Technical Support Document for Emissions from Electric Equipment Manufacture or Refurbishment and Manufacturing of Electrical Components Final Rule for Mandatory Reporting of Greenhouse Gases Revised November 2011


TECHNICAL SUPPORT DOCUMENT FOR
EMISSIONS FROM ELECTRIC EQUIPMENT
MANUFACTURE OR REFURBISHMENT AND
MANUFACTURING OF ELECTRICAL
COMPONENTS

FINAL RULE FOR MANDATORY REPORTING
OF GREENHOUSE GASES

REVISED NOVEMBER 2011

Office of Air and Radiation
U.S. Environmental Protection Agency

-------
Contents

1.	Source Description	3

a.	Total U.S. Emissions	3

b.	Emissions to be Reported	4

2.	Options for Reporting Threshold	4

a. Data Analysis Supporting Proposal	4

3.	Options for Monitoring Methods	5

a.	IPCC Tier 1 Approach	5

b.	IPCC Tier 2 Approach	5

c.	IPCC Tier 3 Approach	5

4.	Issues Related to the Transfer of Equipment from Equipment Manufacturers to Equipment Users	12

5.	Review of Existing Relevant Reporting Programs/Methodologies	13

6.	Procedures for Estimating Missing Data	13

7.	Q A/QC Requirements	14

8.	Reporting Procedures	15

9.	References	16

2

-------
1. Source Description

Sulfur hexafluoride (SF6) is most commonly used in the electrical power industry where it acts as an insulator and
interrupter in equipment that transmits and distributes electricity (RAND 2004). The electric power industry in the
United States has used SF6 gas since the 1950s because of its superior dielectric strength and arc-quenching
characteristics. SF6 has replaced flammable insulating oils in many applications and allows for more compact
substations in dense urban areas. Currently, there are no available or foreseen substitutes for SF6.

High-voltage circuit breakers account for the majority of SF6 use in the United States. Other types of electrical
equipment that use SF6 include gas-insulated substations, switches (other than circuit breakers), high-voltage
transmission lines, and high-voltage bushings.

Original equipment manufacturers (OEMs) purchase bulk SF6 gas to:

¦ install a shipping charge in high-voltage closed-pressure equipment;1

¦ ship alongside closed-pressure equipment for use at electric power system sites;

¦ fill sealed-pressure equipment with its intended lifetime supply of SF6;2 and

¦ test equipment.

SF6 emissions from OEMs typically occur during the testing, manufacturing, and installation or commissioning of
equipment but can also occur when equipment is decommissioned at a manufacturing facility.

Perfluorocarbons (PFCs) are sometimes used as dielectrics and heat transfer fluids in power transformers. PFCs are
also used for retrofitting CFC-113 cooled transformers. One PFC used in this application is perfluorohexane
(CY,Fm). In terms of both absolute and carbon-weighted emissions, PFC emissions from electrical equipment are
generally believed to be much smaller than SF6 emissions from electrical equipment; however, there may be some
exceptions to this pattern (IPCC, 2006). Throughout this Technical Support Document, "SF6" will be used to denote
SF6 and/or PFCs.

a. Total U.S. Emissions

Emissions of SF6 from OEMs in the United States were estimated to be 0.8 Tg C02 Eq. in 2006 (EPA 2010).

The 1990 to 2006 emission estimates for OEMs were derived by assuming that manufacturing emissions equal 10
percent of the quantity of SF6 charged into new equipment. The quantity of SF6 charged into new equipment from
1990 to 2000 was estimated based on statistics compiled by the National Electrical Manufacturers Association
(NEMA). The quantities of SF6 charged into new equipment for 2001 to 2006 were estimated using data reported by
participants in EPA's SF6 Emission Reduction Partnership for Electric Power Systems along with EPA's estimate of
the total industry SF6 nameplate capacity (128.4 Tg C02 Eq. in 2006). Specifically, EPA calculated the ratio of new
nameplate capacity to total nameplate capacity of a subset of Partners for which new nameplate capacity data was
available from 1999 to 2006. EPA then multiplied this ratio by the total industry nameplate capacity estimate to
derive the amount of SF6 charged into new equipment for the entire industry. The 10 percent emission rate is the
average of the "ideal" and "realistic" manufacturing emission rates (4 percent and 17 percent, respectively)
identified in a paper prepared under the auspices of the International Council on Large Electric Systems (CIGRE) in
February 2002 (O'Connell et al. 2002). This method for estimating OEM emissions is the same method used in
EPA's Inventory of Greenhouse Gas Emissions and Sinks: 1990-2006 (EPA 2008).

1 Closed-pressure equipment requires refilling (topping up) gas during its lifetime and generally contains between
five and several hundred kilograms of SF6 per unit (IPCC 2006).

2 Sealed-pressure equipment should not require any refilling (topping up) with gas during its lifetime and generally
contains less than 5 kilograms of SF6 per unit (IPCC 2006).

-------
b. Emissions to be Reported

EPA is requiring facilities to report all SF6 and PFC emissions from testing and manufacturing of new
equipment, from refurbishing and decommissioning and disposal of previously manufactured equipment, from
storage cylinders and other containers, and from the installation of new equipment unless the title of the
equipment has transferred to the electric power system facility.

Additionally, EPA is requiring facilities to report other source categories at the facility for which calculation
methods are provided in the rule, as applicable. For example, facilities must report C02, CH4, and N20
combustion emissions from stationary combustion units by following the requirements of 40 CFR part 98,
subpart C (General Stationary Fuel Combustion Sources).

2. Options for Reporting Threshold

EPA evaluated a range of emission and consumption-based thresholds for electrical equipment manufacturing. The
emission-based thresholds considered were 1,000; 10,000; 25,000; and 100,000 mt C02 Eq. These thresholds
translate to consumption-based thresholds of 922; 9,220; 23,061, and 92,244 lbs of SF6, respectively, assuming an
average manufacture emission rate of 10%.

As shown in Table 1 below, EPA estimates that all ten domestic facilities identified as consumers of SF6 will fall
above the 1,000; 10,000; and 25,000 mt C02 Eq. thresholds and above the 922; 9,220; 23,061 lbs. of SF6 thresholds,
while only the largest five facilities would be captured by the 100,000 mt C02 Eq./92,244 lbs SF6 threshold.
However, while EPA attempted to accurately estimate SF6 consumption per facility, it is conceivable that OEMs
specializing in the production of relatively low-voltage switchgear for the distribution market will fall below both
the consumption and emission threshold.

Additional details on the number of facilities and total emissions that would be captured by each threshold are
shown below in Table 1.

Table 1. Emission Threshold Summary

1-'mission Threshold (Ml CO* K.(|)

1.000

10.000

25.000

100.000

C oiisiimplion Threshold (lbs. ol'SI",,)

>)22

y.220

23.0(il

'>2.244

Number of Facilities Above

Percent of Facilities Above

100%

50%

Total Emissions of Facilities Above (Mt C02 Eq)

814,128

569,890

Percent of Emissions Above

100%

70%

a. Data Analysis to Determine Reporting Threshold

The consumption-based threshold offers an important advantage relative to the emission-based threshold because it
permits OEMs to quickly determine whether they are covered. It also avoids the scenario in which sources drop in
and out of the program based on fluctuating emissions.

EPA performed a threshold analysis to determine how many OEMs would fall above and below the four different
consumption thresholds considered. The analysis was performed by:

¦ Using the Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2006 (EPA 2008) to determine the
estimated amount of SF6 gas purchased by all OEMs, which was estimated as 750,981 lbs in 2006.

¦ Converting the 750,981 lbs of SF6 to metric tons of C02 equivalent, which is 8,141,281 Mt C02 Eq.

-------
¦ Estimating total U.S. emissions by multiplying total purchases by the 10% emission rate, which is 814,128
Mt C02 Eq.

¦ Identifying the ten OEMs in the U.S. that are responsible for most of the SF6 consumption in the
manufacturing sector.

¦ Dispersing the total amount of U. S. OEM emissions of SF6 among each of the ten OEMs based on the
estimated market-share of each OEM.

3. Options for Monitoring Methods

For electrical equipment manufacturers, emissions can occur during equipment testing, manufacturing, installation,
and decommissioning and refurbishing. EPA evaluated a range of options for estimating emissions that result from
these activities. The three primary options reviewed were the Tier 1, Tier 2, and Tier 3 IPCC methods presented in
the 2006 IPCC Guidelines for National Greenhouse Gas Inventories (IPCC 2006). Each of these options is
described below.

a. IPCC Tier 1 Approach

The Tier 1 method, referred to as the default emission factor method, estimates emissions by multiplying the
appropriate default regional emission factor by the amount of SF6 consumed by each equipment manufacturer.

Tier 1 emission factors are usually developed through industry research. However, due to a lack of data availability,
IPCC does not provide a default emission factor for the United States. The default emission factors for emissions
associated with the manufacture of closed-pressure electrical equipment in Europe and Japan are 8.5% and 29%,
respectively.

b. IPCC Tier 2 Approach

The Tier 2 method, referred to as the country-specific emission factor method, also uses emission factors. But these
emission factors are more accurate than the Tier 1 factors because they are developed using criteria specific to the
country where the manufacturer is located and are usually developed by analyzing actual pure-mass balance
emissions that were calculated at the life-cycle level.

While the Tier 1 and Tier 2 approaches are relatively simple, they are likely to result in inaccurate facility emissions
estimates because they do not take into account the variation in SF6 handling and management practices among
OEMs as well as variation in production methods associated with the different types of electrical equipment
produced. Furthermore, due to lack of data availability, IPCC does not provide emission factors for electrical
equipment manufacturing in the United States.

c. IPCC Tier 3 Approach

The Tier 3 method measures SF6 emissions using mass-balance equations for each life-cycle stage of the SF6 use.
The equations below are based on the Tier 3 mass-balance equations for equipment manufacturing emissions and
equipment installation emissions.

The Tier 3 equation for estimating emissions from equipment manufacturing emissions, which includes equipment
testing, equipment manufacturing, and equipment decommissioning and refurbishing is the following:

Equipment Manufacturing Emissions = Decrease in SF6 Inventory + Acquisitions of SF6 -

Disbursements of SF6

Where:

-------
Decrease in SF6 Inventory = SF6 stored in containers at the beginning of the year - SF6 stored in
containers at the end of the year

Acquisitions of SF6 = SF6 purchased from chemical producers or distributors in bulk + SF6 returned
by equipment users + SF6 returned to site after offsite recycling

Disbursements ofSF6= SF6 contained in new equipment delivered to customers + SF6delivered to
equipment users in containers + SF6 returned to suppliers + SF6 sent off-site for recycling + SF6
sent off-site for destruction destroyed

In addition, emissions may occur when the manufacturer fills the equipment off-site from the manufacturing facility,
before transferring custody to the equipment user. These emissions, along with other issues related to the transfer of
equipment from manufacturers to users, are discussed further below.

To quantitatively determine disbursements of SF6 to customers in new equipment (or cylinders), four options were
considered:

¦ Option 1: Disbursements could be estimated by weighing containers before and after gas from the
containers was used to fill equipment or cylinders.

¦ Option 2: Disbursements could be estimated by using flow meters to measure the amount of gas used to fill
equipment or cylinders.

¦ Option 3: Disbursements could be estimated by assuming that the mass of SF6 or PFCs disbursed to
customers in equipment is equal to the nameplate capacity of the equipment or, where the equipment is
shipped with a partial charge, equal to the nameplate capacity of the equipment times the ratio of the
densities of the partial charge and the full charge.

¦ Option 4: Disbursements could be estimated by weighing the equipment filled with SF6 or the PFC from
the container before and after filling. The tare weight of the equipment would then be subtracted from the
weight of the filled equipment to determine the weight of the gas in the equipment, and therefore, the
weight of the actual disbursement.

These options and their advantages and disadvantages are discussed, in turn, below.

¦ Option 1 appears to be a practice currently employed at some manufacturing facilities. A digital scale is
used to weigh the containers storing the gas before and after the equipment is filled (MEPPI 2010). In this
regard, this option does not present new costs or changes to current procedures (at least for some facilities).
However, EPA recognizes that emissions can occur downstream of the container to the equipment being
filled. These emissions can occur through leaks in the hose or line connecting the container to the
equipment, during coupling and decoupling activities, as well as through general mishandling (e.g., a faulty
connection). To accurately estimate disbursements, these emissions should be estimated separately and
subtracted from the disbursement total.

Emissions could be calculated by the following method:

i) Determine annual disbursements by summing individual disbursements as summarized in the
equation below. Individual disbursements are denoted as "Qp" and are summed to determine
annual disbursements.

Da„o=ZQr

P=i

Where:

-------
D; H |;; = The annual disbursement of SF6 or PFCs sent to customers in new equipment or

cylinders or sent off-site for other purposes including for recycling, for destruction or
to be returned to suppliers.

Qp = The mass of the SF6 or PFCs charged into equipment or containers over the period p
sent to customers or sent off-site for other purposes including for recycling, for
destruction or to be returned to suppliers.

n = The number of periods in the year.

ii) Determine Qp, the mass of the SF6 or PFCs charged into equipment or containers over the period
p sent to customers or sent off-site for other purposes including for recycling, for destruction or to
be returned to suppliers, by weighing containers before and after gas from containers is used to fill
equipment or cylinders and subtracting emissions that occur during the filling. This calculation is
summarized in the equation below.

Q=Mb-Me-El

Where:

Qp = The mass of SF6 or the PFC charged into equipment or containers over the period p
sent to customers or sent off-site for other purposes including for recycling, for
destruction or to be returned to suppliers.

Mb = The mass of the contents of the containers used to fill equipment or cylinders at the
beginning of period p.

Me = The mass of the contents of the containers used to fill equipment or cylinders at the end
of period p.

El = The mass of SF6 or the PFC emitted during the period p downstream of the containers
used to fill equipment or cylinders and in cases where a flowmeter is used, downstream
of the flowmeter during the period p (e.g., emissions from hoses or other flow lines that
connect the container to the equipment that is being filled).

iii) Determine the mass of SF6 or PFC emitted downstream of the containers used to fill equipment or
cylinders. These emissions could originate from hoses or other flow lines that connect the
container to the equipment or cylinder that is being filled. This calculation is summarized in the
equation below. To estimate losses from filling events, sum the emissions losses for each of the
different valve-hose combinations during the period p.

El = ^ Fa x EFa

Z=1

Where:

El = The mass of SF6 or the PFC emitted during the period p downstream of the

containers used to fill equipment or cylinders and in cases where a flowmeter is
used, downstream of the flowmeter during the period p (e.g., emissions from hoses
or other flow lines that connect the container to the equipment that is being filled).

FC1 = The total number of fill operations over the period p for the valve-hose combination.

EFC1 = The emission factor for the valve-hose combination.

n = The number of different valve-hose combinations used during the period p.

The emission factor for a given valve-hose combination could be estimated using measurements and/or
engineering assessments or calculations based on chemical engineering principles or physical or chemical
laws or properties. Such assessments or calculations could be based on, as applicable, the internal volume
of the hose or line that is open to the atmosphere during coupling and decoupling activities, the internal

-------
pressure of the hose or line, the time the hose or line is open to the atmosphere during coupling and
decoupling activities, the frequency with which the hose or line is purged and the flow rate during purges.
Such methods could also include the use of leak detection methods (e.g., EPA Method 21 and the Protocol
for Equipment Leak Emission Estimates) to determine a loss factor appropriate to calculate emissions;
however, bagging techniques as described in these methods may introduce feasibility concerns (e.g.,
user/operator error, bag damage) (Dilo 2010a). Unexpected or accidental emissions from the filling lines or
hoses should be included in the total. EPA's understanding is that electrical equipment is at a vacuum and
is sealed prior to being filled with SF6 or PFCs (MEPPI 2010, NEMA Task Force 2001a). However, if any
air or nitrogen is in the equipment and is purged during the filling process, then the method should also
account for SF6 and PFC emissions that occur during such purging.

Facilities could calculate the emission factor or use an industry-developed value for each combination of
hose and valve fitting. No such values have been established by the industry to-date; however, valve
manufacturers, in collaboration with equipment manufacturers could establish industry-wide factors. EPA
inquired with one company that manufactures valves about developing such values (Dilo 2010a). Their
calculations, for two standard fitting sizes—an 8 MM (DN8) and 20 MM (DN20), indicate the following:

Under normal breaker pressure (75 PSI) and at 20 degrees Celsius:

¦ 8 MM valve -80 milligrams; and

¦ 20 MM valve -265 milligrams (Dilo 2011).3

¦ Option 2 presents the same issue as in option 1 in that emissions downstream of the containers would be
required to be estimated separately and subtracted from the disbursement total. One disadvantage in this
approach is that flowmeters may not already be used by the OEM, in which case, the OEM would need to
purchase these devices (Dilo 2010a). OEMs commonly rely only on the pressure rise to monitor the transfer
of the gas from a container to a breaker or other piece of equipment (MEPPI 2010).

Emissions could be calculated by the following method:

Da„a=Y.Q,

p=i

Where:

Dm,. ! = The annual disbursement of SF6 or PFCs sent to customers in new equipment or

cylinders or sent off-site for other purposes including for recycling, for destruction or
to be returned to suppliers.

n = The number of periods in the year.

3 In a November 2011 revision to this Technical Support Document, EPA updated the value for the 8 MM valve
from 10 grams to 80 milligrams and the value for 20 MM valve from 50 grams to 235 milligrams.

-------
QP=Mmr-EL

Where:

Qp = The mass of SF6or the PFC charged into equipment or containers over the period p
sent to customers or sent off-site for other purposes including for recycling, for
destruction or to be returned to suppliers.

Mml = The mass of the SF6 or the PFC that has flowed through the flowmeter during the
period p.

El = The mass of SF6 or the PFC emitted during the period p downstream of the containers
used to fill equipment or cylinders and in cases where a flowmeter is used,
downstream of the flowmeter during the period p (e.g., emissions from hoses or other
flow lines that connect the container to the equipment that is being filled).

iii) Determine the mass of SF6 or PFC emitted downstream of the containers used to fill equipment or
cylinders. These emissions could originate from hoses or other flow lines that connect the
container to the equipment or cylinder that is being filled. This calculation uses the same equation
as in option 1, and is shown again below. To estimate losses from filling events, sum the
emissions losses for each of the different valve-hose combinations during the period p.

El = ^ Fa x EFa

Z=1

Where:

El = The mass of SF6 or the PFC emitted during the period p downstream of the containers
used to fill equipment or cylinders

FC1 = The total number of fill operations over the period p for the valve-hose combination.

EFC1 = The emission factor for the valve-hose combination.

n = The number of different valve-hose combinations used during the period p.

As explained in Option 1, above, facilities could calculate the emission factor or use an industry-developed
value for each combination of hose and valve fitting. No such values have been established by the industry
to-date; however, valve manufacturers, in collaboration with equipment manufacturers could establish
industry-wide factors.

Option 3 equates the mass of SF6 or PFCs disbursed to customers in equipment to the nameplate capacity
of the equipment or, where the equipment is shipped with a partial charge, equal to the nameplate capacity
of the equipment times the ratio of the densities of the partial charge and the full charge. The nameplate
capacity could be based on the manufacturer's current estimate (i.e., the one that appears on the nameplate)
or it could be measured. One disadvantage of assuming that the mass of SF6 or PFCs disbursed to
customers in equipment is equal to the current estimate of the nameplate capacity is that the current
estimate may never have been measured with good precision and accuracy. Even if the nameplate capacity
has previously been carefully measured, the internal volume of the equipment or density to which the
equipment is charged might have changed since the original measurement. Because the mass-balance
approach requires precise inputs, inaccuracies of even two or three percent could lead to very large
inaccuracies in the facility's emissions estimate.

One way of developing a more precise estimate of the nameplate capacity of equipment would be to fill the
equipment with a fluid and then to carefully recover the fluid, measuring what was recovered. This fluid
could be SF6, another gas, or a liquid. If SF6 was used, the equipment would be charged to its operational
or shipping SF6 density using the facility's usual methods and then emptied. The mass of the SF6
recovered, adjusted slightly for the residual pressure of the SF6 that would remain in the equipment even at
a deep vacuum, could be equated to the full or shipping charge, as applicable. One advantage of this

-------
approach is that it would reflect the actual SF6 charging practices of the facility; one disadvantage is that it
could result in small SF6 emissions during the charging and recovery steps.

If a liquid was used, the equipment would be filled carefully, ensuring that the full volume was filled, and
then emptied. The volume of the liquid recovered would be equated to the internal volume of the
equipment. The temperature of the liquid would need to be kept constant throughout this exercise to obtain
an accurate measurement of the volume. This volume multiplied by the SF6 density at the full charge would
yield the nameplate capacity of the equipment.

To account for variability, a certain number of these measurements would need to be performed to develop
a robust and representative average nameplate capacity (or shipping charge) for each make and model.
Equipment samples should be selected so as to reflect predictable variability in the facility's filling
practices and conditions.4 In addition, within a particular set of conditions, equipment should be selected at
random. The specific number of samples would depend on the variability of the nameplate capacity within
each make and model and on the desired level of precision of the estimated mean nameplate capacity of
each make and model. As discussed below, the desired level of precision of the mean would depend on the
level of precision of the emissions estimate as a whole.

A Student T distribution calculation is the appropriate statistical procedure for determining when additional
samples should be taken to achieve a tolerable error for the average nameplate capacity of a certain make
and model of SF6-containing electrical equipment.5 The calculation 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. In principle, the mean of the samples taken approaches the true mean as more
samples are taken; this procedure determines how many samples are needed as a function of the relative
standard deviation of the sample measurements. The following matrix demonstrates this concept using a
95 percent confidence level and an assumed initial sample size of ten. On the left is the relative standard
deviation of the sample population; an initial sample size of ten was chosen for this computation. For
example, an equipment manufacturer conducts a minimum of ten tests to determine the nominal nameplate
capacity of ten samples from a given make and model of a high voltage circuit breaker. In this hypothetical
example, the resulting standard deviation of those samples is 1.5 percent. If the tolerable error is one
percent of the true mean, the manufacturer must increase its sample size to twelve, i.e., conduct two more
tests, according to the chart below. If the standard deviation of the sample population tested was larger, the
number of samples that would need to be taken would also increase.7

4 For example, if equipment is filled to a particular pressure, and the temperature of the facility varies during the day or
seasonally, then the density to which the equipment is filled will vary as well (with higher densities at lower temperatures). In
this case, samples should be taken to reflect the typical range of temperatures at which equipment is filled.

5 See the Technical Support Document for Fluorinated Gas Production for more discussion of the Student's t test.

6 Table 2 is provided as an example; a manufacturer would apply the Student T distribution to determine the appropriate number
of samples based on the variability of the quantity of gas charged into equipment of a given make and model.

7 It is important to note that the reliability of the mean nameplate capacity estimate depends on the accuracy and precision of the
individual measurements as well as on the number of times they are repeated. Accuracy is a measure of the bias (systematic
error) of a measurement; precision is a measure of the random error. The precision includes the random errors from the
measuring devices such as the scales used and the variation in the nameplate capacity for a single make and model. Repeated
measurements using the same measurement device improve the precision of the mean, but they do not improve accuracy because
each measurement has the same systematic error. For example, if the scale used to weigh the SF6 shipping charges is only
guaranteed to be accurate within one percent, and it happens to be biased upward by 0.75 percent, the mean of even 2,000 weight
measurements of the same device will be biased upward by 0.75 percent.

-------
Table 2. Example Lookup Table of Total Sample Size required to

meet Precisions, given Initial Sample Size of 10







Precision

















'J



0.25%

0.50%

1.00%

5.00%

E

0.5%

18

10

10

10

/

1.0%

64

18

10

10

y

1.5%

141

38

12

10



2.0%

249

64

18

10

i

2.5%

389

99

27

10

i

3.0%

560

141

38

10

=
/.

3.5%

763

191

50

10

o

4.0%

996

249

64

10

n

4.5%

1260

315

81

10



5.0%

1556

389

99

10

The actual precision with which the mean nameplate capacity would need to be measured would depend on
(1) the desired level of precision for the emissions estimate for the facility, (2) the precision with which the
other quantities in the mass-balance equation were monitored, (3) the share of the facility's SF6
disbursements accounted for by a particular make and model, (4) the number of makes and models whose
nameplate capacities were measured,8 and (5) the level of emissions. Example calculations indicate that a
facility that measured gas acquisitions with a precision and accuracy of one percent and that had an
emission rate of 10 percent might need to estimate the mean nameplate capacity (or shipping charge) to a
precision of two percent to achieve a precision in its emissions estimate of about 20 percent (expressed as a
95-percent confidence interval). A facility that measured gas acquisitions with a precision and accuracy of
one percent and that had an emission rate of 5 percent might need to estimate the mean nameplate capacity
to a precision of one half of one percent to achieve a precision in its emission estimate of about 25 percent.
(The accuracy of the device used for the nameplate capacity measurement would also need to be one half of
one percent or better.)

For other sources (e.g., fhiorinated GHG production processes using the mass-balance approach), EPA is
proposing a maximum error of 30 percent. However, in general, the tolerable level of error in an estimate
depends in part on the size of the emissions. For smaller emission sources, errors larger than 30 percent
might be tolerable.

To reflect subtle changes in manufacturing methods and conditions over time, it may be appropriate to
require re-measurement of nameplate capacities at some interval, e.g., every five or ten years.

According to industry, equipment manufacturers are filling breakers with SF6 gas to a full charge as part of
their testing procedures; the equipment is then evacuated and filled to a partial charge for shipment at a
later stage in the process (MEPPI 2010). Therefore, an advantage to Option 3 is that it appears that it could
be integrated into normal facility procedures without introducing any significant filling and recovering
related activities. However, if the test charge were used to estimate the nameplate capacity, equipment
manufacturers would need to account for possible variations between the density to which the equipment
was charged during the test and that to which it was charged for shipping.

¦ Lastly, Option 4 would require the weighing of the equipment filled with SF6 or the PFC before and after
the equipment is charged. The tare weight of the equipment would then be subtracted from the weight of
the filled equipment to determine the weight of the gas in the equipment, and therefore, the weight of the
actual disbursement. Although, in theory this option is advantageous because it shifts the emphasis to the

8 If the nameplate capacities of multiple models were measured independently, using multiple measurement devices, then the
"error" (more precisely, the relative standard deviation) of their summed or averaged nameplate capacities would be smaller than
the errors of their individual nameplate capacities. However, if the nameplate capacities were measured using the same
measuring device (with the same systematic error), then the error of the summed nameplate capacities would not be reduced
below the systematic error of the individual nameplate capacity measurements.

11

-------
equipment that is being disbursed (rather than weighing the containers at the intermediary step), this option
is not practical for two reasons. First, the mass of the SF6 or PFC charged into the equipment is likely to be
low relative to the mass of the equipment; thus, it may be difficult to obtain a precise measurement of the
mass of the SF6 or PFC using this method (i.e., within 1 percent) even if the scale is precise and accurate to
within 1 percent of full scale. Second, it would be very difficult and impractical to weigh the shipping unit,
which can be large and difficult to maneuver onto a scale (MEPPI 2010).

Based upon a close examination of these options and review of comments received during the public comment
period, EPA decided to require either Options 1 or 2. Additionally, to increase flexibility, EPA also incorporated
Option 3 into the final rule. For this option, the number of measurements that is required for each make and model
to determine a sufficiently precise estimate of shipping charge or nameplate capacity must be calculated to achieve a
precision of one percent of the true mean, using a 95 percent confidence interval.

4. Issues Related to the Transfer of Equipment from Equipment Manufacturers to
Equipment Users

As noted above, emissions may occur when the manufacturer fills the equipment off-site from the manufacturing
facility, before transferring custody to the equipment user. Such emissions could be estimated using the following
equation:

EI = Mf +Mc — Ni

Where:

EI = Total annual SF6 or PFC emissions from equipment installation at electric transmission or
distribution facilities.

Mf = The total annual mass of the SF6 or PFCs used to fill equipment off-site from the OEM facility.
Mc = The total annual mass of the SF6 or PFCs used to charge the equipment prior to leaving the

electrical equipment manufacturer facility.

Ni = The total annual nameplate capacity of the equipment installed at electric transmission or
distribution facilities.

These emissions could be attributed to the equipment manufacturer. Through EPA's SF6 Emission Reduction
Partnership for Electric Power Systems, a conference call was organized between representatives from the National
Electrical Manufacturers Association (NEMA) and EPS Partners. Participants agreed that the responsibility of
installation emissions varies and discussed points at which the reporting requirement for installation emissions
would transfer from the manufacturer to the user (NEMA representatives 2010b). The final rule delineates the
reporting boundary between the electric equipment manufacturer under this subpart and the electric transmission or
distribution facility (under subpart DD) with respect to emissions during equipment installation. The final rule
specifies that the responsibility of reporting emissions from installation practices is dependent upon the point at
which the title is transferred to the electric power transmission or distribution facility by the electrical equipment
manufacturer or third-party contracted by the manufacturer. The OEM must estimate and report emissions from
equipment installation using the equipment installation mass balance equation if:

1) the electrical equipment manufacturer holds the title; or

2) the title temporarily passes from the OEM to a third party contractor.

In addition, any emissions from equipment that was filled (partially or completely) at the OEM's facility but whose
charge leaked out before being delivered to and installed for the customer should also be the manufacturer's
responsibility to report.

EPA understands that in some cases, manufacturers may exceed the nameplate capacity of equipment when charging
it, either intentionally, to postpone the re-fill of the equipment in the event that the equipment develops a leak, or
unintentionally due to adiabatic cooling that occurs during filling (NEMA Representatives 2010a). If there is an

-------
overcharge, then the actual initial charge of the equipment should be conveyed clearly to the equipment user. If it is
not, the user will underestimate emissions. The underestimate will occur because the user will underestimate the
quantity of gas "purchased with or inside of equipment." Effectively, an extra supply of gas is hidden inside the
equipment rather than conveyed alongside it in a container.

For example, suppose that a piece of equipment is charged to 105 percent of its proper charge (nameplate capacity)
when it is first installed, but that its initial charge is recorded as 100 percent of the proper charge. Suppose further
that the equipment slowly leaks over its lifetime so that it has 90 percent of its proper charge left upon retirement.
When the equipment is retired, only the difference between the proper charge (100 percent) and the actual charge
(90 percent) will be noticed. The loss of the extra five percent will not be recorded because the initial gain of that
five percent was not recorded. In the event that the equipment never leaks during its lifetime and is retired with 105
percent of its charge, the five-percent overcharge, when recovered along with the rest of the SF6 inside the
equipment, will register as an addition to the SF6 in storage that is five percent above the retired nameplate capacity.
Again, this will result in emissions being underestimated by the quantity of the overcharge.

To avoid the underestimate, electrical equipment manufacturers would need to provide equipment users with an
accurate estimate of the quantity of SF6 actually charged into newly commissioned equipment. The user would then
record this as additional gas "purchased with or inside of equipment." Using the standard mass-balance approach,
this extra gas would register as an emission during the first year. This is because the extra gas would be recorded as
part of the total amount of gas purchased during the year, but would not be canceled out by a corresponding increase
to cylinder inventory (since the extra gas is inside the equipment) or by the increase to nameplate capacity (since the
extra gas is by definition the amount of gas remaining once the nameplate capacity of the equipment is subtracted
from the amount of gas actually inside the equipment). Since it would not be accounted for as an addition to
inventory or as part of the new nameplate capacity, it would seem to have been emitted or used to replace emitted
gas. However, this apparent loss would be made up for when the equipment was retired, as discussed above.

An alternative approach to tracking the overcharge would be to record it as stored inside the particular piece(s) of
equipment that is (are) overcharged. Effectively, this extra charge would be tracked as if it were stored inside a
cylinder that happened to be attached to the equipment. The equipment would be checked annually to see whether
and how much of its charge had leaked during the year; if the overcharge had not leaked out, then the overcharge
would still be considered to be in storage and would not be recorded as an emission. Although this approach would
avoid indicating an emission before any occurred, the need for annual checks of overcharged equipment would make
it relatively burdensome to carry out.

Given the potential for an overcharge as well as general data accuracy of inputs required to estimate emissions of
SF6 from electrical equipment use, EPA decided to require in the final rule that the quantity of gas charged into
delivered equipment and added during installation by the manufacturer be certified by the manufacturer and
expressed in pounds of SF6 or PFC. Electrical equipment manufacturers must keep records of certifications of the
quantity of gas, in pounds, charged into equipment at the electrical equipment manufacturer or refurbishment facility
as well as the actual quantity of gas, in pounds, charged into equipment at installation.

5. Review of Existing Relevant Reporting Programs/Methodologies

In addition to the 2006IPCC Guidelines, EPA also reviewed the protocols and guidance in the Inventory of U.S.
Greenhouse Gas Emissions and Sinks, the Technical Guidelines for the Voluntary Reporting of Greenhouse Gases
(1605(b)) Program, EPA's Climate Leaders Program, and The Climate Registry. These protocols and guidance
coalesce around the IPCC 2006 Tier 3 guidelines.

6. Procedures for Estimating Missing Data

It is expected that equipment manufacturers should be able to obtain 100 percent of the data needed to perform the
mass balance calculations for both SF6 and PFCs. The use of the mass-balance approach requires measured values
for all inputs. However, if needed, replace missing data using data from similar manufacturing operations, and from
similar equipment testing and decommissioning activities for which data are available.

-------
7.

QA/QC Requirements

QA/QC methods for reviewing completeness and accuracy of reporting include the following:

¦ Review inputs to the mass balance equation to ensure inputs and outputs to the manufacturer's system are
all accounted for in all appropriate sections.

¦ Ensure no negative inputs are entered and negative emissions are not calculated. However, the change in
storage inventory may be calculated as a negative number.

¦ Ensure that beginning of year inventory matches end of year inventory from previous year.

¦ Ensure that in addition to SF6 purchased from bulk gas distributors, SF6 returned from utilities and received
from offsite recycling are properly accounted as additions to inventory.

QA/QC methods should be employed throughout the year. Important checks/procedures include the following:

¦ Ensure that cylinders returned to the vendor are weighed in a consistent manner.

¦ Ensure that gas suppliers measure the amount of gas remaining in cylinders/tanks returned (residual gas).

¦ Adopt practices such as tracking cylinders leaving and entering storage with check-out sheets and weigh-in
procedures.

a. Analysis to Determine Scale Accuracy Requirements

A ±1 percent (of true weight) relative accuracy requirement for scales was originally proposed; however, based on
comments EPA received during the public comment period indicating that the proposed requirement was too
stringent, EPA reexamined the appropriate level of accuracy and precision for scales used to weigh cylinders.

The first steps undertaken by EPA to reassess scale accuracy requirements were to research scale manufacturer Web
sites and to contact scale manufacturers and electrical equipment users to better understand what scales are available
on the market and the typical specifications of scales designed to weigh cylinders. 9

Subsequently, EPA performed a sensitivity analysis using a variety of scale accuracies to analyze what effect
changes in scale accuracies would have on the relative uncertainty of emission estimates. The analysis was
performed using assumptions for two hypothetical electrical equipment manufacturers—a switch manufacturer
manufacturing low and medium voltage equipment ("OEM 1" in Table 3, below) and a high voltage electrical
equipment manufacturer manufacturing 138 kV circuit breakers ("OEM 2" in Table 3, below). Since the price of
scales tends to increase as scale accuracy increases (all else being equal), EPA's goal was to determine which scale
accuracy requirement would result in the least cost burden while still providing emission estimates with reasonable
uncertainty levels. The summary results from the sensitivity analysis are provided in Table 3 below.10

Table 3. Relative Uncertainties of Emission Estimates for Various Scale Accuracies (95% Confidence Interval)

Level of accuracy applied to mass-balance inputs

OEM 1

OEM 2

Average

± 1% (relative)

± 5% (relative)

42%

24%

± 1 pound (absolute)

± 2 pound (absolute)

16%

± 1% of Ml scale (absolute)3

26%

15%

aAssuming full scale is equivalent to a scale capacity of 330 pounds.

EPA concluded that the incremental increase in relative uncertainty from a scale accuracy requirement of ± 1

9 Documentation of the internet research and correspondence with scale manufacturers can be found in the docket for this rule
(Docket ID No. EPA-HQ-OAR-2009-0927).

10 The analysis, in its entirety, is provided in the docket for this rule (Docket ID No. EPA-HQ-OAR-2009-0927).

-------
percent of true mass or weight to ± 2 pounds absolute scale accuracy was not enough to justify maintaining the
proposed accuracy of ± 1 percent of true mass or weight and its associated burden. Additionally, EPA understands
that some electrical equipment manufacturers draw SF6 from large vessels such as ISO containers to fill equipment
at their facilities (Dilo 2010b). After reviewing the results of the sensitivity analysis as well as public comments
submitted by electrical equipment manufacturers and information received from industry correspondence, EPA
eased the proposed requirement as well as differentiated between flowmeters and scales as follows:

•	For flowmeters, the final rule requires a scale accuracy of ±1 percent of full scale;

•	For scales, the final rule requires a ±1 percent of the maximum weight of the containers (i.e., gas plus tare)
typically weighed on the scale.11

This final requirement was designed taking into consideration that some electrical equipment manufacturers use
large containers to fill equipment. A requirement such as ±2 pounds absolute scale accuracy would be exceedingly
stringent when applied to scales with very large capacities used to weigh very large containers. EPA believes that
the final accuracy requirements will lower the burden on reporters without significant compromise to data quality.

8. Reporting Procedures

The following supplemental data would aid in verifying facilities' emissions estimates, in pounds unless otherwise
specified:

¦	SF6 and PFC stored in containers at the beginning and end of the year

¦	SF6 and PFCs sent off-site for destruction;

¦	SF6 and PFCs sent off-site to be recycled;

¦	SF6 and PFCs purchased in bulk, in pounds;

¦	SF6 and PFCs returned by equipment users with or inside equipment;

¦	SF6 and PFCs returned from off site after recycling

¦	SF6 and PFCs stored in containers at the beginning and end of the year;

¦	SF6 and PFCs inside new equipment delivered to customers;

¦	SF6 and PFCs inside containers delivered to customers;

¦	SF6 and PFCs returned to suppliers;

¦	The nameplate capacity of the equipment delivered to customers with SF6 or PFCs inside, if different from
the quantity of SF6 and PFCs in equipment delivered to customers inside equipment;

¦	A description of the engineering methods and calculations used to determine emissions from hoses or other
flow lines that connect the container to the equipment that is being filled;

¦	The emission factors used for each hose and valve combination and the associated valve fitting sizes and
hose diameters;

¦	The total number of fill operations for each hose and valve combination used to fill equipment or container
disbursements;

¦	The mean value for each make, model, and group of conditions if the mass of SF6 or the PFC disbursed to
customers in new equipment is determined by assuming that it is equal to the equipment's nameplate
capacity or, in cases where equipment is shipped with a partial charge, equal to its partial shipping charge;

¦	The number of samples and the upper and lower bounds on the 95 percent confidence interval for each
make, model, and group of conditions if the mass of SF6 or the PFC disbursed to customers in new
equipment is determined by assuming that it is equal to the equipment's nameplate capacity or, in cases
where equipment is shipped with a partial charge, equal to its partial shipping charge;

¦	SF6 and PFCs, in pounds, used to fill equipment at off-site electric power transmission or distribution
locations;

¦	SF6 and PFCs, in pounds, used to charge the equipment prior to leaving the electrical equipment
manufacturer or refurbishment facility;

¦	The nameplate capacity of the equipment, in pounds, installed at off-site electric power transmission or
distribution locations used to determine emissions from installation; and

11 For scales that are used to weigh cylinders containing 115 pounds of gas when full, this equates to ±1 percent of the sum of
115 pounds and approximately 120 pounds tare, or slightly more than ±2 pounds absolute accuracy.

15

-------
¦ For any missing data, the reason the data were missing, the parameter for which the data were missing, the
substitute parameters used to estimate emissions in their absence, and the quantity of emissions thereby
estimated.

9. References

Dilo (2011) Personal Communication between Lukas Rothlisberger of Dilo and Sally Rand, U.S. EPA, January
2011.

Dilo (2010a) Personal Communication between Lukas Rothlisberger of Dilo and Mollie Averyt of ICF International,
June 2010. Available in EPA docket EPA-HQ-OAR-2009-0927.

Dilo (2010b) Personal Communication between Lukas Rothlisberger of Dilo and Mollie Averyt of ICF International,
July 2010. Available in EPA docket EPA-HQ-OAR-2009-0927.

EPA (2008) Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2006. U.S. Environmental Protection
Agency, Washington, D.C. April 2008. Available at:
http://www.epa.gov/climatechange/emissions/downloads/08_CR.pdf

IPCC (2006) 2006IPCC Guidelines for National Greenhouse Gas Inventories. The National Greenhouse Gas
Inventories Programme, The Intergovernmental Panel on Climate Change, H.S. Eggleston, L. Buendia, K. Miwa, T
Ngara, and K. Tanabe (eds.). Hayama, Kanagawa, Japan. Available at: http://www.ipcc-
nggip.iges.or.jp/public/2006gl/index.html

MEPPI (2010) Personal Communication between Phil Bolin and Dave Giegel of Mitsubishi Electric Power
Products, Inc. and Mollie Averyt of ICF International, February 22-March 1, 2010. Available in EPA docket EPA-
HQ-OAR-2009-0927..

NEMA representatives (2010a) Meeting notes from the National Electrical Manufacturers Association Ad-Hoc
Task Group Meeting, SF6 May 7, 2010. Arlington Virginia. Available in EPA docket EPA-HQ-OAR-2009-0927..

NEMA representatives (2010b) Conference call between representatives of the National Electrical Manufacturers
Association, Partners of EPA's SF6 Emission Reduction Partnership for Electric Power Systems, and EPA, June 3,
2010. Available in EPA docket EPA-HQ-OAR-2009-0927..

O'Connell, P., F. Heil, J. Henriot, G. Mauthe, H. Morrison, L. Neimeyer, M. Pittroff, R. Probst, J.P. Tailebois
(2002) SF6 in the Electric Industry, Status 2000, Cigre. February 2002. Available at:
http://www.cigre.org/userfiles/publications/ELT_200_7.pdf

RAND (2004) RAND Environmental Science and Policy Center, "Trends in SF6 Sales and End-Use Applications:
1961-2003," Katie D. Smythe. International Conference on SF6 and the Environment: Emission Reduction
Strategies. Scottsdale, AZ. December 1-3, 2004. Available at: http://www.epa.gov/electricpower-
sf6/documents/co nf04_smythe.pdf.

-------