Standards of Performance for Petroleum
Refineries
Background Information for Final Standards
Summary of Public Comments and Responses
-------
This page intentionally left blank.
11
-------
Standards of Performance for
Petroleum Refineries
Background Information for Final Standards
Summary of Public Comments and Responses
Contract No. EP-D-06-118
Work Assignment No. 1-12
Project No. 06/09
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emission Standards Division
Research Triangle Park, North Carolina 27711
April 2008
in
-------
iv
-------
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.
v
-------
TABLE OF CONTENTS
Page
Chapter 1: Summary 1
Chapter 2: Overview of Public Comments 2
Chapter 3: Definitions 6
3.1 Definition of Fuel Gas 6
3.2 Other Definition Issues 12
Chapter 4: Fuel Gas Combustion Device Standards 24
4.1 Tighter Ja Fuel Gas S02/H2S Standard 24
4.2 TRS in Fuel Gas 25
4.3 Process Heaters - NOx and S02 29
4.4 Monitoring Exemptions and Other Miscellaneous Fuel Gas Comments 34
Chapter 5: FCCU and FCU Emission Standards 39
5.1 Fluid Catalytic Cracking Units 39
5.1.1 PM Emission Limits and Opacity 39
5.1.2 NOx Emission Limit 58
5.1.3 S02 Emission Limit 60
5.1.4 CO Emissions Limit 64
5.1.5 Operating Parameter Limits 69
5.1.6 Other FCCU-related Comments 75
5.2 Fluid Coking Units 78
Chapter 6: Sulfur Recovery Plant (SRP) Standards 82
Chapter 7: Work Practice Standards 89
7.1 General 89
7.2 Flare Management 90
7.3 Flare Monitoring 99
7.4 Malfunctions of Amine Systems and SRP 101
7.5 Root Cause Analysis 103
7.6 Delayed Coking Depressurization 105
Chapter 8: Small Business Concerns 107
Chapter 9: NSPS Stringency 110
Chapter 10: Other Comments 112
vi
-------
Chapter 1
SUMMARY
On May 14, 2007, the U. S. Environmental Protection Agency (EPA) proposed
amendments to the Standards of Performance for Petroleum Refineries (40 CFR part 60,
subpart J) and separate standards of performance for new, modified, or reconstructed process
units at petroleum refineries (40 CFR part 60, subpart Ja). Public comments on the proposal and
EPA's proposed changes were requested when the proposal was published in the Federal
Register. Comments were received from 38 sources, including petroleum refiners, industry trade
associations and consultants, State and local environment and health departments, environmental
groups, and other interested parties.
On December 7, 2007, EPA published a Notice of Data Availability (NODA) notifying
interested parties that additional information had been added to the docket for the rulemaking.
Public comments on the additional data were requested at the time of publication in the Federal
Register. EPA received eight comments from petroleum refiners and industry trade associations.
All of the comments submitted by these 46 sources, and EPA's responses to the
comments, are summarized in this document or the preamble to the final amendments and new
standards. The summary and EPA's responses form part of the basis for the revisions made to
the standards between proposal and promulgation.
1
-------
Chapter 2
OVERVIEW OF PUBLIC COMMENTS
The public comment period following the May 14, 2007, Federal Register notice of
proposed amendments to the New Source Performance Standards (NSPS) for Petroleum
Refineries (40 CFR part 60, subpart J) and new proposed NSPS for petroleum refineries (40 CFR
part 60, subpart Ja) lasted from May 14, 2007 to August 27, 2007. A total of 38 letters
commenting on the proposed amendments and new standards for petroleum refineries were
received. These letters have been placed in the docket for this rulemaking (Docket No. EPA-
HQ-OAR-2007-0011). Table 1 lists the names of the persons submitting the 38 letters, their
affiliations, and the recorded docket item number assigned to their correspondence (i.e., the four
digits added to the end of the Docket No.).
Many commenters supported the comments submitted by others. By convention, rather
than identify all of the supporting commenters in each discussion of issues raised by a primary
commenter, we identified the supporting commenters only in this chapter. In the remainder of
this document, it is to be understood that each reference to the docket item number of a primary
commenter stands for all of the supporting commenters as well. For example,
Commenters 0123, 0125, 0135, 0136, 0137, 0138, 0139, 0140, 0141, 0142, 0143, 0144, 0145,
0150, 0151, 0153, 0157, 0158, 0159, 0170, and 173 supported the comments submitted by
Commenter 0154. Commenters 0152 and 0155 supported the comments submitted by
Commenter 0149. Commenter 0173 supported the comments submitted by Commenter 0129.
Some of the comments appear to be posted in the docket twice; their comments are subsequently
referred to by the first posting number assigned to the comments. Table 1 includes the secondary
docket number when the comments were posted twice.
The public comment period following the December 7, 2007, Federal Register notice
(NODA) lasted from December 7, 2007, to January 7, 2008. Five letters commenting on the
additional data were submitted. Within a month following the close of the public comment
2
-------
period, three additional comment letters were received.1 All of these letters have been placed in
the docket for this rulemaking (Docket ID No. EPA-HQ-OAR-2007-0011). Table 1 includes the
names of the persons submitting the eight letters, their affiliations, and the recorded docket item
number assigned to their correspondence {i.e., the four digits added to the end of the Docket ID
No.).
Table 1. List of Commenters on the Proposed Amendments to 40 CFR Part 60,
Subparts J and Ja
Docket llem No.
KPA-IIQ-OAU- ('ommenler and Affiliation
2007-001I-
0116
P. Kariuki, Texas Refinery Corporation, International Division, Nairobi,
Kenya
0121 G. Shankle, Texas Commission on Environmental Quality (TCEQ)
0122 Anonymous Public Commenter
0123 R. Metcalf, Louisiana Mid-Continent Oil and Gas Association
0124 L. Zink, Montana Sulphur & Chemical Co
0125 R. Hermanson, BP America, Inc.
L. A. Randel, Industry Professionals for Clean Air; also on behalf of
Mothers for Clean Air and the Galveston-Houston Association for Smog
Prevention
0126
(0126.1 and 0126.2)
0127, 0147 J.P. Broadbent, Bay Area Air Quality Management District
0128, 0134
T. Ballo, Earthjustice; also on behalf of Environmental Integrity Project
and Sierra Club
0129, 0133 S.V. Allen, Gary-Williams Energy Corporation, for Ad Hoc Coalition of
(0129.1, .2, & .3) Small Business Refiners
0130 B. J. Wakefield, Environmental Integrity Project and Sierra Club
0131 G. Garten, Lion Oil Company
0132 R. Smullen, HOVENSA
0135 P. Haid, HESS Corporation
0136 T. Fleming, Ohio Petroleum Council
0137 E. T. Roth, Wisconsin Petroleum Council
1 Two additional public comment letters were received over a month after the end of the public comment period, and
while the information provided in these comments was considered to the extent possible, the comments are not
summarized and responses are not provided in this document or the preamble to the final standards.
3
-------
.'I ltd
IIQ-<
)7-()()
0138
0139
0140
0141
0142
0143
0144
0145
0146
0148
0149
0150
0151
0152
0153
0154
55, 01
0156
0157
0158
0159
0161
0169
0170
0171
Coniiiicnlor siml AI'lllintion
D. Garg, Valero Energy Corporation
L. B. Barry, Chevron Corporation
J. A. Maxwell, New Jersey Petroleum Council
J. M. Griffin, American Petroleum Institute of Michigan (APIM)
S. Smith, Lyondell Chemical Company
M. McShane, Indiana Petroleum Council
R. Ness, North Dakota Petroleum Council (Petroleum Council)
G. B. Patterson, Delaware Petroleum Council
A. Mirzakhalili, Delaware Department of Natural Resources and
Environmental Control (DNERC)
Chuck Feerick, ExxonMobil Refining and Supply Company
B. Hodanbosi and Ursula Kramer, National Association of Clean Air
Agencies (NACAA)
D. F. Hunter, ConocoPhillips
J. K. Sims, U. S. Oil and Gas Association
M. Naxemi, South Coast Air Quality Management District (SCAQMD)
J. Pounds, Ohio Chemistry Technology Council (OCTC)
R. Chittim, American Petroleum Institute (API), National Petrochemical
and Petroleum Refiners Association (NPRA), and Western States
Petroleum Association (WSPA)
M. Asmundson, The Northwest Clean Air Agency (NWCAA)
A. Greene, CITGO Petroleum Corporation
T. Parker, The Arkansas Petroleum Council (APC)
D. M. Hastings, Texas Oil and Gas Association (TXOGA)
Marathon Petroleum Company LLC
T. K. Metrose, Tesoro Hawaii Corporation
K. Comey, affiliation unknown
R.A. Cade, Marathon Petroleum Company LLC
B. Lane, affiliation unknown
4
-------
Docket lU'in No.
KPA-IIQ-OAU- (onimcnler siml AI'lllintion
2007-001I-
0172
R. Chittim, American Petroleum Institute, also submitted on behalf of the
National Petrochemical and Refiners Association
0173 C.G. Swanberg, CVR Energy, Inc.
R. Chittim, American Petroleum Institute (API), National Petrochemical
0174 and Petroleum Refiners Association (NPRA), and Western States
Petroleum Association (WSPA)
0175 R.A. Cade, Marathon Petroleum Company LLC
0176 K.C. Antoine, HOVENSA LLC
5
-------
Chapter 3
DEFINITIONS
3.1 Definition of Fuel Gas
Comment: Commenters 0138, 0142, 0148, 0150, 0154, 0156, 0159 and 0161 supported
the proposed clarifications to the definition of "fuel gas" for both subparts J and Ja and suggested
that additional exclusions/clarifications be added. Commenter 0138 stated that EPA's original
intent was to define refinery fuel gases as gases that are: (1) generated by a refinery process unit;
(2) combusted for the purpose of energy recovery; and (3) amine treatable. Commenter 0154
recommended that EPA adopt these same three items as clarifications to the definition of "fuel
gas." With respect to item 1, Commenter 0154 suggested that fuel gases should be limited to
gases generated by equipment engaged in refinery processing operations and should not include
gases generated by non-processing units or ancillary equipment. The commenter noted that
although the definition of "fuel gas" in the rule only states that fuel gas is gas that is combusted,
the background documents suggest that it was EPA's intent to cover only gases that were burned
as fuels to produce useful work or heat, hence the term "fuel" gas. Commenters 0138, 0150,
and 0154 recommended that EPA include in the definition of "fuel gas" that the gas be
"combusted to produce useful work" similar to the definition in the National Emission Standards
for Organic Hazardous Air Pollutants for Equipment Leaks (HON) (40 CFR part 63, subpart H)
at 40 CFR 63.111. Commenter 0150 noted that it will be difficult and cumbersome to
specifically exempt all possible streams that do not meet this definition of deriving useful work,
so this simple clarification is preferable to a listing of exempt streams. Commenter 0154
suggested EPA define a heating value of 200 to 300 British thermal units per standard cubic feet
(Btu/scf), depending on the hydrogen content, as a means of establishing gas that can produce
useful work. Commenters 0138 and 0154 suggested that the streams that are not amenable to
amine treatment should not be included in the definition of "fuel gas."
6
-------
Commenter 0138 noted that fuel gas must be limited to gases combusted as fuels;
otherwise, defining "process gas" as "any gas generated by a petroleum refinery process unit,
except fuel gas and process upset gas" would have no meaning. The commenter noted that the
proposed rule does not include a new definition of "fuel gas," but the preamble makes clear
EPA's intent to cover what should be considered "process gas."
Response: We have reviewed the original NSPS background documents (Docket Item
No. EPA-HQ-OAR-2007-0011-0080, 0081, and 0082). The background document for the
proposed rule states that the standards apply to emissions from "process heaters, boilers, and
waste gas disposal systems [i.e., flares] that burn process gas generated at the refinery." This
statement is consistent with a relatively broad definition of "fuel gas" as included in subpart J,
but "process gas" as defined in subpart J is specific to gas generated by a "refinery process unit."
The background document for the promulgated standard also states that "Fuel gas is defined as
any gas produced by a process unit within a petroleum refinery and combusted as fueT
(emphasis added). The last two words in this statement of the definition of fuel gas suggest that
the original focus was on gases that were used as fuel, (i.e., were combusted to produce useful
heat or work). While the background information describes fuel gas as suggested by the
commenter, the attention to the detail of the specific words used in a background document are
far less than those used when developing the rule. Difficulties in segregating gases produced by
"refinery process units" and those produced from "non-refinery process units" (e.g., is the
wastewater treatment system a "refinery process unit?") and concerns about what constitutes
useful work (e.g., if combustion in a flare does not produce useful heat or work, does that mean
flares are not fuel gas combustion devices?) are likely to have led to the intentionally broad
definition of fuel gas. Also, the definition of fuel gas combustion system included in the
background document for the promulgated standards is "any equipment such as, but not limited
to, process heaters, boilers, and flares used to burn gases." Thus, in the background document,
fuel gas combustion devices are devices that burn any gas, so that the broad definition of fuel gas
in subpart J appears to be consistent with the background information.
With respect to limiting the fuel gas standards to gases that are amine treatable, there is
no indication that the definition of fuel gas was ever considered to be limited to gases that are
amine treatable. While the best demonstrated technology (BDT) review focused on the
performance of amine treatment systems, the background information clearly indicates that post-
7
-------
combustion controls can be used so long as they meet the alternative sulfur dioxide (SO2) limit
included in the original subpart J. The background document for the proposed rule states that
"the proposed standards will limit sulfur dioxide emissions to the atmosphere from heaters,
boilers, and flares by specifying that the fuel gas burned shall contain not more than
230 milligrams per normal cubic meter (mg/Nm3) of hydrogen sulfide, or 0.10 grains per dry
standard cubic feet (gr/dscf), unless resultant combustion gases are treated in a manner equally
effective in preventing the release to sulfur dioxide to the atmosphere." Similarly in the
background to the promulgated standards, the document states that "Although the standard limits
sulfur dioxide emissions by specifying a limit on the hydrogen sulfide content of fuel gas
combusted, compliance with the standard can be achieved be removing sulfur dioxide from the
combustion effluent gases instead of removing hydrogen sulfide from the fuel gas before
combustion." There is no discussion in any of the background information that the fuel gas
combustion standards were limited to gases that are amine treatable.
Although some specific parts of the background information for subpart J suggests that
fuel gas was intended to be limited to gases combusted as fuel, a more complete review of the
background information supports the Agency's broad interpretation of the definition of fuel gas.
If we were to limit the definition to gases combusted as fuels, then gases combusted for other
reasons would not be subject to any emission standards, and would have the effect of increasing
SO2 emissions. We have finalized our proposed clarifications that certain streams are not
considered fuel gas and others are inherently low in sulfur and do not need to be monitored, but
we do not believe we should relax the standards to the extent that the commenters suggest.
We disagree with the commenter that "process gas" has no meaning if fuel gas includes
all gas that is combusted. Propane produced in the refinery is a process gas. Effectively all
gases generated at a refinery that are combusted and that are not a result of process upset or
malfunction are, by definition, process gases. It appears that the commenter would like to define
process gas as gas that is combusted, but that is not combusted as fuel. We summarily reject this
idea as it would generally exclude gases combusted in flares, as these gases are arguable not
fuels (not producing useful work or heat). The background information is clear that waste gas or
flare gas are fuel gas combustion devices and are subject to the fuel gas standards (unless the
gases combusted are process upsets gases). The definition of process gas may help to clarify that
there are gas streams in the refinery that are not subject to the hydrogen sulfide (H2S)
8
-------
concentration limit; however, we agree that the definition of process gas is unnecessary and have
removed it from subpart Ja.
Comment: Commenter 0127 suggested that the definition of "fuel gas" should also
exclude vapors that are collected and combusted to comply with wastewater provisions or marine
tank vessel loading provisions in State Implementation Plan (SlP)-approved rules (for the same
reasons these sources are excluded when complying with the federal rules). Commenter 0174
asserted that vapors from wastewater and marine vessel loading are not fuel gas; therefore, the
commenter requested that the definition of fuel gas exclude these streams regardless of whether
or not they are subject to one of the named Federal regulations. The commenter noted that the
proposed definition of fuel gas does not account for sources combusting those vapors voluntarily,
or to meet a State regulation or permit, or to comply with Federal regulation not included in the
definition. The commenter also noted that the Bay Area Air Quality Management District flare
rule (EPA-HQ-OAR-2007-0011-0162) excludes flares that exclusively handle emissions from
storage, loading, and wastewater treatment systems. Commenter 0131 suggested an exclusion
from the fuel gas definition for fuel gas vapors from truck and rail loading docks vented to air
pollution control devices; each of the reasons EPA articulated at proposal for exempting
wastewater and marine vessel loading sources are equally applicable to truck and rail loading and
the exemption should be expanded to include them.
Commenter 0130 disagreed with the exemptions in both subparts J and Ja for fuel gas
vapors from control devices complying with 40 CFR part 60, subpart QQQ (wastewater systems)
and 40 CFR part 63, subpart Y (marine vessel loading), as the effect would be to exempt the
emissions from H2S concentration limits. The commenter disagreed with EPA rationale that
these emissions are typically low in H2S, the streams are not cost-effective to amine treat, and
loading sources are often located at the edge of the refinery so it is not economically reasonable
to regulate. The commenter stated that it is technically and economically feasible to regulate the
wastewater and marine vessel loading emissions. Given the hazards to human health and the
environment, all sources of H2S emissions at oil refineries should be considered fuel gas and be
subject to the S02 or H2S concentration limits, if it is economically and technically feasible to do
so. As EPA acknowledges that variability may result in these streams exceeding the current fuel
gas H2S concentration limits, the emissions should be regulated as fuel gas. The commenter
stated that EPA has not explained the meaning of cost effective or economically reasonable,
9
-------
although the commenter noted that an endeavor may be economically feasible without being
cost-effective.
Response: It is beyond the scope of the NSPS review to evaluate each and every State
regulation or SIP to determine if the combustion systems in these rules are comparable to the
Federal regulations cited. If necessary, an applicability determination request can be made so
that the specific State or local requirements can be fully evaluated.
Comment: Commenter 0125 recommended that the following exemptions be included in
the definition of "fuel gas": (1) low-BTU gas; (2) start-up hydrocarbon purges; (3)
depressurizing purges and "steamout steam" for units shutting down; (4) heater tube decoking
purges; (5) reformer catalyst regenerator gas; and (6) green coke calciner process gas.
Commenter 0138 provided a list of 16 streams/gases that they suggested should not be
defined as fuel gas. The commenter supported the clarification that monitoring is not required
for exempt streams and suggested that the 16 streams listed be likewise excluded.
Commenter 0154 suggested 39 streams that they believe should be exempted from the definition
of "fuel gas"; Commenter 0174 added vapors from all wastewater and marine loading operations
to that list. Commenter 0156 listed 10 additional streams and suggested that any stream
controlled by any State or federal regulation that uses a combustion device should be exempted
from the definition of fuel gas. Commenter 0154 recommended that any stream with a sulfur
content and flow rate such that combustion of these gases would be below 500 pounds per day
(lb/day) (the reportable quantity for S02) should be excluded from the definition of fuel gas.
Commenter 0148 provided a list of five additional vent streams that are combusted for emission
control that they believe should be exempted from the definition of "fuel gas." The commenter
noted that the standard for miscellaneous process vents in National Emission Standards for
Hazardous Air Pollutants From Petroleum Refineries (40 CFR part 63, subpart CC) (Refinery
MACT I) specifically stated that there should be no overlap between Refinery MACT I and other
Federal rules, and the commenter recommended that all vents subject to other NSPS or MACT
standards be excluded from the fuel gas requirements in subparts J and Ja.
Commenter 0146 recommended expanding the definition of "fuel gas" to include
synthesis gas produced by gasification of petroleum coke produced by fluid coking operations
because raw synthesis gas can potentially contain significant concentrations of H2S, up to 2 mole
percent. The commenter also noted that current generation amine scrubbing technologies have
10
-------
been demonstrated to desulfurize synthesis gas to levels well below those in the proposed
standard.
Response: Given the numerous comments and different lists of process streams received
on this issue, it is evident that providing a detailed list of every stream exempt from the
definition of fuel gas is not a practical solution, nor do we think it is warranted. The streams
exempted at proposal were exempted based on the costs and technical issues associated with
recovering these remote, oxygen-containing gas streams and adding them into the refinery's fuel
gas system. The commenters did not provide specific rationale along these same lines as why
these additional gas streams should be exempted from the definition of fuel gas. The primary
focus of the comments was that fuel gases should apply only to those gases that are used as fuels.
Comment: Commenters 0150 and 0154 stated that vapor from sulfur pits should not be
considered under the definition of "fuel gas."
Response: Vapors from sulfur pits, if combusted, meet the definition of fuel gas. The
real question is whether the devices that combust these gases are fuel gas combustion devices.
Fuel gas combustion devices exclude the combustion of fuel gas at "facilities in which gases are
combusted to produce sulfur or sulfuric acid." This means that gases combusted in the sulfur
recovery plant (SRP) are not subject to the fuel gas combustion standards, but are subject to the
SRP standards. At proposal, we specifically included the sulfur pit as part of the SRP, but at
promulgation, we only clarified this definition for subpart Ja. We note that there is a difference
in the subpart J definition of a Claus SRP and the exclusion provided by the fuel gas combustion
device so that one can argue that the combustion of sulfur pit vapors at the SRP are not subject to
the fuel gas combustion device standards. It is the Agency's opinion that the combustion of
sulfur pit vapors is subject to the SRP standards and not the fuel gas combustion device
standards.
Comment: Commenter 0125 suggested that EPA specify in the definition of "fuel gas"
that mixing of fuel gas with natural gas is not circumvention by use of gaseous diluents per
§60.12.
Response: Mixing of natural gas and refinery fuel gas as well as mixing of sweet refinery
fuel gas with high-sulfur fuel gas is commonly practiced, in many cases to ensure that the
resulting fuel gas has the desired properties for combustion. These practices are not considered
dilution and are therefore allowed within the regulation. In our proposed amendments, we
11
-------
attempted to make it clear that the affected source is the fuel gas combustion device and that the
SO2 (or alternative H2S or total reduced sulfur (TRS) limits) apply at the affected source. We do
not require that each individual fuel gas stream produced in the refinery that has a concentration
exceeding the alternative H2S concentration limit be treated to meet the applicable concentration
limit; rather, it is the fuel gas fed to the fuel gas combustion device that must meet the
concentration limit. In the final amendments, we clarified that the H2S concentration
measurement can be made at a central location, after fuel gases from several sources have been
mixed, which may include the addition of natural gas.
3.2 Other Definition Issues
Comment: Commenters 0125, 0148, 0150 and 0154 stated that the definition of "process
upset gas" should not be changed in subpart Ja; "process upset gas" should include start-up and
shutdown. Not all shutdowns are planned, and not all flaring during planned shutdowns can be
eliminated. It is not cost-effective (and sometimes infeasible) to eliminate flaring; the flare
management plan (FMP) and general duties under §60.11(d) are sufficient to minimize the start-
up and shutdown emissions.
Response: We agree with the commenter that not all shutdowns are planned and that not
all flaring can be eliminated from planned start-ups and shutdowns. However, we have not
changed the definition of "process upset gas" from proposed subpart Ja. The final work practice
standards for flares include a limit on the flare flow during normal operations and a FMP; further
details on the requirements for affected flares are provided in the preamble to the final standards
and later in this document.
Comment: Commenter 0131 suggested the exemption for process upset gases and fuel
gas released to the flare as a result of relief valve leakage and other emergency malfunctions
should expressly include an exemption for fuel gas released to the flare as a result of flare gas
recovery unit compressor staging because this process meets the definition of startup under
subparts J and Ja.
Response: The premise of the comment is incorrect. Startup is defined in the General
Provisions (subpart A) to part 60 as "the setting in operation of an affectedfacility for any
purpose" (emphasis added). Subparts J and Ja do not alter this definition, and the flare gas
recovery unit is not an "affected facility" under subparts J or Ja.
12
-------
Comment: Commenter 0154 requested that a definition of "auxiliary" (or
"supplemental") fuel be added to both subparts J and Ja and noted that monitoring of the gas
mixture is not needed if the auxiliary fuel meets the fuel gas standards.
Response: We acknowledge that there is no need to monitor the mixture of the fuel gas
fed to a fuel gas combustion device as long as each of the individual fuel gas streams fed to the
fuel gas combustion device meets the fuel gas standard. This suggests that all of the individual
streams are monitored unless they are specifically exempt from monitoring as an inherently low-
sulfur fuel gas. For example, if natural gas is used as an auxiliary fuel and mixed with the
primary fuel gas at a given process heater, the refinery owner and operator does not need to
demonstrate compliance with the standard at the process heater if the primary fuel gas meets the
fuel gas standard. In this case, the mixture would certainly meet the standard. However, when
the primary fuel gas does not meet the standard, questions and uncertainties arise regarding
whether the fuel gas combustion device is in compliance with the standard. It could be that, with
the addition of the auxiliary fuel, the fuel gas combustion device is in compliance with the
standards, even though the primary fuel gas used would exceed the standard if it were the only
fuel combusted. In response to this comment, we expressly allow an exemption from monitoring
at the fuel gas combustion device when all gases that are mixed at the unit meet the fuel gas
requirement. However, we also expressly indicate that, if the owner or operator elects this
option, an exceedance occurs each time any of the individual fuel gas streams that constitute the
total fuel gas fed to the combustion device exceeds the fuel gas standards. Therefore, it may be
desirable for a refinery owner and operator to monitor the fuel gas as fed to the fuel gas
combustion unit; however, if the owner and operator does not, they must report any exceedance
of the fuel gas standard for any fuel gas streams used by the fuel gas combustion device as an
exceedance for that fuel gas combustion device, regardless of the relative quantities of fuels fed
to that fuel gas combustion device.
Comment: Commenters 0150 and 0154 opposed the definition of "other fuel gas
combustion device" and suggested maintaining the definition of "fuel gas combustion device" in
subpart J. The commenters stated that the new definition is needless and confusing.
Commenter 0150 suggested that, if necessary, these devices can be referred to as "fuel gas
combustion devices other than process heaters" without having to add a definition.
Commenter 0146 stated that the new definition is confusing because it suggests that process
13
-------
heaters are excluded and recommended rephrasing the definition of other fuel gas combustion
device to include process heaters as being affected units, along with boilers and flares: "Other
fuel gas combustion device means any equipment, such as process heaters boilers and flares,
used to combust fuel gas. However, facilities, inclusive of process heaters, in which gases are
combusted to produce sulfur or sulfuric acid, are excluded from this definition."
Response: We agree that there is no real need to define "other fuel gas combustion
devices." In all cases where standards apply to "other fuel gas combustion devices," the
standards also apply to "process heaters." Therefore, to clarify the requirements, we have
revised the proposed subpart Ja to either refer to "fuel gas combustion devices," which includes
process heaters, or to refer specifically to process heaters. This revision makes the definition of
fuel gas combustion devices consistent between subparts J and Ja.
Comment: Commenter 0125 suggested that the following equipment be excluded from
the definition of fuel gas combustion device: (1) gas-fired combustion turbines; (2) cogeneration
units; (3) any incinerator installed to comply with control provisions in any 40 CFR part 60, 61,
or 63 rule; (4) any internal combustion device; (5) cutting/welding torches, space heaters, and
similar miscellaneous equipments; and (6) temporary equipment, such as portable flares or
thermal oxidizers, used during maintenance or tank cleaning. Similarly, Commenter 0171 asked
EPA to clarify whether turbines, cogeneration units, welding machines, and internal combustion
engines are fuel gas combustion devices.
Response: We believe fuel gases combusted in gas-fired combustion turbines,
cogeneration units, or internal combustion devices should meet the fuel gas standards. That is,
we clarify in this response that these units are considered fuel gas combustion devices. The
definition of fuel gas combustion device is "any equipment... used to combust fuel gas." The
specific listing of typical types of fuel gas combustion devices in the definition was never
intended to be an exhaustive list. We anticipate that cutting/welding torches, space heaters, and
similar miscellaneous equipments would use fuels that meet the fuel gas sulfur content
specifications and would not combust sour refinery fuel gas. However, if these units are
combusting refinery fuel gas, they should be utilizing fuel gases that are treated to meet the
standard; therefore, we are not excluding them from the standard. We believe the definition of
fuel gas in this final rule adequately addresses these issues without specific exclusions in the
definition of fuel gas combustion device.
14
-------
Comment: Best practices at Commenter 0170's refineries include installation of a Safety
Instrumentation System (SIS) on each process heater to shut down a process heater quickly and
safely if there are unsafe conditions. The commenter noted that a SIS includes shutdown valves,
flow meters, and sensors for pressure, temperature, and flame. The commenter stated that the
system has no effect on the normal operation of the heater (e.g., firing rate or capacity,
efficiency); the SIS would function as a pollution control device if a shutdown or malfunction
occurs. Therefore, the commenter believes that the SIS is not part of the process heater "affected
facility," and the costs of the SIS should not be considered when determining whether a project
is a modification or reconstruction. The commenter requested that EPA clarify that safety
systems are not part of the process heater "affected source."
Response: As the SIS is integral to the safe operation of the process heater, we believe
that SIS is part of the "affected facility."
Comment: Commenter 0139, 0148, 0150, 0154, 0156, 0161, and 0174 objected to the
expanded definition of "sulfur recovery plant" that includes all (multiple) Claus trains, tail gas
units, sulfur pits, and storage tanks. Commenters 0148, 0150, 0156, and 0161 suggested the
revised definition would make (unlawful) retroactive changes to subpart J, so any change in
definition must be restricted to subpart Ja. Commenters 0142 and 0154 asserted that these
amendments to subpart J will cause refineries to be out of compliance and are clearly "ex post
facto" provisions. Commenters 0150 and 0174 noted that applicability determinations are case-
and site-specific, and EPA should not expand one specific applicability determination to a
broader regulatory interpretation. Commenter 0174 added that EPA previously stated that
applicability determinations are not "nationally applicable" actions within the meaning of
section 307(b)(1) of the Clean Air Act (CAA) (68 FR 7373).
Commenter 0161 stated that the sulfur pit should not be included as part of the sulfur
recovery plant (SRP) because EPA's assumption that pit gases are normally recycled to the
beginning of the sulfur recovery train is incorrect. Commenter 0170 stated that changing the
relevant definitions to clarify that sulfur pits are included in the definition of SRP could conflict
with the consent decree requirements. Commenter 0174 noted that although Applicability
Determination 0500042 states that sulfur pits should be included in the SRP in 2004, an earlier
applicability determination (NR71, 1990) clearly indicates that the sulfur pits were not always
considered to be included; therefore, EPA cannot call a national application of that determination
15
-------
a "clarification" to subpart J. Commenter 0170 stated that sulfur pit vents are currently
adequately controlled, and they were not considered as part of the original NSPS {i.e., BDT for
sulfur pits was not determined), so the change in the definition cannot be considered a
clarification. Commenter 0174 agreed, noting that the documentation from the original NSPS
identified vents from sulfur pits and recognized that they were uncontrolled but did not set
standards for those vents. Commenter 0175 noted that at their Louisiana refinery, H2S is a
regulated pollutant, and the H2S emissions from the primary and downstream pits, tanks, and
loading operations combined are less than 8 tons per year (tons/yr); in addition, the refinery uses
less than one-sixth of the 576 hours allowed for maintenance activities per year.
If EPA does include sulfur pits as part of the SRP affected facility in subpart Ja,
Commenters 0148, 0156, and 0159 recommended that EPA include a provision to perform
periodic maintenance (of sulfur pit eductors and transfer piping) without facilities being out of
compliance (Commenter 0148 provided example language). Commenter 0170 suggested that
including sulfur pit vents as part of the SRP with a maintenance allowance for sulfur pit eductors
should be provided as an alternative compliance option rather than the only standard.
Commenter 0174 suggested that EPA provide 240 hours per year of control system outage for
maintenance to prevent plugging. Commenters 0150 and 0156 stated that, if appropriate, sulfur
pits should be a separate affected facility under subpart Ja rather than included in the definition
of the SRP. Commenter 0174 stated that sulfur pit vents can only be included in subpart Ja if the
BDT and other required analyses are completed and provided for public comment.
Commenters 0150, 0156, and 0159 stated that the sulfur pit requirements should apply
only to the primary sulfur pit and that secondary pits, downstream tanks, and loading racks
should be expressly excluded, as it is not cost effective to control these downstream sources.
Similarly, Commenters 0139, 0150, and 0154 urged EPA not to include sulfur storage facilities
in the SRP definition until they can justify the controls via the required BDT analysis.
Commenter 0175 requested that the SRP definition exclude tanks that are downstream of the first
liquid sulfur drop-out point and physically separate from the SRP. Commenters 0139 and 0161
stated that EPA has previously indicated that sulfur storage tanks and loading racks exist to
distribute sulfur product and are not part of the SRP (see EPA Applicability Determination
Control #0500042). A cost-effectiveness evaluation would show EPA's expanded control is not
justified. Commenter 0175 indicated that cost-effectiveness values for a conceptual project are
16
-------
estimated to be $150,000 per ton of H2S for a secondary tank and may increase to $2,000,000 per
ton if a degassing chemical is used; the degassing chemical only reduces H2S from 7 tons/yr to
less than 1 ton/yr. Commenter 0139 stated that emissions from downstream fugitive sources are
de minimis, and EPA has not provided SO2 and H2S emissions data to demonstrate these fugitive
sources are significant. Commenter 0139 further noted that imposing controls on sulfur storage
facilities downstream of Claus trains will adversely impact some consent decree projects.
Commenter 0156 provided permit limits based on actual measurements indicating that
downstream storage tank emissions are 40 times smaller than the primary sulfur pits.
Commenter 0175 also noted that if vapors from sulfur storage tanks are required to be returned to
the Claus unit, the distance and pressure difference between the tank and the Claus unit cause
safety and maintenance concerns (e.g., potential for fire, plugging of the line between the two
units).
Commenter 0150 stated that the definition of SRP could have unintentional consequences
due to the broadness of the definition. The commenter recommended EPA clarify that Merox
units, acid plants, and caustic scrubbers are not part of the SRP by including the word
"elemental" to clarify that an SRP recovers "elemental" sulfur.
Response: We believe the definition as currently written provides for coverage of sulfur
pits. Therefore, we have decided not to amend the definition of "Claus sulfur recovery plant" to
specifically include multiple process units or sulfur pits in subpart J.
The definition of "sulfur recovery plant" in subpart Ja includes sulfur pits but does not
include secondary sulfur storage vessels downstream of the sulfur pits. This definition reflects
the Agency's current understanding of the chemistry and operation of a SRP. The Agency's
position is that the sulfur pits are part of the Claus SRP because removing the elemental sulfur
between Claus reactors is essential in shifting the equilibrium of the "vapor-phase catalytic
reaction" towards completion. If the Claus SRP "process unit" did not include sulfur pits, it
could not effectively recover sulfur. As such, the sulfur pit is an integral part of the overall
process unit described in the definition of a Claus SRP. To determine whether controlling the
sulfur pits is cost-effective, we performed a BDT analysis considering three options: (1) not
including sulfur pits; (2) including primary sulfur pits; and (3) including all sulfur pits and
secondary sulfur storage vessels. Based on the results shown in Table 2, controlling primary
17
-------
sulfur pits is cost effective for both new and modified and reconstructed SRP, but including all
sulfur storage vessels is not cost-effective.
Table 2. National Fifth Year Impacts of Options for Standards Considered for New,
Reconstructed, and Modified Sulfur Recovery Plant Sulfur Pits Subject to 40 CFR Part 60,
Subpart Ja
Option
Capital
Cost
($1000)
Total
Annual Cost
($l,000/yr)
Emission
Reduction
(tons S02/yr)
Cost-
effectiveness
($/ton)
Incremental Cost-
effectiveness
($/ton)
New SRP — Baseline emissions: 31 tons S<
Vyr
1
0
0
0
N/A
N/A
2
660
86
54
1,600
1,600
3
1,300
170
55
3,100
63,000
Modified and F
Leconstructet
SRP — Baseline emissions: 280 tons SCVyr
1
0
0
0
N/A
N/A
2
7,700
940
480
1,900
1,900
3
15,000
1,900
500
3,800
78,000
Based on the public comments, we are including a provision that allows the sulfur pits
240 hours per year of uncontrolled operation to perform preventative maintenance on the
eductors and transfer piping. During times that the sulfur pits are not controlled, refiners have
the general obligation to minimize emissions to the extent practicable, consistent with good air
pollution practices.
Finally, we are not specifically including "elemental" sulfur in the definition of "sulfur
recovery plant" because doing so would exclude Lo-Cat units and other sulfur recovery methods
that should be required to achieve the emission limits for small SRP.
Comment: Commenter 0150 noted that, while EPA may consider multiple Claus trains in
a given sulfur recovery plant to be a single affected source, it should not consider separate,
independent sulfur recovery plants to be a single affected source. Commenter 0150 requested
clarification as to whether truly independent SRP are considered together under the revised
definition of SRP in the subpart Ja proposal. Commenter 0150 also suggested that, if all SRP are
considered together, then completely dismantling and replacing the SRP would not trigger a
modification as long as the total SO2 emissions do not increase. Commenter 0156 stated that
each separate sulfur recovery unit should be considered as separate affected sources.
18
-------
Commenters 0148 and 0154 stated that multiple, independent Claus trains should be separate
affected facilities.
Response: Consistent with current Agency interpretation, we are clarifying in subpart Ja
that multiple sulfur recovery units (SRU) or trains are part of a single SRP only when they have
the same sour gas source. If the SRU receive sour gas from completely segregated sour gas
treatment systems and there is no connection between the SRU, than the SRU are considered to
be part of separate SRP. If there is a connection between SRU such that sour gas could be
transferred from one to the other, these SRU are considered one SRP. In response to the
commenter who suggested that dismantling and rebuilding an SRP would not trigger
modification provisions, we note that the resulting SRP would either become subject to subpart
Ja as a new SRP or a reconstructed SRP, depending on the exact situation.
Comment: Commenter 0125 requested that EPA clarify that the revised definition of
"sulfur recovery plant" as an affected source does not affect the customary exclusion from
compliance during start-up or shutdown of a single amine stripper, Claus train, or tail gas
treatment unit. EPA did not discuss or evaluate the impact of this definitional change on the
affected source as it pertains to §60.2; therefore, EPA should specifically state in both subparts J
and Ja that the new definition of "sulfur recovery plant" does not preclude the use of this
exclusion during a start-up or shutdown of a portion of the affected facility.
Response: The amine stripper is not an affected facility, and is therefore not included in
the definition of a shutdown as it pertains to §60.2. In the preamble to the proposed rule, we
indicated that the sulfur standards for fuel gas combustion devices and the sulfur recovery plant
were to be met at all times. If one Claus train in a Claus SRP shutdown, diversion of flow to the
operating Claus units and implementation of a sulfur shedding plan are considered good air
pollution control practices. However, provided that the root cause analysis is conducted
(presuming the shutdown event causes a release of 500 lbs/day or more of SO2), the definitions
in subpart Ja do not necessarily preclude the use of this exclusion during a start-up or shutdown
of a portion of the affected facility.
Comment: Commenters 0148, 0150, and 0154 opposed changing the definition of any
affected source to include multiple units of any kind and recommended that subpart Ja clearly
state that multiple fluid catalytic cracking units (FCCU) are not one affected facility. Multiple,
19
-------
independent refinery process units, whether they are FCCU or fluid coking units (FCU), should
be separate affected facilities, according to Commenters 0148 and 0154.
Response: In this final rule, we are clarifying that multiple independent FCCU are
separate affected sources so long as they are truly independent. If the FCCU regenerator exhaust
from two separate FCCU share a common exhaust treatment unit (e.g., carbon monoxide (CO)
boiler or wet scrubber), then the FCCU are not independent and considered a single affected
source.
Comment: Commenter 0150 requested that the definition of "fuel gas" be consistent with
the definition in the HON. Commenter 0156 suggested that the original definition of FCCU in
subpart J be retained in subpart Ja. Commenter 0159 objected to the proposed definition of
"fluid catalytic cracking unit" in subpart Ja, which includes the control and heat recovery
equipment. The definition is inconsistent with subpart J and other NSPS and National Emission
Standards for Hazardous Air Pollutants for Petroleum Refineries: Catalytic Cracking Units,
Catalytic Reforming Units, and Sulfur Recovery Units (40 CFR part 63, subpart UUU) (Refinery
MACT II); the commenter noted that subpart J only includes the FCCU catalyst regenerator and
Refinery MACT II includes only the process vents that are associated with regeneration of the
catalyst such as the catalyst regeneration flue vent. The commenter stated that regulatory
precedent has generally limited applicability to the FCCU components generating emissions, and
the preamble did not explain why EPA expanded the definition for subpart Ja. The commenter
stated that EPA stated in 1989 that limiting applicability of the sulfur oxide (SOx) standards to
the regenerator of the FCCU "would lead to bringing replacement equipment under these
standards sooner." At a minimum, the commenter recommended that EPA exclude the pollution
control device if the affected facility is expanded for FCCU.
Response: The definitions in the final subpart Ja have been developed to describe each
affected source as appropriate for this subpart. The definition of the FCCU specifically included
the heat recovery system (CO boiler) as this system is integral to the operation of partial
combustion units. We note that including the whole FCCU as the affected source raises the 50%
cost threshold for reconstruction, but it would require the refinery to consider whether any
change to the FCCU is a modification or reconstruction.
Comment: Commenters 0154 and 0156 suggested deleting the definition of "process
gas" because it in not used in the rule and therefore is not needed.
20
-------
Response: Because the term "process gas" is not used in subpart Ja, we have not
included a definition of the term in §60.101a. Although the term is not used directly in subpart J,
we have not amended any definitions in subpart J, including removing terms, other than the
change to the definition of fuel gas as described in the preamble to the final rule.
Comment: Commenter 0154 suggested that EPA define "petroleum refinery process
unit" (rather than "refinery process unit") as: "a process unit that refines petroleum products as
defined in 'petroleum refinery.'" Commenter 0150 suggested that the definition of "refinery
process unit" be consistent with Refinery MACT I and they requested revision of the term
"petroleum refinery" to include SIC Code 2911. Commenter 0156 stated that "refinery process
unit" should be defined consistent with the definition of process unit in §60.481 to stress the
independence of a process unit.
Response: The suggested definition of "petroleum refinery process unit" would exclude
sulfur plants and amine treatment systems from the definition of refinery process units. This
definitional change would help to clarify that start-up and shutdown of amine strippers or Claus
SRP are not included in the definition of process upset gas. At proposal, we indicated that the
process upset gas exclusion did not apply because these units did not "generate" the sour gas.
This definitional change in refinery process unit would make this point more clear. However, we
do not believe this is the commenter's intent, especially in view of other comments provided by
this commenter. The commenter provided no compelling rationale for the requested changes,
and we do not believe the commenter has thoroughly considered the ramifications of the
definitional changes suggested. As such, we decided not to make the suggested changes.
Comment: Commenter 0154 suggested that EPA define nitrogen oxides (NOx) as
nitrogen dioxide (NO2) in §60.102a(b)(2).
Response: The emission limit applies to all NOx compounds as measured by EPA
Methods 7, 7A, 7C, 7D, or 7E. These methods measure total NOx concentrations as NO2. The
primary NOx species (those specifically listed in these methods) are nitric oxide (NO) and NO2.
We believe the specification of the test methods adequately define nitrogen oxides and that they
are measures as N02.
Comment: Commenter 0154 suggested that EPA define particulate matter (PM) as
determined by Method 5B or 5F (consistent with subpart J). According to the commenter,
Method 5 does not provide predictable, reproducible results.
21
-------
Response: The differences in Methods 5, 5B, and 5F are addressed at length in the
discussion of the PM emission limits for FCCU and FCU. The term "particulate matter" is
already defined in the NSPS General Provisions (40 CFR part 60, subpart A), and we do not see
a need to define it differently in this subpart.
Comment: Commenters 0150 and 0154 suggested that EPA delete the definition of "fuel
gas producing unit."
Response: This comment is addressed at length in the discussion of the work practice
standards. In response to this particular comment, we note that we are deleting the definition of
"fuel gas producing unit" because it is no longer necessary due to changes in the work practice
standards between proposal and promulgation.
Comment: Commenter 0154 suggested that EPA define "amine treatment system" and
specifically state that it is not an affected facility.
Response: We consider the amine treatment system to be a control system used to
comply with the sulfur standards for fuel gas combustion devices (see Section 7.4 for further
details). The amine treatment system is part of the fuel gas system, which has been defined in
subpart Ja for purposes of clarifying when a flare has been modified (see Section 7.2 for further
details).
Comment: Commenter 0127 supported the revised definition of "oxidation control
system" to clarify that thermal oxidizers are not "oxidation control systems." On the other hand,
Commenter 0150 stated that not all oxidative and reductive control systems route the S02 or H2S
to the reactor furnace or first stage reactor. Therefore, these SRP currently do not meet the
proposed definition of SRP in subpart J. If EPA intended to require all SRP control systems to
route emissions to the reactor furnace or first stage reactor, then these SRP will be significantly
impacted by the rule. The commenter recommended no change in the definitions from those in
the existing subpart J and further suggested that this change in definitions in the proposed
amendments to subpart J are unlawful, as they impose substantive requirements to existing units.
Response: As the definition of "oxidative control system" includes the concept that the
control system is reducing emissions from the sulfur recovery plant by converting these
emissions to SO2, we do not believe incinerators or thermal oxidizers qualify as an oxidative
control system. We agree with the commenter that indicated that not all oxidative or reductive
control systems recycle sulfur to the front of the Claus unit. For example, a LoCat system can be
22
-------
used as a tail gas treatment unit for a Claus SRP to meet the subpart J or Ja standards without
recycling the sulfur to the Claus SRP. As such, the proposed amendments needlessly limit the
types of tail gas treatment systems that can be used; therefore, we are not amending these
definitions in the final amendments for subpart J or Ja.
23
-------
Chapter 4
FUEL GAS COMBUSTION DEVICE STANDARDS
Comment: Commenter 0127 supported focusing the fuel gas combustion device
standards in the proposed subpart Ja rule on SO2 rather than on H2S. The commenter noted that
relatively high amounts of S02 are released due to sulfur compounds other than H2S in the fuel
gas. The commenter opposed the use of an H2S or TRS limit (in lieu of the S02 limit) in
§60.102a(h) and suggested that a total sulfur limit be used. Total sulfur can be continuously
monitored, according to the commenter, and will include sulfur species not included in the
definition of "reduced sulfur compounds." The commenter also stated that the need to monitor
total sulfur will become even more important as refineries continue to use heavier crude slates.
On the other hand, Commenter 0156 stated that EPA should focus the standard on H2S (or TRS,
as applicable) and provide an alternative monitoring option for S02 as in subpart J.
Response: The focus of the fuel gas combustion device standards is to limit the S02
emissions from the combustion of refinery fuel gas. As such, we intentionally prepared the
emission limit in terms of S02. We agree that monitoring total sulfur content more directly
correlates with the S02 emissions generated from the combustion of refinery fuel gas. However,
based on our revised BDT analysis for removing non-H2S reduced sulfur compounds from the
fuel gas, we have concluded that the TRS standard is not BDT. We do maintain the more direct
S02 emissions standard in the final rule as an alternative to the H2S standard.
4.1 Tighter Ja Fuel Gas SO2/H2S Standard
Comment: Commenters 0150 and 0156 stated that EPA should revise the standard to the
familiar 162 parts per million by volume (ppmv) limit. Commenter 0156 stated that this is the
appropriate conversion of the 230 milligrams per dry standard cubic meter (mg/dscm) emission
limit because the original temperature basis was 68°F. Commenter 0154 recommended that EPA
either specify the standard conditions that apply to the 0.010 grains per dry standard cubic feet
(gr/dscf) H2S standard or specify that concentration values of 164 ppmv or less are in compliance
24
-------
with the standard (as 164 rounds to 160 with two significant digits). Commenter 0150 noted the
effort and cost to update computer systems and title V permits for this minimal change in the
emission limit (from 162 ppmv to 160 ppmv) is not justified. On the other hand,
Commenter 0161 stated that EPA should explicitly revise subpart J to reflect EPA's intention
that 230 mg/dscm is equivalent to 160 ppmv, not 162 ppmv. That would eliminate a great deal
of debate (over just a few ppmv) on how to properly convert from the current standard to ppmv.
The rule should simply be restated in terms of ppmv, which is also the terms by which it will be
measured and enforced, according to Commenter 0161.
Response: Standard conditions are defined in the NSPS General Provisions at 40 CFR
60.2 as being 68°F and 1 atmosphere. Using these as standard conditions, the subpart J emission
limit is equivalent to 162 ppmv. We did not propose and are not finalizing any revisions to the
subpart J emission limit; however, we are clarifying in this response that the subpart J emission
limit is effectively 162 ppmv. We agree with the commenter that the NSPS emission limit
should be provided in terms of ppmv, as these are the units by which the H2S concentration is
typically determined. Therefore, in our proposed subpart Ja rule, we specified the H2S
concentration in terms of a ppmv limit. We proposed an emission limit of 160 ppmv for a 3-hour
standard and we intended that this limit be applied to three significant digits. However, we agree
with the commenter that there is effort needed to adjust reporting output for just a few ppmv.
Also, to make it clear that the concentration limit was intended to be evaluated to 3 significant
digits and not allow 164 ppmv as a compliant concentration, we are revising the 3-hour standard
to state expressly that the concentration limit is 162 ppmv.
Comment: Commenters 0148, 0150, 0154, and 0174 recommended adding a 500 lb/day
SO2 compliance option specific to flares.
Response: The response to this comment is discussed in detail in Chapter 7 of this
document and in the preamble to the final amendments and standards. In summary, we are not
promulgating the commenter's suggested compliance option.
4.2 TRS in Fuel Gas
Comment: Commenter 0148 provided an example of a treatment system installed to
meet a facility-wide fuel gas standard of 40 ppmv total sulfur. The commenter estimated the
capital cost of the entire system to be $150-million but estimated only the emissions of TRS
25
-------
when reporting the cost-effectiveness for treating the gas from the delayed coking unit (DCU) to
be $60,000/ton SO2 reduced. The commenter also indicated that low-BTU gas from flexicoking
units would need to be specially treated to achieve a total sulfur content of less than 150 ppmv.
The commenter estimated the capital cost to be $61-million and the cost-effectiveness to be
$20,000/ton SO2 reduced and noted that the treatment would increase energy consumption,
resulting in increases in NOx, CO, and carbon dioxide (CO2) emissions. Commenter 0156
provided an estimate of $50-million (in 2007 dollars) to treat TRS down to 45 ppmv (long-term
average), resulting in 142 tons/yr of SO2 reduction to demonstrate that the TRS requirement is
not cost-effective (about $352,000 per ton of SO2).
Response: As described in the preamble to the final rule, re-evaluated BDT for TRS
based on the comments received and concluded that the TRS standard is not BDT.
Comment: Commenter 0174 stated that monitoring TRS in refinery fuel gas is not
technically feasible over a 425 ppmv span or the general operating range of refinery fuel gas
systems. Once calibrated, flame photometric detectors are only accurate over a range of about
50 ppmv, and the TRS concentration in coking unit offgas varies by more than 50 ppmv.
Calibrating multiple monitors over individual sections of the total range would be complex and
operators would be unable to tell which analyzer is correct where the detector ranges overlap.
These difficulties also affect the effectiveness of the suggestion that a refinery could "over treat"
the H2S in other fuel gas streams. If the TRS species and concentrations in coking unit offgas
cannot be determined, then it is impossible to know how much additional H2S would have to be
removed as well as how much additional equipment would be needed to meet the overall TRS
emission limit. Without those data, the cost-effectiveness of the proposal cannot be determined
for a BDT analysis. Therefore, the commenter stated that EPA should eliminate the requirement
to monitor TRS in refinery fuel gas.
Response: Total sulfur monitors could be used and would eliminate the issues described
by the commenter. Nonetheless, as described in the preamble to the final rule, we determined
that the TRS (or a total sulfur) standard was not cost-effective and therefore not BDT. As such,
we are not requiring TRS monitoring except for flares as needed to demonstrate compliance with
the 500 lb/day SO2 root cause analysis. We note again that total sulfur monitors could be used
for these flares if there are operational issues with the reduced sulfur monitors.
26
-------
Comment: Commenter 0156 stated that the new TRS standard would require all new
continuous emission monitoring systems (CEMS) to be installed and these costs were not
included in EPA's cost analysis.
Response: This comment is no longer relevant as we have not promulgated a TRS
standard for fuel gas systems. However, we have required TRS CEMS on newly affected flares.
In our revised analysis, we included the cost of a total reduced sulfur CEMS for newly affected
flare systems.
Comment: Commenter 0161 stated that there should be no automatic exemption or
limitation on the requirement to monitor all TRS compounds for refineries that do not combust
fuel gas from coking units. According to the commenter, other process units can also produce
reduced sulfur compounds. The commenter stated that the TRS monitoring requirement should
be applicable to all sources, unless a refinery can demonstrate the absence (< 3 percent) of other
reduced sulfur compounds, and that the exemption from TRS monitoring should be broadened
and made available to all refineries who can qualify.
Response: Our data indicate that non-H2S sulfur compounds are only a very small
fraction of the total sulfur content of fuel gas streams produced by units other than the coking
unit. The commenter provided no data to support that other gas streams contain appreciable
concentrations of non-H2S sulfur compounds. Based on the data available, we cannot justify the
expense of replacing all existing monitors, and we can only estimate the costs and emission
reductions of reducing TRS from fuel gas generated by coking units. We are interested in
determining the total amount of sulfur emissions from fuel gas combustion devices and are
considering the best way to obtain accurate measurements outside of these NSPS.
Comment: Commenters 0138, 0154, 0156, and 0159 stated that TRS is not defined in the
proposal and it is unclear, based on the test method and monitoring requirements, what
compounds are included in TRS {i.e., Method 16, and Performance Specification 5).
Commenter 0159 also noted that the term "coking unit" as used in the TRS standard is not
defined, leading to additional ambiguity.
Response: We anticipated that essentially all of the sulfur compounds in the fuel gas
would be encompassed by the four sulfur-containing compounds (H2S, methyl mercaptan,
dimethyl sulfide, and dimethyl disulfide) referenced collectively as TRS in Method 16.
27
-------
However, as we are not promulgating standards for TRS, the requested definitions are not
needed.
Comment: Commenter 0154 stated that EPA did not establish "standards of
performance" for TRS; therefore, the proposed TRS provisions do not trigger CAA
section 111(d). The commenter suggested that if any TRS standard is developed, it should be
applicable only to entirely new coking units. Commenter 0156 stated that EPA did not provide
exposure data or health data as to why EPA included a TRS standard.
Commenter 0154 suggested that, if EPA is concerned about TRS in coker fuel gas, the
20 ppmv SO2 concentration limit could be demonstrated initially and every 24 months thereafter
as an alternate monitoring plan to show compliance with the fuel gas combustion standards when
the fuel gas contains coker fuel gas.
Response: We are not promulgating an alternative compliance option for TRS, so the
response below is mainly focused the issue of triggering CAA section 111(d). The TRS
concentration limit was proposed as an alternative compliance method for an SO2 emission limit;
therefore, we agree that this compliance alternative would not trigger CAA section 111(d). As
TRS was proposed as an indicator of the SO2 emissions generated during fuel gas combustion,
we were not compelled to provide exposure or health data for TRS emissions {i.e., TRS is not
emitted, SO2 is).
Comment: Commenter 0161 noted that there are many challenges associated with
conducting a performance evaluation of a H2S or TRS CEMS using Method 16 because of the
different background matrix, which may preclude use of the direct monitoring of fuel gas.
Therefore, the commenter stated that the rule should alternatively allow the performance
evaluation to be conducted by measuring the SO2 concentration and oxygen (O2) content at the
stack and converting back to TRS or H2S on a common basis.
Response: First, we are not including TRS limits for fuel gas in the final rule. We do
specifically allow direct monitoring of SO2 in the fuel gas standards, so we do not see a need to
convert back to TRS or H2S concentration. For flares, we do require reduced sulfur monitors for
the fuel gas to assess the 500 lb/day root cause analysis, but we specify Methods 15 or 15A
rather than Method 16.
28
-------
4.3 Process Heaters - NOx and S02
Comment: Commenter 0154 recommended that EPA either develop NOx and SO2
standards that are fuel-specific or exempt fuel gas combustion devices during periods of natural
gas curtailment (i.e., when fuel oil is the only alternative). According to the commenter, non-
continental U.S. refineries as well as some continental refineries do not have access to natural
gas and therefore must use fuel oil in their process heaters. Commenter 0161 stated that the
extension of the SO2 standards to process heaters that combust liquid fuels is grossly flawed.
The cost to use low-sulfur fuels could be significant (e.g., added hydrotreatment capacity), and
these costs were not accounted for in EPA's BDT assessment. The commenter recommended
that fuel gas combustion devices that also burn liquid fuels be exempt from the 20 ppmv SO2
limit; the commenter noted that the fuel gas used in these units would still need to comply with
the H2S /TRS concentration limits.
Commenter 0154 recommended that the definition of "process heater" and "other fuel gas
combustion device" exclude units firing liquid fuels so they will not be subject to the SO2 limit.
Commenters 0154 and 0174 noted that the 80 ppmv NOx emission limit is not technically
feasible for heaters and boilers firing either fuel oil or a combination of fuel oil and refinery gas
(duel fuel). Based on vendor information, the commenters estimated that 150 ppmv NOx is
achievable when firing ultra low sulfur diesel and about 250 ppmv is achievable when firing
residual fuel oil. The commenter stated that it is not cost-effective to install selective catalytic
reduction (SCR) or other air pollution control devices (APCD) needed to meet the 80 ppmv limit
for the infrequent times that liquid fuels are fired, and EPA did not conclude that the use of SCR
for NOx control was BDT.
Response: Based on information received during meeting with industry representatives,
we expect that most, if not all, new process heaters will be gas-fired. Gas-fired process heaters
typically have lower NOx and S02 emissions than oil-fired units, especially units fired with high
sulfur fuel oil. While we maintain the SO2 emissions limit, we provide the H2S fuel gas standard
as an equal alternative. Also, we revised the definition of fuel gas combustion devices and made
the standards specific to fuel gas combustion devices. If the process heater is using only liquid
fuels, it would not meet the definition of fuel gas combustion device and would not be subject to
the emission limits. If the process heater co fires liquid and gaseous fuels, the fuel gas H2S
29
-------
standard could be used to demonstrate compliance, but the NOx emissions limit would still be
applicable.
Comment: Commenters 0150 and 0154 suggested that process heaters are more
appropriately regulated under a different NSPS because process heaters are universal process
units and are not exclusive to refineries.
Response: Refineries use a significant number of process heaters, more than most other
industries. Additionally, process heaters are the largest contributors to the NOx emissions at a
typical petroleum refinery. As such, we see no reason to delay applying NOx controls to new
refinery process heaters in favor of an NSPS that does not yet and may never exist.
Comment: Commenter 0154 stated that since SO2 emissions from gas turbines (40 CFR
part 60, subparts GG and KKKK) and boilers (40 CFR part 60, subparts Db and Dc) are already
regulated, these units should be exempt from Ja.
Response: As stated previously, gas turbines and boilers are fuel gas combustion
devices. Due to the specific nature and use of refinery fuel gas at petroleum refineries, we
consider it appropriate to develop specific standards for these combustion devices when fired
with refinery fuel gas. Similarly, due to the integrated operation of CO boilers and FCCU,
specific standards in subpart J and Ja are appropriate for these units.
Comment: Commenters 0154 and 0159 supported the 80 ppmv NOx emission limit but
opposed the NOx CEMS requirement, especially for smaller units. Commenter 0159 stated that
a NOx CEMS requirement would add $1 million to the project cost. Commenters 0148, 0154,
and 0159 stated that NOx emissions from process heaters are stable over long time periods;
therefore, compliance is adequately demonstrated using periodic stack tests instead of CEMS.
Commenters 0148, 0150, 0154, 0156, 0159, and 0174 recommended that process heaters less
than 100 million British thermal units per hour (MMBtu/hr) higher heating value be exempt from
the CEMS requirement; Commenter 0174 noted that this level is consistent with the monitoring
threshold in the National Emission Standards for Hazardous Air Pollutants for Industrial,
Commercial, and Institutional Boilers and Process Heaters (40 CFR part 63, subpart DDDDD)
(Boiler MACT) as well as many of the consent decrees. Performance tests on either an annual
(Commenter 0148) or biennial (every 24 months; Commenter 0154) basis could be used to
demonstrate compliance for smaller process heaters.
30
-------
Commenter 0131 suggested an increase in the size of the fuel gas combustion device on
which a NOx CEMS would be placed, from > 20 MMBtu/hr to 150 MMBtu/hr, which is
consistent with the consent decrees under the National Petroleum Refining Initiative. Measuring
NOx emissions requires a separate CEMS on each heater/boiler and is cost prohibitive, while
SO2 CEMS allow measurement of the H2S level in the entire fuel gas system using one CEMS.
Response: In our evaluation of the costs for the proposed NOx emissions limit, we
included the cost of NOx monitors for all process heaters. Although our cost-estimate for NOx
CEMS was significantly less than $1 million dollars, we did assess the relative cost for the
emission controls versus the cost for the CEMS. For smaller process heaters, the cost of the
CEMS was a significant portion (up to 30 percent) of the overall compliance costs. Furthermore,
available data indicate the NOx emissions for process heaters that employ low-NOx burners
(LNB) and ultra low-NOx burners (ULNB) are consistent and stable. Therefore, we agree that
CEMS are not cost-effective means of ensuring compliance for the smaller process heaters.
Based on our evaluation of the costs, and based on public comments, we determined that process
heaters with rated heating capacity of 100 MMBtu/hr or higher should install CEMS to
demonstrate compliance with the NOx emission limit. For process heaters with a rated heating
capacity of less than 100 MMBtu/hr, we provide an alternative biannual testing compliance
option for units that have installed LNB or ULNB.
Comment: Commenter 0154 stated that the capital costs for retrofitting existing process
heaters is understated by EPA's cost analysis because the ULNB are larger than conventional
burners. As such, some process heaters would have to be completely rebuilt in order to
accommodate the larger ULNB.
Response: Although ULNB may be larger than existing burners, there are a suite of
control technologies that are assumed to apply to a refinery's various process heaters.
Depending on the process heaters, the larger ULNB may still fit within the process heater
without significant retrofit issues. In cases where ULNB retrofits are a significant issue, there
are other technologies, such as exhaust gas recirculation, advanced system controls for better
excess air and temperature control, selective non-catalytic reduction systems, and selective
catalytic reduction systems, that are applicable for reducing the NOx emissions that do not have
the same retrofit issues. Furthermore, based on discussions with industry representatives, most
newly affected process heaters will be newly constructed units. There are only a few occasions
31
-------
in which a process heater would be modified. As such, only a very small percentage of newly
affected process heaters would have additional retrofit costs and these costs do not impact the
BDT determination.
Comment: Commenter 0154 noted a discrepancy between the rule language and
preamble discussion regarding the NOx emission limit being a 24-hour rolling average and a 7-
day rolling average. The commenter suggested that the 7-day rolling average is appropriate and
recommended that references to a 24-hour average be corrected.
Response: For process heaters, a 24-hour rolling average NOx limit is appropriate. As
this commenter and others mentioned, the NOx emissions from process heaters are very
consistent and stable when LNB or ULNB are used. Additionally, process heaters do not have
the competing mechanisms of CO control, coke-make, catalyst re-activation, and others of an
FCCU that impact the NOx emissions from FCCU. Therefore, we provided a 7-day average for
NOx emissions from the FCCU, but maintain the 24-hour average is appropriate for process
heaters.
The proposal preamble incorrectly stated that compliance is on a 7-day rolling average
basis at 72 FR 27182. A 24-hour rolling average is correctly indicated in the proposal preamble
at 72 FR 27194 and the proposed rule at 40 CFR 60.102a(g)(3). We have corrected the error in
the summary of the requirements in the preamble to the final standards.
Comment: Commenter 0131 noted that there are conflicting monitoring requirements for
miscellaneous process vents under Refinery MACT I and subpart J. Refinery MACT I allows
routing of miscellaneous process vents to process heater or boiler as a control device, and
subparts J and Ja suggest that CEMS on the heater or boiler would be required to comply with
subparts Ja and Ja fuel gas monitoring requirements. The commenter suggested that subparts J
and Ja should provide an exemption from the monitoring requirements when miscellaneous
process vents are being routed to a process heater or boiler for compliance with Refinery
MACT I.
Response: The Refinery MACT I standards were developed to regulate hazardous air
pollutant (HAP) emissions. For miscellaneous process vents, the primary concern is organic
HAP emissions, which are easily well-controlled by routing these emissions to process heaters
and boilers. The Refinery MACT I standards do not, and in fact cannot, address emissions of
SO2 because SO2 is a criteria air pollutant and not a HAP. We have provided streamlined
32
-------
procedures by which an exemption from the fuel gas monitoring requirements of individual
streams can be obtained if the gases contain low concentrations of sulfur. We see no logical
rationale to exclude miscellaneous process vents from the sulfur monitoring requirements
provided that the process vent gases meet the definition of fuel gas and do not meet other
monitoring exemptions provided in subpart J and Ja for inherently low-sulfur fuel gas.
Comment: Commenter 0150 noted that the span requirement for O2 CEMS used in
conjunction with NOx CEMS should be 10 percent rather than the 25 percent specified in the
proposed rule.
Response: In general, we anticipate that the oxygen content will be well below
10 percent in the process heaters (and other pollutant/emission sources for which an O2
correction is required), so that a 10 percent span value would be preferable. However, subpart J
specifically requires a span value of 25 and we received comments on the specified 10 percent
span requirement for the O2 CEMS in Refinery MACT II as initially promulgated in April 11,
2002. According to these previous comments, many O2 CEMS span checks are made with air,
so that the higher span value was specifically requested. Considering these factors, we have
revised the span requirements for O2 CEMS for process heaters (as well as other sources) to
allow the refinery owners and operators flexibility to set the O2 span at values between 10 and 25
percent, inclusive.
Comment: Commenter 0150 requested that EPA exempt any process heater subject to
the Boiler MACT (if promulgated prior to promulgation of subpart Ja) from the quality assurance
(QA) requirements; at a minimum, these requirements should be harmonized. The commenter
requested clarification related to Performance Specification 4a with respect to dual-range
analyzers.
Response: As noted previously, the Boiler MACT requirements specifically address
HAP, while the NSPS address criteria pollutants. As the pollutants being monitored differ, the
monitoring and QA requirements are also likely to differ. We reviewed the requirements under
the Boiler MACT and the primary requirement for gas-fired boilers and process heaters is a
400 ppmv CO limit. As such, we see little overlap between these rules, especially for process
heaters, although some FCCU CO boilers may be subject to both requirements. This CO limit
may overlap with the CO limit for certain FCCU CO boilers. The proposal for subpart Ja
specifies PS 4, while the Boiler MACT specifies PS 4a. We have revised the requirements for
33
-------
CO monitoring to allow the use of either PS 4 or 4a. Furthermore, we clarify in this response
that dual range analyzers are allowed and encouraged under this final rule. When dual range
analyzers are used, the low range of the analyzer should meet the requirements of this final rule.
Additionally, the zero and span gas checks of the low range are sufficient to satisfy the
performance specifications requirements of this rule.
4.4 Monitoring Exemptions and Other Miscellaneous Fuel Gas Comments
Comment: Commenters 0124, 0127, 0150, 0154, 0156, and 0159 supported the concept
that streams that are "inherently low" in sulfur should not require monitoring. Commenters 0154
and 0156 suggested that "inherently low" should be the same as in EPA's guidance for
alternative monitoring plans (AMP) (i.e., the average plus 3 standard deviations (o) is less than
81 ppmv). According to Commenters 0154 and 0156, the 5 ppmv maximum requirement is too
stringent. Commenter 0154 also suggested a one-time demonstration if the result is less than
30 ppmv. Commenter 0124 suggested that the rule also explicitly exempt gas streams with de
minimis annual sulfur flows, such as pilot gas, regardless of the sulfur concentration in the
stream. Commenter 0156 requested that petitions filed under this exemption be effective
immediately upon submittal rather than the date of approval.
Commenter 0130 disagreed with exempting fuel gas streams inherently low in sulfur
content from monitoring. Since the streams are subject to the concentration limits, monitoring is
necessary to determine compliance, enforce the limit, and verify EPA's expectation that the
stream will not exceed the limit. EPA has stated that without meaningful monitoring, the public,
government agencies, and facility officials are unable to assess compliance.
Response: The 5 ppmv H2S requirement was established because this level is
significantly less than the long-term H2S concentration compliance alternative. Unlike EPA's
guidance for alternative monitoring plans, the exemption provided does not rely on on-going
periodic monitoring. This lower 5 ppmv concentration limit is what provides the assurance
necessary to eliminate the on-going periodic monitoring. That is, setting the acceptance criteria
an order of magnitude below the actual concentration limit and requiring that this level be
demonstrated over 14 samples provides adequate assurance that the concentration limit is being
achieved on a continuous basis. We disagree with the commenter that an acceptance criterion of
81 ppmv or a one-time demonstration of less than 30 ppmv provides an adequate level of
34
-------
compliance assurance with the long-term 60 ppmv concentration limit. If a refinery owner or
operator cannot achieve the 5 ppmv limit, the refinery owner or operator may still elect to
request an alternative monitoring plan following EPA's guidance. The acceptance criteria for the
subpart Ja limit would be average + 3o < 30 ppmv (i.e., half the long-term concentration limit);
on-going periodic monitoring would then be required, as specified in EPA's guidance, to ensure
continuous compliance.
Comment: Commenters 0124, 0127, and 0154 supported the monitoring exemption for
fuel gases generated by process units that are intolerant of sulfur. Commenter 0159 stated that
an HF alkylation process unit should be included as a unit that is "intolerant" of sulfur.
Response: We reviewed data on HF alkylation process units, and we agree that this
process unit should be listed as a unit that is intolerant of sulfur.
Comment: Commenter 0156 requested that §60.105(b)(l)(ii) be modified to allow for
the installation of a blind or valve/car-seal to isolate inherently low sulfur lines from sour gas
lines, when these crossover points are used as emergency back-up only. Recordkeeping and
reporting requirements to explain times when the blind or seal is opened can be used to
demonstrate compliance.
Response: We agree that installation of a blind or valve/car-seal to isolate inherently low
sulfur lines from sour gas lines is adequate to eliminate fuel gas crossover. If the blind or seal is
opened, then monitoring of the fuel gas that is being combusted is needed to demonstrate
compliance with the standard. As such, the blind or seal can only be opened during times of
process upset or malfunction.
Comment: Commenters 0125, 0127, 0138, 0154, and 0159 supported the monitoring
exemption for commercial grade products. Commenters 0138 and 0154 stated that the
commercial grade product limit should be 60 ppmv (rather than 30 ppmv) since the methyl
mercaptan is less volatile than liquefied petroleum gas (LPG). Commenter 0154 also suggested
that direct measurement of vapor-phase concentrations with colored stain tubes is a better
compliance option than the liquid phase specifications. Commenter 0125 supported the
exemption for commercial grade products but requested clarification regarding whether the
commercial grade product specifications relate specifically to sulfur. For example, one refinery
uses a gas that does not meet the commercial grade standards for natural gas since its CO2
content exceeds the pipeline specifications, and EPA should clarify whether this refinery would
35
-------
have to monitor this gas stream even though the sulfur content does meet the pipeline
specifications for natural gas. Commenter 0159 recommended that EPA specifically list
commercial-grade gasoline, diesel fuel, jet fuel, LPG, and propylene as streams that are exempt
from monitoring.
Response: The commenters provided no data to suggest that commercial grade LPG does
not typically meet a 30 ppmv sulfur limit. Based on minimum required odorant additions, it
appears that methyl mercaptan concentrations should be well below the 30 ppmv sulfur limit.
Additionally, the relative vapor pressure of the components in LPG does not really impact the
concentration of the LPG fuel gas. LPG is only liquid under pressure; at lower pressures all of
the LPG constituents will be in the gas phase when fed to the fuel gas combustion device.
Additionally, the rule specifically applies to "gas streams" and it was intended that the
commercial grade specification would be evaluated in the gas phase {i.e., assuming that gas
stream was derived from the complete vaporization of the commercial product). We
intentionally did not list specific commercial-grade products such as gasoline, diesel fuel, jet
fuel, etc. because not all of these commercial grade fuels have to meet the low sulfur limits. It is
sufficient to state that the commercial grade product specification for sulfur must be 30 ppmv.
The commenter makes an interesting point about the high-C02 natural gas stream. To address
these comments as well as to clarify our intent regarding the commercial grade exemption, we
are revising the exemption paragraph as follows: "Gas streams that meet a commercial grade
product specification for sulfur content of 30 ppmv or less. In the case of an LPG product
specification in the pressurized liquid state, the gas phase sulfur content should be evaluated
assuming complete vaporization of the LPG and sulfur containing-compounds at the product
specification concentration."
Comment: Commenter 0125 suggested that EPA provide streamlined alternative
monitoring schemes or "mixing rules" to allow commercial supplemental fuels (natural gas,
propane, or butane) to be added to fuel gas without requiring a fuel gas monitor at each location
that mixing occurs.
Response: We agree with the commenter. If each individual gas stream meets the
concentration limit, then the mixture of gases will meet the concentration limit. Therefore, we
provide explicit language in this final rule at 40 CFR 60.104a(j)(4) to allow the refinery owner or
operator to demonstrate compliance on the mixture as used in the fuel gas combustion unit or by
36
-------
individual streams used in the fuel gas combustion unit. If all individual gas streams fed to a fuel
gas combustion unit meet the provisions of the subpart either by direct measurement or by
inherently low sulfur content provisions, then the gas mixture does not need to be measured in
order to demonstrate compliance at the fuel gas combustion unit. However, if any of the
individually-measured gas streams exceed the allowable concentration limit, then the owner or
operator must report an exceedance for that fuel gas combustion unit. In some cases, the refinery
owner or operator may avoid certain exceedance events by measuring the mixture. For example,
suppose commercial grade natural gas is used to supplement refinery fuel gas at a fuel gas
combustion unit. The new provision clarifies that the owner or operator may monitor only the
refinery fuel gas stream and not the natural gas/fuel gas mixture to demonstrate compliance.
However, if the 3-hour average refinery fuel gas stream H2S concentration exceeds 162 ppmv,
then the owner or operator must report an exceedance for that fuel gas combustion unit.
However, if the owner or operator monitors the fuel gas mixture entering the fuel gas combustion
unit, it may be possible that the fuel gas mixture has an H2S concentration less than 162 ppmv
(depending on the amount of natural gas used). Therefore, while we provide the option to
monitor individual fuel gas streams, the refinery owner or operator may still want to measure the
fuel gas mixture. If the refinery owner or operator elects to measure individual streams, then the
owner or operator must report an exceedance for the fuel gas combustion unit if any one of the
monitored streams exceeds the allowable concentration limit.
Comment: Commenter 0121 recommended that the H2S monitoring exemption for low-
sulfur gas streams be authorized with mandated records and testing rather than Administrator
approval. Detailed piping diagrams, flow and concentration ranges, descriptions of fuel gas
supply, and test data should be maintained on-site where they can be reviewed by an inspector.
The facility should also be required to maintain equipment to allow sampling of the fuel entering
the combustion device at any time. These records, along with the 5 ppmv maximum
concentration limit, should be sufficient for a facility to determine compliance without a review.
Response: We agree with the commenter and have revised the final standards to specify
that if an owner or operator follows the specifications for completing an application for an
exemption from monitoring the sulfur content of a low-sulfur fuel gas stream, the effective date
is the date the application is submitted.
37
-------
Comment: Commenter 0154 requested clarification that "Appendix F to Part 60, Quality
Assurance Procedures, is not a new requirement when compared to the existing CMS -
Continuous Monitoring System - requirements."
Response: We do not expect that the direct specification to Appendix F for the H2S
monitoring in §60.107a(a)(2)(iii) adds additional requirements to the H2S monitoring
requirements in subpart J because the NSPS General Provisions require . .if the continuous
monitoring system (CMS) is used to demonstrate compliance with emission limits on a
continuous basis, (the CMS is subject to) appendix F to this part, unless otherwise specified in an
applicable subpart or by the Administrator." As the H2S monitoring systems in subparts J and Ja
are being used to demonstrate compliance with an S02 standard, we expect that the QA
requirements in Appendix F (daily calibration drift assessments and quarterly data accuracy
assessments) are being performed on these continuous monitoring systems. To clarify that these
assessments are required for the H2S continuous monitoring systems, we specifically indicated
that the QA requirements in Appendix F are to be followed in proposed subpart Ja rule. We
maintain this same language in this final rule as it is our intent that these monitors follow, at a
minimum, the QA requirements in Appendix F.
38
-------
Chapter 5
FCCU AND FCU EMISSION STANDARDS
5.1 Fluid Catalytic Cracking Units
5.1.1 PM Emission Limits and Opacity
Comment: Commenter 0156 noted that the 1 kg/Mg PM emission limit did not appear in
either of the options under §60.102a(b). Commenter 0159 stated that, if EPA does require the
0.5 kg/Mg PM limit for modified and reconstructed units, EPA should include a conditional
exemption for cases "demonstrated to the satisfaction of the Administrator that compliance with
the standard is technologically or economically infeasible" similar to provisions in 40 CFR
60.283(a)(l)(iv).
Response: The co-proposal language in 40 CFR 60.102a(b) did include the correct
limits; the two options should have appeared as described in the preamble (72 FR 27190). As
explained in the preamble to the final standards, we are promulgating a 1 kg/Mg PM emission
limit for modified and reconstructed process units.
Comment: Commenter 0154 suggested that, if EPA pursues a 0.5 lb/1,000 lb coke burn
PM emission limit based on Method 5B or 5F, a new BDT analysis must be performed and a re-
proposal of Ja is needed to allow for public review and comment.
Response: The promulgated limit for new FCCU is 0.5 kg/Mg using Methods 5B and 5F,
and we disagree that another proposal is necessary. Any refiner commencing construction,
reconstruction or modification of a FCCU between proposal and promulgation should have
planned to meet the proposed standards, which were more stringent than the promulgated limits.
Therefore, no additional efforts should be needed to meet the promulgated limits.
Comment: Commenters 0138, 0150, 0154, and 0159 objected to the use of Method 5 in
subpart Ja. The commenters asserted that the proposed PM standards are based on very limited
emissions data, which do not necessarily reflect the industry as a whole. Commenters 0150,
0154, and 0159 stated that they could not locate data for an (electrostatic precipitator) ESP-
39
-------
controlled FCCU or FCU that was tested with Method 5 and achieved the proposed 0.5 kg/Mg
PM emission limit. The commenters concluded that the only basis for the proposed PM limit
was the low filterable PM emissions for "Source 88." Commenter 0159 noted that "Source 88"
uses a state-of-the-art best available control technology (BACT)-level wet gas scrubber (WGS)
that clearly does not represent the broader population of FCCU and control scenarios.
Commenters 0150 and 0154 noted that the sulfuric acid mist concentrations for "Source 88"
were higher than the filterable PM for many test runs, calling into question the test method
results EPA relied upon in its assessment. Commenter 0159 stated that the data set used to
support the standard is insufficient, and even though the average Method 5 is only 20 percent
more than Method 5B or 5F, the range is significant, with up to a 52 percent difference.
Commenter 0148 cited EPA's Fine Particle Implementation Rule (72 FR 20585), in
which EPA provided a transition period for developing emission limits for condensable PM2.5
until 2011, as rationale for not using EPA Method 5. Commenters 0150 and 0154 noted that
early tests conducted using Method 5 exhibited high PM emission concentrations but were
unsure if this was due to test method issues alone or if "changes in process operating conditions"
also contributed to the difference.
Commenters 0150, 0154, and 0159 suggested that EPA Method 5 is not appropriate for
refineries because the sampling line temperature for EPA Method 5 is in the critical zone for
sulfuric acid condensation, especially for streams with moisture content of less than 10 percent.
Commenters 0150 and 0154 provided calculations of the sulfuric acid that would be captured at
various sampling temperatures within the allowable range of EPA Method 5, assuming a
moisture content of 5 percent. The commenters cited a 1982 memo for allowing Test Method 5B
for NSPS D and Da (see also 44 FR 33580). Commenter 0156 stated that sulfuric acid in a wet
gas scrubber would approximately double the amount of PM captured for EPA Method 5 versus
Method 5B and expressed concerns that the condensing sulfates may cause blinding of the filters
using the 250°F sampling temperature of Method 5. Commenters 0150 and 0154 also stated that
EPA should not use vendor guarantees in its BDT determination but should consider the fact that
no vendors provide guarantees based on EPA Method 5.
Commenter 0170 provided results of a PM case study conducted on an FCCU meeting
0.8 lb/1000 lb coke burn based on Method 5F (as required by consent decree) with an ESP. The
study compared filterable PM measured using Method 5F to filterable PM measured using
40
-------
Method 5 to better understand the effects of temperature and sulfate content. The results show
that PM collected by Method 5 ranges from 6 to 16 percent higher than Method 5F; during one
test, that range resulted in a difference in measured filterable PM emissions ranging from 0.05 to
0.1 lb/1000 lb coke burn.
Commenter 0139 noted that one FCCU meets the 0.5 lb PM/1000 coke burn as
determined by EPA Method 5B using a huge two-ESP system, but even this control system could
not meet EPA's proposed total PM limit that includes condensable PM 100 percent of the time.
The commenter also stated that the incremental cost effectiveness of going from 0.5 lb filterable
PM to EPA's proposed standard would be beyond EPA's typical cost effectiveness cutoff for
NSPS and that EPA must provide a cost-effectiveness analysis to justify a tighter standard based
on Method 5.
Commenter 0138 did not know the performance of their recently installed FCCU control
systems in terms of EPA Method 5 and was, therefore, unsure if the new controls could meet the
proposed PM standards. Commenter 0138 stated that "basing the PM standard of FCCU on test
methods such as Method 5F that, by EPA's own admission, generates widely varying results is
neither sound nor reasonable and would lead to numerous non-compliant events." The
commenter also quoted EPA that "coordinating the test method with the pollutant defined by the
emissions limit is critical to an effective regulation."
Response: We note that the commenters did not appear to consider the PM data that were
presented in the background report for South Coast Air Quality Management District
(SCAQMD) Rule 1105.1 (Docket Item No. EPA-HQ-OAR-2007-0011-0031) in their critique of
the data. Based on data reported in this report, we concluded an ESP could meet a 0.5 kg
PM/Mg coke burn limit for filterable PM using test Method 5. The key test data are summarized
in Table G-l of the report; the methods used to generate the data in Table G-l are described in
Appendix E of the SCAQMD report. The recent performance test data for a wet scrubber tested
using the New Jersey method (tested twice, once at 200°F and once at 250°F), similarly suggest
that low levels of PM (i.e., 0.5 kg PM/Mg coke burn) can be achieved with a wet scrubber when
sampling at 250°F.
However, even though the data set is larger than the commenters thought, these data are
relatively limited and it is difficult for us to determine if the proposed limits could be achieved
by all configurations of FCCU, regardless of control device. We agree with the commenter that
41
-------
suggests that the PM test method should be based on the pollutant being controlled. We
endeavor to assess the true impact of the FCCU emissions on the ambient air quality, and we are
especially interested in the impacts of condensable PM. However, we note that Method 5
captures only a portion of those condensable emissions, and since the test data using Method 5
are limited, we have concluded that setting NSPS based on Method 5 is not appropriate at this
time. We are continuing to develop Method 202 to measure condensable PM and will evaluate
the appropriateness of including that method in future rulemaking efforts. (See the response to
the next comment for additional information on Method 202).
Comment: Commenters 0139, 0154, 0156, 0159, and 0174 stated that EPA should not
promulgate a rule based on condensable PM as measured by Method 202 because it is not a
reliable and repeatable method for measuring condensable PM from FCCU and FCU.
Commenters 0138, 0150, and 0154 stated that EPA has not accounted for the present levels of
condensable PM in the FCCU (sulfuric acid, poly cyclic organic matter (POM), and ammonia
salts) and noted that these condensable PM would all be measured using EPA Method 202.
Commenters 0150 and 0154 cited the positive bias in Method 202 due to S02 adsorption and the
limited test data as reasons that condensable PM should not be included. Commenters 0138,
0154, and 0159 also noted that EPA's "average" of 0.5 kg/Mg of coke burn-off (for condensable
PM) is statistically meaningless based on the data (although Commenter 0159 provided data to
suggest condensable PM from one of their FCCU WGS was 0.27 kg/Mg). Commenter 0159
provided source test data for one refinery where the filterable PM by Method 5B was 0.83 kg/Mg
and the total PM emissions by Method 5B and Method 202 combined was 1.1 kg/Mg. The
commenters agreed to work with EPA to develop an appropriate technique for measuring
condensable PM from FCCU and FCU but opposed the inclusion of condensable PM within the
subpart Ja PM emission limit. Commenter 0174 noted that the API Stationary Source Emission
Task Force is considering how best to characterize condensable PM emissions and will discuss
their plans with EPA in the future.
Commenter 0170 stated that particulate matter test methods that include condensable PM
are sensitive and complex, and EPA should not include condensable PM in subpart J. The
commenter provided results of a PM case study conducted on an FCCU meeting 0.8 lb/1000 lb
coke burn based on Method 5F (as required by consent decree) with an ESP. The study
compared condensable PM measured using Method 202 to condensable PM measured using
42
-------
conditional draft Method 20x; the study also compared the results obtained by two different
contractors using Method 202 to assess the method repeatability. The results suggest that
Method 20x has better data repeatability than Method 202. The commenter also stated that
Method 20x appears to correct for "artifact pseudo-particulate formation of absorbed sulfates and
nitrates in the impinger water train" and noted that Method 20x uses indirect cooling, which
causes a controlled condensation that is more representative of atmospheric conditions than the
water train used by Method 202. In addition, the commenter stated that there was significant
variation between the condensable PM measured by the two different testing contractors using
Method 202. Based on these results, the commenter expressed concern over how condensable
PM could possibly be measured reliably enough to ensure compliance with a standard.
Therefore, the commenter opposed any FCCU PM standard including condensable PM. If EPA
does proceed with setting a limit on condensable PM, the commenter urged EPA not to limit the
test method to Method 202 but to allow for the approval and use of other test methods that are
shown to be more accurate and repeatable, such as Method 20x.
Commenter 0121 agreed that it is important to consider both condensable and non-
condensable (filterable) PM when evaluating the performance of new, modified, or reconstructed
FCCU. The commenter noted that the Texas Commission on Environmental Quality (TCEQ)
has considered total PM (filterable and condensable) in their BACT determinations for FCCU in
Texas. The commenter recommended that a total PM limit (filterable and condensable) be set at
1 kg/Mg coke burn and encouraged EPA to continue to improve Method 202. Commenter 0146
recommended incorporation of EPA Reference Method 202 in conjunction with Reference
Method 5. In the commenter's experience, stack test data from FCCU and FCU show that
approximately 5 to 10 percent of total PM is condensable PM.
Response: While we agree with the commenters on the importance of condensable PM,
there are some concerns regarding the current condensable test method (Method 202).
Additionally, we have limited data by which to assess the performance of the BDT with respect
to condensable PM. We are currently working with stakeholders, including industry trade
organizations, to develop a suitable test method for condensable PM. Once a reliable test
method is developed for the condensable PM, we will obtain performance data for the industry,
evaluate the condensable PM emissions from FCCU, and assess alternative PM limits that
43
-------
include condensable PM. However, we are not including a limit for condensable PM in this final
rule.
Comment: Commenter 0152 disagreed with EPA's assertion that a 0.15 kg/Mg coke
burn limit would need an ESP with ammonia (NH3) injection and would result in a much lower
total PM reduction. There is no technical justification that ESP are the only choice of control
equipment to achieve this level and that NH3 injection is a must when ESP are chosen. The
commenter stated that: (1) ESP manufacturers indicate ESP built to specifications do not need to
rely on NH3; (2) wet scrubber manufacturers indicate they can achieve levels close to
0.15 kg/Mg coke burn; and (3) EPA is not correct in assuming condensable PM formed by NH3
injection will form condensable PM with EPA Method 5 because ammonium sulfate starts
condensing below 200°F and will not be captured by EPA Method 5, which captures
condensable PM between 250 and 320°F.
Commenter agreed that Option 5 in Table 2 should be rejected; it results in higher PM
and SO2 emissions than Option 4 because ammonia injection may increase production of
condensable PM as it improves control of filterable PM. However, the commenter also stated
that EPA overestimated the emission reduction for PM because EPA based reductions on both
filterable and condensable PM.
Response: Although it may be possible to design an ESP to meet the 0.15 kg PM/Mg
coke burn emission limit without the use of ammonia injection, ammonia injection is commonly
used. Our assessment of this option relied heavily on the data from the background document
for the SCAQMD PM standard. Specifically, Table G-l indicates that there were three ESP that
were at or below the 0.15 kg PM/Mg coke burn PM limit. Responses in Appendix K indicate
that all three of these refineries were injecting ammonia at the time. The data indicate that the
sulfate and condensable PM values for these units were very high (ranging from 1.82 to
4.23 kg/Mg coke burn). The one facility that did not use ammonia injection had emissions of
only 0.08 kg/Mg coke burn of sulfate and condensable PM. Thus, while this facility could not
meet the 0.15 kg PM/Mg coke burn limit, its total PM emissions, including sulfate and
condensable PM, were a factor of 4 to 8 times lower than the units that were meeting the 0.15 kg
filterable PM/Mg coke burn limit. In other words, this facility could meet a total 0.5 kg/Mg PM
limit with lower emissions of total PM. This facility could reduce its filterable PM emissions by
44
-------
using ammonia injection, which would be a much more economical alternative to building
additional ESP collection area, but condensable PM would likely increase.
The test methods used to generate the data in Table G-l are described in Appendix E of
the SCAQMD report; the filterable data represent AQMD Method 5.2, which maintains the filter
portion of the probe at 200°F. As such, we expected that AQMD Method 5.2 would provide
higher PM filterable concentrations than EPA Method 5. Nonetheless, five of the six ESP tested
met the 0.5 kg PM/Mg coke burn limit, and three of the six ESP tested met the 0.15 kg PM/Mg
coke burn limit using AQMD Method 5.2. However, upon further review, we do note that
AQMD Method 5.2 does allow evaluation and subtraction of sulfate particulate, so that AQMD
Method 5.2 may be more similar to Method 5F than originally considered. This certainly calls
into question the achievability of the 0.15 kg/Mg PM standard when using EPA Method 5.
We have no test data for any wet scrubber achieving a 0.15 kg PM/Mg coke burn limit,
and there is a significant difference between achieving "levels close to" 0.15 kg PM/Mg coke
burn and meeting this limit at all times. Furthermore, when we assessed the impacts of the
control options at proposal, we included projected impacts on condensable PM. We considered
condensable PM to be the PM collected in the impingers of "back-half' of the sampling train.
As the impingers that are maintained in an ice water bath, we consider all of the ammonia sulfate
as condensable PM (i.e., condensable PM is anything that condenses between the sampling train
temperature (250 or 320°F) and the impinger exhaust temperature (approximately 60°F)). Based
on the lack of any wet scrubbers meeting a 0.15 kg/Mg coke burn PM limit and the data
presented in Table G-l, we concluded that: (1) ESP would be required to meet this limit; (2)
ammonia injection would likely be used to meet this limit; and (3) this ammonia injection would
likely increase the overall PM emissions from FCCU (as compared with other regulatory options
being considered). The commenter provided no additional data to support that ESP as operated
by refinery owners and operators are achieving this 0.15 kg/Mg coke burn PM limit without
ammonia injection. We therefore maintain that this 0.15 kg/Mg coke burn PM limit based on
Method 5B and 5F is not BDT.
In response to Commenter 0154, we note that only by considering the impact of
condensable PM can we conclude that a PM limit of 0.15 kg/Mg coke burn is not desirable; if we
consider only filterable emissions, then this option appears to have more merit. Although the
limits that we have analyzed and concluded to be BDT do not include condensable PM, we drew
45
-------
that conclusion by considering all PM. We strove to assess the true impact of the regulatory
alternatives on the ambient air quality in our analysis, and we believe that includes
acknowledging the effect of the options considered on the total PM emissions.
Comment: Commenters 0150 and 0154 stated that EPA made an incorrect assertion that
wet scrubbers are generally more effective than ESP in controlling condensable PM. The
commenter quoted several reports noting that the removal efficiency of wet scrubbers at
removing very fine PM (submicron particles) is relatively poor.
Response: As condensable PM are not included in this final rule, the following response
is somewhat moot and academic but is provided to address the comment received. While we
recognize that wet scrubbers are not necessarily the best control system for removing fine PM,
the commenters did not appear to consider the differences in operating temperatures between
ESP and wet scrubbers as employed in the petroleum refining industry for FCCU control. The
commenter's letter includes the following statement: "Existing particulate control systems do
not effectively remove condensable aerosols because the aerosol precursors are often in the
vapor state when they pass through the control device. Although the vapors usually condense in
a wet scrubber, they often form ultrafine particles which are very difficult to capture" (emphasis
added). That is exactly the point: ESP are expected to be ineffective on gaseous aerosols due to
the hot operating temperatures; wet scrubbers generally condense these aerosols. While these
condensed aerosols "often" form ultrafine PM, they do not always; some of the condensed
aerosols agglomerate and are efficiently removed in the wet scrubber. Even if all of the
condensed aerosols were PM fine and "difficult" to capture, this does not imply the removal
efficiency is zero for the wet scrubber. However, at typical FCCU ESP operating temperatures
of 500°F, these aerosols are gaseous and the ESP would be expected to have a very low, if not
zero, percent removal efficiency for these gaseous aerosols. If ESP were operated at the same
lower temperature as wet scrubbers, it is quite likely that they could be as or more efficient than
wet scrubbers at removing this fine PM. Therefore, the context of this statement is important.
We did not mean to imply that wet scrubbers were more effective at reducing condensable PM
emissions in every application. However, based on the manner in which these control systems
are operated in the petroleum refining industry for FCCU emission control, we maintain that
FCCU wet scrubbers are expected to effect better condensable PM control than FCCU ESP.
46
-------
Furthermore, in the refining industry, ammonia is often injected upstream of the ESP to
"condition" the gas stream and improve PM control. While this conditioning may improve the
removal of PM that exists in the gas stream at 500°F, this practice can exacerbate the amount of
condensable PM formed in the gas stream (or in the atmosphere) once the gas cools to ambient
conditions. As discussed previously, the available test data suggest the condensable PM
emissions from ESP that use ammonia conditioning are much higher than ESP that do not. This
is another reason why wet scrubbers used to control the FCCU emissions are generally expected
to have lower condensable emissions than a comparable FCCU ESP.
Comment: Commenter 0161 requested clarification of whether ammonia sulfate is to be
included in the determination of PM. In some cases, ammonia sulfate may have been excluded
from the PM test data EPA used. The commenter noted that ammonia is often injected to help
control PM (in ESP) and NOx emissions, and stated that, at the least, the mass of ammonia in the
ammonia sulfate PM collected should not be included in the reportable PM emissions.
Response: The final standards in subpart Ja are based on the use of Method 5B and 5F.
We will consider methods to measure and reduce PM not measured by these methods at a future
time.
Comment: Commenter 0150 stated that an adequate BDT analysis must include a
showing that, at effective cost, units can meet the proposed PM limit concurrently while meeting
the new NOx limit, and that this should include a cross-section of possible operating
configurations (e.g., partial-burn versus full-burn regeneration, hydrotreated feed, and other feed
quality variables). Commenters 0138, 0150, and 0154 stated that EPA has not considered the
increased levels of condensable PM that will result from the proposed NOx control requirements.
According to the commenters, the use of SCR or selective non-catalytic reduction (SNCR) to
meet the NOx limit will increase the quantities of ammonia salts. Commenter 0154 stated that an
ammonia slip of 10 ppmv would increase condensable PM by approximately 0.11 kg/Mg coke
burned. The commenters also noted that use of oxygen enrichment may increase sulfuric acid
formation. Commenter 0159 stated that EPA failed to account for the higher PM emissions that
typically result from higher SOx additive use. For all these reasons, the commenter contended
that the BDT determinations are deficient.
Response: In developing the revised impacts for this final rule, we accounted for cross
pollutant impacts to the extent practicable. As described in additional detail regarding the FCCU
47
-------
NOx emissions limit, the NOx emission limit was selected after considering the secondary
impacts {i.e., PM formation from additional electricity generation) of the various options. The
commenters did not provide any additional data to assess the anticipated increase in PM
emissions. Furthermore, as we are not promulgating a PM condensable standard in this final
rule, we do not expect that the NOx controls will have a significant impact on the required
performance of the PM control system to meet both the PM and NOx emission limits. The
commenter did not provide any data by which to assess the magnitude or the validity of the claim
that addition of more SOx additives will increase the PM emissions. Based on the advances in
the performance of SOx additives, use of SOx additives are not expected to significantly impact
PM emissions. Referring again to the Table G-l in the background document for the SCAQMD
Rule 1105.1, there were three facilities that used catalyst additives (as indicated in Appendix K)
and all of these three could meet a 0.5 kg/Mg filterable PM limit.
Comment: Commenter 0154 stated that the PM standard should be 1 kg/Mg (not
including condensable PM) because this is the limit established by the top-performing units in
the Refinery MACT II standard.
Response: At the time the Refinery MACT II standard was being developed, there were
no refineries subject to a 0.5 kg/Mg PM emission limit, so the MACT floor was a PM limit of
1 kg/Mg coke burn for both new and existing sources. Furthermore, the Refinery MACT II
standard is a HAP standard and PM was used as a surrogate for the metal HAP emissions. As
sulfuric acid mist and other condensable PM are not HAP and not correlated with the metal HAP
emissions, it was appropriate to use Methods 5B or 5F and not to include condensable PM in the
surrogate metal HAP PM measurement. Also, due to the low concentrations of metal HAP on
the PM that was emitted, it was not cost-effective to require additional metal HAP emission
control. For example, assuming that the incremental cost effectiveness for a 0.5 kg/Mg PM
emission limit is $5,000/ton of PM reduced, the average cost-effectiveness in terms of metal
HAP reduction would be approximately $3.5-million/ton of metal HAP reduced. As the NSPS
are criteria pollutant standards, it is reasonable to expect that different conclusions are reached
when evaluating the cost-effectiveness of the control system based on different pollutant types
and more recent performance data.
Comment: Commenters 0150 and 0154 noted that there could be difficulty achieving the
Performance Specification (PS)-l 1 requirements due to significant differences in condensable
48
-------
PM present in the stack at typical ESP exhaust temperatures and the cooler sampling
temperatures used in EPA Method 5, making it difficult to use a PM CEMS.
Response: As indicated in previous responses, we recognize that the level of condensable
PM present is strongly influenced by the control device operating temperature and the PM
sampling temperature. As such, we agree with the commenter that there may be significant
issues in demonstrating the performance of the PM CEMS with the PS-11 requirements as
proposed. In efforts address this issue and to promote the use of PM CEMS, we have modified
the PS-11 requirements to allow PM demonstration using EPA Method 5, 51, or 17. Thus, if a
refinery owner or operator elects to install a PM CEMS, then on-going compliance can be
demonstrated using EPA Method 17 instead of EPA Method 5.
Comment: Commenters 0138, 0150, 0154, and 0159 objected to the single "model plant"
approach used in EPA's cost analysis. This model plant approach does not realistically consider
important factors such as the inherent sulfur content of the feed, partial-burn versus full-burn
regeneration, and FCCU/regenerator size. Commenters 0138, 0150, and 0154 asserted that the
purchased equipment costs, which are critical to the overall cost estimates, are underestimated.
Costs should not be escalated from estimates that are 20 to 30 years old; rather, new cost
estimates should be solicited.
Commenters 0138, 0148, 0154, 0156 and 0159 provided estimates of costs and emission
reductions for several actual projects. These data indicate that EPA's costs are significantly
underestimated and that the proposed standards are much less cost-effective than presented by
EPA. Commenter 0154 provided data to show that actual FCCU WGS projects had cost-
effectiveness values ranging from $7,000 to $50,000 per ton of PM removed (not including co-
control of SO2); FCCU ESP projects ranged from $10,000 to $222,000 per ton PM removed.
Commenter 0148 provided cost estimates for a new ESP that is projected to cost $300-million in
2011 and achieve only 168 tons/yr PM reduction. Commenter 0156 recently installed new WGS
at a total capital cost of $100-million to meet a 0.5 kg/Mg PM emission limit based on EPA
Method 5B. The commenter indicated that additional upgrades on these new scrubber systems
would be needed to meet the 0.5 kg/Mg emission limit based on EPA Method 5 at a cost in
excess of $50,000/ton of PM removed. Commenter 0159 estimated that costs of complying with
the 0.5 kg/Mg PM limit based on EPA Method 5 would range from $43,600 to $127,000 per ton
of PM removed.
49
-------
Response: As explained throughout this section, we completely recalculated the cost and
emission reduction impacts of the FCCU PM and SO2 standards to consider all existing FCCU.
Therefore, the commenter's concerns about the "model plant" approach should no longer be an
issue. Although much of the control costs were developed using costs from old cost functions
which were escalated to 2005 dollars, this is a common and accepted costing practice.
Furthermore, we did solicit a wet scrubber vendor quote for the model FCCU plant in developing
the proposed rule. The projected purchased equipment costs (prior to tax and shipping) escalated
to 2005 dollars using our costing algorithm were within 10 percent of the vendor's free on board
(FOB) estimates. This assessment verifies the appropriateness of the wet scrubber costing
algorithm. In the impacts estimated for this final rule, we did increase our auxiliary equipment
estimates so that our wet scrubber FOB estimates exactly match the FOB estimates provided by
the vendor.
In reviewing our ESP costs, we recognized that we did not include costs for ancillary
equipment in our purchased equipment costs. Thus, although we generally disagree with the
commenter regarding the use of escalation factors, we did conclude that the ESP costs were
significantly underestimated at proposal.
For this final rule, we did assess the costs of the PM emission reductions independently
from (or ignoring the) SO2 emission reductions. Based on our revised equipment cost estimates
and nationwide impact assessment, we determined that a 0.5 kg/Mg coke burn PM emission limit
based on Method 5 would achieve nationwide emission reductions of 810 tons of PM at a cost of
$23,000/ton of PM reduced for existing FCCU (compared to maintaining 1.0 kg/Mg coke burn
PM emission limit based on Method 5B or 5F). For new FCCU, the proposed emission limit
would achieve nationwide emission reductions of 300 tons of PM at a cost of $6,700/ton of PM
reduced. For both data availability and cost reasons, we determined that the 0.5 kg/Mg coke
burn PM emission limit based on Method 5 is not BDT (more details are provided in the
preamble to the final standards).
Comment: Commenters 0138 and 0154 indicated that EPA "did not account for
deterioration of the process equipment (e.g., FCCU regenerator catalyst cyclones) and the
associated PM emissions increase that typically occurs between process turnarounds."
Commenter 0154 suggested that meeting the proposed limit would require more frequent
turnarounds, greatly increasing the cost of the proposed rule as well as limiting energy
50
-------
production. Similarly, Commenter 0150 suggested that EPA must demonstrate the ability of
units to meet the limit over the duration of the process and pollutant controls' turnaround cycle,
which will strongly influence the cost-effectiveness calculation. Commenter 0159 quoted the
original NSPS background document (EPA 450/2-74-003) that stated that PM emissions
"generally increase about 35 percent during the run (between turnarounds.
Response: We reviewed Background Information for New Source Performance
Standards: Asphalt Concrete Plants, Petroleum Refineries, Storage Vessels, Secondary Lead
Smelters and Refineries, Brass or Bronze Ingot Production Plants, Iron and Steel Plants, Sewage
Treatment Plants; Volume 3, Promulgated Standards (Refinery NSPS BID), the 1974
background document referenced by the commenter.2 There is significant scatter in the data in
Figure 4-3 on Page 32. Although the linear regression of the data suggests a slight upward slope,
one could not reject the hypothesis that the slope of the line is zero. That is, the referenced
background document does not irrefutably suggest that all units exhibit a deterioration in control
performance. If the performance of the FCCU cyclones does deteriorate, this deterioration
would result primarily in more large particles entering the WGS or ESP. These larger particles
should be removed from the exhaust stream at very high efficiencies by the downstream WGS or
ESP. Nonetheless, although no data are provided regarding when turnarounds occurred, the data
presented in Tables G-l and G-2 of the background document for SCAQMD Rule 1105.1,
especially for Facility F, appears to support this deterioration in performance. Also, FCCU run
lengths have increased from approximately 2 years back in 1974 to approximately 5 years. This
additional run length could further exacerbate the deterioration effect, although it is also likely
that the increases in technology that afford the longer run lengths are also subject to less
deterioration (else they could not sustain the additional run length).
We anticipated that the maintenance and monitoring requirements of the emission control
systems would prevent significant deterioration in the control system performance. However, we
received several adverse comments suggesting that the operating parameters were not good
indicators of performance. Consequently, we are increasing the required frequency of
performance tests (for units not employing a PM CEMS) to annual demonstrations.
2 U.S. Environmental Protection Agency. 1974. Background Information for New Source Performance Standards:
Asphalt Concrete Plants, Petroleum Refineries, Storage Vessels, Secondary Lead Smelters and Refineries, Brass or
Bronze Ingot Production Plants, Iron and Steel Plants, Sewage Treatment Plants; Volume 3, Promulgated
Standards. February 1974. EPA 450/2-74-003 (APTD-1352c). Docket ID No. EPA-HQ-OAR-2007-0011-0082.
51
-------
We also conclude that, if a refinery owner or operator anticipates that the FCCU control
system is prone to deterioration, the control system can be "over" designed to meet the PM limit
at the end of the run. The commenters provided no recent data to support the "deterioration in
performance" theory in current FCCU operations. Subsequent tests at the wet scrubber in New
Jersey indicated that performance was consistent or improved, but these were only 8 months
apart. Based on the lack of additional data provided by the commenters, we maintain that the
proposed PM emission limits are achievable. By requiring annual performance demonstrations,
refinery owners and operators will be able to better assess their control systems' on-going
performance and adjust their maintenance schedule to ensure continuous compliance with this
final rule.
Comment: Commenters 0138, 0154, and 0159 stated that EPA's impact analysis contains
several errors and discrepancies. For example, the commenters suggested that EPA double-
counted the PM2.5 fraction. Commenters 0138 and 0154 stated that EPA used electricity costs
that are in 2004 dollars while all other costs are in 2005 dollars. EPA also stated that labor
estimates for operating and maintenance of air pollution control devices are based on EPA's Air
Pollution Control Manual, but the labor estimates provided in EPA's analysis are not consistent
with the referenced manual. Additionally, Commenter 0138 stated the baseline emissions and
emission reductions were qualitatively estimated from the test data rather than being statistically
derived. These discrepancies raise concerns regarding the overall accuracy of the estimated
impacts. Commenter 0154 stated that EPA did not provide adequate supporting documentation
for key assumptions: (1) input values not referenced (e.g., PM emission factors; costs for
CEMS); (2) spreadsheet calculations not provided; and (3) referenced data not applied correctly
(e.g., meeting minutes indicate four new DCU; EPA assumed five).
Commenters 0138 and 0154 stated that EPA overestimated the number and types of
affected sources in the fifth year after promulgation. The assumption of 15 refineries' worth of
processes is over-simplified and unrealistic. The proportion of new versus
modified/reconstructed is unrealistic with no supporting information for the basis of the estimate.
The commenters believe the number of affected sources is inflated and they believe that the
inflated number of sources causes the proposed standards to appear more cost-effective than they
really are.
52
-------
Response: In general the issues raised by these comments do not change the conclusions
drawn from the analysis results in any way, but we recognize that these analyses are complicated
and difficult to recreate. For example, the commenters alleged that the PM2.5 fraction was
double-counted, which was not the case, but we acknowledge that the column labels in Appendix
B of Data and Assumptions used in the Impacts Analysis for PM and SO 2 Emissions from Fluid
Catalytic Cracking Units and Fluid Coking Units were not clear. There were two columns in the
table used to account for filterable PM, one labeled PMIO-FIL and one labeled PM25-FIL. In
the analysis, PMIO-FIL represented the PM that was between 10 and 2.5 micrometers ([j,m) and
PM25-FIL represented PM that was 2.5 [j,m or less; the labels. Had PMIO-FIL actually included
the PM2.5 fraction, then the PMIO-FIL emissions in tons/yr would have to be, in all cases, greater
than the PM25-FIL emissions; looking at the values in Appendix B, it is apparent that the PM2.5
fraction was not double-counted. For the final analysis, every table in which these or similar
column headings appear includes a footnote that explains the actual meaning of the headings.
We specifically indicated that the electricity costs were in 2004 dollars because, at the
time that the impact estimates were calculated for the proposed rule, average annual electricity
costs were not available for 2005. Electricity and labor costs for 2006, the most recent year now
available, have been included in the revised impact estimate for this final rule. Based on the
relative contribution of the electricity costs, the lack of 2005 electricity costs would not have
been expected to significantly change the impact estimates. Had the 2005 electricity costs been
available at proposal, the revision would have increased the total annualized costs for ESP by
less than 1 percent and increased the total annualized costs for wet scrubbers by less than
0.5 percent. These differences are completely in the noise of the overall costing assessment.
As explained in the impacts memorandum, the labor rates themselves were taken from
the May 2005 National Industry-Specific Occupational Employment and Wage Estimates
published by the Bureau of Labor Statistics. What was not as clear was how those labor rates
were used; the costing algorithm (including labor rates) for venturi-type wet scrubbers was
developed using Handbook: Control Technologies for Hazardous Air Pollutants, which was
included in the docket.3 Costs for CEMS were estimated from EPA's Cost Model.4 We
3 U.S. Environmental Protection Agency. 1991. Handbook: Control Technologies for Hazardous Air Pollutants.
June 1991. EPA/625/6-91/014. Docket ID No. EPA-HQ-OAR-2007-0011-0167.
4 U.S. Environmental Protection Agency, Emissions Measurement Center. 1998. CEMS Cost Model, ver. 3.0, Build
#41. http://www.epa.gov/ttn/emc/cem.html
53
-------
acknowledge that we could have included more extensive documentation for some of the values
and procedures we used to develop the impacts, but most were generally documented at least
minimally in the impacts memorandum or attachments or were simply unit conversions from
allowed emission rates. We have made a conscious effort to correct these deficiencies for the
final analyses.
With respect to the number of DCU, the memorandum actually indicates that there would
be six new coking units, four of which were specifically labeled in the Oil and Gas Journal as
DCU and two that were not specifically designated. This agrees completely with our assumption
of three new refineries' worth of units per year over the next 5 years and our process unit counts,
which yields a total of six new, reconstructed, or modified coking units per year. We assumed
that one of the affected units would be a fluid coking unit, leaving five affected DCU, which is in
perfect agreement with the cited memorandum. Furthermore, the reported number of coking unit
construction projects in the Oil and Gas Journal supports the assumptions used to estimate the
number of processes impacted. The assumption that new processes units would reflect existing
refinery processes was based on input from industry representatives; the number of new versus
modified and reconstructed units was based on industry-reported expansion plans. Although
criticizing the assumptions, the commenter provided no better basis for estimating the number of
newly affected units.
Comment: According to Commenters 0138 and 0154, since the consent decrees typically
focused on larger refineries, the potential emission reductions will occur primarily at smaller
uncontrolled sources. As the control costs associated with smaller processes do not decline as
linearly with size as do emissions, the cost-effectiveness of the controls would be worse if EPA
had properly accounted for the size of the uncontrolled sources, according to the commenter.
Commenter 0138 stated that EPA's proposed PM standard for FCCU under subpart Ja is not
cost-effective. The commenter provided site-specific engineering cost estimates to indicate that
the PM controls are much less cost-effective than EPA estimates. The commenter strongly
supported EPA's co-proposal that modified and reconstructed FCCU comply with the existing
subpart J standards for PM (1 lb/1,000 lb coke burn using EPA Methods 5B or 5F).
Response: In our assessment for the proposed rule, we attempted to account for the
general consent decree requirements. As indicated previously, we completely revamped our
analysis for assessing the impacts of FCCU regulatory alternatives in response to this and other
54
-------
comments. For this final rule, we used an FCCU-specific model that takes into account each
specific FCCU's size, control configuration, and consent decree requirements to develop impacts
for modified and reconstructed FCCU. We also revised our costing analysis for new FCCU.
These revisions led us to conclude that BDT is slightly different than proposed. Modified and
reconstructed units will comply with the PM emission limit currently provided in subpart J. New
FCCU will comply with a PM emission limit of 0.5 kg/Mg coke burn based on Method 5B and
5F. Based on our revised analysis, this standard will reduce PM emissions nationwide by
240 tons/yr at a cost of $5,600 per ton of PM reduced.
Comment: Commenters 0129, 0138, and 0154 stated that the 7 percent interest rate used
in the impacts analysis is inappropriate for industry. The commenters suggested a minimum
interest rate of 10 percent or even 30 to 40 percent used by industry when evaluating new
projects. The commenters also suggested that the equipment lifetimes used for some controls are
over-estimated. Specifically, wet scrubber lifetimes of 15 years should be used rather than
20 years and SCR catalyst lifetimes of 2 to 3 years should be used rather than 5 years.
Response: We are primarily attempting to assess the cost of capital (i.e., if a loan was
required). We acknowledge that interest rates vary, but the 7 percent annual interest rate is our
best estimate for long-term cost of capital. As some wet scrubbers have been in service for
approximately 20 years in FCCU regenerator vent service, we maintain that 20 year lifetime for
wet scrubbers is warranted for this analysis. Although some operational difficulties have been
observed in SCR FCCU applications, as the more experience is achieved using SCR, we
anticipate 5-year catalyst lifetimes can be achieved. Nonetheless, we have revised our cost
estimates for SCR to use a 2-year catalyst lifetime. Given the NOx emission limit established in
this final rule, we do not anticipate any refineries using an SCR, so that this adjustment does not
significantly alter the cost-effectiveness of the NOx emission controls for this final rule.
Comment: Commenters 0138 and 0154 asserted that EPA underestimated the operating
and maintenance labor hours and used inappropriate and out-dated (2005) labor costs that do not
reflect the labor rate increases that resulted after Hurricane Katrina. Commenter 0159 also noted
that price escalation after Hurricane Katrina has been significant (63 to 75 percent based on two
cases cited by the commenter); therefore, they added a 30 percent escalation factor to the costs
that they provided.
55
-------
Response: We reviewed labor rates for the petroleum refining industry for 2006 as
published by the Bureau of Labor Statistics. While labor rates may have jumped immediately
after Hurricane Katrina, this jump appears to have been temporary in nature. The nationwide
average labor rates for production (SOC Code 51-0000) and maintenance (SOC Code 49-0000)
workers in the petroleum and coal product manufacturing industry increased by only 3 to
4 percent from 2005 to 2006. Equipment costs were escalated using the Chemical Engineering
Plant Cost Index (CEPCI); the CEPCI indicated a 6.7 percent cost increase from 2005 to 2006.
Marshall and Swift provides sector-specific equipment cost indices on a quarterly basis. From
second quarter 2005 to second quarter 2006, the Marshall and Swift Equipment Cost Index
(MSECI) for the petroleum product industry increased by 4.9 percent. From third quarter 2005
to third quarter 2006, the MSECI for the petroleum product industry increased by 6.7 percent.
Although details are lacking regarding the time period over which the commenter's price
increases actually occurred, it is common for major expansion projects to take several years. We
note that from 2003 to 2006, the CEPCI increased approximately 25 percent. At the time the
cost analysis was developed for the proposed rule, 2005 data were the most recent annual
average data available. The data now available for 2006 do not indicate a significant escalation
in labor or equipment costs, as suggested by the commenters. Nonetheless, since average 2006
values are now available, we have revised our cost impacts to average 2006 dollars in response
to these comments.
Comment: Commenter 0154 stated that EPA did not fully consider the costs of
wastewater treatment and the potential issues associated with soluble salts, such as sodium
sulfate, in the WGS effluent and National Pollutant Discharge Elimination System (NPDES)
discharge permit requirements.
Response: We recognize that due to state and local wastewater discharge permitting
issues, some significant additional costs may be incurred by some refineries related to the
treatment or regeneration of wet scrubber wastewater. At proposal, we assumed that refineries
faced with high wastewater disposal costs would elect to use catalyst additives and an ESP or
alternate control system, such as a spray dry adsorber and/or baghouse system. Based on
information provided by the commenter, the quantity of catalyst additives needed to comply with
the SO2 emission limit when using high sulfur feed impacts the overall conversion efficiency of
the FCCU, which greatly impacts the economics of this compliance option. However, other
56
-------
alternative compliance options are available for SO2 removal, such as spray dry adsorption, that
can be used to comply with the emission limits in this final rule. We evaluated the costs of a
spray dry adsorption/baghouse system to comply with the combined S02 and PM limits. The
total capital costs of this system are approximately double the costs of a conventional FCCU wet
scrubber (compared to a factor of 10 times higher for a regenerative wet scrubber system). The
total annualized costs of the spray dry adsorption/baghouse system were also approximately
double the costs of a conventional FCCU wet scrubber. While these cost increases are
significant, they are relatively small compared to a installing a regenerative wet scrubber system.
We specifically evaluated the costs of a regenerative wet scrubber system for the FCU because a
wet scrubber is expected to be the only demonstrated control system.
Comment: Commenter 0154 suggested that, since a larger wet gas scrubber is needed to
meet the PM emission limit, the cost of the WGS (or at least the incremental cost of the larger
WGS) should be assigned solely to PM removal. By combining the PM and SO2 emission
reductions, EPA is underestimating the cost-effectiveness of the PM controls.
Response: As the type and performance of the control system selected depends to a large
extent on the emission limits established for both PM and SO2, we considered these factors in
total when estimating the impacts of the proposed rule. Although there are still some cross-
pollutant impacts, we attempted to segregate the costs of removing PM separately from SO2, in
our revised impact estimates. While the incremental cost of controlling PM emissions is higher
than the cost of S02 removal, the incremental PM emission reduction is still cost-effective,
especially considering 83 percent of the incremental PM reduction is fine PM.
Comment: Commenters 0125, 0150, 0154, and 0161 supported the elimination of the
opacity requirement in subpart Ja. Commenters 0125, 0150, and 0154 requested that this change
be extended to subpart J so that FCCU controlled with WGS do not need to apply for Alternative
Monitoring Plans. Commenter 0125 also suggested an exclusion from opacity monitoring
"during times of unusual ambient humidity, which can interfere with the proper operation of
such devices." Commenter 0150 and 0161 stated that the elimination of the opacity requirement
could be contingent on implementation of continuous parameter monitoring system (CPMS) or
direct PM monitoring according to the requirement in either subpart Ja or Refinery MACT II.
57
-------
Commenter 0130 stated that an opacity limit is still necessary along with PM CEMS in
subpart Ja. Since the total PM limit should be 0.5 kg/Mg coke burn, the opacity limit should be
15 percent, or half of the current 30 percent opacity in subpart J.
Response: As wet scrubber control systems were expected to be the primary control
system used to comply with the combined SO2 and PM requirements for subpart Ja and as
continuous opacity monitoring system (COMS) are not appropriate for wet flue gas streams, we
evaluated alternative continuous compliance options for the FCCU regenerator vent. We
concluded that operating parameter monitoring along with periodic performance tests would
provide as an effective means of demonstrating continuous compliance as a COMS at lower cost
for dry flue gas stacks and avoid the technical issues of wet flue gas stacks. We did not modify
the opacity requirements in subpart J because we anticipated existing NSPS units would have
already installed a COMS or applied for and received an alternative monitoring plan. A PM
CEMS is expected to be much more accurate for demonstrating compliance with the PM limit
than a COMS. We did include the use of COMS for systems controlled using cyclones as no
operating parameter would provide as good an indication of the performance of the control
system; the actual opacity operating limit is established based on the results of the site-specific
performance test.
5.1.2 NOx Emission Limit
Comment: Commenter 0159 estimated the cost-effectiveness of meeting a 20 or 40
ppmv NOx limit to be in the range of $28,000 to $125,000 per ton of NOx removed.
Commenter 0154 suggested that LoTOx units have limited operating experience;
therefore, their performance has not been demonstrated across a range of FCCU over a long
period of time, and LoTOx units should not be considered in the BDT analysis.
Commenter 0174 stated that FCC regenerator rebuilds are not likely to provide
significant NOx reductions beyond those achieved by catalyst additives. These rebuilds are also
difficult and costly. Therefore, the commenter stated that regenerator rebuilds are not cost-
effective (especially when looking at the incremental reductions beyond catalyst additives) and
cannot be determined to represent BDT.
According to Commenter 0146, most FCCU and FCU covered by recent consent
agreements are equipped with WGS for PM and SO2 control, and these controls are compatible
58
-------
with cost-effective, demonstrated NOx controls such as LoTOx and WGS Plus technologies that
achieve 85 to 95 percent reduction with cost effectiveness values of $4,000 to $7,000 per ton
(estimated based on 7 percent interest rate and 15 year equipment life, but given that some units
are over 50 years old, a longer equipment life could be used that would further lower the costs).
Commenters 0146 and 0149 stated that FCCU and FCU are typically very large NOx-emitting
sources that will remain unaddressed by other rulemaking efforts; Commenter 0146 cited rules
such as electric generating unit (EGU) multi-pollutant regulation and Clean Air Interstate Rule
(CAIR) and noted that FCCU and FCU will remain some of the largest NOx emitting sources in
the country following the implementation of these rules to address NOx emissions from other
sources.
Response: With respect to costs for existing units, special APCD (such as an SCR) are
not required for most (if any) FCCU to meet the proposed emission limit of 80 ppmv NOx-
There are limited retrofit issues with the use of non-platinum oxidation promoters, advanced
oxidation controls, LNB in CO boilers (if applicable), and SNCR. Additionally, several recent
wet scrubber installations have been equipped for easy inclusion of LoTOx system, so LoTOx
can also be added with minimal retrofit issues for some units. Furthermore, in estimating the
impacts of the NOx emissions limit for existing units, we included a significant retrofit cost
factor in our analysis to account for the added cost of installing these systems on existing units.
We disagree with commenters that suggest that we should ignore the LoTOx system in
our BDT analysis. We did not select any one technology as BDT; we identified a suite of
potential controls that could be used in different circumstances. Initial performance evaluations
showed very high removal efficiencies for the LoTOx system. While we recognize that the
technology has had limited application, the available data certainly suggest that an 80 ppmv limit
can be easily achieved by the system. While we may agree that we should not determine that the
LoTOx system is the sole BDT, it is not logical to ignore this technology as a viable control
option, especially for units with existing compatible wet scrubber control systems.
In re-evaluating the impacts for FCCU NOx controls, we specifically quantified the
secondary impacts of APCD. In addition to the direct PM impacts of SNCR and SCR, SCR and
LoTOx units require additional electrical consumption, and we calculated the secondary PM,
SO2, and NOx emission impacts of the additional electrical consumption. The cost-effectiveness
when looking only at the primary NOx emissions of an emission limit of 20 ppmv was
59
-------
$5,800/ton of NOx removed for new units and $6,800/ton of NOx removed for modified and
reconstructed units, and this control option had considerable secondary impacts. We also
evaluated higher NOx emission limits. A control option based on an 80 ppmv NOx emission
limit provided cost-effective NOx control with little to no secondary impacts. Therefore, as
discussed in further detail in the preamble to the final rule, the 80 ppmv NOx emission limit was
determined to be BDT.
Comment: Commenter 0150 noted that the span requirement for 02 CEMS used in
conjunction with NOx CEMS should be 10 percent rather than the 25 percent specified in the
proposed rule. Also, the commenter requested that the QA requirement to meet Procedure 1,
Appendix F be deleted in §60.105a(e)(5) because the requirement (in Appendix F) is for SO2, not
NOx. Commenter 0154 noted that there is a typographical error in §60.105a(e)(5). The section
refers to "SO2" several times, but it should refer to "NOx"
Response: As discussed previously, we agree that a lower O2 span is acceptable and we
have revised the span requirements for O2 CEMS to allow the refinery owners and operators
flexibility to set the 02 span at values between 10 and 25 percent, inclusive. We disagree that
Procedure 1 is not applicable; Procedure 1 is specifically applicable to NOx CEMS. However,
we do acknowledge an editorial error in that §60.105a(e)(5) incorrectly referenced SO2 monitors
rather than NOx monitors; the same error also appeared in §60.107a(c)(5). We have revised
these paragraphs to correctly refer to NOx monitors, as originally intended.
5.1.3 SO2 Emission Limit
Comment: Commenter 0159 stated that EPA grossly underestimated the costs and relied
on inflated SO2 emission reductions to support the cost-effectiveness of the proposed standards.
According to the commenter, EPA estimated that modified or reconstructed FCCU could meet
the proposed limit with existing controls whereas the commenter estimated that four of their
seven FCCU would need supplemental pollution control technology to meet the 0.5 kg/Mg PM
emission limit. Also, based on Commenter 0159's non-WGS FCCU, which all meet the
50 ppmv limit, the emission reduction achieved by implementing a 25 ppmv limit is on the order
of 20 tons/yr rather than 500 tons/yr. Commenter 0138 asserted that the EPA did not account for
the fact that 76 percent of FCCU are subject to consent decrees and would already be meeting
the proposed 25 ppmv SO2 emission limit. According to the commenter, EPA's baseline
60
-------
calculation for SO2 assumes that all modified and reconstructed units were only subject to J and
not to the 25 ppmv limit typical of the consent decree requirements. Commenters 0138 and 0159
stated that the baseline S02 emissions from the 7.2 modified or reconstructed units should be
3,415 tons/yr, not 4,560 tons/yr.
Response: After reviewing the proposal analyses, we acknowledge that there was an
inadvertent error in the baseline SO2 emission estimates. Although we set up the baseline
analysis to account for reconstructed units that are subject to consent decree requirements and are
already required to meet the 25 ppmv emission limit, we applied the incorrect emission factor to
these units. Although this error did significantly overestimate the proposed rule's anticipated
SO2 emissions reductions from FCCU, the proposed SO2 emission limits for FCCU were still
cost-effective when the error was corrected. The incremental cost-effectiveness of the more
stringent SO2 limit should have been $700/ton of SO2 removed rather than the $220/ton reported
in Table 4 of the preamble to the proposed rule. Regardless, this error does not affect the
analysis for the final standards because the analysis used at proposal has been significantly
revised in response to these and other comments. To estimate the impacts of this final rule, we
used an FCCU-specific model that takes into account each specific FCCU's size, control
configuration, and consent decree requirements. It is important to note that, while some FCCU
with ESP meet the 50 ppmv emission limit, there are non-consent decree FCCU with ESP that do
not meet that limit. Moreover, subpart J does not require that level of performance, so there are
some units that have much higher emission reductions from baseline (subpart J) to the 25 ppmv
SO2 emission limit imposed by this final rule. We also recognize that the actual emission
reductions are subject to the size of these higher emitting units. By using the FCCU-specific
assessment, we can more appropriately account for these variables.
Comment: Commenters 0150, 0154 and 0159 provided data to suggest that the retrofits
of existing sources are not cost effective. The average cost-effectiveness of the reported projects
by Commenter 0154 was $2,400/ton, and cost-effectiveness values ranged from $70 to
$l,000/ton for catalyst additives and from $500 to $8,000/ton for WGS. Commenter 0159
suggested that, if catalyst additives cannot be used, retrofitting an existing FCCU with a WGS
would have costs in excess of $100,000 per ton SO2 removed.
Response: Based on the majority of the costs and SO2 emission reductions submitted by
the commenters, we conclude that the 50 ppmv/25 ppmv SO2 emission limits are cost-effective.
61
-------
Although higher than our estimates, an average cost-effectiveness of $2,400/ton of SO2 removed
confirms our original conclusion that the tighter SO2 emission limits are BDT. While we
recognize that some selected projects will have higher than average costs, the average nationwide
costs by both our analysis and the data provided by the comments support the adoption of the
more stringent SO2 emission limits. As all of the projects for which costs were submitted were
for existing FCCU, we conclude that the tighter SO2 emission limits are BDT for modifications
and reconstructions as well as for new construction. As such, we reject the co-proposed option
to allow existing (modified or reconstructed) FCCU to meet only the subpart J standards {i.e.,
any of the three options allowed under subpart Ja).
Comment: Commenters 0125 and 0171 stated that FCCU with no APCD installed should
be allowed to meet the 50 ppmv compliance option currently provided in subpart J to units with a
control device.
Response: We agree. The direct SO2 concentration limit is easily more stringent than the
other alternative compliance options based on the graph in the background document to the
proposed S02 standards for FCCU, and it is more directly and continuously measured using an
S02 CEMS.
Comment: Commenter 0159 noted that FCCU feed hydrotreating may be an integral part
of the SO2 compliance strategy. In such cases, the SO2 limit may be exceeded when the
hydrotreater is down. Therefore, short-term exceptions to the 50 ppmv 7-day average SO2 limit
should be provided during times of FCCU feed hydrotreater start-up, shutdown, and malfunction
to avoid costly control measures, such as a redundant hydrotreater or a WGS retrofit. Facilities
could prepare a Hydrotreater Outage Plan outlining FCCU SO2 minimization steps during
hydrotreater outages.
Response: We would like to encourage the use of hydrotreaters as a "pretreatment"
option for FCCU feed streams. FCCU feed hydrotreaters can be used to effectively reduce SO2
emissions, as well as metal HAP emissions, from the FCCU regenerator with minimal secondary
environmental impacts. The FCCU feed hydrotreater also generally increases the yield of the
FCCU, and reduced coke make in the FCCU, both of which improve the efficiency and overall
environmental performance of the refinery. However, regardless of the method chosen, the
FCCU must demonstrate continuous compliance with the SO2 emission limit. Therefore, we
have not provided specific provisions for FCCU that pretreat their fresh feed.
62
-------
Comment: Commenter 0154 suggested that the rule should include provisions to use
either a CEMS or a CPMS to demonstrate compliance with the SO2 standard, similar to the
FCCU PM compliance options.
Response: As most FCCU have already installed an SO2 CEMS, the SO2 CEMS
requirement has minimal additional cost implications. Furthermore, SO2 CEMS are needed to
adequately demonstrate compliance with both the 7-day 50 ppmv SO2 emission limit and the
annual 25 ppmv S02 emission limit.
Comment: Commenters 0125, 0150, and 0154 requested that EPA include a monitoring
option in subpart J §60.105(h)(6) to allow SO2 CEMS data to be used in lieu of daily Method 8
testing for demonstrating compliance with the emission standard in §60.104(b)(2). This request
is commensurate with several EPA-approved alternative monitoring plans. Improved
understanding of the SO2 catalyst additives demonstrates that sulfur trioxide (SO3) emissions are
very small and SO2 monitoring adequately demonstrates compliance.
Response: The relative fraction of SO2 and SO3 in the FCCU regenerator exhaust
depends both on the combustion conditions in the FCCU regenerator and on the type of catalyst
additive used. Based on data collected in the development of the original 9.8 kg/Mg emission
limit, up to 60 percent of the SOx emissions could be SO3. The commenters provided no
specific test data to support the assertion that the SO3 emissions are "very small" in all cases. In
order to employ a higher SO2 monitoring level, the refinery owner and operator will have to
develop a site-specific adjustment factor for use in converting the S02 CEMS concentration data
to a total SOx concentration and this request should be made on an individual basis. Essentially,
daily sampling is required following the procedures in §60.106(i) for 21 days. The 7-day
average ratio of SO2 to total SOx emissions are to be calculated for each consecutive 7-day
interval of the 21 day sampling period. The lowest S02 to total SOx emissions ratio is to be used
to convert the 9.8 kg/Mg SOx emission limit to an SO2 emission limit for the SO2 CEMS.
Comment: Commenter 0121 noted that the SO2 concentration standard is stricter for
units using oxygen enrichment and requested clarification whether that was EPA's intent.
Response: No, that was not our intent. We had originally considered a normalizing the
allowable emissions by the process throughput. However, as the SO2 (and NOx) CEMS measure
concentration directly and an emission limit that is normalized by throughput would require
stack flow rate monitors as well as process throughput recordings, the direct concentration
63
-------
measurement was selected for the proposed rule. According to our calculations, a 50 ppmv SO2
emission rate for an air-supplied FCCU regenerator equates to an SO2 emission rate of
1.43 kg/Mg coke burn and a 25 ppmv S02 limit equates to 0.715 kg/Mg coke burn; we have
included these alternative emission limits in the final standards. A similar issue is expected with
NOx, so we have also included appropriate alternative NOx limits in terms of the coke burn rate
specifically for complete combustion regenerators that use oxygen enrichment.
5.1.4 CO Emissions Limit
Comment: Commenter 0154 supported the 500 ppmv CO emission limit but indicated
that the standard should not be changed to 0 percent excess air. The commenter also suggested
that the rule should include provisions to use either a CEMS or a CPMS to demonstrate
compliance with the CO standard, similar to the FCCU PM compliance options.
Commenter 0125 similarly indicated that the 0 percent excess air correction of the CO limit may
be significant and that EPA did not appear to evaluate the impacts of this change.
Commenters 0150, 0154, and 0159 recommended that the 1-hour averaging time for CO
be increased to daily averages (midnight to midnight) because non-platinum combustion
promoters used to reduce NOx are not as efficient at reducing CO, leading to increased CO
emission variability. Commenter 0150 stated that the longer averaging time should also be
provided in subpart J for FCCU using additive NOx control.
Similarly, Commenter 0170 supported the proposed CO limit for subpart Ja but
recommended that the averaging time be extended to a 24-hour rolling average. The commenter
stated that the extension is needed in both subparts J and Ja to provide flexibility to minimize
NOx emissions. The commenter noted that the combination of NOx catalyst additives and
palladium-based CO promoters (instead of platinum-based CO promoters) greatly reduces NOx
emissions but can cause greater variation in CO emissions. Two of the commenter's refineries
using a Grace XNOx™ additive experienced an increase in CO and had to request a modification
of the state CO limits in their permits. The commenter stated that extending the averaging time
will allow process set points to be established at levels that will achieve greater NOx reductions
with only occasional short-term spikes in CO concentrations that will only cause negligible
increases in CO emissions. The commenter also noted that no technical justification for the 1-
hour averaging period was found in the supporting documentation for the original subpart J.
64
-------
EPA stated in the preamble to the proposed subpart J amendments that limiting NOx emissions is
a higher priority than limiting CO emissions, and the commenter's suggestion helps to achieve
the higher priority goal. The commenter noted that the current 1-hour averaging time means an
operator must make quick changes to process operating conditions if the CO spikes, which can
cause longer-lasting changes to NOx levels. According to the commenter, a longer averaging
time would allow an operator to make gradual process changes to bring CO under control while
maintaining the NOx values at lower concentrations.
Commenter 0149 recommended a limit of 300 ppmv over a 1-hour average for CO for
this NSPS as this is demonstrated and achievable in consent decrees. Commenter 0130 stated
that EPA should adopt the Boiler MACT CO limit for liquid and gas-fired boilers of 400 ppmv.
These CO standards are clearly feasible and address the kind of criteria pollutants that the NSPS
is meant to cover.
Commenter 0130 disagreed with EPA's decision to retain the existing CO emissions limit
of 500 ppmv because using complete combustion catalyst regeneration likely results in higher
NOx emissions. EPA has not sufficiently demonstrated or explained the existence, likelihood, or
extent of the trade-off of NOx emissions and has not adequately justified its failure to improve
upon the existing CO limit of subpart J.
Response: We acknowledge that lower CO emission limits are achievable, but we
weighed the trade-off between NOx and CO emission control and determined that lower CO
emission limits would not result in additional overall environmental benefit when considering the
additional NOx formation. Lower CO emission concentrations are achieved by higher
combustion temperatures and higher excess oxygen concentrations, exactly the combination of
factors that lead to NOx formation. As discussed previously, we selected a NOx emission limit
that was achievable for most FCCU using non-platinum catalyst additives and oxidation controls
in part due to the secondary PM formation associated with some of the more advanced NOx
emission control systems. We maintain that proper combustion control is BDT for CO emissions
and an important control option in the suite of controls determined to be BDT for NOx control.
Therefore, it is important to consider the impacts of the likely NOx combustion controls on the
CO emissions.
As the residence time of the gases within the regenerator (and/or CO boiler, if applicable)
is on the order of seconds, a proper combustion control feedback loop should easily be able to
65
-------
adjust the air blower rate, oxygen enrichment rate (if used), and other process parameters to
maintain proper CO control on an hourly average basis. All of the data in the technical record
suggest that a 500 ppmv hourly average CO limit is achievable when using non-platinum
oxidation promoters. Additionally, the commenters' statement that they had to a request a
modification of their state CO limits in their permits is misleading. Both of the state permits in
question had CO limits that were lower than 500 ppmv, and the permits after modifications were
still more stringent than the 500 ppmv CO limit in subpart J. In fact, these refineries were still
required to comply with the Refinery MACT 2 rule, which requires CO concentrations of 500
ppmv or less on a 1-hour average. No commenter has provided any data to indicate that the 500
ppmv 1-hour CO emission limit cannot be met while complying with consent decree NOx limits.
The oxygen correction factor was included to minimize any air dilution effects. Most
complete combustion FCCU operate with O2 concentrations of approximately 0.5 to 1.0 percent,
the lower end of this range is expected to be used in most cases minimize NOx formation.
Partial burn FCCU that use CO boilers typically operate at between 2 and 6 percent outlet O2
concentrations. Again, when considering the co-control of NOx, newNSPS units are expected to
operate at the lower-end of this range. Therefore, the O2 correction factor is expected to be a 2 to
10 percent adjustment on the CO emission limit. Based on the available data, we maintain that
proper combustion control is BDT for CO emissions and that the BDT can achieve an hourly
average emission limit of 500 ppmv CO on a dry basis, corrected to 0 percent excess air, when
considering the co-control of other pollutants from the FCCU.
Additionally, we maintain that the 500 ppmv limit in this final rule is at least as stringent
as the requirements in the Boiler MACT. For gaseous fuels, Boiler MACT requires a 400 ppmv
CO limit corrected to 3 percent oxygen; an equivalent limit corrected to 0 percent oxygen is
467 ppmv. Moreover, the 400 ppmv is assessed on a 30-day average. As seen previously, this is
likely to be a much less stringent standard in terms of CO emissions than a 500 ppmv hourly
average limit. We note that in Boiler MACT, CO is a surrogate for organic HAP emissions. As
discussed in the response to comments in the development of the Refinery MACT II, CO is only
a good surrogate for organic HAP emission reduction down to a certain point. As the Boiler
MACT was using CO as a surrogate for organic HAP emissions, the 400 ppmv 30-day average
limit was appropriate. However, for the direct control of CO emissions, we maintain that the
hourly average emission limit that has been in-place for more than two decades is most
66
-------
appropriate. We expect that the hourly average limit of 500 ppmv will ensure that a much lower
CO emission limit is achieved on a longer-term (24-hour or 30-day average) basis. Therefore,
simply lengthening the CO averaging time is expected to increase the overall mass emissions.
As described previously in assessing BDT for NOx, we attempted to account for secondary
impacts on other pollutants. We concluded that maintaining the 500 ppmv CO emission limit
(rather than lowering this limit) and achieving an 80 ppmv NOx emissions limit provides the
optimal environmental solution.
Comment: Commenter 0154 supported the monitoring exemption for units that
demonstrate average CO emissions of less than 50 ppmv. The commenter recommended that the
averaging interval for the demonstration be 30 days as in subpart J; the commenter further
recommended that the monitoring exemption be valid from the date of submittal rather than the
data of approval (consistent with the sulfur monitoring exemptions in the consent decrees).
Response: The commenter appears to interpret the requirements in subpart J differently
than we do. As the CO emission limit in subpart J is an hourly average limit, the demonstration
of the "average CO emissions" being less than 50 ppmv is somewhat ambiguous when a 30-day
performance demonstration is required. Based on the required span of the CO CEMS when
performing the 30 day demonstration, concentrations exceeding 100 ppmv may not be accurately
measured. If a 30-day average was truly intended, a dual range monitor would be needed to
accurately measure any short-term hourly excursions to fully verify that the 30-day average
emissions were 50 ppmv or less. For example, if a unit typically averages 25 ppmv CO, it could
have 18 hourly averages in 30 days that average 1000 ppmv and still meet the 50 ppmv limit
averaged over the 30-day demonstration. However, a CEMS with a span of 100 ppmv cannot
accurately measure these perturbations, and a demonstration of a 30-day average 50 ppmv would
certainly not justify automatic approval as being compliant with the 500 ppmv hourly average
CO emission limit. Through this preamble, we clarify that the averaging time of the CO
emissions during the 30-day demonstration was intended to be 1-hour averages in subpart J.
However, we are not revising the language for this demonstration in subpart J. In subpart J,
approval of the written request for an exemption is required. This approval is needed if a small
number of 1-hour average CO emissions exceed 50 ppmv dry basis over 30-days, but, to the
satisfaction of the Administrator, the refinery has adequately demonstrated that the FCCU
achieves an hourly average CO emission limit of 50 ppmv or less. Approval of the written
67
-------
request is also needed because no other parameter monitoring is specifically required under
subpart J following the demonstration, but the Administrator may wish to subject the approval of
the request upon some measure of the system to ensure the on-going performance of the
combustion controls is maintained.
To clarify our intent in subpart Ja, we are adding the word "hourly" preceding the
. .average CO emissions.. ."in §60.105a(h)(3). Also, as we include specific CPMS
requirements for demonstrating on-going performance is similar to the performance during the
30-day demonstration period, Administrator approval is not needed for this purpose. We also
note that, if the hourly average CO limit of 50 ppmv or less is achieved during each and every
hour over the 30-day demonstration period, Administrator approval is not necessary.
Consequently, we have added paragraph (h)(3)(iii) to this section to describe that the effective
data of the alternative CPMS monitoring exemption is the date of submittal for demonstrations
where all hourly average CO concentrations are 50 ppmv (dry basis) or less.
Comment: Commenter 0154 suggested that CO generated from firing auxiliary fuel in
the CO boiler should be subtracted from the CO measured to the atmosphere in determining
compliance with the FCCU CO standard because the CO boiler is not part of the affected facility
under subpart J.
Response: This exclusion is not necessary. First, we have defined the FCCU in
subpart Ja to include the CO boiler and other systems used for waste heat recovery, so that the
emissions from these activities are specifically included in the subpart Ja applicability. Second,
with the auxiliary fuel is also auxiliary air, so the concentration-based CO limit already provides
additional mass emissions of CO attributable to the combustion of auxiliary fuel. Finally,
complete CO destruction is easier to achieve in these post-combustion systems (than in complete
combustion FCCU) because the CO boiler does not have the same operating temperature
constraint as in the FCCU regenerator, where catalyst activity may be lost at higher temperatures.
As such, we maintain that BDT is proper combustion control of these systems and that the BDT
can achieve an hourly average emission limit of 500 ppmv CO on a dry basis, corrected to 0
percent excess air.
68
-------
5.1.5 Operating Parameter Limits
Comment: Commenter 0131 requested that FCCU under subpart J be allowed to use
CPMS just as units subject to subpart Ja.
Response: We did not modify the opacity requirements in subpart J because we
anticipated existing NSPS units would have already installed a COMS or applied and received
approval for an alternative monitoring plan. Source owners or operators are already allowed by
the NSPS General Provisions to request alternative monitoring procedures, such as using CPMS
allowed under subpart Ja for sources subject to subpart J. Source owners or operators who wish
to make such as request should follow the process given in 40 CFR 60.13(i).
Comment: Commenter 0150 stated that there are too many variables affecting coke burn-
off rates for the refinery owner/operator to ensure that the maximum coke burn-off rate would
occur during the performance test. As such, EPA should provide a 10 percent allowance or a
factor based on the ratio of the emission limit and performance test results before reporting is
required in the semi-annual reports. This flexibility is also warranted because the daily coke
burn-off rate does not directly correspond to the NSPS emission limit.
Response: We agree with the commenter that coke make is dependent on a great number
of factors, including crude type, catalyst type and additives used, feed rates, and cracking
temperature and pressure. We recognize that not all of these factors are readily controlled, so it
may be impossible to achieve a maximum coke burn-off rate during the initial performance test.
However, the face velocity of the gas within the ESP is a direct function of coke burn rate and
the performance of the ESP is strongly affected by the face velocity. As such, the coke burn-off
rate is a key parameter to monitor, record, and maintain below certain levels. As indicated in
§60.102a(c), we desire that the operating limits not be exceeded on an hourly average basis (we
note that we failed to specify the averaging time for the coke burn-off rate in this and other
sections; we have clarified this by directly specifying the averaging times in this final rule).
However, an exceedance of the hourly coke-burn operating limit is not included in the definition
of excess emissions in §60.105a(h) and is not reportable as an excess emission event in the semi-
annual reports. Records of the coke burn-off rate are required to be kept as specified in
§60.108a(c)(4), but no reporting of the coke burn-off rates is required.
Comment: Commenter 0154 supported the idea that compliance with the PM emission
limit can be demonstrated by either a CEMS or a CPMS. On the other hand, Commenter 0150
69
-------
stated that PM CEMS have not been adequately demonstrated on FCCU, so they do not consider
PM CEMS to be a valid compliance option.
Commenter 0161 stated that many refiners have had to install COMS as part of their
consent decree. The facilities should only have to use a single method for determining
compliance with the FCCU PM limit. Either EPA should rescind the consent decree requirement
for COMS or allow facilities that have opacity monitors a means of demonstrating compliance
using the existing COMS. The commenter also stated that the correlation between other
parameters and the formation of PM is not proven and that surrogate indicators should only be
allowed if the facility makes an equivalency determination.
Response: Although PM CEMS have not been used in the petroleum refining industry,
there are a limited number of PM CEMS operating at electric utilities. Through this rulemaking,
we seek to encourage the use of PM CEMS, as these systems provide a more direct measure of
the actual PM emissions on a continuous basis than operating limits or opacity monitors.
We understand that refiners that are required to use COMS by their consent decree or
State and local agencies would have somewhat duplicative monitoring provisions. However, we
expect that control device operating parameters will provide a better indication of performance
than a COMS for most control systems. We have included a means of establishing an opacity
operating limit for FCCU equipped with a cyclone as no control device operating parameter is a
better indicator of performance for this type of control device. The opacity operating limit will
be determined on a site-specific basis based on the results of the initial performance test. The
alternative COMS operating limit is applicable only to units meeting the limit with cyclones. For
ESP, operating parameters are expected to provide a better indication of control device
performance than a COMS. A COMS is not appropriate for wet scrubber stacks, and continuous
bag leak detectors are preferred to COMS for baghouse control systems.
We note that opacity is a condition, not a pollutant, and opacity has been used as an
indicator of proper control device operation. While it may be possible for a source owner or
operator to develop correlations between opacity and PM emissions, extensive performance
testing - much beyond that required by PS-11 for PM CEMS - would be necessary and such
testing would require many modes of source operation. If the owner or operator chooses not to
use an instrumental approach such as a PM CEMS or bag leak detection system, the owner or
operator could conduct a series of ongoing emissions testing, perhaps daily or weekly.
70
-------
Comment: Regarding the bag leak detection system monitoring requirements,
Commenter 0150 stated that baghouses are not "demonstrated technologies" for FCCU; the
commenter did not know of an FCCU that is controlled using a baghouse. The commenter
asserted that other "assumptions" made in the preamble about appropriate control technologies
for PM and SO2 are not, but must be, demonstrated with performance test data.
Commenter 0159 (the one refiner in the US that operates a baghouse) stated that alleviating the
cause of a bag leak detection system alarm within 3 hours is unreasonable and recommended that
EPA remove this requirement in §60.105a(c)(3) in favor of "action-specific alleviation times to
incorporated into the required site-specific monitoring plans."
Response: In response to the first commenter, we note that there is a refinery that
operates a baghouse control system for the FCCU vent. Baghouses are well-demonstrated
control systems for other similar exhaust streams and one has been in operation at a refinery for
nearly twenty years. The performance achieved by baghouse control systems in similar
continuous service clearly indicates that the proposed emission limits are achievable using a
baghouse control system. Nonetheless, performance data for this refinery baghouse were
requested and the data received indicate that this system can meet the PM emission limit in this
final rule.
Given that a bag leak detection system provides a continuous signal proportional to the
amount of particulate matter it sees, an alarm above a certain set point would be expected to
occur instantaneously. Such quick detection of a problem would allow the source owner or
operator ample time to begin correcting the problem using techniques such as compartment
isolation and bag replacement or process shutdown.
Comment: Commenters 0125 and 0154 requested that the hourly average operating
limits for CPMS be changed to daily averages to be consistent with Refinery MACT II and to
minimize short-term perturbations (e.g., during rapping events). Commenters 0150 and 0154
disagreed with setting the operating limit based on the lowest hourly average value measured
over the three test runs. Instead, the commenters recommended that the operating limit be set as
the average value of the three runs. According to the commenters, using the lowest hourly
average would require the unit to operate below its demonstrated capacity, resulting in lost
production.
71
-------
Commenter 0127 agreed that if parametric monitoring is used, limits must be placed on
the parameters. The commenter requested additional clarification on the procedures used to
approve a performance test and the parameters set during the test. According to the commenter,
§60.104a should include procedures that ensure the permitting authority or EPA Administrator
approves the parameter limits and should require the set limits to remain in-place until a
subsequent performance test is conducted and new parameter limits are approved.
Commenters 0148, 0150, and 0154 addressed the wet scrubber operating parameters.
Commenter 0148 noted during development of Refinery MACT II, EPA agreed that a CEMS or
CPMS can be problematic for WGS, and it is important that the alternative compliance options in
Refinery MACT II be included in subpart Ja. The commenters noted that pressure drop is not a
good indicator for jet ejector type scrubbers and suggested deleting the average pressure drop
limit for non-venturi jet ejector WGS consistent with Refinery MACT II. Commenters 0150
and 0154 suggested that the liquid-to-gas ratio is not a sensitive indicator of performance in the
range of 4 to 20 gallons per 1,000 actual cubic feet per minute (acfm), and above 20 gallons per
1,000 acfm, PM control efficiency can actually decrease in some types of scrubbers. The
commenters suggested monitoring just the liquid recirculation rate (not the liquid-to-gas ratio).
Commenter 0148 recommended that other operating parameters be consistent with those in
Refinery MACT II, specifically: (1) the language for wet scrubber total liquid flow rate; (2)
provisions to calculate exhaust gas flow rate; and (3) not requiring Performance Specification 3
(PS-3) for 02 monitors. In addition, Commenter 0156 stated that the CPMS in §60.102a(c)(2)
for wet scrubbers should be consistent with current refinery AMP and with applicability
determinations issued by EPA Region VI. According to the commenter, these provisions allow
the owner/operator to multiply the average pressure drop or liquid-to-gas flow rate determined
during the source test by 70 percent to develop the operating limits. This factor is needed to
account for differences in these parameters at lower operating rates.
Commenters 0150 and 0154 stated that CPMS operating parameters selected for ESP are
incorrect. According to the commenters, the ESP used in refinery applications often have
multiple fields in series; each field has its own secondary voltage. The secondary voltage in one
specific field can drop without a significant change in PM emissions if the secondary voltage in
the other fields remains high. The commenters also suggested that, while decreases in power
input typically result in higher PM emissions for high resistivity particles, the inverse
72
-------
relationship may be observed for low resistivity particles.5 The commenters suggested that EPA
should review the ESP data to determine if power input is a reasonable indicator of performance,
and, if so, include only total power input as the monitoring parameter (dropping the secondary
voltage monitoring requirement). Commenter 0125 considered the CPMS requirements to be
overly restrictive and suggested that EPA incorporate language similar to the following: "If the
daily average ESP total power input falls below the level measured in the most recent source test
which demonstrated compliance with the emission limit, a source test shall be performed within
90 days at the new minimum daily average ESP total power level."
Response: First, we recognize that it is difficult to achieve worst-case hourly average
operating limits during the performance tests and that the operating limits are only indicators of
the performance of the control device and are not direct measurements of emissions. As such,
we intentionally defined excess emissions in §60.105a(h) as the 24-hour periods in which the
average operating parameter exceeded the operating limit. That is, reportable exceedances are
only those in which the 24-hour average operating parameter falls below the operating limit
derived from the performance test. By providing a longer averaging time to determine if
operating limit exceedances must be reported, we have provided a small allowance for
occasional hourly excursions without having to report each hourly excursion as an excess
emissions event, which addresses several of the commenter's concerns.
Second, we are a bit surprised at the comments regarding using the lowest hourly average
value from the performance tests as the operating limit. The proposed rule acknowledged that
there are variances in the hourly operating characteristics of the control system and attempted to
provide the owner or operator an appropriate level of leniency in establishing the operating
parameters. At proposal, we saw no reason to further limit the range of appropriate operating
parameters by averaging the other test run values in order to establish the operating limit, as long
as the owner or operator selected the operating limit based on compliant test runs. For example,
suppose the liquid-to-gas ratio (LGR) for a wet scrubber was 6, 8, and 10 gallons per 1,000 acfm
for the three performance test runs. Assuming all of the test runs were compliant, the proposed
rule would allow the operating limit to be set at an LGR of 6 gallons per 1,000 acfm rather than
at 8 gallons per 1,000 acfm, which would be the average of the 3 runs. However, the final
5 Richards, J. 2000. "Control of Particulate Matter Emissions." U.S. EPA Air Pollution Training Institute Course
413, Student Manual. August 2000.
73
-------
standards include a requirement to set the operating limit as the average value of the three runs to
be more consistent with the NSPS General Provisions and the public comments.
The commenters also questioned the usefulness of many of the operating limits. We
identified three possible continuous compliance alternatives: 1) a PM CEMS; 2) parameter
operating limits; and 3) opacity monitoring. We selected parameter operating limits because PM
CEMS have not been demonstrated on refinery emission sources and opacity monitors cannot be
used with wet scrubber control systems. Additionally, the 30 percent opacity limit shows poor
correlation with control system performance. Therefore, we selected parameter operating limits
as the best means of demonstrating continuous compliance.
We maintain that liquid-to-gas ratio is the most appropriate operating parameter for wet
scrubbers. Additionally, establishing a minimum liquid-to-gas ratio should provide greater
flexibility to the owner or operator when FCCU processing rates are reduced. We expect and
require that the performance test be conducted near operating system capacity. As such,
establishing a limit only for the liquid flow rate performance test would generally require higher
liquid-to-gas ratios during times of lower FCCU utilization rates than the liquid-to-gas ratio
operating limit.
With respect to wet scrubber pressure drop, we agree that pressure drop is not a good
indicator of performance for wet scrubbers using jet-ejector systems. We also realize that most
FCCU wet scrubber systems employ jet-ejector or similar high pressure nozzle systems to
atomize the scrubbing water whether the systems employs a venturi scrubber design or not. For
nearly all FCCU wet scrubber systems, the venturi is used for mixing rather than atomization,
and the operating pressure drop is not nearly as important an indicator of performance as it is in a
traditional venturi scrubber. Therefore, we provided an alternative to pressure drop as an
operating parameter for jet-ejector or other atomizing spray nozzles (regardless of wet scrubber
type); the alternative requires owners and operators to conduct a daily check of the air or water
pressure, as applicable, to the spray nozzles to ensure proper performance.
We maintain that secondary voltage is an important indicator of ESP control device
performance. We recognize that the secondary currents may vary in different fields of the ESP,
so we clarified that the average secondary voltage for the entire system is to be used as the
operating parameter, not the secondary voltage for each particular field. Also, as FCCU fines are
74
-------
expected to have high resistivity, we maintain that total power input is an appropriate operating
parameter to monitor.
Notwithstanding the previous paragraphs, we acknowledge that control device operating
limits are only indicators of actual PM emissions performance. The refinery owner or operator
may elect to install a PM CEMS is they feel the operating limits are too confining. Also, given
the commenters' concerns regarding the ability of the monitored parameters to indicate
performance and previous comments regarding deterioration of PM emissions performance
between FCCU turnarounds (see Section 5.1.1), we have increased the frequency of PM
performance testing to annually. Additionally, an annual performance demonstration for units
electing operating limits is more commensurate with the quarterly and annual accuracy
determination requirements for units electing to use a PM CEMS.
Comment: Commenter 0159 recommended that EPA include provisions for a source-
specific monitoring plan for other types of control devices (e.g., cyclones) that do not have
specifically-listed control device operating parameter monitoring requirements consistent with
provisions in Refinery MACT II.
Response: We agree, and we have included a provision to develop a source specific
monitoring plan for cyclones or other control systems for which operating parameters are not
specifically provided.
5.1.6 Oth er FCCU-related Comments
Comment: Commenter 0154 opposed expanding the FCCU applicability beyond the
regenerator and air blower (i.e., to include the CO boiler).
Response: There are two general types of FCCU, complete combustion (or complete
burn) systems that fully oxidize the coke within the regenerator to CO2 and partial combustion
(or partial burn) units that convert the coke to a mixture of CO and CO2, which subsequently use
a CO or waste heat boiler to complete the combustion process. As the CO boiler is an integral
part of the partial burn FCCU, it is reasonable to include it as part of the affected source.
Furthermore, all of the data used to develop the FCCU emission limits included the contribution
from the CO boiler, if present. That is, the PM and SO2 emission limits are achievable for FCCU
that have CO boilers. Furthermore, for partial combustion FCCU, the CO boiler is expected to
be the primary contributor to NOx emissions from the FCCU. Allowing partial combustion
75
-------
FCCU to comply with the NOx emission upstream of the CO boiler nullifies the intent of the
NOx emission limits.
Comment: Commenters 0154 and 0156 supported the proposed revisions to the coke
burn-off equation in subpart J (and Ja). Commenter 0121 noted that the definition of K2 in
Equation 2 on page 27210 (coke burn-off rate) appears to incorrectly reference "%" in the units.
Commenter 0125 noted that the time period associated with the coke burn-off rate determination
in §60.102a(c)(l)(ii) was not specified and suggested a 7-day average be used. Commenter 0156
stated that the definition of Qr in the coke burn-off equation should be the same as Refinery
MACT II, which allows measurement prior to a precipitator. The commenter also requested that
the alternative flue gas rate determination in §63.1573(a)(1) or (a)(2), as applicable, be allowed
to calculate Qr. These changes will make the rules consistent and reduce permitting burden for
both the agency and the refineries.
Response: We agree that the units in the definition of K2 were incorrect and we have
corrected the definition units in this final rule. We have also specified the time period for the
coke burn-off determination. It was our intent that the coke burn-off rate be calculated hourly,
using hourly average input values (flow rate and exhaust gas concentrations), and that these
hourly rates would be used to calculate a daily average (calendar day) coke burn-off rate that
would be recorded and maintained along with the hours of operation per calendar day as required
in §60.105a(b)(3). As such, §60.102a(c)(l)(ii) specifies that the daily average coke burn-off rate
must not exceed the level established during the performance test. Only the equation in
§63.1573(a)(2) is applicable for determining Qr and this alternative equation was provided in the
proposed rule and is also included in this final rule. The equation in §63.1573(a)(1) was
provided as an alternative to a CPMS for measuring actual flow rates in the exhaust stream. We
have provided this equation as an alternative to the requirement to install a CPMS to measure
exhaust gas flow rate in §60.105a(b)(l)(ii).
Comment: Commenter 0159 stated that, when it is unsafe to measure gas flow rate
between the FCCU and CO boiler, EPA should provide an alternative monitoring option, such as
their recently granted AMP.
Response: Equation 3 provides a means to calculate the regenerator exhaust gas flow rate
if it is difficult to measure. Unfortunately, the referenced AMP was not included in the docket
submission so an evaluation of the applicability of the cited AMP to other units could not be
76
-------
determined. Individual facilities do have the right to apply for an AMP if the site-specific
conditions warrant alternative monitoring procedures.
Comment: Commenter 0125 indicated that the equation used to calculate the PM
emission rate (Equation 1 in subpart Ja) places the unit conversion factor in the denominator
whereas the corresponding equation in subpart J and Refinery MACT II has it in the numerator.
The same equation should be used in all three rules. The commenter also noted that the equation
uses Cpm but the definition uses Cs.
Response: The unit conversion constant in proposed Equation 1 (promulgated as
Equation 3) of subpart Ja is in the denominator just as it is in §60.106(b) of subpart J. There are
slight differences in the units used to define the emission terms, but we did maintain consistency
with the subpart J equation presentation. Consistency regarding the placement of the unit
conversion term between all three rules is not important as ensuring that the value and units of K
result in the correct value and units for E, which we have done in subpart Ja. We agree with the
commenter that designation for the concentration of total PM was different in the equation and
the definition of terms, and we have used the designation "cs" for both in the final standards.
Comment: Commenter 0156 noted that there is inconsistency with the use of "0 percent
excess air" and "0 percent O2"; the commenter suggested that both subparts J and Ja should
consistently use the term "0 percent O2," which is used in 60.106(h)(6) and 60.104a(d)(8).
Response: We attempted to consistently use to the term "0 percent excess air" as this is
the phrase used in subpart J. However, the equation for correcting to "0 percent excess air" in
subpart J defined the adjusted concentration, Cadj, as the "concentration adjusted to zero percent
oxygen." In attempts to avoid any confusion on this matter, we defined Cadj in Equation 6 of
subpart Ja to indicate that the correction to zero percent oxygen and zero percent excess air are
the same. While we agree with the commenter's preference to refer to the "0 percent 02," we
recognize that existing NSPS units and local permit agencies are likely to have "0 percent excess
air" already in their permits, and we saw no reason to have these permits adjusted simply to
revise this language. We believe that it is sufficient to indicate that, for the purposes of subparts
J and Ja, these terms have the exact same meaning.
77
-------
5.2 Fluid Coking Units
Comment: Commenters 0148 and 0154 opposed the inclusion of standards for FCU
altogether and recommended they be removed from subpart Ja. Commenter 0138 recommended
that, if standards are developed for FCU, the FCU standards should only apply to new units and
modified or reconstructed units should be exempt from these standards.
On the other hand, Commenter 0130 stated that the limits for PM, SO2, and NOx in
subpart Ja should apply to all new, reconstructed, and modified FCU, and the commenter
objected to allowing reconstructed and modified units to meet subpart J. EPA has not justified or
explained the rationale for this "do nothing" proposal.
Response: At refineries that have a FCU, the FCU is often the largest emission source.
The fact that there are only a few FCU in operation in the U.S. is not sufficient rationale to
ignore this very large emission source. Additionally, industry representatives have indicated that
it is unlikely that any new FCU will be built because DCU have found preference in the
petroleum refining industry. Therefore, limiting the rule only to newly constructed FCU has the
potential to be equivalent to no rule at all. FCU are a significant source of PM, SO2, and NOx
emissions, and standards are needed to ensure that these sources are adequately controlled.
Comment: According to Commenter 0138, one of the four existing FCU is slated for
permanent shutdown; the two operated by the commenter have recently installed or are currently
installing controls. Commenter 0148 operates the fourth unit, which is used to fuel a co-
generation unit; the stack gases from the cogeneration boilers are treated in a limestone fluidized
bed scrubber (baghouse) for SO2 and PM control and are subject to emission standard relevant
for the cogeneration/power industry. Commenters 0138, 0148, and 0154 noted that under the
proposed rule, this FCU would have to install controls for instances when the cogeneration unit
is down (about 300 hours per year). The commenters used this facility as an example of why
emission controls would not be cost-effective for PM, S02, or NOx.
Response: The situation of the off-site cogeneration boilers is unique but does not justify
ignoring FCU emissions, as FCU are significant emission sources. We first note that there is
very little difference in a CO boiler and a cogeneration boiler. The commenters attempted to
distance themselves from the relevant standards for the cogeneration/power industry, but the fact
remains that the control devices used for the cogeneration boiler (in this case, a spray dry
adsorber/baghouse system) are applicable to control emissions from an FCU CO boiler and can
78
-------
meet the emission limitations required by this final rule. Second, the "solution" proposed by the
commenter to add a control system for 300 hours per year is not realistic and does not account
for any energy savings the refinery would incur by using the energy produced by the combustion
of the gases to off-set steam consumption or other energy needs. Third, the refinery in question
is required by consent decree to install a wet gas scrubber on its FCCU. If this refinery
anticipated a modification or reconstruction of the FCU was likely in the near future, it could
consider the possibility of designing the FCCU scrubber to handle the FCU exhaust for these
short time periods. The truth is, there are a variety of solutions possible, but only the worst-cost
solution was provided by the commenters. Nonetheless, we developed a cost estimate specific
for this refinery, based upon vendor-quoted wet scrubber cost estimates. Assuming only
300 hours of operation per year, we estimated that the total pollutant control costs for this unit
would be less than $10,000 per ton of PM/SO2 reduced, which is not unreasonable for these
pollutants. Consequently, we conclude that wet scrubbers to control PM and SO2 emissions from
FCU are cost-effective and are, therefore, BDT.
Comment: Commenter 0130 stated that although EPA's PM and S02 analyses appear to
be the same for FCU and FCCU, EPA did not explain why it is considering a total PM limit for
FCCU that includes condensable PM but is not considering the same for FCU. Commenter 0130
stated that any PM limit must include condensable PM, as condensable PM account for a large
portion of refinery PM emissions and are some of the most hazardous PM emissions. The
commenter stated that the total PM limit including both filterable and condensable PM from
FCU as well as FCCU should be 0.5 kg/Mg coke burn, and EPA has not demonstrated that
current BDT cannot achieve this limit.
Response: We were considering including condensable PM for FCU just as we were for
FCCU. However, due to unresolved issues regarding the condensable test method and therefore
a lack of performance data based on such a method, we are not including limits for condensable
PM in this final rule.
At the time of proposal, we only had one source test for fluid coking unit. The initial data
indicated that the system achieved a PM limit of 0.3 kg/Mg coke burn using EPA Method 5B.
The second performance test on the same FCU/wet scrubber system 8 months later indicated that
the system achieved a PM emission rate of 0.83 kg/Mg coke burn. As explained in the preamble
to the final standards, we have determined that the wet scrubber controlling emissions from this
79
-------
FCU represents BDT, and the two data points for this scrubber indicate that 1.0 kg/Mg coke burn
using Method 5B is achievable.
Comment: Commenters 0138 and 0154 stated that the consent decrees should not be
used as the basis for the proposed standards. The consent decrees were negotiated on a site-
specific basis and did not consider cost-effectiveness or technical feasibility.
Response: The consent decrees were not used as the sole basis for the proposed
standards. The consent decrees often demonstrated that certain emission limitations were
technically feasible; separate cost analyses were then conducted to assess whether or not these
achievable emission limits were cost-effective and therefore BDT.
Comment: Commenter 0154 requested clarification of a "basic wet scrubber" and an
"enhanced wet scrubber" as used in EPA's impact analysis.
Response: We received vendor quotes for two wet scrubber systems based on the rate
and inlet loading of the model FCCU used at proposal. The "basic wet scrubber" is an EDV-
1000 (DuPont™ BELCO®) designed to meet a PM limit of 1.0 kg/Mg coke burn, and the
"enhanced wet scrubber" is an EDV-5000 (DuPont™ BELCO®) designed to meet a PM limit of
0.5 kg/Mg coke burn.
Comment: Commenter 0154 suggested that there is an inconsistency between the
monitoring requirements in §63.105a(b)(2) and the definition of "fluid coking unit" in §63.101a.
The commenter believed that monitoring before the CO boiler should be deleted. The
commenter also suggested adding an option to monitor after the CO boiler and subtract the
contribution of CO from the CO boiler from the atmospheric emissions.
Response: The monitoring of CO prior to the CO boiler in §63.105a(b)(2), if needed, is
used only to calculate the coke burn-off rate for compliance with the PM standards. To
demonstrate compliance with the final exhaust CO emission limit for units with a CO boiler, a
separate CO monitor is needed as specified in §63.105a(g). Therefore, we believe that there is
no conflict between §63.105a(b)(2) and the definition of "fluid coking unit" and that no
additional monitoring option is needed.
Furthermore, the CO boiler does not contribute to CO emissions; it acts to reduce the CO
emissions. The CO concentration in the exhaust gas entering a typical CO boiler is
approximately 2 to 5 percent and the CO concentration leaving the CO boiler is required to be
less than 500 ppmv (a 99 percent or greater reduction). When CO boilers are used, the CO
80
-------
CEMS installed to assess the compliance with the CO emission limit must be installed
downstream of the CO boiler.
Comment: Commenter 0130 noted that delayed coking units (DCU) account for
91.6 percent of the industry's coking charge capacity in 2001 to 2006, and the remaining
8.4 percent was attributed to FCU. Therefore, the commenter's recommended FCU NOx limit of
20 ppmv should also apply to new, modified, and reconstructed DCU.
Response: We note that DCU do not have a direct atmospheric vent; therefore, it is not
feasible to apply an emission limit to a DCU. The DCU process heaters are subject to the
40 ppmv NOx emission limit for process heaters in the final standards (assuming that the process
heaters are greater than 40 MMBtu/hr).
81
-------
Chapter 6
SULFUR RECOVERY PLANT (SRP) STANDARDS
Comment: Commenter 0154 supported the proposed 250 ppmv combined SO2 and
reduced sulfur emission limit for SRP > 20 long tons per day (LTD) but requested that EPA
retain the 300 ppmv reduced sulfur limit for units that do not emit S02.
Commenter 0127 stated that the 250 ppmv SO2 standard for SRP is outdated; some SRP
have been achieving SO2 limits of 50 ppmv for more than 10 years. Therefore, the commenter
recommended that the NSPS limit for new SRP be set at 50 ppmv (dry basis, corrected to
0 percent excess air). Commenter 0127 also supported the 10 ppmv H2S limit for SRP but noted
that lower limits have been achieved in practice.
Response: The 250 ppmv limit is effectively based on an overall SRP efficiency of 99.9
percent. As described previously, we clarify in the final standards that emissions from the
primary sulfur recovery pits must be included for SRP that become subject to subpart Ja.
Adequate information was not provided to indicate if the SRP achieving the 50 ppmv SO2 limit
is meeting that limit while including sulfur pit emissions. Furthermore, no data were provided to
verify that this level of performance on a 12-hour basis (versus a longer averaging period).
Based on the data available to the Administrator, the 250 ppmv total sulfur limit (or 99.9 percent
sulfur recovery) is achievable for all modified, reconstructed, or new SRP.
Comment: Commenter 0161 stated that the applicability threshold criteria should be
based on sulfur loading rather than sulfur production because the upstream measurements are
easier to make and are more accurate. The commenter suggested that the "miscellaneous
correction" refining the 20 LTD cut-off should preferentially state "... except Claus plants that
have never processed more than 20 long tons per day (LTD)." Additionally, the requirement to
determine and record the hourly production rate and the hours of operation for each SRP should
be required only for those sources that claim to be exempt from more stringent emissions
standards because they process less than 20 LTD. The commenter also suggested that the time
82
-------
frame for measurement should be once per day rather than once per hour to improve accuracy
when pit levels are used.
Commenter 0154 suggested that the rule should include provisions to use either a CEMS
or a CPMS to demonstrate compliance with the SO2 standard for SRP, similar to the FCCU PM
compliance options. Commenter 0154 also suggested that the CPMS alternative to the H2S
standard should be determined on a 12-hour rolling average basis like the emission limit.
Response: Note that we specifically indicate in the rule that the size threshold is based on
the capacity of the SRP and not its actual production. As these are new source standards,
affected sources must be able to assess their applicability prior to start-up. Therefore, we reject
the suggested clarification that the cut-off should preferentially state "... except Claus plants that
have never processed more than 20 long tons per day (LTD)."
We appreciate the comments regarding the difficulties in determining sulfur production
rates, and we have revised the requirements in this final rule to eliminate most of the
commenters' concerns. First, the emission limits for small SRP have been revised to be 10 times
higher than the limits for large SRP. For example, if 250 ppmv S02 is assumed to be equivalent
to 99.9 percent control (the level or performance assumed for a large SRP), then 2,500 ppmv SO2
should be equivalent to 99 percent control, or the level of performance required for a small SRP.
The concentration emission limits can be met with a CEMS. In addition, the requirement to
record the hourly production rate and the hours of operation for each SRP has been removed
from the final standards.
Comment: Commenter 0135 noted that the adoption of a completely different standard
for small SRP appears to be based on the assumption that these are non-Claus based units;
however, the commenter does operate a small Claus-based sulfur plant where the tail gas from
the plant is treated by an incinerator and WGS; the FCCU regenerator tail gas is also fed through
the WGS. It would be difficult to segregate the sulfur plant tail gas load from the FCCU tail gas
to make an efficiency determination based on LTD of sulfur recovered. Therefore, the
commenter asked that EPA specifically allow the standard to be applied to FCCU regenerator
and sulfur plant tail gas combined and treated in a WGS, given the high removal efficiency of the
unit.
Response: It is inappropriate to grant this combined stack an emission limit of 250 ppmv.
It is expected that the FCCU would provide the bulk of the total volumetric flow and a 250 ppmv
83
-------
limit would likely allow much higher SO2 emissions than a more specifically targeted
compliance option, such as one based on a flow-weighted average. The configuration described
by the commenter is quite unique, and it is beyond the scope of these national standards to
attempt to address compliance provisions for every possible unique refinery and control system
configuration. For this case, the refinery owner and operator should submit an alternative
monitoring plan as allowed under §60.13(i) of the NSPS General Provisions. Alternatively, the
refinery owner and operator can comply with the 50/25 ppmv S02 standard for FCCU on the
combined stream; achieving that level of performance would clearly demonstrate compliance
with both the FCCU and SRP emission limits.
Comment: Commenter 0161 stated that the 10 ppmv H2S limit is not needed because
H2S concentrations are included in the overall S02 emission limit. Commenter 0161 also stated
that other reduced sulfur compounds have a higher oxidation temperature than H2S, so using H2S
as surrogate indicator of tail gas incinerator efficiency for continuing operation may not be
achieving the desired goal of preventing reduced sulfur compound slip.
Response: In subpart J, the 10 ppmv H2S limit existed for reductive control systems even
though H2S is included in the TRS measurement. With the clarification of the inclusion of sulfur
pits as part of the affected source, reduced sulfur releases remain a concern. Considering all of
the variations in the emission sources and controls, we maintain that limiting outlet H2S
concentrations to 10 ppmv or less is needed to ensure a properly operated and maintained SRP.
Furthermore, for the specific case of SRP that employ a tail gas incinerator or similar combustion
device, we recognize that H2S oxidation is efficiently achieved and we allow parameter
monitoring of the combustion device rather than direct H2S monitoring under this circumstance.
Comment: Commenter 0161 stated that small SRP will have to convert flow from a wet
to a dry basis and suggested that the moisture content as determined during the initial
performance test be allowed for use to perform this conversion, as moisture content of this
stream does not vary significantly.
Response: We have revised the performance standard and compliance requirements for
the small SRP since proposal. This comment is no longer relevant.
Comment: Commenter 0161 suggested that the span range for the 02 analyzer of the
SRP should be 10 percent as currently specified under subpart J because most SRP will operate
at less than 5 percent excess 02.
84
-------
Response: As discussed previously, we agree that a 10 percent span is acceptable for all
O2 monitoring systems and we provide flexible limits for the O2 span in this final rule.
Comment: Regarding the temperature and 02 operating limits, Commenter 0161 stated
that the operating limit should be the lowest level at which compliance was previously
demonstrated through successful performance testing (rather than the most recent performance
test).
Response: As with the CPMS requirements on the FCCU, changes to control system
characteristics due to equipment wear are of concern. Therefore, we require the CPMS operating
parameters to be determined by the most recent performance demonstration. However, the
concern for degrading control device performance is less for SRP tail gas incinerators; therefore,
we only require an initial performance demonstration. Additional performance demonstrations
may be requested by the State or local agency, or a refinery owner or operator may elect to
conduct an additional performance demonstration for the express purpose of establishing new
operating parameters. In either case, the owner or operator must determine and use the operating
parameter limits established during the most recent performance demonstration because the unit
could degrade over time so that the most recent performance test provides the best assessment of
allowable operating parameter values needed to meet the emissions limit.
Comment: Commenter 0148 recommended that the monitoring requirements associated
with the SRP be consistent with those in Refinery MACT II. Specifically, many refineries have
CEMS that measure H2S, COS, and CS2. The proposed regulation requires that H2S still be
measured but requires that the stream be combusted to measure the S02 in the combustion
products. These refiners will incur significant costs to install new CEMS that will do nothing to
reduce emissions.
Response: In the Refinery MACT II, we exclude H2S because it is not a HAP. However,
if a source becomes subject to the Refinery NSPS, monitoring of H2S is required.
Comment: Commenter 0154 stated that a compliance option is needed for oxygen-
enriched SRP. As these units feed much less nitrogen into the system, their mass emissions per
unit of sulfur recovery can be comparable to traditional units, but they cannot meet the 250 ppmv
concentration limit. Commenter 0121 also noted that the concentration standard is stricter for
units using oxygen enrichment and requested clarification if this was EPA's intent.
85
-------
Response: We acknowledge that the proposed concentration standard for SRP is not
appropriate for units that use oxygen enrichment. We would like to encourage the use of oxygen
enrichment, as this allows the sulfur plant to run more efficiently and more reliably. We have
developed two correlation equations, one for SRP with capacities greater than 20 LTD
(Equation 1 in §60.102a(f)(l)(iii)), and another for SRP with capacities of 20 LTD or less
(Equation 2 in §60.102a(f)(2)(iii)). The owner or operator of an oxygen-enriched SRP can use
these equations to calculate a unit-specific emission limit based on the oxygen concentration.
This alternative will allow units that use oxygen enrichment a means to comply with a similar
standard as units that use air.
Comment: Commenters 0154 and 0156 suggested that EPA should retain the 300 ppmv
TRS limit for reduction control systems. According to Commenter 0154, the rule appears to
require incineration, which was not included in the cost analysis and which would result in
adverse secondary air emissions. Commenter 0156 cited the 1983 review of the NSPS, which
states that the 300 ppmv reduced sulfur compound (RSC) standard is roughly equivalent to BDT
of 99.8 to 99.9 percent. The higher standard is needed because combustion air used to incinerate
the RSC dilutes the resulting SO2 concentration. According to the commenter, lowering the RSC
emission standard is arbitrary and will unfairly penalize SRP that do not use an incinerator, when
EPA should instead encourage these types of systems, as incinerators consume fuel and increase
PM, NOx and CO emissions.
Response: We did not require incineration in the proposed rule and we did not intend to.
However, it is true that because the proposed SO2 limit is provided in terms of a concentration,
the use of an incinerator would effectively dilute the tail gas stream, making it easier to comply
with the standard. We minimize the amount of dilution that can occur by requiring the
concentration standard to be corrected to 0 percent oxygen, but some dilution does occur as a
result of the combustion fuel and auxiliary air required by incinerator. Therefore, in the final
subpart Ja, we have retained the limits provided to large SRP in subpart J. We have also
included separate emission limits for small SRP in subpart Ja.
Comment: Commenters 0154 and 0156 noted that the subpart Ja proposed standard for
SRP is a combined SO2 and reduced sulfur compound (RSC) emission limit. However, the
monitoring requirement is a total sulfur monitor, which is inconsistent with the emission limit.
In cases of modifications or reconstructions that trigger subpart Ja, new CEMS would be
86
-------
required at significant costs but with little or no environmental benefit. The commenter
suggested that the rule should specify the use of both an SO2 and an RSC CEMS rather than the
total sulfur monitor. If the initial compliance demonstration indicated that RSC concentration
was less than 10 ppmv, then the RSC CEM would not be required. If the initial compliance
demonstration indicated that SO2 concentration was less than 10 ppmv, then the SO2 CEM would
not be required. According to the commenters, this provision will allow most modified units to
maintain their current CEMS. Commenter 0156 stated that total installed cost of an RSC CEMS
was approximately $250,000. Commenter 0148 also expressed concern that new CEMS would
be required for FLEXSORB units at costs of approximately $500,000 for no emission reductions
and recommended adopting the Refinery MACT II provisions for monitoring SRP.
Commenter 0161 also indicated that Method 15 should not be required when a ceramic thermal
oxidizer is employed, as all sulfur compounds would be converted to SO2.
Response: For reasons stated in the preamble, we maintained the form of the SRP
emission limits in subpart J in subpart Ja. Therefore, these comments are no longer relevant.
Comment: Commenter 0161 stated that the rule should allow for site-specific test plans
to be approved by the local regulating agency when there are discrepancies caused by sulfuric
acid mist between wet chemistry methods and CEMS performance demonstrations {i.e., one
method may count it while the other method may not).
Response: Site-specific monitoring plans may be requested under §60.13(i) of the NSPS
General Provisions. Both permitting authorities and the Agency sign off on test plans as part of
title V approval process.
Comment: Commenter 0156 noted that there was a discrepancy between proposed
§60.102a(e)(3) and §60.106a(b)(3). The emission limit in §60.102a(e)(3) is on a 12-hour rolling
average basis; therefore, the excess emission report trigger in §60.106a(b)(3) should also be on a
12-hour rolling average basis.
Response: We agree, and we have clarified §60.106a(b)(3) to require reporting of excess
emissions for all 12-hour periods where the H2S concentration exceeds the applicable emission
limit.
Comment: Commenter 0154 noted that the cross-reference in §60.106a(5) [we expect
that the commenter meant to indicate §60.106a(a)(5)] should reference §60.102a(e)(2) and not
§60.102a(b).
87
-------
Response: The commenter is correct that the cross-reference in proposed §60.106a(a)(5)
should not have referenced proposed §60.102a(b). However, because the percent recovery
compliance option was removed from the final standards, proposed §60.106a(a)(5) does not
appear in the final standards; therefore, there is no need to revise the incorrect reference.
88
-------
Chapter 7
WORK PRACTICE STANDARDS
7.1 General
Comment: Commenter 0154 requested clarification on whether the proposed work
practice standard requirements apply to existing units and whether or not the work practice
standards regarding fuel gas producing units and SSM plans/root cause analysis (RCA) are
intended to apply only to new units and not to modified or reconstructed units. The commenter
interpreted the rule to say that the proposed DCU depressurization standard is the only work
practice standard that applies to modified and reconstructed units (as well as new units).
Response: All of the final work practice standards are intended to apply to new,
modified, and reconstructed sources. We note that a number of these work practice standards
have been revised since proposal; revisions include specifying that the affected facility for the
flare management standards is the flare rather than a fuel gas producing unit.
Comment: Commenter 0130 suggested that given the significant hazards to human
health and the environment posed by volatile organic compound (VOC) emissions, and given
that at least one refinery has enclosed its coke cutting operations, the NSPS should require
enclosure of coke cutting operations. EPA has not justified or explained its rationale for a "do
nothing" proposal.
Response: We further investigated the enclosure for coke cutting operations. The
enclosure is designed specifically to reduce PM emissions; it is not designed to be a gas-tight
enclosure. There is no fan by which the emissions are routed to a central location for
treatment/control. As such, the coke cutting enclosure is not expected to effect a reduction in
VOC emissions. We could not identify a commercially available and proven process by which
the VOC emissions from coke-cutting operations could be cost-effectively reduced. Therefore,
we did not establish a VOC emission limit or work practice standard for coke-cutting operations.
89
-------
Comment: Commenter 0156 supported Option 2 recordkeeping and reporting
requirements under §60.108a.
Response: The only difference between Option 1 and Option 2 recordkeeping and
reporting requirements is that Option 1 includes requirements for an SSM plan and RCA
(consistent with the requirements of Option 1 for §60.103a) and Option 2 does not (consistent
with the requirements of Option 2 for §60.103a). We note that the commenter did not object to
the SSM and RCA requirements in general, rather the commenter provided comments on how to
improve them. The recordkeeping and reporting requirements must be commensurate with the
requirements in the other sections of the rule. The final standards include recordkeeping and
reporting requirements appropriate for the final work practice standards.
7.2 Flare Management
Comment: Commenters 0125 and 0171 requested that EPA clarify what constitutes the
"affected facility" for a flare. Commenter 0171 asked if the knockout drum, flare gas
compressor, and pilot gas system are considered part of the affected facility and noted that it
does not make sense for replacement of a knockout drum to trigger NSPS applicability.
Commenter 0125 suggested that the flare should include only equipment located downstream of
the inlet flange to the flare knockout drum. The affected facility for a flare should not include
flare gas recovery (FGR) systems, amine systems, steam assist systems, fuel gas monitoring
equipment, or any upstream piping or knockout drums not directly located by the flare.
Commenter 0125 suggested that EPA could make this clarification in the NSPS General
Provisions.
Response: The final subpart Ja standards include a definition for "flare" that 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, knockout pots, piping
and header systems. We also recognize that the general determination of when an affected source
is considered to be modified is not easily applied to a flare. Therefore, as explained in the
preamble to the final standards, we have provided clarification regarding when a flare is
considered an affected source through modification.
90
-------
Comment: Commenter 0174 stated that the requirements for the FMP as proposed are
"overly burdensome and unnecessarily broad." The commenter recommended that EPA limit the
plan to: (1) technical information about the flare and the upstream equipment sending sulfur-
containing gas to the flare; (2) a description of the refinery's plan for reducing flaring of the
sulfur-containing gases; (3) plans for minimizing flaring during maintenance, startup and
shutdown, fluctuations in gas quantity or quality, and recurring failure of a control device or
other equipment; and (4) procedure for a RCA. Under the commenter's plan, refineries would
submit a summary of the FMP (to ensure that no confidential information is released) and formal
EPA approval would not be necessary (although EPA would have the right to review the full
FMP and request any necessary changes). The refinery would review and update the FMP (and
the summary, if necessary) each year. The refinery would report any changes to the FMP,
instances when the FMP was not followed, and RCA performed (including results) in its semi-
annual report. The commenter also noted that sources not subject to subpart Ja but electing to
comply with the work practice requirements should be able to include these items in their subpart
J periodic report to avoid duplicate reporting. Commenter 0150 noted that FMP must allow for
flaring during times when there are safety concerns or it is not cost-effective to recover the fuel
gas. In addition, a reasonable timeframe is needed to develop and implement the FMP.
Response: In the final subpart Ja standards, the owner or operator may only send
250,000 standard cubic feet per day (scfd) of fuel gas to an affected flare during normal
operations. The flare also must have an FMP that includes: (1) methods for monitoring the flow
rate to flare; (2) procedures to minimize discharges to the flare system during the planned start-
up or shutdown of the refinery process units that are connected directly to the affected flare; (3)
procedures for conducting a root cause analysis of any process upset or malfunction that causes a
discharge to the flare in excess of 500,000 scfd; (4) procedures to be taken to reduce flaring in
cases of fuel gas imbalance; and (5) records to keep during times of fuel gas imbalance and other
circumstances that keep the flare from meeting the 250,000 scfd flow rate limit. We note that we
view the flare management requirement in the final standards as providing benefits in addition to
S02 reductions (it also reduces VOC and NOx emissions and increases the overall energy
efficiency of the refinery); therefore, we do not agree with the commenter that suggested limiting
the applicability of the flare management practices to those gases that contain significant
quantities or concentrations of sulfur.
91
-------
Comment: Commenters 0154 and 0156 supported the co-proposal of no restrictions on
routine flaring at refineries that are "fuel gas rich." Additionally, Commenter 0154 suggested the
term "fuel gas rich" should be changed to "periods of fuel gas imbalance" to accommodate
instances when significant fuel gas consumers (e.g., catalytic reforming units) are down and the
refinery temporarily produces more fuel gas than it can consume. Similarly, Commenter 0150
noted that the FMP must allow for flaring during times when the refinery is "long on fuel gas."
On the other hand, Commenter 0126 did not believe an exemption for fuel gas-rich refineries
was needed as that would be a very unlikely occurrence.
Response: We anticipate that a FGR system will be a common method chosen to comply
with the flow rate requirements when a flare is currently used on a continuous basis. For
refineries that are not fuel gas-rich, recovered fuel gas offsets natural gas purchases. The cost-
effectiveness of the FGR system is primarily dependent on the quantity of gas that the system
can recover. Based on estimates of current flaring quantities, we anticipate that the cost-
effectiveness of the FMP, including FGR systems when needed, is approximately $500 to $1,500
per ton of S02 reduced. Flare gas recovery will also reduce NOx and VOC emissions. Many
refineries have already implemented similar work practices through consent decrees and local
rules (BAAQMD and SCAQMD), and these requirements have had a demonstrated reduction in
flaring events. Furthermore, we find that these work practices are cost-effective and are,
therefore, BDT.
However, FGR is not practical if there is no place in the refinery where the gases can be
used. If a refinery does not purchase any natural gas and fires all on-site boilers and process
heaters with fuel gas and still has excess fuel gas, then that refinery is considered fuel gas-rich.
In this case, there are only two real options for the excess fuel gas: (1) either burn the excess
fuel gas in an additional boiler or turbine to produce steam and/or electricity (for either on-site or
off-site use, "cogeneration"); or (2) flare the excess fuel gas. Therefore, we have provided
refinery owners and operators a provision to flare excess fuel gas if the refinery can demonstrate
that it is producing more fuel gas than it can use. As mentioned by the commenter, being fuel
gas rich on an on-going basis is rare. It is even more unlikely that the quantity of excess fuel gas
is sufficient to make the cogeneration option economically viable or cost-effective. Any owner
or operator of an affected flare at a refinery that has been or has the potential to be fuel gas-rich
92
-------
should use the FMP for that flare to address procedures to follow and records to keep during
times of fuel gas imbalance.
Comment: Commenters 0148, 0154, 0156, and 0159 stated that there should be no limit
on routine flaring of gas that meets the H2S concentration limits. According to these
commenters, the prohibition on "routine flaring" even for gas that meets the H2S concentration
limits could also be interpreted to ban the use of sweep or purge gas that is needed for safe flare
operations. Commenter 0125 recommended that flare header sweep gas and flare stack purge
gas be specifically excluded from the definition of "fuel gas" or from the fuel gas monitoring
requirements. Commenter 0125 also suggested that EPA provide a regulatory allowance for
flaring gas during "boiler-code mandated inspections of pressure vessels in the flare gas system."
Commenters 0156 and 0159 also stated that product or vessel blanketing gas should be exempt
from the prohibition from flaring. Additionally, some gases, such as high hydrogen or nitrogen
streams, can cause upsets of the combustion units, which would lead to increased flaring
compared to the direct flaring of these gases. Therefore, flaring of these gases should be
allowed. Commenter 0174 added that a requirement to eliminate "routine" flaring also conflicts
with many regulations (Federal and State) that provide standards for the use of flares as a control
device for HAP and VOC.
Response: The final standards do not define or prohibit "routine" flaring. Instead, they
require the owner or operator of an affected flare to monitor the flow rate of that flare and limit
the flow to the flare to 250,000 scfd under normal conditions. Therefore, exemptions for the
specific situations mentioned by the commenters are unnecessary.
While we agree that sweep or purge gases are needed for safe flare operations, we
disagree with the notion that there should be no restrictions on flaring sweet fuel gas unless, as
noted before, the refinery is fuel gas rich. Flaring sweet fuel gas while purchasing natural gas,
steam, or electricity reduces the overall efficiency of the refinery, increases the total energy
consumption, and increases emissions. Sweep gases needed to ensure proper operations of the
flare systems should easily fit into the 250,000 scfd limit. Flaring less than 250,000 scfd of gas
that is low in sulfur content is unlikely to exceed the 500 lb/day S02 trigger for a RCA. As such,
there would be little or no burden to the refinery for this low-sulfur fuel gas flaring, provided that
it is not an on-going or continual event. Similarly, we believe that the once-a-year, short-term
flaring episode that may be associated with a mandated boiler code inspection would not exceed
93
-------
the 500 lb/day SO2 trigger for a RCA and would not greatly burden the refinery owner or
operator. However, as long as a refinery is purchasing natural gas from off-site, we see no
reason to waste natural resources by continuous flare operation when FGR is generally cost-
effective and provides significant environmental benefits. Finally, the change to allow 250,000
scfd flow rather than restricting "routine" flaring should allow flares to be used as control
devices for other standards. If more than 250,000 scfd of gas is burned with an affected flare, the
FMP should address the situation.
Comment: Commenter 0154 indicated that the rule language regarding SSM plans is
unclear and appears to apply to "all equipment," not just to the equipment that is an affected
facility subject to subpart Ja or part of the fuel gas producing unit. Commenter 0156 stated the
term "process equipment," which is used in the SSM plan requirements [§60.103a(b)], is not
defined in the regulation, and the section should instead refer to "affected facilities as defined in
§60.100a(a) " Additionally, amine treatment systems are not affected facilities in §60.100a(a),
so they should not be included in §60.103a(b)(l). Also, Commenter 0156 suggested that the
phrase "refinery process unit subject to the provisions of this subpart" in the last sentence of
§60.103a(b)(l) should be replaced with the phrase "affected facility."
Commenters 0138 and 0154 stated that the Administrator's authority to require changes
to the SSM Plan in §60.103a(b) must be limited to changes that are consistent with the BDT (i.e.,
flare minimization, not flare prohibition during start-up and shutdowns). Commenter 0154
provided suggested language to §60.103a(b)(5)(i) and (ii) to limit revisions to those where the
cost-effectiveness would be no greater than $2,000 per ton of pollutant reduced.
Response: The final standards no longer require a SSM plan, so most of the commenters
concerns should be addressed. We note that as proposed, the SSM plan was to include "a
program of corrective action for malfunctioning process, air pollution control, and monitoring
equipment used to comply with the requirements of this subpart." As such, it did not include "all
process equipment," but only process equipment used to comply with the requirements of this
subpart. In addition, we note that the proposed SSM plan was specifically intended to include
amine treatment systems and other control systems used to meet the S02 standards for fuel gas
combustion devices or the alternative fuel gas standards. We have included a new definition of a
"fuel gas system" which is used, among other things, in the explanation of when a flare is
considered to be modified. This definition specifically includes "units used to remove sulfur
94
-------
contaminants from the fuel gas (e.g., amine scrubbers)" so there should be no confusion that we
intend for amine scrubbers to be considered control systems, but amine treatment systems are not
an affected facility in subpart Ja..
The final standards do not specifically grant the Administrator authority to require
changes to the FMP. We note that in general, we reject the idea that $2,000/ton is not cost-
effective, especially for SO2 reductions when there are other collateral benefits (VOC and NOx
reductions), and if the SSM plan was required in the final standards, we would reject the
suggestion that changes made by the Administrator to a FMP be limited to those where the cost-
effectiveness would be no greater than $2,000 per ton of pollutant reduced.
Comment: Commenter 0150 suggested a voluntary program to incorporate FMP at
refineries. Commenter 0154 suggested that reduction in flaring may be achieved through
voluntary programs, amendments to the NSPS General Provisions, or amendments to Refinery
MACT I. The commenter also suggested that FMP requirements could be added to the
engineering standards for flares in the NSPS General Provisions; in a later comment letter
(0174), the commenter concluded that the appropriate location for refinery flare standards is
subparts J and Ja. Commenter 0154 supported enforceable plans implementing procedures for
pre-planning for flare management practices during planned maintenance events and for
implementing safe and reasonable prevention measures identified from a RCA.
Response: We do not believe a voluntary program is appropriate because refineries will
typically only voluntarily install a FGR system when the fuel gas recovery credits provide a
payback of the capital expense within a year, whereas we consider these systems to be highly
cost-effective even when the payback period is 10 or 15 years. We reviewed the potential to
regulate flares under other subparts, but we conclude that the refinery fuel gas and flare systems
are largely unique to the petroleum refining industry and that regulation of these systems in this
rule is appropriate.
Comment: Commenters 0148, 0150, 0154, and 0174 suggested that a 500 lb/day SO2
standard tied with an FMP requirement should be added as an alternative compliance option (to
the H2S concentration limit) for flares. The commenters recommended that this alternative
compliance option be provided in both subparts J and Ja and noted that it could be used as an
incentive for the FMP to cover all flares. Commenter 0154 also noted that these requirements
should be applicable to flares that receive process gas, fuel gas, or process upset gas; they should
95
-------
not be applicable to flares used solely as an air pollution control device, such as a flare used
exclusively to control emissions from a gasoline loading rack. Commenter 0174 clarified that if
the refinery elects to comply with this alternative for any flare, all flares at the refinery would
need a FMP. The commenter noted that EPA could choose to set the 500 lb/day SO2 limit as a
total for all flares for which the alternative compliance option is chosen (i.e., if the alternative
compliance option is selected for two flares at a refinery, the total emissions from both flares
would be limited to 500 lb/day).
Response: The final standards include a requirement to develop and follow a FMP that
includes all units tied to an affected flare. While 500 lb/day SO2 was established as the trigger
for initiating a RCA, we do not believe an exemption from the fuel gas standards is warranted for
emissions below this level, especially if they are on-going, routine releases, which we have
already indicated should be recovered, and we see no rationale for this emission limit to be BDT.
This alternative would appear to promote the use of flares for low volume gas streams that may
not currently be combusted without amine treatment. Additionally, this emission limit, which
translates to 90 tons/yr of potential S02 emissions, would most likely be implemented on a per
fuel gas combustion device basis, so that the inclusion of this alternative could increase SO2
emissions by hundreds of tons per year per refinery.
Comment: Commenter 0131 suggested an exemption from the fuel gas standard should
apply to such de minimis events that are less than 500 lb of SO2 in a 24-hour period to avoid
reporting of flaring events that do not result in significant emissions or exceedances of permitted
limits. This exemption is consistent with the consent decrees and title V upset condition
reporting and State deviation reporting requirements.
Commenter 0124 recommended that EPA follow a simple flare regulation such as the one
adopted by the State of Montana in its Billings/Laurel state plan. Key aspects of the plan,
according to the commenter, include: (1) reporting of significant flare events (using engineering
calculations, not CEMS); (2) exclusion for tracking de minimis events; and (3) limiting "routine"
flaring to levels demonstrated by modeling to assure compliance with NAAQS. Due to the
variation in flow rates and concentrations of fuel gas sent to a flare, CEMS capable of covering
the range of flows or concentrations are either unavailable or inaccurate.
Response: We reviewed the referenced state plan as well as other flare management
plans, such as those implemented in South Coast and Bay Area Air Quality Management
96
-------
District. We believe that the final standards provides a balanced approach that targets reductions
of major flaring occurrences, allows for safe operations of the flares, and provides a reasonable
means of demonstrating continuous compliance with the final standards.
Comment: According to Commenters 0138 and 0139, the consent decree requirements
are largely similar to the proposed work practice standards and they will be implemented over
the course of several years; the proposed standards interfere with these plans/schedules without
an overall emissions benefit (compared to the reductions achieved under their consent decree
requirements). The requirements and timing of any flaring minimization should be consistent
with individual consent decree requirements. Otherwise, the flare minimization requirements
should not become effective until 3 or 4 years after the commencement of modifications or
reconstructions.
Response: As noted by the commenter, the work practice standards have been well-
demonstrated by refineries subject to consent decrees. We have attempted to make these
requirements consistent with consent decrees, but complete consistency with each individual
consent decree is impossible since the consent decree requirements are not all the same. We
have determined that flare management through FMP and FGR systems is BDT. With respect to
the timing of the requirements, section 111(a)(2) of the CAA states that a new source "means
any stationary source or modification of which is commenced after publication of regulations (or,
if earlier, proposed regulations) prescribing a standard of performance under this section which
will be applicable to such source." Furthermore, section 111(b)(1)(B) states that "Standards of
performance or revisions thereof shall become effective upon promulgation." Therefore, the
CAA does not provide the flexibility desired by the commenter. As such, the flare management
requirements are effective when a new flare is installed or an existing flare source is modified or
reconstructed as defined in this rule, thereby becoming an affected source.
Comment: Commenter 0126 stated that electronic steam flow controls should be BDT
and manual adjustment of steam flow should not be permitted, as proper steam flow is critical to
the flare combustion efficiency. In an attached report, Commenter 0126 also suggested that
enclosed ground flares or thermal oxidizers are more efficient destruction technologies than
elevated flares when FGR is impractical.
Response: We focused on minimizing the use of flaring rather than on trying to achieve
some small, and practically impossible to verify, incremental reduction in VOC emissions.
97
-------
Comment: Commenter 0159 suggested that EPA consider the application of properly
engineered split range control valves to be functionally equivalent to standard pressure relief
valves so that releases from these valves can be sent to the flare.
Response: We disagree that split range control valves to be functionally equivalent to
standard pressure relief valves, and included no specific provision for split range control valves
in the final rule.
Comment: Commenter 0122 requested clarification of how the flare standards apply
when a refinery and chemical plant or other non-refining operations are co-located (same parent
company or joint venture partners). The commenter posed the following questions:
¦ If gas generated at the chemical plant is sent to the refinery flare, is it subject to
the proposed standards?
¦ If gas generated at the refinery is sent to the chemical plant flare, is it subject to
the proposed standards?
¦ If refinery and chemical plant gases are mixed together, does the combined stream
need to meet the emissions standard or would only the refinery fuel gases (prior to
mixing) need to be monitored and to comply with the emission limit?
Commenter 0125 stated that subparts J and Ja should not be applicable to flares not located at the
petroleum refinery or flares associated with chemical process units, regardless of whether the
chemical process unit is located at the refinery or at a separate but adjacent (contiguous) plant.
Response: In this final rule, we have eliminated the fuel gas producing unit as an
affected facility, and the affected facility for these specific work practice standards is the flare.
When a flare at a refinery is modified, then that flare has to meet the provisions in this subpart,
which focus on flare management. The gas generated at the chemical plant could be sent
continuously to an affected refinery flare, as long as the total flow from the flare did not exceed
the flow rate limit. The gases could also be sent to a flare header with a FGR system installed to
recover the gases for use elsewhere in the refinery (or chemical plant). In the second case, the
chemical plant flare would not be an affected facility, so gases sent to the chemical plant flare
would not be subject to the requirements in this subpart. Regarding the third question, if the
mixed fuel gas is sent to a new, modified, or reconstructed fuel gas combustion device at the
refinery, then the mixed fuel gas (as used in the affected fuel combustion device) must meet the
standard.
98
-------
7.3 Flare Monitoring
Comment: Commenters 0125, 0138, and 0154 suggested that operating and maintaining
a properly designed FGR system is sufficient "monitoring" to ensure flares burn only exempt
gases. Commenter 0125 stated that the existing FGR systems were installed due to local
regulations or consent decrees, and the sizing of those systems was subject to extensive review.
Therefore, Commenters 0125 and 0154 suggested that "any refinery that operates a FGR system
should be exempt from any further monitoring, continuous or periodic, of its flare system."
Commenter 0154 also indicated that FGR systems cannot eliminate all flaring during planned
start-ups and shutdowns; therefore, a requirement to operate a FGR system should contain an
exemption for process upsets (including planned start-ups and shutdowns).
Commenter 0149 opposed the exemption from a monitoring requirement when FGR
systems are used; the NSPS should require both FGR and monitoring of the sulfur rate and flow
rate, as well as recordkeeping and reporting. The commenter urged EPA to require flare
monitoring and recording plans to be submitted to permitting authorities for approval; operation
monitoring and recording requirements, including type of flare, operating parameters, and
specific measurement and recording of sulfur rate and flow rate; and testing and monitoring
methods. EPA should use recent SCAQMD rules and the model rule developed by the Mid-
Atlantic Regional Air Management Association (MARAMA) as templates for refinery flaring
NSPS requirements. Commenter 0126 agreed that a monitoring exemption is not needed or
appropriate for systems with FGR systems. Appropriately sensitive flow monitoring should be
required to show that routine flaring is not occurring, according to the commenter.
Commenter 0124 and 0154 supported a monitoring exemption for process upset gas; the
exemption is necessary for the safe refinery operations. The commenters supported the
exemption for process upset gases of fuel gas released to the flare as a result of relief valve
leakage or emergency malfunction and supported the clarification that continuous monitoring is
not required for fuel gas streams that are exempt from §60.104(a)(1) (malfunctions and relief
valve leakage).
Commenter 0127 stated that preamble language regarding EPA's decision not to require
monitoring for flaring of process upset gases is inadequate and that the rule should specifically
address the issue in §60.105. There have been several court cases challenging the need for
monitoring of flares to ensure that they are only used as emergency flares; most interpretations
99
-------
suggest flare monitoring is needed to demonstrate compliance with subpart J. The regulatory
text must clarify the type of monitoring, if any, that is required for these flares. The commenter
also indicated that monitoring of the flares for flow and composition is needed if RCA are
required (so that the facility can adequately determine if the 500 lb/day SO2 RCA threshold has
been exceeded.
Commenter 0130 stated that flares, including upset emissions, should be subject to
continuous monitoring and reporting requirements, even if the upset emissions are not subject to
the sulfur content requirements. Emission from upsets can be a major source of pollution from
refineries, sometimes exceeding their total routine emissions of some pollutants. Studies have
shown that wind and other factors can reduce flare combustion efficiencies, which means more
pollution is actually being released rather than being destroyed. The CAA mandates continuous
compliance with its pollution limits and does not provide exceptions for excess emissions.
EPA's Policy Regarding Excess Emissions During Malfunctions, Startup, and Shutdown states
that all periods of excess emissions must be considered violations.
Commenter 0130 stated the NSPS should require continuous monitoring of flares,
electronic reporting of all excess emissions within 24 hours, the reports to be publicly available
within 72 hours on State websites, and the reports to include: (1) pollutants emitted; (2) amount
of emissions; (3) calculation method; (4) cause of release; (5) regulatory limit; (6) amount
emissions exceed limit; and (7) actions to prevent excess emissions from recurring. The NSPS
should also include automatic penalties, require offset reduction in routine emissions, require
facility shutdown after a certain number of excess emissions, and include excess emissions in
potential to emit (PTE) calculations.
Response: After consideration of these comments as well as our decision to define the
flare as the affected facility for the work practice standards (instead of the fuel gas producing
unit), we agree that monitoring of the flares for flow is absolutely necessary to ensure
compliance with these standards. While we encourage the use of FGR systems, monitoring the
flow to the flare is the only way to ensure compliance with the flare flow rate limit and ensure
that the FGR system is not being overloaded as a refinery expands or debottlenecks.
Additionally, we require sulfur monitoring of flare gas to demonstrate compliance with the SO2
root cause analysis trigger.
100
-------
Comment: Commenter 0150 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 if it is monitored elsewhere (e.g., the fuel gas system).
Response: We agree that the sweep gas in these cases would not have to be monitored
for sulfur content at an affected flare. We do require flow and TRS monitoring of the flare waste
gas as explained above, but an H2S monitor is not needed for the sweep gas provided the sweep
gas meets the definitions of streams that do not require monitoring or it has been monitored
previously at a representative spot in the fuel gas system.
7.4 Malfunctions of Amine Systems and SRP
Comment: Commenter 0154 disagreed with EPA's interpretation that petroleum amine
treating systems and SRP are not petroleum refinery process units and therefore cannot generate
process upset gas. The commenter noted that "generate" is not defined in the NSPS and asserted
that the only reasonable disposition of the gases during amine unit or SRP impairment is a flare.
The commenter indicated that the costs presented by EPA for the control options considered
were low and stated that "most refineries have SRP much larger than the 50 LTD, some as much
as 10 times larger or more." Commenters 0150 and 0154 supported EPA's determination that an
SSM plan is BDT for flare minimization during malfunctions of the amine unit or SRP.
Response: We consider the amine treatment system and SRP to be control systems used
to comply with the fuel gas combustion device SO2 standards. When a control device on a
process unit malfunctions, one cannot generally continue to operate the process unit until the
control device is operational. While the amine treatment systems and SRP are not typical
APCD, their primary function is emissions control. While we agree that it is better to flare the
sour gas rather than emit large quantities of untreated H2S, we prefer that neither occur.
Moreover, the final rule does not include the proposed SSM plan requirements so the comments
are no longer relevant.
Comment: Commenter 0154 suggested that the storage of lean amine would potentially
require the storage of rich amine, which raises serious safety issues and, therefore, should not be
considered a viable option. Commenter 0150 cited costs and safety concerns for rejecting amine
storage as a viable option.
101
-------
Response: We did not require storage of rich or lean amine in the proposed rule; we only
considered it as one possible action a refinery may take to minimize production losses when their
sour gas control system malfunctions. The proposed refinery in Arizona includes requirements
for such storage systems, so we expect that appropriate safety measures can be engineered into
the tank systems. We recognize that there are several viable alternatives, and we encourage
refinery owners and operators to employ a method tailored to their facility and preferences. We
note that while the final standards do not require an SSM plan, we note that an upset of an amine
scrubbing system or SRP is likely to cause a release of 500 lb/day SO2 from a fuel gas
combustion device and require a RCA and corrective action. (If the fuel gas combustion device
is an affected flare, the upset could trigger the 500,000 scfd flow rate RCA as well.)
Comment: Commenter 0154 stated that an additional Claus sulfur recovery train should
not be a method of controlling sour gas flaring, as an additional Claus unit is not a cost-effective
option. According to the commenter's estimates, the cost-effectiveness would exceed
$11,000/ton SO2. Commenter 0150 also supported EPA's conclusion that redundant SRP are not
cost-effective and stated that periods of SRP maintenance must be considered in the final rule to
avoid inadvertently requiring redundant SRP (the commenter referenced an NPRA paper from
Sept. 17, 2006).
Response: Again, we only provided examples of what a refinery could do and did not
mandate a specific course of action in the proposed standards. We believe an additional Claus
unit is cost-effective, especially compared to other alternatives, but again, each refinery can
choose its course of action. While $11,000 per ton may be unacceptable for some pollutants, it is
not entirely unacceptable for SO2 control. Additionally, the cost-effectiveness is highly
dependent on the frequency and duration of upsets, which is difficult to project. If there is an
extended outage, the cost-effectiveness for the additional Claus unit could be less than
$ 1,000/ton. The refinery owner or operator may choose any action to minimize emissions during
an upset, keeping in mind that if an upset or malfunction causes a release large enough to trigger
RCA for either a flow rate or SO2, additional action may be required.
Comment: Commenter 0125 stated that EPA should specifically address switching of the
amine type in the fuel gas treatment (amine) system as it relates to a modification under subpart J
(or subpart Ja). The commenter recommended that amine type switching be excluded from what
is considered a modification of the amine system.
102
-------
Response: We do not believe the situation described by the commenter should be an
issue. In subpart J, the amine system is not affected facility, so we are unclear why the definition
of a modification is relevant. As stated previously, the amine system is a control system and it
must operate in a manner so as to achieve the SO2 emission standard for fuel gas combustion
devices or alternative fuel gas standards. Amine type switching is allowed so long as the new
amine type can achieve the required SO2 emission standard for fuel gas combustion devices or
the alternative fuel gas standards.
7.5 Root Cause Analysis
Comment: Commenters 0126, 0127, and 0154 supported the RCA requirements for SO2
releases greater than 500 lb/day. Commenter 0154 conditioned their support of RCA on the
provision that events less than this threshold are deemed compliant with the provisions of the
subpart.
Commenter 0130 stated that EPA has not justified the 500 lb SO2 cutoff and the threshold
should be lower for a number of reasons, including the health and safety of refinery workers.
EPA has not explained its rationale for casting the threshold in terms of lb/day releases, and the
threshold should be one of total S02 released, not lb/day. A release less than 500 lb/day could go
on for days, weeks, or years without triggering the requirement to perform a RCA.
Response: The 500 lb/day RCA threshold was justified on the basis of costs in the
preamble of the proposed rule. Quite simply, it is not cost-effective to perform a RCA to attempt
to prevent each and every emissions event. The mass of emissions potentially reduced by
investigating small releases does not justify the costs associated with the investigation. The flare
management and monitoring requirements are expected to reduce long-term, on-going releases.
As flow monitors are required for each affected flare, a release that lasts for a long period of time
will be documented, and the refiner should be prepared to reconcile that release with the general
obligation to minimize emissions where practicable. As previously explained, we reject the
suggestion that all emissions less than 500 lb/day SO2 are compliant with the subpart.
Comment: Commenter 0127 stated that monitoring of the flares for flow and
composition is needed so that the facility can adequately determine if the 500 lb/day SO2
threshold has been exceeded.
103
-------
Response: We agree that the flow rate must be monitored. However, as discussed
previously, it is very difficult and expensive to install and operate monitoring systems capable of
quantifying the range of flows and composition expected from a flare. The required flow
monitor is expected to serve as a means of determining when an emission event occurs but may
not be appropriate to quantitatively determine high flow rates. Engineering calculations are
expected to provide reasonable accurate SO2 emission estimates for the purposes of the RCA
threshold.
Comment: Commenter 0149 supported requiring RCA of flaring and other venting
releases greater than 500,000 scfd VOC. Commenter 0126 recommended that RCA be required
for hydrocarbon flaring events because: (1) flare destruction efficiency can be significantly less
than 98 percent if the heating value, velocity, or steam flow to fuel gas ratio are outside of
optimal operating ranges; (2) flares are significant sources of greenhouse gases (GHG); (3) VOC
emissions can still be large even when the flare achieves 98 percent destruction efficiency; (4) by
requiring RCA, the RCA findings will be available to EPA; and (5) EPA has had a policy of
considering investigation of hydrocarbon flaring events to be "good air pollution control
practices." The commenter suggested RCA threshold levels of 100 lb per day of HAP, 5,000 lb
per day of VOC, or 500,000 scfd of flare gas. Commenter 0127 also suggested an RCA be
conducted when the volume of gas released exceeds 500,000 scf per event (in addition to the 500
lb/day SO2 RCA threshold). In contrast, Commenter 0154 disagreed that EPA should require
RCA for hydrocarbon flaring events because it is more difficult to estimate the emissions (some
disagreement regarding the destruction efficiency of the flare) and the hydrocarbon releases are
well-controlled with the flare.
Response: As we stated in the preamble to the proposed rule, we found little direct
environmental benefit from performing RCA for hydrocarbon flaring events because the flares
are efficient control systems for hydrocarbons. However, we do agree that determining the cause
of large releases can reduce those releases in the future. Based on our analysis, RCA of flaring
events in excess of 500,000 scfd will pay for itself in reducing lost product or in offsetting
natural gas purchases. Therefore, the final standards include a requirement to conduct a RCA if
a flow of more than 500,000 scfd is sent to an affected flare.
104
-------
7.6 Delayed Coking Depressurization
Comment: Commenter 0156 provided examples of coker depressurization practices that
use set points or timed cycles that route depressurization gas to the atmosphere at about 10 psig;
the commenter estimated it would cost $12-million in capital, achieving 182 tons/yr S02
reduction, which yields a cost-effectiveness value much higher than that reported by EPA.
Commenter 0176 noted that their cost to meet the proposed standard could range from
$20- to $50-million and would depend on whether: (1) additional compressors would be needed;
(2) pumps could operate down to 5 psig; and (3) the blowdown gases between 5 and 10 psig
would need additional treatment to be useable in the fuel gas system. The commenter also stated
that vents from coking units are difficult to test due to the high water content of the emissions
stream, the short-term nature of the emissions, and the lack of a discrete stack to test. The
commenter provided test data, protocols, and reports that detail the concerns with testing a coker
vent. The data included estimates of VOC emissions for a coking unit the size of the
commenter's ranging from 1.3 to 10.6 tons per year; the commenter noted that SO2 results were
omitted "because of perceived significant 'low bias' in the results." Because the best case cost-
effectiveness values are $9,400/ton S02 (using EPA's emission estimates) and $178,000/ton
VOC, the commenter stated that EPA cannot determine that controls on coker vents is BDT.
Response: While we recognize that the final standards must be technically feasible for all
units, we note that the standards do not have to be requirements that that every existing unit can
meet without any changes to the current operation or controls. We recognize that some existing
units will become affected facilities, but this will be due to a process modification or
reconstruction. When the process is being modified or reconstructed, additional piping and other
equipment modifications can be made to the unit to allow depressurization down to 5 psig. Once
we determine that this requirement is technically feasible, we must also demonstrate that it is
cost-effective. Several of the commenters suggested that depressurization down to 10 or 20 psig
should be acceptable prior to releasing the gases to the atmosphere. In our previous analysis, we
assumed small quantities of VOC reductions from the assumed higher efficiency of process
heaters and boilers versus flares. However, when we evaluate the cost-effectiveness of vessel
depressurization down to 5 psig, assuming the baseline is depressurization down to 15 psig (the
depressurization point above which a delayed coker vent is a miscellaneous process vent in
Refinery MACT I ) and then venting to the atmosphere, we find the 5 psig requirement to be
105
-------
cost-effective for SO2 control (between $1,300 and 5,300/ton SO2 reduced). While the VOC
reduction is minimal, based on our impact estimates for the DCU depressurization work practice
standards, a 5 psig DCU vessel depressurization limit was determined to be BDT.
Comment: Commenter 0154 requested that EPA provide a proposal regarding how
compliance with the 5 psig cutoff would be demonstrated. The commenter also indicated that
compressing the coker blowdown gases may not be cost-effective depending on the sulfur
content of the gases. The commenter indicated that, in addition to compression, "additional
amine adsorption and regeneration equipment and even additional sulfur train capacity may be
required to handle coker blowdown gases. By the time the decoking sequence gets to the venting
stage (i.e., out of the fractionator and out of the scrubber), the concentration of VOC and or
sulfur species is minimal..." Commenter 0154 also asserted that the coker depressurization gas
is not suitable for fuel and requested an option by which an alternate operating limit or parameter
can be established when where it is "problematic to vent delayed coker gas at less than 5 psig."
Response: We provided a portion of the comment verbatim because we are confused by
the commenter's statement. In one sentence, the commenter suggested that there is so much
additional sulfur in the additional coker depressurization gas recovered when depressurizing to
5 psig that additional sulfur recovery capacity would be needed in order to treat this incremental
amount of gas; in the very next sentence, the commenter stated that the concentrations of VOC
and sulfur species are minimal. Therefore; we are not entirely clear on the intent of the
comment. We recognize that the cost-effectiveness of this depressurization requirement is
highly dependent on a number of factors, including current depressurization set point, VOC and
SO2 concentration in the vessel prior to discharge to the atmosphere or flare, and availability of
adequate sulfur removal capacity to handle the additional gases. As such, there will be instances
when the cost-effectiveness of the requirements for DCU depressurization (or other requirements
in this final rule) may exceed some perceived cost-effectiveness threshold when evaluated for an
individual unit. However, we are developing standards of performance for new units; we are not
attempting to develop standards that the worst-performing existing unit can meet with status quo
operations. We evaluate the total costs and total emission reductions of the control options on a
nationwide basis. Based on our analysis, we conclude that coking unit vessel depressurization
down to 5 psig is a cost-effective control strategy for both VOC and SO2 on a nationwide basis
and is BDT.
106
-------
Chapter 8
SMALL BUSINESS CONCERNS
Comment: Commenter 0129 noted there are 33 small refiners operating 41 refineries that
fall within the small business definition, and EPA must observe the Small Business Regulatory
Enforcement Fairness Act of 1996 (SBREFA), 5 U.S.C. §609 to ensure that the interests of small
business entities receive due attention in the consideration and adoption of this agency rule. The
commenter highlighted the special status of small business refiners in that: (1) small business
refiners are important to the economy; (2) small business refiners have limited resources; (3)
small business refiners do not enjoy economies of scale; and (4) small refiners are generally
located in attainment areas, (30 of 41 small refineries, or 73 percent, are in attainment areas).
Response: SBREFA requirements are triggered when there is (or is expected to be) a
significant economic impact on a substantial number of small entities. While other indicators
may also be used, the economic impact is typically evaluated as a ratio of the annualized
regulatory costs to the annual revenue of the small entity. The number of small entities with
cost-to-revenue ratios exceeding certain thresholds (typically 1 or 3 percent) is determined to
assess whether there are significant economic impacts on a substantial number of small entities.
As no small refineries were expected to have costs in excess of 1 percent of their revenues, no
SBREFA panel was needed. However, we did evaluate the impacts for small refiners separately
in many cases, including small SRP and process heaters, to reduce the burden on small refiners
when possible.
Comment: Commenter 0129 noted that the model refinery sizes used by EPA in its cost
analysis are much larger than the average small refinery and the rule will impact many more
small refiners than large refiners. The 41 small refineries have capacities that range from 2,000
to 116,000 barrels per day (BPD) with an average of 31,000 BPD, compared with EPA's model
size of 143,000 BPD. The average small refiner FCCU is 20,800 BPD compared to EPA's
model FCCU of 50,000 BPD. The commenter also stated that the regulatory impacts do not
accurately reflect the number of small refiners that will have to install controls to meet
107
-------
subpart Ja. For FCCU and FCU, EPA assumes 76.5 percent of units are currently subject to EPA
Petroleum Refinery Enforcement Initiative and these refiners will not need to install controls to
meet subpart Ja; in contrast, of the 41 small refiners, only 4 refineries or 9.8 percent are subject
to a consent decree. Therefore, the commenter stated, a much higher percentage of small
refineries will have to install controls, which cost more on a per barrel basis than EPA estimates.
In addition, small refiners that have the ultra-low sulfur diesel delayed compliance options may
be installing new desulfurization capacity to meet the ultra-low standards. The commenter urged
EPA to revise the impacts analysis to account for the disproportionate amount of pollution
controls that will be required to be installed by small refiners.
Commenter 0129 also provided costs, emission reductions, and cost-effectiveness for
small refiners. The cost effectiveness for using wet scrubbers to meet the proposed PM and S02
emissions standards for FCCU at three small refiners ranged from $5,600 to $17,000 per ton,
compared with EPA's estimate of $1,900 per ton for the overall industry average model. The
cost effectiveness for using ULNB on process heaters at five small refiners ranged from $9,800
to $100,000 per ton compared to EPA's estimate of $1,600 per ton. In these instances, EPA has
clearly underestimated the capital and operating costs and the cost effectiveness.
Response: In response to these and other comments, we significantly revised our impacts
estimates. For the FCCU and FCU impacts, we evaluated costs on a per unit basis so that the
costs were developed based on the size of actual units. For FCCU, we directly included consent
decree requirements on a FCCU-specific basis and performed a Monte Carlo analysis in which a
number of randomly selected FCCU were included as affected facilities to develop our best
estimate of the average cost and environmental impact of these final standards. We also
developed additional model plants by which to evaluate a greater range and more accurate
distribution of process heater sizes. Various technical memoranda in Docket ID EPA-HQ-OAR-
2007-0011 provide additional information.
Comment: Commenter 0129 stated that EPA should not require controls for small
refiners or should include less stringent controls with reasonable costs for small refiners. This is
consistent with the NSPS for refineries for Claus Sulfur Recovery Units at 40 CFR 60.100(a)
where existing units greater than 20 LTD are required to be controlled and where the proposed
regulation has a two-tiered system based on capacity with one requirement for large units and
another for small units. The commenter provided suggested cutoffs for FCCU controls for PM,
108
-------
SO2, and NOx only for FCCU units with current maximum capacity of 30,000 BPD or more and
for FCU with input capacity of 20,000 BPD or more. The suggested process heater cutoff should
be process heaters or boilers with a heat input of 40 MMBTU/hr or more, consistent with EPA's
current policy and even the consent decrees. The sulfur recovery plant cutoff should be sulfur
recovery units with a capacity of 20 LTD or more. In addition, the commenter stated that the
new SO2 fuel gas limits and work practice standards for fuel gas production units should apply to
refineries with crude input capacity of 100,000 BPD or more. If EPA does not elect to invoke
the Small Business Advocacy Review Panel process, then EPA should use the data provided by
the small refiners to provide less stringent control options commensurate with the true
incremental costs to small refiners.
Response: In developing the impact estimates for this final rule, we did account for
actual differences in refinery sizes. Although there are economies of scale that generally make
the controls for larger refineries more cost-effective than those for smaller refineries, we did not
identify significant adverse costs for controlling smaller FCCU. After including additional
model plants, which provided a greater range and more accurate distribution of process heater
sizes, and making other revisions to the NOx control cost estimates, we agree with the
commenter that it is not cost-effective to require NOx controls for process heaters less than
40 MMBtu/hr. As such, we have changed the applicability threshold of the NOx emission limits
for process heaters to those process heaters with capacities greater than 40 MMBtu/hr. In
addition, the final standards include a less stringent emission limit for small SRP.
109
-------
Chapter 9
NSPS STRINGENCY
Comment: Commenter 0130 stated that section 111 of the CAA requires that NSPS
reflect the best emission reductions which are feasible {i.e., the BDT). The NSPS must be at
least as stringent as the requirements in consent decrees, whether those limits are technology-
based or emissions-based, as refiners covering approximately 84 percent of refining capacity are
under consent decrees. Any new NSPS provision under either subpart J or subpart Ja which is
not at least as stringent as the most stringent consent decree cannot be justified by EPA.
Commenter 0149 stated that considerable technological progress has occurred in the last
decades and it is appropriate that EPA revise and supplement the standards. The commenter
supported some of the revisions to subpart J and the addition of new subpart Ja standards and
suggested that others be strengthened to reflect consent decrees. The commenter stated that
section 111 of the CAA requires EPA to look "toward what may be fairly projected for the
regulated future." The court decision in Portland Cement I (486 F. 2d 375 at 384 (D.C. Cir.
1973)) stated that section 111 of the CAA does not require that any cement plant currently now
in existence be able to meet the proposed standards.
Commenter 0142 noted that the CAA specifically requires that rules resulting from
review of NSPS be both technically feasible and cost effective. Industry would expect that there
would be relatively few changes in the review revisions because there have been: (1) relatively
little change in the available control technology options; (2) dramatic increases in labor and
material costs, including impacts from Hurricane Katrina; and (3) dramatic emissions reductions
from NSPS, NESHAP, and NSR controls.
Response: The evaluation of BDT must consider both the technical feasibility and the
costs of the alternative emission limits. As several commenters pointed out previously, the
consent decrees were not generally established using cost as a major factor. As such, some
consent decrees may have lower emission limits (particularly for NOx from selected FCCU), but
we found those lower limits to have unacceptably high costs, especially considering the
110
-------
increment to establishing NOx limits at higher concentrations. On the other hand, this final rule
quite often establishes emission limits that are comparable to the most stringent of the consent
decrees. Thus, although we cannot rely solely on the consent decree provisions as BDT, the
consent decrees can often be used to confirm that the emission limits are achievable.
Comment: Commenter 0142 stated that EPA has not correctly performed best
demonstrated technology and cost-effectiveness calculations for subpart Ja.
Response: As described in the comment responses in the specific emission sources
chapters, the cost impacts were significantly revised to provide more accurate costs and cost-
effectiveness estimates in our assessment of BDT.
Ill
-------
Chapter 10
OTHER COMMENTS
Comment: Commenter 0154 requested that subpart Ja specifically state:
"Notwithstanding any provision of this subpart, projects commenced due to a New Source
Review Consent Decree entered into on or before May 14, 2007 are exempt from the
requirements of this subpart Ja." According to the commenter, the consent decree "commenced"
when it became a binding contract; therefore, these projects should not be subject to subpart Ja.
Commenter 0150 concurred with this request and provided alternative regulatory text; the
commenter also requested EPA publish a Federal Register notice as soon as practicable
codifying this "clarification." Commenter 0139 urged EPA to include specific language in
subpart Ja that modified and reconstructed FCCU are subject only to the existing subpart J
requirements as follows: "Any FCCU that commences modification or reconstruction after May
14, 2007 shall be subject to the standards in 40 CFR part 60, subpart J and exempt from the
provisions of this subpart."
Commenter 0154 stated that any substantive proposed changes must apply only to new,
modified, or reconstructed facilities after May 14, 2007; substantive new requirements cannot be
applied retroactively. The commenter also noted that since no impacts or basis was included, no
revisions to 40 CFR part 63, subpart UUU can be made (i.e., none of the work practice standards
or new emission limits can be imposed under subpart UUU without additional rulemaking).
Commenter 0131 noted that requiring subpart Ja to apply retroactively for petroleum refineries
for which construction, reconstruction, or modification commenced after May 14, 2007 is a
violation of due process, and regulated entities would be required to spend money to comply
with subpart Ja while not knowing what the promulgated requirements will be or when the rule
will be promulgated.
Commenter 0130 opposed EPA's proposal that FCCU and FCU that begin modification
or reconstruction remain subject to the emission limits and requirements in subpart J rather than
requirements in subpart Ja. Subpart Ja should apply to all modified, reconstructed, and new
FCCU and FCU, and EPA has provided no justification for exempting these sources from
112
-------
modern, current BDT. The purpose of NSPS is to attain and maintain ambient air quality by
ensuring the best demonstrated emission control technologies are installed as the industry is
modernized. For oil refineries, modernization comes largely from modification and
reconstruction, as no new oil refinery has been constructed in 30 years.
Response: The CAA section 111 does not provide the flexibility in applicability dates
requested by a number of the commenters. The final standards were determined by evaluating
the impacts on new sources separately from modified and reconstructed sources, and the final
emission limits and work practices were established accordingly.
Comment: Commenter 0154 stated that the proposed recordkeeping and reporting
requirements impose a significant burden with no environmental benefit and suggested these
requirements be consistent with the current subpart J.
Response: We have determined that the recordkeeping and reporting requirements
included in the final standards are necessary to determine compliance with the standards.
Comment: Commenter 0116 stated that the proposed NSPS for refineries falls short of
international community-intended goals and suggested EPA work with the Kenya government to
conduct efficacy tests to select the best options for improving energy efficiency at petroleum
refineries.
Response: It is unclear how working with Kenya, with its one refinery, will help the
Agency identify the best options for improving energy efficiency at petroleum refineries for the
150 U.S. refineries. It is also unclear what international community-intended goals the proposed
NSPS is falling short of. The subpart Ja standards are generally the most stringent federal
standards for petroleum refinery operations of any in the world with respect to PM, SO2, NOx
and CO. The commenter appears to be confused regarding the scope of the NSPS program as
authorized by the CAA.
Comment: Commenter 0125 suggested the following clarifications to the applicability of
subpart J and Ja standards. Subparts J and Ja should not apply to: (1) cogen plants that meet the
definition of an Electric Generating Unit; (2) marine vessel loading operations not
adjacent/contiguous to a refinery; (3) sulfur recovery plants not adjacent/contiguous to a refinery;
and (4) green petroleum coke calciner units, regardless of whether or not they are
adjacent/contiguous to a refinery.
113
-------
Response: Item 1 - the only way a cogeneration plant would be subject to J or Ja is if it
is on-site at the refinery and if it burns refinery fuel gas. We see no reason why a fuel gas
combustion device, regardless of its output (boiler steam or electricity), needs to burn sour fuel
gas. Item 2 - we are not sure how a marine vessel loading operation not co-located at a refinery
is affected by subpart J, especially since we now exclude marine vessel loading vapors for the
definition of fuel gas. Item 3 - both subparts J and Ja specify that the sulfur recovery plants is an
affected facility regardless of whether or not the sulfur recovery plant is located within the
boundaries of the refinery. Item 4 - petroleum coke calciners generally produce a gas that is
then combusted. If the calciner is at the refinery, then the generated gas meets the definition of
fuel gas and the device combusting this gas is a fuel gas combustion device that would be subject
to the NSPS standards if it is new, modified, or reconstructed. If the calciner is off-site, the gas
would not be generated at a petroleum refinery and would not be subject to subpart J or Ja.
Comment: Commenter 0125 recommended that EPA provide regulatory exemptions
from emission limits for control device maintenance, stating that less emissions typically occur
when bypassing the control device than during the shutdown and start-up of the associated
process unit as well as safety concerns associated with start-up and shutdown events.
Response: We have provided an exemption for sulfur pit maintenance; any other
exemptions would need to be evaluated on a case-by-case basis.
Comment: Commenter 0125 requested clarification of the applicability of Appendix F to
the subpart J and Ja standards [other than those spelled out in §60.105(a)(12)].
Response: Proposed subpart Ja included a number of paragraphs specifying that the
owner or operator must comply with the quality assurance requirements of procedure 1 in 40
CFR part 60, appendix F, and the final subpart Ja standards also include that requirement for
most of the CEMS required for compliance.
Comment: Commenter 0125 requested clarification on how emissions that may be
released at multiple discharge points should be treated in calculating the long-term average; the
commenter suggested a flow-weighted average be used.
Response: We agree that a flow-weighted average should be used in calculating the long
term average when emissions are discharged at multiple points; in fact, this is clearly specified
for SRP.
114
-------
Comment: Commenter 0121 requested clarification on how compliance with the long-
term (7- and 365-day) emission limits should be determined when a start-up, shutdown, or
malfunction activity occurs during the averaging time. Commenters 0125 and 0161 suggested
that EPA provide specific protocols for missing data in calculating the 365-day average.
Commenter 0125 requested clarification regarding how much data are necessary to calculate the
hourly averages of operating limits. Language regarding 75 percent of the hours, as in
§63.1572(c)(3) would be helpful in §60.105a(b)(a)(i) .
Response: Process upsets are exempt from the emission limit as specified in the NSPS
General Provisions and should not be included in the long-term average.
Comment: Commenter 0125 stated that oxygen analyzer spans should not be set within
the rule but instead by the operator to match the process conditions at a specific application.
Response: We have provided a range of oxygen analyzer spans, from 10 to 25 percent, in
the final standards. The operator may set the span to match the process conditions as long as it is
a value within that range.
Comment: Commenter 0125 stated that units that operate properly at high excess
air/oxygen (e.g., thermal oxidizers) are penalized by the 0 percent O2 correction, and EPA should
provide equitable treatment for these types of equipment.
Response: If a thermal oxidizer is operating with high excess air, the excess air acts to
dilute the SO2 or NOx concentration in the vent stream. The rule does not prevent the use of
higher excess air rates, but the 02 correction is needed to ensure dilution air is not used to meet
the concentration emission limit. An owner operator considering a thermal oxidizer should
consider whether that device will comply with the emission limits including an O2 correction in
that decision.
Comment: Commenter 0125 suggested that EPA grandfather existing CEMS that have
been certified under federally-enforceable rules, even if the certified analyzer parameters (e.g.,
span) differs from those specified in subpart Ja. Commenter 0161 requested clarification of how
to use CEMS data that exceed the certified span when calculating the rolling average (e.g., if the
CEMS says 800 ppmv but the span is 300 ppmv, what value should be used in the rolling
average calculation); the commenter indicated that EPA has established a position in numerous
applicability determinations of using the maximum of the span (300 ppmv rather than 800 ppmv
for the above example).
115
-------
Response: We have specified the required span for each monitor in the final standards.
Equipment developed for other performance specifications may or may not be appropriate for
this rule, and it should not be force fit. Sources with CEMS data outside the certified span
should report CEMS out-of-control and declare their compliance status as other than continuous.
Comment: Commenter 0161 requested clarification of EPA's interpretation of the
April 14, 2000, letter from Judith Katz to Region 3. The April 14, 2000, interpretation suggests
12-hour rolling averages should be computed continuously {i.e., every minute), but the proposed
rule suggests hourly computations. Specifically, the commenter asked if an hour used for
averaging purposes begins at the top of that hour or anytime; the commenter noted that
depending on EPA's interpretation, a 180-minute average might be more clear than a 3-hour
average.
Response: Requiring 12-hour averages 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 using 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.
Comment: Commenter 0125 noted that the cross-reference in §60.105a(b)(3), which
references §60.104a(d)(4)(vii), must be incorrect because paragraph (d)(4) only goes up to (v).
Response: The commenter is correct that the reference was in error; the proposed
equation for coke burn-off rate was located in §60.104a(d)(4)(iii). In the final standards,
§60.105a(b)(l)(iv) correctly refers to the coke burn-off rate equation in §60.104a(d)(4)(iii).
Comment: Commenter 0150 stated that North Slope Topping Plants should be expressly
exempt from NSPS subpart Ja. These facilities do not have many of the affected sources covered
by the rule, only a few fuel gas combustion devices. According to the commenter, it would not
be cost-effective to install amine units (and SRP) for these facilities.
Response: The requirements of subpart Ja would only become applicable if a unit is
modified, reconstructed, or newly constructed. Subpart J has similar fuel gas standards with no
exclusion, so absent subpart Ja, the North Slope Topping Plants would need to install amine units
and SRP when a unit became subject to subpart J. As such, we do not see that the additional
requirements imposed by subpart Ja are not cost-effective.
116
-------
Comment: Commenter 0150 stated that the compliance date for projects commencing
construction between the proposal date and final promulgation date should be 3 years after
promulgation with an allowance for the refinery to request an additional 2-year extension to
avoid an unscheduled turnaround.
Response: The CAA section 111 does not provide the flexibility requested by the
commenter.
Comment: Commenter 0150 stated that proposed timeframes for monitoring system
installation and operation are infeasible; 365 days should be allowed for engineering, ordering,
installing, and testing of monitors. At minimum, 180 days should be allowed with an allowance
for an extension request.
Response: A unit only becomes an affected facility when it is modified, reconstructed, or
newly constructed. When those projects are being planned, the refinery has adequate time to
also engineer, order, and install a monitor. In the preamble to the final standards, we addressed
the situation in which a refinery makes a process change that may result in a fuel gas no longer
being able to demonstrate compliance with the low sulfur exemptions; for that situation only, the
owner or operator may conduct daily monitoring until a CEMS is installed for up to 180 days.
We have not changed the time frames for monitoring for any other process units.
Comment: Commenter 0150 suggested that, due to the deficiencies in the impact
analyses, EPA should only finalize the clarifications to subpart J (after deleting any non-elective
substantive requirements) and re-propose subpart Ja after consideration of the public comments.
Response: We have revised our impact analysis in response to public comments. We do
not see a need to re-propose.
Comment: Commenter 0156 stated that only Methods 11, 15, 15a, or 16 are needed in
§60.104a(j) (i.e., references to Methods 1, 2, and 3 are not needed), which would be consistent
with the requirements in subpart J.
Response: We disagree. Calculations contained in Methods 11, 15, 15a, or 16 rely on
data obtained according to Methods 1, 2, and 3
Comment: Commenter 0156 noted that the span value for H2S in fuel gas in
§60.107a(a)(2)(ii) was 425 ppmv and that the units are not consistent with subpart J. The span
value in subpart J is 425 mg/dscm (which is about 300 ppmv) H2S.
117
-------
Response: The commenter is correct that subpart J does include a span of 425 mg/dscm
in §60.105(a)(4)(i), and the wrong units were included in §60.107a(a)(2)(ii). The span value in
subpart Ja was revised to 320 ppmv in the final rule.
Comment: Commenter 0169 noted that 40 CFR 60.40b(c) (subpart Db) contains an
exemption for sources that are subject to subpart J, and the commenter requested that EPA
extend that exemption to subpart Ja as well.
Response: Specific language currently in Db is: "Affected facilities that also meet the
applicability requirements under subpart J (Standards of performance for petroleum refineries;
§60.104) are subject to the PM and NOx standards under this subpart and the SO2 standards
under subpart J (§60.104)." There are two specific types of boilers that are subject to Db. First,
there are general fuel gas combustion devices that are only subject to S02 standards under
subpart J (or Ja), and the language in subpart Db is specifically applicable to these units. Second,
there are CO or waste heat boilers associated with the FCCU. These boilers are subject to the
PM, NOx and SO2 limits in the final standards for subpart Ja and not those in subpart Db.
118
------- |