Sierra Club v. EPA ( 2019 )

Case: 18-60116 Document: 00515144673 Page: 1 Date Filed: 10/03/2019
IN THE UNITED STATES COURT OF APPEALS
FOR THE FIFTH CIRCUIT United States Court of Appeals
Fifth Circuit
FILED
October 3, 2019
No. 18-60116
Lyle W. Cayce
Clerk
SIERRA CLUB; NATIONAL PARKS CONSERVATION ASSOCIATION;
ENTERGY LOUISIANA, L.L.C.,
Petitioners
v.
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY; ANDREW
WHEELER, Administrator, United States Environmental Protection Agency,
Respondents
On Petitions for Review of Final Administrative Action
of the United States Environmental Protection Agency
Before SMITH, WIENER, and ELROD, Circuit Judges.
WIENER, Circuit Judge:
In a December 21, 2017 Final Rule (“the Final Rule”), the United States
Environmental Protection Agency (“the EPA”) approved Louisiana’s state
implementation plan (“SIP”) for controlling regional haze. Louisiana’s regional
haze SIP had two alleged problems. First, the SIP used an outdated air-
pollution model called “CALPUFF” to measure the visibility impacts of
powerplant emissions. Second, the SIP included a sparse explanation for how
Louisiana weighed five mandatory statutory factors in determining the Best
Available Retrofit Technology (“BART”) for controlling emissions at Unit 6 of
the Roy S. Nelson powerplant (“Nelson”). Despite the EPA’s knowledge of these
Case: 18-60116 Document: 00515144673 Page: 2 Date Filed: 10/03/2019
No. 18-60116
problems, it determined that Louisiana had fulfilled its obligations under the
Clean Air Act.
We consider two petitions for review of the Final Rule. One is from
Petitioners-Appellants Sierra Club and National Parks Conservation
Association (collectively, “Environmental Petitioners”). The other is from
Petitioner-Appellant Entergy Louisiana, L.L.C., the owner of the Nelson
powerplant, and Cleco Power, L.L.C., an intervenor in this case (collectively,
“Industry Petitioners”).
Environmental Petitioners maintain that Louisiana’s SIP does too little
to curb regional haze at federally protected areas. They contend that (1)
Louisiana’s determination that “low-sulfur coal” was the BART for the Nelson
powerplant was deficient in several respects and (2) the EPA acted arbitrarily
and capriciously in approving Louisiana’s SIP because it knew about those
deficiencies.
In contrast, Industry Petitioners insist that Louisiana’s SIP
overestimates the amount of pollution that their powerplants produce. In their
challenge to the EPA’s approval of Louisiana’s “subject to BART”
determinations, Industry Petitioners object to Louisiana’s and the EPA’s use
of the “CALPUFF” model, which they maintain relies on several flawed
technical assumptions.
We deny Industry Petitioners’ petition. We afford “significant deference”
to agency decisions involving analysis of scientific data within the agency’s
technical expertise. The EPA’s selection of a model to measure air pollution
levels is precisely that type of decision. The EPA therefore did not act
2
Case: 18-60116 Document: 00515144673 Page: 3 Date Filed: 10/03/2019
No. 18-60116
arbitrarily and capriciously in relying on the CALPUFF model to approve
Louisiana’s “subject to BART” determinations.
Although Environmental Petitioners’ challenge presents a closer
question, we deny that petition as well. Louisiana’s explanation of its BART
determination for Nelson omitted two of the five mandatory factors and failed
to compare—or even set out—the numbers for the costs and benefits of the
control options Louisiana considered. Louisiana also failed to explain how its
decision accounted for the EPA-submitted analyses that pointed out
substantial flaws in other analyses in the administrative record. But applying
the deferential standards of the Administrative Procedures Act to the facts of
this case, we hold that the EPA’s approval of Louisiana’s SIP was not arbitrary
and capricious.
The petitions for review are denied.
I. BACKGROUND
This case addresses the EPA’s approval of Louisiana’s SIP for controlling
regional haze. The Clean Air Act “requires the states and the federal
government to set and seek to achieve targets for visibility in protected
national parks and wildlife areas by modifying regulations that control air
pollutants in ambient air.” 1 Under the Act, the federal government identifies
air pollutants and sets standards, and the states have “the primary
responsibility” for implementing those standards through SIPs. 2 After a state
1 Texas v. EPA, 829 F.3d 405, 411 (5th Cir. 2016) (citing 42 U.S.C. §§ 71410, 7491,
7492(e)(2)).
2 Id. (citation omitted).
3
Case: 18-60116 Document: 00515144673 Page: 4 Date Filed: 10/03/2019
No. 18-60116
submits its SIP, the EPA reviews the SIP for compliance with the Clean Air
Act.
Powerplants that emit sulfur dioxide (“SO2”) and oxides of nitrogen
(“NOx”) contribute to regional haze in protected federal areas. 3 Louisiana has
five powerplants that cause or contribute to visibility impairments in such
areas. 4 Since 2008, Louisiana has revised its SIP several times and established
emission controls at some of those powerplants. In October 2017, Louisiana
submitted its final SIP revisions, which addressed, inter alia, emission controls
at Unit 6 of the Nelson powerplant.
On December 21, 2017, the EPA promulgated a final rule approving
Louisiana’s SIP. 82 Fed. Reg. 60,520 (Dec. 21, 2017) (“the Final Rule”). The two
petitions for review of the Final Rule address the EPA’s approval of (1)
Louisiana’s determination that Nelson and two units at the Cleco-owned
Brame Energy Center (“Brame”) are subject to BART for controlling emissions,
(2) Louisiana’s selection of low-sulfur coal as BART for controlling SO2
emissions at Nelson, and (3) Louisiana’s reasonable progress goals and long-
term strategy.
Environmental Petitioners first address Louisiana’s BART
determination for Nelson. They contend that the Louisiana Department of
3Id. Regional haze is a “visibility impairment that is produced by a multitude of
sources and activities that are located across a broad geographic area and emit fine
particulates . . . and their precursors.” 77 Fed. Reg. 42,834, 42,837 (July 20, 2012) (describing
regional haze and the history of regional haze regulation).
4 See 82 Fed. Reg. 22,936, 22,942 (May 19, 2017).
4
Case: 18-60116 Document: 00515144673 Page: 5 Date Filed: 10/03/2019
No. 18-60116
Environmental Quality (“LDEQ”) erred in three ways: (1) determining that
low-sulfur coal was BART for Nelson, despite the EPA-submitted analyses that
contradicted the analyses in the record, (2) failing to provide a rational basis
for rejecting a more effective pollution control, and (3) not complying with the
BART guidelines or considering all the mandatory BART factors.
Environmental Petitioners next object to the Final Rule’s approval of
Louisiana’s long-term strategy and reasonable progress goals. They maintain
that Louisiana’s 2017 SIP revisions did not fulfill the state’s obligations to
revise and resubmit its long-term strategy and reasonable progress goals after
the EPA disapproved that strategy and those goals in 2012. Louisiana’s 2017
SIP revisions did not impose additional controls at non-BART “reasonable
progress” sources and did not address the state’s reasonable progress goals or
long-term strategy for achieving natural visibility conditions. According to
Environmental Petitioners, the EPA improperly overlooked these omissions
when it approved Louisiana’s SIP.
Environmental Petitioners object to the type of BART control Louisiana
implemented. Industry Petitioners, in contrast, object to Louisiana’s
determination that Nelson and Brame are subject to BART at all. Industry
Petitioners challenge the technical assumptions underlying the modeling
methods on which Louisiana and the EPA relied. Louisiana relied on the
“CALPUFF” model, and the EPA relied on both the CALPUFF and the “CAMx”
models. According to Industry Petitioners, the CALPUFF model overstates the
visibility effects of powerplant emissions. They also maintain that the EPA
5
Case: 18-60116 Document: 00515144673 Page: 6 Date Filed: 10/03/2019
No. 18-60116
exceeded its authority by using its own CAMx modeling to support Louisiana’s
modeling results.
Each petitioner in this case is also a respondent-intervenor. The
Environmental Petitioners, in addition to petitioning for review of some parts
of the Final Rule, also intervened and filed a responsive brief opposing
Industry Petitioners’ petition. Likewise, Industry Petitioners, in addition to
petitioning for review of some parts of the Final Rule, also intervened and filed
a responsive brief opposing Environmental Petitioners’ petition.
The EPA responded to both petitions, insisting that the Final Rule
should be approved in full. The LDEQ filed an amicus brief supporting the
EPA’s position.
A. Statutory and Regulatory Framework
This case is governed by the Clean Air Act and the regulations
implementing it. In 1977, “in response to a growing awareness that visibility
was rapidly deteriorating in many places, such as wilderness areas and
national parks,” Congress amended the Act by enacting § 169A. 5
That amendment established as a national goal “the prevention of any
future, and the remedying of any existing, impairment in visibility in
mandatory class I Federal areas which impairment results from manmade air
pollution.” 6 Protected class I Federal areas include “all (1) international parks,
(2) national wilderness areas which exceed 5,000 acres in size, (3) national
5 North Dakota v. EPA, 730 F.3d 750, 755 (8th Cir. 2013) (internal quotation omitted).
6 42 U.S.C. § 7491.
6
Case: 18-60116 Document: 00515144673 Page: 7 Date Filed: 10/03/2019
No. 18-60116
memorial parks which exceed 5,000 acres in size, and (4) national parks which
exceed 6,000 acres in size.” 7 Louisiana’s Breton National Wildlife Refuge and
the Caney Creek Wilderness Area in southwest Arkansas are the protected
Class I Federal areas at issue here. 8
The Clean Air Act is “an experiment in cooperative federalism,” in which
the federal government identifies pollutants and sets visibility targets, and the
states implement those standards though SIPs. 9 The Act directed the EPA to
issue regulations requiring: (1) that states submit SIPs to the EPA, (2) the
installation of the “best available retrofit technology, as determined by the
State . . . for controlling emissions” at specified air-pollution sources, and (3)
that each state adopt a long-term strategy “for making reasonable progress”
toward the national visibility goal. 10
Based on the Act’s directive, the EPA promulgated the Regional Haze
Rule in 1999. 11 “The Regional Haze Rule established the guidelines for state
compliance with the air visibility requirements of [the Clean Air Act].” 12 In
2005, the EPA revised that Rule and issued the BART guidelines, which set
out the process for states to establish BART emissions limitations. 13
7 Id. § 7472(a).
8 77 Fed. Reg. 11,839, 11,845 (Feb. 28, 2012).
9 Texas, 829 F.3d at 411.
10 42 U.S.C. § 7491(b)(2)(A)–(B). The EPA has stated that the national visibility goal
is “to attain natural visibility conditions by the year 2064.” 40 C.F.R. § 51.308(d)(1)(i)(B).
11 64 Fed. Reg. 35,714 (July 1, 1999); 40 C.F.R. § 51.308 (“Regional Haze Rule”).
12 Texas, 829 F.3d at 412.
13 70 Fed. Reg. 39,104 (July 6, 2005) (“BART Rule”); 40 C.F.R. Part 51, App. Y (“BART
Guidelines”).
7
Case: 18-60116 Document: 00515144673 Page: 8 Date Filed: 10/03/2019
No. 18-60116
The Clean Air Act and the Regional Haze Rule impose five requirements
for SIPs:
For each affected wilderness and national park, the plan must: (1)
set “reasonable progress goals” toward achieving natural visibility
conditions that ensure improvements in visibility on the most
impaired days over the period of the implementation plan; (2)
calculate baseline visibility and natural visibility conditions; (3)
devise a long-term strategy with enforceable emissions limitations,
compliance schedules, and other measures necessary to achieve
the reasonable progress goals; (4) develop a monitoring strategy
for measuring and reporting visibility; and (5) list the best
available retrofit technology . . . that emission sources in the state
will have to adopt to achieve the visibility goals, along with a
schedule for implementing BART. 14
After a state submits its SIP to the EPA, the agency reviews the SIP for
compliance with the Clean Air Act. If the EPA determines that a SIP does not
comply with the Act, it must promulgate a “Federal implementation plan” that
fixes the SIP’s shortcomings, unless the state corrects the deficiency. 15 The
EPA’s role is confined “to the ministerial function of reviewing SIPs for
consistency with the Act’s requirements.” 16
The parties focus on two of the Clean Air Act’s and Regional Haze Rule’s
requirements: (1) the BART emission limits and (2) the reasonable progress
goals and long-term strategy that each state must implement.
14 Texas, 829 F.3d at 412 (citing 40 C.F.R. § 51.308(d), (e)).
15 42 U.S.C. § 7410(k)(3), (c)(1). Because the EPA approved Louisiana’s SIP, there is
no Federal implementation plan at issue here.
16 Texas, 829 F.3d at 411 (quoting Luminant Generation Co. v. EPA, 675 F.3d 917, 921
(5th Cir. 2012)).
8
Case: 18-60116 Document: 00515144673 Page: 9 Date Filed: 10/03/2019
No. 18-60116
1. BART Emission Limits
The Regional Haze Rule defines BART as “an emission limitation based
on the degree of reduction achievable through the application of the best
system of continuous emission reduction for each pollutant which is emitted by
an existing stationary facility.” 17 The BART process has three steps. First, a
state must identify all “BART-eligible sources.” Second, it must determine
which of those BART-eligible sources are “subject to BART.” Third, for each
source that is “subject to BART,” the state must make a “BART determination”
by analyzing and selecting the appropriate emission control for that source. 18
Step one is identifying all “BART-eligible sources.” That definition
includes all stationary facilities that (1) were in existence before August 7,
1977 but were not in operation before August 7, 1962, (2) have “the potential
to emit 250 tons per year or more of any visibility-impairing air pollutant,” and
(3) fall within one of 26 listed source categories. 19 The parties agree that Nelson
and Brame satisfy these requirements and are “BART-eligible” sources.
Step two is making “subject to BART” determinations. The state must
determine which of the BART-eligible sources emit air pollutants that “may
reasonably be anticipated to cause or contribute to any impairment of
visibility” in a Class I area. 20 The key words are “cause” and “contribute to.”
1740 C.F.R. § 51.301.
18BART Guidelines, 40 C.F.R. Part 51, App. Y at I.F; see also Util. Air Regulatory Grp.
v. EPA, 471 F.3d 1333, 1335–36 (D.C. Cir. 2006) (explaining the process).
19 40 C.F.R. § 51.301.
20 Id.9
Case: 18-60116 Document: 00515144673 Page: 10 Date Filed: 10/03/2019
No. 18-60116
Sources that a state determines may reasonably be anticipated to “cause” or
“contribute to” visibility impairment at a protected area are “subject to
BART.” 21 Under the BART Guidelines, a 1.0 deciview 22 change from an
individual source “causes” visibility impairment, whereas a 0.5 deciview
change from an individual source “contributes to” impairment. 23 In some
instances, states may set a lower threshold for sources that “contribute to”
visibility impairment. 24
At the “subject to BART” screening step, states have discretion to either
(a) determine that all BART-eligible sources are “subject to BART” via
“collective attribution” or (b) conduct individualized testing to determine that
a source, or a group of sources, is exempt from BART. 25 Here, Louisiana opted
for the second option.
21 Id.22 Visibility impairments are measured in deciviews. A deciview “is the unit of
measurement on the deciview index scale for quantifying in a standard manner human
perceptions of visibility.” 40 C.F.R. § 51.301. “A higher deciview measurement indicates more
haze and less visibility. . . . A single deciview is around the increment that the average person
can perceive with the naked eye.” Texas, 829 F.3d at 413 n.2.
23 BART Rule, 70 Fed. Reg. at 39,120 (“[W]e are clarifying that for purposes of
determining which sources are subject to BART, States should consider a 1.0 deciview change
or more from an individual source to ‘cause’ visibility impairment, and a change of 0.5
deciviews to ‘contribute’ to impairment.”). For a fuller background on the “cause” or
“contribute” distinction, see id. at 39,117–39,122.
24 Id. at 39,120–39,121 (“In a regulatory context, we believe that a State’s decision as
to an appropriate threshold for contribution could depend upon the number of sources
affecting a class I area.”).
25 BART Guidelines, 40 C.F.R. Pt. 51, App. Y at III.
10
Case: 18-60116 Document: 00515144673 Page: 11 Date Filed: 10/03/2019
No. 18-60116
The third step is the “BART determination,” in which a state must
identify the appropriate emission control for each “subject to BART” source. 26
At each such source, the state must consider five statutory factors: (1) the costs
of compliance, (2) the energy and nonair quality environmental impacts of
compliance, (3) any existing pollution control technology in use at the source,
(4) the remaining useful life of the source, and (5) the degree of improvement
in visibility which may reasonably be anticipated to result from the use of such
technology. 27
For powerplants with generating capacities greater than 750 megawatts,
like those at issue here, a state’s BART determination must comply with the
BART guidelines, 28 which provide step-by-step instructions for making BART
determinations. 29 In contrast, at the BART-eligibility and “subject to BART”
steps, the BART guidelines are advisory only. 30
2. Reasonable Progress Goals and Long-Term Strategy
The Regional Haze Rule requires a state to “establish goals (expressed
in deciviews) that provide for reasonable progress towards achieving natural
visibility conditions” for each Class I area within that state. 31 The process and
considerations for calculating these goals are as follows:
26Id. at IV.
2742 U.S.C. § 7491(g)(2); Regional Haze Rule, 40 C.F.R. § 308(e)(1)(ii)(A) .
28 Regional Haze Rule, 40 C.F.R. § 51.308(e)(1)(ii)(B); 42 U.S.C. § 7491(b)(2); see also
77 Fed. Reg. 11,839, 11,849 (Feb. 28, 2012).
29 See BART Guidelines, 40 C.F.R. Part 51, App. Y § IV.
30 Util. Air Regulatory Grp., 471 F.3d at 1338–39.
31 40 C.F.R. § 51.308(d)(1).
11
Case: 18-60116 Document: 00515144673 Page: 12 Date Filed: 10/03/2019
No. 18-60116
A state begins by calculating the steady linear rate of decreasing
emissions that would achieve natural visibility in the covered
wildernesses and national parks by the year 2064. 40 C.F.R.
§ 51.308. If a state determines that the linear rate would result in
unreasonable regulations, it must propose an alternative set of
reasonable progress goals and demonstrate both that the linear
rate is unreasonable and that the alternative goals are reasonable.
Id. § 51.308(d)(1)(ii). The Clean Air Act and the Regional Haze
Rule require a state to consider four factors when setting
reasonable progress goals: “the costs of compliance, the time
necessary for compliance, and the energy and nonair quality
environmental impacts of compliance, and the remaining useful
life of any existing source subject to such requirements.” 42 U.S.C.
§ 7491(g)(1); see also 40 C.F.R. § 51.308(d)(1)(i)(A) (repeating the
factors listed in § 7491(g)(1)). 32
The evaluation of the four statutory factors is often referred to as a “‘four-factor
analysis’ or ‘reasonable progress analysis.’” 33
In addition to setting reasonable progress goals, a state must establish a
“long-term strategy” to make reasonable progress toward achieving natural
visibility conditions. 34 This strategy must include “enforceable emissions
limitations, compliance schedules, and other measures necessary to achieve
the reasonable progress goals.” 35 As part of its long-term strategy, a state may
impose additional emission reduction measures at sources that are not “subject
to BART.” 36 Sources that are regulated through the reasonable progress
32 Texas, 829 F.3d at 412–13.
33 81 Fed. Reg. 66,332, 66,360 (Sept. 27, 2016).
34 40 C.F.R. § 51.308(d)(3).
35 Id.; see 42 U.S.C. § 7491(b)(2)(A).
36 40 C.F.R. § 51.308(d)(3).
12
Case: 18-60116 Document: 00515144673 Page: 13 Date Filed: 10/03/2019
No. 18-60116
goals—rather than through BART—are referred to as “reasonable progress
sources.” 37
B. Factual Background
The procedural history of this case began when Louisiana submitted its
regional haze SIP to the EPA in 2008. Since then, Louisiana has revised and
resubmitted its SIP many times. Louisiana submitted its final SIP revisions in
October 2017, and the EPA approved Louisiana’s SIP in its December 2017
Final Rule. 38
1. Louisiana’s 2008 Submittal
Louisiana submitted a SIP for addressing regional haze in June 2008. In
that SIP, Louisiana relied on an emissions-trading program called the Clean
Air Interstate Rule (“CAIR”) to satisfy its BART, reasonable-progress, and
long-term strategy obligations. The EPA reviewed that SIP in two separate
actions in 2012. 39
The first such action, a national rulemaking, was based on a D.C. Circuit
decision that (1) held that states could not rely on CAIR to satisfy their BART
obligations and (2) remanded CAIR to the EPA. 40 In that national rulemaking,
the EPA finalized a “limited disapproval” of Louisiana’s SIP because it had
relied on CAIR. 41
37 Id.38 See Final Rule, 82 Fed. Reg. at 60,521–60,522 (“Our Previous Actions”).
39 See 82 Fed. Reg. 32,294, 32,295 (July 13, 2017) (detailing the procedural history).
40 See North Carolina v. EPA, 531 F.3d 896, 930 (D.C. Cir. 2008), reh’g granted in part,
550 F.3d 1176 (D.C. Cir. 2008).
41 77 Fed. Reg. 33,642, 33,643 (June 7, 2012).
13
Case: 18-60116 Document: 00515144673 Page: 14 Date Filed: 10/03/2019
No. 18-60116
The second such action “partially disapproved” Louisiana’s SIP
submission based on issues other than, and “go[ing beyond],” the SIP’s reliance
on CAIR. In addition to this “partial disapproval” of particular parts of
Louisiana’s SIP, that action also included a “partial limited approval” of other
parts of the SIP.
The “partial limited approval” part of that action was based on the EPA’s
conclusion that specified parts of the SIP, “as a whole, strengthen[ed] the
State’s SIP.” 42 Although the EPA concluded that some of the SIP provisions—
including Louisiana’s reasonable progress goals and long-term strategy—did
not comply with the Clean Air Act, those noncompliant provisions were
included in the partial limited approval. 43
In the “partial disapproval” part of that action, the EPA disapproved the
parts of the SIP related to CAIR, including the BART analyses for non-electric
generating unit (“EGU”) sources 44 and the BART determinations for four
specific non-EGU sources. 45 Those deficiencies are not at issue here.
A definitional note: Limited approvals and partial approvals are
different. If a submittal does not meet all of the Act’s requirements, but a
“separable” part does meet those requirements, a “partial approval may be
42 77 Fed. Reg. 39,425, 39,426 (July 3, 2012) (“2012 Final Rule”).
43 See id.44 “[T]he term ‘electric generating unit’ or ‘EGU’ is used to mean a solid fuel-fired
steam generating unit that serves a generator that produces electricity for sale to the electric
grid.” EPA Whitepaper on Available and Emerging Technologies for Reducing Greenhouse
Gas Emissions from Coal-Fired Electric Generating Units (Oct. 2010),
https://www.epa.gov/sites/production/files/2015-12/documents/electricgeneration.pdf.
45 2012 Final Rule, 77 Fed. Reg. at 39,426.
14
Case: 18-60116 Document: 00515144673 Page: 15 Date Filed: 10/03/2019
No. 18-60116
used to approve that part of the submittal and disapprove the remainder.” 46 In
contrast, if parts of a submittal do not meet the Clean Air Act’s requirements,
but “the submittal as a whole” has a “strengthening effect” on the SIP, the EPA
may use a “limited approval” to enact the entire submittal. 47 The practical
difference is
that under a limited approval the EPA’s approval action goes to
the entire [submittal]. In other words, although portions of a
[submittal] prevent the EPA from finding that the [submittal]
meets a certain requirement of the Act, the EPA believes that the
[submittal], as a whole, strengthens the SIP. Therefore, the EPA
approves the entire [submittal]--even those portions that prohibit
full approval. Likewise, when the EPA issues the limited
disapproval, the disapproval applies to the entire [submittal] as
failing to meet a specific requirement of the Act. The [submittal]
remains a part of the [enforceable] SIP, however, under the limited
disapproval, because the [submittal] strengthens the SIP. The
disapproval only applies to whether the submittal meets a specific
requirement of the Act and does not affect incorporation of the rule
into the approved, federally enforceable SIP. 48
In short, in a partial approval, the approved parts of the submittal go into effect
and the disapproved parts do not. In a limited approval, the entire part that is
“limitedly” approved—including the provisions within that part that do not
46 Processing of State Implementation Plan Revisions at 2, the EPA Memorandum
from John Calcagni (July 9, 1992), https://www.epa.gov/sites/production/files/2015-
07/documents/procsip.pdf.
47 Id. at 2–3.
48 Id. at 3.
15
Case: 18-60116 Document: 00515144673 Page: 16 Date Filed: 10/03/2019
No. 18-60116
comply with the Clean Air Act—goes into effect and are incorporated into the
approved SIP.
As a result of the partial disapproval, the EPA was required to
promulgate a federal implementation plan within two years of that
disapproval, unless it approved a corrected state plan. 49 The EPA did not
promulgate a federal plan or approve a corrected state plan within that
timeframe. Based on that, the Sierra Club sued the EPA, and a federal district
court entered a consent decree requiring the EPA to issue a federal plan or
approve a corrected state plan by December 15, 2017. 50
2. Louisiana’s 2017 Revisions and the EPA’s Approval
To bring its SIP into compliance, Louisiana revised its SIP to, among
other things, address BART for the Nelson and Brame powerplants. Nelson
and Brame are two of the largest sources of SO2 and NOx emissions in
Louisiana, are “BART-eligible,” and have generating capacities greater than
750 megawatts. 51
In August 2016, Louisiana submitted a partial plan addressing BART at
“non-EGU” sources. The EPA proposed to approve that partial plan in October
2016 and finalized the approval in the December 2017 Final Rule. 52 The
approval of that partial plan is not at issue here.
49 42 U.S.C. § 7410(c)(1).
50 See Notice of Modification of Consent Decree, Sierra Club v. Pruitt, No. 15-cv-01555-
JEB (D.D.C. Mar. 13, 2017).
51 Final Rule, 82 Fed. Reg. at 60,525–26.
52 81 Fed. Reg. 74,750 (Oct. 27, 2016) (proposed rulemaking); Final Rule, 82 Fed. Reg.
at 60,522 (“[W]e are finalizing that approval here.”).
16
Case: 18-60116 Document: 00515144673 Page: 17 Date Filed: 10/03/2019
No. 18-60116
In February 2017, Louisiana submitted a revised plan addressing BART
for EGU sources, including Nelson and Brame. In June 2017, Louisiana revised
its SIP to make a BART determination for Nelson. Louisiana revised its SIP
again in October 2017. The EPA’s December 2017 Final Rule explains the
relevant procedural history as follows:
On June 20, 2017, LDEQ submitted a SIP revision for parallel
processing related to Entergy’s Nelson facility. On July 13, 2017,
we proposed to approve this SIP revision along with the remaining
portion of the February 2017 SIP revision that addressed BART
for the Nelson facility. Specifically, we proposed to approve the
LDEQ BART determinations for Nelson Units 6 and 4, and the
Unit 4 auxiliary boiler, and the [administrative order on consent]
that makes the emission limits that represent BART permanent
and enforceable for the purposes of regional haze. We also solicited
comment with respect to any information that would support or
refute the costs in Entergy’s evaluation of SO2 controls for Unit 6.
On June 21, 2017, Entergy submitted a comment to LDEQ on its
proposed SIP revision requesting a three-year compliance deadline
to achieve the proposed SO2 BART limit for Nelson Unit 6.
Entergy’s letter explained that the company has coal contracts in
place for the next three years, so the revised compliance date would
provide the company sufficient time to transition to new mines
with lower sulfur coal. Additionally, Entergy stated that it did not
have the necessary equipment to blend varying fuel supplies. On
August 24, 2017, we received a letter from LDEQ explaining their
intent to revise the compliance date in the SIP revision for Nelson
Unit 6 based on Entergy’s comment letter. On September 26, 2017,
we supplemented our proposed approval of the SO2 BART
determination for Nelson by proposing to approve the three-year
compliance date. On October 26, 2017, we received LDEQ’s final
SIP revision addressing Nelson, including a final [administrative
order of consent] with emission limits and a SO2 compliance date
17
Case: 18-60116 Document: 00515144673 Page: 18 Date Filed: 10/03/2019
No. 18-60116
three years from the effective date of the EPA’s final approval of
the SIP revision. 53
In the October 2017 SIP revisions, Louisiana (1) determined that Nelson
and Brame were “subject to BART” and (2) determined that low-sulfur coal was
BART for controlling SO2 emissions at Nelson.
In reaching those conclusions, Louisiana considered analyses submitted
by Entergy, Cleco, and the EPA. The analyses submitted by the parties
addressed: (1) the modeling methods for making “subject to BART”
determinations and (2) the potential control options for BART at Nelson. The
EPA’s analyses contradicted and sharply criticized significant parts of
Entergy’s and Cleco’s analyses. The LDEQ included all the analyses as
appendices to its revised SIP.
i. The “Subject to BART” Determinations
The LDEQ concluded that Nelson and Brame were “subject to BART.”
The department reached that conclusion based on dispersion modeling that
established that those powerplants produced emissions that surpassed the 0.5
deciview threshold for “contributing to” visibility impairment at the Breton
National Wilderness Area and the Caney Creek Wilderness Area.
The LDEQ relied on the “CALPUFF” model, which “predicts 24-hour
average pollutant concentrations based on source emissions and how they
disperse in the atmosphere” and converts those concentrations to daily
53 82 Fed. Reg. at 60,522.
18
Case: 18-60116 Document: 00515144673 Page: 19 Date Filed: 10/03/2019
No. 18-60116
deciview effects. 54 The BART guidelines provide that a state may “use
CALPUFF or other appropriate model to estimate visibility impacts from a
single source at a Class I area.” 55 When the EPA promulgated those guidelines
in 2005, CALPUFF was “the best regulatory modeling application currently
available for predicting a single source’s contribution to visibility impairment
and [was then] the only EPA-approved model for use in estimating single
source pollutant concentrations resulting from the long range transport of
primary pollutants.” 56 Although the EPA has recently removed CALPUFF as
a “preferred” model for other air-quality modeling applications under the Clean
Air Act, CALPUFF remains a recommended model for making “subject-to-
BART” and BART emission control determinations. 57
Both Entergy’s and the EPA’s CALPUFF modeling showed that Nelson
and Brame each had a greater than 0.5 deciview impact on visibility
impairment at Breton and Caney Creek. Entergy, however, submitted reports
showing that CALPUFF’s reliance on flawed assumptions imposed a too-high
margin of error.
In addition to conducting CALPUFF modeling, Entergy and the EPA
conducted another type of modeling called “CAMx.” Entergy and the EPA
submitted their CAMx analyses to the LDEQ.
54 See BART Rule, 70 Fed. Reg. at 39,122 (describing the CALPUFF model in detail).
55 BART Guidelines, 40 C.F.R. Part 51, App. Y, III.A.3.
56 Id.
57 82 Fed. Reg. 5,182, 5,196 (Jan. 17, 2017) (“[T]his final action does not affect the
EPA’s recommendation that states use CALPUFF to determine the applicability and level of
best available retrofit technology in regional haze implementation plans.”).
19
Case: 18-60116 Document: 00515144673 Page: 20 Date Filed: 10/03/2019
No. 18-60116
The EPA’s CAMx modeling showed that Nelson and Brame exceeded the
0.5 deciview threshold. In contrast, Entergy’s CAMx modeling showed that
those powerplants’ emissions did not exceed the 0.5 deciview threshold. The
EPA, however, concluded that Entergy’s CAMx modeling “was not conducted
in accordance with the BART Guidelines and d[id] not properly assess
maximum baseline impacts, so [the EPA] consider[ed] this CAMx modeling
provided by Entergy to be invalid for supporting a determination of minimal
visibility impacts.” 58
When the LDEQ reviewed the models that the parties had submitted, it
stated that it did “not have the expertise with which to review those [CAMx]
model runs,” and instead relied only on the CALPUFF modeling. The LDEQ
did, however, include the CAMx analyses as appendices to its revised SIP.
ii. BART Determinations for Brame and Nelson
Louisiana’s revised SIP also made BART determinations for Brame and
Nelson. For the two Brame units, the LDEQ determined that “no additional
controls constitute[d] BART.” The LDEQ reached that determination based on
those units’ existing controls. At one of those units, BART was satisfied via the
unit’s earlier conversion from coal to natural gas. The other unit satisfied its
BART obligations based on an earlier dry sorbent injection installation. 59
For Nelson, the LDEQ changed its BART determination between the
February 2017 revision and the June and October 2017 revisions. In the
82 Fed. Reg. at 32,299.
58
Cleco does not challenge the LDEQ’s determination about the BART controls at the
59
Brame units; it challenges only the determination that the units are “subject to BART.”
20
Case: 18-60116 Document: 00515144673 Page: 21 Date Filed: 10/03/2019
No. 18-60116
February 2017 SIP revision, the LDEQ concluded that no additional controls
were necessary for Nelson to satisfy BART. Although the LDEQ stated that
“low sulfur coal presents the most feasible control based on economics and
impacts to visibility,” it also stated that it “believe[d] that the visibility
improvement that would be achieved through the installation and operation of
controls at each of the Nelson units would be negligible, therefore the facility’s
existing controls satisfy the BART requirements and no further controls are
necessary.”
In its June 2017 SIP revision, however, the LDEQ changed course and
determined that low-sulfur coal was BART for controlling SO2 at Nelson. The
October 2017 SIP revision retained that determination and extended the
compliance date for BART at Nelson by three years.
In the October 2017 SIP revision, the LDEQ stated that it evaluated four
different technologies for reducing emissions at Nelson: (1) low-sulfur coal, (2)
dry sorbent injection (“DSI”), (3) dry flue-gas desulfurization (“FGD”), or a “dry
scrubber,” and (4) wet FGD, or a “wet scrubber.” The LDEQ stated its
reasoning for selecting low-sulfur coal as BART at Nelson as follows:
In the Entergy BART five-factor SO2 analysis for the Unit 6
Boiler, a number of emission reduction controls were reviewed.
The reviewed controls included the use of a lower sulfur coal, DSI,
enhanced DSI, dry flue-gas desulfurization (FGD) and wet FGD.
LDEQ has reviewed and weighed the five factors carefully; after a
review of the information that Entergy and EPA provided, LDEQ
has concluded that the appropriate BART for this facility is to
establish an emission limit of .6 lbs/MMBtu based on a 30 day
rolling average as defined in the AOC (see Appendix D). While
additional visibility benefits may be available through the use of
21
Case: 18-60116 Document: 00515144673 Page: 22 Date Filed: 10/03/2019
No. 18-60116
FGD, the lower sulfur coal option results in visibility benefits at a
lower annual cost. Along with the extra cost, FGD use results in
additional waste spent due to spent reagent and has some power
demands to run the equipment. LDEQ believes, at present, that
the use of lower sulfur coal presents the appropriate SO2 control
based on consideration of economics, energy impacts, non-air
quality environmental impacts, and impacts to visibility.
iii. Long-Term Strategy and Reasonable Progress Goals
Louisiana’s revised SIP did not specifically set out a long-term strategy
or impose emission controls at additional “reasonable progress sources.”
Neither did it evaluate the four “reasonable progress factors” for determining
whether any non-BART reasonable progress sources should be controlled.
3. The EPA’s Approval of Louisiana’s SIP
The EPA proposed two separate rules to approve Louisiana’s revised
SIP. In May 2017, the EPA issued a proposed rule to approve the entire
February 2017 SIP revision except for the part about Nelson, on which the EPA
deferred action. 60 This rulemaking included an approval of Louisiana’s “subject
to BART” determination for Brame. 61
In June 2017, Louisiana revised its SIP by changing its previous “no
further controls” BART determination for Nelson to require the use of low-
sulfur coal. In a July 2017 rulemaking (“the Proposed Rule”), the EPA proposed
to approve the remaining part of Louisiana’s SIP addressing Nelson. 62 The
60 82 Fed. Reg. 22,936 (May 19, 2017).
61 Id. at 22,942–22,943.
62 Proposed Rule, 82 Fed. Reg. 32,294, 32,300 (July 13, 2017).
22
Case: 18-60116 Document: 00515144673 Page: 23 Date Filed: 10/03/2019
No. 18-60116
Proposed Rule (1) agreed with the LDEQ that Nelson was “subject to BART”
and (2) stated that the LDEQ had adequately analyzed the five mandatory
BART factors. 63 But the Proposed Rule also criticized several aspects of the
SIP revision, including Entergy’s modeling, cost estimates, and supporting
documentation. The Proposed Rule stated that the EPA performed its own
modeling and analyses to correct those errors. 64
After the EPA issued the Proposed Rule, Louisiana submitted a letter to
the EPA explaining the state’s intention to extend the compliance date in its
draft SIP for Nelson by three years. In response to that letter, the EPA
supplemented the Proposed Rule to reflect such change. 65
Louisiana revised its SIP again in October 2017. As discussed above, that
revision extended the date for Nelson to comply with BART by three years. The
EPA finalized its proposed approval of all of Louisiana’s SIP revisions when it
issued the Final Rule on December 21, 2017. 66
The Environmental and Industry Petitioners timely petitioned for
review of the Final Rule. On the same day that this case was filed, the
Environmental Petitioners also filed an administrative petition for
reconsideration of the Final Rule. This court stayed the case pending the EPA’s
63 Id. at 32,296–32,297.
64 Id. at 32,297–32,298.
65 82 Fed. Reg. 44,753 (Sept. 26, 2017).
66 Final Rule, 82 Fed. Reg. 60,520 (Dec. 21, 2017).
23
Case: 18-60116 Document: 00515144673 Page: 24 Date Filed: 10/03/2019
No. 18-60116
decision on the administrative petition, then lifted the stay after the EPA
denied that petition. 67
II. STANDARD OF REVIEW
We review the EPA’s approval of Louisiana’s regional haze SIP under
the standards set out in the Administrative Procedure Act (“APA”), which
require us to set aside an agency action that is “arbitrary, capricious, an abuse
of discretion, or otherwise not in accordance with law.” 68 An action is arbitrary
and capricious “if the agency has relied on factors which Congress has not
intended it to consider, entirely failed to consider an important aspect of the
problem, offered an explanation for its decision that runs counter to the
evidence before the agency, or is so implausible that it could not be ascribed to
a difference in view or the product of agency expertise.” 69
The arbitrary and capricious standard is “narrow,” and we must “be
mindful not to substitute [our] judgment for that of the agency.” 70 We “must
also ensure that the agency ‘examined the relevant data and articulated a
satisfactory explanation for its action.’” 71 “We consider whether the decision
67 See EPA’s Response to Reconsideration Petition, https://www.regulations.gov/
contentStreamer?documentId=EPA-R06-OAR-2016-0520-0011&contentType=pdf. Although
we take notice of the administrative petition for reconsideration and the EPA’s response to
it, our review of the Final Rule “is limited to the record before the agency at the time of its
decision.” Luminant, 675 F.3d at 925 (quotation omitted).
68 5 U.S.C. § 706(2)(A).
69 Tex. Oil & Gas Ass’n v. EPA, 161 F.3d 923, 933 (5th Cir. 1998) (quoting Motor
Vehicle Mfrs. Ass’n v. State Farm Mut. Auto Ins. Co., 463 U.S. 29, 43 (1983)).
70 10 Ring Precision, Inc. v. Jones, 722 F.3d 711, 723 (5th Cir. 2013) (quoting Motor
Vehicle Mfrs. Ass’n, 463 U.S. at 43).
71 Id. (quoting Motor Vehicle Mfrs. Ass’n, 463 U.S. at 43).
24
Case: 18-60116 Document: 00515144673 Page: 25 Date Filed: 10/03/2019
No. 18-60116
was based on a consideration of the relevant factors and whether there has
been a clear error of judgment.” 72
Additionally, federal agencies “are required to engage in ‘reasoned
decisionmaking.’” 73 The agency’s process must be “logical and rational,” and its
decision “is lawful only if it rests ‘on a consideration of the relevant factors.’” 74
The agency must “articulate a satisfactory explanation for its action including
a ‘rational connection between the facts found and the choice made.’” 75 Under
this standard, courts have vacated agency decisions that created “unexplained
inconsistencies in the rulemaking record.” 76
Under the Clean Air Act’s structure of cooperative federalism, Louisiana
is the entity considering “relevant factors,” and the EPA’s role is confined to
ensuring that Louisiana’s determinations complied with the Clean Air Act. 77
Under this structure, we review whether the EPA was arbitrary and capricious
in approving Louisiana’s compliance with the Act.
72 Id. (quoting Motor Vehicle Mfrs. Ass’n, 463 U.S. at 43).
73 Michigan v. EPA, 135 S. Ct. 2699, 2706 (2015) (quoting Allentown Mack Sales &
Serv., Inc. v. NLRB, 622 U.S. 359, 374 (1998)).
74 Id. (quoting Motor Vehicle Mfrs. Ass’n, 463 U.S. at 43).
75 Motor Vehicle Mfrs. Ass’n, 463 U.S. at 43 (quoting Burlington Truck Lines v. United
States, 371 U.S. 156, 168 (1962)).
76 U.S. Sugar Corp. v. EPA, 830 F.3d 579, 651 (D.C. Cir. 2016); see id. at 650 (collecting
D.C. Circuit authority and stating “[t]his court has ‘often declined to affirm an agency
decision if there are unexplained inconsistencies in the final rule’” (citation omitted)); Gulf
Power Co. v. FERC, 983 F.2d 1095, 1101 (D.C. Cir. 1993) (“[W]hen an agency takes
inconsistent positions . . . it must explain its reasoning.); Gen. Chem. Corp. v. United States,
817 F.2d 844, 846 (D.C. Cir. 1987) (holding agency action to be arbitrary because its analysis
was “internally inconsistent and inadequately explained”).
77 See Texas, 829 F.3d at 411.
25
Case: 18-60116 Document: 00515144673 Page: 26 Date Filed: 10/03/2019
No. 18-60116
III. ANALYSIS
As a threshold matter, Industry Petitioners and Environmental
Petitioners have standing. Industry Petitioners’ standing is based on Entergy’s
and Cleco’s ownership of the powerplants at issue. Environmental Petitioners
have standing because one of the Sierra Club’s members submitted a
declaration adequately asserting that (1) he regularly visits the national parks
in question, (2) he has concrete plans to return in the future, and (3) regional
haze affects his visibility. Although the declarations submitted by the National
Parks Conservation Association do not appear to give it standing in this case,
“one party with standing is sufficient to satisfy Article III’s case-or-controversy
requirement.” 78
Each petition raises two major issues. Environmental Petitioners
challenge the EPA’s approval of (1) Louisiana’s selection of low-sulfur coal as
BART for controlling SO2 emissions at Nelson and (2) Louisiana’s reasonable
progress goals and long-term strategy.
Industry Petitioners challenge the EPA’s approval of Louisiana’s
“subject to BART” determinations at Nelson and Brame. They contend that (1)
the CALPUFF modeling that Louisiana relied on was technically flawed; (2)
the CAMx modeling, on which the EPA relied but Louisiana did not, was
technically flawed; and (3) even if the CAMx modeling were not technically
78Brackeen v. Bernhardt, 937 F.3d 406, 421 (5th Cir. 2019) (quoting Rumsfeld v.
Forum for Acad. & Institutional Rights, Inc., 547 U.S. 47, 53 n.2 (2006)).
26
Case: 18-60116 Document: 00515144673 Page: 27 Date Filed: 10/03/2019
No. 18-60116
flawed, the EPA should not have considered it because Louisiana expressly did
not consider it.
We address each issue in turn.
A. The Environmental Petitioners’ Petition
The first part of Environmental Petitioners’ challenge addresses the
EPA’s approval of Louisiana’s determination that low-sulfur coal was BART
for Nelson. The second part of their challenge addresses the EPA’s approval of
Louisiana’s long-term strategy and reasonable progress goals.
We deny Environmental Petitioners’ petition, and we explain our
reasons below.
1. The EPA’s Approval of the BART Determination for Nelson
In its October 2017 SIP revision, Louisiana selected low-sulfur coal as
BART for controlling SO2 emissions at Nelson. The LDEQ explained its
determination as follows:
In the Entergy BART five-factor SO2 analysis for the Unit 6
Boiler, a number of emission reduction controls were reviewed.
The reviewed controls included the use of a lower sulfur coal, DSI,
enhanced DSI, dry flue-gas desulfurization (FGD) and wet FGD.
LDEQ has reviewed and weighed the five factors carefully; after a
review of the information that Entergy and EPA provided, LDEQ
has concluded that the appropriate BART for this facility is to
establish an emission limit of .6 lbs/MMBtu based on a 30 day
rolling average as defined in the AOC (see Appendix D). While
additional visibility benefits may be available through the use of
FGD, the lower sulfur coal option results in visibility benefits at a
lower annual cost. Along with the extra cost, FGD use results in
additional waste spent due to spent reagent and has some power
demands to run the equipment. LDEQ believes, at present, that
27
Case: 18-60116 Document: 00515144673 Page: 28 Date Filed: 10/03/2019
No. 18-60116
the use of lower sulfur coal presents the appropriate SO2 control
based on consideration of economics, energy impacts, non-air
quality environmental impacts, and impacts to visibility.
Environmental Petitioners contend that the EPA’s approval of this
determination was arbitrary and capricious for three reasons. First,
Louisiana’s selection of low-sulfur coal as BART at Nelson was inconsistent
with the evidence that the state considered. Second, Louisiana did not provide
a “rational basis” for rejecting more effective pollution controls. Third, in its
BART determination, Louisiana did not weigh the Clean Air Act’s factors in
compliance with the mandatory BART guidelines.
i. The Evidence Louisiana Considered
According to Environmental Petitioners, Louisiana’s BART
determination rested solely on analysis by Nelson’s owner, Entergy.
Environmental Petitioners point out that the EPA was “unable to verify any of
the company’s costs” because those costs were based on a proprietary database
to which the EPA was not given access. 79 Similarly, the EPA could not verify
Entergy’s modeling analyses because Entergy did not provide the “inputs” that
were used in the modeling.
For the parts of Entergy’s analyses that the EPA did review, the EPA
concluded that Entergy’s cost and visibility analyses, on which Louisiana
relied, had many errors, including: (1) implementing improper costs, (2)
inflated contingency estimates, and (3) modeling errors. The EPA pointed out
79 Proposed Rule, 82 Fed. Reg. at 39,432.
28
Case: 18-60116 Document: 00515144673 Page: 29 Date Filed: 10/03/2019
No. 18-60116
these errors in the Proposed Rule: (1) “Entergy’s control cost estimates
included costs not allowed under our Control Cost Manual (e.g., escalation
during construction and owner’s costs)”; (2) “Entergy also assumed a
contingency of 25%, which we note is unusually high”; and (3) “Entergy’s
CALPUFF modeling included errors in its estimates of sulfuric acid and PM
emissions.” 80
After noting these perceived errors, the EPA submitted its own BART
analyses that reached markedly different cost-effectiveness estimates for each
proposed control option at Nelson. The following table compares the EPA’s and
Entergy’s cost-effectiveness calculations, in cost per ton of pollutant removed,
for each control option.
Cost-Effectiveness Comparison 81
Technology Entergy’s Calculation the EPA’s Calculation
Low-sulfur coal $597 $2,957
DSI $5,611 $3,578 - $4,302
Wet scrubber $4,413 $2,743
Dry scrubber (SDA) $4,536 $2,706
Most striking is the difference between the EPA’s and Entergy’s
calculations for low-sulfur coal. Entergy estimated that low-sulfur coal would
cost $597 per ton of pollution removed; the EPA estimated that it would cost
$2,957 per ton. Entergy’s estimate for what the parties agree would be the most
80Proposed Rule, 82 Fed. Reg. at 32,298–32,299.
81The Proposed Rule contains a thorough narrative discussion of the EPA’s analyses
and the problems with Entergy’s analyses. 82 Fed. Reg. at 32,298–32,299.
29
Case: 18-60116 Document: 00515144673 Page: 30 Date Filed: 10/03/2019
No. 18-60116
effective option—a dry scrubber—was significantly higher than the EPA’s
estimate. Entergy’s estimate was $4,536 per ton of pollution removed; the
EPA’s was $2,706 per ton.
Similarly, the EPA’s conclusions about the visibility improvements for
each option differed from Entergy’s conclusions. In the Proposed Rule, the EPA
explained that the differences resulted from Entergy’s model’s failure to follow
the BART guidelines. 82 The following chart compares the visibility
improvement calculations (in deciviews) for each control option.
Expected Visibility Improvement Comparison for Caney Creek
Technology Trinity Consultants’ the EPA’s Calculation
Calculation
Low-sulfur coal 0.164 0.411
DSI 0.302 0.511
Dry scrubber (SDA) 0.355 0.831
According to Environmental Petitioners, Louisiana’s reliance on
Entergy’s analyses—which the EPA’s analyses concluded overestimated the
costs and underestimated the benefits of more effective pollution controls—was
irrational. After the EPA informed Louisiana of those errors, that state
reached the same BART determination. In its SIP revision, Louisiana did not
acknowledge the EPA’s criticisms and did not attempt to reconcile the
82 82 Fed. Reg. at 32,299 (“As we discuss above and in the CAMx Modeling TSD,
Entergy also provided additional screening modeling results using CAMx to support its
conclusion that visibility impacts from Unit 6 are minimal. However, this modeling was not
conducted in accordance with the BART Guidelines and does not properly assess maximum
baseline impacts, so we consider this CAMx modeling provided by Entergy to be invalid for
supporting a determination of minimal visibility impacts.”).
30
Case: 18-60116 Document: 00515144673 Page: 31 Date Filed: 10/03/2019
No. 18-60116
conflicting the EPA and Entergy analyses. Environmental Petitioners insist
that Louisiana’s failure to explain why its decision remained unchanged after
reviewing the EPA’s contradictory analyses created an “unexplained
inconsistenc[y] in the rulemaking record” that made the EPA’s approval
arbitrary and capricious. 83
The EPA does not defend these discrepancies. In its brief, it concedes
that the agency “was aware of these errors when it approved the SIP,”
acknowledges that the SIP revision contained “a number of errors,” and
describes Entergy’s submissions as “Entergy’s faulty analyses.”
The EPA contends that, despite these errors, Louisiana “may, to the
extent supported by the record as a whole, reach the same conclusion both
before and after reviewing a particular set of information.” 84 The EPA points
to the fact that its own analyses were in the record and that the LDEQ included
those analyses in an appendix to the revised SIP. The EPA insists that, based
on the agency’s earlier review of the “entirety of” Louisiana’s SIP submission,
Louisiana considered all the information in its October 2017 SIP submission,
weighed that information in arriving at its final BART determination, and
explained the reasons for its decision.
Industry Respondents defend the analyses they submitted to the LDEQ
and insist that those analyses adequately supported Louisiana’s BART
determination. They criticize the analyses that the EPA submitted to the
83U.S. Sugar Corp, 830 F.3d 579 at 651.
84See 42 U.S.C. § 7410(k)(3) (“[the EPA] shall approve such submittal as a whole if it
meets all of the applicable requirements of [the Act].”).
31
Case: 18-60116 Document: 00515144673 Page: 32 Date Filed: 10/03/2019
No. 18-60116
LDEQ and challenge the EPA’s earlier determination that their analyses failed
to comply with the BART guidelines.
Environmental Petitioners have the better arguments on this point.
Entergy and the EPA reached significantly different conclusions about the
costs and visibility improvements of each emission control. But after the EPA
submitted analyses pointing out flaws in Entergy’s analysis, the LDEQ did not
discuss the EPA-identified errors, and it reached the same BART
determination for Nelson as it did before the EPA identified mistakes. Given
the EPA’s analysis showing that Entergy significantly underestimated the cost
of low-sulfur coal and overestimated the cost of a scrubber, Louisiana’s failure
to address or reconcile the conflicting analyses appears to create an
“unexplained inconsistenc[y]” in the rulemaking record. 85 Although that
shortcoming in Louisiana’s SIP is worthy of careful scrutiny, it does not fully
resolve the matter.
ii. Rational Explanation for Rejecting a More Effective
Control
Environmental Petitioners next contend that Louisiana did not provide
a rational explanation for rejecting a more effective pollution control at Nelson.
They point to the EPA’s past actions that reviewed BART determinations for
whether the costs and visibility benefits of a determination were within the
range of the EPA’s prior BART determinations. 86 They also cite the BART
85See U.S. Sugar Corp., 830 F.3d at 651.
86See 82 Fed. Reg. 912,938 (Jan. 4, 2017) (“[T]he cost-effectiveness of all of the controls
that form the basis of our proposed BART determinations are within a range found to be
32
Case: 18-60116 Document: 00515144673 Page: 33 Date Filed: 10/03/2019
No. 18-60116
guidelines, which recommend that “[a] reasonable range would be a range that
is consistent with the range of cost effectiveness values used in other similar
permit decisions over a period of time.” 87
The LDEQ’s SIP revisions did not provide cost-effectiveness numbers for
any of the control options it considered. Based on the LDEQ’s failure to include
those numbers in its SIP, the EPA was not able to compare the instant BART
determination with the EPA’s previously approved BART determinations.
According to Environmental Petitioners, this lack of a comparison to prior
EPA-approved BART determinations amounts to an “unexplained deviation
from past practice.” 88
To bolster this argument, Environmental Petitioners compared the
EPA’s cost-effectiveness numbers for a scrubber with all of the EPA’s prior
BART determinations. Environmental Petitioners included a graph of those
results, which shows that the cost-effectiveness numbers for a scrubber align
with the EPA’s prior BART decisions.
The EPA responds that, although in other actions it has stated that
comparisons to prior BART determinations are helpful, the BART guidelines
do not require such comparisons. The EPA points to the Final Rule’s
statements explaining that BART determinations depend on the unique
acceptable in other case.”); 80 Fed. Reg. 18,944, 18,952 (Apr. 8, 2015) (“[T]he cost effectiveness
. . . is within the range of what we consider to be cost-effective for BART.”).
87 BART Guidelines, 40 C.F.R. part 51, App. Y § IV.D.4.f.
88 See Encino Motorcars, LLC v. Navarro, 136 S. Ct. 2117, 2125–26 (2016); WildEarth
Guardians v. EPA, 770 F.3d 919, 941 (10th Cir. 2014).
33
Case: 18-60116 Document: 00515144673 Page: 34 Date Filed: 10/03/2019
No. 18-60116
circumstances of each source and that “[s]ome variation is to be expected
because SIP actions are highly fact-dependent.” 89 The EPA also cites several
cases holding that states have broad authority to weigh the statutory factors
and to pick appropriate emission controls. 90
The EPA has the better argument on this point. Louisiana has the
authority to select the appropriate BART emission control. The fact that a
scrubber more closely aligns with the EPA’s prior BART determinations did
not preclude Louisiana from choosing low-sulfur coal instead. The guideline
that Environmental Petitioners cite does not require that BART
determinations align with the EPA’s prior BART determinations:
You should provide documentation of any unusual circumstances
that exist for the source that would lead to cost-effectiveness
estimates that would exceed that for recent retrofits. This is
especially important in cases where recent retrofits have cost-
effectiveness values that are within what has been considered a
reasonable range, but your analysis concludes that costs for the
source being analyzed are not considered reasonable. (A
reasonable range would be a range that is consistent with the
range of cost effectiveness values used in other similar permit
decisions over a period of time.) 91
Although the parties do not explain how low-sulfur coal compares to the
EPA’s prior BART determinations, the record is devoid of any “unusual
89 82 Fed. Reg. at 60,534–60,535.
90 See Oklahoma v. EPA, 723 F.3d 1201, 1209 (10th Cir. 2013) (“[S]tates ha[ve] broad
authority to weigh the statutory factors and make BART determinations.”); see also Ariz. ex
rel. Darwin v. EPA, 852 F.3d 1148, 1161 (9th Cir. 2017).
91 BART Guidelines, 40 C.F.R. part 51, App. Y § IV.D.4.f.
34
Case: 18-60116 Document: 00515144673 Page: 35 Date Filed: 10/03/2019
No. 18-60116
circumstances” indicating that low-sulfur coal’s cost-effectiveness estimates
are out of line with prior BART determinations. Therefore, the guideline
addressing “similar permit decisions over a period of time” does not apply here.
iii. Weighing the Statutory Factors and Compliance with
the BART Guidelines
In making a BART determination, a state “shall take into consideration”
five factors: (1) the costs of compliance, (2) the energy and nonair quality
environmental impacts of compliance, (3) any existing pollution control
technology in use at the source, (4) the remaining useful life of the source, and
(5) the degree of improvement in visibility that may reasonably be anticipated
to result from the use of such technology. 92 States have discretion as to “the
weight and significance” of each factor, as well as the appropriate emission
control that qualifies as BART at a given source. 93
In addition to considering those statutory factors, for powerplants with
generating capacities greater than 750 megawatts, like Nelson and Brame, the
state’s BART determination “shall be determined pursuant to [the BART]
guidelines.” 94 The BART guidelines state:
From the alternatives you evaluated . . . we recommend you
develop a chart (or charts) displaying for each of the alternatives:
(1) Expected emission rate . . . ;
92 42 U.S.C. § 7491(g)(2).
93 BART Rule, 70 Fed. Reg. at 39,123; see Oklahoma, 723 F.3d at 1209 (“[S]tates ha[ve]
broad authority to weigh the statutory factors and make BART determinations.”).
94 42 U.S.C. § 7491 (b)(2)(B); see 40 C.F.R. § 51.308(e)(1)(ii)(B). In its brief, the EPA
concedes that the BART Guidelines are mandatory at this step.
35
Case: 18-60116 Document: 00515144673 Page: 36 Date Filed: 10/03/2019
No. 18-60116
(2) Emissions performance level . . . ;
(3) Expected emissions reductions . . . ;
(4) Costs of compliance—total annualized costs ($), cost
effectiveness ($/ton), and incremental cost effectiveness
($/ton), and/or any other cost-effectiveness measures (such
as $/deciview);
(5) Energy impacts;
(6) Non-air quality environmental impacts; and
(7) Modeled visibility impacts. 95
....
You have discretion to determine the order in which you should
evaluate control options for BART. Whatever the order in which
you choose to evaluate options, you should always (1) display the
options evaluated; (2) identify the average and incremental costs
of each option; (3) consider the energy and non-air quality
environmental impacts of each option; (4) consider the remaining
useful life; and (5) consider the modeled visibility impacts. You
should provide a justification for adopting the technology that you
select as the “best” level of control, including an explanation of the
CAA factors that led you to choose that option over other control
levels. 96
The LDEQ’s explanation of its determination that low-sulfur coal was
BART for Nelson states:
In the Entergy BART five-factor SO2 analysis for the Unit 6
Boiler, a number of emission reduction controls were reviewed.
The reviewed controls included the use of a lower sulfur coal, DSI,
enhanced DSI, dry flue-gas desulfurization (FGD) and wet FGD.
LDEQ has reviewed and weighed the five factors carefully; after a
review of the information that Entergy and EPA provided, LDEQ
95 BART Guidelines, 40 C.F.R. part 51, App. Y § IV.E.1 (emphasis added).
96 Id. (emphasis added).
36
Case: 18-60116 Document: 00515144673 Page: 37 Date Filed: 10/03/2019
No. 18-60116
has concluded that the appropriate BART for this facility is to
establish an emission limit of .6 lbs/MMBtu based on a 30 day
rolling average as defined in the AOC (see Appendix D). While
additional visibility benefits may be available through the use of
FGD, the lower sulfur coal option results in visibility benefits at a
lower annual cost. Along with the extra cost, FGD use results in
additional waste spent due to spent reagent and has some power
demands to run the equipment. LDEQ believes, at present, that
the use of lower sulfur coal presents the appropriate SO2 control
based on consideration of economics, energy impacts, non-air
quality environmental impacts, and impacts to visibility.
In the 2017 Final Rule, the EPA approved this explanation as compliant
with the Clean Air Act. The Final Rule relied on the language in the revised
SIP, which states that “LDEQ has weighed the five factors carefully” and
reached its decision “after a review of both Entergy’s and the EPA’s
information.” 97 Referring to this statement, the Final Rule concludes:
This indicates that the State reviewed the information it received
from both Entergy and the EPA, and thus had adequate
information upon which to base its determination. After reviewing
the relevant information contained in LDEQ’s SIP, we determined
that the State’s SIP meets the requirements of the Act and the
applicable regulations and guidance. 98
Environmental Petitioners challenge the EPA’s approval of the LDEQ’s
explanation of its BART determination. Those Petitioners contend that,
because the LDEQ failed to discuss how it weighed each statutory factor, it did
not comply with the BART guidelines and therefore did not satisfy its
97 Final Rule, 82 Fed. Reg. at 60,532.
98 Id.
37
Case: 18-60116 Document: 00515144673 Page: 38 Date Filed: 10/03/2019
No. 18-60116
obligations under the Clean Air Act. They point to the LDEQ’s failure to
evaluate (1) the existing pollution controls at Nelson, (2) the remaining useful
life of Nelson, (3) the cost-effectiveness of pollution controls, and (4) the
visibility benefits of pollution controls.
In response, the EPA concedes that Louisiana “could have expanded its
discussion of its decisionmaking process.” The EPA nonetheless maintains that
it “determined that the path Louisiana took in weighing the five factors and
reaching its BART determination could be reasonably discerned from the
record before it.” The EPA’s and Industry Respondents’ arguments focus on the
fact that Entergy’s and the EPA’s analyses—which were included as
appendices to the SIP revisions—did the appropriate analysis and considered
all the factors.
Environmental Petitioners are correct that the LDEQ’s explanation did
not discuss the Clean Air Act’s factors of (1) the existing controls at Nelson, (2)
Nelson’s remaining useful life, or (3) the BART guidelines factor of “cost-
effectiveness.”
On the “existing control” factor, the LDEQ’s failure to discuss the
existing controls at Nelson highlights a factual dispute between the parties.
Environmental Petitioners maintain that low-sulfur coal is already in use at
Nelson and has been since 2015. The EPA’s June 2017 technical support
document states that “beginning in the Spring of 2015, Entergy began
purchasing coals with lower sulfur contents that occur in a tighter range in
comparison to those purchased in the past.”
38
Case: 18-60116 Document: 00515144673 Page: 39 Date Filed: 10/03/2019
No. 18-60116
Regarding the “remaining useful life” factor, the LDEQ’s explanation of
its BART determination does not mention or address that factor. Neither does
the LDEQ’s explanation mention “cost-effectiveness.” The BART guidelines
require states to consider the cost-effectiveness of potential control options. 99
Those guidelines define that term as follows:
Cost-effectiveness, in general, is a criterion used to assess the
potential for achieving an objective in the most economical way.
For purposes of air pollutant analysis, “effectiveness” is measured
in terms of annualized control costs. We recommend two types of
cost-effectiveness calculations - average cost effectiveness, and
incremental cost effectiveness.
...
Average cost effectiveness means the total annualized costs of
control divided by annual emissions reductions (the difference
between baseline annual emissions and the estimate of emissions
after controls), using the following formula . . . . 100
Louisiana’s explanation mentions “annual cost” but does not mention
“cost-effectiveness.” As the BART guidelines explain, annual cost and cost-
effectiveness are different: Annual cost is the annualized capital cost for a
control; cost-effectiveness is the annual cost of a control divided by the annual
emissions reductions of that control. In general, the least effective control
technology will almost always have the lowest absolute (or annual) cost. In
contrast, the most effective control technology will often have the highest
99 See BART Guidelines, 40 C.F.R. part 51, App. Y § IV.D.4.b; see also id. § IV.E.1.
100 Id. § IV.D.4.(b)–(c).
39
Case: 18-60116 Document: 00515144673 Page: 40 Date Filed: 10/03/2019
No. 18-60116
absolute (or annual) cost. 101 Based on this distinction, the BART guidelines
impose cost-effectiveness, i.e., the cost per ton of pollution removed—instead of
annual cost—as the relevant consideration.
* * *
The Clean Air Act requires states to consider five statutory factors when
making a BART determination. The EPA then must approve or disapprove
that determination. Our role is to decide whether the EPA was arbitrary and
capricious in approving Louisiana’s SIP.
The LDEQ’s key statements on this issue are that (1) it “reviewed and
weighed the five factors carefully,” (2) it reached its decision after “review[ing]
. . . the information that Entergy and EPA provided,” and (3) the “lower sulfur
coal option results in visibility benefits at a lower annual cost.”
The LDEQ’s short explanation is lacking in several respects. It did not
(1) provide numbers supporting its conclusions, (2) address or reconcile the
EPA’s criticism of Entergy’s analyses, (3) compare the costs or benefits of each
control option, or (4) discuss Nelson’s existing controls, remaining useful life,
or the cost-efficiency of the potential control options. The EPA’s and Industry
Respondents’ primary contention is that because Entergy and the EPA did the
appropriate analysis and weighed the statutory factors—and the LDEQ said it
reviewed those analyses and included them as appendices—the LDEQ
appropriately weighed the factors. The parties’ briefing on how Louisiana
weighed the factors focuses almost entirely on the analyses that the EPA and
101 See Proposed Rule, 82 Fed. Reg. at 32,298–32,299.
40
Case: 18-60116 Document: 00515144673 Page: 41 Date Filed: 10/03/2019
No. 18-60116
Entergy submitted. None of the briefs, however, cite any place in the record in
which the LDEQ discusses all five statutory factors.
Moreover, the parties rely on the LDEQ’s statement that it considered
the analyses that Entergy and the EPA provided before making its
determination, as well as the fact that those analyses were included as
appendices to the SIP revisions. But the LDEQ expressly disavowed
“agree[ing] with,” “adopt[ing],” or “incorporat[ing]” any of Entergy’s or the
EPA’s calculations in its revised SIP:
LDEQ submits these comments in response to certain statements
in the preamble [of the 2017 Proposed Rule] that matters were
“adopted and incorporated” into the LDEQ SIP revision. LDEQ
places all documents and information submitted to it in connection
with the development of the SIP in an administrative record . . . .
However, placement in the record does not indicate that LDEQ
agrees with or has “adopted” positions, conclusions, or decisions,
nor has “incorporated” them into the SIP revision submitted to the
EPA. The final SIP document and any enforceable conditions
included therein encompass the final decision by LDEQ.
Based on this express disavowal, we cannot conclude that the analyses Entergy
and the EPA submitted to the LDEQ were incorporated into the terms of
Louisiana’s revised SIP.
And, in contrast to the BART guidelines’ detailed instructions for making
a BART determination, the LDEQ’s explanation of its BART determination for
Nelson provides hardly any details. The guidelines recommend “always”
identifying the average and incremental costs of each option, but the LDEQ
provided neither specific numbers supporting that decision nor a comparison
41
Case: 18-60116 Document: 00515144673 Page: 42 Date Filed: 10/03/2019
No. 18-60116
of the costs and estimated visibility benefits for each control option. Unlike the
detailed analyses that the EPA and Entergy submitted—which did provide cost
and pollution-control estimates for each control option—the LDEQ simply
stated that low-sulfur coal had an unspecified “lower annual cost.”
Despite these shortcomings, we conclude that the EPA’s approval of
Louisiana’s BART determination was not arbitrary and capricious. 102 The
standard of review is narrow and highly deferential. Our review is limited to
“whether the decision was based on a consideration of the relevant factors and
whether there has been a clear error of judgment.” 103
There was a plethora of record evidence—the analyses that Entergy
submitted to the LDEQ—to support the BART determination. There was also
considerable record evidence—the analyses that the EPA submitted—
undermining that determination. The LDEQ stated that (1) it considered both
of these analyses when it made its BART determination and (2) it “reviewed
and weighed the five factors carefully.” Although the LDEQ could have offered
a more thorough explanation of its reasoning, its assurance that it weighed the
five factors carefully indicates that its decision “rest[ed] ‘on a consideration of
the relevant factors’” 104 and that the agency did not “entirely fail[] to consider
102 We note that this would be a much easier case if the LDEQ had (a) explained the
weight it afforded to any of the five factors, (b) expressly compared the costs of each option,
or (c) simply reproduced Entergy’s charts and numbers in the revised SIP.
103 10 Ring Precision, Inc., 722 F.3d at 723 (quoting Motor Vehicle Mfrs. Ass’n, 463
U.S. at 43).
104 Michigan, 135 S. Ct. at 2706 (quoting Motor Vehicle Mfrs. Ass’n, 463 U.S. at 43).
42
Case: 18-60116 Document: 00515144673 Page: 43 Date Filed: 10/03/2019
No. 18-60116
an important aspect of the problem.” 105 The EPA’s approval of that
determination was not arbitrary and capricious. 106
2. Louisiana’s Long-Term Strategy and Reasonable Progress
Goals
The second part of Environmental Petitioners’ challenge is to the EPA’s
approval of Louisiana’s long-term strategy and reasonable-progress goals. The
Clean Air Act requires states to adopt long-term strategies “for making
reasonable progress” toward the national visibility goal. 107 Louisiana’s 2008
SIP included reasonable progress goals and a long-term strategy, but the 2017
SIP revisions did not.
Environmental Petitioners contend that the 2017 SIP revisions did not
comply with the Act because they did not address Louisiana’s long-term
105 Motor Vehicle Mfrs. Ass’n, 463 U.S. at 43.
106 Environmental Petitioners cite no authority, and we are aware of no authority,
demanding that Louisiana prefer the EPA’s analyses to those provided by industry. Indeed,
in a highly technical area such as cost-effectiveness of various control options, a private firm
may well provide more accurate data inputs regarding a specific proposal at the specific plant
it operates. To take one example, the EPA and Entergy disagreed about the cost-effectiveness
of low-sulfur coal. Entergy calculated it to cost $597 per ton of pollutant removed while the
EPA calculated $2,957 per ton. That discrepancy was based on Entergy’s reliance on
proprietary data, its contingency value, and its consideration of costs that the EPA does not
allow in its control cost manual. See 82 Fed. Reg. at 32,298. The technical analyses used
different approaches and presented different strengths and weaknesses. Louisiana
scrutinized Entergy’s analysis and chose to embrace it. It was free to do so. A state has “wide
discretion” in formulating its SIP and “may select whatever mix of control devices it desires”
so long as national standards are met. Union Elec. Co. v. EPA, 427 U.S. 246, 250, 266 (1976).
That is why Congress tied the EPA’s hands during SIP approval: “the Administrator shall
approve such submittal as a whole if it meets all of the applicable requirements of this
chapter.” 42 U.S.C. § 7410(k)(3) (emphasis added).
107 42 U.S.C. § 7491(b)(2)(A)–(B).
43
Case: 18-60116 Document: 00515144673 Page: 44 Date Filed: 10/03/2019
No. 18-60116
strategy, set out reasonable progress goals, or impose emission controls at any
additional “reasonable progress sources,” such as the Dolet Hills powerplant.
Dolet Hills, which is the largest single source of SO2 emissions in Louisiana, is
a non-BART-eligible EGU powerplant that could be subject to emissions
controls as a reasonable progress source.
The EPA points to its actions in 2012 on this issue, including a proposed
rule and a final rule (“the 2012 Proposed Rule” and “the 2012 Final Rule”),
which included a “partial disapproval” and “partial limited approval” of
Louisiana’s 2008 SIP. 108 It insists that Louisiana’s reasonable progress goals
and long-term-strategy obligations were approved as part of the 2012 Final
Rule’s “partial limited approval.” According to the EPA, the 2012 resolution of
this issue time-bars the Environmental Petitioners challenge.
The EPA is correct. Louisiana was not required to re-do its reasonable
progress analysis or re-evaluate the reasonable progress sources because that
part of its 2008 SIP was approved and made effective by the 2012 Final Rule.
The 2012 Final Rule “finaliz[ed] a partial limited approval and a partial
disapproval of a revision to the Louisiana SIP submitted . . . on June 13,
2008.” 109
We resolve this issue based on the distinction between “partial” and
“limited” approvals. A “limited approval” is used when some parts of a SIP do
not meet the Clean Air Act’s requirements, but the EPA concludes that “the
108 77 Fed. Reg. 11,839, 11,840 (Feb. 28, 2012) (“2012 Proposed Rule”); 2012 Final
Rule, 77 Fed. Reg. at 39,425.
109 2012 Final Rule, 77 Fed. Reg. at 39,426.
44
Case: 18-60116 Document: 00515144673 Page: 45 Date Filed: 10/03/2019
No. 18-60116
submittal as a whole” has a “strengthening effect” on the SIP. 110 In a “limited
approval,” the provisions of a SIP that are “limitedly approved,” including any
provisions that do not comply with the Act, are incorporated into the SIP and
go into effect. In contrast, in a “partial approval/disapproval,” the partially
approved parts of a submittal go into effect, and the disapproved parts do not.
The 2012 Final Rule contained a “partial limited approval” and a “partial
disapproval.” This sentence from the Final Rule addresses the parts of the SIP
that were included in the “limited approval”:
EPA grants a partial limited approval of the LA RH SIP
submittal for meeting the requirements of: 51.308(d), for the core
requirements for regional haze SIPs, except for the requirements
of 51.308(d)(3); 51.308(f), for the commitment to submit
comprehensive periodic revisions of regional haze SIPs; 51.308(g),
for the commitment to submit periodic reports describing progress
towards the reasonable progress goals (RPGs); 51.308(h), for the
commitment to conduct periodic determinations of the adequacy of
the existing regional haze SIP; and 51.308(i), for coordination with
state and Federal Land Managers. 111
All provisions of the 2008 SIP referenced in this sentence were “limitedly
approved,” including the provisions after “except,” which did not meet the Act’s
requirements. Through this “partial limited approval,” the EPA concluded that
all of the provisions referenced in that sentence would take effect because the
submittal “as a whole . . . strengthen[ed] the Louisiana SIP” and the EPA did
110 Processing of State Implementation Plan Revisions at 2–3, the EPA Memorandum
from John Calcagni (July 9, 1992), https://www.epa.gov/sites/production/files/2015-
07/documents/procsip.pdf.
111 2012 Final Rule, 77 Fed. Reg. at 39,426.
45
Case: 18-60116 Document: 00515144673 Page: 46 Date Filed: 10/03/2019
No. 18-60116
not specifically “address[] [them in the] partial disapproval.” 112 Those
referenced provisions therefore became effective when the EPA issued the 2012
Final Rule.
The following sentence in the 2012 Final Rule addresses the provisions
of the SIP that were “partially disapproved”:
However, . . . the EPA is also partially disapproving the LA
RH SIP submittal because it does not include fully approvable
measures for meeting the requirements of 40 CFR 51.308(d)(3),
long-term strategy for regional haze as it relies on deficient non-
EGU BART analyses; and 51.308(e), BART requirements for
regional haze visibility impairment with respect to emissions of
visibility impairing pollutants from four non-EGUs. 113
Because the provisions referenced in this sentence were “partially
disapproved,” they did not go into effect when the EPA issued the 2012 Final
Rule. So, after the EPA issued the 2012 Final Rule, Louisiana was required to
re-do only the parts of its SIP referenced in that partial disapproval. Those
were (1) the parts of 2008 SIP’s long-term strategy that “relie[d] on deficient
non-EGU BART analyses” and (2) the parts of the 2008 SIP’s long-term
strategy that involved BART requirements “from four non-EGUs.” These were
the only two provisions that did not go into effect and needed to be remedied
by Louisiana.
Other than these two deficiencies, the 2012 Final Rule “result[ed] in
approval of all of the remaining elements of Louisiana’s [2008 Regional Haze]
112 Id. at 39,425.
113 Id. at 39,426.
46
Case: 18-60116 Document: 00515144673 Page: 47 Date Filed: 10/03/2019
No. 18-60116
SIP.” 114 That included Louisiana’s reasonable progress analysis and long-term
strategy.
Environmental Petitioners focus on one of the EPA’s responses to a
comment. In the first part of that response, the EPA elaborates on the partial
disapproval and identifies the two deficiencies that Louisiana needed to
address in its SIP revisions.
[1] Louisiana must submit and the EPA must approve a
revised SIP submittal to address . . . BART for EGUs to cure the
deficiencies in the SIP resulting from the remand of CAIR. . . . [2]
Louisiana must also submit revisions sufficient to cure the
deficiencies in the non-EGU BART determinations. 115
Environmental Petitioners rely on a different sentence in that response
to contend that Louisiana was required to re-do its evaluation of the non-BART
powerplants that qualify as reasonable progress sources. They point to this
sentence in the EPA’s response: “Louisiana must consider whether EGUs
previously covered by the CAIR, whether subject to BART or not, should be
controlled to ensure reasonable progress to meet the State’s long-term
strategies.” 116 However, in context, that sentence simply restates Louisiana’s
overall obligations under the Clean Air Act. The response reads in full as
follows:
We have evaluated the LA RH SIP submittal as a whole and
at this time we are taking final action on all elements of the LA
RH SIP submittal that were not addressed in the national Better-
114 Id. at 39,426.
115 Id. at 39,427.
116 Id.
47
Case: 18-60116 Document: 00515144673 Page: 48 Date Filed: 10/03/2019
No. 18-60116
than-BART rule. Louisiana must consider whether EGUs
previously covered by the CAIR, whether subject to BART or not,
should be controlled to ensure reasonable progress to meet the
State’s long-term strategies. However, insofar as Louisiana’s LTS
and RPGs are affected by the remand of CAIR, those issues are
addressed in the national Better-than-BART rulemaking and are
outside the scope of this action on the remainder of the LA RH SIP.
Also, the CAA expressly provides authority to the EPA to partially
approve and partially disapprove a SIP revision. 42 U.S.C.
7410(k)(3). The EPA has adopted the partial approval approach
numerous times in SIP actions across the nation over the last
twenty years. Partial approval and partial disapproval is
appropriate here because the EPA has determined that a portion
of Louisiana’s RH SIP meets regional haze requirements and a
portion of it does not. . . . Therefore, Louisiana must submit and
the EPA must approve a revised SIP submittal to address both
NOX and SO2 BART for EGUs to cure the deficiencies in the SIP
resulting from the remand of CAIR. Louisiana may elect to rely on
the Transport Rule for NOX BART for EGUs in that
submittal. However, because Louisiana is not covered under the
Transport Rule for SO2, the State must submit source-specific
SO2 BART evaluations for the subject-to-BART EGUs in
Louisiana. As discussed further in our responses to several
comments below, Louisiana must also submit revisions sufficient
to cure the deficiencies in the non-EGU BART determinations. 117
This context, including the lack of an express directive to “submit” a new
long-term strategy or to impose controls at additional reasonable progress
sources, confirms that the 2012 Final Rule approved Louisiana’s reasonable
progress goals and selection of reasonable progress sources. Additionally, the
117 2012 Final Rule, 77 Fed. Reg. at 39,427 (emphasis added).
48
Case: 18-60116 Document: 00515144673 Page: 49 Date Filed: 10/03/2019
No. 18-60116
2012 Final Rule does not mention the single source Louisiana identified as a
reasonable progress source, Big Cajun II. In three separate parts of the 2012
Final Rule, the EPA identifies the partially disapproved parts and directs
Louisiana to address only those provisions. 118 If the EPA had intended to
disapprove Louisiana’s reasonable progress goals, long-term strategy, or
identification of reasonable progress sources in this response, it would have
included those provisions of the 2008 SIP in the “partial disapproval” sentence.
Because Louisiana’s reasonable progress goals, long-term strategy, and
evaluation of reasonable progress sources were included in the 2012 Final
Rule’s “partial limited approval” of Louisiana’s 2008 SIP, we reject
Environmental Petitioners’ challenge on this issue.
ii. Notice and Comment
Environmental Petitioners alternatively claim that the EPA violated the
APA’s notice and comment requirements.
Under the APA, “notice of proposed rulemaking shall be published in the
Federal Register,” and “shall be accompanied by a statement of its basis and
purpose. . . . The statement of basis and purpose shall include a summary of .
. . the major legal interpretations and policy considerations underlying the
proposed rule.” 119
Environmental Petitioners contend that the EPA did not include its
reasoning addressing Louisiana’s reasonable progress analysis in the July
118 Id. at 39,426–39,427, 39,435.
119 42 U.S.C. § 7607(d)(3).
49
Case: 18-60116 Document: 00515144673 Page: 50 Date Filed: 10/03/2019
No. 18-60116
2017 Proposed Rule. They assert that the EPA’s inclusion of additional
reasoning on that issue in the December 2017 Final Rule amounted to a “major
legal interpretation.”
In the 2017 Proposed Rule, the EPA did not discuss Louisiana’s
reasonable progress analysis, long-term strategy, or evaluation of reasonable
progress sources. Environmental Petitioners commented on the 2017 Proposed
Rule’s failure to address Louisiana’s reasonable progress analysis and long-
term strategy, and the EPA responded to that comment in the 2017 Final Rule.
In its response to Environmental Petitioners’ comments on the 2017 Final
Rule, the EPA directed Environmental Petitioners to the 2012 Final Rule. The
agency explained that it had already approved Louisiana’s reasonable progress
analysis, long-term strategy, and evaluation of reasonable progress sources. 120
We conclude that this explanation is not a major legal interpretation or
policy consideration underlying the 2017 Proposed Rule that required
additional notice and comment. Rather, it is merely a response to a comment
that pointed the Environmental Petitioners to the agency’s earlier resolution
of the issue in the 2012 Final Rule.
* * *
In short, the EPA’s 2012 Final Rule, which “limitedly approved” most of
Louisiana’s 2008 SIP, did not require Louisiana to re-do its reasonable
progress goals or its long-term strategy in its 2017 SIP revisions. Neither did
120 See 82 Fed. Reg. at 60,539–60,540.
50
Case: 18-60116 Document: 00515144673 Page: 51 Date Filed: 10/03/2019
No. 18-60116
the EPA’s response to Environmental Petitioners’ 2017 comment “reopen” the
issue 121 or violate the APA’s notice and comment provision.
Environmental Petitioners’ petition is denied.
B. The Industry Petitioners’ Petition
In contrast to Environmental Petitioners’ challenge to Louisiana’s
selection of emission controls to satisfy BART (step three of the BART
analysis), Industry Petitioners challenge Louisiana’s determinations that
Nelson and Brame are “subject to BART” at all (step two of the BART analysis).
Industry Petitioners object to the EPA’s approval of Louisiana’s “subject to
BART” determinations, which relied on the “CALPUFF” model. They also
challenge the EPA’s reliance on its own “CAMx” modeling (on which Louisiana
did not rely) to approve Louisiana’s SIP.
The Clean Air Act states that a source is “subject to BART” if it “emits
any air pollutant which may reasonably be anticipated to cause or contribute
to any impairment of visibility” in a Class I area. 122 Under the BART
guidelines, a 1.0 deciview change from an individual source “causes” visibility
impairment, whereas a 0.5 deciview change from an individual source
“contributes” to visibility impairment. 123 Under some circumstances, states
121 See West Virginia v. EPA, 362 F.3d 861, 872 (D.C. Cir. 2004) (“‘[W]hether an agency
has in fact reopened an issue, explicitly or implicitly,’ depends on the ‘entire context of the
rulemaking including all relevant proposals and reactions of the agency.’” (citation omitted)).
122 42 U.S.C. § 7491(b)(2)(A).
123 BART Rule, 70 Fed. Reg. at 39,120.
51
Case: 18-60116 Document: 00515144673 Page: 52 Date Filed: 10/03/2019
No. 18-60116
may set a lower threshold for sources that “contribute” to visibility
impairment. 124
In the BART guidelines, the EPA adopted CALPUFF as the preferred
model for making “subject to BART” determinations. 125 The EPA recently
removed CALPUFF as a “preferred” model for other air-quality modeling
applications under the Clean Air Act, but retained CALPUFF as a preferred
model for making “subject-to-BART” determinations. 126
The LDEQ reviewed CALPUFF and CAMx modeling results that
Entergy, Cleco, and the EPA submitted to the department. In the LDEQ’s
February and June 2017 SIP revisions, it determined that Nelson and two
units at Brame were “subject to BART.” Louisiana relied on CALPUFF
modeling to determine that Nelson and Brame exceeded the 0.5 deciview
threshold. The LDEQ acknowledged, but did not rely on, the CAMx modeling
that the parties submitted. The LDEQ stated that it did “not have the expertise
with which to review [the CAMx] model runs.” And, in its amicus brief, the
LDEQ states that it still does not have the technical expertise to review CAMx
modeling.
124 Id. at 39,120–39,121.
125 BART Guidelines, 40 C.F.R. part 51, App. Y, III.A.3. See 69 Fed. Reg. 25,184, 25,194
(May 5, 2004) (“We are proposing that a CALPUFF assessment of an individual source be
used as the preferred approach for determining whether a BART-eligible source may be
exempt from BART.”).
126 82 Fed. Reg. 5,182, 5,196 (Jan. 17, 2017) (“[T]his final action does not affect the
EPA’s recommendation that states use CALPUFF to determine the applicability and level of
best available retrofit technology in regional haze implementation plans.”).
52
Case: 18-60116 Document: 00515144673 Page: 53 Date Filed: 10/03/2019
No. 18-60116
In a May 2017 proposed rule, the EPA proposed to approve the “subject
to BART” determination for Brame. 127 In the July 2017 Proposed Rule, it
proposed to approve Louisiana’s “subject to BART” and BART
determinations. 128 Industry Petitioners submitted comments on those
proposed rules (1) objecting to the LDEQ’s reliance on CALPUFF and (2)
criticizing the EPA’s reliance on its own CAMx modeling. In the instant 2017
Final Rule, the EPA responded to those comments and approved the LDEQ’s
determination that Nelson and Brame have visibility effects greater than 0.5
deciviews at Class I areas and are therefore “subject to BART.” 129
Here, Industry Petitioners contend that the EPA’s approval of
Louisiana’s “subject to BART” determinations at Nelson and Brame was
arbitrary and capricious. The EPA and Environmental Respondents respond
that we lack jurisdiction to consider Industry Petitioners’ challenge. They also
insist that (1) Louisiana permissibly relied on the CALPUFF model in making
its BART determination and (2) the EPA was not arbitrary and capricious in
considering its own CAMx modeling.
As we shall explain in greater detail, we conclude that (1) we have
jurisdiction and (2) the EPA’s approval of Louisiana’s reliance on the
CALPUFF model was not arbitrary and capricious.
127 82 Fed. Reg. 22,936, 22,941 (May 19, 2017).
128 82 Fed. Reg. 32,294, 32,296 (July 13, 2017).
129 Final Rule, 82 Fed. Reg. at 60,520; id. at 60,525–60,526 (response to comments on
modeling).
53
Case: 18-60116 Document: 00515144673 Page: 54 Date Filed: 10/03/2019
No. 18-60116
1. Jurisdiction
a. Environmental Respondents’ Jurisdictional Arguments
Environmental Respondents contend that parts of Industry Petitioners’
challenge address the 2005 BART guidelines rather than the 2017 Final Rule.
Environmental Respondents assert that this court lacks jurisdiction over the
parts of the Industry Petitioners’ claim that address CALPUFF’s use of 24-
hour actual emissions and natural background conditions. According to
Environmental Respondents, these are complaints about the policies
underlying the 2005 BART guidelines, policies to which the EPA responded
when it promulgated those guidelines. According to Environmental
Respondents, these concerns should have been raised (1) in the D.C. Circuit
and (2) within 60 days after those guidelines were published in the Federal
Register. 130
Environmental Respondents liken this case to American Road &
Transportation Builders Ass’n v. EPA, 705 F.3d 453 (D.C. Cir. 2013), in which
the D.C. Circuit considered a challenge to the EPA’s approval of a state’s SIP
in 2011. In that case, the EPA’s approval of a SIP relied on a rule that the
agency had issued in 1994. 131 The petitioner did not claim that the state had
misapplied the EPA’s rule; rather, it contended that the policies announced in
the 1994 rule were unlawful. 132 Because that challenge was to the 1994 rule—
rather than to the 2011 SIP approval—the D.C. Circuit concluded that it lacked
130 42 U.S.C. § 7607(b)(1).
131 705 F.3d at 454–55.
132 Id. at 456.
54
Case: 18-60116 Document: 00515144673 Page: 55 Date Filed: 10/03/2019
No. 18-60116
jurisdiction because the challenge was time-barred under the Clean Air Act’s
60-day period for filing petitions for review. 133
Environmental Respondents maintain that Industry Petitioners’ claims
objecting to CALPUFF’s assumptions are similarly time-barred. They note
that Industry Petitioners have an alternative route to recourse under the
Clean Air Act’s provision that provides for reconsideration of a rule if a
petitioner can demonstrate that it was impracticable to raise an issue during
the public comment period. 134
The jurisdictional issue here is similar to the one that the D.C. Circuit
considered in American Road & Transportation Builders. That court held that
the petitioner’s challenge was to the underlying the EPA rule rather than to
the EPA’s later approval of a SIP based on the standards set out in that
underlying rule. The D.C. Circuit dismissed that challenge to a SIP approval
because it was “duplicative of arguments the agency had already rejected”
many years earlier. 135 The court explained that “[t]here would be no pressure
to challenge regulations within the 60-day period after their promulgation if
any petitioner could simply wait to test the substance of those regulations once
the EPA applies them, for example, in an approval of a state SIP revision—as
[the petitioner] has attempted to do here.” 136
133 Id. at 458.
134 See 42 U.S.C. § 7607(d)(7)(B).
135 Am. Road & Transportation Builders Ass’n, 705 F.3d at 456.
136 Id. at 458.
55
Case: 18-60116 Document: 00515144673 Page: 56 Date Filed: 10/03/2019
No. 18-60116
Similarly, in the instant petition for review, Industry Petitioners
challenge the CALPUFF model based on its use of 24-hour actual emissions
and natural background conditions. The EPA expressly rejected these
arguments when it adopted CALPUFF as the preferred model at the time in
2005 that it issued the BART guidelines. 137
But, unlike American Road & Transportation Builders, the BART
guidelines at issue here recommend—but do not require—that states use
CALPUFF. 138 Louisiana could have picked a different model. Because Industry
Petitioners challenge Louisiana’s choice to use CALPUFF modeling instead of
a different model, Industry Petitioners’ challenge is properly viewed as a
challenge to the application of the BART guidelines rather than a challenge to
the those guidelines themselves. Louisiana had the option to pick a different
model, so the parts of Industry Petitioners’ challenge addressing CALPUFF’s
flaws are not barred by the Clean Air Act’s 60-day statute of limitations.
b. The EPA’s Jurisdictional Arguments
The EPA maintains that, after actively participating in Louisiana’s SIP-
development process, the Industry Petitioners should have sought judicial
137 BART Guidelines, 40 C.F.R. part 51, App. Y III.A.3 (“We recommend that States
use the 24 hour average actual emission rate from the highest emitting day of the
meteorological period modeled.”).
138 BART Rule, 70 Fed. Reg. at 39,123 (“The use of other models and techniques to
estimate if a source causes or contributes to visibility impairment may be considered by the
State, and the BART guidelines preserve a State’s ability to use other models.”).
56
Case: 18-60116 Document: 00515144673 Page: 57 Date Filed: 10/03/2019
No. 18-60116
review of Louisiana’s modeling decisions in state court based on a Louisiana
statute that allows for review of state rules and regulations. 139
This argument is unavailing. Industry Petitioners challenge the EPA’s
approval of Louisiana’s SIP, not the underlying state action. Although
Industry Petitioners might also have been able to challenge Louisiana’s
process for developing its SIP in state court, the Clean Air Act expressly
provides Industry Petitioners the opportunity to petition this court for review
of the EPA’s actions. 140
2. CALPUFF Modeling
“A reviewing court must be ‘most deferential’ to the agency where, as
here, its decision is based upon its evaluation of complex scientific data within
its technical expertise.” 141 And “because ‘a model is meant to simplify reality
in order to make it tractable,’ it is no criticism of a model ‘that it does not fit
every application perfectly.’” 142
139See La. R.S. § 49:963(a)(1)).
14042 U.S.C. § 7607(b)(1).
141 BCCA Appeal Grp. v. EPA, 355 F.3d 817, 824 (5th Cir. 2003) (quoting Baltimore
Gas & Elec. Co. v. NRDC, 462 U.S. 87, 103 (1983)). The D.C. Circuit, setting out a similar
standard, has described this as “extreme deference”:
[W]e will “give an extreme degree of deference to the agency when it is
evaluating scientific data within its technical expertise.” . . . Furthermore, “we
must defer to the agency’s decision on how to balance the cost and complexity
of a more elaborate model against the oversimplification of a simpler model.”
. . . We will “reverse only if the model is so oversimplified that the agency’s
conclusions from it are unreasonable.”
West Virginia, 362 F.3d at 871 (citations omitted).
142 Ass’n of Battery Recyclers, Inc. v. EPA, 308 F.3d 1047, 1063 (D.C. Cir. 2000) (citation
omitted).
57
Case: 18-60116 Document: 00515144673 Page: 58 Date Filed: 10/03/2019
No. 18-60116
Despite the significant deference we afford to decisions involving
analysis of scientific data within an agency’s technical expertise, Industry
Petitioners maintain that the EPA’s approval of Louisiana’s CALPUFF
modeling was arbitrary and capricious. They first point to CALPUFF’s
shortcomings as a model, including that it (1) is unreliable at predicting
visibility impacts at locations farther than 300 km from an emission source, (2)
tends to overestimate visibility impacts, (3) uses “overly simplistic”
assumptions, and (4) has a high margin of error. They next contend that the
EPA’s reliance on its own CAMx modeling was improper because the Clean Air
Act does not give the EPA authority to cure a deficient SIP. Finally, even
assuming that the EPA could rely on its own CAMx modeling, Industry
Petitioners take issue with technical flaws in the EPA’s CAMx modeling,
including that model’s: (1) use of “24-hour maximum emission rates” instead
of “actual emission rates,” (2) reliance on “absolute maximum modeled
concentrations” instead of “relative response factors,” and (3) deficient
performance on a “model performance evaluation.”
i. Distance Limitations
Industry Petitioners contend that the CALPUFF model used at Nelson
and Brame exceeded that model’s distance limitations and was inconsistent
with the EPA’s prior determination that CALPUFF is unreliable when
assessing visibility impacts from sources more than 300 km from a Class I
58
Case: 18-60116 Document: 00515144673 Page: 59 Date Filed: 10/03/2019
No. 18-60116
area. 143 Nelson is 425 km from Breton and 460 km from Caney Creek; Brame
is 422 km from Breton and 352 km from Caney Creek.
Industry Petitioners cite a 1998 statement from the EPA’s Interagency
Workgroup on Air Quality Modeling (“IWAQM”) that “there are serious
conceptual concerns with the use of puff dispersion [models like CALPUFF] for
very long-range transport (300 km and beyond).” In a 2003 final rule, IWAQM
concluded that the “CALPUFF dispersion model . . . performed in a reasonable
manner . . . so long as the transport distance was limited to less than
300km.” 144 According to Industry Petitioners, because CALPUFF does not
perform “in a reasonable manner” at distances greater than 300 km, CALPUFF
may not be used to determine that sources at such distances may “reasonably
be anticipated” to cause or contribute to visibility impairment.
Industry Petitioners next point to several regional haze actions in which
the EPA has stated that it has concerns about using CALPUFF at distances
greater than 300 km. They rely on the Arkansas Regional Haze Federal
Implementation Plan, the New Mexico Regional Haze SIP, and the EPA’s
statement defending its reliance on CAMx for Texas’s and Oklahoma’s regional
haze SIPs. 145
143 Industry Petitioners raise the same concerns they raised in their comments on the
Proposed Rule.
144 68 Fed. Reg. 18,458 (Apr. 15, 2003).
145 Arkansas Regional Haze Federal Implementation Plan, 81 Fed. Reg. 66,332, 66,394
(Sept. 27, 2016) (“[T]here are concerns in using CALPUFF for modeling impacts at distances
much greater than 300 km from the source . . . .”); New Mexico Regional Haze SIP, 79 Fed.
Reg. 60,978, 60,983 (Oct. 9, 2014) (“CALPUFF results are less reliable at distances greater
than 300 km.”).
59
Case: 18-60116 Document: 00515144673 Page: 60 Date Filed: 10/03/2019
No. 18-60116
Industry Petitioners describe these statements and criticisms of
CALPUFF as indicative of the EPA’s failure to follow its own guidance and
past practice. They contend that the Final Rule does not provide a rational
explanation for this departure. And, although the EPA has stated that
CALPUFF is permissible at sources farther than 300 km “with some additional
considerations” such as stack height or size of emissions, Industry Petitioners
maintain that the EPA did not address these “additional considerations” at
Nelson and Brame.
The EPA responds that Industry Petitioners overstate and oversimplify
the BART guidelines and the EPA’s prior actions. The EPA insists, contrary to
Industry Petitioners’ assertions, that it has never stated that: (1) CALPUFF
should not be used to model visibility impacts more than 300 km from the
emissions source, (2) CALPUFF is unable to reliably predict visibility impacts
at distances greater than 300 km, or (3) CALPUFF is “unreliable” and does not
perform in a reasonable manner. The EPA also contends that Industry
Petitioners’ reliance on statements from IWAQM is misplaced, arguing that
IWAQM’s statements (1) set out guidelines for air quality models under the
Clean Air Act for purposes other than regional haze analyses and (2) were
issued in 1998 and 2003, before the EPA issued the BART guidelines in 2005.
The EPA explains that CALPUFF has been used at distances greater
than 300 km, citing South Dakota’s CALPUFF modeling for sources farther
than 400 km, and Nebraska’s CALPUFF modeling for sources between 300 and
60
Case: 18-60116 Document: 00515144673 Page: 61 Date Filed: 10/03/2019
No. 18-60116
600 km. 146 In the EPA’s “Modeling Response to Comments” document, it stated
that “the EPA and [Federal Land Managers] have utilized CALPUFF results
in a number of different situations when the range was between 300–450 km
or more.” 147
The EPA also points out that historically, “the use of CALPUFF was
generally acceptable at 300 km or greater for larger emissions sources with
elevated stacks.” In its Modeling Response to Comments document, the EPA
expressly stated that Nelson and Brame are “larger emissions sources with
elevated stacks.” Nelson has three BART-eligible electric generating units that
have a combined 1200 megawatts of production capacity, 148 and Brame has two
BART-eligible electric generating units with a combined output of 963
megawatts. 149 The EPA’s CALPUFF analysis estimated that both Nelson and
Brame exceeded the 1.0 deciview threshold for “causing” visibility impairment.
These high visibility numbers for Nelson and Brame align with the EPA’s
statement in the BART Rule that it anticipated “that most of these plants are
predicted to have much higher maximum impacts” than 1.0 deciviews. 150
Finally, the EPA maintains that its own CAMx modeling provided
additional support for the CALPUFF model’s results. The EPA’s CAMx
146 Final Rule, 82 Fed. Reg. at 60,525 n.39.
147 See Final Rule, 82 Fed. Reg. at 60,525 (“As discussed in the Modeling RTC
document, EPA and FLM representatives have utilized CALPUFF results in a number of
different situations when the range was between 300-450 km or more.”).
148 Final Rule, 82 Fed. Reg. at 60,522, 60,533.
149 82 Fed. Reg. at 22,944.
150 BART Rule, 70 Fed. Reg. at 39,123.
61
Case: 18-60116 Document: 00515144673 Page: 62 Date Filed: 10/03/2019
No. 18-60116
modeling showed visibility impairments at Caney Creek from Nelson as high
as 2.22 deciviews and from Brame as high as 2.83 deciviews. 151 According to
the EPA, such large visibility-impact numbers provide additional support for
the CALPUFF model’s results.
We conclude that, although the EPA has “a higher confidence level” in
CALPUFF modeling at distances under than 300 km, the agency did not act
arbitrarily and capriciously by using that model at Nelson and Brame. Even
though Nelson and Brame are farther than 300 km from the protected areas
at issue here, the use of CALPUFF modeling aligns with the EPA’s acceptance
of CALPUFF at longer distances for powerplants with larger emissions sources
with elevated stacks like Nelson and Brame. 152
ii. The EPA’s January 2017 Action
Industry Petitioners focus on the EPA’s January 2017 action, which
removed CALPUFF as the preferred model for long-range air quality
assessments. For those assessments, the EPA concluded that photochemical
151 Final Rule, 82 Fed. Reg. at 60,526 (“Entergy Nelson has a maximum modeled
impact of 2.22 dv at Caney Creek, with 31 days out of the 365 days modeled exceeding 0.5 dv,
and 9 days exceeding 1.0 dv. Similarly, Cleco Brame has a maximum modeled impact of 2.833
dv at Caney Creek, with 30 days out of a maximum 365 days modeled exceeding 0.5 dv and
10 days exceeding 1.0 dv.”).
152 Our opinion should not be read to countenance the reliability or accuracy of the
CALPUFF model in general. Although the EPA maintains that it recommends that states
continue to use CALPUFF modeling for “subject to BART” determinations based on
“consistency across and within states in the regional haze program” and because most of the
modeling for this planning period has already been completed, there is reason to question the
EPA’s recommendation that states continue using a model that it appears to know is not
reliable when used at distances such as those involved here.
62
Case: 18-60116 Document: 00515144673 Page: 63 Date Filed: 10/03/2019
No. 18-60116
grid models like CAMx “are generally most appropriate . . . because they
provide a spatially and temporally dynamic realistic chemical and physical
environment for plume growth and chemical transformation.” 153 Industry
Petitioners maintain that the EPA’s 2017 action confirms that states should
rely on CAMx rather than CALPUFF when making BART determinations,
because it more realistically simulates the processes involved in haze
formation.
The EPA’s January 2017 action, however, does not support Industry
Petitioners’ arguments. The EPA removed CALPUFF as a preferred model for
other types of Clean Air Act analyses, but the agency expressly confirmed that
CALPUFF remains a recommended option for BART determinations. “[T]his
final action does not affect the EPA’s recommendation that states use
CALPUFF to determine the applicability and level of best available retrofit
technology in regional haze implementation plans.” 154
But the EPA retained CALPUFF as a preferred model for BART
determinations because of concerns about consistency between the states’
BART determinations:
The proposed changes to the Guideline do not affect the EPA’s
recommendation in the 2005 BART Guidelines to use CALPUFF
in the BART determination process. Given that the overwhelming
majority of BART determinations have been made using
CALPUFF, we consider it appropriate for states (or the EPA) to
continue to use this application for the remaining assessments
153 80 Fed. Reg. 45,340, 45,349 (July 29, 2015), finalized at 82 Fed. Reg. 5,182 (Jan.
17, 2017).
154 82 Fed. Reg. at 5,196.
63
Case: 18-60116 Document: 00515144673 Page: 64 Date Filed: 10/03/2019
No. 18-60116
under the current Guideline with approved protocols. This
approach assures consistency across and within states in the
regional haze program. In addition, in many instances, the
modeling of visibility impacts has already been completed even
though the BART determination process is not yet done. Allowing
states to continue to rely on CALPUFF avoids additional time and
expense in developing a new assessment of visibility impacts for a
SIP initially due in 2007. 155
Because CALPUFF remains the BART guidelines’ recommended model,
the January 2017 action does not undermine the use of CALPUFF for BART
determinations. Rather, the EPA’s explanations about (1) consistency between
states, (2) that most of the modeling for this planning period has already been
completed, and (3) avoiding the “additional time and expense in developing a
new assessment” support the use of that model here.
iii. Overestimation Bias
Industry Petitioners next object to several technical assumptions
underlying the CALPUFF model. The most significant of these is CALPUFF’s
simulation of the transformation of SO2 and NOx emissions into visibility-
impairing sulfates and nitrates. CALPUFF’s simplistic handling of chemical
processes can result in “a systematic bias in the estimated concentrations and
visibility impacts,” in other words, an “overestimation bias.” When compared
with “real-world measurements,” CALPUFF overpredicts the contribution of
nitrates to visibility impairment. According to Industry Petitioners, a
comparison between the “real-world” impairment data and the CALPUFF
155 80 Fed. Reg. at 45,350.
64
Case: 18-60116 Document: 00515144673 Page: 65 Date Filed: 10/03/2019
No. 18-60116
modeling results shows that the CALPUFF modeling here overestimated
nitrate contributions “by nearly 500%.”
Industry Petitioners cite 2005 guidelines from the Central Regional Air
Planning Association (“CENRAP”) warning that CALPUFF’s overestimation
bias may lead to “unwarranted and excessive control of emissions from some
sources.” 156 They maintain that the EPA has not articulated a satisfactory
explanation for ignoring the technical shortcomings in the CALPUFF model. 157
In defending its use of CALPUFF here, the EPA emphasizes the policy
considerations underlying the BART guidelines. When the EPA proposed the
BART guidelines that recommended CALPUFF as the preferred model, some
commenters objected on the same grounds raised here: CALPUFF is unreliable
at long distances, relies on simplified assumptions, and overpredicts visibility
impairment. 158 The EPA responded to those criticisms by describing
CALPUFF as a “conservative” screening model and stating that “conservatism
is needed because the purpose of the [subject to BART] test [is] solely to
determine if a closer look is needed.” 159 At this “screening” step, the EPA
emphasized that models should capture the “maximum” or “worst case”
potential impacts of a source. 160
156 See Alpine Geophysics, LLC, CENRAP BART Modeling Guidelines, at 3–8 (Dec.
15, 2005).
15710 Ring Precision, Inc., 722 F.3d at 723.
158BART Rule, 70 Fed. Reg. at 39,126.
159 Id. at 39,123 (“[the EPA] understand[s] the concerns of commenters that the
chemistry modules of the CALPUFF model are less advanced than some of the more recent
atmospheric chemistry simulations.”).
160 Id. at 39,123, 39,126.
65
Case: 18-60116 Document: 00515144673 Page: 66 Date Filed: 10/03/2019
No. 18-60116
We conclude that the BART guidelines accounted for this
“overestimation” bias. Those guidelines explain the assumptions underlying
the CALPUFF model, including the use of “the 98th percentile visibility
impairment rather than the highest daily impact,” which was meant to
“exclude roughly 7 days per year from consideration” and “minimiz[e] the
likelihood that the highest modeled visibility impacts might be caused by
unusual meteorology or conservative assumptions in the model.” As the EPA
explained in the analysis it submitted to the LDEQ, it “made the decision to
consider the less conservative 98th percentile primarily because the chemistry
modules in the CALPUFF model are simplified and likely to provide
conservative (higher) results for peak impacts.”
iv. Margin of Error
Entergy submitted analyses to the EPA that purported to demonstrate
that Nelson’s CALPUFF-predicted visibility impacts at Caney Creek and
Breton fell within CALPUFF’s margin of error. According to Industry
Petitioners, its margin of error at Caney Creek is 1.38 deciviews, and its
margin of error at Breton is 1.25 deciviews. So, based on Entergy’s CALPUFF
modeling analysis that calculated Nelson’s baseline visibility impact at 0.703
deciviews at Caney Creek and 0.77 deciviews at Breton, the Entergy-calculated
margin of error for CALPUFF would put Nelson below the 0.5 deciview “subject
to BART” threshold.
Entergy submitted comments on this issue, relying on a Ninth Circuit
decision, National Parks Conservation Ass’n v. EPA, 788 F.3d 1134, 1146–47
(9th Cir. 2015), in which the court concluded that the EPA did not
66
Case: 18-60116 Document: 00515144673 Page: 67 Date Filed: 10/03/2019
No. 18-60116
“meaningfully address” a comment about model-predicted visibility
improvements that were within a model’s margin of error. 161 There, the federal
implementation plan for Montana sought to implement a control that
CALPUFF predicted would improve visibility by only 0.085 deciview. 162 A
commenter maintained that “an improvement of 0.085 deciview [was] ‘beyond
the CALPUFF model’s ability to predict with any confidence.’” 163 The EPA
responded to that comment, but the court concluded that the response did “not
meaningfully address [the] comment” because it did not “also suffice as a
reasoned response regarding how CALPUFF could be relied upon to predict an
improvement of as little as 0.085 deciviews when PPL offered reasons to think
that doing so was outside the model’s capabilities.” 164
Entergy submitted its “margin of error” comments, which compared that
case to this one. The EPA addressed Entergy’s comments in its November 2017
Modeling Response to Comments document.
Industry Petitioners contend that the EPA’s response to Entergy’s
comments did not adequately respond to the margin of error issue and
161 722 F.3d at 1146–47.
162 Id. Notably, this small number is significantly smaller than the baseline impact
values at issue here, which are above 1 deciview for both facilities.
163 Nat’l Parks Conservation Ass’n, 788 F.3d at 1146–47.
164 Id. One judge on that panel separately concurred to emphasize the procedural focus
of the court’s holding. “I write separately to underline my understanding that . . . we are not
impugning the EPA’s use of the CALPUFF model generally. Instead we are requiring a
sufficiently reasoned response to a particular comment regarding CALPUFF’s usefulness in
these specific circumstances.” Id. at 1149 (Berzon, J., concurring).
67
Case: 18-60116 Document: 00515144673 Page: 68 Date Filed: 10/03/2019
No. 18-60116
therefore failed to fulfill the agency’s procedural obligation to consider
significant comments made by the public.
The EPA responded to this “margin of error” comment as follows:
The commenter mischaracterizes the Ninth Circuit decision
regarding the “margin of error” of the model. The commenter
suggests that the Court agreed that the anticipated visibility
benefits in that case were within the margin of error of the model.
This is incorrect. The Ninth Circuit decision did not rule on any
specific issue related to CALPUFF. Rather, the court ruled on a
procedural error that the EPA did not respond to the comment
received regarding the CALPUFF margin of error in its
rulemaking as required under the law. Here and elsewhere in our
previous response to comments we address a very similar comment
with respect to CALPUFF modeling for Arkansas sources, as well
as the commenter’s analysis claiming to estimate the “margin of
error.”
We responded to comments concerning a very similar “margin of
error” analysis in our response to comments and final action for
Regional Haze in Arkansas. The Trinity analysis discussed in the
comment above purports to calculate a “margin of error” of the
CALPUFF modeling for Entergy Nelson. In general, the
commenter’s analysis adds CALPUFF model results for a specific
source or sources with CAMx model results and compares this
value to visibility conditions derived from monitored data at each
Class I area. This analysis is flawed for many reasons as discussed
in detail in our Arkansas RTC document that discusses a similar
analysis performed for Entergy Lake Catherine and fails to
provide any assessment of the ability of the CALPUFF model to
evaluate visibility impacts or the degree of visibility improvement
that may be expected from available control technology to inform
BART and reasonable progress evaluations. Whether or not the
modeled visibility impacts or benefits lie below this calculated
“margin of error” is immaterial to any assessment of whether or
68
Case: 18-60116 Document: 00515144673 Page: 69 Date Filed: 10/03/2019
No. 18-60116
not the visibility impairment or benefits from controls can
reasonably be anticipated to occur. We note that the commenter
did not provide any spreadsheets or detailed calculations to
support this analysis, therefore we were unable to fully review and
respond to the individual calculations or specific methodology
underlying the presented values in the commenter’s summary
report and comments.
As discussed elsewhere in this document, we are confident that
CALPUFF distinguishes, comparatively, the relative contributions
from sources such that the differences in source configurations,
sizes, emission rates, and visibility impacts are well-reflected in
the model results. We agree with LDEQ that the CALPUFF model
followed the reviewed protocol is an appropriate tool to evaluate
visibility impacts and benefits to inform a BART determination.
Furthermore, our CAMx modeling of coal-fired sources included in
the LA RH SIP (see Appendix F) further supports the conclusion
that the Entergy Nelson and Cleco Brame sources are subject to
BART.
This explanation is fulsome. In it, the EPA (1) cited its responses on the
same issue in the Arkansas and Texas federal implementation plans (which
are in the administrative record here), (2) disagreed with the commenter’s
interpretation of National Parks Conservation Ass’n, (3) criticized the
commenter’s “calculated margin of error” methodology, (4) disagreed with the
commenter’s method of combining results from two different models, and (5)
noted the commenter’s failure to provide the underlying data. Although
Industry Petitioners might not agree with the conclusion that the EPA
reached, the EPA “meaningfully respond[ed]” to the substance of the comment.
* * *
69
Case: 18-60116 Document: 00515144673 Page: 70 Date Filed: 10/03/2019
No. 18-60116
In short, Industry Petitioners contend that CALPUFF uses
oversimplified and unrealistic assumptions. The EPA agrees in part, but
insists that those assumptions are based on reasoned policy decisions.
Given the conflicting technical contentions, we defer to the EPA’s
approval of Louisiana’s reliance on the CALPUFF model. The EPA’s selection
of modeling methods to measure visibility impacts is exactly the type of
decision for which “significant deference” is appropriate. 165 Although
CALPUFF has documented flaws, (1) the BART guidelines continue to
recommend that model for these determinations, 166 (2) the EPA has recently
directed states to use that model, 167 and (3) it “has been used for almost every
single-source BART analysis in the country.”
We defer to the EPA’s approval of Louisiana’s reliance on the CALPUFF
model. Industry Petitioners have not carried their “considerable burden” to
overcome the “presumption of regularity” that we afford to “the EPA’s choice
of analytical methodology.” 168 Because a model “is meant to simplify reality in
order to make it tractable,” and need not “fit every application perfectly,” 169 we
165 See Appalachian Power Co. v. EPA, 135 F.3d 791, 802 (D.C. Cir. 1998) (“Statistical
analysis is perhaps the prime example of those areas of technical wilderness into which
judicial expeditions are best limited to ascertaining the lay of the land.”).
166 BART Guidelines, 40 C.F.R. part 51, App. Y, III.A.3.
167 82 Fed. Reg. 5,182, 5,196 (Jan. 17, 2017) (“[T]his final action does not affect the
EPA’s recommendation that states use CALPUFF to determine the applicability and level of
best available retrofit technology in regional haze implementation plans.”).
168 BCCA Appeal Grp., 355 F.3d at 832
169 Ass’n of Battery Recyclers, 308 F.3d at 1063.
70
Case: 18-60116 Document: 00515144673 Page: 71 Date Filed: 10/03/2019
No. 18-60116
hold that the EPA’s reliance on the CALPUFF model was not arbitrary and
capricious.
Moreover, the Clean Air Act’s structure gives states the primary
responsibility for implementing the federal standards. The EPA’s role is to
review the states’ plans for compliance with the Act’s requirements. Within
those respective roles, “the States may submit implementation plans more
stringent than federal law requires,” and the EPA “must approve such plans if
they meet the minimum requirements” of the Clean Air Act. 170 At this “subject
to BART” stage—the purpose of which is to determine which of the BART-
eligible sources warrant a “closer look” 171—Louisiana’s reliance on a modeling
method that tends to be overinclusive would result in a SIP that is more
stringent than the federal requirements. The EPA’s approval of such a SIP
would be well within the agency’s role to ensure the SIP “meet[s] the minimum
requirements” of the Clean Air Act. 172
3. Challenge to the EPA’s CAMx Modeling
We have concluded that the EPA’s reliance on the CALPUFF model was
not arbitrary and capricious, so we need not address Industry Petitioners’
contentions about the EPA’s use of CAMx modeling.
170 Union Elec. Co., 427 U.S. at 265; see 42 U.S.C. § 7410(k)(3) (“[the EPA] shall
approve [a SIP] submittal as a whole if it meets all of the applicable requirements of this
chapter.”).
171 BART Rule, 70 Fed. Reg. at 39,127.
172 Union Elec. Co., 427 U.S. at 265.
71
Case: 18-60116 Document: 00515144673 Page: 72 Date Filed: 10/03/2019
No. 18-60116
IV. CONCLUSION
For the foregoing reasons, the petitions for review are DENIED.
72