Standards of Performance for Petroleum
Refineries
Background Information for Final Amendments
Summary of Public Comments and Responses
(Revised)
-------
This page intentionally left blank.
11
-------
Standards of Performance for
Petroleum Refineries
Background Information for Final Amendments
Summary of Public Comments and Responses (Revised)
Contract No. EP-D-11-084
Work Assignment No. 0-11
Project No. 06/09 (438)
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Sector Policies and Programs Division
Research Triangle Park, North Carolina 27711
December 2011
(Revised August 2012)
in
-------
Disclaimer
This report has been reviewed by the Sector Policies and Programs Division of the Office of Air
Quality Planning and Standards, EPA, and approved for publication. Mention of trade names or
commercial products is not intended to constitute endorsement or recommendation for use.
iv
-------
TABLE OF CONTENTS
1.0 Introduction 1
2.0 Applicability 4
2.1 Applicability of Subpart Ja 4
2.1.1 Gasification 4
2.1.2 Crude Oil from Oil Shale and Tar Sands 8
2.1.3 Flares 10
2.1.4 Other Units 12
2.2 Applicability Date 13
2.3 Flare Modification 19
2.4 Notification Schedule 33
2.5 Overlap with and Effect on Other Subparts 34
2.6 Equivalence with State and Local Rules 36
3.0 Definitions 41
3.1 Air Preheat 41
3.2 Co-Fired Process Heater 41
3.3 Corrective Action Analysis 42
3.4 Delayed Coking Unit 43
3.5 Flare 44
3.6 Fuel Gas 46
3.7 Fuel Gas Combustion Device 49
3.8 Natural Draft and Forced Draft Process Heaters 50
3.9 Process Upset Gas 52
3.10 Refinery Process Unit 54
3.11 Sulfur Recovery Plant 57
4.0 Fuel Gas Combustion Devices and Flares 58
4.1 Emissions Limits 58
4.1.1 Long-Term H2S Concentration Limit for Fuel Gas Combusted in Flares 58
4.1.2 Alternative Compliance Options 61
4.2 Elimination of Routine Flaring 62
4.3 Flare Flow Rate Limit 66
4.4 Other Standards for Flares 66
4.4.1 General Comments on Other Flare Standards 66
4.4.2 Flare Management Plan Requirements 68
4.4.3 Root Cause Analysis and Corrective Action Analysis 70
4.5 Monitoring Fuel Gas Combustion Devices 78
4.6 Flow Monitoring for Flares 78
4.7 Sulfur Monitoring for Flares 81
4.8 Compliance Schedule 88
4.9 Startup, Shutdown and Malfunction 93
5.0 Process Heaters 101
5.1 Emissions Limits 101
5.1.1 Equivalence of Emissions Limit Units 101
5.1.2 Limits for New, Modified and Reconstructed Forced Draft Process
Heaters 102
v
-------
5.1.3 Turndown 104
5.1.4 Site-Specific Emissions Limit Clarifications 108
5.1.5 Co-Fired Process Heaters Ill
5.2 Monitoring of Process Heaters 114
5.3 Other Clarifications 118
Compliance Requirements and General Monitoring, Recordkeeping and Reporting 119
6.1 Rolling Averages 119
6.2 Excess Emissions 120
6.3 Monitoring Methods 121
6.4 Reporting Requirements 121
6.5 Impacts on Small Refineries 121
vi
-------
1.0 INTRODUCTION
On June 24, 2008 (73 FR 35838), the U.S. Environmental Protection Agency (EPA)
issued: (1) final amendments to the existing refineries new source performance standards (NSPS)
in 40 CFR part 60, subpart J ("subpart J"); and (2) new petroleum refinery NSPS in 40 CFR part
60, subpart Ja ("subpart Ja"). On September 26, 2008 (73 FR 55751), the effective date of the
NSPS in 40 CFR part 60, subpart Ja, the EPA published a notice granting the request for
reconsideration and a 90-day stay of the following provisions of 40 CFR part 60, subpart Ja:
(1) the flare modification provision; (2) the definition of "flare"; (3) the fuel gas combustion
device sulfur limits; (4) the flow limit for flare systems; (5) the total reduced sulfur (TRS) and
flow monitoring requirements for flares; and (6) the nitrogen oxides (NOx) emissions limit for
process heaters. On December 22, 2008, the EPA proposed to extend the original 90-day stay
until the agency has reached a final decision on all of the issues for which reconsideration was
granted (73 FR 78260, 73 FR 78546, 73 FR 78549).
Also on December 22, 2008 (73 FR 78522), the EPA proposed amendments to the NSPS
in 40 CFR part 60, subparts J and Ja to address the reconsidered issues as well as technical
corrections. The proposal provided a 45-day comment period, ending February 5, 2009. The
EPA received 20 comments on the proposed amendments from refiners, industry trade
associations and consultants, state environmental departments, environmental groups and other
interested parties during the comment period. Following the close of the public comment period,
two additional comment letters were received. All 22 of these comments have been placed in the
docket for this rulemaking (Docket No. EPA-HQ-OAR-2007-0011). Table 1 lists the names and
affiliation of each comment and the assigned docket number.
1
-------
Table 1. List of Commenters on the Proposed Amendments to 40 CFR part 60, subpart Ja
Docket Item No.
KI'A-I IQ-OAR-2007-
0011-
C oniiiKMiU'r :tnil AITiliiilion
0295
B. Lane, Environmental Consultant
0296
RA. Cade, Marathon Petroleum Company LLC
0297
M. Resh, Linde North America, Inc.
0298
D. Copeland, P.E., Praxair Inc.
0299
B. Lane, Environmental Consultant
0300
C.M. Campbell, RTP Environmental Associates Inc., on behalf of Hyperion
Refining, LLC
0301
C. Feerick, ExxonMobil Company
0302
S.L. Sherk, American Refining Group, Inc
0303
M.E. Pugh, Flint Hills Resources, LP (FHR)
0304
R.W. Hermanson, BP America, Inc.
0305
D.F. Hunter, ConocoPhillips Company
0306
L.B. Barry, Chevron Corporation
0307 (Attachments
posted as 0317)
J.S. Peterson, Environmental Integrity Project (EIP), also on behalf of Natural
Resources Defense Council (NRDC) and Sierra Club
0308
C.S. Colman, HESS Corporation, on behalf of HOVENSA LLC
0309
T.K. Metrose, Tesoro Corporation
0310
K.A. Saffell, Valero Energy Corporation
0311
R. Chittim, American Petroleum Institute (API), National Petrochemical and
Refiners Association (NPRA) and Western States Petroleum Association
(WSPA)
0312
A. Pasquale, Thermo Fisher Scientific
0313
B.F. Bateman, Bay Area Air Quality Management District (BAAQMD)
0314
H.A. Green, CITGO Petroleum Corporation
0315
R. Chittim, API, NPRA and WSPA
0316
B.F. Bateman, BAAQMD
Several commenters, in addition to providing specific comments, also expressed support
for comments provided by other commenters (generally, individual companies indicating support
for comments submitted by their trade organization). Specifically, Commenters 0295, 0301,
0302, 0304, 0305, 0306, 0308 and 0310 supported comments submitted by Commenter 0311
(API, NPRA and WSPA). By convention, the commenter numbers presented in the comment
2
-------
summaries throughout the rest of this document indicate the commenter(s) that specifically
provided those comments and not the commenters who supported that comment.
All of the comments and the EPA's responses to the comments are summarized in this
document or the preamble to the final amendments. The EPA's responses to these comments
form part of the basis for the EPA's final decisions.
3
-------
2.0 APPLICABILITY
2.1 Applicability of Subpart Ja
2.1.1 Gasification
Comment: Commenter 0300 stated that the EPA should exempt entrained flow
gasification processes from subpart Ja. The commenter cited a recent applicability determination
provided by the South Dakota Department of Environmental Resources that indicated syngas
produced from coke gasification was not fuel gas subject to subpart Ja and questioned how the
gasification process would be considered a refinery process unit and the associated flare a fuel
gas combustion device under the proposed amendments. The commenter stated that the best
system of emission reduction (BSER)1 determination did not address entrained flow gasification
processes or consider that gas with a low energy content (also referred to as low British thermal
unit [Btu] content) is not suitable for recovery as fuel gas. The commenter noted that both 40
CFR part 60, subpart Da and subpart Ja would apply to their gasification facility, causing
redundant requirements, but the commenter's only specific objection was to the requirement to
conduct a root cause analysis (RCA) for planned startup and shutdown events when there is no
outlet for the low Btu gas other than the flare. The commenter did not express concern with the
fuel gas sulfur limits because, according to the commenter, the syngas is desulfurized prior to
combustion under normal operations.
Commenters 0310 and 0311 objected to the proposed amendments to the definition of
"fuel gas" and "refinery process unit" to include coke gasification. The commenters cited a
Delaware court case that an adjacent coke gasification plant was not part of the refinery, largely
because coke gasification was not an integral refinery process operation, according to the
commenters. Furthermore, the commenters contended that coke gasification was not considered
in the original NSPS (40 CFR part 60, subpart J) because the technology was not developed until
1 The level of control prescribed by CAA section 111 historically has been referred to as "best demonstrated
technology" or BDT. In order to better reflect that CAA section 111 was amended in 1990 to clarify that "best
systems" may or may not be "technology," the EPA is now using the term "best system of emission reduction" or
BSER. Therefore, while commenters referred to "BDT" in their comment letters, we use "BSER" throughout this
document.
4
-------
the mid-1980s, and the subpart Ja rulemaking docket contains no BSER analysis specific to coke
gasification. As such, the commenters asserted that the revision is more than a "clarification," the
rulemaking record failed to satisfy applicable procedural standards for establishing BSER and
the EPA failed to afford interested parties with a meaningful opportunity to comment. According
to Commenters 0310 and 0311, the inclusion of coke gasification under subpart Ja places
unreasonable operational limitations and costs on these sources.
Commenter 0310 noted that much of the sulfur from the coke gasifier is in the form of
carbonyl sulfide (COS) and provided a cost for systems to meet a TRS limit of 50 parts per
million by volume (ppmv) that would comply with the 60 ppmv hydrogen sulfide (H2S)
concentration limit determined daily on a 365 successive calendar day rolling average basis
(hereafter referred to as the long-term 60 ppmv H2S concentration limit). The commenter stated
that the control system is not BSER because the cost effectiveness of the system is roughly
$20,000/ton of sulfur dioxide (SO2) reduced. The commenter also contended that flexicoking
units are "materially different" than coke gasifiers by focusing on the upstream coking processes
associated with a flexicoking unit rather the production and treatment of low energy gas from the
flexicoking gasifier. The commenter noted that FLEXSORB amine treatment units achieve <10
ppmv H2S on flexicoking gas but suggested that this performance is not applicable or
representative of the performance achievable from coke gasification units because flexicoking
gasifiers use air while coke gasification units use oxygen (so the flexicoking gas is diluted with
nitrogen). According to Commenters 0310 and 0311, the proposed rule would discourage use of
gasification technology, contrary to established national security objectives.
Response: The key issues here are: (1) whether the clarifications to the definition of "fuel
gas" and "refinery process units" altered the historic applicability of the refinery fuel gas
standards for coke gasifiers; and (2) whether low-Btu gas produced from coke gasification units
should be subject to the fuel gas standards. With respect to the inclusion of "other gasifiers" in
the definition of fuel gas, we reaffirm our position that this is a clarification rather than a new
requirement. We note that the definition of "fuel gas" at 40 CFR 60.101a is "any gas which is
generated at a petroleum refinery and which is combusted," and the amendments do not alter that
language. The definition does not require the gas to be generated by a "refinery process unit" but
more broadly applies to any gas generated at a petroleum refinery. We also note that the
definition of "petroleum refinery" at 40 CFR 60.101a is "any facility engaged in producing
5
-------
gasoline, kerosene, distillate fuel oils, residual fuel oils, lubricants, asphalt (bitumen) or other
products through distillation of petroleum or through redistillation, cracking, or reforming of
unfinished petroleum derivatives." The primary subject in this definition is "any facility." It is
not "the group of refinery process units" or similarly limiting phrases. The definition of
petroleum refinery includes the entire facility. If a coke gasifier is located at a facility "engaged
in producing gasoline, kerosene, distillate fuel oils, residual fuel oils, lubricants, asphalt
(bitumen) or other products through distillation of petroleum or through redistillation, cracking,
or reforming of unfinished petroleum derivatives," then the coke gasifier is located at a
petroleum refinery. As such, the gas produced from the coke gasification process would be
considered "fuel gas" regardless of whether the gasification process itself is considered a refinery
process unit (see next paragraph for discussion on this issue). Consequently, prior to the newly
proposed clarification to the definition of "fuel gas" in subpart Ja, it appears that a flare or other
unit that combusts gas for a coke gasifier would be an affected facility under subpart J or Ja,
provided that the gasifier is located at a petroleum refinery. Based on the background provided
by Commenter 0300 ("the planned refinery is a 400,000 barrel per day, highly complex, full
conversion refinery that will produce clean, green transportation fuels such as ultra-low sulfur
gasoline and ultra-low sulfur diesel"), the coke gasifier is located at a petroleum refinery and the
gas generated by the gasifier meets the definition of "fuel gas" regardless of the proposed
clarifications to subpart Ja. Consequently, we find that the coke gasification unit described by
Commenter 0300 is located at a petroleum refinery, the syngas produced by the unit is "fuel gas"
and the flare combusting the fuel gas is an affected source subject to subpart Ja. (See letter to
Preston Phillips, Vice President, Hyperion Refining LLC, from Cynthia J. Reynolds, Director,
Technical Enforcement Program, U.S. EPA, Region 8, dated November 20, 2008 (Applicability
Determination Control Number: 0800090) for further detail on this determination.) In addition,
we note that in Star Enterprise v. EPA, 235 F.3d 139, 151 (3rd Cir. 2001), the decision referenced
by Commenters 0310 and 0311, the reason the Court held that the coke gasification plant was not
part of the refinery was because of the physical location of the coke gasification plant, i.e., it was
located next door to the refinery rather than "in" the refinery ("[I]n determining what facilities
are 'affected facilities' that can be regulated under Subpart J, and, specifically, in determining
what facilities are 'in petroleum refineries,' the touchstone of such a determination is the
physical location of the facilities in question."). The Court did not address the factual situation in
6
-------
which a coke gasifier and turbines are physically located "in" the petroleum refinery. See 40
CFR 60.100a(a); Star Enterprise, 235 F.3d at 151-152.
With respect to the proposed clarification of "refinery process unit," the original
definition read as follows: "Refinery process unit means any segment of the petroleum refinery
in which a specific processing operation is conducted." As stated previously, the petroleum
refinery includes the entire facility. Thus, to be considered a refinery process unit, only two
criteria are needed: (1) the unit must be located at a petroleum refinery; and (2) the unit must be
used to conduct "a specific processing operation." The definition does not directly limit the
scope of "processing operations." That is, the definition of refinery process unit does not limit
process operations to distillation, re-distillation, cracking or reforming, and it does not limit it to
only those processes used to produce gasoline, kerosene, fuel oils, etc. As coke gasification is
easily construed as a "specific processing operation," this unit meets the definition of a refinery
process unit if the unit is located at a petroleum refinery. Consequently, we considered the
proposed inclusion of coke gasification (as well as product loading, sulfur recovery and
wastewater treatment) as an example of a refinery process unit to be a clarification of the existing
definition rather than an expansion of the definition. (See Section 3.10 for additional discussion
on the definition of "refinery process unit.")
The more overarching issue is whether or not low-Btu gas produced from coke
gasification units should be subject to the fuel gas standards. The commenters suggest that a
separate analysis of the BSER is needed for coke gasification units and that the requirements of
subpart Ja are not cost effective for coke gasifiers. We find that both the long-term 60 ppmv H2S
concentration limit and the 162 ppmv H2S concentration limit determined hourly on a 3-hour
rolling average basis (hereafter referred to as the short-term 162 ppmv H2S concentration limit)
are achievable based on the information provided by Commenter 0300, as well as the
performance of other integrated gasification combined cycle (IGCC) units. Gasification systems
have received support from the EPA and the U.S. Department of Energy (DOE) largely because
they are considered "clean technologies" because sulfur can be efficiently removed prior to final
combustion. Commenter 0300 only expressed concern regarding the flare flow RCA, simply
stating that the syngas is desulfurized before being flared. Additionally, while we acknowledge
that higher H2S concentrations can be expected from a gasification system using pure oxygen
rather than air, the fact that the FLEXSORB amine treatment units achieves <10 ppmv H2S on
7
-------
flexicoking gas appears to support the case that, correcting for the diluting nitrogen, oxygen-fired
gasification systems should be able to achieve H2S concentrations below 60 ppmv, and this
information was included in the rulemaking record.
Although Commenter 0310 provided cost estimates to support their claim that the long-
term 60 ppmv H2S concentration limit and the short-term 162 ppmv H2S concentration limit in
subpart Ja are not cost effective for gasification, the commenter focused on COS removal rather
than on H2S removal and, as such, mischaracterized the subpart Ja requirements. The limits on
sulfur content of fuel gas are based on the H2S concentration of the fuel gas. While we evaluated
limiting the concentration of other reduced sulfur compounds in fuel gas when we originally
proposed subpart Ja, we did not finalize those standards as BSER due to the high costs of the
technologies needed to reduce the total sulfur content of the fuel gas. Although we encourage the
use of these COS removal technologies, we fail to see how the subpart Ja requirements mandate
them. The commenter provided no evidence that amine scrubbing cannot be used to meet the
H2S concentration limits. Based on our analysis of the costs for amine scrubbing systems already
in the docket, we conclude that amine scrubbing of the syngas from coke gasification to meet the
long-term 60 ppmv Ft2S concentration limit and the short-term 162 ppmv H2S concentration limit
is achievable and cost effective and therefore BSER. In addition, this information supports our
decision to finalize the proposed clarification that fuel gas includes "gases from flexicoking unit
gasifiers and other gasifiers."
Regarding flares associated with coke gasification units, we note that as described in
more detail in Section 4.4.3, we agree that a RCA is unnecessary for planned startups and
shutdowns that are already subject to and following minimization requirements of the flare
management plan (FMP) under 40 CFR 60.103a(a)(5). Eliminating the RCA requirement for
planned startups and shutdowns of units following their FMP appears to adequately address
Commenter 0300's primary concern.
2.1.2 Crude Oil from Oil Shale and Tar Sands
Comment: Commenter 0307 stated that the EPA should regulate facilities that only
produce crude oil from oil shale and tar sands. The commenter acknowledged that these facilities
do not appear to fall within the regulatory definitions for "petroleum refinery" and "refinery
process unit," but it is important for the EPA to set strict emissions limits and monitoring
8
-------
requirements to regulate emissions from these facilities. The commenter noted that extraction of
oil from tar sands and oil shale are energy intensive processes, resulting in nearly three times
more greenhouse gas emissions than extraction of conventional crude oil.2 The commenter also
noted that oil shale extraction releases emissions of SO2 and other sulfur compounds, particulate
matter (PM), carbon monoxide (CO), carbon dioxide (CO2), ozone (O3), NOx, lead, silica,
metals, ammonia (NH3), trace organics and trace elements.3 Commenter 0305 supported the
change to the definition of "petroleum refinery" to exclude oil shale and tar sands-derived crude
but requested that the definition specifically exclude oil production as well, either by referring to
Standard Industrial Classification (SIC) 2911 or adding, "nor is the extraction or production of
crude oil" to the end of the definition.
Response: The definition of "petroleum refinery" does not include, and could not be
reasonably construed to include, conventional oil production facilities. While we also do not
believe oil shale and tar sand processing facilities fall within the regulatory definitions for
"petroleum refinery," the presence of a distillation or separation unit at these oil shale and tar
sands processing facilities led to ambiguities, which is why the clarification was included in the
definition. It may be reasonable to apply certain subpart Ja standards, such as FMP, to these
types of facilities, but we have not completed a BSER analysis specific to these processes,
especially when they are isolated from a refinery. As in the previous discussion regarding coke
gasification, we do consider oil shale and tar sand-processing units that are co-located at a
petroleum refinery facility to be part of the petroleum refinery and subject to the FMP, RCA and
fuel gas and process heater emissions standards. However, more information is needed on the
operations and emissions from these oil production facilities. We are continuing to look at these
processes to determine if regulation of these emission sources is needed. We do note that
upgrading units that employ coking units to process tar sands into a synthetic crude oil do not
meet the exclusion from the definition of a petroleum refinery because these facilities do more
than simply remove diluent from the heavy oil that was added to effect extraction or to improve
flow characteristics for pipeline transport.
2 Wakefield, Ben, and Matt Price, Environmental Integrity Project. Tar Sands: Feeding U.S. Refinery Expansions
with Dirty Fuel. June 2008.
3 Bordetsky, Ann, et al., Natural Res. Def. Council et al. Driving it Home: Choosing the Right Path for Fueling
North America's Transportation Future. June 2007. Available at
http://www.nrdc.org/energv/drivingithome/drivingithome.pdf.
9
-------
2.1.3 Flares
Comment: Commenters 0305 and 0311 objected to the reference to the expanded subpart
Ja definition of "flare" in subpart J and stated that it expands the applicability of subpart J
retroactively. According to Commenter 0311, since subpart J does not specifically define "flare,"
the cross-reference to the subpart Ja definition could be construed as the definition that should
have been applied historically under subpart J.
Response: As stated previously, we do not consider the reference in subpart J to the
definition of flare in subpart Ja as changing the definition of flare under subpart J, regardless of
there being no formal definition included in that subpart. After again reviewing the available
information, we maintain that the cross-reference in subpart J to the definition of "flare" in
subpart Ja only changes the applicability date for flares; however, to avoid confusion, we have
removed the specific reference to 40 CFR 60.101a (the subpart Ja definition of "flare") from
40 CFR 60.100(b) of subpart J.
Comment: Commenter 0311 stated that temporary flares (those placed in service for less
than 6 months) should be excluded from subpart Ja. These temporary flares are needed for flare
maintenance as well as for tank cleanings or similar activities in remote areas of the refinery.
Commenter 0315 clarified that temporary flares might be needed to handle emergency flows
while the main flare is off-line for purposes of installing equipment needed for that main flare to
comply with subpart Ja. According to the commenter, the EPA should encourage maintenance of
existing flares, but these activities would be discouraged if the temporary flares were subject to
the rule. Also, the commenter stated that the EPA did not consider the costs of subjecting
temporary flares to subpart Ja.
Response: With respect to flare maintenance, if the flare being maintained is already
subject to subpart Ja, it appears that the temporary flare should be able to comply with subpart
Ja. If the temporary flare is considered a "new" flare, then routine maintenance of a flare not
currently subject to subpart Ja would effectively trigger the rule. We can see that this could
create an unnecessary barrier to properly maintaining the flare. Given the importance of flares in
the safe operation of the facility and in the destruction of flammable gases in the case of an upset
or malfunction, we agree that a temporary flare installed for the purposes of conducting
maintenance on an existing flare must meet the applicable limits of the flare that it is temporarily
replacing and does not, in itself, trigger applicability to subpart Ja.
10
-------
For tank cleaning operations, we believe that the gases generated can be considered
process upset gases (produced as a result of startup or shutdown). Thus, the rule specifically
allows the use of a flare to combust these gases without subjecting them to H2S concentration
requirements for fuel gas. As such, for tank cleaning operations, these flares would only be
subject to RCA and FMP requirements. As discussed in Section 4.4.3, we have modified the
requirements in 40 CFR 60.103a(d)(3) to exclude RCA for planned startups and shutdowns that
follow the flare minimization procedures in 40 CFR 60.103a(a)(5). We have also modified the
text in 40 CFR 60.103a(a)(5) regarding flare minimization to clarify that the provisions apply to
"planned startup and shutdown of the refinery process units and ancillary equipment that are
connected to the affected flare." We interpret "ancillary equipment" to include storage vessels.
After considering the options, we determined that including this phrase would best clarify our
intent. Provided that the refinery owner or operator prepares FMP for tank cleaning operations
and follows those plans, then the temporary flare would be compliant with the requirements of
subpart Ja. Consequently, the question becomes whether preparing a FMP is necessary and
appropriate {i.e., BSER) for storage vessel cleaning operations. The additional cost is minimal, as
one flare management protocol can cover multiple storage vessels. While adequate purging of
the storage vessels is necessary, protocols to minimize unnecessary purges to the flare are
expected to reduce both SO2 and NOx emissions and minimize auxiliary natural gas use (needed
to maintain Btu content of the high nitrogen content purge gas). As such, we find that it is
reasonable to require temporary flares used for planned startup and shutdown of process units to
comply with subpart Ja by preparing and following a FMP. Consequently, we are not providing a
general exclusion for temporary flares. While there may be a myriad of reasons why a temporary
flare may be used, given the examples provided by the commenter, we find that only the flares
installed for routine maintenance of an existing flare warrants an exclusion from the applicability
of subpart Ja.
Comment: Commenter 0296 stated that the EPA ignored hydrogen flares in their analysis
and stated that no volatile organic compound (VOC) emissions occur as a result of flaring
hydrogen.
Response: While evaluating the impacts of the rule, we found that reducing flaring,
whether it is a hydrogen flare or hydrocarbon flare, reduces NOx emissions. Hydrogen typically
has value to the refinery {e.g., used in hydrotreating and hydrocracking units), and if the
11
-------
hydrogen is nearly pure, recovery and use of this stream has environmental benefits. If the
hydrogen stream is not pure, then flaring reductions can also lead to emission reductions of SO2
and VOC. As such, the final requirements for flares do not specifically exclude hydrogen flares.
2.1.4 Other Units
Comment: Commenter 0311 stated that fluid catalytic cracking unit (FCCU) startup
heaters should be excluded from the process heater definition. These in-line heaters are used for
100 to 125 hours to heat air that then heats the catalyst and catalyst vessel during FCCU startup.
The commenter believes these are not process heaters as they do not "transfer heat indirectly to
process stream materials" and requested that the EPA clarify the point by specifically excluding
such heaters from the process heater definition.
Response: Based on the limited description of the startup FCCU heater provided by the
commenter, it does appear that this type of unit does not meet the definition of process heater and
would not be subject to the NOx emissions limits. Additional information obtained on an FCCU
heater at the ConocoPhillips Trainer Refinery4 indicates that for at least one of these types of
units, there is no process feed during the FCCU startup process, so there are no "process stream
materials" to receive any heat. However, with respect to SO2 limits, if these in-line heaters use
fuel gas to heat the air, they are unequivocally fuel gas combustion devices and would have to
meet either the SO2 emissions limits or the H2S concentration limits.
We are not making any changes to the final rule to address these types of units. The
owner or operator should use the definitions of "process heater" and "fuel gas combustion
device" in 40 CFR 60.101a to determine whether the unit is an affected source and what actions
are needed to meet the requirements of subpart Ja. If there is any question about whether a
specific unit meets the definition of "process heater" in subpart Ja, the owner or operator should
contact their permitting authority or the EPA for assistance.
Comment: Commenter 0299 requested clarification as to whether the gas generated when
degassing and cleaning a tank is considered fuel gas if it is combusted in an internal combustion
engine or thermal oxidizer. The commenter recommended that subparts J and Ja not apply to
these temporary controls.
4 Wallwork, James F. ConocoPhillips Trainer Refinery Additional Comments on Proposed Plan Approval 23-
0003P. Letter to Sachin Shankar, P.E., Pennsylvania Department of Environmental Protection. October 15, 2009.
12
-------
Response: Based on the current definition of "fuel gas" in subpart J and Ja, these gases
would be considered fuel gas and the internal combustion engine or thermal oxidizer would be a
fuel gas combustion device. However, we believe that these gases can be considered process
upset gases (produced as a result of startup or shutdown). Thus, the rule specifically allows the
use of a flare to combust these gases, but the use of an internal combustion engine or thermal
oxidizer would appear to be prohibited without monitoring the H2S concentration of the fuel gas
and possibly treating the fuel gas to reduce the H2S concentration. The different treatment of
these control systems is difficult to justify. We agree that it is preferable that these tank cleaning
emissions be combusted rather than released directly to the atmosphere. We also find that
internal combustion engines would provide a comparable level of control as a flare with the
added advantage of producing useful work, which can further reduce emissions elsewhere in the
refinery by reducing energy consumption demand. A thermal oxidizer, while providing a
comparable level of control, has the disadvantage that it would require significant excess fuel to
maintain proper temperature in the incinerator, so we find that the thermal oxidizer is a less
advantageous control option. Finally, we agree that, for most tanks, based on distance to a fuel
gas line and the frequency of degassing and cleaning (typically once every 5 to 10 years), it
would not be cost effective to have to capture and treat the relatively small quantities of gas that
are produced infrequently due to tank degassing and cleaning. As such, we have added a
provision to 40 CFR 60.102a(g)(l)(iii) to specifically exempt "combustion in a portable
generator of fuel gas released as a result of tank degassing and/or cleaning."
2.2 Applicability Date
Comment: Commenters 0296, 0306, 0310, 0311 and 0314 suggested that December 22,
2008 (the proposal date of the amendments to subpart Ja), should be the applicability date for the
flare. The commenters stated that the December 22, 2008, proposal date is the date that industry
was provided adequate notice of the new requirements. Commenters 0310 and 0311 further
asserted that the EPA has the discretion and authority to extend the applicability date to
December 22, 2008. Commenter 0306 and 0311 stated that the June 24, 2008, trigger date for
flares is inappropriate because the term "modification" was altered and would result in
retroactive application of the rule. Commenter 0310 also stated that the June 24, 2008, trigger
date for flares would be inconsistent with and interfere with consent decree obligations.
13
-------
Commenter 0314 stated that setting an applicability date before the date the regulations were first
published is "an arbitrary and capricious Agency action."
Commenters 0296, 0311 and 0314 also suggested that the December 22, 2008, proposal
date should be the applicability date for process heater provisions. The commenters stated that
the original proposal date is inappropriate because the final standards were more stringent than
those originally proposed on May 14, 2007. Commenter 0296 noted that the June 24, 2008, final
rule notice included NOx limits of 40 ppmv but suggested that December 22, 2008, is the proper
applicability date, as the NOx provisions in the final rule were stayed. Commenters 0311 and
0314 suggested that the December 22, 2008, proposal date is appropriate because the proposed
standards are fundamentally different than the promulgated standards (e.g., establishment of
different classes of process heaters).
Of particular concern for Commenters 0296 and 0314 is that they began construction on a
new process heater after the original proposal date that was designed and permitted to meet 50
ppmv NOx, as they expected the standard to be 80 ppmv NOx. Commenter 0296 cited an
erroneous statement in the Regulatory Impact Analysis suggesting that 80 ppmv would be
allowed for process heaters that commenced construction after May 14, 2007, but prior to June
24, 2008. Although the unit has not started up yet, based on the design of the process heater,
Commenter 0296 does not expect the unit to meet the 40 ppmv emissions limit. Similarly,
Commenter 0314 noted that the only way for their process heater currently under construction to
meet the 40 ppmv emissions limit is to install selective catalytic reduction (SCR), so they are
making provisions to install SCR even though the EPA already determined that SCR is not
BSER. According to both commenters, either changing the effective date of the requirement or
changing the emissions limit for newly constructed forced draft process heaters would eliminate
this "retroactive noncompliance" issue.
Commenters 0310 and 0311 stated that, if the EPA imposes standards on syngas
produced by coke gasification, then the applicability date of those requirements should be no
earlier than the December 22, 2008, proposal date. Commenter 0311 also indicated that the
trigger date for other newly defined refinery process units (e.g., wastewater treatment) should be
the December 22, 2008, proposal date.
Response: Commenters raise two basic questions: (1) What date applies for determining
whether a flare or a process heater is a "new source" and subject to subpart Ja? (2) When must
14
-------
that "new" flare or process heater be in compliance with the applicable requirements in subpart
Ja?
Section 111(a)(2) of the CAA defines a "new source" as "any stationary source, the
construction or modification of which is commenced after the publication of regulations (or, if
earlier, proposed regulations) prescribing a standard of performance under this section which
will be applicable to such source." An important aspect of this definition is the "notice" it
provides thatNSPS will be prospectively applicable after construction, reconstruction or
modification of an affected source. For NSPS subpart Ja, a "new" process heater is one that
commences construction, reconstruction or modification after May 14, 2007, which is the date of
the "proposed regulations." For the reasons explained in the final rule preamble at 73 FR 35856-
35857, a "new" flare is one that commences construction, reconstruction or modification after
June 24, 2008, which is the date of the "publication of regulations." Therefore, in compliance
with CAA section 111(a)(2), refineries are on "notice" that after these respective dates, "new"
process heaters and "new" flares will be subject to subpart Ja. See 40 CFR 60.100a(b).
Commenters, however, are essentially asserting that refineries were not on "notice" and
cannot now be required to "retroactively" comply with subpart Ja. They assert several legal
arguments for this position, but their basic position boils down to a lack of "notice," which they
assert requires the EPA to move the "trigger date" for subpart Ja applicability to December 22,
2008, the proposal date for the proposed amendments.
With respect to flares, we find no validity in the arguments that the June 24, 2008,
applicability date is inappropriate. The limited changes in the flare modification provision
proposed on December 22, 2008, only identified certain connections that are not considered
modifications of a flare. As such, we see no scenario where the revised flare modification
provision would trigger retroactive application of the rule.
For process heaters, while we recognize that the emissions limits and form of the
standard have changed based on the information gathered during the rulemaking process, the
regulated community was notified that NOx standards were being considered for process heaters
on May 14, 2007. The best system of emission reduction should be implemented at this time. We
recognize that refinery owners and operators desire maximum operating flexibility, but the
proposed rule clearly indicated that controlling excess oxygen amounts is part of BSER, and
process heater design and operation at high excess oxygen content was not considered BSER.
15
-------
While we acknowledge that the initially promulgated 40 ppmv 24-hour NOx emissions limit and
even the final amendments to these standards may limit operator flexibility to a greater extent
than the originally proposed emissions limits, we find that the emissions limits in the final
amendments are achievable and appropriate for newly constructed process heaters.
As discussed in greater detail in Section 2.1.1, the proposed clarifications of fuel gas and
refinery process units in no way impacted the applicability of subpart Ja standards to the units in
question. As such, we see no reason to specify applicability dates for syngas {i.e., fuel gas)
produced by coke gasification. The limitations on sulfur in fuel gas apply to the fuel gas
combustion device or flare as the affected source, not the unit producing the fuel gas. In other
words, if a fuel gas combustion device or flare is an affected source based on the date that
construction, reconstruction or modification commenced, then all the fuel gas combusted by that
device must meet the sulfur standards, regardless of its origin, so an applicability date for coke
gasification is irrelevant. Similarly, we see no basis or need to specify applicability dates for
wastewater treatment systems or loading operations. Gases generated from these units were
always considered to be fuel gas (unless they meet the newly established exemption from the
definition of fuel gas for combustion of vapors in a flare or thermal oxidizer to control
emissions). If these gases were not already considered fuel gas, than this exemption would have
been unnecessary and the industry would not have provided so many comments to expand the
exemptions for different types of fuel gases. Furthermore, as we did not specify any specific
regulations for wastewater treatment units or loading racks under subpart Ja, we do not
understand the commenters' request to change the applicability date.
Comment: Commenter 0305 recommended that 40 CFR 60.100a(c) clarify that: "For the
purpose of this subpart, all projects committed to and subject to compliance with a Consent
Decree filed with the Department of Justice prior to May 14, 2007 are deemed to have
commenced prior to May 14, 2007." The commenter stated that this clarification is needed
because subpart Ja could significantly impact (or create uncertainty in) some projects which have
been long planned and committed to in order to comply with a consent decree.
Response: We decline to make this change, which equates to a request that the EPA find
that the filing date of a consent decree is equivalent or should be deemed equivalent to the date
construction "commences" for all refinery projects covered under consent decrees. Instead,
16
-------
sources should continue to use the following definitions from 40 CFR 60.2 in making this
determination:
Commenced means, with respect to the definition of new source in section
111(a)(2) of the Act, that an owner or operator has undertaken a continuous
program of construction or modification or that an owner or operator has entered
into a contractual obligation to undertake and complete, within a reasonable time,
a continuous program of construction or modification.
Construction means fabrication, erection, or installation of an affected facility.
Modification5 means any physical change in, or change in the method of
operation of, an existing facility which increases the amount of any air pollutant
(to which a standard applies) emitted into the atmosphere by that facility or which
results in the emission of any air pollutant (to which a standard applies) into the
atmosphere not previously emitted.
The definitions provide a reasonable set of criteria for determining when construction or
modification commenced and should be applied on a project-specific basis. Sources that have
any questions regarding how the final amendments to subpart Ja may impact their compliance
with their consent decree should contact the EPA personnel within EPA's Office of Enforcement
and Compliance Assurance (OECA) with whom they negotiated the consent decree.
However, we have provided additional compliance time for owners and operators of
certain flares. Owners and operators of modified flares that are have accepted applicability of
subpart J under a federal consent decree shall comply with the subpart J requirements as
specified in the consent decree, but must meet the short-term 162 ppmv H2S concentration limit
no later than 3 years after the effective date of the final amendments. In addition, modified flares
that are already subject to the short-term H2S concentration limit but that have an approved
monitoring alternative under subpart J and do not have the monitoring equipment in-place that is
required under subpart Ja shall be given up to 3 years from the effective date of this final rule to
install the monitors required by subpart Ja (or to obtain an approved monitoring alternative under
subpart Ja). In this interim period, owners and operators of these modified flares shall
demonstrate compliance with the short-term H2S concentration limit using the monitoring
alternative approved under subpart J.
5 Sources should also reference CAA section 111(a)(4) and 40 CFR 60.14 when determining whether a
"modification" has occurred.
17
-------
Comment: Commenter 0315 requested clarification regarding the timing of potential
modifications. The proposed rule text, read literally, could be interpreted to mean that the date
the new piping is connected to the flare header is the date an owner or operator should use to
determine if that connection is a modification. However, in most cases, most of the work on the
project has occurred before the actual connection is made, partly because accomplishing the
connection and allowing flare header hydrocarbon to flow into new equipment poses a major
safety risk. The commenter also noted that this literal interpretation of the date of a modification
is inconsistent with the NSPS General Provisions at 40 CFR 60.2, which states: "Commenced
means... that an owner or operator has undertaken a continuous program of construction or
modification or that an owner or operator has entered into a contractual obligation to undertake
and complete, within a reasonable time, a continuous program of construction or modification."
The commenter requested that the final rule clearly specify that, for projects where a flare
connection is part of an entire project, the potential flare modification is considered to have
begun upon the commencement or contracting of the project (i.e., a continuous program of
construction or modification that happens to include a new connection to a flare).
Response: It is not our intention to apply a definition of "commenced" to subpart Ja that
is inconsistent with that term as it is defined in the General Provisions. As explained in the
previous response above, sources should continue using the definitions of "commenced,"
"construction" and "modification"6 from the General Provisions for determining when a
particular project "commences." Using those definitions, it already appears that the date a source
would use for evaluating whether a new connection to a flare is a "modification" of that flare
would be the date that the entire project "commences," not necessarily the date the piping is
physically connected to the flare header. However, we recognize that the "new connection"
language in the flare modification provision could be misconstrued in this context; therefore, we
have added a sentence to 40 CFR 60.100a(b) that states: "For the purposes of this subpart, a
modification to a flare commences when a project that includes any of the activities in
paragraphs (c)(1) or (2) of this section is commenced."
6 See footnote 4.
18
-------
2.3 Flare Modification
Comment: Commenter 0308 objected to the special modification provision for flares in
subpart Ja. The commenter stated that the provisions in 40 CFR 60.14 can and should apply to
flares. Specifically, the EPA should consider the ".. .hourly emission rate, at maximum physical
capacity, before and after an operational change" (57 FR 32316). Because the maximum physical
capacity of the flare, even when considering the header system as part of the flare, is not changed
by additional tie-ins, new tie-ins should not constitute a modification, according to
Commenters 0308 and 0311. Furthermore, the commenters asserted that the NSPS program was
never intended to apply across an entire industry based on minor changes of an affected facility.
Commenter 0310 stated that the flare modification provision is unlawful because routine
connections generally do not result in an emissions increase and may be seen as having the
primary purpose of reducing emissions, which has been excluded under 40 CFR 60.14(e)(5), a
paragraph that is not limited to pollutants "to which the standard is applicable." Commenter 0311
suggested a narrower flare modification provision stating that only connections that "result in a
net increase in the amount of VOC or SO2 emitted during normal operations" are considered
modifications. Commenter 0308 noted that a single project may remove some tie-ins and add
some tie-ins, so that the net emissions could actually be reduced, and requested clarification if it
was the EPA's intent for such projects to constitute a flare modification.
Commenter 0310 stated that the flare modification provision in 40 CFR 60.100a(c)(2)
should specify that the increase in flow capacity is an increase in the total volume of gas under
standard conditions that could reach the flare and that it is specific to flow increases of VOC or
sulfur compounds. Commenter 0309 suggested that most of the exemptions could be reduced
down by focusing the rule on SO2 emissions and exempting any connection of fuel gas that has a
H2S concentration of less than 162 ppmv. Commenter 0305 stated that "flow" is not a regulated
pollutant, so an increase in flow should not be considered a modification. Commenter 0314
agreed that examining what is a flare modification is warranted and agreed that projects that
increase production through new process vents to flares or projects that increase process vent
flow rates to flares should be subject to subpart Ja. Commenter 0309 stated that the proposed
flare modification provision is too broad and suggested that the EPA should focus the flare
modification provision on connections that provide a primary/routine flow from a process unit to
the flare, such as delayed coking unit offgas.
19
-------
Response: The agency made a conscious decision to promulgate a separate provision for
a flare modification in 40 CFR part 60, subpart Ja (see 40 CFR 60.14(f)) because flares are
operated differently from other refinery process units, making it difficult to apply the
modification provision in the General Provisions (40 CFR 60.14) to them. The physical capacity
of a flare is based on the amount of gas potentially discharged to a flare as a result of emergency
relief. Refiners frequently make connections to existing flares that result in emissions increases
at the flares, but may never approach the physical capacity of the flare system. Contrary to
commenters' assertions, the flare modification provision in 40 CFR 60.100a(c) does meet the
statutory definition of "modification" in CAA section 111(a)(4), which is "any physical change
in, or change in the method of operation of, a stationary source which increases the amount of
any air pollutant emitted by such source or which results in the emission of any air pollutant not
previously emitted." It is axiomatic that the connections to the flare described in 40 CFR
60.100a(c) qualify as physical or operational changes to the flare. Additionally, we explained in
the proposed rule how these connections also resulted in emissions increases from the flare (see
73 FR 78529). Thus, these types of new connections of refinery process units (including
ancillary equipment) and fuel gas systems to the flare qualify as a "modification" of the flare and
trigger subpart Ja applicability for the flare.
Those connections we identified that do not increase emissions from the flare were
specifically excluded from triggering 40 CFR part 60, subpart Ja applicability under this same
provision (see 40 CFR 60.100a(c)(l)). Specifically, we proposed on December 22, 2008, that the
following types of connections to a flare would not be considered a modification of the flare:
(1) connections made to install monitoring systems to the flares; (2) connections made to install a
flare gas recovery (FGR) system; (3) connections made to replace or upgrade existing pressure
relief or safety valves, provided the new pressure relief or safety valve has a set point opening
pressure no lower and an internal diameter no greater than the existing equipment being replaced
or upgraded; and (4) replacing piping or moving an existing connection from a refinery process
unit to a new location in the same flare, provided the new pipe diameter is less than or equal to
the diameter of the pipe/connection being replaced/moved. While we agree that there may be
other connections to a flare that would not result in an emissions increase from the flare (see
response to the next comment for specific details), we disagree with the commenters that the
20
-------
flare modification provision should be further limited beyond what is already provided in the
provision.
We disagree with commenters that we must consider the "net" emissions from the
process unit and the flare when determining whether a flare is modified. The affected facility is
the flare and does not include the process units that are tied into the flare header system. See
Asarco v. EPA, 578 F.2d 319, 325(D.C. Cir. 1978) (holding that emission increases had to be
determined based on emissions from the affected facility). We also disagree that a modification
determination should be limited to emissions increases of VOC or SO2. Flares are known to emit
VOC, SO2, CO, PM and NOx, as well as other air pollutants, all of which are relevant when
determining whether a flare has been modified. See CAA section 111(a)(4). That is, we consider
the standards for flares to be emission standards for VOC, S02, CO, PM and NOx. See,
generally, 73 FR 35838, 35842, 35854-35856 (June 24, 2008); 73 FR 78522, 78533 (December
22, 2008), as well as Table 4 of the preamble to the final amendments. Using the flare to control
VOC emissions at other refinery process units will increase CO, PM and NOx emissions from
the flare and are, therefore, considered modifications of the flare, even if there is a net reduction
in VOC emissions at the refinery.
In evaluating whether a flare has been modified, we consider increases in flow to the flare
to be directly indicative of increased emissions from the flare. While we agree that "flow" is not
a pollutant, we evaluated flow limits as a means to reduce SO2, VOC, CO, NOx and other
emissions from the flare. The emissions from the flare are very difficult, if not impossible, to
measure accurately, but flow to the flare can be measured, and the flow to the flare generates
SO2, VOC, CO, PM, NOx and other emissions. Therefore, a physical or operational change to a
flare that causes an increase of flow to the flare will increase emissions of at least one of these
pollutants and is considered a modification of the flare.
Comment: Commenters 0298, 0301, 0302, 0303, 0305, 0306, 0309, 0311 and 0314
suggested that the EPA should specifically exempt the following situations from the flare
modification provision:
(1) connections made to upgrade or enhance (not just to install) a FGR system
(Commenters 0301 and 0311);
(2) new flare header interconnect lines, since the connection is between flares and not
from a process unit or fuel gas system to a flare (Commenters 0308, 0310 and 0311);
21
-------
(3) connections made for flare gas sulfur removal (Commenter 0311; these reduce
emissions);
(4) connections made to install back-up equipment (Commenter 0311; no impact on
emissions);
(5) all emergency pressure relief valve (PRV) connections from existing equipment
because the rule may create a disincentive to improving safety (Commenter 0298,
0305 and 0311) and the upset gases released by these valves are exempt from the
emissions standard (Commenter 0301);
(6) connections of monitoring system purge gases and analyzer exhausts (Commenters
0298 and 0301) or closed vent sampling systems (Commenter 0314);
(7) purge and clearing vapors, block and bleeder vents and other uncombusted vapors
where the flare is the control device (Commenter 0301 and 0306);
(8) connections made to comply with other federal, state or local rules where the flare is
the control device (Commenters 0301 and 0314);
(9) connections of "unregulated gases" such as hydrogen, nitrogen, ammonia, other non-
hydrocarbon gases or natural gas" (Commenters 0301, 0305, 0309 and 0311) or any
connection that is not "fuel gas" (Commenter 0311);
(10) new connections upstream of an existing FGR system (Commenters 0301, 0303,
0305 and 0314), provided the new connections do not compromise or exceed the FGR
system's capacity (Commenter 0301);
(11) any new, moved or replaced piping connections that do not result in a net increase in
emissions (Commenter 0311 specifically indicated increases in VOC or S02
emissions) from the flare, regardless of piping size (Commenter 0301, 0305, 0306,
0311 and 0314); Commenter 0301 suggested a larger PRV is such an example, stating
that the larger PRV does not change the source or mass emissions, it just allows the
pressure to relieve more quickly;
(12) vapors from tanks used to store sweet or treated products (Commenter 0309);
(13) temporary connections for purging existing equipment (Commenters 0311 and 0314;
these are essentially "existing" connections); and
(14) connections of safety instrumentation systems (SIS) described under Occupational
Safety and Health Administration (OSHA) process safety standards at 29 CFR
1910.119, the EPA's risk management program at 49 CFR 68 and/or American
National Standards Institute (ANSI)/International Society of Automation (ISA)-
84.00.01-2004 (Commenter 0314; the purpose of these systems is for safety and can
often minimize process upsets when they do happen).
Commenter 0315 stated that for existing flares with sufficient FGR capacity, the EPA
should make clear that sites are allowed to make new connections, tie-ins and other changes to
their FGR systems as long as the new connections will not compromise or exceed the system's
capacity to recover the new flow without triggering a flare modification. Commenter 0301 noted
22
-------
that a safety project to add a new pressure relief valve cannot carry the costs of a FGR system or
additional amine treatment. According to the commenters, the EPA should encourage venting
these gases to the flare rather than venting them to the atmosphere or compromising safety.
Response: We carefully reviewed the commenters' suggested changes to the flare
modification provision to determine whether there are additional connections that should not be
considered modifications to the flare. We agree that the first four connections in the commenters'
list should not be considered modifications of a flare. Projects to upgrade or enhance components
of a FGR system (e.g., addition of compressors or recycle lines) will improve the operation of
the FGR system, and connections to these additional components will not result in increased
emissions. Connections made for removal of sulfur from flare gas (Item 2 above) will generally
result in a slight decrease in volumetric flow and a large decrease in emissions of S02.
Connections made to install back-up or redundant equipment (Item 3 above), such as a back-up
compressor, will result in fewer released emissions if there is a malfunction in the main
equipment.
The request to exclude flare interconnections (Item 4 above) is a complicated issue
because interconnecting two separate flares alters what we consider to be the affected facility.
The definition of "flare" specifically includes the flare gas header system as part of the flare.
Prior to interconnecting the flares, presumably each flare header system is independent, and there
would be two separate "flares," each of which could potentially be an affected facility subject to
40 CFR part 60, subpart Ja. However, because the flare includes the flare header system, we
consider that an interconnected flare system is a single affected facility, and we have amended
the definition of "flare" for clarity. We agree that interconnections between flares will not alter
the cumulative amount of gas being flared (i.e., interconnecting two flares does not result in an
emissions increase relative to the two single flares prior to interconnection). We also see cases
where the emissions from a single flare tip will likely be reduced due to the flare interconnect.
For example, when a large release event occurs, this gas will now flow to both of the
interconnected flares rather than a single flare. The maximum emission rate for the original
single flare actually decreases, while the combined emissions from both flares is the same
quantity as prior to the interconnection. Considering this, we agree that the interconnection of
two flares does not necessarily result in a modification of the flare and we have specifically
excluded flare interconnections from the modification provisions.
23
-------
However, we also clarify in this response that when a flare that is subject to 40 CFR part
60, subpart Ja is interconnected with a flare that is not subject to subpart Ja, then the resulting
interconnected flare is subject to subpart Ja. That is, the only case in which an interconnection
between two (or more) flares results in a combined, interconnected flare that is not subject to
subpart Ja is when none of the original individual flares were subject to subpart Ja. Additionally,
we note that if a new connection is made to the interconnected flare, then the flare (including
each individual flare tip within the interconnected flare header system) is modified and becomes
an affected facility subject to subpart Ja.
While we agree that connections that do not increase the emissions from the flare should
not trigger a modification, we disagree with the commenter that their other suggested
connections do not increase the flare's emissions at the time gases are discharged via the new
connection. Each of the commenters' suggestions is discussed in the following paragraphs.
We previously proposed an exemption for emergency pressure relief valve connections
from existing equipment (Item 5 above) if they replace or upgrade existing equipment and do not
increase the instantaneous release rate to the flare {i.e., the new pressure relief valve has a
pressure set point and diameter no greater than the equipment being replaced). As stated
previously in this document, we are finalizing that amendment, as proposed. However, new
connections, even if they are made to "existing equipment," will result in an increase in flow to
the flare during periods of process upset that cause the pressure relief valve to open.
Connections of monitoring system purge gases and analyzer exhausts or closed vent
sampling systems (Item 6 above) will increase the emissions from the flare. Similarly,
connections of purge and clearing vapors and block and bleeder vents (Item 7 above), also
trigger a modification of the flare because the increase of gas flow to the flare will increase the
emissions from the flare.
We recognize that connections to a flare may be made to comply with other federal, state
or local rules where the flare is an emissions control device (Item 8 above). In fact, nearly all
flares could be considered "control devices." We agree that using a flare as an emissions control
device is preferable to venting the process unit to the atmosphere. However, while using the flare
as an emissions control device does decrease emissions from the process unit being controlled,
the increase of gas flow to the flare will increase the emissions from the flare. Therefore, a
24
-------
connection from a process unit to a flare for use as an emissions control device results in a
modification of that flare.
Comments suggesting that connections of "unregulated gases" such as hydrogen,
nitrogen, ammonia, other non-hydrocarbon gases or natural gas or connections that are not "fuel
gas," should not be considered a modification of the flare (Item 9 above) are in conflict with the
statutory definition of "modification." Each of the streams mentioned by the commenter, when
directed to a flare, will increase emissions of at least one pollutant (either PM, CO or NOx) from
the flare (all of which the standard is intended to reduce). That is, we reiterate that we consider
the standards for flares to be emission standards for VOC, SO2, CO, PM and NOx. As such, we
do not agree that the types of gas streams suggested by the commenters should be exempt from
the modification determination.
New connections upstream of an existing FGR system (Item 10 above) will increase the
likelihood of an event that would cause an exceedance of the FGR system's capacity (even if the
new connections "do not exceed the FGR system's capacity" under normal conditions), and the
amount of gases sent to the flare would increase as a result of such an event, thereby increasing
the emissions from the flare.
We reiterate that we proposed an exemption for any moved or replaced piping or pressure
relief valve connections of the same size. However, we disagree with the commenter's
suggestion that any "new, moved, or replaced piping or pressure relief valve connections that do
not result in a net increase in emissions from the flare regardless of piping or pressure relief
valve size" should be exempted (Item 11 above). The premise of the suggested amendment is
that new or larger connections somehow will not increase emissions from the flare. We have
discussed new connections previously, so we will concentrate on the "regardless of piping or
pressure relief valve size" comment in this paragraph. First, the size of the pressure relief valve
or piping does correlate to the discharge rate to the flare, with larger pressure relief valves or
larger diameter piping allowing higher discharge rates to the flare at a given pressure. In fact,
larger pressure relief valves and larger diameter pipes are specifically designed to allow higher
flow rates to the flare. Second, higher flow rates will lead to higher emission rates. For a pressure
relief event that occurs for several hours, the flow rate to the flare during the first hour of relief
using the larger pressure relief valve or larger diameter piping will be larger than the flow rate
experienced using the smaller pressure relief valve or smaller diameter piping and will result in
25
-------
higher emissions from the flare. Therefore, we reject the notion that larger diameter pipes and
larger pressure relief valves do not increase the emissions rate from the flare during a release
event. We are finalizing the proposed exemptions for moved or replaced piping or pressure relief
valves with the size and design restrictions for the new piping or pressure relief valves as
proposed on December 22, 2008.
Commenters suggested that connections of vapors from tanks used to store sweet or
treated products (Item 12 above) should not be modifications because those gas streams have less
than 162 ppmv H2S. We reiterate that SO2 is not the only pollutant emitted from flares and that
the additional flow of sweet gases will increase the emissions of at least one pollutant from the
flare, so we are not exempting these types of connections to the flare from the 40 CFR part 60,
subpart Ja flare modification provision. However, we have amended the sulfur monitoring
requirements for flares to exempt vapors from tanks used to store sweet or treated products from
the flare sulfur monitoring requirements. This monitoring exemption is justified because it is not
needed for the purposes of a RCA or other compliance purpose. For these sweet vapors, the flow
rate RCA threshold will be exceeded well before the S02 RCA threshold.
We carefully considered temporary connections for purging existing equipment (Item 13
above), but we failed to see how these temporary connections are essentially "existing
connections." According to the commenters, "maintenance gases have been routed in some form
or other to the flare for years, and the temporary tie-in to accomplish that is not a change and is
not an increase in emissions when viewed from a before and after perspective." If the
connections already exist, then opening an existing valve to allow for this type of purging would
not trigger a flare modification. If the connection is being relocated and the piping used is the
same diameter as the pre-existing connection, then this scenario is adequately covered by the
proposed exclusion for relocated connections. However, if a new connection is made specifically
to purge an existing piece of equipment, this purge gas unequivocally represents additional gas
flow sent to the flare that did not exist and could not exist prior to the connection being made.
Again, we consider that the increase in gas flow to the flare will result in an increase in
emissions of at least one pollutant from the flare. As such, no exemption is provided for new
connections to existing equipment, regardless if these connections are temporary or permanent.
We also find that these types of flows should be expressly considered in the FMP and that flaring
from these "temporary" connections should be minimized to the extent practicable.
26
-------
The impact of connections of SIS described under OSHA process safety standards at
29 CFR 1910.119, the EPA's risk management program at 49 CFR 68 and ANSI/ISA-84.00.01-
2004 (Item 14 above) should be evaluated on a case-by-case basis to determine whether these
connections result in a flare modification. We expect that, if these connections are made for flare
monitoring purposes, these connections are already excluded in the exemption for flare
monitoring systems. If the "SIS" are process unit analyzers and the new connections are being
made to connect the analyzer exhaust to the flare, these connections would be considered a
modification, as previously discussed. The commenter may also be referring to new connections
for additional pressure relief valves identified in the safety reviews required by the cited rules,
which we would consider to be a modification of the flare.
Following all of the above review and analysis, we are finalizing three of the connections
as proposed, adding three of the connections requested by commenters and revising one of the
proposed connections as requested by commenters in 40 CFR 60.100a(c)(l). Thus, the following
seven types of connections are not considered a modification of the flare:
(1) Connections made to install monitoring systems to the flare.
(2) Connections made to install a FGR system or connections made to upgrade or
enhance components of a FGR system (e.g., addition of compressors or recycle lines).
(3) Connections made to replace or upgrade existing pressure relief or safety valves,
provided the new pressure relief or safety valve has a set point opening pressure no lower and an
internal diameter no greater than the existing equipment being replaced or upgraded.
(4) Connections that interconnect two or more flares.
(5) Connections made for flare gas sulfur removal.
(6) Connections made to install back-up (redundant) equipment associated with the flare
(such as a back-up compressor) that does not increase the capacity of the flare.
(7) Replacing piping or moving an existing connection from a refinery process unit to a
new location in the same flare, provided the new pipe diameter is less than or equal to the
diameter of the pipe/connection being replaced/moved.
Comment: Commenter 0309 stated that the flare modification provision in
40 CFR 60.100a(c)(2) should explicitly exclude the following physical changes made to the
flare: (1) change in flare pilot and igniter systems (including larger pilot lines or those that use
more gas); (2) installation of sweep or purge gas reduction device (e.g., an internal ring or
27
-------
molecular seal); (3) increase in steam capacity (to limit smoke); (4) alteration of flare seal drum
level or control scheme; and (5) connections to increase sweep or purge gas rates provided the
sweep or purge gas has an H2S concentration of less than 162 ppmv on a 3-hour rolling average
basis.
Response: Under 40 CFR 60.100a(c)(2), a flare is modified when it "is physically altered
to increase the flow capacity of the flare." We did not propose to revise this provision, nor do we
think revisions are necessary. We do clarify in this response that the "flow capacity of the flare"
is intended to refer to the capacity of flare gas {i.e., gas from the "flare gas header system,"
which is defined in the amended rule) that can be safely discharged to the flare because this kind
of change most likely results in an emissions increase from the flare. This is consistent with the
definition of "modification" in CAA section 111(a)(4), which is "any physical change in, or
change in the method of operation of, a stationary source which increases the amount of any air
pollutant emitted by such source or which results in the emission of any air pollutant not
previously emitted." Alterations of the pilot and igniter systems or the steam system would not
typically constitute a "modification" of the flare because these alterations do not typically
increase the flare gas flow capacity of the flare. However, if these physical alterations do
increase the flow capacity of the flare {e.g., larger pilots or increased steam ducts installed to
allow higher flare gas flow rates to be handled by the flare, thereby increasing the flow capacity
of the flare), which results in an emissions increase, then these alterations would be deemed a
"modification" of the flare under 40 CFR 60.100a(c)(2). Similarly, installation of sweep or purge
gas reduction devices or flare seal drum controls are not expected to increase the flare gas flow
capacity of the flare and would not generally be a "modification" of the flare. However, if these
physical changes do increase the flare gas flow capacity of the flare, which results in an
emissions increase, then these alterations are "modifications" of the flare.
While the exemptions specified by the commenter may not be flare modifications under
40 CFR 60.100a(c)(2), a few of these examples may be considered flare modifications under
40 CFR 60.100a(c)(l). Specifically, Item 5, connections to increase sweep or purge gas rates,
would be considered a modification of the flare if these gases originate from a process unit or
fuel gas system. However, if these connections are from a separate, independent natural gas line,
for example, these connections would not be considered a flare modification under
40 CFR 60.100a(c)(l). As noted in this response, we disagree that a blanket exemption of the
28
-------
modifications requested by the commenter is appropriate. This response, while clarifying the
intent of the rule's language, also indicates the need to evaluate potential modifications on a
case-by-case basis. If there is any question about whether an alteration of the flare should be
considered a modification of the flare based on the subpart Ja flare modification provision, the
owner or operator should contact their permitting authority or the EPA for assistance.
Comment: Commenters 0298, 0301, 0302, 0309, 0311 and 0314 suggested that the EPA
should specifically state that de minimis emission increases (or net emission decreases) resulting
from new connections to a flare made to control and combust fugitive emissions such as leaks
from compressor seals, valves or pumps, are not flare modifications. Commenter 0302 suggested
allowing site-specific exemptions for connections that do not increase emissions or that result in
a de minimis emissions increase. Conversely, Commenter 0313 objected to setting a de minimis
emissions increase below which a modification would not trigger subpart Ja applicability. The
commenter noted that the specific exclusions proposed by the EPA are acceptable because they
arguably cause no emission increases. However, emissions from a flare are difficult to determine,
particularly at low flow rates, and allowing a de minimis approach would cause confusion over
the applicability of subpart Ja.
Response: In the preamble to our proposed amendments, the EPA specifically requested
comment on using the de minimis exception in the flare modification provision. 73 FR 78522,
78529. Industry Petitioners had suggested some type of de minimis emissions increase should be
allowed without triggering 40 CFR part 60, subpart Ja applicability. Id. The EPA acknowledged
that these exceptions are "permissible but not required" under the modification provision in the
CAA. Id. The EPA also stated: "We request comments on a de minimis approach and on specific
changes that may occur to flares that will result in de minimis increases in emissions. We also
request comments on the type, number and amount of emissions that would be considered de
minimis." Id.
Industry Petitioners continue to recommend that any emissions increases resulting from
"routine connections" to the flare system "will be de minimis" and should not trigger 40 CFR
part 60, subpart Ja applicability at the flare, but they have not provided the comments or data
requested in the proposal preamble that the EPA could consider to evaluate the impacts of such
an approach. Docket Item No. EPA-HQ-OAR-2007-0011-0311 (second attachment), pg 20.
Industry Petitioners again suggest that the EPA exercise its authority and "authorize exceptions
29
-------
from otherwise clear statutory mandates" by promulgating de minimis exemptions for the flare
modification provision. Id:, Alabama Power Co. v. Costle, 636 F.2d 323, 360 (D.C. Cir. 1979).
As explained in Alabama Power, the de minimis exception allows agency flexibility in
interpreting a statute to prevent "pointless expenditures of effort." Id. However, as Industry
Petitioners recognize, nothing mandates that the EPA use its de minimis authority in any given
instance, and courts especially recognize the significant deference due an agency's use of a de
minimis exception. Id. at 400; Shays v. Federal Election Com '//, 414 F.3d 76, 113 (D.C. Cir.
2005); Environmental Defense Fund, Inc. v. EPA, 82 F.3d 451, 466 (D.C. Cir. 1996); Ass'// of
Admin. Law Judges v. Fed. Labor Relations Auth., 397 F.3d 957, 961 (D.C. Cir. 2005).
In exercising that discretion, the EPA must consider the cautionary advice it received
from the Alabama Court regarding its use of the de minimis exception: "EPA must take into
account in any action... that this exemption authority is narrow in reach and tightly bounded by
the need to show that the situation is genuinely de minimis ." Id. at 361. The Court also noted that
exemptions from "the clear commands of a regulatory statute, though sometimes permitted, are
not favored." Id. at 358. The EPA must exercise this authority cautiously, and only in those
circumstances that truly warrant its application.
The EPA has found no basis for promulgating a de minimis exception to the flare
modification provision. Despite its assertions, Industry Petitioners have still provided no data to
support a finding that the emissions increases resulting from the alleged "routine connections" to
a flare system are truly "trivial or [of] no value." Docket Item No. EPA-HQ-OAR-2007-0011-
0311 (second attachment), pg 20. Without the requested information showing that "the situation
is genuinely de minimis," Alabama Power, 636 F.2d at 361 and, therefore, warrants this kind of
exception, we believe such an exemption would be inappropriate.
Additionally, Industry Petitioners' example that "venting a new small storage tank to a
flare system... easily would cost a typical refinery tens of millions of dollars" since "the entire
flare system" (emphasis in original) would be subject to subpart Ja is unavailing for its argument
that the EPA should promulgate a de minimis exception for the flare modification provision.
Docket Item No. EPA-HQ-OAR-2007-0011-0311 (second attachment), pg 21. As the District of
Columbia Circuit specifically states in Shays, authority for promulgating a de minimis exception
"does not extend to a situation where the regulatory function does provide benefits, in the sense
of furthering regulatory objectives, but the agency concludes the acknowledged benefits are
30
-------
exceeded by the costs." Shays, 414 F.3d 76, 114 (emphasis added). By focusing solely on cost,
Industry Petitioners are effectively asking the agency to engage in the type of cost-benefit
analysis prohibited by the Shays Court. Such cost analyses are improper in these types of
decisions. Industry Petitioners generally focus their discussion on VOC emissions and effectively
admit that connecting the small storage tank to the flare system increases emissions from the
flare (e.g., "uncontrolled tank emissions would be essentially eliminated by combustion in a
flare" (Docket Item No. EPA-HQ-OAR-2007-0011-0311 (second attachment), pg 21, emphasis
added)). Furthermore, they disregard additional emissions of NOx and CO resulting from the
combustion of these gases at the flare. Industry Petitioners also provide no data quantifying these
emissions increases and, therefore, cannot demonstrate that they are "trivial or [of] no value" or,
in other words, that the emissions increases are, in fact, de minimis. As releases to the flare are
often event driven, one can envision situations where the release from even a small storage tank
could be significant. On the other hand, the EPA sees a substantial environmental benefit in
requiring controls that will reduce the cumulative emissions from a flare that becomes subject to
40 CFR part 60, subpart Ja because of any of these alleged "routine connections." Thus, given
the nature of releases to the flare, we determined that a de minimis exemption from the
modification provisions for flares is unworkable and unwarranted.
Comment: Commenter 0315 expanded on their original comments (0311) and addressed
the comparison between the flare modification provision in subpart Ja and modification of a
municipal solid waste (MSW) landfill in the direct final amendments to Standards of
Performance for New Stationary Sources and Guidelines for Control of Existing Sources:
Municipal Solid Waste Landfills (63 FR 32743, June 16, 1998). The commenter noted that the
preamble to that rule explains that the modification provision "is specific to landfills but is
consistent with the intent of section 60.14 of the NSPS General Provisions." "Production" is
considered to be the amount of waste placed in the landfill, and "total 'production' for the entire
life of the facility is controlled through the amount of design capacity specified in the permit."
The commenter noted that according to the preamble, only "physical or operational changes that
increase the size of the landfill beyond its permitted capacity" are considered modifications for
purposes of NSPS. The preamble also stated that "operational changes" at a landfill are typically
accomplished without capital expenditure; therefore, they are not included in the modification
provision. Finally, the commenter noted that the preamble to the proposal stated that "if an MSW
31
-------
landfill increased its waste acceptance rate, such a change would be analogous to an increase in
production rate at a manufacturing facility" which is excluded under 40 CFR 60.14. Therefore,
the commenter asserted, the MSW modification provision is consistent with 40 CFR 60.14, while
the subpart Ja flare modification provision is a departure from 40 CFR 60.14.
The commenter made several points in comparing the MSW landfill modification
provision to their recommendations for the flare modification provision in subpart Ja. First, the
commenter stated that the provision that a modification occurs only if the design capacity of the
landfill is changed is consistent with the suggestion that a new connection to a flare is only a
modification if other changes are made to the flare system that would increase the capacity of the
flare. Second, using the landfill in its intended function and continually adding new waste is not
considered a series of physical changes, each adding a small increment to the potential emissions
of the landfill; therefore, incremental additions to a flare system should not constitute
modifications to the flare system for purposes of subpart Ja. Third, since the longstanding
exclusions to the 40 CFR 60.14 definition of modification, such as excluding operational
changes that do not require capital expenditure, were included in the MSW landfill modification
provision, they should be available for the subpart Ja flare modification provision. (The
commenter noted that the MSW landfill rule actually goes beyond the 40 CFR 60.14 exclusion
by allowing any operational changes to be excluded whether or not they actually have a capital
expenditure.) The commenter concluded that the EPA's rationale for the MSW landfill
modification provisions provides support for their suggested modifications to the flare
modification provision in subpart Ja.
Response: In developing the modification provisions for MSW landfills, we evaluated the
special circumstances associated with landfills and developed a modification provision specific
to landfills. In exactly the same manner, we evaluated the special circumstances associated with
flares and developed a modification provision specific to flares. In both cases, we considered the
same underlying principles of what constitutes a "modification" based on the provisions in CAA
section 111(a)(4), 40 CFR 60.2 and 40 CFR 60.14. We arrived at different modification
provisions for MSW landfills and flares because of the many factual and operational differences
for these two very different sources. We therefore do not consider it appropriate to make the
changes to the flare modification provision that are suggested by the commenter.
32
-------
Comment: Commenter 0308 requested clarification with respect to how the flare
modification provision applies to flare interconnects, which are used to direct flow from one
flare to another flare if one flare is temporarily out of service. The commenter requested
concurrence that a new connection to "flare header system A" only modifies "flare A" and not to
the interconnected "flare B" since the systems are generally operated independently.
Response: We disagree that only the first or primary flare in an interconnected flare
system should be subject to subpart Ja when a new connection to a process unit or fuel gas
system is made. New connections to the interconnected flare system would suggest that
additional gas could be sent to any of the flares in the interconnected system. In a cascaded
system, the new connection increases the likelihood that gas will be sent to the secondary flare.
As such, a new connection to the flare gas system would trigger the applicability of subpart Ja
for both (or all) interconnected flares.
For flares that are used only for back-up or redundant systems to the primary flare {i.e.,
for staged flares that are only triggered when the flare gas pressure at the primary flare exceeds a
certain pressure), the final amendments provide a simplified water seal monitoring approach
under 40 CFR 60.107a(g) to ensure gas is not sent to the secondary flare. It is anticipated that
any event that causes a release so large as to trigger gas flow to subsequent flares in the series
would trigger a RCA on the primary flare. The root cause of the flaring event for the back-up
flares would be the same as the primary flare, so it would be redundant to require additional
RCA for the back-up flares. Therefore, the final amendments do not require duplicative RCA for
the secondary flare if a RCA is triggered for the primary flare. However, if flare gas is sent to the
secondary flare (as detected by the water seal monitoring) prior to triggering a RCA on the
primary flare, then the indication of flow to the secondary flare automatically triggers a RCA for
the primary flare. If the interconnection does not have a water seal or other device that prevents
flow to the interconnected flare during normal operation, then a new connection to an
interconnected flare header system would require flow and sulfur monitoring for each flare
within the interconnected header system.
2.4 Notification Schedule
Comment: Commenter 0295 requested clarification that the 40 CFR 60.7 notifications for
construction or startup are not due until the stay is lifted, even if these activities occur during the
33
-------
stay. Similarly, Commenter 0311 requested that the 40 CFR 60.108a and 40 CFR 60.7
notifications for construction or startup be due 60 days after the stay is lifted and the final rule
published.
Response: Notification for construction and startup should be provided at the time of
construction or startup, as applicable. These requirements existed under subpart J and there is no
reason to delay these notifications.
2.5 Overlap with and Effect on Other Subparts
Comment: Commenter 0311 noted that 40 CFR part 63, subpart UUU (Refinery
MACT 2) often references specific subpart J requirements, resulting in a refinery potentially
having to meet both subparts J and Ja for many sources. Commenter 0311 suggested adding the
following sentence to 40 CFR 60.100a(e) of subpart Ja: "Owners or operators may also choose to
comply with the applicable provisions of Subpart Ja of this part to satisfy the requirements of
Part 63 Subpart UUU of this chapter, where Subpart UUU references this subpart J or individual
sections of this subpart."
Response: While we understand the desire to allow compliance with subpart Ja to fulfill
the requirements of Refinery MACT 2, it is inappropriate to effectively put the suggested
Refinery MACT 2 alternative in subpart Ja. It is much more appropriate to include this
alternative directly in subpart UUU. Refinery MACT 2 is currently under review, and we intend
to include subpart Ja as a compliance option for subpart UUU (where appropriate) in future
amendments to subpart UUU.
Comment: Commenters 0304 and 0311 supported the EPA's proposal at
40 CFR 60.100(e) of subpart J to allow affected sources subject to subpart J to elect to comply
with the corresponding provisions of subpart Ja. However, Commenter 0304 requested
clarification as to whether the "election" is only available on a "whole refinery" basis or if the
election can be made on an "affected source" basis.
Response: The election would be made on an individual affected facility basis, not the
entire refinery. For example, if a facility has two FCCU subject to subpart J requirements, the
owner or operator may elect to comply with subpart J requirements for one FCCU and subpart Ja
requirements for the second FCCU.
34
-------
Comment: Commenter 0309 recommended that the EPA explicitly include in subpart J
the fuel gas H2S concentration limit in units of ppmv that is equivalent to 230 milligrams per dry
standard cubic meter (mg/dscm) (as it did in subpart Ja), since concentration is measured in units
of ppmv and the proper conversion has been a source of significant debate.
Response: We specified the fuel gas combustion device H2S concentration limit in
subpart Ja in units of ppmv because the continuous monitors used to demonstrate compliance
generally output the H2S concentration in ppmv. We expect that the inclusion of the short-term
162 ppmv H2S concentration limit in subpart Ja should help to resolve debates about the proper
conversion. However, we decided not to specifically include this limit in subpart J because many
permits specify the H2S concentration limit in terms of ppmv and the concentration limits in
these existing permits may range from 160 to 164 ppmv. We did not want to require re-
permitting of these facilities, reprogramming of monitoring equipment and/or potential
retroactive non-compliance issues that could occur with this "new" emissions limit if we had
specified the H2S concentration limit as 162 ppmv in subpart J.
Comment: Commenter 0311 stated that 40 CFR part 63, subpart LLLLL indicates at
40 CFR 63.8681(e) that subpart J does not apply for asphalt blowing stills subject to subpart
LLLLL. The commenter requested similar clarification for subpart Ja by exempting this process
in 40 CFR 60.100a.
Response: We reviewed the requirement in 40 CFR part 63, subpart LLLLL. Due to the
oxygen (02) content of this process gas, we agree that it is not suitable for recovery as fuel gas
and subsequent amine treatment; therefore, it is not BSER for combustion controls used on
asphalt blowing stills to meet the H2S concentration limits (or alternative S02 emissions limits).
We reviewed 40 CFR 60.100a, but we feel a blanket exemption from 40 CFR part 60, subpart Ja
is not necessary. Instead, we have included an exemption within the definition of fuel gas similar
to the exemptions included for combustion controls on vapors collected and combusted from
wastewater treatment and marine vessel loading operations. Specifically, we amended the
definition of fuel gas in 40 CFR 60.101a to clarify that fuel gas does not include vapors that are
collected and combusted to control emissions from asphalt processing units {i.e., asphalt blowing
stills).
35
-------
2.6 Equivalence with State and Local Rules
Comment: Several commenters addressed the EPA's request for comment on "the
equivalency of the subpart Ja requirements as proposed to be amended today and the [South
Coast Air Quality Management District (SCAQMD) Rule 1118]" and "whether the EPA could
deem a facility in compliance with subpart Ja as proposed to be amended today if that facility
was found to be in compliance with SCAQMD Rule 1118, or other equivalent State or local
rules" (73 FR 78532, December 22, 2008). Commenter 0307 disagreed with the EPA's position,
alleging that the "EPA's suggestion that it can waive compliance with the NSPS in this manner is
contrary to the Clean Air Act." The commenter stated that the EPA's suggestion "that existing
state and local requirements render the federal requirements irrelevant only confirms that EPA's
proposed flaring requirements do not reflect the best technological system of continuous
emission reduction." 42 U.S.C. §7411(h)(1) (emphasis added). The commenter also stated that
the CAA already provides a mechanism for implementation of alternative work practice
standards in narrowly defined circumstances (42 U.S.C. §7411(h)(3)); an owner or operator may
demonstrate to the Administrator that an alternative means of emissions limitation is equivalent
to the federal standard and request permission to comply with the alternative method in lieu of
the federal standard. Therefore, the commenter asserted, the CAA clearly states that the "EPA's
authority to waive federal work practice standards is case specific." Finally, the commenter
stated that the EPA did not explain how emissions reductions achieved through compliance with
SCAQMD Rule 1118 are equivalent to subpart Ja. Further, the commenter asserted that the EPA
neither identified other state rules that could be considered equivalent, nor explained how the
EPA would determine that a specific state or local rule is equivalent to subpart Ja. Therefore, the
commenter stated, it is impossible to fully assess the merit of the EPA's idea and provide
meaningful comments.
Commenter 0315 responded to Commenter 0307's comments. Commenter 0315 noted
that it appears that Commenter 0307 argues that if state regulations are more stringent than the
proposed requirements for subpart Ja, then the subpart Ja proposed requirements cannot be
"best" demonstrated technology. The commenter stated that the existence of more stringent
control measures is not per se fatal to an EPA BSER determination because "most stringent" is
not the criteria that must be applied under the law to determine BSER. In response to Commenter
0307's statement that the EPA must grant permission to use alternative methodologies on a case-
36
-------
by-case basis, Commenter 0315 stated that the EPA has full authority to establish alternative
regulatory standards that are determined to be as stringent as or more stringent than BSER. The
commenter noted that CAA section 111(h)(3) generally applies after the EPA has completed a
national rulemaking and an owner or operator requests approval for a site-specific alternative at a
later date. In this instance, it is logical that, if an alternative method is identified during the
rulemaking process, . .the law would allow EPA to establish a site-specific alternative [in the
rule itself] (especially, as under [CAA section 111], where the alternative would have to be
determined through notice and comment rulemaking - which, in essence, requires EPA to amend
the regulation to accomplish the site-specific alternative)." Finally, Commenter 0315 noted that
Commenter 0307 did not consider the documents on record for the original proposal of subpart
Ja, which demonstrate that the subpart Ja flare requirements are based on and almost entirely
consistent with SCAQMD Rule 1118. The commenter noted that the EPA determined that
SCAQMD Rule 1118 constituted BSER for flares and designed subpart Ja so that affected flares
would be subject to similar requirements.
Commenters 0304, 0306 and 0311 recommended that refineries complying with
California flare rules, specifically SCAQMD Rule 1118 (Commenters 0304 and 0311) and Bay
Area Air Quality Management District (BAAQMD) Regulation 12-12 and 12-11
(Commenter 0311) be deemed in compliance with subparts J and Ja. According to Commenter
0304, SCAQMD Rule 1118 is "in all respects equivalent to or more stringent than the
corresponding requirements" of subparts J and Ja. Although Commenter 0311 did note that the
FMP in SCAQMD Rule 1118 is only required for facilities that fail to achieve sulfur oxides
(SOx) emission reduction targets, the commenter asserted the subpart Ja FMP provided no
greater emission reduction incentive or mechanism than the SCAQMD Rule 1118. Commenter
0311 recommended that the EPA add a paragraph (e) to 40 CFR 60.103a that states: "Local rules
that are equivalent to these work practice standards (Section 60.103a) satisfy the requirements of
this section, including South Coast AQMD Rule 1118 and Bay Area AQMD rules 12-11 and 12-
12." The commenter also requested a similar paragraph be added to 40 CFR 60.107a(e) to allow
the monitoring methods used in SCAQMD Rule 1118 and BAAQMD Regulation 12-12 and 12-
11 to be acceptable under subpart Ja.
Commenter 0315 provided a table comparing each of the six proposed FMP requirements
in 40 CFR 60.103a(a) to the SCAQMD and BAAQMD regulations. The commenter's table
37
-------
identified sections of BAAQMD Regulation 12-12 and 12-11 that are equivalent to the six
subpart Ja FMP requirements, but the commenter also noted that SCAQMD Rule 1118 is only
equivalent to five of the proposed requirements; it does not require an owner or operator to
identify procedures to reduce flaring in cases of fuel gas imbalance (although Commenter 0311
noted that SCAQMD Rule 1118 requires minimization of all flaring, including fuel gas
imbalance). Based on this information, the commenter stated that facilities already complying
with SCAQMD Rule 1118 or BAAQMD Regulation 12-12 and 12-11 are already meeting
requirements equivalent to or more stringent than BSER, so these facilities should not have to
meet different requirements under subpart Ja.
Commenter 0315 also noted that BAAQMD Regulation 12-11 allows periodic sampling,
and the commenter stated that for flares used mostly for startup, shutdown and upset/malfunction
(SSM) events, a sampling system accurately calculates the total sulfur emissions, is relatively
cost effective and is more reliable than the continuous composition monitors in use for this
purpose. Therefore, the commenter requested that the EPA consider BAAQMD Regulation 12-
11 to be equivalent to total sulfur monitoring for SSM flares.
Response: First, we note that there seems to be some misunderstanding regarding how a
determination that SCAQMD Rule 1118 or BAAQMD Regulation 12, Rule 11 and
Regulation 12, Rule 12 are equivalent to 40 CFR part 60, subpart Ja would actually be
implemented in subpart Ja. The EPA will not "waive" the obligation to comply with subpart Ja if
the source is complying with SCAQMD Rule 1118 or BAAQMD Regulation 12, Rule 11 and
Regulation 12, Rule 12. In other words, the EPA will not allow the owner or operator to
"choose" to comply with SCAQMD Rule 1118 or BAAQMD Regulation 12, Rule 11 and
Regulation 12, Rule 12 instead of subpart Ja. Rather, the source must always demonstrate
compliance with subpart Ja. If SCAQMD Rule 1118 or BAAQMD Regulation 12, Rule 11 and
Regulation 12, Rule 12 are determined to be equivalent to subpart Ja, then these requirements
would be provided as an alternative within subpart Ja for the source to demonstrate that it is
meeting the requirements of subpart Ja.
To assess the comments, we reviewed SCAQMD Rule 1118, BAAQMD Regulation 12,
Rule 11, and BAAQMD Regulation 12, Rule 12 and compared these rules to the 40 CFR part 60,
subpart Ja requirements we are finalizing. We have included documentation of this review in
Docket ID No. EPA-HQ-OAR-2007-0011 that shows the sections of each of those rules that we
38
-------
consider are equivalent to the subpart Ja requirements. We determined that SCAQMD Rule 1118
and BAAQMD Regulation 12, Rule 11 and Regulation 12, Rule 12 will result in equivalent to or
greater than the emissions reductions resulting from the subpart Ja FMP requirements. As a
result of our analysis, we have amended subpart Ja, as described in the following paragraphs.
We determined that SCAQMD Rule 1118 is equivalent to the flare requirements and
monitoring, recordkeeping and reporting provisions for determining compliance with the flare
requirements in 40 CFR part 60, subpart Ja. We also determined that the combined provisions of
BAAQMD Regulation 12, Rule 11 and BAAQMD Regulation 12, Rule 12 are equivalent to the
flare requirements and monitoring, recordkeeping and reporting provisions for determining
compliance with the flare requirements in subpart Ja. Therefore, we have added specific
compliance options for flares that are located in the SCAQMD and are in compliance with
SCAQMD Rule 1118, as well as for flares that are located in the BAAQMD and are in
compliance with both BAAQMD Regulation 12, Rule 11 and BAAQMD Regulation 12,
Rule 12. Flares that are in compliance with these alternative compliance options are in
compliance with the flare standards in subpart Ja. Specifically, 40 CFR 60.103a(g) specifies that
flares that are located in the SCAQMD may elect to comply with SCAQMD Rule 1118 and
flares that are located in the BAAQMD may elect to comply with both BAAQMD
Regulation 12, Rule 11 and BAAQMD Regulation 12, Rule 12 to comply with the FMP
requirements of 40 CFR 60.103a(a) and (b) and the RCA and corrective action analysis
requirements of 40 CFR 60.103a(c) though (e). In addition, 40 CFR 60.107a(h) indicates that
flares that are located in the SCAQMD may elect to comply with the monitoring requirements of
SCAQMD Rule 1118 and flares that are located in the BAAQMD may elect to comply with the
combined monitoring requirements of both BAAQMD Regulation 12, Rule 11 and BAAQMD
Regulation 12, Rule 12 to comply with the monitoring requirements of 40 CFR 60.107a(e)
and (f). The owner or operator must notify the Administrator, as specified in 40 CFR 60.103a(g),
that the flare is in compliance with SCAQMD Rule 1118 or both BAAQMD Regulation 12,
Rule 11 and BAAQMD Regulation 12, Rule 12. The owner or operator must also submit a copy
of the existing FMP (if applicable), as specified in 40 CFR 60.103a(g).
We note that, as pointed out by commenters, an owner or operator maintains the ability
under CAA section 111(h)(3) to submit a request to establish, on a case-by-case basis, that "an
alternative means of emission limitation will achieve a reduction in emissions... at least
39
-------
equivalent to the reduction in emissions" achieved under the flare standards of 40 CFR part 60,
subpart Ja. Pursuant to CAA section 111(h)(3), we also included specific provisions within
40 CFR 60.103a for owners or operators to submit a request for "an alternative means of
emission limitation" that will achieve a reduction in emissions at least equivalent to the reduction
in emissions achieved under the final standards in subpart Ja.
40
-------
3.0 DEFINITIONS
3.1 Air Preheat
Comment: Commenter 0311 stated that the definition of "air preheat" is not used in the
rule and should be deleted.
Response: The term "air preheat" is used within the definition of "forced draft process
heater" so we have retained the definition in the final amendments.
3.2 Co-Fired Process Heater
Comment: Commenters 0308 and 0311 stated that the definition of "co-fired process
heater" should be revised to remove the reference to "burners that are designed" to be supplied
by both gaseous and liquid fuels because co-fired process heaters almost always use different
burner tips. The commenters recommended: "Co-fired process heater means a process heater
designed to be supplied by both gaseous and liquid fuels."
Response: It appears that there is some confusion regarding what is meant by "burners"
within this definition. We intended that a "burner" refers to each location in the process heater
designed to produce a flame to impart heat to the system. A single burner that is designed to be
supplied with both gaseous and liquid fuels is expected to have separate burner tips within the
burner, but both burner tips are part of the single burner location. If a process heater has some
burners that can only be fired with gaseous fuels and other burners that can only be fired by
liquid fuels, these process heaters are not considered co-fired process heaters. The gas burners in
such a process heater can employ ultra-low NOx burners capable of meeting the emissions limit
in the same manner as conventional gas-fired process heaters, and these process heaters should
demonstrate compliance with the rule when firing only with the gas burners. Burners that are
designed with multiple tips (so that the single burner can be supplied by both gaseous and liquid
fuels) have different design considerations, because staged fuel is generally needed for reducing
NOx in gas-fired burners, where staged air is used to reduce NOx in oil-fired burners. These
burners {i.e., those with both types of burner tips in the same burner) cannot be designed to stage
both air and fuel. As such, these burners can be either designed for staged air to limit NOx while
41
-------
firing oil but will have conventional gas-fired burners, or the burners can be designed for staged
fuel to limit NOx when firing gaseous fuels but will have conventional burners for fuel oil. If a
process heater has completely separate burners, some dedicated for gaseous fuels and some
dedicated for liquid fuels, each individual burner could be designed and optimized for its fuel
type. As such, we determined that it is inappropriate to revise the definition of co-fired process
heater as requested by the commenter. Furthermore, we find that gas-fired process heaters that
have emergency oil back-up burners in the event of a natural gas curtailment should employ gas
burners that are optimized for firing gas using staged fuel ultra-low NOx burners because they do
not routinely fire oil. Therefore, we revised the definition of co-fired process heaters to clarify
that co-fired process heaters are process heaters that are designed to be supplied by both gaseous
and liquid fuels on a routine basis. We further clarify in the definition that process heaters that
have gas burners with emergency oil back-up burners are not considered co-fired process heaters.
We clarify in this response that a process heater that contains gas burners with emergency oil
back-up burners must comply with the NOx limits for conventional gas-fired process heaters.
3.3 Corrective Action Analysis
Comment: Commenter 0305 stated that the definition of "corrective action analysis"
should not include the word "all" because it places unnecessary burden on the facility to
document "all" potential corrective actions (and subjects the facility to liability if it does not
include all possible actions) rather than focusing on the most promising.
Response: We disagree that the proposed definition of "corrective action analysis"
requires the facility to describe "all" potential corrective actions or otherwise puts undue burden
on the refinery owner or operator. The actual definition requires "a description of all reasonable
interim and long-term measures." The inclusion of the word "reasonable" modifies and limits the
term "all." The phrase "all reasonable" clarifies the requirements to perform a comprehensive
analysis while limiting the need to document unreasonable measures. Deleting the word "all" is
unlikely to alter the agency's interpretation of the requirement, and only acts to make the
requirement less clear. Therefore, we respectfully disagree with the commenter and retain this
portion of the definition of "corrective action analysis" as proposed.
42
-------
3.4 Delayed Coking Unit
Comment: Commenters 0305 and 0311 supported the revised definition of "delayed
coking unit" but stated that, since subpart Ja only sets standards for the coke drums, the
definition should just include the coke drums associated with a single fractionator. The
commenters stated that the definition should not include the fractionator itself because VOC
emissions from the fractionator are covered by NSPS for equipment leaks.
Response: The proposed amendments to the definition of "delayed coking unit"
specifically listed the primary components of the delayed coking unit. In particular, based on the
operation of the delayed coking unit, we find that the fractionator is an integral part of the
delayed coking unit. The fresh feed to the delayed coking unit is generally introduced in the
fractionator tower bottoms receiver. This integral use of the fractionator is different than the use
of fractionators used for other units defined in 40 CFR part 60, subpart Ja, such as the FCCU.
For the FCCU, fresh feed is introduced in the riser, which is part of the affected facility in
subpart Ja. As the feed to the delayed coking unit is to the fractionator, we find that the
fractionator is an integral part of the delayed coking unit, so we specifically include it as part of
the affected facility. While our proposed amendments covered only the major components of the
delayed coking unit, upon our review of the definition based on the comments received, we note
that there are several other components of the delayed coking unit that are integral to the
operation of the delayed coking unit. Additionally, even though the standards are specific to the
coke drum, many of these integral components are interconnected and necessary for the delayed
coking unit to meet the applicable standards. Based on our review of the operation of a delayed
coking unit, we also include coke cutting and blowdown recovery equipment in the final
definition because this equipment is also integral to the overall cyclical operation of the process
unit. The definition of "delayed coking unit" has been amended in the final rule to read as
follows:
Delayed coking unit means a refinery process unit in which high molecular weight
petroleum derivatives are thermally cracked and petroleum coke is produced in a
series of closed, batch system reactors. A delayed coking unit includes but is not
limited to all of the coke drums associated with a single fractionator; the
fractionator, including bottoms receiver and overhead condenser; the coke drum
cutting water and quench system, including the jet pump and coker quench water
tank; process piping and associated equipment such as pumps, valves and
connectors; and the coke drum blowdown recovery compressor system.
43
-------
Since this definition is more specific than the definition included in the amendments
proposed on December 22, 2008, it could affect which delayed coking units are subject to
subpart Ja. For example, an owner or operator may have made a change to a delayed coking unit
that would not be considered a modification under the December 22, 2008, definition, but that
same change could make the delayed coking unit a modified facility subject to subpart Ja using
the definition of "delayed coking unit" above. In other words, in changing the definition of
"delayed coking unit" in the final rule, some delayed coking units that would not have been
affected sources under the proposed requirements might now be covered by the final rule. Under
CAA section 111(a)(2), a "new source" is defined from the date of proposal only if there is a
standard "which will be applicable to such source;" otherwise, a "new source" is defined based
upon the final rule date. In this circumstance, using the proposal date as the new source date for
determining applicability for this group of delayed coking units would be inappropriate, as such
units would not have been on notice that subpart Ja could apply to them. Accordingly, we moved
the "new source" date for this group of delayed coking units so that delayed coking units that are
only defined as such under the final rule are covered by the final rule only if they commence
construction, reconstruction or modification after the promulgation date of these final
amendments. The "new source" date for other delayed coking units will depend on the previous
definitions and when the activities involving the delayed coking unit occurred. See 40 CFR
60.100a(b) for determining applicability of subpart Ja for delayed coking units.
3.5 Flare
Comment: Commenter 0308 and 0311 recommended deleting "knock-out pots, piping
and header system" from the definition of "flare" because piping and header systems are not
"equipment used to combust fuel gas" and should not be part of a fuel gas combustion device.
Commenter 0308 stated that, if the EPA retains this expanded definition, then the EPA should
clarify that the other fuel gas combustion devices are limited to just the "equipment used to
combust fuel gas" and do not include the fuel gas lines. Commenter 0311 stated that if "knock-
out pots, piping and header systems" are included, the flare's knock-out pot, piping and header
system must be clarified since numerous process units have knock-out pots and nearly all will be
connected either directly or indirectly (via the fuel gas system) to a flare. As such, the entire fuel
gas system could be considered part of the flare header system. The commenter also requested
44
-------
that the definition include a primary use clause so as to clarify the affected facility in cases of
interconnected flare systems. Commenter 0311 recommended: "Flare means an open-flame fuel
gas combustion device used for burning off unwanted gas or flammable gas and liquids. The
flare includes the foundation, flare tip, structural support, burner, igniter, flare controls including
air injection or steam injection systems, flame arrestors, and knockout pots, piping and header
systems from process units, fuel gas systems or FGR systems to a flare tip. Knockout pots,
piping and headers shared among multiple flares are assigned on a primary use basis."
Commenter 0315 also stated that the flare should not include the piping within process
units and fuel gas systems. The commenter recommended that the EPA clearly differentiate the
boundary between the refinery process unit, the fuel gas system and the flare. The commenter
noted that main lines from individual refinery process units and fuel gas systems to flares are
generally equipped with battery limit isolation valves, and these valves define the "battery
limits" of the process unit or fuel gas system. The commenter recommended that the EPA revise
the flare definition to state that: (1) a "flare" includes only the knockout pots, piping and headers
outside of a process unit's or fuel gas system's battery limits; and (2) any new piping or
connections that are made inside the process unit's or fuel gas system's battery limits are not
modifications to the flare. The commenter noted that new piping that is added outside of the
process unit's battery limits or the fuel gas system that either results in a new connection from
the process unit or fuel gas system to the flare or increases the capacity of an existing flare
system to handle more flow (e.g. increasing the size of a flare header or a knock out pot) would
be a modification to the flare. The commenter provided a diagram of an example refinery to
illustrate their points.
Response: We agree that a clarification is needed to distinguish between the flare header
system and the fuel gas system. We have added a definition of "flare gas header system" to
clarify this issue; see the response to the next comment for further details. We have also listed
flares as a completely separate affected facility, so we do not believe a clarification to the
definition of fuel gas combustion device is needed to specifically state that the fuel gas system is
not part of the fuel gas combustion device. With respect to interconnected flare systems, we
disagree that a primary use clause is needed because we determined that each flare in an
interconnected flare system is modified when a connection meeting those defined for a flare
modification is made to such a system. Please see Section 2.3 of this document for more details.
45
-------
Comment: Commenter 0313 requested that the definition of "flare" include the FGR
system. The commenter noted that in the BAAQMD, the flare header collects both the episodic
gas that needs to be sent to the flare device for safe disposal as well as the routine gas that is sent
back to the refinery fuel gas system. The commenter noted that refiners in the BAAQMD are
submitting permit applications asserting that all flare header systems upstream of the seal pot are
separate from the flare because during normal operation, these flare headers serve to collect the
gas that is redirected to the fuel gas system by the FGR system. The commenter noted that this
interpretation seems inconsistent with the EPA's stated intention in the preamble to the proposed
amendments to include the flare gas header system in the definition of "flare." The commenter
stated that the EPA should explicitly include the FGR system in the definition to eliminate any
confusion and clarify the agency's intent.
Response: We have added a definition to clarify what is included in the flare gas header
system. We specify that all piping used to transport gas to a flare either directly from a process
unit or directly from a pressure relief valve from the fuel gas system is part of the flare gas
header system, regardless of whether or not the flare gas header system includes a FGR system
{i.e., regardless of whether or not a FGR system draws gas from the flare gas header system).
Consistent with this definition, we note that if a FGR system is present, it is considered to be part
of the flare gas header system.
3.6 Fuel Gas
Comment: Commenters 0304, 0305 and 0311 supported the amended definition of "fuel
gas" as it pertains to coke calciners. Commenter 0304 suggested the same amendment be made
to the definition of fuel gas in subpart J. Commenters 0305 and 0311 suggested that the
definition be clarified to list all types of premium coke (specifically recarburizer coke, needle
coke and metallurgical coke) in addition to anode grade coke.
Response: We agree that the final disposition of the premium grade coke is not important
and have generalized the parenthetical to simply read "used to make premium grade coke." As
any new, modified or reconstructed fuel gas combustion device will be subject to subpart Ja
rather than subpart J going forward, there is no need to revise the definition of fuel gas in subpart
J at this point in time.
46
-------
Comment: Commenter 0309 suggested that the exclusion in the definition of "fuel gas"
be extended to vapors combusted to comply with the storage vessel requirements in 40 CFR 60,
subparts K, Ka, Kb (NSPS for storage vessels) and 40 CFR part 63, subpart CC (Refinery
MACT 1) because these are low pressure vapor streams and most have low sulfur content. The
commenter stated that at least certain storage vessels (reformate, alkylate, gasoline) should be
exempt due to low sulfur content.
Response: We disagree that the definition of fuel gas should exclude vapors from storage
vessels. While some storage vessels may be remotely located, many of these storage vessels are
located near refinery process units or each other so that combined collection and, if necessary,
treatment of the gases can be accomplished cost effectively. The current provisions in subpart Ja
(and subpart J) contain an exemption from the continuous H2S monitoring requirements for gases
that are inherently low sulfur or can be determined to be low sulfur. As such, the current
provisions for low sulfur fuel gas adequately address the commenter's issue without needing to
exempt certain types of storage vessels from the definition of fuel gas.
Comment: Commenter 0311 recommended that the exclusion from the definition of "fuel
gas" be extended to vapors "from marine vessel loading operations or waste management units
that are collected and combusted" without any reference to a federal requirement. At a minimum,
Commenter 0311 stated that marine benzene loading under 40 CFR part 61, subpart BB; the
wastewater provisions of 40 CFR part 63, subpart G; remediation efforts regulated under
Resource Conservation and Recovery Act (RCRA) corrective action; and RCRA 7003 orders
should be added to the exclusion.
Response: We were originally concerned that removing the reference to a federal
standard may inadvertently exempt the use of these vapors when used in process heaters or
boilers. We determined that it was not BSER to require thermal oxidizers used to comply with
the cited federal standards to comply with the H2S concentration limits due to the typically
remote location of the combustion sources (control devices) relative to refinery process units (see
technical memorandum entitled Fuel Gas Treatment of Marine Vessel Loading and Wastewater
Treatment Unit Off-gas, in Docket ID No. EPA-HQ-OAR-2007-0011). However, if these gases
are currently routed to a fuel gas system or directly to a process heater or boiler, treatment of the
fuel gas to meet the SO2 emissions limits or the H2S concentration limits is expected to be
economically viable. Additionally, these gases are expected to be only a small portion of the fuel
47
-------
gas combusted in these units, and the refinery has an option to over-treat the primary fuel gas so
that gases from the wastewater treatment system or marine vessel loading operation can remain
untreated while the fuel gas combustion device itself can comply with the S02 emissions limits
or the H2S concentration limits, based on the mixture of fuels used in the device.
In reviewing the rules suggested by the commenter, as well as those we originally listed,
we noted that acceptable "control devices" or "combustion units" in these rules include process
heaters and boilers. We did not intend to exclude vapors that are collected and routed to a
process heater or boiler to be exempt from the definition of fuel gas. In other words, when
developing this exclusion, we specifically considered the combustion of these gases via a thermal
oxidizer or flare currently located at the marine vessel loading or wastewater treatment location.
These remote combustion devices were really the subject of the analysis, but we did not want to
exclude these combustion units themselves because other fuel gas is often fed to these units to
ensure adequate combustion of the vapors being controlled. It is clear from our rationale and the
description of the exemption included in the preamble of the proposed rule that the exemption
was intended "to exempt vapors that are collected and combusted in an air pollution control
device installed to comply with" specific wastewater or marine vessel loading emissions
standards. (72 FR 27180 and also at 27183; emphasis added.) Process heaters or boilers would
not be "installed" to comply with these provisions, and it was not our intent to exclude vapors
sent to these types of combustion units. However, the regulatory text is more ambiguous and
appears to exclude any vapors collected and combusted, regardless of where they are combusted.
As such, we are amending this exclusion to better represent our original intent.
Additionally, with the added clarity in the regulatory text, it seems appropriate to extend
this exclusion to control devices used at these locations regardless of why the emission controls
were installed. That is, while we originally considered air pollution control devices that were
mandated by the EPA, we see no reason to discriminate against air pollution control devices that
were installed voluntarily to reduce the emissions from these sources. Further, we intend to
clarify that gases off the sour water system, including the sour water stripper, would likely
contain higher amounts of reduced sulfur and would be economically viable to treat. Therefore,
we are also clarifying that the exemption does not extend to the sour water system. Therefore, the
amended definition of "fuel gas" in both 40 CFR part 60, subparts J and Ja states that fuel gas
"does not include vapors that are collected and combusted in a thermal oxidizer or flare installed
48
-------
to control emissions from wastewater treatment units other than those processing sour water,
marine tank vessel loading operations or asphalt processing units (i.e., asphalt blowing stills)."
With respect to remediation efforts conducted under RCRA corrective actions, we are
unwilling to grant such an exclusion from the definition of "fuel gas" in 40 CFR part 60,
subpart Ja. First, we anticipate that most vapors from remediation efforts would be low in sulfur
and, if so, the owner or operator could apply for the alternative monitoring methods provided in
the rule. Also, although some remediation efforts may occur in remote locations, many of the
remediation efforts are conducted in reasonable proximity to existing process units. Finally, the
range of activities included in RCRA remediation efforts is broad, and we have little information
regarding the number and types of RCRA remediation activities that are being conducted. The
commenter provided no description of such activities, nor did they provide a reasonable rationale
as to why the vapors from these activities should be exempted.
Comment: Commenter 0311 noted that the reference to 40 CFR 60.692 was missing in
the revised definition of "fuel gas" and that "through" was misspelled.
Response: We appreciate the comment and acknowledge that the proposed definition did
include an inadvertent error. We have removed the section that contained the error from the final
definition.
3.7 Fuel Gas Combustion Device
Comment: Commenter 0308 recommended that the EPA delete flares from the definition
of "fuel gas combustion devices" and create a separate affected facility covering only flares so
that flares can be regulated in a targeted, coherent and consistent fashion.
Response: We agree. Due to the numerous special definitions and standards that apply to
flares, we determined that the requirements for flares and for fuel gas combustion devices would
be clarified if fuel gas combustion devices and flares were defined as separate affected facilities.
Therefore, the final rule defines a flare as an affected facility rather than including flares as a
type of fuel gas combustion device.
Comment: Commenter 0305 suggested clarifying the definition of "fuel gas combustion
device" to clarify that the fuel gas combustion device is located at the petroleum refinery.
Response: We do not agree that the requested clarification is necessary. A fuel gas
combustion device is defined as "any equipment, such as process heaters and boilers, used to
49
-------
combust fuel gas." Section 60.100a(a) states that "the provisions of this subpart apply to" fuel
gas combustion devices (including process heaters) "in petroleum refineries" (emphasis added).
Therefore, the final rule already specifies that a fuel gas combustion device must be located at a
petroleum refinery to be an affected source subject to subpart Ja.
We do note that the definition of "fuel gas" is "any gas which is generated at a petroleum
refinery and which is combusted." The definition does not specify that fuel gas must be
combusted at a petroleum refinery, and it is not our intention for that definition to be interpreted
as such.
3.8 Natural Draft and Forced Draft Process Heaters
Comment: Commenter 0296 indicated that there are four classes of process heaters and it
is unclear whether balanced draft or induced draft process heaters are considered natural draft or
forced draft process heaters. The commenter suggested that the EPA should clarify the definition
of "forced draft process heater" to include these configurations. Commenter 0305 also requested
clarification that induced draft process heaters are considered forced draft process heaters.
Commenter 0311 noted that in general, induced draft fans do not affect process heater NOx
controls, but for process heaters with downwardly firing burners, the induced draft fan is used to
move the combustion gases through the entire process heater, not just the convection section, and
those process heaters experience NOx control issues similar to forced draft process heaters. As
such, Commenter 0311 recommended that "process heaters with downward firing burners that
are equipped with an induced draft fan" be included in the definition of "forced draft process
heater." Commenter 0311 also requested that the definition be clarified based on the design
rather than the operation. The commenter noted that operators of some forced draft process
heaters may shut down the fan to perform maintenance on the fan or preheat system while still
operating the process heater and requested clarification on whether the process heater is still
subject to the forced draft process heater NOx limits during those periods.
Response: There are essentially four types of process heaters: (1) natural draft process
heaters with no fans; (2) induced draft process heaters with a fan on the exhaust line; (3) supplied
draft process heaters with a fan on the inlet air (more commonly referred to as forced draft
process heaters); and (4) balanced draft process heaters with fans on both the inlet air and on the
exhaust air (typically used for air preheat). The definition of forced draft process heater clearly
50
-------
includes both "supplied draft" and "balanced draft" process heaters as these systems have fans in
the inlet air line prior to the air entering the process heater or air preheat. The definition of forced
draft process heater just as clearly excludes induced draft process heaters because these process
heaters do not have a fan in the inlet air line. The definition of natural draft process heaters
includes induced draft process heaters because induced draft process heaters do not have a fan on
the inlet air (i.e., they are not "forced draft" process heaters). Nonetheless, the definitions are
somewhat confusing given the inclusion of "under positive pressure" for forced draft systems
and "under ambient pressure" for natural draft systems. We have clarified these definitions to
more clearly indicate that, for the purposes of subpart Ja, balanced draft systems are considered
forced draft process heaters and induced draft systems are considered natural draft process
heaters.
The commenters provided no direct data on "downwardly firing burners" or specifics as
to why these units could not meet the 40 ppmv NOx emissions limit. Based on the operations of
these burners, we think that they can meet the same limits as natural draft units, but given the
lack of data for this argument, we have included downwardly firing induced draft burners in the
list of process heaters that that may apply for a site-specific NOx emissions limit. If a process
heater equipped with downwardly firing burners is modified or reconstructed, the owner or
operator of the process heater should strive to meet the 40 ppmv emissions limit but may apply
for a site-specific emissions limit if the process heater cannot meet that limit.
We intended the emissions limits and the definitions of the process heaters to be based on
the operation of the heater (except for co-fired process heaters, the definition of which is design
based). According to comments received, including design performance levels from low-NOx
burner vendors for process heaters that are designed to operate in both natural draft mode and
forced draft mode, the burners can achieve the 40 ppmv (or 0.040 pounds per million British
thermal unit [lb/MMBtu]) NOx emissions limit while operating under natural draft conditions
even though it cannot meet this limit while operating in the forced draft mode, specifically
because of the higher inlet air temperatures when using air preheat. Based on the available
information, we conclude that the natural draft emissions limit is appropriate and BSER when
the process heater is operating in natural draft mode even if the unit is "designed" to operate as a
forced draft process heater. Therefore, we made no changes to the definitions of "forced draft
process heater" and "natural draft process heater" as a result of this comment.
51
-------
3.9 Process Upset Gas
Comment: Commenter 0303, 0305, 0306, 0308 and 0311 supported the amended
definition of "process upset gas" to include startup and shutdown and stated that the FMP is
effective in reducing SSM emissions. Commenter 0315 provided specific examples of planned
startups and shutdowns in which flaring is necessary for the proper operation of the unit and the
process train. Commenter 0302 requested that the definition of "process upset gas" include "gas
generated during periods when the fuel gas desulfurization system is down for maintenance in
accordance with the manufacturer's recommendations and good operating practices" as it is
unclear if these maintenance periods can be considered "malfunctions." If this provision in not
provided, the EPA should revise its cost analysis to include back-up desulfurization systems,
which would be needed if these periods cannot be excluded.
On the other hand, Commenter 0313 requested that the EPA not finalize the addition of
planned startups and shutdowns to the definition of "process upset gas." The commenter stated
that planned startups and shutdowns are a major cause of flare emissions, and unlike emergency
situations, refineries have the ability and resources to mitigate flare emissions during a planned
event. If the EPA adds startup and shutdown to the definition, then a large amount of the
emission reductions associated with subpart Ja will not be realized. In a subsequent letter
(Comment 0316), the commenter clarified this comment, noting that while many refineries in the
BAAQMD have found alternatives to flaring during planned startups and shutdowns by
modifying the design of new flares, it may not be possible for all new flares to be designed to
avoid flaring during all startups and shutdowns. The commenter also noted that BAAQMD
Regulation 12-12 is based on a FMP approach. The commenter stated that if the EPA concludes
that it is not reasonable to promulgate the fuel gas combustion device sulfur standards for new
flares as originally proposed (i.e., if startup and shutdown gases are included in the definition of
process upset gas), then the FMP approach would be an alternative way to address flaring
emissions from startups and shutdowns that would provide significant benefits.
Response: The only impact of including startup and shutdown gases within the definition
of process upset gas is to exclude these gases from the H2S concentration requirement. As an
example, the planned shutdown of a reactor vessel usually involves purging the vessel with
nitrogen. After the first few purge sequences, the purge gas may have very little Btu content, so
that the gas cannot be recovered in the fuel gas system, but the gas may contain concentrations of
52
-------
H2S that exceed 162 ppmv, depending on the reactor vessel contents. Flaring this gas to destroy
H2S is the best environmental practice, and it is not feasible to recover this gas, nor is it cost
effective to treat this gas separately to reduce its H2S concentration to below 162 ppmv.
Therefore, we cannot conclude that requiring refiners to limit the short-term fuel gas
concentration to 162 ppmv H2S determined hourly on a 3-hour rolling average basis is BSER for
planned startup or shutdown for all refinery process units at a petroleum refinery. However, we
have placed requirements on the flares to address releases during startups and shutdowns. The
FMP effectively addresses flare flow from planned startup and shutdowns as it requires
minimization of the amount of gas discharged to a flare under 40 CFR 60.103a(a)(5). The rule
also requires RCA to be performed if more than 500 pounds (lb) of S02 are discharged from a
flare in a 24-hour period under 40 CFR 60.103a(c), and this requirement applies to process upset
gas (including planned startup and shutdowns that do not follow the procedures in the FMP).7
These requirements effectively limit discharges of planned startup and shutdown gases to the
flare, and these provisions in subpart Ja are expected to yield significant emission reductions
compared to the requirements in subpart J. Consequently, we conclude that the inclusion of
startup and shutdown gases in the definition of "process upset gas" is necessary and aligned with
our BSER assessment for flares, and that the BSER for startup and shutdown events is a standard
for flares that requires the owner or operator to have and follow a FMP and to perform root cause
and corrective action analyses when these procedures are not followed.
With respect to maintenance of fuel gas desulfurization units, we find that the fuel gas
desulfurization unit is a control device used to meet the fuel gas standards. Continuing to operate
fuel gas combustion devices unabated when the control system for fuel gas is down for any
reason, including maintenance, startup, shutdown, malfunction or upset, is inconsistent with
good air pollution control practices. The requested maintenance exemptions are not in subpart J,
which is the baseline case for the cost analysis. Additional treatment capacity needed for these
cases is expected to exist under subpart Ja; our cost analysis considered the additional costs of
complying with subpart Ja (i.e., the additional long-term 60 ppmv H2S concentration limit) that
were in addition to those costs already imposed due to subpart J requirements. In our cost
analysis, we did assume some units would not have enough existing excess capacity to meet the
7 While we expect that most process upset gas will be combusted in flares, it is important to note that the rule also
requires RCA to be performed if S02 emissions exceed the short-term 162 ppmv H2S concentration limit for a
fuel gas combustion device by more than 500 pounds.
53
-------
long-term 60 ppmv H2S concentration limit, and would have to install additional amine treatment
systems. We conclude that the cost analysis conducted for fuel gas combustion devices
adequately reflects the incremental costs of subpart Ja. Therefore, the final amendments to
subpart Ja do not include the requested maintenance exemption.
Comment: Commenter 0308 suggested that "startup" be used in the definition of "process
upset gas" (rather than "start-up") to conform with the term "startup" in 40 CFR 60.2.
Response: We agree that "startup" is consistent with the General Provisions of 40 CFR
part 60, and it appears more frequently than "start-up" throughout subpart Ja. For consistency,
we have revised the occurrences of "start-up" to read "startup."
3.10 Refinery Process Unit
Comment: Commenters 0305 and 0311 objected to the change to the definition of
"refinery process unit" to include coke gasification, loading and wastewater treatment, stating the
change makes the term more expansive. The commenters stated that the EPA did not evaluate the
impacts or explain the consequences of the revised definition. Commenter 0311 also stated that
product loading is generally considered part of the refinery process unit to which it is associated
and wastewater treatment is a utility. Commenter 0305 suggested that the definition specify SIC
2911 (as in Refinery MACT 1).
Response: The original definition of "refinery process unit" in 40 CFR part 60, subpart J
and the definition of "refinery process unit" promulgated in 40 CFR part 60, subpart Ja in
June 2008 read as follows: "Refinery process unit means any segment of the petroleum refinery
in which a specific processing operation is conducted." Thus, to be considered a refinery process
unit, only two criteria are needed: (1) the unit must be located at a petroleum refinery; and (2) the
unit must be used to conduct "a specific processing operation." The definition does not directly
limit the scope of "processing operations." That is, the definition of refinery process unit does
not limit process operations to distillation, re-distillation, cracking or reforming, and it is not
limited to only those processes used to produce gasoline, kerosene, fuel oils, etc. In the proposed
amendment to this definition, we listed "operations" that we construed as conducting a "specific
processing operation" when these operations are located at a petroleum refinery. Consequently,
we considered the proposed inclusion of examples of refinery process units to be a clarification
of the existing definition rather than an expansion of the original definition.
54
-------
We reviewed the impact of the proposed revision of this definition on 40 CFR part 60,
subpart Ja as well as its historic use in 40 CFR part 60, subpart J. The term "refinery process
unit" is used primarily in the definitions of certain affected facilities, "process gas" and "process
upset gas" in subparts J and Ja. The term is also used in the flare provisions in subpart Ja. With
respect to the definitional terms, there can be no issue with including the designation of "refinery
process unit" within the definitions for specific process units. "Process gas" is not used at all in
either rule, although it was revised between proposal and promulgation of subpart J. In response
to a comment that the definition of "process gas" "should have included the non-hydrocarbon
gases produced by various process units in a refinery," the EPA responded: "The definition has
been revised to include all gases produced by process units in a refinery except fuel gas and
process upset gas." (See page 127 of Background Information For New Source Performance
Standards, Volume 3, Promulgated Standards (BID Vol. 3), EPA 450/2-74-003 (Feb. 1974),
Docket Item No. EPA-HQ-OAR-2007-0011-0082). The definition had actually been revised to
include "any gas generated by a petroleum refinery process unit." The response in BID Vol. 3
suggests that the EPA considered "refinery process units" and "process units in a refinery" to
have the same meaning, and there is no mention of limiting what is considered to be a "refinery
process unit" or "process units in a refinery."
"Process upset gas" is used only to provide an exemption to the H2S concentration limit
for process upset gas sent to a flare. See 40 CFR 60.104(a)(1), 60.103a(h). Therefore, a narrow
definition of "refinery process unit" would only limit those gases sent to a flare that would
qualify as "process upset gas." For example, if a coke gasifier is not a refinery process unit, then
gases generated during the startup, shutdown or malfunction of a coke gasifier located at the
refinery would not be "process upset gas" and would be required to comply with the requirement
to limit short-term H2S concentration in fuel gas to 162 ppmv if sent to a flare. We find that the
historical application of the "process upset gas" exclusion has considered a broad definition of
what constitutes a "refinery process unit."
For 40 CFR part 60, subpart Ja, the definition of "refinery process unit" also impacts the
flare provisions. Based on the proposed revisions of "refinery process unit," it was clearly our
intent that a broad definition of "refinery process unit" should apply to the flare requirements.
Specifically, we intended that a flare modification occurs when a wide range of equipment at the
petroleum refinery is newly connected to the flare. It was also our intent that the FMP consider
55
-------
flare minimization methods for this broadly defined range of equipment referred to collectively
as "refinery process units."
Based on our review of the impacts of changes to the definition of "refinery process
unit," and considering all of the comments received, we maintain that the existing definition of
"refinery process unit" is broad and should be broadly interpreted. For consistency between
40 CFR part 60, subparts J and Ja, we have elected to maintain the existing definition and not
include an example list of refinery process units within the definition. However, to clarify that a
modification to a flare occurs when these types of equipment are connected to the flare, we
revised the language in the flaring provisions to refer to "refinery process units, including
ancillary equipment." This revision is made to clarify our original intent that coke gasification
units, storage tanks, product loading operations and wastewater treatment systems, as well as
pressure relief valves, pumps, sampling vents, continuous analyzer vents and other similar
equipment, are units from which a connection to a flare would trigger a flare modification and
generate gas streams that should be considered in the FMP. We have included in the final
amendments a definition of "ancillary equipment." Specifically:
Ancillary equipment means equipment used in conjunction with or that serve a
refinery process unit. Ancillary equipment includes, but is not limited to, storage
tanks, product loading operations, wastewater treatment systems, steam- or
electricity-producing units (including coke gasification units), pressure relief
valves, pumps, sampling vents and continuous analyzer vents.
Sulfur recovery plants are also units from which a connection to a flare would trigger a
flare modification and generate gas streams that should be considered in the FMP. We recognize
that on-site sulfur recovery plants are considered refinery process units, and we proposed
amendments to the definitions of "refinery process unit" and "sulfur recovery plant" to clarify
that we consider a sulfur recovery plant to be "a segment of the petroleum refinery in which a
specific processing operation is conducted." However, the strict definition of "refinery process
unit" would only apply to sulfur recovery plants physically located at the refinery. As 40 CFR
part 60, subpart Ja also applies to off-site sulfur recovery plants (see 40 CFR 60.100(a) and
40 CFR 60.100a(a)), we found it potentially contradictory to define a sulfur recovery plant
located outside the refinery as a "refinery process unit," so we are also not finalizing the
proposed amendment to include the term "all refinery process units" in the definition of "sulfur
56
-------
recovery plant." However, while connections to a refinery flare from an off-site sulfur recovery
plant are not expected to be common, off-site sulfur recovery plants are subject to subpart Ja. We
clarify in this response that we would consider such a connection to a flare to be from a "refinery
process unit, including ancillary equipment," such that connecting an off-site sulfur recovery
plant that is subject to subpart Ja to a flare at a refinery would cause that flare to be a modified
flare subject to subpart Ja.
Further, in reviewing the definition of "sulfur recovery plant," we noticed an inadvertent
error that also suggests that the sulfur recovery plant must be located at a petroleum refinery,
which is not consistent with the applicability provisions in 40 CFR 60.100(a) and 40 CFR
60.100a(a). Specifically, we inadvertently omitted the word "produced" in this first sentence, so
we are amending the definition of "sulfur recovery plant" to clarify that a sulfur recovery plant
recovers sulfur from sour gases "produced at the petroleum refinery." Thus, we are amending the
definition of "sulfur recovery plant" to correct inadvertent errors and to clarify that off-site sulfur
recovery plants are included in the definition of "sulfur recovery plant," as these plants are
expressly considered to be affected facilities in 40 CFR part 60, subpart Ja.
3.11 Sulfur Recovery Plant
Comment: Commenter 0305 requested a clarification in the definition of "sulfur recovery
plant" to exclude loading facilities (in addition to secondary storage vessels).
Response: We agree and have revised the final definition. We also corrected an
inadvertent omission of the word "produced" in the definition. The "sulfur recovery plant" need
not be located at a petroleum refinery, but, for the purposes of subpart Ja, is limited to units that
recover sulfur from sour gases "produced at a petroleum refinery."
57
-------
4.0 FUEL GAS COMBUSTION DEVICES AND FLARES
4.1 Emissions Limits
4.1.1 Long-Term H2S Concentration Limit for Fuel Gas Combusted in Flares
Comment: Commenters 0296, 0301, 0305, 0310 and 0311 asserted that the EPA did not
properly demonstrate that the long-term 60 ppmv H2S fuel gas concentration limit is BSER for
flares as fuel gas combustion devices because the EPA underestimated the cost of treating the
flare gas and overstated emissions reductions. According to Commenter 0296, flare gas treatment
would require a compressor system for each flare header and would be more likely to require
additional amine treatment systems and/or sulfur recovery units than the EPA estimated. The
commenters also suggested that the EPA overestimated the value of the recovered flare gas
because the Btu content of flare gas is highly variable and generally lower than that used by the
EPA. Commenter 0311 stated that many refinery streams currently sent to the flare contain H2S
concentrations between 60 and 162 ppmv, and requiring fuel gas to meet a long-term 60 ppmv
H2S concentration limit would result in significant re-piping and capital investment for minimal
emission reductions, which the EPA's impact analysis did not consider. Commenter 0311 also
asserted that the emission reductions were overstated {i.e., the EPA should not use 162 ppmv as
current long-term performance at baseline) and that the model analysis was oversimplified and
neglected to account for the range of costs that are incurred by individual refineries.
Commenter 0301 asserted that the long-term 60 ppmv fuel gas H2S concentration limit for flares
does not meet the BSER cost effectiveness test as required by the CAA, particularly when
evaluating the S02 reductions incremental to those achieved by consent decrees.
Commenter 0301 suggested that a long-term 60 ppmv fuel gas H2S concentration limit
for flares would cost $360-460 million and only reduce S02 emissions by 190 tons per year
(tons/yr), which the commenter estimated to be a cost effectiveness of $326,000 per ton of S02
reduced. Commenter 0310 provided costs estimates for: (1) additional amine systems of actual
projects showing cost effectiveness of $44,000 to $63,000 per ton of S02 reduced; (2) FGR
systems with cost effectiveness of $25,000 to $65,000 per ton of S02 reduced (details of the
58
-------
commenter's estimate are confidential); and (3) FGR system expansions, which included an
additional compressor, surge storage capacity, and amine scrubber capacity, showing cost
effectiveness of recovery system expansions to be $300,000 to $400,000 per ton of S02 reduced.
While most of their comments focused on flares, Commenter 0310 also stated that a long-term
60 ppmv fuel gas H2S concentration limit is improper for fuel gas combustion devices in general
because it could preclude some refineries from processing high-sulfur crude oils, thereby
limiting refining production capacity.
Commenter 0296 noted that the flare will receive both "fuel gas" and exempt gas streams
("process upset gas" as well as gas from pressure relief valve leakage), so it is impossible to
determine if any emissions limit exceedance is caused by the regulated "fuel gas" or by the
exempt gas. Commenter 0296 recommended that the EPA allow Alternative Monitoring Plans to
demonstrate compliance with the fuel gas emissions limits for non-exempt gas streams upstream
of the flare header. Alternatively, Commenter 0296 recommended that the long-term 60 ppmv
H2S fuel gas concentration limit be applicable only to the refinery fuel gas system and not to the
refinery flare system.
Response: We acknowledge that, at proposal, we determined that a long-term 60 ppmv
H2S fuel gas concentration limit was cost effective primarily for process heaters, boilers and
other fuel gas combustion devices that are fed by the refinery's fuel gas system. Based on the
typical configuration at a refinery, adding one new fuel gas combustion device to the fuel gas
system would essentially require the owner or operator to limit the long-term concentration of
H2S in the entire fuel gas system to 60 ppmv, so emission reductions would result from all fuel
gas combustion devices tied to that fuel gas system. Upon review of the BSER analysis
conducted at proposal for fuel gas combustion devices, we now realize that the analysis is not
applicable to flares (See Docket Item No. EPA-HQ-OAR-2007-0011-0289).
Moreover, since we are regulating flares separately from other fuel gas combustion
devices in this final rule, we should separately consider whether a long-term H2S concentration
limit is appropriate for fuel gas sent to flares.
In developing the suite of CAA section 111(h) standards for flares, we considered
whether refineries should be required to optimize management of their fuel gas by limiting the
long-term H2S concentration to 60 ppmv in addition to the short-term H2S concentration of
162 ppmv during normal operating conditions. We determined that, for refineries to demonstrate
59
-------
that their fuel gas complies with a long-term H2S concentration of 60 ppmv, refineries would
have to install a FGR system (which was not needed for other fuel gas combustion devices) and
then upgrade the fuel gas desulfurization system. Alternatively, refineries would have to treat the
recovered fuel gas to limit the long-term concentration of H2S to 60 ppmv with new amine
treatment units on each flare.
While some of the costs provided by the commenters did not include the value of the
recovered gas and appeared, at times, to include equipment not necessarily required by the
regulation, we generally agree with the commenters, based on our own cost estimates, that
optimizing management of the fuel gas system to limit the long-term concentration of H2S to
60 ppmv is not cost effective for flares (see Table 2 below). We note that the costs provided by
the commenters and the costs and emissions reductions in our analysis are the incremental costs
and emissions reductions of going from the short-term 162 ppmv H2S concentration to a
combined short-term 162 ppmv H2S concentration and long-term 60 ppmv H2S concentration.
While we are aware that some consent decrees require refineries to limit the concentration of
H2S in the fuel gas to levels lower than the short-term 162 ppmv H2S concentration, our baseline
when evaluating the impacts of a national standard (in this case, 40 CFR part 60, subpart Ja) is
the national set of requirements to which an affected flare would be subject in the absence of
subpart Ja {i.e., the short-term 162 ppmv H2S concentration limit in 40 CFR part 60, subpart J).
Table 2. National Fifth Year Impacts of Meeting a Long-Term 60 ppmv H2S Concentration
for Flares Subject to 40 CFR part 60, subpart Ja
Total
211111II ill
Kmission
Kniission
Kniission
C'iipiliil
cost
reduction
reduction
reduction
Cost
cost
(SI 000/
(tons
(tons
(tons
effect i\ciiess
(SI.000)
yr)'
S(),/yr)h
VVyr)1'
YOC/vr)1'
(S/ton)
New
80,000
15,000
6
34
130
84,000
Modified /
Reconstructed
860,000
160,000
53
310
1,200
100,000
a Because of the heat content of recovered gas, each scf of recovered gas is assumed to offset one scf of natural gas;
a value of $5/10,000 scf of natural gas was used to estimate recovery credit.
b These emission reductions are based on flares already meeting the short-term 162 ppmv H2S fuel gas concentration
limit in 40 CFR part 60, subpart J (i.e., these are the incremental emission reductions achieved from a baseline of
optimizing management of the fuel gas system to limit the short-term H2S concentration in the fuel gas to 162
ppmv to the originally proposed combined short-term 162 ppmv H2S concentration and long-term 60 ppmv H2S
concentration in the fuel gas).
60
-------
4.1.2 A Itern ative Complian ce Options
Comment: Commenter 0301 requested that the EPA specifically allow alternative means
of compliance included in consent decrees as acceptable compliance alternatives for subpart Ja.
Specifically, certain flares have an alternative 500 pounds per day (lb/day) SO2 emissions limit
in lieu of a requirement to limit the short-term concentration of H2S in fuel gas combusted in a
flare to 162 ppmv, and the commenter requested assurance that this alternative is sufficient to
demonstrate compliance with subpart Ja.
Response: The requirements in consent decrees are negotiated settlements and are not
based on an analysis of BSER, as several commenters have pointed out (although typically to
argue why certain consent decree requirements should not be the basis of subpart Ja
requirements). The BSER analysis conducted for this final rule package supports the
requirements in this national rulemaking. As with any national standard, there will be units
where the costs to reduce a ton of pollutant emissions (in this case SO2) will be less than the
average costs projected for the industry and there will be units where the costs are higher than
the average costs projected for the industry. While we try to consider and, if appropriate,
minimize the impact of the less cost effective units, we generally base our decision on the
nationwide costs and emissions reductions achieved. As such, we do not feel compelled to loosen
the national standard based on the poor cost effectiveness of a very limited number (less than
1 percent) of flares.
We have conducted specific analyses that consider flares as separate and distinct affected
facilities from fuel gas combustion devices. While we concluded that a long-term 60 ppmv fuel
gas H2S concentration limit is not cost effective for flares, we maintain that the requirement to
optimize management of the fuel gas by limiting the short-term concentration of H2S to 162
ppmv H2S during normal operating conditions is cost effective for flares. We also find that the
500 lb/day S02 limit is not, in any way, equivalent to the long-standing short-term 162 ppmv
H2S concentration requirement for these units in subpart J. At 500,000 standard cubic feet (scf)
in a 24-hour period, the threshold at which a RCA is triggered for flare flow, the fuel gas could
contain 6,000 ppmv H2S. With a 500 lb/day standard, flares with flow rates below the RCA
threshold of 500,000 scf in any 24-hour period could exceed the short-term 162 ppmv H2S
61
-------
concentration requirement by a factor of 40 without exceeding the 500 lb/day SO2 emissions
limit requested by the commenters. We do not see this to be, in any way, equivalent to the
existing limit contained in subpart J, and we are not including this significant exemption from the
short-term 162 ppmv H2S concentration requirement in subpart Ja.
4.2 Elimination of Routine Flaring
Comment: Commenter 0307 stated that the EPA's proposed amendments for flares are
illegal and arbitrary. The commenter noted that when the EPA originally proposed subpart Ja for
petroleum refineries (see 72 FR 27178, May 14, 2007), the EPA determined that elimination of
routine flaring through the use of available technologies such as FGR is technologically feasible,
cost effective and BSERunder CAA section 111. However, the final rule (see 73 FR 35838, June
24, 2008) did not include those requirements, and the commenter asserted that the EPA did not
adequately explain why the BSER determination for the final rule changed (see 73 FR 35838,
June 24, 2008). The commenter stated that the EPA did not provide rationale for dismissing
technology that the agency determined was demonstrated and cost effective in favor of standards
such as the FMP to manage flare emissions. The commenter also noted that in the June 2008
final rule, the EPA stated that the prohibition on routine flaring was removed because the EPA
"could not determine how to define 'routine' events." However, the same final rule included a
limit on flow to the flare applicable to "normal" operations (i.e., excluding SSM). The
commenter asserted that "all of these concepts in the final rule require distinguishing between
routine and exceptional events," so the EPA's explanation for the change is "internally
inconsistent."
Commenter 0307 asserted that there is no question that FGR is demonstrated technology;
FGR is commonly used in existing refineries to capture and recycle process gases. Existing
refineries have been able to significantly reduce or eliminate routine flaring through the use of
FGR, and other refineries are planning installations, so FGR "may fairly be projected for the
regulated future."8 Further, recovery technology can significantly reduce air pollution as well as
save owners and operators money through recovery of useful flare gas components such as
hydrocarbons that a refinery can burn as fuel and hydrogen, which is used in several processes
8 Lignite Energy Council v. USEPA, 198 F.3d 930,934 (D.C. Cir. 1999) (citing Portland Cement Ass'n v.
Ruckelshaus, 486 F.2d 375, 391 (D.C. Cir. 1973))
62
-------
and expensive to make.9 In addition, demand for steam generation is reduced when steam-
assisted flaring is reduced. The commenter stated that the EPA, the DOE, the Valero refinery in
Texas and the SCAQMD have found FGR systems to be cost effective. The commenter noted
that the revised emission estimates presented in the preamble to the proposed amendments do
more accurately represent emissions from flares than the estimates in the preamble to the final
standards, but given that increase in emission reductions, the EPA did not explain why both
eliminating routine flaring and reducing flaring via a limit on flare flow rate are suddenly not
cost effective. The commenter stated that the EPA should revise the rule to return to the
originally proposed standard prohibiting all routine flaring because the CAA section 111
standards must reflect the best technology capable of capturing and eliminating the pollutants of
concern. At the least, the EPA must provide a new analysis to clearly demonstrate why a
prohibition on routine flaring using FGR and other technologies is not feasible or cost effective
when considering the revised emissions reductions.
Commenter 0307 also noted that in the preamble to the final standards, the EPA stated
that FGR is not technically feasible if a refinery produces more fuel gas than it needs to operate,
but the commenter stated that the EPA provided no data or rationale to support that claim. The
commenter also stated that even if the refinery produces more fuel gas than needed, there are
alternatives to flaring that the EPA did not consider, such as the possibilities of building on-site
storage for excess gas or selling the gas to third parties; in fact, the EPA stated in the preamble to
the original proposed rule that the agency believed "other options exist, such as building an
electric co-generating unit."
Response: We note that our original proposal (see 72 FR 27178, May 14, 2007) did not
eliminate all routine flaring, only routine flaring from new "fuel gas producing units." The
affected facility in the original proposal was not the flare, but the fuel gas producing unit. For a
new fuel gas producing unit, the proposed requirements were to minimize releases to the flare
during startup and shutdown and to eliminate routine flaring of gases from the fuel gas producing
unit during normal operation. For units that produce significant quantities of fuel gas, refinery
owners and operators routinely install FGR systems to recover the gas for use as fuel gas in the
9 John Zink Company. "Environmental Control With Serious Payback." Printed February 4, 2009, and included as
Attachment A to Comment 0307 (Docket Item No. EPA-HQ-OAR-2007-0011-0317).
63
-------
refinery, so a gas recovery system (compressor) at the fuel gas producing unit was the expected
method of compliance for an affected fuel gas producing unit at the time of the original proposal.
The problems with the initial proposal were that there were ambiguities in the definition
of a fuel gas producing unit and in what was meant by "no routine flaring." As we evaluated the
proposed rule based on the comments received, we determined that it would be more
straightforward to make the flare the affected facility in the final rule (see 73 FR 35838, June 24,
2008). By regulating the flare, given the specific flare modification provision in subpart Ja, we
actually increased the projected emission reductions of the rule as compared to the reductions
achieved by regulating only new fuel gas producing units. In other words, the initial proposal
affected only a portion of the gas sent to a flare, while the final rule affects all of the gas sent to a
flare. As such, we find that the "flare as the affected facility" paradigm being finalized today is
environmentally superior to the "no routine flaring from new fuel gas producing units" paradigm
we originally proposed (see 72 FR 27178, May 14, 2007).
The evaluation of FGR systems included in the docket demonstrates that FGR below
certain flow rates is not cost effective; therefore, a complete ban on "routine" flaring is not
BSER (see the "Flare Options Calculator" included as an attachment to Docket Item No. EPA-
HQ-OAR-2007-0011-0223 and the flare impact memoranda included as Docket Item No. EPA-
HQ-OAR-2007-0011-0223 and Docket Item No. EPA-HQ-OAR-2007-0011-0289). We do agree
with the commenter that FGR systems are technically effective at reducing flare emissions. We
also maintain that it is cost effective to install FGR systems when the flow rate of flare gas
exceeds a certain threshold. However, the flow threshold at which FGR becomes cost effective is
dependent on a number of factors, many of which are variable (such as the Btu content of the
flare gas). The flare gas options calculator was developed to evaluate a range of flow rates, flare
gas Btu content and other factors to determine the likely range of flows above which FGR would
be cost effective. Based on the data available at the time, a recoverable flare gas flow rate of
between 250,000 and 500,000 standard cubic feet per day (scfd) was determined to be the
threshold at which FGR would be cost effective. In the proposed amendments (see 73 FR 78522,
December 22, 2008), we proposed to remove the 250,000 scfd flow limit, but we also proposed
to expand the applicability of the flare flow RCA requirement to routine flaring discharges in an
effort to address uncertainties on how to define "normal operations" and to account for high
64
-------
daily flow rates that might cause exceedances of the 30-day limit even though "routine" flow is
low.
The cost effectiveness analysis provided in the docket is not really applicable to fuel gas
rich facilities because it assumes that the recovered flare gas can be combusted on-site in existing
process heaters. When we consider the flare as the affected facility, it is necessary to provide
provisions for flaring during periods when the facility produces more fuel gas than it needs {i.e.,
is fuel gas rich). If a refinery has no place to use the recovered gas, the recovery compressors
will only re-circulate the gas back to the flare. To eliminate flaring during times when the facility
is fuel gas rich, the owner or operator would have to find somewhere else to send the fuel gas
besides the flare. The owner or operator may be able to store the fuel gas until it is needed, which
would require additional storage tanks, but this alternative has many drawbacks, such as strict
limitations on excess storage capacity and safety hazards associated with storing the explosive
gases. Alternatively, the owner or operator could purchase additional boilers, turbines or other
electric co-generation equipment to extract useful work when combusting the excess gas. In this
alternative, the excess fuel gas can be turned into steam or electricity that could either be used
on-site or sent off-site {e.g., electricity supplied to the electric power grid). The additional costs
associated with the steam- or electricity-generating equipment greatly increases the amount of
flare gas that would need to be recovered for a FGR/co-generation system to be cost effective,
and this additional gas would be needed on a routine basis. However, a refinery is not typically
fuel gas rich consistently all year long. Consequently, it is not cost effective to require FGR
during periods when the refinery is fuel gas rich because of the limited time periods when this
excess gas is available and the costs associated with purchasing steam or electricity generating
equipment needed to utilize the excess fuel gas. Furthermore, the few facilities that are
consistently fuel gas rich typically have made arrangements to sell their excess fuel gas, have
already installed co-generation units or have made other arrangements to utilize this economic
asset. We did not fully evaluate a maximum flow rate or other criteria limiting the applicability
of the fuel gas rich exemption, both because we did not have the data to do so and because we
did not consider this a likely situation. Additionally, there are a significant number of variables
to consider when determining the cost and cost effectiveness of a co-generation unit or other type
of potential control system, making the development of a single national standard highly
problematic. The agency collected additional data from the petroleum refinery industry through
65
-------
an information collection request during the course of 2011 (see Docket ID No. EPA-HQ-OAR-
2010-0682), and we are in the process of analyzing the collected data. If fuel gas rich situations
are found to be more common than we anticipate, we will consider additional emission reduction
opportunities for these situations at that time. However, based on the information currently
available, when we considered the flare as the affected source (rather than the fuel gas producing
unit) and the technology needs to eliminate "routine flaring" during periods when a refinery is
fuel gas rich, we determined that a "no routine flaring" requirement is not BSER because it is not
cost effective. Rather, we determined that the development and adherence to a FMP that
minimizes flaring during periods when the refinery is fuel gas rich is cost effective and therefore
BSER for occasional times when a refinery is fuel gas rich.
4.3 Flare Flow Rate Limit
Comment: Commenter 0303, 0306, 0308 and 0311 supported the removal of the flow rate
limit for flares because the FMP and other requirements will reduce flare flow, and flexibility is
needed in order to accommodate sweep gas requirements for different flares and
weather/seasonal conditions.
Response: We appreciate these comments; however, as seen by the comments received
on the flare flow RCA requirements, there remain issues regarding sweep gas requirements, flare
configurations and weather/seasonal conditions that still need to be addressed with respect to the
flow RCA. These issues are discussed in detail in Section 4.4.3 of this document.
4.4 Other Standards for Flares
4.4.1 General Comments on Other Flare Standards
Comment: Commenters 0305 and 0311 were critical of the impact analysis performed for
the flare standards, stating that: (1) the 2002 National Emissions Inventory (NEI) CO data are not
representative of process flow (Commenter 0305); (2) the flows or recovery rates were not
adjusted to consider upsets that would not be recovered and continuous sweep gas flows
(Commenter 0305); (3) the heating value of the waste gas was too high (Commenters 0305 and
0311); (4) the EPA did not consider net costs when no offset is possible (e.g., when the refinery
is long on fuel gas or flare gas is poor quality) (Commenters 0305 and 0311); (5) the EPA
underestimated capital costs for recovery systems (Commenter 0311); (6) the EPA
66
-------
underestimated capital costs for monitoring systems (Commenter 0311); (7) the EPA did not
consider that for certain types of flares, no sulfur emission reductions are possible (e.g., liquefied
petroleum gas (LPG) tank, hydrogen or emergency-only flares), but owners and operators still
incur the cost of installing sulfur monitors on those flares (Commenter 0305); and (8) limited
data were used to assess emission reductions from RCA (Commenters 0305 and 0311).
Response: We generally disagree with these criticisms of the impact analysis for the flare
standards. As stated in the preamble to the final amendments, we recognized that the flow
threshold at which FGR becomes cost effective is dependent on a number of factors, many of
which are variable (such as the Btu content and flow rate of the flare gas). The flare gas options
calculator was developed to evaluate a range of flow rates, flare gas Btu content and other factors
to determine the likely range of flows above which FGR would be cost effective. The costs of the
systems and monitors came from vendor-reported costs. These costs agree well with or tend to be
slightly higher than costs for FGR systems collected by the BAAQMD, many of which came
from refinery industry representatives to support their flare management requirements. We have
acknowledged that cost credits were used predicated on the ability to productively use the
recovered flare gas, but we have provided specific exemptions from the flow RCA for periods
when the refinery may be fuel gas rich. We did apply an 80-percent flare gas recovery efficiency
to account for upsets that could not be reduced. We have also amended the flare flow RCA
threshold in 40 CFR 60.103a(c)(l)(ii) to be 500,000 scf above a defined baseline in any 24-hour
period to account for sweep and purge gas flow.
The flare requirements were established to reduce SO2, VOC, CO and NOx emissions, so
the analysis was not limited to just sulfur-containing streams. The analysis assumed that the
average H2S concentration of the recovered flare gas was 60 ppmv, which is well below the
short-term H2S concentration of 162 ppmv that we are finalizing for fuel gas combusted in flares,
so this assumption adequately accounts for recovery of gases with very low sulfur content, such
as LPG. Based on the BAAQMD experience with requiring flare flow monitoring, as well as
recent Differential Absorption LIDAR (DIAL) observations, it appears that the flare flow rates
extrapolated from the NEI data may seriously underestimate actual flare usage. While this does
not impact the threshold flow rate at which FGR becomes cost effective, it does impact the
number of flares and the quantity of emissions reduced as a result of the rule. Based on our
review of the impact analyses, we find no merit in the criticism suggested by the commenters. If
67
-------
anything, we find that the flare standards may achieve more emission reductions than previously
projected. We uphold our analysis and the conclusion that the flare standards are cost effective
and BSER.
4.4.2 Flare Management Plan Requirements
Comment: Commenter 0309 suggested that the EPA's intention in promulgating a
different definition of "affected facility" in subpart Ja than the existing definition in subpart J
was to indicate that existing process units are not subject to the flare standards. The commenter
suggested that the EPA should further clarify this point so that it is clear that existing affected
facilities are exempt from the flare standards.
Response: The affected facility discussion is different primarily because the flare
provisions were first published at a different date than the other rule requirements, not to denote
that the FMP does not cover existing sources connected to a newly modified flare. The FMP
covers all refinery process units and ancillary emission sources (existing and new units) that are
connected to an affected flare.
Comment: Commenters 0308 and 0311 suggested that the requirement to minimize
discharges to the flare in proposed 40 CFR 60.103a(a)(l) should specifically address routine
discharges and include a clause to limit the minimization of flaring to those that: (1) are
"consistent with good engineering practices" and (2) consider "the cost of achieving such
emission reduction, and any non-air quality health and environmental impact and energy
requirements." These revisions are needed, according to Commenter 0308, because non-routine
discharges are covered under the RCA and so that the requirement is compliant with the need to
consider cost under section 111 of the CAA.
Response: We agree that the language in proposed 40 CFR 60.103a(a)(l) appears to
require an assessment of flare minimization irrespective of cost or other relevant considerations,
as contained in CAA section 111, which was not our intent. We are clarifying, through this
response, that cost, safety and emissions reductions may be considered when evaluating what
actions should be taken to minimize discharges to a flare, but we disagree that the flare
minimization assessment should be limited to "routine discharges." We have revised the FMP
requirements in 40 CFR 60.103a(a) to more fully describe the types of information that must be
evaluated and included in the plan.
68
-------
As noted in section III.C of the preamble to the final amendments (summary of the rule),
we are finalizing our proposed withdrawal of the 250,000 scfd 30-day rolling average flow limit
for flares. This limitation does not adequately account for site-specific factors regarding flare gas
Btu content, ability to offset natural gas purchase and other considerations. We find that these
factors need to be addressed in a site-specific basis and are more appropriately addressed through
the FMP. In the absence of the specific flow limitation, we have included additional
requirements in the FMP to prompt a thorough review of the flare system so that, as an example,
FGR systems are installed and used where these systems are warranted. We have also revised the
flare minimization requirements to require the FMP to be submitted to the Administrator
(40 CFR 60.103a(b)).
As part of the development of the FMP, refinery owners and operators can provide
rationale and supporting evidence regarding the flare reduction options considered, the costs of
each option, the quantity of flare gas that would be recovered or prevented by the option, the Btu
content of the flare gas and the ability or inability of the reduction option to offset natural gas
purchases. The plan will also include the rationale for the selected reduction option, including
consideration of safety concerns. The owner or operator must comply with the plan, as submitted
to the Administrator. Major revisions to the plan, such as the addition of an alternative baseline
(see next section for further detail on baselines), must also be submitted to the Administrator.
In summary, although we did not incorporate the commenter's suggested language for
limiting the scope of the minimization requirements to actions that are "consistent with good
engineering practices" and that "consider costs and other health and environmental impacts," we
acknowledge that these are valid considerations in the selection of the minimization alternatives
available for a given affected flare. We find that the process of developing and submitting the
FMP will ensure that these factors are considered consistent with CAA section 111 and that the
requirement to minimize discharges to the flare is implemented consistently across all affected
sources.
Comment: Commenter 0308 requested confirmation that the requirement to include
procedures to reduce flaring during periods of fuel imbalance in 40 CFR 60.103a(a)(6) does not
require curtailed production of key fuel gas producing units, which would not be cost effective,
even though it may be the only means of reducing flaring.
69
-------
Response: We agree that the procedures to reduce flaring in cases of fuel gas imbalance
are not intended to require refineries to curtail production of key fuel gas producing units such as
the atmospheric distillation column, the FCCU, coking unit or similar processes. We expect that
refinery owners and operators will consider plans such as procedures for reducing overhead
condenser temperatures (by increased cooling water recirculation or using refrigerated
condensers) to minimize the amount of C3 or C4 hydrocarbons that are introduced to the fuel gas
system, procedures for temporary fuel gas storage if appropriate storage vessels are available on-
site or agreements to sell excess fuel gas to neighboring industrial plants. Owners or operators of
affected flares should document the viable flaring reduction alternatives in their FMP. It is also
recommended that the owner or operator keep records of the alternatives evaluated, including
why certain options were not considered viable options for inclusion in the FMP.
4.4.3 Root Cause Analysis and Corrective Action Analysis
Comment: Commenter 0296 supported the RCA requirements but noted that 45 days,
while reasonable, is a short time period for many complex events and this time period should not
be shortened. Commenters 0301, 0305 and 0311 requested clarification that the RCA was to be
conducted within 45 days but that not all of the corrective actions would have to be completed
within 45 days. Commenter 0301 recommended replacing the word "implement" with the word
"identify" in proposed 40 CFR 60.103a(a)(5), suggesting that this appears to be the EPA's intent
based on the implementation schedule described in proposed 40 CFR 60.103a(c). Commenter
0311 also requested that a provision be added for up to a 6-month extension, with approval, to
perform RCA evaluations for complicated events or disasters.
Response: We consider the 45-day time period to complete the RCA to be appropriate,
and we do not believe that an effective RCA can be conducted 6 months after the event. We do
agree with commenters that some corrective actions may require more time to fully implement. It
is our intent, however, that the corrective actions that can be implemented within the 45-day
period be implemented within that timeframe. Therefore, simply replacing the word "implement"
with the word "identify" in 40 CFR 60.103a(e) (formerly part of 40 CFR 60.103a(a)(5)) does not
comport with our intent. Instead, we have added 40 CFR 60.103a(e)(2) to cover corrective
actions that cannot be implemented within 45 days as follows: "For corrective actions that cannot
be fully implemented within 45 days following the discharge for which the root cause and
70
-------
corrective action analyses were required, the owner or operator shall develop an implementation
schedule to complete the corrective action(s) as soon as practicable."
Comment: Commenter 0311 stated that the S02 and flow RCA should not be triggered on
a per flare basis, but rather a per upset event basis. The commenter contended that flares
complying with the short-term 162 ppmv H2S concentration requirement for fuel gas could
exceed or approach 500 lb SO2 during 24 hours of normal use, and that the per-flare limit
penalizes facilities with consolidated flare systems (larger individual flares). The commenter also
noted that the proposed language for flow RCA could be interpreted to require a RCA for every
minute when a 24-hour accumulation exceeds a threshold and suggested that if the trigger is not
changed to an event, then it should be evaluated on a calendar day basis, which is much easier to
implement.
Response: As the flare is the affected facility, triggering the RCA on a per flare basis is
an obvious and rational approach. With respect to the SO2 RCA, we found that the consent
decree requirements for a RCA did not consider the allowable SO2 emissions from the flare but
rather the direct mass emissions (uncorrected for "allowable" emissions). Furthermore, the
expected "allowable" mass emission correction based on 162 ppmv H2S will be negligible for
events that trigger the SO2 RCA but not the flow RCA. We disagree with the commenter that
fuel gas compliant with the requirement to limit the short-term H2S concentration of fuel gas to
162 ppmv would exceed or approach 500 lb S02 in any 24-hour period during normal operating
conditions. The flare gas flow rate would have to be 20 million scfd at 162 ppmv H2S to exceed
500 lb SO2 in any 24-hour period, and in that case, the flare would have already triggered a RCA
based on the flow rate. We do note that the final rule provides for the establishment of a base
flow rate per flare under 40 CFR 60.103a(a)(4), which we expect to be higher for larger flares. In
this manner, the flare flow RCA considers flow above the baseline, effectively considering
"flaring events" and, to the extent possible, eliminating disincentives for large individual flares.
However, we see no reason why flare gas flow rates of the magnitude described above would be
considered normal operation or baseline flow rates compliant with the FMP requirements in the
final rule, as amended.
We do clarify in the rule under 40 CFR 60.103a(d) that multiple RCA are not required for
a single event. This includes both events that last for more than 24 hours or events that exceed
both the flow and SO2 RCA thresholds. Although not noted by the commenter, the per event
71
-------
method would seem to require accounting for the emissions across all affected flares at the
refinery. We considered this approach, but we concluded that the allowance to develop a flare-
specific baseline flow adequately addresses the disincentives for large individual flares with
respect to the flow limit. However, the final rule does continue to require evaluation of the SO2
and flow RCA thresholds on a 24-hour basis rather than a calendar day to more accurately
consider the emissions from a single discharge that lasts less than 24 hours but occurs over
2 calendar days. For all of the above reasons, we conclude that the final rule requirements (as
amended) appropriately consider different refinery configurations with respect to the number and
size of flares and, although the RCA is evaluated on a per flare basis, the final rule, as amended,
effectively considers the emissions "event" from that flare.
Comment: Commenters 0304 and 0305 stated that the flare flow rate threshold for RCA
of 500,000 scf in any 24-hour period is arbitrary and cannot be fairly applied to all refineries.
The commenters suggested that the flow threshold should be either deleted (Commenter 0304) or
replaced with a site-specific limit which appropriately reflects the size and complexity of the
individual flare system (Commenters 0304 and 0305). Commenter 0304 cited an ultracracker
flare that cycles from 5 million to 25 million scfd. According to the commenter, this would result
in meaningless daily RCA, and the commenter asked if this was the EPA's intent. If the flow
threshold remains, Commenter 0304 suggested that the following gases should be excluded from
the calculation of 500,000 scf: (1) hydrogen and nitrogen (i.e., compounds that are not
hydrocarbons); (2) purge and sweep gas; (3) natural gas added to increase the Btu content of the
flare gas; and (4) gases regulated by other rules (where the flare is the control device).
Commenter 0315 added that pilot gas should be excluded from the calculation of 500,000 scf gas
flow. Commenters 0308 and 0311 requested that the EPA clarify that the flow RCA be restricted
to upsets and malfunctions as initially promulgated on June 24, 2008, because pilot gas, sweep
gas, purge gas and other routine flows would trigger a RCA automatically or with a nominal
release event and because a RCA can only effectively reduce non-routine flows. Commenters
0308 and 0311 also noted that the requirements in proposed 40 CFR 60.103a(a)(l) and (4)
should not be duplicative; proposed 40 CFR 60.103a(a)(l) should focus on routine flows and
proposed 40 CFR 60.103a(a)(4) should focus on non-routine events.
Commenter 0311 suggested the flow meters required in the rule could be used to identify
step changes caused by an event from routine flow. Commenter 0315 recommended that the
72
-------
EPA adopt the approach used in the refinery consent decrees to address RCA for recurring
events. For each "flaring device," sites may prepare and submit a single RCA for one or more
root causes found by that analysis to routinely occur. The consent decrees include a provision
that the owner or operator must inform the EPA that it is electing to report only once on that root
cause, and unless the EPA objects within 30 days of receipt of the RCA, such election is
effective. Examples cited by the commenter include routine maintenance activities such as
purging vessels or clearing equipment, maintenance of a FGR system, flow from alkylation units
which cannot be recycled due to contamination issues, fuel gas imbalance issues and high
concentrations of hydrogen or nitrogen (either with or without FGR since these streams would
not be suitable for fuel gas).
Response: We proposed the flare flow RCA threshold of 500,000 scf in any 24-hour
period because we projected that FGR would be a cost effective emission reduction technique for
flares with fuel gas flows that routinely exceed 500,000 scfd, although we acknowledge that the
threshold at which FGR becomes cost effective is strongly (inversely) correlated to the average
Btu content of the flare gas (i.e., a relatively small reduction in the Btu content of the gas makes
the recovery system significantly less cost effective). Although we did not specifically exclude
sweep or purge gas from the flow, we expected that the flow rates of sweep or purge gas (i.e.,
gases needed to ensure the readiness of the flare and the safety of the flare gas system) would be
negligible when compared to the RCA threshold of 500,000 scf in any 24-hour period. In fact, in
our original analysis of the appropriate flow rate RCA threshold (Docket Item No. EPA-HQ-
OAR-2007-0011-0246), we essentially assumed that the sweep and purge gas flow rates were
zero, and we estimated costs and emissions reductions of the 500,000 scf in any 24-hour period
threshold, based on recovering that amount of gas or eliminating recurring events of that size
(rather than 500,000 scf minus the sweep or purge gas flow).
However, while we do not believe that 5 million scfd10 is a reasonable base flow for a
flare, we do acknowledge that the size of the flare, as well as the flare header system, will greatly
10 Regarding commenter's cited ultracracker flare example, it is difficult to believe that sweep gas alone accounts for
5 million scfd of flare gas flow. Additionally, a compositional analysis of the base flare gas from the normal flow,
based on data provided from a DIAL study of this refinery, suggests that the base flare gas is of sufficient quality
to recover. It also appears, based on the data provided by the commenter, that the hydrogen stream recycle
compressor was off-line approximately half the year. For such huge gas flows, considering the cost of purchasing
or producing additional hydrogen and the emissions associated with that process, it is reasonable to expect that the
facility would have a back-up compressor if the primary compressor is unreliable.
73
-------
impact the required flow needed to maintain the readiness of the flare. Although we can derive
suitable flare flow thresholds for average conditions, these thresholds are not necessarily
reasonable when applied to all flows, and we did not intend for on-going RCA to be conducted
on account of sweep or purge gas.
Therefore, rather than specifying a one-size-fits-all threshold, the final rule requires
facilities to develop their own base flare flow rates as part of their FMP. A flow-based RCA is
triggered if flows measured by the flow monitor exceed 500,000 scf greater than the base flare
flow rate in any 24-hour period. Evaluating the flow rate threshold above a baseline better
reflects our original analysis of the impacts of flow-based RCA when the sweep or purge gas
flow rates are not negligible. We also note that 40 CFR 60.103a(d) allows a single RCA to be
conducted for any single continuous discharge that causes the flare to exceed either the RCA
threshold for SO2 or flow for two or more consecutive 24-hour periods.
The final rule does not limit RCA to upsets and malfunctions of refinery process units
and ancillary equipment connected to the flare, nor does it explicitly allow owners or operators to
use one RCA report for an event that occurs routinely. When we decided to eliminate the
numerical limit on flare flow rate, we specifically increased the scope of the flare flow RCA to
cover more than just upsets and malfunctions. We also decided not to explicitly allow owners or
operators to use one RCA report for an event that occurs routinely as a means to discourage
routine flaring of recoverable gas. However, we recognize that there may be recurring discharges
to the flare that are not recoverable for various reasons. Therefore, the final rule does allow for
several base cases, which could include recurring maintenance; this provision will avoid multiple
RCA for a recurring event. As described above, the FMP (as well as significant revisions to the
plan to include alternative baselines) must be submitted to the Administrator. The Administrator
or delegated authority (e.g., the state) may review the plan, although formal approval of the plan
is not required. Not specifying a formal approval process is intended to minimize the burden
associated with reviewing FMP. Rather, the rule specifies elements of the plan that need to be
addressed in order for the plan to be considered adequate and provides an opportunity for a
delegated authority to find the plan not adequate if they choose to do so.
We expect that a final FMP in compliance with 40 CFR part 60, subpart Ja will possess
the following characteristics: (1) completeness (all gas streams are considered, all required
elements are included and all appropriate flare reduction measures are evaluated); (2) accuracy
74
-------
(the emission reductions and cost estimates for the different options are accurate); and
(3) reasonableness (the selection of reduction options is correct and the baseline flow value is
reasonable). If the Administrator identifies deficiencies in the plan (e.g., the plan does not
contain all the required elements, alternative flare reduction options were not evaluated or
selected when reasonable, the baseline or alternative baseline flow rates are considered
unreasonable), the Administrator will notify the owner or operator of the apparent deficiencies.
The owner or operator must either revise the plan to address the deficiencies or provide
additional information to document the reasonableness of the plan.
Comment: Commenters 0298, 0300 and 0304 requested that an exemption from the flare
flow RCA requirement be provided for planned startup and shutdown events that are consistent
with the facility's FMP because it is redundant and unnecessary.
Response: We agree. There will be some units whose startup and/or shutdown gas flow to
the flare will exceed the flow RCA threshold even though the gas flow to the flare has been
minimized to the extent possible. For example, a hydrotreater may receive most or all of its
hydrogen from a catalytic reforming unit. If this hydrotreater is shut down, the refinery will be
producing significant quantities of hydrogen. While some of this hydrogen may be used in the
fuel gas system, the amount of hydrogen that can be effectively redirected is limited by the
heating value needs of the fuel gas combustion units (hydrogen has a low higher heating value
compared to hydrocarbons; too much hydrogen may adversely affect performance of the process
heaters due to low heat input rates) and the amount of fuel gas that is needed by the refinery (if
the refinery has a fuel gas balance (i.e., limited or no natural gas needs prior to the unit
shutdown), the excess hydrogen caused by the shutdown cannot be used to reduce natural gas
purchases and some amount of hydrogen and/or fuel gas would then have to be flared). As long
as planned startup and shutdowns are performed according to the FMP requirements, a RCA is
not required, and we have revised the regulatory language to exclude these circumstances.
Comment: Commenter 0311 requested clarification regarding the RCA and corrective
action analysis requirements for sulfur recovery plants. The commenter asked the EPA to specify
if the 500 lb of S02 above the emission limit are to be calculated per calendar day or per 24 hour
period. In addition, the commenter noted that sulfur recovery plants with reduction control
systems emit reduced sulfur and not SO2, so a RCA threshold based on mass of SO2 is not
logical. Instead, the commenter recommended that the RCA threshold for these systems be set at
75
-------
500 lb reduced sulfur or total sulfur (depending on the type of CMS used for monitoring the
emissions) above the emission limit.
Response: As stated previously, we elected to evaluate the S02 RCA threshold on a 24-
hour basis rather than a calendar day to more accurately consider the emissions from a single
discharge that lasts less than 24 hours but occurs over 2 calendar days. However, we agree that
the description of the threshold for conducting RCA for sulfur recovery plants was confusing as
proposed, particularly when the RCA threshold for sulfur recovery plants was compared to the
RCA threshold for fuel gas combustion devices. The proposed RCA threshold for both types of
process units was 500 lb above the emission limit, but the proposed amendments directed the
owner or operator to compare the SO2 emissions to "the period of the exceedance" for fuel gas
combustion devices and "the entire 24-hour period" for sulfur recovery plants. Upon further
consideration, we see no reason for these requirements to be different for fuel gas combustion
devices and sulfur recovery plants. In addition, "period of the exceedance" was intended to have
the same meaning as "period of excess emissions" under 40 CFR 60.105a(i) and
40 CFR 60.106a(b); to avoid confusion, we have decided to use the already-defined term "period
of excess emissions." Therefore, under 40 CFR 60.103a(c)(3) of the final amendments, the
owner or operator is required to conduct RCA as described in 40 CFR 60.103a(d) "any time the
SO2 emissions are more than [500 lb] greater than the amount that would have been emitted if
the SO2 or reduced sulfur concentration was equal to the applicable emissions limit in
40 CFR 60.102a(f)(l) or (2) during one or more consecutive periods of excess emissions or any
24 hour period, whichever is shorter." This clarifying amendment is needed to ensure that the
magnitude of the emission limit exceedance is properly compared to what would have been
emitted if the emissions were equivalent to the emissions limit based on the averaging time
allowed for that emissions limit. In other words, if the owner or operator calculates four
consecutive 3- or 12- hour averages (as applicable) that are above the applicable emission limit,
the owner or operator should compare the total emissions from those four averages with the
amount that would have been emitted if the applicable emissions limit had been met over the
duration of those four averages.
The following is an example application of this provision. Consider a sulfur recovery
plant with an emission limit of 250 ppmv SO2 on a 12-hour rolling average basis. The emissions
are constant such that the hourly averages and the 12-hour rolling averages are typically about
76
-------
175 ppmv. Something then causes the hourly emissions to reach 1,000 ppmv SO2 for two
consecutive hours. When the operator calculates the 12-hour averages to determine compliance,
he finds that the high concentrations caused a total of 11 12-hour averages to be above the
250 ppmv SO2 emissions limit. To determine whether a RCA is required, the operator must
compare the mass emissions from those 11 averages with the mass that would have been emitted
if the 12-hour average for each of those 11 periods had been 250 ppmv SO2. If the difference is
500 lb, then RCA is required.
With respect to sulfur recovery plants with reductive control systems, we acknowledge
that the SO2 RCA threshold is somewhat contrived for these systems; however, we conclude that
it is appropriate so that the magnitude of the event that triggers a RCA is essentially the same
regardless of whether an oxidative or reductive control system is used. Specifically, the
concentration measurement for the reduced sulfur compounds is determined as ppm SO2 by
volume. To convert the concentration into mass emissions, the molecular weights of the
pollutants are needed. Most reduced sulfur compound monitors do not determine the
concentration of individual reduced sulfur compounds; they determine the amount of S02
produced when the reduced sulfur compounds are oxidized. The mass of reduced sulfur
compounds emitted cannot be directly determined from the monitored values. One must either
calculate the mass of sulfur emitted in the reduced sulfur compounds and/or calculate the
equivalent mass of SO2 emitted. The molecular weight for sulfur is half that of SO2. If units with
reductive control systems were allowed to calculate the excess emissions in terms of pounds of
sulfur per day rather than pounds of SO2 (equivalents) per day, an event would have to be twice
as large to exceed a 500 lb SO2 RCA threshold as compared to units with an oxidative systems.
Such a discrepancy is not appropriate considering that the emission limits for both systems are
based on the same overall sulfur recovery efficiency. Consequently, we are not revising the RCA
threshold for sulfur recovery plants that employ reductive control systems. Through this
response, we clarify that the SO2 RCA threshold must be determined considering the mass of
SO2 equivalent emissions above the applicable emissions limit for all sulfur recovery plants,
including those with reductive control systems.
77
-------
4.5 Monitoring Fuel Gas Combustion Devices
Comment: Commenter 0305 recommended that the EPA allow alternative monitoring
plans for fuel gas combustion devices approved under subpart J to be deemed compliant with
subpart Ja.
Response: The alternative monitoring plan criteria in subpart Ja were developed based on
the alternative monitoring plan requirements in subpart J, but with the additional intent to ensure
compliance with the long-term 60 ppmv H2S concentration limit. For flares, which are not
required to meet the long-term 60 ppmv H2S concentration limit, alternative monitoring plans
approved under subpart J are deemed compliant with subpart Ja. However, for fuel gas
combustion devices, which are required to meet the long-term 60 ppmv H2S concentration limit,
the subpart Ja criteria must be used to develop a new alternative monitoring plan.
4.6 Flow Monitoring for Flares
Comment: Commenters 0305 and 0311 stated that "emergency only" or "SSM only"
flares should be exempt from flow monitoring requirements, and Commenter 0311 also
requested clarification that secondary and "malfunction only" flares are exempt from the flow
meter requirement. Commenter 0311 stated that flare flow meters are costly and unnecessary
because engineering calculations, which are currently used, are sufficient to evaluate whether the
flare flow rate exceeds the RCA threshold of 500,000 scf in any 24-hour period. At a minimum,
flares that are designed to handle flows less than 500,000 scfd should be excluded from the flow
monitoring requirements, as there is no justification for flow monitors in this case. Commenter
0315 also noted that flow monitors are unnecessary for pilot gas, which has no significant
increase or decrease potential and which does not flow through the main flare header and thus
would require a separate flow monitor. Commenter 0303 stated that, for flares with FGR
systems, the pressure drop across the flare seal drum can be used to calculate flow rate and this
method is a cost effective alternative to continuous flow monitors for these flares.
Response: In the final rule, flow monitoring is used to determine whether a RCA is
required rather than to ensure compliance with a specific flow limit. We have reviewed the
commenters' suggestions and agree that, in certain specific cases, monitoring is not necessary
and should not be required. However, as a general rule, we believe flow monitors are needed, not
only to provide a verifiable measure of exceedances of the flow RCA threshold, but also
78
-------
exceedances of the RCA threshold of 500 lb SO2 in any 24-hour period. In addition, when we
evaluated local rules, such as the initial B AAQMD rule for flare monitoring, we saw that the
measured flare flow rates were several times greater than previously projected by the facilities.
Consequently, we find great value in the flow monitoring requirements for flares. These
monitoring requirements will greatly improve the accuracy of emissions estimates from these
flares. The resulting improved accuracy of flare emissions estimates will also lead to better
decision-making as we conduct future reviews of rules applicable to petroleum refineries. We did
consider each of the commenters' suggested exemptions in light of this fact; our specific
considerations follow.
We did not specifically consider that some flares would not be capable of exceeding the
flow RCA threshold {i.e., designed to handle less than 500,000 scfd of gas). However, these
small flares could still exceed the RCA threshold of 500 lb SO2 in any 24-hour period. As such,
we did not provide an exemption from the monitoring requirements for these small flares.
We agree that the monitoring of pilot gas flow is not needed. In the final rule, a RCA is
required if the gas flow to the flare exceeds 500,000 scf above the baseline in any 24-hour
period. The flow of pilot gas is considered to be part of the baseline flow and is assumed to be
constant. As such, monitoring of pilot gas would not be necessary to determine whether a flare
has exceeded 500,000 scf above the baseline in any 24-hour period. In practice, the actual
baseline flow set for the flare may or may not expressly include the pilot gas flow rate.
Generally, the configuration of the flare header is such that the flare flow monitor would not
measure pilot gas flow. In this case, the baseline flow determined for the flare would not
expressly include the pilot gas flow rate. If the flare flow monitor is configured in such a way
that it does measure pilot gas, then pilot gas would be considered part of the baseline conditions
for that flare.
We agree with commenters that flares with FGR systems do have unique conditions and
these warrant alternative monitoring options. Additionally, we recognize that the monitoring
requirements may be burdensome for flares that are truly "emergency only" {i.e., flares that flare
gas rarely, if at all, during a typical year) or for secondary flares in a cascaded flare system.
These flares are expected to have a water seal that prevents flare use during normal operations
and ensures that the pressure upstream of the water seal (expressed in inches of water) does not
exceed the water seal height during normal operations (hereafter referred to as "properly
79
-------
maintain a water seal"). We find that, for these select types of flares, water seal monitoring as an
alternative to the flow (and sulfur) monitoring provisions is appropriate.
For flares with a FGR system and other emergency or secondary flares that properly
maintain a water seal, the final rule states that an owner or operator may elect to monitor the
pressure in the gas header just before the water seal and monitor the water seal liquid height to
verify that the flare header pressure is less than the water seal, which is an indication that no flow
of gas occurs. If the flare header pressure exceeds the water seal liquid level, a RCA is triggered
unless the pressure exceedance is attributable to staging of compressors. This alternative reduces
the costs associated with installing sulfur and flow monitoring systems for flares that rarely
receive fuel gas. Engineering calculations can be used to estimate the emissions during the event,
but not for determining whether or not a RCA is required.
To ensure that this option is only used for flares that are truly emergency flares and not
for flares that are used for routine discharges, the final rule contains a limit on the number of
pressure exceedances requiring RCA that can occur in one year. Following the fifth reportable
pressure exceedance in any consecutive 365 days, the owner or operator must comply with the
sulfur and flow monitoring requirements of 40 CFR 60.107a(e) and (f). Based on a review of
available flaring data, we expect that gas may be sent to an emergency flare three to four times
per year, on average. Consistent with this information, we are providing in these final
amendments that an "emergency flare" may receive up to four releases to the flare in any
consecutive 365-day period to account for year-to-year variability. However, a flare receiving
more than four discharges in a consecutive 365-day period can no longer be considered an
"emergency flare" and must install the required sulfur and flow monitors.
Comment: Commenter 0305 expressed concern regarding the flow meter range for
emergency-only flares and suggested that for existing monitors, the flow range currently being
used should be deemed acceptable, even if it means additional engineering and process
information must supplement the readings to estimate emissions. The commenter also suggested
that the range for new meters should be determined as appropriate for the specific flare.
Commenter 0311 suggested that the flow meter range requirements should be clarified to
accommodate an average flow rate of 500,000 scfd, and multiple flow meters should not be
required to measure very high or very low flows.
80
-------
Response: Flow meters for flares are considered continuous parameter monitoring
systems (CPMS) and no specific span is specified in the rule. The rule requires that the flow
monitor be placed in a position to provide a representative measurement of the total gas flow rate
and it specifies a minimum sensitivity and calibration requirements. We agree that the flow
meter should accommodate an average flow rate of 500,000 scfd and we did not intend to require
multiple flow monitors to measure exceedingly low or high flow rates. It is our understanding
that ultrasonic flow monitoring systems are typically used for flares and that these devices can
accurately measure flow rates over a large span (e.g., velocities from 1 to 3,000 feet per second
[ft/s]). As such, we anticipate that a single flow meter installed for flares will be able to measure
a wide range of flow rates accurately and that the range measured by a single flow meter device
will be sufficient to assess both the sulfur and flow rate RCA thresholds.
Comment: Commenter 0311 noted that the newly proposed Appendix F QA/QC
procedure (P-4) for CPMS will likely result in significant periods of monitor outages if
applicable to fuel gas flow meters.
Response: This comment lacks any detail supporting the commenter's position.
Presumably, if this is a legitimate concern, the comment should have also provided direct and
more substantive comments on the proposed Procedure 4 to Appendix F. We defer to the
supporting material and responses to comment document for that action; please see Docket No.
EPA-HQ-OAR-2006-0640 for more details regarding that particular rulemaking.
4.7 Sulfur Monitoring for Flares
Comment: Commenter 0301 stated that flares that handle only exempt gases, specifically
flares with FGR systems, should be exempt from H2S monitoring. According to Commenter
0301, process knowledge and engineering calculations are sufficient to determine if the SO2
RCA threshold of 500 lb in any 24-hour period is exceeded. If H2S monitoring is required,
Commenter 0301 requested that the rule allow engineering estimates of the TRS-to-H2S
correction factor due to safety concerns and inability to collect samples for "emergency only"
flares. Commenter 0314 agreed that monitoring is unnecessary for flares with adequately
designed and operated FGR systems and that there are data at the flare gas source to calculate the
amount of S02 emitted and determine if RCA is needed. Commenter 0303 stated that continuous
sulfur monitoring is unreasonable for flares that generally see little or no gas flow, and sulfur
81
-------
emissions can be calculated based on sampling the flare gas (although the commenter also stated
that 42 percent of their flaring events were less than 10 minutes in duration). Commenter 0303
requested that 40 CFR 60.107a(e) and (f) and 40 CFR 60.108a(6)(iii), (iv) and (v) be modified to
allow owners or operators to submit and use an alternative monitoring plan to demonstrate that
the FGR system is operating within its capacity and to calculate SO2 emissions from engineering
calculations and flare gas sampling. Commenters 0305 and 0311 agreed that "emergency only"
or "SSM only" flares should be exempt from sulfur monitoring requirements. Commenter 0315
noted that it is common practice to configure flare systems such that one flare handles normal
loads and another flare receives gas only when the primary flare is overloaded during periods of
SSM. The commenter noted that, in this situation, the continuous sulfur monitor on the primary
flare would often provide data on the sulfur content of the flared gas (depending on the exact
configuration). The commenter asserted that the burdens related to a continuous sulfur analyzer
are not justified for flares where flows are very intermittent and event-related and where sulfur
data is easily available by other means (e.g., grab samples, process data). Commenters 0301,
0305 and 0311 recommended that monitoring exemptions be provided for flares in low sulfur
service (less than 5 ppmv or sulfur-intolerant streams), similar to the exemptions provided in
40 CFR 60.107a(a)(3).
Commenter 0296 stated that properly designed FGR systems should exempt a flare from
the H2S concentration requirements in subpart Ja as only exempt gas streams would be
combusted by the flares. Commenter 0296 also stated that the EPA should provide incentives for
refineries to install costly FGR systems. For example, the EPA should eliminate the TRS
monitoring requirements for flares with FGR systems, allow the use of an existing fuel gas
continuous monitoring system (CMS) if the refinery is long on fuel gas and subsequently must
flare some of the recovered gas and allow 5 years for installation of a FGR system.
Commenter 0305 requested clarification that when sweep gas is routed to an emergency-
only flare, it is not required to be monitored if it meets one of the definitions of streams that do
not require monitoring or is monitored elsewhere (for example within the fuel gas system).
Similarly, Commenter 0305 requested clarification that flare pilot gas is not required to be
monitored for sulfur or flow if it is routed from natural gas or from refinery fuel gas monitored
elsewhere for compliance with subpart J or Ja, nor is it required to be included in the flare flow
for the purposes of RCA.
82
-------
Commenter 0314 noted that, to meet the performance test requirements on flare sulfur
monitors as proposed at 40 CFR 60.107a(e)(l)(ii) or (e)(2)(ii), the owner or operator of a flare
with FGR may have to intentionally flare sour gas. The commenter also noted that the relative
accuracy test audit (RATA) may be enough to trigger RCA. Therefore, the commenter
recommended revising the performance test requirements for flares with FGR to require only a
Cylinder Gas Audit. Commenter 0311 also noted that flares with occasional flows, such as flares
handling only SSM events, would have to flare intentionally to perform required testing.
Response: We have amended the final rule so that gases that are exempt from H2S
monitoring due to low sulfur content are also exempt from sulfur monitoring requirements for
flares. For low-sulfur gases, the flare RCA will always be triggered by an exceedance of the flow
rate threshold well before the S02 threshold is exceeded, so no sulfur monitoring is required.
However, this exemption can only be used for flares that are configured to receive only fuel gas
streams that are inherently low in sulfur content, as described in 40 CFR 60.107a(a)(3), such as
flares used for pressure relief of propane or butane product spheres (fuel gas streams meeting
commercial grade product specifications for sulfur content of 30 ppmv or less) or flares used to
combust fuel gas streams produced in process units that are intolerant to sulfur contamination
(e.g., hydrogen plant, catalytic reforming unit, isomerization unit, hydrogen fluoride alkylation
unit). We already clarified that flare pilot gas is not required to be monitored (see Section 4.6).
Also, 40 CFR part 60, subpart Ja already allows for H2S monitoring at a central location, such as
the fuel mix drum, for all fuel gas combustion devices (and we are finalizing amendments to
ensure it is clear that H2S monitoring at a central location is allowed for flares as well). Thus, we
agree that if a flare only burns natural gas, fuel gas monitored elsewhere or fuel gas streams that
are inherently low in sulfur content (as defined in 40 CFR 60.107a(a)(3)), then no H2S monitor is
needed.
The remaining issue is whether or not sulfur monitoring is necessary for "emergency
only" flares. (An emergency flare is defined as a flare that combusts gas exclusively released as a
result of malfunctions (and not startup, shutdown, routine operations or any other cause) on four
or fewer occasions in a rolling 365-day period. For purposes of the rule, a flare cannot be
categorized as an "emergency flare" unless it maintains a water seal.) We acknowledge that there
are difficulties and costs with installing monitors on flares that rarely operate. However, we are
concerned about how the owner or operator will detect emissions above 500 lb S02 in any 24-
83
-------
hour period during an upset or malfunction of a refinery process unit or ancillary equipment
connected to the flare. Commenters appear to have conflicting opinions regarding the ability to
sample the flare gas to determine the sulfur content (or total sulfur-to-H2S ratio) during a flaring
event. If samples could be taken during the flaring events, then that would be a potential option.
However, during a process upset or malfunction, focus should be on alleviating the problem
rather than taking a special sample. Also, given the duration of some of these events, it appears
unlikely that representative samples can be manually collected.
Taking the difficulties discussed above into account, we have developed an alternative
monitoring option for emergency flares. As noted in the previous response, emergency flares are
expected to properly maintain a water seal. We provide pressure and water seal liquid level
monitoring, as previously described as an alternative to the sulfur and flow monitors. As
described in more detail above, any fuel gas pressure exceeding the water seal liquid level
triggers a RCA and there is a limit to the number of exceedances in one year. Under this option,
a RCA is triggered based on the monitored pressure and water seal height, so accurate
measurements of flow rate and sulfur concentrations are less critical than for flares that must
evaluate these parameters to determine if a RCA is needed. Consequently, for these flares,
engineering calculations can be used to estimate the reported emissions during the flaring event,
but the RCA must be performed regardless of the magnitude of these engineering estimates.
Using this alternative monitoring option, emergency flares are not required to install continuous
sulfur monitoring systems. Flares that do not meet the conditions of an emergency flare are
required to install continuous sulfur monitoring systems and cannot elect this alternative
monitoring option.
We also agree that flaring solely for the purpose of a RATA or other performance test is
not desirable. The "cylinder gas audit" procedures requested by the commenter are described as
alternative relative accuracy procedures in section 16.0 of Performance Specification 2
(referenced from Performance Specification 5). We reviewed the alternative relative accuracy
procedures and considered how they may apply to flares, and we have determined that the
alternative relative accuracy procedures are appropriate for flares. We expect that, for most
affected flares, the variability in flow (including no flow conditions) and sulfur content of the
gases discharged to the flare create significant barriers to the normally required relative accuracy
assessments, particularly if those assessments need to be made over a range of sulfur
84
-------
concentrations potentially seen by the monitor. Therefore, we are amending 40 CFR
60.107a(e)(l)(ii) and 40 CFR 60.107a(e)(2)(ii) to specify that the owner or operator of a flare
may elect to use the alternative relative accuracy procedures in section 16.0 of Performance
Specification 2 of Appendix B to part 60. As required by 40 CFR 60.108a(b), the owner or
operator shall notify the Administrator of their intent to use the alternative relative accuracy
procedures.
Comment: Commenter 0312 requested that the EPA clarify whether the additionally
proposed sulfur monitoring options for flares are for TRS or total sulfur. The commenter noted
that measuring total sulfur is the simplest and most inclusive measurement of SO2 emissions and
it is the method included in SCAQMD Rule 1118. The commenter also requested that methods
for measuring total sulfur in gaseous fuels be included as acceptable options to perform the
relative accuracy evaluations of the CMS.
Commenter 0311 requested that provisions be made in 40 CFR 60.107a(e)(2) to develop
a total sulfur-to-H2S (or TRS-to-H2S) ratio so that the total sulfur monitor can be used for both
the RCA and for compliance with the requirement to limit the short-term H2S concentration in
fuel gas sent to a flare to 162 ppmv without the need for a duplicative H2S CMS. Commenter
0313 supported the addition of alternative monitoring methods for the sulfur content of flare gas
but noted that since the composition of flare gas is highly variable, the alternative methods must
meet continuous monitoring requirements.
Response: We have clarified and consolidated the monitoring requirements to allow TRS
monitoring for flares. For the purposes of evaluating the S02 RCA threshold, total sulfur
monitoring provides the most accurate assessment. However, in most cases, the vast majority of
sulfur contained in gases discharged to the flare is expected to be in the form of TRS compounds,
which include carbon disulfide, COS and H2S. Our test method for measuring TRS includes the
use of EPA Method 15 A as a reference method, and because EPA Method 15A measures total
sulfur, the TRS monitoring requirement is equivalent to a total sulfur monitoring method.
As discussed previously, we are relying on the suite of flare requirements we are
promulgating to limit S02 emissions at the flare. These include optimizing management of the
fuel gas by limiting the short-term concentration of H2S to 162 ppmv during normal operating
conditions. We expected most refineries would already have the H2S monitor and did not
consider the use of a total sulfur monitor for use in complying with the short-term 162 ppmv H2S
85
-------
concentration in the fuel gas. As the H2S concentration will always be less than the TRS
concentration, it is acceptable to use the TRS monitor to verify that the fuel gas does not exceed
the short-term H2S concentration of 162 ppmv. Therefore, we have provided for the use of TRS
monitors, provided the monitor can also meet the 300 ppmv span requirement.
However, we have not provided a correction factor to scale down the TRS concentration
to H2S. The owner or operator using this method must essentially be able to demonstrate they
can achieve a 162 ppmv TRS concentration in the fuel gas. The concentration ratio was provided
for the purposes of the RCA because of the costs of adding a total sulfur monitoring system
when a dual range H2S monitor was already in-place, as well as the expected accuracy needed
for the system to assess the S02 RCA threshold. As few cases would exist where the flaring
event would be right at the S02 RCA threshold of 500 lb in any 24-hour period, inaccuracies
associated with the average total sulfur-to-H2S ratio were not expected to be significant.
On the other hand, the short-term 162 ppmv H2S concentration in the fuel gas must be
continuously maintained, and the total sulfur-to-H2S ratio at these low concentrations is expected
to be highly variable, depending on the efficiency of the amine scrubber systems. As the amine
scrubber systems, according to previous industry comments, are not effective for reduced sulfur
compounds other than H2S, the non-H2S reduced sulfur concentration is expected to be fairly
constant, with most of the fluctuations in total sulfur content being attributable to fluctuations in
H2S concentrations. Consequently, we have determined that the inaccuracies of the ratio
approach are not acceptable for continuously demonstrating that the short-term concentration in
the fuel gas does not exceed 162 ppmv H2S. Therefore, owners or operators of affected flares
may use the direct output of a TRS monitor to assess compliance with the short-term 162 ppmv
H2S concentration in the fuel gas, or they must install a continuous H2S monitor.
Comment: Commenter 0311 supported the proposed amendment revising the span value
for fuel gas H2S analyzers to match the span requirements in subpart J, stating this will save time
and money. However, Commenter 0301 stated that the span value for the flare H2S monitoring
option is too restrictive and suggested that the Appendix F requirements provide sufficient
quality assurance/quality control (QA/QC) without the need for the rule to specify the span
range.
Commenter 0311 requested clarification of the sulfur monitor span for flares, suggesting
that it should be based on the 162 ppmv H2S concentration requirement and that engineering
86
-------
calculations can be used to assess exceedances of the SO2 RCA threshold of 500 lb in any 24-
hour period.
Response: The H2S span value is at 300 ppmv to verify compliance with the H2S
concentration requirement for the fuel gas; the span of the total sulfur monitor needs to be much
greater than that to be able to quantify the sulfur content in streams containing several percent
sulfur. For units that use the H2S analyzers both to assess compliance with the short-term 162
ppmv H2S concentration requirement for the fuel gas and to assess exceedances of the S02 RCA
threshold of 500 lb in any 24-hour period, a dual range monitor will be necessary. For the
purposes of the S02 RCA threshold of 500 lb in any 24-hour period, we intended that the
monitor be capable of accurately determining the sulfur concentration for the range of
concentrations expected to be seen at the flare. We are particularly interested in quantifying the
concentrations of high sulfur-containing streams as these would be the streams most likely to
trigger a RCA at low flows. We proposed that the span for the flare sulfur monitor be selected
from a range of 1 to 5 percent. We agree with the commenter that this may be too restrictive, and
we have revised the span requirements to be determined based on the maximum sulfur content of
gas that can be discharged to the flare (e.g., roughly 1.1 to 1.3 times the maximum anticipated
sulfur concentration), but no less than 5,000 ppmv. A single dual range monitor may be used to
comply with the short-term 162 ppmv H2S concentration requirement for the fuel gas and the
S02 RCA threshold monitoring requirement provided the applicable span specifications are met.
In reviewing the span specifications, we noted that span requirements were inadvertently omitted
from the TRS compound monitoring alternative. The purpose of these monitors is identical to the
H2S monitoring alternative, and the same span considerations apply for these monitors.
We disagree that the QA/QC procedures in Appendix F to part 60 are sufficient without
specifying the span values. Procedure 1 of Appendix F to part 60 defines "span value" as: "The
upper limit of a gas concentration measurement range that is specified for affected source
categories in the applicable subpart of the regulation." The concentrations used for calibration
are based on the span value. Several of the QA/QC procedures in Appendix F are undefined if
the span value is not defined in the rule.
87
-------
4.8 Compliance Schedule
Comment: Commenters 0296, 0301, 0304, 0306, 0310 and 0311 stated that compliance
with a long-term 60 ppmv H2S concentration limit on fuel gas combusted in flares would require
a much longer time period than 2 years. The commenters suggested that 3 to 5 years would be
needed to install a FGR system and/or additional amine treatment needed for compliance.
Commenter 0310 further stated that the compliance schedule in subpart Ja for installing
additional sulfur removal equipment should be consistent with (or allow) established consent
decree schedules. Both Commenters 0310 and 0311 objected to phasing out the additional time
after the rule has been in place for 5 years.
Commenter 0311 stated that all flares should be given 3 years from the date of final
promulgation and lifting of the stay to comply with the flare standards (regardless of whether
new amine or FGR systems are being installed) to allow time for planning, re-piping and other
necessary activities, and the EPA should provide an optional additional 2 years for capital
installation projects (e.g., for permitting, equipment installation). Commenter 0305 stated that
any capital project (not just amine treatment) required to meet any of the subpart Ja flare
standards (not just the long-term 60 ppmv H2S concentration) should trigger a compliance
deadline of up to 3 years after the effective date of the rule, with an allowance provided for
refineries to request an additional 2 years if needed to correspond with an affected-unit
turnaround because of the huge operating costs and product disruptions associated with
unscheduled turnarounds. Commenter 0301 recommended that the extra time to begin RCA
provided to refiners committing to install FGR systems in proposed 40 CFR 60.103a(a)(4) and
(b)(1) should also be provided to refiners committing to expand an existing FGR system.
Commenter 0315 agreed that additional time is needed since a number of the flare requirements
will require shutdowns. The commenter also stated that since not all flares are currently subject
to subpart J, additional time would be needed for all fuel gas combusted in those flares to meet
the short-term H2S concentration requirement of 162 ppmv.
Commenter 0304 stated that the 18-month compliance deadline for sulfur CMS for flares
is not adequate for large refineries with multiple flares because the installation of the CMS will
typically need to be staggered (based on availability of CMS equipment and crew). Commenter
0310 similarly asserted that 3 years would be needed to install monitors due to design and
installation considerations together with the constraints on manufacturers by the sudden demand
88
-------
for several hundred flow and sulfur monitors. Both commenters cited delays experienced in
obtaining monitors to comply with SCAQMD Rule 1118.
Commenter 0295 requested clarification regarding the trigger date from which the
additional time to comply with the flare provisions (e.g., 2 years when installing a FGR system)
begins. The commenter asked whether the trigger date is when construction starts, at startup or
when the stay is removed (or whichever is later). Commenter 0315 agreed that the EPA should
set the compliance time based on the initial startup of the modification. The commenter noted
that the EPA should follow the 40 CFR part 60 General Provisions for performance test timing
and the 40 CFR part 63 General Provisions for compliance timing.
Response: As we are no longer applying the long-term 60 ppmv H2S fuel gas
concentration limit to flares, the comments related to the amount of time needed to comply with
a long-term 60 ppmv H2S fuel gas concentration limit are moot. We do, however, recognize that
a flare modification can occur much more quickly than modifications of traditional process-
related emission sources. Therefore, we evaluated the comments regarding the amount of time
needed to meet the various requirements for flares while keeping the 40 CFR part 60, subpart Ja
flare modification provision in mind. We discuss each requirement and the time for
demonstrating compliance with that requirement in the following paragraphs.
We find it appropriate to require modified flares that already have adequate treatment and
monitoring equipment in place to achieve a short-term H2S concentration of 162 ppmv (resulting
from compliance with 40 CFR part 60, subpart J) to continue to meet that concentration upon
startup of the affected flare or the effective date of this final rule, whichever is later. However,
some flares are not affected facilities subject to 40 CFR part 60, subpart J, and others are
complying with subpart J requirements as specified in consent decrees or have received
alternative monitoring plans by which to demonstrate compliance with the short-term H2S
concentration limit. In these cases, we find it appropriate to allow more time to comply with the
short-term H2S concentration limit and/or the associated monitoring requirements because
additional amine treatment and/or monitoring systems will be required to comply with the rule.
Therefore, the final rule requires all modified flares that are newly subject to 40 CFR part
60, subpart Ja (but were not previously subject to 40 CFR part 60, subpart J) to comply with the
short-term H2S concentration limit and applicable monitoring requirements no later than 3 years
after the effective date of this final rule or upon startup of the affected flare, whichever is later.
89
-------
Modified flares that have accepted applicability of subpart J under a federal consent decree shall
comply with the subpart J requirements as specified in the consent decree but shall comply with
the short-term H2S concentration limit and applicable monitoring requirements no later than
3 years after the effective date of this final rule. Modified flares that are already subject to the
162 ppmv short-term H2S concentration limit under subpart J must meet the short-term H2S
concentration limit under subpart Ja upon startup of the affected flare or the effective date of this
final rule, whichever is later. Finally, modified flares that are already subject to the short-term
H2S concentration limit but that have an approved monitoring alternative under subpart J and do
not have the monitoring equipment in-place that is required under subpart Ja shall be given up to
3 years from the effective date of this final rule to install the monitors required by subpart Ja (or
to obtain an approved monitoring alternative under subpart Ja).
As we noted in the preamble to the proposed amendments, many of the connections that
would trigger applicability to 40 CFR part 60, subpart Ja are critical to the safe and efficient
operation of the refinery. These connections can, and often must, be installed quickly. At the
same time, nearly all refineries will need time for planning, designing, purchasing and installing
(including any necessary re-piping) sulfur and flow monitors that are newly required by subpart
Ja. Some refineries will elect to add FGR and/or sulfur treatment equipment to minimize their
emissions as part of the evaluations conducted as required by the new FMP requirements, and
time will be needed for planning, designing, purchasing and installing these components as well.
Given that many flares will become modified affected sources relatively quickly, owners and
operators will be competing with one another for the services and products of a finite number of
vendors who provide the necessary monitors and other equipment. Several commenters
specifically noted availability of monitors as an issue when complying with SCAQMD Rule
1118. As such, we find that immediate compliance with the requirements for flares, such as the
planning, designing, purchasing and installation of (including any necessary re-piping) sulfur and
flow monitors, may be difficult for operators to meet, especially in situations where quick
connections to the flare are made. A phased compliance schedule allows for the operators to
comply with some requirements associated with flares, such as continuing to achieve a short-
term H2S concentration of 162 ppmv, if the flares are already subject to 40 CFR part 60,
subpart J and have adequate monitoring in place to comply with this final rule, while allowing
90
-------
time to install treatment and processing equipment and monitoring equipment to comply with the
standards where necessary.
A phased compliance schedule will also allow owners and operators to minimize process
interruption by coordinating the installation of monitoring equipment with process shutdowns or
turnarounds. In addition to providing operating flexibility to the refinery, we are taking into
consideration the fact that a process shutdown and subsequent startup can generate significant
emissions, even if the refinery is taking care to minimize those emissions. We consider a phased
compliance schedule that allows owners and operators to avoid startups and shutdowns that are
not necessary to maintain the equipment and process to be environmentally beneficial overall
and the best system of emissions reduction for a quickly modified flare. Considering the time
needed to complete engineering specifications, order and install the required monitoring
equipment, and considering the need to coordinate this installation with process unit shutdown or
turnarounds, we determined that completion of these activities within 3 years is consistent with
the best system of emissions reductions for quickly modified flares.
We note, however, that this phased compliance schedule for the flare requirements in
40 CFR part 60, subpart Ja is intended for those situations when a flare modification occurs
quickly and the owner or operator does not have significant planning opportunities to install the
required monitors or implement the selected flare minimization options without significant
process interruptions. For a future large project on a schedule that includes time for planning,
designing, purchasing and installing equipment and monitors, we expect that the owner and
operator will have time to assess whether or not the refinery flares will become affected sources
through modification. If a project will result in the modification of a flare, we expect that the
owner or operator will then plan how to meet the standards in subpart Ja as part of the project
itself, including the installation of the monitoring systems and the development of a FMP.
Because of the ability to plan ahead, flares that are modified as part of a large project will not
have all of the difficulties meeting the subpart Ja flare requirements upon completion of the
modification as those flares that are modified quickly. Therefore, we find that compliance with
the flare requirements upon startup of the modified flare is appropriate and consistent with the
best system of emissions reduction for large projects resulting in a modification of a flare. Thus,
we determined that the appropriate time period for compliance with the flare standards is either:
(1)3 years from the effective date of these amendments or (2) upon startup of the modified flare,
91
-------
whichever is later.11 In this manner, flares that become subject to subpart Ja quickly, based on a
small safety-related connection (or have already become subject to subpart Ja based on a
modification prior to the effective date of these amendments) will have up to 3 years from the
effective date of these amendments to comply fully with the flare standards, but flares that are
modified as the result of a significant project, such as the installation of a new process unit that
will be tied into an existing flare, will effectively be required to comply with the flare standards
at the startup of the new process unit.
Therefore, for the reasons described above, we are providing flares that become affected
facilities subject to 40 CFR part 60, subpart Ja through modification with a phased compliance
schedule for the flare standards, as described in this paragraph. The final rule requires owners
and operators of modified flares to meet the short-term 162 ppmv H2S concentration requirement
by the effective date of these amendments or upon startup of the affected flare (whichever is
later) only if they are already subject to the short-term 162 ppmv H2S concentration limit in
40 CFR part 60, subpart J. Modified flares that were not affected flares under subpart J prior to
being modified facilities under subpart Ja must comply with the short-term 162 ppmv H2S
concentration requirement within 3 years of the effective date of these amendments or upon
startup of the modified flare, whichever is later. Owners and operators of modified flares that
have accepted applicability of subpart J under a federal consent decree shall comply with the
subpart J requirements as specified in the consent decree, but must meet the short-term 162 ppmv
H2S concentration limit no later than 3 years after the effective date of this final rule. Owners and
operators of modified flares that are already subject to subpart J and that have an approved
monitoring alternative and are unable to meet the applicable subpart Ja monitoring requirements
for the short-term H2S concentration limit must meet the short-term H2S concentration
requirement upon startup of the affected flare or the effective date of this final rule, whichever is
later, but shall be given up to 3 years from the effective date of this final rule to install the
monitors required by subpart Ja. In this interim period, owners and operators of these modified
flares shall demonstrate compliance with the short-term H2S concentration limit using the
monitoring alternative approved under subpart J.
11 For the purposes of this subpart, startup of the modified flare occurs when any of the activities in 40 CFR
60.100a(c)(l) or (2) is completed (e.g., when a new connection is made to a flare such that flow from a refinery
process unit or ancillary equipment can flow to the flare via that new connection).
92
-------
Additionally, we are requiring owners and operators of modified flares to complete and
implement the FMP under 40 CFR 60.103a(a) by 3 years from the effective date of these
amendments or upon startup of the modified flare, whichever is later. We are requiring owners
and operators of modified flares to begin conducting root cause and corrective action analyses
under 40 CFR 60.103a(c) and (d) no later than 3 years from the effective date of these
amendments or the date of the startup of the modified flare, whichever is later, so that the facility
can complete the FMP and establish baseline flow rates prior to performing the root cause and
corrective action analyses. We are also requiring owners and operators of modified flares to
install and begin operating the monitors necessary to demonstrate compliance with these
provisions, as required under 40 CFR 60.107a(e) through (g) within 3 years from the effective
date of these amendments or by the startup date of the modified flare, whichever is later, when
the monitors are not already in place. Compliance with the phased compliance schedule
constitutes compliance with the flare standards as of the effective date.
We note that the final rule does not provide a phased compliance schedule for new and
reconstructed flares. The final rule requires owners and operators of new and reconstructed flares
to meet all the flare requirements, including the short-term 162 ppmv H2S concentration
requirement, upon the effective date of the requirements or upon startup of the affected flare,
whichever is later.
4.9 Startup, Shutdown and Malfunction
Comment: Commenter 0307 stated that exempting flares12 from the H2S concentration
limits during SSM events is illegal because the CAA requires continuous compliance with
standards issued under CAA section 111. See CAA sections 111(a)(1), 302(k). Therefore, the
commenter asserted that it is not BSER to allow startup and shutdown events to be exempt from
compliance, and these emissions may not be regulated through work practice standards. For
support, the commenter cited Sierra Club v. EPA, 551 F.3d 1019 (D.C. Cir. 2008), in which the
Court stated: "When sections 112 and 302(k) are read together, then, Congress has required that
there must be continuous section 112-compliant standards." The commenter noted that the Court
12 The comments submitted referenced "fuel gas combustion devices" as the affected source when describing the
exemption during SSM events. However, the exemption only applies to flares. See 40 CFR 60.103a(h). The
discussion in this document is therefore focused on flares as distinguished from other types of fuel gas
combustion devices that are required to comply at all times with the limits in 40 CFR 60.102a(g)(l).
93
-------
found that the exemption from compliance with CAA section 112 standards during SSM events
violates the CAA because the section 111 general duty to minimize emissions during SSM
events (see 40 CFR 60.11(d)) is not a CAA section 112-complaint standard. The commenter
asserted that the CAA also requires that a section 111-compliant standard that reflects BSER13 be
in effect at all times for flares.
Commenter 0307 further asserted that work practice standards for flares are not CAA
section 111-compliant standards because this is not one of those "limited instances" in which
CAA section 111(h) authorizes such standards. The commenter stated that the EPA must show
that a standard of performance for flares is "not feasible to prescribe or enforce" because "(A) a
pollutant... cannot be emitted through a conveyance designed and constructed to emit or capture
such pollutant, or that any requirement for, or use of, such a conveyance would be inconsistent
with any Federal, State, or local law, or (B) .. .the application of measurement methodology to a
particular class of sources is not practicable due to technological or economic limitations." See
CAA section 111(h)(2). The commenter stated that neither of these exemptions appear to apply
and the EPA cannot claim that it is infeasible to promulgate a standard of performance for
flares,14 so the EPA cannot set a work practice standard for flares. Thus, the commenter asserted
that a CAA section 111-compliant standard does not continuously apply to flares since both the
exemption from the H2S concentration limits during SSM events and the flare work practice
standards are not lawful under the CAA.
Commenter 0315 disagreed and provided several reasons why they believe the EPA may
lawfully exempt flares from the H2S concentration limits during SSM events. First, the
commenter noted that 40 CFR part 60, subpart Ja was promulgated as part of the mandatory
periodic review of 40 CFR part 60, subpart J required by CAA section 111(b)(1)(B), which states
that "the Administrator need not review any such [existing] standard if the Administrator
13 The commenter asserted, without providing support, that it is not BSER to exempt flares from the H2S
concentration limits during startup and shutdown events. The commenter also stated that the EPA, at a minimum,
must demonstrate how the exemption from the H2S concentration limits during SSM events does in fact represent
BSER, but the commenter stated that the EPA has failed to make this demonstration.
14 The commenter cited the EPA's rationale for proposing work practice standards for flaring in which we state: "It
is not feasible to prescribe or enforce a standard of performance for these sources because either the pollution
prevention measures eliminate the emission source, so that there are no emissions to capture and convey, or the
emissions are so transient, and in some cases, occur so randomly, that the application of a measurement
methodology to these sources is not technically and economically practical." 72 FR 27178, 27194-27195 (May
14, 2007). In response, the commenter states: "[T]he plain language of the Act recognizes that standards of
performance leading to the 'capture' of emissions are not infeasible [citation omitted], and EPA has proposed to
apply measurement methodologies to flares in spite of the transience of their emissions."
94
-------
determines that such review is not appropriate in light of readily available information on the
efficacy of such [existing] standard." The commenter noted that subpart J exempts a flare from
the H2S concentration limits when combusting certain gases generated during SSM events (see
40 CFR 60.104(a)(1), 60.101(e)) and stated that the record contains "ample evidence" to support
maintaining that provision in subpart Ja. The commenter asserted that including these same
provisions in subpart Ja is "an appropriate exercise of the EPA's authority to 'not review' this
aspect of the existing standard in light of the efficacy of the existing standard." See CAA section
111(b)(1)(B).
Second, Commenter 0315 noted that the Sierra Club decision was largely grounded in the
Court's determination that Congress amended CAA section 112 out of concern "about the slow
pace of EPA's regulation of HAPs" (551 F.3d at 1028), eliminating much of the EPA's
discretion and requiring sources to "meet the strictest standards" without variance "based on
different time periods." The commenter further explained that the Court pointed to CAA section
112(d)(1) regarding the EPA's authority to "distinguish among classes, types, and sizes of
sources" when promulgating CAA section 112 standards as further evidence for constraining the
EPA's ability to adopt different standards applicable during SSM events. In contrast, the
commenter asserted that "Congress has expressed no such concern about EPA's efforts to
implement section 111" of the CAA despite revisions to CAA section 111 in 1977 and 1990.
Therefore, the commenter asserted, Congress has "effectively ratified EPA's longstanding
approach to SSM under the NSPS program," which includes the exemption for flares from the
H2S concentration requirements during SSM events.
Commenter 0315 also asserted that, regardless of the above and despite the similar nature
of the provisions in CAA sections 111 and 112, the EPA has the discretion to implement them
differently "under the markedly differently context of the NSPS program v. the MACT
program." S qq Environmental Defense v. Duke Energy Corp., 549 U.S. 561, 575-576 (2007). For
example, the commenter asserted that the word "continuous" as used in the NSPS program could
be interpreted and applied differently, as acknowledged by the Court in National Lime Ass 'n v.
EPA, 627 F.2d 416, 434 (D.C. Cir. 1980) (deferring to agency regarding the effect of "the
perplexing implications of Congress' new requirement of systems of continuous emission
reduction" on the agency's longstanding "regulations permitting flexibility to account for
startups, shutdowns, and malfunctions"). The commenter urged the EPA to exercise this
95
-------
discretion and "reassert the many practical, technical and economic factors" that justify
promulgating separate standards for SSM events in the NSPS program.
Third, Commenter 0315 asserted that requiring flares to meet the H2S concentration
limits during SSM events does not represent BSER for this time period. According to the
commenter, "startup and shutdown gases are intermittent streams that cannot be cost-effectively
treated for sulfur removal because of their infrequent occurrence, their scattered points of
generation and their variability." Therefore, for all of the above reasons, Commenter 0315
asserted that exempting a flare from the H2S concentration limits when combusting certain gases
generated during SSM events is lawful under CAA section 111.
Alternatively, Commenter 0315 stated that if a standard must apply during SSM events,
the flare work practice standards are appropriate in lieu of the H2S concentration limit.
Response: Regardless of whether or how the Sierra Club decision under CAA
section 112 applies to NSPS promulgated under CAA section 111, we are promulgating final
amendments for flares that include a suite of standards that apply at all times and are aimed at
reducing S02 emissions from flares. As described previously, this suite of standards requires
refineries to: (1) develop and implement a FMP; (2) conduct RCA and take corrective action
when waste gas sent to the flare exceeds a flow rate of 500,000 scf above the baseline;
(3) conduct RCA and take corrective action when S02 emissions exceed 500 lb in a 24-hour
period; and (4) optimize management of the fuel gas by limiting the short-term concentration of
H2S to 162 ppmv during normal operating conditions. Additionally, refineries must install and
operate monitors for measuring sulfur and flow at the inlet of all of their flares. Together, these
requirements provide CAA section 111-compliant standards that collectively cover all operating
conditions of the flare.
As the commenter notes, CAA section 111(h)(1) allows the EPA to promulgate a design,
equipment, work practice or operational standard or "combination thereof," when "it is not
feasible to prescribe or enforce a standard of performance" which reflects BSER for the
particular affected source. CAA section 111(h)(2) defines the phrase "not feasible to prescribe or
enforce a standard of performance" as follows:
[A]ny situation in which the Administrator determines that (A) a pollutant or
pollutants cannot be emitted through a conveyance designed and constructed to
emit or capture such pollutant, or that any requirement for, or use of, such a
96
-------
conveyance would be inconsistent with any Federal, State, or local law, or (B) the
application of measurement methodology to a particular class of sources is not
practicable due to technological or economic limitations.
We have determined that flares meet the criteria set forth in CAA section 111(h)(2)(A)
because emissions from a flare do not occur "through a conveyance designed and constructed to
emit or capture such pollutant." Gases are conveyed to the flare for destruction and combustion
products, such as S02, are not created until combustion occurs, which happens in the flame that
burns outside of the flare tip. In other words, the SO2, NOx, PM, CO, VOC and other pollutants
generated from burning the gases are only created once the gases pass through the flare and come
into contact with the flame burning on the outside of the flare. The flare itself is not a
"conveyance" that is "emitting" or "capturing" these pollutants; instead, it is a structure designed
to combust the gases in the open air. Thus, setting a standard of performance for SO2 (and other
pollutants) is not "feasible," allowing the EPA to instead promulgate standards under CAA
section 111(h), which will collectively limit emissions from the flare.
The EPA previously promulgated a standard of performance for S02 emissions for fuel
gas combustion devices which also applied to flares. 39 FR 9308, 9315 (March 8, 1974). The
standard is expressed as an H2S concentration limit because it was developed as an alternative to
measuring the SO2 concentration in the stack gases exiting fuel gas combustion devices other
than flares {i.e., boilers and process heaters). That approach is appropriate for fuel gas
combustion devices other than flares because measuring the H2S in the fuel gas combusted in
those devices is directly indicative of the SO2 emitted from the exhaust stacks of those other
devices. As explained in section III of the preamble to the final amendments, we are, for the first
time, designating flares as their own affected facility. As such, in finalizing these amendments
for flares, we considered whether we could also apply a standard of performance for SO2
emissions, expressed as an H2S concentration limit or a total sulfur limit at the inlet to the flare.
However, as explained above, flares are substantially different from other fuel gas combustion
devices so that this approach is not workable for flares. For example, S02 emissions from a flare
are dependent on many factors, including the flow rates of all gases sent to the flare, the total
sulfur content of all gases sent to the flare and the combustion efficiency at the flare. Each of
these factors is also dependent on many variables. For example, combustion efficiency at the
flare is dependent upon the flammability of the gases entering the flare, the turbulence at the
97
-------
flare,15 the wind speed and wind direction and the presence of other pollutants in the gases that
can react with the sulfur to form sulfur-containing pollutants other than SO2. Since so many
factors affect the potential formation of S02 emissions outside the flare tip, we realized that we
could not properly derive an H2S concentration limit or a total sulfur limit at the flare inlet that
would directly correlate with those SO2 emissions. Thus, we determined that we cannot set a
standard of performance for SO2 emissions at the flare.
However, we still recognize that reducing the amount of sulfur that is sent to a flare will
reduce the SO2 emissions at the flare. Even with the uncertainty described above, we understand
the importance of refineries managing the fuel gas sent to their flares in a way that minimizes the
sulfur content so as to ultimately minimize the SO2 emissions. Rather than eliminate the H2S
concentration limit altogether, we are instead requiring under CAA section 111(h) that refineries
limit the short-term concentration of H2S to 162 ppmv in the fuel gas sent to flares during normal
operating conditions. Refineries rely on various methods for optimizing the management of fuel
gas, including the use of amine treatment and FGR systems. Amine treatment removes the H2S
from the flare gas that generates the pollutants before the gas is sent to the flare. Flare gas
recovery systems remove the flare gas altogether and instead treat this gas in a fuel gas treatment
system to be used elsewhere as fuel gas in the refinery. Requiring refineries to meet this
concentration limit at the flare ensures that the fuel gas has been adequately treated and managed
such that it can be used as fuel gas in the fuel gas system elsewhere in the refinery. We are not
requiring refineries to meet this limit during other periods of operation because FGR systems that
capture gases prior to amine treatment can be quickly overwhelmed during high fuel gas flows.
Thus, requiring that flares meet this H2S concentration limit during periods when high fuel gas
flows would likely overwhelm these FGR systems would not fully address the circumstances
refineries face in managing these high flow periods. Designing FGR systems to capture the full
range of gas flows to the flare would not only require the ability to predict the full range of gas
flows in the flare headers, but also would require refiners to install recovery compressors in a
staged fashion such that all events causing high gas flows could be captured and managed,
neither of which are practical. Therefore, promulgating flare requirements that include the H2S
fuel gas concentration limit during normal operating conditions, coupled with requirements for
15 Turbulence is needed to insure good mixing at the flare, but is affected by whether the flare is assisted with air or
steam or non-assisted.
98
-------
refineries to develop and implement a FMP and conduct RCA and take corrective action when
waste gas sent to the flare exceeds a flow rate of 500,000 scf above the baseline or 500 lb of SO2
in a 24-hour period, recognizes these unique circumstances while still requiring the refinery to
take all reasonable measures for reducing or eliminating the flow and sulfur content of gases
being sent to the flares.
We are aware that numeric SO2 emission limits for flares have been established under
state law and in Federal Implementation Plan (FIP) regulatory requirements. Those source-
specific circumstances differ markedly from this nationally applicable rulemaking, necessitating
different decisions in two very different circumstances. For example, the EPA's SO2 FIP for the
Billings/Laurel Montana area includes a SO2 emission limit of 150 lb of SO2 per 3 hours for four
sources that apply to the flares at all times. See 40 CFR 52.1392(d)(2)(i), (e)(2)(i), (f)(2)(i) and
(g)(2)(i). These source-specific limits were appropriately based on dispersion modeling in the
Billings/Laurel area to determine what was needed to meet national ambient air quality standards
(NAAQS) for SO2 in the Billings/Laurel area. In contrast, the nationally applicable standards and
requirements we are promulgating in this rule must represent the BSER achievable for an entire
industry sector scattered across the entire country. This requires that we consider costs and other
non-air quality factors that affect all petroleum refineries nationwide in making that decision and
not just as applied to a particular group of sources in a particular location.
Additionally, those four sources subject to the Billings/Laurel FIP demonstrate
compliance with the 150 lb S02/3-hours emission limit by measuring the total sulfur
concentration and volumetric flow rate of the gas stream at the inlet to the flare. See 40 CFR
52.1392(d)(2)(ii), (e)(2)(ii), (f)(2)(ii), (g)(2)(ii) and (h). Since the FIP must include emissions
limits that insure attainment and maintenance of the NAAQS in the Billings/Laurel area, it was
appropriate, in setting the standards for the Billings/Laurel FIP, to conservatively assume that
100 percent of the sulfur in the gases discharged to the flare is converted to SO2, and based on
this conversion, set the numeric limit as a value that is not to be exceeded. However, that same
assumption is not appropriate when setting national standards for flares. Instead, we must
consider the many factors affecting the formation of S02 at the flare tip and how these factors
affect how much of the sulfur in the gases sent into the flare actually converts to SO2. Therefore,
although setting such source-specific limits was appropriate to satisfy what the modeling showed
99
-------
was necessary to meet the SO2 NAAQS in the Billings/Laurel area, a different analysis and
standard is appropriate for a national rulemaking.
Therefore, for the reasons discussed above, the EPA is finalizing this collective set of
CAA section 11 l(h)-compliant standards for flares, based on our interpretation of CAA
section 111(h) as it applies to flares.
Comment: Commenter 0299 requested that the EPA provide provisions for fuel gas
compressor preventative maintenance to reduce compressor outages as compared to running the
compressor until it fails. Commenter 0311 stated that FGR system outages may occur for safety
reasons (e.g., if oxygen gets in the flare gas) and requested clarification that proposed
40 CFR 60.103a(a)(6) specifically acknowledge this possibility by inclusion of "flare gas
recovery system outages (e.g., for maintenance or for safety reasons)."
Response: We expect that refineries will properly maintain the compressors, and RCA
will need to be conducted for FGR outages and systems put in place to eliminate the recurrence
of such events. Maintenance of the FGR does not constitute a process SSM event that excludes
the fuel gas from the H2S concentration limits. However, inclusion of procedures in the FMP to
minimize the volume of gas flared during FGR system maintenance is reasonable and we have
added this requirement to 40 CFR 60.103a(a)(7) as requested by the commenter.
100
-------
5.0 PROCESS HEATERS
5.1 Emissions Limits
5.1.1 Equivalence of Emissions Limit Units
Comment: Commenters 0296, 0305 and 0311 stated that the limits for the concentration
and heating value emissions limits should be the same numerically; i.e., the process heat rate
limit numerically equivalent to the 40 ppmv concentration limit is 0.04 lb/MMBtu and the
process heat rate limit numerically equivalent to the 60 ppmv concentration limit is 0.06
lb/MMBtu.
Response: In August 2008, Industry Petitioners provided the EPA with suggestions for
revising the process heater standards (Docket Item No. EPA-HQ-OAR-2007-0011-0257). One of
their recommendations was to include emissions limits based on heating value (lb/MMBtu) to
account for hydrogen content variations in the fuel gas. They suggested that, on an annual basis,
most natural draft process heaters could meet 0.035 lb/MMBtu and all other process heaters
could meet 0.055 lb/MMBtu. We evaluated these suggested emissions limits and determined that
they were reasonably equivalent to the concentration-based limits we were proposing. We also
requested comment on their use and their equivalency, as described in the preamble to the
proposed amendments (see 73 FR 78527). Industry commenters now assert that the emissions
limit numerically equivalent to the 40 ppmv concentration limit is 0.040 lb/MMBtu and the
emissions limit numerically equivalent to the 60 ppmv concentration limit is 0.060 lb/MMBtu.
We note that, as discussed in the preamble to the proposed amendments, the exact
conversion from ppmv to lb/MMBtu depends on the hydrogen content of the fuel gas. However,
our calculations generally support the more direct numerical conversion suggested by
commenters over the typical range of hydrogen concentrations expected in the fuel gas (see
Revised NOy Impact Estimates for Process Heaters, in Docket ID No. EPA-HQ-OAR-2007-
0011). Therefore, we are finalizing heating value-based emissions limits of 0.040 lb/MMBtu and
0.060 lb/MMBtu for natural draft process heaters and forced draft process heaters, respectively,
based on direct numerical conversions from the concentration-based emissions limits.
101
-------
We are also clarifying that the owner or operator must demonstrate that the process heater
is in compliance with either the applicable concentration-based or heating value-based NOx
limit. The heating value-based NOx emission rate is calculated using the 02-based F factor,
which is the ratio of combustion gas volume to heat input. Ongoing compliance with this NOx
emissions limit is determined using a NOx CEMS and at least daily sampling of fuel gas heat
content or composition to calculate a daily average heating value-based emissions rate, which is
subsequently used to determine the 30-day average.
Specifically, if the F factor is determined at least daily, the owner or operator may elect to
calculate both a 30-day rolling average NOx concentration (ppmv, dry basis, corrected to
0-percent excess air) and a 30-day rolling average NOx emission factor (in lb/MMBtu) and
demonstrate that the process heater is in compliance with either one of these limits. For most fuel
gas systems, the alternative emissions limits are expected to be identical; however, there may be
instances where a process heater may be complying with one of the emissions limits and not the
other. For example, a process heater combusting fuel gas with very high hydrogen content may
have an average NOx concentration above the 60 ppmv limit, but below the 0.060 lb/MMBtu
limit, largely due to the concentration limit being determined on a dry basis (and understanding
that the combustion of hydrogen produces only water and not carbon dioxide). Provided that the
appropriate monitoring is conducted, an affected source would only be out of compliance if it
exceeds both the concentration-based limit and the heating value-based limit at the same time.
However, to have the option to determine compliance with the alternative heating value-based
emissions limit, the refinery owner or operator must, at least daily, determine the F factor (dry
basis) for the fuel gas according to the monitoring provisions in 40 CFR 60.107a(d). If the F
factor is not determined at least daily, the heating value-based alternative cannot be used.
Generally, fuel gas heating value is important to the overall operation of refinery boilers and
process heaters; as such, refiners maintain their fuel gas within an operating range that they need
to fire these sources, often by mixing with natural gas, etc., so we anticipate that most, if not all,
refiners will already have this information available on a daily basis.
5.1.2 Limits for New, Modified and Reconstructed Forced Draft Process Heaters
Comment: Commenters 0296, 0311 and 0314 suggested that newly constructed forced
draft process heaters should have the same 60 ppmv emissions limit as modified and
102
-------
reconstructed forced draft process heaters. Commenter 0296 stated that the emissions limit for
newly constructed forced draft process heaters was based on only two data points and that it does
not fully consider the operating range of newly constructed units. The commenter included a
John Zink white paper and asserted that this white paper shows that newly constructed forced
draft process heaters (particularly those that use air preheat) cannot meet the 40 ppmv NOx
emissions limit. Commenter 0314 agreed, noting that they designed a new forced draft process
heater with air preheat to meet 50 ppmv NOx, and the "heater vendor and burner supplier have
determined that they cannot get appreciably lower emissions than the current design achieves."
The commenter noted that since burner modifications alone will not be able to achieve 40 ppmv,
this heater would need SCR to meet the 40 ppmv NOx limit, and they provided costs and
emission reductions to support the EPA's finding that SCR is not cost effective. Commenter
0314 also suggested that the EPA could return to the proposed limit of 80 ppmv NOx.
Similarly, Commenter 0297 supported higher emissions limits for process heaters with air
preheat. The commenter noted that meeting a 40 ppmv limit for new forced draft process heaters
with combustion air preheat "and allowing for a comfortable margin to ensure compliance is
extremely difficult, if not impossible, without the use of add-on controls." The commenter also
suggested that the mass emissions of NOx from units with air preheat would not be higher than
units without air preheat due to the reduced flue gas flow rate and recommended the 60 ppmv
NOx limit be extended to new forced draft process heaters equipped with combustion air preheat.
Response: The commenters provided only limited and theoretical data to support their
argument that new forced draft process heaters cannot meet the 40 ppmv (or 0.040 lb/MMBtu)
NOx emissions limit. Specifically, the John Zink white paper cited by Commenter 0296
(submitted as an attachment to Docket Item No. EPA-HQ-OAR-2007-0011-0296) stated only
that the 40 ppmv emissions limit could not be "guaranteed" for a new forced draft process heater,
based on the design conditions, which included air preheat. Actual NOx performance data for
that commenter's new forced draft process heaters are not available, as those particular process
heaters are not yet operational. As such, the actual performance of these forced draft process
heaters is still in question. However, we acknowledge that we only have data for one new forced
draft process heater without air preheat that is currently operating that could meet a 40 ppmv
NOx emissions limit on a 365-day average. We conducted additional data evaluations to
determine appropriate limits and averaging times for all process heaters at normal operating
103
-------
conditions while considering this and other public comments we received. As part of the data
analysis effort, we obtained a year's worth of hourly CEMS data for the new forced draft process
heater without air preheat capable of meeting 40 ppmv on a 365-day average. As discussed later
in this chapter, our analysis of the additional data that we obtained following the proposal
supported revising all NOx emissions limits to be on a 30-day average basis. The data indicate
that the 30-day averages for the new forced draft process heater without air preheat capable of
meeting 40 ppmv on a 365-day average exceeded 40 ppmv 15 percent of the time, but none of
the 30-day averages exceeded 60 ppmv NOx.
Consequently, we are raising the NOx emissions limit (while concurrently reducing the
averaging time) for all new forced draft process heaters to be equivalent to the emissions limit
for modified and reconstructed forced draft process heaters {i.e., 60 ppmv or 0.060 lb/MMBtu
with a 30-day averaging period). Furthermore, based on the information provided by the
commenters, as well as the available performance data for existing forced draft process heaters
with air preheat that have been retrofitted with ultra-low NOx burners, we also conclude that the
60 ppmv (or 0.060 lb/MMBtu) on a 30-day rolling average basis adequately accommodates
forced draft process heaters that use air preheat. Based on our review of CEMS data for new and
retrofitted forced draft process heaters, we conclude that 60 ppmv (or 0.060 lb/MMBtu) on a 30-
day rolling average basis is BSER for new, reconstructed or modified forced draft process
heaters. (For additional details, s qq Revised NOx Impact Estimates for Process Heaters, in
Docket ID No. EPA-HQ-OAR-2007-0011.)
5.1.3 Turndown
Comment: Several commenters addressed the need for the rule to address turndown,
which is a period of time when process heaters are firing below capacity. Commenter 0296 stated
that turndown conditions can extend over a full year, so special provisions are needed for these
turndown conditions. Commenters 0296 and 0305 suggested that a mass emission cap be used as
the sole compliance requirement (lb/MMBtu limit multiplied by the heater's rated capacity).
Alternatively, Commenter 0296 suggested that a mass emission cap be provided for "turndown
operations," which the commenter suggested be defined as operating at 50 percent of the rated
capacity or less. Commenter 0311 requested that the proposed emissions limits only apply to
furnaces that are firing at or above 50 percent of design rates. Days when firing rates are below
104
-------
50 percent of capacity would not be included in the 365-day average and the units would meet a
mass emission rate (pounds per hour [lb/hr]) based on the emissions limit at design capacity,
according to the commenter. Also, according to Commenter 0311, process heaters must often
operate at higher oxygen (O2) levels during turndown, so the proposed maximum O2 operating
limit should not apply when small furnaces (those not required to install continuous emission
monitoring systems [CEMS]) are firing at less than 50 percent of design rates. Commenter 0315
added that limited new data suggest that meeting a NOx concentration limit becomes problematic
at 70 percent of capacity, so the mass limit should apply when a process heater is firing below
70 percent of capacity and the emissions limit should be calculated as the allowed emission
factor for that heater multiplied by the 70-percent firing rate.
Response: In our proposed amendments, we provided a longer averaging time (365-day
average) so that short periods of turn-down would not significantly affect the overall
performance of the unit. However, according to the commenters, the longer averaging time does
not adequately address turndown conditions. Therefore, we re-evaluated the available data,
including our existing data and additional data provided by the industry, to determine the
appropriate emissions limits during different types of operation, including turndown. The
additional data provided by Industry and our evaluation of those data are included in the docket
for the final amendments (Docket ID No. EPA-OAR-HQ-2007-0011). Based on our analysis of
the data (described in greater detail in the next paragraph), we concluded that a 30-day averaging
period is appropriate for the NOx emission limits under most operating scenarios.
Upon examination of all available CEMS data, we determined that, for periods of normal
operation (i.e., firing at 50 percent or more of design capacity), the proposed NOx emissions
limits of 40 and 60 ppmv were not achievable for all process heaters using a 24-hour averaging
period (the averaging period included in the final June 2008 rule). From the available data, short-
term fluctuations in the NOx concentrations of process heaters using ultra-low NOx burners
caused them to exceed a 24-hour average limit somewhat frequently, but a 30-day average
provided adequate time to average out the short-term fluctuations. We note that a few of the
process heaters operated at relatively high excess 02 concentrations at normal conditions (i.e., at
exhaust O2 concentrations of 6 percent or more). These units had periods of excess emissions
above the 30-day average emission limits, but we rejected the performance of these process
heaters as BSER because of the high exhaust O2 concentrations for these units during normal
105
-------
(i.e., non-turndown) firing rates. That is, these process heaters were not being operated optimally
for reducing NOx emissions. Furthermore, when these process heaters were operated at the lower
range of exhaust concentrations for the unit (although generally higher than what would be
considered optimal excess O2 concentrations for reducing NOx emissions), the process heater
could meet the applicable 40 or 60 ppmv emissions limit on a 30-day averaging period. Based on
our review of CEMS data for process heaters with ultra-low NOx burners that operated at excess
02 concentrations less than 6 percent (i.e., operated in a manner consistent with proper low NOx
burner operation), all such process heaters could comply with the final NOx emissions limits on
a 30-day average basis. Consequently, we revised the basic emissions limits to be on a 30-day
average.
As described previously in this section, we conclude that the applicable 40 or 60 ppmv
emissions limit on a 30-day averaging period is achievable for process heaters during periods of
normal operation. Our next step was to evaluate the achievability of the emissions limits during
turndown conditions and alternative approaches for establishing emissions limitations where
necessary. The following paragraphs describe our analysis of the data, including our evaluation
of alternative methods for accommodating turndown conditions and our rationale for providing
the site-specific alternative for extended turndown conditions.
There were very limited CEMS data available for process heaters operating under
turndown conditions (i.e., firing below 50 percent of design capacity). However, two general
trends were observed in the CEMS data that were available: (1) typical exhaust 02
concentrations increase at lower firing rates; and (2) exhaust NOx concentrations (corrected to
0-percent excess O2) increase with increasing O2 concentration (regardless of firing rates). These
data, along with the need to operate the process heater at higher O2 concentrations during low
firing rates to maintain flame stability, suggest that an alternative NOx emissions limit could, in
some instances, be needed to address extended turndown conditions (turndown events lasting a
majority of the 30-day averaging time). As such, we considered alternative compliance options
to address turndown conditions.
One alternative compliance option considered to address turndown was a mass-based
NOx emissions limit that would be equivalent to the mass of NOx emitted from a unit meeting
the 0.040 (or 0.060) lb/MMBtu limit while firing at 50 percent of capacity, as suggested by
commenters. However, for most units for which CEMS data are available, the alternative mass-
106
-------
based emissions limit did not improve the ability of the process heater to meet the emissions
limit. We note that most of the process heaters were able to meet the applicable concentration-
based emissions limit (40/60 ppmv) or the heating value-based (0.040/0.060 lb/MMBtu)
emissions limit during turndown. Therefore, the issue appears to be limited to a few of the
process heaters that must operate at relatively high excess O2 concentrations during turndown
conditions. For these units, the alternative mass-based emissions limit that we were considering
rarely, if ever, provided a means for these units to comply with the performance standard.
We understand that technology providers recommend operating process heaters that are
turned down at higher excess O2 concentrations to improve flame stability and ensure safe
operation of the process heater; however, based on the information provided by the technology
providers, there is still an optimal excess 02 concentration at which flame stability is achieved
while minimizing NOx formation. That is, even when a process heater is operating at less than
50-percent design capacity, excess O2 concentrations should still be controlled to minimize NOx
formation within the safe operating constraints to maintain flame stability. We do not have
specific data on process heaters that are near, but below, the concentration emissions limits when
firing above 50-percent capacity, but cannot meet the concentration limit when firing below
50-percent capacity, so we have no data that show that process heaters operating at less than
50-percent design capacity and controlling excess O2 concentrations cannot meet the emissions
limits. However, we acknowledge that the correlations with firing rates and O2 and/or NOx
concentrations and the need for higher 02 concentrations to maintain flame stability generally
support the commenter's argument that a few marginally compliant process heaters will have
difficulty meeting the basic emissions limit when the unit is turned down. As such, we
acknowledge that there may be periods of turndown in which a process heater is operating as
recommended, but may be unable to meet the concentration or heating value-based emissions
limits in the final rule, especially when the unit is operated at turndown for extended periods
(e.g., for 20 days or more compared to the 30-day averaging time). As the need for an alternative
limit appears to be limited to a few process heaters and the optimal O2 concentration is expected
to vary, based on fuel gas composition, we determined that a site-specific emissions limit was the
best approach to account for these extended turndown conditions. As such, the final rule provides
owners and operators that have a process heater operating in turndown for an extended period of
107
-------
time the option of developing a site-specific emissions limit that would apply to those operating
conditions and requesting approval from the Administrator to use that limit.
For process heaters between 40 and 100 million British thermal units per hour
(MMBtu/hr) capacity that do not install a NOx CEMS, turndown is also expected to be an issue
with respect to achieving the O2 operating limit. As described above, higher O2 concentrations
are generally needed to maintain flame stability at low firing rates. To address potential
turndown compliance issues with the 02 operating limit, we have provided an allowance for
process heater owners or operators to develop an O2 operating curve to provide different O2
operating limits based on the firing rate of the process heater. If a single O2 operating limit is
established, it must be determined when the process heater is being fired at 70 percent or more of
capacity {i.e., far from turndown conditions). For process heaters that routinely operate at less
than 50 percent of design capacity and require additional O2 to maintain flame stability, a
separate O2 operating limit should be established for turndown by conducting a second
performance test while the unit is operating at less than 50 percent of capacity. Additional
performance tests can be conducted to develop 02 operating limits for additional operating
ranges.
5.1.4 Site-Specific Emissions Limit Clarifications
Comment: Commenters 0302, 0305, 0308 and 0311 supported the provision in
40 CFR 60.102a(i) for a site-specific NOx emissions limit for process heaters that cannot be
retrofit with ultra-low NOx burners. Commenters 0308 and 0311 suggested expanding the
applicability of the paragraph to include process heaters for which it may be physically possible
to install ultra-low NOx burners, but other "structural or operability reasons" would prevent the
process heaters from operating properly. Commenter 0308 also noted that island refineries that
have difficulty obtaining natural gas may have fuel gas with higher levels of olefins and propane,
which contribute to higher NOx emissions. The commenter suggested allowing process heaters
that cannot meet the NOx limits for "fuel gas quality reasons" to apply for a site-specific
emissions limit. Commenter 0311 stated that the provisions should also apply to new forced draft
process heaters, especially those using air preheat, that might have trouble meeting a 40 ppmv
limit.
108
-------
Commenter 0311 also stated that the General Provisions provides 180 days to achieve
full process rates, test the unit and submit the results, and this timeframe, rather than the 90-day
time frame proposed, is needed to file the petition. Finally, Commenter 0311 stated that the rule
should specify that the petition's limit applies to the process heater until the petition is acted
upon since the petition is being submitted because the facility cannot meet the emissions limit.
Response: We appreciate the commenters' support of the option to petition for a site-
specific emissions limit. We have clarified the site-specific emissions limit provisions under
40 CFR 60.102a(i) to address most of the commenters' concerns as described in this response.
Regarding process heaters that cannot properly operate ultra-low NOx burners, the
preamble to the proposed rule notes that owners and operators may seek EPA approval for a site-
specific limit for "modified or reconstructed natural draft and forced draft process heaters that
have limited firebox size or other limitations and therefore cannot apply the [BSER] of ultra-low
NOx burners." Our intention was to accommodate process heaters that cannot install and
properly operate combustion modification-based technology due to structural limitations of the
process heater (e.g., size of the heater box, flame impingement on tubes containing process
fluids, spacing of the individual burners). In other words, if ultra-low NOx burners can
physically fit in a process heater but the process heater cannot be operated safely or the ultra-low
NOx burners themselves cannot operate as recommended by the manufacturer, then we consider
there to be physical limitations preventing the use of combustion modification-based technology.
We agree that the proposed language in 40 CFR 60.102a(i) could be interpreted more narrowly
than intended, so we revised the language in 40 CFR 60.102a(i)(l)(i) to specify that "a modified
or reconstructed process heater that lacks sufficient space to accommodate installation and
proper operation of combustion modification-based technology (e.g., ultra-low NOx burners)"
may submit a petition for a site-specific limit.
This provision, while considering "proper operation," was specifically included to
address process heaters with physical (i.e., "structural") constraints in retrofitting a process
heater with ultra-low NOx burners. This provision is not intended to provide unlimited
operational flexibility for the burners. Further, we find that the requested "operability" provision
undermines part of the BSER analysis. As seen by several of the comments received, the process
heater specifications often include maximum radiant section temperatures, maximum preheat
temperatures, maximum excess oxygen content and flexibility in burner design to allow both
109
-------
natural draft or forced draft operation. We recognize that, for existing process heaters, some of
these parameters are fixed, but the question is: should the design and operation of the process
heater be cognizant and reflective of the NOx emissions limits? In developing our BSER
analysis, we included limiting the amount of excess oxygen employed and included costs for
excess oxygen measurement and control. We do not expect that process heaters that are subject
to subpart Ja will operate exactly like process heaters that are not subject to subpart Ja, just with
different burner designs. For example, in reviewing NOx emissions data, we found that several
process heaters were operating with excess oxygen concentrations of 6 to 8 percent while firing
at or near capacity. Discussions with technology vendors, however, suggest that "proper
operation" of a low-NOx burner firing at or near burner capacity would be to operate at excess
oxygen concentrations of 2 to 4 percent. We recognize that there are several factors that
contribute to NOx formation, and it is incumbent on the owner or operator of an affected process
heater to control, to the extent possible, these various parameters so as to minimize NOx
emissions and to comply with the NOx emissions limits. Thus, while we recognize that certain
low-NOx burner retrofits may be infeasible due to structural limitations, we do not want to open
the door for refinery owners and operators to disregard the emissions limits when operating
modified process heaters or when designing new process heaters and then contend that the
emissions limits in 40 CFR 60.102a(g)(2) are not achievable for "operability" reasons.
Therefore, we reject the request to expand the scope of the site-specific emissions limit eligibility
to include "operability" issues per se unconstrained by the "proper operation" of low-NOx
burners.
Similarly, we reject the request to expand the scope of the site-specific emissions limit
eligibility to include fuel gas quality constraints. Based on our analysis of the available data, the
final NOx emissions limits are achievable over a wide range of refinery fuel gas compositions.
Furthermore, we have added alternative NOx emissions limits determined on a heating value
basis, which provides flexibility for times when the composition of refinery fuel gas varies (e.g.,
high hydrogen content).
Regarding new forced draft process heaters, the preamble to the final amendments
explains that we have finalized amendments that do not include new forced draft process heaters
in the list of process heaters eligible to submit a petition for a site-specific emissions limit. As
noted in the preamble to the final amendments, we have revised the NOx emission limit for new
110
-------
forced draft process heaters to be 60 ppmv (on a 30-day rolling average basis), rather than 40
ppmv (on a 365-day average) as provided in the proposed amendments. Based on available
emission data and ultra-low NOx burner performance guarantees, we conclude that all new
forced draft process heaters can meet an emissions limit of 60 ppmv NOx on a 30-day rolling
average basis; therefore, site-specific emissions limits are not warranted for new forced draft
process heaters.
Regarding the specifics of submitting the petition, we agree with the commenter that it is
appropriate to allow an owner or operator up to 180 days after startup to submit a petition for a
site-specific limit. As the commenter noted, 180 days is consistent with the General Provisions,
and it is also consistent with similar performance level determinations for boilers
(40 CFR 60.44b(f)). We also agree with the commenter that clarification is needed regarding
compliance during the time between submittal of the petition and approval by the Administrator.
We have clarified in the final amendments that the owner or operator should comply with the
site-specific limit identified in the submitted petition until the petition is approved by the
Administrator.
5.1.5 Co-Fired Process Heaters
Comment: Commenter 0308 requested clarification regarding the parenthetical qualifier
"(applicable only when the process heater is being co-fired)" in proposed
40 CFR 60.102a(g)(2)(iv)(A). The commenter stated that it is unclear if proposed
40 CFR 60.102a(g)(2)(i)-(iii) apply to these process heaters when they are not co-firing or if only
proposed 40 CFR 60.102a(g)(2)(iv)(B) must be applied. Based on the limitations of co-fired
burner design, the emission limitations in proposed 40 CFR 60.102a(g)(2)(i)-(iii) are
unachievable for co-fired process heaters, and clarifying language is needed to ensure that these
limits are not enforced for co-fired process heaters when only gaseous fuels are used. If the
parenthetical was intended to exclude only those times when 100-percent fuel oil is being
burned, then the parenthetical should be clarified as follows: "(applicable only when the process
heater is being fired in whole or in part with fuel gas)." Commenter 0311 likewise suggested that
the EPA clarify that process heaters with co-firing capabilities (including those with separate
burners dedicated to gas and liquid fuels) would have to meet either 150 ppmv or 0.08
lb/MMBtu when firing only gaseous fuels.
Ill
-------
Commenters 0308 and 0311 requested that the introductory text in proposed
40 CFR 60.102a(g)(2)(iv) be clarified to clearly state that a co-fired process heater must "comply
with either" proposed 40 CFR 60.102a(g)(2)(iv)(A) or (B) as some might interpret the current
language to mean that co-fired process heaters must comply with both paragraphs (A) and (B).
Response: We agree that the language concerning co-fired process heaters is ambiguous,
and we have revised this language to clarify the requirements. The final rule provides two
subcategories of co-fired process heaters: natural draft and forced draft. Each subcategory
includes two options for emissions limits, one on a concentration basis and the other on a mass
basis. The mass-based emissions limit considers the daily average amounts of gas and fuel oil
fired by the heater, but the language no longer includes a qualifier addressing the type of fuel that
must be fired for the limits to apply. In other words, the applicability of the emissions limits is
based on the design of the process heaters and not on the type of fuel being fired at any given
time. As stated previously (see Section 3.2), we do not consider process heaters that have
completely separate gas and oil burners to meet the definition of co-fired process heaters. These
units must demonstrate compliance with the NOx emissions limits when they are fired only with
gaseous fuels. Additionally, the final amendments clarify that the owner or operator of an
affected co-fired process heater may elect to comply with either the heating value-based or the
concentration-based emissions limit, but the owner or operator must comply with the selected
limit at all times. That is, the owner or operator cannot switch back and forth between
compliance methods depending on the fuel burned.
We also note that the final rule does not address process heaters designed to fire only fuel
oil. Because we did not propose requirements specific to oil-fired process heaters, the final rule
does not include amendments addressing oil-fired process heaters. However, we will evaluate
adding requirements for oil-fired process heaters in a future rulemaking.
Comment: Commenter 0308 stated that the emissions limits in proposed
40 CFR 60.102a(g)(2)(iv)(A) and (B) are not achievable ".. .for the conditions that exist at
HOVENSA..and submitted a white paper to support higher emissions limits, which the
commenter recommended be adopted, especially in light of a daily averaging period.
Alternatively, the commenter suggested the EPA defer the promulgation of NOx limits for co-
fired process heaters due to the lack of adequate data to "adequately demonstrate" an achievable
112
-------
emissions limit under section 111 of the CAA. Commenter 0311 stated that the 0.27 lb/MMBtu
limit is not achievable when firing residual fuel oil.
Commenters 0308 and 0311 recommended that the emissions limit alternative in
proposed 40 CFR 60.102a(g)(2)(iv)(B) be determined on a 365-day average as these process
heaters will have the same issues with respect to turndown and variability as other process
heaters. Commenter 0311 also stated that the notation "ENox,hour" in proposed Equation 3 was
confusing and that the "hour" subscript should be dropped or changed to "daily."
Response: The final June 2008 rule included only one emissions limit for all co-fired
process heaters, and Industry Petitioners asserted that differences in the configuration and
operation of different types of process heaters warranted different emissions limits. The proposed
amendments introduced two specific emissions limits for co-fired process heaters, one based on
vendor guarantees for the burners and one based on an average NOx concentration for a
combination of fuel gas and fuel oil. We note that, for purposes of this rule, a co-fired process
heater is defined as a process heater that employs burners that are designed to be supplied by
both gaseous and liquid fuels. In other words, co-fired process heaters are designed to routinely
fire both oil and gas in the same burner. These do not include burners that are designed to burn
gas, but have supplemental oil firing capability that is not routinely used {i.e., emergency oil
back-up).
To respond to the comments requesting higher emissions limits for co-fired process
heaters, we reviewed the white paper provided by one commenter (submitted as an attachment to
Docket Item No. EPA-HQ-OAR-2007-0011-0308), as well as additional burner emissions test
data provided by another commenter16 (conducted under well-controlled conditions using best
available ultra-low NOx burner technologies at the manufacturer's testing facility). This
information indicates that, for co-fired natural draft process heaters, a daily average emissions
limit calculated based on a limit of 0.06 lb/MMBtu for the gas portion of the firing and
0.35 lb/MMBtu for the oil portion of the firing is achievable. Similarly, the information indicates
that, for co-fired forced draft process heaters, a daily average emissions limit calculated based on
a limit of 0.11 lb/MMBtu for the gas portion of the firing and 0.40 lb/MMBtu for the oil portion
of the firing is achievable. As noted above, these values are based on burner performance tests,
16 The commenter providing this data asserted that it is confidential business information (CBI). We will follow our
CBI regulations in 40 CFR part 2 in handling this data. The data has been placed in the docket, but is not publicly
available.
113
-------
which are considered a better source of information than the vendor guarantees that were relied
upon to develop the proposed emissions limit. Therefore, we are revising the NOx emissions
limits for co-fired process heaters to those described above. We note that we have revised the
concentration-based NOx emissions limits to be on a 30-day average basis (same as the limits for
gas-fired process heaters). We have also revised the nomenclature of the daily average emissions
limit in Equations 3 and 4 (proposed Equation 3) to be clear that we intend the limit to be
determined on a daily basis rather than on an hourly basis.
We also note that the burner performance tests were conducted in a controlled
environment at the burner manufacturer's full-scale facilities. While it is incumbent on the owner
or operator of an affected process heater to control certain operating parameters, such as excess
02 concentrations, to the extent possible, we recognize that the performance limits in the final
amendments are based on limited data, none of which are direct test data for a co-fired process
heater operated at a petroleum refinery. We conclude that the low-NOx burner technologies
exist, are demonstrated and are cost effective for co-fired process heaters and they are, therefore,
BSER for co-fired process heaters. However, as the performance limits are based on limited
operational data, we also conclude that it is reasonable to provide an alternative, site-specific
limit in the event that factors outside the influence of the burner design and operation (such as
nitrogen content in the fuel oil) suggests the emission limits in the final rule are inappropriate for
a specific application. Consequently, co-fired process heaters that cannot meet the limits
specified above can request approval for a site-specific emissions limit, as allowed above, for
process heaters that operate for extended periods under turndown.
5.2 Monitoring of Process Heaters
Comment: Commenters 0305 and 0311 noted that CEMS compliance requirements in 40
CFR part 75 are similar to those in 40 CFR part 60, but not identical. The commenters requested
an exemption from the 40 CFR part 60 monitoring and QA requirements for any process heater
or boiler subject to and complying with 40 CFR part 75 monitoring required by Acid Rain rules,
a NOx State Implementation Plan (SIP) call, or other rules. Alternatively, the commenters
requested that the EPA specifically provide 40 CFR part 75 monitoring and QA provisions to be
compliant with 40 CFR part 60, subpart Ja.
114
-------
Response: We reviewed the QA/QC requirements in 40 CFR part 75. It appears that the
40 CFR part 75 NOx emissions limits (provided predominantly for coal-fired units) are much
higher than those provided in these final amendments (which are predominantly for gas-fired
process heaters). The accuracy requirement for the RATA on NOx is ±0.015 or 0.020 lb/MMBtu,
depending on the RATA frequency. We find that these accuracies are not appropriate given the
NOx emissions limits that are in the final rule. Consequently, we do not consider the 40 CFR
part 75 QA/QC requirements sufficient for compliance with the final rule and we are not
amending the rule to allow 40 CFR part 75 monitoring and QA provisions.
Comment: Commenter 0311 asserted that daily determination of fuel oil heating value to
comply with the co-fired process heater NOx emissions limit is excessive and unnecessary if the
fuel oil is fed from a tank because the heating value does not vary that much. The commenter
requested that the EPA allow the 40 CFR part 75 sampling protocols (Sections 2.2.3 and 2.2.4.1
through 2.2.4.3 of Appendix D) instead of daily sampling. The commenter stated that
40 CFR 60.4370 (40 CFR part 60, subpart KKKK) allows these protocols for sulfur sampling of
liquid fuels to turbines.
Response: We reviewed the referenced sections in 40 CFR part 75. While we do not
understand the reference to Section 2.2.3, which seems to indicate daily sampling, we recognize
that heating value of fuel oil fed from a common storage tank is not expected to vary unless
additional shipments are received. As many of the alternatives in the cited sections are sulfur-
centric, we have provided similar alternatives in 40 CFR 60.107a(d)(7)(viii) to reduce the burden
associated with daily sampling when daily sampling will not improve the accuracy of the
measurements.
Comment: Commenter 0307 stated that the EPA should require CEMS for all process
heaters to demonstrate compliance with the NOx emissions limit. The proposed amendments
allow the option of biennial source testing for certain process heaters with low-NOx or ultra-low
NOx burners; however, commenters assert that biennial source testing is not sufficient for
ensuring that these process heaters are meeting their NOx emissions limits. The commenter
stated that section 504(c) of the CAA requires that Title V permits include monitoring
requirements that "assure compliance with the permit terms and conditions," and in Sierra Club
v. USEPA,17 the D.C. Circuit held that the EPA and state permitting agencies must include
17 536 F.3d 673 (D.C. Cir. 2008).
115
-------
additional monitoring requirements in Title V permits if the existing requirements are
insufficient to ensure compliance with an emissions limit. The commenter stated that biennial
source testing for process heaters is not sufficient to demonstrate compliance; instead, the
commenter asserted that CEMS are needed for demonstrating that all process heaters are in
continuous compliance with the NOx emissions limit. The commenter also noted that if the EPA
relies on CEMS data provided by industry to weaken the NOx emissions limits, then the EPA
should at least require CEMS to ensure industry is complying with that weaker limit.
Commenter 0315 countered that subpart Ja does require CEMS for process heaters firing
more than 100 MMBtu/hr. The commenter also noted that Commenter 0307 did not provide data
supporting their assertion that biennial source testing is insufficient for small process heaters, and
the EPA has discretion in determining what monitoring is adequate. Commenter 0315 noted that
the EPA originally proposed CEMS on all process heaters, but based on public comments
received through the subpart Ja rulemaking process, the EPA revised the evaluation and
determined that CEMS were not justified for small process heaters (see Docket Item No. EPA-
HQ-OAR-2007-0011-0222 and the third attachment to Docket Item No. EPA-HQ-OAR-2007-
0011-0222). The commenter noted that the EPA's determination was reasonable and consistent
with the agency's longstanding approach for assuring continuous compliance, and the
commenter cited several sections of 40 CFR part 64 for support.
Commenter 0311 supported the more flexible span range for NOx and the ability for
process heaters between 40 and 100 MMBtu/hr to use either a CEMS or biennial test.
Response: We consider the monitoring requirements in the final rule sufficient for all
process heaters to demonstrate continuous compliance with their applicable emissions limits. The
final rule requires that owners or operators install CEMS for demonstrating compliance with the
NOx emissions limits for all process heaters (40 CFR 60.107a(c) and (d)). As an alternative to
this general requirement, the rule also provides that small process heaters—those with rated heat
capacities of less than 100 MMBtu/hr—may instead conduct biennial testing for demonstrating
compliance with their NOx emissions limits. Commenter 0307 did not make this distinction in
their comments. Thus, we want to reiterate that only smaller process heaters (those with rated
heat capacities of less than 100 MMBtu/hr) may choose the option of biennial tests. This
provision was included after considering available data on these small process heaters as well as
the cost of applying a CEMS for these smaller process heaters. For the available data
116
-------
summarized in the process heater memorandum, only one of 29 process heaters with data
obtained via CEMS has less than 100 MMBtu/hr capacity. For the data from small process
heaters summarized in the process heater memorandum, 9 of 10 data points are a result of one-
time source tests. Regarding the monitoring costs for small process heaters, we reviewed the
estimated costs supporting the process heater impacts memorandum (see the third attachment to
Docket Item No. EPA-HQ-OAR-2007-0011-0222). The cost of a CEMS is fixed, regardless of
the process heater size, so it is a larger portion of the cost for smaller process heaters, and the
effect is to make the control option less cost effective for that process heater. For modified
forced draft process heaters, for example, the CEMS contributes an additional 50 percent to the
annualized cost of the burner retrofits. The cost effectiveness of ultra-low NOx burner retrofits
for a 200 MMBtu/hr process heater with and without CEMS is approximately $1,000 per ton of
NOx reduced and $1,500 per ton of NOx reduced, respectively. In contrast, the cost effectiveness
for a modified 70 MMBtu/hr forced draft process heater (the mid-point of the range of small
process heaters) at a medium-sized refinery with and without a CEMS is about $3,600 and
$2,300 per ton ofNOx, reduced, respectively.
Additionally, this option for small process heaters is not without its own continuous
monitoring requirements. This option is provided specifically for units that use low-NOx or
ultra-low NOx burner technologies because the performance of these control systems is generally
quite stable. As noted previously, one important variable that affects the performance of these
burners is the level of excess oxygen supplied. To verify continuous performance that is
indicative of the conditions present during the biennial source tests, the owner or operator must
establish a maximum excess oxygen operating limit or operating curve during the performance
test and monitor the oxygen content in the exhaust on a continuous basis. What was inadvertently
left out of the requirements for this option is the inclusion of this CPMS operating limit in the list
of conditions that define reportable excess emissions. We are adding that requirement in these
final amendments at 40 CFR 60.107a(h)(5); the effect of this amendment is that the owner or
operator of a small process heater that elects to use biennial testing for compliance must report
any daily average oxygen concentration that exceeds the maximum excess oxygen operating
limit or operating curve as an exceedance of the NOx emissions limit, just as an owner or
operator electing to comply using a CEMS must report any 30-day period during which the
average concentration ofNOx exceeds the applicable emissions limit. Thus, considering all of
117
-------
the above, the monitoring requirements for process heaters—both large and small—are
appropriate and sufficient for an owner or operator to demonstrate continuous compliance with
the applicable NOx emissions limits.
5.3 Other Clarifications
Comment: Commenter 0311 requested that the term "other means" in
40 CFR 60.102a(i)(2) be replaced with "other combustion modification based technology" to
clarify that evaluation or use of post-combustion technologies, such as SCR, are not required.
Response: We agree that the language in 40 CFR 60.102a(i)(2), as worded, could be
interpreted to mean that the petitioner would have to evaluate SCR technology and provide
evidence as to why the emissions limit could not be met using an add-on control technology.
This was not our intent. We meant for the evaluation to include potential combustion
modifications. We have clarified the language in 40 CFR 60.102a(i)(2)(ii) to require that the
owner or operator "demonstrate that ultra-low NOx burners, flue gas recirculation, control of
excess air or other combustion modification-based technology (including combinations of these
combustion modification-based technologies) cannot be used to meet the applicable emissions
limit."
118
-------
6.0 COMPLIANCE REQUIREMENTS AND GENERAL
MONITORING, RECORDKEEPING AND REPORTING
6.1 Rolling Averages
Comment: Commenter 0309 requested clarification regarding how to calculate rolling
averages during periods when a CMS (including CEMS) exceeds the maximum prescribed span
range in the rule. The commenter suggested that the EPA's policy was that the maximum
prescribed span range provided in the rule be used. For example, the commenter noted, for
periods when the fuel gas monitor exceeds 300 ppmv H2S, 300 ppmv should be used in
calculating the average for that time period, regardless of the monitor's reading.
Response: We do not agree that it is the EPA's policy, at least within 40 CFR part 60, to
use the maximum prescribed span range when the monitor's reading exceeds the maximum
prescribed span when calculating rolling averages. It is our policy that sources with CMS data
outside the certified span should report the CMS "out of control" and declare their compliance
status as other than continuous. Additionally, for the purposes of calculating rolling averages, the
direct monitor reading should be used for all periods that the CMS is operational, except for
periods when the CMS data are invalidated according to 40 CFR 60.13((h)(2)(iv). The CMS
must be operational for all periods except for system breakdowns, repairs, calibration checks and
zero and span adjustments as specified in 40 CFR 60.13(e).
Comment: Commenter 0311 requested clarification on how the "365 successive calendar
day rolling average" in 40 CFR 60.102a(g)(l)(ii) and the "rolling 365 successive operating day
average" in 40 CFR 60.102a(g)(2) is to be calculated when: a) the unit is not operating; b) when
the monitor is unavailable (or "out of control" according to Appendix F); and c) turndown occurs
(if a separate emissions limit does not apply to these periods as recommended by the
commenter). According to the commenter, a fixed 365 calendar day period is easier to calculate
than having to determine which days are to be included and having a different and changing
number of calendar days in the calculation of the average.
Response: The final rule includes a requirement in 40 CFR 60.102a(g)(l)(ii) to limit the
long-term H2S concentration of fuel gas "determined daily on a 365 successive calendar day
119
-------
average basis." The final rule also includes NOx limits for process heaters determined on a "30-
day rolling average basis" in 40 CFR 60.102a(g)(2). It is our intention that each 365-day average
would be based on 365 calendar days. Similarly, it is our intention that each 30-day average
would be based on 30 calendar days, regardless of whether the process heater is operating at full
capacity or in turndown. Only the hours that the fuel gas combustion device is operating should
be used when calculating the daily average for a given calendar day. If the fuel gas combustion
device did not operate at all during a given calendar day, that day would have a null value and
would not be counted in the 30-day or 365-day average. As such, the 30-day rolling average may
only contain 20 days of values if the process heater was not operating for 10 days during this 30-
day period. Similarly, CMS "out of control" periods would be considered null values for the
purposes of calculating the 30-day or 365-day average. See also the previous comment and
response regarding CMS "out of control" periods.
6.2 Excess Emissions
Comment: Commenter 0309 requested clarification of "all 12-hour periods" for excess
emissions reporting because the note in 40 CFR 60.106a(b) conflicts with guidance provided by
EPA Region 3 on April 14, 2000. If the April 2000 guidance is the preferred methodology, then
the EPA should delete the note and 3-hour standards should be expressed as 180-minute
averages.
Response: First, we note that the guidance provided by EPA Region 3 is based on the
language as it was in subpart J prior to amendments, where no additional guidance in the rule
was provided on how to calculate the 12-hour averages. As such, Region 3's interpretation is
justified based on the language used in subpart J. However, requiring 12-hour averages to be
calculated every minute greatly increases the computational and data storage requirements for
the refinery with little to no environmental benefit. It is extremely unlikely that a 12-hour
average excursion will be "seen" when evaluating "by minute" 12-hour averages and not "seen"
when calculating hourly averages. While the hourly averages should be calculated using the
continuous data {i.e., by minute), there is no need to store these "by minute" data for 12 hours,
increasing the storage needs by an order of magnitude. We note that the sections describing
excess emissions {e.g., 40 CFR 60.106a(b), 40 CFR 60.107a(h)) already specify that the 3-hour
or 12-hour periods are to be determined by first calculating hourly averages and then taking the
120
-------
arithmetic average of the appropriate number of contiguous hourly averages, and we do not find
it necessary to amend the rule to provide further guidance.
6.3 Monitoring Methods
Comment: Commenter 0311 stated that the reference to Gas Processors Association
(GPA) Standard 2172-96 in 40 CFR 60.17(m)(2) and 40 CFR 60.107a(d)(7)(v) should be
updated to GPA 2172-09 because the standard has been updated.
Response: We agree, and both 40 CFR 60.17(m)(2) and 40 CFR 60.107a(d)(7)(viii) have
been updated to reference the latest version of the standard.
6.4 Reporting Requirements
Comment: Commenter 0311 suggested that the reporting requirement in
40 CFR 60.108a(c)(6) for RCA be revised to quarterly rather than per event.
Response: First, 40 CFR 60.108a(c)(6) includes recordkeeping requirements and should
not be confused with reporting requirements. Section 60.108a(c)(6) requires refinery owners and
operators to keep records of all the RCA info, but it does not specifically require submission of
reports. We also note that 40 CFR 60.103a(d) and (e) use the term "record" or "record and
explain." While the latter suggests reporting, neither of these specifically requires submission of
the record. The actual reporting requirements are summarized in 40 CFR 60.108a(d), which
requires refinery owners and operators to submit excess emission reports on the schedule
required by 40 CFR 60.7(c) and describes the information to be included in the report. Section
60.7(c) requires excess emission reports to be submitted semi-annually: "Each owner or operator
required to install a continuous monitoring device shall submit excess emissions and monitoring
systems performance report (excess emissions are defined in applicable subparts).. .to the
Administrator semiannually... .All reports shall be postmarked by the 30th day following the end
of each six-month period." Therefore, while the report must detail each event that occurred
during the reporting period, these reports only need to be submitted on a semi-annual basis.
6.5 Impacts on Small Refineries
Comment: Commenter 0302 stated that much of the monitoring, recordkeeping and
reporting requirements are independent of the size of the refinery, and small refineries, especially
121
-------
those not subject to a consent decree, will have a greater burden in terms of "cost per gallon"
than larger refineries.
Response: We expect smaller refineries will have fewer flares and fewer process heaters
above 40 MMBtu/hr and/or 100 MMBtu/hr thresholds that require monitoring. The alternative
monitoring for process heaters less than 100 MMBtu was specifically included at
40 CFR 60.107a(c)(6) and 40 CFR 60.107a(d)(8) based on the costs of the NOx CEMS, and
should provide significant relief for small refineries. We were unable to identify reasonable
alternatives to direct monitoring for flares, given the expected variability in flow and
composition of flare gas. As such, we require the same level of monitoring for flares regardless
of the size of the flare or the refinery.
122
------- |