[Federal Register Volume 68, Number 207 (Monday, October 27, 2003)]
[Rules and Regulations]
[Pages 61248-61280]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 03-26320]



[[Page 61247]]

-----------------------------------------------------------------------

Part II





Environmental Protection Agency





-----------------------------------------------------------------------



40 CFR Parts 51 and 52



Prevention of Significant Deterioration (PSD) and Non-Attainment New 
Source Review (NSR): Equipment Replacement Provision of the Routine 
Maintenance, Repair and Replacement Exclusion; Final Rule

  Federal Register / Vol. 68, No. 207 / Monday, October 27, 2003 / 
Rules and Regulations  

[[Page 61248]]


-----------------------------------------------------------------------

ENVIRONMENTAL PROTECTION AGENCY

40 CFR Parts 51 and 52

[FRL-7575-9; RIN 2060-AK28; Electronic Docket OAR-2002-0068; Legacy 
Docket A-2002-04]


Prevention of Significant Deterioration (PSD) and Non-Attainment 
New Source Review (NSR): Equipment Replacement Provision of the Routine 
Maintenance, Repair and Replacement Exclusion

AGENCY: Environmental Protection Agency (EPA).

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: The EPA is finalizing revisions to the regulations governing 
the NSR programs mandated by parts C and D of title I of the Clean Air 
Act (CAA). Today's changes reflect EPA's incorporation of comments from 
the proposed rule for ``Prevention of Significant Deterioration (PSD) 
and Non-attainment New Source Review (NSR): Routine Maintenance, Repair 
and Replacement.'' These changes provide a category of equipment 
replacement activities that are not subject to Major NSR requirements 
under the routine maintenance, repair and replacement (RMRR) exclusion. 
The changes are intended to provide greater regulatory certainty 
without sacrificing the current level of environmental protection and 
benefit derived from the NSR program. We believe that these changes 
will facilitate the safe, efficient, and reliable operation of affected 
facilities.

EFFECTIVE DATE: This final rule is effective on December 26, 2003.

ADDRESSES: Docket. Docket No. A-2002-04 (Electronic docket OAR-2002-
0068), containing supporting information used to develop the proposed 
rule and today's final rule, is available for public inspection and 
copying between 8:00 a.m. and 4:30 p.m., Monday through Friday (except 
government holidays) at the Air and Radiation Docket and Information 
Center (6102T), Room B-108, EPA West Building, 1301 Constitution 
Avenue, NW, Washington, D.C. 20460; telephone (202) 566-1742, fax (202) 
566-1741. A reasonable fee may be charged for copying docket materials.
    Worldwide Web (WWW). In addition to being available in the docket, 
an electronic copy of this final rule will also be available on the WWW 
through the Technology Transfer Network (TTN). Following signature, a 
copy of the rule will be posted on the TTN's policy and guidance page 
for newly proposed or promulgated rules: http://www.epa.gov/ttn/oarpg.

FOR FURTHER INFORMATION CONTACT: Mr. Dave Svendsgaard, Information 
Transfer and Program Integration Division (C339-03), U.S. EPA Office of 
Air Quality Planning and Standards, Research Triangle Park, North 
Carolina 27711, telephone 919-541-2380, or electronic mail at 
[email protected], for questions on this rule.

SUPPLEMENTARY INFORMATION:

Regulated Entities

    Entities potentially affected by this final action include sources 
in all industry groups. The majority of sources potentially affected 
are expected to be in the following groups:

------------------------------------------------------------------------
          Industry group                SIC \a\           NAICS \b\
------------------------------------------------------------------------
Electric Services.................             491  221111, 221112,
                                                     221113, 221119,
                                                     221121, 221122
Petroleum Refining................             291  324110
Industrial Inorganic Chemicals....             281  325181, 325120,
                                                     325131, 325182,
                                                     211112, 325998,
                                                     331311, 325188
Industrial Organic Chemicals......             286  325110, 325132,
                                                     325192, 325188,
                                                     325193, 325120,
                                                     325199
Miscellaneous Chemical Products...             289  325520, 325920,
                                                     325910, 325182,
                                                     325510
Natural Gas Liquids...............             132  211112
Natural Gas Transport.............             492  486210, 221210
Pulp and Paper Mills..............             261  322110, 322121,
                                                     322122, 322130
Paper Mills.......................             262  322121, 322122
Automobile Manufacturing..........             371  336111, 336112,
                                                     336211, 336992,
                                                     336322, 336312,
                                                     336330, 336340,
                                                     336350, 336399,
                                                     336212, 336213
Pharmaceuticals...................             283  325411, 325412,
                                                     325413, 325414
------------------------------------------------------------------------
\a\ Standard Industrial Classification.
\b\ North American Industry Classification System.

Entities potentially affected by this final action also include State, 
local, and tribal governments that are delegated authority to implement 
these regulations.

Outline

    The information presented in this preamble is organized as follows:

I. General Information
    A. How can I get copies of this document and other related 
information?
    1. Docket
    2. Electronic Access
    B. Where can I obtain additional information?
II. Background
    A. What is the RMRR exclusion?
    B. Issues surrounding the RMRR exclusion
    C. Process used to develop this rule
    D. What we proposed
III. Equipment Replacement Provision
    A. Overview and justification for today's final action
    B. What is an identical or functionally equivalent replacement 
and why should such an activity be considered RMRR?
    C. What cost limit has been placed on the equipment replacement 
approach?
    D. What will be the basis of applying the 20-percent threshold?
    E. What basic design parameters are being established to qualify 
for the equipment replacement provision?
    F. What collection of equipment should be considered in applying 
the equipment replacement provision and how should it be defined?
    G. Consideration of non-emitting units as part of the process 
unit
    H. What is the accounting basis for the process unit?
    I. Enforcement
    1. Compliance assurance
    2. General issues
    J. Quantitative Analysis
    K. Consideration of other options
    1. Annual Maintenance, repair and replacement allowance
    2. Capacity-based option
    3. Age-based option
    L. Specific list of excluded activities
    M. Stand-alone exclusion for energy efficiency projects
    N. Legal Basis
    1. How does the NSR program address existing sources and why is 
today's rule consistent with this approach?
    2. Why today's rule appropriately implements the Clean Air Act's 
definition of modification
IV. Administrative Requirements for This Rule
    A. Executive Order 12866--Regulatory Planning and Review

[[Page 61249]]

    B. Executive Order 13132--Federalism
    C. Executive Order 13175--Consultation and Coordination with 
Indian Tribal Governments
    D. Executive Order 13045--Protection of Children from 
Environmental Health Risks and Safety Risks
    E. Paperwork Reduction Act
    F. Regulatory Flexibility Analysis
    G. Unfunded Mandates Reform Act of 1995
    H. National Technology Transfer and Advancement Act of 1995
    I. Executive Order 13211--Actions Concerning Regulations That 
Significantly Affect Energy Supply, Distribution, or Use
    J. Executive Order 12988--Civil Justice Reform
V. Effective Date for Today's Requirements
VI. Statutory Authority

I. General Information

A. How Can I Get Copies of This Document and Other Related Information?

    1. Docket. The EPA has established an official public docket for 
this action under Docket ID No. A-2002-04. The official public docket 
consists of the documents specifically referenced in this action, any 
public comments received, and other information related to this action. 
Although a part of the official docket, the public docket does not 
include Confidential Business Information (CBI) or other information 
whose disclosure is restricted by statute. The official public docket 
is the collection of materials that is available for public viewing at 
the EPA Docket Center, (Air Docket), U.S. Environmental Protection 
Agency, 1301 Constitution Ave., NW., Room: B108, Mail Code: 6102T, 
Washington, DC, 20004. The EPA Docket Center Public Reading Room is 
open from 8:30 a.m. to 4:30 p.m., Monday through Friday, excluding 
legal holidays. The telephone number for the Reading Room is (202) 566-
1742. A reasonable fee may be charged for copying.
    2. Electronic Access. You may access this Federal Register document 
electronically through the EPA Internet under the ``Federal Register'' 
listings at http://www.epa.gov/fedrgstr/.
    An electronic version of the public docket is available through 
EPA's electronic public docket and comment system, EPA Dockets. You may 
use EPA Dockets at http://www.epa.gov/edocket/ to submit or view public 
comments, access the index listing of the contents of the official 
public docket, and to access those documents in the public docket that 
are available electronically. Once in the system, select ``search,'' 
then key in the appropriate docket identification number.
    Certain types of information will not be placed in the EPA Dockets. 
Information claimed as CBI and other information whose disclosure is 
restricted by statute, which is not included in the official public 
docket, will not be available for public viewing in EPA's electronic 
public docket. EPA's policy is that copyrighted material will not be 
placed in EPA's electronic public docket but will be available only in 
printed, paper form in the official public docket. To the extent 
feasible, publicly available docket materials will be made available in 
EPA's electronic public docket. When a document is selected from the 
index list in EPA Dockets, the system will identify whether the 
document is available for viewing in EPA's electronic public docket. 
Although not all docket materials may be available electronically, you 
may still access any of the publicly available docket materials through 
the docket facility identified in section I.A.1. of this preamble. The 
EPA intends to work towards providing electronic access to all of the 
publicly available docket materials through EPA's electronic public 
docket.
    For additional information about EPA's electronic public docket 
visit EPA Dockets online or see 67 FR 38102, May 31, 2002.

B. Where Can I Obtain Additional Information?

    In addition to being available in the docket, an electronic copy of 
today's final rule is also available on the WWW through the Technology 
Transfer Network (TTN). Following signature by the EPA Administrator, a 
copy of this rule will be posted on the TTN's policy and guidance page 
for newly proposed or promulgated rules at http://www.epa.gov/ttn/oarpg. The TTN provides information and technology exchange in various 
areas of air pollution control. If more information regarding the TTN 
is needed, call the TTN HELP line at (919) 541-5384.

II. Background

A. What Is the RMRR Exclusion?

    Title I of the Clean Air Act (CAA) established the New Source 
Review program \1\ to help control airborne emissions from major new 
stationary sources of pollution. Under the program, anyone who seeks to 
construct a new stationary source that will be a major source of 
regulated pollutants must obtain a permit from State authorities (or, 
where a State has not established its own program, from EPA directly) 
before beginning construction of the source. In order to obtain the 
permit, the owner or operator must, among other things, demonstrate 
that the new source will have state-of-the-art pollution control 
devices.
---------------------------------------------------------------------------

    \1\ We broadly use the term ``New Source Review,'' or NSR, to 
encompass both the PSD and the Non-attainment New Source Review 
program.
---------------------------------------------------------------------------

    The NSR program does not generally affect existing sources, but it 
does apply if they undergo a ``modification.'' The NSR provisions of 
the CAA do not create their own definition of ``modification,'' instead 
borrowing the definition of the term established by section 111 of the 
CAA, which defined the term for purposes of the New Source Performance 
Standards (NSPS) program. That definition states that ``[t]he term 
``modification'' means any physical change in, or change in the method 
of operation of, a stationary source which increases the amount of any 
air pollutant emitted by such source or which results in the emission 
of any air pollutant not previously emitted.'' Under 40 CFR parts 51 
and 52, the rules we have promulgated to carry out the NSR program, 
``major modification'' is similarly defined as any physical change in 
or change in the method of operation of a major stationary source that 
would result in: (1) A significant emissions increase of a regulated 
NSR pollutant; and (2) a significant net emissions increase of that 
pollutant from the major stationary source.\2\ The regulations further 
provide that certain activities do not constitute a ``physical change 
or change in the method of operation'' under the definition of ``major 
modification.'' One category of such activities is routine maintenance, 
repair and replacement (RMRR). The regulatory provisions excluding RMRR 
from the definition of change constitute the RMRR exclusion.
---------------------------------------------------------------------------

    \2\ Once a modification is determined to be major, NSR 
requirements apply only to those specific pollutants for which there 
would be a significant net emissions increase.
---------------------------------------------------------------------------

B. Issues Surrounding the RMRR Exclusion

    Until today, the NSR regulations have not further specified what 
types of activities are encompassed by the term RMRR. Heretofore, we 
have applied the RMRR exclusion exclusively on a case-by-case basis 
using a multi-factor test for determining whether a particular activity 
falls within or outside the exclusion. We have made these case-by-case 
determinations both in the context of applicability determinations, 
where a source or permitting authority has requested EPA's guidance 
concerning whether a particular activity falls within the exclusion or 
requires a permit, and

[[Page 61250]]

in the context of enforcement actions, where we have challenged an 
activity undertaken by a source after the fact and the source has 
asserted that the activity was permissible under the exclusion.
    This case-by-case approach has been praised for its flexibility, 
but criticized for hampering activities important to assuring the safe, 
reliable and efficient operation of existing plants. Specifically, some 
of the case-by-case determinations we have made, particularly over the 
past decade, and particularly in a series of enforcement actions, have 
been criticized for giving the exclusion a narrow scope that disallows 
replacement of significant plant components with identical or 
functionally equivalent components. Critics argue that the effect is to 
discourage plant owners or operators from engaging in replacements that 
are important to restoring, maintaining and improving plant safety, 
reliability, and efficiency. They further argue that this effect is 
exacerbated by what they assert are the uncertainties inherent in the 
case-by-case approach.
    To elaborate on the uncertainty issues: Unless an owner or operator 
seeks an applicability determination from his or her reviewing 
authority, it can be difficult for the owner or operator to know with 
reasonable certainty whether a particular activity constitutes RMRR. 
This gives the owner or operator five choices, two of which the owner 
or operator is not likely to select, and the other three of which have 
significant drawbacks for the productivity of the plant.
    First, the owner or operator may simply seek an NSR permit. That 
course, however, is likely to be time-consuming and expensive, since it 
will likely result in a requirement to retrofit an existing plant with 
state-of-the-art pollution controls which often is very costly and can 
present significant technical challenges. Therefore, an owner or 
operator is not likely to select this option if it can be avoided.
    Second, the owner or operator may proceed at risk without a 
reviewing authority determination. That option, however, is also not 
likely to be attractive where a significant replacement activity is 
involved, because if the owner or operator proceeds without a reviewing 
authority determination and if we later find that he or she made an 
incorrect determination on its own, the owner or operator faces 
potentially serious enforcement consequences. Those consequences could 
well include substantial fines (along with the further consequences of 
having been determined to be in violation of the CAA) and penalties and 
a requirement to install the state-of-the-art pollution controls, even 
though those controls present technical issues or represent a 
significant enough expenditure that they likely would have deterred the 
owner or operator from seeking a permit in the first place. The owner 
or operator is not likely to take this risk if he or she believes there 
is a high probability of these kinds of consequences and if he or she 
has other options.
    Third, the owner or operator may seek an applicability 
determination. That process, too, is time-consuming and expensive, 
albeit typically less so than seeking a permit. This path presents a 
potentially significant barrier to today's global, quick-to-market 
industries, such as computer chips, pharmaceuticals, and autos. This 
approach also is likely to result in substantial foregone activities 
that would enhance the safety, reliability and efficiency of the plant 
while awaiting the applicability determination.
    Fourth, the owner or operator may forego or curtail replacements 
that would enhance the safe, reliable, or efficient operation of its 
plant, instead opting to repair existing components even though they 
are inferior to current day replacements because they likely have 
deteriorated with use and probably are less advanced and less efficient 
than current technology. Foregoing the replacement activities 
altogether will reduce plant safety, reliability and efficiency; 
curtailing or postponing them does as well, differing only in the 
degree of these effects.
    Finally, the owner or operator may curtail the plant's productive 
capacity by replacing components with less than the best technology in 
order to be more certain that the replacement is within the RMRR 
regulatory bounds, or he or she may agree to limit the source's hours 
of operation or capacity or install less than state-of-the-art air 
pollution controls to ensure no increase in emissions. Either of those 
courses, however, will also result in loss of plant productivity.
    The uncertainties are also problematic for State and local 
reviewing authorities. They require those authorities to devote scarce 
resources to make complex determinations, including applicability 
determinations, and consult with other agencies to ensure that any 
determinations are consistent with determinations made for similar 
circumstances in other jurisdictions and/or that other reviewing 
authorities would concur with the conclusion.
    Industry commenters strongly echoed these concerns, asserting that 
the expense and delay associated with NSR scrutiny, whether or not the 
activity is ultimately judged to be subject to major NSR, have caused a 
number of facilities to forego needed and beneficial maintenance, 
repair, and replacement activities, including ones that would likely 
have reduced emissions. In our June 2002 report to the President, we 
similarly concluded that the NSR program has impeded or resulted in the 
cancellation of projects that would have maintained and improved the 
reliability, efficiency, or safety of existing energy capacity.
    We are persuaded that we should change the approach to the RMRR 
exclusion that we have been following for equipment replacements. The 
approach we have been taking often has not encompassed the replacement 
of existing components with identical or similar new components that 
serve the same function, that represent a small fraction of the value 
of the process unit of which they are a part, that do not change the 
process unit's basic design parameters, and that do not cause the 
process unit to exceed any emission limitations. For the reasons noted 
above, this approach tends to have the effect of leading sources to 
refrain from replacing components, to replace them with inferior 
components, or to artificially constrain production in other ways. We 
are persuaded that none of these outcomes advanced the central policy 
of the major NSR program as applied to existing sources, which is not 
to cut back on emissions from existing major stationary sources through 
limitations on their productive capacity, but rather to ensure that 
they will install state-of-the-art pollution controls at a juncture 
where it otherwise makes sense to do so. We also do not believe the 
outcomes produced by the approach we have been taking have significant 
environmental benefits compared with the approach we are adopting today 
and, indeed, we believe our new approach may well produce environmental 
improvements as compared to the old one.
    We are also persuaded that uncertainties surrounding the scope of 
the exclusion that are associated with the case-by-case approach tend 
to exacerbate the problem outlined above. These uncertainties can 
discourage replacements that would promote safety, reliability and 
efficiency even in instances where, if the matter were brought to EPA, 
we would determine that the replacement in question was RMRR. Such 
discouragement results in lost capacity and lost opportunities to 
improve energy efficiency and reduce air pollution.
    We believe that these problems will be significantly reduced by the 
rule we

[[Page 61251]]

are adopting today. This rule specifies that the replacement of 
components of a process unit with identical components or their 
functional equivalents will come within the scope of the exclusion, 
provided the cost of replacing the component falls below 20 percent of 
the replacement value of the process unit of which the component is a 
part, the replacement does not change the unit's basic design 
parameters, and the unit continues to meet enforceable emission and 
operational limitations.
    Our new equipment replacement approach will allow owners or 
operators to replace components under a wider variety of circumstances 
than they have been able to do under our prior RMRR approach. It also 
provides more certainty both to source owners or operators who will be 
able better to plan activities at their facilities, and to reviewing 
authorities who will be able better to focus resources on other areas 
of their environmental programs rather than on time-consuming RMRR 
determinations. The effect should be to remove disincentives to 
undertaking RMRR activities falling within the rule, thereby enhancing 
key operational elements such as efficiency, safety, reliability, and 
environmental performance. For example, we anticipate that improved 
safety and reliability will result in more stable process operations 
and reduce periods of startup, shutdown, and malfunction and the 
increased emissions usually associated with them. Accordingly, we 
believe the rule will promote the central purpose of Title I of the 
CAA, ``to protect and enhance the quality of the Nation's air resources 
so as to promote the public health and welfare and the productive 
capacity of its population.'' CAA section 101.
    We note that we continue to believe that our prior narrower and 
entirely case-by-case approach to the RMRR exclusion was consistent 
with the relevant language of the CAA and a reasonable effort to 
effectuate its policies. At the same time, we also believe that the 
final rule's categorical exclusion of certain replacement activities 
and the broader definition of RMRR on which that exclusion is premised 
are likewise consistent with the statute's language and represent a 
better accommodation of the statute's twofold ends. We therefore have 
decided to adopt the final rule.

C. Process Used To Develop This Rule

    In the 1992 ``WEPCO Rule'' preamble, we declared our intent to 
issue guidance on the subject of RMRR. In 1994, as an outgrowth of 
meetings with the Clean Air Act Advisory Committee, we developed, for 
discussion purposes only, a preliminary draft that presented possible 
ways of how RMRR could be defined. We received a substantial volume of 
comments on this document. We subsequently decided not to include this 
preliminary draft approach in our 1996 NSR proposed rulemaking.
    In 2001, the President's National Energy Policy directed EPA in 
consultation with the Department of Energy (DOE) and other Federal 
agencies to review the impact of NSR on investment in new utility and 
refinery generation capacity, energy efficiency and environmental 
protection. Our Report to the President illustrated the problems 
associated with our prior case-by-case approach to identifying RMRR 
activities and underscored the advantages of establishing an objective 
bright-line approach for administering the RMRR provision.
    We held conference calls with various stakeholders during October 
2001 (including representatives from industry, State and local 
governments, and environmental groups) to discuss new ideas that were 
raised as to how the RMRR provision might be improved. The proposed 
RMRR rule reflected many of the ideas discussed in those meetings. 
Today's final rule on the equipment replacement provision is based on 
careful consideration of comments received on the proposed RMRR rule 
(67 FR 80920, December 31, 2002), where we sought comment on all 
aspects of our proposed approaches. Today's rule represents final 
action on only one part of what we proposed in December 2002--the 
equipment replacement provision. We have decided, for now, not to take 
final action on the proposed annual maintenance, repair and replacement 
allowance approach.

D. What We Proposed

    The RMRR proposal offered for comment two cost-based approaches for 
determining what constitutes routine maintenance, repair, and 
replacement. Under the proposal, facilities could have relied on a 
facility-wide annual maintenance, repair and replacement allowance and/
or an equipment replacement cost threshold to determine whether major 
NSR requirements were triggered by performing plant maintenance, repair 
and replacement activities. The proposal additionally outlined two 
options based on the capacity and age of a facility. We solicited 
comment on all aspects of the proposed approaches as well as any other 
viable option for clarifying the term ``routine maintenance, repair, 
and replacement.'' We took public comment on the proposed rule until 
May 2, 2003--120 days following publication in the Federal Register.
    Under the ``annual maintenance, repair and replacement allowance,'' 
an annual maintenance cost allowance would be established for each 
industrial facility based on an industry-specific percentage. For the 
percentage, we considered using the Internal Revenue Service ``Annual 
Asset Guideline Repair Allowance Percentages'' (AAGRAP), which for 
years has been used as an integral part of an exclusion under the New 
Source Performance Standard (NSPS) program. A multi-year allowance 
approach, in addition to the annual approach, was also offered for 
consideration in the proposal.
    Safeguards were proposed to ensure that the types of activities 
undertaken under the annual allowance are not activities that should be 
subject to greater scrutiny. These safeguards include: (1) No new unit 
may be installed; (2) no unit may be replaced in its entirety; and (3) 
changes may not cause an increase in the short-term emission rate of 
any regulated NSR pollutant.
    Under the ``equipment replacement provision,'' or ERP, we proposed 
to streamline the process for determining if major NSR permitting 
requirements apply to replacement of existing equipment with identical 
new equipment or with functionally equivalent equipment. Per-
replacement-of-component(s) thresholds, potentially up to 50 percent of 
the cost of replacing the process unit, were suggested by the proposal. 
As long as the threshold was not exceeded and the basic design 
parameters remained unchanged, the activity would be considered RMRR 
under this approach.
    Under the proposal, all activities that fell within the annual 
maintenance, repair and replacement allowance or the equipment 
replacement threshold and that met all the other criteria for these 
provisions would be considered RMRR without further review. Activities 
that were unable to be accommodated under the annual maintenance, 
repair and replacement allowance or the equipment replacement threshold 
could still qualify for the RMRR exclusion after a case-by-case review 
in accordance with current rules.
    We solicited comments on all aspects of our RMRR proposal.

III. Equipment Replacement Provision

A. Overview and Justification for Today's Final Action

    Today, we are revising certain provisions of the major NSR program 
by

[[Page 61252]]

finalizing the equipment replacement provision (ERP) to specify 
activities that will automatically qualify for the RMRR exclusion. This 
rule is effective on December 26, 2003. At this time, we are not taking 
action on our proposed annual maintenance, repair and replacement 
allowance approach.
    Although many commenters requested that we further clarify the 
case-by-case approach for determining whether an activity is RMRR, we 
are not taking action on this suggestion at this time. We are still 
considering what, if any, changes should be made to that policy. In the 
meantime, the case-by-case approach will remain available for the owner 
or operator of a source to use as an alternative and/or supplement to 
today's ERP.
    Under today's rule, an activity (or aggregations of activities) can 
qualify for the ERP if: (1) It involves replacement of any existing 
component(s) \3\ of a process unit with component(s) that are identical 
or that serve the same purpose as the replaced component(s); (2) the 
fixed capital cost of the replaced component(s), plus costs of any 
activities that are part of the replacement activity (e.g., labor, 
contract services, major equipment rental, and associated repair and 
maintenance activities),\4\ does not exceed 20 percent of the current 
replacement value of the process unit; and (3) the replacement(s) does 
not alter the basic design parameters of the process unit or cause the 
process unit to exceed any emission limitation or operational 
limitation (that has the effect of constraining emissions) that applies 
to any component of the process unit and that is legally enforceable.
---------------------------------------------------------------------------

    \3\ For the sake of clarity, we want to be clear that the term 
``component'' is meant to be applied broadly and read broadly to 
include replacements of both large components, such as economizers, 
reheaters, etc. at a boiler, as well as small items, such as screws, 
washers, gaskets, etc.
    \4\ We note that certain ancillary costs incurred during a given 
replacement activity should not be part of the replacement activity, 
such as replacement power that must be purchased during the 
maintenance shutdown of an electric utility.
---------------------------------------------------------------------------

    Today's final rule specifies the procedures by which the owner or 
operator of a source selects the basic design parameters for steam 
electric generating facilities and for other types of process units. 
Specifically, for steam electric generating facilities, we have 
clarified our proposed approach by specifying maximum hourly heat input 
and fuel consumption rate \5\ as basic design parameters. We are also 
allowing owners or operators of steam electric generating facilities 
the option to select a pair of parameters based on the process unit's 
output--more specifically, maximum hourly electric output rate or 
maximum steam flow rate--as an alternative to the previously proposed 
input-based parameters. Likewise, we are retaining our proposed 
approach of specifying maximum rate of fuel or material input for other 
types of process units, but we also allow you to use maximum rate of 
heat input, or maximum rate of product output if you prefer an output-
based basic design parameter. In addition, we allow you to propose an 
alternative basic design parameter(s), if the above options are 
inappropriate for your process unit.
---------------------------------------------------------------------------

    \5\ Actually proposed as ``fuel consumption specifications.''
---------------------------------------------------------------------------

    We are not specifically defining the basis for determining the 
replacement value of a new process unit. Instead, the final rule 
provides you with the flexibility of using any of the following: (1) 
Replacement cost; \6\ (2) invested cost, adjusted for inflation; (3) 
the insurance value, where the insurance value covers complete 
replacement of the process unit (rather than, for example, lost revenue 
replacement); or (4) another accounting procedure to establish a 
replacement value of the process unit if such accounting procedure is 
based on Generally Accepted Accounting Principles (GAAP). The GAAP are 
the conventions, rules and procedures that define accepted accounting 
practice for recording and reporting financial information, including 
broad guidelines as well as detailed procedures. The basic doctrine was 
set forth by the Accounting Principles Board of the American Institute 
of Certified Public Accountants, which was superseded in 1973 by the 
Financial Accounting Standards Board.
---------------------------------------------------------------------------

    \6\ Replacement cost can be either an estimate of the fixed 
capital cost of constructing a new process unit or the current 
appraised value of the process unit.
---------------------------------------------------------------------------

    If you choose to use options 3 or 4 to determine the replacement 
value for a particular process unit, you must send a notice reflecting 
your decision to your reviewing authority. The first time that an owner 
or operator submits such a notice for a particular process unit, the 
notice may be submitted at any time, but any subsequent notice for that 
process unit may be submitted only at the beginning of the process 
unit's fiscal year. You must continue to use the same basis to evaluate 
any additional activities that you undertake on that process unit 
within that same fiscal year. If you have provided notice of using 
either option 3 or 4, then the reviewing authority will assume that the 
same method will be used for subsequent fiscal years unless you send a 
notice to them declaring your intent to use another method. In the 
absence of providing any notification to your reviewing authority, you 
must use option 1 or 2.
    The final rules also set forth a definition of process unit, 
specifically delineate the boundary of the process unit for certain 
specified industries, and define a functionally equivalent replacement. 
A more detailed discussion of these requirements and our rationale for 
this action is contained in other parts of this preamble section.
    Today's final rules are designed to allow you to engage in 
activities that facilitate the safe, reliable and efficient operation 
of your source. We believe that today's final action broadens the major 
NSR program exclusion for equipment replacements and provides you with 
additional certainty as to what equipment replacement activities 
qualify for the RMRR exclusion. By adding certainty to the process, we 
are removing the disincentives to undertaking routine equipment 
replacements and promoting proper operational planning to facilitate 
safe, reliable and efficient operations. When an activity qualifies for 
the ERP, it will be considered RMRR and excluded from major NSR without 
regard to other considerations. In many cases, we believe that 
maintaining safe, reliable and efficient operations will have the 
corresponding environmental benefit of reducing the amount of pollution 
generated per product produced. The final rules also will reduce the 
resource burden on reviewing authorities resulting from implementation 
of the existing, case-by-case process for determining RMRR. In these 
respects, the final rules are consistent with the central purpose of 
the CAA, ``to protect and enhance the quality of the Nation's air 
resources so as to promote the public health and welfare and the 
productive capacity of its population.'' CAA section 101.

B. What Is an Identical or Functionally Equivalent Replacement and Why 
Should Such an Activity Be Considered RMRR?

    We proposed to exclude the replacement of existing equipment with 
identical or functionally equivalent components. As we observed at the 
time of our RMRR proposal, we believe that most identical and 
functionally equivalent replacements are necessary for the safe, 
efficient and reliable operations of virtually all industrial 
operations; are not of regulatory concern; will improve air quality 
(e.g., by decreasing startup, shutdown, and malfunctions); and thus 
should qualify

[[Page 61253]]

for the ERP under the RMRR exclusion. We believe industrial facilities 
are constructed with the understanding that certain equipment failures 
are common and ongoing maintenance programs that include replacing 
components in order to maintain, restore, or enhance the reliability, 
safety, and efficiency of a plant are routine. Conversely, delaying or 
foregoing maintenance could lead to failure of the production unit and 
may create or add to safety concerns.
    When such equipment replacement occurs, the replaced component is 
inherent to both the design and purpose of the process unit, and there 
is no reason to believe that such activity will cause the unit to emit 
above its original design capacity. Moreover, most of these 
replacements are conducted at industrial facilities to maintain proper 
operations and to implement good engineering practices. For example, if 
a pump associated with a distillation column fails and is replaced with 
an identical new pump, we believe that such a common activity is and 
should be considered an excluded replacement. It is not a ``change'' to 
the plant, since it merely maintains the plant as designed. Instead, it 
is the type of activity expected to occur to maintain the plant. 
Therefore, we think replacements like this properly fall within the 
exclusion for ``routine maintenance, repair and replacement.'' We also 
believe treating them in this fashion is consistent with the basic 
policies of the CAA: that existing plants are subject to major NSR 
permitting requirements only when they engage in an activity that 
constitutes an opportune time to install state-of-the-art pollution 
control equipment.
    We also believe that this principle extends beyond the replacement 
of equipment with identical equipment. When equipment is wearing out or 
breaks down, it often is replaced with equipment that serves the same 
purpose or function but is different in some respects or improved in 
some ways in comparison with the equipment that is removed. To continue 
with the example used above, if, instead of replacing the worn out 
distillation column pump with an identical one, the owner or operator 
replaced it with a new and improved model, it does not seem to us that 
this changes the fundamental reasons for treating that replacement as 
likewise within the scope of ``routine maintenance, repair and 
replacement.''
    This is particularly true since technology is constantly changing 
and evolving. When equipment of this sort needs to be replaced, it 
often is simply not possible to find the old-style technology. Owners 
or operators may have no choice but to purchase and install equipment 
reflecting current design innovations. Even if it is possible to find 
old-style equipment, it seems unnecessary and undesirable to generally 
construe NSR permitting requirements in a manner that is bound to deter 
owners or operators from using the best equipment that suits the given 
need when replacements must be installed.
    The limiting principle here is that the replacement equipment must 
be identical or functionally equivalent and must not change the basic 
design parameters of the affected process unit (e.g., for electric 
utility steam generating units, this might mean heat input and fuel 
consumption specifications). We also believe, however, that we need not 
and should not treat efficiency as a basic design parameter as we do 
not believe NSR was intended to impede industry in making energy and 
process efficiency improvements. We believe such improvements, on 
balance, will be beneficial both economically and environmentally. This 
treatment of efficiency should address the concern and perception that 
the NSR program serves as a barrier to activities undertaken to 
facilitate, restore, or improve efficiency, reliability, availability, 
or safety of a facility.
    Today's rule does not distinguish between the replacement of 
components that are expected to be replaced frequently or periodically 
and the replacement of components that may occur on a less frequent or 
one-time basis. It likewise does not distinguish between the 
replacement of larger and smaller components, instead requiring greater 
scrutiny if the replacement in question is part of an activity that 
exceeds 20 percent of the replacement value of the process unit.
    Our decisions on these points are derived from reflection on the 
function of the exclusion in the context of the CAA. As explained 
above, and as described more fully in our legal analysis set forth 
below, we do not believe that application of the major NSR program to 
``modified'' plants is designed to require existing plants that are 
continuing to operate in a manner consistent with their original design 
to curtail their rate of production or hours of operation beyond 
limitations set forth in their existing permits. We likewise do not 
believe that the program is designed to discourage plants from 
replacing parts or components so as to preserve their ability to 
produce at that rate. Rather, we believe Title I of the Clean Air 
largely leaves to State and local permitting authorities whether to 
require adjustments in the operations of those plants in order to 
reduce emissions to the degree needed to attain or maintain national 
air quality standards, and how to weigh the trade-offs such adjustments 
may produce in terms of potential economic impacts and loss of 
productivity. Instead, we believe the central function of the 
application of major NSR permitting requirements to ``modifications'' 
is to assure that plants install state-of-the-art pollution controls.
    We recognize that on these points, the approach taken by our final 
rule thereby differs in some respects from the multi-factor, case-by-
case approach we have been using in identifying RMRR, and particularly 
from some of our applications of that test to certain equipment 
replacements. We believe, however, that this adjustment in our approach 
is fully warranted for the reasons outlined above, and described more 
fully in our legal analysis below.
    The following examples of functionally equivalent replacements 
under today's rule include:\7\
---------------------------------------------------------------------------

    \7\ As discussed in more detail below, although such activities 
would be functionally equivalent, they would still need to meet 
other criteria to qualify for the ERP. For example, a functionally 
equivalent replacement does not qualify for the ERP if it results in 
a change to a basic design parameter of the affected unit. If an 
activity does not qualify for RMRR under the ERP, the case-by-case 
RMRR approach would still be available to the owner or operator 
under those circumstances. And, of course, even if the activity does 
not qualify for the RMRR exclusion, the activity will not be a 
modification and, hence, will not trigger NSR unless it results in a 
significant emissions increase.
---------------------------------------------------------------------------

    [sbull] Replacing worn out pipes in a chemical process plant with 
pipes that are constructed of different metallurgy (e.g., to help 
reduce corrosion, erosion, or chemical compatibility problems).
    [sbull] Replacing an analog controller with a digital controller, 
even though a similar analog controller can still be purchased and even 
though the new controller would allow for more precise control. A good 
example was presented to us by the forest products industry during our 
review of the NSR program's impacts on the energy sector. A company in 
that sector needed to replace outdated analog controllers at a series 
of six batch digesters. In this case, the original controllers were no 
longer manufactured. The new digital controllers, costing approximately 
$50,000, are capable of receiving inputs from the digester vessel 
temperature, pressure, and chemical/steam flow. The new controllers 
would have more precisely filled and pressurized digesters with chips, 
chemicals, and steam, thus bringing a batch digester on line faster.
    [sbull] Replacing an existing mill or pulverizer (e.g., grinding 
clinker in a

[[Page 61254]]

cement factory or coal for a boiler) with a new one of a different type 
because both new and old equipment serve the same purpose (even if the 
characteristics of the ground material would be different before and 
after the replacement).
    [sbull] Replacing existing spray paint nozzles with new ones that 
might atomize the spray better or have a higher transfer efficiency 
because the ``before'' and ``after'' nozzles serve the same function.
    At the same time, there are numerous activities that occur at 
facilities that may fall within the bounds of the cost threshold 
percentage, basic design parameters, and other backstop features of 
today's rule, but nevertheless cannot qualify for the RMRR exclusion on 
the grounds that the equipment is neither identical nor functionally 
equivalent. An example of this would be a chemical processing facility 
where the owner or operator makes a physical change that allows the 
production of a new end product that physically could not have been 
manufactured with the previous equipment using the same raw materials 
as used before in the same amounts as before. This would not be a 
functionally equivalent replacement activity because the facility is 
able to produce an end product after making the change that the 
facility was not capable of making before the change. Consequently, 
this activity would not qualify as RMRR under today's ERP.
    Several commenters said the equipment replacement provision will 
streamline the major NSR applicability analysis. A number of commenters 
believed the ERP would be easier to implement than the proposed annual 
maintenance, repair and replacement allowance approach. One commenter 
said that allowing identical replacements to be excluded from major NSR 
will codify existing industrial practices, where replacement has no 
impact on emissions and would clearly represent RMRR.
    Many commenters expressed support for the ERP, but recommended 
certain changes that they felt needed to be made to improve the 
proposal. One commenter supported the ERP in combination with a 
capacity-based option, on the assumption that repair and maintenance is 
to be excluded as well as equipment replacement.
    One commenter attempted to collect data from turbine customers and 
found that achieving a level of data collection necessary for the ERP 
was far from simple, because the cost of maintenance activities is 
affected by such things as variability in engine model, package 
technology, and type of maintenance contract. Another commenter gave an 
example of the benefit that the ERP may provide. Without the ERP, the 
commenter said the source is limited to some fraction of boiler tubes 
allowed to be replaced at a given time, whereas with the ERP, 
replacement of all boiler tubes would, in the commenter's opinion, 
rightfully be considered routine. Another commenter said the ERP will 
remove regulatory burdens for types of equipment replacements that are 
in their view ``routine,'' such as replacement of tubes in industrial 
boilers. They added that, without a clearer understanding of which 
activities are RMRR, they may be inclined to delay conducting such 
replacements.
    Many other commenters generally opposed any change to the RMRR 
exclusion, including one based on equipment replacement. Some of these 
commenters believed the ERP was problematic because it would allow a 
source to replace an entire process unit over time. Two of the 
commenters opposed the ERP because they felt it would create 
disincentives for the implementation of Plantwide Applicability Limits 
(PAL) and Clean Unit provisions from the recently finalized rule.
    One commenter said that from an engineering standpoint, for a power 
plant, the difference between routine maintenance and a major plant 
refurbishing project is clear. To further clarify, the commenter made 
the following points. According to the commenter, routine maintenance 
is frequent and follows a predictable pattern. The commenter 
characterized routine maintenance at power plants as: repair of leaking 
pipes, pumps, valves, and fans; cleaning and lubrication of components; 
and inspections. The commenter added that permanent staff do this work 
either while the plant is operating or during only brief periods of 
downtime. The commenter further expressed that activities that are not 
routine require long plant or process unit shutdowns, are done 
infrequently, and are major capital projects for which special funding 
is set aside as a result of years of planning and design work.
    One commenter said the proposal will allow emissions increases that 
will be difficult to offset through other regulations. One commenter 
objected to the ERP for a number of reasons: (1) The provision does not 
prevent replacement with different equipment; (2) it does not promote 
efficiency improvements or application of good air pollution controls; 
and (3) it would allow replacements that would significantly increase 
emissions. This commenter said replacement of air pollution controls 
should trigger best available control technology (BACT) or lowest 
achievable emission rate (LAER) requirements. Two local air pollution 
control agencies in California noted that they currently already 
exclude all replacements with identical equipment from major NSR when 
certain conditions are met.
    Commenters generally had similar viewpoints on allowing both 
identical and functionally equivalent equipment replacements to qualify 
as RMRR. However, some commenters expressed greater concern related to 
excluding the replacement of equipment with functionally equivalent 
equipment. Primarily their concerns were rooted in the fact that a 
functionally equivalent replacement component could lead to increases 
in operational efficiency or productivity, and these commenters 
asserted that these sorts of process enhancements should not be 
excluded as RMRR.
    We agree with the commenters who felt identical and functionally 
equivalent replacement activities generally should be excluded as RMRR. 
We also agree with the commenters who believe that this provision will 
streamline the major NSR applicability process and will bring clarity. 
The provision we are finalizing will allow a source to make a simple 
determination as to whether a replacement piece of equipment qualifies 
as identical or functionally equivalent. This type of determination 
will be straightforward and easier for the source to implement than the 
current case-by-case analysis required to determine a replacement falls 
within the RMRR exclusion. We support the air pollution agencies that 
have already excluded these types of changes from NSR.
    We disagree with those commenters who believe that this provision 
will create disincentives for sources to accept a PAL or have emission 
units designated as Clean Units. A PAL offers a source to bring on 
entirely new emissions units with no Federal preconstruction permit, as 
long as emissions caps are not exceeded. A PAL or a Clean Unit 
designation allows a source to make modifications without performing a 
major NSR applicability test. These advantages will still be the 
driving force for sources to elect to use the PAL or Clean Unit 
provisions, and we do not believe this final rule will significantly 
detract from their appeal.
    We also believe that there is substantial value in facilitating 
equipment replacements to a greater degree than our current approach 
permits and draws a cleaner and more

[[Page 61255]]

easily administered line between equipment replacements that 
categorically do not require a permit and major plant refurbishing 
which will result in increased emissions. For pieces of equipment used 
at industrial facilities, most manufacturers have well-established 
procedures for the inspection and replacement that are part of the 
regular maintenance necessary to provide for the equipment's safe, 
efficient and reliable operation. Some of these replacements are large 
in terms of cost and infrequent, but all are necessary to maintain the 
safe, efficient and reliable use of the process unit. We believe it is 
important to allow for these replacements provided that certain 
safeguards are in place, as discussed below.
    We disagree with suggestions from commenters that the time period 
between activities, standing alone, provides an appropriate or clear 
distinction between activities that should be permissible under the 
RMRR exclusion and those that should not. In fact, some components wear 
out every year, while others wear out every 20 years. Nevertheless, 
both types of changes should fall within the ERP of the RMRR exclusion 
because both allow the facility to operate as designed. By not imposing 
a time limitation, the ERP allows replacement activities to be driven 
by consideration of economic efficiency rather than artificial 
regulatory constraints.
    We disagree with commenters who expressed particular concern about 
functionally equivalent replacements. We continue to believe such 
activities should be encouraged and should qualify as RMRR. Even though 
a functionally equivalent component varies in some respects from the 
replaced component, we feel the most important factor to consider is 
whether the replacement will serve the same purpose as the replaced 
component. We acknowledge that a functionally equivalent replacement 
can result in an increase in efficiency and, consequently, 
productivity. In fact, one of our goals is to promote such outcomes. 
However, we believe that the basic design parameter safeguard is 
appropriate to assure that the ERP only automatically excludes from 
major NSR functionally equivalent replacements that do not result in a 
significant change to the fundamental characteristics of the process 
unit.
    We note that the two local programs in California that exclude the 
replacement of equipment with identical equipment also allow the 
replacement of equipment with functionally equivalent equipment without 
considering such action to be a modification. Due to local air quality 
considerations, the local programs establish minimum pollution control 
requirements that are imposed in some circumstances when functionally 
equivalent equipment replacements occur. Nothing in today's rule would 
prevent a State or local program from imposing additional requirements 
necessary to meet Federal, State or local air quality goals.
    After reviewing the comments on our proposal, we have decided to 
promulgate what we proposed in December 2002 for the RMRR equipment 
replacement provision with relatively minor changes. We decided to 
include another safeguard in addition to those we proposed in order to 
appropriately constrain the meaning of the term ``functionally 
equivalent.'' The additional safeguard is that an excluded replacement 
activity cannot cause the process unit to exceed any emission 
limitation or operational limitation (that has the effect of 
constraining emissions) that applies to the process unit and that is 
legally enforceable.
    Thus, today's final rule allows you to categorize identical and 
functionally equivalent equipment replacements as RMRR if the fixed 
capital cost of such replacement plus the cost of repair and 
maintenance activities that are part of the replacement activity does 
not exceed 20 percent of the replacement value of the process unit, and 
if the replacement does not alter a basic design parameter of the 
process unit or cause the process unit to exceed any emission 
limitation or operational limitation (that has the effect of 
constraining emissions) that applies to the process unit.

C. What Cost Limit Has Been Placed on the Equipment Replacement 
Approach?

    The next concept presented in the proposal is the cost-based 
limitation on the scope of the ERP. The purpose of this threshold is to 
distinguish between those equipment replacement activities that should 
automatically qualify as RMRR without further consideration and those 
activities that should undergo case-specific consideration. This 
concept is akin to the long-established reconstruction provision under 
the NSPS program. For the reasons explained below, we have decided to 
establish a 20-percent cost threshold under the ERP.
    We believe a similar bright-line rule that would obviate the need 
for case-by-case review under our multi-factor test of appropriate 
categories of equipment replacements would be extremely useful in 
addressing many of the problems that we have identified with the 
current operation of the NSR program. Such a rule would be particularly 
useful in avoiding the uncertainty and delay, and consequent postponed 
or foregone equipment replacements, that our multi-factor case-by-case 
review induces. For example, our RIA indicates that it takes a year, on 
average, to obtain a determination whether a proposed replacement is 
routine. That kind of delay obviously creates perverse disincentives to 
refrain from equipment replacements and instead repair existing 
equipment or find some other solution.
    This is the kind of problem that classically leads agencies to 
fashion bright-line tests to provide greater regulatory certainty and 
efficiency. Moreover, because the kind of disincentives that give rise 
to this concern operate largely by economic means, prompting sources to 
take one course of action (cut back on productive equipment 
replacement) rather than another (replace the equipment and incur the 
costs of delay, as well as potentially the costs of installing state-
of-the-art controls), we think a cost-based threshold is a reasonable 
basis on which to create such a bright-line rule.
    In the proposal, we observed that it may sometimes be difficult to 
determine where to draw the line between an activity that should be 
treated as an excluded replacement activity and one that should be 
viewed as a physical change that might constitute a major modification, 
when the replacement of equipment with identical or functionally 
equivalent equipment involves a large portion of an existing process 
unit. We solicited comment on a range of equipment replacement cost 
thresholds such as one based on the NSPS program. Under the NSPS 
program, when the cost of a project at an existing affected facility 
exceeds 50 percent of the fixed capital cost that would be required to 
construct a comparable entirely new unit (that is, the current capital 
replacement value of the existing affected source), then the source 
must notify and provide information to the permitting authority. After 
considering a range of factors, including the cost of the activity, the 
estimated life of the facility after the replacements, the extent to 
which the replaced equipment causes or contributes to the emissions 
from the source, and any economic or technical limitations on 
compliance with the NSPS, the reviewing authority

[[Page 61256]]

determines whether the proposed project is a reconstruction.\8\
---------------------------------------------------------------------------

    \8\ In the proposal, it was incorrectly stated that 
applicability of the NSPS was triggered if a project exceeded 50 
percent of the cost of replacing the affected facility. As stated in 
this notice, if an activity exceeds this cost threshold, that only 
triggers further evaluation, not the automatic application of the 
NSPS to the source.
---------------------------------------------------------------------------

    We observed that, in some respects, an equipment replacement cost 
threshold set at the NSPS reconstruction test could be an appropriate 
approach for distinguishing between routine and nonroutine identical 
and functionally equivalent replacements under the major NSR program. 
As under the NSPS program, we do not believe it is reasonable to 
exclude from major NSR those activities that involve the total 
replacement of an existing entire process unit.
    We also noted, however, that there are other considerations 
pointing in favor of a threshold lower than the 50-percent 
reconstruction threshold that might be appropriate to bound the ERP. 
Under NSPS, when a source undertakes a replacement activity at an 
existing affected facility that constitutes half or more of the 
facility's capital replacement value, our rules require a case-by-case 
determination as to whether such replacements constitute construction. 
We noted that a percentage threshold lower than 50 percent might be 
more appropriate for determining where we would require case-by-case 
consideration of the question whether equipment replacements constitute 
a modification of an existing process unit under major NSR. We 
solicited comments on the appropriate level of any percentage.
    Many commenters supported the threshold of 50 percent of 
replacement value as the upper limit on equipment replacement. They 
felt this number is consistent with existing regulatory requirements 
and would accord the flexibility originally intended under the CAA for 
RMRR activities, while at the same time assuring that major, nonroutine 
projects remain subject to major NSR applicability review, and they 
felt this number is consistent with a common-sense interpretation of 
the regulations.
    They also believed a 50-percent cutoff to be consistent with 
reconstruction definitions used in many NSPS and National Emission 
Standards for Hazardous Air Pollutants regulations. Some commenters 
stated that a 50-percent cutoff for the ERP would be valid for the same 
reason as for the NSPS reconstruction test; significant changes to a 
process unit are necessary before retrofit controls should be 
considered, provided there is no increase in emissions.
    Many other commenters opposed the 50-percent replacement value 
threshold. They believed the capital replacement percentage should be 
much less than 50 percent. One commenter suggested as an appropriate 
threshold that the sum of equipment replacement costs for a single 
process unit over any period of 5 consecutive years should not exceed 
50 percent of the replacement value of the process unit. Another 
commenter said the replacement percentage should not be higher than 25 
percent. Another commenter suggested a replacement percentage of 5 to 
10 percent to reduce the risk of replacement of an entire process unit 
over time without installation of BACT. One commenter said a more 
appropriate percentage for electricity producers is 0.1 to 1.0 percent. 
Another commenter said the threshold should be 5 percent, 1 percent, or 
even less, as shown by an NSR enforcement case against the Tennessee 
Valley Authority (TVA).
    Another commenter believed the 50-percent number has no practical 
effect in protecting public health and the environment, and the 
commenter was not aware of any projects that have exceeded 50 percent 
in cost.
    While opposed to the ERP in general, one commenter said the cost 
threshold should be as high a percentage as possible, so as not to 
promote premature replacement of equipment that is repairable. Another 
commenter said the 50-percent number from the NSPS is archaic and not 
environmentally protective. This commenter suggested that the threshold 
instead be 24 percent. The commenter believed this lower percentage is 
appropriate because the lifetime of high-cost materials will 
considerably exceed 5 years.
    We agree with those commenters who see a relationship between 
establishing a threshold for equipment replacements that we will treat 
as RMRR under the major NSR program and the threshold the NSPS program 
established for reconstruction. However, we disagree that these two 
thresholds should be the same. The NSPS threshold was intended to 
identify those activities that, even though they did not qualify as a 
modification under NSPS, nevertheless are of such magnitude that 
further consideration should be given as to whether they are projects 
tantamount to new construction. The 50-percent NSPS threshold is not a 
bright line in the sense that all projects that exceed 50 percent are 
automatically considered as reconstruction. Rather, as discussed above, 
it is a threshold intended to alert permitting authorities to 
significant projects and allow case-by-case decisions based on a series 
of regulatory factors.
    The ERP replicates the NSPS concept in some ways. It identifies a 
threshold below which there is no need for further inquiry into whether 
an activity qualifies for the ERP and above which there is a need for a 
case-by-case determination. The major difference between the ERP and 
the NSPS reconstruction test is that the ERP deals with modifications, 
not reconstructions. This difference weighs in favor of establishing 
the equipment replacement threshold at something less than the 
reconstruction threshold. It is logical and practical to conclude, as 
some of the commenters do, that by using the word ``modification'' the 
CAA intended to capture activities on a smaller scale than 
reconstructions. As noted above, we have set the ERP cost threshold at 
20 percent. This value is less than one-half of the 50-percent 
reconstruction threshold and, therefore, fits well within this 
conceptual framework.
    A 20-percent cost threshold would be consistent with the decision 
of the U.S. Court of Appeals for the Seventh Circuit in the Wisconsin 
Electric Power Company v. Reilly (``WEPCO'') case, to the extent that 
it would not automatically allow the activities performed there to 
constitute RMRR. See 893 F.2d 901 (7th Cir. 1990). This court decision 
directly addressed the question of what level of ``like kind'' 
replacement activities qualify as changes under the major NSR program.
    In the WEPCO case, the Court considered an activity involving 5 
coal-fired units at WEPCO's Port Washington plant. Each unit was rated 
at 80 megawatts of electrical output capacity. The activity involved 
the replacement of numerous major components. The information submitted 
by WEPCO showed that the company intended to replace several components 
that are essential to the operation of the Port Washington plant. In 
particular, WEPCO sought to replace the rear steam drums on the boilers 
at units 2, 3, 4, and 5. According to WEPCO, these steam drums were a 
type of ``header'' for the collection and distribution of steam and/or 
water within the boilers. WEPCO viewed their replacement as necessary 
to continue operation of the units in a safe condition. In addition, at 
each of the emissions units, WEPCO planned to repair or replace several 
other integral components, including replacement of the air heaters at 
units 1, 2, 3, and 4. WEPCO also planned to renovate major mechanical 
and electrical auxiliary systems and common plant support facilities. 
WEPCO intended to perform

[[Page 61257]]

the work over a 4-year period, utilizing successive 9-month outages at 
each unit. The cost of the activity was estimated in 1988 to be $87.5 
million. The Court noted that EPA concluded at the time this activity 
was unprecedented in that EPA did not find a single instance of 
renovation work at any electric utility generating station that 
approached this activity in nature, scope and extent. The Court 
determined, at our urging, that the changes did constitute a ``physical 
change'' under the NSR rules.
    In the case of a steam electric generating facility, the process 
unit definition provided in today's rule is nearly identical to the 
make-up of the ``comparable new facility'' that was used in the NSPS 
evaluation of the WEPCO renovation project. However, under our rule we 
would not include the cost of pollution control equipment in 
determining the replacement cost of the WEPCO process units. WEPCO had 
electrostatic precipitators on each of its 5 process units, which our 
rule would subtract from the replacement cost. In addition, the WEPCO 
evaluation dealt with 5 boilers, each with its own turbine-generator 
set; to be consistent with today's definition of steam electric 
generating facility, we would likely treat each boiler unit as 
belonging to a different process unit. However, since all of the 
boilers underwent similar renovations, for simplicity we can assume 
that all of the process unit-specific activity costs are equivalent.
    Using 1991 dollars, consistent with the timeframe of the Seventh 
Circuit Court's decision, it appears that the value of the 5 process 
units at the 400-megawatt WEPCO Port Washington facility would be 
approximately $321 million based on 1991 model plant values provided by 
the International Energy Agency. The 1988 project cost of $87.5 million 
scaled up to 1991 dollars would have had an adjusted project cost of 
$92.3 million.\9\ Thus, the capital cost percentage for the replacement 
activities at WEPCO, averaged over its 5 process units, amounted to 29 
percent. Alternatively, using the project cost of ``at least $70.5 
million'' cited in the 1991 decision by the Seventh Circuit, and using 
the same value for process unit cost, we compute at least 22 percent. 
The 20-percent threshold is, therefore, beneath the scope of the 
activities at issue in the WEPCO case and hence not inconsistent with 
that decision.
---------------------------------------------------------------------------

    \9\ Using the Chemical Engineering magazine's Annual Plant Cost 
Index (composite), $87.5 million in 1988 dollars is equal in real 
terms to (361.3/342.5) multiplied by 87.5 million, or $92.3 million 
in 1991 dollars.
---------------------------------------------------------------------------

    The 20-percent threshold also is supported by available data for 
the electric utility sector. We have a robust and detailed set of 
information available on maintenance, repair and replacement activities 
for the electric utility sector. Information about the electric utility 
sector persuades us that we have established the right ERP threshold 
for this sector.
    Information on other industrial sectors beyond electric utilities 
(as well as general economic theory) further supports our 20 percent 
bright line test. Case studies performed by an EPA contractor and 
included in Appendix C of our final regulatory impacts analysis (RIA) 
estimate the overall impact of the rule on six different industrial 
sectors (pulp and paper mills, automobile manufacturing, natural gas 
transmission, carbon black manufacturing, pharmaceutical manufacturing, 
and petroleum refining). The case studies find that routine equipment 
replacement activities generally do not cause emissions increases. The 
case studies also find that equipment replacement activities vary 
widely within these industries. Likewise, the cost of these activities 
as a percent of the process unit replacement value varies widely. We 
recognize that the study addresses specific case examples from only a 
part of regulated industry and that the project cost information is 
derived from a limited inquiry of industry representatives. We believe, 
however, that the study provides a useful scoping assessment that tends 
to support the proposition that the 20 percent threshold derived for 
the utility industry (which is based on robust industry data) should be 
applied to industry as a whole. In short, the study supports our view 
that it is reasonable to assume that equipment replacement activities 
in the utility industry are similar enough to replacement practices in 
other industry that the 20 percent value determined for utilities, is 
appropriate for industry as a whole. This data indicates that most 
typical replacement activities will fall within the 20-percent 
threshold. At the same time, the data indicates that some major 
replacement activities likely will cross the 20-percent threshold and 
will require a case-by-case evaluation under the multi-factor RMRR 
test.
    Two comment letters (from the Utility Air Regulatory Group (UARG) 
and from the American Lung Association (ALA), et al.) were particularly 
helpful in understanding the issues associated with the electric 
utility sector. The UARG provided as an attachment to its comment 
letter a document describing major repair and replacement activities 
that its members believe must be undertaken at utility generating 
stations in order to keep those facilities operational. The UARG noted 
that capital costs incurred for repair and replacement activities at an 
individual process unit additionally include activities more minor than 
those addressed in the document. The UARG grouped repair and 
replacement activities into project families; within each project 
family were per-component costs ($/kW) for numerous equipment 
replacement activities. We have reviewed the list of projects supplied 
by UARG and have concluded that these types of replacement activities 
are important to maintaining, facilitating, restoring or improving the 
safety, reliability, availability, or efficiency of process units. 
Therefore, generally speaking, these types of individual activities and 
groups of activities should qualify for the ERP and be excluded from 
major NSR without case-specific review. We also believe that it is 
reasonably expected in the electric utility industry for groups of 
these activities to be implemented at the same time. Such groupings 
should also be excluded without case-specific review. When we compare 
the 20-percent ERP cost percentage to the UARG data, we find that 
individual replacement activities would, in fact, qualify for the ERP 
and that limited groupings of these activities would qualify. However, 
larger groupings of these activities--groupings that are not usually 
seen in the industry--would not qualify for the ERP. This shows that 
the 20-percent threshold will be effective in distinguishing between 
activities (and aggregations of activities) that should not require 
case-specific review to be excluded from major NSR and those that do.
    The ALA commenters provided with their comments the results of 
their analysis of projects at issue in an NSR enforcement case against 
Tennessee Valley Authority (TVA). As shown in the ALA comment letter, 
the Clean Air Task Force and the Natural Resources Defense Council 
looked at costs for 14 projects on a process unit basis, in year 2001 
dollars, from the publicly available record for the case. For all but 
one of the challenged projects, the ALA commenters calculated a cost of 
less than 4 percent of process unit replacement cost. The ALA 
commenters submitted results of this analysis with their opposition to 
a source-wide, 5-percent maintenance allowance. As noted above, we 
concluded in our 2002 report to the President that the NSR

[[Page 61258]]

program--and the RMRR provision in particular--has in fact resulted in 
delay or cancellation of activities that would have maintained and 
improved the reliability, efficiency, and safety of existing energy 
capacity. The primary purpose of today's rule is to rectify this 
problem. Thus, to the extent the activities addressed by ALA qualify 
for the ERP, we now believe that such activities, if conducted in the 
future, should be excluded from major NSR.
    A final factor that we believe supports our selection of a 20 
percent threshold is the cost of installing state-of-the-art controls 
on existing units. There is obviously no single answer to the question 
of at what point that cost becomes the deciding factor in an owner's 
decision whether to replace a piece of equipment and incur that cost, 
since much will depend on the rate of return on the investment. 
Nevertheless, we think it is reasonable to assume that if the cost of 
the controls is greater than the cost of the replaced equipment, it is 
likely to operate as a substantial deterrent to replacing the equipment 
at issue. That is likely to be the case with respect to electric 
utilities if we set the threshold below 20 percent, which represents 
the approximate cost of retrofitting existing plants with state-of-the-
art controls. The equation is similar for industrial boilers. Notably, 
those sectors represent a substantial fraction of the emissions 
potentially subject to the NSR program. While the relative costs of air 
pollution controls in other industries vary more widely than the costs 
for utility and industrial boilers, we nevertheless believe that the 
costs and technical issues associated with retrofitting air pollution 
controls factor significantly into equipment replacement decisions.

D. What Will Be the Basis of Applying the 20-Percent Threshold?

    In the proposal, we solicited comment on whether implementing the 
ERP on a per-activity basis or on some other reasoned basis, such as 
applying the percentage to components that are replaced collectively 
over a fixed period of time, may be more workable.
    Many commenters stated that the ERP should be implemented on a per-
activity (or aggregation of activities) basis. Two of the commenters 
cited longstanding NSR precedent as the basis of their comments, while 
two other commenters relied on NSPS precedent. Another commenter 
thought the per-activity approach would be less confusing than summing 
activities over a fixed period of time. Other commenters believed the 
equipment replacement threshold should in fact be applied on a 5-year 
rolling average.
    We have decided to apply the percentage threshold on a per-activity 
(or aggregation of activities) basis. This is consistent with how major 
NSR has been applied in the past and will continue to apply in the 
future, with the exception of those sources which establish a PAL. The 
major NSR program is a preconstruction program that requires 
applicability to be determined for a given activity at a facility and, 
as necessary, permitting to occur prior to the time activities are 
commenced. The major NSR program also requires applicability to be 
determined, in the first instance, based on an assessment only of the 
parts of a facility involved in the activity. A per-activity basis 
works well with this approach. We are not going final with a 
``component-by-component'' approach that we solicited comment on 
through our RMRR proposal.
    There would be obvious problems if we chose any of the other 
approaches suggested in the proposal or suggested by commenters (for 
example, annual basis or 5-year rolling average). One of the primary 
concerns with applying the percentage to activities performed over a 
span of time is that we would be restructuring the major NSR program to 
operate based on after-the-fact determinations. This raises the 
difficult question of what happens under this type of approach if you 
learn after commencement of an activity that it does not qualify under 
the ERP. This situation is largely avoided by the per-activity approach 
that we are establishing in today's rule.
    It should be noted that activities that are related must be 
aggregated under the ERP, in the same way as they would have to be 
aggregated for other NSR applicability purposes. Under our current 
policy of aggregation, two or more replacement activities that occur at 
the same time are not automatically considered a single activity solely 
because they happen at the same time. For example, a steam turbine 
rotor replacement project and a boiler tube replacement project would 
not be aggregated simply because they occur during the same maintenance 
outage and on the same process unit. Further inquiry into the nature of 
the activities and their relationship to each other is needed before 
deciding whether the activities must be aggregated under NSR. Also, 
non-replacement activities that are part of a larger replacement 
activity should be included when calculating costs for a replacement 
activity against the capital cost threshold.

E. What Basic Design Parameters Are Being Established To Qualify for 
the Equipment Replacement Provision?

    In the proposal, equipment replacements were only eligible for the 
ERP if they did not change the basic design parameters of the process 
unit. We proposed that maximum heat input and fuel consumption 
specifications for EUSGUs and maximum material/fuel input 
specifications for other types of process units are basic design 
parameters. We solicited comments on limiting the eligibility of the 
ERP this way and on the basic design parameters we proposed.
    Several commenters expressed concerns with either the use of these 
specific parameters, or the restriction of the regulated community to 
only this set of design parameters. Other comments centered around an 
inconsistency in how EPA has accounted for efficiency in the basic 
design parameter safeguard. The commenters stated that, while EPA 
stated in the proposed preamble that efficiency is not a basic design 
parameter, the basic design parameter safeguard, as proposed, has the 
potential to bar equipment replacements that achieve significant gains 
in efficiency.
    Commenters from all sides supported EPA's approach to handling 
activities intended to improve an affected process unit's performance 
beyond its basic design parameters. Commenters asserted that these 
actions would not fall within the RMRR exclusion. Commenters from the 
gas transmission industry concurred and amplified this concept, stating 
that an engine that is ``uprated'' at the time of overhaul should not 
be excluded from major NSR under the RMRR exclusion.
    We recognize that the proposed basic design parameters are 
inconsistent with some industry conventions, and that we should allow 
for industry-specific flexibility or specify additional source 
category-specific parameters. For example, for natural gas transmission 
compressor stations, commenters explained that brake horsepower is the 
conventional design capacity parameter. We received similar comments 
from other industries, including cement and surface coaters, who 
objected to limiting their facilities to the proposed basic design 
parameters. Accordingly, we have decided to provide flexibility by 
providing a menu of choices from which the owners or operators may 
select and also by allowing for owners or operators to propose 
alternative basic design parameters to their reviewing authority which 
would then be made legally enforceable.

[[Page 61259]]

    In addition to this flexibility, there may be a need for additional 
flexibility in using the basic design parameters that are spelled out 
in today's rule. For instance with boilers, maximum steam production 
rate is often used by the industry, and it may make sense in some cases 
to set the design parameters based on those values rather than on 
maximum heat input. Likewise, a crude oil distillation tower may have 
several capacities that are a function of the type of crude that is to 
be processed, and so a refiner may need to have a set of basic design 
parameters for its crude towers. These situations can be addressed by 
the source proposing alternative parameters or sets of parameters to 
their reviewing authority.
    Also, there should be flexibility in how the basic design 
parameters are demonstrated when the owner or operator chooses not to 
rely on the design information for its process unit. For example, in 
order to establish the heat input value that the process unit has 
demonstrated it is capable of achieving, an electric generating unit 
should have the flexibility to reference available credible 
information, such as results of historic maximum capability tests or 
engineering calculations. Results from tests performed by electric 
utilities in the context of providing assurances to generation dispatch 
systems and regional or national power pools may be used to establish 
the process unit's maximum heat input. A review of such data or other 
available operational data or design information can reveal the heat 
input that the process unit is capable of achieving in its ``pre-
activity'' configuration, and this can be compared to a ``post-
activity'' heat input value. Plant operators, where the specified basic 
design parameters are inappropriate for the process, can propose what 
the measure of performance will be for these process units, including 
the use of permit limits on amount of production, to their reviewing 
authority. For process units having multiple end products and raw 
materials, the owner or operator should consider the primary product or 
primary raw material when selecting a basic design parameter.
    Many pieces of equipment are purchased based on their capacity or 
output. Consequently, for both utilities and non-utilities, we have 
modified the proposed basic design parameters to include output-based 
alternatives in today's final rule. For utilities, the owner or 
operator can select maximum hourly electric output rate and maximum 
steam flow rate as its basic design parameters, as an alternative to 
using input-based measures of maximum hourly fuel consumption rate and 
maximum hourly heat input. (We are clarifying from the proposal that 
the correct parameter is maximum hourly heat input, not maximum heat 
input.) Owners or operators may set different design parameters for 
different fuel types (such as coal or oil) or a combustion device that 
can accommodate multiple fuel types: for coal-fired units, owners or 
operators should consider that the fuel consumption rate will vary 
depending on the quality of the coal for a given heat input. When 
establishing fuel consumption specifications in terms of weight or 
volume, the minimum fuel quality based on BTU content should be used 
for coal-fired units.
    Regardless of whether the source selects a basic design 
parameter(s) specified for non-utilities in today's rule or gets 
approval from their reviewing authority to use an alternative 
parameter(s) for any type of source, we have not specified a fixed 
averaging time period for the circumstance because we want the owner or 
operator to have the flexibility to select an averaging time that best 
accommodates their operation. In most cases, we believe that long term 
averaging periods (e.g., a 12-month fixed period) will not be 
appropriate.
    Thus, an equipment replacement that improves a process unit's 
efficiency and thereby enables the unit to return to its design 
parameters can qualify as RMRR even if current actual emissions 
increase as a result. For example, if boiler tubes or refractories are 
replaced on a boiler process unit, and these activities are beneath the 
capital cost threshold and are within the unit's basic design 
parameters, then they would qualify as RMRR under the ERP even if this 
improves the unit's efficiency.
    The manufacturer's design parameters of a process unit are always 
acceptable if an owner or operator chooses to rely on them. In the rare 
cases where a facility does not have established design parameters, we 
believe that a reasonable look back period should be used for 
establishing the pre-activity values for basic design parameters, 
rather than taking the condition of the process unit immediately before 
the activity. We have therefore established a 5-year look back period, 
consistent with that for the NSPS hourly emissions increase test, for 
these situations.
    We were urged by some commenters to incorporate a de minimis 
increase level in the basic design parameters that would allow 
activities to qualify for the ERP even though the activities would 
result in a minor change to the relevant basic design parameters. They 
argued that some effects resulting from the replacement may not be 
apparent before the equipment has been replaced. They argued that 
allowing for small changes in basic design parameters would add greater 
certainty to the ERP because unforeseen small changes would not cause 
an activity to lose the exclusion after the fact. While we sympathize 
with the commenter's concern, we do not see a ready solution to this 
problem under the RMRR exclusion. In fact, we are not persuaded that 
those types of changes can be readily justified under the ERP because 
it is hard to see how an activity that causes basic design parameters 
to change is not ``a change'' under NSR.
    In sum, we continue to believe that an identical or functionally 
equivalent replacement should not qualify for the ERP if the activity 
causes the process unit to exceed its specified basic design 
parameters. Without such a requirement, significant alteration of a 
process unit's fundamental design could be accomplished under the guise 
of the ERP. Such an outcome obviously does not square with the idea 
that identical or functionally equivalent replacements are not 
``changes'' under the major NSR program. Our final rule is different 
from the proposal, however, in that it provides greater flexibility in 
defining basic design parameters for process units. We were persuaded 
by commenters who expressed concerns that the proposed approaches did 
not adequately encompass all affected operations and industry sectors.

F. What Collection of Equipment Should Be Considered in Applying the 
Equipment Replacement Provision and How Should It Be Defined?

    In the proposal, we raised the issue of what collection of 
equipment should be considered in applying the threshold under the ERP. 
We proposed the term ``process unit'' as the appropriate collection to 
accommodate the intended coverage of activities under the ERP. The 
purpose of this term is, to the extent possible, to align 
implementation of the ERP with generally accepted and practical 
understandings of what constitutes a discrete production process. The 
general definition that we proposed was based closely on the definition 
of process unit contained in 40 CFR 63.41 and read as follows:

    Process unit means any collection of structures and/or equipment 
that processes, assembles, applies, blends, or otherwise uses 
material inputs to produce or store a completed product. A single 
facility may contain more than one process unit.

To help illustrate these concepts, we further proposed five industry-
specific

[[Page 61260]]

examples of how this definition of process unit might be applied.
    Some commenters compared the proposal's definition of ``process 
unit'' (``* * * producing or storing a completed product * * *'') to 
the definition that is used by section 112(g) and that appears in 40 
CFR 63.41 (`` * * * producing or storing an intermediate or final 
product * * *''). One of the commenters supported the proposed 
definition. Two commenters said the rule's definition should be 
consistent with that used by section 112(g), which they believe is 
broad enough to encompass interrelated operations. While supporting the 
RMRR proposal's definition, two commenters recommended that EPA provide 
regulatory flexibility by allowing a facility the option to choose 
which definition it will use.
    One commenter generally supported the proposed definition of 
``process unit,'' but this commenter believed that ``the delineation of 
a process unit should be made by regulated entity rather than 
explicitly defined in a rule.''
    Three commenters asserted that pollution control equipment should 
be included in the process unit definition. One industry commenter said 
pollution control equipment is often integral to the process and may 
produce an intermediate product. One environmental commenter believed 
the proposed rule was unclear as to whether pollution control equipment 
is part of the process unit.
    Several commenters said the proposed definition is too vague or 
broad. Another commenter urged EPA to change the definition of process 
unit to limit the scope of what is allowed in the ERP, so that the 
source of emissions (for example, an entire coal boiler) would not be 
allowed to be replaced without major NSR. The commenter asserted that 
the replacement unit's scope should be limited to an emission unit.
    Most commenters agreed that the general process unit definition is 
sufficient. However, a number of commenters suggested that we revise or 
eliminate some of the process unit examples (that is, the industry 
category-specific definitions), and others were concerned that the 
proposed definitions do not support the detailed process unit 
definition for a specific industry because the definitions will never 
capture all possible elements and configurations.
    We received comments from several industry representatives 
suggesting changes to our proposed industry-specific definitions, and 
also to request that we delineate other process unit types explicitly 
in the rule. Definitions were submitted for sugar mills, chemical 
manufacturing plants, surface coating operations, flat glass 
manufacturing, fiberglass manufacturing, and gas compressor stations.
    One industry commenter agreed with our proposed approach to 
proportionately allocate, based on capacity, the cost of those 
components shared by two or more process units. Another commenter 
suggested that, for electric utilities, we allocate the cost of shared 
equipment based on a pro rata share of megawatts produced.
    We agree with the commenters who favor using a process unit as the 
basis for administering the ERP and including a definition of process 
unit in the final rule. We also agree with the commenters who suggested 
that the definition of process unit should be consistent with the 
definition in 40 CFR 63.41, and we have altered the final rule 
definition to include those processes that produce ``intermediates.'' 
We acknowledge that, without further explanation, the term 
``intermediates'' is susceptible to misinterpretation, which can cause 
confusion and lead to less regulatory certainty. Thus, we provide the 
following explanation as to how we intend to interpret today's rule.
    By ``intermediates,'' we mean the intended product of an integrated 
facility operation. For example, for an automotive manufacturing plant, 
while the completed product would be the driveable vehicle ready for 
shipping to the showroom, an intermediate product could be the engine 
or the painted body shell. In this case, we would not consider smaller 
production operations, such as the e-coat, primer surface, or top coat 
operation, to be intermediates in the context of our final rule 
definition for process unit. Our primary goal in defining this term 
``process unit'' is to encompass integrated manufacturing operations 
that produce a completed product, and those operations that produce an 
intermediate as the product of the process unit. In the case of the 
automotive paint shop, series of coating steps together comprise the 
carefully designed and interrelated set of operations, all of which are 
needed to provide a coating system that meets design specifications. 
The individual operations almost never are implemented individually 
and, as a practical matter, simply would serve no meaningful purpose in 
the absence of the others.
    We disagree with the commenters who wish to include all pollution 
control equipment in the definition of process unit. We feel that 
periodic replacement of components of emissions control equipment 
should be encouraged and would rarely lead to actual emissions 
increases. In instances where identical or functionally equivalent 
replacement of pollution control equipment occurs, it is likely you 
will qualify for a Pollution Control Project exclusion. We do agree, 
however, that where the control equipment is an integral component of 
the process it should be included. Therefore, we are excluding 
associated pollution control equipment from the definition of the 
``process unit,'' except for control equipment that serves a dual 
purpose in the process. We know there are industries where pollution 
control equipment performs a dual purpose; for example, condensers 
often serve to control emissions of organic air pollutants while 
serving as an integral component of the operation of a fractionation 
column. A low-NOX burner is another example of a dual-
purpose component. In such cases, to provide clarity and simplify 
administration of the ERP, our rule provides that dual purpose 
equipment should be considered part of the process. We are also 
clarifying in today's rule that administrative buildings (including 
warehousing) are not to be included in the process unit, but other 
types of non-emitting units that are integral to the processing 
equipment should be included.
    We also have included in our final rule industry-specific examples 
of how this definition might be applied. The examples are drawn from 
three selected industrial processing categories--electric utilities, 
refineries, and incinerators. We proposed each of these detailed 
definitions and received mostly support from commenters on their 
accuracy. While we also proposed detailed definitions for two other 
industries--pulp and paper and cement producers--we have decided not to 
finalize those definitions after receiving comments from the relevant 
industry trade association asserting that the definitions did not, and 
could not, capture all of their industry's configurations and they 
believed the generic process unit definition was sufficient for their 
industry. Because of the centrality of the ``process unit'' concept to 
the usefulness of the ERP, it is our desire to include specific 
definitions for steam electric generating facilities, petroleum 
refineries, and incinerators in the final rule to provide as much 
certainty as possible for facilities in these industries. As noted 
above, these definitions also should be useful for those in other 
industries who

[[Page 61261]]

will apply our general definition because the industry specific 
definitions provide clear examples of how we intend the general 
definition to be interpreted and applied. During the public comment 
period on the proposal, several commenters submitted additional 
industry specific definitions and asked us to put them in the final 
rule. We are not finalizing these suggested definitions at this time, 
because we did not include them in the proposed rule. However, provided 
below are the process unit definitions that commenters submitted to us 
and that we think comport well with the general definition of process 
unit promulgated today.
    [sbull] For a natural gas compressor station, each compressor 
system, together with its proportionate share of common support 
equipment is a separate process unit. This would generally consist of 
the air inlet system, accessory drive system, gas producer, fuel 
delivery system, cooling system, lube system, power turbine, power 
shaft, control system, starting system, exhaust system, and support 
facilities (e.g., auxiliary power generating equipment, heating/cooling 
equipment, station and yard pipe, valves, etc.).
    [sbull] For a flat glass manufacturing plant, each production line 
within a facility should be a separate process unit. Flat glass 
production is completed on a continuous line where raw materials are 
added at one end, a continuous ribbon of glass is formed, and finished 
glass is packaged at the other end. The flat glass production line 
consists of: the batch house, where raw materials are stored and 
weighed; the furnace and refiner, where the raw materials are melted; 
the bath, where the glass ribbon is formed; the lehr, where the ribbon 
is annealed; and the cutting and packaging equipment, where the glass 
is removed from the line for sale to customers or for additional 
processing later.
    [sbull] For a fiberglass production facility, each production line 
is a separate process unit. Fiberglass is manufactured on a continuous 
line where raw materials are melted at one end to form a continuous 
strand of fiberglass that is packaged at the other end. The fiberglass 
production line begins with the batch house, where raw materials are 
stored and weighed. In the melter, forehearth, and refiner, the raw 
materials are melted and refined. From the refiner, glass fibers are 
formed through controlled bushings. From the bushings, the continuous 
strand fibers are either directly cut or packaged or wound onto spools 
for packaging for sale to customers or for additional later processing.
    [sbull] For the production of precipitated amorphous silica, the 
process unit includes, but is not limited to: raw material storage and 
handling equipment used for mixing sand and other raw materials prior 
to addition to the furnace; the furnace itself; the raw material 
storage and handling equipment for the cullet dissolving and silica 
precipitation process; all dissolving, precipitation, and filtration 
tanks and equipment; and drying equipment. Further, the process unit 
includes all the product packaging, storage, handling, and transfer 
equipment.
    [sbull] For a chemical manufacturing plant, the process unit would 
include all the equipment assembled and connected by pipes or ducts to 
process raw materials and to manufacture an intended primary product 
and associated byproducts or intermediates. The process unit can 
consist of more than one unit operation. Chemical manufacturing process 
units may include, but are not limited to: raw material storage, and 
air oxidation reactors and their associated product separators and 
recovery devices; reactors and their associated product separators and 
recovery devices; distillation units and their associated distillate 
receivers and recovery devices; associated unit operations; associated 
recovery devices; and any feed, intermediate and product storage 
vessels, product transfer racks, and connected ducts and piping. A 
chemical manufacturing process unit includes pumps, compressors, 
agitators, pressure relief devices, sampling connection systems, open-
ended valves or lines, valves, connectors, instrumentation systems, and 
process control or dual purpose air pollution control devices or 
systems. For a chemical manufacturing facility, there are several types 
of process units: those that separate and distill raw material 
feedstocks; those that change molecular structures through reactions or 
polymerization; those that ``finish'' the reacted or polymerized 
product, through compounding, blending, or similar operations; 
auxiliary facilities, such as boilers and by-product fuel production; 
and those that load, unload, blend, or store products. Process 
equipment that acts to control emissions, such as condensers, recovery 
devices, and oxidizers, is considered part of the process unit.
    We note that we were unable to include some other process unit 
definitions submitted by commenters. While we do not believe that these 
other proposed definitions were necessarily inconsistent with our 
general definition of process unit, we had concerns and questions with 
some of these proposed definitions. We believe that now that this rule 
is issued, we can more fully evaluate those other definitions, 
including communicating with the leading industry officials, and 
determine whether we would approve of their use.
    Finally, we have made some slight corrections to the process unit 
definitions that we proposed based on comments we received on the 
proposed definitions.
    There are numerous industries that have industrial boilers at their 
facility to provide electricity and steam to their operations. As a 
general rule, we would expect these boilers to be treated as a separate 
process unit from the other unit operations occurring at the facility. 
We would expect the boundaries of the process units for such boilers to 
be consistent with the boundaries established under the definition for 
a steam electric generating facility in today's rule, which encompasses 
all equipment from coal handling to the emission stacks.
    We also decided to continue to require that owners or operators who 
have components shared by two or more process units to proportionately 
allocate, based on capacity, the cost of those components. And we agree 
with the commenter that an equitable approach for electric utilities 
having components shared by two or more process units is to allocate 
the cost of shared equipment based on the pro rata share of megawatts 
produced by each process unit.

G. Consideration of Non-Emitting Units as Part of the Process Unit

    Many commenters supported excluding non-emitting equipment from the 
ERP. One commenter stated that triggering the major NSR review process 
for maintenance activities is an impediment to continuous improvement 
projects for certain products and processes, even if actual emissions 
decrease or only non-emitting units on the process line are affected. 
Delays or postponements of project maintenance work adversely affect 
the reliability, safety and productivity of operations and cost control 
efforts. Another commenter recommended that work at clearly non-
emitting units, specifically including foundation regrouting and repair 
and frametop replacement, should be excluded from this rule. Three 
commenters believed that non-emitting units cannot result in an

[[Page 61262]]

increase of emissions and thus do not need to be evaluated under major 
NSR.
    A blanket exclusion for non-emitting units could create problems of 
interpretation because the term ``non-emitting components'' is 
ambiguous when considering certain components. Commenters asserted that 
identifying and separating out non-emitting components can be a complex 
undertaking, and may be contrary to the goal of a clear and 
straightforward option. One commenter provided the following examples: 
(1) Piping systems (although pipe connectors are a source of fugitive 
emissions, the pipe normally is not); and (2) structural supports for a 
process unit (separating out the cost of supports from an investment 
basis throughout a facility will be difficult).
    Another commenter believed it would be difficult to separate the 
costs of emitting and non-emitting equipment when determining the cost 
of the process unit. The commenter also believed it would be difficult 
to determine allocation of shared equipment in the cost analysis.
    We are concerned that, if owners or operators were allowed to strip 
away all of the non-emitting components from a process unit definition, 
it would create significant ambiguity in the rule and could result in 
significant variation in how the rule is applied to similar sources in 
different jurisdictions. In addition, we simply do not think it is 
practical or logical to separate ``non-emitting'' components of a 
process unit from ``emitting'' components. We believe that integrated 
manufacturing operations (that is, process units) typically include 
both types of equipment. Separating emitting from non-emitting 
equipment would create an artificial divide that contrasts sharply with 
physical and operational reality.
    As noted above, however, we do believe that a distinction should be 
made between non-emitting equipment that is part of a process unit and 
non-emitting equipment that is functionally distinct from the process 
unit. For example, most production facilities have buildings or space 
to house administrative offices, such as offices for the plant 
accounting staff. Such non-emitting facilities should not be considered 
part of any process unit under today's rule.

H. What Is the Accounting Basis for the Process Unit?

    In the proposal, the accounting basis for the ERP discussed was the 
same as for the NSPS reconstruction provision, which is the fixed 
capital cost that would be required to construct an entirely new unit. 
We also discussed for the annual maintenance, repair and replacement 
allowance using the invested cost of a unit as the accounting basis. We 
proposed that it would be appropriate to require that costs be 
calculated using an approach along the lines set out in the EPA Air 
Pollution Control Cost Manual (http://www.epa.gov/ttn/catc/dir1/c_allchs.pdf). Finally, we solicited comment on whether the costs 
associated with the unanticipated shutdown of equipment, due to 
component failure or catastrophic failures such as explosions or fires, 
should be included in evaluating costs under the ERP.
    In reviewing comments, we recognized that some commenters appeared 
to direct their comments on the accounting methods at the annual 
maintenance, repair and replacement allowance, and not necessarily the 
ERP. Often, we came to this conclusion simply by the way the commenters 
organized their comments, and not by any specific statements in the 
comment letter. However, since we asked for comment on the accounting 
approaches as they would be applied to both the annual maintenance, 
repair and replacement allowance and the ERP, we believe that comments 
that appeared to be dedicated to the annual maintenance, repair and 
replacement allowance should also apply to our evaluation of the 
accounting for the ERP, except in the case where the commenter 
specified that their comments on the proposed accounting methods 
applied only to the annual maintenance, repair and replacement 
allowance or the ERP. Likewise, for considering whether costs 
associated with unanticipated shutdown of equipment, we considered the 
comments to apply to both the ERP and the annual maintenance, repair 
and replacement allowance unless the commenter specifically noted that 
the comment should not be applied to both of the proposed rule 
provisions.
    Most commenters asked for flexibility on whether a facility should 
use replacement value, invested cost or insurance valuation as the 
basis for the calculations. They felt that all were of equal merit and 
different ones would be available at different facilities so EPA should 
not prescribe only one type.
    Most commenters did not support the sole use of the EPA Air 
Pollution Control Cost Manual (APCCM) to standardize calculations for 
replacement and repair costs for RMRR in general. Most commenters felt 
that the APCCM is a worthy reference for costing but also that sources 
should not be limited to only one manual, because a single manual is 
likely to have shortcomings and not be able to represent every 
situation.
    Many commenters supported an exclusion of costs for unanticipated 
shutdowns and failures. They noted that strong incentives exist to 
avoid fires, explosions and other unanticipated equipment failures 
because of the risk of human injury and production interruptions and 
because of the expense involved in restoring lost capacity. As a 
result, they contend that a catastrophic event already penalizes the 
facility dramatically, but then to impose the case-by-case analysis 
would only exacerbate their troubles. They explained that failures take 
place occasionally and can result in a sudden, unplanned partial or 
total loss of equipment. When such a failure occurs at a natural gas 
compressor station, the turbine or engine concerned must be replaced 
immediately to avoid a disruption in gas supply. Other facilities may 
have similar pressures to maintain their product around the clock. Such 
replacement fits easily within most elements of the equipment 
replacement test. Commenters asserted that replacing a catastrophically 
failed turbine or engine is clearly ``routine,'' since companies will 
always replace such failures.
    Other commenters, however, opposed an exclusion for unanticipated 
shutdowns and failures on the grounds that maintenance activities 
performed during forced outages are simply maintenance and should be 
considered as such, particularly given that the proposed RMRR rule 
approaches and the December 2002 final rules already have given the 
industry a number of exclusion options.
    We are allowing sources to determine the applicability of today's 
rule on the basis of replacement value, with an option for sources to 
notify their reviewing authority in writing if they desire to use 
another option (for example, invested cost or insurance value where the 
insurance value covers only the complete replacement of the process 
unit). The equipment replacement cost should be based on the current 
replacement value of the entire process unit at the time of conducting 
the activity.
    Typically, replacement value is more easily obtained than invested 
cost. Most manufacturers will have information concerning the 
replacement value of a process unit, because such costs are commonly 
used when evaluating various business scenarios relating to 
manufacturing costs. Also, use of replacement value is consistent with 
the NSPS provisions.

[[Page 61263]]

    In addition to determining the replacement value of a process unit, 
in our final rule we allow for the use of several other accepted 
methods in different industries for estimating such values. Replacement 
values are the estimated value of replacing a unit and can be based on 
a current appraisal. In lieu of replacement cost, you can also use 
inflation-adjusted original investment, insurance limits if insured for 
full replacement of the unit, or other cost estimation techniques 
currently employed by the company, as long as the company follows GAAP 
and if approved by the reviewing authority.
    A dollar-per-kilowatt rate for calculating costs may be appropriate 
for utilities. This model is specific to source and fuel type and is 
updated periodically. We allow sources to use insurance valuation 
methods such as the Handy-Whitman Index to determine replacement costs 
for electric utilities. Other sources to compute costs include the 
Nelson Refinery Construction Index Factors, Solomon Refinery Study, and 
licensors of the respective process unit (e.g., Kellogg, UOP).
    In order for a cost-based approach to be equitable, all owners or 
operators must include the same categories of expenses in both the 
process unit replacement value and the replacement activities sought to 
be excluded. Therefore, although the final rule does not mandate any 
particular approach, we believe it is generally appropriate to 
calculate costs using an approach similar to the elements of Total 
Capital Investment as defined in the APCCM. While the manual contains 
basic concepts that could be used to estimate total capital investment 
at a process unit, it is geared toward cost calculations for add-on 
control equipment. On the other hand, the underlying concepts are taken 
from work done by the American Association of Cost Engineers to define 
the components of cost calculations for all types of processes, not 
just emission control equipment. In certain cases, other manuals might 
make more sense depending on their circumstances.
    Under the APCCM, total capital investment includes the costs 
required to purchase equipment, the costs of labor and materials for 
installing the equipment (direct installation costs), costs for site 
preparation and buildings, and certain other indirect installation 
costs. However, any costs that are part of the installation and 
maintenance of pollution control equipment should be excluded from the 
cost calculation, per our discussion in the previous section of this 
preamble. We believe equipment that serves a dual purpose of process 
equipment and control equipment (combustion equipment used to produce 
steam and to control hazardous air pollutant emissions, exhaust 
conditioning in the semiconductor industry, etc. should be considered 
process equipment.
    Direct installation costs include costs for foundations and 
supports, erecting and handling the equipment, electrical work, piping, 
insulation, and painting. Indirect installation costs include such 
costs as: engineering costs; construction and field expenses (costs for 
construction supervisory personnel, office personnel, rental of 
temporary offices, etc.); contractor fees (for construction and 
engineering firms involved in the activity); startup and performance 
test costs; and contingencies.
    We believe there may be merit to the comments we received 
advocating a categorical exclusion for unanticipated shutdowns and 
failures of some kind. When such an outage occurs, there may be a real 
urgency to restore the plant to operation without forcing it to await 
the results of a permitting action or applicability determination. In 
the past, we have handled these situations with case-by-case consent 
orders; however, even that approach may lead to unnecessary delays. It 
may specifically be sensible to relaxing the 20 percent cost threshold 
limitation for such events because it is unlikely that sources would 
incur an outage to avoid controls. We did not propose such a stand-
alone exclusion and hence we believe we should not act upon it at this 
time.

I. Enforcement

1. Compliance Assurance
    We believe that the records developed and maintained in the 
ordinary course of business will provide the primary means of assuring 
compliance with today's rule. We know that, as a general rule, 
companies necessarily generate and keep records related to the types of 
projects covered by today's rule. For example, companies generally have 
comprehensive procedures by which funds are allocated to both capital 
and maintenance expense projects. Many of the records generated by 
these procedures are needed for tax accounting purposes and, by law, 
must be maintained for at least 6 years. Moreover, additional records 
must be maintained in industries regulated for other purposes, such as 
the energy sector (over 90 percent of which, by capacity, is subject to 
FERC regulation). Public utilities, licensees and natural gas companies 
that are subject to FERC jurisdiction must, unless they receive a 
waiver from the Commission, comply with extensive accounting and record 
retention requirements. They must keep financial information according 
to uniform systems of accounts that are set out in 18 CFR part 101 for 
public utilities and licensees, and 18 CFR part 201 for natural gas 
companies. These uniform systems of accounts include hundreds of 
specific accounts, including individual accounts for boiler plant 
equipment, engines and engine-driven generators, turbogenerator units, 
and hundreds of other asset, liability, cost and property items.
    These companies also must retain records according to the schedules 
set forth in 18 CFR part 125 (for public utilities and licensees) and 
18 CFR part 225 (for natural gas companies). The types of records that 
companies must keep include, for public utilities and licensees, for 
example, generation and output logs (records must be kept for 3 years), 
load records (3 years), gauge-reading reports (2 years), maintenance 
work orders and job orders showing entries for labor, materials and 
other charges in connection with maintenance and other work pertaining 
to utility operations (5 years), work order sheets for construction 
work in progress (5 years), appraisals and valuations made of utility 
property or investments (3 years), engineering records, drawings, and 
other supporting data for proposed or as-constructed utility 
facilities, including detail drawings and records of engineering 
studies (must be kept until facilities are retired), contracts or other 
agreements relating to services performed in connection with 
construction of utility plant (6 years after the plant is retired or 
sold), general and subsidiary ledgers (10 years), paid and canceled 
vouchers, and original bills and invoices for materials, services, etc. 
(5 years).
    Altogether, these various sources of information provide more than 
reasonable assurance of compliance with today's rule. This is 
particularly true given EPA's broad authority to inspect affected 
facilities and require submission of compliance related data. 
Accordingly, we are not imposing any recordkeeping requirements in 
today's rule.
2. General Issues
    Today's rule provides revisions to the major NSR program to specify 
categories of equipment replacement activities that we will consider 
RMRR in the future. As recognized by the U.S. Supreme Court, an agency 
may not promulgate retroactive rules absent express congressional 
authority. See Bowen v. Georgetown Univ. Hosp., 488 U.S. 204,

[[Page 61264]]

208, 102 L. Ed. 2d 493, 109 S. Ct. 468 (1988). The CAA contains no such 
expressed grant of authority, and we do not intend by our actions today 
to create retroactive applicability for today's rule. 42 U.S.C. 7401 et 
seq. Today's rule applies only to conduct that occurs after the rule's 
effective date.
    None of today's rule revisions apply to any changes that are the 
subject of existing enforcement actions that the Agency has brought and 
none constitute a defense thereto. Furthermore, prior applicability 
determinations on major modifications that result in control 
requirements in an NSR permit that currently applies to a source remain 
valid and enforceable as to that source.
    As noted above, today we are changing the scope of the RMRR 
exclusion from the major NSR program by taking final action on the ERP. 
If you subsequently undertake an activity that does not meet the 
applicable provisions of these new alternatives and do not obtain a 
preconstruction permit if you are required to do so, you will be 
subject to any applicable enforcement provisions (including the 
possibility of citizens' suits) under the applicable sections of the 
CAA. Sanctions for violations of these provisions may include monetary 
penalties of up to $27,500 per day of violation, as well as the 
possibility of injunctive relief, which may include the requirement to 
install air pollution controls.

J. Quantitative Analysis

    At proposal, we presented a quantitative analysis of the possible 
emissions consequences of the range of different approaches to the RMRR 
exclusion to evaluate if our policy conclusions are correct. Our 
analysis was conducted using the Integrated Planning Model (IPM). This 
analysis was done for electric utilities because we have a powerful 
model to perform such an analysis that we do not have for other 
industries. We stated that the results for electric utilities 
accurately reflect the trends we would see in other industries.
    The IPM analyses of different scenarios showed that the breadth of 
the RMRR exclusion would have no practical impact on, let alone be the 
controlling factor in determining, the emissions reductions that will 
be achieved in the future under the major NSR program. The analyses 
showed that emissions of SO2 are essentially the same under 
all scenarios, but that under today's rule these emission levels will 
be met in a more economically efficient manner than the base case. This 
stands to reason because nationwide emissions of SO2 from 
the power sector are capped by the title IV Acid Rain Program. For 
NOX, these analyses showed modest relative decreases in some 
cases and modest relative increases in other cases. These predicted 
changes represent only a fraction of nationwide NOX 
emissions from the power sector, which hover around 4.3 million tons 
per year (tpy). At this time, we do not have adequate information to 
predict with confidence which modeled scenario is most likely to occur. 
What these analyses indicate, however, is that regardless of which 
scenario is closest to what comes to pass, today's rule will not have a 
significant impact, up or down, on emissions from the power sector. 
However, we expect the rule to result in significant improvements in 
safety, reliability, and other relevant operational parameters.
    The DOE also presented further analysis of the possible emissions 
consequences of the range of different approaches to the RMRR 
exclusion. Using the National Energy Modeling System (NEMS), a variety 
of changes in energy efficiency and availability were evaluated, as 
well as the effect on emissions resulting from these regulatory 
revisions. This analysis concluded that efficiency improvements 
resulting from increased maintenance, repair and replacement are 
expected to decrease emissions, whereas availability improvements are 
expected to increase emissions. In the cases represented in this 
analysis, the emissions reductions from assumed reductions in heat 
rates tended to dominate the corresponding effects of the assumed 
availability increases.
    A number of commenters said that the underlying assumptions EPA 
used in the IPM analysis were flawed and resulted in erroneous 
conclusions regarding the emission reduction potential of the proposed 
RMRR rules. Several commenters stated that EPA's IPM analysis 
incorrectly assumes that no major modifications at any older units 
would ever trigger the requirement to add new pollution controls. In 
addition, according to commenters, EPA also erroneously assumed that 
this lack of major maintenance, repair and replacement will have very 
little impact on the performance of those power plants, when in reality 
their emissions would increase significantly. The commenters cited a 
Clean Air Task Force analysis for power plants, which estimates that 
EPA's rule revisions will result in at least 7 million more tons of 
SO2 and 2.4 million more tons of NOx annually. 
Some commenters also questioned the appropriateness of using EPA's 
analysis for the electric generating sector to draw conclusions about 
non-utilities.
    One commenter said the IPM and DOE NEMS analyses correctly 
demonstrate that EPA's RMRR proposal will have no appreciable impact on 
emissions from the power sector. According to the commenter, this 
conclusion is consistent with EPA's findings in a 1989 report, ``1989 
EPA Base Case Forecasts,'' which demonstrated that continuing to allow 
utilities to undertake activities including ongoing annual operating 
and maintenance activities and a major refurbishment when the unit 
reached 30 years of operating life would have no appreciable impact on 
emissions from the power sector, just as EPA's and DOE's recent 
analysis confirmed.
    One commenter said the proposal lacks any reference to the gains 
accomplished by major NSR, the ongoing enforcement actions, settlements 
reached as a result of those actions, or the potential gains from the 
investigations now pending. The commenter argued that EPA's reliance on 
improvements in productive capacity as the measure of success fails to 
consider that productive capacity must be balanced with the interests 
of health and welfare. The commenter also noted that a critical part of 
EPA's burden is to consider all the relevant factors leading to its 
conclusion that the exclusions are necessary and appropriate and that 
at the very least this includes an assessment of the expected effects 
on emissions, which in turn will determine the public health benefits 
and costs of the proposed rule. Although data on emission reductions 
achieved under the existing program are available, we have stated that 
we cannot precisely quantify the effects the proposed rule will have on 
emissions. Some commenters stated that before promulgating a final 
rule, EPA should provide such a quantitative assessment of the rule.
    We disagree with the commenters who believe that emissions would be 
significantly higher for electric utilities than are estimated under 
the IPM model runs. These commenters' arguments rely on the assumption 
that EPA's base case is invalid because, if major NSR rules were left 
unchanged, eventually all coal-fired utilities would either apply BACT 
or deteriorate so badly that they would have to shut down. We do not 
believe this assumption is accurate. As we have explained, our 
experience suggests that under the current NSR program, managers of 
coal-fired electric generating facilities have available to them a 
number of actions they can take to avoid triggering major NSR, and in 
many instances they will take one of

[[Page 61265]]

these actions to avoid the high retrofit costs and delays in obtaining 
a major NSR permit. If necessary, owners or operators can and will 
limit their activities to those that do not trigger major NSR, and will 
take enforceable restrictions on fuel use or other actions to avoid 
major NSR. This results in some decline in efficiency and capacity, as 
the EPA's base case modeled, but the units would likely remain viable 
electric generating units for years without triggering BACT 
requirements. Thus, we believe our base case represents a far more 
realistic assessment of what would happen under current major NSR rules 
than the dramatic BACT reductions presented by these commenters.
    Furthermore, while some of the facilities may be modified and 
subjected to control, nationwide emissions as estimated in the model 
runs would still rise to the level of the Acid Rain cap for 
SO2. To the degree these modifications come at facilities 
that are otherwise projected to be controlled because of existing 
SO2 and NOX requirements, there would be no 
difference in effect between the model runs and alternative scenarios. 
We agree with the commenter who noted that the recent analysis and the 
estimated impact on emissions is consistent with the previous EPA 
report in 1989. Our recent analysis confirms that efficiency 
improvements have the potential to result in environmental benefits 
that offset (or more than offset) emissions increases from improved 
availability, but that previous major NSR rules discouraged these 
improvements.
    Regarding the applicability of our analysis to non-utility sectors, 
we continue to believe that our conclusions are valid for all sectors, 
and further, that the effects from the electric utility industry 
dominate those from other sectors. We acknowledge that the results for 
the SO2 cap for utilities cannot be extended to non-
utilities that are not similarly capped. However, our model runs for 
NOx reflected the absence of a cap, and are therefore valid 
for other uncapped sectors. Thus in the case of industrial boilers, 
which behave similarly to utilities, we would expect to see similar 
efficiency improvements and availability improvements occurring in 
tandem, resulting in either modest increases or decreases. Because the 
overall emissions from this sector are significantly smaller than for 
utilities, the modeled effects for utilities are expected to dominate 
the analysis.
    For other industrial sectors, we do not anticipate that emissions 
increases will result from equipment replacement activities that 
qualify as RMRR under today's rule. While some efficiency improvements 
may result, the overall effect of these improvements will not be to 
induce greater demand and greater emissions, in contrast to the effect 
shown by the modeling for utilities (i.e., demand for other industrial 
sectors depends on independent factors). Indeed, without increased 
demand, efficiency improvements that lower emissions per unit of output 
would result in a decrease in emissions.
    A number of commenters raised concerns that EPA had not analyzed 
the impact of the final rule on industries other than for electric 
utilities. We have, thus, supported further efforts to analyze 
empirically the effects of this rule. This work is included in the 
Regulatory Impact Analysis (RIA) for the final rule. Even the experts 
involved in this analysis emphasize that empirical assessments of the 
costs, emissions, and other economic and environmental effects of this 
rule are extremely difficult to perform, particularly when generalizing 
beyond the specific industrial sector and type of facility involved. 
The analysis would have to simulate a great many decisions made by each 
plant involving routine maintenance under a variety of policy 
scenarios. There is simply no credible way to make these assessments 
for the entire economy or for an entire sector. Hence, with the 
exception of the electric utility industry model, we relied on a case 
study approach to gain insights as to how this rule affects particular 
industrial sectors.
    A series of case studies were analyzed by an EPA contractor to 
estimate the overall impact of the final rule on six different 
industrial sectors (automobile manufacturing, carbon black 
manufacturing, natural gas transmission, paper and pulp mills, 
petroleum refining and pharmaceutical manufacturing). The analysis was 
designed to examine effects of the final rule, but it is important to 
note that the case studies were performed prior to decisions on the 
exact form and content of the final rule. For example, the selection of 
process units for each of the industries may not be an accurate 
depiction concerning how a particular industry's operations should be 
separated into process units under the final rule. As such, none of 
these characterizations should be taken as EPA's position on 
appropriate process units for a given industry. (Information on that 
subject can be found in Section III.F of the preamble and in the final 
rule for selected industries.) In addition, in costing out replacement 
activities in the different industries, the contractor made assumptions 
regarding which costs needed to be included and how multiple 
replacement activities should be grouped that may not be consistent 
with the final rule. Again, these assumptions on the part of the 
contractor should not be interpreted as EPA's conclusions of how their 
rules should be applied to such replacement activities in these 
industries.
    Even with these caveats, the case studies provide useful insight 
into the potential effects of the final ERP. The six industries are 
significant sources of air pollution emissions and are very diverse in 
terms of their types of operations, their existing maintenance, repair 
and replacement strategies, and the range of potential replacement 
costs at some of their process units. This diversity is important 
because the final rule will impact a great many industrial sectors and 
individual process units which are extremely varied in terms of their 
maintenance, repair and replacement strategies. For example, issues 
related to safety, reliability and availability will vary greatly 
across these industries. The need to assure that the electricity and 
natural gas supply is reliable and available is critical to ensuring 
the safety of the public in the hottest and coldest times of the year, 
and it is critical to the operation of the nation's infrastructure, to 
the degree they do not have backup power generation, devoted to public 
health (e.g., drinking water, sewage treatment, food refrigeration, 
hospitals). Thus, strategies related to maintenance, repair and 
replacement at existing facilities are critical to ensure that vital 
electric utilities and natural gas transmission continue uninterrupted. 
As we are clarifying what activities fall within the ERP, owners or 
operators at these facilities will be able to make decisions on when 
and how to conduct RMRR activities based on engineering judgement.
    The case studies conclude that equipment replacement activities 
vary widely within these industries for the process units selected. 
Across the industries, the studies estimated that equipment replacement 
activities could range in percentage by over an order of magnitude. By 
establishing a threshold at 20 percent of the replacement cost of the 
process unit, we believe we have set a reasonable standard that allows 
most replacements to proceed unimpeded as long as the other safeguards 
are met. At the same time, under the 20 percent threshold, the most 
capital-intensive replacements would be subject to case-by-case review. 
The data from these case studies clearly indicate that 20 percent would 
function well as the dividing line between those replacement activities 
that automatically qualify under the

[[Page 61266]]

ERP and those activities which should be subject to case-by-case 
review.
    The case studies also indicate that replacement activities in these 
industries should not lead to increased emissions at the sources. Based 
on the case studies, we believe that replacement with identical or 
functionally equivalent equipment as the rule requires, will result in 
equivalent or reduced emissions. The decrease in emissions would result 
from efficiency improvements that reduce the amount of air pollution 
emitted per product produced in the process unit. Therefore, if 
operating levels do not change, then total emissions will decrease with 
such identical or functionally equivalent equipment replacements.
    The case studies looked at a wide range of projects. We have 
concluded based on this analysis that replacement activities do not 
generally cause changes in operating levels at the process unit. 
Instead, other factors, like economic downturns or increased demand for 
the product of the process unit, will cause operating levels to 
fluctuate. Efficiency changes, even when they lead to increases in 
product output from the same raw material input will not lead to 
increases in emissions unless an independent factor like increased 
demand for the product also occurs. We strongly support efficiency 
improvements where they can occur as long as the other safeguards in 
the rule are met.
    Our inability to model economy-wide impacts does not mean we cannot 
characterize the effects of this rule. In qualitative terms, the case 
studies further support our conclusion that the old case-by-case 
approach to RMRR is having perverse effects by discouraging projects 
that would improve efficiency. As noted elsewhere, efficiency 
improvements necessarily imply less pollution holding everything else 
constant. For example, the case study on the pulp and paper industry 
finds that:

    ``[A]s [safety, reliability and efficiency] activities begin to 
be reviewed, those that raise * * * questions under the ambiguity of 
the current rules may be postponed, altered, or simply cancelled. 
Under the proposed ERP approach, these activities can be tested 
against a clearer set of criteria, that will allow more activities 
to be executed.
    * * * The new approach provides the regulatory clarity and 
certainty in making applicability decisions that is completely 
absent from the current case-by-case approach. Thus, the manner in 
which mills will handle the processing of equipment replacement 
activities, with regard to assessing their air permit applicability 
assessments, will be able to be streamlined. By definition, a 
``case-by-case'' approach is simply unworkable for a typical pulp 
and paper mill, which may have thousands of maintenance and repair 
related work orders involving equipment replacements executed each 
year, affecting all areas of mill operations. Clearly, only a small 
subset of these equipment replacement activities can be evaluated 
using the complicated and vaguely interpreted multi-factor test 
inherent with the current case-by-case approach. * * * The proposed 
ERP approach helps by setting criteria for the routineness 
determinations. Under the proposed approach, a mill could set up 
more straight-forward guidelines to be followed throughout an 
organization that would allow quick and defensible determinations to 
be made regarding individual maintenance activities.''

Based on the analytical work performed by the contractor for pulp and 
paper, we expect that, at such facilities, the power boiler would be 
the most affected by the ERP, as well as an important or even dominant 
emissions source. We would anticipate that this would be true for many 
of the inorganic and organic chemical subsectors. In fact, we did not 
pursue an analysis of the chlor-alkali sector, in large part because 
the power boiler was the most obvious process unit to analyze, and the 
issues raised overlapped with the pulp and paper analysis. Thus, it is 
logical that the conclusions from the case studies would generalize to 
many other sectors.
    Beyond the case studies, there is also a great deal of research and 
experience that allows for some robust findings. Previous research, 
such as the articles cited below, supports the following findings:
    [sbull] Enhanced efficiency and less pollution in the short run. 
Holding everything else constant, when a plant's efficiency increases, 
pollution must go down. This nation's growing experience with pollution 
prevention, efficiency enhancements, voluntary environmental programs, 
and Environmental Management Systems adoption all reinforce the notion 
that enhanced plant efficiency translates into less environmental 
pollution.\10\ Further, there is an economic incentive to keep plant 
efficiency high. Proper maintenance and the resulting efficiency 
enhancements and pollution prevention reduce resource needs and 
therefore reduce costs.\11\ By providing the certainty needed to plan 
and undertake efficiency investments (economically efficient 
maintenance) this rule will achieve lower pollution.
---------------------------------------------------------------------------

    \10\ By efficiency, we mean unit of input per unit of output, 
for example, amount of energy needed to produce a specific amount of 
output. Another example would be the amount of raw material to 
produce a specific amount of output.
    \11\ A common example illustrates the point well. When one 
``tunes-up'' a car, the automobile gets more miles per gallon, is 
cleaner burning, and is cheaper to operate.
---------------------------------------------------------------------------

    [sbull] The rule will allow firms to take advantage of pollution 
prevention opportunities and new, innovative pollution-reducing 
technologies. As technology advances, plants will be able to replace 
existing components with functionally equivalent components that 
enhance energy efficiency (and reduce pollution).\12\ One example of 
such an opportunity identified by the EPA contractor in one of the case 
studies is the replacement of spray guns on a topcoat operation in 
order to improve the quality of the paint job, while also increasing 
the transfer efficiency, and decreasing coating and associated solvent 
usage. This project could be deemed a physical change and have major 
NSR applicability ramifications if not for the ERP of the RMRR 
exclusion. Under the current case-by-case approach to RMRR, the 
facility may forego the change to the newer spray gun design if there 
is a perceived risk that the determination could be questioned. Under 
the new ERP approach, the change would proceed more definitively as 
RMRR, and thus the emission reductions could be realized.
---------------------------------------------------------------------------

    \12\ For example, energy efficiency is not a design parameter to 
determine functional equivalency for defining routine maintenance. 
Accordingly, a firm could adopt a more efficient ``functionally 
equivalent'' technology without fear of triggering NSR provisions.
---------------------------------------------------------------------------

    [sbull] While firms can operate existing plants efficiently, the 
rule preserves powerful incentives within the CAA to adopt ``leap-
frog'' technologies and production processes that further reduce costs, 
increase efficiencies and reduce pollution. Because of the CAA 
requirements and economic gains associated with improved efficiency, 
producers still have an incentive to invest in these clean technologies 
to replace older facilities.
    In addition, a substantial body of research has explored the 
consequences of environmental regulation that sets more stringent 
control requirements for new sources. This research explores how 
differentiated regulation can affect firm behavior both on theoretical 
and empirical grounds. A listing of some of this literature is included 
in the RIA for the final rule. This literature provides further 
evidence that the NSR can easily distort investment and production 
decisions against more efficient maintenance and replacement.
    Therefore, based on the information evaluated, we affirm the 
overall conclusion of our analysis--that today's rule has no practical 
effect on the environmental benefits of major NSR in the future. We 
have presented

[[Page 61267]]

additional, more detailed supporting information in our final RIA and 
our response to comments document, both of which can be found in the 
docket for today's action.

K. Consideration of Other Options

    In addition to the cost-based approaches that we proposed, we also 
asked for comment on age-based and capacity-based approaches, and any 
other viable option for addressing RMRR.
1. Annual Maintenance, Repair and Replacement Allowance
    We are not taking action on the proposed Annual Maintenance, Repair 
and Replacement Allowance option for the RMRR exclusion, and therefore 
public comments on this option are not addressed at this time. We will 
address comments on our proposed Annual Maintenance, Repair and 
Replacement Allowance if and when we take final action on that 
proposal.
2. Capacity-Based Option
    As mentioned above, we considered the alternative option of 
developing an RMRR provision based on the capacity of a process unit. 
Under such an approach, an owner or operator could undertake any 
activity that does not increase the capacity of the process unit. 
Basing RMRR on capacity has appeal for several reasons. For starters, 
an objective of RMRR is to keep a unit operating at capacity and/or 
availability. In addition, the linkage between capacity and 
environmental impact is more apparent than that between cost and 
environmental impact. Finally, this type of approach might, in 
principle, be easier to use before beginning actual construction than 
some of the cost-based approaches.
    Several commenters were concerned with defining the capacity of a 
process unit. Capacity may be defined based on input or output. 
Nameplate capacity of a process unit may vary greatly from the capacity 
at which the process unit may be able to operate. It may be more 
appropriate in some industries to measure capacity based on input while 
in others on output. Commenters felt that a capacity-based approach 
would not be workable at complex manufacturing sources, because 
``capacity'' as a useful shorthand term for the processing capability 
correlates exactly only with a historical feed or product slate no 
longer available or made. A number of commenters supported a capacity-
based option, generally indicating that a capacity-based option would 
be simpler and less burdensome to use than the other proposed 
approaches.
    Another large concern of commenters was that a capacity-based 
approach could prevent facilities from performing activities that make 
the facilities more efficient. RMRR provisions need to include some 
form of the other approaches to account for energy efficiency projects 
at utilities, which could increase output capacity (i.e., production) 
without necessarily increasing heat input or fuel consumption. Some 
commenters noted that maximum hourly emissions is a more appropriate 
surrogate for a change in capacity, because it is consistent with 
existing NSPS procedures and with averaging periods for ambient air 
quality monitoring and standards.
    We agree that an appropriate capacity-based approach would have to 
be tailored to various types of sources, with capacity based on input 
for some and on output for others. As an example, in a review of 
promulgated and proposed Maximum Achievable Control Technology 
standards, six of eleven standards measured capacity based on process 
unit output while five standards based capacity on input. In fact, the 
NSPS exclusion for increases in production rate at 40 CFR 60.14(e) 
originally was dependent upon the ``operating design capacity'' of an 
affected facility. In proposed revisions to the NSPS program published 
on October 15, 1974, we state (39 FR 36948):

    ``The exemption of increases in production rate is no longer 
dependent upon the ``operating design capacity.'' This term is not 
easily defined, and for certain industries the ``design capacity'' 
bears little relationship to the actual operating capacity of the 
facility.''

    We also agree that a capacity-based approach has its limitations, 
as described by the commenters. We have concluded that the ERP 
eliminates the need to implement the capacity based approach. We have 
decided not to finalize a capacity-based approach.
3. Age-Based Option
    Under our proposed age-based approach, any process unit under a 
specified age could undergo any activity that does not increase the 
capacity of a process unit on a maximum hourly basis without triggering 
the requirements of the major NSR program. However, the activities 
could not constitute reconstruction of the process unit; that is, their 
cost could not exceed 50 percent of the cost of a replacement process 
unit. The age of the process unit would likely be in the range of 25-50 
years. We also proposed that the owner or operator would have to become 
a Clean Unit as defined at 40 CFR 51.165(c)(3), 51.166(t)(3), and 
52.21(x)(3), once the age of a process unit exceeds the age threshold.
    Such an approach would provide an owner or operator a clear 
understanding of RMRR for an extended period of time. It also may 
provide the owner or operator greater flexibility than under the 
current system for a limited period of time. Like the capacity-based 
approach, this approach would, in principle, allow for a fairly simple 
preconstruction determination of applicability.
    Very few commenters expressed any interest in developing this type 
of approach. Their concerns centered around defining capacity and 
establishing the age cut-off (because the useful life of equipment is 
difficult to establish and may vary greatly). Other concerns raised by 
commenters were that some of the activities that would be allowed at 
newer sources do not fit within any ordinary meaning of RMRR and some 
of the activities that would be forbidden at older facilities would 
come within that meaning, and also that some sources may consciously, 
and appropriately, engage in aggressive RMRR as a method of maximizing 
the life span of its process units, and an age-based approach would 
discriminate against them.
    One commenter stated that EPA should establish a normal lifetime, 
tailored to each industry, beyond which industry would need to install 
BACT or shut down. This type of approach would obviously require a 
substantial amount of time and analytical effort.
    The age of a source alone is not a legitimate reason to require the 
addition of pollution control equipment. Age has no direct bearing on a 
unit's environmental impact; some facilities maintain equipment better 
than others. We have decided not to promulgate an age-based approach. 
We have several basic concerns with this approach that we have not been 
able to reconcile. We also believe that the equipment replacement 
approach largely addresses the commenters' concerns regarding the age-
based approach.
    Thus, we have decided not to finalize a rule using this approach.

L. Specific List of Excluded Activities

    Several commenters supported the development of lists of activities 
that are considered RMRR; some of these commenters also supported 
developing lists of activities that do not qualify as RMRR. Commenters 
suggested various ways in which such lists could fit into the overall 
RMRR program. We are concerned, however, that such a list

[[Page 61268]]

would have to be implemented through rulemaking, which would require a 
considerable amount of time, analytical effort, and resources.
    A commenter suggested two ways by which we could develop a list of 
qualifying activities. First, we could review records for ongoing 
enforcement activity, to identify activities that we have and have not 
already alleged to be RMRR. There is an ample body of knowledge for 
electric power plants. Second, we could identify where activities would 
fall with respect to the cost criteria, then adjust the classification 
of each activity based on the WEPCO criteria to prepare lists of 
routine and nonroutine activities.
    Some commenters felt that industry-specific lists of routine and 
nonroutine activities would provide the best interim clarification to 
major NSR until legislative reform is in place. Other commenters 
opposed the development of lists of activities that are considered 
RMRR, contending that such lists would become quickly outdated.
    Some commenters requested that certain activities be specifically 
classified as RMRR. These suggested activities included the following:
    [sbull] The common practice of changing out the engine core in a 
combustion turbine when it is due for overhaul (to reduce downtime). 
The removed engine core is overhauled offline, and is then available to 
be switched in for the next like-kind engine core that reaches the 
point of overhaul. Unless the components are upgraded, the heat input 
remains the same and so does the emissions rate.
    [sbull] Any change that does not increase the achievable hourly 
emissions (as determined based on the permit and/or original design 
parameters) of existing equipment, processes, and emissions units.
    [sbull] Certain activities, for example, boiler tuning and 
maintenance, repair and replacement of air pollution equipment or CEMS 
should be categorically excluded as RMRR.
    [sbull] Any activity that is part of a long-term service agreement 
(primarily gas turbines) should be categorically excluded from major 
NSR.
    [sbull] Any activity involving steam turbine overhaul work should 
be categorically excluded from major NSR.
    Activities such as the above might be RMRR, but we believe there 
are simply too many activities in too many industries to effectively 
improve major NSR implementation through creation of lists. Moreover, 
lists would be a ``snapshot in time'' that would need to be reviewed 
and periodically updated for each industry sector. We have consequently 
decided not to attempt to list activities that are categorically 
excluded as RMRR.

M. Stand-Alone Exclusion for Energy Efficiency Projects

    In the proposal, we acknowledged that certain types of activities 
that improve energy efficiency would not qualify as RMRR. We solicited 
comment on whether there was the need for a ``stand-alone'' exclusion 
for activities that promote energy efficiency.
    Many commenters supported a stand-alone exclusion from major NSR 
for energy efficiency projects. With the following safeguards, they 
favored specifically excluding from the definition of ``major 
modification'' activities that promote energy efficiency and/or 
resource conservation when: (1) The activity results in lower emissions 
per unit of production or lower energy utilization per unit of 
production; (2) the percent decrease in emissions or energy utilization 
per unit of production is greater than the percent increase in maximum 
hourly emission rates; (3) activity costs do not exceed 50 percent of 
the replacement value of the process unit; and (4) the activity does 
not result in an increase in allowable emissions.
    Other commenters pointed out that efficiency upgrades will 
frequently create incentives to further utilize a source and 
subsequently increase mass emissions. One commenter stated that if 
activities that result in small efficiency gains can qualify as RMRR, 
older, dirtier electric generating units will be better able to out-
compete newer, much cleaner plants (that have higher costs due to 
emission controls).
    One commenter stated that EPA is incorrect in stating that energy 
efficiency projects are being discouraged by major NSR, particularly 
under the new actual-to-projected-actual applicability test. This 
commenter added that the only projects that are discouraged by major 
NSR are ones that increase emissions. This commenter felt that the 
December 2002 final major NSR rules provide a broad range of major NSR 
exclusions (including revised baseline determinations, Clean Unit 
designations, pollution control projects, PALS, and combinations of 
these provisions, as well as an RMRR exclusion) under which energy 
efficiency projects will certainly occur.
    We strongly support efforts to improve energy efficiency at 
existing power plants. These activities reduce the amount of air 
pollution emitted per unit of electricity generated. We believe that 
today's ERP supports energy efficiency projects and that the actual-to-
projected-actual applicability test contained in the December 2002 NSR 
final rules also should remove impediments to energy efficiency 
projects. Together, these rules will obviate the need for a specified 
RMRR provision for energy efficiency projects. Thus, at this time we 
are not finalizing a provision to categorically exclude energy 
efficiency projects from major NSR.

N. Legal Basis

1. How Does the NSR Program Address Existing Sources and Why Is Today's 
Rule Consistent With This Approach?
    The core of the NSR program is to require preconstruction permits 
for all new major sources. Congress specifically decided that existing 
sources generally would not be required to obtain permits. These 
considerations are the starting point for understanding its application 
to ``modifications'' and the meaning we should give that term.
    The NSR program's scope is closely related to the scope of the NSPS 
program, created seven years earlier in the CAA Amendments of 1970. In 
section 111 of the CAA, which sets forth the NSPS provisions, Congress 
applied the New Source Performance Standards to ``new sources,'' secs. 
111(b)(1)(B), 111(b)(4). Congress determined that as a general matter 
it would not impose the NSPS standards on existing sources, instead 
leaving to the State and local permitting authorities the decision of 
the extent to which to regulate those sources through ``State 
Implementation Plans'' designed to implement National Ambient Air 
Quality Standards (NAAQS). See sec. 110.
    Congress followed a similar approach in determining the scope of 
the major NSR program established by the 1977 Amendments to the CAA. As 
amended, the CAA specifies that State Implementation Plans must contain 
provisions that require sources to obtain major NSR permits prior to 
the point of ``construction'' of a source. Secs. 172(c)(5); 165 (a). By 
contrast, the CAA generally leaves to State and local permitting 
authorities in the first instance the question of the extent, means and 
timetable for obtaining reductions from existing sources needed to 
comply with National Ambient Air Quality Standards. See secs. 
172(c)(1), 161.
    NSR's applicability to existing sources to which a ``modification'' 
is made is an exception to this basic concept. This exception likewise 
finds its roots in the NSPS program's applicability to 
``modifications'' of existing sources. The 1970 CAA made the NSPS 
program applicable to modifications through its

[[Page 61269]]

definition of a ``new source,'' which it defined as ``any stationary 
source, the construction or modification of which is commenced after 
the publication of regulations * * * prescribing a[n applicable] 
standard of performance * * *.'' Section 111(a)(2). Section 111(a)(4), 
in turn, defined a ``modification'' as ``any physical change in, or 
change in the method of operation of, a stationary source which 
increases the amount of any air pollutant emitted from such source or 
which results in the emission of any air pollutant not previously 
emitted.''
    Congress did not further define the terms ``physical change'' or 
``change in the method of operation'' in the NSPS program. Therefore we 
issued regulations to clarify their meaning. As early as our 1971 NSPS 
regulations, we have made clear that many activities that do not affect 
the contemplated operation of a unit in a manner consistent with its 
original design are not physical or operational changes. Specifically, 
in our 1971 NSPS regulations, we determined that physical or 
operational changes do not include:
    (1) ``Routine maintenance, repair and replacement'' of equipment;
    (2) ``An increase in the production rate, if such increase does not 
exceed the operating design capacity of the affected facility'';
    (3) ``An increase in the hours of operation''; and
    (4) ``Use of an alternative fuel or raw material if * * * the 
affected facility is designed to accommodate such alternative use.''

36 FR at 24877 (Dec. 23, 1971). The premise behind characterizing these 
activities as not being ``changes'' is that they all contemplate that 
the plant will continue to be operated in a manner consistent with its 
original design.
    The 1977 Amendments to the CAA likewise made the NSR program 
applicable to ``modifications.'' The original 1977 Amendments did so 
explicitly only in their provisions dealing with the non-attainment 
portion of the NSR program, see CAA sec. 171(4). But in ``technical and 
conforming'' amendments to the 1977 Amendments, Congress clarified that 
it intended the same result with respect to the prevention of 
significant deterioration provisions, see CAA sec. 169(2)(C).
    Notably, Congress did not enact a new definition of 
``modification'' in either the original 1977 Amendments or the 
``technical and conforming amendments.'' Rather, it incorporated the 
NSPS definition of ``modification'' by cross-reference. See CAA sec. 
169(2)(C); CAA sec. 171(4). In moving the adoption of those amendments, 
the sponsor (who was also the sponsor of the original 1977 Clean Air 
Act Amendments and who indicated that the technical amendments had been 
approved by all members of the original 1977 Amendments conference 
committee) stated in a summary and statement of intent that he placed 
in the Congressional Record that this was a deliberate choice. As that 
summary explained, Congress intended the amendment ``implement[ed] the 
[1977 Clean Air Act Amendments] conference agreement to cover 
``modification'' as well as ``construction'' by defining 
``construction'' in part C to conform to usage in other parts of the 
Act.'' 123 Cong. Rec. 36331 (Nov. 1, 1977). We have understood this to 
be a reference to our preexisting rules interpreting the term 
``modification'' in the NSPS context. 49 FR 43211, 43213 (1984); see 
also 43 FR 26388, 26394, 26397 (June 19, 1978).
    The original 1978 NSR rules concerning modifications that we 
promulgated after enactment of the 1977 Amendments generally tracked 
the NSPS approach by specifying that ``routine maintenance, repair and 
replacement'' was not a change; by specifying that changes in hours of 
operation and rates of production were not a ``change'; and by using 
the same basic approach NSPS used to the question of what constitutes 
an ``increase'' (increase to a source's potential to emit, except that 
the NSR rule used annual potential to emit while the NSPS program used 
short-term potential to emit). 43 FR 26388 (June 19, 1978). Even after 
the D.C. Circuit struck down other portions of our 1978 NSR rules in 
its original per curiam decision in Alabama Power Co. v. Costle, 606 
F.2d 1068 (D.C. Cir. 1979), we continued to propose to retain the RMRR 
provision and the ``potential to emit'' approach to emissions increases 
in our revised rules, although to drop the ``hours of operation and 
rate of production'' provisions because the ``potential to emit'' 
provision made them unnecessary. 44 FR 51924, 51937 (September 5, 
1979). In our final 1980 NSR rules, however, issued after the D.C. 
Circuit's final Alabama Power decision, 635 F.2d 323 (1980), we changed 
our approach to the definition of ``increase'' in the NSR context to 
specify that a change would trigger NSR if it would result in an 
increase over ``actual annual emissions.'' 45 FR 52676 (August 7, 
1980). At the same time, and notably, we restored the provisions 
stating that increases in hours of operation or production rate were 
not ``changes.'' Id. at 52704.
    It is important to understand what we did--and did not--decide in 
those final 1980 NSR rules. What we did decide was that as a general 
proposition, we would better serve the purposes of the NSR program if 
we used ``actual'' rather than ``potential'' emissions as a baseline 
for determining whether an activity at a new source results in an 
emissions increase. What we did not decide was that the purposes of the 
NSR program never allow us to exclude from the definition of ``change'' 
any activity at a plant that may increase its actual emissions but does 
not increase its ``potential'' emissions. In particular, for example, 
we decided to retain the ``hours of operation'' and ``rate of 
production'' exclusions even though such changes might result in 
increases in ``actual'' emissions because not having the provisions 
``would severely and unduly hamper the ability of any company to take 
advantage of favorable market conditions.'' Id. Similarly, we retained 
the exclusion for ``routine maintenance, repair and replacement'' even 
though it too can result in emission increases. Yet there is little 
doubt that increases in hours of operation and rates of production and 
RMRR arguably could be understood to fall within the statutory 
definition of modification, since increases in hours of operation and 
rates of production certainly may be argued to be changes in the 
``method of operation'' of a plant, and RMRR certainly may be argued to 
be a ``physical change'' to a plant. On balance, however, we rejected 
that interpretation and determined that the definition of modification 
should not be read so broadly as to encompass hours of operation or 
production rate increases, at least so long as they are unrelated to a 
physical change.
    In the revisions to the NSR program we announced last December, we 
reiterated our adherence to the view that as a general matter we should 
continue to use ``actual'' rather than ``potential'' emissions in 
determining what activities constitute ``modifications'' under NSR. We 
continue to believe that is correct, but we also believe we should 
amplify our reasons for holding this view and why that view is entirely 
consistent with the rule we are promulgating today. In determining the 
scope to give to ``modification,'' we believe it is important to give 
weight to both aspects of what Congress decided in 1977. Congress 
decided that generally speaking, existing plants would not be subject 
to NSR, but that they would be subject to NSR when they made

[[Page 61270]]

``modifications.'' It is also important to understand why Congress 
chose this point at which to impose NSR on existing plants: to avoid 
the need to impose costly retrofits, but require placement of new 
control technology at a time when it makes the most sense for it to be 
installed. See H.R. Rep. No. 294, 95th Cong., 1st Sess. 185, reprinted 
in 1977 U.S. Code Cong. & Admin. News at 1254; 116 Cong. Rec. 32,918 
(Sept. 21, 1970) (remarks of Sen. Cooper). See also WEPCO, 893 F.2d at 
909-910; National-Southwire Aluminum Co. v. EPA, 838 F.2d 835, 843 (6th 
Cir., Boggs, J., dissenting), cert. denied, 488 U.S. 955 (1988). A 
wholesale exclusion of any activity that restores a plant to its 
potential to emit from the definition of modification is not consistent 
with this balance, since there are many activities that might have that 
effect but the conduct of which would be an extremely effective time 
for the placement for new control technology.
    At the same time, we believe it is also important to give equal 
weight to the converse proposition that existing plants should not have 
to install new control technology in the ordinary course of their 
operations. To require them to do so would fail to give full effect to 
Congress's decision that existing sources generally would not be 
required to obtain permits. It would also subject these plants and the 
consumers who rely on them to enormous dislocation and expense. That is 
why we believe we have rightly excluded increases in hours of operation 
and rates of production from the definition of ``change.'' That is also 
why we believe we have rightly excluded ``routine maintenance, repair 
and replacement'' of existing plants from that definition.
    For similar reasons, we believe today's rule draws an appropriate 
line of demarcation between replacements that should not be treated as 
changes, and those as to which further consideration of the question is 
appropriate. Our rule states categorically that the replacement of 
components with identical or functionally equivalent components that do 
not exceed 20% of the replacement value of the process unit and does 
not change its basic design parameters is not a change and is within 
the RMRR exclusion. On the other hand, the rule contemplates case-by-
case evaluation of identical or functionally equivalent equipment 
replacements that do not have these characteristics.
    We believe this approach is consistent with the intended scope of 
``modification'' under the NSR program. The record of this rulemaking 
demonstrates that there are substantial categories of replacement 
activities undertaken in order to assure the safety, reliability and 
efficiency of existing plants that, if conducted at the same time, cost 
less than the 20-percent replacement cost threshold. It also 
demonstrates that there are sound business reasons why an owner or 
operator may find it makes sense to conduct some of these activities at 
the same time.
    On the other hand, given the costs and technical problems 
associated with installing state-of-the-art pollution controls at 
existing facilities, we do not believe it plausible that, if faced with 
the choice of replacing equipment that has a value less than 20 percent 
of a process unit and having to install those controls, or coming up 
with another solution--such as repairing the existing equipment or 
limiting hours of operation so as to be confident that the activity 
will not trigger NSR--the owner of a source would elect to replace the 
equipment if he also has to install the state-of-the-art controls. 
Rather, we believe he will repair the existing equipment or 
artificially constrain production. Therefore the replacement of that 
equipment is not, in fact, an opportune time for the installation of 
such controls. It follows that treating such replacements as an NSR 
trigger will not lead to the installation of controls. Rather, it will 
merely create incentives to make a plant less productive than its 
design capacity would allow it to be.
    We do not believe it is the policy of the CAA to seek to promote 
emissions reductions by forcing new limits on hours of operation or 
rates of production of existing plants. We made that point clear in 
1980 when we determined that we should retain the hours of operation 
and rate of production exclusions in the NSR context. To the contrary, 
as we said in promulgating the 1980 rules, Congress's decision to 
exclude existing sources because of the dislocation that covering them 
would cause can reasonably be understood as allowing those sources to 
increase hours of operation or production up to permitted levels as 
market conditions dictate. We note that this does not leave such 
activities outside the scope of the CAA: if a State concludes that 
resulting air quality considerations warrant revision to its SIP to add 
further limitations to a permit, it may exercise its authority to 
impose them, even in the absence of anything that constitutes a 
``change'' to an existing plant. But we believe that our 1980 
conclusion that increases in hours of operation or production at 
existing plants should not trigger NSR remains the better construction 
of the CAA. That being the case, we now believe that the fact that such 
increases may occur after replacement of equipment that does not 
present an opportune time for the installation of controls should 
change that conclusion.
    To summarize: with respect to existing sources, the purpose of the 
NSR provisions is simply to require the installation of controls at the 
appropriate and opportune time. The kind of replacements that 
automatically fall within the equipment replacement provision 
established today do not represent such an appropriate and opportune 
time. Accordingly, and given that it is consistent with the meaning of 
``change'' to treat this kind of replacement as not being a ``change,'' 
we believe excluding them on that basis from the definition of 
``modification'' as used in the NSR program is well calculated to serve 
all of the policies of the NSR provisions of the CAA, and is therefore 
a legitimate exercise of our discretion under Chevron, U.S.A. Inc. v. 
NRDC, 467 U.S. 837 (1984), to construe an ambiguous term. Likewise, we 
believe this approach is consistent with the holding in the WEPCO case, 
and with some though not all of that case's reasoning.
    Today's rule treats the activities excluded from the definition of 
``change'' as a category of ``routine maintenance, repair and 
replacement''. We received many comments as to whether we can and 
should adopt the ERP as an expansion of the RMRR exclusion. We believe 
it is appropriate to expand the former RMRR exception. Before 
promulgation of today's rule, we interpreted the phrase ``routine 
maintenance, repair and replacement'' to be limited to the day-to-day 
maintenance and repair of equipment and the replacement of relatively 
small parts of a plant that frequently require replacement. Today we 
are expanding the former definition of RMRR through this rulemaking to 
include other activities covered by the 20 percent cost threshold that 
are needed to facilitate the efficiency, reliability and safety of 
affected sources.
    We believe it is appropriate to add one final note regarding the 
fact that this approach represents a change from the approach we have 
taken in the recent past. As the Supreme Court explained in Chevron, 
where it upheld a considerably more significant shift in the Agency's 
understanding of Title I of the CAA, to wit, the scope of the term 
``stationary source,'' there is nothing inherently suspect about a 
change of approach of this type by an expert Agency seeking to 
interpret a technical statutory term so as best to accommodate 
competing

[[Page 61271]]

interests that Congress has charged the Agency with reconciling.
    In section 101 of the CAA, Congress stated that Title I of the CAA 
has a dual purpose: ``to protect and enhance the quality of the 
Nation's air resources so as to promote the public health and welfare 
and the productive capacity of its population'' (emphasis added). This 
duality is reiterated in the statement of purpose of the PSD provisions 
and in the House Report accompanying the 1977 Amendments in connection 
with the non-attainment provisions. See sec. 160(1) (purposes of the 
PSD program are, inter alia, ``to protect public health and welfare 
from any actual or potential adverse effect'' of air pollution and ``to 
insure that economic growth will continue to occur consistent with the 
preservation of existing clean air resources''); H.R. Rep. No. 95-294, 
p. 211 (The ``two main purposes'' of the non-attainment permitting 
program are ``(1) to allow reasonable economic growth to continue in an 
area while making reasonable further progress to assure attainment of 
the standards by a fixed date; and (2) to allow States greater 
flexibility for the former purpose than EPA's present interpretative 
regulations afford'').
    More specifically, with regard to the question at issue here, 
Congress directed EPA not to apply NSR preconstruction permitting 
requirements to existing plants as a general matter, but to apply them 
to ``modifications.'' Both directives are entitled to receive 
appropriate weight.
    In these circumstances, changes in an Agency's understanding 
informed by greater experience are not only not surprising, they are to 
be expected. Effectuating these underlying Congressional commands 
requires a careful weighing and accommodation of the competing 
considerations underlying them. Sensitivity to unintended consequences, 
and a willingness to adjust policies in a manner informed by a better 
understanding of those consequences, are a central element of the 
responsibilities of an Agency given such a charge. As the Chevron Court 
explained:

    Our review of the EPA's varying interpretations of the word 
``source''--both before and after the 1977 Amendments--convinces us 
that the agency primarily responsible for administering this 
important legislation has consistently interpreted it flexibly--not 
in a sterile textual vacuum, but in the context of implementing 
policy decisions in a technical and complex arena. The fact that the 
agency has from time to time changed its interpretation of the term 
``source'' does not, as respondents argue, lead us to conclude that 
no deference should be accorded the agency's interpretation of the 
statute. An initial agency interpretation is not instantly carved in 
stone. On the contrary, the agency, to engage in informed 
rulemaking, must consider varying interpretations and the wisdom of 
its policy on a continuing basis. Moreover, the fact that the agency 
has adopted different definitions in different contexts adds force 
to the argument that the definition itself is flexible, particularly 
since Congress has never indicated any disapproval of a flexible 
reading of the statute.

467 U.S. at 863-64.
    The Court went on to point out:

    In these cases the Administrator's interpretation represents a 
reasonable accommodation of manifestly competing interests and is 
entitled to deference: the regulatory scheme is technical and 
complex, the agency considered the matter in a detailed and reasoned 
fashion, and the decision involves reconciling conflicting policies. 
Congress intended to accommodate both interests, but did not do so 
itself on the level of specificity presented by these cases. * * *
    [A]n agency to which Congress has delegated policymaking 
responsibilities may, within the limits of that delegation, properly 
rely upon the incumbent administration's views of wise policy to 
inform its judgments. While agencies are not directly accountable to 
the people, the Chief Executive is, and it is entirely appropriate 
for this political branch of the Government to make such policy 
choices--resolving the competing interests which Congress itself 
either inadvertently did not resolve, or intentionally left to be 
resolved by the agency charged with the administration of the 
statute in light of everyday realities. * * *
    We hold that the EPA's definition of the term ``source'' is a 
permissible construction of the statute which seeks to accommodate 
progress in reducing air pollution with economic growth. `The 
Regulations which the Administrator has adopted provide what the 
agency could allowably view as * * * [an] effective reconciliation 
of these twofold ends. * * *'

Id. at 865-66 (citations and footnotes omitted). We believe the same 
reasoning applies here, and makes it entirely appropriate for us to 
adopt the equipment replacement provision today.
2. Why Today's Rule Appropriately Implements the Clean Air Act's 
Definition of Modification
    As noted above, the modification provisions of the NSR program in 
parts C and D of title I of the CAA are based on the definition of 
modification in section 111(a)(4) of the CAA. The term ``modification'' 
means ``any physical change in, or change in the method of operation 
of, a stationary source which increases the amount of any air pollutant 
emitted by such source or which results in the emission of any air 
pollutant not previously emitted.'' As we observed in the notice of 
proposed rulemaking for this rule, that definition contemplates that 
you will first determine whether a physical or operational change will 
occur. If so, then you proceed to determine whether the physical or 
operational change will result in an emissions increase over baseline 
levels.
    Real-world, common-sense usage of the word ``change'' in ``physical 
change'' and ``change in the method of operation'' shows that 
``change'' is susceptible to multiple meanings. As we have noted 
previously, ``EPA has always recognized that Congress did not intend 
that every activity at an existing facility be considered a physical or 
operational change for purposes of NSR.'' 57 FR 32,314, 32,319 (July 
21, 1992). Conceivably, ``change'' could encompass a range of 
activities from periodically replacing filters in production machinery, 
to once in-a-lifetime anticipated replacement of a component, to 
complete replacement of a production unit.
    For example, all cars must periodically have their oil ``changed.'' 
When considered from one perspective, this activity does represent a 
``change'' because old oil is removed and new oil is added. From 
another perspective, however, this activity would not be considered a 
change because it does not alter any significant characteristic of the 
car.
    More to the point, chemical and pharmaceutical manufacturing 
operations often are designed, operated, and permitted as ``multi-
function'' facilities. These facilities have numerous pieces of 
equipment (such as storage tanks, reactors, distillation columns, 
centrifuges, filter dryers, etc.) that can be reconfigured to 
accommodate a wide variety of products and operating conditions. When 
switching from product X to product Y, a plant can make substantial 
``changes'' in the types of equipment used, the processing conditions, 
and the raw materials, reagents, solvents, and other processing 
materials. In this case, the same basic equipment is used to make a 
wide variety of end products. But, as long as the facility is operated 
as designed and permitted, we would not consider (and have not 
considered over the 20+ year life of the NSR program) such changes to 
be physical or operational ``changes'' for purposes of administering 
the NSR program.
    Similarly, manufacturing equipment often is built with expendable 
components. For example, industrial gas turbines, such as those used to 
drive compressors on natural gas pipelines, regularly need to have 
components

[[Page 61272]]

replaced as they wear out due to the high temperature and pressure 
conditions inside the turbine. In fact, these gas turbines are built 
with the knowledge and expectation that such replacements will be 
needed. In recognition of this fact, under the New Source Performance 
Standard for gas turbines, 40 CFR part 60, subpart GG, we have 
concluded that ``replacement of stator blades, turbine nozzles, turbine 
buckets, fuel nozzles, combustion chambers, seals, and shaft packings' 
are not ``changes'' for regulatory purposes. See EPA-450/2-77-017a, 
background support document for Subpart GG. Such replacements are akin 
to getting a new set of brakes on a car--not something that happens 
often, not an activity that is necessarily inexpensive, but plainly an 
activity that is an expected part of maintaining and operating the 
facility and one that does not represent an alteration of the affected 
process unit.
    As the preceding examples suggest, identifying activities that are 
``changes'' for NSR purposes--and thus potentially trigger the need for 
an NSR permit--requires the exercise of Agency expertise. The 
application of agency expertise to the interpretation of this statutory 
term is the classic situation in which an agency is accorded deference 
under Chevron, U.S.A., Inc. v. NRDC, 467 U.S. 837 (1984).
    Historically, we have asserted the power to interpret the relevant 
statutory terms. For example, even though both the NSPS and NSR 
programs incorporate the definition of ``modification'' from section 
111, from the outset EPA has adopted quite disparate readings of the 
term in our rules. See 57 FR 32314, 32316 (July 21, 1992) (WEPCO rule 
discussion of how emission increases are calculated differently for the 
NSPS and NSR programs). The NSPS program requires a change to result in 
an increase in the hourly potential to emit of the facility. 40 CFR 
60.14(a)-(b). In contrast, under NSR, we require an increase in annual 
emissions. E.g., 40 CFR 51.165(a)(1)(x). These disparate tests reflect 
the Agency's view that the statutory term ``modification'' must be 
construed with a view to what makes sense in particular statutory 
context, and are not obvious on their face.
    The exclusions from NSR we adopted in 1980 also reflect the 
exercise of the Chevron discretion. Not only did we adopt the RMRR 
exclusion at that time, but we also adopted exclusions for increases in 
the hours of operation, fuel changes, and raw material changes. Only 
the RMRR exclusion arguably could be justified as de minimis. For 
example, by doubling hours of operation, a 500 tpy emitting plant could 
conceivably double its emissions.\13\ The extra 500 tpy is far above 
any level EPA has ever thought justifiable as de minimis. E.g., 40 CFR 
51.166(b)(23)(i) (definition of ``significant''). Nor is it likely that 
these other exclusions could be based on some inherent power to adopt 
categorical exclusions from the CAA's commands. See Alabama Power 
Company v. Costle, 636 F.2d 323, 359 (D.C. Cir. 1980) (``categorical 
exemptions * * * are not favored''). Accordingly, these other 
exclusions must be justified as an exercise of Chevron discretion.
---------------------------------------------------------------------------

    \13\ As discussed below, our regulations provided a comparable 
exclusion from NSPS at the time of the 1977 Amendments that 
established the NSR program.
---------------------------------------------------------------------------

    As noted previously, in 1977 when Congress incorporated by 
reference into the NSR program the pre-existing NSPS statutory 
definition of modification, EPA had already adopted and had been 
administering regulations and policy under the NSPS program related to 
the meaning of the term ``modification.'' Our rules and policy provided 
that certain significant activities did not constitute physical or 
operational changes under the NSPS program prior to 1977 (or, for that 
matter, under the NSPS program as administered today). In addition to 
the gas turbine example provided above, perhaps the best indication 
that EPA did not consider the terms ``modification'' or ``change'' to 
cover everything other than de minimis activities is the exclusion for 
production rate increases under the NSPS program. 40 CFR 60.14(e)(2).
    Under this provision, projects valued at millions of dollars can be 
implemented--with no limitations on the nature of the project--without 
triggering applicable NSPSs. For example, up to 10 percent of the asset 
value of affected operations at a kraft pulp mill can be invested in a 
project without triggering the applicable NSPS, 40 CFR part 60, subpart 
BB. The affected facilities at a kraft pulp mill typically are valued 
in excess of $100 million. Therefore, an owner or operator can 
implement projects costing millions of dollars without triggering the 
applicable NSPS. This holds true regardless of the nature of the 
project--it can be a ``like-kind'' replacement of the kind addressed by 
today's rule or it can result in a substantial change in the nature of 
the operation. Thus, under the NSPS program that existed when Congress 
enacted NSR and incorporated into NSR the applicable NSPS definitions, 
projects of substantial cost that result in substantial change in 
affected facilities were not considered ``changes.'' The same is true 
under the NSPS program as it stands today.
    We recognize that the Agency previously has not specifically 
asserted that our interpretation of ``change'' and the exclusions from 
NSR are based on an exercise of Chevron discretion. In some instances, 
such as in a decision of the EAB, In re: Tennessee Valley Authority, 9 
E.A.D. 357 (EAB 2000), and in briefs in various enforcement-related 
cases, we have previously interpreted ``change'' such that virtually 
all changes, even trivial ones, are encompassed by the CAA. Thus, we 
generally interpreted the exclusion as being limited to de minimis 
circumstances. However, EPA does have the authority to interpret these 
key terms through rulemaking. Upon further consideration of the history 
of our actions, the statute, and its legislative history, EPA believes 
that a different view is permissible, and, for policy reasons discussed 
above, more appropriate. Therefore, we adopt this view prospectively in 
today's action.\14\
---------------------------------------------------------------------------

    \14\ We have taken positions in numerous court filings 
concerning the proper interpretation and usage of key statutory 
terms, such as ``physical change'' and ``any physical change.'' 
These positions were based on permissible constructions of the 
statute of which the regulated community had fair notice, and 
correctly reflect the Agency's reasonable accommodation of the Clean 
Air Act's competing policies in light of its experience at the time 
it adopted the RMRR exclusion in 1980. The Agency has sought, and 
has obtained, deference for its interpretations, and, 
notwithstanding today's adoption of a revised interpretation of the 
statute and an expansion of the RMRR exclusion, the Agency shall 
continue to seek deference for those prior interpretations in 
ongoing enforcement litigation.
---------------------------------------------------------------------------

    The argument that our authority to exclude certain activities from 
being modifications under new source review can only be based on a de 
minimis rationale sometimes relies on the word ``any'' used to modify 
``physical change'' and ``change in the method of operation,'' pointing 
to the word ``any'' in the definition of ``modification'' as a signal 
from Congress that the term ``change'' must be interpreted as 
encompassing the broadest possible sense of the term. Such an 
interpretation is not compelled by the language and legislative history 
of the statute, as demonstrated by the manner in which we have 
interpreted the word ``change'' under both the NSPS and the NSR 
programs.\15\
---------------------------------------------------------------------------

    \15\ We note that the word ``any'' is simply a modifier that 
does not change the meaning of the word it modifies. For example, 
using the term ``any'' to modify the word ``car'' does not somehow 
change or expand the meaning of the word ``car.'' ``Any'' simply 
means that, once you have decided what a car is, then all objects 
meeting the definition are encompassed.

---------------------------------------------------------------------------

[[Page 61273]]

    Nothing in the appellate case law directly disposes of this issue 
in a manner that prevents a new interpretation today. Two cases, 
Alabama Power and WEPCO, are relied on by some commenters to assert 
that EPA must interpret ``modification'' and ``change'' expansively and 
base all exclusions on a de minimis rationale. However, in Alabama 
Power, the issue before the court was the emissions increase portion of 
the definition of ``modification.'' The court would have allowed de 
minimis increases in emissions to be excluded from requirements 
applying to ``modifications'' under new source review but not emissions 
increases equal to the thresholds set by statute for new construction. 
636 F.2d at 399-400. The court did not have before it the issue of what 
is a ``change'' and did not decide this issue.
    In WEPCO, both parties advanced the view that the statute was clear 
on its face. EPA advanced the view that the term ``modification'' is 
necessarily broad, and that only de minimis departures are appropriate. 
WEPCO asserted that the plain meaning of the term ``physical change'' 
allowed for the five large scale rehabilitation projects it 
contemplated at its Port Washington plant. The WEPCO court held that 
the rehabilitation projects at issue were too large to reasonably 
conclude that they should not be treated as physical changes. The 
court's holding that the statute did not require the interpretation 
advanced by WEPCO does not deny EPA the discretion to decide to adopt a 
different, reasonable interpretation of the term ``modification.''
    While the Court in WEPCO decided that the projects in that case 
were physical changes, the decision in WEPCO does not answer the 
question of where to draw the line between activities that should and 
should not be considered ``changes.'' Nevertheless, contrary to the 
suggestions of several commenters, the projects at issue in WEPCO would 
have cost more than the 20 percent of replacement cost threshold 
selected today and, barring other applicable exclusions, would have 
been subject to case-by-case review in the PSD program. See section 
III.D above.\16\
---------------------------------------------------------------------------

    \16\ We note that decisions recently were rendered in two of the 
Agency's pending NSR enforcement cases in the utility sector. In 
both cases, the Agency asserted that the then existing RMRR 
exclusion should be applied in a narrow fashion such that only de 
minimis projects should be excluded under that rule. In our case 
against Ohio Edison in the U.S. District Court for the Southern 
District of Ohio, the court determined that the disputed projects 
did not qualify for the existing RMRR exclusion. The Agency sought 
and received from the court broad deference with regard to the 
Agency's interpretation of the CAA and the relevant EPA rules. In 
our case against Duke Energy in the U.S. District Court for the 
Middle District of North Carolina, the court issued a decision on 
cross motions for summary judgment. The decision took exception with 
several legal conclusions reached in the Ohio Edison decision and 
determined that the then existing RMRR exclusion must be applied 
from the perspective of what projects are routine within the 
relevant industrial source category. EPA today is adopting 
prospectively a new interpretation of the CAA and is finalizing a 
revision to the RMRR regulation at issue in those cases.
---------------------------------------------------------------------------

    Some commenters argued that, to further the purposes of the 
statute, any interpretation must result in the eventual elimination of 
so-called ``grandfathered'' facilities. We recognize the need to reduce 
emissions from many existing plants--regardless of whether they are 
``grandfathered'' (because they have never gone through NSR) or whether 
they have previously gone through NSR but can further reduce their 
emissions. EPA and States have issued regulations under a variety of 
statutory provisions to accomplish this goal in the past, and we will 
continue to do so in the future. We do not believe, however, the 
modification provisions of the CAA should be interpreted to ensure that 
all major facilities eventually trigger NSR. In fact, such an 
interpretation cannot be squared with the plain language of the CAA.
    An existing source--whether grandfathered or not--triggers NSR only 
if it makes a physical or operational change that results in an 
emissions increase. Thus, a facility can conceivably continue to 
operate indefinitely without triggering NSR--making as many physical or 
operational changes as it desires--as long as the changes do not result 
in emissions increases. This outcome is an unavoidable consequence of 
the plain statutory language and is at odds with the notion that 
Congress intended that every major source would eventually trigger NSR. 
Moreover, there is nothing in the legislative history of the 1977 
Amendments, which created the NSR program, to suggest that Congress 
intended to force all then-existing sources to go through NSR. To the 
extent that some members of Congress expressed that view during the 
debate over the 1990 amendments, such statements are not probative of 
what Congress meant in 1977. Central Bank of Denver, N.A. v. First 
Interstate Bank of Denver, N.A., 511 U.S. 164, 185-86 (1994), and cases 
cited.
    In deciding to incorporate by reference the statutory definition of 
``modification'' in section 111, Congress's intent cannot have been to 
preclude us from adopting an interpretation of ``modification'' or 
``change'' that differs from one that sweeps in all activities at a 
source. Under the NSPS program, this interpretation did not apply at 
the time of the 1977 amendments. When the NSPS definition of 
``modification'' was adopted as part of the NSR program in 1977, the 
Congressional Record explained that this provision, ``[i]mplements 
conference agreement to cover ``modification'' as well as 
``construction'' by defining ``construction'' in part C to conform to 
usage in other parts of the Act.'' 123 Cong. Rec. 36331 (Nov. 1, 1977) 
(emphasis added). Although we do not assert that the NSPS 
interpretation is the only one we could have adopted for NSR purposes 
(we followed quite a different interpretation from 1980 until today) at 
the very least it delineates a zone of discretion within which EPA may 
operate.
    Our interpretation today of physical or operational change in a 
flexible way furthers the purposes of the statute. As noted above, 
Congress made it clear that the CAA in general, and the NSR program in 
particular, should be administered in a manner that protects the 
environment and promotes the productive capacity of the nation. CAA 
section 101(b)(1). The Chevron Court recognized Congress' intent and 
noted that ``Congress sought to accommodate the conflict between the 
economic interest in permitting capital improvements to continue and 
the environmental interest in improving air quality'' when it 
established the NSR program. Chevron, 467 U.S. at 851. Generally, we 
believe that these goals are best accomplished by providing state and 
local governments with discretion to make decisions as to what 
emissions reductions are needed in their jurisdictions to attain and 
maintain good air quality. See CAA section 101(a)(3).
    It is now clear that many power plants and industrial facilities 
must substantially reduce their emissions in order to allow States to 
meet the stringent Federal air quality standards that the Supreme Court 
upheld in 2002. Under the CAA, Congress designed a number of regulatory 
programs that will collectively achieve the necessary reductions. 
Although the NSR program will effectively limit emissions from new and 
modified sources, it was not designed to achieve emission reductions 
from every existing source.

[[Page 61274]]

IV. Administrative Requirements for This Rule

A. Executive Order 12866--Regulatory Planning and Review

    Under Executive Order 12866 [58 FR 51735 (October 4, 1993)], we 
must determine whether the regulatory action is ``significant'' and 
therefore subject to review by the Office of Management and Budget 
(OMB) and the requirements of the Executive Order. The Executive Order 
defines ``significant regulatory action'' as one that is likely to 
result in a rule that may:
    (1) Have an annual effect on the economy of $100 million or more or 
adversely affect in a material way the economy, a sector of the 
economy, productivity, competition, jobs, the environment, public 
health or safety, or State, local, or tribal governments or 
communities;
    (2) Create a serious inconsistency or otherwise interfere with an 
action taken or planned by another agency;
    (3) Materially alter the budgetary impact of entitlements, grants, 
user fees, or loan programs, or the rights and obligations of 
recipients thereof; or
    (4) Raise novel legal or policy issues arising out of legal 
mandates, the President's priorities, or the principles set forth in 
the Executive Order.
    Pursuant to the terms of Executive Order 12866, OMB has notified us 
that it considers this an ``economically significant regulatory 
action'' within the meaning of the Executive Order. We have submitted 
this action to OMB for review. Changes made in response to OMB 
suggestions or recommendations will be documented in the public record. 
All written comments from OMB to EPA and any written EPA response to 
any of those comments are included in the docket listed at the 
beginning of this notice under ADDRESSES. In addition, consistent with 
Executive Order 12866, we consulted with the State, local and tribal 
agencies that will be affected by this rule. We have also sought 
involvement from industry and public interest groups.

B. Executive Order 13132--Federalism

    Executive Order 13132, entitled ``Federalism'' (64 FR 43255, August 
10, 1999), requires us to develop an accountable process to ensure 
``meaningful and timely input by State and local officials in the 
development of regulatory policies that have federalism implications.'' 
``Policies that have federalism implications'' are defined in the 
Executive Order to include regulations that have ``substantial direct 
effects on the States, on the relationship between the national 
government and the States, or on the distribution of power and 
responsibilities among the various levels of government.''
    This final rule does not have federalism implications. 
Nevertheless, as described in section II.C of this notice, in 
developing this rule, we consulted with affected parties and interested 
stakeholders, including State and local authorities, to enable them to 
provide timely input in the development of this rule. This rule will 
not have substantial direct effects on the States, on the relationship 
between the national government and the State and local programs, or on 
the distribution of power and responsibilities among the various levels 
of government, as specified in Executive Order 13132. We expect this 
rule will result in some expenditures by the States, we expect those 
expenditures to be limited to $580,000 for the estimated 112 affected 
reviewing authorities. This estimate reflects the small increase in 
burden imposed upon reviewing authorities in order for them to revise 
their State Implementation Plans (SIP). However, this revision provides 
sources permitted by the States greater certainty in application of the 
program, which should in turn reduce the overall burden of the program 
on State and local authorities. Thus, the requirements of Executive 
Order 13132 do not apply to this rule.

C. Executive Order 13175--Consultation and Coordination With Indian 
Tribal Governments

    Executive Order 13175, entitled ``Consultation and Coordination 
with Indian Tribal Governments'' (65 FR 67249, November 6, 2000), 
requires EPA to develop an accountable process to ensure ``meaningful 
and timely input by tribal officials in the development of regulatory 
policies that have tribal implications.'' We believe that this rule 
does not have tribal implications as specified in Executive Order 
13175. Thus, Executive Order 13175 does not apply.
    The purpose of today's final rule is to add greater flexibility to 
the existing major NSR regulations. These changes will benefit 
reviewing authorities and the regulated community, including any major 
source owned by a tribal government or located in or near tribal land, 
by providing increased certainty as to when the requirements of the 
major NSR program apply. Taken as a whole, today's rule should result 
in no added burden or compliance costs and should not substantially 
change the level of environmental performance achieved under the 
previous rules and guidance.
    We anticipate that initially these changes will result in a small 
increase in the burden imposed upon reviewing authorities in order for 
them to be included in the State's SIP. Nevertheless, these options and 
revisions will ultimately provide greater operational flexibility to 
sources permitted by the States, which will in turn reduce the overall 
burden on the program on State and local authorities by reducing the 
number of required permit modifications. In comparison, no tribal 
government currently has an approved Tribal Implementation Plan (TIP) 
under the CAA to implement the NSR program. The Federal government is 
currently the NSR reviewing authority in Indian country. Thus, tribal 
governments should not experience added burden, nor should their laws 
be affected with respect to implementation of this rule. Additionally, 
although major stationary sources affected by today's rule could be 
located in or near Indian country and/or be owned or operated by tribal 
governments, such affected sources would not incur additional costs or 
compliance burdens as a result of this rule. Instead, the only effect 
on such sources should be the benefit of the added certainty and 
flexibility provided by the rule.
    We recognize the importance of including tribal outreach as part of 
the rulemaking process. In addition to affording tribes an opportunity 
to comment on this rule through the proposal, on which two tribes did 
submit comments, we have also alerted tribes of this action through our 
website and quarterly newsletter. To this point we have not 
specifically consulted with tribal officials on this rule, but we are 
committed to work with any tribal government to resolve any issues that 
we may have overlooked in today's rules and that may have an adverse 
impact in Indian country.

D. Executive Order 13045--Protection of Children From Environmental 
Health Risks and Safety Risks

    Executive Order 13045, ``Protection of Children from Environmental 
Health Risks and Safety Risks'' (62 FR 19885, April 23, 1997) applies 
to any rule that (1) is determined to be ``economically significant'' 
as defined under Executive Order 12866, and (2) concerns an 
environmental health or safety risk that EPA has reason to believe may 
have a disproportionate effect on children. If the regulatory action 
meets both criteria, we must evaluate the environmental health or 
safety effects of the planned rule on children and explain why the 
planned regulation is preferable to other

[[Page 61275]]

potentially effective and reasonable alternatives that we considered.
    This rule is not subject to Executive Order 13045, because we do 
not have reason to believe the environmental health or safety risks 
addressed by this action present a disproportionate risk to children. 
We believe that, based on our analysis of electric utilities, this rule 
as a whole will result in equal or better environmental protection than 
currently provided by the existing regulations, and do so in a more 
streamlined and effective manner.

E. Paperwork Reduction Act

    The information collection requirements in this final rule have 
been submitted for approval to OMB under the requirements of the 
Paperwork Reduction Act, 44 U.S.C. 3501 et seq. An ICR document has 
been prepared by EPA (ICR No. 1230.14), and a copy may be obtained from 
Susan Auby, U.S. Environmental Protection Agency, Office of 
Environmental Information, Collection Strategies Division (2822T), 1200 
Pennsylvania Avenue, NW., Washington, DC 20460-0001, by e-mail at 
[email protected], or by calling (202) 566-1672. A copy may also be 
downloaded off the Internet at http://www.epa.gov/icr. The information 
requirements included in ICR No. 1230.14 are not enforceable until OMB 
approves them.
    The information that ICR No. 1230.14 covers is required for the 
submittal of a complete permit application for the construction or 
modification of all major new stationary sources of pollutants in 
attainment and nonattainment areas, as well as for applicable minor 
stationary sources of pollutants. This information collection is 
necessary for the proper performance of EPA's functions, has practical 
utility, and is not unnecessarily duplicative of information we 
otherwise can reasonably access. We have reduced, to the extent 
practicable and appropriate, the burden on persons providing the 
information to or for EPA. In fact, we feel that this rule will result 
in less burden on industry and reviewing authorities since it 
streamlines the process of determining whether a replacement activity 
is RMRR.
    However, according to ICR No. 1230.14, we do anticipate an initial 
increase in burden for reviewing authorities as a result of the rule 
changes, to account for revising state implementation plans to 
incorporate these rule changes. As discussed above, we expect those 
one-time expenditures to be limited to $580,000 for the estimated 112 
affected reviewing authorities. For the number of respondent reviewing 
authorities, the analysis uses the 112 reviewing authorities count used 
by other permitting ICR's for the one-time tasks (for example, SIP 
revisions).
    Burden means the total time, effort, or financial resources 
expended by persons to generate, maintain, retain, or disclose or 
provide information to or for a Federal agency. This includes the time 
needed to review instructions; develop, acquire, install, and utilize 
technology and systems for the purpose of responding to the information 
collection; adjust existing ways to comply with any previously 
applicable instructions and requirements; train personnel to respond to 
a collection of information; search existing data sources; complete and 
review the collection of information; and transmit or otherwise 
disclose the information.
    An agency may not conduct or sponsor, and a person is not required 
to respond to, a collection of information unless it displays a 
currently valid OMB control number. The OMB control numbers for EPA's 
regulations are listed in 40 CFR part 9 and 48 CFR chapter 15. We will 
continue to present OMB control numbers in a consolidated table format 
to be codified in 40 CFR part 9 of the Agency's regulations, and in 
each CFR volume containing EPA regulations. The table lists the section 
numbers with reporting and recordkeeping requirements, and the current 
OMB control numbers. This listing of the OMB control numbers and their 
subsequent codification in the CFR satisfy the requirements of the 
Paperwork Reduction Act (44 U.S.C. 3501 et seq.) and OMB's implementing 
regulations at 5 CFR part 1320.

F. Regulatory Flexibility Analysis

    We determined it is not necessary to prepare a regulatory 
flexibility analysis in connection with this final rule. We have also 
determined that this rule will not have a significant economic impact 
on a substantial number of small entities. For purposes of assessing 
the impacts of today's rule on small entities, small entity is defined 
as: (1) Any small business employing fewer than 500 employees; (2) a 
small governmental jurisdiction that is a government of a city, county, 
town, school district or special district with a population of less 
than 50,000; and (3) a small organization that is any not-for-profit 
enterprise which is independently owned and operated and is not 
dominant in its field.
    After considering the economic impacts of today's rule on small 
entities, EPA has concluded that this action will not have a 
significant economic impact on a substantial number of small entities. 
In determining whether a rule has a significant economic impact on a 
substantial number of small entities, the impact of concern is any 
significant adverse economic impact on small entities, since the 
primary purpose of the regulatory flexibility analyses is to identify 
and address regulatory alternatives ``which minimize any significant 
economic impact of this rule on small entities.'' 5 U.S.C. Sections 603 
and 604. Thus, an agency may conclude that a rule will not have a 
significant economic impact on a substantial number of small entities 
if the rule relieves regulatory burden, or otherwise has a positive 
economic effect on all of the small entities subject to the rule. 
Today's rule will not have a significant economic impact on a 
substantial number of small entities because it will decrease the 
regulatory burden of the existing regulations and have a positive 
effect on all small entities subject to the rule. This rule improves 
operational flexibility for owners or operators of major stationary 
sources and clarifies applicable requirements for determining if a 
change qualifies as a major modification. We have therefore concluded 
that today's rule will relieve regulatory burden for all small 
entities.

G. Unfunded Mandates Reform Act of 1995

    Title II of the Unfunded Mandates Reform Act of 1995 (UMRA), Public 
Law 104-4, establishes requirements for Federal agencies to assess the 
effects of their regulatory actions on State, local, and tribal 
governments and the private sector. Under section 202 of UMRA, we 
generally must prepare a written statement, including a cost-benefit 
analysis, for proposed and final rules with ``Federal mandates'' that 
may result in expenditures to State, local, and tribal governments, in 
the aggregate, or to the private sector of $100 million or more in any 
one year. Before promulgating an EPA rule for which a written statement 
is needed, section 205 of the UMRA generally requires us to identify 
and consider a reasonable number of regulatory alternatives and adopt 
the least costly, most cost-effective or least burdensome alternative 
that achieves the objectives of the rule. The provisions of section 205 
do not apply when they are inconsistent with applicable law. Moreover, 
section 205 allows us to adopt an alternative other than the least 
costly, most cost-effective, or least burdensome alternative if the 
Administrator publishes with the final rule an explanation why that 
alternative was not adopted.

[[Page 61276]]

    Before we establish any regulatory requirements that may 
significantly or uniquely affect small governments, including tribal 
governments, we must have developed under section 203 of the UMRA a 
small government agency plan. The plan must provide for notifying 
potentially affected small governments, enabling officials of affected 
small governments to have meaningful and timely input in the 
development of our regulatory proposals with significant Federal 
intergovernmental mandates, and informing, educating, and advising 
small governments on compliance with the regulatory requirements.
    We believe these rule changes will actually reduce the regulatory 
burden associated with the major NSR program by improving the 
operational flexibility of owners or operators and clarifying the 
requirements. Because the program changes provided in the rule are not 
expected to result in a significant increase in the expenditure by 
State, local, and tribal governments, or the private sector, we have 
not prepared a budgetary impact statement or specifically addressed the 
selection of the least costly, most cost-effective, or least burdensome 
alternative. Because small governments will not be significantly or 
uniquely affected by this rule, we are not required to develop a plan 
with regard to small governments. Therefore, this rule is not subject 
to the requirements of section 203 of the UMRA.

H. National Technology Transfer and Advancement Act of 1995

    Section 12(d) of the National Technology Transfer and Advancement 
Act of 1995 (NTTAA), Public Law No. 104-113, section 12(d) (15 U.S.C. 
272 note) directs us to use voluntary consensus standards (VCS) in our 
regulatory activities unless to do so would be inconsistent with 
applicable law or otherwise impractical. VCS are technical standards 
(for example, materials specifications, test methods, sampling 
procedures, and business practices) that are developed or adopted by 
voluntary consensus standards bodies. The NTTAA directs us to provide 
Congress, through OMB, explanations when the Agency decides not to use 
available and applicable VCS.
    Although this rule does involve the use of technical standards, it 
does not preclude the State, local, and tribal reviewing agencies from 
using VCS. Today's rule is an improvement of the existing NSR 
permitting program. As such, it only ensures that promulgated technical 
standards are considered and appropriate controls are installed, prior 
to the construction of major sources of air emissions. Therefore, we 
are not considering the use of any VCS in today's rule.

I. Executive Order 13211--Actions Concerning Regulations That 
Significantly Affect Energy Supply, Distribution, or Use

    This rule is not a ``significant energy action'' as defined in 
Executive Order 13211, ``Actions Concerning Regulations That 
Significantly Affect Energy Supply, Distribution, or Use'' (66 FR 28355 
(May 22, 2001)) because it is not likely to have a significant adverse 
effect on the supply, distribution or use of energy.
    Today's rule improves the ability of sources to maintain the 
reliability of production facilities, and effectively utilize and 
improve existing capacity.

J. Executive Order 12988--Civil Justice Reform

    This final rule does not have any preemptive or retroactive effect. 
This action meets applicable standards in sections 3(a) and 3(b)(2) of 
Executive Order 12988, Civil Justice Reform, to minimize litigation, 
eliminate ambiguity, and reduce burden.

V. Effective Date for Today's Requirements

    All of these changes will take effect in the Federal PSD program 
(codified at Sec.  52.21) on December 26, 2003. This means that these 
rules will apply on December 26, 2003, in any area without an approved 
PSD program, for which we are the reviewing authority, or for which we 
have delegated our authority to issue permits to a State or local 
reviewing authority.
    To be approvable under the SIP, State and local agency programs 
implementing part C (PSD permit program in Sec.  51.166) or part D 
(nonattainment NSR permit program in Sec.  51.165) must include today's 
changes as minimum program elements. State and local agencies should 
assure that any program changes under Sec. Sec.  51.165 and 51.166 are 
consistently accounted for in other SIP planning measures. State and 
local agencies must adopt and submit revisions to their part 51 
permitting programs implementing these minimum program elements no 
later than October 27, 2006. That is, for both nonattainment and 
attainment areas, the SIP revisions must be adopted and submitted 
within 3 years from today. The CAA does not specify a date for 
submission of SIPs when we revise the PSD and NSR rules. We believe it 
is appropriate to establish a date analogous to the date for submission 
of new SIPs when a NAAQS is promulgated or revised. Under section 
110(a)(1) of the CAA, as amended in 1990, that date is 3 years from 
promulgation or revision of the NAAQS. Accordingly, we have established 
3 years from today's revisions as the required date for submission of 
conforming SIP revisions.
    Today's rule revises the Federal PSD program located at 40 CFR 
52.21 to include the new equipment replacement provision of the RMRR 
exclusion. The part 52 regulations governing Federal permitting 
programs include the Federal PSD rule at 40 CFR 52.21 as well as the 
various sections of subparts C through DDD of part 52 that incorporate 
the Federal permitting program by reference for those jurisdictions 
where EPA applies part 52.21 as a Federal Implementation Plan because 
such jurisdictions lack an approved SIP to implement the PSD program. 
Because today's final rule adds additional paragraphs to the part 52.21 
rules, we will be revising the references in subparts C through DDD to 
appropriately reflect the program that applies. This final action will 
be taken in a separate Federal Register notice and will not change the 
effective date of today's final changes.

VI. Statutory Authority

    The statutory authority for this action is provided by sections 
101, 111, 114, 116, and 301 of the CAA as amended (42 U.S.C. 7401, 
7411, 7414, 7416, and 7601). This rulemaking is also subject to section 
307(d) of the CAA (42 U.S.C. 7407(d)).

List of Subjects in 40 CFR Parts 51 and 52

    Environmental protection, Administrative practices and procedures, 
Air pollution control, Intergovernmental relations.

    Dated: August 27, 2003.
Marianne Lamont Horinko,
Acting Administrator.

0
For the reasons set out in the preamble, title 40, chapter I of the 
Code of Federal Regulations is amended as follows:

PART 51--[AMENDED]

0
1. The authority citation for part 51 continues to read as follows:

    Authority: 23 U.S.C. 101; 42 U.S.C. 7401-7671q.

Subpart I--[Amended]

0
2. Section 51.165 is amended:
0
a. By revising paragraph (a)(1)(v)(C)(1).
0
b. By adding paragraphs (a)(1)(xliii) through (xlvi) and paragraph (h).
    The revision and additions read as follows:

[[Page 61277]]

Sec.  51.165  Permit requirements.

    (a) * * *
    (1) * * *
    (v) * * *
    (C) * * *
    (1) Routine maintenance, repair and replacement. Routine 
maintenance, repair and replacement shall include, but not be limited 
to, any activity(s) that meets the requirements of the equipment 
replacement provisions contained in paragraph (h) of this section;
* * * * *
    (xliii)(A) In general, process unit means any collection of 
structures and/or equipment that processes, assembles, applies, blends, 
or otherwise uses material inputs to produce or store an intermediate 
or a completed product. A single stationary source may contain more 
than one process unit, and a process unit may contain more than one 
emissions unit.
    (B) Pollution control equipment is not part of the process unit, 
unless it serves a dual function as both process and control equipment. 
Administrative and warehousing facilities are not part of the process 
unit.
    (C) For replacement cost purposes, components shared between two or 
more process units are proportionately allocated based on capacity.
    (D) The following list identifies the process units at specific 
categories of stationary sources.
    (1) For a steam electric generating facility, the process unit 
consists of those portions of the plant that contribute directly to the 
production of electricity. For example, at a pulverized coal-fired 
facility, the process unit would generally be the combination of those 
systems from the coal receiving equipment through the emission stack 
(excluding post-combustion pollution controls), including the coal 
handling equipment, pulverizers or coal crushers, feedwater heaters, 
ash handling, boiler, burners, turbine-generator set, condenser, 
cooling tower, water treatment system, air preheaters, and operating 
control systems. Each separate generating unit is a separate process 
unit.
    (2) For a petroleum refinery, there are several categories of 
process units: those that separate and/or distill petroleum feedstocks; 
those that change molecular structures; petroleum treating processes; 
auxiliary facilities, such as steam generators and hydrogen production 
units; and those that load, unload, blend or store intermediate or 
completed products.
    (3) For an incinerator, the process unit would consist of 
components from the feed pit or refuse pit to the stack, including 
conveyors, combustion devices, heat exchangers and steam generators, 
quench tanks, and fans.
    (xliv) Functionally equivalent component means a component that 
serves the same purpose as the replaced component.
    (xlv) Fixed capital cost means the capital needed to provide all 
the depreciable components. ``Depreciable components'' refers to all 
components of fixed capital cost and is calculated by subtracting land 
and working capital from the total capital investment, as defined in 
paragraph (a)(1)(xlvi) of this section.
    (xlvi) Total capital investment means the sum of the following: All 
costs required to purchase needed process equipment (purchased 
equipment costs); the costs of labor and materials for installing that 
equipment (direct installation costs); the costs of site preparation 
and buildings; other costs such as engineering, construction and field 
expenses, fees to contractors, startup and performance tests, and 
contingencies (indirect installation costs); land for the process 
equipment; and working capital for the process equipment.
* * * * *
    (h) Equipment replacement provision. Without regard to other 
considerations, routine maintenance, repair and replacement includes, 
but is not limited to, the replacement of any component of a process 
unit with an identical or functionally equivalent component(s), and 
maintenance and repair activities that are part of the replacement 
activity, provided that all of the requirements in paragraphs (h)(1) 
through (3) of this section are met.
    (1) Capital Cost threshold for Equipment Replacement. (i) For an 
electric utility steam generating unit, as defined in Sec.  
51.165(a)(1)(xx), the fixed capital cost of the replacement 
component(s) plus the cost of any associated maintenance and repair 
activities that are part of the replacement shall not exceed 20 percent 
of the replacement value of the process unit, at the time the equipment 
is replaced. For a process unit that is not an electric utility steam 
generating unit the fixed capital cost of the replacement component(s) 
plus the cost of any associated maintenance and repair activities that 
are part of the replacement shall not exceed 20 percent of the 
replacement value of the process unit, at the time the equipment is 
replaced.
    (ii) In determining the replacement value of the process unit; and, 
except as otherwise allowed under paragraph (h)(1)(iii) of this 
section, the owner or operator shall determine the replacement value of 
the process unit on an estimate of the fixed capital cost of 
constructing a new process unit, or on the current appraised value of 
the process unit.
    (iii) As an alternative to paragraph (h)(1)(ii) of this section for 
determining the replacement value of a process unit, an owner or 
operator may choose to use insurance value (where the insurance value 
covers only complete replacement), investment value adjusted for 
inflation, or another accounting procedure if such procedure is based 
on Generally Accepted Accounting Principles, provided that the owner or 
operator sends a notice to the reviewing authority. The first time that 
an owner or operator submits such a notice for a particular process 
unit, the notice may be submitted at any time, but any subsequent 
notice for that process unit may be submitted only at the beginning of 
the process unit's fiscal year. Unless the owner or operator submits a 
notice to the reviewing authority, then paragraph (h)(1)(ii) of this 
section will be used to establish the replacement value of the process 
unit. Once the owner or operator submits a notice to use an alternative 
accounting procedure, the owner or operator must continue to use that 
procedure for the entire fiscal year for that process unit. In 
subsequent fiscal years, the owner or operator must continue to use 
this selected procedure unless and until the owner or operator sends 
another notice to the reviewing authority selecting another procedure 
consistent with this paragraph or paragraph (h)(1)(ii) of this section 
at the beginning of such fiscal year.
    (2) Basic design parameters. The replacement does not change the 
basic design parameter(s) of the process unit to which the activity 
pertains.
    (i) Except as provided in paragraph (h)(2)(iii) of this section, 
for a process unit at a steam electric generating facility, the owner 
or operator may select as its basic design parameters either maximum 
hourly heat input and maximum hourly fuel consumption rate or maximum 
hourly electric output rate and maximum steam flow rate. When 
establishing fuel consumption specifications in terms of weight or 
volume, the minimum fuel quality based on British Thermal Units content 
shall be used for determining the basic design parameter(s) for a coal-
fired electric utility steam generating unit.
    (ii) Except as provided in paragraph (h)(2)(iii) of this section, 
the basic design parameter(s) for any process unit that is not at a 
steam electric generating

[[Page 61278]]

facility are maximum rate of fuel or heat input, maximum rate of 
material input, or maximum rate of product output. Combustion process 
units will typically use maximum rate of fuel input. For sources having 
multiple end products and raw materials, the owner or operator should 
consider the primary product or primary raw material when selecting a 
basic design parameter.
    (iii) If the owner or operator believes the basic design 
parameter(s) in paragraphs (h)(2)(i) and (ii) of this section is not 
appropriate for a specific industry or type of process unit, the owner 
or operator may propose to the reviewing authority an alternative basic 
design parameter(s) for the source's process unit(s). If the reviewing 
authority approves of the use of an alternative basic design 
parameter(s), the reviewing authority shall issue a permit that is 
legally enforceable that records such basic design parameter(s) and 
requires the owner or operator to comply with such parameter(s).
    (iv) The owner or operator shall use credible information, such as 
results of historic maximum capability tests, design information from 
the manufacturer, or engineering calculations, in establishing the 
magnitude of the basic design parameter(s) specified in paragraphs 
(h)(2)(i) and (ii) of this section.
    (v) If design information is not available for a process unit, then 
the owner or operator shall determine the process unit's basic design 
parameter(s) using the maximum value achieved by the process unit in 
the five-year period immediately preceding the planned activity.
    (vi) Efficiency of a process unit is not a basic design parameter.
    (3) The replacement activity shall not cause the process unit to 
exceed any emission limitation, or operational limitation that has the 
effect of constraining emissions, that applies to the process unit and 
that is legally enforceable.

0
3. Section 51.166 is amended:
0
a. By revising paragraph (b)(2)(iii)(a).
0
b. By adding paragraphs (b)(53) through (56) and paragraph (y).
    The revision and additions read as follows:


Sec.  51.166  Prevention of significant deterioration of air quality.

    (b) * * *
    (2) * * *
    (iii) * * *
    (a) Routine maintenance, repair and replacement. Routine 
maintenance, repair and replacement shall include, but not be limited 
to, any activity(s) that meets the requirements of the equipment 
replacement provisions contained in paragraph (y) of this section;
* * * * *
    (53)(i) In general, process unit means any collection of structures 
and/or equipment that processes, assembles, applies, blends, or 
otherwise uses material inputs to produce or store an intermediate or a 
completed product. A single stationary source may contain more than one 
process unit, and a process unit may contain more than one emissions 
unit.
    (ii) Pollution control equipment is not part of the process unit, 
unless it serves a dual function as both process and control equipment. 
Administrative and warehousing facilities are not part of the process 
unit.
    (iii) For replacement cost purposes, components shared between two 
or more process units are proportionately allocated based on capacity.
    (iv) The following list identifies the process units at specific 
categories of stationary sources.
    (a) For a steam electric generating facility, the process unit 
consists of those portions of the plant that contribute directly to the 
production of electricity. For example, at a pulverized coal-fired 
facility, the process unit would generally be the combination of those 
systems from the coal receiving equipment through the emission stack 
(excluding post-combustion pollution controls), including the coal 
handling equipment, pulverizers or coal crushers, feedwater heaters, 
ash handling, boiler, burners, turbine-generator set, condenser, 
cooling tower, water treatment system, air preheaters, and operating 
control systems. Each separate generating unit is a separate process 
unit.
    (b) For a petroleum refinery, there are several categories of 
process units: those that separate and/or distill petroleum feedstocks; 
those that change molecular structures; petroleum treating processes; 
auxiliary facilities, such as steam generators and hydrogen production 
units; and those that load, unload, blend or store intermediate or 
completed products.
    (c) For an incinerator, the process unit would consist of 
components from the feed pit or refuse pit to the stack, including 
conveyors, combustion devices, heat exchangers and steam generators, 
quench tanks, and fans.
    (54) Functionally equivalent component means a component that 
serves the same purpose as the replaced component.
    (55) Fixed capital cost means the capital needed to provide all the 
depreciable components. ``Depreciable components'' refers to all 
components of fixed capital cost and is calculated by subtracting land 
and working capital from the total capital investment, as defined in 
paragraph (b)(56) of this section.
    (56) Total capital investment means the sum of the following: all 
costs required to purchase needed process equipment (purchased 
equipment costs); the costs of labor and materials for installing that 
equipment (direct installation costs); the costs of site preparation 
and buildings; other costs such as engineering, construction and field 
expenses, fees to contractors, startup and performance tests, and 
contingencies (indirect installation costs); land for the process 
equipment; and working capital for the process equipment.
* * * * *
    (y) Equipment replacement provision. Without regard to other 
considerations, routine maintenance, repair and replacement includes, 
but is not limited to, the replacement of any component of a process 
unit with an identical or functionally equivalent component(s), and 
maintenance and repair activities that are part of the replacement 
activity, provided that all of the requirements in paragraphs (y)(1) 
through (3) of this section are met.
    (1) Capital Cost threshold for Equipment Replacement. (i) For an 
electric utility steam generating unit, as defined in Sec.  
51.166(b)(30), the fixed capital cost of the replacement component(s) 
plus the cost of any associated maintenance and repair activities that 
are part of the replacement shall not exceed 20 percent of the 
replacement value of the process unit, at the time the equipment is 
replaced. For a process unit that is not an electric utility steam 
generating unit the fixed capital cost of the replacement component(s) 
plus the cost of any associated maintenance and repair activities that 
are part of the replacement shall not exceed 20 percent of the 
replacement value of the process unit, at the time the equipment is 
replaced.
    (ii) In determining the replacement value of the process unit; and, 
except as otherwise allowed under paragraph (y)(1)(iii) of this 
section, the owner or operator shall determine the replacement value of 
the process unit on an estimate of the fixed capital cost of 
constructing a new process unit, or on the current appraised value of 
the process unit.
    (iii) As an alternative to paragraph (y)(1)(ii) of this section for 
determining

[[Page 61279]]

the replacement value of a process unit, an owner or operator may 
choose to use insurance value (where the insurance value covers only 
complete replacement), investment value adjusted for inflation, or 
another accounting procedure if such procedure is based on Generally 
Accepted Accounting Principles, provided that the owner or operator 
sends a notice to the reviewing authority. The first time that an owner 
or operator submits such a notice for a particular process unit, the 
notice may be submitted at any time, but any subsequent notice for that 
process unit may be submitted only at the beginning of the process 
unit's fiscal year. Unless the owner or operator submits a notice to 
the reviewing authority, then paragraph (y)(1)(ii) of this section will 
be used to establish the replacement value of the process unit. Once 
the owner or operator submits a notice to use an alternative accounting 
procedure, the owner or operator must continue to use that procedure 
for the entire fiscal year for that process unit. In subsequent fiscal 
years, the owner or operator must continue to use this selected 
procedure unless and until the owner or operator sends another notice 
to the reviewing authority selecting another procedure consistent with 
this paragraph or paragraph (y)(1)(ii) of this section at the beginning 
of such fiscal year.
    (2) Basic design parameters. The replacement does not change the 
basic design parameter(s) of the process unit to which the activity 
pertains.
    (i) Except as provided in paragraph (y)(2)(iii) of this section, 
for a process unit at a steam electric generating facility, the owner 
or operator may select as its basic design parameters either maximum 
hourly heat input and maximum hourly fuel consumption rate or maximum 
hourly electric output rate and maximum steam flow rate. When 
establishing fuel consumption specifications in terms of weight or 
volume, the minimum fuel quality based on British Thermal Units content 
shall be used for determining the basic design parameter(s) for a coal-
fired electric utility steam generating unit.
    (ii) Except as provided in paragraph (y)(2)(iii) of this section, 
the basic design parameter(s) for any process unit that is not at a 
steam electric generating facility are maximum rate of fuel or heat 
input, maximum rate of material input, or maximum rate of product 
output. Combustion process units will typically use maximum rate of 
fuel input. For sources having multiple end products and raw materials, 
the owner or operator should consider the primary product or primary 
raw material when selecting a basic design parameter.
    (iii) If the owner or operator believes the basic design 
parameter(s) in paragraphs (y)(2)(i) and (ii) of this section is not 
appropriate for a specific industry or type of process unit, the owner 
or operator may propose to the reviewing authority an alternative basic 
design parameter(s) for the source's process unit(s). If the reviewing 
authority approves of the use of an alternative basic design 
parameter(s), the reviewing authority shall issue a permit that is 
legally enforceable that records such basic design parameter(s) and 
requires the owner or operator to comply with such parameter(s).
    (iv) The owner or operator shall use credible information, such as 
results of historic maximum capability tests, design information from 
the manufacturer, or engineering calculations, in establishing the 
magnitude of the basic design parameter(s) specified in paragraphs 
(y)(2)(i) and (ii) of this section.
    (v) If design information is not available for a process unit, then 
the owner or operator shall determine the process unit's basic design 
parameter(s) using the maximum value achieved by the process unit in 
the five-year period immediately preceding the planned activity.
    (vi) Efficiency of a process unit is not a basic design parameter.
    (3) The replacement activity shall not cause the process unit to 
exceed any emission limitation, or operational limitation that has the 
effect of constraining emissions, that applies to the process unit and 
that is legally enforceable.

PART 52--[AMENDED]

0
1. The authority citation for part 52 continues to read as follows:

    Authority: 42 U.S.C. 7401, et seq.

Subpart A--[Amended]

0
2. Section 52.21 is amended:
0
a. By revising paragraph (b)(2)(iii)(a).
0
b. By adding paragraphs (b)(55) through (58) and paragraph (cc).
    The revision and additions read as follows:


Sec.  52.21  Prevention of significant deterioration of air quality.

    (b) * * *
    (2) * * *
    (iii) * * *
    (a) Routine maintenance, repair and replacement. Routine 
maintenance, repair and replacement shall include, but not be limited 
to, any activity(s) that meets the requirements of the equipment 
replacement provisions contained in paragraph (cc) of this section;
* * * * *
    (55)(i) In general, process unit means any collection of structures 
and/or equipment that processes, assembles, applies, blends, or 
otherwise uses material inputs to produce or store an intermediate or a 
completed product. A single stationary source may contain more than one 
process unit, and a process unit may contain more than one emissions 
unit.
    (ii) Pollution control equipment is not part of the process unit, 
unless it serves a dual function as both process and control equipment. 
Administrative and warehousing facilities are not part of the process 
unit.
    (iii) For replacement cost purposes, components shared between two 
or more process units are proportionately allocated based on capacity.
    (iv) The following list identifies the process units at specific 
categories of stationary sources.
    (a) For a steam electric generating facility, the process unit 
consists of those portions of the plant that contribute directly to the 
production of electricity. For example, at a pulverized coal-fired 
facility, the process unit would generally be the combination of those 
systems from the coal receiving equipment through the emission stack 
(excluding post-combustion pollution controls), including the coal 
handling equipment, pulverizers or coal crushers, feedwater heaters, 
ash handling, boiler, burners, turbine-generator set, condenser, 
cooling tower, water treatment system, air preheaters, and operating 
control systems. Each separate generating unit is a separate process 
unit.
    (b) For a petroleum refinery, there are several categories of 
process units: those that separate and/or distill petroleum feedstocks; 
those that change molecular structures; petroleum treating processes; 
auxiliary facilities, such as steam generators and hydrogen production 
units; and those that load, unload, blend or store intermediate or 
completed products.
    (c) For an incinerator, the process unit would consist of 
components from the feed pit or refuse pit to the stack, including 
conveyors, combustion devices, heat exchangers and steam generators, 
quench tanks, and fans.
    (56) Functionally equivalent component means a component that 
serves the same purpose as the replaced component.
    (57) Fixed capital cost means the capital needed to provide all the 
depreciable components. ``Depreciable

[[Page 61280]]

components'' refers to all components of fixed capital cost and is 
calculated by subtracting land and working capital from the total 
capital investment, as defined in paragraph (b)(58) of this section.
    (58) Total capital investment means the sum of the following: all 
costs required to purchase needed process equipment (purchased 
equipment costs); the costs of labor and materials for installing that 
equipment (direct installation costs); the costs of site preparation 
and buildings; other costs such as engineering, construction and field 
expenses, fees to contractors, startup and performance tests, and 
contingencies (indirect installation costs); land for the process 
equipment; and working capital for the process equipment.
* * * * *
    (cc) Without regard to other considerations, routine maintenance, 
repair and replacement includes, but is not limited to, the replacement 
of any component of a process unit with an identical or functionally 
equivalent component(s), and maintenance and repair activities that are 
part of the replacement activity, provided that all of the requirements 
in paragraphs (cc)(1) through (3) of this section are met.
    (1) Capital cost threshold for equipment replacement. (i) For an 
electric utility steam generating unit, as defined in Sec.  
52.21(b)(31), the fixed capital cost of the replacement component(s) 
plus the cost of any associated maintenance and repair activities that 
are part of the replacement shall not exceed 20 percent of the 
replacement value of the process unit, at the time the equipment is 
replaced. For a process unit that is not an electric utility steam 
generating unit the fixed capital cost of the replacement component(s) 
plus the cost of any associated maintenance and repair activities that 
are part of the replacement shall not exceed 20 percent of the 
replacement value of the process unit, at the time the equipment is 
replaced.
    (ii) In determining the replacement value of the process unit; and, 
except as otherwise allowed under paragraph (cc)(1)(iii) of this 
section, the owner or operator shall determine the replacement value of 
the process unit on an estimate of the fixed capital cost of 
constructing a new process unit, or on the current appraised value of 
the process unit.
    (iii) As an alternative to paragraph (cc)(1)(ii) of this section 
for determining the replacement value of a process unit, an owner or 
operator may choose to use insurance value (where the insurance value 
covers only complete replacement), investment value adjusted for 
inflation, or another accounting procedure if such procedure is based 
on Generally Accepted Accounting Principles, provided that the owner or 
operator sends a notice to the reviewing authority. The first time that 
an owner or operator submits such a notice for a particular process 
unit, the notice may be submitted at any time, but any subsequent 
notice for that process unit may be submitted only at the beginning of 
the process unit's fiscal year. Unless the owner or operator submits a 
notice to the reviewing authority, then paragraph (cc)(1)(ii) of this 
section will be used to establish the replacement value of the process 
unit. Once the owner or operator submits a notice to use an alternative 
accounting procedure, the owner or operator must continue to use that 
procedure for the entire fiscal year for that process unit. In 
subsequent fiscal years, the owner or operator must continue to use 
this selected procedure unless and until the owner or operator sends 
another notice to the reviewing authority selecting another procedure 
consistent with this paragraph or paragraph (cc)(1)(ii) of this section 
at the beginning of such fiscal year.
    (2) Basic design parameters. The replacement does not change the 
basic design parameter(s) of the process unit to which the activity 
pertains.
    (i) Except as provided in paragraph (cc)(2)(iii) of this section, 
for a process unit at a steam electric generating facility, the owner 
or operator may select as its basic design parameters either maximum 
hourly heat input and maximum hourly fuel consumption rate or maximum 
hourly electric output rate and maximum steam flow rate. When 
establishing fuel consumption specifications in terms of weight or 
volume, the minimum fuel quality based on British Thermal Units content 
shall be used for determining the basic design parameter(s) for a coal-
fired electric utility steam generating unit.
    (ii) Except as provided in paragraph (cc)(2)(iii) of this section, 
the basic design parameter(s) for any process unit that is not at a 
steam electric generating facility are maximum rate of fuel or heat 
input, maximum rate of material input, or maximum rate of product 
output. Combustion process units will typically use maximum rate of 
fuel input. For sources having multiple end products and raw materials, 
the owner or operator should consider the primary product or primary 
raw material when selecting a basic design parameter.
    (iii) If the owner or operator believes the basic design 
parameter(s) in paragraphs (cc)(2)(i) and (ii) of this section is not 
appropriate for a specific industry or type of process unit, the owner 
or operator may propose to the reviewing authority an alternative basic 
design parameter(s) for the source's process unit(s). If the reviewing 
authority approves of the use of an alternative basic design 
parameter(s), the reviewing authority shall issue a permit that is 
legally enforceable that records such basic design parameter(s) and 
requires the owner or operator to comply with such parameter(s).
    (iv) The owner or operator shall use credible information, such as 
results of historic maximum capability tests, design information from 
the manufacturer, or engineering calculations, in establishing the 
magnitude of the basic design parameter(s) specified in paragraphs 
(cc)(2)(i) and (ii) of this section.
    (v) If design information is not available for a process unit, then 
the owner or operator shall determine the process unit's basic design 
parameter(s) using the maximum value achieved by the process unit in 
the five-year period immediately preceding the planned activity.
    (vi) Efficiency of a process unit is not a basic design parameter.
    (3) The replacement activity shall not cause the process unit to 
exceed any emission limitation, or operational limitation that has the 
effect of constraining emissions, that applies to the process unit and 
that is legally enforceable.

[FR Doc. 03-26320 Filed 10-24-03; 8:45 am]
BILLING CODE 6560-50-P