Accidental Release Prevention Requirements: Risk Management Programs Under Clean Air Act Section 112(r)(7)

[Federal Register: June 20, 1996 (Volume 61, Number 120)]
[Rules and Regulations]
[Page 31667-31730]
From the Federal Register Online via GPO Access [wais.access.gpo.gov]

ENVIRONMENTAL PROTECTION AGENCY
40 CFR Part 68
[FRL-5516-5]
RIN 2050-AD26

Accidental Release Prevention Requirements: Risk Management
Programs Under Clean Air Act Section 112(r)(7)

AGENCY: Environmental Protection Agency.
ACTION: Final rule.

SUMMARY: The Clean Air Act requires EPA to promulgate regulations to
prevent accidental releases of regulated substances and reduce the
severity of those releases that do occur. EPA is promulgating rules
that apply to all stationary sources with processes that contain more
than a threshold quantity of a regulated substance. Processes will be
divided into three categories based on: the potential for offsite
consequences associated with a worst-case accidental release; accident
history; or compliance with the prevention requirements under OSHA's
Process Safety Management Standard. Processes that have no potential
impact on the public in the case of an accidental release will have
minimal requirements. For other processes, sources will implement a
risk management program that includes more detailed requirements for
hazard assessment, prevention, and emergency response. Processes in
industry categories with a history of accidental releases and processes
already complying with OSHA's Process Safety Management Standard will
be subject to a prevention program that is identical to parallel
elements of the OSHA Standard. All other processes will be subject to
streamlined prevention requirements. All sources must prepare a risk
management plan based on the risk management programs established at
the source. The source must submit the plan to a central point
specified by EPA; the plan will be available to state and local
governments and the public. These regulations will encourage sources to
reduce the probability of accidental releases of substances that have
the potential to cause immediate harm to public health and the
environment and will stimulate the dialogue between industry and the
public to improve accident prevention and emergency response practices.

DATES: The rule is effective August 19, 1996.

ADDRESSES: Supporting material used in developing the proposed rule,
supplemental notice, and final rule is contained in Docket No. A-91-73.
The docket is available for public inspection and copying between 8:00
a.m. and 5:30 p.m., Monday through Friday (except government holidays)
at Room 1500, 401 M St. SW, Washington, DC 20460. A reasonable fee may
charged for copying.

FOR FURTHER INFORMATION CONTACT: Craig Matthiessen at (202) 260-8600,
Chemical Emergency Preparedness and Prevention Office, U.S.
Environmental Protection Agency, 401 M St. SW, Washington, DC 20460, or
the Emergency Planning and Community Right-to-Know Hotline at 1-800-
424-9346 (in the Washington, DC, metropolitan area, (703) 412-9810).

SUPPLEMENTARY INFORMATION: Judicial Review. Accidental Release
Prevention Requirements: Risk Management Programs Under Clean Air Act
Section 112(r)(7) were proposed in the Federal Register on October 20,
1993 (58 FR 54190). A supplemental notice was issued on March 13, 1995
(60 FR 13526). This Federal Register action announces the EPA's final
decisions on the rule. Under section 307(b)(1) of the Act, judicial
review of the Accidental Release Prevention Requirements: Risk
Management Programs is available only by the petition for review in the
U.S. Court of Appeals for the District of Columbia Circuit within 60
days of today's publication of this final rule. Under section 307(b)(2)
of the Act, the requirements that are the subject of today's notice may
not be challenged later in civil or criminal proceedings brought by the
EPA to enforce these requirements.

Regulated Entities

Entities potentially regulated by this action are those stationary
sources that have more than a threshold quantity of a regulated
substance in a process. Regulated categories and entities include:

Category Examples of regulated entities

Chemical Manufacturers............ Industrial organics & inorganics,
paints, pharmaceuticals, adhesives,
sealants, fibers
Petrochemical..................... Refineries, industrial gases,
plastics & resins, synthetic rubber
Other Manufacturing............... Electronics, semiconductors, paper,
fabricated metals, industrial
machinery, furniture, textiles
Agriculture....................... Fertilzers, pesticides
Public Sources.................... Drinking and waste water treatment
works
Utilities......................... Electric and Gas Utilities
Others............................ Food and cold storage, propane
retail, warehousing and wholesalers
Federal Sources................... Military and energy installations

This table is not intended to be exhaustive, but rather provides a
guide for readers regarding entities likely to be regulated by this
action. This table lists the types of entities that EPA is now aware
could potentially be regulated by this action. Other types of entities
not listed in the table could also be regulated. To determine whether a
stationary source is regulated by this action, carefully examine the
provisions associated with the list of substances and thresholds under
Sec. 68.130 (59 FR 4478), the proposed modifications (61 FR 16598,
April 15, 1996) and the stay of implementation of the affected
provisions until the proposed modifications are final published
elsewhere in today's Federal Register, and the applicability criteria
in Sec. 68.10 of today's rule. If you have questions regarding the
applicability of this action to a particular entity, consult the person
listed in the preceding FOR FURTHER INFORMATION CONTACT section.
The following outline is provided to aid in reading this preamble:

I. Introduction and Background

A. Statutory Authority

B. Background

II. Discussion of Final Rule

A. Applicability

B. Program Criteria and Requirements

C. Hazard Assessment

D. Prevention Programs

E. Emergency Response

F. Risk Management Plan (RMP)

G. Air Permitting

H. Other Issues

III. Discussion of Comments

A. Tiering

Rationale

Program 1 vs. Program 2 and Program 3 Criteria

[[Page 31669]]

a. Potential for Offsite Impact

b. Accident History

c. Other

3. Program 2 vs. Program 3 Criteria

a. Number of Employees

b. SIC Code

c. Site-specific, Risk-based Criteria

d. Accident History

e. Other

4. Program 1 Requirements

a. Certification of No Environmental Impact

b. Signs

c. Emergency Response Program

d. Other

5. Program 2 Requirements

a. Streamlined Program

b. Other Regulations

c. Emergency Response Program

B. Offsite Consequence Analysis

Worst-Case Release Scenario

Mitigation Systems

a. Worst-Case Release Scenario

b. Alternative Scenarios

Populations Affected

Number of Scenarios

Technical Guidance

Modeling Parameters

a. Endpoints

b. Meteorology

C. Consideration of Environmental Impact

Inclusion of Environmental Impacts

Environments to be Considered

Level of Analysis Required

D. Program 3 Consistency with OSHA PSM Standard

Prevention Program

Enforcement

Exemptions

E. Relationship to Air Permits

General Relationship between the Part 68 and Part 70 programs

Impact of EPA's Proposal on Air Permitting Programs

Part 68 as an ``Applicable Requirement'' under Part 70

Role of the Air Permitting Authority

Air Permit Application Contents

Air Permit Contents

Completeness Review

Interaction of the Implementing Agency and the Permitting
Authority

Designated Agency

Reopening Air Permits to Incorporate Section 112(r)
Requirements

Use of Air Funds

Other Issues

F. General Definitions

Significant Accidental Release

Stationary Source

Process

Offsite

Other Definitions

G. Risk Management Plan (RMP)

Level of Detail

RMP Content

Submission

Other Issues

H. Prevention Program

I. Accident History

J. Emergency Response Program

K. Registration

L. Model Risk Management Programs

M. Implementing Agency Audits

N. Public Participation

O. Inherently Safer Technologies

P. Coverage by Other Regulations

General Issues

DOT Transportation Regulations

Other EPA Regulations

Other Federal Regulations

State and Local Regulations

Q. Industry-Specific Issues

Oil and Gas Facilities

Retail Facilities

a. Propane Retailers

b. Ammonia Retailers

Refrigeration Systems

Other Operations

R. Implementing Agency Delegation

S. Accident Reporting

T. Other Issues

OSHA VPP

Qualified Third Party

Documentation

IV. Section-by-Section Analysis of the Rule

V. Required Analyses

A. E.O. 12866

B. Regulatory Flexibility Act

C. Unfunded Mandate Reform Act

D. Paperwork Reduction Act

E. Submission to Congress and the General Accounting Office

I. Introduction and Background

A. Statutory Authority

This rule is promulgated under sections 112(r), 301(a)(1), Title V
of the Clean Air Act (CAA) as amended (42 U.S.C. 7412(r), 7601(a)(1),
7661-7661f).

B. Background

The CAA Amendments of 1990 amend section 112 and add paragraph (r).
The intent of section 112(r) is to prevent accidental releases to the
air and mitigate the consequences of such releases by focusing
prevention measures on chemicals that pose the greatest risk to the
public and the environment. Section 112(r)(3) mandates that EPA
promulgate a list of regulated substances, with threshold quantities;
this list defines the stationary sources that will be subject to
accident prevention regulations mandated by section 112(r)(7). EPA
promulgated its list of substances on January 31, 1994 (59 FR 4478)
(``List Rule'').

As noted elsewhere in today's Federal Register, EPA has stayed
certain provisions of part 68 that were promulgated as part of the List
Rule. The stayed provisions are being addressed in amendments to the
List Rule, which were proposed in 61 FR 16598 (April 15, 1996).
Therefore, EPA has not taken final action on provisions of the Risk
Management Program rule that apply to regulated substances, mixtures,
and stationary sources addressed by the stayed provisions. Final action
will be deferred until EPA takes final action on the proposed
amendments to the List Rule.

Section 112(r)(7) mandates that EPA promulgate regulations and
develop guidance to prevent, detect, and respond to accidental
releases. Stationary sources covered by these regulations must develop
and implement a risk management program that includes a hazard
assessment, a prevention program, and an emergency response program.
The risk management program must be described in a risk management plan
(RMP) that must be registered with EPA, submitted to state and local
authorities, and made available to the public. On October 20, 1993, EPA
published a Notice of Proposed Rulemaking (NPRM) for the section
112(r)(7) regulations (58 FR 54190). (For a summary of the statutory
requirements of section 112(r) and related statutory provisions, see
the October 20, 1993, NPRM).

Following publication of the proposed rule, EPA held four public
hearings and received approximately 770 written comments. Because of
these comments, EPA issued a supplemental notice of proposed rulemaking
(SNPRM) on March 13, 1995 (60 FR 13526) for comment on: approaches for
setting different requirements for sources that pose different levels
of hazard (tiering); worst-case releases and other hazard assessment
issues; accident information reporting; public participation;
inherently safer approaches; and implementation and integration of
section 112(r) with state programs, particularly state air permitting
programs. EPA held a public hearing on March 31, 1995, in Washington,
DC, and received more than 280 written comments. Today's rule reflects
EPA's consideration of all comments; major issues raised by commenters
and EPA's response are briefly discussed in Section III of this
preamble. A summary of all comments submitted and EPA's response to
them is available in the Docket (see ADDRESSES).
EPA has proposed to delist explosives from Sec. 68.130.
Consequently, explosives are not addressed in this rule. EPA had also
requested at the time of the final List Rule comments on whether
flammable substances, when used as fuel, posed a lesser intrinsic
hazard than the same substance handled otherwise (59 FR 4500, January
31, 1994). The comments submitted lacked data that would justify a
lesser level of hazard consideration for flammable fuels; hence, the
Agency will not adopt a fuel use exemption for purposes of threshold
quantity determination.

With today's rule, EPA continues the philosophy that the Agency
embraced in implementing the Emergency Planning and Community Right-toKnow
Act of

[[Page 31670]]

1986 (EPCRA). Specifically, EPA recognizes that regulatory
requirements, by themselves, will not guarantee safety. Instead, EPA
believes that information about hazards in a community can and should
lead public officials and the general public to work with industry to
prevent accidents. For example, today's rule requires covered sources
to provide information about possible worst-case scenarios. EPA intends
that officials and the public use this information to understand the
chemical hazards in the community and then engage in a dialogue with
industry to reduce risk. In this way, accident prevention is focused
primarily at the local level where the risk is found. Further, today's
rule builds on existing programs and standards. For example, EPA has
coordinated with Occupational Safety and Health Administration (OSHA)
and the Department of Transportation (DOT) in developing this
regulation. To the extent possible, covered sources will not face
inconsistent requirements under these agencies' rules. EPA is
encouraging sources to use existing emergency response programs, rather
than develop a separate and duplicative program under this rule. In
addition, today's rule scales requirements based on the potential risk
posed by a source and the steps needed to address the risk, rather than
imposing identical requirements on all sources.
To accommodate the concerns of small businesses, EPA is providing
guidance with reference tables that covered sources can use to model
the offsite consequences of a release. EPA is providing a model RMP
guidance for the ammonia refrigeration industry, and will develop
similar guidance for propane handlers and drinking water systems. As
today's rule is implemented, EPA hopes that other industry sectors will
work with EPA to develop model RMPs for other processes, thereby
reducing costs for individual sources. Finally, today's rule requires
industry to submit RMPs centrally in a format and method to be
determined by EPA. Working with stakeholders, EPA will develop
mechanisms to allow industry to use appropriate electronic technology
to register with EPA and submit RMPs. In turn, all interested parties
will be able to access electronically the data in RMPs. This method of
submission and access avoids a potentially significant amount of
paperwork for all involved parties and promotes uniformity. Users will
be able to develop databases for specific purposes and compare RMPs for
various sites across the country. In turn, industries' use of the data
will promote continuous improvement, for example, through new safety
technologies. As the method for submitting RMPs is developed, EPA
invites the participation of all stakeholders, including industry,
state and local governments, local emergency planning committees,
environmental groups, and the general public.

II. Discussion of Final Rule

A. Applicability

The owner or operator of a stationary source that has more than a
threshold quantity of a regulated substance in a process must comply
with these requirements no later than June 21, 1999; three years after
the date on which a regulated substance is first listed under
Sec. 68.130; or the date on which a regulated substance is first
present in more than a threshold quantity in a process, whichever is
later.

B. Program Criteria and Requirements

Under today's rule, processes subject to these requirements are
divided into three tiers, labeled Programs 1, 2, and 3. EPA has adopted
the term ``Program'' to replace the term ``Tier'' found in the SNPRM to
avoid confusion with Tier I and Tier II forms submitted under EPCRA,
also known as Title III of the Superfund Amendments and Reauthorization
Act of 1986 (SARA Title III). Eligibility for any given Program is
based on process criteria so that classification of one process in a
Program does not influence the classification of other processes at the
source. For example, if a process meets Program 1 criteria, the source
need only satisfy Program 1 requirements for that process, even if
other processes at the source are subject to Program 2 or Program 3. A
source, therefore, could have processes in one or more of the three
Programs.

Program 1 is available to any process that has not had an
accidental release with offsite consequences in the five years prior to
the submission date of the RMP and has no public receptors within the
distance to a specified toxic or flammable endpoint associated with a
worst-case release scenario. Program 3 applies to processes in Standard
Industrial Classification (SIC) codes 2611 (pulp mills), 2812 (chloralkali),
2819 (industrial inorganics), 2821 (plastics and resins), 2865
(cyclic crudes), 2869 (industrial organics), 2873 (nitrogen
fertilizers), 2879 (agricultural chemicals), and 2911 (petroleum
refineries). Program 3 also applies to all processes subject to the
OSHA Process Safety Management (PSM) standard (29 CFR 1910.119), unless
the process is eligible for Program 1. Owners or operators will need to
determine individual SIC codes for each covered process to determine
whether Program 3 applies. All other covered processes must satisfy
Program 2 requirements. Program requirements and differences are
illustrated on Tables 1 and 2:

Table 1--Program Eligibility Criteria

Program 1 Program 2 Program 3

No offsite accident history. .................. Process is subject
to OSHA PSM.
No public receptors in worst- The process is not Process is in SIC
case circle. eligible for code 2611, 2812,
Program 1 or 3. 2819, 2821, 2865,
2869, 2873, 2879,
or 2911.
Emergency response ..................
coordinated with local
responders.

Table 2--Comparison of Program Requirements

Program 1 Program 2 Program 3

Hazard Assessment:
Worst-case analysis......... Worst-case analysis. Worst-case analysis.
Alternative releases Alternative
releases.
5-year accident history..... 5-year accident 5-year accident
history. history.
Management Program:
Document management Document management
system. system.

[[Page 31671]]
Prevention Program:
Certify no additional steps Safety Information.. Process Safety
needed. Information.
Hazard Review....... Process Hazard
Analysis.
Operating Procedures Operating
Procedures.
Training............ Training.
Maintenance......... Mechanical
Integrity.
Incident Incident
Investigation. Investigation.
Compliance Audit.... Compliance Audit.
.................. Management of
Change.
.................. Pre-startup Review.
.................. Contractors.
.................. Employee
Participation.
.................. Hot Work Permits.
Emergency Response Program:
Coordinate with local Develop plan and Develop plan and
responders. program. program.
Risk Management Plan
Contents:
Executive Summary........... Executive Summary... Executive Summary
Registration................ Registration........ Registration.
Worst-case data............. Worst-case data..... Worst-case data.
5-year accident history..... Alternative release Alternative release
data. data.
Certification............... 5-year accident 5-year accident
history. history.
Prevention program Prevention program
data. data.
Emergency response Emergency response
data. data.
Certification....... Certification.

The owner or operator of a covered process must: (1) prepare and
submit a single risk management plan (RMP), including registration that
covers all affected processes and chemicals; (2) conduct a worst-case
release scenario analysis, review accident history, ensure emergency
response procedures are coordinated with community response
organizations to determine eligibility for Program 1 and, if eligible,
document the worst case and complete a Program 1 certification for the
RMP; (3) conduct a hazard assessment, document a management system,
implement a more extensive, but still streamlined prevention program,
and implement an emergency response program for Program 2 processes;
and (4) conduct a hazard assessment, document a management system,
implement a prevention program that is fundamentally identical to the
OSHA PSM Standard, and implement an emergency response program for
Program 3 processes.

Measures taken by sources to comply with OSHA PSM for any process
that meets OSHA's PSM standard are sufficient to comply with the
prevention program requirements of all three Programs. EPA will retain
its authority to enforce the prevention program requirements and the
general duty requirements of CAA Section 112(r)(1). EPA and OSHA are
working closely to coordinate interpretation and enforcement of PSM and
accident prevention programs. EPA will also work with state and local
agencies to coordinate oversight of worker and public safety and
environmental protection programs.

C. Hazard Assessment

EPA has adopted the worst-case definition proposed in the SNPRM.
For all substances, the worst-case release scenario will be defined as
the release of the largest quantity of a regulated substance from a
vessel or process line failure, including administrative controls and
passive mitigation that limit the total quantity involved or the
release rate. For most gases, the worst-case release scenario assumes
that the quantity is released in 10 minutes. For liquids, the scenario
assumes an instantaneous spill; the release rate to the air is the
volatilization rate from a pool 1 cm deep unless passive mitigation
systems contain the substance in a smaller area. For flammables, the
worst case assumes an instantaneous release and a vapor cloud
explosion.

For the final rule, EPA has adopted the term ``alternative release
scenarios'' to replace the term ``other more likely scenarios'' found
in the NPRM and SNPRM. The non-worst-case accidental releases for the
hazard assessment portion of the risk management plan were presumed
``more likely to occur'' and ``more realistic'' than the worst case.
EPA believes sources should have flexibility to select non-worst-case
scenarios that are the most useful for communication with the public
and first responders and for emergency response preparedness and
planning. Catastrophic accidental releases are typically rare events;
the words ``more likely'' suggests certainty of occurrence.
Consequently, the scenarios other than worst case provided in the
hazard assessment are called alternative release scenarios. For
alternative scenarios, sources may consider the effects of both passive
and active mitigation systems.

One worst-case release scenario will be defined to represent all
toxics, and one worst-case release scenario will be defined to
represent all flammables held above the threshold at the source.
Additional worst-case release scenario(s) must be analyzed and reported
if such a release from another covered process at the source
potentially affects public receptors that would not be potentially
affected by the first scenario. EPA recognizes that this approach may
be problematic for some sources such as batch processors and warehouses
where use of listed substances or inventory may vary considerably
within an RMP reporting period. EPA suggests that owners or operators
of such processes develop a worst-case scenario for future chemical use
and inventory based on past practices to minimize the need for frequent
revision of their worst-case scenario. For alternative release
scenarios, one scenario is required for each toxic substance and one to
represent all flammable substances held in covered processes at the
source.

An endpoint is needed for the offsite consequence analysis.
Appendix A of today's rule lists the endpoints for toxic substances
that must be used in worst-case and alternative scenario assessment. The endpoint for a toxic
substance is its Emergency Response Planning Guideline level 2 (ERPG-2)
developed by the American Industrial Hygiene Association (AIHA). If a
substance has no ERPG-2, then the endpoint is the level of concern
(LOC) from the Technical Guidance for Hazards Analysis, updated where
necessary to reflect new toxicity data. EPA recognizes the limitations
associated with ERPG-2 and LOC values and is working with other
agencies to develop Acute Exposure Guideline Limits (AEGLs). When these
values have been developed and peer-reviewed, EPA intends to adopt them
through rulemaking as the toxic endpoints for this rule. For
flammables, vapor cloud explosion distances will be based on an
overpressure of 1 psi; for alternative flammable releases, radiant heat
distances will be based on an exposure of 5 kW/m\2\ for 40 seconds. For
vapor cloud fires and jet fires, the lower flammability limit provided
by the National Fire Protection Association (NFPA) or other sources
shall be used.

EPA selected 1.5 meter per second (m/s) wind speed and F
atmospheric stability class as the default worst-case scenario
meteorological conditions. If the owner or operator has meteorological
data that show that higher minimum wind speeds or less stable
atmospheric class conditions existed at the source at all times in the
previous three years, then the higher wind speed and different
stability class may be used. Alternative release analyses may use sitespecific,
typical meteorological conditions. If the owner or operator
has no data on typical meteorological conditions, then conditions used
in the RMP Offsite Consequence Analysis Guidance (3 m/s and D
stability), may be used. Although EPA is providing technical guidance
and reference tables for worst-case and alternative release scenario
assessments, owners or operators may use any generally recognized,
commercially or publicly available air dispersion modeling techniques,
provided the modeling parameters specified in the rule are used.
For the hazard assessment and the RMP, populations potentially
affected are defined as those within a circle that has as its center
the point of release and its radius the distance to the toxic or
flammable endpoint. Owners or operators may use Census data to define
this population, and may update those data if they are inaccurate. EPA
suggests that owners or operators use LandView, an electronic
publication of environmental, geographic and demographic information
published by EPA and the Bureau of Census. The presence of schools,
hospitals, other institutions, public arenas, recreational areas, and
large commercial and industrial developments that can be identified on
street maps within this circle must be noted in the RMP, but the number
of people occupying them need not be enumerated. The presence of
environmental receptors within this circle must also be listed. EPA has
defined environmental receptors as natural areas such as national or
state parks, forests, or monuments; officially designated wildlife
sanctuaries, preserves, refuges, or areas; and Federal wilderness
areas, that can be exposed to an accidental release. All of these can
be identified on local U.S. Geological Survey maps or maps based on
USGS data.

The five-year accident history will cover all accidents involving
regulated substances, but only from covered processes at the source
that resulted in serious on site or certain known offsite impacts in
the five years prior to the submission of each RMP. EPA has replaced
the definition of significant accidental release with specific
definitions of the types of releases to be covered under each of the
specific requirements previously associated with this definition.

D. Prevention Programs

EPA has retained the management system requirement proposed in the
NPRM, but only for Program 2 and 3 processes. EPA has moved the
management system requirement from the prevention program section to
the general requirements section because it should be designed to
oversee the implementation of all elements of the risk management
program. The owner or operator must designate a qualified person or
position with overall responsibility for the program and specify the
lines of authority if responsibility for implementing individual
requirements is assigned to other persons or positions.
In the SNPRM, EPA proposed a Program 2 prevention program that
covered training, maintenance, safety precautions, and monitoring, but
did not specify any particular actions. EPA solicited comment on
whether specific prevention activities should be required for Program 2
sources, such as any of the specific activities initially proposed in
the NPRM. For today's rule, EPA has developed seven specific elements
for the Program 2 prevention program: safety information (Sec. 68.48),
hazard review (Sec. 68.50), operating procedures (Sec. 68.52), training
(Sec. 68.54), maintenance (Sec. 68.56), compliance audits (Sec. 68.58),
and incident investigation (Sec. 68.60). Most Program 2 processes are
likely to be relatively simple and located at smaller businesses. EPA
believes owners or operators of Program 2 processes can successfully
prevent accidents without a program as detailed as the OSHA PSM, which
was primarily designed for the chemical industry. EPA combined and
tailored elements common to OSHA's PSM and EPA's NPRM to generate
Program 2 requirements and applied them to non-petrochemical industry
processes. EPA is also developing model risk management programs (and
RMPs) for several industry sectors that will have Program 2 processes.
These model guidances will help sources comply by providing standard
elements that can be adopted to a specific source. EPA expects that
many Program 2 processes will already be in compliance with most of the
requirements through compliance with other Federal regulations, state
laws, industry standards and codes, and good engineering practices.
The Program 3 prevention program includes the requirements of the
OSHA PSM standard, 29 CFR 1910.119 (c) through (m) and (o), with minor
wording changes to address statutory differences. This makes it clear
that one accident prevention program to protect workers, the general
public, and the environment will satisfy both OSHA and EPA. For
elements that are in both the EPA and OSHA rules, EPA has used OSHA's
language verbatim, with the following changes: the replacement of the
terms ``highly hazardous substance,'' ``employer,'' ``standard'' and
``facility'' with ``regulated substance,'' ``owner or operator,''
``part or rule,'' and ``stationary source''; the deletion of specific
references to workplace impacts or to ``safety and health;'' changes to
specific schedule dates; and changes to references within the standard.
The ``safety and health'' and ``workplace impacts'' references occur in
OSHA's PSM standard in process safety information (29 CFR 1910.119
(d)(2)(E)), process hazards analysis (29 CFR 1910.119(e)(3)(vii)), and
incident investigation (29 CFR 1910.119(m)(1)). These changes are
designed to ensure that OSHA retains its oversight of actions designed
to protect workers while EPA retains its oversight of actions to
protect public health and the environment and to remove possible
interpretations that certain elements of process safety management fail
to account for offsite impacts. Commenters were particularly concerned
about the phase-in of process hazard analyses

[[Page 31673]]

(PHAs). Under the final rule, PHAs conducted for OSHA are considered
adequate to meet EPA's requirements. They will be updated on the OSHA
schedule (i.e., by the fifth anniversary of their initial completion).
This approach will eliminate any need for duplicative analyses.
Documentation for the PHA developed for OSHA will be sufficient to meet
EPA's purposes.

EPA anticipates that sources whose processes are already in
compliance with OSHA PSM will not need to take any additional steps or
create any new documentation to comply with EPA's Program 3 prevention
program. Any PSM modifications necessary to account for protection of
public health and the environment along with protection of workers can
be made when PSM elements are updated under the OSHA requirements. EPA
has modified the OSHA definition of catastrophic release, which serves
as the trigger for an incident investigation, to include events ``that
present imminent and substantial endangerment to public health and the
environment.'' As a result, this rule requires investigation of
accidental releases that pose a risk to the public or the environment,
whereas the OSHA rule does not. EPA recognizes that catastrophic
accidental releases primarily affect the workplace and that this change
will have little effect on incident investigation programs already
established. However, EPA needs to ensure that deviations that could
have had only an offsite impact are also addressed.

E. Emergency Response

EPA has adopted the emergency response requirements found in the
statute, without additional specific planning requirements beyond those
necessary to implement the statute. This action is consistent with the
Agency's effort to develop a single Federal approach for emergency
response planning. The Presidential Review of Federal release
prevention, mitigation, and response authorities (required under
section 112(r)(10) of the Clean Air Act) found that there is seldom
harmony in the required formats or elements of response plans prepared
to meet various Federal regulations. Accordingly, EPA has committed not
to specify new plan elements and/or a specific plan format in today's
rule beyond those that are statutorily required. EPA believes that
plans developed to comply with other EPA contingency planning
requirements and the OSHA Hazardous Waste and Emergency Operations
(HAZWOPER) rule (29 CFR 1910.120) will meet most of the requirements
for the emergency response program. In addition, EPA and other National
Response Team agencies have prepared Integrated Contingency Plan
Guidance (``one plan'') (NRT, May 1996). The NRT and the agencies
responsible for reviewing and approving federal response plans to which
the one plan option applies agree that integrated response plans
prepared in the format provided in this guidance will be acceptable and
be the federally preferred method of response planning. An emergency
response plan that includes the elements specified in this guidance can
be used to meet the requirements in today's rule. The final rule also
provides relief for sources that are too small to respond to releases
with their own employees; these sources will not be required to develop
emergency response plans provided that procedures for notifying nonemployee
emergency responders have been adopted and that appropriate
responses to their hazards have been addressed in the community
emergency response plan developed under EPCRA (42 U.S.C. 11003) for
toxics or coordinated with the local fire department for flammables.

F. Risk Management Plan (RMP)

Owners or operators must submit their first RMP by the date
specified in Sec. 68.10. After the RMP is submitted, changes at the
source may require updates to the RMP other than the standard update
every five years. If a new substance or new process is added, the RMP
will need to be revised and submitted by the date the substance is
first in the process above the threshold quantity. If changes to
processes require revised hazard assessments or PHAs, or if a process
changes Program level, the source must submit a revised RMP within six
months.

EPA intends that the RMP will be submitted in a method and format
to a central point as specified by EPA. States, local entities
including local emergency planning committees (LEPCs), and the public
will be able to access all RMPs electronically. This process will
relieve states and local entities of the burden of filing documents and
providing public access to them without limiting these agencies' or the
public's access to the information.

The RMP is a multi-purpose document. The CAA requires that the RMP
indicate compliance with the regulations and also include the hazard
assessment, prevention program, and emergency response program. EPA is
mandated to develop a program for auditing RMPs and requiring
revisions, where appropriate. The RMP, therefore, must include enough
data to allow the implementing agency to determine, through review of
the RMP, whether the source is in compliance with the rule. EPA,
however, believes that the RMP must serve another function; to provide
information to the public in a form that will be understandable and
will encourage the public to use the information to improve the
dialogue with sources on issues related to prevention and preparedness.
To meet both of these purposes, the RMP will consist of the
source's registration; an executive summary that will provide a brief
description of the source's activities as they relate to covered
processes and program elements; and data elements that address
compliance with each of the rule elements. While the public and
implementing agencies could make use of all sections of the RMP, the
executive summary will provide text descriptions and give the source a
chance to explain its programs in a format that will be easy for
communities to read and understand. The data elements will provide the
implementing agency with the basic data it needs to assess compliance
without asking for detailed documentation. The Agency is considering
development of an RMP form where the data elements of the form would
provide the implementing agency with the basic data it needs to assess
compliance without asking for detailed documentation. All data elements
would be checkoff boxes, yes/no answers, or numerical entries.
This approach will provide data that anyone can download or search.
States, communities, trade associations, or public interest groups may
want to use the data or a subset of the data to create databases that
allow them to compare sources in the same industry or same area. For
example, a local entity will be able to download data from all
reporting sources that are similar to ones in its community to
determine whether the quantities stored and process controls used are
typical. The information will provide the public with data that will
enhance their dialogue with sources. It will also help sources and
trade associations to understand practices in their industries and
identify practices that could be used to reduce risks. The risk
management program documentation will remain at the source and will be
available for review by EPA and the implementing agency.

G. Air Permitting

The SNPRM discussed the relationship between section 112(r) and CAA
air permitting requirements for sources subject to both provisions.
Under the CAA, air permitting authorities must ensure that sources are

[[Page 31674]]

in compliance with applicable requirements to issue a permit. Because
section 112(r) is an applicable requirement, EPA has identified in the
final rule the permit conditions and the actions owners or operators
and air permitting authorities must take to ensure compliance. The
permit must identify part 68 as an applicable requirement and establish
conditions that require the owner or operator of the source to submit
either a compliance schedule for meeting the requirements of part 68 by
the date specified in Sec. 68.10(a) or, as part of the compliance
certification submitted under 40 CFR 70.6(c)(5), a certification
statement that, to the best of the owner or operator's knowledge, the
source is in compliance with all requirements of this part, including
the registration and submission of the RMP. The owner or operator must
also submit any additional relevant information requested by the air
permitting authority or designated agency to ensure compliance with the
requirements of this section. If a permit is already issued that does
not contain the provisions described above, then, the owner or operator
or air permitting authority shall initiate permit revision or reopening
according to the procedures in 40 CFR 70.7 or 71.7 to incorporate the
terms and conditions as described above. EPA also allows the state to
assign the authority to implement and enforce these requirements to
another agency or agencies (the ``designated agency'') to take
advantage of resources or accident prevention expertise that might be
available in these other agencies. Finally, the air permitting
authority or designated agency must: (1) Verify that the source owner
or operator has registered and submitted an RMP or a revised plan when
required; (2) verify that the source owner or operator has submitted
the proper certification or compliance schedule; (3) for some or all
sources, use one or more mechanisms such as, but not limited to, a
completeness check, source audits, record reviews or facility
inspections to ensure that permitted sources are in compliance; and (4)
initiate enforcement action, based on the requirements of this section,
as appropriate.

H. Other Issues

In the SNPRM, EPA discussed three other issues raised by
commenters: accident information reporting, public participation, and
inherently safer technologies. EPA has decided not to develop any
requirements related to these issues at this time. Although EPA
continues to believe that accident reports that provide more detail on
the causes and impacts of accidents could be useful, the Agency has
decided to limit such reporting required under this rule to the fiveyear
accident history mandated by the CAA. When necessary, EPA will use
its authority to investigate individual accidents and to seek
additional information to the extent authorized by CAA section 114
(i.e., to determine compliance with this rule and CAA section
112(r)(1), to support further rule development, and to assist research
on hazard assessment).

Secondly, the Agency encourages sources, the public, and local
entities to work together on accident prevention issues, but believes
that the wide variety and large number of sources subject to this rule
make any single mandatory approach to public participation
inappropriate. RMP information should be used as the basis for dialogue
between the community and sources on accidental release prevention,
risk reduction and preparedness for emergency response. Industry and
the public should continue to use the LEPC as a mechanism for this
dialogue.

Finally, EPA does not believe that a requirement that owners or
operators conduct searches or analyses of alternative process
technologies for new or existing processes will produce significant
additional benefits. Many commenters, including those who support these
analyses, indicated that an assessment of inherently safer design
alternatives has the most benefit in the development of new processes.
Industry generally examines new process alternatives to avoid the
addition of more costly administrative or engineering controls
associated with a design that may be more hazardous in nature. Although
some existing processes may be judged to be inherently less safe than
others, EPA believes most of these processes can be safely operated
through management and control of the hazards without spending
resources searching for unavailable or unaffordable new process
technologies. Application of good PHA techniques often reveals
opportunities for continuous improvement of existing processes and
operations without a separate analysis of alternatives. EPA encourages
owners or operators to continue to examine and adopt viable alternative
processing technologies, system safeguards, or process modifications to
make new and existing processes and operations inherently safer.
Through the process and prevention program information in the RMP,
sources can demonstrate, and users of the RMP information can observe
and promote, progress toward safer processes and operations.
EPA is considering the development of incentives and awards to
stimulate inherently safer alternative research and development, public
outreach and education, and risk communication efforts. The Agency
welcomes ideas and participation in this effort.

III. Discussion of Comments

EPA received 1220 comments, including 180 relevant comments
submitted for the List Rule, 757 comments on the NPRM, and 283 comments
on the SNPRM. The commenters represented 92 chemical manufacturers, 81
other chemical users, 111 petroleum industry companies, 174 industry
trade associations, 40 other trade associations, 58 agricultural supply
retailers, 102 propane retailers, 132 explosives users, 29 water
treatment facilities, 26 utilities, 66 state agencies, 63 local
governments, 8 other Federal agencies, 52 academics and consultants, 61
environmental groups, 6 labor unions, and 31 private citizens. The
remaining 88 letters were requests for extensions of the comment
period, interim or duplicate sets of comments, or had been sent to the
incorrect docket. The major issues raised by the commenters are briefly
addressed below; a complete presentation of the Agency's response to
the comments received on this rulemaking is available in the Risk
Management Program Rule: Summary and Response to Comments in the docket
(see ADDRESSES).

Many commenters requested that EPA's list be identical to OSHA's
list of highly hazardous substances and no thresholds should be less
than OSHA's. These comments were addressed in the final list rule (59
FR 4478; January 21, 1994) and background material related to these
issues is available in docket number A-91-74 (see ADDRESSES).

A. Tiering

Commenters on the NPRM suggested that EPA create different levels
of requirements for sources that pose different risks. In the SNPRM,
EPA proposed three tiers: a low hazard tier for sources whose worstcase
release would not affect any public or environmental receptors of
concern; a medium hazard tier for sources that were not eligible or
covered by the low or high hazard tiers; and a high hazard tier based
on either industry sector accident history and number of employees or
simply based on the number of employees. Generally, commenters were
concerned that all processes at a source would need to be eligible for
Program 1 before any process could be. EPA has revised the rule to
clarify that eligibility for any tier

[[Page 31675]]

(Program) is based on process criteria, not source. If a process meets
Program 1 criteria, the owners or operators need only meet Program 1
requirements for that process even if other processes at the source are
subject to Program 2 or Program 3.

Rationale. Only 2 of the 57 commenters opposed tiering arguing
that the CAA mandates that all covered sources be required to complete
a full prevention program and that Congress had considered and rejected
exemptions. One commenter argued that EPA had already accounted for
``differences in size, operations, processes, class and categories of
sources'' in developing the list and thresholds. Most commenters
supported tiering as an appropriate way to recognize different levels
of risks and to allow sources and emergency responders to focus on the
highest risk processes.

EPA disagrees that the CAA requires all covered processes to comply
with the same detailed risk management program. EPA listed regulated
substances because of their inherent hazards, such as toxicity and
volatility. EPA did not consider, nor does the CAA indicate that it may
consider, ``differences in size, operations, processes, class and
categories of sources'' in selecting chemicals or setting thresholds.
In establishing section 112(r)(7) requirements, however, Congress
clearly recognized that a ``one-size-fits-all'' approach may not be
appropriate for these regulations and directed EPA to consider these
factors in the development of the accident prevention regulations.
Furthermore, EPA strongly disputes the assertion that it has exempted
any source from regulation by creating different programs for different
sources. As noted below, all covered processes will be addressed in
RMPs that contain hazard assessment, prevention, and response
information, as required by statute.

Program 1 vs. Program 2 and Program 3 Criteria. Commenters
generally supported Program 1 for low-risk sources, but argued that
few, if any, sources would qualify because the requirements were too
stringent.

a. Potential for Offsite Impact. Commenters generally agreed that
sources that can demonstrate no offsite impact should be eligible for
Program 1, but only public health should be considered, not
environmental impacts. Others stated that only sources posing a threat
of ``considerable'' impacts should not be eligible for Program 1. One
commenter stated that EPA's worst-case scenario is unrealistic and its
use as a Program 1 trigger is unreasonable. Other commenters want EPA
to allow site-specific modeling for the offsite consequence analysis,
rather than look-up tables.

In today's rule, EPA specifically allows owners or operators to use
site-specific air dispersion modeling for their offsite consequence
analyses. EPA disagrees that offsite impacts should be limited to
``considerable'' impacts. When offsite impacts are possible, it may be
reasonable to implement some additional measures to reduce accidental
releases, especially when the burden of measures such as additional
training or safety precautions is low. Programs 2 and 3 provide
flexibility to allow source-specific consideration of the appropriate
level of effort. Program 1 requires no additional prevention measures,
which is only categorically justifiable if such measures would not
reduce offsite impact. It is reasonable to couple a no impact criterion
with a conservative worst-case scenario to conclude categorically the
public would not benefit from additional prevention measures. If no
impact can be demonstrated for a conservative worst-case release, then
no impact is likely to occur for any other release event, and the
process could be judged to pose a low threat to the surrounding area.
EPA has decided that potential impact on environmental receptors
resulting from a worst-case scenario will not be a criterion to
determine eligibility for Program 1. EPA agrees that very little, if
any, data exist on the potential acute environmental impacts or
environmental endpoints associated with listed chemicals upon
accidental release. In addition, the offsite consequence distances
estimated using human acute toxicity or overpressure effects may not be
directly relevant to environmental effects. However, owners or
operators will be required to document in the RMP the presence of such
receptors within the distance determined for the worst case. EPA
believes that natural resource agencies and the public will be able to
benefit from the environmental receptors information in the RMP in
discussions with the source.

b. Accident History for Program 1. Many commenters objected to
accident history as a Program 1 criterion, arguing that a process that
had a significant accidental release in the previous five years may
have been changed to reduce or eliminate future events and public
impact. Several commenters suggested that such processes that otherwise
meet Program 1 criteria should remain eligible, but be required to
justify and document the changes. Some commenters also objected to
EPA's proposed definition of significant accidental release, arguing
that many companies and emergency responders conservatively evacuate or
shelter-in-place during minor incidents. Under the proposed definition,
these actions disqualify a process from Program 1 even if there were no
offsite impacts. Some commenters stated that the accident history
provision was unnecessary because, by definition, a Program 1 process
is not capable of an accidental release that could affect public
receptors.

EPA has decided to retain the accident history criterion for
Program 1 processes, excluding events with evacuations and shelterings
in place, and to drop the definition of significant accidental release.
Program 1 eligibility is not a one-time exercise; owners or operators
must certify in each RMP that no qualifying releases have occurred
since the previous RMP submission and provide current worst-case
release data indicating no offsite impacts are anticipated in the
future. Program 1 criteria and accident history provide owners or
operators an opportunity to demonstrate to the community ongoing
excellence in accident prevention and an incentive to search for and
implement ways, such as inventory reduction, to reduce the potential
for offsite impacts associated with large scale accidental releases.
Further, the unique circumstances surrounding past accidents can
provide a reality check on the theoretical modeling and worst-case
scenario claims used for the offsite consequence assessment and serve
to verify that administrative controls and passive mitigation measures
work as intended. EPA decided to delete public evacuations or
shelterings-in-place as criteria for Program 1 eligibility. EPA is that
inclusion of these criteria in Program 1 eligibility may create a
perverse incentive not to report releases and it may encourage sources
and local emergency officials to take more chances during an event when
there may be potential exposures that do not rise to the endpoint
specified in this rule but would otherwise be worthy of precautionary
actions by the source or by local officials. If the evacuation or
sheltering takes place because of a concern for public exposure to an
endpoint as specified in this rule, then public receptors necessarily
would be under the worst case distance and the process would not be
eligible for Program 1 under the criteria of the rule. Owners or
operators of processes that meet Program 1 eligibility requirements are
required to report a 5 year accident history for that process. If local

[[Page 31676]]

emergency planners, first responders or the public have concerns about
processes in Program 1 because of a past evacuation or sheltering-inplace
event, then mechanisms under EPCRA could be used to gather more
information from the source about its prevention program (such as EPCRA
sections 302(b)(2) [designation of a facility if it does not already
handle extremely hazardous substances listed under section 302] and
303(d)(3) [provision of information to the emergency planning
committee]) and involve the source in emergency planning. Sources and
local first responders should be discussing evacuation and shelteringin
-place criteria and decisions as part of emergency response planning.
c. Other. Many commenters asked that specific industries such as
ammonia refrigeration, retail fertilizer outlets, all flammables, and
all non-PSM sources be assigned to Program 1. EPA disagrees because
each source has unique surroundings that must be considered in the
worst-case assessment and each source must demonstrate favorable
accident history. All ammonia refrigeration units covered by this rule
are already subject to OSHA PSM; many of these have had accidents that
affected the community and should be required to complete the
requirements of the hazard assessment and emergency response program
and provide the community with full RMP information. According to the
industry, a typical ammonia fertilizer retailer handles 200 tons of
ammonia. Some retailers may be very geographically isolated and can
qualify for Program 1, but EPA expects that most will be subject to
Program 2. Given the large quantity of ammonia involved, EPA considers
it important that the community have information on offsite
consequences from these sources and that the owner or operator takes
the necessary steps to address accidental release prevention and
emergency response.

EPA expects that some sources handling flammables will qualify for
Program 1 because the distance to a 1 psi overpressure is generally
less than distances to toxic endpoints. Nonetheless, those sources
handling flammables in sufficient quantity to generate a potential
offsite impact should provide the community with information on hazards
and address prevention and response steps. Many sources handling
flammables are already subject to PSM; the only additional steps
required under this rule are completion of the hazard assessment and
emergency response programs and submission of an RMP.
EPA does not agree that non-PSM sources should be assigned to
Program 1. Many of these sources could have an accidental release that
can affect the community. OSHA exempted retailers because they are
covered by other OSHA or state regulations that address workplace
safety, not because they are incapable of having offsite impacts. All
retailers are in Program 2 unless they can meet Program 1 criteria;
thus, they should be taking prevention steps and will be providing the
community with information. Compliance with other existing Federal and
state programs may satisfy many Program 2 prevention requirements,
thereby limiting the burden. In addition, EPA expects to develop model
risk management programs for these sectors. Public sources in states
without delegated OSHA programs are not covered by OSHA PSM because
OSHA is barred by law from regulating them. Nonetheless, these sources
may pose a threat to the community. Today's rule places these sources
in Program 2.

3. Program 2 vs. Program 3 Criteria. In the SNPRM, EPA's preferred
approach assigned sources to Program 3 based on SIC code and number of
employees; sources in specified SIC codes with 100 or more full-time
employees (FTE) would have been subject to the full program in 3 years;
sources in a subset of these SIC codes with 20 to 99 FTEs would have
been subject to the full program in 8 years. The alternative was to
impose the full program on all sources with more than 100 FTEs. Most
SNPRM commenters submitted suggestions and arguments about this
approach.

a. Number of Employees. Only two commenters supported using the
number of employees as the sole criterion, arguing it would be the
easiest approach to implement with the greatest amount of industry
participation. Commenters opposed it because the number of employees
proposed does not reliably correlate with risk, hazard, or quantity on
site, and because it could act as an incentive to reduce employment. In
addition, some commenters stated that smaller sources may have fewer
resources to manage hazards and, therefore, may pose a greater risk to
the public.

EPA agrees and has deleted the number of employees as a Program 3
criterion. Although size of a source in the manufacturing sectors may
be related to the quantities on site and complexity of the processes,
many other sources may have similar characteristics with fewer
employees. Complexity is more directly associated with the type of
industry (i.e., SIC code) than with number of employees; a highly
automated process may involve fewer employees and be more complex than
a more labor intensive process. Quantity, if relevant, can be directly
measured rather than indirectly by number of employees. In addition,
EPA was concerned that the data on which the Agency based its proposed
approach may not be representative of all accidental releases. These
data, drawn from reports to the National Response Center and EPA
regions, appear to indicate that larger sources have more and larger
accidental releases than do smaller sources. This finding, however, may
in part reflect different levels of reporting, rather than different
levels of accidents. Both Federal and state officials report that the
number of releases has risen in recent years as more sources learn
about their reporting obligations. EPA has decided that, because the
processes within the SIC codes basically handle the same chemicals in
the same way, smaller sources should not be moved to a different
Program based on the number of employees.
b. SIC Code. Fifty-seven commenters, particularly those in the oil
industry, utilities, and public systems, supported the use of SIC codes
based on accident history; 28 commenters opposed it. Supporters argued
that industry accident records represented a reasonable criterion for
identifying high-risk sources. If an entire industry has a long history
without accidental release, it may indicate that the materials handled
and handling conditions generate a smaller potential for serious
releases or that the industry is effectively controlled by government
or industry standards. Some commenters argued that industry accident
histories reflect underlying risk better than individual source
accident histories because accidents are rare events; a source with no
accidental releases over the previous five years is not necessarily
safe.

Commenters opposing the use of SIC codes stated that the approach
is arbitrary, that accidents with only onsite effects should not be
used, that sources in other industry sectors handle similar quantities
and pose similar risks, and that sources within an industry that have
successful risk management practices are penalized by a few isolated
sources within the industry.

EPA has decided to retain the use of SIC codes, adding SIC 2865
based on further review of accident histories, and to add coverage by
the OSHA PSM standard as a separate criterion for Program 3. EPA
selected the SIC codes by analyzing accident data filed by

[[Page 31677]]

sources in response to EPA's request for information in the Accidental
Release Information Program (ARIP). ARIP collects data from certain
sources that report releases under CERCLA section 103. EPA selected the
SIC codes that showed a high frequency of the most serious accidents
across a significant percentage of all sources within the SIC code to
avoid mischaracterizing an industry based on isolated, problematic
sources. Data on the selection criteria were summarized in the SNPRM
and the docket at the time of the SNPRM. The accident history of the
cyclic crudes industry (SIC code 2865) is similar to that of the
categories selected. EPA disagrees that only offsite impacts should be
considered; accidental releases that caused death, hospitalizations, or
injuries on site are also of concern because they indicate significant
safety problems that could lead to releases that cause impacts offsite.
The SIC codes selected by EPA are basically the same ones OSHA selected
for its PSM program inspection focus. EPA disagrees that sources are
``penalized'' by this approach because owners or operators of processes
in these SIC codes have an opportunity to present their safety record,
demonstrate the success of their accident prevention programs, and
communicate with the local community the basis for their risk
management practices. Sources that receive Merit or Star status in the
OSHA Voluntary Protection Program will be favorably distinguished from
others in the same industry when implementing agencies are selecting
sources for audits (see section III.T.1 below).
EPA agrees that serious accidents occur infrequently even at
sources with poor safety practices and that industry-wide accident
records provide a better mechanism than the accident history at a
single source for identifying those sectors whose chemicals and
processes may lead to serious releases. A high proportion of the
sources in some SIC codes reported releases; EPA's analysis
specifically took into account the number of reports from individual
sources to avoid selecting an SIC code because of a small number of
sources with serious safety problems.

The OSHA PSM already applies to most covered processes in the
selected SIC codes. EPA expects that there will be fewer than 400
additional processes assigned to Program 3 that are not already subject
to the OSHA PSM standard at the approximately 1,400 sources in these
SIC codes and that all of these sources will already have other
processes covered by OSHA PSM. Consequently, fulfilling the RMP
requirements imposes little additional burden.
EPA decided to include all covered processes currently subject to
the OSHA PSM standard in Program 3 to eliminate any confusion and
inconsistency between the prevention requirements that the owners or
operators of such processes must meet. EPA's Program 3 prevention
program is identical to the OSHA PSM standard. Including OSHA PSM
processes in Program 3, therefore, imposes no additional burden on
these processes; the only new requirements for such processes are the
hazard assessment, emergency response program, and the RMP, which are
the same under Programs 2 and 3.

c. Site-Specific, Risk-based Criteria. Many commenters stated that
Program assignment should be based on site-specific risk-based
criteria. Accident history is one such criterion and is discussed
separately in Section III.A.3.d. Other criteria suggested include
population density or proximity, quantity on site, number of substances
held above the threshold, process conditions, toxicity, volatility,
alternative release scenario results, or combinations of these factors
as a risk index.

EPA agrees with commenters that Program assignments should be riskbased
to the extent possible; however, as the variety of suggestions
indicates, a considerable number of variables would need to be
considered. EPA knows of no standard approach or equation that is used
and generally accepted. The variety of suggestions indicate the
likelihood that any proposed formula would meet opposition. No
commenter provided a method to comprehensively address these factors on
a nation-wide basis.

An important consideration for EPA in developing the rule
provisions for Program assignment was to avoid undue complexity,
confusion, and resource expenditure by sources and implementing
agencies implementing the rule's criteria. To some extent, EPA has
incorporated risk factors, including site-specific factors, in
determining which sources are eligible for which Program. For example,
Program 1 eligibility already considers the potential for offsite
impacts; any process for which there are no public receptors within the
distance to an endpoint from a worst-case release may be eligible for
Program 1, provided there have been no releases with certain offsite
consequences within the previous five years. Today's rule allows
sources to consider passive mitigation and administrative controls in
conducting the worst-case release analysis. Such site-specific
considerations affect the extent of potential exposure to a worst-case
release, and thus are reflected in the Program 1 eligibility criteria.
Elements of risk such as process complexity and accident history are
also reflected the design of Program 2 and Program 3 requirements and
the assignment of processes to these Programs. Program 2 sources
generally handle and store regulated substances, but do not react or
manufacture them. EPA believes Program 2 sources can take prevention
steps that are less detailed than those in the OSHA PSM standard and
still accomplish accident prevention that is protective of any
population nearby. Program 3 is reserved for processes already subject
to the OSHA PSM standard and processes with high accidental release
histories. The SIC codes with an accident history selected by EPA for
Program 3 are typically complex processes. The PSM standard was
designed for, and is particularly appropriate for, these processes.
EPA takes issue with the appropriateness of some of the suggested
factors. Meteorological conditions vary too much to be considered in
determining a risk level. Chemical quantity alone does not accurately
relate to risk because the location and handling conditions can
dramatically change the potential for exposures.
In addition, EPA has implementation concerns about a detailed,
national, multi-factor, risk-based approach, were it to be feasible.
States such as Delaware have used a simple version of a risk-based
approach and found that it created serious problems for the state and
the sources. Smaller sources and those without technical staff have had
great difficulty in implementing the approach and have had to rely on
state officials to determine applicability for them. Delaware
specifically recommended that EPA not attempt implementing a similar
approach on a national basis because of the burden it imposes on the
state and the confusion and uncertainty it creates for sources.
Delaware has fewer than 100 sources; nationally, EPA estimates that
66,000 sources will be subject to the rule, approximately 62,000 of
which are outside of the chemical and refining sectors. If implementing
agencies had to help most of these sources determine the index score
and Program for each process, not only would the burden on the agencies
be extreme, but implementation would also be delayed. Furthermore, were
EPA to simply identify risk factors without an index and leave the
determination of Program

[[Page 31678]]

level to sources or implementing agencies, the process for such sitespecific
determinations would be even more complex and resource
intensive for sources and implementing agencies; it would create
disincentives for a state to become involved and to take on the role of
an implementing agency. EPA believes it is better to have sources and
agencies focus their resources on prevention activities.
EPA considered, but decided against, a less comprehensive riskbased
approach using proximity or population density as criteria for
distinguishing between Program 2 and 3. EPA recognizes that accidental
releases from sources near or in densely populated areas may harm more
individuals and be perceived to pose a greater risk than other sources.
However, as stated above, EPA believes that the type of process, its
complexity and accident history should be considered for Program 2 or 3
assignment, regardless of the number of people potentially exposed. In
other words, EPA does not believe the streamlined Program 2 prevention
elements should apply to a complex Program 3 process just because fewer
persons could be potentially exposed or that the Program 3 prevention
elements should apply to a Program 2 process because more people could
be potentially exposed. EPA believes that populations offsite should be
protected from harm based on the type of process; the Program 2
prevention elements, properly applied to the expected types of Program
2 processes, serves to protect off-site populations, just as the
Program 3 prevention elements for complex processes serves to protect
offsite populations.

If Program assignments were based on the alternative release
scenario results, sources would not have the flexibility and latitude
in today's rule for these scenarios because more definite criteria
would need to be considered to ensure the proper scenarios and results
are assessed. This places more emphasis and burden for sources on the
offsite consequence assessment rather than on accident prevention and
communication with the public and first responders. Furthermore,
because active mitigation includes process and control equipment that
may fail, considering such equipment in evaluating risk would not be
appropriate without detailed review by the source and oversight by the
implementing agency.

Some commenters suggested yet another variation of a less
comprehensive, ``risk''-based approach that would have EPA use a sitespecific
analysis of likelihood of release to assign Program levels.
Many of the same difficulties in developing a ``risk index'' for
determining Program assignments would apply to an attempt to
incorporate likelihood in a more sophisticated manner than EPA was able
to do in its analysis of accident history by SIC code. In addition to
the substance-specific properties considered as part of the chemical
listing criteria, the site-specific likelihood of a release depends on
a number of factors, including the appropriateness of the equipment in
use, the maintenance of that equipment, operator performance, and
safety systems and their performance. Evaluating site-specific
likelihood of release requires data on each of these items; such data
rarely exist especially for complex processes where a variety of
equipment must be evaluated along with the performance of multiple
operators and maintenance workers. Using surrogate data (e.g.,
manufacturer's failure rate data) introduces error of an unknown
magnitude to the analysis. Such analyses are very costly and produce
results that are, at best, questionable.
EPA also believes that assessing the likelihood of a release at
most sites for site-specific individualized Program-level
determinations is neither technically feasible nor cost-effective. In
most cases, the data do not exist to conduct a meaningful analysis;
where they do exist, the cost of developing a defensible analysis and
overseeing it could well exceed the cost of compliance with the rule.
Such an approach would resemble a permit program, which would be
resource-intensive for sources and implementing agencies. EPA
determined that the simpler approach for assigning sources to Program 1
would provide regulatory relief for those sources that could not affect
the public while allowing other sources to devote their resources to
prevention activities rather than to analyses that would be subject to
legal challenges.

EPA notes that sources have the flexibility to implement
appropriate accident prevention measures based on the hazards and risks
discovered in the hazard review or process hazard analysis. The
structure of Programs 2 and 3, therefore, reflect site-specific risk
criteria. Further, the purpose of the risk management program and RMP
effort is to prevent accidents and facilitate local level dialogue
about the risks, prevention measures, and emergency response effort in
place at the source. The local community and first responders may have
far different concerns that should, and can be addressed better through
today's approach than those reflected by a risk index approach.
d. Accident History. Some commenters argued that EPA should assign
sources to Program 3 based on the accident history of the source. One
commenter suggested that any source with no accidental release that
exceeded a reportable quantity (as defined in CERCLA) for the previous
five years should be in Program 2. Others argued that a source should
be in Program 2 if it had no significant accidental release in the
previous five years. Some commenters said that a one-release standard
was too stringent and that two or more significant accidental releases
should be allowed before a source was assigned to Program 3. Another
commenter suggested that a source with no significant accidental
releases in the past five years and with few potentially impacted
neighbors should be placed in Program 2.
Other commenters opposed this approach, arguing that, in many
cases, sources take steps to prevent recurrences following a serious
release. In some cases, the offsite impacts from releases are minor and
would not justify assigning a source to a particular Program. Other
commenters stated that the absence of an accidental release can be
indicative of lower risk, but it can also simply mean that a release
has not yet occurred. Several commenters noted that a five-year time
period is statistically insignificant because accidental releases are
infrequent events.

EPA agrees that source-specific accident history is not a
reasonable basis for assigning processes to Programs 2 and 3. Given the
relative infrequency of serious accidents, a five- or even ten-year
period without an accident may not be indicative of safe operations. In
addition, the criteria necessary to define the types of past accidental
release for the purposes of program classification would need to be
based on a wide variety of variables and site-specific factors, which
would lead to confusion and unnecessary complexity. Factors such as
weather conditions at the time of the release, rather than the size of
a source or its management practices, often determine whether a release
has offsite consequences. EPA believes that accident history is
appropriately used on an industry-wide basis as described above for
selection of Program 3 sources. If accidental releases with
consequences appear to occur at a large proportion of sources within an
SIC code, where similar processes, equipment and chemicals are used,
then it is reasonable to conclude that

[[Page 31679]]

processes in that SIC code pose a greater likelihood of a high hazard
release than others. This approach removes the need for at least one
accident to occur at every source that EPA believes ought to be
assigned to a particular Program, especially when such accidents are
rare events. EPA is also concerned that using source-specific accident
history as a criterion would create an incentive for sources to fail to
report releases. Finally, as EPA has stated, assignments to Program 2
and 3 also consider the appropriateness of the prevention steps for the
types of sources. EPA believes that both Programs move sources to
greater accident prevention.

e. Other. Some commenters asked that the implementing agency be
given discretion to move a source into a different Program based on
local concerns and knowledge. EPA notes that states have the authority,
under the CAA, to impose more, but not less, stringent standards than
EPA (see CAA section 112(r)(11)).

A few commenters suggested that Program 2 be limited to sources for
which a model risk management program had been developed. The models
would be designed to reflect risks associated with categories of
sources that all use the same type of equipment and handle the
substances in the same way (e.g., propane retailers and users, ammonia
retailers). EPA considered this approach and decided that the Program 2
prevention program provides a better, generic prevention approach for
processes for which the more detailed PSM program would be
inappropriate. Limiting Program 2 to those industrial sectors where
industry-specific models are feasible would place some manufacturing
sources at a disadvantage simply because their chemical uses,
processes, and equipment were too varied to allow development of a
model or because there are too few sources to justify use of EPA or
industry resources to develop a model. In addition, if EPA were to
limit Program 2 to sources with model programs, Program 2 regulations
would need sufficient specificity to enforce the use of these models;
otherwise, sources would be able to ignore both PSM and the models. EPA
is also concerned that codifying the model plans could stifle
innovation in safety practices. If industry codes or other Federal
regulations on which parts of the models may be based were updated, EPA
would have to revise its models; given the time needed to propose and
adopt regulations, sources might have to delay implementation of new
systems and, in some cases, might be caught between complying with a
revised EPA or OSHA regulation or state law or complying with the
model. Consequently, EPA decided it was better to have models available
as guidance, but not require compliance with them. Further, EPA
believes that the key elements of good accident prevention practices
are captured within the requirements of the Program 2 prevention
program. Model programs and plans are likely to build on these
approaches, making it easier for sources in Program 2 to use models
that are later developed by others.

EPA is working with industry to develop model risk management
programs and RMPs for ammonia refrigeration systems, propane
distributors and users, and water treatment systems. EPA also expects
to develop models for ammonia retailers and wastewater treatment
systems. EPA encourages other industrial sectors to work together on
additional model development.

4. Program 1 Requirements. Commenters were generally opposed to
posting signs, and certification of no environmental impact.
a. Certification of No Environmental Impact. Many commenters stated
that it would be ``virtually impossible'' to certify ``no potential for
environmental impacts,'' as required by the SNPRM. Commenters said that
the definition of environmental impact was too vague, that the list of
environments suggested in the SNPRM was too broad, and that the
language seemed to require a full environmental consequence assessment,
making the requirement impossible. One commenter noted that companies
would find it difficult to assert that there could be ``no
environmental impacts'' even after an environmental consequence
assessment reveals insignificant impacts. Two commenters suggested that
EPA substitute ``low potential for environmental impact'' or ``no
potential for long-term, adverse environmental impact.'' Other
commenters requested that environmental impact be dropped or that the
requirement be changed to mirror the Program 1 eligibility criteria
with an indication in the RMP that no environmental receptors of
concern were within the worst-case distance to an endpoint.
As described above in section III.A.2.a. Potential for Offsite
Impact, EPA has decided not to make the presence of environmental
receptors a part of the eligibility criteria for Program 1 and has
deleted the certification requirement. Instead, owners or operators of
all covered processes will have to identify in the RMP any
environmental receptors that are within the distance potentially
affected by the worst case.

b. Signs. Commenters generally opposed the SNPRM requirement that
sources with Program 1 processes post signs warning of the hazards on
site if the only regulated substances present at the site above the
threshold quantity were listed for flammability. Commenters stated that
local and state fire and safety codes often already require such signs.
In addition, sources are already required under EPCRA section 312 to
file annual inventories with the LEPC and fire department that identify
hazards on site. Signs would have fulfilled the emergency response
program requirements for a source. Because Program 1 eligibility will
now be determined on a by-process basis rather than by source-wide
criteria and because EPA has revised the emergency response program
provisions as noted below, EPA has dropped the requirement for signs.
c. Emergency Response Program. In the SNPRM, EPA asked whether
additional emergency response planning and coordination should be
required for Program 1 processes. Some commenters supported this
requirement, while others stated that most sources are already covered
by EPCRA and participate in community response planning. Commenters
stated that because the worst-case release could not reach public
receptors, such efforts were not necessary.
In the final rule, EPA is requiring the owner or operator of a
Program 1 process to ensure that any necessary response actions have
been coordinated with local response agencies. EPA believes that local
responders may become involved in an incident, even if the public is
not threatened. No additional CAA-related planning activities are
required, however.

d. Other. Many commenters stated that, since Program 1 processes
generate no offsite impact, they should be exempt from this rule. One
commenter objected to Program 1 because members of the public,
particularly first responders and business visitors, could still be
hurt by a release. Other commenters suggested that the annual EPCRA
section 312 form could be amended to indicate that a source was covered
by the rule, replacing the RMP registration form.
The CAA requires that all sources with more than a threshold
quantity of a listed substance register an RMP, perform a hazard
assessment, and develop accidental release prevention and emergency
response programs. Therefore, total exemption of processes that meet
Program 1 criteria is not

[[Page 31680]]

possible. See S. Rep. No. 228, 101st Cong., 1st session, at 208
(``Senate Report'') (precursor of RMP provision mandating hazard
assessments for sources that exceed threshold for listed substance);
136 Congressional Record S16927 (daily ed. October 27, 1990) (remarks
of Sen. Durenburger, sources with more than a threshold quantity are
subject to regulations); 136 Cong. Rec. H12879 (daily ed. Oct. 26,
1990)(remarks of Rep. Barton)(all users of hazardous chemicals are
required to plan for accidents). Moreover, even if an exemption for
processes that exceed a threshold were permissible, the owner or
operator would need to take steps that are equivalent to the hazard
assessment to establish eligibility for the exemption. The offsite
consequence analysis is the most significant burden for a Program 1
process under this rule. The minimal additional actions required in
today's rule for Program 1 simply establish a record of eligibility and
a response coordination mechanism.

EPA recognizes that emergency responders and site visitors could be
hurt by an accidental release from any process, but notes that
responder safety is covered by OSHA and EPA under the HAZWOPER
regulations. It is the owners' or operators' responsibility to inform
visitors about the hazards and the appropriate steps to take in the
event of an accidental release from any process subject to today's
rule.

Finally, EPA has based the registration information requirements in
today's rule on the EPCRA section 312 Tier II form. The CAA requires
that the RMP be registered with EPA. Because the EPCRA form is not
submitted to EPA, it would not substitute for registration with EPA
either in its present or amended form. Completion of the registration
portion of the RMP should impose little additional burden on owners or
operators. However, EPA recognizes the information overlap between the
Tier II form and the RMP registration and is considering use of the RMP
registration for the Tier II reporting requirement.
5. Program 2 Requirements. Commenters were generally concerned
about the lack of specific requirements for the Program 2 streamlined
prevention program and emergency response requirements, and how
compliance with other regulations would be incorporated.
a. Streamlined Program. Commenters stated that the Program 2
prevention program does not provide much, if any, regulatory relief
because sources would need to address most of the ten elements of the
Program 3 prevention program. Others said that the majority of the
sources affected by the rule are already covered by OSHA PSM and
chemical industry standards, the Program 2 requirements do not satisfy
the CAA mandate, and that only a full process hazard analysis would
meet the hazard assessment requirements under section 112(r). Another
commenter argued that EPA's statement that sources must comply with the
CAA's general duty clause was inadequate because EPA has not used, and
has no policy about, the clause.

EPA agrees that the preferred approach in the SNPRM did not provide
sufficient detail on Program 2 prevention requirements to distinguish
it from Program 3. EPA solicited comments on whether Program 2 should
require additional, specific prevention steps. Today's rule provides
specific requirements as discussed in section I.D above and in Section
IV below. In the RMP, the owner or operator will be required to report
on other Federal or state regulations, industry codes, and standards
used to comply with prevention elements as well as any major hazards,
process controls, mitigation systems, monitoring and detection systems
examined in the hazard review. This streamlined prevention program
addresses many of the PSM elements as the basis for sound prevention
practices, but is tailored to processes with less complex chemical
uses; this program provides considerable regulatory relief by
substantially reducing the documentation and recordkeeping burden of
PSM. In addition, EPA will provide guidance and model risk management
programs to further assist Program 2 processes in developing and
maintaining good prevention program practices.
EPA disagrees that only a full PHA would meet the requirements of
the Act. Section 112(r) does not contain detailed requirements for the
hazard assessment, beyond the key components of accidental release
scenarios and a five-year accident history. EPA believes that a PHA is
more appropriately considered an element of a prevention program, such
as PSM. The statute does not mandate detailed PHA engineering analyses
for all sources, whether as part of the hazard assessment or the
prevention program. EPA believes PHAs involve a more detailed
engineering analysis than is necessary to prevent accidents at Program
2 sources. The ``hazard review'' provisions of Program 2 should be
sufficient to detect process hazards at these simpler processes. EPA
recognizes that although hazard assessments and PHAs or process hazard
reviews are discreet elements that can be performed independently,
hazard assessment results can enhance PHA or process hazards reviews
and in turn, the results of the PHA or review can enhance the hazard
assessment. EPA encourages owners or operators to make maximum use of
the PHA or review and hazard assessment information to manage risks and
prevent accidents.

Finally, sources with Program 2 requirements, as well as sources
with Program 1 or 3 requirements, must comply with the general duty
clause of CAA Section 112(r)(1). The general duty clause provides that
owners and operators have a general duty to identify hazards that may
result from accidental releases, design and maintain a safe facility,
and minimize the consequences of any releases that occur. The general
duty clause is a self-executing statutory requirement: it requires no
regulations or other EPA action to take effect. The clause provides a
separate statutory mechanism that EPA will use in appropriate
circumstances to ensure the protection of public health and the
environment. To date, EPA has undertaken several inspections designed
in part to determine compliance with Section 112(r)(1). As appropriate
at a future date, EPA may issue policies or guidance on application of
the general duty clause.

b. Other Regulations. Commenters generally agree that OSHA PSM,
HAZWOPER, the OSHA hazard communication standard (29 CFR 1910.1200),
and NFPA-58 are examples of other regulations or voluntary industry
standards that could be cited to meet the requirements of a Program 2
prevention program. Commenters requested that EPA provide a matrix or
crosswalk that indicates which other regulations, standards, and codes
met specific requirements. One commenter opposed the use of other
regulations or referencing of voluntary industry standards, stating
that, other than OSHA PSM, no other OSHA standard addresses safety
precautions or maintenance. Another commenter objected that this
approach creates another documentation burden without any commensurate
benefit.

EPA agrees that the SNPRM preferred approach for Program 2 was not
specific enough and has provided more detailed requirements in this
rule as noted above. EPA continues to believe that many of the Program
2 prevention requirements are already met through industry compliance
with existing regulations and voluntary standards. For example, ammonia
retailers whose processes are designed to meet the OSHA ammonia
handling rule (29 CFR

[[Page 31681]]

1910.111) should be able to meet the Program 2 requirement that the
process design meets good engineering practices. This effectively
allows sources to cite compliance with these other regulations and
standards instead of developing specific, duplicative elements solely
to comply with Program 2. EPA will also use these existing regulations
and standards as it develops model programs.
c. Emergency Response Program. Commenters supported considering
HAZWOPER programs as adequate to meet the Program 2 emergency response
program. A few commenters said that HAZWOPER is inadequate because it
does not consider offsite impacts or the environment. Some commenters
also said that coverage of a source by an EPCRA community emergency
response plan should be sufficient. Others said that any contingency
plan developed under Federal or state law should be considered
sufficient because the requirements under these programs are generally
consistent with EPA's proposed emergency response program; one
commenter noted that, for flammable processes, compliance with 29 CFR
1910.38 should be adequate because the response is usually evacuation
of employees. Five commenters opposed any requirement that sources with
Program 2 processes conduct drills or exercises because they represent
lower hazards.

Consistent with its efforts to consolidate Federal emergency
planning requirements, EPA has included language in the final rule that
will allow any source in compliance with another Federal emergency
response program that includes the elements specified in this rule to
use that program to meet these requirements. In particular, this
applies to response plans prepared in accordance with the National
Response Team's Integrated Contingency Plan Guidance (``one plan'')
(NRT, May 1996). EPA believes that sources should have a single
response plan; creation of multiple response plans to meet slightly
different Federal or state standards is counterproductive, diverting
resources that could be used to develop better response capabilities.
EPA recognizes that some sources will only evacuate their employees
in the event of a release. For these sources, EPA will not require the
development of emergency response plans, provided that appropriate
responses to their hazards have been discussed in the community
emergency response plan developed under 42 U.S.C. 11003 for toxics or
coordinated with the local fire department for flammables.

B. Offsite Consequence Analysis

Worst-Case Release Scenario. EPA proposed in the NPRM to define
the worst-case release as the ``loss of all of the regulated substance
from the process * * * that leads to the worst offsite consequences''
and that the scenario should assume ``instantaneous release.'' Hundreds
of commenters stated that instantaneous loss of the total process
contents is not technically feasible for complex systems and,
therefore, represents a non-credible worst case that would provide no
useful information to the public or the source for risk communication,
accident prevention, and emergency preparedness. Many commenters also
argued that this approach differed from the release modeling
assumptions contained in EPA's Technical Guidance for Hazards Analysis,
which has been the basis for community emergency planning activities
under EPCRA. Although some commenters were generally opposed to the
concept of worst case, most of the commenters were supportive of an
approach similar to that taken in the Technical Guidance.
In response to these comments, EPA proposed in the SNPRM to
redefine a worst-case scenario as the release, over a 10-minute period,
of the largest quantity of a regulated substance resulting from a
vessel or process piping failure. The 10-minute release time is drawn
from the Technical Guidance for Hazards Analysis. EPA believes this
duration is reasonable and accounts for comments arguing that an
``instantaneous'' release is unrealistic for large-scale releases.
EPA has decided to adopt the SNPRM approach for worst-case toxic
vapor releases in the final rule because most of the SNPRM comments
agreed that the redefinition is generally more credible and that the
10-minute time frame particularly applies to vapor releases. Although
some commenters argued that this approach still does not account for
all process-specific conditions, EPA believes it is reasonable and
representative of accident history. EPA notes that owners or operators
may use air dispersion modeling techniques that better account for
site-specific conditions, provided modeling parameters as specified in
the rule are applied. This release scenario will apply to substances
that are gases at ambient conditions, including those liquefied under
pressure. Gases liquefied by refrigeration only may be analyzed as
liquids if the spill would be contained by passive mitigation systems
to a depth greater than 1 cm.

Under the SNPRM, worst-case liquid spills were assumed to form a
pool in 10 minutes, with the release rate to the air determined by
volatilization rate. EPA recognized that this approach differs from the
use of an instantaneous release in the Technical Guidance, which EPA
cited as an alternative to its favored approach. The few comments
received were divided between support of this approach and arguments
that the 10-minute time frame was unrealistic for liquid releases
(particularly for pipelines and connected equipment) and thus did not
properly account for process-specific conditions.
EPA's approach for the liquid worst-case scenario in the final rule
is similar to the Technical Guidance methodology, in which the total
quantity of liquid in a vessel or pipeline is instantaneously spilled
upon failure, considering administrative controls or passive mitigation
discussed below. The rate of release to the air is not instantaneous;
it is determined by the volatilization rate of the spilled liquid,
which depends on the surface area of the pool formed after the spill.
The pool surface area is determined by assuming the spilled liquid
rapidly spreads out and forms a one-centimeter deep pool, unless
passive mitigation systems contain the pool to a smaller area. EPA
believes this approach is reasonable because total vessel or pipeline
failure will generally lead to immediate and rapid spillage followed by
pool volatilization. Further, if the liquid were assumed to spill over
a particular time frame rather than instantaneously, owners or
operators would need to calculate the amount of vapor emitted to the
air as the liquid is spilled, in addition to the volatilization rate as
the pool spreads out and reaches its maximum size. Computer-based
models are available for such calculations, but they are complex and
require considerable data input to use. EPA believes that liquid
spillage from a worst-case scenario is likely to be extremely rapid
such that the most significant portion of the release rate is given by
pool volatilization; consequently, liquid release time is not
necessary. Liquid spill rates and times could be reflected in
alternative scenarios discussed below.

As proposed, the worst-case for flammables assumes that the total
quantity of the substance in the vessel or pipeline vaporizes,
resulting in a vapor cloud explosion. If the vapor cloud explosion is
modeled using a TNT-equivalent methodology, then a 10 percent yield
factor must be used.

EPA requested comment in the SNPRM on whether the worst-case
scenario should include an additional

[[Page 31682]]

amount of substance that could potentially drain or flow from process
equipment interconnected with the failed vessel or pipeline. Many
commenters opposed this option, suggesting that it is technically
uncertain and would have little value in terms of what they saw as
EPA's intended purpose for the worst-case assessment. Other commenters
requested that ``interconnected equipment'' be defined and clarified.
Given the assumption of rapid release associated with initial equipment
failure, EPA agrees that determination of the spill rate from connected
piping and equipment is likely to be technically complex, very
different from that of the quantity in the vessel or failed pipeline,
and likely to extend the duration of volatilization rather than
affecting the rate overall. Therefore, EPA has not included this
requirement in the final rule.

EPA also sought comment in the SNPRM on options for the
determination of the relevant quantity of regulated substance in a
vessel or process piping for a worst-case release scenario: the maximum
possible vessel inventory (design capacity) at any time without regard
for operational practices and administrative controls; the maximum
possible vessel inventory unless there are internal administrative
controls (written procedural restrictions) that limit inventories to
less than the maximum; or historic or projected maximum operating
inventories without regard to administrative controls. EPA preferred
that the maximum vessel inventory including administrative controls
that might limit or raise the vessel quantity to be used in the worstcase
assessment and reported in the worst-case release analysis section
of the RMP. If the quantity used in the assessment were exceeded (e.g.,
an administrative control were ignored), then the source would be in
violation of the rule (i.e., failure to perform a worst-case analysis)
and RMP reporting unless the administrative control was revised, the
worst-case analysis updated to reflect any changes in the analysis, and
a revised RMP submitted. This approach acknowledges the efforts by
sources to increase process safety by intentionally reducing the
inventory of regulated substances (e.g., vessels kept at half capacity
to allow for process upsets, emergency shutdowns, and deinventorying or
maintenance turnarounds). EPA notes that at some sources, as a result
of inventory reduction measures, the largest quantity may be held in a
transportation container that is loaded or unloaded at the source (See
section P.2).

A few commenters supported the other options, noting that
administrative controls may fail, potentially generating a larger
scenario. However, the majority of commenters supported EPA's preferred
approach based on the historical reliability of such controls at many
sources and the role that such a provision could play in encouraging
their use at additional locations. Other commenters asked whether
mechanical controls, alone or in combination with administrative
controls, should be incorporated into the proposal. Although mechanical
controls may also serve to limit the quantity, EPA has decided not to
include them in the quantity determination for the worst-case release
scenario because the definition for administrative control as ``written
procedural mechanisms used for hazard control'' provides a backup for
possible failure of mechanical controls. For more discussion of
mechanical controls, see section III(B)(2), mitigation systems, below.
In the SNPRM, EPA considered providing the implementing agency with
the discretion to determine the appropriate quantity for the worst-case
release scenario on a site-specific or industry-specific basis. EPA
noted in the SNPRM, and most of the few comments received on this issue
agreed, that implementing agency discretion would result in increased
administrative burden on the implementing agency and crossjurisdictional
differences in the methodology used for the worst-case
analyses. EPA has decided not to incorporate this approach in the final
rule. States, however, may impose more stringent requirements, such as
additional modeling, under state authority.
In the NPRM worst-case definition, EPA did not specify what
constitutes or how to determine the worst offsite consequences. Some
commenters indicated that without clear direction, EPA's proposed worst
case might not actually capture the scenario that leads to the most
severe offsite impact. In the SNPRM, EPA indicated that the worst-case
scenario should be the scenario that generates the greatest distance to
a specified endpoint (i.e., the toxic vapor cloud or blast wave from a
vapor cloud explosion that travels the farthest).
EPA recognizes that there may be other release scenarios that could
generate a greater distance than the release from the largest vessel or
pipeline. Consequently, EPA has added paragraph (h) to Sec. 68.25 to
require owners or operators to consider other scenarios if those
scenarios generate greater distances to the endpoint than the distance
generated by the largest vessel or pipeline scenario. Owners or
operators need to consider releases from smaller vessels if those
vessels contain the substance at higher temperature or pressures or if
they are closer to public receptors. In some cases, the largest vessel
will be a storage vessel where the substance is held at ambient
conditions. A reactor vessel may hold a smaller quantity, but at high
pressures and temperatures, generating a release that could travel
farther offsite to an endpoint. Vessel location is important,
especially at large sources. A smaller vessel located nearer to the
stationary source boundary may generate a greater impact distance than
a larger vessel farther away. This difference may be particularly
important for flammables, because impact distances for flammables are
generally shorter than those for toxic releases.

2. Mitigation Systems

a. Worst-case scenario. In the NPRM worst-case scenario, EPA
indicated that sources must assume that both active and passive systems
fail to mitigate the release. Commenters were generally split between
those who wanted passive (as well as certain redundant active)
mitigation systems to be included and those who argued that historical
evidence from catastrophic releases suggests that the worst case should
assume the failure of all such systems. Those who supported mitigation
argued that inclusion provides a more credible scenario for improved
risk communication, accident prevention, and emergency planning.
EPA proposed in the SNPRM to include passive mitigation systems in
the worst-case release scenario as long as the system is capable of
withstanding, and continuing to function as intended during and after a
destructive event, such as an earthquake, storm, or explosion, which
causes a vessel or pipeline to fail. Passive systems such as dikes,
catch basins, and drains for liquids, and enclosures for both liquids
and gases, could be assumed to mitigate the release. Some commenters
opposed this approach, arguing again that the worst case should account
for the possibility of passive mitigation failure. The majority
supported this approach because the assumption that passive systems
specifically designed and installed as protection against a potential
catastrophe fail is unrealistic. Furthermore, the approach recognizes
and encourages prevention through additional passive mitigation and
supports more realistic emergency

[[Page 31683]]

planning. A few commenters also suggested that active mitigation
measures that were unlikely to fail (e.g., redundant or backup systems)
should be considered, for similar reasons. Historical data, however,
indicate that certain events compromise active mitigation systems
(e.g., explosions have destroyed fire water piping systems).

For the final rule, EPA has decided to adopt the SNPRM approach.
Passive mitigation systems would be defined as those systems that
operate without human, mechanical, or other energy input and would
include building enclosures, dikes, and containment walls. EPA also
agrees that reservoirs or vessels sufficiently buried underground are
passively mitigated or prevented from failing catastrophically. In this
case, sources should evaluate the failure of piping connected to
underground storage for the worst case or alternative case scenarios.
In addition to the requirements outlined in Sec. 68.25, EPA provides
guidance on how passive mitigation would affect release rate and
distance to endpoints in its RMP Offsite Consequence Analysis Guidance.

b. Alternative scenarios. EPA initially proposed that sources could
include passive mitigation systems in their alternative scenario
assessments, but that active mitigation systems (e.g., excess flow
valves, fail-safe and automatic shutdown valves, scrubbers, flares,
deluge systems, and water curtains) would be assumed to fail. Some
commenters generally opposed inclusion of any mitigation systems in the
hazard assessment, while other commenters noted that the alternative
release scenario should recognize and encourage industry accident
prevention efforts, specifically the installation of additional
mitigation systems, and support more realistic emergency planning.

EPA proposed in the SNPRM to allow sources to consider passive and
active mitigation measures in the alternative release scenario
assessment. Commenters supported this approach and EPA has decided to
retain it in the final rule. EPA agrees that the assumption that both
passive and active mitigation measures fail when such measures are
specifically designed and installed to mitigate catastrophic releases
is unrealistic for the alternative scenarios. Although not required,
EPA notes that sources may choose to apply passive and active
mitigation measures to a worst-case type scenario to illustrate the
capabilities of such systems to reduce the potential impact of a worstcase
accidental release. In addition to the requirements outlined in
Sec. 68.28, EPA provides guidance in its RMP Offsite Consequence
Analysis Guidance on how passive and active mitigation would affect
release rate and distance to endpoints.

3. Populations Affected. EPA described in the NPRM preamble certain
locations (e.g., schools and hospitals) where sensitive populations
might be present and proposed in the rule that owners or operators
identify potentially exposed populations as part of the offsite
consequence assessment. Commenters generally opposed requirements for
population surveys; several commenters suggested that Census data or
other readily available population information should be sufficient,
while other commenters indicated that the LEPC or other local planning
entities were the appropriate entity to prepare these data.

EPA believes owners or operators need to be aware of the magnitude
of impact on populations associated with the worst-case and alternative
scenarios. However, EPA learned that, although much of this information
is readily available, identification of some sensitive populations
could require considerable effort, especially if the distance to an
endpoint generated in the offsite consequence assessment is large or
crosses several jurisdictions. Consequently, EPA proposed in the SNPRM
that offsite populations be defined using available Census data;
information on the number of children and people over 65 could be
considered a proxy for sensitive populations, thereby accomplishing the
same objective as the proposed rule. EPA also indicated that it has
developed a geographic information system, LandView, that will
facilitate analysis of resident populations. (LandView can be ordered
from the U.S. Bureau of the Census customer service at (301) 457-4100.)
In general, commenters agreed with the SNPRM approach. However, some
commenters questioned the accuracy of potentially ten-year-old Census
data and requested additional flexibility, or a greater role for local
government, in this analysis.

EPA has decided to adopt the approach outlined in the SNPRM for the
final rule. Sources will be allowed to use available Census data to
estimate populations potentially affected. Sources may update these
data if they believe the data are inaccurate, but are not required to
do so. Populations shall be reported to two significant digits. Because
Census data are limited to residential populations, sources will also
have to note in the RMP whether other, non-residential populations,
such as schools, hospitals, prisons, public recreational areas or
arenas, and major commercial or industrial areas, are within the
distance to an endpoint. These institutions and areas are those that
can generally be found on local street maps. Sources will not be
required to estimate the number of people who might be present at these
locations. EPA provides further guidance on the identification of
affected populations in its RMP Offsite Consequence Analysis Guidance.

4. Number of Scenarios In the NPRM. EPA required a worst-case
release scenario for each regulated substance. Commenters requested
clarification, because one substance could be present in more than one
process at the source and sources would need to select the ``worst''
worst case for substances in multiple processes. In addition, one
process may have several, similar listed substances and multiple worstcase
analyses of similar substances (e.g., flammables) would not
provide additional useful information to the public.

EPA proposed in the SNPRM that sources report in the RMP one worstcase
release scenario representative of all toxic substances present at
the source and one worst-case release scenario representative of all
flammable substances present at the source. Even though additional
screening analyses to determine the appropriate worst-case scenario
might be necessary, this approach reduces to a maximum of two the
number of worst-case analyses reported in the RMP by a source. In
general, commenters favored this approach, particularly for flammables,
which do not produce markedly different adverse effects. A few
commenters argued that a single toxic substance should not be
considered representative of all toxic substances at a source, since
there are considerable differences in toxic endpoint and adverse
affect.

EPA has decided to adopt the approach outlined in the SNPRM for the
final rule: report one worst-case release scenario for all flammables
and one worst-case release scenario for all toxics at the source. EPA
notes that the worst-case scenario is designed principally to support a
dialogue between the source and the community on release prevention,
and not to serve as the sole or primary basis for local emergency
planning. The ``worst'' worst-case release scenario will inform the
broadest range of individuals that they may be impacted by the source
so that they may participate in dialogue with the source about
prevention, preparedness, and emergency response actions. Lesser worstcase
release scenarios would not

[[Page 31684]]

inform any person not already within the range of the ``worst'' worst
case even though the health effects may be different; consequently, EPA
believes that only a single toxic worst case is necessary. However,
sources must also analyze and report another worst-case release
scenario (for flammables or toxics) if such a release from another
location at the source potentially affects public receptors different
from those potentially affected by the first scenario (e.g., if a
large-sized source is located between two communities and has a covered
process adjacent to each community).

In the NPRM, EPA did not specify the number of alternative
scenarios to be reported for each regulated substance. EPA noted in the
preamble that this approach, while providing flexibility, may also
create uncertainty about what EPA will consider to be an adequate
number of scenarios. While a few commenters argued against scenarios
beyond the worst case, many commenters supported a requirement for a
maximum of two: the worst case plus one additional scenario; others
supported a maximum of three. Many of the commenters noted that local
entities could request further information under EPCRA section
303(d)(3) authority if they desired. At the same time, a number of
commenters suggested that this determination should be made by the
source based on their scenario analysis, perhaps in coordination with a
local agency.

In the SNPRM, EPA proposed to require one alternative release
scenario for all flammable substances at the source and one alternative
scenario for each toxic substance at the source. As discussed above,
the listed flammable substances behave similarly upon release and have
the same endpoint, while each toxic substance has a different endpoint
and different atmospheric behavior. EPA sought comment on whether one
toxic substance alternative scenario could represent all toxic
substances at a source or in a process. Although commenters generally
agreed with the approach for flammables, only a few argued that a
single alternative scenario for all toxics was also appropriate; most
others supported EPA's proposal.

Upon review of the comments, EPA has decided to adopt the approach
outlined in the SNPRM: an alternative release scenario must be reported
in the RMP for each toxic held above