Response to Comments:
              Fuel Economy Labeling of Motor Vehicles
&EPA
United States
Environmental Protection
Agency
Office of Transportation and Air Quality
           EPA420-R-06-016
            December 2006

-------
                           Response to Comments:
                Fuel Economy Labeling of Motor Vehicles
                               Assessment and Standards Division
                                         and
                            Certification and Innovative Strategies Division

                              Office of Transportation and Air Quality
                              U.S. Environmental Protection Agency
v>EPA
United States                                         EPA420-R-06-016
Environmental Protection                                    December 2006
Agency

-------
                                   Table of Contents


LIST OF COMMENTING ORGANIZATIONS AND ABBREVIATIONS	iv

  COMMENTS SUBMITTED DURING THE COMMENTPERIOD	iv
  COMMENTS SUBMITTED AFTER THE CLOSE OF THE COMMENT PERIOD	vi

PUBLIC HEARING TESTIMONY	vii

INTRODUCTION	1

CHAPTER 1:   GENERAL ISSUES	2

  1.1     GENERAL COMMENTS SUPPORTINGPROPOSAL	2
  1.2     GENERAL COMMENTS OPPOSING PROPOSAL	3
  1.3     ESTIMATES ARE BASED ON CONTROLLED, REPEATABLE TESTS	4
  1.4     THERE is No PERFECT TEST	5
  1.5     CAFEPROGRAM	5
  1.6     SHOULD LABEL ESTIMATE COVER 50 OR 75 PERCENTILE?	7
  1.7     PUBLIC AVAILABILITY OF FUEL ECONOMY TEST DATA	8
  1.8     OVERALL PURPOSE OF FUEL ECONOMY ESTIMATES	8
  1.9     SMALL VOLUME MANUFACTURERS	9
  1.10   TYPOGRAPHICAL ERRORS AND REGULATORY REFERENCES	10
  1.11   MISCELLANEOUS REGULATORY COMMENTS	12
  1.12   DAYTIME RUNNING LIGHTS	13

CHAPTER 2:   LABEL FORMAT AND CONTENT	14

  2.1     GENERAL LABEL PREFERENCES	14
  2.2     LABELFORMAT EFFECTIVE DATE	17
  2.3     LABEL CONTENT ISSUES	19
    2.3.1   Fuel Economy of Comparable Vehicles.	19
    2.3.2   Estimated Annual Fuel Cost	23
    2.3.3    Combined Fuel Economy Calculation	25
    2.3.4   Range of Expected Fuel Economy	26
    2.3.5    "YourMileage Will Vary" Statement	27
    2.3.6    Graphical Updates and Government Logos	28
    2.3.7   Environmental Information on Fuel Economy Labels	29
    2.3.8    Vehicle Descriptor Information	31
    2.3.9    Website Reference	32
    2.3.10     Transition Language	33
    2.3.11     City and Highway Numbers	34
  2.4     LABEL SIZE AND ORIENTATION ISSUES	34
  2.5     IMPLEMENTATION OUTREACH	35
  2.6     COMPARABLE CLASS DESIGNATIONS	36
  2.7     LABELING REQUIREMENTS FOR DUAL FUELED VEHICLES	38

CHAPTERS:   IMPLEMENTATION ISSUES	40

  3.1     IN-USE VALIDATION OF 5-CYCLE COEFFICIENTS	40
  3.2     PERIODIC REVIEW AND UPDATING OF THE MPG-BASED CURVES	44
  3.3     USE OFBAGDATAFACTORS IN THE 5-CYCLE EQUATIONS	45
  3.4     FLEXIBILITY TO APPLY THE MPG-BASED ADJUSTMENTS AT THE MODEL TYPE LEVEL	46
  3.5     CRITERIA FOR ADDITIONAL 20F / SFTP TESTING	47
  3.6     ANALYTICALLY DERIVED FUEL ECONOMY	49
  3.7     CONSUMER EDUCATION ISSUES	50
  3.8     GOVERNMENT-INDUSTRY WORKING GROUPS	51
  3.9     WHEN CITY FUEL ECONOMY is GREATER THAN HIGHWAY FUEL ECONOMY	51
  3.10   TECHNOLOGY-SPECIFIC, MODEL-SPECIFIC, OR MANUFACTURER-SPECIFIC ADJUSTMENT FACTORS	52

-------
  3.11    USE OF FUEL ECONOMY ESTIMATES IN ADVERTISING	55

CHAPTER 4:   OTHER RELATED PROPOSALS	56

  4.1     VOLUNTARY FUEL ECONOMY LABELING FOR VEHICLES EXCEEDING 8500 POUNDS GROSS VEHICLE
  WEIGHT RATING	56
  4.2     ELECTRONIC DISTRIBUTION OF DEALER-SUPPLIED FUEL ECONOMY BOOKLET	59
  4.3     CONSIDERATION OF FUEL CONSUMPTION vs. FUEL ECONOMY AS A METRIC	60
  4.4     WEB-BASED DRIVER-SPECIFIC FUEL ECONOMY CALCULATOR	62

CHAPTER 5:   CITY AND HIGHWAY FUEL ECONOMY ESTIMATES AND METHODOLOGY	64

  5.1     THE PROPOSED FORMULAE ARE COMPLEX	64
  5.2     VALIDATION OF THE 5-CYCLE FORMULAE FOR INDIVIDUAL VEHICLES	65
  5.3     SOME FORMULAE  COEFFICIENTS ARE COUNTER-INTUITIVE	73
  5.4     START FUEL USE MAY NOT BE ACCURATE ON PER VEHICLE BASIS	78
  5.5     RELATIVE ACCURACY OF THE MPG-B ASED EQUATIONS VERSUS THE 5 -CYCLE FORMULAE FOR
  INDIVIDUAL VEHICLES	79
  5.6     MPG-BASED OR "GENERIC" ADJUSTMENTS TO FTP ANDHFET	82
  5.7     THE THREE NEW TEST CYCLES ARE EXTREME	83
  5.8     WEIGHTING OF NEW CYCLES	86
  5.9     INTERPOLATION OF FUEL ECONOMY USING FIVE CYCLES	87
  5.10    MISUSE OF EMISSION FACTOR MODELS	99
  5.11    DERIVATION OF VEHICLE-SPECIFIC S-CYCLE METHOD	102
     5.11.1     Start Fuel Use	102
       5.11.1.1    Start Fuel	102
       5.11.12    Trip Length	104
       5.11.1.3    Formula for Start Fuel Use	109
     5.11.2    Running Fuel Use.	110
       5.11.2.2    Representative Mix of Dynamometer Driving Cycles	110
     5.11.3    Effect of Air Conditioning on Fuel Economy	116
       5.11.3.1    Approximating Excess Fuel Use Due to Air Conditioning	116
       5.11.3.2    Available In-Use Data and Analysis	118
       5.11.3.3    Driver Behavior	120
       5.11.3.4    Accounting for Energy Reduction Technologies	121
       5.11.3.5    Temperature Impact on Compressor Power	124
       5.11.3.6    Estimates of National A/C and Defroster Use	125
     5.11.4    Effect of Cold Temperatures on Fuel Economy	126
     5.11.5    Adjustment Factor for Non-Dynamometer Effects	129
     5.11.6     mpg Approach Equation Tolerances	131
  5.12    ONRO AD FUEL ECONOMY ESTIMATES BY OTHER ORGANIZATIONS	135

CHAPTER 6:   TESTING PROVISIONS	143

  6.1     GENERAL TEST PROCEDURE ISSUES	143
  6.2     TEST REQUIREMENTS FOR VEHICLES CURRENTLY EXEMPT FROM CERTAIN EMISSIONS TESTS	145
     6.2.1   Diesel Vehicles	145
     6.2.2   Alternative-Fueled Vehicles	147
  6.3     MODIFICATIONS TO EXISTING TEST PROCEDURES	147
     6.3.1   Revisions to US06 Bag Measurements.	147
     6.3.2   Heater/Defroster Use During the Cold FTP.	150
     6.3.3   4-Phase, 4-Bag FTP for Gasoline-Electric Hybrid Vehicles	152
  6.4     TESTFUELS	156
  6.5     HYBRID-ELECTRIC VEHICLE BATTERY STATE OF CHARGE	157

CHAPTER 7:   COST ANALYSIS	159

  7.1     TESTINGBURDEN	159
     7.1.1   Testing Burden for Small Manufacturers.	159
     7.1.2   2008- 2010Model Year Testing Costs: Testing Issues	160
     7.1.3   2011 and Later Model Year Testing Costs: Inclusion ofMDPVs	161

-------
7.2 FACILITIES COSTS	162
   7.2.7 Facility Upgrades for Cold FTP Testing	162
7.3 STARTUP BURDEN	163
   7.3.1 Dual Information Systems for CAFE/Gas Guzzler and Label Calculations	163
   7.3.2 Additional Startup Costs	164

-------
List of Commenting Organizations and Abbreviations
The abbreviations and annotations in this list are used throughout this document to refer
to various organizations that submitted written and/or oral comments on the proposed
rulemaking.  We also received comments from more than 2,800 members of the public.

Comments Submitted During the Comment Period
Organization
Alliance of Automobile Manufacturers
American Automobile Association
American Council for an Energy-Efficient
Economy
American Honda Motor Co., Inc.
Association of International Automobile
Manufacturers
Bluewater Network
Consumers Union
DaimlerChrysler
Delphi LLC
Environmental Defense
Ferrari S.p.A.
Fuji Heavy Industries USA, Inc. (Subaru of
America)
General Motors Corporation
Mitsubishi Motors
Montana Coalition for Health, Environmental &
Economic Rights
Abbreviation
AAM
AAA
ACEEE
Honda
AIAM
Bluewater
CU
OCX
Delphi
ED
Ferrari
Subaru
GM
Mitsubishi
Montana
CHEER
Docket Document Number
EPA-HQ-OAR-2005-01 69-01 1 0
EPA-HQ-OAR-2005-01 69-01 10.1
EPA-HQ-OAR-2005-01 69-0105
EPA-HQ-OAR-2005-01 69-01 05.1
EPA-HQ-OAR-2005-01 69-01 1 9
EPA-HQ-OAR-2005-01 69-01 19.1
EPA-HQ-OAR-2005-01 69-01 1 9.2
EPA-HQ-OAR-2005-01 69-01 1 5
EPA- HQ-OAR-2005-0 1 69-01 15.1
EPA-HQ-OAR-2005-01 69-01 1 0
EPA-HQ-OAR-2005-01 69-01 10.1
EPA-HQ-OAR-2005-01 69-01 00
EPA-HQ-OAR-2005-01 69-01 37
EPA-HQ-OAR-2005-01 69-01 21
EPA-HQ-OAR-2005-01 69-01 21.1
EPA-HQ-OAR-2005-01 69-01 16
EPA-HQ-OAR-2005-0169-0116.1
EPA-HQ-OAR-2005-01 69-01 24
EPA-HQ-OAR-2005-0169-0124.1
EPA-HQ-OAR-2005-01 69-01 38
EPA-HQ-OAR-2005-0169-0112.1
EPA-HQ-OAR-2005-01 69-01 14
EPA-HQ-OAR-2005-0169-0114.1
EPA-HQ-OAR-2005-0169-0114.2
EPA-HQ-OAR-2005-01 69-01 06
EPA-HQ-OAR-2005-01 69-01 06.1
EPA-HQ-OAR-2005-01 69-01 25
EPA-HQ-OAR-2005-0169-0125.1
EPA-HQ-OAR-2005-01 69-0072
                                    IV

-------
Organization
National Automobile Dealers Association
National Renewable Energy Laboratory
Natural Resources Canada, Transportation
Energy Use Division
Natural Resources Defense Council
New York State Department of Environmental
Conservation
Nissan Motor Company Limited
PPG Industries
Public Citizen
Toyota Motor Corporation
Union of Concerned Scientists
Volkswagen of America, Inc.
Abbreviation
NADA
NREL
NRC-TEUD
NRDC
NYDEC
Nissan
PPG
PC
Toyota
UCS
VW
Docket Document Number
EPA-HQ-OAR-2005-01 69-01 31
EPA- HQ-OAR-2005-01 69-01 31.1
EPA-HQ-OAR-2005-01 69-01 31 .2
EPA-HQ-OAR-2005-01 69-01 31 .3
EPA- HQ-OAR-2005-01 69-01 31 .4
EPA-HQ-OAR-2005-01 69-01 08
EPA-HQ-OAR-2005-01 69-01 08.1
EPA-HQ-OAR-2005-01 69-01 09
EPA-HQ-OAR-2005-01 69-01 09.1
EPA-HQ-OAR-2005-01 69-01 1 7
EPA-HQ-OAR-2005-01 69-01 17.1
EPA-HQ-OAR-2005-01 69-01 35
EPA- HQ-OAR-2005-01 69-01 28
EPA-HQ-OAR-2005-0169-0128.1
EPA- HQ-OAR-2005-01 69-0078
EPA-HQ-OAR-2005-01 69-0078.1
EPA-HQ-OAR-2005-01 69-01 1 3
EPA-HQ-OAR-2005-01 69-01 13.1
EPA-HQ-OAR-2005-01 69-01 30
EPA-HQ-OAR-2005-0169-0123.1
EPA- HQ-OAR-2005-01 69-01 28
EPA-HQ-OAR-2005-0169-0128.1
EPA-HQ-OAR-2005-01 69-01 1 8
EPA-HQ-OAR-2005-01 69-01 18.1
EPA-HQ-OAR-2005-01 69-01 36
EPA- HQ-OAR-2005-01 69-01 27
EPA-HQ-OAR-2005-0169-0127.1
EPA- HQ-OAR-2005-01 69-01 28
EPA-HQ-OAR-2005-0169-0128.1
EPA- HQ-OAR-2005-01 69-01 20

-------
Comments Submitted after the Close of the Comment Period
Organization
American Honda Motor Co., Inc.
Ford Motor Company
Alliance of Automobile Manufacturers
Selle DorTse LLC
Bluewater Network- compendium of emails
from J. Wilson et al.
Abbreviation
Honda
Ford
AAM

Bluewater
Docket Document Number
EPA- HQ-OAR-2005-01 69-01 43
EPA- HQ-OAR-2005-01 69-01 41
EPA- HQ-OAR-2005-01 69-01 42
EPA-HQ-OAR-2005-0169-0142.1
EPA-HQ-OAR-2005-0169-0142.2
EPA- HQ-OAR-2005-01 69-01 40
EPA- HQ-OAR-2005-01 69-01 39
                                  VI

-------
Public Hearing Testimony
Mantill Williams, Director of Public Affairs
American Automobile Association (AAA)

Ed Tonkin, Chairman of the Regulatory Affairs Committee
Andy Koblenz, Vice President and General Counsel
National Automobile Dealers Association (NADA)

Giedrius Ambrozaitis,
Alliance of Automobile Manufacturers

David Raney, Senior Manager of Environmental and Energy Affairs
John German
American Honda Motor Company

Dr. John DeCicco
Environmental Defense

Dr. Russell Long, Founder
Bluewater Network
Vice President, Friends of the Earth U.S.
                                     VII

-------
Introduction

      On February 1, 2006, we published a Notice of Proposed Rulemaking (NPRM)
that proposed changes to the testing and calculation procedures for fuel economy
estimates that are posted on the window stickers of all new cars and light trucks sold in
the United States.  In addition to proposing a new calculation methodology that would
bring the estimates closer to the real-world experience of consumers, we proposed
revisions to the design and content of the label, as well as to some of the information
that appears on it.

      We held a public hearing on the NPRM  in Romulus, Michigan on March 3, 2006.
At that hearing, oral comments on the NPRM were received and recorded. A written
comment period remained open until April 3, 2006.  A complete list of organizations that
provided comments on the NPRM is shown in the following table, along with
abbreviations for these organizations that are used throughout this document.  In
addition to the organizations shown in the following table, we received comments from
many individual stakeholders. All comments and hearing testimony have been placed
in the docket for this rulemaking. Publicly available docket materials are available either
electronically through www.regulations.gov or in hard copy at the Air and Radiation
Docket,  EPA/DC, EPA West, Room B102, 1301 Constitution Ave., NW, Washington,
DC.  See Docket ID No. EPA-HQ-OAR-2005-0169. The Public Reading Room is open
from 8:30 a.m. to 4:30 p.m., Monday through Friday, excluding legal  holidays. The
telephone number for the Public Reading Room is (202) 566-1744, and the telephone
number for the Air and Radiation Docket is (202) 566-1742.

      This Response to Comments contains a detailed summary of all comments we
received on the NPRM as well as our analysis  of each comment and our response.  In
many cases this document also summarizes what we proposed in the NPRM, but we
can not repeat the same level of detail that was presented in the NPRM. For that
reason we refer readers to the NPRM for detailed descriptions of what we proposed.
The reader should also refer to the final rulemaking notice in the Federal Register and
to the Final Technical Support Document.

-------
Chapter 1:    General Issues

1.1    General Comments Supporting Proposal

What commenters said:

      We received general support from many stakeholders, including automotive
dealers, states, automotive manufacturers, environmental organizations, consumer
groups, and individual consumers. Supporting comments often noted the growing
dissatisfaction with the frequent gap between the EPA label values and real-world
experience. All agreed that consumers deserve the most useful, accurate, and up-to-
date information in order to make educated purchasing decisions. Auto companies
tended to support the general improvements and the conclusion that the new
procedures will result in more accurate fuel economy information, but they expressed
some reservations regarding specific details of the program. We address the specific
comments from the auto industry later in this document.  Comments of this general
nature came from Toyota, DaimlerChrysler, GM, Mitsubishi, Nissan, NADA, NYDEC,
AAA, PC, AAM/AIAM, PPG, NRDC, UCS, many citizens, and multiple anonymous
commenters.

      Bluewater Network stated that the evidence demonstrates "conclusively that
there are staggering inconsistencies between the mileage motorists are actually
achieving, and the values on the labels." They commend EPA "for advancing a strong
proposal to attempt to rectify this problem." Despite their support, they note several
areas which they believe must be strengthened.

      AAA commissioned a study by the Automobile Club of Southern California's
Automotive Research Center. This study tested a number of the most popular vehicles
on the US06 test.  They also collected in-use fuel economy data from more than 40 of
these vehicles. They argue that their results both confirm the shortcomings of the
existing fuel economy labeling tests and indicate that EPA"... is on the right track with
its proposal..."

      Several comments implicitly support our  proposal by virtue of commenting
negatively on the current label's reflection of reality. For example, one commenter
suggested that "EPA allows automobile manufacturers to mislead and defraud
unsuspecting consumers by placing faulty mileage estimates on automobile stickers."
Another stated that "EPA has been hoodwinking american  [sic] consumers for years..."

Our response:

      We agree with and appreciate the general comments of support.  The NPRM
outlined in detail the deficiencies of the current methods and the resulting gap between
the current fuel economy estimates and what consumers are achieving in real-world
driving. We continue to believe that the case for action is strong, and that the proposal
advanced a technically sound approach for dramatically improving the accuracy of the

-------
EPA fuel economy estimates. After consideration of all the comments, we are finalizing
a new program that in many respects matches the basic framework of the proposed
program. However, we are also addressing numerous public comments in the final
program, as discussed throughout this document.
1.2   General Comments Opposing Proposal

What commenters said:

      Only one comment specifically suggested taking no action at all.  His concern is
that the new methods will discourage consumers from considering hybrid vehicles.  He
suggests no action or simply reducing the current labels by a flat 15 percent for all
vehicles.  Environmental Defense urged EPA to proceed with caution, citing conflicting
sources of information regarding the magnitude of the discrepancy between the EPA
label values and real-world values.

Our response:

      As noted above under Section 1.1,  we continue to believe that the case for taking
action is strong, and consequently we are finalizing a program essentially identical to
the proposed approach.  We do not agree with  comments that suggest a flat one-size-
fits-all adjustment factor. Our analysis in the NPRM clearly showed a large amount of
vehicle-to-vehicle variation over the US06, SC03, and Cold FTP tests.  These findings
call into question the appropriateness of the continued use of "one-size-fits-all"
adjustment factors.  These additional emission  test cycles incorporate several critical
factors that are present in real-world driving, and that can have a significant impact on
fuel economy. Use of these test cycles will result in fuel economy estimates that are
essentially tailored to each vehicle's response to the impacts of each test cycle (higher
speeds, more aggressive driving, air conditioning, and cold temperatures).

      We do not believe that the new approach will discourage consumers from
purchasing or considering hybrid vehicles.  While it is true that many vehicles that get
high fuel economy, including hybrids, will see a greater reduction in the label values
than  lower mpg vehicles, we do not believe that this will discourage consumers from
considering those high-mpg vehicles. High mpg vehicles will be more sensitive to
changes in driving conditions for two reasons.  One, because they use relatively little
fuel in the first place, any increase in fuel consumption will be relatively larger  in
percentage. Two, because of the non-linearity of fuel  economy (distance per unit of fuel
volume, e.g. miles per gallon) with respect to fuel consumption (unit of fuel volume per
distance,  e.g. gallons per 100 miles), a fixed increase  in fuel consumption (e.g., due to
air conditioning) will  lower the fuel economy of a high mpg vehicle much more than it will
lower the fuel economy of a low mpg vehicle. For example, the fuel consumption
increase associated  with a 35 mpg rating that actually achieves 30 mpg in the real-world
is the same as a 15  mpg rating that actually achieves  14 mpg.  In addition, we  have no
interest in perpetuating label values that bear little relationship to reality.  Even with the

-------
adjusted label values, hybrid vehicles will remain among the most fuel-efficient vehicles
available in the marketplace.  Hybrids may well be the most significant powertrain
technology innovation driven to market commercialization primarily because of its fuel
economy potential. In addition, the nature of hybrid technology (the addition of a battery
as a second source of on-board power,  sophisticated control systems, sometimes a
smaller engine) suggests that fuel economy will likely be more sensitive to certain
conditions such as high acceleration and deceleration rates, cold ambient temperatures,
etc. Finally, by industry standards, hybrids are a relatively young technology,  and there
is every reason to believe that as the technology matures, hybrid vehicle fuel economy
will become much more robust over a broader range of driver behavior and climate
conditions.
1.3   Estimates Are Based on Controlled, Repeatable Tests

What we proposed:

      The methodology we proposed was based on the use of controlled, repeatable
tests conducted in emission and fuel economy testing facilities that conform to EPA's
stringent regulations on test procedures.

What commenters said:

      Several supporting comments further expressed agreement with EPA's
assessment in the NPRM that it is essential that the fuel economy estimates be based
on controlled, repeatable laboratory tests that provide manufacturers with a level playing
field. Some of these commenters also agreed that the auto manufacturers should
remain the primary parties responsible for perform ing the fuel economy testing  and
calculation of the mpg estimates.

Our response:

      It is essential that our fuel economy estimates continue to be derived from
controlled, repeatable, laboratory tests. We noted in the NPRM that these laboratory
tests provide a level playing field for all vehicles, which is essential for comparing the
fuel economy of one vehicle to another. EPA and manufacturers  test over 1,250 vehicle
models annually and every test is run under identical conditions and under a precise
driver's trace, which assures that the result will be the same for an individual vehicle
model no matter when and where the laboratory test is performed.  Finally, EPA must
preserve the ability to confirm the values achieved by the manufacturers' testing, and
this can only be achieved with a highly repeatable test or set of tests.

      However, the EPA fuel economy test methods need to reflect real world
conditions as well as being a repeatable test. While some organizations have issued
their own fuel economy numbers based on on-road driving, their approach introduces  a
wide number of variables - different drivers, driving patterns, weather conditions,

-------
temperatures, etc. -that make repeatability, and therefore vehicle-to-vehicle
comparisons, impossible.  The new fuel economy test methods are more representative
of real-world conditions than the current fuel economy tests, yet we retain in this final
rule our practice of relying on controlled, repeatable, laboratory tests.
1.4   There is No Perfect Test

What we proposed:

      We emphasized in the NPRM that it is impossible to design a "perfect" fuel
economy test that will provide accurate real-world fuel economy estimates for every
consumer.

What commenters said:

      Some comments expressed agreement with the principle that there is no perfect
test that can accurately predict the fuel economy for every driver under all
circumstances, and that there will inevitably be times when a driver's actual fuel
economy will be higher or lower due to differences in vehicle use, driving styles, and
conditions.  For example, AAA noted that because people's driving habits vary
significantly, there is no single test that will yield 100% accuracy.

Our response:

      We agree with the comments, and will continue to stress this fundamental point.
It is important to emphasize that fuel economy varies from driver to driver for a wide
variety of reasons, such as different driving styles, climates, traffic patterns, use of
accessories, loads, weather, number of passengers, and vehicle maintenance.  Even
different drivers of the same vehicle will experience different fuel economy as these and
other factors vary. Therefore, it is impossible to design a "perfect" fuel economy test
that will provide accurate fuel economy estimates for every consumer. With any
estimate, there will always be consumers that get better or worse actual fuel economy.
The EPA estimates are meant to be a general guideline for consumers, particularly to
compare the relative fuel economy of one  vehicle to another. Nevertheless, we do
believe that the new fuel economy test methods will do a better job of giving consumers
a more accurate estimate of the fuel economy they can achieve in the real-world.
1.5   CAFE Program

What we proposed:

      We noted in the NPRM that the proposal would not impact the test procedures,
driving cycles, measurement techniques, or the calculation methods used to determine
fuel economy values for CAFE compliance calculations.

-------
What commenters said:

      A number of commenters addressed the connection between the CAFE program
and the fuel economy label estimates. NADA stressed that it is important that EPA's
action not have an impact on CAFE.  A number of others expressed disagreement with
this opinion, however. The NYDEC recommended that the CAFE test methods be
updated to match the proposed revised methodology. Their fear, in particular,  is that
"EPA's efforts to inform the public about fuel economy will be thwarted by the use of
different standards for the vehicle labels than those used for [CAFE]..."  In agreement,
M. R. King put it this way: "If the window fuel economy sticker is not accurate, why
should the CAFE  calculation continue to be inaccurate?" If the CAFE test methods
remain fixed, NYDEC requests that the CAFE value assigned to the vehicle be put on
the label.

      Bluewater  Network notes that they have observed "a great deal of confusion"
over the several different sets of fuel economy values published by EPA and NHTSA
(CAFE values, EPA "adjusted" values, and EPA "unadjusted" values). For this reason,
they suggest that EPA end any public dissemination of the unadjusted EPA values,
including in the EPA Trends Report. They find "no compelling justification or public
good" that results from these values, and absent a legal mandate to make these values
public,  they urge us to not make them publicly available.

Our response:

      Our final rule does not alter the FTP and HFET driving cycles, the measurement
techniques or the calculation methods used to determine CAFE.  EPCA requires that
CAFE for passenger automobiles be determined from the EPA test procedures in place
as of 1975 (or procedures that give comparable results), which are  the city and highway
tests of today, with a few small adjustments for minor procedural changes that have
occurred since 1975.  The new method for calculating fuel economy label estimates fall
under regulations that are separate from the CAFE regulations.

      EPCA requires that CAFE for passenger automobiles be determined from the
EPA test procedures in place as of 1975 (or procedures that give comparable results),
which are the city and highway tests of today, with a few small adjustments for minor
procedural changes that have occurred since 1975.1 Today's final rule will not impact
the CAFE calculations.2

      EPA notes the comment by Bluewater Network regarding the release of
unadjusted fuel economy values and will focus future EPA Fuel Economy Trends
reports on information that is useful to consumers.
     See 49 U.S.C. 32904(c).
     See 49 U.S.C. 32904(c).

-------
1.6   Should Label Estimate Cover 50th or 75th Percentile?

What we proposed:

      We proposed to continue to set the city and highway mpg estimates at the
average, or mean, level, consistent with analytical approaches of the past. However,
we expressed our understanding that many drivers expect to achieve or exceed the fuel
economy indicated by these average mpg estimates. By continuing to set the estimates
at the average level, by definition, the average driver will achieve of exceed the fuel
economy label values fifty percent of the time. We invited comment on whether the city
and highway estimates should be set a level that is lower than average- for example,
to ensure that the average driver will achieve or exceed the label value 75 percent of
the time.

What commenters said:

      AAM/AIAM supported EPA's position in the NPRM, which retains the existing
practice of setting the  city and highway fuel economy label estimates at the average,  or
mean, level. AAM/AIAM note that an alternative, such as that suggested above, would
amount to applying an arbitrary discount to the mean value produced from 5-cycle
testing, in addition to the proposed 11% downward adjustment for non-dynamometer
effects, and would add no value to the label as a useful tool for comparative purposes.
The also believe that consumers understand that fuel economy label values represent
the average fuel economy experienced by drivers. Rather than arbitrarily setting a level
that is lower than average to ensure more driver coverage, AAM/AIAM suggest that a
more prudent approach would be to validate the  accuracy of the current 5-cycle
approach through analysis of real-world fuel economy data. NYDEC and  PC concurred
with the view of AAM/AIAM, noting that the average is more consistent with how
consumers understand the label values, and that using an  alternative would result in
label values that are inaccurate by a greater degree for a higher percentage of
consumers. NADA commented that the goal should be to  present label numbers that
are a "hypothetical average." Natural Resources Canada suggests reducing the
weighting given to high speed and aggressive driving, or shifting the estimates to a
percentile whereby more than 50 percent of drivers would not achieve the label value.

Our response:

      We received no comments opposing the proposed approach or supporting an
alternative approach to the one that was proposed.  We agree with the commenters and
continue to believe that the labels should express consumer's average fuel economy
experience, thus we are finalizing the city and highway estimates as average values as
proposed.
                                      7

-------
1.7   Public Availability of Fuel Economy Test Data

What we proposed:

      We did not propose anything specific with respect to making fuel economy test
data publicly available. However, we have historically made fuel economy data
available in several ways.  Fuel economy information for all cars and light trucks has
always been published in the statutorily required Fuel Economy Guide available at
automobile dealerships. We have also historically published the Fuel Economy Trends
report, which presents an annually analysis of fuel economy data. And we have always
made public the raw fuel economy test data, and have posted data going back to 1978
on our web site.

What commenters said:

      ACEEE commented that EPA should continue to make fuel economy information
available with a high degree of transparency.

Our response:

      We plan to continue to make fuel economy data available to the public in the
most useful ways possible.
1.8   Overall Purpose of Fuel Economy Estimates

What we proposed:

      The EPA fuel economy estimates have appeared on the window stickers of all
new cars and light trucks since the late 1970's and are well-recognized by consumers.
In the NPRM we noted that these estimates essentially serve two purposes:  (1) to
provide consumers with a basis on which to compare the fuel economy of different
vehicles, and (2) to provide consumers with a reasonable estimate of the fuel economy
they can expect to achieve. While the EPA fuel economy estimates have generally
been a useful tool for comparing the relative fuel economy of different vehicles, it is also
important that they reflect the fuel economy that consumers can reasonably expect to
achieve in the real world. Consumers need to be provided with accurate, easily
understandable, and relevant information regarding the fuel economy of new vehicles.

What commenters said:

      Natural Resources Canada took some issue with our use of the label estimates
to represent driving that by definition results in poor fuel economy. In particular, they
focus on the  representation of aggressive and fast driving through the inclusion of the
US06 driving cycle. A more appropriate use of the label, from their perspective, is to act
as a benchmark that drivers can reasonably expect to achieve through fuel-efficient
                                      8

-------
driving techniques. They agree that making adjustments to the label estimates to
account for items that the driver has limited control over (e.g., climate and fuel factors)
is appropriate, "...but making adjustments to the label value to correct for poor driving
behavior may not fairly represent the fuel economy that a driver could (and in our view
should) attain."  In their view a significant part of narrowing the gap between the label
estimates and actual in-use fuel economy should be the education of consumers to
promote better driving behavior.  Our approach, they suggest, diminishes the label's
ability to promote good driving behavior.

      UCS comments that the primary goal of the fuel economy ratings should be to
"provide consumers with a reliable basis for comparing the fuel economy and costs of
individual vehicles."  A secondary goal, they state, is to ensure that policymakers and
the public have a reliable means for measuring the average fuel economy performance
of the fleet as a whole.

Our response:

      We continue to believe that the fuel economy labels serve two key purposes: (1)
to provide consumers with a basis on which to compare the fuel economy of different
vehicles, and (2) to provide consumers with a reasonable estimate of the fuel economy
they can expect to achieve.  Natural Resources Canada's point is really about a
philosophical choice between two ways of representing fuel economy. We have always
determined that the labels should represent typical U.S. driving habits and conditions,
without making judgments as to whether driving behavior is poor and inefficient or
conservative and efficient. We do not believe that we are necessarily correcting for
"poor driving behavior," rather, we see it as representing real-world driving behavior.
Representing fuel economy levels that could be achieved with good driving techniques
under good conditions would not solve the current problem (in fact, it would enhance it)
of a large disconnect between label values and real-world fuel economy. We have to
represent the typical driver and typical conditions,  and at the same time ensure that
vehicle owners receive useful information  regarding  what they can do to improve fuel
economy.
1.9   Small Volume Manufacturers

What we proposed:

      We did not propose any specific provisions for small volume manufacturers, as
our analysis indicated that none were necessary.  We proposed features designed to
provide adequate lead time and to minimize test burden that equally apply to both large
and small volume manufacturers.

-------
What commenters said:

      Ferrari, while in general support of our proposal, expressed concern about
additional testing burden needed to run all five test cycles on each model. They
suggest to adopt a manufacturer-specific correlation factor in lieu of performing the
US06, SC03 and Cold FTP tests, or to use the "worst case" fuel economy data to
represent untested  configurations.

Our response:

      Our proposal does not require manufacturers to run all five test cycles on each
model.  First, until 2011, all manufacturers will be able to employ the mpg line method,
which only continues to rely on the same FTP and HFET tests as today. Second, for
2011 and later models, full five-cycle model type testing will  only be required when the
criteria set forth in the final rule are exceeded. Based on these criteria, we project that
only four percent of test groups will need to perform Cold FTP and SFTP tests and only
13 percent will need to perform US06 tests.  However, EPA's regulations allowing data
substitution (worst case fuel economy data substituted for untested configurations) and
for analytically derived data (where the fuel economy from an untested configuration
can be mathematically determined by using the data from a  similar tested vehicle that is
more fuel efficient.) These provisions should address Ferrari's concerns about test
burden.

1.10  Typographical Errors and Regulatory References

What commenters said:

      Some commenters found what they believe to be errors or inconsistencies in the
regulatory text.  Mitsubishi alerted us to the following issues:

   o  In 600.010-08(b)(1) the text should read "The manufacturer shall generate city
      and HFET fuel economy data..." rather than "The manufacturer shall generate
      city and FTP fuel economy data..."
   o  The equation for Running FC in 600.114-08(a) should be the following, where the
      denominator of the second term in the second parenthetical term is Bag32oFE,
      not Bag22oFE as it was in the NPRM:

   Running FC =
   0.70 x
            0.48
      0.41
0.11
         Bag275FE  Bag375FE  US06CityFE
                            0.30x
0.5
0.5
                                  Bag220FE  Bag320FE
           19.9
                     1
             0.61
      0.39
SC03FE  (Bag375FE  Bag275FE
                                      10

-------
   o  In 600.115-08(b) the equation for Derived 5-Cycle Highway Fuel Economy
      should read "HFET FE" where the NPRM read "FTP FE":

      Derived 5 - cycle Highway Fuel Economy =
                                           (Highway Inter cept}+ ^^    ^-
                                              &   J      ** J     HFETFE   )

   o  Section 600. 1 1 6-08(a)(1 ) references 600. 1 1 4-08(b) for the 5-cycle city fuel
      economy equation. Mitsubishi believe that the referenced section actually
      related to the 5-cycle highway fuel economy equation, and that the appropriate
      reference is 600.1 14-08(a).
   o  Section 600. 1 1 6-08(a)(3)(B)(1 ) references 600. 1 1 4-08(c), a section that does not
      exist. Mitsubishi believes that the appropriate reference is 600.1 14-08(b).

      NREL commented that the air conditioning equation as expressed in the Draft
TSD on pages 67 and 75 is incorrect, and that the correct form of the equation appears
on pages 101 and 113.  They believe that the correct form of the equation is the
following:
                   AICFC=
                            SCQ3FE  \Bag3FE75   Bag2FE.
      NREL commented that the coefficient for Ambient Temperature in the equation
on page 71 should be 0.084, not 0.84.

      Mitsubishi encountered ambiguity in some of the regulatory references in
600.116-08. They note that there are two references to 600.010-08 as a method for
calculating 5-cycle city fuel economy and 5-cycle highway fuel economy.  However,
their confusion stems from the fact that 600.010-08 does not define the 5-cycle
equations, but merely the calculations for each specific cycle in the equations.  They
suggest that it may be more appropriate to reference 600.114-08, where the 5-cycle
equations are defined. Mitsubishi recommends that EPA explain these references more
clearly, or change the references to the appropriate section.

      The comments from AAM/AIAM included a long list of incorrect, obsolete, or non-
existent regulatory references which should be changed. They also offer a number of
suggests for correction or clarification of regulatory language.

Our response:

      We appreciate the level of detail with which these commenters reviewed the
proposed regulatory text.  Many of their comments and suggestions are valid, and we
have incorporated their suggestions as appropriate. Readers of the final regulations will
find that these and other errors and typographical mistakes have been corrected.
                                      11

-------
1.11   Miscellaneous Regulatory Comments

What commenters said:

      AAM/AIAM recommended that a number of definitions be added to the definitions
list in 40 CFR 600.002-08. They proposed adding definitions of US06-City and US06-
Highway to Part 600 rather than modifying Part 86, as was done in the proposal.  Their
proposed definitions are as follows:

   o  Two-bag FTP means the FTP for HEVs as described in section 86.1811 -04(n).

   o  US06-City means the combined periods of the US06 Test that occur before and
      after the US06-Highway period.

   o  US06-Highway means the period of the US06 Test that begins at the end of the
      deceleration which is scheduled to occur at 128 seconds of the driving schedule
      and terminates at the end of the deceleration which is scheduled to occur at 493
      seconds of the driving schedule.

   o  Five-cycle means the FTP, HFET, US06, SC03 and cold temperature FTP tests
      as described in Subpart B and C of this Part.

   o  Alcohol means a mixture containing at least 85 percent by volume denaturized
      ethanol, or because of requirements relating to cold start, safety, or vehicle
      functions, a mixture containing at least 70 percent by volume denaturized
      ethanol.

   o  Alcohol dual fuel automobile means an automobile:
         o  Which is designed to operate on alcohol and on gasoline; and
         o  Which provides equal or greater energy efficiency as calculated in
            accordance with section 600.510(g)(1) while operating on alcohol as it
            does while operating on gasoline; and
         o  (iii)Which, in the case of passenger automobiles,  meets or exceeds the
            minimum driving range established by the Department of Transportation in
            49 CFR part 538.

   o  Certification vehicle means a vehicle that is selected under section 86.1828-01 of
      this chapter and used to determine compliance under section 86.1848-01 of this
      chapter for issuance of an original certificate of conformity.

Our response:

      We agree with AAM/AIAM that the majority of these definitions add clarity to the
regulations, and with one exception we have incorporated them into the final regulatory
                                      12

-------
text. We are not finalizing a definition for "Two-bag FTP" because it is not a term that is
used in the final regulations.
1.12  Daytime Running Lights

What we proposed:

      We did not propose anything specific with respect to daytime running lights.

What commenters said:

      An anonymous commenter suggested that EPA "repeal" the manufacturer
guidance letter CD-94-02, which allows manufacturers to disable daytime running lights
(DRLs) while determining fuel economy. Their reasons for this were several:  (1) DRLs
can impact fuel economy; (2) an impact of up to 0.25 mpg  equates to millions of barrels
simply to support DRLs; (3) increased refueling events exposes consumers to toxic
chemicals and fire injury potential; (4) safety advantages of DRLs are unproven; and (5)
manufacturers have made it increasingly difficult to disable DRLs.

Our response:

      We believe that DRLs have generally evolved to minimize power usage,
especially relative to those that were in use when the guidance memo was issued.
Nevertheless, allowing them to be disabled is inconsistent with the intent of the new fuel
economy estimates to better reflect the conditions that vehicles experience  in actual
real-world  use. Thus, we intend to reevaluate the provisions of the guidance memo in
the future, in consultation with NHTSA.
                                      13

-------
Chapter 2:    Label Format and Content

2.1   General Label Preferences

What we proposed:

      We proposed to revise the design and informational content of the fuel economy
window sticker to convey fuel economy information more clearly to consumers.  The
proposal included four label options under consideration3 While Alternatives 1 and 2
preserve some of the "look and feel" of the current label, Alternatives 3 and 4 depart
from the traditional label design, presenting the same information graphically.
Alternative 4 is similar to Alternative 3, but illustrates the comparable class information
graphically.  The four alternatives are shown below:
          EPS Fuf.1 Ecttntimy t|]rNlt*a>
                     EPA Filial Economy Estimates
          20
                          i&MWl' UK
26
              Alternative 1
           CM f""t EcWiertiy CMintVto
                           Alternative 3
                        EM F"*l economy BjttinnHH*
                   20
                   26

                           cn>
                           MFC.
                         HGHVAV''
                           MPC.
20
26

    Note that the NPRM contained four alternatives, printed in the Appendix to the proposed regulations
on pages 5510-5513, labeled Alternative 1, 2, 3, and 4. These same labels were posted on EPA's
website, but in a slightly different order and with different nomenclature (Label A, B, C, and D). In the
following discussion we refer to the labels printed in the NPRM and use that nomenclature.
                                        14

-------
What commenters said:

      Nearly all of the commenters (almost 3000 in total) preferred Alternative 4,
attributing their choice to the graphic expressing the vehicle's fuel economy relative to
other vehicles in its class. Alternative 4 contained the only within-class comparison
graphic, which many likened to the Energy Star graphic comparing energy use of
appliances.  One commenter stated that Alternative 4 "best clarifies the relative fuel
economy against competitors in the same vehicle class."  Another noted that Alternative
4 "looks sufficiently different from the current sticker to make it easy for consumers to
notice the change." [NYDEC, AAA, F.  Schmitt, Bluewater Network, B. Kotlier, multiple
anonymous]

      However, many automobile manufacturers expressed that the "portrait"
orientation of Alternatives 2 and 4 (the current label format is "landscape") would pose
significant implementation issues. [AAM/AIAM, GM] Subaru supported the industry
position, adding that the shadow figure of a full-size gray gas pump (a "watermark")
would conflict with the Subaru  logo watermark currently displayed on their labels.

      NADA noted that each of the proposed label formats offered pluses and minuses
for dealership sales personnel who frequently address consumers' concerns at the
point-of-sale. They stated that an "informal, non-statistical" survey of some dealers
suggested "overwhelming support" for Alternative 1 and a consensus against
Alternatives 3 and 4.  This preference was attributed to Alternative 1's similarity to the
current label, which they saw as an advantage to the public, noting that it offers a
"cleaner, simpler, and more straight-forward approach" than the current label or the
other proposed alternatives.  Alternatives 3 and 4 were viewed by the surveyed dealers
as "...too radical a departure from the time-honored label currently in use."  They added
that the final label should be as simple as possible.

      Natural Resources Canada supported the layouts of Alternatives 3 and 4 and
mentioned that Alternative 3 may be more familiar to consumers because it retains a
landscape orientation and does not diverge dramatically from the current format.  They
preferred the contemporary design of Alternative 4, because although it may require the
consumer to look carefully at first, it clearly highlights the fact that information has  been
updated and improvements have been made.

      Two  anonymous commenters expressed a preference for Alternative 1.  While
one failed to explain why, the other suggested that the similarity to the current label
would make it easier for consumers to locate it on the window sticker. This commenter
also  stated  that Alternative 1 showed the annual fuel cost estimate more prominently
than the other labels.  [Anonymous#1,  Anonymous#3]

      ACEEE advised EPA that consumers may not fully benefit from the label largely
because labels are not optimized for consumers.  Consequently ACEEE recommended
that EPA hire "one or more outside specialists in visual communication or quantitative
                                      15

-------
information to design from scratch a label for EPA's consideration that will maximize
comprehension, motivation, and message effectiveness."

      Bluewater Network commented that they strongly support the approach taken in
Alternative 4 "because it provides a great deal of valuable consumer information very
clearly."

Our response:

      Clearly, no one label design can address every comment, but with the exception
of NADA, the groups generally agreed that a new appearance is desirable.  We
acknowledge NADA's concern that the new label should not be a "radical departure"
from the current design and recognize that dealers are highly involved in answering
public questions about the label.  Although the new label will present the facts
differently, it will retain  most of the content.  Much of the new information will be self-
explanatory, such as the web link to www.fueleconomy.gov, or the added basis for the
estimated annual fuel costs (price per gallon and miles per year.) However, the public
did  not share NADA's "overwhelming [dealer] support" for a more "time-honored,"
conservative label design; while the  pictorial representation of the comparable fuel
economy is the most transformed feature that dealers may need to explain,  commenters
indicated that the new graphic is  a highly desirable improvement over the textual
version. The label design we are adopting does retain some familiar design  aspects,
specifically the location of the city and highway estimates on either side of the fuel pump
graphic. We believe that we can address NADA's concerns about consumer
comprehension by jointly-sponsoring an outreach campaign with car dealers that could
include explanatory materials, such as a brochure that dealers could distribute to
customers. We also plan to conduct outreach  and education targeted at the general
public to raise consumer awareness of the new label  and its contents.

      We acknowledge the ACEEE suggestion that EPA utilize outside experts to help
design the label.  The old fuel economy label, portions of which were developed in the
1970's, was designed by EPA engineers and not by graphic designers. Based on a
limited amount of market research for the new label, we discovered that prospective
buyers did not understand much of the information on the label other than the large  city
and highway numbers, and that we lacked the expertise to develop a label format that
would be more user-friendly.  Therefore, we contracted a professional graphic design
firm before the proposal, to help us both create and market-test new designs, resulting
in the four proposed alternatives.  We also enlisted their assistance after the proposal
in order to market-test alternatives based on comments received.  This contractor then
incorporated suggestions from public comments and  the focus groups into the final  label
design recommendations for EPA's consideration.  Comments and suggestions from
the  focus groups are located in the Focus Group Report and the Docket.4 The final label
design is shown below:
   4 Each of these can be found in the public docket, which is available at www.regulations.gov.
Search under EPA Docket ID No. EPA-HQ-OAR-2005-0169.
                                      16

-------
EPA Fuel Economy Estimates

CITY MPG
18

Expected range
for most drivers
15 to 21 MPG






These estimates reflect new EPA methods beginning with

Estimated
Annual Fuel Cost
^ffl $2,039 1

based on 15,000 miles
at $2. 80 per gallon
 Combined Fuel Economy
This Vehicle

21

All SUVs
2008 models.
HIGHWAY MPG
25

Expected range
for most drivers
21 to 29 MPG
Your actual

mileage will vary
depending on how you
drive and maintain
your vehicle.
|K*r/5i See tne FREE Fuel Economy Guide at dealers or www.fueleconomy.gov ^Bfiw^
2.2   Label Format Effective Date

What we proposed:

      The NPRM proposed changes to the methodology for calculating label values
and changes to the design and content of the label, beginning with 2008 models.

What commenters said:

      AAM/AIAM noted that when new fuel economy labels become effective (the 2008
model year), the auto companies will be simultaneously be implementing the new safety
rating label requirements finalized by the National Highway Traffic Safety Administration
(NHTSA) on September 12, 2006.5

      AAM/AIAM also  argued that lead time for changing the format is inadequate
because the few months between EPAs finalization of the rule and the potential start of
2008 model year production in January of 2007 is not enough time to work with
suppliers on creating new labels.  Consequently, they requested that any changes to
the look, font, orientation, content, or language be made voluntary for the 2008 model
year.  They specifically noted that they are not proposing to delay the presentation of
     See 71 FR 53572 (Sept. 12, 2006).
                                     17

-------
the new label values, only that changes to the appearance be voluntary for the 2008
model year.  They further recommended that EPA allow manufacturers to maintain the
old label format, provided that text is added in the variable text field (under the fuel
pump, above the Estimated Annual Cost) that states "New 2008 Method."

      Subaru commented that manufacturers frequently purchase pre-printed
Monroney pricing labels in bulk, thus making  it likely that some will already have label
stock for the 2008 model year vehicles long before EPA's rule is finalized.

      The industry's view was that if a final rule were published by September 2006,
modifications to the look, font, orientation, content,  or language on the label could be
implemented on vehicles manufactured on or after  September 1, 2007. They indicated
that if the fuel economy label  effective dates are aligned with the statutorily-mandated
timing of the safety label requirements, they require only one round of significant
modification to their overall pricing labels. General Motors reiterated this position in their
comments, citing their staggered model production throughout the calendar year and
the difficulty in applying labels to early 2008 vehicles with the first build event scheduled
for January of 2007.

      NADA suggested that the final rule clarify that manufacturers are free to provide
dealers with replacement labels for installation on undelivered vehicles initially labeled
under the old rules.  Providing this flexibility, they argue, will reduce consumer
confusion.

Our response:

      We believe that the new fuel economy estimates should ideally be accompanied
by a new fuel economy label design that clearly announces this fact to the public.
Recognizing that a typical dealer's lot may contain vehicles from multiple model years
(e.g. 2007 and 2008) for the first or second calendar years of implementation,
consumers would be able to readily distinguish the new estimates by the new look of
the label, and thus more likely to understand that the fuel economy of 2007 models will
be different from that of 2008 models. We also realize that the timing of this final rule
will cause  some significant practical difficulties for those manufacturers introducing 2008
models early in the calendar year 2007.  Manufacturers remarked that the new NHTSA
safety ratings are statutorily required on vehicles manufactured on or after September 1,
2007, and that harmonizing the new fuel economy label effective with that same date
would ease some of their implementation burden.  In general, the traditional "start" of
the new model year  has been October of the  previous calendar year (e.g., model year
2008 would begin in October of calendar year 2007). However, as manufacturers have
indicated,  staggered production schedules throughout the calendar year have become
widespread across the industry.  We now understand that some manufacturers have
very early  model year 2008 (e.g., January/ February, 2007) launches planned for
several vehicle models. We are therefore finalizing a requirement that aligns mandatory
use of the new label format with the NHTSA label date.  This will give manufacturers
needed lead time to  phase in the new labels without multiple changes to the vehicle
                                      18

-------
pricing (Monroney) sticker.  All 2008 models manufactured on or after September 1,
2007 will have the new label format. We agree with NADA's comment that
manufacturers should be allowed to use the new label design as soon as possible, and
we strongly encourage manufacturers to apply them on a voluntary basis to 2008
models for sale before that date.  Those 2008 models using the current fuel economy
label design will be required to include, at a minimum, the following information on the
label:

         1. The city and highway  estimates, based on the new calculation methods
            contained in this final  rule.
         2. A statement in bold print that says "These estimates reflect new EPA
            methods beginning with 2008 models.".
2.3   Label Content Issues

      2.3.1     Fuel Economy of Comparable Vehicles

What we proposed:

      The EPCA statute requires that the label include the fuel economy of comparable
vehicles: This provision was intended to help car shoppers compare the fuel economy
of similar vehicles. EPA's current regulations require that the label include the following
statement: "For comparison shopping, all [vehicles/trucks] classified as [insert category
as determined in 600.315] have been issued mileage ratings ranging from	to	
mpg city and	to	mpg highway." Based on pre-proposal focus group research, it
appeared that car buyers disregarded this statement since we learned in small print and
contained lengthy text. Many perceived it as "fine print," and thus less important. We
showed the pre-proposal focus groups label samples with  less text for comparable fuel
economy, and they generally preferred these versions. A few participants suggested
that a pictorial representation may be even more clear. In our proposal, three of the four
sample labels contained a revised statement to lessen the "fine print" look, and the
fourth was a graphical representation similar in concept to DOE's EnergyGuide label on
which we requested comment.  This graphic based the fuel economy comparison on
combined city/highway fuel economy instead of separate city and highway comparisons,
as demonstrated in the old label and the proposed alternatives 1 through 3.

What commenters said:

      We received many comments on this issue with suggestions and alternatives
provided below.

      AAM/AIAM noted that EPA's current class definitions are quite broad, and as a
result "the range in fuel economy in any one class is generally very large, and the class
leaders are typically lower volume  specialty vehicles." They  remarked that a Toyota
Prius and a Rolls-Royce Phantom  are both classified as mid-size vehicles, yet they are
                                      19

-------
clearly not in the same market segment.  Citing this wide variation, AAM/AIAM
suggested retaining the current label language on the fuel economy of comparable
vehicles. They requested that if EPA were to receive comments on this issue specifying
a substantial preference for a graphical approach, they "would like to work with the
agency...on the correct look, feel, and  format of a graphic that ensures our competitive
concerns are addressed."

      NADA commented that an informal and non-statistical survey of dealers on their
Regulatory Affairs Committee found that the graphical presentation of comparable class
fuel economy "...failed to offer advantages..." over the text used in EPA's proposed
Alternative 1 label.  However, they held that the label should continue to contain this
information in some form.

      Toyota submitted a label concept that included the following recommendation for
a graphic combining both the annual estimated fuel cost and the fuel economy range of
comparable vehicles.

                                         on

                                This
      Public Citizen suggested that customers will less likely dismiss a pictorial
representation because it may more effectively convey information. They also
remarked that non-English speakers may understand the graphic more easily. In
addition to graphically comparing the combined fuel economy, they recommended that
this approach be used for both city and highway fuel economy estimates.  For
implementation, they suggested that EPA conduct focus groups to determine the best
format for a graphical comparison of comparable class fuel economy.

      Bluewater Network noted that "the bar graph is also consistent with the
EnergyGuide Label,  and highly comprehensible to consumers."

      DCS argued that with the recent proliferation of vehicle designs and the "fading
class distinctions" exemplified by crossover vehicles, consumers have increased
shopping across classes and should be presented with information comparing vehicles
on the dealer lot to the full range of vehicles on the market. They proposed the two
following alternative comparable class graphics for the label including comparisons to
the whole fleet and to the specific class.
                                      20

-------
14   All         25
                                                           13
                                                            44

                       14  AH
      Placed on the Alternative 4 label and similar to that of the EnergyGuide program,
commenters generally understood the graphical element and preferred it over the
current text.  Many commenters preferred Alternative 4 because the visual was more
powerful and comprehensible.  [A. Benenson, AAA, R. Appel].  Natural Resources
Canada stated that a graphical representation should be on the final label design. An
anonymous commenter suggested that since this graphic presented useful information
and was the label's only reference to combined fuel economy, it should be displayed in
a font size "nearly as large as the City and Highway numbers." This commenter also
noted that while the within-class range is useful, it would be even more useful to see the
average combined fuel economy for that class of vehicles.

      One commenter remarked that the graphical depiction of comparable class fuel
economy is "statistically lacking" because it does not convey information about how
many vehicles are in the relevant vehicle class and how their fuel economy values are
distributed between the minimum and maximum. Since one of the endpoints of the
range (e.g., the maximum fuel economy in the class) could be an outlier and the bulk of
the vehicles  in the class might be clustered nearer the minimum value, this commenter
suggested making the graphic a small histogram and provided an example of this
concept.  As illustrated in the following graphic, a car could be at the midpoint between
22 and 40 mpg, suggesting that it achieves average or median fuel economy in the
class. The histogram, however, would more accurately  indicate that a midpoint value
would actually indicate a higher combined fuel economy than most vehicles in the class.
[D. Fitz-Randolph]
                                      21

-------
                       Estimated Annual Fuel Cost: $1435
                                on 1S.OOO    at $2,20 per

                            Combined foei economy for this vehicle
                                 23
                            of       fuel       for ail Compacts
                              Placeholder tor Guezfer Tax
                                     Fit infe
Our response:

      While public commenters preferred a graphic representation of comparable fuel
economy, manufacturers were concerned about competitive fairness.  We examined
this element further with post-proposal focus groups, which highly supported the
graphic.  Although many noted that including the combined fuel economy was
important, since most drivers do not drive exclusively in the city or highway style, others
remarked that while the information would be useful to some,  it would be of no use to
them because they did not shop within a specific class of vehicle.  We thought the UCS
comment to show the fuel economy range of all vehicles had merit, and we tested the
following versions of this graphic with the focus groups:

 Version  1: Within-Class Range of Vehicles   Version 2: Added Range of All Vehicles
      Combined fuel economy for this vehicle
                 20
        10 ^^^^^^^^^ 31
        Range of combined fuel economy
              for all SUVs
Combined fuel economy

   This Vehicle

1,0   w    31            83
  All SUVs
                                                     Ail Vehicles
      The focus groups slightly preferred Version 1 because of its simplicity, many
participants noting that they already knew which class of vehicles they would be
considering. Others preferring Version 2 mentioned that it could influence some people
to reconsider vehicles with higher fuel economy. Although some participants thought the
added fuel economy range in Version 2 was useful, many thought it was too much
information or were confused by what it represents.6

      We are requiring a graphic for comparable fuel economy and the scale showing
the range of fuel economy for vehicles in the comparable class because it was weII-
 The comparable classes that EPA designates are related to how the comparable fuel economy is
presented on the label.  Section 2.6 contains a separate discussion of comparable classes.
                                       22

-------
received by the public commenters and focus groups. People responded favorably to
its likeness to the widely-used EnergyGuide graphic. We recognize manufacturers'
concern that showing comparable fuel economy information may be misleading in some
cases, since the class ranges are often broad and sometimes defined by lower sales
volume vehicles.  However, this issue is derived more from how to create class
distinctions, rather than how to present the information.  Some focus group participants
noted that if a certain class of vehicles had a particularly wide fuel economy range, a car
shopper could solicit the informational sources provided on the label (Fuel Economy
Guide or web site) to determine if the more fuel efficient models in that class would
meet their vehicle needs. We agree with this observation and are retaining a simpler
graphic similar to that in Alternative 4, because the label information is more clear and
concise.
      2.3.2     Estimated Annual Fuel Cost

What we proposed:

      The EPCA statute requires that estimated annual fuel cost be reported on the
label. Currently, the EPA requires only the dollar amount in the cost estimate, but
allows manufacturers to voluntarily include the per-gallon fuel cost and annual miles
driven,  items that derive annual fuel cost.  However, since manufacturers typically do
not include this optional information, most labels contain only the annual cost. We
proposed to require also the per-gallon fuel cost and annual miles  driven on the label in
the following statement: "Estimated Annual Fuel Costs = $XXXX (based on XX,XXX
miles at $X.XX per gallon)." We also sought comments on whether the text should
include the combined fuel economy number as part of the derived  Estimated Annual
Fuel Cost.

What commenters said:

      While opinions on the presentation of estimated annual fuel cost differed, no one
opposed including per-gallon fuel cost and annual mileage, the bases for the estimate.
Some commenters suggested showing a range of estimated fuel costs based on either
a five-year projection of fuel costs or an analysis of fuel volatility. [Bluewater Network,
A. Benenson, anonymous]  Bluewater Network suggested a font size for this element at
least as large as the mpg values and also urged EPA  to present a  fuel cost range from
100% city driving to 100% highway driving, because it is useful to those who drive
exclusively in the city or on the highway. A. Benenson stated that  EPA should revise
the per-gallon fuel cost at least annually, and MT-CHEER noted that the label should
use the most realistic fuel economy number, whether  it is the combined, city, or  highway
figure to estimate the annual cost. AAA and NADA generally supported these notions,
and an anonymous commenter stated that annual fuel cost is "one of the most important
pieces  of information for the consumer." Another anonymous commenter noted that the
label lacked a fuel consumption metric, and suggested displaying the estimated annual
fuel cost more prominently, since it is a useful proxy for fuel consumption.
                                      23

-------
      Public Citizen and UCS commented that because the combined fuel economy
number is important in determining the estimated annual fuel cost, it should be included
in the explanation, since it makes the calculation less transparent to consumers.  They
suggested, however, that EPA conduct a focus group study to examine consumer
response.

      AAM/AIAM and Subaru supported including both miles per year and dollars per
gallon of fuel on the label, because this information may be valuable to fuel-economy-
conscious consumers.  They recommended that EPA continue to update the fuel cost
annually and not more often because frequent changes may be problematic, confusing,
and costly. Toyota submitted the following label concept which combines both the
annual estimated fuel cost and the fuel consumption range of comparable vehicles.

                                        on




                    	W
Our response:

      Focus group participants and public commenters expressed interest in fuel cost
information, remarking that the added bases for this calculation (cost-per-gallon and
miles-per-year) substantially improved the label.  We agree that we should enlarge the
fuel cost font.  We tested various sizes with focus groups; most focus group participants
preferred seeing the fuel cost information more prominently on the label. However, the
fuel cost value will not be as large as the city and highway estimates in the final label,
because we believe that it is confusing to display three large numbers of the same size.
There is now a smaller fourth number for the combined fuel economy shown on the
comparable class fuel economy graphic that competes for space on the label.  The final
font size for fuel cost and combined fuel economy is the same, and the city and highway
numbers are slightly larger.

      We appreciated Bluewater's suggestion that adding a range of fuel cost based on
100% city driving or 100% highway driving may be useful to those who drive
predominantly in urban or rural areas. We tested the following graphics with the focus
groups:
                                     24

-------
        Option 1: Proposed fuel cost info  >  Option 2: With added city and highway
                                                     fuel cost

                                                     Estimated
                Estimated                         Annual Fuel Cost
             Annual Fuel Cost                     corar-ed- $2100
             based 01 15000 ni &s                       .      <-->,  
                                                   basea on It. 000 nt
               at 52.33 per gaiter                         aisr.SC perso
The focus groups had mixed reactions to these options, but slightly preferred Option 1
because it was simpler and provided all of the vital information. Others thought that the
combined estimate would be more accurate, since they did not drive exclusively in
either city or highway conditions. Alternatively, those that preferred seeing the added
city/highway fuel costs did so because they did drive under one condition more often
than another; others simply preferred having more information.

      We are finalizing  Option 1 based on positive response from both public
commenters and focus groups.  While the option to include separate city and highway
annual fuel costs may provide additional useful information for some consumers, others
may disregard it altogether because of its complexity.  Furthermore, there is enough
information provided on the simpler graphic that a person could determine their own
customized fuel cost estimate by modifying one or more parameters (e.g. mpg, dollars-
per-gallon, or miles-per-year).
      2.3.3     Combined Fuel Economy Calculation

What we proposed:

      We proposed that the combined fuel economy used in the comparable class
graphic to calculate the estimated annual fuel costs be the weighted average of 43%
city mpg and 57% highway mpg, instead of the current 55% city mpg and 45% highway
mpg.  This change is consistent with the ratio used to determine combined 5-cycle fuel
economy for the fleet.  This 43/57 ratio is based on national driving patterns indicating
average mileage driven (not time spent) in "city" versus "highway" conditions.

What commenters said:

      Bluewater Network noted that the proposed 43/57 city/highway split for
determining combined fuel economy was not accounting for the amount of time spent in
city or highway conditions.  They suggested that EPA use a 50/50 city/highway split
because it more closely aligns with how many customers consider their driving patterns
(in terms of time spent versus miles driven).
                                      25

-------
      Alliance/AIAM agreed that while the proposed 43/57 split may be statistically
valid on a miles-driven basis, it does not account for time spent during the trip. They
believed that people relate their driving habits most closely to the current 55/45 split,
and that the current division may be preferable, but also recommended that EPA collect
more data including the effects of modern-day congestion.

Our response:

      The two public comments  submitted on this topic suggested that most people
intuitively consider how much time they spend driving in either city or highway-type
conditions, rather than the percent of miles driven, and moreover, that most people think
of their driving as 50% city and 50% highway, or something close to  that.  We are
inclined to agree with this, since the label is designed for the general public's use. We
also  note that the combined fuel economy used for corporate average fuel economy
(CAFE)  and the Gas Guzzler tax remains 55% city/45% highway, according to EPCA
and IRS statutory requirements.  Thus, in order to maintain continuity, we are
maintaining the weights for combined fuel economy at the current 55% city and 45%
highway weights.
      2.3.4     Range of Expected Fuel Economy

What we proposed:

      Although not statutorily required, the current label provides a single statement
explaining why actual fuel economy will vary from the EPA estimates and an expected
range of fuel economy for that vehicle, determined by +/-15 percent of the city and
highway estimates. Providing the range of expected city and highway fuel economy on
the label helps consumers predict their fuel economy across a wider spectrum of driving
conditions. The current label presents a few reasons why mileage will vary and
includes the range of expected city and highway fuel economy.  Earlier focus group
research showed that consumers were not easily associating the ranges given with city
and highway numbers with the vehicle. Some thought that this information was
associated with the city number, since it was located directly below it. We proposed to
split this statement into two separate elements: 1) the general statement that fuel
economy will vary, and 2) the numerical range of expected city and highway fuel
economy.  In order to help consumers determine their realistic fuel economy, we
proposed to place the range of expected fuel economy closer to (underneath or on the
side of, depending on the label) the actual city and highway estimates.  We also
proposed to increase the range to +/-17 percent, representing a 10th to 90th  percentile
of expected in-use fuel economy.

What commenters said:

      AAM/AIAM noted that this information is not statutorily required by EPCA and
stated their belief that it adds little value.  They recommended removing this information
                                      26

-------
from the label, "thus making the look of the label less cluttered and making the
remaining information more bold, concise, and noticeable." They also suggested that
the expected fuel economy range be provided in EPA's Fuel Economy Guide and on the
fuel economy web site, www.fueleconomy.gov.

Our response:

      We agree with the industry comment that since the current label's presentation of
expected fuel economy is difficult to comprehend, it is not useful to car buyers in that
form. However, we maintain that it is important to inform consumers that fuel economy
will vary and to provide a reasonable range given this variability. Focus group
participants more favorably received the shortened disclaimer language and its new
location.  The groups recognized that fuel economy would vary based on the driver and
driving conditions. Some stated that since the estimated values can easily be mistaken
for reality, it is necessary to include the expected range as a disclaimer. Others
commented that this information may be particularly useful for younger, first-time car
buyers, who may not be aware of how fuel economy can vary. We acknowledge these
comments and are finalizing this requirement as proposed.
      2.3.5     "Your Mileage Will Vary" Statement

What we proposed:

      We proposed to separate the range of expected fuel economy and the statement
that actual fuel economy will vary from the EPA estimates, in order to emphasize to
consumers that the mpg values are estimates only and that drivers may experience
different fuel economy depending on many factors.  Since many focus group
participants did not notice the current  label's text, which includes reasons why mileage
will vary, we proposed to reword and reformat it to convey the message more
effectively.  The proposed text was "Your actual mileage can vary significantly
depending on how you drive and maintain your vehicle and other factors."

What commenters said:

      NADA commented that this "disclaimer/qualifier" is an "all-important" item and
should continue to be on the label.

      Bluewater Network suggested replacing the proposed language with the
following, in order to educate consumers about maintenance and driving habits that
affect fuel economy: "For best mileage performance, please keep tires fully inflated,
conduct regular tune-ups, use cruise control, avoid rapid acceleration and high speeds,
and reduce air conditioning, defrosting, or heating when possible." They also
recommended that EPA delete the "actual fuel mileage may vary" or relegate it to a
footnote.
                                      27

-------
      Toyota suggested adding detail to the "Your mileage will vary" statement by
listing items a driver typically can and cannot control. Their proposed language was the
following:

                 Your actual mileage can vary significantly:
                 Some factors are beyond your control
                        traffic, geography and climate
                 Some you can control:
                        Speeding, aggressive driving or not
                        maintaining your vehicle well

Our response:

      Most commenters recognized the importance of the disclaimer statement but
took issue with how much detail should be provided in this statement.  We tested
alternative wordings with the focus groups to see if additional explanation, as suggested
by Bluewater Network and Toyota, would  be necessary and the general consensus was
that fewer words were better. Nearly all of the groups independently suggested the
same revision to the disclaimer statement, which was a combination of a simple
statement and a web link. Since all groups were able to list multiple reasons why fuel
economy varies, including weather, driving habits, and maintenance, we considered that
it may not be necessary to provide detailed reasons on  the label.  Many advised that
including too much detail would be "preaching to the choir," and suggested a web link
for young first-time buyers who  may not know these details, since younger drivers may
be more inclined to use the internet. We are finalizing a shorter disclaimer statement:
"Your actual mileage will vary depending on how you drive and maintain your vehicle."
However, we are not adding the web site to  the statement, because the final label
includes this link in the bottom border, directly beneath this text.
      2.3.6     Graphical Updates and Government Logos

What we proposed:

      We proposed to update the appearance of the label.  In the sample labels, we
included a more contemporary fuel pump design. Since many focus group participants
were unaware that the EPA issues the fuel economy estimates, rather than auto
manufacturers or dealers, we proposed label designs with more prominent EPA and
DOE logos to clarify that the mpg values originate from government testing. We also
proposed adding a link to the EPA-DOE Fuel Economy Guide website
(www.fueleconomy.gov) to provide additional information for interested consumers.
                                     28

-------
What commenters said:

      Public Citizen supported making the government logos more prominent to
indicate that EPA and DOE are responsible for the fuel economy labels and testing.

      AAM/AIAM agreed with enhancing the EPA and DOE logos and label heading.

      Subaru initially objected to the gray gas pump watermark in the background of
two of the proposed labels because it would cover the Subaru brand logo artwork that
currently appears on their Monroney label, but agreed with the more prominent display
of the government logos.  In a subsequent comment, Subaru informed EPA that it was
implementing a new design and was no longer concerned with the fuel pump graphic.

Our response:

      We are requiring the enhanced EPA and DOE logos on the label and the "EPA"
designation in the label title because commenters and focus group participants clearly
indicated that the government seals add credibility to the label.
      2.3.7     Environmental Information on Fuel Economy Labels

What we proposed:

      Historically, EPA used the Green Vehicle Guide web site
(www.epa.gov/greenvehicles) to rate vehicle fuel economy and emissions information
on a 0 to 10 scale. Some suggested adding similar information to the label to convey
more completely a vehicle's environmental performance and to provide a graphical
vehicle to vehicle comparison.  Prior to the proposal, we showed examples of fuel
economy labels that included environmental ratings (for Air Pollution and Greenhouse
Gas) to focus group participants and asked for their impressions. Although there was
confusion arising from the newness of the information, they generally agreed that the
ratings could be useful in the future. We requested comment in the NPRM on a
possible voluntary environmental labeling program.

What commenters said:

      AAM/AIAM stated that they "...support making available to the consumer, data
on vehicle emissions and fuel economy but do not support ratings or rankings of
vehicles that must ultimately be based on subjective methodologies." They continued,
"...we believe that fuel economy and exhaust emissions data should be reported
separately to the consumers." They argued that the label should report exhaust
emissions data to the consumer on the vehicle's certification standard and not a
subjective "1 to 10" or other type of scale, because skeptics can raise numerous
questions on how the factors are weighted. They suggested that EPA make factual
information available to the public to ensure that each individual can make informed
                                     29

-------
decisions, although they also noted from the focus groups that additional information
dilutes the meaning of the overall information. They recommended that we exclude this
information on the label, but publish it on the web site for those interested.

      While NADA commented that "no environmental descriptor is necessary," Public
Citizen suggested that EPA require this information and that it would "greatly enhance
consumers' ability to purchase a vehicle that meets their environmental demands."

      UCS commented that EPA should execute its authority under U.S.C. Section
32908(b)(1)(F) to require environmental information on fuel  economy labels.  UCS
preferred numerical and graphical representations of the vehicle's greenhouse gas and
smog scores like those set forth in the NPRM, noting that these scores should compare
vehicles with all  new vehicles, instead of those within a certain class.  UCS also
suggested that EPA include an official "seal of approval" on  the label for the most
environmentally  benign vehicles.

      ACEEE agreed that the label should include greenhouse gas and smog ratings,
since space is available.  They cited their 2005 report, "Environmental Performance
Labels for Vehicles: Findings of Research with Consumers and Stakeholders," as a
potential source for different forms of presenting this information.  They further suggest
that the time is ripe for adding this information to the label because of the label redesign
process that our proposal put  in motion.

      Environmental Defense did not support a voluntary environmental label with
EPA's greenhouse gas rating from 0-10, which they criticized as "too coarse". They
commented that we should  instead educate consumers on the relation of fuel economy
and the environment and emphasize that individuals can reduce impact on the
environment by  choosing a vehicle with high fuel economy.

      An anonymous commenter suggested that EPA include "CO/mile" (but
undoubtedly meaning CCb emissions per mile) to raise awareness about the vehicle's
impact on global warming.

Our response:

      While some environmental groups wanted to add other environmental
information to the label, automobile manufacturers and NADA were reluctant, since they
were concerned with the subjective nature of the 0 - 10 scale. Focus group participants
repeatedly indicated that a simpler label conveyed the fuel economy  message better.
At this point in time, we believe it is premature to add environmental or GHG score
information to the label.  Since there was no clear consensus from the public comments
and several commenters raised concerns about the scales for these scores, we believe
the issue needs  further evaluation.

      However, to fully consider the suggestions from environmental groups, we tested
the following brief statement linking fuel economy and the environment with the focus
                                      30

-------
groups: "Buying a vehicle with better fuel economy helps protect the environment and
reduces dependence on oil." Reaction to this statement was strongly divided; some
declared that the statement was "preachy," obvious, and unnecessary, while others
stated that it was important and reflected EPA's mission. Because public comments
and focus group reactions were strongly divided, and because we would like the label to
be as simple and uncluttered as possible, we are not requiring other environmental
information on the label at this time. We agree with automakers that the EPA Green
Vehicle Guide and Fuel Economy web sites are better places to publish information on
environmental impacts from vehicle ownership, since there is more space for
elaboration.  Thus, we are not currently addressing a voluntary environmental labeling
effort, but may choose to do so  in the future.
      2.3.8     Vehicle Descriptor Information

What we proposed:

      Consistent with the current regulations, we proposed that fuel economy labels
within the "Monroney" price sticker not contain vehicle descriptor information, but those
affixed separately must include it.

What commenters said:

      AAM/AIAM agreed with the deletion of fuel metering, catalyst usage, and
California systems from the vehicle description requirements in 40 CFR 600.307-08(c).
They additionally suggested that we delete "(iv) Number of engine cylinders or rotors"
as a separate line item subsection and include it instead in the category of "additional
engine information, if necessary." Thus, they suggested changing paragraph (v) from
its proposed reading ("Additional engine description, if necessary to distinguish
otherwise identical model types, as approved by the Administrator; and") to the
following: "Additional engine description (for example, number of cylinders or rotors or
number of valves), if necessary to distinguish otherwise identical model types, in
accordance with good engineering judgment; and...."

      NADA suggested that EPA reconsider whether to require all of the vehicle-
specific description information in the proposed section 600.307-08(c) and noted that
EPA has considerable statutory authority in this regard.

      Several commenters recommended that the label continue to contain the vehicle
descriptor information on the label, because they thought the proposed labels omitted
information such as the vehicle model, engine and transmission type, etc.

Our response:

      Our goal for vehicle descriptor information is to provide consumers enough detail
to distinguish between visually identical models on a dealer's lot that may have different
                                      31

-------
fuel economy because of differences in engine design, (such as number of cylinders,
rotors, etc.). We agree with the manufacturers' suggestion and are revising the
regulation accordingly. We also recognize that EPA has considerable statutory
authority regarding vehicle descriptor information, and would like to finalize the
minimum text required for consumers to match the vehicle and its fuel economy. This
information would appear only on those fuel economy labels that are physically
separate from the price stickers, an infrequent occurrence, based on an informal survey
of car makers.

      We are also finalizing details for the location, font size and other design
information.  To assist manufacturers, we are including an example label that contains
the vehicle description information.
      2.3.9     Website Reference

What we proposed:

      We proposed to require that the label include a web link to the jointly sponsored
EPA-DOE website, fueleconomy.gov, which contains information on vehicle fuel
economy matters and includes downloadable versions of the Fuel Economy Guides,
1985 to present.

What commenters said:

      NADA commented that the label should refer prospective purchasers to the
www.fueleconomy.gov website and the DOE/EPA Fuel Economy Guide for more
information, noting that those purchasers who consider fuel economy an important
factor in their choice will likely use these tools to compare options before arriving at the
dealer.  Public Citizen likewise supported placing the website reference on the new
label. UCS commented that "the prominent inclusion of [the website address] on the
label is essential."

Our response:

      We agree with the commenters that the web link is an important addition to the
label. The pre-proposal and post-proposal focus groups confirmed that the web link
was useful and important information, and thus we are finalizing it on the label.
                                      32

-------
      2.3.10    Transition Language

What we proposed:

      We requested comments on the need for a temporary transitional statement on
the label to indicate that that the fuel economy estimates were based on new methods.
We also requested comments on how long it should appear on the label.

What commenters said:

      Several commented that for a period of time after the new labels are introduced,
the label should include a statement explaining that fuel economy estimates are based
on new methods. For example, A. Benenson suggested the following:

      "Starting with 2008 model year vehicles, EPA has improved the way it estimates
      vehicle mileage. Because of this change, the mpg estimates on this label more
      accurately reflect the mileage drivers will actually obtain."

      The Alliance/AIAM agreed that  a brief statement would  be a positive addition to
the label and suggested "New 2008 Methods."

Our response:

      We agree that a transitional statement is needed, although A. Benenson's
statement may be too wordy and the auto makers' statement too brief.  We tested the
following text with the post-proposal focus groups: "These estimates reflect new EPA
methods beginning with 2008 models." The groups understood the sentence but felt
that it left the reader to ask, "What are the new methods?" They then concluded that
they  might find the answer at the web  site provided or in the Fuel Economy Guide.
Since this is the goal of the website reference, we are finalizing the statement tested
with the focus groups.  It will be centered near the top of the label.

      When asked  how long the transitional statement should appear on the label,
focus group response varied widely. Some suggested the expected duration of a
consumer's vehicle  purchase cycle and others recommended  one year, indicating that
this information would become "stale"  after a while. We are requiring this transitional
statement on all fuel economy labels for the 2008 and 2009 model years,  because
dealer lots may contain vehicles with both label designs.  When 2010 models are
offered,  however, all models on dealer lots will have the new label design, and the text
will not be necessary anymore.
                                      33

-------
      2.3.11    City and Highway Numbers

What we proposed:

      We proposed to retain the large "city" and "highway" numbers as the most
prominent information on the label.  The numbers on the four alternate labels were all
the same size, and were slightly larger than the old label.

What commenters said:

      An anonymous commenter, citing a memory that after the 1972 oil embargo the
city estimate had a larger font than the highway estimate, desired to see this again,
believing that the city estimate is a more accurate estimate of overall fuel economy.

Our response:

      The commenter is correct that in the early stages of fuel economy labeling,  an
EPA regulation eliminated the highway estimate from the label, because many
consumers complained that the highway number was unrealistically high When EPA
revised the label in 1984, it reinstated the highway estimate with a 22% downward
adjustment to reflect real world driving more  accurately. The new calculations we  are
finalizing today reflect changes in driving behavior and  technology since 1984, and
better predict today's on-road city and highway fuel economy. Although no single
estimate can ever capture a  person's complete real world experience, we believe that
consumers will be more likely to achieve the new estimates. While we are continuing to
provide consumers with the range of city and highway fuel economy they can expect to
achieve on the road, we are  also introducing a combined fuel economy estimate on the
label. Finally, when we showed pre-proposal focus groups a label having a single range
instead of separate city and highway estimates, they strongly preferred to see both
numbers, because this format was familiar, and the new presentation confusing.
Therefore, we are requiring both the city and highway estimates.
2.4   Label Size and Orientation Issues

What we proposed:

      Of the four proposed labels, two were positioned vertically ("portrait"), and two
horizontally (landscape).  The vertical format was 4.5 inches wide and 7 inches tall,
while the horizontal format (currently used) was 7 inches wide and 4.5 inches tall.

What commenters said:

      Several manufacturers commented that the portrait format would require a
complete redesign of the "Monroney" price sticker label. AAM/AIAM noted that
changing the orientation would require "significant and costly modifications" to
                                      34

-------
manufacturers' fuel economy labeling and vehicle pricing systems, and recommended
that we retain the current landscape format. GM stated that they recently redesigned
their Monroney pricing sticker to include the newly required NHTSA safety ratings, with
a horizontal fuel economy label, and that changes would be costly. Subaru suggested
letting the manufacturers choose between portrait or landscape versions.

      The general public preferred the appearance of Alternative 4, positioned
vertically, but specifically favored its graphic of comparable fuel economy.

Our response:

      We agree that the new estimates should be represented by a more contemporary
label. We acknowledge industry cost concerns about redesigning pre-existing price
stickers and recognize the challenges posed by simultaneously incorporating two new
labels onto the price sticker- the NHTSA crash rating label and the fuel economy label.
Although public commenters preferred the vertical orientation, the primary reasons
provided were more relevant to the design elements (particularly the gray "watermark"
fuel pump design with information it its "window" and the bar graphic showing
comparable fuel economy) rather than the label orientation itself. Therefore, in order to
address both the consumers' needs and the automakers' concerns, we are finalizing a
horizontal label orientation and requiring those elements of the vertical format that were
appealing to the public, such as the new fuel pump graphic, the bar graph of
comparable fuel economy, clearly indicated fuel cost information, expected fuel
economy ranges located close to the city and  highway numbers, a disclaimer statement
that is set apart, a web link, and a transitional  statement announcing the new estimates.

      We do not agree with Subaru that we should let manufacturers choose their
orientation,  because customers we believe it is important for consumers to recognize a
constant label design.
2.5   Implementation Outreach

What we proposed:

      In the preamble to the proposal, we discussed the need for an extensive
outreach campaign to ensure that the public understands that the estimates and label
design are new.

What commenters said:

      AAM/AIAM urged EPA to plan several media announcements in early 2007 to
inform consumers about the changes to the estimates and the label design.

      Toyota recommended that EPA aggressively promote the new fuel economy
label to assure that customers comprehend the changes in fuel economy numbers,
                                      35

-------
during the label transition process.  Toyota specifically suggested that EPA increase
public awareness of the transition through the Fuel Economy Guide, the fuel economy
website, and other methods, such as a pamphlet explaining the label transition.

Our response:

      We concur and will work with the automotive industry and other interested parties
to inform the public on changes to the fuel economy label and estimates.
2.6   Comparable Class Designations

What we proposed:

      We proposed to add SUVs and Minivans to the pre-existing list of comparable
classes, and to expand the "Small Pickup Truck" class weight limit of 4500 Ibs to 6000
Ibs.

      We did not propose a class definition for "crossover" vehicle types, but requested
comments on how to classify vehicle designs that do not clearly fall into a single
category, based on the EPA definitions.  We proposed to retain our current policy of
allowing manufacturers to recommend,, subject to EPA approval, which class they
believe best fits a "crossover" vehicle.

What commenters said:

      AAM/AIAM supported adding minivans and SUVs to the  list of comparable
classes, suggesting that the proposed definition of minivan be modified to read:
"minivan means a light truck which...," instead of "minivan means an automobile
which...," to align with the definition of light truck given in 40 CFR Section 600.315-
08(a)(2)(iv).  They also recommended that we clarify the distinction between minivans
and SUVs by adding a rear sliding door to the minivan definition.  They agreed with the
proposed increase in weight delineation for the small pickup truck class from 4500
pounds GVWR to 6000 pounds GVWR.  For "crossover" vehicles, they agreed that  EPA
should continue to determine the most suitable class on a case-by-case basis.  They
commented that although they did not recommend changing any of the other
comparable class designations, they were concerned about the graphic presentation of
the range of fuel economy for comparable vehicles, as discussed previously in Section
2.3.1.

      UCS commented that the addition of SUVs and minivans to the vehicle class list
was "good in principle," but that since the definitions of the two are similar, it is likely
that many vehicles will fit both definitions.

      UCS commented that class distinctions have been blurring in recent years, and
that any prescribed definition could quickly become obsolete. They further noted that
                                      36

-------
EPA should present information on the label to accommodate consumers, who are
increasingly shopping across a number of classes, such as minivan, SUV, large car,
etc.  They proposed a comparable class fuel economy graphic that gave the range of
fuel economy for all vehicles and the range of the given class.

Our response:

      We agree with auto manufacturers and UCS that we should clarify minivan and
SUV definitions, and are adopting the suggestions from AAM/AIAM.

      We also agree that we should retain our current policy to determine a class for
"crossover" vehicles on a case-by-case basis.  So-called "crossover" vehicles are those
that meet the definition of more than one vehicle class, and thus are difficult to
categorize. EPA currently uses discretion to assign these vehicles to a class on a case-
by-case basis.  For example, we attempt to determine which class assignment makes
sense from a consumer perspective (e.g., is it more likely to be considered by
consumers looking for a minivan or for an SUV) and what marketing segment is being
targeted by the manufacturer. We did not propose to change how we are addressing
the recent proliferation of "crossover" vehicles, but we requested comments on whether
we should create a separate "crossover"  class.  Some public comments supported the
creation of this class, but did not suggest how to define it. Auto companies were
opposed to it, citing the difficulties in creating a meaningful class definition. Lacking
such a definition that would clearly distinguish between a "crossover" vehicle and other
vehicle classes, we are not creating a separate class for crossover vehicles.  It should
also be noted that the EPA-defined vehicle classes are used only to provide consumer
information about fuel economy and serve no other regulatory purpose  (i.e., they are
not used to determine CAFE compliance, and they are not used for the purpose of
determining compliance with EPA emission standards.)

      In portraying the range of fuel economy for comparable vehicles  on the label,
several commenters noted that the comparable class structure does not adequately
provide consumers with meaningful fuel economy comparisons, and that class
distinctions have been  blurring in recent years.  Commenters noted that many
consumers shop across classes. These  commenters did not suggest any specific
revisions to the class structure to address these concerns; rather, their suggestions
relate to the presentation of the  comparable class information on the label, which is
addressed in Section III. Additionally, manufacturers expressed concern that the wide
fuel economy ranges of some classes are not necessarily representative of vehicles that
consumers would normally compare (the example they cite is the midsize class, which
contains the Toyota Prius and the Rolls Royce Phantom). Auto manufactures further
noted that the highest sales vehicles are  typically near the midpoint of the range, and
that vehicles at either end of the range (low and high fuel economy) are typically
vehicles with low sales volume or "niche" vehicles. They suggest that consumers
usually shop within subsets of the defined vehicle classes, and not across the entire
class.  To  address these concerns,  manufacturers recommended against using a
graphical representation of the comparable class fuel economy, and that EPA should
                                      37

-------
continue to use the text that is used today.  However, they did not suggest any specific
changes to the class structure to address these concerns.

      We believe that with the changes we are finalizing today, the comparable class
structure generally represents the distinctions between vehicle types offered in the fleet
today.  Absent suggestions during the public comment period for new comparable
vehicle classifications, we are finalizing the comparable class structure largely as
proposed,  with minor changes as discussed above.  We welcome interested parties to
continue working with EPA in the future on  how to ensure that the comparable classes
are kept current with the dynamic vehicle fleet.  If it becomes necessary in the future to
further modify the comparable class structure, EPA would do so through a rulemaking.
2.7   Labeling Requirements for Dual Fueled Vehicles

What we proposed:

      Currently, manufacturers may voluntarily include the alternative fuel's estimated
fuel economy and annual cost on the label, but are not required to do so.  The EPCA
statute currently requires that the label should:

     "(A) indicate the fuel economy of the automobile when operated on gasoline or
         diesel fuel;
      (B) clearly identify the automobile as a dual fueled automobile;
      (C) clearly identify the fuels on which the automobile may be operated; and
      (D) contain a statement informing the consumer that the additional  information
         required by subsection (c)(2) of this section is published and distributed by
         the Secretary of Energy." Ref. 49 U.S.C. 32908 (c)(3)."

      The current labeling requirements for dual fueled vehicles are consistent with
these EPCA requirements.  We neither proposed changes to these requirements, nor
sought comment on the topic.

What commenters said:

      EPA received a late public comment from several environmental and consumer
groups urging EPA to require manufacturers to include for E85 vehicles the fuel
economy and estimated annual fuel costs of both gasoline and ethanol. With increased
marketing and availability of these vehicles, the late comment  suggested that the label
be required to not only display separate gasoline and ethanol fuel economy and annual
cost estimates, but also to provide EPA smog and greenhouse gas scores and the ratio
of ethanol to gasoline (which is not always 85:1) on the label.  The additions would
inform customers that, when operating an E85 vehicle with E85 fuel, their fuel economy
will be decreased, not a condition that most consumers expect from vehicles touted as
"environmentally friendly". They also noted that these additions would educate
                                      38

-------
customers that although E85 fuel economy may be lower than that of gasoline, the
alcohol-based fuel reduces greenhouse gas emissions.

Our response:

      We believe that a mandatory requirement for manufacturers to display E85 fuel
economy information on the label in addition to gasoline deserves a  more carefully
considered approach. Before requiring the inclusion of E85 fuel economy for FFVs, we
must consider many issues pertaining to the design and placement of this information,
such as:  1) how to  clearly present E85 mpg relative to gasoline; 2) how to educate
consumers that E85 helps reduce greenhouse gases; 3)  how to best convey estimated
annual fuel costs of E85 (given the  varying cost of E85 prices nationwide), and 4) how
to graphically depict comparable class fuel economy for E85 in addition to gasoline.

      Since we did not request comments on this topic, we are not finalizing
requirements today that differ from the current regulations.  However, we agree that it is
important to provide consumers with complete fuel  economy information on  alternatively
fueled vehicles, particularly in light of the rising sale of flex-fueled vehicles and a
developing E85 fuel infrastructure.  We agree that it is important for consumers to
understand that fuel economy on E85 is typically 20-30 percent lower than on gasoline,
because E-85 has a lower energy density.  Consumers can view the gasoline and E85
estimates of all FFVs in the Fuel Economy Guide and on thewww.fueleconomy.gov
web site.

      We strongly  encourage manufacturers to voluntarily  include the E85 (or other
alternative fuel) mpg and estimated annual fuel costs on the label. The final label
design includes a placeholder for such information. In the near future, we plan to work
with interested stakeholders to determine how best to present E85 fuel economy
information on the label.
   7 Based on fuel economies of gasoline and E85 reported in the Model Year 2006 Fuel Economy
Guide, p. 18.
                                      39

-------
Chapters:    Implementation Issues

3.1    In-Use Validation of 5-Cycle Coefficients

What we proposed:

      In the proposal, we expressed an interest in ensuring that the new methods
continue to reflect real-world fuel economy into the future, and we encouraged
stakeholders to submit data that would inform future analysis and potential changes to
the methodology.  While we did not propose a specific program for validation of the 5-
cycle formulae, the NPRM clearly expressed that we would be receptive to new and
additional data. We noted that the 5-cycle formulae are derived from extensive data on
real-world driving conditions, such as driving activity, temperatures, air conditioner
operation, trip length,  and other factors. We requested comment on the formulae and
on the underlying data on which they are based. We encouraged stakeholders to
submit data that might inform the 5-cycle analyses, and we expressed our desire to
ensure that the 5-cycle approach continues in future years to reflect updated conditions
impacting real-world fuel economy.  Therefore, in the NPRM we encouraged
stakeholders to submit data so that EPA may evaluate the need for changes to this
approach over time.

What commenters said:

      AAM/AIAM expressed support for the 2008 mpg-based approach, calling it "...a
good compromise between urgency and use of available data..."  However, they believe
that "EPA has an obligation to validate the proposed five-cycle equations against
available in-use data to  ensure the continual integrity of the labeling program." They
raise a number of concerns with the 5-cycle approach that they believe  warrant
additional validation.  For example, they believe that the equations may be appropriate
for predicting aggregate fleet-wide fuel economy, but that they may break down when
applied to individual vehicles. They also cite technical reasons for questioning the
accuracy, particularly with respect to the inherent variability of the US06 cycle and how
that might carry through to the label values given the high weighting of the US06 in the
highway equation. They further note that the cycles used in the 5-cycle approach were
designed as worst-case emission test cycles, and that the 5-cycle approach
assumptions that these "extreme" cycles can be used to accurately predict fuel
economy performance over a wide range of operation may be incorrect, especially for
new  technologies such as hybrid vehicles and cylinder deactivation. (These specific
criticisms are addressed in Section 5 of this Response to Comments Document.)
AAM/AIMA makes it clear that they are not asking for new driving cycles, but rather for
adjustments to equations and coefficients if warranted by investigation of in-use data.
They believe that EPA must take a leadership role in this effort, and further urge EPA to
"...place a milestone date in the rule (approximately August of 2009) before which the
agency would review  all available data and determine if the available data raise
questions or concerns about the effectiveness of the labeling program."  If this question
                                      40

-------
is answered in the affirmative, AAM/AIAM requests that EPA initiate a supplementary
rulemaking to propose and finalize changes.

      Honda likewise expressed support for the mpg-based methodology, noting that "it
addresses an immediate need for responding to customer concerns..."  They expressed
a preference that the equations be validated, but mitigating this concern is the fact that
any errors in the methodology would be applied equally to all vehicles.

      DaimlerChrysler and other automobile companies commented that the new
regulation must allow for future review and adjustments based on assessments of real-
world fuel economy data.  They asked that EPA affirm a commitment to adjusting the
formulae if data suggests that that the resulting fuel economy values do not reflect real-
world experience.  They noted the importance  of this with respect to assuring that new
technologies receive accurate label values.

      Toyota echoed these comments, stating that "Toyota believes that the formulas
promulgated in the NPRM need to be validated...," expressing their support for an in-
use fuel economy data collection program. Honda similarly encouraged EPA to support
a testing program of in-use vehicles to validate the 5-cycle equations. Nissan
commented that they could support surveys of in-use vehicles for fuel economy
purposes, but that EPA-mandated testing of in-use vehicles by manufacturers  would be
overly burdensome.

      Toyota expressed an opinion that periodic updating of the fuel economy
coefficients for the 5-cycle equations should be done every 3-4 years.  Specifically, they
suggest that the equations should be updated when an innovative technology  to
improve fuel economy has been introduced.

      Bluewater Network stressed that allowing 20 years to transpire between
adjustments to the fuel economy procedure is unacceptable. They asked that EPA put
in place a system to "(1) monitor accuracy of test results across all manufacturers and
models and (2) to make adjustments as necessary to reflect changed technology,
driving conditions, driving habits, speeds, temperatures, etc." They suggest three ways
to monitor accuracy: requiring manufacturers to submit onboard computer data,
instituting a consumer data collection program, and monitoring fuel economy testing
done by other institutions.

      Public Citizen argues for a thorough study of real-world driving behavior and fuel
economy test procedures that ultimately reflect the findings of such a study. They
argue, for example, that the US06 and SC03 tests may no longer be accurate  because
they were developed from driving surveys conducted  in 1992, and driving has  changed
since then (e.g., driving has become more aggressive). They state that "it is critical that
EPA's testing procedures reflect current real-world driving conditions." They argue that
a study is needed both in the short term to feed into new label estimates, and
periodically thereafter, to ensure that the label  estimates continue to reflect real-world
driving conditions.
                                      41

-------
      NRDC likewise commented that "...this nation's overall fuel economy program
efforts...would critically benefit from a statistically robust collection of in-use fuel-
economy data," urging the auto companies and EPA to commit to a public/private
partnership to ensure that such a program is implemented quickly. They suggest that
the current situation is largely due to the lack of an ongoing mechanism to collect and
evaluate fuel economy data. They also ask that EPA commit to modifying the
methodology if in-use data demonstrates that a change is needed. They argue that an
ongoing program of data collection is necessary to "...ensure consumers are receiving
accurate information in the future, to prevent "gaming" of the system by manufacturers,
and to adequately support the nation's overall fuel economy program."  NRDC is also
concerned about some criticisms that have been raised regarding the statistical validity
and accuracy of the proposed 5-cycle methodology, particularly with respect to hybrid
vehicles, and they suggest that the proposal would benefit greatly from the collection of
more data from hybrid vehicles.  Section 3 of the Technical Support Document (TSD)
contains an analysis of the accuracy and validity of the 5-cycle methodology for hybrid
vehicles.

      UCS states that the proposed 5-cycle methodology is "a promising interim
solution," but that EPA needs to take the next step and begin a comprehensive study of
real-world driving patterns and fuel economy.  They further state that EPA should
develop new test cycles based on the results of the in-use data collection. They argue
that by testing vehicles under the same conditions in which they are likely to be used,
the need for vehicle-specific adjustment factors can be eliminated. Further, they
suggest that new test cycles will fairly reflect the potential fuel savings of various
existing and emerging technologies, which may not be true of the proposed 5-cycle
methodology. They specify that a study of in-use driving patterns and fuel economy
would involve collecting data on fuel consumption, speed on a second-by-second basis,
trip start and end times, air conditioner and defroster usage, ambient and engine
temperatures, additional use of accessories, and other factors relevant to fuel economy.

      UCS commented that in addition to updating the mpg-based equations
periodically, which will ensure that the mpg-based adjustments reflect evolution in the
relationship between 5-cycle and FTP/HFET results, EPA should update the weightings
of the various cycles within the 5-cycle equations as data becomes available.

      NADA commented that EPA should implement the first phase of the program,  but
"...reserve the potential implementation of the second phase until appropriate validation
testing suggests it makes sense to do so."

      AAA expressed no opposition to further research, and in fact said that it should
not be precluded, but concluded that the increasingly-important emphasis on gas prices
and fuel-efficiency mandates that EPA "...get this program up and running quickly."
Bluewater Network similarly  stated that there should be no further delay in issuing a
final  rule.
                                      42

-------
      PPG recommended that EPA consider Federal laboratories such as the National
Renewable Energy Laboratory (NREL) as a qualified source of data, and factor any
appropriate data from such agencies into the new mileage ratings.

      Finally, several commenters suggested that EPA conduct an evaluation of the 5-
cycle method prior to model year 2011, when the 5-cycle method becomes required.

Our response:

      In the proposal, we expressed an interest in ensuring that the new methods
continue to reflect real-world fuel economy into the future, and we encouraged
stakeholders to submit data that would inform future analysis and potential changes to
the methodology.  We remain open to reviewing  any valid test data that would show that
any of the 5-cycle assumptions were  inappropriate for a specific vehicle and considering
modifications to the 5-cycle formulae to account for these differences. We believe it is
critical to ensure that the fuel economy methods  are periodically evaluated.  We are
committed to evaluating the 5-cycle method every several years (e.g., five years) to
ensure that it appropriately accounts for advancements in vehicle technology, changes
in driving patterns, and any new data collected on in-use fuel economy.  We welcome
stakeholders to submit any such future data for use in our periodic evaluation of the fuel
economy test methods.

      We are also committed to offering technical guidance to any stakeholder
interested in undertaking an in-use testing and data-collection program.  By seeking our
technical input up front, stakeholders can better ensure that the data is collected in a
way that is ultimately best-suited to evaluate potential changes to the methodology. It is
important to note that collection of in-use fuel economy data alone is interesting, but can
only be used to indicate whether the 5-cycle estimates are too high or too low;  it does
not inform how the 5-cycle methodology could be improved. The 5-cycle approach is
based on emission test results over the five test cycles and on the weighting  of a
number of factors based  on their average impact across all U.S. driving. Data on in-use
fuel economy alone, without complementary driving behavior and activity data, does not
provide any insight on what changes may be appropriate to the 5-cycle weighting
factors.

      If appropriate data is submitted before the end of 2008, we would plan to review
it in a timely manner.  If such data suggests that changes to the 5-cycle approach are
necessary, we would plan to issue a separate rulemaking to address changes to the
methodology, providing adequate lead time to the industry to comply.
                                      43

-------
3.2   Periodic Review and Updating of the mpg-Based Curves

What we proposed:

      We proposed to update the mpg-based curves periodically, using all of the
available five-cycle fuel economy estimates for the previous three or more model years.
The proposal stated that we would publish the equations for the mpg-based approach
by January 1 of the calendar year prior to the model year to which the equations would
first apply (e.g., for model year 2010 fuel economy calculations, the equations would be
made available before January 1, 2009).  We also indicated in the proposal that
updating would take place periodically, but no more than on an annual basis.

What commenters said:

      While expressing general support for periodic and timely updates, AAM/AIAM
describes concerns with lead time and with  the frequency of such updates. AAM/AIAM
commented that making the equations available on January 1 of the model year prior to
the model year for which they first apply provides insufficient lead time for
manufacturers. They note that the start of production for some vehicles can be as early
as January 2 of the calendar year prior, thus giving manufacturers a new labeling
equation after product designs are locked down, leaving manufacturers unable to
respond to the new equations. Consequently, they advocate making the new equations
available no later than January 1  of the model year two years prior to the model year to
which the equations would first apply. AAM/AIAM agrees that EPA should have the
flexibility to update the mpg-based equations periodically (especially if validation data
identifies a specific need), but they believe that doing so on an annual basis would be
costly and time-prohibitive. They recommend that EPA update the mpg-based
equations "only when such an update is warranted based on statistical significance,
rather than at specified intervals."

Our response:

      We plan to update the mpg-based  curves periodically using all  of the available 5-
cycle fuel economy estimates for the previous three or more  model years. We
proposed that these revised mpg-based equations would be issued through the
publication of an EPA guidance document which would be released by January 1 of the
calendar year prior to the model year to which the equations first apply. We suggested
in the proposal that this meant, for example, that mpg equations for the 2012 model
year would be published prior to January  1 of 2011.  However, we now recognize that
the model year for many manufacturers can begin almost a full year before the start of
the identically-named calendar year (i.e., the 2012 model year can begin on January 2,
2011).  Manufacturers commented that issuing guidance applicable to a given model
year potentially mere days or weeks from the start of that model year for some vehicle
lines did not provide adequate lead time.  We agree with manufacturers that more lead
time  is needed than what we proposed. However, we do not believe a full year before
the earliest model year introduction date is necessary. We believe a period of 6 months
                                      44

-------
prior to the first possible model year provides sufficient lead time, and also gives EPA
ample time to issue new mpg-lines based on more current data. We are thus finalizing
regulations that require EPA to issue guidance regarding revisions to the equations by
no later than July 1 of the calendar year prior to the earliest start of the model year that
starts in the following calendar year. In other words, for new equations to be applicable
to the 2010 model year (which can begin as early as January 2, 2009), EPA must issue
guidance prior to July 1, 2008.
3.3   Use of Bag Data Factors in the 5-Cycle Equations

What we proposed:

      In the Draft Technical Support Document, we derived the mpg-based equations
from a fuel economy database that consisted of both bag-specific fuel economy
measurements and estimated bag measurement data. Those vehicles with bag fuel
economy data were used to develop relationships between bag and whole cycle fuel
economy, and from these relationships the bag data was then estimated for those
vehicles lacking complete bag fuel economy data.  We also estimated the relationship
between US06-City and US06-Highway fuel  economy and overall US06 fuel economy
based on test data from 80 vehicles.

What commenters said:

      AAM/AIAM proposed that manufacturers be allowed to use these same factors
for the initial 5-cycle regression equation when using the 5-cycle approach.  They
suggested that this would "aid in the adoption of the five-cycle equation" and "promote
process simplicity going forward."  They noted that both EPA and manufacturer data
systems are not currently capable of incorporating bag data, and that use of these
factors would provide some lead time for getting these systems updated and modified.
They suggested that manufacturers not be precluded from submitting bag data, but that
if they do such data should be posted to the  EPA website that provides supplementary
information to the Test Car List.  They propose new regulatory language that would
implement these factors.

Our response:

      We do not believe that it is necessary to allow the use of these bag data factors
for early use of the 5-cycle equation. The  5-cycle method is optional in the  first several
years of the new program (model years 2008-2010). EPA will be prepared  to receive
bag data in these instances. Manufacturers will have the necessary bag data to
incorporate into the 5-cycle test, regardless of whether the data systems are completely
ready to accept such data.  We do not believe it is appropriate or reasonable to allow
manufacturers to estimate bag data using  these factors, as it would lessen  the accuracy
of the 5-cycle result for a given vehicle.  Moreover, manufacturers would possess the
bag data results. Therefore, we are requiring that manufacturers use actual bag fuel
                                      45

-------
economy data in their 5-cycle fuel economy calculations, and that they submit that
information to EPA.
3.4   Flexibility to Apply the mpg-Based Adjustments at the Model Type Level

What we proposed:

      Section 600.115-08 of the proposed regulations would apply the mpg-based
adjustments at each test result level supporting the fuel economy label instead of at the
model type level. These values then feed into section 600.207, which determines the
fuel economy values for each vehicle configuration, which ultimately feed into section
600.210 to calculate the label value.

What commenters said:

      AAM/AIAM noted that current regulations apply an adjustment factor at the model
type level (sections 600.209a and 600.207b), not at the test result level, and that
manufacturer and EPA data systems are not configured to apply an adjustment factor at
the test result level. According to AAM/AIAM, many manufacturers do not have
separate data systems for fuel economy labeling and CAFE, and unless EPA allows
flexibility to apply the adjustments at the model type level, this change would impact
manufacturer's CAFE databases. They propose specific language for section 600.210
that would accomplish the desired result.

Our response:

      We agree with AAM/AIAM that allowing flexibility in applying an adjustment factor
at the model type level instead of the test result  level will be less costly to implement
and we are modifying the regulatory text to reflect this. The city and highway test
results obtained from certification and fuel economy test vehicles shall first undergo the
vehicle configuration and model type calculations for FTP and HFET-based fuel
economy (sections 600.206-08 and 600.208-08), and then the resulting FTP-based and
HFET-based model type fuel economy values shall undergo the derived 5-cycle
calculations. This change is consistent with the  placement of the current fuel economy
label downward  adjustments (-10% city and -22% highway), which are performed after
the model type fuel economy is determined.

      As a result, Section 86.600.115-08 has been removed, and the remaining
sections in Subpart B  renumbered accordingly, and the appropriate derived 5-cycle
calculations and associated text has been added to section 86. 600.210-08.  The final
label values will  not be impacted in any significant way, except for possible minor
rounding differences.
                                      46

-------
3.5   Criteria for Additional 20F / SFTP Testing

What we proposed:

      Each year, manufacturers must demonstrate compliance with federal emission
standards by performing tests over all five test procedures.  The vehicles on which
these tests are performed are known as "emission data vehicles", which are selected to
represent the "worst-case" emitting vehicle in a group of vehicles, known as a "test
group", which share common engine and emission control designs.8 EPA issues
certificates of emission conformity for each test group of vehicles in each model year.
Thus, for each test group, there exists a set of official certification test data from all five
test cycles - FTP, HFET, US06, SC03 and Cold FTP. The fuel economy measured from
these official certification tests can be inserted into the 5-cycle city and highway
formulae to determine city and highway fuel economy values. Since FTP and HFET
testing is included in the official certification data, the mpg-based city and highway fuel
economy values can also be determined.  Thus, for each emission data vehicle, the 5-
cycle city and highway fuel economy values then can be compared to the mpg-based
city and highway fuel economy values. We believe that it is reasonable to allow
continued use of the mpg-based line when the available 5-cycle fuel economy data
(from emissions certification) indicates that the mpg-based fuel economy determined
from the official FTP and HFET tests performed for the test group are similar enough to
the 5-cycle fuel economy determined from the official FTP, HFET, US06, SC03 and
Cold FTP tests for that same test group.  In that case, the manufacturer can use the
mpg-based method for all model types covered under the EPA certificate of conformity
that is represented by the 5-cycle data submitted to represent those vehicles. The
manufacturer will not need to conduct 5-cycle testing for fuel economy labeling for these
model types.

What commenters said:

      AAM/AIAM expressed concerns that EPA's proposed rule does not sufficiently
address certain common and uncommon certification vehicle test processes  that would
present challenges in determining whether or not the criteria for additional testing are
applicable. They provide the following examples:
    The "emission data vehicle" is the test vehicle chosen to represent a "test group" for emission
certification purposes. A "test group" is made up of vehicles that share common combustion cycle,
engine type, fuel type, fuel metering system, catalyst construction and precious metal content, engine
displacement, number and arrangement of cylinders, and emission standards. The emission data vehicle
is required to be the vehicle within the test group that is expected to be worst-case for exhaust emissions.
In general the criteria that cause the emission data vehicle to be worst-case for emissions will also cause
it to be worst-case for fuel economy (e.g., it will be the  heaviest vehicle in the test group, with an
automatic transmission, four-wheel drive, etc.). In general, the FTP, HFET, US06 and SC03 are
performed on the emission data vehicle to demonstrate that the test group complies with the federal
emission standards. The Cold FTP is performed on the worst-case vehicle within a durability group, which
represents a larger group of vehicles, including those covered in the test group.


                                        47

-------
     When a model type is certified under multiple test groups (e.g., a model type
      certified under both Federal and California test groups);
     When the five cycles  are not conducted on the same vehicle sub-configuration
      (e.g., SC03 tested at  ALVW and US06 tested at LVW);
     When the EDV conducts multiple tests for any of the five cycles (e.g.,
      manufacture-conducted testing and agency confirmatory testing);
     When the test group is dual-certified (at heavier and lighter test weights).

AAM/AIAM provides proposed language to be added to section 600.116-08 which they
believe would add consistency and fairness to the process. They propose that the
regulations specify that:

     In general, manufacturers should evaluate data generated in the same vehicle
      sub-configuration to determine if the criteria for additional testing are met.
     EPA will provide guidance by the end of 2008 on how manufacturers should
      evaluate if the criteria are met when multiple test data exist, when multiple sets  of
      5-cycle data exist, and when model types are certified under multiple test groups.
     Once a model type is approved as not requiring additional 20F/SFTP testing,
      then the manufacturer may proceed with the labeling of products of evaluation of
      meeting the criteria will not be revisited.
     A requirement to conduct additional 20F/SFTP testing for 2011 and later model
      year products is only  to be based on the test group's 2011 and later model year
      five-cycle test data. For example, if a 2010 MY product is  carried over to the
      2011 MY, and the 2010 MY product used the mpg Approach to calculate the
      label, then the 2011 MY product should be allowed  to use carryover label values.
     If the criterion to conduct additional 20F/SFTP testing is not met based on
      manufacturer-conducted testing, but data from subsequent agency-confirmatory
      testing result in the criteria being met, then manufacturers shall be allowed to use
      the mpg-based equations for calculating label values for all model types in the
      same test group as the certification vehicle except for the model type
      represented by the certification vehicle's configuration. That model type would
      then be required to use the five-cycle equations to calculate the label values and
      would be allowed to use appropriate agency-confirmatory test data to calculate
      the label values. On principle, AAM/AIAM believes that evaluations to determine
      if the criteria for additional testing are met should be based on five-cycle test
      results conducted on  the same test vehicle at the same test facility. AAM/AIAM
      is concerned that if agency-confirmatory test data suddenly resulted in a
      requirement for additional testing, manufacturers could not reasonably complete
      all additional five-cycle tests required for affected models in that test group and
      produce fuel economy labels before production.

Our response:

      We agree that whenever possible manufacturers should evaluate data generated
by the same vehicle configuration or sub-configuration. Although we acknowledge the
manufacturer's point that this may not always be possible,  we believe that in most cases
                                      48

-------
the official EPA test data should suffice, without adjustment or substitution. The
manufacturers noted several situations where the results for the five test procedures
might not meet the ideal in terms of comparability (e.g., tests done on different sub-
configurations with differing test weights). It may be appropriate in some of these
situations to allow the manufacturers to make adjustments to fuel  economy test results
in order to achieve results that are comparable across the five tests. For example, if
four tests are  conducted with a test weight of 4500 Ibs and one is  tested at 4750 Ibs, we
expect that an appropriate method could be identified that would enable the
manufacturer to adjust the test result at the heavier test weight to  represent a lower test
weight. To address this situation, a provision has been added to the regulations
allowing the Administrator to approve adjustments or substitutions to the official
certification test data.  We intend to issue guidance that will address a range of
reasonable adjustments and substitutions of fuel economy data, and that will maintain a
level playing field across the industry.

3.6   Analytically Derived Fuel Economy

What we proposed:

      Currently, the US06, SC03 and cold FTP tests are only performed on a sub-set
of new vehicle configurations.  In contrast, for fuel economy purposes, FTP and HFET
tests are performed on many more vehicle configurations.  In order to minimize the
number of additional US06, SC03 and cold FTP tests resulting from  this proposal, we
proposed that manufacturers be allowed to estimate the fuel economy over these three
tests for vehicle configurations that are not normally tested for emission compliance
purposes using the fuel economy measurements that are normally available.  This is
currently done on a more limited basis for both the FTP and HFET, and is referred to as
analytically derived fuel economy (ADFE).

What commenters said:

      AAM/AIAM support the use of ADFE, but expressed concern that no definitive
timetable was established to ensure that a method will be approved  by the EPA before
5-cycle testing is required for some labels beginning in the 2011 MY. They suggest that
the regulations should state clearly that EPA will  provide guidance by the end of the
2008 calendar year on how to estimate five-cycle fuel economy for configurations not
normally tested over all five cycles. They also request that the regulations provide
sufficient flexibility to allow EPA to consider other proposals for determining analytically-
derived estimates that may be offered by manufacturers. As an example, they note a
possible method using the ratio of "five-cycle city fuel economy" to "FTP fuel economy"
of the certification vehicle, applying that quotient  to the FTP fuel economy of the FEDV
in the same test group to estimate five-cycle city  fuel economy of the FEDV. A similar
approach could also be used to estimate highway fuel economy.  They also note that
EPA may want to consider other more sophisticated methods to estimate individual
US06, SC03 and 20F fuel economy results analogous to currently approved methods
to estimate FTP and HFET fuel economy.
                                      49

-------
Our response:

      When a vehicle is required to generate data from all five test cycles, there are
multiple ways for the manufacturer to accomplish this.  One way would be to perform
the three additional tests - the US06, SC03, and cold FTP tests (the FTP and HFET
would be performed under current and future requirements). The other way is to
estimate fuel economy values over the US06, SC03 and cold FTP tests analytically (i.e.,
analytically derived fuel economy, or ADFE) from testing of a similar vehicle over these
three cycles. Under this method, manufacturers will be allowed to estimate the effect of
differences in inertia test weight, road load horsepower, and N/V ratio (the ratio of
engine revolutions to vehicle speed when the vehicle is in  its highest gear) on fuel
economy, and use these estimates to calculate predicted fuel economy over the three
new fuel economy test cycles. A procedure to estimate the effect of these three vehicle
parameters on FTP and HFET fuel economy has already been developed.9 We plan to
work with manufacturers to appropriately analytically derive fuel economy for the US06,
SC03 and cold FTP tests.  We will implement these estimation procedures using
agency guidance, as is currently done for FTP and HFET fuel economy.
3.7   Consumer Education Issues

What we proposed:

      While we did not propose anything specific with respect to consumer education,
our effort in redesigning the label is to improve how we communicate fuel economy
information to consumers.

What commenters said:

      AIAM/AAMA and Subaru suggested that EPA plan several media
announcements in early 2007 to inform consumers that new changes have been made
to the method for estimating the fuel economy values that will appear on the window
stickers of 2008 MY vehicles,  including any early introduction vehicles that may be
available soon.

      Toyota supports and recommends an EPA initiative to improve consumer
education on fuel economy improvements that can be achieved through better vehicle
maintenance and driving techniques which promote fuel conservation.  NADA opined
that it will continue to be necessary to remind vehicle buyers that variations in climate,
traffic, driving styles, accessory use, loads, tire pressure, weather, vehicle age, vehicle
modifications, maintenance, and other factors, the fuel economy actually achieved in-
use may be higher or lower than the EPA label values.
   9 U.S. EPA Memorandum "Updated Analytically Derived Fuel Economy (ADFE) Policy for 2005 MY
and Later," CCD-04-06 (LDVLDT), March 11, 2004. Available in the public docket for review.


                                      50

-------
Our response:

      We are always looking for new and better ways to communicate fuel economy
information to consumers, and our redesigned label and thewww.fueleconomy.gov
website accomplish much in this regard. We are planning to conduct additional public
outreach when the rule is released, and we are working with stakeholders on how to
best communicate the change in fuel economy label values to consumers.  Part of this
outreach will include educating consumers that they can have a great deal of impact on
the fuel economy they achieve.

3.8   Government-Industry Working Groups

What we proposed:

      We did not propose anything specific with respect to a government-industry
working group.

What commenters said:

      General Motors recommended that EPA convene a working group to address a
number of specific details that they believe require further discussion and analysis
before full implementation can be achieved.  GM volunteered to lead and help
coordinate such an effort.

Our response:

      As always, we will work with industry  representatives and other stakeholders as
needed to achieve the smoothest possible implementation of the new program. We
appreciate GM's offer to assist in this effort.
3.9   When City Fuel Economy is Greater Than Highway Fuel Economy

What we proposed:

      Consistent with the language of current regulations, we proposed that if the city
value exceeds the highway value for a model type, the city value will be set equal to the
highway value. In cases where special vehicle design features may result in city values
that exceed highway values, we  proposed that the manufacturer may request EPA
approval to waive this requirement. Our proposal would require that such a request be
accompanied by on-road fuel economy data which demonstrates that the fuel economy
during city-type driving is higher than fuel economy during highway-type driving.
                                     51

-------
What commenters said:

      AAM/AIAM stated that they fail to see the necessity for the proposed provisions
which would require on-road fuel economy data to demonstrate that city fuel economy
exceeds highway fuel economy. They argue that if a vehicle is designed to optimize city
fuel economy and that is reflected in the testing results, there should be no reason why
additional on-road testing is necessary. They comment that"... imposing an additional,
unspecified on-road testing requirement would amount to a significant burden on
manufacturers that  would add cost, complexity, and time burdens which could impede
the development of vehicles  with attributes desirable to millions of commuters who live
and drive in congested urban areas." They consequently recommend that paragraphs
(c) and (d)(2) of the proposed regulations in section 600.210-08 be deleted.

Our response:

      We agree with the comment from manufacturers, and the final rule thus deletes
the paragraphs in 600.210-08.  Hybrid vehicles in particular are more likely to achieve
better city fuel economy than highway fuel economy, and in the context of this
technology we no longer believe that the existing requirements are needed or
reasonable.
3.10  Technology-Specific, Model-Specific, or Manufacturer-Specific Adjustment
      Factors

What we proposed:

      We proposed a methodology that would apply similarly to all vehicles, without
any adjustment factors based on specific technologies, models, or manufacturers. We
did not specifically propose a methodology for making adjustments to fuel economy
results based on vehicle technology, vehicle models, or on the manufacturer.

What commenters said:

      Because  they observed that "a small number of vehicles appear to have real-
world gas mileage that falls significantly below the proposed new mileage values,"
Bluewater Network suggests that EPA consider finalizing regulations that allow separate
adjustments for  individual vehicle models which have test data that differs markedly
from real-world values. One approach they suggest would be allowing EPA to make
these adjustments based on real-world results reported by consumers and "reputable
sources conducting independent testing" although they acknowledge the potential
difficulty with determining a protocol for deciding if a vehicle's real-world performance is
far enough from the label value to warrant an adjustment.  A second approach they
suggest is to allow manufacturers to collect and submit onboard computer data that
could be used to make adjustments for individual vehicle models.  However, they
                                      52

-------
recognize the potential for "gaming" of the system using such data, and urge EPA to
use caution if such an approach is considered.

      Bluewater Network commented that they believe plug-in hybrids will be
commercially available in the future and EPA should consider how this emerging
technology should be addressed with respect to fuel economy labeling and testing.

      Honda suggested that EPA allow manufacturers to collect real-world fuel
economy data which could then feed back directly into fuel economy estimates. Under
their concept, the manufacturer would gather in-use fuel economy data from actual
customer vehicles during the first 6 to 12 months of production, and then label future
production using the average of the in-use data.

      Volkswagen expressed  concerns that the proposed procedures may unfairly
penalize diesel vehicles. They believe that a high percentage of their  diesel vehicle
customers come closer to achieving today's current labels than is suggested by the
proposed methodology, and that additional factors that reduce the label estimate may
represent an unnecessary penalty for diesel technology.  They acknowledge that little
data exists to support or refute this concern and that they intend to collect some and
present it to EPA.

      AAM/AIAM commented that EPA's proposed changes to the fuel economy label
accounts for many variables that can affect fuel economy, but that expanding the
program to account for technology differences would introduce significant complexity
and not be consistent with the  intent of the Fuel Economy Guide as a convenient
consumer tool.  They point out that validation of the technology adjustment factors
would be necessary to ensure  their integrity, and that such validation could only be
accomplished after extensive field experience was demonstrated. They point to EPA's
1984 final rule on labeling, in which EPA explicitly rejected the technology-specific
approach to adjustment factors.10  A rationale for rejecting technology-specific factors at
that time, and one that AAM/AIAM suggests is even more relevant today,  is that "...such
factors could become outdated so fast as to provide no better (and in some cases,
worse) estimates of expected in-use fuel economy." AAM/AIAM points out that
"...hybrid technology is  progressing so rapidly and there are so many different types of
hybrid systems that hybrid-specific factors would be difficult to determine and would
quickly become outdated."  They recommend that EPA should not consider adopting
technology-specific weighting factors at this time.  Toyota also expressed  opposition to
technology-specific solutions.

      UCS recommends that EPA not consider calculation procedures or adjustments
that are specific to certain technologies, or to vehicles deemed to be used by certain
groups of drivers, unless definitive data can demonstrate no other way to  reliably
provide fuel economy estimates. Any specific procedures could undermine the
impartiality of the fuel economy testing, and should be avoidable by developing new
   10 See 49 FR 13834 (April 6, 1984).
                                      53

-------
cycles that accurately reflect the range of in-use driving conditions experienced by most
drivers.  UCS solidified their argument by noting that "The best chance for EPA's fuel
economy ratings to remain the gold standard of objectivity is to ensure that all vehicles
are tested under an appropriately wide set of identical conditions."

Our response:

      We are not finalizing a methodology by which in-use data can feed back into
adjusting the fuel economy estimates.  At this time, we do not believe  that it is possible
to determine a method for this that is fair, equitable, and that  retains a level playing field
across manufacturers. Allowing this would open up the serious potential for unfair
practices - manufacturers could selectively choose the specific vehicles for which in-
use data is collected, thereby giving them an opportunity to boost fuel economy results
where they believe the in-use data to be an advantage,  but simultaneously ignore those
vehicles that might actually be performing below the label values in real-world use.
Additionally,  adjusting the results of individual vehicles or technologies undermines the
credibility of  the fuel economy estimates because vehicles would no longer be tested in
an identical and repeatable way.  Rather, we would encourage manufacturers that
collect in-use data to  share it with EPA such that the data could contribute to our
periodic evaluation of the 5-cycle method.

      Our analyses in the Technical Support Document do not indicate an issue with
respect to diesel vehicles.  However, we always remain open to  hearing from
manufacturers with specific concerns, and we encourage manufacturers with such
concerns to  collect and evaluate appropriate data and bring it to  our attention.

      Commercial hybrids available today ultimately obtain all propulsion energy from
liquid fuel stored in the fuel tank, while a plug-in hybrid uses a combination of liquid fuel
and supplemental energy from the electric grid stored in the battery (i.e., overnight
charge).  Properly accounting for this supplemental electric energy is the central issue in
assessing the performance of  plug-in hybrids. Therefore, plug-in hybrids will require a
more comprehensive assessment in order to determine the proper test procedures for
fuel economy and emissions.  Since plug-in hybrid technology is rapidly advancing, we
will work with key stakeholders in the next few years to assess the appropriateness of
the 5-cycle methodology in capturing the fuel economy impact of the plug-in technology.

      We agree strongly with  those commenters urging against technology-specific
adjustment factors. We are especially concerned that such an approach would create
an uneven playing field across vehicles. We are  aware that technologies are rapidly
developing, and we intend to ensure, as part of our ongoing evaluation of the fuel
economy test methods, that new and developing technologies are represented
appropriately.
                                       54

-------
3.11  Use of Fuel Economy Estimates in Advertising

What we proposed:

      We did not propose anything specific with respect to the use of fuel economy
estimates  in advertising.

What commenters said:

      An anonymous commenter "is appalled" that auto manufactures are allowed to
advertise using solely their highway fuel economy estimates, while the city values are
relegated to the fine print.

Our response:

      Use of the fuel economy estimates in advertising is governed by the Federal
Trade Commission (FTC). In the mid-1970's the Federal Trade Commission (FTC)
"took note of the dramatic increase in the number of fuel economy claims then being
made and of the proliferation of test procedures then being used as the basis for such
claims."11   They responded by promulgating regulations  in 16 CFR Part 259 entitled
"Guide Concerning Fuel Economy Advertising for New Vehicles" ("Fuel Guide").  The
Fuel Guide, adopted in 1975 and subsequently revised twice,  provides guidance to
automobile manufacturers to prevent deceptive advertising and to facilitate the use of
fuel economy information in advertising. The Fuel Guide advises vehicle manufacturers
and dealers how to disclose the established fuel economy of a vehicle, as determined
by the Environmental Protection Agency's rules pursuant to the Automobile Information
Disclosure Act (15 U.S.C. 2996), in advertisements that make representations regarding
the fuel economy of a new vehicle. The disclosure is tied to the claim made in the
advertisement. If both city and highway fuel economy claims are made, both city and
highway EPA figures should be disclosed. A claim regarding either city or highway fuel
economy should be accompanied by the corresponding EPA figure. A general fuel
economy claim would trigger disclosure of the EPA city figure, although the advertiser
would be free to state the highway figure as well. The authority for the Fuel Guide is
tied to the Federal Trade Commission Act (15 U.S.C. 41-58) which, briefly stated,
makes it illegal for one to engage in "unfair  methods of competition in or affecting
commerce and unfair or deceptive acts or practices in or affecting commerce."
   11  See 40 FR 42003 (Sept. 10, 1975).
                                      55

-------
Chapter 4:    Other Related Proposals

4.1    Voluntary Fuel Economy Labeling for Vehicles Exceeding 8500 Pounds
      Gross Vehicle Weight Rating

What we proposed:

      In the NPRM, we sought comment on a voluntary labeling program for vehicles
above 8,500 pounds gross vehicle weight rating (GVWR), and how such a program
might be implemented.  Over the past several years there has been a growing market
for these heavier vehicles, which fall into a number of utility classes, such as SUVs,
pickups, and vans (including heavier versions of such models as Hummer, Ford
Excursion, Chevy Silverado and Dodge Ram).  We believe that consumers would be
interested in using fuel economy estimates for these vehicles when comparison
shopping.  The rising fuel prices of recent times certainly have increased consumer
awareness of the costs associated with owning a vehicle.  When our proposal was
published we did not have the authority under the relevant statutes to require labeling of
these vehicles.

What commenters said:

      NYDEC expressed support for calling for manufacturers to place fuel economy
window stickers on vehicles over 8,500 pounds GVWR, citing the fact that a large
number of these vehicles are frequently used as passenger vehicles for personal
transportation.  Public Citizen argued that "EPA has clear authority to test and label
vehicles above 8,500 Ibs. and should do so," providing a lengthy legal argument
supporting their contention.  UCS agreed, summarizing the Public Citizen legal analysis
in their comments.  UCS stated that EPA should develop a mandatory fuel economy
labeling  program for these vehicles, noting that these vehicles typically  consume more
fuel than other consumer-oriented vehicles and without a label consumers have no idea
what to expect from them. UCS also recommended that EPA work with DOE to gather
"Your mpg" data from real world users of these vehicles and include this information on
the www.fueleconomy.gov website. Bluewater Network commented that, while it was
unclear to them whether EPCA allows EPA to require labeling of these vehicles, it does
not appear that EPA is precluded from requiring that fuel economy information for these
vehicles be provided in other forums.  Bluewater suggested that EPA require
manufacturers to publish fuel economy values for these vehicles in vehicle user
manuals and other sales materials. They also suggested that EPA make the same
information available in EPA's Green Vehicle Guide, on EPA's website, and in other
relevant publications, to the extent that EPCA allows.

      AAM/AIAM argued that EPA's request for voluntary labeling is understandable,
but it may not provide information that is of significant value.  They noted that tests
currently performed for these vehicles can vary dramatically from the tests performed
for light duty regulated vehicles.  For example, current regulations require that many of
these vehicles  be tested using significantly different procedures from light duty test
                                     56

-------
procedures that generate the fuel economy label values (e.g., some are engine-
certified).  They also noted that highway tests are not required for all fuel types, even in
California where most of these vehicles are chassis-certified. As a result, AAM/AIAM
suggested that labeling of these heavier vehicles will not produce information that is
comparable to the labels on  light duty vehicles, and may in fact create more confusion
and misunderstanding than additional value in the marketplace.

      Ford submitted a comment after the close of the comment  period that responded
in detail to the legal analysis presented by Public Citizen.  Ford's analysis was
subsequently supported by comments from the Alliance of Automobile Manufacturers.
Ford detailed why they believe that Public Citizen's assertion that EPA has the authority
to require fuel economy labeling for vehicles over 8,500 Ibs GVWR is faulty. Ford
concludes that Public Citizen's "expansive" reading EPCA is inconsistent with the
statutory language and principles of statutory construction. Further, they note that at a
minimum, "it must be acknowledged that [EPCA] does not explicitly authorize EPA to
issue rules expanding the scope of labeling, in contrast to the authority granted to
NHTSA...with respect to the CAFE program," and that "At most, EPA should allow
NHTSA to determine the scope of the CAFE program under its authority, and follow suit
in its administration of the labeling program." Finally, they conclude that "EPA should
not seek to require fuel economy labeling on any vehicles not required to comply with
CAFE standards, because doing so would ignore Congress' clearly-stated intent to
avoid having the labeling  program drive additional testing requirements, thereby
overburdening manufacturers."

      NADA did not object to EPA's suggestions that manufacturers voluntarily post
fuel economy information on vehicles over 8,500 pounds GVWR.

Our response:

      We are finalizing in this rule  a fuel economy labeling program for Medium-Duty
Passenger Vehicles (MDPVs),  a subset of vehicles between 8,500 and 10,000 Ibs
GVWR. MDPVs were first defined  in the regulation that put in place the "Tier 2"
emission standards.12 This recently-defined class of vehicles includes SUVs and
passenger vans between 8,500 and 10,000 Ibs GVWR,  but excludes large pick-up
trucks. The specific regulatory definition was designed to capture in the light-duty
vehicle emissions program those vehicles that are designed predominantly for
passenger use.13
   12  See 65 FR 6698 (Feb. 10, 2000).
   13 This is the regulatory definition of Medium-Duty Passenger Vehicle, found in 40 CFR 86.1803-01:
   Medium-duty passenger vehicle (MDPV) means any heavy-duty vehicle (as defined in this subpart)
with a gross vehicle weight rating (GVWR) of less than 10,000 pounds that is designed primarily for the
transportation of persons. The MDPV definition does not include any vehicle which:
      (1) Is an "incomplete truck" as defined in this subpart; or
      (2) Has a seating capacity of more than 12 persons; or
      (3) Is designed for more than 9 persons in seating rearward of the driver's seat; or
      (4) Is equipped with an open cargo area (for example, a pick-up truck box or bed) of 72.0 inches


                                       57

-------
      Under the Energy Policy and Conservation Act (EPCA), EPA is required to
establish regulations that require a manufacturer to attach a label to each "automobile"
manufactured in a model year.14  "Automobile" is defined as a vehicle not more than
6,000 Ibs GVWR, and those vehicles between 6,000 and 10,000 Ibs GVWR that DOT
determines are appropriate for inclusion in the CAFE program.15 "Automobile" for the
purposes of labeling also includes vehicles at no more than 8,500 Ibs GVWR whether or
not DOT has included those vehicles in the CAFE program.16 EPA has no authority to
require labels on vehicles that are not automobiles, therefore EPA has no authority to
require labeling of either vehicles above 10,000 Ibs GVWR, or vehicles between 8,500
and 10,000 Ibs GVWR that are not included by DOT in the CAFE program.

      Since the time of EPA's proposal,  DOT has included some vehicles above 8,500
Ibs GVWR and below 10,000 Ibs in its CAFE program, beginning in model year 2011,17
Since these vehicles now meet the definition of automobile, EPA is authorized to
include these vehicles  in labeling program.

      MDPVs are currently subject to emission standards that apply on the existing
Federal Test Procedure, and many also undergo emission testing on the current
Highway Fuel Economy Test due to requirements in California.  Beginning with the 2011
model year, manufacturers will be routinely testing MDPVs over the FTP and the HFET
tests in order to comply with the CAFE program.  However, MDPVs are not today
subject to all of the additional emission tests we are utilizing for the 5-cycle method.18
Specifically, MDPVs are not subject to the 1996 Supplemental Federal Test Procedure
(SFTP) regulations.19 The SFTP regulations include the US06 and SC03 test
procedures, both of which are necessary elements of the 5-cycle fuel economy
methodology. These two test cycles represent high speed and  aggressive driving
(US06), and impacts of air conditioner operation (SC03). We do not believe it is
appropriate to require SFTP testing for MDPVs for fuel economy purposes alone, but
we are not prepared at this time to establish SFTP standards for MDPVs. In the Tier 2
regulations, we acknowledged that MDPVs were not covered by SFTP requirements,
and we specifically noted that SFTP emission standards would  be addressed in a future
regulation.20 We believe that the appropriate time to consider 5-cycle fuel economy
testing for MDPVs is during or after development of appropriate SFTP emission
standards for MDPVs.  We plan to address SFTP emission standards for MDPVs in the
in interior length or more. A covered box not readily accessible from the passenger compartment will be
considered an open cargo area for purposes of this definition.

   14  See 49 U.S.C. 32908(b).
   15  See 49 U.S.C. 32901 (a)(3).
   16  See 49 U.S.C. 32908(a).
   17  See 71  F.R. 17565, (April 6, 2006).
   18 MDPVs are currently required under the Tier 2 program to meet a carbon monoxide standard on
the cold FTP test; compliance with this standard is being phased in over the 2008 and 2009 model years.
   19 See 61 FR 54852 (Oct.  22,  1996).
   20 See 65 FR 6789 (Feb.  10, 2000)


                                      58

-------
near future. At that time, we will also assess the appropriateness of 5-cycle fuel
economy testing for MDPVs.  However, we are finalizing a program that requires
MDPVs to use the mpg-based adjustments to calculate fuel economy estimates.  The
mpg-based approach does not require testing beyond what will be required to meet the
CAFE program in model year 2011.  Manufacturers will simply take their FTP and HFET
test results (conducted for the CAFE program) and apply them to the mpg-based
equation to determine their fuel economy label values.
4.2   Electronic Distribution of Dealer-Supplied Fuel Economy Booklet

What we proposed:

      A statutory provision in EPCA authorizes EPA to prescribe regulations that
require car dealers to provide to consumers a copy of the annual Fuel Economy
Guide.21 Historically, DOE has printed and sent copies of the Guide to dealers, at
government expense, although this is not an EPCA requirement. In recent model years,
dealers have been allowed, on a trial basis, to provide the Guide electronically. Dealers
can have an on-site computer for customers, or they can provide the Guide on a CD or
diskette, or they can print a paper copy for the customer. This has been working  well,
and DOE agrees with our proposal to codify these options effective with the 2008 model
year.

What commenters said:

      NADA stated at the public hearing and in their written comments that EPA's
existing approach appeared to be working very well, and that they support EPA's
proposal.  Public Citizen opposed allowing dealers the option of electronically providing
the Fuel Economy Guide to customers. Their primary opposition is  (1) that some
people are disinclined to use computers, and (2) that EPA presented no evidence that
the trial program was in fact successful. In an oral testimony given  at the public
hearing, NADA supported EPA's proposal, stating that the approach "seemed to be
working very well."

Our response:

      We are finalizing our proposal to allow the electronic distribution of the  Fuel
Economy Guide by dealerships. We proposed adding language to the regulations that
allows dealers to fulfill their requirement to provide customers with copies of the Fuel
Economy Guide booklet by using an on-site computer. This method has been used on
a trial basis in recent years. The National Automobile Dealer Association (NADA)
commented that this proposal should be finalized, because  it is a more efficient,
effective way of providing customers with this information.  EPA's regulations do not
relieve dealerships of the responsibility to make the Guide "available to prospective
   21 See 49 U.S.C. 32908(c)(3).
                                      59

-------
buyers."22 We agree with Public Citizen that there are people who may be disinclined
to use a computer.  However, we would expect dealers who opt to provide the guide
electronically would also provide assistance as needed to those car shoppers in
accessing and viewing the Guide. The electronic version of the Guide is an easy-to-use
PDF file, which can be viewed a page at a time, exactly the same way a paper copy is
viewed.
4.3   Consideration of Fuel Consumption vs. Fuel Economy as a Metric

What we proposed:

      EPCA defines fuel economy as "...the average number of miles traveled by an
automobile for each gallon of gasoline (or equivalent amount of other fuel) used, as
determined by the Administrator under section 32904 (c) of this title."23 Thus, EPA's
fuel economy information program has always expressed fuel efficiency in miles per
gallon.  It is a metric that Americans have come to know and understand.
Notwithstanding this requirement, a few auto manufacturers have suggested that it may
be more meaningful to express fuel efficiency in terms of consumption (e.g., gallons  per
100 miles) rather than in terms of economy (miles per gallon).  A fuel-consumption
metric is currently used  in Canada and in Europe. Fuel consumption  numbers speak
directly to the amount of fuel used, to which a consumer can relate in  terms of cost
when filling up.  Because a few stakeholders have expressed interest in a fuel-
consumption metric, we requested comments on the gallons-per-mile fuel consumption
metric,  and how it could be best used and presented publicly, such as whether it should
be included in the Fuel Economy Guide.

What commenters said:

      Public Citizen noted that, while there is some merit to a fuel consumption metric,
consumers are comfortable today with the miles-per-gallon  metric. Any change, they
argue, should be carefully deliberated and should involve a massive public outreach
campaign to educate consumers. They also suggest that the estimated annual fuel cost
provides information derived from a consumption basis.

      Toyota believes that fuel consumption is a more meaningful measurement than
miles-per-gallon for expressing fuel efficiency, although they recognize EPA's statutory
limitations. They note that the miles-per-gallon metric is fundamentally non-linear in
relation to  issues of consumer interest, such as cost of fuel  or gallons of fuel used, and
they suggest that anecdotal evidence shows that the non-linear aspects of miles-per-
gallon can itself lead to consumer confusion.  They conclude that "...this is a matter on
which the EPA is obligated to educate the public as fuel consumption, not fuel economy,
is a direct reflection of the environmental impact of vehicles in use."
   22  See 49 U.S.C. 32908 (c)(3).
   23  See 49 U.S.C. 32901 (a)(10).
                                      60

-------
      NADA noted no opposition to including fuel consumption information on the
www.fueleconomy.gov website or in the annual DOE/EPA Fuel Economy Guide.

      An anonymous commenter expressed disappointment that none of the EPA's
proposed label designs prominently lists fuel consumption as opposed to fuel economy.
They note that only the consumption metric truly allows for simple comparisons of fuel
savings between different vehicles.

Our response:

      The final rule will continue the past practice of expressing the City and Highway
estimates in fuel economy units of miles per gallon, rather than as fuel consumption
units.  We are statutorily required to report miles per gallon on the label by EPCA, which
defines fuel economy as "...the average number of miles traveled by an automobile for
each gallon of gasoline (or equivalent amount of other fuel) used, as determined by the
Administrator under section 32904 (c) of this title."24 We have always expressed fuel
efficiency in miles per gallon, and it is a metric that Americans have come to know and
understand.

      Our experience  is that consumers are very familiar with the miles-per-gallon
estimates given on  the label. Given that we are obligated  statutorily to report fuel
economy in terms of miles per gallon, we cannot change the metric on the fuel economy
label, and we believe it would be far too confusing to report a fuel consumption metric in
addition to the miles per gallon estimates. We are concerned that consumers would not
understand a different fuel-efficiency metric and, without a long-term, comprehensive
public awareness campaign, it would be very confusing  to the public. We also
understand that some manufacturers plan to pursue some public outreach and
education in regards to using the fuel consumption metric.

      However, the labels do provide an easy way to compare the relative fuel
consumption - and  therefore, fuel and dollar savings - of different vehicles.  The
estimated annual fuel cost information on the label is based on a fuel consumption
metric: it is simply the dollar equivalent of the number of gallons consumed over 15,000
miles. Thus we believe the inclusion of the estimated annual fuel cost on the label is
already a valuable metric for consumers, which relates directly to fuel consumption.
Moreover, we believe it would take  a long-term educational process for consumers to
begin to relate to the fuel consumption metric of gallons  per mile.
   24  See 49 U.S.C. 32901 (a)(10).
                                      61

-------
4.4   Web-based Driver-Specific Fuel Economy Calculator

What we proposed:

      We requested feedback from stakeholders regarding what additional information
could be made available either in the annual Fuel Economy Guide or the
www.fueleconomy.gov web site.  We acknowledged that overall space on the label is
limited, but that we would like to be able to make more information available to
consumers who are interested in a greater depth of detail. The web site and the Fuel
Economy Guide are logical places to include additional detail that does not lend itself to
presentation on the label.  We requested specific comment on  including a fuel economy
calculator on the web site that would enable consumers to calculate an estimated fuel
economy that is tailored to their specific driving conditions and  behavior. Such a fuel
economy calculator could be designed that would allow the user to input their specific
driving conditions, such as the amount of time spent with air conditioning on, what
climate they live in, how much driving is done under higher speed/aggressive driving
conditions, etc. These inputs could go into an algorithm that would estimate the fuel
economy for a specific vehicle under the conditions input by  the user.

What commenters said:

      Public Citizen is supportive of the addition of a fuel economy calculator to allow
consumers to obtain a fuel economy estimate tailored to their specific driving conditions
and styles.  However, they emphasize that EPA must be careful to ensure that such a
calculator produces accurate estimates.  They suggest that such a calculator, if or when
it is finalized, be referenced on the fuel economy label. Although UCS agrees that such
a tool could provide users with valuable insight into the effects  of many factors on fuel
economy, they recommend that EPA wait to establish such a calculator until additional
data is gathered after the final rule is issued.

      Subaru expressed general support for such a calculator, noting that it "could
prove to be an educational tool to consumers." Natural Resources Canada also
expressed support, recommending that such a calculator be designed to suggest an
optimum target fuel economy  and areas where a driver could, through changes in
behavior or technique, improve fuel economy. AAM/AIAM agreed that a more detailed
calculator on the EPA fuel  economy website could prove to be  useful to consumers.
They suggested that the calculator include driving types, temperatures, and accessory
loads.

Our response:

      Based on the analyses that now underlie the new fuel economy estimates, we
believe at this time that we could construct a web-based calculator as described  in the
proposal. We agree with those who commented that such a calculator could be a useful
and educational tool for consumers.  For example, a user could input their vehicle
information, their passenger load, their air conditioner use, and other factors, and
                                      62

-------
receive as outputs a variety of information regarding the impact of various factors on
their fuel economy and fuel economy estimates that are specific to their situation. We
agree with commenters that it is important to ensure the accuracy of such a tool, since
the algorithms required to complete these calculations are non-trivial. We plan to
consider further how to best design and implement a calculator tool, and we may seek
additional input from interested stakeholders.
                                      63

-------
Chapters:    City and Highway Fuel Economy Estimates and
               Methodology

5.1    The Proposed Formulae are Complex

What we proposed:

      We proposed new methods for calculating fuel economy that are in fact more
complex than the previous methods. However, the proposed methods will also better
predict the fuel economy that customers will achieve in real-world use.

What commenters said:

      Natural Resources Canada, which fields consumer complaints and questions
about fuel economy, expressed concern that the proposed formulae will be too complex
to communicate to consumers.  They suggest that "A new system for estimating vehicle
fuel economy would be simpler and more meaningful to consumers if it did not factor in
extreme driver behavior and if the final equations were streamlined to remove any of the
non-significant coefficients or factors."

Our response:

      We do not believe that most consumers will be interested in or concerned about
the  actual mathematical equations used to combine the fuel economy measurements
made over the various dynamometer test cycles in city and highway fuel economy
labels.  The focus groups indicated to us a general aversion on the part of consumers to
complex and detailed information, except in a very few cases (and in those cases, the
information  is available via the website that will be included on  the new label). We
believe that it will be sufficient for most consumers to understand that EPA included
actual fuel economy measurements while the vehicle was driven at high speeds and
rates of acceleration, at colder temperatures and with the air conditioning turned on.
One of the most difficult tasks we currently face is explaining to consumers how we
historically estimated onroad fuel economy  by testing vehicles at a maximum speed of
60 mph only at 75F with the air conditioning off.

      We also believe that consumers are sufficiently intelligent to understand that
engineering equations can sometimes appear complex, even if the principles involved
can be explained rather simply. The alternative is to simplify the equation and lose
accuracy for the subjective advantage  of being able to show a  simple equation  in the
regulations or an article explaining our new label procedure.  We do not believe that this
is a necessary or appropriate trade-off.
                                     64

-------
5.2   Validation of the 5-cycle Formulae for Individual Vehicles

What we proposed:

      We did not propose a specific method for validating the accuracy of the 5-cycle
formulae on an individual vehicle basis.

What commenters said:

      Honda commented that: "the individual (understood to mean 5-cycle) label
adjustments are based primarily upon analyses of driving behavior alone.  ...  By
ignoring the effect of changes in driving behavior and conditions on the in-use fuel
consumption of individual  vehicles, the Proposal implicitly assumes that all vehicles
have the same relationship between fuel consumption and real world driving conditions.
However, the proposal makes no attempt to validate this assumption, either analytically
or with in-use fuel economy data."

      UCS states that "Because the 5-cycle methodology begins to test for a number of
important factors, it is appropriate for use as an interim solution while new test cycles
are being developed."  They argue that because the 5-cycle approach is based on
models (MOBILE and MOVES) that are designed to consider the entire vehicle fleet, the
methodology may improve the average fuel economy ratings for the fleet without
leading to corresponding improvements in the accuracy of individual vehicle ratings.
They suggest the possibility that the 5-cycle approach is delivering the correct results by
virtue of errors canceling one another out.  For example, the relative importance of cold
starts could be exaggerated, but be balanced out by an underestimation of air
conditioner impacts. Consequently, they recommend that EPA strengthen the
methodology through validation with in-use data.

Our response:

      In addressing the comments by Honda and UCS, it is useful to first summarize
certain key aspects of EPA's general approach  to developing the fuel economy label, as
well as describing the basis for the 5-cycle formulae and how it compares to the current
labeling formula and the mpg adjustment approach.

      The fuel economy label on any vehicle model has two basic functions - to provide
information on the city and highway fuel economy that can be expected  from that
individual model, and to allow for a comparison  between different models.  Meeting
these goals has to take into account that there is a very wide variety of different in-use
driving conditions - speed, acceleration rates, deceleration rates, fluctuations in speed
and accelerations/decelerations, vehicle operating conditions such as AC use that
increase the load on the engine, ambient conditions like temperature and wind speed,
road conditions like grading, vehicle conditions  like tire pressure and wear, and so on.
All of these factors can affect the fuel economy  of a vehicle.  In addition, this wide
variety of driving  patterns  reflects the wide variety among drivers and how and where
                                      65

-------
they drive their vehicles. Since a single fuel economy label needs to provide useful
information to a wide variety of persons nationwide, EPA's general approach has been
to provide label values that predict in-use fuel economy for the vehicle assuming that it
will  be driven in a way that is generally representative of average nationwide patterns.
This recognizes that a single label value cannot be developed that identifies for each
person the fuel economy they would experience based on that individual's unique
driving patterns. The use of a  common benchmark of representative in-use driving
patterns, however, provides information that is still quite useful for each person in
gauging the likely fuel economy they can expect for themselves, as well as providing a
consistent basis for comparison of expected fuel economy between different vehicles.

      Fuel economy label values are derived using fuel economy data obtained by
testing a vehicle over a limited number of prescribed driving conditions in the laboratory,
and then making projections of the vehicle's fuel economy when it is operated in-use
under a wide variety of conditions including those not tested in the laboratory. The
current label formula combines data on fuel economy from testing over the driving
conditions found on the FTP and the HFET, and then adjusts these results by a set
percentage. The FTP represents a subset of city driving conditions - relatively low
speed, mild acceleration and deceleration rates, about 75F ambient temperature,
accessories like the AC are off, and the engine is first started up after a long time sitting
at the same 75F temperature.  The FTP fuel economy value is then adjusted by a  set
percentage to then predict city fuel economy over the wide variety of conditions found in
in-use city driving, including the many types of driving conditions that are not included in
the  FTP - higher speeds, more aggressive acceleration and deceleration, AC use, and
the  myriad other variations in in-use driving conditions.

      Under the current label formula, if two vehicles have the same fuel economy
result on the FTP they have the same label value for city fuel economy.  If their FTP fuel
economy data differs - for example, one is a set percentage lower than the other - then
they will have different labeled values for city fuel economy which will continue to show
the  same relative difference as the vehicles showed on the FTP.  This is based on two
assumptions. The first is that a vehicle with a certain fuel economy during specified
laboratory driving conditions will continue to have the same fuel economy during in-use
driving conditions that are basically the same as those under the test. The common
adjustment factor reflects a second assumption - that every vehicle will react the same
way to the wide variety of in-use driving conditions that are not included in the tests, and
for which we therefore do not have vehicle specific data.  In effect, the current label
formula predicts that if a vehicle were driven under conditions that are representative of
the  wide variation in in-use driving conditions, it's in-use city fuel economy would always
be a set percentage below the value it shows when tested on  the FTP driving
conditions.

      The impetus behind this rulemaking was the large amount of in-use data from a
number of sources indicating that the current formula does not do a satisfactory job of
meeting the two informational goals described above. In developing an updated
approach to address these concerns, a significant amount of information on in-use
                                      66

-------
driving conditions is now available to better characterize the variety of in-use driving
conditions. There are also several additional laboratory tests available that allow for
collection of fuel economy test data from individual vehicles over a wider variety of
driving conditions.

      In developing the 5-cycle formulae, EPA uses much of the same basic framework
described above:  (1) the formula is designed to predict fuel economy over
representative in-use driving conditions, using a common benchmark for labeling
different vehicles, (2) the formula assumes that if a vehicles achieves a certain fuel
economy during testing under a certain kind of driving, then the vehicle will continue to
achieve that fuel  economy value under similar in-use driving conditions, and (3) where
we do not have data to allow projection of individual vehicle fuel economy test data to
similar in-use driving conditions,  then a common, generic adjustment factor is used to
account for all conditions not generally represented by the laboratory tests.

      For example, there are five tests that are now available to provide fuel  economy
values for an  individual vehicle.  As a group, they reflect a much wider variety of driving
conditions. They cover several distinct kinds of driving patterns (based on factors such
as average speed and power- Bag 3 FTP, Bag 2 FTP, HFET, US06 City, US06 Hwy),
different temperature conditions  (75F, 20F, 95F), different accessory loads (AC use on,
AC use off) and different amounts of time sitting between engine start (soak time -10
min., 12 hour). There is also a large amount of information  on in-use driving conditions
to use in deriving a characterization of representative in-use driving conditions.

      For example, fuel economy during in-use driving is made up of two parts - the
fuel used during warm-up of the  engine when that engine is started (start fuel use), and
the fuel used  after the engine is warmed up (running fuel use).  For each of these, EPA
developed a comprehensive picture of average conditions.  For example, for running
fuel use EPA developed a description of average U.S. driving from the Draft
MOVES2004 motor vehicle emissions model.  EPA determined that about 41 % of in-
use city driving is under driving conditions that are similar to the driving conditions of
Bag 3 of the FTP, 48% of in-use driving is similar to the driving conditions of Bag 2 of
the FTP, and 11 % is similar to the more aggressive driving conditions in  the USOSCity.
EPA also evaluated temperature conditions during which in-use driving occurs, and
determined that the average temperature in which driving occurs is approximately
58.7F,  and on average the air conditioner compressor is engaged and putting load on
the engine approximately 15.2 percent of the time. EPA then developed a formula that
properly weights  the varied kinds of test driving conditions by their relative contribution
to this benchmark of average in-use driving.  In some cases this required additional
analysis. For example,  the tests are driven at 75F and 20F, but not at 58.7F.  Using
engineering analysis and test data on how fuel economy changes as temperature
changes, EPA developed a relationship for the change in fuel use related to starting the
engine as temperature increases from 20F to 75F. This relationship, along with an
individual vehicle's test data at 20F and 75F, is used to predict the fuel economy at
58.7F.
                                      67

-------
      The 5-cycle formulae therefore reflects EPA's use of a common benchmark for
labeling - average in-use driving conditions nationwide - and the assumption that a
vehicle's fuel economy during a certain kind of driving in the lab reflects the vehicle's
fuel economy during similar in-use driving conditions. For example the running use
formula assigns a rating to Bag 3 of the FTP that is consistent with the fact that the
driving on Bag 3 is similar to 41% of the average city driving. The formula is based on
the assumption that the fuel economy a vehicle achieves during Bag 3 of the FTP is a
good predictor of the fuel economy it will receive during the 41 % of average in-use city
driving that is similar to the driving during Bag 3.  The formula also incorporates the fuel
economy test data at 20F and 75F to numerically provide the proper discount to reflect
that average city driving is at 58.7F.  For conditions that are represented by the test
driving,  EPA assumes that in-use driving will show the same variation in fuel economy
that the test driving does - different vehicles that have different fuel economy during the
test driving will show the same variation in fuel economy when driving in-use over
similar conditions. This is reflected in the weightings used in the 5-cyle formulae.

       The 5-cycle formula also uses a generic adjustment factor to account for the
variety of conditions that occur during in-use driving that are not fairly  represented by
the driving conditions under which individual vehicles are tested, such as variations in
wind speed, road conditions, tire pressure, and the like.  For these conditions EPA
does not have test data for individual vehicles and assumes that all vehicles will show
the same effect notwithstanding differences in  design between different vehicles. This
is reflected in the generic adjustment factor.

      There are therefore two basic differences between the current label formula and
the 5-cycle formula.  The range of in-use conditions over which EPA is able to use test
data to identify differences in fuel economy between different vehicles is a broader
range of driving conditions for the 5 cycle than the current label, and the range of in-use
conditions over which  EPA assumes all vehicles react the same way is a narrower set
of conditions over the 5-cycle than the current  label.

      Honda commented that: "the individual [5-cycle] label adjustments are based
primarily upon analyses of driving behavior alone.  ... By ignoring the effect of changes
in driving behavior and conditions on the in-use fuel consumption of individual vehicles,
the Proposal implicitly  assumes that all vehicles have the same relationship between
fuel consumption and real world driving conditions.  However, the proposal makes no
attempt to valid ate this assumption, either analytically or with in-use fuel economy
data."

      This is not an accurate description of the 5-cycle formula.  It is based upon
analysis of in-use driving behavior, but not primarily on that alone. An analysis of
driving behavior is a central element, but just one element. The analysis of in-use
driving patterns is critical because it allows  EPA to  accurately characterize average
driving conditions across the US, and use this as a benchmark. It also allows EPA to
accurately identify which parts of this average in-use driving are represented by the
various kinds of driving conditions in the laboratory tests, and properly weigh the test
                                       68

-------
driving so it accurately reflects the contribution that that kind of driving makes to the
benchmark of average in-use driving.  Having done that, the 5-cycle then applies the
vehicle specific fuel economy data generated in the test driving, weighing it properly to
reflect the contribution of that driving to overall, representative in-use driving. Thus the
analysis of average in-use driving behavior is a central element of the formula,  but it is
not used alone or in isolation, but instead is combined with individual vehicle specific
fuel economy data generated during testing.

      The 5-cycle formula does not "ignore[e] the effect of changes in driving behavior
and conditions on the in-use fuel consumption of individual vehicles." It does the
opposite.  The 5-cycle formula uses a broader range of driving conditions under which
individual vehicles are tested, which then identifies whether and how much a change  in
driving conditions changes fuel consumption for the individual vehicle. The formula then
assigns that change in fuel economy to similar in-use driving conditions.  It does this
over a broader range of driving conditions than the current label formula,  because the
driving tests cover a broader range of driving conditions. The generic adjustment used
in the 5-cycle formula does ignore differences between individual  vehicles, and
assumes that all vehicles will react the same way to the changes  in in-use driving
conditions covered by the generic adjustment factor.  This applies to the conditions not
fairly represented by the test driving conducted on individual vehicles. However this
assumption covers a much narrower range of driving  conditions for the 5-cycle  formula
than for the current label.

      The 5-cycle formula does not "implicitly assum[e] that all vehicles have the same
relationship between fuel consumption and real world driving conditions." As described
above, for the broad range of driving conditions covered by the test driving conditions, it
identifies an individual vehicle's fuel consumption. This identifies  differences between
vehicles over these driving conditions. These differences in fuel economy are then
properly weighted to reflect their occurrence during in-use driving.  It is only for the
generic adjustment that the 5-cycle formula assumes  all vehicles have the same
relationship between fuel consumption and real world driving conditions.  However for
the 5-cycle formula this is a narrower range of in-use  driving conditions than the current
label formula.

      Honda objects that EPA did not validate this implicit "assumption",  either
analytically or with in-use fuel economy data for individual vehicles, before making the
5-cycle formulae mandatory.  As noted above, EPA does not make the assumptions
Honda suggests, except for the generic adjustment factor.  The technical basis for the
generic adjustment factor is explained in Section III.A.5 of the Final TSD.  Honda does
not appear to object to the technical basis for the generic adjustment, however. Its call
for validation is aimed at the parts of the 5-cycle formula that use vehicle specific test
data, and not the generic adjustment which is not vehicle specific. For the vehicle
specific parts of the formulae, as discussed above, EPA did not make the assumptions
that Honda states, and in fact made distinctly different assumptions.
                                       69

-------
      Honda does not appear to object to several of the basic elements of EPA's
approach.  For example, Honda does not appear to object to the assumption that the
fuel economy an individual vehicle demonstrates when tested under certain driving
conditions also fairly represents the fuel economy the vehicle will show during similar in-
use driving conditions. While in-usefuel economy data for individual vehicles arguably
would be relevant to validate this assumption, it is clearly a reasonable engineering
judgment to make and in any case there is no apparent dispute on this issue.

      Honda does not appear to raise significant objections to the concept of using
average in-use driving conditions as the benchmark forfuel economy labeling. That is,
the fuel economy label should be based on a projection of the fuel economy of the
vehicle over a driving pattern that fairly represents average US driving conditions25
While Honda states there is a need to validate the 5-cycle formula for in-use fuel
economy for individual vehicles, such in-use fuel economy validation would not be
relevant to characterizing this average US driving pattern. The average driving pattern
is not intended to be vehicle specific; it is an average that is then used as the
benchmark for labeling. The weightings of the formula are likewise  not vehicle specific,
but are intended to reflect the correlation between the  kind of driving during testing and
the amount of such  driving in the average US driving.  In use fuel economy data would
not inform you about the validity of this average driving pattern, because it is a driving
pattern,  not a fuel economy value. It is true that information on in-use driving of
individual vehicles would provide additional data about in-use driving patterns,
irrespective of the in-use fuel economy. However absent a broad range of in-use data
over a variety of models, manufacturers, and driving conditions, it would not be a useful
way to identify whether the average driving pattern  needs to be revised. Again, that is
because this benchmark and the related weightings in the formula are not vehicle
specific but are  intended to be an average covering a wide range of in-use conditions.

      The vehicle specific part of the 5-cycle formula is the fuel economy data derived
for an individual vehicle over the kinds of driving covered by the various tests.  In-use
fuel economy validation from individual vehicles is not  needed to validate the laboratory
results themselves,  and as discussed above it is not needed  to validate the assumption
that the individual test data for the vehicle represents the fuel economy it will achieve
when driving under conditions in-use that are similar to those during testing.

      Arguably Honda is referring to EPA's invitation for comment on whether certain
vehicle technologies are generally driven in a way that is distinctly different from the
average, e.g. unusually low incidence of aggressive driving, A/C usage, etc. See 71 FR
5442 (co!2-3). For example, Honda suggests using the 5-cycle formula only until a
manufacturer collects enough fuel economy data on an individual model from in-use
drivers of that model, and then substituting this manufacturer derived fuel economy
value for the 5-cycle value.   As discussed in the preamble of the proposal, in this
situation in-use data would be used to show that a different average benchmark (with
   25  Honda does object to various details of the benchmark average driving pattern developed by EPA,
and those objections are responded to elsewhere.


                                       70

-------
correlated changes to the weightings in the 5-cycle formula) should appropriately be
used for a specific kind of technology. That is a different issue, however, than the need
for in-use validation before using the 5-cycle formula for it's stated purpose - to predict
in-use fuel economy over a broad national average of driving conditions, based on fuel
economy test data for an individual vehicle gathered over a broad but necessarily
limited variety of driving conditions.  The fact that in-use fuel economy data might be
necessary and useful in deciding whether to provide an exception to the benchmark
average driving pattern for certain vehicles or a certain kind of technology does not
indicate that such in-use data is needed or useful in validating the average in-use
driving pattern itself.  As noted above, it generally is not.

      As will be discussed further below, the proposed 5-cycle formulae are based in
part on some assumptions about how vehicles respond to conditions that are not tested
directly by the five dynamometer tests. E.g., the effect on fuel economy of ambient
temperatures between the 20F and 75F is subject to an engineering analysis and
certain test data to establish the relationship between fuel economy at the average US
driving temperature, 58.7F, and fuel economy measured at 20F and 75F.  Likewise,
EPA makes certain assumptions about the AC compressor load during driving patterns
not included in the test driving. However, these assumptions are much less extensive
than those involved in either the current label formula or the proposed mpg adjustment
formula. All the assumptions regarding how fuel economy varies under conditions "in
between" the various tests were supported by the results of dynamometer testing or
engineering analysis. Honda has provided no data to indicate that any particular
assumption was inappropriate for any specific vehicle. EPA remains open to reviewing
any valid test data which would show that any of its assumptions were inappropriate for
a specific vehicle and considering modifications to the 5-cycle formulae to account for
these differences. But the need to perform these analyses is not a basis for delaying
use of the 5-cycle formulae. EPA's analyses are supported in the record, as noted, and
EPA believes the 5-cycle formula  presents significantly less risk of error in labeling than
the much more extensive assumptions that made and not supported in the current label
formula.

      For the reasons noted above, EPA does not believe that it needs in-use
validation data from  individual vehicles prior to making the 5-cycle formula mandatory.
In large part this is based on the extensive technical analyses underlying the 5-cycle
formula, the validity of the representative in-use driving pattern used as a benchmark,
and the increased ability to use individual vehicle test data to represent the vehicle's in-
use fuel economy.   In addition, this is based on the view that the 5 cycle formulae
would still be a better choice for fuel economy labeling that the mpg adjusted formula or
the current label formula.

      It is inconsistent to reject the 5-cycle formula because it has not been validated
through data on in-use fuel economy for individual vehicles, but to accept the mpg
adjusted formula or the current label formula.  The mpg adjusted  formula is basically a
streamlined way to predict the 5-cycle formula result for an individual vehicle. For the
large majority of vehicles, the mpg adjustment formula does a good job of accurately
                                       71

-------
predicting the 5-cycle formula value for that vehicle.  If the 5-cycle formula is
inappropriate because of various problems and the lack of in-use validation, then the
mpg adjustment is inappropriate for the same reasons because the mpg adjustment
formula is just a shorthand way to develop the 5-cycle formula result. There are
vehicles where the mpg adjustment formula does not do a very good job of accurately
predicting their 5-cycle formula results. For this subset of vehicles, the mpg adjustment
basically assigns them the average 5-cycle fuel economy derived from the large
majority of vehicles where the mpg adjustment does a good job of predicting the 5-cycle
value.  For these cars, any basic defect of the 5-cycle formula would also apply to the
mpg adjusted value.  If there is not a valid technical basis to use the 5-cycle formula
then logically there is not a valid technical basis to use the mpg adjustment formula.
EPA believes, however, that there is a valid basis to use the 5 cycle-formula, hence a
valid basis to use the mpg adjustment formulae as well,  especially for the large majority
of vehicles where there is a close correlation between the 5 cycle result the and the
mpg adjustment result.

      It also would be inconsistent to delay the 5-cycle formulae pending further in-use
validation, while continuing to use the current label formula. The current label has not
been validated for individual in-use vehicles, and in general in-use data shows that the
current label formula generally overestimates fuel economy and masks differences in
fuel economy between individual vehicles. This is because it  generates vehicle specific
fuel economy data over a narrow range of laboratory driving conditions, and it uses a
generic factor, that does not discriminate between individual vehicles, to cover all other
driving conditions. The 5-cycle formula generates vehicle specific fuel economy data
over a wider range of test driving conditions, and relies on a generic adjustment factor
to cover a narrower range of the remaining driving conditions. The comparison of 5-
cycle results to the current label, as reflected in the mpg adjustment formula, indicates
that the 5 cycle formula does a better job than the current label formula of predicting in-
use fuel economy and identifying fuel economy differences between different vehicles.

      The real solution to the issue raised by Honda and explicitly suggested by UCS-
greater accuracy in predicting individual vehicle fuel economy - is not to delay use of the
5-cycle formula but to expand the extent of individual vehicle fuel economy testing in the
laboratory (i.e., more driving patterns, intermediate colder and hotter temperatures,
intermediate soak times, etc.), assuming the increase in accuracy was worth the
increased testing. Honda and the other vehicle manufacturers have argued strongly
against  this approach. Absent such an increase in fuel economy testing  of individual
vehicles, for the reasons discussed above EPA believes there is not a valid technical
basis to delay the use of the 5-cycle formulae pending validation by in-use fuel economy
testing of individual vehicles, with reliance on either the mpg adjustment  or the current
label formula in the meantime.
                                       72

-------
5.3   Some Formulae Coefficients are Counter-Intuitive

What we proposed:

      We proposed weightings of the various cycles or bags that were based on
extensive analyses of in-use driving data and vehicle test data.

What commenters said:

      Honda commented that the overall contributions of several of the cycles or bags
in the 5-cycle formulae do not make engineering sense, as some of the contributions
are negative.  In terms of fuel consumption (as opposed to fuel economy), Honda
consolidated the various terms of the 5-cycle formulae  in order to identify the overall
weight given to each cycle or bag. The results are shown in Table 5-1.

Table 5-1:  Proposed Coefficients for the 5-Cycle Fuel Economy

FTP - Bag 1
FTP - Bag 2
FTP - Bag 3
HFET
US06- City Bag
US06- Highway Bag
SC03
Cold FTP - Bag 1
Cold FTP - Bag 2
Cold FTP - Bag 3
City
25.7%
28.0%
-5.8%

7.7%

14.4%
8.1%
15.0%
6.9%
Highway
1.5%
-2.0%
-4.6%
21.3%

79.9%
5.0%
0.5%

-0.5%
Honda presents five concerns related to the figures in this table.

      1)    The negative coefficients for Bag 3 of the FTP in both formulae, and the
            negative coefficient for Bag 2 of the FTP in the highway formula. These
            negative coefficients mean that if a manufacturer increases a vehicle's fuel
            economy during this type of  driving, the vehicle's 5-cycle fuel economy will
            actually decrease, not increase.

      2)    The weighting for SC03 in the city formula is 14.4% but the driving
            conditions represented by this cycle only represent 3% of all driving.

      3)    The total weighting for cold FTP in the city formula is over 30%, but the
            driving conditions represented by this cycle  only represent 2%  of all
            driving.

      4)    In total, the extreme cycles represent over half of the weighting of the city
            formula, while only representing 5% of all driving.
                                      73

-------
      5)    For the highway driving, the extreme cycles account for over 85% of the
            highway label value.

Our response:

      We conducted our own analysis to determine the overall weightings of the
various bags and cycle in the two 5-cycle formulae and confirmed the reasonableness
of the figures shown in Table 5-1. Thus, we agree with Honda that, from a purely
mathematical perspective, the figures in  Table 5-1  indicate how the fuel consumptions
measured over the various bags and cycles contribute to the overall estimate of fuel
consumption during  either city or highway driving and thus, to the 5-cycle city and
highway fuel economy label values.

      Regarding Honda's concern about negative bag coefficients, we believe that this
concern is purely theoretical, and does not represent a real problem with the 5-cycle
formulae. For example, we recalculated the overall coefficients for the various bags and
cycles using a very large average trip length (i.e., 10,000 miles) which effectively
removes any contribution from the cold start towards city or highway fuel economy. The
results are shown in Table 5-2.
Table 5-2:  Proposed Coefficients for the 5-Cycle Fuel Economy with
Infinite Trip Length

FTP - Bag 1
FTP - Bag 2
FTP - Bag 3
HFET
US06- City Bag
US06- Highway Bag
SC03
Cold FTP - Bag 1
Cold FTP - Bag 2
Cold FTP - Bag 3
City
0%
27%
17%

10%

16%
0%
16%
14%
Highway
0%
-2%
-3%
17%

81%
7%
0%

0%
      As can be seen, the negative coefficient for Bag 3 of the FTP in the city formula
changes to a relatively large positive 17% with the removal of cold start fuel use. Thus,
the -5.8% coefficient in the proposed 5-cycle city formula is clearly due to the presence
of cold starts in city driving (i.e., short trips). The method for estimating cold start fuel
consumption is via the difference between fuel consumption over Bags 1  and 3 of either
the FTP or cold FTP.  The driving pattern during both bags is exactly the same.  The
only difference is that Bag 1 begins with a cold start and Bag 3 begins with a hot start.
Thus, the difference in fuel consumption over the two bags is clearly due to the
presence of the cold start in Bag 1.  This difference has been used in  emission
modeling, such as that performed in the EPA MOBILE models for thirty years and is well
established.
                                      74

-------
      Honda did not suggest excluding the impact of cold starts from the 5-cycle
formulae.  Nor do we believe that they meant to imply that the impact of cold starts
should not be included. Still, this indicates that their concern about the negative
coefficient is theoretical, and not practical.  For example, there are a number of ways to
improve fuel economy over Bag 3 of the FTP, such as reducing vehicle weight or
aerodynamic drag, or modifying gear ratios, etc.  However, in the vast majority of these
cases, fuel economy over Bag 1  will improve as well, since it involves exactly the same
type of driving. As seen in Table 5-2, the coefficient for Bag 1 vastly exceeds that of
Bag 3. Thus, even if the change in Bag 1  fuel consumption was somewhat smaller than
that of Bag 3, 5-cycle city fuel economy would improve.  In addition, since Bag 2 of the
FTP and the SC03 test consist of very similar, low speed driving,  the vehicular changes
described above would likely improve fuel consumption over these other bags and
cycles as well, further improving  5-cycle city fuel economy.

      Honda did not provide any examples of technologies or techniques which could
improve fuel consumption over Bag 3, but which would not improve fuel consumption
over any of the other bags and cycles. Thus, we do not believe that the existence of a
negative effective coefficient for an individual cycle or bag is necessarily a problem in
the 5-cycle formulae.

      The negative coefficients for Bags 2 and 3 in the 5-cycle highway formulae are
primarily due to the inclusion of the effect of air conditioning in the formulae. The effect
of air conditioning is estimated via the difference in fuel consumption over the SC03 test
and a mix of Bags 2 and 3 of the FTP. Again, it seems extremely unlikely that a
manufacturer could improve fuel consumption over Bags 2 and 3 of the  FTP without
also improving fuel consumption over SC03, since the driving patterns are quite similar.
Thus, again, we believe that Honda's concern in this  case is more theoretical  than
practical.

      Regarding the effective coefficient of 14% for SC03, Honda states that the driving
conditions represented by this test only occur 3% of the time on the road.  Honda did
not explain or provide data to support the basis for this percentage. However, it is likely
that the 3% represents the percentage of onroad driving which occurs at 95 degrees
Fahrenheit or higher. If so, this characterization  is very biased, because it assumes that
if driving occurs at a temperature below 95 degrees Fahrenheit, it is adequately
represented  by the other bags or cycles. This is clearly not the case. Drivers use their
air conditioning systems at temperatures below 95 degrees Fahrenheit.  None of the
other bags or cycles involves operation of the air conditioning system. Thus, to
completely exclude driving at lower ambient temperatures when the air conditioning is
operational when assessing the applicability of the SC03 test to onroad  driving is
technically unjustifiable.

      The contributions of fuel consumption over SC03 in the 5-cycle formulae
explicitly consider the variation in use of air conditioning over the  range of ambient
conditions, the cycling of the compressor at various ambient conditions and the effect of
                                      75

-------
air conditioning on fuel consumption as a function of vehicle speed.  Honda neither
challenged any of these analyses, nor provided suggestions for their improvement.
Overall, these analyses (which are described in the TSD) indicate that the air
conditioning compressor is engaged about 15% of the time on the road.  Since the
SC03 cycle essentially consists of city driving, it is not surprising that the weighting of
the SC03 test in the 5-cycle city formula is very close to this figure.  In addition, the
SC03 results are combined with other results to approximate the effect of the air
conditioner compressor on fuel economy irrespective of ambient temperature, and this
impact is then weighted to reflect that the compressor is on during approximately 15%
of VMT. Thus, the actual percent of VMT represented by the SC03 driving cycle is not
directly relevant, as the test data is used to isolate the impact of the compressor load
irrespective of driving cycle.

      Starting with the cold CO test, we agree with Honda that only 2% of national VMT
occurs below 20F. However, to cite this statistic as the only indication of the relevance
of the cold FTP test is misleading. This implies that the standard FTP at 75F is a better
representation of the remaining 98% of driving, particularly cold starts. As indicated in
Figure 1 of Honda's comments, this is not the case. The FTP at75F is not a better
indicator of fuel economy at temperatures just slightly above 20F than the cold CO test.
A temperature of 50F is roughly halfway between the nominal temperatures of the two
FTP tests. Figure 1 indicates that roughly 30-35% of national VMT occurs at
temperatures which fall closer to the 20F test than to the 75F test. Thus, when
examined  in greater depth, the 20F cold FTP is not as "extreme" as it appears at first
blush, especially given the fact that the weighting factor for the various bags of the cold
FTP in the 5-cycle formulae are between 0.24 and 0.30.

      This is even more true for the SC03 and US06 tests.  Honda points out that the
SC03 test  is performed at 95F, when again only 2% of national VMT occurs at this
temperature or greater. However, the SC03 test is an air conditioning test, not simply
an FTP at  95F. Our analysis indicates that the air conditioning system is turned on
23.9% of the time in-use (Section III.A.3. of the Final TSD).  As discussed there, the
National Energy Research Laboratory estimates even higher use, 29%. An FTP
performed at 75F with the air conditioning off is not a better indicator of the incremental
fuel use associated with air conditioning than the SC03 test, even if the temperature  is
below 95F.  Thus, the percentage of driving occurring with the air conditioning on is
much more relevant to the benefit of the SC03 test than the percentage of driving
occurring at or above the temperature of the SC03 test.

Drivers rarely use their air conditioning below 70 F. Thus, the amount of driving below
70 F is irrelevant to the representativeness of the  SC03 test. When the air conditioning
system is turned off, it has no effect on fuel economy and our analysis accounts for the
fact that it  has no effect on fuel economy.  The issue  is how to best estimate the impact
of air conditioning on fuel consumption. The 5-cycle formulae isolates the impact of air
conditioning at 95F by comparing the fuel consumption over the SC03 test to that over
a combination of Bags 2 and 3 of the FTP designed to match the driving cycle of the
SC03 test. The 5-cycle formulae then weights this incremental fuel  consumption by 1)
                                      76

-------
the percentage of driving occurring with the air conditioning turned on, 2) the
percentage of time that the air conditioning compressor is on relative to that occurring
during the SC03 test, 3) the relative load of the compressor on the engine at the
temperature at which the driving is occurring, and 4) the speed of city and highway
driving relative to that of the SC03 test. Each adjustment accounts for differences
between the specific conditions occurring during the SC03 test and those existing on
the road. The alternative is to assume that the impact of air conditioning is proportional
to fuel use over the FTP, a test where the air conditioning is not operating.  Honda
presents no data to indicate that this alternative is a better estimate of the impact of air
conditioning. Available data of fuel consumption over the SC03 and FTP tests indicates
that the impact of air conditioning is not proportional to  fuel use over the FTP (see
Figure 111-11 of the Draft TSD).  Thus,  Honda's statement that only 2% of in-use driving
occurs above 95F does not address the major issue, that of air conditioning and its
effect on onroad fuel economy.

      In response to Honda's comment that the vehicle in the SC03 test has been
sitting for 10 minutes in the sun prior to the test, we do  not believe that this is extreme.
Prior to driving, vehicles can be parked under a roof, but often they are parked in the
open. Our in-use data on vehicle starts indicates that only 1 % of all starts in-use occur
after a vehicle soak of 10 minutes or less. Thus, when  vehicles are parked outside in
sunny conditions, their inside temperature is likely much higher than that in the SC03,
even when the ambient temperature is below 95F.  While the test is relatively short, 10
minutes or 3.5 miles, this is exactly our estimate of the  average trip length during "city"
driving.  Thus, with respect to both its  vehicle soak time and its length, the SC03 cycle is
not "extreme".

      Honda's comments about the "extreme" cycles representing half to 85% of the
contributions to  the 5-cycle formulae extend the same thinking to the US06 cycle. Few
if any specific vehicles are driven exactly like any of the bags or cycles.  Thus, they are
all extreme in this sense. As examined in Section III of the TSD, the US06 cycle is also
extreme in the sense that only a small percentage of onroad driving tends to fall  outside
of its speed-acceleration envelope.  But like the contributions of the SC03 and cold FTP
tests, the issue with  the US06 contribution is which bag or cycle best represents onroad
driving, not whether one or the other bag or cycle exactly matches it. Honda neither
presented any new onroad driving activity data, nor suggested any way to improve the
analyses which  determined the relative contributions of the various bags or cycles to
describe onroad driving.  As described in Appendix A of the TSD, recent California
chase car data indicates that 20% of city driving time and 41 % of highway driving time
fell  outside of the speed-acceleration envelope of the FTP and HFET cycles. This does
not imply that the US06 cycle can only represent these percentages of city and highway
driving.  It just means that these percentages of driving in California are not represented
by even one second of driving on either the FTP or HFET cycles. The weighting of
driving at the edges  of the FTP/HFET speed-acceleration "envelope" may be much
lower than those found on the road. The VSP-based cycle analysis of city driving
indicated that the US06 contribution was much lower than the 20% figure based  on the
FTP/HFET speed-acceleration envelope.  In contrast, the VSP-based cycle analysis of
                                       77

-------
highway driving indicated that the US06 contribution was much higher than the 41%
figure based on the FTP/HFET speed-acceleration envelope.  Both types of analysis
show that the contributions of the various bags and cycles, including those referred to
as extreme by Honda, appear to be appropriate at this time.
5.4   Start Fuel Use May Not Be Accurate on Per Vehicle Basis

What we proposed:

      In the proposal we assessed the sensitivity of the 5-cycle equations to several
factors and alternative assumptions. We evaluated the effect of soak time and ambient
temperature on start fuel use using test results from 50 degree  F testing required by
California. Based on this data, we concluded that it was unlikely that uncertainty in the
effect of ambient temperature on start fuel use would significantly affect city 5-cycle fuel
economy.  We also concluded that it would have no effect on 5-cycle highway fuel
economy, due to the extremely low contribution of start fuel use in highway driving.
Hybrids would likely show the greatest variability in this area, due to the greater number
of technological factors that could be affected, but even for these vehicles we found that
the 5-cycle fuel economy for the vehicle reflecting the greatest difference in temperature
sensitivity was only changed 1%.

What commenters said:

      UCS commented that the comparison of start fuel use at 50 degrees and 20
degrees Fahrenheit presented in the Draft Technical Support Document should not be
interpreted too broadly.  They suggest that the model may correctly predict the average
start fuel use for the nine vehicles,  it is not clear that each individual vehicle's start fuel
use is accurately reflected.  They point to the data for hybrid vehicles, which indicates
that the appropriate weighting for the 20 degree start fuel use may vary by a factor of
two between different vehicle models.  They suggest that it will  be important to validate
the approach  for individual vehicles, rather than on the basis of a group average.

Our response:

      UCS' comment about the possibility that the correct weighting factors for 20F
and 75F operation might differ between individual vehicles was addressed with the
previous comment by Honda.  As discussed there,  the current label procedure, as well
as the mpg-based equations, not only assumes the same rate of change of start fuel
use with temperature, but the same overall degree of change between 20F and 75F.
Both the degree of change  occurring between 20F and 75F and the rate at which this
change occurs are important.  The 5-cycle formulae utilize actual vehicle test data to
obtain the degree of change between 20F and 75F, while assuming that the rate of
change between 20F and 75F matches that of the vehicles which have been actually
tested at intermediate temperatures, as well as at 20F and 75F. The other two
approaches assume both factors (degree of change between 20F and 75F, and rate of
                                      78

-------
change between those temperatures), and assume all vehicles react to changes in
temperature in the same way in both regards. The 5-cycle approach is clearly superior
in this regard.
5.5   Relative Accuracy of the mpg-Based Equations Versus the 5-cycle
      Formulae for Individual Vehicles

What we proposed:

      We proposed the use of the mpg-based equations for model years 2008 through
2010, then implementation of the vehicle-specific 5-cycle method starting with the 2011
model year.

What commenters said:

      Honda commented: "We believe EPA has done a credible analysis based upon
air quality modeling methodologies to create an  equation that incorporates new, more
aggressive driving conditions to predict aggregate fuel economy for the fleet in actual
use. Reducing the equation as EPA has done into the mpg-based equation works
because this is not attempting to make the results vehicle specific.  However,  little in-
use data exists to validate the coefficients in the 5-mode equation for use in predicting
vehicle specific results.  Thus, Honda supports the 2008 mpg-based approach but we
do have some concerns with applying the formula to individual vehicles in 2011."

Our response:

      These two statements reflect a basic misunderstanding about the relationship
between the three approaches to developing fuel economy label values discussed in the
NPRM: 1) the current label formulae, 2) the mpg-based equations, and 3) the  5-cycle
formulae. All three approaches produce vehicle specific fuel economy label values.
The foundation of all three approaches is a set of measured fuel economy values based
on the testing of the vehicle being labeled. Thus, all three approaches are "vehicle
specific." All three approaches also include "fleet average" adjustments to the "vehicle
specific" fuel economy measurements. The difference between the three approaches is
in the degree of vehicle specificity, not a fundamental difference where two of the
approaches are appropriately fleet average and  the third  (5-cycle) is vehicle specific.

      Honda appears to be looking only at the portion of the various fuel economy label
formulae beyond the current FTP and HFET testing. Viewed in this way, the current
label formulae apply generic city and highway percentage adjustments to all measured
FTP and HFET fuel economy values.  While slightly more elegant in its mathematics,
the mpg-based equations do essentially the same thing, as the effective percentage
                                      79

-------
adjustment applied to the FTP and HFET fuel economy values varies only slightly
except for roughly three hybrid vehicles with extremely high city fuel economy.26

      The 5-cycle formulae, in contrast, introduce measured fuel economy values from
other test cycles.  Because of this, Honda appears to perceive these fuel economy
values as "vehicle-specific adjustments" to the FTP and HFET fuel economy values.
However, these additional three tests are no more vehicle-specific in concept than those
measured over the FTP and HFET tests. Thus, the distinction being drawn by Honda is
more related to history (how are the current "vehicle specific" fuel economy labels
changing) than based on fundamental engineering or technical considerations. Going
back further historically, the original fuel economy label values were based solely on the
FTP. The use of the HFET for highway label values was introduced at a later date.
This modification was in fact a vehicle specific adjustment to  the original, FTP-based
label values.  (Fuel economy measured over the HFET is obviously vehicle specific.)

      Viewed much more simply, the current label procedures utilize vehicle specific
data on the fuel economy of vehicles driven under relatively mild city and highway
driving conditions at 75F. The driving  cycles for these tests were developed from fleet-
average driving surveys which consisted of much less data than was used to develop
the driving cycles contained in the US06 and SC03 tests. The FTP and HFET tests
have never been justified on a vehicle specific basis in the  way that Honda seems to be
implying needs to be done for the three additional tests.

      However,  no effort was made in the past to determine if the FTP and HFET were
the "best" indicators of city and highway fuel economy, respectively.  The FTP and
HFET were simply two reasonable driving cycles with average speeds in the range
desired and were believed to provide reasonable estimates of onroad fuel economy.
Obviously,  in  1984, it was shown that these cycles significantly over-estimated onroad
fuel economy and required adjustment.  This was done in a fleet-average manner.
However, neither EPA nor any other organization concluded that there were no vehicle
specific factors which contributed to the shortfall indicated by the data. In fact, as
described above, the impact of high speed, aggressive driving, air conditioning and cold
temperature on fuel economy, while being correlated to a vehicle's fuel economy over
the FTP or HFET, has a vehicle  specific component which  can often  be traced to its
specific design (e.g., power to weight ratio, body style, etc.).

      Also, while the FTP fuel economy can appear to be a single fuel economy value,
it is really the result of weighting of three fuel economy values, one for each test "bag."
The weightings of these bags, particularly Bags 1 and 3, were based on the same type
   26  It is important to note that this does not indicate that the mpg adjustment approach and the current
label approach use similar methodologies or similar types of assumptions, as they do not. The mpg
approach is basically a way to estimate an individual vehicle's 5-cycle result, thus it is a surrogate for the
5-cycle approach, and not a different approach relying on different assumptions or data.  It turns out that
for the large majority of cars the 5-cyce results do tend to cluster in ways that are similar to using a
common adjustment factor


                                       80

-------
of "fleet-average" data (i.e., the in-use distribution of vehicle soak times prior to engine
start-up) that were used in developing the weighting factors in the 5-cycle formulae.
The 43% weighting of Bag 1 and 57% weighting of Bag 3 have never been validated on
a vehicle-specific basis, as Honda is saying is required for the 5-cycle formulae.  The
driving cycles which comprise the FTP and HFET tests were based on much less
onroad vehicle driving activity data than those used to develop the weighting factors for
the 5-cycle formulae. Even the weighting factors for Bag 1/3 and Bag 2 of the FTP used
in the current label formulae have never been "validated" on a vehicle specific basis.
Thus, Honda's perception that the current fuel economy label formulae do not need to
be validated, but the 5-cycle formulae require such validation, is clearly arbitrary and not
based on any real technical distinction.

      The same is true for the assumed trip lengths for city and highway driving. The
city fuel economy label is based on the FTP'S length of 7.5 miles (with respect to cold
starts), while the highway fuel economy label is based on an infinite trip length (i.e., zero
cold starts).  Likewise, the FTP and HFET are both performed at a nominal 75F.
These factors have also never been validated on a vehicle-specific basis.

      In fact, all of these factors implicit in today's fuel economy labels and the
proposed 5-cycle labels would differ slightly for individual vehicles. The volume of fuel
needed to start and warm up the engine as a function of soak time and ambient
temperature varies  between vehicles.  If all vehicles were thoroughly tested across the
wide range of situations encountered in-use and the results condensed so that a cold
start at 75F was averaged with a hot start at 75F, the fraction of cold starts would
differ. This can be easily shown using the data on start fuel use presented in the TSD.
The issue is not that the appropriate weighting factor would be exactly the same for
each individual vehicle design, but that the weighting factors do not vary too widely. If
this were the case,  the solution, again, would be to measure fuel economy under more
conditions. The solution would not be to return to generic, fleet-average adjustments.

      It should also be noted that Honda did not provide any data supporting the ability
of the FTP and HFET to accurately reflect any additional fuel consumption  related to
other driving patterns, air conditioning and defroster use,  and cold temperature,
including the effect of heating the passenger compartment. Nor did they provide any
data to indicate that the US06, SC03 or cold FTP tests reflected fuel economy impacts
not indicative of those occurring in-use. In fact, even a simplistic analysis of engine and
vehicle design and their effect on fuel economy would demonstrate that specific
vehicles are going to respond differently to the factors not covered by the FTP and
HFET tests. Honda itself argues that vehicles respond differently to such factors. Thus,
their own arguments, when appropriately applied to the currently label procedures and
the proposed mpg-based equations, highlight the limitations in these two approaches
relative to the proposed 5-cycle formulae.

      The 5-cycle formulae assume that the US06, SC03 and cold FTP tests, in
addition to the FTP and HFET, contain vehicle specific information with respect to
onroad fuel economy. Honda makes a number of specific arguments which attempt to
                                       81

-------
challenge this position which we address below.  However, overall, Honda provides no
vehicle activity or fuel economy data which demonstrate that the US06, SC03, and cold
FTP tests do not provide information which is relevant to estimating a vehicle's onroad
fuel economy.  It is clear that vehicles are often driven faster and more aggressively
than the FTP and HFET and at temperatures below 75F, and drivers often use their air
conditioning. The NPRM also  provide substantial fuel economy data that the impact of
these factors have a differential effect on specific vehicle's fuel economy. Thus,  broadly
speaking, Honda did not provide any justification for not utilizing this information  while
still trusting the information provided by the FTP and HFET tests.  Therefore, we find
this distinction  being made by Honda incorrect and reject their suggestion that the
coefficients in the 5-cycle formulae need a level of justification not required by the
current label procedure and mpg-based equations.
5.6   mpg-based or "Generic" Adjustments to FTP and HFET

What we proposed:

      The mpg-based method that we proposed bears some similarity to the current
method, in that the method makes adjustments to FTP and HFET test results.
However, the mpg-based adjustments we proposed are not flat percentage amounts,
but differ based on the fuel economy achieved on the FTP and HFET tests.

What commenters said:

      Honda strongly supports the application of mpg-based adjustments to the FTP
and HFET fuel economy measurements.  In support of this, they state that "if there is
some error in the methodology, it is applied equally to all products."

Our response:

      We appreciate Honda's support for the first phase of our proposed changes to
the fuel economy label calculation procedures.  However, we disagree that the error in
the mpg-based equations, if they exist, are equivalent for all vehicle models.  A very
simple example demonstrates this. While relatively few vehicles are sold without air
conditioning today, this was more frequent in the past. Manufacturers would often sell
the same vehicle model with and without air conditioning. These two vehicle
configurations would achieve essentially identical fuel economy over the FTP and HFET
tests, since the air conditioning was never operative and the weight of the air
conditioning system itself is quite small compared to that of the vehicle and might not
even move the vehicle up to a higher inertia weight class. However, their onroad fuel
economy would clearly differ, due to the one vehicle's air conditioning use. Both the
current label procedure and the mpg-based adjustments would indicate that both
vehicles would achieve the same onroad fuel economy. This would be obviously
incorrect. The onroad shortfall for the vehicle equipped with air conditioning would be
greater than that for the vehicle without air conditioning.  Thus, applying a generic
                                      82

-------
adjustment to FTP and HFET fuel economy does not automatically minimize the
difference between onroad and label fuel economy.

      In fact, this comment by Honda directly conflicts with a later comment.  Later in
their comments Honda states that: "By ignoring the effect of changes in driving behavior
and conditions on the in-use fuel consumption of individual vehicles, the Proposal
implicitly assumes that all vehicles have the same relationship between fuel
consumption and real world driving conditions."  The proposed 5-cycle formulae does
this to some degree, in that some of its weighting factors are based in part on
interpolations of how fuel economy changes between values measured over two tests,
such as the cold FTP and FTP, or SC03 and the FTP.  However, the generic
adjustments recommended by Honda assume that the fuel economy of all vehicles
respond in the same way to any conditions not included in the FTP and HFET tests.
Ample data was presented in the Draft Technical Support Document which proved this
not to be the case for colder temperatures, air conditioning use and high speed and
aggressive driving.

      Honda appears to be willing to accept the vehicle specific fuel consumption
information which is generated by the FTP and HFET tests, but not any other tests.
This is again odd, since the driving cycles of both tests have never been representative
of any driving studies, even the studies upon which they were based.  For example, all
accelerations and decelerations which were found to occur  in-use which exceeded 3.3
mph per second were reduced to this value, due to problems with operating a vehicle
under these conditions on the older twin-roll dynamometers. These equipment driven
limitations in  the FTP and HFET cycles have continued to this date. Yet Honda makes
no mention of this problem.

      This comment by Honda overlaps significantly with a number of their other
comments. Because of this, we will present additional  analyses below which also bear
on the comment being addressed here.  However, we believe that even this cursory
evaluation shows that Honda did not present any information to justify accepting the
vehicle specific nature of the FTP and HFET tests, while rejecting all additional vehicle
specific information.  Nor did they supply any data to support their assertions in this
area.
5.7   The Three New Test Cycles are Extreme

What we proposed:

      We proposed that three test procedures currently used for emissions compliance
purposes be incorporated into the fuel economy estimate calculations.  These test
procedures are the US06, SC03, and Cold FTP.  These procedures allow the fuel
economy methodology to account for vehicle-specific responses to conditions that are
not represented by the current tests, such as high speed driving, rapid accelerations
and decelerations, the use of air conditioning, and cold temperatures.
                                      83

-------
What commenters said:

      Honda states that the conditions addressed by the three tests are "worst-case",
or in other words, rarely occur in-use.  Honda states, for example, that only 2% of
national vehicle miles traveled (VMT) occurs when the temperature is 20F or less.

Our response:

      Starting with the cold CO test, we agree with Honda that only 2% of national VMT
occurs below 20F.  However, to cite this statistic as the only indication of the relevance
of the cold FTP test is misleading. This implies that the standard FTP at 75F is a better
representation of the remaining 98% of driving, particularly cold starts.  As indicated in
Figure 1 of Honda's comments, this  is not the case. The FTP at 75F is not a better
indicator of fuel economy at temperatures just slightly above 20F than the cold CO test.
A temperature of 50F is roughly halfway between the nominal temperatures of the two
FTP tests. Figure 1 indicates that roughly 30-35%  of national VMT occurs at
temperatures which fall closer to the 20F test than to the 75F test. Thus, when
examined in greater depth, the 20F cold FTP is not as "extreme" as it appears at first
blush, especially given the fact that the weighting factor for the various bags of the cold
FTP in the 5-cycle formulae are between 0.24 and 0.30.

      This is even more true for the SC03 and US06 tests. Honda points out that the
SC03 test is performed at 95F, when  again only 2% of national VMT occurs at this
temperature or greater.  However, the SC03 test is an air conditioning test, not simply
an FTP at 95F. Our analysis indicates that the air conditioning system is turned on
23.9% of the time in-use (Section III.A.3. of the Final TSD).  As discussed there, the
National Energy Research Laboratory estimates even higher use, 29%. An FTP
performed at 75F with the air conditioning off is not a better indicator of the incremental
fuel use associated with air conditioning than the SC03 test, even if the temperature is
below 95F.  Thus,  the percentage of driving occurring with the air conditioning on is
much more relevant to the benefit of the SC03 test than the percentage of driving
occurring at  or above the temperature of the SC03 test.

      Drivers rarely use their air conditioning below 70 F.  Thus, the amount of driving
below 70 F is irrelevant to the representativeness of the SC03 test.  When the air
conditioning  system is turned off, it has no effect on fuel economy and our analysis
accounts for the fact that it has no effect on fuel economy. The issue is how to best
estimate the impact of air conditioning on fuel consumption. The 5-cycle formulae
isolates the impact of air conditioning at 95F by comparing the fuel consumption over
the SC03 test to that over a combination of Bags 2 and 3 of the FTP designed to match
the driving cycle of the SC03  test. The 5-cycle formulae then weights this  incremental
fuel consumption by 1) the percentage of driving occurring with the air conditioning
turned on, 2) the percentage of time that the air conditioning compressor is on relative
to that occurring during the SC03 test, 3) the relative load of the compressor on the
engine at the temperature at which the driving is occurring, and 4) the speed of city and
                                      84

-------
highway driving relative to that of the SC03 test.  Each adjustment accounts for
differences between the specific conditions occurring during the SC03 test and those
existing on the road.  The alternative is to assume that the impact of air conditioning is
proportional to fuel use over the FTP, a test where the air conditioning is not operating.
Honda presents no data to indicate that this alternative is a better estimate of the impact
of air conditioning.  Available data of fuel consumption over the SC03 and FTP tests
indicates that the impact of air conditioning is not proportional to fuel use over the FTP
(see Figure 111-11 of the Draft TSD). Thus,  Honda's statement that only 2% of in-use
driving occurs above  95F does not address the major issue, that of air conditioning and
its effect on onroad fuel economy.

       In  response to Honda's comment that the vehicle in the SC03 test has been
sitting for 10 minutes  in the sun prior to the test, we do not believe that this is extreme.
Prior to driving, vehicles can be parked under a roof, but often they are parked in the
open.  Our in-use data on vehicle starts indicates that only 1 % of all starts in-use occur
after a vehicle soak of 10 minutes or less.  Thus,  when vehicles are parked outside in
sunny conditions, their inside temperature is likely much  higher than that in the SC03,
even when the ambient temperature is below 95F. While the test is relatively short, 10
minutes or 3.5 miles,  this  is exactly our estimate of the average trip length during "city"
driving. Thus, with respect to both its vehicle soak time and its length, the SC03 cycle is
not "extreme".

       Finally, Honda characterizes the US06 test as representing the 99.5 percentile of
onroad speed and acceleration conditions.  Honda did not provide data to support this
statement.  The US06 cycle was constructed primarily from segments of the REP05
cycle, which represented 28% of onroad driving in 3 U.S.  cities in early 1990's.27 (One
hill of the US06 cycle came from a high speed, high load cycle developed by GARB that
was analogous to the REP05 cycle.) Thus, the US06 cycle represents less than 28% of
the urban driving monitored in these three cities at that time.  However, conditions have
changed  dramatically since the early 1990's in at least two important ways.  One,
freeway speed limits  have increased from 55 mph to 70 mph or even higher. Two,  the
power to weight ratio  of vehicles has increased dramatically over the past 15 years. For
example, between 1975 and 1990, the horsepower to weight ratio of new cars ranged
from 0.0320-0.0402 hp/lb.  In contrast, between 1990 and 2005, the horsepower to
                                                                          OQ
weight ratio of new cars ranged from 0.0402-0.0525 hp/lb, an increase of 20-30%.
Thus, it is inappropriate to judge any driving cycle, but especially one focused on high
speeds, using onroad driving activity data developed 15 years ago.

       One indication of how representative the US06 cycle  is for today's highway
driving is the comparison of its vehicle-specific power (VSP) frequency distribution to
   27  U.S. Environmental Protection Agency. Regulatory Impact Analysis (Final Rule) - Federal Test
Procedure Revisions. U.S. EPA, August 15, 1996.
   Website: http://www.epa.gov/otaq/regs/ld-hwy/ftp-rev/sftp-ria.pdf
   28 "Light-Duty Automotive Technology and Fuel Economy Trends: 1975 Through 2005". Appendix E,
U.S. EPA, EPA420-R-05-001, July 2005.


                                       85

-------
estimates of onroad highway driving.  Of particular interest are those VSP bins with the
highest power levels and highest fuel consumption rates.  Table 5-3 presents the VSP
frequency distribution for the five highest power bins for highway driving as represented
in the 5-cycle highway fuel economy formula (79% US06 highway bag plus 21 % HFET),
as well as the VSP frequencies from three independent estimates of current highway
driving onroad. All three sources of onroad driving, the EPA Draft MOVES2004 model,
EPA testing conducted in Kansas City, and California ARB sponsored chase car studies
conducted in California, are  described in the TSD.

Table 5-3: Comparison of  Driving Activity in High Speed, High Power
Modes

VSP
Bin
26
36
37
38
39
Total
Frequency of Driving in VSP Bins During Highway Driving
5-Cycle (79% US06
Highway)
1.7%
10.9%
12.8%
8.1%
12.3%
45.8%
MOVES
4.7%
10.1%
9.8%
8.0%
10.7%
43.3%
Kansas City (non-
hybrids)
1.0%
15.7%
6.5%
4.5%
1 1 .2%
38.9%
California
Chase Car *
0%*
12.5%
12.3%
10.6%
13.8%
49.2%
* Estimated from speed-acceleration frequency distributions. Value of 0% for bin 26 is likely an
artifact of the processing of driving activity between 40 mph and 50 mph into city and highway
driving, which is assumed to have a speed cut-point of 45 mph.

      As can be seen, the frequency of driving in any particular high power VSP bin
can vary by a few percent.  Overall, the total amount of driving in the five highest power
VSP bins in the 5-cycle highway formula varies from the onroad estimates by -3.4% to +
6.9%. Thus, with a 79% contribution to highway driving, the US06 highway bag cannot
be described as extreme with respect to several estimates of current onroad highway
driving.  In contrast, while not shown in the table, only 8.1 % of the HFET cycle falls into
these five high power VSP bins.  Thus, the HFET cycle by itself is clearly not a good
indicator of current highway driving.  However, if one had to choose between the HFET
cycle and US06 highway bag as being an "extreme" description of highway driving, from
the point of view  of fuel consumption, one would have to conclude that the HFET was
the more extreme cycle of the two, being extremely mild in its required  engine power
levels.
5.8   Weighting of New Cycles

What we proposed:

      We proposed weightings of the 5-cycle test data that were based on extensive
analyses of real-world and laboratory test data.
                                      86

-------
What commenters said:

      Honda states that: "The proposed 5-cycle coefficients assign most of the weight
of a given type of in-use operation to the extreme cycles."

Our response:

      Except for the representation of most of highway driving by the highway bag of
the US06 test, Honda's statement is clearly incorrect. The effective weighting factor for
air conditioning use is only 0.144 in city driving and 0.050 in highway driving.  These
factors are clearly well below 50% of vehicle operation. The factors for the various bags
of the cold FTP are between 0.24 and 0.30, again well below 50%. As discussed
above, we also disagree with Honda's claim that these cycles are "extreme."
5.9   Interpolation of Fuel Economy Using Five Cycles

What we proposed:

      Where made possible by available data, the 5-cycle method uses interpolation to
determine the fuel economy at intermediate conditions not represented directly by test
data. We found this to be more accurate and preferable to using extrapolation
techniques.

What commenters said:

      Honda states that "these extreme cycles cannot represent the broadband and
midpoints of driving and fuel economy performance.  Thus, the 5-mode proposal
implicitly assumes that interpolation of fuel consumption can properly represent all
conditions on all vehicles. This assumption is valid only if there is a smooth relationship
between fuel consumption and driving conditions without non-linear steps and if all
vehicles have the same fuel consumption response to changes in driving condition.
Again, these assumptions have not been validated."  Further on, Honda states that:
"The proposed 5-cycle coefficients assign most of the weight of a given type of in-use
operation to the extreme cycles.  [This previous statement is addressed immediately
above.]  This is not necessarily representative of the entire range of the real world
driving conditions and the characteristics of the middle ground between the various
conditions represented in the 5-cycle equation."  Honda also states that the
discontinuities present in a number of newer fuel efficient technologies may cause the
proposed 5-cycle formulae to under value their efficiency (i.e., penalize them relative to
their benefit in the real world).

Our response:

      In this comment, Honda is basically arguing that the US06, SC03 and cold FTP
tests are so extreme that any interpolations based on them is fraught with error.  In fact,
                                       87

-------
as discussed above, Honda is basically arguing that extrapolating fuel economy effects
from only the FTP and HFET fuel economy values is better than interpolating fuel
economy effects using all five cycles.

      It is common knowledge that interpolation is generally more accurate than
extrapolation due to the availability of additional data.  Extrapolation is based on pure
speculation of the effect of the parameter being extrapolated.  Interpolation at minimum
bounds the effect, since the effect under intermediate conditions will fall between the
two conditions tested.  Thus, Honda is taking a very extreme position with this
comment. They are arguing that the three new tests are so extreme (i.e., their
conditions rarely occur in-use) that knowledge of fuel economy under these conditions
provides little value.

      In addition, Honda states that the effect of driving patterns, air conditioning and
colder temperatures on the fuel economy of modern vehicles is extremely non-linear
and even discontinuous in nature.  Therefore, they imply that knowledge and use of fuel
economy data over the three new test cycles will actually misrepresent fuel economy
over conditions  intermediate between these tests and the FTP and HFET tests.

      We have already shown above that the three new tests are not as extreme as
Honda asserts.  In addition, the concept of interpolation is much less applicable to the
US06 and HFET tests, than for the SC03, cold FTP and FTP tests.  For example, the
US06 driving cycle itself contains a wide range of driving, ranging from idle to over 80
mph, high rates of acceleration and deceleration,  as well as relatively steady state
cruising.  On average, the US06  highway and HFET driving patterns are quite different.
However, over 50% of the time spent driving in the two cycles occurs in VSP bins where
both cycles have significant operation (i.e., bins 0, 33, 35, and 36). Thus, it is not the
case where fuel consumption is being averaged at idle and wide open throttle. Both
cycles represent a range of highway driving.  The HFET represents relatively low speed
highway driving coupled with very low rates  of acceleration and deceleration.  The US06
highway bag represents higher speed driving coupled with some more severe
accelerations and decelerations.  Weighted together in the 5-cycle highway formula,
their combined driving  pattern  matches the best available description of onroad driving
patterns contained in the EPA's Draft MOVES2004 model, fairly well.  Thus, the 5-cycle
formula for highway fuel economy does not interpolate between the HFET and US06
driving patterns  as much as it combines them. The concept of interpolation is much
more applicable to the use of the SC03 and cold FTP data in the proposed 5-cycle
formulae.

      First, Honda did not provide any data to support the fact that fuel economy
effects developed from the US06, SC03 and cold FTP tests were not indicative of
changes in fuel economy which would occur at more intermediate conditions. Honda
claims that discontinuities in fuel  economy effects occur with modern vehicles, but again
provides no data to support this contention.  Therefore, we have to consider Honda's
concerns as hypothetical at this point.  Still, we consider the possibility for such a
hypothetical situation to exist below.
                                      88

-------
      In contrast, in many cases where EPA interpolated between the conditions
represented by two of the five dynamometer tests (e.g., ambient temperature, air
conditioning), test data at intermediate levels of the parameter of interest (e.g., cold start
fuel use at 50F) were used directly to develop the linear or curvilinear shape of the
curve depicting the effect of the parameter on fuel consumption. These data at the
intermediate temperatures indicated that vehicles generally responded to lower
temperatures in a similar manner (i.e., start fuel use increased), though there was some
variability between individual vehicle's response.  Some of this  difference may have
been due to the normal variability in fuel economy measurement.  However, some of the
difference in vehicles' response to colder temperatures is likely due to differences in
design.  It is helpful to look at an example.

      In developing the weighting factor for start fuel use, we used the results of FTP
testing 580 vehicles at a  variety of ambient temperatures (Section  III.A.1 of the TSD).
We further evaluated this database and found 295 vehicles which were tested at the
same three ambient temperatures,  15-35 F, 42-60 F, and 68-80 F. To this database,
we added the 13 Honda and Toyota vehicles described in Section III.A.1 of the TSD,
which were tested over the FTP at 20F, SOT, and 75F. We determined the start fuel
use for the cold start of the FTP at each of these temperatures  for each vehicle. The
average start fuel  use for the three  temperature ranges was:

      15-35F           0.0735 gallons
      42-60 F           0.0457 gallons
      68-80 F           0.0246 gallons

      The issue here is how to best estimate the start fuel use at the intermediate
temperature range.  Is it better to apply  a fleet-wide adjustment to the start fuel use at
68-80 F, or to interpolate between the start fuel use estimates for both the high and low
temperature ranges?

      The first approach is that used by the current fuel economy label procedure. This
is basically to determine the average impact of temperature on  start fuel use and apply
this to every vehicle. This can be done  in either of two ways: additive and multiplicative.
We will evaluate both ways here. The additive approach indicates that the start fuel use
at the intermediate range is 0.0211  gallons higher than that at 68-80 F. The
multiplicative approach indicates that the start fuel use at the intermediate range is
1.856 times higher than that at 68-80 F.

      The second approach is that used in the 5-cycle formula. This approach
estimates the start fuel use for the intermediate temperature range on the start fuel use
in the lowest and highest temperature ranges. This approach indicates that the start
fuel use at the intermediate range is 43.1 % of the way from the start fuel at 68-80 F to
that at 15-35 F.
                                      89

-------
      On average, all three approaches work. Any errors made in predicting the start
fuel use for individual vehicles always balance out.  The issue is which approach
provides the best estimate for each vehicle individually. We assess this by calculating
the coefficient of variation of the error between the estimated start fuel use at the
intermediate temperature and that actually measured.  The additive version  of the fleet-
wide approach produced a coefficient of variation of 34%, while the multiplicative
version of this approach produced a coefficient of variation of 37%.  The interpolation
approach produced a coefficient of variation of 24%, well below that of either fleet-wide
approach. The interpolation approach, which is that imbedded in the 5-cycle formulae,
clearly produces the smallest error in predicting start fuel use at intermediate
temperatures.

      We repeated this analysis for the 13 Toyota and Honda late model vehicles.  The
results were the same.  The additive and multiplicative extrapolation approaches
produced coefficients of variation of 20% and 19%, respectively. Interpolation produced
a coefficient of variation of 9%.  It is interesting to note that, while this group of vehicles
was small, it consisted of a wide variety of vehicle types and included four hybrids.
Honda appears most concerned about the application of the 5-cycle formulae to
advanced technology vehicles.  Yet when applied to exactly such a group, the
interpolation approach embedded  in the 5-cycle formulae performed well and much
better than the approach taken in the current label formulae and the mpg-based
equations.

      Thus, it is clear, at least for the effect of ambient temperature on cold start fuel
use, that interpolation is far superior to extrapolation.  Given that start fuel use is by far
the dominant effect of the cold FTP on 5-cycle fuel economy, this example essentially
addresses the main issue related to the use of the cold FTP in the 5-cycle formulae.  If
Honda concern was valid, the "extremely" low temperatures of the lowest temperature
range would have produced misleading predictions of start fuel use for the intermediate
temperature range. That this did not occur indicates that Honda's concern is invalid, at
least for this the effect of ambient temperature on start fuel  use.

      We believe that it is useful to illustrate in a more conceptual  manner the
circumstances which must come together in order for Honda's concerns to become
reality. All three label procedures, current, mpg-based and 5-cycle, incorporate the
effect of a variety of operational and environmental conditions on fuel economy.
However, they do so in different ways. We will focus our discussion on the current and
5-cycle procedures.  The mpg-based procedure basically represents the average effect
of the 5-cycle procedure adjusted for a vehicle's fuel economy over the FTP and HFET.
Thus, conceptually, it differs only slightly from the current procedure with respect to the
issue being addressed here.

      Figure 5-1 illustrates the change in relative start fuel consumption with a drop in
ambient temperature from 75F to 20F for three vehicles; one with a relatively low
response to temperature, one with an average response and one with a high response.
                                       90

-------
For simplicity, we assumed that start fuel use varies linearly with temperature in each
case. We will address the issue of non-linearity below.

  Figure 5-1.  Effect of Ambient Temperature on Cold Start Fuel Use
                                                        *  Low

                                                        A  Average

                                                        --High
             20   30   40    50    60    70    80

                       Temperature (F)
      The current label procedures are based on dynamometer testing at 75F.  Thus,
only start fuel use at75F is known for a specific vehicle. As mentioned above, the 10%
adjustment factor for city driving implicitly assumes that all vehicles' cold start fuel use
changes with temperature to the same degree.  No other projection can be made, since
no other information is available.  We know from available test data at both 20F and
50F that this is not the case.  The start fuel use of individual vehicles responds
differently to temperature. Thus, the current label procedure is clearly in error when it
assumes that this response is identical for all vehicles.

      The 5-cycle formulae base their estimates of start fuel use on testing at both
20F and 75F. Therefore,  it does not need to assume this effect.  This effect is based
directly on the fuel economy measured over both the FTP and cold FTP tests. Honda
does not appear to be acknowledging this direct use of test data in the 5-cycle formulae.

      The weighting factors for start fuel use in the 5-cycle formulae are based on an
assumed shape of the curve of start fuel use in between 20F and 75F.  Figure 5-1
shows this effect to be linear.  However, in reality it can be linear, curvilinear, or even
discontinuous, as pointed out by Honda. Figure 5-2 shows the change in relative start
fuel use versus temperature for three vehicles which all show the same total effect of
moving from 75F to 20F.  One vehicle's response is linear, another's is curvilinear
(increasing slowly at first and then accelerating as the temperature approached 20F),
                                      91

-------
while the third vehicle's is linear with a discontinuity at 50F.  The last vehicle's
discontinuity might be due to the shut-off of some fuel saving technology, like engine
shut off when the vehicle is stopped.

   Figure 5-2. Possible Shape of Start Fuel Use Versus Temperature
      0
          1.8
S  1.6
Li.

?  1.4
0  1.2
            1
                                                         *  Linear

                                                         A  Curvilinear

                                                        -*- Discontinuous
              20   30   40   50   60   70   80
                       Temperature (F)
      In-use, vehicles are operated over a range of ambient temperatures. From a
driving perspective, the average ambient temperature is just under 60 F (see Section
III.A.4 of the TSD).  Driving is more frequent at this temperature that it is at higher or
lower temperatures.  Average start fuel use in-use is then the start fuel use at any
specific temperature weighted by the frequency of starts at that temperature. If vehicles
were tested over small intervals of ambient temperature, for example every 5 F, the
differences in the three vehicles' start fuel use responses shown in Figure 2 would be
fully reflected in the predicted average fuel consumption. However, since vehicles are
only tested at20F and 75F, the weighting of these two test results needed to produce
the average start fuel use with respect to ambient temperature would differ to some
degree.

      For example, the weighting of start fuel use at 20F for the vehicle with the
curvilinear response would be lower than that for the vehicle with the linear response.
This would be the case, because the vehicle with the curvilinear response would have
lower fuel consumption at every temperature in between 20F and 75F and would
necessarily have a lower start fuel use than the vehicle with the linear response. (We
are ignoring what happens below 20F and above 75F in this example, which is
relevant in the real world.)
                                      92

-------
      There is no way to reflect this difference in the current label procedures. The two
vehicles' start fuel use at75F is reflected in their FTP tests and incorporated into the
label values. However, the increase in start fuel  use at lower ambient temperatures is
based on the fleet-average vehicle and comprises a portion of the 10% and 22%
generic adjustment factors. Thus,  as discussed above with respect to the vehicles
depicted in Figure 1, the current label procedure  assumes that all vehicles have the
same relationship between start fuel use versus temperature.

      The 5-cycle procedure includes the ability to reflect differences in vehicles' start
fuel use at both 20F and 75F. However, it, like the current label procedures, cannot
reflect differences in the shape of  the curve in  between these two temperatures. Thus,
the 5-cycle procedure is better than the current procedure in one respect, but still not
ideal. Additional testing at intermediate temperatures would be required to further
improve the accuracy of the label  value.  However, Honda and other manufacturers
clearly  recommend against this, and currently  EPA believes it is unnecessary.

      The mpg-based procedure  converts the individual vehicle responses estimated
by the 5-cycle formulae to their fleet-wide averages as a function of FTP or HFET fuel
economy.  If the effect of ambient  temperature on start fuel use in relative terms
changes with vehicles' FTP fuel economy, then the mpg-based procedure will reflect
this.  However, if the differences in vehicles' response to temperature are due to other
factors, then the mpg-based procedure reduces basically to the current label procedure
with a different adjustment factor.  In fact, the mpg-based equations create a city fuel
economy label by reducing FTP fuel economy by 20-28%; the adjustment to HFET fuel
economy is 29-30%. Thus, the mpg-based equations  reflect a portion of the difference
in individual vehicle responses to various factors which affect city fuel economy.
However,  regarding factors which  affect highway fuel economy, the mpg-based
equation is nearly identical to the current label procedure with a higher downward
adjustment.

      It is impossible to theoretically predict how the average start fuel use for the
vehicle with the discontinuous response to ambient temperature would compare to that
for the  other vehicles. Its average start fuel use,  and thus,  its weighting of start fuel use
at 20F, could be higher or lower depending on where the discontinuity occurs (e.g., 30
F or 60 F) and the degree that fuel consumption changes at the discontinuity. None of
the three label procedures is capable of accurately accounting for such discontinuous
effects of ambient and operation conditions on fuel economy. Aside from testing
vehicles over a much wider and comprehensive set of ambient and operating
conditions, the only issue is how to use the available test results to best estimate
onroad fuel consumption.  Again, we can use the example of the effect of ambient
temperature on start fuel use to evaluate which of the three procedures best uses the
available information to predict onroad fuel economy.

      As  shown in Figures 5-1  and 5-2, vehicles can vary in their total response to  a
change in temperature from 20F to 75F, as well as in the rate of change in this
response as temperature moves from one end to the other.  The 5-cycle formulae
                                      93

-------
clearly can account for the total difference in this response between the dynamometer
tests which are conducted. Thus, the only potential inaccuracy in the formulae is due to
differences in the shape of the curves between the two endpoints.  With the current
label procedure, neither the degree of the change in fuel consumption between 20F
and 75F,  nor the shape of the curve is known. Thus, all vehicles are assumed to have
the same relative change  in fuel consumption at every ambient temperature other than
75F.  The only way that the 5-cycle formulae could produce a worse estimate of onroad
fuel economy would be if vehicles' response to ambient temperature between, for
example, 40 F and 75F was inconsistent with their response at 20F.  Even in this
situation, the degree of error would need to be relatively large to be worse than the
current label procedure, since the latter completely misses all differences in vehicles'
response to any relevant factor.

      While this is theoretically possible, it requires that the change in start fuel use at
20F differ so substantially from that at intermediate temperatures that vehicles with
higher increases in fuel consumption at 20F have lower increases at, for example, 45
F, and vice versa.  In other words, the types of curves shown in Figures 5-1 and 5-2
have to criss-cross. It is highly unlikely that differing degrees of curvature in a vehicle's
response could produce this result.  The most likely situation would involve a severe
discontinuity in a vehicle's response. Figure 5-3 shows this situation.

   Figure 5-3. Potential Source of Error in the 5-Cycle Weighting Factors
      
-------
response. However, over most the temperature range shown, the vehicle with the
discontinuous response has the lower increase in start fuel use. Thus, the onroad fuel
consumption of the vehicle with the discontinuous response is likely to be the lower of
the two vehicles. However, a weighting of the start fuel use at 20F and 75F would
indicate the opposite.  In order for the use of the 20F measurement to cause this
problem, the discontinuity indicated in Figure 3 must be both large and occur close to
20F.

      This is the situation about which Honda appears to be concerned. While
theoretically possible, we have no fuel economy data which indicate such a severe
discontinuity exists with any currently marketed vehicle or any future technology.
Honda submitted no data to support the existence of such  severe discontinuities.
Honda also submitted  no data indicating that such  discontinuities would occur in a way
that the vehicle would  be in high fuel consumption mode frequently over the SC03,
US06 or cold FTP tests,  but less frequently in-use compared to the weighting of these
tests in the 5-cycle formulae. As presented above, the existing data clearly support the
superiority of the interpolation approach followed in the 5-cycle formulae.

      Practically, such a large discontinuity only occurs when a piece of equipment or a
technology turns off or on.  Examples would be engine shut off at vehicle idle or mild
driving and cylinder deactivation at low engine loads.  Such technologies are currently
applied in only a very limited number of models. A practical example related to colder
temperatures could involve use of the heater and its effect on the engine shut-off
strategy for hybrids. Most hybrids save fuel by shutting the engine off when the vehicle
is idling.  However, many current hybrid systems disable this engine shut-off when the
heater is turned on, as the heat comes from the engine.  Thus, the savings  related to
engine shut-off disappear in the range of ambient temperatures when people tend to
turn on their vehicle heater. If people tend to turn on their heater just above 20F, then
the situation indicated  in Figure 3 could occur.  However, people tend to turn on their
heater well above this  temperature.  The weighting factor for the cold  FTP with respect
to start fuel use is 24%, while that for running fuel use is 30%. In the case of the heater
affecting engine shut-off, as long as the heater was turned on 24-30% of the time in-
use,  even if this caused a large discontinuity in fuel use, the 5-cycle formulae would
provide a reasonable estimate of onroad fuel consumption. Roughly 35% of all driving
occurs below 50F, while about 21 % occurs below 40 F. Nearly all people will have
their heater turned on at 40 F.  Many, if not most drivers will turn on their heater at 50F.
Thus, in this case, the  weighting factor for the fuel consumption over the cold FTP
appears quite appropriate with respect to the impact of heater use on  engine shut-off
and fuel consumption.

      In contrast, we have ample evidence that vehicles vary in their fuel economy
response to ambient temperature. The 5-cycle formulae allow these differences to be
incorporated into the label values. The current formulae do not. The mpg-based
equations only allow them to be incorporated to the extent that the differences are a
function of FTP or HFET fuel economy. Other differences  in the response to
temperature are lost with the mpg-based equations. The only advantage of the current
                                      95

-------
or mpg-based equations would be to avoid the hypothetical situations where the
incorporation of the three new tests over-estimates significantly the impact of these
conditions on onroad fuel economy value.

      It is interesting to note that the mpg-based equations have the same potential for
error as the 5-cycle formulae should the hypothetical technologies which reflect severe
discontinuities become the predominant technology in the new vehicle fleet. In this
situation, 5-cycle fuel economy values of most of the vehicle fleet would reflect the
problem asserted by Honda. The mpg-based regression of these 5-cycle fuel economy
values would fully reflect the effect of the 5-cycle weighting of the three new tests.

      In addition to interpolating start fuel use between 20F and 75F, Honda also
cites concerns about interpolating between: 1) the low highway speeds and power
levels of the HFET and the high highway speeds and power levels of theUSOS highway
bag and 2) the 75F non-air conditioning operation of the FTP and the 95F air
conditioning operation of the SC03 test. As we have already discussed, the 5-cycle
formulae does not interpolate between fuel economy over the HFET and the highway
bag of US06, as much as it combines them.  Some in-use highway operation occurs like
that of the HFET and some occurs like that of the US06 highway bag. Both cycles
include a range of vehicle speeds and power levels.  Together, there are no large gaps
in vehicle operation in terms of VSP, unlike that in between the cold  and standard FTP.
Thus, Honda's concern about interpolating in between the HFET and US06 fuel
economy values is unfounded.

      Regarding air conditioning use, there are at least a couple of ways in which the
weighting factors in the 5-cycle formulae were developed utilizing interpolation. We
estimated the percentage of time that the compressor is engaged at temperatures and
humidities both above and below those existing during the SC03 test. We also
estimated the change in load which the compressor places on the engine at
temperatures other than 95F.  Individual vehicles might respond differently to  these
"off-cycle" conditions.  Had we used fuel economy data over these intermediate
conditions for various vehicles, we might have developed somewhat different weighting
factors for the effect of air conditioning use in the 5-cycle formulae. However, as stated
a number of times already, the alternative is to simply apply the same factor to all
vehicles' fuel economy based on its fuel economy over the FTP and HFET. This is
highly uncertain and clearly wrong in several instances.  One, vehicles without air
conditioning would receive the same adjustment as those with air conditioning. Second,
larger vehicles with optional rear passenger air conditioning would receive the  same
adjustment with or without the option. Under the 5-cycle approach, the vehicle with the
larger or second compressor would achieve a lower fuel economy over SC03 and a
lower 5-cycle fuel economy.

      In Section 5 of their comments, Honda also cites data referenced in the Draft
TSD that indicates that the engagement of the air conditioning compressor over SC03
varied between vehicles. Honda states that these data show that the assumption about
air conditioning compressor engagement based on fleet wide averages do not  apply to
                                      96

-------
individual vehicles.  We reevaluated the data cited by Honda and it appears that Honda
misunderstands how the weighting factors of the proposed 5-cycle formula work in
practice for a specific set of vehicles. This same is true for the current label procedure.

      Honda cites data from six vehicles on the percentage of time that the air
conditioning compressor is disengaged during the  SC03 test.  Honda states that these
percentages range from 3% to 37%.  The implication is that a single value cannot
represent both 3% and 37% and therefore, the weighting factor in the proposed 5-cycle
formulae must be wrong.  First, the 5-cycle formulae do not assume any compressor on
time for a specific vehicle operated over SC03. This percentage is implicit in the SC03
test itself.  The vehicle  is operated over the SC03 test, the air conditioning compressor
is engaged to cool the cabin as would be the case onroad under the ambient conditions
of the test and this compressor engagement increases fuel consumption in the same
way. The fundamental impact of running the air conditioning system is determined by
the test, not by some assumed formulae.

      The 5-cycle formulae do assume that the impact of air conditioning under more
moderate conditions is a specific fraction of the impact over a short trip at 95F.  If a
vehicle shows a relatively high impact of air conditioning over SC03, then  the proposal
assumes that this vehicle will also show a relatively high impact at more moderate
conditions. The same assumption is made regarding a vehicle showing a relatively low
impact of air conditioning over SC03. The data cited by Honda actually confirms the
reasonableness of this assumption.

Table 5-4 shows the percentage of time that the air conditioning compressor is engaged
during the SC03 test.9 We show the percentage of time that the compressor is
engaged, rather than disengaged, since air conditioning increases fuel use when the
compressor is engaged, not when it is disengaged. Three test conditions are shown.
The first is the official test temperature of 95F with the required solar load.  The second
set of runs was performed at 75F, again with solar load. The third set of runs was
performed at 75F without any solar load, but with  the heater turned all the way up to
simulate the effect of solar load on compressor engagement.

      Table 5-4. Air Conditioning Compressor Engagement over SC03 (% of
      time)

Escort
Mustang
Town Car
Bronco
Windstar
Corolla
95F
63
93
97
97
97
97
75F With Solar Load
40
53
66
76
68
68
75F With Heater On
44
71
95
95
98
100
   29
      SAE Nam paper on A/C load
                                      97

-------
      As indicated by Honda, the percentage of time which the compressor was
engaged varied between the vehicles.  However, the primary issue with respect to the
weighting factors for air conditioning use in the proposed 5-cycle formulae is the
correlation between compressor engagement over the SC03 test and that at other,
usually less severe conditions. We performed a regression between the percentage of
time which the compressor was engaged over SC03 and that at75F with solar load.
The adjusted r-square was 0.760. The p-value for the factor relating the two
compressor usages was 0.022, indicating a very strong statistically significance. The
same type of regression was performed between compressor engagement over SC03
and that at 75F with the heater turned on and the  results showed an even stronger
correlation (adjusted r-square of 0.827 and a p-value of 0.008.  These strong
correlations indicate that the degree of compressor usage over SC03 is a very good
indicator of compressor usage at lower temperatures. Since this is the precise
assumption made  in developing the weighting factor for the impact of air conditioning on
fuel economy in the 5-cycle formulae, these data are a strong confirmation of the
proposal and do not refute the formulae, as asserted by Honda.

      While these data confirm the proposed 5-cycle formulae, they do refute the
assumption inherent in the current label formulae.  The current label  procedure utilizes
no vehicle-specific information regarding compressor engagement or the impact of air
conditioning on onroad fuel economy. The impact  of air conditioning on onroad fuel
economy is assumed to be the same for all vehicles and is implicitly included in the
generic 10% and 22% adjustment factors applied to the FTP and HFET fuel economy
measurements. The test program producing the above data did not report fuel
economy, so we cannot evaluate the impact of air conditioning use on a vehicle specific
basis for these vehicles. However, we know from the certification database presented
in the NPRM that there is some variability in vehicles' fuel economy response to air
conditioning. The  data in Table 5-4 confirm this difference by indicating that the
compressors of some vehicles run all the time over SC03, while others run much less.
Thus, the implicit assumption in the current label procedures that the impact of air
conditioning is a constant percentage for all vehicles is challenged by the data from this
test program.

      In summary, Honda suggests that we ignore significant amounts of valid fuel
economy obtained with the three new tests in order to avoid a purely hypothetical
situation where the use of these data might produce a less realistic estimate of onroad
fuel economy than the current procedure or the mpg-based equations.  In doing so,
Honda ignores the fact that the current label procedure (and the proposed mpg-based
procedure) involves more extensive assumptions about the response of various
vehicles' fuel economy to ambient and operational conditions than the proposed 5-cycle
formulae. A generic adjustment to FTP and HFET fuel economy produces a fuel
economy label which is just as vehicle specific as the 5-cycle formulae. Assuming that
all ambient and operational conditions which differ from the FTP and HFET have the
same effect on each vehicle's fuel economy does not automatically minimize error.  In
fact, as will be shown below for start fuel use at colder temperatures, a generic
adjustment produces a much larger error than using fuel economy values measured
                                      98

-------
over the FTP and cold FTP tests.  This should be no surprise.  Additional testing
provides additional information which allows more accurate fuel economy estimates to
be developed.  In fact, if cost was  not an issue, even more testing covering an even
wider set of driving patterns and ambient conditions would be beneficial.

      Put a slightly different way,  Honda has pointed out a number of uncertainties
associated with the proposed  5-cycle formulae which we agree are present. What they
fail to acknowledge is that the same type of evaluation of the current label procedure
and the proposed mpg-based equations would reach the same conclusion, only more
so.  Thus, we conclude that Honda's comment that the proposed 5-cycle formulae
assume that all vehicles have the same relationship between fuel consumption and real
world driving conditions is false. In contrast, it is actually true for the current label
procedure which we propose to replace with the 5-cycle fuel economy formulae.
5.10  Misuse of Emission Factor Models

What we proposed:

      We proposed weighting factors in the 5-cycle formulae that were developed
using fleet-average driving activity and fuel consumption data.  The basis for some of
these weighting factors was data that are currently used in our MOBILE and MOVES
emission factor models.

What commenters said:

      Honda goes through the development of the 5-cycle formulae as described in
detail in the Draft TSD and  identifies 15-20 areas where EPA used fleet-average activity
or fuel consumption data to develop the weighting factors  in the formulae. One example
is the change in a vehicle's fuel consumption as the ambient temperature decreases
from 75F to 20F.  Honda believes that EPA should validate that the weighting factors
developed in each of these areas applies to individual vehicles, as well as to the fleet-
average vehicle.  Honda's concerns are increased by the belief that modern computer
controls allow significant non-linearity to be introduced into vehicle operation. While
historically, the fuel consumption of conventional vehicles  might have increased
gradually as ambient temperature decreased, the fuel consumption of modern vehicles
might change suddenly at some specific temperature. A weighting of fuel consumption
at 20F and 75F based on fleet average data reflecting the gradual change might not
be appropriate for a specific vehicle exhibiting the sudden  dramatic change. Most
importantly, Honda believes that without such validation, the 5-cycle formulae as
proposed might produce a worse estimate of onroad fuel economy than the current
label procedures.
                                      99

-------
Our response:

      We do not take issue with Honda's comment that the weighting factors in the 5-
cycle formulae were developed using fleet-average driving activity and fuel consumption
data.  Nor do we claim to have validated these factors on  an individual vehicle basis. In
fact, we presented some data in the Draft TSD which indicates that a set of 5-cycle
formulae "tuned" to an individual vehicle would likely differ to some degree (e.g., the
comparison of appropriate weighting factors for start fuel use at 20F for hybrids versus
conventional vehicles in Section III.A of the TSD). However, we disagree with Honda
that these two facts make the proposed 5-cycle formulae somehow inappropriate or
unjustified. In order to demonstrate this, we address each of Honda's concerns below.

      Regarding Honda's concern that fleet-averages were used to develop the
weighting factors, this approach is the same as that used  to develop the current fuel
economy labels. For example, the city fuel economy value is 90% of the fuel economy
value  measured over the FTP. The 90% factor was based on the difference between
onroad fuel economy estimated for a large number of in-use vehicles (i.e., the fleet) and
the average of these vehicles' FTP fuel economy values.  This factor should have been
larger for some vehicles and smaller for others. However, the only way to know this for
certain for individual vehicles is to measure the onroad fuel economy of each vehicle
and compare it to its FTP fuel economy.  However, once the onroad fuel economy is
available, there is no need for an adjustment factor, one could just use the onroad fuel
economy value and skip measuring fuel economy over the FTP. Of course, the major
problem with this approach is that onroad fuel economy values would only be available
after a considerable number of vehicles of the particular model have been sold and
operated for some time. It is also difficult  to obtain a fully  representative sample.  Thus,
the 90% factor represents an acceptable adjustment factor on average.

      The FTP itself includes at least two aspects which were based on fleet average
estimates.  First, the driving cycle was based on the average speeds of individual
vehicles which were driven over a particular road route in  Los Angeles circa 1970.
Different vehicles of various power-to-weight ratios and other design criteria were likely
driven over this road route differently. Also, vehicles are obviously driven on many
other types of road  routes during "city" driving.  The relationship between each vehicle's
average onroad fuel economy and its fuel economy over LA road route #4 will vary.
Yet, all vehicles' city fuel economy label values are based on their fuel economy over
this single road route. The same is true for the highway fuel economy label.

      The other aspect of the FTP which is based on fleet average estimates is the
57% weighting of fuel consumption following a hot start (i.e., 10 minute soak) and the
43% weighting of fuel consumption following a cold start (i.e., 12 hour soak).  Vehicles
are started in-use after a wide range of soak times. Even  if every vehicle model was
found to have the same distribution of pre-start soak times in-use, the relative weight of
hot start fuel use and cold start fuel use which  produced each vehicle's average start
fuel use in-use would likely vary.  The current FTP weighting factors of 43% and 57%
                                      100

-------
represent factors which were determined to be reasonable on average in the early
1970's.

      Neither of these two aspects of the FTP have these been validated on an
individual vehicle basis in the way Honda asserts must be done in order for the 5-cycle
formulae to be valid. Numerous studies have shown a statistical correlation between
FTP fuel economy and "city" fuel economy and between HFET fuel economy and
"highway" fuel economy.  However,  as will be discussed below regarding fuel economy
data obtained from Strategic Visions, there is a high degree of correlation between the
fuel economy values measured over all five of the driving cycles.  Thus,  the fact that
past studies showed a good correlation between FTP fuel economy and onroad city fuel
economy means very little.  A similar analysis would have indicated a good correlation
between SC03 FTP fuel economy and onroad city fuel economy, even though air
conditioning is used much less than half of the time in-use. The same would be true for
a correlation between US06 fuel economy and onroad highway fuel economy. As the
Honda analysis  indicates, even fuel economy over HFET is a good predictor of city fuel
economy.  No study has been performed which shows or "validates" that the FTP and
HFET are the appropriate tests on an individual vehicle basis.

      The driving patterns contained in the FTP and HFET, as well as the cold/hot start
weighting factors are examples of the use of fleet-average data and relationships
explicitly included in the current label procedure.  However, essentially every
relationship which we used to develop the 5-cycle formulae, and which Honda criticizes,
is implicitly included in the current label procedure, as well.  (Because the mpg-based
equations apply the differences between the 5-cycle label values and  FTP  or HFET fuel
economy values, the mpg-based equations involve these same relationships, as well.)
For example Honda criticizes EPA's development of the weighting factor for the
highway bag of US06 in the 5-cycle highway fuel economy formula.  However, the
current highway label procedure assumes that the effect of driving patterns which differ
from the HFET is the same for all vehicles. This factor is not explicitly identified. It is
simply a portion of the overall 22% downward adjustment applied to HFET fuel
economy.  However, it is assumed to be identical for all vehicles.  Honda's own
comments state that they believe that the fuel economy of individual vehicles will
respond to different driving patterns differently. The assumption that all  vehicles
respond identically has never been validated on an  individual basis.  If such an attempt
were made, it would likely fail. The  differential relationship between fuel  economy over
the HFET and US06 highway bag is also evidence of this difference.

      As stated above,  unless vehicles are tested over all possible modes of operation
and ambient conditions occurring in-use, some judgment must be applied to, in a sense,
estimate fuel economy over the untested conditions. We believe that it is much better to
perform  this estimation via interpolation between actual fuel economy measurements,
as opposed to extrapolating from a single fuel economy measurement, such as the FTP
or HFET. The proposed 5-cycle formulae involve the actual measurement of fuel
economy over ten individual sets of operation and conditions. The current label
procedure involves three measurements. The fuel economy effect of the assessed
                                     101

-------
range of operation and ambient conditions included in these ten distinct measurements
is determined for each vehicle directly from the testing of that vehicle.  These
measurements involve no assumptions. The current label procedures, as well as the
proposed mpg-based equations, do not involve the direct measure of fuel economy at
the extra seven sets of operation and conditions. These approaches must "assume"
these fuel economy effects.  All approaches have to estimate fuel economy for
operation and conditions not tested.  However, with the proposed 5-cycle formulae,
these assumptions are usually bracketed by real test data.  With the current and mpg-
based procedures, there are not extra data points which provide such  bracketing.
Estimation of fuel economy during all operation and conditions not included in the FTP
or HFET are assumed based on fleet-wide relationships of the type criticized by Honda.
Thus, we must reject Honda's repeated assertions that the 5-cycle formulae are
inherently deficient because these relationships have not been "validated" for individual
vehicles.  Honda has simply failed to understand the greater role of these same
relationships in the current and mpg-based procedures.

      Honda has also inconsistently applied its own criticism of the use of fleet-wide
fuel economy relationships  by assuming that fleet-wide adjustments to FTP and HFET
fuel economy, as is done with the current and mpg-based procedures, is inherently
accurate, while criticizing the use of such relationships under more restricted conditions
in the development of the weighting factors for the 5-cycle formulae.
5.11  Derivation of Vehicle-Specific 5-Cycle Method

      5.11.1    Start Fuel Use

            5.11.1.1     Start Fuel

What we proposed:

      We proposed that start fuel consumption for a vehicle be determined by
calculating the difference between its FTP Bag 1  (cold start) fuel consumption and its
FTP Bag 3 (hot start) fuel consumption.

What commenters said:

      UCS commented that the proposed methodology is reasonable in principle, but it
is valid only if the vehicle is fully warmed up by the end of Bag 1 of the FTP.  If the
vehicle is not fully warmed up until Bag 2, then the start fuel use would be the difference
in fuel use between Bags 1 and 2 combined and  that of Bags 3 and 4 combined, with a
4-bag FTP. By only considering the difference between Bags 1 and 3, they argue, the
proposed methodology may omit some start fuel  use that occurs in Bag 2, thus
underestimating the start fuel consumption  and overestimating the running fuel
consumption during Bag 2. As a potential indicator that this might be the case,  UCS
points to the data in Table III.A-22 in the Draft TSD, where in many cases the Bag 4 fuel
                                     102

-------
economy is higher than the Bag 2 fuel economy, especially at 20 degrees F.  They
further note that no data points to hybrids as being unique in this regard, and thus it is
possible, they postulate, that conventional vehicles may also  not be warmed up by the
end of Bag 1. They suggest that a modest test program on a dynamometer could
determine whether vehicles are warmed up after Bag 1, Bag 2, or not until later, and 75
and 20 degrees F.

      They further argue that different vehicles must be tested in similar and consistent
ways in order to ensure fairness. Start fuel for hybrids, tested on the 4-bag FTP and
where the vehicle is "allowed" to warm up through Bags 1 and 2, might produce results
inconsistent with a conventional vehicle tested on a 3-bag FTP, where the vehicle is
assumed to warm up by the end of Bag 1.

Our response:

      As described in the Draft TSD, testing at 75F has assumed that vehicles are
fully warmed up after Bag  1 for over 20 years.  As also discussed in the Draft TSD, it is
less clear that this is true at 20F. If the vehicle is not fully warmed up after Bag 1 at
either temperature, its cold start fuel use  will be under-estimated, as pointed out by
UCS.  However, the vehicle's warmed up fuel use, which is taken in part from Bag 2, will
be over-estimated.  Thus,  the net impact  on estimated fuel use would be less than that
indicated just by the under-estimation of start fuel use.

      Throughout this rule, we have made it a premise to avoid adding to the current
tests which are performed. There have been exceptions to this premise, such as
requiring diesels to be tested at 20F. However,  this change  is the extension of a
testing requirement already applicable to nearly all vehicles to a few vehicles which had
been granted an exception. We have also proposed some changes to the test
procedures, such as splitting the  US06 test in to two bags and requiring the heater to be
turned on during the cold FTP test.  However, these changes would not extend the test
and would have little long term impact on test costs.

      In contrast, requiring all vehicles to be tested over a 4-bag FTP would  increase
the cost of every FTP both in terms of dynamometer and personnel time.  This would
particularly be true for cold FTP testing. The increase in accuracy of the fuel  economy
label due to this change is unknown, but  likely to be small. Table III.A-21  in the Draft
TSD presents the result of a DOE test program performed at  several temperatures. It
found fuel economy over Bag 1 to be  26% lower at 20F than at 75F,  while this
difference was only 8% for both Bags 2 and 3.  Roughly 2% out of these differences
was likely due to difference in summer and winter fuel. Thus, even if all of the
remaining difference in Bag 2 and 3 fuel economy was due to further warm up, this 6%
difference would only represent one-fourth of that found in Bag 1.  Some of this
remaining 6% difference is likely due to factors normally associated with colder
temperatures, such as increased friction in the drivetrain. Thus, the three-bag FTP
could  be capturing more than 80% of  the cold start fuel use and part of the remainder
would be captured through an increase in running fuel use.
                                     103

-------
      We believe that this is adequate for the purpose of fuel economy labeling.  It is
certainly better than the current procedure where we assume that every vehicle has the
same percentage increase in start fuel use at colder temperatures relative to its start
fuel use at 75F.  To the decrease that a small amount of start fuel use is missed by the
current cold FTP test procedure, this is captured in a generic way through the non-
dynamometer factor.
            5.11.1.2     Trip Length

               5.11.1.2.1   City/Highway VMT Split of 43/57 Percent

What we proposed:

      The combined fuel economy using the current label formulae is a 55/45 harmonic
weighting of the current city and highway fuel economy labels. For the proposed 5-
cycle formulae and related analyses, we used a 43/57  harmonic weighting of the 5-cycle
city and highway fuel economies to represent combined overall fuel economy.  This
city/highway split for the 5-cycle fuel economies was based on:
      1) The assumption that driving generally less than 45 mph is city driving and that
above 45 mph  is highway driving, and
      2) The description of onroad driving patterns contained in MOVES.

      Note that there are two separable issues here: one relating to the 5-cycle
equations and  related background analyses that have  little direct impact on the label,
and the other relating to those items on the label that are directly influenced by the
city/highway split and highly visible to consumers (e.g., the average annual fuel cost
and combined fuel economy).  It may in fact be reasonable to use different weightings in
these two different cases. This section addresses the  weightings used in the 5-cycle
analyses; Section 2.0 addresses the use of the combined fuel economy on the label
itself.

What commenters said:

      AAM/AIAM  notes that the proposed change in the EPA city/highway split, from
55/45 to 43/57, suggests that today's  drivers are more  frequently traveling at speeds
greater than 45 miles per hour, or at least more frequently than twenty years ago when
the adjustment factors were first developed. They point out that the proposed change
assumes that drivers commuting on the road would get there more quickly today than
they would have were they to drive the same route twenty years ago. AAM/AIAM finds
that this is statistically valid on a miles-driven basis, but they believe that their analysis
reveals that EPA's proposed change does not take into account modern-day
congestion. As a result, they point out that the methodology produces counter-intuitive
predictions, such as most drivers are driving at highway speeds 57% of the time as EPA
reports in their NPRM. They believe that the current 55/45 split aligns more closely with
                                     104

-------
the way many customers think of their driving patterns - a 50/50 split - than the
proposed split, which hints at a precision that departs from many drivers. Being
unconvinced that a change is needed, AAM/AIAM recommends that EPA pursue further
data collection and validation, which include the effects of modern-day congestion.

Our response:

      The split of U.S. driving into city and highway fractions was based on an analysis
of estimates of U.S. driving as a whole contained in the Draft MOVES2004  emissions
model. This split is inherently subjective, as it can involve both the immediate driving
pattern of the vehicle, as well as the overall driving pattern of the trip.  Various drivers
are highly likely to use different definitions and would estimate different driving splits
even if they were to drive exactly the same route at the same time of day. The
city/highway split of 43/57 was developed using the approximate boundary  of 45 mph,
applied to a segment of driving, and not on a second-by-second basis. This worked
well for the various driving cycles used in MOVES2004 to represent onroad driving
activity.  However, in real life, many trips would contain driving segments in both the city
and highway categories which would be difficult to separate in a quantitative way.

      As described in Section III.E of the TSD pertaining to the comparisons of fuel
economy label values to onroad fuel economy estimates, it is most accurate to compare
combined fuel economy in both cases. This avoids the need to split driving in to city or
highway categories. The most important factor in this case is that the city and highway
driving be re-combined in the same as it was originally split.  As outlined in  Section  III.F
of the TSD, we evaluated alternative definitions of city and highway driving  which in turn
changed the split of VMT between the two categories.  None of these  alternatives
appeared to carry any advantages, particularly in terms of being able to be  simulated by
the various dynamometer cycles.  We have no new data or insight which would  change
this conclusion.

      Regarding the impact of congestion on the split between city and highway
driving, we believe that AAM/AIAM is confusing the impact of congestion on driving time
and mileage.  Congestion increases the length of a trip in terms of time, but does not
necessarily change its mileage. Thus, congestion by itself would not be expected to
increase the city fraction of VMT.  The 43/57 split is based on VMT, not time. It is
possible that congestion would shift  some driving which was formerly at highway
speeds to that which is now considered to be city driving. This would tend to increase
city fraction of all driving.

      The other trend which has been occurring for the last 30 years  is urban and
suburban sprawl.  This tends to increase trip length, as confirmed by the National
Personal Travel Survey results. Urban beltways and other new freeways have allowed
relatively efficient additional travel to occur with city driving at either ends.  At the same
time, some of these freeways become congested at rush hour and convert to city
driving.  This congestion likely increases city driving less than might be thought, again
due to the difference between time and mileage. It takes a long time to  move a mile on
                                      105

-------
a congested freeway. Even those people who spend most of their time in what seems
like a parking lot likely achieve most of their mileage at a much higher speed.

      The final factor affecting the comparison of the current 55/45 and the proposed
43/57 city/highway split is the fact that they were based on different definitions of city
and highway driving, as outlined in  Section III.A.2 of the TSD.  The 55/45 split was
consistent with the urban/rural split of VMT as estimated by FHWA in the late 1970's
and early 1980's.  As indicated by the characterization of driving patterns in
MOVES2004 and  supported by common sense, more highway driving occurs on
freeways in urban areas than low speed driving occurs in rural areas.  Thus, breaking
down city and highway driving by vehicle speed is likely to yield a lower fraction of city
driving than a breakdown based on urban driving. This is likely the biggest factor
yielding the change in the city/highway split.
               5. 1 1. 1.2.2   Highway Trip Length

What we proposed:

      We proposed a highway trip length of 60 miles.

What commenters said:

       Mitsubishi expressed concern over the use of the denominator of "60" in the
following equation, found in proposed 600.1 14-08(b). They commented that EPA did
not clearly explain why the number is used, and how it was derived, arguing that it
appeared to be arbitrarily chosen. UCS likewise expressed some concerns with EPA's
assumptions regarding trip length in the 5-cycle equations  and how they may impact the
calculated fuel economy, especially given EPA's language that the trip length of 60
miles was "somewhat arbitrary."  They agree with EPA's assessment that once highway
trip length is over 30-40 miles the denominator in the equation makes little difference,
but they argue that it has not been clearly established that  the average in-use highway
trip is indeed greater than 30 miles.

           , ,  77         , x         ( (0.76 x StartFueL, + 0.24 x StartFueL
       StartFC (gallons per mile) = 0.330 x  - - ^ -
              V6      f     }
      More importantly, UCS argues, is the fact that highway trip length has an indirect
impact on city fuel economy because EPA has fixed the city/highway mileage ratio and
the average trip length, thus making city trip length a de facto function of highway trip
length. They recommend that EPA focus on determining more accurately the length of
the average city trip, then allowing the highway trip length to  be determined as a
function of the city trip length, if necessary.
                                      106

-------
Our response:

      EPA's derivation of the trip lengths for all driving, city driving and highway driving
is explained in great detail in Section III.A.1 of the TSD.  Many vehicle trips in real life
involve a mix of city and highway driving. There is no single method for assigning these
trips to either type of driving. Thus, it is impossible to determine a precise estimate of
trip lengths for either city or highway driving. One can estimate the length of the
average vehicle trip from studies of vehicle operation. In the TSD, we estimate this
length to currently be 7.68 miles. The estimates of city and highway VMT coupled with
the trip lengths for city and highway driving must be consistent with this overall trip
length of 7.68 miles.

      The city/highway split of VMT in the  U.S. of 43/57 comes from our analysis of the
characterization of driving contained in the  MOVES2004 model and our definition of city
and highway driving.  This estimate is independent of any assumptions regarding trip
length. Given an estimate of either city or highway trip length, the other trip length is
determined by the other information. Table 5-5 shows how the two trip lengths  interact.
Table 5-5. Relationship Between City and Highway Trip Length (miles)*
City
3.0
3.5
4.0
4.5
Highway
-48.7
60**
24.3
16.2
City
5
6
7
7.7
Highway
12.8
9.7
8.3
7.7
   * Figures based on more precise city/highway VMT split of 42.5/57.5.
   ** Rounded, per proposed 5-cycle formulae.

      As can be seen, relatively small changes in city trip length produce large
changes in highway trip length. Highway trip length becomes infinite for a city trip
length of 3.27 miles, so city trips cannot be shorter than this length.  Also, highway and
city trip length become equivalent for a city trip length of 7.7 miles.  Since it is counter-
intuitive that city trip would be longer than highway trips, city trips cannot be longer than
this length.

      Regardless of the specific city and highway trip lengths, the excess fuel use
associated with engine starts will be accounted for in either the city or highway fuel
economy estimate.  The only issue is where to concentrate the start fuel use.  The
current procedure concentrates all measured start fuel  use in city driving.  The HFET is
a hot start test.  Assigning a finite 60 mile trip length to  highway driving decreased 5-
cycle highway fuel economy by 0.5%.  Because of the interaction between city and
highway trip length,  this increased city fuel economy by 0.4%.  Decreasing the highway
trip length from  60 to 20 miles would decrease 5-cycle highway fuel economy by 0.9%,
while increasing city fuel economy by 1 %.  This change would tend to bring the city and
highway fuel economy label values closer together (by  roughly 2%). A change of this
magnitude would only affect the actual label values of a very small fraction of vehicles
given that labels are rounded to whole mpg figures.
                                       107

-------
      Changing the highway trip length to a lower value than 60 miles would be no less
arbitrary than the original 60 mile estimate. Many drivers commute 60 miles to work, so
this distance fits many drivers' situation.  It is also long enough that any "city" driving at
the beginning and end of the trip is likely to be relatively short. Drivers also often
assess their fuel economy on even longer trips. Thus, an estimate made using  a 60 mile
trip length is still very applicable to an all day highway trip, for example. At the same,
many drivers commute 20 miles to work and might appreciate a shorter trip length than
60 miles.  One number will not be preferred by all drivers or interested parties.  At this
time, we believe that a wider difference between city and highway fuel economy
highlights the difference in driving patterns and provides useful information to the
consumer.  Thus, we do not choose to change the trip length for highway driving to a
lower figure.

      As described above, there is an inter-relationship between the trip length for city,
highway and all driving.  As described in Section III.A.1  of the Draft Technical Support
Document, the average trip length for city driving was estimated to be 3.5 miles. This
estimate was based on two estimates of trip length.  One source was  a set of
instrumented vehicle studies conducted in the early  1990's in support of EPA's
Supplemental FTP rulemakings.  The other source was the  National Household Travel
Survey (NHTS), which shows a national average trip length  of 9.8 miles.

      The results of the instrumented vehicle studies are shown in Table 5-6 below.
This is the same information that was presented in Table III.A-7 of the Technical
Support Document for the NPRM. The estimate of average trip length used in
MOBILE6.2 is based on the  average trip lengths found in Baltimore and Spokane, which
is believed to be the most unbiased methodology of those used in the various cities.

 Table 5-6.  Estimates of In-Use Average Trip Length
Location

Baltimore - Exeter
Baltimore - Rossville
Baltimore - Combined
Spokane
Atlanta
Los Angeles
MOBILE6.2
Average Trip Length (miles)
Instrumented Vehicle Studies
4.0
5.9
4.9
3.6
6.0
Not Available
4.49
Chase Car Studies
Not Available
Not Available
7.5
5.8
Not Available
7.8
Not Applicable
      As pointed out in the Draft TSD, there is a fundamental inconsistency between
the MOBILE6.2 trip length of 4.49 miles and the NHTS trip length of 9.8 miles.  The
MOBILE6.2 estimate should be considered to apply to urban driving, while the  NHTS
estimate represents a nationwide average. Still, given that urban driving represents
                                      108

-------
over half of all U.S. driving, the MOBILES.2 trip length results in more urban trips than
the NHTS estimate implies to occur nationwide.

      In the NPRM, we removed this inconsistency by reducing the NHTS trip length to
7.7 miles based on a presumption that a survey of this nature misses trips which are of
a very short duration, or counts a sequence of trips that are interrupted by very short
engine off times as single trips.

      Since the time of the NPRM,  EPA obtained more recent data on the operation of
vehicles in Atlanta from Georgia Tech.  Georgia Tech has been instrumenting vehicles
in the Atlanta area for some time and gathering operational data. As with the previous
instrumented vehicle studies, this type of study captures all vehicle trips no matter how
short. The major difference between this more recent work and the previous studies is
the amount of data being collected.  To date, Georgia Tech has collected data on over
620,000 vehicle trips.  In contrast, the previous instrumented vehicle studies involved
about a week's worth of operation for less than a hundred vehicles per city.  Thus, the
recent data represents almost two orders of magnitude more data than was evaluated in
the instrumented vehicle studies described in Table 5-6 above.

      The average number of trips per day found for these 620,000 trips was 4.62.
Unfortunately, the analysis of the number of miles traveled per trip is still underway and
will not be available until later this year.  Nationally, vehicles travel about 34 miles per
day. Urban vehicles might drive somewhat less than rural vehicles.  However, given
that Atlanta is a sprawling metropolis, its driving is probably close to the national
average in terms of miles traveled per day. Assuming this, the average trip length in
Atlanta is about 7.4 miles.  This is 24%  greater than the average trip length of 6.0 miles
found in the early 1990's.  The series of NHTS  have also found a general increase in
trip length nationally of about 1.0 mile between 1990 and 2001.
            5.11.1.3    Formula for Start Fuel Use

What we proposed:

      We proposed the following formula for start fuel use, using 3.59 and 3.91 as the
distances for the first and second bags of the FTP, respectively.

                 Start Fuelx forvehicles testedover a 4 -bag FTP =


                          7.5                     7.5
                 f    3.59       3.91    1   f   3.59       3.91
                             Bag2FEJ   {Bag3FEx  Bag4FE
                                      109

-------
What commenters said:

      Mitsubishi noted that 40 CFR 86.144-90 describes the distance of the first
bag/phase of the FTP as 3.598 miles, and the distance of the second bag/phase as
3.902 miles. They point out that the proposed equation in 600.114-08(a) uses figures
slightly inconsistent with these values. Mitsubishi recommends that EPA use the
published distances in 86.144-90, rather than the proposed values.

Our response:

      The regulations in 40 CFR 86.144-90 do in fact describe the distances as
Mitsubishi notes, but this is only in the context of an example calculation. For the actual
calculations from a given test, the manufacturer is required to use the measured driving
distance of the appropriate phase as calculated from the measured roll or shaft
revolutions.  This may or may not be equal to the distances shown in the example
calculations in 40 CFR 86.144-90.  In order to be consistent with other terms in the 5-
cycle formulae, we will use the figures from 40 CFR 86.144-90, but we will round off
these figures to one decimal place (i.e., 3.6 and 3.9 miles).
      5.11.2   Running Fuel Use

            5.11.2.2     Representative Mix of Dynamometer Driving Cycles

What we proposed:

      The Draft MOVES2004 emissions model describes on-road driving differently
than previous EPA emission inventory models.  While starting with whole driving cycles,
it goes further by breaking down vehicle operation on a second by second basis into 17
categories or bins.  One bin (Bin 0) contains significant decelerations. Another bin (Bin
1) contains idling operation.  The other 15 bins contain brief or modest decelerations,
cruising operation and accelerations. The 15 bins are broken down into three sets of
bins by vehicle speed: Bins 11-16 contain operation at 1-25 mph, Bins 21-26 contain
operation at 25-50 mph, Bins 33-36 contain operation at 51  mph or faster.  These three
sets of bins are further sub-divided according to the power required of the engine
divided by vehicle mass.  This ratio is termed vehicle specific power, or VSP, and has
the  units of kilowatt per megagram (kW/Mg).

      The VSP distributions for the four complete dynamometer cycles plus individual
bags of the FTP and US06 cycles are shown in Table III.A-15.  As was the case for the
Draft MOVES2004 inventory cycles, these VSP distributions represent a 50-50 mix of
cars and light trucks.
                                      110

-------
    Table 5-7. VSP Distributions for Dynamometer Cycles (% of time)
BinID
0
1
11
12
13
14
15
16
17
18
19
21
22
23
24
25
26
27
28
29
33
35
36
37
38
39
LA4
12.0%
18.6%
6.4%
10.6%
8.0%
5.0%
2.3%
0.7%
0.1%
0.0%
0.0%
4.4%
10.9%
9.5%
2.9%
1.5%
0.7%
0.2%
0.3%
0.2%
1 .6%
2.6%
1.1%
0.1%
0.2%
0.0%
HFET
3.5%
0.7%
0.0%
0.1%
0.3%
0.1%
0.5%
0.1%
0.0%
0.0%
0.0%
4.1%
3.6%
12.4%
17.3%
7.5%
2.5%
0.4%
0.3%
0.0%
6.1%
32.3%
6.0%
2.0%
0.1%
0.0%
US06
16.8%
7.5%
0.7%
0.8%
0.3%
0.5%
0.2%
1.0%
0.8%
0.5%
1.7%
1 .4%
0.7%
0.8%
0.1%
0.4%
0.2%
0.7%
1 .2%
4.5%
7.8%
14.1%
8.6%
10.3%
6.6%
12.0%
US06
City
32.6%
14.3%
1.3%
1.7%
0.9%
1.3%
0.4%
2.2%
2.2%
1.3%
3.5%
3.3%
1.5%
1.7%
0.2%
1.1%
0.4%
1.5%
2.4%
9.1%
4.6%
3.7%
0.4%
0.9%
0.9%
6.5%
US06
Hwy
7.1%
2.5%
0.3%
0.3%
0.0%
0.0%
0.0%
0.3%
0.0%
0.0%
0.5%
0.3%
0.1%
0.1%
0.0%
0.0%
0.0%
0.1%
0.4%
1 .6%
9.9%
20.7%
13.8%
16.2%
10.2%
15.5%
SC03
10.5%
19.7%
3.7%
9.2%
5.0%
2.4%
2.0%
2.1%
0.0%
0.0%
0.0%
7.7%
9.6%
10.3%
6.6%
3.9%
4.0%
0.0%
0.0%
0.0%
1.8%
0.6%
0.5%
0.4%
0.0%
0.2%
Bag3
13.3%
18.8%
2.8%
4.4%
3.0%
2.4%
2.9%
1.7%
0.4%
0.0%
0.0%
4.6%
7.3%
12.4%
4.5%
2.7%
2.0%
0.6%
0.9%
0.6%
4.3%
7.1%
2.9%
0.3%
0.5%
0.0%
Bag 2
1 1 .2%
18.5%
8.5%
14.3%
10.9%
6.5%
1.9%
0.2%
0.0%
0.0%
0.0%
4.3%
13.0%
7.9%
2.0%
0.8%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
What commenters said:

      Honda criticizes the development of the 5-cycle weighting factors for high speed
and aggressive driving because the highest VSP bins (bin numbers 19, 29, and 39 with
21 kw/Mg power or greater) start at a relatively modest level. About 18% of the US06
cycle falls into the highest power bins.  Onroad driving follows a bell shaped curve.
Operation is concentrated at the low end of each range of power levels, while US06
operation in this bin reaches as high as 50 kw/Mg.  Honda argues that this causes an
artificially high weighting for the US06 cycle. Honda also states that EPA misspoke
when it stated that the US06 driving cycle was a concentrated version of the REP05
cycle. US06 contains some operation from California's unified cycle, as well.

      Natural Resources Canada take some issue with the fact that the US06 cycle
contains speeds that are  above those legally permitted in Canada or the U.S., and they
                                     111

-------
question the high weighting factor of the US06 cycle in the highway fuel economy
estimate and whether it is too high for a fair representation of the fuel economy
performance of a vehicle. They don't argue with the methodology or the data sources
used to arrive at the factor, which they note are robust and among the best available,
but they are unable "...to justify [the maximum speed of the US06 cycle] to both industry
and consumers as being  suitable for use in a fuel economy test." They suggest one of
two ways to resolve their concerns: (1) Ignore the fuel consumption associated
seconds 290 through 350 of the US06  cycle  (i.e., a portion of the cycle with vehicle
speeds at roughly 80 mph) when calculating fuel economy; or (2) test a sample of
vehicles with and without the speed excursion between 290 and 350 seconds of the
cycle and evaluate the incremental fuel economy increase due to the 80 mph driving
segment.

Our response:

      As background to this discussion,  it might be helpful to review the discussion of
the concept of VSP and how it was used  to derive the cycle weights in the 5-cycle
formulae in  Section III.A.2. of the Draft Technical Support Document.

      First, Honda points out that 18% of operation during the US06 test falls into the
three highest power bins. Three out of 26 bins is 12%.  Given that US06 focuses on a
specific type of driving  (high speed, aggressive), it is not surprising that the distribution
of driving across the VSP bins would not be equal. For example, as shown in Table
III.A.-15 of the Draft Technical  Support Document, 32% of the operation during HFET
occurs in one VSP bin, bin 35.

      Second, Honda states that the lower limit of the three highest power VSP bins,
21 mW/Mg, is modest. Yet the highest power level achieved over the HFET by a typical
car or light truck (19 mW/Mg) does  not even  reach this level for one second. Only 3-
3.5% of vehicle operation over the HFET is above 12 mW/Mg, the lower limit for VSP
bins 16, 26  and 36. From their comments, Honda clearly prefers to base highway fuel
economy solely on HFET testing plus a generic adjustment factor based on HFET fuel
economy.  Honda does not address the issue of accurately predicting fuel consumption
at higher power levels using a  test with such low power levels.  Again, Honda is
applying criteria selectively to the proposed 5-cycle formulae and failing to apply it to the
current ormpg-based procedures.

      Honda does not provide any data to support its claim that vehicle operation is
concentrated at the lower end of each range of power levels.  Honda's statement
appears to  assume that vehicle operation can  be described by a single statistical
distribution  (i.e., driving activity is concentrated near the mean and diminishes in
frequency as operation varies more from  the mean).  However, driving is not likely to
occur randomly.  Speed limits, particularly those affecting highway driving, are
concentrated in a narrow range (i.e., 55-80 mph).  Thus,  driving activity might not
diminish further from the mean. Two estimates of highway driving, that contained in
Draft MOVES2004 and that based on California chase car data, indicate roughly equal
                                     112

-------
amounts of driving in bins 36, 37, 38, and 39.  Thus, it is doubtful that operation in these
bins is concentrated at the low end of the range of VSP included in each bin. Operation
in bin  39 could be an exception,  since it is open-ended.  Even with this bin, though, the
power level at which the frequency of operation tends to tail off is unclear, as is the rate
at which it tails off.

      Nonetheless, Honda's overall point likely has some validity.  EPA's 15 car study
using  a portable emission measurement unit (PEMS), which was described in the Draft
Technical Support Document, compared fuel consumption measured during
dynamometer tests to those measured onroad. In that study, the highest power VSP
bin was  36 (i.e.,  bin 36 included  all operation contained in VSP bins 36-39 as defined in
the  Draft Technical Support Document). We found that fuel consumption in bin 36 were
6%  higher over US06 than onroad. This was due to a higher average power level in bin
36 over  US06 than onroad. This was one of the reasons we expanded the number of
high power bins  beyond that used in the Draft  MOVES2004 model.  In particular, bins
16,  26 and 36 were each broken down into four VSP  bins, limiting the amount of driving
contained in the final,  open-ended bin.

      As shown in Table 5-3 above,  the 5-cycle formula for highway fuel economy
includes 12% of operation in bin 39, versus 37.5% for bins 36-39. It is unlikely that the
difference between VSP bin 39 fuel consumption over US06 and onroad would  be as
high as 6%, given the more limited range of operation included in bin 39 compared to
the  bin 36 definition used in EPA's 15-car PEMS study.  However, even if this were the
case,  a 6% over-estimation of fuel consumption for 12% of vehicle operation only
results in a 0.7% overestimation of 5-cycle highway fuel consumption. This ignores the
effect of the cold starts, air conditioning operation and operation at colder temperatures,
which all work to lower this percentage. This is a  very small potential source of error.
The benefit of including US06 in the fuel economy label procedure is the direct
measurement of fuel  consumption in  VSP bins 36-39 which  are not addressed by the
HFET and which represent 40% of highway driving in-use.  The net benefit of the trade-
off is very clear.

      We performed an analysis of available second by second fuel economy data
collected for 80 vehicles to directly address this issue further. As discussed in Section
III.B. of the Draft TSD, 61 cars and 19 light trucks were tested over both the FTP and
US06 tests and  had their fuel consumption measured on a second by second basis.
We  grouped each second of each vehicle's operation over the warmed up portions of
the  FTP (Bags 2 and 3) and US06 tests into the 26 VSP bins and calculated the
average fuel consumption for each vehicle in each VSP bin. We then weighted these
VSP fuel consumption values by the frequency of vehicle operation over the HFET cycle
and the  US06 highway bag (as shown in Table 5.3 above) in order to estimate each
vehicle's average fuel consumption over these test cycles. We repeated this weighting
procedure for the MOVES estimate of onroad  highway driving (see Table 5-3 above for
the  weighting factors for the high power VSP bins and Table III.A-12 of the Draft TSD
for all  of the weighting factors).   On average, the fuel  consumption for the 80 vehicles
                                     113

-------
over the HFET, US06 highway and MOVES highway driving patterns were 3.049, 3.609
and 3.443 gallons per 100 miles, respectively.

      The current label approach estimates onroad highway fuel consumption based
on only fuel consumption over the  HFET test.  For these 80 vehicles, average fuel
consumption over the MOVES highway driving pattern was 12.6% higher than that over
the HFET. Per the current label approach, then, onroad highway fuel economy would
be estimated by multiplying HFET fuel consumption by a factor of 1.126.  We applied
this factor to each vehicle's HFET fuel consumption and compared the result to our
estimate of onroad highway fuel consumption  per MOVES. On average, the HFET plus
generic factor over-estimated onroad highway fuel consumption by 0.6%. However,
more importantly, the standard deviation of the percentage differences was 3.2%.  We
could have completely eliminated the average difference by a slight change in the
adjustment factor (i.e., 1.12 versus 1.126). However, the standard deviation would be
unaffected.  It is the standard deviation which  indicates the ability of the approach to
reflect differences in the fuel economy performance of individual vehicles.

      We repeated this using the 5-cycle weighting of 79% US06 highway and 21%
HFET fuel consumption.  On average, the 5-cycle formula over-estimated onroad
highway fuel consumption by 1.7% and the standard deviation of the percentage
differences was 0.5%.  The fact that the 5-cycle approach produced a greater difference
on average is not important here, since the factor accounting for non-dynamometer
effects (i.e., 0.89 in the NPRM) adjusts the final 5-cycle fuel economy values to match
more robust estimates of onroad fuel economy, such as  those made by FHWA. The
key finding is that the combination  of US06 highway and HFET fuel consumption is a
much better predictor of each vehicle's fuel consumption over the MOVES description of
highway driving than the HFET alone. This clearly refutes Honda's concern that the
US06 test is too extreme  to be a useful predictor of onroad fuel economy.

      We agree with Honda that a portion of the US06 cycle came from what was then
called the California ARB-02 cycle. As stated  in EPA's technical report supporting the
development of the US06 cycle:30

      A third control approach involves a hybrid cycle that shares characteristics of
      both  the air-fuel control approach and the representative cycle approach. The
      new cycle, US06, is 600 seconds in duration and is comprised of segments of
      CARB's ARB02 cycle and EPA's REP05 cycle. Similar to the air-fuel control
      method, this method targets specific high emission, non-FTP operation. And like
      the representative cycle, the US06 is based on actual segments ofin-use driving.

And
30
  "Final Technical Report on Aggressive Driving Behavior for the Revised Federal Test Procedure Notice
of Proposed Rulemaking", EPA, January 31, 1995
                                     114

-------
      The ARB02 cycle (figure 1-7) was developed by CARB based on data from their
      Los Angeles chase car study.  The purpose of the cycle is to test vehicles over
      in-use operation outside the boundary of the LA4, including extreme in-use
      driving events.

Comparison of the REP05 and US06 cycles on a second by second basis indicates that
hills 1 and 3 were taken from the ARB02 cycle, while the remainder of US06 was taken
from the REP05 cycle.  It appears that some splicing of portions of REP05 was
performed to remove cruise events and focus on accelerations and decelerations.  For
example, hill 2 of US06 consists of seconds 3-52 and 364-393 of REP05, with one
second added in between to provide time for a reasonable transition between the two
segments.

      Thus, all of the US06 cycle came from monitored vehicle operation via two
representative cycles developed to represent high speed and aggressive driving not
represented by the FTP.  However, in order to develop a shorter test cycle which would
be less costly to run which still required emission control over the  most severe portions
of the representative cycles,  some of the less extreme operation outside of the FTP was
excluded and the more extreme, though still representative, operation  was retained.

      With respect to the comment from Natural Resources Canada, we believe that
various philosophies can  be followed in developing consumer information in general and
fuel economy label values in particular. We desire to provide consumers with as best
an estimate of the fuel economy which they will achieve on the road given the way
consumers actually drive. As can be attested by even brief trips on U.S. freeways,  this
often involves speeds exceeding the speed limit.  Therefore, we do not believe that it is
necessary to develop alternative label values or to assess  the impact of removing
driving which exceeds most speed limits from the estimate. In fact, the state of Texas
recently increased their maximum speed limit to 80 mph in some parts of the state.
Wyoming and other states have had speed limits above 70 mph for some time. Thus,
this type of driving  is not illegal everywhere.

      It is useful to note that the HFET contains speeds as high as 60 mph.  This cycle
was developed when speed limits were no higher than 55 mph. A speed of 60 mph no
longer seems excessive.  However, at the time, it exceeded the allowable speed limit,
because this was how people were found to drive. Therefore, we see  no compelling
need to change this approach at this time.
                                     115

-------
5.11.3 Effect of Air Conditioning on Fuel Economy

5.11.3.1     Approximating Excess Fuel Use Due to Air Conditioning

What we proposed:

      We proposed to estimate the incremental fuel use due to the operation of the air
conditioner as the difference in fuel use measured over the SC03 versus this
combination  of fuel use over Bags 2 and 3 of the standard FTP. This difference in fuel
use between the two tests provides a direct estimate of the impact of air conditioning
use for the conditions present during the SC03 test.  The SC03 test is performed at
95F and 40  percent relative humidity.  The test only lasts 10 minutes and the vehicle is
pre-heated with radiant lamps for 10 minutes prior to the test. Thus, the air conditioning
compressor is generally engaged throughout the entire test. The speed of the vehicle
during the SC03 test is relatively low, at an average speed of 21.5 mph. Of course,
real-world  vehicles operate at different speeds and ambient temperatures and the
compressor may not be engaged 100 percent of the time, particularly during longer
trips. All three of these factors can affect the impact of air conditioning on fuel
economy.  We therefore adjust the estimate of the impact of air conditioning on fuel use
from the SC03 test in  three ways to account for these three factors.

What commenters said:

      UCS commented that while EPA's approach "appears reasonable," they question
the potential  impact of differences between the FTP and SC03 driving schedules.  For
example, they cite the acceleration rates of the SC03 cycle, which, while moderate, are
beyond anything that occurs on the FTP.  They also suggest that the possibility exists
that that a car is not fully warmed up by Bag 2 of the FTP, whereas the SC03  is started
with a fully warmed  up engine. They recommend that EPA validate this approach  by
comparing the weighted FTP results of a representative sample of vehicles with the
results of a set of modified SC03 tests conducted on the same vehicles. The "modified"
SC03 tests would follow the SC03 driving schedule, but would  be conducted with the air
conditioning off and at the same environmental conditions as the  FTP.

      Honda also states that the effect of air conditioning is determined from the
difference  in  fuel consumption over the SC03 test and a blend of  Bags  2 and 3 of the
FTP. Honda points out that the two driving cycles are different and do  not necessarily
yield the same fuel consumption absent air conditioning for all vehicles.

      Honda presents a very general comment about the efficacy of testing which EPA
performed to assess the impact of air conditioning over a number of test cycles.

Our response:

      Honda is generally correct.  The driving pattern in SC03 differs from that
contained  in  the FTP.  Our 39%/61 % combination of Bags 2 and 3 of the FTP
                                     116

-------
approximates the fuel consumption over SC03 at75F with the air conditioning turned
off. However, the blend of Bags 2 and 3 which exactly matches SC03 without air
conditioning is likely to vary somewhat across vehicles.

      Again, however, this potential error in the proposed 5-cycle formulae does not
occur in a vacuum. The alternative is the current label procedure, where air
conditioning is assumed to affect all vehicles equally on  a percentage basis; or the mpg-
based equations, which assume that air conditioning affects vehicles in proportion to
their fuel consumption over the FTP (which  is actually quite similar to the current label
procedure in this regard).  Thus, it is not a matter of the  5-cycle formulae not being
entirely accurate. It is a matter of which approach produces the best estimate.

      We evaluated the potential for error in the various label approaches using data
collected in support of the proposed rule31  In this test program, six vehicles were
tested over a variety of cycles and temperatures with and without the air conditioning
system turned on.  In particular, vehicles were tested over a hot start FTP at 75F
without air conditioning, over a hot start FTP at 95F with air conditioning, and over the
SC03 cycle at 95F with air conditioning.  The testing at  95F was not in an
environmental chamber with solar load, etc.  It was conducted in a standard
dynamometer in  a test cell heated to 95F.  Thus, the SC03 fuel economy values are
not indicative of those which would occur over an official SC03 test.  However, the
consistency of testing methods for all the cycles actually helps isolate the impact of test
cycle  and is satisfactory for the purpose of this comparison.

      The most  direct measure of the impact of air conditioning is the difference in fuel
consumption over a hot start FTP at 95F with air conditioning and a hot start FTP  at
75F without air conditioning. This comparison avoids the  concerns raised by Honda,
since  the driving  cycles were exactly the same in both sets of measurements. Average
fuel consumption over a hot start FTP at95F with the air conditioning on was 0.01267
gallons per mile higher than with the air conditioning turned off.  Average fuel
consumption over the hot  start FTP at 75F  was 0.03867 gallons per mile. Thus, air
conditioning use  over a hot start FTP at95F increased  fuel consumption by 32.7%.

      We repeated this comparison using the 5-cycle approach. Air conditioning fuel
use was determined by subtracting a 39/61% blend of fuel consumption over Bags 2
and 3 of the FTP from SC03 fuel consumption.  Average fuel consumption due to air
conditioning use  was 0.01160 gallons per mile. Thus, air conditioning increased fuel
consumption by 30.0%. Thus, the approach taken in the 5-cycle formulae
underestimated fuel use due to air conditioning by 8%.

      Of greater importance is the ability of the  various  fuel economy label approaches
to predict the impact of air conditioning on a specific vehicle basis. The current label
approach assumes that air conditioning has the  same percentage effect on each
   31 Mitcham, A., Fernandez, A., & Bochenek, D. "Impacts of Ambient Temperature and Air
Conditioning Usage on Fuel Economy" U.S. EPA, Office of Transportation & Air Quality, 2005.


                                      117

-------
vehicle. Thus, air conditioning use under this approach is assumed to be 32.7% of fuel
consumption over the hot start FTP. The 5-cycle approach is exactly that described
above. The mpg based approach uses the results of the 5-cycle formulae, but
correlates the fuel use due to air conditioning against the hot start FTP. Doing so and
performing a least squares regression produces an air conditioning fuel use of:

      0.8581 + 0.0781 * hot start FTP fuel consumption.

      As we have done above, we then  determined the error in predicting each
vehicle's air conditioning fuel  use and calculated the coefficient of variation of the error.

      The current label procedure produces the greatest error. The coefficient of
variation was 48.9% for the six vehicles.  The mpg-based procedure reduced this error
by a factor of two, to 21.2%.  However, the 5-cycle procedure reduced this latter error
again by a factor of two, to 10.6%.  Thus, it is very clear that the error introduced by
using a blend of fuel consumption over Bags 2 and 3 as the basis for assessing air
conditioning fuel use is minor compared to basing this fuel use on fleet-wide estimates.
Thus, we find Honda's assertion incorrect.  UCS' concern appears to be unwarranted.

      With respect to Honda's criticism of our test program, Honda is correct that this
testing did not simulate the flow of air through the  engine compartment in  a manner
representative of onroad operation.  However, the point of this testing was to simply
confirm a very general principle that the air conditioning fuel use was primarily a
function of time and not mileage. Honda did not present any data to indicate that a
more  representative simulation of airflow through the engine compartment would have
changed this conclusion. Thus, we see no need to change this conclusion and its
impact on the proposed 5-cycle formulae.
            5.11.3.2     Available In-Use Data and Analysis

What we proposed:

      The Draft MOVES2004 algorithm of compressor on fraction versus heat index
was developed from the direct measurement of air conditioning operation of over 1000
trips by 20 vehicles in Phoenix, Arizona during the summer and fall of 1992. The
algorithm considers both the frequency that the system is turned on by the driver and
the frequency that the compressor is engaged once the system is turned on.

What commenters said:

      Natural Resources Canada acknowledged that there is no better set of data than
that gathered in Phoenix, Arizona in 1992 to describe the average performance
characteristics of mobile air conditioning systems. However,  they question whether the
data and the resulting conclusions are valid given the advances in compressor  and
climate control system design in the last 14 years. To address this,  they recommend
                                      118

-------
that EPA repeat the Phoenix test program with newer vehicles in different U.S. cities to
capture the effect of new technologies and climatic conditions on air conditioning use.
Alternatively, they suggest that a sensitivity test could be performed using the existing
5-cycle test data, by varying the 13.3 percent air conditioning weighting factor between
estimates of its possible extreme values and noting the impact on overall fuel economy.
This would highlight the potential need (or lack thereof) for a new study.

      NREL commented that the graph on page 69 of the Draft TSD reflects some
unusual trends.  In particular, they find it odd that at 100F, 85% of the vehicles have the
air conditioning on, when personal experience suggests the percentage should be much
higher. They also note the  "dip" at about 85F.

Our response:

      It would be beneficial to repeat the 1992 study of air conditioning use in Phoenix
with late model year vehicles and in other locales. At the same time, the uncertainty
introduced by reliance on this single study appears to be quite small on average. In
Section III.A.3 of the Draft Technical Support Document, we described work by NREL
which estimated air conditioning use using a method which is completely independent of
that relied on by EPA.  In the Draft Technical Support Document, we estimated that
NREL's estimate for air conditioning use (switch turned on) for the U.S. as a whole was
24%.  As described above,  they have updated their analysis since that study we
referenced. Their latest estimate is 29%.  Based on the Phoenix study, we estimate
that air conditioning is in use 23% of the time. While we believe our estimate is more
sound, as it is based on actual measurements of air conditioning use, the NREL
estimate can be used to estimate the  potential error in this figure. Increasing our
estimate of compressor on time of 13.3% by the ratio of 29/23 yields a new compressor
on time of 16.8%. This decreases 5-cycle city and highway fuel economy by 0.8% and
0.4%, respectively.  These are very small changes and would not even result in a
change in the label figure in most cases, due to round off to the nearest mpg. Thus,
uncertainty in the percentage of time that the air conditioning system is turned  on does
not appear to be a significant issue at this time.

      Some newer climate control systems automatically turn on the air conditioning
whenever cooling is desired and trim temperature using heat from the engine coolant
system. Air conditioning use could be higher with such systems, as the driver  might not
even know that the air conditioning system was on. At the present time, such systems
are more common in Europe than the U.S. Should they become more common in the
U.S., their effect on air conditioning use could become an issue for the 5-cycle formulae.

      Regarding NREL's comments about anomalies in the Phoenix data, they are
correct.  However, there are usually anomalies with real data.  One would expect that
essentially 100% of drivers  would utilize their air conditioning at 100 F. However, if the
drivers sampled happened to like driving with the windows down, or had just come out
of a building which was over-cooled, they might have preferred the  warm outdoor air to
                                     119

-------
the air conditioning.  Since the drivers were not asked why they used or did not use air
conditioning for each trip, we will never know the answer.

      Fortunately, whether 85% or 100% of drivers used air conditioning at 100 F had
little impact on the overall projection of air conditioning use, given the high levels of use
at slightly lower temperatures.  As shown above, increasing air condition use by a
relative 25% had little impact on 5-cycle fuel economy values.  Therefore, the anomalies
in the Phoenix data are not a major concern.
            5.11.3.3     Driver Behavior

What we proposed:

      We proposed that the SC03 test procedure be used, without modification, as an
element of the 5-cycle test and calculation procedures.

What commenters said:

      PPG supports changes that consider air conditioner use in the fuel economy
label calculation, and they commented that EPA's current and proposed air conditioner
fuel consumption test methodologies do not replicate actual consumer experiences. As
an example, they note that vehicles  in actual use are typically exposed to a solar load
for substantially longer than the test  requires. They further suggest that the proposed
test protocol requiring that the air conditioner be operated at maximum cooling with high
fan speed for the duration of the test is not representative. They suggest that this could
drive temperatures lower than the consumer's comfort level, and that a more
representative test would be to turn down the fan speed when a set comfort level was
achieved.  They suggest that the test would be more accurate if it replicated more
realistic solar heating conditions by testing after extended heat soaks and by more
accurately accounting for how consumers adjust their air conditioner settings.

Our response:

      PPG's suggestion would amount to a revision to the SC03 emissions test
procedure, which we did not propose to do. We do not necessarily agree or disagree
with PPG's comments regarding solar load  exposure and consumer behavior with
respect to the fan speed settings, but we did not propose to amend the SC03
procedure, especially in ways that could potentially impact emission results or the cost
of conducting the test.  In particular,  an extended heat soak could dramatically impact
the costs for manufacturers.  We did not intend for our proposal to reopen the SC03 test
procedure and change it in any way; the emission standards for the SC03 were
designed with the specific test procedure as it was designed,  and changing the SC03
procedure would have far-reaching impacts. In addition, the impact of air conditioner
usage on fuel economy is based on  a difference in tests  results, one test run with the air
conditioner on and one run with the air conditioner off, with the result  used as the
                                      120

-------
estimate of the fuel economy impact of the air conditioner at the temperature of the
SC03 test (95F).  The impact of temperature on air conditioning use is then used to
determine the percentage of time that the air conditioning system is turned on in-use
and the percentage of time that the compressor is actually engaged. The impact of
temperature on compressor load is then factored in based on other information on the
effect of temperature on load.
            5.11.3.4     Accounting for Energy Reduction Technologies

What we proposed:

      We did not propose any specific provisions to encourage or discourage energy
reduction technologies.

What commenters said:

      Delphi commented that because the compressor is "on" throughout the entire
SC03 test, the test will not reflect any efforts by a vehicle manufacturer to improve the
energy efficiency of their air conditioner systems at lower heat loads.  Because average
conditions in the U.S. are milder than those experienced during the SC03 test, they
point out that "...accounting for the differential in A/C systems performance across
these conditions is critical." Delphi claims that the most energy-efficient systems today
may consume only 50 percent of the energy of the least efficient systems, and that
"...this difference is not likely to be seen in the SC03 test" (emphasis is Delphi's). They
believe that more accurate estimates of compressor usage are available and  that there
is a better way to establish the average energy used by today's A/C systems and future
systems.

      Delphi suggests that EPA  update the "compressor on time" percentages, and
refers to field data from 1987 model year vehicles with fixed-displacement cycling
compressors and from 1997 model year vehicles with variable-displacement
compressors. They also note that since the early 1990's, when the Phoenix data was
collected, automatic climate control systems have become more prevalent, thus
increasing compressor on-time. They present the following summary of their  data,
without additional detail:
Field Test
1987 Fixed Displacement, Manual and Semi-Automatic
Systems
1997 Variable Displacement, Manual Systems
1997 Variable Displacement, Automatic Systems
% Compressor On-
Time (including
defrost)
32%
45%
70%
                                     121

-------
      Additionally, Delphi proposes a series of correction factors that would be
employed within the fuel economy calculation equations. These factors are intended to
reflect the effect of air conditioning technology that may not show up in the SC03 test.
They suggest factors that would adjust the fuel economy results for systems with
variable-displacement systems, electronically controlled compressors, accurate control
of evaporator temperatures, etc. They argue that these factors, while leading to more
accurate fuel use estimates, will also provide manufacturers with the incentive to adopt
some of these technologies.

      NREL likewise noted some technologies that can reduce the fuel used by air
conditioning systems but that would not be accounted for in the new fuel economy
estimates.  These technologies include those that reduce the thermal load in the cabin
(e.g., solar  reflective glass, insulation, ventilation), more efficient equipment (e.g.,
variable-displacement compressors), improved thermal comfort technologies (e.g.,
ventilated or climate control seats), and advanced control strategies (e.g., increased use
of recirculation air). To resolve these issues, they recommend the following:
   o  Increase the soak period of the SC03 test or reduce the compressor-on
      percentage in the adjustment calculation for vehicles with load reduction
      technologies;
   o  Use  a realistic temperature setpoint in the cabin for the test or adjust the
      compressor-on percentage for vehicles with load reduction technologies;
   o  Reduce the compressor-on percentage for vehicles with a control system that
      increases the use of recirculation air.

      PPG also recommended that EPA consider modifying the proposal to consider
the impact of technologies that lead to more efficient air conditioner systems, although
their comments were focused on the potential impact of solar reflective window glazings
on air conditioner fuel use.

Our response:

      The  current label procedure only addresses the impact of air conditioning use on
fuel economy through the 10% and 22% generic adjustment factors which are applied to
the FTP and HFET test results. Since these factors are the same for vehicles with and
without air conditioning, the current procedure actually reduces the fuel economy label
values due to air conditioning use for vehicles not even equipped with air conditioning.
This is the status quo.

      The  proposed  mpg-based equations actually continue this approach.  While the
fuel economy projections used to develop the mpg-based equations are based on the
impact of air conditioning on individual vehicle's fuel economy,  the average adjustment
is applied to vehicles with and without air conditioning.

      The  proposed  5-cycle formulae incorporate the impact of air conditioning on fuel
economy through the SC03 test, with compressor usage and load adjusted for
conditions outside those included in the SC03 test. The SC03  test, by virtue of its short
                                      122

-------
duration and short time of heat load prior to the test, reflects the impact of some aspects
of vehicles' climate control systems realistically, while not doing so with respects to
other aspects.  The adjustments in compressor usage for non-SC03 conditions were
developed from data obtained from vehicles on the road in 1992. The adjustments in
compressor load for non-SC03 conditions were based on air conditioning systems with
fixed displacement compressors which cycle on and off frequently under less than
maximum cooling conditions.

      The comments summarized above point out a number of limitations with the
proposed 5-cycle formulae in terms of granting credit to certain technologies whose
benefits are either not reflected in the SC03 test or are only partially reflected.  Delphi
indicates that some efficient air conditioning systems in use currently could be as much
as 50% more efficient than the systems modeled in the 5-cycle formulae.  At the same
time, Delphi points out that some climate control systems may actually run the
compressor more than that indicated in the Phoenix study. This could increase the fuel
economy impact of these systems relative to those modeled in the 5-cycle formulae.

      Using the 5-cycle certification database for 615 vehicles, reducing the impact of
air conditioning on fuel use (by reducing the effective compressor on percentage from
the 13.3% estimated in the 5-cycle formulae to 6.7%) improved city and highway fuel
economy by 1.6% and 0.7%, respectively.  While significant, such an improvement
would not change the actual label fuel economy for most vehicles due to the rounding of
the label to  one mpg.  At the same, we desire to give credit to all technologies which
affect onroad fuel economy to the fullest degree possible without significantly increasing
the current test burden.

      Both the current and proposed label procedure base the label values entirely on
fuel economy measurements from dynamometer tests coupled with adjustments or
weighting factors applicable to all vehicles. As described elsewhere in this document,
we considered developing separate 5-cycle weighting factors for hybrid vehicles, but
rejected that approach, in part due to the variation in what constitutes "hybrid"
technology. The same concern applies here.  The  effect of each of the fuel-saving
technologies mentioned in the comments would likely differ across vehicle lines and
comprise more of a continuum in technology than discrete steps. Considerable effort
would be needed to develop procedures to appropriately address such technologies.
These efforts would necessarily delay the final rule implementing 5-cycle formulae
without such adjustments for different air conditioning system technologies.  Given the
relatively small impact of these technologies on the label values, and the much larger
impact of the 5-cycle approach as a whole, we do not believe it appropriate to delay the
5-cycle approach until procedures can be  developed which account for differences in air
conditioning systems.  Nonetheless, we encourage those familiar with such systems to
provide further information regarding their operation in-use and their impact on  onroad
fuel economy.  We remain open to working with all  affected parties in developing
procedures which  could allow the  in-use efficiencies of such systems to be appropriately
incorporated into the 5-cycle formulae.
                                      123

-------
            5.11.3.5     Temperature Impact on Compressor Power

What we proposed:

      Using data from the Phoenix program, we identified a relationship between
compressor cycling and ambient temperature. Following this we used a simplified
thermodynamic model to develop a correlation to reduce compressor torque as a
function of ambient temperature.

What commenters said:

      NREL commented that the impact of ambient temperature on compressor power
may be double counted in our analysis.  They note that on page 70 of the Draft TSD we
identify the relationship between compressor cycling and ambient temperature.
Following this we use a simplified thermodynamic model to develop a correlation to
reduce compressor torque as a function of ambient temperature.  NREL suggests that
an actual air conditioning system "...will cycle the compressor to control the high side
pressure and evaporator temperature while the simplified model adjusted the
compressor power," and that using both may be inappropriately adjusting compressor
power twice.

      Additionally, NREL notes that, contrary to statements in the Draft TSD on page
71, there is available data from  the SAE Alternate Refrigerant Cooperative  Research
Program regarding vehicle air conditioning system  design and  how engine speed and
load affect its efficiency.

      Honda mentions that EPA adjusted air conditioner load for ambient temperatures
below 95F, but did not do so for humidity or solar load or vehicle  soak time in the sun
prior to operation.

Our response:

      We do not believe that we are performing any double counting when accounting
for both the time which the compressor is not engaged and the reduced load on the
engine at lower temperatures while the compressor is  engaged. The study which
estimated (in a simplified fashion) the compressor torque (and  power) as a function of
ambient temperature did so in a way which should  be independent of cycling. The
compressor model in the reference was designed to be combined with a  cycling clutch
model, which is briefly introduced in the paper, but not thoroughly developed. The
activity cycling factors used in the AC factors in the present analysis were based on
actual measurements of compressor use in Phoenix and should be independent.

      With respect to Honda's  comment, we stated in the Draft Technical Support
Document that we did not adjust air conditioning compressor load for differences in
                                     124

-------
humidity because we thought the effects were small. Honda provided no data to refute
this.  Thus, we are not adding any further adjustment at this time.

      Regarding solar load and vehicle soak time in the sun prior to operation, it is not
clear that this would affect compressor load. These factors will affect the amount of
time that the compressor is engaged.  This is already accounted for through the use of
measured onroad compressor usage rates from the Phoenix study coupled with an
assumption that the compressor is engaged 100% of the time over the SC03 cycle. The
last assumption  is due substantially to the solar load included in the SC03 test and the
fact that the vehicle was under significant solar load for 10 minutes prior to start of the
test.

      Temperatures under the hood of a moving vehicle could be affected by solar load
and, thus, affect compressor efficiency. EPA has no data which would allow the
quantitative estimation of this effect.  Honda also did not supply any data.  We expect
that the effect of varying solar load would be less than the effect of ambient
temperature.  Accounting for the variation in ambient temperature when the air
conditioning is operating only reduced the effective compressor on time by 15%. Thus,
ignoring the variation in solar load should be much smaller. Given that the overall
impact of air conditioning is only 2% on city fuel economy and less on highway fuel
economy, we expect this to have a negligible impact on 5-cycle fuel economy.
            5.11.3.6     Estimates of National A/C and Defroster Use

What we proposed:

      We used a study from NREL to determine estimates of national average air
conditioning and air conditioning plus defroster use.

What commenters said:

      NREL commented that our estimates of national average air conditioning and air
conditioning plus defroster use (22.9% and 33.5%, respectively), which we derived from
an NREL study, should actually be 28.1% for air conditioning and 32.6% for air
conditioning plus defroster use after weighting by vehicle registrations. NREL also
assumed a mix of 60% cars and 40% trucks, while our analysis used a mix of 65% and
35% cars and trucks, respectively.

      NREL confirms our assumption that in their analysis, the relative fuel economy
impact of air conditioning and defroster includes the impact of ambient temperature on
load and on the operating frequency of the compressor.

      NREL suggested that the 61F air temperature during defroster use (Draft TSD,
page 73) seems high, noting their analysis assumed that the A/C would be used for
dehumidification when the ambient temperature was between 35 and 55 F.
                                     125

-------
Our response:

      We contacted NREL personnel to determine the source of the discrepancy
regarding their estimate of national average air conditioning usage. We determined that
they had since updated the analysis which they performed in 2002 and that the more
recent analysis indicated higher air conditioning usage.  Thus, NREL currently projects
that drivers have the air conditioning system turned on 28.1 % of the time, while our
estimate is 23.8%, based on the Phoenix study. This difference is not surprising  given
the completely different methods used to derive these estimates. They are still quite
similar in magnitude, though obviously not as close as indicated in the Draft Technical
Support Document. We will continue to use our estimate in the developing the
weighting factor in the 5-cycle formulae, due to the fact that it is based on actual vehicle
testing, though this is admittedly limited  in nature.  NREL's estimate is based on a
prediction  of a driver's comfort level and has not been confirmed by any vehicle testing.

      Regarding the use of a 61F air temperature assumed for defroster operation, it is
not clear whether NREL thought this was our assumption, or, as was actually the case,
our description of their assumption. This temperature is stated in their paper (actually
as 16 C).  We agree with NREL that this air temperature seems high given that they
assumed that demister use would  only occur when the ambient temperature was
between 35 and 55 F.  We attributed this to higher underhood temperatures, but could
be wrong in this regard. It could have been a typographical error in the report or simply
an error in their analysis. In either case, a lower temperature for demister use would
decrease the compressor load slightly, which would decrease our estimate that
demister use on average is equivalent to 1.3%  of the excess fuel use per mile
measured  during SC03.

      More importantly, correcting NREL's estimate of air conditioning use from 22.9%
to 28.1% and total air conditioning plus demister use from 33.5% to 32.6% reduces
demister use from 10.6% to 4.5%. This reduction is more than a factor of two and
dwarfs the effect of a lower air temperature in demister load modeling and small
changes to the assumed mix of cars and light trucks.  Thus, the impact of demister use
is likely less than 1 % of the excess fuel use per mile measured during SC03.  This
strongly supports our decision in the Draft Technical Support Document to not include
the impact of demister use in the 5-cycle formulae.
      5.11.4    Effect of Cold Temperatures on Fuel Economy

What we proposed:

      We proposed to model the effect of cold temperature on city fuel economy with a
50/50 weighting of fuel consumption over bags 2 and 3 of the cold FTP.
                                     126

-------
What commenters said:

      Honda comments that EPA did not show that Bag 2 was representative of city
driving at colder temperatures. They also stated that EPA did not perform any analysis
to confirm its assumption that the effect of colder temperatures on running fuel use
changed gradually and linearly.

      Honda also points out that the excess fuel use at colder temperatures also
includes a difference in driving pattern which may be confounding the results. EPA did
not evaluate the impact of this difference in driving pattern for individual vehicles.

      Honda further states that EPA did not validate its assumption that running fuel
use during highway driving at 20F was 4% higher than that at 75F on an individual
vehicle basis.

Our response:

      There was a typographical error in some of the 5-cycle city fuel economy
formulae indicating that city driving at colder temperatures was assumed to be equal to
Bag 2 fuel consumption of the  cold FTP. These equations should have indicated that a
50/50 weighting of fuel consumption over Bags 2 and 3 of the cold FTP.  The text
deriving this portion of the formula correctly indicated the  latter understanding. This
50/50 weighting was the result of the same type of VSP modeling used to develop the
cycle weighting factors at 75F. It is also nearly identical to the relative weighting of the
FTP  itself. Thus, we see no issue with the representation of driving at colder
temperature which would require modification to the 50/50 weighting.

Our response:

      This assumption that the effect of colder temperatures on running fuel use
changed gradually and linearly is used to develop the weighting factor for running fuel
use at 20F. Assuming a linear relationship between running fuel use and ambient
temperature produced an average driving temperature of  58.7 F, which is 30% of the
way from 75Fto 20F.

      We evaluated the potential impact of our assumption that running fuel use at
highway speeds increases linearly with temperature between 75F and 20F. We
analyzed the trend in hot LA4 fuel consumption for the 13 Toyota and Honda vehicles
described in Section III.F.1 of the Draft Technical Support Document.  Three of the four
hybrids in this database were tested over a four-bag FTP.  Thus, for these three
vehicles, we used Bags 3 and  4 to represent a fully warmed up FTP, or hot LA4. This
data shows that running fuel use at 50F was 4.6% higher than at 75F, while that at
20F was 19.5% higher. Thus, for city driving, the relationship between temperature
and running fuel use was non-linear, with the effect increasing gradually at first and then
increasing as the temperature  decreased further from 75F. We fit a second order
                                      127

-------
equation to this data. The result was that running fuel use at temperature "T" relative to
thatat75Fwas:

            1 - 0.00035  *  (T-75F)  + 0.00006 * (T-75F)2

Applying this relative running  fuel use to each hour, month and county in the U.S., as
described  in the Draft Technical Support Document for start fuel use and air
conditioning use, we found that relative running fuel use was on average across the
U.S. 3.6% higher than that at 75F.  This increase of 3.6% is a relative 18% of the
19.5% effect of operating at 20F for these 13 vehicles. Thus, it is possible that the
30% factor based on an assumed linear relationship over-estimates the impact of colder
temperatures.  Therefore, we will change the weighting factor in the 5-cycle formula for
running fuel use at 20F to  0.18 from 0.30. This increases 5-cycle city fuel economy by
0.8% for the 423 vehicles in our certification database, while 5-cycle highway fuel
economy increases by 0.5%.  Combined 5-cycle fuel economy increases by 0.7%.
Because this change affects the relationship between combined fuel economy based on
the FTP and HFET and that based on the 5-cycle formulae,  it will tend to increase the
factor which accounts for non-dynamometer effects.

      Honda is correct in understanding that running fuel use during city driving at 75F
is based on a different composite driving cycle than that at 20F.  The reason for this is
the fact that the US06 test is not performed at 20F. Honda is also correct in pointing
out that this difference of driving pattern is included in the description of excess running
fuel use due to colder temperatures  during city driving at the end of Section III.A.4 of the
Draft Technical Support Document.  However, approaching  the inclusion of the effect of
colder temperatures on running fuel  use during city driving as a simple addition of an
"excess" fuel use is not the best understanding of what was done. We presented this
effect as an excess in order to be consistent with other aspects of the 5-cycle equation,
like start fuel use and fuel  use associated with air conditioning. It is more
straightforward to understand the treatment of colder temperatures in this case as an
averaging  of running fuel use at warm temperatures (i.e., 75F) and that at cold
temperatures (i.e., 20F).  Running fuel use at 75F is estimated using the best fuel
economy data available, which includes the city bag of US06. Running fuel use at20F
is estimated using the best available data, which only consists of Bags 2 and 3 of the
cold FTP.  If fuel economy over US06 was not available, then running fuel  use at 75F
would also be based solely on Bags 2 and 3 of the standard FTP.  It does not improve
the accuracy of the estimate of running fuel use at 75F to exclude the US06 city bag
data. We  could have developed a factor based on the FTP  and cold FTP data which
estimated  the impact of cold temperature on city driving. This alternative approach was
discussed  in the Draft Technical Support Document and rejected in favor of the
proposed approach.  This alternative would have presumed that the change in running
fuel use over the FTP at colder temperatures also applied to US06 city driving.  While
this assumption would not have been unreasonable, the proposed approach avoided
this assumption.  In its place,  we simply have a simplified city driving cycle at colder
temperatures.  Given Honda's preference to base fuel economy labels on procedures
which emphasize the FTP,  this approach would presumably be preferential. Honda did
                                      128

-------
not suggest an alternative approach within the 5-cycle construct. Thus, we see no need
to change this aspect of the 5-cycle city fuel economy formula at this time.

      Honda's comment that EPA did not validate its assumption that running fuel use
during highway driving at 20F was 4% higher than at 75F on an individual basis
seems to be a simple repetition of previous comments that EPA did not validate some
aspect of the proposed 5-cycle formulae.  The impact of the 4% assumption is
essentially a 1.2% reduction in highway fuel economy for each and every vehicle.  If this
factor were not included in the 5-cycle highway formula, combined 5-cycle fuel economy
would have been 0.6% higher.  The comparison of combined 5-cycle fuel economy with
onroad fuel economy estimates made by FHWA, as well as the sum of estimated non-
dynamometer effects would have shown a 0.6% greater shortfall.  Thus, the factor for
non-dynamometer effects would increase by 0.6%.  This would  obviously reduce both
the city and highway fuel economy of each and every vehicle by 0.6%. We have a data-
driven method for accounting for colder temperatures on running fuel  use during city
driving. We believe it is preferable to estimate this effect for highway  driving rather than
account for it in the non-dynamometer factor which is applied to both city and highway
fuel economy. We could attempt to develop separate non-dynamometer adjustment
factors for city and highway fuel economy. However, the methodology to do so would
involve a number of assumptions and uncertainties which would defeat much of the
purpose for the separate factors.

      In addition, the 4% assumption is very analogous to the approach taken in the
current label procedures and the mpg-based equations. It does not rely on the US06,
SC03, or cold FTP test results. To criticize this assumption seems to be inconsistent
with the rest of Honda's comments.
      5.11.5    Adjustment Factor for Non-Dynamometer Effects

What we proposed:

      We proposed a downward adjustment to account for effects that are not reflected
in the proposed 5-cycle approach.  There are many factors that impact fuel economy,
but are difficult to account for in the test cell on the dynamometer. These include
roadway roughness, road grade (hills), wind, tire pressure, heavier loads, hills,
snow/ice, effects of ethanol in gasoline, larger vehicle loads (e.g., trailers, cargo,
multiple passengers), and others.  Current data indicates that these impacts can lower
fuel economy from 9 to 13 percent. We proposed an 11 percent downward adjustment
to account for these non-dynamometer effects.

What commenters said:

      ACEEE commented that EPA appears to have assumed - by using a single
adjustment factor for both city and  highway driving - that the  non-dynamometer factors
as a group affect city and highway  driving equally. They contend that this may not be a
                                     129

-------
valid assumption, especially given the large contribution of wind resistance to the
proposed 11 percent adjustment, and the fact that the impact of wind resistance on fuel
economy changes dramatically with vehicle speed.

      Bluewater Network believes that the fuel economy results for both the mpg-
based approach and the 5-cycle approach are overstated for hybrids.  Consequently,
they suggested that EPA add an additional non-dynamometer factor specifically for
hybrid vehicle city fuel economy, citing a large real-world shortfall that they claim is not
accounted for in the 5-cycle approach or the proposed non-dynamometer factor.

      Honda argued EPA's estimates of the effects of ethanol in the fuel and low tire
pressure seemed reasonable, but estimates for other factors were "problematic." They
suggest that average wind speed data and the fuel economy impacts of wind  speed
seem far too high. They further argue that EPA made errors in comparing the average
5-cycle fuel economy values with fleetwide estimates of in-use fuel economy made by
FHWA, and that the result of these errors leads to an overestimate of the impact of non-
dynamometer effects. They conclude that the overall non-dynamometer effect should
be 6 percent, not 11 percent as proposed.

Our response:

      The comment by ACEEE reflects the common understanding that higher vehicle
speeds produce greater aerodynamic drag and that this leads to an increased impact of
wind on fuel economy.  As discussed in  the Draft Technical Support Document, this is
true for a headwind, but not for wind coming from the side of the vehicle. For a given
wind speed, the slower the vehicle is traveling, the greater the net angle that the vehicle
is traveling through the air.  Changes in  this net angle of movement can have an
enormous effect on total aerodynamic drag, as indicated by the analysis described in
the Draft Technical Support Document.  Thus, the effect of wind is not greater for
highway driving than city driving.

      More importantly, we do not have solid estimates of onroad fuel economy for city
and highway driving separately. The uncertainties in splitting total onroad fuel economy
into city and highway components are sufficiently large to counter any likely
improvement in the accuracy of having separate non-dynamometer factors. Thus, we
continue to believe that a single factor for non-dynamometer effects is appropriate.

It is possible that the proposed mpg-based equations overstate hybrid city fuel economy
relative to that for conventional vehicles, at least as indicated by a comparison of mpg-
based fuel economy to 5-cycle fuel economy. This is described in Section III.B of the
Draft Technical Support Document and depicted in Figure 111-17. Elsewhere, we
address the comment that we develop separate mpg-based equations for hybrid
vehicles and still conclude that a single equation for all vehicles is the most appropriate
approach at this time. The required  use of the 5-cycle formulae starting with the 2011
model year should make any deficiency in this regard a  short term one.
                                     130

-------
      Regarding Honda's comment, there were three steps in our determination of the
11 % downward adjustment to account for factors not represented by the various
dynamometer tests.  First, we determined that the current label values over-estimate
onroad fuel economy per FHWA (with some adjustments by EPA) by 3-4%.  Second,
we determined that the average combined 5-cycle fuel economy for the 414
conventional vehicles in our certification database was 3% higher than current label
values on average. Third, we determined that changes to EPA test procedures
occurring since the vast majority of the vehicles in the 2002 and 2003 in-use fleets were
certified had increased average composite city-highway fuel economy by 3%.  In total,
then, the difference between the average composite 5-cycle fuel economy and FHWA
estimates of onroad fuel economy was 9-11 % on a comparable basis.  We chose the
upper end of this range for inclusion in the mpg-based and 5-cycle formulae.

      Honda presented a number of reasons why changes to EPA test procedures
occurring over the 2000-2004 model years did not increase composite fuel economy by
3%.  Honda pointed out that two other changes occurred in addition to the one cited in
the Draft Technical Support Document which basically compensated for the effect of
removing the elimination of 10% road load adjustment to simulate air conditioning.
Honda references several communications between EPA, NHTSA and industry on this
issue.

      Upon reevaluating this issue, EPA agrees with Honda's assessment that the net
effect of test procedure changes occurring in this timeframe was roughly zero. Thus,
when estimating the difference  between the average composite 5-cycle fuel economy
and FHWA estimates of onroad fuel economy, no adjustment should be made because
of changes in test procedures occurring with the implementation of the Supplemental
FTP standards in the 2000-2004 timeframe.

      However, other factors affecting the estimate of the non-dynamometer factor also
need to be updated.  Specifically, more recent FHWA estimates of onroad fuel economy
are available. Also, the effect of requiring the heater and/or defroster to be turned on
during the cold FTP also needs to be incorporated into the analysis. Thus, a revised
estimate of the non-dynamometer factor will be made in the Final TSD.
      5.11.6    mpg Approach Equation Tolerances

What we proposed:

      Section 600.116-08 of the proposed regulations would require additional
20F/SFTP testing when the certification vehicle's city and highway five-cycle fuel
economy are less than 0.96-times and 0.95-times the certification vehicle's city and
highway mpg-based fuel economy, respectively. Because it is possible for the 5-cycle
fuel economy values to meet either one of the city or highway criteria, but not the other,
we proposed a methodology that accommodates the variety of possible outcomes. If
the 5-cycle fuel economy values for a specific emission data vehicle are more than four
                                     131

-------
percent below the mpg-based estimate for city fuel economy, all the vehicle
configurations represented by that emission data vehicle would be required to use the
5-cycle approach for determining both city and highway fuel economy.  This is because
all five cycles play a significant role in the 5-cycle city fuel economy formula.  In the
case where the 5-cycle highway fuel economy values are more than five percent below
the mpg-based estimate for highway fuel economy, but the 5-cycle city values are not
more than four percent below the  mpg-based city fuel economy estimate, we proposed
that all vehicle configurations represented by that emission data vehicle would be
allowed to use the mpg-based formulae to derive the city fuel economy label value. We
proposed that the highway fuel economy value, however,  would in this case be based
on an alternative, simplified 5-cycle formula, which would  be based only on test results
from the FTP, HFET, and  US06 tests. This is because the impacts of the Cold FTP and
the SC03 tests on the 5-cycle highway formula are relatively small and can be included
in the formula as simple estimations based on existing data.

What commenters said:

      AAM/AIAM evaluated the 423-vehicle database used to generate the mpg-based
equations and concluded that the  proposed 0.95-times mpg-based highway fuel
economy criterion would likely impose additional testing for over ten percent of vehicles.
They suggest that this test burden would exceed the benchmark established in the
CAP2000 rulemaking for requiring additional testing through a manufacturer
confirmatory process. Their analysis of the data set also showed that the  standard
deviation of the five-cycle  city data (about the mpg-based city curve) is 0.61 mpg, and
the standard deviation of the 5-cycle highway data (about the mpg-based highway
curve) is 1.14 mpg. They  conclude from this that the variability of 5-cycle highway
results was about 87% greater than that of 5-cycle city results, which they believe is
reflective of the great variability seen in US06 data. They believe that a criterion  of 0.93-
times the mpg-based highway fuel economy would reflect this increased level of
variability and consequently they request that EPA reconsider this factor.

      Nissan noted that the proposed criterion for the City tolerance band is consistent
with their experience, but they feel that the Highway tolerance band is "overly restrictive
and does not accurately reflect the greater variability of the Highway result (principally
from the greater variability of the US06 component)." They propose a factor of 0.93 x
Highway mpg.

      Toyota expressed support for the mpg-based formula with the applied
"bandwidths," believing that it will  help reduce the need for additional testing. They also
expressed agreement with the proposed alternative highway fuel economy formula
based on the FTP, HFET,  and US06 tests, stating that it will also achieve 5-cycle goals
and require less additional testing. However, Toyota also suggested an additional step,
involving a "3-cycle" fuel economy analysis, before a vehicle line would be committed to
the 5-cycle approach. This interim step would allow a manufacturer whose emission
data vehicle falls outside the mpg-based tolerance limits to re-evaluate fuel economy
with the Toyota 3-cycle formula. If the emission  data vehicle's 3-cycle city or highway
                                      132

-------
values fall within +/-1 mpg of the 5-cycle results, then the other vehicles within that test
group would only have to run an additional US06, which would then be used in the
Toyota 3-cycle equation to determine the label value. If the 3-cycle values fell outside a
+/-1 mpg tolerance, then the full 5-cycle approach would be required.

      UCS commented that the tolerance bands are too wide, and may fail to show
significant differences between different models.  They suggested comparing two
vehicles, one which gets a highway rating of 30.4 mpg from both the 5-cycle test and
the mpg-based adjustment, and the other which gets 28.1 mpg on the 5-cycle test but
29.5 mpg using the mpg-based adjustment. The label for each vehicle would be
rounded to 30 mpg, although, UCS points out, the first vehicle gets 8 percent better fuel
economy. They recommend that EPA tighten the tolerance band requirements to
reduce the number of vehicles that are able to escape 5-cycle testing. They also
suggest that EPA make data publicly available indicating which vehicles are contained
in each test group to add transparency and opportunities to comment on their
appropriateness. UCS also questioned the assumption that all the vehicle types within
a test group can be adequately represented by the emission data vehicle and will return
similar fuel economy test results. They believe that this assertion needs to be validated.

      Toyota proposes that the tolerance levels for evaluating fuel economy estimates
should be numerically expressed as  +/-1  mpg, rather than a percentage-based
tolerance level.  They suggest that this is necessary in order to maintain consistency
among the range of fuel economy data between high mpg and low mpg vehicles.

Our response:

      The criteria for use of the mpg-based approach in model year 2011 and later (5-
cycle city fuel economy above four percent and 5-cycle highway fuel economy above
five percent) are based on the balance of three factors.  First, we designed them to be
sufficiently large so that typical test-to-test variability would not cause a test group to fail
the criteria. This may be a greater concern for the highway fuel economy comparison,
due to the dominance of the US06 fuel economy (which inherently has greater test-to-
test variability than the other tests) in the 5-cycle formula.  Second, we want to minimize
the potential error in the fuel economy label.  Label fuel economy values are rounded to
the nearest whole  mpg. Thus, we felt it important to keep the difference between the 5-
cycle and mpg-based fuel economy values within  roughly  one mpg,  if possible. In other
words, if the difference between the two methods is less than 1 mpg, then the two
methods would produce the same label value. If the difference is more than 1 mpg then
we would expect the 5-cycle method to result in a different label value, and thus it is
more important to  trigger the requirement for additional  testing. Third, we want to avoid
requiring additional fuel economy testing that will have little to no impact on the label
values.

      The four percent tolerance band for city fuel economy is equivalent to roughly
0.6-0.7 mpg on average. Due to the contribution of a number of independent fuel
economy measurements in the 5-cycle city formula, the effect of test to test variability
                                      133

-------
should be much lower than four percent. Based on the 5-cycle test results of 615 recent
model year vehicles, we estimate that about 96 percent of test groups would fall above
the four percent tolerance line. Thus, we believe that this criterion adequately satisfies
the three factors mentioned above.

      The five percent tolerance band for highway fuel economy is equivalent to
roughly 1.1  mpg on average.  Thus, it is slightly higher than the typical error associated
with rounding.  However, due to the dominant contribution of the US06 fuel economy in
the 5-cycle highway formula, and the fact that this test tends to have relatively high
variability, we are concerned that test-to-test variability could be on the order of 3.0
percent in the 5-cycle highway formula. We estimate that about 87 percent of test
groups would fall above the five percent tolerance  line. Thus, again, we believe that this
criterion adequately satisfies the three factors mentioned above.

      Overall, allowing the continued use of the mpg-based approach in this way will
reduce the number of additional SC03 and cold FTP  tests by about 96 percent and
reduce the number of additional US06 tests by about 87 percent.  Moreover, this
significant reduction  in test burden  is achieved with no significant impact on the fuel
economy estimate.

      We believe that Toyota's proposal adds unnecessary complexity with little likely
gain. As we have noted, a large majority of vehicles are expected to pass both the city
and highway criteria and thus be able to  use the mpg-based adjustments.  Our final rule
offers a "3-cycle" option in the event that only the highway criteria is failed, with the end
result that the burden of additional testing over five test procedures will be quite small.

      Each year, manufacturers must demonstrate compliance with federal emission
standards by performing tests over all five test procedures.  The vehicles on which
these tests are performed are known as "emission data vehicles", which are selected to
represent the "worst-case" emitting vehicle in a group of vehicles, known as a "test
group", which share common engine and emission control designs. We expect that
these vehicles will also be the worst-case configuration for fuel economy, given that the
parameters which contribute to higher emissions (e.g., transmission class, vehicle
weight, presence of accessories) also generally contribute to lower fuel economy. We
believe that it is reasonable to allow continued use of the mpg-based line when the
available 5-cycle fuel economy data (from emissions  certification) indicates that the
mpg-based fuel economy determined from the official FTP and HFET tests performed
for the test group are similar enough to the 5-cycle fuel economy determined from the
official FTP, HFET, US06, SC03 and Cold FTP tests  for that same test group. In that
case, the manufacturer can use the mpg-based method for all model types covered
under the EPA certificate of conformity that is represented by the  5-cycle data submitted
to represent those vehicles. The manufacturer will not need to conduct 5-cycle testing
for fuel economy labeling for these model types.
                                      134

-------
5.12  Onroad Fuel Economy Estimates by Other Organizations

What we proposed:

      In the proposal, we compared onroad fuel economy estimates developed by
several organizations to the current, mpg-based and 5-cycle fuel economy label values.

What commenters said:

      Honda provided a number of comments on EPA's presentations of the onroad
fuel economy estimates by outside organizations. Generally, Honda agreed with EPA's
characterization of the relative strengths and weaknesses of the various estimates.
Overall,  Honda believes that none of the studies cited in the proposal comes close to
being a robust source of onroad fuel economy data.

      More specifically, Honda criticized the  Consumer Report estimates due to their
basis on unknown and apparently unrepresentative driving cycles and adjustments for
differences in ambient temperature. Honda was especially critical of the temperature
adjustment applied by Consumer Report to hybrid fuel economy.  Honda also criticized
EPA for comparing EPA label values to the Consumer Report estimates for the vehicle
fleet as a whole and not on an individual vehicle basis.

      Honda also criticized the unknown nature of the basis for the AAA estimates.
Honda also criticized AAA's conclusion that the US06 test better characterized onroad
fuel economy than the current label values.

      With  respect to the Oak Ridge National Laboratory "Yourmpg" database,  Honda
pointed out that their hybrids did not perform any worse than conventional vehicles.

      Honda criticized EPA's use of onroad fuel economy estimates from the
Department of Energy's FreedomCar program, most of  the vehicles in this program are
in commercial service.  Honda cites the need for extensive maintenance of FreedomCar
vehicles as evidence of extreme service.

      Honda also criticized EPA's analyses of its fuel economy data collected recently
in Kansas City, concerned that the presence of the PEMS equipment affected the fuel
economy measurements. Honda also presented a number of concerns with the quality
of the speed data. Honda also believed that comparisons should be performed  in terms
of fuel consumption instead of fuel economy,  that relative performance between vehicle
models is more important than absolute fuel economy comparisons.

      Finally, Honda presented an analysis of onroad fuel economy estimates obtained
from Strategic Vision.  Honda also made a supplemental comment after the end of the
comment period presenting additional  analysis of the Strategic vision data.
                                     135

-------
      In their comments at the hearing, AAA provided additional onroad and
dynamometer fuel economy estimates. AAA borrowed 41 vehicles from their owners in
Southern California, obtained their owners' estimate of onroad fuel economy, and then
tested the vehicles using their own fuel economy test procedures, as well as over the
FTP, HFET and US06 cycles. They compared the owners' fuel economy and their fuel
economy estimates to the current label values and to US06 fuel economy levels. AAA
found that the US06 fuel economy was a better estimate of onroad fuel economy than
the current label value.

Our response:

      Regarding Honda's criticism of various aspects of the various organizations' test
procedures, we presented many of these elements  in the NPRM and/or the Draft TSD.
If sufficient estimates were available from test programs not having the weaknesses
mentioned,  it might be reasonable to avoid presentation of some or most of the
estimates. However, because estimates lacking weaknesses are currently unavailable,
we presented all of the estimates along with a complete description of their relative
strengths and weaknesses. Since Honda believes that the Strategic Vision data are the
best available, they presumably would prefer that these  estimates be emphasized.
However, as discussed below, we have discovered additional concerns with this
database, as well. Thus, we believe that it is still most appropriate to present all of the
organizations' estimates with appropriate caveats.

      In response to Honda's specific criticisms of the Consumer Report estimates,  we
believe their criticism of generic temperature adjustments is exaggerated. As  Honda
noted, the Consumer Report estimates do not include any cold starts or air conditioning
operation, the two conditions most affected by ambient temperature. Ambient
temperature only weakly affects the fuel economy of warmed-up conventional vehicles,
as discussed in the TSD. Hybrids could be more sensitive to temperature effects from
changes in battery charging and discharging efficiency.  Thus, Honda's criticism of
Consumer Report's hybrid estimates could be correct  in this area. However, this
difference in the impact of ambient temperature on hybrid and conventional warmed  up
fuel economy argues for the 5-cycle approach,  which Honda criticizes throughout their
comments.  Both the current and mpg-based label approaches can only apply the same
adjustment factors to all vehicles, hybrid or conventional. If the effect of temperature
differs for these two types of vehicles, or between individual vehicles, these label setting
approaches cannot reflect these differences.

      Regarding Honda's criticism of our analysis of the Consumer Report estimates,
we added a comparison in the final TSD of the relative ability of all three label
approaches to predict onroad fuel economy as estimated by Consumer Report.

      Honda's criticism of AAA's test procedures is very similar to their criticism of the
Consumer Report procedures.  Our response is the same as described  above.  Honda's
criticism of AAA's analysis of their new onroad  and US06 data is similar to their criticism
of our analysis of the Consumer Report data.  Honda believed that an analysis of the
                                     136

-------
ability of the various label approaches to match the onroad fuel economy estimates on
an individual model basis is more important than an analysis of the match-up on
average across all models. We will perform a more thorough analysis of this data in the
Final TSD.  While we will examine average fuel economy differences, we will also
examine the ability of various label approaches to reflect the onroad fuel economy of
individual vehicles.

      Honda presented two criticisms of the  FreedomCar data.  First, the vehicles are
primarily in  commercial service.  Second, no conventional vehicles are in the test
program. Honda was concerned that many vehicles in commercial service are operated
continuously, 6 days a week.  They cited maintenance records that include catalyst and
transmission replacement,  indicating extremely aggressive driving.

      Although commercial operation can differ from personal use, it provides an
opportunity to obtain many hours of vehicle operation over a short period of time. The
hybrid models cited by Honda have not been offered long enough for most vehicles in
personal use to achieve greater than  100,000 miles. Thus,  it is unclear whether the
maintenance received  by the  FreedomCar vehicles is unusual or not, particularly for
vehicles in a test program which likely receive more frequent and careful inspections
than the average vehicle. From Honda's description, it would appear that there are
factors affecting the FreedomCar Program's vehicles which increase fuel economy, as
well as decrease fuel economy relative to the average vehicle.  Continual operation
means few  cold starts. Phoenix certainly experiences little truly cold weather.  Both of
these factors reduce hybrid fuel economy more than conventional vehicles.
Commercial operation  might be more aggressive, but this is at best a subjective
observation. Air conditioning use will be higher in Phoenix than on average across  the
U.S. However, 24-hour a day commercial operation means much more operation at
night than the average vehicle, when  temperatures and solar loading are lower. Thus,
again, commercial operation in Phoenix might not be as extreme as it might look at first
glance.

      The  inclusion of conventional vehicles in the program would have allowed a
comparison of onroad fuel economy relative to label values for both conventional
vehicles and hybrids and would have provided more information. However, this was not
the focus of the DOE study. Overall, we do not believe that Honda has demonstrated
that the FreedomCar data is unrepresentative to the degree that it should not be
included in  the discussion of available onroad fuel economy estimates relative to label
values.

      Honda's criticism of EPA's Kansas City data includes all of the weaknesses
mentioned by EPA in the Draft TSD (e.g., few or no cold starts, little air conditioning
operation, limited monitored operation per vehicle, etc.).  In addition,  Honda noted that
the separate analysis of hybrid and conventional vehicle performance was
inappropriate, due to the low r-squared value for the hybrid vehicles (-0.17). This value
of r-squared is low because this study only included a  few hybrids, all of which fell into a
relatively narrow range of label fuel economy (48-56 mpg, with one vehicle at 64 mpg).
                                      137

-------
The low r-squared value indicates the relatively wide range of onroad fuel economy
values achieved with vehicles with a narrow range of label fuel economy values.

      Honda also criticized EPA for performing a regression of onroad fuel economy
versus label fuel economy, believing that a regression of fuel consumption (inverse of
fuel economy) would be technically preferable. We performed a regression of onroad
fuel consumption per mile versus the inverse of the current fuel economy value for
hybrid vehicles, as Honda suggested. First, we found that the intercept was not
statistically significant (p-value of 0.684). Thus, we performed a new regression with an
intercept of zero. We found an r-squared value of 0.18, which is not much different than
that for the regression of fuel economy.  The slope of the regression was 1.135,
indicating that the hybrids consumed 13.5% more fuel than predicted by the inverse of
their label values.  More importantly, this slope had a p-value  of lO^31, indicating that it
was extremely unlikely to be zero.  The 95% confidence interval for the slope ranged
from 1.09 to 1.18.

      We repeated  the analysis of fuel consumption per mile for conventional vehicles
and found very different results.  First, both the slope and the  intercept were statistically
significant at the 90% confidence level.  (The slope was statistically significant at the
95% confidence interval, but the intercept was not.)  The value of the intercept was
0.008 and the slope  was 0.85. Thus, vehicles with relatively low label values (i.e., 20
mpg) performed well on the road, but conventional vehicles with higher label values
(i.e., 35 mpg) showed a significant shortfall.

      We then combined the conventional and hybrid data and performed a single
regression, including an offset term for hybrids. (We did not evaluate the potential for a
different slope for hybrids.)  The offset term for hybrids was not statistically significant,
indicating that hybrids were not found to perform differently than conventional vehicles
during the Kansas City testing.  At the same  time, this testing  did not include cold starts,
which is the largest factor negatively affecting hybrid fuel economy relative to
conventional vehicles.

      Honda stated that EPA summarized the ORNL "Yourmpg" data well. Honda
presented two criticisms of EPA's analysis of this database. EPA found that hybrids
had a greater discrepancy between label and onroad fuel economy than conventional
vehicles. Honda believed that higher fuel economy vehicles, hybrid or conventional,
would have a greater onroad shortfall. Two,  EPA grouped all  hybrids together in
making this general  observation.  Honda found that this database indicates that not all
hybrids show the same higher degree of onroad fuel economy shortfall.  In particular,
the onroad fuel economy shortfall for the Honda Civic and Insight models were not
greater than those for conventional vehicles.

      In response to Honda's first comment on this database, we compared the onroad
fuel economy shortfall for conventional vehicles with relatively high fuel economy (i.e.,
those which had a current combined label value of at least 33 mpg) to those for all
conventional vehicles and hybrids. As discussed in more detail in Section  II.A.1 of the
                                      138

-------
Final TSD, we performed this comparison on the expanded Yourmpg database as it
existed on August 28, 2006. At this time, it consisted of roughly twice as many entries
as it did when NPRM was published, (8180 versus 4092 fuel economy entries). Using
the expanded database, we found that the onroad fuel economy shortfall for
conventional vehicles, conventional vehicles with current label values of 33 mpg or
greater, and hybrids were 1.4%, 2.3%, and 8.2%, respectively.  Thus, conventional
vehicles with relatively high label values appear to have a higher fuel economy shortfall.
However, the increase is very small. Hybrids, on the other hand, showed a much
greater shortfall.  Therefore, we disagree with Honda that the greater shortfall observed
for hybrids is due to their high label values.

      With respect to the shortfall observed for individual hybrids, Table 5-8 shows the
shortfall observed using the current database.

Table 5-8.  Hybrid Fuel Economy in the Yourmpg Database
Vehicle Model
Insight
Civic
Prius
Silverado
Escape/Mariner
Accord
Lexus
No. of Entries
30
113
226
2
58
22
45
Onroad Fuel
Economy
66.0
46.1
47.9
15.2
29.8
30.8
25.1
Current Label
Value *
65.0
48.9
53.8
17.5
31.9
32.3
29.4
Shortfall
2%
-6%
-1 1 %
-13%
-6%
-4%
-15%
      As can be seen, there is a range of shortfalls observed for the individual hybrid
models. However, there is also a wide range in the number of entries, so some care
needs to be taken in comparing across models. The Insight does show a smaller
shortfall than the other hybrids (actually a negative shortfall). However, there are only
30 fuel economy entries for Insights by 10 unique owners.  All the other hybrids show a
greater shortfall than the high fuel economy, conventional vehicles.  The shortfall for the
Accord hybrid exceeds that for the high fuel economy,  conventional vehicles by only
2%, but with even less entries than the Insight.  Finally, the Civic hybrid is on the lower
end of the shortfall spectrum with many more entries.  However, these 113 are by only
12 owners. The likelihood of significant differences in geographic location, driving
styles, etc. for those making entries for each hybrid model is large. Thus, it is  unclear
whether these differences in shortfall are real or not. Even with all hybrids combined,
there is concern that differences in driving style and location could affect the outcome.

      AAA found that the US06 fuel economy indicated onroad fuel economy more
accurately than  the current label values.  As previously mentioned, Honda disagreed
with this finding  and provided some additional regression analyses of the AAA data.  We
do not believe that the AAA analysis evaluates all the available alternatives for
estimating onroad fuel economy. For example, it only examines two options: the current
label values and the US06 fuel economy, from a sample size of only 17 vehicles.  Still, it
indicates the value of considering fuel economy occurring during aggressive and high
                                      139

-------
speed driving in estimating onroad fuel economy.  We believe that including both the
city and highway portions of the US06 test in the 5-cycle formulae accomplish this task.
We have added the new AAA data to our analysis of onroad fuel economy estimates in
the Final TSD along with AAA fuel economy estimates developed prior to the NPRM.

      Honda presented a number of correlations between the Strategic Vision data and
label values based on the current, mpg-based and 5-cycle approaches. Specifically,
Honda was able to match 95 vehicle models which were surveyed by Strategic Vision
and were also present in EPA's 5-cycle certification database.  First, Honda performed
a regression of onroad city and highway fuel consumption from the Strategic Vision
survey against the label values using the three approaches.  All three approaches
yielded good and similar correlation coefficients for both city and highway fuel economy.
The 5-cycle label values performed best for city fuel economy and the current label
values performed best for highway fuel economy.  However, in all cases, the results
were very similar. As discussed more below, the approximate nature of survey
respondents' estimates of onroad fuel economy may prevent a more precise evaluation
of the various label approaches.

      Second, in their supplemental comments, Honda performed a regression of the
inverse of the average fuel economy for each of the 95 models from the Strategic Vision
survey versus the fuel consumption over each of the nine dynamometer cycles or bags.
Honda used a backward, stepwise regression technique to identify the cycles or bags
which were correlated with city or highway fuel economy at a 90% confidence level.
The results are summarized in Table 5-9 below.
Table 5-9. Dynamometer Cycles Which Correlate with Strategic Vision Fuel
Economy
City Fuel Economy
Significant
correlation
Bag 1 , FTP
Bag1, Cold FTP
HFET
SC03


Non-significant
correlation
Bag 2, FTP
Bag 3, FTP
US06
Bag 2, Cold FTP
Bag 3, Cold FTP

Highway Fuel Economy
Significant
correlation
Bag 1 , FTP
Bag 1, Cold FTP
HFET
SC03
Bag 3, FTP
Bag 3, Cold FTP
Non-significant
correlation
Bag 2, FTP
US06
Bag 2, Cold FTP



      As shown in the table, Honda found that Bag 1 from both the FTP and cold FTP
tests, as well as the HFET and SC03 tests correlated with city fuel economy from the
Strategic Vision survey values. The remaining five cycles and bags did not show
statistically significant correlations once the other four bags and cycles were in the
model. For highway fuel economy, the same four cycles and bags showed a
statistically significant correlation, plus Bag 3 from both the FTP and cold FTP tests.

      Some of these findings appear to be reasonable, while others are surprising.  It is
reasonable that bags that include a cold start should be good predictors of city fuel
                                     140

-------
economy. The same is true for SC03 and the impact of air conditioning on city fuel
economy. However, it is very surprising that the HFET test (basically 45-60 mph cruise-
like driving) would be the best predictor of fuel consumption during relatively low speed,
start-stop, city driving. Honda expressed the same surprise.

      Regarding the regression  results for highway driving, it is surprising that the two
cold start bags would be important predictors of driving with presumably few cold starts.
The inclusion of the two Bag 3's, the HFET and the SC03 tests is more reasonable.
The exclusion of US06 is surprising, given the results of the MOVES driving pattern
analysis presented in Chapter III.A.2 of the Draft TSD and the prevalence of onroad
highway speeds above 60 mph.

      The fact that this type of regression analysis produces both expected and
unexpected findings is itself not surprising.  In fact, such a regression analysis probably
should not have been performed, as it violates an important premise of regression
analysis.  This premise is that the independent variables not be collinear. In this case,
fuel consumption values for all of the various dynamometer cycles and bags are
strongly correlated amongst themselves. Table 5-10 shows the correlation coefficients
(r-squared values) between the fuel economy values over a number of the combinations
of the various cycles and bags for the 95 vehicles that Honda used in their regressions.
Table 5-10. Correlation Between Fuel Consumption Values Over
Dynamometer Cycles
Cycles
Bag 2 FTP vs. Bag 3
FTP
Bag 2 FTP vs. HFET
Bag 2 FTP vs. US06
Bag 2 FTP vs. SC03
Bag 2 FTP vs. Bag 2
Cold FTP
Bag 2 FTP vs. Bag 3
cold FTP
R-Squared
Value
0.974
0.912
0.893
0.943
0.973
0.953
Cycles
Bag 3 FTP vs.
HFET
Bag 3 FTP vs.
US06
Bag 3 FTP vs.
SC03
Bag 2 Cold FTP vs.
Bag 2 Cold FTP
Bag 3 FTP vs. Bag
3 Cold FTP

R-Squared Value
0.971
0.958
0.945
0.966
0.983

      As can be seen in Table 5-10, all of the correlation coefficients are 0.89 or
higher, indicating a very strong degree of collinearity between the various fuel
consumption values.  This is not surprising, since large vehicles will have relatively high
fuel consumption rates over all the cycles and small vehicles will have relatively low fuel
consumption rates over all the cycles. These high correlation coefficients indicate that
one simply cannot regress onroad fuel economy estimates against fuel consumption
values from dynamometer cycles.  There is insufficient independence between the fuel
consumption values over the various dynamometer cycles.
                                      141

-------
      A much more sophisticated regression analysis will be necessary to evaluate
how various dynamometer fuel economy values predict onroad fuel economy.  First, it
will be necessary to utilize high quality onroad fuel economy measurements, as
opposed to survey responses. As described in Section II.A.3 of the Final TSD, there
appears to be a significant bias in the Strategic Vision survey responses towards fuel
economy values which are a multiple of 5.  This indicates that a number of the survey
responders (estimated to be about 13%) were rounding off their estimates to the
nearest 5 mpg.  This is insufficient detail for the type of regression performed by Honda.
Second, it will probably be necessary to incorporate descriptions of the vehicle activity
which led to specific onroad fuel economy measurements. The latter would allow the
direct evaluation of cycles or bags which are aimed  at specific operating conditions, like
cold starts at various ambient temperatures and air conditioning.  This would also avoid
the problems with collinearity present when all the cycles and bags are included in the
regression at the same time.
                                      142

-------
Chapter 6:    Testing Provisions

6.1    General Test Procedure Issues

What we proposed:

      We proposed the following changes to fuel economy test procedures: 1)
collecting the US06 exhaust sample into two bags as opposed to the current one-bag
sample; 2) requiring heater/defroster operation during the Cold Federal Test Procedure
(FTP) testing (currently optional); 3) requiring Cold FTP testing for diesel vehicles; and
4) codifying requirements for four phase FTP testing of gasoline-electric hybrid vehicles
rather than allowing this practice under special test procedures contained in 40 CFR
86.1840-01. As stated in the proposal, these changes were to take effect beginning
with the 2008 model year.

What commenters said:

      AAM/AIAM expressed general concerns with any changes that:

      "Are required for 2008 model year testing without sufficient lead-time;"
      "Could harm data accuracy and repeatability;"
      "Could impact stringency or misrepresent real-world conditions;" or
      "Could add significant cost and complexity without a value-added benefit."

Our response:

      The general concerns expressed by AAM/AIAM apply to the entire set of test
procedure changes set forth in the proposal. As such, our responses to specific
comments on each individual  test procedure change will indirectly address the general
concerns and provide more detail.  However, as background for the responses to
specific comments, the following paragraphs summarize the specific test procedure
changes, identify the associated general concerns, and provide a general response.

      Upon analysis of AAM/AIAM's comments, we acknowledge the importance of
allowing enough lead time to change 2008 model year testing.  We are addressing this
concern for all specific test procedure changes, except the Hybrid four-phase, four-bag
FTP, by extending the applicable effective date until the 2011 model year. We are also
finalizing alternative procedures for the 2008 through 2010 model year interim period for
manufacturers choosing to use the  5-cycle approach. Therefore, the final rule
addresses lead time concerns by providing additional lead time and alternative test
procedures for the interim  period, as proposed by AAM/AIAM.

      AAM/AIAM commented that  dividing the US06 exhaust sample into city and
highway samples will affect data accuracy and test repeatability, although they did not
provide data to support this claim.  We performed a ten vehicle study  examining the
correlation and variability of the one-bag and two-bag US06 exhaust sampling
                                     143

-------
approaches that demonstrated nearly identical fuel economy results and equivalent
variability and measurement accuracy of criteria pollutants for the one-bag and two-bag
US06.32 While we continue to believe the two-bag US06 measurement proposed is a
valid approach that will not lead to significant differences in emission results, we also
believe that the alternative approaches suggested by the auto industry could yield
technically valid results.  We have therefore revised the proposal and are finalizing the
requirements for the two-bag US06 measurement as outlined in Preamble Section IV.

      AAM/AIAM commented that operating the heater/defroster device during the
Cold FTP test may impact stringency or misrepresent real-world conditions.  We
proposed test protocols for climate control settings including specifying appropriate fan
speed settings,  timing of turning on the heater/defroster during the test, and accounting
for various vehicle climate control designs (e.g.,  systems with automatic temperature
adjustment). AAM/AIAM opposed some of the specific test protocols, arguing that a
driver would not likely maintain the fan at maximum speed during an entire driving trip.
AAM/AIAM stated that many electronic systems  automatically control fan speed as the
vehicle warms up, and further noted that some vehicles can not operate the defroster
while the blower is off, which we proposed as a possible test protocol.

      We agree with the comments from AAM/AIAM that the protocol could be modified
to better simulate how a driver would typically operate the heater/defroster system in the
real world, and we acknowledge the design  challenges identified.

      We disagree with the AAM/AIAM comment that operating the heater/defroster
system will impact stringency.  As demonstrated in EPA's docketed study (Docket No.
EPA-HQ-OAR-2005-0169-0067, SwRI, "Fuel Economy Impacts of Interior
Heater/Defroster Usage on Conventional and Hybrid Gasoline-Powered Vehicles"  U.S.
EPA, Office of Transportation and Air Quality, 2005.), the worst case scenario of
activating the heater/defroster immediately upon start of test had an insignificant
emission impact.  In addition, we are waiting until 2 minutes into the test to turn on the
heater/defroster, since,  in reality, it is unlikely that passengers would blow cold air at
maximum speed when first starting the vehicle.  This further reduces any potential
emissions impact, since nearly all catalyst systems should have achieved light off
temperature at that point.

      To address the AAM/AIAM comment that running the heater/defroster system as
proposed may misrepresent real-world  conditions, we are finalizing test protocol that
allows fan speed to be reduced 505 seconds after the start of the test during the idle
period between hill 5 and hill 6 of the FTP, which corresponds to the end of bag 1 and
the start of bag 2.  We are allowing manufacturers having unique climate control
systems to use alternate heater/defroster test protocols, under the special test
procedures contained in 40 CFR 86.1840-01, with prior agency approval.
   32  Mitcham, A. & Fernandez, A., "Revising the US06 Test Procedure into a Two Sample Period Test"
U.S. EPA, Office of Transportation and Air Quality, 2005. Docket No. EPA-HQ-OAR-2005-0169-0058.


                                      144

-------
      Finally, AAM/AIAM commented that the proposed test procedure changes
impose significant additional costs without a value-added benefit. We proposed these
changes to ensure that the fuel economy test procedures more closely represent real-
world conditions, and we disagree with this comment for the following reasons. First,
the proposed changes apply to existing test procedures and do not require new test
procedures or alter any of the drive schedules.  Second, some modifications specifically
for the two-bag US06 exhaust collection, have an initial fixed cost for minor site
modifications, if necessary, including dynamometer analyzer software revisions,
additional analyzer sampling hardware (i.e., exhaust sample bags, exhaust tubing, etc.),
and post-revision validation testing.  In response, we have modified the final  rule to
allow for alternative methods of determining the city/highway ratio using a one-bag
US06.  Third, we have revised certain test procedure changes in the final rule for the
Cold FTP diesel test; as a result, some anticipated hardware changes are no longer
necessary.  We changed these procedures to enable manufacturers to use the 5-cycle
approach without the additional test burden of generating 5-cycle data inputs.
Therefore, we believe that the test procedure changes provide sufficient added benefit
without additional significant cost.
6.2   Test Requirements for Vehicles Currently Exempt from Certain Emissions
      Tests

      6.2.1     Diesel Vehicles

What we proposed:

      Diesel vehicles are not currently subject to Cold CO emission standards and,
thus, do not have a 20 degree F FTP fuel economy result to use in the 5-cycle equation.
Therefore, beginning with the 2008 model year, we proposed that diesel vehicles must
perform a Cold FTP for the purpose of collecting fuel economy data. We proposed the
use of a winter grade diesel fuel for such a test.

What commenters said:

      AAM/AIAM raised numerous concerns with subjecting diesels to the Cold FTP
test in the 2008 to 2010 timeframe, arguing that EPA does not recognize the significant
laboratory changes that will be needed. They cite the potential for major laboratory
retrofitting, which in some cases might not even be possible with the existing lab
configuration. They also remarked that that there are no test protocols developed for
new diesel technologies on the  horizon.  They also noted that the proposed regulations
would require Cold  FTP testing  on all diesel certification vehicles, unlike the gasoline
requirements which require testing on a durability group basis.  For these reasons,
AAM/AIAM recommends that EPA wait until the 2011 model year to require diesels to
perform the Cold FTP, giving them more time to address test facility concerns.
Although they did not oppose the test fuel proposal beyond lead time issues, they
                                      145

-------
proposed to instead calculate fuel economy based on CO and C02 measurements,
excluding HC.

      Volkswagen, DaimlerChrysler, and Nissan reiterated the concerns expressed by
AAM/AIAM. Volkswagen noted that their preliminary testing indicates the possible
presence of some issues with diesels that are not yet fully understood, thus arguing for
additional lead time.  Nissan stated specific technical issues that they could not likely
overcome before the 2010 model year.  DaimlerChrysler supported the notion that fuel
economy values should reflect cold testing, but reiterated AAM/AIAM's lead time
concerns, suggesting that Cold FTP testing for diesels be optional in the 2008-2010
model years.

Our response:

      We agree that significant laboratory changes are necessary for Cold FTP diesel
testing, since diesel vehicles are not currently subject to FTP testing at 20 degree F.
Therefore, we are revising the proposed provisions for Cold  FTP diesel testing as
detailed in the paragraphs below.

      First, we are not requiring the measurement of particulate matter (PM) during the
Cold FTP diesel test, since PM is not part of the fuel economy calculation, and thus has
no impact on fuel economy.  This eliminates the need for a PM diesel tunnel,  an
extensive change that would have been required under the proposal.

      Second, we are delaying the requirement for Cold FTP diesel testing to allow
manufacturers additional lead time to address facility concerns.  For manufacturers that
choose to use the 5-cycle approach during the 2008 - 2010  model years, we  are
requiring diesel Cold FTP testing, but fuel economy may be  reported based on CO and
C02 measurements only. We are not requiring the measurement of hydrocarbons (HC)
during the Cold FTP  diesel test during the 2008-2010 model years, because according
to current HC  data, excluding HC from the fuel economy calculation has a negligible
impact (i.e., less than 0.1%) on diesel fuel economy reporting. This also provides
additional lead time for manufacturers to make laboratory upgrades such as a heated
flame ionization detector (FID), heated sample lines and additional, external drying
capacity for diesel HC measurement.

      Third, in the 2011 model year, we are requiring manufacturers to conduct and
report results from the Cold FTP diesel testing, including HC measurements.  Because
diesel emission control technology is evolving rapidly, we believe that HC contribution
will  warrant future measurement and inclusion in the fuel economy calculation.  This
should provide sufficient lead time for any laboratory changes for cold temperature
capability and specific HC measurement equipment.

      Finally, any Cold  FTP diesel testing performed for optional 5-cycle testing during
model years 2008-2010 and/or for 2011  beyond must use the winter grade fuel
specification discussed in the proposal and finalized in this rulemaking.
                                     146

-------
      6.2.2     Alternative-Fueled Vehicles

What we proposed:

      Flexible-fuel vehicles (FFVs) are subject to the Supplemental Federal Test
Procedure (SFTP; which includes the US06 and SC03 tests) and Cold Carbon
Monoxide (CO) emission standards and test requirements, but only when operating on
gasoline. Thus, we proposed that the fuel economy label values of FFVs when
operating on gasoline be determined using the same mpg-based or 5-cycle approaches
applicable to gasoline vehicles and thus additional testing for US06, SC03 and Cold
FTP while operating on the alternative fuel would not be required. We also proposed a
methodology for how manufacturers of FFVs are to determine and report fuel economy
label values, when a FFV is operated on the alternative fuel.

What commenters said:

      We did not receive any comments  on this proposal.

Our response:

      Since we did not receive comments on this proposal, we are finalizing it as
proposed.


6.3   Modifications to Existing Test Procedures

      6.3.1     Revisions to US06 Bag Measurements

What we proposed:

      The US06 driving schedule contains elements of both city and highway driving,
yet the exhaust sample is collected in only one sample, or "bag." For an emissions test,
the one-bag sample is appropriate, since  emissions are summed over the entire test to
determine compliance. However,  the ideal  methodology for fuel economy would be to
reflect the city portion in the city miles per gallon (mpg) estimate and  the highway
portion in the highway mpg estimate. Consequently, we proposed a test protocol that
would collect the US06 exhaust sample into two-bags. We conducted a test program
that demonstrated that two-bag collection is technically feasible, requiring only some
software reprogramming and minimal hardware reconfiguration. We  also demonstrated
that a split-phase sampling period does not significantly alter the criteria pollutant
emission results.
                                     147

-------
What commenters said:

      AAM/AIAM generally disagreed with EPA's assessment that the costs of
collecting US06 exhaust emissions would be minimal, characterizing the test procedure
change as "substantial," yet they did not provide data to support this comment. Toyota,
Nissan, Mitsubishi, and DaimlerChrysler agreed.

   AAM/AIAM expressed concern with the accuracy and variability of the test when split
into two phases.  They recommended that EPA not adopt the two-bag US06, but
instead allow the use of ratio factors (i.e.,  ratio of the US06 city portion/whole US06 test
and the US06 Highway portion/whole US06 test) or alternative methods to estimate the
city and highway fuel consumption from a traditional single-bag US06.  This approach,
they argue, would result in substantially similar results at truly  minimal cost. Toyota,
Nissan, Mitsubishi, and DaimlerChrysler independently expressed a preference for ratio
factors. They proposed the following:

   o  For conventional gasoline vehicles: Use ratio factors of 0.68 for the city fuel
      economy and 1.16 for the highway fuel  economy, which were used to derive the
      mpg-based formulae.
   o  For hybrid vehicles: Use 0.68 and 1.16  ratio factors for 2008 through 2010 model
      years, and gather data to evaluate  whether continued use of these factors is
      appropriate for hybrids.
   o  For diesels: Confirm the 0.68 and 1.16 ratio factors after compiling 2007 and
      2008 test data.
   o  For all vehicles: Allow manufacturers to propose alternative ratio methods for
      EPA approval (e.g., fuel injector fuel economy data, model CCb data).

Toyota suggested that hybrid manufacturers submit data, and  if it resembles that of
conventional gasoline vehicles, then those factors can be used.  If hybrids are seen to
be substantially different, Toyota suggests that EPA allow each manufacturer to use
their own factors based on data and good engineering judgment.

      AAM/AIAM commented that separating the two-phase bag software from the
one-phase PM sample collection system for diesels  requires significant software
changes, cost, and lead-time.  They also noted that diesel testing will be complicated by
the need to align integrated real-time collection of total hydrocarbon (THC) data for fuel
economy calculations with bag collection of methane (CH4) for emissions compliance.
Nissan and DaimlerChrysler reiterated these diesel issues.

      DaimlerChrysler also noted that this provision as proposed in 40 CFR Part 86
would impact all US06 testing, including certification and IUVP testing, and would result
in additional unnecessary expense.  They suggested that the two-phase US06
provision, if adopted, should be written into Part 60 to ensure that the context is limited
to fuel economy testing.
                                      148

-------
Our response:

      We disagree with the comments from AAM/AIAM, Toyota, Nissan, Mitsubishi,
and DaimlerChrysler that the test procedure changes and the costs of collecting US06
exhaust emissions in two-bags are "substantial."  The changes necessary to enable
two-bag US06 testing involve an initial, one-time, fixed cost software revision and, in
some cases, an additional sample bag, sample line and analyzer plumbing for collection
of US06 emissions in a second bag.

      We also disagree with the comment from AAM/AIAM that splitting the US06
exhaust into two bags decreases accuracy and increases variability. As mentioned
above, EPA performed a ten vehicle study to examine the correlation and variability of
the one-bag and two-bag US06 exhaust sampling approach (Docket No.  EPA-HQ-OAR-
2005-0169-0058, Mitcham, A. & Fernandez, A., "Revising the US06 Test Procedure into
a Two Sample Period Test" U.S. EPA, Office of Transportation and Air Quality, 2005.)
which demonstrated low variability, nearly identical fuel economy results,  and equivalent
variability and measurement accuracy of criteria pollutants for the one-bag and two-bag
US06. AAM/AIAM neither addressed the results of this study nor submitted additional
data to refute the results of this study.

      Regarding the suggestion from AAM/AIAM, Toyota,  Nissan, Mitsubishi, and
DaimlerChrysler to use alternate ratio factors, the ratio factors in the proposal were
used for our 5-cycle assumptions based on the data available at the time. These ratio
factors are clearly inappropriate for hybrid vehicles and other advanced technology
vehicles (i.e.,  vehicles with cylinder deactivation, engine idle shut off operation, etc.).

      Although we disagree with the comments that the cost and procedure changes
are "substantial," we recognize that these changes take time to implement.  We also
agree with the industry concern about accurately timing the real-time PM  collection for
emissions compliance with the bag collection of hydrocarbons for fuel economy
information in the US06 test for diesels.  Because manufactures may need additional
lead  time for gasoline vehicles and because two-bag diesel fuel economy testing is
more complicated, we are revising the provisions for the applicable model years for the
two-bag US06 from the proposal. We are not requiring the two-bag US06 for model
years 2008 through 2010. However, for manufacturers choosing to use the 5-cycle
approach for model years 2008 through 2010, the two-bag US06 must be conducted or
data otherwise supplied in two-bag US06 format (as described further below).
Therefore, for model year 2011 and beyond, we are requiring a city/highway US06
calculation for use in the 5-cycle approach.

      In addition, as suggested by the Alliance and AIAM, we are allowing the use of
additional methods for measuring and/or determining a two-bag US06 fuel economy in
lieu of conducting an actual two-bag US06 test.  Some suggested methods  include: 1)
conducting a one-bag US06 and using emissions analyzer modal data to determine the
appropriate ratio of city and highway operation; or 2) conducting a one-bag US06 and
using On-Board Diagnostic (OBD) fuel rate (e.g.,  grams of fuel  per second) data to
                                     149

-------
determine the appropriate ratio of city and highway operation over the one-bag US06.
Additionally, the manufacturers may use other methods based on good engineering
judgment, with EPA review and approval, as long as these methods achieve equivalent
or more technically valid results, based on manufacturer submitted data.  These
additional options for deriving or measuring the city/highway US06 fuel economy are
allowed for manufacturers optionally using the 5-cycle approach in model years 2008
through 2010, and for all manufacturers in model year 2011 and beyond, when the 5-
cycle approach is required.

      In the case when manufacturers conduct one-bag US06 tests and use modal
emissions and onboard diagnostic (OBD) fuel rate data, the ratio of city and highway
operation over the one-bag US06 is applied to the CO, C02 and HC results in order to
determine the city and highway US06 fuel economy values, constituting a "virtual" two-
bag US06. However, this  option  only applies for determining the city and highway US06
fuel economy and, thus, is not applicable for determining US06 emissions.  This option
should also address the concerns of one-phase PM sample collection systems,
integrated real-time total hydrocarbon data collection for fuel economy calculations, and
the timing of CH4 bag measurements for diesels since a one-bag US06 can still be
used.

      We disagree with the comments from Daimler-Chrysler that writing this provision
as proposed in 40 CFR Part 86 would impact all US06 testing, including certification and
IUVP testing, resulting in additional and unnecessary expense. We proposed using
existing test procedures for emissions testing in order to minimize test burden and any
unnecessary expenses. Writing the requirements for a two-bag US06 test into Part 600
for fuel economy purposes would require manufacturers to conduct a one-bag US06 for
emissions purposes and an additional, two-bag US06 for fuel economy purposes,
adding additional, unnecessary testing and expenses, and duplicating effort.  We are
writing the final regulation  such that 40 CFR part 86 will contain both 1-bag or 2-bag
US06 as an option. Part 600 will require that manufacturers determine the US06
City/Highway ratios based on one of several methods, with 2-bag US06 results as one
of these options.
      6.3.2     Heater/Defroster Use During the Cold FTP

What we proposed:

      The current Cold FTP test conducted at 20 degrees F provides the option to use
the header and/or defroster. Some manufacturers routinely use this option to heat the
vehicle cabin, others do not. In order to more closely reflect real-world operation, and to
ensure a level playing field across manufacturers and vehicle lines when performing this
test, we requested comment on whether we should require operation of the
heater/defroster during this test.  We suggested a test protocol for doing so, but we also
noted that there are many possible approaches. We proposed to set the heat level to
high and direct the airflow to the windshield for defrost.  After two minutes the fan speed
                                     150

-------
would be turned to high and left there for the duration of the test.  Automatic climate
control systems would be set to defrost mode and to a temperature of 72 degrees F.

What commenters said:

      AAM/AIAM proposed that EPA defer consideration of this concept until we have
a better understanding of real-world operation and of the potential impact on the
stringency of existing and upcoming emission standards (e.g., Mobile Source Air
Toxics).   They argued that a driver would not likely maintain a maximum fan speed for
43 minutes (the effective length of the test given the emission weighting equation), and
that this worst-case scenario surpasses EPA's intent to reflect real-world conditions.
AAM/AIAM also noted that many electronic systems automatically reduce fan speed as
the vehicle warms up, and that some vehicles can not operate the defroster without the
blower, as we proposed in the first two minutes of the test. They suggested that we
allow manufacturers choosing the 5-cycle approach in the 2008 - 2010 model years to
use the heater/defroster for testing  of fuel economy data vehicles "per manufacturer
procedure using good engineering judgment and with consideration of expected typical
real-world operating patterns."

      DaimlerChrysler, Mitsubishi, and Nissan independently reiterated these
comments and recommended that  EPA reevaluate how well the concept represents
real-world behavior.  Mitsubishi noted that their test data indicates a far smaller impact
on fuel economy due to defroster/heater operation than EPA estimates in the proposal,
but did not provide data to support  this claim.  Toyota also called for more data and a
better understanding of the fuel economy and emission impacts of heater/defroster
operation, and suggested along with DaimlerChrysler that changes to the cold test
method should be reevaluated in the context of the proposed Mobile Source Air Toxics
regulations.

      Bluewater Network, UCS, NYDEC, Montana-CHEER, and  Public Citizen support
using the heater/defroster during the Cold FTP test. Some suggested that the protocol
could be improved with additional data and study of real-world driving conditions. UCS
suggested that EPA develop a standardized methodology based on realistic usage
patterns. Public Citizen highlighted that without a heater/defroster requirement,
manufacturers who choose to use more realistic test conditions may be penalized
relative to those who do not.

Our response:

      The purpose of requiring heater or defroster use during the Cold FTP is to more
closely reflect real-world operation, and to ensure a level playing field across
manufacturers and vehicle lines when performing this test. Based on the comments we
received, we are finalizing the requirement for mandatory heater/defroster operation
during the Cold FTP. We are revising the approach outlined in the NPRM to reflect
   33 See regulation in 40 CFR 86.1811-10(g) (March 29, 2006).
                                      151

-------
more closely real-world operation, and by accommodating various climate control
systems and addressing issues of lead time with respect to applicable model years for
mandatory heater/defroster operation during the Cold FTP.

      First, we agree with the AAM/AIAM comment that a driver will not likely maintain
a maximum fan speed for 43 minutes since in reality the vehicle interior would reach a
comfortable temperature for occupants and the fan speed would manually or
automatically be reduced to  a lower setting. Therefore, for manual climate control
systems, the final regulations will require that the fan speed setting be reduced to the
lowest possible to maintain air flow at approximately 505 seconds after the start of the
test during the idle period between "hill" 5 and "hill" 6 of the FTP (the point that
corresponds to the end of bag 1 and the start of bag 2), and the temperature setting will
remain at the hottest setting. These settings will be held for the remainder of the test,
including the final bag following the  10 minute soak period.  For automatic climate
control systems, the selector will be set to heater or defroster mode and to a
temperature of 72 degrees F for the duration of the test.  We are finalizing all other
aspects of heater/defroster operation as outlined in the proposal.

      For vehicles with multiple zone climate control systems,  the same fan and
temperature settings should be maintained throughout the zones for both manual and
automatic interior climate control systems. This and other unique designs may require
special, alternative protocols/procedures during the Cold FTP.  Therefore, with prior
agency approval, manufacturers may request and use alternative heater/defroster test
protocols/procedures if they have a  unique climate control system.

      Finally, we are revising the applicable model year for implementation  of
mandatory heater/defroster operation during the Cold FTP from model year 2008 to
model year 2011.  This will allow manufacturers time to fully assess any impacts related
to the Mobile Source Air Toxic (MSAT) cold volatile organic compound (VOC) proposed
standards, which would also be determined based on the Cold FTP test.  However, if
manufacturers choose to perform 5-cycle  testing between the 2008 through 2010 model
year, they must follow the heater/defroster protocols/procedures set forth in this
regulation.
      6.3.3    4-Phase, 4-Bag FTP for Gasoline-Electric Hybrid Vehicles

What we proposed:

      The  FTP consists of two parts, referred to in the regulations as the "cold start"
test and the "hot start" test. Each of these parts is divided into two periods, or "phases":
a "transient" phase and  a "stabilized" phase.  For conventional vehicles, the stabilized
phase of the hot start test has been assumed to be identical to the stabilized phase of
the cold start test, allowing the test to be shortened by not running the hot start
stabilized phase.  However, because gasoline-electric hybrid vehicles have two energy
sources that can be combined in many ways, EPA and manufacturers recognized that
                                      152

-------
the assumption regarding the equivalence of the stabilized phases of the hot and cold
start tests may be invalid for hybrid vehicles. Consequently, we have historically
required vehicles with gasoline-electric hybrid systems to perform the complete set of
four phases of the FTP under existing provisions referenced in the regulations (40 CFR
86.1840-01) for special test procedures.  Rather than continue this practice, we
proposed to develop explicit regulatory language to require full-four phase testing of
hybrid-electric vehicles.  We proposed to require that hybrid vehicles conduct all four
phases of the FTP for both emissions and fuel economy testing, including the Cold FTP.

What commenters said:

      AAM/AIAM cited 40 CFR 86.1811-04(n), which aligns with California, as already
requiring two full Urban Dynamometer Driving Schedules (UDDSs) for hybrid vehicles
and allowing the emissions to be collected in two-bags (one per UDDS).  AAM/AIAM
supported extending the two-UDDS requirement to hybrid vehicles conducting the Cold
FTP. While they agreed that we should require  all four phases of the FTP, they
disagreed with our proposal requiring that emissions be collected in four bags.  They
suggested that we allow manufacturers the option to use four sample bags, because
requiring this methodology would force lead-time issues and costly facility modifications.
They also noted that the four-phase two-bag approach improves accuracy and
alignment with California. They proposed  the following:

   o  Retaining section 86.1811-04(n) as-is;
   o  Consider a new section  86.1811-11 (n) that would delay these requirements for
      the Cold FTP test;
   o  Define both two-Bag FTP and four-Bag FTP in Part 600.002
   o  Add 5-cycle fuel economy equations that clarify the application for both two-bag
      and four-bag testing;
   o  Weight the bag fuel consumption elements of the 5-cycle equations by theoretical
      distance traveled to ensure consistent label adjustments between two- and four-
      bag data.

   Nissan noted that in order to implement a 4-bag FTP, they would need new
hardware, software, and test equipment, thus making the 2008 model year difficult to
achieve.  They also expressed  concern that a 4-bag  requirement might not be
consistent with California, and that auto companies may need to support two test
equipment regimes.

Our response:

      There are two distinct but related protocols for FTP testing of hybrid-electric
vehicles: 1) the driving schedule, which requires performance of the FTP over four
phases (i.e., two consecutive Urban Dynamometer Driving Schedule or UDDS cycles)
versus the three phase FTP for conventional vehicles (i.e., one UDDS and repeat of
only the first 505 seconds (Hot 505) of the UDDS with the emissions from last 864
seconds of the UDDS assumed to be same as in the first UDDS), and 2) the
                                     153

-------
measurement of emissions during the four phase FTP by either collecting emissions in
2-bags, combining emissions from the first two and last two phases, or collecting
emission in 4-bags by measuring emissions during each of the four phases. We
proposed provisions for both protocols and we will address them beginning with the
four-phase FTP and  then the 4-bag versus 2-bag FTP.

      The AAM/AIAM commented that 40 CFR 86.1811 -04(n), which aligns with
California, already requires two full UDDSs for hybrid vehicles.  Regulatory language
contained in 40 CFR 86.1811-04(n)  refers to and incorporates by reference the
California procedure entitled "California Exhaust Emission Standards and Test
Procedures for 2003 and Subsequent Model Zero-Emission Vehicles, and 2001 and
Subsequent Model Hybrid Electric Vehicles, in the Passenger Car, Light-Duty  Truck and
Medium-Duty Vehicle Classes" (August 5, 1999).  In the California procedure,  section
6.2 describes the dynamometer procedures for hybrid electric vehicle emission testing
and states:  "...The dynamometer run consists of two tests, a "cold" start test... and a
"hot" start test..." by  conducting two consecutive UDDS cycles. Before this  statement,
the introductory paragraph for this section also states that "Alternative procedures may
be used if shown to yield equivalent results." In addition, EPA proposed that hybrid
electric vehicles undergo a full four-phase FTP under the special test provisions in 40
CFR 86.1840-01.

      In the proposal, we intended to develop explicit regulatory language that would
require manufactures to use a full four-phase FTP  test for hybrid-electric  vehicles
instead of continuing to reference the special test procedure provisions in 40 CFR
86.1840-01.  However, upon further review, we agree with AAM/AIAM that 40 CFR
86.1811-04(n), which references the GARB procedure requiring the full four-phase
FTP for hybrid-electric vehicles, adequately covers the requirement for a full four-phase
FTP.  Therefore, we  will retain 86.1811-04(n) as written, which addresses the
comments from AAM/AIAM and Nissan and no longer need to develop further explicit
regulatory language  requiring a full four-phase FTP for hybrid-electric vehicles.

      EPA currently collects the emissions  in four bags corresponding  to the following
four phases: cold transient emissions collected in bag 1  (505 seconds), cold stabilized
emissions collected in  bag 2 (864 seconds), hot transient emissions collected  in bag 3
(505 seconds or a Hot 505), and hot stabilized emissions collected in bag 4 (864
seconds. For conventional vehicles, bag 4 is not repeated since we assume that  the
emissions are the same as in the cold stabilized  phase). Accordingly, our test
procedures are aligned with GARB for hybrid-electric vehicles, contrary to the
comments from Nissan, since both tests require the full four phases, or two consecutive
UDDS cycles.  Therefore, the only difference between a four-phase, 2-bag FTP (where
bag 1 +bag 2 =  bag A and bag 3+bag 4  = bag B) and a four-phase, 4-bag  FTP is the
emissions sample collection method and the analysis of the test results.

      The AAM/AIAM commented that it supports extending the use of a four-phase
FTP to hybrid vehicles conducting the Cold FTP. However, after further consideration,
we are not going to extend the requirement for four-phase FTP testing of hybrid vehicles
                                      154

-------
to the Cold FTP as originally stated in the proposal. The lower temperature (20 F) for
the Cold FTP versus the typical temperature (75 F) for the FTP could impact the hot
stabilized phase (i.e., bag 4) fuel economy for the Cold FTP in that the vehicle may not
truly be "hot" or "stabilized" during the normally substituted cold stabilized phase (i.e.,
bag 2) of  a 3 bag test. This is the case for both conventional and hybrid vehicles and
therefore retaining the current Cold FTP 3 bag test procedures maintains a level playing
field between conventional and hybrid vehicles when calculating running fuel
consumption at 20 F. Further, the comments cited that requiring four bags would force
facility modifications with significant costs and lead time issues. We feel that is
particularly relevant in the case of 20 F Cold FTP test facilities which have been strictly
designed around the 3-bag FTP test procedures. Therefore, for hybrid-electric vehicles,
the Cold FTP will continue to omit the fourth phase of the FTP emission test and collect
the emissions for the  remaining three phases (i.e.,  3-bag FTP), as currently required.

      AAM/AIAM disagreed with our proposal requiring that emissions be collected in
four bags,  stating that the four-bag approach would require costly facility modifications
and lead-time issues.  Accordingly, AAM/AIAM suggested that a 4-bag FTP be optional,
indicating that a 2-bag FTP increases accuracy and alignment with GARB. We
recognize that some manufacturers and GARB may require  new hardware, software,
and test equipment to implement a 4-bag test. Since our test procedures are aligned
with GARB requiring full four phase FTP testing for hybrid-electric vehicles, this is an
issue  of how to divide and analyze the emissions results.  Therefore, we agree with the
AAM/AIAM suggestion to optionally allow a 4-bag FTP. As a result, we will accept both
four-phase, 2-bag FTP results and four-phase, 4-bag FTP results.  Accordingly, we will
develop an optional 2-bag 5-cycle fuel economy formula/equation that will use
appropriate 5-cycle, bag fuel consumption weightings based on theoretical distance
traveled to ensure consistent label adjustments for two-bag and four-bag FTP testing,
as suggested by AAM/AIAM.

      We disagree with the AAM/AIAM recommendation to consider a new section in
40 CFR 86.1811-11(n) that would delay the requirement for the four-phase, 4-bag Cold
FTP test and define both the two-bag FTP and four-bag FTP in 40 CFR 600.002. First,
the 3-bag FTP is currently required and used for Cold FTP testing of hybrid-electric
vehicles, and thus it is unnecessary to create a new section in 40 CFR 86.1811 -11 (n)
to delay the four-phase, 4-bag  Cold FTP. Second, the FTP definition in Subpart A -
"Fuel  Economy Regulations for 1977 and Later Automobiles - General Provisions" 40
CFR 600.002-85(a)(7) references "procedures described in part 86," including 40 CFR
86.1811-04(n), which references the California procedure and describes the four-
phase, 2-bag FTP. Therefore,  we also  do not believe it is necessary to further define
the 2-bag and 4-bag FTP in 40 CFR 600.002, since they are completely described in
the referenced part 86 provisions.
                                      155

-------
6.4   Test Fuels

What we proposed:

      With the exception of specifying a fuel to be used for Cold FTP testing of diesel
vehicles, we did not propose changes to the fuels used for current emission or fuel
economy tests.  However, in the proposed rule's discussion of the Energy Policy Act of
2005, we interpreted the statute's reference to "current reference fuels" to mean the
laboratory fuels used to perform the fuel economy tests. We understood the underlying
concern of Congress to be the possibility that the high-quality lab fuels would give
higher fuel economy results that the typical fuel used by consumers.  The test fuel
specifications are detailed in the emission compliance  regulations.  The test fuel is
roughly equivalent to premium, high-octane fuel available at the pump.

What commenters said:

      The NYDEC disagreed with the use of premium, high-octane fuel, and suggested
that actual in-use fuels provide fuel economy estimates closer to what consumers would
actually experience in real world conditions.

Our response:

      We disagree with the comments from NYDEC regarding the use of in-use fuels.
The test fuel used for emissions certification and fuel economy testing is required to
have specific and repeatable characteristics to ensure  consistency in test results. While
certification fuel is roughly equivalent to premium, high octane fuel available to
consumers, EPA has previously established requirements for emissions and fuel
economy testing that ensure that the octane level of certification fuels does not
influence emissions or fuel economy test results.

      Under the provisions of 40 CFR 86.090-27, EPA requires manufacturers to
show prior to certification that knock sensor equipped vehicles, which could  take
advantage of elevated  octane levels, could operate on 91 research octane number
(RON) (equivalent to commercially available 87 octane fuel) without affecting emissions
or fuel economy.  Manufacturers are required to perform testing on both 95 RON
(equivalent to commercially available 93 octane fuel) and 91 RON,  compare the data for
91 versus 95 RON, and are not allowed to have more than a 3% difference between the
emissions and fuel economy results. These procedures were streamlined in 1997, as
detailed in the manufacturer guidance document "VPCD-97-01" available on the EPA
website (http://epa.gov/otaq/cert/dearmfr/vpcd9701.pdf) and still apply today.
Therefore, this provides some measure of confidence that premium, high-octane fuels
do not impact fuel economy results.  Additionally, under our Compliance Assurance
Program 2000 (CAP 2000), the manufacturers are required to  conduct In-Use
Verification  Program (IUVP) FTP, highway, and US06  testing on in-use vehicles at
10,000 and 50,000 miles as-received with the in-use fuel in the fuel tank (40 CFR
86.1845-04). Therefore, in the future, any indications of advantages resulting from
                                      156

-------
testing on high octane, premium fuels may be detected in the IUVP fuel economy data
results.
6.5   Hybrid-Electric Vehicle Battery State of Charge

What we proposed:

      EPA did not propose anything specific regarding the measurement of battery
state of charge (SOC) and fuel economy. However, the manufacturers raised a related
issue.

What commenters said:

      AAM/AIAM expressed concern with the criteria used at EPA's laboratory for
confirmatory testing of hybrid-electric vehicles, because it imposes a maximum absolute
state of charge (SOC) deviation equivalent to 1 % of the battery's rated capacity. This is
inconsistent with the criteria outlined in the Code of Federal Regulations (CFR) and in
California regulations, which include a maximum absolute SOC deviation criterion
equivalent to 1% of fuel energy consumed.  They also stated that the regulations do not
provide flexibility to correct fuel economy when the SOC criterion is exceeded, thus
leading to possibly unnecessary additional testing. They recommended that EPA
immediately align laboratory practice with the SOC criterion in the CFR, and that the
regulations include flexibility to either void tests under certain criteria or validate data by
applying SOC corrections approved by EPA to the fuel economy data.  They noted that
a modified version of the draft SAE J2572 procedure might be useful for determining
SOC corrections.

      Toyota commented that it did not support EPA's proposal to apply SOC criteria to
each bag.  Instead, they suggested  that EPA provide an option to correct fuel economy
results based on battery charge/discharge when the balance exceeds 1 percent.

Our response:

      EPA did not propose any regulatory changes to the measurement of battery state
of charge (SOC).  The comments regarding EPA laboratory practices are tangential to
this rulemaking effort.

      Thus far, all hybrid-electric vehicles tested at EPA's National Vehicle and Fuel
Emissions Laboratory (NVFEL) have passed compliance using the battery SOC limits of
+/-1 %  34 and the current practice of measuring the battery SOC over the entire
   34 Society of Automotive Engineers (SAE) Standard J1711 - "Recommended Practice for Measuring
   the Exhaust Emissions and Fuel Economy of Hybrid-Electric Vehicles," March 1, 1999 and SAE
   J2711 - "Recommended Practice for Measuring Fuel Economy and Emissions of Hybrid-Electric and
   Conventional Heavy-Duty Vehicles," September 1, 2002.
                                      157

-------
procedure for battery SOC neutrality/equilibrium.  In addition, if the SOC
neutrality/equilibrium exceeds the initial SOC by +1%, the maximum allowed SOC
deviation criterion, the manufacturers are afforded the opportunity for a re-test, at their
discretion.

      Regarding comments from the AAM/AIAM and Toyota regarding flexibility to
correct fuel economy results based on battery charge/discharge when the balance
exceeds 1 percent.  We would be willing to evaluate such a proposal to allow data
validation by applying battery SOC corrections to fuel economy results, in addition to the
existing provision for a discretionary re-test. We believe the following considerations
would be important: 1) statistical criteria (e.g., 95% confidence intervals, statistical f-
and t-test, etc.) must be defined and applied to the correction factors, 2) contiguous
data points for plotting the change in charge for the rechargeable energy storage
system (RESS) versus fuel consumption.  EPA would reserve the right to request and
review the underlying data points and other information used to develop the RESS
change in charge versus fuel consumption plot. Further, EPA has been and plans to
continue participating in the development of SAE practices, specifically SAE J257235 as
mentioned by the commenters, and any other committees related to hybrid-electric
vehicle practices.
   35 SAE Standard J2572 - "Recommended Practice for Measuring Fuel Consumption and Range of
   Fuel Cell and Hybrid Fuel Cell Vehicles Powered by Compressed Gaseous Hydrogen," August 1,
   2006.
                                       158

-------
Chapter?:    Cost Analysis

7.1    Testing Burden

      7.1.1     Testing Burden for Small Manufacturers

What we proposed:

      Testing Burden includes an estimate of the number of new tests required by the
rule, and a cost per test in terms of labor and operations and maintenance costs. The
added costs of new facilities to conduct the tests are considered separately as capital
costs, and one-time startup costs are considered separately as startup capital costs.

      The proposal did not include any specific provisions that were designed to lessen
the  compliance burden for small manufacturers.  We did propose and finalize many
features designed to provide adequate lead time and to minimize testing burdens that
apply to both large-volume and small-volume manufacturers.

What commenters said:

      Mitsubishi commented that EPA should recognize that there may be a
disproportionate testing burden for some manufacturers.  Ferrari asked that EPA
recognize the unique position of small-volume manufacturers, noting that they are
especially concerned regarding the potential number of new tests they will be required
to perform.

Our response:

      EPA has a number of provisions in its existing certification regulations that
reduce the burden of Clean Air Act compliance for small-volume manufacturers. For
example, special durability demonstration and in-use vehicle program testing
requirements apply under 40 CFR, Subpart S. Similarly, small manufacturers are
eligible to apply for reduced certification fees.  It is true, however,  that in general existing
emissions and fuel economy requirements are governed by environmental and
compliance considerations that do not vary as a function of manufacturer size,  including
city and highway emissions and fuel  economy testing for certification. This approach is
carried through in this rule. However, we are finalizing a number of provisions designed
to reduce test burdens that apply to large-volume and small-volume manufacturers.
During the Model Year 2008 to 2010 transition period, manufacturers have an
alternative to additional testing: they  may opt to use the mpg-based fuel economy
calculation. Second, in MY 2011 and beyond,  only those vehicles falling below the mpg-
based tolerance bands must undergo full 5-cycle testing. Third, where applicable we
allow fuel economy data to be determined based on analytically derived fuel economy
(ADFE) calculations.
                                     159

-------
      7.1.2    2008 - 2010 Model Year Testing Costs: Testing Issues

What we proposed:

      The proposal set forth specific test procedure changes as follows:  1) collection
of the US06 exhaust sample into two bags as opposed to the current one-bag sample;
2) required heater/defroster operation during the Cold FTP testing as opposed to
current optional operation; 3) requiring Cold FTP testing for diesel vehicles, which is
currently not required; and 4) codifying requirements for four-phase, four-bag FTP
testing of gasoline-electric hybrid vehicles rather than allowing this practice under
special test procedures contained in 40 CFR 86.1840-01. As stated in the proposal, all
of these changes were to take effect beginning with the 2008 model year, which, for a
particular test group, could be as early as January 2, 2007.

What commenters said:

      AAM/AIAM and several manufacturers disagreed with the assessment in the
NPRM that test volume and test facility costs for the 2008 through 2010 time period will
be zero.  They suggest that this assumption did not take into account the cost
associated with performing the proposed Cold FTP on diesel vehicles or performing the
two-bag US06 tests on all applicable vehicles. They believe that major test cell
modifications or new facilities will be required to perform Cold FTP tests on diesel
powered vehicles, noting that from their perspective the requirement to perform Cold
FTP tests is particularly onerous as this test data is not required for the calculation of
fuel economy label values until the 2011 model year.  Concerns were also expressed
about the difficulty of compliance with the heater/defroster provisions during the
transition period, and with the four-bag FTP testing requirement both as a matter of
adequate lead time and as potentially requiring costly facility modifications, especially if
EPA requirements were inconsistent with California's. Lead time concerns were also
expressed  by NADA.

Our response:

      These issues are addressed in detail in Section 6 of the Response to Comments.
The cost implications of these issues and their resolutions are highlighted here.

      The final regulations  clarify that the cold FTP diesel testing requirement only
begins with model year 2011 (except for those vehicles for which manufacturers choose
to voluntarily use the 5-cycle method). Consequently, the final rule continues to
anticipate no test burden or facility costs during the transition period attributable to this
cost item.

      As discussed in detail in Section 6, EPA has offered alternative methods for
calculating two-bag US06 data. These alternatives are available during the transition
period for those manufacturers choosing to use the 5-cycle approach, and we feel this
largely addresses the concerns regarding lead time. EPA's evaluation indicates that the
                                      160

-------
hardware changes to perform two-bag collection would be minimal and that the three-
model year lead time is adequate to implement them. Information system
reprogramming as well as validation tests were already included as startup costs in the
proposal, and these have been retained.

      We are finalizing a provision to require the mandatory use of heater/defroster, but
have extended the effective date until MY 2011 (except for those vehicles for which
manufacturers choose to apply the 5-cycle method). We believe this addresses any
cost concerns regarding this provision.

      As discussed in detail in Section 6, for the model year 2008 four-phase FTP
testing for hybrids, we will allow consistency with California. We believe these
modifications of the proposal adequately address the lead-time and facility modification
cost issues, and consequently no cost has been added to the final cost analysis for this
item.
      7.1.3     2011 and Later Model Year Testing Costs: Inclusion of MDPVs

What we proposed:

      As discussed in Section 4 of the Response to Comments, EPA sought comment
on a voluntary labeling program  for vehicles above 8500 pounds gross vehicle weight
rating (GVWR), and how such a program might be implemented.

What commenters said:

      Environmental groups supported the proposal, and Public Citizen argued that
EPA has the authority to make the requirement mandatory. AAM/AIAM suggested that
labeling of these heavier vehicles will not produce information that is comparable to the
labels on light duty vehicles, and may in fact create more confusion and
misunderstanding  than additional value in the marketplace.

Our response:

      As discussed in Section 4, we are finalizing in this rule a fuel economy labeling
program for MDPVs beginning in model year 2011. Testing will be limited to the FTP
and HFET tests. This is  congruent with regulations finalized by NHTSA's expansion of
the CAFE program to include MDPVs beginning the same model year. The testing
required under the CAFE and fuel economy labeling programs are overlapping, but not
identical; in general, more vehicles must be tested for CAFE, and therefore we do not
expect that inclusion of MDPVs in the labeling program will require additional testing.

      MDPVs are already subject to FTP emissions standards, and may also conduct
the HFET due to California requirements. NHTSA assessed the additional cost burden
                                     161

-------
of testing MDPVs in its light-truck CAFE rule (see Final Regulatory Impact Analysis,
page VIM 9).36
7.2 Facilities Costs

      7.2.1 Facility Upgrades for Cold FTP Testing

What we proposed:

      The proposal did not consider any facility costs for cold diesel testing as distinct
from the capitalized cost of facility capacity to conduct additional Cold FTP tests for all
vehicles covered, including diesel vehicles. The calculation used as a baseline the
number of city/highway fuel economy tests conducted according to EPA's database for
model year 2004. No allowances were made for potential increases in the fleet size in
future years.

What commenters said:

      AAM/AIAM  along with Volkswagen, Nissan, and Daimler Chrysler, believe that
major test cell modifications or new facilities will be required to perform Cold FTP tests
on diesel powered vehicles, noting that from their perspective the requirement to
perform Cold FTP tests is particularly onerous as this test data is not required for the
calculation of fuel economy label values until the 2011 model year. As mentioned
above, they further note that the proposed two-bag US06 test would require significant
cost to change both sample system hardware and software. They do not have specific
estimates of the costs associated with these changes, but they note that the costs "are
likely to be substantial."

Our response:

      EPA agrees that there may be some facility upgrades needed for some
manufacturers conducting cold FTP testing on diesel vehicles for model year 2011 and
later. Based on cost estimates developed by EPA's testing facilities in Ann Arbor,
Michigan, EPA has added a capital cost item corresponding to $55,000 for each of ten
cold testing facilities to conduct diesel hydrocarbon testing, including FIDs (flame
ionization detectors) as well as heated sample probes, sample lines and sample filters.
The number of cold FTP tests to be performed has also been increased to reflect the
possibility of growth in the diesel fleet between the MY 2004 dataset that was the
36 Final Regulatory Impact Analysis: Corporate Average Fuel Economy And CAFE Reform For MY 2008-2011
Light Trucks. U.S. Department of Transportation, National Highway Traffic and Safety Administration, Office of
Regulatory Analysis and Evaluation, National Center for Statistics and Analysis, March, 2006. Available at
http://www.nhtsa.dot.gov/staticfiles/DOT/NHTSA/Rulemaking/Rules/ Associated%20Files/2006_FRIAPublic.pdf
                                       162

-------
baseline for cost calculations and MY 2011, although this number is speculative
(increase from 5 test groups in MY 2006 to 10 in MY 2011). There are permanent
upgrades that will be required on a continuing basis for cold testing facilities conducting
diesel testing, and therefore are treated as facility capital costs rather than one-time
startup capital costs.
7.3 Startup Burden

      7.3.1  Dual Information Systems for CAFE/Gas Guzzler and Label
               Calculations

What we proposed:

      The proposal contemplated no change in the existing system of translating test
vehicle fuel economy values (now to be calculated using the mpg-based or 5-cycle
formulas) into model type fuel economy label values. Under the existing system, fuel
economy label values for test vehicles are arrived at through several steps.  These
ratings are computed as the sales-weighted harmonic mean of the "base levels" within
each model type, which in turn are calculated as the sales-weighted harmonic mean of
the configurations and subconfiguration fuel economy values within each base level.
This procedure is intended to insure that the most representative fuel economy  values
are posted on new vehicles, which are sold by the manufacturer's model designation
rather than categories that correspond to the test groups that are used for generating
fuel economy data as a part of the certification testing process.  Under the proposal, this
same methodology would apply to the mpg-based fuel economy, or the full 5-cycle
values would similarly be plugged in at each level. Thus, the proposal applied the mpg-
based adjustments at the test level. These values would then feed into fuel economy
values for each vehicle configuration,  which ultimately feed into the label value.

What commenters said:

      AAM/AIAM pointed out that the proposal would require one calculating system for
the new label values and a separate system, preserving the existing system, for
calculation of CAFE fuel economy values. This is because current regulations apply
10% and 22% downward adjustment factors for label values at the model type rather
than test level,  whereas these adjustments are not applied to the CAFE calculation. If
label value adjustments are applied at the test level on up, as in the proposal, then the
model level fuel economy value can no longer be applied to CAFE by simply leaving off
the 10% and 22% adjustments. Accordingly, AAM/AIAM requested that the mpg-based
calculation be allowed at the model type level rather than the test level, so that the
existing information processing programming architecture could be preserved for both.
                                     163

-------
Our response:

      EPA agrees that there may be some convenience in applying the derived 5-cycle
equation at the model-type level for manufacturers who wish to use the same data
management system for reporting CAFE and label values.  This may be particularly true
for the early part of the transition period.  However, this approach is not available for the
vehicle-specific 5-cycle label calculations, and any manufacturers who use it during the
phase-in period  during the 2008-2010 model years will encounter an  information cost
not contemplated in the proposal - a dual calculation procedure will be needed.
Similarly, a dual calculating system will be needed for model years 2011 and after.

      The cost  analysis has been updated to account for this increased information
system burden.  Based on a projection of EPA's information development contract
costs, and an estimate of the portion of those costs attributable to the dual information
system possibility, we have increased the industry information startup costs
(unamortized) by $933,450.
      7.3.2 Additional Startup Costs

What we proposed:

      The proposal contained several provisions that had potential cost impacts having
to do with the timing of new requirements or possible future adjustments. The proposed
effective date for the methodology for calculating label values and changes to the
design and content of the label was model year 2008, which for some vehicles can start
as soon as January 2, 2007. Of the four proposed new labels required for model year
2008, two were positioned vertically ("portrait"). The  proposal stated that we would
publish the equations for the mpg-based  approach by January 1 of the calendar year
prior to the model year to which the equations would first apply (e.g., for model year
2010 fuel economy calculations, the equations would be made available before January
1,2009).

What commenters said:

      As discussed in detail in Section 2, commenters argued that the sticker label
requires prior arrangements with suppliers. In addition, the possibility was raised of two
separate rounds of label adjustments to also reflect new NHTSA safety label rating
requirements that apply to vehicles manufactured on or after September 1, 2007. The
industry argued for an effective date for label changes should be September 1, 2007, or
a year after the publication of the final rule. Similarly, several manufacturers noted that a
change from the current landscape label  orientation  would cause costly modifications,
particularly for those like GM, who have redesigned their label in the landscape
orientation to accommodate the new NHTSA label requirements. As for updating of the
mpg-based equations, AAM/AIAM describes concerns with lead time and with the
frequency of such updates: making the equations available on January 1 of the model
                                      164

-------
year prior to the model year for which they first apply provides insufficient lead time for
manufacturers, since the start of production for some vehicles of a model year can be
as early as January 1 of the prior calendar year.  Consequently, they advocate making
the new equations available no later than January 1 of the model year two years prior to
the model  year to which the equations would first apply.

Our response:

      EPA acknowledges the cost concerns about redesigning pre-existing price
stickers and recognizes the challenges posed by simultaneously incorporating two new
labels onto the price sticker - the NHTSA crash rating label and the fuel economy label.
We are therefore aligning the mandatory date for using the new fuel economy label with
that of NHTSA's new crash test rating  label. EPA believes that with these modifications
no significant  new costs will be incurred by industry.

      The final rule requires that all 2008 models manufactured on or after September
1, 2007 have  the new label format. Manufacturers will be allowed to use the new label
design as  soon as possible, and are encouraged to do so. The current label design
can be used until September  1, 2007 with certain additional information included. We
are also finalizing a horizontal label orientation. Thus, the cost concerns mentioned by
industry for redesigning the price sticker would not be realized.

      As discussed in Section 3.2, we are finalizing regulations that require  EPA to
issue guidance regarding revisions to the equations by no later than July 1 of the
calendar year prior to the earliest start of the model year that starts in the following
calendar year. In other words, for new equations to be applicable to  the 2010 model
year (which can begin as early as January 2, 2009), EPA must issue guidance prior to
July 1,2008.
                                      165

-------