alternatives. For this reason guidelines need to
be established to aid in determining these situations in which alternatives
to flaring are reasonable. The following research program is recom-
mended:
8.4.3 Identify Economic Considerations Now Used to Determine Whether
a Given Flared Stream has Sufficient By-Product Value for Recovery
Representative processes would be chosen for evaluation from the process
industries. By-product and waste streams would be listed for chosen
processes. Stream composition and volume and recovery conditions (tem-
perature and pressure) would be listed for each process stream along with
recovery value, capital, operating and utilities costs for recovery and end
use.
114
-------
LMSC-HREC TR D390190
8.4.4 Identify Alternative Uses of Low Pressure Flammable
Hydrocarbon Gases
Waste streams sent to the flare system are usually available at relatively
low pressure. Suggested or potential uses for such streams would be
identified and evaluated. One such suggested use which seems reasonable
is the use of the waste stream for afterburner fuel gas (Ref. 100).
8.4.5 Evaluation of Alternative Disposal Methods
For the processes and waste streams selected for economic analysis
in Section 7.4.3, alternative disposal methods such as incineration,
adsorption, absorption, scrubbing and filtration would be identified.
These would be evaluated for technical and economic feasibility.
8.4.6 Priority
On a scale of A through E, the priority of research described in Sections
8.4.3, 8.4.4 and 8.4.5 is D.
8.5 Emission Factors for Elevated Flare Systems
8.5.1 Summary and Objectives
The objective of the study would be to recommend the best available
method for sampling and analysis of gaseous flare emissions and conduct
field testing of elevated flare systems.
8.5.2 Background
Very little information on elevated flare emissions is available as has
been discussed previously in several sections of this report. Further-
more, the validity of the fragmentary information available is unknown.
Based on our conversations with flare vendors and a major chemical
manufacturing firm, two methods of sampling elevated flare emissions
were identified, direct probe sampling and tracer-assisted probe
sampling. Direct probe sampling involves inserting a probe into the
exhaust plume beyond the flame boundary and is therefore stongly
dependent on probe location. The use of a tracer aids the sampling
technique by allowing a correction for dilution of the plume by ambient
air.
These techniques are preliminary and many other improvements are
foreseen. For example, the use of heavy and light tracers in conjunction
may allow a further correction for buoyant and diffusion forces and a
measurement of reliability; if the measured dilution of both tracers is
the same, the air dilution factor can be calculated without consideration
of the buoyancy factor.
115
-------
LMSC-HREC TR D390190
In addition to air dilution problems, the direct sampling methods are
complicated by accessibility to the plume, and other problems which
typically arise in direct source sampling such as the requirement for
rapid quenching of reaction products, condensation of liquid products
in the probe and correction for the finite sampling and analysis times
involved (Ref. 39). Of these, plume accessibility appears to be the most
difficult obstacle; methods used have involved either a construction
derrick or a long pole to support the sampling probe. Other methods
considered have involved the use of helicopter borne sampling equipment.
In general, these methods tend to be dangerous, cumbersome and ex-
pensive. Improvements envisioned in this area include the use of fixed
supporting structures taller than the flare stack and at a safe distance
from which boom lowering of the probe into the plume would be practical.
For steady-state emissions over long periods of time, the problem of
flare sampling is not significantly different from stack gas sampling using
multiple receptor locations relatively close to ground level. Such re-
ceptor methods normally require a relatively isolated source and require
a relatively large number of points for a statistically reliable estimate
of the source strength. Such requirements are rarely met with flares.
8.5.3 Site Selection and Evaluation of Sampling Methods and Hardware
From a survey of sampling and analytical techniques now in use, a sampling
system would be chosen which is best suited to the problem of monitoring
source emissions from flares, and a program developed for the determina-
tion of emissions factors. Emissions considered would include hydrocarbons
NO , SO , participates and partial oxidation products such as CO and aide-
X 3v
hydes. The sampling and analysis technique would be suitable for emis-
sions monitoring of sudden upsets as well as steady-state flow. The
duration of plant upsets may be from a few minutes up to a maximum of
about one hour (Ref. 10). During major upsets, discharge of several
hundred thousand pounds per hour to the flare is common with resulting
flame lengths of several hundred feet and combustion rates upwards of a
billion Btu's per hour (Ref. 30). Testing sites would be selected from
among industrial locations and experimental flare systems furnished by
manufacturers of combustion equipment. At least one site would be chosen
from the hydrocarbon process industries.
8.5.4 Field Testing of Elevated Flare Systems
Field testing would involve the measurement of emission factors at selected
sites. Analysis of data would include an estimation of precision. Analysis
of the emissions from the selected plant site(s) would include an inventory
of flared streams and measured emissions on a day-to-day basis for a
period of time long enough to give an indication of typical plant flaring
practices .
8.5.5 Priority
On a scale of A through E, the priority for the research outlined in
Sections 8.5.3 and 8.5.4 would be A.
116
-------
LMSC-HREC TR D390190
SECTION IX
REFERENCES
1. Reed, R. D., Furnace Operations, Gulf Publishing Co., Houston,
Texas, 1973.
2. Scheers, A. H., "Veiligheidssystemen en luchtverontreiniging," Pt.
Procestechniek, Vol.28, No. (6)-(13), 1973.
3. Byrd, J. F., "Solving a Major Odor Problem in a Chemical Process,"
Paper No. 63-35, APCA Conference, June 1963.
4. American Petroleum Institute, API Manual on Disposal of Refinery
Wastes, Vol.11, "Waste Gases and Particulate Matter," 5th edition.
New York, 1957.
5. Danielson, J. A., ed., "Air Pollution Engineering Manual," Air Pollu-
tion Control Dist., County of Los Angeles, U.S. Dept. of Health,
Education and Welfare, Public Health Service, Cincinnati, Ohio, 1967.
6. Horton, B. B., Master, N.,and Wedel, R. T., "Approaches to the Design of
Flare Systems," Presented at Sessions, "Offsite Facilities," ASME Petro-
leum Mechanical Engineering Conf., Los Angeles, 17 September 1973.
7. Stewert, R. M., "High Integrity Protective Systems," Instn. Chem.
Engrs. Symposium Series No. 34, London, 1971.
8. Lawler, J. B., "Cut Costs on Flare Stack Design," Hydrocarbon
Process., Vol.46, No. 9, September 1967, pp. 139-142.
9. Vanderlinde, L.G., "State of the Art Review of Smokeless Flares,"
Paper presented at the Regional Meeting, Gulf Coast Gas Processors
Association, Corpus Christi, Texas, 21 February 1974.
10. Communication Between M. G. Klett and John F. Straitz, III, National
AirOil Burner Company, Inc., 26 March 1974.
11. Communication Between M.G. Klett and Mick Sanderson, Flaregas
Corporation, 22 March 1974.
12, American Petroleum Institute, "Guide for Pressure Relief and De-
pressuring Systems," API RP 521, Washington, D. C., 1969.
13. Tan, S. H., "Flare System Design Simplified," Hydrocarbon Process,
Vol.46, No. 1, January 1967, pp.209-212.
14. Kent, G. R., "Practical Design of Flare Stacks," Hydrocarbon Process..
Vol.43, No. 8, August 1964, pp. 121-125.
11?
-------
LMSC-HREC TR D390190
15. Rolke, R.W. et 3.1., "Afterburner Systems Study," EPA-R2-72-062,
Environmental Protection Agency, Washington, D.C., August 1972.
16. Gaydon, A.G., and Wolfhard, H.G., Flames, Their Structure, Radia-
tion and Temperature, Third Edition Revised, Chapman and Hall,
London, 1970.
17. Horton, B.B. et al., "Approaches to the Design of Flare Systems,"
Paper presented at Session 8, Offsite Facilities,11 ASME Petro.
Mech. Engr. Conf., Los Angeles, 17 September 1973.
18. Seebold, J.G., "Curing Flare Systems Noise Provides Major
Challenge," Can. Petrol., January 1973, p. 53.
19. Seebold, J.G., and Hersh, A.S., "Refinery Flare System Injectors
Redesigned for Noise Control," Paper presented at the Petro. Div.,
ASME, New York, 1970.
20. Shore, D., "Toward Quieter Flaring," CEP, Vol.69, No. 10, October
1973, pp.60-64.
21. Bragg, S.L., "Combustion Noise," J.Inst. Fuel, January 1963.
22. London, D.E., "Requirements for Safe Discharge of Hydrocarbons
to Atmosphere," Proc.API Division of Refining, Vol. 43 (III), 1963,
pp.418-433.
23. Kent, G.R., "Make Profit from Flares," Hydrocarbon Process.,
October 1972, pp. 121-123.
24. Werchan, R.E., and Bruce, R.D., "Process Plan Noise Can be
Controlled," CEP Vol.69. No. 10, October 1973, pp. 51-55.
25. Palmer, H.B., Cullis, C.F., "The Formation of Carbon from Gases,"
Chemistry and Physics of Carbon, Vol.1, P. L. Walker, Ed., Marcel
Dekker, Inc., New York, 1965.
26. Parker, W.G., and Woldhard, H.G., J.Chem.Soc., p. 2038, 1950.
27. Communication Between M.G. Keltt and Robert Schwartz, John Zink
Company, 8 February 1974.
28. Lauderback, W., "Unique Flow System Retards Smoke," Hydrocarbon
Process., June 1972, pp. 127-128.
29. Powell, D.T., and Schwartz, R. E., "Borco Uses Low-Level Flare
at Freeport," Oil and Gas Journal, 19 June 1972, pp.^2-63.
118
-------
LMSC-HREC TR D390190
30. Brzustowski, T. A., and E.G. Sommer, Jr., "Predicting Radiant
Heating from Flares," Preprint No. 64-73, API Division of Refining,
Philadelphia, Pa., 1973.
31. Sussman, V. H., Palmer, R. K., Bonanana, F., Steigerwald, B.J.,
and Lunche, R.G., "Emissions to the Atmosphere from Eight
Miscellaneous Sources in Oil Refineries," Report No. 8, Joint District
Federal, and State Project for Evaluation of Refinery Emissions,
Los Angeles County Air Pollution Control District, Los Angeles,
June 1958.
32. Haagen-Smit, A. J., "Chemistry and Physiology of Los Angeles
Smog," T&EC, Vol.44, No. 6, June 1952, pp. 1342-1346.
33. Cantwell, J.E., and Bryant, R.E., "How to Avoid Piping Failures
in: (1) Corrosion by Liquid Metal Attack, (2) Flare Tips by Severe
Cracking," Hydrocarbon Process., May 1973, pp. 114-117.
34, Sternling, C. V., and Wendt, J. D. L.,"On the Oxidation of Fuel Nitrogen
in a Diffusion Flame," AIChE J., Vol.20, No. 1, January 1974, pp.
81-87.
35. Reed, R. D., "Design and Operation of Flare Systems," CEP, Vol.64,
No. 6, June 1968, pp. 53-57.
36. Bodurtha, F. T., Palmer, P. A. and Walsh, W. H., "Discharge of
Heavy Gases from Relief Valves," CEP, Vol.69, No. 4, April 1973,
pp. 37-41.
37. Pasquill, F., "The Estimation of the Dispersion of Windborne Material,
Meteorol. Mag.. Vol.90, No. 1063, February 1961, pp. 33-49.
38. Bosanquet, C.H., and Pearson, J. L., "The Spread of Smoke and
Gases from Chimneys," Trans. Faraday Soc., Vol.32, 1936, pp.
1249-1264.
39, Turner, D. B., Workbook of Atmospheric Dispersion Estimates,
U.S. Dept. Health, Education and Welfare, Pub. Health Service Rept.
No. 999-AP-26, Cincinnati, Ohio, Rev. 1969.
40. Carpenter, S. B. et al., "Principal Plume Dispersion Models.- TVA
Power Plants," APCA J., Vol.21, No. 8, August 1971, pp. 491-495.
41. Coward, H. F., and Jones, G. W., "Limits of Flammability of Gases
and Vapors," U.S. Bureau of Mines, Bulletin No. 503, 1952.
42. Jones, G. W., "Fire and Explosion Hazards of Combustible Gases and
Vapors," in Industrial Hygiene and Toxicology, Vol. 1, 2nd ed., Inter-
science, New York, 1958.
119
-------
LMSC-HREC TR D390190
43, Factory Mutual Engineering Corp,, Handbook of Industrial Loss Pre-
vention, 2nd ed., McGraw Hill, New TorTc, 1967.
44. Zabetakis, M. G., "Flammability Characteristics of Combustible
Gases and Vapors," U.S. Bureau of Mines Bulletin 627, 1965.
45. Sutton, O. G., "A Theory of Eddy Diffusion in the Atmosphere,"
Proc. Roy. Soc., London, A, 135, 1932, pp. 143-165.
46. Constance, J. D., "Calculate Effective Stack Height Quickly,"
Chem. Eng., 4 September 1972, pp. 81-83.
47, Holland, J. Z., "A Meteorological Survey of the Oak Ridge Area,"
AEC Kept. No. ORD-99, Washington, D. C., 1953, p. 584.
48. Briggs, G. A., "Plume Rise," AEC Critical Review Series, TID-
25075, 1969.
49. Heitner, I., "A Critical Look at API RP 521," Hydrocarbon Process.,
November 1970, pp. 209-212.
50. Clean Air Amendments of 1970, PL 91-604.
51. "National Primary and Secondary Ambient Air Quality Standards,"
Federal Register. 36, No. 84, 8186-8201, 30 April 1971.
52. "Proposed Sulfur Oxides Secondary Standards," Federal Register,
Vol.38, No. 87, 7 May 1973, pp. 11355-11356.
53. "Standard of Performance for New Stationary Sources," Federal
Register, Vol.39, No. 47, 8 March 1974.
54. "Title 40 Protection of Environment," Federal Register, Vol. 36,
No, 228, 25 November 1971, pp. 22369-22480.
55. "National Emission Standards for Hazardous Air Pollutants,"
Federal Register, Vol.30, No. 234, 7 December 1971, pp.23239-
23256.
56. King, W. R., "Air Pollution Control," Riegel's Handbook of Industrial
Chemistry, James Kent, ed., Van Nostrand Reinhold Co., New York,
1974.
57. Dubko, S. L., "Air Pollution Control at Gas Processing and Sulfur
Recovery Plants," Department of Health, Government of Alberta,
Canada, February 1969.
58. "Compilation of Air Pollutant Emission Factors," U.S. Environmental
Protection Agency, Office of Air Programs, OAP Publication No.
AP-42, Research Triangle Park, N, C., March 1973,
59. "Guide for Compiling a Comprehensive Emission Inventory," U. S.
Environmental Protection Agency, Publication No. APTD-1135,
Research Triangle Park, N. C., March 1973.
120
-------
LMSC-HREC TR D390190
60. Cross, F, L,., "Environmental Aspects of Site Selection for a Petroleum
Refinery," Proceedings of the Fourth Annual Northeastern Regional
Antipollution Conference, University of Rhode Island, 15 July 1971.
61. International Petroleum Encyclopedia 1973, Petroleum Publishing
Co., Tulsa, Okla., 1973.
62, Husa, H. W., "How to Compute Safe Purge Rates," Hydrocarbon
Process. Petrol. Refiner, Vol.43, No. 5, May 1964, pp. 179-182.
63. Bluhm, W.C., "Safe Operation of Refinery Flare Systems," Paper
presented at 26th Meeting, API Division of Refining, Houston, Texas,
10 May 1961.
64. Reed, R. D., US 3, 055, 417, Assigned to John Zink Company.
65. John Zink Company Sales Folder MS12732MA.
66. National AirOil Burner Co., Inc., Bulletin No. 38, 1972.
67. Gibson, R. O., and D. J. Vinson, "Smokeless Flares Installed at
Gasoline Plants," Oil and Gas Journal, 4 December 1972, pp. 75-77.
68. Peterson, P., "Explosions in Flare Stacks," Presented at AIChE
Petrochemical Exposition, Houston, Texas, 21 February 1967.
69. Patterson, G. C., "Fundamentals of Engineering Offsites and Utilities
for the HPI, Part 6, Pressure Relief and Blowdown," Petrol. Chem.
Engr.. May 1967, pp. 53-58,
70, Lapple, C, E., and Shepherd, C. B,, "Calculation of Particle Trajectories,"
Ind.Eng.Chem.. Vol.32, 1940, pp. 605-617.
71. Craven, A.D., "Thermal Radiation Hazards from the Ignition of Emer-
gency Vents," Paper presented at Symp.on Process Hazards, Inst. of
Chem. Engineers, N. W. Branch, Manchester, England, April 1972.
72. Zabetakis, M. G., and Burgen, D. S., "Research on the Hazards Asso-
ciated with the Production and Handling of Liquid Nitrogen," Bureau
of Mines Report, Invest. 5707, U.S. Dept. Interior, Washington, D. C.,
1961.
73. Kent, G. R., "Find Radiation Effect of Flares," Hydrocarbon Process..
Vol. 47, No. 6, June 1968, pp. 119-130.
74. Hawthorne, W. R,, Weddell, D. S., and Hottel, H. C., Third Symposium
on Combustion and Flame and Explosion Phenomena, Williams & Wilkins,
Baltimore, 1949.
121
-------
LMSC-HREC TR D390190
75. Lewis, B., and von Elbe, G., Combustion, Flames and Explosiona
of Gases, Second Edition, Academic Press, New York, 1961
76. Kent, G. R., "Practical Design of Flare Stacks," Hydrocarbon
Process, and Petrol. Refiner, Vol. 43, No. 8, August 1964, pp.
121-125.
77. Hoehne, V. D., and Luce, R. G., " The Effect of Velocity, Temperature,
and Molecular Weight on Flammability Limits in Wind-Blown Jets of
Hydrocarbon Gases," API Preprint No. 56-70, 1970.
78. Brzustowski, T. A., "A Model for Predicting the Shapes and Lengths
of Turbulent Flames Over Elevated Industrial Flares," Presented at
the 22nd Can. Chem. Eng. Conf., Toronto, 1972.
79. Ban, S. D., Summary Report on Zelten Flare Radiation Study, Battelle
Memorial Inst., Columbus, Ohio, 1970 (draft).
80, Grumer, J. et al., "Hydrogen Flare Stack Diffusion Flames: Low
and High Flow Instabilities, Burning Rates, Dilution Limits, Tempera-
tures, and Wind Effects," Bureau of Mines Rept., Invest. 7457, U.S.
Dept. Interior, Washington, D. C., 1970.
81. Chomiak, J., "A Possible Propagation Mechanism of Turbulent Flames
at High Reynolds Numbers," Combustion and Flame, Vol. 15.
82. Hottel, H, C., "Radiant Heat Transmission," in W. H. McAdams,
Heat Transmission, 3rd Edition, McGraw-Hill, New York, 1954,
pp. 51-125.
83. Chemical Engineers' Handbook, ed. R. H. Perry, C. H. Chilton and
S. D. Kirkpatrick, 4th ed., 5-133, McGraw-Hill, New York, 1963.
84. "Flow of Fluids Through Valves, Fittings, and Pipe," Tech. Paper
410, Crane Co., 1957.
85. Leithe, W., The Analysis of Air Pollutants. Ann Arbor-Humphrey
Science Publishers, Ann Arbor, Mich., 1970.
86. Morrow, N.L., Brief, R.S.,and Bertrand, R . R ., "Air Pollution
Analysis," Chemical Engineering. 24 January 1972.
87. Lengelle, G., and Verdier, L., Gas Sampling and Analysis in Com-
bustion Phenomena, NATO Advisory Group for Aerospace Research
and Development, AGARDograph No. 168, July 1973.
88. Coloff, S.G., "Ambient Air Monitoring of Gaseous Pollutants,"
American Laboratory, July 1973.
122
-------
LMSC-HREC TR D390190
89. West, P. W.. andGaeke, A.G., Anal. Chem. Vol. 28. p. 1816, 1956.
90. Kildal, H., and Byer, R . L., "Comparison of Laser Methods for the
Remote Detection of Atmospheric Pollutants," Proc. IEEE, Vol. 59,
No.12,p.1645,
91. Measures, R.M., "A Comparative Study of Laser Methods of Air
Pollution Mapping," Institute of Aerospace Studies, University of
Toronto, Canada, UTIAS Report No. 174.
92. Hodgeson, J, A., McClenny, W. A., and Haust, P. L,, "Air Pollution
Monitoring by Advanced Spectroscopic Techniques," Science, Vol.82,
19 October 1973.
93. Davis, S.I., Applied Optics. Vol.5, p. 139, 1966.
94, Snowman, L.R., and Gillmeister, R. J., Paper 71-1059, presented
at the Joint Conference on Sensing of Environmental Pollutants,
Palo Alto, Calif., November 1971.
95. American Gas Association Monthly, 57, No. 2, February 1975.
96. Chase, R.L., and George, R. E.,"Contaminant Emissions from the
Combustion of Fuels," JAPA, Vol. 10, No. 1, February I960.
97. Mikatarian, R.R., and Benefield, J, W., "Turbulence in Chemical
Lasers," Paper No. 74-148, AIAA 12th Aerospace Sciences Meeting,
Washington, D. C., February 1974.
98. Kent, J. H., and Bilger.R.W., "Turbulent Diffusion Flames," TN
F-37, The University of Sydney, Austrailia, 1972.
99. Battelle Memorial Institute, "The Federal R&D Plan for Air Pollution
Control by Combustion-Process Modification," Final Report, EPA
Contract CPA 22-69-147, Columbus, Ohio, 11 January 1971.
100. Weekley, G.H., Jr., and Sheelon, F.R., "Jet Compressors Recover
Waste Gases," Hydrocarbon Process., Vol.49, No. 10, October 1966,
pp.165-170.
123
-------
TECHNICAL REPORT DATA
(Please read /HUfuetitMis mi the reverse before completing)
1. REPORT NC
EPA- 600/2 -76-079
4. TITLE ANDSUBT1TLE
Flare Systems Study
2.
7. AUTHOR(S)
M. G. Klett and J. B. Galeski
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Lockheed Missiles and Space Co. , Inc.
Huntsville Research and Engineering Center
4800 Bradford Drive
Huntsville, Alabama 35807
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
15, SUPPLEMENTARY NOTES PrOieCt
Ext 2547.
3. RECIPIt NT'S ACCESSION-NO.
5. REPORT DATE
March 1976
6. PERFORMING ORGANIZATION CODE
8. PCRHORMING ORGANIZATION REPORT UO.
LMSC-HREC TR D390190
10. PROGRAM ELEMENT NO
1AB015; ROAP 21AXM-030
11. CONTRACT/GRANT NO.
68-02-1331, Task 3
13. TYPE OF REPORT AND PEBIOD COVER!- D
Final; 8/74-5/75
14, SPONSORING AGENCY CODE
EPA-ORD
officer for this report is Max Samfield, Mail Drop 62,
is. ABSTRACT rp^ repOrt gjves results of a study of industrial flare technology for control
of gaseous combustible emissions from stationary sources. The study included eval-
uation of present engineering technology, evaluation of existing flare systems,
assessment of present practices and problems, determination of major sources, and
development of research recommendations including potential applications. The
report summarizes emissions data, and gives emission factors for hydrocarbon waste
streams, based on limited available data. It recommends the selection of applicable
flare systems and components for flaring given waste streams. It discusses poten-
tial problems which may affect design and component selection. It gives cost
guidelines based on discussions with flare vendors and users.
17.
a. DESCRIPTORS
KEY WORDS AND DOCUMENT ANALYSIS
b.lDENTIFIERS/OPEN ENDED TERMS C. CQ£>~~ , I'icid/Group
Air Pollution Iron and Steel Industry Air Pollution Control 13B 11F
Flammable Gases Chemical Engineering Stationary Sources 11G 07A
Exhaust Gases Cost Effectiveness Industrial Flares 21B 14A
Hydrocarbons 07C
Industrial Processes 13H
Refineries
EPA Foim 2?20-1 (9-73)
124
------- |