United States Court of Appeals
         FOR THE DISTRICT OF COLUMBIA CIRCUIT



Argued March 6, 2023                   Decided May 16, 2023

                        No. 22-1148

GPA MIDSTREAM ASSOCIATION AND AMERICAN PETROLEUM
                   INSTITUTE,
                  PETITIONERS

                              v.

   UNITED STATES DEPARTMENT OF TRANSPORTATION AND
       PIPELINE AND HAZARDOUS MATERIALS SAFETY
                    ADMINISTRATION,
                     RESPONDENTS


           On Petition for Review of a Final Rule
            of the Department of Transportation



     Keith J. Coyle argued the cause for petitioners. With him
on the briefs was Christina Manfredi McKinley.

    Anna O. Mohan, Attorney, U.S. Department of Justice,
argued the cause for respondent. With her on the brief were
Brian M. Boynton, Principal Deputy Assistant Attorney
General, Abby C. Wright, Attorney, Paul M. Geier, Assistant
General Counsel, U.S. Department of Transportation, and
Charles E. Enloe, Senior Trial Attorney.
                               2
    Before: CHILDS, Circuit Judge, and EDWARDS and
GINSBURG, Senior Circuit Judges.

    Opinion for the Court filed by Senior Circuit Judge
GINSBURG.

     GINSBURG, Senior Circuit Judge: The Pipeline and
Hazardous Materials Safety Administration (PHMSA)
prescribes safety standards for pipelines on behalf of the
Secretary of Transportation. 

(a)(2); 

(a)(1). Two oil and gas associations, GPA
Midstream and the American Petroleum Institute, petition for
review of a safety standard requiring their members to install
remote-controlled or automatic shut-off valves in some types
of new or replaced gas and hazardous liquid pipelines. 

 (2022). The petitioners challenge the standard as
it applies to “gathering” pipelines used to collect raw gas or
crude oil from a well. They argue the PHMSA unlawfully
failed to disclose the economic basis for regulating gathering
pipelines when it proposed the standard, and also failed to
make a reasoned determination that regulating these pipelines
was appropriate.

     We agree. The PHMSA said nothing about the
practicability or the costs and benefits of the standard for
gathering pipelines until promulgating the final rule, even
though the law required it to address those subjects when
publishing the proposed rule for public comment and peer
review. The PHMSA also ultimately failed to make a reasoned
determination that the benefits of regulating gathering
pipelines would exceed the costs, and that doing so would be
practicable, as required by law. We therefore grant the petition
for review.
                              3
                      I.   Background

    We begin with some regulatory and factual background.

A. Regulated Gathering Lines

     In order to collect raw gas or crude oil from one or more
wells, oil and gas companies rely upon so-called gathering
pipelines, which pass mostly through rural areas. 

, 195.2. Oil and gas companies then rely upon a large
network of transmission or “main” lines to transport gas, crude
oil, and petroleum products long distances further down the
supply chain, as illustrated by this graphic from the PHMSA.




     The PHMSA regulates these pipelines under the
Secretary’s longstanding authority to “prescribe minimum
safety standards for pipeline transportation and for pipeline
facilities.” 

(a)(2). Although it has broad
authority, the agency must follow detailed rulemaking
procedures. As relevant here, a 1996 law provides the PHMSA
must (1) “identify the costs and benefits associated with the
proposed standard” in a risk assessment, and (2) submit this
risk assessment to an advisory committee of experts for peer
                               4
review, and to the public for comment. Accountable Pipeline
Safety and Partnership Act of 1996, § 4, 

Stat. 3794, 3795, codified at 

(b)(3)(B), (4).
Ultimately, the PHMSA must regulate “only upon a reasoned
determination that the benefits, including safety and
environmental benefits, of the intended standard justify its
costs.” 

(b)(5).

     Transmission pipelines have long been subject to federal
safety standards. By contrast, for many years, gathering
pipelines in rural areas were not subject to federal safety
standards. In the 2000s, however, the PHMSA defined a new
class of “regulated gathering line” subject to federal safety
standards in rural areas. See 

 (2006)
(defining regulated gathering line for gas); 

(2008) (defining regulated rural gathering line for hazardous
liquids); see also 

(b)(2)(A) (authorizing the
PHMSA to define the term “regulated gathering line”). This
definition depends upon the proximity of a pipeline segment to
an area where a rupture may cause serious harm.

     A gas pipeline carries methane, which is a safety hazard.
When a cloud of methane ignites, the resulting fireball and heat
can kill and burn anything nearby. Since 2006, the PHMSA,
therefore, has regulated segments of gas gathering pipeline
near more than ten buildings intended for human occupancy,
where burning gas may take lives and destroy property. 

(c)(1), 192.5(b)(2)–(4). The agency no longer
treats rural and non-rural gas gathering pipelines differently.

     Also in 2006, the PHMSA created two tiers of regulated
gas gathering line—Types A and B—based upon “hoop
stress,” that is, the force on the wall of a steel pipe as fluid
                                  5
pushes outward against it. * Hoop stress is an important physical
variable. When hoop stress exceeds the stress specified by the
manufacturer, the pipe may permanently expand like an
overstretched rubber band, thus thinning out the wall, which
may eventually burst. This important specification is known as
the “specified minimum yield strength” (SMYS) of the pipe.
Thomas O. Miesner & William L. Leffler, Oil and Gas
Pipelines in Non-Technical Language 232 (2006). A Type A
line operates at or above 20% of the SMYS of the pipe, while
a Type B line operates at a lower stress. 

(c).
The PHMSA determined the stress of a Type A line was
“indicative of onshore gathering lines whose operating
pressure presents a significant enough risk in certain
circumstances to warrant the same amount of regulation as
transmission lines,” with exceptions not here relevant. 71 Fed.
Reg. at 13,296/3; see also 

(c) (“An operator
of a Type A regulated onshore gathering line must comply with
the requirements of this part applicable to transmission lines,
except . . .”). By contrast, a low stress Type B line must meet
fewer safety standards. 

(d).

     Hazardous liquid gathering pipelines, which largely
transport crude oil, may contaminate drinking water or natural
resources. In 2008, the PHMSA therefore regulated rural
gathering pipelines located within a quarter mile of a source of
drinking water or of an important ecological resource (e.g., a
wetland frequented by migratory birds or by an endangered
species), as long as the pipeline falls within a certain diameter
range and operates at or above 20% of the SMYS of the steel
pipe. 

 § 195.11(a). A regulated rural gathering line must

*
  Office of Pipeline Safety, Interpretation Response PI-70-024
(1970), https://perma.cc/GEW6-SQBQ. Hoop stress is determined
by S = D × P / 2t, where S is hoop stress, D is diameter, P is internal
operating pressure, and t is wall thickness. Id.
                                6
comply with many of the safety standards that apply to a
transmission line carrying hazardous liquids. See id.
§ 195.11(b); see also 73 Fed. Reg. at 31,640/3 (describing the
overlapping rules). The PHMSA also continued regulating any
gathering “pipeline located in a non-rural area.” 

(a)(4)(i). As a result, regulated gathering lines and
transmission lines are now generally subject to many of the
same safety standards.

B. The 2011 Act

     In 2010, the rupture of a gas transmission pipeline
destroyed a neighborhood in San Bruno, California. The
ignition blast and fire resulting from the rupture killed eight and
injured 51 persons, and destroyed or damaged more than 100
homes. It took 95 minutes to stop the flow of gas from the
rupture, as firefighters struggled to contain the flames.

     The National Surface Transportation Board investigated
and prepared a report. Pacific Gas and Electric Company
Natural Gas Transmission Pipeline Rupture and Fire,
NTSB/PAR-11/01 (2011), https://perma.cc/868M-ASNR. As
relevant here, the Board found property damage could have
been mitigated had two nearby valves been equipped with
automatic or remote-controlled shut-off valves, which do not
require dispatching a local mechanic to the site of the valve. 

at 103–104, 125. The Board accordingly recommended that the
PHMSA require automatic or remote-control shut-off valves in
high-consequence areas and high-density locations. 

.

     Within a few months, the Congress passed the Pipeline
Safety, Regulatory Certainty, and Job Creation Act of 2011, § 4
of which directs the PHMSA to require automatic or remote-
controlled valves “if appropriate” in “transmission pipeline
facilities”:
                               7

    Not later than 2 years after the date of enactment of this
    subsection, and after considering the factors specified in
    subsection (b)(2), the [PHMSA], if appropriate, shall
    require by regulation the use of automatic or remote-
    controlled shut-off valves, or equivalent technology,
    where economically, technically, and operationally
    feasible on transmission pipeline facilities constructed or
    entirely replaced after the date on which the [PHMSA]
    issues the final rule containing such requirement.

(n)(1).

C. The Proposed Rule

    In 2020, the PHMSA published a notice of proposed
rulemaking to implement § 4. The agency proposed requiring
automatic or remote-controlled shut-off valves or an equivalent
technology “on all newly constructed or entirely replaced
natural gas transmission and hazardous liquid pipelines that
have nominal diameters of 6 inches or greater.” 

, 7164/1 (2020). As required by law, the PHMSA also
prepared a risk assessment purporting to “identify the costs and
benefits associated with the proposed standard.” 

(b)(3)(B). Notably, however, the notice of proposed
rulemaking and risk assessment said nothing about the costs
and benefits of applying the standard to gathering pipelines.

     Pursuant to certain pre-existing rules, however, new or
replaced regulated gathering lines would have been subject to
the proposed standard unless expressly carved out by the rule.
In their comments, the petitioners accordingly sought an
exemption for gathering pipelines. Among other things, they
argued the risk assessment lacked the cost-benefit data needed
to justify applying the rule to gathering pipelines.
                                8

D. The Advisory Committee Recommendation

     As required by law, the PHMSA also made the risk
assessment available for public comment and for peer review
by two federal advisory committees, one for gas pipelines and
one for pipelines carrying hazardous liquids. 

(b)(4)(A)–(B), 60115. Because the risk assessment
said nothing about the costs and benefits of applying the safety
standard to gathering pipelines in particular, however, the
public committees had no economic data or analysis about
gathering pipelines to review and analyze.

     Nevertheless, when the PHMSA presented the proposal to
the advisory committees, the agency revealed for the first time
that it would be applying the standard to at least some regulated
gathering lines. Members of the committees argued this was
inappropriate because they had been given no analysis or data
for gathering pipelines. Because of this gap, the committees
recommended deferring the safety standard for gathering
pipelines until a future rulemaking.

E. The Final Rule

     The PHMSA plowed ahead anyway. In the final rule, the
PHMSA required automatic or remote-controlled shut-off
valves in many new or replaced pipelines with a diameter of
six or more inches, including Type A lines and regulated rural
gathering lines that carry hazardous liquids across a body of
water wider than 100 feet. 68 Fed. Reg. at 20,941/3, 20,949.

     The PHMSA addressed some objections in the preamble
to the final rule. Id. at 20,949. It pointed out that the proposed
rule never said regulated gathering lines would be exempt—
which is correct because the proposed rule said nothing at all
                                9
about gathering lines. Id. at 20,949/1. The PHMSA also said
regulating Type A lines made sense because “ruptures on these
pipelines will generally present a higher risk of public safety
consequences, similar to gas transmission lines.” Id.

     The risk assessment accompanying the final rule included
some data about gathering pipelines. To identify the costs, the
PHMSA estimated the number of new or replaced miles of
pipeline that would be subject to the standard each year—
including, for the first time, the number of miles of gathering
pipelines. Still, when the PHMSA then estimated the number
of valves needed, the annualized cost of the equipment, and
other costs, it used no data for gathering pipelines; nor did it
discuss the benefits of applying the standard to regulated
gathering lines. It filled the gap by assuming the methodology
and data for transmission pipelines were also valid for
gathering lines.

     The PHMSA said it was “not able to quantify the benefits,”
because “a detailed projection of avoided incidents and avoided
costs of those incidents is not available”; therefore, it discussed
benefits in qualitative terms. The PHMSA claimed the standard
could avoid “significant” property and environmental
damages. To support this claim, the PHMSA relied upon a
foundational technical study the agency commissioned from
Oak Ridge National Laboratory, modeling the property and
environmental damages that could be avoided by using
automatic or remoted-controlled valves in specific scenarios.
See Studies for the Requirements of Automatic and Remotely
Controlled Shutoff Valves on Hazardous Liquids and Natural
Gas Pipelines with Respect to Public and Environmental
Safety, ORNL/TM-2012/411 (2012), https://perma.cc/QZP7-
6FKX (Oak Ridge). The study, however, covered only
transmission pipelines. Id. at 6 (“The results of this study apply
to natural gas and hazardous liquid transmission lines.”).
                                 10

    Relying upon this final risk assessment, the PHMSA
concluded, in the preamble to the final rule, that the
“(unquantified) public safety, environmental, and equity
benefits of the final rule . . . justify the costs of the final rule.”
68 Fed. Reg. at 20,943/1.

                           II. Analysis

     We begin by considering whether the petition for review
is timely. Finding it is, we proceed to the merits, applying the
familiar standards of review in the Administrative Procedure
Act (APA). 

(a)(3). To the extent the
petitioners argue the safety standard exceeds the PHMSA’s
authority, we deny the petition. We agree, however, that the
PHMSA prescribed the standard “without observance of
procedure required by law.” 

(2)(D). We further
agree that the PHMSA ultimately failed to show that regulating
gathering pipelines would be “appropriate.” 

(b)(2)(B).

A. The Petition for Review Is Timely

     A petition for review of a safety standard must be filed
“not later than 89 days after the regulation is prescribed.” 

(a)(1). A final rule is “prescribed” when it is
“established authoritatively.” Nat. Res. Def. Council v. Nat’l
Highway Traffic Safety Admin., 

 (2d Cir.
2018) (brackets omitted) (quoting Black’s Law Dictionary
(10th ed. 2014)). A legislative rule is established
authoritatively when it is duly fixed and so becomes binding on
the public, “even if it sets a future effective date.” Humane
Soc’y v. Dep’t of Agric., 

 (D.C. Cir. 2022).
                                 11
     A final rule is not duly fixed at least until it is filed for
public inspection with the Office of the Federal Register. See

. Until then, it may be withdrawn without explanation
or notice and comment and is “not valid” and enforceable
against the public at large. 

; Humane Soc’y,

. Because a legislative rule that is unenforceable
and may be withdrawn at will is not “duly fixed,” we hold a
standard is not “prescribed” by the PHMSA at least until the
agency files the final rule for public inspection. Cf. James
Madison, The Federalist, No. 62, at 381 (Clinton Rossiter ed.,
1961) (“Law is defined to be a rule of action; but how can that
be a rule, which is little known, and less fixed?”).

     That was done on April 7, 2022. The petitioners filed their
petition on July 1, 2022, 85 days later. The petition is therefore
timely. Because the petition is timely, we have no occasion to
decide whether the filing clock started running only after the
rule was first published in the Federal Register, as the
petitioners argue, citing Natural Resources Defense Council,

 (“[A] regulation is not ‘prescribed’ until it has
legal effect, and it does not have legal effect until it is published
in the Federal Register.”). Nor have we occasion to decide
whether the 89-day deadline is jurisdictional.

B. The PHMSA Did Not Exceed Its Legal Authority

     In their briefs, the petitioners argued § 4 of the 2011 Act,

(n)(1), strips the PHMSA of power to require
automatic or remote-controlled shut-off valves in gathering
pipelines under its general rulemaking authority. 

§ 60102(a)(2). During oral argument, however, counsel for
petitioners conceded the PHMSA retains this power. Any
argument to the contrary is, therefore, waived.
                               12
     The petitioners now press a narrower argument. They
claim the PHMSA needed to find automatic or remote-
controlled shut-off valves “appropriate” for new transmission
pipelines pursuant to § 4 before it could require them for
gathering pipelines. Now that the PHMSA has made that
finding under § 4, however, the petitioners concede the agency
may “use [its] general rulemaking authority tomorrow to go out
and do a rule” for gathering pipelines. In other words, they
argue that § 4 creates only a condition precedent, not a ban.

     Section 4 creates neither a condition precedent nor a ban.
As the petitioners themselves explain at length in their opening
brief, § 4 does not apply to gathering pipelines. Section 4 by its
plain terms applies only to “transmission pipeline facilities.”
We do not understand how § 4 could plausibly be read to create
a condition precedent for a different type of pipeline facility.
Nor do we see any reason to think the condition precedent, if
there were one, would have to be met through a separate
rulemaking. More to the point, we have no discretion to order
a separate rulemaking unless a statute clearly ordains a
particular sequence of procedures. See Perez v. Mortg. Bankers
Ass’n, 

 (2015) (“Agencies are free to grant
additional procedural rights in the exercise of their discretion,
but reviewing courts are generally not free to impose them if
the agencies have not chosen to grant them.” (quoting Vt.
Yankee Nuclear Power Corp. v. Nat. Res. Def. Council, Inc.,

 (1978))).

C. The PHMSA Failed To Observe the Rulemaking
   Procedures

     Although the PHMSA had the power to regulate, the
agency had to follow the hybrid rulemaking procedures laid out
in the APA, 

(c), and the pipeline safety laws. 

(b)(3)–(4). The petitioners argue the PHMSA
                                13
did not follow those procedures because the risk assessment
made available for peer review and public comment said
nothing about regulating gathering pipelines.

     Indeed, the notice of proposed rulemaking and
accompanying risk assessment were all about transmission
lines. They contained no data, analysis, or conjecture about the
costs and benefits of applying the proposed safety standard to
gathering facilities. The study by Oak Ridge, upon which the
agency relied to show the feasibility and benefits of the safety
standard, also addresses only transmission pipelines. The
proposed rule did not even make a finding that the standard
would be “appropriate” for gathering pipelines. Gathering
pipelines, it seems, were a mere afterthought when they
surfaced during the advisory committees’ deliberation.

     Both the pipeline safety laws and the APA require more.
Under the former, the PHMSA must consider “the
appropriateness of the standard for the particular type of
pipeline     transportation      or    facility.”     

(b)(2)(B). To be appropriate for a particular type of
pipeline facility, the standard must be “practicable,” and the
benefits must justify the costs. 

 § 60102(b)(1)(A), (5); cf.
Michigan v. EPA, 

 (2015) (“No regulation is
‘appropriate’ if it does significantly more harm than good.”).
The PHMSA, therefore, had to explain why the safety standard
is practicable and makes sense for regulated gathering lines, but
it made no effort to do so until issuing the final rule, when there
could be no peer review or public comment.

     That was a serious error. We have long held that, in order
to provide the public with a meaningful chance of participating
in the rulemaking process, as required by the APA, see 

(c), an agency must disclose critical information
justifying the proposal in time for public comment. See Owner-
                               14
Operator Indep. Drivers Ass’n, Inc. v. Fed. Motor Carrier
Safety Admin., 

 (D.C. Cir. 2007) (citing
cases). The procedures required by the pipeline safety laws are
more specific and still more demanding. As noted above, the
PHMSA must submit for peer review and make available for
public comment a risk assessment identifying “the costs and
benefits associated with the proposed standard.” 

(b)(3)(B), (4)(A). It must also “identify technical data
or other information upon which the risk assessment
information and proposed standard is based.” 

§ 60102(b)(3)(D). In sum, the risk assessment made available
for comment and peer review had to contain the technical and
cost-benefit information critically needed to justify the safety
standard for each type of pipeline transportation and facility.
The risk assessment did not comply with this requirement
because it said nothing about the practicability or the costs and
benefits of regulating the gathering sector of the pipeline
industry.

     The PHMSA tries to make something out of nothing, but
that is an impossible task. Cf. Kooritzky v. Reich, 

 (D.C. Cir. 1994) (“Something is not a logical outgrowth
of nothing.”).

     The agency first argues the analysis for transmission
pipelines was good enough. The petitioners do not dispute this.
The agency then points out the methodology and data for
transmission pipelines were exposed to comment and
refutation. Again, no dispute. Because it later used the same
methodology and data to calculate the costs for gathering
pipelines, the PHMSA argues, in essence, that the final risk
assessment merely “expands on and confirms data in the
rulemaking record,” which is permitted under our precedent.
Competitive Enter. Inst. v. Dep’t of Transp., 

, 920
                               15
(D.C. Cir. 2017); see also Chamber of Commerce v. SEC, 

 (D.C. Cir. 2006).

     That is not an accurate account of the final rule. For one,
the PHMSA’s estimate of the number of miles of gathering line
subject to the safety standard “was entirely new.” Owner-
Operator Indep. Drivers Ass’n, 

. That datum is
critical, as it determines the number of valves needed to comply
and hence the cost. More important still, the PHMSA had never
before exposed its methodological assumption that the analysis
and data regarding transmission lines would be equally
applicable to gathering lines. This was not a “minor
modification used to check or confirm prior analysis.” 

 If the
PHSMSA thought the information for transmission pipelines
was valid for gathering lines, then it should have said so in time
for peer review and public comment. By remaining silent about
this critical assumption until the final risk assessment, the
PHMSA sandbagged the advisory committees and the public,
sidestepping the process of public deliberation required by law.

     The PHMSA argues the final rule came as no surprise
because transmission and regulated gathering lines are treated
alike by default under preexisting rules. That is beside the
point. The petitioners do not dispute the rule was a logical
outgrowth of the proposal; they cheerfully concede they knew
regulated gathering lines would be regulated unless carved out.
Their gripe is with the agency’s failure to do an adequate risk
assessment in time for peer review and public comment. The
agency may not avoid these “troublesome rulemaking
procedures . . . simply by announcing its independence in a
general rule.” United States v. Picciotto, 

(D.C. Cir. 1989).

     The PHMSA argues it satisfied the requirements, at least
for Type A pipelines, because of its longstanding judgment that
                               16
Type A lines present “a significant enough risk in certain
circumstances to warrant the same amount of regulation as
transmission lines.” 71 Fed. Reg. at 13,296/3. The petitioners
do not quibble with this general proposition about a similarity
in risk, and it may have support in the record. Operating at 20%
of the SMYS may increase the risk of a rupture. See, e.g., 87
Fed. Reg. at 20,960/2 (“PHMSA is aware of data that would
indicate that pipelines operating at pressures lower than 20
percent of SMYS are at less risk of rupturing.”). Hoop stress is
also related to internal pressure and diameter, which increase
the rate of gas flow and thus the magnitude of the expected
harm of a rupture. See Oak Ridge at 8, 11. If a finding that a
Type A line presents a risk comparable to that of a transmission
line were enough, then perhaps the PHMSA would be in the
clear.

     A similarity in risk, operating pressures, or diameters,
however, does not mean the safety standard is practicable or
has similar benefits and costs when applied to a different sector
of the pipeline industry. For example, according to the
PHMSA’s risk assessment, the cost of the equipment needed to
comply with the final rule doubles when a pipeline operator
must upgrade from a manual valve rather than from a valve
already equipped with an “automating actuator” (a motor or
equivalent device, which still requires adding a
communications system to operate a valve remotely or
automatically). If gathering operators use actuators at a lower
rate than do transmission operators, then they would face
greater costs per mile to upgrade their equipment in order to
comply with the standard, regardless of similarity in risk,
operating pressure, or diameter.

    Finally, the PHMSA argues it actually did say something
about gathering pipelines in its proposal. Here the PHMSA
points to a Leak Detection Study, which it cited in the preamble
                               17
to the proposal, that discusses past accidents in both
transmission and gathering pipelines. The study expressly says,
however, it “does not address th[e] issue of shut-off valves.”
David Shaw et al., Leak Detection Study 2-2 (Dec. 2012),
https://perma.cc/SNG6-6GAQ; see also 87 Fed. Reg. at
20,945/2–3 (summarizing the report). A report that does not
address the relevant safety technology is not a starting point for
a risk assessment of the proposed standard.

     We conclude that, although the PHMSA was required by
law to identify the costs and benefits of requiring automatic or
remote-controlled shut-off valves for gathering pipelines in a
risk assessment of the proposed rule, it did not even attempt to
do so. By remaining mum, the PHMSA flouted the pipeline
safety laws and a cardinal rule of administrative law.

D. The PHMSA’s Procedural Error Is Prejudicial

     The petitioners have shown the PHMSA erred, but they
also must show the error is prejudicial to them, as we do not
right wrongs that make no difference. 

. To show
prejudice, the petitioners must raise a credible argument about
the merits of the rule. Owner-Operator Indep. Drivers Ass’n,
Inc., 

. They need not show the agency, had it
adhered to the procedural requirements of the law, “would have
reached a different result.” Chamber of Commerce, 

. They need only show they “had something useful to say.”

     We are convinced the petitioners do have something useful
to say to the PHMSA, and that they raise a credible argument
on the merits. Indeed, as we explain next, the petitioners have
shown the final rule was arbitrary and capricious, so they
necessarily have satisfied their burden of proof. See Owner-
Operator Indep. Drivers Ass’n, Inc., 494 F.3d at 202–03
                               18
(finding the procedural error prejudicial because the petitioners
also showed the agency did not reasonably explain the analysis
in the final rule).

E. The PHMSA            Failed      To   Make    a    Reasoned
   Determination

     Apart from following the statutory procedures, the
PHMSA was ultimately required to show regulating gathering
pipelines would be “appropriate.” 

(b)(2)(B).
To do so reasonably, it had to consider the relevant factors and
explain why regulating gathering pipelines would be
practicable, and show that the benefits would justify the costs.
It did not do so.

     The PHMSA claims it satisfied its statutory duty by
asserting in a footnote of the final risk assessment that
regulated gathering lines are similar to transmission lines “in
terms of design and operating characteristics, and risks to
public safety and the environment,” and therefore “similar
behavior by operators is expected.” It argues we must defer to
this “general analysis based on informed conjecture.” Chamber
of Commerce v. SEC, 

 (D.C. Cir. 2005).
Because we conclude the PHMSA’s conjecture was not
“informed,” we do not defer to it.

     At best, the record shows regulated gathering lines and
transmission pipelines pose a comparable risk. As we have
already explained, however, this does not mean gathering
pipelines are similar in all important respects. Indeed, the
petitioners point out the PHMSA failed to consider how
gathering pipelines are different from transmission pipelines in
a number of other respects. For example, they submit
declarations with their opening brief stating that regulated
segments of gathering line are typically short and regulated
                               19
only intermittently, which they argue will make compliance far
more difficult and expensive than the PHMSA recognized. The
PHMSA never considered this aspect of the problem during the
rulemaking and never disputed the merits of this argument in
its brief, so we assume it is true. During oral argument, counsel
for the PHMSA did say the agency could take these difficulties
into account later because the rule authorizes site-specific
exemptions when an operator shows compliance is not feasible.
See 

(g); 195.258(e). Because this
argument would impermissibly shift the burden of proof to the
petitioners and other operators, we must reject it.

     The operators point to several other differences between
gathering and transmission lines. They tell us, for example, that
transmission pipelines are run from “evolved control rooms”
that complement automatic or remote-controlled shut-off
valves, whereas the gathering sector is made up of smaller
players that rely more upon on-site personnel. They also tell us
gas transmission operators behave differently because they are
price-regulated public utilities, while the gas gathering sector
relies upon market prices to recover costs. Although we cannot
fully evaluate the importance of these asserted differences
precisely because the agency failed to develop an adequate
administrative record in time for comment, they surely seem
relevant to the agency’s decision making, and at a minimum
show the agency’s procedural error was prejudicial.

     The PHMSA concedes there may be a difference in cost.
In particular, it concedes gathering operators may install
automatic or remote-controlled shut-off valves at lower rates,
or not at all. By contrast, many transmission operators already
install compliant valves, and their high baseline rate of
compliance means they will incur little additional cost for
equipment. A lower baseline rate of compliance for gathering
pipelines, therefore, “could in turn increase compliance costs
                               20
for those lines,” as the PHMSA conceded in a footnote to its
final risk assessment. This would seem a relevant factor as
well. The PHMSA nonetheless downplayed its significance on
the ground that few gathering pipelines will be subject to the
rule, so the increase in the total cost of the final rule will be
small and, if the baseline rate of compliance is lower for
gathering lines, then “the benefits of the rulemaking would be
higher.”

     We are not reassured. The relevant question under the law
is whether the benefits of regulating gathering lines justify the
costs, and that question cannot be answered by comparing costs
against costs. As to the “higher” benefits, we would perhaps
find this assertion more persuasive had the agency made the
requisite “thorough” assessment of the benefits it claims to
have made. Far from finding the assessment “thorough,”
however, we find it inadequate, and conclude the agency failed
to make a “reasoned determination that the benefits, including
safety and environmental benefits, of the intended standard
justify its costs.” 

(b)(5).

     The risk assessment does not quantify any of the benefits
of the standard. This is troubling enough, as a reasoned
decision would explain why any unquantified benefits cannot
reasonably be quantified. 

 § 60102(b)(2)(D), (b)(5); see also
Bus. Roundtable v. SEC, 

 (D.C. Cir. 2011)
(agency “failed adequately to quantify the certain costs or to
explain why those costs could not be quantified”). Quantifying
benefits always requires making projections, so it is no answer
to say “a detailed projection of avoided incidents and avoided
costs is not available.” The agency even conceded “it could be
possible [to] estimate the benefit of this rule for hazardous
pipelines,” and still, it did not do so. Without quantified
benefits to compare against costs, it is not apparent just how
                               21
the agency went about weighing the benefits against the costs.

(b)(5).

     Even the qualitative discussion of the benefits does not say
anything about gathering pipelines. The risk assessment relies
upon estimates of avoided damages modeled in the Oak Ridge
study but, as mentioned above, that study by its terms does not
cover gathering pipelines. The PHMSA now argues the same
results and hypothetical pipeline configuration would be
representative for gathering pipelines of a similar pressure and
diameter, but that is not obvious; the study relies upon a
“hypothetical” pipeline configuration where, among other
things, “pump stations are located at 100 mile intervals along
the pipeline.” Oak Ridge at 34, 150. That may be typical of
long-distance transmission pipelines, but it would seem highly
unusual among gathering pipelines. Be that as it may, the
agency did not make this point in the administrative record so
for us to consider it would “contradict[] the foundational
principle of administrative law that a court may uphold agency
action only on the grounds that the agency invoked when it
took the action.” Michigan, 

 (citing SEC v.
Chenery Corp., 

 (1943)).

     Because the PHMSA invites us to rely upon the estimates
of avoided damages in the Oak Ridge study, we note two
additional problems lest the agency overlook them in any
further rulemaking applying the safety standard to gathering
lines.

     First, the avoided damages touted by the agency ignore the
probability of a rupture. Oak Ridge modeled avoided damages
“based on the premise that the releases occur (100% failure
likelihood),” and, to boot, assuming the worst type of rupture—
a “guillotine” break that slices a pipeline, exposing the entire
diameter of the pipe. Oak Ridge at 34, 150. As Oak Ridge
                                22
carefully noted, however, the benefit of the technology is also
“a function” of the “probability of failure,” and the probability
of a rupture, let alone the worst type of rupture, “is low.” Id. at
33. In order to identify the expected benefits of avoiding a
rupture, a rational analysis would have to consider the
probability of a rupture, not just the magnitude of the harm
avoided. Cf. Cass Sunstein, Worst-Case Scenarios 2 (2007)
(“People who are sensible, or even sane, do not treat a 1 percent
risk of loss the same as a certainty of a loss.”). The agency
keeps detailed data on the frequency of pipeline accidents, so
the limitations of the Oak Ridge study do not excuse the
agency’s failure to consider the low probability of a rupture
when reporting avoided costs.

     Second, and more important still, the study does not help
justify the standard for gas gathering pipelines. Far from it.
According to Oak Ridge, the cost-benefit ratio for automatic or
remote-controlled shut-off valves on gas pipelines is generally
“negative” because most of the severe damage from a gas fire
happens “before valve closure can isolate the damaged pipeline
and begin limiting the amount of natural gas that escapes and
burns.” Oak Ridge at 181. Even assuming the worst type of
rupture occurs with certainty, id. at 34, Oak Ridge found no net
safety benefit in any scenarios it modeled for a gas pipeline
(even for a high-pressure, large-diameter gas transmission
pipeline located in a densely populated area) unless the valves
close within “10 minutes or less after the break.” Id. at 184.

     The final rule gives operators 30 minutes to close the
valves “measured from an operator’s identification of a rupture
after notification of a potential rupture,” which may be well
after the rupture occurred. 68 Fed. Reg. at 20,941/3, codified at

(b). As the PHMSA itself observed,
“potential property damages are likely to be largely complete
before the 30-minute limit.” The PHMSA says it expects
                               23
operators will overcomply, but it does not explain why or by
how much, or whether they will usually be able to shut off
valves within 10 minutes after a break. Therefore, even if the
results of the Oak Ridge study applied to gathering pipelines,
the agency fails adequately to explain why the study supports
the agency’s position on the merits.

    Considering asserted differences between transmission
and gathering pipelines with respect to their operations and the
cost of compliance; the PHMSA’s failure to quantify any
benefits; its weak qualitative analysis; and the record as a
whole, we conclude the agency has not reasonably explained
why the rule is appropriate for gathering pipelines.

F. The Remedy Is a Limited Vacatur

     That leaves the question of remedy. The petitioners argue
for vacatur of the rule only as it applies to gathering pipelines.
The PHMSA agrees with the proposed remedy, asking us,
should we find a prejudicial error, to “sever the portion of the
rule applicable to transmission lines and vacate the rule only as
applied to gathering lines.” Although the text of the final rule
is not divisible in this way, we may “invalidate only some
applications even of indivisible text.” Nat. Res. Def. Council v.
Wheeler, 

 (D.C. Cir. 2020). We therefore vacate
the rule in its entirety as it applies to gathering pipeline
facilities.

                        III. Conclusion

    As the Supreme Court has said, “the Government should
turn square corners in dealing with the people.” Dep’t of
Homeland Sec. v. Regents of the Univ. of Cal., 

 (2020). The PHMSA did not turn square corners here. It
cut corners to the prejudice of the petitioners, the
                               24
administrative process, and thus the public. We therefore grant
the petition for review and vacate the final rule as it applies to
gathering pipeline facilities.
                                                     So ordered.