[Federal Register Volume 89, Number 35 (Wednesday, February 21, 2024)]
[Proposed Rules]
[Pages 13022-13032]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 2024-03528]



[[Page 13022]]

=======================================================================
-----------------------------------------------------------------------

ENVIRONMENTAL PROTECTION AGENCY

40 CFR Part 52

[EPA-R03-OAR-2024-0047; FRL-9920-01-R3]


Air Plan Disapproval; Pennsylvania; Reasonably Available Control 
Technology Case-by-Case Permits for Keystone, Conemaugh and Homer City 
Generating Facilities for the 1997 and 2008 Ozone National Ambient Air 
Quality Standards

AGENCY: Environmental Protection Agency (EPA).

ACTION: Proposed rule.

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

SUMMARY: The Environmental Protection Agency (EPA) is proposing to 
disapprove state implementation plan (SIP) revisions submitted by the 
Pennsylvania Department of Environmental Protection (PADEP) on behalf 
of the Commonwealth of Pennsylvania (Pennsylvania). PADEP submitted SIP 
revisions for the Keystone, Conemaugh and Homer City electric 
generating facilities on May 26, 2022 to address certain reasonably 
available control technique (RACT) requirements for the 1997 and 2008 
ozone national ambient air quality standards (NAAQS). EPA is proposing 
to disapprove the May 26, 2022 SIP revisions for these facilities as 
the SIPs contain problematic provisions and fail to justify the 
selection of permit limits as RACT consistent with applicable 
requirements and case law. This action is being taken under the Clean 
Air Act (CAA).

DATES: Written comments must be received on or before March 22, 2024.

ADDRESSES: Submit your comments, identified by Docket ID No. EPA-R03-
OAR-2024-0047 at https://www.regulations.gov, or via email to 
[email protected]. For comments submitted at Regulations.gov, follow 
the online instructions for submitting comments. Once submitted, 
comments cannot be edited or removed from Regulations.gov. For either 
manner of submission, EPA may publish any comment received to its 
public docket. Do not submit electronically any information you 
consider to be confidential business information (CBI) or other 
information whose disclosure is restricted by statute. Multimedia 
submissions (audio, video, etc.) must be accompanied by a written 
comment. The written comment is considered the official comment and 
should include discussion of all points you wish to make. EPA will 
generally not consider comments or comment contents located outside of 
the primary submission (i.e. on the web, cloud, or other file sharing 
system). For additional submission methods, please contact the person 
identified in the FOR FURTHER INFORMATION CONTACT section. For the full 
EPA public comment policy, information about CBI or multimedia 
submissions, and general guidance on making effective comments, please 
visit https://www.epa.gov/dockets/commenting-epa-dockets.

FOR FURTHER INFORMATION CONTACT: Sean Silverman, Planning & 
Implementation Branch (3AD30), Air & Radiation Division, U.S. 
Environmental Protection Agency, Region III, 1600 John F. Kennedy 
Boulevard, Philadelphia, Pennsylvania 19103. The telephone number is 
(215) 814-5511. Mr. Silverman can also be reached via electronic mail 
at [email protected].

SUPPLEMENTARY INFORMATION:

I. Background

    The RACT requirements in CAA section 182(b)(2) apply to all ozone 
nonattainment areas classified as Moderate or higher (i.e. Serious, 
Severe, or Extreme). Section 184(b)(1)(B) of the CAA also applies RACT 
to all areas located within ozone transport regions. The entire 
Commonwealth of Pennsylvania is part of the Ozone Transport Region 
(OTR) established by section 184 of the CAA and therefore subject 
statewide to RACT requirements.
    On May 16, 2016, Pennsylvania submitted a SIP revision intended to 
satisfy CAA sections 182(b)(2)(C), 182(f), and 184 for the 1997 and 
2008 8-hour ozone NAAQS for all major sources of nitrogen oxides 
(NOX) and volatile organic compounds (VOCs) in Pennsylvania 
not subject to control techniques guidelines (CTGs), with a few 
exceptions not relevant to this action. On May 9, 2019, EPA published a 
final action fully approving certain provisions and conditionally 
approving other portions of Pennsylvania's May 16, 2016, SIP submission 
to implement RACT for the 1997 and 2008 Ozone NAAQS (hereafter the 
``RACT II rule''). 84 FR 20274 (May 9, 2019). Specifically, EPA's 
action fully approved sections 121.1, 129.96, 129.97, and 129.100 of 
Title 25 of the Pennsylvania Code (25 Pa. Code) as meeting certain 
aspects of major stationary source RACT in CAA sections 172, 182, and 
184 for the 1997 and 2008 ozone NAAQS, and conditionally approved 25 
Pa. Code sections 129.98 and 129.99 following a commitment provided by 
Pennsylvania to submit additional SIP revisions to address the 
deficiencies identified by EPA in the May 16, 2016 SIP revision. Id. at 
20290.
    On August 27, 2020, the Third Circuit held unlawful and vacated 
EPA's approval of certain SIP provisions challenged by the Sierra Club. 
Sierra Club v. EPA, 972 F.3d 290 (3rd Cir. 2020) (``Sierra Club''). The 
case related to EPA's approval of only that portion of the RACT II rule 
applicable to coal-fired electricity generating units (EGUs) equipped 
with selective catalytic reduction (SCR) for control of NOX. 
Specifically at issue was EPA's approval of the presumptive RACT 
NOX limit for these EGUs of 0.12 pounds of NOX 
per Million British Thermal Units (MMBtu) of heat input (lbs/MMBtu) 
when the inlet temperature to the SCR was 600 degrees Fahrenheit or 
above, found at 25 Pa. Code 129.97(g)(1)(viii); the application of the 
less stringent NOX limits of 25 Pa Code 129.97(g)(1)(vi) to 
EGUs with SCR when the inlet temperature to the SCR was below 600 
degrees Fahrenheit; and the failure of the RACT II rule at 25 Pa. Code 
129.100(d) to specifically require these EGUs to keep temperature data 
for the inlet temperature to the SCRs and report that data to PADEP.
    The Court explained that, while RACT does not require the lowest 
achievable emissions limit, Pennsylvania's adoption of a limit derived 
from the average historical NOX emissions of the units at 
these EGUs, without more, was insufficient. The record showed that 
certain units within Pennsylvania were capable of achieving 
significantly lower rates of NOX emissions. The Court found 
that EPA did not sufficiently explain why a lower standard was 
infeasible. Sierra Club, 972 F.3d at 299-303. Second, the Court held 
that Pennsylvania's standard acted as a loophole because it permitted 
unlimited operations without the use of SCR controls if exhaust gas 
temperature was kept below what the Court considered an arbitrary 
temperature threshold of 600 degrees Fahrenheit. Id. at 303-07. Third, 
the Court held that Pennsylvania's reporting requirements were not 
enforceable. Id. at 307-09.
    Consequently, the Court vacated EPA's approval of this portion of 
the 2016 SIP and ordered EPA either to approve a revised, compliant SIP 
or promulgate a FIP within two years (i.e., by August 27, 2022). Sierra 
Club at 309. The Court stated that the new standard--SIP or FIP--``must 
be technology forcing, in accord with [EPA's] RACT standard, and lack 
the gaping loophole found in the [2016 SIP's] enforcement regime.'' Id. 
On August 16, 2022, EPA took final action

[[Page 13023]]

to disapprove the vacated portions of the May 19, 2020 approval. 87 FR 
50257. EPA published its proposed FIP on May 25, 2022. 87 FR 31798. EPA 
issued a FIP on August 31, 2022. 87 FR 53381.
    Following the Court's decision, PADEP required that by April 1, 
2021, each source within a facility which had been subject to the 
presumptive 0.12 lb/MMBtu limit submit a permit application in 
accordance with 25 Pa. Code 129.99 setting forth a RACT analysis for 
each unit at the facility.\1\ On or about April 1, 2021, Conemaugh, 
Homer City, Keystone, and Montour submitted permit applications to 
PADEP with RACT analyses.\2\ PADEP found the permit applications to be 
technically deficient and therefore issued technical deficiency letters 
to each of these sources seeking additional information. Although the 
sources submitted additional information, PADEP decided that it would 
do its own case-by-case RACT analysis for each EGU at each facility and 
propose new RACT limits for each EGU in amended title V permits. Once 
these permits became final, PADEP intended to submit each permit to EPA 
as a SIP revision to meet the RACT requirement for each source. EPA 
also continued to regularly discuss with PADEP their efforts to develop 
case-by-case RACT/title V permits for these sources.
---------------------------------------------------------------------------

    \1\ See ``Conemaugh RACT II Review Memo'' p. 2, ``Homer City 
RACT II Review Memo'' p. 3 and ``Keystone RACT II Review Memo'' p. 
2, available in the docket of this action.
    \2\ The Bruce Mansfield EGUs ceased all operations prior to 
April 1, 2021 and therefore did not submit a RACT permit 
application.
---------------------------------------------------------------------------

    From September 11, 2021, through November 6, 2021, PADEP serially 
issued draft RACT/title V permits for four sources, while Allegheny 
County issued a draft RACT/title V permit for Cheswick in December 
2021.\3\ EPA submitted timely comments on each draft permit. Many of 
the concerns and issues identified in EPA's first set of comments 
(which was on the Keystone permit) appeared again in the draft permits 
for the other sources. EPA's comments raised significant concerns over 
the approvability of each permit because each remained inconsistent 
with the court's decision, and PADEP did not address those concerns 
with each subsequent draft permit it published for comment. On May 26, 
2022, PADEP submitted case-by-case RACT determinations to EPA as a 
revision to the Pennsylvania SIP which still contained the 
approvability issues EPA had flagged in its comments, for Keystone, 
Conemaugh, and Homer City. PADEP submitted a case-by-case RACT 
determination for Montour as a revision to the Pennsylvania SIP on June 
9, 2022, but subsequently formally withdrew it.\4\ In addition, the 
Cheswick facility permanently ceased operations and surrendered all of 
its air permits to the Allegheny County Health Department.\5\ The Homer 
City facility also ceased all coal-burning operations on July 1, 
2023.\6\ Prior to July 1st, only Unit 3 at Homer City was operating. 
However, because Homer City has not formally surrendered its CAA 
permits, which would demonstrate that the shutdown is permanent, and 
because PADEP has not withdrawn the SIP submission with regard to Homer 
City, EPA will continue to consider the approvability of the RACT 
NOX limits for Homer City.
---------------------------------------------------------------------------

    \3\ See 51 Pa.B. 5834, September 11, 2021 (Keystone); 51 Pa.B. 
6259, October 2, 2021 (Conemaugh); 51 Pa.B. 6558, October 16, 2021 
(Homer City); 51 Pa.B. 6930, November 6, 2021 (Montour); Allegheny 
County Health Department Public Notices, December 2, 2021 
(Cheswick).
    \4\ See document dated October 26, 2022 from EPA Region III to 
Acting Secretary Ramez Ziadeh of PADEP available in the docket of 
this action.
    \5\ See document dated April 15, 2022 from Allegheny County to 
Lee Bahl of GenOn Holdings LLC available in the docket of this 
action.
    \6\ See https://www.pjm.com/planning/service-requests/gen-deactivations.
---------------------------------------------------------------------------

    EPA notes that the May 2022 permits for Keystone and Conemaugh also 
contain case-by-case RACT limits for certain gas or oil-fired auxiliary 
boilers at these facilities. However, EPA is not taking action at this 
time on the case-by-case RACT limits in these permits for two auxiliary 
boilers at Keystone (Source IDs 037 and 038) and the two at Conemaugh 
(Source IDs 039 and 041). These auxiliary boilers were not subject to 
the presumptive RACT limit in 25 Pa. Code 129.97(g)(1)(viii) for which 
EPA issued a final disapproval in August 2022.

II. Summary of the Case-by-Case Permit SIP Revisions

    EPA notes that the RACT limits in PADEP's May 2022 SIP submittal 
addressed RACT limits for the large Electric Generating Units (EGUs) at 
Conemaugh, Keystone, and Homer City for only the 1997 and 2008 ozone 
NAAQS. These source-specific limits were established pursuant to 25 Pa. 
Code 129.99, which was conditionally approved by EPA in March 2019. 
Section 129.99 of 25 Pa. Code allows a source to apply for an 
alternative RACT limit (a.k.a. ``case-by-case'' RACT limit) to the 
otherwise default (a.k.a. ``presumptive'') RACT limits where 
appropriate. In response to comments,\7\ PADEP affirmed that the RACT 
limits for the EGUs at these three sources do not address the 2015 
ozone NAAQS.\8\
---------------------------------------------------------------------------

    \7\ All three Response to Comments (RTC) documents are in the 
docket for this matter. The Conemaugh Response to Comments (Con RTC) 
and Keystone Response to Comments (Key RTC) are both dated May 12, 
2022. The Homer City Response to Comments (HC RTC) in the official 
SIP submission is marked ``Draft'' and does not contain a date.
    \8\ See, e.g., Conemaugh Response to Comments, p. 2: 
``incorporates the provisions and requirements contained in the 
amended RACT II approval for the facility, which are intended to 
satisfy the [CAA] RACT requirements for the 1997 and 2008 . . . 
ozone [NAAQS].''
---------------------------------------------------------------------------

Summary of Pennsylvania's Process for Setting Limits

    PADEP developed the NOX limits for each of the EGUs at 
each facility using a similar methodology, which included using similar 
years of data. Table 1 in this document summarizes the three 
NOX emission rates applicable to each unit at each facility 
as proposed by Pennsylvania for public comment, and the final limits in 
the permits submitted by PADEP for approval as SIP revisions.

                                            Table 1--Proposed Rates and Final Rates in 2022 PA SIP Submission
--------------------------------------------------------------------------------------------------------------------------------------------------------
                                                                          Proposed limits                        Submitted as SIP revision (final)
                                                         -----------------------------------------------------------------------------------------------
           Facility               Unit       Capacity       SCR on lb/    All conditions  All conditions    SCR on lb/    All conditions  All conditions
                                                            MMBtu daily   lb/MMBtu daily   lb/hr 30-day     MMBtu daily   lb/MMBtu daily   lb/hr 30-day
                                                               avg.            avg.            avg.            avg.            avg.            avg.
--------------------------------------------------------------------------------------------------------------------------------------------------------
Conemaugh.....................  1.......           8,280           0.070            0.27             700           0.070            0.27             700
                                2.......           8,280           0.070            0.27             700           0.070            0.27             700
Keystone......................  1.......           8,717           0.080            0.30             800           0.080            0.30             770
                                2.......           8,717           0.080            0.30             800           0.080            0.30             770
Homer City....................  1.......           6,792           0.080            0.45             550           0.080            0.45             600
                                2.......           6,792           0.080            0.45             550           0.080            0.45             600
                                3.......           7,260           0.070            0.27             510           0.070            0.27             560
--------------------------------------------------------------------------------------------------------------------------------------------------------


[[Page 13024]]

    A technical evaluation memo (TEM) accompanying each draft permit 
issued for public comment provided an initial explanation for PADEP's 
methodology for determining the proposed RACT level of controls for 
each facility.\9\ For each unit at each facility, PADEP states that it 
followed a ``top-down'' approach to determine NOX emissions 
limits, which included searching for and identifying the ``best 
methodology, technique, technology, or other means for reducing 
NOX while factoring environmental, energy and economic 
considerations into the analysis.'' Con TEM, p. 2; Key TEM, p. 2; HC 
TEM, p. 2. This included identifying the controls installed on coal-
fired units in some other states. PADEP then used the EPA Control Cost 
Manual (sixth edition), June 12, 2019, and sometimes vendor's quotes, 
to determine whether control options PADEP identified as technically 
feasible were also cost effective. Con TEM, pp. 2-3; Key TEM, p. 2; HC 
TEM, p. 3. PADEP performed some type of analysis for multiple 
NOX control technologies \10\ for each facility before 
``determin[ing] that no additional controls are cost effective.'' Con 
TEM at 3. See, e.g., Con TEM pp. 3-19. In lieu of new controls, PADEP 
determined that for each of the three facilities, changes to the way 
the facilities operated their SCRs and changes to how they ``tuned'' 
the boilers were the only technically available and cost-effective 
controls for reducing NOX emissions. Con TEM, pp. 2-3; Key 
TEM, pp. 2-3; HC TEM, pp. 2-3. This approach resulted in PADEP adopting 
three separate but related limits for each EGU at each of the three 
facilities. The proposed and final rates are in Table 1 of this 
document. The method that PADEP used to arrive at each of the three 
rates is summarized below.
---------------------------------------------------------------------------

    \9\ All three technical evaluation memos (TEMs) are in the 
docket for this matter. The Homer City technical evaluation memo (HC 
TEM) is dated October 14, 2021. The Conemaugh technical evaluation 
memo (Con TEM) is dated September 28, 2021. The Keystone technical 
evaluation memo (Key TEM) is dated August 25, 2021.
    \10\ Potential controls evaluated included: Precombustion 
Controls (Switching to Natural Gas, Switching from high to low 
emitting or zero emitting units), Combustion Controls (Partial or 
full oxy firing, Oxygen enhanced combustion, LNB installation, LNB 
Optimization, LNB Upgrade, Flue Gas Recirculation (FGR), Separated 
overfired air, Rotating opposed fire air) Post Combustion Controls 
(Additional SCR, SCR Optimization, Economizer Bypass during low 
load, startup, and shutdown to allow SCR operation, V-temp 
economizer during low load, startup, and shutdown to allow SCR 
operation, Flue gas reheat during low load, startup, and shutdown to 
allow SCR operation, Dry sorbent injection prior to SCR during low 
load conditions to allow SCR operation, addition of Selective Non-
Catalytic Reduction (SNCR), SNCR Optimization, Return of partially 
operating SCR and SNCR systems to full operation) Station Wide 
Improvements (Installation/improvement of digital process controls 
on equipment to minimize NOX emissions and detect 
equipment in need to maintenance, Improved/increased equipment 
cleaning and maintenance practices). See Con TEM p. 4-19; Key TEM, 
pp 3-17; HC TEM pp. 3-21.
---------------------------------------------------------------------------

Selection of SCR-on lb/MMBtu Daily Average Rates

    To determine the ``SCR-on'' limit representing RACT for when the 
SCRs are operating, PADEP ``analyzed daily NOX emissions 
rates from EPA's Clean Air Markets Division (CAMD) database at varying 
operating load conditions'' for most of the units at each facility. Con 
TEM, p. 3; HC TEM pp. 4-13; Key TEM p. 3. PADEP examined data for each 
facility ranging from 2016 to 2020, depending on the facility.\11\ For 
certain months within the 2016-2020 time frame, PADEP states that it 
analyzed the percentage of daily heat input and corresponding 
percentage of daily reagent injection for a unit or units at each 
facility to ascertain how heat input and reagent input affected daily 
NOX emissions, and to determine the lowest emission limit 
each unit could technically and economically achieve with the SCRs. 
PADEP's analysis included examining the percentage of maximum heat rate 
input for the unit for each day of certain months and the corresponding 
percentage of maximum ammonia (the reagent used) input observed per day 
for the same month. Con TEM, p. 4; Key TEM, pp. 3-8; HC TEM pp. 4-13. 
From this data, PADEP identified a ``load'' or heat input level at 
which it seemed that Conemaugh stopped injecting ammonia into the flue 
gas stream, see Con. TEM, p. 6, but did not identify loads or heat 
inputs at which Homer City's or Keystone's units stopped injecting 
ammonia. See HC TEM pp. 4-13, Key TEM pp. 3-8. PADEP also presented 
NOX emission rate data for certain months for each source 
during various ozone seasons, and for some periods outside of ozone 
seasons.\12\ For each facility, PADEP found that the automated controls 
that run the SCRs seemed to be set at an emissions ``set point,'' 
expressed as pounds of NOX per million Btus of heat input 
(lb NOX/MMBtu), and that these set points varied over 
time.\13\ For Conemaugh, PADEP concluded that ``additional emission 
reductions would be achieved if the operator operated the SCR with a 
lower emissions setpoint while the SCR is running.'' Con. TEM, p. 6. 
For Keystone, PADEP stated ``[b]oth units at Keystone seem to be able 
to achieve a NOX rate of 0.060 lb/MMBtu on a daily average 
basis,'' but cited ``varying load conditions and other factors'' as 
affecting SCR performance and therefore proposed (and finalized) a 
daily average SCR-on rate of 0.08 lb/MMBtu for both. Key TEM, p. 6. 
PADEP does not provide further information on what these other factors 
are or what impact they and load conditions would have that lead to the 
selection of the 0.08 lb/MMBtu limit. For Homer City units 1 and 2, 
PADEP's analysis concluded that the facility seemed to be targeting a 
NOX emission rate of 0.10 lb/MMBtu when the SCRs were 
operating, but identified ``rare'' periods where the units achieved 
rates below 0.05 lb/MMBtu. HC TEM, p. 5. However, PADEP concluded that 
``[d]espite the fact that emissions under 0.10 lb/MMBtu are possible 
under at least some operating conditions, accounting for other 
operating condition requires a limit above the minimum achievable.'' HC 
TEM, p. 8. PADEP therefore proposed (and finalized) a 0.080 lb/MMBtu 
daily average operating rate for Units 1 and 2 when the SCR is 
operating. For Unit 3, PADEP found that it was also targeting a 
NOX emission rate of 0.10 lb/MMBtu, but during July 2019 was 
able to consistently achieve NOX rates between 0.08 and 0.09 
lb/MMBtu despite daily load swings. HC TEM, p. 9-10. PADEP identified 
two other instances where Unit 3 was capable of achieving 
NOX rates lower than 0.08, but did not identify the lowest 
achievable SCR-on rate before determining that other factors require a 
limit above the lowest achievable NOX rate.\14\ Without 
identifying the lowest achievable NOX emission rate or 
explaining how the other factors affect that rate, PADEP proposed (and 
finalized) a rate of 0.07 lb/MMBtu when the SCR is operating. HC TEM, 
pp. 12-13.
---------------------------------------------------------------------------

    \11\ Note that Key TEM p. 3, Con TEM p. 3 and HC TEM p. 4, state 
that the years reviewed are 2017-2020, but Key RTC p. 8, Con RTC p. 
9 and HC RTC p. 7 state years reviewed were 2016-2020.
    \12\ For Keystone, the months examined were May 2017 and April 
2018. Key TEM pp. 3-6. For Homer City, the months were June 2019 
(unit 1), July of 2019 and 2020 (unit 2), and December 2017, July 
2019, September 16, 2019 and Dec. 4, 2019 (unit 3). HC TEM, pp. 4-
14. For Conemaugh, dates examined included May 2017, September 5, 
2019, and April 4, 2020. Con TEM pp. 3-8.
    \13\ Con TEM, pp. 3-8; Key TEM, pp. 3-7; HC TEM, pp. 4-13.
    \14\ The other factors PADEP cites are varying loads, operating 
load, catalyst condition, exhaust temperature and velocity, moisture 
level, initial NOX levels in the exhaust, and other 
unnamed factors. HC TEM, p. 13.
---------------------------------------------------------------------------

    Certain changes made to PADEP's proposed rates for each source in 
response to comments received are discussed in EPA's analysis of the 
final rates.

[[Page 13025]]

Selection of All Conditions lb/MMBtu Daily Average Rate

    The lb/MMBtu limits in the ``All Conditions lb/MMBtu Daily 
Average'' columns of Table 1 in this document, represent the daily 
average NOX limits that PADEP determined each unit at each 
facility could achieve solely through the operation of its existing 
low-NOX burners with overfire air, so long as the sources 
``tuned'' their boilers to optimize the reduction of NOX 
rather than to obtain the highest heat output. PADEP describes boiler 
tuning as making a number of adjustments to the boiler operating 
parameters that affect the generation of NOX in the boiler 
fire box, including excess air levels, secondary air biasing, fuel/
auxiliary air damper adjustments, burner tilt, fuel flow biasing, and 
changes to primary air flows. See, e.g., Con. TEM, pp. 14-15. As stated 
in the technical evaluation memo for Conemaugh, ``[g]enerally boiler's 
regular inspection, preventive maintenance, tuning, practicing during 
shutdown and upset conditions to prevent excess emissions, inspections 
and testing of Over Fire Air (OFA) components, and adjusted of burner 
angle to minimize NOX emissions results in lowering 
NOX emissions by 5-15% or at an average of 10. %. [sic]'' 
Con. TEM, p. 15. For each of the EGU boilers (units) at each of the 
facilities, PADEP determined that the boiler burners had not been tuned 
to minimize NOX emissions, but rather had been tuned to 
maximize output. Key TEM, p. 13; HC TEM, p. 15; Con TEM, pp. 14-15. For 
each facility, PADEP concluded that tuning the boilers to minimize 
NOX emissions could result in lowering NOX 
emissions by 5% to 15%, so PADEP elected to apply an average 
NOX reduction of 10% when setting the ``All Conditions lb/
MMBtu Daily Average'' rate. Id.

Selection of All Conditions 30-Day Rolling Average lb/hr Rate

    Regarding the 30-day rolling average pounds of NOX/hour 
limits in the column in Table 1 labeled ``All Conditions lb/hr 30-day 
Average,'' there is some ambiguity in how PADEP arrived at the final 
rates for Keystone and Conemaugh. In the Keystone RTC, PADEP states the 
30-day lb/hr limit was ``derived from the emission level at 0.08 lb/
MMBTU at full load . . . with an additional small margin to account for 
the fact that it is impossible to completely avoid all periods of 
operation when complying with the 0.080 lb/MMBtu is technically 
infeasible.'' Key RTC, p. 10. Similar language stating that the 30-day 
lb/hr rate was derived from the daily SCR-on rates is also in Con RTC 
p. 11 and HC RTC p. 9. PADEP's explanation for how the 30-day lb/hr 
limits were derived in the Technical Evaluation Memos is more ambiguous 
and does not explicitly state the 30-day lb/hr rate is derived from the 
daily SCR-on lb/MMBtu rate. A description of what PADEP did in the 
Technical Evaluation Memos is outlined below.
    PADEP seems to have generally performed a similar analysis of 
similar years of data for all three facilities, but used a different 
method to set the 30-day lb/hr rates for Conemaugh and Homer City than 
for Keystone. For each source at each facility, PADEP says it analyzed 
``mass-based NOX emission rate in pounds per hour on a 30 
operational day rolling average basis using EPA's CAMD database at all 
operating conditions for [the units] from 2017-2020. Mass based 
emission rate on a 30 operational day rolling average basis is 
dependent on number of hours a unit is operated, on average, at high 
load vs low load for the past 30-days [sic].'' \15\ Con TEM p. 15; see 
also Key TEM pp. 13-14, HC TEM pp. 17-19.
---------------------------------------------------------------------------

    \15\ The Conemaugh TEM does not show the results of the full 
analysis of the 2017-2020 data. For example, Figure 6 in the Con TEM 
shows only ozone season operating load versus 30-day rolling average 
NOX emissions on a lb/hr basis for the 2017 ozone season. 
Con TEM, p. 6.
---------------------------------------------------------------------------

    Following this analysis for each facility, for Conemaugh and Homer 
City PADEP used the SCR-on lb/MMBtu rate for each unit at each 
facility, then multiplied that SCR-on rate by each unit's maximum MMBtu 
per hour rating to arrive at the number of pounds per hour that each 
unit would emit if they ran at their full heat input rating while 
complying with that unit's SCR-on lb/MMBtu rate. For example, the 
technical review memo for Conemaugh explains that:

    ``Each of Conemaugh's units emits about 580 lb NOX 
per hour assuming an emission level of .070 lb/MMBtu and 100% load. 
The impact to the environment should never exceed this level on a 
long-term basis. The Department is proposing a limit of 700 lb/hr 
limit on a 30 operational day rolling basis which accounts for all 
operating scenarios including situations during which the SCR is not 
able to operate. The compliance buffer also accounts for the fact 
that both units at Conemaugh operate as much as 10% over their rated 
capacity.'' (Con TEM, p.15).

    For Conemaugh, PADEP concluded that Units 1 and 2 were operating 
between 55% and 100% load during this time and both were able to 
achieve at or below 625 lb/hr on a 30-operating day basis. PADEP found 
that during this time period both units operated at around a 0.075 lb/
MMBtu NOX emissions rate, with occasional higher spikes in 
rate. Based on this data, PADEP concluded:

    ``Given that the Department believes that NOX rates 
below .07 are readily achievable with the SCR in operation, and the 
fact that both units were able to achieve a 30-day rolling 
NOX rate of under 625 lb/hr despite operating at a rate 
between .075 and .1, DEP believes that Conemaugh Generating Station 
can achieve a NOX rate of 700 lb/hr on a 30-day rolling 
basis. Even if the facility were to operate at low load for a 
significant time during a 30-day averaging period--generating 
significantly more mass emissions than operation at higher loads 
with SCR, emission rates at high load should be significantly below 
700 lb/hr allowing the facility to ``make up'' for higher emissions 
during times of low load, assuming the facility operates to the 
NOX rate of .045-.05 lb/MMBtu it is usually capable of 
meeting when the SCR is operating.'' Con TEM, pp. 16-17.

Thus, for Conemaugh, PADEP proposed and finalized an all conditions 30-
day rolling average lb/hr limit of 700 lb/hr.
    For Homer City, PADEP used the proposed SCR-on daily average 
NOX limit of 0.08 lb/MMBtu for Units 1 and 2, multiplied by 
the maximum MMBtu per hour for each of these units, to arrive at a 30-
day rolling average limit of 550 lb of NOX per hour for each 
unit. For Unit 3, PADEP used the proposed SCR-on lb/hr daily average 
limit of 0.07 lb/MMBtu multiplied by the maximum heat input for Unit 3 
to arrive at a rolling 30-day average limit of 510 lbs of 
NOX per hour. HC TEM, p. 17. In response to a comment from 
Homer City, PADEP raised the 30-day rolling average lb/hr limits to 600 
lb/hr for Units 1 and 2 and 560 lb/hr for unit 3.
    For Keystone, PADEP appears to have arrived at its proposed and 
final 30-day rolling average lb/hr limit through a different method. 
PADEP's TEM states that PADEP analyzed the mass-based NOX 
emission rate pounds per hour on a 30-day rolling average at all 
operating conditions for Units 1 and 2 from 2017-2020. Key TEM, p. 13. 
The TEM then provides Figure 5, which graphs the 30-day rolling 
NOX rates for Units 1 and 2, but only for the 2017 ozone 
season. Key TEM. P. 14. From Figure 5, the TEM concludes that both 
units were able to achieve at or below 800 lbs/hr on a 30-day rolling 
average basis, continuously. Key TEM, p.14. The TEM then asserts that 
based on the CAMD data, ``DEP believes that by managing combination of 
hours of operations when a unit is operating at loads supporting SCR 
and at lower loads with [low NOX burners], Unit 1 and Unit 2 
can achieve 800 lbs/hr on a 30-day operating day rolling average basis 
despite the changes in utilization of the boiler.'' Id. From this,

[[Page 13026]]

PADEP concluded that the 800 lb/hr 30-day rolling average limit under 
all operating conditions is RACT. Id. However, in response to comments, 
this limit was changed to 770 lbs/hr for both units. Key RTC pp. 6.
    In the response to comments document for Conemaugh, PADEP explained 
that the 30-day rolling average lbs/hr all conditions rate ``. . . is 
the glue that holds the three emission limits together and ensures that 
the emission reductions from the two Conemaugh Generating Station units 
are maximized . . . This emission limit applies at all times and in all 
circumstances, without exception.'' Con RTC, p. 6.\16\ PADEP further 
asserts that the SCR-on lb/MMBtu daily average rate minimizes the 
emissions that occur when operating with the SCR, while also claiming 
that the 30-day rolling average lb/hr all conditions rate minimizes 
``both the amount of time that the units can be operated when the SCR 
is technically unavailable, as well as forces the load (and therefore 
mass emission rate) to the lowest rate possible when it is not being 
operated due to technical unavailability.'' See, e.g. Con RTC, p. 6. 
The RTC further explains that ``[a]t any load above approximately 30%-
40%, operation without control by the SCR results in emissions greater 
than 700 lbs/hr. As the load climbs, the emissions per hour climb 
proportionately.'' Id. PADEP asserts that the 700 lb/hr rolling 30-day 
average limit ``ensures that the operator will maximize operating hours 
with the SCR and minimize heat input (and total mass emissions) when 
operation of the SCR is technically infeasible.'' Id.
---------------------------------------------------------------------------

    \16\ Nearly identical statements are in Key RTC p. 6 and HC RTC 
p. 6.
---------------------------------------------------------------------------

III. EPA's Evaluation of the RACT Permit Limits in the SIP Submittals

    EPA's review of the RACT permit limits in each of the three case-
by-case RACT permits submitted as SIP revisions by PADEP has identified 
several issues appearing in each permit which preclude approval of the 
SIP submissions as satisfying RACT requirements. In summary, EPA has 
determined that there are issues regarding the enforceability of the 
SCR-on permit limits, Director's discretion issues related to the SCR-
on limits, and an inadequate justification for why the SCR-on limits 
meet the definition of RACT for each source. Moreover, because some of 
the 30-day rolling hourly average pound per hour mass limits appear to 
be derived from the daily lb/MMBtu SCR-on limits, the failure of the 
SCR-on limit to meet the criteria for RACT calls into question whether 
the 30-day limits are RACT.\17\ Also, EPA cannot verify from PADEP's 
submitted SIPs whether these 30-day rolling average pound per hour mass 
limits actually act as a constraint on operation of the EGUs without 
operation of the SCRs in a way that represents RACT. In addition, PADEP 
has added a ``compliance margin'' buffer to the 30-day rolling average 
pound per hour limits without an adequate explanation of why that 
buffer is necessary to make the limits technologically or economically 
feasible. Each of these issues is discussed below. As a result, EPA is 
proposing to disapprove this SIP revision.
---------------------------------------------------------------------------

    \17\ In Key RTC p. 10 PADEP states the 30-day lb/hr limit was 
``derived from the emission level at 0.08 lb/MMBTU at full load . . 
. with an additional small margin . . .'' Similar language stating 
that the 30-day lb/hr rate was derived from the daily SCR-on rates 
is also in Con RTC p. 11 and HC RTC p. 9. PADEP's explanation for 
how the 30-day lb/hr limits were derived in Key TEM pp. 13-14, Con 
TEMP pp. 15-17 and HC TEM pp. 17-19 is more ambiguous and doesn't 
explicitly state the 30-day lb/hr rate is derived from the daily 
SCR-on lb/MMBtu rate as noted under the ``Selection of All 
Conditions 30-day Rolling Average lbs/hr Rate'' heading in section 
II.
---------------------------------------------------------------------------

Lack of Enforceability of the ``SCR-On'' Limits for Each EGU at Each 
Facility

    Neither the permits nor the background information submitted with 
the SIP set forth clear, objective criteria for determining when 
emissions from each EGU are subject to the SCR-on lb/MMBtu daily 
average limit(s). As such, it is not possible in all circumstances for 
EPA or the public to determine whether this limit applies, and 
therefore whether the sources are in noncompliance with that limit. As 
a result, EPA is proposing to disapprove the PADEP SIP revision on this 
basis.
    Each permit includes language stating the NOX emissions 
are limited at a certain level, but that certain emissions are excluded 
when evaluating whether the limitations are met. Specifically, the 
permits contain exclusions for:

``. . . emissions during start-up, and shut-down; operation pursuant 
to emergency generation required by PJM, including any necessary 
testing for such emergency operations; and during periods in which 
compliance with this emission limit would require operation of any 
equipment in a manner inconsistent with technological limitations, 
good engineering and maintenance practices, and/or good air 
pollution control practices for minimizing emissions.'' See, e.g., 
Conemaugh final permit, Section E, Restrictions, #001, p. 176. 
Keystone Final Permit, p. 169, and Homer City Final Permit p. 
134.\18\
---------------------------------------------------------------------------

    \18\ PJM is the Pennsylvania-New Jersey-Maryland 
Interconnection, a regional transmission organization operating in 
the midatlantic states.

    EPA has determined that the exclusion during ``Operation pursuant 
to emergency generation required by PJM'' is problematic. This 
condition is not defined in the permit for Homer City but is defined in 
the final permits for Conemaugh (p. 176) and Keystone (p.169), stating 
that ``the emissions limit remains in effect unless the permittee 
demonstrates that compliance with the [applicable emission limitation] 
is technically infeasible.'' There are no bounds or explanation in the 
permit regarding what would equate to technical infeasibility, nor is 
there information on whom the permittee would demonstrate this 
infeasibility to or how EPA or the public could determine whether such 
an adequate demonstration was made. In response to comments, PADEP 
---------------------------------------------------------------------------
stated:

``the Conemaugh Station permit includes a process where emissions 
can be requested for exclusion from calculation of the 0.070 lb/
MMBtu emission limit if the owner/operator makes a demonstration of 
technical infeasibility to the Department's satisfaction. The 
general factors that may lead to technical infeasibility are 
included in the Conemaugh Station permit, and mirror SIP-approved 
RACT regulations in neighboring states. In fact, the list of general 
factors in Conemaugh's permit is more limited than the factors 
listed in regulations promulgated by one commentator. See COMAR 
26.11.38.04 section 4.'' \19\ Con RTC, p. 3.
---------------------------------------------------------------------------

    \19\ COMAR is the Code of Maryland Regulations.

EPA did not find the suggested list of ``general factors'' which may 
lead to a determination of technical infeasibility.
    EPA also notes that this type of post-hoc determination allowing 
the director to grant exemptions from a SIP-approved emission limit 
during periods of startup, shutdown or other periods is the type of 
director's discretion prohibited by the CAA, for the reasons set forth 
in EPA's 2015 startup, shutdown and malfunction (SSM) SIP Action. 80 FR 
at 33840, 33917 (June 12, 2015). As stated in the 2015 SSM SIP Action, 
``SIP provisions cannot contain director's discretion to alter SIP 
requirements, including those that allow for variances or outright 
exemptions for emissions during SSM events.'' 80 FR at 33917. In the 
case of the permits submitted as part of Pennsylvania's 2022 SIP 
revision, each contains language that allows the director to decide 
whether or not emissions from a source during any hour should be 
counted towards the more stringent SCR-on emission limits of 0.07-0.08 
lb/MMBTU or to the less stringent emission limits of 0.27-0.45 lb/
MMBTU. Although the rates would not change, the director would be 
making a

[[Page 13027]]

decision as to whether certain emissions should be exempted from the 
more stringent SCR-on lb/MMBtu 24-hour average rate. This is the type 
of unilateral, ad hoc (or post hoc) decision by the director which 
could negate the possibility of enforcement of an otherwise enforceable 
SIP emission limit by EPA or the public and which is barred by EPA as 
first established in the 1999 SSM Guidance. 1999 SSM SIP guidance at 3, 
80 FR 33840 at 33917.
    In addition, pursuant to EPA's responsibilities under sections 
110(k)(3), 110(l) and 193 of the CAA, the Agency cannot approve a SIP 
provision that automatically preauthorizes the state to unilaterally 
revise the SIP emission limit (in this case by making determinations 
that it did not apply at certain times) without meeting the applicable 
procedural and substantive statutory requirements for SIP revisions. 80 
FR at 33918. As stated in EPA's 2015 SSM SIP Action, ``[i]t is a 
fundamental tenet of the CAA that states cannot unilaterally change SIP 
provisions, including the emission limitations within SIP provisions, 
without the EPA's approval of the change through the appropriate 
process.'' Id.
    In the quoted response to comments on this issue, PADEP claims that 
the list of general factors in the permits (which EPA could not locate) 
are more limited than factors listed in Maryland's regulations. EPA 
notes that it has not approved the cited Maryland regulation, COMAR 
26.11.38.04, as RACT for EGUs, so the cited example does not carry any 
weight in EPA's analysis of this SIP revision.\20\ PADEP claims that 
the list of general factors (which again, EPA could not locate) 
``mirror SIP-approved RACT regulations in neighboring states,'' but 
PADEP does not identify these other SIP-approved RACT regulations and 
EPA is not aware of what PADEP may be referencing. Without knowing 
which SIP-approved RACT regulations PADEP is referring to, EPA cannot 
judge the relevance of this argument.
---------------------------------------------------------------------------

    \20\ See the final document at 82 FR 24546 (May 30, 2017) 
approving the NOX limits for Maryland's EGUs as SIP 
strengthening measures, and the final document at 84 FR 5004 
(February 20, 2019) approving Maryland's RACT regulations for 
controlling VOC major sources for the 2008 ozone NAAQS, which notes 
that Maryland will address major sources of NOX in 
another SIP. None of the VOC regulations approved included the 
language in COMAR 26.11.38.04.
---------------------------------------------------------------------------

    The exclusion for ``periods in which compliance with this emission 
limit would require operation of any equipment in a manner inconsistent 
with technological limitations, good engineering practices, and/or good 
air pollution control practices . . .'' is also problematic. No permit 
provides additional definitions or instruction on how this provision 
should be interpreted or applied. Similar to other provisions at issue 
here, this lack of definition makes this exemption provision difficult 
or impossible to enforce.
    Although the permits require that the sources keep certain data and 
submit a monthly report to PADEP, it is in the sources' discretion to 
identify in these monthly reports ``whether or not they believe they 
are subject to the [SCR-on] lb NOX/MMBtu limit'' and 
``clearly document how [they] determined whether or not they believe 
they are subject to the [SCR-on] lb NOX/MMBtu hourly 
limit.'' \21\ But this does not explain how PADEP will determine 
whether certain hours of NOX emissions from the sources 
should be counted towards the SCR-on daily average lb/MMBtu limits for 
each source, or the circumstances under which these emissions would be 
excluded from the limit. It is even more difficult to understand how 
EPA or the public would discern which hours of emissions should be 
counted towards the SCR-on limit. If it is unknown which hours of 
emissions count, it is impossible to determine whether a source 
complied with the SCR-on limit. In other words, without clear and 
objective criteria for excluding these emissions, neither EPA nor the 
public could determine whether the sources were complying with the SCR-
on limit at each source. Although this situation is somewhat different 
than the situation faced by the Third Circuit in the Sierra Club appeal 
(lack of adequate recordkeeping), the lack of objective criteria for 
determining compliance in this situation leads to the same problem 
identified by that court, which is that there is no way for interested 
members of the public or EPA to conduct oversight. Sierra Club at 307.
---------------------------------------------------------------------------

    \21\ Conemaugh final permit, p. 177. The final unredacted 
permits for all three facilities also state that the monthly reports 
should include the hourly load levels, heat input, ammonia injection 
rates, NOX rates, total NOX emissions, the SCR 
emission set point, SCR inlet and outlet temperature, and clearly 
indicate any days which the SCR-on lb/MMBtu emission limit is 
exceeded. For days exceeding the SCR-on lb/MMBtu limit, the above 
information must be provided on an hourly basis and the permittee 
must give a detailed explanation for why they exceeded their 
emission limit. Conemaugh permit, p. 176, Keystone permit pp. 170-
171, Homer City permit p. 137.
---------------------------------------------------------------------------

Pennsylvania's Inadequate Justification of Certain Limits as RACT

    EPA understands the PADEP's submission to argue that RACT for these 
facilities is comprised of: (1) a low daily SCR-on lb/MMBtu limit with 
exclusions as outlined in the prior section; (2) a much higher all 
conditions daily lb/MMBtu limit that provides a permissible emissions 
level under all operating conditions including when the SCR is not 
operating; and (3) the 30-day rolling average all conditions lb/hr 
limit, which is intended to provide some restriction on the extent to 
which the source could claim exclusions from the SCR-on rate. EPA has 
identified issues with each of these limits as discussed in the 
subsections below. EPA does allow for the possibility that different or 
alternative emissions limits (AELs) can apply during different modes of 
operation in the manner that PADEP has done here for the three 
different limits described.\22\ However, EPA has stated that those AELs 
``must be clearly stated components of the emission limitation, must 
meet the applicable level of control required for the type of SIP 
provision (e.g., be RACT for sources located in nonattainment areas) 
and must be legally and practicably enforceable.'' \23\ Accordingly, 
here EPA must evaluate whether this combination of limits satisfies the 
OTR RACT requirement. PADEP did not provide any justification for why 
these limits appropriately function as alternative emission limits. In 
the 2015 SSM SIP Action, EPA recommended states consider seven criteria 
when developing alternative emission limits.\24\ These recommended 
criteria assure the alternative emission limitations meet basic CAA 
requirements. PADEP did not explain why the alternative emission 
limitations included in this SIP revision meet CAA requirements, 
including RACT, and EPA cannot approve alternative emission limitations 
without such a showing.
---------------------------------------------------------------------------

    \22\ ``State Implementation Plans: Response to Petition for 
Rulemaking; Restatement and Update of EPA's SSM Policy Applicable to 
SIPs; Findings of Substantial Inadequacy; and SIP Calls To Amend 
Provisions Applying to Excess Emissions During Periods of Startup, 
Shutdown and Malfunction'' 80 FR 33840, section XI.D.
    \23\ Ibid. P 33913.
    \24\ Ibid. p. 33914.
---------------------------------------------------------------------------

    Further, PADEP developed the emissions limits for the Keystone, 
Conemaugh and Homer City Facilities by reviewing only operating data 
and emissions rates from a limited number of years.\25\ PADEP claims 
that using emissions and operating data from a limited set of 
relatively recent years is justified because these years reflect what 
is currently possible due to aging

[[Page 13028]]

equipment and changes in operating patterns, including the impact of 
changes made to the catalyst in the SCR system in order to meet the 
requirements of the 2011 Mercury Air Toxics Standard (MATS) (Key RTC p. 
8, Con RTC p. 9, and HC RTC p. 7).
---------------------------------------------------------------------------

    \25\ EPA also notes an inconsistency in how PADEP discusses the 
data that was considered in developing the limits at issue in this 
SIP revision. In the RTCs, PADEP references data from 2016-2020. 
(Key RTC p. 8, Con RTC p. 9, and HC RTC p. 7). However, in the TEMs, 
PADEP references data from 2017-2020.
---------------------------------------------------------------------------

    However, PADEP presented no data or analysis showing that aging 
equipment, particularly the SCR control systems, have deteriorated such 
that data from earlier years are unreliable. PADEP's submittals have 
also not justified a rate selection methodology that relies on a 
limited set of years, nor have they explained why the selected years 
represent the lowest rate that can now be achieved when accounting for 
such changes. Stated differently, the RACT limits (regardless of 
averaging time) must reflect levels that represent periods of good 
emissions control, not business as usual (e.g., a 5-year average of 
past results) or higher-emitting periods.

Selection of the SCR-On lb/MMBtu Daily Average Emission Rates

    PADEP's own data and analysis calls into question whether the final 
SCR-on daily average lb/MMBtu rates for Conemaugh (0.070 lb/MMBtu), 
Keystone (0.080 lb/MMBtu) and Homer City (0.080 lb/MMBtu for Units 1 
and 2, and 0.070 lb/MMBtu for Unit 3) are RACT. Based on PADEP's SIP 
submission, EPA cannot determine whether the SCR-on rates for any of 
the three facilities are the lowest rates that can be achieved 
considering technological and economic feasibility. Although PADEP 
makes a general determination that optimization of the existing SCRs at 
each facility is RACT, the data PADEP provided in its SIP submission do 
not support a claim that these rates are the lowest achievable rates 
that can reasonably be obtained at each unit when the SCRs are 
operating, considering technological and economic feasibility. In 
addition, PADEP then applies an upward adjustment to these rates to 
account for factors, such as lag time, changes in boiler operating 
patterns, and aging of equipment, that PADEP states it has already 
accounted for by using data from 2017 to 2020 in their analyses for 
setting the RACT limits. As such, there should be no needed upward 
adjustment to account for these factors. Also, PADEP consistently 
applies a compliance margin to its rates without explaining what the 
margin is, in many cases, or why such a margin is needed to make the 
selected limit technologically or economically feasible.

Conemaugh

    For Conemaugh, PADEP asserts that it examined CAMD emissions and 
other data for Units 1 and 2 for the years 2017-2020, but because both 
units are similar, assumed that data from unit 1 applied to unit 2 and 
therefore only discussed unit 1 data.\26\ Con TEM, pp. 3-4. Figure 1 in 
the TEM is a graph showing percentage of heat input, NOX 
emission rates and percentage of ammonia injection rates during May 
2017. From this graph, PADEP determined that Conemaugh Unit 1 
maintained a NOX emission rate of 0.045 lb/MMBtu from May 
5th through May 18th, which PADEP attributed to an ammonia injection 
control system operating at a set point of 0.045 lb/MMBtu. Con TEM, p. 
4. From May 19th through the end of May 2017, PADEP observed that the 
``relative difference between the ammonia injection rates and heat 
input rates have increased,'' leading to a steady NOX 
emission rate around 0.08 lb/MMBtu. Id. PADEP then notes that following 
May 2017, unit 1 only operated with varying set points between 0.065 
and 0.08 lb/MMBtu throughout the 2017 and 2018 ozone seasons. Con TEM, 
p. 5. PADEP further observed that NOX rates increased 
significantly in 2019 and provided a graph (Figure 2) which PADEP 
asserts shows that during this month, Conemaugh ceased injecting 
ammonia for NOX control at around 50% heat input, and even 
when operating at 100% of heat input, the NOX emission rates 
stayed around 0.1 lb/MMBtu. PADEP concluded that ``this strongly 
suggests that additional emission reductions would be achieved if the 
operator operated the SCR with a lower emission set point while the SCR 
is running.'' Con TEM, p. 6. In addition, PADEP identified an April 
2020 example when the SCR was not operating despite the boiler 
operating at loads ``clearly supporting'' SCR operation, with 
NOX emissions close to 0.3 lb/MMBtu during this time. Con 
TEM, p. 7. From this PADEP concluded that ``[s]imply choosing not to 
operate the SCR is not indicative of the control level achievable by 
the system.'' Id. Based on this data, PADEP then selected an SCR-on 
rate of 0.07 lb/MMBtu for Conemaugh. Id. The only explanation given for 
this specific rate is that it ``includes a factor to provide an 
appropriate compliance margin, fluctuations in load, any lag in the 
control system as well as to account for other factors in the 
facility's future operations.'' Con TEM, p. 8.
---------------------------------------------------------------------------

    \26\ In response to a comment submitted on Conemaugh, PADEP 
replied that during the 2018 ozone season, with a few exceptions, 
Conemaugh's unit 2 was consistently able to achieve daily emission 
levels in the .055-.07 lb NOX/MMBtu range. Con. RTC, p. 
7.
---------------------------------------------------------------------------

    The response to comments (RTC) document for Conemaugh adds 
discussion of a 2017 study performed on unit 1 in May 2017 that 
suggested that running the SCR with a set point of 0.04 lb/MMBtu caused 
a spike in mercury emissions, and also discusses a 2016 study at the 
end of ozone season on unit 2 that suggested running the SCR at a 0.050 
lb/MMBtu set point also caused an increase in mercury emissions. RTC, 
pp. 6-8. Based on further analysis, PADEP concluded that ``a setpoint 
of 0.06 lb NOX/MMBtu . . . is achievable by [Conemaugh].'' 
RTC p. 8. However, the RTC states, without explanation, that PADEP is 
choosing to keep the 0.07 lb NOX/MMBtu daily average 
emission rate. RTC p. 8.
    EPA finds that PADEP's explanation of why this limit meets the 
definition of RACT is inadequate. Having concluded in the RTC that a 
0.06 setpoint is achievable at Conemaugh, PADEP provides no explanation 
as to why it selected 0.07 lb/MMBtu as the daily average SCR-on rate. 
Nor is there any explanation of why a compliance margin is necessary, 
what compliance margin was applied in this instance, how fluctuations 
in load or lag in the control system affect the lowest achievable 
emissions rate, and how or why the rate must be adjusted to account for 
future operations. In the absence of an explanation of how PADEP 
selected the specific 0.07 lb/MMBtu rate and how any of these other 
factors affect the technical and economic feasibility of the lowest 
rate identified, EPA cannot support PADEP's conclusion that the 0.07 
lb/MMBtu daily average rate is RACT for when Conemaugh's SCRs are 
operating.

Keystone

    Like Conemaugh, PADEP's analysis for Keystone's SCR-on daily 
average rate of 0.08 lb/MMBtu does not adequately explain why this rate 
represents the lowest emission limit that Keystone's two units are 
capable of meeting based on technological and economic feasibility. In 
the TEM for Keystone, PADEP explains that it analyzed EPA's CAMD data 
for Keystone Units 1 and 2 from 2017-2020. Key TEM, p. 3. The TEM then 
includes a graph (Figure 1) showing certain daily operating statistics 
for unit 2 for the month of May 2017, from which PADEP concludes that 
unit 2 was able to maintain a NOX emission rate below 0.06 
while the SCR was operating. TEM, p. 4. The TEM then shows a graph 
(Figure 2) plotting certain

[[Page 13029]]

daily operating parameters for unit 1 during May 2017. TEM, p. 5. From 
Figure 2, PADEP concludes that unit 1 was able to achieve a 0.05 lb/
MMBtu rate for ten days, but this rate increased to 0.09 lb/MMBtu for 
the rest of the month because the operator elected to inject less 
ammonia into the SCR system even though the heat input remained almost 
constant at levels supporting SCR operation. Key TEM, p. 5. From this 
and other data, PADEP concludes that both Keystone units can achieve an 
SCR-on rate of 0.06 lb/MMBtu on a daily average basis.
    The Keystone response to comments contains a long discussion of a 
study Keystone submitted at some point in time purporting to show the 
effects of trying to operate the SCRs at a NOx emission rate setpoint 
of 0.05 to 0.06 lb/MMBtu during May 2017. Keystone RTC, p. 26. The RTC 
notes that when Unit 2 attempted to operate at 0.055 lb/MMBtu for two 
months in 2017, pressure drop across the air preheater increased to a 
level requiring measures--in this case raising the SCR setpoint to 0.08 
lb/MMBtu--to reduce the pressure drop. Id. at 27. The same study found 
that operating unit 1's SCR at a 0.05 lb/MMBtu setpoint for only 15 
days resulted in SCR catalyst fouling which prevented the SCR from 
operating under 0.08 to 0.09 lb/MMBtu rates for the rest of the test 
period. Id. The Keystone RTC then discusses at length the meaning of 
the study and information submitted by another source and the effect of 
different SCR set points on pressure drop, catalyst fouling, and the 
ability to meet certain NOX emission rates. Key RTC, pp. 27-
30. PADEP concluded from these studies that Keystone should conduct a 
future setpoint study to determine that optimal emission levels from 
the SCR are achieved, but that based on the current evidence, the SCR 
controls setpoint should be changed from 0.06 to 0.07 lb 
NOX/MMBtu.\27\ Key RTC, p. 32. However, PADEP set a 
NOX emission rate of 0.08 lb/MMBtu because ``varying load 
conditions and other factors can and do affect SCR performance and 
resulting NOX emission rates.'' TEM, p. 6.
---------------------------------------------------------------------------

    \27\ The final unredacted permit does not mention this setpoint 
study. Instead, Section E, Source Group Restrictions, subsection VI, 
Work Practice Requirements, condition #012 requires that Keystone 
submit a technical evaluation to PADEP on the possibility of heating 
the flue gas prior to the SCR inlet to allow SCR operation at low 
load levels. Keystone final permit, p. 172. This condition does not 
appear in the redacted final permit submitted for inclusion into the 
SIP.
---------------------------------------------------------------------------

    EPA acknowledges that catalyst fouling and other similar factors 
may affect the feasibility of SCR to achieve low rates. However, 
similar to EPA's review of the Conemaugh limit, in the absence of an 
explanation of how any of these other factors affect the technical and 
economic feasibility of the lowest rate identified, EPA cannot support 
PADEP's conclusion that the 0.08 lb/MMBtu daily average rate represents 
RACT.

Homer City

    Similar to EPA's assessment of the rates for Conemaugh and 
Keystone, PADEP does not provide adequate justification for Homer 
City's final SCR-on daily average rates of 0.08 lb/MMBtu for units 1 
and 2 and 0.07 lb/MMBtu for unit 3 are the lowest emission limit that 
these sources can meet based on technological and economic feasibility. 
The TEM for Homer City explains that PADEP evaluated data from 2017-
2020 for all three units. PADEP notes that the unit 1 and 2 SCRs were 
upgraded in 2018, and ``NOX emission rates significantly 
improved,'' TEM, p. 6, but fails to explain why, in light of this, 
PADEP thought consideration of 2017 data was appropriate. For units 1 
and 2, PADEP notes that during 2019 and 2020 the SCRs were operated to 
generally keep NOX emission rates at 0.10 lb/MMBtu, but also 
identified periods of time when the NOX emission rate for 
unit 2 went as low as 0.05 lb/MMBtu because more ammonia was being 
injected. TEM, p. 5. The TEM states that other instances of between 
0.05 and 0.10 lb/MMBtu were identified. TEM, p. 5. Looking at 
additional data following the upgrade, PADEP suggested that ``had July 
of 2020's ammonia injection rates matched that of July 2019, 
significantly [sic] emissions reductions could have been achieved 
during that timeframe.'' TEM p. 6. The TEM then states that ``[d]espite 
the evidence presented,'' other factors such as load, exhaust 
temperature, etc., and other unspecified factors ``can and do affect 
SCR performance'' and require an operating limit above the never 
specified achievable minimum. TEM, p. 8. PADEP then selected an SCR-on 
rate of 0.08 lb/MMBtu as a daily average for units 1 and 2 but provided 
no analysis or explanation why 0.08 lb/MMBtu is the lowest rate that 
these units could meet based on technological and economic feasibility. 
See TEM, pp. 4-9. The TEM also states that the rate includes an 
unspecified factor to include a compliance margin, account for load 
fluctuations, control system lags, and projected future changes in 
operations. TEM, p. 9. In the RTC, PADEP seems to apply the 
``findings'' from Keystone's attempt to operate the SCRs with a low 
0.05 lb/MMBtu setpoint that such a setting leads to fouling of the air 
preheater, high pressure drops, and SCR catalyst fouling before 
determining that an emission rate of 0.08 lb/MMBtu will not cause these 
problems at Homer City. RTC, p. 11.
    For unit 3, the TEM states that there is evidence that unit 3 can 
meet a NOX emission rate between 0.08 and 0.09 lb/MMBtu, but 
limited evidence that it can meet a lower limit under certain 
circumstances. TEM, p.10. Citing the same factors affecting SCR 
performance as it cited for units 1 and 2, PADEP then concludes a value 
above the minimum SCR rate is needed, but without explanation sets the 
SCR-on daily average rate at 0.07 lb/MMBtu. TEM, p. 13. In their 
response to comments document, PADEP seems to rely upon Keystone's 
study of operating the SCRs at a low set point to support their 
selection of the SCR-on limits for all the units. However, there is no 
discussion of why the Keystone study can be applied to Homer City, 
particularly given that Homer City seems to use an economizer bypass to 
keep the SCRs operating at lower temperatures than might be possible at 
Keystone.

Selection of the All Conditions 30-Day Rolling Average lb/hr Rate

    The PADEP permits allow significant emissions to be excluded from 
the daily lb/MMBtu SCR-on rate under a variety of conditions, and it is 
necessary to evaluate whether the alternative emissions limits 
applicable during these excluded conditions constitute RACT. Although 
the PADEP permits contain a daily lb/MMBtu no-SCR rate, PADEP suggests 
that the 30-day rolling average lb/hr rate is ``the glue'' that holds 
the emissions limits together, and EPA acknowledges that it is a 
critical component to the RACT justification because it establishes the 
practical limitation on the extent to which the source can operate 
without SCRs over an extended period of time. Accordingly, EPA must 
evaluate whether PADEP's 30-day rolling average limit satisfies the 
RACT requirement. EPA's assessment is that PADEP fails to clearly 
demonstrate that the All Conditions 30-day rolling average lb/hr rate 
necessitates that these facilities operate their SCRs to achieve the 
lowest emission rate that is technologically and economically feasible, 
which is required to meet the definitions of RACT.
    PADEP asserts that the 30-day rate represents RACT because ``[a]t 
any load above approximately 30%-40%, operation without control by the 
SCR results in emissions greater than 700 lbs/hr.'' See, e.g., Con RTC 
p. 6. This suggests that the SCR would be

[[Page 13030]]

necessary at higher loads, but it does not address the question of 
whether it meets the RACT requirements when the facilities could run at 
30-40% without using SCR. The EPA believes that it is possible that the 
sources could operate at low loads while simultaneously meeting the 
daily All Conditions lb/MMBtu rate and the 30-day lb/hr rate, thereby 
creating a permissible way to avoid operating the SCR for long periods 
of time. This resembles the 600-degree temperature SCR ``loophole,'' 
which the Third Circuit was highly critical of, that allowed facilities 
to operate just below the temperature threshold at night when demand 
was low and to avoid running the SCR. Sierra Club, 306. The 30-day lb/
hr all conditions rate does not appear to resolve this issue.
    Furthermore, PADEP's justification for the All Conditions 30-day 
lb/hr rates leave many specifics about the justification of the 
selected rates unanswered. Each of PADEP's technical evaluation memos 
have similar language stating ``[PADEP] evaluated and analyzed mass-
based NOX emission rate in pounds per hour on a 30-day 
rolling average basis from EPA's Clean Air Markets Division (CAMD) 
database at all operating conditions . . . from 2017-2020.'' PADEP then 
presents a set of graphs for each unit depicting the 30-day average 
rolling NOX emissions (lb/hr) overlaid with percentage of 
maximum heat input (MMBtu). The time frames explored in these graphs is 
summarized in table 2 of this document.

                         Table 2--Time Frames for 30-Day Average Data Provided by PADEP
----------------------------------------------------------------------------------------------------------------
 Location in technical evaluation memo           Facility               Unit                 Time frame
----------------------------------------------------------------------------------------------------------------
pg. 16, figure 6......................  Conemaugh................               1  May 2017-September 2017.
pg. 16, figure 6......................  Conemaugh................               2  May 2017-September 2017.
pg. 14, figure 5......................  Keystone.................               1  May 2017-September 2017.
pg. 14, figure 5......................  Keystone.................               2  May 2017-September 2017.
pg 18, Figure 11......................  Homer City...............               1  January 2019-December 2021.
pg 18, Figure 11......................  Homer City...............               2  January 2019-December 2021.
pg 18, Figure 11......................  Homer City...............               3  January 2019-September 2020
                                                                                    (approximate).
----------------------------------------------------------------------------------------------------------------

    None of these graphs displays data for the full 2017-2020 timeframe 
PADEP evaluated and analyzed. Only the graphs from Homer City units 
display data for more than a single year. PADEP's analysis then 
consists of a qualitative description of the 30-day lbs/hr average the 
units were able to achieve in the time frames in table 2 of this 
document, but lacks data or description of what 30-day lb/hr all 
conditions rates were observed outside of those time frames. Without 
additional information about the 30-day average lb/hr rates achieved 
during the four years PADEP analyzed, EPA cannot determine whether the 
lb/hr limit selected for each unit represent an average of these years 
of data, which the Sierra Club court found problematic, or the lowest 
emissions in lb/hr which these sources achieved in this time frame, 
considering technological and economic feasibility.
    PADEP may or may not have reviewed a complete set of data from 
2017-2020, but the analysis of this was not included in the technical 
evaluation memos or the response to comment documents. As such, EPA 
could not determine whether or this 30-day all conditions lb/hr rate 
``. . . ensures that the operator will maximize operating hours with 
the SCR and minimize heat input (and total mass emissions) when 
operation of the SCR is technically infeasible.''
    Additionally, the compliance buffer added to the 30-day lb/hr all 
conditions rate does not appear to be sufficiently justified. PADEP 
states in its Technical Evaluation Memo for Conemaugh ``[e]ach of 
Conemaugh's units emits about 580 lb NOX per hour assuming 
an emission level of .070 lb/MMBtu and 100% load. The impact to the 
environment should never exceed this level on a long-term basis.'' Con 
TEM p. 15. It would appear PADEP arrived at this number simply by 
multiplying the daily SCR-on (lb/MMBtu) Rate by each boiler's rated 
capacity (MMBtu/hr). Similar statements were made in the memos for 
Keystone and Homer City. See table 3 of this document, for this 
calculation for each boiler at Keystone, Conemaugh and Homer City. The 
table also compares this to the permit limits contained in PADEP's 2022 
SIP Submission, as well as a simple calculation of the percent increase 
in those limits (a compliance buffer added by PADEP). However, no 
explanation is given for why compliance buffers of 10-21% are needed, 
or why certain units should receive more than double the buffer of 
others.

                                        Table 3--Compliance Buffers for PADEP's 30-Day All Conditions lb/hr Rates
--------------------------------------------------------------------------------------------------------------------------------------------------------
                                                                               Rated       Permit  limit   Calculated at   Permit limit
                        Facility                               Unit          capacity      daily  SCR-on  capacity  (lb/    30-day avg.     Compliance
                                                                            (MMBtu/hr)      (lb/MMBtu)          hr)           (lb/hr)       buffer  (%)
--------------------------------------------------------------------------------------------------------------------------------------------------------
Conemaugh...............................................               1           8,280           0.070             580             700              21
                                                                       2           8,280           0.070             580             700              21
Keystone................................................               1           8,717           0.080             697             770              10
                                                                       2           8,717           0.080             697             770              10
Homer City..............................................               1           6,792           0.080             543             600              10
                                                                       2           6,792           0.080             543             600              10
                                                                       3           7,260           0.070             508             560              10
--------------------------------------------------------------------------------------------------------------------------------------------------------


[[Page 13031]]

EPA Approval Would Not Be Consistent With CAA Section 110(l)

    Section 110(l) of the CAA prohibits the Administrator from 
approving any SIP revision ``. . . if the revision would interfere with 
any applicable requirement concerning attainment and reasonable further 
progress (as defined in section 7501 of this title), or any other 
applicable requirement of this chapter.'' For over 15 years, EPA has 
interpreted section 110(l) as permitting approval of a SIP revision as 
long as ``emissions in the air are not increased,'' thereby preserving 
``status quo air quality.'' Ky. Res. Council, Inc. v. EPA, 467 F.3d 
986, 991 (6th Cir. 2006); see also Indiana v. EPA, 796 F.3d 803, 806 
(7th Cir. 2015); Ala. Env't Council v. EPA, 711 F.3d 1277, 1292-93 
(11th Cir. 2013); Galveston-Houston Ass'n for Smog Prevention v. EPA, 
289 F. App'x 745, 754 (5th Cir. 2008). This turns on EPA's interpreting 
``interfere'' as meaning ``to hinder or make worse.'' Ky. Res. Council, 
467 F. 3d at 995. The court in a recent Third Circuit decision 
confirmed that a 110(l) analysis is not a one-size-fits-all provision 
and the variables that must be analyzed depend on the particular 
interference the SIP revision poses. Center for Biological Diversity v. 
EPA, 75 F.4th 174, 181 (3rd Cir. 2023). Here, with the information 
available to EPA, EPA could not determine that approval of the SIP 
revisions at issue would not result in interference. Therefore, EPA 
approval of these SIP revisions would not be consistent with section 
110(l).

IV. Proposed Action

    EPA's review of these materials indicates that Pennsylvania's May 
2016 SIP Submittals for Keystone, Conemaugh and Homer City Generating 
facilities: (1) do not adequately support Pennsylvania's justification 
for the selection of RACT limits for the large EGU boilers; (2) lack 
enforceable objective clear criteria for determining when emissions 
from each EGU are subject to the SCR-on 24-hour average limit; and (3) 
contain unbounded director's discretion provisions. For these, and 
other reasons described above, EPA is proposing to disapprove 
Pennsylvania's May 26, 2022 SIP revisions. EPA is soliciting public 
comments on the issues discussed in this document. These comments will 
be considered before taking final action.

V. Statutory and Executive Order Reviews

    Additional information about these statutes and Executive Orders 
can be found at https://www.epa.gov/laws-regulations/laws-and-executive-orders.

A. Executive Order 12866: Regulatory Planning and Review and Executive 
Order 13563: Improving Regulation and Regulatory Review

    This action is not a ``significant regulatory action'' as defined 
by Executive Order 12866 and was therefore not submitted to the Office 
of Management and Budget for review.

B. Paperwork Reduction Act (PRA)

    This proposed action does not impose an information collection 
burden under the PRA because it does not contain any information 
collection activities.

C. Regulatory Flexibility Act (RFA)

    I certify that this action will not have a significant economic 
impact on a substantial number of small entities under the RFA. This 
action merely proposes to disapprove a SIP submission as not meeting 
the CAA.

D. Unfunded Mandates Reform Act (UMRA)

    This action does not contain any unfunded mandate as described in 
UMRA, 2 U.S.C. 1531-1538, and does not significantly or uniquely affect 
small governments. The action imposes no enforceable duty on any state, 
local or tribal governments or the private sector.

E. Executive Order 13132: Federalism

    This action does not have federalism implications. It will not 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.

F. Executive Order 13175: Consultation and Coordination With Indian 
Tribal Governments

    This action does not have tribal implications as specified in 
Executive Order 13175. This action does not apply on any Indian 
reservation land, any other area where the EPA or an Indian tribe has 
demonstrated that a tribe has jurisdiction, or non-reservation areas of 
Indian country. Thus, Executive Order 13175 does not apply to this 
action.

G. Executive Order 13045: Protection of Children From Environmental 
Health Risks and Safety Risks

    The EPA interprets Executive Order 13045 as applying only to those 
regulatory actions that concern environmental health or safety risks 
that the EPA has reason to believe may disproportionately affect 
children, per the definition of ``covered regulatory action'' in 
section 2-202 of the Executive Order. This action is not subject to 
Executive Order 13045 because it merely proposes to disapprove a SIP 
submission as not meeting the CAA.

H. Executive Order 13211, Actions That Significantly Affect Energy 
Supply, Distribution or Use

    This action is not subject to Executive Order 13211, because it is 
not a significant regulatory action under Executive Order 12866.

I. National Technology Transfer and Advancement Act

    This rulemaking does not involve technical standards.

J. Executive Order 12898: Federal Actions To Address Environmental 
Justice in Minority Populations and Low-Income Populations

    Executive Order 12898 (Federal Actions to Address Environmental 
Justice in Minority Populations and Low-Income Populations, 59 FR 7629, 
February 16, 1994) directs Federal agencies to identify and address 
``disproportionately high and adverse human health or environmental 
effects'' of their actions on minority populations and low-income 
populations to the greatest extent practicable and permitted by law. 
EPA defines environmental justice (EJ) as ``the fair treatment and 
meaningful involvement of all people regardless of race, color, 
national origin, or income with respect to the development, 
implementation, and enforcement of environmental laws, regulations, and 
policies.'' EPA further defines the term fair treatment to mean that 
``no group of people should bear a disproportionate burden of 
environmental harms and risks, including those resulting from the 
negative environmental consequences of industrial, governmental, and 
commercial operations or programs and policies.''
    Under the CAA, the Administrator is required to approve a SIP 
submission that complies with the provisions of the Act and applicable 
Federal regulations. 42 U.S.C. 7410(k); 40 CFR 52.02(a). Thus, in 
reviewing SIP submissions, the EPA's role is to review state choices, 
and approve those choices if they meet the minimum criteria of the Act. 
Accordingly, this proposed action disapproves state law as meeting 
Federal requirements and does not impose additional requirements beyond 
those imposed by state law.
    The air agency did not evaluate environmental justice 
considerations as

[[Page 13032]]

part of its SIP submittal; the CAA and applicable implementing 
regulations neither prohibit nor require such an evaluation. EPA did 
not perform an EJ analysis and did not consider EJ in this action. Due 
to the nature of the action being taken here, this action is expected 
to have a neutral to positive impact on the air quality of the affected 
area. Consideration of EJ is not required as part of this action, and 
there is no information in the record inconsistent with the stated goal 
of E.O. 12898 of achieving environmental justice for people of color, 
low-income populations, and Indigenous peoples. This action merely 
proposes to disapprove a SIP submission as not meeting the CAA.

List of Subjects in 40 CFR Part 52

    Environmental protection, Air pollution control, Incorporation by 
reference, Intergovernmental relations, Nitrogen dioxide, Ozone, 
Reporting and recordkeeping requirements, Volatile organic compounds.

Adam Ortiz,
Regional Administrator, Region III.
[FR Doc. 2024-03528 Filed 2-20-24; 8:45 am]
BILLING CODE 6560-50-P