American Public Gas Association v. United States Department of Energy

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Argued January 20, 2023 Decided July 7, 2023
No. 22-1107
AMERICAN P UBLIC GAS ASSOCIATION ,
P ETITIONER
v.
UNITED S TATES D EPARTMENT OF ENERGY,
R ESPONDENT
AMERICAN G AS ASSOCIATION,
INTERVENOR
Consolidated with 22-1111, 22-1117
On Petitions for Review of a
Final Action of the Department of Energy
Barton Day and Jason Neal, argued the causes for
petitioners. With them on the joint briefs were John P. Gregg
and Christopher J. Wright.
John Starcher, Attorney, U.S. Department of Justice,
argued the cause for respondent. With him on the brief were

-- 1 of 29 --

2
Brian M. Boynton, Principal Deputy Assistant Attorney
General, and Michael S. Raab, Attorney.
Michelle Wu, Ian Fein, Aaron Colangelo, Timothy D.
Ballo, Letitia James, Attorney General, Office of the Attorney
General for the State of New York, Brian Lusignan, Assistant
Solicitor General, Lisa S. Kwong, Assistant Attorney General,
Kwame Raoul, Attorney General, Office of the Attorney
General for the State of Illinois, Elizabeth Dubats, Assistant
Attorney General, Aaron M. Frey, Attorney General, Office of
the Attorney General for the State of Maine, Katherine E.
Tierney, Assistant Attorney General, Brian E. Frosh, Attorney
General, Office of the Attorney General for the State of
Maryland, at the time the brief was filed, John B. Howard, Jr.,
Special Assistant Attorney General, Maura Healey, Attorney
General, Office of the Attorney General for the
Commonwealth of Massachusetts, at the time the brief was
filed, Turner Smith, Assistant Attorney General, Keith Ellison,
Attorney General, Office of the Attorney General for the State
of Minnesota, Peter N. Surdo, Special Assistant Attorney
General, Aaron Ford, Attorney General, Office of the Attorney
General for the State of Nevada, Heidi Parry Stern, Solicitor
General, Matthew J. Platkin, Attorney General, Office of the
Attorney General for the State of New Jersey, Paul Youchak,
Deputy Attorney General, Ellen F. Rosenblum, Attorney
General, Office of the Attorney General for the State of
Oregon, Paul Garrahan, Attorney in Charge, Steve Novick,
Special Assistant Attorney General, Karl A. Racine, Attorney
General, Office of the Attorney General for the District of
Columbia, at the time the brief was filed, Caroline S. Van Zile,
Solicitor General, and Christopher Gene King were on the
brief for amici curiae State, Municipal, and Non-Profit in
support of respondent. Barbara D. Underwood, Solicitor
General, Office of the Attorney General for the State of New
York, entered an appearance.

-- 2 of 29 --

3
Before: H ENDERSON and WILKINS , Circuit Judges, and
R OGERS , Senior Circuit Judge.
Opinion for the Court filed by Circuit Judge WILKINS .
WILKINS , Circuit Judge: Last year our Court ordered
Respondent United States Department of Energy (“DOE” or
“Agency”) to address three different categories of comments
raised during its informal rulemaking establishing more
stringent energy efficiency standards for commercial packaged
boilers. See Am. Pub. Gas Ass’n v. DOE, 22 F.4th 1018 (D.C.
Cir. 2022) [hereinafter “APGA I”]; see also Energy
Conservation Program: Energy Conservation Standards for
Commercial Packaged Boilers, 85 Fed. Reg. 1592 (Jan. 10,
2020) [hereinafter “Final Rule”]. On remand, the DOE
published a supplement to the Final Rule responding to our
order. See Energy Conservation Program: Energy
Conservation Standards for Commercial Packaged Boilers;
Response to United States Court of Appeals for the District of
Columbia Circuit Remand in American Public Gas Association
v. United States Department of Energy, 87 Fed. Reg. 23421
(Apr. 20, 2022) [hereinafter “Supplement”]. Petitioners are
trade associations and natural gas utilities who assert that they
are negatively affected by the Final Rule as supplemented and
contend that the Agency failed yet again on remand to properly
support its reasoning. They argue further that the DOE failed
to provide notice and comment as required under the
Administrative Procedure Act (“APA”), see 5 U.S.C. § 553(b),
(c), despite relying upon additional literature and new
empirical evidence in the Supplement.
We agree that the DOE should have provided notice and
comment given its reliance on new literature and evidence and
that the DOE again failed to offer a sufficient explanation in

-- 3 of 29 --

4
response to the comments challenging a key assumption in its
analysis. Accordingly, we grant the petitions and vacate the
Final Rule and Supplement.
I.
“The Energy Policy and Conservation Act, as amended in
1992, prescribes energy efficiency standards for certain
commercial and industrial equipment.” APGA I, 22 F.4th at
1022 (citing 42 U.S.C. § 6313). Under the Act, the Secretary
of Energy is authorized to amend the national efficiency
standards to correspond to the industry standards developed by
the American Society of Heating, Refrigerating and Air-
Conditioning Engineers (“ASHRAE”), a private professional
association that writes standards and guidelines for the heating,
air conditioning, and refrigeration industry. ASHRAE’s
standards are known as the ASHRAE/IES Standard 90.1. The
Act also allows the Secretary to adopt a more stringent standard
than what ASHRAE provides if she determines that there is
“clear and convincing evidence” that “adoption of a uniform
national standard more stringent than the amended
ASHRAE/IES Standard 90.1 for the product would result in
significant additional conservation of energy and is
technologically feasible and economically justified.” 42
U.S.C. § 6313(a)(6)(A)(ii)(II).
Originally, the Secretary could not amend the national
“energy efficiency standard[s] for equipment covered by
Section 6313” except “in response to a corresponding
amendment of Standard 90.1 by the ASHRAE.” APGA I, 22
F.4th at 1022. Congress amended the Act in 2007, adding a
“lookback” provision that required the Secretary to “evaluate
whether a more stringent standard is necessary for [any]
category of equipment” for which AHSRAE had failed to
provide an updated standard for six years. Id. (citing

-- 4 of 29 --

5
42 U.S.C. § 6313(a)(6)(C)(i)). “[E]ven under the ‘lookback’
provision, the Secretary may establish a more stringent
standard only if she determines by clear and convincing
evidence that the standard will result in significant
conservation of energy, is technologically feasible, and is
economically justified.” Id. As provided in APGA I:
In determining whether a more stringent
standard is “economically justified,” the
Secretary is required to consider “to the
maximum extent practicable” (1) “the economic
impact of the standard on the manufacturers and
on the consumers of the products subject to the
standard”; (2) “the savings in operating costs
throughout the estimated average life of the
product in the type (or class) compared to any
increase in the price of, or in the initial charges
for, or maintenance expenses of, the products
that are likely to result from the imposition of
the standard” or, in other words, the difference
in the life-cycle cost (LCC) of equipment with
and without a more stringent standard; (3) “the
total projected quantity of energy savings likely
to result directly from the imposition of the
standard”; and other factors not relevant here.
Id. (quoting 42 U.S.C. § 6313(a)(6)(B)(ii)).
This case concerns a rule promulgated by the DOE under
the Act’s “lookback” provision which modified the national
energy efficiency standards for commercial packaged boilers.
A.
Commercial packaged boilers are commonly used to heat
commercial and institutional buildings such as schools, hotels,

-- 5 of 29 --

6
office and apartment buildings, and hospitals. To be defined as
a commercial packaged boiler under DOE regulations, a boiler
must meet certain criteria, including having “a rated input of at
least 300 kBtu/h.” APGA I, 22 F.4th at 1023 (citing 10 C.F.R.
§ 431.82). “Rated input means the maximum rate at which the
commercial packaged boiler has been rated to use energy.” 10
C.F.R. § 431.82. As provided in the Final Rule, “[t]he DOE
categorizes packaged boilers based upon their size (small,
large, and very large), the type of fuel they use (gas-fired or oil-
fired), and their heating medium (hot water or steam).” APGA
I, 22 F.4th at 1023.
ASHRAE updated the standards for commercial packaged
boilers in January 2008 with the release of ASHRAE Standard
90.1-2007. See Energy Conservation Program for Certain
Industrial Equipment: Energy Conservation Standards and Test
Procedures for Commercial Heating, Air-Conditioning, and
Water-Heating Equipment, 74 Fed. Reg. 36312, 36315 (July
22, 2009). In 2009, “the DOE promulgated a Final Rule for
commercial packaged boilers” that adopted the ASHRAE
amendment. APGA I, 22 F.4th at 1023. More than six years
passed after the release of ASHRAE Standard 90.1-2007
without updates to the ASHRAE standards for commercial
packaged boilers, leading to the DOE’s proposal for “new,
more stringent energy efficiency standards for eight of the
twelve categories of commercial packaged boilers” in 2016.
Id. (citing Energy Conservation Program: Energy Conservation
Standards for Commercial Packaged Boilers (Proposed Rule),
81 Fed. Reg. 15836 (Mar. 24, 2016) [hereinafter “2016
Proposed Rule”]).
In the 2016 Proposed Rule, the DOE “tentatively
concluded that there [was] . . . clear and convincing evidence
to support more stringent standards for most types of
commercial packaged boilers.” Id. (quotation marks omitted).

-- 6 of 29 --

7
After notice and comment, the DOE “published its Final Rule
[in 2020], which was, as relevant here, substantively equivalent
to its Proposed Rule.” Id. at 1024. The Final Rule expanded
the different classes of commercial packaged boilers from 10
to 12 and amended the standards to “prescribe [more stringent]
minimum thermal efficiencies (ET ) or combustion efficiencies
(EC )” that “apply to all equipment listed in [the Rule] and
manufactured in, or imported into, the United States on and
after the compliance dates” set by the Agency. 85 Fed. Reg. at
1594. Compliance was mandated for the boilers subject to the
amended energy conservation standards by January 10, 2023,
three years after the Final Rule was published. Id.
As discussed above, the Act requires the DOE to account
for “‘the economic impact of the proposed standard . . . on the
consumers of the products subject to the standard’ and the
difference in [life-cycle cost] savings the standard would bring
about.” APGA I, 22 F.4th at 1023 (quoting
42 U.S.C. § 6313(a)(6)(B)(ii)(I)–(II)). In the Final Rule, the
DOE sought to meet this statutory obligation by developing a
statistical model to compare a valuation of the life-cycle cost
assuming the Agency did not impose a new standard (the “Base
Case”) with a valuation of the life-cycle cost that the market
would bear should the Agency impose a new standard (“New
Standards Case”). Id. “The [life-cycle cost] of any piece of
equipment is the sum of (a) the purchase price (including
installation cost and sales tax) and (b) the lifetime cost of
operating it (fuel, maintenance, and repair), discounted to
present value.” Id. The DOE also calculated the payback
period to further understand the costs and savings associated
with the proposed standards. The payback period is expressed
in years and “is the amount of time it takes the consumer to
recover the additional installed cost of
more-efficient equipment, compared to baseline equipment,
through energy cost savings.” 81 Fed. Reg. at 15875.

-- 7 of 29 --

8
As provided in the DOE’s Technical Support Document,
the Agency’s statistical model used “Microsoft Excel®
spreadsheets combined with Crystal Ball®, a commercially
available simulation add-in, to conduct probability analyses”
that employed a “Monte Carlo simulation and probability
distributions.” J.A. 351. A “Monte Carlo simulation . . .
randomly generates values for uncertain variables again and
again to simulate a model.” Id. at 352. These simulations “can
consist of as many trials (or scenarios) as desired—hundreds or
even thousands.” Id. “During a single trial, Crystal Ball
randomly selects a value from the defined possibilities (the
range and shape of the probability distribution) for each
uncertain variable and then recalculates the spreadsheet.” Id.
The DOE’s “Monte Carlo simulation consists of 10,000 [life-
cycle cost] and [payback period] calculations using input
values that are either sampled from probability distributions
and building samples or characterized with single point
values.” 85 Fed. Reg. at 1626.
The Agency used the Energy Information
Administration’s 2012 Commercial Building Energy
Consumption Survey and the 2009 Residential Energy
Consumption Survey to compile a representative sample of
commercial and residential buildings. 87 Fed. Reg. at 23422.
“[B]oth are national sample surveys that collect information on
the stock of commercial and residential buildings, including
both building characteristics and energy usage data (including
consumption and expenditures).” Resp’t Br. 17–18 (quotation
marks omitted). Next, the DOE had to “estimate . . . the
efficiency of the boilers that would be sold absent the Rule.”
Id. at 16–17. It used shipment data provided by Petitioner Air-
Conditioning, Heating, and Refrigeration Institute (“AHRI”)
“to analyze trends within equipment classes, as it relate[d] to
efficiency levels, to determine the anticipated [Base Case]
efficiency distribution in 2020, the assumed compliance year

-- 8 of 29 --

9
for amended standards.” 85 Fed. Reg. at 1635. Since “the
DOE had historical shipping data for only two of the eight
relevant categories of boilers . . . it assumed the distribution of
efficiency levels in shipped equipment was the same as the
distribution of efficiency levels among models listed in the
database maintained by the AHRI.” APGA I, 22 F.4th at 1023.
In APGA I, we held that “DOE’s reliance upon th[is] proxy
data” was reasonable since it had been “empirically validated”
by the Agency. Id. at 1030.
With both the representative market and an estimate of the
various boilers of different efficiency levels that would be sold
in the Base Case, the Agency had to predict which boilers—
based on efficiency level—would be purchased for each
sample building. To do so, the DOE randomly assigned boilers
to the sample buildings given the share of boilers that would be
sold, such that, for example, “[a]n efficiency level associated
with 30 per cent of the models listed in the AHRI data base had
a 30 per cent chance of being selected for any given
boiler/building combination.” Id. at 1024. Accordingly, the
assignment, while random, was “constrained by the shipment
and model data collected by DOE and submitted by AHRI.” 87
Fed. Reg. at 23423.
The DOE also made several other assumptions and
analytical choices to calculate costs. The Agency assumed for
“the heat load (the amount of heat energy per unit of time that
is needed to maintain a certain temperature in a defined space)
of the sample buildings” “that for every square foot of heated
area, a building uses an average of 30 Btu/h.” APGA I, 22 F.4th
at 1024 (citing 85 Fed. Reg. at 1624). This assumption allowed
the DOE “to calculate the burner operating hours and the
energy use of a given boiler in any boiler/building
combination” for both the Base Case and New Standards Case.
Id. The DOE also assumed that the lifetime of a boiler would

-- 9 of 29 --

10
be 24.8 years and used “energy prices forecasted in the Energy
Information Administration’s 2016 Annual Energy Outlook”
to “estimate the operating cost associated with energy use for
any given boiler/building combination.” Id. “For electricity
and natural gas prices, the DOE . . . applied ‘seasonal marginal
price factors’ to obtain marginal fuel prices, which it said better
represent the cost to the consumer of changes in energy
consumption.” Id. at 1028. “For oil, however, the DOE used
the average prices, because it did not have sufficient data to
convert average prices into marginal prices.” Id.
The Agency found that “[t]he average [life-cycle cost]
savings [were] positive for all equipment classes, and the
[payback period] [was] less than the average lifetime of the
equipment.” 85 Fed. Reg. at 1594. The Agency’s analysis,
including a discussion of the DOE’s various assumptions and
analytical methods, is explained in further detail in the Final
Rule and Technical Support Document.
B.
Petitioners American Public Gas Association (“APGA”)
and AHRI are both trade associations. APGA represents retail
natural gas distribution entities owned by local governments,
and AHRI’s members manufacture commercial packaged
boilers. Petitioner Spire, Inc. owns and operates natural gas
distribution companies, and its subsidiaries—Petitioners Spire
Alabama Inc. and Spire Missouri Inc.—are natural gas utilities
that serve residential, commercial, and institutional customers
in Alabama and Missouri. Originally, Petitioners, excluding
Spire Alabama Inc., submitted several challenges to the Final
Rule, the “most meritorious” of which “target[ed] the
assumptions and data the DOE used to conclude that more

-- 10 of 29 --

11
stringent efficiency standards were economically justified by
clear and convincing evidence.” APGA I, 22 F.4th at 1026.
First, Petitioners challenged the Agency’s random
assignment of boilers to sample buildings, arguing that “the
DOE failed to recognize that a purchaser of commercial
packaged boilers would rationally consider the costs and
benefits of its investment and is likely to buy the boiler that
produces the best economic performance for its building.” Id.
at 1027. In the Final Rule, the Agency “noted that
‘development of a complete consumer choice model, to support
an alternative to random assignment in the no-new-standards
case, for boiler efficiency would require data that are not
currently available, as well as recognition of the various factors
that impact the purchasing decision.’” Id. (quoting 85 Fed.
Reg. at 1638). The DOE also “list[ed] several possible market
failures as ‘problems that this standards [sic] address.’” Id.
(second alteration in original) (quoting 85 Fed. Reg. at 1676).
The Court found that “[t]he significant concerns the petitioners
raised about [random] assignment . . . demand[ed] a more
complete response,” especially given the importance of boiler
assignment to the life-cycle cost analysis. Id. The Court
faulted the DOE for “essentially sa[ying] it did the best it could
with the data it had” “[i]nstead of producing evidence of some
market failure in this specific market.” Id. Given the lack of
“a cogent and reasoned response to the substantial concerns the
petitioners raised about this crucial part of its analysis,” the
Court held that it “[could not] say it was reasonable for the
DOE to conclude that clear and convincing evidence
support[ed] the adoption of a more stringent standard.” Id. at
1028.
Second, Petitioners challenged the DOE’s prediction of
energy prices, claiming “the average prices the DOE used d[id]
not reflect the marginal prices paid by purchasers of

-- 11 of 29 --

12
commercial packaged boilers.” Id. Petitioners noted that
“operators of commercial packaged boilers . . . receive volume
discounts and enter into hedging contracts, and therefore pay
significantly less” because they “are among the largest
purchasers of fuel from energy utilities.” Id. They asserted
that “DOE significantly overstated the savings associated with
promulgation of a stricter standard” since the use of “predicted
average energy prices” did not capture the significantly lower
cost paid by purchasers of commercial packaged boilers. Id.
“The DOE responded that the data sets it used ‘[were] the best
aggregate sources for energy prices currently available’ and it
‘incorporated many adjustment factors to the average price data
and the price trend data to account for the price differences due
to variations in locations, seasons, and market sectors and to
ensure that the energy prices are properly accounted for in the
economic analysis.’” Id. (quoting 85 Fed. Reg. at 1632). The
Court found the DOE’s response to be “conclusory, not
explanatory.” Id. Specifically, the Court noted that the
response failed to address “the lower prices for fuel allegedly
paid by those who operate commercial packaged boilers.” Id.
Finally, as relevant here, Petitioners “challenge[d] the
DOE’s estimates for burner operating hours.” Id. at 1029.
Since it lacked actual burner operating hour data, the Agency
estimated them based on certain building data “and
assumptions about heat load, including the adoption of a rule
of thumb that for every square foot of heated area, a building
uses 30 Btu/h.” Id. During notice and comment, the following
“purported anomalies in the DOE’s estimates” were raised by
a consultant for AHRI, as discussed in APGA I:
“[C]ommercial buildings are generally cooling
load dominated so it would be highly unusual to
have one thousand system operating hours per
year,” yet according to DOE’s estimates, the

-- 12 of 29 --

13
median burner operating hours for six of eight
categories of burners was more than 1000
hours, the 90th percentile of two of the eight
categories was more than 2000 operating hours,
and the maximum burner operating hours in all
categories was well over 2000 hours. Further,
DOE “surprisingly,” he said, estimated that the
median, 90th percentile, and maximum burner
hours for large boilers are lower than the
median, 90th percentile, and maximum burner
hours for small boilers of the same type. These
results, the consultant argued, should have
alerted the DOE to the possibility that either its
assumption about heat load or the data from
[Commercial Building Energy Consumption
Survey] were faulty.
Id. While the DOE “twice acknowledged these comments in
the Final Rule,” it “did not respond to them.” Id. Without
providing a reason, “DOE reiterated that it ‘ha[d] high
confidence that its building load estimation is representative of
the building loads in the field.’” Id. (quoting 85 Fed. Reg. at
1624). Similar to aforementioned responses, the Agency
“explained that ‘[it] ha[d] not identified a source of
comprehensive burner operating hour data for commercial
boilers that could be used for such an analysis nor was such
identified to DOE by stakeholders.’” Id. (quoting 85 Fed. Reg.
at 1637). The Court noted that “[u]sing data ill-suited to the
task is not excused by failure—even good faith failure—to
locate suitable data, particularly considering that the Congress
here required clear and convincing evidence before the
Secretary can disturb the regulatory status quo.” Id.
Accordingly, we ordered the DOE to provide a “reasoned
response to these concerns as well.” Id.

-- 13 of 29 --

14
The Court characterized the “deficiencies of the [Final
Rule] . . . as failures to explain, the type of deficiency most
readily remedied on remand.” Id. at 1031. Accordingly, the
Court found that “remanding the Final Rule to the DOE to
reevaluate it within a limited time [was] the proper remedy.”
Id. at 1030–31. The opinion issued on January 18, 2022, and
gave the DOE 90 days “to take appropriate remedial action” or
“the Final Rule [would] automatically be vacated unless the
agency demonstrate[d] within ten days of the issuance of th[e]
decision the need for additional time.” Id. at 1031. We
originally withheld issuance of the mandate to allow time for
the parties to petition for rehearing, and—after no petition
materialized—issued the formal mandate to the DOE on March
14, 2022.
On March 23, 2022, Petitioners filed a joint submission on
the DOE docket, discussing their view on “the issues DOE
faced on remand” and requesting that the Agency defer
enforcement of the new standards or stay the Rule pending
judicial review arising from any appeal of the DOE’s final
action. Pet’rs Br. 11; see also J.A. 497–98. On April 20, 2022,
the Agency published the Supplement to the Final Rule. While
the Supplement did not discuss the Petitioners’ March 2022
request, it did respond to the three challenges raised in AGPA
I.
Initially, Petitioners tried to challenge the Final Rule as
supplemented by filing a motion to vacate in the original appeal
docket. See Joint Mot. to Vacate, Am. Pub. Gas. Ass’n v. DOE,
No. 20-1068 (D.C. Cir. Apr. 28, 2022). The Court denied the
motion since the case was remanded and our rules require “a
new . . . petition for review . . . [for] . . . a party [to] seek[]
review of the proceedings conducted on remand.” D.C. C IR . R.
41(b); see also Order, Am. Pub. Gas. Ass’n v. DOE, No. 20-
1068 (D.C. Cir. June 1, 2022) (per curiam). APGA filed its

-- 14 of 29 --

15
petition on June 14, 2022, and AHRI and Spire filed separate
petitions for review on June 15, 2022, and June 16, 2022,
respectively. After the petitions were consolidated, Petitioners
unsuccessfully moved the Court to stay the enforcement of the
Final Rule pending appeal. See Order, Am. Pub. Gas. Ass’n v.
DOE, No. 22-1107 (D.C. Cir. Aug. 17, 2022) (per curiam).
Petitioners assert that the DOE failed to adequately explain
its reasoning as required on remand. Further, they contend the
Agency should have provided an opportunity for notice and
comment prior to filing the Supplement and, regardless, that
the Final Rule as supplemented fails to meet the clear and
convincing standard required by the Energy Policy and
Conservation Act.
II.
This Court has jurisdiction to review the Final Rule as
supplemented pursuant to 42 U.S.C. §§ 6306(b) and 6316.
Petitioners have demonstrated standing through declarations
attesting to their expectations of economic losses caused by the
Final Rule that may be remedied by vacatur of the rule. See
generally Sierra Club v. Morton, 405 U.S. 727, 733–34 (1972).
The Court reviews the Final Rule and Supplement under the
APA. See APGA I, 22 F.4th at 1024–25 (citing 5 U.S.C. §
706(2)(A)).
The APA “requires [the Court] to hold unlawful agency
action that is ‘arbitrary, capricious, an abuse of discretion, or
otherwise not in accordance with law.’” Susquehanna Int'l
Grp., LLP v. SEC, 866 F.3d 442, 445 (D.C. Cir. 2017) (quoting
5 U.S.C. § 706(2)(A)). Under this standard, “this Court is
highly deferential to the agency’s decision and presumes that
the agency action is valid.” Oceana, Inc. v. Ross, 920 F.3d 855,
863 (D.C. Cir. 2019) (quotation marks omitted). The Court is
“not a ‘rubber stamp,’” however, and “must ensure the agency

-- 15 of 29 --

16
considered all of the relevant factors.” Id. (quoting Ethyl Corp.
v. EPA, 541 F.2d 1, 34 (D.C. Cir. 1976) (en banc)). “[T]he
focal point for judicial review should be the administrative
record already in existence, not some new record made initially
in the reviewing court.” Camp v. Pitts, 411 U.S. 138, 142
(1973) (per curiam).
Under the Energy Policy and Conservation Act, “the
Secretary is not authorized to establish a more stringent
efficiency standard for commercial packaged boilers . . . unless
there is clear and convincing evidence that the standard would
result in significant additional conservation of energy, would
be technologically feasible, and is economically justified.”
APGA I, 22 F.4th at 1025 (citing 42 U.S.C.
§ 6313(a)(6)(C)(i)(I)–(II)). “[C]lear and convincing evidence
requires a factfinder (in this case the Secretary) to have an
‘abiding conviction’ that her findings (in this case that a more
stringent standard would result in significant additional
conservation of energy, would be technologically feasible, and
is economically justified) are ‘highly probable’ to be true.” Id.
(quoting Colorado v. New Mexico, 467 U.S. 310, 316 (1984)).
“Even where clear and convincing evidence is required before
an agency can act, however, judicial review of agency action
remains deferential.” Id. at 1025–26. “The court asks itself
only whether it was reasonable for the agency to determine it
met the standard.” Id. at 1026.
We discuss each of Petitioners’ challenges in turn.
A.
APGA I required the DOE to provide a cogent and
reasoned response to Petitioners’ challenges to the Agency’s
use of random assignment to model boiler purchases in its life-
cycle cost model. See APGA I, 22 F.4th at 1027–28.
Petitioners contend that the DOE should have provided notice

-- 16 of 29 --

17
and comment since the DOE’s response to our order relied on
new studies and documentation. We agree.
On remand, the DOE supported its use of random
assignment with a more detailed explanation of the various
market failures and behavioral biases it contends lead to
“irrational” energy investment decisions in the market for
commercial packaged boilers, such as purchasing a less
efficient boiler even when a more efficient model might have
lower upfront or lifetime costs. In the Supplement, the Agency
referenced studies that it claimed “demonstrate the existence of
market failures preventing the adoption of energy-efficient
technologies in a variety of commercial sectors around the
world, including office buildings, supermarkets, and the
electric motor market.” 87 Fed. Reg. at 23425. It also cited
corroborating datasets to demonstrate that boilers of various
efficiency levels “are installed in a variety of building types and
that the building characteristics do not correlate strongly with
the existing boiler efficiency.” Id. at 23427. These datasets
included: (1) “data from the Federal Energy Management
Program (‘FEMP’) on commercial gas-fired hot water boiler
installations in government buildings from 2000 to 2013”; (2)
“recent installation data and case studies for areas within the
North region”; and (3) a regional study published in 2020
“characterizing the energy consumption and building
characteristics of commercial buildings throughout the
Northwest region of the country.” Id. at 23425–23426.
Generally, “the ‘technical studies and data’ upon which the
agency relies” “must be revealed for public evaluation.”
Chamber of Com. of U.S. v. SEC, 443 F.3d 890, 899 (D.C. Cir.
2006) (quoting Solite Corp. v. EPA, 952 F.2d 473, 484 (D.C.
Cir. 1991)). This requirement remains binding on the agency
even after our Court has remanded a rule for further
explanation, including when an “agency determines that

-- 17 of 29 --

18
additional fact gathering is necessary” on remand. Id. at 900.
While we have recognized certain exceptions to this
requirement, see id. (collecting cases), none apply here.
First, the DOE contends that notice and comment was
unnecessary on remand because the Final Rule merely
“advanced ‘a hypothesis’ and some supporting explanation,”
and the Supplement “provided additional support for that
hypothesis . . . but . . . did not reject or modify the hypothesis
such that additional comment was necessary.” Resp’t Br. 56
(quoting Bldg. Indus. Ass’n of Superior Cal. v. Norton, 247
F.3d 1241, 1246 (D.C. Cir. 2001)). As we held in Building
Industry, “a final rule that is a logical outgrowth of the proposal
does not require an additional round of notice and comment
even if the final rule relies on data submitted during the
comment period.” 247 F.3d at 1246. In Building Industry,
however, the agency provided more than an unsupported
explanation to bolster its hypothesis. Instead, the agency’s
“proposal advanced for comment a hypothesis and some
supporting data.” Id. (emphasis added). The additional study
relied upon, “released after the proposal,” id. at 1245, provided
“additional support for that hypothesis—indeed, better support
than was previously available,” id. at 1246.
Such was the case in International Fabricare Institute v.
EPA, another authority cited by the DOE. 972 F.2d 384 (D.C.
Cir. 1992) (per curiam). In that case, the petitioner challenged
the EPA’s newly adopted “regulations establish[ing]
permissible concentration levels for contaminants occurring in
drinking water.” Id. at 387. The agency had to determine
which method it would use to “ascertain how low a
concentration of [a regulated] chemical reliably [could] be
measured when testing water to determine compliance with the
limit.” Id. at 398. In its original notice of the rulemaking, “the
EPA acknowledged . . . that [the chosen method’s] accuracy

-- 18 of 29 --

19
had been verified in only one laboratory” and sought comments
on the proposed approach. Id. at 399 (citations omitted). After
receiving comments challenging the reliability of the method,
the EPA promulgated the regulations, relying upon additional
studies “conducted by private laboratories [that] [i]n the EPA’s
view . . . adequately confirmed the reliability of [its chosen
method].” Id. As in Building Industry, the EPA’s original
notice referenced some documentation in support of the
challenged approach, specifically the verification of the chosen
method by one laboratory. We held that notice and comment
was unnecessary since the Fabricare “petitioners had fair
notice of, and full opportunity to comment on, the issue
actually decided by the EPA.” 972 F.2d at 399.
Here, the new studies and datasets referenced in the
Supplement did not “address[] ‘alleged deficiencies’ in [any]
pre-existing data.” Solite Corp., 952 F.2d at 484 (quoting
Cmty. Nutrition Inst. v. Block, 749 F.2d 50, 58 (D.C. Cir.
1984)). Instead, the additional materials referenced in the
Supplement provided “entirely new information ‘critical’ to the
[Agency’s] determination” of life-cycle costs. Block, 749 F.2d
at 58. In APGA I, we explained that “we [could not] say it was
reasonable for the DOE to conclude that clear and convincing
evidence support[ed] the adoption of a more stringent
standard” absent a “cogent and reasoned response to the
substantial concerns the petitioners raised” about the Agency’s
use of random assignment. APGA I, 22 F.4th at 1028. Absent
the cited studies and corroborating documentation, the DOE
fails to adequately explain how there is a “rational relationship”
between their model and the purchasing behavior in the market
for commercial packaged boilers. Am. Iron & Steel Inst. v.
EPA, 115 F.3d 979, 1004 (D.C. Cir. 1997) (per curiam)
(quoting Chem. Mfrs. Ass’n v. EPA, 28 F.3d 1259, 1265 (D.C.
Cir. 1994)). The cited materials were necessary to respond to
our order and justify a key input in the life-cycle cost analysis.

-- 19 of 29 --

20
Because they were relied upon for the first time on remand, the
Agency should have provided an opportunity for notice and
comment as required by the APA. See 5 U.S.C. § 553(b), (c).
Second, the DOE argues that it should be excused from the
APA’s notice and comment requirements because Petitioners
have failed to demonstrate that they were “prejudiced by [the]
lack of opportunity to comment.” Chamber of Com., 443 F.3d
at 904. Specifically, the Agency contends that Petitioners fail
to “specify what objectionable new information the
Department relied on” and “what they would have submitted in
response to that information beyond what was already
submitted during earlier notice and comment.” Resp’t Br. 62–
63 (citing 5 U.S.C. § 706 and Chamber of Com., 443 F.3d at
904).
Petitioners do not have a high burden in demonstrating
“prejudice in notice-and-comment cases.” Chamber of Com.,
443 F.3d at 904. In general, “an utter failure to comply with
notice and comment cannot be considered harmless if there is
any uncertainty at all as to the effect of that failure.” Sugar
Cane Growers Co-op. of Fla. v. Veneman, 289 F.3d 89, 96
(D.C. Cir. 2002). Accordingly, those objecting to an agency’s
late reference to critical documents can demonstrate prejudice
by creating “enough ‘uncertainty [as to] whether petitioner’s
comments would have had some effect if they had been
considered.’” Chamber of Com., 443 F.3d at 904 (quoting
McLouth Steel Prods. Corp. v. Thomas, 838 F.2d 1317, 1324
(D.C. Cir. 1988)).
Petitioners make several objections to the studies and
datasets cited in the Supplement. They note that the referenced
studies are too “generic” to apply to the market for commercial
packaged boilers. Pet’rs Br. 37. Petitioners also contend that
the datasets which the DOE references to “demonstrate[] [the]

-- 20 of 29 --

21
relevant market failures would—at most—have incremental
impacts insufficient to justify random assignment.” Id. at 38.
Petitioners have “indicate[d] with ‘reasonable specificity’”
“the nature of [their] objection[s]” and “how [they] might have
responded if given the opportunity.” Air Transp. Ass’n of Am.
v. FAA, 169 F.3d 1, 8 (D.C. Cir. 1999) (quoting Air Transp.
Ass’n of Am. v. CAB, 732 F.2d 219, 224 n.11 (D.C. Cir. 1984)).
These objections provide enough uncertainty as to whether the
Petitioners’ comments would have influenced the Agency’s
decision had they been given the opportunity to comment.
Further, Petitioners “had no knowledge of the new information
until” the Supplement was published “and had no subsequent
opportunity to provide comments.” Id. Under these
circumstances, Petitioners have demonstrated prejudice from
the DOE’s failure to provide notice and comment.
Third, the Agency contends that the APA’s “good cause”
exception to notice and comment should apply. It claims that
“a new round of notice and comment was not expected by this
Court and would have been impracticable” given the court-
imposed 90-day deadline with only the first 10 days after the
mandate issued made available for the Agency to request an
extension. Resp’t Br. 59. It compares the situation to
Methodist Hospital of Sacramento v. Shalala, an appeal in
which this Court held that notice and comment was
impracticable due, in part, to the tight deadlines imposed by
Congress. 38 F.3d 1225, 1237 (D.C. Cir. 1994). The Agency
contends that our “Court has found good cause where there is
no indication that the agency ‘had a substantial period of time
within which to propose regulations’ or that the agency abused
the deadline by ‘simply waiting until the eve of . . . the deadline,
then raising up the “good cause” banner.’” Resp’t Br. 60
(quoting Methodist, 38 F.3d at 1237) (emphasis omitted).

-- 21 of 29 --

22
The Agency misreads our holding in Methodist. In that
case, we reiterated that “strict congressionally imposed
deadlines, without more, by no means warrant invocation of the
good cause exception.” Methodist, 38 F.3d at 1236 (quoting
Petry v. Block, 737 F.2d 1193, 1203 (D.C. Cir. 1984)).
However, we found the good cause exception applied since the
“congressional deadlines [were] very tight and . . . the statute
[was] particularly complicated.” Id. Among other things, we
noted that the situation in Methodist differed from other “cases
where the court has found strict deadlines alone insufficient to
establish good cause” because in the relevant statutory scheme
“Congress ha[d] expressed its clear intent that APA notice and
comment procedures need not be followed.” Id. at 1237
(citations omitted). The DOE makes no claim that the Energy
Policy and Conservation Act imposes a similarly complex set
of procedures as in Methodist. A tight “statutory, judicial, or
administrative deadline” alone, Council of S. Mountains, Inc.
v. Donovan, 653 F.2d 573, 581 (D.C. Cir. 1981) (per curiam),
“by no means warrant[s] invocation of the good cause
exception,” Methodist, 38 F.3d at 1236.
“[T]he good cause exception is to be narrowly construed
and only reluctantly countenanced.” Mack Trucks, Inc. v. EPA,
682 F.3d 87, 93 (D.C. Cir. 2012) (quotation marks omitted).
We have typically applied the good cause exception to
“excuse[] notice and comment in emergency situations, where
delay could result in serious harm, or when the very
announcement of a proposed rule itself could be expected to
precipitate activity by affected parties that would harm the
public welfare.” Chamber of Com., 443 F.3d at 908 (citations
omitted). Here, where none of the aforementioned
circumstances applied and the Agency declined to seek an
available extension of its compliance deadline, the DOE lacked
good cause to adopt the Final Rule as supplemented absent
public comment on the new studies and documentation it relied

-- 22 of 29 --

23
upon. Since we find that the DOE lacked good cause to
dispense with the required notice and comment procedures, we
do not reach Petitioners’ alternative argument that the Agency
failed again on remand to provide a cogent and reasoned
response to the concerns raised about the DOE’s use of random
assignment.
B.
In APGA I, the Court required the DOE to address
Petitioners’ concerns with the Agency’s use of certain fuel
prices that informed its life-cycle cost analysis, specifically
whether the data sets it used captured “the lower prices for fuel
allegedly paid by those who operate commercial packaged
boilers.” 22 F.4th at 1028. The Court characterized the DOE’s
original response as “conclusory, not explanatory,” specifically
describing the response as follows:
The DOE responded that the data sets it used
“are the best aggregate sources for energy prices
currently available” and it “incorporated many
adjustment factors to the average price data and
the price trend data to account for the price
differences due to variations in locations,
seasons, and market sectors and to ensure that
the energy prices are properly accounted for in
the economic analysis.”
Id. (quoting 85 Fed. Reg. at 1632).
On remand, the Agency did more than simply say the U.S.
Energy Information Administration (“EIA”) data it relied on
was “the best aggregate source[] for energy prices.” Id. It
explained specifically how the data captured the prices paid by
all consumers. In the Supplement, the DOE “emphasize[d] that
the EIA data provide[d] complete coverage of all utilities and

-- 23 of 29 --

24
all customers, including larger commercial and industrial
utility customers that may have discounted energy prices.” 87
Fed. Reg. at 23428. Though the DOE conceded that it “was
unable to identify data to provide a basis for determining a
potentially lower price for larger commercial and industrial
utility customers, either on a state-by-state basis or in a
nationally representative manner,” the Supplement explained
that “the historic data on which DOE did rely includes such
discounts.” Id. at 23429. The DOE finally asserted that any
adjustment to its analysis—such as “to adjust downward the
marginal energy price for a small subset of individual
customers in the [life-cycle cost] Monte Carlo sample as
suggested by commenters”—would “yield[] substantially the
same overall average [life-cycle cost] savings result as DOE’s
current estimate” since that data already includes “actual utility
rates paid by all customers.” Id. In our view, the Agency
provided a “cogent response” that adequately addressed
Petitioners’ concerns about “the lower prices for fuel allegedly
paid by those who operate commercial packaged boilers.”
APGA I, 22 F.4th at 1028.
For the first time in rebuttal, however, Petitioners fault the
natural gas prices data used by the DOE for failing to include
one category of large consumers in its analysis: industrial and
manufacturing facilities. They point to the Final Rule’s citation
to an EIA website describing that the natural gas pricing data
used by the DOE “indicate[s] that it is limited to
‘nonmanufacturing establishments.’” Pet’rs Reply Br. 14 n.4
(citing J.A. 360–61); see also J.A. 373.
“We require petitioners and appellants to raise all of their
arguments in the opening brief to prevent ‘sandbagging’ of
appellees and respondents and to provide opposing counsel the
chance to respond.” Corson & Gruman Co. v. NLRB, 899 F.2d
47, 50 n.4 (D.C. Cir. 1990). There is an exception to this

-- 24 of 29 --

25
general rule allowing a petitioner, “in a reply brief, [to] respond
to arguments raised for the first time in the [respondent’s]
brief.” United States v. Powers, 885 F.3d 728, 732 (D.C. Cir.
2018) (quotation marks omitted). This exception applies when
a petitioner could not be “required to assume in [their] opening
brief that the [opposing party] would rely on” a specific
argument. Id. Here, however, the DOE’s argument was
foreseeable since the Supplement provided that the “DOE’s
current approach . . . captures the impact of actual utility rates
paid by all customers, including those that enjoy lower
marginal rates for whatever reason, in an aggregated fashion.”
87 Fed. Reg. at 23429. Petitioners did not have to “assume”
the Agency’s argument, because they had it right before them
from the start. Powers, 885 F.3d at 732.
Accordingly, Petitioners have forfeited this argument. We
find that the Final Rule as supplemented provides a sufficient
response to Petitioners’ concerns about the fuel prices used in
the DOE’s life-cycle cost analysis.
C.
In APGA I, the Court required the DOE to address
Petitioners’ concerns that the estimated burner operating hours
used in the life-cycle cost analysis were anomalous and the
“possibility that either [the DOE’s] assumption about heat load
or the data from [the Commercial Building Energy
Consumption Survey] were faulty.” 22 F.4th at 1029.
Petitioners argue that the Final Rule as supplemented still
ignores these comments by failing to address its “energy-use
assumption” and erroneously suggesting that “burner operating
hours have minimal impact on the results of its analysis.”
Pet’rs Br. 43–44.
On remand, the Agency provided a more detailed
description of how the underlying Commercial Building

-- 25 of 29 --

26
Energy Consumption Survey data reflects real-world building
energy use. See 87 Fed. Reg. at 23429–23430. It also
described how the seemingly anomalous burner operating hour
results were actually reflective of real world conditions in
which both boiler size and other factors like the climate or the
age of the building require longer and higher burner operating
hours. Id. at 23430. At the same time, however, the DOE
failed to address the impact of its underlying assumption “that
for every square foot of heated area, a building uses an average
of 30 Btu/h.” APGA I, 22 F.4th at 1024. In a colloquy with the
Court, the DOE’s counsel conceded that there was no
explanation at all in the Supplement to support the Agency’s
30 Btu/h assumption. Oral Arg. Tr. 28.
Even if burner operating hours may be a “derived
quantity” as the Agency argues, see Resp’t Br. 51 (quoting 87
Fed. Reg. at 23430), the “30 Btu/h” figure is an input into the
DOE’s calculation that commenters complained caused
erroneous building load estimations separate from the concerns
raised about the Commercial Building Energy Consumption
Survey data. See, e.g., J.A. 202, 389. In APGA I, we held that
these concerns required a “reasoned response,” yet DOE
provided no additional explanation for the assumption on
remand. 22 F.4th at 1029.
We review the DOE’s “rulemaking under the ‘lookback’
provision” under the APA, id. at 1024, with the “further
wrinkle” that “clear and convincing evidence is required before
[the] [A]gency can act,” id. at 1025. The Court must
“ask[] . . . whether it was reasonable for the agency to
determine it met the [clear and convincing evidence] standard.”
Id. at 1026. “The ‘scope of review’ provisions of the APA, 5
U.S.C. § 706(2), are cumulative.” Ass’n of Data Processing
Serv. Orgs., Inc. v. Bd. of Governors of Fed. Rsrv. Sys., 745
F.2d 677, 683 (D.C. Cir. 1984). “Thus, an agency action which

-- 26 of 29 --

27
is supported by the required substantial evidence may in
another regard be ‘arbitrary, capricious, an abuse of discretion,
or otherwise not in accordance with law’—for example,
because it is an abrupt and unexplained departure from agency
precedent.” Id. Similarly, the Court may find the DOE’s
failure to explain and justify the “30 Btu/h” assumption on
remand to be “arbitrary and capricious” without determining
whether that omission demonstrated the Agency acted without
the requisite “clear and convincing” evidence.
Again, the DOE has “fail[ed] to engage the arguments
raised before it.” APGA I, 22 F.4th at 1027 (quotation marks
omitted).
III.
“‘[V]acatur is the normal remedy’ when a rule is found
unlawful.” APGA I, 22 F.4th at 1030 (quoting Allina Health
Servs. v. Sebelius, 746 F.3d 1102, 1110 (D.C. Cir. 2014)).
Under certain circumstances, however, the Court may remand
without vacatur and allow the agency to “fix the deficient rule.”
Id. “The decision to vacate depends on two factors: the
likelihood that ‘deficiencies’ in an order can be redressed on
remand, even if the agency reaches the same result, and the
‘disruptive consequences’ of vacatur.” Black Oak Energy, LLC
v. FERC, 725 F.3d 230, 244 (D.C. Cir. 2013) (quoting Allied-
Signal v. Nuclear Regul. Comm’n, 988 F.2d 146, 150–51 (D.C.
Cir. 1993)).
For the first factor, we have discussed two significant
deficiencies with the Final Rule as supplemented. The Agency
failed to provide notice and comment despite its reliance on
new studies and data critical to supporting its use of random
assignment to assign boilers in the life-cycle cost analysis. The
DOE also failed to address challenges to its 30 Btu/h

-- 27 of 29 --

28
assumption in calculating burner operating hours for the life-
cycle cost analysis for the second time.
As to the second factor, the DOE contends that vacatur of
the Final Rule would result in “significant disruption of the
status quo” since the Final Rule went into effect on January 10,
2023, and some consumers and manufacturers would have to
manage switching back to the prior standards after several
years of preparing to comply with the new, more stringent
standards. Resp’t Br. 65–66. Further, the Agency argues that
vacatur would harm the public given the loss of the significant
environmental and health benefits expected from the new
efficiency standards. It contends that the harm of losing these
benefits would be “long-lived” since noncompliant boilers,
with an expected lifetime use of approximately 25 years, would
be manufactured and sold. Id. at 67–68. Even should the
Agency repromulgate the rule, it asserts that there is little
chance noncompliant boilers manufactured and sold in the
interim would be replaced with compliant boilers in the near
term. Petitioners also concede that the effect of the Final Rule
has resulted in “[s]ome manufacturers hav[ing] already
suffered irreparable injuries.” Pet’rs Reply Br. 28.
The disruptive consequences of vacatur are apparent, and
we are “sensitive to the risk of interfering with environmental
protection, which is one potential disruptive consequence”
raised by the DOE. North Carolina v. EPA, 531 F.3d 896, 929
(D.C. Cir.) (per curiam), modified on reh’g in part, 550 F.3d
1176 (D.C. Cir. 2008). However, none of the DOE’s
arguments demonstrate that “[t]he egg has been scrambled and
there is no apparent way to restore the status quo ante,” namely
the state of affairs under the prior, less stringent standards.
Sugar Cane Growers, 289 F.3d at 97. As Petitioners state,
vacatur would allow “manufacturers to resume production of
boilers” that meet either standard, and it is undisputed that

-- 28 of 29 --

29
“many manufacturers already sell, and a significant number of
consumers already purchase, boilers that meet DOE’s
more-stringent standard.” Pet’rs Reply Br. 28. Separately,
even though we have found the DOE’s explanation regarding
fuel prices sufficient, “leaving the regulations in place during
remand would ignore [P]etitioners’ potentially meritorious
challenges” related to the use of random assignment, which we
have chosen not to reach given the lack of notice and comment,
and regarding the 30 Btu/h assumption that the DOE failed to
explain. Cement Kiln Recycling Coal. v. EPA, 255 F.3d 855,
872 (D.C. Cir. 2001) (per curiam).
“[T]he [C]ourt typically vacates rules when an agency
entirely fails to provide notice and comment.” Daimler Trucks
N. Am. LLC v. EPA, 737 F.3d 95, 103 (D.C. Cir. 2013)
(quotation marks omitted). The DOE’s “fail[ure] to comply
with our remand order” also counsels toward vacatur, since it
has yet again “come up with insufficient support” for the 30
Btu/h assumption. Tex Tin Corp. v. EPA, 992 F.2d 353, 356
(D.C. Cir. 1993). We see no reason to break from our
established practice when for the second time “we are not
persuaded it was reasonable for the Secretary to conclude the
Final Rule was supported by clear and convincing evidence.”
APGA I, 22 F.4th at 1022.
In sum, we grant the petitions and vacate the Final Rule as
supplemented. We remand to DOE for further proceedings
consistent with this opinion.
So ordered.

-- 29 of 29 --