United States
Environmental Protection
Agency
Office Of Air Quality
Planning And Standards
Research Triangle Park, NC 27711
EPA-453/R-01-011
June 2001
FINAL REPORT
Air
     Petroleum Refineries:  Catalytic
   Cracking Units, Catalytic Reforming
    Units, and Sulfur Recovery Units -
       Background Information for
       Promulgated Standards and
          Response to Comments

               Final Report

-------
(This page is intentionally blank)

-------
                                             EPA-453/R-01-011
          Petroleum Refineries:  Catalytic Cracking Units,
      Catalytic Reforming Units, and Sulfur Recovery Units -
      Background Information for Promulgated Standards and
                       Response to Comments
                   U.S. Environmental Protection Agency
                 Office of Air Quality Planning and Standards
                 Waste and Chemical Process Group, MD-13
                     Research Triangle Park, NC 27711
                       Prepared Under Contract By:

                        Research Triangle Institute
                     Center for Environmental Analysis
                     Research Triangle Park, NC 27711
                             June 2001
D.
Ul
I/)

-------
                                      Disclaimer

This report has been reviewed by the Emission Standards Division of the Office of Air Quality
Planning and Standards of the United States Environmental Protection Agency and approved for
publication. Mention of trade names or commercial products is not intended to constitute
endorsement or recommendation for use. Copies of this report are available through the Library
Services (MD-35), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711,
or from the National Technical Information Services 5285 Port Royal Road, Springfield, VA
22161.
                                          11

-------
                            Environmental Protection Agency

Petroleum Refineries:  Catalytic Cracking Units, Catalytic Reforming Units, and Sulfur
Recovery Units - Background Information for Promulgated Standards and Response to
Comments

                                      Prepared by:
 Sally Shaver                                                       (Date)
 Director, Emission Standards Division
 U.S. Environmental Protection Agency
 Research Triangle Park, NC  27711

1.      The promulgated standards will regulate emissions of hazardous air pollutants (HAP)
       emitted from Petroleum Refinery process vents. Only those process vents that are part of
       major sources under section 112(d) of the CAA are regulated. The final standards will
       reduce emissions of several organic, inorganic, reduced sulfur, and metallic compounds
       identified in the CAA list of hazardous air pollutants.

2.      Copies of this document have been sent to the following Federal Departments: Labor,
       Health and Human Services, Defense, Office of Management and Budget, Transportation,
       Agriculture, Commerce, Interior, and Energy; the National Science Foundation; and the
       Council on Environmental Quality. Copies have also been sent to members of the State
       and Territorial Air Pollution Program Administrators; the Association of Local Air
       Pollution Control Officials; EPA Regional Administrators; and other interested parties.

3.      For additional information contact:

             Mr. Robert Lucas
             Waste and Chemical Process Group (WCPG)
             U.S. Environmental Protection Agency
             Research Triangle Park, NC 27711
             Telephone: (919) 541-0884

4.      Copies of this document may be obtained from:

             U.S. EPA Library (MD-35)
             Research Triangle Park, NC 27711
                                          111

-------
                           TABLE OF CONTENTS


Chapter                                                                 Page

LIST OF ACRONYMS	viii

LIST OF COMMENTERS  	x

1.0   CONTROL REQUIREMENTS FOR CCU CATALYST REGENERATOR VENTS
       	 1-1
      1.1    Subcategorization of Thermal CCU	 1-1
      1.2    Use of PM and CO as Surrogates for HAP	 1-1
      1.3    NSPS as MACT Floor	 1-3
      1.4    Extension of the Compliance Date	 1-4
      1.5    Hg Emission Control	 1-6
      1.6    Fabric Filters as MACT Floor or Beyond the Floor Technology  	1-8
      1.7    Ni Alternative Standard (Ibs/hr) 	 1-9
      1.8    Ni Alternative Standard (Ibs per 1,000 Ibs of coke burn-off)	  1-12
      1.9    Use of the Arithmetic Mean, Median, Geometric Mean, 90th Percentile
            Value, or Highest Value as the Representative Concentration Used
            in the Factor for Conversion of PM to Ni  	  1-15
      1.10   Format of HAP Metal Standards	  1-17
      1.11   Level of Proposed CO Limit for HAP Organics	  1-18
      1.12   Format of Proposed CO Limit for HAP Organics  	  1-20
      1.13   Organic HAP Alternative to Proposed CO Standard 	  1-21
      1.14   SOX and NOX Emission Control 		  1-21
      1.15   Control of D/F Emissions 	  1-22
      1.16   Allowance or Exclusion of Emissions Created by CO Control Device
            	  1-23
      1.17   Pollution Prevention Technologies for PM and HAP Metals from FCCU
            	  1-24
      1.18   Compliance Date for Retrofit Particulate Controls	  1-24
      1.19   Hydrotreatment of CCU Feed	  1-25
      1.20   Other Associated Emission Sources 	  1-27

2.0   CONTROL REQUIREMENTS FOR CRU CATALYST REGENERATOR VENTS
       	 2-1
      2.1    MACT floor for Semi-Regenerative CRU vs Exclusion of Control
            Requirements  	 2-1
      2.2    Three Percent O2 Correction 	 2-2
      2.3    CRU Cutoff for Depressurization and Purging  	 2-3
      2.4    Clarification of TOG Standard	 2-3
      2.5    Achievability of HC1 Limits Based  on Scrubbers  in the Steel Pickling
            Industry	 2-4
      2.6    Evaluate D/F Emissions  	 2-4

                                    iv

-------
      2.7   Format of Proposed Standard for Organic HAP 	  2-5
      2.8   New Combustion Technologies	  2-6
      2.9   Opacity Limit for Flares	  2-6
      2.10  Requirements for Final Purge Vent	  2-6
      2.11  Alternative Standard for Inorganic HAP Emissions	  2-6

3.0   CONTROL REQUIREMENTS FOR SRU	  3-1
      3.1   NSPS as MACT Floor	  3-1
      3.2   Parallel Unit Configurations as MACT Floor or NSPS	  3-2
      3.3   Off-site Sulfur Plants  	  3-3
      3.4   Hydrotreatment as MACT Floor  	  3-3
      3.5 Thermal Oxidizers for NSPS SRU	  3-4
      3.6   Consistent Definition of TRS	  3-4
      3.7   Format of Proposed Standard	3-5
      3.8   Incinerator for TRS Control	  3-5
      3.9   Calculation of TRS Limit 	  3-6
      3.10  Sulfur Recovery Pits, Stretford Solution Tanks, and Non-Glaus SRU
             	  3-6
      3.11  NSPS Exemption for Small SRU	  3-7

4.0   CONTROL REQUIREMENTS FOR BY-PASS LINES	  4-1
      4.1   Flow Meter Alternatives for CCU Regenerator By-Pass Lines	4-1
      4.2   Installation Requirement for Flow Meter  	  4-2
      4.3   Continuous Monitoring Option for CCU By-Pass Valves	4-2

5.0   MONITORING REQUIREMENTS	  5-1
      5.1   CO Boiler Monitoring Requirements for Full-Burn FCCU	5-1
      5.2   Process Data and Secondary Measurement Devices As Alternatives to
            Flow Monitoring Requirements for Wet Scrubbers  	  5-3
      5.3   Monitoring Requirements for Other Wet Scrubber Designs 	5-4
      5.4   Continuous O2 Monitor for Full Burn CCU Regenerators  	5-5
      5.5   Monitoring of Uncontrolled CCU 	  5-5
      5.6   Repeat Performance Tests for CCU Catalyst Regenerators Subject to Ni
            Alternative	  5-7
      5.7   Calibration of Temperature Measurement Device for a Boiler or Process
            Heater Less than 44 MW Where  the Vent Stream is Not Introduced into
            the Flame Zone	  5-8
      5.8   Monitoring Exemption for a Boiler or Process Heater Greater than 44
            MW Heat Input	  5-8
      5.9   Accuracy and Calibration Requirements for CCU and CRU with Wet
            Scrubbers	 5-10
      5.10  Monitoring Requirements for Catalytic Incinerators	 5-10
      5.11  Monitoring Requirements for CRU With Internal Scrubbers	5-11
      5.12  Continuous Emission Monitoring 	 5-12
      5.13  Annual Stack Tests	 5-13
      5.14  ESP Plate Area and Conditioning Agents  	 5-13

-------
      5.15  Opacity Monitoring for Non-NSPS CCU	  5-14
      5.16  Daily Averages for Monitoring Systems 	5-14
      5.17  Draeger Tubes for Monitoring of Scrubbers 	  5-15
      5.18  Method 26 vs Method 26A for HC1 Emissions from CRU 	5-15
      5.19  Monitoring Requirements for SRU without Combustion Device ....  5-16
      5.20  Monitoring Requirements for Flares 	  5-16
      5.21  Exceedances and Excursions	5-17

6.0   PERFORMANCE TEST REQUIREMENTS	  6-1
      6.1    Determine Reactor Pressure  During Performance Test  	6-1
      6.2    Determine Maximum Amount of Recycle During Performance Test  .  6-1
      6.3    Determine All HAP Metals During Performance Test  	6-1
      6.4    Conditions Requiring New Performance Tests	  6-2
      6.5    Test Conditions for CCU Regenerator Vent	  6-3
      6.6    Use of Engineering Analysis to Establish Limits for Process/Operating
            Parameters	6-3
      6.7    Performance Test for Organic HAP from CCU	  6-4
      6.8    Early Compliance Certification	6-4
      6.9    Equation 2 for Calculating Coke Burn-off Rate	  6-5
      6.10  Alternative Coke Burn Rate Equation	6-5
      6.12  Method 5B and 5F for PM 	  6-5

7.0   STARTUP, SHUTDOWN, MALFUNCTION, AND MAINTENANCE	  7-1
      7.1    Provisions for Planned Maintenance	  7-1
      7.2    Storage of Intermediate Products Through Duration of Maintenance
             	  7-3
      7.3    Maintenance Plan Requirements	  7-3
      7.4    HAP Emissions from Startup, Shutdown and Upset Conditions	  7-3
      7.5    Reporting Malfunction Events	  7-4

8.0   RELATIONSHIP TO NSPS AND OTHER RULES	  8-1
      8.1    Relationship of NSPS to MACT Standard	8-1
      8.2    Definition of Affected Facility vs Definition of Affected Source	8-4
      8.3    Triggering the NSPS due to Emissions from Flares and Combustion
            Devices 	  8-5
      8.4    State or Local Requirements  	  8-5

9.0   IMPACT ANALYSES	  9-1
      9.1    Ni Emission Estimates for CCU Regenerator Vents  	9-1
      9.2    Emission Estimate Methodology	  9-2
      9.3    Selection of Pollutants in Database	  9-3
      9.4    Impacts of Additional Pollutants	  9-3
      9.5    Cost Estimates for CCU Catalyst Regenerator Vents	9-4
      9.6    Cost Effectiveness Estimates for CCU Catalyst Regenerator Vents  . . .  9-4
      9.7    Low Health Risk Does Not Warrant Proposed Controls	9-5
      9.8    Database Weaknesses	  9-6

                                     vi

-------
      9.9   CRU Catalyst Regenerator Vent Emission Rates	  9-7
      9.10  CRU Emissions Table	  9-7
      9.11  Effect of Hydrotreating on Emission Estimates	  9-8
      9.12  Effect of ESP Collection Area on Cost Estimates  	  9-9
      9.13  Effect of Scrubber Pressure Drop on Costs  	  9-9
      9.14  Effect of CO Monitoring on Compliance Costs for Full Burn CCU . . 9-10
      9.15  Selection of Control Equipment for Costing	 9-10
      9.16  Energy Impacts for Incinerators	9-11
      9.17  Additional Environmental Impacts from Bioaccumulative HAP .... 9-11
      9.18  Economic Analysis	9-12

10.0  ADMINISTRATIVE REQUIREMENTS	 10-1
      10.1  Executive Order 13045 on Children's Health	 10-1
      10.2  Executive Order 12898 on Environmental Justice 	 10-2
      10.3  Executive Order 12866	 10-3

11.0  MISCELLANEOUS	 11-1
      11.1  Section 112(j) MACT Hammer	 11-1
      11.2  Notification Requirements	 11-1
      11.3  Reporting Requirements	 11-2
      11.4  Implementation of Final Rule  	 11-2
      11.5  Source-specific Regulatory Approach	 11-3
      11.6  Language Clarification	 11-3
                                     Vll

-------
                            LIST OF ACRONYMS
API	  American Petroleum Institute
bbl  	  Barrel
BID	  Background information document
bpd	Barrels per day
CAA	  Clean Air Act
GARB	  California Air Resources Board
CCU 	  Catalytic cracking unit(s)
GEMS	  Continuous emission monitoring system(s)
CO  	  Carbon monoxide
COM3	  Continuous opacity monitoring system(s)
COS  	  Carbonyl sulfide
CPMS	  Continuous parameter monitoring system(s)
CRU	  Catalytic reforming unit(s)
CS2	  Carbon disulfide
D/F	  Dioxin and furan
E-Cat	  Equilibrium catalyst
EDV  	  Electrodynamic Venturi
EPA  	  Environmental Protection Agency
ESP	  Electrostatic precipitator(s)
FCCU                 Fluid catalytic cracking unit(s)
HAP	  Hazardous air pollutant(s)
HC1	  Hydrogen chloride
Hg	  Mercury
H2S	  Hydrogen sulfide
MACT  	  Maximum achievable control technology
                                    vin

-------
mg/yr                  Milligram(s) per year
MW  	  Megawatt
NAAQS  	  National Ambient Air Quality Standard(s)
NESHAP  	   National Emission Standard for Hazardous Air Pollutants
Ni	  Nickel
NOX	  Nitrogen oxide
NSPS  	  New source performance standard
O2	  Oxygen
OAQPS	  Office of Air Quality Planning and Standards
pH  	  percent hydrogen (i.e., acidity or alkalinity)
PM  	  Particulate matter
POM	  Polycyclic organic matter
ppm  	  Parts per million
ppmv  	  Parts per million by volume
ppmw	  Parts per million by weight
psig	  Pounds per square inch gauge
RFC  	  Reformulated gasoline
SO2	  Sulfur dioxide
SOX	  Sulfur oxide(s)
SRU  	  Sulfur recovery unit(s)
SSMP	  Startup, Shutdown, and Malfunction Plan
THC  	  Total hydrocarbons
TOG  	  Total organic carbon
tpy  	  Tons per year
TRS  	  Total reduced sulfur
VE	  Visible emissions
VOC	  Volatile organic compound(s)
                                     IX

-------
LIST OF COMMENTERS
Name
Chuck Ferrick
Robert Morris
Donald Parus
Prasad Rao
J. David Thorton
Edward J.
Campobenedetto
Manisha D. Blair
William O'Sullivan, P.E.
Ray Bishop
Lois N. Epstein, P.E.
LaNell S. Anderson
Eugene D. Becker
Joel F. Wilson
G.T. Theriot
Jerry E. Thompson
Thomas H. Jackson
Ann Farner Miller
Affiliation
American Petroleum Institute
National Petrochemical & Refiners
Association
Amoco Petroleum Products, Yorktown
Refinery
Consultant/Economist
Minnesota Pollution Control Agency
Institute of Clean Air Companies
Colorado Department of Public Health
and the Environment
New Jersey Department of
Environmental Protection
Oklahoma Department of Environmental
Quality
Environmental Defense Fund
Grandparents of East Harris County
Global Sulfur Systems, Inc.
Conoco, Inc.
Exxon Company, USA
Citgo Petroleum Corporation
Equiva Services LLC
Tosco Corporation
Docket No.
IV-F-3.1
IV-F-3.2
IV-F-3.3
IV-D-55
IV-D-24
IV-D-30
IV-D-32
IV-D-41
IV-G-1
IV-D-57
IV-D-25
IV-D-26
IV-D-27
IV-D-28
IV-D-29
IV-D-31
IV-D-38
IV-D-33
IV-D-35
IV-D-36
IV-D-37
IV-D-39
IV-D-40
IV-D-42

-------
Name
Gary C. Furlong
James A. Ross
Richard J. Glaser
Emily Green
William R. Beck
Allen R. Ellett
Norbert Dee, Ph.D
Phillip T. Cavanaugh
Steven K. Pavel
Karen Ritter
R. Hermanson
Neil Carman, PhD
Kent Veron, P.E.
Anne-Marie Ainsworth
S. Eric Vrolenburg
Affiliation
Sunoco, Inc.
Phillips 66 Company
Amoco Petroleum Products/Mandan
Refinery
Sierra Club Great Lakes Program
Mobil Corporation
BP Oil Company/Toledo Refinery
National Petrochemical & Refiners
Association
The Chevron Companies
Coastal Catalyst Technology, Inc.
American Petroleum Institute
Amoco Corporation
Sierra Club, Lone Star Chapter et al
Marathon Ashland Petroleum
Lyondell-Citgo Refining Company Ltd
Pennzoil-Quaker State Company
Docket No.
IV-D-43
IV-D-44
IV-D-45
IV-D-46
IV-D-47
IV-D-48
IV-D-34
IV-D-49
IV-D-51
IV-D-52
IV-D-53
IV-D-54
IV-D-56
IV-D-58
IV-D-59
IV-G-2
XI

-------
(This page is intentionally blank)

-------
1.0   CONTROL REQUIREMENTS FOR CCU CATALYST REGENERATOR VENTS

      1.1   Subcategorization of Thermal CCU

      Comment: Commenters IV-F-3.2, IV-D-42, IV-D-49, IV-D-54, and IV-D-59
believe there are process and emission differences between the thermal (i.e.,
nonfluidized-bed) CCU regenerator vent and the fluidized-bed CCU regenerator vent
that warrant Subcategorization. They describe differences in process operations (flow
pattern, operational severity, catalysts, regenerator methods) and control technology
(technology for fluidized bed units is not applicable and there is no comparable
proven control technology available).  According to the commenters, designing and
installing new, unproven control equipment would be difficult because the unit
operates under very low pressure and the costs would be much higher. EPA has
subcategorized CCU in all previous rulemakings, CCU monitoring requirements for
organic emissions based on full burn vs partial burn regeneration, and
semi-regenerative CRU based on operating processes and emission controls during
regeneration processes.  Subcategorization of CCU also would be consistent with
European rulemakings.

      Response:  After a more in-depth examination and evaluation of this particular
technology, we agree with the commenters that there are  substantial technical
differences in process operations of nonfluidized-bed CCU and fluidized-bed CCU
that warrant Subcategorization. These process differences include catalyst size and
composition, as well as reactor operating characteristics (e.g., plug downflow versus
fluidized riser processes, operating pressures). We also agree that the older nonfluid
process of thermal cracking is not amenable to conventional  particulate control
technology, in large part because of the potential for back pressure as a result of
control device operation. In addition, none of the three (3) existing thermal units
identified in our database are equipped with air emission controls and we have been
unable to identify any applicable air emission control technology that could be
transferred to these type of CCU. The MACT floor for a thermal CCU subcategory is
"no control," for both existing and new units, although it should be noted that it
appears highly unlikely that any new units of this type would ever come on-line; the
current trend is that these units are closing operations. Rather than subcategorize
and set a standard based  on no control for these units, the EPA has excluded non-fluid
CCU from the requirements of the final rule.

      1.2   Use of PM and CO as Surrogates for HAP

      Comment: Commenter IV-D-46 believes EPA violates section 112 of the CAA
by using PM and CO as surrogates for metal HAP and organic HAP. Section 112
requires emission standards for HAP listed for regulation, which does not include PM
or CO. Although EPA calculates an alternative limit assuming that Ni is a surrogate
for all other metal HAP, the agency selects the NSPS as the MACT floor without
characterizing the proportion of emitted PM that are hazardous metals, non-hazardous
metals, and unburned carbonaceous materials or the metals emission control
performance of the PM emission controls in light of varying metal feed rate

                                      1-1

-------
parameters.  EPA did not characterize the relationship of HAP organics to CO.  The
commenter points out that during the Industrial Combustion Coordinated
Rulemaking, EPA consultants explicitly state that CO alone could not be related to
emissions of chlorinated dibenzo-dioxins/furans from combustion sources.

      Response: It is important to note that the determination of MACT floors for
CCU organic HAP and metallic HAP were based on the control technologies used in
the industry, complete combustion of vent gases for control of organic HAP and an
ESP or Venturi scrubber for control of metallic HAP.  Surrogates were used in the
standards only to characterize the performance of these best performing technologies.
We have used surrogates for listed HAP in several rules because this simplifies
compliance demonstrations by allowing the use of well-known methods, i.e., methods
used to comply with the other CAA standards such as NSPS, and reduces costs
associated with constituent analyses. In this case, we believe there is a strong
justification for the use of surrogates. For the units regulated in this source category,
the HAP emissions are contained in the same emission stream as the criteria
pollutants subject to the NSPS and the control technologies and methods are the
same. As discussed in Chapter 4 of the BID for the proposed standards, the CCU
metal HAP emissions are primarily associated with the catalyst particles entrained in
the CCU catalyst regenerator flue gas and the particulate emission control devices
used to comply with the NSPS or other rules also provide metal HAP control. The
EPA has documented that HAP metals that exist in PM form are readily controlled by
PM control devices.  PM is a reasonable indicator of HAP metals relative to control
device performance. The final rule retains the optional use of PM as a surrogate for
HAP metals because the MACT floor equipment and level of control for HAP metals,
i.e., ESP and Venturi scrubbers, is identical to that for PM. Using PM as a surrogate
for specific HAP metals  eliminates the cost of performance testing to comply with
numerous standards for individual metals, and achieves exactly the same level of
HAP metal emissions limitation.  The same basic argument holds true for organic HAP
and use  of surrogates.

      The organic HAP emissions from CCU are contained in the same exhaust gas
stream from CCU catalyst regenerator as any CO, THC or VOC emissions and all of
these carbon-based emissions are a result of incomplete combustion in the catalyst
regeneration step.  The control methods for these pollutant types, including organic
HAP, are the same, i.e., "complete combustion." The EPA has determined that for
CCU, CO is a reasonable indicator of complete combustion. As for the comment that
during EPA's Industrial Combustion Coordinated Rulemaking, EPA consultants
explicitly state that CO alone could not be related to emissions of chlorinated
dibenzo-dioxins/furans from combustion sources, the EPA feels that is point is not
relevant to CCU because it is not expected that these units have D/F emissions. The
currently available source test data all  show D/F HAP levels that are below the test
method detection limits; there are no data that confirm the presence of D/F HAP in the
refinery CCU vent stream.
      As for the other type of units regulated in this source category, no surrogate is
used for HC1 from CRU.  COS and CS2 are the main HAP emitted from the final sulfur

                                     1-2

-------
plant vent and TRS is a reasonable indicator of these sulfur HAP. These reduced
sulfur compounds are contained in the same sulfur plant vent emission stream as the
SOX subject to the NSPS and the control methods are the same (thermal and catalytic
oxidizers).  TRS is an excellent surrogate for these sulfur HAP compounds since
reduced sulfur compounds consist of the two sulfur HAP plus H2S  (the nonHAP).
TRS (expressed as S02) is a reasonable indicator of reduced sulfur HAP.

      There are not sufficient data to establish a firm statistical relationship between
PM and metal HAP, CO and organic HAP, and SO2 and reduced sulfur HAP.
However, since the MACT floor determinations were made based on the technologies
in use at the units and the surrogates were used only to characterize the proper
performance of the units,  a precise statistical relationship between surrogate and HAP
is not necessary in this particular situation.

      1.3   NSPS as MACT Floor

      Comment:  Commenters IV-D-26, IV-D-31, IV-D-46, and FV-D-56 believe EPA
violates the CAA by using the NSPS as the MACT floor for existing  and new sources.
The EPA analysis does not articulate the average performance of the top 12 percent of
units (by PM/CO or metal/organic HAP) or the best performing source. EPA instead
sets the standard for new and existing sources as the NSPS rate on a not-to-be-
exceeded basis. For existing sources, EPA should establish specific metal HAP
emission rates based on the average of the best performing five metal emission rates
for each metal HAP and for the control for organic HAP. For new sources, EPA should
establish the standard based on the best performing unit.

      Commenters IV-D-26,  IV-D-31, and IV-D-56 state that the proposed HAP metal
emission limit adopted from the NSPS (1.0 Ib of PM per 1,000 Ibs of coke burn-off) is
less stringent than the allowable MACT floor and does not reflect the level of control
already met by more than 12 percent of existing sources using an ESP, Venturi
scrubber, or fabric filter. They point to EPA data for 26 units which reflect a median
emission rate of 0.81  and a mean of 0.86 Ib per 1,000 Ibs of coke burn-off. These
Commenters urge EPA to lower the proposed limit to 0.81 lb/1,000 Ib of coke, the
median PM emission rate  in the database.

      Response: In determining MACT, the EPA can consider alternative approaches
for establishing the MACT floor; these include:  (1) source test data  that characterize
actual emissions discharged by the sources,  (2) use of a technology and an
accompanying demonstrated achievable emission level that characterizes the
technology and accounts for process and air pollution control device variability, and
(3) information on Federal and State regulations and/or permit conditions that apply
to the source.

      If the information gathered indicates that more than 12 percent of the existing
units or sources are currently subject to the NSPS for that source category and no
more stringent rules apply, the NSPS thus represents the average emission limitation

                                     1-3

-------
achieved, in terms of a regulatory requirement, by the best performing 12 percent of
existing sources.

      The commenters' points were recently addressed by the B.C. Circuit in Sierra
Club v. EPA (March 2, 1999). That case holds that EPA can reasonably interpret the
MACT floor methodology so long as the Agency's methodology in a particular rule
allows it to "make a reasonable estimate of the performance of the top 12 percent of
units" and that evaluating how a given MACT technology performs is a permissible
means of estimating this performance. In developing emission limits, EPA cannot
establish a limit based on the median value only, without allowing for the
achievability of that limit in practice. The EPA believes that the NSPS levels selected
to characterize the MACT floor performance adequately account for the variability
inherent in the processes themselves and the air pollution control technologies, and
indicates what levels  are consistently achievable in practice.

      Based on the information and data available, the EPA concluded that the
MACT floor determination for existing CCU sources of metallic HAP (i.e., use of a PM
control device such as an ESP or Venturi scrubber) also represents the HAP emission
control that is achieved in practice by the best-controlled similar source in the source
category. Therefore, the MACT floor for new sources is the same as that for existing
sources of metallic HAP. No technology has been demonstrated in this industry to
provide a level of control more stringent than the MACT floor for metallic HAP.

      1.4   Extension of the Compliance Date

      Commenters IV-F-3.2, IV-D-35, IV-D-39, IV-D-43, IV-D-44, IV-D-47, IV-D-49,
IV-D-53, IV-D-54, IV-D-59, and IV-G-2 urge EPA to defer or delay promulgation of the
inorganic HAP standards for CCU catalyst regeneration vents to allow time to
coordinate the rule with development of the Tier 2 Motor Vehicle Emissions
Standards and Gasoline Sulfur Control Requirements other mlemakings such as RFC
Phase II,  and NAAQS PM standards. Achieving the sulfur reductions under Tier 2
will, in many cases, result in additional modifications to the unit and possibly
pretreatment of its feed to reduce the sulfur or nitrogen content, which is expected to
reduce metal HAP emissions if the reductions are severe enough. Other
recommendations include extending the compliance date until the compliance date
for Tier 2 fuels or 1 year afterwards, extending the compliance date to 6H years from
the compliance date for the final rule, and deferral for an unspecified period.  The
major concern voiced by the commenters is that plants will be required to install
expensive controls that may be extraneous as soon as they are installed depending on
the outcome of these and other rulemakings.  Because of the low emissions and risk,
the impact on the environment would be minimal, if any at all.

      Response: We agree that the Tier 2 rule will also affect the refinery production
processes and may affect the volume, characteristics, and environmental fate of
pollutants now emitted by refineries.
                                     1-4

-------
      To comply with the Tier 2 gasoline sulfur control requirements, individual
refineries ultimately will need to produce gasoline with an average sulfur content of
30 ppm. The majority of refineries will need to undertake major construction projects
to meet this limit.  Since these projects could require modification of CCU and other
affected sources, we revised the schedule to delay promulgation of this rule until
completion of the Tier 2, which was promulgated on February 10, 2000 (65 FR 6698).

      For some refineries, the Tier 2 rule significantly impacts its CCU.  These
refineries will have construction projects adding hydrotreating of the feed to the CCU.
For these refineries, we extended the compliance date to allow more time for
construction projects. We believe that this will encourage refinery owners and
operators to employ hydrotreating of the feedstock to comply with the Tier 2 rule.  As
discussed in more detail below, we believe that hydrotreating the feedstock has
increased environmental benefits relative to other methods of reducing gasoline
sulfur.

      The extended compliance date for existing CCU is based on when and how a
refinery produces low sulfur gasoline to meet the Tier 2 limit.  Hydrotreating the feed
to the CCU is one of the  means of producing low sulfur gasoline. As discussed further
below, hydrotreating the feedstock provides environmental benefits not realized with
other methods of producing low sulfur gasoline. It is also, unfortunately, significantly
more expensive than other methods of reducing the sulfur content of gasoline.

      A refinery owner  or operator must determine which technology to use in
reducing gasoline sulfur to meet the fuel standards. A number of alternatives are
available. Refineries may elect to hydro treat after the CCU, hydrotreat the CCU
feedstock, or implement some other form of desulfurization technology.
Hydrotreating the feedstock removes metals as well as sulfur. While hydrotreating the
feedstock to the unit would allow greater flexibility within the overall refinery
operations  and  would better position the refinery for any additional sulfur fuel
standards that might be promulgated in the future, such as standards to reduce sulfur
in diesel fuel (64 FR 26142, May 13, 1999), the cost of hydrotreating the CCU feed is
considerably more than post-unit hydrotreating for desulfurization. Thus, despite  the
greater flexibility realized through hydrotreating the feedstock, there is an economic
bias against its  use to reduce gasoline sulfur to meet the fuel  standards. We believe
that this bias could increase substantially if we do not coordinate the compliance
dates for this NESHAP and  the Tier 2 rule. A substantial increase in the economic
bias against hydrotreating the feedstock would likely result in less refineries
implementing this method of reducing gasoline sulfur, thereby foregoing a potentially
significant environmental benefit.

      Some facilities will take longer than 3 years to comply with the Tier 2
standards.  Should these facilities elect to install hydrotreatment units for the feed to
the  CCU, these  new units will not be operating at the compliance date for the MACT
standard, 3 years after promulgation. To avoid noncompliance, an owner or operator
would be required to install expensive PM controls  to comply with the MACT

                                      1-5

-------
standard.  These new controls might then become redundant with the later startup of
the hydrotreatment unit for the feed to the CCU. Therefore, if the owner or operator
elects to install a hydrotreatment unit for the feed to the CCU, the MACT compliance
date for the CCU becomes the same as the Tier 2 compliance date.

      Linking the compliance dates for the two rules, in this particular instance for
those refineries that elect to hydrotreat the CCU feedstock, will allow the refinery to
coordinate both decision making and the actual construction projects and, thus,
minimize  disruption to the refinery operations. We believe that not linking the
compliance dates for the two rules could result in an environmental benefit being
foregone and that linking them will result in a net environmental benefit because the
number of process unit shutdowns and startups would be minimized. Shutdowns
and startups can result in considerably more emissions to the atmosphere than
operations under normal conditions. An estimate of the emissions reductions that
would result from linking the compliance dates for the CCU standards and Tier 2 fuel
standards  is not possible at this time. This is because we lack information regarding
how the refineries will choose to comply with the fuel standards and the uncertainties
associated with startup and shutdown of these refinery operations.

      Linking the CCU compliance date to the Tier 2 fuel standards' compliance date
(i.e., the date the refinery produces low sulfur gasoline at 30 ppm) will not result in an
overall or  complete delay of the MACT standards for all CCU. While we believe that
linking the compliance dates will serve as an incentive to hydrotreat the CCU
feedstock, we nevertheless  expect that the majority of facilities will comply with the
fuel standards without implementing CCU feedstock hydrotreating. In some  cases,
even those that elect to hydrotreat the feedstock will comply in 5 years or less to take
advantage of the various pooling, averaging, banking, and trading options provided in
the final Tier 2 standards.  The remainder of refineries will begin production of low
sulfur gasoline over  the next  10-year period, although most are expected to be in full
compliance (i.e.,  producing gasoline at the 30 ppm annual average) by the year 2006.
In no case will refineries be allowed any later than December 31, 2009, to comply
with the standard for CCU, which corresponds to the final Tier 2 compliance date.

      1.5    Hg Emission Control

      Comment: Six commenters (IV-D-30, IV-D-25, IV-D-26, IV-D-31, IV-D-46, and
IV-D-56) urge EPA to reconsider its determination not to regulate Hg emissions,
conduct a  review of  available data and technology, and establish standards for Hg
emissions  from CCU regenerator vents.  According to Commenters IV-D-30, IV-D-31,
and IV-D-56, EPA is  incorrect in its determination that no technology that controls Hg
has been shown to be applicable to emissions from CCU regeneration vents. EDV
scrubbers  may control Hg when appropriate modules are attached. Commenter
IV-D-26 agrees that while PM controls will reduce a certain portion of the Hg,
emerging technologies are becoming available to control gaseous Hg emissions; EPA
should review available data, correlate emissions to coke-burn rate, exhaust rate,
throughput, the presence or absence of upstream pretreatment such as hydrotreating,

                                     1-6

-------
and to the Hg content of the crude oil.  Otherwise, EPA should defer the Hg portion of
the rulemaking until more research is conducted.

      Commenters IV-D-31 and IV-D-56 specifically ask EPA to evaluate MACT for
existing sources based on the lowest crude Hg content used by the top 12 percent of
existing refineries and MACT for new sources based on the lowest Hg content crude
oil available that still meets refinery performance characteristics.  This commenter
also notes that since the preamble  states that use of feedstocks with lower metal
content is a form of pollution prevention, this rationale also applies to Hg.

      Commenters IV-D-46 and IV-D-56 believe EPA erred under section 112 and
abused its discretion in determining that uncontrolled Hg emissions constitute the
MACT floor. These commenter argue:

      The MACT floor for existing sources should be the average of the best
      performing 5 sources or 2.75 x 10"2 Ibs of Hg per mm bbl. New source MACT
      should be equal to the best performing source in the Hg emissions database,
      1.00 x 10"3 Ibs of Hg per mm bbl. [Note: EPA database shows 0.443 and 0.07,
      respectively for these two numbers]

      EPA must consider technology transfer and pollution prevention control
      techniques as part of a beyond the floor determination. In support, the
      commenter cites language from the Senate Committee Report on the standard
      setting process, with emphasis on the "top-down" process.  According to the
      commenter, consideration of "top-down" strategies in setting a beyond the floor
      standard is highly appropriate in cases of persistent, bioaccumulative toxicants,
      and technology transfer is a component in top-down PSD decisions, as noted in
      the legislative history. EPA can not dismiss the possibility of using technology
      transfer to control Hg emissions because no technology is presently used on the
      vents. The proposal contains no analysis to show that the application of
      methods derived by technology transfer would be technically infeasible. Use of
      spray dryer/fabric filter/carbon absorption technology should be technically
      feasible to control Hg emissions. Environmental benefits from more effective
      control of other toxic heavy metals also would occur. Baghouse particulate
      emission controls also would contain emissions which ESP can not control
      during upsets. Sodium sulfide injection was not considered or were
      pre-combustion Hg removal processes for gas-oil feeds to the CCU.  EPA's
      failure to consider pollution prevention process and feedstock changes as they
      affect Hg emissions also violates the provisions of section 112(d)(2)(A), which
      states that EPA must consider in part, measures which reduce the volume of or
      eliminate emissions of such pollutants  through process changes, substitution of
      materials or other modifications.  Refiners using high Hg feed materials could
      switch to low Hg feedstocks or consider pre-CCU combustion and/or
      pre-distillation techniques for Hg removal.
                                      1-7

-------
      Response: The EPA has conducted a review of available data and technology
and has concluded that there are no demonstrated technologies that control volatile or
gaseous Hg emissions from CCU regenerator vents. A supplier of EDV scrubber
technology was contacted and source test and performance data documenting Hg
removal efficiencies were requested. No data were received to substantiate or
document the commenters claim of Hg removal.  There are a number of emerging
technologies that show promise in the control of these emissions but none have been
shown to be applicable to CCU vents. (See 1999 Critical Review - Mercury
Measurement and Its Control, T.D. Brown, et al, AWMAJune 23, 1999.) Therefore, the
MACT floor for Hg emissions for new and existing units is no control. The comment
that EPA should simply take the average of the best performing sources to determine
the MACT floor for both new and existing sources ignores the  fact that no control
technique could be identified for the units with the lowest Hg  emissions. Without
identification of a Hg control technique, a standard based on the best five performing
units for existing sources and the best single performing source in the Hg emissions
database would not be achievable in practice across the industry. In addition, the
values or limits reported by the  commenter are inconsistent with the available source
test data in the current refinery vent database.

      The EPA has also examined pollution prevention and source reduction control
techniques for Hg emissions control. EPA agrees that pre-treatment techniques such
as hydrotreating the CCU feed would have an impact on Hg and other HAP metal
emissions.  We collected  additional data after proposal to assess the environmental
impacts and performance of hydrotreatment, including Hg removal, but currently
available data are not adequate to characterize the precise Hg removal that could be
expected and what factors in the processes influence  these removal rates.

      1.6    Fabric Filters as MACT Floor or Beyond the Floor Technology

      Comment: Commenters IV-D-30, IV-D-31, and IV-D-56  do not agree with EPA's
preamble statement that "no additional control technology options were identified
that had been demonstrated to be more effective than the MACT floor technologies
that would achieve significant additional reductions in HAP emissions." They point
out that EPA only acknowledges wet scrubbers and ESP for controlling particulates
and metal HAPs when fabric filters are in wide use and far more effective for
particulate and metal HAP. The EPA must acknowledge this technology as a potential
control device and evaluate its applicability for metal HAP control on FCCU
regenerator vents.

      Response: We agree with the commenters that fabric filters are typically used
for the control of particulates and HAP metals. However, this  technology is not in use
in this industry in large part for technical reasons such as the high temperature of the
gas stream that impacts operation and operating costs, and the more frequent down
time for this type of control device again associated with maintenance.  In fact, we are
aware of only one plant that is reportedly using this control method.  At this plant,
the CCU vent control configuration includes heat recovery, a dry lime scrubber, and a

                                     1-8

-------
baghouse. The smaller CCU at the refinery is a residual oil CCU (referred to a the
ROSE unit) and was designed to process very poor quality crudes. The dry lime
scrubber was included in the process design to remove sulfur (i.e., SOx) from the CCU
catalyst regeneration vent flue gas. The baghouse was included in the process design
to remove the lime from the flue gas after the dry lime scrubber. That is, the baghouse
was not installed to control PM emissions from the CCU but rather lime dust
emissions from the lime scrubber.  The PM loading to the baghouse, for this control
scenario, is primarily lime dust entrained from the dry lime scrubber.  The baghouse
itself is a 12 compartment baghouse with 400+ bags per compartment. The facility
operator sights operational difficulties both with the dry lime scrubber (lime handling
area and disposal) as well as baghouse operation (leak detection and repair) in
addition to high operating and maintenance costs. Based on this information and
other factors EPA does not consider this control system to be the"best of the best"
performing technologies for reducing HAP emissions from the CCU regenerator vent.
These technical considerations would make the application of this air pollution
control technology to this industry somewhat uncertain. Performance in this type of
application is not demonstrated or documented. Wet scrubbers and ESP are the
dominant control methods for metal HAP (PM) emissions from FCCU and clearly
comprise the MACT floor technology.

      1.7   Ni Alternative Standard (Ibs/hr)

      Comment: Three commenters (IV-D-45, IV-D-53, and IV-D-54) believe EPA
should relax the proposed CCU metal HAP standard that is formatted! in terms of
Ib/hr. For small refineries that emit relatively small levels of Ni, they recommend a
limit in the range of 100 to 200 grams per hour.  Commenter IV-D-45 suggests this
limit should be averaged over a rolling 12 month averaging period, with compliance
determined by periodic E-Cat analyses, analyzed by third parties, and material
balance on the FCCU catalyst. Since catalyst is added in batch steps, weekly or
monthly measurements are not reliable and a longer averaging period is needed to
smooth the data to a reliable average. According to the commenters, the proposed
limit (13 grams per hour) is based on a select group of the best-performing FCCU with
low Ni feed that use feed hydrotreaters  and are subject to other stringent regulatory
requirements.  It cannot be concluded that the performance of this subset is
equivalent to the MACT floor technology.

      According to Commenters IV-D-53 and IV-D-54, EPA's variability analysis also
is flawed because EPA uses: (1) the z-statistic rather than the student's t-statistic,
which is appropriate for small samples from  populations, (2) the average relative
standard deviation instead of the more representative maximum relative standard
deviation, (3) data known to be false or problematic,  and (4) the 95% confidence level
rather than the 98% interval, which the commenter claims is an EPA precedent. This
commenter also believes the level of emissions excluded by the standard is trivial and
of little environmental significance. Raising the alternative standard would allow
some refineries to avoid installing controls that are not cost effective and provide real
                                      1-9

-------
de minimus relief.  The commenters support a standard of 0.38 tpy based on this
approach.

      Coimnenter IV-D-49 and associated industry commenters supports the
alternative standard but believes that the limit should be revised to include the
variability of Ni concentrations within the same crude.  They contend that this option
is not available to all facilities because a refinery can not easily change its crude oil or
the feed composition to the unit to one with low Ni to meet the standard. The
standard is set at the extreme low end of the range of Ni emissions from California
refineries based on the combination of low Ni crude oil and hydrotreating of
feedstocks. This commenter cites U.S. Department of Energy data showing that
outside California, 30% of the industry hydrotreat CCU feed.  The percentage of units
with the combination of hydrotreating and low Ni crudes is far less than 30% due to
the limited availability of low Ni crude.  For these reasons, EPA needs another
alternative that is not more restrictive on large units than small units.

      Commenters IV-D-30,  IV-D-31, and IV-D-56 oppose the proposed alternative
(0.029 Ib/hr of Ni). According to these commenters, this option does not control
particulate emissions,  the use of Ni as a surrogate poses difficulties in monitoring and
ensuring compliance, and it inadvertently allows medium and large units to escape
control. They point out that, while this option provides flexibility to small units,
units with a coke burn capacity less than 25,000 Ibs/hr  will not need to install
controls and even medium to large-sized units could comply with a third-stage
cyclone.  EPA appears to be exempting the very population of units that is currently
uncontrolled (small and medium-sized refineries that have been grandfathered or
otherwise have escaped regulation). Since most large units have at least a high
efficiency third-stage cyclone, they also could comply with this option.  If it is EPA's
intent to provide relief to small operators, the alternative should be limited  to units
with a coke burn capacity less than 15,000-20,000 Ibs/hr at facilities that are classified
as small entities. Or, EPA could promulgate a PM Ib/hr standard which would
control both metal HAP and particulates and make more sense for units with low Ni
content in the feed.  The commenter suggests that a standard of 20 Ibs/hr per 1,000 Ibs
of coke burn-off would ensure that only the smaller units (less than 20,000 Ibs of coke
burn per  hour) are provided flexibility. Another solution is to supplement the Ni
standard with a PM standard (e.g., 0.029 Ib/hr of Ni and 20 Ibs/hr of PM) since this
would focus only on small units, effectively control metal HAP and PM, and yet is
more lenient and less expensive than the PM NSPS.

      Response:  The data available to EPA indicates that 12 percent of the refinery
industry meets an emission limitation of 0.029 Ib/hr (0.125 tpy), which was used to
characterize performance of the MACT floor technology. The emission limitation in
this format was selected to provide additional flexibility to the own/operator and was
meant to  provide a means for the owner  or operator to demonstrate compliance
regardless of what combination of treatment and operational practices were used to
achieve the emissions limit. It is also  important to note that each compliance option
within a particular standard does not need to be available to every refinery; all

                                     1-10

-------
refiners can comply by installing the MACT technology.  The EPA rejects the 100 to
200 grams per hour limit suggested by some commenters. These high limits are
provided without any supporting documentation and appear to be selected on the
basis of emission values higher than 90% of the refinery industry.

      We acknowledge the quality assurance concerns regarding the results of certain
Ni emission measurements and the use of larger confidence intervals about the
average emission value in setting an emission limit that reflects use of the MACT floor
technology. However, we also believe that the analysis must use the  average of the
top 12 percent or the 6th percentile facility, rather than the emissions of the 12th
percentile facility. There are 124 FCCU in the U.S. and its territories; the 6th
percentile of the industry would be represented by the emissions reductions achieved
by the 7th and 8th ranked units. Reanalysis of the data, considering the reviewer's
comments on the statistical approach while using the 6th percentile unit, yields an
emission limit nearly identical to the proposed limit.

      In response to this comment, we examined the emission rates of the top
performing unit for which we have documented source test results. We found that the
average emission rates, as well as each individual test run result for the top 8 ranked
units, are all below 200 Ibs/yr. The 9th and 10th ranked units have similar average
emission rates, but a wider fluctuation in the individual test run results. From the test
data available, we determined that the proposed emission limit of 250 Ibs/yr
adequately characterizes the performance of the MACT floor technologies while
taking into account process variability.  For these reasons, we made no change in the
proposed Ni Ib/hr emission limit.

      The EPA considers the Ni emission limit alternative standard to be both useful
and necessary.  The purpose of the MACT program is to control HAP  emissions.
Although PM control is strongly related to metal HAP emissions, it is only one of the
two primary factors.  If the refinery operates with low metal HAP E-Cat concentrations
and relatively low PM emissions, additional PM emission controls may be an
ineffective  and unnecessary requirement to ensure low metal HAP emissions.
Consequently, no limitation of this second metal HAP emission limit is required or
justified in order to exact some additional level of PM control on facilities that
otherwise meet the metal HAP MACT floor (as measured using Ni as  the surrogate).
In short, no PM controls are required if a unit meets the Ni standard.

      1.8   Ni Alternative Standard (Ibs per 1,000 Ibs of coke burn-off)

      Comment: Fourteen commenters (IV-F-3.1, IV-F-3.2, IV-F-3.3,  IV-D-37,
IV-D-39, IV-D-40, IV-D-43, IV-D-44, IV-D-47, IV-D-48, IV-D-49, IV-D-53, IV-D-54, and
IV-D-59) urge EPA to include a rate-based Ni alternative of 0.006 Ni/1,000 Ibs of coke
burned in the final rule.  The commenters support the approach of using a direct
conversion factor based on FCCU fines data to convert the proposed PM emission
limit in Ibs PM /1,000 Ibs of coke burn to a Ni emission limit alternative. According to
the commenters, this format avoids penalizing large units with low HAP emissions

                                     1-11

-------
and the conversion factor approach better equates to the NSPS PM standard. The first
Ni alternative (a mass-based format in Ib/hr) is representative of the lowest emitting
sources, regardless of processing capacity and cannot be related to the rate-based
(lb/1,000 Ibs of coke burn-off) PM standard that recognizes the wide variety of
processing capacity within the industry. Most of the units that can comply with the
PM standard cannot comply with the mass-based standard due to their greater size.
Larger units should not be subject to a more restrictive Ni limit than smaller units due
to their  greater processing capacity.

      Commenter IV-D-49 explains that the relationship between the PM Ni content
and the  E-Cat Ni content is roughly 1 to 1, the key factor being the Ni E-Cat
concentration. In the proposal preamble, the EPA recommended using a value of
1,300 ppmw which is the average of Ni fines concentrations and results in an
emission rate of 0.0013 Ib Ni/1,000 Ib of coke which the commenters contend is not
achievable by all facilities equipped with MACT floor technology. They note that
some refineries can not easily change crude oil or feed composition to low Ni to meet
the standard at this level as the Ni content of crude oils varies widely, as do
emissions. The commenters argue that to ensure the standard is achievable by all
facilities with an ESP or scrubber, EPA must use the highest Ni content of any
operating FCCU, (6,833 ppm). Based on industry data, the Ni on the fines is about
0.87 times the Ni on the E-Cat, which yields the 0.006 value for the alternative Ni
standard in Ibs/ 1000 Ibs coke burn. Commenter IV-D-33 agrees with the need for an
alternative standard due to the high cost of control for a CCU regenerator with low Ni
emissions and supports the industry proposal for an alternative standard of 0.006 Ib of
Ni per 1,000 Ibs of coke burn calculated using the highest or worst-case Ni E-Cat value
for the conversion of PM to Ni.  The commenter believes this approach is technically
and environmentally defensible as the MACT floor standard.  Commenter IV-D-33
would support a slightly more stringent standard provided EPA accepts industry
comments on the methodology used to calculate the Ni mass standard. Commenter
IV-D-44 supports the industry-recommended alternative,  even though 6,833 ppmw
E-Cat Ni concentration used to establish this standard is not a maximum and may
limit operations. Commenters IV-D-47 and FV-D-48 supports the rate-based option
provided it reflects current operating conditions with the industry including normal
variability in the crudes that are processed.

      On the need for and usefulness  of a second nickel standard, Commenter
IV-D-48 believes the API-recommended standard determined using the worst-case
conversion factor is reasonable. Using the EPA recommendation of the average E-Cat
value, this commenter estimates 45 refineries would be required to install controls at
a cost of $48 million/yr to reduce HAP by 38 tons per year.  With an alternative Ni
standard at the level recommended by the industry (using the highest or worst-case
E-Cat), 6 refineries would install controls to reduce HAP by 19 tons per year at a cost
of $6.5 million/yr. Use of the worst-case value will still require large HAP sources to
control emissions and would be more cost effective.
                                     1-12

-------
      Commenters IV-D-30, IV-D-31, and IV-D-56 strongly disagree that EPA should
provide the second Ni alternative at the level suggested by industry (i.e., 0.006 Ibs per
1,000 Ibs of coke burn). They state that this alternative is not technically equivalent
to the MACT floor, is not protective of the environment as it is set at a level that
allows all refiners to process heavy feeds with no control device, and it is difficult to
enforce. Many refiners with nickel E-Cat levels an order of magnitude below 7,000
ppmw (the highest Ni on E-Cat level in the database) would not require any
particulate controls and most others could comply with only a third stage high
efficiency cyclone rather than the MACT floor technology. Based on EPA data and
analyses, the average nickel on E-Cat is less than 500 ppmw. Additionally, many
refiners process virgin gas oil or hydrotreated feed low in metals. The Ni on E-Cat in
these cases is less than 1,000 ppmw. The commenters contend that the actual Ni
emission rate will increase under the industry's recommended worst-case approach
and that a standard at the worst-case level could result in  increased emissions of
metal HAP and other pollutants (SOX, CO2, and PM) over time because refiners could
process heavier feeds with higher Ni level, given the overall industry trend towards
processing of dirtier crude.

      Commenters IV-D-30, IV-D-31, and IV-D-56 also point out that the alternative
format may present compliance and monitoring problems because refiners must
change the feed frequently depending on product demand, crude prices and operating
requirements and the Ni emission rate varies widely as a result. Under these
conditions the refiner can not directly measure the Ni emission rate. To do so
requires information on the coke burn rate, PM emission rate, and the instantaneous
metal content of the catalyst inventory or the Ni content of the feed (all of which vary
with time).  Refineries also may trade the catalyst inventory so it may not be relied
upon for computing the HAP metals emission rate. These commenters believe this
additional option is an invitation to non-compliance and that a unit that demonstrates
initial compliance may not be in compliance if it later processes a heavier feed.

      Response: After careful review of all the information and data collected
following proposal and received as part of the public comments, the EPA has decided
to include an additional metal HAP alternative for CCU formatted in terms of Ni
emissions per 1,000 Ibs of coke burn. It has been concluded that this particular
format (i.e., Ib Ni/ 1,000 Ibs coke burn) does account for the wide variation of
processing capacity within the industry and provides adequate means of determining
continuous compliance. However, it also has been concluded that the approach of
using E-Cat Ni concentration to make a direct conversion of the PM emission standard
to a Ni limit is not appropriate. The conversion factor approach does not lend itself to
determining a Ni emission value that is technically equivalent or relates back to use of
the MACT floor technology. The EPA has not been able to establish a rationale that
would support the determination or derivation of a second alternative Ni emission
limit value based on using the PM limit and the Ni concentration in the CCU catalyst
fines to convert the PM mass to an equivalent Ni mass. Use of the Ni fines (or E-Cat)
concentration as a conversion factor would not result in a emission limit that is
technically equivalent to the MACT floor since the E-Cat Ni concentration in no way

                                     1-13

-------
reflects the performance of the MACT floor technology. The Agency has no data or
information to relate E-Cat metals concentration to the best performing facilities (i.e.,
E-Cat metal concentration does not reflect or relate to control device performance). In
rejecting this conversion approach, the EPA is also rejecting the argument that a
worst-case conversion factor is necessary or appropriate.

      Nonetheless, the EPA feels that the alternative format in terms of Ibs Ni/1,000
Ibs coke burn has considerable merit even though the recommended approach to
determining or calculating the specific emission limit based on an E-Cat conversion
factor is not appropriate. This particular format allows for flexible compliance on the
part of the FCCU owner/operator. A compliance option in this format is suitable for
those large units that cannot, in some part because of their size, meet the Ib/hr limit.
An emission limit expressed in this format can be met by using front-end
hydrotreating, in-process operational changes, or end of pipe add-on controls alone or
in combination.  We have therefore used the refinery database to develop a Ni
emission limit in this particular format using an analysis based on the available
emissions test data normalized in terms of coke burn rate. Although the currently
available source test data is somewhat limited and is generally assumed to be
representative of the lowest Ni emitters across the industry, it does allow an analysis
to determine an appropriate emission limit to characterize the performance of the
MACT floor technology following EPA's  basic MACT floor determination criteria.
This analysis thus provides an emission limit in the alternative format (Ni emissions
per unit coke burn) that is reflective of the MACT floor technology.

      As there are 124 FCCU in the U.S. and its territories; the 6th percentile of the
industry would be represented by the emission reductions achieved by the 7th and 8th
ranked units.  Through review of the emission data, we found that the average
emission rates as well as each individual test run result for the top 8 ranked CCU are
all below 0.001 Ib Ni/1,000 Ibs coke burn. The average emission rate for the 9th ranked
facility is more than twice that of the 8th ranked facility. Therefore, only the top 8
units are considered in setting the emission limit. Based on our data analysis, we
determined that the emission limit of 0.001 Ib Ni/1,000 Ibs coke burn adequately
characterizes performance of the MACT  floor technology while taking into account
process and measurement variability (see Figure 1).  This analysis provides an
                                     1-14

-------
emission limit in the alternative format (Ni emissions per unit coke burn) that is
reflective of the MACT floor technology. This emission limit is included in the final
rule as an alternative to the PM or Ni Ib/hr limit.

      1.9    Use of the Arithmetic Mean, Median, Geometric Mean, 90th Percentile Value,
            or Highest Value as the Representative Concentration Used in the Factor for
            Conversion of PM to Ni

      Comment:  According to Commenters IV-D-30 and IV-D-31, the relationship
between PM emissions and Ni (or any other metal HAP) is best characterized by
material balance equations which should be used instead of the statistical measures
under consideration. Industry commenters generally contend that the highest value
must be used.  Commenter IV-D-48 contends that since all sources must be able to
achieve a standard, EPA must choose the highest value of Ni as the representative
concentrations used in the conversion from PM to Ni. The choice of the 90th
percentile, the arithmetic mean, or the geometric mean would automatically exclude
sources that have higher Ni concentrations from meeting any standard based on those
levels without controlling their PM emissions to a lower level than the control
technology of choice would allow.  This would force those refiners to choose the PM
standard instead of the Ni standard.

      Response:  As noted previously in comment/response 1.8, the EPA has decided
to include an additional metal HAP alternative for FCCU formatted in terms of Ibs of
Ni emissions per 1,000 Ibs of coke burn. It has been concluded that this particular
format (i.e., Ib Ni/ 1,000 Ibs coke burn) accounts for the wide variation of processing
capacity within the industry and provides adequate means of determining continuous
compliance. However, it also has been concluded that the approach of using E-Cat Ni
concentration data to make a direct conversion of the PM emission standard to a Ni
limit is not the appropriate methodology to establish the emission limit in this format.
The conversion factor approach is not technically equivalent nor does it relate to
performance of the MACT floor technology. No rationale has been established that
would support the determination or derivation of the second alternative Ni emission
limit value based on using the PM emission limit that reflects the performance of the
MACT control technology and the Ni concentration in the CCU catalyst fines to
convert the PM mass to an equivalent Ni mass. Use of the Ni fines (or E-Cat)
concentration at any value  (highest, lowest, mean, or 90th percentile) as a conversion
factor would not result in an emission limit that is technically equivalent to the
MACT floor since the E-Cat Ni concentration does not reflect the performance of the
MACT floor technology, as does PM emissions. The Ni E-Cat concentration of a FCCU
are dependent on a complex mixture of operating and economic considerations.
                                     1-15

-------
Figure 1. Available Ni Emission Test Data for FCCUs in Ibs Ni/1000 Ib coke burn

.DOE-U3
.5UE-03
_ 3.00E-D3
s
o
o
 2.50E-D3
2.00E-03
o
u
40

^ 1 .50E-03 "
z
.OOE-03
.DOE-04



" ,.;>*V
dul^t:R
uns
, ' , f" "'-,' '<~"J ' '"' J .'- 
Average Emissions ;






t
4




4
1
4
. . . I J
t 4
, J J
O.OOE+00 i i  i i i
12345
<
, i
i i
1
6 7 E
FCCU Rank
4
i

9
i

i
^

i
t
t
\
\



I



t
10 11 12

-------
The Agency has no data or information to relate E-Cat metals concentration to the best
performing facilities (i.e., E-Cat metal concentration does not reflect or relate to
control device performance).

       The EPA has concluded that use of the worst case Ni E-Cat in a direct
conversion of the PM limit to a Ni limit is in no way reflective of or linked to Ni
emission control or the MACT floor technology. The  EPA analysis indicates that an
alternative standard based on the highest Ni E-Cat concentration to convert PM to Ni
would require at most two facilities to add controls; the industry analysis shows the
number of facilities adding controls dropping from 45 to six facilities. In addition, a
standard at this high level could in the long term result in increased emissions of
metal HAP rather than their control given the general trend in the industry to process
lower quality crudes.

      We have therefore developed a Ni emission limit in the particular format (Ibs
Ni/1,000 coke burn) that characterizes the performance of the MACT floor technology.
The emission limit was determined based on the available emissions test data
normalized in terms of coke burn rate using an analysis following basic EPA criteria
for determination of the MACT floor. See comment/response 1.8 for more information
on the alternative emission limit formatted in Ib Ni per 1,000 Ibs of coke burn.

      1.10  Format of HAP Metal Standards

      Comment: Commenters IV-D-30, IV-D-25, IV-D-28, IV-D-31, IV-D-56, and
IV-D-57 urge EPA to change the format of the HAP metal standard for PM (Ibs of PM
per 1,000 Ibs of coke burn-off) adopted from the NSPS, the HAP metal standard for Ni
(lb/hr), and/or the HAP metal standard for Ni (Ibs of Ni per 1,000 Ibs of coke burn-off)
on which EPA requested comments.  All the commenters recommend that the
standards use an output-based format expressed in pounds of pollutant per 1,000
barrels of feed or throughput to replace or supplement the proposed standards.
Commenters IV-D-30 and IV-D-31 explain that the current format does not distinguish
between units processing light feeds and those processing heavier feeds. This is
important because units processing heavier feeds burn more coke per bbl of feed
processed, which generates more emissions. The standard should be expressed in
terms of emissions per bbl of feed based on the median quality of feed to the unit (i.e.,
the median pounds of  coke generated per bbl). This would treat all units (including
fluid and non-fluid units) processing feeds of different qualities equally and achieve
better control of metal HAP, PM, and other pollutants such as CO and greenhouse
gases.  This would treat FCCU and non-fluid CCU equally and allow refiners
additional flexibility in that they could use control devices, process better quality
feed, and/or use catalysts that reduce coke formation.  Commenter IV-D-25 points to
the NOX rule for new utility boilers, which ties the emission limit to the quantity of
electricity produced rather than the quantity of fuel burned as precedent. According
to the commenter, this revised format would allow facilities to address emissions from
a broader viewpoint, encourage redesign of the process to achieve the emissions limit,
and allow comparisons among different types of control equipment. Commenter

                                     1-17

-------
IV-D-28 does not believe the amount of coke burn-off, even if used in the NSPS, is the
proper surrogate for HAP metals because it does not correlate well with the metal
content of the feed. This commenter recommends a limit based on the HAP metal
content of the feed (such as Ni), amount (bbl/hr) of feed processed, limits on catalyst
fines exiting the regenerator, and particulate removal (over 99.9 percent).  The
maximum emission limit in Ib/hr of Ni for each unit can be derived from maximum
allowable E-Cat of Ni on emitted particulates and the maximum allowable Ib/hr of
particulate emitted.

      Response:  The EPA primarily requested comment on the format of the
additional HAP metal standard being considered by the Agency, and as previously
discussed, the EPA has decided to include an additional metal HAP alternative for the
CCU in the format suggested at proposal, i.e., Ni emissions per unit of coke burn.
With regard to the format recommended by the commenter, the EPA agrees that using
CCU feed rate to normalize the emission rates from the CCU has certain advantages,
as described by the commenters, for an additional metal HAP alternative emission
limit. However, there also are disadvantages to this format as a replacement to the
proposed PM and Ni emission limits and the new alternative format in terms of Ni per
unit of coke burn. We evaluated a Ni emission limit in the Ib/bbl format but
concluded that the performance of the MACT floor technology is tied closely to the
exhaust gas flow rate which is more directly a function of the coke burn rate than the
CCU oil feed rate. The intent of the Ni emission limits, both in terms of Ib/hr and Ibs
/1,000 Ibs coke burn, is to require MACT  control technology for CCU with high mass
Ni emissions without undue burden on small refineries or those refineries that have
low Ni emissions. The proposed formats and the new alternative Ni format are
considered to accomplish this as effectively as possible. As such, no further change in
the format of the proposed standards were made in the final rule.

      1.11  Level of Proposed CO Limit for HAP Organics

      Comment: Commenters IV-D-26, IV-D-28,  and Pvf-D-56 believe the 500 ppmv
CO limit is too high. Commenter IV-D-26 explains that although over 12 percent of
the existing CCU are required to comply with the NSPS, no data is presented to
indicate the actual performance level (according to the commenter). The commenter
contends that this is contrary to section 112,  which requires that MACT limits to be
based on operating data from the top 12 percent. Because sources typically operate
under their permitted limit, actual performance is probably under 500 ppmv. CO
levels under 100 ppmv are readily available which is evidenced by the provision
exempting them from monitoring if they demonstrate operation at less than 50 ppmv
for 30 days. The 500 ppmv also fails to account for the lower operating efficiency that
results  in higher operating costs.  Commenter IV-D-28 agrees and provides data in
support showing that actual CO emissions from four refineries are all 50 ppm or
under;  actual CO emissions from one refinery with a 500 ppmv limit is 10 ppm.
Organic HAP emissions even at the 50 ppmv level are significant as stack test results
for a 54,000 bbl/day CCU show cyanide compounds of 4.8 to 5.2 Ibs/hr and benzene
emissions up to 1.6 Ibs/hr. This commenter recommends a limit of about 100 ppm for

                                    1-18

-------
normal operation. These levels can be achieved by using a low concentration
oxidation catalyst in the regenerator in addition to complete combustion.

      Commenter IV-D-54 disagrees. While EPA correctly concludes that more than
12 percent of the CCU are subject to the CO limit in the NSPS, there are no data in the
record to demonstrate that these units can reliably and continuously meet the NSPS
limit. EPA should review the periodic reports required by 40  CFR 60.7(c) to better
evaluate this issue and consider averaging times longer than 1-hour for the final
standard.

      Response:  We do not agree with the commenters who assert that the MACT
floor must be based  solely on operating data. The CAA requires a minimum level or
"floor" for existing sources for categories or subcategories with 30 or more total
sources that can be less stringent than the standards for new sources but cannot be
less stringent than the average emission limitation achieved by the best-performing 12
percent of existing sources. After the floor has been determined, EPA must set
standards that are technically achievable and no less stringent than the floor that can
be met by all sources within the category or subcategory.

      As discussed  in the preamble to the proposed standard (see 63 FR 48899,
September 11, 1998), individual constituent data for organic HAP were not sufficient
to establish a MACT floor and  could not be considered representative of the entire
industry. For this reason, we reviewed emissions data on VOC, THC, and CO since
these data are indicative of individual organic HAP. The emission data for CCU
catalyst regeneration vents indicate that complete burn/combustion CCU and partial
burn/combustion CCU followed by a CO boiler or other combustion device achieve
similar organic HAP emission rates. As a result, both are considered types of
"complete combustion." The NSPS CO emission limit requires FCCU to have
complete combustion characterized and demonstrated by limiting the CO
concentration to less than 500  ppmv on a not-to-be-exceeded basis. This CO
concentration limit is determined by specific performance tests using EPA reference
methods. Well over 12 percent of the existing CCU are subject to this NSPS limit. We
feel that this limit represents the short-term average emission limitation achieved by
the best performing  12 percent of existing sources.

      We disagree that the limit needs to be set at a lower limit (e.g., 100 ppmv).
First, a requirement  for a lower CO limit presupposes that organic HAP emissions are
further reduced at lower CO levels. The available emissions data do not support that
supposition. Although the organic HAP emissions for CCU operating at or below 500
ppmv CO were significantly less than the organic HAP emissions for CCU operating
above 500 ppmv CO, no reduction in total organic HAP is seen from 500 to 100 ppmv,
based on the available data. As demonstrated by the data presented by the
commenters, HAP emissions occur at all levels of CO within this concentration range.
At the very low CO levels, it appears that aromatic HAP concentrations may be
reduced but other HAP (e.g., formaldehyde) concentrations increase. Consequently,
                                     1-19

-------
there appears to be little to no HAP emission reduction effected by requiring CO limits
of 50 or 100 ppmv rather than the 500 ppmv limit.

      Under the structure of the final rule, FCCU without an add-on control device, a
combustion device in this case, would be required to install and operate a GEMS to
monitor the CO emissions directly for compliance with the 1-hour standard of 500
ppmv.  The 500 ppmv 1-hr average value was set in the 1973  NSPS specifically to
accommodate complete-burn units. From the available CO source test data for non-
NSPS units, it has been concluded that existing non-NSPS units should be subject to
the same standard, i.e., they routinely achieve the 500 ppmv 1-hr average. The EPA
went to the States to obtain additional data on partial-burn units and the limited data
we received did not support a longer averaging time for these units. A longer
averaging time would allow significant periods of operation at CO levels above 500
ppmv, which would reduce the HAP emission control efficiency for the unit.

      The 500 ppmv limit is based on a short-term averaging time of 1-hour and is set
at this value to account for process fluctuations and source test variability. That is,
given a 500 ppmv limit evaluated on a 1-hour basis, refiners axe forced to operate at
much lower CO levels (e.g., 50 -100 ppmv) to comply with the standard during
process fluctuations. If the limit were based on a longer-term average, a lower limit
could be selected that would be much closer to actual operating levels over the long
term. However, commenters did not provide the historical data needed to analyze the
alternative CO or averaging time limits.  For these reasons, we did not revise the limit
or the averaging time.

      1.12   Format of Proposed CO Limit for HAP Organics

      Comment: Commenter IV-D-25 urges EPA to adopt an output-based format
using throughput to replace or supplement the proposed standard, pointing to the NOX
rule for new utility boilers as a precedent. This rule ties the emission limit to the
quantity of electricity produced rather than the quantity of fuel burned.  According to
the commenter, this revised format would allow facilities to address emissions from a
broader viewpoint, encourage redesign of the process to achieve the emissions limit,
and allow comparisons among different types of control equipment.

     Response: The EPA has determined that the concentration format (i.e.,
500 ppmv for CO) that was proposed for control of organic HAP provides a reasonable
and adequate characterization of the performance of complete combustion processes
applicable to CCU, the MACT floor technology. This format is also consistent with
the regulatory format of the NSPS for these units. Furthermore, the concentration
format lends itself to direct measurement using a GEMS.  The format recommended by
the commenter, using unit throughput to normalize the emission limit, does not
provide any advantages for this source category. It complicates compliance
monitoring and it does not correlate as directly to the performance of the control
technology. Therefore, we are confident that the proposed format for the organic HAP
limits is adequate to characterize performance of the unit and to achieve a long term

                                    1-20

-------
control of organic HAP emissions without restricting the operating flexibility of the
unit.

      1.13  Organic HAP Alternative to Proposed CO Standard

      Comment:  Commenter IV-D-54 believes EPA should include an organic HAP
standard as an alternative to the CO emission limit for CCU. This standard could be
in the range of 20  to SOppmw, measured by Method  18 or 25. This would be a direct
limit on organic HAP that bypasses the uncertainty of relying on CO as a surrogate.
While the low CO content may serve as an indicator of complete combustion, it is not
necessarily true that high CO content is a good indicator of high HAP emissions. This
would be consistent with the MACT standard. The commenter notes that the lack of
data precludes an opinion on whether the standard would be achievable or whether
their company might elect to comply with an organic HAP alternative.

      Response:  We agree that a direct limit would be preferable to a surrogate
approach. We considered the organic HAP standard  as well as THC as an alternative
surrogate but determined that we did not have adequate data to characterize all HAP
species present in the CCU catalyst regenerator vent emissions.  That is, few
refineries tested for the complete array of possible organic HAP emitted from the CCU
vent. As discussed earlier, we selected CO as a surrogate for organic HAP and the 500
ppmv limit of the  NSPS to characterize or represent  complete combustion, the MACT
floor technology.  The available data suggest that units operating below this CO level
have comparable levels of organic HAP emissions.

      1.14  SOX and NOX Emission Control

      Comment:  Commenters IV-D-30, IV-D-26, IV-D-31, and IV-D-56 urge EPA to
include the NSPS  standards for SOX control in the MACT standards for CCU catalyst
regenerator vents; two of the three commenters also advocate obtaining NOX
reductions. According to Commenters IV-D-30, IV-D-31, and IV-D-56, FCCU are one
of the largest emitters of SOX from refineries. An uncontrolled unit emits about 5,694
tpy of SOX for a medium-sized unit with a feed sulfur content of 1% that burns
50,000 Ibs/hr of coke. FCCU, particularly those operating in full combustion mode,
also are large NOX emitters. While the EPA SOX program has established control
requirements for SOX from utilities and mobile sources, emissions from petroleum
refineries remain unabated and may be increasing due to the use of more sour crude.
They point to the SOX health effects (morbidity as well as mortality) among sensitive
population groups and to the environmental and economic benefits and that EDV
scrubbers will control SOX, NOX, and volatile and semi-volatile metal HAP (e.g.,
cadmium, selenium, and mercury) that otherwise will  escape control.  The burden on
small refineries could be reduced by allowing use of SOX and NOx-reducing catalysts
or additives rather than the more costly add-on control equipment.

      Response: While we understand that SOX and NOX emissions are of concern,
EPA can not designate HAP standards under section 112 for the control of criteria or

                                     1-21

-------
ambient pollutants regulated elsewhere under the CAA, e.g., criteria pollutants are
subject to the NSPS requirements of section 111. Therefore, we did not include SOX
control requirements for FCCU.

      We did investigate the EDV scrubber to determine its performance for HAP
reduction. As previously discussed, no data were obtained to support the commenters
contention that EDV scrubbers are effective for volatile metal HAP control.

      1.15   Control of D/F Emissions

      Comment: Commenters IV-D-38 and IV-D-56 urge EPA to incorporate a MACT
standard for the control of D/F emissions from CCU regenerator vents. EPA is not
meeting the requirements of the CAA if these  emissions are not addressed in the final
rule.

      Response: First, the EPA was unable to confirm appreciable D/F emissions
from the CCU vent. With the support of EPA, GARB conducted a detailed emissions
source test at  one complete combustion FCCU with no other post-combustion device
in order to assess the potential of FCCU to emit dioxins, PCB  and a variety of other
HAP (both metallic and organic HAP).  In this test, 6-hour sampling runs were
employed to enhance the lower detection limits resulting from the analyses.
Nonetheless, only the octachloro isomers of dioxin and furan were detected in the
measurements.  These  are the least toxic of the D/F isomers; the mass emission rate of
dioxins in terms of 2,3,7,8-TCDD toxicity equivalents (TEQ) for the detected D/F
isomers ranged from 2  to 13 micrograms per year Gug/yr) for the three source test runs
performed.  Based on detection limits for those D/F isomers not detected during the
test, the 2,3,7,8-TCDD TEQ is calculated to be 3 to 5 mg/yr. None of the PCB isomers
were detected in any of the three stack samples collected for PCB analysis during the
source test.

      Similar test results were observed during a source test  of a CO boiler processing
FCCU exhaust gases. This CO boiler also received waste sludge from the wastewater
treatment facility at the refinery; as a result this boiler was tested as a hazardous
waste incinerator. The 2,3,7,8-TCDD TEQ emissions rate reported, largely based on
detection limits, ranged from 2 to 10 mg/yr. The test summary data provided was not
adequate to calculate the TEQ for detected isomers only, but the TEQ based on
detection limits was similar to those measured during the GARB source test.

      Based on the lack of significant detectable amounts of D/F and PCB measured
during these source tests, and the lack of any  other data to support that the FCCU
contains a source of chlorine, which is necessary to generate dioxins, the FCCU was
not considered to be a significant source of D/F emissions. Additionally, no
significant difference was seen between complete combustion units with no post-
combustion device and a partial combustion FCCU that employs a post-combustion
device. Again, the lack of a source of chlorine is thought to be the primary reason for
the low D/F formation in the CCU regenerator vent gas.

                                     1-22

-------
      Second, the proposed MACT standard, although not directly developed for D/F
emissions (because of the lack of verified D/F emissions), is considered by EPA to
provide substantive D/F emission control. Section 112 authorizes the development of
technology-based standards; MACT standards are based on the technology in use at
the best-controlled facilities.  D/F emission control may be effected by limiting
chlorine, providing complete combustion, and/or effective PM removal (much of the
D/F congeners condense onto PM). As chlorine is an unwanted contaminant in
gasoline and may adversely affect CCU catalyst performance, the refineries have a
built-in economic incentive to minimize the amount of chlorine that enters the CCU
process.  Beyond limiting chlorine, the MACT standard effectively establishes the
requirements that should minimize D/F emissions, if any D/F compounds are
generated. As none of the existing units regulated by the source category had specific
D/F or other control systems in-place that had a higher demonstrated D/F removal
efficiency than those required by the MACT standard, no HAP specific emission limits
were proposed for these particular organic HAP.

      Risk is not considered in determining the MACT technology. However, the
CAA recognizes the high toxicity of 2,3,7,8 TCDFs and 2,3,7,8 TCDD in section
112(c)(6). In the event that D/F emissions were found to be higher than the current
data indicate or if the currently projected levels of D/F emissions are found to have an
adverse impact on human health of the environment (based on subsequent risk
analyses), the residual risk posed by these HAP will be addressed in accordance with
section 112(f)(2) within 8 years following promulgation of these standards.

      1.16  Allowance or Exclusion of Emissions Created by  CO Control Device

      Comment:  Commenter IV-D-49 and associated industry  commenters ask EPA
to include an allowance in the CCU catalyst regenerator vent standard for emissions
created by the CO control device as is done in the NSPS. Combustion of the  CO in a
process heater or boiler will increase emissions of CO2, NOX, particulates, and possibly
SO2. The NSPS includes a variance for the additional particulates created by the
control device.  Without this, i.e., the allowance for secondary PM emissions, EPA
costs for the CO control system will be underestimated by not accounting for the
additional PM control. One commenter (IV-D-58) has a CCU catalyst regenerator
vented to a CO boiler. The CO boiler uses a fluidized bed with  limestone injection to
remove SO2. According to the commenter,  the addition of limestone does not affect
emissions of HAP metals so any nonHAP particulates resulting from the limestone
addition should be excluded from compliance determinations.

      Response: We have included an option in the final rule that allows the owner
or operator to elect to comply with the NSPS limit and monitoring requirements,
which includes the variance for additional PM generated by combustion operations
used as control devices. We also revised the test requirements of the rule to allow
measurements of PM ahead of the CO boiler or other process equipment rather than
only at the exhaust vent. We believe this should resolve the commenter's concern
                                     1-23

-------
about nonHAP particulates resulting from limestone addition as part of an air
pollution control system.

      1.17  Pollution Prevention Technologies for PM and HAP Metals from FCCU

      Comment: Commenter IV-D-52 asks EPA to address the ACT and DEMET
technologies as available and documented pollution prevention technologies that
reduce the source of metal HAP emissions from fluid CCU.  Technical papers are
provided as support.  According to the commenter, hydrotreating of the CCU feed was
identified as a source reduction technique; although, it is an energy intensive high
pressure hydrogen process which generates a hazardous waste and is physically
disconnected from the CCU (as are these two noted technologies}. The technologies
are available and have been incorporated in the FCCU circulating catalyst system at
one refinery to reduce PM and metal emissions.

      Response: The EPA recognizes the ACT and DEMET technologies as potential
technologies for reducing metal HAP content of the E-Cat and thereby reducing metal
HAP CCU emissions. Similar to hydrotreating, these processes may help a refinery to
meet the Ni emission limit alternatives (i.e.,  the Ni Ib/hour limit or the Ni lb/1,000 Ib
coke burn limit) but no requirement specific for their use is  included in the final rule.

      1.18  Compliance Date for Retrofit Particulate Controls

      Comment: Commenters IV-D-36, IV-D-43, IV-D-47, IV-D-49, IV-D-53, IV-D-54,
and IV-D-59 recommend that a facility be allowed to schedule its compliance date for
particulate standards for the CCU catalyst regeneration vent so that it coincides with
the next unit turnaround after promulgation. According to Commenter IV-D-49, most
facilities will request an extra year to schedule both the tie-in of controls and the
normal turnaround at the same time.  Although the extra year that can be requested
will reduce some of the problem, there will still be a large number of facilities that
will require an additional nonscheduled turnaround. It would not be desirable  to
force a significant part of the industry to shutdown because  of the potential impacts
on gasoline supply. Extra shutdowns, either to install equipment, new control tie-in
points, or new control systems exposes the environment to additional emissions from
startup and shutdowns and is very costly. Commenters IV-D-49 and IV-D-53 say EPA
has granted similar delays in other rules and should allow 3 years plus 150 days.
Commenter IV-D-43 suggests requiring controls for existing  sources within 3 years of
promulgation or the next scheduled major turnaround but no later than 6 years  after
promulgation. One commenter (IV-D-36) cites the typical turnout schedule of 3 to 4 +
years, concerns regarding the availability of control equipment and qualified
contractors, and similar extensions for other rules previously granted by EPA and
requests 6^2 years after promulgation. Commenter IV-D-54 recommends the first
turnaround following three years of the date of publication of the final rule or by
seven years from the date of publication.
                                    1-24

-------
      Response:  We have responded to these concerns in a number of ways.  First,
the new regulatory approach tied to Tier 2 , i.e., the switch of this rule to the 10-year
bin provides plants additional time and the extended compliance date for those CCU
that commit to use of hydro treating to comply with the 30 ppm Tier 2 fuel standard.
See comment/response 1.4 for additional discussion on the compliance date.

       We also revised the performance test requirements to expand the period of
time available for conducting the initial performance test. Under the final rule, you
may conduct the initial performance test anytime after the effective date of the rule
(the date of promulgation) rather than waiting until after the compliance date (i.e., 3
years after promulgation or 4 years if an extension is approved). As proposed, the
notification of compliance status (which must include a copy of the performance test
report and performance evaluation report if applicable) must be submitted no later
than 150 days after the applicable compliance date.

      1.19  Hydrotreatment of CCU Feed

      Comment:  Commenters IV-D-38 and IV-D-56 request that EPA evaluate
catalytic hydro treatment of CCU feed for inclusion in the MACT standards.  As
discussed in the BID for the proposed standards, this can reduce HAP as well as
increase yield and catalyst life. Because this technology is in place at more than 12
percent of existing refineries according to U.S. Energy Information Administration
data, it should be required as the MACT standard for all new and existing sources.
Commenter IV-D-49 and associated commenters oppose EPA's description as a
pollution prevention technique for the control of Ni emissions because hydrotreating
increases emissions of CO2, VOC, particulate, NOX, and any HAP emissions associated
with the combustion of carbon-based fuels. It also generates a RCRA-listed hazardous
waste and the Ni removed from the feedstock to the CCU feed is transferred to the
waste. This commenter also believes that under the definitions in the proposed rule,
a hydrotreater meets the  definition of a CCU.

      Response:  Since proposal of this rule, we have met with industry
representatives many times to discuss how they will change the refinery production
processes to meet the requirements of the Tier 2 and related distillates rule. It is clear
that a variety of methods may be used. For example, plants  that currently hydrotreat
or hydrocrack the feed to the CCU may increase the severity of treatment to reach the
30-ppm sulfur level required by  the Tier 2 rule.  Other plants may treat the distillate
stream after the CCU to reduce the sulfur content which will not impact emissions of
HAP metals. Either approach will increase the throughput of sulfur discharged to the
SRU.

      We have collected additional data and information since proposal to better
assess the environmental impacts and performance of hydrotreatment. The metal
HAP emission rate from the CCU catalyst regeneration vent can be attributed to two
general factors:  (1) the metal HAP content in the circulating CCU catalyst particles
(i.e., the E- Cat concentration); and (2) the mass rate at which these catalyst particles

                                     1-25

-------
are entrained in the flue gas and emitted to the atmosphere. As discussed previously,
PM emission controls have a direct bearing on the second factor, i.e., PM emission
rate, and therefore the metal HAP emission rate at a constant E-Cat concentration.
Hydrotreating has a much less direct affect on the metal HAP emission rate. As
discussed in the BID and as pointed out in the comments, hydrotreating both reduces
the metal HAP content of the CCU feed and it increases catalyst life. With the
increased catalyst life, the operator, in some situations, can lower the catalyst
replacement rate which tends to increase E-Cat metal HAP concentrations, and tends
to negate or off-set much of the effect of the lower metal HAP feed content's impact on
CCU metal HAP emissions. E-Cat Ni concentrations were examined for refineries that
hydrotreat their CCU feed.  The data varied over a considerable range and were
comparable to the Ni E-Cat concentrations for nonhydrotreating units. More
importantly, hydrotreating technology is not applicable or available to all refinery
configurations. As such, hydrotreating alone could not be considered a MACT floor
technology to reduce metal HAP emissions from the CCU catalyst regenerator vent.

      The EPA also  gathered additional technical information regarding
hydrotreating as a potential source reduction measure (or technology for removal of
metal HAP prior to the CCU). Under the Tier 2 emission standards for vehicles and
gasoline sulfur standards for refineries, refineries would be required to make cleaner
gasoline, i.e., with a lower sulfur content. Hydrotreating of CCU feedstock is one way
of reducing the sulfur content of gasoline and the refinery industry's decision on how
to meet the fuel standards also impacts their decision on how to meet the MACT
standards for CCU.

      Hydrotreating (and hydrorefining) is a process used primarily to remove sulfur
from various refinery process feedstocks.  The process involves mixing the
hydrocarbon stream with hydrogen in the presence of a catalyst at high pressures
(800 to 2,000 psig) and high temperature (700C). The level of sulfur removal is
generally 80  to 95% depending on the "severity" of the process. Severity is increased
with increased operating temperature and hydrogen partial pressure, and lower space
velocity (which increases contact time with the catalyst). With respect to catalytically
removed components, sulfur is the preferentially removed. After approximately 80 to
90% sulfur removal,  nitrogen removal begins, and after that, saturation of aromatic
hydrocarbons.

      1.20   Other Associated Emission Sources

      Comment: Commenters IV-D-30 and IV-D-56 identify catalyst particles
released during loading and unloading of the catalyst inventory as a potential
emission source to be evaluated.

      Response: We have no data on emissions from this activity and can not assess
it quantitatively. However, the EPA did examine these sources during the information
gathering stage of this rulemaking and it was judged that these sources were not
significant and as such did not warrant inclusion in the standards. We will gather

                                     1-26

-------
additional information following promulgation of this rule as part of the rule
development process for residual risk standards.
                                     1-27

-------
(This page intentionally blank)

-------
2.0   CONTROL REQUIREMENTS FOR CRU CATALYST REGENERATOR VENTS

      2.1   MACT floor for Semi-Regenerative CRU vs Exclusion of Control
            Requirements

      Comment: Commenter IV-D-26 opposes the MACT floor determination for
semi-regenerative units, which he believes is substantially less stringent than for the
cyclic or continuous type units. The proposed limit is erroneously based on the 92%
efficient single stage scrubber (the lowest HC1 control efficiency) rather thari 97%
efficient multi-stage scrubber used by the best-performing 3 percent of the
semi-regenerative units.  EPA should establish the floor based on the section 112
requirements as well as state-of-the-art technology.

      Commenter IV-D-59 states that controls for semi-regenerative units are not
warranted at all because of low emissions of HAPs (100 Ibs/yr per unit) and total
pollutants (under 1,000 Ibs/yr per unit) and high costs to control emissions under 5
psig. This plant uses an internal scrubber spray system of caustic injection prior to
the CRU heat exchanger bundles which they did not identify as control equipment in
earlier responses to our information collection requrest. According to the commenter,
this type of control is common practice for semi-regenerative units.  If EPA does not
exclude semi-regenerative units, EPA should determine the MACT floor with respect
to HC1 emissions as internal caustic spray injection.

      Response: As discussed in the preamble to the proposed rule, we
subcategorized semi-regenerative and continuous/cyclic CRU based on operational
differences in the rejuvenation process (i.e., primarily annual hours the system is
regenerating). However, the lack of HC1 emissions data available to characterize
emissions from semi-regenerative units led EPA to base their MACT floor
determination based on current industry control technology practices. Two classes of
scrubbers were designated to characterize the general types of scrubbers used to
control emissions from CRU catalyst regeneration vents during the coke burn-off step:
single stage and multiple stage scrubbers. The single stage system reflects the use of
internal scrubbing systems such as caustic spray injection, spray circulating solution,
hydroclyclones, and once-through spray scrubbers. Multiple stage scrubbers are
generally external to the process (i.e., an add-on control device)  and may include
packed tower, packed column, plate, spray, and Venturi systems.

      A summary of control system data show that 28 percent of continuous CRU use
multiple stage scrubbers compared to 6 percent using a single stage scrubber. For
cyclic CRU,  36 percent use multiple stage scrubbers compared to 11 percent using
single scrubbers. Seventy-two percent of the semi-regenerative units use a
single-stage scrubber while only 3 percent use a multiple stage system. Based on
these data, EPA determined that the MACT floor for continuous and cyclic CRU is the
multiple-stage scrubber while the floor for semi-regenerative  units is the single stage
scrubber.
                                      2-1

-------
      Due to the limited data available on the performance of HC1 scrubbers for CRU,
we characterized the performance of HC1 scrubbers based on industry surveys and
source test data on HC1 scrubbers used in the steel pickling industry. In the preamble
to the proposed standard, we discussed the similarities of the HC1 emission streams
and the basis for the 97% removal efficiency for multiple stage scrubbers and the 92%
efficiency for semi-regenerative scrubbers. We selected the 92% efficiency for semi-
regenerative processes based on the available data and engineering design
considerations of the various types of single stage scrubbers and the efficiency that
can be reasonably expected for all semi-regenerative CRU.

      The MACT floor for new and existing continuous and cyclic CRU are the same.
This is because the catalyst regeneration vent on the top-performing continuous and
cyclic CRU apply the same work practices or control devices as the top 12 percent of
existing continuous and cyclic CRU. In other words, the floor for existing units is
based on the top 12 percent of existing cyclic and continuous CRU and the floor for
new units is based on the top-performing units, which are the same in this case.

      The MACT floors for new and existing semi-regenerative CRU differ. The
MACT floor for existing semi-regenerative units is based on the top 12-percent, which
use single-stage scrubbers (including internal caustic spray systems). The
concentration limits are included in the rule to allow for the use of these systems.
Our data shows that two semi-regenerative CRU use multiple stage scrubbers to
control catalyst regeneration coke burn vents. As these are the best-controlled
semi-regenerative units, we determined that multiple-stage scrubbers constitute the
MACT floor for new semi-regenerative units.  Requiring high efficiency HC1 scrubbers
for existing semi-regenerative units  is clearly beyond the MACT floor and would
require replacement of existing controls at 72% of the units for nominal additional
HAP removal. Therefore, requiring "state-of-the-art" control devices  on this
intermittent vent at existing units was not included in the final rule.

      The EPA refinery information and data regarding HAP emissions from semi-
regenerative CRU indicates that these units can be significant sources of HAP
emissions during the various regeneration cycles and that air emissions controls and
operational practices of one type or another typically are used to reduce these
emissions.

      2.2   Three Percent O, Correction
      Comment:  Commenter IV-D-26 believes the 3% O2 correction in the proposed
limits for TOG and HC1 is inappropriate.  According to the commenter, a correction to
3% O2 is appropriate for boiler applications where O2 levels are typically less than 6%,
but are not appropriate for typical VOC oxidation systems that may operate with 18 to
20% O2 levels. While the correction would not affect the VOC destruction efficiency,
it would drastically affect the absolute default concentrations specified in the limits
(i.e., 20 ppmv on a dry basis for TOG, 10 or 30 ppmv on a dry basis for HC1).
                                      2-2

-------
      Response: The CRU operates under a reducing atmosphere. As such the
primary depressurization and purging cycles are expected to have very low if any O2
(due to safety considerations, i.e., explosions). In addition, the CRU catalyst
regeneration is a controlled burn process that operates with very low O2 concentration
(typically less than 3%). Therefore, the 3% O2 correction in the TOG and HC1 ppm
limits are not inappropriate because they are needed to account for inordinate
amounts of excess air.

      2.3    CRU Cutoff for Depressurization and Purging

      Comment: Commenters IV-F-3.1, IV-D-37, IV-D-47, IV-D-48, IV-D-49, IV-D-53,
IV-D-54, and IV-D-59 recommend changing the proposed cutoff level for control
requirements (differential pressure under 1 psig or reactor vent pressure of 1 psig or
less) to a reactor vent pressure of under 5 psig. The commenter state that 5 psig is the
level used in those States which have the facilities representing the MACT floor
(California, Texas, Louisiana). Also, it is unrealistic to measure a differential pressure
under 1 psig; 5 psig is the minimum pressure differential that could be determined
with any confidence as these vents typically start at pressures around 200 psig.
Commenter IV-D-40 says EPA should clarify that control requirements during initial
depressuring do not apply after the reactor has been initially depressured to 5 psig.

      Response: The EPA agrees with the commenters that since 5 psig is the limit in
those States with facilities representing the MACT floor, a limit of 5 psig is
appropriate.  We have revised the final rule to eliminate the differential pressure limit
and change the reactor vent pressure cutoff level from 1  psig to 5 psig.  The control
requirements apply to depressuring and purging operations until the reactor is
depressurized to this level.

      2.4    Clarification of TOC Standard

      Comment: Commenter IV-D-27 asks EPA to clarify 63.1562(b)(l)(ii)  of the
proposed rule. If a boiler or process  heater is used to control TOC, the vent stream
must be introduced into the flame zone  or any other location that will achieve the
required percent reduction or concentration. Does this mean the stream can be routed
to the fuel gas line? If so, this would make a stack test nearly impossible because the
fuel gas line feeds all boilers and beaters within a refinery. Also most refineries bring
in clean natural gas into the fuel gas line to dilute the H2S prior to the H2S monitor to
comply with the NSPS.  Diluting the gas stream and sending the excess gas to a flare
would make it easy to show less than 20 ppm TOC from a boiler.

      Response: Section 63.1560(c) of the proposed rule exempts streams routed to a
fuel gas system.  This exemption remains in the final rule.
                                      2-3

-------
      2.5   Achievability of HC1 Limits Based on Scrubbers in the Steel Pickling
            Industry

      Comment: Commenter IV-F-3.1 points out that the percent reduction standards
for HC1 emissions during coke burn-off and catalyst regeneration in CRU are based on
the performance of scrubbers in the steel pickling industry.  According to the
commenter, there are technical differences in scrubber characteristics, such as the
number of trays and packing depth and it is not clear that the proposed reductions are
being achieved by scrubbers for refinery CRU. The industry is in the process of
conducting tests to determine if the proposed standards represent the floor and will
submit the results.

      Response: Scrubber performance for HC1 emission control is well-documented.
While there are certainly some minor differences in scrubber and vent stream
characteristics, these differences have been taken into account in analysis the steel
pickling data relative to HC1 removal for CRU vent streams.  Pertinent summary data
are included in the docket and additional test reports supporting the performance
level are in the docket for the steel pickling rule. The commenter did not provide
additional data or information in support of his assertions; however, at subsequent
meetings industry representatives have suggested the sources tested by industry
achieved the required HC1 reduction.  No change was made  to the performance
requirements for HC1 scrubbers relative to the percent reduction standards for HC1
emissions during coke burn-off and catalyst regeneration in  CRU.

      2.6   Evaluate D/F Emissions

      Comment: Commenters IV-D-30, IV-D-29, IV-D-31, and IV-D-56 request that
EPA evaluate D/F emissions from CRU catalyst regenerator vents.  Commenter IV-D-31
ask EPA to include a MACT standard to prevent D/F emissions from CRU catalyst
regeneration vents.  Commenter IV-D-29 points out that the  emissions from CRU
include organic hydrocarbons, chlorinated compounds, and inorganic chlorides.  At
the temperature range of 482F to 842F, the inorganic chlorides from the coke
burn-off process can easily form D/F.  At the 20 ppmv outlet concentration, the
concentration of D/F may expose humans to a level higher than the EPA's dose limit of
0.006 picograms per kilogram per day or the maximum tolerable dose of 1 picogram
per kilogram per day proposed by the Agency for Toxic Substances and Disease
Registry. These limits still can be exceeded even if the 98% emission reduction
standard is  achieved. Since combustion of vent emissions at 1,400F provides direct
control of D/F, this requirement should be added to the standard with associated
monitoring and recordkeeping requirements.  Commenter IV-D-30 asks that EPA note
the presence of D/F in the CRU and CCU emission streams and the extent to which the
proposed standards control them.  Data collection in conjunction with GARB on D/F
emissions should be extended to cover emissions from CRU in their various cycles
(chloriding cycle, sulfiding step, and purge cycle).
                                     2-4

-------
      Response: The EPA acknowledges the presence of D/F emissions in the CRU
catalyst regeneration vent stream and the EPA's refinery database contains relevant
information and data on these emissions. We have also worked closely with Regional,
State, and local agencies to collect additional information on D/F emissions
throughout the course of this rule making; and those efforts are continuing. For
example, we collaborated  with GARB on source testing of CRU vent streams to test
directly for D/F during coke burn. Based on the temperature and compositional
characteristics of the purge cycle venting, D/F emissions during/from this cycle are not
expected.

      The EPA did not  establish a specific emission standard for D/F in this rule.
Although, it should be noted that the MACT floor technology determined for CRU
inorganic emissions, wet scrubbers, also is anticipated to effect some control of D/F
emissions.

      A preliminary risk assessment was performed for the CRU vent based on the
D/F emissions measured during the GARB source test. No significant risks were found
that justified implementing controls beyond the floor. However if, after collection of
additional information,  it is later determined that the D/F emissions pose an
unacceptable risk we will  regulate the CRU catalyst regeneration stream under the
risk standards.

      The presences (or absence ) of D/F emissions  from the CCU vent are discussed
in comment/response 1.15.

      2.7   Format of Proposed Standard for Organic HAP

      Comment: Commenter IV-D-25 urges EPA to adopt an output-based format
using throughput to replace or supplement the proposed standard, pointing to the NOX
rule for new utility boilers as a precedent. This rule ties the emission limit to the
quantity of electricity produced rather than the quantity of fuel burned. According to
the commenter, this  revised format would allow facilities to address emissions from a
broader viewpoint, encourage redesign of the process to achieve the emissions limit,
and allow comparisons  among different types of control equipment.

      Response: A throughput format is not feasible or meaningful for the CRU
because the equipment  standard required for control of organics is based on
conventional combustion technology (i.e., venting emissions to a flare or combustion
device). Since semi-regenerative and cyclic units must shutdown to regenerate and
the period of operation between regeneration cycles is dependent on a variety of
parameters not just unit throughput, a throughput format would not be appropriate in
this case.
                                     2-5

-------
      2.8   New Combustion Technologies

      Comment:  Commenter IV-D-25 suggests EPA change the wording in the
proposed rule from "a flare that meets the requirements for control devices in..." to "a
control device that meets the requirements in..."  It is possible that in certain
applications or as new technologies are introduced that other options will be available
to control emissions.

      Response: We agree with the commenter's suggestion and revised the proposed
rule to refer to "use of a control device that meets the requirements in 40 CFR 63.11."
If a facility uses a combustion technology other than flares that achieves 98%
destruction efficiency,  they can request approval of an alternative standard along with
performance test, monitoring, and recordkeeping/reporting requirements.

      2.9   Opacity Limit for Flares

      Comment:  Commenters IV-D-27, IV-D-31, and IV-D-56 urges EPA to include an
opacity limit for flares  that combust TOG streams as an indicator of good combustion.

      Response: We did not include an opacity limit for flares directly in the CRU
standards because the rule requires flares used to comply with the rule meet the
design and operating requirements in 40 CFR 63.11 of the NESHAP General
Provisions, which in turn requires flares to operate with no visible emissions.

      2.10  Requirements for Final Purge Vent

      Comment:  Commenter IV-D-40 asks EPA to clarify requirements applicable to
vent emissions from the final purge cycle. Guidance is needed as to when the
regeneration cycle has  moved from the catalyst rejuvenation phase to the final purge.
Many operators have different criteria for when they go back to the flare. The
commenter suggests that the final purge be defined as commencing when hydrogen is
readmitted to the reactor (which would denote when control equipment would again
apply to the vent).

      Response: The catalyst rejuvenation phase is characterized by pressurizing the
system with air (excess O2) to evenly redistribute the metal catalyst on the catalyst
particles. Subsequent  depressurization and system purges, typically a nitrogen purge
followed by a hydrogen purge, are considered purge cycles that require appropriate
control.

      2.11  Alternative Standard for Inorganic HAP Emissions

      Comment:  Commenter IV-D-54 believes some CRU can demonstrate by
material balance that only a few pounds of HC1 are emitted each day. These units
should not be subject to the same control requirements  as units with higher
emissions.  EPA should add an alterative standard which limits annual mass

                                     2-6

-------
emissions of HC1 using the same rationale as for the alternative Ni standard (Ibs/hr).
This commenter suggests a limit of one to two tpy of HC1.

      Response: Available data show scrubbers are a well-established control
technology for HC1 emissions from CRU catalyst regeneration vents.  Although we
agree that HC1 emissions from continuous CRU are generally lower in terms of mass
per unit time over the short term, the annual emissions from these units appear to be
as high or higher than cyclic or semi-regenerative units due to differences in operating
hours.  Consequently, a Ib/hr limit becomes problematic both to evaluate (i.e., data
are not available to establish a separate, mass cut-off applicable to the different types
of CRU regenerator units) and to enforce (i.e., inaccuracies in the mass balance
approach causes problems).  Furthermore, during site visits to 10 refineries, none of
the operators believed that they could perform an accurate inventory on either the
chloriding agent or the caustic consumption (for scrubbers) to assess HC1 emissions or
emission reductions using a material balance.

      As discussed in the preamble to the proposed rule, we subcategorized
semi-regenerative and continuous/cyclic CRU based on operational differences in the
regeneration process  (i.e., primarily annual hours the system is regenerating).
However, the lack of HC1  emissions data available to characterize emissions from
these units led EPA to base their MACT floor determination on current industry
control technology practices. Two classes of scrubbers were designated to
characterize the general types of scrubbers used to control emissions from CRU
catalyst regeneration vents during the coke burn-off step:  single state and multiple
stage scrubbers.  The performance of HC1 scrubber systems is best  characterized by
removal efficiency or a concentration limit when inlet concentrations are low.  The
proposed rule provided concentration limits as an alternative to removal efficiency
requirements. The concentration limits were established to provide operational
flexibility to owner and operators of affected CRU with low HC1 exhaust
concentrations (low HC1 emissions).  Based upon these considerations, EPA does not
consider  the suggested regulatory alternative to be either necessary or appropriate.
                                      2-7

-------
(This page intentionally blank)

-------
3.0   CONTROL REQUIREMENTS FOR SRU

      3.1   NSPS as MACT Floor

      Comment: Since substantive data were not available, EPA selected the NSPS
limit as the MACT floor for SRU. Commenter IV-D-26 believes EPA should instead
collect data to determine the actual emissions level of the top performing units to
establish limits based on proven performance reflecting current control technology.
Historically, units subject to the NSPS operate with emissions lower than the required
limits. Thus, the actual level of performance of the top performing units is probably
lower than the proposed rule.  Commenter IV-D-56 believes that selection of the NSPS
as the MACT floor does not comply with section 112(d)  or Congressional intent in
establishing MACT floors.  Specifically, as established the MACT floor: (1) must be at
least as stringent if not more stringent than BACT, (2) does not consider sulfur
pretreatment via catalytic hydrotreatment, (3) does not require backup units and/or
parallel trains, (4) does not require continuous monitoring systems required by the
NSPS, (5) relied on limited data from only one refinery to establish the floor, and (6)
will not reduce the elevated levels of SO2 resulting from upset conditions and
associated health threats to surrounding communities. According to the commenter,
the EPA must collect more data on COS and CS2 (and H2S) from refineries and make
the data publicly available for review.

      Response: As previously discussed in comment/responses 1.2 and 1.3, in
determining MACT, the EPA can consider alternative approaches for establishing the
MACT floor.  These  include:  (1) source test data that characterize actual emissions
discharged by the sources, (2) use of a technology and an accompanying demonstrated
achievable emission level that characterizes the technology and accounts for process
and air pollution control device variability, and (3) information on Federal and State
regulations and/or permit conditions that apply to  the source. If the information
gathered indicates that more than 12%  of the existing units or sources are currently
subject to the NSPS  for that source category and no more stringent rules apply, the
NSPS thus represents the average emission limitation achieved, in terms of a
regulatory requirement, by the best performing 12% of existing sources.

      We agree that most SRU will operate a lower emission level over the long term
in order to comply with the short term not to be exceeded standard of the NSPS. This
issue also had been discussed previously in comment/response 1.11. The selection of
an operating limit for unit emissions is  to a large extent dependent on the averaging
time used to characterize the emissions or unit operation. In general, emission limits
formatted in the short term, not to be exceeded values must take into account
variations in the process operations and the test methods used to measure emissions.

      With regard to the commenters recommendation to require back-up units
and/or parallel sulfur recovery unit trains to limit SO2 emissions resulting from upset
and unit shutdowns, EPA does not have the authority under section 112 to regulate
criteria pollutants such as SO2  It should also be pointed out that hydrotreatment

                                      3-1

-------
would not reduce the emissions from the SRU. Hydrotreatment reduces or removes
the sulfur from refinery streams up-stream of the SRU and the removed sulfur
(typically in the form of H2S) is sent to the SRU for sulfur recovery.

      With regard to the comment that EPA established the MACT floor based on
limited data from one refinery, this comment is baseless and blatantly false as the
establishment of the MACT floor was based on data collected from a nationwide
survey of SRU.

      3.2   Parallel Unit Configurations as MACT Floor or NSPS

      Comment: Commenters IV-D-31, IV-D-35, IV-D-46, and IV-D-56 urge EPA to
include requirements for multiple, parallel SRU to ensure that backup units to prevent
flaring of unscrubbed, raw refinery fuel gases should any units fail.  According to
commenter IV-D-31, many refineries use multiple parallel units to ensure continuous
removal of sulfur even when one of the SRU is not operating. EPA should determine
whether this configuration is in place in at least 12% of the existing refineries and, if
so, require this technology to ensure emission reductions that meet MACT
requirements. Commenter IV-D-35 recommends that EPA include this approach in
the NSPS to allow for full spare capacity of SRU, as currently practiced in California.
Rule 66 from the Los Angeles Air Pollution Control District requires full sparing and
backup of SRU.  Complete spare plants allow for full compliance even during
shutdowns and upset conditions.

      Commenter IV-D-56 states that more than 12% of the sour crude processing
refineries in Texas, the nation's largest oil refining State are required by permit to
operate SRU with backup units and/or parallel trains. The commenters maintain that
EPA must require as MACT for sulfur removal 100% backup SRU in parallel trains or
multiple  trains especially since SRU experience significant downtime and operating
problems. In support, the commenter describes serious upset and maintenance
problems at a number of local refineries, and for one refinery in particular, the high
number of NSPS violations resulting from an undersized and poorly designed SRU
and use of off-spec refinery fuel gas, the 2,000 tons or more of excess SO2 emissions
from malfunctions,  NAAQS exceedances of primary and secondary standard, and
repeated  citizen complaints.

      Response: As noted in the previous response, we do not have the statutory
authority to regulate criteria pollutants such as SO2 or to revise the NSPS under
section 111 to require backup units/parallel trains for improved control of SO2.
However, we do have the statutory authority to require plants to take the proper steps
needed to minimize HAP emissions resulting from startups, shutdowns, and
malfunctions. We will expect the SSMP submitted as part of the notification of
compliance status to address the situations that result from the conditions described
by the commenter. The rule requires combustion of sulfur plant gases to destroy the
sulfur HAP compounds (COS and CS2) to less than 300 ppmv. Therefore, the rule
requires HAP control when there are SRU upsets.  We also are requiring plants to

                                     3-2

-------
account for releases through by-pass lines. However, EPA cannot address the need for
a back-up SRU requirement under this MACT rulemaking; this would be an issue
covered in the State Implementation Plan relating to criteria pollutant emissions.

      3.3   Off-site Sulfur Plants

      Commenters IV-D-47, IV-D-48, and IV-D-53 ask EPA to exempt sulfur plants
are not part of a refinery but which receive gases to provide redundant emergency or
maintenance backup for a refinery sulfur plant.  If this change is not made, these
third-party plants are likely to terminate their agreements with adjacent refineries
instead of spending the money to install controls for a plant that receives flow a
couple of times a year during emergencies, maintenance, or upsets. This would result
in the refinery having to flare the gases and would increase emissions of sulfur.

      Response:  We do not have, and the commenter did not provide, specific data
to know the volume of sulfur containing gases (or the resulting emissions) sent to
off-site facilities for redundant emergency or maintenance backup.  The NSPS states
that a sulfur recovery plant need not be onsite to be subject to the rule. However,
under section 112, the definition of major source refers to any stationary source or
group of stationary sources located within a contiguous area and under common
control that emits a specified level of HAPs individually (10 tpy or more) or
collectively (25 tpy or more). We are not certain that the off-site facility clearly meets
the criteria for "contiguous area" and that the lease agreement denotes "under common
control." Based on the information presented, we do not believe that gases received by
a third party plant solely for redundant emergency or maintenance backup would
meet the HAP threshold levels for major source status. For these reasons, we
excluded off-site sulfur recovery plants receiving gases solely for redundant
emergency or maintenance backup as an affected source under the rule. These plants
may be subject to NSPS requirements, however.

      3.4   Hydrotreatment as MACT Floor

      Comment:  Commenter IV-D-56 states that EPA failed to consider sulfur
pretreatment before the SRU as an acceptable technology for reducing the sulfur feed
to the unit and controlling S02 emissions to below the 300 ppmv limit in the NSPS.
Several refineries (Texas and California) use catalytic hydrotreatment units or
hydrotreaters in addition to  SRU technology which can effectively reduce emissions
from the SRU/tail gas incinerator below the NSPS level.  Catalytic hydrotreatment
must be required at all new  and existing SRU units  since it is in place at more than
12% of all existing refineries in the US. EPA should conduct a review at all existing
refineries and add this to the final standard.  MACT needs to require 100% backup
systems whether it be 100 percent by backup SRU or a combination of backup
catalytic hydrotreatment and suitable-sized SRU.

      Response: The commenter did not provide any data to support the contention
that hydrotreatment reduces SO2 emissions from SRU. Based on the information at

                                     3-3

-------
hand, hydrotreatment would not reduce the emissions from the SRU but rather, it is
anticipated to increase the sulfur load to the sulfur recovery plant.  Hydrotreatment
reduces or removes the sulfur from refinery streams up-stream of the SRU and the
removed sulfur (typically in the form of H2S) is sent to the SRU for sulfur recovery.
The EPA is examining hydrotreatment as a part of the efforts to reduce emissions of
HAP metals and in coordination of the MACT standard with the Tier 2 fuel standards.
We are aware of the wide spread use of hydrotreatment in the refining industry but
are not requiring its use as a part of the HAP control scheme for refinery CCU and
SRU vents for a number of reasons which have been discussed previously (see
comment/response 1.19). We anticipate refineries to expand hydrotreatment capacity
in complying with the  Tier 2 fuel standards, but we also anticipate that this will
increase current SRU sulfur loads  and cause many refineries to expand  or build new
SRU.

      3.5  Thermal Oxidizers for  NSPS SRU

      Comment:  Commenter IV-D-28 believes the proposed MACT standard for TRS
emissions from SRU is appropriately more stringent than the NSPS where it requires a
thermal oxidizer following the tail gas unit, consistent with best industry practices.
This control technique should apply to all SRU. The commenter recommends that
the MACT standard be applied to NSPS units. He suggests a 3-5 year phase in period
to correspond with refinery turnaround periods. Commenter IV-D-46 interprets the
preamble (63 FR 48896) to claim that tail gas treatment at SRU is equivalent to a fume
incinerator in the level of control.  The commenters does not accept this statement
because COS and CS2 emissions will be higher from tail gas units that do not have a
fume incinerator in-line before discharge.

      Response: The MACT standard for SRU is based on the NSPS in 40 CFR Part 60
and should be considered equivalent in stringency. The concentration standard is
used to allow owner or operators a greater degree of flexibility that would be
permitted if an equipment standard (i.e., incinerator) was used. The concentration
standard reflects a high efficiency  SRU. Some tail gas treatment units get high
recoveries and thus there is no need for incineration of the off-gas.  Incineration
would be required for those units with a low recovery efficiency.

      3.6    Consistent Definition of TRS

      Comment: Commenter IV-D-25 believes EPA should use a definition of TRS
consistent with the NSPS for Kraft Paper  Mills (H2S, methyl mercaptan, dimethyl
sulfide, and dimethyl disulfide) and/or the NSPS for Petroleum Refineries (H2S, COS,
and CS2). Commenter IV-D-54 requests that EPA eliminate TRS as a surrogate for
sulfur HAP. The major component of TRS is likely to be H2S and most if not  all
variances in TRS will be the result of H2S variance. This should not be  a violation
since HAP emissions (COS and CS2) will not necessarily change. The EPA should
establish the limit using only COS and CS2.
                                     3-4

-------
      Response: We agree that it would be more effective to have a single definition
of TRS in all rules. However, in this particular situation, we have decided not to use
the cited NSPS definitions because they include compounds that are not listed HAP,
e.g., H2S, which is not a listed HAP. With regard to the commenter who states that the
TRS limit should be based solely on COS and CS2 because exceedance of the limit is
more likely the result of H2S increases, no data were provide to substantiate this claim
and this does not appear to be the case base on the limited information available to
the Agency.  In the NSPS, there is a 10 ppmv limit on H2 S emissions in addition to
the 300 ppmv TRS limit. Therefore, the 300 ppmv limit must pertain predominantly
to COS and CS2. We did account for the H2S increment in the NSPS limit in
establishing the MACT standard and the did not include H2S in the definition of
reduced sulfur HAP compounds.

      3.7   Format of Proposed Standard

      Comment: Commenter IV-D-25 urges EPA to adopt an output-based format for
the final TRS standards to replace or supplement the proposed standard, pointing to
the NOX rule for new utility boilers as a precedent.  This rule ties the emission limit to
the quantity of electricity produced rather than the quantity of fuel burned.
According to the commenter, this revised format would allow facilities to address
emissions from a broader viewpoint, encourage redesign of the process to achieve the
emissions limit, and allow comparisons among different types  of control equipment.

      Response: We did not change the reduced sulfur HAP standard to incorporate
an output based format because, in this case, the  revised format would require an
additional conversion step. We are not aware of any advantages that would result
from this additional complication; there are, however, some disadvantages. The
additional conversion step would create opportunities for potential errors and the
revised format would conflict with the NSPS.

      3.8   Incinerator for TRS Control

      Comment: Commenter IV-D-27 does not believe that incineration is adequate
for TRS control because complete oxidation may not occur.  She points to one refinery
in her State that frequently runs the SRU stack with less than 1% O2.

      Response: In response to the commenter's concerns, we added requirements
for a continuous O2 monitor to the TRS requirements.  These requirements are the
same as those in the NSPS for this source category.

      3.9   Calculation of TRS Limit

      Comment: Commenter IV-D-35 believes it is inaccurate to deduct the 10 ppm
from the  300 ppm TRS limit and assume the remainder is 290 ppm COS.  The
commenter explains that Beavon-Stretford sulfur plants operate without incineration
of the tail gas, thus eliminating the additional CO2 and CO emissions common with

                                     3-5

-------
amine scrubbing of the tail gas sulfur. The amount of COS in the hydrogenated tail
gas is determined by the equilibrium of the following reaction:  CO2 + H2S
< = = = = = > COS + H2O. The equilibrium concentration level of COS is normally
50-60 ppmv. If COS levels are over the chemical equilibrium level of about 60 ppm of
COS, it is due in part to low partial pressure of H2.  Normally, the COS is not
considered a HAP at this level and it seldom is close to 300 ppmv.

      Response: In response to the commenter's concerns, it is important to note that
as the NSPS defines TRS to include COS plus CS2 plus H2S concentrations.  The NSPS
H2S limit is 10 ppmv; therefore it is reasonable to conclude that the COS and CS2
portion of the 300 ppmv limit is between 290 and 300 ppmv depending on the H2S
concentration. Also it should be pointed out that the HAP designation is not
contingent upon the compound being present at a particular concentration; COS is a
HAP at any concentration.

      3.10  Sulfur Recovery Pits, Stretford Solution Tanks, and Non-Glaus SRU

      Comment: Commenters IV-D-38 and IV-D-56 urges EPA to include a MACT
standard for sulfur recovery pits, Stretford Solution Tanks, and nonClaus SRU in the
final rule.  EPA is not meeting the requirements of the CAA if the final rule does not
address these emissions which are documented in the BID for the proposed standards.

      Commenters IV-D-39, IV-D-47, IV-D-53, and IV-D-54 disagree. Commenter
IV-D-53 states that emissions from sulfur pits can occur at each SRU reactor when
elemental  sulfur is condensed and removed from the SRU gas and the liquid sulfur is
collected and stored in bins. Several refineries purge the pits to prevent the buildup
of gases; emissions are controlled by combining the purged gases with the SRU or tail
gas treatment off-gas and venting to an incinerator. Although data are not available to
accurately assess HAP emissions, the commenter believes H2S, COS, and CS2
emissions  are much lower than the TRS standard of 300 ppmv. The Commenters
believe the emissions should be considered de minimus.

      Certain types of tail gas treatment units such as Stretford units use a series of
open vessels as part of the solution circulation loop and a direct air contact cooling
tower to cool the solution. Emission data are not available to accurately assess HAP
emissions  from these units. Based on process considerations there are no significant
H2S, COS,  or CS2 emissions under normal operation according to Commenter IV-D-53.
Commenter rV-D-39 explains that his company does not vent vapors from Stretford
tanks or Stretford solution cooling towers because there is little opportunity for the
formation of COS or CS2 in the process which reduces all sulfur species to H2S.
Controls would be infeasible because the tanks are large and use a large quantity of air
in the oxidation step.  The air also bears the elemental sulfur formed at the top of the
tank.  If the tanks were covered and sufficient freeboard could be added, it is unclear
what technology might be used to control the low level of COS and CS2 in the stream.
Similar problems pertain to the enclosure of the direct-contact cooling tower.
                                     3-6

-------
Commenters IV-D-39, IV-D-53, and IV-D-54 believe emissions are expected to be
lower than the proposed standard for SRU and should be considered de minimus.

      Response: In the preamble to the proposed rule, we requested that commenters
submit specific data and information regarding the extent and type of emissions from
these sources.  Commenters submitted no data to characterize quantity of emissions
from these units. We can make no determination at this time without specific data.
We will collect additional information and data needed to assess these and other
potential sources in conjunction with the development of residual risks standards.

      3.11  NSPS Exemption for Small SRU

      Comment: Commenter IV-G-2 believes the final standard should include the
NSPS exemption for SRU with capacities less than 20 long tons of sulfur per day.
According to the commenter, State and local regulations include adequate control
requirements.  In support, the commenter cites the Agency's rationale for the
exclusion in 1978 as to lessen the economic impact on small refineries and encourage
installation of sulfur plants at small refineries. The cost for a small unit to meet the
sulfur standard is at least $2 million according to the commenter for a very small
incremental increase in control.

      Response: We did not include the exemption for small SRU because MACT
standards under section 112 are technology-based rather than cost-based.  The NSPS
cut-off established in 1978 was based on a cost-effectiveness analysis. Technology for
the control of emissions from these source is well-demonstrated and our economic
analysis does not indicate any economic impact on the industry as a result of
controlling these units.
                                     3-7

-------
(This page intentionally blank)

-------
4.0   CONTROL REQUIREMENTS FOR BY-PASS LINES

      4.1   Flow Meter Alternatives for CCU Regenerator By-Pass Lines

      Comment: Commenters IV-D-37, IV-D-40, IV-D-47, IV-D-48, and IV-D-53
recommend revising the proposed requirements for by-pass lines to allow other
alternatives. According to the commenters, the flow meter option does not work for
the type of by-pass lines used for most CCU regenerators. Flow monitors are difficult
to maintain and operate for a stack that can suddenly receive large flows of gas in the
event of a by-pass. For some refineries, car-sealing closed the by-pass valve is not an
option because the valve must be kept open in the event an emergency bypass is
required.

      The commenters suggest a water seal pot that has sufficient head pressure in
the drum to prevent flow through the by-pass stack with monitoring of the liquid level
of the water seal.  This provides a continuous measure of where gas is being diverted,
so there is no need for an hourly visual inspection. They also suggest electronic
bypass valve position monitoring (which provides continuous documentation of the
valve position such that hourly inspections would not be needed), a flow indicator or
a level indicator (or other alternative device that determines at least hourly whether a
vent stream is present in the by-pass line) or a blind flange.  A blind flange can assure
a no-flow condition as does the chain-locked or car-sealed valve. As with a closed
valve, a monthly visual inspection could be required. Suggested language for
63.1565(h)(l) of the rule is included in their comments to be more consistent with
the MACT I rule.

      Response: We agree that the requirements for a by-pass line can be more
flexible to accommodate different types of systems that provide the same information
as a flow indicator. The purpose of the monitoring is to ensure that each by-pass
event is recorded and reported to EPA.  This can be done just as well using electronic
by-pass valve position monitoring that provides a continuous record of the valve
position, or by a flow indicator or level indicator that determines on an hourly basis
whether or not vent stream flow is present.  If a continuous record of the valve
position for each hour is done by a recording system, hourly visual inspections would
not be needed. If a continuous recorder is not used, you would need to make visual
inspections every hour and record whether a vent stream is present. A blind flange
also can be used to assure a no-flow condition as suggested by the commenter as long
as it is bolted in place.  Plants that want to use a system other than a flow indicator,
level recorder, electronic valve position monitor must request the use of an alternative
control system and provide documentation supporting their equivalency to the rule
requirements.

      4.2    Installation Requirement for Flow Meter

      Comment:  Commenters IV-D-37, IV-D-47, and IV-D-53 request that EPA revise
the proposed requirements for by-pass lines by deleting the provision for installation

                                      4-1

-------
of a flow monitor at the entrance to any by-pass line that could divert the vent stream
from the control device to the atmosphere or revise the language to allow installation
at or as near as practical to the entrance to any by-pass line. The requirement for
installation at the entrance of the by-pass line needlessly restricts the location of the
device but also in effect requires that the device be a flow indicator.  Site specific
piping configurations may preclude installation at the entrance, but this does not
impair the ability to determine whether or not flow is present in the line.

      Response: The EPA agrees with the comment and has revised the final rule
accordingly. It was not our intention to needlessly restrict the type of monitoring
device used (if equivalent) or the location of the device as long as the owner or
operator maintains the ability to determine whether or not flow is present in the line.

      4.3    Continuous Monitoring Option for CCU By-Pass Valves

      Comment: Commenter IV-D-40 asks EPA to include in the final rule an option
for the use of continuous monitoring of both the CO boiler stack and the by-pass stack
to verify that the combined emissions do not exceed the organic  and inorganic HAP
standards for CCU. The commenter requests an alternative because the size of the
by-pass  line and location/design of the by-pass valve make it difficult to achieve a zero
percent  leakage rate even with redesign or replacement, resulting in a small flow
during normal operations when the valve is in a closed position.

      Response: To further understand the comments regarding compliance with the
by-pass  requirements, we discussed the issue with the commenter on a number of
occasions. The commenter's original concern related to the technical issue that the
CCU unit regenerator vent configuration would cause problems in demonstrating
compliance with the by-pass provisions of the rule. After discussions with the facility
staff regarding the by-pass requirements of the final rule and further analysis by the
facility,  it was concluded that the facility can comply with the requirements of the
final rule by use of conventional gas flow technologies that can be incorporated into
the current configuration.
                                      4-2

-------
5.0   MONITORING REQUIREMENTS

      5.1   CO Boiler Monitoring Requirements for Full-Burn FCCU

      Comment: Twelve industry commenters (IV-F-3.1, IV-F-3.2, IV-D-39, IV-D-43,
IV-D-44, IV-D-47, IV-D-48, IV-D-49, IV-D-51, IV-D-53, IV-D-54, IV-D-59) recommend
that the final standard  include consistent monitoring requirements for full-burn and
partial-burn CCU or eliminate them. Commenters IV-D-51 and IV-D-53 believe their
nonNSPS units will not be able to meet the 1-hour averaging time without high costs
and/or production capacity losses because the units are not specifically designed for
this level of performance. According to the commenters, CCU that use complete
combustion (full-burn) should have no monitoring or monitoring based on 24-hour
averages rather than 1-hour averages as required in the NSPS. In full-burn units, the
coke deposited on the catalyst is completely combusted within the regenerator at high
temperature (over 1,300F). In partial-burn units, coke is partially combusted within
the regenerator which operates at lower temperatures and the flue gas is burned in a
separate boiler commonly called a CO boiler.  The proposed monitoring requirements
exempt partial-burn CCU that vent emissions to a CO boiler above 44 MW based on
the robustness of these boilers to manage fluctuations in inlet flue gas CO content.
Daily averages apply to a CCU with emissions vented to combustion devices other
than furnaces or boilers.  However, a full-burn CCU must monitor using a 1-hour
averaging period. Because emissions from full-burn and partial-burn CCU are
comparable, full-burn units are as "robust" as  the partial burn units, and operators
need more time than 1-hour to make changes  if need, EPA should exempt full-burn
units from monitoring requirements like partial-burn units or allow the 24-hour
averaging. Commenter IV-D-54 states that EPA has provided no  data to support a
contention that there is a difference in organic HAP emissions from partial and
full-combustion units or any theoretical basis to believe that combustion is less robust
in control performance. Thus, there is no technical basis to require more frequent
monitoring.  Several of the commenters contend the cost of complying is very high
and has not been included in the cost analysis.

      Commenters IV-D-30 and IV-D-31 acknowledges  that 1-hour averaging period
for a 500  ppmv limit may create compliance problems for many units with complete
combustion. While changes in feed do induce some operational instability which
takes some time to correct, good operating practices, use of combustion promoters,
more combustion air, better distribution of combustion air, and minimizing feed
changes are options  that can be tried to control this problem. If the problem persists,
EPA could reformat the standard using an 8-hour average with a statistical approach
to accommodate the variability of CO emissions. Units  could be required to
simultaneous ensure that the  90th percentile CO concentration as measured by a
GEMS is under 500 ppmv and the 99th percentile is below 1,000 ppmv. Such a
standard would ensure control of organic pollutants while allowing for occasional
excursions due to changes in  operating conditions.
                                     5-1

-------
      Response: The EPA acknowledges the commenters concerns regarding
monitoring requirements for full-burn CCU. However, the commenters did not
provide adequate historical data to support their contention.  It also appears
contradictory that a "robust" system cannot meet the 1-hour limit. The CO boilers or
other furnaces and boilers employ auxiliary fuel and air to promote and maintain
combustion; differentiating between these processes is technically defensible.
Furthermore, based on available data from States and regions, we maintain the
position that CO emissions (i.e., the ability to maintain and control complete
combustion) is not a multiple hour control variable as suggested.

      Under the structure of the final rule, a FCCU without an add-on control device,
a combustion device in this case, would be required to install and operate a GEMS to
monitor the CO emissions directly for compliance with the 1-hour standard of 500
ppmv.  The 500 ppmv 1-hr average value was set in the 1973 NSPS specifically to
accommodate complete burn units. (See Background Information For Proposed New
Source Performance Standards, Volume 1, MAIN TEXT, APD-1352a, Technical Report
No. 7.) Without any data that shows otherwise, it has been concluded that existing
nonNSPS units  should be subject to the same standard, i.e., the 500 ppmv 1-hr
average rather than a higher standard that would result from incorporating a longer
averaging time.  The EPA went to the States to obtain data on partial burn units and
the limited data we received did not support a longer averaging time for these units.
The data previously on hand supported a 1-hour limit for these units. Other
commenters stated that the 500 limit was too high; which generally agrees with the
NSPS analysis that showed a value of about 25 ppmv as appropriate for partial burn
units with a CO boiler.

      Another factor that would have to be considered if the Agency were to consider
a change in the averaging time of the standard is that the NSPS limit of 500 ppm on a
1-hour basis takes into consideration fluctuations in CO concentration resulting from
process operation and test method variability,  as discussed previously in
comment/response 1.11.  Changing the averaging time would also require a change in
the numerical emission limit to correspond to  the increase averaging time. We do not
have the historical data to make this determination at this time.  The 500 ppmv limit
is based on a short-term averaging time of 1-hour and is set at this value high to
account for process fluctuations and source test variability.  That is, given a 500 ppmv
limit evaluated on a 1-hour basis, refiners are forced to  operate at much lower CO
levels (e.g., 50 -100 ppmv) to comply with the standard during process fluctuations.
If the limit were based on a longer-term average, a lower limit could be selected that
would be much closer to actual operating levels over the long term. However,
commenters did not provide the data needed to analyze the alternative CO or
averaging time limits for existing non-NSPS full-burn units. For these reasons, we did
not revise the limit or the averaging time.

      Regarding the difference in total organic HAP emissions from complete versus
partial combustion units, the EPA concluded that complete combustion units had
similar total organic HAP emissions as partial combustion units that are followed by a

                                     5-2

-------
CO boiler.  However, all of the organic HAP data used in that analysis were for full-
burn or complete combustion units with CO levels below 500 ppmv.

      5.2   Process Data and Secondary Measurement Devices As Alternatives to
            Flow Monitoring Requirements for Wet Scrubbers

      Comment: Commenters IV-D-37, IV-D-47, IV-D-49, IV-D-53, and IV-D-59
request alternative monitoring requirements for wet scrubbers.  Commenter IV-D-49
opposes monitoring flow to the unit where a wet scrubber is used to comply with the
standards for the CCU catalyst regenerator vent because of difficulties posed by the
heavy coke burn. Others oppose the proposed monitoring requirements for wet
scrubbers (pressure  drop across the scrubber, gas flow rate, and total water or
scrubbing liquid flow rate to the scrubber) because these direct measurement method
poses technical difficulties due to the erosive and corrosive nature of the hot gases
and liquids. Consequently, direct flow measurement devices are costly and
problematic to install and maintain. These devices also require unit downtime for
installation. According to the commenters, few facilities with wet scrubbers for CCU
directly measure gas flow or scrubber liquid flow rate.  Gas flow rate is a problem
because the ducting is very large and tends to be irregularly shaped with extreme
conditions of heat and particulates all of which affect instrument reliability. Instead,
the gas flow rate is determined based on process data such as the O2 and CO content
of the regenerator outlet flue gas. Liquid flow rate is determined based on such
parameters as the liquid pressure at the inlet to the scrubber and the wet gas scrubber
liquid purge flow rate. These methods for gas and liquid flow rates have been
accepted by Louisiana, Texas, and New Jersey and provide a very good correlation
with actual measured flow rate data. The commenters  recommend that the rule
clearly allow the use of other secondary measurement devices and calculation
procedures for this purpose. They suggest revising the rule to require a measurement
device equipped with a continuous recorder to  measure and record the average daily
pressure drop across the scrubber, the average daily gas flow rate to or through the
scrubber, and the average daily total water (or scrubbing liquid) flow rate to the
scrubber. As an alternative to direct measurement devices, these parameters can be
determined using process data and/or other secondary measurement devices.
Commenter IV-D-59 asks EPA to adopt either pressure drop across the scrubber or
scrubber liquid flow rate and gas flow rate as the monitoring requirements but not
both.

      Response: In the "top-down" approach used by EPA, we first look at methods
for the  direct measurement of emissions and if this is not technically  or economically
feasible, we look at direct measurement of operating parameters. The calculation
method suggested by the commenter is the least preferred  monitoring method because
it may not show when the control device is not operating at the  design level necessary
to continuous comply with the standards.  However, direct volumetric gas flow rate
monitoring is problematic for typical CCU because of the size and configuration of the
duct vent. In this case (i.e., for CCU gas flow rate), volumetric gas flow rate as
determined by process control monitors for air  blast rate (as used in the NSPS)

                                      5-3

-------
combined with gas stream temperature provide a more reliable and equally accurate
measure of the parameter. Therefore, this type of monitoring is allowed in the rule
under certain provisions.

      Without specific information and data on the secondary measurement devices
and calculation procedures, we can not assess the validity of the commenter's
assertions regarding direct measurement of liquid flow rates for wet scrubbers and
whether the recommended approach would provided consistent and equivalent
results when compared to direct measurement techniques.  We believe the
commenters will need to request approval of an alternative monitoring method from
the Administrator in this case rather than State and local permitting authorities as a
major monitoring change. A major change to monitoring is a modification to
federally-required monitoring that uses unproven technology or procedures or is an
entirely new method (sometimes necessary when the required monitoring is
unsuitable). Such a change maybe site-specific  or may apply to one or more source
categories and will usually set a national precedent. One example is a new
monitoring approach developed to apply to a control technology not contemplated in
the applicable regulation. Procedures for requesting an alternative monitoring method
are described in 63.8(f) of the NESHAP General Provisions.

      5.3    Monitoring Requirements for Other Wet Scrubber Designs

      Comment: Commenter IV-D-37 states that the proposed monitoring
requirements are not appropriate for jet ejector wet scrubbers. In these units, the
liquid (not the gas) is injected/atomized via nozzles, or jet ejectors, into the flue gas.
Consequently, the pressure drop of the flue gas is not an appropriate operating
parameter for monitoring. The pressure drop of the scrubbing liquid across the
nozzles is  relevant, but that is a function of the pump pressure or the liquid flow rate
which the proposed rule already requires to be monitored.  The commenter suggests
that the rule state that  wet scrubbers of the jet-ejector design for CCU are not required
to monitor pressure drop.

      Response: Pressure drop is an appropriate monitoring parameter for any
Venturi-type scrubber. Venturi wet scrubbers typically inject water ( or other
scrubbing  liquid) near  the Venturi throat, which atomizes, collides with the
particulate matter, and thus improves scrubbing performance. We agree with the
commenter that pressure drop of the flue gas is not an appropriate operating
parameter for monitoring jet ejector wet scrubbers that do not use a Venturi design for
PM removal. In response, we have revised the proposed rule to not require pressure
drop monitoring for nonVenturi, jet ejector type of wet scrubbers for CCU. However,
the owner or operator must still monitor the gas  flow rate to the scrubber and the total
scrubbing  liquid flow rate to the scrubber as indicators of proper operation and overall
performance.
                                     5-4

-------
      5.4   Continuous O2 Monitor for Full Burn CCU Regenerators

      Comment: Commenters IV-D-49 and IV-D-59 opposes the alternative
monitoring provisions for the use of a continuous O2 monitor for full burn CCU
regenerators. Like flow monitors, these are difficult to design and maintain due to the
concentration of particulates in the stream. The commenter recommends allowing the
placement of the monitor downstream of the particulate control device or the use of
alternate periodic monitoring once per day rather than on a continuous basis.

      Response:  The O2 monitor is needed to assess full burn operation of the unit
and have been in use by NSPS units for over 25 years. While we do not believe that
monitoring this parameter once per day would provide the information necessary to
ensure continuous compliance, we do not oppose alternative placement of the
monitor downstream, of the particulate control device provided no  air is introduced
that would corrupt measurements.

      5.5   Monitoring of Uncontrolled CCU

      Comment:  Commenters IV-F-3.1, IV-D-47, IV-D-48, IV-D-49, IV-D-53, and
IV-D-59 urge EPA to include specific provisions for the monitoring  of uncontrolled
CCU rather than requiring the permitting authority to establish requirements. They
recommend monitoring the coke burn rate  based on a compliance test or engineering
analysis due to the strong correlation of cyclone performance (and therefore PM
emissions) with coke burn (i.e., exhaust gas flow rate). The commenters suggest
monitoring the daily average coke burn rate for CCU that will meet  the PM limit
without an ESP or scrubber. For CCU that will meet the Ni standard without an ESP
or scrubber, the monthly average E-Cat Ni concentration should also be  a monitored
parameter. This concentration can be used along with a correlation between coke
burn rate and PM emissions to demonstrate compliance with the Ni standard.

      Commenter FV-D-54 opposes calculating and recording the coke burn rate if the
source is electing to  comply with the alternative Ni limit, where it is irrelevant.  He
agrees with the need to monitor and record the coke burn rate as an operating
parameter for a source subject to the PM standard, but the rate is only a  calculational
parameter and should not be enforceable in its own right in that there should not be a
violation if the rate is exceeded so long as the standard is met.

      Response: We agree with the commenters1 recommendations for  adding
specific monitoring requirements for uncontrolled CCU regeneration vents (i.e., units
without add-on control devices). Having specific requirements will facilitate the
permitting process for both the permit applicant and the permit authority. In
response to these comments we have revised the proposed rule to include specific
requirements.

      We also agree with the commenters  regarding the role of coke burn rate as a
calculated parameter rather than a monitored parameter where a control device is

                                     5-5

-------
present.  However, the performance of most PM control devices are affected by the
gas flow rate (which is directly proportional to the coke burn rate). For example, gas
velocity through an ESP is an important operational parameter that affects ESP
performance.  As a result, we have revised the proposed requirements for CCU
catalyst regeneration vents for those units that are not subject to (or electing to comply
with) the PM standard of 1 lb/1,000 Ibs of coke burn-off. These units can use gas flow
rate monitors or other parameters to determine flow rate (as discussed in
comment/response 5.2), rather than coke burn rate as their monitored parameter.

      We also agree with the commenters that a CCU catalyst regeneration vent that
is subject to the Ni standard (Ib/hr), must monitor the E-Cat Ni concentration as well
as control device operating parameters. The final rule requires you to measure the
E-Cat Ni concentration to determine a site-specific operating limit for the unit.
Following the performance test, you must determine and record the weekly E-Cat Ni
concentration. We did not select a monthly average E-Cat operating limit as suggested
by the commenters because this is too infrequent and does not correlate to coke
emission limit. Little or no cost is associated with weekly determinations because
most plants already determine the E-Cat on a more frequent basis as a process control
parameter. An equation is provide in the rule to calculate the site specific operating
limit value for the unit. An excursion from the established E-Cat range or established
process/control device operating parameter values would be reported as a deviation
from the operating standard.

      We do not agree with the commenters that coke burn rate should be the
monitored parameters for CCU catalyst regenerator vents not equipped with a control
device. For these situations, we have also added specific monitoring requirements.
This includes use of a COMS because there appears to be a stronger correlation
between particulate emissions and opacity than coke burn rate and PM emissions.
For uncontrolled units electing to comply with the Ni standard, gas flow rate and E-
Cat Ni concentration are also required monitoring parameters.

      If a COMSis used,  and you are subject to (or elect to comply with) the PM
standard of 1 lb/1,000 Ibs  of coke burn-off, you must measure the opacity of emissions
during the initial performance test and the coke burn rate. This is the same
requirement as for NSPS units. If you elect to comply with the Ni standard (Ib/hr),
you must measure the opacity of emissions  during the initial performance test and the
E-Cat Ni concentration and gas flow rate.  Using the opacity data from the test, you
must establish a site-specific operating standard based on the values at the time of the
source test. Any 1-hour period over the site-specific value determined for your unit
(for non-NSPS units) must be reported as a deviation from the operating standard.

      However with regard to overall monitoring requirements, the final rule has
been revised to reflect the fact that Federal requirements for State implementation
plans (40 part 51, appendix P) require COMS for all CCU with process capacity greater
than 20,000 bpd. Consequently, COMS were added to the final rule for all CCU
greater than 20,000 bpd.  However, as water vapor in wet scrubber exhausts interfere

                                     5-6

-------
with COMS, parameter monitoring is still the only monitoring method applicable to
CCU using wet scrubbers. Those units below the 20,000 bpd capacity cut-off have the
choice of COMS or parameter monitoring.

      5.6   Repeat Performance Tests for CCU Catalyst Regenerators Subject to Ni
            Alternative

      Comment:  Commenters IV-D-30 and IV-D-31 request that EPA reinstate the
requirement for biennial performance tests included in the P-MACT document
particularly for any unit subject to a Ni standard. Compliance with the Ni alterative
can only be verified after confirming the particulate emission rate, the coke burn rate,
the E-Cat, and the feed quality.  The commenters also recommends that EPA require
monitoring of the  CCU feed rate on a regular basis and report the quality of feed to the
unit on a monthly or quarterly basis.

      Response: In the final rule, we are not requiring repeat performance test for
those units that elect to comply with the CCU standard formatted in terms of Ni
emissions. These  units must perform and initial performance test to determine their
Ni emissions and  must determine their Ni E-Cat concentration during the test.  For
units greater than  20,000 bpd capacity, this is used as input to determining a
maximum operating value (based on opacity, gas flow rate, and E-Cat Ni
concentration) that is not to be exceeded during operation of the unit. Although CCU
feed rate and feed  quality are important parameters that influence  the Ni E-Cat
concentration, it is also necessary to know the catalyst replacement rate to calculate a
value for E-Cat or  Ni emissions. Direct weekly E-Cat Ni concentration measurements
are a more direct and timely assessment of the E-Cat Ni concentration and adding
these additional parameters is unnecessary and burdensome. For those units that are
less than 20,000 bpd and are using an ESP or wet scrubber as a control device,
continuous compliance also will be based on monitoring of control device operating
values. For those  units that are less than 20,000 bpd and that do not have these
control devices, continuous compliance will be determined based  on maintaining a
site-specific opacity as measured by a COMS.  Given these requirements, we do no
feel that it is necessary to require repeat performance test to ensure continuous
compliance. For those situations where there is any uncertainty on the part of the
Agency regarding  compliance, the EPA has the authority under section 114 of the
CAA to request a source test to determine compliance.

      5.7   Calibration of Temperature Measurement Device for a Boiler or
            Process Heater Less than 44 MW Where the Vent Stream is Not
            Introduced into the Flame Zone

      Comment:  Commenter IV-D-26 refers to the requirements of 63.1565(a)(3)(ii)
of the proposed rule which requires the plant to verify the calibration of the
temperature measurement device every 3 months. According to the commenter,
calibration of the device will not ensure accuracy; calibration should be required of
                                     5-7

-------
the entire temperature recording system. The commenter suggests that a 6-month
frequency would be more reasonable for this requirement

      Response: We agree with the commenter's suggestion and revised the proposed
rule to require calibration of the temperature recording system (device and recorder)
every 6 months.

      5.8   Monitoring Exemption for a Boiler or Process Heater Greater than 44
            MW Heat Input

      Comment: Commenter IV-D-53 questioned why the proposed rule allows CCU
with a CO (waste heat) boiler that has a design heat input capacity of at least 44 MW
to be exempt from the requirement to install and operate a GEMS or CPMS.

      Response: As proposed, the rule does allow CCU with a CO  (waste heat) boiler
that has a design heat input capacity of at least 44 MW to be exempt from the
requirement to install and operate a GEMS or CPMS. The EPA has re-examined
whether these CO boilers operate in a manner similar to typical industrial boilers,
boilers that served as the basis for the exemption, and whether they achieve
equivalent organic compound destruction efficiencies.

      A boiler (or process heater furnace) is comparable to a vapor incinerator (a.k.a.
thermal oxidizer or afterburner) where the average furnace temperature and residence
time, for the most part, determine the combustion efficiency. Thermal oxidizers
generally operate in the range of 1,200 to 1,600 F and require a minimum residence
time of 0.3 seconds in the oxidizing zone. An analysis of test results, along with
kinetics calculations, indicate that for a nonhalogenated VOC, a 98% destruction
efficiency is achieved by vapor incinerators with a combustion temperature of 1,600
F and a residence time of 0.75 seconds.

      A review of the technical basis for the original boiler-size exemption shows that
a mathematical model was developed to estimate the furnace residence time and
temperature profiles for a variety of industrial boilers. This model predicts mean
furnace residence times of from 0.25 to 0.83 seconds for natural gas-fires watertube
boilers in the size range from 4.4 to 44 MW (15tol50xl06 Btu/hr). In industrial
boilers at or above the 44 MW size, residence times and operating temperatures
ensure a 98% VOC destruction efficiency. Furnace temperatures for this range of
industrial boiler sizes are at or above 2,200 F, with peak furnace temperatures
occurring in excess of 2,810 F.

      In the petroleum industry, the operation of partial-burn FCCU produces gases
rich in CO.  To reclaim the thermal energy represented by these gases (i.e., the heat of
combustion of CO and other organics , and the sensible heat of the regeneration
gases), the unit can be designed to include a CO boiler that uses the CO as fuel to
generate steam.  From the air pollution viewpoint, the CO boiler oxidizes the CO and
other combustibles to carbon dioxide and water.

                                     5-8

-------
      In most cases, auxiliary or supplemental fuel is required in addition to the CO
and may be either fuel oil, refinery process gas, or natural gas.  The CO boiler may be
a vertical structure with either a rectangular or circular cross-section with water-
cooled walls. The following design criteria have been established:

1.    The basic firing rate should produce a temperature of 1,800 F in the furnace, to
      provide safe and stable combustion of the fuels.

2.    Air is supplied by the forced draft fan to provide 2% O2 leaving the unit when
      burning gases and supplemental fuel.

3.    Supplemental firing equipment is provided capable of raising the temperature
      of the FCCU gases to 1,450 F which is the temperature needed for ignition of
      the gases.

      The literature also reports that since there are only slight variations in the
operation of the FCCU, the boiler is normally base loaded. It handles all the gases
from the regenerator regardless of the CO2 /CO ratio. A change in this ratio merely
affects the quantity of supplementary fuel necessary to maintain the required furnace
temperature of 1,800 F. This temperature provides a reasonable operating margin for
possible variation in the operation of the regenerator or the boiler.

      Regeneration gases from the FCCU are normally delivered to the inlet of the CO
boiler ductwork at about 1,100 F, and 2 psig.  When the regenerator gases first pass
through an ESP, the inlet gas to the precipitator must be cooled below 500 F.  The CO
boiler would then receive regeneration flue gas between 450 and 500 F. The CO
boiler's  firebox temperature is also reported in this reference to be between 1,800 and
2,000 F. The firebox temperatures for CO boilers do not appear to be generally as
high as  those reported for conventional natural gas fired industrial boilers; however,
the CO boiler operating temperatures are above the typical operating range for most
high efficiency thermal oxidizers.

      Based on the  design and operating characteristics of refinery CO boilers for
FCCU, it appears that this "control unit operates  the same way as an ordinary process
unit boiler" and that CO boilers would operate with residence times and operating
temperatures that ensure a 98 percent VOC destruction efficiency similar to
conventional industrial boilers. Therefore, allowing these units the standard size-
based exemption for monitoring is considered  appropriate.

      5.9   Accuracy and Calibration Requirements for CCU and CRU with Wet
            Scrubbers

      Comment:  Commenters IV-D-49, FV-D-53, and IV-D-59 oppose the 3-month
calibration requirement for cyclic and semi-regenerative CRU that may not regenerate
quarterly. Also, some orifice plates would have to be removed from service to be
calibrated, thus requiring a shutdown of the unit. Commenter IV-D-53 suggests

                                      5-9

-------
calibration every 3 months for cyclic and continuous CRU and prior to regeneration
for semi-regenerative CRU (or quarterly, whichever is less stringent). Commenter
IV-D-49 suggests calibration at the same time as the CRU regeneration for cyclic and
continuous CRU and prior to regeneration for semi-regenerative CRU.

      Commenters IV-D-47 and IV-D-53 explain that existing monitoring equipment
for CRU with wet scrubbers will not meet the proposed calibration and accuracy
requirements. CRU pressure drop monitors are typically in excess of 200 psig and the
pressure monitors and ranged accordingly.  Five psig is the minimum pressure drop
that can be determined with any confidence. Flow rate monitors are typically orifice
plates which will not demonstrate accuracy at the 5 percent level. These devices
typically are assessed to be accurate to within 10 percent.  More accurate flow
monitoring devices (e.g., insertable turbine meter) are difficult to install and maintain
and unlikely to withstand the corrosive characteristic of the  gas.  He recommends that
flow monitors using orifice plates may be calibrated physically and assumed to be
accurate to within _W percent based on monitor-specific design and condition,
inspected during scheduled maintenance once  every 3 years or inspected immediately
prior to regeneration for semi-regenerative CRU.

      Response:  We agree with the commenters' suggestion on calibration
requirements for monitoring devices for a semi-regenerative  CRU equipped with a wet
scrubber. In response, we have revised the proposed rule to allow calibration of
monitoring devices for a semi-regenerative CRU equipped with a wet scrubber prior to
each regeneration. Calibration requirements for monitoring  devices for other types of
CRU remain at 3 months. We also have increase the accuracy specification for wet
scrubber monitoring devices to 10 percent as suggested by the commenters.
Additionally, the pressure drop monitoring requirement has been removed as this
parameter is specific to Venturi-type scrubbers used for PM control and is not relevant
to adsorptive scrubbers, such as those used to control HC1.

      5.10  Monitoring Requirements for Catalytic Incinerators

      Comment:  Commenter IV-D-26 does not consider the daily average upstream
temperature and the daily average temperature across the catalyst bed is the best way
to monitor catalytic incinerators.  If the VOC loading of the process is variable, the
temperature rise across the catalyst will vary accordingly. The temperature rise is
roughly directly proportional to the  VOC concentration entering the catalyst.  If the
VOC loading is low for a considerable period of time, the temperature rise will also be
low even though the catalyst may be performing at the required destruction efficiency.
He suggests annual testing of the catalyst and inspection of the oxidizer for
mechanical integrity. However, maintaining a preset catalyst inlet temperature is the
most common control mode for a catalytic oxidizer.

      Response: We agree with the  comment in general; however, we do not expect
any one to use this type of control technology for their CCU, CRU, or SRU vents and
therefore, have dropped the provisions from the rule.

                                     5-10

-------
      5.11  Monitoring Requirements for CRU With Internal Scrubbers

      Comment:  Commenters IV-D-47 and IV-D-53 state that pressure drop and gas
flow rate are not appropriate operating parameter to be monitored for CRU with
internal scrubbers. Internal scrubbers are different because they do not use trays or
packing to promote liquid-gas contact. Typically, these unit are characterized by
liquid injection into the regeneration gas stream followed by liquid removal in a
gas-liquid separator vessel. Pressure drop can be caused by fittings, heat exchanger
tubes, and other obstructions instead of a scrubber tower containing trays or packing.
Measurement of gas flow rate is a problem.  This is typically measured using orifice
plates. These type of plates will not provide an accurate measurement of gas flow
during the regeneration process. In some systems, the location of the plate will not
provide a measurement of total flow to the scrubber.  In these systems, the plate
would have to be relocated or additional flow measurement devices would have to be
installed to account for the addition of other streams. They recommend revising the
proposed rule to distinguish between an external scrubber using trays or packing and
an internal scrubber where only the gas flow rate and total water (or scrubbing liquid)
flow rate to the  scrubber would be monitored.

      Response: The proposed standard did not require monitoring of pressure drop
and gas flow rate for CRU with no add-on control devices.  The proposed standard
required the owner or operator to include recommended monitoring requirements in
the part 70 or part 71 permit application. As discussed in comment/response 5.7, the
pressure drop monitoring requirement for CRU wet scrubbers was also dropped.  In
addition, we have added specific requirements to the final rule and clarified the
language in the  rule to distinguish between units with an add-on control  device (i.e,
wet scrubber) and no add-on control device (internal scrubbing system).  These
requirements include a initial performance test to measure HC1 emissions with EPA
Method 26 and  during subsequent regeneration periods monitoring of the HC1
emission concentration every 4-hours during the coke burn and rejuvenation cycles
using colorimetric methods such Draeger tubes.

      5.12  Continuous Emission Monitoring

      Comment:  Commenters IV-D-26, IV-D-31, FV-D-46,  and IV-D-56 urge EPA to
require GEMS.  Commenter IV-D-31 supports requirements for HC1, TRS, and either
CO or THC or TOG for all new sources rather than the limited requirements in the
proposed standard because they are feasible (as stated in the BID), ensure standards
are met at all times,  and provide better HAP monitoring. If these technically-proven
systems are in place at more than 12% of existing refinery sources, then they should
be required for existing sources as well as new sources. Commenter IV-D-26 also
urges EPA to require GEMS where proven in similar service. The cost of these
systems is decreasing and they may no longer be too costly. The continuous
monitoring of a  process allows the operator greater flexibility in operation which
could result in increased output, improved efficiency, and overall cost savings.
Commenters IV-D-46 and IV-D-56 specifically request GEMS for TRS  limits.  GEMS

                                     5-11

-------
for TRS are commercially available currently used at a number of Texas refineries to
demonstrate compliance with the NSPS. Due to the TRS emissions from refineries
and numerous exceedances, more accurate information is needed to assess
compliance than operating parameter values.

      Response: We agree with the commenters' recommendations that NSPS
experience with GEMS demonstrate their technical and economic feasibility for this
industry, provide better data, and need to be encouraged.  In determining monitoring
requirements, EPA looks at the various options from a top-down approach.  One of the
options examined was requiring GEMS or COMS for all affected sources under this
rule. We did not select this option because of the high capital and operating costs.
However, we have re-examined these options after proposal and looked carefully for
ways to encourage their use or require their use if needed. As a result, we have
included options in the rule allowing plants to choose to comply with the NSPS limits
and the corresponding monitoring requirements. We have also included requirements
in the rule for the  use of COMS  for catalyst regeneration vents for CCU  with
throughput capacity greater than 20,000 bpd (and not using wet scrubbers) and have
added COMS as a  monitoring option for smaller CCU (see comment/response 5.5).
COMS are required for the larger units under State implementation plan requirements
in 40 Part 51, Appendix P; therefore these costs are not attributable to this rule.

      A GEMS for TRS emissions also is required for SRU with no add-on control
device. The cost of GEMS for these units is reasonable and does not pose any
economic hardship for plants that do not use a control device. The Agency  is also
confident that in those cases where process or control device parameter monitoring is
allowed in place of a GEMS, that this provides adequate assurance of continuous
compliance.

      5.13   Annual Stack Tests

      Comment:  Commenters IV-D-31 and IV-D-56 recommend that annual stack
testing should be required to confirm the integrity of the process values used for
parametric monitoring wherever continuous emission monitoring is not used. An
example is premature wear of components within the CCU due to the abrasive nature
of the catalyst fines which can lead to changes in particle size  distribution and
increased quantities of fines to the control equipment. Either of these conditions
could lead to decreased efficiency of the collection device with no measurable change
in the monitored process values.

      Response: The EPA feels  that the monitoring requirements in the final rule
which are based on the "top down"  monitoring approach are adequate to ensure
continued compliance for these refinery units.  The use of GEMS, COMS,  and CPMS
as appropriate  eliminates the need for repeat source tests. Therefore, the final rule
does not include requirements for annual or repeat stack tests. The EPA regions can
use the section 114 authority under the CAA to request a source test should they
believe there is premature wear of process or control device components that would

                                    5-12

-------
result in increased loading to the control device and/or increased emissions to the
atmosphere.

      5.14  ESP Plate Area and Conditioning Agents

      Comment:  Commenters IV-D-41, IV-D-31, and IV-D-56 recommend that EPA
request data on the ESP plate area (square feet per 1,000 actual cubic feet per minute
of gas flow). EPA could also require a minimum plate area of 300 to 350 square feet
per 1,000 actual cubic feet per minute for reliable performance with the PM NSPS.
EPA also should request information on the use of humidification or commercial
conditioning agents since they enhance ESP performance to ensure that such agents
are used routinely and not merely during performance tests.

      Response: The EPA did collect considerable information on the ESP plate area
used for these devices when applied to CCU catalyst regeneration vents. The
information and data were used in the control device designing and costing efforts.
However, we did not include a design or equipment specification standard in the CCU
standards because the prescription of a performance standard was feasible and thus
takes priority over establishing an equipment standard.  We have taken the
commenters recommendation and revised the performance test and recordkeeping
requirements to require the owner or operator to maintain records on subsequent use
of any condition agents used during and after the performance test. Additionally, we
have modified the ESP monitoring requirement to include actual gas flow rate through
the ESP as this is a key parameter in ESP control efficiency given a fixed plate area.

      5.15  Opacity Monitoring for Non-NSPS CCU

      Comment: Commenter IV-D-43 believes opacity monitoring, coupled with
control device operating parameter monitoring, is a better measure of compliance with
the PM standard for CCU regenerator vents as compared to coke burn rate.  NonNSPS
units should have the option to elect opacity monitoring identical to the NSPS
requirements even though they are not subject to the NSPS.

      Response: We agree with the  commenter and included the NSPS requirements
as an option for "non-NSPS" affected sources in the final rule. Because data from
these systems will be used to demonstrate initial and continuing compliance,
continuous emission and opacity monitoring systems must meet the operation,
maintenance, and quality assurance requirements in the NESHAP General Provisions
and the requirements of each applicable performance specification in Appendices B
and F of 40 CFR Part 60. As discussed in other responses, opacity monitoring
requirements have been added to the final rule such that those CCU with a
throughput greater than 20,000 bpd and not using a wet scrubber are required to
install and operate a COMS. Again, this requirement is based on the state
implementation plan requirement in 40 CFR Part 51, Appendix P.
                                    5-13

-------
      5.16  Daily Averages for Monitoring Systems

      Comment: Commenter IV-D-46 opposes daily averages for combustion device
operating parameters, such as combustion temperatures. Use of daily averages for
combustion unit performance averaging times will allow short term high emission
transients associated with combustion upsets. These short term events will frustrate
the entire purpose of establishing a standard for organic HAPs. All averaging times
for compliance purposes on both emissions and parameter monitoring should be
revised to be no longer than 1-hour.

      Response: The EPA agrees that the monitoring of process and control device
operating parameters should be done in terms of hourly averages and we have revised
the final rule to specify monitoring data collection in terms of hourly averages.  We
have not  however revised the designations of the operating standards (i.e., previously
referred to as excess emissions for control devices) that were proposed in terms of a
24-hour average. The short term fluctuations of the control device operating
parameters is not expected to have a significant influence on the overall emission
levels for these units. Therefore, as was done for this source category in the initial
refinery standards (i.e., MACT I) for process vents and other organic source control
devices, the designation of operating standard is made in terms of 24-hour averages in
most cases. Any deviation from these operating standard values is required to be
reported to the Administrator.

      5.17  Draeger Tubes for Monitoring of Scrubbers

      Comment: Commenters IV-D-47, IV-D-49, IV-D-53, and IV-D-59 recommend
allowing  the use of Draeger tubes for the monitoring of emissions from CRU
scrubbers.  This is a common practice in the refining industry and can be easily
incorporated in the monitoring requirements.  Commenter IV-D-59 specifically
requests use of Draeger-type colorimetric tubes or monitoring the condensate pH to
determine compliance with HC1 emission standards for semi-regenerative CRU. The
commenter contends that outlet monitoring using the appropriate EPA test methods is
technically infeasible because the vent is at the top of the vessel and thus, not
designed for worker access.  Worker exposure to pressure relief valve discharges
would be dangerous and would not be allowed by company or federal occupational
safety rules. Also, the vent outlets do not have the geometry needed for stack testing.
This commenter operates a semi-regenerative unit with an internal scrubber for
caustic spray injection.

      Response: The proposed standard required the owner or operator to include
recommended monitoring requirements for CRU with no add-on control device such
as the configuration described by the commenter (i.e., semi-regenerative unit with an
internal scrubbing system). Outlet monitoring may not be a feasible approach for the
reasons described by the commenter. For this reason, we revised the rule to require
Method 26 rather than 26A for performance tests and to include specific  procedures
for monitoring CRU with no add-on control device that allow use of colorimetric

                                     5-14

-------
methods.  Colorimetric (e.g., Draeger) tubes can provide reasonable performance
monitoring in this particular set of circumstances.  The monitoring procedure varies
according to the concentration range to be measured. For example, a continuous
pump may be required over a hand pump in some cases. Monitoring the pH of the
scrubbing solution condensate is also considered to be an appropriate measure of
scrubber performance for these system applications, and pH monitoring requirements
have been added to the rule for CRU.

      5.18  Method 26 vs Method 26A for HC1 Emissions from CRU

      Comment: Commenters IV-D-47 and IV-D-53 believe Method 26A for
determining HC1 concentration and mass flow rate may not be feasible for CRU due to
potentially hazardous process conditions at sampling locations. For example,
determination HC1 removal efficiency by a CRU internal scrubber using Method 26A
may be difficult and dangerous due to system operating conditions at the scrubber
inlet (high temperature and high pressure). Safety considerations prohibit the use of
26A for quantifying the HC1 content of the inlet gas. Alternative sampling procedures
to determine HC1 and mass flow rate are needed. The commenter recommends the
non-isokenetic Method 26 as an alternative with a sampling rate of at least 0.001 dry
standard cubic meters per minute. To calculate HC1 mass flow rate, the gas flow rate
at the inlet and outlet of the control device may be determined using the calibrated
process flow monitoring equipment.

      Response: We agree that it may be technically difficult to measure a percent
reduction for a CRU that use internal scrubbing systems, i.e., with no add-on control
device. Concentration standards were included in the rule for this reason. We also
agree that the non-isokenetic Method 26 with the sampling rate recommended by the
commenter should be used for performance tests for this situation and have revised
the rule accordingly.

      5.19  Monitoring Requirements for SRU without Combustion Device

      Comment: Commenters IV-D-47 and IV-D-53 recommend that monitoring
requirements for SRU without a combustion device (i.e., a tail gas cleanup unit) be
included in the rule  rather than requiring the owner/operator to submit a
recommended CPMS. The commenter recommends that EPA adopt the monitoring
requirements in 60.105(a)(6) and (a)(7) of the NSPS for this source.

      Response: We agree that specific monitoring requirements should be included
for SRU with no add-on control device and have revised the proposed rule to include
the provisions recommended by the commenter. These are the NSPS monitoring
requirements for Glaus SRU with reduction control systems not followed by
incineration.  Section 60.105(a)(6) and (a)(7) require a GEMS to measure the
concentration of reduced sulfur and O2 emissions with a span value of 450 ppm
reduced sulfur and 10 percent O2. Provisions also are included for performance
evaluations by Performance Specification 5 for the reduced sulfur and O2 monitor and

                                     5-15

-------
for using an air or O2 dilution or oxidation system that converts the reduced sulfur to
SO2 as an alternative to a reduced sulfur monitor.

      5.20  Monitoring Requirements for Flares

      Comment:  Commenter IV-D-56 recommends that the rule require routine
testing of all emergency and process flares rather than routine monitoring since
facilities claim a combustion efficiency of 95% or more, but do not base their claims
on actual emission test results. The actual combustion efficiency needs to be
demonstrated and not merely assumed. EPA needs to address methods for flare
testing with available technology such as remote sensing systems, ultraviolet, infrared
lasers, and other technologies.

      Response: We agree with the commenter that there are more modern methods
for flare testing that are rapidly becoming available and EPA will be looking at these
in the near future. However, the flare provisions in 63.11 of the NESHAP General
Provisions do include requirements to ensure that flares are properly sized and
operate according to their design and these design requirements were based on a 98%
destruction efficiency.  Testing of flares is not being required in the final rules.

      5.21  Exceedances and Excursions

      Comment:  Commenter IV-D-25 does not support allowing one exceedance or
excursion by the same  control device during a 6-month period. This may allow the
facility to exceed the standard once regardless of whether it can justify the
exceedance. Problems that occur regularly but infrequently may also be ignored.
Further, the "one exceedance" standard eliminates enforcement discretion in
addressing these types of exceedances or excursions. EPA should adopt the proposed
rule without the condition that requires one or more exceedances or excursions in a
semi-annual period to be a violation.  Commenter IV-D-56 strongly opposes allowing
potential exemptions for one excursion per semi-annual reporting period. The
commenter describes several problems with specific refineries in his community
regarding exceedances to SRU upsets.  EPA must not allow such excursions to take
place without appropriate enforcement, penalties, and corrective actions.

      Commenters IV-D-39, IV-D-47,  and IV-D-53 believe the proposed rule should
be made consistent with the NESHAP for Hazardous Organics from the Synthetic
Organic Chemical Manufacturing Industry (HON rule).  Here, six exceedances  or
excursions are allowed during the first semi-annual period. The number of excused
exceedances or excursions is then reduced by one for each successive reporting period
until the sixth period.  From the sixth period onward, only one exceedance  or
excursion is allowed. Commenter IV-D-59 requests that EPA adopt the NSPS
requirements for reporting emission exceedances and monitoring system
malfunctions. Here, additional information is requested once a reasonable percentage
threshold has been exceeded.
                                    5-16

-------
      Commenter IV-D-53 asks that "exceedance", "excursion", and "violation" be
defined, that the requirement for 75% percent data availability not automatically
trigger an excess emissions report, and that provisions be added for gaps in
monitoring data used to verify compliance. This can be done in a way similar to the
NSPS where if the duration of monitoring malfunctions is greater than 5% of the
operating time for the period, a higher level of reporting detail is required.

      Response:  In the final rule, the EPA has revised the format and terminology
used in the rule and no longer uses the terms "exceedances", "excursions", "excess
emissions", and "violations". The final rule uses the term "deviation" rather than
"exceedance," or "excursion," etc.  A deviation is any instance in which an affected
source or the owner or operator or an affected source: (1) fails to meet any obligation
or requirement in the rule, (2) fails to meet any term or condition in the operating
permit, or (3) fails to meet any emission limitation or work practice standard during
startup, shutdown, or malfunction regardless of whether or not the failure is
permitted by the rule.  Consistent with 63.6(e), we (the EPA) will determine if a
deviation is a violation of the NESHAP.  Under the new NESHAP format, these
provisions are being included in new MACT standards to improve the understanding
and consistency of our rules.

      The final rule retains the 75% data availability requirement for CPMS.
Obtaining the required monitoring data for 75% of the hours the process operates is
the basic method of demonstrating continuous compliance if you use a CPMS.  The
final rule does not include provisions  for filing data gaps using alternative monitoring
procedures. Plant owners and operators can use a backup monitoring system for this
purpose or apply to EPA for approval of an alternative parameter or monitoring
method for use when the primary system fails.

      The information to be  reported  in the semiannual reports required by the final
rule is nearly identical to the information required for the NSPS excess emission
reports except that there is no 5% trigger for the duration of monitoring malfunctions.
Any deviation must be reported, including a startup, shutdown, or malfunction.

      Under the new format, we have established operational standards for
continuous compliance with emission limits and parameter monitoring. For example,
the owner or operator must establish a minimum operating temperature for any
thermal vapor incinerators used as control devices. Similar parameter values must be
established for other control  device types and process parameters as specified in the
rule.  In establishing the operating limits, we strongly considered the NSPS
definitions of conditions that constitute excess emissions or violation that take into
account process and control device fluctuations over time.  In addition, the final rule
does not provide for a fixed (or declining) number of deviations over each reporting
period as is done in the HON rule. No data were available to define or establish the
conditions, or a typical number of exceedances or deviations, that occur over a given
time period under what could be termed normal operations.
                                     5-17

-------
6.0    PERFORMANCE TEST REQUIREMENTS

      6.1    Determine Reactor Pressure During Performance Test

      Comment: Commenter IV-D-27 recommends that CCU reactor pressure be
recorded during the initial performance test. It is not uncommon for a source to
continue operation with one or two fields down in an ESP. A worst case stack test
would include determining how many fields can be down before maintenance is
performed or load must be dropped.

      Response: We agree that it is important to know ESP performance and
operating conditions especially during a performance test. However, reactor pressure
is not the most direct measure of control system performance for ESP applied to CCU
vents and it is not being utilized as a performance measure in the final standard. We
have revised the rule to require recording of the total number of fields in the ESP and
how many are operated during the performance test. We are also requiring that the
owner or operator maintain records of any change in the number of fields operating of
the ESP over time.

      6.2    Determine Maximum Amount of Recycle During Performance Test

      Comment: Commenter IV-D-27 says  a wet scrubber can operate with 100%
fresh feed during a stack test but run some percentage recycle during normal
operation. A worst case stack test would be the maximum amount of recycle a source
would use.  This could also be included as a parameter for excess emission
determinations.

      Response: We agree that the amount of make-up scrubbing liquid used is  an
important factor in scrubber performance for HC1 scrubbers; we  do not believe that
the efficiency of the CCU PM scrubber will be strongly related to the amount of
recycle used, however. We considered  having recycle scrubber liquid as a monitored
parameter but decided against it since it is not a direct measure of control device
performance and it limits operational flexibility. We have concluded that the most
critical parameter that relates to scrubber efficiency for a HC1 scrubber while not
limiting operational flexibility is pH of the scrubbing liquid.  We have revise the  rule
to record the pH of the scrubbing liquid during the performance test and to require a
minimum pH operating parameter limit that must be maintained.

      6.3    Determine All HAP Metals During Performance Test

      Comment: Commenter IV-D-28 urges EPA to require the CCU catalyst
regenerator vent performance test to include all relevant HAP metals because of the
current lack of emission data and the need to develop residual risk standards in the
future.
                                    6-1

-------
      Response: The EPA agrees with the commenter that additional metal HAP data
would be useful in developing the residual risk standards in the future.  However it
did not seem reasonable to require that those facilities that chose to comply with the
PM standard be required to analyze for the HAP metals.  On the other hand, facilities
conducting a performance test to comply with the Ni standard could obtain this data
at the lowest incremental cost. Therefore, the performance test requirements for
facilities complying with the Ni standard include determining and reporting the
concentrations for other metal HAP in addition to Ni.

      6.4   Conditions Requiring New Performance Tests

      Comment: Commenters IV-D-43, IV-D-49, and IV-D-59 oppose new
performance tests due to changes in process or operating conditions while Commenter
IV-D-27 recommends that change in catalyst or switching from a clean catalyst to a
used catalyst should be added to the list of examples because either of these can result
in increased PM and/or opacity.  Commenter IV-D-43 opposes the requirement to
retest because it is redundant to the prevention of signification deterioration and new
source review rules that require an emissions impact analysis for unit operations and
operational changes that might increase emissions. Also, feedstock change and
capacity increase are not always precursors to an increase in particulate emissions.
Events most likely to create an increase are those where particulate-contact elements
of the process or control device are replaced or modified. Commenter IV-D-49
opposes the requirement based on feedstock changes. He explains that CCU operate
over a range of process conditions, some of which are seasonal and some are
frequently repeated. Feedstock changes by themselves do not cause changes in HAP
emissions and this  is not required by the NSPS. This commenter believes the
provisions can be interpreted by enforcement agencies to require retesting for any
number of process  condition changes.

      Response: The final standards do not contain any requirements that would
directly call for a repeat performance test. Under the final rule, a repeat performance
test may be conducted as part of the owner or operator efforts to change the
established level of control device or process operating parameters. When changing
these  values the owner or operator may conduct a performance test, a performance
test in conjunction with an engineering assessment, or simply conduct an engineering
assessment to verify that, at the new control device or process parameter level, the
unit complies with the applicable emission standard.  Under the final rule, you must
establish a revised value for the monitored process or operating parameter to
determine or demonstrate compliance under the new operating conditions if any
change to the process or operating conditions (including but not limited to changes in
catalyst, feedstock, capacity, control device or capture system) that could result in a
change in the control system performance has been made since the last performance
or compliance tests or assessments were conducted.  The repeat performance test
requirements were dropped because the monitoring requirements for the CCU were
enhanced (COMS, etc., see comment/responses 5.4 and 5.10) so that compliance with
the emissions standards can be assured without requiring additional source tests.

                                     6-2

-------
      6.5   Test Conditions for CCU Regenerator Vent

      Comment: Commenter IV-D-43 opposes the provisions in 63.1564(e) of the
proposed rule for conducting the initial performance test at the maximum operating
capacity for the process and while operating the control device at conditions which
result in the lowest emission reduction.  The maximum operating capacity does not
represent severe conditions for the control device and operating the control device at
conditions which result in the lowest emission reduction implies testing under
end-of-run conditions. The commenter recommends stating that the tests should be
conducted  under conditions that characterize the throughput and operating
conditions projected for the control device under the current configuration of the
process.

      Response: The EPA agrees with the comment and has revised the performance
test requirements such that the test can now be conducted under conditions that
characterize the unit operating conditions that are representative for the unit and
control device under the current configuration of the process operations.  Monitored
operating parameters that are measured during the test to establish the appropriate
operating range for the process or control device can be adjusted or revised based on
the test data at representative conditions and engineering analysis to set the maximum
conditions for the parameter values used to show continuous compliance. The final
rule also includes equations to be used to adjust certain operating limits when the
control variables are expected to be linearly related.

      6.6   Use of Engineering Analysis to Establish Limits for Process/Operating
            Parameters

      Comment: Commenters IV-D-47,  IV-D-49, IV-D-53,  and IV-D-59 believe that
engineering analyses are needed to adjust performance test results used to establish
process or operating parameter values. According to the commenters, it is unlikely
that tests will be conducted under worst cases conditions.  It is more likely that an
affected source will operate at a level below the standard during the compliance test
and the operating parameters established during the test will be unduly low.
Therefore, the results of the performance test need to be adjusted to be equivalent to
the proposed standard. It also is unlikely that the control device can be operated  at
worst case conditions. Without a GEMS, the operator really doesn't know at what
point the control device has been turned down too far and exceeded the limit.

      Response: As noted in the previous response, under the final rule EPA allows
that monitored operating parameters measured during the  performance test to
establish the appropriate operating range for the process or control device can be
adjusted or  revised based on the test data at representative conditions and engineering
analysis to set the maximum conditions for the parameter values used to show
continuous  compliance.  For certain conditions, appropriate equations have been
added to the rule to aide in making these adjustments.
                                     6-3

-------
      6.7   Performance Test for Organic HAP from CCU

      Comment:  Commenter IV-D-43 believes that the performance test should be
the same for NSPS and nonNSPS CCU. An initial Method 10 test should be the
measure of compliance for the MACT just as for the NSPS. Under the proposed
MACT, EPA treats the CO GEMS readings (of units required to install them) as if they
are a compliance measure. This is not the case for NSPS compliance where an
instantaneous reading over 500 ppm is only an indication that the unit has gone out of
equilibrium due to a process fluctuation; and is not a violation or non-compliance
event. The commenter also noted that corrective measures typically require more
than an hour to restore equilibrium.

      Response:  It is important to note that NESHAP requirements under section 112
differ greatly from NSPS provisions under section 111.  For one thing, the CAA
requires a reasonable assurance of continuous  compliance for MACT standards under
section 112. And, continuous monitoring data provide this evidence of compliance.
Therefore in the final MACT standard provisions are included that establish
requirements for continuous compliance or operating limits.  Any deviation from the
operating limits must be reported to the permitting authority in the semi-annual
report. Consistent with 63.6(e), the EPA will  determine if a deviation is to be
considered a violation of the standard. Under section 112, a violation can be assessed
with a financial penalty for these operating standard violations. The commenter is
correct in that a violation of the NSPS emission limit can only be determined through
a source test.

      6.8   Early Compliance Certification

      Comment:  Commenter IV-D-43,  IV-D-49, and IV-D-59 asks that the final rule
allow the performance test to be scheduled anytime during the 3-year compliance
window after promulgation, plus 150 days. This will alleviate any potential
scheduling problems due to the shortage of qualified stack testing firms. A facility
should not have to wait until the proposed testing period because most facilities will
see if they can achieve compliance without controls and such testing must be done
well before the final compliance date.  Testing for newly installed controls also needs
to be performed prior to the compliance date so that adjustments can be made if
needed.

      Response: The point made by the commenter is valid. We have revised the
rule to change requirements for the performance test report to provide for the initial
performance test and report anytime from the date of promulgation of the final rule to
the date 150 days  after the compliance date (3 years from the promulgation date). In
this way, you, the owner or operator, can do your performance test and test reports
anytime from the date of promulgation or the next 3 year period. However, the
performance test and report must not be any later than 150 days after the 3-yr period.
                                     6-4

-------
      6.9   Equation 2 for Calculating Coke Burn-off Rate

      Comment: Commenters IV-D-47, IV-D-48, and IV-D-53 state there should not
be a percent in the denominator of the constant K2. In Equation 2 for coke burn rate in
the proposed rule.

      Response: We agree with the comment; the inclusion of the percent symbol
was a typographical error and it has been removed from the final equation.

      6.10  Alternative Coke Burn Rate Equation

      Comment: Commenters IV-D-49 and IV-D-59 request use of site-specific
equations for coke burn rate. Commenter IV-D-49 states a several methods are used in
the industry to calculate the coke burn rate; the usefulness of the calculation depends
on its repeatability. The rule needs to allow any reasonable alternative calculation
method.  This would provide flexibility in monitoring emissions and be consistent
with site-specific current practices.  The alternative calculation would be calibrated to
the EPA method during the performance test. For example, if emissions during the
stack test are 1 Ib particulate/M Ibs coke with a coke burn rate of 28,000 Ibs/hr using
EPA coke burn calculations and the  equivalent alternative process calculations show a
coke burn rate of 30,000 Ibs/hr, the target coke burn rate is 30,000 Ibs/hr and not
28,000 Ibs/hr.

      Response: We feel that it is important for both implementation and
enforcement that there be a single equation for this parameter. Therefore we are
maintaining the single equation to provide a consistent method of making the
determination.

      6.12   Method  5B and 5F for PM

      Comment: Commenters IV-D-49 and IV-D-59 support use of Method 5B and 5F
to measure PM.  According to the commenter, these methods measure the portion of
the PM that relates to HAP by subtracting out the condensible sulfate particulate.
There is no Ni in the condensible particulate portion of the CCU regenerator vent
emissions. State standards for PM vary according to their purposes and specifying 5B
or 5F regardless of the State requirements would insure consistency.

      Response: EPA has maintained the use of Method 5B and 5F in the final rule to
allow the owner or operator to deduct or subtract the mass of sulfate PM measure
during the performance test.  This fraction of the sample would not have any HAP
metals and therefore should not be counted in the results. We should also point out
that the final rule does not include use of Method  18 for organic HAP because of
technical limitations with this method as applied to CCU.
                                     6-5

-------
(This page intentionally blank]

-------
7.0   STARTUP, SHUTDOWN, MALFUNCTION, AND MAINTENANCE

      7.1   Provisions for Planned Maintenance

      Comment: Commenters IV-D-37, IV-D-42, IV-D-44, IV-D-47, IV-D-49, IV-D-53,
and IV-D-59 state that provisions are needed to accommodate planned maintenance of
control equipment, particularly for the CCU regenerator. The commenters feel that
the CCU should be allowed to operate during periods when the control device is out
of service for maintenance overhauls because the use of preventative maintenance
results in less environmental impacts. Commenter IV-D-43 adds there should be
provisions in the rule for unanticipated maintenance downtime for control devices
while the CCU is running. Full unit shutdown and startup to repair such devices
results in a greater risk of excess emissions than performing an on-line repair where
possible.

      According to the commenters the industry average for CCU turnarounds is 4-6
years and up to 10 years for an ESP or wet scrubber.  Preventative maintenance is
needed more frequently. The turnaround process typically takes about four to six
weeks. Provisions for planned maintenance have been adopted by Texas, California,
New Mexico, and Louisiana and Montana has a provision calling for annual shutdown
of the CO boiler for routine inspection and maintenance. Commenter IV-D-37 also
discusses cases where a control device, such as a wet scrubber, is common to two
CCU. Facilities with a unit sharing common control equipment can not turn around
both units at the same time. For this reason, plants need to shutdown the control
equipment during every other scheduled turnaround. The commenter does not
believe it is reasonable to require redundant control systems due to the costs, the
infrequent nature of downtimes, the resulting emissions, and the economic penalty to
the refinery associated with shutdown of a major process unit. This commenter
recommends the final rule include provisions allowing, subject to approval by the
applicable permitting authority, that would require the SSMP to include specific steps
to minimize emissions during a planned maintenance period.  Such a plan might
specify that the site conduct ambient air quality monitoring to ensure that CO and PM
standards are not exceeded.

      Commenter IV-D-42 explains how flue gas diversion devices and by-pass stacks
are used to allow continued operation of the CCU when the control devices are out of
service for mechanical repairs. The commenter also discussed how malfunctions
require a by-pass of control devices or may cause a control device to shutdown while
the unit continues to operate.  The commenter suggests that short duration
exceedances be allowed for control device startups, shutdowns, and malfunctions.
These emissions may remain estimated but not monitored; the magnitude of the
emissions can be limited by setting duration limits.

      Commenters IV-D-30, IV-D-31, and IV-D-56 believe EPA should include
requirements for industry to estimate uncontrolled emissions of criteria and metal
HAPs during such a maintenance period and require the facility to make up for the

                                     7-1

-------
released emissions during periods of normal operation by "over controlling". For
example, additional emission reductions can be achieved by fabric filters, improving
ESP performance with more plate area or increasing the power supply/voltage,
increasing pressure differential on scrubbers, using SOX reducing catalysts, and
reducing CO2 emissions by employing catalysts that reduce coke make.  The
commenters also recommend that the rule require plants to collect data during
abnormal operations because it is important for the regulatory authority to know the
volume of uncontrolled emissions as well as the frequency and duration of emissions.
Commenter IV-D-29 asks EPA to consider limiting the number of times a facility
would be allowed to have excess emissions  resulting from non operation of a CCU
control device during planned routine maintenance.  Commenter IV-D-46 believes
monitoring averages should include periods of non-operation of emission control
devices. Failure to operate a control device  should be discounted by allowing the
source to escape the deterioration of a measured parameter implicit in such a
circumstance.

      Response: The EPA would like to encourage planned maintenance related to
both processes and control devices,  especially when that maintenance yields an
environmental benefit.  However, only one of the  commenters provided any
substantive information on what particular planned maintenance events they
specifically would like to perform and the net environmental benefit achieved by
these actions.  Consequently, in response to  these comments, we revised the rule to
include provisions allowing the permitting authority to approve a period of planned
routine maintenance for a refinery with multiple CCU served by a single wet scrubber
emission control device. During this pre-approved time period, the refinery may take
the control device and/or one of the  process  units out of service for maintenance while
the remaining process unit(s) continues to operate. To obtain approval, you must
submit a written request at least 6 months before the planned maintenance is
scheduled to begin that contains the specified information and data. This includes:

     A description of the planned routine  maintenance and why it is necessary;
     The date the maintenance will begin  and end;
     A quantified estimate of the emissions (including HAP and criteria pollutants)
      that would be released with an analysis of the environmental benefits (i.e.,
      emission reduction) that would result as opposed to delaying the maintenance
      until the next unit turnaround;
     Actions to be taken to minimize emissions  during the period.

You must include a copy of the request in the compliance report due for the period
before the planned maintenance is scheduled to begin. In the compliance report due
after the routine planned maintenance is complete, you must provide followup
information on the maintenance including the number of hours the control device did
not operate.

      Other than the situation discussed above, the rule does not excuse or  exempt
the refinery owner or operator from  meeting the specified emission limits and

                                     7-2

-------
monitoring requirements during periods when the control device is not in operation;
this would include periods of control device startup, shutdown, and malfunction.

      7.2   Storage of Intermediate Products Through Duration of Maintenance

      Comment: Commenter IV-D-27 explains that during SRU maintenance
shutdowns, the sour gas was routed to the flare for about 2 weeks every 1 or 2 years
and the refinery was subsequently required to permit the emissions. Afterwards,
when the maintenance was scheduled, the refinery would store intermediate products
that had already been cracked to last the duration of the maintenance. The final rule
should address this type of scenario.

      Response: The Agency does not have adequate industry-wide data on the
operating practices for these  units during process turnarounds to formulate the type of
work practice standards that the commenter recommends.

      7.3    Maintenance Plan Requirements

      Comment: Commenters IV-D-37, IV-D-47, and FV-D-53 oppose the proposed
requirement for a maintenance schedule consistent with the manufacturer's
instructions and recommendations for routine and long-term maintenance in the
maintenance schedule. According to the  commenters, manufacturer
recommendations are designed to support their warranties and to protect from claim;
emissions are not their primary consideration.  EPA does not have the data to support
their position that strict adherence to the  manufacturers recommendations results in
decreased HAP emissions or that the best plants in the industry are strictly adhering
to these practices. The commenter finds the proposed requirement unnecessary and
overly restrictive. Maintenance requirements should be consistent with practices
needed to  ensure good air pollution control as required by the NESHAP General
Provisions.

      Response: We have maintained the language in the requirement that
maintenance should be consistent with the manufacturers' recommendations; but
have clarified that this does not mean that the facility must strictly adhere to the
manufacturers instructions or that the facility maintenance plan must be identical to
the manufacturers' recommendations.

      7.4    HAP Emissions from Startup, Shutdown and Upset Conditions

      Comment: Commenter IV-D-28 recommends that the potential  HAP emissions
associated with startup, shutdown, and upset conditions of the SRU and CCU be
addressed in the final rule. At minimum  there should be test requirements to
determine emissions when elevated levels of HAP are likely because these data will be
needed in future residual risk assessments.
                                     7-3

-------
      Response:  The EPA, under the General Provisions to 40 CFR Part 63, specifies
requirements associated with startup, shutdown and malfunctions. We did not
include specific provisions in this rule for testing or reporting of HAP emissions
during these periods and did not include them in the impact analysis for the rule.

      7.5   Reporting Malfunction Events

      Comment:  Commenters IV-D-46 and IV-D-56 do not agree that the proposed
rule should allow a source to report only those malfunction events that were not
managed in accordance with the startup, shutdown, and malfunction event. All
malfunction periods should be reported immediately and in quarterly rather than
semi-annual reports. All excess emission events should be considered as potential
violations.

      Response:  The EPA's Part 63 General Provisions do this (i.e., allow a source to
report only those malfunction events that were not managed in accordance with the
SSMP) to reduce reporting burden.  However, these events must be reported in the
next semi-annual  report. The EPA will determine whether the deviation that occurs
during a startup, shutdown, or malfunction is a violation according to 63.6(e).
                                     7-4

-------
8.0   RELATIONSHIP TO NSPS AND OTHER RULES

      8.1   Relationship of NSPS to MACT Standard

      Comment: Comments on the proposed rule revealed a wide range of
interpretations on the relationship of the refinery NSPS to the MACT standard and
questions regarding which requirements apply under different situations. These
comments seem to fall into three categories: (1) whether NSPS are or are not affected
sources under the MACT standard, (2) are excess emissions and malfunctions by an
NSPS unit subject to the provisions under section 111 or section 112,  (3) is GEMS
data from an NSPS unit directly enforceable as it would be under a MACT standard
and are NSPS units required to comply with Appendix F (not presently required), and
(4) can plants select to comply with the NSPS or MACT standard and do the various
options under the MACT standard apply to NSPS units.

      Response: Because of the wide range of questions and interpretations about the
status and requirements for NSPS units, we have revised the proposed rule to
explicitly state all requirements. First, all FCCU and SRU are affected sources under
this NESHAP.  This includes units subject to the NSPS and those that are not.
Second, the requirements of this rule in no way change the NSPS requirements. A
unit that is subject to the NSPS must comply with all NSPS requirements. To reduce
regulatory overlap, the requirements of this rule for the control of HAP contain
portions of the emission standards and monitoring requirements of the NSPS. For a
FCCU catalyst regenerator vent subject to the NSPS for PM emissions, the HAP metal
emission limits are the same as the PM emission limits in 40 CFR 60.102; the HAP
metal monitoring requirements for FCCU catalyst regenerator vents are the same as
the NSPS requirements in 40 CFR 60.105(a)(l), 40 CFR 60.105(c), and 40 CFR
60.105(d). For a FCCU catalyst regeneration vent subject to the NSPS for CO
emissions, the HAP organic emission limit is the same as the NSPS limit in 40 CFR
60.103 and the monitoring requirements are the same as the NSPS requirements in 40
CFR 60.105(a)(2). For a Glaus SRU over 20 long tons per day subject to the NSPS for
SOX, the HAP organic (sulfur) limits are the same as the NSPS emission limits in 40
CFR 60.104(a)(2) and the monitoring requirements are the same as the NSPS
requirements in 40 CFR 60.105(a)(5).

      No performance test or performance evaluation for COMS or GEMS are
required to demonstrate initial compliance with MACT rule for units currently subject
to the NSPS. However, the owner or operator must certify in the notification of
compliance status report that each unit subject to the NSPS is in compliance with the
applicable emission limit and monitoring requirements in this MACT standard.  The
EPA or State permitting authority may request a test to verify compliance with the
NSPS if there is any question about the certification.

      The NSPS does not require that data from COMS or GEMS be used to
determine compliance. Under NSPS, a source test typically is required for this
purpose.  Under NESHAP, however, data from the COMS and GEMS are used to

                                    8-1

-------
determine compliance. For this reason, the COM3 and GEMS must be operated to
meet the applicable performance specifications in appendix B to 40 CFR Part 60 and
the quality assurance requirements in appendix F to 40 CFR Part 60. Appendix F
provides detailed procedures for conducting daily calibration drift checks and
quarterly relative accuracy audits for GEMS. These may be a new requirement for
some plants. Other requirements for continuous monitoring systems in  63.8 of the
NESHAP General Provisions plants are:

4     Conduct daily calibration drift checks (low-level and high-level) and adjust the
      low-level and high-level drifts whenever the 24-hour low-level drift exceeds
      two times the limit of the applicable performance  specification. Clean all
      optical and instrumental surfaces exposed to effluent gases before making
      adjustments and whenever the cumulative automatic zero compensation (if
      applicable) exceeds 4 percent opacity

+     If system is out of control, take corrective action and repeat all necessary tests
      until system meets all applicable performance specifications

+     Keep necessary parts for routine repairs readily available and immediately
      repair or replace parts to correct routine or otherwise predictable malfunctions
      as defined in the startup, shutdown, and malfunction plan

+     Develop and implement quality control program including written protocol that
      describes procedures for calibrations, calibration drift determinations and
      adjustments,  preventative maintenance, data recording/calculations/reporting,
      accuracy audit procedures, and corrective action program

      Several commenters asked how excess emissions and by-passes from NSPS
units would be treated. It is true that under section 112 standards, excess emissions
determined not to be result of startup, shutdown, or malfunctions may be found to be
violations subject to financial  penalties. Under the final  rule, the EPA has revised the
format and eliminated use of the terms "exceedances", "excursions", "excess
emissions", and "violations". Under the new format, we  have established operational
standards for continuous compliance with emission limits and parameter monitoring.
For example, the owner or operator must establish a minimum operating temperature
for any thermal vapor incinerators used as control devices and must monitor the
combustion temperature continuously. Similar performance related parameter values
must be established (and monitored) for other control device types and/or processes as
specified in the rule. The rule establishes the emission and operating limits but the
rule does not specify situations or define the conditions where a "deviation" from the
operating standard has occurred.  Going outside the established range for the
operating parameter or exceeding the emission limit under any circumstance is a
deviation, which must be reported to the permitting authority. In addition, the final
rule does not establish specific situations where "deviations" from the emission limits
or operating requirements are  considered "violations" of  the rule.  In establishing the
operating limits, we fully considered the appropriate NSPS definitions of conditions

                                      8-2

-------
that constitute excess emissions or violation that take into account process and
control device fluctuations over time.

      We also revised the rule to make the reporting requirements the same for these
units to eliminate duplication and added a provision allowing the State permitting
authority to consolidate reports to reduce any other duplication. In other words, the
same report may be submitted for NSPS and NESHAP reporting purposes, but
deviations from the operating standards (formerly referred to as excess emissions)
under section 112 determined not to be the result of startup, shutdown, or
malfunction events may be assessed as violations that are subject to financial
penalties.  Emissions from any by-pass for an affected unit (including a unit subject to
the NSPS) would be identified in the periodic report required by this rule.

      Some commenters (IV-D-30, IV-D-31, IV-D-56) believe if NSPS units are
brought under the MACT standard, then the flexibility provided under the MACT
rules applies equally to the NSPS units which results in dilution of the NSPS
requirements. It is true  that we considered allowing NSPS units to select to comply
with either the NSPS or MACT standard early in the rule development process, but
we rejected this  alternative prior to proposal because of the reasons the commenters
suggested. An NSPS unit must comply with the specified requirements  (summarized
above) and is not afforded the flexibility for non-NSPS units.

      Commenters IV-D-25 and IV-D-44 believe that the proposed rule does or should
exempt FCCU or Claus SRU already subject to and in compliance with the NSPS.
These units are affected sources under the MACT rule. If we exempted NSPS units
from the MACT  rule, we would not know if they are meeting NESHAP requirements.
These units are not exempted from MACT requirements in the final rule.

      Commenters IV-D-39, IV-D-43, and IV-D-44 believes that non-NSPS units
should have the option to comply with NSPS requirements in lieu of the MACT
standard requirements so that units in a plant will be subject to the same
requirements. We agree with the commenters1 suggestions and have added this as an
option under the final rule. An affected source that is not subject to the NSPS may
choose to comply with the specified NSPS requirements.

      Commenters IV-D-47, IV-D-48, and IV-D-53 request that an affected source be
able to opt to demonstrate compliance with an NSPS standard to which it becomes
subject after the compliance date in lieu of demonstrating compliance with both
standards. We can not revise the NSPS requirements in this rulemaking. However, it
is likely that a relatively recent performance test conducted to demonstrate
compliance with this rule would provide evidence of compliance with the NSPS.

      8.2   Definition  of Affected Facility vs Definition of Affected Source

      Comment: Commenter IV-D-25 urges EPA to make Subpart UUU more clear
than the NSPS, particularly in the definitions of affected facility and existing facility.

                                     8-3

-------
The rule should include provisions defining whether an existing source can be
reconstructed by the addition of a second unit or whether only the added unit
becomes a new source.  He explains that a refinery can install a new sulfur recovery
plant to process H2S and TRS from the sources controlled by an existing sulfur
recovery plant. The NSPS fails to clarify if the addition of the second plant
constitutes the addition of a new unit or the modification of an existing unit.  If
Subpart UUU includes separate standards for new/reconstructed sources, a provision
clarifying this situation needs to be added.

      Response: The affected sources under Subpart UUU are each existing, newly
constructed, or reconstructed FCCU, CRU, and SRU. The only case under the MACT
standards where standards differ for existing versus new or reconstructed sources is
for inorganic HAP (i.e., HC1) from CRU. Under the example given for sulfur recovery
plants, a newly constructed Glaus sulfur recovery plant would be subject to the MACT
standard for (sulfur) organic HAP and the NSPS  for sulfur oxides. The NSPS
standards are for the control of criteria pollutants rather than HAP.  Under NESHAP,
the addition of a second unit triggers the new source standard for the unit because
reconstruction entails the replacement of existing components. In the sulfur plant
example, the new sulfur plant (whether Glaus or other type) would be subject to the
MACT limit for organic HAP and the associated  monitoring requirements. These
standards are the same for new and existing affected sources. We can not modify the
NSPS under this rulemaking to make the clarification requested by the commenter.
We can, however, clarify the role of a reconstructed affected unit under NESHAP.
Under the NESHAP General Provisions "reconstruction" means "the replacement of
components of an affected or a previously unaffected stationary source to such an
extent that:

      The fixed capital cost of the new components exceeds 50 percent of the fixed
      capital cost that would be required to construct a comparable new source; and
      It is technologically and economically feasible for the reconstructed source to
      meet the relevant standard(s) established by the Administrator (or a State)
      pursuant to section 112 of the Act.  Upon  reconstruction, an affected source, or
      a stationary source that becomes an affected source, is subject to relevant
      standards for new sources, including compliance dates, irrespective of any
      change in emissions of hazardous air pollutants from that source." It is clear
      under this definition that a reconstruction is not the addition of a new unit;
      addition of a new unit is new construction subject to the new source review
      process.

      8.3    Triggering the NSPS due to Emissions from Flares and Combustion
            Devices

      Comment: Commenters IV-D-47, IV-D-48, IV-D-49, TV-D-5 3, IV-D-54, and
IV-D-59 request the rule state that emissions from flares and combustion devices do
not trigger NSPS requirements. Commenter IV-D-49 does not agree that compliance
with MACT II triggers the applicability of the NSPS based on the definition of

                                      8-4

-------
"modification" under 40 CFR 60.14. Commenters IV-D-48 and IV-D-54 support the
exclusion of streams routed to fuel gas systems and urge EPA to clarify that any
stream routed to a fuel gas system does not trigger the NSPS. Commenters IV-D-47
and IV-D-53 believes the rule should be clarified to specifically exclude triggering the
NSPS by venting TOG emissions to a flare.

      Response: One of the options for the control of TOG emissions from CRU
allows venting emissions from the regenerator to a combustion device or a flare that
meets the requirements for control devices in 63.11 of the NESHAP General
Provisions. The commenter is correct in that the sulfur oxide standards in 40 CFR
60.104 of the NSPS (Subpart J) include a limit for H2S emissions from fuel gas
combustion. GEMS for SO2 or H2S are required, as well as associated recordkeeping
and reporting requirements.  However, it should be pointed out that it is currently
common industry practice to vent the CRU emissions to the fuel gas system or flare.
As such, the number of units that would significantly increase the amount of fuel gas
combusted as a result of this rule is considered to be small.  Also, the sulfur content of
the vented gases is very low in sulfur  concentration; sulfur deactivates the CRU
catalyst and is typically removed from the feed stream prior to the CRU. The EPA is
reviewing the issue of triggering NSPS requirements as a result of MACT compliance
efforts and is considering different solutions which may including revising the NSPS
monitoring requirements or issuing a  policy directive. It must also be pointed out that
applicability to NSPS is a separate determination that is made on a case by  case basis
and no blanket exclusions are included in the final refinery MACT standards for
venting CRU emissions to flares.

      8.4   State or Local Requirements

      Comment: Commenter IV-D-40 explains that state  and local requirements and
permit conditions often provide equivalent emission limits that are as stringent or
more stringent than the proposed MACT standard. EPA should modify the
applicability language to allow a units subject to State/local regulations or permit
conditions that are equivalent to or more stringent than the MACT standard to be
considered in compliance with the MACT standard and all associated requirements,
in the same way provided to NSPS units.

      Response: We appreciate the general idea behind the comment (i.e., avoiding
duplicative or regulatory overlap and  unproductive regulatory requirements);
however, determining whether rules are equivalent (or more stringent) is not
necessarily that easy. Factors such as test methods, averaging times, format/units are
only some of the items to consider in making such a determination. The EPA held
several meetings with State regulators in an attempt to identify State requirements
that were equivalent to or more stringent than the proposed MACT standard. No such
State requirements were identified. Considerable resources would be required to
make  such determinations on a generic basis on all the regulations covering the units
affected by this rule.  Equivalency is a case-by-case determination made by the
applicable permitting authority.

                                     8-5

-------
9.0   IMPACT ANALYSES

      9.1   Ni Emission Estimates for CCU Regenerator Vents

      Comment: Industry commenters (IV-F-3.1, IV-F-3.2, IV-F-3.3, IV-D-39, and
IV-D-40, IV-D-44, IV-D-45, IV-D-47, IV-D-48, 24, IV-D-53) believe EPA has
overestimated baseline Ni emissions from CCU. According to the commenters,
baseline emissions from 120 CCU are overestimated by an order of magnitude based
on the selective use of the average of only two data points.  EPA then uses this
controlled emission factor to compute the uncontrolled Ni emission factor by dividing
it by 95%- the assumed level of control. This ignores the industry-supplied estimates
of between 9.6 and 33.5 tpy of Ni based on the mean value of the entire database
compared to EPA's estimate of 124 tpy with total baseline HAP of 81 tpy compared to
EPA's estimate of 380 tpy. In support, they cite industry-reported data in the 1996  TRI
(about 33 tpy of Ni) and EPA's Report to Congress on HAP emissions from the electric
utility industry.  This report includes estimates of Ni emissions from the generation of
electricity using coal, oil, and natural gas. These HAP emissions are at least a factor of
100 below baseline emission estimates. Other commenters point to their low
site-specific Ni emission rates.  Commenter IV-F-3.3 submits monthly Ni emission
data in support.  Others believe control of these emissions would be arbitrary and
capricious because they are lower than the Ni emissions from the utility industry
where EPA determined not to regulate based on the low level of emissions and low
risk. Overall, the commenters contend that emissions are de minimus and do not
warrant control.

      Response:  In its initial estimates, the EPA used a mid-range emission factor
approach because industry representatives argued that the refineries for which metal
HAP emissions  data were available were predominantly located in California which
has strict fuel standards, had low Ni feed concentrations, and were all well controlled
for PM emissions. Therefore, using a direct average or median value of the data that
was considered by industry to be biased and not representative of the industry
nationwide was inappropriate.

      After much corroboration with industry representatives, more detailed data on
CCU operations were provided to the Agency. Based on this newly provided
information and data, the EPA subsequently has revised the refinery vent impact
estimates and emission estimation methodology based on information and data
received since proposal. Current estimates of HAP emissions are in general agreement
with those made by most petroleum industry representatives. With regard to the
comment that refinery Ni emissions are lower than the Ni emissions from the utility
industry and that EPA determined not to regulate utilities based on the low level of
emissions and low risk, it must be pointed out that this statement does not accurately
represent EPA policy relative to Ni emissions from the utilities industry.  To date, no
formal regulatory determination has been made regarding regulation of any HAP
emitted from the utility industry. The EPA is in the process of collecting additional

                                      9-1

-------
data on metal HAP emissions and will make a determination in the future.
Additionally, Ni emission factors for utility boilers burning liquid fuels are not
applicable to CCU regenerators as the emission mechanism for PM and metal HAP are
different. Use of these utility boiler emission factors is not appropriate for estimating
CCU regenerator vent emissions.

      9.2   Emission Estimate Methodology

      Comment: Commenters IV-D-30 and IV-D-56 disagree with the emission
estimate methodology- specifically, the use of Ibs per million bbl as the unit for
measuring metal HAP emission. This methodology used by EPA does not take into
account which one of the three major variables is driving the estimates and can not
accurately project emissions in terms of the changing quality of feed to the CCU.  The
commenter recommends generating emission factors that are unit and feed specific.
Using participate control equipment data available from the API database, EPA may
assign appropriate particulate emission values (e.g., 1 Ib per Kg coke for an ESP with
specific collection area greater than 350 square feet per 1,000 acfm) for each type of
particulate control. EPA then should calculate the total metal HAP content on the
E-Cat using the equation supplied by the commenter. Feed quality data (concarbon
and API gravity) is available from the APR database or can be assumed. The final
variable,  (1,000 pounds of coke per million barrels) is a function of the heaviness of
the feed.  EPA should estimate a relationship between coke generation and concarbon
(adjusting for other variables such as the mode of unit operation and the type of
catalyst used) or choose values for coke generation that are consistent with the
concarbon number of the feed (using a linear proportional relationship).  This
approach allows EPA to evaluate the impact on changes in feed quality on metal HAP
emissions at the unit level and industry-wide.

      Response: We have developed a more site-specific approach to the impact
estimates. However, this analysis does not include a factor for feed quality per se.
Consideration of site-specific E-Cat Ni concentration, the most recent approach, is an
alternative means of accounting for both feed quality and operational differences on a
site-specific basis.

      The API database, as provided to EPA, does not contain site specific data on
CCU feed quality (e.g., Ni content, concarbon, or API gravity) or control device
specific data (such as unit-specific plate area for ESP).  These data are considered to
be confidential business information by many refiners; and the industry/trade
association work group that collected the plant data were unable to share this type of
information and data with the EPA in a collective manor.

      Based on recent data obtained by EPA regarding Ni E-Cat concentrations, more
site specific and unit specific emission estimates have been calculated. These E-Cat
based estimates provide an accurate account of current baseline emissions and
projected emission reductions.  Data regarding trends in Ni content of FCCU feeds is
limited and inconclusive. Analysis of feed quality changes is at this time highly

                                      9-2

-------
speculative given the uncertainty associated with the industry's operational response
to the Tier 2 fuel standards.  Depending on industry's Tier 2 compliance approach,
CCU feed metal HAP content could see a downward trend in spite of increases in
crude metal HAP content. To the extent possible, we are coordinating compliance
efforts between this rule and the Tier 2 fuel standards.

      9.3   Selection of Pollutants in Database

      Comment: Commenters IV-D-41 and IV-D-56 disagree with EPA's approach of
excluding compounds reported only by one facility that was not verified by any other
information. The commenter feels that this procedure may have eliminated data
where it was most needed and asks if this excluded data on D/F, cyanide, and Hg
where tests have been conducted at only a few units? Also, since emission tests for
HAP in the various CRU regeneration cycles are scarce, this procedure may have
eliminated any of this data for consideration. Dropping data because other units did
not conduct the test is not defensible.  If a pollutant is detected in a test, then it
should be accepted in the analysis unless there are valid and documented reasons to
exclude it. EPA should review its procedures in this regard.

      Response: First, the technical information document contains emission factors
for D/F, cyanide, and Hg; these data were not excluded from the analysis. To clarify,
the methodology used in developing emission factors for the units of concern involved
excluding data only in the case where there was a single test that showed the presence
of the compound and there were multiple test from the same and other facilities that
reported non-detect for the same compound. This methodology regarding treatment
of non-detect values was  clearly explained in the BID for the proposed rule. The only
compounds for which emission factors were not developed and reported based on this
criteria are CS2 and COS from CCU.  In the case of POM, these data were treated as a
class of HAP and all POM data were considered in developing the reported emission
factor even when one specific POM compound was measured at only on facility in a
single test.

      9.4   Impacts of Additional Pollutants

      Comment: Commenters rV-D-30, IV-D-26, W-D-31, and IV-D-56 ask EPA to add
the impacts of decreases in SOX emissions that would occur with wet scrubbers. In
some cases, SO3 may be considered a HAP. They also request impact analyses of the
effect of the standard on D/F and Hg emissions.

      Response: We agree with the commenter that SOX emissions are expected to be
reduced when a wet scrubber is used to control CCU PM emissions. Concurrent SOX
emissions reductions have been estimated and are reported in chapter 6 of the BID for
the proposed standards..  However, SOX is not a listed HAP nor is it "considered" a
HAP for regulatory purposes. We have not included an analysis of the impact of the
standard  on D/F and Hg emissions; an impacts analysis was not done because there
                                     9-3

-------
are no data that indicate that there are appreciable D/F emissions from FCCU or data
regarding Hg emissions removal efficiency for FCCU scrubbers.

      9.5   Cost Estimates for CCU Catalyst Regenerator Vents

      Comment: Commenters IV-F-3.1, IV-F-3.2, IV-D-39, IV-D-40, IV-D-47,
IV-D-48, IV-D-49, IV-D-54, and IV-D-59 believe that the capital costs of controls (ESP
or scrubber) ranges from about $5 million to $20 million each for each CCU,
depending on the unit size and site configuration compared to EPA's estimate of $31
million for the entire industry. According to Commenter IV-D-49, the parameters
used in the OAQPS model to estimate the cost of inorganic controls are not
representative and are underestimated because EPA did not use site preparation costs
(which are significant),  the exhaust gas temperature is too  low, and the selected
control efficiency is not consistent with the required removal efficiency of the unit.
EPA has indicated that they would modify the input parameters using a retrofit cost
factor of 1.35, an exhaust gas temperature of 500 degrees,  a mean particle diameter
based on Region V source test data, and an ESP selected control efficiency of 500 ft
sq/m-acfm.

      Response: The EPA has modified the control cost input parameters using a
retrofit cost factor of 1.35, an exhaust gas temperature of 500 degrees, a mean particle
diameter based on Region V  source test data, and an ESP selected control efficiency of
95%.  With the revised inputs, the resulting ESP  design (used in the costing) has a
specific collection area (SCA) of 500 ftsq/m-acfm, which is consistent with the SCA's
for ESP that meet the NSPS PM emission limit as confirmed by industry supplied
data.

      9.6   Cost Effectiveness Estimates for CCU Catalyst Regenerator Vents

      Comment: Commenters IV-F-3.1, IV-F-3.2, IV-F-3.3, IV-D-39, IV-D-40,
IV-D-44, IV-D-45, IV-D-47, IV-D-48, IV-D-49, IV-D-53, IV-D-54, and IV-D-59 do not
believe that controls for CCU are cost effective. Based on the overestimated emissions
and underestimated costs, most commenters believe that the cost effectiveness of the
controls for CCU would exceed $1 million dollars per ton of HAP reduced using
industry estimates of 38 tpy of total inorganic HAP reduced with annual costs of $48
million or about $200,000 per ton of HAP reduced if controls for organic HAPs are
included. Some believe the cost effectiveness for their facility is even higher.
According to the commenters,  the high cost effectiveness is beyond Congressional
intent and not consistent with  Administration policy.  In support,  they cite the
Presidential Memorandum of July 16, 1997 to Administrator Browner on
implementation of NAAQS which establishes control costs to under $10,000 per ton.
According to the commenter, EPA's cost effectiveness (about $97,000 per ton
reduced) does not compare well to the policy set by this Administration or to the cost
effectiveness for other MACT rulemakings for the petroleum refinery industry (i.e.,
the HON and Refinery MACT I), which ranges from several hundred thousand dollars
                                     9-4

-------
to $10,000 per ton reduced. EPA should use its latitude under the CAA to develop a
regulation that is not unnecessarily restrictive or costly.

      On the other hand, Commenters IV-D-30, IV-D-31, and IV-D-56 disagree with
the cost effectiveness as a decision criteria, particularly when estimates are based on
a single pollutant because they do not include the substantial reductions in other
pollutants or the associated benefits. These ancillary reductions are an important
component of aggregate risk reduction and monetary benefits.  When these reductions
are considered, the cost effectiveness is very low and the rule is beneficial.

      Response: As noted in the preceding response, we have revised the costs and
emission impacts of the proposed MACT rule based on the information and data
received during the comment period.  The cost-effectiveness numbers especially for
the inorganic (i.e., metal) HAP are high; however, the cost-effectiveness for the entire
rule is moderate at about $4,200/ton of pollutants controlled, as is the combined cost-
effectiveness for both the refinery MACT standards atabout $2,000/ton. These values
are comparable to the cost-effectiveness of other rules for the number of sources
impacted. And although high, the costs are not considered unreasonable.

      With regard to the comments concerning Congressional intent, it must be
pointed out that under the CAA Amendments of 1990 passed by Congress the MACT
standards are technology based standards and are not to be based on
cost-effectiveness. The cost effectiveness values cited by the commenter are
applicable to criteria pollutants and not HAP; if one looks at the cost effectiveness of
the final rule considering the concurrent reductions in criteria pollutants, then the
values are much more reasonable and, as noted above, are in the general range of
values mentioned.

      We agree that the concurrent and ancillary reductions associated with the
control of HAP by the proposed MACT standards are an important component of
aggregate risk reduction and monetary benefits and these reductions should be
considered in assessing the merits of the rule.

      9.7   Low Health Risk Does Not Warrant Proposed Controls

      Comment: Commenters IV-F-3.1, IV-F-3.2, IV-D-39, and IV-D-40, IV-D-44,
IV-D-45, IV-D-47, IV-D-48, IV-D-49, IV-D-53, IV-D-54, and IV-D-59 cite an industry
screening study of 22 CCU that shows low health risk (less than one cancer case in
one million) due to Ni emissions. According to these Commenters, the low emissions
and health risk do not warrant control. Commenters FV-D-30, IV-D-31, and IV-D-33
believe that the emissions and health risk do warrant regulation due to the HAP and
the high volume of non-HAP emitted.  Commenter FV-D-33  describes the lung cancer
rates in communities downwind of the 14 refineries in her area which are 100 percent
above expected incidence. According to the commenter, the refineries are a leading
source of the PM that contributes to these premature deaths. Commenter IV-D-56
speaks to the numerous SO2 and H2S exceedances from particular refineries, the

                                     9-5

-------
numerous carcinogenic, teratogenic, and mutagenic substances emitted, and the
overall concerns in communities exposed to clusters of refineries regarding emissions
of persistent bioaccumulative toxic substances and the effect on food chains such as
fruit and vegetable gardens.

      Response: We have discussed with the industry their screening study of 22
CCU that shows low health risk (less than one cancer case in one million) due to Ni
emissions. Preliminary discussion of the industry risk analysis indicated that the
facilities examined are considered high emitting sources with high nickel feed;
however, upon further inquiry, it appears that all the facilities included in  the initial
risk assessment study submitted as part of the API comments  currently use either an
ESP or wet scrubber for control of PM which is the MACT floor technology for CCU.
As a result it is difficult to draw any industry wide conclusions on the risk posed by
uncontrolled CCU.  The final report will be reviewed in detail and will be considered
in the risk standard development phase of the NESHAP standard setting process
under section 112.

      Again, it also must be emphasized that the MACT standards are not  risk based
but are technology based standards. Risk and cost-effectiveness arguments are
appropriate for control options beyond the floor or for the inclusion of area sources
neither of which is the case with this source category rulemaking. Risk will be
evaluated and considered in the second phase of the NESHAP standard setting
process.

      9.8   Database Weaknesses

      Comment:  Commenters IV-D-41, IV-D-31, and IV-D-56 believe the data set on
emissions tests was not adequate. Only a few of the nearly 70 pollutants were tested
and tests were  made on too few units. Test data on CCU came from a set of 8
facilities; no data were available for emission rates at other units and for other
pollutants. This is why emission factors had to be extrapolated to other units.
Information on particulate emission rates from CCU (the basis for projecting model
unit and nationwide emission estimates) also are  absent. Data on other variables also
were incomplete (e.g., feed quality data and ESP plate areas).  Feed quality  data is
crucial to understanding the relationship between feed quality and HAP metal
emissions.  The database for CRU was even more limited and pertained only to the
coke burn cycle even though there are many different cycles (chloriding cycle,
sulfiding step, and purge cycle).  EPA should add this information to the BID for the
final standards. Some of the emission factor information and additional data are
available in the API report. "Characterization of Hazardous Air Pollutant Emissions
from FCCU, CRU,  and SRU Refinery Process Vents, Final Report, Volumes  I-ffl-
Process and HAP Emissions Survey."

      Response: EPA is aware of the lack of HAP data and test data; however, the
available data in the EPA refinery vent database is adequate to support the  analyses
conducted as the basis for the proposed standards. We have requested the

                                     9-6

-------
document/data referenced by the commenter (the report has been added to the project
docket) but data referenced by the commenter is not in the version of report as
released by the industry. As discussed in comment/response 9.2, much of the process
operating data cited by the commenters is considered by some refiners to be
confidential business information which made the transfer of data from the industry
work group a much more complicated undertaking.

      9.9   CRU Catalyst Regenerator Vent Emission Rates

      Comment: Commenters IV-D-41 and IV-D-56 state that EPA should review the
emission estimates and the suitability of the controls for criteria pollutants for CRU
catalyst regenerator vents.  According to the commenter, emissions of particulates and
SOX from a large continuous CRU are comparable to those of a CCU.

      Response: The commenter did not provide any data to support the comment
that particulates and SOX emissions from a large continuous CRU are comparable to
those of a CCU. The data available in the EPA database do not support this
contention and a comparison of vent flow rates for continuous CRU regenerators and
CCU regenerators makes this assertion very dubious. As mentioned  previously, EPA
does not have the authority to regulate criteria pollutants under section 112 of the
CAA.

      9.10  CRU Emissions Table

      Comment: Commenters IV-D-41 and IV-D-56 ask EPA to expand the table of
emissions from CRU catalyst regenerator vents in the BID for the  proposed standards.
Since the variety and magnitude of CRU catalyst regenerator vent emissions varies in
the different cycles of catalyst regeneration, EPA should specify the cycle to which the
data pertain and ensure that emission factors for all cycles are presented.

      Response: We have made  some basic differentiation of the pollutant types
emitted during the various CRU regeneration cycles in the BID for the proposed rule.
This effort was limited however by the available data. We did not under take the
extensive effort needed to gather additional data on HAP emissions for the CRU
regeneration cycles other than the coke-burn cycle, in large part because it was
obvious that any additional data collected on emissions during the other cycles would
not change the regulatory outcome or level of control required. This  is because all
units are required to utilize controls under the proposed rule during these CRU
catalyst regeneration cycles. It was not necessary to have additional constituent
specific emission factors for the other CRU catalyst regeneration cycles to  establish
the MACT floor or estimate the emission reductions because all units for which
site-specific data were gathered during the standard development process  were
already using the controls required by the proposed rule. And, we have not received
any information that any other CRU are doing otherwise. Gathering any additional
information and data on  this particular point would not be a prudent or cost effective
use of the Agency's limited resources.

                                     9-7

-------
      9.11  Effect of Hydrotreating on Emission Estimates

      Comment: Commenters IV-D-41 and IV-D-56 disagree with EPA's conclusion
in the BID for the proposed standards that "data were inconclusive as to whether
hydrotreating had any effect on the emission factors for HAP metals" and the
subsequent decision to use the same emission factors for units that hydrotreat as for
those that do not hydrotreat.  They argue that a CCU processing hydrotreated feed
should emit less metal HAP (and less SOX and NOJ per unit throughput than a unit
processing non-hydrotreated feed, although both may emit the same amount of Ni per
1,000 pounds of coke bum. This is because the relatively lower Ni content of the
hydrotreated feed may be offset by other factors such as relatively higher particulate
emissions and higher coke generation rates compared to a unit processing
non-hydrotreated feed. Different hydrotreatment units process feeds of different
heaviness and the degree of hydrotreatment may vary among units depending on
factors such as the mode of operation of the unit and the availability of
metal-poisoning resistant catalyst.  Rather than partition emission factors according to
whether hydrotreating is conducted, EPA should use empirical and material balance
equations to link the metal HAP emission factor to the quality of feed and other
factors. If the Ni content of the feed is available, then Ni and total metal HAP on the
E-Cat may be computed.  With the metal content of E-Cat and particulate emission
rate, the metal HAP emission rate can be computed as their product. The key to
computing metal HAP emissions is information on the quality of feed to the unit.

      Response: We agree with the commenter that hydrotreating should result in a
metal HAP reduction in the CCU feed. We also agree that some additional attributes
of hydrotreating the CCU feed stream may tend to mask or minimize the anticipated
metal HAP emission reduction (see discussion in comment/response 1.19).  The EPA
conducted additional information gathering on hydrotreating following proposal of
the rule.  Although  refinery representatives confirm that the catalyst used in
hydrotreatment should effectively adsorb metal HAP and thus reduce the metal HAP
content in the CCU feed,  there was no general agreement that the metal HAP content
of the CCU E-Cat would be reduced (as an industry trend) since this depends on
economic as well as process considerations specific to each individual CCU.

      As noted previously, the API database, as provided to EPA, does not  contain
site specific data on CCU feed quality (e.g., Ni content, noncarbon, or API gravity).
Based on recent data obtained by EPA regarding Ni E-Cat concentrations, more site
specific and unit specific emission estimates have been calculated.  These E-Cat based
estimates provide an accurate account of current baseline emissions and projected
emission reductions. Data regarding FCCU feeds quality changes are among those
operating parameters that certain refiners consider to be confidential and are
unavailable to EPA at this time.

      Using the data available to the Administrator, the EPA could not quantitatively
demonstrate that hydrotreating of CCU feed significantly reduced metal HAP
emissions from the  CCU regenerator vent. Instead, the revised emission approach

                                     9-8

-------
employs E-Cat Ni concentrations which are more directly linked to Ni emissions (as a
surrogate for total metal HAP emissions).

      In addition, analysis of CCU feed quality at this time is highly speculative given
the uncertainly associated with the industry's operational response to the Tier 2 fuel
standards.  Depending on industry's Tier 2 compliance approach, FCCU feed's metal
HAP content could see a downward trend in spite of increases in crude metal HAP
content.  To the extent possible, we are coordinating compliance efforts between this
rule and the Tier 2 fuel standards.

      9.12  Effect of ESP Collection Area on Cost Estimates

      Comment:  Commenters IV-D-41 and IV-D-56 urge EPA to reestimate the costs
for ESPs needed to comply with the standards based on a lower specific collection
area.  The EPA used a specific collection area of 717 ft2/!,000 acfm, which seems high
especially given the 90% efficiency assumption. EPA should use an efficiency of 98%
or higher for effective control of the finer catalyst particles. Industry data indicates
that existing units have ESP with specific  collection areas ranging from under 100 to
about 550 ft2/l,000 acfm; data provided by API indicate that the PM NSPS is
achievable  by ESP with specific collection area of about 350 ft2/!,000 acfm. Cost
estimates with an ESP double the required size overestimates the control costs.

      Response: Upon further review, it was found that the primary reason for the
low removal efficiency while having a high specific collection area was the use of the
mean particle size distribution in the design rather than the mean mass particle
distribution. The ESP cost estimates were revised to correct this error; see
comment/response 9.5.

      9.13   Effect of Scrubber Pressure Drop on Costs

      Comment:  Commenters IV-D-41 and IV-D-56 believe a pressure drop  of 10
inches of water is too low for effective control of particulates and SOX. The Agency
should also add the cost of scrubbing the gas flow rate with caustic as an option to
control SOX.

      Response: The design pressure drop is dependent on the mean particle size
diameter of the PM emissions.  Available data regarding particle size distribution for
CCU emissions suggests that the particle diameter used in our calculations is
reasonable  (i.e., 4 ^m).  The estimated costs for Venturi scrubbers used in the impacts
analysis agrees well with design cost estimates provided by a wet scrubber vendor.
Alternatively, using 1 |im, the design pressure drop is 25 inches of water, which
increases the estimated operating costs by approximately 15 percent.

      We did not include the cost of scrubbing with caustic as an SO2 control option
because the rule is intended to control HAP and any additional cost of controlling any
criteria pollutants is typically not included in the impacts analysis.

                                      9-9

-------
      9.14   Effect of CO Monitoring on Compliance Costs for Full Burn CCU

      Comment: Commenters IV-D-51 and IV-D-54 believe the compliance costs for
complete combustion units are potentially very large and have not been considered in
the proposed rule.  Some potential compliance methods include:  (1) returning unit to
partial combustion regeneration and recommission existing CO boiler or build new
one if needed, (2) retain complete combustion regeneration but reroute flue gas to a
boiler or process heater (existing or new), (3) raise normal stack O2 by reducing unit
feed rate, catalyst circulation rate, and conversion, and raising regen temperature, (4)
raise normal stack O2 by increasing regen blower rate, or (5) raise  normal stack O2 by
enriching regen blower air with 02. Commenter IV-D-53 believes  up to 40% of
existing non-NSPS  complete combustion units would need to consider one of these
modifications to comply, with an overall industry wide cost of $130 to 260 million per
year.

      Response: We agree that some limited number of existing nonNSPS complete
combustion units may need to consider one of these modifications to comply with the
proposed rule. However, neither the commenter nor the industry has  provided any
relevant data to fully document the issues or problems associated with complete burn
units meeting the 1-hour limit or to substantiate the actual number of CCU that
cannot currently comply with the CO 1-hour standard. We do not agree on the likely
number of affected  units or the cost to the industry that are reported in the comment
because the most costly approach was used to make the cost estimates.  Based on the
limited data provided, one facility was projected  in our revised cost estimates to
install and use an auxiliary fan to increase the regenerator blower rate by up to 20%.

      9.15  Selection of Control Equipment for Costing

      Comment: Commenters IV-D-41 and IV-D-56 state that EPA should explain in
the BID for the final rule how the control equipment were selected for costing. For
example, did EPA use a cost minimization algorithm to decide between different ESP,
scrubbers, and other potential control? Also,  did EPA cost a combined control device
(combining flue gases from multiple CCU or CRU)?

      Response: Based on information and data in the EPA/API database which
includes a number of sources, ESP and wet scrubbers were used almost exclusively
for FCCU control devices. We developed control costs for both ESP and wet scrubbers
but the estimates indicated that ESP were always less costly on the basis of PM
control, although wet scrubbers are potentially cost competitive if SOX removal is also
required. We did not cost combined flue gas control of multiple CCU.  Application of
combined stack control are rare in this industry and lead to significant potential
challenges in control device maintenance.
                                    9-10

-------
      9.16  Energy Impacts for Incinerators

      Comment: Commenter IV-D-35 asks EPA to revise the estimates for annual
natural gas requirements for incinerators.  The commenter explains that Beavon
Stretford sulfur plants normally do not require incineration of the tail gas due to the
low H2S concentration except during plant upsets. In comparison, typical tail gas
plants with amine scrubbing plants (i.e., SCOT) require incineration under California
rules if the tail gas H2S concentration exceeds 10 ppm. Therefore, the requirement for
1.5 billion cubic feet of natural gas annual (pg. 48905 of preamble) for incineration
applies to the amine scrubbing plants and not to Beavon-Stretford plants.

      Response:  According to the EPA database, there are Beavon-Stretford sulfur
plants that employ an incinerator.  Therefore, these incineration costs were included
for the uncontrolled units to allow for a conservative cost projection.  The incineration
requirement only applies to SRU that have reduced sulfur compound (i.e., COS and
CS2) emissions greater than 500 ppm regardless of the type of tail gas treatment unit
employed.  It is reasonable to conclude that some of the uncontrolled Beavon Stretford
sulfur plant unit will not require incineration but the exact number of these units is
not known.

      9.17  Additional Environmental Impacts from Bioaccumulative HAP

      Comment:  Commenter IV-D-56 states that the bioaccumulation potential of
reduced sulfur HAP and other air toxics from refineries on fruit trees  and vegetable
gardens in communities surrounding refineries must be more carefully reviewed by
EPA.  This commenter describes concerns in affected communities about the health
and general pollution impacts from persistent bioaccumulative toxic substances.
According to the commenter, refineries, and clusters of refineries in particular, emit
numerous carcinogenic, teratogenic and mutagenic substances which need to be
comprehensively addressed in the standard.

      Response:  The MACT standards under section 112 are technology based
standards.  The second phase of the NESHAP standard setting process involves
examining the risk associated with the HAP source categories.  The health and general
pollution impacts from persistent bioaccumulative toxic substances released from
refineries, and clusters of refineries in particular, will be comprehensively addressed
in that standard setting effort.

      9.18  Economic Analysis

      Comment:  Commenter IV-D-30 believes the economic impact analysis should
include the benefits  of non-HAP pollutant reductions.
                                     9-11

-------
      Response: This rule does not require a benefits analysis for an official
regulatory impacts analysis however we have estimated the non-HAP pollutant
reductions and have considered the benefits of non-HAP pollutant reductions
indirectly in establishing the level of the standard and other associated regulatory
requirements.
                                     9-12

-------
10.0  ADMINISTRATIVE REQUIREMENTS

      10.1  Executive Order 13045 on Children's Health

      Comment: Commenter IV-D-56 states that this Executive Order applies to any
rule determined to be "economically significant" as defined under Executive Order
12886 and that concerns an environmental health or safety risk that EPA has reason to
believe may have a disproportionate effect on children. Due to the large population of
children living near refineries in Texas and the other parts of the U.S., Commenter
IV-D-56 urges EPA to require more stringent MACT standards in compliance with this
order.

      Response: Executive Order 13045, "Protection of Children from Environmental
Risks and Safety Risks" (April 21, 1997), directs Federal Agencies to include an
evaluation of the health or safety effects of planned regulations on children.  This
Order applies to economically significant rules initiated after April 21,1998, which
concern an environmental risk or safety risk that an Agency has reason to believe may
disproportionately affect children. An "economically significant" rule is  defined by
Executive Order 12866 as any rulemaking that has an annual effect on the economy of
$100 million or more, or would adversely affect the economy, productivity,
competition, jobs, the environment, public health  or safety, or state, local, or tribal
governments or communities in a material way. For rules subject to the  Executive
Order, agencies must explain why the planned regulation is preferable to other
potentially effective and reasonably feasible alternatives considered by the Agency.

      These standards are not covered by the Executive Order because the final
standards are not economically significant and because EPA is precluded from
considering health or safety risks in the development of MACT standards under
section 112(d)  of the CAA. Section 112(d) requires determination of the  minimum
level of stringency (i.e., the MACT floor) to be based solely on the performance of
technology.

      There is no doubt that refineries are major source emitters of HAP and also
release high levels of non-HAP criteria/ambient pollutants. The technology-based
standards developed under section 112(d) for petroleum refineries greatly reduce
these  emissions. Implementation of the first stage of air toxic rules for this industry,
published in 1995 (60 FR 43244) began in 1998. When fully implemented, this rule
will reduce emissions of eleven HAP by 59 percent from current levels and non-HAP
VOC by over 60 percent. Today's final rule, which covers process vents from units not
subject to the first rule, will reduce metal and organic HAP from these units by 87
percent, with a total emission reduction of HAP and ambient pollutants of well over
100,000 tons per year. In addition, it may be reasonable to expect further emission
reductions as the industry increases the use of hydrotreatment in response to the Tier
2 fuel standards.
                                     10-1

-------
      We have not conducted a risk assessment to estimate the health effects of
emissions after the implementation of both sets of standards on either adults or
children.  We expect to begin work on this type of analysis over the next few year.
The results of these studies may lead to additional standards under section 112(f).
The residual risk standards under section 112(f) will be risk-based and will consider
any disproportionate impact on children's health as required by the Executive Order
and EPA policy.

      10.2  Executive Order 12898 on Environmental Justice

      Comment:  Commenter IV-D-56 requests that EPA take into consideration
Executive Order 12898 and Title VI of the 1964 Civil Rights Act and implementing
regulations applicable since most of the communities where refineries are located in
are populated by people of color who are also low-income.  Executive Order 12898
prescribes fundamental requirements for federal agencies to insure that all programs
and agencies are not allowed to increase the disproportionate burden of
environmental hazards in communities of color and low income such as most refinery
communities  in Texas. According to the commenter, refineries are among the dirtiest
industrial operations among the major sources of toxic and hazardous air pollutants
and are ranked first among all large industrial and small business sections for citizen
air pollution complaints in Texas.  Most of these facilities are located in or close to
heavily populated areas and are concentrated in clusters. More than 6 million
citizens (including more than 3 million people of color or over 51 percent) live in the
14 Texas counties where refineries are located. TRI data show that refineries are high
emitters of HAPs that create health problems for those in daycare centers, schools,
and nearby homes.

      Response: Attention to the impact of environmental pollution on particular
segments of our society has been steadily increasing. Concern that minority
populations and/or low-income populations bear a disproportionate amount of
adverse health and environmental effects led President Clinton to issue Executive
Order 12898 (59 FR 7629, February 16, 1994). The EPA's Outreach and Special
Project Staff in the Office of Solid Waste and Emergency Response serves to
coordinate and implement the Agency's principles and new initiatives, including
Environmental Justice. The Agency's Environmental Justice homepage provides a
wide range of information on contacts, publications,  and resources (see
http://es.epa.gov/oeca/main/ej/index.html).

      We understand  many of the concerns cited by this commenter, but can not
resolve them in the context of this rulemaking as the effect of this rule will certainly
be to decrease emissions and any associated disproportionate burden on special
segments of the population.  The proper avenue for investigating this issue lies in the
environmental permitting process.  The EPA's "Interim Guidance for Investigating
Title VI Administrative Complaints Challenging Permits" provides detailed
information on the process for filing complaints under Title VI of the Civil Rights Act
                                     10-2

-------
alleging discriminatory effects resulting from the issuance of pollution control permits
by state and local government agencies that receive EPA funding.

      10.3  Executive Order 12866

      Comment: Commenter IV-D-39 believes the costs of the rule are significant and
will exceed $100 million. Because the rulemaking is significant, analysis under
Executive Order 12866 is required.  Commenters IV-D-30, IV-D-31, and IV-D-56 claim
that the full review procedures of Executive Order 12866 apply to this rulemaking if
EPA adopts the industry-recommended Ni alternative because this option is likely to
"adversely affect in a material way... the environment."

      Response: Executive Order 12866 is based on annual costs and not capital
costs. The annual costs of this rule  are under $60 million and well below the $100
million/yr criterion for a major rule  under the Executive Order.  In response to the
concerns voiced by other commenters, EPA did not adopt the industry-recommended
industry alternative and  cannot agree with the commenters that the final rule
"adversely affects in a material way... the environment."
                                     10-3

-------
(This page intentionally blank}

-------
11.0  MISCELLANEOUS

      11.1  Section 1120) MACT Hammer

      Comment:  Commenters IV-F-3.1, IV-D-47, IV-D-49, and IV-D-59 ask EPA to
extend the due date for applications for case-by-case MACT applications due to the
uncertainties in promulgating this rule before the MACT hammer date of May 15,
1999. The commenters explain that it would be unnecessarily burdensome for
industry and permit authorities when a final rule probably will be promulgated
shortly after the hammer date if not before. EPA has previously made these
extensions for  similar reasons. Commenter rV-F-3.2 may ask for a time extension
because, in his view, his organization did not have a full 60-day review of the rule
because the docket was not complete. However, an extension may jeopardize EPA's
ability to promulgate the rule before the hammer date.

      Response:  This comment has  already been resolved. We decided to delay
promulgation of this rule until December 1999 when the Tier 2 rule is expected to be
finalized to bring the compliance dates under the two rules more  closely in line with
each other. Details of this change are included in the Federal Register notice (64 FR
26743, 17/05/99).  The official source category schedule has been changed to
promulgation no later than Nov. 15, 2000 (i.e., the 10-year bin data).

      The Office of Mobile Sources has recently finalized the Tier 2 standards that
will limit the amount of sulfur in gasoline. Some petroleum refineries may comply
with the gasoline sulfur standards by removing both sulfur and metals from the feed
to the CCU) and thereby reduce metallic HAP emissions from the CCU regeneration
vent.  We have moved the Petroleum Refineries-Catalytic  Cracking (Fluid and Other)
Units, Catalytic Reforming Units, and Sulfur Plant Units source category to the
10-year bin to gain understanding of the effects of the gasoline sulfur standards on
refineries, decide how our final MACT rule should address these  effects, and
coordinate the implementation and compliance aspects of the MACT rule with the
schedule for implementation of the gasoline sulfur program.

      11.2  Notification Requirements

      Comment:  Commenter IV-D-29 recommends that the due date of initial
notifications should be specified for area sources that become major sources and for
new and reconstructed sources that had an initial startup after the effective date
where an application for approval or  reconstruction is not  required.  This is help the
source owner/operator and the implementing agency know when a submittal is
required.

      Response: An area source that subsequently increases its HAP emissions or
potential to emit HAP emissions such that the source is a major source becomes
subject to all applicable notification requirements in 40 CFR 63.9 of the NESHAP
General Provisions.  In this case, the notification that an affected source is subject to

                                     11-1

-------
the relevant standard must be submitted within 120 days after the source becomes
subject to the relevant standard. If you have a new or reconstructed source that had
an initial startup after the effective date where an application for approval or
reconstruction is not required, you must notify the Administrator EPA that the source
is subject to the relevant standard within 120 days after startup.  In response to the
request, we have added these requirements to the table in the final rule summarizing
the initial notification requirements.

      11.3   Reporting Requirements

      Comment: Commenter IV-D-56 believes that EPA should require quarterly
reporting for the MACT standards rather than semi-annual reporting. Citizens have a
right to know more than twice a year if their local refinery is not complying with the
law and want more frequent access to information.

      Response: Under the current NESHAP General Provisions, compliance reports
are typically submitted semi-annually . Only under certain circumstances are
quarterly reports required. Consistent with the General Provisions, this rule requires
semi-annual reports of any deviation from the emission limitations (including
operating limits) and work practice standards.  As part of our efforts to reduce the
paperwork burden on States, industry, and the federal government, we are not
requiring separate startup, shutdown, and malfunction reports when actions taken to
respond to the incident are consistent with the SSMP. If actions taken are not
consistent with the plan, detailed information must be included in the next
compliance report.  Plants also must submit specific information to obtain approval
of any planned maintenance activity that could cause a deviation from an emission
limitation.  The semi-annual reports will provide a comprehensive view of activities
and operating problems the facility is experiencing. We believe the information in the
semi-annual reports is sufficient to assess the compliance status of a facility and
whether an inspection is warranted.

      We understand the need for more frequent and up-to-date facility data. EPA is
working to resolve this problem by forming the Office of Environmental Information.
In the future, we believe this will help you obtain more and better environmental data
on the facilities in your area.

      11.4   Implementation of Final Rule

      Comment: Commenters IV-D-30, IV-D-31, and rV-D-56 recommend that EPA
track the various types and numbers of control devices installed in response to the
rule and report the information to the public.  This would verify the stringency of the
standard, confirm the anticipated emission reduction, and reveal the extent to which
estimated costs match up with actual expenditures.

      Response: We think this is a good idea particularly due to interest in the Tier 2
rule. It also is a good way to review the effectiveness of the standard and emission

                                     11-2

-------
reduction and actual costs. We will follow-up on this suggestion after the rule is
implemented should our resources permit.

       11.5   Source-specific Regulatory Approach

       Comment: Commenters IV-D-30, IV-D-31, and IV-D-56 advocate a regulatory
approach that considered all the pollutants from all media from a source or group of
sources. The uncoordinated control of different groups of pollutants from a single
source by different regulations developed at different times limits the effectiveness of
environmental programs.

      Response: We agree with your suggestions as to a more effective regulatory
approach. We have learned much from our effort to develop consolidated air and
water rules for the pulp and paper industry. We have also taken a broader view on
this rulemaking by delaying promulgation to coordinate with the larger Tier 2 effort
due to the cross-media impacts. The Agency is using a sector-based approach to
environmental protection more and more. In many cases, OAQPS has used and
continues to use a sector-based approach in developing MACT standards. In its
MACT Partnership Program, OAQPS forms partnerships with industry, trade
associations,  state agencies, environmental groups, and the general public to provide
information and expertise for developing standards.

      In general, EPA's current sector-based approach to environmental protection
takes a strategic view of problems within the overall regulatory system. This
integrated approach allows EPA to deal with issues encountered across a particular
industrial or economic group.  In using a sector-based approach, EPA works across
media, program, and Agency lines to more effectively address the problems inherent
and common to each sector. This sector-based approach is flexible; it encompasses
regulatory and non-regulatory activities, single media and multimedia considerations,
and single and multi-stakeholder involvement. The sector-based approach promotes
pollution prevention, discourages cross-media pollution transfers, and eliminates
duplication and inconsistencies. Employing a collaborative process within a sector
results in better information, better understanding of the problems, identification of
innovative solutions, and recognition of sensitive issues, which in turn results in more
cost-effective and better environmental results.

      11.6  Language Clarification

      Comment: Commenter IV-D-49, supported by Commenter IV-D-59, believes
EPA needs to add definitions for "process vent," "fuel gas," "boiler," "coke," "thermal
incinerator," and "catalytic incinerator;" be consistent when referring to the CCU
regenerator vent; clarify the definition of "catalytic cracking unit" and the capacity
basis for process heaters and boilers; differentiate between non-fired and fired boilers;
and remove the reference to TOC when referring to CCU regenerator vent. Commenter IV-D-53
includes several recommendations for improved clarity in the rule and changes in Appendix A to
be more consistent with the MACT I rule for petroleum refineries.

                                      11-3

-------
       Response:  We incorporated many of the commenters'suggestions. The final rule adds
new definitions for "process vent", "fuel gas, and "fuel gas system". The new definition of
"process vent" is based on the definition used in the MACT I rule modified for the affected
sources subject to this rule.  We did not include language from the MACT I rule relating to rule
exemptions because these exemptions, where applicable, are covered under the applicability
section of the rule and need not be repeated in this definition. The new definition of "fuel gas"
and "fuel gas system" are directly from MACT I. We clarified the definition of "incinerator" to
describe thermal and catalytic types and added the definition of "boiler" to distinguish "fired" vs
"non-fired" types, as suggested.

       Commenters also suggested changes that make the rule consistent when referring to the
CCU catalyst regenerator vent. As discussed in a previous response to comment, Item 1.1, the
final rule applies only to process vents on FCCU catalyst regenerators. We revised the proposed
definition of "catalytic cracking unit" to be identical to the NSPS which defines "fluid catalytic
cracking unit" and "fluid catalytic cracking unit catalyst regenerator".  We also corrected the
proposed rule to remove the inadvertent reference to TOC.

       Reconciling comments on the applicability of the General Provisions was more difficult.
There are cases, however, where we disagree with the commenter's recommendations. Our
reasons are summarized below.

       (1)  Performance test notification requirements under  63.7(b). We do not agree
that these notifications should not apply because they are not required under MACT I. The
MACT I rule does require performance tests for some process vents, but many of the other types
of affected  sources (storage tanks, loading racks, wastewater, etc.) have different requirements.
The process units regulated under today's final rule have much higher emissions and the tests are
more complex  particularly if the plant is complying with the PM or Ni standards for CCU.
This notification gives notice to the permitting authority in the case that he/she wants to observe
the test.

       (2) Quality assurance program requirements under  63.7(c). We do not agree that
these requirements should not apply because they are not required under MACT I and not
requiring them would reduce recordkeeping/reporting burden. These Q/A requirements are
necessary to ensure the validity of performance tests and continuous monitoring data.

       (3)  Requirements for operation  and maintenance of continuous monitoring systems
under  63.8(c)(6)-(c)(8). Many plants now operate COMS and CEMS because of NSPS
requirements and State Implementation Plan requirements. Both the proposed rule and the final
rule contain requirements for continuous monitoring systems for certain affected sources under
the MACT  standard and as such, plants containing these sources must demonstrate compliance
using data from these systems. These paragraphs establish basic requirements for meeting
applicable performance specifications, adjusting the calibration drift, and other efforts that ensure
proper operation and maintenance. The requirements  are the same as the NSPS. We have tried
to make the rule clear in that these requirements are applicable to COMS and CEMS, but not to
CPMS. The final rule establishes accuracy and calibration requirements for CPMS. Operation


                                          11-4

-------
and maintenance requirements are not included in MACT I because there are no requirements for
COMS or CEMS.

       (4) Quality control program requirements under  63.8(d). The results of a quality
control program are considered in determining the validity of monitoring data.  Like the general
provisions, the rule requires a site-specific performance evaluation test plan prior to a
performance evaluation conducted for a COMS or CEMS. Like the General Provisions, the rule
also requires a written quality control program as part of the notification of compliance status
report that describes procedures that will be used for calibrations, drift adjustments, preventative
maintenance, data recording, calculations, and reporting, accuracy audit procedures, and
corrective action for a malfunctioning monitoring system. The quality control program covers all
monitoring systems (whether a COMS, CEMS, or CPMS) and requires a written protocol that
describes procedures for calibrations, determination and adjustment of calibration drift,
preventative maintenance, data recording/calculations/reporting, and accuracy audit procedures,
including sampling and analysis methods.  The program for corrective action for a
malfunctioning continuous monitoring system can be included in this quality control  plan or in
the SSMP. We believe these requirements are necessary to ensure the proper operation and
maintenance of monitoring systems and are not burdensome. As explained above, the MACT I
sources and monitoring requirements differ from this rule  and not requiring a quality control plan
under MACT I is not relevant to this rule.

       For burden reduction purposes, we are not requiring a site specific test plan prior to any
performance test required by this rule as described in the quality assurance program requirements
in 40 CFR 63.7(c)(2)(i) of the NESHAP General Provisions and we are not requiring a
site-specific performance evaluation test plan as described in the quality  control program
requirements in 40 CFR 63.8(e)(2 through (e)(3). We are  requiring that you report the results of
the performance test and performance evaluation in the notification of compliance status report
and we are requiring that you prepare and implement a written quality control program as
described in 40 CFR 63.8(d).

       (5) Requirements for reduction of monitoring data under  63.8(g). This  provisions
relates to data reduction for continuous emission monitoring systems and continuous opacity
monitoring systems which are used by numerous plants  in this industry as a result of the NSPS
requirements. Separate provisions are included in the rule for the reduction of monitoring data
from continuous parameter monitoring systems on a 1-hour or 24-hour averaging period.  We
added an explanatory note to the entries for 40 CFR 63.8(g)(l) through (g)(4) of Table 44 to
clarify the applicability of this provision. The provisions of 40 CFR 63.8(g)(5) apply to all types
of monitoring systems at all plants.

       (6) Notification requirements for performance tests under  63.9(e).  We have
retained this notification requirements in the  final rule but added an explanatory note to
Appendix A. The rule requires notification of the performance test so that EPA can have an
observer present if desired. However, a site-specific test plan is not required.
                                          11-5

-------
       (7) Notification requirements for opacity and visible emission observations under
 63.9(f). We have retained this requirement in the final rule because plants using flares as a
means of compliance must do a Method 22 test to demonstrate no visible emissions are present.
While the final rule also includes opacity standards for plants opting to meet the NSPS
requirements and provisions for site-specific opacity standards for fluid catalytic cracking unit
catalyst regenerator vents that do not use a wet scrubber as an add-on control device, compliance
with these requirements is demonstrated using a COMS rather than Method 9 and the results of
these tests are included in the notification of compliance status report along with the results of
performance tests. We do not agree that the notification requirement should not be included
because it was not included in MACTI because MACTI does not include opacity and visible
emission standards.  We also do not agree with the commenter's assertions that this requirement
should not be included because the provisions of 40 CFR 63.6 (compliance with opacity and VE
standards) do not apply. They do apply if VE observations are made.  For this reason, the final
rule also retains the requirement in  63.10(d)(3) for reporting the results of opacity or visible
emission observations.

       (8) Recordkeeping requirements under  63.10(b)(2)(i)-(b)(2)(xiv) and 63.10(c).
The recordkeeping requirements of this rule summarize these requirements in the NESHAP
General Provisions.  The MACT I rule does not include requirements for COMS or CEMS or
quality control requirements and their recordkeeping requirements differ for this reason and
because of the differences in the types of emission sources. The requirements of the General
Provisions are not burdensome; they have been approved by OMB.  The recordkeeping
requirements provide the minimum level of information needed by EPA to determine if
compliance is being achieved and maintained.
                                         11-6

-------
                                      TECHNICAL REPORT DATA
                               (Please read Instructions on reverse before completing)
  i. REPORT NO.
   EPA-453/R-01-011
                                                                    3. RECIPIENTS ACCESSION NO.
  4. TITLE AND SUBTITLE
  National Emission Standards for Hazardous Air Pollutants
  (NESHAP) for Petroleum Refineries:  Catalytic Cracking Units,
  Catalytic Reforming Units, and Sulfur Recovery Units -
  Background Information for Promulgated Standards and
  Response to Comments	
                                                                    5. REPORT DATE
                                                                     June 2001
                  8. PERFORMING ORGANIZATION CODE
  7. AUTHOR(S)
  Robert Zerbornia, Jeff Coburn, and Marsha Branscome ,RTI
  and Robert Lucas, EPA
                                                                    8. PERFORMING ORGANIZATION REPORT NO.
 9. PERFORMING ORGANIZATION NAME AND ADDRESS

   U.S. Environmental Protection Agency
   Office of Air Quality Planning and Standards
   Research Triangle Park, NC  27711
                                                                    10. PROGRAM ELEMENT NO.
                  11. CONTRACT/GRANT NO.
                  68-D6-0014
  12. SPONSORING AGENCY NAME AND ADDRESS

   John Seitz, Director
   Office of Air Quality Planning and Standards
   Office of Air and Radiation
   U.S. Environmental Protection Agency
   Research Triangle Park. NC 27711	
                  13. TYPE OF REPORT AND PERIOD COVERED
                  Final
                  14. SPONSORING AGENCY CODE
                  EPA/200/04
  15. SUPPLEMENTARY NOTES
  16. ABSTRACT
  This report provides the background information for the final NESHAP to control hazardous air pollutants
  (HAP) from catalytic cracking units, catalytic reforming units, and sulfur recovery units at petroleum
  refineries. This document contains summaries of public comments received on the proposed rule and EPA
  responses.	"	
 17.
                                       KEY WORDS AND DOCUMENT ANALYSIS
                    DESCRIPTORS
                                                  b. IDENTIFIERS/OPEN ENDED TERMS
                                                                                       c. COSATI Field/Group
 emission controls
 environmental impacts
 estimates of air emissions
Air Pollution Control
Petroleum Refineries
Catalytic Cracking Unit, Catalytic
Reforming Unit, and Sulfur
Recovery Unit
Hazardous Air Pollutants
  18. DISTRIBUTION STATEMENT

   Release Unlimited
19. SECURITY CLASS (Report)
  Unclassified
21. NO. OF PAGES
115
                                                  20. SECURITY CLASS (Page)
                                                    Unclassified
                                                                                       22. PRICE
EPA Form 2220-1 (Rev. 4-77)
                        PREVIOUS EDITION IS OBSOLETE

-------