United Sta:es
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 2771-'
EPA-450/3-80-033b
June 1982
Air
VOC Fugitive
"E
Synthetic
o
o
Manufacturing
Industry
Background
Information for
Promulgated
Standards
-------
-------
TECHNICAL REPORT DATA
(Please read Insmicnons on the reverse before completing)
1. REPORT NO.
EPA-450/3-8Q-033b
3. RECIPIENT'S ACCESSION NO,
4. TITLE AND SUBTITLE
VOC Fugitive Emissions in Synthetic Organic Chemicals
Manufacturing Industry - Background Information for
Promulgated Standards of Performance
5. REPORT DATE
February 1983
6. PERFORMING ORGANIZATION CODE
7. AUTHORIS)
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Office of Air Quality Planning and Standards
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-3058
12. SPONSORING AGENCY NAME AND ADDRESS
OAA for Air Quality Planning and Standards
Office of Air, Noise, and Radiation
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/200/04
1S. SUPPLEMENTARY NOTES
Standards of performance to control equipment leaks of VOC from new modified, and
Standards of performance to control equipment leaks of VOC from new modified, and
reconstructed Synthetic Organic Chemical Manufacturing Industry (SOCMI) plants are
being promulgated under the Authority of Section 111 of the Clean Air Act. These
standards apply only to equipment in process units producing one or more SOCMI
chemicals and for which construction or modification began on or after the date
of proposal of the standards. This document contains a summary of public comments,
EPA responses, and a discussion of differences between the proposed and promulgated
standards of performance.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI f-ield/Gtoup
Air Pollution
Pollution Control
Standards of Performance
Volatile Organic Compounds
Organic Chemical Industry
Air Pollution Control
18, DISTRIBUTION STATEMENT
Release unlimited. Available from EPA
Library (MD-35), Research Triangle Park,
North Carolina 27711
19 SECURITY CLASS I TlliS Report/
unclassified
21. NO. Of PAGES
20. SECURITY CLASS (This page)
unclassified
22. PRICE
Form 2220-1 (R«». 4-77) OBEVIOUS EOI T ION i s o asOLE TE
-------
-------
EPA-450/3-80-033b
VOC Fugitive Emissions
in Synthetic Organic Chemicals
Manufacturing Industry-
Background Information
for Promulgated Standards
Emission Standards and Engineering Division
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
June 1982
t-OU
-------
This report has been reviewed by the Emission Standards and Engineering Division of the Off ice of Air
Quality Planning and Standards, EPA, and approved for publication. Mention of trade names or
commercial products is not intended to constitute endorsement or recommendation for use. Copies of
this report are available through the Library Services Office (MD-35), U.S. Environmental Protection
Agency, Research Triangle Park, N.C. 27711, or from the National Technical Information Services,
5285 Port Royal Road, Springfield, Virginia 22161.
-------
ENVIRONMENTAL PROTECTION AGENCY
Background Information
Final Environmental Impact Statement for
Equipment Leaks of VOC in the Synthetic Organic
Chemical Manufacturing Industry
Prepared by:
JaaRlR. Farmer ' (bate)
Director, Emission Standards and Engineering Division
U. S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
1. The promulgated standards of performance will limit emissions of
volatile organic compounds (VOC) from new modified, and reconstructed
synthetic organic chemical manufacturing industry (SOCMI) process
units. Section 111 of the Clean Air Act (42 U.S.C. 7411) as amended,
directs the Administrator to establish standards of performance for any
category of new stationary source of air pollution that ". . .causes
or contributes significantly to air pollution which may reasonably be
anticipated to endanger public health or welfare."
2. Copies of this document have been sent to the following Federal
Departments: Labor, Health and Human Services, Defense, Transporta-
tion, Agriculture, Commerce, Interior, and Energy; the National Science
Foundation; the Council on Environmental Quality; members of the State
and Territorial Air Pollution Control Officials; EPA Regional
Administrators; and other interested parties.
3. For additional information contact:
Mr. Gil Wood
Standards Development Branch (MD-13)
U. S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
Telephone: (919) 541-5578
4. Copies of this document may be obtained from:
U. S. EPA Library (MD-35)
Research Triangle Park, N.C. 27711
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
m
-------
-------
TABLE OF CONTENTS
Page
CONTENTS . . . iv
LIST OF TABLES xi
LIST OF FIGURES x-jv
Chapter 1 - SUMMARY . . 1-1
1.1 SUMMARY OF CHANGES SINCE PROPOSAL 1-1
1.1.1 Alternative Standards for Valves 1-2
1.1.2 Standards for Pumps 1-4
1.1.3 Requirements for Inaccessible Valves 1-5
1.1.4 Requirements for Existing Reciprocating
Equipment 1-5
1.1.5 Standards for Control Devices 1-6
1.1.5.1 Flares. . 1-6
1.1.5.2 Combustion Device . 1-7
1.1.6 Delay of Repair Provisions . 1-8
1.1.7 Reporting Requirements 1-10
1.1.8 Requirements for Equipment in Vacuum Service ... 1-10
1.1.9 Definition of In VOC Service 1-10
1.1.10 Requirements for Units With No Equipment
in VOC Service 1-11
1.1.11 Exemption of Beverage Alcohol Producers. ..... 1-12
1.1.12 Calibration Gas Requirements 1-12
1.1.13 Equivalency Provisions ; 1-13
1.1.14 Production Rate Cutoff 1-13
1.1.15 Definition of No Detectable Emissions. . 1-13
1.2 SUMMARY OF IMPACTS OF PROMULGATED ACTION. ......... 1-14
1.2.1 Alternatives to Promulgated Action 1-14
1.2.2 Environmental Impacts of Promulgated Action. . . . 1-14
1.2.3 Energy and Economic Impacts of Promulgated
Action . . . 1-15
1.2.4 Other Considerations .......... 1-15
1.2.4.1 Irreversible and Irretrievable
Commitment 1-15
1.2.4.2 Environmental and Energy Impacts of
Delayed Standards 1-15
1.3 SUMMARY OF PUBLIC COMMENTS 1-17
Preceding page blank
-------
TABLE OF CONTENTS (CONTINUED)
Page
Chapter 2 - NEED FOR THE STANDARD 2-1
2.1 SIGNIFICANCE OF EMISSIONS AND ENVIRONMENTAL
IMPACT OF THEIR CONTROL 2-1
2.2 CONTRIBUTION OF CURRENT REGULATIONS TO FUGITIVE
EMISSION CONTROL 2-8
2.3 STANDARDS' BENEFIT TO PUBLIC HEALTH; CONTROLLING VOC TO
CONTROL OZONE . 2-20
Chapter 3 - BASIS FOR THE STANDARDS 3-1
3.1 SELECTION OF FINAL STANDARDS 3-1
3.1.1 Basis For the Final Standards 3-1
3.1.2 Other Comments Concerning the Selection
of Standards 3-5
3.2 EMISSION FACTOR DEVELOPMENT 3-7
3.3 MODEL UNITS . 3-13
3.4 EMISSION REDUCTIONS 3-13
3.5 FORMAT OF STANDARDS 3-15
3.5.1 Performance Standards for Valves 3-17
3.5.2 Equivalent Equipment and Work Practice Standards
for Valves 3-22
3.5.3 Work Practice Standards for Pumps and
Compressors 3-22
3.5.4 Performance Standards for Pumps and Compressors. . 3-23
3.5.5 Performance Standards for Sampling Systems .... 3-24
3.5.6 Emission Limit vs. Concentration Limit 3-25
3.5.7 Complexity of the Standards 3-26
Section 4 - CONTROL TECHNOLOGY 4-1
4.1 FLARES 4-1
-------
TABLE OF CONTENTS (CONTINUED)
Page
4/2 LEAK DETECTION AND REPAIR PROGRAMS . ". 4-15
4.2.1 Monitoring Interval 4-15
4.2.2 Time Estimates 4-31
4.2.3 Repair Requirements . 4-35
4.2.4 Accessibility 4-39
4.2.5 Other Monitoring Methods ............. 4^42
4.2.6 Potential Problems 4-44
4.3 LEAK DEFINITION 4-48
4.4 SAFETY CONSIDERATIONS ........ 4-61
4.5 PRESSURE RELIEF DEVICES 4-67
4.6 COMBUSTION DEVICE 4-76
4.7 NO LEAK EQUIPMENT 4-83
4.8 DUAL SEALS '. 4-88
4.9 SAMPLING SYSTEMS . 4-99
4.10 CLOSED VENT SYSTEM 4-103
4.11 OPEN-ENDED LINES 4-104
4.12 RECIPROCATING PUMPS AND COMPRESSORS .... 4-105
Section 5 - APPLICABILITY OF STANDARDS 5-1
5.1 SOCMI LIST 5-1
5.2 VAPOR PRESSURE CUTOFF . . . . . 5-16
5.3 PERCENT VOC CUTOFF 5-22
5.4 PROCESS UNIT DEFINITION 5-23
5.5 SMALL MANUFACTURERS '. 5-28
5.6 VOC DEFINITION 5-30
5.7 PILOT PLANT AND R&D FACILITIES. 5-31
vii
-------
TABLE OF CONTENTS (CONTINUED)
Page
5.8 FLANGES 5-32
5.9 VACUUM SERVICE 5-34
5.10 ENCLOSED BUILDINGS 5-35
Section 6 - ENVIRONMENTAL IMPACT 6-1
6.1 EMISSIONS ANALYSIS . . . . 6-1
6.2 ENVIRONMENTAL IMPACT. . . 6-11
Section 7 - ECONOMICS 7-1
7.1 COST ESTIMATES 7-2
7.2 GENERAL ECONOMIC ISSUES 7-24
7.3 COST-BENEFIT EVALUATION 7-23
Section 8 - LEGAL 8-1
8.1 EPA'S REGULATORY RESPONSIBILITY 8-1
8.2 PRIORITY LIST 8-9
8.3 EXECUTIVE ORDERS 8-12
8.4 COURT DECISIONS 8-15
8.5 UNIT OPERATION STANDARDS 8-20
8.6 STATES AUTHORITY . 8-21
8.7 REQUEST FOR WITHDRAWAL OR DELAY 8-22
8.8 STATUTORY TIME REQUIREMENTS FOR PROPOSAL 8-24
8.9 TECHNOLOGY-FORCING STANDARDS 8-25
Section 9 - MODIFICATION AND RECONSTRUCTION 9-1
9.1 CLARIFICATION 9-1
9.2 EMISSION CRITERIA FOR MODIFICATION 9-4
vi n
-------
TABLE OF CONTENTS (CONTINUED)
Page
9.3 AFFECTED FACILITY DEFINITION FOR MODIFIED SOURCES'. . . . 9-6
9.4 TECHNICAL PROBLEMS WITH RETROFIT 9-9
9.5 CAPITAL EXPENDITURE CRITERION - 9-11
9.6 COVERAGE OF RECONSTRUCTIONS 9-14
Section 10 - EQUIVALENCY . 10-1
Section 11 - RECORDKEEPING AND REPORTING 11-1
11.1 REPORTING BURDEN 11-1
11.2 RECORDKEEPING : . 11-5
11.3 TAGGING 11-9
11.4 OTHER PROVISIONS 11-10
11.5 BURDEN ESTIMATES 11-10
Section 12 - TEST METHOD / 12-1
12.1 CALIBRATION GAS 12-1
12.2 INSTRUMENTATION 12-3
12.3 VARIABLE RESPONSE TO DIFFERENT CHEMICALS 12-5
12.4 OTHER COMMENTS 12-7
Section 13 - ENFORCEMENT AND COMPLIANCE CONCERNS 13-1
Section 14 - ALTERNATIVE STANDARDS 14-1
Section 15 - MISCELLANEOUS 15-1
APPENDIX A - RESPONSES TO COMMENTS ON THE ADDITIONAL INFORMATION
DOCUMENT r A-l
A.I EMISSION FACTORS A-4
A.2 MODEL UNITS A-22
IX
-------
TABLE OF CONTENTS (CONTINUED)
Page
A.3 EMISSION REDUCTIONS . . . A-26
A.3.1 The LDAR Model A-26
A.3.2 Model Input Parameters . .- A-52
A.3.3 Modeling Results A-59
A.3.4 Control Device . A-64
A.4 COSTS A-68
A. 5 ECONOMICS . A-75
A.6 COMMENTS ON SUBJECTS NOT COVERED IN THE AID A-77
A.7 PREVIOUSLY SUBMITTED COMMENTS A-99
APPENDIX 3 - MONITORING METHODS B'1
-------
LIST OF TABLES
Table Page
1-1 IRRETRIEVABLE LOSSES WHICH WOULD OCCUR IF STANDARD
IMPLEMENTATION WERE DELAYED 1-17
1-2 LIST OF COMMENTS ON PROPOSED STANDARDS OF PERFORMANCE
FOR FUGITIVE EMISSION SOURCES IN SOCMI. . . . . ... .. 1-18
3-1 CONTROL COSTS PER MEGAGRAM OF VOC REDUCED . .... ... 3-3
3-2 COMPARISON OF EMISSION FACTORS, kg/hr/source 3-9
3-3 EQUIPMENT COUNTS FOR FUGITIVE VOC EMISSION SOURCES IN
SOCMI MODEL UNITS . ........ 3-14
3-4 SUMMARY OF CONTROL EFFECTIVENESS FOR SOCMI FUGITIVE
EMISSION SOURCES '. 3-16
4-1 FLARE EMISSION STUDIES COMPLETED AS OF OCTOBER 1982 ... 4-4
4-2 LEAK OCCURRENCE/RECURRENCE RATE ESTIMATES FROM BID. . . . 4-16
4-3 OCCURRENCE RATES AND RECURRENCE RATES FOR VALVES
DETERMINED IN SOCMI UNITS ... . . . . . ... ... . . . 4-19
4-4 EMISSIONS REDUCTION AND COST EFFECTIVENESS OF LEAK
DETECTION AND REPAIR PROGRAMS FOR VALVES WITH
14 PERCENT RECURRENCE, $/Mg SOCMI MODEL UNIT C. ... . 4-27
4-5 ESTIMATED VERSUS ACTUAL MONITORING TIMES FOR VARIOUS
SOCMI PROCESS UNITS ...... ...... ........ 4-33
4-6 SUMMARY OF PERCENT OF SOURCES DISTRIBUTION CURVES AND
PERCENT OF MASS EMISSION CURVES AT VARIOUS ACTION
LEVELS .................... 4-52
6-1 EMISSION FACTORS FOR SOCMI EMISSIONS ANALYSES . . . . . . 6-3
6-2 EQUIPMENT COUNTS USED IN ESTIMATING SOCMI UNIT
EMISSIONS ............... 6-4
6-3 EXAMPLE OF EMISSIONS ESTIMATED FOR MODEL UNIT B IN
ABSENCE OF STANDARDS, ...... ..... 6-6
6-4 SUMMARY OF ESTIMATED EMISSIONS IN THE ABSENCE OF
STANDARDS BY SOCMI UNIT ........... 6-7
-------
LIST OF TABLES (CONTINUED)
Table Page
6-5 SUMMARY OF STANDARDS AND ESTIMATED EFFICIENCIES FOR
NEW SOURCES OF FUGITIVE VOC EMISSIONS "IN SOCMI 6-8
6-6 SUMMARY OF EMISSIONS ESTIMATES FOR AVERAGE SOCMI UNITS
IN MG/YR 6-9
6-7 NATIONWIDE IMPACT OF SOCMI NEW SOURCE STANDARDS FOR
FUGITIVE VOC EMISSIONS IN THE FIFTH YEAR 6-10
7-1 SUMMARY OF CAPITAL COST ESTIMATES OF SOCMI STANDARDS,
1978$ 7-3
7-2 ANNUALIZED COST ESTIMATES OF SOCMI STANDARDS, 1978$ ... 7-4
7-3 CAPITAL AND ANNUALIZED COST SUMMARIES: NATIONWIDE
PROJECTIONS 7-5
7-4 THRESHOLD PLANT CAPACITIES OF CHEMICALS POTENTIALLY
PRODUCED BY SMALL FIRMS AT CURRENT PRICES 7-19
7-5 COMPARISON OF NEW PLANT AND THRESHOLD PLANT CAPACITIES OF
CHEMICALS POTENTIALLY PRODUCED BY SMALL FIRMS 7-21
7-6 PRODUCT PRICE AND PROFITABILITY DATA, 1978-1980 7-23
12-1 COMPARABLE ESTIMATES FOR PERCENT LEAKING (VALVES)
(24 SOCMI Process Units). 12-8
14-1 OVERALL LEAK FREQUENCIES FOR VALVES IN THE SOCMI
24-UNIT STUDY 14-5
A-l LIST OF COMMENTERS ON THE ADDITIONAL INFORMATION
DOCUMENT .FOR FUGITIVE VOC EMISSION SOURCES A-2
A-2 COMPARISON OF EQUIPMENT COUNTS FOR SOCMI PROCESS UNITS. . A-13
A-3 EMISSION FACTORS (KG/HR/SOURCE) FOR VALVES: CALCULATED
FROM EPA'S MAINTENANCE STUDY DATA A-30
A-4 SUMMARY OF INPUTS AND RESULTS OF LDAR MODEL SENSITIVITY
TO OCCURRENCE RATE, LEAK FREQUENCY, AND EMISSION
FACTOR FOR VALVES IN AN AVERAGE MODEL UNIT B. . ; . . . A-42
A-5 INITIAL LEAK REPAIR LABOR-HOURS REQUIREMENT FOR VALVES
BY MODEL UNIT A-71
xn
-------
LIST OF TABLES (CONTINUED)
Table
A- 6
TOTAL COSTS FOR
MODEL UNIT
INITIAL LEAK REPAIR FOR VALVES BY
Page
A-72
A-7 ANNUAL MONITORING AND LEAK REPAIR LABOR REQUIREMENTS
(Monthly/Quarterly Leak Detection and Repair Program
for Valves) A-73
A-8 ANNUAL MONITORING AND LEAK REPAIR COSTS FOR MONTHLY/
QUARTERLY MONITORING OF VALVES BY MODEL UNIT ...... A-74
A-9 COMPARISON OF LEAK FREQUENCIES FOR EQUIPMENT IN LIGHT
LIQUID SERVICE IN SOCMI AND SPOCMI A-95
A-10 CROSS-REFERENCE OF COMMENT AND SECTION CONTAINING
RESPONSE A-100
XI 1 1
-------
LIST OF FIGURES
Figure Page
4-1 Fraction of valves operating improperly at the end of
the Initial year under leak detection and repair
programs of various monitoring intervals for an
average SOCMI unit 4-18
4-2 Cost effectiveness of leak detection and repair programs
for valves as a function of monitoring interval for an
average SOCMI unit 4-24
4-3 Cost effectiveness of valve leak detection and repair
programs as a function of emission reduction for
average SOCMI model unit C 4-26
5-1 Example for process unit definition 5-26
5-2 Cost effectiveness for low volume SOCMI production
units assuming valve standards only 5-33
7-1 Minimum price and capacity combinations 7-17
14-1 Cost effectiveness as a function of leak frequency for
monthly leak detection and repair programs for valves
in SOCMI units 14-4
A-l Emission factor for leaking valves in gas service as
a function of leak frequency A-31
A-2 Emission factor for leaking valves in light liquid
service as a function of leak frequency A-32
A-3 Comparison of emission factors computed for leaking
valves in gas service A-35
A-4 Comparison of emission factors computed for leaking
valves in light liquid service A-36
A-5 Thirty-day occurrence rate as a function of leak
frequency for valves in gas and light liquid
service A-40
A-6 Cost effectiveness as a function of leak frequency for
monthly leak detection and repair programs for valves
in SOCMI units A-44
A-7 Comparison of valve occurrence rates assuming a linear
relationship and an exponential model A-50
xiv
-------
1. SUMMARY
On January 5, 1981, the Environmental Protection Agency (EPA) proposed
standards of performance for fugitive emission sources in the Synthetic
Organic Chemical Manufacturing Industry (46 FR 1136) under the authority of
Section 111 of the Clean Air Act. Public comments were requested on the
proposal in the Federal Register. There were 52 commenters. Most of the
commenters were industry representatives. Also commenting were representa-
tives of state and local air pollution agencies, vendors of equipment used
to control fugitive emissions, and a representative of an environmental
group. The comments that were submitted, along with responses to these
comments, are summarized in this document and in the Additional Information
Document (AID) published on April 26, 1982. The .AID contains a technical
discussion of methodologies and estimates of emissions, emission reductions,
and costs. Comments on the AID and EPA's responses to those comments are
presented in Appendix A. This summary of comments and responses and the AID
serve as the basis for the revisions made to the standards between proposal
and promulgation.
1.1 SUMMARY OF CHANGES SINCE PROPOSAL
The proposed standards were revised as a result of review of public
comments. Significant changes were made in the following areas:
Alternative standards for valves
Standards for pumps
Requirements for inaccessible valves
Requirements for existing reciprocating equipment
Standards for control devices
Delay of repair provisions
Reporting requirements
Requirements for equipment "in vacuum service"
« Definition of "in VOC service"
1-1
-------
Requirements for units with no equipment in VOC service
Exemption of beverage alcohol producers
Calibrantion gas requirements
Equivalency provisions
Production rate cutoff
Definition of "no detectable emissions"
1.1.1 Alternative Standards for Valves
At proposal, two alternative standards were provided for valves in
gas/vapor and light liquid service. Both of these alternatives called for
one year of monthly monitoring to obtain data on which to base the alter-
native standard. The first alternative standard was based on an allowable
percentage of valves leaking. Since an industry-wide allowable leak
percentage was not possible due to variability of leak frequency among
process units, an allowable percentage of valves leaking was to be deter-
mined for that unit based on data collected on that unit. The allowable
percentage was to be the sum of the monthly baseline percentage and the
monthly incremental percentage. A minimum of one performance test was
required annually. The second alternative standard for valves allowed the
development of work practices that would achieve the same result as the
proposed leak detection and repair program. This alternative would allow a
unit to vary the monitoring interval and to use valves with a low proba-
bility of leaking in order to achieve an overall goal of emission
reductions.
Based on comments received on the proposed alternative standards and on
analysis of the results from SOCMI screening and maintenance studies, the
alternative standards for valves were reexamined. As a result, these
standards were clarified in their intent and refined to reflect the informa-
tion gathered on SOCMI units.
The first alternative standard was simplified to a 2 percent limitation
as the maximum percent of valves leaking within a process unit, determined
by a minimum of one performance test annually. This alternative will
provide a cutoff for valves to eliminate unreasonable costs. It will also
provide an incentive to maintain a good performance level and promote
1-2
-------
low-leak unit design as was indicated by one commenter. Inaccessible valves
that would not be monitored on a routine basis under §60.482-7 would be
included in the annual test since an annual test of these valves is not
considered burdensome. The standard could be met by implementing any type
of leak detection and repair program and engineering controls chosen at the
discretion of the owner or operator. This alternative standard would allow
an affected facility to comply with an allowable percentage of valves
leaking without having to determine a specific performance level by a year-
long monthly monitoring program. If the results of a performance test show
a percentage of valves leaking higher than 2 percent, however, the process
unit would not be in compliance with the alternative standard. Finally, if
an owner or operator determines that he no longer wishes to comply with this
alternative standard, he can submit a notification in writing to the
Administrator stating that he will comply with the work practice standard in
§60.482-7.
EPA also recognized benefits which may be derived from statistically-
based skip-period leak detection and repair programs. Under the skip-period
leak detection provisions in the final standards, an owner or operator could
skip from routine monitoring (monthly) to less frequent monitoring after
completing a number of successful sequential monitoring intervals.
Considering a performance level of less than 2 percent leaking and
better than 90 percent certainty that all periods have this performance
level, the two following sets of consecutive periods and fractions of
periods skipped were determined for SOCMI units:
(1) two consecutive quarterly periods achieved to skip to
semi-annual monitoring, and
(2) five consecutive quarterly periods achieved to skip to
annual monitoring.
This alternative also requires that, if an owner or operator does not meet
the required statistical level of performance, he/she must revert to the
monthly leak detection and repair program that is specified in §60.482-7.
Compliance with this alternative work practice standard would be determined
by inspection and review of records.
1-3
-------
1.1.2 Standards for Pumps
Since proposal, EPA has analyzed the annualized cost of controlling VOC
emissions and the resultant VOC reduction for each alternative control
technique. The resultant cost effectiveness ratios were used as the basis
for the selection of the final standards. (Section 3.1)
For pumps the control costs incurred for each megagram of VOC emissions
reduced and emission reductions achieved were determined for two leak
detection and repair programs and for the use of dual mechanical seals with
controlled degassing vents. Both leak detection and repair programs incur
lower costs than the costs which would be incurred with equipment installa-
tion. The lowest average and incremental costs per Mg are associated with a
monthly leak detection and repair program. The monthly program achieves a
higher degree of control than the quarterly program, but it achieves a lower
degree of control than installation of control equipment. Even though
control equipment provides for the greatest amount of VOC reduction, the
costs to obtain this reduction are high. Because the incremental costs for
equipment are unreasonably high in light of the resulting incremental
emission reductions, monthly leak detection and repair was selected as the
basis for the standard for pumps.
The leak detection and repair program requires monthly leak detection
of pumps in light liquid VOC service. Leak detection is to be performed
with a portable VOC analyzer according to Reference Method 21. If a reading
of 10,000 ppmv or greater is obtained, a leak is detected. Initial repair
of the leak must be attempted within 5 days and the repair must be completed
within 15 days. Delay of repair in order to equip a leaking pump with dual
seal systems (required by the standards) is allowed for a period of
6 months. Delay of repair also would be allowed for pumps that could not be
repaired without a process unit shutdown. Delay of repair is not expected
to be needed for most situations, however, because pumps are commonly spared
in SOCHI.
The equipment standards also have been incorporated into the final
standards, since they are equivalent to the monthly leak detection and
repair program that is the basis of the standards for pumps in light liquid
1-4
-------
service. An owner or operator may comply with the equipment standards
(which have not been changed since proposal) if he or she prefers.
1.1.3 Requirements for Inaccessible Valves
Some valves are difficult to monitor because access to .the valve is
restricted. Difficult to monitor valves can be eliminated in new process
units but can not be eliminated in existing process units. In new units,
all valves will be subject to the proposed leak detection and repair
program. However, for process units that become affected by a modification
or reconstruction, the standards have been changed since proposal to allow
an annual leak detection and repair program for valves which are difficult
to monitor. Valves which are difficult to monitor are defined as valves .
which would require elevating.the monitoring personnel more than two meters
above any permanent available support surface. The intent of this
definition is that ladders should be used to elevate monitoring personnel
under safe conditions. However, valves that cannot be safely monitored by,
at least, the use of ladders are classified as difficult to monitor, and,
therefore, they may be monitored annually rather than monthly.
In addition, some valves are unsafe to monitor. Valves which are
unsafe to monitor cannot be eliminated in new or existing units. The final
standards have been changed to allow an owner or operator to submit a plan
that defines a leak detection and repair program conforming with the routine
monitoring requirements of the standards as much as possible given that
monitoring should not occur under unsafe conditions. Valves which are
unsafe to monitor are defined as those valves which could, based on the
judgment of the owner or operator, expose monitoring personnel to imminent
hazards from temperature, pressure, or explosive process conditions.
1.1.4 Requirements for Existing Reciprocating Equipment
Even though reciprocating pumps and compressors are not common in
SOCMI, they do exist in some SOCMI units and may even be necessary in some
applications. In the proposed standards a provision was made that required
reciprocating pump and compressors to enclose the seal area and vent any
emissions to a suitable control system. This provision remains in the final
standards. Based on a review of public comments, EPA has concluded that
1-5
-------
this option is feasible for reciprocating pumps, as well as for most
reciprocating compressors.
However, there may be potential problems associated with retrofitting a
seal vent system to some existing reciprocating compressors. On older
compressors, the distance piece between the cylinder and driver may not be
enclosed and vented. In such cases, retrofitting a vent system to the
compressor in order to comply with the standards could require recasting of
the distance piece or even replacement of the compressor. The cost of doing
recasting or replacement was determined to be unreasonable. Therefore, in
the final regulation EPA has provided an exemption for reciprocating
compressors within facilities which have become affected by virtue of
modification and reconstruction. The exemption applies only to those
specific instances where the distance piece or the compressor must be
replaced. Such compressors will be exempt from the standards until they are
replaced by new compressors or the distance pieces are replaced.
1.1.5 Standards for Control Devices
1.1.5.1 Flares. At proposal, flares were not specifically listed as
an acceptable control option for the reduction of fugitive VOC emissions.
The results of available flare efficiency studies were not considered
relevant. The gas streams tested were not considered representative of the
streams to be controlled in SOCMI. In some cases the flare design was not
representative of flares in the industry. In others the analytical method
was questionable. No method for measuring flare efficiency (evaluating
flare performance) was available. Theoretical calculations indicated that
flare efficiency could be as low as 60 percent for destruction of VOC in
low-flow intermittent streams sent to a large flare. This efficiency was
cited in several background documents (Ethylbenzene/Styrene, Benzene
Fugitive, SOCMI Fugitive VOC) and served as a primary consideration in not
specifically allowing the use of flares.
Since proposal, the use of flares was reconsidered for the SOCMI
standards. Further actual flare measurement results have become available,
most notably from the CMA-EPA study (IV-A-32) , since the 60 percent
*References to Docket Entry Numbers for Docket No. A-79-32 are
presented in this manner throughout this document.
1-6
-------
theoretical estimate was made. In the CMA-EPA study, steam-assisted flares,
air-assisted flares, and flares operated without assist were investigated
over a wide range of exit velocity, gas composition, and flare gas heat
content conditions. After review of available flare efficiency data (see
Section 4.1), EPA has concluded that smokeless flares operated with a flame
present and exit velocities less than 18 m/sec (60 ft/sec) with flare gas
heat contents greater than 11.2 MJ/scm (300 Btu/scf) for steam-assisted
flares or exit velocities less than 18 m/sec (60 ft/sec) and flare gas heat
contents greater than 7.45 MJ/scm (200 Btu/scf) for flares operated without
assist are acceptable alternatives to enclosed combustion devices
(incinerators, boilers, process heaters) and vapor recovery systems such as
carbon adsorbers and condensation units. Air-assisted flares operated
smokelessly with a flame present are also permitted if the heat content of
the flared gas is above 11.2 MJ/scm (300 Btu/scf) and the exit velocity
mee's maximum velocity criterion which is dependent upon the heat content of
the gas. They may be applied to control of emissions from pump seals (or
degassing reservoirs), compressor seals (or degassing reservoirs), and
pressure relief devices. The requirement for the presence of a flame can be
ensured by monitoring the flare's pilot light with an appropriate heat
sensor, such as a thermocouple. .To ensure smokeless operation, visible
emissions from a flare would be limited to less than 5 minutes in any 2-hour
period.
1.1.5.2 Combustion Device. The temperature and residence time
specified for combustion devices in the proposed regulation were based on
data analyzed in an EPA memo ("Thermal Incinerators and Flares") dated
August 22, 1980 (II-B-31). The data base contained in this memo included
Union Carbide laboratory studies, EPA and industry field tests, and 147
tests on existing incinerators in Los Angeles county. These data indicate
that greater than 98 percent efficiency is attainable by incinerators
operating at 1500°F (816°C) and 0.75 seconds residence time. The memo
concludes that 98 percent efficiency, or less than 20 ppmv in the exhaust
stream, is achievable in many situations at less than 1600°F (871°C) and
0.75 seconds residence time. Furthermore, the data indicate that greater
1-7
-------
than 93 percent efficiency is achievable in many situations at 1400°F
(760°C) with 0.75 seconds residence time.
While thermal incinerators are proven control devices for destruction
of VOC emissions, they are not the only enclosed combustion devices that
could be used to achieve 95 percent destruction efficiency. In fact,
boilers and process heaters already existing on-site are expected to be used
for eliminating the small VOC streams covered by the standards. In order to
ensure that these combustion systems achieve the requisite degree of
control, the temperature-residence time requirements for enclosed combustion
devices have been retained in the final regulation. By meeting the require-
ments of 1500°F (816°C) and 0.75 seconds residence time, an owner or
operator will ensure that his combustion device attains the required
95 percent efficiency.
However, other combustion systems, such as catalytic incinerators, are
also applicable to the control of small VOC streams. Systems which employ
catalysts typically operate at lower temperatures and would not be able to
meet these operating requirements. Therefore, the temperature-residence
time requirements will not apply to combustion systems which employ
catalysts. Such systems will have to meet a design requirement which
assures a destruction efficiency of 95 percent. This change will permit the
use of catalytic combustion units for control of fugitive VOC emissions
without an equivalency determination.
1.1.6 Delay of Repair .Provisions
EPA recognizes that, in a few cases, repair of leaking sources may need
to be delayed for technical reasons. Based on comments concerning the
proposed delay of repair provisions, the delay of repair provisions have
been expanded in the final standards. Five provisions for delay of repair
are included in the final standards. The first provision allows delay of
repair where repair is technically or physically infeasible without a
process unit shutdown (complete or partial). An example of such a situation
would be a leaking valve that could not be isolated from the process stream
and requires complete replacement or replacement of internal parts. When a
valve cannot be physically isolated from the process stream, the process
1-8
-------
unit must be shutdown to effect certain repairs on the valve. Thus, because
EPA believes shutdown to effect repair of valves is unreasonable, EPA allows
delay of repair when repair is infeasible without a process unit shutdown.
The second provision allows for delay beyond a process unit shutdown
in limited instances due to the lack of spare valve assemblies. The delay .
will be allowed if valve assembly replacement is necessary, valve assembly
supplies are depleted,, and valve assembly supplies were sufficiently stocked
before they are depleted.
A third provision has been added to avoid causing unreasonable delays
in returning a process unit to production if a unit goes down briefly due to
unforeseen circumstances. Delay of repair beyond,an unscheduled process
unit shutdown will be allowed.if the shutdown is less than 24 hours in
duration. This provision allows for delay of repairs until the next
shutdown if an unscheduled shutdown is too short to allow repair or
replacement of equipment which cannot be repaired on-line. However, repair
of the leaking equipment would be required at the next process unit
shutdown.
A fourth provision also has been added to. clarify the applicability of
the standards to sources that are out of service (usually spare equipment).
Delay of repair of spare fugitive emission sources for which leaks have been
detected will be allowed for sources which are isolated from the process and
which do not remain in VOC service. This provision is applicable only.to
those pieces of equipment that have been isolated from VOC service and
properly purged. Delay of repair will not be allowed for spare equipment
that is pressurized and prepared to be placed on-line; such equipment is
considered to be in (VOC) service.
In addition, a fifth provision has been added to allow delay of repair
for certain leaking pumps. Sometimes, leaking pumps cannot be repaired
under the leak detection and repair program unless the owner or operator
installs dual seals with barrier fluid systems. For these leaking pumps, a
delay of repair for a period of six months will be allowed to install the
required equipment.
1-9
-------
1.1.7 Reporting Requirements
The proposed standards included reporting provisions requiring periodic
reports of leak detection and repair efforts within a process unit. The
reported information was regarded by EPA as a good way to judge how
diligently the required leak detection and repair program had been
implemented. The reporting requirements were considered a means of reducing
in-plant inspections. The costs of reporting were assessed and judged
reasonable.
EPA continues to believe that reporting requirements would reduce
in-plant inspections as a means of determining compliance. But the
quarterly reporting requirements have been reduced to semiannual require-
ments in the final standards. The semiannual reports will consist of data
recorded on leak detection and repair of valves, pumps, and other equipment.
The semiannual reporting requirements may be waived for process units in
States where the program has been delegated for enforcement, if EPA, in the
course of delegation, approves reporting requirements or an alternative
means of compliance surveillance adopted by the State and if the process
unit complies with the requirements established by the State. EPA maintains
the authority for discretionary use of Section 114 of the Clean Air Act to
obtain records and make in-plant inspections. One-time reporting require-
ments, such as the notification requirements in the General Provisions, have
also been retained in the final standards.
1.1.8 Requirements for Equipment "in Vacuum Service"
The proposed regulation defined a source to be in vacuum service if it
is operating at an internal pressure which is continuously less than
100 kPa. It should be noted that 1 atmosphere equals about 100 kPa.
Fugitive emission sources may operate at a pressure below 100 kPa. It would
be inappropriate, in EPA's judgement, to cover such sources because sources
operating even at a slight vacuum would have little if any potential to emit
VOC. There was some confusion over the 100 kPa definition because it is so
near atmospheric pressure. Therefore, to avoid any further misunderstanding
about the standard, the definition for vacuum service has been changed as
follows:
1-10
-------
"In vacuum service" means that equipment is operating with an
internal pressure which is at least 5 kPa below ambient pressure.
1.1.9 Definition of "in VOC Service"
EPft proposed a 10 percent VOC cutoff to avoid covering those sources
that have only small amounts of ozone forming substances in the line. An
additional provision has been added to clarify EPA's intent that streams
fluctuating above and below 10 percent VOC will be covered by the standards.
VOC is defined as any reactive organic compound. Organic compounds are
considered to participate in atmospheric photochemical reactions unless thay
are specifically designated by the Administrator as not participating in
atmospheric photochemical reactions. The following compounds are considered
nonTphotochemically reactive by EPA:
methane
ethane
1,1,1-trichloroethane
methylene chloride
trichlorofluoromethane
dichlorodifluoromethane
trifluoromethane
trichlorotrifluoroethane
dichlorotetrafluoroethane
chloropentafluoroethane
Quantities of these compounds present in the line may be excluded from the
total quantities of organic materials present in determining whether the
piece of equipment is covered by the standards.
1.1.10 Requirements for Units With No Equipment in VOC Service
EPA believes it appropriate to grant an exemption to any SOCMI unit
which does not process VOC. A few SOCMI process units may produce their
products without the use of VOC; however, these units are expected to be the
exception rather than the rule. SOCMI units which do not process VOC would
not have any potential to emit VOC. Therefore, a provision has been added
to the final standards which exempts an owner or operator of a facility
producing a chemical listed in 40 CFR 60.489 from the standards if that
1-11
-------
facility processes no VOC. SOCMI units which cease processing of VOC would
be able to show when VOC processing ceased by maintaining records of
processing operations.
1.1.11 Exemption of Beverage Alcohol Producers
During the public comment period, beverage alcohol producers said that
they should be exempt from coverage by the standards because beer and
whiskey producers were exempted from the priority list. EPA concluded that
process units within beer and whiskey plants that are producing fermented
beverages solely.for purposes of human consumption^should be exempt from the
standards. However, any process units (e.g., a distillation train to
produce industrial grade alcohols from fermentation products) in beer and
whiskey plants that are used to manufacture nonbeverage fermented products
are subject to the standards. Therefore, a specific exemption has been
included in the final regulation for beverage alcohol, producers.
1.1.12 Calibration Gas Requirements
There are two candidate calibration gases for the use of Reference
Method 21 in screening fugitive emission sources in SOCMI: hexane and
methane. Prior to proposal, hexane was specified as the calibrant for the
draft of Reference Method 21. At proposal, the calibration gas was changed
to methane because methane is more readily available in the 'required
concentration range and because SOCMI test data were gathered using methane
as a calibrant. The change was made in response to public comments on the
draft regulatory package.
During the public comment period for the proposed standards, other
public commenters objected to the change from hexane, saying the change
would mean that more leaks were detected. They also noted that the change
eliminated the feasibility of using photoionization monitors that would be
allowed by Reference Method 21.
EPA has considered the differences in the results which would be
obtained with the two calibrants and has found the differences
insignificant. The differences are in the same range as the variability
seen in repeated emission measurements from the same source. Data collected
using hexane and methane can be used interchangeably within ±30 percent at
1-12
-------
the action level. Therefore, because the differences are insignificant in.
terms of the long-term number of leaks detected and because the allowance of
hexane as a calibration gas will provide for the use of photoionization
monitors, EPA has changed the leak detection requirements to allow hexane as
an alternate calibration material. Methane may still be used where it is
the appropriate calibrant. In addition, EPA will likely use methane as the
calibrant for determining compliance with the alternative standards for
valves and for other such determinations.
1.1.13 Equivalency Provisions
The final standards have been changed to allow a vendor or manufacturer
to apply for equivalency for control systems or equipment (see
Section lll(h)(3) of the Clean Air Act). This change was made to increase ,
efficiency in the equivalency process and to provide SOCMI owners and
operators the incentive to purchase improved control systems and equipment
as they are developed. Even though the provision allowing vendors and
manufacturers to apply for equivalency have been added, it should be
remembered that equivalency determinations, where the effectiveness of the
alternative means of emission reduction depends on the owner or operator, .
not the vendor or manufacturer, must be submitted by the plant owner or
operator.
1.1.14 Production Rate Cutoff
Because the costs of the standards for process units with low
production rates are exorbitantly high in comparison to the emissions
reduction achievable, EPA provided an exemption for low production rate
units. Facilities with production rates of 1,000 Mg/yr or less are exempt .
from the standards. It is expected that, even though this cutoff will
exempt most R&D facilities, facilities producing on a semi-commercial or
commercial scale would still be covered.
1.1.15 Definition of "No Detectable Emissions"
The screening level associated with a "no detectable emissions" deter-
mination using the instrument testing techniques described in Reference
Method 21 has been revised to 500 ppmv in the final standards. The
standards for various fugitive emission sources, including no-leak equipment
1-13
-------
and pressure relief devices, require maintaining a condition of "no
detectable emissions." At proposal, the level associated with-this
condition was set at 200 ppmv, based on considerations of the calibration
procedures and instrument reading variability at low screening levels.
While no comments were received on the "no detectable emissions" level,
several comments were received on the instrument and calibration gas (see
Chapter 12). The reference method was subsequently revised to allow other
instruments and calibration gases to be used. With these changes in the
reference method, EPA provided additional latitude in its definition of "no
detectable emissions." Using five percent of the leak definition (which is
10,000 ppmv) as an allowance for calibration procedures and instrument
variability, the definition of "no detectable emissions" for the final
standards became a VOC concentration indicated by a screening value of less
than 500 ppmv above background concentration at the leak interface.
1.2 SUMMARY OF IMPACTS OF PROMULGATED ACTION
1.2.1 Alternatives to Promulgated Action
The regulatory alternatives are discussed in Chapter 6 of the
Background Information Document for the proposed standards. These
regulatory alternatives reflect the different levels of emission control.
They were used to help in selection of the best demonstrated technology,
considering costs, nonair quality health, and environmental and economic
impacts for fugitive emission sources in the synthetic organic chemical
manufacturing industry. These alternatives remain the same.
1.2.2 Environmental Impacts of Promulgated Action
Environmental impacts of the proposed standards are described in
46 FR 1136. Most of the changes in the standards (described above) will
have no effect on the environmental impacts ascribed to the standards.
However, the change in the standards for pumps will reduce the overall
emission reductions and costs of the standards. In addition, EPA has
revised the estimate of emissions of VOC to the atmosphere which will be
reduced by the standards. The revisions were made as a consequence of
revising methods of analysis and numerical estimates as described in the
Additional Information Document (AID).
1-14
-------
The standards will reduce fugitive emissions of VOC from process units
constructed, modified, or reconstructed between 1981 and 1985 from about
83,000 Mg/yr to about 37,000 Mg/yr. This reduction represents about a
55 percent decrease in emissions from the current industry baseline level of
emissions. The water quality and solid waste impacts are the same as those
presented in the Background Information Document (BID) for Proposed
Standards (III-B-1).
Based on the estimates of emissions, emission reductions, and costs of
controlling fugitive VOC emissions presented in the AID, the analysis of the
environmental impact in the fifth year after promulgation presented in the
proposal BID (IH-B-1) was revised. With these revisions noted, the
environmental impact analysis now becomes the final Environmental Impact
Statement for the promulgated standards.
1.2.3 Energy and Economic Impacts of Promulgated Action
Section 7.4 of the BID (III-B-1) describes the energy impacts of the
standards. The changes made in the standards have no effect on these
impacts. Chapters 8 and 9 of the BID (III-B-1) describe the cost and
economic impacts of the proposed standards. In general, there has been
little change to the cost and economic impacts of the standards since
proposal. With the revised cost analysis presented in the AID and the
changes in the level of control required by the standards, the cost and
economic impacts of the final standards are less than the impacts presented
at proposal. The net annualized cost (including recovery credits) will be
about $14.6 million, with a cumulative capital cost of about 544 million for
the five-year period considered. These costs are not expected to result in
industry-wide price increases.
1.2.4 Other Considerations
1.2.4.1 Irreversible and Irretrievable Commitment. Section 7.5.1 of
the BID (III-B-1) for the proposed standards concludes that the standards
will not result in any irreversible or irretrievable commitment of
resources. It was also concluded that the standards should help to save
resources due to the energy savings associated with the reduction in
emissions. These conclusions remain unchanged since proposal.
1-15
-------
1.2.4.2 Environmental and Energy Impacts of Delayed Standards.
Tables 1-1 and 7-7 in the BID (III-B-1) for the proposed standards summarize
the environmental and energy impacts associated with delaying promulgation
of the standards. The revised air and energy impacts are shown in
Table 1-1. The emission reductions and associated energy savings shown
would be irretrievably lost at the rates shown for each of the five years.
1.3 SUMMARY OF PUBLIC COMMENTS
A total of 56 letters commenting on the proposed standards and the
background information document for the proposed standards were received.
Comments from the public hearing on the proposed standards were recorded,
and a transcript of the public hearing was placed in the project docket. At
the request of some of the commenters, the comment period was extended to
allow them more time for review and comment. A list of commenters, their
affiliations, and the EPA docket item number assigned to their
correspondence is given in Table 1-2.
The comments have been categorized under the following topics:
Need for the Standards (Section 2)
Basis for the Standards (Section 3)
Control Technology (Section 4)
Applicability of Standards (Section 5)
Environmental Impact (Section 6)
Economics (Section 7)
Legal (Section 8)
Modification and Reconstruction (Section 9)
Equivalency (Section 10)
Recordkeeping and Reporting (Section 11)
Test Method (Section 12)
Enforcement and Compliance Concerns (Section 13)
Alternative Standards (Section 14)
Miscellaneous (Section 15)
The comments, the issues they address, and EPA's responses are
discussed in the following sections of this document.
1-16
-------
TABLE 1-1. IRRETRIEVABLE LOSSES WHICH WOULD OCCUR IF STANDARD IMPLEMENTATION WERE DELAYED
Year
1981
1982
1983
1984
1985
5-year Total
Emission Reductions Achievable
Under the Standards, Gg
8
17
26
36
46
133
Energy Value, of Emission
Reductions, Terajoules
124
264
403
558
713
2,062
Crude Oil Equivalent
of Emission Reductions,
Thousand Barrels
20
43
66
91
116
337
is the same as presented in Chapter 7 of the Background Information Document.
Based on 1.55 x 10 joules/kg ; this may be slightly over estimated if safety/relief valves and
closed vent systems are controlled by a flare system.
cBased on 5.8 x 106 Btu/bbl crude oil.
-------
TABLE 1-2. LIST OF COMMENTERS ON PROPOSED STANDARDS OF PERFORMANCE
FOR FUGITIVE EMISSION SOURCES IN SOCMI
COMMENTER AND AFFILIATION DOCKET ITEM No.
Mr. J.D. Martin IV-F-1
Chemical Manufacturers Association
c/o Union Carbide
Box 186
Port Lavaca, Texas 77979
Mr. D.E. Ellison IV-F-1
Synthetic Organic Chemicals Manufacturing Assoc.
c/o Virginia Chemicals
3340 W. Norfold Road
Portsmouth, Virginia
Mr. John T. Barr ? . IV-F-1, IV-D-28; IV-D-43
Air Products & Chemicals, Inc.
Box 538
Allentown, Pennsylvania 18105
Mr. A.M. Nickolaus IV-F-1, IV-D-40
Texas Chemical Council
1000 Brazos, Suite 200
Austin, Texas 78701
Mr. G. Wells IV-F-1
The Fertilizer Institute
1015 18th St., N.W.
Washington, D.C.
Mr. Glenn Hoffman IV-F-1
Hercofina
P.O. Box 327
Wilmington, N.C. 28402
Mr. R.B. Dickson IV-F-1
SOCMA Counsel
Cleary, Gottlieb, Steen & Hamilton
1250 Connecticut Avenue, N.W.
Washington, D.C. 20036
1-18
-------
TABLE 1-2. (CONTINUED)
COMMENTER AND AFFILIATION DOCKET ITEM No.
Mr. M.R. Keller IV-F-1
John Zink Company
4401 S. Peon a Street
Tulsa, Oklahoma 74105
Mr. Joseph C. Ledvina IV-F-1, IV-D-18
Conoco Chemicals
P.O. Box 2197
Houston, Texas 77001
Mr. T.A. Kittleman IV-D-1
Senior Engineer
E.I. DuPont de Nemours & Company
Wilmington, Delaware 19898
Mr. Louis R. Harris IV-D-2, IV-D-33
Marketing Manager
BS&B Safety Systems, Inc.
P.O. Box 45590
Tulsa, Oklahoma 74145
Mr. Bruce C. Grefrath, Manager IV-D-3
Environmental Affairs
Synthetic Organic Chemical Manufacturers Assoc.
1075 Central Park Avenue
Scarsdale, New York 10583
Mr. F.L. Piguet IV-D-4
Plant Manager
Allied Chemical
Post Office Box 761
Hopewell, Virginia 23860
Mr. J.F. Cooper IV-D-5
Vice President
Texaco Chemical Company
4800 Fournage Place
Bellaire, Texas 77401
1-19
-------
TABLE 1-2-.- (CONTINUED)
COMMENTER AND AFFILIATION DOCKET ITEM No.
Mr. Jerry M. Schroy IV-D-6, IV-D-36
Monsanto Company
800 N. Lindbergh Boulevard
St. Louis, Missouri 63166
Mr. A.G. Smith, Manager IV-D-7
Environmental Affairs
Shell Oil Company
P.O. Box 4320
Houston, Texas 77210
Mr. David W. Carroll IV-D-8
Assistant General Counsel
Chemical Manufacturers Association
2501 M Street, N.W.
Washington, D.C. 20037
Mr. Edwin M. Wheeler IV-D-9
President
The Fertilizer Institute
1015 18th Street, N.W.
Washington, D.C. 20036
Mr. James F. McAvoy IV-D-10
Director
State of Ohio Environmental Protection Agency
Box 1049
361 E. Broad Street
Columbus, Ohio 43216
Mr. Simon Feigenbaum IV-D-11
Air Pollution Engineer
Allegheny County Health Department
Bureau of Air Pollution Control
301 Thirty-ninth Street
Pittsburg, Pennsylvania 15201
Mr. Robert E. Jones IV-D-12
Technical & Environmental Affairs
The Quaker Oats Company
Merchandise Mart Station
P.O. Box 3514
Chicago, Illinois 60654
1-20
-------
TABLE 1-2. (CONTINUED)
COMMENTED AND AFFILIATION DOCKET ITEM No.
Mr. Richard J. Samelson, Manager IV-D-13
Environmental Programs
PPG Industries, Inc.
One Gateway Center
Pittsburgh, Pennsylvania 15222
Mr. T.C. Owen, Corporate Director IV-D-14
Office of Environmental Affairs
Union Camp Corporation
P.O. Box 1391
Savannah, Georgia 31402
Mr. D.P. Mykytiuk, Manager IV-D-15
Environmental, Health & Safety
ARCO Chemical Company
3801 West Chester Pike
Newtown Square, Pennsylvania 19073
Mr. D.B. Rathbun IV-D-.16
Vice President
American Petroleum Institute
2101 L Street, Northwest
Washington, D.C. 20037
'Mr. Edmund B. Frost IV-D-17
Vice President & General Counsel
Chemical Manufacturers Association, Inc.
2501 M Street, N.W.
Washington, D.C. 20037
Mr. .R.6. Dillard, President IV-D-17
Texas Chemical Council
c/o Shell Chemical Company
One Shell Plaza
Houston, Texas 77001
Mr. Paul J. Sienknecht, Manager IV-D-19
Environmental Regulatory Activities for Air
The Dow Chemical Company
2020 Dow Center
Midland, Michigan 48640
1-21
-------
TABLE 1-2. (CONTINUED)
COMMENTER AND AFFILIATION DOCKET ITEM No.
Mr. Herb Schuyten IV-D-20
Manager, Environmental Programs
Chevron U.S.A., Inc.
P.O. Box 3069
San Francisco, California 94119
Mr. J.C, Edwards, Manager IV-D-21
Clean Environment Program
Tennessee Eastman Company
Eastman Kodak
Kingsport, Tennessee 37662
Mr. John W. Drake IV-D-22
Staff Environmental Engineer
Kerr-McGee Corporation
Kerr-McGee Center
Oklahoma City, Oklahoma 73125
Mr. W.F. Blank, Manager IV-D-23
Pollution Control
Corporate Environmental Affairs
Allied Chemical
P.O. Box 2332R
Morristown, New Jersey 07960
Mr, James W. Tracht, Director IV-D-24
Energy and Environmental Affairs
Pennwalt Corporation
P.O. Box 0
900 First Avenue
King of Prussia, Pennsylvania 19406
Mr. J.J. Moon, Manager IV-D-25, IV-D-41
Environment and Consumer Protection Division
Phillips Petroleum Company
Bartlesville, Oklahoma 74004
Mr. Thomas A. Robinson, Director IV-D-26
Environmental Affairs
Chemicals Division
Vulcan Materials Company
P.O. Box 12283
Wichita, Kansas 67277
1-22
-------
TABLE 1-2. (CONTINUED)
COMMENTER AND AFFILIATION
DOCKET ITEM No.
Mr. Lawrence D. Vanell
Product Line Manager
Foxboro Analytical
Box 5449
South Norwalk, Connecticut
IV-D-27
06856
Mr. Jesse Coates
Chloromethanes Technology Center
Dow Chemical Company
Louisiana Division
Plaquemine, Louisiana 70764
Ms. Janet S. Matey, Manager
Air Programs
Chemical Manufacturers Association
2501 M Street, N.W.
Washington, D.C. 20037
IV-D-29
IV-D-30-, IV-D-31, IV-D-50,
IV-D-53
Mr. Edward Palgyi
Bureau of Air Quality Management
Florida Department of Environmental Regulation
Twin Towers Office Building
2600 Blair Stone Road
Tallahassee, Florida 32301
Mr. D.E. Park, Director
Environmental Affairs .
Ethyl Corporation
P.O. Box 341
Baton Rouge, Louisiana 70821
IV-D-32
IV-D-34
Mr. Henry L. Ramm
Environmental Engineer
Government and Regulatory Affairs Dept.
Rohm and Haas Company
Independence Mall West
Philadelphia, Pennsylvania 19105
Mr. V.J. Marchesani, Director
Energy & Environmental Quality
ICI Americas Inc.
Concord Pike & New Murphy Road
Wilmington, Delaware 19897
IV-D-35, IV-D-51, IV-D-52,
IV-J-3
IV-D-37
1-23
-------
TABLE 1-2. (CONTINUED)
COMMENTER AND AFFILIATION DOCKET ITEM No.
Mr. Ronald A. Lang IV-D-38, IV-D-38a
Executive Director
Synthetic Organic Chemical Manufacturers Assoc.
1612 K Street, N.W., Suite 308
Washington, D.C. 20006
Mr. Joseph Gordon, Director IV-D-39
Environmental Affiars
Lubrizol Corporation
29400 Lakeland Boulevard
Wickliffe, Ohio 44092
Mr. Russell W. Shannon IV-D-42
Associate General Counsel
Distilled Spirits Council of the
United States, Inc.
1300 Pennsylvania Building
Washington, D.C. 20004
Mr. J.S. Cerrito IV-D-44
Environmental Protection Operation
General Electric Company
One River Road
Schenectady, New York 12345
Mr. M.J. Rhoad IV-D-45
Managing Director
International Institute of Synthetic Rubber
Producers, Inc.
2077 South Gessner Road
Houston, Texas 77063
Mr. David D. Doniger IV-D-46
Senior Project Attorney
Natural Resources Defense Council, Inc.
1725 I Street, N.W.
Suite 600
Washington, D.C. 20006
1-24
-------
TABLE 1-2. (CONTINUED)
COMMENTER AND AFFILIATION DOCKET ITEM No.
Mr. Steven A. Tasher IV-D-47
Legal Department
E.I. DuPont de Nemours & Company, Inc.
Wilmington, Delaware 19898
Mr. Thomas V. Malorzo IV-D-48, IV-D-48a
Senior Regulations Analyst
Diamond Shamrock Corporation
717 North Harwood Street
Dallas, Texas 75201
Mr. J.L. McGraw, Chairman IY-D-49
Environmental Impact Committee
Rubber Division
American Chemical Society
P.O. Box 32960
Louisville, Kentucky 40232
Mr. W.F. Muller, Jr., Vice Chairman IV-J-1
Environmental Impact Committee
Southern Rubber Group
c/o The Goodyear Tire & Rubber Co.
P.O. Box 5397
Houston, Texas 77012
Mr. Kenneth E. Blower IV-J-2
Corporate Environmental Consultant
The Standard Oil Company
Midland Building
Cleveland, Ohio 44115
aThe docket number for this project is A-79-32. Dockets are on file at
EPA headquarters in Washington, D.C., and at the Office of Air Quality
Planning and Standards in Durham, N.C.
1-25
-------
An Additional Information Document (AID) (III-B-2) was published in
April, 1982. A notification of its availability and a request for comments
was published in the Federal Register (47 FR 19724, May 7, 1982). A list of
persons commenting on the AID, their comments, and EPA's responses are
presented in Appendix A of this document. Comments on the AID have been
categorized under the following topics:
Emission Factors (Section Al)
Model Units (Section A2)
Emission Reductions (Section A3)
Costs (Section A4)
Economics (Section A5)
Comments on Subjects not Covered in the AID (Section A6)
Previously Submitted Comments (Section A7)
1-.26
-------
2. NEED FOR THE STANDARDS
2.1 SIGNIFICANCE OF EMISSIONS AND ENVIRONMENTAL IMPACT OF THEIR CONTROL
Comment:
Commenters questioned the need for standards of performance for SOCMI
on the grounds that SOCMI emits small quantities of VOC (IV-F-1, No. 1;
IV-F-1, No. 2; IV-F-1, No. 3; IV-D-28; IV-D-26; IV-D-24; IV-D-23; IV-D-21;
IV-D-18; IV-D-12; IV-D-19; IV-D-17; IV-D-7; IV-D-34; IV-D-38; IV-D-40;
IV-D-47; IV-D-48; IV-D-50). One commenter (IV-F-1, No.2) said that EPA had
estimated total VOC emissions (including emissions from natural sources) at
39,100 Gg with stationary sources contributing 19,100 Gg. Emissions from
SOCMI were estimated at only 1,000 Gg, of which 300 Gg were fugitive VOC
emissions. Several commenters added that SOCMI fugitive emissions
represented a small percentage of total VOC emissions-
Contrasted with this comment, another commenter (IV-D-46) emphasized
the significance of VOC emissions from SOCMI (including fugitive emissions)
and the need to reduce these emissions. He said that expressing VOC
emissions from SOCMI as a percentage of total VOC emissions nationwide is
misleading since it dilutes the importance of the industry's emissions in
major nonattainment areas and elsewhere where SOCMI is concentrated. He
said that SOCMI plants make a substantial contribution to VOC emissions in
major nonattainment areas. VOC emissions come from many diverse types of
sources and each source category may account for only a relatively small
percentage of emissions. But all these sources need to be controlled since,
together, they account for much VOC.
Another commenter (IV-D-47) supported the implementation of appropriate
regulatory programs to minimize fugitive emissions for new and modified
facilities pursuant to Section 111 of the Clean Air Act. However, he
challenged EPA's conclusion that fugitive emission leaks from SOCMI
facilities cause or contribute significantly to air pollution which may
reasonably be anticipated to endanger public health or welfare.
2-1
-------
Response:
Emissions of VOC from SOCMI represent a significant source of VOC
emissions to the atmosphere. EPA estimates that 540 Gg of VOC/year
(540,000 Mg/yr) of VOC are emitted to the atmosphere from all sources in
SOCMI (IV-B-24). This estimate of emissions is based on detailed studies of
individual process source types including air oxidation processes, distilla-
tion operations, storage operations, carrier gas processes, equipment leaks,
and secondary sources. Because VOC emissions come from many, diverse source
categories, each source category contributes a relatively small percentage
to the large overall total. Because of this diversity, the relevant figures
to consider are the total emissions, not emissions expressed as a percent.
540 Gg of VOC/year is a significant quantity of VOC to be emitted as air
pollution. This quantity is large in absolute terms and is large relative
to other VOC source categories.
The SOCMI source category ranked first on the Priority List,
40 CFR 60.16 (44 FR 49222, August 21, 1979), of 59 major source categories
for which standards of performance are to be promulgated by 1982. The
Priority List consists of categories of air pollution sources that, in EPA's
judgment, cause or contribute significantly to air pollution which may
reasonably be anticipated to endanger public health or welfare. In
developing the priority list, major source categories were ranked according
to three criteria specified in section lll(f) of the Act: (1) the quantity
of emissions from each source category, (2) the extent to which each
pollutant endangers public health or welfare, and (3) the mobility and
competitive nature of each stationary source category.
The commenters expressing concern over new source standards for SOCMI
have not presented any new information which would change the decision
ranking SOCMI on the Priority List. Therefore, standards of performance
will be promulgated for the SOCMI source category. As discussed in the
response to the next comment, the decision to regulate the fugitive emission
source subcategory of SOCMI is based not on the significance of the contri-
bution of fugitive emission sources (although that contribution is indeed
large), but rather on EPA's ability to identify the best demonstrated
technology (considering costs) for SOCMI fugitive emission sources.
2-2
-------
Comment:
Commenters stated that the total mass of VOC emissions would not be
reduced significantly as a result of implementation of the standards and
there was, therefore, no need for the standards (IV-D-2.4; IV-D-38; IV-D-18;
IV-D-17; IV-D-19; IV-F-1, No. 2). Several commenters (IV-D-24; IV-D-18;
IV-D-17; IV-D-19; IV-F-1, No. 2; IV-F-1, No. 3) pointed out (1) that VOC
emissions from stationary sources would be reduced by less than 1 percent in
1985, (2) that total VOC emissions would be reduced by only 0.4 percent if
the standards were implemented, and (3) that recent SOCMI data indicated
that VOC emissions reductions due to the proposed standards would be only
about 0.15 percent of the total VOC emissions in 1985. They questioned the -
need for a standard which would reduce emissions of VOC by only one tenth of
the total VOC emissions on.a national level. Another commenter (IV-D-47)
referred to a recent EPA report (EPA-450/3-80-023) which shows the reduction
expected in 1982 from promulgation of the proposed standards to be only
0.25 percent.
Response:
Since SOCMI has been listed as a significant contributor to air
pollution under Section lll(f), EPA must promulgate standards of performance
for those new sources within this source category for which the EPA can
identify the best demonstrated technology (considering costs). The amount
of emission reduction achievable is plainly an important factor in
identifying best demonstrated technology (considering costs). It is.
conceivable that for certain source categories or subcategories there may be
no technology that achieves emission reductions at reasonable costs. In
such a case, EPA would not be required to establish standards under
Section 111 for those groups of sources. By contrast, EPA has identifed
several alternative systems of control capable of achieving additional
emission reduction at reasonable cost at SOCMI fugitive emission sources.
EPA must therefore establish standards based on the most effective of those
systems.
Although the specific bases for comparison of the numbers cited by the
commenters are unclear, they are comparing large numbers to still larger
2-3
-------
numbers which results in percentages which are small. In addition, the
commenters are comparing emission reductions to emissions. By 1985 up to
approximately 830 newly constructed, modified and reconstructed facilities
would be affected by these standards. These facilities would contribute an
additional 83 Gg/yr of VOC to the atmosphere if left uncontrolled. These
numbers represent large quantities of organic material being emitted into
the atmosphere where ozone is formed. By implementing the final standards,
approximately 46 Gg/yr of emission reduction is achieved. This represents a
56 percent decrease in uncontrolled fugitive emissions.
In summary, the standards of performance for SOCMI fugitive emission
sources of VOC serve the intent of Section 111 of the Clean Air Act. They
minimize VOC emissions at a reasonable cost. The standards also fulfill, in
part, the requirements of Section lll(f) by providing standards for the
SOCMI source category for which standards must be promulgated.
Comment:
Two commenters (IV-F-1, No. 2; IV-D-38) said that emissions from small
manufacturers in SOCMI were insignificant and that reductions in this small
amount of emissions would also be insignificant. The commenter said that
the relative contribution to VOC fugitive emissions from new and modified
facilities by 1985 would be only 14.2 percent for Model A plants,
36.2 percent for Model B plants, and 49.5 percent for Model C plants. The
commenter interpreted these figures to mean that 52 percent of the affected
facilities would contribute less than 15 percent of the emissions. This
argument was used to support the contention that small businesses should not
be covered by the standards.
Response:
About 10 to 30 percent of SOCMI process units are owned by small
businesses. However, there is no known relationship between small
businesses and process units which are similar to Model Unit A. The three
different types of model units will likely be owned by any size business.
As a consequence, the level of VOC emissions from process units owned by
small businesses is no different than the level of VOC emissions from
process units owned by other businesses. As shown previously, emissions of
2-4
-------
VOC from SOCMI units are large. Small businesses which own and operate
SOCMI units are a part of this industry and, as such, contribute to the air
pollution which EPA considers to be significant. Therefore, model-unit
size, emissions, or emissions per process unit do not provide an objective
basis for exempting small businesses.
The only objective basis for exempting small businesses from the
standards would be to decide whether the level of the standards is appro-
priate for small businesses that is, whether best demonstrated technology
(considering costs) is available for small businesses. . In making this
decision, EPA must consider whether the control technology is demonstrated
and whether the costs are reasonable for small businesses. As discussed
elsewhere in this document, EPA has concluded that the control technology is
demonstrated (Section 4) and that the costs are reasonable for small
businesses (Section 7). Therefore, EPA has not provided an exemption from -
the standards for small businesses.
An exemption is provided where costs are exorbitant in comparison to
the emission reduction benefit achieved. As explained in Section 5.5, an
exemption is provided for process units with low production volumes. This
exemption is based on a cost that is unreasonably high in comparison to the
VOC emission reduction achieved for low production volumes. Some small
businesses may be exempted if they own process units that qualify for the
low production volume exemption.
Comment:
One commenter (IV-D-42) wrote that emissions from natural alcohol
fermentation processes were low and that natural alcohol plants should not
be regulated.
Response:
SOCMI consists of numerous processes that produce numerous chemicals
each of which may by itself emit a relatively small amount of VOC. The
total amount of VOC emissions contributed by SOCMI is, however, significant.
The fact that one process in the industry contributes a small amount of the
total does not provide a basis for exempting that one isolated part, e.g.,
anhydrous alcohol. As before, the only basis for exempting processes from
2-5
-------
the standards would be to decide whether the control technology is
demonstrated and has reasonable costs. Based on this type of analysis, EPA
has provided an exemption for processes w.ith low production volumes.
However, EPA has concluded that the control technologies which form the
basis of the standards are demonstrated and that they have reasonable costs
for SOCMI process units, including anhydrous alcohol process units.
Comment:
One commenter (IV-D-50) stated that uncontrolled SOCMI fugitive
emissions might well approach the regulatory goal of 26 Gg/yr and thus no
standards would really be needed. He made an estimate of uncontrolled
emissions of 55 Gg/yr based on data collected recently in SOCMI. He
considered the estimate low because it was based on data from plants in the
high-leak category, and SOCMI is largely comprised of low-leak and non-leak
processes.
Response:
EPA's intent is not to set a regulatory goal of 26 Gg/yr. Rather, the
intent is to develop standards based on the use of the best demonstrated
technology (considering costs) [BDT]. At proposal, EPA estimated that use
of BDT on affected facilities would limit emissions of VOC in 1985 to
26 Gg/yr. Since proposal, the estimate has been revised to 36 Gg/yr. This
is the estimated effect of the standards five years after proposal of the
standards. However, as discussed in other responses, the purposes of these
standards and Section 111, in general, look beyond any specific effect
within a short time period Tike five years.
With regard to the specific data cited by the commenter, the high leak
category of processes referenced by the commenter was presented in the SOCMI
Data Analysis report (IV-A-14). The category was one of three artifically
derived categories that were derived for purposes of statistical analysis.
The categories do not reflect distributions of leak frequencies among the
process units in SOCMI. However, as discussed in the section on Alternative
Standards, EPA has provided alternative standards which would establish BDT
for low-leak process valves and provide an incentive to design and operate
process units which have low-leak frequencies. In addition, owners of
2-6
-------
process units capable of achieving emission reductions at least equivalent
to BDT may apply for a determination of equivalency.
Comment:
One commenter (IV-D-46) wrote that emissions of volatile organic
compounds (VOC) from SOCMI are important contributors to levels of ozone and
other photochemical oxidants. In many nonattainment areas, meeting the
health standard for ozone requires effective control of these emissions. He
felt that this standard will provide a meaningful floor level of emissions
control a reasonable starting point for analysis of the need for
additional controls under state and federal nonattainment programs. He
added that NSPS for these emissions is also needed to prevent deterioration
of air quality in the large number of areas now nearly violating the health
standard.
This commenter said that control of VOC from this industry will also
help curb particular chemicals which are known or suspected to cause cancer
and other serious long-term illnesses. Here too, an NSPS for this industry
will provide a floor level of control, to which further hazardous air
pollutant controls can be added.
Response:
EPA agrees with this commenter.
Comment:
One. commenter (IV-D-46) requested that EPA calculate the contribution
of SOCMI plants to VOC emissions in the major nonattainment areas for ozone.
The commenter stated that this should be easy to do from readily available
information. He also stated that this calculation would show that the
effect of fugitive emissions from this industry is considerably greater than
the one percent national average figure indicates. The commenter emphasized
that it would make it even clearer that fugitive emissions from SOCMI plants
are a significant source of VOC emissions which need to be controlled.
Response:
EPA has made no calculations of ozone contribution due to VOC emissions
because of uncertainties in new plant locations and currently available
models, and because the formation of ozone and oxidant depends on meteorolo-
gical and topographical factors as well as chemical reactivity. However,
2-7
-------
for purposes of new source standard preparation, it is sufficient to know
that VOC emissions contribute to ozone formation and that control of VOC
emissions will reduce ozone formation. These facts have been documented in
Docket No. IV-A-17.
2.2 CONTRIBUTION OF CURRENT REGULATIONS TO FUGITIVE EMISSION CONTROL
Comment:
Several commenters questioned the need for the standards because, in
their opinions, fugitive emissions of VOC in this industry are adequately
controlled by other regulations. They stated that other environmental
regulations and Occupational Safety and Health Administration regulations
adequately reduce fugitive emissions of VOC.
Comment 1. Several commenters questioned the need for the standards in
light of current environmental regulations imposed on the industry (IV-F-1,
No.l; IV-F-1, No.3; IV-D-23; IV-D-22; IV-D-19; IV-D-20; IV-D-18; IV-D-17;
IV-D-47). The commenters asserted that fugitive emissions of VOC from SOCMI
are now or will be brought under control by such regulatory efforts as State
Implementation Plans (SIP's), the Control Techniques Guideline (CTG) for
SOCMI fugitives, National Emission Standards for Hazardous Air Pollutants
(NESHAP's), prevention of significant deterioration (PSD) requirements, and
lowest achievable emission rate (LAER) requirements. One of the commenters
(IV-D-22) pointed out that the CTG includes more industrial process unit
types and applies to both new and existing units,.making the new source
performance standard (NSPS) unnecessary. Another (IV-D-17) said that
reasonably available control technology as defined by States (RACT) would
achieve almost the total reductions expected under the NSPS and that the
incremental cost and emissions reductions between the two levels were unjus-
tifiable. Another commenter (IV-D-19) said that PSD and non-attainment
requirements, both of which apply to new sources, would accomplish the same
emission reductions. He explained that the Clean Air Act requires that best
available.control technology (BACT) for attainment area sources and LAER for
nonattainment area sources must be at least as stringent as NSPS and that
BACT and LAER may even be more stringent and are set on a case-by-case
basis. One commenter (IV-D-17) said that the State Implementation Plans
2-8
-------
have been designed to meet the National Ambient Air Quality Standards
(NAAQS) for ozone. With the NAAQS achieved, he could see no need for
additional regulations.
Comment 2. Other commenters stressed the potential overlap of OSHA
regulations with the proposed standards (IV-D-6; IV-D-12; IV-D-38; IV-D-32).
One of the commenters (IV-D-6) asserted that OSHA regulations for control of
contaminant levels in the workplace are in fact fugitive emission control
regulations. Pointing out that OSHA regulations require engineering
controls and work practices unless such methods are unavailable, he said
that the engineering controls considered by OSHA are identical to those
described by EPA in the BID (EPA 450/3-80-033a). Another commenter
(IV-D-12) said that many of the chemicals listed in Section 60.489 are
listed in Table Z-l of OSHA regulation 29 CFR. The levels required by the
OSHA regulations were, he said, lower than 10,000 ppmv in most cases.
Citing a particular example from his company's furfural plant, he said that
three of the four products or byproducts produced in the furfural process
unit were closely controlled by OSHA. All of the four were controlled by
industrial hygienists in routine surveillance. Particular attention was
said to be paid to leak-prone areas. In all cases concentrations were in
the range of 0.1 to 15 ppmv for compounds which vary widely in volatility.
A third commenter (IV-D-38) pointed out that nearly one-half of the organic
substances subject to the proposed NSPS are already subject to OSHA require-
ments that result in most new or modified plants being designed to control
emissions to low levels. The commenter continued by citing a survey in
which thirty-four companies reported producing 183 out of the 358a chemicals
listed. Most of the 183 were listed in the 1980 bulletin of the American
Council of Government Industrial Hygienists (ACGIH). He further explained
that many of the chemicals listed by ACGIH have very low volatility and,
therefore, pose little emission risk. Based on the survey and the ACGIH
listings, the commenter felt that the proposed NSPS would do little more
than impose unnecessary operational requirements on plants already designed
to minimize emissions.
aThe list actually contains 378 entries. The count of 358 may have been
a typographical error.
2-9
-------
Response:
Comment 1: Background. These commenters are referring to four types of
environmental regulations: National Ambient Air Quality Standards (NAAQS),
State Implementation Plans (SIP's), National Emission Standards for
Hazardous Air Pollutants (NESHAP's) and New Source Performance Standards
(NSPS). These commenters also are referring to two types of occupational
standards. Each of these statutory programs play a uniquely different role
in meeting the goals of the Clean Air Act and the Occupational Safety and
Health Act. The main thrust of these comments is that fugitive emissions of
VOC within SOCMI do not significantly contribute to air pollution, and that
there is no need for the standards of performance. However, none of these
statutory programs negates the need for new source standards, as explained
below.
National Ambient Air Quality Standards (NAAQS) (Section 109 of the
Clean Air Act) set a ceiling for public exposure to criteria pollutants
(S09, NO , ozone, CO, particulates, lead) by establishing an ambient
L. A
concentration level that must not be exceeded more than one time anywhere in
the United States. States implement plans (Section 110 of the Clean Air
Act) to attain NAAQS. Based on projections that show attainment of NAAQS,
States determine the degree of control that will be imposed on existing
sources and on new sources, depending on whether air quality is better than
or worse than the NAAQS in the area where the source is or will be located.
State Implementation Plans (Section 110 of the Clean Air Act) (SIP's)
are required to include a program for preconstruction review of new or
modified stationary sources to ensure that such sources do not interfere
with attainment or maintenance of the NAAQS. This statutory program is set
forth in Parts C and D of Title 1 of the Clean Air Act. Part C, "Prevention
of Significant Deterioration of Air Quality" (PSD), is for areas of the
country that have attained the NAAQS. The PSD rules require certain new
sources to meet the "best available control technology" (BACT) considering
energy, environmental, and economic impacts and other costs. This type of
emission limitation must be determined on a case-by-case basis. In no event
may the application of BACT result in emissions of any pollutants which will
2-10
-------
exceed the emissions allowed by applicable standards of performance
established pursuant to Section 111 (or 112) of the Clean Air Act.
Part D, "Plan Requirements for Nonattainment Areas" applies to areas of
the country that have not attained the NAAQS. Most existing sources in non-
attainment areas are required to install, at a minimum, "reasonably avail-
able control technology" (RACT). RACT is defined by the State typically
with reference to a Federal control techniques guideline document (CTG).
Certain sources in nonattainment areas are required to install the control
equipment that will result in the lowest available emission rate (LAER) as
defined by the State. Applicable costs do not necessarily play as prominent
a role in determining LAER as they do in determining BACT. The Clean Air
Act defines LAER as that rate of emissions based on the following, whichever
is more stringent:
(A) The most stringent emission limitation which is contained in the
implementation plan of any State for such class or category of
source, unless the owner or operator of the proposed source
demonstrates that such limitations are not achievable, or
(B) The most stringent emission limitation which is achieved in
practice by such class or category of source. In no event can the
emission rate exceed any applicable new source performance
standard.
CTG's provide guidance to States in developing RACT-based environmental
regulations. These regulations are established to correct existing air
pollution problems and affect existing sources in particular. The draft CTG
entitled "Control of Volatile Organic Fugitive Emissions from Synthetic
Organic Chemical, Polymer, and Resin Manufacturing Equipment" was presented
to the National Air Pollution Control Techniques Advisory Committee on
April 30, 1981. This CTG, once it is published in a final form, will
represent Federal guidance to States for RACT-based provisions applicable to
SOCMI facilities. The CTG discusses control techniques which are completely
compatible with the techniques considered for the SOCMI NSPS.
National Emission Standards for Hazardous Air Pollutants (NESHAP's), as
mandated under Section 112 of the Clean Air Act, are distinctly separate
2-11
-------
from NAAQS or standards of performance. NESHAP's were developed to control
pollutants that are hazardous because they are carcinogens or cause other
serious diseases. Some of the individual SOCMI chemicals have been
identified as hazardous pollutants and some SOCMI units may be affected by
NESHAP regulations. However, SOCMI fugitive emissions as a class have not
been identified as hazardous pollutants and, therefore, are not subject to
NESHAP's.
Response:
Comment 1. Standards of performance required by Section 111 play a unique
role under the Clean Air Act. The main purpose of standards of performance
is to require new sources, wherever located, to reduce emissions to the
level achievable by the best technological system of continuous emission
reduction considering the cost of achieving such emission reduction, any
nonair quality health and environmental impact, and energy requirements. . .
[(Section lll(a)(l)]. Congress recognized that establishing such standards
would minimize increases in air pollution from new sources, thereby
improving air quality as the nation's industrial base is replaced over the
long-term. NSPS's thereby serve as a distinct means of achieving the Act's
goals, supplementing the role played by RACT-based requirements for existing
and new sources within state implementation plans developed for the purpose
of attaining the NAAQS.
Where RACT-level control is already in place, however, the impact of
NSPS will be smaller than calculated. RACT and the systems chosen as the
best demonstrated technology for this industry's standards of performance
for new stationary sources are not conflicting types of control; therefore,
where RACT already applies, the standards of performance will supplement
RACT-level control. EPA has determined that existing RACT-level facilities
that become subject to the standards of performance (e.g., through modifica-
tion) can achieve the additional reduction required at a reasonable cost.
Congress also intended NSPS to play an integral role in the new source
review programs of the Act. Standards of performance required by
Section 111 also serve as the minimum level of emission control for BACT and
LAER, which are determined case-by-case. Promulgation of these standards
2-12
-------
therefore assures that BACT and LAER for individual sources are not less
stringent than the "best demonstrated technology" for the class of sources
into which those individual sources fall. Absent identification of "best
demonstrated technology" through promulgation of NSPS's, BACT and LAER might
be less stringent than BDT-level control.
Standards of performance have other benefits in addition to achieving
emissions reductions. Standards of performance establish a degree of
national uniformity to air pollution standards, and, therefore, preclude
situations in which some States may attract new industries as a result of
having relaxed standards relative to other States. Further, standards of
performance provide documentation that reduces uncertainty in evaluations of
available control technology. This documentation includes identification
and comprehensive analyses of alternative emission control technologies,
development of associated costs, evaluation and verification of applicable
emission test methods, and identification of specific emission limits
achievable with alternate technologies. This documentation also provides an
economic analysis that reveals the affordability of controls in a study of
the economic impact of controls on an industry.
After EPA considered the statutory requirements and concluded that
SOCMI is a significant contributor to air pollution, within the meaning of
Section 111, standards for fugitive emission sources of VOG within SOCMI
were selected. These standards are based, as required by Section 111, on a
demonstrated system of continuous emission reduction considering costs,
nonair quality health and environmental impacts and energy requirements.
The selection of the standards was based on technological, cost, energy, and
environmental factors.
EPA was aware of other Clean Air Act programs during preparation of the
standards. As discussed in the next response, the level of control under
existing environmental regulations was considered in estimating emissions
from SOCMI. Further, as discussed in the next section, the selection of the
final standards was based on a comparative analysis of the incremental costs
and emission reductions for the different levels of control considered. It
should be noted that one of the regulatory alternatives considered
2-13
-------
(Regulatory Alternative II) was very similar to the draft CTG. The
existence of other environmental regulations was considered during selection
of BDT, but their existence does not lead EPA to conclude that standards
reflecting better control technology cannot be applied at reasonable costs.
Response:
Comment 2. The commenter also referred to various occupational standards.
Many of the chemicals listed in Section 60.489 of the SOCMI fugitive
emission standards are also listed in Table 2-1, Toxic and Hazardous
Substances, in the general provisions for the Occupational Safety and Health
Act (OSHA) (29 CFR 1910.1000), and some of these .chemicals are also covered
by more specific health standards under OSHA and may be listed by other
groups such as the American Council of Government Industrial Hygienists
(ACGIH). As a consequence, the SOCMI standards and the OSHA standards have
the potential to impact the same fugitive emission sources. However, the
SOCMI standards of performance do not conflict with OSHA standards; they
ensure emission reductions, and they do not require duplicate equipment. In
fact, in many cases they may be supplementary.
Standards of performance and OSHA regulations have different purposes
and may result in different environmental benefits. New source standards
serve to limit directly mass emission rates. In contrast, implementation of
OSHA standards for toxic chemicals does not necessarily limit emissions
directly. Under OSHA, control of emission sources.may include substitution
with less hazardous materials, process modification, worker rotation,
process or worker isolation, ventilation controls, or modification of work
practices. These controls may reduce occupational exposures, but they do
not necessarily reduce the mass rate of VOC emissions to the atmosphere.
Furthermore, the OSHA regulations would require control to different
concentration levels, depending on the toxicity of a specific chemical,
while NSPS regulations would require emission control based on BDT for all
VOC. Fairly high emission rates from fugitive emission sources may be
diluted to the extent necessary to protect workers, but the emissions would
still be released to the atmosphere, adding to the air pollution burden.
Relying on indirect controls that may or may not reduce emissions that would
2-14
-------
degrade air quality would be an unreasonable approach to reducing fugitive
emissions of VOC. .
One commenter had stated that many chemicals listed by ACGIH have low
volatility and thus pose little emission risk. The NSPS for SOGMI fugitive
VOC emissions considers the volatility of the SQCMI chemicals. Standards
apply to sources that have a great tendency to leak, and exemptions are
provided for these sources that have a low potential to emit VOC. Data from
fugitive emission studies show that more volatile chemicals (those with
higher vapor pressures) have a greater tendency to leak and at a higher mass
emission rate. Thus, fugitive emission sources are classified in the
standards by volatility (vapor pressure) service: in heavy liquid,-in-light-
liquid, or in gas or vapor services. These data on which the classifica-
tions were developed were collected in process units that are currently
subject to OSHA regulations.
The commenter gave examples of chemicals which were controlled by OSHA
to concentration limits well below 10,000 ppmv. Many of the chemicals
j
listed in Table Z-l of the OSHA regulations do have concentration limits
under 10,000 ppm. However, those concentration levels are not comparable to
the screening value concentration levels measured with Reference Method 21.
The screening value concentrations are obtained at the leak surface. The
concentration levels required by TLV's are time-weighted average concen-
trations in the workers' air. As noted before, a fairly large emission rate
can be diluted sufficiently to obtain workplace air within certain limits.
The magnitude of dilution effects can be seen in measurements of various
distances from a leak surface. An experiment of this type was reported in
Evaluation of the Maintenance Effect On Fugitive Emissions from Refineries
in the South Coast Air Quality Management District (IV-A-30). The data
presented in this report show the dramatic effect of moving just 20 cm from
the leak surface for pump and compressor seals. All but a few leaks of more
than 10,000 ppmv measured at the surface could no longer be detected at a
distance of 20 cm from the surface and area monitors were typically much
further removed than 20 cm.
Even though the standards affect the same sources, the SOCMI standards
of performance do not require duplicate equipment. The substantive
2-15
-------
equipment or engineering controls that can be required by OSHA standards are
not duplicated by the standards of performance. The same equipment, even
though it may be employed to meet both standards, would not be installed in
duplicate. More importantly, the standards may be supplementary. In many
cases the fugitive emission standards may facilitate compliance with the
OSHA regulations and vice versa. By keeping chemical substances out of the
air, the workers' health and safety and the public welfare are protected.
The NSPS would apply to newly constructed, modified, and reconstructed
facilities. If the two standards require different levels of control and
apply to the same source, an owner of the source would meet the most
stringent standard which would insure that both standards are met. This
choice would be easily made by an owner or operator. In addition, to the
extent that an OSHA standard results in low leak frequency or low emission
rates, the NSPS provides alternative standards which would provide an
incentive to design and operate such process units. For example, if an
owner uses engineering controls for valves (such as sealless valves) in a
process unit and the unit has few valves which leak, then an owner could
select one of two alternative standards for valves which would substantially
reduce the effort required to comply with the standards for valves (see
Section 14). In addition, if an owner uses engineering controls for pumps
that result in low emission rates (such as dual mechanical seals and a heavy
liquid barrier fluid system), an owner would have a pump that would be
exempt from the routine leak detection and repair requirements of the
standards. An owner could select which of the two standards to meet.
Another potential area of conflict may arise in the leak detection and
repair programs. Leak detection and, especially, repair may require workers
to complete tasks near VOC emission sources. Exposure for these workers
could be increased. Work practices including provisions for insuring that
employees work upwind from any leaks would be sufficient to control
exposures during repair of leaking equipment^" In some cases personal
protective equipment may be required. However, this type of situation would
occur from time to time whether the standards were in effect or not. The
same practices could be used during leak detection and repair as are used
during routine plant maintenance and repair.
2-16
-------
In summary, the OSHA regulations and the standards of performance may
impact the same fugitive emission sources. However, the two regulations
have different purposes and have different environmental effects. The
standards are not duplicative, nor is there a conflict between them.
Comment:
Two commenters (IV-F-1, No.2; IV-D-28) said that EPA had failed to
consider adequately the effects of other regulations affecting SOCMI when
determining the costs and benefits of the proposed standards.
Response:
EPA has considered the costs and benefits of other regulations in
developing the promulgated standards. Cost considerations are discussed in
Chapter 8 of the BID and in Section 7 of this document. Benefits of other
regulations are discussed here.
The benefits of other regulations were considered in establishing the
baseline level of control. The baseline was established by examining the
actual level of control which exists in the industry. The actual level of
control reflects the effects of all circumstances acting to affect control,
whether they are regulatory (e.g., OSHA regulations) or economic circum-
stances. The only relevant circumstances are regulatory circumstances that
control release of VOC into the atmosphere from fugitive emission sources.
As discussed above, there are two potential types of regulations which have
the potential to accomplish this control: air pollution regulations and
Occupational Safety and Health Administration regulations.
SOCMI units that are located in areas currently attaining the NAAQS for
ozone probably would not be subject to any VOC fugitive emission regula-
tions. Facilities located in nonattainment areas would be subject to
applicable state implementation plans (SIP's). However, only a few states
have developed or are considering near-term development of these specific
regulations. NESHAP's are being developed for vinyl chloride and benzene,
but these standards apply to only a small portion of SOCMI. For the most
part, there are currently no environmental regulations applicable to
fugitive emission sources in SOCMI.
In addition to environmental regulations, OSHA standards and provisions
have the potential for effecting control of fugitive emissions. There does
2-17
-------
not seem to be a clear correlation between leak frequencies and chemicals
controlled by OSHA. Analysis of TLV's and leak frequencies found in SOCMI
showed no recognizable trend. However, there might be a situation in which
OSHA programs effectively control fugitive emissions. Where OSHA standards
achieve what the NSPS would require, no additional control would be
necessary. Where OSHA standards would achieve somewhat less control,
fugitive emission standards would serve as an important supplement. In
addition, where an OSHA standard results, as discussed in the previous
response, in low leak frequencies or low emission rates the NSPS provides
relief from any overlap in requirements. Thus, since the NSPS is written in
a manner which allows flexibility in the approach to control, both
regulatory aims can be accomplished without conflict.
Provisions of insurance policies for fire and .explosion protection also
have the potential for effecting control of fugitive emissions. Lately,
economics have also been a factor contributing to the control of fugitive
emissions because of the increase in prices for petroleum-derived products.
However, as noted previously, the emission estimates for baseline are based
on the current status of fugitive emission sources which already reflect
these impacts.
Comment:
One commenter (IV-D-32) said that if the material is hazardous, the
Toxic Substances Control Act (TSCA) could control it.
Response:
Even though several of the VOC's covered by the standards are toxic, as
discussed in a previous response, fugitive emissions of VOC from SOCMI have
not been declared hazardous as a class under TSCA. TSCA requires that
substances presenting a risk to man or his environment by virtue of their
toxicity be controlled by other statutes if possible. NESHAP's developed
under Section 112 of the Clean Air Act would control emissions of those
particular pollutants determined hazardous under that section. NSPS's,
however, apply more broadly to new sources in industrial categories deter-
mined to contribute to air pollution which may reasonably be anticipated to
endanger public health or welfare. This application includes significant
2-18
-------
contributors of all types of VOC emissions. For this reason, EPA has listed
SOCMI sources for regulation under Section 111.
Comment:
One commenter (IV-D-12) said that control of odors had caused control
so stringent in'his plant that the quantities of VOC in the air are too
small to be identified. Even though the streams are so dilute, they have
been subjected to special incinerators which consume all organics.
Response:
Many organic compounds have offensive odors and some have extremely low
odor thresholds. A detectable odor is a sign of VOC in the air, even though
in cases of chemicals with low odor thresholds the concentration may be very
low. On the other hand, even a low concentration, by the time it reaches
the public, indicates an emission rate of some magnitude.
Odor control and VOC emissions control may accomplish comparable
emission reductions. Also, the control techniques applicable are virtually
the same. Therefore, compliance with the fugitive emission standards should
aid an owner or operator in his efforts to control, odor; although, for some
chemicals odor control may require even more stringent measures. In
addition, where odor control results in low leak frequencies and low
emission rates, the NSPS, as discussed in a previous response, eliminates
the burden from any overlap in control programs.. .
Comment:
One commenter (IV-D-42) wrote that it would be a waste of government
funds and enforcement resources to have EPA regulate the fermentation
alcohol industry. He argued that this industry is already regulated
stringently by the Bureau of Alcohol, Tobacco and Firearms (BATF) which has
rules requiring distillers to control leaks and spills and account for all
product in order to protect the federal revenue. The commenter cited
27 CFR, Part 19, Subpart I in support of his argument. He also noted that
BATF rules require that all tanks, pipes, valves, and other equipment used
for the production, storage or handling of alcohol be constructed not only
to prevent leaks and spills, but also to prevent plant personnel from
gaining access to any alcohol. The commenter concluded that EPA would.be
duplicating what BATF already does.
2-19
-------
Response:
The regulations in 27 CFR, Part 19, Subpart I require equipment be
installed and operated in a manner which protects the revenue derived from
spirits. The purpose of the regulation is, thus, different from the
standards of performance which are aimed at protecting the environment. In
27 CFR, Part 19 Subpart I there are no provisions specifically relating to
fugitive emission sources. But, protecting the environment by preventing
leakage from fugitive emission sources might provide protection for
revenues. The standards are not duplicative, nor are they conflicting. In
fact, they are complementary because attainment of two different sets of
goals may be facilitated by the same measures.
2.3 STANDARDS' BENEFIT TO PUBLIC HEALTH; CONTROLLING VOC TO.CONTROL OZONE
Comment:
Several commenters questioned the need for the standards to protect the
public health and welfare. One commenter (IV-F-1, No.3) said that the
ambient air quality standard for ozone which is set to protect the public
has recently been raised. He concluded that because most of the country is
in compliance with the ozone standard, the public health and welfare is
protected and the standards of performance are unnecessary.
The commenter (IV-F-1, No.3) continued his argument that the standards
would have no beneficial effect on the public health and welfare by consid-
ering the two cases of people living in attainment areas and those living in
nonattainment areas. He said that those persons residing in attainment
areas would derive no benefit because the ozone levels are below the level
requisite to protect public health and welfare. In considering the other
case, he said that those residing in nonattainment areas would not benefit
because of the insignificance of the emissions being controlled when
compared to the other sources present. He repeated this argument in a set
of written comments (IV-D-28).
In a similar vein he continued by saying that 24 states have requested
extensions beyond 1982 for achieving compliance with the ozone NAAQS. Those
areas within the 24 states which are not expected to be in compliance are
primarily large urban areas. To add to the uneven distribution of ozone
2-20
-------
problem areas addressed by this standard, the SOCMI industry is not evenly
distributed throughout the U.S. He concluded that there were no benefits
from the standards for large portions of the nation either because there is
no SOCMI industry or because the area is already in compliance.
Response:
The ozone standard referred to by one of the commenters, is the
national ambient air quality standard (NAAQS) for ozone. Compliance with
NAAQS does not preclude the development of new source performance standards
(NSPS). New source performance standards are not directly designed to
achieve the ambient air quality goals. An overriding purpose and long range
goal of a NSPS is to minimize emissions at all new and modified sources,
wherever they are located, in order to prevent new pollution problems from
developing and to enhance air quality as the Nation's industrial base is
replaced.
The standards will limit VOC emissions from all new, modified, or
reconstructed SOCMI process units and will result in emission reductions
well into the future. Even though these reductions may not bear directly on
attainment or nonattainment of NAAQS for ozone, they will make room for
future industrial growth while preventing future air quality problems.
Clearly, residents in both attainment and nonattainment areas would benefit
from these standards. The NSPS complements the PSD and nonattainment rules
as a means of achieving and maintaining the NAAQS, while on a broader basis
they prevent new sources from making air pollution problems worse regardless
of the existing quality of ambient air. Therefore, while new source
standards may help in the attainment of NAAQS, the consideration of
compliance or noncompliance with NAAQS does not influence directly the
decision to set standards of performance.
Comment:
Another commenter (IV-D-42) added that EPA had cited little or no
evidence that ethanol emissions from distilling endanger public health or
welfare.
Response:
The Clean Air Act was developed to establish national air quality'and
environmental goals that would protect and enhance the quality of the
2-21
-------
nation's own resources to promote public health and welfare and the
productive capacity of its population. The Administrator clearly documented
the need to regulate VOC in order to protect public health and welfare in
the EPA publication "Air Quality Criteria for Ozone and Other Photochemical
Oxidants" (IV-A-17). VOC emissions are precursors to the formation of ozone
and other oxidants (ozone). Ozone results in a variety of adverse impacts
on health and welfare, including impaired respiratory function, eye
irritation, necrosis of plant tissues, and the deterioration of synthetic
rubber. An independent determination for each of the SOCMI chemicals as
suggested by the commenter calling for an individual finding in the case of
ethanol is unnecessary. VOC emissions as a class have been determined to
contribute to ozone formation. Since ethanol is a VOC and may be produced
with and from other VOC, it remains on the list of SOCMI chemicals.
Comment:
One commenter (IV-D-12) said that complying with the standards would
contribute nothing to air quality in the area of his plant where the major
air problems are carbon monoxide and particulates.
Response:
New source performance standards are aimed at preventing air quality
from deteriorating due to an increase in the number of industrial sources.
If ozone is not a problem in a particular area, the SOCMI fugitive VOC
standards will help, as discussed above, to ensure that ozone levels do not
become a problem in that area. The standards will help in other ways as
well, such as in reducing odors and hazardous air pollutants.
Comment: .
One commenter (IV-D-43) wrote that the Proposed Notice of Revocation
for the NAAQS for hydrocarbons (46 FR 25655; May 8, 1981) destroys the basic
premise for the priority listing of VOC from SOCMI as a class of substances
that endanger publ.ic health and welfare. He wrote further that only those
individual substances whose emissions have a measureable impact on health
and welfare are appropriate for regulation.
Response:
As discussed in Section 8 of this document, the revocation of the NAAQS
for hydrocarbons does not prevent the regulation of VOC emissions as
2-22
-------
precursors to ozone.for which there is a NAAQS. In establishing the level
of the NAAQS for ozone, EPA made the determination that ozone .may endanger
health and welfare, and established the need for controlling VOC as
precursors to ozone (IV-A-17).
Comment:
Two commenters (IV-F-1, No.3; IV-D-50) said that EPA's calculations
assume that there is a direct correlation between hydrocarbon emissions and
ozone levels. They said that this assumption is not necessarily correct.
The commenters continued, saying that there had been no demonstration that
any of the chemicals regulated are directly related to ozone. They said
that EPA had speculated on this relationship.
Response:
EPA has determined that VOC contribute to ozone formation through
photochemical reactions in the atmosphere. These findings were published in
Air Quality Criteria for Ozone and Other Photochemical Oxidants (iy-A-17).
In this same document there are descriptions of several models which relate
emissions of VOC to oxidaht levels. It should be noted, however, that at no
point during this standards-setting process has EPA attempted to relate
quantitatively the emissions of VOC which would be affected to the resultant
air quality. It is sufficient for the purposes of new source standards to
aim at preventing degradation of air by new sources, knowing that emissions
of VOC contribute to ozone formation and, therefore, degradation of the air.
Comment:
One commenter in several sets of comments (IV-F-1, No.3; IV-D-43;
IV-D-28) questioned EPA's controlling ozone on the one hand while it has
work underway on the other hand to prevent the loss of ozone from the upper
atmosphere. The commenter asked how the agency could decide that chloro-
fluorocarbon or chloroform or carbon tetrachloride, for example, must be
regulated because they destroy ozone in the stratosphere and also must be
included in a category for regulatory action because they form ozone in the
troposphere. The commenter referred to 45 FR 66726 as support for his
argument.
2-23
-------
Response:
As discussed in Section 5.1, the chlorofluorocarbons, carbon tetrachlo-
ride, chloroform and other negligibly photochemically reactive compounds
remain on the SOCMI list because photochemically reactive VOC's are
processed in producing these chemicals using fugitive emission sources. The
criteria for removing them from the list would be that none of the raw
materials, additives, intermediates, or finished products is a photochem-
ically reactive VOC. Since the chemicals cited by the commenter are all
produced from photochemically reactive substances, they are covered by the
standards. However, EPA has added provisions to the standards that would
allow an owner or operator to eliminate coverage of fugitive emission
sources that do not contain VOC.
EPA's program to control ozone depletion by chlorofluorocarbons is
described in the Federal Register notice referenced by the commenter.
Congress required that EPA undertake this control effort in the Toxic
Substances Control Act. Controlling emissions of chlorofluorocarbons and
their reactants to prevent ozone formation in the lower levels of the
atmosphere does not conflict with controlling chlorofluorocarbon emissions
to protect degradation of the upper levels of the atmosphere. Both purposes
are served by controlling chlorofluorocarbon emissions.
Chlorofluorocarbons (produced from such substances as perchloroethy-
lene, carbon tetrachloride, and fluorinated derivatives of acetylene) are
not, being regulated because they form ozone but because they are produced
from chemicals that form ozone in the troposphere. In any case, to the
extent that chlorofluorocarbons are controlled, the standards will reduce
the destruction of ozone in the stratosphere.
2-24
-------
3. BASIS FOR THE STANDARDS
Section 3 presents the basis for the final standards and summarizes
the comments and responses on the basis of the proposed standards. The
selection of the final standards is presented in Section 3.1 and the
selection of the formats of the standards is discussed in Section 3.5. The
remaining sections summarize the conclusions from the Additional Information
Document (AID) (III-B-2) on emission factors (Section 3.2), model units
(Section 3.3), and emission reductions (Section 3.4).
3.1 SELECTION OF FINAL STANDARDS
3.1.1 Basis for the Final Standards
Comment:
Many people (IV-D-17; IV-D-24; IV-D-28; IV-D-40; IV-F-1, No.3)
commented on the basis for selection of the proposed standards. The
commenters questioned the choice of Regulatory Alternative IV; they said
that it was not cost effective. Some of the commenters recommended the
selection of Regulatory Alternative II; some recommended Regulatory
Alternative III. Another commenter recommended adoption of Regulatory
Alternative II with the addition of closed loop sampling systems (part of
Regulatory Alternative IV) which he considered cost effective. Some
commenters said that the incremental cost-effectiveness of Regulatory
Alternative IV was unreasonable.
Response:
Section 111 of the Clean Air Act, as amended, requires that standards
of performance be based on the best system of continuous emission reduction
that has been adequately demonstrated, considering costs, nonair quality
health and environmental impacts, and energy requirements. The control
techniques for fugitive emissions have been adequately demonstrated as
discussed in Section 4, Control Technology. The magnitude of fugitive
emissions of VOC from SOCMI and the emission reduction achievable if
fugitive emission control techniques are implemented are discussed in
3-1
-------
Section 6, Environmental Impact. The nonair quality health and environ-
mental impacts are also discussed in Section 6 of this document and
Chapter 7 of the BID for the proposed standards.
Since proposal, EPA has decided to accept the suggestions of commenters
and focus further on cost-effectiveness in selecting the basis for the
selection of final standards. In making this decision, EPA is accepting the
suggestions of commenters to base the standards on cost-effectiveness
considerations. After considering the cost-effectiveness of control
techniques for each fugitive emission source covered by the standards, EPA
analyzed the economic impact on the industry of the control techniques
selected on a cost-effectiveness basis.
Cost-Effectiveness Considerations. EPA analyzed the annualized cost of
controlling VOC emissions and the resultant VOC reduction for each
alternative control technique. Costs for implementing the standards are
presented in Section 7. Emission reductions are presented in Section 6.
The control costs per megagram of VOC reduced are presented in Table 3-1 for
each fugitive emission source covered by the standards. These costs do not
represent the actual amounts of money spent at any particular plant site.
The cost of VOC emission reduction systems will vary according to the
chemical product being produced, production equipment, plant layout,
geographic location, and company preferences and policies. However, these
costs and emission reductions are considered typical of control techniques
for fugitive emission sources within SOCMI units,and can be used in
selecting the level of control to be required by the standards.
Pressure Relief Devices. The annualized costs and VOC emission reductions
achieved for monthly and quarterly leak detection and repair programs (LDRP)
and for the use of control equipment (rupture disks) were determined for
pressure relief devices in gas service. As Table 3-1 shows, both the
quarterly and monthly leak detection and repair programs are less expensive
than installation of rupture disks. Leak detection and repair programs
result in average credits of $240/Mg and $150/Mg of VOC for quarterly and
monthly programs. A monthly leak detection and repair program achieves an
additional 0.7 Mg/yr of emission reduction at a cost of $500/Mg compared to
3-2
-------
TABLE 3-1. CONTROL COSTS PER MEGAGRAM OF VOC REDUCED'
Fugitive
Emission
Source
Pressure
relief
devices
Compressors
Open-ended lines
Sampling Systems
Valves
Pumps
Emission
Control . Reduction
Technique (Mg/yr)
Quarterly leak detection
and repair
Monthly leak detection
and repair
Rupture disks? '3
Controlled degassing vents?
Caps on open-ended lines? >~
Closed purge sampling?
Semi-annual leak
detection and repair.
Quarterly repair leak
detection and repair
Monthly leak detection
and repair*
Quarterly leak detection
and repair
Monthly leak detection
and repair?
Dual mechanical seal
systems vented to a flare
4.4
5.1
10.0
4.0
r 6.2h
3.4
17.1
26.9
33.6
4.1
7.6
12.6
Average
$/Mge '
credit6
credit6
510
credit6
400h
590
credit6
credit6
62
1200
610
2300
Incremental
$/Mgd
500
1200
credit6
400
. 590
credit6
480
credit6
5600
Costs and emission reductions based on fugitive emission source counts in
Model Unit B. See Section 3.2.
Further discussion of control techniques can be found in Section 4.
°Average dollars per megagram (cost effectiveness) = (net annualized cost
per component) * (annual VOC emission reduction per component).
Incremental dollars per megagram = (net annualized cost of the control
technique - net annualized cost of the next less restrictive control
technique) * (annual emission reduction of control technique - annual
emission reduction of the next less restrictive control technique).
6Values indicated as a credit denote savings. The annualized savings are
presented in the text.
Control technique selected as the basis for the standard.
"These would be the costs and 'emission reductions for those sources not
already controlled: 75 percent of the safety/relief valves per process
unit and nearly all open-ended lines are controlled in the absence of
standards.
This cost and emission reduction represent the values if open-ended lines
were not controlled in the absence of standards.
-------
a quarterly leak detection and repair program. Rupture disks achieve an
additional 4.9 Mg/yr emission reduction at a cost of $1200/Mg compared to a
monthly leak detection and repair program. The $1200/Mg incremental cost of
achieving this 4.9 Mg/yr of emission reduction is reasonable. The 4.9 Mg/yr
is about 7 percent of the total emission that can be reduced by the
standards in a model unit B. Thus, the control equipment was selected as
the basis for the pressure relief device standard.
Compressors. Only one control technique can be considered for compressor
seals: the installation of equipment such as control of barrier fluid
systems. As explained in the AID, if a compressor is found .leaking, the
repair procedure would be the installation of control equipment. Because
compressors are not generally spared, repair would be delayed until the next
turnaround, thereby reducing the effectiveness of a leak detection and
repair program to essentially zero. The installation of control equipment
results in a cost savings of $100/Mg, indicating that the value of product
retained by controlling the barrier fluid system exceeds the cost of the
control equipment. This cost is reasonable, and, therefore, control
equipment was selected as the basis for the standard for compressors.
Open-ended Lines. EPA considered caps or closures as
the control technique for the standard for open-ended lines. Caps and
closures are in wide-spread use in SOCMI and are expected to be used even
more frequently in new SOCMI units. The cost and emission reduction
presented in Table 3-1 are the cost and emission reduction which would be
realized for open-ended lines that are not controlled. The $400/Mg cost for
controlling the fugitive emissions of VOC from open-ended lines is
reasonable.
Sampling Systems. Closed purge sampling is the control technique for the
standard for sampling systems. Closed purge systems are becoming
increasingly common in the chemical industry. The $590/Mg cost for fugitive
emissions of VOC from sampling systems is reasonable.
Valves. Several leak detection and repair programs were considered for
valves. The programs differed in the monitoring frequency which would be
implemented. As Table 3-1 shows, the lowest average cost per Mg of VOC and
3-4
-------
the lowest incremental cost per Mg of VOC is associated with the quarterly
program (a cost savings of $41/Mg of VOC on the average). However, the
largest emission reduction is associated with the monthly program at an
average cost per Mg of VOC of $62. Furthermore, the incremental cost per Mg
VOC emissions reduced for the monthly program is $480 per Mg with an
incremental emission reduction of 6.7 Mg/yr. EPA considers these costs to
be reasonable. Therefore, EPA selected a monthly leak detection and repair
program as the basis for the standard for valves.
Pumps. The control costs incurred for each megagram of VOC emissions
reduced and emission reductions achieved were determined for two leak
detection and repair programs and the use of dual mechanical seals with
controlled degassing vents. Both leak detection and repair programs incur
lower costs than the costs which would be incurred with equipment installa-
tion. The lowest average and incremental costs per Mg are associated with a"
monthly leak detection and repair program. The monthly program achieves a
higher degree of control than the quarterly program at an incremental cost
of $25/Mg of additional VOC, but it achieves a lower degree of control than
installation of control equipment. However, even though control equipment
provides for the greatest amount of VOC reduction, the $5000/Mg incremental
cost to obtain the 5 Mg/yr is judged unreasonably high. Because the costs
for equipment are unreasonably high, monthly leak detection and repair was
selected as the basis for the standard for pumps.
Economic Impact Considerations. At proposal, an economic analysis was
performed which evaluated the economic impacts of the standards. None of
the comments on this analysis showed significant adverse impacts on SOCMI
due to the standards. Since proposal, EPA has reconsidered the economic
impact of the standards. The results continue to show no unreasonable
impact. The economic impact analysis is discussed in Chapter 9 of the BID
for the proposed standards and in Section 7.2 of this document.
3.1.2 Other Comments Concerning the Selection of Standards
Comment:
One commenter (TV-D-17) interpreted EPA's choice of the most stringent
Regulatory Alternative to mean that EPA deemed 85 to 90 percent control
3-5
-------
acceptable, and, therefore, EPA chose Regulatory Alternative IV because it
achieved 87 percent control.
Response:
The commenter misinterpreted EPA's actions in choosing Regulatory
Alternative IV for the proposed standards. The choice did not depend on a
particular level of control or on a predetermined, acceptable cost limit. As
explained before, the most stringent controls which were economically
reasonable were chosen. The controls for each Regulatory Alternative were
chosen first. Then, the expected emissions reductions and the costs were
estimated. The emission reduction estimates presented in the BID for that
level of control happened to be about 87 percent and resulted in a
reasonable economic impact.
As indicated in Section 3.5.1, in choosing the final standards EPA has
looked at revised cost estimates and revised emission reduction estimates
for each fugitive emission source to which the standards would apply. The
cost-effectiveness values were considered in selecting the final standards.
After an initial selection of final standards, the economic impacts
associated with the selected standards were analyzed and were found to be
reasonable.
Comment:
Another comment concerned the establishment of the baseline level of
control. A commenter (IV-D-28) wrote that EPA presumes that the equipment
prescribed by the proposed rule would not be used in its absence. This
presumption is incorrect because of the greater use of these methods in new
plants than in older designs. The commenter indicated that this fact
reduces the expected benefits of the proposal. He added that EPA offers no
data on the current extent of usage for dual seals, rupture discs, or
monitoring programs. Any benefit projected as a result of implementation
is, therefore, only speculative. The commenter also stated that various
other regulations are in effect which will also control VOC emission
independently of this regulation. The use of Regulatory Alternative I,
i.e., results of no action, baseline is, therefore, incorrect.
3-6
-------
Response:
The commenter is apparently confused about the definition.of baseline.
The baseline level is representative of current industry practice. EPA's
data are very recent and do represent current industry practice. Therefore,
the baseline is appropriately chosen.
Baseline control was discussed in the BID, p. 3-17 through 3-21.
Included is a discussion of the extent to which monitoring programs are
currently used. Information on current control levels for pumps and rupture
discs is presented in Chapter 3 of the AID.
It should also be noted that in the final selection of the standards,
comparisons of cost effectiveness were made on an incremental basis, not on
a baseline basis. Because of the method used in the final choice of the
standards, the baseline level has little, if any effect on the selection of
the standards. J
3.2 EMISSION FACTOR DEVELOPMENT
Numerous comments4 were received on the interpretation and use of
available fugitive VOC emissions data in assessing impacts of regulatory
alternatives on SOCMI. An analysis of the available studies is presented in
the Additional Information Document (AID) for fugitive VOC emissions
(III-B-2) released previously. The studies were compared and considered
with regard to their applicability to fugitive emissions and to SOCMI.
As discussed in the AID, the percent of fugitive emission sources which
leak and the quantity of emissions from these leaking sources are the
primary factors which influence the quantity of VOC emissions from fugitive
emission sources. EPA still considers data from petroleum refineries
(II-A-26) appropriate in estimating the quantity of VOC emissions from
sources which leak, except for valves in gas service. The data from
petroleum refineries were gathered for the purpose of developing emissions
estimates from fugitive emission sources. Even though data gathered during
the Maintenance Study (IV-A-10) were not gathered for the purpose of
estimating emissions, they have been used to estimate emissions from valves
in gas service.
aIV-D-l; IV-D-6; IV-D-7; IV-D-13; IV-D-15; IV-D-17; IV-D-21; IV-D-26;
IV-D-28; IV-D-40; IV-D-43; IV-D-50; IV-F-1, No.l; IV-F-1, No.3.
3-7
-------
The percent of fugitive emission sources which leak was based on data
gathered during the SOCMI 24-Unit screening study (IV-A-11). The 24 units
screened are a cross-section of process units in SOCMI but are not a
representative selection of process units. Even though the 24 units
screened are not necessarily representative of all SOCMI units, EPA decided
that the percent of leaking sources determined in this study could be used
in combination with quantities of mass emissions from leaking sources to
develop average emission factors.
The average emission factors used in the Background Information
Document (BID) (III-B-1) are compared in Table 3-2 to those factors
developed in the AID. The complete data analysis, evaluation of studies,
and comparison of emission factors are presented in Section 2 of the AID.
Comments on the AID and EPA's responses to those comments are presented in
Appendix A.
Comment:
Several commenters (IV-D-26; -IV-D-28; IV-D-13) cited a paper by
Monsanto Research Corporation (MRC) in supporting their contention that
petroleum refineries and SOCMI were different. One commenter (IV-D-28)
pointed out the fact that the report had been disregarded by EPA because the
results were not comparable.
Response;
As explained in the BID and the AID, the results of the MRC study were
analyzed. The study was performed in a manner that prevented comparisons
between it and other studies. Therefore, the results were not useful in
performing an analysis for regulatory purposes. The usefulness of the MRC
study was in the fact that it pointed out the necessity for doing more work,
which EPA did (see II-B-34 for a discussion of limitations of the study).
Comment:
Two commenters cited work done in their own plants (IV-D-6; IV-D-13)
showing leak frequencies for an acrylonitrile unit and a chlorinated hydro-
carbon unit which were lower than leak frequencies determined in petroleum
refineries.
3-8
-------
TABLE 3-2. COMPARISON OF EMISSION FACTORS, kg/hr/source
Source
AID
BID
Pumps - light liquid
- heavy liquid
Valves - gas
- light liquid
- heavy liquid
Compressors
Pressure relief devices
Flanges
Oper-ended lines
Sampling connections
- gas
0.0494
0.0214
0.0056
0.0071
0.00023
0.228
0.104
0.00083
0.0017
0.0150
0.114
0.021
0.0268
0.0109
0.00023
0.636
0.16
0.00025
0.0023
0.0150
3-9
-------
One of the commenters (IV-D-13) said that studies done in his own plant
showed emissions from pumps to range from 6.8 to 9.5 g/hr. He pointed out
that both these estimates are significantly different from EPA's rate of
120 g/hr.
Response:
It is certainly conceivable that these lower leak frequencies and
emission rates exist. As discussed in the AID, SOCMI contains units with
very low fugitive emission rates and units with very high fugitive emission
rates. Individual units would be expected to fall within a range of leak
frequencies and mass emission rates. The selection of leak frequency and
leak rate information to be used as the basis for estimating impacts of the
standards on SOCMI are discussed in detail in the AID.
Comment:
Inherent differences in the operations and the materials handled were
also offered as supporting arguments for SOCMI's being different from
petroleum refineries. As an example of the differences between refinery and
SOCMI plants, one commenter (IV-D-15) cited the large number of sources
tested in SOCMI that had visible solid residue. Such residue was considered
an indication of possible leaks, but no VOC emissions were measured.
Another comment letter (IV-D-17) said because of the broad spectrum of
materials within SOCMI, having a wide variation in physical and chemical
properties, the emission rates would be expected to vary from SOCMI sector
to sector as well as from the refining industry rate.
Another commenter (IV-D-6) cited differences in toxicity and hazardous-
ness of SOCMI chemicals as compared to refinery streams. With chemicals,
the commenter said, the exclusion of Q~ anc* explosive concerns dictate
operating and design practices. In SOCMI facilities the toxicity of
chemicals often controls design and operating practices. These design and
operating practices, he said, are even more different because they are
influenced by OSHA regulations.
Another commenter (IV-F-1, No. 3) said refineries are characterized by
much more strenuous conditions, larger equipment, higher temperatures, and
outdoor continuous processes. The chemical industry on the other hand was
3-10
-------
said to be characterized by smaller batch-type equipment, indoor operation,
more valuable products, and less strenuous conditions. Another commenter
(IV-D-21) added that these differences are borne out in EPA's own data. He
said leak frequencies for high volume SOCMI chemicals such as ethylene are
somewhat similar to refinery data. According to the commenter, smaller
volume chemicals often produced in batch operations, such as ethylene
dichloride processes, have significantly Tower emission rates.
Another commenter (IV-D-13) said the value of products in SOCMI is
higher than the value of products in refineries; therefore, the fugitive
losses are kept under better control in SOCMI. " : :
Response:
As the commenters pointed out, there are some apparent differences
between petroleum refinery and SOCMI data. These apparent differences may
be due to reasons suggested by the commenters or to other unknown reasons'.
As stated in the AID, the reasons for the differences are not clear. There
is not conclusive evidence to show why such differences are seen. The
reasons cited by the commenters are generalizations which do not adequately
describe differences between the industries. Many SOCMI processes are
outdoor, high-temperature high-pressure processes.
No matter what the reasons are that the most recent data suggest lower
emissions from SOCMI, EPA recognized the difference and the estimates of
impacts of the standards have been revised. In general, the comparison
between SOCMI and petroleum refineries is not appropriate. The determina-
tion of best demonstrated technology was performed separately for the two
industries. Since the determinations of what constitutes best demonstrated
technology were performed independently for the two industries, comparisons
of the two industries do not yield useful information.
Comment:
One commenter (IV-D-6) noted an apparent error in the data interpreta-
tion in the EPA data base. He pointed out that different values of K, the
factor used to correct the units for variables in the emission rate
equation, is reported by the EPA contractor in separate documents. The
commenter indicated that these values are different from those used by his
3-11
-------
company. He stated that if these erroneous factors have been used, the
reported emission rates are five times higher than actual. The commenter
added that evaluation of pump seal literature suggests that this error
exists throughout the entire data base.
Another commenter (IV-D-21) wrote that EPA has used inconsistent and,
in some cases, inaccurate conversion constants (K,) In evaluating both the
SOCMI and refining study data. He charged that these inaccuracies and
inconsistencies call into question the validity of both the refining and
SOCMI data used as the basis for the proposed NSPS.
Response:
As noted in Docket Item Entry IV-B-7, the computations have been
checked and the rates calculated in the refinery assessment study are
correct. Unfortunately, the equation as written on page 124 of Appendix A
of the refinery assessment report (EPA-600/2-80-076b) contains several
errors. The corrected equation is printed below.
CORRECTED EQUATION FOR HYDROCARBON EMISSION RATE
-------
3.3 MODEL UNITS
The model units developed for SOCMI serve as the basis for aggregating
emissions estimates to evaluate the overall impact of regulatory alterna-
tives. They also provide a means of estimating nationwide impacts. A few
conmenters questioned the development of model units based on process
complexity instead of production rate.
The basis for development of the model units was detailed in the AID.
A comparison was made between the equipment counts in the units screened in
the -SOCMI 24-unit screening study (IV-A-11) and in the model units. EPA did
not change the model units presented in the BID. But EPA did revise some
equipment counts to represent the current level of emissions control
reported in SOCMI and to clarify some previous confusion. Comments on the
development of model units presented in the AID and EPA's responses to those
comments are included in Appendix A. The model units used in the BID and
thosa presented in the AID are compared in Table 3-3. The equipment counts
embodied in the model units are used with the emission factors in Table 3-2
to determine the environmental impacts presented in Section 6.
3.4 EMISSION REDUCTIONS
The effectiveness of the standards is evaluated, in part, by the
emissions reductions achievable by the various regulatory options for the
different fugitive emission sources in SOCMI. The control techniques
comprising these options are discussed in Section 4. The basis for
estimating the effectiveness of control techniques was presented in detail
in Section 4 of the AID. Comments on the technical analysis presented in
the AID and EPA's response to those commenters are included in Appendix A.
Key elements in the standards are leak detection and repair programs.
Such programs are useful in reducing emissions from valves and pumps. The
technique used in estimating the effectiveness of these programs in the BID
was based primarily on some engineering judgments concerning occurrence,
recurrence, and repairability of leaking valves.
During the development of these standards, additional data were
gathered to permit an improved evaluation of these phenomena with respect to
reducing fugitive VOC. An evaluation was made of the available data on leak
aIV-D-38; IV-D-40.
3-13
-------
TABLE 3-3. EQUIPMENT COUNTS FOR FUGITIVE VOC EMISSION SOURCES IN SOCMI MODEL UNITS
CO
I
Equipment Component
Pump Seals
Light Liquid Service
Single mechanical
Dual mechanical
Seal less
Heavy Liquid Service
Single mechanical
Packed
Valves
Vapor service
Light liquid service
Heavy liquid service
Safety/relief valves
Vapor service
Light liquid service
Heavy liquid service
Open-ended lines
Vapor service
Light liquid service
Heavy liquid service
Compressor seals
Sampling connections
Controlled
Uncontrolled
Flanges
Model Unit
A
5
3
0
5
. 2
90
64
64
11
1
1
g
47
46
1
26
600
BID Analysis
Model Unit
B
19
10
1
84
6
365
335
335
42
4
4
37
169
189
2
104
2400
Number of Equipment Components
Hodel Unit
C
60
31
1
73
20
1117
1037
1037
130
13
14
115
681
S81
8
320
7400
Hodel Unit
A
5
3
0
5
2
99
131
132
llc
1
1
104d
1«
f
19
7
600
Revised Analysis
Hodel Unit
B
19
10
1
24
6
402
524
524
42°
4
4
415"
28
78
26
2400
Hodel Unit
C
60
31
1
73
20
1232
1618
1618
130C
13
14
1277d
8e
240
BD
7400
'Equipment components In VOC service only.
b52I of existing units are similar to Hodel Unit A.
33» of existing units are similar to Hodel Unit B.
151 of existing units are similar to Hodel Unit C.
Seventy-five percent of gas safety/relief valves are assumed to be controlled at baseline; therefore, the emissions
estimates are based on the following counts: A.3; B,ll; C.33.
All open-ended lines are considered together with a single emission factor; 1001 are controlled at baseline.
Emission factor estimate incorporates 60 percent control; cost estimates are based on the following counts: A. 0.4-.
B, 0.8; C, 3.2.
*75t controlled.
-------
occurrence, recurrence, and repair effectiveness in the AID. In addition, a
model based on recursive equations (LDAR) was developed for evaluating leak
detection and repair programs (IV-A-22). In the AID, this model was
compared to the model used in the BID (ABCD) and to the BID model with an
improvement suggested by industry commenters. EPA determined that the LDAR
model more appropriately represented the leak detection and repair programs
that are part of the SOCMI standards.
In order to evaluate the effectiveness of leak detection and repair
programs for pressure relief devices, some estimate of the efficiency of
these programs must be made. The LDAR model, presented in the AID, is a
better indicator of effectiveness for leak detection and repair programs
than the approach used in the BID for estimating program effectiveness.
However, the LDAR model, which was used for valves and pumps, requires
occurrence and recurrence rates, which were not determined for pressure
relief devices. The ABCD model presented in the BID was based on estimates
that may not be representative of the actual situation, considering the ;
comparison of the results of these models for valves and pumps. Therefore,
the LDRP effectiveness for pressure relief devices was estimated using the
effectiveness for gas service valves based on the LDAR model multiplied by
the ratio of the effectiveness for pressure relief devices based on the ABCD.
approach to effectiveness for gas service valves based on the ABCD approach
(IV-B-I9).
The efficiencies of controlling the other emission sources were not
changed from the ones presented in the BID. The control effectiveness of
the techniques on which the emissions reductions are based are summarized in
Table 3-4. These values are used in Section 6 to determine the overall
effectiveness of the standards in reducing fugitive VOC emissions.
3.5 FORMAT OF STANDARDS
Comments on the format of the standards included several requests for
regulations in different formats. These requests included:
performance standards for valves
equivalent equipment and work practice standards for valves
work practice standards for pumps
3-15
-------
TABLE 3-4. SUMMARY OF CONTROL EFFECTIVENESS FOR SOCMI FUGITIVE EMISSION SOURCES
OJ
1
G1
Emission Source
Pump seals
- light liquid
Valves
- gas
- light liquid
Pressure relief devices
- gas
Open-ended lines
- all services
Compressor seals
Sampling connections
Control device
Type of Standard
Work Practice
Work Practice
Work Practice
Performance
Equipment
Equipment
Equipment
Design
Control Effectiveness
Control Technique Applied (decimal percent)
Monthly leak detection and 0.61
repair
Monthly leak detection and 0.73
repair
Monthly leak detection and 0.46
repair
Tie to flare; rupture disk 1.0
Caps, plugs, blinds 1.0
Mechanical seals with vented 1.0
degassing reservoirs
Closed purge sampling systems 1.0a
Incinerator, vapor recovery 0.95a
system, flare
Where a control device is applied as supplement to equipment, e.g. for compressor seals, the
control effectiveness of the equipment is reduced from nearly 100 percent by the 95 percent
effectiveness of the control device.
-------
performance standards for pumps and compressors
performance standards for sampling systems
Other comments regarding format of the standards included one concerning
complexity of the standards, several suggesting.that concentration limits
are a performance standard format, and one requesting separate standards for
separate equipment types.
3.5.1 Performance Standards for Valves
Comment:
Two commenters (IV-F-1, No.4; IV-D-21) requested that a performance
standard be set for valves in addition to the work practice standard.
Response:
A performance standard offers more flexibility to industry and, in
that regard, allows for more innovative control techniques. However, as
explained in the preamble to the regulation (46 FR 1136), for most SOCMI
fugitive emission sources, it is not feasible to prescribe an emission limit
performance standard. Except in those cases in which a standard can be set
at "no detectable emissions," the only way to measure emissions from SOCMI
fugitive emission sources such as valves would be to use a bagging technique
for each of the valves in a process unit. The great number of valves and
their dispersion over large areas would make such a requirement economically
impracticable. Therefore, EPA did not select this format for the standards.
Another approach to prescribing a performance standard would be to .
specify a number or percent of fugitive emission sources (valves) that would
be allowed to leak. This approach would be more qualitative than an
approach based on quantitative emission measurements such as bagging. This
format would be based on a leak frequency limit rather than an emission
limit and would have some of the same benefits of flexibility. The only
fugitive emission source for which a leak frequency limit would be
applicable is valves because other fugitive emissions are too few in number
to allow a meaningful percent to be determined. The variability in the
percentage of valves leaking among process units precludes the setting of an
allowable percentage of valves leaking which could be achieved by all
process units within SOCMI (see Section 14). This variability is observed
3-17
-------
even among units in which leak detection and repair programs are being
implemented. Even so, establishing an allowable percentage of. valves
leaking may be feasible for some process units. EPA has effectively
provided the flexibility afforded by a performance standard while allowing
for the variability within the industry, by providing an alternative
standard which is an allowable percentage of valves leaking. Alternative
standards are explained in more detail in Section 14.
Comment:
One commenter (IV-F-1, No.4), in requesting that a performance standard
be set for valves in addition to the work practice standard, recommended a
leak frequency of 2 percent be considered as equivalent to the standard as
proposed. He cited work by Snee and Kittleman of DuPont as the basis for
the 2 percent figure.
Response:
Mr. Kittleman and Dr. Snee have been very active proponents of the use
of statistical sampling plans called skip-period plans (II-B-26; II-D-87;
IV-D-1). One of the requirements of the plans is the establishment of a
"good performance level." Based on data presented in the SOCMI BID for
quarterly monitoring, they have recommended that a leak frequency of
2 percent be considered a good performance level for such plans. The
commenter is recommending that this recommended good performance level be
adopted as the compliance level for a performance standard which specifies
an allowable percentage of valves leaking.
A good level of performance based on the percent of valves found
leaking cannot be established for all process units. But it may be
achievable by some units and may be readily achievable by employing a less
frequent leak detection and repair program than is required under the
non-optional standards. Therefore, EPA has set an allowable percentage of
valves leaking for the alternative standard at 2 percent. Alternative
standards are discussed in detail in Section 14.
While agreeing with Mr. Kittleman and Dr. Snee that the compliance
level should be 2 percent, EPA differs in the manner in which the
determination was made. Using the estimates for occurrence and recurrence
3-18
-------
of valve leaks .presented in .the BID, (p.4-10), Mr. Kittleman and Dr. Snee
compared the estimates of initial leak frequency and leak frequency after a
quarterly leak detection and repair program had begun to arrive at
90 percent reduction in the number of leakers. They assumed this level of
control as an acceptable level which should be targeted. The quarterly
comparison was chosen from a table showing monthly, quarterly, and annual
estimates. The estimates were made solely for purposes of comparison of
regulatory alternatives and were based on engineering judgment. They were
not presented as acceptable or actual levels.
The standards for valves are based on a monthly leak detection and ,'
repair program. Using the LDAR model (IV-A-22) for leak detection and :
repair programs, EPA examined the cost and effectiveness of monthly programs
applied to process units exhibiting various leak frequencies for valves.
Leak frequencies associated with high cost effectiveness were identified. '-
And the corresponding statistical level of performance was computed. The -
compliance level of 2 percent that is the basis of the alternative standards
was selected to exempt units exhibiting low leak frequencies and
consequently high cost/effectiveness ratios. ,
Comment:
Another commenter (IV-D-21) who requested a performance standard for
valves wrote that the proposed work practice standards fail to provide for
innovative technology and fail to provide true incentives to reduce VOC
fugitive emissions. He said that a facility would be judged in compliance
if all monitoring was performed, all records kept, and all reports made even
if all monitored equipment was found to leak at each subsequent inspection.
The commenter concluded that the work practice approach would not-result in
significant control of VOC fugitive emissions.
Response:
EPA believes that effective emission reductions can be achieved through
the required work practices. Records of the activities performed under such
a standard will serve as an indicator of the diligence.of the owner or
operator in performing the required work practices. Using these records as
an indicator, compliance with the work practices can be judged. The
3-19
-------
commenter is correct in stating that using the work practice standard
approach, a unit would not be in violation if a leak were found. Rather,
the unit clearly would be in violation if no attempt was made to repair it.
This requirement for repair is the incentive for reducing emissions. Test
results (IV-A-10) show that attempting repair of leaking valves results in
71 weight percent reduction in emissions. Furthermore, successful repair of
leaks (i.e., reducing the screening value below 10,000 ppmv) reduced mass
emissions by 98 percent and even unsuccessful repair (i.e., attempted repair
not reducing the screening value below 10,000 ppmv) reduced mass emissions
by 63 percent. Therefore, effective emission reduction is expected under
the work practice standard.
EPA has provided two alternative standards for valves (see Section 14).
These alternative standards provide flexibility to the owner or operator in
meeting good performance levels based on the number (or percent) of leaking
valves found in a process unit. They are similar to performance standards
in providing initiatives for innovative control techniques. The owner or
operator may use any other program for leak detection and repair, provided
equivalency with the valve standards is established.
Comment:
One commenter (IV-D-17) recommended a performance standard in the form
of a required percentage of reduction in fugitive VOC emissions. He
reasoned that this approach would allow a unit to develop a program incor-
porating equipment or monitoring or both to achieve the required reduction
at the least cost. He referred to this approach as the "bubble concept."
The commenter noted that the relationship between screening values and
emission rates could be used to determine total emissions and emission
reductions.
Response:
A major goal of the SOCMI fugitive VOC NSPS is to reduce emissions of
VOC from all fugitive emission sources throughout the industry. Since these
are new source standards, an inherent purpose is to build: new process units
that would have low emissions. To implement a performance standard based on
a percentage reduction in fugitive VOC emissions, the total uncontrolled
3-20
-------
emissions for the new process unit would need to be determined. This
procedure seems inappropriate for a new process unit that, in order to meet
the objectives of these standards, would have incorporated some fugitive
emissions control devices in the original design. Under the suggested
performance standard, a new process unit that incorporates effective
fugitive emissions control features into its original design would be
penalized by having to meet a more stringent emissions level than a process
unit that did not include such controls originally.
As discussed in the proposal and in the previous responses, a perfor-
mance standard for valves was not possible as the basis of the standards for
valves. EPA has provided an alternative standard for valves that-allows a
unit to meet and maintain a performance level of 2 percent valves leaking
(see Section 14). EPA believes this alternative standard provides the type
of standard recommended by the commenter.
The commenter also suggested using the relationship between screening
value and emission rate to determine emissions and emission reductions.
These relationships were developed for all sources in petroleum refineries
'. and for pumps and valves (gas and light liquid) in SOCMI. The uncontrolled
- emissions on which the suggested performance standard would be based would
: have to be determined for each affected unit. Determining total emissions
in this manner would be extremely time consuming and the results would be
inaccurate. Furthermore, SOCMI represents a wide variety of processes, and
results of fugitive emissions studies indicate that emissions (leak
frequency and emissions rate) vary with process type (IV-A-14). EPA has
considered this variability in emissions characteristics by source and
process in setting the standards. Flexibility in the emissions control
techniques is provided by considering each source individually and providing
a format appropriate to that source (see other responses in this section).
For example, alternative standards for valves have been provided that permit
a process unit to comply with a performance standard based on the percentage
of leaks in the unit.
3-21
-------
3.5.2 Equivalent Equipment and Work Practice Standards for Valves
Comment:
One commenter (IV-F-1, No.4) recommended that if a performance standard
for valves is not possible, at the least, equipment and equivalent work
practice standards should be set. He said that allowing equivalent
compliance alternatives would allow the industry the flexibility necessary
to design and implement the most efficient and economical compliance
programs.
Response:
As previously discussed, compliance with a performance standard for
valves is not being required. The alternative standards for valves (see
Section 14), however, do provide a performance standard as requested by the
commenter. In addition, an alternative work practice standard provides for
a statistical skip-period leak detection and repair program. The plans
would allow skipping inspections as long as a good performance level has
been maintained for a series of inspections and continues to be maintained
at each subsequent inspection.
An equipment standard for valves was considered as explained in the
preamble to the proposed regulation (46 FR 1145). Leakless equipment, such
as diaphragm valves and bellows-sealed valves were not selected as the
standard for valves because of their limited applicability. However, as.
noted, use of these valves would be at least equivalent and is allowed.
Valves of this type would be required to operate with no detectable
emissions and would be subject to an annual performance test, but they would
be exempt from the monthly leak detection and repair requirements for
conventional valves.
3.5.3 Work Practice Standards for Pumps and Compressors
Comment:
Several commenters requested that an alternative work practice standard
be set for pumps and compressors (IV-F-1, No.4; IV-D-17; IV-D-16; IV-D-17).
The commenters cited flexibility, efficiency, and economics as reasons for
the request.
3-22
-------
Response:
EPA has determined that a work practice standard for pumps which
requires monthly leak detection and repair within 15 days is a reasonable
alternative to dual seals equipped with a non-VOC barrier fluid system. The
technical applicability of such programs for pumps is discussed in
Section 4. The effectiveness and cost of leak detection and repair programs
for pumps were examined in the AID (III-B-2).
EPA also considered a work practice standard for compressors. Examina-
tion of information on existing compressors indicated most were already
using the required equipment and only retrofits of seals to existing
reciprocating equipment would be technically impractical (see Section 4.12).
The number of compressors in a process unit was small compared to the number
of pumps as well, and compressors did not generally have spares. The
absence of a spare would make repairs difficult, if not impossible, without
a unit shutdown, and allowing a compressor to leak until shutdown would
severely reduce the effectiveness of a leak detection and repair program.
EPA concluded that the equipment standard proposed for compressors would
provide the highest degree of control at a reasonable cost and the equipment
standard was selected as the final standard.
3.5.4 Performance Standards for Pumps and Compressors
Comment:
Several commenters requested a performance standard for pumps and
compressors (IV-D-6; IV-D-20; IV-D-17). The reasons given included
flexibility, efficiency, and cost-effectiveness.
The wide variability within the industry was cited by one of these
commenters (IV-D-6) as one reason why a single standard was technically
infeasible. Another reason given by this commenter for a performance
standard was that it would make the VOC regulations compatible with OSHA
actions and prevent duplication of federal activities and needless cost to
industry. The same commenter cited an example for an acrylonitrile plant
owned by his company where the equipment standards as proposed for pumps
would be difficult to comply with.
3-23
-------
Response:
As noted in the preamble to the proposed regulations (46 FR 1143-1145),
performance standards based on emission limits were not possible for pumps
and compressors because of technological and economic limitations. More-
over, it was not possible to set concentration limit performance standards
based on concentration limits. Even pumps and compressors using the
equipment required by the standards have the potential to leak and any leak
would be a violation of such a performance standard. Additionally, there
are too few pumps and compressors to set a meaningful percent leaking
performance standard.
EPA has provided a work practice standard for pumps and has established
the equipment standards as alternative control techniques for pumps. In
addition, the equipment standards as proposed allow for several options.
Dual mechanical seals of any configuration with pressurized or non-pres-
surized barrier fluid systems or an enclosed and vented seal area were all
offered for pumps. Compressors can be controlled by using any type seal
with a barrier fluid or enclosed and vented seal areas.
EPA believes that the standards for pumps and compressors are
reasonable and allow all owners or operators of affected facilities to
comply. The standards incorporate provisions for complying by using
sealless equipment, dual seals with barrier fluid systems, or vented seal
areas without requesting permission or an equivalency determination. EPA
has also provided equivalency procedures for these standards that permit an
owner or operator of a process unit to comply with other requirements if the
other requirements are shown to provide emissions reductions equivalent to
the required equipment standards. The technical problem cited by the
commenter in Docket Item No. IV-D-6 is addressed in Sections 4.8 and 4.12.
3.5.5 Performance Standard for Sampling Systems
Comment:
One commenter (IV-D-6) recommended that a performance standard be
established for sampling systems. He felt there were better methods of
sampling which could not be used under the proposed regulations which could
be used without discussion if a performance standard were established.
3-24
-------
Response:
As explained in the preamble to the proposed standards (46 FR 1145), an
emission limit was not specified because measuring mass emissions from each
sampling system would require bagging each system, a measurement method
which is time-consuming, costly, and impractical. The standard for
emissions from sampling systems concerns the material purged prior to
collection of the sample; it does not cover the sample material. Emissions
from this purged material cannot be easily measured. Furthermore, a no
detectable emissions limit is not feasible because some VOC could be emitted
during the transfer of sample to a collection device or during its disposal.
The standard for sampling systems was proposed as an equipment
standard. The final regulation has been modified, slightly to take the form
of an operational standard. Essentially, any sampling system which collects
purged material and returns it to the process or disposes of it properly and
which eliminates emissions of purged material to the atmosphere would be
acceptable. EPA recognizes the fact that some VOC may be emitted when
disconnecting a sample container. These small amounts of emissions are
allowed, but the sample purge must be destroyed or recycled to the process.
Discarding the sample purge to an open drain system is not allowed under the
standards. Sampling systems are treated in more detail in Section 4.9.
3.5.6 Emission Limit vs. Concentration Limit
Comment:
Several commenters (IV-D-16; IV-D-7; IV-D-17) pointed out that a
performance standard need not be in the form of an emission limit. They said
that a performance standard could also take the form of a concentration
limit.
Response:
Performance standards establish a numerical emission limit that place
an upper limit on the amount of pollutant mass allowed from a source. The
amount of mass is generally set as a mass rate per unit time, as a mass per
unit of production, or as a mass per unit of exhaust gases (concentration).
In some cases, these limits can be closely correlated with other measure-
ments such as opacity as is the case for particulate matter. Opacity and
3-25
-------
mass emission rates are related by physically measureable parameters found
in exhaust gases from many industrial processes. These parameters can be
reasonably defined for opacity and mass emission rates.
These parameters are not precisely defined for concentrations measured
by Reference Method 21 and fugitive emission rates. The variability among
process units makes it impracticable to set a limit or allowable percent
leaking for all emission sources. It may be practicable to set a lower
limit for valves, however, to remove unreasonable costs. If the variability
for valves is controlled by operators to reduce the longer-term average
percent leaking to less than 2.0 percent, then routine monthly leak detec-
tion and repair have unreasonable costs. And a performance level can be
s
established to exclude units with too few leaks for control with reasonable
cost. Therefore, a performance standard generally is not technically
practicable for many sources of fugitive emissions, but it is possible to
set a lower limit which excludes low-leaking plants.
The intent of the standards is to reduce fugitive VOC emissions by
finding and repairing fugitive emission leaks or preventing them from
occurring. It is not the intent to allow fugitive emission sources to
continue leaking, thereby emitting VOC. Standards have been provided that
allow the owner or operator to meet the objectives of the standards with a
variety of control options. For example, EPA has provided a performance
standard for valves by setting a performance level of 2 percent leaking that
can be met as an alternative to the normal valve standard. EPA believes
that this was what the commenter was seeking in making his recommendations.
3.5.7 Complexity of the Standards
Comment:
One commenter (IV-D-11) asked that the regulations be simplified. He
asked if the desired objectives could be achieved with simpler regulations.
Response:
The basic concept underlying the fugitive VOC emission standards is one
of finding leaks and repairing them or of preventing them from occurring.
The concept is simple and no one has challenged the desirability or appro-
priateness of this philosophy. Complexity arises in incorporating this
3-2G
-------
philosophy in the regulatory framework. For regulatory purposes, it is
necessary to lay out rules governing methods, frequency, and degree of
control. It is also necessary to write the rules in sufficient detail to
prevent misunderstandings between industry and EPA. The required detail is
increased by considering the range of affected facilities to which the
standard applies. Still further complexity is introduced in endeavoring to
fit the regulation to various situations which provide more effective
control than the requirements in the standards, and thereby allow industry
sufficient flexibility.
An optimum balance between simplicity and flexibility is desirable.
EPA has sought to achieve this balance in the standards. For example, since
proposal, EPA has provided a work practice standard for pumps and has
simplified and clarified alternative standards for valves.
3-27
-------
-------
4. CONTROL TECHNOLOGY
This section addresses comments received on the technical aspects of
the control technologies considered for reducing fugitive VOC emissions from
SOCMI. Twelve technology areas are discussed in the following sections:
4.1 Flares
4.2 Leak Detection and Repair Programs
4.3 Leak Definition
4.4 Safety Considerations
4.5 Pressure Relief Devices
4.6 Combustion Device
4.7 No Leak Equipment
4.8 Dual Seals
4.9 Sampling Systems
4.10 Closed Vent System
4.11 Open-ended Lines
4.12 Reciprocating Pumps and Compressors
4.1 FLARES
Several commenters expressed the desire to use flares as alternatives
to enclosed incinerators or vapor recovery systems. The comments focused on
six areas of concern: (1) data base support of low flare efficiency;
(2) high efficiencies reported for flares on refinery gases; (3) alternative
flare designs for low-flow applications; (4) safety considerations in
choosing control systems; (5) equivalency; and (6) Executive Order 12291.
Each of these areas is discussed in the following comments and responses.
But an explanation of EPA's analysis and final decision on flare usage for
fugitive VOC control is presented first because all the responses to the six
areas of concern are prefaced by this analysis and final decision.
At proposal, flares were not considered an acceptable control option
for elimination of fugitive VOC emissions. The results of studies that were
available were considered inapplicable to the streams to be controlled in
4-1
-------
SOCMI. In some studies the flare design was not representative of flares in
the industry. In others the analytical method was questionable. At that
time no approved method for measuring flare efficiency (evaluating flare
performance) was available .
Theoretical calculations indicated that flare efficiency could be as
low as 60 percent for destruction of VOC in low-flow intermittent streams
sent to a large flare. This efficiency was cited in several background
documents (Ethylbenzene/Styrene, Benzene Fugitive, SOCMI Fugitive VOC) and
served as a primary consideration in not allowing the general use of flares.
However, this theoretical computation was based on assumptions that may not
be applicable to the design situation under study.
The use of flares, therefore, was reconsidered for the SOCMI standards.
Commenters pointed out potential operational difficulties associated with
the use of incinerators that could be avoided with the use of flares. A
major difficulty seen was in designing systems for the low-volume and
intermittent flow to the control device. In addition, consideration was
given to the extensive use of flares by industry to handle emergency
releases. Since flares are currently in widespread use in SOCMI, they
represent a large investment in control by the industry.
The following presents a review of flares and operating conditions used
in five studies of flare combustion efficiency. Each study can be found in
complete form in the docket, .
Palmer (IV-M-8) experimented with a 1/2-inch ID flare head, the tip of
which was located 4 feet from the ground. Ethylene was flared at 50 to
250 ft/sec at the exit, (0.4 x 106 to 2.1 x 106 Btu/hr). Helium was
added to the ethylene as a tracer at 1 to 3 volume percent and the effect of
steam injection was investigated in some experiments. Destruction
efficiency (the percent ethylene converted to some other compound) was
97.8 percent.
Siege! (IV-D-17) made the first comprehensive study of a commercial
flare system. He studied burning of refinery gas on a commercial flare head
manufactured by Flaregas Company. The flare gases used consisted primarily
of hydrogen (45.4 to 69.3 percent by volume) and light paraffins (methane to
4-2
-------
butane). Traces of H^S were also present in some runs. The flare was
operated from 0.03 to 2.9 megagrams of fuel/hr (287 to 6,393 Ib/hr), and the
maximum heat release rate was approximately 235 x 10 Btu/hr. Combustion
efficiencies (the percent VOC converted to 002) averaged over 99 percent.
Lee and Whipple (IV-M-18) studied a bench-scale propane flare. The
flare head was 2 inches in diameter with one 13/16-inch center hole
surrounded by two rings of 16 1/8-inch holes, and two rings of 16 3/16-inch,
holes. This configuration had an open area of 57.1 percent. The velocity
through the head was approximately 3 ft/sec and the heating rate was
0.3 M Btu/hr. The effects of steam and crosswind were not investigated in
this study. Destruction efficiencies were 99.9 percent or greater.
Howes, et al. (IV-A-27) studied two commercial flare heads at
John Zink's flare test facility. The primary purpose of this test (which
was sponsored by the EPA) was to develop a flare testing procedure. The
commercial flare heads were an LH air-assisted head and an LRGO (Linear
Relief Gas Oxidizer) head manufactured by John Zink Company. The LH flare
burned 2,300 Ib/hr of commercial propane. The exit gas velocity based on
the pipe diameter was 27 ft/sec and the firing rate was 44 x 10 Btu/hr.
The LRGO flare consisted of 3 burner heads 3 feet apart. The 3 burners
combined fired 4,200 Ibs/hr of natural gas. This corresponds to a firing
rate of 83.7 x 10 Btu/hr. Steam was not used for either flare, but the :
LH flare head was in some trials assisted by a forced draft fan. Combustion
efficiencies for both flares during normal operation was greater than
99 percent.
An excellent detailed review of all four studies was done by Joseph,
et al. (IV-M-20), and a summary of the studies is given in Table 4-1. A
fifth study by McDaniel, et al. (IV-A-32) determined the influence on flare
performance of mixing, Btu content and gas flow velocity. A steam-assisted
flare was tested at the John Zink facility using the procedures developed by
Howes. The test was sponsored by the Chemical Manufacturers Association
(CMA) with the cooperation and support of the EPA. All of the tests were
with an 80 percent propylene, 20 percent propane mixture diluted as required
with nitrogen to give different Btu/scf values. This was the first work
4-3
-------
TABLE 4-1. FLARE EMISSION STUDIES COMPLETED AS OF OCTOBER 1982
-fi
1
-Ti
Investigator
Palmer (1972)
lee A Uh1|>ple (1981)
Slegel (19BO)
Howes et al. (1981)
HcOanlel et al. (1982)
Sponsor
E.I. du Pont
Union Carbide
Ph.D. Dissertation
University of Karlsruhe
EPA
CMA-EPA
Docket Ho.
IV-H-8
IV-M-1B
IV-D-17
'V-A-Z7
IV-A-32
Flare Tip Design
0.5" dla.
Discrete Hales In
2" dla. cap.
Comnerclal Design
(27.6- die. steam)
Commercial Design
(6" dla. air assist)
Comnerclal Design H.P.
(3 tips t 4* dla.)
Comnerclal Design
Flared Gas
Ethyl one
Propane
=501 H, plus
light hydro-
carbons
Propane
Natural Gas
Propylene
Throughput
10° Btu/hr
0.4 - 2.1
0.3
49 - 17B
44
28 (per tip)
0.01 - 57
Flare
Efficiency 1
97. B - >9g
>99.9
>99
>99
>99
83 - 99.9
-------
which determined flare efficiencies at a variety of nonideal conditions
where lower efficiencies had been predicted. All previous tests were of
flares which burned gases which were very easily combustible and did not
tend to soot. This was also the first test which used the sampling and
chemical analysis methods developed for the EPA by Howes.
The steam-assisted flare was tested with exit flow velocities ranging
from 0.02 to 60 ft/sec, with Btu contents from 200 to 2,183 Btu/scf and with
steam to gas (weight) ratios varying from 0 (no steam) to 6.8611. Steam-
assisted and air-assisted flares were tested with fuel gas heat contents as
low as 300 Btu/scf. Flares without assist were tested down to 200 Btu/scf.
This efficiency was also found to be achievable for air-assisted flares
combusting gases with heat contents over 300 Btu/scf and with exit gas
velocities below a maximum value (depending upon the heat content of the gas
stream). All of these tests, except for those with very high steam to gas
ratios, showed combustion efficiencies of over 98 percent. Flares with high
steam to gas ratios (about 10 times more steam than that required for
smokeless operation) had lower efficiencies (69 to 82 percent) when
combusting 2,183 Btu/scf gas.
After consideration of the results of these five tests, EPA has
concluded that 98 percent combustion efficiency can be achieved by steam-
assisted flares with exit flow velocities less than 60 ft/sec combusting
gases with heat contents over 300 Btu/scf and by flares operated without
assist with exit flow velocities less than 60 ft/sec gases with heat
contents over 200 Btu/scf. Flares are not normally operated at the very
high steam to gas ratios that resulted in low efficiency in some tests
because steam is expensive and operators make every effort to keep steam
consumption low. Flares with high steam rates are also noisy and may be a
neighborhood nuisance.
EPA has a program under way to determine more exactly the efficiencies
of flares used in the petroleum/SOCMI industry and a flare test facility has
been constructed. The combustion efficiency of four flares (1 1/2 inches to
12 inches ID) will be determined and the effect on efficiency of flare
operating parameters, weather factors, and fuel composition will be
established. The efficiency of larger flares will be estimated by scaling.
4-5
-------
According to the current knowledge of flare design, the best available
flare design (i.e., the state-of-the-art flare design) is the smokeless
flare. The smokeless flare introduces air into the flame by injection of
steam or air. This injection of steam or air increases the mixing of the
flared compounds within the flame zone thereby increasing the destruction of
the compounds. Smoking flares are environmentally less desirable because
they emit particulate. It is difficult, however, to maintain smokeless
operation unless the off-gas flow to the flare is constant. When the
off-gas flow rate increases, there is a short period of time before the
smoke sensor responds and additional steam (or air) reaches the flare tip.
During this period,.the flare smokes. Smoking may also occur during large
emergency discharges because insufficient steam (or air) is available in the
plant to make these infrequent discharges nonsmoking. A number of
engineering practices currently used in industry help to achieve continuous
smokeless operation. These include staged elevated flares, dual flare tips
(small tip for low-flow, large tip for emergency releases), and continuous
flare gas recovery systems. These systems are further discussed later in
this section.
Taking all these factors into consideration, EPA decided to allow use
of smokeless flares operated with a flame present to control fugitive VOC
emissions in SOCMI. In order to ensure that the smokeless flare operates
with a flame present, the flare's pilot light is to be monitored with an
appropriate heat sensor, such as a thermocouple. To ensure smokeless
operation, visible emissions from a flare would be limited to less than five
minutes in any 2-hour period. In addition, steam-assisted flares would have
to be operated with exit velocities less than 60 ft/sec combusting gases
with heat contents greater than 300 Btu/scf. Flares operated without assist
would have to be operated with exit velocities less than 60 ft/sec combus-
tion gases with heat contents greater than 200 Btu/scf. Air-assisted flares
would have to be operated with exit velocities below a maximum value,
depending upon the gas heat content which must be greater than 300 Btu/scf.
Flares operated within these requirements are considered as acceptable
alternatives to enclosed combustion devices (incinerators, boilers, process
4-6
-------
heaters) and vapor recovery systems such as carbon adsorbers and condensa-
tion units. They may be applied to control of emissions from pump seals (or
degassing reservoirs), compressor seals (or degassing reservoirs), and
pressure relief devices.
As mentioned above, EPA has a program under way to determine the
effectiveness of flares not studied to date. As this data and information
are collected and evaluated, EPA plans to update the requirements for
flares. It is not expected that the requirements would become more restric-
tive. Until the requirements are updated, plant owners and operators are
allowed to determine whether other flare systems are equivalent to the
systems required in the standards.
Comment:
One commenter (IV-F-1, No.4, p.61) objected to EPA's taking the
position, without supporting data, that flares do not achieve good control.
Another commenter (IV-D-16) stated that EPA had presented no data in support
of the argument that flares may only achieve 60 percent efficiency. Another
commenter (IV-D-17) agreed, adding that EPA has stated without qualifica-
tions, on page 4-19 of the proposed NSPS VOC Fugitive Emission Sources in
SOCMI* [sic], that flaring efficiency is 60-99 percent. The commenter
quoted the following from page 4-20 of this document: "This efficiency
(60 percent) reflects the fact that many flare systems are not of optimum
design. As a result, flares that are designed to handle large volumes of
vapors associated with overpressure releases are used to handle Tow-volumes
of fugitive emissions. With such designs, optimum mixing is not achieved
because the vent gas exit velocity is low and large flares generally cannot
properly inject steam into low-volume streams."
In a previous letter, the commenter (IV-D-17) questioned the relation-
ship between steam injection into a low-volume stream and burning
efficiency. He pointed out that, even though improper balance of steam may
cause flare smoking, low steam injection does not appear to influence
burning efficiencies of flares.
*VOC Fugitive Emissions in the Synthetic Organic Chemical Manufacturing
Industry - Background Information for Proposed Standards of Performance.
4-7
-------
Response:
In the Background Information Document for Ethylbenzene/Styrene (EB/S)
(EPA-450/3-79-035a), the efficiency of a flare system operated alone to
control a small vent stream was estimated to be 60 percent. The estimates
of destruction rates were based on the "Afterburner Systems Study" by Shell
Development Company (II-I-13) and represented a generalized correlation for
hydrocarbons combusted at 1410°F. ..., .
Further actual flare measurement results have become available, most
notably from the CMA-EPA study, since the 60 percent theoretical estimate
was made. In the CMA-EPA study, steam-assisted flares and flares operated
without assist were investigated over a wide range of exit velocity,
composition, and flare gas heat content conditions. After review of
available flare efficiency data, EPA has concluded that smokeless flares
operated with a flame present and exit velocities less than 60 ft/sec with
flare gas heat contents greater than 300 Btu/scf for steam-assisted flares
or exit velocities less than 60 ft/sec and flare gas heat contents greater
than 200 Btu/scf for flares operated without assist are acceptable alterna-
tives to enclosed combustion devices or vapor recovery systems.
Comment:
Several commenters argued that the use of flares should not be excluded
as. a means of controlling barrier fluid degassing emissions. Two commenters
(IV-D-18; IV-D-26) pointed out that flares are common in most SOCMI
processes and that a final decision on the use of flares should not be made
until the current John Zink flare study by Battelle Memorial Laboratories
has been completed. One of these commenters (IV-D-26) also maintained that,
although flare technology may not be suitable for the burning of certain
chemicals (e.g., chlorinated hydrocarbons), the use of this technology
should not be precluded where appropriate. This commenter further stated
that there is evidence that properly designed and operated flares will
achieve 95 to 99 percent efficiency.
Several commenters (IV-D-7; IV-D-15; IV-D-16; IV-D-17; IV-D-23;
IV-D-48) cited the German flare study Degree of Conversion of Flare Gas in
Refinery High Flares by K.D. Siegel as the most recent study on flare
4-8
-------
systems. This study indicated efficiencies for flares to be greater than
95 percent.
One of these commenters (IV-D-17) specifically referenced data from the
study to indicate better than 95 percent efficiency for the almost 1300 test
samples measured over a wide range of operating conditions: 42 mass rates,
23 flare gas densities, and 114 steam/gas ratios. Conversion efficiency was
found to be independent of mass flow, wind speed, or gas composition for the
refinery gases studied. The commenter (IV-D-17) had previously submitted "
Dr. Siege!'s dissertation as a total rebuttal to EPA's position on flare
efficiencies. He presented the dissertation's conclusions as:
(1) In soot-free flare flames, the organically-bound carbon of the
flare gas is converted to carbon dioxide to at least 99 percent.
(2) The emission factor for flames containing soot or soot-free
flames, independent of the optical flame picture, comprises a maximum
of one percent of the organically-bound carbon in the flare gas.
(3) The mass concentration of^the organically-bound carbon at the
flame end is less than 50 mg/m , even in the case of sooty flare
flames.
(4) The bulk of the organically-bound carbon at the flame end consists
of methane and acetylene.
(5) The nitrogen oxide emission of flare flames is low.
Also citing Dr. Siegel's work and the John Zink study by Battelle, one
commenter (IV-D-48) stated that minimum efficiencies for flares are greater
than 95 percent. Another commenter (IV-D-15) agreed and acknowledged that
EPA has conducted an evaluation of Siegel's work which concluded that
universal application of the 99 percent conversion to all flares is
doubtful. He stated that, even though there are questions regarding
validity and interpretation of results, these questions should not preclude
the use of flares as acceptable VOC emissions control systems.
Another commenter (IV-D-17) also noted that Battell'e Memorial Laborato-
ries has conducted a study for EPA to demonstrate measuring techniques for
use at flare towers. The study was conducted over a three-day period using
a John Zink facility flaring propane. Although the test has-long been
4-9
-------
completed, the Battelle study has not been made available, even in draft
form, for public review and comment. The commenter stated that although not
a specific objective of the study, data exist demonstrating that the flare
system was able to achieve a destruction efficiency of greater than
95 percent even with a smoking flare.
One commenter (IV-D-41) suggested that, as a viable alternative, the
standards allow any combustion device providing that 95 percent efficiency
can be maintained. He wrote that flares would be precluded with the current
requirement of 0.75 seconds as a minimum VOC residence time.
Response:
As discussed initially in this section, EPA has determined that
smokeless flares operated with a flame present and exit velocities less than
60 ft/sec with flare gas heat contents greater than 300 Btu/scf for steam-
assisted flares or exit velocities less than 60 ft/sec and flare gas heat
contents greater than 200 Btu/scf for flares operated without assist may be
considered as acceptable alternatives to enclosed combustion devices or
vapor recovery systems for controlling fugitive VOC emissions .in SOCMI.
Comment:
Although disagreeing with EPA's 60 percent efficiency statement, one
commenter (IV-D-17) stated that there are a number of engineering practices
currently in use within industry to deal with flaring low-flow continuous
emissions. One such system involves the use of staged flare systems where a
small diameter flare is operated in tandem with a large diameter flare. The
system is designed such that the small flare takes the continuous low-flow
releases and the larger flare accepts emergency releases. A second system
involves the use of a separate conveyance line to the flare tip for
continuous low-volume, low-pressure releases. A third system, sometimes
used in conjunction with either of the above systems, involves the use of
continuous flare gas recovery. In the latter system, a compressor is used
to recover the continuously generated flare gas base load. The compressor
is sized to handle the base load and any excess gas is flared.
4-10
-------
Response:
The commenter is correct in pointing out systems that provide smokeless
flare operation. The techniques noted are particularly applicable to
handling low-flow streams. Use of a separate conveyance line to the flare
tip for continuous low-flows would reduce explosion and flammability
potential resulting from air seepage into large vent lines. But smaller
lines may necessitate the addition of an auxiliary fan to overcome the
increased pressure drop. The third system described by the commenter is an
effective means of recovering flare header gases and is currently used by
industry.
Comment:
A group of commenters (IV-D-17; IV-D-6) cited that several emergency
situations, including releases from pressure relief devices, exist in which
enclosed combustion devices would be unable to handle flow and pressure "
loads safely. They contended that flare systems are designed to handle such
widely ranging feed conditions as cold liquids and hot gases. One of the
commenters (IV-D-6) stated that his company avoids the use of enclosed
incinerators, specifying flares instead, for control of relief valve
emissions since incinerators involve complex design to supply adequate
combustion air and to handle widely varying flows.
One commenter (IV-D-34) felt that flares should not be precluded from
use to control emissions. He agreed with another commenter (IV-D-17) that
the proposed system is unsafe, wastes energy, and is not cost effective. He
further remarked that his studies show adequate combustion efficiencies from
flares to meet the present requirements. And he noted that, where enclosed
burning is currently required, two enclosed incineration systems are
maintained simultaneously at operating temperature to avoid destruction of
the units' ceramic lining.
Response:
The new source standards for fugitive VOC emissions in SOCMI do not
cover situations such as emergency releases from pressure relief devices.
In fact, the standard for relief devices requires that a performance level
of no detectable emissions (less than 500 ppmv above background) be met;
4-11
-------
there are no equipment requirements for fugitive emissions from relief
devices. One method of meeting this performance standard is to pipe the
relief device to a flare which is a common practice in the industry.
There are other potential emergency situations, such as catastrophic
pump seal failure that must be considered. In such cases, as the commenters
noted, incinerators present a difficult design and operation consideration
resulting from the rapid changes in vent flow rate and temperature. For the
reasons noted in the introduction to this section, EPA has.decided to allow
the use of smokeless flares for controlling fugitive VOC emissions in SOCMI,
provided that the flares are operated with a smokeless flame present.
Steam-assisted flares would have to be operated with exit velocities less
than 60 ft/sec combusting gases with heat contents greater than 300 Btu/scf.
Flares operated without assist would have to limit exit velocities to less
than 60 ft/sec combusting gases with heat contents greater than 200 Btu/scf.
Air-assisted flares would have to operate below a maximum exit velocity
dependent upon the gas heat content which must be greater than 300 Btu/scf.
Other emergency conditions may occur with control device systems. For
example, during failure of a compressor, a flare system may be used to
combust the process fluid from the compressor. When this occurs, the flare
may not be operating in compliance with the requirements in the standards
for flares. Such conditions may be representative of startups, shutdowns,
and malfunctions as discussed in the General Provisions of 40 CFR Part 60.
However, at all times, including periods of startup, shutdown, and malfunc-
tion,* owners and operators shall, to the extent practicable, maintain and
*"Startup" means the setting in operation of an affected facility for any
purpose.
"Shutdown" means the cessation of operation of an affected facility for
any purpose.
"Malfunction" means any sudden and unavoidable failure of air pollution
control equipment or process equipment or of a process to operate in a
normal or usual manner. Failures that are caused entirely or in part by
poor maintenance, careless operation, or any other preventable upset
condition or preventable equipment breakdown shall not be considered
malfunctions.
4-12
-------
operate any affected facility including associated air pollution control
equipment in a manner consistent with good air pollution control practice
for minimizing emissions [40 CFR 60.11(d)]. Determination of whether
acceptable operating and maintenance procedures are being used will be based
on information available to EPA which may include, but is not limited to,
monitoring results, opacity observations review of operating and maintenance
procedures, and inspection of the source. It should be noted that closed
vent systems and control devices used in complying with the standards are
part of the affected facility.
Comment:
One comtnenter (IV-D-17) expressed the understanding that it is EPA's
intention not to preclude flare systems but to require any company choosing
such a control system to demonstrate equivalency pursuant to §60.484. He
wrote that this is an unreasonable, costly, and time-consuming process in
light of the significant and representative data and information that
industry has already submitted demonstrating the equivalency of flare
systems. The commenter pointed out that EPA has an absolute obligation to
include flare systems as an appropriate control system for purposes of
regulating fugitive emissions. He further added that the clear language of
the proposed regulations contradicts statements made by EPA's OAQPS staff,
the preamble, and various support documents that fugitive emissions can be
transported by a closed vent system to an enclosed combustion device or
vapor recovery system, as well as a flare system, and other equivalent
control devices. Failure to correct this inconsistency and revise the
regulatory language could result in unanticipated enforcement initiatives
based on the language of the proposed regulations.
In a previous letter that had been attached, the commenter (IV-D-17)
accused EPA' of taking the position that the burden of proof of high
efficiency of flares is on the industry. He disagreed, arguing that since
flares are standard abatement devices of long-standing in both the chemical
and petroleum refining industries, the burden of proof, with data, is on
EPA. He added that SOCMI has a heavy investment in flares and will strongly
resist EPA's position that flares are not acceptable emission control
4-13
-------
devices. In a later letter, the commenter (IV-D-50) stated that flares are
common and efficient control devices used in SOCMI. He was concerned that
flares had been excluded from use on the basis of lack of efficiency data.
Another commenter (IV-D-48) stated that, in light of the data from the
German flare study by Siege! and the John Zink flare study by Battelle
Memorial Laboratory, an effective equivalency determination had been made.
Therefore, the Administrator should authorize the use of flares to control
fugitive VOC emissions.
Response:
After considering this and previous comments on the use of flares for
controlling fugitive VOC emissions in the SOCMI, EPA has determined that
smokeless flares operated with flames present and exit velocities less than
60 ft/sec with flare gas heat contents greater than 300 Btu/scf for steam-
assisted flares or exit velocities less than 60 ft/sec and flare gas heat
contents greater than 200 Btu/scf for flares operated without assist are
acceptable alternatives to enclosed combustion devices or vapor recovery
systems. In addition, air-assisted smokeless flares may be used provided
they operate below a maximum exit velocity that is based on gas heat content
which must also be greater than 300 Btu/scf. Their use does not require
further demonstration of equivalency. The determination to allow smokeless
flares was based on EPA's belief that smokeless flares can achieve about
98 percent efficiency and techniques are well established that help flares
maintain smokeless operation.
Comment:
One commenter (IV-D-17) wrote that, in order for EPA to be consistent
with the spirit, if not the express language, of the recently issued
Executive Order No. 12291, the Agency is under an affirmative duty to allow
those control options that data demonstrate achieve the environmental objec-
tives of the regulation at a lower cost to industry. In this regard EPA
should not preclude technically sound and cost-effective regulatory options,
such as flares, unless an administrative record is established that clearly
documents that these cost-effective regulatory options will offset a
significant environmental benefit that could otherwise result.
4-14
-------
Response;
Since proposal, the determination was made that smokeless.flares should
be allowed as acceptable control devices, when steam-assisted flares are
operated with exit velocities less than 60 ft/sec with flare gas heat
contents greater than 300 Btu/scf or when flares without assist are operated
with exit velocities less than 60 ft/sec with flare gas heat contents
greater than 200 Btu/scf or when air-assisted flares are operated below a
maximum exit velocity based on gas heat content which must be below
300 Btu/scf.
4.2 LEAK DETECTION AND REPAIR PROGRAMS
A number of comments were received concerning the leak detection and
repair program which is incorporated in the valve standard. The main
subject areas of the comments were the monitoring interval, the repair
requirements, and the estimates made by EPA of time required for leak
detection and repair. Other comments gave suggestions for alternate
approaches and pointed out potential problems which might be encountered in
performing the leak detection and repair program.
4.2.1 Monitoring Interval
Several commenters requested that the monitoring intervals for the leak
detection and repair program be lengthened. Various monitoring intervals
were recommended and a variety of reasons were cited for the recommended
changes. A recommendation for a shorter monitoring interval was also made.
Comment:
Some commenters (IV-F-1, No.l; IV-F-1, No.4; IV-D-17; IV-D-40; IV-D-48)
challenged the occurrence/recurrence relationship assumed by EPA in devel-
oping the monitoring strategy. One of these commenters (IV-F-1, No.4) said
that EPA had assumed a complex leak occurrence rate which is biased to favor
monthly monitoring. He stated that using a linear leak occurrence rate
would show quarterly monitoring to achieve the same results as the proposed
program. He cited recent data published by EPA; in An Evaluation of
Maintenance for Fugitive VOC Emissions ControT whtch seemed to support a
linear leak occurrence rate.
4-15
-------
Another commenter echoed these objections in two sets of comments
(IV-F-1, No.l; IV-D-17) and said that EPA had not offered any rationale or
supportable basis for imposing a monthly monitoring program. The commenter
objected to the assumptions in the BID and said that, until representative
data were made available, it would be logical to assume a linear leak
occurrence rate and that the recurrence rate is proportional to the
occurrence rate.
One commenter (IV-D-48) recommended quarterly .monitoring for valves
based on the assumption that leak occurrence is linear and the recurrence
rate for SOCMI is much lower than that for refineries.
Response:
The commenters are challenging two technical assumptions made by EPA in
the development of the standard for valves. The first one is the rate at
which the number of leaks found in a process unit will increase with time.
The estimates of leak accumulations with time as shown in the BID on
page 4-15 are shown here in Table 4-2.
TABLE 4-2. LEAK OCCURRENCE/RECURRENCE RATE ESTIMATES FROM BID
Monitoring
Interval
1 month
3 months
1 year
a
nm
m
0.1NC
0.2N
0.4N
b
nm
m
0.05N
0.1N
0.2N
an = Total number of leaks which occur, recur, and remain between
monitoring intervals.
nm = Average number of leaks ewer the monitoring interval.
N = Total number of sources at or above the action level.
As pointed out by the commenters, the numbers assumed form a non-linear
relationship with time for accumulated leaks. However, the numbers include
4-16
-------
(as noted in the footnote) the number of leaks which occur, recur, and
remain between monitoring intervals. They are not simple occurrence rate
estimates.
The commenters are partially correct in stating that occurrence rates
should be linear. Occurrence rates have been found to be essentially linear
in recent SOCMI studies as reported in An Evaluation of Maintenance for
Fugitive VOC Emissions Control (IV-A-7.) In this report, the leak
occurrence rate is described by an exponential distribution model and the
leak recurrence rate is described by a mixed distribution model, which
incorporates an exponential model to describe long-term leak recurrences.
Both models are non-linear in format. But, as applied to the data collected
in the SOCMI studies, the models result in a nearly linear relationship with
time. In fact, only slightly non-linear leak occurrence and recurrence
rates are noted when considering a monitoring interval of one year.
Since proposal, analysis of the results of the SOCMI maintenance study
(IV-A-7) led to the development of a new model describing leak detection and
repair programs. This model is described in detail in Docket Item
No. IV-A-22 and in the recently distributed AID (III-B-2). The results of
the model evaluating various possible leak detection and repair programs for
valves and using inputs from the SOCMI maintenance study are shown in
Figure 4-1. The fraction of valves operating improperly (occurring,
recurring, and unrepairable) is presented as a function of monitoring
interval.
The second assumption being challenged is the relationship of valve
leak occurrence to valve leak recurrence. As explained in the preamble to
the proposed regulation (46 FR 1146), the proposed monitoring program for
valves included an allowance to monitor valves that leak infrequently on a
quarterly basis. This was based on the assumption that recurrence of leaks
is a significant contributor to the total number of leaks. Data from SOCMI
fugitive emission studies do not conclusively confirm this assumption.
Valve leak occurrence and recurrence rates are shown in Table 4-3.
These numbers indicate the difference between occurrence and recurrence
rates.
4-17
-------
0.15
s.
-------
TABLE 4-3. OCCURRENCE RATES AND RECURRENCE RATES FOR
VALVES DETERMINED IN SOCMI UNITS
Occurrence
Recurrence
30 Days
(Percent)
1.3
17.2
90 Days
(Percent)
3.8
23.9
180 Days
(Percent)
7.4
32.9
Data taken from Evaluation of Maintenance for Fugitive VOC Emission
Control (IV-A-7).
Recurrence rates are most evident within the first five days after attempted
repair. Beyond that time period, the recurrence rate is essentially equal
to the occurrence rate. Therefore, it is appropriate to discuss recurrence
only in terms of early recurrence, i.e., leaks Which recur within the first
five days after attempted repair. Also, even though the recurrence rate
seems high compared to the occurrence rate, it must be applied to only those
valves on which repair was attempted, not the entire valve population.
This, coupled with an assumed lower emission factor for early recurrence
leaks, reduces the impact of leak recurrence compared to leak occurrence.
However, leak recurrence does contribute significantly to the total number
of leaks.
In selecting the basis of the promulgated standards, EPA mainly
considered two regulatory alternatives for valves monitoring at monthly
intervals and monitoring at quarterly intervals. The incremental cost of
monthly versus quarterly monitoring was judged to be reasonable for the
additional emission reduction achieved by monthly valve monitoring.
Consequently, monthly monitoring was selected as the basis of the standard.
This judgement was based on emission reductions and costs calculated at the
rate at which valve leaks typically occur at SOCMI process units. However,
EPA recognizes that some valves have lower leak occurrence rates than
others. Monthly monitoring of valves that do not leak for 2 consecutive
months was judged to be unreasonable when compared to the additional
emission reduction achieved by monthly monitoring over quarterly monitoring.
Therefore, although EPA is proposing that leak detection and repair programs
4-19
-------
include monthly monitoring for valves, the standard would allow quarterly
monitoring for valves which have been found not leaking for 2 .successive
months. .
Comment:
Another argument given for extending the monitoring interval was that
EPA has improperly calculated the emission reductions achievable under
Alternative IV.' One cbmrnenter (IV-F-1, No.1)'said that quarterly monitoring
would result in a 97.8 percent control efficiency and not 86 percent as
reported in the BID. He said that an analysis had been submitted previously
(IV-D-17) which 'documented this improper"calculation. The commenter urged
that EPA adopt quarterly monitoring since his calculations showed that,a
quarterly monitoring interval would achieve the desired environmental goals
at a lower cost to industry.
Response:
The commenter did submit the referenced analysis on June 30, 1980. It
may be found in 'Docket Item No. II-D-72. The same analysis was resubmitted
as a part of written comments on the proposed standards (IV-D-17).
The commenter's analysis differs from EPA1s in two major areas. The
first major area of difference is in the emission sources included in the
fugitive emissions estimate calculations. EPA's methodology includes
contributions from fugitive emission sources which are not regulated as well
as contributions from the emission sources which are regulated. This
methodology is clearly documented in an example on page 7-6 of the BID. As
the table shows, the contributions from heavy liquid equipment and flanges
are in'cluded.
On the Other hand, the commenter neglected to include the contribution
to emissions from fugitive emission sources which are not controlled by the
regulation. -
" The second major area of difference is in assumptions made. The BID
described a method for estimating control efficiency for a leak detection
and repair program. The model describing such a program incorporated four
factors defined as following.
4-20
-------
Reduction Efficiency =AxBxCxD
where:
A = Theoretical Maximum Control Efficiency = fraction of total mass
emissions for source type with VOC concentrations greater than the
action level.
B = Leak Occurrence and Recurrence Correction Factor = correction
factor to account for sources which start to leak between
inspections (occurrence); for sources which are found to be
leaking, are repaired and start to leak again before the next
inspection (recurrence); and for known leaks which are not
repaired.
C = Non-Instantaneous Repair Correction Factor = correction factor to
account for emissions which occur between detection of a leak and
subsequent repair; that is, repair is not instantaneous.
D = Imperfect Repair Correction Factor = correction factor to account
for the fact that some sources which are repaired are not reduced
to zero emission levels. For computational purposes, all sources-;
which are repaired are assumed to be reduced to a 1000 ppmv
emission level.
The commenter and EPA made different assumptions for these four factors.
The commenter assumed a C factor of 4.5 days; EPA assumed 7.5 days.
EPA assumed for the D factor that valves would be repaired on the average to
1,000 ppmv. The commenter assumed that 25 percent of the valves would be
repaired to a level of 0 ppmv, and the remainder would be repaired to
1000 ppmv. The commenter assumed a linear increase for B, while EPA assumed
a non-linear function. (This factor was discussed in the previous comment
and response.)
Two areas of differing assumptions caused the difference in the control
efficiency estimates: (1) consideration of both controlled and uncontrolled
fugitive emission sources and (2) assumptions regarding the effectiveness of
leak detection and repair programs (primarily with respect to leak occur-
rence rate estimates). When these differences are taken together, the
effect on overall control effectiveness is compounded. The results indicate
a control effectiveness of 97.8 percent (using the commenter1s assumptions)
vs. 86 percent (under EPA's assumptions).
4-21
-------
The LDAR model describing leak detection and repair programs (IV-A-22)
has been used to evaluate the overall effectiveness of such programs for
valves. Using the inputs to the model detailed in the AID, valve leak
detection and repair programs were evaluated for the average SOCMI unit, as
well as for the three process types tested in.SOCMI. The quarterly .moni-
toring with monthly follow-up program that was part of the proposed
standards results in an overall control efficiency of 57 percent for an
average SOCMI unit. The impact of leak detection and repair programs on the
overall effectiveness of the new source standards was discussed in the AID
and is presented here in Section 3.3.
Comment:
In three sets of comments (IV-D-17; IV-D-26; IV-D-48).it was argued
that since leak frequencies in SOCMI units are less than in refineries, the
monitoring interval should be lengthened to quarterly intervals.
Response:
The selection of a monitoring interval was not based on a comparison of
industries and emissions from them. It was selected as a part of the best
system (considering costs) of continuous emission reduction, or best
demonstrated technology (BDT) [see Section 3.1]. As discussed in response
to the next comment, the determination of BDT for valves (in terms of
monitoring interval of leak detection and repair) was based not only on cost
and cost effectiveness, but also on the total emissions reduction achiev-
able. These considerations were made for this source category, independent
of comparison to standards development for other source categories.
Furthermore, leak frequency was taken into consideration in the standards in
the form of alternative standards (see Chapter 14). For instance, annual
leak detection and repair is allowed for units demonstrating and maintaining
a leak frequency for valves of less than 2 percent.
Comment:
One commenter (IV-D-17) said that the quarterly plus monthly monitoring
program cannot be justified based on VOC emissions reductions, in light of
the tremendous time and effort required to locate, tag, record, and
remonitor leaking valves. Similar concerns were expressed by another
4-22
-------
commenter (IV-D-7) who wrote that the leak detection and repair program is
very time-consuming and labor-intensive.
Another commenter (IV-D-18) suggested that the monitoring requirements
for valves be reduced to quarterly monitoring for the first year. Those
valves which are not found leaking would be monitored once annually after
that time. The commenter explained that this lengthening of the monitoring
interval would allow a two-man full-time monitoring team to complete the
first year monitoring requirements. In subsequent years the team would be
free to perform other tasks in the interest of productivity.
Two commenters (IV-D-34; IV-D-50) stated that monitoring on a semi-
annual basis would be adequate. But another commenter (IV-D-46) expressed
concern that the monitoring intervals were too long and would result in
large leaks going unrepaired for too long. Because of this, the commenter -
felt longer inspection intervals could slow attainment of health standards
in many areas, especially if similar concessions were made to other VOC-
emitting industries.
Response:
The commenters are expressing concern over the monitoring interval
chosen and the justification for that choice. The proposed standards
required monthly monitoring because it would provide the greatest emission
reduction potential without imposing difficulties associated with a more
frequent leak detection and repair program (LDRP). Since proposal,
additional data from SOCMI screening and maintenance studies (IV-A-7;
IV-A-11) led to the development of the LDAR model for evaluating the effec-
tiveness of various LDRPs. The details of the LDAR model are given in
Docket Item No. IV-A-22 and its application is discussed in the AID
(III-B-2).
Several monitoring plans for valves were evaluated using the LDAR
model: annual (A), semiannual (SA), quarterly (Q), quarterly with monthly
follow-up on leaking valves (M/Q), and monthly (M). Each of these plans was
then compared in terms of cost effectiveness of the LDRP and the emissions
reduction achievable.
The cost effectiveness of valve LDRPs is presented as a function of
monitoring interval in Figure 4-2, assuming 14 percent early recurrence. A
4-23
-------
100
to ^
in
01
01
-l->
01
o
LJ
50
Monthly
Quarterly
Monitoring Inteyal
Semiannual
Figure 4-2. Cost effectiveness of leak detection and repair programs for valves as
a function of monitoring interval for an average SOCMI unit.
aF
Parentheses indicate net credit of program due to product recovery.
-------
quarterly monitoring interval is noted as the most cost effective plan.
Quarterly monitoring with monthly follow-up is of comparable cost effective-
ness, however. These two programs resulted in a net credit considering the
saved product, while semiannual and monthly monitoring resulted in positive
cost effectiveness.
While cost effectiveness of the LDRP was an important consideration in
selecting the monitoring interval, the emissions reduction achievable was of
equal importance in making a final determination. Figure 4-3 presents cost
effectiveness of the various LDRPs examined versus the emissions reductions
achievable for SOCMI model unit C. The curves for model units A and B have
the same shape, but span a lower range of emission reductions due to a
smaller number of valves. The figure clearly shows the increase in
emissions reduction between semiannual and quarterly programs (57 percent
increase) and between quarterly and monthly programs (26 percent increase).
It also shows the small difference in emission reduction between the
quarterly plan and the more complex plan requiring quarterly monitoring with
monthly follow-up.
The incremental effectiveness of going to increasingly more frequent
monitoring programs was examined in Table 4-4. The incremental cost effec-
tiveness of going from quarterly or monthly/quarterly to monthly is seen as
resulting in a net cost. The other cases indicated the value of increasing
the frequency of monitoring intervals since credits are still obtained with
each increase in frequency. Even though a cost is incurred to increase
monitoring frequency from quarterly to monthly, the cost effectiveness is
considered reasonable.
Based on the analysis of the effect of monitoring interval on costs and
emissions reduction, EPA determined that a monthly monitoring program is to
be used for the SOCMI fugitive VOC emissions standards. While less frequent
programs were more cost-effective, monthly monitoring also had reasonable
cost effectiveness, reasonable incremental cost effectiveness, and yielded
the largest emissions reduction of the programs examined.
4-25
-------
ro
100
50
in
-------
TABLE 4-4. EMISSIONS REDUCTION AND COST EFFECTIVENESS
OF LEAK DETECTION AND REPAIR PROGRAMS FOR VALVES
WITH 14 PERCENT RECURRENCE, FOR SOCMI MODEL
UNIT C
Monitoring EMISSIONS REDUCTION. Mg/Yr COST EFFECTIVENESS, $/Mg
Interval
M 103.3 62 ,
M/Q 85.3 (41)
Q 82.7 (41)
SA 52.6 25
A (4.6)
Monitoring Interval INCREMENTAL EMISSIONS INCREMENTAL COST
Change REDUCTION, Mg/Yr EFFECTIVENESS, $/Mg
From Tp_
A SA 57.2 (252)
SA Q 30.1 (156)
Q M/Q ' 2.5 (37)
M/Q M 18.0 550
KEY: M = monthly; M/Q = quarterly with monthly follow-up of repaired
leaks;
Q = quarterly; SA = semiannual; A = annual.
Parentheses indicate credits.
4-27
-------
Comment:
One of the same commenters (IV-D-17) alleged that EPA's proposed
program is slanted more toward gathering data concerning repair efficiency
than controlling emissions.
Response:
The standards for valves are designed to reduce VOC emissions at a
reasonable cost. Data show that valves in SOCMI process units contribute a
large portion of the VOC emissions from equipment in SOCMI. Moreover, leak
detection and repair programs have been found to reduce emissions from
valves effectively. The proposed leak detection and repair program was -
selected as best demonstrated technology (BDT) for valves in gas and light
liquid service based on the cost effectiveness of controlling fugitive VOC
emissions and the emissions reduction achievable.
The proposed standards required quarterly reports to aid in determining
compliance with the standards. Since proposal, though, considering comments
and in an effort to reduce paperwork, EPA decided that these reports are
beneficial in determining compliance, but they are not necessary on a
quarterly basis. Therefore, semiannual reports are required in the final
regulation; States that are delegated the authority to enforce the
standards, however, may waive such reports through their own programs if
EPA, in delegating the program to the State, approves the reporting require-
ments or an alternative means of compliance surveillance adopted by the
State and if the process units comply with the requirements adopted by the
State. In addition to the semiannual reporting requirements, the standards
still require notifications (construction; anticipated startup; initial
startup; physical operational changes; use of alternative standards;
performance test) and performance test results according to the General
Provisions.
Even though reporting has been streamlined, the recordkeeping require-
ments have not been changed. Recordkeeping has been deemed necessary for
determining compliance because the standard is a work practice standard.
Section 9 addresses in detail the comments and concerns on reporting and
recordkeeping.
4-28
-------
Comment:
One commenter (IV-D-17) recommended that monthly monitoring be required
only for equivalency demonstrations. Another commenter (IV-D-25) said that
the proposed frequency of inspection appears satisfactory for valves that
operate daily; however, for those that operate less frequently he suggested
less frequent inspection, e.g., yearly.
Response:
There are many ways to demonstrate equivalency. Certainly, one
conceivable way would be to monitor monthly and show how a program meets or
exceeds the valve standards. Monthly monitoring is not burdensome in the
judgement of the Agency. Moreover, the proposed and final standards require
monthly monitoring of only those valves which are found leaking. Therefore,
the monitoring interval will not be changed based on frequency of valve
operation. :;
Just as there are many ways to demonstrate equivalency, there are many-
ways to design leak detection and repair programs. If an owner/operator can
devise a more efficient program, he may choose to comply with an alternative
standard. Data from the Analysis Report (IV-A-14) indicate that control
valves, as a class, exhibit higher leak frequencies than block valves.
Since control valves are generally operated more frequencly than block
valves, it is reasonable to believe that less frequently operated valves
leak less frequently. Therefore, EPA has allowed alternative standards to
consider this variability. An owner or operator could make use of a trend
of this type in developing an alternative standard if he can demonstrate,
such standards would achieve equivalent emission reductions. Leak frequency
was selected as the basis of the alternative standards, rather than
frequency of operation, since it is more readily measured and its effect on
leak detection and repair program effectiveness can be examined (see
Chapter 14 and Appendix A).
Comment:
One commenter (IV-D-17) said that it is impractical to monitor more
frequently than once every three months without resulting in a situation
where a detected leak could not be repaired before the next monitoring cycle
4-29
-------
began. He noted that it is obvious from the proposed standards, though not
made clear in the document, that EPA has assumed only one two-man team to
conduct the monitoring of valves in a large SOCMI model unit. He argued
that using EPA estimates of 116 man-hours to monitor and repair 2800 valves
and adding some time for scheduling of repairs, a two-man team could just
barely detect and repair all the leaking valves before it would be time to
start the next monitoring.
Response:
A detected leak need not necessarily be repaired before the next
monitoring cycle begins. Also, since monthly/quarterly monitoring schemes
are available, monitoring cycles are not necessarily contiguous. A leak
must be repaired within 15 days of its detection. There is no requirement
for repairing a leak before monitoring activities resume in the process
unit. Using an estimate of two minutes of monitoring time per valve, it
would take about 93 hours to monitor the unit completely. And the estimated
time required for maintenance is about 53 hours per month. The combined
time is within the monthly time limit of 172 working hours.
Comment:
Another commenter (IV-D-40) presented data to show a reduction in
effectiveness for frequent leak detection and repair programs. To support
his: claims, the commenter cited the essentially linear leak occurrence/
recurrence rates determined for SOCMI. He stated that, due to the limited
amount of recurrence data and the overlapping confidence intervals, only the
occurrence rate can be used. He also cited on-line valve repair efficien-
cies lower than assumed in -the BID among the reasons for the reduced
effectiveness. Using additional data from a high-density polyethylene
plant, the commenter argued that more frequent inspection and maintenance
did not reduce the percentage of valves leaking.
Response:
The comments concerning leak occurrence and recurrence rates were
discussed in the initial responses on monitoring intervals for leak
detection and repair programs and in the AID (III-B-2). As discussed in the
SOCMI Maintenance Study report (IV-A-10), the recurrence rate for valve
4-30
-------
leaks was sufficiently different from the occurrence rate to be determined
as a distinct value. But only a single recurrence rate was determined for.
the three process types tested because recurrence data was scant.
The efficiencies for valve leak detection and repair programs examined
in the BID ranged from 0.86 to 0.90 for gas valves and from 0.62 to 0.74 for
light liquid valves. These efficiencies considered imperfect repair of some
valves, that is, repair that did not reduce emissions to 0 ppmv. The
commenter accurately pointed out that the on-line repair efficiencies,
determined during SOCMI studies were lower than assumed in the BID. In
fact, a repair efficiency of only 29 percent was determined as a result of
maintenance. But this efficiency was found for only simple on-line
maintenance (tightening bolts). And more importantly, this 29 percent
repair efficiency resulted in about 71 .weight percent reduction in fugitive
emissions. Repair efficiencies are also discussed in the next section on ,-
leak definition.
4.2.2 Time Estimates
Comment:
One commenter (IV-D-24) pointed out that EPA has used a figure of
2 man-minutes per source for monitoring time even though on page C-9 of the
BID, the results of a test run indicate 3 to 4 man-minutes per source. He
argued that the labor estimates are low by a minimum factor of two.
Response:
The commenter (IV-D-24) is comparing average times for monitoring
fugitive emission sources as determined in the field with an estimate of
time required to monitor valves. The comparison is not on the .same basis
and is, therefore, invalid.
On page C-9 of the BID an average screening time per source of
1.7 minutes is presented. This time was determined in actual field studies
and includes time to measure not only valves but also pumps, compressors,
safety relief valves, and flanges. A two-man team performed the monitoring,
so the average time per source was 3.4 man-minutes.
The time estimates used for costing purposes may be found on page 8-8
of the BID. The monitoring time estimate for valves is 2 man-minutes per
4-31
-------
valve. However, time estimates for other equipment are much higher. The
average manpower requirement per source would undoubtedly be higher than the
2 man-minutes estimated for valves. For example, applying the time=
estimates "to the equipment distribution in the model units, the average
monitoring requirement per source would be 2.9 man-minutes.
This number compares very favorably with the 3.4 man-minute number
determined in the field, especially considering differences in the
monitoring activities. The value determined in the field also includes time
for instrument calibration, maintenance, etc.; these instrument-related
items were considered in the instrument maintenance cost of $2700/yr and in
the 40 percent overhead charge (see cost estimates Section 5 of AID). In
addition, the average time determined in the field was for a research effort
which required more data gathering'and recording than routine monitoring
will require. The researchers were required to obtain a numerical reading
and record it, while routine monitoring would require only ascertaining
whether the reading was on- or off-scale. Furthermore, the researchers were
recording many extra data concerning valve type, process conditions, and
ambient conditions which would not be required during routine monitoring.
All of this extra effort is included in the 3.4 man-minute average.
Considering these extra activities, the time estimates used for costing
purposes are suitably conservative.
Another check on the validity of the time estimates can be made by
comparing the actual time spent in the field with the time which would have
been predicted by EPA's estimates in the BID. Table 4-5 shows such a
comparison. EPA's estimates applied to the sampled units were again seen to
be conservative.
Comment:
A commenter (IV-D-17) questioned the one minute estimate for valve
monitoring time. He argued that it is not even possible to travel from one
valve to another in one minute. He accused EPA of using the data provided
by the industry out of context to come up with this estimate. The commenter
argued that this figure was generated as a ballpark number for initial
comparison purposes at the beginning of a fugitive emissions study a few
4-32
-------
TABLE 4-5. ESTIMATED VERSUS ACTUAL MONITORING TIMES
FOR VARIOUS SOCMI PROCESS UNITS3
Unit
Number
1
2
3
4
5
6'
11
12
20,21
22
28,29
31
32
33
34
35
60,61,62
64
65
66
Total
Number of
Sources
Chemical Monitored
Vinyl Acetate
Ethyl ene
Vinyl Acetate
Ethylene
Cumene
Cumene
Ethylene
Acetone/Phenol
Ethylene Dichloride/
Vinyl Chloride
Formaldehyde
Ethylene Dichloride/
Vinyl Chloride
Methyl Ethyl Ketone
Methyl Ethyl Ketone
Acetaldehyde
Methyl
Methacrylate
Adipic Acid
Chlorinated
Ethanes
Adipic Acid
Acrylonitrile
Acrylonitrile
TOTALS 43
1391
5078
2713
5278
1025
1573
3685
3207
2298
230
3363
585
679
1148
2019
1577
3332
664
1406
1864
,115
EPA- Estimated
Monitoring Time
(hours)
54
176
98
182
36
55
143
128
91
9
123
22
26
44
77
53
121
26
51
68
1583
Actual
Monitoring
Time
(hours)
46
110
42
132
15
26
117
171 "
100
7
90
16
25
23
30
18
89
21
59
59
1196
Contractor
Radian
Radian
Radian
Radian
Radian
Radian
TRW "-
TRW
PEDCo
PEDCo
PEDCo
Acurex
Acurex
Acurex
Acurex
Acurex
PEDCo
PEDCo
PEDCo
PEDCo
aFrom Frequency of Leak Occurrence for Fittings in Synthetic Organic Chemical Plant
Process Units, by Radian Corp., for U.S. Environmental Protection Agency, Research
Triangle Park,
I.C., September 1980. (IV-A-11).
4-33
-------
years ago. More recent detailed data show monitoring times of 3 to
4 minutes. In addition, the commenter cited an industry report which shows
that monitoring time varies from a minimum of 3 minutes to a maximum of
12 minutes per valve depending upon the accessibility of the valve. He
concluded that the overall average is 4 minutes.
Response:
. The one minute per valve time estimate was taken from information
provided by Exxon Company, USA (II-D-21). The data pre~sented were presented
as the results of "an in-depth study to determine the monitoring manpower
requirements." A review of the letter, the data presented and its
application to cost estimates failed to show that the data were used out of
context or inappropriately.
The newer data referred to by the commenter is also contained in an
Exxon document (II-D-72, Appendix B). The numbers cited by the commenter
were inaccurately cited as manpower estimates per valve. The manpower
requirements given were actually estimates of time which would be required
to monitor pump seals, compressor seals, valves, drains, and pressure relief
valves. As noted in the response to the previous comment, EPA's estimates
allow more monitoring time for other types of equipment, so that the average
time per source would be higher. Furthermore, Exxon's estimates also
included manpower requirements for doing some minor valve maintenance, while
EPA's estimates account for this manpower requirement separately.
The travel time for valve to valve is included in the monitoring time
estimates. If valves were each distantly located from another, the travel
time component would be larger. However, many valves are commonly found
clustered together in one location, requiring no travel time between them.
Comment:
Another commenter (IV-D-18) wrote that the estimate of 16 man-hours per
month to fulfill maintenance requirements is too low. He stated that since
EPA did not provide a breakdown of time requirements, an evaluation could
not be made of time estimated to service both readily accessible and
inaccessible valves, nor was there an estimate of the number of valves and
seals which would require service or replacement per month. The commenter
4-34
-------
further stated that the replacement of a single large valve alone could
consume in excess of the allotted 16 man-hours.
Response:
EPA did provide estimates of time requirements for maintenance in the
BID, p.8-8, and has reexamined these in the AID (III-B-2). While it is true
that no distinction in time estimates is made between accessible and
inaccessible valves, the total time estimate resulted from an assessment of
maintenance requirements for valves, including inaccessible as well as
accessible valves. The commenter is correct in saying that a single large
valve may take as much as sixteen hours to replace. However, replacement of
large valves is not expected to be a frequent occurrence. The average time
for off-line repair of all sizes of valves used by EPA in its cost estimates
has been estimated to be 4 hours per repair (see AID Section 5). The
replacement of smaller valves is expected to take less than this allotted
time for off-line maintenance, on the average.
The standards for inaccessible valves require annual monitoring and
repair, compared to the monthly program for accessible valves. Considering
an annual program for inaccessible valves, the repair requirements would be
relatively infrequent.
4.2.3 Repair Requirements
Comment:
Several coirmenters (IV-F-1, No.l; IV-D-21; IV-D-26; IV-D-34) said that
the requirement for repair at the next unit shutdown is too inflexible and
ignores situations where replacement parts for leaking equipment may not be
available until after the next shutdown. In two sets of these comments
(IV-F-1, No.l; IV-D-21) it was stated that an extended shutdown could happen
due to abnormal near term demand for replacement parts and for unforeseen
manufacturer's or delivery delay. The third commenter (IV-D-26) also
expressed concern that unscheduled outages, not related to maintenance could
also create a situation where once the process unit was down it could not be
restarted because of the inavailability of repair or replacement parts. He
wrote that although this would not be a frequent occurrence, some provisions
4-35
-------
should be made to allow the process unit to start up again even though some
unrepaired fugitive emission sources might still be leaking^
The same commenter cited another similar example involving custom-made
equipment, e.g., a compressor built to certain design specifications that is
not generally available. Due to the age of the unit, replacement seal parts
may no longer be available and it may be necessary to replace the compressor
with a new unit. The commenter suggested that if the existing unit does not
present any occupational safety hazards, it should be allowed to continue
operation during acquisition and fabrication of the new unit regardless of
process shutdown.
Another commenter (IV-D-46) felt that there were significant points
where back-up equipment could enable quick repair of leaks. Although not
practical for all fugitive emission sources, the commenter said that some
sources (control valves, block valves, pressure relief devices, pumps,
compressors) could have back-up systems. The commenter continued, saying
that back-up could be minimized by sharing between emission sources; for
example, piping could be arranged so that a spare pump could serve as a
back-up for more than one pump.
Response;
EPA agrees that there may be occasions when the lack of spare parts
might prevent repair of all leaking valves during a unit shutdown. To allow
for this eventuality, provisions have been made in the regulation to allow
delay of repair beyond a shutdown if certain conditions are met. The
conditions are that valve assembly replacement is necessarys valve assembly
supplies have been sufficiently stocked, and.the supplies have been deleted.
Custom-order, unique parts should also be stocked to avoid delays of repair
due to inavailability.
Spare parts, such as valve packing and pump seals, are items that are
typically stocked and can be stocked without unreasonable burden. For
example, assuming a unit maintains a stock of 8 spare pump seals, the annual
cost of maintaining that stock is about $360. This value assumes $226/seal
and an estimated carrying charge for the stock of 20 percent.
4-36
-------
In allowing delays of repair, EPA recognizes that there may be
instances where equipment cannot be repaired on-line. Although certain
types of equipment are commonly spared in SOCMI, EPA determined it
unreasonable to require spare equipment, where it is not the norm, so that
leaks could be repaired quickly. The costs of requiring such redundancy
would be prohibitive.
The delay of repair provisions are discussed in detail in Section 13 on
Enforcement and Compliance Concerns.
Comment:
One commenter (IV-D-20) wrote that it would be almost impossible to
inspect and repair units in 15 days.
Response:
The standards do not require that a unit be inspected and repaired in
fifteen days, the standards require that all valves be monitored monthly :
but that nonleaking valves may be monitored quarterly. The fifteen-day
requirement is for an individual valve not for a process unit. The require-
ment is that a valve found leaking must be repaired within 15 days of
finding the leak, not within 15 days of the start date for monitoring in the
process unit.
Comment:
In another set of comments (IV-D-17) the need for flexibility in
scheduling repairs was stressed. The commenter said consuming an entire
maintenance force to repair 10 percent of the components may result in
allowing nonleaking equipment to deteriorate. He was concerned that the
fifteen-day repair requirement might prevent timely maintenance on
nonleaking equipment, thereby fostering a situation conducive to causing
more leaks. Another commenter (IV-D-46), however, disagreed, stating that
15 days was too long a period. This commenter felt that 5 days would be
adequate to effect repairs, since personnel and supplies should be at hand.
Response;
The fifteen days is considered adequate for repair of all but those
valves which are critical and cannot be by-passed. The fifteen days
provides sufficient time to schedule and effect on-line repairs that a
4-37
-------
shorter time period might not allow. Provisions have been made for delaying
repair of those valves until the unit is shutdown.
.EPA recognizes the fact that maintenance labor will be required for the
implementation of this leak detection and repair program. The extra
manpower required had been estimated and included in the cost estimates in
Chapter 8 of the BID. These requirements were also included in the
estimates presented in Chapter 5 of the AID (III-B-2). The additional
maintenance manpower should allow efficient scheduling of preventive
maintenance while the leak detection and repair program is underway.
Comment:
One set of comments (IV-D-17) expressed agreement with the estimates of
ten minutes on-line and four hours off-line repair times. But the commenter
expressed concern that EPA has not allowed any unit downtime for repair of
valves which must be taken off-line for repair. The commenter pointed out
that while many leaks can be repaired soon after they are found, and while
the leaking equipment remains in service, it must be recognized that
situations exist where quick repairs will, in fact, result in increased
emissions. As an example, the commenter mentioned a situation where a
critical component of a process unit is leaking. Repair of this critical
component requires a special shutdown of the process unit. In order to
safely shut down the unit, more emissions enter the atmosphere than would
have been emitted from the leaking component. The commenter suggested that
the proposed standard should not only allow, but strongly encourage the
application of realistic judgement in these cases so that total emissions
are reduced. He added that if EPA insists upon repair which requires unit
shutdown, it must include a debit for lost production.
Another commenter (IV-D-32) said that some leaks are better left
unrepaired. He explained that, to change a valve with a small leak,
transfer lines would have to be purged. Even after purging, the line would
contain enough material to pollute the atmosphere.
Response:
EPA recognizes the fact that it would be impractical to shutdown a
process unit to repair a valve. The standard does not require that a unit
4-38
-------
be shutdown for repair. Allowance is made for repairing such critical
valves at the next unit shutdown. In addition, those valves that can be
isolated but would require considerable purging of VOC to the atmosphere
would be exempted from repair until the next unit shutdown. Because
shutdowns are not required, a debit for lost production is not necessary.
Comment:
One commenter (IV-D-46) was concerned that the focus of the standard
appears to be on detecting and repairing leaks after they have occurred. He
noted that the preamble devotes scant attention to preventive maintenance
that could minimize development of leaks in the first place. For example,
the commenter pointed out, EPA has apparently not considered the possibility
of replacing the packing in valves at regular intervals, before it becomes
brittle and subject to leakage. He suggested that automatic replacement at
regular intervals may be justified.
Response:
Certainly the ideal way to eliminate fugitive VOC emissions is to
prevent them from occurring altogether. Some of the equipment and
performance requirements in the final standards provide for this where
possible. One means of reducing leaks from valves is through scheduled
preventative maintenance. Owners and operators have incentive under the
standards to increase preventative maintenance efforts in order to reduce
the number of valves found leaking. This procedure would reduce the monthly
monitoring burden. This type of program would not, however, eliminate leaks
from occurring due to the numerous variables affecting the valve leak
occurrence rate. Although regular valve packing changes may reduce the leak
occurrence rate, it would not eliminate leaks from occurring altogether; a
leak detection and repair program would still be needed to find these other
leaks.
As illustrated by the data collected on fugitive emissions, most valves
do not leak. And in some instances, attempting repair of a nonleaking valve
can result in creating a leaking source. Thus, there may not necessarily be
a positive benefit for routine packing replacement in valves.
4-39
-------
4.2.4 Accessibility
Comment:
Several commenters (IV-F-1, No.l, p.11; IV-D-5; IV-D-15; IV-D-17;
IV-D-21; IV-D-23; IV-D-26; IV-D-29; IV-D-34; IV-D-43; IV-D-50) expressed
concern about the inaccessibility of some valves in SOCMI units. Four sets
of the comments (IV-F-1, No.l; IV-D-17; IV-D-21; IV-D-50) gave safety
considerations, configuration, and elevation constraints as the possible
reasons why a valve may be inaccessible for monitoring. Commenters said
that many of these valves can be eliminated in an entirely new plant but
become a problem when an older plant becomes subject to the regulations due
to modifications. Two commenters (IV-D-21; IV-D-34) recommended that such
inaccessible equipment be exempted or excluded from the proposed NSPS.
Another commenter (IV-D-17) suggested that two new Sections 60.482(f)(7) and
60.482(f)(8) be added as follows:
(7)(i) An owner or operator of a new or modified source subject
to the requirements of §60.482(f)(l)-(6) may for valves that are
routinely inaccessible for safety reasons monitor each inacces-
sible valve for leaks after a process unit overhaul prior to
startup by pressuring with nitrogen to the system process pressure
or 100 psig, whichever is less, and checking with a soap solution
for bubbles, or other equivalent test method pursuant to §60.484.
(ii) When a leak is detected, it shall be repaired as soon as
practicable, but no later than the next scheduled shutdown, or
consistent with §60.482(h).
(iii) For purposes of §§60.483 or 60.484, inaccessible valves
shall not be included.
(8)(i) An owner or operator of a modified source subject to the
requirements of §60.482(f)(l)-(6) may for valves that are
. routinely inaccessible because of elevation or configuration
monitor each inaccessible valve annually using test methods
pursuant to §60.485 or a soap solution for bubbles.
(ii) When a leak is detected, it shall be repaired as soon as
practicable, but no later than the next scheduled shutdown, or
consistent with §60.482(h).
4-40
-------
(ill) For purposes of §§60.483 or §60.484, inaccessible valves
shall not be included.
Another conunenter (IV-D-23) recommended either an exclusion or an
alternative method of sampling for inaccessible valves. He referred to an
alternate test procedure suggested by an industry group (IV-D-17). He
stated that the costs would increase significantly if inaccessible valves
were required to be sampled routinely.
One other commenter (IV-D-26) referred to EPA reports which indicated
some accessibility problems experienced by the contractors during
monitoring. The commenter expressed concern that despite this contractor
experience, EPA wishes to impose on SOCMI such a frustrating and potentially
hazardous task.
Response:
EPA recognizes that some valves may be difficult to monitor because
access to the valve bonnet is restricted or the valves are located in
elevated pipe racks. In addition, some valves may be unsafe to monitor \/
because process conditions include extreme temperatures, pressures, or i
chemicals which could be explosive or hazardous. Difficult to monitor
valves can be eliminated in new process units but may not be eliminated in
existing process units. Therefore, the proposed standard has been amended
to provide for these circumstances.
For process units that become affected by a modification or reconstruc-
tion, EPA is requiring an annual leak detection and repair program for
valves which are difficult to monitor. Valves which are difficult to
monitor are defined as valves which require safely elevating monitoring
personnel more than two meters above any permanent available support
surface. This means that ladders may be required to elevate monitoring
personnel safely, but scaffolds will not be required.
Valves which are unsafe to monitor cannot be eliminated in new or
existing units. These valves are required in certain process units and
would be unsafe to monitor under certain process conditions. These valves
can be monitored at times when the process conditions that indicate the
unsafe conditions are not occurring. Owners or operators will be required
4-41
-------
to demonstrate that valves are unsafe to monitor on a routine basis and to
prepare a plan for monitoring those valves which are unsafe to monitor
routinely. Valves which are unsafe to monitor are defined as those valves
which could, based on the judgement of the owner or operator, expose
monitoring personnel to imminent hazards from temperature, pressure, or
explosive process conditions. A plan is required that defines a leak
detection and repair program conforming with the routine monitoring
requirements of the proposed standards as much as possible given that
monitoring should not occur when it is unsafe to monitor valves that would
expose monitoring personnel to imminent hazards from temperature, pressure,
or explosive process conditions.
4.2.5 Other Monitoring Methods
Comment:
One commenter (IV-D-24) suggested that in light of the high vapor
pressure characteristics of the chemicals listed in Appendix E (§60.489) of
the regulations, area monitoring of processes enclosed by buildings will
suffice in determining the presence of a leak. He recommended that
provisions for this special case should be included in the regulations.
Response:
As discussed in the BID in Appendix D, EPA performed a limited evalua-
tion of fixed-point monitoring systems. The results of these tests
indicated that fixed point systems were not capable of sensing all the leaks
that were found by individual component testing. As a result, fixed-point
area monitors were not incorporated in Reference Method 21.
The application suggested by the commenter, while it does not conform
to Reference Method 21, may be useful to an owner or operator who has
elected to comply with an alternative standard. It is possible that fixed-
point monitors could be used for surveillance in addition to individual
source monitoring.
Comment:
One commenter (IV-D-13) stated that application of any control
resources other than visual inspection to equipment handling liquids in the
vapor pressure range of concern would not be cost effective. He noted that
4-42
-------
almost any amount of leakage will occur in the liquid form. Evaporation
rates are so slow that accumulation of liquid becomes readily evident long
before serious impact upon the environment occurs.
Response:
Evaluations of monitoring methods were made in selecting the method of
individual point monitoring with portable VOC analyzers. Visual inspections
were discarded as being too subjective to be reliable. Experience in the
field has shown that leaks are found with a portable VOC analyzer Which
would not be detected by visual, audible, or olfactory means. Therefore,
the portable VOC analyzer was chosen as the monitoring method.
Comment:
In objecting to a leak detection and repair program, one commenter
(IV-D-32) said that most large plants have safety and housekeeping
inspection teams that perform maintenance checks for the kinds of leaks to -
be regulated by the proposed rule. He felt that only training and close
supervision could correct sloppy operational practices.
Response:
While it may be true that some plants have safety and housekeeping
inspection programs that detect visible leaks, this practice is not
uniformly found in all SOCMI units. The leak detection and repair programs
that are part of the final rule are intended to identify VOC leaks that may
not be detected by visual, audible, or olfactory means which tend to be very
subjective.
Comment:
One commenter (IV-D-1) wrote that data, currently not available, will ~
show that leak frequency will vary with different types of valves. It
would, therefore, be appropriate to group them into different categories
according to leak frequency. He suggested that it would also be possible to
have different inspection plans for the different groups. He added that it
is also appropriate to vary the protection level depending on the toxicity
or hazardous nature of the leaking chemical. According to the commenter,
the effort should be focused where leaks occur most often.
4-43
-------
Response:
The philosophy of focusing effort where leaks occur most frequently is
a sound one which is endorsed by EPA. In fact, the standard for valves
incorporates features resulting from the application of this philosophy.
Several types of fugitive emission sources have been exempted from coverage
because of low leak frequencies, e.g., flanges and heavy liquid equipment.
Furthermore, the scheme of less frequent monitoring of valves which do not
leak for two consecutive months is consistent with this philosophy.
Recent analyses of fugitive emissions data gathered by EPA in SOCMI
units indicate several factors which may influence leak frequency, such as
valve type, line pressure, primary material in the line, and chemical unit
type. These factors may be useful to an owner or operator who chooses to
develop his own plan to comply with an alternative standard. It would be
impossible, however, to write the.nonoptional valve standard (§60.482f) in a
manner that allows for all of those factors. The resulting standard would
be excessively complex and unmanageable.
Protecting people and the environment from toxic chemicals is certainly
an important goal. EPA has determined that VOC compounds contribute to air
pollution which may endanger man's health and welfare and that, therefore,
all VOC should be prevented from entering the atmosphere, including those
chemicals which are toxic.
4.2.6 Potential Monitoring Problems
Comment:
Two commenters (IV-D-15; IV-D-43) wrote that instrument calibration,
operational reliability (an inventory of spare parts is needed), and
difficulties encountered in the field appear to be major problems in
monitoring VOC's in SOCMI plants. Chapter 4 of the BID notes "portable
hydrocarbon detection instruments are the best method for identifying leaks
of VOC from equipment components" (p.4-2). However, the commenter noted
that there is no discussion in the BID of the many difficulties encountered
during contractor sampling in SOCMI plants. The difficulties as detailed by
the commenter are:
4-44
-------
"Nuisance" malfunctions such as failure of the battery pack,
preamplifier and readout meter. Hydrogen leaks also were
frequently encountered in the monitoring equipment.
Another example of a "nuisance" malfunction occurred when exces-
sive amounts of moisture or organic liquids were drawn into the
probe of the hydrocarbon detector. The contractor's solution to
the problem was to allow the instrument to run and dry out for
several hours before reuse. This delay is distruptive and
certainly is more than a "nuisance" when attempting to minimize
sampling costs associated with compliance with this regulation.
The EPA contractor also observed that water, drawn into the
instrument probe and internals during rainy weather or from icy
surfaces, often produced random electrical signals and subsequent
erratic behavior. Again, the contractor's solution to this
problem was to allow the instrument to run and dry out for several
hours.
With Century Systems' instruments the hydrogen flame was
extinguished often when too rich (i.e., very high concentrations)
compositions of VOC's were introduced into the instrument. This
necessitates bringing the instrument to an area where safe
ignition of the hydrogen can be made.
During EPA's survey of an adipic acid manufacturing plant, the
sampling team used teflon tubing packed with glass-wool to prevent
participates and liquids from contaminating the OVA probe. In
addition, it was reported "when cyclohexane contaminated the
probe, we used an elaborate wash system to purge the OVA." (The
use of the glass-wool and an "elaborate wash system" are not
described in Reference Method 21.)
Instrument response to some organic compounds (phenol, for
example) was very slow (10-30 seconds) and background zero was
obtained only after 2 to 3 minutes. Thus, average sampling
time/source can be adversely affected. The EPA contractor
sampling team also noted that a very sluggish response to fugitive
aromatic emissions (particularly cumene) was observed.
4-45
-------
g. Apparently, adverse weather conditions, such as high winds, heavy
rain or severe cold, hampered the screening efforts of the
contractor at several process units. None of these problems was
discussed in the BID. Further, the proposed regulation makes no
allowance in the economics for sampling that is terminated because
of inclement weather.
In summarizing his concerns, the commenter stated that the instrumen-
tation used in VOC monitoring programs may not be reliable, and will result
in significant additional manpower requirements not considered by EPA in
developing the economics for each alternative.
Another commenter (IV-D-26) referred to the EPA contractor report
"Frequency of Leak Occurrence for Fittings in Synthetic Organic Chemicals
Plant Process Units." He stated that this report describes in considerable
detail the delays in monitoring caused by instrument problems, failure of
major parts, and slow repair. The commenter expressed the opinion that the
OVA-108 does not seem to be really designed for rugged and continued use.
Response:
It is true that portable VOC monitors require some maintenance and care
in their use. It is also true that a certain number of spare parts should
be kept on hand to insure uninterrupted service. The instruments are
certainly not perfect, but they are the best method for identifying leaks of
VOC from equipment components, and it is expected that as more development
work is done, the durability and reliability of the instrument will be
improved. Alternatives to portable VOC monitors would be confined to
expensive, nonportable analytical instruments not suited to field applica--
tion, not explosion-proof and even more fragile than portable VOC analyzers.
In other words, portable VOC analyzers are the only choice.
Recognizing that maintenance would be required for the portable VOC
monitor, as it would for any analytical instrument, an allowance for
maintenance was made in the cost analysis. The cost estimate for instrument
maintenance including parts and labor was $2,700 per year. The estimate is
shown in Tables 8-9, 8-10, 8-11, and 8-14 of the Background Document.
4-46
-------
Some of the nuisance malfunctions mentioned by the first commenter can
be avoided by careful operation of the instrument. Avoiding drawing liquids
or particulates into the probe tip will eliminate the need to let the
instrument run and dry out for several hours.
Contrary to the information presented in the first comment, extin-
guishing of the flame by VOC compositions which are too rich presents no
problem to the, Century System's instrument. The flame can be reignited
safely within seconds without removing the instrument from the plant
environment. (See the instrument manual for a description of its safety
features.)
While it is true that use of glass-wool and an "elaborate wash system,"
are not given as part of Reference Method 21, these procedures were
presented in the 24-Unit study (IV-A-11) as methods used to circumvent
specific problems encountered during sampling by contractors. These -
procedures may not be necessary in all units, but there may be instances
where the owner or operator deems it valuable to implement these procedures
to avoid contamination of the probe by particulates or liquids.
In response to the comment concerning sluggish response to certain
chemicals, it is true that sluggish response of an instrument to certain
chemicals could affect monitoring time for an individual unit. However, the
average monitoring time reported by EPA contractors incorporates those
delayed response times. So, the average time reported is unaffected by
sluggish response.
As the first commenter indicated, it is also true that inclement
weather can delay or prevent monitoring with portable VOC monitors.
However, an owner or operator.of an affected facility will have much wider
latitude in scheduling for inclement weather than contractors working under
tight schedules. Interruption or termination of screening will not require
starting over, merely resuming where the interruption occurred,, so that no
allowance was made for extra cost. Furthermore, scheduling of outdoor
activities to meet weather conditions is an administrative problem
frequently encountered and successfully solved in SOCMI.
4-47
-------
The second commenter generally echoed the first commenter's complaints
about portable VOC monitoring instruments. However, he also complained
about slow repair service. Slow repair service by the factory was
encountered as noted in the referenced report. However, as also noted in
this report, a major contributing factor in the delayed repair service was
the fact that the factory was being moved at the time. The report also
noted that regional factory representatives did much to alleviate delays by
providing access to back-up instruments. '
Comment:
One commenter (IV-D-18) expressed concern that tightening the packing
on certain types of valves may make emissions worse. He pointed out that,
as mentioned in the preamble, overtightening will frequently cut or shear
the packing. The commenter stated that overtightening will frequently
result in increases in emissions, shorter valve life, and will be
counter-productive to the intent of the regulation.
Response:
EPA recognizes the fact that some valves cannot be repaired on line.
Overtightening is not advisable and is not encouraged. The regulation does
not require that valves be overtightened. It allows for valves which cannot
be repaired on-line to be repaired off-line at the next unit shutdown.
Comment:
One commenter (IV-D-32) stated that keeping track of numerous fugitive
emission sources (250 in one unit, for example) would be impossible with
limited program resources.
Response:
The final standards require maintaining records on only those fugitive
emission sources found leaking. As part of its regulatory analysis, EPA
considered reporting and recordkeeping requirements. As a result of this
analysis the reporting requirements have been reduced in the final rule to
semiannual reporting, in addition to the notifications required by the
General Provisions and the regulation. Reporting and recordkeeping is
addressed in detail in Section 11.
4-48
-------
4.3 LEAK DEFINITION
Numerous comments were received regarding the selection of the leak
definition for the valve leak detection and repair program. Most comments
suggested that the leak definition be raised. A variety of reasons were
cited for the increase including emissions reductions, maintenance reduc-
tions, and reasons related to the monitoring instrument. Comments on the
leak definition for pumps were also received.
Comment:
A commenter (IV-D-17) wrote that it was his understanding that the
10,000 ppmv action level was chosen because it is the top of the scale on
the Century Organic Vapor Analyzer. The commenter noted that apparently,
EPA felt that 10,000 ppmv would make compliance monitoring easier for the
chemical industry. He expressed significant disagreement with this
assumption. He said that the scale on the Century instrument is a minor
consideration. The commenter said that readings higher than 10,000 ppmv can
be obtained readily with a dilution apparatus. In addition, he expressed
confidence that equipment manufacturers will be able to supply instruments
with direct scales to whatever level is required.
Response:
It is true that one consideration in selecting 10,000 ppmv as the leak
definition for the SOCMI fugitive VOC emissions standards was the monitoring
instrument characteristics. Data on which the standards are based were
collected using common hydrocarbon detectors that are readily available.
These instruments provide direct measurements of VOC up to 10,000 ppmv; in
order to measure higher concentrations with the instruments most commonly
used, additional care and calibration for devices such as dilution probes
are required to obtain accurate results. As a result, additional costs are
associated with measuring concentrations higher than 10,000 ppmv. Although
instruments that directly measure higher VOC concentrations may be available
in the future, the standards are based on the least complicated and best
established portable hydrocarbon detection technique currently available.
4-49
-------
Comment:
Several comments were received which said that the 10,000 ppmv action
level could not be justified by the emission reductions achieved. One
commenter (IV-F-1, No.3) called the choice of leak definition arbitrary.
Another commenter (IV-D-17) argued that using a 20,000 ppmv action level
instead of 10,000 ppmv would result in only 1 percent more emissions.
Comments were also received which stated that the SOCMI Maintenance
Study data suggest a higher leak definition. One commenter (IV-D-7), using
leak frequency/mass emissions data correlations developed in the Maintenance
Study as his basis, wrote that if EPA set a leak definition on the basis of
mass emissions equivalent to those defined in the refinery CTG, an
appropriate leak definition would be 35,000 - 99,500 ppmv (based on
methane). He added that using the leak definition as proposed by EPA with
no consideration of its relationship to mass emission rates results in much
wasted motion by attempting to correct leaks which are insignificant and
unrepairable. The commenter suggested that EPA examine the SOCMI mainte-
nance data to determine if it would support a higher leak definition and
still effect a high percentage reduction of mass emissions. The first
commenter (IV-D-17) said that recent SOCMI data point to a screening value
of 40,000 ppmv. Furthermore, he stated that his preliminary analysis of the
data shows that a majority of emissions could be controlled with a
definition of 100,000 ppmv or higher.
The commenter went on to say that, if the leak rates from equipment in
SOCMI were lower than originally predicted by EPA or if the leak frequency
were lower, a lower percentage of mass emissions would be controllable for a
given action level. Moreover, total uncontrolled emissions would be lower
and control based on a low action level would not be justified.
In a later letter, the commenter (IV-D-50) recommended using an annual
monitoring plan, a 100,000 ppmv screening value for valves in gas service,
and a 10,000 ppmv screening value for other sources. He presented data
(Table II, IV-D-50) to demonstrate control efficiencies of various plans.
The commenter estimated that, under the recommended plan, NSPS emissions
would be below 26 Gg/yr.
4-50
-------
Response:
The leak definition of 10,000 ppmv was selected based on the results of
testing on SOCMI sources. Table 4-6 presents a summary of the percent of
sources screening above the action level (leak definition) for various
action levels (an indication of the number of leaks) and percent of mass
emissions attributable to these action levels. Analysis of the results of
the SOCMI study (IV-A-14) demonstrates two important conclusions. First,
between 4 and 10 percent less mass emissions, not an insignificant quantity,
would be detected with an action level of 20,000 ppmv instead of 10,000 ppmv
for the SOCMI sources tested. Secondly, the percentage of valves found
leaking would be only 0.1 to 4 percent fewer. For a model unit B, this
would mean a minor difference of one to 21 fewer valves found leaking
initially, at an annualized cost difference of $4 to $84 for the initial
repairs.
EPA has determined that, along with other reasons discussed in this
section, the potential emissions reductions attributed to a leak definition
of 10,000 ppmv warrant the small amount of additional work required to
maintain the few extra leaks detected. The work required to monitor and
repair the extra valves found leaking between 10,000 and over 20,000 ppmv
was also not considered burdensome. An analysis of valve leak detection and
repair programs for different leak definitions indicated that a leak
detection and repair program based on 10,000 ppmv would result in a credit
of $60/Mg of additional VOC recovered over a program based on 20,000 ppmv.
Therefore, 10,000 ppmv was selected as the leak definition.
Comment:
Another commenter (IV-D-46), however, stated that EPA is not justified
in exempting from the repair requirements all leaks below 10,000 ppmv. He
quoted a passage from the proposed standards to demonstrate that EPA is not
claiming that it knows that making repairs in 1,000-10,000 ppmv range will
result in net increases in emissions. The commenter noted that EPA is only
speculating that they will. He also pointed out that Table C-16 of the BID
shows that most repaired leaks result in lower emission levels.
4-51
-------
I
in
ro
TABLE 4-fr. SUMMARY OF PERCENT OF SOURCES DISTRIBUTION CURVES
AND PERCENT OF MASS EMISSIONS CURVES AT VARIOUS
ACTION LEVELS3
b Percent of Mass Emissions Attributable
Percent of Sources Screening Above to Sources Screening Above '
Valves
Gas
Ethyl ene
Cumene
Vinyl Acetate
Light Liquid
Ethyl ene
Cumene
Vinyl Acetate
Pump Seals
Light Liquid
Ethyl ene
Cumene
Vinyl Acetate
10,000
15
16
3.7
26
12
0.2
30
14
1.7
20,000
12
13
2.8
22
9
0.1
24
11
1.0
40.000
10
10
2.0
18
6
0.1
18
8
0.5
100,000
7
6
1.2
13
4
0
12
5
0.2
10,000
94
94
90
89
80
25
96
89
67
20.000
90
89
84
83
71
16
92
83
57
40.000
84
83
77
75
61
10
86
75
46
100,000
71
69
62
60
45
4
73
61
31
Curves are based on models derived from data collected during 24-un1t SOCMI study.
Screening values are in ppmv.
cThese values were based on the original leak rate/screening value correlations presented in the
Maintenance Study and have not been changed to reflect the new correlations developed in the
Technical Note on the revision of SOCMI emission factors. Based on a comparison of empirical
data, these values are not expected to change significantly.
-------
This commenter also wrote that even if the 1,000 ppmv level is inappro-
priate, EPA should have examined whether an intermediate cutoff (e.g. 8,000,
5,000,. or 3,000 ppmv) would better maximize net reductions. Neither the BID
nor the proposal discusses the possibility of different cutoff levels for
different types of emission sources.
Response:
The commenter (IV-D-46) correctly pointed out that there would be only
a potential for a net increase in emissions if an action level between 1,000
and 10,000 ppmv were selected. The potential increase in emissions that
could result from attempted repair of a valve with a screening value between
1,000 and 10,000 ppmv. But it is unknown precisely at what action level
maintenance efforts begin to result in increased emissions. Thus, EPA has
determined to use the 10,000 ppmv leak definition as its lower bound for
leak definition. Clearly, considerable emission reduction is achievable
using a 10,000 ppmv leak definition, and the key criterion in selecting a
leak definition is the mass emissions reduction achievable. Any leak
definition chosen would only be an indicator of whether a source was
emitting VOC in quantities large enough to warrant action (repair). A rise
in the screening value above the leak definition is an indication that steps
(such as repacking of valves or replacement of pump seals) must be taken to
lower the screening value below the leak definition and, thus, reduce the
emissions from the source. In this regard, certainly a leak definition of
10,000 ppmv accomplishes this goal and, based on the findings of the
Maintenance Study, results in an overall 71 weight percent reduction of
emissions using only simple on-line maintenance. Furthermore, the
monitoring technique selected for the standards results in a leak or no leak
determination and the leak definition should be easily implemented in this
technique. The 10,000 ppmv definition fulfills this requirement. EPA has
determined that using a lower leak definition would not increase emissions
reduction significantly, and the potential net benefit of a lower leak
definition is questionable. Therefore, a leak definition of 10,000 ppmv was
selected instead of a lower level.
4-53
-------
Comment:
One commenter (IV-D-17) said that raising the leak definition to
20,000 ppmv would decrease the maintenance required by 30 percent. He said
that this factor is of major significance since Section 111 of the Clean Air
Act requires the Administrator to take cost into consideration. The
commenter had previously (IV-D-17) submitted data to show how he had arrived
at his estimate which was based on valves in gas service. And in a later
letter, the commenter (IV-D-50) reiterated that the proposed standards were
not cost effective, citing as one reason the unreasonably low action level
for leaks.
Response:
Under any assumed leak detection and repair program, the maintenance
requirements would be directly related to the number of sources found
leaking at the selected action level (leak definition). For all 24 units,
only about 2 percent fewer gas valves were found leaking at a 20,000 ppmv
action level than at an action level of 10,000 ppmv. Again, relating this
to a model unit B used in the background document, only about 8 fewer gas
valves would be identified as leaking if the higher action level were used.
For valves in gas and light liquid service combined, about 21 percent fewer
valves, or 16 valves in a model unit B, would be found leaking. This
decreased number of valves represents a minor savings, as noted above, in
maintenance costs attributed to the fewer hours associated with screening
and repairing these valves. However, as noted in the previous response, the
incremental cost effectiveness of using a 10,000 ppmv leak definition
instead of 20,000 ppmv is a credit of about $60/Mg of additional VOC
recovered. Therefore, EPA has determined the emissions reductions for a
leak definition of 10,000 ppmv attainable at a reasonable cost (costs were
discussed in detail in Chapter 5 of the AID).
Comment:
The same commenter (IV-D-17) wrote that the change in calibration gas
from hexane to methane has effectively lowered the trigger point and
recommended 20,000 ppmv as a leak definition. Another commenter (IV-D-7)
objected to the change, noting that EPA's justification was that methane is
available. He found this justification inadequate (See Section 12.1).
4-54
-------
Response:
The final regulatory analysis for the SOCMI valve leak detection and
repair program was based on SOCMI screening data measured with an instrument
calibrated to methane. EPA's determinations regarding regulations for SOCMI
fugitive emission sources were made based on these SOCMI data, not on other
data obtained with another instrument calibrated to a different gas.
Therefore, EPA sees no reason to alter the leak definition for differences
in calibration gases.
The availability of methane was not the sole reason for its selection
as the calibration gas for Reference Method 21 when used for SOCMI
screening.. It is true that methane is more readily available than hexane in
the concentrations required for calibration. As discussed in Section 12.1
Test Methods, the SOCMI data gathered to support these standards were
collected using methane as a calibration gas. Furthermore, as explained
more fully in Section 12, the differences between the results obtained by.
the OVA*-methane system (used in SOCMI studies) and the TLV*-hexane system
(used in refinery studies) are not really significant at the 10,000 ppmv
action level. Therefore, data collected using either the OVA-methane system
or the TLV-hexane system can be used interchangeably at the 10,000 ppmv
action level.
For some types of monitoring instruments allowed in the Reference
Method, however, methane cannot be used as the calibration gas. Photoioni-
zation instruments, for example, may be useful in certain SOCMI units, these
instruments do not respond to methane. An alternative calibration gas has
been added in the regulation to allow for this situation but is not
restricted to a single type of analyzer. Based on the comparison presented
above, hexane has been specified as the alternative calibration material.
Moreover, one industry study (IV-A-17) indicates a comparable number of
leaks determined using hexane and methane with the Century OVA* and
Bacharach TLV.*
Comment:
Variation in repeat measurements of the same valve were also cited as
the basis for raising the leak definition to 20,000 ppmv (IV-D-5; IV-F-1,
*Mention of trade names does not represent endorsement by EPA.
4-55
-------
No.4). In a later letter, the commenter (IV-D-40) stated that the level
should be high enough to allow for variations and still identify only
significant leakers, if a single leak definition is to be set for the whole
industry.
Another commenter (IV-D-46), while agreeing with a focus primarily on
large leakers, expressed the belief that the cutoff levels should be lowered
to include more leaks. He further stated that ..EPA has chosen the most
cost-effective, sensible approach.
Response:
Repeat screening was addressed in a report filed in Docket Item
No. IV-A-7. In this report, the variability in repeating precise screening
values was given as X/5.6 to 5.6X. While this does represent a wide range
of values, it relates to obtaining a discrete value. Reproducibility of
leak/no leak determination is a more appropriate consideration when applying
Reference Method 21 since a precise screening value need not be determined.
In Docket Item No. II-B-24, reproducibility of leak/no leak determination at
a 10,000 ppmv action level was given as 90 percent. In another research
study (IV-A-30), reproducibility of leak/no .leak determination at a
10,000 ppmv action level was given as greater than 94 percent. EPA,
therefore, finds no basis for changing the leak definition for the SOCMI
standards.
As discussed in response to previous comments, EPA has determined that
lowering the leak definition is not warranted. The standards are based upon
emission reduction achievable. The leak definition is only an indicator of
whether a source is emitting VOC in quantities significant enough to require
action. The utility of the 10,000 ppmv leak definition to this purpose has
been thoroughly presented in this section. In lowering the leak definition,
the emissions reduction achieved would probably not be increased much and
the point may be approached that attempted repairs could even result in
increasing emissions.
Comment:
Variability in response factors was also cited as supporting evidence
that the leak definition should be raised (IV-D-7). The commenter objected
4-56
-------
to EPA's reasoning that a variability of 2 to 3 would not have a significant
impact on the decision as to whether a leak exists when compared to the
total number of potential leak sources. He said that the comparison should
be made to the number of leaking sources and not the total number of
potential leaking sources. He further stated that the variability in
response is 0 to 571, not 2 to 3 as reported by EPA. He charged that EPA
was aware of this variability before proposal because it was reported in
September, 1980, in EPA Report No. 600/2-81-002. The commenter admitted
that calibration of the instrument for each chemical involved would be
impractical but suggested that an allowance should be made for the
variability in response factors by defining leak concentrations on the basis
of the type of unit surveyed.
Response:
The commenter accurately points out that a wide variation in response
factors (0-571) was determined in laboratory testing of two portable VOC
analyzers used for a number of organic chemicals. But it is important to
point out that, even though this wide variation existed for all chemicals
tested, 90 percent of these chemicals had response factors between 0.1 and
10 (IV-A-8). In addition, many of the chemicals falling outside of this
range either are not SOCMI chemicals or are heavy liquids. And other
monitoring instruments may demonstrate better response factors for specific
chemicals that are to be monitored.
The monitoring requirements of Reference Method 21 result only in a
determination of leak or no leak at an action level (leak definition), not a
concentration measurement. As a result, the effect of response factor
variation on percent leaking estimates is dampened since precise concentra-
tion measurements are not required. These effects were analyzed in the EPA
report on the 24-Unit SOCMI study (IV-A-14). The conclusion of this
analysis was that only a small reduction (approximately 3 percentage points)
in the estimated leak frequencies was evident for gas valves in high leak
service, while no distinguishable differences were seen in all other cases.
EPA, therefore, has determined that no allowances are needed in the
leak definition to account for variation in response factors.
4-57
-------
Comment:
One commenter (IV-D-17) wrote that EPA has defined a leak based simply
on the ability to repair the leaking components. Although he commenter
admitted that data may show that attempted repair of valves leaking at about
1,000 to 3,000 ppmv could result in an increase in emissions, it was
impossible to understand how these data restrict the upper limit to
10,000 ppmv as proposed by EPA. The commenter suggested that it can be
easily concluded from the present data that repairing leaks with screening
values of 1,000 - 2,000 ppmv may not result in net reductions of emissions.
On this basis, the commenter agreed that it is certainly important not to
cause small leaks by requiring tightening of valves with a low-level leak
definition. However, he also concluded from all available data that leaking
valves could be defined as 1,000 - 20,000 ppmv which would significantly
reduce the percent of unsuccessfully repaired valves and almost eliminate
the increased emission rate of some attempted repairs. According to the
commenter, there is no justification in these data to restrict the upper
limit of a leak definition to 10,000 ppmv as proposed by EPA.
Response:
The standards are based on the emission reduction achievable, not on
the ability to repair. The leak definition provides an indication of the
significance of leaks in terms of mass emissions. Using a leak definition
of 1000 ppmv, the emissions may actually increase as a result of attempted
repair. This phenomenon was discussed earlier in this section. But, at a
leak definition of 10,000 ppmv, simple on-line maintenance resulted in
71 weight percent reduction of emissions. For those valves that can be
isolated, off-line maintenance will be done. For valves that cannot be
repaired on-line and cannot be isolated, repair will be done at turnarounds
(shutdowns). These efforts will result in a higher weight percent reduc-
tion. Therefore, EPA decided to use 10,000 ppmv as the leak definition.
Comment:
Comments were also received concerning the leak definition applied to
pumps. One commenter (IV-D-15) argued that the proposed requirements to
inspect dual seals for visible leakage of fluid and the assumption that any
4-58
-------
visual detection of liquid dripping will constitute a leak which must be
repaired may result in significant expenditure of time and effort without
any effect on VOC emissions. In most instances, leakage of the process
fluid cannot occur across the inboard seal due to the higher pressure of the
barrier fluid in the stuffing box, and liquid dripping from the outer seal
will be only the barrier fluid. No light liquid VOC will be present and no
fugitive emissions will occur. Therefore, the commenter stated, defining
any visible dripping from the seal area as a leak is unjustified. Further-
more, since by their very nature all mechanical seals will leak some barrier
fluid when in operation, the proposed definition of a leak would have most
pumps in a perpetual state of leakage.
The commenter asked for a more specific definition of a leak. He
suggested that specifying the number of drops per minute would be one method
to avoid this situation. Rule 466 of the South Coast Air Quality Management
District in California specifies a rate of 3 drops per minute as consti-
tuting a leak for pump seals. The commenter recommended that a similar
provision should be included in the final standards. In addition, a
suspected leak of VOC observed by visual inspection should be confirmed by
monitoring prior to initiating repair work. He stated that if the VOC is
below 10,000 ppmv, pump repairs should not be made.
Another commenter (IV-D-17) wrote that in many instances, the barrier
fluid could be water, so there would be no reason to repair the seal where
the barrier fluid pressure is higher than the pump pressure. He
recommended rewording Section 60.482(a)(4) as follows:
Each pump shall be checked by visual inspection each calendar week for
indications of liquids dripping from the pump seals. If indications of
liquids dripping from the pump are seen, the vapor emissions shall be
monitored by the methods specified in 60.485. A vapor concentration
containing greater than ten percent VOC at greater than 10,000 ppmv
above background shall constitute a leak.
Response.:
At proposal, the standards for pumps in light liquid service required
the use of dual seals with a non-VOC barrier fluid systems. A weekly visual
4-59
-------
inspection of the pump seal was required to identify a leak. As discussed
in Section 4.8 on Dual Seals, the final standards for light liquid pumps
include a work practice standard in addition to the equipment standards
presented at proposal. The work practice standard uses a leak definition of
10,000 ppmv to be determined in accordance with Reference Method 21.
Under the work practice standards pumps would be monitored monthly to
detect leaks using a 10,000 ppmv action level (leak definition). Pumps
screening above 10,000 ppmv would be defined as a leak, requiring repair as
soon as practicable. The first attempt at repair would be made within
five days of detection and repair would be required within 15 days. As an
alternative to the work practice standard, the equipment requirements (dual
seals, etc.) could be met.
A leak under the equipment standard for light liquid pumps is still
defined as any visible leakage from the seal area. The "drops per minute"
format was considered but was not chosen because leakage from pump seals is
often in the form of spray or mist, or is exhibited by ice formation around
the seal area.
Under the equipment standard, a leak is also detected upon failure of
the seal system and/or the barrier fluid system which would be indicated by
a sensor. Again, when a leak is detected, repair must be effected within 15
days of detection. Regardless of seal type/arrangement, visible leakage
from the seal area is generally indicative of seal wear. To prevent exces-
sive wear that could possibly result in catastrophic seal failure, the seal
should be repaired soon after leakage is detected. Therefore, visible leak-
age from the seal area is defined as a leak under the standards for pumps.
4.4 SAFETY CONSIDERATIONS
Some of the comments received on the proposed standards dealt
specifically with safety considerations of the control technologies
presented in the support documents. The four areas commented on include
closed vent systems and incinerators, rupture disk/pressure relief device
installations, hazards during monitoring, and double valve requirements.
Some other comments concerning safety considerations are included under the
individual equipment types involved because the comments focus more on the
control technology for specific equipment types than on the safety aspects.
4-60
-------
Comment:
One commenter (IV-D-17) stated that, as presently drafted, the
requirement for closed vent systems is potentially unsafe. The following
reasons were given for these safety concerns.
First, the conveyance of explosive vapors or gases must be conducted
under lean (<40 percent of the lower explosive limit) or rich (0 percent
oxygen) conditions. Fugitive emissions from compressor and pump seals would
not normally be present. The design of a sealed conveyance system which
will not have hydrocarbon concentrations mixed with oxygen in the explosive
range will be a difficult, if not impossible, engineering job. The
engineering requirements for such a system include:
either a sealed system without oxygen or a system which
conveys vapor gas at <40 percent of the lower explosive limit
at all times,
the use of a water seal or equivalent to provide flashback
protection between the conveyance system and the incinerator,
the use of a blower/compressor to provide motive force to
overcome seal and system pressure drop, and
a turn-down capability in the order of magnitude range.
Secondly, the incinerator designed to burn the anticipated fugitive
emissions efficiently will be unable to handle the volumes and pressures of
volatile liquids and gases discharged under emergency conditions.
Thirdly, enclosed combustion devices are sensitive to conditions of
feed. Under emergency conditions the system could see hot gases or cold
(-150°F) liquids. An enclosed combustion device would have some difficulty
meeting such operating requirements.
The commenter noted that the above concerns are best understood by
considering an example. A chemical plant pumps liquid propylene at 400 psi.
On occasion the pump seals blow, which triggers an automatic emergency pump
shutdown. Under such conditions of emergency releases, an enclosed
combustion device designed for fugitive emission loads is not believed to h~
able to handle the flow and pressure3 loads created by the abo"p emergency
situation. Such flow and pressure loads are said to be orders of magnitude
higher than the design load fo" fugitive
4-61
-------
Another commenter (IV-D-18) wrote that venting of emergency releases
from rupture disks and emergency relief valves on larger tanks and vessels
to incineration devices will create operational hazards due to design
1 imitations.
Response:
The requirement of a closed vent system for fugitive emissions presents
a difficult, but not insurmountable, design situation for safely conveying
explosive vapors or gases if the system is used .for transmission of fugitive
emissions only. Balancing purge gas flowrates with anticipated releases
(e.g., seal failures) must be done to avoid exceeding 40 percent of the
lower explosive limit. This could potentially result in overdilution of
"normal" fugitive emissions. But this same type design problem is encoun-
tered for any vent system where variable flows of explosive materials must
be handled. Large variations in flowrates can typically be handled with
flare systems which are now allowed, since these can achieve turndown ratios
as high as 100:1.
Incinerators have limitations with regard to normal vent flows and
flows experienced under emergency conditions. The equipment requirement is
not for control of emergency flows; it is for disposal of low-volume
fugitive emission streams. Incinerators, along with vapor recovery systems
and properly designed and operated smokeless flares, are acceptable control
devices for fugitive VOC emissions. VOC discharged under emergency
conditions, such as overpressure relief, are not covered under the SOCMI
fugitive VOC standards.
The safety and operational problems associated with incinerators cited
by the comments relate to incinerators designed for control of fugitive
emissions streams only. Incinerators are generally installed to control
much larger streams than those expected for fugitive emissions sources. EPA
believes that the incinerators designed for these larger streams, when
available, will be used to control fugitive emissions streams as well.
Since they would be designed for larger total flow, these incinerators would
be capable of handling larger variations in stream characteristics resulting
from process upsets. Therefore, it is unlikely that variations in the
4-62
-------
fugitive emissions streams will affect the operation of the incinerator.
Even large changes in fugitive emissions streams would be small due to the
small contribution of fugitive streams to the total flow. For example, the
severe temperature changes referred ,to by one commenter would be dampened by
dilution of the fugitive emissions stream into the main flow of the
incinerator. Likewise, concentration differences and shifts in fugitive
emissions will be dampened by the main incinerator stream so that
incinerator operation will not be affected.
Comment:
One commenter (IV-D-6) stated that safety professionals at his organi-
zation strongly recommended against installing a rupture disk ahead of a
relief valve. This combination increases the complexity of a relief system,
reducing the reliability and increasing the probability of relief system
failure. The commenter pointed out that a small leak in the rupture disk
could lead to accumulation of pressure between the disk and the relief
valves, causing the disk pressure to be increased significantly. A small
leak of this type is believed to have contributed to the rupture of a nitro-
aniline reactor at one of the commenter's plants. The commenter further
stated that safety audits of facilities which use rupture disk/relief device
combinations often find full operating pressure downstream of the rupture
disks.
Response:
The potential safety hazard resulting from leakage into the space
between a rupture disk (RD) and a pressure relief device (PRO) can be
minimized by proper installation of the disk assembly (ensuring the right
seal without damaging the disk) and maintenance of the safety devices
(ensuring that no leakage results from corrosion of the disk). Engineering
codes (ASME and API) recommend as minimal requirements that safety indica-
tors such as pressure indicators, petcock vent valves, etc., be installed
between the RD and PRO to avoid this problem. An indicator of this type was
included in the cost of RD/PRD installations in the background document.
This issue is discussed in greater depth in Section 4.5 on pressure relief
devices.
4-63
-------
Moreover, the standards for pressure relief devices require that no
detectable emissions be maintained. The standards do not specify how this
level of emissions are to be met. Other means, such as piping to a flare,
could be used as long as this level of emissions is met.
Comment:
One commenter (IV-D-26) noted that the EPA report "Frequency of Leak
Occurrence for Fittings in Synthetic Organic Chemical Plant Process Units"
points out that adverse weather, heavy rain, high winds, and extreme cold
often delayed the monitoring studies at several process units. In addition,
he said that each appendix of the report contains a section of sampling
problems and the problems of instrument unreliability and the hazards of
climbing up and down process units. The commenter wrote that with this
contractor experience, he found it alarming that EPA wishes to impose on
SOCMI such a frustrating and potentially hazardous task on a monthly (winter
and summer) basis.
Another commenter (IV-D-48) stated that EPA has ignored the fundamental
questions of worker safety and monitorability of valves in proposing the
regulation. He wrote that for safety purposes, a valve should not be
considered accessible if the valve cannot be reached safely from ground
level or from a fixed platform. A valve should not be considered accessible
for monitoring purposes if: (a) monitoring must be done from a movable
ladder or "cherry picker," (b) it is located in high temperature and/or high
pressure locations which are not routinely accessible for monitoring,
(c) they are located in sealed or barricaded operating areas because of
toxicity or explosive concerns, or (d) they are totally enclosed by
insulation or drip covers to protect against corrosive leaks.
The commenter suggested an exemption to the periodic monitoring
requirements for valves which are inaccessible for b'ne or more of the
reasons above. He said that, as an alternative to periodic testing, such
inaccessible valves could be tested during shutdown periods by pressurizing
the system with an appropriate gas to 100 psi, or to the process pressure,
whichever is less, and checking the inaccessible valves for leaks by the
appropriate leak detection procedure. However, such periodic checks of
inaccessible valves should not be required more than once per quarter.
4-64
-------
Response:
Contractor screening studies were conducted under time constraints
that required sampling under adverse weather conditions. The timing of
the periodic monitoring within a specific monitoring interval as required
under the proposed standards, on the other hand, will be left to the
discretion of the owner/operator. Plant personnel can schedule monitoring
around unfavorable meterological conditions. Instrument reliability is
attained through proper maintenance of the instrument and use of the correct
sampling techniques. Training of personnel doing the monitoring will help
achieve this result.
EPA recognizes that some valves are difficult to monitor, due to safety
or access. Valves with difficult access can be virtually eliminated in new
plants and, thus, would be subject to the leak detection and repair program
given in the standards. But, for modified or reconstructed process units,
EPA is requiring an annual leak detection and repair program for valves
which are difficult to monitor. (See also Section 4.2.4.)
Valves which are unsafe to monitor will exist in new process units, as
well as in modified or reconstructed units. For these valves a plan is
required that defines a leak detection and repair program to conform as much
as possible with the routine monitoring requirements, given that monitoring
should not be undertaken under unsafe conditions. Valves that are unsafe
to monitor are those, judged by the owner/operator, which could expose
monitoring personnel to potential hazards from temperature, pressure, or
explosive process conditions. (See also Section 4.2.4.)
Comment:
One commenter (IV-D-12) wrote that installation of double valves, which
are required for open-ended vent valves that are not capped, would present
severe safety complications in a major furfural step at his facility.
Another commenter (IV-D-17) stated that many processes used block-and-
bleed techniques to avoid process contamination or where explosive or
reactive mixtures are present. This commenter recommended a change to
§60.482(e)(2) to exempt such bleed (vent) valves from the standards.
4-65
-------
Response:
The first comment was made in the context of furfural production units,
not other process units using furfural as a solvent in refining or distilla-
tion operations. The commenter (IV-B-8; IV-E-4) subsequently supplied
further clarification, indicating that his concerns were the large discharge
valves used to dump the waste solids remaining after completion of the batch
reaction cycle. These valves were described by the commenter to be more
like manhole covers than open-ended valves. Manhole covers and flanges are
not regulated by the NSPS for SOCMI fugitive VOC emissions.
The block-and-bleed system described by the second commenter has been
provided for in §60.482-6(a)(2). Where a block-and-bleed system is
installed and in use, the bleed valve (second valve) must "seal the open end
at all times except during operations requiring process fluid flow through
the open-ended lines." The function of bleed valve of such a system is to
vent the space between the two block valves; when venting this space, the
bleed valve is "in service." Therefore, the bleed valve of a block-and-
bleed system can remain open when it is venting the space between the block
valves.
4.5 PRESSURE RELIEF DEVICES
Several comments were received concerning the proposed standards for
pressure relief devices (PRO) in SOCMI. The five major areas receiving
comments were: (1) the proposed standards and pressure relief system design,
specifically relating to fragmentation of rupture disk (RD); (2) testing of
inaccessible PRDs; (3) retrofitting a RD to a PRO; (4) increased costs due
to RD installation; and (5) complexity of fugitive emissions control
devices. Each of these areas is addressed separately.
Comment:
Several comments were received concerning rupture disks. One commenter
(IV-D-24) wrote that the use of rupture disks upstream of pressure relief
devices introduces the possibility of debris from a fractured disk
preventing a relief device from properly reseating. Such an incident could
increase VOC emissions, outweighing any advantage of a rupture disk. The
commenter requested further study on this item.
4-66
-------
Another commenter (IV-D-17) wrote that the chemical industry normally
objects to the identification of rupture disks as the sole control device
for limiting fugitive emissions from relief valves. While there are certain
conditions which require installation of rupture disks under relief valves
(corrosive service, for example), there are better solutions under most
circumstances. Relief valve designs are available today which utilize an
elastomeric 0-ring seat as a backup to the conventional metal-to-metal seat.
The high differential pressure which exists in many relief valve applica-
tions is still sealed primarily by the metal-to-metal seat while any leakage
is controlled by the elastomeric 0-ring seat. The elastomers of choice are
expensive but do not appreciably add to the cost of new relief valves.
There are conversion kits for retrofitting existing relief valves with
0-ring seats which, although expensive, are an attractive alternative to the
rupture disk approach. Relief valves with 0-ring seats have been tested and
found to be bubble tight up to over 95 percent of set pressure and reseat to
this condition through several cycles. EPA has already accepted the 0-ring
design as an alternative design under the vinyl chloride regulation, and
this concept should be recognized for control of fugitive hydrocarbon
emissions in the BID.
One commenter (IV-D-18) further noted that a serious problem with the
recommended use of rupture disks is that, due to corrosion or fatigue,
rupture disks often fail prematurely. This will ultimately cause problems
with downtime and associated costs, as well as the increase in VOC emissions
due to the otherwise unnecessary opening of the vessel to replace the
rupture disks. Based on the above considerations, the commenter recommended
that pressure relief mechanisms be serviced during scheduled turnaround or
downtime to minimize those emissions.
Two comment letters (IV-D-6; IV-D-17) disagreed with the assumption
that the installation of a rupture disk beneath a relief valve eliminates
leakage. The commenters1 experience had indicated that this is not the
case. They wrote that considerable leakage has occurred at the gaskets
between the disk holder and the mating of the valve when using the .
manufacturer's instructions for maximum allowable torque on installation.
4-67
-------
After flow testing it became worse because of loosening of assembly bolts
due to forces transmitted from the valve thrusts. The commenters stated
that the obvious answer is to ignore the manufacturer's recommendations and
use higher torque to stop leakage. However, both commenters stated that
this can cause deformation of the disk and affect bursting pressure.
On the other hand, another commenter (IV-D-2; IV-D-33) pointed out that
new technology since 1966 allows the use of reverse buckling rupture disks
under relief valves. These disks do not fragment as the tension loaded
disks do. When using these disks, the piping changes to avoid fragmentation
into the valve are unnecessary. In fact, it would be preferrable to have
the rupture disk in a direct line with the relief valve to give the valve
the opening. Furthermore, the flow coefficients are seen with the two
directly in line and not offset.
One commenter (IV-D-6) was pleased that a performance standard has been
proposed for safety/relief valves in place of the specification standard
previously proposed. The commenter wrote that relief valve/rupture disks
(in specific appropriate applications), vapor collection and control
systems, and improved relief valve design should all be allowed. This
flexibility is needed to meet the wide variety of process requirements in
SOCMI facilities.
The commenter further wrote that the prevention of fugitive emissions
from a relief valve needs to be achieved by proper design and maintenance of
the valve. There are many equipment options including pilot operated
valves, valves constructed of soft materials, and valves with elastomeric
0-ring seats.
Response:
As presented in the preamble to the proposed standards, PRD's are one
of the few fugitive VOC emissions sources for which a standard of perfor-
mance can be established. There are various alternatives for complying with
the "no detectable emissions" performance standard proposed for fugitive VOC
emissions from PRD's in gas service in SOCMI. In proposing a performance
standard, any alternatives (such as RD/PRD combinations and venting to a
flare) may be used if they result in no detectable emissions. The
4-68
-------
particular equipment chosen is dependent upon the process and material being
processed.
The use of a RD/PRD combination was used in the BID as an equipment
specification for the regulatory analysis to evaluate the viability of a
standard for PRO fugitive emissions. Other means also exist to eliminate
fugitive emissions (leakage) from PRD's.
RD/PRD combinations have been commonly used in pressure relief systems.
Two examples of RD/PRD applications are elimination of leaks of toxic
substances and prevention of fouling of PRO internals in polymer processing
units. But, as noted in one of the comments, existing PRDs could also be
converted to soft-seat design, in some cases at considerably less cost than
installing a RD assembly. These would be acceptable alternatives if they
resulted in no detectable emissions.
From a safety standpoint, some RD designs may not be suitable for the
RD/PRD arrangement. But newer RD technology (e.g., reverse buckling RD's)
eliminates many of the potential problems of PRO malfunction due to RD frag-
mentation. Where other RD designs are used, vent piping may be modified to
avoid the problems resulting from fragmentation. Offset piping was
considered in the BID analysis.
Leakage may occur around the gaskets of an improperly installed RD or
through pinholes in a RD resulting from corrosion. This results in a
potential safety hazard since the bursting pressure of the RD is shifted.
Leakage problems, however, can be minimized by proper installation and
maintenance practices. Increases in pressure between the RD and PRO
resulting from such leakage would be indicated by the devices suggested by
ASME standards for RD/PRD installations. The ASME code requires that the
space between RD and PRO be provided with a pressure gage, try cock, free
vent, or suitable telltale indicator to permit detection of disk leakage or
rupture.
Comment:
One commenter (IV-D-15) said pressure relief devices are frequently
inaccessible. He showed data from contractor studies which showed how many
were inaccessible during recent emissions testing. He said the proposed
4-69
-------
standards make no allowance for excluding inaccessible sources and the
economic analysis did not address the extra cost incurred in monitoring
these sources.
Another commenter (IV-D-18) agreed, writing that approximately
90 percent of all pressure relief devices are inaccessible and would require
scaffolding to be constructed for sampling purposes. This represents an
enormous cost consideration. Further, the commenter noted that requiring
monitoring personnel to access pressure relief devices areas on a regular
basis makes monitoring unnecessarily hazardous work. The low level of
emission control achieved does not justify the hazards inherent to
monitoring. This requirement should, therefore, be removed.
Another commenter (IV-D-17) suggested that EPA make it clear in the
proposed standards that pressure relief devices which are tied into a closed
flare header system do not need to be monitored for "no detectable
emissions" levels after each discharge.
Response:
The standard for pressure relief devices in SOCMI requires a
performance test (monitoring) on an annual basis only, in order to verify
that the PRO is maintained at 500 ppmv or less. Annual performance testing
is not considered to be so frequent that it is burdensome. Annual testing
could be scheduled during periodic PRO inspections which are typical of many
industry safety practices.
Additionally, PRDs must be checked within five days after each relief
discharge to ensure that a condition of 500 ppmv or less is maintained. The
outlet of the PRO does not have to be monitored if it is piped to a closed
vent system. The definition of a closed vent system includes control
systems such as enclosed combustion devices (incinerators, boilers, process
heaters), vapor recovery systems, and elevated smokeless flares. Thus,
pressure relief devices that are connected to a closed flare system are in
compliance with the standards for pressure relief devices. For example, for
RD/PRD combinations, since the RD would need replacing after such a relief,
the PRO could be monitored when the system was put back in service. For a
PRO alone, the PRO would have to be monitored after a relief to ensure that
4-70
-------
the valve is reseated properly. Monitoring of PRD's is not expected to be
frequent. PRD's that are relieving frequently indicate improper PRO design
or process operation too close to its limitations.
Comment:
A commenter (IV-D-17) wrote about the use of rupture disks. He said
this practice can lead to problems which may not be readily apparent,
especially on existing facilities. The solution to these problems is not
straightforward nor simple. Significantly higher costs than projected by
EPA will undoubtedly result. For example, the ASME pressure vessel code
(Section VIII) requires one of the following two options in sizing relief
valves for such cases:
1) The rated relieving capacity of the relief valve must be
reduced to 80 percent, or
2) The combination rupture disk/relief valve shall be tested to
establish the official "combination capacity factor."
If the use of rupture disks were required at existing facilities under
Option 1, the capacity of the relief valve would be reduced by 20 percent.
This would likely not be acceptable in many cases and would thus require a
change to a new,.larger relief valve.
Manufacturers are testing and developing the combination capacity
factors, but since there are many such possible combinations, Option 2 is
currently very limited in availability. Even if available, it would likely
require a change in the relief valve. Furthermore,, the commenter expressed
the understanding that once a combination valve and disk has been installed,
no substitution is allowed for either element of the combination since, the
combination factor is good for that particular combination of type and
manufacturer. In either option, it is likely that relief valves with a
capacity greater than that required would be installed. This may lead to
chattering of the valve and extreme vibration of the piping leading to a
flare or other control device. There have been cases where this vibration
has led to failure of the piping. In order to resolve this, larger flare
header piping could be required.
4-71
-------
One cotnmenter (IV-D-2; IV-D-33) pointed out that for retrofit applica-
tions, if a rupture disk/relief valve combination has been flow tested by
the National Board of the ASME Code, a larger relief valve will not be
required. Recent testing of flow coefficients for rupture disk/relief valve
combinations have shown the coefficients to be between 0.95-0.99. For these
combinations it would not be necessary to use a larger relief valve. For
these combinations which have not been tested, it wi-11 be necessary to down
rate it to 0.8.
The commenter also wrote that in retrofit cases, the user may have
fixed piping upstream and downstream. The valves cannot be easily moved up
several inches to allow the space for a rupture disk holder without
thousands of dollars worth of piping changes. The recommended solution is a
rupture disk welded into a flange that fits down in the piping and has only
a minimal height of 1/8 to 1/4 inch. The price of this weld disk and holder
is about half the price of a standard disk and holder.
The commenter also noted that it would be an excellent idea to
reference the ASME code. In addition to requiring a monitor on the space
between the rupture disk and relief valve, the ASME code requires the space
to be vented to maintain safety, since the rupture disk is a pressure
differential device.
Response:
These comments primarily focus on the regulatory analysis presented in
the BID. One of the regulatory alternatives was based on use of RD/PRD
combinations to eliminate fugitive VOC emissions from relief devices and was
used to evaluate the feasibility of standards for relief valves in SOCMI.
It is important to note, however, that the performance standard for relief
devices is "no detectable emissions," providing the owner/operator the
flexibility to choose the means to achieve that level.
The costs presented in the BID for retrofitting a rupture disk to an
existing PRO included a new PRO. In addition, no credit was assumed for the
old PRO that was replaced. As the second commenter noted above, however, a
larger relief device may not be needed in all cases. Therefore, the costs
presented in the BID are expected to be conservative.
4-72
-------
Since proposal, the costs of new and retrofitted RD/PRD systems have
been reevaluated. These costs, in last quarter 1978 dollars, were
determined to be $3485/new system and $4025/retrofit system. Additional
labor was required for retrofit applications. Details of the revised cost
analysis were presented in the AID.
The comments also deal with rating of RD/PRD combinations with respect
to relieving capacity. As noted by the commenter, ASME codes provide design
criteria for relief systems that incorporate RD's and PRD's. If a larger
PRO is required when retrofitting a RD to the system, the potential problems
with valve chattering and pipe vibration attributable to the oversized PRO
could be avoided through good overall design of the relief system. The ASME
Code also provides other requirements for system design. Many of these
requirements are current industry practice and many have been incorporated
into the standard for pressure relief devices.
Comment:
Two comment letters (IV-D-6; IV-D-17) noted that rupture disks have
been presented as cost savers because they allow testing of relief valves
while they are in place. They stated that only one type of rupture disk can
stand up to these pressures. This type of testing yields inadequate infor-
mation because only the pressure at which leakage occurs is obtained and no
information on relieving capacity is obtained. Testing in this manner is an
unacceptable substitute. One comment letter (IV-D-17) added that, on this
basis, industry will continue to test safety valves in shops from a safety
standpoint, and therefore, the claim that rupture disks are cost saving
installations is not supported by actual facts.
This commenter (IV-D-17) further stated that rupture disk
installations, when used, are much more complex and expensive than systems
that are usually presented in order to overcome adequately safety concerns.
Tests have indicated that a pinhole leak through a rupture disk can equalize
pressures on both sides of the disk and lead to vessel overpressure followed
by greatly reduced relieving capacity when the disk finally ruptures. For
this reason, at least one company requires all rupture disk-relief valve
installations to include a pressure switch in the space between disk and
4-73
-------
relief valve along with a control room alarm when high pressure is sensed.
To meet safety standards then, the rupture disk installation becomes much
more expensive.
Another commenter (IV-D-2; IV-D-33) pointed out that the cost analysis
is conservative. He said, in the BID, a block valve was assumed under a
pressure relief device/disk combination. The commenter stated that, in
practice, about half of all oil companies use block valves, while half do
not. He went on to say that the cost of a new installation and retrofit
should be similar since the safety relief valve will seldom need replacing
in a retrofit, due to derating. In addition, a major cost to move or change
downstream piping to accommodate a RD assembly in a retrofit can now be
avoided with some RD designs that fit inside existing piping.
Response:
The proposed standard for PRD's requires a performance level of no
detectable emissions. It is not an equipment requirement. In the BID, one
control alternative (RD/PRD combination) for attaining this level of
performance was presented to evaluate the relative benefits of NSPS for PRO
fugitive emissions. In practice, any control techniques limiting emissions
to less than 500 ppmv above background are acceptable.
Typical industry safety practices call for periodic inspection and
testing of PRD's. These practices would be considered the baseline control
level (current industry practice). When a RD is installed below a PRO, this
testing will probably be done in a shop. This procedure could cost about
$120/PRD for removal, inspection, and reinstallation of the PRO. Although
probably more expensive than field testing, this cost is certainly
affordable and would provide an opportunity for a more thorough maintenance
check than possible under field test circumstances.
As discussed previously in the AID, the costs for RD/PRD systems (new
and retrofit) were reevaluated after proposal. Both the BID estimates and
the current cost estimates included indicators for the space between the RD
and PRO, as required by ASME codes. The costs presented for RD/PRD
combinations demonstrate the affordability of the performance standard for
pressure relief devices. But it is not the only control option applicable
4-74
-------
to this source. Piping the discharge to a flare header is another alter-
native that could be used to meet the "no detectable emissions" limit.
Comment:
One commenter (IV-D-6) stated that design, and use of incinerators and
vapor collection systems are a complication for relief valves because of the
wide range of flow which must be controlled. Further complexities are added
in the need to minimize 0- in the flare header.
Response:
Incinerators and vapor collection systems are not specifically required
by the standard for fugitive VOC emissions from PRD's. The performance
standard requires no detectable VOC.emissions (less than 500 ppmv. above
background), leaving to the owner/operator the choice of method to use to
attain this performance. In addition, the emergency relief conditions,
which add complexity to control system design, are not covered by the
standards for fugitive VOC emissions.
4.6 COMBUSTION DEVICE
The comments received concerning combustion devices focused mainly on
the following areas: (1) incinerator operating specifications, (2) limited
choices of control alternatives, (3) preclusion of catalytic incinerators,
(4) bypass of control system, (5) incinerator design complexities, and
(6) the choice of the 95 percent control level. The bulk of the comments
relate to comments received on other technology areas including flares,
pressure relief devices, and closed vent system.
Comment:
Three comments addressed the operating specifications for the
combustion device. One commenter (IV-D-17) wrote that the EPA incinerator
requirement is wasteful from an energy standpoint. He felt that the
incinerator would probably require supplemental fuel to maintain temperature
at minimal firing. The commenter also cited an inconsistency between the
proposed residence time and those presented in the technical support
document. Another commenter (IV-D-28) called the temperature-residence time
specification for incinerators arbitrary since the standards apply to
substances having a wide variety of thermal characteristics.
4-75
-------
One cormnenter (IV-D-6) noted that the design specifications for
incinerators are incorrect because they do not specify an excess air or
oxygen concentration in the off-gas. The commenter suggested that the
incinerator design specifications should be changed to a performance
standard because the failure to set all necessary parameters can lead to
installation of improper and ineffective control equipment. He further
suggested that since the alternative to an incinerator is a vapor recovery
system of 95 percent efficiency, the incinerator standard should be changed
to 95 percent design efficiency.
Response:
The BID for fugitive VOC emissions from SOCMI sources reported the
limited data available on VOC control efficiency for various combinations of
temperatures and residence times. The temperature and residence time
specified in the proposed regulation were based on data analyzed in an EPA
memo "Thermal Incinerators and Flares," dated August 22, 1980 (II-B-31).
The data base contained in this memo included Union Carbide laboratory
studies, EPA and industry field tests, and 147 tests on existing incinera-
tors in Los Angeles county. These data indicate that greater than
98 percent efficiency is attainable by incinerators operating at 1500°F
(816°C) and 0.75 seconds residence time. The memo concludes that 98 percent
efficiency, or less than 20 ppmv in the exhaust stream, is achievable in
many situations at less than 1600°F (871°C) and 0.75 seconds residence time.
And destruction efficiencies of better than 93 percent are possible for
1400°F (760°C) and 0.75 seconds residence time.
While thermal incinerators are proven control devices for destruction
of VOC emissions, they are not the only enclosed combustion devices that
could be used. In fact, boilers and process heaters already existing
on-site are expected to be used for eliminating the small VOC streams
covered by the standards. In order to ensure "that these combustion systems
achieve the requisite degree of control, temperature-residence time require-
ments for enclosed combustion devices have been retained in the final
regulation.
4-76
-------
Based on an analysis of the performance achievable by control devices
examined for use, EPA decided to retain a control efficiency level of
95 percent for control devices used to comply with the standards. As an
alternative to demonstrating that this performance level is met, an owner or
operator can comply with the standards by maintaining temperature-residence
time requirements. The requirements in the proposal [1500°F (816°C) and
0.75 seconds residence time] have been retained for the final standards. By
meeting these operating requirements, a performance level well over the
requisite 95 percent is ensured. Other temperature-residence time combina-
tions may be used instead of these stated requirements if the owner or
operator demonstrates that the alternative combination achieves the
requisite 95 percent efficiency.
Other combustion systems, such as catalytic incinerators, are also
applicable to the control of small VOC streams. Systems which employ
catalysts, however, typically operate at lower temperatures and would not be
able to meet these operating requirements. Therefore, the temperature-
residence time requirements would not apply to combustion systems which
employ catalysts. Such systems would need to meet the required destruction
efficiency of 95 percent.
Comment:
Several commenters felt'that the choices for control devices were
restricted by the proposed standards. Two commenters cited costs of the
control device. One of the commenters (IV-D-23) suggested that EPA not
preclude the use of more cost effective devices that demonstrate equivalent
emissions reduction. The other commenter (IV-D-21) wrote that the installa-
tion of a combustion device or vapor recovery system for a limited number of
pumps is not cost effective and suggested that provisions for a variance be
made where these alternatives are unreasonable due to the small VOC volumes
involved.
Another commenter (IV-D-17) said the proposed standards exclude
transporting fugitive emissions to other equivalent control devices or
flares. This commenter also felt that §60.482(a)(7) was written on the
assumption that seal leakage from pumps in liquid service is vapor. The
4-77
-------
commenter wrote that the major weight fraction of the leakage will be liquid
and should be recovered in the process or disposed of safely. .The vapors
given off by these liquids can be ducted to destruction or capture devices.
The commenter asked that room be made in Section 60.482(a)(7) to capture and
treat VOC vapors by devices such as packed towers (water scrubbers). The
commenter also noted that the captured vapors may, in some cases, not be
recovered but treated as chemical waste in appropriate water treatment
facilities.
Response:
EPA recognizes that alternative control systems should not be precluded
from use where applicable. The equivalency provisions of §60.484 allows
other control techniques to be used if they are adequately demonstrated.
Any control technique allowed through equivalency must demonstrate at .least
95 percent efficiency in eliminating VOC emissions. This efficiency is
considered reasonable since it has been demonstrated by control technologies
expected to be used in complying with the standards.
The use of packed towers (water scrubbers) for treating fugitive VOC
emissions, therefore, would be allowed if equivalent control is adequately
demonstrated. But the applicability of this technology to VOC control is
expected to be limited by the solubility and vapor pressure of the VOC that
must be controlled.
Normal pump seal leakage is expected to be gas or vapor. The presence
of liquid seal leakage would probably indicate severe seal or barrier fluid
system failures. This liquid seal leakage should be .collected and disposed
of properly. .
The commenters pointed out that, for some applications, an enclosed
combustion system as required by the proposed standards may not be the most
cost-effective means of eliminating fugitive VOC emissions. As previously
discussed in this section, temperature-residence time requirements (816°C or
1500°F; 0.75 seconds) for enclosed combustion devices have been retained in
the final regulation as an alternative to ensure adequate destruction
efficiency for various combustion systems (incinerator, boiler, process
heater). But to provide the flexibility of using catalytic incinerators,
4-78
-------
these operating requirements are not mandatory for incinerators employing
catalysts. In addition, EPA has decided to allow the use of smokeless
flares operated under certain restrictions with a flame present as an
alternative control device. The costs of emissions control options were
presented in the AID.
Comment:
Another commenter (IV-D-17) wrote that, as presently drafted, catalytic
incinerators were initially precluded from use for control of fugitive
emissions from pumps unless their equivalency could be demonstrated. The
commenter felt this process would be time-consuming and costly, especially
since several companies have indicated that catalytic incinerators achieve
equal or better control than the two systems presently recognized in
§60.482(a)(3)(ii) and §60.482(a)(7).
Response:
As initially proposed, catalytic incineration units were inadvertently
precluded from use by the temperature requirements for enclosed combustion
devices. The specified temperature for incinerators (1500°F) was far above
the maximum recommended operating temperature for catalytic units (1000 to
1200°F). As previously discussed, operating requirements (temperature and
residence time) for combustion devices have been retained in the final rule
as one compliance alternative. The main requirement for control devices is
a demonstrated efficiency of 95 percent, which allows use of catalytic
incinerators without equivalency determination. Moreover, the operating
requirements will not apply to combustion units employing catalysts. This
change will permit the use of catalytic combustion units for control of
fugitive VOC emissions without an equivalency determination.
Comment:
Three comments (IV-D-17; IV-D-20; IV-D-50) were concerned with the
requirement that all control systems for fugitive VOC emissions be operated
at all times VOC emissions may occur. They felt that the provision would
necessitate shutdown of the pump or process unit whenever the control system
was being serviced or was experiencing an emergency outage. They noted that
the process emissions resulting from process unit shutdown and startup would
4-79
-------
result in greater VOC emissions than the fugitive loss during a short-term
outage of the control system. One cotmnenter (IV-D-20) wrote that this
requirement, specified in §60.482(a)(7), was not included in
§60.482(a)(3)(ii) and should be deleted. The other comment letter (IV-D-17)
recommended the following change to §60.482(a)(9) to allow a by-pass of the
control system when this system is out of service for maintenance or .during
emergencies and when .the net VOC emissions by-passed do not exceed emissions
resulting from process unit shutdown and startup:
A source, however, may bypass the applicable control device
set forth in §§60.482(a)(3)(ii) and (iii) where emissions from
. an associated shutdown and start-up of the process unit would
be greater than the emissions from the fugitive emission source(s).
Response:
The proposed standards require that the equipment installed to comply
with the standards be operated at all times VOC emissions may occur. For
example, this requirement is given in §§60.482(a)(7) and (9) for pumps
(redesignated §§60.482-2 and -10). Control systems designed to handle only
normal fugitive emission flow rates would probably be incapable of handling
emergency venting situations such as seal failure. This would be especially
true for enclosed combustion devices (incinerators) which have difficulty
handling streams of widely varying compositions, flows, and temperatures.
Thus, to protect the incinerator against damage due to excess loading,
additional safety mechanisms are expected to be used for emergency venting
situations. Such is not the situation where pressure relief devices are
connected to a closed vent system, e.g., vented to a flare.
An example of this kind of arrangement would be an incinerator used to
handle the normal flows expected connected to a seal drum that would release
excess emergency-venting flow (or pressure) to a plant flare system. A
system of this type would not be precluded from use under the standards.
The changes to §60.482(a)(9) [redesignated §60.482-10(f)] recommended
by the commenters present no methods for determining those situations where
process emissions from unit startup and shutdown are greater than the
associated fugitive emissions. Without such a method, a regulation allowing
4-80
-------
for those situations is impossible. However, the control device
requirements since proposal should allow a backup flare system- to be used,
so that outage or service shutdown of a control device should be no problem.
Comment;
One commenter (IV-D-6) wrote that an incinerator introduces a permanent
ignition source and requires sophisticated control for reliable operation.
Design and use of incinerators and vapor collection systems are also a
complication for relief valves because of the wide range of flow which must
be controlled. Further complexities are added in the need to minimize
oxygen in the flare header. The commenter added that providing adequate
combustion for widely varying flows entails complex design problems.
Response:
Design of an incineration system to handle properly the low and
variable flow situations encountered in fugitive emissions control is
difficult and has been addressed in other sections (4.1 Flares; 4.5 Safety
Considerations; 4.6 Pressure Relief Devices). These standards are
applicable to fugitive VOC emissions only and do not cover emergency vent
situations. And even though the range of flow rates covered by the
standards may be large, the final regulations allow more types of control
devices to be used, i.e., enclosed combustion devices (incinerators,
boilers, process heaters), vapor recovery systems (carbon adsorbers,
condensation units), and smokeless flares. Moreover, other systems may be
allowed if equivalency is demonstrated.
Comment:
One commenter (IV-D-16) stated that no supporting data were presented
to justify the need for the arbitrarily selected 95 percent control level
for control devices other than it is attainable by boiler furnaces,
incinerators, process heaters, and carbon adsorption units. This is not an
adequate basis for the definition of a control level, particularly when it
results in the exclusion of a practical and effective control option such as
flares.
4-81
-------
Response:
Based on the data presented in the August 22, 1980 EPA memo on "Thermal
Incinerators and Flares" (II-B-31), incinerator operating parameters of
1500°F (816°C) and 0.75 seconds residence time were selected as the basis of
the proposed standards. An evaluation of three incinerator control levels
was presented for an incinerator at similar conditions: 99 percent
efficiency, or less than 10 ppmv in the exhaust stream (99/10); 98/20; and
95/30. The 98/20 level was considered to be the highest achievable control
level for all new thermal incinerators, considering available technology,
costs, and energy use.
The temperature-residence time requirements for a thermal incinerator
to meet this control level, as required by the proposed standards, precluded
the use of alternate combustion devices such as catalytic incinerators.
These requirements, therefore, were dropped for combustion units employing
catalysts but were retained to reflect the requisite control efficiency. In
addition, EPA has decided to allow the use of smokeless flares operated
under certain restrictions with a flame present (see Section 4.1).
Other control options determined to be applicable to control of
fugitive VOC emissions include vapor recovery systems, such as carbon
adsorbers and condensation units. These systems have demonstrated
95 percent efficiency in removing VOC. Therefore, to allow use of vapor
recovery systems, the 95 percent efficient requirement was made for vapor
recovery systems. This efficiency requirement is also applied to any
control system considered under equivalency.
4.7 NO LEAK EQUIPMENT
Several comments were received concerning requirements of no leak
equipment. The comments, presented below, specifically address leakless
seal technology, canned pumps, and sealed bellows valves. A single response
is provided for all of these comments.
Comment:
One commenter (IV-D-6) pointed out that the performance of zero leak
equipment is cited as 100 percent when in fact this type of equipment
contains static seals which will not permit 100 percent control.
4-82
-------
Another commenter (IV-D-17) objected to the requirement In
Section 60.482(a)(8)(i) that externally actuated shafts penetrating the pump
housing cannot be considered leakless. This requirement was thought to
preclude the future development of leakless seal technology. The commenter
wrote that an individual operator or vendor should be allowed to demonstrate
compliance with the leakless pump provisions using future leakless seal .
technology and, therefore, this provision should be deleted in any revision
of the NSPS.
Comment:
The same commenter earlier (IV-D-17) had the following to say about
sealless pumps.
A number of pumps are marketed today which, fulfill the definition of a
sealless pump, and they all have a broad application to process fluid
handling problems. In most cases if the fluids being handled, the pump
capabilities and design limitations, and the process requirements are compa-
tible, then sealless pumps offer a good method for control of emissions.
Unfortunately, the limitations and requirements are difficult to match.
Canned pumps have limitations in that the motor cooling function must
be accomplished by the process fluid. For this reason, hot fluids and
fluids that will degrade on heating (e.g., monomers) are not good candidates
for application of canned pumps. Fluids that are flammable are also not
recommended for use with canned pumps because of possible failure of the
electrical insulation or power supply cord seal. The potential for a severe
safety problem does not justify their use. The pumping of slurries is also
not recommended because of the high potential for damage to the motor due to
cooling problems caused by poor fluid circulation. The use of canned pumps
in corrosive service may lead to high maintenance requirements and also to
high emission and exposure concerns. Even for canned pumps which seem to
handle clean, noncorrosive fluids, high emissions and/or exposures can
result. The cleanout of the pump prior to disassembly is nearly impossible.
Chemical industry experience with sealless pumps is rather limited and
in many instances unacceptable. Many pumps of standard vendor design are
incapable of handling potentially destructive factors such as dirt in the
4-83
-------
process stream, or precipitation of harmful chemical elements at high rotor
chamber temperature, or vacuum conditions. Canned pumps have been available
for many years, but due to their high initial and operating cost they are
seldom used. The largest pump currently available is 100 kW (135 hp) which
is too small to satisfy many of the chemical industry requirements. In
addition, all sealless pumps are, to various degrees, less efficient than
conventional pumps, and are limited by low differential head.
Potentially troublesome areas associated with practices currently
employed by vendors for design and fabrication of canned pumps are described
in the following paragraphs. These specific areas are: bearing lubrication
and design, motor rotor fabrication, stator liner material selection and
fabrication, and balancing of axial thrust.
a. Bearing lubrication and design. Carbon bearings are most commonly
used by the canned pump vendors. Although they are extremely susceptible to
dirt in the lubricating fluid, a clean environment is not always provided.
In applications where this problem is recognized, a thermal barrier
separating the pump end from the motor is utilized to provide a clean
operating condition for the carbon bearings. For lack of standards, "clean"
fluid is assumed to have solid particles no more than one percent by weight
and no longer than 10 mm.
b. Motor rotor fabrication. To protect the laminations and the core
against corrosion and erosion of the pumped fluid, the rotor is enclosed in
an airtight nonmagnetic can. In general, the rotor sleeve is in the
neighborhood of 0.38 mm (0.015 in.) and the end covers are 1.6 to 3.2 mm
(1/16 in. to 1/8 in.) thick, and are made of 316 stainless steel. To keep
the air gap to a minimum, the sleeve is shrunk onto the rotor. End covers
are welded on the shaft and sleeve by inert gas metal-arc welding. Prior to
the final weld closure, the rotor is preheated to about 340°C (650°F) for
about three hours to remove all traces of moisture. If moisture is not
removed, the can will bulge and rub in service.
c. Stator liner material selection and fabrication. The liner which
is normally 0.38 mm (0.015 in.) thick requires an extremely high quality
control in fabrication and fitting because it must withstand relatively high
4-84
-------
pump design pressures. In some designs, the liner is field-removable for
stator rewinding; in more efficient designs the liner is press-fitted in the
vendor shop to reduce the air gap.
The stator liner is fabricated from nonmagnetic materials, such as 316
stainless steel, Carpenter-20, Monel, or various grades of Hastelloy,
depending on the process fluid and the requirement of the users.
d. Balancing of axial thrust. The present method of balancing the
axial thrust in canned pumps is not satisfactory for the wide range of
process applications in the petrochemical industry. One of the existing
designs manufactured in the U.S.A. has a thrust bearing to absorb the axial
hydraulic forces. Some designs provide thrust collars with very limited
load capacity, but they are mainly intended to absorb momentary forces
generated during pump startup or shutdown. Consequently, the vendors are
relying on hydraulically balanced impellers and on flow control orifices to
position the rotor within the casting without producing an axial rub.
Correct function of the balancing system depends on the cleanliness of the
process fluid and maintenance of various clearances. This undesirable
feature excludes the use of canned pumps in services containing abrasive
particles.
Comment:
One commenter (IV-D-46) expressed concern that seal less pumps, which
are extremely effective, are not required because, according to the
proposal, they can be used only at a limited number of emission points. He
suggested that seal less pumps be required wherever they can be used.
The commenter also noted that sealless compressors are not required
because they are said not to be widely available. He said that if they were
required for future plants, production of sealless compressors would
increase and they would become more readily available. The commenter added
that this is precisely the forcing of technology that Congress has mandated
for new source performance standards.
Comment:
One commenter (IV-D-6) pointed out that although bellows valves have
been used in blocking valve service in the nuclear industry, they have not
4-85
-------
been applied widely in the chemical industry. The commenter stated that
concerns with corrosion and mechanical failure have yet to be resolved for
many of the chemicals listed in Appendix E (§60.489).
Another commenter (IV-D-17) wrote that of all the suggested approaches
to stopping fugitive losses, the use of the sealed bellows valves is by far
the most effective. The commenter said that EPA has estimated that about
75 percent of all fugitive losses occur through valve leakage and only a
small number of these valves (4/0 account for 70 percent of the total
fugitive emissions. Therefore, he concluded that the selected use of sealed
bellows valves could have a significant impact on the reduction of fugitive
losses.
The commenter continued by saying that sealed bellows valves are not
necessarily impermeable. If the bellows are elastomer or fiber-reinforced
elastomer materials, the bellows will allow permeation of vapors and will
therefore leak. If the bellows are metal, their durability is highly
questionable if the valve is operated frequently. When the bellows fail,
bellows valves will result in significant emissions. For this reason they
are not recommended for general service. They do offer good service in
critical areas which are compatible with their limitations.
The commenter added that these types of valves do cost considerably
more than conventional valves. In the smaller range (2 inches or less), the
bellows valves are twice the cost of conventional valves ($100 vs. $50). In
the larger size, the cost becomes somewhat more competitive, but is never
closer than about $2,000 vs. $1,000 for a 6-inch valve. Not only do they
cost more but they are not readily available in quantity (2 inch and below)
and not available at all above six inches.
He said that the use of diaphragm valves should be discouraged.
According to the commenter, it had been found that both temperature and
process liquids tend to damage or destroy the diaphragm in the valve. In
addition, operating pressures will reduce the application of this valve to
mostly pumping and product storage facilities. These valves can be obtained
with different bellows materials, but extreme care should be taken in the
selection of the proper material.
4-86
-------
Response:
EPA recognizes that leakless (no leak) equipment has limited applica-
tion in the SOCMI. Because of the limited applicability, leakless equipment
is not required by the regulations. But since leakless equipment demon-
strates 500 ppmv or less, leakless equipment is allowed under the standards
without equivalency determination. Furthermore, because of the equipment's
low leak potential, it is exempted from the monitoring requirements by
setting a performance standard of 500 ppmv or less with annual testing. It
is important, however, to note that leakless equipment represents but one
available control alternative under the standards. Other alternatives
include equipment standards (dual seals, capture/conveyance/control systems)
and work practice standards (such as leak detection and repair programs).
The provisions for leakless technology are presented in §60.482-2(e)
for pumps. The provisions permit only pumps whose shafts do not penetrate
the pump housing, i.e., sealless and canned pumps. These pumps are excluded
only if operated with emissions less than 500 ppmv above background as
measured by the methods set forth in the regulation. Similar provisions are
given in §60.482-3(i) for compressors and §60.482-7(f) for valves in gas/
vapor and light liquid service. Leakless seal technology is not precluded
by these provisions. Other types of leakless seal technology that may be
developed could become an accepted control alternative if adequately
demonstrated through the equivalency provisions of §60.484.
4.8 DUAL SEALS
Several comments were received regarding the^dual seal requirements
for pumps and compressors. The majority of these comments dealt with the
alledged stringency of uniformly requiring dual seals. Another large group
of comments discussed barrier fluid systems, specifically focusing on
potential chemical hazards and product contamination. Other areas of
concern regarding dual seals included applicability of dual seals to
reciprocating pumps and compressors, guidance on sensors to be used to
indicate seal failure, and inconsistencies in wording of the proposed
regulation.
4-87
-------
Comment:
One commenter (IV-D-25) wrote that the requirement for dual mechanical
seal systems with barrier fluids for pumps and barrier fluid systems for
compressors appears to be unnecessarily stringent. Adequate control of
emissions should be possible by requiring monthly monitoring of less
elaborate sealing systems.
Another commenter (IV-D-17) also felt it inappropriate to require use
of dual mechanical seals for pumps. This commenter was joined by another
commenter (IV-D-26) in recommending that the requirement of dual mechanical
seals for pumps be revised to permit the use of any properly designed dual
seal, thus allowing other dual seal arrangements (e.g., tandem seals). One
of these commenters (IV-D-17) stated that this change would allow the use of
either pressure or level control devices to detect an inner seal failure.
This commenter (IV-D-17) stated that the general belief that a dual
mechanical seal is better than a single seal is not always true. Depending
on the specific requirements of the service, seal design becomes very
complicated and costly and the efficiency cannot be generalized. The
commenter disagreed with the statement in the proposal that dual mechanical
seals will be 100% effective in controlling fugitive emissions. He further
wrote that in the draft final "The Assessment of Environmental Emissions
From Oil Refineries," February, 1980, Appendix B, no data is presented that
would indicate any statistically significant difference for percent of leaks
on packed vs. mechanical seals or single vs. dual seals. Thus, these data
would not support the EPA position. The commenter suggested that, before
such a significant investment in dual seals is required on such a question-
able basis, a detailed technical review should be conducted to establish the
data base for fugitive emission controls for pumps and compressors.
Another commenter (IV-D-15) wrote that in its evaluation of alternative
control technologies for pumps, EPA had not compared and contrasted the
relative merits of single mechanical seals versus dual mechanical seals. In
the BID, data provided in Table 3-1 show emissions from single and double
sealed pumps to be identical. The commenter, however, noted that the BID
does not mention the condition under which the pumps surveyed were operating
or whether a barrier fluid system was employed.
4-88
-------
The commenter stated that although they are not commonly employed,
single mechanical seals can be provided with a barrier fluid system and
auxiliary stuffing box to reduce and even eliminate fugitive emissions from
the pump seal. A barrier fluid system operating at a pressure higher than
the process liquid would control emissions at the 100* efficiency level.
The commenter concluded that since single seals can be more economical to
install and maintain, EPA should have included them in its alternative
control scenarios and provided justification for the choice of dual seals
over single seals.
Another commenter (IV-D-51) agreed, citing a test that was conducted on
over 20 pumps equipped with single mechanical seals that were in light
liquid service in an acrylic acid process. The results showed that single
mechanical seals provided effective control of fugitive emissions using the
10,000 ppm leak definition used for valves in light liquid service. The
commenter recommended that a leak definition of 10,000 ppmv be established
for pumps in light liquid service by adding a section to §60.482 that
permits an inspection and maintenance plan using this definition. He also
wrote that the inspection requirement should take into account a skip-period
plan for pumps demonstrated by six successive measurements showing no leak.
The commenter stated that semiannual leak measurement of qualified pumps is
adequate to insure effective emissions control.
Response:
The proposed standards required that pumps be controlled with dual
mechanical seal systems which included non-VOC barrier fluids and closed
vents to control devices. If this equipment could not be installed, the
owner or operator had the option of enclosing and venting the seal area to a
control device. The control techniques involved were found to be techni-
cally feasible and the costs to control the model units (including control
cost for pumps) were found reasonable.
Since proposal, the costs to control each fugitive emission source have
been scrutinized more closely on an individual basis in the AID (III-B-2).
In looking at the cost effectiveness for each fugitive emission source
(Chapter 3), EPA determined that the costs associated with the equipment
4-89
-------
required for pumps may be unreasonable in some instances when compared to a
leak detection and repair program. Based on this determination, the less
stringent option of a leak detection and repair program has been added as an
alternate standard for pumps. Leak detection and repair programs for pumps
were discussed in Section 4 of the AID.
The final regulation requires a monthly leak detection and repair
program using a 10,000 ppmv leak definition. No skip-period monitoring is
used in the final regulation for pumps since the number of pumps that must
be monitored is not large enough to justify such a program. As alternatives
to the work practice standard, three choices remain: (1) installation of a
properly designed dual mechanical seal system with an associated barrier
fluid system as specified in §60.482-2(d), (2) installation of an enclosed
capture/conveyance/control system as described in §60.482-2(f), and (3) use
of leakless equipment as provided in §60.482-2(e).
However, some pumps in the 24-Unit Study did employ dual mechanical
seals with barrier fluid systems. EPA believes if a pump is equipped with a
dual mechanical seal/barrier fluid system that is operating properly with a
non-VOC or heavy liquid VOC barrier fluid, emissions will be reduced.
The data collected during SOCMI screening studies on pump seal leak
frequency are inconclusive with regard to seal type. No statistical
differences were seen for seal type (single mechanical vs. double mecha-
nical), either on-line or off-line. Off-line pumps, in general, leaked at
about one-third the frequency of on-line pumps. The presence or type of
barrier fluid associated with those seal systems screened was not recorded
for all pumps during the pump seal screening studies conducted for the SOCMI
and Petroleum Refinery Fugitive VOC emission programs. Thus, these
inconclusive results should be considered carefully. For instance, if a VOC
barrier fluid were used on a double seal, the seal would effectively be a
single seal, and classified improperly for leak frequency evaluation. In
addition, no significant effects of temperature and pressure could be seen
from the screening results.
In some instances, a dual mechanical seal/barrier fluid system may be
required to reduce emissions. For pumps that cannot be successfully
4-90
-------
repaired by routine maintenance, a monthly leak detection and repair program
is ineffective in reducing emissions. Furthermore, since these pumps would
be leaking on an almost continuous basis, the cost effectiveness of leak
detection and repair is unreasonable ($3,000/Mg), while the cost effective-
ness of using the required equipment is reasonable ($26Q/Mg - $430/Mg).
Thus, a provision has been added to the standards for pumps that requires
such pumps that cannot be repaired by repeated maintenance efforts to use
the prescribed equipment to reduce emissions. A delay of repair provision
was- also added to permit an owner or operator the time (six months) to
install the equipment on such pumps.
Comment:
Several comments were received regarding the barrier fluid requirements
for dual seal systems. One commenter (IV-D-17) wrote that barrier fluid
systems are not commercially available for many pumps and compressors, on the
market, especially for smaller sizes. In a previous letter, the commenter -
(IV-D-17) pointed out differences of understanding of requirements for a
pressure reservoir sealing system. EPA did not include in its cost
estimates: (1) a flushing oil pump/spare requirement, (2) strainers which
are a critical component to ensure that foreign matter does not destroy the
seals, and (3) instrumentation even though regulations clearly require such
installations.
This commenter also stated that the requirement of a non-VOC barrier
fluid is not technically feasible for many process applications within
SOCMI. He cited, as an example, synthetic alcohol and alkylation processes
(methyl ethyl ketone). Nonvolatile barrier fluids leaking across the seals-
would be converted to coke/tar solids resulting in possible seal damage and
an acid sludge, thus creating hazardous waste disposal problems. Another
commenter (IV-D-6) agreed, noting that dual seals with barrier fluid systems
are not suitable for monomers such as acrylonitrile. Polymerization on the
hot seal faces will cause failure, excess emissions, and increased
maintenance.
One commenter (IV-D-41) wrote that the requirements for dual mechanical
seal systems with barrier fluids for pumps and barrier fluid systems for
4-91
-------
compressors appear to be unnecessarily stringent. In addition, the
commenter maintained that the requirement that a heavy fluid be utilized as
the barrier fluid in dual mechanical seal systems seems excessive.
- Several commenters (IV-D-7; IV-D-15; IV-D-17; IV-D-21; IV-D-26;
IV-D-34; IV-D-37; IV-D-50) expressed concern over potential product
contamination by leakage of barrier fluid into the pumped fluid. One
commenter (IV-D-21) suggested a provision be made for a variance from the
barrier fluid requirement where product requirements preclude use of a
barrier fluid. Another commenter (IV-D-15) suggested that an equitable
exemption from §60.482(a)(3)(i) be offered. The degassing reservoir -
enclosed combustion device exemption given in §60.482(a)(7) was thought to
be prohibitively expensive when there are only a few process pumps which
must meet tight product purity specifications.
A commenter (IV-D-17) added that the use of non-VOC barrier fluids
across the board, pursuant to EPA's requirements, is not possible. He
stated that, since §60.482(a)(3)(ii) and (iii) require the barrier fluid
reservoir to be connected to a closed vent system to recover or destroy VOC
or to be purged back to the process, it seemed immaterial what barrier fluid
was used since it would be controlled.
. One of the commenters (IV-D-37) said process fluids, which are often
light liquids, are generally used as barrier fluids to avoid product
contamination or undesirable side effects. Although alternatives may be
possible, they are not, in general, cost effective and product quality could
suffer as a result. This commenter also felt that allowing more leeway with
the "light liquid" requirement for barrier fluids would be a good applica-
tion of the "bubble" philosphy, allowing industry to choose the degree of
individual control to achieve an overall emission level while maintaining
cost effectiveness. Another commenter (IV-D-7) wrote that unless it were
feasible to allow the liquid to leak across the inner seal into the material
being pumped, the barrier fluid itself could become contaminated to the
point where it would be a light liquid and, thus, require replacement. The
contaminated fluid would have to be moved somehow (presumably in another
dual seal pump) to a process for cleanup or destruction.
4-92
-------
Response:
As described in previous responses in this section, the standards for
light liquid pumps have been revised since proposal. Under the proposed
standards, fugitive VOC emissions would have been eliminated by installing
equipment. This equipment was composed of a dual seal system with its
associated non-VOC barrier fluid system vented to a suitable control device.
The costs of installing such a system are discussed in Section 5 of the AID.
As an alternative for pumps which could not use mechanical seals, the seal
area was to be enclosed and vented to a suitable control device.
The final regulation, in addition to the equipment alternatives previ-
ously specified, allows the use of a work practice standard for pumps. This:
standard calls for a monthly leak detection and repair program with a leak
definition of 10,000 ppmv, as determined by the leak detection monitoring ';
specified in §60.482-2. The addition of the work practice standard for
pumps allows the owner or operator to choose the best means of controlling
emissions from pumps at his process unit. At higher leak frequencies, a
program incorporating both alternatives could prove to be the most cost
effective for any given process unit.
Also, for any given process unit, an owner or operator may choose to
use equipment for some pumps that are chronic leakers and apply the work
practice standard to the remaining pumps. If he chooses to use the
equipment on a pump, he would not have to include that pump in a monthly
leak detection and repair program.
Barrier fluid systems for the seals required under the proposed
standards and allowed under the final standards can be of varying designs. -
The specific design of each system would be dictated by the particular
application under consideration (pump/seal system). API Standard 610
(IV-M-21) presents several barrier system designs for various seal types.
For smaller seal flow applications, an individual pressurized barrier fluid
tank could be used instead of a pump/recirculation loop arrangement.
Some barrier fluid tanks may be as small as two or three gallons;
replacement of barrier fluid contaminated by pumpage leaking across a failed
inner seal would not present a major disposal problem. In addition, a
4-93
-------
properly designed barrier system would provide indicators (sensors) to
indicate a seal failure so that severe contamination of the barrier fluid
would probably be avoided. The type indicator is not specified, but, in
most .cases, it would be a pressure or level sensor.
The costs for double mechanical seal systems used in the BID included
the double mechanical seal, pressurized barrier fluid system, and a barrier
fluid cooler. These costs are considered adequate for the required seal
applications and were not revised in the AID.
In the case of possible product contamination by barrier fluid, each
case should be examined separately. There are too numerous product/barrier
fluid combinations possible in the SOCMI to warrant examination of every
combination. Specific compatibility determinations are expected to be made
by the owner/operator on a case-by-case basis.
A non-VOC barrier fluid is required to comply with the dual seal
equipment standard for pumps if an owner or operator chooses to install
equipment in lieu of monitoring. If a VOC barrier fluid were used instead,
the dual seal would effectively be only a single seal, in terms of sealing
VOC from the atmosphere. Failure of the outer seal would result in VOC
emissions.
Furthermore, the use of an inert material such as water is allowed
under the standards. One commenter noted that dual seals could not be used
for monomers such as acrylonitrile due to problems with polymerization on .
seal faces. Two acrylonitrile units were surveyed in the 24-Unit Study
(IV-A-11); dual seals were being used successfully in one of these process
units.
At proposal, EPA believed that barrier fluids for dual mechanical
seal/barrier fluid systems were available that would effectively reduce VOC
emissions from pump seals. Such barrier fluids would include heavy final
products, heavy oils, water, etc. But comments received on the proposed
standards pointed to a few limitations of barrier fluids. These comments
were considered and the data from fugitive emissions studies were reviewed.
As a result, EPA still maintains that barrier fluids are available that will
4-94
-------
serve the purpose of reducing emissions from pump seals when the fluid is
judiciously selected and the system is properly operated.
The barrier fluid requirements for compressor seal barrier systems are
not necessarily heavy liquids. The requirements prohibit the use of light
liquid or gaseous VOC as barrier fluids. This requirement would not,
however, prohibit the use of an inert material such as nitrogen or steam as
the barrier fluid for such control systems.
Comment: .
One commenter (IV-D-17) noted that the definition of "fugitive emission
source" includes compressors and pumps. The commenter stated that, unfortu-
nately, EPA has not included a definition as to the types of pumps and com-
pressors not subject to the fugitive emissions standards. 'As a result of
this oversight, reciprocating pumps and compressors are included in the :
regulatory program. According to the commenter, the potential problem with
the inclusion of reciprocating pumps and compressors is their inability to
accommodate dual seals, or any type of seal that EPA may ultimately require:
He recommended that, in order to avoid sealing problems associated with
reciprocating pumps and compressors, these equipment be expressly excluded
from the definition of "fugitive emission sources."
Response:
Reciprocating pumps and compressors do present a more complex fugitive
emissions control problem than centrifugal and rotary designs. Recipro-
cating equipment typically requires packed stuffing boxes for sealing, since
mechanical seals are generally unusable. Under the proposed standards for
SOCMI fugitive VOC emissions, reciprocating equipment and other equipment
that cannot use mechanical seals would be required to have an enclosed seal :
area, with the captured gases vented to a suitable control system (smokeless
flare, enclosed combustion device or vapor recovery system).
As the standards have been finalized, reciprocating pumps could comply
with a monthly leak detection and repair program or the seal areas could be
enclosed and vented. API standards covering the design of reciprocating
equipment require that the distance piece (the piece separating the cylinder
and the driver) be enclosed, sealed, and vented (IV-M-10). Since this piece
4-95
-------
is vented, it is technically feasible to vent the seal area of new recipro-
cating equipment. In a meeting with EPA, one industry group stated that for
new equipment, venting is not only possible, but is general practice for
some of its member companies. No specific venting arrangement has been
adopted as a standard by their industry (IV-E-12).
Problems were noted, however, by this group for older existing recipro-
cating compressors. Older compressors may not have enclosed and vented
distance pieces, making retrofitting difficult and expensive. The cost for
compliance was estimated to be a new compressor in some cases. The number
of reciprocating compressors that would be affected under the proposed
standards is unknown. During the screening studies conducted in SOCMI
process units, only three compressors with packed seals (assumed to be
reciprocating compressors) were encountered out of the 22 compressors in VOC
service. The impact of the equipment requirements for new reciprocating
compressors is expected to be small since the compressors are built to
accomodate venting systems. However, to alleviate potential problems
associated with older existing reciprocating compressors, a restricted
exemption from the equipment requirements for compressors has been made for
existing compressors that come under the standards through the modification
provisions. This is discussed in detail later in Section 4.12 on
Reciprocating Pumps and Compressors.
Comment:
One commenter (IV-D-26) noted that the type of sensor required on the
barrier fluid reservoir was not identified. He recommended that some
guidance be given, since a pressure sensor might not be applicable when
using a barrier fluid reservoir that is vented by a closed system to a
combustion or recovery device.
Another commenter (IV-D-17) also indicated that different sensors would
be needed to indicate seal failure for the different seal arrangements. The
commenter cited the differences between two dual seal systems (double mecha-
nical and tandem seals) and indicated that both systems should be allowed
under the proposed standards. The double mechanical seal could require a
4-96
-------
pressure sensor to detect an inner seal failure, but a level control alarm
would be better suited for a tandem seal.
Response:
As a requirement of Section lll(h) of the Clean Air Act, provisions
must be made to ensure the proper operation and maintenance of control
systems required by equipment standards. If an owner or operator chose to
install equipment rather than comply with a work practice standard for
pumps, outer seal failure would be noted through periodic visual inspec-
tions. But some form of indicator would be needed to indicate catastrophic
failure of an inner seal of a dual seal arrangement. The choice of the type
of sensor to be used would necessarily be based on engineering and design
considerations and is, therefore, not specified in the regulation. By
allowing any type of sensor, the owner or operator is afforded the
flexibility of choosing one best suited to his situation.
As an example, consider a double mechanical seal, where the seal faces
are mounted back to back forming a cavity in between. In this case, the
barrier fluid flushed through the cavity is generally maintained at a pres-
sure greater than the pump stuffing box pressure. With a failure of the
inner seal, the barrier fluid would be flushed into the pumped fluid.
Although no fugitive VOC emissions result from the inner seal failure,
continued operation could result in total seal failure. And loss of barrier
fluid would then result in VOC emissions. The initial loss of barrier fluid
could be indicated by a low level alarm on the barrier fluid tank, thereby
avoiding total seal failure and potential VOC emissions. A low pressure
alarm may also be applicable to this system.
Another example is a tandem seal arrangement, where both seal faces are
aligned in the same direction. For tandem seals, the pump stuffing box
pressure is generally higher than the barrier fluid pressure. Therefore,
failure of the inner seal would be more readily indicated by an increase in
barrier system pressure, avoiding potential VOC emissions.
4-97
-------
Comment:
One commenter (IV-D-26) pointed out that the words "stuffing box
pressure" should be followed by the word "or" in the proposed Section
60.482(a)(3)(i).
Response:
The three provisions in the proposed §60.482(a)(3) for pumps and in the
proposed §60.482(b)(3) for compressors are in series, using the conjunction
"or" between the second and third items to indicate that only one of the
three items (i, ii, or iii) would apply. Since this construction results in
some confusion, the addition of "or" has been made between the first and
second items to clarify the intent of exclusive alternatives. With the
redesignation of section numbers for the regulation^ these exclusive
alternatives are found in §60.482-2(d) for pumps and in §60.482-3(b) for
compressors.
Comment:
Another commenter (IV-D-17) pointed out one inconsistency in the
compressor standards which should be clarified. Sections 60.482(b)(l) and
(3) only require a seal system with certain characteristics to be used for
all compressors. However, Section 60.482(b)(4) reverts to referring to dual
mechanical seal systems.
Response:
The most commonly applied seals for compressors in SOCMI are labyrinth
seals. Although the basic labyrinth seal and its more effective variations
(staggered labyrinth, honeycomb labyrinth, rotating labyrinth) are
considered mechanical seals, they are not dual seal systems, as described
for pump applications. Therefore, this portion of §60.482(b)(4) [redesig-
nated §60.482-3(d)] is being revised to read "seal system," not "dual
mechanical seal system."
4.9 SAMPLING SYSTEMS
The comments received regarding requirements for sampling systems dealt
primarily with four concerns: (1) emissions from closed-loop sampling
systems, (2) the cost effectiveness of closed-loop sampling systems,
4-98
-------
(3) potential safety risks of closed-loop sampling, and (4) alternatives to
closed-loop sampling systems.
Comment:
One commenter (IV-D-26) wrote that the phrase "without VOC emissions to
the atmosphere" in Section 60.482(d)(2) would preclude the removal of a
sampling vessel from a closed purge system, since some VOC emissions would
occur when the fittings are disconnected.
Another commenter (IV-D-17) also expressed the concern that providing
the option of returning the fluid directly to the process line or collecting
the purged process fluid for recycle or disposal without VOC emissions to
the atmosphere amounts to a zero emissions requirement. In a previous
letter, the commenter (IV-D-17) stated that closed-loop sampling, where the
sample container is part of the flow path, is not an emission free system.
The coupling points around the sample container will retain liquid when the
container is isolated. Special equipment, vents and vacuum systems can be
installed to remove this liquid, but these precautions only, minimize
exposure, not emissions.
Response:
In the Background Information Document, it is assumed that closed-purge
sampling systems are approximately 100 percent effective in eliminating
sampling purge emissions. But, as noted in the preamble of the proposed
standards (46 FR 1145, January 5, 1981), some VOC could be emitted during
sample transfer to a closed collection device. This was the reason a "no
detectable emissions" or zero emissions limit was not considered feasible
for sampling systems.
The potential for a small amount of VOC to be emitted during sampling
procedures is recognized by EPA. The intent of the standard is to eliminate
sample line purging to the atmosphere, ground, or sewer drain. A zero
emissions limit is not intended, as noted by the establishment of an equip-
ment standard instead of an emissions limit. To clarify this intent in the
regulation, §60.482(d)(2) [redesignated §60.482-5(b)] has been revised to
read:
(b) Each closed purge system,as required in §60.482-5(a) shall:
4-99
-------
(1) Return the purged process fluid directly to the process line
without VOC emissions to the atmosphere, or
(2) Collect the purged process fluid for recycle or disposal without
VOC emissions to the atmosphere; or
(3) Be designed and operated to capture and transport all the
purged process fluid to a control device complying with the
requirements of §60.482-10.
Comment:
One commenter (IV-D-6) pointed out that flow-through sampling systems
may not be the most cost effective sampling procedure for all liquids.
Another commenter (IV-D-13), however, noted that data show closed-loop
sampling systems to be very cost effective and suggested that Alternative II
be modified to include closed-loop sampling systems as the most appropriate
control option.
Response:
Based on the costs estimated for carbon steel sampling systems
presented in the BID, the cost effectiveness of closed purge sampling
systems is approximately $890/Mg VOC. This value assumes 100 percent
efficiency of VOC control using this sampling system. This cost effec-
tiveness could be higher if stainless steel materials are necessary. The
overall economic impact to SOCMI is expected to be overstated, however,
since industry has stated that 75 percent of SOCMI sampling systems use some
comparable sampling system currently (II-E-20). The costs of sampling
systems are discussed in Section 5 of the AID.
The standards for fugitive VOC emissions from SOCMI sources are based
or Alternative IV. The effectiveness of closed-purge sampling systems is
adequately demonstrated by comparing reductions achieved in Regulatory
Alternatives III and IV. Also, the costs of control were considered on a
source by source basis in the AID. Therefore, there is no need to modify
Alternative II to include closed-purge sampling systems.
Comment:
Three comments were received on the safety aspects of closed-loop
sampling systems. One commenter (IV-D-6) said two concerns for liquid
4-100
-------
sampling systems were loss of liquid between valves and pressure at the
valves. Another commenter (IV-D-17) noted a serious safety risk with
closed-loop sampling by collecting "bomb" samples. He cited an example of a
liquid-full sample container that ruptured due to volumetric expansion as
the temperature increased. Another commenter (IV-D-32) said that a double
valve sampling system cannot be used for materials in traced lines since, if
both valves were closed, pressure increases would rupture the pipe.
Response:
The proposed BID recognizes that closed-loop sampling systems have
limitations with respect to low pressure processes or tankage and, in some
instances, safety requirements. The regulation, therefore, does not specify
a "closed-loop sampling system," but does require a "closed purge system."
In some cases, such as low pressure systems, a closed-loop sampling system
would not permit sample collection. Under these circumstances, the sample
and purge would be collected in containers separate from the process piping.
Thus, a "closed purge system" would allow any system that collects all VOC
purged during sampling and recycles or destroys the collected VOC in a
control device. Sampling purge material is not to be discarded in the plant
drain systems. Closed-loop sampling systems are used in the BID to evaluate
the feasibility of controlling fugitive emissions from sampling systems.
Comment:
Another commenter (IV-D-17) noted that the technical support document
does not support the proposed standard for sampling systems. He recommended
the use of closed vent systems be included to minimize emissions from
sampling systems by revising §60.482(d)(2) as follows:
(2) Each closed purged system as required by §60.482(d)(l)
shall return the purged process fluid directly to the process
line, or shall collect the purged process fluid for recycle or
disposal by means of a closed vent system.
In a previous letter, the commenter (IV-D-17) recommended a better
system for use where line pressure drop is available. Partially evacuated
sample containers are connected, and a sample of the flowing stream sucked
into the sample container. Use of a partially evacuated sample container
4-101
-------
prevents the risk of "liquid filling." The commenter further stated that
the most effective means of minimizing emissions and exposures is by
sampling only the quantity of material needed. He referred to the descrip-
tion of various sampling systems presented in papers by Bruce C. Lovelace.
According to the -commenter, one commercial sampler allows sampling only the
quantity desired and has no vented emissions. The vapor and gas displaced
by the liquid are pumped back into the process.
Another commenter (IV-D-6) suggested that a better method would be the
collection of the desired volume of liquid from a sampler that reduces
sample pressure to atmospheric as it is collected. He expressed concern
that under the proposed regulation these techniques would not be permitted.
With a performance standard they could be used without extensive technical
discussion,, the commenter noted. He further recommended defining "closed
purge systems" and "in-situ sampling systems."
Response:
Some of the concerns expressed by the commenters stem from the require-
ment that sampling be conducted "without VOC emissions to the atmosphere."
The intent of the standard for sampling systems is not a zero emissions
requirement. The intent is to minimize VOC emissions during sampling
through improved system design practices. The systems described by Lovelace
certainly provide safety for the individual collecting the samples and also
lend themselves easily to the design practices intended by the proposed
standards for minimizing VOC emissions.
The intent for sampling systems has been clarified in the final
standard with the changes presented in response to other comments. By
adding definitions for "closed purge systems" and "in-situ sampling
systems," the proper equipment design criteria for sampling systems have
been better described.
4.10 CLOSED VENT SYSTEM
Comment:
One commenter (IV-D-34) felt .that the definition of "closed vent
system" shoul.d not preclude the use of a flare as an acceptable control
alternative. He gave as reasons for including flares previous experience
4-102
-------
with flare systems, as well as the inefficient, costly, and unsafe operation
of closed incinerator systems. -
Another commenter (IV-D-17) suggested that the definition of closed
vent system be modified as follows:
'Closed Vent System1 means a system which is not open to
the atmosphere and which is composed of piping, connec-
tions, and, if necessary, flow inducing devices that
transport gas or vapor from a fugitive emission source to
an enclosed combustion device, vapor recovery system, a
flare system, or an equivalent control device as
determined by §60.484.
Response:
As discussed in Section 4.1, the final regulation allows elevated
smokeless flares for control of fugitive VOC emissions from SOCMI if the
flare is operated under certain restrictions with a smokeless flame present.
Thus, some modification to the proposed definition of "closed vent system"
has been made to reflect this decision. Another change has been made to
incorporate a reference to equivalent control devices. Since some
equivalency determinations made through provisions in §60.484 will be
applicable to the industry as a whole, it is also appropriate to include
some reference to these equivalency determinations in the definition of
"closed vent system."
4.11 OPEN-ENDED LINES
Comment:
Two comments were received on the requirements for open-ended lines.
One commenter (IV-D-12) wrote that installation of double valves, which are
required in lieu of caps, plugs, or blinds, would present severe safety
complications in a major furfural step at his facility. Another commenter
(IV-D-17) stated that many processes use block-and-bleed techniques to avoid
process contamination or where explosive or reactive mixtures are present.
Two examples of such conditions were cited: (1) water removal systems using
hot and cold dry gases and (2) fuel gas supply lines to combustion units.
4-103
-------
To exempt these bleed (vent) valves from the standards, the commenter
recommended the following change to §60.482(e)(2):
(2). . . .through the open-ended valves, except where
the open-ended valve or where the valve is vital to
process safety or contamination. -
Response:
With regard to the safety concerns of double valves in furfural produc-
tion units, the commenter provided additional clarification (IV-B-8;
IV-E-4). His concerns were the large discharge valves used to dump waste
solids after completion of the batch reaction cycle. As the commenter
described them, these valves are more like manhole covers than open-ended
valves. In general, manhole covers and flanges, including blind flanges,
are,not regulated by. these standards. Flanges are subject only to < '
§60.482-8.
Currently, standard engineering codes do not cover open-ended lines and
double valve installations, but there are instances where safety require-
ments might impact them. Provisions have been made in the standards for the
block-and-bleed techniques described in the second comment.
Section 60.482-6 requires:
(a)(l) Each open-ended valve or line shall be equipped with a cap,
blind flange, plug, or a second valve.
(2) The cap, blind flange, plug, or second valve shall seal the
open end at all times except during operations requiring
process fluid flow through the open-ended valve or line.
(b) Each open-ended valve or line equipped with a second valve shall
be operated in a manner such that the valve on the process fluid
end is closed before the second valve is closed.
^Where a block-and-bleed system is being used, the bleed valve (second valve)
must "seal the open end at all times except during operations requiring
process fluid flow through the open-ended line." The bleed valve,
therefore, can remain open when venting the space between the two block
valves.
4-104
-------
4.12 RECIPROCATING PUMPS AND .COMPRESSORS
Comment:
One comment letter (IV-D-17) noted that the definition of fugitive
emission source includes compressors and pumps. It was viewed as
unfortunate that the Agency had not included a definition of the types of .
pumps and compressors not subject to the proposed requirements. As a
result, the comment, letter continued, reciprocating pumps and compressors
are included in the regulatory program. He was joined by another commenter
(IV-D-34) in stating that their inclusion was seen as a problem because of.
their inability to accommodate double seals or any type of seal that EPA
may ultimately require. Both commenters recommended expressly excluding
reciprocating pumps and compressors from the definition of "fugitive
emission source." In a later letter (IV-D-50), the first commenter
expressed concern that the proposed standards were not cost effective for a
variety of reasons including coverage of reciprocating pumps and compressors
that are unable to accommodate the required controls.
Response:
Reciprocating equipment was intentionally included under coverage of
this regulatory program. It should be noted that reciprocating equipment is
not as common within this industry as within petroleum refineries and
natural gas plants. In fact, surveys of SOCMI process units indicate that
reciprocating pumps comprise between 5 and 9 percent of the total pump
population in SOCMI and reciprocating compressors represent from 6 to
16 percent of the compressor population. The actual reciprocating
compressor count, however, is small since the number of compressors (in
comparison to pumps) in SOCMI is relatively small.
Because reciprocating equipment does exist in some units and may even
be necessary in some applications, a provision was made in the proposed and
final standards to allow enclosure of the seal area and venting of any
emissions to a suitable control system (see Sections 4.1 Flares, 4.6
Combustion Device, and 4.11 Closed Vent Systems). This option is feasible
for reciprocating pumps, as well as for most reciprocating compressors. New
reciprocating equipment is provided with enclosed and vented seal areas, as
4-105
-------
required by API standards. This requirement was initiated by OSHA
regulations governing allowable hydrogen sulfide levels in the workplace (an
important consideration in petroleum refineries and gas plants). API
representatives and compressor manufacturers have noted that, for new
facilities, the seal area of reciprocating equipment can be vented. One
manufacturer stated that venting the seal area to a flare is currently his
company's practice for new reciprocating equipment ("IV-B-9; IV-B-10;
IV-E-12; IV-M-10.)
But these representatives further noted that there are technical
problems associated with retrofitting a seal vent system to some existing
reciprocating compressors. On older compressors, the distance piece between
the cylinder and driver may not be enclosed and vented. In such cases,
retrofitting a vent system to the compressor in order to comply with the
standards could require recasting of the distance piece or even replacement
of the compressor. The costs of new reciprocating compressors were deter-
mined to be excessive (up to $210,000 for a 2-stage 3000 CFM unit),
especially in the light of the emission reductions that could be gained for
reciprocating compressors. Therefore, EPA decided to examine two alterna-
tives for reciprocating compressors: (1) exemption of some reciprocating
compressors from modification requirements and (2) designation of
compressors as a separate affected facility.
An exemption from modification provisions was considered for those
existing reciprocating compressors whose distance pieces could not
accommodate control equipment. Providing an exemption which is narrowly
limited to such cases would mean that any new reciprocating compressors in
units affected by the standards would be covered. But relief would be
provided for those units affected by virtue of modification provisions which
would encounter severe retrofit difficulties. One 'disadvantage to this
approach is that it would require enforcement personnel to exercise some
judgement concerning whether a reciprocating compressor meets the criteria
for the exemption.
Since the compressor is a major piece of equipment and represents the
major portion of the capital costs of the facilities considered under these
4-106
-------
standards, compressors could be designated as a separate affected facility.
In doing so, some reciprocating compressors would probably be eliminated
from coverage under the modification provisions. Providing an exemption for
specific circumstances, however, also accomodates the problem of the high
cost of controlling these specific compressors. Providing an exemption does
not needlessly change the definition of the affected facility given at
proposal.
Because of the necessity to eliminate coverage of existing compressors
which cannot be retrofitted without replacement of the distance piece or
compressor, EPA in the final regulation, has provided an exemption for
reciprocating compressors which cannot accommodate vents within facilities
which have become affected by virtue of modification. The exemption applies
only to those specific instances where the seal area cannot be enclosed and
vented without recasting the distance piece or replacing the compressor.
Furthermore, at such time that the distance piece or compressor is replaced,
the compressor will no longer be exempted from the standards.
4-107
-------
-------
5. APPLICABILITY OF STANDARDS
This section contains comments and responses concerning which fugitive
emission sources, process units, and plants should be covered by the
standards. Comments concerning applicability of the standards have been
divided into ten subsections:
5.1 SOCMI List
5.2 Vapor Pressure Cutoff
5.3 Percent VOC Cutoff
5.4 Process Unit Definition
5.5 Small Manufacturers
5.6 VOC Definition
5.7 Pilot Plant and R&D Facilities
5.8 Flanges
5.9 Vacuum Service
5.10 Enclosed Buildings
5.1 SOCMI LIST
Comment:
Several comments were received which challenged the basis for the list
of chemicals proposed as the definition of SOCMI in proposed Appendix E
(promulgated §60.489). The comments focused on photochemical reactivity of
listed chemicals, their degree of toxicity, and the applicability of a unit
operations approach for regulating a wide variety of chemical processes-
Some commenters said that the list should logically have been compiled
based on photochemical reactivity and potential to form ozone. They saw no
evidence of this basis in the list (IV-D-17; IY-D-28). Another (IV-D-34)
said that any control requirements for VOC should not include nonreacttve
hydrocarbons. A few chemicals were noted as being universally considered
photochemically unreactive even by EPA and were, therefore, inappropriately
placed on the list: methylene chloride (IV-D-26); chlorofluorocarbons
(IV-D-26; IV-D-15); 1,1,1-trichloroethane (IV-D-26); methanol (IV-D-26;
5-1
-------
IV-D-17); acetone (IV-D-26; IV-D-17); and acetylene (IV-D-26). Federal
Register notices were cited in support of removing these chemicals from the
list: 42 FR 35314, 45 FR 48941 (IV-D-26; IV-D-28). Still another commenter
(IV-F-1, No.3) called decisions relating to the choice of substances to be
included in SOCMI arbitrary. .. ,
One of the commenters pointed out (IV-D-28) that EPA was aware of
differences in photochemical reactivity and differences in contribution to
ozone formation with differences in chemical structure. He cited an EPA
report; Fate of Toxic and Hazardous Materials in the Air Environment.
600/3-80-084, December 1980; and a Federal Register notice, 45 FR 48941. He
also cited the transcript of the public hearing (pp. 52 and 70) as proof of
the Agency's awareness of this fact. He went on to say that the Agency has
compiled the list of SOCMI chemicals based on volume instead of photo-
chemical reactivity even though they were aware of reactivity differences.
He also said that the list was arbitrarily selected with no explanation
about its compilation or why chemicals were or were not included.
One commenter (IV-D-11) said the different degrees of toxicity of the
chemicals required tailoring the "operating and maintenance" procedures for
the individual situation. Another (IV-D-6) wrote that a performance
standard is needed instead of equipment standards because of the wide
variety of chemicals involved having a wide variety of chemical, physical,
and biological properties. The commenter added that a single solution for
all of the chemicals listed is not sound technically.
Two other commenters (IV-D-38; IV-D-40) were also concerned with the
selection of chemicals for the SOCMI list. Both commenters felt that SOCMI
was comprised of a wide variety of chemical processes. They said EPA should
evaluate the individual processes before including them in the standards.
One commenter (IV-D-40) stated that, in taking a unit operations approach to
regulations, EPA needed to ensure that the regulations were reasonable and
technically sound for all members within the class. The other commenter
(IV-D-38) pointed out that, of the 600 different processes within the
category, EPA had evaluated only 27 processes as "the most likely candidates
for NSPS or NESHAP coverage through generic standards."
5-2
-------
Response:
Section 60.489 [Appendix E of the proposed regulation] is not a list of
ozone-forming (photochemically reactive) chemicals although many chemicals
on the list are photochemically reactive. The list consists of chemicals
whose production requires equipment such as valves and pumps for processing
photochemically reactive VOC. The photochemically reactive VOC may be
products, reactants, additives, or intermediates. As explained in Docket
Item No. IV-B-21 as of January 1983, EPA considers eleven organic compounds
nonphotochemically reactive:
methane
ethane
1,1.,1-trichloroethane
methylene chloride
trichlorofluoromethane
dichlorodifluoromethane
chlorodifluoromethane
trifluoromethane
trichlorotrifluoroethane
dichlorotetrafluoroethane
chloropentafluoroethane
As described in the BID for the proposed standards, the chemicals produced
are the building block chemicals for many of the downstream industries
producing synthetic products such as plastics, Pharmaceuticals, textiles,
and specialty chemicals with a wide variety of uses. This basis for
selection of the SOCMI chemicals was explained at proposal (40 FR 1136,
January 5, 1981). All other organic substances are considered VOC now.
As further explained in Docket Item No. IV-B-21, some of the chemicals
cited by. the commenters are photochemically nonreactive; however, their
production requires fugitive emission sources to process photochemically
reactive substances. For example, methylene chloride may be produced by the
hydrochlorination of methanol, a VOC. Therefore, such chemicals remain on
the list because photochemically reactive substances are used in their
production processes. The criteria for their removal include not only
5-3
-------
whether the product Is photochemically reactive, but also whether reactants,
additives, or intermediates are photochemically reactive. Because the
chemicals mentioned fail to meet all of the criteria, they remain on the
list.
As the commenter (IV-D-28) pointed out, EPA is aware of differences in
photochemical reactivity and differences in contribution to ozone formation
with differences in chemical structures. However, the magnitude of these
differences is not fully understood and depends on factors that are not
fully understood. Therefore, it would be extremely difficult, and at least
impractical, to establish regulations based on degree of photochemical
reactivity. It should be noted that some photochemically reactive chemicals
react quickly in the atmosphere to form ozone. Others take longer, but they
are present In sufficient quantity and exist long enough in the atmosphere
to contribute to ozone formation.
Similarly, it is not necessary to establish separate regulations based
on degree of toxicity or chemical, physical, and biological properties.
Other regulations such as National Emission Standards for Hazardous Air
Pollutants (NESHAP) [Section 112 of the Clean Air Act] and OSHA regulations
are aimed at regulating specific chemicals based on their serious toxic"
effects in humans.
The basis of the SOCMI list must be viewed in the context of the use
of the list. The list defines the extent of coverage of standards of
performance for fugitive emission sources of VOC. These standards of
performance are designed to protect air quality by reducing emissions of VOC
from equipment within SOCMI. In doing so, the environment is further
protected against the toxic effects of some of the chemicals found in SOCMI.
Physical and chemical properties of the chemicals have been considered in
composing the SOCMI list. For example, vapor .pressure distinctions have
been established to eliminate routine requirements for equipment which has a
low tendency to leak. As discussed in other portions of this section, EPA
has established other limitations on the use of the list. The chemicals on
the list or chemicals associated with their production participate in the
formation of ozone, and equipment which has been shown to leak is used to
5-4
-------
process these chemicals. .Therefore, their release to the atmosphere should
be control led.
To date, EPA has studied, in detail, fugitive emissions from about
20 chemical processes in SOCMI. In developing the model units for SOCMI,
over 50 process units were surveyed to determine counts of fugitive emission
sources. These process units included the 20 chemical processes studied in
detail. The results of this work support the general technical judgments
made in developing the standards for fugitive VOC from SOCMI. In another
study (IV-A-11) EPA selected a cross section of units for study. The
process units tested in SOCMI have provided sufficient information to
confirm major conclusions concerning fugitive emission sources in VOC
service. As other fugitive emissions data indicated, SOCMI fugitive
emission sources leak. Just as expected, equipment in heavy-liquid service
leaked less frequently with lower emission rates than equipment in light-
liquid and gaseous service. Since process fluid vapor pressure (a factor
common to all process units) is the overriding consideration in predicting
leak frequencies and leak rates, testing all SOCMI units is unnecessary.
Economic data on SOCMI were also collected and examined in developing
the economic impact to the industry. Economic and financial data on
100 chemical firms were studied to develop the models used to evaluate the
cost and economic impact to the industry.
Comment:
Three commenters noticed that chlorofluorocarbons were being regulated
for their tendency to destroy ozone in the upper atmosphere. (IV-F-1,
No. 3; IV-D-15; IV-D-28). They thought it illogical to control a compound
because it destroys ozone and because it creates ozone. One of the
commenters (IV-D-28) referred to 45 FR 66776 as support for removing
chlorofluorocarbons from SOCMI.
Response:
Chlorofluorocarbons (produced from such substances as perchloro-
ethylene, carbon tetrachloride, and fluorinated derivatives of acetylene)
are not being regulated because they form ozone but because they are
produced from chemicals that form ozone in the troposphere. In any case, to
5-5
-------
the extent that chlorofluorocarbons are controlled, the standards will
reduce the destruction of ozone in the stratosphere.
Comment:
In three sets of comments (IV-F-1, No. 5; IV-D-15; IV-D-9) it was
suggested that urea should be withdrawn from the list of chemicals in
Appendix E (§60.489) since the manufacture of urea does not involve VOC.
The commenter said that urea is produced by reacting ammonia and carbon
dioxide. This reaction produces ammonium carbonate which then decomposes to
urea. The urea produced is very pure, containing only trace amounts of a
co-product, biuret. Decomposition of urea yields biuret, ammonia, and
cyanuric acid. Decomposition of biuret yields ammonia and cyanuric acid.
Response:
. The process for manufacturing urea involves* a combination, of up to
seven major unit operations. These major operations are:
(1) solution synthesis (solution formation)
(2) solution concentration
(3) solids formation
- prilling
- granulation
(4) solids cooling
(5) solids screening
(6) solids coating
(7) bagging and/or bulk shipping
The combinations of processing steps are determined by the desired end
products. Plants producing urea solutions alone are comprised of only the
first and seventh unit operations, solution formation and bulk shipping.
Facilities producing solid urea employ these two operations and various
combinations of the remaining five operations depending upon the specific
end product.
Emissions from urea processes include particulate matter and ammonia.
In addition, formaldehyde, a VOC, may also be emitted in some urea processes
when the product is urea solids. Small amounts of formaldehyde are used as
an additive to reduce dust emissions and to prevent solid urea product from
5-6
-------
caking during storage. In view of the potential for fugitive emissions of
VOC, urea plants are appropriately covered.
Even though all urea plants produce an aqueous urea solution, not all
plants would have a solids formation operation. Plants not producing urea
solids have no formaldehyde addition step and, therefore, no potential for
fugitive emissions of VOC. Thus, it would be appropriate to grant an
exemption to urea plants that do not use formaldehyde. In addition, it
would be appropriate to grant an exemption to any SOCMI unit that does not
process VOC. Therefore, an owner or operator of a facility producing a ;
chemical listed in §60.489 by a process in which no VOC are processed will
be granted an exemption from the standards. An exemption for such
situations is provided in §60.480 of the regulation (applicability and
designation of affected facility).
Comment: :
Two commenters (IV-D-19; IV-D-18) objected to including chemicals on
the SOCMI list which were also covered by NESHAP. Vinyl chloride and " r";
benzene were cited as examples. One of the commenters (IV-D-18) felt that
facilities which have established monitoring programs under NESHAP standards
should be allowed to expand the program to those areas within the source to
which VOC NSPS would apply. He said that this would eliminate the costs
associated with having two separate redundant programs in effect for the
source. He preferred the NESHAP program because the monitoring and record-
keeping program afforded a greater level of control but is less burdensome.-
The other commenter (IV-D-19) also recommended that a change be made to
exclude units covered by NESHAP's.
Another commenter (IV-D-48) stated that the inclusion of vinyl chloride
as a VOC and the regulation of facilities which produce vinyl chloride by
the proposed regulations is redundant and in conflict with EPA regulations
currently in force. He added that the proposed regulation is in conflict
with the NESHAP for vinyl chloride in that the NESHAP sets a lower maximum
discharge concentration than the proposed regulation. The commenter pointed
out that the NESHAP also sets standards for controlling fugitive emissions
from loading and unloading lines, pumps, compressors, relief valves, and the
5-7
-------
opening of equipment. The NE5HAP also contains provisions for leak detec-
tion and elimination. The commenter suggested deletion of vinyl chloride
from Appendix E (§60.489).
Response;
The vinyl chloride and benzene NESHAP's and SOCMI NSPS are aimed at
regulating emission sources for different purposes. For example, benzene
fugitive emissions occur in the petroleum refining and chemical manufac-
turing industries as a result of the production and use of materials that
contain benzene. The proposed benzene NESHAP would regulate only the
components containing 10 or more percent by weight benzene (see proposed
§61.111, 46 FR 1165). Fugitive emission sources containing more than
10 percent would be regulated by the benzene NESHAP only. Fugitive emission
sources containing less' than 10 'percent'benzene and located in an affected
facility covered by the SOCMI standards for fugitive emission sources would
be regulated only under the SOCMI (or refinery) VOG fugitive NSPS. It
should be noted that at proposal both the benzene NESHAP for new sources and
the SOCMI NSPS had the same requirements.
The SOCMI standards are applicable only to new vinyl chloride process
units. There are no conflicting requirements in the two standards. For
example, compressors are required by both standards to be equipped with seal
systems. Similarly, an alternative permitted by both standards is the use
of seal!ess pumps. Both standards have semiannual reporting requirements,
but the requirements are not duplicative. Therefore, redundancy should not
be a problem.
Comment:
One commenter (IV-D-4) said the list of chemicals in Appendix E
(§60.489) should be confined to those processes which have been tested and
shown to be significant contributors to air pollution. He suggested that
the list could be amended as new test data became available.
Two commehters representing the same company (IV-D-4; IV-D-23) said
adipic acid should be deleted from the list because test data showed no
leaks in their adipic acid plant. Only four components of the 775 screened
gave positive readings on the screening instrument. The readings were 32,
230, 580, and 2,000 ppmv.
5-8
-------
One of the commenters (IV-D-23) said phenol should also be excluded
from coverage because fugitive emissions were only detected in-those parts
of the phenol unit which were handling acetone.
Response:
As discussed in a response to a previous comment in this section, the
fugitive emissions data collected in SOCMI units confirm the fact that
fugitive emissions occur in SOCMI and that vapor pressure is the overriding
consideration in predicting leak rates and leak frequencies. A cross
section of highrleaking and low-leaking units was tested in SOCMI. The test
results are sufficient to indicate that fugitive emission sources in SOCMI
units do emit substantial quantities of VOC.
Fugitive emissions data show that there is considerable variation in
the leak frequencies of different units of the same process type. Thus, it
is not unexpected that an adipic acid unit could have leaks. It would be
inappropriate to characterize a process as a low-leaking or high-leaking
process based on the results of tests conducted on a limited number of
units. The same argument holds for the case of the phenol unit where only a
part of the unit had leaks. It is precisely because of the production of
acetone and the use of VOC reactants that phenol is covered. One possible
reason for the absence of leaks in some parts of the phenol and adipic acid
units may be due to the fugitive emission sources being in heavy-liquid
service only. In any case, not all phenol or adipic acid units would
necessarily have the same characteristics. EPA has, therefore, concluded
that the standards are appropriate for units producing the above chemicals.
However, EPA believes that incentives should be provided for process
units which have very few leaks. For this reason, EPA has provided
alternative standards for valves. The standards are based on maintaining a
performance level of 2 percent or less of valves leaking within a process
unit. These alternative standards are discussed in more detail in'
Chapter 14.
Comment:
Many of the appendix categories were called overly vague in one comment
letter (IV-D-17). Acrylic acid and esters, ethanolamines, phenol sulfonic
5-9
-------
acids, polybutenes, tetrachloroethanes, toluene sulfonic acids, toluidines,
and trichlorobenzenes were listed as examples. The commenter was joined by
another commenter (IV-D-51) in recommending that each compound to be covered
by the NSPS be listed separately. Another commenter (IV-D-35) pointed out
that the list includes acrylic acid and esters as one entry, but gives
separate entries for n-butyl acrylate and ethyl acrylate.
Response:
As suggested by the commenter, the duplication of the acrylic acid
esters has been removed by removing the words "and esters" from the entry
for acrylic acid. However, it should be noted that some entries include
more than one chemical compound. For example, the entry for trichloro-
benzenes indicates inclusion of all isomers, and EPA1s intent is to include
all isomers. .
Comment:
One commenter .(IV-D-28) objected to EPA's alleged selection of
chemicals on the basis of production volume. However, another commenter
(IV-D-7) argued that the list should be based on production volume and
number of units. This commenter continued, saying that raw materials have
little or no bearing on emissions and, therefore, should not be the basis
for selection of the chemicals. He also noted that many of the chemicals on
the list are produced in insignificant quantities and should be removed from
the list. Another commenter (IV-D-35) questioned whether low-volume
monomers such as hydroxypropyl acrylate would be covered by the rules.
Response:
The segment of the synthetic organic chemical industry covered by the
proposed standards is a readily identifiable subgroup of the organic
chemical industry. The products of this industry segment are derived from
about ten basic petrochemical feedstocks and are used as feedstocks in a
number of synthetic products industries. Many of these products are high-
volume chemicals. Production volume alone is, however, not the basis for
selection of these chemicals, as alleged by the commenter (IV-D-28). EPA
studies of VOC fugitive emissions showed little or no predictable relation-
ship between emissions and line size and capacity (II-A-9). The volatility
5-10
-------
and/or the phase of the process stream is the equipment or process variable
which greatly influences fugitive emission rates. Unless data can be
produced to the contrary, fugitive emissions cannot be assumed to be a
function of production volume. All chemicals listed in §60.489 will,
therefore, continue to be covered under the regulation, regardless of their-
production volume.
There are units, however, with production rates so small that emissions
would also be very small. The cost to control a very small amount of
emissions would be exorbitant when compared to the emission reduction
achieved. Therefore, EPA has excluded units producing less than 1,000 Mg/yr
from coverage. The production rate cutoff is explained in more detail in
Section 5.7. :
Even though EPA agrees with the commenter (IV-D-7) who stated that-'raw
materials have little or no bearing on emissions, EPA believes that the 'raw
materials that are processed in fugitive emission sources should be the
basis for §60.489. Emission factors have been developed for each equipment
type (valves, pumps, etc.) for each of the three kinds of service ('gas, '--"
light liquid, and heavy liquid) by measuring emissions of the raw
material(s) in the line. Moreover, the intent of standards is to reduce
emissions of substances which are photochemically reactive and the
substances emitted by these sources are the substances processed by the
fugitive emission sources.
Comment: ,
One commenter (IV-D-12) noted that a very large fraction of the 400 to
500 chemicals on the list are also on the list in Table Z-l'of 1910.1000 of
OSHA Law CFR29. He said these materials in air from all sources, including
fugitive emission sources.are enforced by OSHA at levels which are in most
cases far below the present no effect levels and in all cases below the
10,000 ppmv level by several orders of magnitude. He commented that
exemption of these duplicated chemicals from these standards would yield
significant savings in capital expense and enforcement costs without compro-
mising air quality in any significant way.
5-11
-------
Response:
The above comments are based on a premise that the SOCMI NSPS and OSHA
regulations are duplicative. The two regulations have different objectives
and different approaches to emissions control (see Section 2.2). The
standards, however, do provide sufficient flexibility so that units that
have very few fugitive leaks due to compliance with OSHA can realize cost
savings by choosing to comply with the several allowable alternative
standards (see Chapter 14).
Comment:
One commenter (IV-D-12) said furfural plants were not like the oil
refineries and petrochemical plants for which the proposed standards have
been developed.
Other commenters (IV-D-48; IV-D-42) wrote that another chemical produc-
tion process that EPA has unintentionally regulated is the production of
ethanol (grain alcohol) via biological processes from grain feedstocks.
They suggested that the proposed definition of "Synthetic Organic Chemicals
Manufacturing Industry" be amended. They suggested, including a qualifica-
tion to the SOCMI list, that the products be produced from petrochemical
feedstocks. One of these commenters added that the beverage distilled
spirits industry, which produces alcohol from natural food products by the
natural process of fermentation, has never been thought of as part of the
Synthetic Organic Chemicals Manufacturing Industry and does not produce
synthetic alcohol. The commenter wrote that the EPA Draft Environmental
Impact Statement (EIS) used to justify regulation of SOCMI establishes that
fermentation alcohol cannot be regulated in this rulemaking. The commenter
also claimed that the EIS is based upon studies of what is conventionally
thought of as the chemical industry.
- This commenter further stressed that fermentation alcohol is not a
synthetic organic chemical. . He wrote that synthetic alcohol typically is
produced by a very complex process whereby ethylene (a VOC) is produced
from a petrochemical feedstock and then is converted to alcohol by esteri-
fication-hydrolysis. The process requires complex equipment containing
numerous valves, pumps, connections and other potential sources of VOC. By
5-12
-------
contrast, the production of alcohol fermented from grain and the like was
said to be a relatively simple process involving few sources of potential
VOC emissions. The commenter submitted process diagrams for synthetic and
natural alcohol processes in support of his argument.
The commenter continued that the production of alcohol represents only
a miniscule portion of the total chemical production EPA intends to
regulate. Ethanol produced from grain, molasses, fruit and whey accounted
for only 478.2 thqusand Mg out of a total of 319,835 thousand Mg VOC
produced in 1976. Thus, only 0.15 percent of the total was fermentation
alcohol.
The commenter cited EPA studies to support his argument that EPA had
previously recognized that alcohol evaporation as a result of the fermenta-
tion and distillation process is too small a factor to justify regulation.
These EPA studies related to the whiskey distilling industry and whiskey
warehousing. The commenter pointed out that EPA stated in those reports
that the fugitive ethanol emissions from production were low. Also, in
those reports EPA did not even consider any controls on beverage alcohol
fermentation and distillation, but focused entirely on whether controls on
aging warehouses were warranted.
Response:
The background information document for the proposed standards pointed
out that SOCMI chemicals are produced from a variety of raw materials in a
wide range of processes (.over 600 for the industry). Although much of the
data presented in the background document was for petrochemical process
units, the conclusions drawn from the data in setting the standards are
applicable to SOCMI in general. The standards have not been developed
specifically for oil refineries or petrochemical plants. Nor have the
standards been developed specifically for processes based on the criteria of
whether a synthetic process is involved. The standards are aimed at plants
producing certain products contained in the SOCMI list which involve the use
and handling of VOC in the production process and have fugitive emission
sources (e.g., pumps, compressors, valves, etc.) in VOC service. Synthetic
organic chemicals are produced by physical, chemical, and biological
5-13
-------
processing methods which involve many operations including the handling of
process fluids. As noted in the proposal BID, most of the synthetic organic
chemicals produced in the United States are derived from crude petroleum and
natural gas, with oil, shale, coal and biomass also serving as primary
feedstocks. Furfural and grain alcohol processes use agricultural materials
to produce, through biological synthesis, organic chemicals that are photo-
chemical ly reactive. Operations in the processing plants contain process
equipment types which are fugitive emission sources (e.g. pumps, valves,
etc). The regulation, therefore, appropriately covers furfural and grain
alcohol. However, as explained in Sections 8.1 and 8.2, these standards do
not cover certain facilities in the whiskey distillation and beer
manufacturing industries.
Comment:
Several comments (IV-D-20; IV-D-15; IV-D-16; IV-D-48) were received
concerning potential application of the standards to refineries. It was
noted that some process units in refineries would qualify as affected
facilities under the proposed definition. One commenter (IV-D-16) requested
that refineries be specifically excluded. Another (IV-D-20) said that it
was inappropriate to cover such units because their primary purpose was not
the manufacture of chemicals. He recommended that if they were covered
under the proposed rules for SOCMI, they should be exempted from the
refinery rule.
One commenter noted that refinery products are often used as feedstocks
in SOCMI, and certain petroleum refineries produce only feedstock materials
for SOCMI, and therefore by definition, are properly a part of SOCMI.
However, the commenter objected to the fact that these regulations also
apply to petroleum refineries whose principal products are motor fuels or
related products. He said these facilities, by definition, are not part of
SOCMI and should not be included in this regulation simply by virtue of the
fact that one or more chemicals which are split as separate fractions during
the processing of crude oil into gasoline are on the list. (Preamble at
46 FR 1461.)
5-14
-------
He concluded that if the suggested exclusion is not granted to petro-
leum refineries whose principal products are motor fuels and related pro-
ducts, these regulations will have a significant economic impact upon this
segment of the industry. By way of illustration, the commenter estimated
that at one of his company's small refineries, producing slightly more than,
50,000 BPD's of gasoline products, the initial capital expenditure will be
$100,000 with annual operating and maintenance costs of a similar magnitude.
Response: .
The primary purpose of most refineries is manufacture of petroleum-
products, such as motor oils. However, some refineries do produce organic
chemicals. EPA is regulating process units that produce one or more of the .
chemicals on the SOCMI list. Because some refineries have sources of ;
fugitive VOC emissions (such as pumps and valves) involved in producing one
or more SOCMI chemicals, EPA believes that the standards appropriately apply
to process units in these refineries that produce these chemicals.- EPA has
considered the impact on SOCMI units located in refineries as well as SOCMI
units located in chemical plants. The impacts are not different, and as
shown in Section 7.2 they are reasonable. In light of these facts, it is
appropriate to regulate affected facilities located in refineries. However,
to eliminate any potential redundancy or confusion, process units that are
covered under the SOCMI standards would be exempted from refinery standards
presently being developed.
Comment:
A commenter (IV-D-43) recommended that some chemicals be eliminated
from the SOCMI list. He said some could be eliminated based on their vapor
pressure. Others, he said, could be eliminated because they do not contri-
bute to ozone formation.
Response:
Any chemicals that may be classified as heavy liquids, based on their:
vapor pressures, remain on the list because light liquid VOC are present in
the process, as reactants, additives or by-products. Similarly, some
chemicals that may not themselves contribute to ozone formation are on the
list because they have ozone forming reactants, additives, or by-products in
5-15
-------
the production process. However, process units may be excluded if they
process only heavy liquid VOC or non-VOC in producing chemicals on the list.
5.2 VAPOR PRESSURE CUTOFF
Comment:
One commenter (IV-D-6) said the proposed definition of light liquid
service was not consistent with the test work referenced in the background
information document. He did not cite any specific inconsistencies.
Response:
The background information document data classify a stream as light
liquid if the following.conditions apply:
(1) The vapor pressure of one or more of the components is greater
than 0.3 kPa at 20°C.
'(2) The total concentration of the pure components having a vapor
pressure greater than 0.3 kPa at 20°C is equal to or greater than
20 percent by weight.
(3) The fluid is .a liquid at operating conditions.
An emission source "in VOC service" was.defined as a source containing
a process fluid that is at least 10 percent VOC by weight. During the
development of the regulations, several industry representatives called the
two definitions confusing and suggested making the percent cutoff for light
liquid and VOC the same. Therefore, at proposal the percent cutoff for
light liquid was changed to 10 percent. As the commenter pointed out, this
has caused an inconsistency between the proposed, definition and the test
data base. Because EPA considers it more appropriate to maintain consis-
tency between this definition and its test data base than to reduce a
potential confusion in another definition, the light liquid percent cutoff
has been changed to 20 percent.
Connent:
One commenter (IV-D-6) objected to applying a vapor pressure cutoff
developed on refinery plant streams to SOCMI streams. He said the majority
of SOCMI processes utilize pure components to produce pure products, while
the refining industry processes mixtures. He cited a model which he had
developed for predicting evaporation of spills and emissions from pump seals
5-16
-------
as a function of temperature. He said the model shows significant
differences between chemicals of concern in the refining industry and SOCMI.
Another commenter (IV-D-19) argued against using the proposed vapor pressure
cutoff because EPA had apparently used refinery data to derive the vapor
pressure cutoff point. Noting that petroleum fractions are mixtures of
compounds with widely varying vapor pressures, he said the data had been
inappropriately interpreted. One other commenter (IV-F-1, No.3) called the
definition of light liquid arbitrary.
Response:
Although the vapor pressure cutoff of 0.3 kPa was derived from
petroleum refinery data, it does not represent misinterpretation of data.
It should be noted in reference to the comment about differences between
SOCMI and petroleum refineries that SOCMI streams can also be mixtures of
compounds. On the other hand, several streams in refineries are pure
chemicals. The differences between the industries are not as clearly
defined as the commenter has indicated. However, an analysis of the vapor
pressures and emission rates had shown that substances with vapor pressures
of 0.3 kPa or higher had significant emission rates while those with lower
vapor pressures did not. This represents the split between kerosene and
naphtha (II-A-7) and is the criterion used by EPA to distinguish between
light liquid and heavy liquid substances. This criterion was used in
collecting the SOCMI data, and EPA is maintaining consistency by using the
same criterion in the standards. .. . . -
Comment:
The previous commenter (IV-D-6) argued further that there is only one
study of fugitive emissions of pure components in the literature, a pump
seal study by Summerfield. He said the study showed that for chemicals with
boiling points above 20°C there was little in the way of emissions. The .
average for 20 tests was 2.8 g/hr. The data were further reported to show
that 90 percent of the pumps tested with chemicals that boil at a
temperature above 20°C had emission rates of less than 4 g/hr. He pointed
out that this emission rate is 1/3 the emission rate reported by EPA for
Alternative A.*
Interpreted by EPA to mean Alternative I.
5-17
-------
He recommended that the definition of light liquid service be changed
to a stream having 20 percent or greater concentration of a component with
vapor pressure greater than 760 mm Hg at operating temperature.
Response:
EPA made a thorough review of the data base for fugitive VOC emissions
in the recently released AID (47 FR 19724, May 7, 1982). This review points
out the limited usefulness of the Summerfield study with respect to applica-
bility to actual SOCMI operating units. The review supports the decisions
made by EPA in defining light liquid service.
Comment!
One commenter (IV-D-19) recommended a vapor pressure breakpoint of
3.5 kPa. He thought the 0.3 kPa criteria unreasonably stringent. He said
regulations for VOC storage require floating roofs or equivalent control for
tanks 75 m or larger with liquids having vapor pressures greater than
3.5 kPa and less than 76.6 kPa.
Three commenters (IV-D-13; IV-D-18; IV-D-20) recommended a vapor
pressure cutoff of 1.5 psi for light liquid service. One (IV-D-13) said
the SIP's uniformly recognize 1.5 psi for purposes of establishing vapor
controls on storage tanks and other units which are considered major sources
of VOC emissions. He considered 1.5 psi a reasonable cutoff level and added
that there were no data to justify a vapor pressure cutoff of 0.3 kPa for
SOCMI fugitive sources.
Another commenter (IV-D-41) proposed that 10.3 kPa (1.5 psi) be used as
the cutoff vapor pressure for the liquid required in dual mechanical seals.
He wrote that as defined, a heavy liquid would have a vapor pressure of less
than 0.3 kPa (0.04 psi) at 20°C which is 35 times more stringent than the
10.3 kPa specified for seals on floating roof storage tanks.
Response:
VOC in storage tanks are at ambient temperature and pressure condi-
tions. VOC processed by fugitive emission Sources, on the other hand, are
at elevated temperature and pressure conditions. The storage tank
criterion, therefore, has no bearing on the fugitive emissions standard.
The vapor pressure cutoff for fugitive emission sources was established to
eliminate requirements for sources which had low leak or emission potentials.
5-18
-------
Comment:
A commenter (JV-D-19) said there were chemicals (cumene, normal propyl-
benzene, styrene, ethylbenzene) which are light liquids by the proposed
definition but which would cause less than a 10,000 ppmv reading even if the
air were saturated with them at 20°C. Another commenter (IV-D-18) said low
vapor pressure materials would not register on a VOC detector.
Response:
As pointed out by the commenter, it may be true that certain chemicals
which are light liquids by the proposed definition would cause .less than a
10,000 ppmv reading even if the air were saturated with.them at 20°C.
However, not many streams in SOCMI are expected to be encountered at 20°C.
At higher temperatures the chemicals mentioned by the commenter will show
readings of 10,000 ppmv or greater. For example, the EPA 24 unit study
(IV-A-11) included 2 cumene units. Several sources in both units were found
leaking (i.e., 10,000 ppmv or greater reading). In fact, 16 percent of all
gas valves and 14 percent of all pumps in these cumene units were found to
have readings of 10,000 ppmv or greater.
Comment:
One commenter (IV-D-28) said the vapor pressure cutoff would classify
water as a light liquid.
Response:
Water is not a VOC. The standards do not require equipment processing
mostly water to be screened with the monitoring instruments.
Comment:
One commenter (IV-D-28) said the vapor pressure criteria have no health
or environmental basis.
Response:
The SOCMI NSPS are technology based standards. They are aimed at
prevention of degradation of air quality due to the emission of VOC by new,
modified, and reconstructed sources. There is no need for the vapor
pressure criterion to have a health basis. The purpose of the vapor
pressure cutoff was to exempt heavy liquids from the regulation, since they
had very little or no leakage.
5-19
-------
Comment;
One commenter (IV-D-13) objected to using vapor pressure data for pure
components in the line. He thought the vapor pressure criteria should be
applied to actual conditions in the line. He said this approach would also
avoid the problem of determining the percent of VOC that qualifies a mixture
as a VOC.
Response:
. The regulations define a source as being in light liquid service if:
(1) The vapor pressure of one or more of the components is
greater than 0.3 kPa at 20°C.
(2) The total concentration of the pure components having a vapor
pressure greater than 0.3 kPa at 20°C is equal to or greater
than 20 percent by weight.
(3) The fluid is a liquid at operating conditions.
The above definition is consistent With the SOCMI NSPS data base. In
addition, vapor pressures of pure components are easily available from
standard reference texts (IV-A-6). Applying the vapor pressure criteria to
the actual conditions in the line would cause unnecessary complications.
For example, to determine the actual vapor pressure of the mixture in line,
an owner or operator would have to conduct vapor pressure tests on all
mixtures of chemicals in each line at the temperature conditions of the
mixture. In EPA's judgement such complications are unwarranted, especially
in light of the simple, yet reasonable approach of summing the
concentrations of pure components.
Comment:
One commenter (IV-D-43) said EPA has not justified its present
arbitrary criteria for dividing substances into light and heavy liquids.
Another (IV-D-18) said 1.5 psi was the standard level for vapor pressure
cutoff. He said that any effort to lower this level should be explained and
justified. He called EPA's selection of 0.3 kPa completely arbitrary. A
third commenter (IV-D-20) recommended that the level be adjusted to 1.5 psi
RVP because that is the definition used by South Coast Air Quality Manage-
ment District (SCAQMD). He said if it were kept at 0.3 kPa, it would
include nearly every hydrocarbon stream in the plant.
5-20
-------
Response:
As explained previously, the 0.3 kPa vapor pressure criterion was based
on fugitive emission data gathered in petroleum refinery studies. Equipment
processing VOC with vapor pressures above 0.3 kPa leaked at significantly
higher rates and frequencies than equipment processing VOC with vapor
pressures below 0.3 kPa. Therefore, EPA elected to exempt equipment
processing lower vapor pressure VOC substances from the routine leak
detection and repair requirements of the standards.
Available data do not show that if the cutoff were kept at 0.3 kPa it
would include nearly every hydrocarbon stream in the plant. EPA's 24 unit
SOCMI study shows several sources to be in heavy liquid service. For
example, 13 percent of all pumps and about 17 percent of all liquid service
valves screened were found to be in heavy liquid service. However, if all
the equipment in a process unit is in light liquid service, the resulting
higher leak frequency could provide further evidence that these emission
sources should be controlled.
Comment:
One commenter (IV-D-19) challenged the concept of a vapor pressure
cutoff. He said it was inappropriate to characterize the vapor pressure of
a mixture by the vapor pressure of its least volatile component and then to
imply that streams containing 10 percent or greater of this least volatile
component should be treated like naphtha.
Response:
As discussed in an earlier response in this section, the vapor
pressure of a stream is not characterized by the vapor pressure of its least
volatile component. A stream containing 10 percent or more of liquids whose
vapor pressure is 0.3 kPa or greater at 20°C is classified as a light
liquid. The 0.3 kPa cutoff is based on fugitive emissions data gathered in
petroleum refineries. The data show that fugitive emission sources
processing VOC streams with vapor pressures lower than 0.3 kPa have low (but
measureable) probabilities of leaking. The cutoff was established to
eliminate those sources from some of the requirements of the standards.
5-21
-------
5.3 IN VOC SERVICE
Comment:
One commenter (IV-D-20) suggested that EPA exempt streams containing
less than 20 percent by weight VOC instead of 10 percent. He said the
20 percent figure was used by SCAQMD.
Another commenter (IV-D-46) stated that the 10 percent VOC cutoff
should be conditioned to prevent evasion. He asked EPA to make provision
for the possibility of plant designers diluting certain streams so as to
avoid the need for controls. The commenter suggested that the rules should
also make clear that where streams fluctuate above and below 10 percent VOC,
the whole stream and its associated valves, seals, and pumps are subject to
the control requirements of the standards.
Response:
The purpose of the 10 percent cutoff is to avoid covering those sources
that have only small amounts of photochemically reactive substances in the
line. In any case, very few sources are expected to have streams containing
between 10 and 20 percent VOC. In view of the strict purity requirements of
most chemicals, EPA does not consider dilution, for the purpose of evasion,
to be a potential problem. However, the rules have been changed to make
clear the fact that streams fluctuating above and below 10 percent VOC will
be covered by the standards.
There has also been confusion over the compounds to be considered in
computing the percent VOC. Volatile organic compounds (VOC) are defined as
organic compounds that participate-in atmospheric photochemical reactions.
EPA considers several organic compounds to be nonphotochemically reactive
(methane; ethane; 1,1,1-trichloroethane; methylene chloride; trichlorofluo-
romethane; dichlorodifluoromethane; chlorodifluoromethane; trifluoromethane;
trichlorotrifluoroethane; dichlrotetrafluoroethane; chloropentafluoro-
ethane). In determining the percent VOC in a process line (as a prerequi-
site to determining whether a piece of equipment is in VOC service),
quantities of the nonphotochemically reactive compounds (these considered
such by EPA)present in the line may be excluded from the total quantity of
organic material.
5-22
-------
5.4 PROCESS UNIT DEFINITION
Comment: . .
Five comment letters (IV-D-15; IV-D-17; IV-D-20; IV-D-14; IV-D-21) said
that the definition of process unit was vague. Two of the commenters
(IV-D-14; IV-D-21) cited potentially confusing situations in which the
standard might be misapplied to solvent recovery operations. One ('IV-D-21)
recommended that the definition include the conditions that the chemical be
produced by chemical synthesis.
Another commenter (IV-D-51) wrote that, as proposed, "process unit"
means equipment assembled to produce, as intermediates or final products,
one or more of the chemicals listed in Appendix E (§60.489).. The key word
of the definition in the context of the overall intent of the Subpart VV is
the word "produce." The commenter added that as the Administrator explains..
in the preamble the applicability of Subpart VV is for facilities that
produce the listed chemicals and not facilities that use them. He expressed
concern that unless this point is clarified it may be interpreted that
process units that purify or recover the listed chemicals are producing
them. The commenter stressed that to produce a chemical means to convert
raw materials by one or more reaction steps to the desired chemical which
may be either an intermediate or a final product. He recommended that the
following sentence be added to the definition of process unit: "Process
units that handle, but do not form by chemical reactions,, the chemic.als
listed in Appendix E (§60.489) are excluded ^rom this subpart."
Another commenter (IV-D-14) recommended that a definition of inter-
mediate products be included to avoid the potential misunderstanding.
Response:
Process unit is defined as equipment assembled to produce, as inter-
mediate or final products, one or more of the SOCMI chemicals (listed in
§60.489). A process unit can operate independently if supplied with
sufficient feed or raw materials and sufficient storage facilities for the
product.
The definition was drafted to provide a common sense, practical way to
determine which equipment are included in an affected facility. There are
5-23
-------
no specific physical boundries or size criteria. The definition instead
depends upon several operational factors, including chemical produced and
the configuration of the processing equipment. The configuration of the
processing equipment may be different for different producers of the same
chemical, and, therefore, it may be fairly site-specific. However, in
practice, the definition will implement the selection of a process unit
basis as the "source" covered by the standards.
The intent of the standards is to cover process units that produce the
chemicals listed in §60.489, either by chemical reaction or by other
processing means, such as separation and purification techniques. EPA sees
no justification in specifically excluding solvent recovery operations. VOC
fugitive emissions occur from equipment in VOC service. Therefore, if there
are any fugitive emission sources in VOC service in a process unit producing
one of the chemicals listed in §60.489, they should be covered by the
standards.- These equipment -components would be present in a process regard-
less of whether it is a chemical synthesis or separation process. It would,
therefore, be inappropriate to define process unit by requiring that a
chemical be produced by chemical synthesis or by a separation process.
Solvent recovery operations will be covered by the SOCMI standards if
they are producing chemicals listed in §60.489. EPA has considered the
impact on producers of all these chemicals. The impact on SOCMI units
located in solvent recovery plants is no different from the impact on SOCMI
units located in chemical plants.
Intermediate chemicals are typically those chemicals produced from raw
materials which are then used captively to generate a final product(s). The
equipment assembled to produce an intermediate chemical constitutes a
process unit if it can be operated independently when provided sufficient
storage for raw materials as well as the intermediate chemical itself.
Thus, any process unit producing a chemical listed in §60.489 as an
intermediate chemical would be covered by the standards.
Furthermore, chemicals listed in §60.489 that are produced as
coproducts in a process unit would also result in coverage of that unit by
the standards. Examples of coproducts are phenol and acetone. Chemicals
5-24
-------
listed in §60.489 produced as by-products, on the other hand, would not
result in coverage of that unit. By-products are produced as a consequence
of producing other chemicals and are not gathered together for any subse-
quent purpose. By-products may be found as trace contaminants in the final
product of a chemical production process unit.
Comment:
Another commenter (IV-D-15) cited a potential problem in defining when
an upstream process unit is an affected facility. He cited as an example a
crude unit in a petroleum refinery which produces intermediate refinery
streams which may contain 10 percent or more VOC per the proposed ASTM
methods. These streams may feed a unit which will eventually produce a
SOCMI chemical. In such instances it was not clear to the commenter which
streams would be covered. He recommended that only the downstream unit be
covered.
Response:
The SOCMI NSPS for fugitive VOC sources regulate process units that
produce chemicals that are either photochemically reactive substances, use
photochemically reactive chemicals as additives or reactants in the
production process, or have photochemically reactive co-products. The
chemicals covered by these standards are listed in §60.489. Therefore, if
the intermediate product is one of the chemicals listed in §60.489, the unit
producing it is covered by the standards. If the downstream unit also
produces one or more chemicals on the list, it will also be covered by the
standards.
The following example is offered for purposes of clarification.
Figure 5-1 shows a configuration of a hypothetical petroleum refinery or
chemical plant. In this hypothetical complex, there are three processes.
Process X uses SOCMI and non-SOCMI raw materials to produce a non-SOCMI
chemical and non-SOCMI co-products. Process unit X would not be covered by
the final SOCMI standards because SOCMI chemicals are not produced.
Process unit Y uses non-SOCMI chemicals to produce another non-SOCMI
chemical and a SOCMI chemical co-product. Process unit Y would be covered
alf this production is less than 1,000 Mg/yr, the unit may be excluded
under the lower production volume cutoff. See Section 5.7.
5-25
-------
rv>
O\
soc
che
non
cher
MI
nical
-SOCM
nical
Process
Unit X
I
r
non-SOCMI
chemical
non-
cher
L
Process
Unit Y
-SOCMI
nical
T
SOCMI
i
SOCMI
co-product
->
1
non-
chem
pu
SOCMI SO
ical ch
L
Process
Unit Z
ch
i
rchased '
CMI non
emical che
emical
i
-SOCMI
mi cal
NOTE: Process units enclosed in dashed boxes would 'be affected facilities
under the SOCMI standards.
Figure 5-1. Example for process unit definition.
-------
by the standards because it produces a SOCMI chemical. There is an
exception to this coverage, however. If the production of the.SOCMI
chemical co-product is less than 1,000 Mg/yr, the unit may be excluded under
the lower production volume cutoff (see Section 5.7). And as discussed in
the previous response, if the SOCMI chemical is a by-product (not a
co-product) that is not collected for any purpose (such as product recovery)
and remains in the final non-SOCMI chemical, product in trace quantities as
an impurity, the unit would not be covered by the standards (see previous
response).
Process unit Z uses the SOCMI coproduct produced in Unit Y and ,
purchased SOCMI materials to produce another SOCMI chemical and a non-SOCMI
co-product. Process unit Z would be covered by the standards. However, the
non-SOCMI co-product stream would not be covered after the point of,removal
from the reaction products. . ,
Comment: ;
Another potential problem with the process unit definition was cited by
one of the commenters (IV-D-20). .He cited an example of an FCC unit which
produces streams containing significant amounts of propylene. The major
purpose of the unit is to produce gasoline. However, the propylene produced
is fed directly into a polymer plant. He did not feel it appropriate to
cover that plant by the proposed standards because the propylene is a
transient intermediate which is not stored or sold as a finished product.
Response:
The fact that a product is not stored or sold as a finished product has
no effect on fugitive emissions. A unit that produces propylene would be
covered by the SOCMI fugitives NSPS unless none of the fugitive emission
sources are "in VOC service."
In the example cited by the commenter, propylene is a co-product of the
gasoline production operations. Under the final standards fugitive emission
sources from the point of production of propylene to the point of raw
material storage or reaction in the polymer unit would be covered if they
are in VOC service.
5-27
-------
5.5 SMALL MANUFACTURERS
Comment:
One commenter (IV-F-1, No.2) objected to small manufacturers being
subject to the standards. He made three arguments in support of his
objections:
1. He said the EPA's assumption that fugitive emissions are
proportional to the number of fugitive emission sources is
invalid. He stated that he believed that even relatively complex
small plants will have fewer emissions than the larger ones.
2. Reductions in emissions from small plants would be insignificant.
3. The economic impact on small producers would be disproportionately
large when compared to emission reductions. (See also
Section 2.1).
Another comment letter (IV-D-3) said the record is lacking in data on
small chemical manufacturers.
Another commenter (IV^D-12) said expenditures required by the standards
would not be warranted for furfural units, considering the small size of the
furfural business. He said the total added expenditures during the first
year would be $109,375.00.
A commenter (IV-D-38), representing a group that had commented
previously (IV-F-1, No.2), stated that additional comment on the proposed
definitions of "small facilities" should be solicited. He used the term
''small business" as defined in the rules issued under Section 8(a) of the
Toxic Substances Control Act. He provided this definition in a subsequent
comment (IV-D-38a).
This eommenter stated that the data base for the proposed standards
consisted primarily of large continuous-process petrochemical plants. These
plants are not considered representative of the majority of the industry's
manufacturing processes, such as small or batch process units. In addition,
it is common for a batch process facility to manufacture a SOCMI chemical
for a limited time, thereafter producing a non-SOCMI chemical. Despite the
limited applicability of the NSPS to only a few products, such a manufac-
turer might incur substantial additional costs in monitoring, repairing, and
5-23
-------
recordkeeping. The commenter, therefore, felt EPA had not given adequate
consideration to such problems in applying NSPS to all industry processes.
Response:
Data from fugitive emissions test work do not show any definite rela-
tionship, between emissions and production volumes above some minimal
qualities (see Section 5.1). Fugitive,emissions are proportional to the
number of sources in a plant rather than to.the plant size or production
rate. Emission reductions from small production volume plants would,
therefore, be no smaller than those from larger ones. On the other hand,
the cost analysis shows that the cost effectiveness for the least complex
process units (model unit A) is $533/Mg VOC as compared to $252/Mg for the
most complex units (model unit C). The difference is obviously not
disproportionately burdensome even for the less complex process units. As
discussed in Section 7.2, the economic impact of the standards does not have
a significant adverse impact on small facilities.
However, there are some units (e.g., R&D facilities) which have
production rates so small that their fugitive emissions are likely to be
very small and the.cost to control fugitive emissions from such a small unit
would be unreasonably high compared to the emission reduction achievable.
Therefore, EPA has excluded from coverage by the standards units producing
less than 1,000 Mg/yr. The lower production rate cutoff is explained in
detail in Section 5.7. .
The record is not lacking data on small,manufacturers as suggested by
the commenter (IV-D-3). EPA has collected data on such statistics as total
number of SOCMI product site locations, number of emission sources versus
unit capacity, and cost estimates of installation of control devices for
small units (II-C-30). All of this information was taken into account in
the analysis of the economic impact on SOCMI. No unreasonable economic
impact on small facilities was found. EPA feels that enough time was
provided for public comment.
If a batch unit produces more, than 1,000 Mg/yr of a SOCMI chemical, it
will have VOC fugitive emissions and would be appropriately covered under
the final standards. The costs for controlling fugitive emissions of VOC in
5-29
-------
such cases would not be any more than for a unit producing a SOCMI chemical
on a continuing basis.
5.6 VOC DEFINITION
Comment:
Two commenters (IV-D-20; IV-D-6) said the definition of VOC should be
clarified. One (IV-D-20) said it. was not clear to him whether methane and
ethane would be excluded as photochemically unreactive. He said a reference
to EPA policies published in 42 FR 35314 would be helpful.
Another (IV-D-6) said the definition of VOC does not adequately reflect
the capabilities of the reference method (see Section 12, Test Method). He
further said that the definition is meaningless for those chemicals which
participate in atmospheric photochemical reactions but cannot be measured by
all instruments allowed by Reference Method 21. He recommended the
following as a substitute for the definition of VOC:
'Volatile Organic Compounds' means any organic compound,
which participates in atmospheric photochemical reactions
and is measurable by the applicable test methods described
in Reference Method 21 which can be calibrated by a
saturated straight chain hydrocarbon.
Response:
Methane and ethane were not intended to be classified as photo-
chemically reactive. The standards have been clarified to allow for the
exclusion of substances not considered photochemically reactive by EPA when
determining the percent VOC in the process fluid (i.e., determining whether
a piece of equipment is in VOC service). The VOC content is to be
determined by the referenced ASTM methods, not by Reference Method 21. As
discussed in Section 5.3 on "In VOC Service," some compounds may be excluded
from the total quantity of organic compounds contained in a process line.
These compounds are the eleven organic compounds considered at this time by
EPA to be nonphotochemically reactive.
As the commenter (IV-D-6) pointed out, no single monitoring instrument
will measure all photochemically reactive chemicals. However, each chemical
can be measured by at least one instrument. Therefore, the selection of the
monitoring instrument will depend partly on the chemicals in the line.
5-30
-------
The intent of the standards is to reduce emissions of VOC. In selecting the
basis for the standards that is, in selecting the best system of
continuous emission reduction (considering costs and other impacts) (BDT),
EPA considered VOC emission reductions and costs for three emission
reductions. However, in implementing'the controls represented by BDT, EPA '
only used the quantity of VOC to determine whether a piece of equipment
should be covered by the standards, as discussed in Section 5.3.
In contrast, the leak definition (10,000 ppmv), the criteria for
control devices, and other criteria are based on total organic compounds.
This is done to ensure that the standards reflect BDT. The data used in
selecting BDT are based on total organic compounds. Thus, while methane and
ethane are considered nonreactive but are measured by the leak detection'
monitor, they can not be subtracted from the criteria such as the
10,000 ppmv leak definition. To do so would not be consistent with
reflecting BDT.
5.7 PILOT PLANT AND R&D FACILITIES
Comment:
One set of comments (IV-D-17) recommended that research and development
facilities be exempted from applicability of the proposed regulations. The
reasoning presented was that operation of such facilities and their contri-
bution of VOC are de minimi's in nature.
Response:
Under Section 111, EPA may exempt units where the costs of the
standards are unreasonably high in comparison to the minimal emissions
reduction achievable. Thus, any exemption would be based on a cost versus
emission reduction analysis. Such an analysis indicates that units
producing less than 1,000 Mg/yr have such low fugitive emissions that the
resulting control is unreasonably high. For this reason, EPA has exempted
process units producing less than 1,000 Mg/yr.
To implement an exemption_on this basis, this potential emission
reduction would have to be translated into a usable format. Two approaches
to an exemption cutoff were considered: (1) processing rate and
(2) production rate.
5-31
-------
An exemption based on processing rate would be hard to determine. The
processing rate for a process unit is the amount of materials that move
through the process within a year, including raw materials, finished
products, intermediate materials and by-products. As a result, this rate is
not only difficult to establish but also may vary greatly.
A production rate cutoff is specified during'the design of the process
unit. It is much easier to determine since the production rate is merely
the amount of material moving out of a process in a year. The production
rate is less apt to change greatly as well. Because it provides the most
easily applied cutoff, exemption based on low production rate was chosen.
An analysis of the cost effectiveness of the standards was made with
particular emphasis on units with a low number of fugitive emission sources
(IV-B-20). At Tow equipment counts, low emission reductions are achieved -
and the cost effectiveness of the standards becomes unreasonably high. A
product of this analysis was Figure 5-2, which presents the cost effective-
ness of the standards (considering valves only) as a function of production
rate for low production volumes. The cost effectiveness of the standards
becomes unreasonable at production rates between 600 and 800 Mg/yr. There
is some uncertainty in the computations, however. Thus, EPA decided to set
a lower production rate cutoff of 1,000 Mg/yr.
The result of providing such a cutoff is to exempt smaller research
facilities directed toward research and development alone. At the same
time, those facilities that are producing chemicals on a scale that would be
considered semi commercial or commercial would be covered by the standards.
5.8 FLANGES
Comment:
One comment letter (IV-D-17) said the term "other connector" in the
phrase "flange or other connector" is excessively vague. It was said to be
impossible to tell from this term what EPA proposed to regulate. It was
requested that EPA define "other connector" more precisely or exclude the
phrase from the regulation.
5-31
-------
Reproduced from
best available copy.
3000
V*
cu
c
>
-M
OJ
M-
4-
0)
2000
1000
o
o
5GO " 1000 2000
Production rate,. Mg/yr
5000
Figure 5-2. Cost effectiveness for low volume SOCMI production
units assuming valve standards only (IV-B-20).
5-33
-------
Response:
Connectors are such items as flanged, screwed, or welded connections or
any other connectors used to connect two pipe lines or a pipe line and a
piece of equipment. This explanation has been included in the regulation
[§60.482(-8)],
Comment:
A commenter (IV-D-43) said flanges were underrepresented in the
sampling reported in EPA-600/2-81-003. He said an effort was made to test
only 20 percent of the total and even this goal could not be achieved
primarily because of inaccessibility. Thus, the problems of real life in a
chemical plant were demonstrated.
Response:
It is precisely because of the large number of flanges in a facility,
some of which may be difficult to access, that flanges are generally
excluded from specific coverage under the standards. EPA has made every
attempt to take into consideration such problems in an effort to make the
regulation as reasonable as possible. There is, however, limited coverage
of flanges under §60.482-8 of the standards that requires monitoring and
repair if evidence of a potential leak is found.
5.9 VACUUM SERVICE
Comment:
In one comment letter (IV-D-17) it was stated that the criteria for
determining vacuum service (100 kPa) was obviously not intended. The
proposed definition was interpreted to mean that when the ambient pressure,
(barometer) is below 100 kPa, atmospheric vented tanks are in vacuum
service. It was recommended that a better definition .would be when a
fugitive emission source is operating at an internal pressure which is
continuously 200 kPa or more below ambient pressure.
Response:
The proposed regulation defines a source to be in vacuum service if it
is operating at an internal pressure which is continuously less than
100 kPa. 'It should be noted that 1 atmosphere equals about 100 kPa. Many
sources may be operating at a pressure below 100 kPa. However, if the
5-34-
-------
source's internal pressure is less than 200 kPa below the ambient pressure,
it will not be classified as being in vacuum service by the commenter's
recommended definition. That, in EPA's judgement, would be inappropriate
because vessels operating even at a slight vacuum would have little if any
potential to emit VOC. When asked for a clarification of the above comment,
the commenter indicated that his comment was in error and that he had not
realized that 100 kPa is atmospheric pressure. Therefore, to avoid any
further misunderstanding about the standard, the definition for vacuum
service has been changed as follows:
'In vacuum service1 means that a fugitive emission source is
operating at an internal pressure which is at least 5 kPa
below ambient pressure.
5.10 ENCLOSED BUILDINGS
Comment:
One commenter (IV-D-51) wrote that OSHA regulates fugitive emissions
released inside totally enclosed buildings by stipulating the maximum
exposures permitted in the workplace. The act also calls for engineering
controls to correct problems of overexposure. Breathing apparatus is only
permitted as a corrective measure if engineering controls are not feasible.
The commenter stated that additional regulations to control fugitive
emissions within the workplace are unnecessary.
Response:
The OSHA and NSPS regulations do not have identical objectives.
However, the SOCMI NSPS does not impose any duplicative requirements. They
only supplement the OSHA standards. As such it would be improper to exclude
enclosed buildings. The respective coverage of OSHA and NSPS regulations is
discussed in greater detail in Section 2 on the Basis of the Standards and
in Section 5.1 on the SOCMI list.
b-35
-------
-------
6. ENVIRONMENTAL IMPACT
Many comments on the proposed standards for VOC fugitive emissions from
SOCMI addressed emissions estimates and emission reductions achievable under
the standards. Such comments appear throughout this document. EPA's
analysis of emissions and emission reductions achievable under the final
standards are presented in this section.
The development of emission factors and model units, as discussed in
Chapter 3, was detailed in the Additional Information Document (AID) for
fugitive VOC emissions (III-B-2). Also presented in the AID was an analysis
of emission reductions achievable under various control options (e.g.,
installation of equipment or implementation of work practices such as leak
detection and repair programs). Discussion and responses to comments
received on the AID are presented in Appendix A of this document.
The analyses presented in Chapter 3 and in the AID were made for
individual fugitive emission sources (single equipment components) only. In
this section the emissions analyses aggregate fugitive emission sources into
model units (see Section 3.2) and examine the overall impact of the
standards on process unit-wide fugitive emissions. Model unit emissions are
then extrapolated to the national level to estimate national impacts.
6.1 EMISSIONS ANALYSIS
Two parallel analyses of SOCMI fugitive emissions are presented in this
section. One analysis presents estimates of emissions and estimates of
emission reductions under the standards using the model units, emission
factors, and estimated control efficiencies presented in the AID. The other
analysis estimates emissions and emission reductions achievable for three
types of SOCMI units tested (ethylene, cumene, vinyl acetate) during the
Maintenance Study (IV-A-10).
Estimating the impact of the standards on emissions from SOCMI units
and from SOCMI nationwide requires the following steps:
6-1
-------
(1) development of emission factors for fugitive emission sources (see
Section 3.1 of this document and Section 2 of the AID);
(2) development of model units (see Section 3.2 of this document and
Section 3 of the AID);
(3) evaluation of the effectiveness of control techniques (see
Section 3.3 of this document and Section 4 of the AID);
(4) aggregation of emissions (with and without standards) by model
unit;
(5) projection of model unit emissions (with and without standards) to
the anticipated growth in the industry.
The development of emission factors for fugitive emission sources in
SOCMI was detailed in the AID. Briefly, the technique uses emission factors
generated in petroleum refining and SOCMI studies for leaking and non-
leaking equipment and the leak frequency determined for each type of equip-
ment during SOCMI screening studies. For the purposes of the parallel
analysis of emissions from the three types of SOCMI units tested, the
emission factors generated for three equipment types (gas valves, light
liquid valves, light liquid pumps) in the SOCMI studies were used. Emission
factors for the remaining fugitive emission sources were estimated using the
approach detailed in the AID. The emission factors are shown in Table 6-1.
The model units used to describe SOCMI were also presented in the AID.
The equipment counts for the model units.have not been changed since
proposal, but some revisions have been made to clarify some confusion and to
account for the current level of control in SOCMI. These equipment counts
are presented for model units A, B, and C in Table 6-2. Also presented in
this table are the average equipment counts found in the ethylene units,
cumene units, and vinyl acetate units tested. These average equipment
counts are used in developing estimates in the parallel.analysis of the
SOCMI units tested.
Estimates of emissions from model units are merely the aggregation of
the emissions due to the various equipment components in a process unit.
These estimates are generated by applying the emission factor for a single
component to the total equipment counts and extrapolating emissions to a
6-2
-------
TABLE 6-1. EMISSION FACTORS FOR SOCMI EMISSIONS ANALYSES
Emissions
Source
Pui.i|) Seals
Light liquid
Heavy liquid
Valves
Gas
Light liquid
llt-avy liquid
Pressure Relief Device
Gas
Open-ended lines
Compressors
|Sampl Ing
connections
Flanges
SOCHI
Leak
Frequency,
I
8.8
2.1
11.4
6.5
0.4
s
3.6
3.9
9.1
2.1
Emission
Factor,
kg/hr
0.0494
0.0214
0.0056
0.0070
0.00023
0.1040
0.0017
0.228
0.0150
0.00083
Ethyl ene
Leak
Frequency,
*
*
0
*
*
l.l
3.9
12. B
5.9
5.7
Emission
Factor,
kg/hr
0.058a
0.0135
0.0086a
0.0183
0.00023
0.1092
0.0028
0.179
0.0150
0.0022
Cumene
Leak
Frequency ',
%
*
0
*
*
0
<3.6)b
9.1
(9.1)b
2.9
Emission
Factor,
kg/hr
0.01B3
0.0135
0.007a
0.0063
0.00023
0.1040
0.0025
0.228
0.0150
0.0011
Vinyl Acetate
Leak
Frequency ,
, *
*
0
*
*
0
(3.6)b
3.7
0
1.0
Emission
Factor,
kg/hr
0.002a
0.0135
0.0014a
0.00023a
0.00023
0.1040
0.0019
O.OB94
0.0150
0.00043
en
i
OJ
Emission factors determined from results of 6-un1t SOCMI Maintenance Study.
Twenty-four unit leak frequency assumed where insufficient or biased data existed.
-------
TABLE 6-2. EQUIPMENT COUNTS USED IN ESTIMATING SOCMI UNIT EMISSIONS
Emissions
Source
Pumps
Light Liquid
Heavy Liquid
Valves
Gas
Light Liquid '
Heavy Liquid
Aa
8
7
99
131
132
Ba
29
30
402
524
524
Pressure Relief Devices
Gas 3C 11C
Open-ended Lines
Compressors
Sampling Connections
Flanges
104
1
7
600
415
2
26
2400
ca
91
93
1232
1618
1618
33C
1277
8
80
7400
Unit Equipment
Ethyl ene
29
8
2494
1854
683
16C
165
5
41
10660
Counts
Cumene
13
2
273
464
124
3C
11
1
3
1933
Vinyl Acetateb
47
3
625
1179
64
5C
278
4
70
1685
SOCMI model units as presented in the SOCMI AID.
Average equipment counts of the units tested in the Maintenance Study by process type.
The seventy-five percent of safety/relief valves in gas service EPA assumed to be controlled at
baseline (tied into flare header) are not included in these numbers.
All open-ended lines assumed to be controlled in the absence of standards.
-------
full year (8760 hours). An example of this procedure 1s shown in Table 6-3.
A summary of unit emissions for the three model units (A, B, and C) and for
the three SOCMI unit types tested (ethylene, cumene, and vinyl acetate) are
presented in Table 6-4.
Emissions reduction estimates have been previously discussed in
Section 3 of this document and in Section 4 of the AID. The efficiency
estimated for each type of equipment depends upon the type of standard
applicable to that type of equipment. For instance, the equipment standards
*
for sampling connections, compressor seals, and open-ended lines are
assumed to be 100 percent effective in eliminating fugitive VOC. Leak
detection and repair programs are more cost-effective alternatives for other
fugitive emission sources (such as valves and pumps), but have lower
efficiencies associated with them. The efficiencies used in evaluating the
overall effectiveness of standards are summarized in Table 6-5.
These efficiencies are applied to the emissions estimated for SOCMI
sources in the absence of any fugitive emissions program. Baseline
emissions, emissions under the standards, and emissions reductions are
summarized for the SOCMI model units in Table 6-6. The overall percent
reduction in fugitive VOC emissions is about 56 percent when computed on a
model unit basis.
The nationwide impact of the standards in the fifth year after proposal
was determined by applying the growth projection for SOCMI to the model
units. The same model unit projections presented in the BID, based on a
5.9 percent growth rate and replacement based on a 20-year equipment life,
were used in this analysis of nationwide impacts. The same percentages of
model units (52% A, 33% B, 15% C) that were used before proposal were
applied to the 831 new, modified, or reconstructed units for this analysis.
As shown in Table 6-7, the total VOC emission reduction estimated for the
fifth year after implementation of the standards is about 46 gigagrams.
*Seals can achieve at least 95 percent emission reduction and the control
device can achieve greater than 95 percent efficiency, for a combined
efficiency of at least 99.7 percent. In addition, compressor seals are
typically vented back to suction. Considering all these factors, an
assumed efficiency of 100 percent for compressor seals is reasonable.
6-5
-------
TABLE 6-3. EXAMPLE OF EMISSIONS ESTIMATED FOR MODEL UNIT B
IN ABSENCE OF STANDARDS
Emissions
Source
Pump seals
Light Liquid
Heavy Liquid
Valves
Gas
Light Liquid
Heavy Liquid
Pressure Relief Devices
Gas
Open-ended lines
Compressors
Sampling connections
Flanges
Total
Number of
Sources
29
30
402
524
524
11
415
2
26
2400
Emission Factor,
kg/hr/source
0.0494
0.0214
0.0056
0.0070
0.00023
0.1040
0.0017
0.228
0.0150
0.00083
Annual Emissions, Mg/yr
12.55
5.62
19.72
32.59
1.06
10.02
<
6.91
3.99
3.42
17.45
106.42
For estimating purposes, one operating year was assumed to be 8760 hours.
-------
TABLE 6-4. SUMMARY OF ESTIMATED EMISSIONS IN THE ABSENCE
OF STANDARDS BY SOCMI UNIT
en
i
Unit Emissions Estimates,
Emissions
Source
Pump seals
Light Liquid
Heavy Liquid
Valves
Gas
Light Liquid
Heavy Liquid
Pressure Relief Devi
Gas
Open-ended lines
Compressors
Sampling connections
Flanges
Total
A
3.46
1.31
4.86
8.15
0.27
ces
2.73
Ob
2.00
d 0.92
4.36
28.06
B
12.55
5.62
19.72
32.59
1.06
10.02
Ob
3.99
3.42
17.45
: 106.42
C
39.38
17.43
60.44
100.63
3.26
30.06
Ob
15.98
. 10.51
53.80
331.49
Vinyl
Acetate
0.82
0.35
7.67
2.38
0.13
4.56
4.63C
3.13
9.20
6.35
39.22
Mg/yra
Cumene
2.05
0.24
16.74
24.39
0.25
2.73
0.24C
2.00
0.39
18.63
67.66
Ethylene
14.73
0.95
187.89
292.34
1.38
15.31
4.05C
7.84
5.39
205.44
735.32
For estimating purposes, one operating,year was assumed to be 8760 hours.
Nearly all open-ended lines were assumed controlled at baseline for the model units.
GEmissions estimates are based on no control in the absence of standards.
The equipment count for sampling connections is taken as 25 percent of the open-ended line count.
-------
TABLE 6-5. SUMMARY OF STANDARDS AND ESTIMATED EFFICIENCIES FOR
NEW SOURCES OF FUGITIVE VOC EMISSIONS IN SOCMI
Equipment
Pumps - Light Liquid
Valves - Gas
- Light Liquid
Pressure Relief Devices
Open-ended Lines
Sampling Connections
Compressors
Type of Standard
Work Practice*
Equipment
Work Practice^
Work Practice
Performance
Equipment
Equipment
Equipment
Estimated Efficiency
0.61
1.0
0.73
0.59
1.0 "
1.0
1.0
1.0
The work practice standards are monthly leak detection and repair
programs. Efficiencies were computed using the LDAR model described
in IV-A-22.
Seals can attain 95 percent efficiency and the control device can attain
at least 95 percent efficiency, for a combined efficiency of at least
99.7 percent. Thus, the efficiency for the combination was assumed to be
1.0.
The performance level for safety/relief valves is no detectable emissions
and the estimated efficiency of 1.0 is based on equipment (e.g., rupture
disks) being used.
-------
TABLE 6-6. SUMMARY OF EMISSIONS ESTIMATES FOR AVERAGE
SOCMI UNITS IN MG/YR
Ch
I
Source
Pumps
Light Liquid
Heavy Liquid
Valves
Gas
Light 1 iquid
Heavy Liquid
Regulated
By
Work Practice
Work Practice
Work Practice
Pressure Helief Devices
Gasa Performance
Open-ended lines
Flanges
Sampling connections
Compressors
TOTAL
% Reduction
Equipment
_
Equipment
Equipment
Baseline Emissions
ABC
3.46
1.31
4.86
0.15
0.27
2.73
0
4.36
0.9?
2.00
28,06
12.55
5.62
19.72
32.59
1.06
10.02
0
17.45
3.42
3.99
106.42
39.38
17.43
60.44
100.63
3.26
30.06
0
53.80
10.51
15. 9B
331.49
Control
Efficiency
0.61
0.73
0.59
1.0
1.0
1.0
1.0
Controlled Emissions
A B C
1.36
1.31
1.31
3.34
0.27
0
0
4.36
0
0
11.95
4.92
5.62
5.32
13.36
1.06
0
0
17.45
0
0
47.73
15.44
17.43
16.32
41.26
3.26
0
0
63.80
0
0
147.51
Emissions Reduction
ABC
2.10
0
3.55
4.81
0
2.73
0
0
0.92
2.00
16.11
57
7.63
0
14.40
19.23
0
10.02
0
0
3.42
3.99
58.69
55
23.94
0
-14.12
59.37
0
30.06
0
0
10.51
15.98
183.98
56
aThis estimate assumes 75 percent of the pressure relief devices (gas service) are controlled (e.a., tied into a flare header) in the absence
of standards.
As discussed in ttie AID and in Section 3 of this document, nearly all open-ended lines are assumed to be controlled at baseline. If
90 percent of the open-ended lines are assumed to be controlled at baseline, the baseline emissions (and emission reductions) in Mg/yr
would be as follows: A. 0.15; B, 0.61; 1.91.
-------
TABLE 6-7. NATIONWIDE IMPACT OF SOCMI NEW SOURCE STANDARDS FOR
FUGITIVE VOC EMISSIONS IN THE FIFTH YEAR
Model Unit Number of Units Emission Reductions, Mg/yr
Per Unit Total
A 432 16.1
;B 274 ' 58.7
C 115 184.0
831
6-10
-------
6.2 ENVIRONMENTAL IMPACT:
Several comments were received regarding the environmental impacts of
the standards in general. The comments ranged from concern over estimates
of fugitive VOC emissions in SOCMI and the potential benefits of achievable
emissions reductions to questioning the impacts of other regulatory
programs, such as SIP., PSD, and LAER. Overlapping coverage of standards,
such as RCRA incinerator requirements, also received comment.
Comment:
Several commenters objected to the estimation of emissions from SOCMI
made by EPA, based on emission rates from petroleum refinery equipment. One
commenter (IV-D-7) said this estimation method was not justified since
studies have now been completed for SOCMI. Another (IV-D-6) said emission
rates are lower than expected from application of refinery data. Another
(IV-D-28) said the background document has no data on emissions for sources
except equipment data from petroleum refiners., He said fugitive emissions
estimates for SOCMI based on petroleum refineries were wrong because the
refining industry is not a proper model for SOCMI.
Jtes_p_ons_e_:
EPA's approach to quantifying fugitive VOC emissions from SOCMI is to
use the best fugitive emissions data, available. At the time of proposal.,
the best available data for SOCMI fugitive emissions were data from
petroleum refineries. As noted in Section 6.1 and explained in the AID, the
estimates of environmental impacts of the standards have been revised. The
previous estimate of 200 Gg/yr uncontrolled emissions from SOCMI facilities
which, will become affected by the standards through 1985 has been revised to
83 Gg/yr.
Comment:
There was disagreement among the commenters concerning.the percentage
of VOC emissions contributed by SOCMI fugitives. Two commenters (IV-D-17; .
IV-D-48) cited information from the Background Information Document which ,
indicated that SOCMI fugitives contributed 2 percent of stationary VOC
emissions. One of these two commenters (IV-D-17) indicated some .confusion
over the numbers and said the percentage should actually be 1 percent.
6-11
-------
Another estimate of 5 percent was quoted from the preamble to the proposed
regulations (IV-D-21). The estimate was also said to be 0.5 percent
(IV-D-17; IV-D-48) because of an apparent error on EPA's part. The
commenters said that EPA had estimated that fugitive emissions from 1000
existing plants total 200 Gg/yr and that 800 new plants would contribute
200 Gg/yr. The commenters saw this as an obvious-error and concluded that
estimates of current contributions of fugitive emissions were obviously
overstated by a factor of 2.
Response:
Percentages can often be confusing because they require consideration
of the number being compared as well as the base to which it is being
compared. The most recently available total for VOC emissions from
stationary sources is 17,000 Gg/yr (IV-A-26). A recent estimate for VOC
emissions from SOCMI (all sources) is 540 Gg/yr with fugitive emissions
contributing about 35 percent of the total or about 190 Gg/yr (IV-B-24).9
Comparing the fugitive emission contribution to VOC emissions from
stationary sources yields roughly 2 percent. Comparing VOC emissions from
all SOCMI sources to total stationary source VOC emissions yields roughly
5 percent.
Another confusing aspect of percentages of emissions is the fact that
the numbers are not static. Emissions change from year to year, most
recently in a downward trend. The numbers presented at proposal were based
on earlier emissions estimates of 19,000 Gg/yr VOC emissions from all
stationary sources, 1000 Gg/yr VOC emissions from SOCMI, and 400 Gg/yr from
fugitive emission sources in SOCMI. Comparisons of these numbers yielded
roughly a 5 percent contribution of SOCMI VOC emissions to stationary source
emissions of VOC and a 2 percent contribution of SOCMI fugitive emissions of
VOC to total stationary source emissions of VOC. Even though the numbers
have changed somewhat since proposal, the percentages have remained
essentially constant.
Previous estimates were based on petroleum refinery emission factors
(IV-A-19). Using the refinery emission factors, fugitive emissions
from SOCMI were estimated to be about 320 Gg/yr, or about 40 percent
of the total 800 Gg/yr estimated for all sources in SOCMI.
6-12
-------
The source of the error pointed out by two of the commenters came from
two industry-wide equipment estimates presented in a draft survey document
of fugitive emissions from .SO'CMI. One of the estimates presented in the
draft document was based on a direct industry scale-up of equipment counts.
This scale-up ,of equipment counts resulted in estimated emissions of
200 Gg/yr; the estimate was discarded since the scale-up was performed
incorrectly and was, therefore, removed from the final document. The other
estimate (400 Gg/yr) was based on the same model unit approach employed in
the BID and AID analyses to estimate number of pieces of equipment. This
approach in estimating emissions is retained in the final document.
Comment:
One commenter. (IV-D-28) expressed some confusion over .what the SOCMI
VOC emissions estimates made by EPA actually are. He said the preamble
gives total annual stationary source VOC emissions as 19,000 Gg and those .'
from SOCMI as 1,000 Gg, or about 1 million tons, for some unspecified year.
He cited another EPA report "Cost and Economic Impact Assessment for
Alternative Levels of the NAAQS for Ozone" (June 1978 draft), which shows
in Table 3-1 a total of 18.6 million tons from 90 nonattainment ACQR's for
1975 with chemical manufacturing responsible for 0.43 million tons.
Environmental Outlook 1980 does not list the chemical industry in Table 4-8
of "Principal Sources of Net Hydrocarbon. Emissions" for 1975 with a total of
13.5 million tons. "Other sources" have a value of 3.5 million tons.
Report EPA 600/8-78-004. gives in Table 5-10 a total for 1975 of 28 million
tons, and chemicals as 1.5. Industrial processes total 3.2. Report OPA
48/8, June 1979, agrees that the total is 28 million tons, but industrial '
processes are responsible for 10.1, or 36 percent. Table 6-3 of "Facts and-
Issues Associated with the Need for a Hydrocarbon Criteria Document,"
February 1980 External Review Draft, agrees with a total of 28.3, and indus-
trial contribution of 10.1, but puts the chemical industry at 2.7 million
tons/yr. Thus, he said that he had some problem with determining the Agency
position on the amount of emissions under consideration.
6-13
-------
Response:
The commenter is correct in stating that several different estimates
appear in the literature. These different estimates are largely due to
EPA's frequent updates using emissions data. They are also in some measure
due to different calculation or modeling methods. It should also be kept in
mind that the numbers are estimates, not absolute measurements.
The numbers cited in the preamble were from an EPA report on the
chemical industry (II-A-22). The total VOC emissions number has since been
revised downward to 800 Gg/yr in a subsequent edition of the report.3
This number compares very closely with the latest published estimates (1977)
cited by the commenter.
Two of the references cited by the commenter present the same 1977
estimates of VOC emissions: Cleaning the Air (OPA 48/8) and Facts and
Issues Associated With the Need for a Hydrocarbon Criteria Document
(External Review Draft). These are the most current of the estimates cited
and are probably the most reflective of today's emission levels. It should
be noted that SOCMI is only a part of the chemical industry classification
listed in these references.
Environmental Outlook 1980 contains earlier projections of emissions
made by the SEAS model. The report explains on page 99 that a major
discrepancy exists between these projections and those made by NEDS. The
report states that the SEAS estimates are about one-third the NEDS
estimates. It goes on to explain that the single most important contributor
to this difference is the fact that SEAS does not account for evaporative
losses of solvent. The report states, "as a result of these discrepancies,
forecasts of hydrocarbon emissions by SEAS are probably low in each projec-
tion year."
The estimates of total VOC emissions from sources in SOCMI have been
revised recently according to the best information available from the
various standards development programs. The new estimate is about
540 Gg/yr for 6 source groups in SOCMI (IV-B-24).
G-14
-------
The estimates presented in Air Quality Criteria for Ozone and Other
Photochemical Oxidants (EPA-600/8-78-004) are siightly older estimates for
1974 and 1975 and, therefore, are not considered as .valid as the more recent
ones.
The estimates contained in Cost and Economic Impact Assessment for
Alternative Levels of the National Ambient Air.Quality Standards for Ozone
(EPA-450/5-79-002), as the commenter pointed out, are estimates for 90 AQCRs
and are, therefore, not comparable to national estimates.
Comment: . ... . . .
EPA's estimates of fugitive emissions of VOC contributed by SOCMI .were
said to be overstated (IV-F-1, No.l; IV-D-23; IV-D-21; IV-D-17; IV-D-7;
IV-D-50; IV-D-38; IV-D-24; IV-D-15). Several commenters said more recent
SOCMI data indicate a more realistic percentage of stationary source VOC of ;
0.25-0.33 percent (IV-D-17; .IV-D-21; IV-D-23; IV-D-50; IV-D-20). Another
commenter (IV-D-38) said the VOC emissions from SOCMI were actually
45 percent of the estimate EPA made at proposal. .Two of the commenters
(IV-D-50; IV-D-24) said the emissions estimates are still further overstated
because they are based on existing units. He said that for regulatory and
economic reasons, new plants have lower emissions than the existing units
sampled.
Response:
EPA's estimate of fugitive emissions of VOC contributed by newly
constructed, reconstructed, or modified process units within SOCMI at
proposal was 200 Gg/yr. This number has been revised to 83 Gg/yr (see
Section 6.1).
The more recent SOCMI data referred to by the commenters is assumed to
consist of the SOCMI maintenance report (IV-A-10) and the SOCMI 24-Unit
Study (IV-A-11). The basis for their percentages is unclear, but it is
probable that the commenters are comparing leak frequencies or emission
factors to those presented in the BID (III-B-1). EPA's latest estimate of
83 Gg/yr reflects this new SOCMI data and represents a decrease of .
79 percent over the original estimate. (See Section 3.1 and the AID for a
discussion of comparisons of petroleum refinery and SOCMI data.)
6-15
-------
Hopefully, the commenter is correct in stating that the emissions are
overstated because they are based on existing units and new units will be
cleaner. Unfortunately, it is not possible to measure emissions from units
which have not yet been built. Data from existing units represents the best
data available. These data were col-lected during the' recent testing program
at SOCMI process units. Thus, the data should reflect the impact of current
regulations and economic environments.
Comment:
Another comment received concerned emission reduction estimates made by
EPA. One commenter (IV-D-15) wrote that the overriding concern with the
proposed standards concerns the doubtful validity of the emission reductions
and economic benefits which are claimed. He said EPA had taken uncontrolled
fugitive hydrocarbon emissions estimates which were originally derived from
a survey of similar process equipment used in refineries, applied
"guesstimated" reduction factors relating to the effectiveness of either
specific control equipment or monitoring and maintenance programs, and
predicted that these will result in an 87 percent reduction in fugitive VOC
emissions from new and modified units. The commenter stated that the claim
that the standards will reduce the total emissions from 200,000 metric tons
to 26,000 metric tons over the next five years in questionable. He asserted
that at this point no quantitative fugitive emission source data exist to
make such claims. Another commenter (IV-D-7) said the reduction in
emissions will be less than that assumed by EPA.
Response:
The commenters are critical of the methods and results obtained in
making estimates of emission reductions for SOCMI under the standards. They
have not offered a better method, however. The data and methods used to
develop estimates of emission reductions achievable under the standards are
described in the AID and in Docket Item No. IV-A-22.
Comment:
One commenter (IV-D-18) alleged that insufficient development and study
has gone into the regulatory package as proposed. He questioned whether the
regulation in present form will provide any significant environmental
6-16
-------
benefit. The commenter indicated that increases in emissions will result
from:
A. Purging of lines so that a valve may be accessed. (It is
physically impossible to purge all lines to a combustion
device.) Replacement of valves during a non-turnaround
period will cause emissions which are l.ikely to exceed
any benefit achieved.
B. Closed sampling systems will require at least four valves
to operate in most instances. The emissions from these
valves will exceed the emissions from the sampling system,
according to the emission factors.-
C. Damage to valves will occur due to overtightening, shearing,
or otherwise destroying the packing. In a typical chemical
facility, if replacement of the valve would mean a process
shutdown and the leakage from the valve does not present a . '-''
hazard, the valve would be allowed to remain in operation,
thereby increasing emissions. Several "unrepairable" valves
could conceivably negate any benefit the proposed regulation
could have had. .
Response:
As shown in Section 6.1, the final regulations for SOCMI fugitive VOC
emissions will indeed result in significant environmental benefit. The
commenter, however, presented three comments for why emissions could
increase as a result of the new source standards, as proposed. Each comment
is addressed separately below.
The proposed standards for valves required that repair be made to a
leaking valve that could be isolated without process unit shutdown. EPA
recognized that some valves, although capable of being isolated, may cause
emissions resulting from purging that could outweigh the benefits of repair
to that valve. Therefore, a specific exemption was added to the standards
to allow valves that could be isolated but that would require purging to the
extent that resultant emissions would exceed benefits of repair to delay
repair until the next shutdown. Prior to repairing such valves at the
shutdown, the purged material from preparing the valves for repair must be
"controlled" in that the purged material must be collected arid disposed of
properly in a control device. The purged material (gas or liquid) should be
6-17
-------
collected in a suitable container for transport to the control device or
should be conveyed through piping or ductwork to the control device for
disposal. Such practices may include hooding ("bagging") of the equipment
component(s) being purged for repair.
There will be cases where the valve can be isolated and repair effected
without the extensive purging and emissions increases described above. An
example of this situation would be the addition of a packing ring to a valve
while it is in place. In these cases, the exemption would not apply and
repair must be effected as prescribed in the standards.
Emissions from an uncontrolled sampling apparatus were estimated based
on the potential purge taken before extracting the sample. The analysis of
emissions, therefore, considered the sampling apparatus alone; any valves
that might be associated with a sampling system were considered in the total
valve count, not as part of the sampling apparatus. If a closed purge
sampling system is used, there may be four valves associated with the
sampling system, but only two of the valves are considered to be in VOC
service. For usual sampling systems, a single valve is considered to be j_n
VOC service. Thus, only one additional valve in VQC service would be needed
to comply with the standards. The uncontrolled emissions from a single
valve (0.0056 kg/hr for a gas valve; 0.0071 kg/hr for a light liquid valve)
are;less than the captured purge from sampling (0.015 kg/hr). Furthermore,
the additional valves in VOC service associated with closed-purge sampling
systems would be subject to the standards for valves (achieving some degree
of emissions reduction) and would represent lower mass emissions than those
for uncontrolled valves. Controlled emissions per valve are estimated to be
0.0015 kg/hr for gas valves and 0.0029 kg/hr for light liquid valves under a
monthly leak detection and repair program.
The emissions increase associated with "unrepairable" valves was not
directly accounted for in the original emissions estimates. It was
indirectly accounted for in the B-factor estimate used in the ABCD model.
Unrepairable valves are not necessarily valves that would, cause operating or
safety problems due to high leakage or failure. Such valves would probably
be repaired without the standards. Unrepairable valves, rather, are valves
6-18
-------
for which attempted repair as defined in the standards repeatedly fails.
The results of the Maintenance Study (IV-A-10) show that even with
unsuccessful repair (repair attempts failed to reduce the screening value
below 10,000 ppmv), emissions from valves were reduced by 63 percent from
the uncontrolled level. These emissions were included in the analysis
presented in Section 6.1. The standards were shown to demonstrate an
emissions reduction of 64 percent or greater for valves only, and about
56 percent for the model unit overall even when unrepairable valves were
included.
Comment:
One commenter (IV-D-17) was concerned that the achievable emissions
reductions were based on maintenance frequency, emission rates, and leak
occurrence/recurrence rates. Changes in these rates for SOCMI will impact ,
emissions reduction and cost effectiveness analyses. The commenter also
pointed out that no assessment was made of mass emissions from equipment
that could not be repaired within a given interval.
Response:
The analysis presented at proposal was based on assumptions and data
available at that time. The analysis has been reconsidered using data
collected on SOCMI units during screening and maintenance studies (IV-A-10;
IV-A-11; IV-A-14). The results of this revised analysis, as detailed in
Section 6.1, indicated an overall emissions reduction of about 56 percent.
This analysis estimated the efficiency of leak detection and repair programs
for valves and pumps, in accordance with the final standards. Emissions
from valves which could not be repaired on-line were included in the
original estimates presented at proposal. These emissions were also
included in the latest emissions analysis. Also, where a repair interval is
specified in the regulation, half of the interval was assumed as the time
during which emissions occurred.
Comment:
Two commenters (IV-D-17; IV-D-18) disliked the assumptions that EPA
made for predicting the effects of a leak detection and repair program for
valves. The assumptions the commenters disagreed with or thought illogical
6-19
-------
include the leak occurrence/recurrence rate (IV-D-17), the monitoring time
requirements (IV-D-18), and the ABCD factors for the valve model printed in
the BID (IV-D-17).
Response;
The technical substance of this comment is treated in detail-in the --
AID (III-B-2).
Comment:
One commenter (IV-D-28) wrote that the impact on the air quality has
not been spelled out. He noted that no estimate is given for reduction in
ozone generation. The only reference is that an 87 percent emission
reduction will be obtained from new and modified sources compared to
emissions which would occur in the absence of the regulation. The commenter
added that this estimate is incorrect because it does not consider LAER, .
BACT, or SIP.
Response:
The concept of controlling VOC emissions to control ozone generation
has been addressed previously in Chapter 2. VOC emissions have been identi-
fied as precursors to the formation of ozone and other oxidants which result
in adverse impacts on health and welfare (IV-A-17). At no point in this
standards-setting process has EPA attempted to relate quantitatively the
emissions of VOC which would be affected to the resultant air quality.
Under Section 111, it is sufficient for the purposes of new source standards
to aim at preventing degradation of air by new sources by controlling VOC to
the level achievable by the best demonstrated technology, knowing that VOC
emissions contribute to oxidant formation.
At proposal, an estimated 87 percent reduction in VOC emissions was
presented for SOCMI under the regulations. Based on the final regulations,
and considering SOCMI'data, the SOCMI standards resulted in an estimated
56 percent reduction in fugitive VOC emissions.
The emissions and emissions reductions achievable by the standards do
consider the effects of other factors, including other regulations. The
benefits of other regulations were considered in establishing the baseline
level of control, or the actual level of control existing in the industry.
6-20
-------
Regulatory circumstances, such as SIP (for facilities in nonattainment
areas), NESHAP (for facilities processing vinyl chloride and benzene), and
OSHA, already affect the levels of emissions control practiced in the
industry. These other regulatory programs are discussed further in
Chapter 2. Other circumstances also have an effect on the levels of control
in the industry. These include the provisions of insurance policies for
fire and explosion protection and the economics of recovering products made
more valuable by increases in prices for petroleum-derived products. EPA
has considered these circumstances in establishing baseline emission
estimates.
Comment:
One commenter (IV-D-26) said EPA's Office of Solid Waste's incinerator,
standards (40 CFR 264 Subpart 0) may well apply to "enclosed combustion
devices" such as those required in the proposed NSPS. The permitting
requirements, he said, would also apply, as would required test burns and
the achievement of a destruction and removal efficiency of 99.99 percent.
The application of RCRA requirements to combustion devices and similar units
under other air quality regulations were said to be most detrimental to
placing such equipment in service. He urged OAQPS to seek a general class
exemption for this type of equipment.
Response:
The commenter appropriately pointed out a potential overlap in the
standards. Some compounds on the SOCMI list are also designated as
hazardous materials under RCRA (Resource Conservation and Recovery Act).
RCRA covers hazardous materials up to their final disposal, and, where an
incinerator is used, a destruction and removal efficiency of 99.99 percent
must be met for all principal organic hazardous constituents.
An example of overlap would be coverage of discarded commercial
chemical products, off-specification species, and container residues by
RCRA; in some instances, material collected during sampling could be
considered under both RCRA and SOCMI standards. In this case, the require-
ments for sampling systems under SOCMI standards might result in a small
total quantity of hazardous material that might need to be disposed of in
6-21
-------
compliance with the requirements of RCRA. When this overlap occurs,
compliance with RCRA requirements will already be needed. Thus, the addi-
tion of a small quantity of material for disposal would not be unreasonable.
By meeting the disposal requirements of RCRA, particularly by incineration,
the requirements of the SOCMI standards would also be met and even
surpassed.
6-22
-------
7. ECONOMICS
This section discusses the comments received on the cost and economic
analyses conducted in support of the standards. The cost methodology and
cost estimates by piece of equipment were discussed in detail in the AID and
are summarized here in Section 7.1. Comments on the AID and EPA's responses
to those comments are presented in Appendix A. Comments related to the
economics methodology are presented in Section 7.2. The economic impact of
the standards is presented in Section 7.3. Comments on the cost benefit
considerations are addressed in Section 7.3 and in Section 3.1 (Selection of
Final Standards).
7.1 COST ESTIMATES ; ;
Several comments were received concerning various aspects of the
estimated costs of the control options for SOCMI. These comments concerned
the general methodology, the results of the analysis, and specific input
data used to evaluate the control techniques considered in selecting the
proposed standards. EPA reviewed these factors and made changes in certain
circumstances as explained in the AID.
In the AID, EPA reviewed the cost estimating techniques (cost metho-
dology) used in the BID and found them valid. After reviewing comments on
the AID, EPA continues to conclude that the general cost methodology used in
the BID is valid. In the AID, EPA also reviewed specific input data for the
various control techniques. EPA concluded that, while most of the data
contained in the BID was correct, some cost input data should be revised to
reflect comments on the BID and data gathered during the SOCMI 24-Unit
Screening Study and the SOCMI Maintenance Study. These input data were
changed accordingly and are discussed in detail in Section 5 of the AID for
each emission source.
Emission source costs are aggregated into model unit costs by using the
equipment counts for each model unit (IV-B-31). Model units are presented
aIV-D-l; IV-D-2; IV-D-6; IV-D-13; IV-D-15; IV-D-17; IV-D-18; IV-D-20;
IV-D-28; IV-D-46; IV-D-50; II-E-20.
7-1
-------
in Section 3.3. Table 7-1 presents a summary of the estimated capital costs
of the standards for the model units. Table 7-2 presents a breakdown of the
annualized cost estimates by model unit. Using the growth projections
presented in the BID and used in Section 6 to project nationwide emissions
reductions, capital and annualized costs to the industry were projected for
the fifth year. The cumulative capital costs and the annualized costs in
the fifth year are summarized in Table 7-3.
7.2 GENERAL ECONOMIC ISSUES
Comment:
Commenters (IV-D-17; IV-D-48) expressed the belief that EPA has, in its
industry growth projections, failed to account adequately for current
economic trends such as the effects of foreign competition on the growth of
SOCMI. The commenters stated that the domestic SOCMI industry had
previously enjoyed some technological advantage and, in the recent past, had
the advantage of raw material and energy costs that are lower than those of
producers in Europe or Japan. They felt that EPA's 5.9 percent growth
factor is apparently an extension of the historic 6-percent factor and does
not adequately address current economic trends. They felt that external
forces deserve further consideration because they will decrease the
projected number of units affected by the standards and, therefore, decrease
the projected increase in VOC emissions.
Response:
The growth projection presented in the background information document
(BID) is 5.9 percent annually as estimated by McGraw-Hill (IV-M-35). The
McGraw-Hill projection was selected primarily because it was calculated for
a group of chemicals that closely corresponds to the SOCMI chemicals and
because the projection methodology was appropriate. In response to these
comments, a review of the literature was conducted to determine the availa-
bility of other projections for synthetic organic chemical production. It
was found that the U.S. Department of Commerce has projected a 6-percent
growth rate for the synthetic organic chemicals industry (IV-M-39).
However, both this projection and those in the BID are now somewhat dated.
7-2
-------
TABLE 7-1. SUMMARY OF CAPITAL COST ESTIMATES OF SOCMI STANDARDS, 1978$
Emission Source
Pump seals, light liquid3
Valves, gas and light liquid3
Safety/relief valves, gas
Compressor seals0
Sampling connections
Monitoring instruments
A
320
470
7,820
1,590
3,220
8,500
21,910
Model Unit
B
1,170
1,900
28,680
3,170
11,960
8,500
55,380
C
3,590
5,830
86,050
12,690
36,800
8,500 "
153,460
aThese are the amortized costs of initial start-up of the leak detection
and repair program (initial repair of leaks).
Assumes a 50/50 split between systems using 3-way valves and systems
using block valves.
Assumes a 50/50 split between systems tied to an enclosed combustion
device and systems tied to a flare; also assumes 60 percent of the
compressors in the industry are already controlled.
7-3
-------
TABLE 7-2. ANNUALIZED COST ESTIMATES OF SOCMI STANDARDS, 1978$
Emission Source
Pump seals, light liquid
Annual ized capital cost
Annual operating cost
Valves, gas and light liquid
Annual ized capital cost
Annual operating cost
Safety/relief valves, gas
Annual ized capital cost
Annual operating cost
Compressor seals
Annual ized capital cost
Annual operating cost
Sampling connections
Annual ized capital cost
Annual operating cost
Monitoring instruments
Annual ized capital cost
Annual operating cost
Total annual ized costs
Product recovery credit3
(55,200)°
Net annual ized costs
A
50
1,890
80
2,970
1,520
700
260
140
530
290
1,960
3,040
13,430
(4,830)b
8,600
Model Unit
B
190
6,770
310
11,970
5,570
2,580
520
280
1,950
1,070
1,960
3,040
36,210
(17,610)b
18,600
C
590
21,290
950
36,840
16,700
7,750
2,060
1,140
6,000
3,280
1,960
3,040
101,600
46,400
Product recovery credit is based on $300/Mg of VOC recovered or saved as a
result of the controls implemented.
Parentheses indicate credits due to the saved/recovered product.
7-'
-------
TABLE 7-3. CAPITAL AND ANNUALIZED COST SUMMARIES:
. NATIONWIDE PROJECTIONS
Capital Costs
Model Unit Cost per Unit, $1000
A 21.9
B 55.4
C 153.5
Total
Annual ized Costs
Model Unit Cost per Unit, $1000
A 8.6
B 18.6
C 46.4
Total
Number of Units
432
. , 274 .
125
831
Number of Units
432
274
125
831
Total , SMillion
9.5
, .15.2.
19.2
43.9
Total , $Million
3.7-
5~.l '" '
5.8
14.6
7-5
-------
Since their publication, it appears that the growth opportunities in the
synthetic organic industry may have decreased somewhat. Retention of the
original growth rate is, however, warranted for several reasons.
First, to develop a more current industry output projection would
require an elaborate study, which could be justified only if the projection
played a significant role in deciding either (1) whether there should be
standards or (2) which standards should be adopted. As discussed in
Section 2.1 of this document, plants that would be affected by the standards
would be significant contributors to air pollution. EPA has selected
standards with reasonable control costs. Thus, even if the growth rate
projection is high, EPA has still met the major criterion used to decide
whether a standard is needed and is reasonable within the context of
Section 111 of the Clean Air Act. Secondly, use of a high growth rate
results in a projection of higher industry compliance cost than would result
from use of a lower rate. A higher growth rate projection tends to repre-
sent a worst-case result in terms of the total fifth-year cost to society.
Thirdly, reduction in the industry growth projection would have the same
proportional effect on compliance costs and on the potential benefits of the
regulation. That is, any projection of the ratio of benefits to costs or
cost to effectiveness for the standards would be left unchanged. Therefore,
because these points indicate that there is insufficient reason to adjust
the growth projections, EPA has retained the original growth rate pro-
jection.
Retaining the growth rate projection used in the BID for the proposed
standards is consistent with recent revisions to the Priority List
(40 CFR 60.16). As expressed in revisions to the Priority List of source
categories (47 FR 950), Congress did not intend, in EPA's judgment, that
source categories showing insignificant growth should be listed under the
significant contributor list of Section lll(b)(l)(A) of the Clean Air Act.
In the context of the Priority List, EPA considers insignificant growth to
be indicated by one or no newly constructed, modified, or reconstructed
plants within the next 5 or 10 years. In the context of SOCMI fugitive
emission sources, EPA considers the growth rate projectionincluding the
7-6
-------
new information associated with current trends in this industryto be
significant and clearly in excess of the criteria used for the Priority
List.
Comment:
EPA, according to a commenter (IV-D-17), stated that both the degree
of control and the cost of control are not the primary basis for setting
standards; he stated that EPA considers reasonableness the primary basis.
The commenter claimed that the standards are based on reasonableness. He
stated that even though higher costs might not change EPA's decision to
issue an NSPS, the NSPS would have a large effect on industry. The
commenter pointed out that one company calculated a compliance cost of more
than $4.5 million per year for 36 capital projects scheduled for completion
between 1981 and 1985. When EPA estimates are used to determine the costs
for these same projects, a net savings of more than $100,000 per year
results. The commenter stated that, if these standards could be implemented
at a net savings for these projects, there would be no need for the
standards.
Response:
EPA considers costs and achievable degrees of control and then applies
reason to select a standard.' In particular, the commenter questioned the
accuracy and reasonableness of the cost estimates and the product recovery
credits. EPA believes that its final cost estimates are reasonable and that
its estimates of product recovery credits are based on realistic estimates
of the average value of the recovered materials. Details were not provided
on the compliance costs of $4.5 million annually that the commenters
indicated would be incurred. Therefore, it is not possible to review this
estimate.
The commenter also contends that, given EPA estimates, the producers of
synthetic organic chemicals would have already installed the control equip-
ment prescribed by standards that result in net benefits. However, there
are several reasons that a firm might not invest in cost-saving technology.
7-7
-------
First, a firm may not know about the technology. Secondly, if the firm is
aware of the technology, potential cost savings may not be perceived as
significant enough to tempt the company to introduce the technology,
especially if, for specific firms, the management is unfamiliar with the
technology. Thirdly, a firm may currently be using equipment that would be
too costly to scrap and replace by a new technology in the immediate future.
However, when the time comes to scrap the firm's old equipment, it could
choose to replace outmoded plant equipment with the previously unavailable
and less costly pollution control equipment. Fourthly, on occasion, invest-
ments in cost-saving pollution control equipment may have too low a return
to be considered worthwhile. Accordingly, the fact that SOCMI existing
plants have not adopted the recommended technology does not prove that the
estimates of savings are incorrect. These savings may not have been avail-
able or perceived when existing plants were constructed. Even new plants
would not necessarily use the recommended technology if owners or operators
are uncertain about potential savings because they lack experience with the
new technology.
Comment:
One commenter (IV-D-28) stated that cost estimates should not be for
the first 5 years only. He said this is an inadequate time frame. The
commenter said this rule would not end on the fifth, year, but would continue
in perpetuity, and that these costs would continue to rise forever.
Response:
The 5-year time horizon used in the environmental and economic impact
analyses is the typical time horizon over which impacts are calculated for
most regulatory actions. The 5-year time horizon facilitates comparison of
the costs and associated emission reductions with those of other standards
affecting both the SOCMI and other industries." EPA recognizes that costs
and benefits associated with the standards accrue over longer time horizons
and that these costs and benefits could be examined in addition to those
accruing over the 5-year period. However, it must be recognized that,
7-8
-------
although annualized costs increase over time as a result of the standards
(in the long run, at approximately the same rate as industry output),
benefits also increase proportionately. Because of the relatively small
cost of the standards and the constraints imposed by the need to develop
timely standards, additional effects were not investigated. The 5-year time
horizon may in fact be most appropriate for the assessment of cost and
benefits for these standard because, within the 5-year time horizon, a
relatively large number of existing facilities may be required to comply
with the standards through modifications or reconstruction. Consequently, a
longer time horizon for impact assessments would not likely alter the
implications of the regulatory analyses because both costs and benefits
would be increasing.
Comment:
One commenter (IV-D-28) noted that most of the calculations show the
after-tax cost to industry. He argued that, if a tax rate of 50 percent is
assumed, the pre-tax cost is twice that shown and that the pre-tax cost is
the true cost to society. The commenter said industry pays half of this
cost and society pays the other half in foregone taxes. If the tax rate is
lower, society pays a smaller portion.
The commenter added that the statement that no significant price
increases are to be expected from this proposal is naive; all costs of all
regulations are paid eventually by society as a whole as price increases,
foregone profits (lost dividends or capital accumulation), or foregone
taxes. Thus, society will pay all of those costs in the long run, in
addition to suffering an inflationary impact from unproductive expenditures.
Response:
There is an important distinction to be made between industry impacts
and social impacts. Industry impacts will result from firms' responses to
the standards. Examples of industry impacts include changes in the market
price for the industry's output, changes in firm and industry output levels,
and changes in firm and industry profit levels. In formulating decisions,
7-9
-------
individual firms consider the after-tax costs of standards. It is after-tax
profit that is available to stockholders, and stockholder wealth maximiza-
tion is the goal of the firm. Thus, when impacts on industry output, market
price, and certain other variables are computed, after-tax calculations are
in fact relevant.
In the "full cost pass through" study, the price of synthetic organic
chemicals is assumed to increase by an amount equal to the compliance cost
increase. Thus, an affected firm's taxable income will not decline as a
result of the standard. Corporate shareholders will not suffer a loss in
wealth, and the government will not lose tax revenue. The entire cost of
the regulation will be paid by chemical users in the form of higher prices.
An alternative estimate would assume full cost absorption. Affected
firms would incur the added cost of compliance but would receive no higher
price for their chemicals. If a 50-percent tax rate is assumed, half of the
cost would be paid by shareholders in the form of a loss in profits and half
would be paid by taxpayers in the form of foregone tax revenue. No cost
would be paid by chemical users through higher prices. The total cost so
calculated would be very close to the total cost calculated based on full-
cost pricing. Either is an estimate of the real resource cost of compli-
ance. However, full cost absorption is generally not a reasonable assump-
tion for standards of performance. Investors will not undertake the
construction of a new facility unless the market price is at a level that
will permit recovery of all costs.
Comment:
One commenter (IV-D-28) expressed concern over the use of 1978 dollars
for cost estimates. He stated that inflation has already increased the
costs by nearly 50 percent and will, by 1985, have caused an increase of
100 percent. Thus, all costs are underestimated. He added that the true
annualized costs by 1985 will be well over $100 million annually when the
effects of the current inflation and the total cost to society are con-
sidered. He stated that the initial capital costs for 1985 will be over
$125 million. Another commenter (IV-D-15) made a similar statement
7-10
-------
regarding the cost of capital. He suggested that the cost estimates be
escalated to reflect current capital costs.
Response:
The cost and impact analyses were conducted in constant dollars as
opposed to current dollars. Current dollar values differ from constant
dollars in that current dollars include the effects of inflation. Constant
dollars are corrected for inflation. The correction involves stating dollar
values for any time period in terms of some base year dollar's purchasing
power. -The base year for this analysis is 1978. Thus, all monetary impacts
are expressed in constant, 1978 dollars.
The use of constant dollars provides a standard of reference for
evaluating the real resource costs of the standards when such costs will be
incurred at different points of time. However, even with a doubling of the
current dollar value of the costs as suggested by the commenter, the compli-
ance costs would still not reach $100 million annually by the fifth year.
This is partially due to the fact that product recovery credits are included
in the cost estimate. If costs are inflated, the product recovery credits
would also be inflated.
When performing discounting operations, e.g., annualizing capital
costs, the discount rate (interest rate) and the flow of costs and revenues
to be discounted must reflect the same assumptions. In particular, if cost
and revenue streams are in constant (real) dollars, a real interest rate
must be used. A nominal interest rate (including inflation) is appropriate
only if nominal dollars are used to estimate cost and revenue streams.
Since real dollar values were used to analyze the economic impact of the
standards, a real interest rate was used also. The methodology employed to
estimate this real interest rate is outlined in Appendix A of the AID.
In summary, the real economic costs of the standard are needed to
assess the economic impacts of the standards. The appropriate way to
compute costs is to (1) allow all costs, revenues, and any recovery credits
to increase over time with expected inflation or (2) remove the effects of
inflation from the estimate. The method typically employed and the one used
here is the latter.
7-11
-------
Comment:
One commenter (IV-D-24) charged that EPA considers capital availability
to be unlimited. He stressed that it is not.
Another commenter (IV-D-15) wrote that the 10-percent interest rate
used to determine capital recovery costs is significantly low. He recom-
mended that a more realistic value of 15 percent should be used in a re-
assessment of the economic analysis.
Response:
The use of a positive interest rate reflects the scarcity of capital.
Two interest rates are used in the analysis. One is used to compute the
cost of capital to the SOCMI. This rate i's 10.8 percent and is based on an
analysis of the real cost of capital to the industry. It is assumed in the
analysis that the price of the chemicals will increase sufficiently to cover
all compliance costs plus earn a normal return (10.8 percent) on compliance
capital expenditures.
The second interest rate employed is 10 percent. It is used to compute
the social cost of the standard. It is also taken to represent a reaV value
of either the social rate of time preference or the social opportunity cost
of capital. The use of this rate is strongly recommended by the Office of
Management and Budget.
Both interest rates appear low compared to current rates. However,
when constant dollar values are used to estimate cost, as is the case in the
analysis, a real interest rate must be used (as discussed in the previous
response). As real rates, both 10 percent and 10.8 percent are quite
conservative and lead to a worst-case estimate for product price increases
and annualized compliance costs. -
Comment:
One commenter (IV-D-15) felt that the economic analysis was severely
distorted by amortizing the labor costs incurred during the initial year of
monitoring over a 10-year period. He suggested that EPA did this to make
the overall calculation easier because there is no sound basis for this
manipulation of figures. The commenter said labor costs are incurred during
7-12
-------
the year in which the work is performed. Therefore, by spreading out these
costs, the economics for the first year of operation become very favorable.
Response:
The commenter has apparently confused the treatment of one-time labor
costs and recurring labor costs. In the analysis for the proposed stan-
dards, labor costs are included in the year incurred. All costs are then
discounted to the present and annualized. EPA assumed that the labor cost
associated with the initial investment is paid along with all other invest-
ment costs. As part of the initial investment outlay, it. is thus recovered
over the entire life of the project at the appropriate discount rate. This
amortization procedure is appropriate and is used in investment-type
analyses.
However, the procedure used to estimate the labor cost has been
changed; Section 5 of the AID contains a discussion of the new procedure.
The procedure was changed because EPA is now using a Leak Detection and
Repair (LDAR) Model to estimate the amount of labor needed and, therefore,
labor costs.
Comment:
Two comment letters (IV-D-24; IV-D-17) objected to an economic analysis
that does not consider the cost of all environmental regulations on the
SOCMI. One commenter (IV-D-24) said EPA has not considered the effects of
NPDES, pretreatment, RCRA, Superfund, TSCA, etc. The commenter said when
all costs are summed, the number becomes very significant. The second
comment letter (IV-D-17) said the only meaningful economic analysis is one
of the total impact of all EPA regulations on an industry. This commenter
estimated that the economic impact of the standards to be about 0.25«t/kg on
average for a new SOCMI plant. Since fugitives are about 20 percent of
total VOC emissions, total regulation at the same cost per unit of pollutant
would be 1.25
-------
be on the order of 4.4
-------
plant, would incur a compliance cost of $520/Mg; Model Unit B, $445/Mg; and
Model Unit C, the most complex plant, $443/Mg; and further, recovery credits
for all three units would be $360/Mg. Thus, the smaller, less complex plant
suffers a relatively larger impact than does the larger, more complex plant.
This commenter also stated that compliance requirements could possibly
require hiring additional employees or result in decreased productivity,
especially for small companies. He presented one estimate for the annual
operating cost of compliance of about $40 for each valve or fitting
affected. He said the reduction in VOC emissions would be small, the loss
of productivity would be high, and, hence, the social costs would be great,
especially for smaller firms. This commenter requested that EPA withdraw
the standards so that resources .could be expended for greater benefits at
reasonable costs in other areas. This commenter felt the standards, if not
withdrawn, should exclude small facilities, since these facilities have
fewer emissions than do larger plants.
Response:
SOCMI fugitive emission standards are standards of performance estab-
lished pursuant to Section 111 of the Clean Air Act and hence would apply to
all new, modified, and reconstructed sources. Only firms that plan to build
new facilities or modify or reconstruct their existing facilities would be
affected by the standards. The Regulatory Flexibility Act (Public Law
96-354, September 19, 1980) requires that special consideration be given to
the impacts of proposed regulations on "small" entities. As a criterion for
extending'loans and related assistance, the Small Business Administration
defines a "small" business in the synthetic organic chemicals industry as
one that employs fewer than 1,000 workers (13 CFR Part 121, Schedule ft).
(For some chemicals a smaller number is used.) The Regulatory Flexibility
Act also applies to small organizations and small governments. However,
there is none that would be affected by the SOCMI NSPS.
The major basis for the commenter's contention that small businesses
would be disproportionately affected is the difference in Model Unit A
(sma.ll size) and Model Units 8 and C (medium and large sizes) cost
7-15
-------
effectiveness. There is no known relationship between model unit type and
small businesses. Thus, the commenter's comparison is not likely to be
relevant. Even if the comparison is relevant, the differences in cost
effectiveness cited by the commenter do not convince EPA that the impacts on
small businesses are unreasonable. However, because of the concern over the
small business impacts of governmental regulations, this issue was subse-
quently examined by EPA.
A two-step approach has been used to develop insights regarding the
possible impacts of the standards on small businesses. First, data on the
capacity of existing plants and employment of their parent firms were used
to identify the chemicals produced in plants owned by small businesses with
plant capacities small enough that production costs would be increased by
more than 2 percent by the standards. Second, trade publications were
reviewed to determine the plant sizes actually scheduled for construction
over the last several years for the identified chemicals. This review was
made to see if new plants producing the identified chemicals would be in
fact small enough to have an increase in unit costs of 2 percent. The
methodology and findings are presented below.
Compliance costs for a combination of chemical prices and plant capa-
cities were estimated for each type of model unit (A, B, and C). The
minimum capacity levels and product prices for each model unit that could
represent a 2-percent increase in average total cost (which is assumed to
equal the change in product price) are shown in Figure 7-1 for Model
Units A, B, and C. Product price-plant capacities falling to the left of
these curves would have a cost increase of more than 2 percent due to the
standards.
Data on organic chemical producers from the Organic Chemical Producers
Data Base, 1976 (IV-A-33) were then examined to see if existing small
businesses are producing chemicals with a plant size and product price that
fall in the area bounded by the price and capacity axes and the 2-percent
curves in Figure 7-1. In essence, existing sources were assumed to repre-
sent possible new sources. This assumption was made because of the uncer-
tain nature of any projection of new facilities and their ownership. To the
7-1C
-------
50H
p 408
r
i
c
300
i
n
200-
P
e
r
M 188
9
10080 26088 30000
capacity(Mg. per year)
40090
50080
Figure 7-1. Minimum price and capacity combinations.
-------
extent that new plants and firms are like current ones, data for current
plants can provide some insight into the possible impacts of the standards
on new plants and firms.
The data base contains information on 138 chemicals produced in 1,020
SOCMI plants by 69 firms. Of these, 30 chemicals were produced in 138
plants, which, if Model Unit C were employed, could have a unit cost
increase greater than 2 percent. However, at least 54 of the 69 firms
employ more than 1,000 workers. Thus* they would not be classified as small
businesses. Employment data are not available for the remaining 15 firms,
so, in a worst-case situation, all could be small businesses. Ten chemicals
are produced by these 15 firms. These 10 chemicals, their 1978 market
price, the threshold capacity below which the standards would increase the
average total cost of production by more than 2 percent, and the capacity of
the smallest existing plant producing each chemical are all provided in
Table 7-4.
Information is available on the sizes of process units used to produce
most of the 10 chemicals shown in Table 7-4. Information concerning the
sizes of process units (equipment counts) is contained in a series of
documents concerning organic chemical manufacturing (IV-A-19) and in the
background information document for proposed standards for petroleum refin-
eries (IV-A-34). This information indicates that none of these chemicals
would be produced as the sole product from a Model Unit C. Thus* based on
existing minimum capacities and the expected size of model units, only two
chemicals, methanol and toluene, might experience more than a 2 percent
increase in the cost of production. It seems reasonable to conclude that if
new facilities were constructed that mirrored the current facilities in
capacity and chemicals produced, the impacts on small firms would be mini-
mal. Only two chemicals might be impacted by a price increase (cost of
production) of more than 2 percent, and it is unlikely that even these two
would be significantly affected unless firms constructed Model Unit C (which
is unlikely) to produce those chemicals.
This initial review focused on the existing chemical plants and
identified 10 chemicals that might be affected by the standard. However,
7-18
-------
Table 7-4. .Threshold Plant Capacities of Chemicals Potentially
Produced by Small Firms at Current Prices
Chemical
Mg/yr)
Acetaldehyde
Benzene
Cyclohexane
Formaldehyde
Isobutylene
Methanol
Phosgene
Propylene
Toluene
Urea
1978
Market
Price
($/Mg)
420
220
250
110
240
130
370
210
170
300
Threshold
A
1.0
2.7
2.3
6.5
2.4
5.3
1.2
2.9
3.8
1.8
Capacity by Model Unit Type1*
(103 Mg/yr)
B
1.8
.6.6
5.4
16.8
5.8
13.6
2.5
7.1
9.6
3.9
C
4.0
17.6
14,2
46.3
15.2
37.5
6.0
19.0
16.0
10.0
Existing
Minimum
Capacities
(103
2.3
10.0
10.0
36.3
6.8 ;'
3.6
3.6
11.3
3.6
4.5
*The threshold capacity of a model unit is the capacity where the increase
in average total cost is exactly equal to 2 percent.
7-19
-------
another perspective can be gained by examining the types and production
sizes of plants for these 10 chemicals that are actually being constructed.
Such an examination provides additional insight into the characteristics of
plants that would be affected by the standards and the possible economic
implications of the standards. .
Data from Chemical Engineering (IV-M-28 to 53) and the Directory of
Chemical Producers (IV-M-54) were used to determine the capacities of plants
scheduled for construction over the 1975-1981 period for the 10 chemicals
identified in Table 7-4. The new capacity of the chemical and threshold
capacity of the affected chemicals are shown in Table 7-5 for comparison.
As shown in Table 7-5, new plant construction was not reported for acetal-
dehyde, isobutylene, or phosgene. This lack of information-is not a problem
because these chemicals .are not likely to be produced in a Model Unit C.
Except for formaldehyde, the new plant capacities are several times the
existing minimum capacities of the respective chemicals. Thus, because
expected new plant capacities for methanol and toluene are considerable
longer than their "threshold capacities," a price increase of more than
2 percent is not likely for these chemicals. To the extent that new plants
are representative of future industry construction patterns, these results
support the conclusion that the impacts on the industry in general and on
small firms in particular are likely to be quite small, i.e., price
increases will be less than 2.percent in all cases and, based on a review of
the costs for all chemicals, will average less than 0.25 percent for the
industry.
Comment:
One commenter (IV-D-17) said the product recovery credit was over-
stated. He said the average market value of $360/Mg was based upon very
pure finished products and that a more realistic estimate is $110/Mg.
Emission reductions will occur on raw material and semifinished streams,
which have a lower product value.
-------
Table 7-5. Comparison of New Plant and Threshold Plant Capacities
of Chemicals Potentially Produced by Small Firms
Existing
Minimum
Capacities
Chemical (10J Mg/yr)
Acetaldehyde
Benzene
Cyclohexane
Formaldehyde
Isobutylene
Methanol
Phosgene
Propylene
Toluene
Urea
2,3
10.0
10.0
36.3
6.8
3.6
3.6
11.3
3.6
4.5
Smallest New
Plant Capacity
Resorted*
(103 Mg/yr)
NA
65
376
32
NA
299
NA
136
38
70 .
Threshold Capacity
by Model -jUnit Type**
(10J Mg/yr)
A B C
1,0
2.7
2.3
6.5
2.4
5.3
1.2
2.9
3.8
1.8
1.8.
6.6
5.4
16.8
5.8
13.6
2.5
7.1
9.6
3.9
4.0
17.6
14.2
. 46.3
15.2 ,
37.5
6.0
19.0
16.0
10.0
*These are the smallest plants built during the 1975-1981 period.
**The threshold capacity of a model unit is the capacity where the increase
in average total cost is exactly equal to 2 percent.
Note: The symbol 'NA1 means that no new plants were built during the
1975-1981 period.
7-21
-------
Response:
The average market value of SOCMI chemicals was reestimated using
primary, intermediate, and final products, and the.price has been subse-
quently revised to $300/Mg. This value is an average product value and
reflects the average value of the materials conserved by the standard. Some
chemicals will have a value exceeding the average; others will be below it.
The methodology and data sources used to estimate the price are included in
Appendix A of the AID.
Comment:
One commenter (IV-D-28) provided the data presented in Table 7-6 in
response .to a request by EPA for information ,on change in prices to his
customers over the past 3 years. The domestic price index shown in-the
table is a weighted average of all domestic sales, adjusted for the annual
sales index, and is based on 1974 = 100.
However, the commenter also added that price indices are not a proper
measure of the profitability of an organization in times of spiraling infla-
tion such as these. Table 7-6 also shows that the operating income of the
chemicals groupwhich he said was the sector more heavily affected by
existing EPA rules.and would be the area most affected by this proposal--
fell nearly 22 percent as a share of sales and that operating income as a
percentage of invested capital fell even more, by over 25 percent. He added
that the capital investment to meet environmental regulations in this group
had risen to about 13 percent of all new invested capital in 1980 and that
the added operating costs because of these rules was now about 12 percent of
total operating costs, excluding raw materials and energy. These incremen-
tal operating costs due to environmental regulations amounted to 35 percent
of the operating income of the chemicals group in fiscal year 1980.
Response:
The commenter claims that his firm has not been able to recover the
costs of environmental regulations and other inflationary forces by price
increases to customers, and he implies that the same situation is likely to
7-22
-------
Table 7-6. Product Price and Profitability Data, 1978-1980
Year
Domestic Price Index*
All Products, 1974 = 100
Consumers Price Index
1967 = 100
Producers Price Index
1967 = 100
Chemicals Group
Operating Income as
% of Sales* . ..
Operating Income as
% of Assets*
1978
148
195.4
194.6
9.6
12.5
1979
156
217.4
215.9
7.7
10.7
1980
176
246.8
244.8
7.5
9.3
% Change
+ 18.9
+26.3
+25.8
-21.9
-25.6
*These data relate, to the commenters firm.
7-2:
-------
hold for these fugitive emission standards. There is not enough information
in the comment letter to verify either the claim or the implication.
Fluctuations in costs, revenues, and profits are a normal feature of most
industries. Furthermore, accounting procedures used in measuring impacts of
environmental regulations vary widely. For example, it sometimes happens
that costs incurred for non-environmental reasons are listed as costs of
environmental regulations, and it also happens that some costs are counted
twice if they are incurred to comply with two different environmental
regulations. These are legitimate accounting procedures in some situations,
but they would be inappropriate here. Also, unlike these fugitive emission
standards, some environmental regulations the commenter may be referring to
are applied to all existing facilities, and some regulations affect only
certain parts of the country; the costs of such regulations cannot be
recovered easily by price increases when there is a buyer's market.
Notwithstanding these considerations, it is EPA's position that the net
increase in costs attributed to the standards will be very small, averaging
less than 0.25 percent, and that the costs will be recoverable in the
market. The costs apply only to facilities that are new, reconstructed, or
substantially modified. Normally, a firm will delay the construction of a
new plant, or the modification or reconstruction of existing facilities,
until the market price enables recoupment of all applicable costs, including
a normal return on compliance capital expenditures.
7.3 COST-BENEFIT EVALUATION
Comment:
Two cormienters (IV-D-34; IV-D-38) considered the impacts of the
proposed program to be overly demanding in terms of dollars, manpower, and
time. Another commenter (IV-D-17) also said the standards would have a
significant adverse impact on SOCMI. He estimated that it will cost SOCMI
$80 million in 1981 to $100 million in 1985, or greater than $475 million
over the 5-year period, to meet the standards. He estimated that the emis-
sion reductions in 5 years due to the standards would not exceed 50 Gg and
that the net annualized cost, after credit for the recovered emissions,
7-24
-------
would vary from $10 million in.1981 to $35 million by 1985, or approximately
$135 million over a 5-year period. Considering that less than-1 percent of
the total national VOC emissions were being addressed, the commenter called
the costs excessive, inflationary, and unjustified. Another commenter
(IV-D-13) wrote that EPA's statement that control costs are insignificant
compared to the annual operating cost of the process unit itself fails to
recognize that all costs are significant and that each added increment of
expense in providing a product is an addition to inflationary pressures.
Addressing the overall cost/benefit of the standards, a group of
commenters (IV-0-17; IV-D-28; IV-D-47; IV-D-48; IV-D-19) said they did not :
see any need for the standards nor d-id they see any discernible benefit.
They said that the standards are counter to the spirit of Executive Order
12291, the purpose of which is to eliminate wasteful and unnecessary regula-
tions that place undue burdens on industry and the economy. They concluded
that the small impact on air quality of the standards does not justify what
they felt were the extreme costs of implementation.
Response:
The annualized compliance costs for each model unit have been estimated
and presented in Section 7.1 of this document for the standards summarized
in Section 1.1., also of this document:
Model unit Compliance cost ($/yr)
A 8,600
B" ' 18,600
C 46,400
Multiplying the total number of new units of each type projected over
the 5-year period 1981-86 by the cost per unit presented above gives the
following fifth-year annualized compliance costs: ,
Compliance Cost
Model Unit Projected New Units (106 $/yr)
A 432 3.7
B 274 5.1' : '
C 125 5.8
Totals 831 14.6
7-25
-------
The 831 new units include 274 replacement units, which encompass both
modifications and reconstructions. The costs of retrofitting some equipment
in facilities that are being modified is greater than the same equipment in
new and reconstructed facilities. The difference between retrofit and
ordinary compliance costs is small and even though retrofit costs were
considered in establishing the standards, this difference is not accounted
for here.
Executive Order 12291 specifies that a regulatory action, to the extent
permitted by law, must not be undertaken unless the potential benefits to
society from the regulation outweigh the potential costs to society. An
exhaustive benefit-cost analysis is not appropriate here due to the small
cost of the standards. However, a simple comparison of the costs and
benefits of the standards 'is presented below.
The standards would benefit society by reducing the release of VOC's to
the atmosphere. The following emission reductions, which are equal to the
amount of recovered material, are for each model unit and for the industry
in the fifth year:
Emission Reductions Industry Emissions
Model Unit (Mg/yr) per Unit Reductions (103 Hg/yr)
A . 16.1 7.0
B ..-. 58.7 16.0
C 184.0 " 23.0
Total 46.0
The compliance cost per Mg of VOC emissions reductions is therefore:
14'6x 1Q6- = $317/Mg
46.0 x 10J
This analysis is quite straightforward and ignores some complexities
that could be included. However, the basic implications would remain. In
view of the damages to the environment caused by VOC emissions and the
7-2C
-------
compliance cost estimates, EPA believes that implementation of these stan-
dards is justified. VOC emissions are a precursor to ozone formation in the
atmosphere. Section 2.1 of this document provides a discussion of the
benefits of ozone reductions.
7-27
-------
-------
8. LEGAL
This section contains comments concerning EPA's regulatory actions in
light of policies and rules established by the Clean Air Act, recent court
decisions, and executive orders on regulation. The comments have been
divided into nine major subject areas:
8.1 EPA's Regulatory Responsibility
8.2 .Priority List
8.3 Executive Orders
8.4 Court Decisions
8.5 Unit Operation Standards
8.6 States Authority
8.7 .Requests for Withdrawal or Delay
8.8 Statutory Time Requirement for Proposal
8.9 Technology-Forcing Standards
8.1 EPA'S REGULATORY RESPONSIBILITY
Comment:
Several commenters (IV-F-1, No.3; IV-D-24; IV-D-28; IV-D-19; IV-D-17)
concluded that EPA had not complied with requirements of the Clean Air Act
for setting new source standards. They argued that EPA had not made the
necessary considerations nor presented adequate justification for setting a
standard. According to the commenters, EPA failed to demonstrate that
fugitive VOC emissions from SOCMI could reasonably be anticipated to endan-
ger public health or welfare. They also noted that no effort was made to
evaluate the potential health effects which would result from the standard.
One commenter (IV-F-1, No.3; IV-D-28) continued, saying that he
believed that any decision to regulate a substance as a significant contri-
butor to pollution which may endanger health should include consideration of
at least these criteria:
8-1
-------
(1) Is health being endangered? Is the national standard being
exceeded for that pollutant?
(2) If it is, are the substances being considered contributing a
significant amount to the concentration of that pollutant in the
areas exceeding the standard?
(3) If health is being endangered by VOC emissions, which of these
emissions makes a significant contribution to that risk, and how
can they best be controlled?
\ (4) Are there other current regulatory activities which will have an
effect on this potential problem, and, if so, what will their
effect be?
He asserted that the Agency had not evaluated these questions. He did not
believe that the Agency could be justified in controlling a substance simply
because it is there.
Response:
The Administrator clearly documented the need to regulate VOC in order
to protect public health and welfare in the EPA publication "Air Quality
Criteria for Ozone and Other Photochemical Oxidants" (EPA-600/8-78-004,
April 1978). VOC emissions are precursors to the formation of ozone and
other oxidants. Photochemical oxidants result in a variety of adverse
impacts on health and welfare, including impaired respiratory function, eye
irritation, necrosis of plant tissues, and the deterioration of selected
synthetic materials such as rubber. Since there are estimated to be over
600 process units in SOCMI in the aggregate that have the potential to emit
a significant quantity of VOC, SOCMI was included on the priority list for
which new source performance standards (NSPS) are to be investigated. The
proposed standards will reduce VOC emission by 46 Gg/yr in the fifth year
after implementation.
The commenter's list of criteria indicates some confusion over the
difference between National Ambient Air Quality Standards (NAAQS) and
Standards of Performance for New Stationary Sources (NSPS). NAAQS are set
for certain criteria pollutants. Criteria air pollutants are those sub-
stances in the air which are reasonably anticipated to endanger public
health or welfare and which are released by numerous or diverse sources.
8-2
-------
SIP standards aimed at attaining the NAAQS in a particular area differ from
standards of performance which limit emissions from specific categories of
sources regardless of their locations. Standards, of performance reflect the
best technology for controlling a particular source and are not directly
designed to achieve any ambient air quality or public health or welfare
goals. Therefore, the commenter's first two criteria are not appropriate
except for the fact that NSPS are developed for sources of emissions which
may endanger public health or welfare.
The commenter's third criterion has been met. All VOC endanger pub-lie
health and welfare. Section lll(f) requires that EPA set standards of
performance for new stationary sources of VOC (within listed source cat-
egories) for which the best demonstrated technology (considering cost,
energy and nonair environmental impacts) can be identified.
With regard to the commenter's fourth criterion, there are basically
four types of environmental regulations (NAAQS, SIP's, NSPS, NESHAP's) and
some occupational health and safety standards which have the potential to
affect VOC emissions. As explained in Section 2.1, each of these statutory
programs plays a uniquely different role in meeting the goals of the Clean
Air Act and the Occupational Safety and Health Act. None of these statutory
programs negates the needs for standards of performance for fugitive
emissions from new SOCMI units.
Comment:
EPA was said to have given no thought to the geographical distribution
of the industry, that is, whether"the industry is located in attainment or
nonattainment areas or in rural or urban areas (IV-F-1, No.3).
Response:
In setting new source performance standards, location of the industry
in attainment or nonattainment areas is not relevant. Location of an
industry in an attainment or nonattainment area is relevant to achieving the
national ambient air quality standards (NAAQS) under Section 109 of the
Clean Air Act. The intent of Congress in establishing NSPS was to establish
a single minimum level of stringency for all state limits, thereby
8-3
-------
preventing States from soliciting industry with lenient air pollution
requirements and causing increased air pollution from new sources.
Comment:
EPA was charged with failing to address the relationship of the
emissions from the industry to other similar man-made and natural emissions
(IV-F-1, No.3: IV-D-19).
Response:
EPA has determined that emissions from SOCMI are significant
(540 Gg/yr), that emissions may endanger public health and .welfare
(EPA-600/8-78-004), and that the emissions can be controlled at reasonable
costs. This is sufficient basis for establishing an NSPS.
In the development of a NSPS, EPA is not charged -with examining the
relationship of the emissions from the industry to other similar man-made or
natural emissions in the area. This type of consideration is necessary for
the Prevention of Significant Deterioration (PSD) or nonattainment area
permitting process.
Comment:
A commenter (IV-D-17) said the standard was being.proposed out of the.
proper sequence. He felt that the more logical sequence is to establish
RACT and then evaluate the effectiveness of NSPS against the RACT control
baseline. He also noted that it is unreasonable to go from an "uncontrolled
baseline" to an NSPS control alternative and evaluate the cost effectiveness
of the NSPS against this "uncontrolled basis." The commenter believe that
RACT will achieve almost all of the reductions being proposed by EPA under
the NSPS and that the cost to achieve the additional NSPS control would be
unreasonable.
Response:
In the development of a NSPS the baseline level represents the level of
control in the absence of any standards of performance. Since there were
not Federal fugitive emission regulations for SOCMI and since the South
Coast Air Quality Management District in California was the only area of the
country that had developed such regulations, the assumed baseline for SOCMI
was uncontrolled. If several States had had time to develop RACT prior to
8-4
-------
the development of the proposed standards, RACT might have been used as the
baseline. . .
Where RACT-level control is already in place, however, the impact of
NSPS will be smaller than calculated. RACT and the systems chosen as best
demonstrated technology for this industry's standards of performance for new
stationary sources are not conflicting types of control; therefore, where
RACT already applies, the standards of performance will supplement RACT-
level control. EPA has .determined that existing RACT-level facilities that
become subject to the standards of performance (e.g., through modification)
can achieve the additional reduction required at a reasonable cost.
Comment: . . . , .
One commenter (IV-D-42) wrote that Appendix A of the EIS, ."Evolution of
the Proposed Standards," cites testing at the.information received from
Stauffer Chemical Company, Phillips Petroleum, Exxon Chemical Company, Shell.
Oil Company, Vulcan Materials Company, American Cyanamid, B.F. Goodrich,
Atlantic Richfield,-DuPont, Chevron, and Dow Chemical Company, among others.
Trade associations consulted were the Chemical Manufacturers Association and
the Texas Chemical Council.
The commenter noted that no beverage distilling companies or trade
associations representing them were studied or consulted, nor were any
background information documents sent to them (such documents were mailed on
7 November 1979 according to Appendix A of the EIS).
Response: . . :
EPA attempted to get,several SOCMI .representatives involved early in
the development of the standards with review and comment of preliminary
documents including those mailed on November.7, 1979. Later in the process
of developing the standards, the Agency sent the Federal Register notice of,
the proposed .standards (46 FR 1136, January 5, 1981) to additional SOCMI :
representatives as well. This notification was made to ensure that all
industries which might include affected facilities had been notified and
were given ample opportunity for review and comment. ,
Although the beverage distilling industry was not involved in the
preliminary review of the Background Information Document (BID), Volume 1, -.
8-5
-------
this industry did have the opportunity to participate in the public comment
period following proposal. All comments received during the public comment
period after proposal of the standards are considered by EPA in the final
rulemaking. Both the public notification, the individual notification of
various industry groups, and commenting period have provided ample oppor-
tunity for all interested parties to comment on the proposed regulations.
Comment:
One commenter (IV-D-42) wrote that Section lll(f)(l) of the Clean Air
Act, 42 U.S.C. 7411(f)(l), required the Administrator to publish a priority
list of stationary VOC sources to be regulated. That list as published on
August 21, 1979, did not include the distilling industry process units in
source categories.
, SOCMI was included on the priority list on the basis of the Adminis-
trator's determination that:
Sources within this industry contribute significantly
to air pollution which may reasonably be anticipated
to endanger public health, or welfare (46 Fed. Reg. 1152).
The commenter pointed out that no such determination was made for the
production of fermentation alcohol. He said that in the August 21, 1979
priority list EPA specifically stated that fermentation alcohol was not
being included in the sources to be regulated because:
Beer manufacture has a much lower emission level than
had been assumed in the background report, and whiskey
manufacture was deleted due to a lack of any demonstrated
control technology (44 Fed_. Reg. 49224).
In addition, in Section III, in studies conducted prior to 1979 EPA has
concluded that emissions from fermentation, distillation and other pro-
duction facilities were so low that they did not warrant regulation. Thus,
the commenter concluded, it could not have been reasonably expected that EPA
intended to regulate this industry when it listed SOCMI in the August 21,
1979 priority list.
One commenter (IV-D-42) wrote that the EIS cited the following factors
studied as a basis for regulating SOCMI and other emission sources, none of
8-6
-------
which have been considered in connection with fermentation alcohol: projec-
tions of growth and replacement of existing facilities, estimated incre-
mental amount of air pollution that could be prevented in future years, cost
of compliance, potential inflationary or recessionary effects of regulation,
effects on small businesses with respect to competition effects on consumer
costs, effects on energy use, and a thorough study of the profitability and
price-setting mechanisms of the industry (EIS, pp. 2-10). The commenter
stressed that fermentation alcohol plants, the structure of the industry,
its finances, the nature of the markets, its competitive position with
respect to foreign producers, and the economic impact of the regulation on
the industry bear no resemblance to the studies and conclusions about the
chemical industry discussed in the EIS.
He added that it is clear from the list of companies consulted,
(Appendix A of the EIS) and from a careful examination of the entire EIS,
that at no time did EPA ever consider distilleries producing alcohol from
grain, molasses, fruit or other agricultural food products in the analyses
leading to this rulemaking. Not even the description of the industry to be
regulated bears any resemblance to the fermentation alcohol industry; it
does not even mention the production of products which can be consumed as
food (EIS, pp.3-1 through 3-3).
The commenter concluded that since EPA did not consider the fermenta-
tion alcohol industry in the EIS, as required by law, it cannot be regulated
under the January 5, 1981 notice. .
Response:
The promulgated priority list (45 FR 49225, August 21, 1979) excluded
beer and whiskey manufacture source categories. However, the list included
the SOCMI category which was identified in the listing notices and the
background information as including 600 processes producing a wide variety
of chemicals. Some of these chemicals can be produced by fermentation
processes used in beer and whiskey manufacturing plants. While the listing
notices did not define beer and whiskey manufacturers, under the most
straightforward reading, those terms include only the processes used to
produce the fermented beverages, beer and whiskey, for human consumption.
8-7
-------
The draft background study entitled "Fermented Beverage Industry" confirms
the fact that this definition was intended by EPA. That study addresses
only the fermentation and distillation of fermented beverages produced for
human consumption; it does not treat the production of chemicals from
fermentation processes.
After reviewing the promulgated priority list and draft background
study, EPA has concluded that process units within beer and whiskey plants
that are producing fermented beverages solely for human consumption were not
included on the priority list within SOCMI or, another source category.
Therefore, they are not covered by the standards. However, any process
units in beer or whiskey plants that are not used for beer or whiskey
manufacture but rather to manufacture nonbeverage fermented products will be
subject to the standards.
There are several reasons to consider regulating nonbeverage fermented
products that are produced in beer or whiskey plants. First, the purpose of
the proposed standards is to reduce VOC emissions which are precursors to
ozone. Currently, there are readily available work practice procedures,
equipment standards, and performance standards that would reduce these
emissions. Therefore, not regulating these sources may be contrary to the
intent of the Clean Air Act. Secondly, EPA has concluded that the effec-
tiveness of the standards and the cost and energy impacts are equitable.
Finally, it is not necessary to regulate the fermented beverage industry as
a whole. Only those process units in a beer or whiskey plant that produce
one of the chemicals listed in 40 CFR 60.489 as final products could be
regulated.
The analysis of emissions, emission reductions, and costs presented in
the BID for proposal are applicable to fermentation alcohol units. Further-
more, the economic analysis performed for SOCMI units.included units which
produce chemicals from biomass, as discussed in Chapter 9 of the BID. The
inclusion of fermentation alcohol processes and other processes using
biomass as a raw material is further documented in II-A-12 and II-A-8.
8-8
-------
8.2 PRIORITY LIST
Comment:
The priority list was said to be preliminary and not final, meaning
that the decision to list the various industries was made without public
comment and without justification or generation of support that the
industries listed affect the public health and welfare (IV-F-1, No.3).
The same commenter, in another, set of comments (IV-D-28) said there was
no evidence of a further consideration of the preliminary assessment. He
also saw no support that the preliminary assessment was a proper, final
conclusion. . ,,
Response: , ;
While the priority listing included a preliminary screening of source
categories, it is a comprehensive screening, forming the basis for .the
Administrator's official determination under §lll(b)4(f); ft is not a
preliminary list. The final priority list was promulgated at 45 FR 49225 on
August 21, 1979, after full notice and comment on the proposed priority
list. This list reflects the Administrator's determination that, based on
preliminary assessments, emissions from the listed source categories con-
tribute significantly to air pollution which may endanger public health or
welfare. As mandated under Section lll(f)(3) of the Clean Air Act, the
Administrator consulted with Governors and State air pollution control. ,
agencies and conducted a public hearing in order to discuss the proposed ,
priority list. The public hearing was held on September 29, 1978. While new
information has led EPA to delete 12 categories from the list and further
information may lead to deleting other categories, nothing EPA has learned
since promulgation of the priority list suggests that SOCMI does not belong'
on the list.
Comment:
One commenter (IV-D-28) said the priority list ranking was improper.
He said the "potential to emit" criteria was used.instead of considering the
effectiveness of abatement devices which are in place. This error was said
to have occurred because the study was completed before the Alabama Power
decision.
8-9
-------
Response:
The commenter suggests that EPA misinterpreted the phrase ."emits, or
has the potential to emit" in the Section 302(j) definition of "major
stationary source" that applies to the standards through Section lll(f)(l).
However, even if EPA had considered existing control levels in deciding
which industries are "major stationary source" categories for purposes of
the Priority List (40 CFR 60.16), SOCMI would have remained a "major source
category" under the Section 302(j) 100-ton cutoff. More importantly, if EPA
had used the commenter1s interpretation and if SOCMI had become a "minor"
source category as a result, that would not change EPA's finding under
Section lll(b)(l) that SOCMI "contributes significantly" to ozone pollution.
Because SOCMI has been listed as a significant contributor, EPA must
promulgate an NSPS, regardless of whether SOCMI is one of the numbered
"major stationary source" categories on the category list at 40 CFR 60.16.
See Priority List promulgation, 44 FR 49223-49224 (August 21, 1979)
(discussing EPA's decision to list three categories not considered "major"
under EPA's analysis). Thus, the commenter's interpretation of "emits, or
has the potential to emit" under Section 302(j) would not have exempted
SOCMI from regulation under Section 111.
Beyond that, EPA disagrees with the commenter's interpretation of
Alabama Power Co. v. Costle. 636 F.2d 323, 352 (D.C. Cir. 1979). In that
case, the Court did hold that EPA must consider the level of emissions with
pollution controls operating when calculating whether a source "emit[s], or
has the potential to emit," 100 tons per year, under the definition of
"major emitting facility" in Section 169(1). That decision, however,
addresses only the question of which sources are subject to new source
review requirements under the prevention of significant deterioration (PSD)
program established in Part D of the Act. The Court based its decision
largely on its view of the intent of the PSD program. The Court has not
addressed whether EPA has properly focused on uncontrolled emissions levels
for the very different purpose of determining categories of "major
stationary sources" subject to inclusion on the NSPS Priority List,
60 CFR 60.16. See 43 FR 38874 (August 31, 1978).
8-10
-------
Although the definition of "major stationary source" that applies to
the NSPS program is similar to the definition of "major emitting facility"
used for PSD purposes, EPA believes that the two definitions may be
interpreted differently, in light of the different purposes of the two
programs. In EPA's judgment, focusing on uncontrolled emissions levels
rather than emissions levels with pollution controls operating is consistent
with the intent of the NSPS program. In contrast to the relatively
cumbersome and uncertain case-by-case PSD review requirements, NSPS's set
minimum standards applicable to broad categories of sources. Congress
intended PSD requirements to apply only to a portion of sources intended to
be covered by minimum NSPS requirements.^/ A focus on uncontrolled
emissions for NSPS listing purposes serves this intent.
Moreover, the Court in Alabama Power recognized, as EPA had in that
rulemaking, that focusing on uncontrolled emissions for PSD applicability
purposes would place an intolerable burden on both EPA and all but the very
largest sources. 636 F.2d at 354. Neither the Court nor EPA has found that
focusing on uncontrolled emissions for purposes of the NSPS Priority list
will impose similar burdens. Nor did any party raise this issue in
commenting on or petitioning for judicial review of EPA's use of
uncontrolled emissions in developing the Priority List.**/
VThe Alabama Power decision itself suggests this difference in applica-
bility. The Court cited NSPS's when it noted that, since Congress was
aware that many major sources were already operating pollution controls
pursuant to legal requirements, it must not have intended all such sources
to be brought under PSD. requirements by virtue of EPA's ignoring these
existing controls. 636 F.2d at 353. This implies the view that at least
some sources that should be covered by NSPS's should not be covered by PSD
review under the Section 169(1) "major emitting facility" definition.
Agency notes in this regard that the commenter did not comment -or
present a judicial challenge on this point. The initial Alabama Power
decision, 606 F.2d 1068 (1979), established the D.C. Circuit's preliminary
view on the definition of "major emitting facility". The Court issued
that opinion before EPA promulgated the Priority List. This permitted the
commenter ample opportunity to cite the decision in support of a timely
comment or judicial petition, ^ee Section 307(b); Oljato Chapter of
Navajo Tribes v. Train, 515 F.2d 654 (D.C. Cir. 197TT
8-11
-------
It is also important to note that EPA did consider the level of
\
emissions with existing controls operating in ranking the source categories
included on the Priority List. Section lll(f) establishes three criteria
EPA was required to use in that ranking. As described in the Priority List
proposal, supra, in the listing rulemaking EPA represented the first
criterion, quantity of emissions, as the emissions an NSPS would prevent
after being in effect for a specified period of time (in this case,
10 years). Emissions for 1990 were calculated first, assuming that the 1980
level of control continued to be applied to new sources. EPA termed this
level TS. Next, EPA calculated 1990 emissions assuming that a best level
of control, representative of an NSPS, were applied to all new sources
constructed between 1980 and 1990; this 1990 emission level was termed T^.
The "potential" emissions that could be prevented by an NSPS over the span
of 10 years equaled the difference between T~ and T... Based on this
emissions level, as well as the potential impact of VOC emissions on public
health and welfare and the mobility and competitive nature of this source
category, EPA properly ranked SOCMI first on the Priority List.
Comment:
Another commenter (IV-D-24) was concerned that EPA has a list of
sources and once it establishes an NSPS for one, it just continues on down
the list ad infinitum contrary to the intention of the Clean Air Act.
Response:
Section lll(f) of the Clean Air Act required the Administrator to
publish a list of categories of stationary sources for which NSPS are to be
promulgated. These source categories represent sources of pollutants that
in the judgment of the Administrator cause, or contribute significantly to,
air pollution which may reasonably be anticipated .to endanger public health
and welfare.
Section lll(f) requires EPA to promulgate standards of performance for
all categories on the list. In doing so, EPA analyzes in detail the alter-
native levels of control achievable for each industry before setting
standards for that industry. The process does not constitute blindly
establishing regulations. In fact, EPA has deleted categories when new
8-12
-------
information has suggested that deletion would be in keeping with the intent
of the Clean Air Act.
Comment:
One commenter (IV-D-38) said the proposed NSPS should be withdrawn and
the source category removed from the priority list since the NSPS would have
an insignificant effect on emissions. He cited, as support for this
request, the provisions in 44 FR 49222-23 (August 21, 1979) that, said:
[I]f further study indicates that an NSPS would have little or no
effect on emissions, or that an NSPS would be impractical, a source
category would be given a lower priority or removed from the list.
Response:
Neither this commenter nor any other submitted information
demonstrating that the contribution of SOCMI sources to harmful air '
pollution is significant. For example, projected growth obviously affects
the significance of the contribution a category of new sources makes to
harmful air pollution. If further study had indicated that NSPS's for the .-
SOCMI category would have little effect on emissions because there would be
no new SOCMI sources in the foreseeable future, this source category could
have been removed from the list. However, EPA has concluded that the
emissions contribution from both existing and projected new sources' in this
industry are significant. Therefore, SOCMI remains on the priority list.
Furthermore, EPA has identified at least one system of control that will
achieve additional emission reduction from the fugitive emissions
subcategory of SOCMI at a reasonable cost. For this reason, EPA is required
to promulgate these standards.
8.3 EXECUTIVE ORDERS
Comment:
One commenter (IV-D-38) considered the proposed NSPS to be contrary to
cost-effective regulation pursuant to Executive Order 12291. He summarized
the requirements of the Executive Order as follows: - . -
(1) Regulations should be based upon adequate information .
concerning the .need for and consequences of proposed
government action;
8-13
-------
(2) Regulatory action should not be undertaken unless the
potential benefits to society outweigh the potential
costs;
(3) Alternatives should be chosen so that the net benefits
to society are maximized and the net costs are minimized;
(4) Priorities should be set so as to maximize net benefits,
taking into account the condition of the economy and
other contemplated regulatory actions.
Two other commenters (IV-F-1, No.3; IV-D-28) said the real cost of the
proposed program in current dollars will be far more than the hundred
million dollar threshold for a regulatory analysis by 1985 or, perhaps,
earlier, and therefore, a regulatory analysis must be performed. The com-
menter (IV-F-1, No.3) also said the proposal must be classified as a major
rule under the recent E.O. 12291 and must, therefore, be reevaluated and
reproposed to be in compliance with that order.
Response:
Executive Order (E.O.) 12291 requires that a regulatory impact analy-
sis, thoroughly examining costs and benefits of a rule, be prepared in
connection with every major rule. A major rule is any regulation which is
likely to result in:
(1) An annual effect on the economy of $100 million or more;
(2) A major increase in costs or prices for consumers,
individual industries, Federal, State, or local
government agencies, or geographic regions; or
(3) Significant adverse effects on competition, employment,
investment, productivity, innovation, or on the ability
of United States-based enterprises to compete with
foreign-based enterprises in domestic or export markets.
An economic analysis of this standard was prepared. -Economic impact
estimates presented in the Background Information Document, Volume I, and
summarized in the preamble to the proposed regulation (46 FR 1136;
January 5, 1981) showed that no unreasonable economic impacts are expected.
Since proposal, changes in the standards have caused the economic impacts to
be reduced; thus, the discussion in Section 7 also shows no unreasonable
8-14
-------
economic impacts. Because no unreasonable economic impacts are expected and
the criteria for a major rule have not been met, no additional regulatory
impact analysis has been prepared.
Comment:
The same commenter (IV-D-28) said Executive Order (E.O.) 12044 required
consideration of the effectiveness of alternative levels of control. This
requirement was said to have been reemphasized by E.O. 12291. E.O. 12291
was also said to require that the lowest cost alternative be chosen. The
commenter said that EPA had failed to comply with these mandates by not
considering health data for the alternatives and by not providing an ade-
quate explanation for the choice of the most stringent alternative.
Response:
Alternatives were considered during the development of the SOCMI
regulation. They are discussed in the preamble to the proposed standards
and in the Background Information Document (BID), Volume I. In selecting
the final standards EPA considered alternative levels of control for each
fugitive emission source covered by the standards.
Health factors are considered prior to the decision to regulate the
source, indicating whether a pollutant endangers public health and welfare
and whether a particular source category's contribution of that pollutant is
significant. In development of .an NSPS, EPA is not specifically required to
analyze health data for each alternative. Each alternative control level is
evaluated in terms of emission reduction, impacts on water quality and solid
waste, secondary environmental impacts, energy requirements, and economic
impacts.
Comment:
In another set of comments (IV-D-17) EPA was urged to allow the use of
flares as control devices to be consistent with the spirit, if not the
express language of E.O. 12291. It was said that EPA should not preclude
technically sound and cost-effective regulatory options unless the Agency
establishes a record clearly documenting that the options will offset a
significant environmental benefit that could otherwise result.
8-15
-------
Response:
For the reasons described in Section 4.1, the regulation has been
revised to allow the use of smokeless flares that are designed and operated
within the exit velocity and heat content criteria established for the
following three flare types: steam-assisted flares, air-assisted flares,
and nonassisted flares.
Comment: .
A commenter (IV-D-43) said that E.O. 12291 requires that emissions be
calculated on a substance-by-substance basis in order to evaluate the-
benefit of various alternatives. In addition, this type of analysis is
needed to determine the incremental cost of ozone to the local burden, and
to estimate the results in terms of human health.
Response:
This commenter asserts that the emissions from the chemicals listed in
40 CFR 60.489 must be individually calculated to meet the requirement of
E.O. 12291, and that based on these calculations, adverse effects of each
chemical on public health and welfare must be based on local burden not on a
nationwide burden.
E.O. 12291 requires that potential societal benefits of a regulation
outweigh potential costs. This executive order does not require emissions
to be calculated on a substance-by-substance basis in order to evaluate the
benefit of various alternatives. Rather, the aggregate cost and resulting
benefits must be considered. Therefore, individual calculations for each
chemical are not necessary for these standards. As discussed in
Section 5.1, the chemicals listed in 40 CFR 60.489 are chemicals whose
production requires the use of fugitive emission sources to process VOC.
All of the processes represented have the potential to emit VOC which result
in photochemical ozone formation. The VOC may be contributed from raw
materials, additives, products, or co-products. The chemicals contribute to
ozone in varying rates and in different amounts depending not only on the
chemical's reactivity but also on meteorological conditions and the mix of
pollutants in the atmosphere.
8-16
-------
For purposes of setting an NSPS, it is sufficient to know that the
processes emit significant quantities of VOC which have the potential to
form ozone. The purpose of the NSPS is to control emissions of VOC from
new, modified, and reconstructed sources to prevent ozone levels from
becoming higher. Section 111 does not require that EPA establish standards
of performance according to the direct effect a given amount of VOC emitted
from the regulated source will have on the public health and welfare.
In response to the commenter's second point, NSPS are developed to set
a nationwide level of control which will keep air pollution problems from
worsening due to industrial growth, not to control VOC emissions on a local
level. Other parts of the Clean Air Act such as Section 110, State Imple-
mentation Plans, define programs for local regulation of industries keyed to
air quality in the immediate area.
8.4 COURT DECISIONS
Comment:
One commenter (IV-F-1, No.3; IV-D-28) argued that the proposal was in
sharp conflict with existing case law. Specifically, the commenter said
that EPA's action is contrary to the Supreme Court's decision in IUP v. API,
100 S.Ct. 2844 (1980). He noted that in IUD, which involved an Occupational
Safety and Health Agency (OSHA) benzene standard, the Court ruled that
before setting such standards OSHA must make a threshold determination that
the workplace currently poses a significant health risk and that a new, more
stringent standard is reasonably necessary or appropriate to provide safety
or to preserve the health of those affected. He commented further that the
Court rejected the idea that if there were some risk at one workplace
exposure level, it was proper for OSHA to regulate to a lower workplace
exposure level simply on the assumption that the risk also would be lowered
thereby.
The commenter noted the Court's holding that OSHA's action, by not
resting on a determination of significant existing risk and predicted
benefit, would unlawfully impose enormous costs that might produce little,
if any, discernible benefit. He also cited a concurring opinion by Chief
Justice Burger stating that OSHA must show something more than a de minimi's
8-17
-------
relationship between cause and effect, and also cited Alabama Power v.
Costle, 636 F.2d 323 (D.C. C1r. 1979).
The commenter suggested that even though the OSHA benzene case involved
a statutory provision different from Section 111 of the Clean Air Act, the
basic requirements are the same and have not been fulfilled in EPA's
establishment of this NSPS.
Response:
EPA does not agree that the legal requirements for establishing
NSPS's are the same as those governing OSHA standards. The statute at issue
in the OSHA benzene case states, among other things, that OSHA may establish
standards "reasonably necessary and appropriate to provide safe and health-
ful employment." (Occupational Safety and Health Act, §3(8)). Standards
involving harmful materials, in particular, also have to "assured, to the
extent feasible, .... that no employee will suffer material impairment of
health or functional capacity." (_Id., §6(b)(5)). The Court invalidated the
benzene standard OSHA promulgated under these provisions. A plurality of
the Justices held that to comply with §3(8) « as required in setting
§6(b)(5) standards ~ OSHA must first determine that workplaces in
compliance with existing standards are not "safe." The plurality held that
this in turn must mean that the workplaces pose a "significant risk" of harm
to the employee. The plurality ruled that, having made that finding, OSHA
may then set standards only if it shows that such standards will eliminate
or lessen that risk.
The legal standard under Section 111 of the Clean Air Act is quite
different from the standard at issue in the OSHA benzene case. Section 111
requires EPA to identify categories of sources that "contribute[] signifi-
cantly to[] air pollution which may reasonably be anticipated to endanger
public health or welfare." (Section lll(b)(l)(A)). For sources within such
"significant contributor" categories, the Agency must establish NSPS's that
other factors the Administrator determines has been adequately demon-
strated." (Section lll(a)(l)). Thus, in contrast to OSHA standards, each
NSPS is predicated, not on the Administrator's finding of a specific level
of existing harm caused by the affected source and the effect of more
8-18
-------
stringent standards on that level, but rattier on the findings that (1) the
source category in which, the affected sources fall "contributes signifi-
cantly" to the pollution found harmful under the Section lll(b)(l)(A)
standard discussed above and (2) the selected standard of performance
reflects the performance of the best demonstrated technology. Section 111
simply does not require NSPS's to reflect the degree of control shown
necessary to avoid directly a particular level of harm that the
Administrator has found to be significant.
Contrary to the commenter1s statements, EPA has complied with the
requirement in Section 111. First, EPA made the general finding that
ozone "may reasonably be anticipated to endanger public health and welfare."
EPA has found that ozone causes a variety of adverse impacts on health and
welfare, including impaired respiratory function, eye irritation, necrosis
of plant tissues, and the deterioration of selected synthetic material,
such as rubber. EPA clearly documented these adverse effects when it
promulgated the national ambient air quality standard (NAAQS) for ozone.
(See "Air Quality Criteria for Ozone and Other Photochemical Oxidants,"
EPA-600/8-78-004.) EPA has also documented the finding that VOC, the
pollutant regulated by this NSPS, is a precursor to the formation of ozone.
(_Id_.) The ozone NAAQS has been upheld by the D.C. Circuit. (665 F.2d 1176
(D.C. Cir. 1981), cert, denied, 102 S.Ct. 1737 (1982)). Next, through a
separate rulemaking, the. Administrator identified SOCMI as a "significant
contributor" source category. (NSPS Priority List, 40 CFR 60.16). As
explained in an earlier response, that listing was adequately documented and
was not challenged in court. V Finally, through today's final NSPS rule-
making, EPA is selecting the standard of performance reflecting the applica-
tion of the best demonstrated technology for the equipment leaks subcategory
of SOCMI. In sum, contrary to the commenter's claim, EPA is meeting the
applicable requirements in promulgating this NSPS.
V Moreover, no new facts have arisen to suggest that the SOCMI category
is no longer a "significant contributor" to ozone.
8-19
-------
The commenter also incorrectly suggests that this NSPS unlawfully
constitutes de minimi's regulation. In the case cited by the commenter,
Alabama Power Co. v. Costle. 636 F.2d 323, 400 (1979), the D.C. Circuit
stated that:
EPA does have discretion, in administering the statute's
"modification" provision, to exempt from PSD review some
emission increases on grounds of de minimi's or administrative
necessity.
As this language shows, the Court was simply noting the Agency's discretion
to avoid de minimi's -regulation in defining the term "modification" for PSD
applicability purposes. The Court did not require EPA to avoid de minimis
regulation in the PSD area. Nor did the decision address whether such
regulation might be effected under other provisions of the Clean Air Act,
such as Section 111. Thus, Alabama Power simply has no bearing on EPA's
promulgation of the NSPS for SOCMI equipment leaks.
Moreover, even if EPA were prohibited from promulgating an NSPS having
only de minimis effects, this NSPS would meet the applicable requirements.
As stated earlier in this document, these standards will reduce fugitive
emissions of VOC from process units constructed, modified, or reconstructed
between 1981 and 1985 from about 83,000 Mg/yr to about 46,000 Mg/yr. This
reduction will in turn reduce ozone levels from what they would be, absent
this NSPS. In the Agency's judgment, this is not a de minimis effect.
8.5 UNIT OPERATION STANDARDS
Comment:
One commenter in two comment letters (IV-D-28; IV-D-43) questioned the
Agency's authority to set unit operation standards. It was clear to him
that EPA could regulate in this fashion only if each substance or process
within the scope of such a regulation is shown by scientific evidence to
present a significant risk of that hazard which causes the Agency to act.
He could not see that the Agency had presented such evidence.
This same commenter, in a second letter, questioned the Administrator's
authority to set unit operations standards. In support of his statement,
the commenter cited the Proposed Revocation Notice for the National Ambient
8-20
-------
Air Quality Standards (NAAQS). for hydrocarbons (46 FR 25665, May 8, 1981)
which says: (1) that class standards cannot be set simply because members
of that class are suspect, and (2) that a case-by-case study of the need for
and benefits from any regulation must be performed prior to action.
Response;
Section lll(b)(2) of the Clean Air Act gives the Administrator the
authority to establish unit operation new source performance standards for a
class of facilities, provided those standards reflect best demonstrated
technology for that class. EPA has concluded that it can develop a unit
operation standard reflecting best demonstrated technology for over 600
SOCMI process units. This results in a substantial savings to the indus-
tries being regulated as well as to the taxpayer.
As previously explained in this section, all of the processes repre-
sented have the potential to emit VOC which result ,in ozone formation. The
VOC may be contributed from raw.materials, additives, products, or
by-products. The chemicals contribute to ozone at varying rates and in
different amounts depending not only on the chemical's reactivity but also
on meteorological conditions and the mix of pollutants in the atmosphere.
For purposes of setting an NSPS, it is sufficient to know that SOCMI
sources emit significant quantities of VOC which have the potential to form
ozone. The purpose of the NSPS is to control emissions of VOC from new,
modified, and reconstructed process unit fugitive emission sources to
prevent ozone levels from becoming higher as a result of industrial growth.
The Proposed Revocation Notice for the hydrocarbon (HC) NAAQS does not
directly affect the development of this NSPS. As explained in that notice,
the NAAQS for HC were intended only as a guide in the development of state
implementation plans to attain the original NAAQS for photochemical oxidants
(Recast as NAAQS for ozone in 1979). The revocation of the NAAQS for HC was
proposed because EPA determined that there is no single, universally appli-
cable relationship between HC and ozone and that HC as a class apparently do
not produce any health or welfare effects in or near ambient levels.
However, the proposed revocation was in no way intended to restrict EPA or
state authority to limit VOC emissions, including HC as a class, where
8-21
-------
necessary to limit the formation of ozone or particularly HC that are found
to pose a threat to health and welfare. Since VOC are precursors to ozone
and ozone has been determined harmful to the public health and welfare,
significant sources of VOC are subject to regulation under Section 111 of
the Clean Air Act (46 FR 25656; May 8, 1981).
8.6 STATES AUTHORITY
Comment:
One commenter (IV-D-22) said the promulgation of the standard would be
contrary to Section 101(a)(3) of the Clean Air Act which placesfprimary
responsibility for prevention and control of air pollution at its source
with State and Local governments.
Response:
The Clean Air Act requires the Federal government and State governments
to work together in controlling air pollution. With the 1970 Clean Air Act,
the Federal Environmental Protection Agency (EPA) was created and was
congressionally mandated to establish national air quality and regulatory
goals. In addition, the Act gave State and Local governments primary
authority for all regulatory efforts needed to achieve the national air
quality and regulatory goals.
Section 101(a)(4) of the Clean Air Act acknowledges the need for
Federal financial assistance and leadership in the development of cooper-
ative Federal, State, regional, and local programs to prevent and control
air pollution. In the case of new source performance standards, EPA is
mandated to establish national standards of performance for new, modified,
and reconstructed stationary sources. However, as delineated in Sec-
tion lll(c)(l) of the Clean Air Act, a State may develop and submit to the
Administrator a procedure for implementing and enforcing standards of
performance for the new source located in that State. Federal standards are
important because they set the minimum requirements for new sources.
8.7 REQUEST FOR WITHDRAWAL OR DELAY
Comment:
Three commenters (IV-D-38; IV-D-43; IY-D-50) requested that the
proposed standards be withdrawn. All three commenters claimed that the
8-22
-------
proposed standards would be costly with an insignificant effect on
emissions. One of the commenters said that based on SOCMI data, the uncon-
trolled emissions approach the level of control proposed in the regulation.
The estimated uncontrolled emissions, presented by the commenter, are
55 Gg/yr, compared to the 200 Gg/yr based upon refinery data presented in
the Background Information Document (BID), Volume I.
One of these commenters urged the Agency to withdraw the proposal until
EPA can supply the public with the relevant documents so that informed
public comment can be made.
Response:
EPA finds no logical or legal basis for withdrawing or delaying the
proposed NSPS for SOCMI. EPA has concluded that, as Section 111 requires,
the proposed standards reflect application of technology that will achieve
the most additional reduction at reasonable cost. The emissions reduction
based on SOCMI data are discussed in Section 3.4, and Section 7.2 discusses
the cost effectiveness of these standards.
The reports referenced in the FEDERAL REGISTER (46 FR 21789; April 14,
1981) provide an analysis of SOCMI fugitive emissions data. Although EPA
"was unable to fill all the requests for these reports, the reports are in
the docket in Washington and available for public review. In addition,
complete reports were sent to and made available through the Chemical
Manufacturers Association, the Texas Chemical Council, the American
Petroleum Institute, and the Synthetic Organic Chemical Manufacturers
Association. Furthermore, the AID was prepared and submitted to the public
for comment. Comments on the AID will be considered in preparing the final
standards.
Comment:
One commenter (IV-D-5) said promulgation should be deferred until
questions relating to the ozone standard are resolved. He said certain
questions relating to the ozone standard are currently before the U.S. Court
of Appeals for the District of Columbia Circuit. He also commented that the
number of allowable exceedances of the ambient standard is under review by
the Administration's Task Force for Regulatory Reform. He added that the
8-23
-------
National commission on Air Quality has recommended that the dates for
achievement of the ambient standard be lifted from the Clean Air Act.
Response:
This comment is not relevant to developing new source performance
standards. Section 111 requires EPA to set standards of performance for new
sources in source categories contributing significantly to harmful air
pollution. EPA has already determined that ozone is harmful to the public
health and welfare. Based on this finding and the judgment that SOCMI's
contribution of VOC, precursors to ozone, is significant, EPA is now setting
these standards (see Section lll(f)). Section 111 requires that standards
of performance for new stationary sources reflect the "best demonstrated
technology," not the technology that will achieve a particular level of
pollutants-in the ambient air. For this reason, the exact level of the
ozone NAAQS will not affect the level of stringency of any NSPS aimed at
controlling VOC emissions.
Comment:
Another commenter (IV-D-28; IV-F-1, No.3) said the proposal should be
withdrawn or reclassified as an advance notice of proposed rtilemaking. He
thought that it should be revised and reissued after serious procedural and
factual errors had been corrected.
Response:
EPA is unaware of any procedural or factual errors made during the
rulemaking process. In addition, the purpose of an advance notice of
proposed rulemaking for new source performance standards is to announce the
development of a standard. Both the proposed priority list (43 FR 38876;
August 31, 1981) and the promulgated priority list (44 FR 49222; August 21,
1979) were advance notices of proposed rulemaking for all new source
performance standards (NSPS). The proposed standards resulted from EPA's
detailed study of emissions, control technology, and control costs for SOCMI
fugitive emission sources. The proposal provided adequate notice of the
standards that EPA is promulgating in this action. It, therefore, served
the purposes of a notice of proposed rulemaking, not merely an advance
notice.
8-24
-------
8.8 STATUTORY TIME REQUIREMENTS FOR PROPOSAL
Comment:
Two cormienters (IV-D-28; IV-D-46) noted that the Administrator was late
in proposing the standards. One of the commenters said the Administrator is
required by Section lll(b)(l) of the Clean Air Act to make the judgment that
a category of stationary sources may reasonably be anticipated to endanger
public health or welfare. He reported that the Administrator made that
determination, on a preliminary basis, at 44 FR 49222 on August"21, 1979.
The commenter noted that the Administrator failed the statutory requirement
for proposal within 120 days. Publication of the proposal occurred about
500 days later. The second commenter questioned why NSPS had not been
developed sooner for the SOCMI industry, the number one source on the
priority list.
Response:
The source categories identified on the priority list are not subject
to the provisions of Section lll(b)(l)(B) of the Clean Air Act which would
require proposal of new source performance standards (NSPS) within 120 days
of adoption of the priority list. The promulgation of NSPS for sources on '
the priority list is to follow the time schedule prescribed in Section
lll(f)(l) of the Clean Air Act (44 FR 49225; August 21, 1979). EPA is
endeavoring to adhere to this schedule. As discussed in a previous comment
and response, the priority list ranking does not indicate the order in which
standards will be promulgated.
8.9 TECHNOLOGY-FORCING STANDARDS
Comment:
One commenter (IV-D-46) wrote that the technology-forcing requirements
of Section 111 bear special emphasis. The law requires the NSPS to reflect
the performance (or in certain cases, the design, equipment, work practices,
or operational measures) determined to be "achievable" by the best
adequately demonstrated controls. The commenter emphasized that what is
achievable is not the same as what had been achieved by the plants designed
and built years ago. The law requires new installations to apply knowledge
gained from past installations, not just to initiate the status quo of old
designs. Otherwise, the standards would be perpetually out of date.
8-25
-------
The commenter cited National Lime Association v. EPA where the court
explicitly reiterated the technology-forcing authority of EPA. He concluded
that if any doubt remained, it was dispelled by the subsequent power plant
(Sierra Club v. Costle] and diesel auto cases, which make it absolutely
clear that EPA is supposed to set technology-forcing NSPS.
Response:
This comment raises the issue of whether this standard has gone far
enough in requiring the SOCMI industry to develop and adopt the best
technology for the control of VOC emissions from new sources.
Although Section 111 of the Clean Air Act is technology-forcing to the
extent that it requires new, modified, and reconstructed facilities to
comply with a standard based on application of the best demonstrated control
technology, the setting of that standard and the selection of that best
demonstrated technology are constrained by the provisions of the Act.
As discussed in Essex Chemical v. Ruckelshaus, EPA's determination of
what contributes the best demonstrated technology must be based on solid
technical data. Pure speculation or prediction concerning the performance
of technologies which are not currently in use anywhere within an industry
cannot be used as the basis of the standard. Under this rule, technologies
or procedures which are not currently in use anywhere within the industry
can be considered demonstrated only if the available data show clearly that
they are capable of being successfully transferred from one application to
an industry being addressed by an NSPS. Although innovative or untried
technologies may in some cases form the basis for an NSPS, in many cases a
technology can only be demonstrated as a result of actual application and
experience somewhere within an industry. Therefore, to be consistent with
both National Lime and Sierra Club, technology-forcing under this standard,
and many others, can only be construed to extend to the application of the
best technologies and procedures currently available, not to untried or
developing technologies whose performance is unproven.
In the development of this standard, a wide range of technologies and
procedures were reviewed to determine their availability and performance,
including both those currently practiced within the SOCMI industry and those
8-26
-------
which are developing. Based on the terms of Section 111, and considering
demonstrated performance, availability, cost, energy impacts, and environ-
mental costs and benefits, the work practices and procedures which comprise
the basis of this standard, the equipment standards and the performance
standards were determined to constitute the most effective application of
control technology consistent with the requirements and constraints of the
Act. Should other technologies or practices be developed which surpass
these in effectiveness, costs, and benefits, they may be included in the
subsequent 4-year review of this standard.
The diesel auto cases referred to by this commenter are not applicable
to new source performance standards, but rather apply to the Vehicle
Emission Standards under Title II, Part A of the Clean Air Act.
8-27
-------
-------
9. MODIFICATION AND RECONSTRUCTION
Comments received concerning modification and reconstruction provisions
(40 CFR 60.14 and 60.15) as they apply to the standards have been divided
into six main subject areas: (1) Clarification,(2) Suggestions for Emission
Criteria for Modifications, (3) Suggestions for Changes in the Affected
Facility Definition for Modifications, (4) Technical Problems with Retrofit;
(5) The Capital Expenditure Criterion; (6) Coverage of Reconstructions.
9.1 CLARIFICATION
Comment:
Several commenters requested clarification of the modification and
reconstruction provisions. (IV-D-26; IV-F-1, No.9; IV-F-1, No.6)
Response:
Section 111 of the Clean Air Act directs the Administrator to establish
standards of performance for any new stationary source which causes or
contributes significantly to air pollution which may reasonably be antici-
pated to endanger public health or welfare. Under Section lll(a)(2) and
40 CFR 60.14 and 60.15 a new source is defined as any stationary source, the
new construction, reconstruction or modification of which is commenced after
the date of proposal of the standards (January 5, 1981). For purposes of
these standards, the stationary source is the group of fugitive emission
sources within a SOCMI process unit. Each such group of fugitive emission
sources on which new construction, reconstruction, or modification commences
after proposal is subject to the standards, i.e. is an affected facility.
An existing facility is the group of fugitive emission sources within a
SOCMI process unit on which construction has not commenced after proposal.
The General Provisions (40 CFR Part 60, Subpart A) outline the
procedures for implementing the Act, including the rules governing modifica-
tion and reconstruction (40 CFR 60.14 and 60.15). With some exceptions, a
modification is defined as any physical or operational change to an existing
facility which results in an increase in emissions from that facility. The
9-1
-------
exceptions include changes such as routine maintenance, repair, and replace-
ment; an increase in production rate accomplished without a capital expendi-
ture; and an increase in the hours of operation. The preamble to the
proposed standards pointed out that these standards are not intended to
cover existing process units making routine and minor additions. The
standards (40 CFR 60 Supart VV) exclude such process improvements. Such
routine changes and additions at existing SOCMI units would not cause an
existing facility to become an affected facility if the changes and
additions are made without incurring a capital expenditure. The General
Provisions define "capital expenditure" (40 CFR 60.2); the capital
expenditure criterion is discussed in a subsequent comment and response.
Section 60.15 defines reconstruction as replacement of components of an
existing facility 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 entirely new facility. It must also be
determined that compliance is technologically and economically feasible.
The General Provisions allow for the Administrator's case-by-case determi-
nation of whether proposed replacements constitute reconstruction.
Both of these provisions apply to existing facilities that would, if
affected by these provisions, become affected facilities. Increases in
emissions will be determined based, in general, on the number of fugitive
emission sources; if the number of fugitive emission sources increases,
emissions increase unless an owner or operator takes specific actions to
offset the increase due to the additional fugitive emission sources. The
capital costs associated with replacing an existing facility is limited to
the costs associated with replacing the fugitive emission sources only.
Emissions and capital costs associated with other sources, such as process
vent emissions, are not considered when evaluating the applicability of
these provisions.
Comment;
One commenter (IV-D-51) wrote that the proposed NSPS is unique in the
NSPS subparts because it applies not to a process but rather to equipment
within a category of processes. Because of this unique difference, the
9-2
-------
commenter said it is difficult to connect the content of 40 CFR 60.14 in the
General Provisions with the proposed rules. He recommended the following
addition to §60.480:
Modification of a process unit accomplished with
no net increase in emissions from the group of
fugitive emission sources within the process unit
shall not be considered a modification under this
subpart.
Response:
The commenter is correct in his interpretation of the modification
provisions. A change must indeed increase emissions to be considered a
modification. By definition, if a change is accomplished without an
emissions increase, it is not a modification. Certain changes must also
require a capital expenditure to be considered a modification.
The affected facility regulated by the standards is the group of
fugitive emission sources within a SOCMI process unit. The definition of
the affected facility is clearly stated in the regulation [§60.480(a)]. The
General Provisions (40 CFR 60.14) use the same term, "affected facility,"
(and the corresponding term, "existing facility") in explaining the pro-
visions for modification. Therefore, EPA has concluded that there is no
need to clarify the modification provisions specifically in Subpart VV.
Comment:
While supporting the provision in the proposed rules that process
improvements achieved without a capital expenditure are not considered
modifications, one commenter (IV-D-51) offered one suggestion to improve
this provision. He asked that the definition of Capital Expenditure in the
General Provisions1 be restated in the proposed subpart with changes to make
it clear for this subpart. The commenter expressed the understanding that a
facility's basis used to calculate the modification investment limit [sic,
"capital expenditure"] is only the affected facility and not the entire
process unit. He said the present definition of "capital expenditure" in
the General Provisions only fits cases where the affected facility.is the
same as the entire facility.
9-3
-------
Response:
The function of the General Provisions is to define terms and list
requirements that are common to all new source performance standards. The
standards use the definition of "capital expenditure" as intended in the
General Provisions and, therefore, it is neither necessary nor appropriate
to redefine that term in Subpart VV. The definition of "capital expendi-
ture" in the General Provisions uses the term "existing facility" and does
not imply that, in the case of Subpart VV, this term means the entire
process unit, and it:is this facility to which the capital expenditure
criterion-would be applied. Subpart VV defines a "facility" (which is
either an affected facility or an existing facility) as the group of all
fugitive emission sources within a SOCMI unit. Thus, the General Provisions
definition of "capital expenditure" fits the case of fugitive emission
sources within SOCMI. A comment related to this one concerning capital
availability is presented in Section 7.3.
9.2 EMISSION CRITERIA FOR MODIFICATION
Comment:
EPA was urged in three sets of comments (IV-D-7; IV-D-17; IV-F-1, No.l,
p.12) to consider an allowance for a de minimi's emissions increase criterion
in its revisions of the proposed regulations. It was suggested that it
would be reasonable to exempt sources increasing emissions by 10-20 tons per
year from coverage under the modification provisions. The 10-20 tons per.
year cutoff was said to be supported by data contained in an EPA contractor
study: Impact of Proposed and Alternative De Minimi's Levels for Criteria
Pollutants, June, 1980. One commenter recommended setting the cutoff at
either 10 tons per year or 10 percent of the baseline emissions, whichever
is greater. In one of the sets of comments (IV-D-17) further arguments for
establishing an emission increase criterion were given. PSD and nonattain-
ment reviews for permits were said to provide ample opportunity for
stringent control of modified sources. According to the comment, all net
increases in emissions from existing major sources must be accumulated.
Then, when a 40 tons per year level is reached, a review for a permit is
required. The commenter continued, saying that either BACT or LAER would be
9-4
-------
required on the last changed unit triggering review. This commenter and
another commenter (IV-D-15) basically requested that EPA consider an
emissions increase criterion as a possible solution to the problem of
covering plants making small changes.
Response:
tinder the definition in Section lll(a)(4), any physical or operational
change resulting in an increase in emissions constitutes a "modification."
EPA has exempted certain small emissions increases from consideration in
deciding whether there has been an increase in emissions constituting a
"modification" for purposes of PSD applicability (40 CFR 52.21(b}(2) and
(b)(231)). This action" followed the decision in Alabama Power Co. v.
Costle. 636 F.2d 323 (D.C. Cir. 1979), in which the D.C. Circuit held that
EPA has authority to interpret the definition of "modification" so as to
exempt sources with small emissions increases from PSD review'on grounds of
administrative necessity (Id. at 400). -
As explained in a previous response, the Alabama Power decision does
not require EPA to provide a de minimi's exemption from application of the
"modification" definition for NSPS applicability purposes. Nor has EPA's
experience in implementing the NSPS program suggested an administrative need
for relieving existing sources from NSPS applicability when they undergo
changes resulting in only a small increase in emissions. This differs ;
somewhat from EPA's implementation of the definition of "modification" for
PSD applicability purposes. In that area, the Agency has determined that
the administrative burden of applying the full preconstruction review
process to a source with only a small emissions increase may be unreasonable
(45 Federal Register 52705 (Aug. 7, 1980)). The administrative burden
associated with the NSPS program, however, is relatively minimal. In
contrast to PSD requirements, NSPS's are categorically applicable
technology-based requirements only; they do not involve an assessment of
ambient effects and do not require case-by-case review.
Furthermore, EPA believes that the current straightforward application
of the "modification" definition for NSPS purposes best serves Section Ill's
intent. One key purpose of the NSPS program is to prevent new pollution
9-5
-------
problems from arising. One way that the statute seeks to achieve this is by
requiring application of the best demonstrated technology at, and thereby
minimizing emissions from, existing facilities with increased emissions.
The current NSPS approach of not providing an exemption from the "modifica-
tion" provision based on the size of the emissions sources are not intended
to cover existing plants making routine and minor additions. The "modifica-
tion" provisions in the General Provisions of 40 CFR Part 60 exempt changes
such as additions made to increase production rate (if they can be accom-
plished without capital expenditure, as defined in the General Provisions)
and routine replacements (40 CFR 60.14(e)). In addition, these standards
would exempt additions made for process improvements if they are made
without incurring a capital expenditure.
9.3 AFFECTED FACILITY DEFINITION FOR MODIFIED SOURCES
Comment:
One commenter (IV-D-15) was concerned that the financial guidelines
provided by EPA to define a capital expenditure could result in inappro-
priate application of performance standards. The commenter thought this
might especially be a problem for small production facilities with limited
assets. He cited an example. If a distillation column were added to an
existing unit, the entire unit and all associated equipment would be subject
to the NSPS. The situation appeared unreasonable to him in view of the fact
that the fugitive emissions sources associated with the new column would
contribute only a small incremental increase in the overall VOC emissions
from the unit. He further commented that upgrading and retrofitting all the
fugitive emission sources within the unit would substantially increase the
capital required for the project and could possibly result in abandoning it
altogether. As a solution, the commenter suggested limiting the scope of
the affected facility to individual equipment being added or replaced for
modified facilities. He offered another alternative: an emissions increase
criterion (see Section 9.2).
Response:
The Clean Air Act requires that standards of performance be applied to
existing sources which increase emissions. EPA has carefully analyzed the
9-6
-------
modification and reconstruction provisions which might apply to existing
SOCMI units. The Agency has concluded that the provisions offer sufficient
flexibility for continued productive operation and reasonable protection
against inappropriate coverage of existing facilities within the context of
the Clean Air Act. In developing the standards, EPA considered retrofit
costs which would be incurred if an existing unit were covered under modifi-
cation provisions. In selecting the final standards, EPA selected standards
based on cost-effectiveness considerations,. Furthermore, the economic
analysis for this regulation showed no capital availability problem, nor did
it show any unreasonable economic impact.
The commenter's example and interpretation of the modification and
reconstruction provisions which would apply shows some confusion about the
provisions themselves and the affected facility to which they apply.
Because the determination of a capital expenditure is based on a percentage
of the purchase price of the existing facility, small and large production
facilities should be affected equally. The commenter may be referring to a
production facility owned by a small business instead of a small production
facility. Still, as shown in Section 7.3, the economic analysis showed no
adverse impact on small businesses.
Furthermore, the affected facility to which the standards apply is the
group of all fugitive emission sources (pumps, compressors, sampling
connections, safety/relief valves and open-ended lines) within a process
unit. If a new distillation column were added to an existing process unit,
the cost basis used to determine whether a capital expenditure had been made
is the cost of the added fugitive emission sources compared to the cost of
the existing fugitive emission sources. The cost basis does not include the
cost of the column or the entire process unit.
The existing modification provisions in 40 CFR 60.14 and the process
improvement exemption in the SOCMI standards prevent small increases in
emissions due to routine activities from causing modification to occur.
Finally, it is important to note that to be considered a modification, a
change made must increase emissions from the existing facility. If a change
is made that would increase emissions, but that potential emissions increase
9-7
-------
is offset so there is no net increase, then the change is not considered a
modification.
The preamble to the proposed regulation (40 CFR.1136, January 5, 1981)
explained that the commenter's suggestion of limiting the scope of the
affected facility to the individual equipment being added or replaced was
considered. However, such a designation would mean that replaced equipment
components in existing units would be subject to the new source standards,
while adjacent components would not be subject to the standards. Deter-
mining which components were subject to the requirements of the standard
would be impracticable for the owner/operator and for EPA. Therefore, the
individual equipment definition was not selected for the affected facility.
The commenter's suggestion of an emissions increase criterion was also
considered at proposal. This issue is discussed in Section 9.2.
9.4 TECHNICAL PROBLEMS WITH RETROFIT
Comment:
One commenter (IV-D-24) wrote that some existing SOCHI plants might be
subject to the rule by virtue of modification or reconstruction. In these
cases an enclosed combustion device would be required. For safety reasons,
the combustion device should be located at a distance from the process area.
However, the commenter said there might not be enough space to install a
combustion device in an appropriate location. For this reason, he said
modified and reconstructed facilities should be relieved of the requirement
to install a combustion device.
Another commenter (IV-D-15) cited the same problem of space
limitations. He said existing plants often would not have room for barrier
fluid systems and combustion devices. He added that locating these pieces
of equipment at a distance from pumps and compressors would increase
construction and operating costs by requiring more piping to interconnect
the systems and increasing the energy costs to run the system.
Response:
Regardless of whether a unit is a new or existing unit, EPA expects
that an existing control device will be used. The control device need not
be a combustion device. Furthermore, since proposal, some of the provisions
9-8
-------
which would have required a control device have been changed. The changes
effectively mean that a control device is not necessarily required; in fact,
most units will not require a control device.
The proposed standards required that pumps and compressors be
controlled with a barrier fluid/dual seal system. The vent streams from
this system were to be destroyed in an enclosed combustion device or
95 percent efficiency vapor recovery system. However, the final standards
allow an owner or operator of an affected facility to Choose among several
methods for compliance with the final standards for pumps and compressors.
First, the final standards for pumps do not require a control device. An
owner/operator may choose to comply with a work practice standard for pumps
instead of the equipment standard as originally proposed. The work practice
standard eliminates the need for a control device. The standards allow vent
streams from pump and compressor barrier fluid systems to be controlled by
flares, enclosed combustion devices, or any other control device designed
and operated for 95 percent efficiency. The provision for allowing flares :
means that if a control device is required, additional options are available
to the owner or operator. The provision for allowing flares also means that
a control device will most assuredly be present in all SOCMI process units.
As discussed in Section 4.1, smokeless flares are only allowed if they are
operated within certain exit velocity and gas heat content criteria.
Another change in the standards was made specifically to avoid retrofit
problems with existing compressors. If an existing 'compressor becomes
subject to the rules by virtue of modification and reconstruction and it
cannot be vented because vents cannot be provided in the existing hardware
surrounding the seal area, it is exempt from complying with the compressor
standards. This provision is discussed more fully in Section 4.12.
Comment:
One commenter (IV-D-15) gave another example of a retrofit problem with
rupture disks. He said the increased bulk and weight of the new pressure
control system may exceed the stress limits of existing tower welds and
connections. Increasing the strength of these new connections would be
costly and would require additional downtime for installation.
9-9
-------
Another commenter (IV-D-2) offered some solutions to retrofit problems
for pressure relief devices. He said if a combination rupture disc/relief
valve that had been flow-tested by the National Board of the ASME code were
selected, a larger relief device would not be required. He also offered a
solution to the problem of having fixed piping upstream and downstream of
the relief device. This situation prevents the device from being moved up
several inches to allow space for a rupture disc. He recommended a rupture
disc welded into a flange that fits down in the piping and has only a
minimal height of 1/8 to 1/4 inch. The price of the welded disk and holder
is about half the price of a standard disk and holder.
Response:
The standards do not require rupture disks as a control device. The
standards require "no detectable emissions" (defined as no higher a concen-
tration than 500 ppmv measured by a portable VOC monitor) from safety/relief
valves except during emergency venting. This standard may be met in any way
which results in no detectable emissions. Two ways, for example, would be
by using rupture disks or by piping to a flare header.. If rupture disks are
chosen as the method of control, there are some new methods for overcoming
some of the retrofit problems as described by the commenter in IV-D-2.
. The problem posed by the commenter regarding strengthening tower welds
would be an unusual circumstance from an engineering point of view.
However, if bracing or supporting is required, it can be done while the unit
is undergoing modification or reconstruction. No additional downtime would
be required. EPA recognizes that in retrofit situations more complicated
installation procedures may be required. These expected complications have
been accounted for in retrofit installation costs which are higher than
costs for new installations.
9.5 CAPITAL EXPENDITURE CRITERION
Comment:
One commenter (IV-D-7) described some problems with the capital
expenditure criterion applied to determine modifications. He said such a
criterion would be difficult to administer. He commented that a company's
records may not be detailed enough to reflect the cost basis under Internal
9-10
-------
Revenue Code 1012. An even bigger complication was seen in the fact that
the concept of "annual asset guideline repair allowance percentage" is a
part of the Asset Depreciation Range System (ADR) under the Internal Revenue
Code. According to the commenter, President Reagan's tax proposals would
repeal the ADR for property placed in service in 1981 and later years.
Other proposals would make the ADR optional. However, under any of the new
proposals, a company would have to maintain additional records for post-1980
assets, solely for the purpose of the proposed regulation.
Response:
This comment describes two possible problems with using a capital
investment criterion to determine exemptions from the standards for routine
replacement and additions made to increase the production rate. The first
problem is that a company may not have kept accounting records that are in
sufficient detail to reflect the cost basis of the existing facility. This "
cost is necessary to determine if a particular expenditure constitutes a
capital investment. It is true that this difficulty may be a larger problem
for the SOCMI fugitive VOC regulation than for many other new source stan-
dards. The difficulty arises in the fact that the affected facility is the
group of fugitive emission sources within the process unit rather than the
entire unit so that the total cost is composed of many small cost items.
However, there are ways to handle this problem. For example, one way to
estimate the cost of emission sources would be to prorate costs, based on
the present ratio of fugitive emission source costs to total costs. The use
of replacement costs with an appropriate adjustment to reflect original
costs would be considered by EPA. Therefore, because estimation methods can
provide a sufficiently accurate cost estimate of the existing sources, EPA
sees no reason to make the suggested change.
The second problem discussed in this comment with the capital invest-
ment criterion is the tax reform which changes the depreciation rules for
new facilities . The annual asset guideline repair allowance percentage,
referenced by the General Provisions (40 CFR 60.2) for the purpose of
defining a capital investment remains in effect for existing units. It was
included in IRS Publication 534 for 1981 tax returns. Existing units can
9-11
-------
still use the allowable percentage for determination of whether a capital
expenditure has occurred.
SOCMI units built in the future will use the accelerated depreciation
system for which there is no allowable repair percentage; however, they will
be covered by the standards in any case. EPA is currently considering
changes to the General Provisions which would solve the potential problems
which would arise with facilities which would become subject to NSPS in
future years.
Comment:
Another comment letter (IV-D-17) stated that the commenter's
understanding was that process improvements costing up to 12.5 percent
annually of the original investment would not by themselves be considered a
modification.
Response:
The commenter's statement is correct with some qualifications. The
figure of 12.5 percent of the original investment for fugitive emissions
sources represents the maximum expenditure that could be made annually for
routine replacement and minor additions without triggering application of
the modification provisions.
The calculation of this limit is explained in the definition "capital
expenditure" in 40 CFR 60.2. It is the product of the applicable "annual
asset guideline repair allowance percentage" specified in the latest edition
of Internal Revenue Service (IRS) Publication 534 and the existing
facility's basis. For 1980 and 1981, the annual asset guideline repair
percentage was 12.5 percent;, however, this is subject to change annually.
In 1979, for example, the repair percentage was 5.5.
It is important to note that an increase in emissions is the primary
test for determining if some physical or operational change to an existing
facility constitutes a modification. Moreover, changes made for process
improvements [and other changes as described in 40 CFR 60.14(e)] which
increase emissions would not be considered a modification if the capital
expenditure limit has not been exceeded. For example, an owner or operator
might add a control loop consisting of about 5-10 valves for purposes of
9-12
-------
improved process efficiency. If the cost of these valves did not exceed the
12.5 percent allowance of the original fugitive emission source costs, the
change would not be considered a modification. ~
Comment: - . , . ,
One commenter (IV-D-34) agreed with EPA's expression that "NSPS fo.r
SOCMI sources are. not intended to cover existing plants making routine and ,
minor additions." He further agreed that exemptions .should apply for
routine replacement and for additions made for-production rate increases if .<
such increases can be accomplished without capital expenditures.
Response:
The provisions for exempting production rate increases made without
capital expenditures may be found in 40 CFR 60, Subpart A.
Comment: :
Another commenter (IV-D-26) favored allowing capital expenditures if
there were no emissions increase. He said that eliminating the capital . .
expenditure criterion would allow updating of existing facilities while
limiting VOC emissions. Unless some provision is made for capital
expenditures, he added, older existing plants may not be upgraded into more
efficient facilities which incorporate current technologies.
Response:
Capital expenditures are allowed to some extent under modification and,
reconstruction provisions. Existing facilities may spend, up to 50 percent-
of the cost of a comparable new facility and not be covered by the standards
if emissions are not increased. However, if emissions are increased and a
capital expenditure is made (i.e., is greater than the product of the
applicable annual asset guideline repair allowance percentage and the
existing facility's basis), then the facility becomes a modified facility.
The intent of the Modification and Reconstruction Provisions is discussed in
a previous comment in Section 9.1. .
9.6 COVERAGE OF RECONSTRUCTIONS
Comment:
One set of comments (IV-D-17) said that EPA should revise 40 CFR
60.15 to limit the applicability of the proposed NSPS requirements to
9-13
-------
reconstructions. Reconstructions are essentially more comprehensive
modifications and to be consistent with the definition of modification,
should not be subject to NSPS unless there is an increase in net emissions
at the process unit. The comment cited a brief, filed by the steel industry
in nonattainment litigation, which demonstrates that the Clean Air Act would
not require application of NSPS to reconstructions where no emission
increases occur. The comment said that, in apparent recognition of the
reasonableness and legal support for industry's position, EPA proposed to
exclude reconstructions from nonattainment area new source review
(46 FR 16280). This argument continued that it is completely within EPA's
authority to limit the applicability for reconstruction of existing
facilities from any new source review requirements. The comment said this
change would offer the benefit of permitting full use of the bubble concept
at the process unit without any undue restrictions based on air quality.
Another commenter (IV-D-48) wrote that the definition of "reconstruc-
tion" at 40 CFR 60.15(b) is ambiguous in light of the statement of
applicability of 40 CFR 60.480(a) and the definition of "process unit" at
40 CFR 60.481.
He argued that, as these definitions are written in the proposal, it is
possible for a facility, over a period of years, to reach the 50 percent of
capital cost of construction of a new facility and thus become an affected
facility. Accordingly, the definition of "reconstruction" at
40 CFR 15(b)(l) is inadequate and should be amended to prevent this
occurrence. The commenter proposed that one or more of the following
suggestions be implemented to accomplish this amendment:
(1) Provide for a specified time limit during which accrual of capital
expenditures should occur. This time limit would have to be of
sufficient duration, e.g., two years, to prevent stretching such
repairs to avoid coverage.
9-14
-------
(2) Amend the definition to provide that, for purposes of determining
whether or not the 50 percent test is met, the cost of equipment
which is accrued be a single piece of process equipment and
directly related support equipment, e.g., pumps. Where such
support equipment is shared in common by two or more pieces of
process equipment, the capital replacement cost of the support
piece could be apportioned either equally or according to pro rata
usage among the various pieces of process equipment served.
(3) Amend the definition to provide that a repair qualifies as a
reconstruction only where there is a net increase in emissions
from the process unit.
This third approach was said to be, by far, the most desirable approach
since it provides a measure of pollution control which is unaffected by
external forces, such as inflation.
Response:
Since in enacting Section 111 Congress did not define the term
"construction," the question arose whether NSPS would apply to facilities
being rebuilt. Noncoverage of such facilities would have produced the
incongruity that NSPS would apply to completely new facilities, but not to
facilities that were essentially new because they had undergone reconstruc-
tion of much of their component equipment. This would have undermined
Congress's intent under Section 111 to require strict control of emissions
as the Nation's industrial base is replaced.
EPA promulgated the reconstruction provisions in 1975, after notice and
opportunity for public comment (40 FR 58420, December 16, 1975), to fulfill
this intent of Congress. Since this turnover in the industrial base may
occur independently of whether emissions from the rebuilt sources have
increased, the reconstruction provisions do not focus on whether the changes
that render a source essentially new also result in increased emissions.
Congress did not attempt to overrule EPA's previous promulgation of
Section 60.15 in passing the Clean Air Act Amendments of 1977. This
indicates that Congress viewed the reconstruction provisions' focus on
component replacement, rather than emissions level, as consistent with
9-15
-------
Section 111. See, e.g., Red Lion Broadcasting Co. v. FCC, 395 U.S. 367
(1969); NLRB v. Bell Aerospace Division, 416 U.S. 267 (1974). Nor has any
Court questioned the Agency's authority to subject reconstructed sources to
new source performance standards. In fact, in ASARCo v. EPA, 578 F.2d 319,
328 n.31 (D.C. Cir. 1978), the D.C. Circuit suggested that the reconstructed
provisions may not go far enough toward preventing possible abuses by owners
seeking to avoid NSPS by perpetuating the useful lives of their existing
facilities indefinitely. Failure to cover facilities that have undergone
extensive component replacement over a long period of time similarly post-
pones the enhancement of air quality Congress sought under Section 111. The
D.C. Circuit recognized this when it expressed concern in the ASARCo case
that, absent a provision for aggregating replacement expenditures "over the
years," owners could evade the reconstruction provisions by continually
replacing obsolete or worn-out equipment, [578 F.2d 319, 328 n.31 (O.C. Cir.
1978)].
Section 60.15 currently defines "reconstruction" as the replacement of
components of an existing facility 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 entirely new facility and
EPA determines that it is technologically and economically feasible to meet
the applicable NSPS. Subsection (d) indicates that the "new components"
whose cost would be counted toward the 50 percent threshold include those
components the owner "proposes to replace." It is unclear under this
wording whether a reconstruction has occurred in the case of an owner who
first seeks to replace components of an existing facility as a cost equal to
30 percent of the cost of an entirely new facility and then, shortly after
commencing or completing those replacements, seeks to replace an additional
30 percent. Specifically, it is uncertain whether the owner should be
deemed to have made two distinct proposals, or instead a single proposal.
If EPA would take the former view, owners could avoid NSPS coverage under
Section 60.15 simply by characterizing their replacement projects as
distinct proposals, even where the component replacement is completed within
a relatively short period of time.
9-16
-------
EPA did not intend, in promulgating the reconstruction provisions, that
the term "propose" exclude from NSPS coverage facilities undergoing this
type of extensive component replacement. Failure to cover these sources
serves to undermine Congress's intent that air quality be enhanced over the
long term by applying best demonstrated technology with the turnover in the
Nation's industrial base.
To eliminate the ambiguity in the current wording of Section 60.15 and
to further the intent underlying Section 111 (as described above), the
Administrator is interpreting proposed replacement components to include all
replacements which are required by programs of construction of alteration
which commence (but are not necessarily completed) during any 2-year period
for determinations of reconstruction. Stated differently, the Agency will
count toward the 50 percent reconstruction threshold the fixed capital cost
of all depreciable components (except those described below and elsewhere in
this document) required for replacement in all continuous programs of
reconstruction which commence within any 2-year period following
December 17, 1980. In the Administrator's judgment, the 2-year period
provides for this industry a reasonable, objective method of determining
whether an owner is actually proposing extensive component replacement,
within the Agency's original intent in promulgating Section 60.15.
The Administrator must decide on a case-by-case basis if proposed
replacements constitute reconstruction. This decision is based on the
following: (1) fixed capital cost of replacement and estimated life of the
facility after replacement compared to the cost and life of a comparable
entirely new facility, (2) the extent to which the components being replaced
cause or contribute to emissions from the facility, and (3) economic or
technical limitations on compliance inherent in the proposed replacements.
The brief and the Federal Register notice cited in one of the comments
both deal with nonattainment provisions rather than NSPS. The Federal
Register citation (46 FR 16280, March 12, 1981) proposed to standardize the
definitions of "source" for new source review in attainment and nonattain-
ment areas. Among other changes, the proposal seeks to drop the requirement
that reconstructions be subject to nonattainment new source review.
9-17
-------
However, as part of the rationale for dropping reconstructions from this
review, EPA stated that NSPS would assure the use of the most up-to-date
pollution control techniques, regardless of the applicability of nonattain-
ment area new source review. Thus, it is consistent with the brief and the
Federal Register notice to continue reconstruction as is for these
standards.
9-18
-------
10. EQUIVALENCY
Three commenters (IV-F-1, No. 4; IV-D-17; IV-D-18) were concerned with
equivalency provisions in the proposed standards. Two comments referenced
the equivalency provisions of standards already promulgated. The other
comments involved equivalency for test methods and equivalency demonstrated
by vendors.
Comment:
According to one commenter (IV-F-1, No. 4), although the proposed
equivalency procedures are detailed, they are not specific enough to insure
that, having met the requirements, equivalency is virtually automatic. The
comm?nter said the approach taken in the regulations for storage vessels for
petroleum liquids constructed after May 18, 1978 was simpler and more
straightforward and would be preferred over the present proposal.
Response:
The equivalency provisions for storage vessels for petroleum liquids
constructed after May 18, 1978 (40 CFR 60.114a) provide that the Administra-
tor may approve the use of equipment and/or procedures that have been
demonstrated to his or her satisfaction to be equivalent in terms of
reducing VOC emissions to that level required by the standards. The
following four items must be provided in the written application for equiva-
lency by the owner/operator: (1) emissions data, including a description of
the measurement method; (2) design specifications and estimated emissions
reduction capability of the control equipment; (3) an operation and mainte-
nance plan for the control equipment; and (4) any additional information
that would aid in evaluating equivalency.
Section 60.484 allows an owner or operator to apply to the Admin-
istrator for determination of equivalency for any alternative means of
emission limitation that achieves a reduction in emissions of VOC at least
equivalent to the reduction achieved by the required controls. Guidelines
10-1
-------
for application for equivalency are also presented in Section 60.484. The
requirements are essentially the same as those for petroleum storage
vessels. The guidelines are more extensive and detailed for the SOCMI
standards because several different sources are regulated by different
formats. Guidelines are presented for equivalency to equipment standards,
as well as for work practice standards.
After reviewing the equivalency provisions of the proposed standards,
EPA concluded that the detail included in the equivalency provisions is
necessary to provide an understanding of what an owner or operator will have
to demonstrate to obtain equivalency. It was also concluded that the
proposed provisions were explicit enough to convey what was needed without
being so explicit as to limit options of an owner/operator applying for
equivalency. Therefore, after these considerations were made, EPA decided
to promulgate the equivalency provisions as proposed except for the addition
of a provision for allowing manufacturers and vendors to apply for equiv-
alency. This addition is discussed in a later comment in this section.
Comment:
Two comments were made concerning equivalency and monitoring. One
commenter (IV-D-18) said monitoring systems which are equivalent to those
required by the vinyl chloride NESHAP (40 CFR 61.55(b)(8)) are equivalent to
the monitoring requirements for the NSPS. He felt that these equivalent
monitoring systems should be written into the final regulations as alterna-
tives.
Response:
Section 61.65(b)(8) of the vinyl chloride NESHAP addresses "leak
detection and elimination" of fugitive emissions from ethylene dichloride,
vinyl chloride (VC) and polyvinyl chloride plants. This section calls for
VC emissions "to be minimized by instituting and implementing a formal leak
detection and elimination program." The program, developed by the
owner/operator and subject to the approval of the Administrator, must
incorporate the following features:
10-2
-------
(1) A VC monitoring system (fixed point monitors) for detection of
major leaks and identification of the general area of the plant
where a leak is located;
(2) A portable hydrocarbon detector to be used routinely to find small
leaks and to pinpoint the major leaks indicated by the fixed point
system. The sensitivity of such a device must be at least
10 ppmv;
(3) A calibration and maintenance schedule for the items in (1) and
(2).
The specific requirements of each program must be appropriate for the
individual plant. That is., the definition of leak is made based on back-
ground concentration measurements and the features of the overall monitoring
program (location and number of fixed points, as well as frequency of
monitoring) depend upon the size, layout, etc. of the plant.
In choosing a monitoring system/procedure for the SOCMI fugitive
emissions standards, several alternative approaches were considered. One
such, system was the same system specified by the vinyl chloride NESHAP.
This system, however, was not selected for the SOCMI NSPS since monitoring
with a portable VOC detection instrument would be required in both cases and
the fixed point/portable monitoring system was more capital intensive.
Moreover, EPA believes the fixed point system is less efficient in detecting
leaks due to possible meteorological interference. Thus, the monitoring
system may not be equivalent to that required by the SOCMI NSPS.
EPA has recently reviewed the programs developed by polyvinyl chloride
and vinyl chloride manufacturers. The portable monitoring programs vary
considerably among the owners and process units affected by the vinyl
chloride standard. Large variations are seen in leak definition, repair
interval, and monitoring interval. Some of the variations are so large that
equivalency to the SOCMI NSPS is questionable. EPA believes that this large
variation should be minimized. Therefore, EPA is considering clarifying the
vinyl chloride standards to include specific monitoring requirements. In
general, the VC Standard is more restrictive than the SOCMI NSPS. However,
as mentioned above, some aspects of the VC standard may not have been
10-3
-------
adequately implemented. Where the VC standard is more restrictive, the
SOCMI NSPS provides automatic equivalency within the standards. For
example, double mechanical seals are required for pumps in the VC standard,
even though monthly monitoring generally is required in the SOCMI NSPS,
double mechanical seals are explicitly allowed, and if used monthly moni-
toring is not required in the SOCMI NSPS. Where the VC standard is unclean
or less restrictive than the SOCMI NSPS, the additional clarity or emission
reductions are warranted. Thus, based on these considerations, EPA decided
not to remove VC from the SOCMI list or draft the VC monitoring procedure
into the NSPS as alternatives. If a plant has a good system in place, that
plant could apply for permission to use an alternative plan.
Comment:
In another comment letter (IV-D-17) equivalency provisions were
requested for test methods and procedures, as well as for control tech-
niques. The commenter's concern was that the regulations do not provide the
necessary flexibility to approve in the future the use of new instruments
that may use different calibration systems which may provide equivalent or
more accurate results.
Response:
Equivalency determinations are allowed for performance tests and
monitoring by the General Provisions (40 CFR Sections 60.8 and 60.13).
Also, Reference Method 21 gives specifications for the monitoring instrument
that are general enough so as to allow new analytical developments. There-
fore, new instruments using different calibration procedures would be
allowed when the results of the equivalent method have been demonstrated to
be at least as accurate as the results obtained by the required methods.
Comment:
Another commenter (IV-D-17) requested that vendors and manufacturers be
allowed to apply for equivalency. The commenter said that in the real world
context few companies would purchase an innovative control system unless the
manufacturer stated that EPA had approved the system.
10-4
-------
Response:
The equivalency provisions of the proposed standards (§60.484) did not
directly allow a vendor or manufacturer to apply for equivalency. Since
proposal, the proposed equivalency provisions have been changed to allow a
vendor or manufacturer to apply for equivalency. However, it should be
remembered that efficiency of some control techniques can only be shown to
be equivalent by the owner or operator. For those, EPA will not grant
equivalency determinations to vendors or manufacturers. The ultimate
responsibility for insuring the equivalency of alternate methods is on the
owner or operator, not on the vendor or manufacturer.
10-5
-------
-------
11. RECORDKEEPING AND REPORTING
Comments were received on the recordkeeping and reporting provisions of
the proposed standards. They fall roughly into four areas. The first area
of comments concerns the burden of recordkeeping and reporting provisions.
Another area concerns specific recordkeeping provisions, mainly centering on
the requirement for having records readily accessible. The other two areas
concern tagging requirements and provisions for delaying repair past
15 days.
11.1 REPORTING BURDEN
Comment: . " ' " ;
Several commenters (IV-F-1, No".4; IV-D-24; IV-D-5; IV-D-17; IV-D-20;
IV-D-22; IV-D-7; IV-D-34; IV-D-32) said the proposed reporting requirements
were excessive and burdensome. One commenter (IV-D-24) said quarterly
reporting should be reduced to annual reporting. He noted that annual
reports were the precedent at both State and Federal levels. He further
cited dual advantages to this recommended change: a reduction in manpower
required at the plant and a reduction in the size of the bureaucracy needed
to review the reports. Another commenter (IV-D-20) suggested that reports
be eliminated. He said that records were required and that they were
available at any time to EPA personnel. He argued that reporting would do
nothing to further compliance.
Three comment letters (IV-D-17;'IV-F-1, No.l; IV-D-22) said State
reporting requirements overlapped with the proposed reporting requirements
for the new source standard. It was suggested (IV-D-17; IV-D-22) that
reports be submitted to the States; EPA could then request information from
the States.
One commenter (IV-D-21) said that requiring certification of the
quarterly reports by the owner or operator is excessively burdensome and
inconsistent with other new source performance standards (NSPS) reporting
requirements. He cited as an example the fact that industrial boiler NSPS
11-1
-------
excess emission reports do not require owner or operator certification. He
argued that such statements were more properly signed by an authorized
representative or designate.
Two provisions were suggested as additions to the proposed requirements
(IV-0-17). One suggested addition was a provision for reporting the number
of leaking sources not repaired at shutdown. The other suggested addition
was a provision for reporting the number of pump and compressor leaks not
repaired within 15 days.
Response:
The proposed standards included reporting provisions requiring periodic
reports of leak detection and repair efforts within a process unit. The
reported information was regarded by EPA as a good way to judge how
diligently the required leak detection and repair program had been
implemented. The reporting requirements were considered a means of reducing
in-plant inspections. The costs of reporting were assessed and judged
reasonable.
EPA continues to believe that reporting requirements would reduce
in-plant inspections as a means of determining compliance. But the
quarterly reporting requirements in the proposed standards have been reduced
to semiannual reporting in the final standards. The semiannual reports may
be waived for process units in States with the delegated authority to
enforce the standards provided (1) EPA approves the reporting requirements
or an alternative means of compliance surveillance adopted by the State and
(2) the process unit complies with the requirements established by the
State. The one-time notification requirements given in the General
Provisions, however, have not been eliminated from the final standards.
Therefore, any reporting overlap (notifications, etc.) between State and
Federal requirements can be determined when the States request information.
As explained at proposal, three alternatives were considered for
reporting requirements. The three alternatives represented trade-offs
between varying amounts of in-plant inspections and report preparation for
enforcement. The first alternative required minimum reporting and relied on
inspections for enforcement. The third alternative relied almost totally on
11-2
-------
reports and would require a minimum of inspections to judge compliance. The
second alternative, the one selected for proposal, represented a compromise
with some reporting and some inspections required. However, EPA has
concluded that periodic reporting requirements are necessary and has
included semiannual reporting requirements in the final standards. The
reporting requirements have been streamlined to reporting of data on leak
detection and repair of pumps, valves, and other equipment types only. In
addition, the requirement of certification of periodic reports has been
eliminated. This alternative requires an intermediate amount of reporting
and relies also on in-plant inspections for assessing compliance with the
standards. However, the semiannual reporting requirements may be waived for
process units in any State that is delegated authority to enforce these .
standards provided certain criteria of reporting requirements are met.
In addition to periodic reporting, the following reports are required
for compliance with the final regulation:
(1) Notification of construction [40 CFR 60.7(a)(l)] or reconstruction
[40 CFR 60.15];
(2) Notification of anticipated startup [40 CFR 60.7(a)(2)];
(3) Notification of initial startup of affected facilities
[40 CFR 60.7(a)(3)];
(4) Notification of physical operational changes in equipment
specifications [40 CFR 60.7(a)(4)].
If an owner or operator opts for compliance with an alternative standard,
the following additional reports may be required, depending on the standard
selected:
(1) Notification of intent to comply with an alternative standard
[§60.487(b)];
(2) Performance test results [40 CFR 60.8 except as noted in
§60.487(c)].
Comment:
Several commenters felt that some or all of the data requirements of
the proposed reporting provisions are irrelevant for determining compliance
(IV-F-1, No.4; IV-D-17; IV-D-21; IV-D-23; IV-D-7). Specifically cited as
irrelevant for compliance purposes, and therefore unnecessary, were the
11-3
-------
number of valves repaired (IV-D-17; IV-D-23); the number of valves in the
process unit (IV-D-17; IV-D-7); the number of valves, pumps, and compressors
found leaking (IV-D-17; IV-0-21); and the number of leakless valves
(IV-D-21). Three of the commenters interpreted the proposed reporting
requirements as being devised solely for data collection and enforcement
(IV-F-1, No.l; IV-D-17).
Section 114 of the Clean Air Act was cited by one commenter (IV-D-17;
IV-D-50) in speaking of reporting requirements. This commenter said that if
data/information required by the proposed reporting provisions were not
relevant to ascertaining a source's compliance, the Agency will have
exceeded its authority under Section 114.
Response:
The commenters are concerned that the information required in periodic
reports was irrelevant to determining compliance and, therefore, exceeded
EPA's authority of collecting data under Section 114 of the Clean Air Act.
Section 114 established EPA's authority to gather information to enforce
regulations. This authority applies to enforcing promulgated standards, as
well as to gathering information on which standards will be based and later
enforced. Therefore, EPA's authority under Section 114 would not be
exceeded by their required data since EPA deemed that data necessary for
determining compliance with the standards.
As noted in the previous response, however, the routine periodic
reporting requirements have been reduced to semiannual reporting in the
final standards. The data that must be reported in the semiannual reports
must also be maintained as part of the recordkeeping requirements.
Inspection of records will also be used to determine compliance with work
practice standards.
Comment:
The meaning of §60.487(a) was said to be unclear and duplicative of the
sections that followed (IV-D-17). Deletion of this section was recommended.
Response: .
Section 60.487(a) of the proposed standards required that reports
contain the information recorded under §60.486. The meaning of this section
11-4
-------
was neither unclear nor dupTicative. The information required to be
reported in the semiannual reports that are part of the final standards has
been streamlined. Moreover, those data that must be reported are clearly
defined in §60.487(b) for the initial semiannual report and in §60.487(c)
for the subsequent semiannual reports. The data required in the routine
reports include information summarized from records on leak detection and .
repair of pumps, valves, and other equipment.
Comment:
Another commenter (IV-D-22) thought that the recordkeeping and
reporting requirements were not in concert with the Paper Work Reduction Act
of 1980 (44 CFR Chapter 35).
Response:
The recordkeeping and reporting requirements for compliance with the
new source standards have been reviewed and approved by the Office of
Management and Budget as required by the Paper Work Reduction Act of 1980
(44 USC 35). As stated previously, quarterly reporting has been reduced to
semiannual reporting in the final standards. This reduction is consistent
with the intent of the Paperwork Reduction Act of 1980. As explained below,
records documenting the performance of the work practices are also necessary
for judgement of compliance and are, therefore, required by the standards.
Comment;
Two comment letters (IV-D-17, IV-D-23) said that a single certification
statement should be allowed for those plants containing more, than one
affected facility.
Response:
Certification of a process unit's full compliance with the standards is
not required under the promulgated standards. However, if States or EPA
Regional Offices request reports under Section 114 of the Clean Air Act, a
single certification would be allowed.
11.2 RECORDKEEPING
Comment:
Several commenters (IV-D-7; IV-D-17; IV-D-21; IV-D-48) objected to the
proposed requirement for keeping records, technical data, and logs available
11-5
-------
for EPA inspection at the manufacturing site. One commenter (IV-D-17) went
on to refer to Section 114 of the Clean Air Act saying that the Adminis-
trator has exceeded his authority by requiring records to be kept on-site
and available for EPA scrutiny. According to the commenter, these data are
not commonly kept on-site, but may be housed at some central corporate
location. The industry was willing to make applicable records available
upon request by EPA. One .commenter noted that, for the purposes of
demonstrating compliance, it is not necessary to keep the required data
together at one location. Another of the commenters (IV-D-48) asked that
the requirements be modified to allow an owner/operator to supply the
information within a reasonable period of time. He went on to say that EPA
had not demonstrated that the failure to keep such information in the manner
proposed will frustrate the purpose and intent of the Clean Air Act.
Response:
Section 114 provides that, with respect to determining whether any
person is in violation of a standard, EPA may require any person who owns or
operates any emission source or who is subject to any requirement of the
Clean Air Act [other than a manufacturer subject to the provisions of
section 206(c) or 208] to establish and maintain records; to make reports;
to install, use, and maintain monitoring equipment or methods; to sample
emissions (in accordance with methods, locations, and intervals prescribed
by the Administrator); and to provide other information EPA may reasonably
require. Section 114 further provides that EPA shall have a right of entry
to, upon, or through any premises owned or operated by persons subject to
requirements of the Clean Air Act or to premises in which records required
to be maintained are located. EPA may at reasonable times have access to
and copy any records, inspect any monitoring equipment and methods required
and sample any emissions which an owner or operator is required to sample.
EPA has determined that records are necessary to determine compliance with
the standards. The standards state (§60.486) that the required records must
be kept in a readily accessible location. Section 114 gives the
Administrator authority to require and inspect such records at the site and
the General Provisions (§60.7(d)) require that the records must be recorded
11-6
-------
in a permanent form, suitable for inspection. EPA's intention in calling
for a readily accessible location is to facilitate plant inspections by
having records which can be used to judge compliance on the plant site.
After consideration of the burden associated with these records, EPA
concluded that the recordkeeping requirements are reasonable.
Information needed to document compliance with the final standards
consists of various sets of information, depending upon the type of
standard. For equipment and design standards, detailed information of
system design and design changes must be maintained to ensure that the
desired design criteria are met. For equipment standards such in
§60.486{e), criteria for barrier system failure must be recorded and updated
if that criteria changed. Recording the required information for work
practice standards (leak detection and repair programs) in particular, is
not burdensome because the manager of such programs needs this information
to manage the program effectively.
Keeping the records at the plant site does not represent a hardship
because the information must be generated at the plant. This does not mean
that the records must be maintained within the process unit, but they should
be readily available for review during an unannounced inspection. If the
records are to be stored at another facility in another location, they would
have to be shipped there from the plant for storage and would not be
available at such an inspection.
Comment:
Another commenter (IV-D-24) preferred recordkeeping as a method of
measuring compliance instead of reporting. He said that rather than EPA
specifying reporting requirements, each plant should be allowed leeway to
formulate its own recordkeeping methods to assure compliance with the
regulation.,
Response:
As discussed in the previous section on reporting, the periodic
reporting requirements have been reduced to semiannual reporting in the
promulgated regulation.
11-7
-------
As authorized under Section 114 of the Clean Air Act, Section 60.486 of
the standards specifies the information that must be recorded in order to
determine compliance. The information to be recorded is information that
would be required by the manager of the leak detection and repair program in
administering it. Furthermore, this information is needed for enforcement
personnel to determine compliance with work practice standards. These
records serve as the primary tool to prepare the semiannual reports upon
which to base a compliance determination. While the information to be
recorded is specified, the standards do not prescribe a format for recording
this information. The required information may be recorded in any useful
form.
Comment:
Another commenter (IV-D-7) said that EPA's need to review records
should not be a continuing requirement and that EPA should request
information under Section 114 of the Act.
Response:
Compliance with work practice standards is determined by inspection and
review of records. Thus, to judge compliance, EPA will need to review the
records. Section 114 of the Clean Air Act gives the Administrator
authority, beyond requesting information, to inspect the required records at
any time. The required recordkeeping is a continuous process and, as
explained in the response to the previous comment, EPA would inspect the
records when performing a plant inspection. Contrary to the apparent belief
of the commenter, EPA has no reason to believe that information would be
recorded if it is not required by the standards.
Comment:
Records pertaining to equipment installed for compliance with equipment
standards were considered unnecessary by one commenter (IV-D-17). This
commenter recommended that inspection of equipment for proper operation is
all that is necessary to ensure compliance. He said that the equipment
records would be an additional cost to industry and would result in no
decrease of fugitive emissions.
11-8
-------
Response:
It is not clear to what recordkeeping provisions the commenter is
referring. The only recording requirements associated with equipment
standards is in §60.486(g) [§60.486(e) of the proposed standards] which
requires that the criterion for failure of a barrier fluid system be
recorded and any changes in that criterion noted. The commenter may have
been referring to the recordkeeping provisions [§60.486(d); §60.486(c) of
the proposed standards] for design standards which require detailed
schematics, design specifications, and piping and instrumentation diagrams
for closed vent systems, enclosed combustion devices, and vapor recovery
systems.
As required by both sections, the records document how the equipment
will achieve the required design level and describe any design changes. The
records are necessary since compliance cannot be determined without them.
For example, the specific design arrangements of a dual seal system cannot
be established without a shutdown of equipment; thus, without the
information on the seal system design, compliance cannot be determined.
Therefore, these records are necessary for EPA to determine whether the
equipment is in compliance with the standards.
Comment:
One commenter (IV-D-48) wrote that, since the purpose of the proposed
standards is to control fugitive VOC emissions, he failed to understand the
logic or the necessity for requiring data recording beyond information about
the date of testing for monitored sources which do not leak.
Response:
It appears that the commenter is referring to the recordkeeping
provisions for leak detection and repair programs. The final standards
require recordkeeping for only those sources found leaking under a leak
detection and repair program. For those sources complying with equipment
standards, inspection reports would be unnecessary; therefore, routine
recordkeeping requirements like those for work practices are not required
for sources complying with equipment standards.
11-9
-------
11.3 TAGGING
Comment:
Two commenters (IV-D-17; IV-D-25; IV-D-50) objected to the tagging
requirements in the proposed standards. Two of the comment letters
(IV-D-17; IV-D-50) from the same commenter recommended tagging only for
those valves which were not repaired immediately upon detection of a leak.
The commenter also objected to logging that valve as a leaker. The other
commenter (IV-D-25) thought that tagging was altogether unnecessary,
preferring to rely on records to insure that leaking sources receive the
required attention.
Response:
Tagging is not specifically required by the standards, as proposed; any
form of weatherproof and readily visible identification is acceptable. This
identification is required to allow ready location of leaking sources by the
plant personnel or by EPA inspectors. Identification of leaking sources is
an integral part of an inspection program. Without identification it would
be very difficult to locate each valve which requires follow-up monitoring.
It would also be difficult to find valves which are leaking but awaiting
shutdown for repair. Tagging appears to be a useful method of identi-
fication, and tags have been used in leak detection and repair programs, but
any form of weatherproof and readily visible identification is acceptable.
If a process unit has a system of identifying markings on valves and a
diagram is available which allows easy location of the marked valves, the
system would be acceptable.
11.4 OTHER PROVISIONS
Comment:
The provision for requesting permission to delay a repair past 15 days
was said to be unreasonable (IV-D-17). The paper work to be processed was
estimated to be unmanageable. The commenter suggested that records of
reasons for unsuccessful repair, date of detection, and expected repair date
would serve to insure compliance.
11-10
-------
Response:
There is no provision requiring the plant owner or operator to request
permission to delay a repair past 15 days. However, the standards require
that "repair delayed" be recorded in the log for that particular fugitive
emission source if the repair is delayed beyond 15 calendar days after the
date of detection. These records would be needed to assess compliance with
the work practice standards.
11.5 BURDEN ESTIMATES
Based on analyses of industry projections and current methods of
operation, an estimated 166 new respondents per year will be required to
submit reports and begin maintaining records. The annual average number of
respondents for 1983 and 1984 will be 581. The average annual labor
requirement for 1983 and 1984 will be about 65 person-years. Costs to the
Federal government are expected to be $565,342.
The Information Collection Request, SF-83, and Supporting Statement
submitted to OMB are filed in IV-H-2.
11-11
-------
-------
12. TEST METHOD
The comments received regarding the proposed test method were concerned
with three major areas: selection of calibration gas, instrumentation, and
instrument responses to different chemicals. In addition to these principal
areas, comments were received on equivalency for test methods, leak defini-
tion, detection methods, and screening methods.
12.1 CALIBRATION GAS ' ,-
Comment: , ; , ,
Two commenters (IV-D-6; IV-D-7) objected to the use:of methane ,as the
calibration gas for Reference Method 21. One commenter (IV-D-6) said that,;-.
since the. refinery studies on which the proposed SOCMI studies were based
had been conducted using hexane, hexane should also be the basis of the
SOCMI standards. This commenter said the use of methane would result in
more leaks than would the use of hexane. Another commenter (IV-D-17) agreed
and suggested that the leak definition should be revised.to compensate for
the difference. According to one commenter (IV-D-21), the calibration . .;
gas/instrument differences between the petroleum refinery and SOCMI studies,
added to the lower leak frequencies found in SOCMI units, clearly show
fundamental differences between the two industries. . ,
Response: L
The SOCMI fugitive emissions standards are based on data collected on
SOCMI units (see Section 3.2). These data were collected using an
instrument (OVA) calibrated with, methane. The final regulatory analysis was
based on SOCMI screening data measured with an instrument calibrated to ,
methane. This analysis supplements the preliminary one based on refinery
screening data measured with an instrument (TLV) calibrated to hexane.
Since EPA's final determinations were made based on data gathered in SOCMI
using methane as the calibration gas for an OVA, no compensation in. leak
frequency or leak definition is necessary.
12-1
-------
Ultimately, however, the differences are not significant. The vari-
ability seen in repeat sampling of the same source was 23 percent (IV-A-7).
This variability is in the same range as the 30 percent difference seen in
response between the TLV-hexane system and the OVA- methane systems at the
10,000 ppmv action level (IV-A-8). Because the variability in repeat
sampling is so similar to the differences in response_.at 10,000 ppmv, the
data can be used interchangeably within ±30 percent at the action level.
Accordingly, EPA has added hexane as an alternate calibrant.
Comment:
One commenter (IV-D-6) said that VOC definition does not adequately
reflect the capabilities of the reference method. He said the method allows
the use of photoionization devices which will not respond to the calibration
gases specified in the method. He continued his argument saying that, if an
instrument cannot be calibrated, it should not be used to measure emissions.
Response:
The reference method has been written to be applied to fugitive
emission source screening in general, with specific application requirements
being established in each regulation. The method states only that photo-
ionization instruments might meet the requirements, but it does not state
categorically that photoionization instruments may be used. However, since
this type analyzer may be useful in certain SOCMI process units, and it does
not respond to methane, an alternate calibration material has been added in
the regulation.
As discussed in the previous comment response, the variability in
response between a hexane-calibrated instrument and a methane-calibrated
instrument is similar to the variability in repeat sampling of the same
source. Data gathered using either systems were determined to be inter-
changeable. Furthermore, hexane is an appropriate calibrant for photoioni-
zation analyzers since they will respond to hexane at an ionization
potential of 11.7 - 11.8 yeV. Therefore, adding hexane as a specified
alternate calibration material allows calibration of photoionization
instruments.
12-2
-------
The use of the alternate calibration material is not limited to a
single type of analyzer. The.owner or operator may choose to use either
calibrant with any allowable instrument. But in any performance test by EPA
or a State agency, methane may be used as the calibrant even if it,is
different from the one selected by the owner or operator.
Comment:
Another commenter (IV-D-27) suggested.that the gas specification.
section be amended to include a turnover of calibration gas standards every
3 months since calibration gases can deteriorate significantly over time.
Response:
Calibration gas mixtures could be subject to deterioration with time.
Good analytical procedures would dictate periodic checks of gas concen-
trations and changing or turn-over of the calibration gases when necessary
to assure the quality (integrity),of the monitoring. Since proposal, a
provfsion has been added to Reference Method 21 which addresses shelf-life
of calibration gases and procedures to follow to ensure that calibration gas
concentrations are accurate.
12.2 INSTRUMENTATION
Comment:
Two comments (IV-D-15; IV-D-17) concerned the instrumentation require-
ments of Reference Method 21. One commenter (IV-D-17) stated that only two
instruments on the market today could be considered-, and neither one would
meet the specifications of the reference method entirely:, the first
instrument fails the calibration accuracy and the second instrument does not
meet the response time requirements. The commenter further stated that,
considering pros and cons, the first instrument is better suited to finding
fugitive leaks, resulting in five percent more leaks than the second. The
commenters considered this difference in detection capability to be more"
important in determining overall efficiency than leak definition or
inspection interval.
Response:
Since proposal, the instrument specifications have been revised. The
instrument specifications given in the revised Reference Method 21 are based
12-3
-------
on performance during SOCMI screening studies and on comments received by
EPA during the development of the method. The maximum instrument response
time is 30 seconds, and the calibration precision must be less than or equal
to 10 percent of the calibration gas value. This means that the instruments
referenced by the commenters can achieve the specifications.
As discussed above, in Section 3 and in the AID, the standards are
based on data collected on SOCMI emission sources using the OVA. But any
instrument which meets the requirements discussed by the commenter should
provide adequate leak detection when used in accordance with Reference
Method 21.
Comment:
Another commenter (IV-D-15) stated that 20 percent of the time during
screening studies was devoted to calibration and maintenance of the instru-
ment. The potential instrument problems indicated by such high time utili-
zation were not discussed in the BID or in the reference method.
Response:
The data presented in the referenced problem-oriented report,
"Frequency of Leak Occurrence for Fittings in Synthetic Organic Chemical
Process Plant Units," indicated that from 1 to 1-1/2 hours per day were
required for calibration and maintenance of the monitoring instrument.
Based on the number of sources that were screened in each unit, this cali-
bration time amounted to 16-25 percent of the total time for screening.
The calibration time during the EPA studies was expected to be longer
than -for screening alone since concentration measurements were being
recorded that would be .used in further analysis. This necessitated
calibration with more than one standard concentration on a more frequent
basis (2-3 times daily). Also, because concentrations up to 100,000 ppmv
were being measured, a dilution probe had to be calibrated several times
daily. Routine screening would require calibration with only one standard,
one time per day. Also, a dilution probe would not be required.
Calibration time for routine screening is estimated to require about 10 to
25 percent of that required during EPA tests, or about 2 to 6 percent of the
total time for screening.
12-4
-------
The referenced problem-oriented report also listed a number of
problems, equipment-related and procedure-related, encountered during the
SOCMI 24-Unit Study. Procedure- related problems, as well as equipment
problems, to a Targe extent were due to performing work in the field which
was remotely located from laboratories and repair facilities. Problems with
instruments in the field take longer to fix because of shipping delays.
Furthermore, many inefficiencies were encountered because the personnel
performing the studies were research staff who spend only short periods of
time on projects of this type, thereby lacking the experience necessary for
troubleshooting.
Time lost due to equipment failure is expected to be minimized by
maintaining the critical spare instrument parts (including readout meter,
battery pack, regulator repair kit, pressure gauges, hydrogen flow valvesi
and filters) identified during the 24-unit study. Additionally, personnel :
familiar with troubleshooting procedures should facilitate instrument 'Opera-
tion and repair. The proximity of instrument shops and labs should also
improve routine screening efficiency.
The cost estimates in the BID include costs for two instruments, one of
which was considered a spare. Having a spare should decrease instrument
downtime. Moreover, an ample allowance of time was made for calibration and
maintenance of instruments. The 40% administrative and overhead charge
allotted includes time for calibration, and an additional $3,000 per year
was allotted for instrument maintenance. In view of these differences
between research field studies and routine screening activities, the
calibration and maintenance costs allotted in the BID are reasonable and no
adjustments have been made.
12.3 VARIABLE RESPONSE TO DIFFERENT CHEMICALS
Comment;
Several cotnmenters (IV-D-6; IV-D-7; IV-D-17) noted that the instruments
used during screening studies responded differently for different chemicals.
One commenter (IV-D-6) stated that the actual response factor was poorly
related to the theoretical response factor and cited inconsistent responses
for nonane and decane, as well as no response for some chemicals, to support
12-5
-------
his claims. Another commenter (IV-D-7) suggested that the leak concentra-
tion for the standards should vary according to the unit since such wide
variability (0-571) in response factors have been determined for the
industry. He disagreed with EPA's use of a response factor of 2-3 applied
to potential leaking sources. Any comparison should be made to the number
of leaking sources, not potential leaking sources. Another commenter
(IV-D-17) stated that aromatic compounds such as benzene, toluene, and
xylene demonstrate a non-linear response close to 10,000 ppmv.
Response:
Reference Method 21 gives specifications for the instrument to be used
in monitoring fugitive VOC emission sources. The technique is intended to
classify leaks only, not to provide a rigorous analytical concentrations of
VOC. A specific statement has, therefore, been added to Reference Method 21
to clarify the intention to classify leaks only.
The variation in response factor, due to compound or instrument, is not
expected to affect significantly the number of leaks determined through
screening because screening values are usually greater than 10,000 ppmv for
leaks and much less than 10,000 ppmv for non-leaks. Two industry commenters
concur with EPA in this position (IV-D-17; II-D-72). However, to remove
some of the wide variability, a definition, specification, and test
procedure for response factors has been added to Reference Method 21.
Laboratory experiments using two VOC analyzers indicated a wide
variation in response factors for a number of organic chemicals. The range
as indicated by the commenter was 0-571. However, 90 percent of the
chemicals tested had responses between 0.1 and 10 (IV-A-8; IV-A-12;
IV-A-15). Most of the remaining 10 percent are solids or heavy liquids.
Differences were also seen between the two types of analyzers tested. When
considered in analyzing leak frequencies (IV-A-14), the response factor
variation, however, did not produce significant changes in the percent
leaking estimates resulting from the SOCMI 24-Unit Study (II-A-21).
Although a small reduction in the estimated leak frequencies is indicated
for gas valves in high leak service, the estimates in all other cases were
almost indistinguishable from the unadjusted estimate. Furthermore, the
12-G
-------
differences when present were in the same range as the variation in reprodu-
cibility described in IV-A-7. Table 12-1, reproduced from the analysis
report (IV-A-14), illustrates the effects of response factor variation on
leak frequency.
12.4 OTHER;COMMENTS
Comment:
One comment letter (IV-D-17) expressed concern that no provision was
made for the use of new instruments or calibration procedures which would
provide equivalent or more accurate results. They asked that equivalency
provisions be added for test methods and procedures.
Response:
Reference Method 21 gives specifications for the monitoring instrument
that are general enough so as not to preclude new analytical developments.
-,;,
In addition, the General Provisions (40 CFR Part 60, Subpart A) allow for
equivalent methods and procedures to be used for performance testing and
monitoring when the results of the equivalent method have been demonstrated
to be at least as accurate as results obtained by the required methods.
Comment:
One commenter (IV-D-27) suggested that use of a windscreen upwind of
the component being screened would prevent meteorological effects on the
instrument readings.
Response:
The selection of a measurement location at the surface of the source
was made to minimize meteorological effects. During the data collection
efforts, no further provisions were found necessary to obtain repeatable
screening values. Therefore, all of the field data were collected without a
windscreen. In view of these facts it is unnecessary to require that a
windscreen be used while monitoring.
Comment:
One comment letter (IV-D-17) cited wide variability in repeat screening
values from an EPA Report (contract #68-03-2776) as justification for
raising the leak definition. The commenter said that the trigger point
should be set high enough to insure isolating only the bad leakers.
12-7
-------
TABLE 12-1. COMPARABLE ESTIMATES FOR PERCENT LEAKING (VALVES)
(24 SOCMI Process Units)
Process
Stream
Gas
Light
Liquid
Reference:
Number ,
Streened
9374
18133
Percent Percent
Leaking Leaking
Based on Based on
OVA Method 1 .
Readings Adjustments
11.3 10.1
6.1 5.3
Percent
Leaking
Based on
Method 2 -
Adjustments
10.2
5.6
Percent
Leaking
Based on
Method 3 4
Adjustments
10.3
5.5
Analysis of SOCMI VOC Fugitive Emissions Data,
EPA-600/2-81-111 (IV-A-14).
"Method 1 is the adjustment to the OVA Reading based on the response of the
primary chemical in the line.
Method 2 is the mixed chemical-weighted logarithmic average technique for
the primary and secondary chemicals in the line (IV-A-15).
j
Method 3 is the mixed chemical-weighted average technique for the primary
and secondary chemicals in the line (IV-A-15).
-------
Response:
The commenter is referring to a report on maintenance for control of
fugitive VOC emissions filed in IV-A-7. Quality Control/Quality Assurance
procedures and measures for the study are reported in Appendix C of the
report. The statistical measures generated for repeatability of screening
values are measures of repeatability of precise screening values (g^r to
5.6X). However, a more relevant measure of repeatability is one which
measures the reproducibility of an on-off (leak-no leak) measurement, not
one which measures reproducibility of a precise number. This reproduci-
bility was quantified and reported in a memo filed in II-B-24. The
information presented in this docket entry shows a 90% repeatability of a
leak-no leak determination at a 10,000 ppmv action level. In a recent study
of fugitive emission sources in the South Coast Air Quality Management
District in California, greater than 94 percent repeatability of leak-no
leak determination was found for duplicate screening. Thus, EPA finds no
basis for raising the leak definition because of the test method. The issue
of leak definition is addressed in more detail in Section 4.3.
Comment:
One commenter (IV-D-29) submitted several comments about fixed-point
area monitors. He described his plant's fixed-point area monitoring system
as a fugitive emissions detection device which protects their people and
their environment. He explained that the alarm points are based on TLV's
which should reflect exposure levels below which no harm results.
The commenter disagreed with EPA's conclusions about fixed-point area
monitors. He objected to the analysis of testing data for fixed-point
monitors gathered in his plant. He said one problem existed in the fact
that the study was conducted over a short period of time. Therefore, the
area monitors were not subjected to significant wind shifts. The system
depends on periodic wind shifts to detect small leaks such as the ones under
consideration.
The commenter voiced a strong objection to EPA's conclusions that the
area monitoring system did not pick up leaks detected on the walk-through
survey. He said that this was misleading because the walk-through did hot
12-9
-------
pick up some leaks detected by the area monitors. He thought that this
situation existed because the survey was done at ground level.. He pointed
out that there were fugitive emission sources on structures and columns
which were not accessible without scaffolding.
Response:
EPA's evaluation of a fixed-point area monitoring system is filed in a
test report on fugitive emissions from chloromethanes production (IV-A-5).
Three monitoring methods were evaluated in the tests described: individual
component survey, walk-through, and fixed-point area monitoring. The report
clearly confirms the commenter's statement that since some equipment items
were not readily accessible, the fixed monitors might indicate leaks where
individual component testing could not detect them. Nevertheless, the
report also indicates that out of 22 leaks identified by the individual
component survey, only 4 were detected by the area monitors. As indicated
in the report, the low leak detection effectiveness may have been due in
part to the use of TLV's as alarm points; setting the alarms at lower
concentrations may have increased the effectiveness.
This decreased leak detection effectiveness was only part of the
reasoning for not selecting fixed point area monitors as the test method for
use in detecting fugitive emissions. As explained in the BID, the reasons
that area monitors were not selected included meteorological influences, the
expense of the equipment, and the necessity of also performing individual
equipment surveys. Furthermore, it was impossible to specify a system which
would effectively detect leaks in all equipment configurations and under all
meteorological conditions found in the industry.
This does not mean that area monitors have no place in the determi-
nation of fugitive emissions. There may be cases in which area monitors may
be effective emission detection devices because they continuously monitor
for high concentrations. An owner/operator may find a system of fixed point
area monitors especially useful in combination with other fugitive emission
control measures if he chooses to comply with an alternative.
The walk-through survey mentioned by the commenter was also considered
as a test method, but as explained in the BID, it was not selected. The
12-10
-------
results obtained in walk-through surveys were mixed, and are affected by
meteorological conditions (IV-A-18). Furthermore, an individual equipment
survey is required when a high concentration is indicated. However, as is
the case with fixed-point monitors, a walk-through survey system may be a
useful complement to other leak detection methods if an owner/operator
chooses to comply with an alternative.
Comment:
The same commenter (IV-0-29) said that screening data was in no way
related to exposure data. Screening data, he said, was literally a sample
taken in direct contact with a leak. The commenter did not see how a
screening value determination could be extrapolated to effect on the
environment.
Response:
Screening values have been shown to be accurate indicators of leakage
of VOC from equipment. On an average basis, mass emission rates can be
correlated with screening values. If the proportion of sources with
screening values in excess of the leak definition can be reduced, then it
follows that reductions in mass emissions will be achieved. Reductions in
mass emission rates are used to determine best-demonstrated technology
(considering cost), which is the basis for the Standards of Performance.
12-11
-------
-------
13. ENFORCEMENT AND COMPLIANCE CONCERNS
Several commenters expressed interests in enforcement and compliance
concerns. Basically, three areas of comment were presented: (1) resource
requirements for local enforcement, (2)- compliance with the standards for
pressure relief devices, and (3) delay of repair provisions. Three facets
of the delay of repair provisions received comment: unavailable spare
parts, technically infeasible situations, and out of service spare
equipment.
Comment:
One commenter (IV-D-11) stated that local enforcement of the regulation
would require a more intensive use of resources than envisioned.
Response:
In assessing the relative impacts of reporting alternatives, both
industry and enforcement agency requirements were evaluated. The background
information presented in the Federal Register with the proposed standards
(46 FR 1136) gave the industry requirements for reporting as 53 man-years in
1985 (the fifth year) for 830 affected facilities. The corresponding
requirements for enforcement agency review in 1985 were estimated at less
than 4 man-years, or about 7 percent of the estimated requirement for
industry. This estimate included 8 man-hours annually for review of the
quarterly reports submitted for each affected facility. These figures were
filed in. the Docket in the Reports Impact Analysis (II-A-30).
Since proposal, however, the reporting requirements for the SOCMI
fugitive VOC NSPS have been reduced. Records must still be maintained to
the extent necessary to demonstrate compliance with the standards. These
revised recordkeeping and reporting requirements have been estimated at an
average of 65 person-years annually for 1983 and 1984. Even with the
reduction in reporting requirements, however, local enforcement activities
such as on-site inspections may not increase. Thus, no additional burden is
13-1
-------
expected to be placed on local enforcement. The burden calculations for VOC
fugitive emissions in SOCMI are filed in the Docket (IV-H-2). .
Comment:
One commenter (IV-D-17) stated that there was a loophole in the
standard for pressure relief devices. As proposed, the standard requires
each pressure relief device in gas/vapor service to return to a state of "no
detectable emissions" no later than five days after an emergency release.
The commenter pointed out that if a process unit is down five days after the
release, the relief device, by definition, will be in compliance. The
regulation should say that the pressure relief device should be returned to
"no detectable emissions" five days from resumption of normal operation
after each episode of pressure release.
Response:
The standards for pressure relief devices (§60.482-4) are:
(a) Except during pressure releases, each pressure relief device in
gas/vapor service shall be operated with no detectable emissions, as
indicated by an instrument reading of less than 500 ppmv above
background, as determined by the methods, specified in §60.485(c).
(b) (1) After each pressure release, the pressure relief device shall
be returned to a condition of no detectable emissions, as
indicated by an instrument reading of less than 500 ppmv above
background, as soon as practicable, but no later than 5 calendar
days after the pressure release, except as provided in §60.482-9.
(2) No later than 5 calendar days after the pressure release, the
pressure relief device shall be monitored to confirm the
conditions of no detectable emissions, as indicated by an
instrument reading of less than 500 ppmv above background, by the
methods specified in §60.485(c).
(c) Any pressure relief device that is equipped with a closed vent
system capable of capturing and transporting leakage through the
pressure relief device to a control device as described in §60.482-10
is exempted from the requirements of §60.482-4(a) and (b).
13-2
-------
The pressure relief device has the potential to emit VOC whether the
unit in which it is located is in operation or not, because the equipment
and lines may still contain VOC. Therefore, if the process unit is down
five days after the release, the pressure relief device may well be in
compliance with the regulation. But even if the unit is down, the potential
still exists for a pressure relief device to emit VOC in excess of 500 ppmv
above the background.
The standards state that the pressure relief device shall be returned
to a state of no detectable emissions after each pressure release as soon as
practicable. Clearly, this requirement is intended whether the unit has
been down or continued to operate after the release. For example, if the
unit has been down for several days after an incident of pressure release,
the pressure relief device must still meet the requirements of no detectable
emissions when the unit starts up. The pressure relief device must always
achieve the 500 ppmv limit except for the time (no greater than 5 days)
required to return the pressure relief device to less than 500 ppmv after a
release. Of course, if a unit is scheduled to be down for a long period of
time, an owner or operator could elect to remove the device from VOC
service.
Comment:
Several commenters (IV-D-17; IV-D-48; IV-D-50; IV-D-51; IV-D-34) were
concerned that the proposed standards do not allow repairs to be delayed
past a shutdown. Their concerns related to delayed repairs due to
unavailable replacement parts and technically infeasible situations and on
the repair requirements for spare equipment.
They recommended a limited extension provision to avoid those limited
instances where replacement of leaking equipment might not be possible until
after a process unit shutdown. The consequences of not including such a
provision, they said, would be unanticipated and costly continuances of
shutdown or risk of criminal or civil penalties for resuming operation
without repairing all leaks.
Several reasons for occasional necessary delays were offered by
commenters. One reason cited was the delay associated with obtaining
13-3
-------
custom-made or unique parts (IV-D-51; IV-D-48; IV-D-17). Another (IV-D-51;
IV-D-48; IV-D-17) was unforeseen depletion of inventoried parts just prior
to a shutdown. Still another (IV-D-48; IV-D-51) was the occurrence of
unscheduled and unforeseen shutdowns of a duration too short to allow the
opportunity for repairs. One commenter (IV-D-51) pointed out that repairs
normally require the process be cooled, emptied, and inspected to permit
safe working conditions. He said that these procedures can require 24 or
more hours to complete.
Since extensions are allowed for situations qualifying as technically
infeasible, two commenters (IV-D-17; IV-D-48) recommended that a definition
of "technically infeasible" be included to minimize uncertainties and to
reduce unwarranted enforcement proceedings resulting from ambiguities in the
regulations. They recommended that the following definition be added:
For purposes of §60.482(h), technically infeasible shall
mean where a repair within 15 days of leak detection would
constitute an unsafe practice, could result in premature
total process failure, could cause an unscheduled complete
or partial process unit shutdown or if the temporary emission
resulting from the repair would exceed the emissions from the
continued leak.
i Another commenter (IV-D-51) recommended the following amendment to the
proposed §60.482(h):
Delay of repair will not be allowed beyond the first
scheduled process unit shutdown unless bonafide attempts
to secure spare parts in time for the scheduled, shutdown
have failed.
In a related matter one of the commenters (IV-D-17; IV-D-50) in two
comment letters said that if a leak is detected in a pump or compressor with
an installed spare, the five day-fifteen day repair requirements should not
apply to the pump or compressor taken out of service, if the equipment is
properly purged and isolated from the process. They recommended that for
this case the regulation should simply state that the equipment be repaired
prior to coming on-line and be tested within fifteen days of startup.
13-4
-------
Response:
The intent of the standards is to reduce fugitive VOC emissions. To
meet this end, all available" approaches for repair of leaking sources should
be used. EPA recognizes that, in a few cases, repair of leaking sources may
have to be delayed beyond a process unit shutdown. Several provisions have
been added to the standards to accommodate those cases.
The first provision allows delay of repair where repair is technically
or physically infeasible without a process unit shutdown. Technically or
physically infeasible means that all safe repair procedures short of
shutting down the unit have been tried and the valve is still leaking. An
example of such a situation would be a leaking valve that could not be
isolated from the process stream to replace internal parts that would likely
repair the valve. Tn this case, the process unit would have to be shutdown
to effect repairs on the valve, since the valve could not be physically
isolated from the process stream. Once the process unit is shut down for
any reason, the valve must be replaced.
The second provision was added to clarify EPA's intent for spare
equipment that is out of service. This provision would be applicable only
to those pieces of equipment that have been isolated from VOC service and
properly purged. Delay of repair would not be allowed for spare equipment
that was pressurized and prepared to be placed on-line; such equipment is
considered to be in (VOC) service.
A third provision was added for those situations in which it is
possible to isolate a valve from the process for repair, but doing so would
cause higher emissions than allowing the leak to continue. An owner or
operator, to delay repair under this provision, must demonstrate that1
emissions of purged material resulting from immediate repair are greater
than the fugitive emissions that are likely to result from delay of repair.
Furthermore, when repair procedures are effecte.d, the purged material must
be collected and diverted to a control device.
Provisions have also been added for delay of repair for pumps operating
under work practice standards. In a case in which replacement of mechanical
seals does not "repair" the pump, the only way to repair it is to install
13-5
-------
equipment (dual seals with barrier fluid systems or vented seal areas). Six
months is a reasonable length of time for installation of such equipment
(IV-B-Z9); therefore, the standards allow for a 6-month delay in installing
equipment necessary to "repair" the pump.
A final provision has been made for delay in limited instances due to
the lack of spare valve assemblies. It is not, however, the intent of the
standards to permit delays resulting from poor administrative practices.
Two such practices that are not sufficient to substantiate a delay of repair
are unavailable maintenance personnel and the lack of readily available
valve assemblies that were not sufficiently stocked. Custom order or unique
valve assemblies must be sufficiently stocked to avoid delays due to their
inavailability. The additional cost of keeping a sufficient stock of
readily available spare parts is not unreasonable because the additional
stock of spare parts needed for this standard are small in comparison to the
stocks of spare parts commonly stored within this industry. Stocking or
contracting for "quick" supply of unique parts is an economic necessity
within' this industry. Even though EPA believes that this delay of repair
provision would be appropriate only under unusual circumstances, the
commenters1 concerns are valid and the provisions have been added. In
addition, if spare valve assemblies are not in stock, they must be obtained
and must be used at the next process unit shutdown which occurs 6 months
after the initial delay of repair (IV-B-30).
To make allowance for delay of repairs beyond an unscheduled shutdown
in the case of shutdowns of too short a duration to effect repairs, a
process unit shutdown has been defined as longer than 24 hours.
"Process unit shutdown" means a work practice or operational
procedure that stops production from a process unit or part
of a process unit. An unscheduled work practice or opera-
tional procedure that stops production from a process unit
or part of a process unit for less than 24 hours is not
a shutdown. The use of spare equipment and technically
feasible bypassing of equipment without stopping production
are not process unit shutdowns.
13-6
-------
In cases of unscheduled shutdowns of shorter than 24 hour duration,
repair of leaking equipment would be required at the next scheduled process
unit shutdown.
13-7
-------
-------
14. ALTERNATIVE STANDARDS
Several sets of comments were received which addressed the alternative
standards for valves in gas/vapor and light liquid service. The comments
fall into basically three categories: (1) clarification of the provisions1
and requirements of the alternative standards, (2) procedures for
determining and demonstrating equivalency, and (3) skip-period monitoring.
Comment: >
Two commenters (IV-D-17; IV-D-23) asked that the criteria for'approval
or disapproval of a request for an alternative standard be made clear in the
regulation. In a previous letter, one of the commenters (IV-D-17) stressed
that these guidelines be specific so that approval decisions would not be :
open to interpretations. The commenter said difficulties had been ;
experienced in the past when EPA Regional office decisions required inter-
pretation. This same commenter (IV-D-17) requested clarification in the
regulation concerning what constitutes a violation of an alternative
standard and recommended specific language to be added to §60.483. to provide
the clarification.
The commenter also noted that, for any number of reasons, at some time.
it may no longer be cost effective for a source to comply with an
alternative standard. He pointed out that a mechanism should be provided to
allow a source to return to the valve standard given in §60.482(f) by
notifying the Administrator in writing. One comment (IV-D-17) stated that
inaccessible valves create problems in calculating meaningful percentages of
valves leaking and, therefore, should be regulated under another provision.
The same commenter stated that a monthly average was the wrong measure
for judging equivalency. He recommended a 95 percent confidence bound,
determined during the year-long data collection program, as the measure of
equivalency. Without such a provision, equivalency would not be possible.
The commenter also said that, since the BID acknowledges differences between
14-1
-------
annual and monthly numbers, a correction factor should be provided in order
to determine equivalency from annual monitoring data. He suggested the
annual average be divided by 12 to compare with the monthly allowable
average.
In a subsequent letter, the same commenter (IV-D-50) recommended
reduced monitoring requirements for plants demonstrating consistent
attainment of leak rates of 2 percent or less. In so doing, incentives
would be provided for the adoption of more cost-effective alternative
good-performance levels and low-leak plant designs.
Several comments were received on statistical approaches to leak
detection programs. Many of the commenters (IV-F-1, No. 4; IV-D-1; IV-D-17;
IV-D-28; IV-D-47) suggested that statistical inspection plans, such as
skip-period monitoring plans, be considered. These programs have been
demonstrated to be effective quality control techniques and would provide
adequate protection against leaks. The commenters also said skip-period
plans would minimize costs by reducing inspections when good performance was
achieved and demonstrated.
Several of these commenters (IV-F-1, No. 3; IV-F-1, No. 4; IV-D-17;
IV-D-47) pointed out that quality in reducing emissions from a unit is
obtained through good design. And they felt that leak detection and repair
programs were a disincentive to designing and installing low-leak units.
Two of these commenters (IV-F-1, No. 4; IV-D-17) noted that the currently
proposed alternative standards would be more difficult for current low-Teak
units to meet, since the standards call for improving current performance
levels.
One commenter recommended adding a third alternative standard for
valves based on papers presented at the NAPCTAC meeting. These papers
indicated that a 2 percent good performance level-was adequate to ensure low
levels of emissions. In another letter, the commenter (IV-D-47) suggested
skip-period plans be considered for SOCMI since they had been included in
the NAPCTAC package for petroleum refineries (June 2-3, 1981).
Two sets of comments were concerned with various aspects of determining
equivalency with the alternative standards.
14-2
-------
Response:
At proposal, two alternative standards were presented for valves in
gas/vapor and light liquid service. Both of these alternatives called for
one year of monthly monitoring to obtain data on which to base the alterna-
tive standard. Section 60.483(a) in the proposed standards was a
performance standard based on an allowable percentage of valves leaking.
Since an industry-wide allowable leak percentage was not possible due to
variability of leak frequency among process units, an allowable percentage
of valves leaking was to be determined for that unit based on data collected
on that unit. The allowable percentage was to be the sum of the monthly
baseline percentage and the monthly incremental percentage. A minimum of
one performance test was required annually. Section §60.483(b) in the
proposed standards allowed the development of work practices that would
achieve the same result as the proposed leak detection and repair program
for valves. This alternative would allow a unit to vary the monitoring :
interval and to use valves with a low probability of leaking in order to
achieve an overall goal of emissions reductions.
Based on comments received on the proposed alternative standards and on
analysis of the results from SOCMI screening and maintenance studies
(IV-A-10; IV-A-11; IV-A-14), the alternative standards for valves were
reexamined. As a result, these standards were changed and refined in
response to comments and to reflect the information gathered on SOCMI units.
The first alternative standard was reconsidered by (1) looking at the
cost effectiveness of a monthly leak detection and repair program (see
Section 4.2 for a detailed discussion on monitoring interval selection) as a
function of the percentage of valves leaking initially (IV-B-26) and (2)
comparing the leak frequencies for gas/vapor and light liquid valves
determined in the 24-Unit Study (IV-A-11). Figure 14-1 presents the results
of the first analysis for an average SOCMI unit. Table 14-1 presents the
overall leak frequencies (excluding inaccessible valves) for gas/vapor and
light liquid valves for the 24 SOCMI units screened. As shown in this
table, fifteen of the 24 units demonstrated overall leak frequencies for
valves of less than 2 percent. Whereas shown in Figure 14-1, the cost
14-3
-------
10,000
Note: This analysis includes varia-
tion of occurrence rate and
average input emission factor
with leak frequency. Varia-
tion of leak frequency while
holding these other factors
constant produces spurious
results. See comments and
responses on Chapter 4 of the
AID presented in Apppendix A.
8,000
V)
V)
5 6,000
u
-------
TABLE 14-1. OVERALL LEAK FREQUENCIES FOR VALVES
IN THE SOCMI 24-UNIT STUDY
Percent Leaking of Valves Screened t
Unit
1
2
3
4
5
6
11
12
20
21
22
28
29
31
32
33
34
35
60
61
62
64
65
66
Process Gas and
Vinyl acetate
Ethylene
Vinyl acetate
Ethylene
Cumene
Cumene
Ethylene
Acetone/phenol
Ethylene dichloride
Vinyl chloride monomer
Formaldehyde
Ethylene dichloride
Vinyl chloride monomer
Methyl ethyl ketone
Methyl ethyl ketone
Acetaldehyde
Methyl methacryl ate
Adipic acid
Tri- and
perch! oroe thy lene
1,1,1-trichloroethane
Ethylene dichloride
Adipic acid
Acrylonitrile
Acrylonitrile
Light Liquid"
1.2
18.8
1.5
22.1
9.4 .
13.1
12.6
0.3
0.9
0
0.6
5.8
1.5
8.0
4.4
1.5
0.1
0
0.1
1.1
0
0
3.0
1.2
Gas, Light and Heavy Liquid
1.1
14.2
1.4
21.9
8.2
11.2
12.5
0.3
0.9C
Oc
0.6C
5.8C
1.5C
; 8.0C
4.4C "
1.5C
o.ic '; ; :
0
O.lc '
1.1C
oc
0
2.7
1.2C
Percent is percent leaking of light liquid and gas valves only.
Percent is percent leaking of light liquid, heavy liquid, and gas valves.
GNo heavy liquid valves screened in this unit.
14-5
-------
effectiveness of a monthly leak detection and repair program begins to
increase rapidly for overall valve leak frequencies of one percent or less.
Considering the variability inherent in determining leak frequencies, a
2 percent maximum allowable percentage of valves leaking was determined. As
one commenter suggested, this provided an owner or operator a risk of less
than 5 percent that an average of 1 percent was being exceeded.
Therefore, Section 60.483(a) (renumbered §60.483-1) was simplified to a
2 percent limitation as the maximum percent of valves leaking within a
process unit, determined by a minimum of one performance test annually.
This will provide a standard for valves where the costs of monthly
monitoring would be unreasonable. It would also provide a cost-effective
incentive to maintaining a good performance level and promote low-leak unit
design as was indicated by one commenter. Inaccessible valves, that would
not be monitored on a routine basis under §60.482-7, would be included in
the annual test since an annual test of these valves is not considered
burdensome. By incorporating some of the features of the proposed
§60.483(b), this standard (§60.483-1) provides the flexibility of a
performance level that could be met by implementing any type of leak
detection and repair program and engineering controls chosen at the
discretion of the owner or operator. Even though an industry-wide allowable
leak percentage was not possible for valves, this alternative standard would
allow an affected facility to comply with an allowable percentage of valves
leaking without having to determine a specific performance level by a
year-long monthly monitoring program. If the results of a performance test
show a percentage of valves leaking higher than 2 percent, however, the
process unit would be in violation of the alternative valve standard.
Finally, if an owner or operator determines that he no longer wishes to
comply with this alternative standard, he can submit a notification in
writing to the Administrator stating that he will comply with the work
practice standard in §60.482-7.
EPA also recognizes benefits which may be derived from statistically-
based skip-period leak detection and repair programs. Under skip-period
leak detection, an owner or operator could skip from routine monitoring
K-6
-------
(monthly) to less frequent monitoring after completing a number of
successful sequential monitoring intervals with the prescribed performance
level achieved for each interval within 90 percent certainty. Two options
were considered for an alternative standard based on skip-period monitoring:
(1) a mathematically presented skip-period plan which sophisticated owners
or operators could use for their process units and (2) two readily under-
stood, but specified skip-period programs that conform to the established
performance criteria. Only one of the two set skip-period programs may be
selected for any given process unit and the selection must be made in a
notification of intent to the Administrator prior to implementation of the
alternative standard.
The first skip-period monitoring option provides maximum flexibility
when applied to the widely different process units throughout SOCMI.
Equations have been developed (IV-B-28), based on common skip-lot quality
control plans, that allow a straightforward determination of a monitoring
schedule that incorporates skip-period features. By applying these
equations, an owner or operator can optimize the monitoring schedule (a
combination of consecutive periods monitored and skipped periods allowed) to
suit the particular requirements of his process unit.
In addition, this option allows for a variable performance level that
must be maintained. Since the performance level is based on an inverse
relationship of valve count, a higher performance level would be allowed for
process units with low valve counts.. A fixed performance level, on the
other hand, favors the unit with a large number of valves. A disadvantage
to this option, however, is the .increased burden on enforcement personnel in,
maintaining records on process units with variable skip-period monitoring
plans.
This disadvantage of the first option is eliminated under the second .
option. A single performance level would be selected and two set programs
would be presented to conform to the selected performance level. The
reference leak detection and repair program for skipping is monthly/
quarterly leak detection as allowed by the standards. This means that
monitoring of valves would be the same as required in §60.482-7 but that,
14-7
-------
once skipping begins, monitoring would be based on skipping quarterly
periods. When the performance level is not achieved, however, the monthly/
quarterly program would be reinstituted. In addition, monitoring of
inaccessible valves would be handled as under the reference program required
by §60.482-7. Inaccessible valves would not be counted toward determining
the percent of valves leaking under §60.483-2.
Based on less than 2 percent leaking with clearly greater than
90 percent certainty that in all periods less than 2 percent are leaking,
the following set of consecutive periods and fraction of periods skipped
were determined for SOCMI units:
(1) two consecutive quarterly periods achieved to skip to
semiannual monitoring, and
(2) four consecutive quarterly periods achieved to skip to
annual monitoring.
By providing only two specific programs, some of the flexibility afforded by
tailoring a program to the specific process unit is lost. Another disadvan-
tage to this approach is the establishment of a single performance level,
regardless of the number of valves in the process unit.
However, because of the difficulties which EPA would face in keeping
track of many different skip-period programs and the difficulties in
determining which programs are appropriate for control of leaking valves,
EPA decided to provide two set skip-period plans with a single performance
level of 2 percent leaking as its second alternative standard (§60.483-2)
for gas/vapor and light liquid valves. This alternative (§60.483-2) would
provide two choices and would thereby provide flexibility while decreasing
the potential confusion for enforcement personnel. The standard also
requires that, if a process unit does not meet the prescribed performance
level, it must revert to the monthly leak detection and repair program that
is specified in §60.482-7. Compliance with this work practice standard
would be determined by inspection and/or records.
While two set skip-period programs are provided as an alternative
standard for valves in gas and light liquid service, there is no provision
for skipping from semiannual leak detection to annual leak detection. If
K-8
-------
such a change is desired, the owner or operator must submit a notification
of intent to the Administrator and demonstrate compliance with 5 consecutive
leak detection intervals (quarters) at a performance level of 2 percent
leaking or less.
Comment:
One commenter (IV-D-51) had recommended a leak detection and repair
program for pumps in light liquid service based on the good performance of
single mechanical seals in an acrylic acid unit. He further recommended a
v.
skip-period approach for such a program that would allow semiannual
monitoring after six successive periods of finding no leaks.
Response:
As discussed in Section 4.8, the standards for pumps in light liquid
service have been revised to allow implementation of work practices (leak
detection and repair programs) as well as using the equipment required in
the proposed standards. Although statistical sampling techniques such as
skip-period sampling are useful tools in alleviating some of the monitoring
burden for valves, the number of pumps per process unit prohibits the
reasonable use of these techniques because there are too few pumps to use
these techniques. And the amount of monitoring is reasonable considering
the amount of leakage from the leaking pump and the cost of monitoring.
Therefore, skip-period monitoring was not included as an alternative to the
standards for pumps given in §60.482-2.
14-9
-------
-------
15. MISCELLANEOUS
Several comments were received that were not related to the other topic
areas presented in the previous sections. These comments are discussed in
this section. They include comments on EPA's bubble policy and on an
apparent omission from Appendix A of the Background Information Document.
Comment:
Several comments were received which referred to EPA's bubble policy.
The comments reflect a good deal of confusion about the bubble policy and
its application to new source performance standards.
Several commenters (IV-D-6; IV-D-17; IV-D-26; IV-D-34; IV-D-37;
IV-D-38) said that they supported the use of the "bubble" concept for the
SOCMI standards. One of the commenters (IV-D-37) cited page 1139 of'the
SOCMI preamble which said that in most cases fugitive emissions could be
controlled from some other fugitive emission sources within an existing
facility to keep the (overall) fugitive emissions to the original level.
This same commenter said that allowing more leeway with the light liquid
requirement is one application of the bubble concept that would help
industry maintain cost effectiveness and make the proposed NSPS more
realistic.
Three commenters (IV-D-17; IV-D-26; IV-D-34) supported the application
of the "bubble" concept to modified facilities within a SOCMI plant. One of
the commenters (IV-D-26) said that the application of the bubble concept to
a process unit would avoid having to apply the fugitive VOC emission NSPS to
the entire process unit. This commenter requested that the bubble concept
also be applied to modifications. ' -
Another commenter (IV-D-17) said that he supported the proposed
application of the modification definition, which would allow combining
emissions from fugitive sources within the process unit to avoid a net
increase in emissions and thereby avoid having to apply the subject NSPS to
the entire process unit.
15-1
-------
One commenter (IV-D-38) said that the proposed requirements are incon-
sistent with the trend toward more flexible, cost effective emissions
control. He felt that extensions of the "bubble" concept to cover fugitive
emissions could provide adequate limitations for entire plants or individual
facilities. He said that this approach is consistent with Section 111 of
the Clean Air Act.
Another commenter (IV-D-6) strongly recommended setting a performance
standard as the minimum regulation for the SOCMI NSPS. He went on to say
that as a performance standard the VOC regulations would be compatible with
OSHA actions and prevent duplication of federal activities and duplicate
costs for industry. A performance standard would also be consistent with
the "bubble" concept of VOC emission controls.
Response:
The bubble concept is a fairly new emission reduction option that was
recommended by EPA at 44 FR 71780 on December 19, 1979, and is currently
implemented under Section 110 and/or Part D of State implementation plan
(SIP).
The bubble policy allows an existing plant to decrease emission
reductions from an affected facility with high control costs while
simultaneously increasing an equal amount of emission reductions from an
other affected facility in the same defined area with a relatively low
control cost. The end result is a zero net increase in total emissions
within that defined area.
EPA is considering the application of a similar bubble policy in
implementing Section 111 of the Clean Air Act. At this time, EPA has not
endorsed the application of a bubble policy to new source standards. If the
decision is made to incorporate a bubble policy in the implementation of
Section 111, the potential would exist for applying it to SOCMI facilities
covered by these standards even after the standards are promulgated.
However, EPA does not know how a bubble policy would be implemented for work
practice standards. It would be difficult, if not impossible, to implement
a bubble policy within the format of work practice standards.
15-2
-------
A number of commenters have confused the modification and reconstruc-
tion provisions for NSPS with EPA's- bubble policy. Under the SOCMI
standards an affected facility is defined as a process unit which includes
numerous fugitive emission sources that are assembled to produce one or more
of the chemicals listed in 40 CFR 60.489. In order to avoid routine changes
being considered as a modification and subjecting the facility to the - --
standards, the owner or operator may "balance out" an emission increase from
a routine change or addition to one of the fugitive emission sources within
an affected facility by simultaneously decreasing the fugitive emissions
from another source within that same affected facility. This is not an
application of the bubble policy because the balancing of emissions is
within the same affected facility, not between separate plants, although the
concept is similar.
The commenter's relating the "light liquid requirement" (assumed to
mean the vapor pressure cutoff for the definition of light liquid) to the
bubble concept does not seem logical. The vapor pressure cutoff was devised
to eliminate those sources which do not have a tendency to leak (heavy
liquid) from routine monitoring requirements. The vapor pressure definition
would remain the same for different units trying to bubble emissions.
It is also difficult to understand the commenter's point about a
performance standard's being consistent with the bubble concept. A
performance standard is not a prerequisite for applying the bubble policy.
It should be noted that a performance standard has been provided as an
alternative for valves in gas and light liquid service.
Comment:
One commenter (IV-D-17) noted that a letter from the Texas Chemical
Council to Walter Barber regarding a meeting held on July 18, 1980 with EPA
had been omitted from Appendix A of the Background Document.
Response:
The July 28, and 20, 1980 letters from the Texas Chemical Council to
Mr. Walter Barber regarding the July 18, 1980 EPA meeting arrived after the
Background Information Document (BID) was printed. That is why these
letters were not included in Appendix A of the BID. However, both of these
15-3
-------
letters as well as the minutes of the July 18, 1980 meeting have been
considered in developing the regulation and were entered into the docket.
The docket entries for the July 29 and the July 30 letters are II-D-78 and
II-D-79, respectively. The docket number for the minutes of the meeting is
II-E-20.
15-4
-------
APPENDIX A
RESPONSES TO COMMENTS ON THE ADDITIONAL INFORMATION DOCUMENT
-------
-------
APPENDIX A
RESPONSES TO COMMENTS ON THE ADDITIONAL INFORMATION DOCUMENT
Executive Summary
In a Federal Register notice on May 7, 1982 (47 FR 19724), the Environ-
mental Protection Agency (EPA) announced the availability of Fugitive
Emission Sources of Organic Compounds - Additional Information on Emissions,
Emission Reductions, and Costs (EPA-450/3-82-010) and requested comment on
the technical content of this Additional Information Document (AID). The
AID is part of the background information considered in selecting the" -
standards. The 14 comment letters received on the AID were all from ;
industry representatives of companies and associations. A list of : . . ...
commenters, their affiliations, and the EPA docket number assigned to their
correspondence is given in Table A-l.
After review of the comments on the technical content of the AID, EPA
has concluded that the procedures (presented in the AID) for estimating
emissions, emission reductions and costs are appropriate. Many of the
comments on the AID were based on intuitive evaluations of fugitive emission
sources. EPA has attempted to understand these intuitive evaluations and
has prepared more detailed analysis in order to determine the appropriate-
ness of the commenters' suggestions. In some cases, these detailed analyses
have led EPA to change the standards since proposal. A few commenters
presented detailed analyses. These analyses were evaluated by EPA and
considered in selecting the procedure for estimating emissions, emission
reductions and costs. - .
EPA considered carefully the comments on the proposed standards and. on
the technical judgments presented in the AID, particularly when determining
the best demonstrated technology (BDT) for fugitive emission sources of VOC
in SOCMI. For example, EPA has set alternative standards to reflect BDT for
process units using low leak techniques. Process units with less than
2 percent of their valves leaking may opt for an annual performance test
A-l
-------
TABLE A-l. LIST OF COMMENTERS ON THE ADDITIONAL INFORMATION
DOCUMENT FOR FUGITIVE VOC EMISSION SOURCES
Commenter and Affiliation Docket Item No.
Mr. Ronald A. Lang IV-N-1, IV-N-6
Executive Director
Synthetic Organic Chemical Manufacturers
Association
1612 K Street, N.W., Suite 308
Washington, D.C. 20006
Mr. Charles P. Blahous, Vice President IV-N-2
Environment, Health, and Safety
PPG Industries, Inc.
One Gateway Center
Pittsburgh, Pennsylvania 15222
Ms. Geraldine V. Cox, Vice President IV-N-3, IV-N-9
Chemical Manufacturers Association
2501 M Street, N.W.
Washington, D.C. 20037
Mr. C. D. Malloch .IV-N-4, IV-N-13
Regulatory Management Director
Monsanto Company .
800 N. Lindbergh Boulevard
St. Louis, Missouri 63166
Mr. Richard H. Smith IV-N-5
Environmental Specialist
El Paso Products Company
P.O. Box 3986
Odessa, Texas 79760
Ms. Judith A. Feldman IV-N-7
Government Affairs Coordinator
Chevron Chemical Company
P.O. Box 3883
San Francisco, California 94119
Mr. A. H. Nickolaus IV-N-8
Texas Chemical Council
1000 Brazos, Suite 200
Austin, Texas 78701
A.2
-------
TABLE A-l. LIST OF COMMENTERS ON THE ADDITIONAL INFORMATION
DOCUMENT FOR FUGITIVE VOC EMISSION SOURCES (CONTINUED)
Commenter and Affiliation Docket Item No.
Mr. T. A. Kittleman IV-N-10
Senior Engineer
E. I. DuPont de Nemours & Company
Wilmington, Delaware 19898
Mr. Milton J. Rhoad, Managing Director IV-N-11
International Institute of Synthetic
Rubber Producers, Inc.
2077 South Gessner Road, Suite 133
Houston, Texas 77063
Mr. C. T. Seay IV-N-12
Chairman, Air Conservation Committee
Texas Chemical Council
1000 Brazos, Suite 200
Austin, Texas 78701
Mr. William P. Gulledge IV-N-14
Manager, Environmental/Scientific Programs
Chemical Manufacturers Association
2501 M Street, N.W.
Washington, D.C. 20037
A-3
-------
rather than monthly leak detection and repair. It is important to note that
even though EPA's consideration of these comments focused on their technical
aspects, EPA attempted to understand and analyze the underlying concerns
presented in the comments.
This appendix presents summaries of the comments received and the
responses to those comments. The comments and responses are grouped
according to the topic area referred to in the AID. Section A.I addresses
comments on emission factors and emission factor development that was
discussed in Chapter 2 of the AID. Section A.2 presents comments and
responses on model units (Chapter 3 of the AID). Comments on the Leak
Detection and Repair (LDAR) model and other emission control techniques are
discussed in Section A.3, and comments on the costing methodology presented
in Chapter 5 of the AID are discussed in Section A.4. Section A.5 contains,
responses to the comments on the economics data given in Appendix A of the
AID.
A number of comments were received on material that was not addressed
in the AID. The responses to these comments appear in Section A.6.
Finally, there were many comments received that had been submitted
previously on the proposed standards and background document; These
comments are already addressed in various chapters of this Background
Information Document. Thus, the responses are not repeated in this
appendix. Section A.7, however, presents a listing of comment summary,
docket item number of comment correspondence, and a reference listing of the
portions of the BID that address the comment.
A.I EMISSION FACTORS
Comment: Several commenters asserted that SOCMI data is preferable to data
from petroleum refineries and should be used wherever possible (IV-N-8;
IV-N-4; IV-N-3; IV-N-1). One commenter referred to the Maintenance Study,
saying (1) that petroleum refinery data indicate higher mass emission rates
at a given screening value than do SOCMI data; (2) that for valves in gas
service, even the 95 percent confidence intervals do not overlap; and
ft-4
-------
(3) that the number of leaking light liquid valves and pump seals in SOCMI
are lower than those in petroleum refineries. He further argued that
emission factors for all three processes in the Maintenance Study are lower
than those from the Refinery Study. .
Response: The commenter is correct in his assertion that the data from the
Maintenance Study exhibit some differences from the data collected in the
petroleum refinery studies. As explained in the AID, EPA recognized that
fact, and accordingly, changed the estimates of emissions, emission reduc-
tions, and costs used to evaluate the standards. EPA has made the changes
implicitly requested by the commenters.
As further explained in the AID, the reasons for the differences are
not fully understood and could not be explained conclusively. Even though
the Maintenance Study data may be the most complete set of data, all the
data are not necessarily high quality, nor are the data complete. Thus,
some of the data used to evaluate the standards originated from studies of
refinery equipment. But in EPA1s judgment, the data selected in the AID can
be used to reflect the impact of the standards.
To summarize the analysis presented in the AID, EPA concluded that the
best method for arriving at a complete set of emission factors was by using
-leak frequencies determined for SOCMI units in the 24-Unit Study (IV-A^ll)
to weight the emission factors determined in the petroleum refinery study.
These studies were the two most comprehensive sets of fugitive emissions
data available at that time. The refinery study represents the best
available data on fugitive emissions rates (emission factors) from different
equipment types. The 24-Unit SOCMI Study represents the best data available
on leak frequencies for different types of equipment in SOCMI.
The emission factors resulting from combining these data sets were
compared to the factors generated from Maintenance Study data. Only the
emission factor generated for valves in gas service appeared to be abnor-
mally high, as noted by the commenter. Further analysis led to the use of
the leaking and nonleaking emission factors found for gas valves in three
SOCMI process types. This resulting emission factor for gas valves fell
A-5
-------
within the range described by the gas valve emission factors found for the
three SOCMI process types.
The emission factor data from the Maintenance Study were not chosen as
emission factors for SOCMI because the study was designed to evaluate the
effects of maintenance on emission characteristics of pumps and valves in
VOC service; establishing emission factors for these equipment types was not
the primary goal of the study. Furthermore, the average of emission factors
from three unit types were not representative of an average for SOCMI and
the emission factors generated were for only three types of equipment.
Comment: One commenter (IV-N-8) took issue with the criteria EPA used to
choose data for estimating SOCMI fugitive emissions. About the first
criterion, "relevance-to'estimating fugitive emissions from SOCMI," the .
commenter said that the SOCMI Screening and Ma-intenance studies (along with
the analysis report) are the only data of unquestioned applicability and
unquestioned validity collected specifically for this purpose. He asserted
that all other data sets require assumptions, adjustments, or transforma-
tions to be applicable. He said that the SOCMI studies provide a body of
high quality data which were taken by the same procedures and instruments
the EPA has proposed for use in the SOCMI standards.
Concerning EPA's second criterion, "validity of testing and analytical
methods used," the commenter said that the data from petroleum refineries
were taken with different instruments, different calibrants and, in some
cases, different monitoring methods. He cited an example of the Exxon
Cyclohexane study which used soap solutions for leak detection.
The third criterion, "comparability to other work," the commenter
called the antithesis of scientific investigation since it insures that
original errors and inaccuracies are never resolved. He said that the real
concern should not be comparability but applicability to SOCMI. The
commenter did not understand why SOCMI data must be comparable to refining
data nor why they must support EPA's assumption that all refinery data.are
applicable to the chemical industry. The commenter said he could understand
EPA's wanting a single approach to both petroleum refining and SOCMI, but he
A-6
-------
said there are large differences between the two and the differences mean
that portions of the refinery approach are not correct for SOCMI.
Response: The cormnenter presented a strong preference for the data in the
Maintenance Study and the 24-Unit Study. However, where the commenter
disagreed with EPA's judgment, the commenter did not present an objective
rationale for the validity of doing so. In contrast, EPA evaluated each
study and set of data and presented this analysis in the AID.
This evaluation presented in the AID makes it clear that the data
contained in the Maintenance Study and 24-Unit Study are not the highest
quality, best available data on fugitive VOC emissions, nor are they the
sole source of data on fugitive emissions. Numerous studies have been
conducted and these were reviewed in the AID, pointing out both strengths
and weaknesses associated with each study. To gain maximum utility of the
data from these studies, interpretation of the data is required, drawing
upon the strong points of a study while considering its weaknesses. This
evaluation and interpretation of the data was done in the context of the
whole base of fugitive emissions work; it was not done just for isolated
studies. Based upon this review and analysis, it was determined that the
"relevant data from widely different studies had to be merged and transformed
into a useful means of estimating emissions.
A thorough evaluation of all the available studies on fugitive
emissions provided a good basis for selecting the relevant data for estima-
tion of emission factors. An example of this evaluation procedure is the
Maintenance Study. The goal of the Maintenance Study was the determination
of the effectiveness (success rate and emission reduction) of maintenance
techniques in reducing fugitive emissions from valves and pumps in VOC
service. Its purpose was not to generate emission factors for equipment in
SOCMI. However, since the generation of emission factors was a secondary
objective of the study, these factors were used as a guide to evaluate the
emission factor estimation procedure.
A-7
-------
The data used in developing emission factors had to be not only
relevant to this fugitive VOC emissions work, but also valid with respect to
acceptable testing procedures. Thus, another part of the data evaluation
procedure involved the examination of sampling/testing procedures as a test
of their validity. EPA agrees with the commenter that it is important to
consider the differences between both testing and analytical techniques used
in compiling the reported data. For instance, in comparing the reported
emission factors, some important items to evaluate include (1) how the
samples are collected, (2) how the measurements are made (field or labora-
tory determination), and (3) what type of equipment does the reported
emission factor represent (e.g., only leaking equipment, complete distribu-
tion of equipment, etc.).
The same sort of comparisons were made for the various screening
studies. Of particular importance for these kinds of studies was the method
of determining leaks (soaping or instrument screening) and the leak defini-
tion used (bubble count or ppmv level). Soaping data were evaluated in a
qualitative manner with instrument screening data. Analysis of data
collected using two different monitoring instruments with different
calibrants (TLV-hexane and OVA-methane) indicated that the differences are
not relevant to the determination of leak frequency (see Chapter 12).
Comparability of the studies is not, as suggested by the commenter, the
antithesis of scientific investigation. The various studies considered must
be of a basis common enough to allow comparison of the results of the
studies. Otherwise, no valid comparison of the studies can be made, and
differences (including errors, etc.) cannot be determined. For example, the
SOCMI and refinery studies could not be justly evaluated against one another
if the data collected were not comparable. While comparability is an
important criterion for evaluating fugitive emissions data, the primary
criterion used is applicability, i.e., relevance, of the studies to VOC
emissions from equipment in SOCMI. This criterion was discussed earlier in
this section.
A-8
-------
Comment: One commenter (IV-N-10) said the emission factors reported in
Revision of Emission Factors for Nonmethane Hydrocarbons from Valves and
Pump Seals in SOCMI Processes were in reasonable agreement with emission
factors he had calculated independently from the same data. The same
commenter continued, saying that the best estimators for industry-wide
emissions from SOCMI are the average factors he calculated independently
from Maintenance Study data:
EMISSION FACTORS {KG/HR/SOURCE) FOR:VALVES:*
CALCULATED FROM ERA'S MAINTENANCE STUDY DATA
Gas Valves Light Liquid Valves
Plant No. Leaking Nonleaking Leaking NonleakTrfcf
3 0.0468 0.00009 0.0282 0.00009"
4 0.0391 0.00034 . 0.0253 - ' 0.00180
6 0.9561 0.00045 0.0265 . 0.00170
5 0.0698 0.00017 0.0249 0.00125
2 0.0375 0.00019 0.0321 0.00138
Average 0.0498 0.00025 0.0274 ' 0.00124'
*Emission factors were reported in Ibs/hr/source and converted.to
kg/hr/source.
Another commenter (IV-N-8) recommended the following set of emission
factors:
Leaker Rate NonLeaker Rate
(kg/hr/Source) (kg/hr/Source)
Gas Valves 0.0497 0.000247 .
Liquid Valves 0.0273 0.00124
Pumps 0.0772 0.0070
Response: EPA acknowledges the commenter's verification (IV-N-10) of the
revised emission factors reported in Docket Item:No. IV-A-29. EPA does not
agree, however, with the contention of both commenters that averaged
emission factors based on the results of the Maintenance Study are the best
estimators for industry-wide emissions from SOCMI. In the.AID, EPA
developed emission factors for different types of equipment in SOCMI. The
A-9
-------
rationale and support for the development procedure were clearly presented
in that document.
EPA analyzed the available data on fugitive VOC emissions from SOCMI
and other industries. It was determined that the SOCMI 24-Unit data were
appropriate for establishing the percent of equipment components leaking.
However, it was not the purpose of the Maintenance Study to develop emission
factors. Consequently, the emission factors which can be derived from the
Maintenance Study are not the best. EPA believes, as detailed in the AID,
that data collected for equipment in petroleum refineries are more accurate
and, in most cases, appropriate for determining the quantity of mass
emissions from equipment components that leak, except in the case of valves
in gas service. Based on this, emission factors for equipment in SOCMI were
developed and used to estimate emissions for equipment types and services.
An,additional, although less important, consideration in this discussion is
that the factors reported in Docket Item No. IV-A-29 described a range of
emissions for gas valves, light liquid valves, and light liquid pumps. An
arbitrary method of aggregating these emissions estimates would have been
needed in determining nationwide impacts. Furthermore, emission factors
determined in the SOCMI studies were limited to three types of equipment.
Emission factors would have had to be generated for those sources not
represented in IV-A-29 from the refinery data in any case.
Comment: Two commenters (IV-N-9; IV-N-10) said a source's leak rate is
constant for all leak frequencies.
Response: As stated by the commenters, the emission factors for leaking
equipment appear to remain fairly constant with varying leak frequency.
Data from fugitive VOC emission studies tend to support this observation.
This constancy of values for leaking component emission factors provides the
basis for the approach used by EPA in developing the average emission
factors. EPA's approach was presented in the AID in Chapter 2.
A-10
-------
Comment: Three commenters (IV-N-8; IV-N-1; IV-N-14) took issue with the
assumption of the distribution of units in SOCMI. Two of the commenters
(IV-N-8; IV-N-14) said the distribution is weighted too heavily with high
leak-rate ethylene plants. One commenter (IV-N-1) said the 15 process types
were not selected as a representative sample of the true distribution.
Another (IV-N-8) said 37 percent of the valves screened were from ethylene
plants.
The commenter recommended that a more reasonable way to arrive at a
distribution would be to average the 15 processes, giving equal weight to
each process. He showed a comparison of this method with EPA's method.
EPA Method Equal-Weight Method
Ethylene Plant .
Weighting in Averages Below 37%. 6.7%
Average % Leaking
Gas Valves 11.4 4.2%
- Light Liquid Valves 6.5 . 2.9%
Pump Seals 8.8 7.7%
Response: The commenters are in disagreement with EPA over how to determine
the leak frequency that is assumed to represent the behavior of equipment in
SOCMI. In the absence of more definitive information, EPA assumed that the.
total number of equipment components in the 24-Unit Study is applicable for
the estimation of emission factors for the industry. This assumption and
its known .weaknesses were noted in the AID. EPA believes these assumptions
to be reasonable. The relatively high proportion of valves in ethylene
units represented in the sample is not considered inconsistent with the
number of large, complex ethylene units that supply the majority of input
chemicals for the rest of SOCMI.
The commenters1 recommended procedure also entails the acceptance of
some assumptions. The assumption of equal weight throughout SOCMI for the
15 process types represented in the 24-Unit Study is no more realistic than
the assumptions EPA has made. The assumption of equal weighting may, i-n
fact, be -less accurate due to the prevalence of large complex ethylene units
in SOCMI.
A-ll
-------
Comment: One of the same commenters (IV-N-8) said EPA should have obtained
data to characterize the distribution of leak rates in SOCMI from contractor
studies such as the Hydroscience report. He said the Hydroscience report
shows less than 15 percent of valves in ethylene plus propylene service.
Response: The leak frequencies determined for the total equipment counts in
the 24-Unit Study represent a cross-section of SOCMI. The assumption was
made (as discussed in the previous response) that the 24-Unit Study repre-
sented the distribution of leaking and nonleaking equipment components in
SOCMI. These data do not characterize the distribution of leak rates in
SOCMI; they relate to the number of equipment components found leaking or
not leaking. Furthermore, leak rates are characterized for equipment in
specific VOC services (e.g., valves in gas service, pumps in light liquid
service, etc.), not for industries.
The version of the Hydroscience report referenced by the commenter was
a draft report; the data cited by the commenter were not incorporated in the
final report. The proportion of valves in ethylene units in the Hydro-
science surveys is a gross underestimate of valves in ethylene units. Using
the draft report, approximately 1013 valves per unit were estimated for
ethylene units from P&ID diagrams. However, based on the 24-Unit Study in
which actual physical counts were made, an average of over 4800 valves were
found in each ethylene unit.
" If the valve counts reported in the Hydroscience study were adjusted
using the average valve count found in the 24-Unit Study for the 15 process
types, valves in ethylene units would comprise about 34 percent of the total
count for the 415 units in the Hydroscience study. This is the same
percentage attributed to valves in ethylene units in the 24-Unit Study (see
Table A-2).
Comment: The validity of EPA's data base on fugitive emissions was called
questionable by one commenter (IV-N-4). The commenter said that his
concerns about the data were highlighted by Table 2-20 and by the use of
-------
TABLE A-2. COMPARISON OF EQUIPMENT COUNTS FOR 50CMI PROCESS UNITS
Valves in VOC Service8
Number Of
Plant SUes
Chemicals(s) (Hydrosdence)
Acetaldehyde
Acetic acid
Acetic anhydride
Acetone cyanohydrin
Acrolein/glycerln
Acrylic acid
Acrylic acid esters
Acrylonitrile
Adiplc add
Aniline/nitrobenzene
Benzene
Butadiene
Cap ro lac tarn
Chlorobenzenes
Chi orwne thanes
Cumene
Cyclohexane
Cyclohexanone/cyclohexanol
Dimethyl terephthalate
Ethyl acetate
Ethyl benzene/styrene
Ethylene/propylene
Ethylene dichloride
Ethylene oxide
Formaldehyde
Glycol ethers
Linear alkyl benzene
Maleic anhydride
Hethanol
Methyl methacrylate
Phenol/acetone/methyl styrene
Terephthalic acid
Urea
Vinyl acetate
Vinyl chloride
Other Units
Total ,
4
9
'6
3
4
3.
5
6
5
7
14
20
3
' 6
17
14
11
8
6
B
19
37
17
16
53
9
4
9
12
5
11
3
40
7
14
_-_,_
415
Average Number
Per Unit
(24-Unlt Study)
1,080
1,604
1,030
1 .468
855
4.842
1.212
230
1,844
3,337
Z.176
1,862
Total Number
As Org-lnally
Reported By
Hydrosctence
7,098
6,199
4,000
1,263
8,000
3,156
7 ,796
7,101
4,734
6,629
5,779'
33,941
17.713
5,657
17,000
12,765
3,245
8,626
15,428
2,140
8,087
52,690
18.778
10,635
11,953
2.574 ..
10,042
3,724
6,616
3,236
11,207
2,055
2,685
12,732
22,750
«^M^»
358,034
Total
Number With
Revisions From
24-Un1t Study
4,320
6,199
4,000
1,263
8,000
3,156.
7.796
9,624
5.153
6.629
5,779
33,941
17,713
5,657
24,965
12.397 ..
3.245
8,626
15,428
2,140
8.087
179,166
20,604
10,635
12,190
2.574
10,042
3.724
6.616
9,220
36,707
2,055
2,685
15,236 ,
26,075
j_'
528.705
Total
Number In
24-Unit Study
1,080
3,208
2,061
... . ,
2,937 '--'-'
1,771 : : :
. .
14.527 -: ; "
-- 2.424 ;- .
230
. . '- - ^..';
1,844-
3.337
4 ,353
3,725
1 .285
. 42.782
aValves Includes pressure relief devices, open-ended valves, and in-line valves.
A-13
-------
ambient temperature vapor pressure data to characterize chemicals which will
result in VOC leaks. He said that according to the AID, cumene has the same
leak frequency as ethylene, but no data are provided on processing condi-
tions involved in the processes tested. He said that the cumene processes
had to have been high temperature operations or the loss of cumene to the
workplace would not have been detected. According to the commenter, at
10,000 ppmv, the vapor pressure of cumene would have to be 7.6 mm Hg or
greater. Processing temperatures would have to be above ambient (97.3°F or
33.5°C) for this to occur, he said. He compared these temperatures to those
for ethylene at 7.6 mm Hg vapor pressure (-103.9°C) and concluded that
ethylene leaks would be detected at temperatures below ethylene's boiling
point.
Response: Studies of fugitive VOC emissions (II-A-26; IV-A-14) have shown
that the major factor affecting the percent of leaks detected (or leak
frequency) for any equipment type is the vapor pressure of the substance in
the line. This finding forms the basis for separation of different types of
equipment by service (gas, light liquid, heavy liquid). The substance in
the line does not necessarily mean the substance produced as the final
product. In the example cited by the commenter, the leaks found in the
cumene process units were not identified as cumene; there are other
substances involved in the manufacture of cumene, notably benzene. The
vapor pressure of these other compounds must be considered if they are the
substances contained in the line being tested.
The commenter also stressed the importance of temperature on leak
frequency. The effect of temperature (in-line and ambient) on leak
frequency was also examined in Docket Item No. IV-A-14. Temperature was
found to have little effect on leak frequency. Where any effects were
noted, they proved to be inconsistent.
At atmospheric pressure (760 mm Hg), a vapor pressure of 7.6 rran Hg for
any substance would result in a concentration of 10,000 ppmv. Boiling
points are defined as the temperature at which the vapor pressure of a
substance is equal to the atmospheric pressure (i.e., vapor pressure equals
A-14
-------
760 mm Hg). The temperature at which a substance has a vapor pressure of
7.6 iron Hg, therefore, will always be less than its normal boiling point
where its vapor pressure will be about 760 mm Hg.
Comment: One commenter (IV-N-4) found EPA's using emission rate data from
another industry which does not face the same control practices hard to
understand. This commenter said that the correction of the emission rate
data for differences in leak frequencies is not an acceptable alternative to
the use of SOCMI emission rate data. He continued, saying that EPA's
proposal to use leak rate data from petroleum refinery studies instead of
actual SOCMI data penalizes facilities which have achieved low emission
rates.
Response: Techniques for controlling fugitive emissions were designed for
control of emissions from equipment in VOC service. They are not specific.
to a given industry such as SOCMI or petroleum refining. In any case,
refineries are subject to as strict control practices as chemical plants.
As stated above and in the AID, mass emissions data for equipment in VOC
service.in petroleum refineries represent the best available data for
establishing mass emissions from VOC-emitting equipment. When combined with
leak frequency data from the 24-Unit Study,. these data, in. general, provide
estimates of emission factors that are bounded by the range of values found
in SOCMI studies (IV-A-29). The notable exception was valves in gas
service; data from SOCMI studies led to the development of an emission
factor for this equipment type and service that is also bounded by the
emission factors generated for three SOCMI .process types.
EPA's use of emission rate data from petroleum refining studies has
resulted in emission factors that are similar to the values found for three
types of equipment in SOCMI units. . Units that achieve low emission rates
are not penalized as suggested by the commenter. Process units that have
low leak frequencies (indicating low leak rates) have been provided for in
the standards with alternative standards for valves (see Chapter 14). These
alternative standards allow reduced monitoring efforts for units that meet
certain low leak frequency criteria.
A-15
-------
Comment: Another commenter (IV-N-1) said that the original proposal was
premised on the assumption that emissions from petroleum refineries are
similar to emissions from SOCMI units and that data showing the need for
control in petroleum refineries would support the need for controls in
chemical plants. This commenter contended that the data presented in the
AID refuted this assumption. He said the effort to extrapolate from
refinery data to SOCMI fails to produce realistic emission factors for
SOCMI. Another commenter (IV-N-3) said data referenced in Section 2 further
substantiated his position that SOCMI fugitive emissions are significantly
less than those encountered in the petroleum refining industry.
Response: It is not assumed that the emissions from the two industries are
similar. Rather, the standards are premised on an analysis of data related
to the substances processed and the equipment used, regardless of the
industry in which the equipment is being used. Similarly, the need for the
standards, as discussed in Chapter 2, is not based on an assumption that
SOCMI and refinery emissions are similar. The need for fugitive VOC
emission standards for equipment in SOCMI is determined independently from
the need for standards in the refining industry as discussed in Chapter 2.
Control technologies for equipment leaks in the petroleum refining and SOCMI
industries are similar because equipment used and substances processed are
similar. EPA believes the data in the AID support rather than refute the
basis of the standards.
As detailed in the AID and discussed above, the emission factors used
to estimate emissions from equipment in SOCMI are considered by EPA to be
the most realistic factors available to date. They are the result of a
reasoned examination and analysis of available fugitive emissions studies.
Comment: One commenter (IV-N-1) contended that emission factors from batch
processes used by small producers are different from emission factors for
continuous processes. He said that a single set of emission factors could
not describe differences in production methods.
A-16
-------
Response: This same commenter made this statement in an earlier set of
comments (IV-D-38) submitted on the proposed standards. As is noted in
Section 5.5 on "Small Manufacturers," the data base includes information on
small manufacturers. EPA has used these data in developing the model units
emission characteristics, and has found no reason to believe that emissions
from leaking equipment in small batch units would be any different from the
same emissions in other units. The equipment processing VOC is performing
the same function in both types of processes. . .
EPA has provided a low production volume (1,000 Mg/yr) exemption for
small facilities. If a unit produces more than this quantity of .chemicals,'
on either a continuous or batch basis, it will have fugitive emissions that
can be controlled at a reasonable cost. Furthermore, since batch operated.
units are, by design, shutdown on a more frequent basis than continuous.
units, effective repairs can be made more quickly. This would tend to
increase the emissions reduction achievable since_unrepairable equipment
would not accumulate for longer periods as in a .continuous unit.
Comment: Two commenters (IV-N-8; IV-N-1) objected to SOCMI's characterize- .
tion by a single average unit. The commenters said a single unit failed to-
capture the variability in the industry. One of the commenters (IVrN-l): .
said the use of a single average unit leads to serious distortions.of
emissions from individual processes. . ....-,.
As support for this perceived distortion he referenced a chart
distributed at the APCA meeting in.June 1982. The commenter said the
average unit is not representative of the processes that make up the
industry because leak rates for some processes are overstated, while leak ...
rates for other processes are understated. He further stated that data in ,
the AID show that "average units do not characterize the industry because of
differences .in the characteristics of the substances produced." He cited
variations in volatility, safety considerations, odor considerations, and
OSHA regulations.
The commenter cited two passages in the AID as support for his conten-
tion that an average unit cannot represent SOCIMI:
A-17
-------
There are no mass emissions data which can be considered
representative of emission factors for SOCMI as an industry.
(AID at 2-51)
Reviewing the available studies of fugitive emissions from SOCMI
units, no studies were found that resulted in a full set of
emission factors applicable to SOCMI in general. Furthermore, no
study had been designed to establish a single set of emission
factors for SOCMI fugitive emission sources. (AID at 2-58)
He objected to the fact that despite these acknowledged data limitations,
the AID attempts to extrapolate from the petroleum refinery data to estimate
emission factors for SOCMI.
The other conunenter (IV-N-8) said SOCMI cannot be characterized by a
single average leak frequency for the purpose of determining monitoring
frequency. He said that use of a single average leak frequency leads to
absurd results for low-leak plants. He gave as an example vinyl acetate
plants which would have to monitor every two weeks just to maintain their
initial or uncontrolled emission levels.
Response: In developing standards and assessing impacts, several types of
emission factors are required. One average emission factor is needed to
determine the industry-wide effect of the standards, and a second more
detailed set of emission factors is required to evaluate the effect of
controls on categories of facilities affected by the standards.
EPA has acknowledged that there is no average SOCMI unit which can
serve as a predictor of the performance of individual process units.
However, the concept of industry averages is appropriate and has been used
historically by EPA to estimate the impacts of control alternatives on the
industry as a whole. Furthermore, the task of compiling estimates, which
would be a formidable and impractical undertaking, is unnecessary.
In preparing emission estimates for the industry, EPA is not predicting
emissions for any given process unit. Rather, the average values are used
with the knowledge that there are units with emissions lower than the
average, and units with emissions greater than the average. EPA's average
emission factors fall within the range determined for some equipment types
A-18
-------
in three SOCMI process types (IV-A-29). Averaging techniques are applied as
a straightforward means of arriving at industry-wide emissions
estimates. The results of these estimating procedures give a reasonable
approximation of fugitive emissions from SOCMI.
In evaluating the effects of.alternative control approaches on .
individual categories of facilities .affected by the standards, the limita-
tions of the available data on fugitive emissions must be carefully
assessed. This need was clearly acknowledged in the AID. Categorization of
factors is unnecessary. The commenter has quoted two passages from the AID
without giving the same qualifications presented in that document. While it
is true that there are limitations on SOCMI fugitive emission data, and on
fugitive emission data i.n general, these limitations-were cons-idered in
determining which data were applicable to establishing a set of average
emission factors for SOCMI. It was due to these limitations that judgments .-
concerning the data had to be made. The limitations of fugitive emissions
data were restricted to estimating emissions; they did not relate to the
achievability of the fugitive emission control techniques being applied.
Based on the studies of fugitive VOC emissions, it is obvious that fugitive
emissions do occur and that control techniques are avail-able that
effectively reduce them.
As discussed in the AID, the study of fugitive emissions from petroleum
refining was specifically designed to generate emission factors for
different types of equipment. The Maintenance Study on the other hand
focused-on the effects of maintenance in reducing emissions; establishing
emission factors was not a primary goal. When differences in the two sets
of emission factors were noted, and after noticing that leaking emission
factors remain remarkably constant among units, EPA used the leak frequen- ...
cies found in the 24-Unit Study to adjust the emission factors from the
refining"work for the differences and another comparison was made. The only
average emission factor falling outs.ide of the range described for selected
SOCMI equipment types (IV-A-29) was for gas valves. Upon closer examination
of the data, differences in the leaking and nonleaking emission factors from
gas valves in the SOCMI units were found to be markedly different and to
A-19
-------
have tighter confidence bands. Therefore, they were used as the basis for
determining the average emission factor for gas valves in SOCMI. The
resulting factor was verified since it is within the range described in
Docket Item No. IV-A-29.
The use of average emission factors is typically used in estimating
impacts of standards. The technique has been found to yield reasonable
approximations. But in establishing standards, variability of data is also
considered. Variability of leak frequency, emission factor, and leak
occurrence rate", among other things, has been considered in setting
standards for SOCHI fugitive emitting equipment. For example, the standards
provide for less frequent monitoring (monthly/quarterly leak detection and
repair, and other alternative standards) should low leak status be
maintained.
Comment: One conmenter (IV-N-8) thought EPA's rationale for using a single
average unit was not convincing. The commenter said that estimation of
national impacts does not have to be done by extrapolating emissions from a
single typical unit to the entire U.S. population of units in SOCMI. He
said national impacts could be determined from the sum of any number of
subsets instead. Referring to EPA's rationale in the AID, the commenter
said the situation in SOCMI is not similar to the refining industry because
of the wide variation in SOCMI products. He said the fact that EPA used a
single set of emission factors for estimating impacts in the petroleum
refining industry was not a reason; it was simply what EPA did. He said
data in Table 2-2 and the 24-Unit Study showed much more variability in
SOCMI than in petroleum refineries. As an example the commenter said the
leak frequencies for gas valves were 5.6 to 27.3 percent in refineries, a
factor of 4.9; while in SOCMI, leak frequencies were 1.0 to 14.8 percent, a
factor of 14.8. Another commenter (IV-N-1) said the effort to extrapolate
from a fictional average refinery unit to a fictional average SOCMI unit
fails to produce realistic emission factors for SOCMI.
A-20
-------
Response: The estimation of national impacts does not have to rely on the
extrapolation of a single typical unit to the expected population. It is
true that estimates can be based on an aggregation of subsets different from
the current model unit concept. (The method used currently, however, is
appropriate and reasonable.)
EPA has defined its subsets as three model units, which describe three
degrees of process unit complexity. Each model unit is composed of various
numbers of equipment components (valves, pumps, compressors, etc.) in
different services (gas, light liquid, heavy liquid) based on the vapor
pressure of the substances being handled. In previous studies, vapor
pressure of the substance in the line was found to be the major factor
affecting leak characteristics (i.e., leak frequency). Therefore, equipment
types were considered by service, and emission factors were developed
accordingly. Emissions estimates were then based on aggregated emissions
for three model units and extrapolated to the projected number of model
units.. While the commenters suggest other methods for defining subsets-for .
determining emissions impacts, they do not present their methodology nor the
results of their recommended procedure. Additional discussion of emissions
aggregation methods is presented in the comments and responses to Chapter 3
(Model Units) of the AID.
The comparison of ranges of leak frequencies for petroleum refinery
data and SOCMI data made by the commenter is inappropriate. The factors . . .
presented.by the commenter are affected by the absolute.value .of the lower
bound of leak frequency. Since the lower.bound for SOCMI data was less than
the lower bound for petroleum refining data, the range factor for SOCMI
computed by the commenter appears to be much larger. The appropriate
comparative ranges of leak frequencies would have been 21.7 percent and .;:.
13.8 percent for refineries and SOCMI, respectively. This comparison does
not necessarily indicate a higher degree of variability associated with
emissions from SOCMI equipment. .
As discussed previously in this section, the emission factors developed
for estimating emissions from SOCMI were based upon a reasoned examination
of the available studies of fugitive emissions. They are not a mere
A-21
-------
extrapolation from an average refinery unit to an average SOCMI unit. In
fact, the emission factors developed for SOCMI equipment do provide
appropriate estimates for nationwide emissions.
A.2 MODEL UNITS
Comment: One commenter (IV-N-8) maintained that model units should be based
on low, medium, and ethylene leak rates instead of on process complexity.
He said that basing model units on process complexity does not lead to
meaningful consequences because all three model units have the same cost per
component monitored, the same emission reduction per component monitored,
and the same cost per Mg VOC emission reduction. He said the thing that
does make a difference in costs per component and cost effectiveness is leak
rate. The commenter included some calculations for Model B units under
quarterly monitoring of valves with low, medium, and ethylene leak
frequencies. He concluded that leak frequency is the thing that makes a
difference in costs and cost effectiveness per component, and, therefore,
the model units should be based on varying leak frequencies.
Response: By following the commenter's recommendation, it would be possible
to define a large and complex set of'model units. Although the commenter's
analysis was confined to varying leak frequencies for valves, several leak
frequencies for each type of equipment could be modeled for several units of
varying complexity. The model units could also include different trends in
leak frequency for different types of equipment. It is easy to see that
this type of analysis would very quickly result in a large matrix of model
units. However, adding all of this complexity would result in little
benefit in terms of estimating regulatory impacts. Obtaining estimates of
impacts would still require averaging. The same can be said for adding the
complexity required to model three different leak frequencies for valves as
suggested by the commenter. The result would be an expansion of the number
of model units from three to nine, and estimating impacts would still
require averaging. The added complexity would not improve the analysis used
A-22
-------
in establishing the standards; therefore, for purposes of estimating impacts
of the standards, EPA has assumed average leak frequencies for each of the
different types of equipment and three levels of complexity. .
Variations in leak frequency are important and have been considered in
the formulation of the standards, however. Notable examples of this type of
consideration are the alternative standards for valves. These alternative
standards address the leak frequency variation of valves pointed out by the
commenter, providing less costly options for owners or operators with low
leak frequencies. (See comments in Miscellaneous.for other examples of
options for compliance.)
. Finally, the commenter is correct in his assertion that leak frequency
affects costs on a per component basis. However, costs and emissions for
process units depend on the number and kinds of fugitive emission sources
present as well as on leak frequencies.
Comment: The same commenter (IV-N-8) said EPA's reasons for not using .
different leak rate models were unconvincing. The commenter responded to
EPA's assertion that it was impractical because there was no distribution of
the number of units in each leak category. He said that EPA had assumed a
distribution based on the 24-Unit Study to get the proposed average industry
emission rates, a distribution he thought poor (see Section 2); The
commenter said that EPA had thought it impracticable to do a cost analysis
for model plants based on leak frequencies, but that he had been able to
perform a cost analysis on a hand calculator. To the objection EPA raised:
concerning the impracticality of categorizing all the process units, the
commenter said he was not asking EPA to categorize all the processes. He
was simply stating that SOCMI processes have a wide range of leak frequen-
cies and was requesting that EPA's regulations be reasonably appropriate for
all units in SOCMI. He finally responded to EPA's reasoning that the
complexity introduced was unwarranted because of leak rates varying among
different components and,with time. He said that these complexities were
associated with the real world and that EPA's respons.ibil ity was to develop
regulations applicable to the real world, not some hypothetical case.
A-23
-------
Response: The first issue raised by the commenter concerning the distribu-
tion assumed for SOCMI is addressed in Section 2 as noted in the comment.
Turning to the issue of practicability of construction of varying leak
frequency model units, the commenter claims to have performed the analysis
on a hand calculator. However, it should be noted .that the.commenter's
calculations included varying leak frequencies for valves only. His
analysis does not indicate how the varying leak frequencies would be
aggregated. EPA has performed an analysis similar to the commenter's but
not within the context of model units. As EPA has noted, modeling varying
leak rates for different types of equipment would constitute a much more
complex problem, complexity which could not be justified with respect to the
results achieved.
The problem of categorizing SOCMI units was dismissed by the commenter
as unnecessary for such an analysis. However, he did not offer any solution
to the problem of assigning units to categories. It is difficult to see how
an analysis such as the one requested by the commenter could be performed -
otherwise. The approach used by EPA avoids the problem and achieves the.
result of defining appropriate standards which account for ranges of leak
frequencies, as the commenter requests. Variability in the industry has
been taken into account in arriving at reasonable standards, applicable to
all of SOCMI. Evidence of these considerations may be. found in Chapter 14
(Alternative Standards) and in the Miscellaneous Section of this Appendix.
EPA realizes that there is an obligation to develop real world
standards. EPA has accepted that responsibility and believes this respon-
sibility has been met. Complexities associated with standards for equipment
in SOCMI that emit fugitive VOC have been analyzed thoroughly and standards
have been developed to provide for those complexities.
Comment: The same commenter (IV-N-8) recommended that regulatory scenarios
be analyzed for occurrence rates of 0.2, 1, and 2 percent per month.
Response: As in similar comments submitted by this same commenter, his
recommendation concerns valves only. The commenter has apparently over-
A-24
-------
looked the fact that varying occurrence rates for different types of equip-
ment would result in a very complex analysis which is not justified by the
end result. .Therefore, as explained in the preceding comments and
responses, EPA has not incorporated varying leak frequencies and occurrence.
rates in the model units. , .
However, an analysis of varying leak frequency and occurrence rate for
valves has been performed. It is included in Chapter 14. This analysis was
used in developing standards for valves which are reasonable and appropriate
for low leak frequency units as well as.high leak frequency units. ;
Comment: Another commenter (IV-N-1) said that model units should be based
on such items as unit size, system pressure, system temperatures,
volatility, odor threshold, existing control measures, and size of producer.
Response: Construction of model units for this long list of parameters
would result in an extremely complex set of model units. As stated \,
previously, this type of complexity is unnecessary. The variations, if
added to the analysis, would add an undue amount of complexity which would
then be averaged out in aggregating the impact estimates. Furthermore, of
the parameters listed by the commenter, only volatility has been shown to
have a significant effect on. leak frequency. It would be inappropriate;tq
define model units by varying parameters which have not been shown to have a
large effect on fugitive emissions. To the extent the commenter was .
suggesting that EPA tailor the standards to model units with various.items
(as mentioned above), EPA has done so where data indicate a need to do so.
Comment: A commenter (IV-N-1) said that EPA's models fail to present an
accurate portrayal of the industry.
Response: Analysis of all of the information available to EPA indicates -
that the model units present an accurate picture in aggregate. (See AID for
review of available data.) That is, the national averages for emissions and
costs are. reasonably accurate insofar as accuracy can be determined. The
A-25
-------
only means to a more accurate estimate would be by sampling every unit in
SOCMI, an unreasonably expensive undertaking. EPA's information covers
batch, continuous, simple, complex, hazardous, benign, and malodorous
chemicals.
EPA realizes that, just as in the case with averages of any kind, some
units will have higher leak frequencies and some will have lower leak
frequencies; some will have more equipment, and some will have less equip-
ment. However, averages are useful tools, and they are necessary in
aggregating total impacts of the standards. With any average it is
important to keep the range represented by the average in mind. EPA
recognizes this fact, and the standards have been written to allow for
differences in units represented by the model units.
In summary, the model units appropriately represent the range of units
in SOCMI. Furthermore, they reflect variations in the main factor which
influences costs, emissions, and emission reductions, i.e., number of
equipment components.
A.3 EMISSION REDUCTIONS
A.3.1 The LDAR Model
Comment: Several commenters (IV-N-10; IV-N-9; IV-N-11; IV-N-8) favored
using the LDAR model, although they said it needed further work. Commenters
felt that the LDAR represented an improvement over the ABCD model which they
recommended be dropped from consideration.
Response: The selection of the LDAR model in the AID as an evaluation tool
indicates EPA's agreement with the commenters that the LDAR model represents
an improvement over the ABCD model. Comment on the LDAR model was requested
with release of the AID for public review. Suggestions for improvement have
been considered. Although the LDAR model does represent an improvement over
the ABCD model, its proper implementation requires more data than the ABCD
model. As a result, the ABCD model may still be the only mode of evaluating
leak detection and repair programs for equipment for which only limited data
(of certain kinds) are available.
A-26
-------
For example, sufficient data were not available to use the LDAR model
to evaluate leak detection and repair programs for pressure relief devices.
Therefore, the ABCD model was used to make this estimate. The ABCD results
were adjusted, however, based on LDAR model results for other equipment (see
Section 4.3 of the AID). Since the results of the LDAR model for gas valves
indicated a lower effectiveness than estimated by the ABCD model for gas
valves, a ratio of these two results was used to adjust the ABCD model
results for pressure relief devices in gas service (IV-B-19). Thus, where
the LDAR model is not readily applicable, the ABCD model still provides a
means of evaluating leak detection and repair programs. In some cases, the
ABCD approach can be supplemented with information generated by the LDAR
model for other equipment types.
Comment: Two commenters (IV-N-10; IV-N-8) said that the model results
should be reported and evaluated in terms of mass emissions rather than
fractional reduction. Both commenters made the point that a small
fractional reduction may give a large emission reduction and vice versa.
One (IV-N-10) added further arguments, saying that since emission rate is
the value of interest, it would be more appropriately reported, and that
emission rate figures are not likely to cover up apparent anomalies in the
model (as discussed in the previous comment).
Response: The LDAR model, as presented in Docket Item No. IV-A-22, provides
sufficient information for the computation of mass emissions and mass
emissions reduction. In the summary tables in the computer output, both the
mean emission factor and fractional emissions reduction associated with the
LDAR program are presented. Either value can be used to compute the
controlled emissions rate and the emissions reduction when coupled with the
number of equipment components considered and the average input emission
factor. Furthermore, the computer programming can be modified by the user
to print out these results in any format that may be desired.
The commenters are correct in pointing out that fractions and
percentages should be used with caution. It is good practice to present
A-27
-------
both the fractional values and the actual computed emissions levels since
both of these numbers are important. Both values are presented in the AID
in Table 4-15 for valves and in Table 4-20 for pumps in light liquid
service.
Comment:: One commenter (IV-N-10) agreed with the gist of EPA's discussion
of "assumed baseline levels" and "uncontrolled levels" of emissions. He
said he recognized the difficulties inherent in developing a good basis for
comparison of emissions results for the NSPS. The commenter saw that both
contained major uncertainties for typifying the industry. However, the
commenter went on to say that he saw a definite advantage to the "uncon-
trolled" level basis for comparing costs and effectiveness of various
regulatory strategies. The advantage the commenter saw was that the
occurrence rate appears directly in the comparison basis. The commenter
said this was not the case with the assumed baseline level basis. He said
using an assumed baseline level basis apparently avoids assumptions about
typical SOCMI occurrence rates. However, since an occurrence rate is
assumed in the LDAR model for regulated plant emission rate calculations, it
is likely that an occurrence rate is implicitly involved in the baseline
level.
Response: There are problems associated with comparison of one level of
control (e.g., monthly leak detection and repair) to some other level,
whether that level be an "assumed baseline level" or an "uncontrolled
emissions level." Any comparisons among levels of control will be based on
some degree of uncertainty because data collected to compare levels of
control with the LDAR model are collected either in different plants or at
different times.* However, when comparisons of different levels of control
are based on appropriate data, relevant comparisons can be made.
*The feasibility of obtaining truly uncontrolled data within an operating
unit is questionable. An operating unit's primary purpose is to produce
chemicals. Interference or interruption of normal operating and
maintenance practices to collect truly uncontrolled occurrence rate data
may inhibit production and, thus, would not be reasonable.
A-28
-------
When using the LDAR model, it is appropriate to use information and
data collected during the Maintenance Study unless better information and
data are available. The occurrence and recurrence rate data are the most
relevant and useful data from the Maintenance Study; they have been used in
executing the LDAR model. Furthermore, analysis of the impacts of various
levels of control using these data (and data outlined in the AID) is an
appropriate means of selecting the level of control required by these
standards. The comparisons used in selection of the standards included a
comparison to the level of control that would be evident in the absence of
the standards and comparisons to incrementally more and less restrictive
levels of control. The comparisons used in selecting the standards, are on a
common basis.
Comment: Three commenters (IV-N-9; IV-N-8; IV-N-10) said that the equation
used in the LDAR to correlate mass emission rates and leak frequencies is
inappropriate. One commenter (IV-N-8) said he understood that the model "
splits a fixed emission factor with 98 percent to leakers and 2 percent to
nonleakers, regardless of leak frequency.. The commenter recommended that
the model be corrected to reflect the concept presented in Section. 2 of the
AID; that is, that leakers have a certain.emission rate per source and that
nonleakers also have a certain emission rate per source.
Another of the three commenters (IV-N-10) described the model's
workings, saying that instead of using a constant emission factor for each
valve category, the LDAR calculates different emission factors, as an
intermediate step, for each category and application. He said the model
assumes average emissions for each valve category based on the number of
valves initially in the category. The commenter described four valve
categories defined by the model as leaking gas valves, leaking light liquid.
valves, nonleaking gas valves, and nonleaking light liquid valves. He
provided emission factors he calculated from the Maintenance Study data as
shown in Table A-3.
A-29
-------
TABLE A-3. EMISSION FACTORS (KG/HR/SOURCE) FOR VALVES:
CALCULATED FROM EPA'S MAINTENANCE STUDY DATA*
Gas Valves Light Liquid Valves
Plant No.
3
4
6
5
2
Average
Leaking
0.0468
0.0390
0.0561
0.0698
0.0375
0.0498
Nonleaking
0.00009
0.00034
0.00045
0.00017
0.00019
0.00025
Leaking
0.0282
0.0253
0.0265
0.0249
0.0321
0.0274
Nonleaking
0.00009
0.00180
0.00170
0.00125
0.00138
0.00125
*Values were reported by commenter in Ibs/hr/source and converted to
kg/hr/source.
He said that, overall, these data show little difference among units in
spite of the fact that the leak frequencies ranged from 1 percent to more
than 25 percent.
The commenter contrasted these emission factor calculations with the
procedure used in the LDAR model. He said that the model calculates
emission factors for each of the four valve categories described in
Table A-3 based on the initial leak frequency of the valves. He then
presented a graphical analysis, Figures A-l and A-2, showing his under-
standing of the LDAR emission factor calculation method superimposed on the
emission factors presented in Table A-3. He concluded that the relationship
used in the LDAR model is not supported by the Maintenance Study data.
Response: The commenters have misunderstood several aspects of the LDAR
model mechanics and the way EPA used the model -to arrive at estimates of
impacts. First, the categories defined by the LDAR model are not those
given by one of the commenters (IV-N-10). They are, rather
(1) Nonleaking sources (sources screening < action level),
(2) Leaking sources (sources screening >_ action level),
(3) Leaking sources which cannot be repaired on-line and are
awaiting a shutdown for repair, and
(4) Repaired sources with early leak recurrence,
A-30
-------
VAPOR VALUES
25
20
> 15
o
z
LLJ
O
UJ
cc
u.
10
UJ
PLANT NO. 4
PLANT NO. 6
PLANT NO. 2
LDAR model
calculated values
PLANT NO. 3
| PLANT NO. 5
1
1
0.11 0.2 0.3
AVERAGE LEAKER LEAK RATE (*/hr)
Figure A-l. Emission factor for leaking valves "in gas service as
a function of leak frequency (-'IV-N-10).
A-31
-------
LIGHT LIQUID VALUES
25
O
z
LLI
D
O
HI
cc
b.
UJ
-J
20
15
10
PLANT NO. 2
PLANT
NO. 4
PLANT
NO. 5
PLANT NO. 6
LOAR model
calculated values
/
NO. 3
j_
I
I
0.1 0.2 0.3
AVERAGE LEAKER LEAK RATE (*/hr)
Figure A-2. Emission factor for leaking valves in light liquid service
as a function of leak frequency UV-N-10),
A-32
-------
as given on page 3 of Model for Evaluating Effects of Leak Detection and
Repair Programs on Fugitive EmissiOJTJ (IV-A-22).
Furthermore, the commenters have misunderstood the mechanics of the
LDAR model used in partitioning emissions among the four categories. The
model does not arbitrarily split an emission factor with 98 percent to
components that leak and 2 percent to components that do not leak regardless
of the leak frequency, as one commenter (IV-N-8) suggested. In the LDAR
model, emissions per component for leaking and nonleaking components are
calculated, based on the average input emission factor for all components
and the initial leak frequency of the components:
(IFL)(EL) + (1 -IFLK-Ep) = Ej,
where: TFL is the initial leak frequency;
E. is the emission factor leaking components;
Ep is the emission factor for nonleaking components; and
E, is the average input emission factor for all components.
The emission factors are all on a per component basis.
The derivation of EL and Ep is detailed in Docket Item No. IV-A-22
on page 9. Another emission factor, E^, is also defined on that page.
This emission factor is for components exhibiting early leak recurrence and
for components which cannot be repaired. In this way, the LDAR model
assigns emission factors to the four defined categories (two categories
having the same emission factor) prior to initiating the model simulation.
These three emission factors, E, , Ep, and Er, are applied to the
number of components
assigned to each category as a result of initial leak frequency, occurrence
rate, recurrence rate, and maintenance effectiveness, both for normal and
turnaround maintenance efforts.
One commenter (IV-A-10) has misinterpreted the results of the LDAR
model through the use of improper (inconsistent) inputs to the model. He
used a constant average input emission factor, Ej, with varying leak
frequencies. By holding the average input emission factor constant,
regardless of the leak frequency, he produced the curves in Figures A-l and
A-33
-------
A-2 showing the emission factor for leaking components as a function of leak
frequency. The commenter called the illustrated relationships.unreasonable,
especially in view of the constant emission factors he had determined for
leaking components.
EPA agrees that the emission factor for leaking components remains
relatively constant with changing leak frequency as the commenter shows. In
fact, this is the premise for generating average input emission factors for
all components that is presented in Chapter 2 of the AID. Therefore, in the
analysis presented by the commenter, the average input emission factor,
Ej, should have depended upon the leak frequency as given in that chapter.
The LDAR model's approximation of a constant emission factor for
leaking components is shown in Figures A-3 and A-4 for valves in gas and
light liquid service. Each figure presents three curves for the emission
factor for leaking components determined by different methods. Curve "A" is
the constant emission factor used in the development of average input
emission factors. This is the value presented in Chapter 2 of the AID.
Curve "B" represents the LDAR model approximation of the emission factor for
leaking components assuming a constant average input emission factor. This
curve is similar to the one presented by one of the commenters (IV-N-10) in
Figures A-l and A-2. Finally, curve "C" is the LDAR model approximation of
the constant emission factor for leaking components. The average input
emission factors used in generating this curve were determined by the method
presented in Chapter 2 of the AID, based on a constant emission factor for
leaking components (Curve "A") and a leak frequency.
In comparing curves "A" and "C" there is a small deviation from
linearity noted for curve "C." The deviation is the result of the relation-
ship between the emission factor for leaking components and the emission
factor for nonleaking components assumed in the LDAR model. As seen in the
figures, however, this simplifying assumption is justified since the
resulting emission factors for leaking components are a close approximation
of those used to generate the average input emission factors.
A-34
-------
OJ
en
25
20
c
OJ
o
I
u
c.
OJ
13
CT
£ 10
Legend
A Constant emission factor for leaking component
B Emission factor for leaking equipment computed
by LDAR model using a constant average input
emission factor
C Emission factor for leaking component computed
by LDAR model using average Input emission factor
developed by method In Chapter 2 of the AID
0
0.05
Figure A-3.
0.10 0.15 0.20
Emission factor for leaking component, kg/hr/source
Comparison of emission factors computed for leaking valves
in gas service.
-------
i
CO
01
25
Legend
A Constant emission factor for leaking component
B Emission factor for leaking equipment computed
by LDAR model using a constant average input
emission factor
C Emission factor for leaking component computed
by LDAR model using average Input emission factor
developed by method 1n Chapter 2 of the AID
0.05 ' 0.10 0.15
Emission factor for leaking component, kg/hr/source
Figure A-4. Comparison of emission factors computed for leaking valves
In light liquid service.
0.20
-------
Comment: Two commenters (IV-N-9; IV-N-10) objected to the comparison of the
LDAR model results to the ABCD and modified-ABCD model results. One
(IV-N-9) called the comparison an apples and oranges comparison. The other
(IV-N-10), said that, instead,, the validity of the LDAR assumptions, inputs,
and results should be investigated separately.
Response: EPA agrees that there are differences between the ABCD, modified-
ABCD, and LDAR models. These differences stem primarily from the assump-
tions made in their development and the data required to implement them.
The comparisons presented in the AID were not intended as a validation or
discrediting of one method or another. The comparisons were presented as an
aid in understanding how each model was applied. EPA agrees^that each
method should be validated on its own merits. In Chapter 4 of the AID, EPA
has examined the assumptions, inputs, and results of the LDAR model
separately from the other models for evaluating leak detection and repair
programs. . .
Comment: One commenter (IV-N-10) felt it essential that the model be able
to evaluate regulatory impacts on industry segments. The commenter said ,
that this capability is necessary for development of a standard that avoids
gross inequities and insures real emission reductions.
Response: The LDAR model is not constrained by industry segmenting. With
it, any particular case can be examined, given the appropriate input values..
If analysis of any aggregate is desired, the LDAR model is capable of
evaluating the effects of leak detection and repair programs based on the
input values.
As discussed in Chapter 2 of the AID and in the responses to comments
on that chapter, segmenting of an industry as diverse as SOCMI is not
practical. Instead of segmenting the industry, input values were varied to
analyze various situations in examining regulatory alternatives for the
standards. This analysis of the LDAR model input parameters provided the
support for standards with reasonable costs per unit emission reduction. An
A-37
-------
example of this approach is given in Chapter 14 which discusses alternative
standards for valves.
Comment: One commenter's (IV-N-10) analysis showed occurrence rate to be
one of the few parameters to which the LDAR model is sensitive. The
commenter asserted that most emissions come from new leaks occurring between
inspections.
Response: Many of the input parameters to the LDAR model have an impact on
the results of the model. Since input parameters can have different degrees
of effect on the model results, sensitivity analyses were performed by
varying many of the input parameters of the LDAR model, including:
monitoring interval
unsuccessful repair rate for normal maintenance
unsuccessful repair rate for turnaround maintenance
percent emissions reduction associated with unsuccessful repair
turnaround frequency
early recurrence rate
Most of the listed parameters (in fact, all but monitoring interval) show
little effect on the model results. All of these parameters were examined
in the classical sense for sensitivity (i.e., all other parameters were held
constant while only the parameter of interest was varied). The results of
these tests are presented in Docket Item No. IV-B-22. The results of a
sensitivity analysis on the effect of monitoring interval indicated a larger
effect on model results than the other parameters listed above; the results
are detailed in Section 4.2 on "Leak Detection and Repair Programs."
The effects of occurrence rate changes on LDAR model results were also
examined in a sensitivity analysis. As the commenter noted, the occurrence
rate was found to be one of the few parameters that appear to have a
noticeable impact on the results of the LDAR model. The analysis of the
model's sensitivity to occurrence rate, however, is complicated by the
relationship of occurrence rate to two other model input parameters, leak
frequency and emission factor. As a result, the classical approach to a
sensitivity analysis cannot be used for occurrence rate.
A-38
-------
In a physical, real world sense, occurrence rate and leak frequency are
related, i.e., one parameter affects the other. Most likely, the occurrence
rate significantly influences the relative level of leak frequency measured
in the field. For instance, in the absence of any maintenance efforts, a
high occurrence rate will result in a high leak frequency measurement.
Indeed, this trend appears to be substantiated by the data presented in
Figure A-5. These data are from the Maintenance Study (IV-A-10) and the
24-Unit Study (IV-A-11). They represent occurrence rate/leak frequency data
for all valves in three process types and in the composite data set. The
line shown in the figure is the result of a linear regression of the four
data points. Although this regression is not a rigorously developed
relationship for all SOCMI units, it was used to examine the relationship
between model results and occurrence rate.
In the LDAR model, there is a derived relationship between the emission
factors for various categories (defined within the model), the average input:
emission factor, and the leak frequency. This relationship was discussed in
detail in response to the first comment on the LDAR model. Furthermore, as
discussed in Chapter 2 of the AID, the average input emission factor used as
an input to the LDAR model is dependent upon the leak frequency that is
used, since an average emissions rate can be viewed as being composed of
emissions from equipment that is leaking and of emissions from equipment
that is not leaking. As discussed previously in the comment and response on
the LDAR model, different average input emission factors (generated using
leak frequency) must be used as inputs when the LDAR model is executed using
different leak frequencies. The average input emission factors (for valves)
used in this sensitivity analysis were, therefore, varied according to leak
frequency. Furthermore, since an average SOCMI unit with hypothetical leak
frequencies for all valves was considered, a single average input emission
factor was used for gas and light liquid valves combined. This factor
represents the relative distribution of valves by service in the model units
(43 percent in gas service; 57 percent in light liquid service).
As described in the preceding paragraphs, the analysis of the LDAR
model sensitivity to occurrence rate involves variation of occurrence rate,
A-39
-------
I
-p»
o
c
QJ
Ol
0.
O)
*->
ia
O)
(J
c
01
i_
3
(J
(J
O
TD
I
2
1-
Figure A-5.
5 10 15 20
Leak frequency, percent
Thirty-day occurrence rate as a function of leak frequency
for valves In gas and light liquid service.
-------
leak frequency, and average input emission factor due to the interrelation
of these three parameters. Table A-4 presents the inputs and results of the
LDAR model executed for this test of sensitivity. The results for monthly
leak detection and repair are presented for a model unit B containing 926
valves in gas and light liquid service. Several observations are apparent
from the results presented in the table:
(1) The fraction of valves screened remains relatively constant
for the inputs tested;
(2) The fraction of valves maintained decreases with decreasing
occurrence rate since fewer leaks appear with lower
occurrence rate;
(3) The cost of the leak detection and repair program decreases
with decreasing occurrence rate largely as a function of the
maintenance required;
(4) The cost of the leak detection and repair program begins to
stabilize at low occurrence rates since the amount of
maintenance required becomes small at these low occurrence rates;
(5) The effectiveness of the leak detection and repair program
declines rapidly with decreasing, occurrence rate, illustrating
a strong relationship between these two variables;
(6) The emissions reduction achievable also decreases markedly
with decreasing occurrence rate, due to the strong dependence
of effectiveness on occurrence rate and the relationship '.,.;
between average input emission factor and leak frequency.
Cost effectiveness (i.e., cost per unit VOC controlled) usually
provides a good means of comparing control options, regulatory alternatives,
and even the sensitivity of model input values. The net cost effectiveness,
which accounts for product recovery credits resulting from the program being
examined, is also used to compare results, As seen in Table A-4, the
emissions reduction achievable has the largest effect on net cost effective-
ness since a relatively small variation in gross annualized costs is noted
for the range of inputs analyzed. The net cost effectiveness of monthly
A-41 .
-------
-is.
ro
TABLE A-4. SUMMARY OF INPUTS AND RESULTS OF LDAR MODEL SENSITIVITY TO
OCCURRENCE RATE, LEAK FREQUENCY, AND EMISSION FACTOR FOR
VALVES IN AN AVERAGE MODEL UNIT B
Case
1
?
3
4
5
Uunrlerly
Occurrence
Rate , Percent
3.72
1.06
1.30
1.08
0.930
Leak Frequency,
Percent
10
4
2
1
0.5
Average. Input Emission
Factor , kg/hr/valve
0.007B7
0.00385
0.0025
0.00183
0.0015
Fraction
Screened
11.79RB
11.0935
11.9251
11.9414
11.9492
of Valves
flalnblned
0.1706
0.0906
0.0639
0.0500
0.0435
Reduction
Effectiveness
0.705
0.52Z
0.349
0.106
0.061
Emissions
Reduction , Hg/yr
45.0
16.3
7.00
2.76
0.74
Gross
Annua. 11 zed
Costc. $/yr
11760
9840
9210
0880
B720
Net Cost .
Effectiveness .
t/»ig
(39)e
304
1000
2910
11450
Occurrence rale (30-day) taken as a function of leak frequency: OCC - 0.0976 (LF) + 0.264( Quarterly rate taken as three tines the 30-day rate.
Averaqe Input emission factor taken as a function of leak frequency (using emission factors for leaking valves and for nonleaklng valves from Chapter 2
of the AID) and based on the 43/57 split between gas and light liquid valves In the model units: I
Average Input emission factor = [LEF - HLEF](LF) * t«LEF]
where: I.EF » 0.0683 kg/hr/valve
MI.EF - 0.00116 kg/hr/valve
Emissions reduction and gross annual 1 zed costs based on 926 valves (gas and light liquid) In model unit B.
dNel cost effectiveness Includes a credit of $300/Mg of VOC reduced.
Parentheses Indicate an overall credit of the program due to recovered or saved product.
-------
leak detection and repair of valves increases with decreasing occurrence
rate. Again, this change is a result of changing all three input parameters
(occurrence rate, leak frequency, and average input emission factor), not
just a single parameter.
Since all of these parameters act together, EPA considered how these
results should be evaluated in selecting the standards. Of the three
choices for the independent variable, the leak frequency would be of the
most utility. Certainly, the results in Table A-4 indicate the importance
of occurrence rate in establishing the number of leaks in a unit and the
purpose of any leak detection and repair program is to reduce the number of
leaks, thereby reducing emissions. But measuring occurrence rate is time-
consuming and costly. While they provide an accurate measure of emissions,
mass emission rates (emission factors), too, are costly to establish. Leak
frequency, on the other hand, is a quantity readily measurable in any
process unit. It provides an indicator of occurrence rate and can be used -
to approximate emission factors (as described in Chapter 2 of the AID).
Therefore, leak frequency was selected as the independent variable against
which the cost effectiveness should be presented.
Figure A-6 presents the net cost effectiveness for the cases (presented
in Table A-4) as a function of leak frequency. As illustrated in the
figure, the net cost effectiveness of monthly leak detection and repair of
valves increases dramatically as the leak frequency decreases below about
1 percent. This relationship lends support to establishing alternative
standards for valves as discussed in Chapter 14.
Increases in emissions from inspection to inspection also are affected
by various parameters. Emissions increases resulting from leak occurrences
may in some cases (as stated by the commenter).comprise the largest portion
of emissions from equipment. This is,- in general, reasonable since the
emission factor for equipment which leaks is much larger than the emission
factor for nonleaking equipment. There may, however, be other factors which
can result in large portions of emissions being attributed to other
equipment categories. For example, if a high rate of unsuccessful repair is
A-43
-------
10,000
3,000
I 6,000
u
0>
246
Leak frequency, percent
Figure A-6. Cost effectiveness as a function of leak frequency
for monthly leak detection and repair programs for
valves in SOCMI units.
A-44
-------
considered, a significant number of unrepairable components can accumulate
prior to turnaround repair. In this case, the emissions associated with
unrepairable components can account for the larger share of total emissions.
Identification of which category of equipment components account for .the
greatest portion of emissions aids the operator in determining specific
improvements for his leak detection and repair program. But this identifi- .
cation should not overshadow the fact that all of these factors contribute ;
to increases in VOC emissions. Leak detection and repair programs are
effective in reducing both the number of leaks and the total mass emissions.
Improving repair effectiveness also results in enhanced emissions reduction.
Comment: The same commenter (IV-N-10) concluded that emissions carried over.
from inspection to inspection are small because the model provides.that
emissions are substantially reduced at each inspection. . . -
Response: It is true that under certain conditions the emissions, estimated
by the LDAR model, that are carried over from inspection to inspection can
be small. But this is not a function of the model itself. Rather it is
dependent upon the values selected as input parameters for the specific .<
cases evaluated. For instance, the amount of emissions carried over from.,...
inspection to inspection is affected by the number of equipment .components ,v
that cannot be repaired on-line successfully and by the proximity of the ._.--
inspection period to turnaround. If there are large numbers of unsuccess^-.-..
fully repaired equipment components, then the emissions carried over can ,be;,
relatively large since these components will continue to emit VOC at a g.iven
rate. The model assumes no change in emission rate for unrepairables at
each subsequent inspection prior to turnaround.
Comment: A commenter (IV-N-10) recommended that leaks which recur
immediately (i.e., early recurrences) should be treated as unsuccessful
repairs. The commenter said that applying the same repetitive emission
reduction to leaks which recur is inappropriate. .
A-45
-------
Response: The LDAR model does, in some ways treat early recurrences like
unsuccessful repairs. It assigns the same emission factor to these two
categories. However, EPA makes a distinction between the categories. Early
recurrences are pieces of equipment found leaking, repaired, and found
leaking again shortly thereafter. Unsuccessful repairs are pieces of equip-
ment for which repair attempts failed to reduce the screening value below
10,000 ppmv. Since it is possible to repair early recurrences, it is
important (in reducing VOC emissions) to locate them and attempt repair
again.
EPA has no reason to believe that emission reductions achievable
through maintenance on early leak recurrences are any different from
emission reductions achievable through maintenance attempts on unsuccessful
repairs. Therefore, the same emission factor is assigned to both
categories. The emission factor represents a 63 percent reduction over the
emission factor for uncontrolled equipment components. As explained in the
AID, this emission reduction is based on the Maintenance Study (IV-A-10).
The 63 percent reduction is not applied repetitively as indicated by
the commenter. The model computes emission factors for four categories
(leaking, nonleaking, early recurrences, and unrepairable) and assigns
equipment components to each category. If a valve, for example, exhibited
early recurrence for two inspection periods, it would be assigned to the
same category with the same emission factor; its emission factor would not
change. These emission factors for each category are then applied to the
fraction of the population of valves in each category. The resulting
products are summed to yield the average controlled emission factor for
valves operating under the leak detection and repair program.
Comment; One commenter (IV-N-8) generally agreed with EPA concerning
uncertainties tn determining uncontrolled emission rates, such as why they
vary from process to process, and why some regulatory scenarios give
negative results. This commenter thought that the differences were caused
less by cyclic maintenance than by differences in leak occurrence,
maintenance practices, and repair effectiveness, however.
A-46
-------
vVrKt^-;"
The negative results obtained for some monitoring intervals caused
another commenter (IV-N-7) to doubt that the model could be used in a
meaningful way.
Response: There are many variables that affect the emission level of a
process unit. All the items cited by the commenter (IV-N-8) contribute to
this level. Differences in these factors from unit to unit contribute to
the differences seen in the levels of various units. The uncertainties
associated with determining uncontrolled emission rates are further
complicated by the interrelation of these factors. Admittedly, this makes
understanding negative results difficult. While some technical questions
(such as uncontrolled emissions levels) are not fully answered, the LDAR
model does represent EPA's most current understanding of fugitive emissions
behavior and incorporates the most reliable information on fugitive
emissions and effectiveness of maintenance in reducing emissions. r
The technical problems involved in fine-tuning the baseline level of
emissions do not alter the facts that fugitive emissions of VOC do exist in
SOCMI and that maintenance reduces those emissions.
Comment: . A commenter (IV-N-8) questioned the assumption that repaired
leakers experiencing early failure have the same probability of repair as
initial leakers.
Response: The studies of fugitive VOC emissions and maintenance effects on
these emissions do not specifically address the repair efficiency for early
recurrences. This phenomenon of early recurrence for valves was noted in
the Maintenance Study (IV-A-10) and was found to constitute 14 percent of
the valves on which repairs were attempted. The number of valves exhibiting
early. recurrence during .this study was; small (about 22 of the 155 attempted
has no- reason °b '"be'-T.ieve tfrait the. successful: repair rate ; arid
;, emissions reduction for. .early recurring leaks are any- different from these
, valves for occuring leaks. Therefore, a simplifying assumption was made in
A-47
-------
the development of the model to assign the same repair effectiveness to
leaks which occur and recur and to those which recur early. The value used
for emissions reduction associated with successful repair is based on data
from the Maintenance Study (IV-A-10), including occurrences, recurrences,
and early recurrences.
The LDAR model was also used to look at the effect of the fraction of
equipment exhibiting early leak recurrence. Based on the LDAR results for a
single model unit B operating under a monthly leak detection and repair
program, only two valves are found to exhibit early recurrence when
14 percent recurrence is used. Furthermore, there is less than three
percent difference in the emissions reduction for the leak detection and
'repair programs where 14 and zero percent early recurrence are used.
Comment: One commenter (IV-N-3) asked for clarification of how shutdown
repair efficiency is handled in the LDAR. He said information concerning
shutdown repair efficiency was not provided in the AID.
Response: The repair efficiency of equipment at turnaround is assumed to be
100 percent in the LDAR model. On page 4-22 of the AID, this is indicated
as "all unrepaired sources [equipment designated as unrepaired] are repaired
at'.the turnaround." Repair at turnaround is expected to be more comprehen-
sive than normal repair, but the effectiveness is.not expected to be exactly
100 percent. The LDAR model does have the capability of examining other
values for turnaround repair effectiveness. In varying the turnaround
repair effectiveness from 100 percent to 90 percent, only a three percent
drop in emissions reduction is seen for a model unit B using a monthly leak
detection and repair program for valves. At 95 percent turnaround repair
effectiveness, there is only a one percent difference in emissions reduc-
tion. Since turnaround repair effectiveness is expected to be higher than
95 percent, the emissions reduction presented for the 100 percent case is
indicative of that expected for a unit operating under a rule.
A-48
-------
Comment: One commenter in two comment letters (IV-N-3; IV-N-9) requested
that confidence intervals which the LDAR is capable of generating be
published with the modeling results. The commenter said that the end points
of the confidence intervals should be considered in formulating a regulatory
strategy and that they should be used to distinguish between various
regulatory options.
Response: The use of confidence intervals in distinguishing between various
regulatory options is addressed in a memorandum filed at Docket Item
No. IV-B-27. The memorandum explains that the use of confidence intervals
generated by the LDAR model in this manner is inappropriate. The confidence
intervals generated quantify uncertainty in individual output variables.
They are useful in examining the output properties for cases that employ
completely independent inputs, such as Plant A vs. Plant B. But these
confidence intervals do not represent the uncertainty of differences between
cases using common inputs. Since the comparison of the regulatory options
(monitoring intervals) involves the use of common inputs, the use of
confidence intervals generated by the LDAR model to examine differences
between the regulatory options is inappropriate.
Comment; One commenter (IV-N-7) felt that the main problem with the model
is in the assumption for occurrence and recurrence rates. The rate appeared
unrealistically high to the commenter who felt that the rates should not be
extrapolated linearly. He found the assumption of the same number of leaks
occurring in the first and twelfth months unrealistic. He attributed the
negative efficiencies associated with annual programs to the assumption of
linearity.
Response: The occurrence and recurrence rates used in the AID for valves
were^deye'loped in the Maintenance, Study .(IV-A-vlQf)... An exponential model* was
fit to the experimental data" to develop, the estimates. The; results? of this -
procedure for val.ve.s are compared to a linear extrapolation of the quarterly
occurrence rate in Figure A-7. (The rates used' in the LDAR model, producing
A-49
-------
15
Linear assumption
a
-------
the results in the AID were linearly approximated from the quarterly rate.)
As noted by the small difference between the estimates, even at an annual
monitoring interval, an assumption of linearity for valves is a close
approximation of the modeled occurrence rate. Based on this small
difference, an assumption of linearity for the valve leak occurrence rate is
not the reason for the negative efficiencies associated with annual
monitoring programs. In fact when the LDAR model was reexecuted using the
corrected occurrence rate for annual monitoring, a negative efficiency still
resulted for valves in light liquid service (IV-B-23).
In analyzing data on leak occurrences for valves, it was found that
leaks occur randomly. This concept formed the basis for estimating leak
occurrence rates from the data. Since leaks do occur in a random fashion,
the probability of a leak occurring in any given month will be equal and
independent from the probability of leaks occurring in other months.
Comment: A commenter (IV-N-9) objected to subjecting the population of
valves maintained at turnaround to the 14 percent probability of leak
recurrence.
Response: The question of leak recurrence after turnaround maintenance has
not been answered by the available studies on fugitive VOC emissions. There
is, however, no reason to believe that early leak recurrence after turn-
around maintenance does not happen. The best indicator (currently
available) of the rate of recurrence is 14 percent. The importance of the
assumption of 14 percent leak recurrence after turnaround can be addressed,'
however, by analyzing the magnitude of the assumption. The recurrence rate
is applied to only a small population of equipment (previous early recur-
rences and occurrences), resulting in a smaller number of leak recurrences
after turnaround. For example, for a typical model unit B operating under a
monthly leak detection and repair program, 12 valve leaks would occur and
seven valve leaks would- recur compared to a total of 926 valves in the unit.
Furthermore, under this assumption, the cost and emission reductions
computed are conservative. By applying a 14 percent recurrence rate at
turnaround, more maintenance is performed; thus, the costs are overestimated
A-51
-------
Since the sources that experience early leak recurrence do not return to the
higher leaker emission rate, the emission reductions are underestimated.
Comment: The same commenter (IV-N-9) said the basis for comparing
controlled and uncontrolled regulatory programs is inappropriate for
predicting emission reductions for leak detection and repair programs
applied to new sources. The commenter explained that EPA has chosen to
calculate fugitive emissions from present uncontrolled model plants and
emission reductions over a one-year fugitive emission control program. Mass
emissions before and after the one-year period are then used to calculate
the emission reductions achieved. The commenter called this method
inappropriate and said that a better method consisted of comparing emission
reductions before and after a turnaround.
Response: The commenter has apparently misunderstood how the emissions and
emission reductions are computed by the LDAR model. The model does average
the emissions and reductions over the turnaround period, and includes the
impact of turnaround repair. The comparison made in the model is the time-
averaged value over the turnaround period compared to the pre-program level
(the initial, field-measured conditions). This is the comparison presented
by EPA in the AID. The data used to calculate emissions are from plants
which represent a cross-section of SOCMI and, thus, should reflect the
emissions expected from new plants in the absence of standards.
The LDAR model presents another comparison to the pre-program level.
This other comparison is by monitoring interval. It permits evaluation of a
LDAR program performance with time.
A.3.2 Model Input Parameters
Comment: The use of a single industry-wide occurrence rate was considered
unsupported by the SOCMI data (IV-N-9; IV-N-10). One of the commenters
(IV-N-10) stressed the importance of treating the occurrence rate parameter
appropriately because it is one of the most important parameters in the
A-52
-------
model. The commenter asserted that using a single occurrence rate could
lead to meaningless results. He said, the maintenance study showed a wide
range of occurrence rates, 0.2 to 4.1 percent per month. The differences
were said to have persisted for the length of the study and were reported by
the commenters to be significant at the 95 percent confidence level. The
other commenter (IV-N-9) recommended that a range of leak occurrence rates .
be used to examine emission reductions achievable from alternative leak
detection and repair programs.
Another commenter (IV-N-14) said the 1.3 percent per month occurrence
rate is too high for the industry as a whole because it was determined from
high leak plants. . .
Response: Certainly, the occurrence rate is one input parameter that has a
large impact on the cost and effectiveness of leak detection and repair
programs, as evaluated by the LDAR model. A single occurrence rate estimate
for all valves was used in the analyses since it .incorporated all of the
occurrence rate data collected for valves and, thus, represented the .best v.
overall estimate.
The commenter is correct in pointing out that a range of occurrence ..
rate estimates was described for the different process types included in the
Maintenance Study (IV-A-10). In recognizing the variation of occurrence
rates throughout the industry, EPA has further examined the cost and effec-,
tiveness of leak detection and repair programs for valves with varying
occurrence rates. The LDAR results of this analysis are shown in Chapter 14
which discusses alternative standards. (See Section A.I for discussion of.
the distribution of plant leak levels in SOCMI.) , . , . -
Comment: Commenters (IV-N-10; IV-N-8; IV-N-3;. IV-N-9) disagreed with EPA's
selection of values for repair effectiveness and emission reductions
achievable by maintenance on nonrepaired valves.. The commenters favored
using 29 .percent repair effectiveness and a 63 percent emission reduction
for unsuccessful repairs. The commenters contended that if 10 percent are
unrepaired instead of 71 percent as found in the Maintenance Study, it is
A-53
-------
likely that 63 percent emission reduction might not be achievable for the
10 percent unrepaired. One commenter (IV-N-10'} speculated that the
unrepaired valves might even have emission rates greater than average
leakers. Another commenter (IV-N-8) recommended 29 percent repair
effectiveness be used because that number came from the only complete data
which are consistent within themselves. The commenter found it inconsistent
to use SOCMI data for everything else and then use a successful repair
number of 90 percent from petroleum refining data.
Another of the commenters (IV-N-3) contended the data used to arrive at
90 percent repair effectiveness were not valid. The commenter referred to
Table 4-11, saying that the Shell, Union, and Refinery Maintenance data
should not be considered. He said differences in monitoring methods and
instruments made comparisons invalid. He was disappointed that the
selection was based entirely on petroleum refinery data. The commenter
favored using 29 percent repair effectiveness instead of 90 percent because
none of the tests listed in Table 4-11 were designed to develop an average
leak repair as was done in the EPA Maintenance Study. He believed that use
of higher repair efficiencies is inappropriate, as the valves studied were
repaired or removed from service. He called this practice atypical, and
said that it tended to bias repair efficiency data. The commenter
challenged EPA's use of 90 percent repair effectiveness (because he said it
forced a conclusion) that quarterly monitoring is desirable for emission
reductions and cost effectiveness.
Response: The values selected as inputs to the LDAR model for repair
effectiveness and emission reduction due to unsuccessful repair were based
on an examination of all available data. These values resulted from a
reasoned extrapolation of these data to the conditions expected under a
regulation; they are not merely averages of the available data. Moreover,
they are not values based on petroleum refinery data. The data used in
establishing the input values for the LDAR model are data from equipment in
VOC service.
A-54
-------
The commenters suggest that a 90 percent repair effectiveness is high,
especially in view of the 29 percent effectiveness determined in the
Maintenance Study (IV-A-10). However, as discussed further below, the state
of California has regulations in place governing fugitive emissions from
valves. The experience in California is that better than 90 percent repair
effectiveness (essentially 100 percent for valves) is possible for process
units operating under a regulation. Furthermore, the 29 percent repair
effectiveness determined in the Maintenance Study was for simple on-line
maintenance only. It does not reflect the effectiveness of a process unit
operating under standards which require more than simple on-line
maintenance.
The 63 percent emission reduction used for unsuccessfully repaired
valves is at present the best value available. There are no data or other
bases for considering that this value is affected by the effectiveness of
repair. Thus, it is reasonable to assume that the 63 percent reduction is
appropriate for a repair effectiveness of 90 percent, 50 percent, or
29 percent. The only valves that can remain unrepaired are those valves
that necessitate a shutdown for repair. There is no reason to believe that
the unsuccessful repair of these valves would be any less effective in
reducing emissions than unsuccessful repair of valves in general. There-
fore, 63 percent reduction is appropriate to use in estimating impacts of
leak detection and repair programs.
The data used to arrive at the 90 percent repair effectiveness are for
valves in VOC service. The data used in arriving at this value include data
from process units in California which operate under a regulation governing
fugitive emissions. One of the commenters (IV-N-3) expressed concern over
the handling of the repair effectiveness data from one of the process units
in California. He felt that these data did not reflect the real repair
efficiency for the unit since some valves (those that were not repairable
and those requiring off-line repair) were not included in the repair
effectiveness value. As noted in Table 4-11 of the AID, only those valves
for which simple on-line repairs were attempted were included in the data.
The remaining valves that the commenter is concerned about were removed from
A-55
-------
consideration since they were subjected to more elaborate repair procedures
(that would expectedly result in a higher degree of repair effectiveness).
The effect of removing these valves from the data was to lower the overall
repair effectiveness for all valves, not increase it, since off-line repairs
had a higher rate of success (in fact, 100 percent effectiveness) than
on-line repairs.
The data on valves that were repaired off-line were excluded from
Table 4-11, to provide consistency and allow comparison with other data on
maintenance effectiveness. The data in the table represent the success rate
of simple on-line maintenance attempts only; off-line repair is expected to
be more successful. Removal and off-line repair of valves may be atypical
in an uncontrolled process unit. It will, however, be common practice under
a fugitive emissions regulation.
Finally, EPA did not select data to prove a certain result, as
suggested by one of the commenters. The inputs selected for use in the LDAR
model were based on a reasoned examination of all available fugitive
emission studies. The results obtained from the analysis do not indicate
that a quarterly leak detection and repair program should be used for
valves. They show that monthly monitoring, as proposed, achieves a high
degree of emission reduction at a reasonable cost (see Chapter 4).
Comment: In a slightly different view, one commenter (IV-N-8) said the
emission reductions from unsuccessful repair parameter is assumed to remain
constant with time. He said this assumption is apparently based on
Section 3.3 of the Maintenance Study where immediate reductions from
maintenance held up over a six-month period. The commenter admitted to
being puzzled about the data since emissions from a companion control group
of 60 nonleaking, nonmaintained valves actually decreased over a median
77-day period. This behavior he called contrary to his theories. However,
he said, based on present information, he had no option but to agree with
EPA's assumption.
A-56
-------
Response: The assumption of a constant emissions reduction for unsuccess-
fully repaired equipment reflects EPA's best-understanding of fugitive
emissions at this time. The use of this assumption in the LDA.R model is
reasonable since it reflects how the leak detection and repair programs
should be implemented under a rule.
Comment: With respect to input parameters for pumps, one commenter (IV-N-8)
said that with the exception of the emission factor, the input parameters
given in the AID, Table 4-18, are satisfactory.
Response: EPA has examined the available data on pumps in developing the
inputs used in the LDAR model. Emission factor development is discussed
separately in the comments on Chapter 2 (Emission Factors) of the AID.
Comment: Two commenters (IV-N-9; IV-N-10) said that 63 percent emission
reduction should not be used for recurring leaks. The commenters said -';
subsequent unsuccessful repair attempts would result in little, if any,
emission reduction. The commenters recommended that recurring .leaks be-
treated as unsuccessful repairs..
Response: The 63 percent emission reduction used for unsuccessfully , -
repaired valves is based on the results of the Maintenance Study (I-V-A-10);.
This figure was derived from the data collected on all unsuccessfully
repaired valves, whether they were originally leak occurrences or recur-
rences. As discussed in response to a similar comment on the LDAR model,
the same emission reduction was applied to unsuccessfully repaired valves
that had experienced early leak recurrence.
The fractional emission reductions associated with successful repair
and with unsuccessful repair are not applied repetitively for each
monitoring/repair interval as suggested by the commenters. Instead, these
factors are used to determine emission factors which are assigned to
specific categories (leaking equipment, equipment which exhibits early leak
A-57
-------
recurrence, equipment which cannot be repaired, and nonleaking equipment).
The same emission factor would apply to recurring leaks found in sequential
monitoring intervals. The same emission factor is used for equipment which
cannot be repaired and equipment which exhibits early leak recurrence; the
emission factor is determined from the leaking emission factor and the
fractional emission reduction for unsuccessful repair. The resulting
emission factor is then applied to that fraction of the components
determined to be unrepairable or early recurring leaks.
Early recurring leaks, as mentioned above, are treated in some respects
like unsuccessful repairs in the LDAR model. However, leaks that can be
repaired, technically, are cost-effective to repair, whether they have
recurred or occurred. Thus, it would not be logical to consider a valve
which had been repaired and then began to leak again (i.e., recurred) to be
a valve which could not be repaired.
Comment: Several commenters (IV-N-9; IV-N-10) discussed "effective"
occurrence rates versus "unmaintained" occurrence rates. An "effective"
occurrence rate was defined as the rate of occurrence of leaks for valves
operating under a plant's usual maintenance practices. Referring to the
Maintenance Study, the commenters said that the occurrence rates determined
were based on valves for which normal maintenance was not allowed during the
test period. This occurrence rate was called an "unmaintained" occurrence
rate which the commenters said could be quite different from an "effective"
occurrence rate. They said that EPA has used an "unmaintained" occurrence
rate which is inappropriate for representing SOCMI. The commenters asserted
that appropriate "effective" occurrence rates must be used.
Response: The differences between "effective" and "uncontrolled" occurrence
rates, as inputs for the LDAR model, are not expected to be large. In the
Maintenance Study (IV-A-10) of fugitive emission equipment, occurrence rate
data were collected in the most effective manner possible within existing,
operating units. The data collection plan was set up to cause the least
amount of interruption to normal plant operations. For valves, it is
A-58
-------
unlikely that any additional maintenance would have been performed since
only extremely large leaks and improperly operating valves would be repaired
under current industry practices. Given these practices and the large
degree of variability in leak occurrence rate from process unit to unit, the
use of the occurrence rate from the Maintenance Study is appropriate for
analyzing the impacts of leak detection and repair programs on SOCMI.
Moreover, leak detection and repair programs for valves using various
occurrence rates were examined as part of the analysis of alternative
standards for valves (see Chapter 14). ,
Comment: According to two cdmmenters (IV-N-8; IV-N-10), the real world is
probably better represented by less than 100 percent repair efficiency at
turnaround. One of the commenters (IV-N-8) said that it is not normal to
fix any but the obviously malfunctioning valves during turnaround.
Therefore, the commenters said that turnaround maintenance in the sense
contemplated by the proposed regulation does not exist for fugitive
emissions in uncontrolled plants.
Response: EPA recognizes that in existing uncontrolled process units the
effectiveness of turnaround maintenance is less than 100 percent. In
uncontrolled units, only those large leaks"or malfunctioning valves
(especially those critical to process operation) would be maintained at
turnaround. New process units, however, would be subject to standards.
Under the standards, new process units are expected to achieve close to
100 percent turnaround repair effectiveness on equipment that was previously
unsuccessfully repaired.
A.3.3 Modeling Results
Comment: One commenter (IV-N-7) said that negative modeling results
indicate that faulty data could have been used in the model. The same
commenter suggested that negative results may have arisen because the units
tested actually had programs in place which were better than the program
A-59
-------
being evaluated. Another potential reason for negative results provided by
the commenter is that the baseline leak level was measured at a point when
emissions were at a low point in the maintenance cycle.
Response: The possible causes of negative results of the LDAR model can be
examined with an understanding of the parameters used in defining the model.
EPA examined all available data on fugitive VOC emissions carefully in
deriving the inputs used for the LDAR model. The best available data were
selected for use in the LDAR model analysis, and they appear consistent and
reasonable with the current understanding of fugitive emissions. The
analysis and conclusions drawn were presented in the AID.
Some ideas on the causes of negative results of the LDAR model were
presented briefly in the AID (pp. 4-23, 4-24). The commenter presents two
possible reasons for the negative results. Another related reason is the
level to which the performance of a leak detection and repair program is
compared. For instance, a program can be compared to the baseline level
(represented by the measured leak frequency). Alternatively, it can be
compared to an uncontrolled level where leaks accumulate without any action
taken.
The fact that negative modeling results occur does not preclude the use
of-the LDAR model. First, no decisions have been based on any negative
modeling results obtained using the LDAR model. Secondly, the standards are
based on comparisons between levels of control rather than on the absolute
values obtained from the model. Thus, even though some results are
negative, the LDAR model is used in an appropriate manner in selecting the
standards.
Comment: Two commenters (IV-N-I1; IV-N-10) concluded that the success of
any leak prevention strategy will be highly dependent on what happens to a
plant's effective leak occurrence rate. The commenters stressed that there
are ways other than mandated inspections to achieve low effective occurrence
rates.
A-60
-------
Response: At present, there are limited demonstrated means for achieving
low effective occurrence rates. Apart from the use of leakless equipment,
no other control methods have been demonstrated to eliminate leak occur-
rences. Under the standards, however, other methods that will effectively
lower the leak occurrence rate, such as installation of better equipment
(e.g., valves with improved packing), may be used. In fact, process units
choosing this course of action and effectively reducing their leak frequency
may comply with one of the alternative standards and thereby avoid monthly
monitoring. " . ,.
Comment: The same two commenters (IV-N-11; IV-N-10) said reliance on .
mandated inspections to lower effective occurrence rates will increase costs
and emissions because there is no incentive for design, and, operating
practices to minimize effective occurrence rates. The commenters emphasized
the desirability of formulating flexible standards that encourage-building -:
and operating plants to achieve low effective occurrence rates. -
Response: Based on the LDAR analysis presented in the AID, the leak deteo-w
tion and repair programs required under the proposed standards result in :
lowering the cumulative number of leaks in a process unit. This is ' '-;-'.'
evidenced by the emissions reduction achieved by the program. But the :
standards also allow alternative methods for achieving reductions in
fugitive emissions. These methods include the use of leakless equipment
(Chapter 4) and alternative standards for valves (Chapter 14). These r
provisions allow owners or operators to avoid the mandated leak detection
and repair program if they design and operate process units with low- -
effective occurrence rates. Owners and operators have complete flexibility.
in choosing any such approach which effectively achieves low leak
frequencies. . ' . ' . r
The standards are based on the leak detection and repair programs .
described in the AID because these are methods that are currently being used
and can be used by all SOCMI units to reduce fugitive VOC emissions. Other
A-61
-------
methods may be applicable to a given segment of the industry but may not be
applicable universally and, therefore, would not be appropriate for
standards. Moreover, arriving at cost estimates for control techniques
which have not been demonstrated would not be possible.
Comment: One commenter (IV-N-10) recommended that improving turnaround
repair efficiencies be evaluated as a regulatory strategy.
Response: Improving turnaround repair efficiency would lead to a reduction
in fugitive VOC emissions. This is part of EPA's regulatory strategy, as
evidenced by the standards. The standards require the use of all possible
repair efforts at turnaround to remedy those leaks that could not be
repaired on-line or in-place during normal unit operation. As discussed in
Chapter 13, only valves which require valve assembly replacement and for
which replacement parts are not available can remain unrepaired after
shutdown.
Comment: A commenter looking at the LDAR modeling results (IV-N-10)
concluded that if initial percent leaking differences resulted from
different operating and design characteristics, the effective occurrence
rates must also differ, and the emissions for the same inspection period
must be different.
Response: The commenter's conclusion is logical. Many of the characteris-
tics used to describe fugitive emissions are interrelated, making it
difficult to determine which characteristics actually affect the other
measurable characteristics. Supporting evidence for the commenter's thesis
is found in the data for the different process units presented in the
Maintenance Study (IV-A-10) and the Analysis Report (IV-A-14). Each process
type exhibited different leak frequencies and the occurrence rates reported
for each process type also differed.
Regardless of which factor causes the emissions characteristics to
differ, the number of leaks in a process unit will depend on the occurrence
A-62
-------
rate for that unit. If occurrence rates differ, the number of leaks will
differ. Furthermore, the emissions will differ as well, since the total
emissions for a process unit depend on the number of equipment components in
each category (leaking equipment, nonleaking equipment, unrepairable
equipment, and equipment exhibiting early leak recurrence). (The emissions
associated with each category of equipment were discussed earlier in this
section in response to comments on the LDAR model.)
An analysis incorporating some of these ideas is presented in evalua-
tion of alternative standards for valves (Chapter 14). The analysis
considers the cost effectiveness of leak detection and repair programs for
one model unit size over a range of leak frequencies and associated
occurrence rates.
Comment: Negative modeling results were attributed to low occurrence rates
by one commenter (IV-N-8). The same commenter also said that low-leak
plants cannot be explained on the basis of maintenance cycles or maintenance
efficiency. An explanation which seemed more reasonable to the commenter is
that the occurrence rate is low.
Response: As discussed in response to previous comments in the section,
there are several reasons why the LDAR model yields negative results for
some leak detection and repair programs. A low occurrence rate of leaks is
one possible explanation. Based on the mechanics of the LDAR model,
negative effectiveness values for leak detection and repair programs can
result from an occurrence rate estimate that is low in relation to the leak
frequency measured for the process unit. But an important question is does
the low occurrence rate happen naturally or do other factors involved result
in the low occurrence rate. There are several possible explanations of how
low occurrence rates might be measured.
Two factors, relating to field measurements, that could result in low
occurrence rates are (1) too short a period sampled for adequate occurrence
rate determination and (2) interim maintenance being performed on the source
population being sampled. For at least one of the plants sampled during the
A-63
-------
Maintenance Study (IV-A-10), premaintenance and interim maintenance may have
impacted the results. Other explanations of low occurrence rates/negative
results include low or no leak equipment, high maintenance efficiency, and
the position in time of the maintenance study relative to maintenance
activities. The potential impact of the position in the maintenance cycle
was mentioned in Chapter 4 of the AID.
Comment: A commenter (IV-N-8) compared emissions reductions for two cases
by assuming constant controlled emission rates applied to different uncon-
trolled emission levels. Using this method, he obtained large differences
in the emissions reductions obtained using his estimates when compared to
the emissions reductions obtained using EPA's estimates.
Response: The comparative analysis of a quarterly leak detection and repair
program for light liquid valves presented by the commenter is complicated by
his assumption of an emission factor and leak frequency that are vastly
different from those presented in the AID. In his estimates, the controlled
emissions from light liquid valves are apparently taken to be controlled
emissions using EPA estimates, even though his uncontrolled emissions are
almost 40 percent less than the EPA estimate. EPA does not agree with the
analysis presented by the commenter. The controlled emissions level is not
constant as apparently assumed by the commenter. It is dependent upon many
factors, including leak frequency and average emission factor.
A.3.4 Control Device
Comment: Two commenters (IV-N-3; IV-N-8) stated they were not convinced
that local burnout efficiencies were not as acceptable a measure of flare ,
efficiency as an overall or global efficiency measurement. One of the
commenters pointed out that Siege! sampled a planar array of measurement
points above the flame plume. This commenter also pointed out that there is
no EPA approved calculation methodology to estimate flare efficiencies.
A-64
-------
Response: EPA has not specifically discounted the use of local burnout.
measurement to characterize flare efficiency. In the Siege! study, there
was no evidence to show that flare efficiency decreased as the edge of the
plume was approached. Siegel's results support the use of a single point
local burnout as a reasonable measure of flare efficiency. Also, even a
single local burnout sampling point represents an integrated sample since
the plume moves considerably underneath it during sampling even in a light
wind. At the present time, there is no method for determining global
efficiency of flares operated under normal field circumstances. Therefore,
flare efficiency based upon local burnout measurement is the only
practicable method that can be used to measure flare efficiency under actual
field conditions without making the sampling methodology prohibitively
expensive.
An EPA research study presently being conducted will utilize an
experimental method that will allow determination of overall global
efficiency. The commenters agreed that global efficiency measurement is an"
appropriate objective of the EPA research project.
EPA does not plan to develop a method for estimating flare efficiency.
The CMA-EPA study (IV-A-32) has provided sufficient data under conditions
representative of chemical industry operations to allow specification of
conditions that will assure 98 percent or greater combustion efficiency.
Therefore, the need for an estimation method does not exist.
Comment: Two commenters (IV-N-3; IV-N-8) questioned an EPA statement that
flares used for fugitive emission control might be better represented by a
lower flow region than the turbulent region investigated in the DuPont .
study.
Response: The CMA-EPA flare test (IV-A-32) that was completed recently,
investigated a very wide range of flow conditions from very low to a high of
18 m/sec (60 ft/sec). For steam-assisted flares, air-assisted, or flares
operating without assist, flare efficiencies in excess of 98 percent were
measured over this entire range of flow conditions.
A-65
-------
Comment: Two commenters (IV-N-3; IV-N-8) disagreed with the EPA's statement
that the results from the Siege! study were of questionable utility and
applicability to flares used for fugitives control. The limitations of the
Siege! study elicited by the EPA were: (1) a flare tip design allowing
air-fuel prefixing atypical for the majority of chemical industry flares in
the U.S. and (2) high concentrations of hydrogen/low concentrations of VOC
and relatively high Btu content (greater than 1,000 Btu/scf), characteris-
tics not typical of flare gases found in the chemical industry.
Response: Results from the CMA-EPA flare test (IV-A-32) have become
available since the AID (III-B-2) was published. A flare tip without
provision for air-fuel premixing was operated with and without steam assist;
very high efficiencies in the range from 98 to 99+ percent were measured
under the conditions of the test series. One of the commenters (IV-N-3)
also cited these efficiency results in a later comment letter (IV-N-14).
Also during the CMA-EPA flare test, a range of flare gas heat contents,
compositions, and exit velocities were investigated. These characteristics
of the streams are judged to be more representative of flare gases found in
the chemical industry than those tested by Siegel. There were consistently
high efficiencies measured; the only result observed less than 98 percent
occurred at a heat content less than 200 Btu/scf for flare operation without
assist.
Comment: Two commenters (IV-N-3; IV-N-8) made the following points
concerning the EPA position that plant flares may not achieve 95 percent
efficiency:
(1) The CMA-EPA study will provide data on flaring of low Btu gases;
(2) There are no data to show that large flares are not as effective
as small flares;
(3) There are no data to quantify effect of maintenance on flare
performance;
A-66
-------
(4) One company has reported to a trade association that flare
maintenance needs are low while durability of flare tips is high.
Response: The CMA-EPA study (IV-A-32) did investigate the flaring .of low
Btu gas. Data from this test series did show that high flare efficiencies
(i.e., greater than 98 percent) are achievable on streams down to
300 Btu/scf when operated with steam assist and down to 200 Btu/scf when
operated without assist.
At the present time, EPA does not have any data that shows that larger
flares are not as effective as smaller flares in combusting flare gases.
This question will be addressed in the EPA flare research project to
determine if there is any relationship between flare size and combustion
efficiency.
EPA is not aware of any data to quantify the effect of flare
maintenance on flare efficiency. However, flares maintained as required by
the 40 CFR Part 60 General Provisions, operated smokelessly with a flare
flame present, and operated with an exit velocity less than 18 m/sec
(60 ft/sec) [with flare gas heat content greater than 300 Btu/scf for steam
assisted flares or with flare gas heat content greater than 200 Btu/scf for
flares operated without assist] will obtain at least a 98 percent combustion
efficiency. Air-assisted flares will also obtain 98 percent combustion
efficiency on streams with heat contents greater than 300 Btu/scf that meet
certain maximum exit velocity criteria.
Comment: One commenter (IV-N-8) stated that EPA had been hostile to the use
of flares compared to other control devices. This commenter indicated that
EPA should judge flares by the same criteria used for boilers and incinera-
tors. While admitting that flare efficiency is not readily measured, this
commenter thought the most important test of a control device should .be its
cost effectiveness and efficiency, not ease of enforcement.
Response: EPA agrees that control equipment should be .evaluated objectively
in terms of performance and cos*. However, it is also important that means
A-67
-------
be available both to owners or operators and to EPA to assure reasonably
that the expected performance is continuously achieved. In the case of
flares where performance tests are not possible at this time (except in
costly research projects), it is particularly important that performance be
fully characterized before their installation or use is encouraged.
EPA has been aware for some time that flares under some conditions
demonstrate very high measured efficiencies., i.e., as high as 99+ percent.
Older studies have shown much lower efficiencies. However, the CMA-EPA
study (IV-A-32) made available for the first time efficiency measurements
for conditions of flare gas composition, flare gas heat content, and exit
velocity that are representative of the conditions found in the chemical
industry. This study has been instrumental in the adoption of EPA's
position that steam assisted flares, air-assisted flares, or flares operated
without assist achieve at least 98 percent combustion efficiency when
operated smokelessly and within flare gas heat content and exit velocity
specifications.
A.4 COSTS
Comment: One commenter (IV-N-8) recommended that EPA use vendor prices for
estimating monitoring costs. He maintained that EPA's monitoring costs were
too low. The commenter said that monitoring is now offered on a contract
basis by such firms as Espy-Huston and Radian Corporation. He quoted their
monitoring price at $2 per source and up.
Response: EPA contacted both Espy-Huston (IV-E-13) and Radian Corporation
(IV-E-14) regarding the prices quoted by the commenter. Neither organiza-
tion offers services on a per valve basis. Both companies only bid on an
overall job. Furthermore, they frequently offer more than screening
services. Both companies offer data analysis, program design, and other
services in addition to routine screening. EPA has thoroughly checked the
cost basis presented in the BID and the AID and believe it to be the best
cost estimate available.
A-68
-------
Comment: The same commenter (IV-N-8) said that EPA's overhead and adminis-
trative charges are too low. -He provided monitoring labor cost estimates
with all benefits and nonproductive time factored in at $1.05 per valve as
opposed to EPA's estimate of $0.70 per valve. The commenter1s figures
included 1.8 hours per day of nonproductive time and 35 days of leave per
year.
Response: The commenter in hi-S calculations has inappropriately charged all
nonproductive time charges associated with one workeryear to valve leak
detection activities. It is highly unlikely that valve leak detection
activities in a SOCMI process unit will require an entire worker year of
productive effort. For a process unit similar to a model C unit,
1,140 hours (28.5 workweeks) of productive monitoring time would be required
for monthly monitoring of valves in one year. Allowances for nonproductive
time like those cited by the commenter would reduce the total productive
hours available to 1,440 hours per work year. The total monitoring time
estimated for valves in a model C unit would, therefore, amount to
79 percent of the total productive time available in one worker year.
Adding 79 percent of the nonproductive labor charges to the direct labor
requirements for monitoring results in a cost of $0.72 per valve. This
estimate is close to the estimate of $0.70 per valve for direct and indirect
labor charges calculated by applying a factor of 1.4 to the direct labor
requirement.
Comment: The commenter (IV-N-8) cited another reason he felt EPA's
monitoring costs were too low. He said that 2 man-minutes per valve is
optimistic for the average industrial worker. He attributed the low ,
estimate to the fact that EPA's data were obtained under special test
conditions or by contractors paid on a piece work basis who were, therefore,
motivated to complete the job as soon as possible.
He cited data from other SOCMI companies monitoring in petroleum
refineries. He said their monitoring rates range from 100 valves per day
including on-line maintenance to as high as 350 per day without maintenance.
A-G9
-------
He cited an average (based on limited data) of between 200 and 300 valves
per day without on-line maintenance.
Response: The basis for the estimates submitted by the commenter is not
clear. The monitoring time estimates made by EPA are clearly documented in
the BID and the AID. Those estimates were verified by actual field studies
made by EPA. As previously pointed out, the time estimates are very liberal
because they include time spent in collecting data which are not required of
owners or operators .complying with the standards.
Comment: The cost relationships between monthly/quarterly and quarterly
monitoring intervals were questioned (IV-N-3). The commenter asked for
clarification of the assumptions used and the cost numbers presented.
Response: Since the cost basis, assumptions, and methodology for
calculating costs were set forth in the BID and in the AID, the commenter is
apparently requesting that EPA provide another example calculation for
monthly/quarterly monitoring like the one provided for monthly monitoring in
the AID. Accordingly, tables similar to Tables 5-3 through 5-6 of the AID
have been assembled and presented here in Tables A-5 through A-8 for
monthly/quarterly monitoring.
Comment: One commenter (IV-N-3) judged the cost effectiveness of quarterly
monitoring prohibitive if 29 percent repair efficiency is assumed.
Response: As the sensitivity analysis (IV-B-22) shows, varying repair
efficiency has little effect on the results of the LDAR model. The cost for
quarterly monitoring using 29 percent repair efficiency is hardly prohibi-
tive at $452/Mg, nor is the cost of monthly monitoring prohibitive at
$589/Mg.
Comment: One commenter (IV-N-2) stated that the cost of a control device
should be included in the cost analysis. Along similar lines, another
A-70
-------
TABLE A-5. INITIAL LEAK REPAIR LABOR-AMOURS REQUIREMENT FOR VALVES BY MODEL UNIT
No. of Valves Initial Leak Estimated No. Of
Per Model Unit Frequency Initial Leaks
ABC ABC
99 402 1232(G) 0.114 11.3 45.8 140.4
131 524 1618(LL) 0.065 8.5 34.1 105.2
Repair Time, Labor Hours
Man-Hours Required
ABC
1.13 12.8 51.8 158.7
1.13 9.6 38.5 118.9
22.4 90.3 277.6
Based on 75 percent valves repaired on-line in 10 man-minutes and 25 percent repaired off-line in
4 man-hours.
E»
I
J
-------
TABLE A-6 . TOTAL COSTS FOR INITIAL LEAK REPAIR FOR VALVES BY MODEL UNIT
-vl
rxj
Initial Leak Repair Labor
Admin. & Support Costs (0.
Total Costs
Charges ($15/hour)
4 x labor charges)
Annual ized charges for initial leak repair
(0.163 x total costs)
A
$336
134
$470
$77
B
1355
542
1897
309
C
4164
1666
5830
950
alnitial leak repair costs amortized over 10 years at 10 percent interest (CRF = 0.163).
-------
TABLE A-7. ANNUAL MONITORING ANO LEAK REPAIR LABOR REQUIREMENTS
(Monthly/Quarterly Leak Detection and Repair Program for Valves)
Flo.
I'd
A
99
III
of Valves Monitoring Times MonHorimj No. of Lea^s
Hoild Unit Type of Time Monitored labor-Hours Required Per Year
H C Monitoring Man-MIn* Per Year A B C A B C
402 IZ'.Ji'tr.) Instrument Z
«4 1618{l( ) Instrument 2
4.24 14. D
4.23 IB. 5
32.6
56.8
74.0
130. B
174.1 18.6 75.5 231.4
228.3 24.5 98.2 303.2
402.4
Repair Leak Repair
Time, Labor-Hours Required
Han-hours ABC
1.13 21.0
1.13 27.7
48.7
85. 3
111.0
196.3
261.4
342.6
604.0
'instrument monitoring time is 1 minutes for a 2 man team.
e number of leaks found over turnaround 2 from the LDAR model, based on monthly occurrence rate of 1.3 percent.
I.I1AR modeling results. .
-------
TABLE A-8. ANNUAL MONITORING AND LEAK REPAIR COSTS FOR
MONTHLY/QUARTERLY MONITORING OF VALVES BY MODEL UNIT
v- - - - - - -
Monitoring labor-hours
fi;' Repair labor hours
Total labor-hours (Monitoring & Repair)
Labor charges ($15 x total labor-hours)
Admin. & support costs (0.4 x labor charges)
Annual 1zed charge for Initial leak repair
Total costs ($/year)
Product recovery credit3 ($/year)
Net annual 1 zed costs ($/year)
A
32.48
48.75
81.23
$1218
487
77
$1782
(2070)
($290)
B
130.75
196.27
327.02
4905
1962
309
7176
(8240)
(1060)
C
402.41
604.08
1006.49
15097
6039
950
22086
(25300)
(320D)
Product recovery credit is calculated at $300/Mg. The emission reductions are shown in Section 4
of the AID.
Note: Figures in parentheses Indicate credits.
-------
commenter (IV-N-8) stated that a utility allowance is necessary for the
control device.
Response: As discussed in the AID (III-B-2), the costs associated with
operation of a control device are not included in the costs estimated for
the standards. For instance, where an owner or operator chooses to use
equipment to comply with the standards for pumps in light liquid service,
the costs of the standards include the dual seals, the barrier fluid/
degassing reservoir system, and the piping required to, connect the degassing
reservoir with the combustion device on vapor recovery system. The
combustion system (especially considering that smokeless flares are included
in the standards) are assumed to be available in the process unit.
A.5 ECONOMICS
Comment: One commenter (IV-N-2) alleged that the analysis did not treat
each chemical as a profit center within the corporation as a whole as should
have been done. He stated that the analysis equated the economic impact on
a particular chemical as equaling the impact on the corporation as a whole.
Response: Two related issues are addressed by the commenter. The first is
that the analysis focuses on chemicals and not firms. The second is that
the financial and management structure of each firm influences how it would
react to new air pollution standards, but corporate structure was not
expressly considered in evaluating the economic impact of the standard. The
first issue relates to the way EPA develops the cost estimates; the second,
how it evaluates the size of the costs.
The compliance costs for the proposed regulation have been developed
based on an analysis of industry practices and production technology. Where
necessary and appropriate, worst-case scenarios are formulated and the
compliance costs estimated. If significant impacts are projected under such
scenarios, then the impacted chemicals are reviewed in further detail. This
involves an examination of existing facilities and their operating
A-75
-------
characteristics and usually results in the refinement of the initial conser-
vative assumptions. Because most of this effort focuses on chemicals, and
not on corporations, it is easy to get the incorrect impression that EPA
assumes, as the commenter suggests, that economic impacts on a particular
chemical equal the impact on the corporation as a whole. EPA assumes only
that a chemical is not likely to be produced unless all costs, including
compliance costs, are covered. If product prices are sufficient to cover
all costs, including a normal return on investment, then no corporation
producing that chemical should be adversely affected.
As regards the second issue, EPA analyzes the potential economic
impacts of standards on a model unit that can be built and operated
independently, and that, therefore, represents a single investment decision.
Such an approach is completely compatible with the profit center concept
mentioned by the commenter. The only time the size of the corporation
enters the analysis is to determine whether an affected facility would be
operated by a small business. EPA accepts the Small Business Administra-
tion's definition of a small business for the purpose of analyzing the
potential economic impacts on small businesses. However, this does not
detract from the compatability of the economic analysis with the corporate
structure mentioned by the commenter.
Comment: A commenter (IV-N-1) has asserted that the costs of compliance for
small producers are likely to be high because they are unlikely to have
excess labor to devote to VOC emission controls. Further, he also asserts
that controls on small facilities will have low VOC recovery. He concludes
by suggesting that new or modified small facilities be excluded from the
proposed regulation.
Response: Monitoring and repair costs for each model unit have been
estimated and are included in the annualized cost of the standards. EPA
estimates that the annual leak detection and repair labor-hours requirements
for each model unit will be:
A-76
-------
Model Unit A: 214
Model Unit B: 829
Model Unit C: 2,569
These estimates are believed to be reasonable.
The commenter asserts that small producers, presumably as compared to
large producers, have relatively few excess man-hours to devote to this
activity. This comment implies that such excess man-hours are, in essence,
/
free and could be used to comply with the proposed regulation. EPA has not
treated any of the resource requirements, including labor, that are
necessary to comply with the proposed standard, as being free. In all
cases, the requirements and their associated costs have been estimated. To
the extent that any facility has underemployed resources that could be
applied to meeting the standard, then the compliance costs would be lower.
However, EPA has made no such assumption in developing the compliance cost
estimates. Finally, the commenter has not provided any evidence why he
expects that the financial burden on small companies would be great.
The commenter has also proposed a facility size cutoff for the proposed
regulation although he has provided no accompanying data why such a cutoff
is reasonable. It is anticipated, of course, that small facilities will
have smaller reductions in emissions than larger facilities from applying
the controls. EPA is proposing that facilities with an annual capacity of
1,000 Mg or less be exempted from compliance with the proposed regulation.
A.6 COMMENTS ON SUBJECTS NOT COVERED IN THE AID
Comment: Commenters (IV-N-8; IV-N-9; IV-N-3) recommended that EPA adopt a
standard of annual/turnaround monitoring instead of the monthly monitoring
proposed for valves. The commenters maintained that an annual/turnaround ,
monitoring program would be a cost-effective approach. They predicted that
control efficiencies would exceed 90 percent. In a later letter (IV-N-12)
one of the commenters clarified his comments. The program was seen as an
annual leak detection and repair program combined with screening before
A-77
-------
shutdown, attempted repair, and repair of all remaining leakers during
shutdown. The commenter stressed that he did not mean that every process
should be shutdown once a year; although, the industry average for shutdowns
is once a year. The commenter also explained his basis for 90 percent
control efficiency. He said that on-line repair followed by immediate
off-line repair (at shutdown) would approach 100 percent repair efficiency.
Response: EPA estimates for an annual monitoring program coupled with an
annual turnaround show 1.93 Mg/yr reduction in VOC emissions from a model
unit B, a decrease of 3 percent over uncontrolled (baseline) emissions.
Estimates for annual monitoring coupled with a two-year turnaround show a
1 percent decrease in emissions over uncontrolled (baseline). This low
level of emissions reduction yields high cost effectiveness ratios: $1900
and $6700 per Mg for annual monitoring with one- and two-year turnarounds,
respectively. EPA does not consider either of these programs reasonable
alternatives to the standards.
The commenter's estimate of 90 percent control efficiency is actually
an estimate of repair efficiency. EPA agrees that repair efficiency when
repair uis attempted should exceed 90 percent. That is, 90 percent of the
valves found leaking should be repairable to a no-leak status. However,
90 percent repair does not mean 90 percent emission reduction (control
efficiency). The control efficiency numbers presented by EPA are
comparisons of emissions for one year under a leak detection and repair
program with emissions which would occur if the program were not in effect.
Comment: A commenter (IV-N-8), referring to previous submittats, asked that
the percent leaking alternative standard for valves be clarified with
respect to what constitutes a violation. He also asked for a mechanism
through which low-leak plants could be dropped out of the program.
Response: If an owner or operator elects to comply with the percent leaking
alternative standard for valves, a violation would occur when more than
2 percent of the valves in light liquid and gas service within an affected
A-78
-------
facility leak. A leak is defined as a concentration at or above 10,000 ppmv
as measured by Reference Method 21. Affected facilities with a low
percentage of valves leaking (plants with leak frequencies maintained at a
level lower than 2 percent) need only monitor valves for leaks once a year
to determine compliance with this alternative standard. Additional
compliance tests may-be required by the Administrator. Repair must be
attempted on valve leaks detected at the annual-monitoring or at any other
time leaks are detected. Thus, while valves in affected facilities having
low percentages of valves leaking are not exempted from the standards, the
leak detection and repair effort for valves is substantially reduced for
them. .
Dropping process units with low frequencies of valve leaks from
coverage would not be justified for several reasons. First, there would be
no way-for either the owner or^operator or for EPA to insure that the valves
in the affected facility were maintained at a low leak frequency without
continuing requirements for low leak frequency maintenance and periodic
verification. Furthermore, other fugitive emission sources are affected by
the standards, not just valves. It would be unreasonable to drop a unit
from coverage based on valve leak frequency when.emissions from other
fugitive emission sources may be significant. However, there is provision
for waiving the annual performance test (40 CFR 60.8(b)(4)) by demonstrating
by other means to the Administrator that the affected facility is in
compliance with the standards.
Comment: A commenter (IV-N-8) asked that the determination of percent
leaking be changed. He said that the percent leaking should be determined
by dividing the number of valves leaking by the number of accessible valves
in the unit.
Response: It is not clear why the commenter requested this change since its
effect would be to make the percentage of leaking valves in a process unit
higher, thereby making compliance with the alternative standard more
difficult. Inaccessible valves are part of the affected facility and must
A-79
-------
be monitored annually, or,..in the case of unsafe to monitor valves, as often
as safety practices allow. Thus, all light liquid and gas valves should be
included in the total number of valves with the exception of leakless valves
complying with the performance standard of no detectable emissions.
Comment: A commenter (IV-N-8) asked that provisions be made for a soap
bubble type test to be used for inaccessible valves.
Response: Reference Method 21 when promulgated will allow soaping (not soap
scoring) as a method of leak detection for accessible as well as
inaccessible valves. Therefore, soaping may be used for inaccessible valves
as appropriate.
Comment: A commenter (IV-N-8) disliking EPA's monthly/quarterly monitoring
strategy requested that EPA use the LDAR model to evaluate whether it is
worth the cost and nuisance entailed.
Response: The standard for valves is monthly leak detection and repair.
EPA has provided owners and operators of affected facilities the flexibility
of minimizing the time and resources involved in monthly monitoring by
concentrating monitoring effort on leaking valves, i.e., monthly/quarterly
monitoring. However, implementation in this manner is not required.
Monthly monitoring of all valves is acceptable. Also acceptable,is
compliance with one of the two alternative standards for valves.
Implementation of a monthly leak detection and repair program by
concentrating on leaking valves has been evaluated. Results of the evalua-
tion were published in the AID. Costs of various monitoring intervals for
valves were presented in Table 5-7. The costs indicate that monthly
monitoring of all valves in light liquid and gaseous service has reasonable
cos.ts. Therefore, EPA believes that quarterly monitoring of all valves in
light liquid or gaseous VOC service is not an appropriate basis for the
standard. However, EPA believes that implementation of monthly monitoring
by screening demonstrated leakers is effective and also has reasonable costs
A-80
-------
because it allows effort to be concentrated on those valves that leak. In
allowing implementation of monthly monitoring in this way, EPA is
recognizing that some valves leak infrequently. However, it is not EPA's
intention to implement quarterly monitoring of all valves in light liquid or
gaseous VOC service. . .
Comment: A conmienter (IV-N-8) said that anyone using cost effectiveness
numbers and especially incremental cost effectiveness numbers should keep in
mind the uncertainties in the data used to make the estimates. He cited
differences in .EPA's and his estimates of emissions and costs as evidence of
variability. He presented cost effectiveness estimates for light liquid
valves in a Model Unit B performing quarterly leak detection and repair.
His estimate was $l,286/Mg which he compared to an EPA estimated credit of
$55/Mg. _' ^
Response: The commenter is correct in .his assertion that estimates should-.
always be treated with good judgement, and variability in data and in
physical processes represented by data should be considered when estimates.
are used. EPA has followed this philosophy.
Estimates of emissions and costs for SOCMI were developed to assess the
impact of the standards on SOCMI. But the analysis did not stop there. In
developing standards, EPA has also considered the variability within SOCMI.
The standards provide for control of fugitive emissions with reasonable
costs in all situations in SOCMI of which EPA is aware. Providing this .
reasonable cost effectiveness for all parts of SOCMI has resulted in
providing many options from which an owner or operator may.select to comply
with the standards.
The issue raised by the commenter is not a variability issue. Instead,
the issue raised by the commenter is how emissions, emission reductions, and
costs should be calculated. EPA disagrees with the commenter's method of
arriving at a percentage of valves leaking and emission factors as discussed
in Section A.I. EPA also disagrees with his method of arriving at .
controlled emission levels as discussed in Section A.3.
A-81
-------
Comment: One commenter (IV-N-8) concluded from reviewing the confidence
intervals associated with the petroleum refinery emission factors that there
is a sizeable unexplained variance in EPA's correlation of refinery data.
The commenter said that EPA should be trying to rationalize the differences
to insure that their petroleum refining fugitive emission standards are
technically sound.
Response: It is not clear just what variability the commenter is referring
to. A variance component analysis was performed on the petroleum refinery
data. The analysis is presented in Docket Item No. II-A-26. Large portions
of the variability in screening values were found to be due to the effect of
data for pumps and valves. For valves, large portions of the variability
were seen to be due to individual valves. Similarly, an analysis of
variability was performed on SOCMI data (II-A-7). Again, large portions of
the total variation in valve screening values were seen to be due to
variations between valves and day-to-day variations.
Rationalizing this variability as the commenter requests would in this
case be technically infeasible and inappropriate because the variability is
inherent in the valve and its operation. Because the commenter referred to
confidence intervals, he may have been referring to uncertainty attached to
estimates of emission factors. This type of uncertainty arises in any case
where a mean or other statistical estimator is extrapolated from a sample to
a population.
Both the SOCMI and petroleum refining fugitive emission standards are
based on technically sound data collection and analysis methods. Neither
uncertainty attached to estimates nor variability in a data set prevent the
use of that data in developing standards as long as the uncertainty and
variability are considered. These considerations have been made in develop-
ment of the petroleum refinery fugitive emission standards in much the same
manner as the considerations have been made for the SOCMI standards.
Comment: A commenter (IV-N-8) recommended a regulatory approach which
differed from the one proposed by EPA. Instead of a fixed monitoring
A-32
-------
period, the commenter recommended that EPA evaluate a variable monitoring
frequency to maintain..a fixed percentage of valves leaking. The commenter
suggested annual, semiannual, quarterly, and monthly monitoring intervals.
He said an owner/operator would start monitoring on some frequency and if
the number of new leaks exceed some "X" percent for two consecutive times,
he would go to the next more frequent period until he met the "X" percent
leaking figure or reached monthly monitoring. Conversely, if the unit were
below "X" percent leaking for two consecutive times he would drop to the
next longer monitoring period.
The commenter said that such an approach would insure that all affected
facilities maintained an average of X/2 repairable leaks. According to the
commenter, the system would make monitoring frequency a function of leak
rate so that low-leak plants would monitor less frequently than high-leak
units, and it would encourage owners/operators to make engineering changes ...
to rsduce emissions and reward those that did so. He called this approach a
rare feature in federal^regulations.
Two commenters (IV-N-8; IV-N-14) said that an equitable basis for ;:.
controlling chemical industry emissions would be to have controlled emission
rates comparable to those obtained by the NSPS for the petroleum refining
industry. One of the commenters (IV-N-8) judged that the most equitable
basis for setting an allowable percent leaking would be to set "X" percent
leaking times the emission factor equal to mass emissions which would result
from 2 percent leaking in a petroleum refinery. The commenter computed the
allowable percent leaking which would result using SOCMI emission factors
which he recommended (see Section A.I). The resulting allowable percentages
were 14 percent for gas valves and 9.6 percent for light liquid valves. He
said that these values for leak frequency are in the range of unregulated
SOCMI units, and, therefore, the standards are unnecessary.
Response: The commenter's recommended approach incorporates some good
ideas. In fact, many of the aspects of his recommended program have been
incorporated in EPA's alternative standards for valves. One of the alterna-
tive standards incorporates a varying monitoring interval. The other
A-83
-------
incorporates a fixed percentage of valve leaks not to be exceeded. (See
Chapter 14 for a discussion of alternative standards.) The alternative
standards do, as the commenter requested, encourage owners or operators to
reduce emissions and to reward those who succeed in doing so.
It should be remembered that the objective of valve standards is to
reduce emissions to the lowest level which can be attained with reasonable
costs. This objective has dictated an important difference in the two
approaches. Instead of starting with a monitoring interval and shortening
it as recommended by the commenter, the variable monitoring interval
incorporated in the alternative standard has been implemented by starting
with the shortest monitoring interval, lengthening the interval as the
target is successfully met. There is another difference in the percentage
of valves leaking standard. In the context of a single standard (as opposed
to an alternative or optional standard), it would be inappropriate to set a
fixed percentage leaking target for all affected facilities because of
variability in the industry. The fixed percentage in the alternative
standard is a minimum level below which a leak detection and repair program
would not have reasonable costs. It is not a percentage set by a predeter-
mined emission level as recommended by the commenter.
The attempted normalization of emissions from SOCMI to those from
petroleum refineries is not a valid concept within the standard-setting
context. The objective is to achieve the lowest emissions attainable with
reasonable costs, not to achieve the same level of emissions in all
industries. Impacts of the standards for both industries were evaluated
independently. Costs and impacts for best demonstrated technology for both
industries are reasonable. Comments concerning the need for the standards
are addressed in Chapter 2.
Comment: A commenter (IV-N-8) said that many difficulties he had noted in
the proposed standards for SOCMI fugitive emission sources would be eased or
eliminated if the standards were revised to be similar to those for
petroleum refinery fugitive emission sources presented at the July 1981
meeting of the National Air Pollution Control Techniques Advisory Committee.
A-84
-------
He recommended that the appropriate changes be made to make the SOCMI
standard similar to the refinery standards.
Response: Several revisions to the SOCMI fugitive standards have been made
since proposal. The SOCMI fugitive emission standards as well as the
petroleum refining fugitive emission standards represent EPA's best under-
standing of fugitive emissions of VOC. The similarities between the two
sets of standards are not surprising since they affect similar equipment in
similar services. Best demonstrated control techniques are understandably
similar.
Comment: According to a commenter (IV-N-8), EPA seems preoccupied with
methods for estimating industry-wide impacts. He contrasted this concern
with industry's major concern, saying that industry's major concern is for
technically sound, reasonable and equitable standards. ,',';'''
Response: EPA is concerned with technically sound and reasonable standards
as well as their industry-wide impacts. The standards are technically
sound and reasonable and at the same time, they are cost-effective and
affordable for the industry. Since these same criteria are applied when
standards are developed for other source categories, the standards are '
equitable.
Comment: A commenter (IV-N-8) emphasized the fact that in taking a unit
operations approach, EPA has the obligation to see that the standards are
reasonable for all members of a class.
Response: EPA agrees with this comment. EPA is obligated to develop
standards consistent with Section 111 of the Clean Air Act for new,
modified, and reconstructed stationary sources. EPA has fulfilled this
responsibility by developing standards for fugitive emissions of VOC from
SOCMI which reflect best demonstrated technology taking into consideration
costs, nonair environmental impacts, and energy requirements. Inherent in
A-85
-------
this is the need to assure that the standards are reasonable for all members
of a class. In considering the impacts on SOGMI, EPA has evaluated the
impact on all members of SOCMI, and, as appropriate, adjustments have been
made in the standards for all SOCMI members to assure that there are not
unreasonable impacts. For example, an exclusion has been made for units
with production rates so low that compliance with the standards would not be
cost effective. For units with less than 2 percent of their valves leaking,
an option has been provided for eliminating routine leak detection and
repair requirements. Pumps may comply with one of several options,
e.g., leak detection and repair, barrier fluid systems, or enclosed seal
areas. Related comments and responses concerning applicability of the
standards to SOCMI members have been addressed in Chapters 5 and 7.
Comment: A commenter (IV-N-4) complained that the AID did not address some
technical comments he had submitted previously. He said that his comments
concerning vapor pressure/temperature relationships had been ignored.
Response: The comments presented previously by the same commenter have not
been ignored. EPA considers and responds to all comments submitted. Only a
part of the comments were included in the AID, i.e., those which dealt
specifically with methods and data to be used in calculating impacts. The
scope of the AID was clearly defined in the Federal Register Notice of
Availability (IV-I-2) and did not include the topic of vapor pressure/
temperature relationships. The remaining comments are included in the
background document and the comment which the commenter felt was excluded is
discussed in Section 5.2 of of this background document.
Comment: A commenter (IV-N-4) referring to the AID, said the Agency's
statement that, "For reasons only partially understandable, the SOCMI
fugitive emissions data showed a difference in the number of leaking and
non-leaking sources (leak frequency) when compared to the data derived from
petroleum refineries," illustrates a lack of technical expertise .in the
field of workplace exposure control and limited willingness by EPA to learn.
A-S6
-------
He asserted that the Agency needs to develop or contract for the expertise
needed to attain full understanding.
Response: The- commenter has a totally incorrect perspective of EPA desires,
interests and technical expertise. EPA has led the efforts to characterize
the extent of and control approaches for fugitive emissions. Prior to this
activity which has been carried out over the past 5- to 6-years, fugitive
emission rates had not been adequately evaluated and were significantly
underestimated as was the utility.and effectiveness of well-designed leak
detection and repair programs. Throughout this period EPA has sought to
raise the level of understanding in this area and has done so in close .
cooperation with the affected industries.
More specifically, ,EPA has a long history,of research in the field of
fugitive emissions. The chronology of work on the development of standards
for fugitive emissions of VOC from SOCMI was presented in Appendix A of the
Background Information Document (III-B-1) for the proposed standards. Work. ,
began on development of these standards in December 1978 and continued until
the present time (promulgation). EPA studies of fugitive VOC emissions
referenced in the AID include:
. Petroleum Refinery Study (II-A-26)
Four Unit EPA Study (II-B-34).
EPA 6-Unit Study (II-A-23, 24, 25, 28; IV-A-1; IV-A-5)
EPA 24-Unit Study (IV-A-11)
Maintenance Study (IV-A-10)
Analysis Report (IV-A-14)
Analysis of Allied HOPE Unit Data (IV-A-16)
SCAQMD Study (IV-A-30) .
Coke Oven By-product Recovery Plant Study (IV-A-31)
Gas Plant Study (IV-A-28)
Revision of Emission Factors for Non-Methane Hydrocarbons
from Valves and Pump Seals in SOCMI Processes (IV-A-29) .
A-87
-------
Furthermore, the docket includes information concerning fugitive emissions
of VOC collected as early as 1957 (II-I-Z).
Other studies performed by EPA include:
Response factors for VOC analyzers (II-A-20; IV-A-8; IV-A-12;
IV-A-15)
Emission Control Options for the Synthetic Organic Chemical
Industry (II-A-22)
Vapor Pressure Distribution of Selected Chemicals (IV-A-6)
Evaluation of Walkthrough Survey (IV-A-18)
This history of research in the area of fugitive emissions does not
indicate an unwillingness to learn about fugitive emissions and their
control. Although the thrust of the experimental efforts has been protec-
tion of the environment, control of fugitive emissions has an associated
benefit of reducing workplace exposure and saving both money and energy.
Collaboration between EPA and OSHA in control of emissions is evidenced by
the NIOSH document Control of Emissions from Seals and Fittings in Chemical
Process Industries (IV-M-3). In this document, fugitive emissions data
collected by EPA were applied to the subject of workplace exposure. The
report says that EPA's standards should be considered minimum requirements
for areas where workers may be exposed to highly toxic or carcinogenic VOC
fugitive emissions. The report recommends that in these cases additional
equipment, which EPA considers technically feasible but not sufficiently
cost-effective from an environmental point of view, should be incorporated
based on worker exposure considerations.
Comment: A commenter (IV-N-1) asserted that leak frequency is undoubtedly
influenced by vapor pressure. He said that highly volatile compounds are
more likely to leak than are compounds with low volatility. He referred to
Tables 2-12 and 2-19 of the AID as support for his statement. The commenter
said that a comparison of the tables showed that compounds with lower vapor
pressures have lower leak frequencies than compounds with higher vapor
pressures. He presented a table of normal boiling points and vapor
A-ca
-------
pressures at 25°C for ethylene, vinyl acetate, and cumene. He said that EPA
had used the leak rates for these substances with very different vapor
pressures to derive average unit emission factors.
The commenter further asserted that the differences would be much
greater if they were evaluated at their operating temperatures. Another "
commenter (IV-N-4) expressed the same view and maintained that the
definition of light liquid should be changed to reflect vapor pressures -at
actual operating temperatures. He said that fugitive emissions will be
minimized for highly volatile organics operating at low temperatures and for
low volatility substances processed at high temperatures. He said the
standards should apply only to those processes likely to have significant
emissions. He recommended the following definition of light liquid: a
fugitive emission source is in light liquid service if the following
conditions apply:
(1) The vapor pressure of one or more components, present in
concentrations greater than 20 percent by weight, is greater than
760 mm Hg at operating temperature. Vapor pressure may be
obtained from standard reference texts or may be determined
by ASTM Method D-2879.
(2) The fluid is a liquid at operating temperature.
Response: The first commenter (IV-N-1) is correct in his assertion that
highly volatile compounds are more likely to leak than are compounds with
low volatility. Fugitive emissions data indicate that vapor pressure is the
most important factor influencing the frequency of leaks. However, the
commenter is apparently confused concerning the derivation of emission
factors. The quantity of mass emissions were derived from the petroleum
refining study except in the case of gas valves. Furthermore, the emission
factors used were leaking emission factors and nonleaking emission factors.
(See analysis in AID and in Docket Item No. IV-N-10.) This consistency
supports the use of leaking and nonleaking average emission factors to
characterize SOCMI.
A-S9
-------
Although a correlation of leak frequency with operating temperature
seems reasonable, that hypothesis, suggested by both commenters, is not
supported by the data (II-A-26; IV-A-14).. These data show that the effect
of operating temperature was not consistent and was significant in only a
few cases. Based on these data, the standards were designed to eliminate
those sources which tend to leak infrequently (heavy liquid sources)
regardless of operating temperature, and EPA has not been provided any data
nor found any support for changing the standards.
Comment: A commenter (IV-N-9) requested that the model runs used to select
and evaluate the proposed regulatory options be placed in the public docket
for inspection.
Response: The computer print out used in preparation of the final standards
is archived in the docket (IV-B-23).
Comment: A commenter (IV-N-3) complained that Tables 4-12 and 5-7 are
difficult to understand. The commenter concluded that some data were
missing, especially data relating to monthly/quarterly and quarterly
monitoring intervals.
Response: EPA tried to make very clear the methods and data used to arrive
at the estimates in the BID. It is true that the subject of fugitive
emissions is highly complex, and reproducing the estimates presented in the
AID requires some time and effort. The computer print-out generated in
making the estimates in the AID has been placed in the docket (IV-B-23).
Thus, as requested, any numbers the commenter was having difficulty
reproducing are available.
It is not clear just what data the commenter felt were missing with
regard to monthly and monthly/quarterly monitoring intervals. It could be
that they are referring to blank spaces in Table 4-12. There are entries
left blank in Table 4-12 for the monthly/quarterly monitoring intervals
because the ABCD and modified ABCD models cannot evaluate that monitoring
A-90
-------
scheme. However, the numbers used in the estimates are shown in the column
labeled "LDAR." With that exception the same information was presented for
all the monitoring intervals considered. Another possibility is that the
commenter is requesting an example of cost calculations for the monthly/
quarterly program. An example of the cost calculations for a monthly/
quarterly leak detection and repair program for valves is provided in
Section A.4. . .. . .
Comment: A commenter (IV-N-3) complained that the AID failed to provide
enough information to analyze fully and to critique the LDAR model. He,
therefore, .requested an extension in the comment period. He further
complained that the technical note describing the LDAR model was of such
poor printing quality that it could not be used to reproduce the computer
runs.
Response: An extension in the comment period was granted. As previously
noted, EPA recognizes the complexity of the subject of fugitive emissions
and realizes that analysis requires some time and effort. EPA made every
effort to make the necessary information available to the public for review
and comment. A very legible copy of "Model for Evaluating the Effects of
Leak Detection and Repair Programs on Fugitive Emissions" has been placed in
the docket (IV-A-22). In addition, a copy of the document was sent to those
commenters requesting it. It should be .noted that another commenter .
(IV-N-10). was able to run the LDAR model (computer program) while this
commenter was having difficulty.
Comment: The same commenter (IV-N-3) protested that the LDAR technical note
did not use the same input data as that used in the AID.
Response: The technical note (IV-A-22) is a description of the LDAR model
and the software designed to implement it. The input data shown in the
technical note were used for illustrative purposes. The model and software
can be implemented with any input data. To evaluate impacts of the
A-91
-------
standards, EPA has chosen input'data which are the most reasonable for that
purpose. These selections of input data are detailed in the AID.
Comment: A commenter (IV-N-1) submitted that data presented in the AID
confirm the fact that uniform application of control measures in an industry
as diverse as SOCMI is not supported. He said that the regulatory approach
must be sufficiently flexible to reflect differences in substances
regulated, processes used, and sizes of units.
Response: The commenter1s vague reference to data presented in the AID
makes it difficult to know just what data he is referring to. EPA is aware
that variability exists within SOCMI due to different processes, different
chemical substances, and different levels of unit complexity. EPA is also
aware that such variability required consideration when establishing the
standards. As a result, the standards accommodate this variability. For
example, an owner or operator may elect to comply with a fixed percentage of
valves leaking or an alternative work practice standard inste.ad of a monthly
leak detection and repair program according to which plan best suits his
unit's operations. Additionally, control device requirements allow the :use
of several different types of control devices, including elevated smokeless
flares.
EPA has evaluated each fugitive emission source in terras of variability
in the industry. The standards accommodate all of the variability EPA is
aware of. All control techniques known to be equivalent have been included
in the standards. Also, equivalency provisions have been provided so that
new control techniques may be used as they are demonstrated.
Comment: A commenter (IV-N-6) submitted a copy of a draft report,
"Frequency of Leak Occurrence,in the Specialty Organic Chemical Manufac-
turing Industry" prepared by S , as evidence that there is no fugitive
emissions problem among specialty chemical producers. He further commented
that the proposed NSPS for SOCMI "is based on data that do not accurately
characterize this industry." The commenter referred to the five process
A-92
-------
units studied as a representative cross-section of the specialty chemicals
industry. He said .that the results showed a negligible frequency of leaks.
Response: For several reasons the S report and the comments received do
not contain information which is relevant or useful for the SOCMI fugitive
emission standards.
First, the industry for which the S study was designed is the
specialty organic chemical manufacturing industry (SPOCMI), not SOCMI.
There is no indication that any of the five processes screened are SOCMI
processes. Furthermore, the basis for comparison of the two industries is
not clear and is made even more difficult to understand by the fact that .,
SPOCMI has not been well-defined.
A serious limitation on the data itself is the.large proportion of
heavy liquid equipment screened.. Even though previous fugitive emission -
studies have shown negligible emissions from heavy liquid sources, four of
3
the five process units screened in the S study were heavy liquid ...
processes. It is not surprising that few leaks were detected; 86 percent of
the fugitive emission sources screened were in heavy liquid service. The .
fact that few leaks were found merely affirms the previously, determined fact
that fugitive emission sources in heavy liquid service leak infrequently..-
3
It should also be noted that the S study was a small sampling
effort. Only five process units were screened. Furthermore., of the 1034 ,.
fugitive emission sources screened, only 149 were in light liquid service r
and only 82 would be affected by fugitive emission standards if they were in
process units affected by fugitive emission standards. Table A-9. presents a
comparison of leak frequencies for equipment in light liquid service in
SOCMI (Analysis Report, Docket Item No. IV-A-14) and SPOCMI (S3 Report, :
Docket Item No. IV-N-6). The large confidence intervals noted for pumps and
flanges in light liquid service in SPOCMI reflect the small number of these .
equipment that was sampled. . .
A-93
-------
TABLE A-9. COMPARISON OF LEAK FREQUENCIES FOR EQUIPMENT IN
LIGHT LIQUID SERVICE IN S.OCMI AND SPOCMI
SOCMI Results SPOCMI Results
(Docket Item No. IV-A-14) (Docket Item No. IV-N-6)
95% Confidence Percent95% Confidence
Leaking Interval Leaking Interval
Valves
Pumps
Flanges
6
8
1
.5
.8
.3
(6
(6
(0
.1,
.4,
.9,
6.
11
1.
9)
.0)
7)
1.
0
0
1
(0
(0
(0
.3, 6
, 70.
, 12.
.1)
8)
3)
Finally, the report is a draft contractor report which has not been
released to the public. The report is deficient is several respects and
will be substantially revised before it is completed.
EPA recognizes that low-leak units do exist and they will be subject to
the standards. For such low-leak units, alternative standards have been
allowed. These alternative standards would result in no unreasonable impact
on low-leak units.
Comment: One commenter (IV-N-1) objected to the burden the standards place
on small companies. He requested a 10 million pound annual capacity cutoff
to relieve the burden.
Response; An exemption for low production rate units has been provided.
The standards exempt process units producing less than 1,000 Mg/yr of SOCMI
chemicals. EPA's production rate cutoff differs from the commenter's
recommended cutoff level. The commenter's recommendation of 10 million
pounds per year is equivalent to 4,500 Mg/yr. The commenter's basis for
choosing this number is not clear nor did he provide any convincing indica-
tion of an unreasonable burden. EPA's choice of 1,000 Mg/yr is based on an
analysis of cost effectiveness and is presented in Section 5.5 of this
background document.
Comment; A commenter (IV-N-5) submitted data from his plant to show that
greater than 90 percent control of fugitive emissions can be achieved by
A-94
-------
capping and plugging open-ended lines and valves. The commenter said that
the regulatory approach of capping open-ended lines and valves has been
overlooked by EPA.
Response: The commenter1s calculations apparently reflect an inordinately
large number of open-ended lines in his process unit, although it was
difficult to discern just how many from his calculations. The commenter's
uncontrolled emissions estimates show 60 percent of the total uncontrolled
fugitive emissions from open-ended lines. In contrast, uncontrolled
emission estimates for EPA's model units include 6 percent from open ends
only, or 40 percent from open-ended lines and associated valves.
EPA realizes that covering open-ended lines is important in reducing
fugitive emissions of VOC from SOCMI units. Covering open-ended lines has
been made a part of the standards. However, EPA does not agree that
plugging and capping open-ended lines is enough. Control of other fugitive-
emission sources is also important in significantly reducing fugitive ?
emissions of VOC. Furthermore, industry representatives have said that
plugging open ends is now standard practice in new plants. In that regard,
covering open-ended lines is a part of the baseline control level. As shown
in Chapter 6, since capping open ends is a part of the baseline level of
control, no emission reductions are assumed.
Comment: A commenter (IV-N-10) emphasized the desirability of formulating
flexible regulations that encourage building and operating plants to achieve
low "effective" occurrence rates. The commenter said that plants can either
keep "effective" occurrence rates low by good design and maintenance
practices or allow high "effective" occurrence rates and monitor frequently.
He said that frequent monitoring is analogous to treating symptoms, whereas
lowering the "effective" occurrence rate treats the disease.
Response: EPA agrees and has incorporated the commenter's suggestion. The .
commenter is expressing good ideas in his recommendations for concentrating
control strategies on occurrence rates (rates at which leaks occur over
A-95
-------
time). However, his strict use of occurrence rates has practical limita-
tions. There are two major problems with developing control strategies
which reduce occurrence rates. First, to date, the technology for reducing
occurrence rates, while possible, has not been demonstrated, and it is
unlikely that equipment substitution or changes in operating procedures
could be applied throughout SOCMI. The second problem is one of measure-
ment. Measuring occurrence rates is an expensive proposition, requiring
several months of intensive work to arrive at a satisfactory estimate. The
cost of requiring such measurements would be unjustifiably high, just as it
would be unreasonable to expect an owner or operator to bag all of his
fugitive emission sources to estimate his emissions.
For practical reasons, the control strategy embodied in the standards
uses the easily measurable manifestation of the occurrence rate, or leak.
frequency, as an indicator of leak rate. Also included in the strategy is
the demonstrated control technology of maintenance for reducing emission
rates. This approach is reasonable in cost effectiveness terms and can be
applied throughout SOCMI. However, the standards achieve the purposes
sought by the commenter of allowing those units that can reduce their
occurrence rates to a level, as indicated by a low frequency of leaks, to
comply with an alternative standard. The alternative standards allow units
with low occurrence rates (and, therefore, low leak frequencies) to reduce
the amount of leak detection required. Equivalency provisions are also
available to any owner or operator who can demonstrate control by any other
method as effective as the method provided in the standards.
Comment: A commenter (IV-N-4) disagreed with EPA's judgement that the
Four-Unit study should be eliminated from consideration. He said that the
data had been reviewed and found valid in a review by Jones, et al., in
1982. , .
Response: EPA has reviewed the paper referred to by the commenter,
Jones, Alan, Sydney Lipton, and Jeremiah Lynch. "Critical
Review of Fugitive Emission Data." Presented at AICHE meeting
in Orlando, Florida. February 28 - March 3, 1982.
A-96
-------
The paper (IV-M-55) includes the four-unit study in the review of fugitive
emissions data. However, EPA could find no evaluative judgements in the
paper concerning the validity of the four-unit study.
Comment: The same commenter voiced concerns about the validity of EPA's
data in three letters (IV-D-6; IV-N-4; IV-N-13). He had two major concerns:
(1) the K-factor used in the petroleum refinery studies and (2) the statis-
tical distribution model used in analyzing fugitive emissions data. He
cited the following two references in support of his objections:
Jones, A. !_., Lipton, Sydney, and Lynch, Jeremiah, "Critical
Review of Fugitive Emission Data," AICHE meeting,
March 3, 1982, Orlando, Florida.
Harvey, P. A., and Jones, A. !_., "Fugitive Emission Data:
Statistical Treatment," UK Health and Safety Executive and
Esso Europe Inc. (Received as Private Communication from
B. C. Davis, Exxon Chemical Co.)
Specifically, the commenter said that the paper by Jones, et al. reinforced
questions about the handling of refinery data previously raised in his
earlier comments. He used the Harvey and Jones paper to support his
contention that the statistical analysis was not satisfactory and that a
better statistical distribution could be found.
Response: EPA has responded to the commenter's previously submitted
comments concerning the confusion over published K-factors. His questions
about the validity of the data base apparently center on a typographical
error in the K-factor generated in the petroleum refinery studies. (See
Chapter 3 for a response to his comment.) A letter explaining the
discrepancy noted by the commenter has been filed in the docket (IV-B-7).
Furthermore, copies of this letter have been sent to the commenter on two
separate occasions. The reasons for the commenter's remaining concerns are
unclear in the face of actions taken by EPA to respond to his concerns.
A-97
-------
EPA requested a copy of the Harvey and Jones paper from the commenter
(IV-C-78). In one of his letters, the commenter informed EPA that he did
not have permission to release this paper and that the paper would have to
be obtained from the originating company. He subsequently arranged to have
a copy of the Harvey and Jones work sent to EPA (IV-0-55).
EPA reviewed the paper; the three main points made by Harvey and Jones
were:
The methods used by EPA usually fit the data quite well.
The lognormal distribution, used by the EPA to derive fugitive
emission factors, is not the only distribution function which can
be used to describe the fugitive emission data.
The Gamma distribution and other distribution functions can also
be used to fit fugitive emission data.
Preliminary work indicates that a better prediction of fugitive
emission factors may be obtained for some types of fugitive
emission sources using the Gamma distribution rather than the
lognormal distribution.
While other distributions may also be applicable in certain cases, the
lognormal distribution was chosen for describing fugitive emissions in the
Petroleum Refining studies since it was the simplest distribution that
adequately described the data and it was the only distribution that allowed
a closed form expression to be written. In Appendix C of the Petroleum
Refining studies, this selection was documented. Furthermore, it was noted
that only three of the twelve data sets examined failed a test for normality
using this assumed distribution. The commenter draws the conclusion from
the paper that the gamma distribution better describes fugitive emissions
data, but he does not present any comparison of the two distributions, even
though such a comparison is possible.
Though the two papers referenced by the commenter do review and discuss
fugitive emissions data and analysis procedures, it is important to note the
emphasis of these particular papers. Both Harvey and Jones, and Jones,
A-98
-------
et al. address workplace exposure and industrial hygiene. The papers are
written from the standpoint of using fugitive emissions data collected for
environmental purposes to study industrial hygiene. These different
applications require different data analysis techniques in some instances.
Comment: The commenter (IV-N-4; IV-N-13) demanded that EPA audit the data
collection and analysis performed by Radian Corporation, a contractor for
EPA. He said that he had requested an audit previously, and it had yet to
be performed.
Response: EPA continuously audits all contractor work as it is performed as
well as when It is finalized. In addition, EPA work is open to public
scrutiny. An additional EPA audit is not warranted.
A.7 PREVIOUSLY SUBMITTED COMMENTS
A number of comments were received in the correspondence pertaining to
the AID that had been submitted previously on the background documents and
proposed standards. The comment summaries are listed in Table A-10 along
with the docket entry number of the correspondence containing the comment
and the section In this document that responds to the comment.
A-99
-------
TABLE A-10. CROSS-REFERENCE OF COMMENT AND SECTION CONTAINING RESPONSE
COfWENTER
IV-N-1
IV-N-3
IV-H-8,
IV-M-9
IV-N-fl,
IV-N-9 ,
IV-N-1
IV-N-8,
IV-N-9
IV-N-8
IV-N-8
IV-N-8
IV-N-fl
IV-N-8
IV-M-8
IV-H-8,
IV-N-9
IV-N-8,
IV-N-9
IV-N-8,
IV-N-9
IV-N-8,
IV-N-9
IV-N-fl,
IV-N-9
CHAPTER/SECTION
CQWENT SUMMARY CONTAINING RESPONSE
1. The standards are disproportionately burdensome 5.5
for small producers. '
2. Flares that have been properly designed and 4.1
maintained are an acceptable control technology for
reducing VOC fugitive emissions.
3. The regulation should be reproposed based on a 3.7
rewritten BID.
4. Modification should be limited to Increases 1n 9.2
emissions of more than 10-20 tons/yr.
5. Reconstruction should be limited to cases of 9.6
Increased emissions.
6. The definition of fugitive emission source Includes 4.8, 4.12
reciprocating pumps and compressors which cannot be
fitted with double seal technology.
7. Categories of chemicals 1n the SOCMI 11st are overly S.I
vague. Specific compounds should be listed.
8. The leak definition should be higher than 10,000 ppn. 4.3
The recommended figure 1s 20,000 ppm. . '
9. Consnenters object to requiring mechanical seals on 4.8, 4.12
pumps and compressors.
10. The requirements for action when liquid' Is dripping 4.3
from seal should be changed. Since the liquid may be
a non-VOC, the requlrments should be changed to require
monitoring by Instrument for VOC concentrations when
a visual leak is detected. ''',
11. Pressure relief device requirements should be that 4.5, 13
the pressure relief device is returned to no detectable
emissions within 5 calendar days after resumption of . ;
normal operation after each episode of pressure
release.
12. The phrase "without VOC emissions to the atmosphere" 4.9
implies a zero emissions requirement. The standard
should be changed to include "by means of a closed
vent system.'
13. Commenters recommended exclusions for Inaccessible 4.2.4, 4.4
valves due to safety reasons and due to restricted
physical access.
14. Delay for repair due to technical 1nfeas1bil1ty should 4.2.3, 13
be clearly defined.
15. The repair time requirements should not apply to 4.2.3, 13
equipment taken out of service.
16. Commenters requested a provision for delay of repair 4.2.3, 13
beyond shutdown for lack of parts for all fugitive
emission sources.
A-100
-------
TABLE A-10. CROSS-REFERENCE OF COMMENT AND SECTION CONTAINING RESPONSE
(CONTINUED)
COMMENTER
iv-*-8,
IV-N-9
iv-M-8.
[V-M-9
IV-M-8
IV-fl-8
IV-N-8
IV-H-4
IV-fl-1
IV-M-1
[V-N-1
IV-IM
IV-M-1.
tv-M-a
IV-H-Z
iv-n-2
IV-K-7
IV-K-7
17.
13.
19.
20.
21-.
22.
23.
24.
25.
26.
27.
28.
29.
30.
COHOT SUHHARY
Vendors and manufacturers should b« able to obtain
equivalency rulings.
The definition of "connector* Is unclear.
Model units should be low. medium, and high leak
occurrence rates.
EPA should develop standards that account for
variability In SOCHI.
In slde-by-flde measurements a TLV calibrated to
hexane gives Measurements different froai those
obtained with an OVA calibrated to ethane.
SOCHI 1s already regulated by OSHA. so there 1s no
need for the standards.
Safety, odor, and toxlclty problems associated with
SOCHI chemicals dictate stringent control measures.
Small producers are not represented In the data base.
Commenters request Implementation of the bubble policy
for not source standards.
Emissions from SOCHI are Insignificant, so there Is
no need for the standards
The vapor pressure cutoff should be changed to 1.5 ps1
to be consistent with State SIP's.
The selection of 0.3 kPa Is arbitrary.
Commenters prefer performance standards to equipment
standards.
Retrofit may not be feasible for existing reciprocating
compressors*
0 lap reduced from J«!fe
>est available copy, wisl?'
CHAPTER/SECTION
CONTAINING RESPONSE
10
5.8
3.3
3.5. 14
12.1
2.2
2.2.5.1
5-5
15
2.1. 8.Z. 8.7
5.2
5.2
3.5
4.6. 4.12. 9.4
-
A-101
-------
-------
APPENDIX B
MONITORING METHODS
-------
-------
APPENDIX B .,
MONITORING METHODS
The standards require that some fugitive emission vent streams be
vented through a closed vent system to a control device (that is designed
and operated for greater than 95 percent control), such .as an incinerator,
flare, boiler, or process heater. The.standards also require that the
control device be monitored to ensure that it is properly operated and
maintained. This appendix presents methods for monitoring control devices:
incinerators, boilers and process heaters, flares, or product recovery
equipment, such as condensers or carbon adsorbers.
Incinerators .
Incinerators must be maintained and operated properly if the standard
is to be achieved on a continuous basis. .The operating parameters that
affect performance are temperature, type of compound being incinerated,
residence time, inlet concentration, and flow regime. Of these variables,
the last two have the smallest effect.on the performance of an incinerator.
Residence time is a design criterion and is not easily altered after the
incinerator is constructed, unless, of course, the vent stream flowrate is
changed. At temperatures above 760°C, the type of compound being burned has
little effect on the efficiency of combustion.
Continuous monitoring of the .incinerator inlet and outlet would be
preferred because it would provide a continuous, direct measurement of
actual emissions and destruction efficiency. However, EPA is aware of no
continuous monitor being used to measure, total VOC at incinerators which
control fugitive vent streams, probably because each of the many different
compounds would have to be identified separately and their concentrations.
determined. Such a monitoring system would be extremely complex for the
determination of individual component concentration and mass flow rates.
Moreover, it would be.relatively expensive since both inlet and outlet
monitors are required to verify that a certain destruction efficiency is
maintained.
J-l
-------
Monitoring of the incinerator operating temperature provides a reliable
measure of the efficiency of the incinerator in destroying organic
compounds. Both theoretical calculations and results of monitoring or
performance tests show that lower incinerator operating temperatures can
cause a significant decrease in VOC destruction efficiency. Temperature
recorders are relatively inexpensive, costing less than $5,000 installed.
They are easily and cheaply operated. Given the large effect of temperature
on efficiency and the reasonable cost of temperature monitors, EPA believes
that temperature is clearly easy to monitor and would provide some measure
of the uniformity of the operation of the incinerator.
Where a combustion device is used to incinerate only waste VOC streams
(and not multiple waste streams from the process unit), flowrate can also be
an indirect indication of changes in destruction efficiency since it relates
directly to residence time in the combustion device. Flowrates of fugitive
emission vent streams are typically small and thus would probably be ducted
with other larger streams to the same incinerator. Under these circum-
stances, the vent stream flowrate (for fugitive emissions) may not always
give a reliable indication of the residence time of the fugitive emission
vent stream in the incinerator. Simple indication of fugitive emission vent
stream flowrate to the incinerator does, however, provide verification that
VOC is being routed to the incinerator. Flow recorders, at an estimated
installed cost of less than $2,000, are inexpensive and require little
maintenance. Therefore, since flow recorders provide verification that
organics-laden streams are being routed to the incinerator for destruction
and they are inexpensive, flowrate is also a reasonable parameter to monitor
the constancy of performance of an incinerator. Flow recorders should be
installed, calibrated, maintained, and operated according to the
manufacturer's specifications.
Boilers
If a fugitive emissions vent is piped to the flame zone of a boiler (or
process heater), it is only necessary to know that the boiler (or heater) is
operating and that the waste gas is flowing to the boiler (or heater).
Records presently maintained for plant operation, such as steam production
B-2
-------
records, would indicate operation. Flow recorders could be installed to
verify flow of the vent stream to the boiler (or heater). For smaller heat
producing units (less than 44 MW (150 million Btu/hr heat input)),
combustion temperature should also be recorded to enable verification of
optimum operation. Boilers (or heaters) with heat input design capacities
greater than 44 MW would not be required to install temperature recorders.
These larger units always operate at high temperatures (>1100°C) and stable
flowrates to avoid upsets and to maximize steam generation rates. Records
that indicate onstream time would be sufficient for these larger boilers (or
heaters).
Flares
Because flares are not enclosed combustion devices, it is not
practically feasible to measure combustion parameters continuously.
Temperatures and residence times are more variable throughout the combustion
zone for flares than for enclosed devices and, therefore, such measurements
would not necessarily provide a good indicator of flare performance even if
measurable. Monitoring of flow rate to the flare is generally unacceptable
from a safety point of view since the flow measurement would present an
obstruction in an emergency vent line. As a result, flare operation is
usually verified by examination of more prominent characteristics.
The typical method of verifying continuous operation of a flare is
visual inspection. However, if a flare is operating smokelessly, it can be
difficult to determine if a flame is present, and it may take several hours
to discover. The presence of a flame can be determined through the use of a
heat sensing device, such as a thermocouple or ultra-violet (U-V) beam
sensor on a flare's pilot flame. The loss or absence of a flame would be
indicated by a low temperature measurement. The cost of available
thermocouple sensors ranges in price from $800 to $3,000 per pilot. (The
more expensive sensors in this price range have elaborate automatic relight
and alarm systems.) Thermocouples used on flares may, however, burn out if
not installed properly. The cost of a U-V sensor is approximately $2,000.
A U-V system is not as accurate as a thermocouple in indicating the presence
of a flame. The U-V beam is influenced by ambient infrared radiation that
5-3
-------
could affect the accuracy. Furthermore, interference between different U-V
beams makes it difficult to monitor flares with multiple pilots. By design,
U-V sensors are primarily used to verify the existence of flames within
enclosed combustion devices. Therefore, based on cost and applicability,
EPA believes thermocouples provide adequate verification of flare operation.
Product Recovery Equipment
Three types of product recovery equipment which might be used in
controlling fugitive emissions vents are absorbers, condensers, and carbon
adsorbers.
Two operating .parameters are the primary determinants of product
recovery device operation for an absorber: the temperature and specific
gravity of the absorbing liquid. Facilities which have installed an
absorber to recover product which otherwise would be lost will generally
monitor a parameter which indicates the degree of saturation of the
absorbing liquid with respect to the product. Specific gravity is commonly
used for this purpose. Devices for measuring the temperature and specific
gravity are available at reasonable cost. The estimated one-time combined
capital investment for such equipment is $8,000. It is considered
reasonable for an operator of a process unit to install, calibrate,
maintain, and operate according to manufacturer's specifications the
requisite devices to monitoring continuously temperature and specific
gravity or such alternate parameters which would indicate the degree of
saturation of the absorbing liquid.
In constrast, the exit temperature of the offgas is the primary
determinant of the efficiency of a condenser. Again, suitable temperature
recorders are available at a reasonable cost. The estimated one-time
capital investment is $3,000. A record of the outlet temperature would
verify that the condenser is properly operated and maintained. EPA believes
an operator can install, operate, calibrate and maintain according to the
manufacturer's specifications a temperature recorder to verify proper
operation of a condenser.
The operation of a carbon adsorber can be monitored by the carbon bed
temperature and the amount of steam used to regenerate the bed. Steam flow
B-4
-------
meters and temperature recorders are available at reasonable cost. The
estimated one-time capital investment for such equipment is $10,000. These
parameters could be monitored to reflect whether the carbon adsorption unit
has been consistently operated and properly maintained. Therefore, EPA
believes that an operator of a carbon adsorber used as a pollution control
or product recovery device could install, calibrate, maintain, and operate
according to manufacturer's specifications an integrating steam flow
recorder and a carbon bed temperature recorder. Some operators may install
vent stream analyzers to aid in maximizing the recovery of organic com-
pounds. No widely accepted performance specifications have been developed
for such analyzers. If an analyzer is Installed without a recorder, the
vent stream should be sampled at the end of the adsorption cycle (at least
once during every 4 hours of operation) and the concentration recorded as a
means of verifying that operational modes remain consistent with the
conditions under which the performance test was conducted.
B-5
-------
------- |