TECHNICAL SUPPORT DOCUMENT
FOR EMISSIONS FROM
PRODUCTION OF FLUORINATED GASES:
FINAL RULE FOR MANDATORY REPORTING OF
GREENHOUSE GASES
Office of Air and Radiation
U.S. Environmental Protection Agency
November 5, 2010
-------
Table of Contents
1. Source Description 1
A. Facilities Subject to the Rule 1
B. How Fluorinated GHGs are Generated and Released at Fluorinated Gas Production
Facilities 1
C. Background on Uses of Fluorinated Gases 2
D. Potential Overlap Emission Points for Production and Suppliers 3
E. Total U.S. Production and Associated Emissions 4
2. Options for Reporting Threshold 5
A. Uncontrolled Emissions Threshold 6
B. Controlled Emissions Threshold 6
C. No Emissions Threshold (i.e., All-in) 7
3. Monitoring Methods, Data Collection and Current Plant Practices 7
A. Scoping speciation and Options for Characterizing Emitted Streams 8
i. Reliance on previous bench- and pilot-scale testing 8
ii. Initial Scoping speciation (including discussion of analytical methods) 8
B. Options for Developing Emissions Estimates 10
i. Default Emission Factor 10
ii. Mass Balance 11
1. Error Limits 12
2. Mass Balance Approach Equations 15
3. Choice of Reactant Whose Yield Is Measured Error! Bookmark not defined.
4. Frequency of Measurement and Calculation 23
5. Reactant and Byproduct Emissions 25
6. Alternative approach based on measurements of balanced elementError! Bookmark not defined
7. Current Plant Practices for Mass Balance 26
iii. Monitoring of Process Vents 27
1. Process-Vent-Specific Emission Factor Approach 27
2. Process-Vent-Specific Emission Calculation Factor Approach 32
3. Obtaining and Maintaining Representative Emission Factors 35
4. Updates to emission factors and emission calculation factors 39
5. Potential Process Vent Emission Threshold 39
6. Process Vent Preliminary Emission Estimates 41
7. Current Plant Practices for Process Vent Monitoring 41
8. Potential use of continuous emissions monitors to measure emissions from
vents 43
9. Equipment Leak Emissions Estimates 43
10. Destruction Efficiency Testing 45
C. Other Potentially Significant Emission Points 50
4. Procedures for Estimating Missing Data 51
5. QA/QC Requirements 52
6. Reporting and Recordkeeping Procedures 52
7. References 54
i
-------
1. Source Description
A. Facilities Subject to the Rule
This source category, fluorinated gas production, covers emissions of fluorinated
greenhouse gases (GHGs) that occur during the production of fluorinated gases. It also covers
emissions of fluorinated GHGs that occur during transformation of fluorinated gases, destruction
of fluorinated GHGs, and venting of residual fluorinated GHGs from returned containers when
those processes are co-located with fluorinated gas production processes. Fluorinated GHGs
include hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), sulfur hexafluoride (SF6),
nitrogen trifluoride (NF3), and other fluorinated GHGs such as fluorinated ethers. Fluorinated
gases include fluorinated GHGs, chlorofluorocarbons (CFCs) and hydrochlorofluorocarbons
(HCFCs). However, the source category excludes generation and emissions of HFC-23 during
the production of HCFC-22, which is covered by subpart O of part 98 of the Mandatory
Reporting Rule.
Producing a fluorinated gas includes the manufacture of a fluorinated GHG, CFC, or
HCFC from any raw material or feedstock chemical. This includes the manufacture of
fluorinated gases for use in a process that will result in their transformation either at or outside of
the production facility, including manufacture of a CFC or HCFC as an isolated intermediate for
use in a process that will result in the transformation of the CFC or HCFC either at or outside of
the production facility. Producing a fluorinated gas does not include the reuse or recycling of a
fluorinated gas, the creation of HFC-23 during the production of HCFC-22, the creation of
intermediates that are created and transformed in a single process with no storage of the
intermediates, or the creation of fluorinated GHGs that are released or destroyed at the
production facility.
B. How Fluorinated GHGs are Generated and Released at Fluorinated Gas
Production Facilities
Fluorinated GHGs can be generated and emitted from production, transformation,
destruction, and other processes at production facilities in a number of ways. Emissions from
fluorinated gas production include fluorinated GHG products that are emitted upstream of the
production measurement and fluorinated GHG by-products that are generated and emitted either
without or despite recapture or destruction. Emissions from fluorinated gas transformation
include emissions of fluorinated GHG feedstocks and possibly by-products of the transformation
process. Emissions from fluorinated GHG destruction include fluorinated GHGs that survive the
destruction process (or that are created as products-of-incomplete-combustion (PICs) from
destruction). Other emissions sources are discussed further below.
Many reactions producing fluorinated GHGs, CFCs, and HCFCs also generate significant
quantities of chemically related by-products, e.g., through over-fluorination or side reactions.
Table 1 provides an overview of some commonly produced fluorinated GHGs, CFCs, and
HCFCs and their known fluorinated GHG by-products. Note that production of CFCs and
HCFCs can generate and emit fluorinated GHGs such as various HFCs and some PFCs. These
fluorinated GHG by-products occur due to the chemical similarities between HFCs, PFCs,
HCFCs, and CFCs and the common use of halogen replacement chemistry to produce them.
Also note that Table 1 is not exhaustive. It omits the most commonly known product and
byproduct, HCFC-22 and HFC-23, which are covered by subpart O of the Mandatory Reporting
1
-------
Rule. In addition, it does not include all of the by-products and products that have been
identified by fluorinated gas producers.
Table 1. Some Fluorinated Products and their By-products
Product
By-product(s)
Source of information
HFC-134a
HFC-143a
Italy National Inventory Report, 2009
sf6
T
LL
o
Italy National Inventory Report, 2009
Electrochemical
fluorination plant
that produces a
broad range of
fluorochemical
products, mainly
for electronics
industry
SFe, CF4,
C2F6, C3F8,
C4F10, C5F12,
C6Fl4, CF3SF5,
C7F16, c8F18
and CsFleO
Belgium National Inventory Report, 2009
httD://unfccc.int/files/national reDorts/annex i aha inventories/national inven
tories submissions/application/zip/bel 2009 nir 15apr.zip
CFC-11, CFC-12
cf4
2006 IPCC Guidelines
CFC-115
c2f6
Italy National Inventory Report, 2009
http://unfccc.int/files/national reports/annex i aha inventories/national inven
tories submissions/application/zip/ita 2009 nir 15apr.zip
Trifluoroacetic acid
(TFA)
CF4, HFC-125
French National Inventory Report, 2009
http://unfccc.int/files/national reports/annex i aha inventories/national inven
tories submissions/application/zip/fra 2009 nir 7apr.zip
The quantities of by-products generated, as a share of the product, depend on the product
and the process. Based on conversations with fluorinated gas producers, by-product generation
rates generally range between one and five percent but can sometimes be higher or lower.
Emissions of products, by-products, and feedstocks may occur from process vents, from
equipment leaks from flanges, connectors, and other equipment pieces in the production line,
from storage tanks storing either raw materials or products, from wastewater streams, from
control devices (e.g., thermal oxidizers), and during the filling of tanker trucks, railcars,
cylinders, or other containers that are distributed by the producer. Undesired by-products may be
deliberately vented, and some product (or reactant) may be vented at the same time due to
imperfect separation of by-products, products, and reactants. Emissions can also occur during
scheduled maintenance and occasional service work on production equipment, during the
blending and recycling of fluorinated GHGs, and during the evacuation of residual fluorinated
GHGs from containers used to distribute products. EPA estimates that total emissions from this
source category were approximately 10.6 million metric tons C02e (mt CChe) in 2006,'"2
C. Background on Uses of Fluorinated Gases
Fluorinated gases are man-made gases used in several sectors. As noted above, they
include fluorinated GHGs (HFCs, PFCs, SF6, NF3, and a number of fluorinated ethers), CFCs,
and HCFCs, all of which are manufactured through various chemical processes.
1 Fluorinated GHG production data were from 2006; CFC and HCFC production data were from 2008.
2 Memorandum from Schaffner, K. and Hancy, C., RTI International, to Ottinger, D., EPA/OAR/CCD. Threshold
Analysis for Emissions, Promulgation of 40 CFR Part 98, subpart L, Fluorinated Greenhouse Gas Production.
November 4, 2010.
2
-------
Hydrofluorocarbons (HFCs) are the most commonly used fluorinated GHGs, used primarily to
replace ozone-depleting substances in a number of applications, including air-conditioning and
refrigeration, foams, solvents, and aerosols. PFCs are used in fire fighting and to manufacture
semiconductors and other electronics. SF6is used in a diverse array of applications, including
electrical transmission and distribution equipment (as an electrical insulator and arc quencher)
and in magnesium casting operations (as a cover gas to prevent oxidation of molten metal).
Nitrogen trifluoride (NF3) is increasingly used in the semiconductor industry, to reduce overall
semiconductor GHG emissions through processes such as NF3 remote cleaning and NF3
substitution during in-situ cleaning. Fluorinated ethers (HFEs and HCFEs) are used as
anesthetics (e.g., isofluorane, desflurane, and sevoflurane) and as heat transfer fluids (e.g., the H-
Galdens). The manufacture of CFCs and HCFCs for emissive uses is being phased out under the
Montreal Protocol, but production of these materials for use as feedstocks is permitted to
continue indefinitely. As discussed further below, the volume of CFCs and HCFCs used as
feedstocks in the U.S. is significant; even excepting HCFC-22, this volume is actually larger than
the volume of fluorinated GHGs produced in the U.S. Fluorinated GHGs are powerful
greenhouse gases whose ability to trap heat in the atmosphere is often thousands to tens of
thousands of times greater than that of CO2, on a pound-for-pound basis. Some fluorinated
GHGs are also very long lived; SF6 and the PFCs have lifetimes ranging from 3,200 to 50,000
years.3
D. Potential Overlap Emission Points for Production and Suppliers.
The facilities that are covered under the fluorinated gas production source category are
many of the same facilities that are covered under the industrial gas supply source category. In
general, the industrial gas supply source category is intended to cover or capture the quantities of
fluorinated GHGs that are entering and leaving the U.S. supply of industrial gases, (i.e., amount
of product), while the fluorinated gas production source category is intended to cover or capture
the quantities of fluorinated GHGs emitted at fluorinated gas production facilities. Specifically,
the industrial gas suppliers source category is meant to track the quantities of fluorinated GHGs
that are (1) produced, (2) transformed, (3) destroyed, (4) imported, and (5) exported. The
industrial gas suppliers source category would essentially track the amount of final product made
(produced) at a production facility but not the emissions of fluorinated GHG from the production
steps from raw material to final product.
There are several areas of potential overlap between the emissions that are reported under
the industrial gas suppliers source category and those that could be reported under the fluorinated
gas production source category. The areas of overlap concern "downstream" emissions, meaning
they occur at the fluorinated GHG production facility after the fluorinated GHG product
measurement. Downstream emissions include those from container filling (if this occurs after
the production measurement), fluorinated GHG transformation processes, destruction of
fluorinated GHGs that are removed from the U.S. supply, recycling or reclamation of fluorinated
GHGs, and evacuation of fluorinated GHG heels from returned cylinders or containers.
("Upstream" emissions occur at the fluorinated GHG production facility prior to the fluorinated
GHG product measurement.)
3 IPCC, 2006.
3
-------
Fluorinated gas production facilities sometimes make a fluorinated GHG product that is
later used as a raw material in making another product. For example, fluorinated GHG product is
used as a raw material in transformation processes to produce fluoropolymers (the fluoropolymer
product would not be considered a fluorinated GHG). Emissions of fluorinated GHGs that occur
from transformation processes are an area of potential overlap, since the quantities produced for
transformation may be reported under the industrial gas suppliers source category.
Transformation processes may occur at the same site where fluorinated GHG products are
produced, or they may occur at another fluorinated GHG production facility to which the
fluorinated GHG feedstock is shipped.
The emissions of fluorinated GHG from destruction processes used to remove fluorinated
GHG from the U.S. supply is another area of overlap. The fluorinated GHG is returned to the
facility and subtracted from the U.S. supply, however some emissions may occur from the
destruction process, either as fugitives from the handling of the fluorinated GHG or as
undestroyed emissions at the outlet of the destruction device.
Filling cylinders or other containers (loading emissions) with fluorinated GHG product
and blending of fluorinated GHG with other gases will most likely occur after the production
measurement, classifying emissions for cylinder filling (loading) and blending as downstream
processes.
Recycling of used gas may be performed by the producers of new gas or by offsite
recycling firms. Emissions may occur during handling and purification of old gas and during
packaging of recycled gas.4 Since the used gas has already been counted as produced (used gas
is removed from, and then added back into, the gas supply), recycling emissions are considered
to be downstream of the production measurement.
Evacuation of fluorinated GHG from returned cylinders or containers results in emissions
that vary depending on process type and the composition and amount of the heel (residual gas)
present.
In theory, it might be possible to track emissions from transformation and destruction
simply using quantities reported under 40 CFR part 98 subpart 00. However, this would require
that (1) fluorinated GHGs that are produced only to be transformed or destroyed be tracked
separately, (2) production, transformation, and destruction be measured to very good precision
and accuracy (e.g., 0.2 percent), and (3) that no by-products be formed or emitted during these
processes. If all of these conditions were met, emissions could be equated to the differences
between production and transformation and production and destruction. In practice, however, it
would be difficult to meet all of these conditions.
E. Total U.S. Production and Associated Emissions
The production of fluorinated gases includes production of fluorinated GHG (HFCs,
PFCs, SF6, and NF3, HFEs), CFCs and HCFCs (except for HCFC-22 production processes). In
2006 (some data are from 2003 and 2007), 12 U.S. facilities collectively produced over 350
million mtC02e (170,000 tons) of HFCs, PFCs, SF6, and NF3. EPA estimates that an additional
4 2006 IPCC, Chapter 3.10.
4
-------
six facilities collectively produced 1 million mtCC^e of fluorinated anesthetics (HFEs) in 2005.
The total U.S. production of CFCs and HCFCs (except HCFC-22) in 2008 was estimated to be
approximately 266,000 tons. The majority of the CFC and HCFC production processes are co-
located at facilities that also produce fluorinated GHGs. Another 2 facilities that produce CFC
and HCFC products were identified that were not already included in the 12 facilities identified
above for HFC, PFC, SF6, and NF3 production. Fluorinated gas production processes are
sometimes co-located with fluorinated gas transformation processes or fluorinated GHG
destruction processes. The total amounts of fluorinated gases transformed into other products or
destroyed are not known.
Fluorinated GHG emissions (actual emissions, following application of control
techniques where applicable) from U.S. facilities producing fluorinated GHGs are estimated to
range from 0.8 percent to 2 percent of the amount produced, depending on the facility. An
emission rate of 1.5 percent is assumed for fluorinated GHG production processes. At an
emission rate of 1.5 percent, the 12 fluorinated GHG facilities together are estimated to have
emitted approximately 5.3 million mtCC^e of HFCs, PFCs, SF6, andNF3. The six additional
HFE production facilities are estimated to have emitted approximately 17,000 mtC02e of
fluorinated anesthetics, using an emission rate of 1.5 percent.
The quantity of emissions of fluorinated GHGs from production of CFCs and HCFCs is
uncertain. However, the magnitude of by-product generation during F-GHG production (e.g.,
quantities equivalent to one to five percent of the mass of the product) suggests that significant
quantities of by-product F-GHGs may be generated during production of chemically similar
substances. Given the substantial amounts of CFCs and HCFCs that are produced in the U.S.,
emission rates could be relatively low and still result in significant emissions. In 2008, the
combined tonnage of U.S. HCFC and CFC production was higher than that of all fluorinated
GHG production in 2006 (the latest year for which data were available), with CFCs produced as
feedstocks comprising the majority. (Although production of HCFCs and CFCs is limited under
the regulations implementing Title VI of the CAA, production of these substances for use as
feedstocks is permitted to continue indefinitely.) Assuming that this production resulted in
emissions of fluorinated GHG by-products equal to one percent of the mass of CFCs and HCFCs
produced,5 and that these byproducts had an average global warming potential (GWP) of 2000,
emissions from CFC and HCFC production are estimated to have totaled 5.3 million mtCC^e in
2008, the same as estimated emissions from fluorinated GHG production in 2006. (No estimate
of fluorinated GHG emissions from transformation processes, or fluoropolymer production
processes, is available.) These emissions, combined with those from fluorinated GHG
production, lead to a total for the source category of 10.6 million mtCC^e.
2. Options for Reporting Threshold
EPA evaluated a range of threshold options for fluorinated gas production facilities.
These included 1,000 mt C02e, 10,000 mtCC^e, 25,000 mtCC^e, and 100,000 mtCC^e.
Emission levels on both an uncontrolled and a controlled basis were evaluated. Because EPA
5The emission rate from CFC and HCFC production is assumed to be lower than that from fluorinated GHG
production because only by-products are included in the former, while both products and by-products are included in
the latter.
5
-------
has little information on combustion-related emissions at fluorinated gas production facilities,
these emissions were not included in the threshold analysis.
A. Uncontrolled Emissions Threshold
Facility-specific uncontrolled emissions (i.e., pre-control) were estimated for fluorinated
GHG production processes (including fluorinated anesthetics production processes) by
multiplying a factor of 3.0 percent by the estimated production at each facility. For CFC and
HCFC production processes (except for HCFC-22 production processes), uncontrolled emissions
were estimated by multiplying a factor of 2 percent by the estimated production at each facility.
Uncontrolled emissions are strongly influenced by by-product generation rates, which are known
to vary between zero and several percent for fluorinated gas production processes; thus, these
estimates are uncertain. The results of the analysis for uncontrolled emissions from the
production of HFCs, PFCs, SF6, NF3, CFCs, and HCFCs are shown in Table 2. (No emissions of
fluorinated GHG from transformation processes, such as fluoropolymer production processes,
are included in the analysis.)
All or most production facilities would be covered by uncontrolled emission thresholds,
i.e., 14 of 14 or 13 of 14 facilities would be covered depending on the cutoff level. Use of the
uncontrolled emissions threshold would not allow for applying reductions from destruction
devices, which may or may not be achieving the demonstrated destruction efficiency on a
continuous operating basis. Under the uncontrolled emissions threshold option, emissions
reductions would not be overstated and emissions to the atmosphere would not be understated.
(It is uncertain where the emission levels of the individual anesthetic facilities would fall, and the
anesthetic facilities and emissions are not included in Table 2. There are 6 anesthetic facilities
that are estimated to generate a total of approximately 30,000 mt C02e of emissions, by
multiplying a factor of 3.0 percent by the estimated production.) Emissions from combustion
sources at facilities are not included in the emissions estimate shown in Table 2.
Table 2. Threshold Analysis for Fluorinated GHG Emissions from Production of HFCs, PFCs, SF6,
NF3, CFCs, and HCFCs (Uncontrolled Emissions)
Threshold
Level
(mt C02e/r)
Total National
Emissions
(mt C02e)
Number of
Facilities
Emissions Covered
Facilities Covered
mt C02e
Percent
Number
Percent
1,000
10,600,000
14
10,600,000
100%
14
100%
10,000
10,600,000
14
10,600,000
100%
14
100%
25,000
10,600,000
14
10,600,000
100%
14
100%
100,000
10,600,000
14
10,600,000
100%
13
93%
B. Controlled Emissions Threshold
Facility-specific controlled emissions (i.e., those following the control device, if in place)
were estimated for fluorinated GHG production processes (including anesthetics production
processes) by multiplying a factor of 1.5 percent by the estimated production at each facility.
For CFC and HCFC production processes (except for HCFC-22 production processes),
controlled emissions were estimated by multiplying by a factor of 1.0 percent by the estimated
production at each facility. Controlled emissions were assumed to be half of uncontrolled
6
-------
emissions at each facility. The results of the analysis for controlled emissions from the
production of HFCs, PFCs, SF6, NF3, CFCs, and HCFCs are shown in Table 3. No emissions of
fluorinated GHG from transformation processes (other than those that produce fluorinated
GHGs) are included in the analysis.
All or most of the production facilities would be covered by controlled emission
thresholds, depending on the cutoff, although fewer facilities would be covered under the
controlled emission threshold option (10 facilities) than under the uncontrolled emission
threshold option (13 facilities) at the highest cutoff (i.e., 100,000 mtC02e). Use of the controlled
emissions threshold option would allow facilities to account for emission reductions achieved by
destruction devices that had a demonstrated destruction efficiency. It is possible that emission
levels may be understated using the controlled emissions option. (It is uncertain where the
emission levels of the individual anesthetic facilities would fall, and the anesthetic facilities and
emissions are not included in Table 3. There are 6 anesthetic facilities that are estimated to emit
a total of approximately 15,000 mt C02e of emissions, by multiplying a factor of 1.5 percent by
the estimated production.) Emissions from combustion sources at facilities are not included in
the emissions estimate shown in Table 3.
Table 3. Threshold Analysis for Fluorinated GHG Emissions from Production of HFCs, PFCs, SF6,
NF3, CFCs, and HCFCs (Controlled Emissions)
Threshold
Level
(mt CO e/r)
Total
National
Emissions
(mt CO e)
Number of
Facilities
Emissions Covered
Facilities Covered
mt CO e
Percent
Number
Percent
1,000
10,600,000
14
10,600,000
100%
14
100%
10,000
10,600,000
14
10,600,000
100%
14
100%
25,000
10,600,000
14
10,600,000
100%
14
100%
100,000
10,600,000
14
10,300,000
97%
10
71%
C. No Emissions Threshold (i.e., All-in)
Under an "all-in" approach, no emissions threshold level would be included in the rule
and all facilities in the fluorinated gas source category, including facilities producing anesthetics,
would be subject to the rule regardless of emission levels. The all-in approach would ensure that
all facilities identified and quantified their fluorinated GHG emissions, even if they initially
believed those emissions to be small. Some facilities that initially believed their emissions to be
small could find that those emissions actually exceeded 25,000 mt C02e (e.g., because a
previously unidentified fluorinated GHG with a high GWP was being generated and emitted as a
by-product). Facilities whose emissions remained below 25,000 mt C02e or 15,000 mt C02e
could cease reporting after five or three years respectively. In addition, an all-in approach would
provide an essentially level playing field because all facilities in the source category would be
subject to the rule and be subject to the same economic impact from the reporting rule.
However, an all-in approach would also increase the burden on facilities with uncontrolled
emissions below 25,000 mt C02e. This could result in higher reporting costs per metric ton C02e
for these facilities than for higher emitting facilities.
3. Monitoring Methods, Data Collection and Current Plant Practices
7
-------
A. Scoping Speciation and Options for Characterizing Emitted Streams
/'. Reliance on previous bench- and pilot-scale testing
During research and development of a fluorinated gas product, companies develop
laboratory, bench-scale, and pilot-scale processes to manufacture new products. During
laboratory and bench-scale development of the process, companies generate small quantities of
the product. The companies monitor and measure all aspects of the process, including rates and
extent of reaction, reactant and full product and byproduct identity and compositions, and
physical characteristics of the compounds. This testing and analysis of compounds in the various
process streams generally identifies the product and all byproducts generated in the process.
During these laboratory or bench-scale processes, process optimization is addressed by defining
operating conditions that generate the best product yields and least byproducts, considering the
most efficient use of resources. Following the laboratory and bench-scale level of process
development, a process may be scaled to a larger basis, i.e., pilot-scale basis, to continue
research and development of the process and to continue to refine process optimization. The
pilot-scale process, meant to simulate a full-scale production process, generates a larger quantity
of product than the laboratory scale process but less than a full-scale production process. The
equipment used in pilot-scale processes is similar in design to what the full-scale process might
be, and is used to continue research on the process, as the data available from earlier laboratory
and bench-scale systems are applied. However, some data may not be directly applicable to the
pilot-scale system. As part of the continued research, the company continues to test and analyze
the products and byproducts from the pilot-scale process to confirm what compounds are
generated in the process as operating conditions are refined (i.e., changed).
The findings regarding the identities of the compounds generated during these stages of
process research and development may be used to help understand the identities of the
compounds present in emission and waste streams for the full-scale production process. The
analyses are typically rigorous as the company is eager to know and learn as much as possible
and limit side reactions and increase yields. A potential drawback to use of the laboratory and
pilot scale compound and composition analyses is that there may be some differences in the
products and byproducts as the process is scaled from lab oratory/bench-scale to pilot-scale to
full-scale production units. The earlier data may not be fully applicable to the larger scale
processes. Differences are likely to occur between the laboratory and pilot-scales, and streams
may need to be tested and analyzed again to detect changes between the pilot- and full-scale
production processes.
Initial Scoping Speciation
To ensure that all fluorinated GHGs that occur in emitted streams are accurately
identified, an initial scoping speciation or test on vents and streams from fluorinated gas
processes may be conducted. The initial scoping speciation would be conducted using methods
that allow detection and identification of all compounds in the vent or stream. The identity of the
specific compounds that are generated during the fluorinated gas production process is important
to the emissions estimate. Some facilities have indicated that even with extensive laboratory and
bench-scale compositional and quantitative analysis from research, development, and design
stages of the process, unexpected F GHGs can be found in emissions streams.6 In some cases,
6 Based on conversations with Fluorinated GHG producers, as referenced in section 7.
8
-------
this may have occurred because the analytical methods used at the pilot- or laboratory-scale were
not sensitive enough to detect fluorinated GHGs that occurred at low (but still higher than trace)
concentrations; in other cases, it may have occurred because subtle differences between the
laboratory- and full-scale processes led to the generation of new or different by-products.
The results of the scoping speciation can be used to inform both the mass-balance and
emission-factor approaches (described further below). In the mass balance approach, a total
fluorine balance is conducted to account for all fluorine into the process (i.e., in the form of
fluorine-containing reactants) and all fluorine out of the process (i.e., in the form of the fluorine-
containing product, and destroyed or recaptured fluorine-containing by-products, unconsumed
reactants, or products). The difference, in terms of fluorine, is assumed to be emitted. The
facility would determine what fraction of the total fluorine is emitted as the fluorine-containing
reactant, by-products, or product by conducting emission characterizations for process vents by
measurement to speciate the total fluorine. Because byproducts are created, knowledge of these
informs the emissions estimate so that the byproduct compounds are appropriately represented in
the emissions. For example, for facilities using the mass-balance approach, the scoping
speciation can be used to determine whether some emissions are correctly assumed to occur in
the form of the reactant, product, or by-products. For those facilities using an emissions factor
approach, the results from the scoping speciation will ensure that emissions factors are
appropriately developed for all byproduct compounds. The scoping speciation will identify
byproducts that should be measured in subsequent emissions testing to develop emission factors.
Note that some facilities may find it most convenient to conduct the initial scoping speciation,
the emission measurements for mass balance emission characterization, and/or the emission
testing for emission factor development at the same time).
Accurately identifying the compounds emitted provides for a better fluorinated GHG
emissions estimate and also a better C02e emissions estimate because the appropriate GWPs are
applied to the emitted compounds. If there is a difference between the GWPs of a product and a
byproduct and if the emissions are assumed to consist exclusively of the product, the C02e
emissions could be overestimated or underestimated. Underestimation is a particular concern
where the process unexpectedly generates a fluorinated GHG that is difficult to destroy (e.g.,
CF4). In this case, applying the destruction efficiency (DE) of the destruction device to the
emissions may overestimate destruction and underestimate emissions.
Facilities would conduct an initial scoping speciation on each fluorinated gas process at
the facility that has at least one process vent over a threshold level.EPA evaluated emissions
cutoffs on a process vent basis to be consistent with other approaches where emission cutoffs are
also on a process vent basis. Only those processes with at least one process vent with fluorinated
GHG emissions above a cutoff level would conduct the initial scoping speciation. We
considered a limit based on C02e on an uncontrolled basis for each process vent, and note that a
cutoff based on fluorinated GHG, rather than C02e, emissions may be appropriate because the
identity and the GWP of some fluorinated GHG compounds may not be known. The fluorinated
GHG emissions cutoff for process vents would be 1 metric ton fluorinated GHG. The estimate
of emissions from each process vent could be determined using standard engineering
calculations, previous measurements, and engineering assessments (i.e., similar to preliminary
emission estimates discussed under process-vent-specific emissions calculation factors below).
Because the specific fluorinated GHG byproducts may not be known, it may be appropriate to
9
-------
apply the 1-metric-ton-fluorinated-GHG-emission cutoff to uncontrolled emissions. As noted
above, the application of the destruction device's DE may underestimate emissions (and
therefore fail to trigger testing) if difficult-to-destroy fluorinated GHGs occur unexpectedly in
the stream.
A facility would not have to measure every stream or process vent to complete the initial
scoping speciation. For each process with fluorinated GHG emissions above the cutoff level for
process vents, facilities could select which process vents or streams to measure, focusing on
those streams that are most likely to contain all of the fluorinated GHG (e.g., byproducts)
anticipated to be generated or emitted. It is preferable that these fluorinated GHGs be at their
maximum concentrations, although streams with smaller concentrations of byproducts may be
tested if the scope and sensitivity of the analytical method allow detection of the compounds at
lower concentration levels .
Sampling and analytical methods capable of detecting and speciating fluorinated GHG
compounds and capable of identifying multiple compounds simultaneously would be the best
choices for the initial scoping speciation. For example, methods that use gas chromatography
with mass spectrometry (GC/MS) are capable of detecting and identifying multiple fluorinated
GHG compounds. Another example of an appropriate sampling and analytical method includes
Fourier transform infrared (FTIR) analysis. There are several EPA reference methods and other
consensus vetted methods that may be used to sample and analyze fluorinated GHG; these
methods include EPA Method 18 - Measurement of Gaseous Organic Compound Emissions by
Gas Chromatography (GC/ECD, GC/MS), EPA Test Method 320 Measurement of Vapor Phase
Organic and Inorganic Emissions by Extractive Fourier Transform Infrared (FTIR)
Spectroscopy, and ASTM D6348-03 (FTIR) Standard Test Method for Determination of
Gaseous Compounds by Extractive Direct Interface Fourier Transform Infrared (FTIR)
Spectroscopy. Each of these is briefly described below in the section on Process-Vent-Specific
Emission Factor Approach. If speciation measurements are conducted at the stack, the facility
would also conduct an EPA Method 1 series (Sample and Velocity Traverses for Stationary
Sources), Method 2 series (Determination of Stack Gas Velocity and Volumetric Flow Rate),
Method 3 series (Gas Analysis for Carbon Dioxide, Oxygen, Excess Air, and Dry Molecular
Weight), and Method 4 (Determination of Moisture Content in Stack Gases). If a facility is not
able to conduct EPA Method 2, alternative flow rate methods could be used, such as OTM 24
(Tracer Gas Protocol For the Determination of Volumetric Flow Rate Through the Ring Pipe of
the Xact Multi-Metals Monitoring System), or ALT-012 (Emission Measurement Center
Approved Alternative Method). Other validated methods that are capable of detecting the
analyte of interest at the concentration of interest could also be used.
B. Options for Developing Emissions Estimates
/'. Default Emission Factor
One option that was considered for characterizing fluorinated GHG is the use of default
emission factors. These factors would provide a rough estimate of the quantity of fluorinated
GHGs emitted. Although the default emission factor approach is simple and easily-implemented,
it is also highly imprecise; emissions of fluorinated GHG products in U.S. plants are estimated to
10
-------
vary from 0.8 percent to 2 percent of production, more than a factor of two (RTI, 2008).7 Thus,
applying a default factor (1.5 percent, for example) is likely to significantly overestimate
emissions at some plants while significantly underestimating them at others.
Under a somewhat more rigorous default emission factor approach, emission factors
would be developed for similar types of emission points at multiple facilities within a source
category. The emission factors for each type of emission point would be averaged to develop an
industry-wide emission factor for that type of emission point. Because these EFs would represent
average values over multiple facilities, some facilities would be above the EF and some would be
below. As a result, such factors may not provide an accurate emissions determination for a
single facility. Default EFs would not account for inter-facility variability that can occur due to
the age of facility and equipment, equipment sizes, equipment design, differences in raw
materials and suppliers, different typical operating parameters, and variations in the process
operation that may not be readily apparent or known. In addition, default EFs would not provide
information that EPA may need to compare the best and worst performers, or provide insights
into which facilities have the most or fewest opportunities to obtain additional reductions.
Mass Balance
In the total fluorine mass-balance approach, emissions are determined in terms of total
fluorine8 rather than the fluorinated GHG product. Facilities determine the total fluorine emitted
from the process by determining the total fluorine fed to the process, subtracting the total
fluorine in the product resulting from the process, and subtracting the total fluorine that is
destroyed or recaptured. Facilities would weigh or meter the reactants fed into the process, the
product resulting from the process, and any streams that are destroyed and recaptured that
contain reactants, byproducts, and product. Destroyed streams include those sent to the thermal
oxidizer or other equipment, and recaptured streams may be sold, sent to another facility for
destruction, or held for another purpose.9 Facilities calculate the total fluorine emitted as the
difference in the total fluorine mass fed into the process, the total fluorine mass of the main
product, and the total fluorine mass that is destroyed or recaptured.10 As discussed further
below, it is EPA's understanding that some facilities perform similar types of measurements and
calculations to monitor their processes and yields.
The difference is then assigned to loss of reactants, loss of product, and/or loss of
byproducts. Facilities could assume that all the fluorine is emitted in the form of the fluorinated
GHG that has the most significant GWP, or facilities could determine from measurements the
7 The emission rates cited were estimated for HCFC-22, which is an ozone-depleting substance and therefore
excluded from the proposed rule. However, the production processes for many fluorinated GHGs are similar to
those for HCFC-22 (2006 IPCC Guidelines, Volume 3, section 3.10.2.1), and therefore their emission rates are
likely to be similar as well.
8 Facilities could use another element as long as it occurred in all of the fluorinated GHGs fed into or generated by
the process.
9 Note that if these recaptured materials are fed back into the process, they must be counted at that time as fluorine
additions to the process.
10 The fluorine itself, of course, is not destroyed, but any fluorinated GHGs containing the fluorine are (to the
destruction efficiency of the device). In the equations below, non-GHG fluorine-containing compounds that are
removed from the process are treated as completely destroyed, since none of the fluorine in them is emitted as a
fluorinated GHG. Fluorinated GHGs are assumed to be destroyed to the DE of the device for each fluorinated GHG.
11
-------
fractions of the total fluorine emissions that consist of reactants, product, and byproducts. There
may be only one reactant that contains fluorine, or if there are several reactants, the fluorine
would be summed for the reactants. There may be processes with one by-product, or cases with
several by-products, where the fluorine would be summed for all the by-products. There is likely
one product.
Because the mass-balance approach relies on the calculation of relatively small
differences between relatively large numbers, measurements must have good precision to avoid
large uncertainties in the emissions estimates. EPA evaluated methods to estimate and limit the
error associated with the mass balance approach.
1. Error Limits
Precision and Accuracy Requirements for Individual Measurements
One approach to limiting the error associated with emissions estimates developed using
the mass-balance approach is to establish specific precision and accuracy requirements for each
measurement used in the mass-balance equation. These measurements include mass
measurements for reactants, products, and fluorinated GHG by-products, including feed streams,
final product streams, emitted streams, and destroyed or recaptured streams. In addition, they
include concentration measurements where products or by-products occur in streams with other
substances. To limit the uncertainty of the emission estimate to ± 30 percent of the estimate,
devices used to measure masses (e.g., flowmeters and scales) would need to be able to achieve
precisions and accuracies near ± 0.2 percent or better, at least for streams with a large impact on
the mass-balance calculation. Concentration measurements would need to be conducted with
precisions and accuracies of ± 10 percent or better. This approach limits error and is
straightforward to implement and enforce. However, it may require substantial expenditures to
obtain accurate and precise measurements of quantities whose errors have little impact on the
overall error of the emissions estimate. For example, under this approach, facilities might be
required to upgrade flowmeters in streams that affect the emission estimate (and its uncertainty)
only slightly. The approach also limits flexibility. Even a facility with a relatively large error in
one stream may be able to bring the total error of its emissions estimate to a tolerable level by
improving the accuracy and precision of other measurements that are used in the mass-balance
equation, such as the mass flows of reactants and products. Thus, while this approach may be
appropriate as an additional option for ensuring that emissions estimates are accurate and precise,
it is probably not appropriate as the sole option. (See section 6 below for more discussion of this
alternative approach.)
Error Limit for the Emission Estimate as a Whole.
Another approach to limiting the error of the mass-balance approach is to establish a limit
on the error of the overall emission estimate. To estimate the statistical error associated with use
of the mass-balance approach, facilities would be required to use error propagation, considering
the accuracy and precision of their measurements and the calculation methods of the mass-
balance approach. Under this approach, EPA would not specify precision and accuracy
requirements for individual mass or concentration measurements. Instead, EPA would require
that the error associated with the overall estimate of fluorinated GHG emissions fall under a
specific limit. This limit could be expressed in terms of a fraction of the emissions estimate
12
-------
(relative error) or as a specific quantity (absolute error). Facilities could achieve this level of
precision and accuracy however they chose.
A maximum relative error of ± 30 percent would be comparable to that cited by a facility
that has used an emission factor approach to estimate its fluorinated GHG emissions. It is also
comparable to the error that EPA calculates for a facility with an emission rate of two percent
and with good precisions and accuracies for its mass flow measurements (± 0.2 percent) and for
its concentration measurement (±10 percent) of a waste stream constituting five percent of the
process's fluorinated GHG output flow. For facilities whose emissions constitute a very small
share of their inputs and outputs (e.g., one percent or less), a relative error of ± 30 percent would
be very difficult to achieve using a mass-balance approach. At the same time, the absolute error
of such a facility's estimate may be smaller than the absolute error of a facility that meets the
relative error test but that has a higher emission rate. It may therefore be appropriate to establish
a maximum permissible absolute error of 3,000 mtCC^e for facilities whose estimates have
relative errors greater than 30 percent. This absolute error is equivalent to 30 percent of the
10,000 mtC02e threshold that is used elsewhere in the subpart to establish requirements for
different sources (e.g., process vents). Under this approach, processes whose emissions were
lower than 10,000 mtCC^e could have relative errors higher than 30 percent so long as they met
the limit on absolute error. This approach would avoid penalizing processes and facilities with
low emissions.
The absolute and relative errors associated with using the mass balance approach on a
process may be estimated using Equations L-l, L-2, L-3, and L-4 in conjunction with either
Equations L-7 through L-10 or Equation L-17. Alternatively, facilities may estimated these
errors based on the variability of previous process measurements (e.g., the variability of
measurements of stream concentrations), provided these measurements are representative of the
current process and current measurement devices and techniques. Once errors have been
calculated for the quantities in these equations, those errors would be used to calculate the errors
in Equation L-6 and L-5. Where the measured quantity is a mass, the error in the mass would be
equated to the accuracy or precision (whichever is larger) of the flowmeter, scale, or combination
of volumetric and density measurements at the flowrate or mass measured. Where the measured
quantity is a concentration, the error of the concentration would be equated to the accuracy or
precision (whichever is larger) with which the concentration measurements estimate the mean
concentration of that stream component, accounting for the variability of the process, the
frequency of the measurements, and the accuracy or precision (whichever is larger) of the
analytical technique used to measure the concentration at the concentration measured. If the
variability of process measurements is used to estimate the error, this variability would be
assumed to account both for the variability of the process and the precision of the analytical
technique. Facilities may use standard statistical techniques such as the student's t distribution to
estimate the error of the concentration measurements as a function of process variability and
frequency of measurement.
Equation L-l provides the general formula for calculating the absolute errors of sums and
differences where the sum, S, is the summation of variables measured, a, b, c, etc. (e.g., S = a + b
+ c):
13
-------
eSA =[(«*02+(6*eJ2 +(C*ec)2\'2
(Eq. L-l)
Where:
esA = absolute error of the sum, expressed as one half of a 95 percent confidence interval.
ea = relative error of a, expressed as one half of a 95 percent confidence interval,
eb = relative error of b, expressed as one half of a 95 percent confidence interval.
ec = relative error of c, expressed as one half of a 95 percent confidence interval.
Equation L-2 provides the general formula for calculating the relative errors of sums and
differences:
eSR = ( 6f x (Eq. L-2)
\a + b + c)
Where:
eSR = relative error of the sum, expressed as one half of a 95 percent confidence interval.
esA = absolute error of the sum, expressed as one half of a 95 percent confidence interval,
a+b+c = sum of the variables measured.
Equation L-3 provides the general formula for calculating the absolute errors of products
(e.g., flow rates of GHGs calculated as the product of the flow rate of the stream and the
concentration of the GHG in the stream), where the product, P, is the result of multiplying the
variables measured, a, b, c, etc. (e.g., P = a*b*c):
ePA
(a'b'ciel+el+ef}'11 (Eq. L-3)
Where:
epA = absolute error of the product, expressed as one half of a 95 percent confidence
interval.
ea = relative error of a, expressed as one half of a 95 percent confidence interval,
eb = relative error of b, expressed as one half of a 95 percent confidence interval.
ec = relative error of c, expressed as one half of a 95 percent confidence interval.
Equation L-4 provides the general formula for calculating the relative errors of products:
ePR=TT(Eci-L-4)
[a* b*c)
Where:
epR = relative error of the product, expressed as one half of a 95 percent confidence interval.
epA = absolute error of the product, expressed as one half of a 95 percent confidence
interval.
14
-------
a*b*c = product of the variables measured
To estimate the annual C02e emissions of the process for use in the error estimate,
facilities would apply the mass-balance equations below to representative process measurements.
If these process measurements represented less than one year of typical process activity, facilities
would adjust the estimated emissions to account for one year of typical process activity. To
estimate the terms FERd, FEP, and FEBk for use in the error estimate for Equations L-l 1, L-12,
and L-l 3 of this section, facilities could use emission testing, monitoring of emitted streams,
and/or engineering calculations or assessments, or in the alternative assume that all fluorine is
emitted in the form of the fluorinated GHG that has the highest GWP among the fluorinated
GHGs that occur in more than trace concentrations in the process. To convert the fluorinated
GHG emissions to C02e, facilities could use Equation A-l of §98.2. For fluorinated GHGs
whose GWPs are not listed in Table A-l to subpart A of this part, facilities could use a default
GWP of 2,000.
2. Mass Balance Approach Equations.
Estimating Fluorine Emissions
The following set of equations summarizes the mass-balancing method for estimating
fluorinated GHG emissions based on a total fluorine balance (another element could also be used
in place of fluorine).
The total mass of each fluorinated GHG emitted annually from each fluorinated gas
production process and each fluorinated GHG transformation process would be estimated by
using Equation L-5:
EFGHGf = ^ ^Rp-FGHGf + ^pP-FGHGf + ^Bp-FOHOf) (ECl- L"5)
p=1
Where:
EFGHGf = Total mass of each fluorinated GHG f emitted annually from production or
transformation process i(metric tons).
ERp-FGHGf = Total mass of fluorinated GHG reactant f emitted from production process
i over the period p (metric tons, calculated in Equation L-l 1).
Epp.FGHGf = Total mass of the fluorinated GHG product f emitted from production
process i over the period p (metric tons, calculated in Equation L-12).
EBp-FGHGf = Total mass of fluorinated GHG by-product f emitted from production
process i over the period p (metric tons, calculated in Equation L-13).
n = Number of concentration and flow measurement periods for the year.
The total mass of fluorine emitted from process i over the period p would be estimated at
least monthly by calculating the difference between the total mass of fluorine in the reactant(s)
(or inputs, for processes that do not involve a chemical reaction) and the total mass of fluorine in
15
-------
the product (or outputs, for processes that do not involve a chemical reaction), accounting for the
total mass of fluorine in any destroyed or recaptured streams that contain reactants, products, or
by-products (or inputs or outputs). An element other than fluorine may be used in the mass-
balance equation, provided the element occurs in all of the fluorinated GHGs fed into or
generated by the process. In this case, the mass fractions of the element in the reactants,
products, and by-products must be calculated as appropriate for that element. This calculation
must be performed using Equation L-6:
Er=Y. Wi * MFFHi) - P * mffp - fd (Eq. L-6)
1
Where:
Ef = Total mass of fluorine emitted from process i over the period p (metric tons).
Rd = Total mass of the fluorine-containing reactant d that is fed into process i over the
period p (metric tons).
P = Total mass of the fluorine-containing product produced by process i over the period p
(metric tons).
MFFRd = Mass fraction of fluorine in reactant d, calculated in Equation L-14.
MFFp = Mass fraction of fluorine in the product, calculated in Equation L-15.
Fd = Total mass of fluorine in destroyed or recaptured streams from process i containing
fluorine-containing reactants, products, and by-products, calculated in Equation L-7.
v = Number of fluorine-containing reactants fed into process i.
The mass of total fluorine in destroyed or recaptured streams containing fluorine-
containing reactants, products, and by-products would be estimated at least monthly using
Equation L-7. In Equation L-7, the concentration of individual fluorinated GHGs present in each
destroyed or recaptured stream (i.e., speciated approach) is measured and determined, and the
amount of fluorine present is then calculated. (An alternative approach to determining the total
fluorine in destroyed or recaptured streams, as shown using Equations L-17 and L-18, is
discussed below in section 3.B.ii.3.)
fd=Zp,*mfff+1l
j-i
k-1
'( 1 \
IX+IX,
Vi-1 '-1
* MFF,
Bk
v f q
+ Z IX
d-\ \]-\
\
*MFF,
Rd
(Eq. L-7)
Where:
Fd = Total mass of fluorine in destroyed or recaptured streams from process i
containing fluorine-containing reactants, products, and by-products over the
period p.
Pj = Mass of the fluorine-containing product removed from process i in stream j and
destroyed over the period p (calculated in Equation L-8 or L-9).
Bkj = Mass of fluorine-containing by-product k removed from process i in stream j
and destroyed over the period p (calculated in Equation L-8 or L-9).
16
-------
By = Mass of fluorine-containing by-product k removed from process i in stream 1
and recaptured over the period p.
Rdj = Mass of fluorine-containing reactant d removed from process i in stream j and
destroyed over the period p (calculated in Equation L-8 or L-9).
MFFRd = Mass fraction of fluorine in reactant d, calculated in Equation L-14.
MFFp = Mass fraction of fluorine in the product, calculated in Equation L-15.
MFFBk = Mass fraction of fluorine in by-product k, calculated in Equation L-16.
q = Number of streams destroyed in process i.
x = Number of streams recaptured in process i.
u = Number of fluorine-containing by-products generated in process i.
v = Number of fluorine-containing reactants fed into process i.
The mass of each fluorinated GHG removed from process i in stream j and destroyed
over the period p (i.e., Pj, By, or Rdj, as applicable) would be estimated by applying the
destruction efficiency of the device that has been demonstrated for the fluorinated GHG f to
fluorinated GHG f using Equation L-8:
MFGHGfj = DEFGHGf * CFGHGfj * SJ L"8)
Where:
MFGHGfj = Mass of fluorinated GHG f removed from process i in stream j and destroyed
over the period p. (This may be Pj, By, or Rdj, as applicable.)
DEpGHGf = Destruction efficiency of the device that has been demonstrated for fluorinated
GHG f in stream j (fraction).
CFGHGfj = Concentration (mass fraction) of fluorinated GHG f in stream j removed from
process i and fed into the destruction device over the period p. If this
concentration is only a trace concentration, Cf-ghgij is equal to zero.
Sj = Mass removed in stream j from process i and fed into the destruction device over the
period p (metric tons).
The mass of each fluorine-containing compound that is not a fluorinated GHG and that is
removed from process i in stream j and destroyed over the period p (i.e., Pj, By, or Rdj, as
applicable) would be estimated using Equation L-9:
MFcgj=cFcg]*S1 (Eq. L-9)
Where:
Mpcgj = Mass of non-fluorinated GHG fluorine-containing compound g removed from
process i in stream j and destroyed over the period p. (This may be Pj, By, or
Rdj, as applicable).
CFCgj = Concentration (mass fraction) of non-fluorinated GHG fluorine-containing
compound g in stream j removed from process i and fed into the destruction
device over the period p. If this concentration is only a trace concentration,
CFCgj is equal to zero.
17
-------
Sj = Mass removed in stream j from process i and fed into the destruction device over the
period p (metric tons).
The mass of fluorine-containing by-product k removed from process i in stream 1 and
recaptured over the period p would be estimated using Equation L-10:
Ba=cm*Sl (Eq. L-10)
Where:
Bki = Mass of fluorine-containing by-product k removed from process i in stream 1 and
recaptured over the period p (metric tons).
CBki = Concentration (mass fraction) of fluorine-containing by-product k in stream 1
removed from process i and recaptured over the period p. If this concentration is only
a trace concentration, cbh is equal to zero.
Si = Mass removed in stream 1 from process i and recaptured over the period p (metric
tons).
Attributing Fluorine Emissions to Reactants, Products, and By-Products (Characterizing
Emissions)
To estimate the terms FERd, FEP, and FEBk for Equations L-l 1, L-12, and L-13,
facilities could assume that the total mass of fluorine emitted, EF, estimated in Equation L-6,
occurs in the form of the fluorinated GHG that has the highest GWP among the fluorinated
GHGs that occur in more than trace concentrations in the process unless facilities possess
emission characterization measurements showing otherwise. The sum of the terms FERd, FEP,
and FEBk would equal 1. Facilities would document the data and calculations that are used to
speciate individual compounds and to estimate FERd, FEP, and FEBk. Facilities should exclude
from calculations the fluorine included in FD, because this quantity has already been subtracted
from emissions in Equation L-6. For example, they should exclude fluorine-containing
compounds that are not fluorinated GHGs and that result from the destruction of fluorinated
GHGs by any destruction devices (e.g., the mass of HF created by combustion of an HFC).
However, they should include emissions of fluorinated GHGs that survive the destruction
process.
It may be appropriate scale emission characterization requirements to the size of the
emissions from the process and process vent. For processes and process vents that have total
emissions greater than a certain threshold, facilities would conduct measurements to determine
what fraction of the emissions are each individual fluorinated GHG (FERd, FEP, and FEBk). For
processes and process vents with emissions below the threshold, facilities could conduct
measurements or rely on previous measurements.
Facilities could estimate the total process emissions by relying on the estimate of total
process emissions developed for the error estimate. To estimate emissions from process vents,
facilities could conduct emission calculations or engineering assessments. For example, if the
facility calculated that a process emitted 25,000 metric tons C02e or more, the facility would
then estimate the emissions from each process vent, considering controls, using engineering
18
-------
calculations or engineering assessments. The facility would characterize by measurement any
process vent that emits 25,000 metric tons C02e or more and determine the fraction of each
fluorinated GHG.
For other vents, including vents from processes that emitted less than 25,000 metric tons
CC>2e, facilities would characterize emissions either by taking measurements or by using
previous measurements on the process, bench-scale, or pilot-scale process to determine the
fraction of each fluorinated GHG.
If the emissions in either of the above cases were controlled and if the facility speciated
the destroyed streams to estimate FD, the facility could use these monthly (or more frequent)
measurements, along with the appropriate DEs of the destruction device, to characterize
emissions after the destruction device. For vents that were uncontrolled or for which the facility
quantified total fluorine, the facility would use previous or new measurements, as appropriate.
If there was only one process vent associated with the process, the measurement would
require only the fraction of each fluorinated GHG. When there is more than one process vent,
the facility would have to include a flow rate for each process vent and develop a weighted
average across the process for the fraction of each fluorinated GHG. The flow rates could be
measured or estimated. For fluorine emissions that are not accounted for by vent estimates (i.e.,
fugitive emissions), facilities would apply the weighted average of the emission characterization
from process vents to the emissions.
The total mass of fluorine-containing reactant d emitted would be estimated at least
monthly based on the total fluorine emitted and the fraction that consists of fluorine-containing
reactants using Equation L-l 1. If the fluorine-containing reactant d is a non-GHG, facilities
assume that FERa is zero.
E
'R—ip
FERd*EF
_ Rd *MFFRd + FEP *MFFP + £ FEBk *MFFBk ^
Kd=1 k=1
(Eq. L-l 1)
J
Where:
ER-ip -
FERd =
FEP
FEBk
MFFRd
MFFp
MFFek
Total mass of fluorine-containing reactant d that is emitted from process i over the
period p (metric tons).
The fraction of the mass emitted that consists of the fluorine-containing reactant
d.
Total mass of fluorine emissions from process i over the period p (metric tons),
calculated in Equation L-6.
The fraction of the mass emitted that consists of the fluorine-containing product.
The fraction of the mass emitted that consists of fluorine-containing by-product k.
Mass fraction of fluorine in reactant d, calculated in Equation L-14.
Mass fraction of fluorine in the product, calculated in Equation L-l 5.
Mass fraction of fluorine in by-product k, calculation in Equation L-l6.
19
-------
u = Number of fluorine-containing by-products generated in process i.
v = Number of fluorine-containing reactants fed into process i.
The total mass of fluorine-containing product emitted would be estimated at least
monthly based on the total fluorine emitted and the fraction that consists of fluorine-containing
products using Equation L-12. If the fluorine-containing product is a non-GHG, facilities may
assume that FEP is zero.
FFP * F
EP-, = T~v " ; " (Eq- L"12)
Y^FER, *MFFRd + FEP *MFFP + ^FEBk *MFFBk
\d=1 k=1
Where:
Ep.ip = Total mass of fluorine-containing product emitted from process i over the period
p (metric tons).
FEP = The fraction of the mass emitted that consists of the fluorine-containing product.
EF = Total mass of fluorine emissions from process i over the period p (metric tons),
calculated in Equation L-6.
FERd = The fraction of the mass emitted that consists of fluorine-containing reactant d.
FEBk = The fraction of the mass emitted that consists of fluorine-containing by-product
k.
MFFRd = Mass fraction of fluorine in reactant d, calculated in Equation L-14.
MFFp = Mass fraction of fluorine in the product, calculated in Equation L-15.
MFFBk = Mass fraction of fluorine in by-product k, calculation in Equation L-16.
u = Number of fluorine-containing by-products generated in process i.
v = Number of fluorine-containing reactants fed into process i.
The total mass of fluorine-containing by-product k emitted would be estimated at least
monthly based on the total fluorine emitted and the fraction that consists of fluorine-containing
by-products using Equation L-13. If fluorine-containing by-product k is a non-GHG, facilities
may assume that FEBk is zero.
F FEBk*EF (Eq. L-13)
Bk-ip / v „
Y^FER, *MFFm + FEP *MFFP +^FEBk *MFFBk
V 1 k= 1
Where:
EBk-ip = Total mass of fluorine-containing by-product k emitted from process i over the
period p (metric tons).
FEBk = The fraction of the mass emitted that consists of fluorine-containing by-product
k.
FERd = The fraction of the mass emitted that consists of fluorine-containing reactant d.
FEP = The fraction of the mass emitted that consists of the fluorine-containing product.
Ef = Total mass of fluorine emissions from process i over the period p (metric tons),
calculated in Equation L-6.
20
-------
MFFRd
MFFp
MFFek
u
v
Mass fraction of fluorine in reactant d, calculated in Equation L-14.
Mass fraction of fluorine in the product, calculated in Equation L-15.
Mass fraction of fluorine in by-product k, calculation in Equation L-16.
= Number of fluorine-containing by-products generated in process i.
= Number of fluorine-containing reactants fed into process i.
The mass fraction of fluorine in reactant d would be estimated using Equation L-14:
AW „
MFFRd =MFRd *
MW,
(Eq. L-14)
Rd
Where:
MFFRd
MFRd
AWf
MWRd
= Mass fraction of fluorine in reactant d (fraction).
= Moles fluorine per mole of reactant d.
= Atomic weight of fluorine.
= Molecular weight of reactant d.
The mass fraction of fluorine in the product would be estimated using Equation L-15:
(Eq. L-15)
* AWf
MI = MI'',, * £-
MW„
Where:
MFFp
MFP
AWf
MWP
= Mass fraction of fluorine in the product (fraction).
= Moles fluorine per mole of product.
= Atomic weight of fluorine.
= Molecular weight of the product produced.
The mass fraction of fluorine in by-product k would be estimated using Equation L-16:
AWV
MFFBk=MFBk*
MW,
(Eq. L-16)
Bk
Where:
MFFek
MFek
AWf
MWBk
= Mass fraction of fluorine in the product (fraction).
= Moles fluorine per mole of by-product k.
= Atomic weight of fluorine.
= Molecular weight of by-product k.
3. Mass Balance Equations - Alternative Approach for Determining Total
Fluorine in Destroyed and Recaptured Streams.
As an alternative to the approach in Equation L-7 for determining the mass of total
fluorine destroyed or recaptured, facilities could use the approach in Equation L-17. In Equation
L-17, rather than determine the individual fluorinated GHGs present in each destroyed or
recaptured stream (i.e., speciated approach) and then determine the amount of fluorine present
21
-------
using Equation L-7, facilities may use a measurement technique on destroyed or recaptured
streams that provides the total mass of fluorine (i.e., no speciation of individual fluorinated
GHGs). This approach may be particularly useful for process with multiple byproducts.
The total mass of fluorine in destroyed or recaptured streams containing fluorine-
containing compounds (including all fluorine-containing reactants, products, and by-products)
would be estimated at least monthly using Equation L-17.
Fn = t.DE™ * cm * S, + * s, (Eq L-17)
j=1 1=1
Where:
Fd = Total mass of fluorine in destroyed or recaptured streams from process i
containing fluorine-containing reactants, products, and by-products.
DEavgj = Weighted average destruction efficiency of the destruction device for the
fluorine-containing compounds identified in destroyed stream j under
§98.124(b)(4)(ii) and (5)(ii) (calculated in Equation L-18)(fraction).
CxFj = Concentration (mass fraction) of total fluorine in stream j removed from process
i and fed into the destruction device over the period p. If this concentration is
only a trace concentration, CxFj is equal to zero.
Sj = Mass removed in stream j from process i and fed into the destruction device
over the period p (metric tons).
Ctfi = Concentration (mass fraction) of total fluorine in stream 1 removed from process
i and recaptured over the period p. If this concentration is only a trace
concentration, Cbm is equal to zero.
Si = Mass removed in stream 1 from process i and recaptured over the period p.
q = Number of streams destroyed in process i.
x = Number of streams recaptured in process i.
The weighted average destruction efficiency that is applicable to a destroyed stream and
that is used in Equation L-17 would be calculated using Equation L-18. Equation L-18 treats
non-GHG fluorine-containing compounds as being completely destroyed, since none of the
fluorine in them is emitted as a fluorinated GHG. Fluorinated GHGs are assumed to be
destroyed to the DE of the device for each fluorinated GHG.
w y
T.DEpohoi *cpohoi *Ss *MFFpoaof +-£cFCt *S, *MFFS
DE„t = ^ (Eq. L-18)
Zc *S *MFF +Vc *S *MFF
FGHGf °j 1V1± 1 FOHOf ^ C FCg °j 1V1± 1 g
f=1 2=1
Where:
DEavgj = Weighted average destruction efficiency of the destruction device for the
fluorine-containing compounds identified in destroyed stream j under
§98.124(b)(4)(ii) or (b)(5)(ii), as appropriate.
22
-------
DEpGHGf = Destruction efficiency of the device that has been demonstrated for fluorinated
GHG f in stream j (fraction).
CFGHGfj = Concentration (mass fraction) of fluorinated GHG f in stream j removed from
process i and fed into the destruction device over the period p. If this
concentration is only a trace concentration, CF-GHGfj is equal to zero.
CFCgj = Concentration (mass fraction) of non-fluorinated GHG fluorine-containing
compound g in stream j removed from process i and fed into the destruction
device over the period p. If this concentration is only a trace concentration,
CFCgj is equal to zero.
Sj = Mass removed in stream j from process i and fed into the destruction device
over the period p (metric tons).
MFFFGHGf = Mass fraction of fluorine in fluorinated GHG f, calculated in Equation L-14, L-
15, orL-16, as appropriate.
MFFpcg = Mass fraction of fluorine in non-fluorinated-GHG fluorinated compound g,
calculated in Equation L-14, L-15, or L-16, as appropriate,
w = Number of fluorinated GHGs in destroyed stream j.
y = Number of non-fluorinated-GHG fluorine-containing compounds in destroyed
stream j.
4. Frequency of Measurement and Calculation.
EPA considered several measurement and calculation frequencies for the mass balance
approach. The total mass of each fluorinated GHG emitted from production processes could be
estimated on a daily, weekly, or monthly basis by calculating the total fluorine in the reactants,
subtracting the measured total fluorine in the product, and subtracting the measured total fluorine
that is destroyed or recaptured. The daily measurement requires additional man hours to conduct
more frequent measurements; however, in situations where input or output flows are estimated
by multiplying concentrations by total mass flow rates of streams, and where one or both of these
vary, frequent measurements will lead to more accurate and precise estimates than less frequent.
A number of fluorocarbon producers have noted that daily measurements are burdensome
and lead to large errors in the estimates of daily emissions. Many streams may contain acidic
and reactive constituents such as HF, and sampling from these streams can create safety hazards.
Daily product measurements can also vary significantly (sometimes exceeding 100 percent) for
three reasons. First, when continuous processes are initially started, there is a lag time between
the time the reactants are fed into the process and the time products emerge. Second, even after
the process has been running for a while, the quantity of material in the process can vary based
on changes in production rates and other conditions. Third, the relatively large errors in
measurements of in-process product holding tanks (e.g., based on sight-glass readings) have a
significant impact on daily mass balances. Over time, all of these effects smooth out, making
longer term mass balances more reliable than daily mass balances.11 In view of these
considerations, a requirement to perform measurements daily may not be justified, particularly
for processes where concentrations show little variability.
11 Note, however, that the sum of the daily mass-balances would also smooth out.
23
-------
Because different processes operate with different ranges of process variability, it would
be reasonable to require more frequent measurements for more variable processes. The
appropriate frequency of measurement could be determined using the error calculation described
above along with standard statistical techniques such as the Student's t test.
As discussed below in the context of emission testing, the more variable a parameter is,
the more samples must be taken to precisely characterize the mean of that parameter. The
number of samples required to estimate the mean of the parameter with a given level of
confidence can be calculated based on the relative standard deviation (RSD) of the samples and
on some assumptions about the distribution of the parameter. Where the set of samples is
relatively small and only the sample standard deviation "s" is known (rather than the true
standard deviation of the parameter), the appropriate statistic is often the Student's t test.
Monthly sampling will generate twelve concentration measurements per stream per year.
If the variability among the concentration measurements is high enough that these 12
measurements per year (i.e., monthly measurements) result in relative and absolute errors above
the 30-percent and 3,000-mtC02e limits, based on the Student's t test, then the facility would
need to increase the frequency of its measurements to meet the error limits. For example,
facilities may find that weekly measurements are necessary to meet the error criteria for certain
processes. By including process variability in the error calculation, facilities could make more
frequent measurements if this were necessary to address process variability, but they would not
need to incur the costs and risks of sampling for processes where less frequent measurements
yielded precise results.
Facilities that use the alternative to the error calculation (discussed further below) could
make weekly measurements and calculations. EPA calculates that at a weekly frequency, these
measurements will lead to reasonably accurate emission estimates (with an error near 30 percent)
even if the concentrations in the process are highly variable (e.g., even if the RSD of the
concentration measurements is 50 percent, which would be unusually high.)
There are two other measurement practices that could address process variability. First,
facilities could ensure that they made concentrations measurements that reflect the full range of
conditions within the process, e.g., catalyst age. Facilities could also account for emissions that
occur during process startups, shutdowns, and malfunctions, either recording fluorinated GHG
emissions during these events or documenting that these events do not result in significant
emissions. Together, these practices would limit the impact of sampling bias on emissions
estimates.
5. Alternative Precision and Accuracy Requirements
An alternative to requiring facilities to perform the error calculation would be to specify
conditions under which the error of the estimate would be expected to fall under the 3,000
mtC02e error limit. (The 3,000 mtC02e absolute error was identified as the limit because it is
consistent with the absolute error limit in the error calculation approach.) These conditions
would include specific accuracy, precision, and frequency requirements for measurements and a
process throughput limit.
24
-------
To determine the appropriate requirements, one must consider the five factors that affect
the absolute accuracy and precision of mass-balance estimates: (1) the relative accuracy and
precision of the mass measurements, (2) the relative accuracy and precision of the methods used
to measure concentrations (irrespective of process variability), (3) process variability, (4) the
frequency of the measurements, and (5) the quantity of the fluorinated GHG throughput of the
process, that is, the total masses of the fluorinated GHG reactants, products, and by-products fed
into and generated by the process. A number of different combinations of measurement
precisions, accuracies and frequencies would result in an emissions estimate with an acceptable
absolute error; but some combinations are likely to be more easily achieved than others. Based
on the research performed for this rule, one robust combination of requirements that is likely to
be achievable for a number of processes is (1) to measure the masses identified in the rule with
an accuracy and precision of ±0.2 percent of full scale or better, (2) to measure the
concentrations identified in the rule using analytical methods with an accuracy and precision of
±10 percent or better, (3) to conduct these measurements at least weekly, and (4) to limit the
fluorinated GHG throughput of the process (including fluorinated GHG reactants, products, and
by-products) to 500,000 mtC02e or less. At an emission rate of two percent, the 500,000-
mtC02e throughput limit would result in emissions of 10,000 mtCC^e. In combination with the
precision, accuracy, and frequency requirements for measurements, this throughput would be
expected to result in a maximum absolute error of 3,000 mtCChe.
Although some of the requirements could be relaxed and still result in an error near or
below 3,000 mtC02e, this would require a tightening of the other requirements. For example, if
the requirements for the precision and accuracy of the mass measurements were relaxed to ±0.4
percent, the requirements for the precision and accuracy of the concentration measurements
would have to be tightened, e.g., to ±5 percent, and the frequency of these measurements would
have to be increased. Comments received on the April 10, 2009 and April 12, 2010 proposed
rules indicate that achieving precisions and accuracies of ±5 percent for concentration
measurements, and conducting these measurements more often than weekly, could be quite
challenging. The precision and accuracy requirements for concentrations and for the frequency
of measurement are less stringent than those initially proposed in the April 10, 2009 rule; those
for masses are the same as those initially proposed.
Facilities that could not or chose not to meet these requirements would remain free to use
the error calculation to demonstrate compliance with the mass-balance error limits. Under the
error calculation, facilities have the flexibility to focus on improving the accuracy and precision
of those measurements that have a significant impact on the overall error of the estimate rather
than expending resources to improve the accuracy and precision of measurements that are not as
important to the accuracy and precision of the emissions estimate.
6. Comparison Between Proposed and Final Mass-Balance Approaches
The version of the mass-balance approach adopted in the final rule is based on the same
basic principle as the version that was proposed, but it also differs from the proposed version in
some key respects. Both the proposed and final mass-balance approaches use the difference
between the fluorine (or other element) fed into the process and the fluorine removed from the
process to estimate the mass emitted. However, they differ in how they characterize the mass
emitted. The proposed version of the mass-balance approach required calculation of the quantity
25
-------
of by-product generated, which in turn required monitoring of any stream where by-products
occurred in more than trace concentrations. Emissions of the by-product were assumed to equal
the difference between the quantity of by-product generated and the quantity of by-product
destroyed. Any remaining emissions were assumed to consist of the product. The final version
of the mass-balance approach requires emission characterization measurements, which focus on
the actual emissions from the process.
The mass-balance approach in the final rule is fundamentally more flexible than that in
the proposed rule. First, it allows for the possibility that emissions can consist of reactants as
well as products and byproducts. Thus, the finalized version of the mass-balance approach is
capable of quantifying emissions from transformation processes in which fluorinated GHGs are
reactants as well as production processes in which fluorinated GHGs are products. Fluorinated
GHG producers indicated in their comments that they would prefer a version of the mass-balance
approach that was capable of quantifying emissions from transformation processes.
Second, because emissions are characterized at the vent, the finalized mass-balance
approach does not require facilities to speciate the flows in their streams. Instead, they are
required to track the total fluorine in destroyed or recaptured streams. (While they may speciate
these streams to do so, they may also use an analytical method that tracks total fluorine, such as
ASTM D7359-08.)
The facility must conduct emission characterization measurements, or use previous
measurements, to develop the break out among the fluorinated GHG emitted. This emission
measurement would focus on partitioning emissions among the various fluorinated GHGs (i.e.,
speciating the compounds).
7. Current Plant Practices for Mass Balance
A few facilities indicated that they conduct a form of the mass balance approach for
determining fluorinated GHG emissions. Facilities noted overall that the difficulty of the mass
balance approach depends on how sophisticated or complex the process is, i.e., how many input
and output streams there are, and whether there are by-products generated and how many.
Facilities that use a form of the mass balance approach indicated that it is often used in
conjunction with an emissions factor based approach, to track yields and subtle or incremental
changes in the process. These facilities conduct a "material balance" on the process, tracking the
product yield, using the emission factor for some emission points, and the composition and
concentration of the streams to determine if something in the process has changed. Generally,
facilities have a good understanding of the chemistry for all processes and understand what raw
materials are fed to the process, what side reactions occur, and what products and by-products
are created. Facilities have laboratory-scale, pilot-plant, and full-scale stream analysis results.
Facilities stated that with the original plant design, the mass balance equation is predicted.
Facilities noted in particular that at the bench-scale phase of product and process development,
they analyze every/all streams to identify all species and compounds. However, one facility
indicated that while they do regularly measure byproduct concentrations in the product as a
quality check, they do not regularly measure byproduct concentration across all process vents
and streams.
26
-------
Some facilities indicated that they conduct the mass balance on the fluorinated organic
component only.
Facilities noted, in particular, that they reconcile and calculate the production levels on a
monthly basis rather than on shorter time frames such as daily or weekly.
/'//. Monitoring of Process Vents
The emissions from process vents may be determined by a number of methods. These
include emissions testing, engineering assessments, or calculations based on chemical
engineering principles. All three methods may also be used to develop a site-specific emissions
factor based on the emissions estimate and the process activity observed during the test or
assumed for the calculations. These methods or approaches to determining emissions from
fluorinated gas production processes provide emissions estimates of varying accuracy and
precision.
1. Process-Vent-Specific Emission Factor Approach.
One method for estimating fluorinated GHG emissions is based on monitoring or
estimating the composition and flow rate of the process vent streams along with the production
activity parameter to develop a site-specific, process-vent-specific emission factor. Fluorinated
gas producers have indicated that they develop and apply emission factors to estimate emissions
from a range of processes. Typically a facility conducts an initial emissions test or engineering
calculation to develop an emission factor or emissions calculation factor and updates the
emissions factor as needed with process changes. Site-specific, process-vent-specific emission
factors can be developed based on either measurement of process vent emissions or calculation
of process vent emissions. The measurement of process vent emissions to develop an emissions
factor is addressed in this section, and the calculation of process vent emissions is discussed
below in Process-Vent-Specific Emission Calculation Factor approach.
Under this approach, facilities would develop a process-vent-specific emission factor for
each fluorinated GHG from each process vent by emissions testing. Facilities would use EPA,
ASTM, industry-developed, and other testing methods, including consensus standards and
industry-validated methods for determining the process emissions for emission factor
development. The methods used would be quality-assured methods, capable of detecting the
analyte of interest at the concentration of interest, and validated with the compound of interest at
the concentration of interest. There are several EPA test methods available to sample and
measure fluorinated GHG from processes in the fluorinated gas production source category,
including EPA Method 320 (ASTM D6348-03) and EPA Method 18. Method 1 or 1 A, Method
2, 2A, 2B, 2C, or 2D, 2F, or 2G, Method 3, 3A, or 3B, and Method 4 must be conducted along
with EPA Method 18.
Method 320 and ASTM D6348-03. For Method 320, a gas sample is extracted through a
heated gas transport and handling system and analyzed using FTIR spectroscopy to identify and
quantify the gas stream components. It is applicable only for vapor phase analysis of organic or
inorganic compounds. The sensitivity of this method can be adjusted based on the reference
standards used to configure the FTIR, but also depends on the gas composition, moisture content,
27
-------
and sampling losses. The ASTM D6348-03 FTIR method utilizes the same principles as Method
320 but has some improved calibration requirements.
Method 18. Method 18 utilizes gas chromatography (GC) and flame ionization,
photoionization, electron capture, mass spectroscopy, or another applicable detection method to
identify and quantify gaseous organic compounds. For the fluorinated GHG industry, electron
capture and mass spectroscopy and possibly photoionization are the most appropriate detectors
for identifying fluorinated GHG compounds. The identity and concentration levels are
determined by comparison to known standards, which are used to calibrate the GC. Stack gas
samples may be collected through direct interface sampling, dilution interface sampling,
adsorption tube sampling, or bag sampling. One needs to consider whether significant temporal
changes in flow rate and/or concentration occur in the process (e.g., batch processes) when
selecting and applying the sampling approach for this method. The direct interface, dilution
interface, and adsorption tube sampling are not designed for sources that vary in concentration.
Instead, an integrated bag sample with proportional sampling (sample rate adjusted
proportionally with changes in stack gas flow rate) would be appropriate for an emission source
with a varying flow rate and concentration. If the direct interface method is used, the timeframe
of temporal changes of concentration in relation to the analytical timeframe needs to be
considered. For example, if significant changes in concentration are expected to occur rapidly
(e.g., 10-minute time frame), but the GC analytical cycle is 30 minutes, a representative sample
analysis may not be obtained; on the other hand, for this situation, if the analytical cycle is 5
minutes, a representative sample would be obtained. Variations in vent gas flow rate also need
to be considered; for example, if vent flow rates and concentration are varying with time and the
direct interface method is being used, the vent flowrate should be continuously monitored so that
the flowrate and concentration can be correlated to determine mass emissions. For any of the
sampling methods, all samples must be analyzed within a few hours of collection.
Method 320, ASTM D63248-03 and Method 18 also require that the flow rate be
relatively constant during sampling. If the flow rate varies significantly over the course of the
sampling, then this fluctuation must be accounted for when measuring concentration and
calculating mass emissions.
OTM 24, the Tracer Gas Protocol For the Determination of Volumetric Flow Rate
Through the Ring Pipe of the Xact Multi-Metals Monitoring System (OTM 24) uses a GC
equipped with an FID to measure the concentration at a specific point. A tracer gas is introduced
into the process to perform a volumetric flow rate determination, which depends on the tracer
flow rate, concentration at the measurement point and calibration and analysis data.
The Emission Measurement Center Approved Alternative Method (ALT-012) is designed
for measuring particulate emissions in situations where gas velocities are low and difficult to
measure and subject to variations over time. ALT-012 can also be used to measure ducts and
stacks that are not subject to stratification. First, locations are identified upstream, for injecting
the tracer gas, and downstream, for the tracer gas measurement. For a complete measurement,
three ten-minute long sampling intervals must be completed at the downstream location.
Nuclear magnetic resonance (NMR) spectroscopy is a commonly used analytical
technique in organic chemistry. Organic compounds are placed in a magnetic field and exposed
28
-------
to electro-magnetic energy, usually in the form of radio frequency (RF), and the nuclei of certain
atoms absorb the radiation, which causes the nuclear spin to realign in the higher energy
direction. The wavelength the atom absorbs is dependent on the neighboring atoms in the
compound, and the resulting spectrum is called the NMR spectrum. This is a complex laboratory
analytical method used to determine the structure of molecules. Much like using infrared
spectroscopy (IR) to identify functional groups, analysis of a NMR spectrum provides
information on the number and type of chemical entities in a substance. Also, the dependence of
the transition energy on the position of a particular atom in a molecule makes NMR spectroscopy
extremely useful for determining the structure of molecules.
Using the applicable method(s), process activity (e.g., production or feed) would be
monitored at the same time as emissions; the emission factor would be developed by dividing the
emissions by the activity. Both continuous and batch processes are used in the fluorinated gas
production source category. Typically, for continuous processes, facilities would measure and
determine the fluorinated GHG emission rate based on test runs of at least 1 hour duration. For
continuous processes, facilities would use Equation L-19.
f = ^pv * MW * O *—!— *—!— *— CEa L-19^
ContPV jq6 1V1VV *1PV SJ/ IQl I ^ '
Where:
Econtpv = Mass of fluorinated GHG f emitted from process vent v from process i during the
emission test during test run r (kg/hr).
Cpv = Concentration of fluorinated GHG f during test run r of the emission test (ppmv).
MW = Molecular weight of fluorinated GHG f (g/g-mole).
Qpv = Flow rate of the process vent stream during test run r of the emission test (m3/min).
SV = Standard molar volume of gas (0.0240 m3/g-mole at 68°F and 1 atm).
1/103 = Conversion factor (1 kilogram/1,000 grams).
60/1 = Conversion factor (60 minutes/1 hour).
To develop the site-specific, process-vent-specific emission factor, facilities would track
the process activity during the emissions testing and combine it with the emissions. For
continuous processes, a facility would divide the fluorinated GHG emissions during the test by
the process activity during testing. Equation L-20 would be used to estimate the average
emission factor.
z
EFpy = ¦
E*
Activity
EmissionTest J
(Eq. L-20)
Where:
EFpy = Average emission factor for fluorinated GHG f emitted from process vent
v during process i (e.g., kg emitted/kg activity).
29
-------
EPV = Mass of fluorinated GHG f emitted from process vent v from process i
during the emission test during test run r (kg emitted/hr for continuous).
ActivityEmissionTest = Process feed, process production, or other process activity rate during the
emission test during test run r (e.g., kg product/hr).
r = Number of test runs performed during the emission test.
Development of the emissions testing program requires testing during representative
process operation. To develop a representative emission factor for the process vent, the facility
should give some consideration to what process variability might be present in the process. For
each operating scenario of the process, the emissions test data must be representative of the
typical process operation while also including process variability.12 The full range of process
operating conditions should be included. The typical process operation conditions would include
yield, temperature, pressure, and flow rates. The operating conditions and the process activity
should be measured during the emissions test. See section 3.B.iii.3 below for more information
on representativeness.
Emission factors may be developed on an uncontrolled or a controlled basis depending on
the emission level of the process vent. Using the process-vent-specific emissions factor method,
a facility would estimate emissions using the emission factor and would track process activity on
an on-going basis. To estimate annual fluorinated GHG emissions from each vent, facilities
would multiply each uncontrolled emission factor by the appropriate activity data and account
for the use (and uptime) of destruction devices. Where controlled emission factors are
developed, a facility would multiply each controlled emission factor by the appropriate activity
data for periods when the process vent is vented to the destruction device. For periods when the
process vent is not venting the destruction device (i.e., bypassing), the facility would develop an
emission factor for these periods as well (i.e., an uncontrolled emission factor) or would develop
an emission calculation factor.
For facilities that conduct emissions testing on a controlled vent, the emissions of each
fluorinated GHG for each process vent would be calculated using Equation L-21.
Epv = EFPV C * Activityc + ECFPV U * Activityv (Eq. L-21)
Where:
EpV = Mass of fluorinated GHG f emitted from process vent v from process i,
for the year (kg).
EFpy-c = Average emission factor for fluorinated GHG f emitted from process
vent v during process i based on testing after the destruction device (kg
emitted/activity)(e.g., kg emitted/kg product).
Activityc = Total process feed, process production, or other process activity during
the year for which emissions are vented to the properly functioning
destruction device (i.e., controlled).
12 Operating scenarios and other issues associated with developing representative emission factors are discussed
further below under "Obtaining and Maintaining Representative Emission Factors."
30
-------
ECFpv-u = Emission calculation factor for fluorinated GHG f emitted from process
vent v during process i during periods when the process vent is not
vented to the properly functioning destruction device (kg
emitted/activity)(e.g., kg emitted/kg product).
Activityu = Total process feed, process production, or other process activity during
the year for which the process vent is not vented to the properly
functioning destruction device (e.g., kg product)
For facilities that conduct emissions testing on an uncontrolled vent, the emissions would
be calculated by applying the destruction efficiencies of the device that have been demonstrated
for the fluorinated GHGs in the vent stream to the fluorinated GHG emissions for the process
vent, using Equation L-22 of this section. Facilities would apply the destruction efficiency only
to the portion of the process activity during which emissions are vented to the properly
functioning destruction device (i.e., controlled).
Epv = EFPV_V * {Activityj + Activityc * (l - DE)) (Eq. L- 22)
Where:
Epv
EFpv-u
Activityu
Activityc
DE
Mass of fluorinated GHG f emitted from process vent v from process i,
for the year, considering destruction efficiency (kg).
Emission factor (uncontrolled) for fluorinated GHG f emitted from
process vent v during process i (kg emitted/kg product).
Total process feed, process production, or other process activity during
the year for which the process vent is not vented to the properly
functioning destruction device (e.g., kg product).
Total process feed, process production, or other process activity during
the year for which the process vent is vented to the properly functioning
destruction device (e.g., kg product).
Demonstrated destruction efficiency of the destruction device (weight
fraction).
When there is more than one operating scenario associated with a process, the emissions factor
developed for the first operating scenario of the process may be adjusted for other operating
scenarios depending of the magnitude of emissions. Facilities would use Equation L-23 to
develop an adjusted process-vent-specific emission factor.
EFPn„ = ^3- *EFfv (Eq. L-23)
£LL,rT
Where:
31
-------
EFpvadj = Adjusted process-vent-specific emission factor for an untested operating scenario.
ECFut = Emission calculation factor for the untested operating scenario developed under
paragraph (c)(4) of this section.
ECFt = Emission calculation for the tested operating scenario developed under paragraph
(c)(4) of this section.
EFpy = Process-vent-specific emission factor for the tested operating scenario.
Facilities may sum the emissions of each fluorinated GHG from all of the process vents
in a process for the year to estimate the total process vent emissions of each fluorinated GHG.
Facilities would use Equation L-24.
EPfi=ZEpv (Eq. L-24)
1
Where:
-pfi
ipv
Mass of fluorinated GHG f emitted from process vents for process i, for
the year (kg).
Mass of fluorinated GHG f emitted from process vent v from process i, for
the year, considering destruction efficiency (kg).
Number of process vents in process i.
2. Process-Vent-Specific Emission Calculation Factor Approach.
As mentioned above, another method for estimating fluorinated GHG emissions is based
on use of engineering calculations to determine the composition and flow rate of the process vent
streams. These emissions estimates are divided by the production activity parameter to develop
a site-specific, process-vent-specific emission calculation factor. A variation of this method
includes use of engineering assessments to develop process-vent-specific emission calculation
factor. Facilities have indicated that they regularly use engineering calculations and engineering
assessments to estimate fluorinated GHG emissions.
Engineering calculations use chemical engineering principles and component property
data to calculate emissions. Acceptable calculation methods for uncontrolled and controlled
process vent emissions are included in a number of rule texts and EPA guidance documents.
Other promulgated rule texts that contain appropriate calculation equations include the
Miscellaneous Organic NESHAP (MON) NESHAP in §63.2460(b)(l) through (4) (the MON
rule text also refers to equations in the Pharmaceutical NESHAP in §63.1257(d)(2)(i) for
uncontrolled and §63.1257(d)(3)(i) for controlled emissions with some noted caveats to the
calculations). These rule texts include process calculation methods and equations for vapor
displacement, purging, heating, depressurization, vacuum systems, gas evolution, air drying, and
32
-------
empty vessel purging. An even more recent guidance document that provides appropriate
calculations for chemical manufacturing operations is the Methods for Estimating Air Emissions
from Chemical Manufacturing Emissions, Volume II: Chapter 16, of the Emission Inventory
Improvement Program.
Engineering assessments may be conducted using previous emissions test data or other
information available on the process. Facilities typically have component and composition data
available from the original research and development activities on the process and from the
process' lifetime of operation. If the facility performed a scoping speciation as described in
section 3. A.ii above, the results of that test should also be considered. For process vent
emissions determinations, facilities could conduct an engineering assessment to calculate
uncontrolled emissions for each process vent and for each emission episode. Various rules for
other regulatory programs include use of engineering assessments to estimate emissions from
processes and process vents. Other promulgated rules that allow use of engineering assessments
include NESHAP rules in 40 CFR part 63, such as subpart GGG for Pharmaceutical
Manufacturing in §63.1257(d)(2)(ii) and subpart FFFF for MON in §63.2460(b). Engineering
assessments include use of previous emissions test data where the emissions are representative of
current operating practices; bench-scale or pilot-scale streams analysis results for identifying
stream components and composition for products, byproducts, and wastes; design analysis based
on chemical engineering principles, measurable process parameters, or physical or chemical laws
or properties. These previous emissions data, analysis, and calculations may be used to identify
fluorinated GHG compounds and their concentrations in various streams.
To develop the site-specific, process-vent-specific emissions calculation factor, facilities
would first calculate the fluorinated GHG emissions and divide them by the appropriate process
activity associated with the emissions calculation. Equation L-25 would be used to estimate the
emission calculation factor.
ECFPV =
E.
Activity
(Eq. L-25)
Re presentative
Where:
ECFpy = Emission calculation factor for fluorinated GHG f emitted from process
vent v during process i (e.g., kg emitted/kg product).
EpV = Average mass of fluorinated GHG f emitted, based on calculations, from
process vent v from process i during the period or batch for which
emissions were calculated, for either continuous or batch (kg emitted/hr
for continuous, kg emitted/batch for batch).
ActivityRepresentative = Process feed, process production, or other process activity rate
corresponding to average mass of emissions based on calculations (e.g., kg
product/hr for continuous, kg product/batch for batch).
As with the process-vent-specific emissions factor, a facility would estimate emissions
using the emissions calculation factor and would track process activity on an on-going basis. To
estimate annual fluorinated GHG emissions from each vent, facilities would multiply each
emission calculation factor by the appropriate activity data and account for the use (uptime, by-
33
-------
pass, and/or downtime) of destruction devices. Equation L-26 would be used to estimate
emissions for each individual process vent if the process vent is not vented to a destruction
device, and Equation L-27 would be used if the process vent is vented to a destruction device.
Epv = ECFPV * Activity (Eq. L-26)
Where:
Epv = Mass of fluorinated GHG f emitted from process vent v from process i, for
the year (kg).
ECFpy = Emission calculation factor for fluorinated GHG f emitted from process
vent v during process i (kg emitted/activity) (e.g., kg emitted/kg product).
Activity = Process feed, process production, or other process activity during the year.
Epy = ECFPV * (Activity v + Activityc * (l - DE)) (Eq. L-27)
Where:
Epv
ECFpy
Activityu
Activityc
DE
Mass of fluorinated GHG f emitted from process vent v from process i, for
the year considering destruction efficiency (kg).
Emission calculation factor for fluorinated GHG f emitted from process
vent v during process i (e.g., kg emitted/kg product).
Total process feed, process production, or other process activity during the
year for which the process vent is not vented to the properly functioning
destruction device (e.g., kg product).
Total process feed, process production, or other process activity during the
year for which the process vent is vented to the properly functioning
destruction device (e.g., kg product).
Demonstrated destruction efficiency of the destruction device (weight
fraction).
The emissions of each fluorinated GHG from all process vents in the process would be
summed to obtain the total emissions from process vents for the facility as a whole. Equation L-
28 would be used to sum fluorinated GHG emissions over all process vents for the process, and
Equation L-29 would sum the emissions from all processes. Other emission points at the facility
such as equipment leaks, storage tanks, or wastewater would have to be accounted for separately
and added in to the facility total.
EPfi=f,EPV (Eq.L-28)
1
Where:
34
-------
V
Epfi
Epv
Mass of fluorinated GHG f emitted from process vents for process i, for
the year (kg).
Mass of fluorinated GHG f emitted from process vent v from process i, for
the year, considering destruction efficiency (kg).
Number of process vents in process i.
3. Obtaining and Maintaining Representative Emission Factors
Process conditions
Emission rates from a process depend on the operating conditions of that process. Thus,
to ensure that emission factors are representative, facilities should define the operating scenario
that encompasses the typical range of operating conditions for the process. If the process has
multiple operating scenarios, the facility should develop a representative emissions factor for
each operating scenario. To define the process operating scenario, a facility could include
information on the process description, the specific process equipment used, and the range of
operating conditions; the process vents, emission episodes and durations, and the quantity of
uncontrolled fluorinated GHG emissions; the control device or destruction device used to control
emissions; and the manifolding of process vents within the process and from other processes.
Other operating scenarios should also be defined for differences in operating conditions that
affect emissions. Examples of situations where process differences may warrant separate
operating scenarios include the following: making small volumes of a product in one set of
batch process equipment part of the year and making larger volumes in larger batch process
equipment part of the year; use of two different types of catalyst in the same process; deliberate
alterations in process conditions such as temperature or pressure to shift the reaction to a
particular product; and making small volumes of a product in a batch process part of the year and
making large volumes in a continuous process part of the year. A facility would need to develop
a representative emissions factor for each process operating scenario because each operating
scenario for a process will result in different emissions levels.
Prior to developing an emissions factor, facilities should observe the operating conditions
of the process over a period of time and note how changes to the operating conditions affect
emissions rates. The time periods may vary from facility to facility and perhaps from process to
process, but typically may include an examination period of one year or one month and could
include examination of startup and shutdown periods. An examination of the previous year of
process operating conditions might take note of whether there is month-to-month or seasonal
variability in the process. For example, seasonal variability might be caused by production rate
changes or might be caused by changes in the ambient temperature from summer to winter. An
examination of one month of process operating conditions might reveal some short-term process
variability. A review of startup and shutdown process operating conditions may include an
assessment of how frequently startups and shutdowns occur, what effect startups and shutdowns
have on emission rates, and whether or not the impact on emissions can be quantified.
In general, emissions testing during process startups and shutdowns would not be
expected to lead to representative emission factors, because emission rates tend to fluctuate
during such events. Exceptions to this could include long-term monitoring that would not over-
35
-------
represent startup or shutdown conditions in the resulting emission factor, and monitoring
specifically to obtain emission factors for startups and shutdowns conditions. Several companies
indicated that they have analyzed the emissions profile during startup events and during
shutdown events. They found that the emission rates during these events departed from those at
steady state conditions, but that emissions profiles were consistent between one startup event and
another.
Number of samples
Emission samples should be taken over the full range of process conditions within a
given operating scenario. To assess the variability of emission rates within the operating
scenario, a minimum of three emission test runs is appropriate. If the emission factors based on
the three test runs show relatively little variability, i.e., if their relative standard deviation (RSD)
is low, then the three sample emission factors can probably be averaged to obtain a robust,
representative emission factor for the process. However, if the emission factors based on the
three test runs show high variability, that is, a high RSD, then the average of the three emission
factors may be a poor estimator of the actual mean emission rate from the process. In this case,
additional samples should be taken to develop a robust emission factor.13 The equation to
estimate the RSD is shown in Equation RSD-1.
RSD = =
EF
(Eq. RSD-1)
Where:
RSD =
s =
EFbar =
Relative standard deviation, fraction
sample standard deviation, calculated in Eq. RSD-2
average emission factor for all test runs combined
Equation RSD-2 is used to estimate the sample standard deviation, s.
5 =
J(£F,-EFJ4
ef2 - efJ + (/¦;/-; - efJ + ...
n-1
(Eq. RSD-2)
Where:
s
EF;
EFbar
sample standard deviation
the individual emission factor value from test run 1, 2, 3, etc.
average emissions factor across all test runs (sample mean emission factor),
calculated in Equation RSD-3
13 The fact that emission rates from a process show high variability does not mean that a representative emission
factor cannot be developed; it simply means that more samples must be taken to do so. Note, however, that if the
variability of the process can be attributed to the variability of a controllable and measurable process parameter, it is
probably appropriate to develop one or more additional operating scenarios, each with its own emission factor, to
capture this variability.
36
-------
n = number of test runs
Eq. RSD-3 (which corresponds to equation L-20 in the regulatory text) is used to estimate
the mean emissions factor, EFbar.
~EF = EF' +EF2 +EFj+... (Eq. RSD-3)
n
Where:
EFbar = average emissions factor for all test runs combined (sample mean emission factor)
EF; = the individual emission factor value from test run 1, 2, 3, etc.
n = number of test runs
In general, more samples are required to characterize the mean of a population or
parameter that has high variability than are required for a population or parameter that has low
variability. The number of samples required to estimate the mean of the parameter with a given
level of confidence can be calculated based on the RSD of the samples and on some assumptions
about the distribution of the parameter. Where the set of samples is relatively small and only the
sample standard deviation "s" is known (rather than the true standard deviation of the
parameter), the appropriate statistic is often the Student's t test. The Student's t test presumes an
approximately normal, bell-shaped distribution of the underlying parameter, which implies that
the distribution of the average is the Student's t distribution.
For this situation, the Student's t test can be written:
A
^'statistic 9v'(/7 ^
yfn
< -\~t
statistic-95 % CI
(Eq. RSD-4)
Where:
EFbar = Average emission factor across all test runs (sample mean emission factor)
EFt = True mean emission factor of the process
s = Sample standard deviation
n = Number of samples
tstatistic-95%ci = Value of student's t statistic chosen to represent a 95-percent confidence
interval with n-1 degrees of freedom. For a sample size of 3, this value is
4.30.
Dividing through by the sample mean emission factor, EFbar, we obtain:
37
-------
^ statistic-95 % CI ^
(ef-eft)/ef
RSD
4n
^ statistic-95%CI 0^1 • RSD-5)
To choose a RSD above which we will require more than three samples, we first decide
how much error we are willing to tolerate in our emission factor. Suppose we want the emission
factor to have a 95 percent chance of being within 30 percent of the true mean emission rate of
the process. The requirement for the emission factor to be within 30 percent of the true mean
emission rate can be expressed as follows:
((EF -EFt)/Ef)< 0.3 (Eq. RSD-6)
Equation RSD-5 can also be rearranged as follows:
((ef-eft)/ef)<^B-*
yjn
Subtracting Equation RSD-7 from Equation RSD-6, we obtain:
0<0.3-^,
4n
statistic-95%CI
Plugging in the values for t and n and solving for RSD, we obtain
0 3*V3
V > RSD
4.3
0.12 > RSD
We conclude that if the RSD exceeds 0.12 (or 12 percent), we should test more than three
samples. Unfortunately, it may be difficult to obtain an RSD of less than 0.12 even when the
real emission rate is nearly constant. This is because the error of the concentration
measurements used in the testing can be near 10 percent. Thus, we might want to consider
whether we can tolerate a somewhat higher error in our emission factor.14 If we require that the
emission factor be within 40 percent of the true mean emission rate and solve the above
equation, we arrive at the conclusion that we should test more if the RSD exceeds 0.16 (16
percent). If we require that the emission factor be within 50 percent of the true mean emission
rate (with 95-percent confidence) and solve the above equation, we arrive at the conclusion that
we should test more if the RSD exceeds 0.20 (20 percent).
14 Another approach to minimizing the error is to require more samples initially, e.g., 4 instead of 3. With 4 initial
samples, the emission factor has a 95 percent chance of being within 30 percent of the true mean emission rate if the
RSD is less than 19 percent.
38
-------
With an additional three samples (n=6), the uncertainty of the emission factor declines
significantly. For n = 6, Equation RSD-5 becomes:
-2.57 <
(ef-eft)i
RSD
V<5
EF
<+2.57
Plugging an RSD of 0.20 (20 percent) into this equation, we find that after six samples,
there is a 95 percent chance that the sample mean (emission factor) is within 21 percent of the
true mean (average emission rate of the process).
4. Updates to emission factors and emission calculation factors.
The emission rate from a process is likely to change as the process evolves or is altered.
Process changes that can affect the fluorinated GHG emission rate include changes in raw
materials, equipment, production levels, or operating conditions. These changes can occur all at
once or gradually, over time. Facilities that have measured and re-measured their emission
factors over a period of several years have found that gradual, incremental changes to the process
(e.g., to improve yields) have significantly changed emission factors15. To ensure that emission
factors and emission calculation factors continue to accurately represent emission rates, they
should be updated when significant process changes occur. In addition, it may be appropriate to
update emission and emission calculation factors at regular intervals (e.g., five to ten years) to
capture the cumulative impact of small changes to the process.
5. Potential Process Vent Emission Threshold.
In discussions with EPA, representatives of fluorinated gas producers stated that more
effort and resources should be expended to accurately estimate the size of large emission streams
than to estimate the size of small ones. Consistent with this principle, EPA evaluated process
vent emissions cutoffs above which facilities would be required to conduct emissions testing to
develop a site-specific emissions factor and below which facilities could conduct emissions
calculations to develop a site-specific emissions factor.
There are various rules for other regulatory programs that have established emission
cutoffs above and below which there are different control, monitoring, and reporting
requirements. For example, there are process or process vent cutoffs in various NESHAP rules
in 40 CFR part 63, such as subpart GGG for Pharmaceutical Manufacturing, subpart FFFF for
Miscellaneous Organic NESHAP (MON), and subpart VVVVVV for Chemical Manufacturing
Area Source rule. Because some producers mentioned the rule requirement cutoffs in subpart
FFFF, Miscellaneous Organic NESHAP (MON), the initial step in the cutoff development was
reviewing the 10,000 lb/yr (i.e., 5 ton/yr) of uncontrolled organic HAP emissions cutoff for
processes in that rule. (While these data and facility practices for the MON source category may
not be completely analogous to the data and practices at fluorinated gas production facilities, this
information was used to inform the reporting rule-making process.) Similar to the fluorinated
gas production industry, the MON source category includes both continuous and batch processes.
15 Based on conversations with Fluorinated GHG producers, as referenced in Section 7.
39
-------
At the MON facilities, we looked at the emissions levels of MON process vents to identify the
number of vents covered and the emissions covered by various cutoffs. For continuous
processes, approximately 26 percent of process vents have emissions greater >10,000 lb/yr and
more than 99 percent of the total organic HAP emissions from continuous process vents occur
from these vents. For batch processes, 13 percent of batch process vents have emissions >10,000
lb/yr and the organic HAP emissions from these vents are 96 percent of the total organic HAP
emissions from batch process vents. The 10,000 lb/yr organic HAP cutoff for requiring control
of the process vent in the MON rule provides a good balance for minimizing the number of
process vents required to implement control and maximizing the emissions covered or captured
by the rule.
In evaluating cutoffs for the process vents in the fluorinated gas production source
category, we began by analyzing a 10,000 lb/yr fluorinated GHG uncontrolled emissions cutoff.
Because the quantities of these emissions in terms of C02e would vary significantly depending
on the specific fluorinated GHG compound emitted and its GWP, we decided to evaluate the
cutoff in terms of C02e. We assumed an average GWP of approximately 2,000 for the
fluorinated GHG, and given the cutoff value of 10,000 lb/yr fluorinated GHG, we developed a
cutoff of 10,000 mt CC>2e. Process vents above the 10,000 lb/yr limit would be required to
conduct an emissions test to develop a site-specific emissions factor for the vent, and process
vents below the 10,000 mt C02e could develop a site-specific emissions calculation factor based
on engineering calculations or engineering assessments.
Depending on the specific fluorinated GHG emitted, the emissions of the fluorinated
GHG in terms of the actual gas could vary significantly. In Table 4, the emissions in terms of
the fluorinated compounds that are translated from the C02e cutoff are shown. The fluorinated
compounds are shown in order of increasing GWP. For example, for fluorinated GHG
compounds as diverse as HFC-134a to SF6, the cutoff of 10,000 mt C02e would translate into
fluorinated GHG emissions ranging from 16,962 lb/yr to 923 lb/yr, respectively. However, the
detection limits of fluorinated GHG, depending on process vent flow rates, are likely sufficiently
low to allow detection and quantification of these compounds at these emission levels.
The emission threshold for process vents could also be based on controlled emissions, for
example, emission levels considering destruction efficiency are less than 10,000 mtC02e on an
annual basis.
Another potential approach for selecting a process vent cutoff might include a combined
approach, i.e., meeting both a C02e emissions cutoff and also a cutoff in terms of pounds of
fluorinated GHG. For example, if a process vent had emissions less than 25,000 mt C02e and
also less than 10,000 lb of fluorinated GHG emissions, i.e., the process vent meets both criteria,
then it would be considered below the cutoff.
An issue related to setting a process vent cutoff based on C02e is related to fluorinated
GHGs that do not have published GWP. In this instance, a lower level cutoff could be
established for those fluorinated GHG whose GWP does not appear in Table A-l of 40 CFR part
98 subpart A, for example, less than 1 metric ton for emissions that include a compound without
GWP. In addition, an average GWP, e.g., average GWP of 2,000, could be applied for all
fluorinated GHG whose GWP does not appear in Table A-l of 40 CFR part 98 subpart A.
40
-------
Table 4. Analysis of Uncontrolled Process Vent Cutoffs: Translate Emissions from C02e to
Pounds of Fluorinated GHG, and vice versa.1
Vent limit in
mtC02e
In pounds of fluorinated GHG:
HFC-134a
HFC-125
HFC-143a
CF4
C2F6
NF3
SF6
Convert C02e vent cutoff to lb of fluorinated GHG
25,000
42,404 *
19,688*
14,507 *
8,481
5,992
3,205
2,306
20,000
33,923 *
15,750*
11,605 *
6,785
4,793
2,564
1,845
10,000
16,962 *
7,875
5,803
3,392
2,397
1,282
923
5,000
8,481
3,938
2,901
1,696
1,198
641
461
Vent limit in
pounds of
fluorinated
GHG
In mtC02e:
HFC-134a
HFC-125
HFC-143a
CF4
C2F6
NF3
SF6
Convert lb of fluorinated GHG to mt C02e
10,000
5,896
12,698
17,234
29,478*
41,723*
78,005*
108,390 *
5,000
2,948
6,349
8,617
14,739
20,862
39,002
54,195
1 The GWP for fluorinated GHG are HFC-134a is 1,300; HFC-125 is 2,800; HFC-143a is 3,800; CF4 is 6,500 C2F6 is
9,200; NF3 is 17,200; and SF6 is 23,900.
* denotes that the level exceeds a 10,000-lb cutoff (in top table) or a 10,000-mtCC>2e cutoff (in bottom table).
6. Process Vent Preliminary Emission Estimates.
To determine whether a process vent is above or below the 10,000 mt C02e threshold,
facilities could use any of several emission estimation methods discussed above. In this
preliminary step, it seems reasonable to allow use of engineering calculations or engineering
assessments similar to those identified in Section 3.B.iii.2 to determine the emissions of
fluorinated GHG from the process. Acceptable calculation methods for uncontrolled and
controlled process vent emissions may include those from various rule text or guidance
documents, such as the MON, Pharmaceutical, and other examples provided in Section 3.B.iii.2
above.
Engineering assessments are another reasonable approach to determining preliminary
emissions from process vents. Engineering assessments include the types of information
discussed in Section 3.B.iii.2 above (previous test data, bench-scale or pilot-scale test data,
design analysis, etc.) Facilities typically have component and composition data available from
the original research and development activities on the process and from the process' lifetime of
operation. For process vent emissions determinations, facilities could conduct an engineering
assessment to calculate uncontrolled emissions for each process vent and for each emission
episode, and apply destruction efficiencies for periods of control, as applicable.
7. Current Plant Practices for Process Vent Monitoring
Several facilities with fluorinated gas production processes have indicated that they
conduct monitoring and analysis on individual process vents and that they characterize streams
from processes. They typically develop facility-specific or process-specific EFs and track
production activity data to estimate fluorinated GHG emissions.
Facilities indicated that a variety of approaches are used to estimate fluorinated GHG
emissions and develop process-specific emissions factors, including sampling and measurement
41
-------
(i.e., direct testing) and chemical engineering calculations and estimates. At most facilities, a
combination of these approaches is used to estimate fluorinated GHG emissions from the various
process vents at the site. Some facilities indicated, for example, that facility staff measure the
concentrations and flows for reactor vessel process vents and conduct engineering calculations to
estimate concentration and flows from process dryers. Engineering calculations are based on
common chemical engineering principles and empirically based data and process knowledge.
The engineering calculations may also be based on emission estimation equations contained in
rules, for example in the Pharmaceutical NESHAP, or may be conducted using purchased
process modeling software packages, such as ASPEN and EmissionsCalculator.
Facilities indicated that each of these approaches, emission testing and engineering
calculations, are valid and produce reasonable fluorinated GHG emission estimates and that
flexibility in which approach is applied to any particular process vent is appropriate. Facilities
particularly noted that they match the level of effort and "rigorousness" in the emissions
determination approach with the emissions source size. That is, they devote more effort to
characterizing emissions from relatively large fluorinated GHG emissions points, pursuing
emissions testing for these sources, and less effort to characterizing emissions from relatively
small fluorinated GHG emissions points, applying engineering calculations to these sources.
Some facilities noted that certain process vents are not measured directly because the emission
stream contains a compound that is hazardous for plant and personnel safety, e.g., if HF is
present in a stream. The facilities noted that emission points that are not safe to measure are
typically estimated from chemical engineering calculations and from process design data. In
using FTIR instruments, some vendors have noted that specialty glass sample cells are available
that are more resistant to HF. Some facilities noted that measuring fluorinated organics can be
done easily if the HF can be removed from the sampling stream. One option may be to flow the
sample stream through containers of water to remove HF prior to measuring the fluorinated
organic. While this approach may be successful at measuring the fluorinated organic, the flow
rate of the stream cannot be determined with any accuracy. This approach also does not alleviate
the personnel safety concerns.
Facilities further expanded on the use of previous stream analysis results as appropriate
data to be used. Generally, facilities have a good understanding of the chemistry for all
processes and understand what raw materials are fed to the process, what side reactions occur,
and what products and by-products are created. Facilities have laboratory scale, pilot plant, and
full scale streams analysis results. Facilities noted that in particular, at the bench-scale phase of a
product and process development, they analyze all streams to identify all species and compounds
such as product, byproducts, and waste. The process design from when the process was
originally developed and built is detailed, and more recently installed processes will have
excellent design and chemistry documentation available. Older processes may tend to have less
detail.
Facilities indicated that they typically conduct an initial analysis to develop the process-
specific emission factor or emission calculation factor and do not continuously measure and
calculate emissions from process vents and streams. When there is a process change, i.e., a
change in the operating conditions or a change in the process equipment used, they will conduct
a reanalysis of the process. Facilities pointed out that prior to making process change, they
analyze the theoretical chemistry and predict the likely changes in the set of fluorinated GHG
42
-------
compounds generated and in the fluorinated GHG emissions levels. They typically begin with
engineering calculations to determine the impacts of the change on species generated or emission
levels, and follow that with retesting and recalculations when appropriate. Some facilities noted
that after a process change, they will sample and analyze to identify every species or compound
generated, in either process vents or process streams.
8. Potential use of continuous emissions monitors to measure emissions from vents.
Another potential monitoring option EPA considered was requiring that facilities measure
emissions from fluorinated gas production facilities using continuous emissions monitors
(CEMS). If properly selected and maintained, CEMS would be expected to provide estimates of
emissions more accurate than either the mass-balance or the process-vent approach. Under this
approach, facilities would install and operate CEMS capable of measuring fluorinated GHGs to
measure process emissions. The requirements for the CEMs would be similar to those in subpart
C, adjusted, as appropriate, to accommodate CEMS for fluorinated gases. One possible option is
to use Fourier Transform Infrared Spectrometers (FTIRs) CEMS in scrubber stacks to measure
emissions. FTIR spectroscopy is presently used to conduct short-term fluorinated GHG emission
measurements from processes. In recent discussions with FTIR CEMS vendors, they noted that
the units can operate continuously. Some units require liquid nitrogen cooling and the liquid
nitrogen vessel (dewar) has a limited capacity. These nitrogen cooled units can operate for up to
12 hours on a single charge (nitrogen), and continuous operation of the unit could be achieved
through the use of an extended dewar with an automated fill cycle. With the extended dewar, the
unit can operate for approximately 3 days. (The detector is nitrogen cooled, and a dewar, which
is an insulated container with a vacuum between its inner and outer layers, is used to hold the
liquid nitrogen.) FTIR CEMS vendors noted that the sample cell can be adjusted to
accommodate concentration levels, where a shorter cell length can be used for higher
concentrations and longer cell lengths can be used for lower concentrations. Another option for
high concentrations might include use of dilution and a standard cell. The vendor typically sets
up a software package based on the compounds expected in the measured vent, and the software
is revised for other compounds as needed.
However, potential drawbacks to requiring CEMS are that they would be relatively
expensive to install and they may not tolerate the acidic and reactive environments (e.g., high HF
concentration) found in vents at many fluorinated gas production facilities. Some FTIR CEMS
vendors utilize specially configured gas cells that run hot and utilize mirror coatings to protect
the cell from high HF concentrations and use special types of unit seals (o-rings) that are inert to
corrosive compounds. In any case, vendors noted that HF in the vent streams is a common
problem, and that HF concentrations in the percent range are a difficult issue to deal with. The
latter concern might be mitigated by installing CEMS after a scrubber, if this is practicable.
Given these potential concerns, it may be appropriate to require CEMS for particularly large
emission streams, e.g., those resulting in emissions of more than 50,000 mtCC^e annually.
9. Equipment Leak Emissions Estimates.
Process vents are only one of several sources of fluorinated GHG emissions at fluorinated
gas production facilities. Another potentially significant source is equipment leaks, whose
emissions do not occur through process vents. Emissions from equipment leaks may be
estimated using EPA Method 21, EPA Protocol for Equipment Leak Emission Estimates (EPA-
43
-------
453/R-95-017) (EPA Protocol for Equipment Leaks), and other validated methods and
procedures capable of detecting the analyte of interest at the concentration of interest. Leak
monitoring could be performed periodically, which typically would be quarterly, semi-annually,
or annually. Due to the nature and scope of the reporting rule, which is focused on monitoring
and emission estimation rather than on leak repair (control), equipment leak monitoring on an
annual basis should be sufficient. The EPA Protocol for Equipment Leaks includes four methods
for estimating equipment leaks. These are, from least to most accurate, the Average Emission
Factor Approach, the Screening Ranges Approach, EPA Correlation Approach, and the Unit-
Specific Correlation Approach. The Average Emission Factor Approach relies on emission
factors and equipment counts, the service of each component, the composition of the stream, and
the length of time the equipment was in service to estimate emissions, and does not require any
real-world data be collected from the processes. The remaining three methods, The Screening
Ranges Approach, EPA Correlation Approach and the Unit-Specific Correlation Approach all
depend on the collection and use of facility-specific and process-specific screening data to
estimate emissions.
To use these three screening data methods, the facility would need to have (or develop)
Response Factors relating concentrations of the target fluorinated GHG (or surrogate gas co-
occurring in the stream) to concentrations of the gas with which the leak detector is calibrated.
Our understanding is that flame ionization detectors (FIDs) are generally insensitive to
fluorinated GHGs, and that they are therefore not likely to be effective for detecting and
quantifying fluorinated GHG leaks. An exception to this would be a situation in which the
fluorinated GHG occurred in a stream along with a substance (e.g., a hydrocarbon) to which the
FID was sensitive; in this case, the other substance could be used as a surrogate to quantify leaks
from the stream. We understand that at least two fluorocarbon producers currently use methods
in the EPA Protocol for Equipment Leaks to quantify their emissions of fluorinated GHGs with
different levels of accuracy and precision. Other analytical techniques that are sensitive to
fluorinated compounds may be available to monitor concentrations of equipment leaks, including
photoionization, ultraviolet, infrared, and others.
A preliminary review revealed a handful of detectors capable of detecting fluorinated
GHGs. While some fluorinated GHGs can be detected using instruments that meet EPA Method
21 specifications, others cannot. These detectors use various operating principles, including,
infrared, photoionization and ultraviolet detection. Although instruments for detecting leaks of
HFCs and SF6 from air-conditioning, refrigeration, and electrical equipment have existed for
some time, most of these instruments do not quantify emissions and/or detect only one or two
gases. In many cases, therefore, these instruments are not capable of quantifying emissions of
the broad range of fluorinated GHGs that can leak from process equipment in fluorinated gas
production facilities. One issue that arose with these detectors was their high sensitivity. For
some of the equipment, maximum readings were well below the 10,000 ppm leak/no-leak
threshold of the Screening Ranges Approach, meaning that the Screening Ranges Approach
could not be used with these detectors. Those capable of reading the 10,000 ppm concentration
had sampling rates that exceeded the rates established by Method 21. However, the EPA
Correlation Approach and the Unit-Specific Correlation Approach could be used with these
detectors, assuming that the detectors do not hit their maximum readings (peg) in the facility.
44
-------
Under the preliminary review of detectors, EPA did not identify equipment capable of
detecting and quantifying all fluorinated GHG emissions that can also meet all of the Method 21
specifications, or found that the equipment reaches its maximum ("pegs") at relatively low
concentrations. In general, in identifying the need for alternatives to EPA methods, EPA
discovered that the equipment and methods for detecting and quantifying emissions of
fluorinated GHGs from equipment leaks have not advanced as far as those for monitoring
emissions of VOC from equipment leaks. Additional alternatives for monitoring of equipment
leaks beyond Method 21 and the EPA Protocol for Equipment Leaks are likely warranted. For
example, facilities could monitor leaks using other approaches such as soap bubble testing
combined with the Average Emission Factors or the Screening Ranges Approach; use of
alternatives allowed under other regulatory programs, such as alternative leak detection methods
in 40 CFR 63, 63.1036, alternative means of emission limitations: batch processes; or use of
alternative work practices using optical imaging combined with Average Emission Factors. One
facility noted that they conducted room air exhaust monitoring to identify emissions from
equipment leaks and simultaneously conducted direct monitoring of pieces of equipment in
service within the room; the analytical results from the room air exhaust showed that
concentrations were at non-detect levels for all but one or two F GHG compounds, and the
monitoring of pieces of equipment showed non-detect levels for all but one or two F GHG. In a
monitoring approach for room air exhausts, facilities would need to explain how the approach
accounts for variations in concentrations in the room air and possibility that some emissions may
not be accounted for in the measurements (i.e., concentrations below non-detect levels due to
large room air volume or room air exhaust gas flow).
Another approach for monitoring leaks from pieces of equipment may include use of the
Alternative Work Practice (AWP) for EPA Method 21 (similar to monitoring requirements under
40 CFR part 60, subpart A, §40 CFR part 60.18; 40 CFR part 63, subpart A, §40 CFR part 63.11;
or 40 CFR part 65, subpart A, §40 CFR part 65.7). This approach would include monitoring
leaking equipment with an optical gas imaging instrument. Emissions from those pieces of
equipment found to be leaking could be estimated based on emission factors. Under this
approach, facilities could image each piece of equipment in fluorinated GHG service, and all
emissions imaged by the optical gas imaging instrument would be considered leaks and would be
subject to emissions estimation.
Several fluorinated gas producers have indicated that equipment leaks account for a small
share of facility-wide fluorinated GHG emissions. Although this generalization is largely based
on experience with VOC and HAP, two fluorinated gas producers have surveyed at least some of
their process equipment with detectors sensitive to fluorinated GHGs and have found a similar
low level of emissions.
10. Container Heel Venting
In addition to the emission sources mentioned above, the venting of container or cylinder
heels is an additional source of fluorinated GHG emissions. Emissions from returned containers
and from recycling depend heavily on the practices of individual facilities. If the facility
recovers container heels or simply refills containers on top of heels, then emissions from
returned containers will be small. However, if the facility vents heels, then emissions can be
significant, even dominating overall facility emission rates.
45
-------
According to the 2006 IPCC Guidelines 16, SF6 production operations in Germany whose
customers did not require highly purified gas had emission rates in the range of 0.2 percent of the
total quantity of SF6 produced. On the other hand, SF6 production operations in Japan whose
customers required highly purified gas had emission rates of 8 percent due to the venting of heels
whose purity was in doubt.
An analysis of the possible emissions from venting of residual fluorinated GHGs in
containers showed that these emissions could be significant. Cylinder heel venting was assumed
to occur only for fluorinated GHG that are used in etch and chemical vapor deposition chamber
clean processes in electronics. Because the electronics industry requires extremely high purity
for these etch and chamber clean gases, fluorinated gas producers may be reluctant to recycle
them, because recycling could inadvertently introduce impurities. Chemicals made in U.S. that
are used in etch and chamber clean processes include CF4, C2F6, C3F8, c-C4F80, CHF3 (HFC-23),
NF3, SF6.17 A total of 6 companies produce these etch and chamber clean gases at 7 facilities in
the U.S. To estimate the annual emissions from cylinder heel venting, it was assumed that 100
percent of the production of each of these chemicals, except SF6, was used in electronics. For
SF6, it was assumed that approximately 10 percent of SF6 production was used in electronics,
based on historical SF6 usage patterns. For 2006, the total production of etch and chamber clean
fluorinated GHG listed above, including 10 percent of the SF6 production, was over 60 million
mtC02e. Assuming that half of the producing facilities are venting cylinders and emit a 10
percent heel, the emissions from cylinder heel venting are estimated to be approximately 3
million mtCC^e.
The situation is similar for recycling. Emissions during recycling include emissions
during transfer of gas from containers to process equipment, but they also include emissions
associated with ridding the gas of contaminants, such as noncondensables (air, nitrogen, etc.) and
oil. (Since lubricants are chosen to be miscible with refrigerants, a significant quantity of
refrigerant can remain mixed into the oil when the oil is removed from the system or container.)
If facilities carefully recover and purify gas that is returned for recycling, emissions are likely to
be in the neighborhood of one percent or less of the quantities returned.18 However, if facilities
use crude methods to purge noncondensables from containers, e.g., venting from the headspace
of the cylinder, or if they don't attempt to minimize fluorinated GHG emissions from lubricants,
then emissions can be closer to five or ten percent of the quantities returned. (Early standards for
recycling equipment allowed emissions from purging as high as five percent;19 these limits have
been lowered considerably since.) In the worst case, facilities could simply vent returned
fluorinated GHGs, e.g., in cases where they are irretrievably contaminated.
16 From IPCC 2006, 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by the National
Greenhouse Gas Inventories Programme, Eggleston H.S., Buendia L., Miwa K., Ngara T. and Tanabe K. (eds).
Published: IGES, Japan., page 3,104 of Volume 3, Industrial Processes and Product Use.
17 CH2F2 (HFC-32) is also made in the U.S. and is used in etch and clean, but the fraction of the gas used for this
purpose was not known and therefore it was not included in the total (although it is expected to comprise a small
share of the total).
18 From IPCC Volume 3, Chapter 8. http://www.ipcc-
nggip.iges.or.jp/public/2006gl/pdf/3_Volume3/V3_8_Ch8_Other_Product.pdf
19 SAE J1990, 1989 Extraction and Recycle Equipment for Mobile Automotive Air-Conditioning Systems; ARI
Standard 740-1993, For Performance of Refrigerant Recovery and/or Recycling Equipment.
46
-------
To estimate fluorinated GHG emissions from container venting, facilities could either
measure the emissions vented from each container or develop site-specific emission factors, i.e.,
a heel factor, for each combination of fluorinated GHG, container size, and container type that
are vented. To measure the fluorinated GHG emissions, a facility would make a measurement
for each container, either by weighing the container or by measuring the pressure of the container
and calculating the mass using an equation of state.
The total emissions for venting of residual fluorinated GHG from containers may be
summed at the facility based on the number of containers received and vented. Equation L-32
could be used to calculate the annual emissions of each fluorinated GHG...
Ev=tHm~£Hm
' 1 (Eq L-32)
Where:
ECf = Total mass of each fluorinated GHG f emitted from the facility through venting of
residual fluorinated GHG from containers, annual basis (kg/year).
HBfj = Mass of residual fluorinated GHG f in container j when received by facility.
HEfj = Mass of residual fluorinated GHG f in container j after evacuation by facility.
(Facility may equate to zero.)
n = Number of vented containers for each fluorinated GHG f.
A facility could develop heel factors based on representative samples of the containers
received by the facility from fluorinated GHG users. A facility would select a representative
sample of containers for each combination of fluorinated GHG, container size, and container
type that a facility vents. A representative sample would reflect the full range of quantities of
residual gas returned in that container size and type. To determine the residual weight or
pressure, facilities could monitor the mass or the pressure of your cylinders/containers. If the
facility monitors the pressure, the ideal gas law in equation L-33 would be used to convert the
pressure to mass.
pV=ZnRT (Eq. L-33)
Where:
P
Absolute pressure of the gas (Pa)
V
Volume of the gas (m3)
z
Compressibility factor
n =
Amount of substance of the gas (moles)
R
Gas constant (8.314 Joule/Kelvin mole)
T
Absolute temperature (K)
47
-------
Facilities would use the container residual mass for each specific fluorinated GHG to
determine what mass fraction of the initial mass in the container is vented.
Facilities would then calculate the annual fluorinated GHG emissions based on the gas-
specific heel factor and the number of containers that are returned to the facility, using Equation
L-34.
Ecf=ilh£*N£*F£ (Eq. L-34)
j=i
Where:
ECf = Total mass of each fluorinated GHG f emitted from the facility through
venting of residual fluorinated GHG from containers, annual basis (kg/year).
h[j = Facility-wide gas-specific heel factor for input gas f (fraction) and container
size and type j, as determined in §98.124(j) of this subpart.
N|j = Number of containers of size and type j returned to the fluorinated gas
production facility.
Fg = Full capacity of containers of size and type j containing fluorinated GHG f
(kg).
n = Number of combinations of container sizes and types for fluorinated GHG f.
11. Destruction Efficiency Testing
It may be appropriate to require that fluorinated gas producers that destroy F-GHGs
conduct emissions testing periodically, for example every five or ten years, to determine the
destruction efficiency (DE) of the destruction device. The testing for determining the DE would
be similar to the emissions testing that could be conducted to develop site-specific process-vent-
specific emission factors. Facilities would need to conduct testing that shows the destruction
device can successfully destroy emissions at worse-case operating conditions, for example when
operating at high loads reasonably expected to occur and when destroying the most-difficult-to-
destroy fluorinated GHG fed into the device (or when destroying a surrogate that was more
difficult to destroy than that fluorinated GHG). The last point is particularly important because
some fluorinated GHGs (e.g., CF4, SF620 and other PFCs) are extremely difficult to destroy; DEs
determined for other fluorinated GHGs (or for typical Class 1 POHCs) would overestimate the
destruction of these fluorinated GHGs.
For destruction of fluorinated GHG compounds to occur, temperatures must be quite
high, fuel must be provided, flow rates of fuels and air (or oxygen) must be kept above certain
limits, flow rates of fluorinated GHG must be kept below others, and for some particularly
20 For example, SF6 ranks 4th in the Class 1 incinerability rankings of the Office of Research and Development and
Office of Solid Waste and Emergency Response of U.S. Environmental Protection Agency's Guidance on Setting
Permit Conditions and Reporting Trial Burn Results Handbook, Volume II. EPA Publication No. EPA/625/6-
89/019. January 1989.
48
-------
difficult-to-destroy chemicals, pure oxygen must sometimes be fed into the process. If one or
more of these process requirements is not met, destruction efficiencies can drop sharply (in some
cases, by an order of magnitude or more), and fluorinated GHGs will simply be exhausted from
the device. For more discussion of this issue, see the Technical Support Document for Industrial
Gas Supply: Production, Transformation, and Destruction of Fluorinated GHGs and N2O, Docket
number EPA-HQ-OAR-2008-0508-041.
In the destruction and removal efficiency testing that is performed at hazardous waste
combustors pursuant to part 63, subpart EEE, facilities that demonstrate 99.99 percent DRE for a
POHC within a thermal stability class are typically allowed to assume that 99.99 percent DRE
would also be achieved for the other compounds in that class and for compounds in other thermal
stability classes with lower thermal stability rankings. This approach is based on the general
conclusion that, for POHCs that are in the same class and that occur in significant volumes,
differences in DREs tend to be small, and that compounds in other thermal stability classes with
lower stability rankings are easier to destroy.
However, it would be a misapplication of the thermal stability index to conclude that a
combustor that has demonstrated 99.99 percent DRE for any Class 1 compound21 would also
achieve 99.99 percent DRE for SF6, a Class 1 compound, and for perfluoromethane (CF4).
While achieving 99.99 percent DRE for SF6 ensures 99.99 percent DRE for other Class 1
compounds, the converse may not be true. SF6 is substantially more thermally stable than other
Class 1 compounds (and CF4 is substantially more thermally stable than SF6).
The theoretical considerations that support the conclusion that fluorinated GHGs are
extremely thermally stable relate to the high energies of the C-F and S-F bonds. These energies
make it difficult to break the bonds through reaction with oxygen, hydrogen, or the hydroxyl
radical, the typical means of destroying other class 1 compounds. Essentially, the only path
available to destroy these fully fluorinated compounds in hazardous waste combustors or thermal
oxidizers is through thermal decomposition at very high temperatures.22 These temperatures are
significantly higher than those required for the thermal decomposition of most other class 1
compounds. For SF6, the thermal stability index indicates that the temperature to achieve 99
percent destruction with a two-second residence time is 1,090°C; for CF4, we project that the
temperature would be on the order of 1,380°C.23 Researchers have suggested that CF4 may
break down only in the flame zone.24
Experimental evidence supports the idea that SF6 and CF4 are difficult to destroy. Due in
part to the theoretical considerations outlined above, several studies have evaluated the use of
SF6 as a possible surrogate for POHCs in evaluating DREs. Most studies have verified that the
DRE measured for SF6 is likely to be lower than that for POHCs, i.e., that it is likely to yield a
21 Class 1 is the group of POHCs and surrogates with the highest thermal stability, meaning they are the most
difficult compounds to destroy.
22W. Tsang et al make this case for perfluoromethane in Tsang, W., Burgess Jr., D. R., and Babushok, V. (1998)
"On the Incinerability of Highly Fluorinated Organic Compounds," Combustion Science and Technology. 139:1,
385-402. An analogous argument can be made for sulfur hexafluoride.
23 SF6 temperature is from Appendix VIII ranking of POHCs; CF4 temperature is estimated based on the rate
constant provided in Tsang, p. 393.
24 Tsang, p. 387.
49
-------
conservative estimate of the DREs for POHCs under most conditions. In one experiment at a
full-scale hazardous waste incinerator, the investigators found that even at high-temperature
conditions, SF6 had a DRE that led to emissions approximately an order of magnitude higher
than those of other POHCs, including both class 1 and class 2 compounds. At lower-temperature
conditions, SF6 had a DRE that was over 100 times lower than those of other POHCs.25 As
noted above, CF4 is even more difficult to destroy than SF6. This has been confirmed in testing
of point-of-use thermal abatement devices used in electronics manufacturing, which destroyed
CF4 with an efficiency that was significantly lower (sometimes orders of magnitude lower) than
the efficiency with which they destroyed SF6.26
Other fluorinated compounds are not likely to be as stable as CF4 and SF6 because they
can be dissociated at C-H and C-C bonds (which are weaker than C-F and S-F bonds).
Nevertheless, higher molecular weight perfluorocarbons such as C2F6 are still expected to be
relatively difficult to incinerate.27 As is true for CF4, the mechanism of destruction is expected
to be thermal decomposition rather than attack by radicals, although the decomposition
temperature will be lower than for CF4 due to the fact that the C-C bond is weaker than the C-F
bond.
It may be appropriate to allow facilities to rely on data from destruction device testing
that has been conducted within the five or ten years prior to the effective date of the rule.
Facilities that have conducted an emissions test on their destruction device within the five or ten
years prior to the effective date of the rule may be allowed to use the DE determined during that
test if the test was conducted in accordance with the proposed test requirements. Facilities could
also potentially use the DREs determined during principal organic hazardous constituent testing
and hazardous waste combustor testing, provided those tests determined the DRE based on the
most-difficult-to-destroy fluorinated GHG fed into the device (or based on a surrogate that was
more difficult to destroy than the most-difficult-to-destroy fluorinated GHG).
C. Other Potentially Significant Emission Points
In addition to the fluorinated GHG emissions captured via the Mass Balance method or
either of the Process Vent methods, other potentially significant emission points exist both
upstream and downstream of the production measurement. These other emission points under
this source category may include fluorinated GHG emissions from equipment leaks, storage
tanks, wastewater, container filling (loading), refrigerant blending, and reclamation and recycling
processes,, particularly where these emissions occur before the production measurement at
25 A. Trenholm, C. Lee, and H. Jermyn, "Full-Scale POHC Incinerability Ranking and Surrogate Testing," 17th
Annual RREL Hazardous Waste Research Symposium, EPA Office of Research and Development, EPA/600/9-
91/002 April, 1991, pp. 79-88.
26 USEPA, "Developing a Reliable Fluorinated Greenhouse Gas (F-GHG) Destruction or Removal Efficiency
(DRE) Measurement Method for Electronics Manufacturing: A Cooperative Evaluation with Qimonda," March
2008, EPA 430-R-08-017; USEPA, "Developing a Reliable Fluorinated Greenhouse Gas (F-GHG) Destruction or
Removal Efficiency (DRE) Measurement Method for Electronics Manufacturing: A Cooperative Evaluation with
IBM," June 2009, EPA 430-R-10-004; and USEPA, "Developing a Reliable Fluorinated Greenhouse Gas (F-GHG)
Destruction or Removal Efficiency (DRE) Measurement Method for Electronics Manufacturing: A Cooperative
Evaluation with NEC Electronics, Inc.," December 2008, EPA 430-R-10-005.
27 Tsang, p. 401.
50
-------
fluorinated GHG production facilities. (Equipment leak emissions and container heel evacuation
are discussed above in separate sections.)
Fluorinated GHG emissions from wastewater will depend directly on the concentration
and volume of the contaminants that are present in the wastewater. Based on an evaluation using
EPA's Estimation Program Interface (EPI) Suite, approximately 90 percent or more of all
fluorinated GHG compounds present in wastewater will be emitted to the atmosphere. The
operating permits for some facilities referred to wastewater treatment activities for fluorinated
compounds, although it is unclear what level of emissions might occur from wastewater and
whether there would be wastewater emissions at all facilities.
The fluorinated GHG emissions from storage tanks are anticipated to be small to
insignificant due to the use of pressurized tanks for storage. Emissions from blending activities
are also likely to be relatively small. In both cases, however, emissions can be more significant
if leaks or catastrophic failures occur.
Currently, it is EPA's understanding that most fluorinated GHG production facilities
measure their production before container filling (loading), e.g., by using flowmeters just
upstream of the container connection to measure the mass flowing into the containers. If this is
the case, emissions that occur during or after filling (e.g., from hoses and connections) would
have been included in the production (supply) measurement. However, if production is
measured by weighing containers before and after filling, then emissions during container filling
would not have been included in the production measurement. In these cases, facilities using the
emission factor approach would need to quantify container filling emissions for completeness.
The IPCC Guidelines (Volume 3, Chapter 7) estimate that emissions related to container
management (including refrigerant transfers between containers and handling of returned
containers) range between 2 and 10 percent of the refrigerant market.
In the event that the other potentially significant emission points were included under the
rule, possible methods for tracking these emissions include engineering estimates (e.g., for
container filling), equipment leak testing (e.g., for storage tanks and blending activities), default
or site-specific emission factors (all sources), and mass balances (e.g., weighing returned
container heels).
4. Procedures for Estimating Missing Data
In the event that a scale or flowmeter normally used to measure reactants, products, by-
products, or wastes fails to meet a test to verify its accuracy or precision, malfunctions, or is
rendered inoperable, facilities may estimate these quantities using other measurements where
these data are available. For example, facilities that ordinarily measure production by metering
the flow into the day tank could use the weight of product charged into shipping containers for
sale and distribution as a substitute. It is our understanding that the types of flowmeters and
scales used to measure fluorocarbon production (e.g., Coriolis meters) are generally quite
reliable, and therefore that it should rarely be necessary to rely solely on secondary production
measurements. In general, production facilities rely on accurate monitoring and reporting of the
inputs and outputs of the production process. In the event that a secondary mass measurement
51
-------
for the stream is not available, producers can use a related parameter and the historical
relationship between the related parameter and the missing parameter to estimate the flow.
If concentration measurements are unavailable for some period, we are proposing that the
facility use the average of the concentration measurements from just before and just after the
period of missing data.
5. QA/QC Requirements
Typical QA/QC requirements for measuring devices include initial and periodic
verification and calibration. (For example, see the requirements of EPA's Acid Rain regulations
at 40 CFR Part 75.) In this case, it may be appropriate to require an initial verification of
flowmeters and weigh scales and periodic calibration in accordance with the applicable industry
standards or manufacturer directions. If the flowmeter manufacturer performed this verification
at the flowmeter factory and did not recommend a second verification upon installation, the
factory verification should suffice. Calibration of flowmeters and scales could be performed
prior to the reporting year; after the initial calibration, recalibration could be performed as
frequently as specified by the manufacturer. Under this approach, producers could perform the
verification and calibration of their weigh scales during routine product line maintenance.
For the gas chromatography analytical method described under the monitoring section of
this document, monthly calibration, using known certified standards should be used. The
calibration involves validating accurate measurement of these fluorocarbon standards across a
range of possible concentrations, depending on which process streams are being measured.
For development of process-vent-specific emission factors, facilities would need to
conduct and meet the QA/QC procedures specified in the test methods used.
6. Reporting and Recordkeeping Procedures
To verify and document their emissions estimates, owners and operators of facilities
producing fluorinated gases would report the following information, depending on their activities
and on the estimation approach that they used. All facilities would report both their fluorinated
GHG emissions and the quantities used to estimate them on a process-specific basis. They
would also report the results of each initial scoping speciation, specifically, the chemical
identities of the contents of potentially emitted streams. Facilities using the mass-balance
approach would report the masses of the reactants, products, by-products, and wastes, and, if
applicable, the quantities of any product in the by-products and/or wastes (if that product is
emitted at the facility). They would also report the chemical identities of reactants, products, and
by-products, along with the chemical equations used to estimate emissions. Facilities using the
emission factor approach would report the activity data used to calculate emissions (e.g., the
quantity produced, transformed, or destroyed) and the emission factors used to estimate them.
Owners and operators would report annual totals of these quantities by process and facility.
Where fluorinated GHG production facilities have estimated missing data, the facility
would report the reason the data were missing, the length of time the data were missing, the
method used to estimate the missing data, and the estimates of those data.
52
-------
Maintaining records of the information used to determine the reported GHG emissions
would enable us to verify that the GHG emissions monitoring and calculations were performed
correctly. Owners and operators of facilities producing fluorinated GHGs would retain records
documenting the data reported, including records of monthly emission estimation calculations,
all data that went in to the calculations, calibration records for flowmeters, scales, and gas
chromatographs, and documentation of emission factor development activities.
53
-------
7. References
3M 2009. Meeting Documentation for Telephone Calls on October 9 and November 12, 2009
with 3M Representatives Regarding Fluorinated Gas Production. U.S. Environmental Protection
Agency.
Air Products 2009. Meeting Documentation for 17 November 2009 Telephone Call with Air
Products Representatives Regarding Fluorinated Gas Production. RTI International.
Arkema 2009. Meeting Documentation for Various Telephone Calls During Week of 7 October
2009 with Arkema Representatives Regarding Fluorinated Gas Production. U.S. Environmental
Protection Agency.
DuPont 2009. Meeting Documentation for 13 November 2009 Telephone Call with DuPont
Fluoropolymer Representatives Regarding Fluorinated Gas Production. RTI International.
INEOS Fluor 2009. Meeting Documentation for October 2009 Telephone Calls with INEOS
Fluor Representatives Regarding Fluorinated Gas Production. RTI International.
3M Site Visit 2009. Documentation for October 2009 Site Visit to 3M Cordova, Illinois
Regarding Fluorinated Gas Production. RTI International.
DuPont Site Visit 2009. Documentation for October 2009 Site Visit to DuPont Corpus Christi,
Texas Regarding Fluorinated Gas Production. RTI International.
Centre Interprofessionnel Technique d'Etudes de la Pollution Atmospherique. Inventaire Des
Emissions De Gaz A Effect De Serre En France Au Titre de la Convention Cadre Des Nations
Unies Dur Les Changements Climatiques. Mars 2009.
http://unfccc.int/files/national reports/annex i ghg inventories/national inventories submission
s/application/zip/fra 2009 nir 7apr.zip
Daniela Romano, Chiara Arcarese, Antonella Bernetti, Antonio Caputo, Rocio D. Condor, Mario
Contaldi, Riccardo De Lauretis, Eleonora Di Cristofaro, Sandro Federici, Andrea Gagna, Barbara
Gonella, Riccardo Liburdi, Ernesto Taurino, Marina Vitullo. ISPRA - Institute for
Environmental Protection and Research. Italian Greenhouse Gas Inventory 1990-2007. National
Inventory Report, 2009.
http://unfccc.int/files/national reports/annex i ghg inventories/national inventories submission
s/application/zip/ita 2009 nir 15apr.zip
EPA Office of Research and Development and Office of Solid Waste and Emergency Response
of U.S. Environmental Protection Agency. Guidance on Setting Permit Conditions and Reporting
Trial Burn Results Handbook, Volume II. EPA Publication No. EPA/625/6-89/019. January
1989.http://www.epa.gov/nrmrl/pubs/625689019/625689019.pdf
EPA Office of Pollution Prevention Toxics and Syracuse Research Corporation (SRC). EPI Suite
Version 4.00 January 2009. http://www.epa. gov/oppt/exposure/pub s/epi suite .htm
54
-------
Flemish Environment Agency, Flemish Institute for Technological Research, Walloon Public
Service, Brussels Institute for the Management of the Environment, Interregional Cell for the
Environment, ECONOTEC. Belgium's Greenhouse Gas Inventory (1990-2007). April 2009.
http://unfccc.int/files/national reports/annex i ghg inventories/national inventories submission
s/application/zip/bel 2009 nir 15apr.zip
IPCC 2006, 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by the
National Greenhouse Gas Inventories Programme, Eggleston H.S., Buendia L., Miwa K., Ngara
T. and Tanabe K. (eds). Published: IGES, Japan. Chapters 3, 6, 7 and 8. http://www.ipcc-
nggip.iges.or.ip/public/2006gl/vol3.html
IPCC 2006, 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by the
National Greenhouse Gas Inventories Programme, Eggleston H.S., Buendia L., Miwa K., Ngara
T. and Tanabe K. (eds). Published: IGES, Japan, http://www.ipcc-
nggip.iges.or.ip/public/2006gl/index.html
Memorandum from Schaffner, K. and Hancy, C., RTI International, to Ottinger, D.,
EPA/OAR/CCD. Threshold Analysis for Emissions, Promulgation of 40 CFR Part 98, subpart
L, Fluorinated Greenhouse Gas Production. November 4, 2010.
Tsang, W., Burgess Jr., D. R., and Babushok, V. (1998) "On the Incinerability of Highly
Fluorinated Organic Compounds," Combustion Science and Technology. 139:1, 385-402.
Trenholm, A., Lee, C., and Jermyn, H. "Full-Scale POHC Incinerability Ranking and Surrogate
Testing," 17th Annual RREL Hazardous Waste Research Symposium, EPA Office of Research
and Development, EPA/600/9-91/002 April, 1991, pp. 79-88.
U.S. Environmental Protection Agency (EPA), "Developing a Reliable Fluorinated Greenhouse
Gas (F-GHG) Destruction or Removal Efficiency (DRE) Measurement Method for Electronics
Manufacturing: A Cooperative Evaluation with Qimonda," March 2008, EPA 430-R-08-017.
http://www.epa.gov/semiconductor-pfc/documents/qimonda report.pdf
U.S. Environmental Protection Agency (EPA), "Developing a Reliable Fluorinated Greenhouse
Gas (F-GHG) Destruction or Removal Efficiency (DRE) Measurement Method for Electronics
Manufacturing: A Cooperative Evaluation with IBM," June 2009, EPA 430-R-10-004.
http://www.epa.gov/semiconductor-pfc/documents/ibm report.pdf
U.S. Environmental Protection Agency (EPA), "Developing a Reliable Fluorinated Greenhouse
Gas (F-GHG) Destruction or Removal Efficiency (DRE) Measurement Method for Electronics
Manufacturing: A Cooperative Evaluation with NEC Electronics, Inc.," December 2008, EPA
430-R-10-005. http://www.epa.gov/semiconductor-pfc/documents/nec report.pdf
U.S. Environmental Protection Agency (EPA). Emission Inventory Improvement Program
Volume II: Chapter 16. Methods for Estimating Air Emissions From Chemical Manufacturing
Facilities. August 2007. www.epa.gov/ttn/chief/eiip/techreport/volume02/ii 16 aug2007final.pdf
U.S. Environmental Protection Agency (EPA) Office of Air and Radiation and Office of
Atmospheric Programs, Climate Change Division. Protocol for Measuring Destruction or
55
-------
Removal Efficiency (DRE) of Fluorinated Greenhouse Gas Abatement Equipment in Electronics
Manufacturing. March 2010.
U.S. Environmental Protection Agency (EPA) Office of Air Quality Planning and Standards.
Protocol for Equipment Leak Emission Estimates. EPA-453/R-95-017. November 1995.
www.epa.gov/ttnchiel/efdocs/equiplks.pdf
U.S. Environmental Protection Agency (EPA). Eli Lilly and Company, Airtech Environmental
Services Inc.. Tracer Gas Protocol for the Determination of Volumetric Flow Rate Through the
Ring Pipe of the Xact Multi-Metals Monitoring System. September 2006.
www.epa.gov/ttnemc01/prelim/otm24.pdf
U.S. Environmental Protection Agency (EPA). Emission Measurement Center Approved
Alternate Method (ALT-012)- An Alternate Procedure for Stack Gas Volumetric Flow Rate
Determination (Tracer Gas), http://www.epa.gov/ttn/emc/approalt/alt-012.pdf
U.S. Environmental Protection Agency (EPA). Method 1- Sample and Velocity Traverses for
Stationary Sources, http://www.epa. gov/ttnemcO l/promgate/m-01 .pdf
U.S. Environmental Protection Agency (EPA). Method 18 - Measurement of Gaseous Organic
Compound Emissions by Gas Chromatography, http://www.epa.gov/ttnemc01/promgate/m-
18.pdf
U.S. Environmental Protection Agency (EPA). Method 205 - Verification of Gas Dilution
Systems for Field Instrument Calibrations http://www.epa. gov/ttnemcO l/promgate/m-205 .pdf
U.S. Environmental Protection Agency (EPA). Method 21 - Determination of Volatile Organic
Compound Leaks, http://www.epa.gov/ttnemc01/promgate/m-21.pdf
U.S. Environmental Protection Agency (EPA). Method 2A - Direct Measurement of Gas
Volume Through Pipes and Small Ducts, http://www.epa.gov/ttnemcOl/promgate/m-02a.pdf
U.S. Environmental Protection Agency (EPA). Method 2B- Determination of Exhaust Gas
Volume Flow Rate from Gasoline Vapor Incinerators.
http://www.epa.gov/ttnemc01/promgate/m-02b.pdf
U.S. Environmental Protection Agency (EPA). Method 2C- Determination of Gas Velocity and
Volumetric Flow Rate in Small Stacks or Ducts (Standard Pitot Tube)
http://www.epa.gov/ttnemcQl/promgate/m-02c.pdf
U.S. Environmental Protection Agency (EPA). Method 2D - Measurement of Gas Volume Flow
Rates in Small Pipes and Ducts, http://www.epa.gov/ttnemc01/promgate/m-02d.pdf
U.S. Environmental Protection Agency (EPA). Method 2F- Determination of Stack Gas
Velocity and Volumetric Flow Rate With Three-Dimensional Probes.
http://www.epa.gov/ttnemc01/promgate/Methd2F.pdf
56
-------
U.S. Environmental Protection Agency (EPA). Method 2G- Determination of Stack Gas Velocity
and Volumetric Flow Rate With Two-Dimensional Probes.
http://www.epa.gov/ttnemc01/promgate/Methd2G.pdf
U.S. Environmental Protection Agency (EPA). Method 301—Field Validation of Pollutant
Measurement Methods from Various Waste Media, http://www.epa.gov/ttnemc01/promgate/m-
301.pdf
U.S. Environmental Protection Agency (EPA). Method 3B- Gas Analysis for the Determination
of Emission Rate Correction Factor or Excess Air. http://www.epa.gov/ttnemc01/promgate/m-
03b.pdf
U.S. Environmental Protection Agency (EPA). Method 3-Gas Analysis for the Determination of
Dry Molecular Weight, http://www.epa.gov/ttnemc01/promgate/m-03.pdf
U.S. Environmental Protection Agency (EPA). Method 4 - Determination of Moisture Content
in Stack Gases, http://www.epa.gov/ttnemc01/promgate/m-04.pdf
U.S. Environmental Protection Agency (EPA). Test Method 320- Measurement of Vapor Phase
Organic and Inorganic Emissions by Extractive Fourier Transform Infrared (FTIR)
Spectroscopy, http://www.epa.gov/ttnemc01/promgate/m-320.pdf
U.S. Environmental Protection Agency Center for Environmental Research Information, Office
of Research and Development. Compendium Method TO-15- Determination Of Volatile
Organic Compounds (VOCs) In Air Collected In Specially-Prepared Canisters and Analyzed By
Gas Chromatography/Mass Spectrometry (GC/MS). January 1999.
www.epa.gov/ttnamtil/files/ambient/airtox/to-15r.pdf
U.S. Environmental Protection Agency Office of Air and Radiation. Technical Support
Document for Emissions From Production of Fluorinated GHGs: Proposed Rule for Mandatory
Reporting of Greenhouse Gas. February 2, 2009.
57
-------
Appendix A.
Mass Balance Approach: Equations from Re-Proposal TSD (March 22, 2010).
58
-------
Appendix A
Mass Balance Approach: Equations from Re-Proposal TSD (March 22, 2010).
In the mass-balance approach, facilities weigh or meter the reactants fed into the process,
the product resulting from the process, and any byproducts and wastes that are removed from the
process (i.e., sent to the thermal oxidizer or other equipment, not immediately recycled back into
the process). Facilities then calculate the difference between the mass of reactants fed into the
process and the sum of the masses of the main product and those of any byproducts and/or
wastes. This difference is then assigned to loss of reactants, loss of product, and/or conversion to
byproducts.
The following set of equations was included in the re-proposal TSD for the mass-
balancing method for estimating fluorinated GHG emission.
The total mass of each fluorinated GHG product emitted annually from all fluorinated gas
production processes would be estimated by using Equation L-5:
n m
Ep=yZyZEP'P (Eq. L-5)
p=1 2=1
Where:
EP
Total mass of each fluorinated GHG product emitted annually from all production
processes (metric tons).
Epip -
Total mass of the fluorinated GHG product emitted from production process i over
the period p (metric tons, defined in Equation L-7).
n =
Number of concentration and flow measurement periods for the year.
m =
Number of production processes.
The total mass of fluorinated GHG by-product k emitted annually from all fluorinated gas
production processes shall be estimated by using Equation L-6:
n m
EBk=Y,HEBMP (Eq. L-6)
p=1 z=l
Where:
EBk = Total mass of fluorinated GHG by-product k emitted annually from all production
processes (metric tons).
Eekip = Total mass of fluorinated GHG by-product k emitted from production process i over
the period p (metric tons, defined in Equation L-8 on this section),
n = Number of concentration and flow measurement periods for the year,
m = Number of production processes.
59
-------
The total mass of each fluorinated GHG product emitted from production process i over
the period p would be estimated by calculating the difference between the expected production of
the fluorinated GHG based on the consumption of one of the reactants (e.g., HF or a
chlorofluorocarbon reactant) and the measured production of the fluorinated GHG, accounting
for yield losses related to by-products and wastes. This calculation shall be performed using
Equation L-7.
Where:
Epip = Total mass of each fluorinated GHG product emitted from production process i over
the period p (metric tons).
P = Total mass of the fluorinated GHG produced by production process i over the period
p (metric tons).
R = Total mass of the reactant that is consumed by production process i over the period p
(metric tons, defined in Equation L-8).
MWr = Molecular weight of the reactant.
MWp = Molecular weight of the fluorinated GHG produced.
SCr = Stoichiometric coefficient of the reactant.
SCp = Stoichiometric coefficient of the fluorinated GHG produced.
Cp = Concentration (mass fraction) of the fluorinated GHG product in stream j of
destroyed wastes. If this concentration is only a trace concentration, cp is equal to
zero.
WDj = Mass of wastes removed from production process i in stream j and destroyed over the
period p (metric tons, defined in Equation L-9).
LBkip = Yield loss related to by-product k for production process i over the period p (metric
tons, defined in Equation L-10).
q = Number of waste streams destroyed in production process i.
u = Number of by-products generated in production process i.
The total mass of the reactant that is consumed by production process i over the period p
shall be estimated by using Equation L-8:
R = Rp-Rr (Eq. L-8)
Where:
R = Total mass of the reactant that is consumed by production process i over the period p
(metric tons).
Rf = Total mass of the reactant that is fed into production process i over the period p
(metric tons).
Rr = Total mass of the reactant that is permanently removed from production process i
over the period p (metric tons).
60
-------
The mass of wastes removed from production process i in stream j and destroyed over the
period p would be estimated using Equation L-9:
WDJ = WFj * DE (Eq. L-9)
Where:
W|)j = The mass of wastes removed from production process i in stream j and destroyed over
the period p (metric tons).
WFj = The total mass of wastes removed from production process i in stream j and fed into
the destruction device over the period p (metric tons).
DE = Destruction Efficiency of the destruction device (fraction).
Yield loss related to by-product k for production process i over period p would be
estimated using Equation L-10:
(B,p *MWr *MEj
(MW„ * MEp)
£*,, = (Eq. L-10)
Where:
LBkip = Yield loss related to by-product k for production process i over the period p (metric
tons).
Bkip = Mass of by-product k generated by production process i over the period p (metric
tons, defined in Equation L-l 1).
MWp = Molecular weight of the fluorinated GHG produced.
MWisk = Molecular weight of by-product k.
MEBk = Moles of the element shared by the reactant, product, and by-product k per mole of
by-product k.
MEP = Moles of the element shared by the reactant, product, and by-product k per mole of
the product.
If by-product k is responsible for yield loss in production process i and occurs in any
process stream in more than trace concentrations, the mass of by-product k generated by
production process i over the period p would be estimated using Equation L-l 1:
K = £
-------
CBkj = Concentration (mass fraction) of the by-product k in stream j of production process i
over the period p. If this concentration is only a trace concentration, cBkj is equal to
zero.
Sj = Mass flow of process stream j of production process i over the period p.
q = Number of streams in production process i.
If by-product k is responsible for yield loss, is a fluorinated GHG, occurs in any process
stream in more than trace concentrations, and is not completely recaptured or completely
destroyed; the total mass of by-product k emitted from production process i over the period p
would be estimated using Equation L-12:
Ebmp =BMP~Y, cBk,- * WDj - S CBU * Sm (Eq- L-12)
j=i 1=1
Where:
Eekip = Mass of by-product k emitted from production process i over the period p (metric
tons).
Bkip = Mass of by-product k generated by production process i over the period p (metric
tons).
CBkj = Concentration (mass fraction) of the by-product k in stream j of destroyed wastes
over the period p. If this concentration is only a trace concentration, cBj is equal to
zero.
W|)j = The mass of wastes that are removed from production process i in stream j and that
are destroyed over the period p (metric tons, defined in Equation L-9).
Cbw = The concentration (mass fraction) of the by-product k in stream 1 of recaptured
material over the period p. If this concentration is only a trace concentration, Cbh is
equal to zero.
Sri = The mass of materials that are removed from production process i in stream 1 and that
are recaptured over the period p.
q = Number of waste streams destroyed in production process i.
x = Number of streams recaptured in production process i.
Choice of Reactant Whose Yield Is Measured.
EPA has considered use of one reactant to estimate emissions under the mass-balance
approach (rather than both as originally proposed). Some fluorinated GHG producers have noted
that, for various reasons, it is sometimes considerably more difficult to track the yields of some
reactants than others (e.g., HF vs. an organic feedstock). EPA notes that facilities estimating
their emissions based on the yield of one reactant would still need to be able to demonstrate that
their estimate met the statistical error test that might be required.
If an alternate reactant was measured, it would be necessary for facilities to be extremely
thorough in the monitoring approach and calculations in order to account for all potential
emissions. The reactant whose yield is measured and its relationship to fluorinated GHG
62
-------
emissions for a particular process must be fully analyzed to ensure that it provided an accurate
representation of the emissions.
63
------- |