TECHNICAL SUPPORT DOCUMENT FOR
PROCESS EMISSIONS FROM PRIMARY
PRODUCTION OF ALUMINUM:
PROPOSED RULE FOR MANDATORY
REPORTING OF GREENHOUSE GASES
Office of Air and Radiation
U.S. Environmental Protection Agency
February 4, 2009
-------
Contents
1. Source Description 3
a. Total U.S. Emissions 3
b. Emissions to be Reported 3
2. Options for Reporting Threshold 3
3. Options for Monitoring Methods 4
a. Monitoring Methods for PFCs 5
b. Monitoring Methods for CO2 6
4. Procedures for Estimating Missing Data 8
5. QA/QC Requirements 9
6. Reporting Procedures 9
7. References 10
-------
1. Source Description
Aluminum is a light-weight, malleable, and corrosion-resistant metal that is used in many manufactured products,
including aircraft, automobiles, bicycles, and kitchen utensils. As of last reporting, the United States was the fourth
largest producer of primary aluminum, with approximately eight percent of the world total (USGS 2006). The
United States was also a major importer of primary aluminum. The production of primary aluminum—in addition
to consuming large quantities of electricity—results in process-related emissions of carbon dioxide (CO2) and two
perfluorocarbons (PFCs): perfluoromethane (CF4) and perfluoroethane (C2F6). Only these process-related emissions
are considered here.
CO2 is emitted during the aluminum smelting process when alumina (aluminum oxide, A12O3) is reduced to
aluminum using the Hall-Heroult reduction process. The reduction of the alumina occurs through electrolysis in a
molten bath of natural or synthetic cryolite (Na3AlF6). The reduction cells contain a carbon lining that serves as the
cathode. Carbon is also contained in the anode, which can be a carbon mass of paste, coke briquettes, or prebaked
carbon blocks from petroleum coke. During reduction, most of the carbon in the anode is oxidized and released to
the atmosphere as CO2.
In addition to CO2 emissions, the aluminum production industry is also a source of PFC emissions. During the
smelting process, when the alumina ore content of the electrolytic bath falls below critical levels required for
electrolysis, rapid voltage increases occur, which are termed "anode effects." These anode effects cause carbon
from the anode and fluorine from the dissociated molten cryolite bath to combine, thereby producing fugitive
emissions of CF4 and C2F6. For any one smelter, the magnitude of emissions for a given level of production
depends on the frequency and duration of these anode effects. As the frequency and duration of the anode effects
increase, emissions increase. In addition, even at constant levels of production and anode effect minutes, emissions
vary among smelter technologies (e.g., Center-Worked Prebake vs. Side-Worked Prebake) and among individual
smelters using the same smelter technology due to differing operational practices.
a. Total U.S. Emissions
Process emissions of CO2 from the 14 aluminum smelters in the United States were estimated to be 3.9 million
metric tons of CO2 equivalent (MMTCO2e) in 2006. Process emissions of CF4 and C2F6 from aluminum smelters
were estimated to be 2.5 MMTCO2e in 2006. Total greenhouse gas (GHG) emissions from primary aluminum
production in the United States are estimated to be 6.4 MMTCO2e in 2006 (EPA 2008) In 2006, 13 of the 14
aluminum smelters in the United States accounted for the majority of process emissions. The remaining smelter was
shut down for most of 2006, restarting only at the end of that year.
b. Emissions to be Reported
On-site combustion emissions from aluminum production facilities are not addressed within this document; see the
background Technical Support Document for Stationary Combustion (EPA-HQ-OAR-2008-0508-004). This
document addresses process emissions of PFCs and CO2. Process CO2 emissions can come from the following
processes during primary aluminum production:
• Consumption of the anode during electrolysis (for both Prebake and Soderberg cells);
• Anode baking process (for Prebake cells only); and
• Calcining emissions (from coke calcining).
EPA's current understanding is that all prebake smelters in the United States operate their own anode baking
furnaces on site. EPA does not believe that any U.S. smelters operate calcining furnaces on site.
2. Options for Reporting Threshold
EPA evaluated a range of threshold options for primary aluminum production facilities. These included emission-
based thresholds of 1,000, 10,000, 25,000 and 100,000 mtCO2e and capacity-based thresholds equivalent to these.
EPA also evaluated a requirement that all primary aluminum production facilities be required to report.
The capacity thresholds were developed based on IPCC default emission factors and 100 percent capacity
utilization. These are somewhat conservative assumptions, since capacity utilization is often below 100 percent and
-------
emission rates (particularly PFC emission rates) are generally lower than the IPCC defaults. However, both
production and emission rates fluctuate; aluminum smelters sometimes shut down for long periods and then begin
production again. The conservative assumptions would ensure that plants that have a reasonable chance of emitting
more than the threshold quantity were covered.
Table 2-1 illustrates the emissions and facilities that would be covered under these various thresholds.
Table 2-1: Threshold Analysis for Primary Aluminum Production Based on 2006 Emissions and Facility
Production Capacity
Emission
Threshold Level mtCO2e/yr
1,000
10,000
25,000
100,000
Production Capacity
Threshold
mt Al/year
64
640
1,594
6,378
Total National
Emissions
6,403,000
6,403,000
6,403,000
6,403,000
6,403,000
6,403,000
6,403,000
6,403,000
Total
Number of
Facilities
14
14
14
14
14
14
14
14
Emissions Covered
mtCO2e/yr
6,403,000
6,398,000
6,398,000
6,398,000
6,403,000
6,403,000
6,403,000
6,403,000
Percent
100
99
99
99
100
100
100
100
Facilities Covered
Facilities
14
13
13
13
14
14
14
14
Percent
100
93
93
93
100
100
100
100
All smelters that operated throughout 2006 would be covered at all capacity and emissions-based thresholds
considered in this analysis. This consideration supports either a capacity-based threshold or a requirement that all
plants report. A requirement that all plants report would have the additional advantage of simplicity.
3. Options for Monitoring Methods
EPA reviewed a range of protocols for estimating PFC and CO2 process emissions from primary aluminum
production. These protocols include the 2006 IPCC Guidelines, EPA's Voluntary Aluminum Industrial Partnership
(VAIP), the Inventory of U.S. Greenhouse Gas Emissions and Sinks, the International Aluminum Institute's (IAI)
Aluminum Sector Greenhouse Gas Protocol, the Technical Guidelines for the Voluntary Reporting of Greenhouse
Gases (1605(b)) Program, EPA's Climate Leaders Program, and The Climate Registry.
The methods described in these protocols and guidance coalesce around the methods described by the lAI's
Aluminum Sector Greenhouse Gas Protocol and the IPCC guidelines. These methods range from Tier 1 approaches
based on aluminum production to Tier 3 approaches based primarily on smelter-specific data. The IPCC Tier 3 and
IAI methods are essentially the same. For PFCs, they both require smelter-specific data on anode effect frequency
and duration, smelter-specific slope factors, and aluminum production. For CO2, they require smelter-specific data
on anode consumption and anode characteristics (chemical contents)1.
1 The IAI protocol includes an alternate approach for calculating CO2 emissions. However, although this equation
does not refer to impurities, it is assumed that these are already factored in and thus not included in the calculation.
-------
a. Monitoring Methods for PFCs
1. Option 1: Default Emission Factor
Option 1, which is the same as the IPCC Tier 1 approach, uses the product of aluminum production and a
technology-specific default emission factor to estimate PFC emissions from primary aluminum production. Though
this methodology is simple, the default emission factors for PFCs have large uncertainties due to the variability in
anode effect frequency and duration. Based on 1990 data, the IPCC Guidelines give an uncertainty range of -99
percent to +380 percent for the default factor for the Center- Work Prebake technology, the most common smelter
technology in the United States. Moreover, since 1990, all U.S. smelters have sharply reduced their anode effect
frequency and duration; through 2006, average AE minutes per cell day have declined by approximately 85%,
lowering U.S. smelter emission rates well below those of the IPCC Tier 1 defaults.
2. Option 2: Smelter-Specific Anode-Effect Minutes
Option 2, which is the same as the IPCC Tier 2 approach, uses smelter-specific data on anode effect frequency and
duration. Option 2 also uses data on aluminum production and technology-specific slope coefficients.
The slope coefficient represents kilograms of CF4/metric ton of aluminum produced divided by anode effect minutes
per cell-day. The cell-day is the number of cells operating multiplied by the number of days of operation (IPCC
2006). The following equations describe how to calculate CF4 and C2F6 emissions based on the slope method.
ECF4 = SCF4 X AEM
Ec2F6 = ECF4 X FC2F6/CF4
where,
ECF4 is emissions of CF4 from aluminum production (kg CF4)
EC2F6 is emissions of C2F6 from aluminum production (kg C2F6)
SCF4 is the slope coefficient ([kg CF4/metric ton Al]/[AE-Mins/cell-day])
AEM is anode effect minutes per cell-day (AE-Mins/cell-day)
MP is metal production (metric tons Al)
Fc2F6/cF4 is the weight fraction of C2F6/CF4 (kg C2F6/kg CF4)
Although Option 2 results in estimates that are considerably more accurate than those based on Option 1, Option 2 is
significantly less precise than Option 3, as discussed below.
3. Option 3: Smelter-Specific Anode-Effect Minutes and Slope Coefficients
Option 3 uses the same set of equations and parameters as Option 2. The critical distinction between the two
methods is that Option 3 requires recent smelter-specific slope coefficients while Option 2 relies on default,
technology-specific slope coefficients. Of the currently operating U.S. smelters, all but one has measured a smelter-
specific coefficient at least once. However, to use Option 3, some smelters would need to update these
measurements if they occurred more than 3 years ago.
Use of Option 3, which is similar to the IPCC Tier 3 approach, leads to significantly more precise PFC emissions
estimates than use of Option 2. For individual facilities using the most common smelter technology in the United
States., the uncertainty (95% confidence interval) of estimates developed using the Option 2 approach is ±50
percent,2 while the uncertainty of estimates developed using the Tier 3 approach is approximately ±15 percent
(Marks 2008). For a typical U.S. smelter emitting 175,000 metric tons of CO2-eq in PFCs, these errors result in
The most common smelter technology in the United States is the center-work prebake (CWPB) technology. The
2006 IPCC Guidelines provide a 95% confidence interval of ±6 percent for the CWPB default slope coefficient.
However, this range is not the range within which the slope coefficient from a single CWPB has a 95 percent chance
of falling. Instead, it is the range within which the true mean of all CWPB slope factors has a 95 percent chance of
falling.
-------
absolute uncertainties of ±88,000 mtCO2e and ±26,000 mtCO2e, respectively. The reduction in uncertainty
associated with moving from Option 2 to Option 3, 62,000 mtCO2e, is as large as the emissions from many of the
sources that would be subject to the rule.
Option 3 requires that slope coefficients be measured using a method similar to the USEPA/IAI Protocol for
Measurement ofTetrafluoromethane and Hexafluoroethane from Primary Aluminum Production. This protocol was
first published in 2003 and updated in 2008. The protocol establishes guidelines to ensure that measurements of
smelter-specific slope-coefficients are consistent and accurate (e.g., representative of typical smelter operating
conditions and emission rates). These guidelines include recommendations for documenting the frequency and
duration of anode effects, measuring aluminum production, sampling design, measurement instruments and
methods, calculations, quality assurance and quality control, and measurement frequency.
Both the Protocol and industry experts currently recommend that smelter operators re-measure their slope
coefficients at least every three years, and more frequently if they adopt changes to process control algorithms or
observe changes to typical anode effect duration (Marks, 2008a). Specifically, the Protocol recommends that
operators repeat measurements of slope coefficients for CF4 and C2F6 if one or more of the following apply:
1. Thirty-six months have passed since the last measurements (i.e. triennial measurements are
recommended);
2. A change occurs in the control algorithm that affects the mix of types of anode effects or the nature of
the anode effect termination routine;
3. Changes occur in the distribution of duration of anode effects (e.g. when the percentage of manual kills
changes or if, over time, the number of anode effects decreases and results in a fewer number of longer
anode effects)
Changes to process control algorithms or to the typical duration of anode effects can change the relationship
between anode effect minutes, production, and emissions, that is, they can change slope coefficients. In addition,
more subtle changes can also change slope coefficients over time. According to industry experts, the rate of these
more subtle changes has not been sufficiently studied to specify a frequency for re-measurement of less than once
every three years. Thus, Option 3 requires that slope factors be re-measured at least once every three years.
During the past few years, multiple U.S. smelters have adopted changes to their production process which are likely
to have changed their slope coefficients (Marks, 2008a). These include the adoption of slotted anodes and
improvements to process control algorithms. Although some U.S. smelters, such as those operated by Alcoa, have
recently updated their measurements of smelter-specific coefficients, others may not have.
While the cost to implement Option 3 is significantly greater than the cost to implement Option 2, the benefit of
reduced uncertainty is considerable, as noted above. The costs that would be incurred by smelters measuring slope
factors are discussed in the Regulatory Impact Analysis for this rulemaking (EPA-HQ-OAR-2008-0508-002).
Another Tier 3 method included in the IPCC Guidelines is the Overvoltage Method. This method relates PFC
emissions to an overvoltage coefficient, anode effect overvoltage, current efficiency, and aluminum production. The
overvoltage method was developed for smelters using the Pechiney technology. It is EPA's understanding that no
U.S. smelters have used the Pechiney technology for at least a decade.
b. Monitoring Methods for CO2
CO? emitted during electrolysis
1. Option 1: Default Emission Factor
Option 1, which is the same as the IPCC Tier 1 approach, uses the product of aluminum production and a
technology-specific default emission factor to estimate CO2 emissions during electrolysis. This methodology is
simple, and the difference in accuracy between emission estimates developed using Option 1 and Option 2 (five to
ten percent) is notably lower for U.S. smelters than the difference for the PFC estimates. (The IPCC Guidelines
note, "the reactions leading to carbon dioxide emissions are well understood, and the emissions are very directly
connected to the tons of aluminum produced through the fundamental electrochemical equations for alumina
reduction.") However, as part of typical operations, facilities regularly monitor inputs to higher Tier methods (e.g.,
-------
consumption of anodes); consequently, the incremental cost to develop an Option 2 or Option 2/3 hybrid estimate
will be small.
2. Options 2 and 3: Smelter-specific anode consumption
Options 2 and 3, which are the same as the IPCC Tier 2 and 3 approaches, respectively, estimate CO2 emissions
from electrolysis based on metal production and net anode consumption. Options 2 and 3 are mass balance
approaches that assume that all carbon from net anode consumption is ultimately emitted as CO2. Both anode
consumption and aluminum production are collected as part of typical facility operating processes. Other terms in
the Option 2/3 equation make minor adjustments for non-carbon components of the anodes (e.g., sulfur and ash).
The distinction between Option 2 and Option 3 is that Option 2 uses default values for these minor components
while Option 3 uses specific facility operating data for these components. Since the concentrations of these
components are small (typically less than one percent to five percent), facility-specific data on them is not as critical
to the precision of emission estimates as is facility-specific data on net anode consumption. Option 3 improves the
accuracy of the results but the improvement in accuracy is not expected to exceed 5 percent (IPCC 2006).
The following equation describes how to calculate emissions based on these parameters for each technology type.
For Prebake cells:
EC02 = NAC x MP x ([100 - Sa - Ash]a / 100) x (44/12)
where,
ECo2 is CO2 emissions from prebaked anode consumption (metric tons CO2)
MP is total metal production (metric tons Al)
NAC is net prebaked anode consumption per metric ton Al (metric tons C/metric tons Al)
Sa is sulfur content in baked anodes (percent weight)
Asha is ash content in basked anodes (percent weight)
44/12 is CO2 molecular mass: carbon atomic weight ratio (dimensionless)
For Soderberg cells:
EC02 = (PC x MP - [CSM x MP]/1000 - BC/100 x PC x MP x [Sp + Ashp+ Hp] / 100 - [100-BC]/100 x PC x MP
x [So + Ashc] 7100 - MP x CD) x (44/12)
where,
ECo2 is CO2 emissions from paste consumption (metric ton CO2)
MP is total metal production (metric ton Al)
PC is paste consumption (metric ton/metric ton Al)
CSM is emissions of cyclohexane soluble matter (kg/metric ton Al)
BC is binder content in paste (percent weight)
Sp is sulfur content in pitch (percent weight)
Ash p is ash content in pitch (percent weight)
Hp is hydrogen content in pitch (percent weight)
Sc is sulfur content in calcined coke (percent weight)
Ash o is ash content in calcined coke (percent weight)
CD = carbon in skimmed dust from Soderberg cells (metric ton C/metric ton Al)
44/12 is CO2 molecular mass: carbon atomic weight ratio (dimensionless)
The data reported by companies participating in EPA's Voluntary Aluminum Industrial Partnership (VAIP) has
generally not included smelter-specific values for each of these variables. However, most participants in VAIP have
-------
used either data on paste consumption (for Soderberg cells) or on net anode consumption (for Prebake cells), along
with some smelter-specific data on impurities, to develop a hybrid Option 2/3 estimate (i.e., combination of smelter-
specific and default factors).
CO? emitted during anode baking
1. Options 2 and 3: Smelter-specific anode consumption
Options 2 and 3, which are the same as the IPCC Tier 2 and 3 approaches, respectively, estimate CO2 emissions
from combustion of materials during anode baking. (There is no Tier 1 approach for estimating these emissions.)
The Option 2/3 equations rely on a mass balance approach involving the chemical contents of the anodes and
packing materials. No anode baking emissions occur when using Soderberg cells, since these cells are not baked
before aluminum smelting, but rather, bake in the electrolysis cell during smelting. The following equations
describe how to calculate emissions from anode baking of Prebake cells.
EC02PV = (GA - Hw - BA - WT) x (44/12)
EC02PC = PCC x BA x ([100 - Spc - Ashpc] /100) x (44/12)
where,
ECo2pv is CO2 emissions from pitch volatiles combustion (metric tons CO2)
ECo2pc is CO2 emissions from bake furnace packing material (metric tons CO2)
GA is initial weight of green anodes (metric tons)
Hw is hydrogen content in green anodes (metric tons)
BA is baked anode production (metric tons)
WT is waste tar collected (metric tons)
PCC is packing coke consumption (metric tons/metric ton BA)
Spc is sulfur content in packing coke (percent weight)
Ashpc is ash content in packing coke (percent weight)
As is the case for CO2 emitted during electrolysis, the Option 2 approach relies on industry-wide defaults for minor
anode components, requiring smelter-specific data only for the initial weight of green anodes (GA) and for baked
anode production (BA), while Option 3 requires smelter-specific values for all parameters. Again, the
concentrations of minor components are small, limiting their impact on the estimate of CO2 emissions from anode
baking. In addition, anode baking emissions account for approximately 10 percent of total CO2 process emissions,
so reducing the uncertainty in this estimate will have only a minor impact on the overall CO2 process estimate. For
EPA's VAIP program, many smelters report only some smelter-specific values for the concentrations of minor
anode components.
4. Procedures for Estimating Missing Data
Where anode effect minutes per cell day data points are missing, the average anode effect minutes per cell day of the
remaining measurements within the same reporting period may be applied. However, these parameters are typically
logged by the process control system as part of the operations of nearly all aluminium production facilities and the
uncertainties in these data are low.
It is assumed that aluminum production levels will be known, since businesses rely on accurate monitoring and
reporting of production levels. Consequently, there is less than 1 percent uncertainty in the data for the annual
production of aluminum. The likelihood for missing data is low.
For CO2 emissions, the uncertainty in recording anode consumption as baked anode consumption or coke
consumption is estimated to be only slightly higher than for aluminium production, less than 2 percent (IPCC,
2006). This is also an important parameter in smelter operations and is routinely/continuously monitored. The
likelihood for missing data is low.
-------
5. QA/QC Requirements
As noted above, Option 3 for estimating PFC emissions would require that smelter-specific slope coefficients be
measured at least every thirty-six months in accordance with the 2003 (for measurements through 2008) or 2008 (for
measurements after 2008) EPA/LAI Protocol for Measurement ofTetrafluoromethane and Hexafluoroethane
Emissions from Primary Aluminum Production. As noted in the Protocol, key monitoring issues that should be
addressed before undertaking a study include the following.
• Measurement conditions should reflect typical operations at the smelter.
• Measurements should be conducted at locations that capture cells where anode effect data is being collected and
where there is good flow homogeneity in the gas being sampled.
• Measurements should account for background PFC concentrations, and assess the need to quantify fugitive
losses (specifically if collection system efficiencies are less than 90 percent).
• The sampling time should be at least 72 hours; however, a "rule of thumb" should be that the slope coefficient
running average does not change by more than 10 percent from the previous average value.
6. Reporting Procedures
Reporting of the following data would be useful for confirming emissions calculations and/or calculating emission
rates that could be compared across facilities and over time for data quality control purposes:
• Aluminum production amount in metric tons aluminum
• Smelter technology used
• PFC-specific information:
o Anode effect minutes per cell-day
o Smelter-specific slope coefficient
o Last date when smelter-specific slope coefficient was measured
o Certification by owner/operator that measurements of slope coefficients were conducted
in accordance with the 2003 (for measurements through 2008) or 2008 (for measurements
after 2008) EPA/IAI Protocol for Measurement ofTetrafluoromethane and
Hexafluoroethane Emissions from Primary Aluminum Production.
o Criteria used by the smelter to measure the frequency and duration of anode effects
• CO2-specific information:
o Anode consumption.
o Smelter-specific inputs to the CO2 process equations (e.g., levels of impurities) that were
used in the calculation. Exact data elements required will vary depending on smelter
technology.
-------
7. References
EPA (2008) Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2006. U.S. Environmental Protection
Agency, Washington, DC.
EPA/IAI Protocol for Measurement of CF4 and C2F6 Emissions from Primary Aluminum Production, April 2008.
IPCC (2006) 2006 IPCC 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.
Marks, J. (2008), China PFC Management Workshop, Asia Pacific Partnership, April 2008.
Marks, J. (2008a), Telephone conversation with Deborah Ottinger of USEPA, April 30, 2008.
USGS (2006) Mineral Commodity Summaries. U.S. Geological Survey, Reston, VA.
10
------- |