Draft Joint Technical Support Document:

            Proposed Rulemaking for 2017-2025
            Light-Duty Vehicle Greenhouse Gas
            Emission Standards and Corporate
            Average Fuel Economy Standards
&EPA
United States
Environmental Protection
Agency
*****
NHTSA
   www.nhtsa.gov

-------
             Draft Joint Technical Support Document:

                Proposed Rulemaking for 2017-2025
                Light-Duty Vehicle Greenhouse Gas
                 Emission Standards and Corporate
                  Average Fuel Economy Standards
                         Office of Transportation and Air Quality
                         U.S. Environmental Protection Agency

                                  and

                       National Highway Traffic Safety Administration
                          U.S. Department of Transportation
&EPA
United States
Environmental Protection
Agency
*****
NHTSA
EPA-420-D-11-901
November 2011
                                  Avw.nhtsa.gov

-------
                                                                      Contents

EXECUTIVE SUMMARY	I
CHAPTER 1:    THE BASELINE AND REFERENCE VEHICLE FLEET	1-1
1.1   Why do the agencies establish a baseline and reference vehicle fleet?	1-1
1.2   The 2008 baseline vehicle fleet	1-2
  1.2.1    Why did the agencies choose 2008 as the baseline model year?	1-2
  1.2.2    On what data is the baseline vehicle fleet based?	1-3
1.3   The MY 2017-2025 Reference Fleet	1-10
  1.3.1    On what data is the reference vehicle fleet based?	1-11
  1.3.2    How do the agencies develop the reference vehicle fleet?	1-13
  1.3.3    How was the 2008 baseline data merged with the CSM data?	1-13
  1.3.4    How were the CSM forecasts normalized to the AEO forecasts?	1-16
  1.3.5    What are the sales volumes and characteristics of the reference fleet?	1-24
CHAPTER 2:    WHAT ARE THE ATTRIBUTE-BASED CURVES THE AGENCIES
ARE PROPOSING, AND HOW WERE THEY DEVELOPED?	2-1
2.1   Why are standards attribute-based and defined by a mathematical function?	2-1
2.2   What attribute are the agencies proposing to use, and why?	2-2
2.3   What mathematical functions have the agencies previously used, and why?	2-4
  2.3.1    NHTSA in MY 2008 and MY 2011 CAFE (constrained logistic)	2-4
  2.3.2    MYs 2012-2016 Light Duty GHG/CAFE (constrained/piecewise linear)	2-4
  2.3.3    How have the agencies changed the mathematical functions for the proposed
  MYs 2017-2025 standards, and why?	2-5
2.4   What are the agencies proposing for the MYs 2017-2025 curves?	2-6
  2.4.1    What concerns were the agencies looking to address that led them to change
  from the approach used for the MYs 2012-2016 curves?	2-6
  2.4.2    What methodologies and data did the agencies consider in developing the
  2017-2025 curves?	2-8
2.5   Once the agencies determined the appropriate slope for the sloped part, how did
the agencies determine the rest of the mathematical function?	2-46
  2.5.1    Cutpoints for PC curve	2-46
  2.5.2    Cutpoints for LT curve	2-47
  2.5.3    Once the agencies determined the complete mathematical function shape,
  how did the agencies adjust the curves to develop the proposed standards and
  regulatory alternatives?	2-50
CHAPTER 3:    TECHNOLOGIES CONSIDERED IN THE AGENCIES' ANALYSIS
               3-1

-------
                                                                        Contents

3.1   What Technologies did the agencies consider for the proposed 2017-2025
standards?	3-2
3.2   How did the agencies determine the costs of each of these technologies?	3-7
  3.2.1    Direct Costs	3-7
  3.2.2    Indirect Costs	3-11
  3.2.3    Cost reduction through manufacturer learning	3-18
  3.2.4    Costs Updated to 2009 Dollars	3-23
3.3   How did the agencies determine effectiveness of each of these technologies?	3-23
  3.3.1    Vehicle simulation modeling	3-24
  3.3.2    Lumped parameter Modeling	3-65
3.4   What cost and effectiveness estimates have the agencies used for each technology?
     3-72
  3.4.1    Engine technologies	3-72
  3.4.2    Transmission Technologies	3-97
  3.4.3    Vehicle electrification and hybrid electric vehicle technologies	3-107
  3.4.4    Hardware costs for charging grid-connected vehicles	3-181
  3.4.5    Other Technologies Assessed that Reduce CO2 and Improve Fuel Economy
           3-185
3.5   How did the agencies consider real-world limits when defining the rate at which
technologies can be deployed?	3-213
  3.5.1    Refresh and redesign schedules	3-213
  3.5.2    Vehicle phase-in caps	3-216
3.6   How are the technologies applied in the agencies' respective models?	3-223
CHAPTER 4:   ECONOMIC AND OTHER ASSUMPTIONS USED IN THE
AGENCIES'ANALYSIS	4-2
4.1   How the Agencies use the economic and other assumptions in their analyses	4-2
4.2   What assumptions do the agencies use in the impact analyses?	4-3
  4.2.1    The on-road fuel economy "gap"	4-3
  4.2.2    Fuel prices and the value of saving fuel	4-8
  4.2.3    Vehicle Lifetimes and Survival Rates	4-9
  4.2.4    VMT	4-12
  4.2.5    Accounting for the fuel economy rebound effect	4-19
  4.2.6    Benefits from increased vehicle use	4-27
  4.2.7    Added costs from increased vehicle use	4-28

-------
                                                                       Contents

  4.2.8   Petroleum and energy security impacts	4-29
  4.2.9   Air pollutant emissions	4-38
  4.2.10  Reductions in emissions of greenhouse gases	4-50
  4.2.11  The Benefits due to reduced refueling time	4-54
  4.2.12  Discounting future benefits and costs	4-63
CHAPTER 5:   AIR CONDITIONING, OFF-CYCLE CREDITS, AND OTHER
FLEXIBILITIES	5-2
5.1   Air conditioning technologies and credits	5-2
  5.1.1   Overview	5-2
  5.1.2   Air Conditioner Leakage	5-4
  5.1.3   COi Emissions and Fuel Consumption due to Air Conditioners	5-24
  5.1.4   Air Conditioner System Costs	5-51
5.2   Off-Cycle Technologies and Credits	5-55
  5.2.1   Reducing or Offsetting Electrical Loads	5-57
  5.2.2   Active Aerodynamic Improvements	5-62
  5.2.3   Advanced Load Reductions	5-65
  5.2.4   Summary of Proposed Credits	5-74
5.3   Pick-up Truck Credits	5-74
  5.3.1   Pick-up Truck Definition	5-76
  5.3.2   Hybrid Pickup Technology	5-77
  5.3.3   Mild and Strong HEV Pickup Truck Definitions	5-78
  5.3.4   Performance Based Pickup Truck Incentive Credit Thresholds	5-81

-------
                                                                Executive Summary
Executive Summary

       The Environmental Protection Agency (EPA) and the National Highway Traffic
Safety Administration (NHTSA) are issuing a joint proposal to establish new standards for
light-duty highway vehicles that will reduce greenhouse gas emissions and improve fuel
economy. The joint proposed rulemaking is consistent with the Presidential Memorandum
issued by President Obama on May 21, 2010, requesting that NHTSA and EPA develop
through notice and comment rulemaking a coordinated National Program to reduce
greenhouse gas emissions and improve the fuel economy of light-duty vehicles for model
years  2017-2025.  This proposal, consistent with the President's request, responds to the
country's critical need to address global climate change and to reduce oil consumption. .
EPA is proposing greenhouse gas emissions standards under the Clean Air Act, and NHTSA
is proposing Corporate Average Fuel Economy standards under the Energy Policy and
Conservation Act, as amended.  These standards  apply to passenger cars, light-duty trucks,
and medium-duty passenger vehicles, covering model years 2017 through 2025. They require
these vehicles to meet an estimated combined average emissions level of 163 grams of COi
per mile in MY 2025 under EPA's GHG program, and 49.6 mpg in MY 2025 under
NHTSA's CAFE program and represent a harmonized and consistent national program
(National Program). These standards are designed such that compliance can be achieved with
a single national vehicle fleet whose emissions and fuel economy performance improves each
year from MY2017 to 2025.  This document describes the supporting technical analysis for
areas of these jointly proposed rules which are consistent between the two agencies.

       NHTSA and EPA have coordinated closely to create a nationwide joint fuel economy
and GHG program based on consistent compliance structures and technical assumptions.  To
the extent permitted under each  Agency's statutes, NHTSA and EPA have incorporated the
same compliance flexibilities, such as averaging, banking, and trading of credits, off-cycle
credits, and the same testing protocol for determining the agencies' respective fleet-wide
average proposed standards.  In  addition, the agencies have worked together to create a
common baseline fleet and to harmonize most of the costs and benefit inputs used in the
agencies' respective modeling processes for this joint proposed  rule.

              Chapter 1 of this Draft TSD provides an explanation of the agencies'
methodology used to develop the baseline and reference case vehicle fleets, including the
technology composition of these fleets, and how the agencies projected vehicle sales into the
future. One of the fundamental  features of this technical analysis is the development of these
fleets, which are used by both agencies in their respective models. In order to determine
technology costs  associated with this joint rulemaking, it is necessary to consider the vehicle
fleet absent a rulemaking as a "business as usual" comparison. In past CAFE rulemakings,
NHTSA has used confidential product plans submitted by vehicle manufacturers to develop
the reference case fleet. In responding to comments from these previous rulemakings that the
agencies make these fleets available for public review, the agencies created a new
methodology for  creating baseline and reference fleets using data, the vast majority of which
is publicly available.

-------
                                                                 Executive Summary
              Chapter 2 of this document discusses how NHTSA and EPA developed the
mathematical functions which provide the bases for the proposed car and truck standards.
NHTSA and EPA worked together closely to develop regulatory approaches that are
fundamentally the same, and have chosen to use an attribute-based program structure based
on the footprint attribute, similar to the mathematical functions used in the MYs 2012-2016
rule. The agencies revisited other attributes as candidates for the standard functions, but
concluded that footprint remains the best option for balancing the numerous technical and
social factors.  However, the agencies did adjust the shape of the truck footprint curve, in
comparison to the MYs 2012-2016 rule. The agencies also modified the way the car and
truck curves change from year to year compared to the MYs 2012-2016 rule.  In determining
the shape of the footprint curve, the agencies considered factors such as the magnitudes of
COi reduction and fuel savings, how much that shape may incentivize manufacturers to
comply in a manner which circumvents the overall goals  of the joint program, whether the
standards' stringencies are technically attainable, the utility of vehicles, and the mathematical
flexibilities inherent to the statistical  fitting of such a function.

       Chapter 3 contains a detailed  analysis of NHTSA and EPA's technology assumptions
on which the proposed regulations were based. Because the majority of technologies that
reduce GHG emissions and improve fuel economy are identical, it was crucial that NHTSA
and EPA use common assumptions for values pertaining  to technology availability, cost, and
effectiveness.  The agencies collaborated closely in determining which technologies would be
considered in the rulemaking, how much these technologies would cost the manufacturers
(directly) in the time frame of the proposed rules, how these costs will be adjusted for learning
as well as for indirect cost multipliers, and how effective the technologies are at
accomplishing the goals of improving fuel efficiency and GHG  emissions.

       Chapter 4 of this document provides a full description and analysis of the economic
factors considered in this joint proposal. EPA and NHTSA harmonized many inputs
capturing economic and social factors, such as the discount rates, fuel prices, social costs of
carbon, the magnitude of the rebound effect, the value of refueling time, and the social cost of
importing oil and fuel.

       Chapter 5 of this draft TSD discusses proposed adjustments and credits to reflect
technologies that improve air conditioner efficiency, that improve efficiency under other off-
cycle driving conditions, and that reduce leakage of air conditioner refrigerants that contribute
to global warming. The air conditioner credits are similar to the MYs 2012-2016 rule, with
two notable exceptions: NHTSA is proposing to allow A/C efficiency improvements to help
come into compliance with fuel economy standards, and a new air conditioner test procedure
is introduced to help capture efficiency credits. NHTSA  is now also allowing off-cycle
improvements to help  manufacturers  come into compliance with fuel economy standards. A
list of some technologies and their credits and a streamlined methodology is provided by the
agencies to help simplify the credit generating process. Chapter 5 also discusses proposed
adjustments to encourage "game changing" technologies  (such as hybridized powertrains) for
full-size pickup trucks.
                                             11

-------
                  Executive Summary
111

-------
                                              The Baseline and Reference Vehicle Fleet
Chapter 1:   The Baseline and Reference Vehicle Fleet

       The passenger cars and light tracks sold currently in the United States, and those
which are anticipated to be sold in the MYs 2017-2025 timeframe, are highly varied and
satisfy a wide range of consumer needs. From two-seater miniature cars to 11-seater
passenger vans to large extended cab pickup trucks, American consumers have a great
number of vehicle options to accommodate their needs and preferences. Recent volatility in
oil prices and the state of the economy have demonstrated that consumer demand and choice
of vehicles within this wide range can be sensitive to these factors. Although it is impossible
to precisely predict the future, the agencies need to characterize and quantify the future fleet
in order to assess the impacts of rules that would  affect that future fleet. The agencies have
examined various publicly-available sources, and then used inputs from those sources in a
series of models to project the composition of a baseline and reference fleet for purposes of
this analysis. This chapter describes this process.

       The agencies have made every effort to make this analysis transparent and duplicablea.
Because both the input and output sheets from our modeling are public, stakeholders can
verify and check NHTSA's and EPA's modeling, and perform their own analyses with these
datasets.
1.1 Why do the agencies establish a baseline and reference vehicle fleet?

    In order to calculate the impacts of the proposed future GHG and CAFE standards, it is
necessary to estimate the composition of the future vehicle fleet absent those proposed
standards in order to conduct comparisons. EPA and NHTSA have developed a comparison
fleet in two parts. The first step was to develop a baseline fleet based on model year 2008
data, discussed further below.  NHTSA and EPA create a baseline fleet in order to track the
volumes and types of fuel economy-improving and CCVreducing technologies which are
already present in the existing vehicle fleet. Creating a baseline fleet helps to keep, to some
extent, the agencies' models from adding technologies to vehicles that already have these
technologies, which would result in "double counting" of technologies' costs and benefits.
The second step was to project the baseline fleet sales into MYs 2017-2025. This is called the
reference fleet, and it represents the fleet volumes (but, until later steps, not levels of
technology) that the agencies believe would exist in MYs 2017-2025 absent any change due
to regulation in 2017-2025.
a In endeavoring to be transparent and duplicable in every aspect of the analysis supporting the joint proposed
rules discussed in this TSD, the agencies seek to facilitate public participation in the rulemaking process
consistent with Executive Order 13563 (76 Fed. Reg. 3821, Jan. 21, 2011) and OMB Circular A-4 (September
17, 2003, available at http://www.whitehouse.gov/omb/circulars_a004_a-4/ (last accessed Aug. 15, 2011)).

                                              1-1

-------
                                             The Baseline and Reference Vehicle Fleet
    After determining the reference fleet, a third step is needed to account for technologies
(and corresponding increases in cost and reductions in fuel consumption and COi emissions)
that could be added to MY 2008-technology vehicles in the future, taking into account
previously-promulgated standards, and assuming MY 2016 standards are extended through
MY2025. This step uses the Omega and Volpe models to add technologies to that MY 2008-
based market forecast such that each manufacturer's car and truck CAFE and average COi
levels reflect baseline standards.  The models' output, the reference case, is the light-duty fleet
estimated to exist in MYs 2017-2025 without new GHG/CAFE standards covering MYs
2017-2025. All of the agencies'  estimates of emission reductions/fuel economy
improvements, costs,  and societal impacts for purposes of this NPRM are developed in
relation to the agencies' reference cases. This chapter describes the first two steps of the
development of the baseline and reference fleets. The third step of technology addition is
developed separately  by each agency as the outputs of the OMEGA and Volpe models (see
Chapter 3 of the TSD for an explanation of how the models apply technologies to vehicles in
order to evaluate potential paths to compliance).

1.2 The  2008 baseline vehicle fleet

       1.2.1  Why did the agencies choose 2008 as the baseline model year?

       The baseline that EPA developed in consultation with NHTSA for the 2012-2016 final
rule was comprised of model year 2008 CAFE compliance data (specifically, individual
vehicles with sales volumes disaggregated at the level of specific engine/transmission
combinations) submitted by manufacturers to EPA, in part because full MY 2009 data was not
available at the time.  For this NPRM, the agencies chose again to use MY 2008 vehicle data
as the basis of the baseline fleet,  but for different reasons than in the 2012-2016 final rule.
First, MY 2008 is now the most recent model year for which the industry had what the
agencies  would consider to be "normal" sales. Complete MY 2009 data is now available for
the industry, but the agencies believe that the model year was disrupted  by the economic
downturn and the bankruptcies of both General Motors and Chrysler. CAFE compliance data
shows that there was a significant reduction in the number of vehicles sold by both companies
and by the industry as a whole.  These abnormalities led the agencies to conclude that MY
2009 data was likely not representative for projecting the future fleet for purposes of this
analysis.  And second, while MY 2010 data is likely more representative for projecting the
future fleet, it was not complete and available in time for it to be used for the NPRM analysis.
Therefore, for purposes of the NPRM analysis, the agencies chose to use MY 2008 CAFE
compliance data for the baseline since it was  the latest, most representative transparent data
set that we had available.  However, the agencies plan to use the MY 2010 data, if available,
to develop an updated market forecast for use in the final rule.  To the extent the MY 2010
data becomes available within the time frame of the comment period for this proposal the
agencies  will place a copy of this data into each  agencies docket.
                                            1-2

-------
                                              The Baseline and Reference Vehicle Fleet
       1.2.2   On what data is the baseline vehicle fleet based?

       As part of the CAFE program, EPA measures vehicle COi emissions and converts
them to mpg, and generates and maintains the federal fuel economy database.  See 49 U.S.C
32904 and 40 CFR Part 600. Most of the information about the vehicles that make up the
2008 fleet was gathered from EPA's emission certification and fuel economy database, most
of which is available to the public. These data included, by individual vehicle  model
produced in MY 2008, vehicle production volume, fuel economy rating for CAFE
certification (i.e., on the 2-cycle city-highway test), carbon dioxide emissions (equivalent to
fuel economy rating for CAFE certification), fuel type (gasoline, diesel, and/or alternative
fuel), number of engine cylinders, displacement, valves per cylinder, engine cycle,
transmission type, drive (rear-wheel, all-wheel, etc.), hybrid type  (if applicable), and
aspiration (naturally-aspirated, turbocharged, etc.).  In addition to this information about each
vehicle model produced in MY 2008, the agencies also need information about the fuel
economy-improving/CCVreducing technologies already on those vehicle models in order to
assess how much and which technologies to apply to determine a path toward future
compliance. However, EPA's certification database does  not include a detailed description of
the types of fuel economy-improving/CCh-reducing technologies  considered in this  NPRM
because this level of information was not reported in MY  2008 for emission certification or
fuel economy testing.  Thus, the agencies augmented this description with publicly-available
data which includes more complete technology descriptions from Ward's Automotive
Group.b'c The agencies also need information about the footprints of MY 2008 vehicles in
order to create potential target curves (as discussed in Chapter 2 of the TSD, vehicles are
plotted as data points defined by (footprint, fuel economy) or (footprint, CO2 emissions). In a
few instances when relevant vehicle information (such as, for example, vehicle footprint) was
not available from these two sources, the agencies obtained this information principally from
publicly-accessible internet sites such as Motortrend.com  or Edmunds.com, and occasionally
from other sources (such as articles about specific vehicles revealed from internet search
engine research).d'e'f

       The baseline vehicle fleet for the analysis informing these proposed rules is highly
similar to the baseline vehicle fleet used in the MYs 2012-2016 rulemaking, and like that
baseline, is comprised of publicly-available data to the largest extent possible.  Whereas some
of the technology data included in the MYs 2012-2016  analysis' baseline fleet  was based on
confidential product plan information about MY 2008 vehicles, specifically, data about which
vehicles already have low  friction lubricants, electric power steering, improved accessories,
b WardsAuto.com: Used as a source for engine specifications shown in Table 1-1.
c Note that WardsAuto.com, where this information was obtained, is a fee-based service, but all information is
public to subscribers.
d Motortrend.com and Edmunds.com: Used as a source for footprint and vehicle weight data.
e Motortrend.com and Edmunds.com are free, no-fee internet sites.
f A small amount of footprint data from manufacturers' MY 2008 product plans submitted to the agencies was
used in the development of the baseline.

                                              1-3

-------
                                             The Baseline and Reference Vehicle Fleet
and low rolling resistance tires applied, the agencies no longer consider that information as
needing to be withheld, because by now all MY 2008 vehicle models are already in the on-
road fleet. As a result, the agencies are able to make public the exact baseline used in this
rulemaking analysis.

       As explained in the MYs 2012-2016 TSD, creating the 2008 baseline fleet Excel file
was an extremely labor-intensive process. EPA in consultation with NHTSA first considered
using EPA's CAFE certification data, which contains most of the required information.
However, since the deadline for manufacturers to report this data did not allow enough time,
in the MYs 2012-2016 rulemaking, for early modeling review, the agencies began to create
the baseline fleet file using an alternative data source.

       The agencies ultimately relied on a combination of EPA's vehicle emissions
certification data, data from a paid subscription to Ward's Automotive Group, and  CAFE
certification data. EPA's vehicle emissions certification data contains much of the
information required for creating a baseline fleet file, but it lacked the production volumes
that are necessary for the OMEGA and Volpe models, and also contains some vehicle models
that manufacturers certified but did not produce in MY 2008. The data from Ward's
contained production volumes (which were not ultimately used, because they did not have
volumes for individual vehicles down to the resolution of the specific engine and transmission
level) and vehicle specifications, and eliminated extraneous vehicles.

       The EPA vehicle emissions certification dataset came in two parts, an engine file and a
vehicle file, which the agencies combined into one spreadsheet using their common index.
The more-specific Ward's data also came in two parts, an engine file and a vehicle file, and
also required mapping, which was more difficult than combining the EPA vehicle emissions
certification dataset files because there was no common index between the Ward's files. A
new index was implanted in the engine file and a  search equation in the vehicle file, which
identified most of the vehicle and engine combinations.  Each vehicle and engine combination
was reviewed and corrections were made manually when the search routine failed to give the
correct engine and vehicle combination. The combined Ward's data was then mapped to the
EPA vehicle emissions certification data by creating a new index in the combined Ward's
data and using the same process that was used to combine the Ward's engine and vehicle files.

       In the next step, CAFE certification data had to be merged in order to fill out the
needed production volumes.  NHTSA and EPA reviewed the CAFE certification data for MY
2008 as it became available in the MYs 2012-2016 rulemaking.  The CAFE certification set
could have been used with the Ward's data without the EPA vehicle emission certification
data set, but was  instead appended to the combined Ward's and EPA vehicle emission
certification dataset. That combined dataset was then mapped into the CAFE dataset using
the same Excel mapping technique described above. Finally EPA and NHTSA obtained the
remaining attribute and technology data, such as footprint, curb weight, and others (for a
complete list of data with sources see Table 1-1 below) from other sources, thus completing
the baseline dataset.
                                            1-4

-------
                                              The Baseline and Reference Vehicle Fleet
       We note that, besides the use of updated AEO and CSM information, the baseline fleet
for this rulemaking is different from the fleet used in the MYs 2012-2016 rulemaking in one
fairly minor way. Specifically, in the MYs 2012-2016 the agencies aggregated full-size
pickup data in the baseline by using average values to represent all variants of a given pickup
line. While full-size pickups might be offered with various combinations of cab style (e.g.,
regular, extended, crew) and box length (e.g., 5 ¥2', 6 Vi\ 8'), and therefore multiple footprint
sizes, CAFE compliance data for MY 2008 did not contain footprint information, and
therefore could not reliably be used to identify which pickup entries correspond to footprint
values estimable from public or commercial sources. Therefore, the agencies used the known
production levels of average values to represent all variants of a given pickup line (e.g., all
variants of the F-150, or all variants of the Sierra/Silverado) in order to calculate the sales-
weighted average footprint/fuel  economy value for each pickup family.  In retrospect, this
may have affected how we fit the light truck target curve, among other things, so the agencies
have since created an expanded  version of the fleet to account for the variation in
footprint/wheelbase for the large pickups of Chrysler, Ford, GM, Nissan and Toyota. In MY
2008, large pickups were available from Nissan with 2, Chrysler and Toyota with 3, and Ford
and GM with 5 wheelbase/footprint combinations.  The agencies got this footprint data from
MY 2008 product plans submitted by the various manufacturers, which  can be made public at
this time because by now all MY 2008 vehicle models are  already in production, which makes
footprint data about them essentially public information.

       The agencies created the expanded fleet by replicating original records from a single
pickup footprint model into multiple pickup models with distinct footprint values, in order to
reflect the additional pickup model footprints just noted. For example, an F-150 in the MY
2008 baseline used in the MYs 2012-2016 rulemaking analysis with a footprint value of 67
square feet, is disaggregated by  replicating 2 times in all respects, except with footprint values
of 58, 67, and 73 square feet.  Sales volumes of these pickups from the original record were
distributed to each of the "58 square feet" and "73 square feet" duplicates based on the
distribution of MY 2008 sales by these pickups' wheelbase/footprint, which the agencies took
from product plan data submitted by the manufacturers in 2008/2009 in response to requests
to support the MYs 2012-2016 rulemaking analysis.  The agencies were able to distribute the
sales for each of the original pickups by wheelbase/footprint by matching each of the pickups
in the baseline fleet with pickups in the product plans on the basis of drive type, transmission
type, and engine displacement, cylinders/configuration and HP, and then sorting and summing
the sales of the matched pickups in the product plans by wheelbase/footprint.

       Both agencies used this fleet forecast to populate input files for the agencies'
respective modeling systems.  The structure of the market forecast input file used for DOT's
CAFE Compliance and Effects Modeling System (a.k.a. "the Volpe model") is described the
model documentation.  To help readers who wish to directly examine the baseline fleet file
for EPA's OMEGA model, and  to provide some idea of its contents for  those readers who do
not, Table 1-1 shows the columns of the complete fleet file, which includes the MY 2008
baseline data that was compiled. Each column has its name,  definition (description) and
source.  Most elements shown in Table 1-1 also appear in the market forecast input file for
                                             1-5

-------
                                             The Baseline and Reference Vehicle Fleet
DOT's modeling system, which also accommodates some additional data elements discussed
in the model documentation.

                          Table 1-1 Data, Definitions, and Sources
Data Item
Index
Manufacturer
CERT
Manufacturer Name
Name Plate
Model
Reg Class
Our Class
CSM Class
Vehicle Type
Number
Vehicle Index From
Sum Page
Traditional
Car/Truck
NHTSA Defined
New Car/Truck
Total Production
Volume
Fuel Econ.
(mpg)
C02
Area (sf)
Fuel
Fuel Type
Disp
(lit.)
Effective Cyl
Actual Cylinders
Valves Per Cylinder
Definition
Index Used to link EPA and NHTSA baselines
Commnon name of company that
manufactured vehicle. May include more
name plates than Cert Manufacturer Name.
Certification name of company that
manufactured vehicle
Name of Division
Name of Vehicle
EPA Fuel Economy Class Name
If a car's Footprint<43 then "SubCmpctAuto"
If a car's 43<=Footprint<46 then
"CompactAuto"
If a car's 46<=Footprint<53 then
"MidSizeAuto"
If a car's Footprint >=53 then "LargeAuto"
If a S.U.V.'s Footprint < 43 then "SmallSuv"
If a S.U.V.'s 43<=Footprint<46 then
"MidSizeSuv"
If a S.U.V's Footprint >=46 then "LargeSuv"
If a Truck's Footprint < 50 then "SmallPickup"
If a Truck's Footprint>=50 then "LargPickup"
If a Van's Structure is Ladder then
"CargoVan"
If a Van's Structure is Unibody then
"Minivan"
CSM Worldwide' s class for the vehicle. Used
to weight vehicles based on CSM data.
Vehicle Type Number assigned to a vehicle
based on its number of cylinders, valves per
cylinder, and valve actuation technology
Number to be used as a cross reference with
the Sum Pages.
Traditional Car Truck value for reference.
New NHTSA Car Truck value as defined in
201 1 Fuel economy regulations. Used in
calculations.
Total number of vehicles produced for that
model.
EPA Unadjusted Fuel Economy
CO2 calculated from MPG. CO2 weighted
1.15 times higher for diesel vehicles.
Average Track x Wheelbase
Gas or Diesel
Gas or Diesel or Electric
Engine Cylinder Displacement Size in Liters
Number of Cylinder + 2 if the engine has a
turbo or super charger.
Actual Number of Engine Cylinders
Number of Valves Per Actual Cylinder
Where The Data is From
Created
Certification data
Certification data
Certification data
Certification data
Certification data
Derived From Certification data and Footprint
CSM Worldwide
Defined by EPA staff
NA
Certification data
NHTSA
Certification data
Certification data
Certification data
Calculated from track width and wheel base
Wards
Certification data
Wards/Certification data
Derived From Certification data.
Certification data
Certification data
                                            1-6

-------
 The Baseline and Reference Vehicle Fleet
Valve Type
Valve Actuation
VVT
VVLT
Deac
Fuel injection
system
Boost
Engine Cycle
Horsepower
Torque
Trans Type
Trans
Num of Gears
Transmission
Structure
Drive
Drive with AWD
Wheelbase
Track Width
(front)
Track Width
(rear)
Footprint: PU
Average
Theshold Footprint
Curb
Weight
GVWR
Stop-
Start/Hybrid/Full
EV
Import Car
Towing Capacity
(Maximum)
Engine Oil
Type of valve actuation.
Type of valve actuation with values compatible
with the package file.
Type of valve timing with values compatible
with the package file.
Type of valve lift with values compatible with
the package file.
Cylinder Deactivation with a value that is
compatible with the package file.
Type of fuel injection.
Type of Boost if any.
As Defined by EPA Cert. Definition
Max. Horsepower of the Engine
Max. Torque of the Engine
A=Auto AMT=Automated Manual M=Manual
CVT= Continuously Variable Transmission
Type Code with number of Gears
Number of Gears
Transmisison definition. Matches the cost
definition.
Ladder or Unibody
Fwd, Rwd, 4wd
Fwd, Rwd, Awd, 4wd
Length of Wheelbase
Length of Track Width in inches
Length of Track Width in inches
Car and Large Truck Footprints are normal
(Average Track x Wheelbase). Medium and
Small Truck footprints are the production
weighted average for each vehicle.
Footprint valve that will be set to 41 for values
less than 41, Will be set to 56 for car values >
56, and will be set to 74 for truck values >74
Curb Weight of the Vehicle
Gross Vehicle Weight Rating of the Vehicle
Type of Electrification if any. Blank = None
Cars Imported
Weight a vehicle is rated to tow.
Ratio between the applied shear stress and the
Wards (Note:Type E is from Cert Data)
Wards
Wards
Wards
Wards
Wards
Wards
Wards
Wards
Wards
Certification data
Certification data
Certification data
Certification data
General Internet Searches
Certification data
Certification data
Some from Edmunds.com or Motortrend.com,
Others from product plans with a subset verified
with Edmunds.com or Motortrend.com for
accuracy.
Some from Edmunds.com or Motortrend.com,
Others from product plans with a subset verified
with Edmunds.com or Motortrend.com for
accuracy.
Some from Edmunds.com or Motortrend.com,
Others from product plans with a subset verified
with Edmunds.com or Motortrend.com for
accuracy.
Derived from data from Edmunds.com or
Motortrend.com. Production volumes or specific
footprints from product plans.
Derived from data from Edmunds.com or
Motortrend.com. Production volumes or specific
footprints from product plans.
Some from Edmunds.com or Motortrend.com,
Others from product plans with a subset verified
with Edmunds.com or Motortrend.com for
accuracy.
Some from Edmunds.com or Motortrend.com,
Others from product plans with a subset verified
with Edmunds.com or Motortrend.com for
accuracy.
Certification data
Certification data
Volpe Input File
Volpe Input File
1-7

-------
                                                        The Baseline and Reference Vehicle Fleet
Viscosity
Volume 2009
Volume 2010
Volume 20 11
Volume 20 12
Volume 2013
Volume 2014
Volume 2015
Volume 20 16
Volume 2017
Volume 20 18
Volume 2019
Volume 2020
Volume 2021
Volume 2022
Volume 2023
Volume 2024
Volume 2025
Low drag brakes
Electric Power
steering
Volpe Index
rate of shear, which measures the resistance of
flow of the engine oil (as per S AE Glossary of
Automotive Terms)
Projected Production Volume for 2009
Projected Production Volume for 2010
Projected Production Volume for 2011
Projected Production Volume for 2012
Projected Production Volume for 2013
Projected Production Volume for 2014
Projected Production Volume for 2015
Projected Production Volume for 2016
Projected Production Volume for 2017
Projected Production Volume for 2018
Projected Production Volume for 2019
Projected Production Volume for 2020
Projected Production Volume for 2021
Projected Production Volume for 2022
Projected Production Volume for 2023
Projected Production Volume for 2024
Projected Production Volume for 2025
See Volpe Documentation
See Volpe Documentation
Number used to reorder the vehicles in the
EPA baseline in the same order as the Volpe
input file.

Calculated based on 2008 volume and Annual
Energy Outlook and CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Calculated based on 2008 volume and AEO and
CSM adjustment factors.
Volpe Input File
Volpe Input File
Volpe Input File
Notes:
1.  For engines not available in the WardsAuto.com
engine file, an internet search was done to find this
information.
2.  These data were obtained from manufacturer's product
plans. They were used to block (where possible) the
model from adding technology that was already on a
vehicle.
3. Ward's Automotive Group data obtained from "2008
Light Vehicle Engines."
                                                       1-8

-------
                                              The Baseline and Reference Vehicle Fleet
       The sales volumes for the MY 2008 baseline fleet are included in the section below on
reference fleet under the MY 2008 columns.  Table 1-2 displays the engine technologies
present in the baseline fleet.  Again, the engine technologies for the vehicles manufactured by
these manufacturers in MY 2008 were largely obtained from Ward's Auto online.

                        Table 1-2 2008 Engine Technology Percentages
Manufacturer
All
All
All
Aston Martin
Aston Martin
BMW
BMW
Chrysler/Fiat
Chrysler/Fiat
Daimler
Daimler
Ferrari
Ferrari
Ford
Ford
Geely/Volvo
Geely/Volvo
GM
GM
Honda
Honda
Hyundai
Hyundai
Kia
Kia
Lotus
Lotus
Mazda
Mazda
Mitsubishi
Mitsubishi
>
H
0)
"o
3
0)
Both
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Trucks
Cars
Trucks
Cars
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
•a
a)
I
U
o
_a
VH
3
3%
4%
1%
0%
0%
33%
5%
1%
0%
2%
16%
0%
0%
0%
0%
0%
49%
0%
1%
0%
4%
0%
0%
0%
0%
0%
0%
11%
24%
6%
0%
Super Charged
0%
0%
0%
0%
0%
1%
0%
0%
0%
0%
1%
0%
0%
1%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
77%
0%
0%
0%
0%
0%
Single Overhead Cam
20%
17%
24%
0%
0%
14%
0%
21%
39%
55%
36%
0%
0%
15%
65%
0%
0%
0%
0%
57%
64%
0%
0%
0%
0%
0%
0%
0%
1%
100%
100%
Dual Overhead Cam
63%
73%
48%
100%
0%
86%
100%
72%
4%
45%
64%
100%
0%
85%
32%
100%
100%
31%
56%
43%
36%
100%
100%
100%
100%
100%
0%
99%
99%
0%
0%
Overhead Cam
17%
9%
29%
0%
0%
0%
0%
8%
57%
0%
0%
0%
0%
0%
3%
0%
0%
69%
44%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Variable Valve Timing
Continuous
8%
9%
6%
0%
0%
14%
0%
0%
0%
72%
35%
0%
0%
4%
28%
0%
0%
5%
29%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
38%
Variable Valve Timing
Discrete
22%
24%
19%
100%
0%
86%
100%
42%
4%
4%
17%
100%
0%
0%
1%
100%
100%
17%
31%
27%
4%
0%
0%
0%
0%
100%
0%
7%
13%
0%
0%
Variable Valve Timing
Intake Only
30%
35%
23%
0%
0%
0%
0%
0%
0%
13%
47%
0%
0%
47%
9%
0%
0%
14%
1%
20%
28%
100%
100%
10%
17%
0%
0%
92%
87%
0%
0%
Variable Valve Lift and
Timing Continuous
0%
0%
0%
24%
0%
0%
0%
0%
0%
0%
0%
29%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Variable Valve Lift and
Timing Discrete
12%
13%
10%
0%
0%
13%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
100%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Cylinder Deactivation
6%
3%
11%
0%
0%
0%
0%
5%
4%
0%
0%
0%
0%
0%
0%
0%
0%
40%
4%
11%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Direct Injection
5%
7%
3%
0%
0%
33%
6%
0%
0%
2%
16%
0%
0%
0%
0%
0%
0%
0%
6%
0%
4%
0%
0%
0%
0%
0%
0%
11%
24%
0%
0%
                                             1-9

-------
                                              The Baseline and Reference Vehicle Fleet
Nissan
Nissan
Porsche
Porsche
Spyker/Saab
Spyker/Saab
Subaru
Subaru
Suzuki
Suzuki
Tata/JLR
Tata/JLR
Tesla
Tesla
Toyota
Toyota
Volkswagen
Volkswagen
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
0%
0%
17%
12%
100%
0%
15%
3%
0%
0%
0%
0%
0%
0%
0%
0%
43%
1%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
20%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
69%
70%
0%
0%
0%
0%
0%
0%
0%
0%
85%
0%
100%
100%
100%
100%
100%
62%
31%
30%
100%
100%
100%
100%
0%
0%
100%
100%
15%
100%
0%
0%
0%
0%
0%
38%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
4%
0%
100%
100%
17%
0%
0%
23%
0%
0%
76%
0%
0%
0%
29%
61%
48%
99%
96%
100%
0%
0%
0%
62%
31%
7%
0%
0%
24%
100%
0%
0%
71%
39%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
0%
0%
1%
27%
0%
0%
0%
0%
0%
0%
0%
0%
1%
79%
0%
0%
0%
0%
0%
28%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
17%
100%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
8%
6%
89%
100%
       The data in Table 1-2 indicates that manufacturers had already begun implementing a
number of fuel economy/GHG reduction technologies in the baseline (2008) fleet. For
example, VW stands out as having a significant number of turbocharged direct injection
engines, though it is uncertain whether their engines are also downsized.  Some of the valve
and cam technologies are quite common in the baseline fleet: for example, nearly half the
baseline fleet already has dual cam phasing, while Honda and GM have considerable levels of
engines with cylinder deactivation.  Honda also has already implemented continuously
variable valve lift on a majority of their engines. Part of the implication of these technologies
already being present in the baseline is that if manufacturers have already implemented them,
they are therefore not available in the rulemaking analysis for improving fuel economy and
reducing COi emissions further, requiring the agencies to look toward increasing penetration
of these and other technologies and increasingly advanced technologies to project continued
improvements in stringency over time.

       The section below provides further detail on the conversion of the MY 2008 baseline
into the MYs 2017-2025 reference fleet. It also describes more of the data contained in the
baseline spreadsheet.

1.3  The MY 2017-2025 Reference Fleet

       The reference fleet aims to reflect the current market conditions and expectations
about conditions of the vehicle fleet during the model years to which the agencies' rules
apply. Fundamentally, constructing this fleet involved projecting the MY 2008 baseline fleet
into the MYs 2017-2025 model years. It also included the assumption that none of the vehicle
                                            1-10

-------
                                             The Baseline and Reference Vehicle Fleet
models had changes during this period. Projecting this future fleet is a process that is
necessarily uncertain. NHTSA and EPA therefore relied on many sources of reputable
information to make these projections.

       1.3.1  On what data is the reference vehicle fleet based?

       EPA and NHTSA have based the projection of total car and light truck sales on the
most recent projections available made by the Energy Information Administration (EIA). EIA
publishes a projection of national energy use annually called the Annual Energy Outlook
(AEO).3 EIA published its Early Annual Energy Outlook for 2011 in December 2010. EIA
released updated data to NHTSA in February (Interim AEO). The final release of AEO for
2011 came out in April 2011, but by that time EPA/NHTSA had already prepared modeling
runs for potential 2017-2025 standards using the interim data release to NHTSA. EPA and
NHTSA will use the newest version of AEO available in projecting the reference fleet for the
final rule.

       Similar to the analyses supporting the MYs 2012-2016 rulemaking, the agencies have
used the Energy Information Administration's (EIA's) National Energy Modeling System
(NEMS) to estimate the future relative market shares of passenger cars and light trucks.
However, NEMS methodology includes shifting vehicle sales volume, starting after 2007,
away from fleets with lower fuel economy (the light-truck fleet) towards vehicles with higher
fuel economies (the passenger car fleet) in order to facilitate compliance with CAFE and
GHG MYs 2012-2016 standards (the car and truck volumes based on this analysis are shown
in Table 1-3). Because we use our market projection as a baseline relative to which we
measure the effects of new standards, and we attempt to estimate the industry's ability to
comply with new standards without changing product mix (i.e., we analyze the effects of the
proposed rules assuming manufacturers will not change fleet composition as a compliance
strategy, as opposed to changes that might happen due to market forces), the Interim AEO
2011-projected shift in passenger car market share as a result of required fuel economy
improvements creates a circularity. Therefore, for the current analysis, the agencies
developed a new projection of passenger car and light truck sales shares by running scenarios
from the Interim AEO 2011 reference case that first deactivate the above-mentioned sales-
volume shifting methodology and then hold post-2017 CAFE standards constant at MY 2016
levels.  Incorporating these changes reduced the projected passenger car share of the light
vehicle market by an average of about 5% during  2017-2025. This case is referred to as the
"Unforced Reference Case," and the values are shown below in Table 1-4.
                                           1-11

-------
                                              The Baseline and Reference Vehicle Fleet
                         Table 1-3 AEO 2011 Reference Case Values
Model Year
2017
2018
2019
2020
2021
2022
2023
2024
2025
Cars
8,984,200
8,998,200
9,170,900
9,553,600
9,801,100
10,056,600
10,244,500
10,483,400
10,739,600
Trucks
6,812,000
6,552,200
6,391,300
6,336,200
6,380,000
6,384,600
6,396,500
6,407,700
6,470,200
Total Vehicles
15,796,100
15,550,400
15,562,200
15,889,800
16,181,100
16,441,200
16,641,000
16,891,100
17,209,800
                  Table 1-4 AEO 2011 Interim Unforced Reference Case Values
Model Year
2017
2018
2019
2020
2021
2022
2023
2024
2025
Cars
8,440,703
8,376,192
8,464,457
8,725,709
8,911,173
9,123,436
9,344,051
9,580,693
9,836,330
Tracks
7,365,619
7,200,218
7,114,201
7,170,230
7,277,894
7,316,337
7,311,438
7,353,394
7,414,129
Total Vehicles
15,806,322
15,576,410
15,578,658
15,895,939
16,189,066
16,439,772
16,655,489
16,934,087
17,250,459
       In 2017, car and light truck sales are projected to be 8.4 and 7.4 million units,
respectively.  While the total level of sales of 15.8 million units is similar to pre-2008 levels,
the fraction of car sales in 2017 and beyond is projected to be higher than in the 2000-2007
time frame. Note that EIA's definition of cars and tracks follows that used by NHTSA prior
to the MY 2011 CAFE final rale. The MY 2011 CAFE final rale reclassified approximately 1
million 2-wheel drive sport utility vehicles from the track fleet to the car fleet. EIA's sales
projections of cars and trucks for the 2017-2025 model years under the old NHTSA track
definition are shown above in Table  1-3 and Table 1-4.

       In addition to a shift towards  more car sales, sales of segments within both the car and
truck markets have also been changing and are expected to continue to change in the future.
Manufacturers are continuing to introduce more crossover models which offer much of the
                                            1-12

-------
                                             The Baseline and Reference Vehicle Fleet
utility of SUVs but use more car-like designs and unibody structures. In order to reflect these
changes in fleet makeup, EPA and NHTSA used a custom long range forecast purchased from
CSM Worldwide (CSM). CSM Worldwide (CSM)8 is a well-known industry analyst, that
provided the forecast used by the agencies for the 2012-2016 final rule. NHTSA and EPA
decided to use the forecast from CSM for several reasons. One, CSM uses a ground up
approach (e.g., looking at the number of plants and capacity for specific engines,
transmissions, and vehicles) for their forecast, which the agencies believe is a robust
forecasting approach.  Two, CSM agreed to allow us to publish their high level data, on which
the forecast is based, in the public domain.  Three, the CSM forecast covered all the
timeframe of greatest relevance to this analysis (2017-2025 model years). Four, it provided
projections of vehicle sales both by manufacturer and by market segment. And five, it
utilized market segments similar to those used in the EPA emission certification program and
fuel economy guide, such that the agencies could include only the vehicle types covered by
the proposed standards.

       CSM created a forecast that covered model years 2017-2025. Since the agencies used
this forecast to generate the reference fleet (i.e., the fleet expected to be sold absent any
increases in the stringency regulations after the 2016 model year), it is important for the
forecast to be independent of increases during 2017-2025 in the stringency of CAFE/ GHG
standards. However, CSM assumed that CAFE and GHG standards would continue to
increase in  stringency after 2016, although CSM did not use specific future standards as
quantitative inputs to its model. In its quantitative analysis, CSM used fuel price, industry
demand, consumer demand and other economic factors to project the composition of the
future fleet. In response to question by the agencies, CSM indicated that their assumption of
future standards had a negligible (non-discernable) impact on their forecast since it was not a
direct quantitative input to the model such that CSM's forecast would have been essentially
the same had CSM assumed no stringency increases  after 2016.

       The agencies combined the CSM forecast with data from other sources to create the
reference fleet projections. This process is discussed in sections that follow.

       1.3.2   How do the agencies develop the reference vehicle fleet?

       The process of producing the 2017-2025 reference fleet involved combining the
baseline fleet with the projection data described above.  This was a complex multistep
procedure, which is described  in this section.

       1.3.3   How was the 2008 baseline data merged with the  CSM data?

       EPA and NHTSA employed the same methodology as in the 2012-16 rule for mapping
certification vehicles to CSM vehicles. Merging the 2008 baseline  data with the 2017-2025
g CSM World Wide, CSM World Wide is a paid service provider.

                                            1-13

-------
                                             The Baseline and Reference Vehicle Fleet
CSM data required a thorough mapping of certification vehicles to CSM vehicles by
individual make and model. One challenge that the agencies faced when determining a
reference case fleet was that the sales data projected by CSM had different market
segmentation than the data contained in EPA's internal database. In order to create a common
segmentation between the two databases, the agencies performed a side-by-side comparison
of each vehicle model in both datasets, and created an additional "CSM segment" modifier in
the spreadsheet to map the two datasets.  The reference fleet sales based on the "CSM
segmentation" was then projected.

The baseline data and reference fleet volumes are available to the public. The baseline Excel
spreadsheet in the docket is the result of the merged files.4 The spreadsheet provides specific
details on the sources and definitions for the data. The Excel file contains several tabs.  They
are: "Data", "Data Tech Definitions", "SUM", "SUM Tech Definitions", "Truck Vehicle
Type Map", and "Car Vehicle Type Map". "Data" is the tab with the raw  data.  "Data Tech
Definitions" is the tab where each column is defined and its data source named. "SUM" is the
tab where the raw data is processed to be used in the OMEGA and Volpe models.  The
"SUM" tab minus columns A-F and minus the Generic vehicles is the input file for the
models.  The "Generic" manufacturer (shown in the "SUM" tab) is the sum of all
manufacturers and is calculated as a reference, and for data verification purposes.  It is used to
validate the manufacturers' totals.  It also gives an overview of the fleet.
                                            1-14

-------
                                           The Baseline and Reference Vehicle Fleet
      Table 1-5 shows the sum of the models chosen. The number of models is determined
by the number of unique segment and vehicle type combinations.  These combinations of
segment and vehicle type (the vehicle type number is the same as the technology package
number) are determined by the technology packages discussed in the EPA RIA.  "SUM Tech
Definitions" is the tab where the columns of the "SUM" tab are defined. The "Truck Vehicle
Type Map" and "Car Vehicle Type Map" map the number of cylinder and valve actuation
technology to the "tech package" vehicle type number.
                                          1-15

-------
                                                The Baseline and Reference Vehicle Fleet
                          Table 1-5 Models from the SUM Tab Model
                      Model
                      Car Like LargeSuv >=V8  Vehicle Type: 13
                      Car Like LargeSuv V6  Vehicle Type: 16
                      Car Like LargeSuv V6  Vehicle Type: 12
                      Car Like LargeSuv V6  Vehicle Type: 9
                      Car Like LargeSuv 14 and 15 Vehicle Type: 7
                      Car Like MidSizeSuv V6  Vehicle Type: 8
                      Car Like MidSizeSuv V6  Vehicle Type: 5
                      Car Like MidSizeSuv 14 Vehicle Type: 7
                      Car Like SmallSuv V6  Vehicle Type: 12
                      Car Like SmallSuv V6  Vehicle Type: 4
                      Car Like SmallSuv 14 Vehicle Type: 3
                      Large Auto >=V8 Vehicle Type: 13
                      Large Auto >=V8 Vehicle Type: 10
                      LargeAuto >=V8 Vehicle Type: 6
                      LargeAuto V6 Vehicle Type: 12
                      LargeAuto V6 Vehicle Type: 5
                      MidSizeAuto >=V8  Vehicle Type: 13
                      MidSizeAuto >=V8  Vehicle Type: 10
                      MidSizeAuto >=V8 (7 or >)  Vehicle Type: 6
                      MidSizeAuto V6  Vehicle Type: 12
                      MidSizeAuto V6  Vehicle Type: 8
                      MidSizeAuto V6  Vehicle Type: 5
                      MidSizeAuto 14 Vehicle Type: 3
       In the combined EPA certification and CSM database, all 2008 vehicle models were
assumed to continue out to 2025, though their volumes changed in proportion to CSM
projections. Also, any new models expected to be introduced within the 2009-2025
timeframe are not included in the data.  These volumes are reassigned to the existing models
to keep the overall fleet volume the same.  All MYs 2017-2025 vehicles are mapped to the
existing vehicles by a process of mapping to manufacturer market share and overall segment
distribution. The mappings are discussed in the next section. Further discussion of this
limitation is discussed below in section 1.3.4. The statistics of this fleet will be presented
below since further modifications were required to the volumes as the next section describes.

       1.3.4  How were the CSM forecasts normalized to the AEO forecasts?

       The next step in the agencies' generation of the reference fleet is one of the more
complicated steps to explain.  Here, the projected CSM forecasts for relative sales of cars and
trucks by manufacturer and by market segment was normalized (set equal) to the total sales
estimates of the Early Release of the 2011 Annual Energy Outlook (AEO). NHTSA and EPA
used projected car and truck volumes for this period from Early AEO 2011. However, the
AEO projects  sales only at the car and truck level, not at the manufacturer and model-specific
level, and the agencies' analysis requires this further level of detail.  The CSM data provided
year-by-year percentages of cars and trucks sold by each manufacturer as well as the

                                              1-16

-------
                                             The Baseline and Reference Vehicle Fleet
percentages of each vehicle segment. Using these percentages normalized to the AEO-
projected volumes then provided the manufacturer-specific market share and model-specific
sales for model years 2017-2025 (it is worth clarifying that the agencies are not using the
model-specific sales volumes from CSM, only the higher-level volumes by manufacturer and
segment).  This process is described in greater detail in the following paragraphs.

      In order to determine future production volumes, the agencies developed multipliers
by manufacturer and vehicle segment that could be applied to MY 2008 volumes.  The
process for developing the multipliers is complicated, but is easiest to explain as a three-step
process, though the first step is combined with both the second and third step, so only one
multiplier per manufacturer and vehicle segment is developed.

      The three steps are:

          1) Adjust total car and truck sales to match AEO projections.

          2) Adjust car sales to match CSM market share projections for each manufacturer
             and car segment.

          3) Adjust truck sales to match CSM market share projections for each
             manufacturer and truck segment.

      The first step is the adjustment of total car and truck sales in 2008 to match AEO
projections of total car and truck sales in 2017-2025. The volumes for all of the trucks in
2008 were added up (TruckSum2008), and so were the volumes of all the cars (CarSum2008).
A multiplier was developed to scale the volumes in 2008 to the AEO projections.  The
example equation below  shows the general form of how to calculate a car or truck multiplier.
The AEO projections are shown above in Table 1-3.
       Example Equation :

       TruckMultiplier(Year X) = AEOProjectionforTrucks(Year X) / TruckSum2008

       CarMultiplier(Year X) = AEOProjectionforCars(Year X) / CarSum2008

       Where: Year X is the model year of the multiplier.

       The AEO projection is different for each model year. Therefore, the multipliers are
different for each model year. The multipliers can be applied to each 2008 vehicle as a first
adjustment, but multipliers based solely on AEO have limited value since those multipliers
can only give an adjustment that will give the correct total numbers of cars and trucks without
the correct market share or vehicle mix.  A correction factor based on the CSM data, which
does contain market share and vehicle segment mix, is therefore necessary,  so combining the
                                            1-17

-------
                                             The Baseline and Reference Vehicle Fleet
AEO multiplier with CSM multipliers (one per manufacturer, segment, and model year) will
give the best multipliers.

       There were several steps in developing an adjustment for Cars based on the CSM data.
CSM provided data on the market share and vehicle segment distribution.  The first step in
determining the adjustment for Cars was to total the number of Cars in each vehicle segment
by manufacturer in MY 2008. A total for all manufacturers in each segment was also
calculated.  The next step was to multiply the volume of each segment for each manufacturer
by the CSM market share. The AEO multiplier was also applied at this time.  This gave
projected volumes with AEO total volumes and market share correction for Cars.  This is
shown in the "Adjusted for 2017AEO and Manufacturer Market Share" column of Table 1-6.

       The next step is to adjust the sales volumes for CSM vehicle segment distribution.
The process for adjusting for vehicle segment is more complicated than a simple one step
multiplication. In order to keep manufacturers' volumes constant and still have  the correct
vehicle segment distribution, vehicles need to move from segment to segment while
maintaining constant manufacturers' totals.  Six rules and one assumption were applied to
accomplish the shift. The assumption (based on the shift in vehicle sales in 2008 and 2009) is
that people are moving to smaller vehicles in the rulemaking time frame independently of
regulatory requirements. A higher-level (less detailed) example of this procedure is provided
in Section II of the preamble.

       Vehicles from CSM's "Luxury Car," "Specialty Car," and "Other Car" segments, if
reduced, will be equally distributed to the remaining four categories ("Full-Size Car," "Mid-
Size Car," "Small Car," "Mini Car"). If these sales increased, they were taken from the
remaining four categories so that the relative sales in these four categories  remained constant.

       Vehicles from CSM's "Luxury Car," "Specialty Car," and "Other Car" segments, if
increased will take equally from the remaining categories ("Full-Size Car," "Mid-Size Car,"
"Small Car," "Mini Car").

       All manufacturers have the same multiplier for a given segment shift based on moving
all vehicles in that segment to achieve the CSM distribution.  Table 1-6 shows how the 2017
vehicles moved and the multipliers that were created for each adjustment.  This does not mean
that new vehicle segments will be added (except for Generic Mini  Car described in the  next
step) to manufacturers that do not produce them. Vehicles within each manufacturer will be
shifted as close to the distribution as possible given the other rules. Table  1-7 has  the
percentages of Cars per CSM segment.  These percentages are multiplied by the total number
of vehicles in a given year to get the total sales in the segment.  Table 1-6 shows the totals for
2017 in the "2017 AEO-CSM Sales Goal" column.

       When "Full-Size Car," "Mid-Size Car," "Small Car" are processed, if vehicles need to
move in or out of the segment, they will move into or out of the next smaller segment.  So, if
Mid-Size Cars are being processed they can only move to or be taken from Small Cars.  Note:
In order to accomplish this, a "Generic Mini Car" segment was added to manufacturers who


                                           1-18

-------
                                             The Baseline and Reference Vehicle Fleet
did not have a Mini (type) Car in production in 2008, but needed to shift down vehicles from
the Small Car segment.

       The data must be processed in the following order: "Luxury Car," "Specialty Car,"
"Other Car," "Full-Size Car," "Mid-Size Car," "Small Car." The "Mini Car" does not need to
be processed separately. By using this order, it works out that vehicles will always move
toward the correct distribution. There are two exceptions, BMW and Porsche only have
"Luxury Car," "Specialty Car," and "Other Car" vehicles, so their volumes were not changed
or shifted since these rules did not apply to them.

       When an individual manufacturer multiplier is applied for a segment, the vehicles
move to or from the appropriate segments as specified in the previous rules and as shown  in
Table 1-6.
                         Table 1-6 2017 Model Year Volume Shift*
CSM Segment
All Full-Size Car
All Luxury Car
All Mid-Size Car
All Mini Car
All Small Car
All Specialty Car
All Others
2008 MY
Sales
829,896
1,048,341
2,103,108
617,902
1,912,736
469,324
0
Adjusted for
2017 AEO and
Manufacturer
Market Share
830,832
1,408,104
2,500,723
868,339
2,548,393
627,425
0
Luxury,
Specialty,
Other
Adjustment
818,226
1,423,691
2,475,267
851,234
2,513,350
702,048
0
Full Size
Adjustment
347,034
1,423,691
2,946,459
851,234
2,513,350
702,048
0
Midsize
Adjustment
347,034
1,423,691
2,431,715
851,234
3,028,094
702,048
0
Small Car
Adjustment
347,034
1,423,691
2,431,715
1,439,985
2,439,343
702,048
0
2017
AEO-
CSM
Sales Goal
347,034
1,423,691
2,431,715
1,439,985
2,439,343
702,048
0
Number Vehicles that shift and Where
All Full-Size Car
All Luxury Car
All Mid-Size Car
All Mini Car
All Small Car
All Specialty Car
All Others














(12,606)
15,587
(25,456)
(17,105)
(35,043)
74,623
0
(471,192)
0
471,192
0
0
0
0
0
0
(514,744)
0
514,744
0
0
0
0
0
588,751
(588,751)
0
0







Individual Manufacturer Multiplier
All Full-Size Car
All Luxury Car





0.973
0.42







                                            1-19

-------
                                             The Baseline and Reference Vehicle Fleet
All Mid-Size Car
All Mini Car
All Small Car
All Specialty Car
All Others













0.963
1





0.97





1.55
0.96







                       Table 1-7 CSM - Percent of Cars per Segment*
CSM Segment
Compact Car
Full-Size Car
Luxury Car
Mid-Size Car
Mini Car
Small Car
Specialty Car
Others
2017
0.00%
3.95%
16.70%
27.68%
15.33%
27.77%
8.56%
0.00%
2018
0.00%
3.56%
16.87%
27.77%
15.46%
27.57%
8.76%
0.00%
2019
0.00%
3.35%
17.14%
27.47%
1 5.45%
27.74%
8.84%
0.00%
2020
0.00%
4.10%
17.23%
26.94%
1 5.46%
27.99%
8.27%
0.00%
2021
0.00%
3.59%
17.05%
27.18%
15.59%
28.29%
8.29%
0.00%
2022
0.00%
3.03%
17.02%
27.82%
15.67%
28.43%
8.03%
0.00%
2023
0.00%
2.97%
17.10%
28.51%
1 5.47%
28.18%
7.77%
0.00%
2024
0.00%
2.46%
17.40%
28.11%
15.23%
28.49%
8.31%
0.00%
2025
0.00%
2.46%
17.40%
28.11%
15.23%
28.49%
8.31%
0.00%
       Mathematically, an individual manufacturer multiplier is calculated by making the
segment the goal and dividing by the previous total for the segment (shown in Table 1-7). If
the number is greater than 1, the vehicles are entering the segment, and if the number is less
than 1, the vehicles are leaving the segment.  So, for example, if Luxury Cars have an
adjustment of 1.5, then for a specific manufacturer who has Luxury Cars, a multiplier of 1.5 is
applied to its luxury car volume, and the total number of vehicles that shifted into the Luxury
segment is subtracted from the remaining segments to maintain that company's market share.
On the other hand, if Large Cars have an adjustment of 0.7, then for a specific manufacturer
who has Large Cars, a multiplier of 0.7 is applied to its Large Cars, and the total number of
vehicles leaving that segment is transferred into that manufacturer's Mid-Size Cars.

       After the vehicle volumes are shifted using the above rules, a total for each
manufacturer and vehicle segment is maintained.  The total for each manufacturer segment for
a specific model year (e.g., 2017 General Motors Luxury Cars) divided by the MY 2008 total
for that manufacturer segment (e.g., 2008 General Motors Luxury Cars) is the new multiplier
used to determine the future vehicle volume for each vehicle model.  This is done by taking
the multiplier (which is for a specific manufacturer and segment) times the MY 2008 volume
for the specific vehicle model (e.g., 2008 General Motors Luxury Car Cadillac CTS). This
process is repeated for each model  year (2017-2025).

       The method used to adjust CSM Trucks to the AEO market share was different than
the method used for Cars. The process for Cars is different than Trucks because it is not
                                            1-20

-------
                                             The Baseline and Reference Vehicle Fleet
possible to predict how vehicles would shift between segments based on current market
trends.  This is because of the added utility of some trucks that makes their sales more
insensitive to factors like fuel price. Again, CSM provided data on the market share and
vehicle segment distribution. The process for having the fleet match CSM's market share and
vehicle segment distribution was iterative.

       The following totals were determined:

       •  The total number of trucks  for each manufacturer in 2008 model year.

       •  The total number of trucks  in each truck segment in 2008 model year.

       •  The total number of truck in each segment for each manufacturer in 2008 model
          year.

       •  The total number of trucks  for each manufacturer in a specific future model year
          based on the AEO and CSM data. This is the goal for market share.

       •  The total number of trucks  in each truck segment in a specific future model year
          based on the AEO and CSM data. This is the goal for vehicle segment
          distribution. Table 1-8 has the percentages of Trucks per CSM segment.

                       Table 1-8 CSM - Percent of Trucks per Segment
CSM Segment
Full-Size CUV
Full-Size Pickup
Full-Size SUV
Full-Size Van
Mid-Size CUV
Mid-Size MAV
Mid-Size Pickup
Mid-Size SUV
Mid-Size Van
Small CUV
Small MAV
Small SUV
2017
5.9%
16.8%
1.9%
1.2%
18.0%
4.5%
6.1%
4.1%
11.6%
26.0%
2.5%
1.3%
2018
6.3%
16.5%
1.5%
1.2%
17.4%
4.6%
6.1%
4.8%
11.9%
25.9%
2.6%
1.2%
2019
6.8%
15.9%
1.3%
1.1%
17.6%
4.9%
6.1%
4.8%
11.9%
25.7%
2.8%
1.1%
2020
7.5%
16.1%
1.0%
1.4%
17.2%
5.4%
5.6%
4.5%
11.7%
25.6%
2.9%
1.2%
2021
8.3%
15.4%
0.9%
1.3%
16.9%
5.9%
5.7%
4.7%
11.6%
25.1%
3.0%
1.1%
2022
8.8%
15.1%
0.8%
1.3%
16.8%
6.2%
5.7%
4.8%
11.6%
24.9%
3.1%
1.1%
2023
9.5%
14.3%
0.5%
1.3%
16.8%
6.5%
5.8%
4.8%
11.6%
24.7%
3.1%
1.1%
2024
9.2%
13.8%
0.5%
1.2%
17.0%
7.1%
5.9%
4.6%
11.3%
25.3%
3.2%
1.0%
2025
9.1%
13.5%
0.6%
1.2%
17.0%
7.4%
5.8%
4.6%
11.3%
25.3%
3.2%
1.0%
       To start, two different types of tables were created. One table had each manufacturer
with its total sales for 2008 (similar to Table 1-10).  This table will have the goal for each
                                            1-21

-------
                                              The Baseline and Reference Vehicle Fleet
manufacturer, and a column added for each iteration with the current total.  The second table
has a truck segment total by manufacturer.  The second table starts out with a "Generic"
manufacturer (Table 1-10) which is the table where the goal resides. Each manufacturer
(BMW for example is shown in Table 1-11) is then listed below the "Generic" manufacturer.
With each iteration, a new total is added for each segment that is calculated and added to the
table. This is not shown in the tables below. The agencies then engaged in a process of first
adjusting the numbers in the tables to the goal for market share distribution. This was
followed by  adjusting to the goal for vehicle segment distribution. Each time an adjustment
was done a new column was added.  An adjustment was done by creating a multiplier (either
segment distribution-based or manufacturer distribution-based) and applying it to each vehicle
segment total in the current iteration.  A manufacturer-based multiplier is calculated by taking
the goal total for a manufacturer and dividing by the current total (starting with 2008 model
year volumes) for a manufacturer. A segment distribution-based multiplier is calculated by
taking the goal distribution volumes in the Generic manufacturer set and dividing them by the
current volume.  Table 1-9, Table 1-10, and Table 1-11 below illustrates two iterations using
BMW as an  example.

                            Table 1-9 Manufacturer Truck Totals

BMW
2008 Model Year Sales
61,324
Manufacturer Distribution 2017 Volume Goal
138.053
Multiplier for Iteration 1
138,053/61324=2.25
                                            1-22

-------
                                              The Baseline and Reference Vehicle Fleet
                 Table 1-10 Segment Specific Truck Totals for All Manufacturers
Manufacturer
Generic**
Generic
Generic
Generic
Generic
Generic
Generic
Generic
Generic
Generic
Generic
Generic
CSM Segment
Full-Size Pickup
Mid-Size Pickup
Full-Size Van
Mid-Size Van
Mid-Size MAV
Small MAV
Full-Size SUV
Mid-Size SUV
Small SUV
Full-Size CUV
Mid-Size CUV
Small CUV
2008 Model Year Sales
1,332,335
452,013
33,384
719,529
110,353
231,265
559,160
436,080
196,424
264,717
923,165
1,612,029
Segment Distribution 2017
Volume Goal
1,240,844
452,017
85,381
855,022
331,829
186,637
138,821
305,382
94,657
433,683
1,327,905
1,913,439
Multipliers
0.931
1.000
2.558
1.188
3.007
0.807
0.248
0.700
0.482
1.638
1.438
1.187
 ** Generic means all manufacturers.
                      Table 1-11 Segment Specific Truck Totals for BMW
Manufacturer
BMW
BMW
BMW
BMW
BMW
BMW
BMW
BMW
BMW
BMW
BMW
BMW
CSM Segment
Full-Size Pickup
Mid-Size Pickup
Full-Size Van
Mid-Size Van
Mid-Size MAV
Small MAV
Full-Size SUV
Mid-Size SUV
Small SUV
Full-Size CUV
Mid-Size CUV
Small CUV
Total BMW Vehicles
2008 Model Year
Sales




3,882





36,409
21,033
61,324
Iteration 1 Adjust for
Market Share




2.25*3,882=8,739





2.25*36,409=81,964
2.25*21,033=47,350
138,053
Iteration 2 Adjust for Segment
Distribution




2.85*8,739=24,907





1.1*81,964=90,134
1.. 02*47,350=48,306
163,347
       Using this process, the numbers will get closer to the goal of matching CSM's market
share for each manufacturer and distribution for each vehicle segment after each of the
iterations. The iterative process is carried out until the totals nearly match the goals.

       After 19 iterations, all numbers were within 0.01% of CSM's distributions. The
calculation iterations could have been stopped sooner, but they were continued to observe
how the numbers would converge.

       After the market share and segment distribution were complete, the totals need to be
used to create multipliers that could be applied to the original individual 2008 model year
vehicle volumes (each unique manufacture models volume). The total for each manufacturer
                                             1-23

-------
                                             The Baseline and Reference Vehicle Fleet
segment divided by the 2008 model year total for each manufacturer segment gives a
multiplier that can be applied to each vehicle based on its manufacturer and segment.

       The above process is done for each model year needed (2017-2025). The multipliers
are then applied to each vehicle in 2008 model year, which gives a volume for each vehicle in
2017 through 2025 model year.

       1.3.5  What are the sales volumes and characteristics of the reference  fleet?

       Table 1-12 and Table 1-14 below contain the sales volumes that result from the
process above for MY 2008 and 2017-2020.  Table 1-13 and Table 1-15 below contain the
sales volumes that result from the process above for MY 2021-2025.
                           Table 1-12 Vehicle Segment Volumes3
Reference Class Segment
LargeAuto
MidSizeAuto
CompactAuto
SubCmpctAuto

LargePickup
SmallPickup
LargeSUV
MidSizeSUV
SmallSUV
MiniVan
CargoVan
Actual and Projected Sales Volume
2008
562,240
3,098,927
1,979,461
1,365,833

1,582,226
177,497
2,783,949
1,263,360
285,355
642,055
110,858
2017
376,107
3,311,268
2,347,980
2,458,222

1,514,619
156,227
3,194,489
1,358,755
148,251
754,562
185,841
2018
356,768
3,290,408
2,325,393
2,454,112

1,443,766
157,932
3,150,101
1,309,212
149,933
739,551
199,234
2019
353,609
3,303,621
2,369,301
2,489,208

1,383,190
160,752
3,177,868
1,267,394
154,675
717,065
201,974
2020
394,864
3,381,785
2,448,021
2,553,350

1,386,195
146,029
3,203,244
1,285,822
162,677
714,323
219,628
a Volumes in this table are based on the pre-2011 NHTSA definition of Cars and Trucks.
                           Table 1-13 Vehicle Segment Volumes3
Reference Class Segment
LargeAuto
MidSizeAuto
CompactAuto
SubCmpctAuto

LargePickup
SmallPickup
LargeSUV
MidSizeSUV
SmallSUV
MiniVan
CargoVan
Actual and Projected Sales Volume
2021
380,192
3,442,116
2,520,977
2,626,364

1,368,301
150,123
3,312,914
1,281,240
167,223
729,078
210,539
2022
358,295
3,548,263
2,592,199
2,687,167

1,349,421
147,138
3,362,608
1,283,244
169,643
738,982
202,812
2023
362,672
3,692,533
2,632,926
2,721,102

1,301,293
151,315
3,412,753
1,268,288
170,239
740,785
201,585
2024
356,173
3,751,496
2,744,634
2,796,061

1,271,751
154,627
3,475,873
1,292,662
173,191
720,720
196,900
2025
368,843
3,814,941
2,843,069
2,878,288

1,260,389
154,838
3,520,992
1,305,362
175,713
726,256
201,768
                                            1-24

-------
                                               The Baseline and Reference Vehicle Fleet
a Volumes in this table are based on the pre-2011 NHTSA definition of Cars and Trucks.
                Table 1-14 2011+ NHTSA Car and Truck Definition Based Volumes
Vehicle Type
Trucks
Cars
Cars and Trucks
Actual and Projected Sales Volume
2008
5,621,193
8,230,568
13,851,761
2017
5,818,655
9,987,667
15,806,322
2018
5,671,046
9,905,364
15,576,410
2019
5,582,962
9,995,696
15,578,658
2020
5,604,377
10,291,562
15,895,939
                Table 1-15 2011+ NHTSA Car and Truck Definition Based Volumes
Vehicle Type
Trucks
Cars
Cars and Trucks
Actual and Projected Sales Volume
2021
5,683,902
10,505,165
16,189,066
2022
5,703,996
10,735,777
16,439,772
2023
5,687,486
10,968,003
16,655,489
2024
5,675,949
11,258,138
16,934,087
2025
5,708,899
11,541,560
17,250,459
       Table 1-16 and Table 1-17 below contain the sales volumes by manufacturer and
vehicle type for MY 2008 and 2017-2025.
               Table 1-16 NHTSA Car and Truck Definition Manufacturer Volumes
Manufacturers
All
All
All
Aston Martin
Aston Martin
BMW
BMW
Chrysler/Fiat
Chrysler/Fiat
Daimler
Daimler
Ferrari
Ferrari
Ford
Ford
Vehicle Type
Both
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
2008
Baseline
Sales
13,851,761
8,230,568
5,621,193
1,370
-
291,796
61,324
703,158
956,792
208,195
79,135
1,450
-
956,699
814,194
2017
Projected
Volume
15,806,322
9,987,667
5,818,655
1,035
-
313,022
138,053
418,763
409,702
284,847
86,913
6,676
-
1,299,899
763,549
2018
Projected
Volume
15,576,410
9,905,364
5,671,046
1,051
-
322,939
131,942
397,538
387,858
276,409
83,651
6,700
-
1,311,467
748,829
2019
Projected
Volume
15,578,658
9,995,696
5,582,962
1,072
-
346,075
131,373
391,689
366,447
281,425
88,188
6,794
-
1,332,039
717,773
2020
Projected
Volume
15,895,939
10,291,562
5,604,377
1,034
-
357,942
128,339
415,319
360,677
290,989
92,919
6,916
-
1,378,789
717,037
                                              1-25

-------
                               The Baseline and Reference Vehicle Fleet
Ford
Geely/Volvo
Geely/Volvo
GM
GM
HONDA
HONDA
HYUNDAI
HYUNDAI
Kia
Kia
Lotus
Lotus
Mazda
Mazda
Mitsubishi
Mitsubishi
Nissan
Nissan
PORSCHE
PORSCHE
Spyker/Saab
Spyker/Saab
Subaru
Subaru
Suzuki
Suzuki
Tata/JLR
Tata/JLR
Tesla
Tesla
Toyota
Toyota
Volkswagen
Volkswagen
Cars
Trucks
Cars
Trucks
Cars
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
956,699
32,748
65,649
1,507,797
1,587,391
1,006,639
505,140
337,869
53,158
221,980
59,472
252
-
246,661
55,885
85,358
15,371
717,869
305,546
18,909
18,797
21,706
4,250
116,035
82,546
79,339
35,319
9,596
55,584
800
-
1,260,364
951,136
291,483
26,999
1,299,899
41,887
88,234
1,362,761
1,462,204
1,154,600
596,481
592,027
152,885
322,044
98,702
240
-
253,540
51,788
65,099
37,632
870,797
444,938
35,093
13,233
20,024
2,871
224,112
78,242
90,708
22,109
55,881
57,579
27,986
-
1,849,196
1,330,511
551,638
128,819
1,311,467
42,187
89,394
1,438,355
1,474,076
1,138,087
544,619
578,373
151,461
312,370
98,280
243
-
262,512
57,535
63,671
36,300
849,678
412,383
35,444
12,001
20,007
3,596
216,598
75,152
89,932
21,385
56,222
56,606
28,435
-
1,834,181
1,223,415
540,036
145,491
1,332,039
43,125
91,575
1,505,025
1,493,511
1,144,639
527,535
582,971
155,642
314,879
100,679
250
-
266,951
57,494
63,826
35,454
854,400
398,559
36,116
11,469
20,144
3,826
217,095
72,832
90,568
20,692
57,267
57,854
28,990
-
1,836,306
1,142,104
537,114
146,891
1,378,789
42,615
93,003
1,530,755
1,544,983
1,163,666
525,089
598,283
154,173
323,676
96,535
266
-
270,078
58,154
65,080
35,215
882,791
397,869
35,963
11,141
21,069
3,509
223,466
72,458
93,548
20,675
58,182
56,213
27,965
-
1,883,734
1,154,304
554,822
146,700
Table 1-17 NHTSA Car and Truck Definition Manufacturer Volumes
Manufacturers
All
All
All
Aston Martin
Aston Martin
Vehicle Type
Both
Cars
Trucks
Cars
Trucks
2020
Projected
Volume
16,189,066
10,505,165
5,683,902
1,058
-
2021
Projected
Volume
16,439,772
10,735,777
5,703,996
1,049
-
2022
Projected
Volume
16,655,489
10,968,003
5,687,486
1,041
-
2023
Projected
Volume
16,934,087
11,258,138
5,675,949
1,141
-
2024
Projected
Volume
17,250,459
11,541,560
5,708,899
1,182
-
                              1-26

-------
 The Baseline and Reference Vehicle Fleet
BMW
BMW
Chrysler/Fiat
Chrysler/Fiat
Daimler
Daimler
Ferrari
Ferrari
Ford
Ford
Ford
Geely/JLR
Geely/JLR
GM
GM
HONDA
HONDA
HYUNDAI
HYUNDAI
Kia
Kia
Lotus
Lotus
Mazda
Mazda
Mitsubishi
Mitsubishi
Nissan
Nissan
PORSCHE
PORSCHE
Spyker/Saab
Spyker/Saab
Subaru
Subaru
Suzuki
Suzuki
Tata/JLR
Tata/JLR
Tesla
Tesla
Toyota
Toyota
Volkswagen
Volkswagen
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
Cars
Trucks
359,098
128,724
421,013
348,613
300,378
99,449
7,059
-
1,401,617
714,181
1,401,617
41,768
92,726
1,530,020
1,564,277
1,198,880
535,916
613,355
156,466
331,319
95,432
278
-
274,740
59,227
65,851
35,309
912,629
408,029
36,475
11,242
21,294
3,560
230,780
72,773
95,725
20,767
58,677
58,153
28,623
-
1,903,706
1,215,539
585,607
148,734
360,034
128,899
424,173
363,008
304,738
100,935
7,138
-
1,415,221
714,266
1,415,221
41,686
92,512
1,507,653
1,578,556
1,237,504
539,235
627,964
157,493
339,102
94,694
290
-
281,150
60,307
67,261
35,227
937,447
411,883
36,607
11,385
21,709
3,461
238,613
72,736
97,599
20,734
59,349
58,590
28,369
-
1,986,077
1,235,052
593,314
146,750
360,561
127,521
423,882
361,064
312,507
105,315
7,227
-
1,474,797
700,005
1,474,797
42,031
96,840
1,496,819
1,606,495
1,265,564
536,898
634,308
161,189
342,746
95,688
299
-
296,910
61,966
67,680
35,469
954,340
417,121
36,993
11,370
22,410
3,435
241,612
73,022
99,263
20,803
60,639
58,865
28,150
-
2,036,992
1,224,980
596,749
153,927
388,193
146,525
426,017
344,962
332,337
107,084
7,441
-
1,503,670
688,854
1,503,670
42,461
99,181
1,493,597
1,636,805
1,307,851
536,994
657,710
166,092
351,882
96,119
308
-
300,614
61,971
70,728
36,001
982,771
422,217
39,504
11,409
22,800
3,426
248,283
74,142
100,447
21,162
63,728
57,981
30,862
-
2,080,528
1,208,013
605,336
156,939
405,256
145,409
436,479
331,762
340,719
101,067
7,658
-
1,540,109
684,476
1,540,109
42,588
101,107
1,524,008
1,673,936
1,340,321
557,697
677,250
168,136
362,783
97,653
316
-
306,804
61,368
73,305
36,387
1,014,775
426,454
40,696
11,219
23,130
3,475
256,970
74,722
103,154
21,374
65,418
56,805
31,974
-
2,108,053
1,210,016
630,163
154,284
1-27

-------
                                             The Baseline and Reference Vehicle Fleet
       Table 1-18 also shows how the change in fleet make-up may affect the footprint
distributions over time.  The resulting data indicate that footprint will not change significantly
between 2008 and 2025. There will be an increase in the number of cars sold, which will
cause the average footprints for cars and trucks combined to be slightly smaller (about 2%).
This is the result of AEO projecting an increased number of cars, and CSM predicting that
most of that increase will be in the subcompact segment. Again, we note that in order to
ensure that our baseline inputs were not influenced by the proposed regulations, agencies re-
ran AEO to hold standards constant after 2016 (the reader will remember from the text above
that CSM had indicated that its projections were not sensitive to assumptions about new
standards).

                          Table 1-18 Production Foot Print Mean
Model Year
2008
2017
2018
2019
2020
2021
2022
2023
2024
2025
Average
Footprint of all
Vehicles
48.9
48.2
48.1
48.0
48.0
48.0
47.9
47.9
47.7
47.7
Average Footprint
Cars
45.4
44.9
44.9
44.9
44.9
44.9
44.9
44.9
44.9
44.9
Average
Footprint
Trucks
53.9
53.8
53.7
53.6
53.7
53.6
53.6
53.5
53.4
53.3
       Table 1-19 below shows the changes in engine cylinders over the model years.  The
current assumptions show that engines will be downsized over the model years to which these
proposed rules apply. This shift is a projected consequence of the expected changes in class
and segment mix as predicted by AEO and CSM, and does not represent engine downsizing
attributable to the 2012-2016 light-duty CAFE and GHG standards.
                                            1-28

-------
                                              The Baseline and Reference Vehicle Fleet
                 Table 1-19 Percentages of 4,6, 8 Cylinder Engines by Model Year

Model
Year
2008
2017
2018
2019
2020
2021
2022
2023
2024
2025
Trucks
4
Cylinders
10.3%
10.9%
10.6%
10.4%
10.3%
10.3%
10.3%
10.3%
10.5%
10.5%
6
Cylinders
56.4%
63.7%
64.5%
65.5%
65.6%
66.3%
66.7%
67.7%
68.1%
68.2%
8
Cylinders
33.3%
25.4%
24.8%
24.1%
24.1%
23.4%
23.0%
22.0%
21.4%
21.3%
Cars
4
Cylinders
56.9%
60.6%
60.7%
60.7%
60.3%
60.6%
61.1%
60.9%
61.0%
61.1%
6
Cylinders
37.8%
34.5%
34.4%
34.3%
34.7%
34.4%
34.2%
34.3%
34.1%
34.0%
8
Cylinders
5.3%
5.0%
5.0%
5.0%
5.0%
4.9%
4.8%
4.8%
4.8%
4.8%
For the final rule, the agencies intend to use a more recent version of EIAs AEO, and we also
will consider using MY 2010 for the baseline, and potentially an updated future market
forecast.
                                             1-29

-------
                                           The Baseline and Reference Vehicle Fleet
References:

1 EPA's Omega Model and input sheets are available at
http://www.epa.gov/oms/climate/models.htm; DOT/NHTSA's CAFE Compliance and Effects
Modeling System (commonly known as the "Volpe Model") and input and output sheets are
available at http://www.nhtsa.gov/fuel-economy.

2http://www.nhtsa.gov/Laws+&+Regulations/CAFE+-
+Fuel+Economy/CAFE+Compliance+and+Effects+Modeling+System:+The+Volpe+Model

3 Department of Energy, Energy Information Administration, Annual Energy Outlook (AEO)
2011, Early Release. Available at http://www.eia.gov/forecasts/aeo/ (last accessed Aug. 15,
2011).

4 The baseline Excel file ("2008-2025 Production Summary Data _Definitions Docket
08_27_2009") is available in the docket (Docket EPA-HQ-OAR-2010-0799).
                                         1-30

-------
                      What are the Attribute-Based Curves the Agencies are Proposing


Chapter 2:  What are the Attribute-Based Curves the Agencies are
             Proposing, and How Were They Developed?

2.1 Why are standards attribute-based and defined by a mathematical function?

       As in the MYs 2012-2016 CAFE/GHG rales, and as NHTSA did in the MY 2011
CAFE rale, NHTSA and EPA are proposing to set attribute-based CAFE and COi standards
that are defined by a mathematical function. EPCA, as amended by EISA, expressly requires
that CAFE standards for passenger cars and light tracks be based on one or more vehicle
attributes related to fuel economy, and be expressed in the form of a mathematical function.1
The CAA has no such requirement,  although such  an approach is permissible under section
202 (a) and EPA has used the attribute-based approach in issuing standards under analogous
provisions of the CAA (e.g., criteria pollutant standards for non-road diesel engines using
engine size as the  attribute,2 in the recent GHG standards for heavy duty pickups and vans
using a work factor attribute,3 and in the MYs 2012-2016 GHG rule itself which used vehicle
footprint as the attribute). Public comments on the MYs 2012-2016 rulemaking widely
supported  attribute-based standards  for both agencies' standards.

       Under an attribute-based standard, every vehicle model has a performance target (fuel
economy and CO2 emissions for CAFE and CO2 emissions standards, respectively), the level
of which depends  on the vehicle's attribute (for this proposal, footprint, as discussed below).
The manufacturers' fleet average performance is determined by the production-weighteda
average (for CAFE, harmonic average) of those  targets.

       The agencies believe that an attribute-based standard is preferable to a single-industry-
wide average standard in the context of CAFE and COi standards for several reasons. First, if
the shape is chosen properly, every manufacturer is more likely to be required to continue
adding more fuel efficient technology each year across their fleet, because the stringency of
the compliance obligation will depend on the particular product mix of each manufacturer.
Therefore  a maximum feasible attribute-based standard will tend to require greater fuel
savings and COi emissions reductions overall than would a maximum  feasible flat standard
(that is, a single mpg or COi level applicable to every manufacturer).

       Second, depending on the attribute, attribute-based standards reduce the incentive for
manufacturers to respond to CAFE and COi standards in ways harmful to safety.   Because
each vehicle model has its own target  (based on  the attribute chosen), properly fitted attribute-
a Production for sale in the United States.
b The 2002 NAS Report described at length and quantified the potential safety problem with average fuel
economy standards that specify a single numerical requirement for the entire industry. See 2002 NAS Report at
5, finding 12. Ensuing analyses, including by NHTSA, support the fundamental conclusion that standards
structured to minimize incentives to downsize all but the largest vehicles will tend to produce better safety
outcomes than flat standards.

                                            2-1

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

based standards provide little, if any, incentive to build smaller vehicles simply to meet a
fleet-wide average, because the smaller vehicles will be subject to more stringent compliance
targets.0

       Third, attribute-based standards provide a more equitable regulatory framework for
different vehicle manufacturers.11 A single industry-wide average standard imposes
disproportionate cost burdens and compliance difficulties on the manufacturers that need to
change their product plans to meet the standards, and puts no obligation on those
manufacturers that have no need to change their plans. As discussed above, attribute-based
standards help to spread the regulatory cost burden for fuel economy more broadly across all
of the vehicle manufacturers within the industry.

       Fourth, attribute-based standards better respect economic conditions and  consumer
choice, as compared to single-value standards.  A flat, or single value standard, encourages a
certain vehicle size fleet mix by creating incentives for manufacturers to use vehicle
downsizing as a compliance strategy.  Under a footprint-based standard, manufacturers are
required to invest in technologies that improve the fuel economy of the vehicles they sell
rather than shifting the product mix, because reducing the size of the vehicle is generally a
less viable compliance strategy given  that smaller vehicles have more stringent regulatory
targets.
2.2 What attribute are the agencies proposing to use, and why?

       As in the MYs 2012-2016 CAFE/GHG rules, and as NHTSA did in the MY 2011
CAFE rule, NHTSA and EPA are proposing to set CAFE and CC>2 standards that are based on
vehicle footprint, which has an observable correlation to fuel economy and emissions. There
are several policy and technical reasons why NHTSA and EPA believe that footprint is the
most appropriate attribute on which to base the standards, even though some other vehicle
attributes (notably curb weight) are better correlated to fuel economy and emissions.

       First, in the agencies' judgment, from the standpoint of vehicle safety, it is important
that the CAFE and CC>2 standards be set in a way that does not encourage manufacturers to
respond by selling vehicles that are in any way less safe. While NHTSA's  research of
historical crash data also indicates that reductions in vehicle mass that are accompanied by
reductions in vehicle footprint tend to compromise vehicle safety, footprint-based standards
provide an incentive to use advanced lightweight materials and structures that would be
discouraged by weight-based standards, because manufacturers can use them to improve a
vehicle's fuel economy and COi emissions without their use necessarily resulting in a change
in the vehicle's fuel economy and emissions targets.
c Assuming that the attribute is related to vehicle size.
d Id. at 4-5, finding 10.
                                             2-2

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

       Further, although we recognize that weight is better correlated with fuel economy and
   i emissions than is footprint, we continue to believe that there is less risk of "gaming"
(changing the attribute(s) to achieve a more favorable target)  by increasing footprint under
footprint-based standards than by increasing vehicle mass under weight-based standards—it is
relatively easy for a manufacturer to add enough weight to a vehicle to decrease its applicable
fuel economy target a significant amount, as compared to increasing vehicle footprint. We
also continue to agree with concerns raised in 2008 by some commenters on the MY 2011
CAFE rulemaking that there would be greater potential for gaming under multi-attribute
standards, such as those that also depend on weight, torque, power, towing capability, and/or
off-road capability.  The agencies agree with the assessment first presented in NHTSA's MY
2011 CAFE final rule4 that the possibility of gaming is lowest with footprint-based standards,
as opposed to weight-based or multi-attribute-based standards.  Specifically, standards that
incorporate weight, torque, power, towing capability, and/or off-road capability in addition to
footprint would not only be more complex, but by providing degrees of freedom with respect
to more easily-adjusted attributes, they could make it less certain that the future fleet would
actually achieve the average fuel economy and CC>2 reduction levels projected by the
agencies.

       The agencies recognize that based on economic and consumer demand factors that are
external to this rule, the distribution of footprints in the future may be different (either smaller
or larger) than what is projected in this rule. However, the agencies continue to believe that
there will not be significant shifts in this distribution as a direct consequence of this proposed
rule. The agencies also recognize that some international attribute-based standards use
attributes other than footprint and that there could be benefits for a number of manufacturers
if there was greater international harmonization of fuel economy and GHG standards for light-
duty vehicles, but this is largely a question of how stringent standards are and how they are
tested and enforced.  It is entirely possible that footprint-based and weight-based systems can
coexist internationally and not present an undue burden for manufacturers if they are carefully
crafted. Different countries or regions may find different attributes appropriate for basing
standards, depending on the particular challenges they face—from fuel prices, to family size
and land use, to safety concerns, to fleet composition and consumer preference, to other
environmental challenges besides climate change.  The agencies anticipate working more
closely with other countries and regions in the future to consider how to address these issues
in a way that least burdens manufacturers while respecting each country's need to meet its
own particular challenges.

       The agencies continue to find that footprint is the most appropriate attribute upon
which to base the proposed standards, but recognizing strong public interest in this issue, we
seek comment on whether the agencies should consider setting  standards  for the final rule
based on another attribute or another combination of attributes. If commenters suggest that
the agencies should consider another attribute or another combination of attributes, the
agencies specifically request that the commenters address the concerns raised in the
paragraphs above regarding the use of other attributes, and explain how standards  should be
developed using the other attribute(s) in a way that contributes more to fuel savings and
reductions than the footprint-based standards, without compromising safety.

                                             2-3

-------
                       What are the Attribute-Based Curves the Agencies are Proposing



2.3 What mathematical functions have the agencies previously used, and why?

       2.3.1   NHTSA in MY 2008 and MY 2011 CAFE (constrained logistic)

       For the MY 2011 CAFE rule, NHTSA estimated fuel economy levels after
normalization for differences in technology, but did not make adjustments to reflect other
vehicle attributes (e.g., power-to-weight ratios).6 Starting with the technology adjusted
passenger car and light truck fleets, NHTSA used minimum absolute deviation (MAD)
regression without sales weighting to fit a logistic form as a starting point to develop
mathematical functions defining the standards.  NHTSA then identified footprints at which to
apply minimum and maximum values (rather than letting the standards extend without limit)
and transposed these functions vertically (i.e., on a gpm basis, uniformly downward) to
produce the promulgated standards. In the preceding rule, for MYs 2008-2011 light truck
standards, NHTSA examined a range of potential functional forms, and concluded that,
compared to other considered forms, the  constrained logistic form provided the expected and
appropriate trend (decreasing fuel economy as footprint increases), but avoided creating
"kinks" the agency was concerned would provide distortionary incentives for vehicles with
neighboring footprints/

       2.3.2   MYs 2012-2016 Light Duty GHG/CAFE (constrained/piecewise linear)

       For the MYs 2012-2016 rules, NHTSA and EPA re-evaluated potential methods for
specifying mathematical functions to define fuel economy and GHG standards. The agencies
concluded that the constrained logistic form, if applied to post-MY 2011 standards, would
likely contain a steep mid-section that would provide undue incentive to increase the footprint
of midsize passenger cars.5 The agencies judged that a range of methods to fit the curves
would be reasonable, and used a minimum absolute deviation (MAD) regression without sales
weighting on a technology-adjusted car and light truck fleet to fit a linear equation.  This
equation was used as a starting point to develop mathematical functions defining the standards
as discussed above.  The agencies then identified footprints at which to apply minimum and
maximum values (rather than letting the standards extend without limit) and transposed these
constrained/piecewise linear functions vertically (i.e., on a gpm or COi basis, uniformly
downward) to produce the fleetwide fuel economy and CC>2 emission levels for cars and light
trucks described in the final rule.6
e See 74 FR 14196, 14363-14370 (Mar. 30, 2009) for NHTSA discussion of curve fitting in the MY 2011 CAFE
final rule.
f See 71 FR 17556, 17609-17613 (Apr. 6, 2006) for NHTSA discussion of "kinks" in the MYs 2008-2011 light
truck CAFE final rule (there described as "edge effects").  A "kink," as used here, is a portion of the curve where
a small change in footprint results in a disproportionally large change in stringency.

                                             2-4

-------
                        What are the Attribute-Based Curves the Agencies are Proposing

       2.3.3  How have the agencies changed the mathematical functions for the
      proposed MYs 2017-2025 standards, and why?

       By requiring NHTSA to set CAFE standards that are attribute-based and defined by a
mathematical function, Congress appears to have wanted the post-EISA standards to be data-
driven - a mathematical function defining the standards, in order to be "attribute-based,"
should reflect the observed relationship in the data between the attribute chosen and fuel
economy.8  EPA is also proposing to set attribute-based CO2 standards defined by similar
mathematical functions,  for the reasonable technical and policy grounds discussed below and
in section II of the preamble to the proposed rule, and which supports a harmonization with
the CAFE standards.

       The relationship between fuel economy (and GHG emissions) and footprint, though
directionally clear (i.e., fuel economy tends to decrease and CC>2 emissions tend to increase
with increasing footprint), is theoretically vague  and quantitatively uncertain; in other words,
not so precise as to a priori yield only a  single possible curve.h There is thus a range of
legitimate options open to the agencies in developing curve shapes. The agencies may of
course consider statutory objectives in choosing among  the many reasonable alternatives. For
example, curve shapes that might have some theoretical basis could lead to perverse  outcomes
contrary to the intent of the statutes to conserve energy and protect human health  and the
environment.1 Thus, the decision of how to set the target curves cannot always be just about
most "clearly" using a mathematical function to define the relationship between fuel economy
and the attribute; it often has to have a normative aspect, where the agencies adjust the
function that would define the relationship in order to avoid perverse results, improve equity
of burden across manufacturers, preserve consumer choice, etc. This is true both  for the
decisions that guide the mathematical function defining  the sloped portion of the target
curves,  and for the separate decisions that  guide the agencies' choice of  "cutpoints" (if any)
that define the fuel economy/COi levels  and footprints at each end of the curves where the
curves become flat.  Data informs these decisions, but how the agencies define and interpret
the relevant data, and then the choice of  methodology for fitting a curve to the data, must
include a consideration of both technical data and policy goals.
g A mathematical function can be defined, of course, that has nothing to do with the relationship between fuel
economy and the chosen attribute - the most basic example is an industry-wide standard defined as the
mathematical function average required fuel economy = X, where X is the single mpg level set by the agency.
Yet a standard that is simply defined as a mathematical function that is not tied to the attribute(s) would not meet
the requirement of EISA.
h In fact, numerous manufacturers have confidentially shared with the agencies what they describe as "physics
based" curves, with each OEM showing significantly different shapes, and footprint relationships. The sheer
variety of curves shown to the agencies further confirm the lack of an underlying principle of "fundamental
physics" driving the relationship between  CO2 emission or fuel consumption and footprint, and the lack of an
underlying principle to dictate any outcome of the agencies' establishment of footprint-based standards.
1 For example, if the agencies set weight-based  standards defined by a steep function, the standards might
encourage manufacturers to keep adding weight to their vehicles to obtain less  stringent targets.

                                               2-5

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

       The next sections examine the policy concerns that the agencies considered in
developing the proposed target curves that define the proposed MYs 2017-2025 CAFE and
CO2 standards, new technical work (expanding on similar analyses performed by NHTSA
when the agency proposed MY 2011-2015 standards, and by both agencies during
consideration of options for MY 2012-2016 CAFE and GHG standards) that was completed in
the process of reexamining potential mathematical functions, how the agencies have defined
the data, and how the agencies explored statistical curve-fitting methodologies in order to
arrive at proposed curves.

2.4 What are the agencies proposing for the MYs 2017-2025 curves?

       The proposed mathematical functions for the proposed MYs 2017-2025 standards are
somewhat changed from the functions for the MYs 2012-2016 standards, in response to
comments received from stakeholders and in order to address technical concerns and policy
goals that the agencies judge more significant in this 9-year rulemaking than in the prior one,
which only included 5 years. This section (2.4) discusses the methodology the agencies
selected as, at this time, best addressing those technical concerns and policy goals, given the
various technical inputs to the agencies' current analyses. Section 2.5 discusses how the
agencies determined the cutpoints and the flat portions of the MYs 2017-2025 target curves.
We also note that both of these sections  address only how the target curves were fit to fuel
consumption and CO2 emission values determined using the city and highway test procedures,
and that in determining respective regulatory alternatives, the agencies made further
adjustments to the resultant curves in order to account for adjustments for improvements to
mobile air conditioners.

       Thus, recognizing that there are many reasonable statistical methods for fitting curves
to data points that define vehicles in terms of footprint and fuel economy, the agencies have
chosen for this proposed rule to fit curves using an ordinary least-squares formulation, on
sales-weighted data, using a fleet that has had technology applied, and after adjusting the data
for the effects of weight-to-footprint, as described below. This represents a departure from
the statistical approach for fitting the curves in MYs 2012-2016, as explained in the next
section. The  agencies considered a wide variety of reasonable statistical methods in order to
better understand the range of uncertainty regarding the relationship between fuel
consumption (the inverse of fuel economy), CO2 emission rates, and footprint, thereby
providing a range within which decisions about standards would be potentially supportable.

       2.4.1  What concerns were the agencies looking to address that led them to
      change from the approach used for the MYs 2012-2016 curves?

       During the year and a half since the MYs 2012-2016 final rule was issued, NHTSA
and EPA have received a number of comments from stakeholders on how curves should be
fitted to the passenger car and light truck fleets.  Some limited-line manufacturers have argued
that curves should generally be flatter in order to avoid discouraging small vehicles, because
steeper curves tend to result in more stringent targets for smaller vehicles. Most full-line
manufacturers have argued that a passenger car curve similar in slope to the MY 2016


                                            2-6

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

passenger car curve would be appropriate for future model years, but that the light truck curve
should be revised to be less difficult for manufacturers selling the largest full-size pickup
trucks. These manufacturers argued that the MY 2016 light truck curve was not "physics-
based," and that in order for future tightening of standards to be feasible for full-line
manufacturers, the truck curve for later model years should be steeper and extended further
(i.e., made less stringent) into the larger footprints. The agencies also do not agree that the
MY 2016 light truck curve was somehow deficient in lacking a "physics basis," or that it was
somehow overly stringent for manufacturers selling large pickups—manufacturers making
these arguments presented no "physics-based" model to explain how fuel economy should
depend on footprint^ The same manufacturers indicated that they believed that the light truck
standard  should be somewhat steeper after MY 2016, primarily because, after more than ten
years of progressive increases in the stringency of applicable CAFE standards, large pickups
would be less capable of achieving further improvements without compromising load carrying
and towing capacity.

       In developing the curve shapes for this proposed rule, the agencies were aware of the
current and prior technical concerns raised by OEMs concerning the effects of the stringency
on individual manufacturers and their ability to meet the standards with available
technologies, while producing vehicles at a cost that allowed them to recover the additional
costs of the technologies being applied. Although we continue to believe that the
methodology for fitting curves for the MY2012-2016 standards was technically sound, we
recognize manufacturers' technical concerns regarding their abilities to comply with a
similarly shallow curve after MY2016 given the anticipated mix of light trucks in MYs 2017-
2025. As in the MYs 2012-2016 rules, the agencies considered these concerns in the analysis
of potential curve shapes. The agencies also considered safety concerns which could be
raised by curve shapes creating an incentive for vehicle downsizing, as well as the potential
loss to consumer welfare should vehicle upsizing be unduly disincentivized.  In addition, the
agencies  sought to improve the balance of compliance burdens among manufacturers. Among
the technical concerns and resultant policy trade-offs the agencies considered were the
following:
       Flatter standards (i.e., curves) increase the risk that both the weight and size of
       vehicles will be reduced, compromising highway safety.
       Flatter standards potentially impact the utility of vehicles by providing an incentive for
       vehicle downsizing.
       Steeper footprint-based standards may incentivize vehicle upsizing, thus increasing the
       risk that fuel economy and greenhouse gas reduction benefits will be less than
       expected.
       Given the same industry-wide average required fuel economy or CCh standard, flatter
       standards tend to place greater compliance burdens on full-line manufacturers
 See footnote h.

                                             2-7

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

   •   Given the same industry-wide average required fuel economy or CO2 standard, steeper
       standards tend to place greater compliance burdens on limited-line manufacturers
       (depending of course, on which vehicles are being produced).
   •   If cutpoints are adopted, given the same industry-wide average required fuel economy,
       moving small-vehicle cutpoints to the left (i.e., up in terms of fuel economy, down in
       terms of CC>2 emissions) discourages the introduction of small vehicles, and reduces
       the incentive to downsize small vehicles in ways that would compromise highway
       safety.
   •   If cutpoints are adopted, given the same industry-wide average required fuel economy,
       moving large-vehicle cutpoints to the right (i.e., down in terms of fuel economy, up in
       terms of COi emissions) better accommodates the unique design requirements of
       larger vehicles—especially large  pickups—and extends the size range over which
       downsizing is discouraged.

       All of these were policy goals that required trade-offs, and in determining the curves
they also required balance against the comments from the OEM comments discussed in the
introduction to this section. Ultimately, the agencies do not agree that the MY 2017 target
curves for this proposal, on a relative basis, should be made significantly flatter than the MY
2016 curve,k as we believe that this would undo some of the safety-related incentives and
balancing of compliance burdens among manufacturers—effects  that attribute-based
standards are intended to provide.

       Nonetheless, the agencies recognize full-line OEM concerns and have tentatively
concluded that further increases in the stringency of the light truck standards will be more
feasible if the light truck curve is made steeper than the MY 2016 truck curve and the right
(large footprint) cut-point is extended over time to larger footprints.  This conclusion is
supported by the agencies' technical analyses of regulatory alternatives defined using the
curves developed in the manner described below.

       2.4.2  What methodologies and data did the agencies  consider in developing the
              2017-2025 curves?

       In considering how to address the various policy concerns discussed in the previous
sections, the agencies revisited the data and performed  a number  of analyses using different
combinations of the various statistical methods, weighting schemes, adjustments to the data
and the addition of technologies to make the fleets less  technologically heterogeneous. As
discussed in 2.3.3, in the agencies' judgment, there is no single "correct" way to estimate the
relationship  between COi or fuel consumption and footprint - rather, each statistical result is
based on the underlying assumptions about the particular functional form, weightings and
error structures embodied in the representational approach. These assumptions are the subject
k While "significantly" flatter is subjective, the year over year change in curve shapes is discussed in greater
detail in Section 2.5.3.1.

                                             2-8

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

of the following discussion. This process of performing many analyses using combinations of
statistical methods generates many possible outcomes, each embodying different potentially
reasonable combinations of assumptions and each thus reflective of the data as viewed
through a particular lens. The choice of a standard developed by a given combination of these
statistical methods is consequently a decision based upon the agencies' determination of how,
given the policy objectives for this rulemaking and the agencies' MY 2008-based forecast of
the market through MY 2025, to appropriately reflect the current understanding of the
evolution of automotive technology and costs, the future prospects for the vehicle market, and
thereby establish curves (i.e., standards) for cars and light trucks.

       2.4.2.1 What information did the agencies use to estimate a relationship between
              fuel economy, COi and footprint?

       For each fleet, the agencies began with the MY 2008-based market forecast developed
to support this proposal (i.e., the baseline fleet), with vehicles' fuel economy levels and
technological characteristics at MY 2008 levels.1 The development,  scope, and content of this
market forecast is discussed in  detail in Chapter 1 of the joint Technical Support Document
supporting this rulemaking.

       Figure 2-1 shows the MY 2008 COiby car and truck class as it exists in the EPA
OMEGA and NHTSA CAFE model data files (for a gasoline-only fleet, fuel consumption—
the inverse of fuel economy—is directly proportional to CCh). This  dataset is the base fleet
which is the starting point for all analysis in this proposal.
1 While the agencies jointly conducted this analysis, the coefficients ultimately used in the slope setting analysis
are from the CAFE model.

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
                                    2008 CO2 v Footprint







CN
O
0
CO
o
0












f
t



i




f .

• 'i s
0

iff*





6

c


% %



T'"
s-i •
"»




















































•
**.
..£,
>"\


.•fZ
".ff>
A'

a •
•


T
•

* **• *1+
* !°:-°'-
,'•$***
**-"•;•-

« ^ •
8








.-:.
Vf't?
Js. 4°













. s.















2008 S.iles
0

o 50000
0 100000
O 150000

O 200000
O 250000


                           40   50   60    70      40   50   60    70
                                         Footprint
                      Figure 2-1 2008 CO2 vs. Footprint by Car and Truck
         2.4.2.1      What adjustments did the agencies evaluate?

    The agencies believe one possible approach is to fit curves to the minimally adjusted data
shown above (the approach still includes sales mix adjustments, which influence results of
sales-weighted regressions), much as DOT did when it first began evaluating potential
attribute-based standards in 2003.7  However, the agencies  have found, as in prior
rulemakings, that the data are so widely spread (i.e., when graphed, they fall in a loose
"cloud" rather than tightly around an obvious line) that they indicate a relationship between
footprint and COi and fuel consumption that is real but not particularly strong (Figure 2-1).
Therefore, as discussed below, the agencies also explored possible adjustments that could
help to explain and/or reduce the ambiguity of this relationship, or could help to produce
policy outcomes the agencies judged to be more desirable.
              2.4.2.1.1
Adjustment to reflect differences in technology
    As in prior rulemakings, the agencies consider technology differences between vehicle
models to be a significant factor producing uncertainty regarding the relationship between
COi/fuel consumption and footprint.  Noting that attribute-based standards are intended to
encourage the application of additional technology to improve fuel efficiency and reduce COi
                                             2-10

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

emissions, the agencies, in addition to considering approaches based on the unadjusted
engineering characteristics of MY 2008 vehicle models, therefore also considered approaches
in which, as for previous rulemakings, technology is added to vehicles for purposes of the
curve fitting analysis in order to produce fleets that are less varied in technology content.

    The agencies adjusted the baseline fleet for technology by adding all technologies
considered, except for the most advanced high-BMEP (brake mean effective pressure)
gasoline engines, diesel engines, strong HEVs, PHEVs, EVs, and FCVs. The agencies
included 15 percent mass reduction on all vehicles. Figure 2-2 shows the same fleet, with
technology adjustment and 2021 sales applied, and the baseline diesel fueled vehicles, HEV
and EVs removed from the fleet. Of note, the fleet is now more closely clustered"1 (and lower
in emissions), but the same basic pattern emerges; in both figures, the COi emission rate
(which, as  mentioned above, is directly proportional to fuel consumption for a gasoline-only
fleet) increases with increasing footprint, although the relationship is less pronounced for
larger light trucks.

                           Max  ICE Tech - CO2 v. Footprint
                  500 -
                  409 -
                 8
                 o
                 '§300 -
                  100 -
2021 Projected Sale


 •  50000
 • 100000
 • 150000
 • 200000
 • 250000
 • 300000
                        40  50  60   70  80  40  50   60   70  80
                                    Footprint
     Figure 2-2 2008 CO2 vs. Footprint by Car and Truck, after Adjustment Reflecting Technology
                  Differences, and removing diesel fueled vehicles, HEVs and EVs
m For cars, the standard deviation of the CO2 data is reduced from 8 1 to 54 through the technology
normalization. For trucks, the standard deviation is reduced from 62 to 36.
                                              2-11

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
         2.4.2.2      Adjustments reflecting differences in performance and "density"

       As discussed in Section 2.4.1, during stakeholder meetings the agencies held while
developing this NPRM, some manufacturers indicated that they believed that the light truck
standard should be somewhat steeper after MY 2016.  As a means to produce a steeper light
truck curve, the agencies considered adjustments for other differences between vehicle
models (i.e., inflating or deflating the fuel economy of each vehicle model based on the extent
to which one of the vehicle's attributes, such as power, is higher or lower than average).
Previously, NHTSA had rejected such adjustments because they imply that a multi-attribute
standard may be necessary, and the agencies judged multi-attribute standard to be more
subject to gaming than a footprint-only standard."'8 Having considered this issue again for
purposes of this rulemaking, NHTSA and EPA conclude the need to accommodate in the
target curves the challenges faced by manufacturers of large pickups currently outweighs
these prior concerns.  Therefore, the agencies also evaluated curve fitting approaches through
which fuel consumption and COi levels were adjusted with respect to weight-to-footprint
alone, and in combination with power-to-weight. While the agencies examined these
adjustments for purposes of fitting curves, the agencies are not proposing a multi-attribute
standard; the proposed fuel economy and COi targets for each vehicle are still functions of
footprint alone. No adjustment would be used in the compliance process.

       The agencies also examined some differences between the technology-adjusted car
and truck fleets in order to better understand the relationship between footprint and COi/fuel
consumption in the agencies' MY 2008 based forecast. More direct measures (such as
coefficients of drag and rolling resistance), while useful for vehicle simulation, were not
practical or readily available at the fleet level. Given this issue, and  based on analysis
published in the 2012-2016 rule,9 the agencies investigated a sales-weighted (i.e., treating
every vehicle unit sold as a separate observation) regression equation involving power to
weight ratio and vehicle weight (Equation  2-1).°  This equation provides for a strong
n For example, in comments on NHTSA's 2008 NPRM regarding MY 2011-2015 CAFE standards, Porsche
recommended that standards be defined in terms of a "Summed Weighted Attribute", wherein the fuel economy
target would calculated as follows: target =f(SWA), where target is the fuel economy target applicable to a
given vehicle model and SWA = footprint + torque1'1'5 + weight1'2'5.  (NHTSA-2008-0089-0174). While the
standards the agencies are proposing for MY 2017-2025 are not multi-attribute standards, that is the target is
only a function of footprint, we are proposing curve shapes that were developed considering more than one
attribute.
0 These parameters directly relate to the amount of energy required to move the vehicle. As compared to a
lighter vehicle, more energy is required to move a heavier vehicle the same distance. Similarly, a more powerful
engine, when technology adjusted, is less efficient than a less powerful engine.

                                              2-12

-------
                        What are the Attribute-Based Curves the Agencies are Proposing

correlation between HP/WT, weight and COi emissions (R2=0.78, Table 2-1) after accounting
for technology adjustments.13
       Equation 2-1 - Relationship between vehicle attributes and emissions or fuel consumption

                                      /Horsepowen
               C02i or GPMi = (3hp/wt (———J + (BweightWeight; + C
Where:
       HP/Weight= the rated horsepower of the vehicle divided by the curb weight
       Weight = the curb weight of the vehicle in pounds
       C = a constant.
            Table 2-1 - Physical Regression Coefficients against Technology Adjusted CO2

Rz
F-test p
Php/wt
^ weight
C
Cars
0.78
<0.01
1.09*10J
3.29*10~z
-3.29
Light Trucks
0.78
<0.01
1.13*10'
3.45*10"2
2.73
                      *In this gasoline only fleet, these coefficients can be divided by 8887 (the amount of
                      CO2 produced by the combustion of a gallon of the fuel used to certify the fuel
                      economy and emissions of gasoline vehicles) to yield the corresponding fuel
                      consumption coefficients.

       The coefficients above show, for the agencies' MY 2008-based market forecast, strong
correlation between these vehicle attributes and the fuel consumption and emissions of the
vehicle, as well as strong similarity between car and truck coefficients. Given these very
similar parameters, similar distributions of power and weight would be expected to produce
similarly arrayed plots of COi (or equivalently, fuel consumption) by footprint, regardless of
car or truck class. Based on the differences seen in the technology-adjusted plot (Figure 2-2),
the agencies further investigated these particular attributes and their relationship to footprint
in the agencies' MY 2008-based market forecast, to  examine the differences across the
footprint distribution.   Figure 2-3 shows vehicle curb weight charted against footprint, with
sales weighted ordinary least squares sales fit (blue) and sales-weighted LOESS fit (red)
p As R2 does not equal 1, there are remaining unaccounted for differences beyond technology, power and weight.
These may include gear ratios, axle ratios, aerodynamics, and other vehicle features not captured in this
equation.
                                               2-13

-------
                        What are the Attribute-Based Curves the Agencies are Proposing

imposed. For cars, the LOESS fit, which weights nearby points more heavily,q is nearly
identical to the linear fit in the data filled region between about 40 and 56 sq ft (with the gray
bar showing standard error on the Loess fit). For this market forecast, average car curb
weight is linearly proportional to car footprint between 40 and 56 sq ft, or in other words, cars
progress in weight in a regular fashion as they get larger (Figure 2-3). By contrast, a linear fit
does not overlap with the LOESS fit on the truck side, which indicates that for this market
forecast, truck curb weight does not linearly increase with footprint, at least not across the
entire truck fleet. The LOESS fit shows that larger trucks (those on the right side of the data
bend in Figure 2-2) have a different trend than smaller trucks, and after about 55 sq ft, no
longer proportionally increases in weight. The  same pattern is seen in Figure 2-1 and Figure
2-2 above.
q: "In a [LOESS] Fit, fitting is done locally. That is, for the fit at point x, the fit is made using points in a
neighborhood of x, weighted by their distance from x (with differences in 'parametric' variables being ignored
when computing the distance). The size of the neighborhood is controlled by a For a < 1, the neighborhood
includes proportion a of the points, and these have tricubic weighting (proportional to (1 - (dist/maxdist)A3)*3.
For a > 1, all points are used, with the 'maximum distance' assumed to be a^l/p times the actual maximum
distance forp explanatory variables."
A span of 1 was used in these images, http://cran.r-project.org/doc/manuals/fullrefman.pdf

                                                2-14

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
                           WT v. FP - Weighted OLS and Loess Fit
                   4000
  Figure 2-3 Relationship between Weight and Footprint in Agencies' MY2008-Based Market Forecast
       To further pursue this topic, weight divided by footprint (WT/FP) can be thought of as
a "density" of a vehicle (although dimensionally it has units of pressure). As seen in Figure
2-4, the trend in WT/FP in the agencies' MY2008-based market forecast is different in trucks
than in cars. The linear trend on cars is an increase in WT/FP as footprint increases (Figure
2-4).  In contrast, light trucks do not consistently increase in WT/FP ratio as the vehicles grow
larger, but WT/FP actually decreases (Figure 2-4).
                                              2-15

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
                              WT/FP v. FP - Weighted OLS
                   120 -
                   110 -
                   80 -
                   70 -
                   60 -
                                     70      40
                                      Footprint
 Figure 2-4 Relationship between Weight/FP and Footprint in Agencies' MY2008-Based Market Forecast
       The heterogeneity of the truck fleet explains part of the WT/FP trend, where the
pickup truck fleet is largest in footprint, but is also relatively light for its size due to the flat
bed (Figure 2-5). Note that  the two light truck classes with the smallest WT/FP ratios are
small and large pickups.  Further, as the only vehicle class with a sales-weighted average
footprint above 60 square feet, the large pickup trucks have a strong influence on the slope of
the truck curve.  As the correlation between weight and COi is strong (Table 2-1), having
proportionally lighter vehicles at one extreme of the footprint distribution can bias a curve fit
to these vehicles. If no adjustment is made to the curve fitted to the truck fleet, and no other
compensating flexibilities or adjustments are made available, manufacturers selling
significant numbers of vehicles at the large end of the truck distribution will face compliance
burdens that are comparatively more challenging that those faced by manufacturers not
serving this part of the light  truck market.  As noted further below, this consideration
underlies the agencies' proposal to change the cutpoint for larger light trucks from 66 feet to
74 feet, and to steepen the slope of the light truck curve for larger light trucks.
                                              2-16

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
                                    WT/FP by Vehicle Class
                    D_
                    LL
                    i
                                      SO     55     60
                                          Footprint
                         Figure 2-5 Class and the WT/FP distribution
       The agencies also investigated the relationship between HP/WT and footprint in the
agencies' MY2008-based market forecast (Figure 2-6). On a sales weighted basis, cars tend
to become proportionally more powerful as they get larger.  In contrast, there is a minimally
positive relationship between HP/WT and footprint for light trucks, indicating that light trucks
become only slightly more powerful as they get larger, but that the trend is not especially
pronounced.
                                             2-17

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
                                HP/WT v. FP - Weighted OLS
                                                                2021 Sales


                                                                — 50000
                                                                • 100000
                                                                   150000
I                                                                   200000
                                                                   250000
                                                                   300000
I
                         40   50   60   70     40    50
                                        Footprint
                                  Figure 2-6 HP/WT v. FP

       One factor influencing results of this analysis is the non-homogenous nature of the
truck fleet; some vehicles at the smaller end of the footprint curve are different in design and
utility from others at the larger end (leading to the observed bend in the LOESS fit, Figure
2-6).  There are many high volume four-wheel drive vehicles with smaller footprint in the
truck fleet (such as the Chevrolet Equinox, Dodge Nitro, Ford Escape, Honda CR-V, Hyundai
Santa Fe, Jeep Liberty, Nissan Rogue, Toyota RAV4, and others) exhibit only select truck
characteristics/ By contrast, the largest pickup trucks in the light truck fleet have unique
aerodynamic and power characteristics that tend to increase COi emissions and fuel
consumption.  These disparities contribute to the slopes of lines fitted to the light truck fleet.

       The agencies technical analyses of regulatory alternatives developed using curves
fitted as described below supported OEM comments that there will be significant compliance
challenges for the manufacturers of large pickup trucks, and supported the agencies' policy
goal of a steeper slope for the light truck curve. Consequently, the agencies considered
options including fitting curves developed using results of the analysis described above-
Specifically, the agencies note that the WT/FP ratio of the light duty fleet potentially has a
r In most cases, these vehicles have four wheel drive, but no significant towing capability, and no open-bed.
Many of these vehicles are also offered without four wheel drive, and these two wheel drive versions are
classified as passenger cars, not light trucks.
                                              2-18

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

large impact on a sales weighted regression.8 The increasing trend in WT/FP versus footprint
for cars in the 2008 MY baseline would steepen the slope of the car curve, while the
decreasing trend in WT/FP would flatten the truck slope, as compared to a WT/FP adjusted
fleet. This result was reflected in the MYs 2012-2016 final rulemaking,10 where the agencies
noted the steep car curves resulting from a weighted least squares analysis.

       Based on the above analysis, the agencies also considered adjustments for other
differences between vehicle models.  Therefore, utilizing the coefficients derived in Equation
2-1, the agencies also evaluated curve fitting approaches through which fuel consumption and
COi levels were adjusted with respect to weight-to-footprint alone, and in combination with
power-to-weight. This adjustment procedure inflates or deflates the fuel economy or CC>2
emissions of each vehicle model based on the extent to which one of the vehicle's attributes,
such as power, is higher or lower than average. As mentioned above, while the agencies
considered this technique for purposes of fitting curves, the agencies are not proposing a
multi-attribute standard, as the proposed fuel economy and COi targets for each vehicle are
still functions of footprint alone. No adjustment would be used in the compliance process.

       The basis for the gallon-per-mile (GPM) adjustments is the sales-weighted linear
regression discussed in 2.4 (Equation 2-1, Table 2-1). The coefficients to this equation give
the impact of the various car attributes on COi emissions and fuel consumption in the
agencies' MY 2008-based market forecast.  For example, (Bweight gives the impact of weight
while holding the ratio horsepower to weight constant. Importantly, this means that as weight
changes, horsepower must change as well to keep the power/weight ratio constant.  Similarly,
Php/wt gives the COi impact of changing the performance of the vehicle while keeping the
weight constant. These coefficients  were used to perform  an adjustment of the gallons per
mile measure for each vehicle to the respective car or truck—i.e., in the case of a HP/WT
adjustment, to deflate or inflate the fuel consumption of each vehicle model based on the
extent to which the vehicle's power-to-weight ratio is above or below the regression-based
value at that footprint.

       The agencies performed this  normalization to adjust for differences in  vehicle weight
per square foot observations in the data discussed in Section 2.4. This adjustment process
requires two pieces of information:  the weight coefficient from Equation 2-1  and the average
weight per footprint (i.e., pounds per square foot) for that vehicle's group.  Two groups,
passenger cars and light trucks, were used. For each group, the average weight per footprint
was calculated as a weighted average with the weight being the same as in the above
regression (projected sales by vehicle in 2021). The equation below indicates how this
adjustment was carried out.
s As mentioned above, the agencies also performed the same analysis without sales weighting, and found that the
WT/FP ratio also had a directionally similar effect on the fitted car and truck curves.

                                             2-19

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
                              Equation 2-2 WT/FP adjustment
                                                           Weight            \
  Weight per Footprint Adjusted GPM; or C02i = GPM; - Weight;	x Footprint;  x (Bweieht
                                                y         Footprint           j      B

       The term in parentheses represents the vehicle's deviation from an "expected weight."
That is, multiplying the average weight per footprint for a group of vehicles (cars or trucks)
by a specific vehicle's footprint gives an estimate of the weight of that specific vehicle if it's
density were "average," based on the MY 2008 fleet.  Put another way, this factor represents
what the weight is "expected" to be, given the vehicle's footprint, and based on the MY 2008
fleet. This "expected weight" is then subtracted from the vehicle's actual weight.  Vehicles
that are heavier than their "expected weight" will receive a positive value (i.e., a deflated fuel
economy value) here, while vehicles that are lighter than their "expected weight" will receive
a negative number (i.e., an inflated fuel economy value).

       This deviation from "expected weight" is then converted to a gallon value by the
regression coefficient. The units on this coefficient are gallons per mile per pound, as can be
deduced from equation 1. This value is then subtracted from the vehicle's actual gallons per
mile measure. Note that the adjusted truck data no longer exhibits the bend seen in Figure 2-1
and Figure 2-2.
                40      50      60      70     80    40      50      60      70     80
                   Figure 2-7 WT/FP Adjusted Fuel Consumption vs. Footprint

                                             2-20

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
       This adjustment serves to reduce the variation in gallons per mile measures caused by
variation in weight in the agencies' MY 2008-based market forecast. Importantly, this
adjustment serves to reduce the fuel consumption (i.e., inflate fuel economy) for those
vehicles which are heavier than their footprint would suggest while increasing the gallons per
mile measure (i.e., deflating fuel economy) for those vehicles which are lighter.  For trucks, a
linear trend is more evident in the data cloud.1 The following table shows the degree of
adjustment for several vehicle models:
' Using EPA's dataset, R2 for the sales weighted ordinary least squared linear fit between footprint and CO2
improved from 0.38 (technology adjusted CO2) to 0.64 (technology and weight / footprint adjusted CO2)

                                              2-21

-------
     What are the Attribute-Based Curves the Agencies are Proposing




Table 2-2 - Sample Adjustments for Weight to Footprint, Cars
Manufacturer
HONDA
TOYOTA
FORD
GENERAL
MOTORS
HONDA
NISSAN
GENERAL
MOTORS
FORD
TOYOTA
VOLKSWAGEN
FORD
HONDA
HYUNDAI
HONDA
Model
HONDA FIT
TOYOTA
COROLLA
FORD FOCUS
CHEVROLET
MALIBU
HONDA
ACCORD
INFINITIG37
CHEVROLET
CORVETTE
FORD
MUSTANG
TOYOTA
CAMRY
VOLKSWAGEN
JETTA
FORD FUSION
HONDA
ACCORD
HYUNDAI
SONATA
HONDA CIVIC
Name Plate
FIT
COROLLA
FOCUS FWD
MALIBU
ACCORD 4DR
SEDAN
G37 COUPE
CORVETTE
MUSTANG
CAMRY
SOLARA
CONVERTIBLE
JETTA
FUSION FWD
ACCORD 2DR
COUPE
SONATA
CIVIC
Weight/
Footprint
64.4
61.3
62.9
73.5
69.6
76.7
69.3
74.7
75.6
78.0
72.2
71.6
70.7
59.9
Footprint
39.5
42.5
41.7
46.9
46.6
47.6
46.3
46.7
46.9
42.4
46.1
46.6
46.0
43.2
GPM
0.01
0.01
0.02
0.02
0.02
0.02
0.02
0.03
0.02
0.02
0.02
0.02
0.02
0.02
MPG
69.40
69.94
61.94
53.70
57.57
47.83
40.84
31.32
50.87
46.77
59.96
56.92
61.72
64.25
Adjusted
GPM
0.0157
0.0164
0.0177
0.0185
0.0179
0.0200
0.0251
0.0316
0.0191
0.0211
0.0168
0.0178
0.0166
0.0177
Adjusted
MPG
63.73
60.80
56.34
54.08
55.73
50.08
39.83
31.67
52.27
47.47
59.61
56.26
60.34
56.38
GPM %
Adjustment
8.9%
15.0%
9.9%
-0.7%
3.3%
-4.5%
2.5%
-1.1%
-2.7%
-1.5%
0.6%
1.2%
2.3%
14.0%
                          2-22

-------
                      What are the Attribute-Based Curves the Agencies are Proposing


                Table 2-3 - Sample Adjustments for Weight to Footprint, Trucks
Manufacturer
FORD
GENERAL
MOTORS
FIAT
HONDA
TOYOTA
FORD
FIAT
TOYOTA
TATA
GENERAL
MOTORS
GENERAL
MOTORS
GENERAL
MOTORS
TOYOTA
Model
FORD ESCAPE
CHEVROLET
CIS
JEEP GRAND
CHEROKEE
HONDA PILOT
TOYOTA
HIGHLANDER
FORD F150
DODGE RAM
TUNDRA
LAND ROVER
RANGE
ROVER SPORT
CHEVROLET
UPLANDER
HUMMERH3
PONTIAC
TORRENT
TACOMA
Name Plate
ESCAPE FWD
CIS
SILVERADO
2WD 119WB
GRAND
CHEROKEE
4WD
PILOT 4WD
HIGHLANDER
4WD
F150 FFV
4WD145WB
RAM 1500
PICKUP 4WD
140 WB
TOYOTA
TUNDRA
4WD145WB
RANGE
ROVER
SPORT
UPLANDER
FWD
H34WD
TORRENT
FWD
TOYOTA
TACOMA
4WD
Weight/
Footprint
80.1
85.9
103.7
85.2
79.6
73.8
78.1
79.3
118.6
114.4
99.9
84.2
74.8
Footprint
65.2
55.9
47.1
51.3
49.0
67.4
66.3
68.7
47.5
49.2
50.7
48.2
53.4
GPM
0.02
0.03
0.02
0.02
0.02
0.03
0.03
0.03
0.03
0.02
0.03
0.02
0.02
MPG
51.00
39.76
41.45
40.95
45.90
32.70
33.75
32.07
33.17
45.46
36.71
46.64
43.01
Adjusted
GPM
0.0181
0.0248
0.0222
0.0243
0.0227
0.0334
0.0316
0.0325
0.0239
0.0163
0.0242
0.0215
0.0252
Adjusted
MPG
55.11
40.29
44.98
41.22
44.05
29.97
31.65
30.73
41.92
61.34
41.30
46.56
39.63
GPM %
Adjustment
-7.5%
-1.3%
-7.9%
-0.6%
4.2%
9.1%
6.6%
4.3%
-20.9%
-25.9%
-11.1%
0.2%
8.5%
       Based on Equation 2-1, the agencies also evaluated an adjustment of GPM and CC>2

based on HP/WT.

                         Equation 2-3 -Adjustment based on HP/WT
                   HP                             HPj   HP
                   -—adjusted GPM; or C02; = GPM; - (— - —) x (BHP/WT
                   WT
"WT;   WT
       Figure 2-8 shows the adjusted data and the estimated relationship between the adjusted
GPM values and footprint.
                                            2-23

-------
What are the Attribute-Based Curves the Agencies are Proposing
                    2-24

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

       Table 2-4 shows the degree of adjustment for several vehicle models. Those vehicles
which have more power than average for their actual curb weight are adjusted downward (i.e.,
fuel economy ratings are inflated), while those that have less power than average are adjusted
upward (i.e., fuel economy ratings are deflated).
                        50
60
70     80    40
    Footprint
50
60
70
80
                  Figure 2-8 HP/WT Adjusted Fuel Consumption v. Footprint
                                            2-25

-------
       What are the Attribute-Based Curves the Agencies are Proposing




 Table 2-4 - Sample Adjustments for Horsepower to Weight, Cars
Manufacturer
HONDA
TOYOTA
FORD
GENERAL
MOTORS
HONDA
NISSAN
GENERAL
MOTORS
FORD
TOYOTA
VOLKSWAGEN
FORD
HONDA
HYUNDAI
HONDA
Model
HONDA FIT
TOYOTA
COROLLA
FORD FOCUS
CHEVROLET
MALIBU
HONDA
ACCORD
INFINITIG37
CHEVROLET
CORVETTE
FORD
MUSTANG
TOYOTA
CAMRY
VOLKSWAGEN
JETTA
FORD FUSION
HONDA
ACCORD
HYUNDAI
SONATA
HONDA CIVIC
Name Plate
FIT
COROLLA
FOCUS FWD
MALIBU
ACCORD 4DR
SEDAN
G37 COUPE
CORVETTE
MUSTANG
CAMRYSOLARA
CONVERTIBLE
JETTA
FUSION FWD
ACCORD 2DR
COUPE
SONATA
CIVIC
Horsepower
109
126
140
169
190
330
400
500
225
170
160
190
162
140
Footprint
39.5
42.5
41.7
46.9
46.6
47.6
46.3
46.7
46.9
42.4
46.1
46.6
46.0
43.2
GPM
0.01
0.01
0.02
0.02
0.02
0.02
0.02
0.03
0.02
0.02
0.02
0.02
0.02
0.02
MPG
69.40
69.94
61.94
53.70
57.57
47.83
40.84
31.32
50.87
46.77
59.96
56.92
61.72
64.25
Adjusted
GPM
0.0157
0.0164
0.0177
0.0185
0.0179
0.0200
0.0251
0.0316
0.0191
0.0211
0.0168
0.0178
0.0166
0.0177
Adjusted
MPG
63.73
60.80
56.34
54.08
55.73
50.08
39.83
31.67
52.27
47.47
59.61
56.26
60.34
56.38
GPM %
Adjustment
8.9%
15.0%
9.9%
-0.7%
3.3%
-4.5%
2.5%
-1.1%
-2.7%
-1.5%
0.6%
1.2%
2.3%
14.0%
Table 2-5 - Sample Adjustments for Horsepower to Weight, Trucks
Manufacturer
FORD
GENERAL
MOTORS
FIAT
HONDA
TOYOTA
FORD
FIAT
TOYOTA
TATA
GENERAL
MOTORS
GENERAL
MOTORS
GENERAL
MOTORS
TOYOTA
Model
FORD ESCAPE
CHEVROLET
CIS
JEEP GRAND
CHEROKEE
HONDA PILOT
TOYOTA
HIGHLANDER
FORD F150
DODGE RAM
TUNDRA
LAND ROVER
RANGE ROVER
SPORT
CHEVROLET
UPLANDER
HUMMER H3
PONTIAC
TORRENT
TACOMA
Name Plate
ESCAPE FWD
CIS SILVERADO
2WD 119WB
GRAND CHEROKEE
4WD
PILOT 4WD
HIGHLANDER
4WD
F150 FFV 4WD
145 WB
RAM 1500 PICKUP
4WD140WB
TOYOTA TUNDRA
4WD145WB
RANGE ROVER
SPORT
UPLANDER FWD
H34WD
TORRENT FWD
TOYOTA TACOMA
4WD
Horsepower
153
195
210
244
270
300
345
381
300
240
242
185
236
Footprint
65.2
55.9
47.1
51.3
49.0
67.4
66.3
68.7
47.5
49.2
50.7
48.2
53.4
GPM
0.02
0.03
0.02
0.02
0.02
0.03
0.03
0.03
0.03
0.02
0.03
0.02
0.02
MPG
51.00
39.76
41.45
40.95
45.90
32.70
33.75
32.07
33.17
45.46
36.71
46.64
43.01
Adjusted
GPM
0.0181
0.0248
0.0222
0.0243
0.0227
0.0334
0.0316
0.0325
0.0239
0.0163
0.0242
0.0215
0.0252
Adjusted
MPG
55.11
40.29
44.98
41.22
44.05
29.97
31.65
30.73
41.92
61.34
41.30
46.56
39.63
GPM%
Adjustment
-7.5%
-1.3%
-7.9%
-0.6%
4.2%
9.1%
6.6%
4.3%
-20.9%
-25.9%
-11.1%
0.2%
8.5%
                            2-26

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

       The agencies seek comment on the appropriateness of these adjustments, particularly
regarding whether these adjustments suggest that standards should be defined in terms of
other attributes in addition to footprint, and whether they may encourage changes other than
encouraging the application of technology to improve fuel economy and reduce COi
emissions.  The agencies also seek comment regarding whether these adjustments effectively
"lock in" through MY 2025 relationships that were observed in MY 2008.

       The above approaches resulted in three data sets each for (a) vehicles without added
technology and (b) vehicles with technology added to reduce technology differences, any of
which may provide a reasonable basis for fitting mathematical functions upon which to base
the slope of the standard curves: (1) vehicles without any further adjustments; (2) vehicles
with adjustments reflecting differences in "density" (weight/footprint); and (3) vehicles with
adjustments reflecting differences in "density," and adjustments reflecting differences in
performance (power/weight).
         2.4.2.3     What statistical methods did the agencies evaluate?

       Using these data sets, the agencies tested a range of regression methodologies, each
judged to be possibly reasonable for application to at least some of these data sets.

2.4.2.3.1  Regression Approach

       In the MYs 2012-2016 final rules, the agencies employed a robust regression approach
(minimum absolute deviation, or MAD), rather than an ordinary least squares (OLS)
regression.11  MAD is generally applied to mitigate the effect of outliers in a dataset, and thus
was employed in that rulemaking as part of our interest in attempting to best represent the
underlying technology.   NHTSA had used OLS in early development of attribute-based
CAFE standards, but NHTSA (and then NHTSA and EPA) subsequently chose MAD instead
of OLS for both the MY 2011 and the MYs 2012-2016 rulemakings. These decisions on
                                                                               -i (-J
regression technique were made both because OLS gives additional emphasis to outliers   and
because the MAD approach helped  achieve the agencies' policy goals with regard to curve
slope in those rulemakings.13 In the interest of taking a fresh look  at appropriate regression
methodologies as promised in the 2012-2016 light duty rulemaking, in developing this
proposal, the agencies gave full consideration to both OLS and MAD. The OLS
representation, as described, uses squared errors, while MAD employs absolute errors and
thus weights outliers less.

       As noted, one of the reasons stated for choosing MAD over least square regression in
the MYs 2012-2016 rulemaking was that MAD reduced the weight placed on outliers in the
data.  As seen in Figure 2-1, there clearly are some outliers in the data, mostly to the high COi
and fuel consumption side. However, the agencies have further considered whether it is
appropriate to classify these vehicles as outliers. Unlike in traditional datasets, these vehicles'
performance is not mischaracterized due to errors in their measurement, a common reason for
                                            2-27

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

outlier classification. Being certification data, the chances of large measurement errors
should be near zero, particularly towards high CC>2 or fuel consumption. Thus, they can only
be outliers in the sense that the vehicle designs are unlike those of other vehicles. These
outlier vehicles may include performance vehicles, vehicles with high ground clearance,
4WD, or boxy designs.  Given that these are equally legitimate on-road vehicle designs, the
agencies concluded that it would appropriate to reconsider the treatment of these vehicles in
the regression techniques.

      Based on these considerations as well as the adjustments discussed above, the agencies
concluded it was not meaningful to run MAD regressions on gpm data that had already been
adjusted in the manner described above. Normalizing already reduced the variation in the
data, and brought outliers towards average values. This was the intended effect, so the
agencies deemed it unnecessary to apply an additional remedy to resolve an issue that had
already been addressed, but we seek comment on the use of robust regression techniques
under such circumstances.

2.4.2.3.2 Sales Weighting

      Likewise, the agencies reconsidered employing sales-weighting to represent the data.
As explained below, the decision to sales weight or not is ultimately based upon a choice
about how to represent the data, and not by an underlying statistical concern.  Sales weighting
is used if the decision is made to treat each (mass produced) unit sold as a unique physical
observation. Doing  so thereby changes the extent to which different vehicle model types are
emphasized  as compared to a non-sales weighted regression.  For example, while total
General Motors Silverado (332,000) and Ford F-150 (322,000) sales differ by less than
10,000 in MY 2021 market forecast, 62 F-150s models and 38 Silverado models  are reported
in the agencies baselines.  Without sales-weighting, the F-150 models, because there are more
of them, are  given 63 percent more weight in the regression despite  comprising a similar
portion of the marketplace and a relatively homogenous set of vehicle technologies.

        The  agencies did not use sales weighting in the 2012-2016 rulemaking analysis of the
curve shapes. A decision to not perform sales weighting reflects judgment that each vehicle
model provides an equal amount of information concerning  the underlying relationship
between footprint and fuel economy.  Sales-weighted regression gives the highest sales
vehicle model types  vastly more emphasis than the lowest-sales vehicle model types thus
driving the regression toward the sales-weighted fleet norm. For unweighted regression,
vehicle sales do not matter.  The agencies note that the light truck market forecast shows MY
2025 sales of 218,000 units for Toyota's 2WD Sienna, and shows 66 model configurations
with MY 2025 sales of fewer than 100 units.  Similarly, the  agencies' market forecast shows
MY 2025 sales of 267,000 for the Toyota Prius, and shows 40 model configurations with
MY2025 sales of fewer than 100 units. Sales-weighted analysis would give the Toyota
Sienna and Prius more than a thousand times the consideration of many vehicle model
configurations. Sales-weighted analysis would, therefore, cause a large number of vehicle
model configurations to be virtually ignored in the regressions.14
                                            2-28

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

       However, the agencies did note in the MYs 2012-2016 final rales that, "sales weighted
regression would allow the difference between other vehicle attributes to be reflected in the
analysis, and also would reflect consumer demand." 15 In reexamining the sales-weighting for
this analysis, the agencies note that there are low-volume model types account for many of the
passenger car model types (50 percent of passenger car model types account for 3.3 percent of
sales), and it is unclear whether the engineering characteristics of these model types should
equally determine the standard for the remainder of the market.

       In the interest of taking a fresh look at appropriate methodologies as promised in the
last final rule, in developing this proposal, the agencies gave full consideration to both sales-
weighted and unweighted regressions.

2.4.2.3.3 Analyses Performed

       We performed regressions describing the relationship between a vehicle's CO2/fuel
consumption and its footprint, in terms of various combinations  of factors: initial (raw) fleets
with no technology, versus after technology is applied; sales-weighted versus non-sales
weighted; and with and without two sets of normalizing factors applied to the observations.
The agencies excluded diesels and dedicated AFVs because the agencies anticipate that
advanced gasoline-fueled vehicles are likely to be dominant through MY2025.

       These are depicted graphically in Figures 2-9 through 2-16, below.

       Thus, the basic OLS regression on the initial data (with no technology applied) and no
sales-weighting represents one perspective on the relation between footprint and fuel
economy.  Adding sales weighting changes the interpretation to  include the influence of sales
volumes, and thus steps away from representing vehicle technology alone. Likewise, MAD is
an attempt to reduce the impact of outliers, but reducing the impact of outliers might perhaps
be less representative of technical relationships between the variables, although that
relationship  may change over time in reality. Each combination of methods  and data reflects
a perspective, and the regression results simply reflect that perspective in a simple
quantifiable  manner, expressed as the coefficients determining the  line through the average
(for OLS) or the median (for MAD) of the data.  It is left to policy  makers to determine an
appropriate perspective and to interpret the consequences of the  various alternatives.

       We invite comments on the application of the weights as described above,  and the
implications for interpreting the relationship between fuel efficiency and footprint.
        2.4.2.4     What results did the agencies obtain?

       Both agencies analyzed the same statistical approaches. For regressions against data
including technology normalization, NHTSA used the CAFE modeling system, and EPA used
EPA's OMEGA model.  The agencies obtained similar regression results, and have based
today's joint proposal on those obtained by NHTSA.

                                            2-29

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

       For illustrative purposes, the set of figures below show the range of curves determined
by the possible combinations of regression technique, with and without sales weighting, with
and without the application of technology, and with various adjustments to the gpm variable
prior to running a regression. Again, from a statistical perspective, each of these regressions
simply represents the assumptions employed.  Since they are all univariate linear regressions,
they describe the line that will result from minimizing the residuals or squared residuals.
Figures show the results for passenger cars, then light trucks, for ordinary least squares (OLS)
then similar results for MAD regressions for cars and light trucks, respectively.  The various
equations are represented by the string of attributes used to define the regression.  See the
table, Regression Descriptors, below, for the legend. Thus, for example, the line representing
ols_LT_wt_ft_adj_init_w should be read as follows: an OLS regression, for light trucks,
using data adjusted according to weight to footprint, no  technology added, and weighted by
sales.
                              Table 2-6 Regression Descriptors
Notation
ols or mad
PC or LT
hp_wt_adj
wt_ft_adj
wt_ft_hp_wt_adj
init or final
u or w
Description
Ordinary least squares or mean absolute deviation
Passenger car or light truck
Adjustment for horsepower to weight
Adjustment for weight to footprint
Adjustment for both horsepower to weight and weight to
Vehicles with no technology (initial) or with technology
footprint
added (final)
Unweighted or weighted by sales
Thus, the next figure, for example, represents a family of curves (lines) fit using ordinary least
squares on data for passenger cars, not modified for technology, and which therefore permits
comparisons of results in terms of the factors that change in each regression. These factors
are whether the data are sales-weighted (denoted "w") or unweighted (denoted "u"), as well as
the adjustments described above. Each of these adjustments has an influence on the
regressions results, depicted in the figures below.
                                             2-30

-------
                What are the Attribute-Based Curves the Agencies are Proposing
         o.oi
             25
              olsPC	init_w                    olsPC	init_u
              olsPC_hp_wt_adj	init_w          olsPC_hp_wt_adj	init_u
          ^^— ol s PC_wt_ft_h p_wt_a d j	i n it_w ^^— ol s PC_wt_ft_h p_wt_a d j	i n it_u
              olsPC_wt_ft_adj	init_w           olsPC_wt_ft_adj	init_u
           O No Technology Fleet            D  Technology Fleet
Figure 2-9 Best Fit Results for Various Regressions: Cars, No Added Technology, OLS
                                       2-31

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

       Figure 2-10, below, shows comparable results, this time with data representing the
additional technology that has been added to reduce technological heterogeneity. Note that
the data now pass through the relevant data "cloud" for the fleet with the technology
adjustment applied. The slopes of the lines are somewhat more clustered (less divergent) in
the chart depicting  added technology (as discussed in footnote m)
                                            2-32

-------
      What are the Attribute-Based Curves the Agencies are Proposing
 o.i
0.09
0.08
0.07
0.06
0.02
0.01
     25
                          Sya|    ooo
                         35
45
55
65
75
                                Footprint
                                     olsPC  final u
85
               olsPC	final_w                •         	    _
               olsPC_hp_wt_adj	final_w           olsPC_hp_wt_adj	final_u
               ols PC_wt_ft_h p_wt_a d j	f i n a l_w ^^— ol s PC_wt_ft_h p_wt_a d j	f i n a l_u
               ol s PC_wt_ft_a d j	f i n a l_w            ols PC_wt_ft_a d j	f i n a l_u
            O  No Technology Fleet              D  Technology Fleet
Figure 2-10 Best Fit Results for Various Regressions: Cars, with Added Technology, OLS
                              2-33

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

       Similar to the figures displaying the results for passenger cars, the figures below
display regression lines for trucks, first with no technology added, then subsequently, for the
case where technology has been added.  Slopes appear more similar to each other here than of
passenger cars.
                                            2-34

-------
       What are the Attribute-Based Curves the Agencies are Proposing
o.oi
     25
                          35
45
55
65
75
                                Footprint
                                      olsLT  init u
85
                 olsLT	init_w                •         	
                 olsLT_hp_wt_adj	init_w           olsLT_hp_wt_adj	init_u
                 ol s LT_wt_ft_h p_wt_a d j	init_w      olsLT_wt_ft_hp_wt_adj	init_u
                 olsLT_wt_ft_adj	init_w            olsLT_wt_ft_adj	init_u
             O  No Technology Fleet            D  Technology Fleet
Figure 2-11 Best Fit Results for Various Regressions: Trucks, No Added Technology, OLS
                              2-35

-------
       What are the Attribute-Based Curves the Agencies are Proposing
o.oi
     25
                           35
45
55
65
75
                                Footprint
                                      olsLT  final  u
85
                 olsLT	final_w                •         	
                 olsLT_hp_wt_adj	final_w           olsLT_hp_wt_adj	final_u
            — ol s LT_wt_ft_h p_wt_a d j	f i n a l_w ^^ ol s LT_wt_ft_h p_wt_a d j	f i n a l_u
                 olsLT_wt_ft_adj	final_w            olsLT_wt_ft_adj	final_u
             O  No Technology Fleet            D  Technology Fleet
Figure 2-12 Best Fit Results for Various Regressions:  Trucks, With Added Technology, OLS
                              2-36

-------
                      What are the Attribute-Based Curves the Agencies are Proposing
       Figure 2-13, below, displays regression results for the passenger car MAD fitted
curves. The technology adjustment does not have, however, the same degree of impact in
reducing the difference in the attained slopes (between those with and without the addition of
technology) evidenced in the OLS regressions.
                                           2-37

-------
             What are the Attribute-Based Curves the Agencies are Proposing
o.oi
    25
                   35
45
55
65
                                Footprint
                                     madPC init  u
75
85
         madPC_init_w                •          _  _
         madPC_hp_wt_adj_init_w          madPC_hp_wt_adj_init_u
         madPC_wt_ft_hp_wt_adj_init_w     madPC_wt_ft_hp_wt_adj_init_u
       — madPC_wt_ft_adj_init_w           madPC_wt_ft_adj_init_u
      O  No Technology Fleet            D  Technology Fleet
Figure 2-13 Best Fit Results for Various Regressions: Cars, No Added Technology, MAD
                                    2-38

-------
               What are the Attribute-Based Curves the Agencies are Proposing
    o.i
   0.09
   0.08
   0.07
   0.06
   0.02
   0.01
      0
                            B    ya|    ooo
        25        35        45        55        65        75        85
                                  Footprint
        madPC_final_w                     madPC_final_u
   ^^— m a d PC_h p_wt_a d j_f i n a l_w       ^^— m a d PC_h p_wt_a d j_f i n a l_u
   — mad PC_wt_ft_h p_wt_a d j_f i n a l_w — mad PC_wt_ft_h p_wt_a d j_f i n a l_u
        mad PC_wt_ft_a d j_f i n a l_w             mad PC_wt_ft_a d j_f i n a l_u
     O  No Technology Fleet               n  Technology Fleet
Figure 2-14 Best Fit Results for Various Regressions: Cars, Added Technology, MAD
                                      2-39

-------
                        What are the Attribute-Based Curves the Agencies are Proposing

       The MAD regression results below in Figure 2-15 show a grouping of the fitted lines
similar to that displayed in the OLS fits for trucks. As expected, an additional reduction in
divergence is seen in the case where technology has been added, in Figure 2-15, which can be
ascribed to the reduction in heterogeneity of the fleet brought about by the addition of the
technology.
        .0
        3
             o.i
            0.09
            0.08
            0.07
            0.06
            0.05
            0.04
            0.03
            0.02
            0.01
                25
                   35
45
55
65
75
85
                                           Footprint
                                                madLT init  u
          madLT_init_w

          madLT_hp_wt_adj_init_w          madLT_hp_wt_adj_init_u

          mad LT_wt_ft_h p_wt_a d j_i n i t_w ^^ mad LT_wt_ft_h p_wt_a d j_i n it_u

        — madLT_wt_ft_adj_init_w           madLT_wt_ft_adj_init_u

       O  No Technology Fleet           D  Technology Fleet

Figure 2-15 Best Fit Results for Various Regressions: Trucks, No Added Technology, MAD
                                              2-40

-------
            What are the Attribute-Based Curves the Agencies are Proposing
o.oi
    25
                    35
45
55
65
75
                               Footprint
                                     madLT final u
85
          madLT_final_w
      ^^ mad LT_h p_wt_a d j_f i n a l_w      ^^ mad LT_h p_wt_a d j_f i n a l_u
      ^^— mad LT_wt_ft_h p_wt_a d j_f i n a l_w ^^— mad LT_wt_ft_h p_wt_a d j_f i n a l_u
          mad LT_wt_ft_a d j_f i n a l_w           mad LT_wt_ft_a d j_f i n a l_u
       O  No Technology Fleet             D  Technology Fleet
Figure 2-16 Best Fit Results for Various Regressions: Trucks, with Added Technology, MAD
                                   2-41

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
         2.4.2.5      Which methodology did the agencies choose for this proposal, and
         why is it reasonable?

       The choice among the alternatives presented above was to use the OLS formulation,
on sales-weighted data, using a fleet that has had technology applied, and after adjusting the
data for the effect of weight-to-footprint, as described above. The agencies believe that this
represents a technically reasonable approach for purposes of developing target curves to
define the proposed standards, and that it represents a reasonable trade-off among various
considerations balancing statistical, technical, and policy matters, which include the statistical
representativeness of the curves considered and the steepness of the curve chosen. The
agencies judge the application of technology prior to curve fitting to provide a reasonable
means—one consistent with the rule's objective of encouraging manufacturers to add
technology in order to increase fuel economy—of reducing variation in the data and thereby
helping to estimate a relationship between fuel consumption/COi and footprint.

       Similarly, for the agencies' current MY 2008-based market-forecast and the agencies'
current estimates of future technology effectiveness, the inclusion of the weight-to-footprint
data adjustment prior to running the regression also helps to improve the fit of the curves by
reducing the variation in the data, and the agencies believe that the benefits of this adjustment
for this proposed rule likely outweigh the potential that resultant curves might somehow
encourage reduced load carrying capability or vehicle performance (note that the we are not
suggesting that we believe these adjustments will reduce load carrying capability or vehicle
performance). In addition to reducing the variability, the truck curve is also steepened, and
the car curve flattened compared to curves fitted to  sales weighted data that do not include
these normalizations.  The agencies agree with manufacturers of full-size pick-up trucks that
in order to maintain towing and hauling utility, the engines on pick-up trucks  must be more
powerful, than their low "density" nature would statistically suggest based on the agencies'
current MY2008-based market forecast and the agencies' current estimates of the
effectiveness of different fuel-saving technologies.  Therefore, it may be more equitable (i.e.,
in terms of relative compliance challenges faced by different light truck manufacturers) to
adjust the slope of the curve defining fuel economy and COi targets.

       The results of the normalized regressions are displayed in Table, below.
                                Table 2-7 Regression Results
Vehicle
Passenger cars
Light trucks
Slope
(gallons/mile)
0.000431
0.0002526
Constant
(gallons/mile)
-0.00052489
0.01121968
                                             2-42

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

       As described above, however, other approaches are also technically reasonable, and
also represent a way of expressing the underlying relationships. The agencies plan to revisit
the analysis for the final rale, after updating the underlying market forecast and estimates of
technology effectiveness, and based on relevant public comments received. In addition, the
agencies intend to update the technology cost estimates, which could alter the NPRM analysis
results and consequently alter the balance of the trade-offs being  weighed to determine the
final curves.

       As shown in the figures below, the line represents the sales-weighted OLS regression
fit of gallons  per mile regressed on footprint, with the data first adjusted by weight to
footprint, as described above. This introduces weight as an additional consideration into the
slope of the footprint curve, although in a manner that adjusts the data as described above, and
thus maintains a simple graphical interpretation of the curve in a  two dimensional space
(gallons per mile and footprint).
                                     GPM vs. Footprint - Cars
                                          50      55
                                               Footprint
                      Figure 2-17 Gallons per Mile versus Footprint, Cars

                            (Data adjusted by weight to footprint).
                                             2-43

-------
What are the Attribute-Based Curves the Agencies are Proposing
                    2-44

-------
                       What are the Attribute-Based Curves the Agencies are Proposing
                                   GPM vs. Footprint - Trucks
         S.

         I,
                     Figure 2-18 Gallons per Mile versus Footprint, Trucks

                            (data adjusted by weight to footprint).
       In the preceding two figures, passenger car and light truck data is represented for the
specification chosen, with the size of the observation scaled to sales.  The agencies note with
regard to light trucks that for the MYs 2012-2016 analysis NPRM and final rule analyses,
some models of pickups had been aggregated together, when, for example, the same pickup
had been available in different cab configurations with different wheelbases.16 For the current
analysis, these models have been disaggregated and are represented individually, which leads
to a slightly different outcome in the regression results than had they remained aggregated.

         2.4.2.6      Implications of the proposed slope compared to MY 2012-2016

       The proposed slope has several implications relative to the MY 2016 curves, with the
majority of changes on the truck curve.  With the agencies' current MY2008-based market
forecast and the agencies' current estimates of technology effectiveness, the combination of
sales weighting and WT/FP normalization produced a car curve slope similar to that finalized
in the MY 2012-2016 final rulemaking (4.7 g/mile in MY 2016, vs. 4.5 g/mile proposed in
MY 2017).  By contrast, the truck curve is steeper in MY 2017 than in MY 2016 (4.0 g/mile
in MY 2016 vs. 4.9  g/mile in MY 2017).  As discussed previously, a steeper slope relaxes the
                                            2-45

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

stringency of targets for larger vehicles relative to those for smaller vehicles, thereby shifting
relative compliance burdens among manufacturers based on their respective product mix.
2.5 Once the agencies determined the appropriate slope for the sloped part, how did the
       agencies determine the rest of the mathematical function?

       The agencies continue to believe that without a limit at the smallest footprints, the
function—whether logistic or linear—can reach values that would be unfairly burdensome for
a manufacturer that elects to focus on the market for small vehicles; depending on the
underlying data, an unconstrained form could result in stringency levels that are
technologically infeasible and/or economically impracticable for those manufacturers that
may elect to focus on the smallest vehicles.  On the other side of the function, without a limit
at the largest footprints, the function may provide no  floor on required fuel economy. Also,
the safety considerations that support  the provision of a disincentive for downsizing as a
compliance strategy apply weakly, if at all, to the very largest vehicles.  Limiting the
function's value for the largest vehicles thus leads to  a function with an inherent absolute
minimum level  of performance, while remaining consistent with safety considerations.

       Just as for slope, in determining the appropriate footprint and fuel economy values for
the "cutpoints," the places along the curve where the  sloped portion becomes flat, the
agencies took a fresh look for purposes of this proposal, taking into account the updated
market forecast and new assumptions about the availability of technologies.  The next two
sections discuss the agencies'  approach to cutpoints for the passenger car and light truck
curves separately, as the policy considerations for each vary somewhat.
       2.5.1   Cutpoints for PC curve

       The passenger car fleet upon which the agencies have based the target curves for MYs
2017-2025 is derived from MY 2008 data, as discussed above.  In MY 2008, passenger car
footprints ranged from 36.7 square feet, the Lotus Exige 5, to 69.3 square feet, the Daimler
Maybach 62. In that fleet, several manufacturers offer small, sporty coupes below 41 square
feet, such as the BMW Z4 and Mini, Honda  S2000, Mazda MX-5 Miata, Porsche Carrera and
911, and Volkswagen New Beetle.  Because such vehicles represent a small portion (less than
10 percent) of the passenger car market, yet often have performance, utility, and/or structural
characteristics that could make it technologically infeasible and/or economically
impracticable for manufacturers focusing on such vehicles to achieve the very challenging
average requirements that could apply in the absence of a constraint, EPA and NHTSA are
again proposing to cut off the sloped portion of the passenger car function at 41 square feet,
consistent with the MYs 2012-2016 rulemaking. The agencies  recognize that for
manufacturers who make small vehicles in this size range, putting the cutpoint at 41 square
feet creates some incentive to downsize (i.e., further reduce the size, and/or increase the
                                            2-46

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

production of models currently smaller than 41 square feet) to make it easier to meet the
target. Putting the cutpoint here may also create the incentive for manufacturers who do not
currently offer such models to do so in the future.  However, at the same time, the agencies
believe that there is a limit to the market for cars smaller than 41 square feet -- most
consumers likely have some minimum expectation about interior volume, among other things.
The agencies thus believe that the number of consumers who will want vehicles smaller than
41 square feet (regardless of how they are priced) is small, and that the incentive to downsize
to less than 41 square feet in response to this proposal, if present, will be at best minimal.  On
the other hand, the agencies note that some manufacturers are introducing mini cars not
reflected in the agencies MY 2008-based market forecast, such as the Fiat 500, to the U.S.
market, and that the footprint at which the curve is limited may affect the incentive for
manufacturers to do so.

       Above 56 square feet, the only passenger car models present in the MY 2008 fleet
were four luxury vehicles with extremely low sales volumes—the Bentley Arnage and three
versions of the Rolls Royce Phantom. As in the MYs 2012-2016 rulemaking, NHTSA and
EPA therefore are proposing again to cut off the sloped portion of the passenger car function
at 56 square feet.

       While meeting with manufacturers prior to issuing the proposal, the agencies received
comments from some manufacturers that, combined with slope and overall stringency, using
41 square feet as the footprint at which to cap the target for small cars would result in unduly
challenging targets for small cars. The agencies do not agree. No specific vehicle need meet
its target (because standards apply to fleet average performance), and maintaining a sloped
function toward the smaller end of the passenger car market is important to discourage unsafe
downsizing, the agencies are thus proposing to again "cut off the passenger car curve at 41
square feet, notwithstanding these comments.

       The agencies seek comment on setting cutpoints for the MYs 2017-2025 passenger car
curves at 41  square feet and 56 square feet.
       2.5.2   Cutpoints for LT curve
       The light truck fleet upon which the agencies have based the target curves for MYs
2017-2025, like the passenger car fleet, is derived from MY 2008 data, as discussed in
Section 2.4 above.  In MY 2008, light truck footprints ranged from 41.0 square feet, the Jeep
Wrangler, to 77.5 square feet, the Toyota Tundra. For consistency with the curve for
passenger cars, the agencies are proposing to cut off the sloped portion of the light truck
function at the same footprint, 41 square feet, although we recognize that no light trucks are
currently offered below 41 square feet. With regard to the upper cutpoint,  the agencies heard
from a number of manufacturers during the discussions leading up to this proposal that the
                                            2-47

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

location of the cutpoint in the MYs 2012-2016 rules, 66 square feet, meant that the same
standard applied to all light trucks with footprints of 66 square feet or greater, and that in fact
the targets for the largest light trucks in the later years of that rulemaking were extremely
challenging.  Those manufacturers requested that the agencies extend the cutpoint to a larger
footprint, to reduce targets for the largest light trucks which represent a significant percentage
of those manufacturers light truck sales. At the same time, in re-examining the light truck
fleet data, the agencies concluded that aggregating pickup truck models in the MYs 2012-
2016 rule had led the agencies to underestimate the impact of the different pickup truck model
configurations above 66 square feet on manufacturers' fleet average fuel economy and CO2
levels (as discussed immediately below). In disaggregating the pickup truck model data, the
impact of setting the cutpoint at 66 square feet after model year 2016 became clearer to the
agencies.

       In the agencies' view, there is legitimate basis for these comments.  The agencies'
market forecast includes about 24 vehicle configurations above 74 square feet with a total
volume of about 50,000 vehicles or less during any MY in the 2017-2025 time frame. While
a relatively small portion of the overall truck fleet, for some manufacturers, these vehicles are
non-trivial portion of sales. As noted above, the very largest light trucks have significant
load-carrying and towing capabilities that make it particularly challenging for manufacturers
to add fuel economy-improving/COi-reducing technologies in a way that maintains the full
functionality  of those capabilities.

       Considering manufacturer CBI and our estimates of the impact of the 66 square foot
cutpoint for future model years, the agencies have initially determined to adopt curves that
transition to a different cut point.  While noting that no  specific vehicle need meet its target
(because standards apply to fleet average performance),  we believe that the information
provided to us by manufacturers and our own analysis supports the gradual extension of the
cutpoint for large light trucks in this proposal from 66 square feet in MY 2016 out to a larger
footprint square feet before MY 2025.
                                            2-48

-------
                      What are the Attribute-Based Curves the Agencies are Proposing
                2000000 -
                1500000 -
                1000000 -
                 500000 -
               CO
                2000000 -
                1500000 -
                1000000 -
                 500000 -
                                      50         60
                                        Footprint
                     Figure 2-19 Footprint Distribution by Car and Truck*
           *Proposed truck cutpoints for MY 2025 shown in red, car cutpoints shown in green

       The agencies are proposing to phase in the higher cutpoint for the track curve in order
to avoid any backsliding from the MY 2016 standard. A target that is feasible in one model
year should never become less feasible in a subsequent model year—manufacturers should
have no reason to remove fuel economy-improving/CO2-reducing technology from a vehicle
once it has been applied.  Put another way, the agencies are proposing to not allow "curve
crossing" from one model year to the next.  In proposing MYs 2011-2015 CAFE standards
and promulgating MY 2011 standards, NHTSA proposed and requested comment on avoiding
curve crossing, as an "anti-backsliding measure."17 The MY 2016 2 cycle test curves are
therefore a floor for the MYs 2017-2025 curves. For passenger cars, which have minimal
change in slope from the MY 2012-2016 rulemakings and no change in cut points, there are
no curve crossing issues in the proposed standards.

       The minimum stringency determination was done using the two cycle curves.
Stringency adjustments for air conditioning and other credits were calculated after curves that
did not cross were determined in two cycle space. The year over year increase in these
adjustments cause neither the GHG nor CAFE curves (with A/C) to contact the 2016 curves
when charted.
                                            2-49

-------
                       What are the Attribute-Based Curves the Agencies are Proposing

       2.5.3   Once the agencies determined the complete mathematical function shape,
      how did the agencies adjust the curves to develop the proposed standards and
      regulatory alternatives?

       The curves discussed above all reflect the addition of technology to individual vehicle
models to reduce technology differences between vehicle models before fitting curves. This
application of technology was conducted not to directly determine the proposed standards, but
rather for purposes of technology adjustments, and set aside considerations regarding
potential rates of application (i.e., phase-in caps), and considerations regarding economic
implications of applying specific technologies to specific vehicle models.  The following
sections describe further adjustments to the curves discussed above, that affect both the shape
of the curve (section 2.5.3.1), and the location of the curve (2.5.3.2), that helped the agencies
determine curves that defined the proposed standards.

         2.5.3.1     Adjusting for Year over Year Stringency

       As in the MYs 2012-2016 rules, the agencies developed curves defining regulatory
alternatives for consideration by "shifting" these curves.  For the MYs 2012-2016 rules, the
agencies did so on an absolute basis, offsetting the fitted curve by the same value (in gpm or
g/mi) at all footprints. In developing this proposal, the agencies have reconsidered the use of
this approach, and have concluded that after MY 2016, curves should be offset on a relative
basis—that is, by adjusting the entire gpm-based curve (and, equivalently, the CO2 curve) by
the same percentage rather than the same absolute value.  The agencies' estimates of the
effectiveness  of these technologies are all expressed in relative terms—that is, each
technology (with the exception of A/C) is estimated to reduce fuel consumption (the inverse
of fuel economy) and COi emissions by a specific percentage of fuel consumption without the
technology. It is, therefore, more consistent with the agencies' estimates of technology
effectiveness  to develop the proposed standards and regulatory alternatives by applying a
proportional offset to curves expressing fuel consumption or emissions as a function of
footprint. In addition, extended indefinitely (and  without other compensating adjustments),
an absolute offset would eventually (i.e., at very high average stringencies) produce negative
(gpm or g/mi) targets. Relative offsets avoid this potential outcome. Relative offsets do
cause curves to become, on a fuel consumption and COi basis, flatter at greater average
stringencies; however, as discussed above, this outcome remains  consistent with the agencies'
estimates of technology effectiveness.  In other words,  given a relative decrease in average
required fuel consumption or CCh emissions, a curve that is flatter by the same relative
amount should be equally challenging in terms of the potential to achieve compliance through
the addition of fuel-saving technology.

       On this basis, and considering that the "flattening" occurs gradually for the regulatory
alternatives the agencies have evaluated, the agencies tentatively conclude that this approach
to offsetting the curves to develop year-by-year regulatory alternatives neither re-creates a
situation in which manufacturers are likely to respond to standards in ways that compromise
highway safety, nor undoes the attribute-based standard's more equitable balancing of
compliance burdens among disparate manufacturers. The agencies invite comment on these

                                            2-50

-------
                      What are the Attribute-Based Curves the Agencies are Proposing

conclusions, and on any other means that might avoid the potential outcomes—in particular,
negative fuel consumption and COi targets—discussed above.

        2.5.3.2     Adjusting for anticipated improvements to mobile air conditioning
        systems

       The fuel economy values in the agencies' market forecast are based on the  2-cycle
(i.e., city and highway) fuel economy test and calculation procedures that do not reflect
potential improvements in air conditioning system efficiency, refrigerant leakage, or
refrigerant Global Warming Potential (GWP).  Recognizing that there are significant and cost
effective potential air conditioning system improvements available in the rulemaking
timeframe (discussed in detail in Chapter 5 of the draft joint TSD), the agencies are increasing
the stringency of the target curves based on the agencies' assessment of the capability of
manufacturers to implement these changes.  For the proposed CAFE standards and
alternatives, an offset is included based on air conditioning system efficiency improvements,
as these improvements are the only improvements that effect vehicle fuel economy. For the
proposed GHG standards  and alternatives, a stringency increase is included based on air
conditioning system efficiency, leakage and refrigerant improvements. As discussed above in
Chapter 5 of the join TSD, the air conditioning system improvements affect a vehicle's fuel
efficiency or CC>2 emissions performance as an additive stringency increase, as compared to
other fuel efficiency improving technologies which are multiplicative. Therefore, in adjusting
target curves for improvements in the air conditioning system performance, the agencies are
adjusting the target curves by additive stringency increases (or vertical shifts) in the curves.

       For the GHG target curves, the offset for air conditioning system performance is being
handled in the same manner as for the MY 2012-2016 rules.  For the CAFE target curves,
NHTSA for the first time  is proposing to  account for potential improvements in air
conditioning system performance.  Using this methodology, the agencies first use a
multiplicative stringency adjustment for the sloped portion of the curves to reflect the
effectiveness on technologies other that air conditioning system technologies, creating a series
of curve shapes that are "fanned" based on two-cycle performance.  Then the curves are offset
vertically by the air conditioning improvement by  an equal amount at every point.
                                            2-51

-------
                     What are the Attribute-Based Curves the Agencies are Proposing




References:





1 49 U.S.C. 32902(a)(3)(A).




2 69 FR 38958 (June 29, 2004).




3 76 FR 57106, 57162-64, (Sept. 15, 2011).




4 See 74 FR at 14359 (Mar. 30, 2009).




5 75 FR at 25362.




6 See generally 74 FR at 49491-96; 75 FR at 25357-62.




7 68 FR 74920-74926.




8 74 FR 14359.




9 See 75 FR at 25458




10 75 FR at 25363




11 See 75 FR at 25359.




12 l(L at 25362-63.




13 kL at 25363.




14 75 FR at 25362 and n. 64




15 75 FR at 25632/3.




16 See 75 FR at 25354




17 74 Fed. Reg. at 14370 (Mar. 30, 2009).
                                         2-52

-------
                                     Technologies Considered in the Agencies' Analysis
Chapter 3:   Technologies Considered in the Agencies' Analysis

       This Chapter of the joint TSD describes the technologies NHTSA and EPA evaluated
as potential inputs in their respective models and provides estimates of the technologies'
costs, effectiveness and availability. This Chapter also describes, in general terms, how the
agencies use these inputs in their respective models.

       The agencies assume, in this analysis, that manufacturers will add a variety of
technologies to each of their vehicle model platforms in order to improve their fuel economy
and GHG performance.  In order to evaluate proposed CAFE and GHG standards and
regulatory alternatives, it is essential to understand what is feasible within the timeframe of
the proposed rule. Determining the technological feasibility of proposed 2017-2025 standards
requires a thorough study of the technologies available to the manufacturers during that
timeframe.  This chapter includes an assessment of the cost, effectiveness, and the
availability, development time, and manufacturability of the technology within either the
normal redesign periods of a vehicle line or in the design of a new vehicle.  As we describe
below, when a technology can be applied can affect the cost as well as the technology
penetration rate (or phase-in caps) that are assumed in the analysis.

       The agencies considered technologies in many categories that manufacturers could use
to improve  the fuel economy and reduce CC>2 emissions of their  vehicles during the MYs
2017-2025 timeframe. Many of the technologies described in this chapter are available today,
are well known, and could be incorporated into vehicles once product development decisions
are made.  These are "nearer-term" technologies and are identical or very similar to those
considered in the MYs 2012-2016 final rule analysis (of course,  many of these technologies
will likely be applied to the light-duty fleet in order to achieve the 2012-2016 CAFE and
GHG standards; such technologies would be part of the 2016 reference case for  this
analysisa).  Other technologies considered, may not currently be  in production, but are under
development and are expected to be in production in the next five to ten years. Examples of
these technologies are downsized and turbocharged engines operating at combustion pressures
even higher than today's turbocharged engines, and an emerging hybrid architecture mated
with an 8 speed transmission—a combination that is not available today. These are
technologies which the agencies believe can, for the most part, be applied both to cars and
trucks, and  which are expected to achieve significant improvements in fuel  economy and
reductions in CO2 emissions at reasonable costs in the MYs 2017 to 2025 timeframe. The
agencies note that we did not consider in our analysis technologies that are currently in an
initial stage of research because of the uncertainties involved in estimating their costs and
effectiveness and in assessing whether the technologies will be ready to implement at
significant penetration rates during the timeframe of this proposal.  Examples of such
a The technologies in the 2016 reference fleet are projections made by EPA's OMEGA model and NHTSA's
CAFE model respectively. Some technologies may be significantly represented in this reference fleet and these
details can be found in each agency's respective RIAs.

                                             3-1

-------
                                     Technologies Considered in the Agencies' Analysis
technologies would be camless valve actuation and fuel cell vehicles.13 The agencies
acknowledge that due to the relatively long period between the date of this proposal and the
rulemaking timeframe, the possibility exists that new and innovative technologies not
considered in this analysis will make their way into the fleet (perhaps even in significant
numbers). The agencies plan to re-assess these technologies, along with all of the
technologies considered in this proposal, as part of our mid-term evaluation.

3.1 What Technologies did the agencies consider for the proposed 2017-2025 standards?

       The technologies considered for this NPRM analysis by NHTSA and EPA are briefly
described below.  They fit generally into five broad categories:  engine, transmission, vehicle,
electrification/accessory, and hybrid technologies. A more detailed description of each
technology, and the technology's costs and effectiveness, is described in greater detail in
section 3.4 of this TSD.

       Types of engine technologies applied in this NPRM analysis to improve fuel economy
and reduce CC>2 emissions include the following:
       •   Low-friction lubricants - low viscosity and advanced low friction lubricants oils
          are now available with improved performance and better lubrication.

       •   Reduction of engine friction losses - can be achieved through low-tension piston
          rings, roller cam followers, improved material coatings, more optimal thermal
          management, piston surface treatments, and other improvements in the design of
          engine components and subsystems that improve engine operation.
       •   Second level of low-friction lubricants and engine friction reduction - As
          technologies advance between now and the rulemaking timeframe, there will
          further developments enabling lower viscosity and lower friction lubricants and
          more engine friction reduction technologies available.

       •   Cylinder deactivation - deactivates the intake and exhaust valves and prevents fuel
          injection into some cylinders during light-load operation.  The engine runs
          temporarily as though it were a smaller engine which substantially reduces
          pumping losses

       •   Variable valve timing - alters the timing or phase of the intake valve, exhaust
          valve,  or both, primarily to reduce pumping losses, increase specific power, and
          control residual gases.
b Fuel cell vehicles may be especially useful in lieu of full battery electric technology for the larger trucks. We
may consider this possibility for the final rule.

                                             3-2

-------
                           Technologies Considered in the Agencies' Analysis
Discrete variable valve lift - increases efficiency by optimizing air flow over a
broader range of engine operation which reduces pumping losses. Accomplished
by controlled switching between two or more cam profile lobe heights.

Continuous variable valve lift - is an electromechanically controlled system in
which cam period and phasing is changed as lift height is controlled. This yields a
wide range of performance optimization and volumetric efficiency, including
enabling the engine to be valve throttled.

Stoichiometric gasoline direct-injection technology - injects fuel at high pressure
directly into the combustion chamber to improve cooling of the air/fuel charge
within the cylinder, which allows for higher compression ratios and increased
thermodynamic efficiency.

Turbocharging and downsizing - increases the available airflow and specific
power level, allowing a reduced engine  size while maintaining performance. This
reduces pumping losses at lighter loads in comparison to a larger engine. In this
NPRM, the agencies considered three levels of boosting, 18 bar brake mean
effective pressure (BMEP), 24 bar BMEP and 27 bar BMEP, as well as four levels
of downsizing, from 14 to smaller 14 or 13, from V6 to 14 and from V8 to V6 and
14. 18 bar BMEP is applied with 33 percent downsizing, 24 bar BMEP is applied
with 50 percent downsizing and 27 bar BMEP is applied with 56 percent
downsizing. To achieve the same level  of torque when downsizing the
displacement of an engine by 50 percent, approximately double the manifold
absolute pressure (2 bar) is required. Accordingly, with 56 percent downsizing,
the manifold absolute pressure range increases up to 2.3 bar.  Ricardo states in
their 2011 vehicle simulation project report that advanced engines in the 2020-
2025 timeframe can be expected to have advanced boosting systems that increase
the pressure of the intake charge up to 3 bar1. Refer to Section 3.3.1.2.22.2 for
examples of Ricardo-modeled displacements used for turbocharged and downsized
engines in each vehicle class.

Exhaust-gas recirculation boost - increases the exhaust-gas recirculation used in
the combustion process to increase thermal efficiency and reduce pumping losses.
Levels of exhaust gas recirculation approach 25% by volume in the highly boosted
engines modeled by Ricardo (this, in turn raises the boost requirement by
approximately 25%). This technology is only applied to 24 bar and 27 bar BMEP
engines in this NPRM.

Diesel engines - have several characteristics that give superior fuel efficiency,
including reduced pumping losses due to lack of (or greatly reduced) throttling,
and a combustion cycle that operates at  a higher compression ratio, with a  very
lean air/fuel mixture, than an equivalent-performance gasoline engine.  This
technology requires additional enablers, such as NOX trap catalyst after-treatment
or selective catalytic reduction NOX after-treatment.
                                   3-3

-------
                              Technologies Considered in the Agencies' Analysis
Types of transmission technologies applied in this NPRM include:
•  Improved automatic transmission controls - optimizes shift schedule to maximize
   fuel efficiency under wide ranging conditions, and minimizes losses associated
   with torque converter slip through lock-up or modulation.
•  Six- and seven-speed automatic transmissions - the gear ratio spacing and
   transmission ratio are optimized to enable the engine to operate in a more efficient
   operating range over a broader range of vehicle operating conditions.
•  Dual clutch transmission (DCT) - are similar to a manual transmission, but the
   vehicle controls shifting and launch functions. A dual-clutch automated shift
   manual transmission uses separate clutches for even-numbered and odd-numbered
   gears, so the next expected gear is pre-selected, which allows for faster, smoother
   shifting.

•  Eight-speed automatic transmissions - the gear transmission ratio are optimized to
   enable the engine to operate in a more efficient operating range over a broader
   range of vehicle operating conditions. This technology is applied after 2016.
•  Shift  Optimization - tries to keep the engine operating near its most efficient point
   for a  give power demand. The shift controller emulates a traditional CVT by
   selecting the best gear ratio for fuel economy at a given required vehicle power
   level  to take full advantage of high BMEP engines.
•  Manual 6-speed transmission - offers an additional gear ratio, often with a higher
   overdrive gear ratio, than a 5-speed manual transmission.

•  High  Efficiency Gearbox (automatic, DCT or manual) - continuous improvement
   in seals, bearings and clutches, super finishing of gearbox parts, and development
   in the area of lubrication, all aimed at reducing frictional and other parasitic load in
   the system for an automatic, DCT or manual type transmission.

Types of vehicle technologies applied in this NPRM analysis include:
•  Low-rolling-resistance tires - have characteristics that reduce frictional losses
   associated with the energy dissipated in the deformation of the tires under load,
   therefore reducing the energy needed to move the vehicle. There are two levels of
   rolling resistance reduction considered in this  NRPM analysis targeting at 10
   percent and 20 percent rolling resistance reduction respectively.
•  Low-drag brakes - reduce the sliding friction  of disc brake pads on rotors when
   the brakes are not engaged because the brake pads are pulled away from the rotors.
•  Front or secondary axle disconnect for four-wheel drive systems - provides a
   torque distribution disconnect between front and rear axles  when torque is not
   required for the non-driving axle.  This results in the reduction of associated
   parasitic energy losses.
                                      3-4

-------
                                     Technologies Considered in the Agencies' Analysis
       •   Aerodynamic drag reduction - is achieved by changing vehicle shape or reducing
          frontal area, including skirts, air dams, underbody covers, and more aerodynamic
          side view mirrors. There are two levels of aerodynamic drag reduction considered
          in this NPRM analysis targeting 10 percent and 20 percent rolling resistance
          reduction respectively.
       •   Mass reduction- Mass reduction encompasses a variety of techniques ranging
          from improved design and better component integration to application of lighter
          and higher-strength materials. Mass reduction can lead to collateral fuel economy
          and GHG benefits due to downsized engines and/or ancillary systems
          (transmission, steering, brakes, suspension, etc.).  The maximum mass reduction
          level considered in this NPRM is 20 percent.

       Types of electrification/accessory and hybrid technologies applied in this NPRM
include:
       •   Electric power steering (EPS) and electro-hydraulic power steering (EHPS) - is
          an electrically-assisted steering system that has advantages over traditional
          hydraulic power steering because it replaces a continuously operated hydraulic
          pump, thereby reducing parasitic losses from the accessory drive.
       •   Improved accessories (IACC) - may include high efficiency alternators,
          electrically driven (i.e., on-demand) water pumps and cooling and even
          regenerative braking. This excludes other electrical accessories  such as electric oil
          pumps and electrically driven air conditioner compressors. There are two levels of
          IACC applied in this NPRM analysis. The second level of IACC includes
          alternator regenerative braking on top of what are included in the first level of
          IACC.

       •   Air Conditioner Systems - These technologies include improved hoses, connectors
          and seals for leakage control. They also include improved compressors, expansion
          valves, heat exchangers and the control of these components for the purposes of
          improving tailpipe CC^ emissions and fuel economy when the A/C is operating.
          These technologies are covered separately in Chapter 5 of this draft joint TSD.

       •   12-volt Stop-start - also known as idle-stop or 12V micro hybrid and commonly
          implemented as a 12-volt belt-driven integrated starter-generator, this is the most
          basic hybrid system that facilitates idle-stop capability.  Along with other enablers,
          this system replaces a common alternator with an enhanced power starter-
          alternator, both  belt driven, and a revised accessory drive system.

       •   P2 Hybrid - P2 hybrid is a newly emerging hybrid technology that uses a
          transmission integrated electric motor placed between the engine and a gearbox or
          CVT, much like the IMA system described above except with a  wet or dry
          separation clutch which is used to decouple the motor/transmission from the
          engine.  In addition, a P2 Hybrid would typically be equipped with a larger electric
          machine. Engaging the clutch allows all-electric operation and more efficient

                                             3-5

-------
                              Technologies Considered in the Agencies' Analysis
   brake-energy recovery.  Disengaging the clutch allows efficient coupling of the
   engine and electric motor and, when combined with a DCT transmission, reduces
   gear-train losses relative to power-split or 2-mode hybrid systems.

•  Plug-in hybrid electric vehicles (PHEV) - are hybrid electric vehicles with the
   means to charge their battery packs from an outside source of electricity (usually
   the electric grid).  These vehicles have larger battery packs with more energy
   storage and a greater capability to be discharged. They also use a control system
   that allows the battery pack to be substantially depleted under electric-only or
   blended mechanical/electric operation.

•  Electric vehicles (EV) - are vehicles with all-electric drive and with vehicle
   systems powered by energy-optimized batteries charged primarily from grid
   electricity. EVs with 75 mile, 100 mile and 150 mile ranges have been included as
   potential technologies.

Types of accessory/hybridization/electrification technologies discussed but not applied
in this NPRM analysis include:
•  Higher Voltage Stop-Start/Belt Integrated Starter Generator (BISG) - sometimes
   referred to as a mild hybrid, BISG provides idle-stop capability and uses a high
   voltage battery with increased energy capacity over typical automotive batteries.
   The higher system voltage allows the use of a smaller, more powerful electric
   motor and reduces the weight of the motor, inverter, and battery wiring  harnesses.
   This system replaces a standard alternator with an enhanced power, higher voltage,
   higher efficiency starter-alternator, that is belt driven and that can recover braking
   energy while the vehicle slows down (regenerative braking).

•  Integrated Motor Assist (IMA)/Crank integrated starter generator (CISC) -
   provides idle-stop capability and uses a high voltage battery with increased energy
   capacity over typical automotive batteries.  The higher system voltage allows the
   use of a smaller, more powerful electric motor and reduces the  weight of the
   wiring harness. This  system replaces a standard alternator with an enhanced
   power, higher voltage and higher efficiency starter-alternator that is crankshaft
   mounted and can recover braking energy while the vehicle slows  down
   (regenerative braking).  The IMA technology is not included as an enabling
   technology in this analysis, although it is included as a baseline technology
   because it exists in the 2008 baseline fleet.

•  Power-split Hybrid (PSHEV) - is a hybrid electric drive system that replaces the
   traditional transmission with a single planetary gearset and a motor/generator.
   This motor/generator  uses the engine to either charge the battery or supply
   additional power to the drive motor. A second, more powerful motor/generator is
   permanently connected to the vehicle's final drive and always turns with the
   wheels. The planetary gear splits engine power between the first motor/generator
   and the drive motor to either charge the battery or supply power to the wheels.

                                       3-6

-------
                                     Technologies Considered in the Agencies' Analysis
          The power-split hybrid technology is not included as an enabling technology in
          this analysis, although it is included as a baseline technology because it exists in
          the 2008 baseline fleet.

       •   2-Mode Hybrid (2MHEV) - is a hybrid electric drive system that uses an
          adaptation of a conventional stepped-ratio automatic transmission by replacing
          some of the transmission clutches with two electric motors that control the ratio of
          engine speed to vehicle speed, while clutches allow the motors to be bypassed.
          This improves both the transmission torque capacity for heavy-duty applications
          and reduces fuel consumption and CO2 emissions at highway speeds relative to
          other types of hybrid electric drive systems. The 2-mode hybrid technology is not
          included as an enabling technology in this analysis, although it is included as a
          baseline technology because it exists in the 2008 baseline fleet.

3.2 How did the agencies determine the costs of each of these technologies?

3.2.1   Direct Costs

       3.2.1.1 Costs from Tear-down Studies

       There are a number of technologies that have been costed using a rigorous  tear-down
method described in this section. As a general matter, the agencies believe that the best
method to derive technology cost estimates is to conduct studies involving tear-down and
analysis of actual vehicle  components. A "tear-down" involves breaking down a technology
into its fundamental parts  and manufacturing processes by completely disassembling actual
vehicles and vehicle subsystems and precisely determining what is required for its production.
The result of the tear-down is  a "bill of materials" for each and every part of the vehicle or
vehicle subsystem. This tear-down method of costing technologies is often used by
manufacturers to benchmark their products against competitive products. Historically,
vehicle and vehicle component tear-down has not been done on a large scale by researchers
and regulators due  to the expense required for such studies. While tear-down studies are
highly accurate at costing technologies for the year in which the study is intended, their
accuracy, like that of all cost projections, may diminish over time as costs are extrapolated
further into the future because of uncertainties in predicting commodities (and raw material)
prices, labor rates, and manufacturing practices. The projected costs may be higher or lower
than predicted.

       Over the past several years, EPA has contracted with FEV, Inc. and its subcontractor
Munro & Associates to conduct tear-down cost studies for  a number of key technologies
evaluated by the agencies in assessing the feasibility of future GHG and CAFE standards.
The analysis methodology included procedures to scale the tear-down results to smaller and
larger vehicles, and also to different technology configurations. FEV's methodology was
documented in a report published as part of the MY 2012-2016 rulemaking process, detailing
the costing of the first tear-down conducted in this work (#1 in the below list).   This report
was peer reviewed  by experts  in the industry and revised by FEV in response to the peer


                                             3-7

-------
                                    Technologies Considered in the Agencies' Analysis
review comments.3 Subsequent tear-down studies (#2-5 in the below list) were documented
in follow-up FEV reports made available in the public docket for the MY 2012-2016
rulemaking.4

      Since then, FEV's work under this contract work assignment has continued.
Additional cost studies have been completed and are available for public review.5  The most
extensive study, performed after the MY 2012-2016 Final Rule, involved whole-vehicle tear-
downs of a 2010 Ford Fusion power-split hybrid and a conventional 2010 Ford Fusion.  (The
latter served as a baseline vehicle for comparison.) In addition to providing power-split HEV
costs, the results for individual components in these vehicles were subsequently used to cost
another hybrid technology, the P2 hybrid, which employs similar hardware. This approach to
costing P2 hybrids was undertaken because P2 HEVs were not yet in volume production at
the time of hardware procurement for tear-down. Finally, an automotive  lithium-polymer
battery was torn down and costed to provide supplemental battery costing information to that
associated with the NiMH battery in the Fusion. This HEV cost work, including the extension
of results to P2 HEVs, has been extensively documented in a new report prepared by FEV.6
Because of the complexity and comprehensive scope of this HEV analysis, EPA
commissioned a separate peer review focused exclusively on it.  Reviewer comments
generally supported FEV's methodology and results, while including a number of suggestions
for improvement which were subsequently incorporated into FEV's analysis and final report.
The peer review comments and responses are available in the rulemaking docket.7 8

      Over the course of this work assignment, teardown-based studies were performed on
the technologies listed below. These completed studies provide a thorough evaluation of the
new technologies'  costs relative to their baseline (or replaced) technologies.

       1.  Stoichiometric gasoline direct injection (SGDI) and turbocharging with engine
          downsizing (T-DS) on a DOHC  (dual overhead cam) 14 engine, replacing a
          conventional DOHC  14 engine.
      2.  SGDI and T-DS on a SOHC (single overhead cam) on a V6 engine, replacing a
          conventional 3-valve/cylinder SOHC V8 engine.
      3.  SGDI and T-DS on a DOHC 14 engine, replacing a DOHC V6 engine.
      4.  6-speed automatic transmission (AT), replacing a 5-speed AT.
      5.  6-speed wet dual clutch transmission (DCT) replacing a 6-speed AT.
      6.  8-speed AT replacing a 6-speed AT.
      7.  8-speed DCT replacing a 6-speed DCT.
      8.  Power-split hybrid (Ford Fusion with 14 engine) compared to a conventional
          vehicle (Ford Fusion with V6). The results from this tear-down were extended to
          address P2 hybrids. In addition, costs from individual components in this tear-
          down study were used by the agencies in developing cost estimates for PHEVs and
          EVs.
      9.  Mild hybrid with stop-start technology (Saturn Vue with 14 engine), replacing a
          conventional 14 engine.
                                           3-8

-------
                                     Technologies Considered in the Agencies' Analysis
       10. Fiat Multi-Air engine technology. (Although results from this cost study are
          included in the rulemaking docket, they were not used by the agencies in this
          rulemaking's technical analyses.)

       In addition, FEV and EPA extrapolated the engine downsizing costs for the following
scenarios that were based on the above study cases:
       1.  Downsizing a SOHC 2 valve/cylinder V8 engine to a DOHC V6.
       2.  Downsizing a DOHC V8 to a DOHC V6.
       3.  Downsizing a SOHC V6 engine to a DOHC 4 cylinder engine.
       4.  Downsizing a DOHC 4 cylinder engine to a DOHC 3 cylinder engine.

       The agencies have relied on the findings of FEV for estimating the cost of the
technologies covered by the tear-down studies.  However, we note that FEV based their costs
on the assumption that these technologies would be  mature when produced in large  volumes
(450,000 units or more for each component or subsystem). If manufacturers are not able to
employ the technology at the volumes assumed in the FEV analysis with fully learned costs,
then the costs for each of these technologies would be expected to be higher.  There is also the
potential for stranded capital0 if technologies are introduced too rapidly for some indirect
costs to be fully recovered.  While the agencies consider the FEV tear-down analysis results
to be generally valid for the 2017-2025 timeframe for fully mature, high sales volumes, we
have had FEV perform supplemental analysis to consider potential stranded capital costs, and
have included these in our cost estimates. The issue of stranded capital is discussed in detail
in Section 3.2.2.3 of this draft TSD.

       3.2.1.2 Costs of HEV, PHEV, EV, and FCEVs

       The agencies have also reconsidered the costs for HEVs, PHEVs, EVs, and FCEVs
since the MY 2012-2016 rulemaking and the TAR as the result of two issues.  The first issue
is that electrified vehicle technologies are developing rapidly and we sought to capture the
results from the most recent analyses. The second issue is that the analysis for the MYs 2012-
2016 final rule employed a single  $/kWhr estimate,  and did not consider the specific vehicle
and technology application for the battery when we  estimated the cost of the battery.d
Specifically, batteries used in HEVs (high power density applications) versus EVs (high
energy density applications)  need  to be considered appropriately to reflect the design
differences, the chemical material usage differences, and the differences in cost per  kW-hr as
the power to energy ratio of the battery changes for  different applications. To address these
issues for this proposal, the agencies have used a battery cost model developed by Argonne
National Laboratory (ANL) for the Vehicle Technologies Program of the U.S. Department of
c The potential for stranded capital occurs when manufacturing equipment and facilities cannot be used in the
production of a new technology.
d However, we believe that this had little impact on the results of the cost analyses in support of the MYs 2012-
2016 final rule, as the agencies projected that the standards could be met with an increase of less than 2 percent
penetration of hybrid technology and no increase in plug-in or full electric vehicle technology.

-------
                                    Technologies Considered in the Agencies' Analysis
Energy (DoE) Office of Energy Efficiency and Renewable Energy.9 The model developed by
ANL allows users to estimate unique battery pack cost using user customized input sets for
different types of electrified powertrains, such as strong hybrid, PHEV and EV. Since the
publication of the TAR, ANL's battery cost model has been peer-reviewed and ANL has
updated the model to incorporate suggestions from peer-reviewers.10 This newly updated
model is used in this NPRM analysis and we discuss our updated battery costs in section in
Section  3.4.3.9.  The agencies also added new configurations of HEV, PHEV and EV to the
analysis that include the P2 HEV configuration,  two different all-electric mileage ranges for
PHEVs  (20 and 40 in-use miles) and three different mileage ranges for EVs (75, 100 and 150
in-use miles). Details regarding these vehicle technologies are discussed in sections 3.4.3.6.4
and 3.4.3.6.5.

      3.2.1.3 Direct Manufacturing Costs

        Building on the MYs 2012-2016 final rule, the agencies took a fresh look at
technology cost and effectiveness values for purposes of this joint NPRM.  For costs, the
agencies reconsidered both the direct or "piece"  costs and indirect costs of individual
components of technologies. For the direct costs that were not developed through the FEV
tear-down studies, the agencies generally followed a bill of materials (BOM) approach. A bill
of materials, in a general sense, is a list of components that make up a system—in this case,
an item  of fuel economy-improving technology.  In order to determine what a system costs,
one of the first steps is to determine its components and what they cost.

      NHTSA and EPA estimated these components and their costs based on a number of
sources  for cost-related information.  The objective was to use those sources of information
considered to be most credible for projecting the costs of individual vehicle technologies.  For
those cost estimates that are fundamentally unchanged since the 2012-2016 final rule and/or
the 2010 TAR (we make note of these in Section 3.4, below), we have a full description of the
sources  used in Chapter 3 of the final joint TSD  supporting that rule.11'12  For those costs that
have been updated since those analyses (e.g., battery pack cost, costs based on more recent
tear down analyses, etc.), we note their sources in Section 3.4, below. We have also
considered input from manufacturers and suppliers gathered either through meetings
following the 2010 TAR or in comment submitted in response to the 2010 TAR, some of
which cannot be shared publicly in detailed form but, where used, we make note of it without
violating its confidentiality.  Note that a summary of comments on the 2010 TAR, with the
agencies' responses, was published as a "Supplemental Notice of Intent" in December of
2010.13  As discussed throughout this chapter, the agencies have reviewed, revalidated or
updated cost estimates for individual components based on the latest information available.

        Once costs were determined, they were adjusted to ensure that they were all expressed
in 2009  dollars using the GDP price deflator as described in section 3.2.4. Indirect costs were
accounted for using the ICM approach developed by EPA and explained below. NHTSA and
EPA also considered how costs should be adjusted to reflect manufacturer learning as
discussed below. Additionally, costs were adjusted by modifying or scaling content
assumptions to account for differences across the range of vehicle sizes and functional

                                            3-10

-------
                                     Technologies Considered in the Agencies' Analysis
requirements, and adjusted the associated material cost impacts to account for the revised
content, although these adjustments were different for each agency due to the different vehicle
subclasses used in their respective models.

3.2.2   Indirect Costs

       3.2.2.1 Indirect Cost Multiplier Changes

       As discussed in greater detail below, the agencies have revised the markups used to
estimate indirect costs. The first change was to normalize the ICM values to be consistent
with the historical average retail price equivalent (RPE) of 1.5, rather than the single year that
the RTI study examined.  This was done by applying a factor of .57.46 to all indirect cost
elements.  The second change was to re-consider the markup factors and the data used to
generate them.  The result on this new thinking is to increase the markup in all cases. The
final change is the way in which the ICM factors are applied. In previous analyses ICMs
were applied to the learned value of direct costs.  However, since learning influences direct
costs only, the agencies were concerned that this could overstate the impact of learning  on
total costs.  Indirect costs are thus now established based on the initial value of direct costs
and held constant until the long-term ICM is applied.  This is done for all ICM factors except
warranties, which are influenced by the learned value of direct costs.

       3.2.2.2 Cost markups to account for indirect costs

       To produce a unit of output, auto manufacturers incur direct and indirect costs.  Direct
costs include the cost of materials and labor costs. Indirect costs may be related to production
(such as research and development [R&D]), corporate operations (such as salaries, pensions,
and health care costs for corporate staff), or selling (such as transportation, dealer support, and
marketing).  Indirect costs are generally recovered by allocating a share of the costs to each
unit of goods sold. Although it is possible to account for direct costs allocated to each unit of
goods sold, it is more challenging to account for indirect costs allocated to a unit of goods
sold. To make a cost analysis process more feasible, markup factors, which relate total
indirect costs to total direct costs, have been developed. These factors are often referred to as
retail price equivalent (RPE) multipliers.

       Cost analysts and regulatory agencies including EPA and NHTSA have frequently
used these multipliers  to estimate the resultant impact on costs associated with manufacturers'
responses to regulatory requirements. The best approach to determining the impact of
changes in direct manufacturing costs on a manufacturer's indirect costs would be to actually
estimate the cost impact on each indirect cost element. However, doing this within the
constraints of an agency's time or budget is not always feasible, and the technical, financial,
and accounting information to carry out such an analysis may simply be unavailable.

       RPE multipliers provide, at an aggregate level, the relative shares of revenues
(Revenue = Direct Costs + Indirect Costs + Net Income) to direct manufacturing costs.  Using
RPE multipliers implicitly assumes that incremental changes in direct manufacturing costs
                                             3-11

-------
                                     Technologies Considered in the Agencies' Analysis
produce common incremental changes in all indirect cost contributors as well as net income.
A concern in using the RPE multiplier in cost analysis for new technologies added in response
to regulatory requirements is that the indirect costs of vehicle modifications are not likely to
be the same for different technologies. For example, less complex technologies could require
fewer R&D efforts or less warranty coverage than more complex technologies. In addition,
some simple technological adjustments may, for example, have no effect on the number of
corporate personnel and the indirect costs attributable to those personnel.  The use of RPEs,
with their assumption that all technologies have the same proportion of indirect costs, is likely
to overestimate the costs of less complex technologies and underestimate the costs of more
complex technologies.

       To address this concern, the agencies have developed modified multipliers.  These
multipliers are referred to as indirect cost multipliers (ICMs). In contrast to RPE multipliers,
ICMs assign unique incremental changes to each indirect cost contributor


               ICM = (direct cost + adjusted indirect cost + profit)/(direct cost)

       Developing the ICMs from the RPE multipliers requires developing adjustment factors
based on the complexity of the technology and the time frame under consideration. This
methodology was used in the cost estimation for the MYs 2012-2016 final rule.  The ICMs
were developed in a peer-reviewed report from RTI International and were subsequently
discussed in a peer-reviewed journal article.14 Note that the cost of capital (reflected in profit)
is included because of the assumption implicit in ICMs (and RPEs) that capital costs are
proportional to direct costs, and businesses  need to be able to earn returns on their
investments. The capital costs are those associated with the incremental costs of the new
technologies.

       As noted above, for the analysis supporting this proposed rulemaking, the agencies are
again using the ICM approach but have made some changes to both the ICM factors and to
the method of applying those factors to arrive at a final cost estimate.  The first of these
changes was done in response to continued thinking among the EPA-NHTSA team about how
past ICMs have been developed and what are  the most appropriate data sources to rely upon
in determining the appropriate ICMs. The second change has been done both due to staff
concerns and public feedback suggesting that the agencies were inappropriately applying
learning effects to indirect costs via the multiplicative approach to  applying the ICMs.

       Regarding the first change - to the ICM factors themselves - a little background must
first be provided. In the original work done under contract to EPA by RTI International,15
EPA experts had undertaken a consensus approach to determining the impact of specific
technology changes on the indirect costs of a company.  Subsequent to that effort, EPA
experts conducted a blind survey to make this determination on a different set of technology
changes. This  subsequent effort, referred to by EPA as a modified-Delphi approach, resulted
in slightly different ICM determinations.  This effort is detailed in a memorandum contained
in the docket for this rule.16  Upon completing this effort, the EPA  team determined that the
                                            3-12

-------
                                     Technologies Considered in the Agencies' Analysis
original RTI values should be averaged with the modified-Delphi values to arrive at the final
ICMs for low and medium complexity technologies and that the original RTI values would be
used for high complexity level 1 while the modified-Delphi values would be used for high
complexity level 2. These final ICMs as described were used in the MYs 2012-2016 light-
duty GHG/CAFE rulemaking.

       More recently, EPA and NHTSA  decided that the original light-duty RTI values,
because of the technologies considered for low and medium complexity, should no longer be
used and that we should rely solely on the modified-Delphi values for these complexity levels.
The original light-duty RTI study used low rolling resistance tires as a low complexity
technology example and a dual clutch transmission as a medium complexity technology.
Upon further thought, the technologies considered for the modified Delphi values (passive
aerodynamic improvements for low complexity and turbocharging with downsizing for
medium complexity)  were considered to better represent the example technologies.  As a
result, the modified-Delphi values became the working ICMs for low and medium complexity
rather than averaging those values with the original RTI report values. NHTSA and EPA staff
also re-examined the  technology complexity categories that were assigned to each light-duty
technology and modified these assignments to better reflect the technologies that are now
used as proxies to determine each category's ICM value.

       A secondary-level change was also made as part of this ICM recalculation to the light-
duty ICMs.  That change was to revise upward the RPE level reported in the original RTI
report from an original value of 1.46 to 1.5 to reflect the long term average RPE.  The original
RTI study was based  on 2007 data.  However, an analysis of historical RPE data indicates
that, although there is year to year variation, the average RPE has remained roughly 1.5.
ICMs will be applied to future year's data and therefore NHTSA and EPA staff believe that it
would be appropriate to base ICMs on the historical average rather than a single year's result.
Therefore, ICMs in this proposed rulemaking were adjusted to reflect this average level. As a
result, the High  1  and High 2 ICMs have also changed.

       Table 3-1 shows both the ICM values used in the MYs 2012-2016 final rule and the
new ICM values used for the analysis supporting these proposed rules. Near term values
account for differences in the levels of R&D, tooling, and other indirect costs that will be
incurred. Once the program has been fully implemented, some of the indirect costs will no
longer be attributable to the standards and, as such, a lower ICM factor is applied to direct
costs.

                    Table 3-1 Indirect Cost Multipliers Used in this Analysis3

Complexity
Low
Medium
Highl
High2
20 12-20 16 Rule
Near term
1.17
1.31
1.51
1.70
Long term
1.13
1.19
1.32
1.45
This Proposal
Near term
1.24
1.39
1.56
1.77
Long term
1.19
1.29
1.35
1.50
          a Rogozhin, A., et. al., "Using indirect cost multipliers to estimate the total cost of
                                            3-13

-------
                                      Technologies Considered in the Agencies' Analysis
          adding new technology in the automobile industry," International Journal of Production
          Economics (2009); "Documentation of the Development of Indirect Cost Multipliers
          for Three Automotive Technologies," Helfand, G., and Sherwood, T., Memorandum
          dated August 2009; "Heavy Duty Truck Retail Price Equivalent and Indirect Cost
          Multipliers," Draft Report prepared by RTI International and Transportation Research
          Institute, University of Michigan, July 2010

       The second change made to the ICMs has to do with the way in which they are
applied.  To date, we have applied the ICMs, as done in any analysis that relied on RPEs, as a
pure multiplicative factor. This way, a direct manufacturing cost of, say, $100 would be
multiplied by an ICM of 1.24 to arrive at a marked up technology cost of $124. However, as
learning effects (discussed below) are applied to the direct manufacturing cost, the indirect
costs are also reduced accordingly. Therefore, in year two the $100 direct manufacturing cost
might reduce to $97, and the marked up cost would become $120 ($97 x 1.24). As a result,
indirect costs would be reduced from $24 to $20. Given that indirect costs cover many things
such as facility-related costs, electricity, etc., it is perhaps not appropriate to apply the ICM to
the learned direct costs, at least not for those indirect cost elements unlikely to change with
learning.  The EPA-NHTSA team believes that it is appropriate to allow only warranty costs
to decrease with learning, since warranty costs are tied to direct manufacturing costs (since
warranty typically involves replacement of actual parts which should be less costly with
learning). The remaining elements of the indirect costs should  remain constant year-over-
year, at least until some of those indirect costs are no longer attributable to the rulemaking
effort that imposed them (such as R&D).

As a result, the ICM calculation has become more complex with the analysis supporting this proposal. We
must first establish the year in which the direct manufacturing costs are considered "valid." For example,
a cost estimate might be considered valid today, or perhaps not until high volume production is reached—
 which will not occur until MY 2015 or later. That year is known as the base year for the estimated cost.
That cost is the cost used to determine the "non-warranty" portion of the indirect costs. For example, the
non-warranty portion of the medium complexity ICM  in the short-term is 0.343 (the warranty versus non-
                          warranty portions of the ICMs are shown in

       Table 3-2).  For the dual cam phasing (DCP) technology on an 14 engine we have
estimated a direct manufacturing cost of $70 in MY 2015.  So the non-warranty portion of the
indirect costs would be $24.01 ($70 x 0.343).  This value would be added to the learned direct
manufacturing cost for each year through 2018, the last year of short term indirect costs.
Beginning in 2019, when long-term indirect costs begin, the additive factor would become
$18.13 ($70 x 0.259). Additionally, the $70 cost in 2015 would become $67.90 in MY 2016
due to learning ($70 x (1-3%)). So, while the warranty portion of the indirect costs would be
$3.15 ($70 x 0.045) in 2015, indirect costs would decrease to $3.06 ($67.90 x 0.045) in 2016
as warranty costs decrease with learning. The resultant indirect costs for the DCP-I4
technology would be $27.16 ($24.01+$3.15) in MY 2015 and $27.07 ($24.01+$3.06) in
MY2016, and so on for subsequent years.
                    Table 3-2 Warranty and Non-Warranty Portions of ICMs
                     |           Near term           |          Long term

                                              3-14

-------
                                     Technologies Considered in the Agencies' Analysis
Complexity
Low
Medium
Highl
High2
Warranty
0.012
0.045
0.065
0.074
Non-warranty
0.230
0.343
0.499
0.696
Warranty
0.005
0.031
0.032
0.049
Non-warranty
0.187
0.259
0.314
0.448
       There is some level of uncertainty surrounding both the ICM and RPE markup factors.
The ICM estimates used in this proposal group all technologies into three broad categories
and treat them as if individual technologies within each of the three categories (low, medium,
and high complexity) will have exactly the same ratio of indirect costs to direct costs.  This
simplification means it is likely that the direct cost for some technologies within a category
will be higher and some lower than the estimate for the category in general. Additionally, the
ICM estimates were developed using adjustment factors developed in two separate occasions:
the first, a consensus process, was reported in the RTI report; the second, a modified Delphi
method, was conducted separately  and reported in an EPA memorandum. Both these panels
were composed of EPA staff members with previous background in the automobile industry;
the memberships of the two panels overlapped but were not the same.  The panels evaluated
each element of the industry's RPE estimates and estimated the degree to which those
elements would be expected to change in proportion to changes in direct manufacturing costs.
The method and the estimates in the RTI report were peer reviewed by three industry experts
and subsequently by reviewers for  the International Journal of Production Economics.17
However,  the ICM estimates have not yet been validated through a direct accounting of actual
indirect costs for individual technologies. RPEs themselves are also inherently difficult to
estimate because the accounting statements of manufacturers do not neatly categorize all cost
elements as either direct or indirect costs. Hence, each researcher developing an RPE
estimate must apply a certain amount of judgment to the allocation of the costs.  Since
empirical estimates of ICMs are ultimately derived from the same data used to measure RPEs,
this affects both measures. However,  the value of RPE has not been measured for specific
technologies, or for groups of specific technologies.  Thus applying a single average RPE to
any given  technology by definition overstates costs  for very simple technologies, or
understates them for advanced technologies.

       3.2.2.3 Stranded capital

       Because the production of automotive components is capital-intensive, it is possible
for substantial capital investments in manufacturing equipment and facilities to become
"stranded" (where their value is lost, or diminished).  This would occur when the capital is
rendered useless (or less useful)  by some factor that forces a major change  in vehicle design,
plant operations, or manufacturer's product mix, such as a shift in consumer demand for
certain vehicle types. It can also be caused by new standards that phase-in at a rate too rapid
to accommodate planned replacement or redisposition of existing capital to other activities.
The lost value of capital equipment is  then amortized in some way over production of the new
technology components.
                                            3-15

-------
                                     Technologies Considered in the Agencies' Analysis
       It is difficult to quantify accurately any capital stranding associated with new
technology phase-ins under the proposed standards because of the iterative dynamic involved
- that is, the new technology phase-in rate strongly affects the potential for additional cost due
to stranded capital, but that additional cost in turn affects the degree and rate of phase-in for
the same or other individual competing technologies. In addition, such an analysis is very
company-, factory-, and manufacturing process-specific, particularly in regard to finding
alternative uses for equipment and facilities. Nevertheless, in order to account for the
possibility of stranded capital costs, the agencies asked FEV to  perform an analysis, using
conservative assumptions, of the potential stranded capital costs associated with rapid phase-
in of technologies due to new standards, using data from FEV's primary teardown-based cost
analyses.18

       The assumptions made in FEV's stranded capital analysis with potential for major
impacts on  results are:

       •  All manufacturing equipment was bought brand new when the old technology
          started production (no carryover of equipment used to make the previous
          components that the old technology itself replaced).
       •   10-year normal production runs:  Manufacturing equipment used to make old
          technology components is straight-line depreciated over a 10-year life.
       •  Factory managers do not optimize capital equipment phase-outs  (that is, they are
          assumed to routinely repair and replace equipment without regard to whether or
          not it will soon be scrapped due to adoption of new vehicle technology).
       •  Estimated stranded capital is amortized over 5 years of annual production at
          450,000 units (of the new technology components).  This annual production is
          identical to that assumed in FEV's primary teardown-based cost analyses. The 5-
          year recovery period is  chosen to help ensure a conservative analysis; the actual
          recovery would of course vary greatly with market conditions.

       FEV assembled a team of manufacturing experts to perform the analysis, using a
methodology with the following key steps for each vehicle technology scenario:

       1) Identify all of the old technology components that are no longer used or that are
          modified in the new technology vehicles (from the comparison bills of materials
          developed in the primary teardown-based analyses).

       2) For each of these components identify the manufacturing equipment and tooling
          needed to make it.

       3) Estimate the new-purchase $ value of each item identified in step 2.

       4) Assign an "Investment  Category" to each equipment item identified in step 2,
          based on an assessment by FEV's experts of recoverable value:

          •   Flexible: Equipment can be used to manufacture new technology or other parts
              (0% stranded)

                                            3-16

-------
                                     Technologies Considered in the Agencies' Analysis
          •   Re-Useable: Equipment can be used in alternative industries, sold at 50% of its
              remaining value (50% stranded)
          •   Semi-Dedicated: Estimate that 50% of equipment is flexible (50% stranded)
          •   Dedicated: Custom manufacturing equipment (100% stranded)

       5)  Assign an "Investment Category" to each tooling item identified in step 2, based
          on an assessment by FEV's experts of recoverable value:

          •   Flexible: Can be used for manufacturing new technology parts (0% stranded)
          •   Perishable: Frequent replacement of tooling (0% stranded)
          •   Semi-Dedicated Tooling: Estimate that 50%  of tooling is dedicated (50%
              stranded)
          •   Dedicated: Commodity-specific (100% stranded)

       6)  Multiply the %  stranding values from steps 4 and 5 by the $ values from step 3.

       7)  Multiply the results in step 6 by 70%, 50%, and 20% for 3-, 5-, and 8-year
          stranding scenarios, respectively. That is, an old technology, for which production
          is truncated prematurely after only 8 years, will experience the stranding of 20%
          (the last 2 years of its 10-year normal production run) of its associated remaining
          capital value.

       8)  Sum the results in step  7 to obtain overall stranded capital costs.

       9)  Divide the results in step 8 by 2,250,000  (5 years x 450,000 units/year) to obtain
          $/vehicle values, applicable to new technology vehicles for the 1st 5 years of their
          production due to the assumed 5-year recovery period.

       The stranded capital analysis was performed for three transmission technology
scenarios, two engine technology scenarios, and one hybrid  technology scenario, as shown in
Table 3-3. The methodology used by EPA in applying these results to the technology costs is
described in Chapter 3 of EPA's draft RIA. The methodology used by NHTSA in applying
these results to the technology costs is described in NHTSA's preliminary RIA section V.

                Table 3-3 Stranded Capital Analysis Results (2009 dollars/vehicle)
Replaced
technology
6-speed AT
6-speed AT
6-speed DCT
Conventional V6
Conventional V8
Conventional V6
New
technology
6-speed DCT
8 -speed AT
8-speed DCT
DSTGDI 14
DSTGDI V6
Power-split HEV
Stranded capital cost per vehicle
when replaced technology's production is ended
after:
3 years
$55
$48
$28
$56
$60
$111
5 years
$39
$34
$20
$40
$43
$79
8 years
$16
$14
$8
$16
$17
$32
   DSTGDI=Downsized, turbocharged engine with stoichiometric gasoline direct injection.
                                            3-17

-------
                                     Technologies Considered in the Agencies' Analysis
3.2.3   Cost reduction through manufacturer learning

       For this proposal, we have not changed our estimates of learning and how learning
will impact costs going forward from what was employed in the analysis for the MYs 2012-
2016 light-duty vehicle rule. However, we have updated our terminology in an effort to
clarify that we consider there to be one learning effect—learning by doing—which results in
cost reductions occurring with every doubling of production.6 In the past, we have referred to
volume-based and time-based learning.  Our terms were meant only to denote where on the
volume learning curve a certain technology was—"volume-based learning" meant the steep
portion of the curve where learning effects are greatest, while "time-based learning" meant
the flatter portion of the curve where learning effects are less pronounced. Unfortunately, our
terminology led some to believe that we were implementing two completely different types of
learning—one based on volume of production and the other based on time in production. Our
new terminology—steep portion of the curve and flat portion of curve—is simply meant to
make more clear that there is one learning curve and some technologies can be considered to
be on the steep portion while others are well into the flatter portion of the curve. These two
portions of the volume learning curve are shown in Figure 3-1.
e Note that this new terminology was described in the recent heavy-duty GHG final rule (see 76 FR 57320). The
learning approach used in this analysis is entirely consistent with that used and described for that analysis.

                                            3-18

-------
                                     Technologies Considered in the Agencies' Analysis
                          Volume Learning Curve - Steep & Flat Portions
          120%
          100%
                            Steep portion of volume learning curve
                                           Flat portion of volume learning curve
           0%
                                        Cumulative Production
                 Figure 3-1 Steep & Flat Portions of the Volume Learning Curve

       For some of the technologies considered in this analysis, manufacturer learning effects
would be expected to play a role in the actual end costs.  The "learning curve" or "experience
curve" describes the reduction in unit production costs as a function of accumulated
production volume. In theory, the cost behavior it describes applies to cumulative production
volume measured at the level of an individual manufacturer, although it is often assumed—as
both agencies have done in past regulatory analyses—to  apply at the industry-wide level,
particularly in industries like the light duty vehicle production industry that utilize many
common technologies and component supply sources. Both agencies believe there are indeed
many factors that cause costs to decrease over time.  Research in the costs of manufacturing
has consistently shown that, as manufacturers gain experience in production, they are able to
apply innovations to simplify machining and assembly operations, use lower cost materials,
and reduce the number or complexity of component parts.  All of these factors allow
manufacturers to lower the per-unit cost of production. We refer to this phenomenon as the
manufacturing learning curve.

       NHTSA and EPA included a detailed description of the learning effect in the MYs
2012-2016 light-duty rule  and the more recent heavy-duty rule.19  Most studies of the effect of
experience or learning on production costs appear to assume that cost reductions begin only
after some initial volume threshold has been reached, but not all of these studies specify this
threshold volume.  The rate at which costs decline beyond the initial threshold is usually
                                             3-19

-------
                                       Technologies Considered in the Agencies' Analysis
expressed as the percent reduction in average unit cost that results from each successive
doubling of cumulative production volume, sometimes referred to as the learning rate. Many
estimates of experience curves do not specify a cumulative production volume beyond which
cost reductions would no longer occur, instead depending on the asymptotic behavior of the
effect for learning rates below 100 percent to establish a floor on costs.

       In past rulemaking analyses, as noted above, both agencies have used a learning curve
algorithm that applied a learning factor of 20 percent for each doubling of production volume.
NHTSA has used this approach in analyses supporting recent CAFE rules.  In its analyses,
EPA has simplified the approach by using an "every two years" based learning progression
rather than a pure production volume progression (i.e., after two years of production it was
assumed that production volumes would have doubled and, therefore, costs would be reduced
by 20 percent)/

       In the MYs 2012-2016 light-duty rule and the recent heavy-duty GHG final rule, the
agencies employed an additional learning algorithm to reflect the volume-based learning cost
reductions that occur further along on the learning curve.  This additional learning algorithm
was termed "time-based" learning in the 2012-2016 rule simply as a means of distinguishing
this algorithm from the volume-based algorithm mentioned above, although both of the
algorithms reflect the volume-based learning curve supported in the literature.  As described
above, we are now referring to this learning algorithm as the "flat portion" of the learning
curve.  This way, we maintain the clarity that all learning is, in fact, volume-based learning,
and that the level of cost reductions depend only on where on the learning curve a
technology's learning progression is.  We distinguish the flat portion of the curve from the
steep portion of the curve to indicate the level of learning taking place in the years following
implementation of the technology (see Figure 3-1).  The agencies have applied learning
effects on the steep portion of the learning curve for those technologies considered to be
newer technologies likely to experience rapid cost reductions through manufacturer learning,
and learning effects  on the flat portion learning curve for those technologies considered to be
more mature technologies likely to experience only minor cost reductions through
manufacturer learning. As noted above, the  steep portion learning algorithm results in 20
f To clarify, EPA has simplified the steep portion of the volume learning curve by assuming that production
volumes of a given technology will have doubled within two years time. This has been done largely to allow for
a presentation of estimated costs during the years of implementation, without the need to conduct a feedback
loop that ensures that production volumes have indeed doubled.  If we were to attempt such a feedback loop, we
would need to estimate first year costs, feed those into OMEGA, review the resultant technology penetration rate
and volume increase, calculate the learned costs, feed those into OMEGA (since lower costs would result in
higher penetration rates, review the resultant technology penetration rate and volume increase, etc., until an
equilibrium was reached. To do this for all of the technologies considered in our analysis is simply not feasible.
Instead, we have estimated the effects of learning on costs, fed those costs into OMEGA, and reviewed the
resultant penetration rates. The assumption that volumes have doubled after two years is based solely on the
assumption that year two sales are of equal or greater number than year one sales and, therefore, have resulted in
a doubling of production. This could be done on a daily basis, a monthly basis, or, as we have done, a yearly
basis.

                                               3-20

-------
                                     Technologies Considered in the Agencies' Analysis
percent lower costs after two full years of implementation (i.e., the MY 2016 costs would be
20 percent lower than the MYs 2014 and 2015 costs). Once two steep portion learning steps
have occurred, flat portion learning at 3 percent per year becomes  effective for 5 years.
Beyond 5 years of learning at 3 percent per year, 5 years of learning at 2 percent per year,
then 5 at 1 percent per year become effective.

       Learning effects are applied to most but not all technologies because some of the
expected technologies are already used rather widely in the industry and we therefore assume
that learning impacts have already occurred. The steep portion learning algorithm was
applied for only a handful of technologies that are considered to be new or emerging
technologies. Most technologies have been considered to be more established given their
current use in the fleet and, hence, the lower flat portion learning algorithm has been applied.
The learning algorithms applied to each technology and the applicable timeframes are
summarized in Table 3-4.

         Table 3-4 Learning Effect Algorithms Applied to Technologies Used in this Analysis
Technology
Engine modifications to accommodate low
friction lubes
Engine friction reduction - level 1 & 2
Lower rolling resistance tires - level 1
Low drag brakes
Secondary axle disconnect
Electric/Plug-in vehicle battery charger
installation labor
Variable valve timing
Variable valve lift
Cylinder deactivation
Stoichiometric gasoline direct injection
Aggressive shift logic - level 1 & 2
Early torque converter lockup
5/6/7/8 speed auto transmission
6/8 speed dual clutch transmission
High efficiency gearbox
Improved accessories - level 1 & 2
Electronic/electro-hydraulic power steering
Aero improvements - level 1 & 2
Conversion to DOHC without reducing # of
cylinders
Air conditioner related hardware
Air conditioner alternative refrigerant
Cooled EGR
Conversion to Atkinson cycle
Turbocharging & downsizing
Steep learning




















2016-2020



Flat learning




2012-2025

2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-2025
2012-20205
2021-2025
2012-2025
2012-2025
2012-2025
No learning
2012-2025
2012-2025
2012-2025
2012-2025

2012-2025


















                                            3-21

-------
                                      Technologies Considered in the Agencies' Analysis
Mass reduction
Advanced diesel
Hybrid/Electric/Plug-in vehicle non-battery
components
P2 Hybrid vehicle battery-pack components
Electric/Plug-in vehicle battery-pack
components
Electric/plug-in vehicle battery charger
components
Stop-start
Lower rolling resistance tires - level 2



2012-2016
2012-2025a
2012-2025a
2012-2015
2017-2021
2012-2025
2012-2025
2012-2025
2017-2025


2016-2025
2022-2025








 Note that the steep learning effects have for EV and PHEV battery packs and charger components have been
carried through 5 learning cycles but at a decelerated pace as described in the text.

        The learning effects discussed here impact the technology costs in that those
 technology costs for which learning effects are considered applicable are changing throughout
 the period of implementation and the period following implementation.  For example, some of
 the technology costs considered in this analysis are taken from the MY 2012-2016 light-duty
 rule.  Many of the costs in the 2012-2016 light-duty rule were considered "applicable" for the
 2012 model year. If flat-portion learning were applied to those technologies, the 2013 cost
 would be 3 percent lower than the 2012 cost, and the 2014 model year cost 3 percent lower
 than the 2013  cost, etc. As a result, the 2017-2025 costs for a given technology used in this
 analysis reflect those years of flat learning and would not be identical to  the 2012 model year
 cost for that same technology presented in the 2012-2016 light-duty rule.

        Because of the nature of battery pack development (i.e., we are arguably still in the
 research phase for the types of batteries considered in this proposal, and  cost reduction
 through manufacturer-based learning has only just begun, if it has begun at all), the agencies
 have carried the learning curve through five steep based learning steps although at a
 somewhat slower pace than every two years. This has been done in an effort to maintain the
 shape of a traditional learning curve.  This curve was developed by using the ANL BatPaC
 model costs as direct manufacturing costs applicable in the 2025 MY.  We have then
 unlearned those costs back to 2012 using the curve shown in Figure 3-2. This is the same
 curve used in the 2010 TAR (see 2010 TAR at page B-22).  This allows the agencies to
 estimate costs in MYs 2017 through 2025, as well as those costs in each  year back to MY
 2012, if desired. As noted, this learning curve consists of 5 full learning steps on the steep
 portion of the  learning curve, each of which results in costs being reduced 20 percent relative
 to the prior step. These learning steps are shown occurring every two years beginning in 2012
 until 2020, at which time a 5 year gap is imposed until 2025 when the fifth steep learning step
 occurs. Beyond 2025, learning on the flat portion of the curve begins at  3 percent per year
 cost reductions.  The smooth line shows a logarithmic curve fit applied to the learning curve
 as the agencies' cost model would apply learning.
                                              3-22

-------
                                     Technologies Considered in the Agencies' Analysis
        4.000
        3.500
        0.000
              2012
2016
                             •Cost Model
2020          2024
                   • Log. (Cost Model)
2028
     Figure 3-2 Learning Curve used for EV & PHEV Battery-Packs and In-Home Charger Costs
3.2.4   Costs Updated to 2009 Dollars

       This change is simply to update any costs presented in earlier analyses to 2009 dollars
using the GDP price deflator as reported by the Bureau of Economic Analysis on January 27,
2011. The factors used to update costs from 2007 and 2008 dollars to 2009 dollars are shown
below. For the final rule, we may move to 2010 dollars but, for this analysis, given the timing
of conducting modeling runs and developing inputs to those runs, the factors for converting to
2010 dollars were not yet available.

Price Index for Gross Domestic Product
Factor applied to convert to 2009 dollars
2007
106.3
1.031
2008
108.6
1.009
2009
109.6
1.00
Source: Bureau of Economic Analysis, Table 1.1.4. Price Indexes for Gross Domestic Product, downloaded
1/27/2011, last revised 12/22/2010.
3.3 How did the agencies determine effectiveness of each of these technologies?

       The agencies determined the effectiveness of each individual technology with a
process similar to the one used for the 2012-2016 light duty vehicle GHG and CAFE
standards. The individual effectiveness of several technologies discussed in this rule that
were present in the earlier rule were left largely unchanged while others were updated.  EPA
                                            3-23

-------
                                     Technologies Considered in the Agencies' Analysis
and NHTSA reviewed recent confidential manufacturer estimates of technology effectiveness
and found them to be generally consistent with our estimates. Additionally, EPA used vehicle
simulation modeling to gain further insight on existing and new technologies for this
rulemaking. EPA conducted a vehicle simulation project (described in 3.3.1) that included a
majority of the proposed technologies, the results of which:

       •  informed existing individual technology effectiveness values,
       •  provided data for newly introduced technologies, and
       •  most importantly, provided an interactive data source with which to update and
          calibrate the new LP model

       The lumped parameter model then  served as the primary tool in evaluating the
individual technology effectiveness estimates, the combined effectiveness of groups of
technologies (or packages) and synergy factors, as described in 3.3.2.  The effectiveness
values, in conjunction with costs, were then applied to vehicles across the fleet for use in the
Agencies' respective compliance models.

3.3.1   Vehicle simulation modeling

       3.3.1.1 Background

       For regulatory purposes, the fuel economy of any given vehicle is determined by
placing the vehicle on a chassis dynamometer (akin to a large treadmill that puts the vehicle's
wheels in contact with one or more rollers, rather than with a belt stretched between rollers) in
a controlled environment, driving the vehicle over a specific driving cycle  (in which driving
speed is specified for each second of operation), measuring the amount of carbon dioxide
emitted from the vehicle's tailpipe, and calculating fuel consumption based on the density and
carbon content of the fuel.

       One means of determining the effectiveness of a given technology as applied to a
given vehicle model would be to measure the vehicle's fuel economy on a  chassis
dynamometer, install the new technology,  and then re-measure the vehicle's fuel economy.
However,  most technologies cannot simply be "swapped out," and even for those that can,
simply doing so without additional engineering work  may change other vehicle characteristics
(e.g., ride, handling, performance, etc.), producing an "apples to  oranges" comparison.

       Some technologies can also be more narrowly characterized through bench or engine
dynamometer (i.e., in which the engine drives a generator that is, in turn, used to apply a
controlled load to the engine) testing.  For example, engine dynamometer testing could be
used to evaluate the brake-specific fuel consumption (e.g., grams per kilowatt-hour) of a
given engine before and after replacing the engine oil with  a less viscous oil. However, such
testing does not provide a direct measure of overall vehicle fuel economy or changes in
overall vehicle fuel economy.
                                            3-24

-------
                                     Technologies Considered in the Agencies' Analysis
       For a vehicle that does not yet exist, as in the agencies' analyses of CAFE and GHG
standards applicable to future model years, even physical testing can provide only an estimate
of the vehicle's eventual fuel economy.  Among the alternatives to physical testing,
automotive engineers involved in vehicle design make use of computer-based analysis tools,
including a powerful class of tools commonly referred to as "full vehicle simulation."  Given
highly detailed inputs regarding vehicle engineering characteristics, full vehicle simulation
provides a means of estimating vehicle fuel consumption over a given drive cycle, based on
the explicit representation of the physical laws governing vehicle propulsion and dynamics.
Some vehicle simulation tools also incorporate combustion simulation tools that represent the
combustion cycle in terms of governing physical and chemical processes. Although these
tools are computationally intensive and required a great deal of input data, they provide
engineers involved in vehicle development and design with an alternative that can be
considerably faster and less expensive than physical experimentation and testing.

       Properly executed, methods such as physical testing and full vehicle simulation can
provide reasonably (though not absolutely) certain estimates of the vehicle fuel economy of
specific vehicles to be produced in the future. However, when analyzing potential CAFE and
GHG standards, the agencies are not actually designing specific vehicles. The agencies are
considering implications of new standards that will apply to the average performance of
manufacturers' entire production lines.  For this type of analysis, precision in the estimation
of the fuel economy of individual vehicle models is not essential; although it is important that
the agency avoid systematic upward or downward bias, uncertainty at the level of individual
models is mitigated by the fact that compliance with CAFE and GHG standards is based on
average fleet performance.

       DOT's CAFE model and EPA's OMEGA are not full vehicle simulation models.
Both models use higher-level estimates of the efficacy of different technologies or technology
packages.  Both models apply methods to avoid potential double-counting of efficacy
addressing specific energy loss mechanisms  (e.g., pumping losses), and for this NPRM, both
agencies applied estimates using EPA's lumped parameter model, which was updated using
results of full vehicle simulation performed by Ricardo, PLC.  Although full vehicle
simulation could, in principle, be fully integrated into the agencies' model-by-model analyses
of the entire fleet to be projected to be produced in future model years, this level of
integration would be infeasible considering the size and complexity of the fleet. Also,
considering the forward-looking nature of the agencies' analyses, and the amount of
information required to perform full vehicle  simulation, this level of integration would
involve misleadingly precise estimates of fuel consumption and COi emissions.

       Still, while the agencies have used results of full vehicle simulation to inform the
development of model inputs for performing fleet-level analysis, information from other
sources  (e.g.,  vehicle testing) could be considered when developing such model inputs.
Before performing  analysis to support the evaluation and finalization of post-2016 CAFE and
GHG standards, the agencies will revisit estimates of technology efficacy for use DOT's
CAFE model and EPA's OMEGA model, and invite comment on the use of information from
full vehicle simulation and other sources. Related, DOT has, as discussed above, contracted

                                            3-25

-------
                                     Technologies Considered in the Agencies' Analysis
with Argonne National Laboratory (ANL) to provide additional full vehicle simulation
modeling support for this MYs 2017-2025 rulemaking, and anticipates that results will be
available for use in developing inputs for the final rule.

       3.3.1.2 2011 Ricardo Simulation Study

       For this rule EPA built upon its 2008 vehicle simulation project20 used to support the
2012-2016 light duty vehicle GHG and CAFE standards (reference). As in the initial project,
the technical work was conducted by the global engineering consulting firm, Ricardo, Inc.
(under subcontract to SRA Corporation), using its MSC.EASY5 dynamic vehicle simulation
model. This section is intended to supplement the main report which (has been) recently
published and peer-reviewedl. While this project represents a new round of full-scale vehicle
simulation of advanced technologies, the scope has also been expanded in several ways to
broaden the range of vehicle classes and technologies considered, consistent with a longer-
term outlook through model years 2017-2025. The expanded scope also includes a new
analytical tool (complex systems analysis tool) to assist in interpolating the response surface
modeling (RSM) data and visualizing technology effectiveness.  This tool was especially
useful in isolating effectiveness trends during development of the updated Lumped Parameter
model.

       The agencies try to use publicly available information as the basis for technical
assessments whenever possible. Because this rulemaking extends to MY 2025, and includes
some technologies that are not currently in production and for which there is limited
information available in the literature, some of the technology inputs used to estimate
effectiveness are based on confidential business information. This includes the inputs related
to the technologies  listed below which were based on confidential business information
belonging to Ricardo, Inc, and  their expert judgment that contributed to projecting how these
technologies might improve in the future. The agencies have also considered information
which is in the public domain, in particular for turbo-charged, downsized GDI engines as
discussed in Section 3.4.1.8, as well as confidential information on engine and transmission
technologies from automotive suppliers which directionally was in line with the information
considered by Ricardo. The agencies encourage commenters to submit technical information,
preferably that may be released publicly, related to these technologies, particularly on their
effectiveness and ability to be implemented in a way that maintains utility. The agencies
welcome data and information on the technologies individually or in combinations.

          •   Advanced turbocharged and downsized, Atkinson, advanced diesel (e.g.
              projected BSFC maps)Hybrid powertrain control strategies
          •   Optimized transmission shift control strategies
          •   Transmission efficiency improvement.

       Below is a summary of the significant content changes from the 2008 simulation
project to the 2011  simulation project that supports the proposed rule:
                                            3-26

-------
                                      Technologies Considered in the Agencies' Analysis
3.3.1.2.1 More Vehicle Classes

       Two additional vehicle classes were considered, for a total of seven classes:  a small
car (subcompact) and a medium/heavy duty truck class. The inclusion of the small car class
increased the fidelity of the results by capturing engineering differences unique to the smallest
vehicles in the market. The inclusion of the medium/heavy duty truck was meant primarily to
support EPA's analysis for the Heavy Duty GHG Rule21. It is worth noting that these vehicle
classes are for simulation purposes only and are not be confused with regulatory classes or
NHTSA's technology subclasses.

3.3.1.2.2 More engine and vehicle technologies

       The original 2008 project modeled several engine and transmission technologies that
were expected to become commercially available within the 2012-2016 time frame. These
technologies included advanced valvetrain  technologies (such as variable valve timing and
lift, cylinder deactivation), turbocharged and downsized engines, as well as 6 speed automatic
transmissions, CVTs8 and dual-clutch transmissions. The current project built on top of this
effort with the inclusion of several new engine and vehicle technologies.  Highlighted
examples included:

       •   Advanced, highly downsized, high BMEPh turbocharged engines
       •   High efficiency transmissions with 8  speeds and  optimized shift strategies to
           maximize vehicle system efficiency
       •   Atkinson-cycle engines for hybrids
       •   Stop-start (or idle-off) technology

       A discussion of these technologies is included Section 3.3.3, and also in the 2011
vehicle simulation report 1.

3.3.1.2.3 Includes hybrid architectures

       For the first time, this new work includes modeling of hybrid architectures for all
vehicle classes.  Two main classes of hybrids were considered:

       •   Input powersplit hybrids. Examples of input powersplits in the market today
           include the Ford Fusion HEV and the Toyota Prius.
       •   P2 hybrids. An example of the P2 hybrid is the Hyundai Sonata Hybrid.
g Continuously variable transmissions
h BMEP refers to brake mean effective pressure, a common engineering metric which describes the specific
torque of an engine, as a way of comparing engines of different sizes. It is usually expressed in units of bar, or
kPa, Current naturally aspirated production engines typically average 10-12 bar BMEP, while modern
turbocharged engines are now exceeding 20 bar BMEP with regularity. Simply put, a 20 bar BMEP
turbocharged engine will provide twice the torque of an equivalent sized engine that achieves 10 bar BMEP.

                                             3-27

-------
                                     Technologies Considered in the Agencies' Analysis
       While input powersplit hybrids remain a very likely hybrid architecture choice for
some manufacturers, the agencies focused solely on P2 hybrids compared to powersplit
hybrids due to their apparent cost-effectiveness advantage in future years.

       Ricardo proprietary methodology was used to develop the control strategies were
developed for each architecture, the details of which can be found in section 6.8 of the 2011
project report1.
3.3.1.2.4 Complex systems tool for data analysis

       In the original 2008 project, EPA staff selected unique technology packages, based on
engineering judgment, to cover a representative subset of possible vehicle options ending in
MY 2016.  The expanded project time horizon (through MY 2025) and increased complexity
of potential vehicle technology interactions (including hybrids) made package selection much
more difficult.  To account for unforeseen results and trends which might exist, EPA and
Ricardo adopted a complex systems approach, which is a rigorous computational strategy
designed to mathematically account for multiple input variables and determine the
significance of each (the complex systems approach is described in further detail in the 2011
Ricardo report). As a comparison, in the 2008 study, twenty-six unique technology packages
spanning five vehicle classes were selected by EPA staff and then modeled. For this project a
set of core technology packages were chosen for each vehicle class, constituting a total of 107
unique vehicle packages ("nominal runs"), which are shown as Table 3-3 and Table 3-6 in
3.3.1.2.8. A neural network Complex Systems approach to design of experiments (DOE) was
then applied to generate a set of response surface models (RSM), in which several input
parameters were varied independently over a specified range to identify the complex
relationship between these inputs and the vehicle performance. Using these methods, the
vehicle simulation was run for a set of discrete input variables chosen based on a full factorial
analysis, using a computationally efficient algorithm to select each input variable within the
design space, allowing for subsequent statistical regression of the output variables.  This
approach resulted in an average of approximately two thousand independent simulation runs
for each of the 100+ vehicle packages, the outputs of which were interpolated in the data
analysis tool developed for this modeling activity.  For each of these nominal and DOE runs
Ricardo provided detailed  10-hz output data csv files for review1.

       An interactive Complex Systems analysis and visualization tool was developed to
interpret the vast arrays of RSM data generated as part of the project. It was created to sample
a selected portion of the design space populated using the DOE approach described above,
and then interpret the RSM data set in a form that could be used to calibrate the lumped
1 Stakeholders wishing to obtain this data may contact EPA to arrange for transfer of the data. Due to the
considerable size of the files (2 terabytes), stakeholders must supply their own storage media.

                                             3-28

-------
                                     Technologies Considered in the Agencies' Analysis
parameter model (reference the equivalent-performance results in Section 3.3.1.2.16). For
more detail on the use of the RSM tool, refer to the 2011 Ricardo report1.

3.3.1.2.5 Process

       The core technical work, completed in February 2011, consisted of the following
steps:
          Definition of project scope
          Selection of vehicle classes and baseline vehicle characteristics
          Selection of vehicle architectures and individual technologies
          Selection of swept variables for use in the RSM matrix
          Selection of vehicle performance metrics
          Review and revision of the input assumptions and modeling process
          Build and run the baseline EASY5 vehicle models
          Review of baseline runs and checking for errors
          Build and run the nominal technology package EASY5 vehicle models
          Review results and debug
          Run complete DOE matrix for each technology package
          Incorporation of DOE results into RSM tool
3.3.1.2.6 Definition of project scope

       At project initiation, an advisory committee was formed and led by EPA to help guide
the analysis. The advisory committee consisted of technical experts from CARB and The
ICCT, the latter of which co-founded the project.  A complete list of advisory committee
members is found in the vehicle simulation project report1. The committee agreed upon the
underlying ground rules, reviewed modeling assumptions and identified the desired vehicle
architectures and selected technologies for review. The boundaries for the project are
highlighted (quoted) below:

       •  A total of seven vehicle classes will be included:  small car, standard car, large car,
          small and large MPVs (multi-purpose vehicles), truck and HD truck

       •  LDV technologies must have the potential to be commercially deployed in the
          2020-2025  timeframe

       •  Vehicle sizes (footprint and interior space) for each class will be largely
          unchanged  from 2010 to 2020-2025

       •  Hybrid vehicles will use an advanced hybrid control strategy, focusing on battery
          state-of-charge management, but will not compromise vehicle drivability
                                            3-29

-------
                                     Technologies Considered in the Agencies' Analysis
       •  Vehicles will use fuels equivalent to 87 octane pump gasoline and 40 cetane pump
          diesel

       •  It is assumed that 2020-2025 vehicles will meet future California LEV III
          requirements for criteria pollutants, approximately equivalent to current SULEV II
          (or EPA Tier 2 Bin 2) emissions levels

       •  Changes in vehicle road loads including mass, aerodynamic drag, and rolling
          resistance, will not be accounted for in any of the modeled technologies. Instead,
          changes in vehicle road loads may be addressed through user-specified continuous
          input variables in the Complex Systems tool.

       The committee also decided that the following technologies fell outside the scope of
the project, either due to project resource limitations, lack of sufficient input data, or a low
potential to be commercially deployed in the timeframe considered:

       •  Charge-depleting powertrains (e.g. plug-in hybrids and electric range-extended
          vehicles) and electric vehicles

       •  Fuel cell-powered  vehicles

       •  Non-reciprocating  internal combustion engines or external combustion engines

       •  Manual transmissions and single-clutch automated manual transmissions (AMTs)

       •  Kinetic energy recovery systems other than battery systems

       •  Intelligent vehicle-to-vehicle (V2V) and vehicle-to-infrastructure optimization
          technology

       •  Bottoming cycles (such as organic Rankine cycles) for energy recovery

       •  Vehicle safety systems or structures will not be explicitly modeled for vehicles, as
          it is beyond the scope of the study

       The committee also selected  a set of swept input variables (vehicle parameters) which
were considered most important to vehicle fuel economy and performance (swept variables
are continuously variable input values that affect vehicle output efficiency in a smooth
function for the response surface  model).  These variables consisted of engine displacement,
final drive ratio, electric drive motor size (for hybrids), as well as road load factors (vehicle
mass, aerodynamic drag, and rolling resistance). All of these input variables were
randomized in each vehicle design of experiment matrix and then incorporated into the post-
processing RSM data visualization tool.
                                            3-30

-------
                                     Technologies Considered in the Agencies' Analysis
3.3.1.2.7 Selection of vehicle classes and baseline vehicle characteristics

       In order to estimate both technology costs and CO2 reduction estimates, it is necessary
to describe the baseline vehicle characteristics as the basis from which comparisons may be
drawn. In the 2012-2016 light-duty vehicle rule the vehicle baseline was defined as having a
naturally aspirated gasoline engine with a port-fuel injection system, two intake and two
exhaust valves and fixed valve timing and lift; the baseline transmission was a conventional
4-speed automatic, with no hybrid systems. These vehicles are referred to throughout this
section as the 2008 baselines. For the present study, EPA and Ricardo elected to include a set
of 2010 "baseline" technology vehicles, which reflect MY2010 trends in engine and vehicle
technology as well as some technologies that are expected to be widespread within a few
years. It is important to note that the 2010 baseline vehicles in the Ricardo study do not reflect
the technology content of the baseline fleet vehicles used by each agency in their respective
compliance modeling. The Ricardo 2010 baseline vehicles are only used in the analysis
required to establish effectiveness and synergies in the lumped parameter model. The 2010
baseline vehicles all include an engine with dual overhead camshaft and dual-independent
intake/exhaust valve timing, a six-speed automatic transmission, 12-volt idle off (stop-start)
functionality and an alternator with partial  energy regeneration capability.  There is no change
in the engine displacement or vehicle road  load coefficients between the 2008 baseline and
the 2010  baseline vehicles. For a table showing the 2010 baseline vehicle characteristics refer
to Appendix 3 of the 2011 Ricardo report1.

       In the Ricardo study, seven vehicle classes were selected for the analysis, in order to
more fully represent the broad groupings of a wide variety of products offered in the US
passenger car and light-duty truck market.  The seven vehicle categories chosen were as
follows:

   •   Small car: a subcompact car typically powered by a small 4 cylinder engine.
   •   Standard car:  a midsize car typically powered by a small 6 cylinder engine.
   •   Large car: a large passenger car typically powered by a large 6 cylinder engine.
   •   Small MPV:  a  small  multi-purpose vehicle (MPV) or "crossover" vehicle typically
       powered by a 4 cylinder engine
   •   Large MPV:  a minivan or large  MPV or "crossover" unibody constructed vehicle
       with a large frontal area, typically powered by a 6 cylinder engine,  capable of carrying
       ~ 6 or more passengers.
   •   Large truck (1/2 ton):  large sports-utility vehicles and large pickup trucks, typically a
       ladder-on-frame construction, and typically powered by an 8 cylinder engine.
   •   Class 2b/3 truck (3/4 ton): a large pickup truck (although with a GVW no greater  than
       8.500 pounds)  with a heavier  frame  intended  to provide  additional utility (a.k.a.
       "work" truck), typically powered by a larger 8 cylinder gasoline or diesel engine
                                            3-31

-------
                                      Technologies Considered in the Agencies' Analysis
3.3.1.2.8 Technology selection

       Ricardo presented the committee with an array of potential technologies that might
become commercially viable and present in the light-duty market by 2025. EPA and the
Advisory Committee suggested additional other technologies, e.g. Atkinson engines for
hybrids, fast engine warm-up strategies, etc, to consider in the selection process. The
complete set of potential technologies can be found in Appendix 2 of the 2011 Ricardo
report1.  After further deliberation within the committee and by Ricardo, a subset of
technologies considered most promising (from a technical feasibility and cost effectiveness
standpoint) was selected by the committee and Ricardo for inclusion in the project test matrix.
The technologies were distributed among four distinct vehicle architectures.  These
architectures represented unique EASY5 model structures, and are listed below:

       •  2010 Baseline vehicles: intended to represent physical replicas of existing vehicle
          models, although some minor additional content was included (as described in
          Section 3.3.1.2.7)

       •  Conventional stop-start: vehicles for the 2020-2025 timeframe that included
          advanced engines but did not incorporate an electric drive or braking energy
          recovery. These vehicles all contained a 12 volt stop-start (or idle-off) capability,
          along with the following technologies further detailed in the 2011 Ricardo
          simulation studyj:

                     o  higher efficiency gearbox (2020 timeframe)
                     o  optimized  shift strategy (best BSFC)
                     o  alternator regeneration (during braking)
                     o  high-efficiency alternator
                     o  advanced engine warmup technologies
                     o  engine friction reduction (+3.5% fuel consumption reduction over
                        2008 baseline)
       •  P2 hybrid: represent a class of hybrids in which the electric drive motor is coupled
          via a clutch directly to the transmission input shaft.  An existing vehicle in the
          market which most closely represents this architecture is the 2011 Hyundai Sonata
          Hybrid except that Ricardo recommended a P2 hybrid with a more efficient and
          cost effective dual clutch  transmission in lieu of a planetary gear transmission.
          Additional examples of a P2 hybrid approach are the 2011 Volkswagen Touareg
          Hybrid, the 2011 Porsche S Hybrid, and the 2012 Infiniti M35 Hybrid.  Each of
          these are examples of "first generation" P2 systems, as compared to for example
          the powersplit hybrid systems offered by Ford, Toyota and  or the IMA systems
          from Honda which are in  their second, third or even fourth  generation.  The
j The technologies included in all of the conventional stop-start packages were expected to be widespread by
years 2017-2025.  Some "anytime technologies" such as aerodynamic drag and rolling resistance reduction were
excluded from the nominal runs, but were incorporated in the complex systems portion of this project.

                                             3-32

-------
                                     Technologies Considered in the Agencies' Analysis
          agencies are aware of some articles in trade journals, newspapers and other
          reviews that some first generation P2 hybrid vehicles with planetary gear
          transmissions have trade-offs in NVH and drivability - though these reviews do
          not cover all of the P2 systems available today, [and a number of reviews are very
          positive with respect to NVH and drivability].  For this analysis we are projecting
          that these issues with some first generation P2 systems can be addressed with no
          hardware cost increase or reduction in efficiency for future generations of P2
          systems developed for the 2017-2025 time frame.  The agencies seek comment on
          our assumptions in this regard, and we request comment on the applicability of
          DCTs to P2 hybrid applications, including any challenges associated with NVH or
          drivability. Key technology assumptions included:
                     o   Lithium-ion battery
                     o   DCT transmission
                     o   Electric drive motor which provides, when combined with a less
                        powerful engine, equivalent 0-60 performance to the baseline
                        vehicle.
                     o   Engine displacement for the P2 hybrids were assumed to be 20%
                        less than their conventional stop-start equivalents

       •  Input powersplit hybrid: represent a class of hybrids with both an electric drive
          motor and a separate generator linked to a planetary gearset which effectively
          controls the overall gear ratio and distribution of tractive and electrical power.
          Example vehicles in  the market include the Toyota Prius and the Ford Fusion
          hybrid.  Key technology assumptions are consistent with those for the P2 hybrid,
          with the exception of the power split device, which functions as a CVT-type
          transmission (as is the case in real world examples), and replaces the DCT
          transmission in the P2 design.  As stated previously while this technology was
          simulated it was not used in this NPRM analysis.

       Some architectures that seemed less appropriate for certain vehicle classes were
omitted.  For example, in the Ricardo modeling of the medium/heavy duty truck (a Class 3
vehicle with a GVWR >10,000 pounds, and thus not subject to the proposed standards in this
rulemaking), no P2 or input powersplit hybrids were included. Other technologies that did
not seem reasonable for some vehicle classes (such as dry-clutch DCTs for Large MPVs and
Trucks) were also excluded in the Ricardo simulations.

       In summary, 4 distinct vehicle architectures (including the baselines as an
"architecture"), across 7 vehicle classes, and a number of engine and transmission
combinations, represented the complete set of vehicle combinations.  The test matrices'^ can
k For each vehicle class, each advanced engine option is combined with each advanced transmission. Baseline
runs are not combined with other transmissions.

                                            3-33

-------
                                      Technologies Considered in the Agencies' Analysis
be found below in Table 3-5 (for 2010 baselines and conventional stop-start vehicles) and
Table 3-6 (for hybrids).
                      Table 3-5: Nominal Package Matrix for Non-Hybrids






Vehicle Class
Small Car
Standard Car
Small MPV
Full Size Car
Large MPV
LOT
LHDT
08
(D
O) TJ
mSM

1 <° E
8? 2
re o 3
CD CM <
X
X
X
X
X
X
X
o .2
1— *-
o w
J
X


X
X
X
X
Advanced Transmission


_O
*o re

Q. 0
to <
X







>,
Q
TJ
SI
Q- 1-
W O
to O
X








_o
*o re
0) c
fli C
Q. 0
CO <

X
X
X
X
X
X

>,
Q
TJ

0. \-
00 0

X
X
X




0)

TJ
(D
fl)
Q- 1-
06 0




X
X
X
              Table 3-6: Nominal package matrix for P2 and Input Powersplit hybrids








Vehicle Class
Small Car
Standard Car
Small MPV
Full Size Car
Large MPV
LOT
LHDT
Hybrid Architecture


i
2 H
.Q U
X 0

CL S
X
X
X
X
X
X


Q.
fl)

Q.
D
Q.
C
X
X
X
X
X


Advanced Engine
o
.a
3

•c O
•— .E
o ~
53 5
X
X
X
X
X
X


o
.a
i- 
<2 Q
1 -=

< i
X
X
X
X
X
X

                                             3-34

-------
                                     Technologies Considered in the Agencies' Analysis
3.3.1.2.9 Selection of the swept input variables and their ranges

       The advisory committee agreed upon a set of continuous input variables to be swept in
each vehicle package response surface. These variables consisted of both powertrain
characteristics (engine displacement, final drive ratio, and electric machine size for hybrids)
and road load parameters (rolling resistance coefficient, aerodynamic drag force, and vehicle
mass). They were included in the DOE matrix for each vehicle architecture and powertrain
configuration, and also serve as inputs to the complex systems visualization tool. Table 3-7
and Table 3-8  show the swept variables used (and their ranges) for the conventional stop-start
and hybrid packages, respectively.  The ranges represent a percentage of the default value
used in the nominal runs.
       Table 3-7:  Continuous input parameter sweep ranges for conventional stop-start vehicle
Parameter
Engine Displacement
Final Drive Ratio
Rolling Resistance
Aerodynamic Drag
Mass
DoE Range (%)
50 125
75 125
70 100
70 100
60 120
      Table 3-8: Continuous input parameter sweep ranges for P2 and Powersplit hybrid vehicles
Parameter
Engine Displacement
Final Drive Ratio
Rolling Resistance
Aerodynamic Drag
Mass
Electric Machine Size
DoE Ra
P2 Hybrid
50 150
75 125
70 100
70 100
60 120
50 300
nge (%)
Powersplit
50 125
75 125
70 100
70 100
60 120
50 150
       The ranges were intended to include both the (unknown) optimal value for each
technology case, but also wide enough to capture the range of values as they depart from the
optimal value (in engineering parlance this is often referred to as finding the "knee" in the
curve).
                                             3-35

-------
                                     Technologies Considered in the Agencies' Analysis
       From these variables, a user can determine the sensitivity of each input variable to the
vehicle fuel economy and performance. For example, the effect of engine displacement on
fuel economy was evaluated for several packages. A more elaborate discussion of engine
displacement effects is provided in Section 3.3.1.2.22.2.

3.3.1.2.10   Selection of vehicle performance metrics

       For both effectiveness and cost estimates in these rulemakings, the agencies are
assuming that vehicles will maintain utility (performance) comparable to the models in the
baseline fleet1. It was therefore important to maintain equivalent performance in the vehicle
simulation modeling of future vehicle technology. The resulting effectiveness estimates were
in the context of equivalent performance, which carried over into the lumped parameter model
and into the OMEGA and Volpe packages.

       Consistent with the 2008 simulation project, a set of vehicle (acceleration)
performance metrics were selected by the advisory committee as a way of measuring
"equivalent" vehicle performance. When quantifying vehicle efficiency, it is important that
certain other vehicle performance metrics are maintained, such that there are no other
competing factors contributing or detracting from the vehicle efficiency. Other vehicle
characteristics that could impact or detract from vehicle efficiency (e.g., noise, vibration and
harshness (NVH), drivability, durability, etc) were also considered during the generation of
model inputs.  However, they were not analyzed explicitly, with the expectation that
manufacturers would ultimately be able to meet vehicle refinement levels necessary for
commercial acceptability of these new technologies. These metrics, shown below in Table
3-9, include time at full load to reach given speeds (0-10 mph, 0-30 mph, etc), maximum
grade capability, and distance traveled at a given time (e.g., after 3 seconds).  Ultimately, the
measure of equivalent performance is up to the reader or user of the Complex Systems tool.
For EPA's analysis baseline vehicle 0-30 mph and 0-60 mph acceleration times were used as
a benchmark for equivalent performance for the advanced vehicle packages. These estimated
acceleration times are included in Table 3-11 through Table 3-18. Detailed results that
include all performance metrics including those for baseline vehicles are provided in the full
2011 simulation report1.
1 The only exception to this is a subset of hybrids explicitly listed as "non-towing" vehicles. For further details
and background, reference Section 1.3 of EPA's RIA.

                                             3-36

-------
                                     Technologies Considered in the Agencies' Analysis
              Table 3-9: Vehicle performance metrics produced by the EASY5 model
Launch (WOT)

0-10mph

0-30mph

0-50mph

0-60mph
0-70mph
Distance @ 1.3 sec
Distance @ 3 sec
Speed @ 1.3 sec
Speed @ 3 sec
Passing (WOT)

30-50 mph

50-70mph









Gradeability/
torque reserve
Max Speed @ 5%
grade
Max Speed @ 10%
grade
Max Grade @ 70
mph (non-towing)
Max Grade @ 60
mph (towing)





3.3.1.2.11   Review and revision of inputs

       For any system modeling in which the results extend beyond the bounds of known
physical examples (and therefore direct data validation is impossible), it is imperative that the
inputs be carefully constructed and thoroughly examined to minimize the potential for
uncertainty-related errors. Prior to coding of the models, Ricardo presented the following
inputs for review and approval to EPA. For each topic, EPA reviewed the material
considering the rationale of Ricardo's technical experts, the appropriateness of the inputs in
relation to the assumed time horizon, the required emissions levels, and the known literature
in the field today. Listed below are several of the model inputs that were jointly reviewed by
Ricardo and EPA:

       •  Engine maps

                    o  Stoichiometric GDI turbo
                    o  Lean-burn GDI turbo
                    o  Cooled EGR turbo
                    o  Advanced diesel maps
       •  Transmission efficiency tables (by gear) including torque  converter efficiency
       •  Engine warm-up strategy (cold start modifiers)
       •  Alternator regeneration strategy
       •  Transmission shift optimizer
                                            3-37

-------
                                     Technologies Considered in the Agencies' Analysis
       •  Engine friction reduction level
       •  P2 hybrid controls
       •  Input powersplit hybrid controls
       •  Hybrid battery assumptions
       •  Hybrid motor/generator efficiency maps

       EPA technical experts recommended several changes and iterated with Ricardo to
establish a consensus set of inputs that were plausible and met the ground rules of the project.
Some of these changes resulted in higher efficiencies, while others lowered efficiency.
Highlighted below are a few key examples, starting with development of the engine maps:

       Engine maps carry perhaps the most significance of any of the sets of inputs needed to
build vehicle simulation models. They provide the brake specific fuel consumption, or BSFC
(typically in g/kWh) for a given engine speed and load.  Typically these maps show an
optimum speed and load band (or minimum BSFC "island") that is the most efficient
condition in which to operate the engine. Ricardo generated engine maps for both the
baseline vehicles (through benchmarking data) and proposed future engine maps for the
various turbocharged and diesel engines. Figure 3-3 shows an example engine map for a
baseline vehicle. It was constructed from EPA's analysis of a baseline vehicle model run
output file. The contours represent lines of equivalent brake-specific fuel consumption."1

3.3.1.2.11.1  Engine Technologies

       Ricardo developed the engines for the 2012-2025 timeframe in two ways. The first
was to take current boosted SI research engines and project these would represent the level of
performance which could be achieved by production engines in the 2020-2025 timeframe.
The second method took current production Atkinson cycle SI and diesel engines and then
included 2020-2025 timeframe technology improvements. Both methods extrapolated current
engine design and development trend to the 2020-2025 timeframe. These current trends
include engine friction reduction, improved fuel injection systems (e.g., spray guided for the
SI, and higher injection pressures for the diesels), more advanced engine controls, and
improved engine design for faster  engine warm-up.  EPA reviewed the engine maps
recommended by Ricardo and generally concurred they were appropriate for the study time
frame based on EPA's review of maps for current production engines and for research engines
described in the literature.
m BSFC is measured in units of grams of fuel per kw-hour of energy and is an indicator of engine efficiency.
Lower numbers indicate more efficient operating regions. As in this case, an engine typically has an "island' or
region of best efficiency, in this case between 2000-3000 RPM and 150-180 Nm of torque. This island becomes
much larger with the advent of advanced technologies such as boosting and downsizing, as well as advanced
valve train technologies.

                                             3-38

-------
                                     Technologies Considered in the Agencies' Analysis
              220
              200
              1BO
               BO
               4O
               2O

                             2SS
                          330
                     1OOO
                               2OOO
                                          3OOO       4OOO
                                            Speed (rprn)
                                                               5OOO
                                                                          6OOO
                       Figure 3-3: Example baseline engine BSFC map
3.3.1.2.11.2  Stoichiometric GDI

       The original Stoichiometric GDI map that Ricardo proposed was based on laboratory
data they had published in 2007, showing a peak brake-specific load of just under 20 bar
BMEP and a minimum BSFC of approximately 235 g/kWhr, obtained using a compression
ratio of 10.5:1.22 However, based on input from manufacturers and from other, more recent
published data on developmental and research engines, EPA asked Ricardo to raise the load
capability of the engine to approximately 27 bar BMEP.23'24'25'26 This allowed a greater degree
of engine downsizing, which resulted in a downsizing of a 1.5 liter engine to  a 0.74 liter
engine for the nominal small car and a 5.4 liter to a 1.94 liter engine for the nominal large
truck. A compression ratio of 10.5:1 was maintained for improved efficiency. At the same
time, EPA asked that Ricardo eliminate the use of high-load enrichment, since water-cooled
exhaust manifolds, in some cases integrated into the cylinder head, can be incorporated in
next-generation designs to mitigate the need for fuel enrichment in lowering turbine inlet
temperatures to 950 degrees C and thus avoid the added costs of high-temperature materials in
the turbocharger.27'28  By reducing  the need for fuel enrichment fuel consumption is reduced
                                            3-39

-------
                                     Technologies Considered in the Agencies' Analysis
over the more aggressive portions of the drive cycle, and PM emissions control at high load is
improved.
3.3.1.2.11.3  Lean-burn GDI

       Ricardo's initial lean-burn GDI map was based on their single-cylinder research
engine data,  in which they operated in lean stratified charge mode at all speeds and loads,
without due  consideration of the potential limitations in lean exhaust NOx aftertreatment
systems.  To address concerns in this area, EPA examined the boundaries of operation of lean-
NOx catalysts, assuming that manufacturers would adopt either LNTs or metal-zeolite urea
SCR systems.  EPA therefore asked Ricardo to place a constraint on the maximum allowable
catalyst space velocity (at high engine power) and exhaust gas temperature entering the
catalyst (at high load, low engine speed conditions) to maintain catalyst efficiency at high
load and to reduce thermal sintering of PGM under high-temperature, lean operating
conditions. More specifically, EPA recommended that engine operation switch away from
lean operation (at air/fuel equivalence ratios up to approximately X=1.5) to stoichiometric
operation at  turbine outlet temperatures above 600C, and at total exhaust flows corresponding
to space velocities of 60,000/hour, assuming a catalyst volume of 2.5 times engine
displacement.  This marginally diminished the engine brake thermal efficiency to
stoichiometric GDI levels over this region of the map, but it provided more certainty that the
engine would be able to adhere to the emissions levels as assumed in the project ground rules
by the Advisory Committee.  Figure 3-4 shows the engine speed and load region EPA
proposed as  suitable for lean stratified operation.
                                            3-40

-------
                                     Technologies Considered in the Agencies' Analysis
          -700-
         -40000-
Temperature
NOx Catalyst Space Velocity
Transition from lean to A=l operation
       0       1000     2000     3000     4000     5000     6000     7000
                               Engine Speed [rpm]

       Figure 3-4  Proposed lean/stoichiometric operating threshold for lean-burn GDI engines
3.3.1.2.11.4 Cooled EGR GDI

       EPA provided technical information from the literature which enabled Ricardo to
assume a dual loop (both low pressure and high pressure EGR loops), cooled EGR system in
addition to the stoichiometric turbocharged engine. The development of engine maps for this
engine configuration was heavily informed by recently published data.26'27' 8'29. Cooled EGR
allowed the use of "X=l" operation at the same compression ratio with more aggressive spark
timing at high load and reduced pumping losses at part load while maintaining acceptable
turbocharger inlet temperatures.
                                            3-41

-------
                                    Technologies Considered in the Agencies' Analysis
3.3.1.2.11.5  Motor/generator and power inverter efficiency maps

       EPA recommended that Ricardo update the efficiency maps of the motor and
generator (referred to as "electric machines" throughout the project), which they had proposed
based on current best-in-class technology. The baseline motor/generator+inverter efficiency
map is taken from a 2007 Camry and shown in Figure 3-5 below.
          300
          250-
          200-
         .150-
          100-
              500   1000   1500   2000   2500   3000   3500   4000   4500  5000  5500  6000
                                         Speed (RPM)

        Figure 3-5: 2007 Camry Hybrid motor-inverter efficiency map (Burress, et al, 200830)

       EPA requested that Ricardo provide their assessment of where they believed
efficiency improvements might be made, based upon trends in research and development for
both electric machines and power electronics.  Ricardo and EPA generally agreed that these
efficiency improvements were likely to be modest, particularly given the competitive
pressures on manufacturers to reduce the cost of hybrid components.  However, EPA and
Ricardo assumed that today's best-in-class efficiency would likely be marginally improved
through continuous incremental reductions in parasitic losses. To account for this, EPA and
Ricardo agreed to reduce the losses in the motor/generator by 10% (in other words, raising the
efficiency of a 90% efficient motor to 91%) and to reduce the losses in the power electronics
by 25% (mainly through continued improvements in inverter development and electronic
control systems).

3.3.1.2.11.6  Battery

       Battery packs were assumed to consist of spinel LiMnO2 cathode chemistry, which is
consistent with the current state of technology. EPA recommended a maximum usable state of
charge of 40% (from 30% charge to 70% charge) be incorporated as an operating window in
92


90


88


86


84


82


80


78
                                            3-42

-------
                                    Technologies Considered in the Agencies' Analysis
Ricardo's hybrid control logic.  This range may increase in subsequent real world examples as
manufacturers gain more field experience with long term battery durability.  Additionally
there will likely be more advances in battery construction and chemistry by 2025, so EPA
considers these assumptions as conservative in view of the long term research currently
underway in many battery research companies.

3.3.1.2.12   Additional Technologies Modeled by Ricardo for 2011 Report

       The previous section discusses in detail those areas of the Ricardo simulation inputs
which EPA provided recommendations to Ricardo on and which Ricardo agreed and made
modifications to their initial suggestions.  EPA did review modeling inputs for many other
technologies modeled by Ricardo, but for which we generally agreed with the reasonableness
of Ricardo's approach and did not request any changes.  This section summarizes at a high
level some of the additional technologies considered by Ricardo. Additional detail on these
technologies is contained in the 2011 Ricardo final report.

       Diesel engines - Ricardo started with existing production engines and identified
technology advances that would lead to further advances in fuel consumption. These included
many of the same technologies considered for advanced gasoline engines, such as engine
friction reduction, improved fuel injection systems with higher injection pressures and more
advanced controls, and better engine design to improve engine warm-up rate.

       Transmission Technologies - Taking a systems approach in the vehicle simulation
modeling, Ricardo also introduced additional transmission and driveline oriented technologies
that may be pathways to increased efficiency. Some of these key technological enablers
include: shift optimization schedules, advanced clutches, torque converter design and lockup
schedules.

       Automatic and Dual Clutch Transmissions - For the study timeframe, Ricardo
assumed that eight-speed automatic transmissions will be in common use, as this supports
more efficient operation, except for small cars, with energy losses expected to be about 20-
33% lower than in current automatic transmissions. Energy losses in both wet clutch and dry
clutch DCTs are expected to be 40-50% lower than in current automatic transmissions.

       Transmission Shift Optimization - This advanced transmission shift optimization
strategy tries to keep the engine operating near its most efficient point for a given power
demand in effort to emulate a CVT. To protect against operating conditions out  of normal
range, several key parameters were identified, such as maximum engine speed, minimum
lugging speed, and minimum delay between shifts. During development of this strategy,
Ricardo estimated that fuel economy benefits of up to 5% can be obtained when compared to
typical MY 2010 shift maps.

       Torque Converter Technology - Ricardo utilized a lockup clutch model with a multi-
damper system to provide earlier torque converter clutch engagement. The advanced
                                            3-43

-------
                                     Technologies Considered in the Agencies' Analysis
automatic transmission applications allow torque converter lockup in any gear except first
gear, up to sixth for the Small Car or eighth for the other LDV classes.

        Shifting Clutch Technology - Shift clutch technology improves the thermal capacity
of the shifting clutch to reduce plate count and lower clutch losses during shifting. Reducing
the number of plates for the shifting process and reducing the hydraulic cooling requirements
will increase the overall transmission efficiency for similar drivability characteristics.

       Dry Sump Technology - A dry sump lubrication system provides benefits by keeping
the rotating members out of oil, which reduces losses due to windage and churning. This
approach will provide a GHG emissions benefit across all vehicle classes, with the best
benefits at higher speed.
3.3.1.2.13   Baseline models built and run

       Once all of the inputs were established, Ricardo built the baseline models:  For these
new (2010) baseline models Ricardo added a group of minor technologies, most of which
already exist today in the market. The technologies included 12V stop-start, 6-speed
automatic transmission, a high efficiency (70% efficient) alternator, and a strategy -
"alternator regen" - that charges the 12V battery more aggressively by increasing the
alternator field upon vehicle deceleration .

       In the 2008 study Ricardo validated their baseline models with 2008 MY certification
data. Ricardo's 2010 baseline model results provided effectiveness data for EPA to calibrate
the lumped parameter model for some  of the newly applied technologies.  These technologies
included alternator regeneration, high efficiency alternator, and stop-start.

       For all model runs - the baselines and each of the advanced package nominal runs -
EPA reviewed an extensive set of detailed intermediate output data for each model run.  The
parameters that were reviewed are shown in Table 3-10.
                                            3-44

-------
                                           Technologies Considered in the Agencies' Analysis
                         Table 3-10: Vehicle simulation output data reviewed
          Ricardo outputs
          vehicle speed
          throttle position
          engine torque
          engine power
          transmission input shaft torque
          wheel torque
          transmission gear
          torque converter slip ratio
          current engine BSFC
          accessory power
          engine speed
          road load
          N/V
          electric power of motor generator
          mechanical power of motor generator
          motor generator speed
          motor generator torque
          motor generator current
          motor generator voltage
          power flow through battery
          battery state of charge
          battery voltage
          regenerative braking power
          vehicle foundation braking power
          driver braking force
          fuel mass flow rate
          transmission mechanical loss power
          idle off status
EPA-calculated outputs
engine operating point distribution
engine load (BMEP)
total accessory energy
round-trip battery loop losses
torque converter lockup time
total road load
total engine brake thermal energy
EPA-calculated metrics
cycle-average BSFC
average brake thermal efficiency
average engine power
average engine speed
average engine torque
#of idle-off events
% of engine time off
average accessory power
time in each gear
average gear efficiency
average torque converter efficiency
battery state-of-charge statistics
battery efficiency
% of vehicle braking energy recovered
average motor efficiency
average generator efficiency
average motor and generator operating speeds
average motor and generator operating torque
total vehicle tractive energy	
        From this data, a set of summary statistics was generated to compare each baseline and
nominal package ran as a quality check.  This information was used as the starting point in the
dialogue between EPA and Ricardo to identify technical issues with the models. An example
summary table (or "snapshot") for the 2010 Standard Car baseline is provided in Figure 3-6.
                                                    3-45

-------
                                     Technologies Considered in the Agencies' Analysis
Vehicle
CO2 Emissions (g/mi)
Fuel Economy (mpg)
2007 Base Vehicle CO2 (g/mi)
% CO2 Reduction
Engine
Avg Brake Thermal Efficiency
Cycle Avg BSFC (g/kWh)
A\g Engine Power (HP)
A\g Engine Speed (RPMJ
A\gLoad(BMEP-barJ
A\g Torque (Nm)
Total Fuel (g)
Idle Off Events
% Time Off
Accessory Loss
Avg accessory power (W)
Avg BSFC temp mult (20 F)
Avg BSFC temp mult (75 F)
Transmission
Time in gear 1
Time in gear 2
Time in gear 3
Time in gear 4
Time in gear 5
Time in gear 6
Time in gear 7
Time in gear 8
Avg. n (gear)
Avg. n (TC)
Avg. n (driveline)
Battery
SOCAvg
Std Deviation
Max SOC
Min SOC
Max SOC Swing
Battery Efficiency (%J
Average Voltage (V)
Std Dev Voltage (V)
Battery Energy Change (kWh)
% of braking energy recovered
%batt charge via brake recov
%batt charge via engine
MG1
Test-Avg Motor Power (hp)
Avg Motor Eff
Avg Generator Eff
Avg Torque-Motor (N-m)
Avg Torque-Generator (N-m)
Avg RPM-Motor
Avg RPM-Generator
Mech Energy -Motor (kWh)
Mech Energy-Gen (kWh)
MG2
Avg Motor Power (hp)
Avg Motor Eff
Avg Generator Eff
Avg Torque-Motor (N-m)
Avg Torque-Generator (N-m)
Avg RPM-Motor
Avg RPM-Generator
Mech Energy-Motor (kWh)
Mech Energy-Gen (kWh)
Round-trip MG efficiency
Buck/Boost Converter
Avg Discharge Eff
Avg Charging Eff
Avg Bus Voltage (V)
LHV (fuel)
SG (Lei)
Specific CO2
Vehicle Energy Audit (kWh)
Total fuel energy
Total indicated energy
Engine pumping energy
Engine friction energy
Engine braking energy
Total accessory energy
Net brake thermal energy
Torque converter losses
Transmission losses
Battery loop losses
PE losses
Losses to MG devices
Total driveline losses
Vehicle tractive energy
Total road load energy
Foundation braking energy
Alternator regen decel energy
Total reqd. braking energy
FTP
303.8
29.9
337.8
10.1%
FTP
21 .7%
376
7.0
1993
2.21
42.1
1026.4
20
18.0%
0.0%
8.2
1.32
1.20
FTP
30%
9%
16%
27%
9%
9%
0%
0%
87.4%
88.9%
77.7%
FTP
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.0%
#DIV/0!
#DIV/0!
FTP
n/a
n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.00
FTP

n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.00
#DIV/0!
FTP
n/a
n/a
n/a
44
0.739
9087
FTP
12.54
4.48
0.69
0.86
0.20
0.00
2.73
0.30
0.31
0.00
0.00
0.00
0.61
2.12
1.29
0.50
0.32
0.82
Hwy
209.0
43.5
217.5
3.9%
Hwy
27.8%
295
14.1
1833
3.27
62.5
657.8
1
0.5%
0.5%
198.0
n/a
n/a
Hwy
2%
1%
2%
6%
35%
54%
0%
0%
88.0%
97.8%
86%
Hwy
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.0%
#DIV/0!
#DIV/0!
Hwy
n/a
n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.00
Hwy

n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.00
#DIV/0!
Hwy
n/a
n/a
n/a
kJ/g

g/gai
Hwy
8.04
3.38
0.57
0.48
0.03
0.04
2.23
0.05
0.26
0.00
0.00
0.00
0.31
1.92
1.76
0.11
0.06
0.16
Combined
261.2
34.8
283.7
7.9%
Combined
23.8%
344
10.2
1921
2.69
51.3
860.5
n/a
10.1%
0.3%
93.6
n/a
n/a
Combined
17%
5%
10%
18%
21%
29%
0%
0%
87.7%
92.9%
81.5%
Combined
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.0%
#DIV/0!
#DIV/0!
Combined
n/a
n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.00
Combined

n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.00
#DIV/0!
Combined
n/a
n/a
n/a



Combined
10.52
3.98
0.63
0.69
0.12
0.02
2.50
0.19
0.29
0.00
0.00
0.00
0.47
2.03
1.50
0.32
0.20
0.53
US06
312.2
29.1


US06
30.6%
267
23.0
2453
5.19
99.1
764.8
5
6.5%
0.0%
12.4
n/a
n/a
US06
13%
5%
7%
8%
10%
57%
0%
0%
87.9%
95.4%
83.8%
US06
n/a
n/a
n/a
n/a
n/a
n/a
n/a
' n/a
0.00
0.0%
#DIV/0!
#DIV/0!
US06
n/a
n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.00
US06

n/a
n/a
n/a
n/a
n/a
n/a
0.00
0.00
" #DIV/0!

n/a
n/a
n/a



US06
9.35
4.22
0.76
0.52
0.07
0.00
2.86
0.13
0.33
0.00
0.00
0.00
0.46
2.40
1.75
' 0.49
0.12
0.62
                                                                     Powertrain Architecture
Engine
Disp
L
Engine
Torque
Nm
Trans
Type

#
of
gears
MG1
size
kW
MG2
size
kW
Battery
size
kWh
                                                               220  base auto
                                                                      Performance Metrics
                                                      | 0-10mph| 0-30mph | 0-60mph [base 0-60|30-50mph|50-70mph|dist @ 3s|
                                                         1.0     3.1     8.3    8.3     3.2     5.1     20.5
                                              for using Ricardo maps
                                                                  Shift Optimizer Evaluation Tables
Gear

1
2
3
4
5
6
7
8

Gear

1





7
8
Avg BMEP (bar)
FTP
1.7
3.0
2.4
1.6
2.7
2.3
#DIV/0!
#DIV/0!
Hwy
2.3
3.9
4.5
3.1
3.7
2.8
#DIV/0!
#DIV/0!
US06
4.2
7.1
6.5
6.7
6.7
4.0
#DIV/0!
#DIV/0!

Avg BSFC (g/kWh)
FTP
338
328
359
482
361
388
0
0
Hwy
330
282
268
298
279
311
0
0
US06
256
255
264
265
251
279
0
0
Avg RPM
HP
1421
2309
2088
2160
2028
1827
0
0
Hwy
1710
2463
2395
1978
1869
1737
0
0
US06
2155
2881
2974
3209
2561
2137
0
0

Total Energy (%)
IP
6%
5%
1%
4%
2%
1%
0%
0%
Hwy
1%
1%
3%
7%
42%
46%
0%
0%
US06
8%
9%
10%
10%
16%
49%
0%
0%
                                              MG1=sun on planetary
                                              Recovered energy returned to wheels
                                              Gross recovered braking energy
                                              MG2=carrier (tractive)
                                              From alt regen braking (extra alternator load) x °,
Figure 3-6  Sample output summary sheet for Standard Car (Camry) baseline
                                               3-46

-------
                                     Technologies Considered in the Agencies' Analysis
       Summary statistics were used as a first-order quality check on the model. Sample
checks included:

       •  were average engine speed and load within or close to the best BSFC region for
          the vehicle's engine map?
       •  was transmission gear distribution reasonable and consistent between engine
          types?

3.3.1.2.14   Nominal runs

       The Ricardo "nominal" runs refer to the initial set of vehicle simulation models built
for each vehicle architecture and vehicle class. These runs were used by EPA to assess the
validity of the detailed model outputs (and hence the models themselves) prior to proceeding
with the full design of experiment runs. Table 3-11 shows the summary results from the raw
nominal runs for the conventional stop-start vehicles (including 12V stop-start, 70% efficient
alternator, shift optimizer and alternator regen, as well as a 3.5% improvement due to engine
friction reduction). Conventional automatic transmissions are assumed in all nominal runs.
No road load reductions are included in these results. GHG reductions are in reference to the
2008 baseline vehicles.
                 Table 3-11: Nominal Conventional Stop-Start modeling results
Vehicle Engine
Class Type
Small Car
Std Car
La rge Ca r
Small MPV
LargeMPV
Truck
HDTruck
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
2020 Diesel
Displ.
L
0.74
0.74
0.74
1.23
1.04
1.04
1.04
1.41
1.41
1.41
2.85
1.13
1.13
1.13
1.31
1.31
1.31
2.61
1.94
1.94
1.94
4.28
2.3
2.3
2.3
6.6
Torque
Nm
157
157
157
221
220
220
220
298
298
298
503
239
239
239
277
277
277
460
410
410
410
694
486
486
486
895
Trans
Type
AT6
AT6
AT6
AT6
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
FTP
mpg
53.2
55.1
55.1
55.8
44.8
46.6
46.4
37.1
38.8
38.6
38.2
38.8
40.3
40.3
34.8
36.0
36.2
37.3
23.8
24.6
24.8
26.4
16.5
16.8
17.2
19.8
HW
mpg
55.1
56.0
57.4
59.4
54.5
55.5
56.7
43.2
44.0
44.9
46.5
42.6
43.1
44.4
39.2
39.8
40.9
43.3
26.6
27.0
27.7
30.4
18.3
18.4
19.1
21.5
Comb
mpg
54.0
55.5
56.1
57.4
48.7
50.2
50.5
39.6
41.0
41.2
41.5
40.4
41.5
42.0
36.7
37.6
38.2
39.8
25.0
25.6
26.0
28.1
17.3
17.5
18.0
20.5
0-30mph
s
4.0
4.0
4.0
3.7
3.1
3.1
3.1
3.0
3.0
3.0
2.9
3.3
3.3
3.3
3.2
3.2
3.2
3.0
3.0
3.0
3.0
2.9
3.2
3.2
3.2
2.9
0-60mph
s
10.0
10.0
10.0
9.8
8.5
8.5
8.5
7.4
7.4
7.4
7.5
8.9
8.9
8.9
8.6
8.6
8.6
8.6
8.1
8.1
8.1
8.0
9.8
9.8
9.8
8.8
%GHG
Reduction
20%
22%
23%
16%
28%
31%
31%
31%
33%
33%
27%
25%
27%
28%
31%
33%
34%
30%
26%
28%
29%
26%
27%
28%
30%
31%
                                            3-47

-------
                                       Technologies Considered in the Agencies' Analysis
       Table 3-12 shows the results from the nominal runs for the P2 hybrid vehicles.  Dual-
clutch transmissions are assumed in all nominal runs.  No road load reductions are included in
these results. GHG reductions are in reference to the 2008 baseline vehicles.
                        Table 3-12:  Nominal P2 Hybrid modeling results
Vehicle Engine
Class Type
Small Car
Std Car
La rge Ca r
Small MPV
LargeMPV
Truck
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
Displ.
L
0.59
0.59
0.59
1.66
1.66
0.83
0.83
0.83
2.4
2.4
1.13
1.13
1.13
3.8
3.8
0.9
0.9
0.9
2.6
2.6
1.05
1.05
1.05
3.15
3.15
1.55
1.55
1.55
4.6
4.6
Torque
Nm
124
124
124
138
138
176
176
176
200
200
238
238
238
317
317
190
190
190
217
217
221
221
221
263
263
327
327
327
384
384
EM size
kW
14
14
14
14
14
24
24
24
24
24
28
28
28
28
28
20
20
20
20
20
25
25
25
25
25
50
50
50
50
50
Battsize
kWh
0.70
0.70
0.70
0.70
0.70
1.00
1.00
1.00
1.00
1.00
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.15
1.15
1.15
1.15
1.15
1.50
1.50
1.50
1.50
1.50
Trans
Type
DCT6
DCT6
DCT6
DCT6
DCT6
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
FTP
mpg
68.2
68.4
70.2
70.8
71.7
61.9
62.9
65.1
64.6
65.9
49.8
50.4
51.7
49.9
51.1
50.1
50.8
52.0
52.9
54.1
47.7
47.4
47.6
48.3
48.8
32.5
33.0
33.8
33.2
33.9
HW
mpg
57.3
57.7
59.9
59.0
60.5
57.2
58.0
59.7
59.7
61.0
46.5
46.8
48.3
46.2
47.4
44.2
44.5
46.1
45.5
46.8
42.2
42.6
43.0
42.4
43.5
28.4
28.6
29.6
29.0
29.7
Comb
mpg
62.8
63.2
65.2
64.9
66.2
59.7
60.6
62.5
62.3
63.6
48.2
48.7
50.1
48.1
49.4
47.2
47.8
49.2
49.3
50.5
45.0
45.1
45.4
45.4
46.2
30.5
30.9
31.8
31.2
31.8
0-30mph
s
3.8
3.8
3.8
3.7
3.7
3.6
3.6
3.6
3.4
3.4
3.4
3.4
3.4
3.0
3.0
3.9
3.9
3.9
3.7
3.7
3.8
3.8
3.8
3.6
3.6
3.3
3.3
3.3
3.1
3.1
0-60mph
s
9.6
9.6
9.6
10.0
10.0
8.6
8.6
8.6
8.6
8.6
7.7
7.7
7.7
7.1
7.1
9.4
9.4
9.4
9.3
9.3
9.1
9.1
9.1
8.8
8.8
7.9
7.9
7.9
7.8
7.8
%GHG
Reduction
31%
31%
33%
33%
35%
42%
42%
44%
44%
45%
43%
44%
45%
43%
44%
36%
36%
38%
38%
40%
44%
44%
44%
45%
45%
39%
40%
42%
40%
42%
       Table 3-13 shows the results from the nominal runs for the input powersplit vehicles".
No road load reductions are included in these results.  GHG reductions are in reference to the
2008 baseline vehicles.
n While input powersplit hybrids remain a very likely hybrid architecture choice for some manufacturers, the
Agencies focused on P2 hybrids compared to powersplits due to their apparent cost-effectiveness advantage in
                                              3-48

-------
                                      Technologies Considered in the Agencies' Analysis
                     Table 3-13: Nominal Powersplit hybrid modeling results
Vehicle Engine
Class Type
Small Car
Std Car
La rge Ca r
Small MPV
Large MPV
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
Displ.
L
0.59
0.59
0.59
1.66
1.66
0.83
0.83
0.83
2.4
2.4
1.13
1.13
1.13
3.8
3.8
0.9
0.9
0.9
2.6
2.6
1.05
1.05
1.05
3.15
3.15
Torque
Nm
124
124
124
138
138
176
176
176
200
200
238
238
238
317
317
190
190
190
217
217
221
221
221
263
263
EM size
kW
14
14
14
14
14
80
80
80
80
80
28
28
28
28
28
20
20
20
20
20
25
25
25
25
25
Battsize
kWh
0.70
0.70
0.70
0.70
0.70
1.00
1.00
1.00
1.00
1.00
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.15
1.15
1.15
1.15
1.15
Trans
Type
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
PS
FTP
mpg
64.7
65.8
67.7
64.2
67.3
55.6
57.9
58.0
53.3
56.4
46.6
48.0
47.9
40.3
43.0
49.1
50.8
51.3
44.3
49.3
44.8
45.7
47.0
41.7
44.3
HW
mpg
57.2
57.4
60.1
59.5
60.0
51.7
53.5
54.8
51.7
53.3
42.0
41.8
43.6
38.7
40.8
42.2
42.7
44.9
39.6
42.3
39.3
40.6
41.5
38.6
39.6
Comb
mpg
61.1
61.7
64.0
62.0
63.8
53.8
55.8
56.5
52.6
55.0
44.4
45.0
45.9
39.6
42.0
45.8
46.8
48.2
42.1
45.9
42.1
43.3
44.4
40.3
42.0
0-30mph
s
4.8
4.8
4.8
4.7
4.7
3.7
3.7
3.7
3.6
3.6
3.2
3.2
3.2
3.2
3.2
4.7
4.7
4.7
4.6
4.6
4.3
4.3
4.3
4.2
4.2
0-60mph
s
10.4
10.4
10.4
9.8
9.8
8.7
8.7
8.7
8.0
8.0
7.8
7.8
7.8
7.1
7.1
10.3
10.3
10.3
9.1
9.1
9.7
9.7
9.7
8.8
8.8
%GHG
Reduction
29%
30%
32%
30%
32%
35%
38%
38%
34%
37%
38%
39%
40%
31%
35%
33%
35%
37%
28%
34%
40%
42%
43%
37%
40%
3.3.1.2.15   Response Surface Model matrix runs

       After the nominal runs were completed according to the agreed-upon methodology,
Ricardo set up a design of experiment matrix for each vehicle architecture.  The continuously
swept variables were randomized in a Latin hypercube fashion to achieve a representative
sample within each matrix (reference the Ricardo report for more details on the complex
systems modeling approach used). After a data review and removal of runs with errors0 (as
needed) Ricardo then generated Response Surface Models (RSM) for use in the complex
systems tool. EPA used  the tool to evaluate a range of potential engine displacements, final
drive ratios and electric motor sizes (hybrids only) for each vehicle package, in an effort to
future years. As a result the powersplit nominal runs did not receive the same level of engineering scrutiny as
the P2 hybrid nominal runs.
0 e.g., model runs in which the vehicles were underpowered to the point where they could not follow the
prescribed vehicle speed trace, rendering an invalid test or "error". These configurations were then excluded
from the data sets.
                                              3-49

-------
                                     Technologies Considered in the Agencies' Analysis
find the combination that would provide the greatest effectiveness while meeting EPA's
definition of "equivalent performance".

3.3.1.2.16   Equivalent performance definition

       The Ricardo output data provides several performance metrics, as discussed in
3.3.1.2.10.  For simplicity, EPA assumed that a range of acceleration times for both a 0-60
mph test and also a 0-30 mph test (emphasizing launch character) would provide a simple yet
representative measure of a vehicle's equivalent performance. A range was chosen rather
than assuming a single point value equal to the baseline. This provided more acceptable data
points and reduced error due to "noise" in the datasets. The acceptable acceleration times
were as follows with respect to the baseline:

       0-60 mph: 5 percent slower to 15 percent faster as compared to baseline
       0-30 mph: 10 percent slower to 20 percent faster as compared to baseline

       The range above reflects a deviation from the actual baseline value that is well within
the normal variation of acceleration times for different vehicle models within a given vehicle
class.

3.3.1.2.17   Treatment of "turbo lag" in performance runs for turbocharged engines

       A common critique of comparisons of the modeled performance of highly
turbocharged engines with naturally-aspirated engines is that consideration must be given to
the delay in producing full engine load associated with the turbocharger, commonly referred
to as "turbo lag".  In technical discussions, Ricardo's engine experts assured EPA that the
dual-sequential designs of the turbocharger systems in the engines in this study should
mitigate most of this phenomenon  often seen on older-model vehicles.  However,  due to the
heavy reliance on turbocharged engines as a significant source of motive force for the high
BMEP engines evaluated in this project, EPA took this sensitivity further into account.

       Ricardo's initial model of WOT operation was based on a steady-state model of
engine torque, assuming that the engine would be able to instantaneously reach a desired level
of output torque,  without consideration of the intake manifold filling dynamics or the
mechanical inertia of the engine. EPA raised this as an issue, more in terms of properly
representing vehicle performance than for effectiveness differences. EPA reviewed its own
engine development data and proposed a somewhat conservative time constant for both the
naturally aspirated engines (0.3 s) and the turbocharged engines (1.5 s), to apply to the engine
torque response in the vehicle performance runs (these are shown below in Figure  3-7).  In
turn, Ricardo recalculated the acceleration times for the 0-30  and 0-60 mph runs to reflect the
slower time constants. As a result, EPA used these two performance metrics exclusively in
determining "equivalent performance". A transient engine/turbo model would have improved
the accuracy of the model somewhat; however, it was beyond the scope of this project.
                                            3-50

-------
                                     Technologies Considered in the Agencies' Analysis
          120.0%
          100.0%
                                                    — Ricardo-assumed

                                                     EPA-Nat.Asp.

                                                     EPA-Boosted
           0.0% -
              -0.5
                     0.5     1.5     2.5     3.5     4.5     5.5
                               Elapsed Time from WOT Command (sec)
                                                               6.5
                                                                      7.5
   Figure 3-7:  EPA proposed time constants and resulting effect on torque rise time for turbocharging

3.3.1.2.18   Treatment of engine response and "turbo lag" in cycle simulations and
            control logic algorithms

       The EASY5 model used in the Ricardo simulations included engine and driveline
inertia effects which account for some of the real-world transient torque delays. However, the
simulation modeling did not include an adjustment to account for transient engine response
delays (e.g. inclusion of time constant offsets), to simulate naturally aspirated and
turbocharged engine response delays associated with intake manifold gas dynamics and
turbocharger response delay. Consideration of engine response delay might affect how
transmission shift optimization control logic and advanced HEV control logic is structured,
and potentially affect GHG and fuel economy projections, particularly for boosted and
downsized engines. EPA and Ricardo believe that the impact is small over the city and
highway fuel economy test cycles.  However, the agencies seek comment on the fuel
economy impact of transient delays over the test cycles not accounted for in the Ricardo
modeling.
3.3.1.2.19   "Equivalent performance" results for conventional stop-start vehicles

       The following tables show the results from the complex systems tool, when
displacement, final drive ratio and electric motor size are varied to optimize GHG and fuel
consumption reduction effectiveness at equivalent performance for conventional stop-start, P2
and powersplit hybrids. Most of the vehicles show little change in performance between the
                                            3-51

-------
                                      Technologies Considered in the Agencies' Analysis
nominal runs and the equivalent performance results from the complex systems tool. Table
3-14 through Table 3-18 illustrate the various effects of changing road loads on the various
vehicle package configurations. Table 3-14, Table 3-16, and Table 3-16, respectively, show
the equivalent performance results for the conventional stop-start (for both automatic
transmissions and DCTs) and the P2 hybrid vehicles (modeled only as DCTs). No road load
reductions are included in Table 3-14 through Table 3-16.  For comparison, a second set of
tables (Table 3-17 and Table 3-18) give equivalent performance results for conventional stop-
start vehicles and P2 hybrids, each including example road load reductions13 of 20%  mass
reduction, 20% aerodynamic drag reduction and 10% rolling resistance reduction.

       The package effectiveness results from the equivalent performance runs were used in
the datasets to calibrate the individual technology effectiveness values within the lumped
parameter model. The development of the lumped parameter model is described in detail in
Section 1.5 of EPA's RIA.
Table 3-14:  Equivalent performance results for conventional-stop start vehicles (no road load reductions)
Vehicle Engine
Class Type
Small Car
Std Car
La rge Ca r
Small MPV
LargeMPV
Truck
HD Truck
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
2020 Diesel
Displ.
L
0.86
0.90
0.72
1.19
1.13
1.26
1.09
1.48
1.50
1.56
2.57
1.32
1.41
1.40
1.57
1.51
1.47
2.74
2.30
2.06
2.28
4.12
2.72
2.69
2.71
5.64
Torque
Nm
183
190
154
213
240
266
230
314
317
330
454
280
297
296
332
319
312
483
486
435
482
669
575
568
573
764
Trans
Type
AT6
AT6
AT6
AT6
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
FTP
mpg
53.1
56.3
55.2
57.3
44.4
47.0
46.2
37.0
39.2
38.6
39.1
38.9
41.1
40.0
34.8
36.2
36.4
36.7
24.0
25.0
24.8
26.8
16.6
17.2
17.3
21.0
HW
mpg
56.5
57.5
59.1
64.2
54.5
56.0
57.0
43.4
44.3
45.0
47.1
42.4
43.9
45.1
39.5
40.6
40.9
44.0
26.8
26.9
28.1
31.2
18.6
18.8
19.4
24.6
Comb
mpg
54.6
56.9
56.9
60.2
48.5
50.6
50.5
39.6
41.3
41.2
42.3
40.4
42.3
42.1
36.8
38.0
38.3
39.7
25.2
25.8
26.2
28.6
17.4
17.9
18.2
22.5
0-30mph
s
4.1
4.1
4.1
3.8
2.9
2.8
3.1
3.0
2.9
3.0
3.0
3.2
3.2
3.2
2.9
3.0
2.9
3.0
2.8
2.9
2.9
2.9
3.0
2.9
2.9
3.2
0-60mph
s
9.1
8.9
10.1
10.0
7.9
7.2
8.3
7.2
7.1
7.0
8.1
8.0
7.7
7.7
7.4
7.7
7.6
8.4
7.0
7.6
7.2
8.3
8.4
8.4
8.4
10.3
%GHG
Reduction
21%
24%
24%
20%
28%
31%
31%
31%
34%
34%
28%
25%
28%
28%
31%
34%
34%
29%
26%
28%
29%
28%
27%
29%
30%
37%
p Note that in the regulatory fleet analysis, levels of road load reduction technologies (e.g., mass reduction) will
vary by vehicle class.  These tables are illustrative in nature.
                                             3-52

-------
                                      Technologies Considered in the Agencies' Analysis
Table 3-15: Equivalent performance results for conventional-stop start vehicles with DCT transmissions
                                  (no road load reductions)
Vehicle Engine
Class Type
Small Car
Std Car
La rge Ca r
Small MPV
Large MPV
Truck
HD Truck
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
2020 Diesel
Displ.
L
0.91
0.92
0.89
1.13
1.08
1.29
1.17
1.53
1.66
1.48
2.44
1.30
1.32
1.33
1.53
1.56
1.56
2.42
2.23
2.26
2.25
3.78
2.55
2.62
2.58
5.45
Torque
Nm
193
196
188
204
229
273
248
324
352
313
431
276
280
282
324
330
330
427
472
478
475
613
538
554
544
739
Trans
Type
dry DCT6
dry DCT6
dry DCT6
dry DCT6
dry DCT8
dry DCT8
dry DCT8
dry DCT8
dry DCT8
dry DCT8
dry DCT8
dry DCT8
dry DCT8
dry DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
wet DCT8
FTP
mpg
55.0
58.0
57.2
61.4
46.4
48.7
48.1
38.4
40.5
40.0
41.0
40.1
42.1
41.7
36.0
38.0
37.6
39.2
24.8
25.9
25.8
28.1
17.3
17.8
18.0
21.8
HW
mpg
58.8
59.8
61.3
69.4
55.0
57.5
57.6
44.0
45.4
45.6
48.4
43.6
44.7
45.6
40.2
41.1
41.8
45.2
27.1
27.7
28.1
32.1
18.1
18.7
19.0
24.2
Comb
mpg
56.7
58.8
59.0
64.8
49.9
52.3
51.9
40.7
42.6
42.3
44.0
41.6
43.2
43.3
37.8
39.4
39.4
41.7
25.8
26.7
26.8
29.8
17.6
18.2
18.4
22.8
0-30mph
s
3.9
3.9
3.9
3.9
3.1
3.0
3.0
2.9
2.9
3.0
3.0
3.1
3.2
3.1
3.1
3.0
3.0
3.1
3.0
3.0
3.0
3.0
3.1
3.1
3.1
3.3
0-60mph
s
8.6
8.5
8.7
10.4
8.0
7.1
7.6
6.8
6.5
7.0
8.1
7.7
7.7
7.6
7.4
7.3
7.3
9.0
7.1
7.0
7.0
8.6
8.5
8.4
8.5
10.3
%GHG
Reduction
23%
26%
27%
26%
30%
33%
33%
33%
36%
35%
31%
27%
30%
30%
33%
36%
36%
33%
28%
31%
31%
31%
28%
30%
31%
38%
                                              3-53

-------
                             Technologies Considered in the Agencies' Analysis
Table 3-16: Equivalent performance results for P2 hybrids (no road load reductions)
Vehicle Engine
Class Type
Small Car
Std Car
La rge Ca r
Small MPV
LargeMPV
Truck
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
Displ.
L
0.68
0.68
0.67
1.72
1.68
1.00
0.95
1.04
2.54
2.31
1.39
1.37
1.38
3.73
3.33
1.40
1.39
1.41
3.87
3.59
1.31
1.30
1.29
3.13
3.00
1.87
1.92
1.92
5.34
5.34
Torque
Nm
144
144
142
143
140
213
202
219
212
193
292
289
291
311
278
295
293
297
322
299
276
274
272
262
250
394
404
405
445
445
EM size
kW
21
21
21
17
19
26
27
26
27
28
29
29
29
30
30
34
37
38
38
39
30
31
32
34
34
50
48
48
53
56
Battsize
kWh
0.70
0.70
0.70
0.70
0.70
1.00
1.00
1.00
1.00
1.00
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.15
1.15
1.15
1.15
1.15
1.50
1.50
1.50
1.50
1.50
Trans
Type
DCT6
DCT6
DCT6
DCT6
DCT6
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
FTP
mpg
68.9
70.1
72.0
72.0
74.4
62.2
63.2
64.8
64.6
65.7
50.6
51.3
52.6
48.6
50.7
52.3
53.0
54.4
53.6
55.2
48.5
49.0
49.2
48.0
48.5
33.3
33.6
34.6
32.3
32.7
HW
mpg
58.7
59.2
61.2
60.8
62.0
57.7
58.3
60.4
59.5
60.7
47.3
47.9
49.0
46.1
47.7
45.5
45.9
47.2
46.2
47.4
42.3
42.6
42.7
42.3
43.0
29.0
29.3
30.2
28.8
29.4
Comb
mpg
63.9
64.7
66.7
66.5
68.2
60.1
60.9
62.7
62.2
63.4
49.1
49.7
50.9
47.5
49.3
49.0
49.6
50.9
50.0
51.4
45.5
45.9
46.0
45.3
45.9
31.2
31.5
32.4
30.6
31.1
0-30mph
s
3.7
3.7
3.7
3.9
3.8
3.4
3.4
3.4
3.4
3.4
3.3
3.4
3.4
3.2
3.3
3.6
3.5
3.4
3.6
3.7
3.2
3.2
3.2
3.2
3.2
3.3
3.4
3.3
3.1
3.0
0-60mph
s
8.5
8.5
8.5
9.6
9.6
7.9
8.0
7.8
8.6
8.7
7.2
7.3
7.2
7.5
8.0
8.1
8.0
7.9
9.0
9.3
7.4
7.4
7.5
8.2
8.3
7.3
7.2
7.2
7.2
7.1
%GHG
Reduction
32%
33%
35%
35%
36%
42%
43%
44%
44%
45%
44%
45%
46%
42%
44%
38%
39%
40%
39%
41%
45%
45%
45%
44%
45%
40%
41%
43%
39%
40%
                                     3-54

-------
                                     Technologies Considered in the Agencies' Analysis
Table 3-17 Equivalent performance results for conventional-stop start vehicles (with 20% mass, 20%
                   aerodynamic drag and 10% rolling resistance reductions)
Vehicle Engine
Class Type
Small Car
Std Car
La rge Ca r
Small MPV
LargeMPV
Truck
HD Truck
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
2020 Diesel
STDI
LBDI
EGRB
2020 Diesel
Displ.
L
0.68
0.89
0.69
0.91
1.04
1.27
0.98
1.00
1.49
1.00
2.05
1.20
1.40
1.13
1.00
1.26
1.02
1.98
1.44
1.89
1.44
3.20
2.21
2.24
2.19
4.45
Torque
Nm
145
189
146
164
220
268
207
212
315
212
362
253
296
238
212
266
216
349
303
399
305
518
466
473
463
603
Trans
Type
AT6
AT6
AT6
AT6
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
ATS
FTP
mpg
65.0
68.9
67.6
71.8
53.9
57.3
56.2
46.5
48.4
48.5
48.5
46.3
49.1
48.4
42.4
44.2
44.2
46.4
29.4
30.2
30.5
32.8
20.0
20.5
20.9
25.3
HW
mpg
70.0
72.4
73.1
83.2
67.6
70.6
70.1
53.8
55.0
55.9
59.7
51.8
53.5
53.6
46.8
48.1
48.7
54.0
32.1
32.9
33.6
38.8
22.2
22.6
23.1
30.1
Comb
mpg
67.2
70.4
70.0
76.5
59.3
62.6
61.7
49.5
51.2
51.6
53.0
48.6
51.0
50.6
44.3
45.9
46.2
49.6
30.6
31.3
31.8
35.3
20.9
21.4
21.8
27.3
0-30mph
s
4.1
4.2
4.1
3.7
2.9
2.8
3.0
3.1
3.0
3.1
3.0
3.2
3.3
3.2
3.2
2.9
3.2
3.0
3.1
2.8
3.1
3.0
3.0
3.0
3.0
3.2
0-60mph
s
9.2
8.4
9.2
10.4
7.2
6.4
7.6
8.1
6.5
8.1
8.1
7.4
6.9
7.7
8.8
7.3
8.7
9.0
8.6
7.0
8.6
8.6
8.4
8.4
8.4
10.3
%GHG
Reduction
35%
38%
38%
37%
41%
44%
43%
45%
46%
47%
42%
37%
40%
40%
43%
45%
45%
43%
39%
41%
42%
41%
39%
41%
42%
48%
                                             3-55

-------
                                      Technologies Considered in the Agencies' Analysis
 Table 3-18: Equivalent performance results for P2 hybrids (with 20% mass, 20% aerodynamic drag and
                              10% rolling resistance reductions)
Vehicle Engine
Class Type
Small Car
Std Car
La rge Ca r
Small MPV
LargeMPV
Truck
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
STDI
LBDI
EGRB
ATKCS
ATKDVA
Displ.
L
0.68
0.68
0.68
1.60
1.52
0.90
0.91
0.92
2.36
2.03
1.21
1.25
1.25
3.52
3.29
1.25
1.22
1.24
3.71
3.44
1.01
1.04
1.02
2.91
2.84
1.57
1.60
1.58
4.16
4.15
Torque
Nm
143
144
143
133
127
191
194
194
197
169
254
263
263
293
274
265
257
262
309
287
213
219
215
243
237
330
337
334
347
346
EM size
kW
11
11
11
11
11
18
18
18
18
18
22
21
21
21
21
21
22
21
21
21
28
28
26
21
22
41
38
40
38
39
Battsize
kWh
0.70
0.70
0.70
0.70
0.70
1.00
1.00
1.00
1.00
1.00
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.10
1.15
1.15
1.15
1.15
1.15
1.50
1.50
1.50
1.50
1.50
Trans
Type
DCT6
DCT6
DCT6
DCT6
DCT6
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
DCT8
FTP
mpg
85.8
87.6
89.5
89.4
93.9
78.1
79.7
81.4
82.2
83.5
63.2
64.9
65.7
61.1
63.9
63.9
65.2
66.5
65.0
67.5
59.5
61.0
60.6
58.9
60.1
39.4
40.3
41.0
39.9
41.4
HW
mpg
72.2
73.1
75.4
74.9
76.9
71.1
72.2
74.2
73.8
76.2
57.3
58.5
59.8
57.0
59.3
53.4
53.9
55.7
55.1
56.7
50.2
50.9
51.6
51.1
52.4
34.4
35.0
36.0
34.9
35.9
Comb
mpg
79.1
80.4
82.5
82.2
85.4
74.8
76.2
78.0
78.2
80.0
60.4
61.9
62.9
59.2
61.7
58.7
59.5
61.1
60.1
62.1
54.9
56.0
56.2
55.1
56.3
37.0
37.7
38.6
37.5
38.7
0-30mph
s
3.7
3.7
3.7
3.8
3.8
3.2
3.3
3.2
3.1
3.3
3.1
3.1
3.1
3.0
3.0
3.5
3.5
3.5
3.4
3.6
3.2
3.2
3.2
3.2
3.2
3.2
3.3
3.2
3.0
3.0
0-60mph
s
7.9
7.9
8.0
8.9
9.0
7.2
7.2
7.1
7.5
8.3
6.6
6.5
6.6
6.7
6.8
7.7
7.7
7.8
8.2
8.7
7.4
7.3
7.3
7.5
7.7
7.0
7.0
7.0
7.1
7.2
%GHG
Reduction
45%
46%
47%
47%
49%
53%
54%
55%
55%
56%
55%
56%
56%
54%
56%
48%
49%
50%
49%
51%
54%
55%
55%
54%
55%
50%
51%
52%
50%
52%
3.3.1.2.20   Validation of vehicle simulation results

       Ricardo described the process used to validate the baseline vehicles in its report1.
Ideally it would be desirable to validate the simulation results with actual vehicle certification
test data. However, due to the nature and intended time frame (10+ years into the future) of
the technologies modeled within the vehicle classes, it is difficult to find many real-world
examples of specific technologies at the level of development reflected within the latest
simulation models.  Furthermore, there are no current vehicles in production that contain all
(or even a majority) of the multiple advanced technologies embedded within the models so it
is difficult to make meaningful direct comparisons between actual vehicles and model results.
Finally, there is no direct way to disaggregate the various advanced technologies and isolate
only the relevant pieces for evaluation (e.g., an advanced turbocharged engine at an interim
BMEP level with a baseline-level transmission without stop-start): the lumped parameter
model was developed for this very analytical capability.  A full description of the lumped
parameter model (including example comparisons of existing vehicle models to lumped
parameter estimates) is provided in 3.3.2.
                                             3-56

-------
                                      Technologies Considered in the Agencies' Analysis
3.3.1.2.21     The "efficient frontier" capability in Complex Systems tool

       A powerful feature of the Complex Systems tool is the "efficient frontier" function,
which provides a graphical representation of the RSM data for the vehicle configuration of
interest.  The user can identify the combination of various attributes (engine displacement,
final drive ratio, motor size, etc) which project the best model effectiveness.  Figure 3-8
below is an example of the efficient frontier for a Standard Car with a cooled EGR
turbocharged engine and a dry clutch DCT. The light red line along the top of the data set
represents the best fuel economy at each 0-60 mph acceleration time within the desired
window. The solid dark blue points represent the combinations that achieve  both the desired
0-60 and 0-30 mph criteria for equivalent performance. In this way, it is easy to quantify the
best effectiveness for a given technology package.
                              Efficient Frontier: Camry_ConvSS CamryJEGRBJKT
          ,:, |J

          59.9

          59.8

          58.7
          59.4

          59, -.

          w.j

          59.1

          59*

          58.9
          58.4

          58.3

          r- 1.2

          r :.'.

          58,1

          5 '.8

          57,8
          573

          57.4

          'J ", !

          572
                   Acceptable 0-60 mph time window
                                      8.5   9.0   9.5
                                         0-6O mph
                 Figure 3-8:  "Efficient Frontier" function in complex systems tool
                                              3-57

-------
                                    Technologies Considered in the Agencies' Analysis
3.3.1.2.22   Significance of the Complex Systems tool

       The complex systems tool was used not only to identify the optimal combination of
input variables for each vehicle architecture, but also to analyze trends in the input variables
for quality assurance (i.e., to make sure the response surface models made engineering sense),
and to establish numerical relationships between these variables for the lumped parameter
model calibration. Shown below are a few examples of the types of inquiries made via the
complex systems tool:
                                           3-58

-------
                                     Technologies Considered in the Agencies' Analysis
3.3.1.2.22.1 Effects of motor size (HEVs)

       EPA reviewed the effects of motor size on hybrids. As motor size is increased, there
is more opportunity to recapture energy during braking (because more powerful motors can
recover all of the energy in more severe braking events). However, oversized motors also
experience reduced efficiency as they operate in a less efficient operating region. This is
shown in Figure 3-9 below, which shows a sweep of motor size vs. fuel economy for both the
FTP/HWFE combined and also the high speed/load US06 cycle. Note that the optimum
motor size increases with respect to the US 06 cycle due to more severe braking and
acceleration rates.
                      Monte Carlo Results
                                                             Plot 2: Monte Carlo Results
                                                    -T

                                                    46


                                                    45
                                                    44

                                                    43

                                                    42-

                                                    41

                                                    40
            0.5 0.6 0.7 D.B  0.9 1.0 1.1 1.2 1.3  1.4 1.5 1.6 1.7
                           EMSIZE
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3  1.4 1.5 1.6  1.7
              EMSIZE
Figure 3-9: Electric motor sweeps for Standard Car class, P2 hybrid with stoichiometric GDI engine (left
                            = FTP/HWFE test; right = US06 test)
3.3.1.2.22.2 Effects of engine displacement

       EPA reviewed the effects of engine displacement at equivalent performance to
determine if there would be an "optimal" range of downsizing for best effectiveness.
Surprisingly, there was little benefit beyond downsizing the engine past a minimal point.
Shown in Figure 3-10 is an example complex systems tool graph with fuel economy plotted
against engine displacement multiplier (compared to the "nominal" engine displacement) for
the Truck class for three gasoline turbocharged engine packages and one diesel engine
package (note all packages included 20% weight reduction, 20% aerodynamic drag reduction,
and 10% rolling resistance reduction):
                                             3-59

-------
                                    Technologies Considered in the Agencies' Analysis
       •   The diesel engine result shows that the nominal engine in this case was originally
          oversized because it was scaled on engine power not more accurately on engine
          torque and continued displacement reduction would improve fuel economy. For
          this package, the displacement for optimal fuel economy is smaller than 50% of
          the nominal value, however; when considering equivalent vehicle performance, the
          minimum diesel displacement increases to roughly 70% of the nominal value.

       •   In contrast, the gasoline turbo engine results shown reflect a relative insensitivity
          of displacement to fuel economy for these advanced vehicles.
            39

            38

            37

         -36

         5 35
         I
         a 34
         LL
         ffl
            32

            31

            30

            29

            28
                           F-150 Monte Carlo Results
               0.5   0.6   0.7   0.8   0.9   1,0    1,1
                          DISPLACEMENT MULTIPLIER
1.2    1.3
         Figure 3-10: Example displacement sweep for Truck class in complex systems tool
       Figure 3-10 shows that as modeled, the swept displacement range is not large enough
for the advanced gasoline turbocharged engines. The displacement multiplier for these
engines must be greater than 1.3x the nominal displacement before the fuel economy would
degrade substantially. As the displacement drops below about 65% of the nominal (already
downsized) value, the efficiency decreases, as the engine load must be much higher to provide
the same required power. Regardless, the total fuel efficiency decrease from optimal is rather
                                           3-60

-------
                                      Technologies Considered in the Agencies' Analysis
small compared to today's engines. A 27-bar cooled EGR turbocharged GDI engine map for
a large carq was reverse-engineered from the Ricardo 10 hz output data, and is provided in
Figure 3-11.  The efficiency of this family of engines is very robust to changes in engine
displacement because the highlighted BSFC region of interest (the second one out from the
minimum BSFC "island") spans a large speed and load range. As a result, significant changes
in displacement do not greatly reduce fuel efficiency.  As displacement increases, the average
operating points for the engine over a given test cycle will trend towards the lower left (lower
speed, lower loadr) portion of the map.  In this case the points on the plot exist within the
same BSFC contour, so there is little degradation in engine efficiency with increasing
displacement (and drivetrain efficiency may improve at higher gears, potentially resulting in a
fuel economy increase).  Were the displacement to be increased much further, the operating
region would cross the contour and fuel efficiency would begin to drop much more
dramatically.
q The 27 bar, cooled EGR turbocharged engine maps are similar for all classes as they originated from a common
reference map and scaled according to engine displacement, as described in Section 6.3 of the 2011 Ricardo
report.
r Load decreases as it is reflective of a % of the maximum achievable torque and torque is increasing with
increased displacement.  Speed decreases because of the greater torque available combined with the shift
optimizer algorithm (allowing for a greater propensity to operate in higher gears).

                                              3-61

-------
                                 Technologies Considered in the Agencies' Analysis
                                      Avg. load (bmep) and
                                      speed decrease with
                                      increasing displacement
                 1000
2000
3000
  RPM
5000
6000
 Figure 3-11: Advanced engine BSFC map (27-bar cooled EGR turbocharged GDI engine for large car)

3.3.1.2.23   Effects of mass reduction

      With the complex systems tool EPA isolated the effectiveness of mass reduction on
advanced vehicle technology packages.  Figure 3-12 below shows a mass reduction sweep
plot of the Large MPV class for a conventional STDI and P2 hybrid vehicle with an Atkinson
engine.
                                       3-62

-------
                                     Technologies Considered in the Agencies' Analysis
                       Large MPV - weight sweep at equivalent performance

        "S
                                    Atkinson-P2 hybrid
                                                 -4.6% per 10% WR
                                             Stolen GDI engine
                                                          ~5.2% per 10% WR
              0.725 0.750 0,775 0.800  0.825  0.850  0.875 0.900 0.925 0.950 0.975  1.000  1.025  1.050  1,075 1.100 1.125
                                         Weight Factor

  Figure 3-12: Mass reduction sweep for Large MPV class at baseline equivalent performance. Engine
                      displacement and motor size (hybrids) held constant.

       The mass reduction effectiveness, originally estimated at roughly 6% GHG reduction
for a 10% reduction in mass, has been revised to reflect data such as that shown above.
Isolated from benefits due to engine downsizing opportunities, the effectiveness of weight
reduction for the non-hybrid packages is on the order of 5% per 10% weight reduction, while
mass reduction for the P2 hybrid (or any hybrid) is reduced, on the order of 4.5% per 10%
reduction due to the synergies with brake energy recovery (less braking energy is recoverable
because the vehicle weighs less). The lumped parameter tool was also revised to  incorporate
the synergies of weight reduction and hybrids.

3.3.1.2.24   Vehicle simulation report peer review process

       As previously discussed, vehicle  simulation modeling is a very detailed,
mathematically intensive approach which relies heavily on numerical engineering inputs.
These inputs (e.g., engine maps, transmission efficiency, control logic, etc.) are the heart of
the model and are derived  directly from proprietary engineering knowledge of components
and subsystems.  To simulate advanced engine and vehicle concepts, state-of-the-art
knowledge must be applied and converted into modeling inputs.  Public domain information
is rarely at the forefront of technology, and of little use in modeling vehicles in the 2017-2025
time frame.
                                             3-63

-------
                                     Technologies Considered in the Agencies' Analysis
       Engineering details on advanced vehicle technologies are closely guarded in industry,
and engineering services companies which develop and generate this confidential information
rely on it to remain competitive in the marketplace. Therefore, it is difficult, if not
impossible, to be completely transparent with an advanced vehicle simulation model and
make all of the inputs available for public review. EPA commissioned an external peer
review of the 2011 Ricardo simulation project and report. The peer reviewers selected were
highly respected members of academia and industry, all with substantial backgrounds in
automotive technology. The list of peer reviewers and their credentials is provided in the
associated peer review report31.

       EPA charged the peer reviewers to thoroughly evaluate the body of work with respect
to the following topics:

       •  Adequacy of the numerical inputs (engine technology selection, battery inputs,
          accessory load assumptions, etc) and highlight any caveats or limitations that
          would affect the final results.
       •  Validity and applicability of the simulation methodology, and if it adequately
          addresses synergies
       •  The results, and their validity and applicability to the light-duty vehicle fleet in the
          2020-2025 timeframe.
       •  Completeness of the report (does  it offer enough detail of the modeling process)
       •  The overall adequacy of the report for predicting the effectiveness of these
          technologies, and suggest recommendations for improvement
       The first round of comments was reflective of the reviewer's lack of access to model
inputs. Because the confidential inputs were initially withheld (for reasons described above),
"lack of transparency" was a consistent theme amongst the reviewers, so much that they
expressed frustration with their ability to evaluate the model methodology and the quality of
the inputs.  Additionally, due to the lack of access to Ricardo proprietary input data the peer
reviewers expressed concern that they could not adequately judge the validity or accuracy of
the input information or the simulation results.  EPA worked with Ricardo to provide the peer
reviewers with access to all of the detailed confidential modeling inputs under non disclosure
agreements. With this necessary information, 3 of the 5 peer reviewers submitted a second
round of comments which were generally more specific.  In turn, Ricardo modified the report
to address some of the comments, and they developed a response to comments document
which covered the comments from the peer review. One common theme called for increased
detail in how the inputs were generated. To address these requests, Ricardo provided the
detailed case studies that were used in the development of the engine maps for the cooled
EGR boosted engines and the Atkinson engines for hybrids. Ricardo also elaborated on the
hybrid control strategy, complete with state flow diagrams of operating modes, as well as a
discussion of how hybrid control strategy was optimized.  Additional transmission input
details were provided, including an overview of the development of advanced gear
efficiencies and how the optimized shift strategy was applied.
                                            3-64

-------
                                    Technologies Considered in the Agencies' Analysis
       The docket to this proposal contains Ricardo's response to comment document (which
includes the first version of the Ricardo report that was peer reviewed and both rounds of peer
                                        QOQQ
review comments), and Ricardo's final report.    The agencies seek comment on the all of
these references and  on the responsiveness of the final report to the peer review comments.

       3.3.1.3 Future Argonne National Laboratory Simulation Study

       The U.S. D.O.T. Volpe Center has entered into  a contract with Argonne National
Laboratory (ANL) to provide full vehicle simulation modeling support for this MYs 2017-
2025 rulemaking.  While modeling was not completed in time for use in this NPRM, NHTSA
intends to use this modeling to validate/update technology effectiveness estimates and
synergy factors for the final rulemaking analysis. This simulation modeling will be
accomplished using ANL's full vehicle simulation tool called "Autonomie," which is the
successor to ANL's Powertrain System Analysis Toolkit (PSAT) simulation tool, and ANL's
expertise with advanced vehicle technologies.

3.3.2   Lumped parameter Modeling

       3.3.2.1 Overview of the lumped parameter model

       As a more practical alternative to full vehicle simulation, EPA developed a "lumped
parameter model" that estimates the effectiveness of various technology combinations or
"packages," in a manner that accounts for synergies between technologies.  In the  analysis
supporting the MYs 2012-2016 light duty vehicle GHG and CAFE rule, EPA built over 140
packages for use in its OMEGA model, which spanned 19 vehicle classes and over 1100
vehicle models.  Vehicle simulation modeling performed for EPA by Ricardo, PLC, was used
to calibrate the lumped parameter model. Although DOT's analysis  supporting the MYs
2012-2016 CAFE rule applied technologies incrementally, rather than specifying packages in
advance, DOT calibrated CAFE model inputs, using EPA's lumped parameter model, to
harmonize as fully as practical with estimates produced by EPA's lumped parameter model.

       To support this rulemaking, EPA has updated its lumped parameter model and
calibrated it with updated vehicle simulation work performed  for EPA by Ricardo, PLC. As
in the MYs 2012-2016 rulemaking, DOT has calibrated inputs including synergy factors, to
the CAFE model to as fully as practical align with estimates produced by EPA's lumped
parameter model.

       Both agencies are continuing to conduct and sponsor vehicle  simulation efforts to
improve inputs to the agencies' respective modeling systems,  and both agencies expect that
the final rules will be informed by this  ongoing work.

       The basis for EPA's lumped parameter analysis is a first-principles energy balance
that estimates the manner in which the  chemical energy of the fuel is converted into various
forms of thermal and mechanical energy on the vehicle. The analysis accounts for the
dissipation of energy into the different  categories of energy losses, including each of the
following:

                                           3-65

-------
                                      Technologies Considered in the Agencies' Analysis
          Second law losses (thermodynamic losses inherent in the combustion of fuel),
          Heat lost from the combustion process to the exhaust and coolant,
          Pumping losses, i.e., work performed by the engine during the intake and exhaust
          strokes,
          Friction losses in the engine,
          Transmission losses, associated with friction and other parasitic losses of the
          gearbox, torque converter (when applicable) and driveline
          Accessory losses, related directly to the  parasitics associated with the engine
          accessories,
          Vehicle road load (tire and aerodynamic) losses;
          Inertial losses (energy dissipated as heat in the brakes)
The remaining energy is available to propel the vehicle. It is assumed that the baseline vehicle
has a fixed percentage of fuel lost to each category. Each technology is grouped into the
major types of engine loss categories it reduces. In this way, interactions between multiple
technologies that are applied to the vehicle may be determined. When a technology is applied,
the lumped parameter model estimates its effects by modifying the appropriate loss categories
by a  given percentage. Then, each subsequent technology that reduces the losses in an already
improved category has less of a potential impact than it would if applied on its own.
Using a lumped parameter approach for calculating package effectiveness provides necessary
grounding to physical principles. Due to the mathematical structure of the model, it naturally
limits the maximum effectiveness achievable for a family of similar technologies8. This can
prove useful when computer-simulated packages are compared to a "theoretical limit" as a
plausibility check. Additionally, the reduction of certain energy loss categories directly
impacts the effects on others. For example, as mass is reduced the benefits of brake energy
recovery decreases because there is not as much inertia energy to recapture.
Figure 3-13 is an example spreadsheet used by EPA to estimate the package effectiveness and
the synergistic impacts of a technology package for a standard-size car.
s For example, if only 4% of fuel energy is lost (in a baseline engine) to pumping work, leveraging multiple
technologies to theoretically eliminate all pumping losses would yield an aggregate reduction of no more than
15% in fuel consumption.

                                             3-66

-------
                                              Technologies Considered in the Agencies' Analysis
                                    EPA Stall Deliberative Materials-Do Not Quote
                      Vehicle Energy Effects Estimator
Vehicle Type
Standard car





%oftractiveenergy
Reduction
% of NEW fuel
Rated Power 1 Rated Torque
158 hp 161 ft-Ib
0 0
Gross Indie
Brake Energy
Road Loads
Mass Drag Tires
Braking / Aero Rolling
Inertia Load Load
23% 37% 40%
o% m 7%
ET
' 3625
0
ated Energ

Gearbox,
T.C
Losses

22.3%
4.4%
W | SOnph RL
Ib 11.3 hp
^^^^H

Total Engine Frictbn


Access Frictbn Punping
Losses Losses Losses

41.7% 15.4% 81.2%


Heat
Lost To
Exhaust &
Coolant
IndEff
Losses

320%
Package Notes
12VStop-Start
Stoich GDI Turbc

IiTeverc ibilities,
etc.

Second
Law

30%
                                                                              Evaluate New
                                                                              Reset LPModel
                      59.6%
                      76.5%
 80.6%
 84.9%
                               Tractiv
100.0%  17.3%
100.0%  24.7%
100.0%
94.2%
              Current Results
           66.1%  Fuel Consumption (GGE/mile)
           33.9%  FCRedictionw no-techs
           512%  FElmprovement (npgge)
           51.2%  FE Improvement (npg)
         |  305% ~| GHGreductbn vs 2008 Ricardo baselim
           335%  GHGreductbn vs no-techs
                          fodepemfent
         Technology           FC Estimate*
Original frictbn/brake rat
Based on PMEP/IMEP
(GM study)
         Vehicle mass reductb
         Aero Drag Reductb
                                     iking/stopped, inertia, rolling rt
                                             Implementation into estimator
                                                                         Check

                                                                         100.0%
 2008 Ricard) baseline ralues  includes some tec!
     Fuel Econon^ 320  npg (combined)
   Fuel Cbnsunptbn 0.031  gal/mi
    GHGemissions  284  g/miCO2E
 Regressed baseline ralues   assumes no techs
   reqd fueleneigy 11.95  kWh
     fueleconon^ 30.4  npg (unadj)
   fuelconsunptbn 0.033  gal/mi
    GHGemissions  299  g/miCO2E
  Current package values
     fueleconon^ 46.03  npg (unadj)
   fuelconsunptbn 0.022 ^gal/ni
    GHGemissions  197 'g/miCO2E
% or   User Picklist
Level  hcluda? (0/1) Devstatus
                                             20.5% pumping, -25% fric
                           4.0%  total WT Punping
                                             13.5% punping, +0.2% IE, -3.5% Me
                                             23.5% punping, +0.2% IE, -2.5% Me
                                     is,punping	punping,-5% tr
         Early upshift (formerly ASL)
         High voltage SS, with launch (BAS)
                                             11%B/L3% P, 3% F,2'
                                             +1.9% IE, 41% punping
                                    Ihd. Eff, punping  +1.3% IE, 41% punping
                                    Ihd. Eff, - punping +6% IE, -30% punping
                                                                    1
Pick one max
«—

Pick one
Additive to
Included
in P2
Pbkonemax
                                                        %EV=|  50% |
                        Figure 3-13 Sample lumped parameter model spreadsheet

         The LP model has been updated from the MYs 2012-2016 final rule to support the
MYs 2017-2025 proposed standards.  Changes were made to include new technologies for
2017 and beyond, improve fidelity for baseline attributes  and technologies, and better
represent hybrids based on more comprehensive vehicle simulation modeling. EPA RIA
Chapter 1 provides details of the methodology used to update and refine the model.

         3.3.2.2 Calibration of Lumped Parameter model to vehicle simulation data

         The LP model includes a majority of the new technologies being considered as part of
this proposed rulemaking. The results from the Ricardo vehicle simulation project (3.3.1)
were used to successfully calibrate the predictive accuracy and the synergy calculations that
                                                       3-67

-------
                                     Technologies Considered in the Agencies' Analysis
occur within the LP model. When the vehicle packages Ricardo modeled are estimated in the
lumped parameter model, the results are comparable. All of the baselines for each vehicle
class, as predicted by the LP model, fall within 3% of the Ricardo-modeled baseline results.
With a few exceptions (discussed in Chapter 1 of EPA's draft RIA the lumped parameter
results for the 2020-2025 "nominal" technology packages are within 5% of the vehicle
simulation results.  Shown below in Figures x-y are Ricardo's vehicle simulation package
results (for conventional stop-start and P2 hybrid packages1) compared to the lumped
parameter estimates.
                            Small Car Nominal Results
                                                                        Ricardo

                                                                        LP results
              Figure 3-14:  Comparison of LP to simulation results for Small Car class
 Refer to 3.3.1for definitions of the baselines, "conventional stop-start" and "P2 hybrid" vehicle architectures.

                                            3-68

-------
                           Technologies Considered in the Agencies' Analysis
                 Standard Car Nominal Results
                                                              I Ricardo
                                                              I LP results
              I    ConventionaISS   |  \
                                             P2 Hybrid
  Figure 3-15: Comparison of LP to simulation results for Standard Car class
                   Large Car Nominal Results
  60
  50
  40
oo
=• 30
  20

  10

   0
I    I   I   I
I    I   I   I
III
III
III
III
I Ricardo
I LP results
                       ConventionaISS
                                         P2 Hybrid
    Figure 3-16: Comparison of LP to simulation results for Large Car class
                                   3-69

-------
                           Technologies Considered in the Agencies' Analysis
                 Small MPV Nominal Results
  60






  50






  40
=• 30
  20






  10
n   •    •    •    •   •    •
     •   •    •    •    •   •    •
     •   •    •    •    •   •    •
I Ricardo



I LP results
     X   X
                      ConventionaISS
                                       P2 Hybrid
   Figure 3-17: Comparison of LP to simulation results for Small MPV class
                 Large MPV Nominal Results
                                                             Ricardo



                                                             LP results
   Figure 3-18: Comparison of LP to simulation results for Large MPV class
                                  3-70

-------
                                     Technologies Considered in the Agencies' Analysis
                              Truck Nominal Results
               Figure 3-19: Comparison of LP to simulation results for Truck class
       3.3.2.3 Comparison of results to real-world examples

       To validate the lumped parameter model, representations of actual late-model
production vehicles exhibiting advanced technologies were created.  Shown below in Table
3-19 are a set of select vehicle models containing a diverse array of technologies: included are
the pertinent technologies and vehicle specifications, along with actual vehicle certification
fuel economy test data compared to the lumped parameter fuel economy estimates.  For the
vehicles and technologies shown, the predicted fuel economy is within about 3% of the actual
data.
     Table 3-19: Production vehicle certification data compared to lumped parameter predictions
Vehicle
Vehicleclass
Engine
Transmission
HEV motor (kW)
ETW(lbs)
City/HWFE(mpg)
LP estimate (mpg)
Key technologies applied
in LP model
2011 Chevy Cruze ECO
Small Car
1.4LI4
turbo GDI
6 speed auto
n/a
3375
40.3
40.2
GDI (stoich.)
turbo (30% downsize)
ultra low R tires
active grill shutters
2011 Sonata Hybrid
Standard Car
2.4LI4
Atkinson
6 speed DCT
30
3750
52.2
51.7
P2 hybrid
aero improvements
2011 Escape Hybrid
Small MPV
2.5LI4
Atkinson
CVT
67
4000
43.9
44.0
Powersplit hybrid
2011 F-150 Ecoboost
Truck
3.5LV6
turbo GDI
6 speed auto
n/a
6000
22.6
21.9
GDI (stoich)
turbo (37% downsize)
                                            3-71

-------
                                     Technologies Considered in the Agencies' Analysis
3.4 What cost and effectiveness estimates have the agencies used for each technology?

       As discussed in the previous sections, many the effectiveness estimates for this
proposal, including the estimates for the technologies carried over from the MYs 2012-2016
final rule, been derived from the 2011 Ricardo study and corresponding updated version of
the lumped-parameter model.  It is important to note that when referencing the effectiveness
estimates from the MYs 2012-2016 the final rule the agencies used the average of the range
presented. If for example, the effectiveness range for technology X was determined to be 1 to
2 percent then the agencies used a value of 1.5 percent in their respective analyses.  However,
the effectiveness ranges that are presented for the MYs 2017-2025 analysis, as informed by
the Ricardo 2011 study, define the range of estimates used by the agencies for the different
vehicle types.  Again using technology X as an example, if the range is now defined as 2.0 to
2.5 percent then for small passenger cars (subcompact or compact) the estimated effectiveness
might be 2.0 percent but for large cars an estimate of 2.5 percent might be used.

3.4.1   Engine technologies

       One thing that is immediately clear from the cost tables that follow is that we have
updated our costing approach for some technologies in an effort to provide better granularity
in our estimates.  This is easily seen in Table 3-21, among others, where we list costs for
technologies by engine configuration—in-line or "I" versus "V"—and/or by number of
cylinders. In the 2012-2016 final rule, we showed costs for a small car, large car, large truck,
etc. The problem with that approach is that different vehicle classes can have many different
sized engines.  This will be especially true going forward as more turbocharged and
downsized engines enter the fleet.  For example, we project that many vehicles in the large car
class some of which, today, have V8 engines would have highly turbocharged 14 engines
under the proposal. As such, we would not want to estimate the large car costs of engine
friction reduction—which have always and continue to be based on the number of cylinders—
assuming that all large cars have V8 engines.
       3.4.1.1 Low Friction Lubricants

       One of the most basic methods of reducing fuel consumption in gasoline engines is the
use of lower viscosity engine lubricants. More advanced multi-viscosity engine oils are
available today with improved performance in a wider temperature band and with better
lubricating properties. This can be accomplished by changes to the oil base stock (e.g.,
switching engine lubricants from a Group I base oils to lower-friction, lower viscosity Group
III synthetic) and through changes to lubricant additive packages (e.g., friction modifiers and
viscosity improvers). The use of 5W-30 motor oil is now widespread and auto manufacturers
are introducing the use of even lower viscosity oils, such as 5W-20 and OW-20, to improve
cold-flow properties and reduce cold start friction. However, in some cases, changes to the
                                            3-72

-------
                                      Technologies Considered in the Agencies' Analysis
crankshaft, rod and main bearings and changes to the mechanical tolerances of engine
components may be required. In all cases, durability testing would be required to ensure that
durability is not compromised. The shift to lower viscosity and lower friction lubricants will
also improve the effectiveness of valvetrain technologies such as cylinder deactivation, which
rely on a minimum oil temperature (viscosity) for operation.

       Several manufacturers have previously commented confidentially, that low friction
lubricants could have an effectiveness value between 0 to 1 percent.  The agencies used the
average effectiveness of 0.5 in the MYs 2012-2016 final rule. For purposes of this proposal,
the agencies relied on the lump parameter model and the range for the effectiveness of low
friction lubricant is 0.5 to 0.8 percent.

       In the 2012-2016 rule, the 2010 TAR and the recent HD GHG rule,  EPA and NHTSA
used a direct manufacturing cost (DMC) of $3 (2007$) and considered that cost to be
independent of vehicle class since the engineering work required should apply to any engine
size. The agencies continue to believe that this cost is appropriate and have updated it to $3
(2009$) for this analysis. No learning is applied to this technology so the DMC remains $3
year-over-year. The agencies have used a low complexity ICM of 1.24 for this technology
through 2018 and 1.19 thereafter.  The resultant costs are shown in Table 3-20" Note that low
friction lubes are expected to exceed 85 percent penetration by the 2017 MY.

       Table 3-20 Costs for Engine Modifications to Accommodate Low Friction Lubes (2009$)
Cost
type
DMC
1C
TC
Engine
type
All
All
All
2017
$3
$1
$4
2018
$3
$1
$4
2019
$3
$1
$4
2020
$3
$1
$4
2021
$3
$1
$4
2022
$3
$1
$4
2023
$3
$1
$4
2024
$3
$1
$4
2025
$3
$1
$4
         DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are
         incremental to the baseline.
       3.4.1.2 Engine Friction Reduction

       In addition to low friction lubricants, manufacturers can also reduce friction and
improve fuel consumption by improving the design of engine components and subsystems.
Approximately 10 percent of the energy consumed by a vehicle is lost to friction, and just
over half is due to frictional losses within the engine.34 Examples include improvements in
low-tension piston rings, piston skirt design, roller cam followers, improved crankshaft design
and bearings, material coatings, material substitution, more optimal thermal management, and
piston and cylinder surface treatments.  Additionally, as computer-aided modeling  software
u Note that the costs developed for low friction lubes for this analysis reflect the costs associated with any
engine changes that would be required as well as any durability testing that may be required.
                                             3-73

-------
                                     Technologies Considered in the Agencies' Analysis
continues to improve, more opportunities for evolutionary friction reductions may become
available.

       All reciprocating and rotating components in the engine are potential candidates for
friction reduction, and minute improvements in several components can add up to a
measurable fuel economy improvement. In MYs 2012-2016 final rule, the agencies relied on
the 2002 NAS, NESCCAF and EEA reports as well as confidential manufacturer data that
suggested a range of effectiveness for engine friction reduction to be between 1 to 3 percent.
Because of the incremental nature of the CAFE model, NHTS A used the narrower range of 1
to 2 percent, which resulted in an average effectiveness of 1.5 percent. Based on the 2011
Ricardo study the effectiveness for engine friction reduction range has been changed to 2.0 to
2.7 percent..  For this proposal the agencies have added a second level of incremental
improvements in engine friction reduction over multiple vehicle redesign cycles. This second
level of engine friction reduction includes  some additional improvements to low friction
lubricant, relative to the low friction lubricant technology discussed above. The technologies
for this second level of engine friction reduction and low friction lubricants is considered to
be mature only after MY 2017. The effectiveness for this  second level, relative to the base
engine, is 3.4 to 4.8 percent based on the lump parameter model. Because of the incremental
nature of the CAFE model, NHTS A used the effectiveness range of 0.83 to 1.37 percent
incremental to the first level of engine friction reduction and low friction lubricants.

       In the 2012-2016 rule, the 2010 TAR and the HD GHG final rule, NHTS A and EPA
used a cost estimate of $11 (2007$) per cylinder direct manufacturing cost, or $12 (2009$) per
cylinder in this analysis. No  learning  is applied to this technology so the DMC remains $12
(2009$) year-over-year.  The agencies have used a low complexity ICM of 1.24 for this
technology through 2018 and 1.19 thereafter. The resultant costs are shown in Table 3-21.
Note that the first level of engine friction reduction is expected to exceed 85 percent
penetration by the 2017 MY.

                 Table 3-21 Costs for Engine Friction Reduction - Level 1 (2009$)
Cost
type
DMC
DMC
DMC
DMC
1C
1C
1C
1C
TC
TC
TC
TC
Engine
type
13
14
V6
V8
13
14
V6
V8
13
14
V6
V8
2017
$35
$47
$70
$93
$8
$11
$17
$23
$44
$58
$87
$116
2018
$35
$47
$70
$93
$8
$11
$17
$23
$44
$58
$87
$116
2019
$35
$47
$70
$93
$7
$9
$13
$18
$42
$56
$84
$111
2020
$35
$47
$70
$93
$7
$9
$13
$18
$42
$56
$84
$111
2021
$35
$47
$70
$93
$7
$9
$13
$18
$42
$56
$84
$111
2022
$35
$47
$70
$93
$7
$9
$13
$18
$42
$56
$84
$111
2023
$35
$47
$70
$93
$7
$9
$13
$18
$42
$56
$84
$111
2024
$35
$47
$70
$93
$7
$9
$13
$18
$42
$56
$84
$111
2025
$35
$47
$70
$93
$7
$9
$13
$18
$42
$56
$84
$111
         DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are
         incremental to the baseline.
                                            3-74

-------
                                      Technologies Considered in the Agencies' Analysis
       The agencies have estimated the DMC of this technology—a second level of friction
reduction with a second level of low friction lube—at double the combined DMCs of the first
level of engine friction reduction and first level of low friction lube (double the DMC relative
to the baseline).  As a result, the costs of the second level of engine friction reduction are as
shown in Table 3-22. For EFR2 the agencies have used a low complexity ICM of 1.24
through 2024 and 1.19 thereafter

                 Table 3-22 Costs for Engine Friction Reduction - Level 2 (2009$)
Cost
type
DMC
DMC
DMC
DMC
1C
1C
1C
1C
TC
TC
TC
TC
Engine
type
13
14
V6
V8
13
14
V6
V8
13
14
V6
V8
2017
$76
$100
$146
$193
$18
$24
$35
$47
$95
$124
$182
$240
2018
$76
$100
$146
$193
$18
$24
$35
$47
$95
$124
$182
$240
2019
$76
$100
$146
$193
$18
$24
$35
$47
$95
$124
$182
$240
2020
$76
$100
$146
$193
$18
$24
$35
$47
$95
$124
$182
$240
2021
$76
$100
$146
$193
$18
$24
$35
$47
$95
$124
$182
$240
2022
$76
$100
$146
$193
$18
$24
$35
$47
$95
$124
$182
$240
2023
$76
$100
$146
$193
$18
$24
$35
$47
$95
$124
$182
$240
2024
$76
$100
$146
$193
$18
$24
$35
$47
$95
$124
$182
$240
2025
$76
$100
$146
$193
$15
$19
$28
$37
$91
$119
$175
$230
         DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are
         incremental to the baseline.
       3.4.1.3 Cylinder Deactivation

       In conventional spark-ignited engines throttling the airflow controls engine torque
output.  At partial loads, efficiency can be improved by using cylinder deactivation instead of
throttling.  Cylinder deactivation (DEAC) can improve engine efficiency by disabling or
deactivating (usually) half of the cylinders when the load is less than half of the engine's total
torque capability - the valves are kept closed, and no fuel is injected - as a result, the trapped
air within the deactivated cylinders is simply compressed and expanded as an air spring, with
reduced friction and heat losses.  The active cylinders combust at almost double the load
required if all of the cylinders were operating.  Pumping  losses are significantly reduced as
long as the engine is operated in this  "part-cylinder" mode.

       Cylinder deactivation control strategy relies on setting maximum manifold absolute
pressures or predicted torque within which it can deactivate the cylinders. Noise and
vibration issues reduce the operating range to which cylinder deactivation is allowed,
although manufacturers continue exploring vehicle changes that enable increasing the amount
of time that cylinder deactivation might be suitable.  Some manufacturers may choose to
adopt active engine mounts and/or active noise cancellations systems to address NVH
concerns and to allow a greater operating range of activation (and the agencies have estimated
the costs for doing so, as noted below). Manufacturers have legitimately stated that use of
                                             3-75

-------
                                    Technologies Considered in the Agencies' Analysis
DEAC on 4 cylinder engines would cause unacceptable NVH; therefore, as in the 2012-2016
rule and the TAR, the agencies are not applying cylinder deactivation to 4-cylinder engines in
evaluating potential emission reductions/fuel economy improvements and attendant costs.

       Cylinder deactivation has seen a recent resurgence thanks to better valvetrain designs
and engine controls.  General Motors and Chrysler Group have incorporated cylinder
deactivation across a substantial portion of their V8-powered lineups. Honda (Odyssey, Pilot)
offers V6 models with cylinder deactivation.

       Effectiveness improvements scale roughly with engine displacement-to-vehicle weight
ratio: the higher displacement-to-weight vehicles, operating at lower relative loads for normal
driving, have the potential to operate in part-cylinder mode more frequently.

       NHTSA and EPA reviewed estimates from the 2012-2016 final rule, TAR, the RIA for
the heavy-duty GHG and fuel consumption rule. The OMEGA model, which is based on
packages, applied  a 6 percent reduction in CO2 emissions depending on vehicle class. The
CAFE model, due to its incremental nature, used a range depending on the engine valvetrain
configuration.  For example, for DOHC engines which are already equipped with DCP and
DVVLD, there is little benefit that can be achieved from adding cylinder deactivation since
the pumping work has already been minimized and internal Exhaust Gas Recirculation (EGR)
rates are maximized, so the effectiveness is only up to 0.5 percent for DEACD. For SOHC
engines which have CCP and DVVLS applied, effectiveness ranged from 2.5 to 3 percent for
DEACS. For OHV engines, without VVT or VVL technologies, the effectiveness  for
DEACO ranged from 3.9 to 5.5 percent.

       For this proposal the agencies, taking into account the additional review and the work
performed for the  Ricardo study, have revised the estimates for cylinder deactivation. The
effectiveness for relative to the base engine is 4.7 to 6.5 percent based on the lump parameter
model. Because of the incremental nature of the CAFE model, NHTSA used the
effectiveness range of 0.44 to 0.66 percent incremental for SOHC and DOHC applications.
For OHV applications, the effectiveness was increased slightly with a range of 4.66 to 6.30
percent.

       In the 2012-2016 rule and  the 2010 TAR, the agencies used a DMC estimate of $140
(2007$) and $157  (2007$) for cylinder deactivation technology on V6 and V8 engines,
respectively. The  DMC's become $144 (2009$) and $162 (2009$) for this analysis and are
considered applicable in the 2015  MY. This technology is considered to be on the flat-portion
of the learning curve. The agencies have applied a low complexity ICM of 1.24 to this
technology through 2018 and 1.19 thereafter. The resultant costs are shown in Table 3-23.

                      Table 3-23 Costs for Cylinder Deactivation (2009$)
Cost
type
DMC
DMC
1C
Engine
type
V6
V8
V6
2017
$137
$154
$55
2018
$134
$151
$55
2019
$131
$148
$41
2020
$129
$145
$41
2021
$126
$142
$41
2022
$124
$139
$41
2023
$121
$136
$41
2024
$119
$134
$41
2025
$116
$131
$41
                                           3-76

-------
                                    Technologies Considered in the Agencies' Analysis
1C
TC
TC
V8
V6
V8
$62
$192
$216
$62
$189
$213
$46
$173
$194
$46
$170
$191
$46
$167
$188
$46
$165
$185
$46
$162
$182
$46
$160
$180
$46
$157
$177
         DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are
         incremental to the baseline.

       If lost motion devices are on the engine, the cost of DEAC as applied to SOHC and
DOHC engines could be as low as $32 in MY 2017.  This $32 accounts for the potential
additional application of active engine mounts on SOHC and DOHC engines and can only be
applied on 50 percent of the vehicles.  Further, this SOHC and DOHC engine estimate is
relevant to the CAFE model only because the OMEGA model does not apply technologies in
the same incremental fashion as the CAFE model.

       3.4.1.4 Variable Valve Timing (VVT)

       Variable valve timing (VVT) encompasses a family of valve-train designs that alter
the timing of the intake valve, exhaust valve, or both, primarily to reduce pumping losses,
increase specific power, and control the level of residual gases in the cylinder. VVT reduces
pumping losses when the engine is lightly loaded by controlling valve timing closer to an
optimum needed to sustain horsepower and torque. VVT can also improve volumetric
efficiency at higher engine speeds and loads. Additionally, VVT can be used to alter (and
optimize) the effective compression ratio where it is advantageous for certain engine
operating modes (e.g., in the Atkinson Cycle).

       VVT has now become a widely adopted technology: in MY 2010, approximately 86
percent of all new cars and light trucks had engines with some method of variable valve
timing.35  Manufacturers are currently using many different types of variable valve timing,
which have a variety of different names and methods. Manufacturers are currently using
many different types of variable valve timing, which have a variety of different names and
methods. Therefore, the degree of further improvement across the fleet is limited by the level
of valvetrain technology already implemented on the vehicles. Information found in the 2008
baseline vehicle fleet file is used to determine the degree to which VVT technologies have
already been applied to particular vehicles  to ensure the proper level of VVT technology, if
any, is  applied. The three major types of VVT are listed below.

       Each of the three implementations of VVT uses a cam phaser to adjust the camshaft
angular position relative to the crankshaft position, referred to as "camshaft phasing." The
phase adjustment results in changes to the pumping work required by the engine to
accomplish the gas exchange process.  The majority of current cam phaser applications use
hydraulically-actuated units, powered by engine oil pressure and managed by a solenoid that
controls the oil pressure supplied to the phaser.

3.4.1.4.1 Intake Cam Phasing (ICP)

       Valvetrains with ICP, which is the simplest of the cam phasing technologies, can
modify the timing of the inlet valves by phasing the intake camshaft while the exhaust valve
                                            3-77

-------
                                    Technologies Considered in the Agencies' Analysis
timing remains fixed. This requires the addition of a cam phaser on each bank of intake
valves on the engine.  An in-line 4-cylinder engine has one bank of intake valves, while V-
configured engines have two banks of intake valves.

      In the MYs 2012-2016 final rale and TAR, NHTSA and EPA assumed an
effectiveness range of 2 to 3 percent for ICP. Based on the 2011 Ricardo study and updated
lumped-parameter model the agencies have fined tuned the range to be 2.1 to 2.7 percent.

      In the 2012-2016 rale and the 2010 TAR, the agencies estimated the DMC of a cam
phaser needed for VVT-intake at $37 (2007$). This DMC becomes $38 (2009$) for this
analysis and is considered applicable in the 2015 MY.  This cost would be required for each
cam shaft controlling intake valves, as such an overhead cam 14 would need one phaser, an
overhead cam V6 or V8 would need two phasers, and an overhead valve V6 or V8 would
need just one. This technology is considered to be on the flat-portion of the learning curve.
The agencies have applied a low complexity ICM of 1.24 to this technology through 2018 and
1.19 thereafter. The resultant costs are shown in Table 3-24.

                         Table 3-24 Costs for VVT-intake (2009$)
Cost
type
DMC
DMC
DMC
1C
1C
1C
TC
TC
TC
Engine type
OHC-I4
OHC-V6/V8
OHV-V6/V8
OHC-I4
OHC-V6/V8
OHV-V6/V8
OHC-I4
OHC-V6/V8
OHV-V6/V8
2017
$36
$73
$36
$9
$18
$9
$46
$91
$46
2018
$36
$71
$36
$9
$18
$9
$45
$90
$45
2019
$35
$70
$35
$7
$15
$7
$42
$84
$42
2020
$34
$68
$34
$7
$15
$7
$42
$83
$42
2021
$34
$67
$34
$7
$15
$7
$41
$82
$41
2022
$33
$66
$33
$7
$15
$7
$40
$80
$40
2023
$32
$64
$32
$7
$15
$7
$40
$79
$40
2024
$32
$63
$32
$7
$15
$7
$39
$78
$39
2025
$31
$62
$31
$7
$15
$7
$38
$76
$38
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; OHC=overhead cam;
OHV=overhead valve; all costs are incremental to the baseline.
3.4.1.4.2 Coupled Cam Phasing (CCP)

       Valvetrains with coupled (or coordinated) cam phasing can modify the timing of both
the inlet valves and the exhaust valves an equal amount by phasing the camshaft of a single
overhead cam (SOHC) engine or an overhead valve (OHV) engine. For overhead cam
engines, this requires the addition of a cam phaser on each bank of the engine. Thus, an in-
line 4-cylinder engine has one cam phaser, while SOHC V-engines have two cam phasers.
For overhead valve (OHV) engines, which have only one camshaft to actuate both inlet and
                                           3-78

-------
                                    Technologies Considered in the Agencies' Analysis
exhaust valves, CCP is the only VVT implementation option available and requires only one
cam phaser.v

       The agencies' MYs 2012-2016 final rule estimated the effectiveness of CCP to be
between 1 to 4 percent.  Due to the incremental nature and decision tree logic of the Volpe
model, NHTSA estimated the effectiveness for coupled cam phasing on a SOHC engine to be
1 to 3 percent and 1 to 1.5 percent for coupled cam phasing on an overhead valve engine.

       For this proposal the agencies, taking into account the additional review and the work
performed for the 2011 Ricardo study, have revised the estimates for cylinder deactivation.
The effectiveness relative to the base engine is 4.1 to 5.5 percent based on the lump parameter
model. Because of the incremental nature of the CAFE model, NHTSA used the incremental
effectiveness range of 4.14 to 5.36 percent for SOHC applications; an increase over the 2012-
16 final rule and 2010 TAR. For OHV applications, CCP was paired with discrete variable
valve lift (DVVL) to form a new technology descriptor called variable valve actuation
(VVA). VVA will be discussed later in Chapter 3.

       The same cam phaser has been assumed for intake cam phasing as for coupled cam
phasing, thus CCPs are identical to those presented in Table 3-24.
3.4.1.4.3 Dual Cam Phasing (DCP)

       The most flexible VVT design is dual (independent) cam phasing, where the intake
and exhaust valve opening and closing events are controlled independently. This allows the
option of controlling valve overlap, which can be used as an internal EGR strategy.  At low
engine loads, DCP creates a reduction in pumping losses, resulting in improved fuel
consumption/reduced COi emissions. Increased internal EGR also results  in lower engine-out
NOx emissions. The amount by which fuel consumption is improved and  COi emissions are
reduced depends on the residual tolerance of the combustion system. Additional
improvements are observed at idle, where low valve overlap could result in improved
combustion stability, potentially reducing idle fuel consumption.

       For the 2012-2016 final rule and TAR the EPA and NHTSA assumed an effectiveness
range for DCP to be between 3 to 5 percent relative to a base engine or 2 to 3 relative to an
engine with ICP. The agencies have updated this range, based on the updated lumped-
parameter model to be 4.1 to 5.5 percent relative to a base engine or 2.0 to 2.7 percent relative
to an engine with ICP.
v It is also noted that coaxial camshaft developments would allow other VVT options to be applied to OHV
engines. However, since they would potentially be adopted on a limited number of OHV engines NHTSA did
not include them in the decision tree.

                                            3-79

-------
                                     Technologies Considered in the Agencies' Analysis
       The costs for VVT-dual cam phasing are the same per phaser as described above for
VVT-intake. However, for DCP, an additional cam phaser is required for each camshaft
controlling exhaust valves. As a result, an overhead cam 14 would need two phasers, an
overhead cam V6 or V8 would need four phasers, and an overhead valve V6 or V8 would
need two. This technology is considered to be on the flat-portion of the learning curve. The
agencies have applied a medium complexity ICM of 1.39 to this technology through 2018 and
1.29 thereafter. The resultant costs are shown in Table 3-25.

                     Table 3-25 Costs for VVT-Dual Cam Phasing (2009$)
Cost
type
DMC
DMC
DMC
1C
1C
1C
TC
TC
TC
Engine type
OHC-I4
OHC-V6/V8
OHV-V6/V8
OHC-I4
OHC-V6/V8
OHV-V6/V8
OHC-I4
OHC-V6/V8
OHV-V6/V8
2017
$67
$143
$73
$27
$58
$29
$94
$201
$102
2018
$65
$140
$71
$27
$58
$29
$92
$198
$101
2019
$64
$138
$70
$20
$43
$22
$84
$181
$92
2020
$63
$135
$68
$20
$43
$22
$83
$178
$90
2021
$61
$132
$67
$20
$43
$22
$81
$175
$89
2022
$60
$130
$66
$20
$43
$22
$80
$173
$87
2023
$59
$127
$64
$20
$43
$22
$79
$170
$86
2024
$58
$124
$63
$20
$43
$22
$78
$167
$85
2025
$57
$122
$62
$20
$43
$22
$77
$165
$84
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; OHC=overhead cam;
OHV=overhead valve; all costs are incremental to the baseline.
       3.4.1.5 Variable Valve Lift (VVL)

       Controlling the lift of the valves provides a potential for further efficiency
improvements.  By optimizing the valve-lift profile for specific engine operating regions, the
pumping losses can be reduced by reducing the amount of throttling required to produce the
desired engine power output.  By moving the throttling losses further downstream of the
throttle valve, the heat transfer losses that occur from the throttling process are directed into
the fresh charge-air mixture just prior to compression, delaying the onset of knock-limited
combustion processes. Variable valve lift control can also be used to induce in-cylinder
mixture motion, which improves fuel-air mixing and can result in improved thermodynamic
efficiency. Variable valve lift control can also potentially reduce overall valvetrain friction.
At the  same time, such systems may also incur increased parasitic losses associated with their
actuation mechanisms.  A number of manufacturers  have already implemented VVL into their
fleets (Toyota, Honda, and BMW), but overall this technology is still available for most of the
fleet. There are two major classifications of variable valve lift, described below:

3.4.1.5.1 Discrete Variable Valve Lift (DVVL)

       Discrete variable valve lift (DVVL) systems  allow the selection between two or three
discrete cam profiles by means of a hydraulically-actuated mechanical system. By optimizing
the cam profile for specific engine operating regions, the pumping losses can be reduced by
reducing the amount of throttling required to produce the desired engine power output.  This
increases the efficiency of the engine. These cam profiles consist of a low and a high-lift
lobe, and may include an inert or blank lobe to incorporate cylinder deactivation (in the case
                                            3-80

-------
                                    Technologies Considered in the Agencies' Analysis
of a 3-step DVVL system). DVVL is normally applied together with VVT control.  DVVL is
also known as Cam Profile Switching (CPS). DVVL is a mature technology with low
technical risk.

NHTSA's and EPA's MY 2012-16 final rale, previously-received confidential manufacturer
data, and report from NESCCAF, all estimated the effectiveness of DVVL to be between 1 to
4 percent above that realized by VVT systems. Based on the 2011 Ricardo study, NHTSA
and EPA have revised the effectiveness range of DVVL systems to 2.8 to 3.9 percent above
that realized by VVT systems.

       In the 2012-2016 rale and the 2010 TAR, the agencies estimated the DMC of DVVL
at $116 (2007$), $169 (2007$) and $241 (2007$) for an 14, V6 and V8 engine, respectively.
These DMCs become $120 (2009$), $174 (2009$) and $248 (2009$) for this analysis all of
which are considered applicable in the 2015MY. This technology is considered to be on the
flat-portion of the learning curve and is applicable only to engines with overhead cam
configurations.  The agencies have applied a medium complexity ICM of 1.39  to this
technology through 2018 and 1.29 thereafter. The resultant costs are shown in Table 3-26.

                    Table 3-26 Costs for Discrete Variable Valve Lift (2009$)
Cost
type
DMC
DMC
DMC
1C
1C
1C
TC
TC
TC
Engine type
OHC-I4
OHC-V6
OHC-V8
OHC-I4
OHC-V6
OHC-V8
OHC-I4
OHC-V6
OHC-V8
2017
$114
$165
$236
$46
$67
$96
$160
$232
$332
2018
$112
$162
$231
$46
$67
$95
$158
$229
$327
2019
$109
$159
$227
$34
$50
$71
$144
$209
$298
2020
$107
$156
$222
$34
$50
$71
$142
$205
$293
2021
$105
$152
$218
$34
$50
$71
$139
$202
$289
2022
$103
$149
$213
$34
$50
$71
$137
$199
$284
2023
$101
$146
$209
$34
$50
$71
$135
$196
$280
2024
$99
$143
$205
$34
$49
$71
$133
$193
$276
2025
$97
$141
$201
$34
$49
$70
$131
$190
$271
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; OHC=overhead cam;
OHV=overhead valve; all costs are incremental to the baseline.
3.4.1.5.2 Continuously Variable Valve Lift (CVVL)

       In CVVL systems, valve lift is varied by means of a mechanical linkage, driven by an
actuator controlled by the engine control unit.  The valve opening and phasing vary as the lift
is changed and the relation depends on the geometry of the mechanical system. BMW has
considerable production experience with CVVL systems and has sold port-injected
"Valvetronic" engines since 2001.  Fiat is now offering "MultiAir" engines enabling precise
control over intake valve lift.  CVVL allows the airflow into the engine to be regulated by
means of intake valve opening reduction, which improves engine efficiency by reducing
pumping losses from throttling the intake system further upstream as with a conventionally
throttled engine.
                                           3-81

-------
                                    Technologies Considered in the Agencies' Analysis
       Variable valve lift gives a further reduction in pumping losses compared to that which
can be obtained with cam phase control only, with CVVL providing greater effectiveness than
DVVL, since it can be fully optimized for all engine speeds and loads, and is not limited to a
two or three step compromise. There may also be a small reduction in valvetrain friction
when operating at low valve lift, resulting in improved low load fuel consumption for cam
phase control with variable valve lift as compared to cam phase control only.  Most of the fuel
economy effectiveness is achieved with variable valve lift on the intake valves only. CVVL
is only applicable to double overhead cam (DOHC) engines.

       The 2012-2016 final rule estimated the effectiveness for CVVL at 1.5 to 3.5 percent
over an engine with DCP, but also recognize that it could go up as high as 5 percent above
and beyond DCP to account for the implementation of more complex CVVL systems such as
BMW's "Valvetronic" and Fiat "MultiAir" engines.  Thus, the effectiveness range for CVVL
in this joint TSD ranges from 1.5 to 7 percent depending on the complexity level of the
application. NHTSA and EPA believe this estimate continues to be applicable for this
proposal.

       For this rulemaking, NHTSA has increased the incremental effectiveness values for
this technology to a range of 3.6 to 4.9 percent from  1.5 to 3.5 percent in the MYs 20120-
2016 final rule.

       In the 2012-2016 rule and the 2010 TAR, the agencies estimated the DMC of CVVL
at $174 (2007$), $320 (2007$), $349  (2007$), $866 (2007$) and $947 (2007$) for an OHC-
14, OHC-V6, OHC-V8, OHV-V6 and OHV-V8  engine, respectively.  These DMCs become
$180 (2009$), $330 (2009$), $360 (2009$), $893 (2009$) and $977 (2009$) for this analysis
all of which are considered applicable in the 2015MY. This technology is considered to be on
the flat-portion of the learning curve.   The agencies have applied a medium complexity ICM
of 1.39 to this technology through 2018 and 1.29 thereafter.  The resultant costs are shown in
Table 3-27.

                  Table 3-27 Costs for Continuous Variable Valve Lift (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
TC
TC
TC
TC
Engine type
OHC-I4
OHC-V6
OHC-V8
OHV-V6
OHV-V8
OHC-I4
OHC-V6
OHC-V8
OHV-V6
OHV-V8
OHC-I4
OHC-V6
OHC-V8
OHV-V6
2017
$171
$313
$342
$849
$928
$69
$127
$139
$344
$376
$240
$440
$480
$1,193
2018
$168
$307
$335
$832
$910
$69
$127
$138
$343
$375
$237
$434
$473
$1,175
2019
$164
$301
$328
$815
$892
$52
$95
$103
$256
$280
$216
$396
$432
$1,072
2020
$161
$295
$322
$799
$874
$52
$94
$103
$256
$280
$212
$389
$425
$1,055
2021
$158
$289
$315
$783
$856
$51
$94
$103
$255
$279
$209
$383
$418
$1,038
2022
$155
$283
$309
$767
$839
$51
$94
$103
$255
$279
$206
$377
$412
$1,022
2023
$151
$278
$303
$752
$822
$51
$94
$102
$254
$278
$203
$372
$405
$1,006
2024
$148
$272
$297
$737
$806
$51
$94
$102
$254
$278
$200
$366
$399
$991
2025
$145
$267
$291
$722
$790
$51
$94
$102
$253
$277
$197
$360
$393
$976
                                           3-82

-------
                                     Technologies Considered in the Agencies' Analysis
|  TC  |   OHV-V8   |  $1,304 |  $1,285 | $1,172 | $1,154 | $1,136 | $1,118 | $1,101 | $1,084 |  $1,067 |
 DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; OHC=overhead cam;
 OHV=overhead valve; all costs are incremental to the baseline.

       3.4.1.6 Variable Valve Actuation (VVA)

       For this proposal, NHTSA has combined two valve control technologies for OHV
engines. Coupled cam phasing (CCPO) and discrete valve lift (DVVLO) into one technology
defined as variable valve actuation (VVA). The agency estimates the incremental
effectiveness for VVA applied to and OHV engine as 2.71 to 3.59 percent. This effectiveness
value is slightly lower than coupled cam phasing for overhead cam applications (CCPS) based
on the assumption that VVA would be applied to an OHV engine after cylinder deactivation
(DEAC). For more information on combining these technologies please refer to the NHTSA
specific Preliminary Regulatory Impact Analysis (PRIA).

       3.4.1.7 Stoichiometric Gasoline Direct Injection (SGDI)

       Stoichiometric gasoline direct injection (SGDI), or Spark Ignition Direct injection
(SIDI), engines inject fuel at high pressure directly into the combustion chamber (rather than
the intake port in port fuel injection).  SGDI requires changes to the injector design, an
additional high pressure fuel pump, new fuel rails to handle the higher fuel pressures and
changes to the cylinder head and piston crown design.  Direct injection of the fuel into the
cylinder improves cooling of the air/fuel charge within the cylinder, which allows for higher
compression ratios and  increased thermodynamic efficiency without the onset of combustion
knock. Recent injector design advances, improved electronic engine management systems
and the introduction of multiple injection events per cylinder firing cycle promote better
mixing of the air and fuel, enhance combustion rates, increase residual exhaust gas tolerance
and improve cold start emissions. SGDI engines achieve higher power density and match
well with other technologies, such as boosting and variable valvetrain designs.

       Several manufacturers are manufacturing vehicles with SGDI engines, including
VW/Audi, BMW, Toyota (Lexus IS 350), Ford (Ecoboost), and General Motors (Chevrolet
Impala and Cadillac CTS 3.6L). BMW, GM, Ford and VW/Audi have announced their plans
to increase dramatically the  number of SGDI engines in their portfolios.

       NHTSA and EPA reviewed estimates from the 2012-2016 final rule and TAR, which
stated an effectiveness range of SGDI to be between 2 and 3 percent. NHTSA and EPA
reviewed estimates from the Alliance of Automobile Manufactures, which projects 3 percent
gains in fuel efficiency  and a 7 percent improvement in torque. The torque increase provides
the opportunity to downsize the engine allowing an increase in efficiency of up to a 5.8
percent. NHTSA and EPA also reviewed other published literature,  reporting 3 percent
effectiveness for SGDI.36  Confidential manufacturer data reported an efficiency effectiveness
range of 1 to 2 percent.  Based on data from the recent Ricardo study and reconfiguration of
                                            3-83

-------
                                     Technologies Considered in the Agencies' Analysis
the new lumped parameter model, EPA and NHTSA have revised this value to 1.5 percentw.
Combined with other technologies (i.e., boosting, downsizing, and in some cases, cooled
EGR), SGDI can achieve greater reductions in fuel consumption and CCh emissions
compared to engines of similar power output.

       The NHTSA and EPA cost estimates for SGDI take into account the changes required
to the engine hardware, engine electronic controls, ancillary and Noise Vibration and
Harshness (NVH) mitigation systems.  Through contacts with industry NVH suppliers, and
manufacturer press releases, the agencies believe that the NVH treatments will be limited to
the mitigation of fuel system noise, specifically from the injectors and the fuel lines and have
included corresponding cost estimates for these NVH controls. In the 2012-2016 FRM, the
agencies estimated the DMC for SGDI at $213  (2007$), $321 (2007$) and $386 (2007$) for
13/14, V6 and V8 engines, respectively. These DMCs become $220 (2009$), $331 (2009$)
and $398 (2009$) for this analysis all of which are considered applicable in the 2012MY.
This technology is considered to be on the flat-portion of the learning curve.  The agencies
have applied a medium complexity ICM of 1.39 to this technology through 2018 and 1.29
thereafter. The resultant costs are shown in Table 3-28.

               Table 3-28 Costs for Stoichiometric Gasoline Direct Injection (2009$)
Cost type
DMC
DMC
DMC
1C
1C
1C
TC
TC
TC
Engine type
13/14
V6
V8
13/14
V6
V8
13/14
V6
V8
2017
$191
$287
$345
$84
$126
$152
$274
$413
$497
2018
$187
$281
$339
$84
$126
$152
$270
$407
$490
2019
$183
$276
$332
$62
$94
$113
$246
$370
$445
2020
$179
$270
$325
$62
$94
$113
$242
$364
$438
2021
$176
$265
$319
$62
$94
$113
$238
$359
$431
2022
$172
$260
$312
$62
$94
$113
$234
$353
$425
2023
$169
$254
$306
$62
$94
$112
$231
$348
$418
2024
$165
$249
$300
$62
$93
$112
$227
$343
$412
2025
$162
$244
$294
$62
$93
$112
$224
$338
$406
  DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are incremental to the baseline.

       3.4.1.8 Turbocharging and Downsizing (TRBDS)

       The specific power of a naturally aspirated engine is primarily limited by the rate at
which the engine is able to draw air into the combustion chambers. Turbocharging and
supercharging (grouped together here as boosting) are two methods to increase the intake
manifold pressure and cylinder charge-air mass above naturally aspirated levels. Boosting
increases the airflow into the engine, thus increasing the specific power level, and with it the
ability to reduce engine displacement while maintaining performance.  This effectively
reduces the pumping losses at lighter loads in comparison to a larger, naturally aspirated
engine.
w However, because GDI is a key enabler for modern, highly downsized turbocharged engines, this difference
will be overshadowed by the higher effectiveness for turbocharging and downsizing when they are combined
into packages.
                                            3-84

-------
                                    Technologies Considered in the Agencies' Analysis
       Almost every major manufacturer currently markets a vehicle with some form of
boosting. While boosting has been a common practice for increasing performance for several
decades, turbocharging has considerable potential to improve fuel economy and reduce COi
emissions when the engine displacement is also reduced.  Specific power levels for a boosted
engine often exceed 100 hp/L, compared to average naturally aspirated engine power densities
of roughly 70 hp/L.  As a result, engines can be downsized roughly 30 percent or higher while
maintaining similar peak output levels. In the last decade, improvements to turbocharger
turbine and compressor design have improved their reliability and performance across the
entire engine operating range.  New variable geometry turbines and ball-bearing center
cartridges allow faster turbocharger spool-up (virtually eliminating the once-common "turbo
lag") while maintaining high flow rates for increased boost at high engine speeds. Low speed
torque output has been dramatically improved for modern turbocharged engines. However,
even with turbocharger improvements, maximum engine torque at very low engine speed
conditions, for example launch from standstill, is increased less than at mid and high engine
speed conditions.  The potential to downsize engines may be less on vehicles with low
displacement to vehicle mass ratios for example a very small displacement engine in a vehicle
with significant curb weight, in order to provide adequate acceleration from standstill,
particularly up grades or at high altitudes.

       Use of GDI systems with turbocharged engines and air-to-air charge air cooling also
reduces the fuel octane requirements for knock limited combustion and allows the use of
higher compression ratios. Ford's "Ecoboost" downsized, turbocharged GDI engines
introduced on MY 2010 vehicles allow the replacement of V8 engines with V6 engines with
improved in 0-60 mph acceleration and with fuel economy improvements of up to 12
percent.37

       Recently published data with advanced spray-guided injection systems and more
aggressive engine downsizing targeted towards reduced fuel consumption and COi emissions
reductions indicate that the potential for reducing COi emissions for turbocharged, downsized
GDI engines may be as much as 15 to 30 percent relative to port-fuel-injected
engines.14'15'16'17'18 Confidential manufacturer data suggests an incremental range of fuel
consumption and COi emission reduction of 4.8 to 7.5 percent for turbocharging and
downsizing. Other publicly-available sources suggest a fuel consumption and CO2 emission
reduction of 8 to 13 percent compared to current-production naturally-aspirated engines
without friction reduction or other fuel economy technologies: a joint technical paper by
Bosch and Ricardo suggesting fuel economy gain of 8 to  10 percent for downsizing from a 5.7
liter port  injection V8 to a 3.6 liter V6 with direct injection using a wall-guided direct
injection  system;38 a Renault report suggesting a 11.9 percent NEDC fuel consumption gain
for downsizing from a 1.4 liter port injection in-line 4-cylinder engine to a 1.0 liter in-line 4-
cylinder engine, also with wall-guided direct injection;3  and a Robert Bosch paper suggesting
a 13 percent NEDC gain for downsizing to a turbocharged DI engine, again with wall-guided
injection.40 These reported fuel economy benefits show a wide range depending on the SGDI
technology employed.
                                            3-85

-------
                                     Technologies Considered in the Agencies' Analysis
       NHTSA and EPA reviewed estimates from the 2012-2016 final rale, the TAR, and
existing public literature. The previous estimate from the MYs 2012-2016 suggested a 12 to
14 percent effectiveness improvement, which included low friction lubricant (level one),
engine friction reduction (level one), DCP, DVVL and SGDI, over baseline fixed-valve
engines, similar to the estimate for Ford's Ecoboost engine, which is already in production.
Additionally, the agencies analyzed Ricardo vehicle simulation data for various turbocharged
engine packages.  Based on this data, and considering the widespread nature of the public
estimates, the effectiveness  of turbocharging and downsizing is highly dependent upon
implementation and degree  of downsizing.

       In alignment with these variances, for this proposal the agencies evaluated 4 different
levels of downsized and turbocharged high Brake Mean Effective Pressure (BMEP)X. engines;
18-bar, 24-bar, 24-bar with  cooled exhaust gas recirculation (EGR) and 21-bar with cooled
EGR   All engines are assumed to include gasoline direct injection (SGDI) and effectiveness
values include the benefits of this technology. In addition, the agencies believe to implement
in production a 27 bar boost level, it is necessary to incorporate cooled exhaust gas
recirculation (EGR) and also require a 2-stage turbocharger as well as engine changes to
increase robustness. The cooled EGR technology is discussed later in this section.

       NHTSA and EPA have revised the effectiveness to reflect this new information and
assume that turbocharging and downsizing, alone, will provide a  12 to 24.6 percent
effectiveness improvement  (dependent upon degree of downsizing and boost levels) over
naturally aspirated, fixed-valve engines. More specifically, 12.1 to 14.9 percent for 18-bar
engines, which is  equal to the boost levels evaluated in the MYs 2012-2016 final rule,
assuming 33 percent downsizing, 16.4 to 20.1 percent for 24-bar engines, assuming 50
percent downsizing, 19.3 to 23.0  percent for 24-bar engines with cooled EGR, assuming 50
percent downsizing and 20.6 to 24.6 percent for 27-bar engines with cooled EGR, assuming
56 percent downsizing. For comparison purposes an 18-bar engine with low friction lubricant
(level one), engine friction reduction (level one), DCP, DVVL and SGDI,  which is equivalent
to MYs 2012-2016 assumed turbocharging and downsizing technology, now results in a 16.8
to 20.9 percent effectiveness improvement.  Coupling turbocharging and downsizing with low
friction lubricant (level one and two), engine friction reductions (level one and two), DCP,
DVVL and SGDI, for the MYs 2017-2025 timeframe, yields 18.0 to 22.4 percent for 18-bar
engines 20.4 to 25.2 percent for 24-bar engines, 23.2 to 27.9 percent for 24-bar engine with
cooled EGR and 24.0 to 28.8 percent for 27-bar with cooled EGR over naturally aspirated,
fixed-valve engines.
x Brake Mean Effective Pressure is the average amount of pressure in pounds per square inch (psi) that must be
exerted on the piston to create the measured horsepower. This indicates how effective an engine is at filling the
combustion chamber with an air/fuel mixture, compressing it and achieving the most power from it. A higher
BMEP value contributes to higher overall efficiency.
                                            3-86

-------
                                    Technologies Considered in the Agencies' Analysis
       As noted above, the agencies relied on engine teardown analyses conducted by EPA,
FEV and Munro to develop costs for turbocharged GDI engines.41 In the 2012-2016 FRM,
the agencies estimated the DMC for turbocharging to 18 bar BMEP at $404 (2007$) and $681
(2007$) for 14 and V6/V8 engines, respectively, where the higher cost for the V-configuration
engines represents twin turbochargers versus the single turbocharger in the I-configuration
engine.  These DMCs become $417 (2009$) and $702 (2009$), respectively, for this analysis.
In the 2010 TAR, the agencies presented costs for 24 bar BMEP turbocharging at 1.5x the
cost of the 18 bar BMEP technology. This additional cost covered the incremental cost
increase of a variable geometry turbocharger (see 2010 TAR at page B-12). Thus, the DMC
for 24 bar BMEP would be $625 (2009$) and $1,053 (2009$) for I-configuration and V-
configuration engines,  respectively. Note also for this proposal, the agencies  are estimating
the DMC of the 27 bar BMEP technology at 2.5x the 18 bar BMEP technology, or $1,042
(2009$) and $1,756 (2009$) for I-configuration and V-configuration engines, respectively. All
of these turbocharger-related DMCs are considered applicable in the 2012MY.  The agencies
consider each turbocharger technology to be on the flat portion of the learning curve and have
applied a medium complexity ICM of 1.39 through 2018 for 18 bar and through 2024 for 24
and 27 bar, then 1.29 to each thereafter.  The resultant costs are shown in Table 3-29.

                         Table 3-29 Costs for Turbocharging (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
Technology
(BMEP)
18 bar
18 bar
24 bar
24 bar
27 bar
27 bar
18 bar
18 bar
24 bar
24 bar
27 bar
27 bar
18 bar
18 bar
24 bar
24 bar
27 bar
27 bar
Engine
type
I-engine
V-engine
I-engine
V-engine
I-engine
V-engine
I-engine
V-engine
I-engine
V-engine
I-engine
V-engine
I-engine
V-engine
I-engine
V-engine
I-engine
V-engine
2017
$361
$609
$542
$914
$904
$1,523
$159
$268
$238
$402
$397
$669
$520
$877
$780
$1,316
$1,301
$2,193
2018
$354
$597
$531
$896
$886
$1,493
$159
$267
$238
$401
$396
$668
$513
$864
$769
$1,296
$1,282
$2,161
2019
$347
$585
$521
$878
$868
$1,463
$119
$200
$237
$400
$396
$667
$466
$785
$758
$1,278
$1,263
$2,130
2020
$340
$573
$510
$860
$850
$1,434
$118
$199
$237
$399
$395
$665
$459
$773
$747
$1,259
$1,245
$2,099
2021
$333
$562
$500
$843
$833
$1,405
$118
$199
$236
$399
$394
$664
$451
$761
$736
$1,241
$1,227
$2,069
2022
$327
$551
$490
$826
$817
$1,377
$118
$199
$236
$398
$393
$663
$445
$749
$726
$1,224
$1,210
$2,040
2023
$320
$540
$480
$810
$800
$1,349
$118
$198
$236
$397
$393
$662
$438
$738
$716
$1,207
$1,193
$2,011
2024
$314
$529
$471
$793
$784
$1,322
$117
$198
$235
$396
$392
$661
$431
$727
$706
$1,190
$1,176
$1,983
2025
$308
$518
$461
$778
$769
$1,296
$117
$198
$176
$297
$293
$494
$425
$716
$637
$1,074
$1,062
$1,790
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are incremental to the baseline.

       The costs for the downsizing portion of the turbo/downsize technology is more
complex. The agencies have described those cost and how they were developed—based
primarily on FEV teardowns but some were scaled based on teardowns to generate costs for
downsizing situations that were not covered by teardowns—in both the 2012-2016 FRM and
the 2010 TAR. The DMCs used for this analysis are identical to those used in the 2010 TAR
except that they have been updated to 2009 dollars. Notable is the fact that many of the
downsizing costs are negative because they result in fewer parts and less material than the
engine from which they are "derived."  For example a V8 engine could be replaced by a
                                            3-87

-------
                                       Technologies Considered in the Agencies' Analysis
turbocharged V6 engine having two fewer cylinders and as many as eight fewer valves (in the
case of a V8 DOHC downsized to a V6 DOHC). Importantly, the agencies have used an
approach to calculating indirect costs that results in positive indirect costs regardless of
whether the DMC is positive or negative. This is done by calculating indirect costs based on
the absolute value of the DMC, then adding the indirect cost to the DMC to arrive at the total
cost.  This way, the agencies are never making a negative DMC "more negative" when
accounting for the indirect costs. This approach has been used in the 2012-2016 final rule and
the 2010 TAR.  Given the history of the downsizing costs used by the agencies, many are
considered applicable in the 2012MY and many in the 2017MY.y All are considered to be on
the flat portion of the learning curve. The agencies have applied a medium complexity ICM
of 1.39 through 2018 and 1.29 thereafter. The resultant costs are shown in Table 3-30.

                        Table 3-30 Costs for Engine Downsizing (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
Technology
14 DOHC to 13
14 DOHC to 14
V6 DOHC to 14
V6 SOHC 2V to 14
V6 OHV to 14
V8 DOHC to 14
V8 DOHC to V6
V8 SOHC 2V to 14
V8 SOHC 3V to 14
V8 SOHC 2V to
V6
V8 SOHC 3V to
V6
V8 OHV to 14
V8 OHV to V6
14 DOHC to 13
14 DOHC to 14
V6 DOHC to 14
V6 SOHC 2V to 14
V6 OHV to 14
V8 DOHC to 14
V8 DOHC to V6
V8 SOHC 2V to 14
V8 SOHC 3V to 14
V8 SOHC 2V to
V6
V8 SOHC 3V to
V6
V8 OHV to 14
V8 OHV to V6
14 DOHC to 13
2017
-$171
-$75
-$485
-$339
$276
-$839
-$243
-$645
-$718
-$74
-$138
-$237
$322
$75
$33
$213
$149
$107
$325
$107
$250
$278
$33
$60
$92
$125
-$96
2018
-$168
-$74
-$475
-$332
$268
-$814
-$238
-$625
-$696
-$73
-$135
-$230
$312
$75
$33
$213
$149
$106
$324
$106
$249
$277
$33
$60
$92
$124
-$93
2019
-$164
-$72
-$466
-$325
$260
-$789
-$233
-$607
-$675
-$71
-$132
-$223
$303
$56
$25
$159
$111
$79
$241
$80
$186
$207
$24
$45
$68
$93
-$108
2020
-$161
-$71
-$457
-$319
$252
-$766
-$228
-$588
-$655
-$70
-$130
-$217
$294
$56
$25
$159
$111
$79
$241
$79
$185
$206
$24
$45
$68
$92
-$105
2021
-$158
-$69
-$447
-$313
$244
-$743
-$224
-$571
-$635
-$68
-$127
-$210
$285
$56
$25
$158
$111
$79
$240
$79
$184
$205
$24
$45
$68
$92
-$102
2022
-$155
-$68
-$438
-$306
$237
-$720
-$219
-$554
-$616
-$67
-$124
-$204
$276
$56
$25
$158
$111
$79
$239
$79
$184
$205
$24
$45
$68
$92
-$99
2023
-$152
-$67
-$430
-$300
$232
-$706
-$215
-$543
-$604
-$66
-$122
-$200
$271
$56
$25
$158
$110
$79
$239
$79
$184
$204
$24
$45
$68
$92
-$96
2024
-$149
-$65
-$421
-$294
$227
-$692
-$211
-$532
-$592
-$64
-$119
-$196
$265
$56
$24
$158
$110
$78
$238
$79
$183
$204
$24
$45
$67
$91
-$93
2025
-$146
-$64
-$413
-$288
$223
-$678
-$207
-$521
-$580
-$63
-$117
-$192
$260
$56
$24
$157
$110
$78
$238
$79
$183
$204
$24
$45
$67
$91
-$90
y The engine downsize costs based on actual FEV teardowns were considered applicable to the 2012MY, as was
explained for some downsize costs in the 2012-2016 final rule and others in the 2010 TAR. For other downsize
costs—the two changes from OHV engines to DOHC engines—the agencies did not use FEV teardowns or
extrapolations from FEV teardowns, and instead used the methodology employed in the 2008 EPA Staff Report,
a methodology determined by both agencies to result in cost estimates more appropriate for the 2017MY. The
new downsize costs—those for V8 engines downsized to 14 engines—use a combination of V8 to V6 then V6 to
14 downsize costs and are considered applicable to the 2017MY within the context of this analysis.
                                              3-88

-------
                                         Technologies Considered in the Agencies' Analysis
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
14 DOHC to 14
V6 DOHC to 14
V6 SOHC 2V to 14
V6 OHV to 14
V8 DOHC to 14
V8 DOHC to V6
V8 SOHC 2V to 14
V8 SOHC 3V to 14
V8 SOHC 2V to
V6
V8 SOHC 3V to
V6
V8 OHV to 14
V8 OHV to V6
-$42
-$272
-$190
$383
-$514
-$136
-$395
-$440
-$42
-$77
-$145
$446
-$41
-$263
-$183
$374
-$490
-$131
-$377
-$419
-$40
-$75
-$139
$436
-$48
-$307
-$214
$339
-$548
-$154
-$421
-$469
-$47
-$87
-$155
$395
-$46
-$298
-$208
$331
-$525
-$149
-$403
-$449
-$46
-$84
-$148
$386
-$45
-$289
-$202
$323
-$503
-$145
-$386
-$430
-$44
-$82
-$142
$377
-$43
-$280
-$196
$316
-$481
-$140
-$370
-$412
-$43
-$80
-$136
$368
-$42
-$272
-$190
$311
-$467
-$136
-$359
-$400
-$42
-$77
-$132
$362
-$41
-$263
-$184
$306
-$453
-$132
-$348
-$388
-$40
-$75
-$128
$357
-$40
-$255
-$178
$301
-$440
-$128
-$338
-$376
-$39
-$72
-$124
$351
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are incremental to the baseline; all resultant engines are
DOHC.

         Note that the V8 to 14 engine downsize is new for this proposal.  This level of engine
  downsizing is considered for this analysis only if it also includes 27 bar BMEP turbo boost
  which, in addition, requires the addition of cooled EGR (discussed below).  As a result, any
  27 bar BMEP engine in this analysis will be 14 configuration and will include cooled EGR.

         With the information shown in Table 3-29 and Table 3-30, the costs for any
  turbo/downsize change can be determined.  These costs are shown in Table 3-31.

                         Table 3-31 Total Costs for Turbo/Downsizing (2009$)
Downsize
Technology
14 DOHC to 13
14 DOHC to 13
14 DOHC to 13
14 DOHC to 14
14 DOHC to 14
14 DOHC to 14
V6 DOHC to 14
V6 DOHC to 14
V6 DOHC to 14
V6 SOHC 2V to
14
V6 SOHC 2V to
14
V6 SOHC 2V to
14
V6 OHV to 14
V6 OHV to 14
V6 OHV to 14
V8 DOHC to 14
V8 DOHC to 14
V8 DOHC to 14
V8 DOHC to V6
V8 DOHC to V6
V8 DOHC to V6
V8 SOHC 2V to
14
V8 SOHC 2V to
Turbo
Technology
(BMEP)
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
2017
$424
$685
$1,205
$478
$738
$1,259
$248
$509
$1,029
$330
$591
$1,111
$903
$1,163
$1,683
$6
$266
$787
$741
$1,180
$2,057
$125
$385
2018
$420
$677
$1,189
$472
$728
$1,241
$250
$507
$1,019
$329
$586
$1,098
$887
$1,143
$1,656
$23
$279
$792
$733
$1,165
$2,029
$136
$393
2019
$357
$650
$1,155
$418
$710
$1,216
$159
$451
$957
$251
$544
$1,049
$805
$1,097
$1,602
-$82
$210
$716
$631
$1,124
$1,976
$45
$337
2020
$353
$642
$1,140
$412
$701
$1,199
$161
$449
$948
$250
$539
$1,037
$789
$1,078
$1,576
-$66
$222
$720
$624
$1,110
$1,950
$55
$344
2021
$350
$635
$1,126
$407
$692
$1,183
$163
$448
$939
$250
$535
$1,026
$775
$1,060
$1,551
-$51
$234
$725
$616
$1,097
$1,925
$65
$350
2022
$346
$627
$1,111
$401
$683
$1,167
$164
$446
$930
$249
$530
$1,014
$760
$1,042
$1,526
-$36
$245
$729
$609
$1,084
$1,900
$75
$356
2023
$342
$620
$1,097
$396
$674
$1,151
$166
$444
$921
$248
$526
$1,003
$749
$1,026
$1,504
-$29
$249
$726
$602
$1,071
$1,875
$79
$357
2024
$338
$613
$1,083
$390
$665
$1,135
$168
$442
$913
$247
$522
$992
$737
$1,012
$1,482
-$22
$252
$723
$595
$1,058
$1,851
$83
$357
2025
$335
$547
$972
$385
$598
$1,022
$170
$382
$807
$246
$459
$884
$726
$938
$1,363
-$15
$197
$622
$588
$946
$1,662
$87
$299
                                                3-89

-------
                                      Technologies Considered in the Agencies' Analysis
14
V8 SOHC 2V to
14
V8 SOHC 3V to
14
V8 SOHC 3V to
14
V8 SOHC 3V to
14
V8 SOHC 2V to
V6
V8 SOHC 2V to
V6
V8 SOHC 2V to
V6
V8 SOHC 3V to
V6
V8 SOHC 3V to
V6
V8 SOHC 3V to
V6
V8 OHV to 14
V8 OHV to 14
V8 OHV to 14
V8 OHV to V6
V8 OHV to V6
V8 OHV to V6

27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar
18 bar
24 bar
27 bar

$906
$81
$341
$861
$835
$1,274
$2,151
$800
$1,238
$2,116
$375
$635
$1,155
$1,323
$1,762
$2,639

$905
$94
$350
$863
$824
$1,256
$2,121
$790
$1,222
$2,086
$374
$631
$1,143
$1,301
$1,733
$2,597

$842
-$3
$289
$795
$738
$1,231
$2,083
$698
$1,191
$2,043
$311
$603
$1,108
$1,180
$1,673
$2,525

$842
$9
$298
$796
$727
$1,214
$2,053
$688
$1,175
$2,015
$310
$599
$1,097
$1,159
$1,646
$2,485

$841
$21
$306
$797
$717
$1,197
$2,025
$679
$1,160
$1,987
$309
$594
$1,085
$1,138
$1,618
$2,446

$840
$33
$314
$799
$707
$1,181
$1,997
$670
$1,144
$1,960
$309
$590
$1,074
$1,118
$1,592
$2,408

$834
$38
$316
$793
$697
$1,165
$1,969
$661
$1,130
$1,934
$306
$584
$1,061
$1,101
$1,569
$2,373

$828
$43
$318
$788
$687
$1,149
$1,943
$652
$1,115
$1,908
$303
$578
$1,048
$1,084
$1,547
$2,340

$724
$48
$261
$686
$677
$1,035
$1,751
$644
$1,002
$1,718
$300
$513
$938
$1,067
$1,426
$2,142
All costs are total costs (Direct manufacturing costs + Indirect costs); all costs are incremental to the
baseline; all resultant engines are DOHC; note that costs are shown for 27 bar BMEP engines with V6
engines. In fact, the agencies do not believe that manufacturers will employ 27 bar BMEP technology on V6
engines to comply with the proposed standards, instead using the additional boost to allow for downsizing
V6 engines to smaller 14 engines than would be used for 18 bar BMEP or 24 bar BMEP 14 engines and/or
downsizing V8 engines to 14 engines. As a result, whenever a 27 bar BMEP engine is chosen by either
agency's model, the engine configuration will be an 14 and will include cooled EGR, as discussed in section
3.4.1.8.
       3.4.1.9 Cooled Exhaust-Gas Recirculation (EGR)

       While not considered in the technology packages used for assessing potential
compliance pathways in the 2012-2016 light-duty rule, the agencies have considered an
emerging technology referred to as cooled exhaust gas recirculation (cooled-EGR) as applied
to downsized, turbocharged GDI engines. In the 2010 TAR, the agencies considered this
technology as an advanced gasoline technology since it was considered an  emerging and not
yet available technology in the light-duty gasoline market. While a cooled or "boosted" EGR
technology was discussed in the 2012-2016 light-duty rule record, the technology considered
here is comparatively more advanced as described in the 2010 TAR.  As such, the agencies
have considered new costs and new effectiveness values  for it.  The effectiveness values used
for vehicle packages with cooled EGR within this analysis reflect a conservative estimate of
system performance at approximately 24-bar BMEP. Vehicle simulation modeling of
technology packages using the more highly boosted and downsized cooled EGR engines (up
                                              3-90

-------
                                    Technologies Considered in the Agencies' Analysis
to 27-bar BMEP, and utilizing EGR rates of 20-25%) with dual-stage turbocharging has been
completed as part of EPA's contract with Ricardo Engineering as described in 3.3.1.2.  For
this NPRM, the agencies have updated the effectiveness of vehicle packages with cooled EGR
using the new Ricardo vehicle simulation modeling runs.

       Cooled exhaust gas recirculation or Boosted EGR is a combustion concept that
involves utilizing EGR as  a charge diluent for controlling combustion temperatures and
cooling the EGR prior to its introduction to the combustion system.  Higher exhaust gas
residual levels at part load conditions reduce pumping losses for increased fuel economy.  The
additional charge dilution  enabled by cooled EGR reduces the incidence of knocking
combustion and obviates the need for fuel enrichment at high engine power. This allows for
higher boost pressure and/or compression ratio and further reduction in engine displacement
and both pumping and friction losses while maintaining performance.  Engines of this type
use GDI and both dual cam phasing and discrete variable valve lift.  The EGR systems
considered in this proposal would use a dual-loop system with both high and low pressure
EGR loops and dual EGR  coolers. The engines would also use single-stage, variable
geometry turbocharging with higher intake boost pressure available across a broader range of
engine  operation than conventional turbocharged SI engines.  Such a system is estimated to be
capable of an additional 3  to 5 percent effectiveness relative to a turbocharged, downsized
GDI engine without cooled-EGR.42'43 The agencies have also considered  a more advanced
version of such a cooled EGR system that employs very high combustion  pressures by using
dual stage turbocharging.  This modeling work has been completed by Ricardo Engineering.
The simulation modeling is similar to work that Ricardo conducted for EPA for its 2008 staff
report on GHG effectiveness of light-duty vehicle technologies.44 The agencies have
considered this more advanced cooled EGR approach for this proposal.

       For the MYs 2012-2016 final rule and TAR, NHTSA  and EPA assumed a 5
percent fuel consumption effectiveness for cooled EGR compared to a conventional
downsized DI turbocharged engine.45 46 Based on the data from the Ricardo and Lotus
reports, NHTSA and EPA estimate the incremental reduction in fuel consumption for
EGR Boost to be 5 percent over a turbocharged and downsized DI engine. Thus, if
cooled  EGR is applied to 24-bar engine, adding the 19.3 percent from the  turbocharging
and downsizing to the 5 percent gain from cooled EGR results in total fuel consumption
reduction of 22.1 percent.  This is in agreement with the range suggested in the Lotus
and Ricardo reports.

       In the 2010 TAR, the agencies estimated the DMC of the cooled EGR system at $240
(2007$, see 2010 TAR at page B-12)).  This DMC becomes $242 (2009$) for this analysis.
This DMC is considered applicable in the 2012MY.  The agencies consider cooled EGR
technology to be on the flat portion of the learning curve and  have applied a medium
complexity ICM of 1.39 through  2024 then 1.29 thereafter. The resultant  costs are shown in
Table 3-32.
                         Table 3-32 Costs for Cooled EGR (2009$)
       fCost  | Engine type | 2017 | 2018 | 2019 | 2020  | 2021 |  2022 | 2023 |  2024 | 2025 ]


                                           3-91

-------
                                     Technologies Considered in the Agencies' Analysis
type
DMC
1C
TC

All
All
All

$210
$92
$303

$206
$92
$298

$202
$92
$294

$198
$92
$290

$194
$92
$285

$190
$91
$281

$186
$91
$277

$182
$91
$274

$179
$68
$247
         DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are
         incremental to the baseline.

       Note that, in the 2010 TAR, the agencies presented the cooled EGR system costs
inclusive of turbo charging costs (see 2010 TAR, Table B2.2-1 at page B-12). For this
analysis, the agencies are presenting the cooled EGR costs as a stand-alone technology that
can be added to any turbo/downsized engine provided sufficient boost is provided and
sufficient engine robustness is accounted for. As such, the cooled EGR system is considered
applicable only the 24 bar BMEP and 27 bar BMEP engines. Further, the agencies believe
that 24 bar BMEP engines are capable of maintaining NOx control without cooled EGR, so
each agency's respective models may choose 24 bar BMEP engines with and/or without
cooled EGR. However, as noted above, 27 bar BMEP engines are considered to require
cooled EGR to maintain NOx emission control. As such, neither agency's model is allowed
to choose 27 bar BMEP technology without also adding cooled EGR.
       3.4.1.10
Diesel Engine Technology (DSL)
       Diesel engines have several characteristics that give them superior fuel efficiency
compared to conventional gasoline, spark-ignited engines.  Pumping losses are much lower
due to lack of (or greatly reduced) throttling in a diesel engine.  The diesel combustion cycle
operates at a higher compression ratio than does a gasoline engine. As a result, turbocharged
light-duty diesels typically achieve much higher torque levels at lower engine speeds than
equivalent-displacement naturally-aspirated gasoline engines.  Future  high BMEP
turbocharged and downsized engines, mentioned above, are projected to improve torque
levels at lower engine speeds thus reducing the diesel advantage in this area. Diesels also
operate with a very lean air/fuel mixture. These attributes - reduced pumping losses, higher
compression ratio and lean/air fuel mixture - allow the engine to extract more energy from a
given mass of fuel than a gasoline engine, and thus make it more efficient.  Additionally,
diesel fuel has higher energy content per gallon than does gasoline.  While diesel fuel has a
higher energy content than gasoline, it also contains more carbon per gallon than does
gasoline: diesel produces 22.2 pounds of CC>2 per gallon when burned, while gasoline
produces 19.4 pounds of CO2 per gallon. This higher carbon content slightly offsets the GHG
emissions benefit of diesel fuel relative to gasoline, however, the disbenefit is more than
compensated by the greater efficiency of the diesel engine. Since diesel engines are more fuel
efficient than current naturally aspirated PFI gasoline engines, the agencies anticipate that
manufacturers will evaluate and potentially invest in diesel engine production as a way to
comply with more  stringent CAFE standards. However, there are two primary reasons why
manufacturers might not choose to invest significantly in diesel engine technologies as a way
to comply with the CAFE and GHG standards for MYs 2017-2025.

       As discussed above, even  though diesel has higher energy content than gasoline it also
has a higher carbon density that results in higher amounts of CCh emitted per gallon,
                                            3-92

-------
                                    Technologies Considered in the Agencies' Analysis
approximately 15 percent more than a gallon of gasoline. This is commonly referred to as the
"carbon penalty" associated with using diesel fuel - a diesel vehicle yields greater fuel
economy improvements compared to its CO2 emissions reduction improvements, so a
manufacturer that invests in diesel technology to meet CAFE standards may have more
trouble meeting the GHG standards than if it used a different and more cost effective (from a
GHG perspective) technology.

       And second, diesel engines also have emissions characteristics that present challenges
to meeting federal Tier 2 NOx emissions standards. By way of comparison for readers
familiar with the European on-road fleet, which contains many more diesel vehicles than the
U.S. on-road fleet, U.S. Tier 2 emissions fleet average requirement of bin 5 require roughly
45 to 65 percent more NOX reduction compared to the Euro VI standards.

       Despite considerable advances by manufacturers in developing Tier 2-compliant diesel
engines, it remains somewhat of a systems-engineering challenge to maintain the full fuel
consumption advantage of the diesel engine while meeting Tier 2 emissions regulations
because some of the emissions reduction strategies can increase fuel consumption (relative to
a Tier 1 compliant diesel engine), depending on the combination of strategies employed. A
combination of combustion improvements  (that reduce NOx emissions leaving the engine)
and aftertreatment (capturing and reducing NOx emissions via a NOx adsorption catalyst, or
via selective catalytic reduction (SCR) using a reductant such as urea) that have left the
engine before they leave the vehicle tailpipe) are being introduced on Tier 2 compliant light-
duty diesel vehicles today.  However, recently there have been a small number of
announcements that diesel engines will be added to some passenger cars, in some cases a
segment first for a manufacturer47, or that new passenger car diesel engines are being
designed to meet all global emissions regulations.48 This suggests to the agencies that some
manufacturers may be planning to use diesel engines in their plans to meet the tighter CAFE
standards in the mid-term, which may be enabled by advances in diesel engine and emission
control technology. Manufacturers that focus on diesel engines have also stated to the
agencies their expectation that diesel engines will continue to  be a viable technology for
improving fuel economy and GHG emissions in the future.

       We spend time here discussing available emissions reduction technologies for diesel
engines as part of this rulemaking because  of the potential they have to impact fuel economy
and GHG emissions for the vehicles that have them.  With respect to combustion
improvements, we note that several key advances in diesel engine combustion technology
have made it possible to reduce emissions coming from the engine prior to aftertreatment,
which reduces the need for aftertreatment.  These technologies include improved fuel systems
(higher injection pressure and multiple-injection capability), advanced controls and sensors to
optimize combustion and emissions performance, higher EGR levels and EGR cooling to
reduce NOX, and advanced  turbocharging systems.  These systems are available today and
they do not adversely impact fuel efficiency. However, additional improvements in these
technologies will be needed to reduce engine emissions further, should future emissions
standards become more stringent. Further development may also be needed to reduce the fuel
efficiency penalty associated with EGR.

                                           3-93

-------
                                        Technologies Considered in the Agencies' Analysis
       With respect to catalytic exhaust emission control systems, typical 3-way exhaust
catalysts without NOx storage capability are not able to reduce NOx emissions from engines
operated lean of stoichiometry (diesel or lean-burn gasoline)=.  To reduce NOx,
hydrocarbons, and particulate emissions, all diesels will require a catalyzed diesel particulate
filter (CDPF) and sometimes a separate diesel oxidation catalyst (DOC), and either a lean
NOx trap (LNT)z or the use of a selective catalytic reduction system, typically base-metal
zeolite urea-SCRaa.

       The increased cost of diesel emissions control technologies relative to powertrains
with stoichiometric gasoline engines that are approaching comparable efficiency may also
make diesels less attractive to manufacturers as a technology solution for more stringent
CAFE and GHG standards.  However, recognizing that some manufacturers may still employ
diesel technology to meet the future standards, the agencies have included diesels in our
analysis as follows:

       The agencies sought to ensure that diesel engines would have equivalent performance
to comparable gasoline engine vehicles.  For the Subcompact, Compact, and Midsize
Passenger Car, Performance Subcompact Car, and Small Light Truck vehicle subclasses, the
agencies assumed that an 14 gasoline  base engine would be replaced by an in-line 4-cylinder
diesel engine with displacement varying around 2.0 liters. For the Performance Compact,
Performance Midsize, Large Passenger Car, Minivan, and Midsize Truck vehicle subclasses
for the CAFE model, the agencies assumed that a V6 gasoline base engine would be replaced
by an in-line 4-cylinder diesel engine with displacement varying around 2.8 liters.   For the
Large Truck and Performance Large Car vehicle subclasses for the CAFE model, the agencies
assumed that a V8 gasoline base engine  would be replaced with a V6 diesel engine with
z A lean NOX trap operates= by oxidizing NO to NO2 in the exhaust and storing NO2 on alkali sorbent material,
most often BaO.  When the control system determines (via mathematical model and typicalla NOX sensor) that
the trap is saturated with NOX, it switches the engine into a operating mode just rich of stoichiometry that allow
NOx to be released from the alkali storage and temporarily allow three-way function of the catalyst similar to
three-way catalysts used in stoichiometric gasoline applications. LNTs preferentially store sulfate compounds
from the fuel, whichreduces NOx storage capacity over time, thus the system must undergo periodic
desulfurization by operating at a net-fuel-rich condition at high temperatures in order to retain NOX trapping
efficiency.
aa An SCR aftertreatment system uses a reductant (typically, ammonia derived from urea) that is injected into the
exhaust stream ahead of the SCR catalyst. Ammonia is a strong reductant even under net lean conditions. It
combines with NOX in the SCR catalyst to form N2 and water. The hardware configuration for an SCR system is
sometimes more complicated than that of an LNT, due to the onboard urea storage and delivery system (which
requires a urea pump and injector to inject urea into the exhaust stream), which generally makes an SCR system
cost more than an LNT system.  While a rich engine-operating mode is not required for NOX reduction, the urea
is typically injected at a rate of approximately 3 percent of the fuel consumed. The agencies understand that
manufacturers designing SCR systems intend to align urea tank refills with standard maintenance practices such
as oil changes as more diesel vehicles are introduced into the market.  For diesel vehicles currently on the
market, this is generally already the practice, and represents an ongoing maintenance cost for vehicles with this
technology.
                                               3-94

-------
                                     Technologies Considered in the Agencies' Analysis
displacement varying around 4.0 liters to meet vehicle performance requirements. It was also
assumed that diesel engines for all of these classes would utilize SCR aftertreatment systems
given recent improvements in zeolite-based SCR systems and system efficiency. These
assumptions impacted our estimates of the costs of implementing diesel engines as compared
to the base gasoline engines.

       Diesel engines are more costly than port-injected spark-ignition gasoline engines.
These higher costs result from more costly components, more complex systems for emissions
control, and other factors.  The vehicle systems that are impacted include:

       •  Fuel systems (higher pressures and more responsive injectors);

       •  Controls and sensors to optimize combustion and emissions performance;

       •  Engine design (higher cylinder pressures require a more robust engine, but higher
          torque output means diesel engines can have reduced displacement);

       •  Turbocharger(s);

       •  Aftertreatment  systems, which tend to  be more costly for diesels;

       In the MYs 2012-2016 final rule, the agencies estimated the DMC for converting a
gasoline PFI engine with 3-way catalyst aftertreatment to a diesel engine with diesel
aftertreatment at $1,697 (2007$), $2,399 (2007$), $1,956 (2007$) and $2,676 (2007$) for a
small car, large car, medium/large MPV & small truck, and large truck, respectively (see final
Joint TSD, Table 3-12 at page 3-44).  All of these costs were for SCR-based diesel systems,
with the exception of the small car, which was a LNT-based system. For this proposal, we are
using the same methodology as used in the MYs 2012-2016 final rule, but have made four
primary changes to the cost estimates. First, the agencies have not estimated costs for a LNT-
based system, and instead have estimated costs for all vehicle types assuming they will
employ SCR-based systems.  Second, the agencies assumed that manufacturers would meet a
Tier 2 bin 2  average rather than a Tier 2 bin 5 average, assuming that more stringent levels of
compliance will be required in the future. In order to estimate costs for Tier 2 bin 2 compliant
vehicles, catalyst volume costs were estimated based on an assumed increase in volume of 20
percent. This was the estimated necessary increase needed to meet Tier 2, bin 2 emission
level of 0.02 grams of NOx per mile.  Increased catalyst volume resulted in a higher cost
estimate for diesel aftertreatment than was estimated for the MYs 2012-2016 final rule.  The
third is to update all platinum group metal costs from the March 2009 values used in the
2012-2016 final rule to February 2011 values.1* The February 2011 values were used for
bb As reported by Johnson-Matthey, the March 2009 monthly average costs were $1,085 per Troy ounce and
$1,169 per Troy ounce for platinum (Pt) and rhodium (Rh), respectively. As also reported by Johnson-Matthey,
the February 2011 monthly average costs were $1,829 per Troy ounce and $2,476 per Troy ounce for Pt and Rh,
respectively. See www.platinum.matthey.com.

                                             3-95

-------
                                        Technologies Considered in the Agencies' Analysis
purposes of this NPRM analysis because they represented the most recent monthly average
prices available at the time the agencies "locked-down" all cost estimates for the purposes of
moving into the modeling phase of analysis.cc  The forth is to include an additional $50 DMC
for all costs to cover costs associated with improvements to fuel and urea controls. All of the
diesel costs are considered applicable to MY 2012.  The agencies consider diesel technology
to be on the flat portion of the learning curve and have applied a medium complexity ICM of
1.39 through 2018, and then an ICM of 1.29 thereafter. The resultant costs are shown in Table
3-33.
                    Table 3-33 Costs for Conversion to Advanced Diesel (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
Vehicle class
Subcompact/Small car
Large car
Minivan
Small truck
Large truck
Subcompact/Small car
Large car
Minivan
Small truck
Large truck
Subcompact/Small car
Large car
Minivan
Small truck
Large truck
2017
$2,039
$2,498
$2,044
$2,061
$2,858
$896
$1,098
$898
$906
$1,256
$2,936
$3,596
$2,942
$2,967
$4,114
2018
$1,999
$2,448
$2,003
$2,020
$2,800
$895
$1,096
$897
$904
$1,253
$2,893
$3,544
$2,900
$2,924
$4,053
2019
$1,959
$2,399
$1,963
$1,980
$2,744
$669
$819
$670
$676
$937
$2,627
$3,218
$2,633
$2,656
$3,681
2020
$1,919
$2,351
$1,924
$1,940
$2,690
$668
$818
$669
$675
$935
$2,587
$3,169
$2,593
$2,615
$3,625
2021
$1,881
$2,304
$1,885
$1,901
$2,636
$666
$816
$668
$674
$934
$2,547
$3,120
$2,553
$2,575
$3,570
2022
$1,843
$2,258
$1,848
$1,863
$2,583
$665
$815
$667
$672
$932
$2,509
$3,073
$2,515
$2,535
$3,515
2023
$1,807
$2,213
$1,811
$1,826
$2,531
$664
$813
$666
$671
$931
$2,471
$3,026
$2,477
$2,497
$3,462
2024
$1,770
$2,168
$1,774
$1,790
$2,481
$663
$812
$664
$670
$929
$2,433
$2,980
$2,438
$2,460
$3,410
2025
$1,735
$2,125
$1,739
$1,754
$2,431
$662
$811
$663
$669
$927
$2,397
$2,936
$2,402
$2,423
$3,358
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are incremental to the baseline.

        For the MYs 2012-016 final rule and TAR, NHTSA and EPA estimated the fuel
consumption reduction of a SCR-based diesel engine to be between 20 to 25 percent over a
baseline gasoline engine.  NHTSA  and EPA have revisited these values and have now
estimated, based on the Ricardo 2011 study, the effectiveness of a SCR-based diesel engine to
be 28.4 to 30.5 percent. For purposes of COi reduction, EPA estimates a 7  to 20 percent for
light-duty diesels equipped with SCR.
cc Note that there is no good way of determining what PGM prices to use when conducting cost analyses.  Spot
prices are inherently dangerous to use because spot prices, like stock prices on the stock market, can vary
considerably from day to day. One could argue that an average price is best, but average prices can vary
considerably depending on the length of time included in the average.  And if too much time is included in the
average, then average prices from a time prior to PGM use in diesel engines may be included which would lead
some to conclude that we had cherry picked our values. Given no good option, it seems most transparent and
least self serving to simply choose a price and report its basis. In the end, the PGM costs represent 16-23 percent
of the diesel DMC in this analysis. Further, diesels play very little to no role in enabling compliance with the
proposed standards.
                                               3-96

-------
                                    Technologies Considered in the Agencies' Analysis
3.4.2   Transmission Technologies

       NHTSA and EPA have also reviewed the transmission technology estimates used in
the 2012-2016 final rale and the 2010 TAR. In doing so, NHTSA and EPA considered or
reconsidered all available sources and updated the estimates as appropriate. The section
below describes each of the transmission technologies considered for this rulemaking.

       3.4.2.1 Improved Automatic Transmission Control (Aggressive Shift Logic and
       Early Torque Converter Lockup)

       Calibrating the transmission shift schedule to upshift earlier and quicker, and to lock-
up or partially lock-up the torque converter under a broader range of operating conditions can
reduce fuel consumption and COi emissions. However, this operation can result in a
perceptible degradation in noise, vibration, and harshness (NVH).  The degree to which NVH
can be degraded before it becomes noticeable to the driver is strongly influenced by
characteristics of the vehicle, and  although it is somewhat subjective, it always places a limit
on how much fuel consumption can be improved by transmission control changes.
Aggressive Shift Logic and Early  Torque Converter Lockup are best optimized
simultaneously when added to an  automatic transmission due to the fact that adding both of
them requires only minor modifications to the transmission mechanical components or
calibration software. As a result, these two technologies are combined in the modeling when
added to an automatic transmission. Since a dual clutch transmission (DCT) has no torque
converter, the early torque converter lockup technology is not included when adding ASL to
the DCT.

       3.4.2.2 Aggressive  Shift Logic

       During operation, a transmission's controller manages the operation of the
transmission by scheduling the upshift or downshift, and, in automatic transmissions, locking
or allowing the torque converter to slip based on a preprogrammed shift schedule. The shift
schedule contains a number of lookup table functions, which define the shift points and torque
converter lockup based on vehicle speed and throttle position, and other parameters such as
temperature. Aggressive shift logic (ASL) can be employed in such a way as  to maximize
fuel efficiency by modifying the shift schedule to upshift earlier and inhibit downshifts under
some conditions, which reduces engine pumping losses and engine friction. The application
of this  technology does require a manufacturer to confirm that drivability, durability, and
NVH are not significantly degraded.

       For this proposal, the agencies are considering two levels of ASL. The first level is
that discussed in the 2012-2016 final rale and the 2010 TAR. ASL-level  1 is an early upshift
strategy whereby the transmission shifts to the next higher gear "earlier" (or at lower RPM
during a gradual acceleration) than would occur in a traditional automatic transmission. This
early upshift reduces fuel consumption by allowing the engine to operate  at a lower RPM and
higher load, which typically moves the engine into a more efficient operating region.
                                           3-97

-------
                                     Technologies Considered in the Agencies' Analysis
       ASL-level 2 is a shift optimization strategy whereby the engine and/or transmission
controller(s) continuously evaluate all possible gear options that would provide the necessary
tractive power (while limiting the adverse effects on driveline NVH) and select the gear that
lets the engine run in the most efficient operating zone. Ricardo acknowledged in its report
that the ASL-level 2 ("shift optimization") strategy currently causes significant implications
for drivability and hence affects consumer acceptability. However, Ricardo recommended the
inclusion of this technology for the 2020-2025 time frame with the assumption that
manufacturers will develop a means of yielding the fuel economy benefit without adversely
affecting driver acceptability.  The agencies believe these drivability challenges could include
shift business - that is, a high level of shifting compared to current vehicles as perceived by
the customers. The agencies note that in confidential discussions with two major transmission
suppliers, the suppliers described transmission advances which reduce shifting time and
provide smoother torque transitions than today's designs, making the shifting event less
apparent to the driver, however these improvements will not influence the customer's
perception of shift business related to the changes in engine speed.

       In addition, the agencies note that several auto companies and transmission firms have
announced future introduction of transmissions into the U.S. market with even a higher
number of gears than were included in the Ricardo simulation and in the agencies' feasibility
assessment for this proposal (which is 8 forward speeds). These announcements include both
9 and 10 speed transmissions which may present further challenges with shift business, given
the availability of one or two additional gears. At the same time, the associated closer gear
spacing will generally result in smaller engine speed changes during  shifting that may be less
noticeable to the driver.

       The agencies are including shift optimization in the analysis under the premise that
manufacturers and suppliers are developing means to mitigate these drivability issues by MY
2017, as assumed in the 2011 Ricardo study (more information on Ricardo's treatment of the
optimized shift strategy is described in Section 6.4 of the 2011 Ricardo report).  If
manufacturers are not able to solve these drivability issues, the assumed effectiveness could
be lower and the cost could be higher or both. The agencies are seeking comment on  the
feasibility of ASL-level 2 and the likelihood that manufacturers will be able to overcome the
drivability issues.

       In MYs 2012-2016 final rule, the agencies estimated an effectiveness improvement of
1 to 2 percent for aggressive shift logic which was supported by the 2002 NAS and
NESCCAF reports as well as confidential manufacturer data.  The agencies updated the
effectiveness of ASL-level 1 ranging from 1.9 to 2.7 based on 2010 Ricardo study. In CAFE
model an incremental effectiveness ranging for both ASL and early torque converter  lockup
ranging from 2.3 to 3.1 percent is applied (Early torque converter has effectiveness of 0.5
percent).

       ASL-level 2 is new to this analysis which is based on the shift optimization algorithm
in 2011 Ricardo study. The effectiveness for ASL-level 2 ranges from 5.1 to 7.0 percent
improvement over transmission with unimproved  shift logic or roughly 4 to 5 percent over a

                                            3-98

-------
                                     Technologies Considered in the Agencies' Analysis
transmission that already incorporates aggressive shift logic. In the CAFE model, an
incremental effectiveness ranging from 3.27 to 4.31 percent is applied.

       In the 2012-2016 rale, the agencies estimated the DMC at $26 (2007$) which was
considered applicable to the 2015MY.  This DMC becomes $27 (2009$) for this analysis.
The agencies consider ASL-level 1 technology to be on the flat portion of the learning curve
and have applied a medium complexity ICM of 1.39 through 2018 then  1.29 thereafter. For
ASL-level 2, the agencies are estimating the DMC  at an equivalent $27  (2009$) except that
this cost is considered applicable to the 2017MY.  Essentially this yields a nearly negligible
incremental cost for ASL-level 2 over ASL-level 1. The agencies consider ASL-level 2
technology to be on the flat portion of the learning  curve and have applied a medium
complexity ICM of 1.39 through 2024 then 1.29 thereafter. The timing  of the ASL-level 2
ICMs is different than that for the level 1 technology because the level 2 technology is newer
and not yet being implemented in the fleet. The resultant costs are shown in Table 3-34.
Note that both levels of ASL technology are incremental to the baseline system, so ASL-level
2 is not incremental to ASL-level 1.

                 Table 3-34 Costs for Aggressive Shift Logic Levels 1 & 2 (2009$)
Cost type
DMC
DMC
1C
1C
TC
TC
Technology
ASL-level 1
ASL-level 2
ASL-level 1
ASL-level 2
ASL-level 1
ASL-level 2
Transmission
type
All
All
All
All
All
All
2017
$26
$27
$6
$7
$32
$33
2018
$25
$26
$6
$7
$32
$33
2019
$25
$25
$5
$6
$30
$32
2020
$24
$25
$5
$6
$29
$31
2021
$24
$24
$5
$6
$29
$30
2022
$23
$23
$5
$6
$28
$30
2023
$23
$23
$5
$6
$28
$29
2024
$22
$22
$5
$6
$27
$29
2025
$22
$22
$5
$5
$27
$27
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are incremental to the baseline.
       3.4.2.3 Early Torque Converter Lockup

       A torque converter is a fluid coupling located between the engine and transmission in
vehicles with automatic transmissions and continuously-variable transmissions (CVT). This
fluid coupling allows for slip so the engine can run while the vehicle is idling in gear (as at a
stop light), provides for smoothness of the powertrain, and also provides for torque
multiplication during acceleration, and especially launch. During light acceleration and
cruising, the inherent slip in a torque converter causes increased fuel consumption, so modern
automatic transmissions utilize a clutch in the torque converter to lock it and prevent this
slippage. Fuel consumption can be further reduced by locking up the torque converter at
lower vehicle speeds, provided there is sufficient power to propel the vehicle, and noise and
                                            3-99

-------
                                      Technologies Considered in the Agencies' Analysis
vibration are not excessive.dd  If the torque converter cannot be fully locked up for maximum
efficiency, a partial lockup strategy can be employed to reduce slippage. Early torque
converter lockup is applicable to all vehicle types with automatic transmissions. Some torque
converters  will require upgraded clutch materials to withstand additional loading and the
slipping conditions during partial lock-up. As with aggressive shift logic, confirmation of
acceptable drivability, performance, durability and NVH characteristics is required to
successfully implement this technology.

       Regarding the effectiveness of Early Torque Converter Lockup, the 2012-2016 final
rule, TAR, and the 2010 Ricardo study estimated an effectiveness improvement of 0.4 to 0.5
percent.

       In the 2012-2016 rule, the agencies estimated the DMC at $24 (2007$) which was
considered applicable to the 2015MY. This DMC remains $24 (2009$) for this analysis.66
The agencies consider early torque converter lockup technology to be on the flat portion of
the learning curve and have applied a low complexity ICM of 1.24 through 2018 then 1.19
thereafter.  The resultant costs are shown in Table 3-35.

                   Table 3-35 Costs for Early Torque Converter Lockup (2009$)
Cost type
DMC
1C
TC
Transmission
type
Automatic
Automatic
Automatic
2017
$23
$6
$29
2018
$23
$6
$29
2019
$22
$5
$27
2020
$22
$5
$27
2021
$21
$5
$26
2022
$21
$5
$26
2023
$21
$5
$25
2024
$20
$5
$25
2025
$20
$5
$24
      DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are incremental to
      the baseline.

       3.4.2.4 High Efficiency Gearbox

       For this rule, a high efficiency gearbox refers to some or all of a suite of incremental
gearbox improvement technologies that should be available within the 2017 to 2025
timeframe. The majority of these improvements address mechanical friction within the
gearbox. These improvements include but are not limited to: shifting clutch technology
improvements (especially for smaller vehicle classes), improved kinematic design, dry sump
lubrication systems, more efficient seals, bearings and clutches (reducing drag), component
superfinishing and improved transmission lubricants. More detailed description can be found
in the 2011 Ricardo report49. Note that the high efficiency gearbox technology is applicable
to any type of transmission.
dd Although only modifications to the transmission calibration software are considered as part of this technology,
very aggressive early torque converter lock up may require an adjustment to damper stiffness and hysteresis
inside the torque converter.
ee As is true throughout this presentation of cost estimates, the agencies round costs to the nearest dollar. In the
actual model input files, the cost in 2007$ would have been $23.68 and the cost in 2009$ is $24.42. So an
impact of the dollar-year conversion is reflected in the analysis even when it does not appear so in this
presentation.
                                             3-100

-------
                                     Technologies Considered in the Agencies' Analysis
       EPA analyzed detailed transmission efficiency input data provided by Ricardo and
implemented it directly into the lumped parameter model. Based on the LP effectiveness
resulting from these inputs, EPA and NHTSA estimate that a high efficiency gearbox can
provide a GHG or fuel consumption reduction in the range of 3.8 to 5.7 percent (3.8% for
4WD trucks with an unimproved rear axle) over a baseline automatic transmission in
MY2017 and beyond.

       The agencies estimate the DMC of the high efficiency gearbox at $200 (2009$).  We
have based this on the DMC for engine friction reduction in a V8 engine which, as presented
in Table 3-22 is $193 (2009$). We have rounded this up to $200 for this analysis. This DMC
is considered applicable for the 2017MY.  The agencies consider high efficiency gearbox
technology to be on the flat portion of the learning curve and have applied a low complexity
ICM of 1.24 through 2024 then 1.19 thereafter.  The resultant costs are shown in Table 3-36.

                     Table 3-36 Costs for High Efficiency Gearbox (2009$)
Cost type
DMC
1C
TC
Transmission
type
Automatic/Dual
clutch
Automatic/Dual
clutch
Automatic/Dual
clutch
2017
$200
$48
$248
2018
$194
$48
$242
2019
$188
$48
$236
2020
$183
$48
$231
2021
$177
$48
$225
2022
$172
$48
$220
2023
$168
$48
$216
2024
$165
$48
$213
2025
$162
$38
$200
     DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; all costs are incremental to the
     baseline.
       3.4.2.5 Automatic 6-, 7- and 8-Speed Transmissions (NAUTO and 8SPD)

       Manufacturers can also choose to replace 4- and 5-speed transmission with 6-, 7-, or
8-speed automatic transmissions. Additional ratios allow for further optimization of engine
operation over a wider range of conditions, but this is subject to diminishing returns as the
number of speeds increases. As additional planetary gear sets are added (which may be
necessary in some cases to achieve the higher number of ratios), additional weight and friction
are introduced.  Also, the additional shifting of such a transmission can be perceived as
bothersome to some consumers, so manufacturers need to develop strategies for smooth
shifts.  Some manufacturers are replacing 4- and 5-speed automatics with 6-speed automatics,
and 7- and 8-speed automatics have also entered production.  While a six speed transmission
application was most prevalent for the 2012-2016 final rule, eight speed transmissions are
expected to be readily available and applied in the 2017 through 2025 timeframe.

       As discussed in the MY 2011 CAFE final rule, confidential manufacturer data
projected that 6-speed transmissions could incrementally reduce fuel consumption by  0 to 5
percent from a baseline 4-speed automatic transmission, while an 8-speed transmission could
incrementally reduce fuel consumption by up to 6 percent from  a baseline 4-speed automatic
transmission. GM has publicly claimed a fuel economy improvement of up to 4 percent for
                                           3-101

-------
                                    Technologies Considered in the Agencies' Analysis
its new 6-speed automatic transmissions.50 The 2008 EPA Staff Technical Report found a 4.5
to 6.5 percent fuel consumption improvement for a 6-speed over a 4-speed automatic
transmission.51 Based on this information, NHTSA estimated in the MY 2011 rule, that the
conversion to a 6-,7- and 8-speed transmission (NAUTO) from a 4 or 5-speed automatic
transmission with IATC would have an incremental fuel consumption benefit of 1.4 percent to
3.4 percent, for all vehicle classes.  From a baseline 4 or 5 speed transmission without IATC,
the incremental fuel consumption benefit would be approximately 3 to 6 percent, which is
consistent with the EPA Staff Report estimate. In MYs 2012-2016 final rule, NHTSA and
EPA reviewed these effectiveness estimates and concluded that they remain accurate. While
the CAFE model follows the incremental approach discussed above, the GHG model
estimates the packaged effectiveness of 4.5 to 6.5 percent

       In this NPRM analysis, the agencies divided the improvement for this technology into
two steps, first from 4 or 5 speed transmission to 6  or 7  speed transmission (NAUTO), then
from 6 or 7 speed transmission to 8 speed transmission (8SPD). The effectiveness estimates
for NAUTO and 8SPD are based on 2011 Ricardo study. In this NPRM analysis, the
effectiveness for a 6-speed transmission relative to  a 4-speed base transmission ranges from
3.1 to 3.9 percent (2.1 percent for large truck with unimproved rear axle) including 7 percent
of transmission gearbox efficiency improvement that the agencies assumed accompanying the
new 6 speed transmission after MY 2010. NHTSA  incorporated this effectiveness estimate
into the CAFE model as incremental improvement  over IATC ranging from 1.89 to 2.13
percent. In this NPRM analysis, the agencies assumed that 8-speed transmission will not start
to phase in until MY2017. NHTSA applied 8-speed automatic transmission succeeding 6-
speed automatic transmission to vehicles with towing requirement, such as Minivan, Midsize
light truck and large light truck. All other vehicle subclasses use 8-speed DCT to succeed 6-
speed DCT. The effectiveness for an 8-speed DCT  relative to a 4-speed DCT transmission
ranges from 11.1 to 13.1 percent for subcompact car, small car and small light truck. The
effectiveness for an 8-speed automatic transmission relative to 4-speed automatic
transmission ranges for large CUV and large truck ranges from 8.7 to 9.2 percent in Lump
parameter model. This translates into effectiveness  in the range of 3.85 to 4.57 percent for an
8-speed DCT relative to a 6-speed DCT and 4.9 to 5.34 percent for 8-speed automatic
transmission relative to 6-speed automatic transmission in CAFE model.

       In the 2010 TAR, the agencies estimated the DMC at -$13 (2008$) for a 6 speed
automatic transmission relative to a 4 speed auto transmission, applicable in the 2017MY (see
2010 TAR, Table B2.1-1 at page B-10). For the 2012MY, that DMC was -$15  (2008$),
although that value was not presented in the TAR.  The latter DMC remains -$15 (2009$) for
this analysis which is considered to be applicable in the 2012MY. The agencies consider 6
speed automatic transmission technology to be on the flat portion of the learning curve and
have applied a low complexity ICM of 1.24 through 2018 then 1.19 thereafter.  The resultant
costs are shown in Table 3-37.

       New for this analysis is the cost of an 8 speed automatic transmission. For the cost of
this technology, the agencies have relied on a tear-down study completed by FEV since
publication of the TAR.52 In that study, the 8 speed auto transmission was found to be $62

                                          3-102

-------
                                     Technologies Considered in the Agencies' Analysis
(2007$) more costly than the 6 speed auto transmission. This DMC becomes $64 (2009$) for
this analysis. Adding the $64 (2009$) to the -$15 (2009$) DMC for a 6 speed relative to a 4
speed, the 8 speed auto transmission relative to a 4 speed auto transmission would be $49
(2009$). The agencies consider this DMC to be applicable to the 2012MY.  The agencies
consider the 8 speed auto transmission technology to be on the flat portion of the learning
curve and have applied a medium complexity ICM of 1.39 through the 2018MY then 1.29
thereafter.ff The resultant costs for both 6 speed and 8 speed auto transmissions are shown in
Table 3-37.
               Table 3-37 Costs for 6 and 8 Speed Automatic Transmissions (2009$)
Cost
type
DMC
DMC
DMC
1C
1C
1C
TC
TC
TC
Transmission type
6spAT from 4spAT
SspAT from 6spAT
SspAT from 4spAT
6spAT from 4spAT
SspAT from 6spAT
SspAT from 4spAT
6spAT from 4spAT
SspAT from 6spAT
SspAT from 4spAT
2017
-$13
$55
$43
$4
$24
$19
-$9
$80
$61
2018
-$12
$54
$42
$4
$24
$19
-$9
$78
$60
2019
-$12
$53
$41
$3
$18
$14
-$9
$71
$55
2020
-$12
$52
$40
$3
$18
$14
-$9
$70
$54
2021
-$12
$51
$39
$3
$18
$14
-$9
$69
$53
2022
-$12
$50
$38
$3
$18
$14
-$9
$68
$52
2023
-$11
$49
$38
$3
$18
$14
-$8
$67
$52
2024
-$11
$48
$37
$3
$18
$14
-$8
$66
$51
2025
-$11
$47
$36
$3
$18
$14
-$8
$65
$50
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; sp=speed; AT=automatic transmission

       Note that the cost for the 8 speed automatic transmission relative to the 6 speed
automatic transmission is lower here than that used in the recent heavy-duty GHG rule. In
that rule, we remained consistent with the proposal for that rule which carried  an estimated
DMC of $210 (2008$). That DMC was based on an estimate derived by NAS (see NAS
2010, Table 7-10). For this proposal, we have chosen to use the more recent DMC shown in
Table 3-37 which is based on a tear-down analysis done by FEV.

       3.4.2.6 Dual Clutch Transmissions / Automated Manual Transmissions
       (DCTAM)

       An Automated Manual Transmission (AMT) is mechanically similar to a conventional
manual transmission, but shifting and launch functions are automatically controlled by the
electronics. There are two basic types of AMTs, single-clutch  and dual-clutch (DCT).  A
single-clutch AMT is essentially a manual transmission with automated clutch and shifting.
Because of shift quality issues  with single-clutch designs,  DCTs are far more common in the
U.S. and are the basis of the estimates that follow. A DCT uses separate clutches (and
separate gear shafts) for the even-numbered gears and odd-numbered gears. In this way, the
next expected gear is pre-selected, which allows for faster and  smoother shifting.  For
example, if the vehicle is accelerating in third gear, the shaft with  gears one, three and five
has gear three engaged and is transmitting power. The shaft with gears two, four, and six is
ff This ICM would be applied to the 6 speed to 8 speed increment of $64 (2009$) applicable in 2012. The 4
speed to 6 speed increment would carry the low complexity ICM.
                                            3-103

-------
                                     Technologies Considered in the Agencies' Analysis
idle, but has gear four engaged. When a shift is required, the controller disengages the odd-
gear clutch while simultaneously engaging the even-gear clutch, thus making a smooth shift.
If, on the other hand, the driver slows down instead of continuing to accelerate, the
transmission will have to change to second gear on the idling shaft to anticipate a downshift.
This shift can be made quickly on the idling shaft since there is no torque being transferred on
it.

       In addition to single-clutch and dual-clutch AMTs, there are also wet clutch and dry
clutch designs which are used for different types of vehicle applications. Wet clutch AMTs
offer a higher torque capacity that comes from the use of a hydraulic system that cools the
clutches. Wet clutch systems are less efficient than the dry clutch systems due to the losses
associated with hydraulic pumping. Additionally, wet AMTs have a higher cost due to the
additional hydraulic hardware required.

       Overall, DCTs likely offer the greatest potential for effectiveness improvements
among the various transmission options presented in this report because they offer the
inherently lower losses of a manual transmission with the efficiency and shift quality
advantages of electronic controls. The lower losses stem from the elimination of the
conventional lock-up torque converter, and a greatly reduced need for high pressure hydraulic
circuits to hold clutches or bands to maintain gear ratios (in automatic transmissions) or hold
pulleys in position to maintain gear ratio (in Continuously Variable Transmissions).
However, the lack of a torque converter will affect how the vehicle launches from rest, so a
DCT will most likely be paired with an engine that offers sufficient torque at low engine
speeds to allow for adequate launch performance or provide lower launch gears to
approximate the torque multiplication of the torque converter to provide equivalent
performance.

In MYs 2012-2016 final rule, EPA and NHTSA estimated a 5.5 to 9.5 percent improvement
in fuel consumption over a baseline 4/5-speed automatic transmission for a wet clutch DCT,
which was assumed for all but the smallest of vehicle subclasses, Subcompact and Compact
cars and small LT.  This results in an incremental effectiveness estimate of 2.7 to 4.1 percent
over a 6-speed automatic transmission with IATC. For Subcompact and Compact Cars and
small LT, which were assumed to use a dry clutch DCT, NHTSA estimated an 8 to 13 percent
fuel consumption improvement over a baseline 4/5-speed automatic transmission, which
equates to a 5.5 to 7.5 percent incremental improvement over the 6-speed transmission.

Based on the 2011 Ricardo study, EPA and NHTSA have concluded that 8 to  13 percent
effectiveness is appropriate for 6-speed DCTs and 11 to 16  percent is appropriate for 8-speed
DCTs for this proposal.  These values include not only the DCT but also the increase in
stepped gears and also a high efficiency  gearbox (mentioned later).  Independent of other
technologies, this translates to an effectiveness  for the DCT, alone, of 4 to 5% (for wet-clutch
designs) and 5 to 6% (for dry-clutch designs) compared to a baseline automatic transmission
of similar vintage and number of fixed gears.
                                           3-104

-------
                                    Technologies Considered in the Agencies' Analysis
In this NPRM analysis, NHTSA applied an incremental effectiveness of 4 percent for a 6-
speed dry DCT and 3.4 to 3.8 percent for a wet DCT compared to a 6-speed automatic
transmission based on the lumped parameter model which includes the accompanied
transmission efficiency improvement for MY 2010 and after transmissions. This translates to
an effectiveness range of 7.4 to 8.6 percent compared to a 4 speed automatic transmission for
dry clutch design and 7.4 to 7.9 percent for a wet clutch design. NHTSA did not apply DCTs
to vehicles with towing requirements, such as Minivan, Midsize light truck and large pickup
truck. EPA did not apply DCTs to vehicle types classified as towing as described in Chapter 1
of EPA's Draft RIA.

       In the 2010 TAR, the agencies estimated the DMC at -$234 (2008$) for a 6 speed dry-
clutch DCT and -$165 for a 6 speed wet-clutch DCT with both DMCs applicable in the
2017MY (see 2010 TAR, Table B2.1-1 at page B-10) and both incremental to a 4 speed
automatic transmission.  In the 2010 TAR, we pointed to Chapter 3 of the 2012-2016 final
joint TSD where we noted that the DCT costs of -$147 (2007$ and incremental to a 6-speed
automatic transmission) were based on a FEV tear-down study that assumed 450,000 units of
production. We went on to state that we did not consider there to be sufficient US capacity in
the  2012-2016 timeframe to produce 450,000 units and for that reason we were adjusting the
tear-down values accordingly. The TAR timeframe for consideration was 2017-2025, and in
the  TAR  we argued that production capacity would exist and that the FEV tear-down results
we  valid without adjustment.  We continue to believe that to be the case.  In the final joint
TSD supporting the 2012-2016 rule we also noted that the negative tear-down estimates found
by FEV were not surprising when considering the relative simplicity of a dual-clutch
transmission compared to an automatic transmission. Again, we continue to consider this to
be true.

       For this analysis, we consider the 2010 TAR DMCs to be applicable to the 2012MY,
thus the DMCs become  -$236 (2009$) and -$167 (2009$) for 6 speed dry- and wet-clutch
DCTs, respectively, both applicable in the 2012MY and incremental to a 4 speed auto
transmission. The agencies consider the 6 speed DCT technology to be on the flat portion of
the  learning curve and have applied a medium complexity ICM of 1.39 through 2018 then
1.29 thereafter.  The resultant costs are shown in Table 3-38.

       New for this analysis is costing for an 8 speed DCT.  For the cost of this technology,
the  agencies have relied on a tear-down study completed by FEV since publication of the
TAR.53 In that study, the 8 speed DCT was found to be $198 (2007$) more costly than the 6
speed DCT.  This DMC increment becomes $202 (2009$) for this analysis. Adding the $202
(2009$) to the -$236 (2009$)  DMC and the -$167 (2009$) DMC for a 6 speed dry- and wet-
clutch DCT, the 8 speed dry- and wet-clutch DCTs relative to a 4 speed auto transmission
would be -$32 (2009$) and $38 (2009$), respectively. The agencies consider this DMC to be
applicable to the 2012MY.  The agencies consider the 8 speed DCT technology to be on the
flat portion of the learning curve and have applied a medium complexity ICM of 1.39 through
the  2024MY then  1.29 thereafter. The 8 speed DCT has a later switch to long term ICMs
because it is a newer technology that is not currently implemented in the fleet.  The resultant
costs for both 6 speed and 8 speed DCTs are shown in Table 3-38.

                                          3-105

-------
                                     Technologies Considered in the Agencies' Analysis
                Table 3-38 Costs for 6 & 8 Speed Dual Clutch Transmissions (2009$)
Cost
type
DMC
DMC
DMC
DMC
1C
1C
1C
1C
TC
TC
TC
TC
Transmission type
6spDCT-dry
6sp DCT-wet
8sp DCT-dry
8sp DCT-wet
6spDCT-dry
6sp DCT-wet
8sp DCT-dry
8sp DCT-wet
6spDCT-dry
6sp DCT-wet
8sp DCT-dry
8sp DCT-wet
2017
-$205
-$145
-$28
$33
$90
$64
$12
$14
-$115
-$81
-$16
$47
2018
-$201
-$142
-$27
$32
$90
$63
$12
$14
-$111
-$78
-$15
$46
2019
-$197
-$139
-$27
$31
$67
$47
$12
$14
-$130
-$91
-$14
$46
2020
-$193
-$136
-$26
$31
$67
$47
$12
$14
-$126
-$89
-$14
$45
2021
-$189
-$133
-$26
$30
$67
$47
$12
$14
-$122
-$86
-$13
$44
2022
-$185
-$131
-$25
$30
$67
$47
$12
$14
-$118
-$84
-$13
$44
2023
-$182
-$128
-$25
$29
$67
$47
$12
$14
-$115
-$81
-$12
$43
2024
-$178
-$125
-$24
$28
$67
$47
$12
$14
-$111
-$79
-$12
$43
2025
-$174
-$123
-$24
$28
$67
$47
$9
$11
-$108
-$76
-$15
$38
DMC=Direct manufacturing cost;
Note that all costs are relative to a
IC=Indirect cost; TC=Total cost; sp=speed;
4 speed automatic transmission.
dry=dry clutch; wet=wet-clutch
       3.4.2.7 6-Speed Manual Transmissions (6MAN)

       Manual transmissions are entirely dependent upon driver input to shift gears: the
driver selects when to perform the shift and which gear to select. This is the most efficient
transfer of energy of all transmission layouts, because it has the lowest internal gear losses,
with a minimal hydraulic system, and the driver provides the energy to actuate the clutch.
From a systems viewpoint, however, vehicles with manual transmissions have the drawback
that the driver may not always select the optimum gear ratio for fuel economy. Nonetheless,
increasing the number of available ratios in a manual transmission can improve fuel economy
by allowing the driver to  select a ratio that optimizes engine operation more often.  Typically,
this is achieved through adding overdrive ratios to reduce engine speed at cruising velocities
(which saves fuel through reduced engine pumping  losses) and pushing the torque required of
the engine towards the optimum level. However, if the gear ratio steps are not properly
designed, this may require the driver to change gears more often in city driving, resulting in
customer dissatisfaction.  Additionally, if gear ratios are selected to achieve improved launch
performance instead of to improve fuel economy, then no fuel saving effectiveness is realized.

       The 2012-2016 final rule estimated an effectiveness increase of 0.5 percent for
replacing a 5-speed manual with a 6-speed manual transmission, which was derived from
confidential manufacturer data. Based on the updated LPM, NHTSA has found that an
effectiveness increase of 2.0 to 2.5 percent is possible when moving from a 5-speed to a 6-
speed manual transmission with improved internals. NHTSA updated costs to reflect the
ICM low complexity markup of 1.11 which resulted in an incremental compliance cost of
$250  as compared to $338 for MY 2012. This represents a DMC of $225 (2007$) which
becomes $232 (2009$) for this analysis, applicable in the 2012MY.  NHTSA continues to
consider a 6 speed manual transmission to be on the flat portion of the learning curve and has
applied a low complexity ICM of 1.24 through 2018 then 1.19 thereafter.  NHTSA's resultant
costs  for a 6 speed manual transmission are shown in Table 3-39.
                                            3-106

-------
                                       Technologies Considered in the Agencies' Analysis
               Table 3-39 Costs for 6 Speed Manual Transmission (2009$)
Cost
type
DMC
1C
TC
Transmission type
6sp manual
6sp manual
6sp manual
2017
$202
$56
$257
2018
$197
$56
$253
2019
$194
$45
$238
2020
$190
$44
$234
2021
$186
$44
$230
2022
$182
$44
$227
2023
$179
$44
$223
2024
$175
$44
$219
2025
$171
$44
$216
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost; sp=speed; dry=dry clutch; wet=wet-clutch
Note that all costs are relative to a 5 speed manual transmission.

3.4.3  Vehicle electrification and hybrid electric vehicle technologies

For the costs presented in this electrification and hybrid vehicle section, we have estimated
costs for vehicle classes since the technologies are closely linked to the size of the vehicle as
opposed to the number of cylinders on the engine or its valvetrain configuration. The vehicle
classes for which we have estimated costs are consistent with the seven vehicle classes
developed for the lumped parameter model.  Each agency has used the vehicle class specific
costs and mapped those into their respective model-specific vehicle classes or types as shown
in Table 3-40.  This table simply presents the mapping of lumped parameter model vehicle
classes (or cost vehicle classes) into model-specific vehicle classes (or vehicle types in the
case of EPA's OMEGA model, please refer to Chapter 1 of EPA's draft RIA for more details)
to help the reader understand how the vehicle classes used for costing relate to the vehicle
classes used for modeling.

    Table 3-40 Mapping of Vehicle Class into each Agency's Model-Specific Vehicle Classes or Types
EPA Vehicle
Class for Cost
Purpose
Subcompact Car
Small Car
Large Car
Minivan
Small Truck
Minivan with
Towing
Large Truck
Lump
Parameter
Classification
Small Car
StdCar
Large Car
Large MPV
Small MPV
Large MPV
Truck
Example
Yaris
Camry
Chrysler 300
Dodge Grand
Caravan
Saturn Vue
Dodge Grand
Caravan
FordF150
OMEGA Model
Vehicle Type*
1
2,3
5,6,15
4,7
8
9, 10, 16, 17
11, 12, 13,14, 18, 19
NHTSA/CAFE
Model
Classification
Subcompact
Subcompact Perf PC
Compact
Compact Perf PC
Mid-size PC
Mid-size Perf PC
Large PC
Large Perf PC

Small LT
Midsize LT
MinVan LT
Large LT
 * OMEGA uses 19 vehicle types as shown here and described in detail in Chapter 1 of EPA's draft RIA.
                                              3-107

-------
                                    Technologies Considered in the Agencies' Analysis
       3.4.3.1 Electrical Power Steering (EPS) / Electrohydraulic Power Steering
       (EHPS)

       Electric power steering (EPS) and Electrohydraulic power steering (EHPS) provide a
potential reduction in CC>2 emissions and fuel consumption over hydraulic power steering
because of reduced overall accessory loads. This eliminates the parasitic losses associated
with belt-driven power steering pumps which consistently draw load from the engine to pump
hydraulic fluid through the steering actuation systems even when the wheels are not being
turned.  EPS is an enabler for all vehicle hybridization technologies since it provides power
steering when the engine is off. EPS may be implemented on most vehicles with a standard
12V system.  Some heavier vehicles may require a higher voltage system or EHPS which may
add cost and complexity.

       The 2012-2016 final rule, EPA and NHTSA estimated a 1 to 2 percent effectiveness
for light duty vehicles based on the 2002 NAS report, Sierra Research Report and confidential
OEM data. The 2010 Ricardo study also confirmed this estimate.  NHTSA and EPA
reviewed these effectiveness estimates and found them to be accurate, thus they have been
retained for this proposal. For large pickup truck the agencies used EHPS due to the utility
requirement of these vehicles. The effectiveness of EHPS is estimated to be 0.8 percent.

       In the MY 2012-2016 final rule, the agencies estimated the DMC at $88 (2007$).
Converting to 2009$, this DMC becomes $90 for this analysis, consistent with the recent
heavy-duty GHG rule, which is considered applicable in the 2015MY.  The agencies use the
same DMC for EPS as for EHPS. Technically, EHPS is less costly than EPS.  However, we
believe that EHPS is likely to be used, if at all, on the largest trucks and utility vehicles. As
such, it would probably need to be heavier-duty than typical EPS systems and the agencies
consider the net effect to place EHPS on par with EPS in terms of costs.  The agencies
consider EPS/EHPS technology to be on the flat portion of the learning curve and have
applied  a low complexity ICM of 1.24 through 2018 then 1.19 thereafter.  The  resultant costs
are shown in Table 3-41.

              Table 3-41 Costs of Electrical/Electro-hydraulic Power Steering (2009$)
Cost type
DMC
1C
TC
2017
$86
$22
$108
2018
$84
$22
$106
2019
$82
$17
$100
2020
$81
$17
$98
2021
$79
$17
$96
2022
$78
$17
$95
2023
$76
$17
$93
2024
$74
$17
$92
2025
$73
$17
$90
            DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                                           3-108

-------
                                    Technologies Considered in the Agencies' Analysis
       3.4.3.2 Improved Accessories

       The accessories on an engine, including the alternator, coolant and oil pumps are
traditionally mechanically-driven.  A reduction in CO2 emissions and fuel consumption can be
realized by driving them electrically, and only when needed ("on-demand").

       Electric water pumps and electric fans can provide better control of engine cooling.
For example, coolant flow from an electric water pump can be reduced and the radiator fan
can be shut off during engine warm-up or cold ambient temperature conditions which will
reduce warm-up time, reduce warm-up fuel enrichment, and reduce parasitic losses.

       Indirect benefit may be obtained by reducing the flow from the water pump
electrically during the engine warm-up period, allowing the engine to heat more rapidly and
thereby reducing the fuel enrichment needed during cold starting of the engine. Further
benefit may be obtained when electrification is combined with an improved, higher efficiency
engine alternator. Intelligent cooling can more easily be applied to vehicles that do not
typically carry heavy payloads, so larger vehicles with towing capacity present a challenge, as
these vehicles have high cooling fan loads. Both agencies also included a higher efficiency
alternator in this category to improve the cooling system. Both agencies also included a higher
efficiency alternator in this category to improve the cooling system.

       The agencies considered whether to include electric oil pump technology for the
rulemaking.  Because it is necessary to operate the oil pump any time the engine is running,
electric oil pump technology has insignificant effect on efficiency.  Therefore, the agencies
decided to not include electric oil pump technology for this proposal.

       In MYs 2012-2016 final rule, the agencies used the effectiveness value in the range of
1 to 2 percent based on technologies discussed above. NHTSA did not apply this technology
to large pickup truck due to the utility requirement concern for this vehicle class.

       For this proposal, the agencies are considering two levels of improved  accessories. For
level one of this technology (IACC1) NHTSA now incorporates a high efficiency alternator
(70 percent efficiency). The second level of improved accessories  (IACC2) adds the higher
efficiency alternator and incorporates a mild regenerative alternator strategy, as well as
intelligent cooling. NHTSA and EPA jointly reviewed the estimates of 1 to 2  percent
effectiveness estimates used in the 2012-2016 final rule and TAR for level IACC1. More
precisely, the agencies used effectiveness value in 1.2 to 1.8 percent range varying based on
different vehicle  subclasses. The incremental effectiveness for this  technology in relative to
EPS in the CAFE model is 0.91 to  1.61 percent. The combined effectiveness for IACC1 and
IACC2 ranges from 3.1 to 3.9 percent and NHTSA applied incremental effectiveness of
IACC2 in  relative to IACC1 ranging from 1.74 to 2.55 percent.

       In the 2012-2016 rule, the agencies estimated the DMC of IACC1 at $71 (2007$).
Converting to 2009$, this DMC becomes $73 for this analysis, applicable in the 2015MY, and
consistent with the heavy-duty GHG rule.  The agencies consider IACC1 technology to be on
                                           3-109

-------
                                     Technologies Considered in the Agencies' Analysis
the flat portion of the learning curve and have applied a low complexity ICM of 1.24 through
2018 then 1.19 thereafter.

       Cost is higher for IACC2 due to the inclusion of a higher efficiency alternator and a
mild level of regeneration. The agencies estimate the DMC of the higher efficiency alternator
and the regeneration strategy at $45 (2009$) incremental to IACC1, applicable in the
2015MY. Including the costs for IACC1 results in a DMC for IACC2 of $118 (2009$)
relative to the baseline case and applicable in the 2015MY. The agencies consider the IACC2
technology to be on the flat portion of the learning curve.  The agencies have applied a low
complexity ICM of 1.24 through 2018 then 1.19 thereafter. The resultant costs are shown in
Table 3-42.

            Table 3-42 Costs for Improved Accessory Technology - Levels 1 & 2 (2009$)
Cost type
DMC
DMC
1C
1C
TC
TC
IACC
Technology
IACC1
IACC2
IACC1
IACC2
IACC1
IACC 2
2017
$70
$113
$18
$29
$87
$141
2018
$68
$110
$18
$29
$86
$139
2019
$67
$108
$14
$23
$81
$131
2020
$66
$106
$14
$23
$80
$129
2021
$64
$104
$14
$23
$78
$127
2022
$63
$102
$14
$23
$77
$124
2023
$62
$100
$14
$23
$76
$122
2024
$61
$98
$14
$23
$75
$120
2025
$59
$96
$14
$23
$73
$118
      DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
      Note that both levels of IACC technology are incremental to today's baseline case.
       3.4.3.3 Air Conditioner Systems

       We have a detailed description of the AC program in Chapter 5 of this draft joint TSD.
The reader is directed to that chapter to learn the specifics of the program, the credits
involved, and details behind the costs we have estimated.  Table 3-43 is a copy of Table 5-17
showing the total costs for A/C controls used in this proposal.

               Table 3-43 Total Costs for A/C Control Used in This Proposal (2009$)
Car/
Truck
Car
Truck
Fleet
Cost type
TC
TC
TC
TC
TC
TC
TC
Rule
Reference
Control
Both
Reference
Control
Both
Both
2017
$75
$25
$100
$57
$2
$60
$85
2018
$74
$40
$114
$56
$46
$102
$110
2019
$69
$56
$126
$53
$73
$126
$126
2020
$68
$65
$133
$52
$81
$133
$133
2021
$67
$78
$145
$51
$94
$145
$145
2022
$66
$76
$142
$50
$92
$142
$142
2023
$65
$72
$137
$50
$87
$137
$137
2024
$64
$70
$134
$49
$85
$134
$134
2025
$63
$69
$132
$48
$84
$132
$132
     TC=Total cost
                                            3-110

-------
                                     Technologies Considered in the Agencies' Analysis
       3.4.3.4 Stop-start (12V Micro Hybrid)

       The stop-start technology we consider for this proposal—also known as idle-stop or
12-volt micro-hybrid—is the most basic hybrid system that facilitates idle-stop capability.
When vehicle comes to a stop, the system will automatically shut down the internal
combustion engine and restarts the engine when vehicle starts to move again. This is
especially beneficial to reduce emission and fuel consumption when vehicle spends
significant amount of time stopping in traffic jam. Along with other enablers, this system
typically replaces the standard 12-volt starter with an improved unit capable of higher power
and increased cycle life.  These systems typically incorporate an improved battery to prevent
voltage-droop on restart. Different from MY 2012-2016 rule, this technology is applied to all
vehicle classes, including large pickup truck. In MYs 2012-2016 final rule, even though EPA
did not use 12 volt stop-start technology, NHTSA and EPA jointly reviewed the assumption.
The effectiveness NHTSA used in the CAFE model for MYs 2012-2016 final rule ranged
from 2 to 4 percent, depending on whether the vehicle is  equipped with a 4-, 6- or 8-cylinder
engine, with the 4-cylinder engine having the lowest range and the 8-cylinder having the
highest88. In this NPRM analysis, when combining IACC1, IACC2 and 12V stop-start system,
the estimated effectiveness based on 2010 Ricardo study ranges from 4.8 percent to 5.9
percent. The agencies applied this effectiveness in the NPRM analysis. For CAFE modeling,
the incremental effectiveness for 12V stop-start relative to IACC2 is 1.68 to 2.2 percent.

       In the 2012-2016 rule, the agencies estimated the DMC at $282 (2007$) to $350
(2007$) for small cars through large trucks, respectively.  Converting to 2009$, these DMCs
become $290 (2009$) through $361 (2009$) for this analysis which are considered applicable
in the 2015MY.  The agencies consider 12V stop-start technology to be on the steep portion
of the learning curve in the 2012-2016 timeframe and flat thereafter and have applied a
medium complexity ICM of 1.39 through 2018 then 1.29 thereafter. The resultant costs are
shown in Table 3-44.

         Table 3-44 EPA and NHTSA Costs for 12V Micro Hybrid or 12V Stop-Start (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
Vehicle
Class
Subcompact/
Small car
Large car
Minivan
Small truck
Large truck
Subcompact/
Small car
Large car
Minivan
Small truck
Large truck
2017
$282
$319
$319
$319
$350
$112
$127
$127
$127
$139
2018
$273
$310
$310
$310
$340
$112
$127
$127
$127
$139
2019
$265
$300
$300
$300
$329
$83
$94
$94
$94
$104
2020
$257
$291
$291
$291
$320
$83
$94
$94
$94
$103
2021
$249
$283
$283
$283
$310
$83
$94
$94
$94
$103
2022
$242
$274
$274
$274
$301
$83
$94
$94
$94
$103
2023
$235
$266
$266
$266
$292
$82
$93
$93
$93
$102
2024
$228
$258
$258
$258
$283
$82
$93
$93
$93
$102
2025
$221
$250
$250
$250
$274
$82
$93
$93
$93
$102
gg For a description of how Stop Start is considered for off-cycle credits refer to TSD Chapter 5.2.3.1.

                                           3-111

-------
                                     Technologies Considered in the Agencies' Analysis
TC
TC
TC
TC
TC
Subcompact/
Small car
Large car
Minivan
Small truck
Large truck
$394
$446
$446
$446
$490
$385
$436
$436
$436
$479
$348
$395
$395
$395
$433
$340
$385
$385
$385
$423
$332
$376
$376
$376
$413
$324
$368
$368
$368
$403
$317
$359
$359
$359
$394
$310
$351
$351
$351
$385
$303
$343
$343
$343
$376

   DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
       3.4.3.5 Mild Hybrid

       Mild hybrid systems, also called Higher Voltage Stop-Start and Belt Mounted
Integrated Starter Generator (BISG) systems are similar to a micro-hybrid system, offering
idle-stop functionality, except that they utilize larger electric machine and a higher capacity
battery, typically 42 volts or above, thus enabling a limited level of regenerative braking
unavailable for a MHEV. The larger electric machine and battery also enables a limited
degree of power assist, which  MHEV cannot provide. However, because of the limited torque
capacity of the belt-driven design, these systems have a smaller electric machine, and thus
less capability than crank-integrated or stronger hybrid systems. These systems replace the
conventional alternator with a belt-driven starter/alternator and may add high voltage
electrical accessories (which may include electric power steering and an auxiliary automatic
transmission pump).  The limited electrical requirements of these systems allow the use of
lead-acid batteries or supercapacitors for energy storage.

       The MY 2012-2016 final rule estimates the effectiveness for these technologies range
from 3.0 to 7.5 percent depending on vehicle subclass. The CAFE model, which applies this
effectiveness incrementally to the prior 12 Volt MHEV technology, uses estimates of 4 to 6
percent.

       EPA estimates an incremental compliance cost range of $549 (small car) to $682
(large truck) for a MY 2012 vehicle and including a medium complexity ICM of 1.25
(2007$). With volume-based learning applied, these become $351 (small car) and $437 (large
truck) for a MY 2016 vehicle (2007$). The cost estimate in the CAFE model is incremental
to the 12 Volt micro hybrid systems as noted above, and therefore is adjusted upwards to $286
to reflect the additional battery capacity, wiring upgrades, and a larger optimized electric
machine only.  The $286 reflects volume-based learning factors and the ICM medium-
complexity markup of 1.25. This technology is not applied in this NPRM analysis.
3.4.3.5.1 Integrated Motor Assist (IMA)/Crank Integrated Starter Generator (CISC)
                                                     ,54
       IMA is a system developed and marketed by Honda  and is similar to CISG.  They
both utilize a thin axial electric motor bolted to the engine's crankshaft and connected to the
transmission through a torque converter or clutch. The axial motor is motor/generator that
typically operates above 100 volts (but lower than the stronger hybrid systems discussed
                                           3-112

-------
                                     Technologies Considered in the Agencies' Analysis
below, which typically operate at around 300 volts) and can provide sufficient torque for
launch as well as generate sufficient current to provide significant levels of brake energy
recovery.  The motor/generator also acts as the starter for the engine and can replace a typical
accessory-driven alternator. Current IMA/CISG systems typically do not launch the vehicle
on electric power alone, although some commercially available systems can cruise on electric
power and dual-clutch IMA/CISG systems capable of all-electric drive are under
development. IMA and CISG could be applied to all classes of vehicles. This technology is
not used as an enabling technology in this  NPRM analysis by either EPA or NHTSA.

       EPA relied on a combination of certification data (comparing vehicles available with
and without a hybrid system and backing out other components where appropriate) and
manufacturer-supplied information to determine that the effectiveness of these systems in
terms of COi reduction is 30 percent for small cars, 25 percent for large cars, and 20 percent
for minivans and small trucks  similar to the range estimated by NHTSA for the respective
vehicle classes.  The effectiveness for small cars assumes engine downsizing to maintain
approximately equivalent performance.  The large car, minivan, and small truck effectiveness
values assume less engine downsizing in order to improve vehicle performance and/or
maintain towing and hauling performance.

       In the 2012-2016 final rule, the agencies estimated the DMC at $1,973, $2,497,
$2,508, $2,366 and $3,063 (all values in 2007$) for a small car, large car, minivan, small
truck and large truck, respectively. These  DMCs become $2,034, $2,575, $2,586, $2,440 and
$3,159 (all values in 2009$) for this analysis. All of these DMCs are considered applicable in
the 2015MY. The agencies consider the IMA technology to be on the steep portion of the
learning curve and have applied a highl complexity ICM of 1.56 through 2018 then  1.35
thereafter.  The  resultant costs are as shown in Table 3-45.  As noted earlier, the IMA
technology is not included as an enabling technology in this analysis, although it is included
as a baseline technology because it exists in the 2008 baseline fleet. The agencies moved
away from this technology and applied P2  hybrid instead because comparing to IMA, P2 is
more cost effective.

                         Table 3-45 Costs for IMA Hybrids (2009$)
Cost
type
DMC
DMC
DMC
DMC
1C
1C
1C
1C
TC
TC
TC
Vehicle Class
Subcompact/Small
car
Large car
Minivan
Small truck
Subcompact/Small
car
Large car
Minivan
Small truck
Subcompact/Small
car
Large car
Minivan
2017
$1,973
$2,498
$2,508
$2,367
$1,143
$1,446
$1,452
$1,370
$3,116
$3,944
$3,961
2018
$1,914
$2,423
$2,433
$2,296
$1,139
$1,441
$1,447
$1,366
$3,053
$3,864
$3,880
2019
$1,857
$2,350
$2,360
$2,227
$697
$882
$886
$836
$2,554
$3,233
$3,246
2020
$1,801
$2,280
$2,289
$2,160
$695
$880
$884
$834
$2,496
$3,160
$3,173
2021
$1,747
$2,211
$2,221
$2,095
$694
$878
$882
$832
$2,440
$3,089
$3,102
2022
$1,695
$2,145
$2,154
$2,032
$692
$876
$879
$830
$2,386
$3,021
$3,033
2023
$1,644
$2,081
$2,089
$1,971
$690
$874
$877
$828
$2,334
$2,954
$2,967
2024
$1,594
$2,018
$2,027
$1,912
$689
$872
$875
$826
$2,283
$2,890
$2,902
2025
$1,547
$1,958
$1,966
$1,855
$687
$870
$873
$824
$2,234
$2,827
$2,839
                                           3-113

-------
                                     Technologies Considered in the Agencies' Analysis
  TC   |     Small truck    |  $3,737 |  $3,662 |  $3,063 |  $2,994 |  $2,927 |  $2,862 |  $2,799 |  $2,738 | $2,679
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
       3.4.3.6 HEV, PHEV, EV and Fuel Cell Vehicle Technologies

       A hybrid vehicle is a vehicle that combines two or more sources of propulsion energy,
where one uses a consumable fuel (like gasoline), and one is rechargeable (during operation,
or by another energy source). Hybrid technology is well established in the U.S. market and
more manufacturers are adding hybrid models to their lineups. Hybrids reduce fuel
consumption through three major mechanisms:

       •  The internal combustion engine can be optimized (through downsizing, modifying
          the operating cycle, or other control techniques)  to operate at or near its most
          efficient point more of the time. Power loss from engine downsizing can be
          mitigated by employing power assist from the secondary power source.

       •  Some of the energy normally lost as heat while braking can be captured and stored
          in the energy storage system for later use.

       •  The engine  is turned off when it is not needed, such as when the vehicle is coasting
          or when stopped.

       Hybrid vehicles utilize some combination of the three above mechanisms to reduce
fuel consumption and COi emissions.  A fourth mechanism to reduce petroleum fuel
consumption, available only to plug-in hybrids, is by substituting the petroleum fuel energy
with energy from another source, such as the electric grid.  The effectiveness of fuel
consumption and COi reduction depends on the utilization of the above mechanisms  and how
aggressively they are pursued.  One area where this variation is particularly prevalent is in the
choice of engine size and its effect  on balancing fuel economy and performance. Some
manufacturers choose not to downsize the engine when applying  hybrid technologies. In
these cases,  performance is vastly improved, while fuel efficiency improves  significantly less
than if the engine was downsized to maintain the same performance as the conventional
version.  While this approach has been used in cars  such as  the Lexus 600h luxury vehicle, it
is more likely to be used in the future for vehicles like trucks where towing and/or hauling are
an integral part of their performance requirements.  In these cases, if the engine is downsized,
the battery can be quickly drained during a long hill climb with a heavy load, leaving only a
downsized engine to  carry the entire load. Because towing  capability is currently a heavily-
marketed truck attribute, manufacturers are hesitant to offer a truck with downsized engine
which can lead to a significantly diminished towing performance when the battery state of
charge level is low, and therefore engines are traditionally not downsized for these vehicles.

       Although hybrid vehicles using other energy storage concepts (flywheel, hydraulic)
have been developed, the automotive systems in production for passenger cars and light
trucks are all hybrid electric vehicles (HEV) that use battery storage and electric drive

                                            3-114

-------
                                     Technologies Considered in the Agencies' Analysis
systems. This appears likely to be the case for the foreseeable future. HEVs are part of a
continuum of vehicles using systems with differing levels of electric drive and electric energy
storage. This range of vehicles includes relatively basic system without electric energy
storage such as engine start/stop systems; HEV systems with varying degrees of electric
storage and electric drive system capability including mild-hybrid electric vehicles (MHEV)
with limited capability but lower cost; strong hybrid electric vehicles (SHEV) with full
hybridization capability such as the P2 hybrid technology which the agencies evaluate as a
compliance option in this NPRM; plug-in hybrid electric vehicles (PHEV) with differing
degrees of all electric range and battery electric vehicles (EV) that rely entirely on electric
drive and battery electric energy storage.

       Different HEV, PHEV and EV concepts utilize these mechanisms differently, so they
are treated separately for the purposes of this analysis.  In many applications, particularly with
PHEV and EV, the battery represents the most costly and system-limiting sub-component of
the hybrid system. Currently, there are many battery chemistries being developed and refined
for hybrid applications that are expected to enhance the performance of future hybrid
vehicles. Section 3.4.3.6.4 contains a discussion of battery energy storage and the major
hybrid concepts that were determined to be available during the MY 2017-2015 timeframe.

       Fuel cell vehicles are a separate category of electric vehicle  that rely entirely on
electric propulsion with electricity produced on-board the vehicle using a proton-exchange-
membrane fuel cell (PEMFC) fueled with hydrogen.  Fuel cell vehicles under development
are typically configured as a hybrid with battery storage used to provide brake energy
recovery and improved response to fast transients in vehicle energy demand.

3.4.3.6.1 Power-split hybrid

       Power-split hybrid (PSHEV) - a hybrid electric drive system that replaces the
traditional transmission with a single planetary gear set and a motor/generator. This
motor/generator uses the engine to either charge the battery or to  supply additional power to
the drive motor.  A second, more powerful motor/generator is permanently connected to the
vehicle's final drive and always turns with the  wheels.  The planetary gear splits engine power
between the first motor/generator and the drive motor to either charge the battery or supply
power to the wheels. Power-split hybrids are not used as an enabling technology in this
proposal.

       In MYs 2012-2016 final rule, EPA and NHTSA used a combination of manufacturer-
supplied information and a comparison of vehicles available with and without a hybrid system
from EPA's fuel economy test data to determine that the effectiveness is 19 to 36 percent for
the classes to which it is applied. The estimate would depend on whether engine downsizing
is also assumed.  In the CAFE incremental model, the range of effectiveness used was 23 to
33 percent as engine downsizing is not assumed (and accounted for elsewhere).

       For this analysis, in order to estimate baseline costs, the agencies are using power-split
HEV costs generated by FEV as part of a tear-down study. In that study, FEV found the


                                           3-115

-------
                                     Technologies Considered in the Agencies' Analysis
DMC of the entire power-split system (battery-pack and non-battery components) to be
$2,853 (2007$), $3,175 (2007$), $3,435 (2007$), $4,168 (2007$) for vehicle sized, for
example, like a Ford Fiesta, Ford Focus, Ford Fusion and Ford Flex, respectively.  For this
analysis, these values become $2,942, $3,274, $3,542 and $4,298, respectively, all in 2009
dollars. In the 2012-2016 final rule, the agencies estimated the DMC of a large truck power-
split system at $5,137 (2007$) which becomes $5,299 for this analysis (2009$) and we are
using this value for the minivan-towing vehicle class.  All of these DMCs are considered
applicable in the 2015MY.  The agencies consider the power-split technology to be on the flat
portion of the learning curve and have applied a highl complexity ICM of 1.56 through 2018
then 1.35 thereafter.  The resultant costs are as shown in Table 3-46. As noted earlier, the
IMA technology is not included as an enabling technology in this analysis, although it is
included as a baseline technology because it exists in the 2008 baseline fleet.

                       Table 3-46 Costs for Power-Split Hybrids (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
Vehicle Class
Subcompact
Small car
Large car
Minivan
Small truck
Minivan-towing
Subcompact
Small car
Large car
Minivan
Small truck
Minivan-towing
Subcompact
Small car
Large car
Minivan
Small truck
Minivan-towing
2017
$2,797
$3,112
$3,367
$4,086
$3,393
$5,037
$1,649
$1,835
$1,985
$2,409
$2,000
$2,970
$4,445
$4,947
$5,352
$6,494
$5,392
$8,007
2018
$2,741
$3,050
$3,300
$4,004
$3,325
$4,937
$1,645
$1,831
$1,981
$2,403
$1,996
$2,963
$4,386
$4,881
$5,281
$6,407
$5,320
$7,900
2019
$2,686
$2,989
$3,234
$3,924
$3,258
$4,838
$1,008
$1,122
$1,214
$1,473
$1,223
$1,816
$3,694
$4,111
$4,448
$5,396
$4,481
$6,654
2020
$2,632
$2,929
$3,169
$3,845
$3,193
$4,741
$1,006
$1,120
$1,212
$1,470
$1,221
$1,813
$3,638
$4,049
$4,381
$5,315
$4,414
$6,554
2021
$2,579
$2,871
$3,106
$3,768
$3,129
$4,646
$1,005
$1,118
$1,210
$1,468
$1,219
$1,810
$3,584
$3,989
$4,315
$5,236
$4,348
$6,456
2022
$2,528
$2,813
$3,044
$3,693
$3,067
$4,553
$1,003
$1,116
$1,208
$1,465
$1,217
$1,807
$3,531
$3,930
$4,251
$5,158
$4,283
$6,360
2023
$2,477
$2,757
$2,983
$3,619
$3,005
$4,462
$1,001
$1,114
$1,206
$1,463
$1,215
$1,804
$3,479
$3,872
$4,188
$5,082
$4,220
$6,266
2024
$2,428
$2,702
$2,923
$3,547
$2,945
$4,373
$1,000
$1,113
$1,204
$1,461
$1,213
$1,801
$3,428
$3,815
$4,127
$5,007
$4,158
$6,174
2025
$2,379
$2,648
$2,865
$3,476
$2,886
$4,286
$998
$1,111
$1,202
$1,458
$1,211
$1,798
$3,377
$3,759
$4,067
$4,934
$4,097
$6,084
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
3.4.3.6.2 2-mode hybrid

       2-mode hybrid (2MHEV) - is a hybrid electric drive system that uses an adaptation of
a conventional stepped-ratio automatic transmission by replacing some of the transmission
clutches with two electric motors that control the ratio of engine speed to vehicle speed, while
clutches allow the motors to be bypassed. This improves both the transmission torque
capacity for heavy-duty applications and reduces fuel consumption and CO2 emissions at
highway speeds relative to other types of hybrid electric drive systems. 2-mode hybrids have
not been considered in this proposal.  Depending on the comments that the agencies received
for this NPRM, the agencies might re-consider this hybrid technology in vehicles with towing
requirement, such as pickup trucks, in the final rule.
                                            3-116

-------
                                      Technologies Considered in the Agencies' Analysis
        For MYs 2012-2016 final rule, the CAFE model considered a range of 23 to 33
 percent with a midpoint of 28 percent, assuming no engine downsizing to preserve the utility
 nature of medium and large trucks (e.g., maintaining full towing capability even in situations
 with low battery charge) and EPA estimates COi emissions reduction effectiveness to be 25
 percent for large trucks (LDT3 and LDT4 categories) based on vehicle certification data.
 EPA estimates an effectiveness of 40 percent for smaller vehicles.

        The agencies have estimated the costs for 2-mode hybrids using costs used in the 2010
 TAR.  For this analysis, the 2-mode battery pack DMC is estimated at $1,078 (2009$) and the
 DMC of non-battery components is estimated at $2,938 (2009$). The battery pack DMC is
 considered to be applicable for the 2025MY while the non-battery pack DMC would be
 applicable for the 2012MY.  The agencies consider the 2-mode battery packs to be on the
 steep portion of the learning curve during the 2017-2025 timeframe. The agencies have
 applied a highl complexity ICM of 1.56 through 2018 then 1.35 thereafter.  For 2-mode non-
 battery components, the agencies consider them to be on the flat portion of the learning curve
 in the 2017-2025 timeframe and have applied a highl complexity ICM of 1.56 through 2018
 then 1.35 thereafter. The resultant 2-mode hybrid costs are presented in Table 3-47.

                          Table 3-47 Costs for 2-Mode Hybrids (2009$)
Cost type
Vehicle Class
2017
2018
2019
2020
2021
2022
2023
2024
2025
Battery-pack
DMC
TC
Minivan/Minivan-
to wing/Large
truck
Minivan/Minivan-
to wing/Large
truck
$2,105
$2,779
$1,684
$2,331
$1,684
$2,076
$1,348
$1,728
$1,348
$1,728
$1,348
$1,728
$1,348
$1,728
$1,348
$1,728
$1,078
$1,450
Non-battery pack components
DMC
1C
TC
Minivan/Minivan-
to wing/Large
truck
Minivan/Minivan-
to wing/Large
truck
Minivan/Minivan-
to wing/Large
truck
$2,549
$1,631
$4,180
$2,498
$1,628
$4,126
$2,448
$999
$3,448
$2,399
$998
$3,397
$2,351
$996
$3,348
$2,304
$995
$3,299
$2,258
$993
$3,252
$2,213
$992
$3,205
$2,169
$990
$3,159
Battery-pack and non-battery pack components
TC
Minivan/Minivan-
to wing/Large
truck
$6,960
$6,457
$5,524
$5,126
$5,076
$5,027
$4,980
$4,933
$4,610
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
 3.4.3.6.3 P2 Hybrid

        A P2 hybrid is a vehicle with an electric drive motor coupled to the engine crankshaft
 via a clutch. The engine and the drive motor are mechanically independent of each other,
 allowing the engine or motor to power the vehicle separately or combined.  This is similar to
                                             3-117

-------
                                    Technologies Considered in the Agencies' Analysis
the Honda HEV architecture with the exception of the added clutch, and larger batteries and
motors. Examples of this include the Hyundai Sonata HEV and Infiniti M35h. The agencies
believe that the P2 is an example of a "strong" hybrid technology that is typical of what we
will see in the timeframe of this rule. The agencies could have equally chosen the power-split
architecture as the representative HEV architecture. These two HEV's have similar average
effectiveness values (combined city and highway fuel economy), though the P2 systems may
have lower cost due to the lower number of parts and complexity.

       The P2 Hybrid is a newly emerging hybrid technology that uses a transmission
integrated electric motor placed between the engine and a gearbox or CVT, much like the
IMA system described above except with a wet or dry separation clutch which is used to
decouple the motor/transmission from the engine. In addition, a P2 Hybrid would typically be
equipped with a larger electric machine.  Disengaging the clutch allows all-electric operation
and more efficient brake-energy recovery. Engaging the clutch allows efficient coupling of
the engine and electric motor and, when combined with a DCT transmission, reduces gear-
train losses relative to PSHEV or 2MHEV systems.

       For purposes of this rulemaking analysis, the agencies are assuming that P2 hybrids
will become the dominant technology in the MYs 2017-2025 timeframe, replacing costlier
power-split or 2-mode architectures while providing substantially similar efficiency
improvement.  At the present time, P2 hybrids are relatively new to the market and the
agencies have not attempted to quantify any measurable performance differential between
these technologies. As mentioned, the 2011 Hyundai Sonata, 2011 Volkswagen Touareg
Hybrid, the 2011 Porsche S Hybrid, and the 2012 Infiniti M35 Hybrid are examples of P2
hybrids currently in production and available to consumers.  The agencies are aware of some
articles in trade journals, newspapers and other reviews that some first generation P2 hybrid
vehicles with planetary gear transmissions have trade-offs in NVH and drivability - though
these reviews do not cover all of the P2 systems available today, and a number of reviews are
very positive with respect to NVH and drivability. The agencies recognize that manufacturers
will have several years to test, develop and improve P2 technology in the years before 2017.
We expect that manufacturers will address any perceived integration issues in early
production models. However, we believe it is important to continue to monitor development
of P2 hybrids and market acceptance of this technology. We will continue to  gather
information on these issues and consider them as part of the mid-term evaluation.

       The agencies request comment regarding the potential of P2 hybrids to overcome
these issues or others, and we specifically seek comment from automakers developing and
considering P2 technology on whether they believe these to be significant impediments to
deployment and how they may be addressed.

       The effectiveness used for vehicle packages with the P2-hybrid configuration within
this analysis reflects a conservative estimate of system performance.  Vehicle  simulation
modeling of technology packages using the P-2 hybrid has recently been completed under a
contract with Ricardo Engineering. The agencies have updated the effectiveness of hybrid
                                           3-118

-------
                                      Technologies Considered in the Agencies' Analysis
electric vehicle packages using the new Ricardo vehicle simulation modeling runs for this
analysis.

       Due to the lower cost and comparative effectiveness of P2 hybrid in relative to other
strong hybrid technologies, such as power-split hybrid and 2-mode hybrid, the agencies
assume P2 hybrid application for all vehicle sub-classes in this NPRM analysis and increased
HEV effectiveness by approximately 2% comparing to 2012-2016 light duty GHG/CAFE
final rule based on published data for new HEVs that have entered into production, such as
2011  Hyundai Sonata hybrid, 2010 Hyundai Elantra LPI HEV (Korean market only), 2011
Infiniti G35 Hybrid and 2011 Volkswagen Touareg Hybrid).  In addition, for the Large Car,
Minivan and Small Truck subclasses, the agencies further increased HEV effectiveness by
assuming that towing capacity could be reduced from their current ratinghh to approximately
1,500 pounds for some vehicles in these subclasses without significantly impacting
consumers' need for utility in these vehicles.11 - The agencies believe that consumers for
these vehicles who require higher towing capacity could acquire it by purchasing a vehicle
with a more capable non-hybrid powertrain (as they do today) .jj  Moreover, it is  likely that
some fraction of consumers who purchase the larger engine option do so for purposes of
hauling and acceleration performance,  not just maximum towing.

       A reduction in towing capacity allows greater engine downsizing, which increases
estimated overall HEV system incremental effectiveness by 5  to 10 percent for Large Cars,
Minivans, and Small Trucks, similar to the HEV effectiveness value assumed for Small Cars
and Compact Cars.kk

       Based on the recent Ricardo study, the effectiveness for P2 hybrid used in this NPRM
is 46.2 percent for subcompact and compact passenger cars, 48.6 percent for midsize
passenger car, 49.4 percent for large passenger car, 46.1 percent for small light truck, 45.7
percent for midsize SUV, truck and minivan and 45.1 percent  for large pickup truck.

       The agencies have applied a high complexity ICM to both the battery and non-battery
component costs for P2 hybrid. But for battery for P2 hybrid, the ICM switches  from short
14 Current small SUVs and Minivans have an approximate average towing capacity of 2000 Ibs (without a
towing package), but range from no towing capacity to 3500 pounds.
11 We note that there are some gasoline vehicles in the large car/minivan/small truck segments sold today which
do not have any towing rating.
jj The agencies recognize that assuming that certain consumers will choose to purchase non-hybrid vehicles in
order to obtain their desired towing capacity could lead to some increase in fuel consumption and CO2 emissions
as compared to assuming that towing capacity is maintained for hybrid vehicles across the board. However, the
agencies think it likely that the net improvement in fuel consumption and CO2 emissions due to the increased
numbers of hybrids available for consumers to choose will offset any potential increase in fuel consumption and
CO2 emissions resulting from consumers selecting the higher-performance non-hybrid powertrain vehicles.
^ The effectiveness of HEVs for heavier vehicles  which require conventional towing capabilities is markedly
less because the rated power of the 1C engine must be similar to its non-hybrid brethren. As such, there is less
opportunity for downsizing with these vehicles.

                                             3-119

-------
                                    Technologies Considered in the Agencies' Analysis
term value of 1.56 to long term value of 1.35 at 2024 while for the non-battery component the
switch happens at 2018.

      The costs for P2 hybrids without mass reduction as used in the Volpe model are listed
in Table 3-48. NHTSA accounts the cost impact from the interaction between mass reduction
and sizing of the electrification system (battery and non-battery  system) as a cost synergy as
described in section 3.4.3.9. Estimated costs for P2 HEVs with  mass reduction as used in the
OMEGA model are presented in Sections 3.4.3.9 and 3.4.3.10 below.
                                           3-120

-------
                                    Technologies Considered in the Agencies' Analysis
Table 3-48 NHTSA Costs for P2 Hybrid Applied in Volpe Model without Mass Reduction (2009$)
Tech.
Battery
Battery
Battery
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Non-battery
Non-battery
Non-battery
Battery
Battery
Battery
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Non-battery
Non-battery
Non-battery
Battery
Battery
Battery
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Non-battery
Non-battery
Non-battery
Cost type
DMC

DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
NHTSA Vehicle Class
Subcompact PC/Perf PC
Compact PC/Perf PC
Midsize PC/Perf PC
Large PC/Perf PC
Midsize LT
Mini van
Small LT
Large LT
Subcompact PC/Perf PC
Compact PC/Perf PC
Midsize PC/Perf PC
Large PC/Perf PC
Midsize LT
Mini van
Small LT
Large LT
Subcompact PC/Perf PC
Compact PC/Perf PC
Midsize PC/Perf PC
Large PC/Perf PC
Midsize LT
Mini van
Small LT
Large LT
Subcompact PC/Perf PC
Compact PC/Perf PC
Midsize PC/Perf PC
Large PC/Perf PC
Midsize LT
Mini van
Small LT
Large LT
Subcompact PC/Perf PC
Compact PC/Perf PC
Midsize PC/Perf PC
Large PC/Perf PC
Midsize LT
Mini van
Small LT
Large LT
Subcompact PC/Perf PC
Compact PC/Perf PC
Midsize PC/Perf PC
Large PC/Perf PC
Midsize LT
Mini van
Small LT
Large LT
2017
$716
$758
$864
$929
$822
$964
$1,467
$1,537
$1,775
$1,756
$1,690
$1,803
$404
$427
$487
$523
$463
$543
$939
$983
$1,136
$1,124
$1,081
$1,154
$1,120
$1,185
$1,350
$1,452
$1,285
$1,508
$2,406
$2,520
$2,911
$2,880
$2,772
$2,957
2018
$695
$735
$838
$901
$797
$935
$1,438
$1,506
$1,739
$1,721
$1,656
$1,767
$402
$426
$485
$522
$462
$542
$937
$981
$1,133
$1,121
$1,079
$1,151
$1,097
$1,161
$1,323
$1,423
$1,259
$1,477
$2,375
$2,487
$2,873
$2,843
$2,736
$2,919
2019
$674
$713
$813
$874
$773
$907
$1,409
$1,476
$1,705
$1,687
$1,623
$1,732
$401
$424
$483
$520
$460
$540
$575
$602
$696
$688
$663
$707
$1,075
$1,137
$1,296
$1,394
$1,233
$1,447
$1,984
$2,078
$2,401
$2,375
$2,286
$2,439
2020
$654
$692
$788
$848
$750
$880
$1,381
$1,446
$1,671
$1,653
$1,591
$1,697
$400
$423
$482
$518
$459
$538
$574
$601
$695
$687
$662
$706
$1,053
$1,114
$1,270
$1,366
$1,209
$1,418
$1,955
$2,048
$2,365
$2,340
$2,252
$2,403
2021
$634
$671
$765
$822
$728
$854
$1,353
$1,417
$1,637
$1,620
$1,559
$1,663
$398
$421
$480
$517
$457
$536
$573
$601
$694
$686
$661
$705
$1,032
$1,092
$1,245
$1,339
$1,185
$1,390
$1,927
$2,018
$2,331
$2,306
$2,220
$2,368
2022
$615
$651
$742
$798
$706
$828
$1,326
$1,389
$1,604
$1,588
$1,528
$1,630
$397
$420
$479
$515
$456
$535
$572
$600
$693
$685
$660
$704
$1,012
$1,071
$1,220
$1,313
$1,162
$1,363
$1,899
$1,989
$2,297
$2,273
$2,187
$2,334
2023
$597
$631
$719
$774
$685
$803
$1,300
$1,361
$1,572
$1,556
$1,497
$1,597
$396
$419
$477
$513
$454
$533
$572
$599
$692
$684
$659
$703
$992
$1,050
$1,197
$1,287
$1,139
$1,336
$1,871
$1,960
$2,264
$2,240
$2,156
$2,300
2024
$579
$612
$698
$750
$664
$779
$1,274
$1,334
$1,541
$1,525
$1,467
$1,566
$395
$418
$476
$512
$453
$531
$571
$598
$691
$683
$658
$702
$973
$1,030
$1,174
$1,262
$1,117
$1,311
$1,845
$1,932
$2,232
$2,208
$2,125
$2,267
2025
$561
$594
$677
$728
$644
$756
$1,248
$1,307
$1,510
$1,494
$1,438
$1,534
$242
$257
$292
$314
$278
$326
$570
$597
$690
$682
$657
$701
$804
$850
$969
$1,042
$922
$1,082
$1,818
$1,904
$2,200
$2,177
$2,095
$2,235
                        DMC=Direct manufacturing cost; IC=Indirect cost; TC=Totalcost
                                           3-121

-------
                                     Technologies Considered in the Agencies' Analysis
3.4.3.6.4 Plug-In Hybrid

       Plug-In Hybrid Electric Vehicles (PHEVs) are very similar to Hybrid Electric
Vehicles, but with three significant functional differences. The first is the addition of a means
to charge the battery pack from an outside source of electricity (e.g., the electric grid).
Second, a PHEV would have a larger battery pack with more energy storage, and a greater
capability to be discharged. Finally, a PHEV would have a control system that allows the
battery pack to be significantly depleted during normal operation.

       Table 3-49 below, illustrates how PHEVs compare functionally to both hybrid electric
vehicles (HEV) and electric vehicles (EV). These characteristics can change significantly
within each vehicle class/subclass, so this is simply meant as an illustration of the general
characteristics. In reality, the design options are so varied that all these vehicles exist on a
continuum with HEVs on one end and EVs on the other.

                  Table 3-49 Conventional, HEVs, PHEVs, and EVs Compared

Attribute
Drive Power
Engine Size
Electric Range
Battery Charging
Increasing Electrification
Conventional
Engine
Full Size
None
None
HEV
Blended
Engine/Electric
Full Size or Smaller
None to Very Short
On-Board
PHEV
Blended
Engine/Electric
Smaller or Much
Smaller
Short to Medium
Grid/On-Board
EV
Electric
No Engine
Medium to Long
Grid Only
       Deriving some of their propulsion energy from the electric grid provides several
advantages for PHEVs.  PHEVs offer a significant opportunity to replace petroleum used for
transportation energy with domestically-produced electricity. The reduction in petroleum
usage does, of course, depend on the amount of electric drive the vehicle is capable of under
its duty cycle. PHEVs also provide electric utilities the possibility to increase electric
generation during "off-peak" periods overnight when there is excess generation capacity and
electricity prices are lower. Utilities like to increase this "base load" because it increases
overall system efficiency and lowers average costs. PHEVs can lower localized emissions of
criteria pollutants and air toxics especially in urban areas by operating on electric power. The
emissions from the power generation occur outside the urban area at the power generation
plant which provides health benefits for residents of the more densely populated urban areas
by moving emissions of ozone precursors out of the urban air shed.  Unlike most other
alternative fuel  technologies, PHEVs can initially use an existing infrastructure for refueling
(charging and liquid refueling) so investments in infrastructure may be reduced.

       In analyzing the impacts of grid-connected vehicles like PHEVs and EVs, the
emissions from the electrical generation can be accounted  for if a full upstream and
downstream analysis is desired. While this issue is being studied on an on-going basis,
                                            3-122

-------
                                    Technologies Considered in the Agencies' Analysis
upstream COi emissions are not unique to grid-connected technologies and so are not
included in this analysis.  Sec II of the Preamble has more information on upstream emissions.

       PHEVs will be considerably more costly than conventional vehicles and some other
advanced technologies due to the fact that PHEVs require both conventional internal
combustion engine and electrical driving system and the larger expensive battery pack. To
take full advantage of their capability, consumers would have to be willing to charge the
vehicles during electricity off-peak hours during the night, and would need access to electric
power where they park their vehicles. For many urban dwellers who may park on the street,
or in private or public lots or garages, charging may not be practical.  Charging may be
possible at an owner's place of work, but that would increase grid loading during peak hours
which would eliminate some of the benefits to utilities of off-peak charging versus on-peak.
Oil savings will still be the same in this case assuming the vehicle can be charged fully.

       The effectiveness potential of PHEVs depends on many factors, the most important
being the energy storage capacity designed into the battery pack. To estimate the fuel
consumption and tailpipe CO2 reduction potential of PHEVs, EPA has developed an in-house
vehicle energy model (PEREGRIN) to  estimate the fuel consumption/CO2 emissions
reductions of PHEVs. This model is based on the PERE (Physical Emission Rate Estimator)
physics-based model used as a fuel consumption input for EPA's MOVES mobile source
emissions model.

       How EPA Estimates PHEV Effectiveness

       The PHEV small car, large car, minivan and small trucks were modeled using
parameters from a midsize car similar to today's hybrids and scaled to each vehicle's weight.
The large truck PHEV was modeled separately assuming no engine downsizing.  PHEVs can
have a wide variation in the All Electric Range (AER) that they offer. Some PHEVs are of
the "blended" type where the engine is on during most of the vehicle operation, but the
proportion of electric energy that is used to propel the vehicle is significantly higher than that
used in a PSHEV or 2MHEV. Each PHEV was modeled with enough battery capacity for a
20-mile-equivalent AER and a power requirement to provide similar performance to a hybrid
vehicle. 20 miles was selected because it offers a good compromise for vehicle performance,
weight, battery packaging and cost. Given expected near-term battery capability, a 20 mile
range represents the likely capability that will be  seen in PHEVs in the near-to-mid term.

       To calculate the total energy use of a PHEV, the PHEV can be thought of as operating
in two distinct modes, electric (EV) mode, and hybrid (HEV) mode. At the tailpipe, the CO2
emissions during EV operation are zero.  The EV mode fuel economy can then be combined
with the HEV mode fuel economy using the Utility Factor calculation in SAE J1711 to
determine a total MPG value for the vehicle. (See Table 3-50)

           Table 3-50 Sample Calculation of PHEV Gasoline-Equivalent CO2 Reduction

| EV energy comb (0.55 city / 0.45 hwy)
Midsize Car
0.252 kwh/mi
Large Truck
0.429 kwh/mi |
                                          3-123

-------
                                    Technologies Considered in the Agencies' Analysis
EV range (from PEREGRIN)
SAEJ1711 utility factor
HEV mode comb FE (0.55 city / 0.45 hwy)
Total UF-adjusted FE (UF*FCEV + (1-UF)*FCHEV)
Baseline FE
Percent FE gain
Percent CO2 reduction
20 miles
0.30
49.1 mpg
70.1 mpg
29.3 mpg
139%
-58%
20 miles
0.30
25.6 mpg
36.6 mpg
19.2 mpg
90%
-47%
       Calculating a total fuel consumption and tailpipe COi reduction based on model
outputs and the Utility Factor calculations results in a 58 percent reduction for small cars,
large cars, minivans, and small trucks. For large trucks, the result is a 47 percent reduction.
The lower improvement is due to less engine downsizing in the large truck class.

       How NHTSA Estimates PHEV Effectiveness

       For CAFE calculation, PHEV is treated as a dual fuel vehicle. NHTSA needs to
consider using dual fuel vehicle calculation for PHEV and uses a petroleum equivalency
factor as stated in 49 U.S.C. 32904 and 32905.

       When deciding PHEV and EV effectiveness, NHTSA referenced the fuel economy of
3 pairs of vehicles for which NHTSA has fuel economy data in the CAFE database. These
three vehicles pairs are MiniE electric vehicle versus gasoline powered Mini with automatic
transmission, Tesla Roadster electric vehicle versus gasoline powered rear-wheel-drive Lotus
Elise Sedan with a 6-speed manual transmission, and Nissan Leaf electric vehicle versus
gasoline powered Nissan Sentra with automatic transmission. The fuel economy and fuel
consumption for the first two pairs are shown in Table 3-51. Nissan Leaf information is used
but not shown in the table because it is confidential information. Because technologies are
applied in the CAFE model in an incremental manner, the effectiveness for each technology is
incremental to the previous technology. In the electrification decision tree of the CAFE
model, the order of technology selection starts from gasoline only powertrain, then moves to
strong hybrid, to plug-in hybrid electric vehicle, and finally to electric vehicle. So the
incremental effectiveness for each step has to be defined.
                     Table 3-51 EV Fuel Economy and Fuel Consumption
104 Mile Range (Mini Website)
MiniE (mpg)
Mini Gas ATX (mpg)
Fuel Economy
[mpg]
342.4
38.6
Fuel Consumption
[gpm]
0.0029206
0.0259067
                     227 Mile Range (EPA)
Tesla Roadster
Lotus Elise Sedan M6 RWD
346.8
30.6
0.0028835
0.0326797
                                           3-124

-------
                                   Technologies Considered in the Agencies' Analysis
      In order to calculate the effectiveness of PHEV for purposes of a CAFE standard, fuel
economy for strong hybrid electric vehicle (SHEV) is calculated first using the incremental
effectiveness of strong hybrid from LPM model which is around 46 percent. For an example,
the derived fuel economy for SHEV based on Mini Gas ATX is 71.7 mpg. Then the fuel
economy from gasoline source for PHEV is assumed to be the same as SHEV fuel economy,
i.e. 71.7 mpg in the case of Mini E. The petroleum equivalent fuel economy from the
electricity source is set to be the same as the EV fuel economy, i.e. 342.4 mpg in the case of
Mini E. The combined fuel economy for PHEV is calculated using the 50-50 weighting factor
as follows.

      PHEV Combined Fuel Economy
                     Gasoline FE Weighing Factor   Electric FE Weighing Factor
                        Gasoline Fuel Economy           EV Fuel Economy
                               0.5    0.5
                              71.7 + 342.4

      NHTSA decided to use a 50-50 weighing factor in the calculation above by modeling
a 30-mile range PHEV. According to SAE standard J171 1, a vehicle with 27.4 to 28.2 mile
charge depleting range has a 0.5 utility factor. This utility factor value of 0.5 is equivalent to
50-50 weighting for dual fuel vehicle calculation. In the NPRM analysis, EPA models a 20-
mile range and a 40-mile range PHEV.

      The incremental fuel consumption reduction for PHEV  is then calculated in relative to
strong HEV. Using the example of Mini E, the incremental fuel consumption reduction for
PHEV relative to SHEV is 39.5 percent as shown below.
         Incremental Fuel Consumption Reduction for PHEV
                                   ___
                          PHEV Fuel Economy  SHEV Fuel Economy}
                                    SHEV Fuel Economy
                          118.6   7l.7J_xWO% = _39_5o/0
                               1
                             7T7

      Table 3-52 lists the incremental effectiveness calculation for two pairs of vehicles,
MiniE and Tesla Roaster. Incremental fuel consumption calculation for PHEV based on
Nissan Leaf is not shown in Table 3-52 due to confidentiality of the fuel economy rating. The
derived incremental effectiveness for Nissan Leaf is 40.6%.  The average incremental
effectiveness of these three pairs of vehicles is 40.65 percent which is used in CAFE
modeling.
                                         3-125

-------
                                   Technologies Considered in the Agencies' Analysis
         Table 3-52 Incremental Effectiveness Calculation for purposes of CAFE modeling
          Mini E

Combined Fuel Economy [mpg]
Gasoline Fuel Economy [mpg]
Electric Petroleum Equivalent Fuel Economy [mpg]
Combined Fuel Consumption[gpm]
Gasoline Fuel Consumption [gpm]
Incremental Combined Fuel Consumption [%]
Gasoline Weighing Factor[%]
Electricity Weighing Factor [%]
Gasoline
38.6







SHEV2
71.7
71.7

0.0139414
0.0139414



PHEV1
118.6
71.7
342.4
0.0084310
0.0139414
39.5%
50%
50%
EV1
342.4


0.0029206

65.4%
0%
100%
          Tesla

Combined Fuel Economy [mpg]
Gasoline Fuel Economy [mpg]
Electric Petroleum Equivalent Fuel Economy [mpg]
Combined Fuel Consumption[gpm]
Gasoline Fuel Consumption [gpm]
Incremental Combined Fuel Consumption [%]
Gasoline Weighing Factor[%]
Electricity Weighing Factor [%]
Gasoline
30.6







SHEV2
56.7
56.7

0.017647
0.017647



PHEV1
97.4
56.7
346.8
0.0102653
0.0176471
41.8%
50%
50%
EV1
346.8


0.0028835

71.9%
0%
100%
      Once the fuel economy of the PHEV is calculated, the effectiveness of PHEV
incremental to EV can be calculated similarly using the formula below.
           Incremental Fuel Consumption Improvement for EV
                           EV Fuel Economy   PHEV Fuel Economy-'
                                                                :rlOO%
                                    PHEV Fuel Economy
      The average effectiveness for the three pairs of vehicles of 68.54% is used in CAFE
modeling.

      The cost of PHEV consists of three parts, the cost for battery, the cost for non-battery
systems and the cost for charger and the labor to install it. Costs for PHEVs without mass
reduction as used in the Volpe model are listed in Table 3-53 to Table 3-55. NHTSA accounts
the cost impact from the interaction between mass reduction and sizing of the electrification
system (battery and non-battery system) as a cost synergy as described in section 3.4.3.9.
Sections 3.4.3.9 and 3.4.3.10 contain the cost for PHEVs with mass reduction as used in
EPA's OMEGA model. PHEV20 and PHEV40 are sized by EPA with the methodologies
discussed in section 3.4.3.8.

      The battery pack DMCs for PHEV20 and PHEV40 are calculated using ANL's BatPac
model. NHTSA modeled a PHEV 30 for this proposal, for which NHTSA averaged the costs
of PHEV20s and PHEV40s.
                                         3-126

-------
                                    Technologies Considered in the Agencies' Analysis
       The agencies have applied a high complexity ICM to non-battery component cost for
PHEV and PHEV charger, which switch from short term value of 1.56 to long term value of
1.35 at 2018. The agencies applied a higher ICM factor to the battery of PHEV due to the fact
that it a more complex technology. The ICM for PHEV battery switches from short term
value  of 1.77 to long term value of 1.50  at 2024.
   Table 3-53 NHTSA Costs for PHEV20 Applied in the Volpe Model with No Mass Reduction (2009$)
Tech.
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor
Battery
Cost
type
DMC

DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
TC
NHTSA
Vehicle Class
Subcompact
PC/PerfPC
Compact
PC/PerfPC
Midsize
PC/PerfPC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize
PC/Perf PC
Large PC/Perf
PC
All
All
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize
PC/Perf PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize
PC/Perf PC
Large PC/Perf
PC
All
All
Subcompact
PC/Perf PC
Compact
2017
$5,082
$5,363
$6,505
$2,556
$2,820
$3,903
$59
$1,000
$2,186
$2,307
$2,798
$1,635
$1,804
$2,497
$19
$0
$7,269
2018
$4,066
$4,291
$5,204
$2,505
$2,764
$3,825
$47
$1,000
$2,112
$2,228
$2,703
$1,632
$1,801
$2,492
$18
$0
$6,177
2019
$4,066
$4,291
$5,204
$2,455
$2,709
$3,749
$47
$1,000
$2,112
$2,228
$2,703
$1,002
$1,106
$1,530
$18
$0
$6,177
2020
$3,253
$3,433
$4,163
$2,406
$2,654
$3,674
$38
$1,000
$2,052
$2,165
$2,626
$1,000
$1,104
$1,528
$17
$0
$5,304
2021
$3,253
$3,433
$4,163
$2,358
$2,601
$3,600
$38
$1,000
$2,052
$2,165
$2,626
$999
$1,102
$1,525
$17
$0
$5,304
2022
$3,253
$3,433
$4,163
$2,311
$2,549
$3,528
$38
$1,000
$2,052
$2,165
$2,626
$997
$1,101
$1,523
$17
$0
$5,304
2023
$3,253
$3,433
$4,163
$2,264
$2,498
$3,458
$38
$1,000
$2,052
$2,165
$2,626
$996
$1,099
$1,521
$17
$0
$5,304
2024
$3,253
$3,433
$4,163
$2,219
$2,448
$3,389
$38
$1,000
$2,052
$2,165
$2,626
$995
$1,097
$1,519
$17
$0
$5,304
2025
$2,602
$2,746
$3,331
$2,175
$2,399
$3,321
$30
$1,000
$1,292
$1,364
$1,654
$993
$1,096
$1,517
$10
$0
$3,894
                                          3-127

-------
                                   Technologies Considered in the Agencies' Analysis

Battery
Battery

Non-battery


Non-battery
Non-battery
Charger
Charger
Labor

TC
TC

TC


TC
TC
TC
TC
PC/PerfPC
Midsize
PC/PerfPC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC

PC/Perf PC
Large PC/Perf
PC
All
All

$7,671
$9,303

$4,191


$4,625
$6,401
$77
$1,000

$6,519
$7,907

$4,137


$4,565
$6,318
$65
$1,000

$6,519
$7,907

$3,457


$3,814
$5,279
$65
$1,000

$5,598
$6,789

$3,406


$3,758
$5,202
$55
$1,000

$5,598
$6,789

$3,357


$3,704
$5,126
$55
$1,000

$5,598
$6,789

$3,308


$3,650
$5,052
$55
$1,000

$5,598
$6,789

$3,260


$3,597
$4,979
$55
$1,000

$5,598
$6,789

$3,214


$3,546
$4,907
$55
$1,000

$4,110
$4,985

$3,168


$3,495
$4,837
$40
$1,000
                  DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
Table 3-54 NHTSA Costs for PHEV40 Applied in the Volpe Model with No Mass Reduction (2009$)
Tech.
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor
Battery
Battery
Battery
Non-battery
Cost
type
DMC

DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
NHTSA
Vehicle Class
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
2017
$7,126
$7,884
$10,140
$2,557
$2,820
$3,902
$357
$1,000
$3,066
$3,392
$4,362
$1,636
2018
$5,701
$6,307
$8,112
$2,506
$2,763
$3,824
$286
$1,000
$2,112
$2,228
$2,703
$1,632
2019
$5,701
$6,307
$8,112
$2,455
$2,708
$3,748
$286
$1,000
$2,112
$2,228
$2,703
$1,002
2020
$4,561
$5,046
$6,490
$2,406
$2,654
$3,673
$229
$1,000
$2,052
$2,165
$2,626
$1,001
2021
$4,561
$5,046
$6,490
$2,358
$2,601
$3,599
$229
$1,000
$2,052
$2,165
$2,626
$999
2022
$4,561
$5,046
$6,490
$2,311
$2,549
$3,527
$229
$1,000
$2,052
$2,165
$2,626
$998
2023
$4,561
$5,046
$6,490
$2,265
$2,498
$3,457
$229
$1,000
$2,052
$2,165
$2,626
$996
2024
$4,561
$5,046
$6,490
$2,220
$2,448
$3,388
$229
$1,000
$2,052
$2,165
$2,626
$995
2025
$3,649
$4,037
$5,192
$2,175
$2,399
$3,320
$183
$1,000
$1,292
$1,364
$1,654
$993
                                          3-128

-------
                                  Technologies Considered in the Agencies' Analysis

Non-battery
Non-battery
Charger
Charger
Labor
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor

1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
PC/PerfPC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All

$1,804
$2,496
$114
$0
$10,191
$11,276
$14,502
$4,192
$4,624
$6,399
$472
$1,000

$1,800
$2,491
$110
$0
$7,812
$8,536
$10,815
$4,138
$4,564
$6,316
$396
$1,000

$1,105
$1,530
$110
$0
$7,812
$8,536
$10,815
$3,458
$3,813
$5,277
$396
$1,000

$1,104
$1,527
$106
$0
$6,612
$7,211
$9,116
$3,407
$3,758
$5,200
$335
$1,000

$1,102
$1,525
$106
$0
$6,612
$7,211
$9,116
$3,357
$3,703
$5,124
$335
$1,000

$1,100
$1,523
$106
$0
$6,612
$7,211
$9,116
$3,309
$3,649
$5,050
$335
$1,000

$1,099
$1,520
$106
$0
$6,612
$7,211
$9,116
$3,261
$3,596
$4,977
$335
$1,000

$1,097
$1,518
$106
$0
$6,612
$7,211
$9,116
$3,214
$3,545
$4,906
$335
$1,000

$1,096
$1,516
$63
$0
$4,941
$5,400
$6,846
$3,168
$3,494
$4,836
$246
$1,000
                   DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
Table 3-55 NHTSA Costs Applied in Volpe Model for PHEV30 with No Mass Reduction (2009$)
Tech.
Battery
Battery
Battery
Non-
battery
Non-
battery
Non-
battery
Charger
Cost
type
DMC

DMC
DMC
DMC
DMC
DMC
NHTSA
Vehicle Class
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
All
2017
$6,104
$6,624
$8,323
$2,556
$2,820
$3,903
$208
2018
$4,883
$5,299
$6,658
$2,505
$2,764
$3,825
$166
2019
$4,883
$5,299
$6,658
$2,455
$2,708
$3,748
$166
2020
$3,907
$4,239
$5,327
$2,406
$2,654
$3,673
$133
2021
$3,907
$4,239
$5,327
$2,358
$2,601
$3,600
$133
2022
$3,907
$4,239
$5,327
$2,311
$2,549
$3,528
$133
2023
$3,907
$4,239
$5,327
$2,265
$2,498
$3,457
$133
2024
$3,907
$4,239
$5,327
$2,219
$2,448
$3,388
$133
2025
$3,125
$3,391
$4,261
$2,175
$2,399
$3,320
$107
                                         3-129

-------
                                     Technologies Considered in the Agencies' Analysis
Charger
Labor
Battery
Battery
Battery
Non-
battery
Non-
battery
Non-
battery
Charger
Charger
Labor
Battery
Battery
Battery
Non-
battery
Non-
battery
Non-
battery
Charger
Charger
Labor
DMC
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
All
Subcompact
PC/PerfPC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
All
All
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
All
All
$1,000
$2,626
$2,849
$3,580
$1,635
$1,804
$2,497
$67
$0
$8,730
$9,473
$11,903
$4,192
$4,624
$6,400
$275
$1,000
$1,000
$2,112
$2,228
$2,703
$1,632
$1,800
$2,492
$64
$0
$6,995
$7,527
$9,361
$4,137
$4,564
$6,317
$230
$1,000
$1,000
$2,112
$2,228
$2,703
$1,002
$1,105
$1,530
$64
$0
$6,995
$7,527
$9,361
$3,457
$3,814
$5,278
$230
$1,000
$1,000
$2,052
$2,165
$2,626
$1,001
$1,104
$1,528
$62
$0
$5,958
$6,404
$7,952
$3,407
$3,758
$5,201
$195
$1,000
$1,000
$2,052
$2,165
$2,626
$999
$1,102
$1,525
$62
$0
$5,958
$6,404
$7,952
$3,357
$3,703
$5,125
$195
$1,000
$1,000
$2,052
$2,165
$2,626
$998
$1,100
$1,523
$62
$0
$5,958
$6,404
$7,952
$3,308
$3,649
$5,051
$195
$1,000
$1,000
$2,052
$2,165
$2,626
$996
$1,099
$1,521
$62
$0
$5,958
$6,404
$7,952
$3,261
$3,597
$4,978
$195
$1,000
$1,000
$2,052
$2,165
$2,626
$995
$1,097
$1,518
$62
$0
$5,958
$6,404
$7,952
$3,214
$3,545
$4,907
$195
$1,000
$1,000
$1,292
$1,364
$1,654
$993
$1,096
$1,516
$37
$0
$4,418
$4,755
$5,915
$3,168
$3,495
$4,837
$143
$1,000
                      DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost

3.4.3.6.5 Electric vehicles

       Electric vehicles (EV) - are vehicles with all-electric drive and with vehicle systems
powered by energy-optimized batteries charged primarily from grid electricity. While the
2016 FRM did not anticipate a significant penetration of EVs, in this analysis, EVs with
several ranges have been included. The GHG effectiveness is unchanged from estimates used
for 2016 model year vehicles in the 2012-2016 final rule which is 100 percent GHG
reduction.  NHTSA uses petroleum equivalency factor in calculating the effectiveness for EVs
as stated in the section above for PHEV.
                                            3-130

-------
                                    Technologies Considered in the Agencies' Analysis
       Once the fuel economy of the PHEV is calculated as shown in the previous section,
the effectiveness of PHEV incremental to EV can be calculated similarly using the formula
below.
         Incremental Fuel Consumption Improvement for EV
                         c,        '
Fuel Economy   PHEV Fuel Economy-1
                                                                   xWQ%
                                    PHEV Fuel Economy
       The average effectiveness for the three pairs of vehicles of 68.54% is used in CAFE
modeling.

       Battery costs assume that battery packs for EV applications will be designed to last for
the full useful life of the vehicle at a useable state of charge equivalent to 80% of the nominal
battery pack capacity. NHTSA applied a 75-mile range EV and a 150-mile range EV in this
NPRM analysis. As this technology is entering the market, the OEM will try to keep the cost
low at the beginning so that there will be more penetration. Due to the high cost of the battery
packs at this early stage of EVs, OEM will try to limit the battery pack size to reduce cost.
Also the early adopters for this technology are normally urban drivers and range anxiety is not
a big concern to them. Therefore NHTSA  applied a 75-mile range EV for early adoption of
this technology in the market. As the technology develops and as the market penetration
increases, OEMs need to help the consumers overcome the range anxiety and longer driving
range will be expected. NHTSA applied 150-mile EV for this broad market adoption of this
technology.

The cost of an EV consists of three parts, cost of battery pack, cost of non-battery systems,
and cost of charger and charger installation labor. The agencies have applied a high
complexity ICM to non-battery component cost for EVs and EV chargers, which switch from
short term value of 1.56 to long term value of 1.35 at 2018. The agencies applied a higher
ICM factor to the battery of EVs due to the fact that it a more complex technology. The ICM
for EV battery switches from short term value of 1.77 to long term value of 1.50 at 2024. The
agencies present costs of EVs in Sections 3.4.3.9 and  3.4.3. 10. The costs of EVs without mass
reduction as applied in Volpe model are listed in Table 3-56  to Table  3-58. NHTSA accounts
the cost impact from the interaction between mass reduction and sizing of electrification
system (battery and non-battery system) as cost synergy as described in section 3.4.3.9.
     Table 3-56 NHTSA Costs Applied in Volpe Model for EV75 with No Mass Reduction (2009$)
Tech.
Battery
Cost
type
DMC
NHTSA
Vehicle Class
Subcompact
PC/PerfPC
Compact
PC/PerfPC
2017
$10,594
2018
$8,475
2019
$8,475
2020
$6,780
2021
$6,780
2022
$6,780
2023
$6,780
2024
$6,780
2025
$5,424
                                          3-131

-------
                  Technologies Considered in the Agencies' Analysis
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor

DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All
$11,500
$14,009
$411
$749
$1,255
$391
$1,000
$4,557
$4,947
$6,027
$317
$577
$966
$114
$0
$15,152
$16,447
$20,036
$728
$1,326
$2,221
$505
$1,000
$9,200
$11,207
$399
$727
$1,217
$313
$1,000
$4,401
$4,778
$5,820
$316
$575
$963
$110
$0
$12,877
$13,978
$17,028
$714
$1,302
$2,180
$422
$1,000
$9,200
$11,207
$387
$705
$1,181
$313
$1,000
$4,401
$4,778
$5,820
$315
$574
$961
$110
$0
$12,877
$13,978
$17,028
$702
$1,278
$2,141
$422
$1,000
$7,360
$8,966
$375
$684
$1,145
$250
$1,000
$4,277
$4,642
$5,655
$314
$572
$958
$106
$0
$11,057
$12,002
$14,621
$689
$1,256
$2,103
$356
$1,000
$7,360
$8,966
$364
$663
$1,111
$250
$1,000
$4,277
$4,642
$5,655
$313
$570
$956
$106
$0
$11,057
$12,002
$14,621
$677
$1,234
$2,066
$356
$1,000
$7,360
$8,966
$353
$643
$1,077
$250
$1,000
$4,277
$4,642
$5,655
$312
$569
$953
$106
$0
$11,057
$12,002
$14,621
$665
$1,212
$2,031
$356
$1,000
$7,360
$8,966
$346
$630
$1,056
$250
$1,000
$4,277
$4,642
$5,655
$312
$568
$952
$106
$0
$11,057
$12,002
$14,621
$658
$1,198
$2,007
$356
$1,000
$7,360
$8,966
$339
$618
$1,035
$250
$1,000
$4,277
$4,642
$5,655
$311
$567
$950
$106
$0
$11,057
$12,002
$14,621
$650
$1,185
$1,985
$356
$1,000
$5,888
$7,173
$332
$605
$1,014
$200
$1,000
$2,694
$2,924
$3,562
$200
$365
$611
$63
$0
$8,118
$8,812
$10,735
$533
$970
$1,625
$263
$1,000
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                         3-132

-------
                                Technologies Considered in the Agencies' Analysis
Table 3-57 NHTSA Costs for EV100 Applied in Volpe Model with No Mass Reduction (2009$)
Tech.
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor
Battery
Battery
Battery
Non-battery
Non-battery
Non-battery
Charger
Charger
Labor
Battery
Battery
Battery
Non-battery
Cost
type
DMC

DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
NHTSA
Vehicle Class
Subcompact
PC/PerfPC
Compact
PC/PerfPC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All
Subcompact
PC/Perf PC
Compact
PC/Perf PC
Midsize PC/Perf
PC
Large PC/Perf
PC
Subcompact
PC/Perf PC
Compact
2017
$12,422
$13,679
$15,823
$413
$748
$1,253
$391
$1,000
$5,344
$5,884
$6,807
$318
$576
$965
$125
$0
$17,766
$19,563
$22,630
$730
2018
$9,938
$10,943
$12,658
$400
$726
$1,216
$313
$1,000
$5,161
$5,683
$6,574
$317
$574
$962
$120
$0
$15,099
$16,626
$19,232
$717
2019
$9,938
$10,943
$12,658
$388
$704
$1,179
$313
$1,000
$5,161
$5,683
$6,574
$316
$573
$959
$120
$0
$15,099
$16,626
$19,232
$704
2020
$7,950
$8,755
$10,127
$377
$683
$1,144
$250
$1,000
$5,015
$5,522
$6,387
$315
$571
$957
$116
$0
$12,965
$14,276
$16,514
$692
2021
$7,950
$8,755
$10,127
$365
$662
$1,109
$250
$1,000
$5,015
$5,522
$6,387
$314
$570
$954
$116
$0
$12,965
$14,276
$16,514
$680
2022
$7,950
$8,755
$10,127
$354
$642
$1,076
$250
$1,000
$5,015
$5,522
$6,387
$313
$568
$952
$116
$0
$12,965
$14,276
$16,514
$668
2023
$7,950
$8,755
$10,127
$347
$630
$1,055
$250
$1,000
$5,015
$5,522
$6,387
$313
$567
$950
$116
$0
$12,965
$14,276
$16,514
$660
2024
$7,950
$8,755
$10,127
$340
$617
$1,033
$250
$1,000
$5,015
$5,522
$6,387
$312
$566
$949
$116
$0
$12,965
$14,276
$16,514
$653
2025
$6,360
$7,004
$8,101
$334
$605
$1,013
$200
$1,000
$3,159
$3,478
$4,023
$201
$365
$611
$69
$0
$9,519
$10,482
$12,125
$535
                                       3-133

-------
                                 Technologies Considered in the Agencies' Analysis

Non-battery
Non-battery
Charger
Charger
Labor

TC
TC
TC
TC
PC/PerfPC
Midsize PC/Perf
PC
Large PC/Perf
PC
All
All

$1,324
$2,218
$516
$1,000

$1,300
$2,178
$432
$1,000

$1,277
$2,139
$432
$1,000

$1,254
$2,101
$366
$1,000

$1,232
$2,064
$366
$1,000

$1,211
$2,028
$366
$1,000

$1,197
$2,005
$366
$1,000

$1,183
$1,982
$366
$1,000

$969
$1,623
$269
$1,000
              DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
Table 3-58 NHTSA Costs for EV150 Applied in Volpe Model with No Mass Reduction (2009$)
Cost
type
DMC

DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
NHTSA Vehicle
Class
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
All
All
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
All
2017
$16,195
$17,944
$21,463
$415
$746
$1,254
$391
$1,000
$6,967
$7,719
$9,233
$320
$574
$966
$125
2018
$12,956
$14,355
$17,170
$403
$723
$1,216
$313
$1,000
$6,728
$7,455
$8,917
$319
$573
$963
$120
2019
$12,956
$14,355
$17,170
$391
$702
$1,180
$313
$1,000
$6,728
$7,455
$8,917
$318
$571
$960
$120
2020
$10,365
$11,484
$13,736
$379
$681
$1,144
$250
$1,000
$6,538
$7,243
$8,664
$317
$570
$958
$116
2021
$10,365
$11,484
$13,736
$368
$660
$1,110
$250
$1,000
$6,538
$7,243
$8,664
$316
$568
$955
$116
2022
$10,365
$11,484
$13,736
$357
$640
$1,077
$250
$1,000
$6,538
$7,243
$8,664
$315
$567
$953
$116
2023
$10,365
$11,484
$13,736
$350
$628
$1,055
$250
$1,000
$6,538
$7,243
$8,664
$315
$566
$951
$116
2024
$10,365
$11,484
$13,736
$343
$615
$1,034
$250
$1,000
$6,538
$7,243
$8,664
$314
$565
$949
$116
2025
$8,292
$9,187
$10,989
$336
$603
$1,014
$200
$1,000
$4,118
$4,562
$5,457
$202
$363
$611
$69
                                        3-134

-------
                                     Technologies Considered in the Agencies' Analysis
1C
TC
TC
TC
TC
TC
TC
TC
TC
All
Subcompact
PC/PerfPC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
Subcompact
PC/Perf PC
Compact PC/Perf
PC
Midsize PC/Perf
PC
Large PC/Perf PC
All
All
$0
$23,162
$25,663
$30,696
$735
$1,320
$2,220
$516
$1,000
$0
$19,684
$21,810
$26,087
$722
$1,296
$2,179
$432
$1,000
$0
$19,684
$21,810
$26,087
$709
$1,273
$2,140
$432
$1,000
$0
$16,902
$18,727
$22,400
$696
$1,250
$2,102
$366
$1,000
$0
$16,902
$18,727
$22,400
$684
$1,228
$2,065
$366
$1,000
$0
$16,902
$18,727
$22,400
$672
$1,207
$2,029
$366
$1,000
$0
$16,902
$18,727
$22,400
$664
$1,193
$2,006
$366
$1,000
$0
$16,902
$18,727
$22,400
$657
$1,180
$1,984
$366
$1,000
$0
$12,410
$13,750
$16,446
$538
$966
$1,625
$269
$1,000
3.4.3.6.6 Fuel cell electric vehicles

       Fuel cell electric vehicles (FCEVs) - utilize a full electric drive platform but consume
electricity generated by an on-board fuel cell and hydrogen fuel.  Fuel cells are electro-
chemical devices that directly convert reactants (hydrogen and oxygen via air) into electricity,
with the potential of achieving more than twice the efficiency of conventional internal
combustion engines. High pressure gaseous hydrogen storage tanks are used by most
automakers for FCEVs that are currently under development. The high pressure tanks are
similar to those used for compressed gas storage in more than 10 million CNG vehicles
worldwide, except that they are designed to operate at a higher pressure (350 bar or 700 bar
vs. 250 bar for CNG).  Due to the uncertainty of the future availability for this technology,
FCEVs were not included in any OMEGA or Volpe model runs.
       3.4.3.7 Batteries for MHEV, HEV, PHEV and EV Applications

       The design of battery secondary cells can vary considerably between MHEV, HEV,
PHEV and EV applications.

       MHEV batteries: Due to their lower voltage (12-42 VDC) and reduced power and
energy requirements, MHEV systems may continue to use lead-acid batteries even long term
(2017 model year and later). MHEV battery designs differ from those of current starved-
electrolyte (typical maintenance free batteries) or flooded-electrolyte (the older style lead-acid
batteries requiring water "top-off) batteries used for starting, lighting and ignition (SLI) in
                                           3-135

-------
                                     Technologies Considered in the Agencies' Analysis
automotive applications. Standard SLI batteries are primarily designed to provide high-
current for engine start-up and then recharge immediately after startup via the vehicle's
charging system. Deeply discharging a standard SLI battery will greatly shorten its life.
MHEV applications are expected to use:

          •   Extended-cycle-life flooded (ELF) lead-acid batteries
          •   Absorptive glass matt, valve-regulated lead-acid (AGM/VRLA) batteries -or -
          •   Asymmetric lead-acid battery/capacitor hybrids (e.g., flooded ultrabatteries)

MHEV systems using electrolytic double-layer capacitors are also under development and
may provide improved performance and reduced cost in the post-2017 timeframe.

       HEV batteries: HEV applications operate in a narrow, short-cycling, charge-
sustaining state of charge (SOC). Energy capacity in HEV applications  is somewhat limited
by the ability of the battery and power electronics to accept charge and by space and weight
constraints within the vehicle design. HEV battery designs tend to be optimized for high
power density rather than high energy density, with thinner cathode and anode layers and
more numerous current collectors and separators (Figure 3-20).

       EV batteries:  EV batteries tend to be optimized for high energy density and are
considerably larger and heavier than HEV batteries in order to provide sufficient energy
capacity. EV battery cells tend to have thicker cathode and anode layers and fewer collectors
and separators  than HEV cells.  This reduced the specific cost on a per-kW-hr basis for EV
battery cells relative to HEV battery cells.

       PHEV batteries:  PHEV battery designs are intermediate between power-optimized
HEV and energy-optimized EV battery cell designs. PHEV batteries must provide both
charge depleting operation similar to an EV and charge sustaining operation similar to an
HEV. Unlike HEV applications, charge-sustaining operation with PHEVs occurs at a
relatively low battery state of charge (SOC) which can pose a significant challenge with
respect to attaining acceptable battery cycle life.  In the case of the GM Volt, this limits
charge depleting operation to a minimum SOC of approximately 30 percent.55 An alternative
approach for PHEV applications that has potential to allow extension of charge depletion to a
lower battery SOC is using energy-optimized lithium-ion batteries for charge depleting
operation in combination with the use of supercapacitors for charge sustaining operation.56
                                           3-136

-------
                                     Technologies Considered in the Agencies' Analysis
               Figure 3-20: Schematic representation of power and energy optimized

                               prismatic-layered battery cells
                Collector (-)

                Cathode (-)

                Separator

                Anode (+)

                Collector (+)
                                 HEV Power-optimized Battery Cell
                                  EV Energy-optimized Battery Cell
       Power-split hybrid vehicles from Toyota, Ford and Nissan (which uses the Toyota
system under license), integrated motor assist hybrid vehicles from Honda and the GM 2-
mode hybrid vehicles currently use nickel-metal hydride (NiMH) batteries. Lithium-ion (Li-
ion) batteries offer the potential to approximately double both the energy and power density
relative to current NiMH batteries, enabling much more electrical-energy-intensive
automotive applications such as PHEVs and EVs.

       Li-ion batteries for high-volume automotive applications differ substantially from
those used in consumer electronics applications with respect to cathode chemistry,
construction and cell size.  Li-ion battery designs currently in production by CPI (LG-Chem)
for the GM Volt PHEV and by AESC and  GS-Yuasa (respectively) for the Nissan Leaf and
Mitsubishi i-Miev use large-format, layered-prismatic cells assembled into battery modules.
The modules are then combined into battery packs.

       Two families of cathode chemistries are used in large-format,  automotive Li-ion
batteries currently in production - LiMn2O4-spinel (CPI, GS-Yuasa,  AESC) and LiFePO4
(A123 Systems). Current production batteries typically use graphite anodes. Automotive Li-
ion batteries using lithium nickel manganese cobalt (NMC) oxide cathodes with graphite
anodes are in advanced stages of development for PHEV and EV applications.  The agencies
expect large-format Li-ion batteries to completely replace NiMH batteries for post-2017 HEV
applications.  We also expect that large-format stacked and/or folded  prismatic  Li-ion cell
                                            3-137

-------
                                     Technologies Considered in the Agencies' Analysis
designs will continue to be used for PHEV and EV applications and that NMC/graphite Li-ion
batteries will be a mature technology for 2017-2025 light-duty vehicle applications.
       3.4.3.8 HEV, PHEV and EV System Sizing Methodology

       Battery packs are (and will continue to be) one of the most expensive components for
EVs, PHEVs and HEVs. To obtain reasonable cost estimates for electrified vehicles, it was
therefore important to establish a reliable approach for determining battery attributes for each
vehicle and class.  Both battery energy content ("size") and power rating are key inputs used
to establish costs per ANL's battery costing model. For EVs and PHEVs in particular, battery
size and weight are closely related, and so battery weight must be known as well. The
following section details the steps taken to size a battery for

       a)     EVs and PHEVs (at various all-electric ranges),

       b)     a more simplified separate approach for MHEVs and HEVs.

3.4.3.8.1 Battery Pack Sizing for EVs and PHEVs

       Calculation of required battery pack energy requirements for EVs and PHEVs  is not
straightforward. Because vehicle energy consumption is strongly dependent on weight, and
battery packs are very heavy, the weight of the battery pack itself can change the energy
required to move the vehicle. As vehicle energy consumption increases, the battery size must
increase for a given range (in the case of EVs and PHEVs) - as a result, vehicle weight
increases, and per-mile energy consumption increases as well, increasing the battery size, and
so on.

       EPA built spreadsheets to estimate the required battery size for each vehicle and class
(reference here?) Listed below are the steps EPA has taken in these spreadsheets to estimate
not only battery size, but associated weight for EVs and PHEVs of varying ranges and
designs.

          1. Establish baseline FE/energy consumption
          2. Assume nominal weight of electrified vehicle (based on weight reduction
             target)
          3. Calculate vehicle energy demand at this target weight
          4. Calculate required battery energy
          5. Calculate actual battery and vehicle weight
          6. Do vehicle weight and battery size match estimated values?
       Iterate steps 2-6 until assumed weight reduction target (and nominal vehicle weight)
reconciles with required battery size and calculated weight of vehicle.

                                           3-138

-------
                                      Technologies Considered in the Agencies' Analysis
       Baseline vehicle energy consumption is estimated based on a fitted trendline for FE
vs. inertia weight, or ETW (from FE Trends data for 2008 MY vehicles, table M-80) and
converting to Wh/mi. It is shown in Figure 3-21.
                    2008 Fuel Economy vs. Inertia Weight
                     (source: Fuel Economy Trends Report, Table M-80)
         a.
         si

         J
                                    y - O.OOOOOlSOx2 - 0.02194637X + 85.81284974
                                           Rz = 0.99565332
                                 3000     4000

                                 Inertia wt(lbs)
       Figure 3-21: Average fuel economy based on inertia weight (ETW) from FE Trends data

       Then, fuel economy was converted into energy consumption (assuming 33700 Wh
energy in 1 gallon of gasoline) and used to populate a range of test weights between 2000 and
6000 Ibs. A linear trendline was used to fit this curve and then applied for estimating generic
energy consumption for baseline vehicles  of a given ETW, and is shown below in Figure
3-22.
                                             3-139

-------
                                      Technologies Considered in the Agencies' Analysis
                 2008 Energy Consumption vs. Inertia Weight
                                 3000     4000

                                 Inertia wt(lbs)
            Figure 3-22: Equivalent energy consumption (in Wh/mi) for baseline vehicles

       To calculate battery pack size, the electrified vehicle weight must first be known; to
calculate vehicle weight, the battery pack size must first be known. This circular reference
required an iterative solution. EPA assumed a target vehicle glider (a rolling chassis with no
powertrain) weight reduction and applied that to the baseline curb weight. The resulting
nominal vehicle weight was then used to calculate the vehicle energy demand. To calculate
the energy demand (efficiency) of an electric vehicle in Wh/mi, the following information
was needed:

       •  Baseline energy consumption / mpg

       •  Efficiency (r|) improvement of electric vehicle

       •  Change in road loads

       In Table 3-59 below, the following definitions apply:

       •  Brake eff (brake efficiency) - the %  amount of chemical fuel energy converted to
          energy at the engine crankshaft (or, for batteries, the amount of stored electrical
          energy converted to shaft energy entering the transmission)

       •  D/L eff (driveline efficiency) - the % of the brake energy entering the transmission
          delivered through the driveline to the wheels

       •  Wheel eff (wheel efficiency) - the product of brake and driveline efficiency
                                            3-140

-------
                                     Technologies Considered in the Agencies' Analysis
          Cycle eff (cycle efficiency) - the % of energy delivered to the wheels used to
          overcome road loads and power the vehicle (it does not include energy lost as
          braking heat)

          Vehicle efficiency - the product of wheel and cycle efficiency

          Road loads - the amount of resistant energy the vehicle must overcome during a
          city/highway test. Composed of vehicle weight (inertia), aerodynamic drag and
          rolling resistance

                   Table 3-59: EV efficiency and energy demand calculations


Overall EV efficiency calculations,

Class Brake eff
Baseline gas ICE 24%
Subcompact 85%
Small car 85%
Large car 85%
Small Truck 85%
Minivan 85%
Track 85%

D/Leff
81%
93%
93%
93%
93%
93%
93%




by vehicle class

Wheel eff
20%
79%
79%
79%
79%
79%
79%

Cycle eff'
77%
97%
97%
97%
97%
97%
97%
Vehicle
efficiency
15%
77%
77%
77%
77%
77%
77%
Road Energy fiiergy
Loads Reduction Efficiency
100% Increase
91%
91%
91%
91%
91%
91%
82%
82%
82%
82%
82%
82%
464%
464%
464%
464%
464%
464%
IW-based
base ICE
nominal
mpgge

37
32
26
26
24
21
Base
FTP
Onroad
fuel energy fuel energy fuel energy
reqd
W-hr/mi

911
1060
1279
1314
1401
1597
reqd
W-hr/mi

161
188
227
233
248
283
reqd
W-hr/mi

230
268
324
333
355
404
       The energy efficiency of a baseline vehicle (around 15%), as indicated in the table
above, was estimated using efficiency terms derived from EPA's lumped parameter model
(engine/battery brake efficiency, driveline efficiency, cycle efficiency and road load ratio to
baseline).  To calculate the energy consumption of an EV (or PHEV in charge-depleting
mode), the following assumptions were made:

       •  "Brake" efficiency (for an EV, the efficiency of converting battery energy to
          tractive energy at the transmission input shaft) was estimated at 85% - assuming,
          roughly a 95% efficiency for the battery, motor, and power electronics,
          respectively.

       •  The driveline efficiency (including the transmission) was comparable to the value
          calculated by the lumped parameter model for an advanced 6-speed dual-clutch
          transmission at 93%.

       •  The cycle efficiency assumes regenerative braking where 95% recoverable braking
          energy is recaptured. As a result, most of the energy delivered to the wheels is
          used to overcome road loads.
                                           3-141

-------
                                    Technologies Considered in the Agencies' Analysis
       •   The road loads were based on the weight reduction of the vehicle. In the case of a
          100 mile EV with a 10% weight reduction, road loads (as calculated by the LP
          model) are reduced to 91% of the baseline vehicle11.

          The energy consumption of the EV includes ratio of the roadloads of the EV to the
          baseline vehicle, and the ratio of the efficiency of the EV compared to the baseline
          vehicle. It is expressed mathematically as shown below in Equation 3-1:  EV
          energy consumption

Equation 3-1: EV energy consumption

                  f..n  ,  .,    „            i%Roadloadnew  rivehicle old \
           EEV FTp(Wh/mi) = Ebaseline FTP * —-—-—-— *	~-—
                                           \%Roadloadold   r]vehicle_newj

       In the table 3-x, the baseline energy required (in Wh/mi) is in the column labeled
"Base fuel energy reqd". The energy required for each vehicle class EV over the FTP is in the
column "FTP fuel energy reqd W-hr/mi" and incorporates the equation above. This energy
rate refers to the laboratory or unadjusted test cycle value, as opposed to a real-world
"onroad" value. EPA assumes a 30% fuel economy shortfall, based loosely on the 5-cycle
Fuel Economy Labeling Rule (year) which is directionally correct for electrified vehicles.
This corresponds to an increase in fuel consumption of 43%. Applying this 43% increase
gives the onroad energy consumption values for EVs as shown in the far right column of the
previous table. From this value, one can determine an appropriate battery pack size for the
vehicle.

       The required battery energy for EVs equals the onroad energy consumption,
multiplied by the desired range, divided by the useful state-of-charge window of the battery.
It is calculated as follows in Equation 3-2

Equation 3-2: Required battery pack energy (size) for EVs

                                   P       c    \ \f T ciYi QP CTYI /^
                                    OTLT^OCLCt \. -vy* / J       c_y  V  J
                       sp(Wh)=	fe	
       Assumed usable SOC (battery state-of-charge) windows were 80% for EVs (10-90%)
and 70% for PHEVs (15%-85%).  The battery pack sizes are listed in orange in Table 3-60 for
the 100-mile EV case and show both the onroad energy consumption ("EV adj Wh/mi"
column) and the nominal battery energy content or "battery pack size".
1 Included in this example road load calculation is a 10% reduction in rolling resistance and aerodynamic drag.

                                          3-142

-------
                                    Technologies Considered in the Agencies' Analysis
              Table 3-60: Battery pack sizes for 100-mile EV based on inertia weight

Category
Subcompact
Small car
Large car
Small Truck
Minivan
Truck
BASELINE
curb wt
Ibs
2628
3118
3751
3849
4087
4646
Inertia
wt
Ibs
2928
3418
4051
4149
4387
4946
EV EV 100 mi
unadj
Wh/mi
161
188
227
233
248
283
adj
Wh/mi
230
268
324
333
355
404
batt pack
size kWh
28.8
33.5
40.5
41.6
44.3
50.5
       EPA used the following formula to determine weight of an EV (Equation 3-3):

Equation 3-3: EV weight calculation
W   = W    —
  EV
                                      — W
                                          1CE_powertrain
       Any weight reduction technology was applied only to the glider (baseline vehicle
absent powertrain) as defined in Equation 3-4.

       Equation 3-4: Weight reduction of the glider

                              = %WR * (Wbase - WICE_powertrain)
       In the case of a PHEV's, it was assumed that the base ICE powertrain remains so it is
not deducted; the proper equation for PHEVs is shown in Equation 3-5.

       Equation 3-5: Weight calculation for PHEV

                      WPHEV
              = Wbase - WRglider + Wf
electric _drive
       Listed in Table 3-61 are the assumed baseline ICE-powertrain weights, by vehicle
class:
                                           3-143

-------
                                      Technologies Considered in the Agencies' Analysis
                 Table 3-61: Baseline ICE-powertrain weight assumptions, by class
ICE powertrain weight estimates
Class
Subcompact
Small car
Large car
Small Truck
Minivan
Truck
Engine
250
300
375
300
400
550
Trans (diff
not included)
125
150
175
150
200
200
Fuel system
(50% fill)
50
60
70
60
80
100
Engine
mounts/NVH
treatments
25
25
25
25
25
25
Exhaust
20
25
30
25
30
40
12V battery
25
30
35
30
40
50
Total ICE
powertrain
weight
495
590
710
590
775
965
       EPA then estimated the weight of the electric drive subsystem using the energy
content of the battery pack as an input.  EPA scaled the weight by applying a specific energy
for the electric drive subsystem - including the battery pack, drive motor, wiring, power
electronics, etc.  - of 120 Wh/kg (or 18.33 Ib/kWh).  This specific energy value is based on
adding components to an assumed battery pack specific energy of 150 Wh/kgmm. Then, the
gearbox (the only subsystem excluded from the electric drive scaling) was added to the
weight of the electric drive subsystem; this total was included into the electric vehicle weight
calculation as Weiectric_drivenn • A summary table of electric drive weights for 100-mile EVs is
shown as  Table 3-62.

                     Table 3-62: Total electric drive weights for 100-mile EVs
EV powertrain weight estimates - 100 mile
Class
Subcompact
Small car
Large car
Small Truck
Minivan
Truck
Battery pack
size (kWh)
28.8
33.5
40.5
41.6
44.3
50.5
2020 electric
content (Ibs)
528
615
742
762
813
926
range
Gearbox
(power-split
or other)
50
60
70
60
80
100

2020 EV
powertrain
total
578
675
812
822
893
1026
       The difference between the actual weight and the predicted or nominal weight should
be zero.  However, if not then a revised weight reduction was used for another iteration of
steps 2-6 until the two vehicle weights match. Spreadsheet tools such as "solver" in MS
Excel were used for automating this iterative process.
mm 150 Wh/kg is a conservative estimate for year 2017 and beyond: outputs from ANL's battery cost model
show specific energy values of 160- 180 Wh/kg for a similar timeframe.
™ Applies only to the EV. Because the baseline ICE powertrain weight (which includes gearbox weight) was not
deducted from the PHEV, it is not added back in for the PHEV.
                                             3-144

-------
                                     Technologies Considered in the Agencies' Analysis
       Table 3-63 shows example results for 100-mile range EVs; in this case a 10% applied
glider weight reduction for a variety of vehicle classes.

                     Table 3-63: Sample calculation sheet for 100-mile EVs
Class


Subcompact
Small car
Large car
Small Truck
Minivan
Minivan w/ tov
Truck
Baseline curb
weight
Ibs
2628
3118
3751
3849
4087
f* 4087
4646
Baseline
power/wt
ratio
0.0487
0.0496
0.0710
0.0545
0.0570
0.0570
0.0566
Powertrain
weight
Ibs
1 495
590
710
590
775
' 775
965
Base glider
weight
Ibs
2133
2528
3041
3259
3312
3312
3681
WR
of glider
Ibs
427
506
608
652
662
662
736
NewEVwt
(nominal)
Ibs
2201
2612
3143
3197
3425
3425
3910
Energy cons. Battpacksize Electric drive
adjusted lOOmirange weightflb)
Wh/ni kWh
225 28.1
260 325
314 39.3
329 41.1
^346^ 43.3
346 43.3
390 48.7
566
656
790
813
874
874
994
NewEV
weight

2272
2679
3223
3421
3523
3523
3938
Error


0
0
0
0
0
0
0
% WR %RL
from curb vs.
base
13.5% 88%
14.1% 88%
14.1% 88%
11.1% 89%
13.8% 88%
13.8% 88%
15.2% 87%
       Table 3-64 shows the effect on net electric vehicle weight reduction after 20% glider
weight reduction was applied to EVs and PHEVs. As battery pack size increases for larger-
range EVs and PHEVs, the overall realized vehicle weight reduction decreases (because it
requires more energy to carry the extra battery weight). In this example, EVs with a 150 mile
range require almost 20% weight reduction to the glider to make up for the additional weight
of the electric drive  and battery pack compared to a conventional ICE-based powertrain.

Table 3-64: Actual weight reduction percentages for EVs and PHEVs with 20% weight reduction applied
                                       to glider



Subcompact
Small car
Large car
Small Truck
Minivan
Truck (w/ towing)
75 Mile EV
actual % WR
vs. base vehicle
19%
19%
19%
16%
19%
19%
100 Mile EV
actual % WR
vs. base vehicle
14%
14%
14%
11%
14%
14%
150 Mile EV
actual % WR
vs. base vehicle
2%
2%
2%
-1%
2%
2%
20 Mile PHEV
actual % WR
vs. base vehicle
12%
12%
12%
12%
12%
10%
40 Mile PHEV
actual % WR
vs. base vehicle
7%
7%
7%
8%
7%
6%
       Because there is no "all-electric range" requirement for HEVs, battery pack sizes were
relatively consistent for a given weight class. Furthermore, because battery pack sizes are at
least an order of magnitude smaller for HEVs than for all-electric vehicles, the sensitivity of
HEV vehicle weight (and hence energy consumption) to battery pack size is rather
insignificant. For these reasons, a more direct approach (rather than an iterative process)
works for battery sizing of HEVs.

       •   HEV batteries were scaled similar to the 2010 Fusion Hybrid based on nominal
          battery energy per Ib ETW (equivalent test weight), at 0.37 Wh/lb.

       •   A higher usable SOC window of 40% (compared to 30% for Fusion Hybrid)
          reduced the required Li-Ion battery size to 75% of the Fusion Hybrid's NiMH
          battery. This resulted in a 0.28 Wh/lb ETW ratio.
                                            3-145

-------
                                    Technologies Considered in the Agencies' Analysis
       •  In comparing anecdotal data for HEVs, the agencies assumed a slight weight
          increase of 4-5% for HEVs compared to baseline non-hybridized vehicles. The
          added weight of the Li-Ion pack, motor and other electric hardware were offset
          partially by the reduced size of the base engine.

       3.4.3.9 HEV, PHEV and EV battery pack design and cost analysis using the ANL
       BatPac model

       The U.S. Department of Energy (DOE) has established long term industry goals and
targets for advanced battery systems as it does for many energy efficient technologies.
Argonne National Laboratory (ANL) was funded by DOE to provide an independent
assessment of Li-ion battery costs because of their expertise in the field as one of the primary
DOE National Laboratories responsible for basic and applied battery energy storage
technologies for future HEV, PHEV and EV applications. A basic description of the ANL Li-
ion battery cost model and initial modeling results for PHEV applications were published in a
peer-reviewed technical paper presented at EVS-2457. ANL has extended modeling inputs
and pack design criteria within the battery cost model to include analysis of manufacturing
costs for EVs and HEVs as well has PHEVs.58 In early 2011,  ANL issued a draft report
detailing the methodology, inputs and outputs of their Battery Performance and Cost (BatPac)
model.59 A complete independent peer-review of the BatPac model and its inputs and results
for HEV, PHEV and EV applications has been completed60. ANL recently provided the
agencies with an updated report documenting the BatPac model that fully addresses the issues
raised within the peer review.61 Based on the feedback from peer-reviewers, ANL has updated
the model  in the following areas.

          1. Battery pack price is adjusted upward. This adjustment is based on the
             feedback from several peer-reviewers, and changes are related to limiting
             electrode thickness to 100 microns,  changing allocation of overhead cost to
             more closely represent a Tier 1 auto supplier, increasing cost of tabs, changing
             capital cost of material preparation, etc;
          2. Battery management system cost is increased to represent the complete
             monitoring and control needs for proper battery operation and safety as shown
             in Table 5.3  in the report;
          3. Battery automatic and manual disconnect unit cost is added based on safety
             considerations as  shown in Table 5.3 in the report;
          4. Liquid thermal management system is added. ANL states in the report that the
             closure design it uses in the model does not have sufficient surface area to be
             cooled by air effectively as shown in Table 5.3 in the report.

       This model and the peer review report are available in  the public dockets for this
rulemaking 60'62.

                                          3-146

-------
                                     Technologies Considered in the Agencies' Analysis
       NHTSA and EPA have decided to use the updated ANL BatPac model, dated July 17,
2011, for estimating large-format lithium-ion batteries for this proposal for the following
reasons.  First, the ANL model has been described and presented in the public domain and
does not rely upon confidential business information (which would therefore not be
reviewable by the public).  The model was developed by scientists at ANL who have
significant experience in this area.  The model uses a bill of materials methodology which the
agencies  believe is the preferred method for developing cost estimates.  The ANL model
appropriately considers the vehicle applications power and energy requirements, which are
two of the fundamental parameters when designing a lithium-ion battery for an HEV, PHEV,
or EV. The ANL model can estimate high volume production costs, which the agencies
believe is appropriate for the 2025 time frame. Finally, the ANL model's cost estimates,
while generally lower than the estimates we received from the OEMs, is consistent with some
of the supplier cost estimates the agencies received from large-format lithium-ion battery pack
manufacturers.  A portion of those data was received from on-site visits to vehicle
manufacturers and battery suppliers done by the EPA in 2008.

       The ANL battery cost model is based on a bill of materials approach in addition to
specific design criteria for the intended application  of a battery pack. The costs include
materials, manufacturing processes, the cost of capital equipment, plant area, and labor for
each manufacturing step as well as the design criteria include a vehicle  application's power
and energy storage capacity requirements, the battery's cathode and anode chemistry, and the
number of cells per module and modules per battery pack. The model assumes use of a
laminated multi-layer prismatic cell and battery modules consisting of double-seamed rigid
containers. The model also assumes that the battery modules are liquid-cooled. The model
takes into consideration the cost of capital equipment, plant area and labor for each step in the
manufacturing process for battery packs and places relevant limits on electrode coating
thicknesses and other processes limited by existing  and near-term manufacturing processes.
The ANL model also takes into consideration annual pack production volume and economies
of scale for high-volume production.

       Basic user inputs to BatPaC include performance goals (power and energy capacity),
choice of battery chemistry (of five predefined chemistries), the vehicle type for which the
battery is intended (HEV, PHEV, or EV), the desired number of cells and modules, and the
volume of production. BatPaC then designs the cells, modules, and battery pack, and
provides  an itemized cost breakdown at the specified production volume.

       BatPaC provides default values for engineering properties and material costs that
allow the model to operate without requiring the user to supply detailed technical or
experimental data.  In general, the default properties and costs represent what the model
authors consider to be reasonable values representing the state of the art expected to be
available to large battery manufacturers in the year  2020.  Users are encouraged to change
these defaults as necessary to represent their own expectations or their own proprietary data.

       In using BatPaC, it is extremely important that the user monitor certain properties of
the cells, modules, and packs that it generates, to ensure that they  stay within practical design

                                           3-147

-------
                                   Technologies Considered in the Agencies' Analysis
guidelines, adjusting related inputs if necessary. In particular, pack voltage and individual
cell capacity should be limited to appropriate ranges for the application.  These design
guidelines are not rigidly defined but approximate ranges are beginning to emerge in the
industry.

       Also inherent in BatPaC are certain modeling assumptions that are still open to some
uncertainty  or debate in the industry. For some, such as the available portion of total battery
energy (aka "SOC window") for a PHEV/EV/HEV, the user can easily modify a single
parameter to represent a value other than the default.  For others, such as the type of thermal
management employed (BatPaC is limited to liquid cooling and does not support passive or
active air cooling), or the packaging of cells and modules in a pack (parallel modules are not
supported),  changes can often be made by modifying the relevant cost outputs or performing
workarounds in the use of the model.

       The  cost outputs used by the agencies to determine  2025 HEV, PHEV and EV battery
costs were based on the following inputs and assumptions.

       EPA selected basic user inputs as follows. For performance goals, EPA used the
power and energy requirements derived from the scaling analysis described in the previous
section. Specifically, these covered each of the seven classes of vehicles (Subcompact, Small
Car, Large Car, Small Truck, Minivan, Minivan with Towing, Large Truck) under each of the
five weight  reduction scenarios (0%, 2%,  7.5%, 10%, and 20%). The chosen battery
chemistries  were NMC441-G (for EVs and PHEV40) and LMO-G (for P2 HEVs and
PHEV20). Vehicle types were EV75, EV100, EV150 (using the BatPaC "EV" setting);
PHEV20 and PHEV40 (using the "PHEV" setting), and P2 HEV (using the "HEV-HP"
setting). All modules were composed of 32 cells each, with each pack having a varying
number of modules. Cost outputs were generated for  annual production volumes of 50K,
125K, 250K, and 450K packs. The cost outputs for the 450K production volume are used in
the NPRM analysis.

       For engineering properties and material costs, and for other parameters not identified
below, EPA used the defaults provided in the model.

       For design guidelines regarding pack voltage  and cell capacity, EPA chose guidelines
based on knowledge of current practices and developing trends of battery manufacturers and
OEMs, supplemented by discussions with the BatPaC authors. Specifically: (1) allowable
pack voltage was targeted to approximately 120V for HEVs and approximately 350-400V for
EVs and PHEVs (with some EV150 packs for larger vehicles allowed to about 460-600V); (2)
allowable cell capacity was limited to less than 80 A-hr.

       EPA made several modeling assumptions that differed from the default model: (1) The
SOC window for PHEV20 was limited to 50% rather than the default 70%. (2) The SOC
window for HEVs was increased to 40% rather than the default 25%. (3) EV packs were
modeled as  two half-packs to avoid exceeding pack voltage guidelines. Although the model
provided for a potential solution by placing parallel cells within modules, EPA felt that likely


                                          3-148

-------
                                    Technologies Considered in the Agencies' Analysis
industry practices would be better represented by placing parallel modules within a pack, or
by dividing the pack into two parallel packs for packaging flexibility. Because the model did
not support parallel modules, each EV pack was modeled as two half-packs, each at half the
target power and energy, to be installed in parallel. Per ANL recommendation, half-packs
were modeled at twice the full-pack production volume, the projected half-pack cost was then
doubled, and costs for the battery management system (BMS), disconnects, and thermal
management were added only once, and module controls added twice. (4) HEV packs were
assumed to be air cooled instead of liquid cooled (except for large work trucks and minivans
with towing, which are still modeled as liquid-cooled). Because the model did not support air
cooling, EPA replaced the model's projected cost for liquid cooling with a cost for air cooling
(blower motor, ducting, and temperature feedback) derived from FEV's teardown studies.
EPA is working with ANL and investigating the potential for modifying the BatPac model to
include air cooling as an option

       Additionally, EPA did not include warranty costs computed by BatPaC in the total
battery cost because these are accounted for elsewhere by means of indirect cost multipliers
(ICMs).

               Table 3-65  Summary of Inputs and Assumptions Used with BatPaC
Category of
input/Assumptions
Annual production
volume
Battery chemistry
Allowable pack voltage
Allowable cell capacity
Cells per module
SOC window for HEVs
SOC window for
PHEV20
Thermal management
EV pack configuration
BatPaC Default or
Suggested Values
n/a
n/a
for HEV: 1 60-260 V
for PHEV, EV: 290-360 V
< 60 A-hr
16-32
25%
70%
Liquid
• Single pack, cells in
series
• Single pack, some
parallel cells
Agency Inputs for NPRM
Analysis
450,000
for HEV, PHEV20: LMO-G
for PHEV40, EV: NMC441-G
for HEV: ~ 120V
for PHEV, EV: ~ 360-600 V
< 80 A-hr
32
40%
50%
Air, for small/medium HEVs
Liquid for all others
Two packs, cells in series,
packs in parallel
                                           3-149

-------
                                     Technologies Considered in the Agencies' Analysis
       The cost projections produced by BatPaC are sensitive to the inputs and assumptions
the user provides. Significant uncertainty remains regarding which will best represent
manufacturer practice in the year 2020. The battery pack cost projection from BatPac model
ranges from $167/kWh for EV150 large truck to $267/kWh for PHEV40 large car with NMC
as chemistry and to $375/kWh for PHEV20 sub-compact car as shown in Table 3-66to Table
3-71. The agencies note that costs used in the analysis are lower than the costs generally
reported in stakeholder meetings, which ranged from $300/kW-hour to $400/kW-hour range
for 2020 and $250 to $300/kW-hour range for 2025. A comparison of BatPac modeling
results to the costs used in the 2012-2016 final rule and to cost estimates compiled by EPA
from battery suppliers and auto OEMs is shown in Figure 3-24.  The agencies also reviewed
publically available PHEV and EV battery cost literature including reports from Anderman63,
Frost & Sullivan64, TIAX65, Boston Consulting Group66, NRC67 etc. EPA and NHTSA
anticipate that public comment or further research may lead to the use of different inputs and
assumptions that may change the cost projections used for the final rule. Due to the the
uncertainties inherent in estimating battery costs through the 2025 model year, a sensitivity
analysis will be provided in each agency's RIA using a a range of costs estimated by DOE
technical experts to represent a reasonable outer bounds to the results from the BatPaC model.
In a recent report to NHTSA and EPA, DOE and  ANL suggested the following range for the
sensitivity study  with 95%  confidence interval after analyzing the confidence bound using the
BatPac model. The agencies plan to use this suggested range for the sensitivity study.

        Summary Table 1. Suggested confidence bounds as a percentage of the calculated point estimate
        for a graphite based Li-ion battery using the default inputs in BatPaC,
                                          Confidence Interval
         Battery type    Cathodes              lower    upper
         HEV          LMO, LFP, NCA, NMC       -10%      10%
         PHEV, EV      NMC, NCA              -10%      20%
         PHEV, EV      LMO, LFP                -20%      35%
                       Figure 3-23 Table from ANL Recommendation
                                           3-150

-------
                                      Technologies Considered in the Agencies' Analysis
                                 Estimated Battery Pack Costs
$1,200.00

1/1
ro
_Q
^
u
ro
Q.
.c
_*i
^.
vi
w
e




Total Range of OEM Supplier Estimates (>5sources) collected by EPA from 2008 -2010
| Range from the majority of OEM stakeholder meetings in June-August 2010
D2012-2016 EPA Cost Estimate



O BatPac Model (PHEV20] « BatPac Model (PHEV40]
+ BatPac Model (EV75] A BatPac Model (EV150) Q BatPac Model (EV100)



'' "i t f
1
2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2C
Calendar Year
 Figure 3-24 Comparison of direct manufacturing costs per unit of energy storage ($/kW-hr) between the
estimates used by EPA in the 2012-2016 GHG final rule, the BatPaC model results for PHEV20, PHEV40,
  EV75, EV100 and EV150 packages compared to estimates from OEM battery suppliers (2009 dollars,
  markups not included). Multiple points shown for the BatPaC model results for PHEV 20, PHEV40,
 EV75, EV100 and EV150 reflect the range of energy-specific costs for EPA's subcompact through large-
car package categories (see Table x for details). A range of OEM estimated battery costs from stakeholder
 meetings is also shown for comparison (red bars) which may or may not reflect additional cost markups.

       While it is expected that other Li-ion battery chemistries with higher energy density,
higher power density and lower cost will likely be available in the 2017-2025 timeframe, the
specific chemistries used for the cost analysis were chosen due to their known characteristics
and to be consistent with both public available information on current and near term HEV,
PHEV and EV product offerings from Hyundai, GM and Nissan as well as confidential
business information on future products currently under development.69'70'71'72 The cost
outputs  from the BatPaC model used by the agencies in this analysis are shown in Table 3-66
through Table 3-71 for different levels of applied weight reduction technology.  We
differentiate between "applied" weight reduction and "net"  weight reduction in this analysis
due to the fact that in order to achieve the same amount of mass reduction, more mass
reduction technologies might need to be applied to vehicles with electrifications than with
traditional powertrain because of the added weight of the electrification systems such as
battery, and in an effort to make clear that we have estimated vehicle level battery pack
costs—and motor and other electrified vehicle specific costs—based on the net weight
reduction of the vehicle.  For example, a typical EV150 battery back and associated motors
                                             3-151

-------
                                      Technologies Considered in the Agencies' Analysis
and other EV-specific equipment increases vehicle weight roughly 18 percent. As a result, an
EV150 that applied 20 percent mass reduction technology (see section 3.4.5.5 for a full
discussion of mass reduction technologies and costs) would have a net weight reduction of
only 2 percent.  In such a case, the agencies would estimate mass reduction costs associated
with a 20 percent applied mass reduction, and EV150 costs associated with only a 2 percent
net mass reduction (lower net mass reduction results in higher battery pack and motor costs).
Similarly, HEV battery packs increase vehicle weight by roughly 5 or 6 percent. Therefore,
for an HEV with 20 percent applied mass reduction technology—and costs associated with 20
percent applied mass reduction—would have HEV costs associated with a 15 percent net
mass reduction. Furthermore, such an HEV would have an effectiveness level improvement
associated with a  15 percent net mass reduction rather than a 20 percent net reduction.
  Table 3-66 Direct Manufacturing Costs for P2 HEV battery packs at
                     weight reduction (2009 dollars, markups not
different levels of applied vehicle
included)
P2 HEV (LMO)
@ 450K/yr
volume
Subcompact
Small Car
Large Car
Minivan
Minivan+towing
Small Truck
Large Truck
0% weight
reduction
Pack $/kWh
$716
$757
$864
$847
$928
$822
$964
$886
$802
$772
$699
$766
$717
$706
2% weight
reduction
Pack $/kWh
$713
$754
$858
$842
$923
$817
$958
$898
$813
$781
$708
$776
$727
$715
7.5% weight
reduction
Pack $/kWh
$704
$743
$843
$828
$909
$802
$942
$934
$845
$809
$734
$806
$752
$742
10% weight
reduction
Pack $/kWh
$700
$739
$836
$821
$902
$796
$934
$951
$861
$823
$747
$821
$765
$755
20% weight
reduction
Pack $/kWh
$691
$725
$819
$803
$887
$781
$920
$997
$900
$859
$779
$851
$801
$783
  Table 3-67 Direct Manufacturing Costs for PHEV20 battery packs at different levels of applied vehicle
                     weight reduction (2009 dollars, markups not included)
PHEV20
(LMO)
@ 450K/yr
volume
Subcompact
Small Car
Large Car
Minivan
Minivan+towing
Small Truck
Large Truck
0% weight
reduction
Pack $/kWh
$2,602
$2,746
$3,331
$3,296
$3,296
$3,143
$3,522
$375
$340
$342
$309
$309
$314
$290
2% weight
reduction
Pack $/kWh
$2,585
$2,726
$3,299
$3,267
$3,267
$3,116
$3,470
$377
$342
$343
$310
$310
$315
$289
7.5% weight
reduction
Pack $/kWh
$2,539
$2,671
$3,213
$3,188
$3,188
$3,042
$3,381
$381
$345
$343
$311
$311
$316
$289
10% weight
reduction
Pack $/kWh
$2,516
$2,647
$3,176
$3,153
$3,153
$3,010
$3,342
$382
$345
$343
$311
$311
$317
$289
20% weight
reduction
Pack $/kWh
$2,501
$2,628
$3,145
$3,126
$3,139
$2,974
$3,334
$384
$347
$344
$312
$313
$319
$290
                                            3-152

-------
                                       Technologies Considered in the Agencies' Analysis
Table 3-68 Direct Manufacturing Costs for PHEV40 battery pack at
                     weight reduction (2009 dollars, markups not
different levels of applied vehicle
included)
PHEV40
(NMC)
@ 450K/yr
volume
Subcompact
Small Car
Large Car
Minivan
Minivan+towing
Small Truck
Large Truck
0% weight
reduction
Pack $/kWh
$3,655
$4,043
$5,193
$5,041
$5,041
$4,788
$5,512
$264
$251
$267
$236
$236
$239
$227
2% weight
reduction
Pack $/kWh
$3,622
$3,986
$5,128
$4,985
$4,985
$4,737
$5,449
$264
$250
$266
$236
$236
$239
$227
7.5% weight
reduction
Pack $/kWh
$3,590
$3,883
$4,969
$4,883
$4,905
$4,602
$5,345
$268
$250
$266
$238
$239
$239
$226
10% weight
reduction
Pack $/kWh
$3,590
$3,888
$4,969
$4,893
$4,916
$4,598
$5,345
$268
$251
$266
$237
$238
$239
$226
20% weight
reduction
Pack $/kWh
$3,590
$3,888
$4,969
$4,893
$4,916
$4,598
$5,345
$268
$251
$266
$237
$238
$239
$226
 Table 3-69 Direct Manufacturing Costs for EV75 battery packs at different levels of applied vehicle
                     weight reduction (2009 dollars, markups not included)
EV75 (NMC)
@ 450K/yr
volume
Subcompact
Small Car
Large Car
Minivan
Minivan+towing
Small Truck
Large Truck
0% weight
reduction
Pack $/kWh
$5,418
$5,892
$7,180
$7,198
$7,198
$6,827
$7,613
$238
$223
$225
$206
$206
$208
$191
2% weight
reduction
Pack $/kWh
$5,384
$5,842
$7,102
$7,128
$7,128
$6,763
$7,764
$239
$223
$225
$206
$206
$208
$197
7.5% weight
reduction
Pack $/kWh
$5,340
$5,731
$6,907
$6,942
$6,942
$6,592
$7,557
$244
$225
$225
$206
$206
$209
$197
10% weight
reduction
Pack $/kWh
$5,306
$5,692
$6,822
$6,864
$6,864
$6,520
$7,468
$246
$226
$225
$206
$206
$209
$197
20% weight
reduction
Pack $/kWh
$5,155
$5,494
$6,509
$6,528
$6,528
$6,306
$7,116
$252
$232
$228
$209
$209
$211
$200
 Table 3-70 Direct Manufacturing Costs for EV100 battery packs at different levels of applied vehicle
                     weight reduction (2009 dollars, markups not included)
EV100 (NMC)
@ 450K/yr
volume
Subcompact
Small Car
Large Car
Minivan
Minivan+towing
0% weight
reduction
Pack $/kWh
$6,360
$7,001
$8,101
$8,414
$8,414
$210
$198
$190
$180
$180
2% weight
reduction
Pack $/kWh
$6,316
$6,951
$8,016
$8,348
$8,348
$211
$199
$190
$181
$181
7.5% weight
reduction
Pack $/kWh
$6,206
$6,782
$7,802
$8,183
$8,183
$213
$200
$191
$182
$182
10% weight
reduction
Pack $/kWh
$6,162
$6,727
$7,711
$8,116
$8,116
$214
$201
$191
$183
$183
20% weight
reduction
Pack $/kWh
$6,074
$6,600
$7,526
$7,980
$7,980
$216
$203
$192
$184
$184
                                              3-153

-------
                                     Technologies Considered in the Agencies' Analysis
Small Truck
Large Truck
$8,047
$9,232
$184
$174
$7,981
$9,158
$184
$174
$7,825
$8,970
$186
$175
$7,763
$8,895
$187
$176
$7,700
$8,671
$187
$178
  Table 3-71 Direct Manufacturing Costs for EV150 battery packs at different levels of applied vehicle
                     weight reduction (2009 dollars, markups not included)
EV150 (NMC)
@ 450K/yr
volume
Subcompact
Small Car
Large Car
Minivan
Minivan+towing
Small Truck
Large Truck
0% weight
reduction
$/kW
Pack h
$8,292
$9,189
$10,99
1
$11,74
7
$11,74
7
$11,25
3
$13,33
7
$182
$174
$172
$168
$168
$170
$167
2% weight
reduction
$/kW
Pack h
$8,260
$9,115
$10,90
2
$11,65
0
$11,65
0
$11,25
3
$13,22
7
$183
$174
$173
$168
$168
$170
$168
7.5% weight
reduction
Pack $/kWh
$8,260
$9,115
$10,902
$11,650
$11,650
$11,253
$13,172
$183
$174
$173
$168
$168
$170
$168
10% weight
reduction
Pack $/kWh
$8,260
$9,115
$10,902
$11,650
$11,650
$11,253
$13,172
$183
$174
$173
$168
$168
$170
$168
20% weight
reduction
Pack $/kWh
$8,260
$9,115
$10,902
$11,650
$11,650
$11,253
$13,172
$183
$174
$173
$168
$168
$170
$168
       Specifically for modeling purposes, both agencies wanted HEV/PHEV/EV battery
pack costs based on net weight reduction rather than applied weight reduction as shown in
Table 3-66 through Table 3-71 above.  The agencies did this by first determining the average
weight differences (applied weight reduction vs net weight reduction) for each of the 7 major
vehicle classes (subcompact, small car, large car, minivan, small truck, minivan+towing &
large truck) and each of the electrification types (P2 HEV, PHEV & EV).  Due to the weight
increases of adding electrification system and battery pack and the weight decreases by
applying smaller or no conventional internal combustion engine, the net mass reduction for
HEV, PHEV and EV varies for different electrification package and vehicle classes. For an
example, for a 20-mile subcompact PHEV, 5% mass reduction of the glider is offset by the
additional weight of electrification system, i.e. 5% mass reduction needs to be applied to
glider to achieve a net 0% overall vehicle mass reduction for a PHEV20 subcompact
passenger car. Those weight reduction differences are shown in Table 3-72.

   Table 3-72 EPA and NHTSA Weight Reduction Offset Associated with Electrification Technologies
Vehicle Class
Subcompact
Small car
Large car
Minivan
Small truck
Minivan+towing
Large truck
P2HEV
5%
5%
5%
5%
5%
6%
6%
PHEV20
7%
7%
7%
7%
7%
8%
9%
PHEV40
13%
12%
13%
13%
12%
14%
14%
EV75
0%
-1%
-1%
-1%
3%
-1%
-2%
EV100
6%
5%
5%
6%
8%
6%
4%
EV150
18%
17%
18%
18%
20%
18%
16%
             Notes:
                                            3-154

-------
                                        Technologies Considered in the Agencies' Analysis
              For example, PHEV40-specific technologies add 12-14% to vehicle weight so that a
              20% applied weight reduction would result in a 6-8% net weight reduction.
              While an EV75 can actually reduce vehicle weight by 1-2% (i.e., battery packs and
              motors weigh less than the removed internal combustion engine and transmission), the
              agencies used a value of 0% where negative entries are shown.
       The agencies then generated linear regressions of battery pack costs against percentage
net weight reduction using the costs shown in Table 3-66 through Table 3-71 and the weight
reduction offsets shown in Table 3-72. These results are shown in Table 3-73.

   Table 3-73 EPA and NHTSA Linear Regressions of Battery Pack Direct Manufacturing Costs vs Net
                                   Weight Reduction (2009$)
Vehicle
Class
Subcompact
Small car
Large car
Minivan
Small truck
Minivan+towing
Large truck
P2HEV
-$177x+$716
-$218x+$758
-$300x+$864
-$294x+$848
-$277x+$822
-$294x+$929
-$317x+$964
PHEV20
-$862x+$2,602
-$998x+$2,746
-$l,568x+$3,331
-$l,439x+$3,296
-$l,338x+$3,143


PHEV40
-$867x+$3,649
-$2,093x+$4,037
-$3,152x+$5,192
-$2,090x+$5,035
-$2,444x+$4,787


EV75
-$l,350x+$5,424
-$2,033x+$5,888
-$3,460x+$7,173
-$3,480x+$7,201
-$3,148x+$6,828


EV100
-$2,064x+$6,360
-$2,849x+$7,004
-$4,019x+$8,101
-$3,090x+$8,414
-$2,971x+$8,045


EV150
-$2,019x+$8,292
-$3,100x+$9,187
-$3,770x+$10,989
-$4,566x+$l 1,746
$11,253


Notes:
"x" in the equations represents the net weight reduction as a percentage, so a subcompact P2 HEV battery pack with a 20% applied weight
reduction and, therefore, a 15% net weight reduction would cost (-$177)x(15%)+$716=$689.
The small truck EV150 regression has no slope since the net weight reduction is always 0 due to the 20% weight reduction hit.
The agencies did not regress PHEV or EV costs for the minivan+towing and large truck vehicle classes since we do not believe these vehicle
classes would use the technologies.
       For P2 HEV battery packs, the direct manufacturing costs shown in Table 3-73 are
considered applicable to the 2017MY.  The agencies consider the P2 battery packs technology
to be on the flat portion of the learning curve during the 2017-2025 timeframe.  The agencies
have applied a highl complexity ICM of 1.56 through 2024 then 1.35 thereafter.  For PHEV
and EV battery packs, the direct manufacturing costs shown in Table 3-73 are considered
applicable to the 2025MY.  For the PHEV and EV battery packs, the agencies have applied
the learning curve discussed in Section 3.2.3.  The agencies have applied a high2  complexity
ICM of 1.77 through 2024 then 1.50 thereafter.  The resultant costs for P2 HEV, PHEV20,
PHEV40, EV75, EV100 and EV150 battery packs are shown in Table 3-74 through Table
3-79, respectively.

                        Table 3-74 Costs for P2 HEV Battery Packs (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
Vehicle class
Subcompact
Subcompact
Subcompact
Small car
Small car
Applied
WR
10%
15%
20%
10%
15%
Net
WR
5%
10%
15%
5%
10%
2017
$707
$699
$690
$747
$736
2018
$686
$678
$669
$725
$714
2019
$666
$657
$649
$703
$693
2020
$646
$638
$629
$682
$672
2021
$626
$618
$611
$661
$652
2022
$607
$600
$592
$641
$632
2023
$589
$582
$574
$622
$613
2024
$572
$564
$557
$603
$595
2025
$554
$547
$541
$585
$577
                                               3-155

-------
Technologies Considered in the Agencies' Analysis
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Minivan-
towing
Minivan-
towing
Minivan-
towing
Large truck
Large truck
Large truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Minivan-
towing
Minivan-
towing
Minivan-
towing
Large truck
Large truck
Large truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
4%
9%
14%
4%
9%
14%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
4%
9%
14%
4%
9%
14%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
$725
$849
$834
$819
$833
$818
$804
$808
$794
$780
$917
$902
$888
$952
$936
$920
$399
$394
$389
$421
$415
$409
$478
$470
$461
$470
$461
$453
$455
$448
$440
$517
$509
$500
$536
$527
$518
$1,106
$1,092
$1,078
$1,168
$1,151
$1,134
$1,327
$1,304
$1,280
$1,303
$1,280
$1,257
$1,264
$1,242
$703
$823
$809
$794
$808
$794
$779
$784
$770
$757
$889
$875
$861
$923
$908
$892
$397
$392
$387
$420
$413
$407
$477
$468
$460
$468
$460
$451
$454
$446
$438
$515
$507
$499
$534
$526
$517
$1,083
$1,070
$1,056
$1,144
$1,127
$1,111
$1,300
$1,277
$1,254
$1,276
$1,253
$1,231
$1,238
$1,217
$682
$798
$784
$770
$784
$770
$756
$760
$747
$734
$863
$849
$835
$895
$880
$866
$396
$391
$386
$418
$412
$406
$475
$467
$458
$466
$458
$450
$452
$445
$437
$513
$505
$497
$533
$524
$515
$1,062
$1,048
$1,035
$1,121
$1,105
$1,088
$1,274
$1,251
$1,229
$1,250
$1,228
$1,206
$1,213
$1,192
$662
$775
$761
$747
$760
$747
$733
$738
$725
$712
$837
$823
$810
$868
$854
$840
$395
$390
$385
$417
$411
$405
$474
$465
$457
$465
$457
$448
$451
$443
$435
$512
$503
$495
$531
$522
$513
$1,040
$1,027
$1,014
$1,098
$1,082
$1,066
$1,248
$1,226
$1,204
$1,225
$1,203
$1,182
$1,188
$1,168
$642
$751
$738
$725
$737
$724
$711
$715
$703
$691
$812
$799
$786
$842
$828
$814
$393
$389
$384
$415
$409
$403
$472
$464
$455
$463
$455
$447
$449
$442
$434
$510
$502
$494
$529
$520
$512
$1,020
$1,007
$994
$1,077
$1,061
$1,045
$1,223
$1,202
$1,180
$1,201
$1,180
$1,158
$1,165
$1,145
$623
$729
$716
$703
$715
$703
$690
$694
$682
$670
$787
$775
$762
$817
$804
$790
$392
$387
$382
$414
$408
$402
$471
$462
$454
$462
$454
$446
$448
$440
$433
$508
$500
$492
$528
$519
$510
$1,000
$987
$975
$1,056
$1,040
$1,025
$1,199
$1,178
$1,157
$1,177
$1,156
$1,136
$1,142
$1,122
$604
$707
$694
$682
$694
$682
$669
$673
$662
$650
$764
$752
$739
$793
$779
$766
$391
$386
$381
$413
$407
$401
$469
$461
$453
$461
$452
$444
$447
$439
$431
$507
$499
$491
$526
$517
$509
$980
$968
$956
$1,035
$1,020
$1,005
$1,176
$1,155
$1,135
$1,154
$1,134
$1,114
$1,120
$1,101
$586
$686
$674
$661
$673
$661
$649
$653
$642
$631
$741
$729
$717
$769
$756
$743
$390
$385
$380
$412
$406
$400
$468
$459
$451
$459
$451
$443
$445
$438
$430
$505
$497
$489
$524
$516
$507
$961
$949
$937
$1,015
$1,000
$986
$1,153
$1,133
$1,113
$1,132
$1,112
$1,092
$1,098
$1,080
$568
$665
$653
$642
$653
$641
$630
$633
$622
$612
$719
$707
$696
$746
$733
$721
$239
$236
$233
$253
$249
$245
$287
$282
$277
$282
$277
$272
$274
$269
$264
$310
$305
$300
$322
$317
$311
$794
$784
$774
$838
$826
$814
$952
$936
$919
$935
$918
$902
$907
$891
      3-156

-------
                                          Technologies Considered in the Agencies' Analysis
TC
TC
TC
TC
TC
TC
TC
Small truck
Minivan-
towing
Minivan-
towing
Minivan-
towing
Large truck
Large truck
Large truck
20%
10%
15%
20%
10%
15%
20%
15%
4%
9%
14%
4%
9%
14%
$1,220
$1,434
$1,411
$1,388
$1,488
$1,463
$1,438
$1,195
$1,405
$1,382
$1,359
$1,457
$1,433
$1,409
$1,171
$1,376
$1,354
$1,332
$1,428
$1,404
$1,380
$1,148
$1,349
$1,327
$1,305
$1,399
$1,376
$1,353
$1,125
$1,322
$1,301
$1,279
$1,372
$1,349
$1,326
$1,103
$1,296
$1,275
$1,254
$1,345
$1,322
$1,300
$1,081
$1,271
$1,250
$1,230
$1,319
$1,297
$1,275
$1,061
$1,246
$1,226
$1,206
$1,293
$1,272
$1,250
$876
$1,029
$1,013
$996
$1,068
$1,050
$1,032
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                          Table 3-75 Costs for PHEV20 Battery Packs (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
Vehicle
class
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Applied
WR
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
NetWR
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
2017
$5,032
$4,948
$4,864
$5,305
$5,208
$5,110
$6,413
$6,260
$6,107
$6,353
$6,212
$6,072
$6,059
$5,929
$5,798
$2,165
$2,128
$2,092
$2,282
$2,240
$2,198
$2,759
$2,693
$2,627
$2,733
$2,672
$2,612
$2,607
$2,550
$2,494
$7,197
$7,076
$6,956
$7,587
$7,448
$7,308
$9,172
$8,953
$8,734
2018
$4,026
$3,958
$3,891
$4,244
$4,166
$4,088
$5,131
$5,008
$4,886
$5,082
$4,970
$4,857
$4,847
$4,743
$4,638
$2,091
$2,056
$2,021
$2,204
$2,164
$2,123
$2,664
$2,601
$2,537
$2,639
$2,581
$2,523
$2,517
$2,463
$2,409
$6,116
$6,014
$5,911
$6,448
$6,330
$6,211
$7,795
$7,609
$7,423
2019
$4,026
$3,958
$3,891
$4,244
$4,166
$4,088
$5,131
$5,008
$4,886
$5,082
$4,970
$4,857
$4,847
$4,743
$4,638
$2,091
$2,056
$2,021
$2,204
$2,164
$2,123
$2,664
$2,601
$2,537
$2,639
$2,581
$2,523
$2,517
$2,463
$2,409
$6,116
$6,014
$5,911
$6,448
$6,330
$6,211
$7,795
$7,609
$7,423
2020
$3,220
$3,167
$3,113
$3,395
$3,333
$3,270
$4,104
$4,006
$3,908
$4,066
$3,976
$3,886
$3,878
$3,794
$3,711
$2,031
$1,997
$1,963
$2,142
$2,102
$2,063
$2,589
$2,527
$2,465
$2,565
$2,508
$2,451
$2,446
$2,393
$2,340
$5,252
$5,164
$5,076
$5,537
$5,435
$5,333
$6,693
$6,533
$6,374
2021
$3,220
$3,167
$3,113
$3,395
$3,333
$3,270
$4,104
$4,006
$3,908
$4,066
$3,976
$3,886
$3,878
$3,794
$3,711
$2,031
$1,997
$1,963
$2,142
$2,102
$2,063
$2,589
$2,527
$2,465
$2,565
$2,508
$2,451
$2,446
$2,393
$2,340
$5,252
$5,164
$5,076
$5,537
$5,435
$5,333
$6,693
$6,533
$6,374
2022
$3,220
$3,167
$3,113
$3,395
$3,333
$3,270
$4,104
$4,006
$3,908
$4,066
$3,976
$3,886
$3,878
$3,794
$3,711
$2,031
$1,997
$1,963
$2,142
$2,102
$2,063
$2,589
$2,527
$2,465
$2,565
$2,508
$2,451
$2,446
$2,393
$2,340
$5,252
$5,164
$5,076
$5,537
$5,435
$5,333
$6,693
$6,533
$6,374
2023
$3,220
$3,167
$3,113
$3,395
$3,333
$3,270
$4,104
$4,006
$3,908
$4,066
$3,976
$3,886
$3,878
$3,794
$3,711
$2,031
$1,997
$1,963
$2,142
$2,102
$2,063
$2,589
$2,527
$2,465
$2,565
$2,508
$2,451
$2,446
$2,393
$2,340
$5,252
$5,164
$5,076
$5,537
$5,435
$5,333
$6,693
$6,533
$6,374
2024
$3,220
$3,167
$3,113
$3,395
$3,333
$3,270
$4,104
$4,006
$3,908
$4,066
$3,976
$3,886
$3,878
$3,794
$3,711
$2,031
$1,997
$1,963
$2,142
$2,102
$2,063
$2,589
$2,527
$2,465
$2,565
$2,508
$2,451
$2,446
$2,393
$2,340
$5,252
$5,164
$5,076
$5,537
$5,435
$5,333
$6,693
$6,533
$6,374
2025
$2,576
$2,533
$2,490
$2,716
$2,666
$2,616
$3,284
$3,205
$3,127
$3,253
$3,181
$3,109
$3,102
$3,035
$2,969
$1,279
$1,258
$1,237
$1,349
$1,324
$1,299
$1,631
$1,592
$1,553
$1,615
$1,580
$1,544
$1,541
$1,507
$1,474
$3,856
$3,791
$3,727
$4,065
$3,990
$3,916
$4,914
$4,797
$4,679
                                                 3-157

-------
                                            Technologies Considered in the Agencies' Analysis
TC
TC
TC
TC
TC
TC
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
10%
15%
20%
10%
15%
20%
3%
8%
13%
3%
8%
13%
$9,086
$8,885
$8,684
$8,666
$8,479
$8,292
$7,722
$7,551
$7,380
$7,365
$7,206
$7,047
$7,722
$7,551
$7,380
$7,365
$7,206
$7,047
$6,630
$6,484
$6,337
$6,324
$6,188
$6,051
$6,630
$6,484
$6,337
$6,324
$6,188
$6,051
$6,630
$6,484
$6,337
$6,324
$6,188
$6,051
$6,630
$6,484
$6,337
$6,324
$6,188
$6,051
$6,630
$6,484
$6,337
$6,324
$6,188
$6,051
$4,868
$4,760
$4,653
$4,643
$4,543
$4,443
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                            Table 3-76 Costs for PHEV40 Battery Packs (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
Vehicle
class
Sub compact
Sub compact
Small car
Small car
Large car
Large car
Minivan
Minivan
Small truck
Small truck
Subcompact
Subcompact
Small car
Small car
Large car
Large car
Minivan
Minivan
Small truck
Small truck
Subcompact
Subcompact
Small car
Small car
Large car
Large car
Minivan
Minivan
Small truck
Small truck
Applied
WR
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
Net
WR
2%
7%
3%
8%
2%
7%
2%
7%
3%
8%
2%
7%
3%
8%
2%
7%
2%
7%
3%
8%
2%
7%
3%
8%
2%
7%
2%
7%
3%
8%
2017
$7,092
$7,007
$7,761
$7,557
$10,017
$9,709
$9,752
$9,548
$9,206
$8,967
$3,051
$3,014
$3,339
$3,251
$4,309
$4,177
$4,195
$4,107
$3,960
$3,858
$10,143
$10,022
$11,100
$10,808
$14,326
$13,886
$13,947
$13,655
$13,166
$12,825
2018
$5,674
$5,606
$6,209
$6,046
$8,014
$7,767
$7,802
$7,638
$7,365
$7,174
$2,946
$2,911
$3,225
$3,140
$4,162
$4,034
$4,052
$3,967
$3,825
$3,726
$8,620
$8,517
$9,434
$9,185
$12,175
$11,801
$11,853
$11,605
$11,190
$10,899
2019
$5,674
$5,606
$6,209
$6,046
$8,014
$7,767
$7,802
$7,638
$7,365
$7,174
$2,946
$2,911
$3,225
$3,140
$4,162
$4,034
$4,052
$3,967
$3,825
$3,726
$8,620
$8,517
$9,434
$9,185
$12,175
$11,801
$11,853
$11,605
$11,190
$10,899
2020
$4,539
$4,485
$4,967
$4,836
$6,411
$6,214
$6,241
$6,111
$5,892
$5,739
$2,863
$2,829
$3,133
$3,051
$4,044
$3,919
$3,937
$3,854
$3,716
$3,620
$7,402
$7,313
$8,100
$7,887
$10,455
$10,133
$10,178
$9,965
$9,608
$9,359
2021
$4,539
$4,485
$4,967
$4,836
$6,411
$6,214
$6,241
$6,111
$5,892
$5,739
$2,863
$2,829
$3,133
$3,051
$4,044
$3,919
$3,937
$3,854
$3,716
$3,620
$7,402
$7,313
$8,100
$7,887
$10,455
$10,133
$10,178
$9,965
$9,608
$9,359
2022
$4,539
$4,485
$4,967
$4,836
$6,411
$6,214
$6,241
$6,111
$5,892
$5,739
$2,863
$2,829
$3,133
$3,051
$4,044
$3,919
$3,937
$3,854
$3,716
$3,620
$7,402
$7,313
$8,100
$7,887
$10,455
$10,133
$10,178
$9,965
$9,608
$9,359
2023
$4,539
$4,485
$4,967
$4,836
$6,411
$6,214
$6,241
$6,111
$5,892
$5,739
$2,863
$2,829
$3,133
$3,051
$4,044
$3,919
$3,937
$3,854
$3,716
$3,620
$7,402
$7,313
$8,100
$7,887
$10,455
$10,133
$10,178
$9,965
$9,608
$9,359
2024
$4,539
$4,485
$4,967
$4,836
$6,411
$6,214
$6,241
$6,111
$5,892
$5,739
$2,863
$2,829
$3,133
$3,051
$4,044
$3,919
$3,937
$3,854
$3,716
$3,620
$7,402
$7,313
$8,100
$7,887
$10,455
$10,133
$10,178
$9,965
$9,608
$9,359
2025
$3,631
$3,588
$3,974
$3,869
$5,129
$4,971
$4,993
$4,889
$4,713
$4,591
$1,803
$1,782
$1,973
$1,921
$2,547
$2,469
$2,480
$2,428
$2,341
$2,280
$5,434
$5,370
$5,947
$5,791
$7,676
$7,440
$7,473
$7,316
$7,054
$6,871
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                             Table 3-77 Costs for EV75 Battery Packs (2009$)
Cost
type
DMC
DMC
DMC
DMC
Vehicle
class
Subcompact
Subcompact
Subcompact
Small car
Applied
WR
10%
15%
20%
10%
Net
WR
10%
15%
20%
10%
2017
$10,331
$10,199
$10,067
$11,103
2018
$8,264
$8,159
$8,054
$8,882
2019
$8,264
$8,159
$8,054
$8,882
2020
$6,612
$6,527
$6,443
$7,106
2021
$6,612
$6,527
$6,443
$7,106
2022
$6,612
$6,527
$6,443
$7,106
2023
$6,612
$6,527
$6,443
$7,106
2024
$6,612
$6,527
$6,443
$7,106
2025
$5,289
$5,222
$5,154
$5,685
                                                   3-158

-------
                                            Technologies Considered in the Agencies' Analysis
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
$10,904
$10,706
$13,333
$12,996
$12,658
$13,385
$13,045
$12,705
$12,722
$12,414
$12,107
$4,444
$4,387
$4,331
$4,776
$4,691
$4,605
$5,736
$5,591
$5,445
$5,758
$5,612
$5,466
$5,473
$5,340
$5,208
$14,775
$14,586
$14,398
$15,879
$15,595
$15,311
$19,069
$18,586
$18,103
$19,143
$18,657
$18,170
$18,194
$17,755
$17,315
$8,723
$8,565
$10,667
$10,396
$10,126
$10,708
$10,436
$10,164
$10,177
$9,931
$9,685
$4,292
$4,237
$4,182
$4,613
$4,530
$4,448
$5,540
$5,399
$5,259
$5,561
$5,420
$5,278
$5,285
$5,158
$5,030
$12,556
$12,396
$12,236
$13,495
$13,254
$13,012
$16,206
$15,796
$15,385
$16,269
$15,855
$15,442
$15,463
$15,089
$14,715
$8,723
$8,565
$10,667
$10,396
$10,126
$10,708
$10,436
$10,164
$10,177
$9,931
$9,685
$4,292
$4,237
$4,182
$4,613
$4,530
$4,448
$5,540
$5,399
$5,259
$5,561
$5,420
$5,278
$5,285
$5,158
$5,030
$12,556
$12,396
$12,236
$13,495
$13,254
$13,012
$16,206
$15,796
$15,385
$16,269
$15,855
$15,442
$15,463
$15,089
$14,715
$6,979
$6,852
$8,533
$8,317
$8,101
$8,566
$8,349
$8,131
$8,142
$7,945
$7,748
$4,170
$4,117
$4,064
$4,482
$4,402
$4,322
$5,382
$5,246
$5,110
$5,403
$5,266
$5,129
$5,135
$5,011
$4,887
$10,782
$10,644
$10,507
$11,588
$11,381
$11,173
$13,916
$13,563
$13,211
$13,969
$13,615
$13,260
$13,277
$12,956
$12,636
$6,979
$6,852
$8,533
$8,317
$8,101
$8,566
$8,349
$8,131
$8,142
$7,945
$7,748
$4,170
$4,117
$4,064
$4,482
$4,402
$4,322
$5,382
$5,246
$5,110
$5,403
$5,266
$5,129
$5,135
$5,011
$4,887
$10,782
$10,644
$10,507
$11,588
$11,381
$11,173
$13,916
$13,563
$13,211
$13,969
$13,615
$13,260
$13,277
$12,956
$12,636
$6,979
$6,852
$8,533
$8,317
$8,101
$8,566
$8,349
$8,131
$8,142
$7,945
$7,748
$4,170
$4,117
$4,064
$4,482
$4,402
$4,322
$5,382
$5,246
$5,110
$5,403
$5,266
$5,129
$5,135
$5,011
$4,887
$10,782
$10,644
$10,507
$11,588
$11,381
$11,173
$13,916
$13,563
$13,211
$13,969
$13,615
$13,260
$13,277
$12,956
$12,636
$6,979
$6,852
$8,533
$8,317
$8,101
$8,566
$8,349
$8,131
$8,142
$7,945
$7,748
$4,170
$4,117
$4,064
$4,482
$4,402
$4,322
$5,382
$5,246
$5,110
$5,403
$5,266
$5,129
$5,135
$5,011
$4,887
$10,782
$10,644
$10,507
$11,588
$11,381
$11,173
$13,916
$13,563
$13,211
$13,969
$13,615
$13,260
$13,277
$12,956
$12,636
$6,979
$6,852
$8,533
$8,317
$8,101
$8,566
$8,349
$8,131
$8,142
$7,945
$7,748
$4,170
$4,117
$4,064
$4,482
$4,402
$4,322
$5,382
$5,246
$5,110
$5,403
$5,266
$5,129
$5,135
$5,011
$4,887
$10,782
$10,644
$10,507
$11,588
$11,381
$11,173
$13,916
$13,563
$13,211
$13,969
$13,615
$13,260
$13,277
$12,956
$12,636
$5,583
$5,481
$6,827
$6,654
$6,481
$6,853
$6,679
$6,505
$6,513
$6,356
$6,199
$2,627
$2,593
$2,560
$2,823
$2,773
$2,722
$3,390
$3,304
$3,218
$3,403
$3,317
$3,230
$3,235
$3,156
$3,078
$7,916
$7,815
$7,714
$8,508
$8,356
$8,203
$10,217
$9,958
$9,699
$10,256
$9,996
$9,735
$9,748
$9,513
$9,277
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                             Table 3-78 Costs for EV100 Battery Packs (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
Vehicle
class
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Applied
WR
10%
15%
20%
10%
15%
20%
10%
Net
WR
4%
9%
14%
5%
10%
15%
5%
2017
$12,261
$12,059
$11,858
$13,401
$13,122
$12,844
$15,430
2018
$9,809
$9,648
$9,486
$10,721
$10,498
$10,275
$12,344
2019
$9,809
$9,648
$9,486
$10,721
$10,498
$10,275
$12,344
2020
$7,847
$7,718
$7,589
$8,576
$8,398
$8,220
$9,875
2021
$7,847
$7,718
$7,589
$8,576
$8,398
$8,220
$9,875
2022
$7,847
$7,718
$7,589
$8,576
$8,398
$8,220
$9,875
2023
$7,847
$7,718
$7,589
$8,576
$8,398
$8,220
$9,875
2024
$7,847
$7,718
$7,589
$8,576
$8,398
$8,220
$9,875
2025
$6,278
$6,174
$6,071
$6,861
$6,719
$6,576
$7,900
                                                   3-159

-------
                                            Technologies Considered in the Agencies' Analysis
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
4%
9%
14%
2%
7%
12%
4%
9%
14%
5%
10%
15%
5%
10%
15%
4%
9%
14%
2%
7%
12%
4%
9%
14%
5%
10%
15%
5%
10%
15%
4%
9%
14%
2%
7%
12%
$15,038
$14,645
$16,192
$15,890
$15,588
$15,597
$15,307
$15,017
$5,275
$5,188
$5,101
$5,765
$5,645
$5,525
$6,638
$6,469
$6,300
$6,966
$6,836
$6,706
$6,710
$6,585
$6,460
$17,536
$17,247
$16,959
$19,165
$18,767
$18,370
$22,068
$21,507
$20,946
$23,158
$22,726
$22,294
$22,307
$21,892
$21,477
$12,030
$11,716
$12,954
$12,712
$12,471
$12,478
$12,246
$12,013
$5,094
$5,010
$4,926
$5,567
$5,452
$5,336
$6,411
$6,248
$6,085
$6,727
$6,602
$6,476
$6,480
$6,359
$6,239
$14,903
$14,658
$14,413
$16,288
$15,950
$15,612
$18,755
$18,278
$17,801
$19,681
$19,314
$18,947
$18,958
$18,605
$18,252
$12,030
$11,716
$12,954
$12,712
$12,471
$12,478
$12,246
$12,013
$5,094
$5,010
$4,926
$5,567
$5,452
$5,336
$6,411
$6,248
$6,085
$6,727
$6,602
$6,476
$6,480
$6,359
$6,239
$14,903
$14,658
$14,413
$16,288
$15,950
$15,612
$18,755
$18,278
$17,801
$19,681
$19,314
$18,947
$18,958
$18,605
$18,252
$9,624
$9,373
$10,363
$10,170
$9,977
$9,982
$9,796
$9,611
$4,950
$4,868
$4,787
$5,410
$5,297
$5,185
$6,229
$6,071
$5,912
$6,536
$6,415
$6,293
$6,296
$6,179
$6,062
$12,797
$12,586
$12,376
$13,986
$13,696
$13,405
$16,104
$15,695
$15,285
$16,899
$16,584
$16,269
$16,278
$15,976
$15,673
$9,624
$9,373
$10,363
$10,170
$9,977
$9,982
$9,796
$9,611
$4,950
$4,868
$4,787
$5,410
$5,297
$5,185
$6,229
$6,071
$5,912
$6,536
$6,415
$6,293
$6,296
$6,179
$6,062
$12,797
$12,586
$12,376
$13,986
$13,696
$13,405
$16,104
$15,695
$15,285
$16,899
$16,584
$16,269
$16,278
$15,976
$15,673
$9,624
$9,373
$10,363
$10,170
$9,977
$9,982
$9,796
$9,611
$4,950
$4,868
$4,787
$5,410
$5,297
$5,185
$6,229
$6,071
$5,912
$6,536
$6,415
$6,293
$6,296
$6,179
$6,062
$12,797
$12,586
$12,376
$13,986
$13,696
$13,405
$16,104
$15,695
$15,285
$16,899
$16,584
$16,269
$16,278
$15,976
$15,673
$9,624
$9,373
$10,363
$10,170
$9,977
$9,982
$9,796
$9,611
$4,950
$4,868
$4,787
$5,410
$5,297
$5,185
$6,229
$6,071
$5,912
$6,536
$6,415
$6,293
$6,296
$6,179
$6,062
$12,797
$12,586
$12,376
$13,986
$13,696
$13,405
$16,104
$15,695
$15,285
$16,899
$16,584
$16,269
$16,278
$15,976
$15,673
$9,624
$9,373
$10,363
$10,170
$9,977
$9,982
$9,796
$9,611
$4,950
$4,868
$4,787
$5,410
$5,297
$5,185
$6,229
$6,071
$5,912
$6,536
$6,415
$6,293
$6,296
$6,179
$6,062
$12,797
$12,586
$12,376
$13,986
$13,696
$13,405
$16,104
$15,695
$15,285
$16,899
$16,584
$16,269
$16,278
$15,976
$15,673
$7,699
$7,498
$8,290
$8,136
$7,981
$7,986
$7,837
$7,689
$3,118
$3,066
$3,015
$3,407
$3,337
$3,266
$3,923
$3,824
$3,724
$4,117
$4,040
$3,964
$3,966
$3,892
$3,818
$9,395
$9,241
$9,086
$10,268
$10,055
$9,842
$11,824
$11,523
$11,222
$12,407
$12,176
$11,945
$11,951
$11,729
$11,507
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                             Table 3-79 Costs for EV150 Battery Packs (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
Vehicle
class
Subcompact
Small car
Large car
Minivan
Small truck
Subcompact
Small car
Large car
Minivan
Small truck
Applied
WR
20%
20%
20%
20%
20%
20%
20%
20%
20%
20%
Net
WR
2%
3%
2%
2%
0%
2%
3%
2%
2%
0%
2017
$16,116
$17,762
$21,316
$22,763
$21,979
$6,933
$7,641
$9,170
$9,792
$9,455
2018
$12,893
$14,210
$17,052
$18,210
$17,583
$6,696
$7,379
$8,856
$9,457
$9,131
2019
$12,893
$14,210
$17,052
$18,210
$17,583
$6,696
$7,379
$8,856
$9,457
$9,131
2020
$10,314
$11,368
$13,642
$14,568
$14,066
$6,506
$7,170
$8,605
$9,189
$8,872
2021
$10,314
$11,368
$13,642
$14,568
$14,066
$6,506
$7,170
$8,605
$9,189
$8,872
2022
$10,314
$11,368
$13,642
$14,568
$14,066
$6,506
$7,170
$8,605
$9,189
$8,872
2023
$10,314
$11,368
$13,642
$14,568
$14,066
$6,506
$7,170
$8,605
$9,189
$8,872
2024
$10,314
$11,368
$13,642
$14,568
$14,066
$6,506
$7,170
$8,605
$9,189
$8,872
2025
$8,251
$9,094
$10,914
$11,655
$11,253
$4,098
$4,516
$5,420
$5,788
$5,588
                                                   3-160

-------
                                          Technologies Considered in the Agencies' Analysis
TC
TC
TC
TC
TC
Subcompact
Small car
Large car
Minivan
Small truck
20%
20%
20%
20%
20%
2%
3%
2%
2%
0%
$23,049
$25,403
$30,485
$32,555
$31,434
$19,588
$21,589
$25,908
$27,668
$26,714
$19,588
$21,589
$25,908
$27,668
$26,714
$16,820
$18,538
$22,247
$23,757
$22,939
$16,820
$18,538
$22,247
$23,757
$22,939
$16,820
$18,538
$22,247
$23,757
$22,939
$16,820
$18,538
$22,247
$23,757
$22,939
$16,820
$18,538
$22,247
$23,757
$22,939
$12,349
$13,610
$16,333
$17,443
$16,842
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
            Because CAFE Volpe model does not use pre-built package and it applies
     technologies as necessary to meet the fuel consumption reduction requirement, cost
     interaction between any particular technology and other technologies has to be flexible so that
     when a technology is picked, the model will automatically look through the cost synergy
     defined in a table and apply cost adjustment accordingly. The total cost for mass reduction
     and electrification is composed of the following four parts.

               (1) Cost of net mass reduction;
               (2) Cost of electrification with zero mass reduction;
               (3) Mass  reduction cost synergy  for increased or  decreased  amount  of mass
                  reduction due  to  switching from conventional powertrain to electrification
                  systems as define  in Figure 3-25. For an example, if a midsize passenger car
                  needs  both 10 percent net mass reduction and  P2 hybrid to meet the  CAFE
                  target, the model will  need to find the cost of additional  5 percent of mass
                  reduction to consider the vehicle weight  increase due to switching from
                  conventional powertrain system to P2 electrification packages. This additional
                  5  percent of mass  reduction  is calculated starting  from  10 percent mass
                  reduction, not zero as shown in Figure 3-25 because mass  reduction cost versus
                  mass reduction percent is not a linear function. The cost increases  faster as the
                  amount of mass reduction becomes higher.
               (4) Electrification system cost  synergy (battery and non-battery components) due
                  to mass reduction as defined  in Table  3-73 and  Table 3-86: Continuing the
                  example in the steps above, if a midsize passenger car needs both 10 percent
                  net mass  reduction and P2 hybrid to meet the CAFE target, after calculating
                  the costs above, the model will need  to find the cost of electrification systems,
                  including battery system and non-battery system, with the required net amount
                  of mass reduction  using the equations in Table 3-73 and Table 3-86. Then the
                  delta cost between this cost and the cost calculated in step 2, i.e. electrification
                  system cost with zero applied mass reduction is calculated and treated as a cost
                  synergy. These cost deltas  are normally a negative, i.e. cost reduction,  due to
                  the downsizing of electrification system resulting from mass reduction.

     The sum of item (3) and  (4) in the above list are calculate as cost synergy and store in the cost
     synergy table as defined  in NHTSA's RIA.
                                                3-161

-------
                                    Technologies Considered in the Agencies' Analysis
          Figure 3-25 Mass Reduction Cost Example for Applied and Net Mass Reduction
$600
S $500
8 $400
'•§ $300
3
| $200
U)
| $100
$-
0
Example of Applied and Net Mass Reduction
Costs



\tf°*f

J^

^
<3
'
iX<°^ ^>^

^^
cP^I
^t^J^
^^/^x"
' s'
y^


/
T




% 5% 10% 15% 20%
Amount of Mass Reduction [%]
       The agencies have also carefully reconsidered the power and energy requirements for
each electrified vehicle type, which has a significant impact on the cost estimates for HEVs,
PHEVs, and EVs as compared to the estimates used in the 2012-2016 rulemaking.

       The agencies note that, for this analysis, the agencies have assumed batteries will be
capable of lasting the lifetime of the vehicle, which is consistent with the expected customer
demands from this technology (as manufacturers have confirmed).  Lastly, the agencies have
focused attention on an emerging HEV technology known as a P2-hybrid, a technology not
considered in the 2012-2016 light-duty rule.

       The agencies have also considered, for this analysis, the costs associated with in-home
chargers expected to be necessary for PHEVs and EVs. Further details on in-home chargers
and their estimated costs  are presented in Section 3.4.4.
       3.4.3.10
Non-battery costs for MHEVs, HEVs, PHEVs, EVs and FCEVs
       This section addresses the costs of non-battery components which are required for
electric drive vehicles. Some of these components are not found in every electric-drive
vehicle (e.g. an HEV does not have an on-board battery charger as found in a PHEV or EV).
Others are found in all electric drive vehicles and/or must be scaled to the vehicle type or
class to properly represent the cost.  The agencies derived the costs of these components from
the FEV teardown study and the 2010 TAR. Where appropriate, costs were scaled to vehicle
                                           3-162

-------
                                    Technologies Considered in the Agencies' Analysis
class and in the case of the motor and inverter, the sizing methodology used for battery sizing
was applied.

       The electric drive motor and inverter provide the motive power for any electric-drive
vehicle converting electrical energy from the battery into kinetic energy for propulsion. In an
electric-drive vehicle, energy stored in the battery is routed to the inverter which converts it to
a voltage and wave form that can be used by the motor.

       In many cases, such as HEVs, the combined cost of the motor and inverter exceed the
battery cost. As batteries become larger in PHEVs and EVs, the battery cost grows faster than
motor and inverter cost. For this analysis, the agencies used the vehicle power requirement
calculation discussed in 3.4.3.8 to calculate the required motor and inverter size for each
vehicle class at each weight reduction point. Then, for the HEVs and PHEVs, a regression
was created from the FEV teardown data for motors and inverters and this regression was
used to calculate the motor and inverter cost for each combination of vehicle class and weight
reduction. This regression was $14.48x(motor size in kW)+$763.54.  The results are shown
as the "Motor assembly" line item in Table 3-80, Table 3-81 and Table 3-82 which show our
scaled DMC for P2 HEV, PHEV20 and PHEV40, respectively.

       For EVs, the agencies used the motor and inverter cost regression from the 2010 TAR
(see TAR at page B-21). Since the FEV teardown was conducted on an HEV Ford Fusion,
the agencies believe the technology for an EV  is different enough to warrant using the TAR
regression. The regression presented in the TAR showed  the DMC being equal to
$8.28x(motor size in kW)+$181.43. The results are presented as separate line items for
"Motor inverter" and "Motor assembly" in Table 3-83, Table 3-84 and Table 3-85 which
show our scaled DMC for EV75, EV100 and EV150, respectively.

       In addition to electric drive motors and inverters, there are several other components
in electric drive vehicles that are required. These components include the following:

       •   Body Modifications which are required on HEVs and PHEVs include changes  to
          sheet metal to accommodate electric drive components and the addition of
          fasteners to secure components such as electric cables.  These costs come from the
          FEV teardown and are scaled by vehicle class. For EVs, these costs are assumed
          to be included in the base vehicle because they are less likely to be adapted from
          conventional vehicles.

       •   Brake System changes include the addition of a braking system that can control the
          vehicle's regenerative braking system—a key enabler of electric drive vehicle
          efficiency. The brake system costs are from the FEV teardown and are scaled to
          vehicle class.

       •   Climate Control System includes components such as an electric air conditioning
          compressor that enables operation while the engine is off for HEVs and PHEVs as
                                           3-163

-------
                          Technologies Considered in the Agencies' Analysis
well as for an EV which has no engine.  Climate control system costs come from
the FEV teardown and are scaled to vehicle class.

Conventional vehicle battery and alternator are deleted in these vehicles, for a cost
savings, replaced by the DC-DC converter which converts the high-voltage
traction battery to a nominal 12V DC to operate the vehicle's accessories. This
credit comes from the FEV teardown study and is scaled to vehicle class.

DC-DC converter converts the high-voltage battery voltage to a nominal 12V
battery voltage to run vehicle accessories such as the radio, lights and wipers. This
cost comes from the FEV teardown study and is scaled to vehicle class.

Power distribution and Control consists of those components which route
electricity to the motor, inverter and contains the controllers to operate and
monitor the electric drive system.  This cost applies to HEVs and PHEVs and
comes from the FEV teardown study. It is scaled to vehicle class.

On-Vehicle Charger consists of the components necessary to charge a PHEV or
EV from an outlet.  It includes the charging port, wiring and electronics necessary
to convert a 120V or 240V AC input to the high-voltage DC power necessary to
charge the battery.  Because the FEV teardown study subject vehicle did not have
an on-vehicle charger,  the costs from the TAR were used for this item. It is not
scaled to vehicle class, however the EV charger is assumed to cost twice the
amount of the PHEV charger to account for a higher current capacity.  This cost
does not include off-vehicle charger components which are discussed in  Section
3.4.4, below.

Supplemental heating is required for passenger comfort on PHEVs and EVs which
may operate for long periods with no engine heat available.  This cost comes from
the FEV teardown study and is scaled to vehicle class. The supplemental heater on
the EV is assumed to be three times more costly than the PHEV because the entire
cabin  comfort is dependent on the supplemental heater.  In a PHEV, it is assumed
that in extreme conditions, the internal combustion engine will start to provide
additional cabin heat and defrost functions.

High Voltage Wiring is an item used on EVs only.  It includes the high voltage
cabling from the battery to the inverter and motor as well as control components.
It is equivalent to the power distribution and control used on HEVs and PHEVs
and comes from the FEV teardown study. It is scaled to vehicle class.

Delete Internal  Combustion Engine and Transmission For EVs, the engine and
transmission are deleted and a credit is applied. These credits come from work
done in support of the 2010 TAR and are scaled to vehicle class.
                                 3-164

-------
                                    Technologies Considered in the Agencies' Analysis
       The results of the scaling exercise applied to non-battery components are presented in
Table 3-80 through Table 3-85 for P2 HEVs, PHEV20, PHEV40, EV75, EV100 and EV150,
respectively.

    Table 3-80 Scaled Non-battery DMC by Applied Vehicle Weight Reduction for P2 HEV (2009$)
System
0%WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
Motor assembly
Total
2%WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
Motor assembly
Total
7.5% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
Motor assembly
Total
70% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
Motor assembly
Total
20% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
Motor assembly
Total
Subcompact

$6
$233
$154
-$62
$115
$203
$1,038
$1,688

$6
$233
$154
-$62
$115
$203
$1,038
$1,688

$6
$233
$154
-$62
$115
$203
$1,024
$1,673

$6
$233
$154
-$62
$115
$203
$1,009
$1,659

$6
$233
$154
-$62
$115
$203
$995
$1,644
Small car

$6
$238
$164
-$67
$126
$207
$1,096
$1,771

$6
$238
$164
-$67
$126
$207
$1,096
$1,771

$6
$238
$164
-$67
$126
$207
$1,067
$1,742

$6
$238
$164
-$67
$126
$207
$1,067
$1,742

$6
$238
$164
-$67
$126
$207
$1,053
$1,727
Large car

$6
$242
$176
-$85
$157
$210
$1,342
$2,048

$6
$242
$176
-$85
$157
$210
$1,327
$2,034

$6
$242
$176
-$85
$157
$210
$1,298
$2,005

$6
$242
$176
-$85
$157
$210
$1,284
$1,990

$6
$242
$176
-$85
$157
$210
$1,255
$1,961
Minivan

$7
$242
$220
-$89
$167
$210
$1,270
$2,027

$7
$242
$220
-$89
$167
$210
$1,255
$2,013

$7
$242
$220
-$89
$167
$210
$1,226
$1,984

$7
$242
$220
-$89
$167
$210
$1,226
$1,984

$7
$242
$220
-$89
$167
$210
$1,197
$1,955
Minivan+towing

$7
$242
$220
-$89
$167
$210
$1,270
$2,027

$7
$242
$220
-$89
$167
$210
$1,255
$2,013

$7
$242
$220
-$89
$167
$210
$1,226
$1,984

$7
$242
$220
-$89
$167
$210
$1,226
$1,984

$7
$242
$220
-$89
$167
$210
$1,197
$1,955
Small truck

$6
$240
$202
-$89
$167
$207
$1,212
$1,946

$6
$240
$202
-$89
$167
$207
$1,212
$1,946

$6
$240
$202
-$89
$167
$207
$1,183
$1,917

$6
$240
$202
-$89
$167
$207
$1,168
$1,903

$6
$240
$202
-$89
$167
$207
$1,154
$1,888
Large truck

$7
$248
$194
-$97
$183
$219
$1,327
$2,082

$7
$248
$194
-$97
$183
$219
$1,313
$2,067

$7
$248
$194
-$97
$183
$219
$1,284
$2,038

$7
$248
$194
-$97
$183
$219
$1,284
$2,038

$7
$248
$194
-$97
$183
$219
$1,255
$2,009
    Table 3-81 Scaled Non-battery DMC by Applied Vehicle Weight Reduction for PHEV20 (2009$)a

                                           3-165

-------
                                            Technologies Considered in the Agencies' Analysis
System
0%WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
2%WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
7.5% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
70% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
20% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
Subcompact

$6
$233
$154
-$62
$115
$203
$103
$42
$2,151
$2,947

$6
$233
$154
-$62
$115
$203
$103
$42
$2,122
$2,918

$6
$233
$154
-$62
$115
$203
$103
$42
$2,036
$2,831

$6
$233
$154
-$62
$115
$203
$103
$42
$2,007
$2,802

$6
$233
$154
-$62
$115
$203
$103
$42
$1,978
$2,773
Small car

$6
$238
$164
-$67
$126
$207
$103
$45
$2,426
$3,249

$6
$238
$164
-$67
$126
$207
$103
$45
$2,397
$3,220

$6
$238
$164
-$67
$126
$207
$103
$45
$2,310
$3,133

$6
$238
$164
-$67
$126
$207
$103
$45
$2,267
$3,090

$6
$238
$164
-$67
$126
$207
$103
$45
$2,238
$3,061
Large car

$6
$242
$176
-$85
$157
$210
$103
$48
$3,640
$4,498

$6
$242
$176
-$85
$157
$210
$103
$48
$3,583
$4,440

$6
$242
$176
-$85
$157
$210
$103
$48
$3,424
$4,281

$6
$242
$176
-$85
$157
$210
$103
$48
$3,351
$4,209

$6
$242
$176
-$85
$157
$210
$103
$48
$3,294
$4,151
Minivan

$7
$242
$220
-$89
$167
$210
$103
$60
$3,279
$4,200

$7
$242
$220
-$89
$167
$210
$103
$60
$3,221
$4,142

$7
$242
$220
-$89
$167
$210
$103
$60
$3,091
$4,012

$7
$242
$220
-$89
$167
$210
$103
$60
$3,033
$3,954

$7
$242
$220
-$89
$167
$210
$103
$60
$2,990
$3,911
Small truck

$6
$240
$202
-$89
$167
$207
$103
$55
$3,019
$3,911

$6
$240
$202
-$89
$167
$207
$103
$55
$2,975
$3,868

$6
$240
$202
-$89
$167
$207
$103
$55
$2,860
$3,752

$6
$240
$202
-$89
$167
$207
$103
$55
$2,802
$3,694

$6
$240
$202
-$89
$167
$207
$103
$55
$2,744
$3,637
a The agencies have not estimated PHEV or EV costs for the minivan+towing and large truck vehicle classes since we do not believe these
                                                   3-166

-------
                                        Technologies Considered in the Agencies' Analysis
vehicle classes would use the technologies.
                                               3-167

-------
                              Technologies Considered in the Agencies' Analysis
Table 3-82 Scaled Non-battery DMC by Applied Vehicle Weight Reduction for PHEV40 (2009$)
System
0%WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
2%WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
7.5% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
70% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Total
20% WR
Body system
Brake system
Climate controls
Delete electrical
DC-DC converter
Power Distr & control
On vehicle charger
Supplemental heater
Motor assembly
Subcompact

$6
$233
$154
-$62
$115
$203
$103
$42
$2,151
$2,947

$6
$233
$154
-$62
$115
$203
$103
$42
$2,122
$2,918

$6
$233
$154
-$62
$115
$203
$103
$42
$2,050
$2,845

$6
$233
$154
-$62
$115
$203
$103
$42
$2,050
$2,845

$6
$233
$154
-$62
$115
$203
$103
$42
$2,050
Small car

$6
$238
$164
-$67
$126
$207
$103
$45
$2,426
$3,249

$6
$238
$164
-$67
$126
$207
$103
$45
$2,397
$3,220

$6
$238
$164
-$67
$126
$207
$103
$45
$2,310
$3,133

$6
$238
$164
-$67
$126
$207
$103
$45
$2,310
$3,133

$6
$238
$164
-$67
$126
$207
$103
$45
$2,310
Large car

$6
$242
$176
-$85
$157
$210
$103
$48
$3,640
$4,498

$6
$242
$176
-$85
$157
$210
$103
$48
$3,583
$4,440

$6
$242
$176
-$85
$157
$210
$103
$48
$3,438
$4,296

$6
$242
$176
-$85
$157
$210
$103
$48
$3,438
$4,296

$6
$242
$176
-$85
$157
$210
$103
$48
$3,438
Minivan

$7
$242
$220
-$89
$167
$210
$103
$60
$3,279
$4,200

$7
$242
$220
-$89
$167
$210
$103
$60
$3,221
$4,142

$7
$242
$220
-$89
$167
$210
$103
$60
$3,106
$4,026

$7
$242
$220
-$89
$167
$210
$103
$60
$3,106
$4,026

$7
$242
$220
-$89
$167
$210
$103
$60
$3,106
Small truck

$6
$240
$202
-$89
$167
$207
$103
$55
$3,019
$3,911

$6
$240
$202
-$89
$167
$207
$103
$55
$2,975
$3,868

$6
$240
$202
-$89
$167
$207
$103
$55
$2,860
$3,752

$6
$240
$202
-$89
$167
$207
$103
$55
$2,845
$3,738

$6
$240
$202
-$89
$167
$207
$103
$55
$2,845
                                     3-168

-------
                                              Technologies Considered in the Agencies' Analysis
Total	|	$2,845  |	$3,133 |	$4,296 |	$4,026 |	$3,738
a The agencies have not estimated PHEV or EV costs for the minivan+towing and large truck vehicle classes since we do not believe these
vehicle classes would use the technologies.
                                                      3-169

-------
                                Technologies Considered in the Agencies' Analysis
Table 3-83 Scaled Non-battery DMC by Applied Vehicle Weight Reduction for EV75 (2009$)
System
0%WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
2%WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
7.5% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
70% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Subcompact

$233
$154
-$62
$115
$203
$85
$309
$693
$119
-$1,565
-$877
$1,007
$415

$233
$154
-$62
$115
$203
$85
$309
$679
$119
-$1,565
-$877
$990
$384

$233
$154
-$62
$115
$203
$85
$309
$635
$119
-$1,565
-$877
$939
$289

$233
$154
-$62
$115
$203
$85
$309
$621
$119
-$1,565
-$877
Small car

$238
$164
-$67
$126
$207
$90
$309
$830
$119
-$1,565
-$877
$1,169
$745

$238
$164
-$67
$126
$207
$90
$309
$816
$119
-$1,565
-$877
$1,152
$713

$238
$164
-$67
$126
$207
$90
$309
$773
$119
-$1,565
-$877
$1,101
$619

$238
$164
-$67
$126
$207
$90
$309
$751
$119
-$1,565
-$877
Large car

$242
$176
-$85
$157
$210
$97
$309
$1,437
$119
-$2,418
-$877
$1,887
$1,254

$242
$176
-$85
$157
$210
$97
$309
$1,408
$119
-$2,418
-$877
$1,853
$1,191

$242
$176
-$85
$157
$210
$97
$309
$1,328
$119
-$2,418
-$877
$1,759
$1,017

$242
$176
-$85
$157
$210
$97
$309
$1,292
$119
-$2,418
-$877
Minivan

$242
$220
-$89
$167
$210
$120
$309
$1,256
$119
-$2,347
-$877
$1,673
$1,005

$242
$220
-$89
$167
$210
$120
$309
$1,227
$119
-$2,347
-$877
$1,639
$942

$242
$220
-$89
$167
$210
$120
$309
$1,162
$119
-$2,347
-$877
$1,562
$800

$242
$220
-$89
$167
$210
$120
$309
$1,134
$119
-$2,347
-$877
Small truck

$240
$202
-$89
$167
$207
$110
$309
$1,126
$119
-$1,849
-$877
$1,520
$1,186

$240
$202
-$89
$167
$207
$110
$309
$1,105
$119
-$1,849
-$877
$1,494
$1,139

$240
$202
-$89
$167
$207
$110
$309
$1,047
$119
-$1,849
-$877
$1,426
$1,013

$240
$202
-$89
$167
$207
$110
$309
$1,018
$119
-$1,849
-$877
                                       3-170

-------
                                             Technologies Considered in the Agencies' Analysis
Motor assembly
Total
20% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
$922
$258

$233
$154
-$62
$115
$203
$85
$309
$563
$119
-$1,565
-$877
$853
$132
$1,075
$571

$238
$164
-$67
$126
$207
$90
$309
$671
$119
-$1,565
-$877
$982
$398
$1,716
$938

$242
$176
-$85
$157
$210
$97
$309
$1,155
$119
-$2,418
-$877
$1,554
$639
$1,528
$737

$242
$220
-$89
$167
$210
$120
$309
$1,018
$119
-$2,347
-$877
$1,392
$485
$1,392
$950

$240
$202
-$89
$167
$207
$110
$309
$946
$119
-$1,849
-$877
$1,306
$792
a The agencies have not estimated PHEV or EV costs for the minivan+towing and large truck vehicle classes since we do not believe these
vehicle classes would use the technologies.
                                                      3-171

-------
                                 Technologies Considered in the Agencies' Analysis
Table 3-84 Scaled Non-battery DMC by Applied Vehicle Weight Reduction for EV100 (2009$)
System
0%WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
2%WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
7.5% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
70% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Subcompact

$233
$154
-$62
$115
$203
$85
$309
$693
$119
-$1,565
-$877
$1,007
$415

$233
$154
-$62
$115
$203
$85
$309
$679
$119
-$1,565
-$877
$990
$384

$233
$154
-$62
$115
$203
$85
$309
$635
$119
-$1,565
-$877
$939
$289

$233
$154
-$62
$115
$203
$85
$309
$621
$119
Small car

$238
$164
-$67
$126
$207
$90
$309
$830
$119
-$1,565
-$877
$1,169
$745

$238
$164
-$67
$126
$207
$90
$309
$816
$119
-$1,565
-$877
$1,152
$713

$238
$164
-$67
$126
$207
$90
$309
$773
$119
-$1,565
-$877
$1,101
$619

$238
$164
-$67
$126
$207
$90
$309
$751
$119
Large car

$242
$176
-$85
$157
$210
$97
$309
$1,437
$119
-$2,418
-$877
$1,887
$1,254

$242
$176
-$85
$157
$210
$97
$309
$1,408
$119
-$2,418
-$877
$1,853
$1,191

$242
$176
-$85
$157
$210
$97
$309
$1,328
$119
-$2,418
-$877
$1,759
$1,017

$242
$176
-$85
$157
$210
$97
$309
$1,292
$119
Minivan

$242
$220
-$89
$167
$210
$120
$309
$1,256
$119
-$2,347
-$877
$1,673
$1,005

$242
$220
-$89
$167
$210
$120
$309
$1,227
$119
-$2,347
-$877
$1,639
$942

$242
$220
-$89
$167
$210
$120
$309
$1,162
$119
-$2,347
-$877
$1,562
$800

$242
$220
-$89
$167
$210
$120
$309
$1,134
$119
Small truck

$240
$202
-$89
$167
$207
$110
$309
$1,126
$119
-$1,849
-$877
$1,520
$1,186

$240
$202
-$89
$167
$207
$110
$309
$1,105
$119
-$1,849
-$877
$1,494
$1,139

$240
$202
-$89
$167
$207
$110
$309
$1,047
$119
-$1,849
-$877
$1,426
$1,013

$240
$202
-$89
$167
$207
$110
$309
$1,018
$119
                                        3-172

-------
                                             Technologies Considered in the Agencies' Analysis
Delete 1C engine
Delete transmission
Motor assembly
Total
20% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
-$1,565
-$877
$922
$258

$233
$154
-$62
$115
$203
$85
$309
$599
$119
-$1,565
-$877
$896
$210
-$1,565
-$877
$1,075
$571

$238
$164
-$67
$126
$207
$90
$309
$715
$119
-$1,565
-$877
$1,033
$493
-$2,418
-$877
$1,716
$938

$242
$176
-$85
$157
$210
$97
$309
$1,235
$119
-$2,418
-$877
$1,648
$812
-$2,347
-$877
$1,528
$737

$242
$220
-$89
$167
$210
$120
$309
$1,083
$119
-$2,347
-$877
$1,468
$627
-$1,849
-$877
$1,392
$950

$240
$202
-$89
$167
$207
$110
$309
$1,004
$119
-$1,849
-$877
$1,374
$918
a The agencies have not estimated PHEV or EV costs for the minivan+towing and large truck vehicle classes
vehicle classes would use the technologies.
                                                                                  since we do not believe these
                                                      3-173

-------
                                 Technologies Considered in the Agencies' Analysis
Table 3-85 Scaled Non-battery DMC by Applied Vehicle Weight Reduction for EV150 (2009$)
System
0%WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
2%WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
7.5% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
70% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Subcompact

$233
$154
-$62
$115
$203
$85
$309
$693
$119
-$1,565
-$877
$1,007
$415

$233
$154
-$62
$115
$203
$85
$309
$679
$119
-$1,565
-$877
$990
$384

$233
$154
-$62
$115
$203
$85
$309
$679
$119
-$1,565
-$877
$990
$384

$233
$154
-$62
$115
$203
$85
$309
$679
$119
-$1,565
-$877
Small car

$238
$164
-$67
$126
$207
$90
$309
$830
$119
-$1,565
-$877
$1,169
$745

$238
$164
-$67
$126
$207
$90
$309
$816
$119
-$1,565
-$877
$1,152
$713

$238
$164
-$67
$126
$207
$90
$309
$809
$119
-$1,565
-$877
$1,144
$697

$238
$164
-$67
$126
$207
$90
$309
$809
$119
-$1,565
-$877
Large car

$242
$176
-$85
$157
$210
$97
$309
$1,437
$119
-$2,418
-$877
$1,887
$1,254

$242
$176
-$85
$157
$210
$97
$309
$1,408
$119
-$2,418
-$877
$1,853
$1,191

$242
$176
-$85
$157
$210
$97
$309
$1,401
$119
-$2,418
-$877
$1,844
$1,175

$242
$176
-$85
$157
$210
$97
$309
$1,401
$119
-$2,418
-$877
Minivan

$242
$220
-$89
$167
$210
$120
$309
$1,256
$119
-$2,347
-$877
$1,673
$1,005

$242
$220
-$89
$167
$210
$120
$309
$1,227
$119
-$2,347
-$877
$1,639
$942

$242
$220
-$89
$167
$210
$120
$309
$1,227
$119
-$2,347
-$877
$1,639
$942

$242
$220
-$89
$167
$210
$120
$309
$1,227
$119
-$2,347
-$877
Small truck

$240
$202
-$89
$167
$207
$110
$309
$1,141
$119
-$1,849
-$877
$1,537
$1,218

$240
$202
-$89
$167
$207
$110
$309
$1,141
$119
-$1,849
-$877
$1,537
$1,218

$240
$202
-$89
$167
$207
$110
$309
$1,141
$119
-$1,849
-$877
$1,537
$1,218

$240
$202
-$89
$167
$207
$110
$309
$1,141
$119
-$1,849
-$877
                                        3-174

-------
                                         Technologies Considered in the Agencies' Analysis
Motor assembly
Total
20% WR
Brake system
Climate controls
Delete electrical
DC-DC converter
High voltage wiring
Supplemental heater
On vehicle charger
Motor inverter
Controls
Delete 1C engine
Delete transmission
Motor assembly
Total
$990
$384

$233
$154
-$62
$115
$203
$85
$309
$679
$119
-$1,565
-$877
$990
$384
$1,144
$697

$238
$164
-$67
$126
$207
$90
$309
$809
$119
-$1,565
-$877
$1,144
$697
$1,844
$1,175

$242
$176
-$85
$157
$210
$97
$309
$1,401
$119
-$2,418
-$877
$1,844
$1,175
$1,639
$942

$242
$220
-$89
$167
$210
$120
$309
$1,227
$119
-$2,347
-$877
$1,639
$942
$1,537
$1,218

$240
$202
-$89
$167
$207
$110
$309
$1,141
$119
-$1,849
-$877
$1,537
$1,218
a The agencies have not estimated PHEV or EV costs for the minivan+towing and large truck vehicle classes since we do not believe these
vehicle classes would use the technologies.

        Similar to the approach taken for battery pack costs, the agencies generated linear
regressions of non-battery system costs against percent of net mass reduction and the results
are shown in Table 3-86. This was done using the same weight reduction offsets as used for
battery packs as presented in Table 3-72.  The agencies separated battery pack costs from the
remainder of the systems for each type of electrified vehicle.  The advantage of separating the
battery pack costs from other system costs is that it allows each to carry unique indirect cost
multipliers and learning effects which are important given that battery technology is an
emerging technology, while electric motors and inverters are more stable technologies.

 Table 3-86 Linear Regressions of Non-Battery System Direct Manufacturing Costs vs Net Mass reduction
                                             (2009$)
Vehicle
Class
Subcompact
Small car
Large car
Minivan
Small truck
Minivan+towing
Large truck
P2HEV
$323x+$l,691
$321x+$l,771
$581x+$2,046
$466x+$2,024
$428x+$l,948
$492x+$2,024
$488x+$2,079
PHEV20
$l,478x+$2,946
$l,602x+$3,251
$2,930x+$4,499
$2,433x+$4,196
$2,186x+$3,912


PHEV40
$l,473x+$2,947
$l,613x+$3,250
$2,860x+$4,498
$2,441x+$4,196
$2,201x+$3,912


EV75
-$l,505x+$411
-$l,803x+$749
$3,180x+$l,255
$2,687x+$ 1,002
$2,390x+$l,188


EV100
-$l,535x+$413
-$l,787x+$748
$3,137x+$l,253
$2,696x+$ 1,002
$2,383x+$l,187


EV150
-$l,976x+$415
-$l,924x+$746
$3,278x+$l,254
$2,969x+$ 1,005
$1,218


Notes:
"x" in the equations represents the net weight reduction as a percentage, so the non-battery components for a subcompact P2 HEV with a
20% applied weight reduction and, therefore, a 15% net weight reduction would cost (-$323)x(15%)+$l,691=$l,643.
The small truck EV150 regression has no slope since the net weight reduction is always 0 due to the 20% weight reduction hit.
The agencies did not regress PHEV or EV costs for the minivan+towing and large truck vehicle classes since we do not believe these vehicle
classes would use the technologies.
        For P2 HEV non-battery components, the direct manufacturing costs shown in Table
3-86 are considered applicable to the 2017MY. The agencies consider the P2 non-battery
                                                 3-175

-------
                                   Technologies Considered in the Agencies' Analysis
component technologies to be on the flat portion of the learning curve during the 2017-2025
timeframe. The agencies have applied a highl complexity ICM of 1.56 through 2018 then
1.35 thereafter. For PHEV and EV non-battery components, the direct manufacturing costs
shown in Table 3-86 are considered applicable to the 2025MY.  The agencies consider the
PHEV and EV non-battery component technologies to be on the flat portion of the learning
curve during the 2017-2025 timeframe. The agencies have applied a high2 complexity ICM
of 1.77 through 2024 then 1.50  thereafter. The resultant costs for P2 HEV, PHEV20,
PHEV40, EV75, EV100 and EV150 non-battery components are shown in Table 3-87
through Table 3-92, respectively.
                Table 3-87 Costs for P2 HEV Non-Battery Components (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
Vehicle class
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Minivan-
towing
Minivan-
towing
Minivan-
towing
Large truck
Large truck
Large truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Minivan-
Applied
WR
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
Net
WR
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
4%
9%
14%
4%
9%
14%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
4%
2017
$1,453
$1,439
$1,425
$1,523
$1,509
$1,495
$1,750
$1,725
$1,699
$1,735
$1,715
$1,695
$1,672
$1,653
$1,635
$1,739
$1,718
$1,697
$1,786
$1,765
$1,744
$930
$921
$912
$974
$965
$956
$1,119
$1,103
$1,087
$1,110
$1,097
$1,084
$1,069
$1,058
$1,046
$1,113
2018
$1,424
$1,410
$1,397
$1,492
$1,479
$1,465
$1,715
$1,690
$1,665
$1,701
$1,681
$1,661
$1,638
$1,620
$1,602
$1,704
$1,684
$1,663
$1,751
$1,730
$1,709
$928
$919
$910
$972
$963
$954
$1,117
$1,101
$1,085
$1,108
$1,095
$1,082
$1,067
$1,055
$1,044
$1,110
2019
$1,396
$1,382
$1,369
$1,463
$1,449
$1,436
$1,681
$1,656
$1,632
$1,667
$1,647
$1,628
$1,605
$1,588
$1,570
$1,670
$1,650
$1,629
$1,716
$1,695
$1,675
$570
$564
$559
$597
$591
$586
$686
$676
$666
$680
$672
$664
$655
$648
$641
$682
2020
$1,368
$1,355
$1,341
$1,433
$1,420
$1,407
$1,647
$1,623
$1,599
$1,633
$1,614
$1,595
$1,573
$1,556
$1,538
$1,637
$1,617
$1,597
$1,681
$1,661
$1,641
$569
$563
$558
$596
$591
$585
$685
$675
$665
$679
$671
$663
$654
$647
$640
$681
2021
$1,340
$1,327
$1,315
$1,405
$1,392
$1,379
$1,614
$1,591
$1,567
$1,601
$1,582
$1,563
$1,542
$1,525
$1,508
$1,604
$1,585
$1,565
$1,648
$1,628
$1,609
$568
$562
$557
$595
$590
$584
$684
$674
$664
$678
$670
$662
$653
$646
$639
$680
2022
$1,314
$1,301
$1,288
$1,377
$1,364
$1,351
$1,582
$1,559
$1,536
$1,569
$1,550
$1,532
$1,511
$1,494
$1,477
$1,572
$1,553
$1,534
$1,615
$1,596
$1,576
$567
$562
$556
$594
$589
$583
$683
$673
$663
$677
$669
$661
$652
$645
$638
$679
2023
$1,287
$1,275
$1,262
$1,349
$1,337
$1,324
$1,550
$1,528
$1,505
$1,537
$1,519
$1,501
$1,481
$1,464
$1,448
$1,541
$1,522
$1,503
$1,582
$1,564
$1,545
$566
$561
$555
$593
$588
$582
$682
$672
$662
$676
$668
$660
$651
$644
$637
$678
2024
$1,262
$1,249
$1,237
$1,322
$1,310
$1,298
$1,519
$1,497
$1,475
$1,506
$1,489
$1,471
$1,451
$1,435
$1,419
$1,510
$1,491
$1,473
$1,551
$1,532
$1,514
$565
$560
$554
$592
$587
$582
$681
$671
$661
$675
$667
$659
$650
$643
$636
$677
2025
$1,236
$1,224
$1,212
$1,296
$1,284
$1,272
$1,489
$1,467
$1,446
$1,476
$1,459
$1,442
$1,422
$1,406
$1,391
$1,480
$1,462
$1,443
$1,520
$1,502
$1,484
$565
$559
$554
$592
$586
$581
$680
$670
$660
$674
$666
$658
$649
$642
$635
$676
                                          3-176

-------
                                          Technologies Considered in the Agencies' Analysis

1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
towing
Minivan-
towing
Minivan-
towing
Large truck
Large truck
Large truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Minivan-
towing
Minivan-
towing
Minivan-
towing
Large truck
Large truck
Large truck

15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%

9%
14%
4%
9%
14%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
5%
10%
15%
4%
9%
14%
4%
9%
14%

$1,099
$1,085
$1,143
$1,129
$1,116
$2,383
$2,360
$2,337
$2,497
$2,474
$2,451
$2,869
$2,828
$2,787
$2,846
$2,812
$2,779
$2,741
$2,711
$2,680
$2,852
$2,817
$2,782
$2,929
$2,895
$2,860

$1,097
$1,083
$1,141
$1,127
$1,113
$2,352
$2,329
$2,307
$2,465
$2,442
$2,419
$2,832
$2,791
$2,750
$2,809
$2,776
$2,743
$2,706
$2,675
$2,645
$2,815
$2,780
$2,746
$2,891
$2,857
$2,823

$673
$665
$700
$692
$684
$1,965
$1,946
$1,927
$2,059
$2,041
$2,022
$2,366
$2,332
$2,298
$2,347
$2,320
$2,292
$2,261
$2,236
$2,211
$2,352
$2,323
$2,294
$2,416
$2,387
$2,359

$672
$664
$699
$691
$683
$1,936
$1,918
$1,899
$2,029
$2,011
$1,992
$2,332
$2,298
$2,265
$2,312
$2,286
$2,259
$2,228
$2,203
$2,178
$2,318
$2,289
$2,261
$2,381
$2,352
$2,324

$671
$663
$698
$690
$682
$1,908
$1,890
$1,871
$2,000
$1,981
$1,963
$2,298
$2,265
$2,232
$2,279
$2,252
$2,226
$2,195
$2,171
$2,146
$2,284
$2,256
$2,228
$2,346
$2,318
$2,290

$670
$662
$697
$689
$681
$1,881
$1,862
$1,844
$1,971
$1,953
$1,935
$2,264
$2,232
$2,199
$2,246
$2,220
$2,193
$2,163
$2,139
$2,115
$2,251
$2,223
$2,196
$2,312
$2,284
$2,257

$669
$661
$696
$688
$680
$1,853
$1,836
$1,818
$1,942
$1,925
$1,907
$2,232
$2,200
$2,168
$2,213
$2,188
$2,162
$2,132
$2,108
$2,085
$2,218
$2,191
$2,164
$2,279
$2,252
$2,224

$668
$660
$695
$687
$679
$1,827
$1,809
$1,792
$1,914
$1,897
$1,879
$2,200
$2,168
$2,136
$2,182
$2,156
$2,131
$2,102
$2,078
$2,055
$2,187
$2,160
$2,133
$2,246
$2,219
$2,193

$667
$659
$694
$686
$678
$1,801
$1,784
$1,766
$1,887
$1,870
$1,853
$2,168
$2,137
$2,106
$2,151
$2,126
$2,100
$2,072
$2,049
$2,026
$2,155
$2,129
$2,102
$2,214
$2,188
$2,161
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                     Table 3-88 Costs for PHEV20 Non-Battery Components (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
Vehicle
class
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Applied
WR
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
NetWR
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
2017
$2,518
$2,454
$2,389
$2,779
$2,709
$2,640
$3,827
$3,700
$3,573
$3,577
$3,471
$3,366
$3,337
$3,243
$3,148
$1,611
$1,570
2018
$2,467
$2,404
$2,342
$2,723
$2,655
$2,587
$3,751
$3,626
$3,502
$3,505
$3,402
$3,298
$3,271
$3,178
$3,085
$1,607
$1,566
2019
$2,418
$2,356
$2,295
$2,669
$2,602
$2,535
$3,676
$3,554
$3,431
$3,435
$3,334
$3,232
$3,205
$3,114
$3,023
$987
$962
2020
$2,370
$2,309
$2,249
$2,615
$2,550
$2,484
$3,602
$3,482
$3,363
$3,366
$3,267
$3,168
$3,141
$3,052
$2,963
$985
$960
2021
$2,322
$2,263
$2,204
$2,563
$2,499
$2,435
$3,530
$3,413
$3,296
$3,299
$3,202
$3,104
$3,078
$2,991
$2,903
$984
$959
2022
$2,276
$2,218
$2,160
$2,512
$2,449
$2,386
$3,459
$3,345
$3,230
$3,233
$3,138
$3,042
$3,017
$2,931
$2,845
$982
$957
2023
$2,230
$2,173
$2,117
$2,461
$2,400
$2,338
$3,390
$3,278
$3,165
$3,168
$3,075
$2,982
$2,956
$2,872
$2,788
$981
$956
2024
$2,186
$2,130
$2,074
$2,412
$2,352
$2,292
$3,322
$3,212
$3,102
$3,105
$3,014
$2,922
$2,897
$2,815
$2,733
$980
$955
2025
$2,142
$2,087
$2,033
$2,364
$2,305
$2,246
$3,256
$3,148
$3,040
$3,043
$2,953
$2,863
$2,839
$2,759
$2,678
$978
$953
                                                 3-177

-------
                                          Technologies Considered in the Agencies' Analysis
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
3%
8%
13%
$1,529
$1,778
$1,733
$1,689
$2,448
$2,367
$2,286
$2,288
$2,221
$2,153
$2,135
$2,074
$2,014
$4,128
$4,023
$3,918
$4,556
$4,442
$4,328
$6,276
$6,067
$5,859
$5,865
$5,692
$5,519
$5,472
$5,317
$5,161
$1,526
$1,774
$1,730
$1,685
$2,443
$2,362
$2,281
$2,284
$2,216
$2,149
$2,131
$2,070
$2,010
$4,075
$3,971
$3,867
$4,497
$4,385
$4,272
$6,194
$5,988
$5,783
$5,789
$5,618
$5,447
$5,401
$5,248
$5,094
$937
$1,089
$1,062
$1,035
$1,500
$1,450
$1,401
$1,402
$1,361
$1,319
$1,308
$1,271
$1,234
$3,405
$3,318
$3,231
$3,758
$3,664
$3,570
$5,176
$5,004
$4,832
$4,837
$4,695
$4,552
$4,513
$4,385
$4,257
$935
$1,088
$1,060
$1,033
$1,498
$1,448
$1,398
$1,400
$1,359
$1,317
$1,306
$1,269
$1,232
$3,355
$3,270
$3,184
$3,703
$3,610
$3,517
$5,100
$4,931
$4,761
$4,766
$4,626
$4,485
$4,447
$4,321
$4,195
$934
$1,086
$1,059
$1,032
$1,496
$1,446
$1,396
$1,398
$1,357
$1,315
$1,304
$1,267
$1,230
$3,306
$3,222
$3,138
$3,649
$3,558
$3,466
$5,026
$4,859
$4,692
$4,697
$4,558
$4,420
$4,382
$4,258
$4,133
$932
$1,084
$1,057
$1,030
$1,493
$1,444
$1,394
$1,396
$1,355
$1,313
$1,302
$1,265
$1,228
$3,258
$3,175
$3,092
$3,596
$3,506
$3,416
$4,953
$4,788
$4,624
$4,629
$4,492
$4,356
$4,319
$4,196
$4,073
$931
$1,083
$1,056
$1,028
$1,491
$1,442
$1,392
$1,394
$1,353
$1,311
$1,300
$1,263
$1,226
$3,211
$3,129
$3,048
$3,544
$3,455
$3,367
$4,881
$4,719
$4,557
$4,562
$4,428
$4,293
$4,257
$4,136
$4,015
$930
$1,081
$1,054
$1,027
$1,489
$1,440
$1,390
$1,392
$1,351
$1,309
$1,298
$1,262
$1,225
$3,165
$3,085
$3,004
$3,493
$3,406
$3,318
$4,811
$4,652
$4,492
$4,497
$4,364
$4,231
$4,196
$4,076
$3,957
$928
$1,080
$1,053
$1,026
$1,487
$1,438
$1,388
$1,390
$1,349
$1,308
$1,297
$1,260
$1,223
$3,120
$3,041
$2,961
$3,443
$3,357
$3,271
$4,743
$4,585
$4,428
$4,433
$4,302
$4,171
$4,136
$4,018
$3,901
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                     Table 3-89 Costs for PHEV40 Non-Battery Components (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
Vehicle
class
Subcompact
Subcompact
Small car
Small car
Large car
Large car
Minivan
Minivan
Small truck
Small truck
Subcompact
Subcompact
Small car
Small car
Large car
Large car
Minivan
Minivan
Small truck
Small truck
Applied
WR
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
Net
WR
2%
7%
3%
8%
2%
7%
2%
7%
3%
8%
2%
7%
3%
8%
2%
7%
2%
7%
3%
8%
2017
$2,531
$2,467
$2,778
$2,708
$3,853
$3,729
$3,598
$3,492
$3,336
$3,241
$1,619
$1,578
$1,777
$1,732
$2,465
$2,385
$2,302
$2,234
$2,134
$2,073
2018
$2,481
$2,418
$2,722
$2,654
$3,776
$3,654
$3,526
$3,422
$3,270
$3,176
$1,616
$1,575
$1,773
$1,729
$2,460
$2,381
$2,297
$2,230
$2,130
$2,069
2019
$2,431
$2,370
$2,668
$2,601
$3,700
$3,581
$3,456
$3,354
$3,204
$3,113
$992
$967
$1,089
$1,061
$1,510
$1,462
$1,410
$1,369
$1,308
$1,270
2020
$2,382
$2,322
$2,614
$2,549
$3,626
$3,509
$3,387
$3,287
$3,140
$3,050
$991
$966
$1,087
$1,060
$1,508
$1,459
$1,408
$1,367
$1,306
$1,268
2021
$2,335
$2,276
$2,562
$2,498
$3,554
$3,439
$3,319
$3,221
$3,077
$2,989
$989
$964
$1,086
$1,058
$1,506
$1,457
$1,406
$1,365
$1,304
$1,266
2022
$2,288
$2,230
$2,511
$2,448
$3,483
$3,370
$3,252
$3,157
$3,016
$2,930
$988
$963
$1,084
$1,057
$1,503
$1,455
$1,404
$1,363
$1,302
$1,265
2023
$2,242
$2,186
$2,461
$2,399
$3,413
$3,303
$3,187
$3,094
$2,955
$2,871
$986
$961
$1,082
$1,055
$1,501
$1,453
$1,402
$1,361
$1,300
$1,263
2024
$2,197
$2,142
$2,411
$2,351
$3,345
$3,237
$3,124
$3,032
$2,896
$2,813
$985
$960
$1,081
$1,054
$1,499
$1,451
$1,400
$1,359
$1,298
$1,261
2025
$2,153
$2,099
$2,363
$2,304
$3,278
$3,172
$3,061
$2,971
$2,838
$2,757
$983
$959
$1,079
$1,052
$1,497
$1,449
$1,398
$1,357
$1,296
$1,259
                                                 3-178

-------
                                          Technologies Considered in the Agencies' Analysis
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
Subcompact
Subcompact
Small car
Small car
Large car
Large car
Minivan
Minivan
Small truck
Small truck
15%
20%
15%
20%
15%
20%
15%
20%
15%
20%
2%
7%
3%
8%
2%
7%
2%
7%
3%
8%
$4,150
$4,046
$4,555
$4,440
$6,317
$6,114
$5,900
$5,726
$5,471
$5,314
$4,096
$3,993
$4,496
$4,382
$6,235
$6,035
$5,824
$5,652
$5,400
$5,245
$3,423
$3,337
$3,757
$3,662
$5,210
$5,043
$4,866
$4,723
$4,512
$4,383
$3,373
$3,288
$3,702
$3,608
$5,134
$4,969
$4,795
$4,654
$4,446
$4,319
$3,324
$3,240
$3,648
$3,556
$5,059
$4,896
$4,725
$4,586
$4,381
$4,256
$3,276
$3,193
$3,595
$3,504
$4,986
$4,825
$4,657
$4,519
$4,318
$4,194
$3,228
$3,147
$3,543
$3,454
$4,914
$4,756
$4,589
$4,454
$4,255
$4,134
$3,182
$3,102
$3,492
$3,404
$4,844
$4,688
$4,524
$4,390
$4,194
$4,074
$3,137
$3,058
$3,442
$3,356
$4,775
$4,621
$4,459
$4,328
$4,135
$4,016
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                       Table 3-90 Costs for EV75 Non-Battery Components (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
Vehicle
class
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Applied
WR
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
Net
WR
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
2017
$261
$185
$110
$569
$479
$389
$937
$778
$619
$733
$599
$464
$949
$829
$710
$201
$143
$85
$438
$369
$299
$721
$599
$476
$565
$461
$358
$731
$639
$547
$461
$328
$195
$1,007
$847
$688
$1,658
$1,376
2018
$253
$180
$107
$552
$464
$377
$909
$754
$600
$711
$581
$451
$920
$805
$689
$200
$142
$85
$437
$368
$298
$719
$597
$475
$563
$460
$357
$729
$637
$545
$453
$322
$191
$989
$832
$675
$1,628
$1,351
2019
$245
$174
$104
$535
$450
$366
$881
$732
$582
$690
$563
$437
$893
$780
$668
$200
$142
$84
$436
$367
$298
$717
$595
$474
$561
$459
$356
$727
$635
$544
$445
$316
$188
$971
$817
$663
$1,598
$1,327
2020
$238
$169
$100
$519
$437
$355
$855
$710
$565
$669
$547
$424
$866
$757
$648
$199
$142
$84
$434
$366
$297
$715
$594
$472
$560
$457
$355
$725
$633
$542
$437
$311
$185
$954
$802
$651
$1,570
$1,304
2021
$231
$164
$97
$504
$424
$344
$829
$688
$548
$649
$530
$411
$840
$734
$629
$198
$141
$84
$433
$365
$296
$713
$592
$471
$558
$456
$354
$723
$632
$541
$429
$305
$181
$937
$788
$640
$1,543
$1,281
2022
$224
$159
$95
$488
$411
$334
$804
$668
$531
$630
$514
$399
$815
$712
$610
$198
$141
$84
$432
$364
$295
$712
$591
$470
$557
$455
$353
$721
$630
$539
$422
$300
$178
$921
$775
$629
$1,516
$1,258
2023
$219
$156
$93
$479
$403
$327
$788
$654
$521
$617
$504
$391
$799
$698
$598
$198
$141
$83
$431
$363
$295
$710
$590
$469
$556
$454
$352
$720
$629
$538
$417
$297
$176
$910
$766
$622
$1,499
$1,244
2024
$215
$153
$91
$469
$395
$320
$772
$641
$510
$605
$494
$383
$783
$684
$586
$197
$140
$83
$431
$362
$294
$709
$589
$468
$555
$453
$352
$718
$628
$538
$412
$293
$174
$900
$757
$615
$1,482
$1,230
2025
$211
$150
$89
$460
$387
$314
$757
$629
$500
$593
$484
$375
$767
$670
$574
$127
$90
$54
$277
$233
$189
$456
$379
$301
$357
$292
$226
$462
$404
$346
$338
$240
$143
$737
$620
$503
$1,213
$1,007
                                                 3-179

-------
                                           Technologies Considered in the Agencies' Analysis
TC
TC
TC
TC
TC
TC
TC
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
20%
10%
15%
20%
10%
15%
20%
20%
10%
15%
20%
10%
15%
20%
$1,095
$1,298
$1,060
$822
$1,680
$1,468
$1,257
$1,075
$1,274
$1,041
$807
$1,649
$1,441
$1,234
$1,056
$1,251
$1,022
$793
$1,619
$1,416
$1,212
$1,037
$1,229
$1,004
$779
$1,591
$1,390
$1,190
$1,019
$1,207
$986
$765
$1,563
$1,366
$1,169
$1,001
$1,187
$969
$752
$1,536
$1,342
$1,149
$990
$1,173
$958
$743
$1,518
$1,327
$1,136
$979
$1,160
$947
$735
$1,501
$1,312
$1,123
$801
$950
$776
$602
$1,229
$1,075
$920
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                       Table 3-91 Costs for EV100 Non-Battery Components (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
TC
Vehicle
class
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Minivan
Small truck
Small truck
Small truck
Subcompact
Subcompact
Subcompact
Small car
Small car
Small car
Large car
Large car
Large car
Minivan
Minivan
Applied
WR
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
20%
10%
15%
Net
WR
4%
9%
14%
5%
10%
15%
5%
10%
15%
4%
9%
14%
2%
7%
12%
4%
9%
14%
5%
10%
15%
5%
10%
15%
4%
9%
14%
2%
7%
12%
4%
9%
14%
5%
10%
15%
5%
10%
15%
4%
9%
2017
$351
$275
$198
$659
$569
$480
$1,096
$939
$783
$894
$759
$625
$1,140
$1,020
$901
$271
$211
$152
$507
$438
$370
$844
$723
$603
$689
$585
$481
$877
$786
$694
$622
$486
$350
$1,166
$1,008
$850
$1,940
$1,663
$1,385
$1,583
$1,344
2018
$341
$266
$192
$639
$552
$466
$1,063
$911
$759
$867
$737
$606
$1,105
$990
$874
$270
$211
$152
$506
$437
$369
$842
$721
$601
$687
$583
$480
$875
$783
$692
$611
$477
$344
$1,145
$990
$834
$1,905
$1,632
$1,360
$1,554
$1,320
2019
$331
$258
$186
$620
$536
$452
$1,031
$884
$736
$841
$714
$588
$1,072
$960
$848
$269
$210
$151
$504
$436
$368
$839
$719
$599
$685
$581
$478
$873
$781
$690
$600
$469
$338
$1,124
$972
$819
$1,871
$1,603
$1,335
$1,526
$1,296
2020
$321
$251
$181
$601
$520
$438
$1,001
$857
$714
$816
$693
$570
$1,040
$931
$822
$268
$210
$151
$503
$435
$367
$837
$717
$598
$683
$580
$477
$870
$779
$688
$589
$460
$332
$1,104
$955
$805
$1,838
$1,575
$1,312
$1,499
$1,273
2021
$311
$243
$175
$583
$504
$425
$970
$832
$693
$792
$672
$553
$1,009
$903
$798
$268
$209
$151
$502
$434
$366
$835
$715
$596
$681
$578
$476
$868
$777
$686
$579
$452
$326
$1,085
$938
$791
$1,805
$1,547
$1,289
$1,473
$1,251
2022
$302
$236
$170
$566
$489
$412
$941
$807
$672
$768
$652
$536
$979
$876
$774
$267
$209
$150
$500
$433
$365
$833
$714
$594
$679
$577
$474
$866
$775
$685
$569
$444
$320
$1,066
$922
$777
$1,774
$1,520
$1,266
$1,447
$1,229
2023
$296
$231
$166
$554
$479
$404
$923
$791
$659
$752
$639
$526
$959
$859
$758
$266
$208
$150
$500
$432
$364
$831
$712
$593
$678
$576
$474
$864
$774
$683
$562
$439
$317
$1,054
$911
$768
$1,754
$1,503
$1,252
$1,431
$1,215
2024
$290
$226
$163
$543
$470
$396
$904
$775
$645
$737
$626
$515
$940
$842
$743
$266
$208
$150
$499
$431
$363
$830
$711
$592
$677
$575
$473
$863
$773
$682
$556
$434
$313
$1,042
$901
$759
$1,734
$1,486
$1,238
$1,414
$1,201
2025
$284
$222
$160
$532
$460
$388
$886
$759
$632
$723
$614
$505
$921
$825
$728
$171
$134
$96
$321
$277
$234
$534
$458
$381
$436
$370
$304
$555
$497
$439
$455
$356
$256
$853
$738
$622
$1,420
$1,217
$1,014
$1,158
$984
                                                  3-180

-------
                                          Technologies Considered in the Agencies' Analysis
TC
TC
TC
TC
Minivan
Small truck
Small truck
Small truck
20%
10%
15%
20%
14%
2%
7%
12%
$1,105
$2,017
$1,806
$1,595
$1,085
$1,980
$1,773
$1,566
$1,066
$1,945
$1,741
$1,538
$1,047
$1,910
$1,710
$1,511
$1,029
$1,877
$1,680
$1,484
$1,011
$1,844
$1,651
$1,458
$999
$1,823
$1,633
$1,442
$988
$1,803
$1,614
$1,426
$809
$1,476
$1,322
$1,167
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                       Table 3-92 Costs for EV150 Non-Battery Components (2009$)
Cost
type
DMC
DMC
DMC
DMC
DMC
1C
1C
1C
1C
1C
TC
TC
TC
TC
TC
Vehicle
class
Subcompact
Small car
Large car
Minivan
Small truck
Subcompact
Small car
Large car
Minivan
Small truck
Subcompact
Small car
Large car
Minivan
Small truck
Applied
WR
20%
20%
20%
20%
20%
20%
20%
20%
20%
20%
20%
20%
20%
20%
20%
Net
WR
2%
3%
2%
2%
0%
2%
3%
2%
2%
0%
2%
3%
2%
2%
0%
2017
$376
$688
$1,188
$946
$1,218
$289
$530
$915
$728
$938
$665
$1,218
$2,104
$1,674
$2,155
2018
$365
$667
$1,153
$917
$1,181
$289
$528
$913
$726
$935
$653
$1,196
$2,065
$1,643
$2,116
2019
$354
$647
$1,118
$890
$1,146
$288
$527
$910
$724
$932
$641
$1,174
$2,028
$1,614
$2,078
2020
$343
$628
$1,085
$863
$1,111
$287
$525
$907
$722
$930
$630
$1,153
$1,992
$1,585
$2,041
2021
$333
$609
$1,052
$837
$1,078
$286
$524
$905
$720
$927
$619
$1,133
$1,957
$1,557
$2,005
2022
$323
$591
$1,021
$812
$1,046
$285
$523
$903
$718
$925
$608
$1,114
$1,923
$1,530
$1,971
2023
$316
$579
$1,000
$796
$1,025
$285
$522
$901
$717
$924
$601
$1,101
$1,901
$1,513
$1,948
2024
$310
$567
$980
$780
$1,004
$285
$521
$900
$716
$922
$594
$1,088
$1,880
$1,496
$1,926
2025
$304
$556
$961
$764
$984
$183
$335
$579
$461
$593
$487
$891
$1,540
$1,225
$1,578
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
     3.4.4   Hardware costs for charging grid-connected vehicles

            Grid-connected vehicles such as EVs and PHEVs require a means to charge their on-
     board batteries to enable their electric range capabilities. These vehicles require certain
     hardware to charge, both on-vehicle and off-vehicle. The agencies' September 2010 Technical
     Assessment Report contains an in-depth analysis of the topic of charging and infrastructure.
     The TAR analysis and assumptions did not receive any significant comment, and a review of
     the current state of the industry indicates the assumptions in the TAR are still valid.
     Therefore, the assumptions for the cost of Electric Vehicle Support Equipment (EVSE) are
     unchanged. Additionally, while some of the characteristics of the modeled grid-connected
     vehicles such as battery size and energy demand have changed somewhat due to further
     analysis, the application of Level  1 and Level 2 charging by vehicle type based on charge time
     has not changed.

            Three charging levels are currently under consideration. Level 1 charging uses a
     standard 120 volt (V), 15-20 amps (A) rated (12-16 A usable) circuit and is available in
     standard residential and commercial buildings.  Level 2 charging uses a single phase, 240 V,
     20-80 A circuit and allows much shorter charge times.  Level 3 charging—sometimes
     colloquially called "quick" or "fast" charging—uses a 480 V, three-phase circuit, available in
     mainly industrial areas, typically providing 60-150 kW of off-board charging power. It is
                                                 3-181

-------
                                     Technologies Considered in the Agencies' Analysis
expected that 97 to 99% of charging will take place at home, so a cost for a home charger,
appropriate to the duty cycle of the vehicle, is added to the vehicle cost. Level 3 charging is
available to commercial users and vehicles that charge at Level 3 stations will be assumed to
pay at the charge station for the convenience of fast charging. Therefore Level 3 charger
costs are not included in overall vehicle cost.

       The specific equipment required for charging a grid-connected vehicle consists of the
following:

        Charger:  A charger that converts electricity from alternating current (AC) from the
electricity source to direct current (DC) required for the battery, and also converts the
incoming 120 or 240 volt current to 300 or higher volts.  Grid-connected vehicles carry an on-
board charger capable of accepting AC current from a wall plug (Level 1 circuit) or, from a
Level 2 charging station. On-board charger power capability ranges from 1.4 to  10 kW and is
usually proportional to the vehicle's battery capacity. The lowest charging power, 1.4 kW, is
expected only when grid-connected vehicles are connected to 120 volt (Level 1)  outlets, and
all currently known PHEV and EV on-board chargers are expected to provide at  least 3.3 kW
charging when connected to a Level 2 (220 volt, 20+ A) charging station.  The latest SAE
connection recommended practice, J1772, allows for delivery of up to -19 kW to an on-board
vehicle charger. For higher capacity charging under Level 3, a charging station that delivers
DC current directly to the vehicle's battery is incorporated off-board in the wall or pedestal
mounted.

       Charging Station: The charging station needed to safely deliver energy from the
electric circuit to the vehicle, called electric vehicle support equipment (EVSE).  The EVSE
may at a minimum, be a specialized cordset that connects a household Level 1/120V socket to
the vehicle; otherwise, the EVSE will include a cordset and a charging station (a wall or
pedestal mounted box incorporating a charger and other equipment). Charging stations may
include optional advanced features such as timers to delay charging until off-peak hours,
communications equipment to allow the utility to regulate charging, or even electricity
metering capabilities. Stakeholders are working on which features are best located on the
EVSE or on the vehicle itself, and it is possible that redundant capabilities and features may
be present in both the vehicle and EVSEs in the near future until these issues are worked out.
EVSE and vehicle manufacturers are also working to ensure that current SAE-compliant
"basic" EVSEs are charge-compatible with future grid-connected vehicles.

       Dedicated Circuit: A Level 1 circuit is standard household current,  120V AC, rated at
15 or 20 A (12 or 16 A usable). A Level 2 circuit is rated at 208 to 240V and up to 80 A and
is similar to the type of circuit that powers electric  stoves (up to 50 A) and dryers (usually 30
A). Generally, Level 1 and 2 circuits used for electric vehicle recharging must be dedicated
circuits, i.e., there cannot be other appliances on that circuit. For a Level 2 circuit, the
homeowner or other user must install a charging station and will need a permit. A homeowner
may choose to install the charger on a separately-metered circuit to take advantage of special
electrical rates for off-peak charging, where available.
                                            3-182

-------
                                        Technologies Considered in the Agencies' Analysis
       In addition to the costs of purchasing and installing charging equipment, charging
station installation may include the costs of upgrading existing electrical panels and installing
the electrical connection from the panel to the desired station location. These costs may be
dramatically lowered if new construction incorporates the panel box and wiring required for
charging stations, or even includes charging stations or outlets for charging stations as
standard equipment.

       The current costs of charging stations are highly variable depending on the level of
service (and alternative power capabilities within these categories), location (individual
residence, grouped residences, retail or business, parking lot or garage), level of sophistication
of the station, and installation requirements, including electrical upgrading requirements.
Estimated costs for charging  stations are included in Table 3-93 below.

           Table 3-93: Estimated Costs for Charging Stations Used in the 2010 TAR (2008$)
Level
1










2




Location
Single
Residence









Residential,
Apartment
Complex,
or Fleet
Depotb
Equipment
$30- $200 (charge cord only,
included at no cost to consumer
with EV/PHEV) when an
accessible household plug (e.g.,
in a garage or adjacent to a
driveway) with a ground fault
interrupter is already available




3.3 kW EVSE (each): $300-
$4,000

6.6 kW EVSE (each): $400-
$4,000
Installation
$400-$1000+ may be necessary
depending on difficulty of
installing a new circuit at the
desired location, but in most
cases, owners with sufficient
panel capacity would opt for a
more capable 220 VAC Level 2
installation instead of a Level 1
dedicated circuit because the
additional installation cost is
only marginally higher
3.3- 6.6 kW installation cost:
$400-$2,300 without
wiring/service panel upgrade, or
$2,000-$5,000 with panel
upgrade
rets: 73,74,7i,76,a
a Detailed information on charger cost for each charging level and location and specific sources for cost
estimates are available in the TAR, Appendix G.
b Level 2 EVSE installation costs vary considerably for single-family residences, multi-family residences, and
fleet depots, depending upon the need for wiring and service panel upgrades. The range depicted here reflects
the anticipated variability of these costs.  However, EPRI estimates that the typical residential Level 2
installation costs to be approximately $1,500. See the TAR, Appendix G for additional information.

       3.4.4.1  Application of charging level by vehicle type

       The home charging availability for a specific consumer will need to be differentiated
among EV/PHEVs with different battery capacity. The electric  outlets in existing homes are
                                               3-183

-------
                                    Technologies Considered in the Agencies' Analysis
most likely ready for Level 1 charging, which is about sufficient for fully recharging a
PHEV20 SUV during normal nighttime, provided the outlet is not being heavily utilized by
other loads.  Shorter available charging time or owning a PHEV or an EV with a larger
battery make the capability to fully charge overnight with a Level 1 system less likely, but
upgrading to a Level 2 system in such cases will allow full recharge to happen more quickly.

       Table 3-94 shows the application of charge level by vehicle type and range. Charging
types were chosen based on nominal time to charge a fully-depleted battery in a vehicle with
0% net weight reduction.  Charge times exceeding 9 hours for Level 1 were deemed
unacceptable and Level 2 charging was specified.  For charge times between 6 hours and 9
hours on Level 1, a mix of Level 1  and Level 2 was specified. This was done to recognize the
varying consumer value of faster, but more expensive, Level 2 charging over Level 1
charging.

                   Table 3-94: Charger Type by Vehicle Technology and Class

Subcompact
Small Car
Large Car
Minivan
Small Truck
Large Truck
PHEV20
100% LI
100% LI
100% LI
100% LI
100% LI
50% LI
50% L2
PHEV40
25% LI
75% L2
10% LI
90% L2
100% L2
100% L2
100% L2
100% L2
EV75
100% L2
100% L2
100% L2
100% L2
100% L2
100% L2
EV100
100% L2
100% L2
100% L2
100% L2
100% L2
100% L2
EV150
100% L2
100% L2
100% L2
100% L2
100% L2
100% L2
       For this proposal, the resultant costs associated with in-home chargers and installation
of in-home chargers are included in the total cost for an EV and or PHEV. However, here we
summarize specially the costs for chargers and installation labor.  The agencies have
estimated the DMC of a level 1 charge cord at $30 (2009$) based on typical costs of similar
electrical equipment sold to consumers  today and that for a level 2 charger at $202 (2009$).
Labor associated with installing either of these chargers is estimated at $1,009 (2009$).
Further, we have estimated that all PHEV20 vehicles (PHEVs with a 20 mile range) would be
charged via a level 1 charger and that all EVs, regardless of range, would be charged via a
level 2 charger. For the PHEV40 vehicles (PHEVs with a 40 mile range), we have estimated
that: 25% of subcompacts would be charged with a level 1 charger with the remainder
charged via a level 2 charger; 10% of small cars would be charged with a level 1 charger with
the remainder charged via a level 2 charger; and all remaining PHEV 40 vehicles would be
charged via a level 2 charger. All costs  presented here are considered applicable in the  2025
model year. The agencies have applied  the learning curve presented in Section 3.2.3 to all
                                           3-184

-------
                                     Technologies Considered in the Agencies' Analysis
charger costs.  The agencies have also applied a Highl ICM of 1.56 through 2024
thereafter. Installation costs, being labor costs, have no learning impacts or ICMs
The resultant costs are shown in Table 3-95.

                    Table 3-95 Costs for EV/PHEV In-home Chargers (2009$)
then 1.34
applied.
Cost
type
DMC
DMC
DMC
1C
1C
1C
TC
TC
TC
TC
Technology
PHEV20
Charger
PHEV40
Charger
EV
Charger
PHEV20
Charger
PHEV40
Charger
EV
Charger
PHEV20
Charger
PHEV40
Charger
EV
Charger
Charger
labor
Vehicle
Class
All
Subcompact
Small car
Larger car
Minivan
Small truck
All
All
Subcompact
Small car
Larger car
Minivan
Small truck
All
All
Subcompact
Small car
Larger car
Minivan
Small truck
All
All
2017
$59
$311
$361
$394
$394
$19
$99
$115
$126
$126
$78
$410
$476
$521
$521
$1,009
2018
$47
$248
$289
$315
$315
$18
$95
$111
$121
$121
$65
$344
$399
$437
$437
$1,009
2019
$47
$248
$289
$315
$315
$18
$95
$111
$121
$121
$65
$344
$399
$437
$437
$1,009
2020
$38
$199
$231
$252
$252
$18
$92
$107
$117
$117
$55
$291
$338
$369
$369
$1,009
2021
$38
$199
$231
$252
$252
$18
$92
$107
$117
$117
$55
$291
$338
$369
$369
$1,009
2022
$38
$199
$231
$252
$252
$18
$92
$107
$117
$117
$55
$291
$338
$369
$369
$1,009
2023
$38
$199
$231
$252
$252
$18
$92
$107
$117
$117
$55
$291
$338
$369
$369
$1,009
2024
$38
$199
$231
$252
$252
$18
$92
$107
$117
$117
$55
$291
$338
$369
$369
$1,009
2025
$30
$159
$185
$202
$202
$10
$55
$64
$70
$70
$41
$214
$249
$272
$272
$1,009
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
3.4.5   Other Technologies Assessed that Reduce CO2 and Improve Fuel Economy

       In addition to the technologies already mentioned, the technologies generally
considered in the agencies' analysis are described below. They fall into five broad categories:
engine technologies, transmission technologies, vehicle technologies, electrification/accessory
technologies, hybrid technologies and mass reduction

       3.4.5.1 Lower Rolling Resistance Tires

       Tire rolling resistance is the frictional loss associated mainly with the energy
dissipated in the deformation of the tires under load and thus influences fuel economy and
                                            3-185

-------
                                      Technologies Considered in the Agencies' Analysis
COi emissions.  Other tire design characteristics (e.g., materials, construction, and tread
design) influence durability, traction (both wet and dry grip), vehicle handling, and ride
comfort in addition to rolling resistance.  A typical low rolling resistance tire's attributes
could include: increased specified tire inflation pressure, material changes, and tire
construction with less hysteresis, geometry changes (e.g., reduced aspect ratios), and
reduction in sidewall and tread deflection. These changes would generally be accompanied
with additional changes to vehicle suspension tuning and/or suspension design.

       The agencies expect that greater reductions in tire rolling resistance will be possible
during the rulemaking timeframe than are currently available, as tire manufacturers continue
to improve their products in order to meet increasing demand by auto OEMs for tires that
contribute more  to their vehicles'  fuel efficiency. Thus, for this proposal, the agencies are
considering two  "levels" of lower rolling resistance tires.  The  first level ("LRR1") is defined
as a 10 percent reduction in rolling resistance from a base tire,  which was estimated to be a 1
to 2 percent effectiveness improvement MYs 2012-2016 final rule. Based on the 2011
Ricardo study the agencies are now using 1.9%  for all classes.  LRR1 tires are widely
available today,  and appear to comprise a larger and larger portion of tire manufacturers'
product lines as the technology continues to improve and  mature.  The second level ("LRR2")
is defined as a 20 percent reduction in rolling resistance from a base tire, yielding an
estimated 3.9 percent effectiveness.  In the CAFE model this results in a 2.0% incremental
effectiveness increase form LRR1.  LRR2 represents an additional level of rolling resistance
improvement beyond what the agencies considered in the MYs 2012-2016 rulemaking
analysis.

       In the 2012-2016 light duty vehicle rule, the agencies estimated the incremental DMC
at an increase of $5 (2007$) per vehicle.  This included costs associated with five tires per
vehicle, four primary and one spare with  no learning applied due to the commodity based
nature of this technology.  Looking forward from 2016, the agencies continue to apply this
same estimated DMC adjusted for 2009 dollars.00 The agencies consider LRRT1 to be fully
learned out or "off the learning curve (i.e., the  DMC does not change year-over-year) and
have applied a low complexity ICM of 1.24 through 2018, and then 1.19 thereafter, due to the
fact that this technology is already well established in the marketplace.

       To analyze the feasibility and cost for a  second level of rolling resistance
improvement, EPA, NHTSA, and CARB met with a number of the largest tire suppliers in the
United States. The suppliers  were generally optimistic about the ability of tire rolling
resistance to improve in the future without the need to sacrifice traction (safety) or tread life
(durability). Suppliers all generally stated that rolling resistance levels could be reduced by
20 percent relative to today's tires by MY 2017. As such, the agencies agreed, based on these
00 As noted elsewhere in this chapter, we show dollar values to the nearest dollar. However, dollars and cents are
carried through each agency's respective analysis. Thus, while the cost for lower rolling resistance tires in the
2012-2016 final rule was shown as $5, the specific value used in that rule was $5.15 (2007$) and is now $5.31
(2009$).  We show $5 for presentation simplicity.

                                             3-186

-------
                                     Technologies Considered in the Agencies' Analysis
discussions, to consider LRR2 as initially available for purposes of this analysis in MY 2017,
but not widespread in the marketplace until MYs 2022-2023. In alignment with introduction
of new technology, the agencies limited the phase-in schedule to 15 percent of a
manufacturer's fleet starting in 2017, and did not allow complete application (100 percent of a
manufacturer's fleet) until 2023. The agencies believe that this schedule aligns with the
necessary efforts for production implementation such as system and electronic systems
calibration and verification.

       LRR2 technology does not yet exist in the marketplace, making cost estimation
challenging without disclosing potentially confidential business information. To develop a
transparent cost estimate, the agencies relied on LRR1 history, costs, market implementation,
and information provided by the 2010 NAS report.  The agencies assumed low rolling
resistance technology ("LRR1") first entered the marketplace in the 1993 time frame with
more widespread adoption being achieved in recent years, yielding approximately 15 years to
maturity and widespread adoption.

       Then, using MY 2017 as the starting point for market entry for LRR2 and taking into
account the advances in industry knowledge and an assumed increase in demand for
improvements in this technology, the agencies interpolated DMC for LRR2 at $10 (2009$)
per tire, or $40 ($2009) per vehicle. This estimate is generally fairly consistent with CBI
suggestions by tire suppliers. The agencies have not included a cost for the spare tire because
we believe manufacturers are not likely to include a LRR2 as a spare given the $10 DMC. In
some cases and when possible pending any state-level requirements, manufacturers have
removed spare tires replacing them with tire repair kits to reduce both cost and weight
associated with a spare tire.77 The agencies consider this estimated cost for LRR2 to be
applicable in MY 2021.  Further, the agencies consider LRR2 technology to be on the steep
portion of the learning curve where costs would reduce quickly in a relative short amount of
time. The agencies have applied a low complexity  ICM of 1.24 through 2024, and then 1.19
thereafter. The ICM timing for LRR2 is different from that for LRR1 because LRR2 is brand-
new for this rulemaking and is not yet being implemented in the fleet.  The resultant costs are
shown in Table 3-96. Note that both LRR1 and LRR2 are incremental to the baseline system,
so LRR2 is not incremental to LRR1.

              Table 3-96 Costs for Lower Rolling Resistance Tires Levels 1 & 2 (2009$)


Cost type


DMC
DMC
1C
1C
TC
TC
Lower
Rolling
Resistance
Tire
Technology
Level 1
Level 2
Level 1
Level 2
Level 1
Level 2


2017


$5
$63
$1
$10
$6
$72


2018


$5
$63
$1
$10
$6
$72


2019


$5
$50
$1
$10
$6
$60


2020


$5
$50
$1
$10
$6
$60


2021


$5
$40
$1
$10
$6
$50


2022


$5
$39
$1
$10
$6
$48


2023


$5
$38
$1
$10
$6
$47


2024


$5
$37
$1
$10
$6
$46


2025


$5
$35
$1
$8
$6
$43
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
                                           3-187

-------
                                     Technologies Considered in the Agencies' Analysis
Note that both levels of lower rolling resistance tires are incremental to today's baseline tires.
       Given that the proposed standards cover such a long timeframe, the agencies also
considered introducing a third level of rolling resistance reduction ("LRR3"), defined as a 30
percent reduction in rolling resistance. The agencies evaluated the potential of LRR3 entering
the marketplace during this proposed rulemaking timeframe.

       Tire technologies that enable improvements of 10 and 20 percent have been in
existence for many years. Achieving improvements up to 20 percent involves optimizing and
integrating multiple technologies, with a primary contributor being the adoption of a silica
tread technology.78  This approach was based on the use of a new silica along with a specific
polymer and coupling  agent combination.  The use of the polymer, coupling agent and silica
was known to reduce tire rolling resistance at the expense of tread wear, but new approach
novel silica reduced the tread wear tradeoff.

       Tire suppliers have indicated there are one or more innovations/inventions that they
expect to occur in order to move the industry to the next quantum reduction of rolling
resistance. However, based on the historical development and integration of tire technologies,
there appears to be little evidence supporting improvements beyond LRRT2 by 2025.
Therefore, the agencies decided not to incorporate LRRT3 at this time.

       The agencies seek comment, however, on whether we should consider application of a
30 percent reduction from today's rolling resistance levels being available for mass
production implementation by MY 2025 or sooner. The agencies seek comment on the
viability of this technology,  maturity by MY 2025, as well as market introduction timing and
the technological ways that this level of rolling resistance improvement will be achieved
without any tradeoffs in terms of vehicle handling capability and tire life from what
consumers expect today. Finally, the agencies appreciate any cost information regarding the
potential incorporation of LRRT3 relative to today's costs as well as during the timeframe
covered by this proposal.

       3.4.5.2 Low Drag Brakes

       Low drag brakes reduce the sliding friction of disc brake pads on rotors when the
brakes are not engaged because the brake pads are pulled away from the rotating disc either
by mechanical or electric methods

       The 2012-2016 final rule and TAR estimated the effectiveness of low drag brakes to
be as much as 1 percent.  NHTSA and EPA have slightly revised the effectiveness down to
0.8% based on the 2011 Ricardo study and updated lumped-parameter model.

        In the 2012-2016 rule, the agencies  estimated the DMC at $57  (2007$). This DMC
becomes $58 (2009$) for this analysis. The agencies consider low drag brake technology to
be off the learning curve (i.e., the DMC does not change year-over-year) and have applied a

                                            3-188

-------
                                     Technologies Considered in the Agencies' Analysis
low complexity ICM of 1.24 through 2018 then 1.19 thereafter. The resultant costs are shown
in Table 3-97.

                        Table 3-97 Costs for Low Drag Brakes (2009$)
Cost type
DMC
1C
TC
2017
$58
$14
$73
2018
$58
$14
$73
2019
$58
$11
$70
2020
$58
$11
$70
2021
$58
$11
$70
2022
$58
$11
$70
2023
$58
$11
$70
2024
$58
$11
$70
2025
$58
$11
$70
            DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
       3.4.5.3 Front or Secondary Axle Disconnect for Four-Wheel Drive Systems

       Energy is required to continually drive the front, or secondary, axle in a four-wheel
drive system even when the system is not required during most operating conditions.  This
energy loss directly results in increased fuel consumption and CO2 emissions. Many part-time
four-wheel drive systems use some type of front axle disconnect to provide shift-on-the-fly
capabilities. The front axle disconnect is normally part of the front differential assembly. As
part of a shift-on-the-fly four-wheel drive system, the front axle disconnect serves two basic
purposes.  First, in two-wheel drive mode, it disengages the front axle from the front driveline
so the front wheels do not turn the front driveline at road speed, saving wear and tear.
Second, when shifting from two- to four-wheel drive "on the fly" (while moving), the front
axle disconnect couples the front axle to the front differential side gear only when the transfer
case's synchronizing mechanism has spun the front driveshaft up to the same speed as the rear
driveshaft. Four-wheel drive systems that have  a front axle disconnect typically do not have
either manual- or automatic-locking hubs. To isolate the front wheels from the rest of the
front driveline, front axle disconnects use a sliding sleeve to connect or disconnect an axle
shaft from the front differential side gear.  NHTSA and EPA are not aware of any
manufacturer offering this technology in the U.S. today on unibody frame vehicles; however,
it is possible this technology could be introduced by manufacturers within the MYs 2017-
2025 time period.

       The 2012-2016 final rule estimated an effectiveness improvement of 1.0 to 1.5 percent
for axle disconnect. Based on the 2011 Ricardo report, NHTSA and EPA refined this range to
1.2 to 1.4 percent.

       In the 2012-2016 rule, the agencies estimated the DMC at $78 (2007$) which was
considered applicable to the 2015MY. This DMC becomes $81 (2009$) for this analysis.
The agencies consider secondary axle disconnect technology to be on the flat portion of the
learning curve and have applied a low complexity ICM of 1.24 through 2018 then 1.19
thereafter.  The resultant costs are shown in Table 3-98.
                                            3-189

-------
                                      Technologies Considered in the Agencies' Analysis
                     Table 3-98 Costs for Secondary Axle Disconnect (2009$)
Cost type
DMC
1C
TC
2017
$77
$19
$96
2018
$75
$19
$94
2019
$74
$15
$89
2020
$72
$15
$88
2021
$71
$15
$86
2022
$69
$15
$85
2023
$68
$15
$83
2024
$66
$15
$82
2025
$65
$15
$81
            DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
       3.4.5.4 Aerodynamic Drag Reduction

       Many factors affect a vehicle's aerodynamic drag and the resulting power required to
move it through the air.  While these factors change with air density and the square and cube
of vehicle speed, respectively, the overall drag effect is determined by the product of its
frontal area and drag coefficient. Reductions in these quantities can therefore reduce fuel
consumption and COi emissions.  Although frontal areas tend to be relatively similar within a
vehicle class (mostly due to market-competitive size requirements), significant variations in
drag coefficient can be observed.  Significant changes to a vehicle's aerodynamic
performance may need to be implemented during a redesign (e.g., changes in vehicle shape).
However, shorter-term aerodynamic reductions, with a somewhat lower effectiveness, may be
achieved through the use of revised exterior components (typically at a model refresh in mid-
cycle) and add-on devices that are currently being applied. The latter list would include
revised front and rear fascias, modified front air dams and rear valances, addition of rear deck
lips and underbody panels, and lower aerodynamic drag exterior mirrors.

       The 2012-2016 final rule estimated that a fleet average of 10 to 20 percent total
aerodynamic drag reduction is attainable which equates to incremental reductions in fuel
consumption and COi emissions of 2 to 3 percent for both cars and trucks.  These numbers
are generally supported by the Ricardo  study and public technical literature and therefore NHTSA and
EPA are retaining these estimates, as confirmed by joint review, for the purposes of this proposal.

       For this proposal, the agencies are considering two levels of aero improvements. The
first level is that discussed in the 2012-2016 final rule and the 2010 TAR and includes such
body features as air dams, tire spats, and perhaps one underbody panel. In the 2012-2016
rule, the agencies estimated the DMC of aero-level 1 at $39 (2007$). This DMC becomes
$40 (2009$) for this analysis, applicable in the 2015MY. The agencies consider aero-level 1
technology to be on the flat portion of the learning curve and have applied a low complexity
ICM of 1.24 through 2018 then  1.19 thereafter.

       The second level of aero—level 2 which includes such body features as active grille
shutterspp, rear visors, larger under body panels or low-profile roof racks —was discussed in
the 2010 TAR where the agencies estimated the DMC at $120 (2008$) incremental to the
pp For details on how active aerodynamics are considered for off-cycle credits, see TSD Chapter 5.2.2.

                                            3-190

-------
                                     Technologies Considered in the Agencies' Analysis
baseline vehicle. The agencies inadvertently used that cost as inclusive of aero-level 1
technologies when it should have been incremental to aero-1 technologies.  As a result, the
agencies now consider the TAR cost to more appropriately be incremental to aero-level 1 with
a DMC for this analysis of $121 (2009$).  The agencies consider this cost to be applicable in
the 2015MY. Further, the agencies consider aero-level 2 technology to be on the flat portion
of the learning curve. The agencies have applied a medium complexity ICM of 1.39 through
2024 then 1.29 thereafter. The timing of the aero-level 2 ICMs is different than that for the
level 1 technology because the level 2 technology is newer and not yet being implemented in
the fleet.  The resultant costs are shown in Table 3-99.

           Table 3-99 Costs for Aerodynamic Drag Improvements - Levels 1 & 2 (2009$)
Cost
type
DMC
DMC
1C
1C
TC
TC
TC
Aero
Technology
Level 1
Level 2
Level 1
Level 2
Level 1
Level 2
Level 2
Incremental
to
Baseline
Aero-level 1
Baseline
Aero-level 1
Baseline
Aero-level 1
Baseline
2017
$38
$115
$10
$47
$48
$162
$210
2018
$38
$113
$10
$46
$47
$159
$207
2019
$37
$110
$8
$46
$45
$157
$201
2020
$36
$108
$8
$46
$44
$155
$198
2021
$35
$106
$8
$46
$43
$152
$195
2022
$35
$104
$8
$46
$42
$150
$192
2023
$34
$102
$8
$46
$42
$148
$190
2024
$33
$100
$8
$46
$41
$146
$187
2025
$33
$98
$8
$34
$40
$132
$173
DMC=Direct manufacturing cost; IC=Indirect cost; TC=Total cost
       3.4.5.5 Mass Reduction

       Over the past 20 years, there has been a generally increasing trend in the weight of the
light duty vehicle fleet as shown in Figure 3-26 from EPA's Fuel Economy Trends Report.79
There have been a number of factors contributing to this weight increase including
manufacturers choosing to build and consumers choosing to purchase larger vehicles
including heavier trucks, SUVs, and CUVs. Also contributing to this weight increase has been
an increase in vehicle content including; safety features (air bags, antilock brakes, energy
absorbent and intrusion resistant vehicle structures, etc.), noise reduction (additional damping
material), added comfort (air conditioning), luxury features (infotainment systems, power
locks and windows), etc.

       This increased weight in the fleet has been partially enabled by the increased
efficiency of vehicles, especially in engines and transmissions. The impressive improvements
in efficiency during this period have allowed for greater weight carrying and volume capacity
(and towing), safety, consumer features and vehicle refinement, as well as greater acceleration
performance.
                                            3-191

-------
                                     Technologies Considered in the Agencies' Analysis
           4600
           4200

           3800
            3400
            3000
              1975
                      1980
                               1985
                                       1990     1995

                                        Model year
                                                       2000
                                                                2005
                  Figure 3-26 Light Duty Fleet Weight characteristics 1975-2010

       Since 1987, on average, the overall fleet has become heavier and faster while fuel
economy has not shown marked or consistent increases. A calculation by University of
California Davis80 shows the combined impact of the fleet getting heavier while having
approximately stable fuel economy from 1987 to 2009 in ton-mpg terms. The improvement in
the fleet's technical efficiency is illustrated below in Figure 3-27. During the same period,
there are many improvements in vehicle performance, such as faster vehicle acceleration
shown in Figure 3-27 and reduced fatality in the fleet as shown in Figure 3-28.
                                            3-192

-------
                                           Technologies Considered in the Agencies' Analysis
          Vehicle weight
          and acceleration
          trends
          Fuel economy and
          weight-adjusted
          efficiency trends
                                                          3- Vehicle test weight
                                                          • 0-60 mph acceleration time
                                                    1990     1995
                                                      Model year
                                    Fuel economy (mpg)

                                    "Mass-adjusted fuel economy (ton-mpg)
                                                   increased vehicle efficiency, stable mpg
                                     1980    1985
                                                   1990     1995     2000     2005     2010
                                                     Model year
Figure 3-27 U.S. Light duty Fleet trends for weight, acceleration, fuel economy and  weight-adjusted fuel
                                 economy for model years 1975-2009
                    Motor vehicle crash deaths per billion miles traveled
                    1950-2009
                        60
                        40
                        20
                          1950 55   60  65   70   75  80   85  90   95  2000  05
                        Figure 3-28 U.S. Vehicle Fatality for the past 60 years1
                                                                           ,81
                                                   3-193

-------
                                     Technologies Considered in the Agencies' Analysis
       Reducing a vehicle's mass, or "down-weighting" the vehicle, decreases fuel
consumption by reducing the energy demand needed to overcome forces resisting motion.
Mass reduction can be also achieved by vehicle "downsizing" where a vehicle is physically
reduced in size by reducing exterior dimensions, such as  shifting from a midsize vehicle to a
compact vehicle. Both vehicle down-weighting and vehicle downsizing can yield lower GHG
emission and reduce fuel consumption. But vehicle downsizing is dependent on the consumer
choices which are influenced by many factors, such as the consumer's utility needs, fuel
prices, economic conditions, etc. qq In this NPRM analysis, the agencies are not analyzing
downsizing since we are assuming that the attribute based standards will not exert any
regulatory pressure for manufacturers to change the size of vehicles in order to come into
compliance with the proposed standards (as described in  Section II.F of the Preamble and
Chapter 2 of the joint TSD). Instead we are assuming that manufacturers will favor down-
weighting of a vehicle through material substitution, design optimization and adopting other
advance manufacturing technologies while not compromising a vehicle's attributes and
functionalities, such as occupant or cargo space, vehicle safety, comfort, acceleration
performance, etc. While keeping everything else constant, the lighter a vehicle is, the less fuel
is needed to drive the vehicle over a driving cycle. Researchers and industry have used a rule
of thumb, based on testing and simulation,  that 10 percent reduction in vehicle mass can be
expected to generate  a 6 to 7 percent increase in fuel economy if the vehicle powertrain and
other components are also downsized accordingly.82 A recent 2010 Ricardo study, funded by
EPA, updated this range to 5 to 8 percent increase in fuel economy.

       Mass reduction has an important relationship with vehicle powertrain selection and
sizing. Vehicle powertrain selection depends on an OEM's product strategy, and may include
a variety of options such as: naturally aspirated, boosted and downsized gasoline, diesel, or
vehicle electrification (P/H/EV). Regardless of the strategy selected, vehicle mass reduction
for non-powertrain systems is an important enabler to further reduce vehicle fuel consumption
and reduce the size of the powertrain system. Often times the term "glider" is used to include
all of the vehicle parts except for the powertrain of the vehicle. Figure 3-29 illustrates a
typical vehicle system mass breakdown83. Normally the non-powertrain systems account for
75 percent of vehicle weight and this is what the agencies are focusing on for this discussion.
  Vehicle mass reduction is very different that vehicle "down-sizing". Vehicle downsizing can confuse or confound the
analysis of mass-reduction technology trends; however these are distinctly different factors.

                                            3-194

-------
                                     Technologies Considered in the Agencies' Analysis
Approximate vehicle
mass breakdown"
Misc.;
(teurc, ™*
fenders;
8* Body;
23-23% A
10-15%
Powertrain;
Suspension/chassis;
22-27%
System
Body-in-white
Powertrain
Chassis
Interior
Closures
Miscellaneous
Major components in system
Passenger compartment frame, cross and side beams, roof
structure, front-end structure, underbody floor structure, panels
Engine, transmission, exhaust system, fuel tank
Chassis, suspension, tires, wheels, steering, brakes
Seats, instrument panel, insulation, trim, airbags
Front and rear doors, hood, lift gate
Electrical, lighting, thermal, windows, glaring
           "Based on Stodobky etal, 199 5a; Bjelkengren, 2008; Lotus Engineering, 2010; the actual system definitions and system
            component inclusion can vary, and percentage weight breakdown can vary substantially by vehicle

                        Figure 3-29 Vehicle system mass approximation
       Mass reduction can potentially be applied to any of a vehicle's subsystems, including
the engine, exhaust system, transmission, chassis, suspension, brakes, body, closure panels,
glazing, seats and other interior components, engine cooling systems, and HVAC systems.
Manufacturers generally tend to undertake larger amounts of mass reduction systematically
and more broadly across all vehicle systems when redesigning a vehicle. For example, if a
manufacturer applies a smaller, lighter engine with lower torque-output to a vehicle, this can
allow the use of a smaller, lighter-weight transmission and drive line components, because
those components need not be as heavy and robust to  support equivalent performance in the
redesigned vehicle with a smaller engine. Likewise, the combined mass reductions of the
engine, drivetrain, and body in turn reduce stresses on the suspension components, steering
components, wheels, tires, and brakes, which can allow further reductions in the mass of these
subsystems. Reducing the unsprung masses  such as the brakes, control arms, wheels, and
tires further reduce stresses in the suspension mounting points which will allow for further
optimization and potential mass reduction. When redesigning vehicles, OEMs normally set
weight targets by benchmarking other vehicles in the  same segment and projecting weight
trends into the future, and then identifying  targets for  all components and subsystems that
support achieving the target.  The agencies believe this holistic approach, taking into
consideration of all secondary mass savings, is the most effective way for OEMs to achieve
large amount of mass reduction.  During a  vehicle redesign where mass reduction is a
strategic vehicle program goal, OEMs can  consider modular systems design, secondary mass
effects, multi-material concepts, and new manufacturing processes to help optimize vehicles
for much greater potential mass reduction.  Figure 3-30 illustrates  an example of this approach
and how significant mass reduction opportunities can  be achieved when a complete vehicle
redesign is undertaken.
                                            3-195

-------
                                          Technologies Considered in the Agencies' Analysis
          Mau-eeducbon
* Redesign coovtnHoocl midHUH vehicle fee BBS cphmtation, whh two redesign •rchilecttvet
* Low Development % chicle technology wilh tihiuiJry-leading luanufacliHWft technique* thai Mete deemed
  feasible for 2014 (for model year 2017 production) for avvembljf »t existing facilities
• Hijtls [)f.clt]|JiikvTit i,clutlt (cx.hmjbgy. «JABafldilictf ito lo titaiV cfib&xti jdoiii^ tokds*i4ffife]y pracctiii,
  ttul u-«m: (kemcs! feaM^k fnr ^017 (for mock) vsar 20i20 prodniclian)
* Exlemive maf of nulerul 4 ubslitutiou Viilh hi^L-itrengtli ^[ceL advanced hijjL-iticagtli steel, aluommui,
  tttafccmiuni, pla.dic« and ccmpoulei Omtughoul veiuclei
* { unvci t am c u&e of cnicTgmg dftign ami paris uMegr«u«n concept* ie nunuimc I cciautil mk
• I"s4ji|i tyticryittic luul-vdiick Miln4a»1ial HUM reductidn ufipurTLUutici. found at mumin/ctl piece cotb
* Tie Low Dci'el
-------
                                     Technologies Considered in the Agencies' Analysis
redesign affects the ability to fully optimize a new vehicle to achieve all of the theoretically
possible secondary mass reduction. While there is agreement in the literature that primary
mass reduction can enable secondary mass reduction, the agencies recognize that care must be
taken when reviewing reports on mass reduction methods and practices to ascertain if
compounding effects have been considered and how.

       Mass reduction can occur through a variety of techniques available to manufacturers.
As summarized by NAS in its 2011 report, there are two key strategies for reducing vehicle
mass, changing the design to use less material or substituting light-weighting materials for
heavier materials while maintaining performance (safety and stiffness).88 The first approach is
to use less material comparing to the baseline component by optimizing the design and
structure of the component, system or vehicle structure. For an example, a "body on frame"
vehicle can be redesigned with  a lighter "unibody" construction by eliminating the number of
components and reducing the weight of the overall body structure, resulting in significant
mass reduction and related cost reduction. The unibody design dominates  the passenger car
segment and has an increasing penetration into what used to be body-on-frame vehicles, such
as SUVs. This technique was used in the 2011 Ford Explorer redesign in addition to extensive
use of high strength steels89. Figure 3-31 depicts body-on-frame and unibody designs for two
sport utility vehicles...
                            Body-QR-frame
Unibody
         Figure 3-31 Illustration of Body-on-Frame (BoF) and Unibody vehicle construction
       Manufactures can also continue to utilize Computer Aided Engineering (CAE) tools to
further reduce inefficiencies in vehicle design. For example, the Future Steel Vehicle (FSV)
project90 sponsored by the WorldAutoSteel, used three levels of optimization, topology
optimization, low fidelity 3G (Geometry Grade and Gauge) optimization and sub-system
optimization, to achieve 30 percent mass reduction in vehicle body structure with a unibody
design. Designs similar to some used in the FSV project have been applied in production
vehicles, such as the B-pillar of new Ford Focus.91 An example of this process in shown in the
Future Steel Vehicle project shown in Figure 3-32.
                                            3-197

-------
                                Technologies Considered in the Agencies' Analysis
 i.-t    T4: Body Structure Sub-System Optimisation
The filial desert attained fram the LF3G 6pt>mis^ti6rt was u&ed as the- basis ftw the &ub-sy$te«i
as well as the source of the boundary conditions. Load path mapping was conducted on the model to
identify the mast dominant structural sub-systems in the body structure. Load path mapping considers the
dominant loads in the structural sub-systems for each of the load cases as shown in Figure 2-7.
                                                                             . LIIl

                                                                             iff*
                                                                             • fin;
                  Figure 3-7: T4 Load Path Mapping — Major Load Path Components


Based  on toad  path  mapping, seven  structural sub-systems (Figure 2-S) were selected for further
optimisation using the spectrum of FSV's potential manufacturing technologies.
                 Front
                              JL
                              M^snorcun

                           Figure 2-8. Structural Sub-System* Selected
FutureStee [Vehicle
                                                                     WorldAutoStccI
   Figure 3-32 Example of vehicle body load path mapping for mass optimization
                                        3-198

-------
                                     Technologies Considered in the Agencies' Analysis
       Vehicle manufacturers have long used these continually-improving CAE tools to
optimize vehicle designs. But because any design must maintain component and system
functionality, there are practical limitations to the amount of additional design improvement
and mass reduction that can be achieved through optimization.  Additionally, ultimate
optimization of vehicle design for mass reduction may be limited by OEMs' typical use of a
common platform for multiple vehicle models.  While optimization may concentrate of the
vehicle that has the largest production volume for a platform, designs must also support the
most demanding functional requirements of all of the vehicles that share that platform. In
addition, the engineering resources and capital for tooling and equipment that would be
needed to optimize every vehicle component at each redesign affects the ability to fully
optimize a new vehicle to achieve all of the theoretically possible secondary mass reduction.
Therefore, it is inherent that some level of mass inefficiency will exist on many or all of the
vehicles that share a platform. The agencies seek comment and information on the degree  to
which shared vehicle components and architectures affect the feasible amount of mass
reduction and the cost for mass reduction relative to what could be achieved if mass reduction
was optimized for a single vehicle design.

       Using less material can also be achieved through improving the manufacturing
process, such as by using improved joining technologies and parts consolidation. This
method is often used in combination with applying new materials. For example, more precise
manufacturing techniques, such as laser welding, may reduce the flange size necessary for
welding and thus marginally decrease the mass of an assembly.  Also, when complex
assemblies are constructed from fewer pieces, the mass of the assembly tends to be lower.
Additionally, while synergies in mass reduction certainly exist, and while certain technologies
(e.g., parts consolidation and molding of advanced composites) can enable one another, others
(e.g., laser welding and magnesium casting) may be incompatible.

       The second key strategy to reduce mass of an assembly  or component involves  the
substitution of lower density and/or higher strength materials. Table 3-100 shows material
usage typical to high-volume vehicles. Material substitution includes replacing materials, such
as mild steel, with advanced and regular higher-strength steels,  aluminum, magnesium and/or
composite materials. The substitution of advanced high strength steel (AHSS)  can reduce the
mass of a steel part because AHSS has higher strength than mild steel  and therefore less
material is needed in strength-critical components despite the fact that its density is not
significantly different from mild steel.  Some manufacturers are considering even more
advanced materials for many applications, but the advanced microstructure  and limited
industry experience with some materials may make these longer-term  solutions. For example,
advanced composite materials (such as carbon fiber-reinforced  plastic), depending on the
specific fiber, matrix, reinforcement architecture, and processing method, can be subject to
dozens of competing damage and failure mechanisms that may  complicate a manufacturer's
ability  to ensure equivalent levels of durability and crashworthiness. As the industry gains
experience with these materials, these concerns will inevitably diminish, but may remain
relevant during the timeframe of this rulemaking. Material substitution also tends to be quite
manufacturer and situation specific in practice; some materials  work better than others  for
some vehicle components and a manufacturer may invest more  heavily in adjusting its

                                            3-199

-------
                                      Technologies Considered in the Agencies' Analysis
manufacturing to a particular type of advanced material and complicate its ability to consider
others. The agencies recognize that like any type of mass reduction, material substitution has
to be conducted not only with consideration to maintaining equivalent component strength,
but also to maintaining all the other attributes of that component, system or vehicle, such as
crashworthiness, durability, and noise, vibration and harshness (NVH).

       Automobiles also utilize a wide range of plastic types, including polypropylenes,
polyesters, and vinyl esters. These materials are utilized in hatches, roofs, interior panels,
instrument panels, and hundreds of other parts. Although primarily replacing nonstructural
vehicle components, plastics have continued to make in-roads in bumper systems and in
composite beam applications and a number of studies have found potential to supplant
structural beams and frame component. Additionally included in this general category are the
more costly composites, like glass fiber and carbon fiber reinforced polymers. These
materials, to date, are used primarily in limited applications in low-production-volume
vehicles.
Table 3-100 Distribution of Material in Typical Contemporary Vehicles (e.g. Toyota Camry and Chevrolet
                                         Malibu)

                                                                                Approximate Content in Cars
 Material                       Comments                                            Today, by Weight (percent'f
 Iron and mild steel
 High-strength steel
 Aluminum
 Plastic
 Other (magnesium, titanium, rubber, elc.i
Under 480 Mpa                                         55
> 480 Mpa (in body structure)                                15
No aluminum closure panels; aluminum engine block and head and wheels    10
Miscellaneous parts, niostfv interior trim, tight lenses, facm, instrument panel  10
Miscellaneous pans                                      10
       If vehicle mass is reduced sufficiently, a manufacturer may use a smaller, lighter, and
potentially more efficient powertrain while maintaining vehicle acceleration performance. If
a powertrain is downsized, approximately half of the mass reduction may be attributed to the
reduced torque requirement which results from the lower vehicle mass.  The lower torque
requirement enables a reduction in engine displacement, changes to transmission torque
converter and  gear ratios, and changes to final drive gear ratio. The reduced powertrain torque
enables the downsizing and/or mass reduction of powertrain components and accompanying
reduced rotating mass (e.g., for transmission, driveshafts/halfshafts, wheels, and tires) with
similar powertrain durability.

       All manufacturers are using some or all of these methods to some extent to reduce
mass in the vehicles they are producing today, and the agencies expect that the industry will
continue to learn and improve the application of these techniques for more vehicles during the
rulemaking timeframe. We consider mass reduction in net percentage terms in  our analysis
not only because effectively determining specific appropriate mass reduction methods for
each vehicle in the baseline fleet is a large task beyond the scope of this rulemaking, but also
because we recognize that even as manufacturers reduce mass to make vehicles more
                                             3-200

-------
                                     Technologies Considered in the Agencies' Analysis
efficient, they may also be adding mass in the form of increased vehicle content, some of
which is feature and safety content in response to market forces and other governmental
regulations.  For these reasons, when the agencies discuss the amount of mass reduction that
we are assuming is feasible for purposes of our analysis, we are implicitly balancing both the
considerable opportunities that we believe exist for mass reduction in the future, and the
reality that vehicle manufacturing is complex and that mass reduction methods must be
applied thoughtfully and judiciously as safety and content demands on vehicles continue to
increase over time. Despite our considerable discussion of the topic, the agencies' application
of mass reduction in our analysis is fairly simplified. As applied in our models, the percentage
reduction for a given vehicle that is  assumed for a given year is an abstraction for the use of
all the mass reduction  methods described above (and in the literature search portion of the
above cost discussion). This represents the significant complexity of mass reduction
technologies for improving fuel economy and reducing CCh emissions.
How much mass reduction do the agencies believe is feasible in the rulemaking timeframe?

       Feasibility, if narrowly defined as the ability to reduce mass without any other
constraints, is nearly unbounded. However, the feasible amount of mass reduction is affected
by other considerations. Cost effectiveness is one of those constraints and is discussed in the
cost section, above.  In the analysis for the MYs 2012-2016 rulemaking, NHTSA assumed
different amounts of mass reduction (defined as net reduction of a percentage of total vehicle
mass) were feasible  for different vehicle subclasses in different model years. In addition, it
was assumed that more mass was taken out at a redesign and/or later in the rulemaking
timeframe than at a refresh and/or earlier in the rulemaking timeframe. More specifically,
NHTSA assumed that mass could be reduced  1.5 percent at any refresh or redesign, and that
mass could be reduced an additional incremental 3.5-8.5 percent (3.5 for smaller vehicles, 8.5
for the largest vehicles) at redesigns after MY 2014 to provide leadtime for these larger mass
reduction amounts.  The amount (percentage) of mass reduction that the NHTSA used in the
analysis generally aligned with information that the agencies received, during the MY 2012-
2016 rulemaking, from manufacturers related to their plans to reduce mass of larger vehicles
more than smaller vehicles in the 2012-2016 timeframe. Based on the NHTSA's analysis, it
was estimated that mass reduction in response to the MY 2012-2016 program would achieve a
safety-neutral result.

       In the analysis for the current rulemaking for MYs 2017-2025, the agencies reviewed
a number of public reports and accompanying data, as well as confidential information from
manufacturers and believe that mass reduction of up to 20 percent can be achieved in a cost
effective manner using technologies currently in production.  More detail on studies reviewed
by the agencies and  additional studies currently in progress by the agencies is located in Table
3-103 and Preamble section II.G.

       From a general planning perspective, nearly all automakers have made some public
statement regarding  vehicle mass reduction being a core part of the overall technology
strategy that they  will utilize to achieve future fuel  economy and CO2 emission standards.

                                           3-201

-------
                                     Technologies Considered in the Agencies' Analysis
Estimates from Ducker Worldwide indicate that the automobile industry will see an annual
                                         09
increase in AHSS of about 10% through 2020  . Ford has stated that it intends to reduce the
weight of its vehicles by 250-750 Ib per model from 2011 to 202093. For context, the midpoint
of that range of reductions would correspond to a 12% reduction from the current Ford new
light duty vehicle sales fleet. Similarly, Nissan has a target of a 15% mass reduction per
vehicle by 201594. This reduction would represent over a 500-lb reduction from their 2008
light duty vehicle average. Mazda's has released a statement about achieving a 220-lb
reduction per vehicle by  201695. This is equivalent to about a 6% reduction for the company's
current fleet. Toyota stated that it could end up reducing the mass of the Corolla and mid-size
models by 30% and 10%, respectively, in the 2015 timeframe. The low end of those targets,
10%, is equivalent to 350 Ib per Toyota vehicle in 2008. Land Rover remains committed to a
goal of reducing curb weights of its S.U.V.'s by as much as 500 kilograms over the next 10
years96. Several reports are summarized in the University of California study as shown in
Table 3-10197.98
   Table 3-101 Automaker industry statements regarding plans for vehicle mass-reduction technology
Afiffliatioa
General
Motors
Ford
Nissan
BMW
Volkswagen
Fiat
Volkswagen
BMW
BMW
Quote
"We use a lot of aluminum today-about 300 pounds per vehicle-and are likely to use more lightweight
materials in the future."
"The use of advanced materials such as magnesium, aluminum and ultra high-strength boron steel offers
automakers structural strength at a reduced weigh! to help improve fuel economy and meet safety and
durability requirements"
"We are working to reduce the thickness of steel sheet by enhancing the strength, expanding the use of
aluminum and other lightweight materials, and reducing vehicle weight by rationalizing vehicle body
structure"
"Lightweight construction is a core aspect for sustainable mobility improving both fuel consianptiGii and CO;
emissions, two key elements of our EfficientDynamics strategy. . . .we will be able to produce carbon fiber
enhanced components in large volumes at competitive costs for the first time. This is particularly relevant
for electric -powered vehicles."
"Material design and manufacturing technologies remain key technologies in vehicle development. Only
integrated approaches that work on these three key technologies will be successful in the future. In addition to
the development of metals and light metals, the research on fibre-reinforced plastics will play a major role."
"A reduction of fuel consumption attains big importance because of the possible economical savings. In order
to achieve that, different ways are followed: alternative engine concepts (for example electric engines instead
of combustion ones) or weight reduction of the vehicle structure. Using lightweight materials and different
joining techniques helps to reach this aim"
"Lightweight design is a key measure for reducing vehicle fuel consumption, along with power train
efficiency, aerodynamics and electrical power management"
"A dynamic vehicle with a low fuel consumption finally demands a stiff body with a low weight. To achieve
die initially mentioned targets, it is therefore necessary to design a body which offers good stiffness values and
a high level of passive safety at a low weight.
"Light weight design can be achieved by engineering light weight, manufacturing light weight and material
light weight design"
Source
Keith. 2010
Keith. 2010
Keith. 2010
BMW and
SGL, 2010
Goede et al.
2ffi»
Nunez. 2009
Krmke. 2009
Prestorf.
2009
Prestorf.
2009
The agencies also believe the practical limits of mass reduction will be different for each
vehicle model as each model starts with a different mix of conventional and advanced
materials, components, and features intended to meet the function and price of a particular
market segment. A vehicle that already has a significant fraction of advanced high strength
steel (AHSS) or any other advanced material in its structure, for example, will not have the
                                           3-202

-------
                                     Technologies Considered in the Agencies' Analysis
opportunity to realize the same percentage of mass reduction as a vehicle of more traditional
construction.  Given the myriad methods of achieving mass reduction, and the difficulty in
obtaining data, accounting for the current level of mass reduction technology for every model
in production in a baseline model year would be an impractical task.  However, the agencies
believe that reducing vehicle weight to reduce fuel consumption has a continuum of solutions
and the technologies employed will have levels of effectiveness and feasibility that will vary
by manufacturers and by vehicle.  In estimating the amount of mass reduction for this
analysis, the agencies also consider fleet safety effects for mass reduction.  See the Preamble
II.G for a detailed discussion of the safety considerations in establishing CAFE and GHG
standards. In the CAFE and OMEGA analyses, the agencies considered several levels  of
mass reduction to all of the models in each subclass as discussed below.

       Based on the many aspects of mass reduction (i.e. feasibility, cost and safety), for the
proposal the agencies believe that mass reduction of up to 20 percent is feasible on light
trucks, CUVs and minivans, but that less mass reduction should be implemented on other
vehicle types to avoid increases in societal fatalities. While the agencies continue to examine
mass reduction further, we remain alert to safety considerations and seek to ensure that any
CAFE and CO2 standards can be achieved in a safety-neutral manner.

       In the CAFE model, NHTSA applied the amounts of mass reduction shown in Table
3-102, which enabled us to achieve overall fleet fatality estimates of close to zero.

                Table 3-102 MASS REDUCTION AMOUNT APPLIED IN CAFE MODEL
Absolute
%
MR1*
MR2
MRS
MR4
MRS
Subcompact
and
Subcompact
Perf. PC
Compact and
Compact
Perf. PC
Midsize PC
and Midsize
Perf. PC
Large PC and
Large Perf.
PC
Mini van LT
Small,
Midsize and
Large LT
0.0% 2.0% 1.5% 1.5% 1.5% 1.5%
0.0% 0.0% 5.0% 7.5% 7.5% 7.5%
0.0% 0.0% 0.0% 10.0% 10.0% 10.0%
0.0% 0.0% 0.0% 0.0% 15.0% 15.0%
0.0% 0.0% 0.0% 0.0% 20.0% 20.0%
           Notes:
           *MR1-MR5: different levels of mass reduction used in CAFE model

       The amounts of mass reduction shown in Table 3-102, however, are for conventional
vehicles. The amount of mass reduction applied in the OMEGA model follows the safety
neutral analysis approach described in Section II.G of the Preamble. The results are described
on a variety of tables within EPA's draft RIA (Chapter 3.8.2). The agencies assume that
vehicles with hybrid and electric powertrain are heavier than conventional vehicles because of
the mass of battery systems. In comparing anecdotal data for HEVs, EPA and NHTSA
assumes a slight weight increase of 4-5% for HEVs compared to baseline non-hybridized
vehicles. The added weight of the Li-ion pack, motor and other electric hardware were offset
partially by the reduced size of the base engine as stated in TSD section 3.4.3.8. This
assumption, which we believe accurately, reflects real-world HEV, PHEV and EV
                                           3-203

-------
                                      Technologies Considered in the Agencies' Analysis
construction, as an example, for a subcompact PHEV with 20 mile range operating on
electricity, because of the additional weight of the electrification system, the agencies assume
that to achieve no change in total vehicle mass, it would be necessary to reduce the mass of
the glider (the vehicle without the powertrain), by 6 percent.  The mass reduction for P/H/EVs
can be found section 3.4.3.9 in TSD, section 3.4.3.8 and in EPA's RIA Chapter 1 and
NHTSA's RIA Chapter V section E.3.h.4.
How much do the agencies estimate mass reduction will cost in the rulemaking timeframe?
       Automakers are currently utilizing various mass reduction techniques across the light-
duty vehicle fleet, and will continue to use and in some cases expand these approaches for the
2017 to 2025 time frame.  These approaches may include optimized design,  geometry, part
consolidations, and materials substitution. Unlike the other technologies described in this
chapter, mass reduction is potentially more complex in that we cannot define it as a single
piece of equipment or hardware change to implement the technological improvement. Mass
reduction, depending upon the level of reduction targeted, has the potential to impact nearly
every system on the vehicle.  Because of this complexity, there are unique challenges to
estimating the cost for mass reduction and for demonstrating the feasibility of reducing
vehicle mass by a given amount. This section describes the cost estimates used for the
agencies' analysis.

       In the analysis for the MYs 2012-2016 rulemaking, the agencies assumed a constant
cost for mass reduction of $1.32 for each pound reduced up to a mass reduction level of 10
percent (or $1.48/lb using an ICM factor of 1.1 for a low-complexity technology).  The
$1.32/lb estimate was based on averaging three studies: the 2002 NAS Report, a 2008 study
by Sierra Research, and a 2007 study by MIT researchers.n

       Since the MYs  2012-2016 final rule, the agencies have given further consideration to
the cost of mass reduction, and now believe that a cost that varies with the level of mass
reduction provides a better estimate.  The agencies believe that as the vehicle fleet progresses
from lower to higher levels of mass reduction and becomes increasingly optimized for mass
and other attributes, the cost for mass reduction will progressively increase.  The higher levels
of mass reduction may, for example, require applying more advanced materials and
technologies than lower levels of mass reduction, which means that the cost  of achieving
those higher levels may increase accordingly. The unit cost of mass reduction versus the
n Specifically, the 2002 NAS Report estimated that vehicle weight could be reduced by 5 percent (without
engine downsizing) at a cost of $210-$350, which translates into $1.50/lb assuming a 3,800 Ib base vehicle and
using the midpoint cost; Sierra Research estimated that a 10 percent reduction (with compounding) could be
accomplished for $1.01/lb, and MIT researchers estimated that a 14 percent reduction (with no compounding)
could be accomplished for $1.36/lb. References for these studies are available in endnotes to Chapter 3 of the
TSD for the MYs 2012-2016 final rule.

                                            3-204

-------
                                      Technologies Considered in the Agencies' Analysis
amount of mass reduction might be linear, parabolic, or some other higher order relationship.
In the 2017-2025 Notice of Intent, 75 FR 62739 (Oct. 13, 2010),  CARB, EPA and NHTSA
derived a second  order curve based on a study with two vehicle redesigns conducted by Lotus
Engineering completed in 2010, such that zero mass reduction had zero cost, and the dollars
per pound increased with greater levels of mass reduction. Since the publication of the TAR,
the agencies have identified a number of additional studies in the literature relating to the
costs of vehicle mass reduction, which are discussed below.  The studies show that for low or
high mass reduction, the costs can range from small cost savings to significant cost increases.
The economic costs associated with mass reduction are difficult to determine conclusively
due to the broad range of methods s employed to achieve mass reduction. The costs on a
specific vehicle or component depend on many factors, such as the design, materials selected,
raw material price,  appropriate manufacturing processes, production volume, component
functionality, required engineering and development, etc. Cost data thus varies widely in the
literature. Of the various studies reviewed by the agencies, not all are equal in their original
intent, rigor, transparency, or applicability to this regulatory purpose. The individual studies
range from complete vehicle redesign to advanced optimization of individual components,
and were conducted by researchers with a wide range of experience and background. Some
of the studies were  literature reviews, while others developed new designs for lighter
components or complete lighter vehicles, while yet others built physical components or
systems, and conducted testing on those components and systems. Some of the studies
focused only on a certain sub-system (which is a building block for the  overall vehicle
design), while some of them took a systematical approach and re-designed the whole vehicle
to achieve the maximum mass reduction and cost reduction. The latter studies typically
identified a specific baseline vehicle, and then utilized different engineering approaches and
investigated a variety of mass-reduction concepts that could be applied  to that vehicle. Some
of the differences between studies emanate from the characteristics of the baseline vehicle and
its adaptability to the new technology or method, and the cost assumptions relating to the
original components and the redesigned components. Assumptions regarding the degree and
cost of any associated mass decompounding can also confound comparisons.ss  Despite this
variation in the literature, in actual practice, we believe manufacturers will choose a target
mass reduction for  a whole vehicle and for each sub-system, and work to find the lowest total
cost method to achieve those targets.  Such a process would consider numerous primary and
ss The concept of secondary weight savings or mass compounding (also called mass decompounding) derives
from the qualitative understanding that as vehicle weight decreases, other vehicle systems can also decrease in
mass while maintaining the original vehicle level of performance and function. For instance, following a primary
weight reduction in the vehicle (e.g. Body in White), the designs of some of the other dependant vehicle
subsystems (tires, suspensions, brakes, powertrain, body structure) may be redesigned and reduced in mass to
account for the overall lighter vehicle. The lighter vehicle is also associated with lighter loads, less friction and
drag, and may require less power to be accelerated, and the powertrain may therefore be scaled down in size with
a potential for reduced mass, even while maintaining equivalent acceleration performance and functionality. The
compounded or secondary mass savings from these additional systems may then drive further mass reductions in
the original primary weight reduction (e.g. Body in White). Mass compounding factors found in literature are
rough estimates of the secondary mass reduction amount.

                                             3-205

-------
                                       Technologies Considered in the Agencies' Analysis
secondary cost factors (including engineering, facilities, equipment, tooling, and retraining
costs) as well as technological and manufacturing risks."

       Regardless of the confidence in specific estimates, the agencies must select a curve
that will be applied to the whole fleet that will define the average cost per pound of mass
reduction as a function of total percentage of mass reduction. There are many significant
challenges that make it difficult for the agencies to establish an estimated cost curve based on
the literature, such as the differences in the baselines used in the studies, whether the studies
considered platform sharing and powertrain sharing, and other considerations. The agencies
initially considered using the flat rate cost estimate that was used for the last rulemaking,
$1.32/lb, but as discussed above, there are appropriate reasons to consider a variable cost
curve.  The agencies then considered the cost estimates from the TAR, but have noted that
there is more data available at present that could potentially be useful in informing our
estimates. Nonetheless, coalescing these disparate datasets into a single curve has limitations
since the various studies are not directly comparable.

       With these challenges in mind, and because the agencies have not finished the
significant mass reduction studies targeted for the CAFE and GHG rulemaking (described
below), the agencies examined all the studies in Table 3-103 including information supplied
by manufacturers (during meetings held subsequent to the TAR) when deciding the mass
reduction cost estimate used for this NPRM.UU The agencies considered three major factors in
examining these studies. First, whether a study was rigorous in terms  of how it evaluates and
validates mass reduction from technological and design perspectives.  This includes
consideration of a study's comprehensiveness, the technical rigor of its methodology, the
validation methods employed, and the relevance of the technologies evaluated in the study
given our rulemaking time frame. Second, whether  a study was rigorous in terms of its
estimation of costs, including the completeness and  rigor of the methodology,  such as whether
the study includes data for all categories of direct  manufacturing costs, and whether the study
presents  detailed cost information for both the baseline and the light-weighted design. And
third, the degree of peer review, including if the study is peer-reviewed, and whether it has
effectively addressed any critical technical, methodological, and cost issues raised by the
peer-review, if this information is available.

       Some of the variation may be attributed to the complexity of mass reduction as it is
not one single discrete technology and can have direct as well as indirect effects on other
" We also note that the cost of mass reduction in the Volpe model is quantified on a per pound basis that is a
function of the percentage decrease in vehicle mass. We assume that OEMs would find the most cost-effective
approach to achieve such a mass reduction. Realistically, this would depend heavily on the baseline vehicle as
well as the size and adaptability of the initial design to the new technology. Thus, the Volpe model strives to be
realistic in the aggregate while recognizing that the figures proposed for any specific model may be debatable.
uu The agencies considered confidential cost information provided by OEMs that covered a range of components,
systems, designs and materials. Some of these cost estimates are higher than some of the literature studies, and
manufacturers provided varying levels of detail on the basis for the costs such as whether mass compounding is
included, or whether the costs include markup factors.

                                             3-206

-------
                                     Technologies Considered in the Agencies' Analysis
systems and components. The 2010 NAS study speaks to this point when it states on page 7-1
that "The term material substitution oversimplifies the complexity of introducing advanced
materials, because seldom does one part change without changing others around it." These
variations underscore that there is not a unique mass reduction solution as there are many
different methods with varying costs for taking mass out of vehicles, and every manufacturer,
even every vehicle, could have a different approach depending on the specific vehicle,
assembly plant and model year of implementation. The agencies recognize that there are
challenges to characterizing the mass reduction plans for the entire future fleet due to the
complexity and variety of methods available. So far the agencies have not found any study
that addresses how to generalize the mass reduction that is achievable on a single vehicle to
the whole fleet.

       Table 3-103 contains a summary of the data contained in the studies, and the OEM
CBI data, which the agencies reviewed. There is a degree of uncertainty associated with
comparing the costs from the range of studies in the literature  when trying to summarize them
in a single table, and we encourage interested stakeholders to carefully review the information
in the literature. For some of the cost estimates presented in the papers there are unknowns
such as: what year the costs are estimated for, whether mass decompounding (and potential
resultant cost savings) was taken into account, and whether mark-ups  or indirect costs were
included. The agencies tried to normalize the cost estimations from all these studies by
converting them to 2009 year dollar, applying mass compounding factor of 1.35 for mass
reduction amount more than 10 percent if it has not been applied in the study and factoring
out the RPE specified in the study to derive direct manufacture costs for comparison. There
are some papers that give cost for only component mass reduction, others that have more
general subsystem costs and others yet that estimate total vehicle mass reduction costs (which
often include and present data at the subsystem level). Other studies have multiple scenarios
for different materials, different vehicle structures and mass reduction strategies. Thus, a
single study which contains more than one vehicle can be broken down into a range of vehicle
types, or at the subsystem level, or even at the component level.  While Table 3-103 is
inclusive of all of the information reviewed by the agencies, for the reasons described above
the technical staff for the  two agencies applied various different approaches in evaluating the
information. The linear mass-cost relationship developed for this proposal and presented
below is the consensus assessment from the two agencies of the appropriate mass cost for this
proposal.
                                            3-207

-------
                                    Technologies Considered in the Agencies' Analysis
Table 3-103 Mass Reduction Studies Considered for Estimating Mass Reduction Cost for this NPRM
Studies
Cost Year
Cost Information from Studies
Mass Reduction [Ib]
Compounding Factor
Mass Reduction with
Compounding [Ib]
Baseline Vehicle Weight [Ib]
Mass Reductioning
w/Compounding [%]
J/V
4-1
ifi
O
u
u
Q_
ee.
Dollar Multiplierto 2009
2009 Direct Manufacturing
Cost [$]
Unit Cost of Mass Reduction
[$/lb]
Individual Cost Data Points
AISI, 1998(ULSAB)
AISI, 2000(ULSAC)
Austin et al, 2008 (Sierra Research) - ULS Unibody
Austin et al, 2008 (Sierra Research) - AL Unibody
Austin et al, 2008 (Sierra Research) - ULS BoF
Austin et al, 2008 (Sierra Research) - AL BoF
Bull et al, 2008 (Alum Assoc.) - ALBIW
Bull et al, 2008 (Alum Assoc.) - ALCIosure
Bull et al, 2008 (Alum Assoc.) - Whole Vehicle
Cheah et al, 2007 (MIT) - 20%
Das, 2008 (ORNL) - AL Body & Panel
Das, 2008 (ORNL)-FRPMC
Das, 2009 (ORNL) - CF Body & Panel, AL Chassis
Das, 2010 (ORNL) - CF Body & Panel, Mg Chassis
EEA, 2007- Midsize Car- Adv Steel
EEA, 2007- Midsize Car- Plast/Comp
EEA, 2007- Car-Avg. AI/Mg
EEA, 2007 -Midsize Car- Al
EEA, 2007 -Midsize Car- Mg
EEA, 2007 - Light Truck - Adv Steel
EEA, 2007 - Light Truck - Plast/Comp
EEA, 2007 - Light Truck - Al
EEA, 2007 - Light Truck - Mg
Gecketal, 2008 (Ford)
Lotus, 2010 - LD
Lotus, 2010 - HD
Montalbo et al, 2008 (GM/MIT) - Closure - HSS
Montalbo et al, 2008 (GM/MIT) - Closure - AL
Montalbo et al, 2008 (GM/MIT) - Closure - Mg/AL
Plotkin et al, 2009 (Argonne)
1998
2000
2008
2008
2008
2008
2008
2008
2008
2007
2008
2008
2009
2010
2007
2007
2007
2007
2007
2007
2007
2007
2007
2008
2010
2010
2008
2008
2008
2009
103
6
320
573
176
298
279
70
573
712
637
536
933
1173
236
254
657
586
712
422
456
873
1026
1310
660
1217
25
120
139
683
i
i
i
i
i
i
i
i
i
i
i
1.0
i
i
i
i
1.35
1.35
1.35
1
1
1.35
1.35
1
1
1
1
1
1
1
103
6
320
573
176
298
279
70
573
712
637
536
933
1173
236
254
887
791
961
422
456
1179
1385
1310
660
1217
25
120
139
683
2977
2977
3200
3200
4500
4500
3378
3378
3378
3560
3363
3363
3363
3363
3350
3350
4500
3350
3350
4750
4750
4750
4750
5250
3740
3740
4000
4000
4000
3250
3.5%
0.2%
10.0%
17.9%
3.9%
6.6%
8.3%
2.1%
17.0%
20.0%
19.0%
15.9%
27.7%
34.9%
7.0%
7.6%
14.6%
23.6%
28.7%
8.9%
9.6%
24.8%
29.2%
25.0%
17.6%
32.5%
0.6%
3.0%
3.5%
21.0%
-$32
$15
$209
$1,805
$171
$1,411
$455
$151
$122
$646
$180
-$280
$1,490
$373
$179
$239
$1,411
$1,388
$1,508
$291
$398
$1,830
$1,976
$500
-$121
$362
$10
$110
$110
$1,300
1.0
1.0
1.61
1.61
1.61
1.61
1.0
1.0
1.0
1.0
1.5
1.5
1.5
1.5
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.28
1.24
1.01
1.01
1.01
1.01
1.01
1.01
1.03
1.03
1.01
1.01
1.00
1.00
1.03
1.03
1.03
1.03
1.03
1.03
1.03
1.03
1.03
1.01
1.00
1.00
1.01
1.01
1.01
1.00
-$41
$18
$131
$1,134
$107
$887
$460
$153
$126
$667
$121
-$189
$993
$248
$185
$247
$1,458
$1,434
$1,558
$301
$411
$1,891
$2,042
$506
-$120
$360
$10
$111
$111
$1,300
-$0.40
$2.99
$0.41
$1.98
$0.61
$2.98
$1.65
$2.17
$0.22
$0.94
$0.19
-$0.35
$1.06
$0.21
$0.78
$0.97
$1.64
$1.81
$1.62
$0.71
$0.90
$1.60
$1.47
$0.39
-$0.18
$0.30
$0.41
$0.92
$0.80
$1.90
                                          3-208

-------
                                     Technologies Considered in the Agencies' Analysis
Table 3-103 (... Continue) Mass Reduction Studies Considered for Estimating Mass Reduction Cost for
                                      this NPRM
Studies
Cost Year
Cost Information from Studies
Mass Reduction [Ib]
Compounding Factor
Mass Reduction with
Compounding [Ib]
Baseline Vehicle Weight [Ib]
Mass Reductioning
w/Compounding [%]
W
1
LLJ
Q_
ee.
Dollar Multiplier to 2009
2009 Direct Manufacturing
Cost [$]
Unit Cost of Mass Reduction
[$/lb]
Cost Curves
NAS, 2010- Average
NAS, 2010
OEMl-Average
OEM1
OEM2-Average
OEM2
OEMS-Average
OEMS
OEM4
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2010
2011
2011
2011
































































































































10.0%
1.0%
2.0%
5.0%
10.0%
20.0%
9.5%
8.0%
9.0%
9.5%
10.0%
11.0%
3.1%
0.4%
0.9%
1.9%
2.3%
2.4%
3.1%
3.6%
4.0%
4.1%
4.5%
4.8%
5.0%
7.2%
4.0%
7.5%
10.0%
6.9%
8.1%
16.4%
































































































































$1.50
$ 1.41
$ 1.46
$ 1.65
$ 1.52
$ 1.88
$11.60
$ 6.00
$ 7.00
$ 8.00
$ 12.00
$ 25.00
$0.61
$
$ 0.10
$ 0.20
$ 0.33
$ 0.38
$ 0.60
$ 0.76
$ 0.85
$ 0.88
$ 0.98
$ 1.09
$ 1.17
$1.03
$ 0.57
$ 1.01
$ 1.51
$ 0.97
$ 1.02
$ 1.95
                                            3-209

-------
                                     Technologies Considered in the Agencies' Analysis
       EPA and NHTSA scrutinized the various available studies in the literature as well as
confidential information provided by several auto firms based on the kinds of factors
described above for purposes of estimating the cost of mass-reduction in the 2017-2025
timeframe. We determined that there was wide variation across the studies with respect to
costs estimates, applicability to the 2017-2025 time frame, and technical rigor. The mass cost
curve that was developed this proposal is defined by the following equation:

       Mass Reduction Direct Manufacturing Cost ($/lb) = 4.32 x Percentage of Mass
Reduction

For example, this results in an estimated $173 cost increase for a 10% mass reduction of a
4,0001b vehicle (or $0.43/lb), and a $390 cost increase for 15% reduction on the same vehicle
(or $0.65/lb).

       Because of the wide variation in data used to select this estimated cost curve, the
agencies have also conducted cost sensitivity studies in their respective RIAs using values of
+/-40%. The wide variability in the applicability and rigor of the studies also provides
justification for continued research in this field, such as the agency studies discussed below.
The assessment of the current studies highlights the importance of these agency studies, as
they are expected to be amongst the most comprehensive ever conducted in the literature, and
to be more informative than other studies for estimating the cost of mass reduction for
purposes of rulemaking.

       The agencies consider this DMC to be applicable to the 2017MY and consider mass
reduction technology to be on the flat portion of the learning curve in the 2017-2025MY
timeframe. To estimate indirect costs for applied mass reduction of up to 15%, the agencies
have applied a low complexity ICM of 1.24 through 2018 and 1.19 thereafter. To estimate
indirect costs for applied mass reduction of 15% to 25%, the agencies have applied a medium
complexity ICM of  1.39 through 2024 and 1.29 thereafter. To estimate indirect costs for
applied mass reduction greater than 25%, the agencies believe it is appropriate to apply a
highl complexity ICM of 1.56 through 2024 and 1.35  thereafter.

       The agencies seek detailed comment regarding options for realistically and
appropriately assessing the degree of feasible mass reduction for vehicles in  the rulemaking
timeframe and the total costs to achieve that mass reduction.  For example, the agencies seek
comments on what practical limiting factors need to be considered when considering
maximum feasible amount of mass reduction; the degree to which these limiting factors will
impact the amount of feasible mass reduction (in terms of the percent of mass reduction); the
best method(s) to assess  an appropriate and feasible fleet-wide amount mass reduction amount
(because each study mainly focuses on a single vehicle); etc.  If commenters wish to submit
additional studies for the agencies' consideration, it would assist the agencies if commenters
could address how the studies also contribute to the agencies' understanding of the issues
enumerated above.  The agencies also note that we expect to refine our estimate of both the
                                            3-210

-------
                                    Technologies Considered in the Agencies' Analysis
amount and the cost of mass reduction between the NPRM and the final rule based on the
ongoing work described below.

How effective do the agencies estimate that mass reduction will be?

       In the analysis for the MYs 2012-2016 final rule, NHTSA and EPA estimated that a
10 percent mass reduction with engine downsizing would result in a 6.5 percent reduction in
fuel consumption while maintaining equivalent vehicle performance (i.e., 0-60 mph time,
towing capacity, etc.), consistent with estimates in the 2002 NAS report. For small amounts
of mass reduction, such as the 1.5 percent used at vehicle refresh in NHTSA's modeling, no
engine downsizing was used, so a 10 percent mass reduction without engine downsizing was
assumed to result in a 3.5 percent reduction in fuel consumption.  In this NPRM, both
agencies have chosen to use the effectiveness value for mass reduction from EPA's lumped
parameter model to maintain consistency. EPA's lumped parameter model-estimated mass
reduction effectiveness is based on a simulation model developed by Ricardo, Inc. under
contract to EPA.  The 2011 Ricardo simulation results show an effectiveness of 5.1
percent for every 10 percent reduction in mass. NHTSA has assumed that for mass
reduction less than 10 percent the effectiveness is 3.5 percent. For mass reduction greater
than 10 percent, NHTSA estimates the effectiveness is 5.1 percent which avoids double
counting benefits - because the effectiveness of engine downsizing  is included in the
effectiveness of the engine decision tree when applying engine downsizing, it should
appropriately be removed from the mass reduction effectiveness value in the mass reduction
decision tree. EPA applies an effectiveness of 5.1 percent for every 10 percent mass
reduction, and this scales linearly from 0 percent mass reduction, up to the maximum
applied mass reduction for any given vehicle, which in this proposal is never larger than
20 percent.

What additional studies are the agencies conducting to inform our estimates of mass
reduction amounts,  cost, and effectiveness?

       In the MYs 2012-2016 final rule, the agencies stated that there are several areas
concerning vehicle mass  reduction and vehicle safety on which the agencies will focus their
research efforts and undertake further study.  Some studies focus on the potential safety
effects of mass reduction through fleetwide analyses, and thus help  to inform the agencies
with regard to how much mass reduction might appropriately be deemed feasible in the
rulemaking timeframe, while others focus on the cost and feasibility of reducing mass in
specific vehicles.  The results of all of these studies are currently expected to be available for
the final rule, and  should contribute significantly to informing the agencies' estimates of the
costs and feasible  amounts of mass reduction to be included in that analysis.  The following is
an update for the status of those studies.

       The agencies and independent researchers have several vehicle level projects to
determine the maximum potential for mass reduction in the  MY 2017-2021 timeframe by
using advanced materials and improved designs while continuing to meeting safety
                                           3-211

-------
                                    Technologies Considered in the Agencies' Analysis
regulations and maintaining functionality of vehicles, and one study that will investigate the
effects of resultant designs on fleet safety:

   •  NHTSA has awarded a contract to Electricore, with EDAG and George Washington
      University (GWU) as subcontractors, to study the maximum feasible amount mass
      reduction for a mid-size car - specifically, a Honda Accord. The study tears down a
      MY 2011 Honda Accord, studies each component and sub-system, and then redesigns
      each component and sub-system trying to maximize the amount of mass reduction with
      technologies that are considered feasible for 200,000 units per year production volume
      during the time frame of this rulemaking. Electricore and its sub-contractors  are
      consulting industry leaders and experts for each component and sub-system when
      deciding which technologies are feasible. Electricore and its sub-contractors  are also
      building detailed CAD/CAE/powertrain models to validate vehicle safety, stiffness,
      NVH, durability, drivability and powertrain performance. For OEM-supplied parts, a
      detailed cost model is being built based on a Technical Cost Modeling (TCM)
      approach developed by the Massachusetts Institute of Technology (MIT) Materials
      Systems Laboratory's research" for estimating the manufacturing costs of OEM parts.
      The cost will be broken down into each of the operations involved in the
      manufacturing, such as for a sheet metal part production by starting from blanking the
      steel coil, until the final operation to fabricate the component. Total costs are then
      categorized into fixed cost, such as tooling, equipment, and facilities; and variable
      costs such as labor, material, energy, and maintenance. These costs will be assessed
      through an interactive process between the product designer, manufacturing engineers
      and cost analysts. For OEM-purchased parts, the cost will be estimated by consultation
      with experienced cost analysts and Tier 1 system suppliers. This study will help to
      inform the agencies about the feasible amount of mass reduction and the cost
      associated with it. NHTSA intends to have this study completed and peer reviewed
      before July 2012, in time for it to play an integral role in informing the final rule.

   •  EPA has awarded a contract to FEV, with EDAG and Monroe & Associates,  Inc. as
      subcontractors, to study the maximum feasible amount of mass reduction for  a mid-size
      CUV (cross over vehicle) specifically, a Toyota Venza.  The study tears down a MY
      2010 vehicle, studies each component and sub-system, and then redesigns each
      component and sub-system trying to maximize the amount of mass reduction with
      technologies that are considered feasible for high volume production for a 2017 MY
      vehicle. FEV in coordination with EDAG is building detailed CAD/CAE/powertrain
      models to validate vehicle safety, stiffness, NVH, durability, drivability and powertrain
      performance to assess the safety of this new design. This study builds upon the low
      development (20% mass reduction) design in the 2010 Lotus Engineering study "An
      Assessment of Mass Reduction Opportunities for a 2017-2020 Model Year Vehicle
      Program". This study will undergo a peer review. EPA intends to have this study
      completed and peer reviewed before July 2012, in time for it to play an integral role in
      informing the final rule.
                                           3-212

-------
                                     Technologies Considered in the Agencies' Analysis
   •  California Air Resources Board (CARB) has awarded a contract to Lotus Engineering,
      to study the maximum feasible amount mass reduction for a mid-size CUV (cross over
      vehicle) specifically, a Toyota Venza. The study will concentrate on the Body-in-
      White and closures in the high development design (40% mass reduction) in the Lotus
      Engineering study cited above. The study will provide an updated design with crash
      simulation, detailed costing and manufacturing feasibility of these two systems for a
      MY2020 high volume production vehicle. This study will undergo a peer review.
      CARB intends to have this study completed and peer reviewed before July 2012, in
      time for it to play an integral role in informing the final rule.

   •  NHTSA has contracted with GWU to build a fleet simulation model to study the
      impact and relationship of light-weighted vehicle design and injuries and fatalities.
      This study will also include an evaluation of potential countermeasures to reduce any
      safety concerns associated with lightweight vehicles. NHTSA will include three light-
      weighted vehicle designs in this study: the one from Electricore/EDAG/GWU
      mentioned above, one from Lotus Engineering funded by California Air Resource
      Board for the second phase  of the study, evaluating mass reduction levels around 35
      percent of total vehicle mass, and one funded by EPA and the  International Council on
      Clean Transportation (ICCT). This study will help to inform the agencies about the
      possible safety implications for light-weighted vehicle designs and the appropriate
      counter-measures,vv if applicable, for these designs, as well as the feasible amounts of
      mass reduction. All of these analyses are expected to be finished and peer-reviewed
      before July 2012,  in time to inform the final rule.

       Safety considerations in establishing CAFE/GHG standards along with discussion of
NHTSA's February 25, 2011, mass-size-safety workshop at DOT headquarters, can be found
in Section II.G of the preamble for this proposal. NHTSA plans  to host additional workshops
when the studies have reached a sufficient level of completion, to share the results with the
public and seek public comments.
3.5 How did the agencies consider real-world limits when defining the rate at which
       technologies can be deployed?

3.5.1   Refresh and redesign schedules

       During MYs 2017-2025 manufacturers are expected to go through the normal
automotive business cycle of redesigning and upgrading their light-duty vehicle products, and
in some cases introducing entirely new vehicles not in the market today.  The MY 2017-2025
standards timeframe allows manufacturers the time needed to incorporate GHG reduction and
vv Countermeasures could potentially involve improved front end structure, knee bags, seat ramps, buckle
pretensioners, and others.

                                           3-213

-------
                                     Technologies Considered in the Agencies' Analysis
fuel-saving technologies into their normal business cycle while considering the requirements
of the MY 2012-2016 standards.. This is important because it has the potential to avoid the
much higher costs that could occur if manufacturers need to add or change technology at
times other than their scheduled vehicle redesigns. This time period also provides
manufacturers the opportunity to plan for compliance using a multi-year time frame, again
consistent with normal business practice.  Over these 9 model years, and the 5 prior model
years that make up the 2012-2016 standards, there will be an opportunity for manufacturers to
evaluate, presumably, every one of their vehicle platforms and models and add technology in
a cost effective way to control GHG emissions and improve fuel economy. This includes all
the technologies considered here and the redesign of the air conditioner systems in ways that
will further reduce GHG emissions and improve fuel economy.

       Because of the complexities of the automobile manufacturing process, manufacturers
are generally only able to add new technologies to vehicles on a specific schedule; just
because a technology exists in the marketplace or is made available, does not mean that it is
immediately available for application on all of a manufacturer's vehicles.  In the automobile
industry there are two terms that describe when technology changes to vehicles occur:
redesign and refresh (i.e., freshening).  Vehicle redesign usually refers to significant changes
to a vehicle's appearance, shape, dimensions, and powertrain. Redesign is traditionally
associated with the introduction of "new"  vehicles into the market, often characterized as the
"next generation" of a vehicle, or a new platform. Across the industry, redesign of models
generally takes place about every 5 years.  However, while 5 years is a typical design period,
there are many instances where redesign cycles can be longer or shorter. For example, it has
generally been the case that pickup trucks  and full size vans have longer redesign cycles,
while high-volume cars have shorter redesign cycles in order to remain competitive in the
market. There are many other factors that can also affect redesign such as availability of
capital and engineering resources and the extent of platform and component sharing between
models, or even manufacturers.

       Vehicle refresh usually refers to less extensive vehicle modifications, such as minor
changes to a vehicle's appearance, a moderate upgrade to a powertrain system, or small
changes to the vehicle's feature or safety equipment content. Refresh is traditionally
associated with mid-cycle cosmetic changes to a vehicle, within its current generation, to
make it appear "fresh." Vehicle refresh generally occurs no earlier than two years after a
vehicle redesign or at least two years before a scheduled redesign. For the majority of
technologies discussed today, manufacturers will only be able to apply them at a refresh or
redesign, because their application would be significant enough to involve some level of
engineering, testing, and calibration work.

       Most vehicles would likely undergo two redesigns during this period. Even with the
potential of multiple of refresh and redesign cycles, it is still likely that some of the more
advanced and costly technologies (such as cooled boosted EGR engines, or advanced
(P)HEVs) may not be able to be fully implemented within the timeframe of this rule. These
limitations are captured in "phase-in caps," discussed in the next section, and "maximum
technology penetration rates" within the modeling analysis.

                                            3-214

-------
                                     Technologies Considered in the Agencies' Analysis
       The broad technology classes evaluated for purposes of this analysis are defined below
and a brief discussion of the limiting factors considered are presented.

       •  Conventional Spark Ignition (SI) - This technology category includes all
          technologies, such as gasoline direct injection engines, cylinder deactivation, six
          and eight speed automatic and dual clutch transmissions, and start-stop micro-
          hybrid technology, that are not contained in other categories.  Many of these
          technologies were anticipated as being available in the MYs 2012-2016 time frame
          in the recent NHTSA and EPA final rule, and it is expected manufacturers could
          expand production to all models by model year 2025. Conventional SI also
          includes turbocharged and downsized engines and turbocharged and downsized
          engines that include cooled EGR with additional levels of boost and a larger
          degree of engine downsizing than seen in the current light-duty gasoline fleet.
          These latter technologies are similar to the technologies that many OEMs indicated
          were underdevelopment and which they anticipate will be introduced into the
          market in the 2017-2025 time frame.

       •  Hybrid - While the agencies recognize there are many types of full-hybrids either
          in production or under development, for the purposes of this analysis we have
          specifically modeled the P2 type hybrid, as explained in section 3.4.3.6.3. While
          the agencies expect the proliferation of these vehicles to increase in this timeframe,
          the maximum technology penetration rate and phase-in caps are set at less than
          100% in MY 2025 due to industry-wide engineering and capacity constraints, for
          converting the entire new vehicle fleet to strong hybrids (like P2 and others) in this
          time frame. As described these technologies (along with PHEVs and EVs) require
          a significant cost and complexity, and thus are  not expected to be able to be fully
          phased into the 2017-2025 fleet like other more conventional (but advanced)
          engines.

       •  Plug-in Hybrid (PHEV) - This technology includes PHEVs with a range  of 20 and
          40 miles. The maximum technology penetration rates and phase-in caps are set at
          less than 100% in MY 2025 due to the same general potential constraints  as listed
          for the HEVs, but are lower for PHEVs due to  the current status of the
          development of these advanced vehicles and the higher cost relative to HEVs. In
          addition, some consumers may have limited or no access to charging
          infrastructure, and for those consumers, the PHEV offers little benefit over an
          HEV at a higher cost.  Further, we project (based on what we know today) that
          PHEV technology is not available to some vehicle types, such as large pickups.
          While it is technically possible to electrify such vehicles, there are tradeoffs in
          terms of cost, electric range, and utility that may reduce the appeal of the vehicle
          to a narrower market.  However, the agencies are interested in promoting
          innovation to overcome these potential obstacles and are thus incentivizing more
          HEV and PHEV pickup trucks with credit flexibilities as described in the preamble
          for this proposed rule.
                                           3-215

-------
                                     Technologies Considered in the Agencies' Analysis
       •  Electric Vehicle (EV) - This technology includes vehicles with actual on-road
          ranges of 75, 100, and 150 miles. The actual on-road range was calculated using a
          projected 30% gap between two-cycle and on-road range. These vehicles are
          powered solely by electricity and are not powered by any liquid fuels.  The
          maximum technology penetration rates and phase-in caps are set at less than 100%
          in MY 2025 due to the same general potential constraints as discussed for PHEVs.
          EVs have additional constraints due to limited infrastructure and range as well.
          Further, as with PHEVs, we assume that EV technology is not available to some
          vehicle types, such as large pickups.  While it is possible to electrify such vehicles,
          there are tradeoffs in terms of cost, range, and utility that would reduce the appeal
          of the vehicle to a narrower market. These trade-offs are expected to reduce the
          market for other vehicle types as well, and for this analysis we have considered
          this in the development of the maximum technology penetration rates.

       •  Mass Reduction - This technology includes material substitution, smart design,
          and mass reduction compounding. NHTSA and EPA have conducted a thorough
          assessment of the levels of mass reduction that could be achieved which is both
          technologically feasible and which can be implemented in a safe manner for this
          joint federal NPRM (as described earlier in this Chapter).

3.5.2   Vehicle phase-in caps

       GHG-reducing and fuel-saving technologies for vehicle applications vary widely in
function, cost, effectiveness and availability. Some of these attributes, like cost and
availability vary from year to year. New technologies often take several years to become
available across the entire market. The agencies use phase-in caps to manage the maximum
rate that the CAFE and OMEGA models can apply new technologies.

       Phase-in caps are intended to function as a proxy for a number of real-world
limitations in deploying new technologies in the auto industry. These limitations can include
but are not limited to, engineering resources  at the OEM or supplier level, restrictions on
intellectual property that limit deployment, and/or limitations in material or component supply
as a market for a new technology develops. Without phase-in caps, the models may apply
technologies at rates that are not representative of what the industry is actually capable of
producing, which would suggest that more stringent standards might be feasible than actually
would be.

       EPA applies the caps on an OEM vehicle platform basis for most technologies.  For a
given technology with a cap of x%, this means that x% of a vehicle platform can receive that
technology.  On a fleet average basis, since all vehicle platforms can receive x% of this
technology, x% of a manufacturer's fleet can also receive that technology. EVs and PHEVs
are an exception to this rule.  Unlike other technologies, which are applicable to all classes of
vehicles, EPA only allows non-towing vehicle types to be electrified in the OMEGA model.
As a result, the PHEV and EV cap was applied so that the average manufacturer could
produce to the cap levels. Manufacturers that make fewer non-towing vehicles have a lower

                                           3-216

-------
                                     Technologies Considered in the Agencies' Analysis
potential maximum production limit of EVs and PHEV, while those that make more non-
towing vehicles have a higher potential maximum limit on EV and PHEV production.

       NHTSA applies phase-in caps in addition to refresh/redesign cycles used in the CAFE
model, which constrain the rate of technology application at the vehicle level so as to ensure a
period of stability following any modeled technology applications, Unlike vehicle-level cycle
settings, phase-in caps, defined on a percent per year basis, constrain technology application
at the OEM level. As discussed above phase-in caps are intended to reflect a manufacturer's
overall resource capacity available for implementing new technologies (such as engineering
and development personnel and financial resources) thereby ensuring that resource capacity is
accounted for in the modeling process. At a high level, phase-in caps and refresh/redesign
cycles work in conjunction with one another to avoid the CAFE modeling process out-pacing
an OEM's limited pool of available resources during the rulemaking time frame, especially in
years where many models may be scheduled for refresh or redesign.  This helps  to ensure
technological feasibility and economic practicability in determining the stringency of the
standards.

       Phase-in caps do not define market penetration rates and they do not define the rate at
which a particular technology will be applied, rather they simply present an upper limit, or
ceiling at which the agencies' computer models can apply new technologies to vehicles to
raise their fuel economy and reduce their CO2 emissions.  Ultimately, phase-in caps are
determined by the agencies using engineering judgment. However, there are several sources
of information on technology penetration that the agencies consider in assigning phase-in caps
to various technologies:
              Confidential OEM submissions indicate the rate at which an individual
              manufacturer can deploy a particular technology. Manufacturer information is
              especially helpful if multiple manufacturers indicate similar technology
              penetration rates. The agencies consider these CBI submissions along with
              other sources of information.
              Historical data from EPA's annual Fuel Economy Trends Report100 is used to
              inform the agencies about typical historical rates of adoption of technologies.
              However, historical data does not necessarily indicate the rates of future
              technology penetration.  Increased competition is driving faster vehicle
              redesigns and  faster adoption of new technologies.  On the other hand, some
              new technologies such as EVs  are significantly more complex than most other
              historical technologies, which must also be considered in defining a phase-in
              rate.
              Trade press articles, company publications, press releases, and other reports
              often discuss new technologies, how quickly they will be deployed and
              manufacturing strategies that enable faster penetration rates. These articles
              provide a useful glimpse into how manufacturers are changing in order to
              become more  competitive.

                                            3-217

-------
                                     Technologies Considered in the Agencies' Analysis
       3.5.2.1 Trends Report and Industry Data

       For over 30 years, EPA's Fuel Economy Trends report has tracked the fuel economy
of light duty vehicles and the technology used by automakers to improve fuel economy. A
particularly interesting aspect of the Trends data is how technology is adopted by the industry
and how this changes over time. Trends data shows that industry-wide, it has typically taken
up to 15-20 years for a technology to penetrate the entire fleet. Some technologies such as
port fuel injection and variable valve timing start slowly and then rapidly progress. Others,
like torque converter lockup and front wheel drive penetrate rapidly after their first
appearance on the market. Figure 3-33, below shows these trends.
CO
.c
CO
                     100%
                     80%
                     60%
                 o   40%
                 T3
                 Q
                 °-   20%
                      0%
                                                 Port FL
                                Front Wheel
                              Lockup
                                                                i-Valve
        ^Vx-

Variable Valve Timing
                                                         CVT
                                  5      10      15     20     25
                                       Years After First Use
                                                                  30
                                                                         35
                                                                    101
                 Figure 3-33 Technology Penetration After First Significant Use

       There are several cases where technologies have penetrated the fleet rapidly,
sometimes beginning with significant market penetration, sometimes beginning with
relatively small market penetration.  For example, six speed automatic transmissions were in
7% of the industry-wide fleet in 2006 and by 2010, they were in 36% of the fleet, for an
increase of 29% in 4 yearsww. Port fuel injection went from about 12% of the fleet in 1984 to
88% in 1994. Front wheel drive, a technology that requires a complete change in vehicle
architecture, increased from 9% in 1979 to 60% by 1988102.

       Recent academic literature has also used deployment rate data from the EPA Fuel
Economy Trends Report, Wards Factory Installed data, and other sources to report to describe
historical deployment rates of a variety of technologies. (DeCicco, 2010 and Zoepf, 2011).
DeCicco, for example, cites conversion to fuel injection and front wheel drive in passenger
cars as having seen maximum growth in adoption of 17% and 11% per year respectively.103
ww EPA staff calculated the penetration rate of 6-speed automatic transmissions from 2010 Trends data.
Aggregated source data can be seen on page 54 of the 2010 Fuel Economy Trends Report.
                                            3-218

-------
                                     Technologies Considered in the Agencies' Analysis
Zoepf examines a broader array of automotive technologies and notes a span of maximum
growth rates in passenger cars from 4% to nearly 24% per year with variance based on feature
type.104

      While these examples show that the industry is capable of adopting certain new
technologies rapidly industry-wide, considering the rate of introduction of technology by
individual OEMs  shows that the pace of technology introduction can in some cases be even
faster. Table 3-104 below shows how individual manufacturers can apply technologies rapidly
to a large fraction of their fleet. Although not typical for most manufacturers and
technologies, the data below shows that manufacturers have chosen to deploy some
technologies very rapidly.

                 Table 3-104: Historical Phase-In Rates of Selected Technologies
Manufacturer
General Motors
Ford
Honda
Chrysler
Toyota-cars only
Nissan-cars only
Toyota-cars only
Ford
Nissan
Volkswagen
Hyundai
General Motors
General Motors
Technology
Lockup Transmission
Fuel Injection
Fuel Injection
Fuel Injection
Multi-Valve
Multi-Valve
Variable Valve Timing
Multi-Valve
Continuously Variable
Transmission
Gasoline Direct Injection
Variable Valve Timing
Variable Valve Timing
Gasoline Direct Injection
Technology Market Share Increase
1980-1982: 83% in 3 years
1983-1987: 91% in 5 years
1986-1990: 91% in 5 years
1988: 37% in 1 year
1987-1989: 85% in 3 years
1989-1990: 71% in 2 years
2000-2003: 87% in 4 years
2004-2005: 36% in 2 years
2007: 45% in 1 year
2008: 52% in 1 year
2009: 48% in 1 year
2006-2010: 75% in 5 years
2010: 27% in 1 year
       Often, a rapid application of technology is helped by having similar vehicle
architecture, or by sharing major components such as engines or transmissions across multiple
products. As discussed below, platform sharing combined with improvements in platform
and manufacturing flexibility is expected to further enable faster implementation of new
technologies.

       3.5.2.2 The rate of technology adoption is increasing

       The agencies recognize that new technologies may not achieve rapid deployment
immediately and that small-scale production is a part of the technology learning process. To
this end the phase-in caps distinguish between technologies that have been successfully
applied in existing vehicles and those that under development but are anticipated on
production vehicles in the near future.

       The rate of technology adoption appears to be increasing as manufacturers increase
model turnover and decrease the numbers of unique vehicle platforms. This facilitates a
steady stream of new products, increased sales and optimized vehicle redesigns allowing and
                                            3-219

-------
                                     Technologies Considered in the Agencies' Analysis
fuel consumption-reducing technologies to be applied to as many vehicles as possible. In
today's globally competitive market, and certainly for the U.S., market share and
competitiveness is strongly influenced by a manufacturer's ability to turn over their product
line-up. Merrill Lynch's Car Wars Report105  shows that replacement rate is speeding up and
showroom age is dropping as manufacturers are striving to be more competitive in the market.
Increased model turn-over creates more opportunity for manufacturers to deploy new
technologies faster than in the past.

       Zoepf, cited above, reports that the developmental time, from first production
application to maximum growth rate, has been declining exponentially as manufacturers bring
innovations to market progressively faster. Ellison et al. (1995)106 indicate that U.S. and
European automakers reduced overall product development time by more than a year in the
1990s. Ellison et al. point to the increased role that suppliers have had in product
development process during the same time, potentially commoditizing innovations more
quickly.

       Vehicle platforms are the basic underpinnings of vehicles and are often shared across
several unique products.  By reducing the number of platforms, and making these platforms
flexible, manufacturers can better deploy resources to serve a wider market with more
products. Utilizing a modern, flexible platform architecture, a manufacturer can produce a
sedan, wagon, minivan, and a crossover, or SUV on a single platform and all of  these
products can be assembled in a single vehicle assembly plant.  Basic components can be
developed and purchased at high volumes, while enabling the manufacturer to exploit what
would otherwise be niche markets. This commonization of platforms does have the potential
to increase the mass for lighter vehicle models within the platform because the platform needs
to be designed for the more severe duty cycle of the SUV  and/or larger engine. Volkswagen
has recently launched a new platform  called MQB, which will be used world-wide by up to 60
unique models from VW, Audi, Seat,  and Skoda. This structure will replace 18 "engine
mounting architectures" with just two.

             It gives us the possibility to produce models from different segments and in
       varying sizes using the same basic front-end architecture," .... "We  can  go from a
       typical hatchback to a saloon,  cabriolet and SUV with only detailed changes to the
       size of the wheel carriers." ...  it will be used on  every model from the new Lupo all the
       way through to the next-generation Sharan.107

       One of the key enablers of this drive to reduce platforms and increase model turn-over
                                   i ns
is increased manufacturing flexibility.    For example, in 2004, Ford invested in flexible
manufacturing technology for their Cleveland No. 1 engine plant. Although the plant was
shut down for two years after this investment, Ford was able to retool and reopen the plant at
a low cost to produce their new 3.5L EcoBoost turbocharged, direct injection engine as well
as their 3.7L V6.109  In their December, 2008 business plan submitted to Congress,110 Ford
further stated,
                                           3-220

-------
                                    Technologies Considered in the Agencies' Analysis
              ...nearly all of our U.S. assembly plants will have flexible body shops by 2012
       to enable quick response to changing consumer demands and nearly half of our
       transmission and engine plants will be flexible, capable of manufacturing various
       combinations of transmission and engine families.

       Like VW, Ford is also striving to reduce their platforms and complexity. In Ford's
2008 business plan submitted to Congress, they stated that in addition to divesting themselves
from certain luxury brands like Jaguar, Land Rover, Volvo, and Aston Martin, they were
working to consolidate their vehicle platforms from 25 in 2005 to 9 by 2012. Having more
vehicles per platform frees up resources to deploy new technologies across a greater number
of vehicles more quickly and increases the rate at which new technologies can be introduced.
We believe GM's recent restructuring will also enable faster vehicle redesigns and more rapid
penetration rates in the 2010-plus time frame compared to the 1990s and 2000s. In the past
seven years, GM has eliminated five brands (Saturn, Hummer, Saab, Pontiac, and
Oldsmobile), significantly reducing the number of unique products and platforms the
company needed to devote engineering resources to. GM has set a goal to halve its number of
vehicle platforms by 2018 and boost manufacturing efficiency by 40%.m

       3.5.2.3 Phase-in Rates Used in the Analysis

       Table 3-105 below shows phase-in rates for the technologies used in the OMEGA
model. OMEGA calculations are based on five year intervals, so phase-in caps are derived
for model years 2016, 2021 and 2025.  Table 3-106  shows phase-in rates for the technologies
used in the CAFE model.  The CAFE model calculations are  annual, so phase in rates are
derived for every year of the program.  Where possible, phase-in rates for OMEGA and
CAFE were harmonized, but there are some differences mainly where technologies differ
between the agencies.

       Most technologies are available at a rate of either 85% or 100% beginning in 2016.
Some advanced technologies expected to enter the market in  the near future such as EGR
Boost follow a 3% annual cap increase from 2016 to 2021, then, approximately 10% from
2021 to 2025. Diesels follow an annual 3% increase in phase-in cap through 2025. Hybrids
follow a 3% annual increase from 2016 to 2012, then 5% from 2021 to 2015. PHEVs and EVs
follow a 1 % annual cap increase.

       Lower phase-in caps for Alternate Fueled Vehicles (AFVs) reflect additional
investment in infrastructure that is required to achieve high levels of conversion to a new fuel
type.  These limited phase-in caps also reflect as yet unknown consumer responses to HEVs,
PHEVs and BEVs.

                    Table 3-105 Phase-In Caps used in the OMEGA model
Technology
Low Friction Lubricants
Engine Friction Reduction - level 1
2016
100%
100%
2021
100%
100%
2025
100%
100%
                                           3-221

-------
Technologies Considered in the Agencies' Analysis
Early Torque Converter lockup
Aggressive Shift Logic - Level 1
Improved Accessories - Level 1
Low Rolling Resistance Tires - Level 1
Low Drag Brakes
VVT - Intake Cam Phasing
VVT - Coupled Cam Phasing
VVT - Dual Cam Phasing
Cylinder Deactivation
Variable Valve Lift - Discrete
Variable Valve Lift - Continuous
Conversion to DOHC
Stoichiometric Gasoline Direct Injection (GDI)
Turbocharging (18 bar BMEP) and Downsizing
Continuously Variable Transmission
6-speed Automatic Transmission
6-speed Dual Clutch Transmission - dry & wet clutch
Electric & Electric/Hydraulic Power Steering
12V Stop-Start
Secondary Axle Disconnect
Aero Drag Reduction - Level 1
Aggressive Shift logic - Level 2 (Shift Optimizer)
8-speed Automatic Transmission
8-speed Dual Clutch Transmission - dry & wet clutch
Improved Accessories - Level 2
Aero Drag Reduction - Level 2
Low Rolling Resistance Tires - Level 2
Engine Friction Reduction - level 2 (inc. low friction lubes - level 2)
High Effiency Gearbox
Turbocharging (24 bar BMEP) and Downsizing
Cooled EGR
P2 Hybrid Electric Vehicle (HEV)
Turbocharging (27 bar BMEP) and Downsizing
Conversion to Advanced Diesel
Full Electric Vehicle (EV)
Plug-in HEV
100%
100%
100%
100%
100%
85%
85%
85%
85%
85%
85%
85%
85%
85%
85%
85%
85%
85%
85%
85%
85%
0%
30%
30%
30%
30%
0%
0%
0%
15%
15%
15%
0%
15%
6%
5%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
80%
80%
80%
80%
75%
60%
60%
30%
30%
30%
15%
30%
11%
10%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
75%
75%
50%
50%
42%
15%
14%
      3-222

-------
                                                     Technologies Considered in the Agencies' Analysis
                               Table 3-106 Phase-In Caps used in the CAFE Model
                  Technology
                                                    MY
                                                    2009
           MY MY
           2010 2011
 MY
2012
 MY MY
2013 2014
 MY
2015
 MY MY
2016 2017
 MY
2018
 MY
2019
 MY MY
2020 2021
 MY MY
2022 2023
 MY MY
2024 2025
 Low Friction Lubricants - Level 1
 Engine Friction Reduction - Level 1
 Low Friction Lubricants and Engine Friction Reduction -Lev el 2
 Variable Valve Timing (WT) - Coupled Cam Phasing (CCP) on SOHC
 Discrete Variable Valve Lift (DWL) on SOHC
 Cylinder Deactivation on SOHC
 Variable Valve Timing (WT) - Intake CamPhasing (ICP)
 Variable Valve Timing (WT) - Dual Cam Phasing (DCP)
 Discrete Variable Valve Lift (DWL) on DOHC
 Continuously Variable Valve Lift (CWL)
 Cylinder Deactivation on DOHC
 Stoichiometric Gasoline Direct Injection (GDI)
 Cylinder Deactivation on OHV
 Variable Valve Actuation - CCP and DWL on OHV
 Stoichiometric Gasoline Direct Injection (GDI) on OHV
 Turbocharging and Downsizing - Level 1 (ISbarBMEP)
 Turbocharging and Downsizing - Level2 (24barBMEP)
 Cooled Exhaust Gas Recirculation (EGR) - Level 1 (24bar BMEP)
 Cooled Exhaust Gas Recirculation (EGR) - Level2 (27bar BMEP)
 Advanced Diesel
                                                                  75%  75%
                                                                  75%  75%
                                                                  45%  50%
                                                                  6%  6%
 6-Speed Manual/Improved Internals
 High Efficiency Gearbox(Manual)
 Improved Auto. Trans. Controls/Externals
 6-Speed Trans with Improved Internals (Auto)
 6-speedDCT
 8-Speed Trans (Auto orDCT)
 High Efficiency Gearbox(Auto orDCT)
 Shift Optimizer
                                   85%  95% 100%
                                   0%  12% 24%
                                   85%  95% 100%
                                   85%  95% 100%
                                   85%  95% 100%
                                   30%  40% 50%
                                   0%  12% 24%
                                                                                           100% 100% 100%
 Electric Power Steering
 Improved Accessories - Level 1
 Improved Accessories - Level2
 12V Micro-Hybrid (Stop-Start)
 Integrated Starter Generator
 Strong Hybrid - Leve
 Conversion fromSHEVl to SHEV2
 Strong Hybrid - Level2
 Plug-in Hybrid - 30 mi range
 Plug-in Hybrid
 Electric Vehicle (Early Adopter) - 75 mile range
 Electric Vehicle (Broad Market) - 150 mile range
 Fu el Cell Vehicle
 Mass Reduction - Level 1
 Mass Reduction - Level 2
 Mass Reduction - Level 3
 Mass Reduction - Level 4
 Mass Reduction - Level 5
 Low Rolling Resistance Tires - Level 1
 Low Rolling Resistance Tires -Level2
 Low Drag Brakes
 Secondary Axle Disconnect
 Aero Drag Reduction, Level 1
 Aero Drag Reduction, Level2
AER01
AER02
                                                                            50% 60% 70% 80% 90%
3.6  How are the technologies applied in the agencies' respective models?
          Although both NHTSA and EPA are basing their fuel economy and emission
modeling on the same baseline vehicle fleet and cost and effectiveness estimates for control
technologies, differences in the CAFE and OMEGA models result in   this common
information being processed  in different ways prior to its  use in the respective models.  With
respect to the vehicle fleet, the CAFE Model evaluates the addition of technology to
                                                              3-223

-------
                                     Technologies Considered in the Agencies' Analysis
individual vehicle configurations or models, while the OMEGA model does so for vehicle
platforms broken down further by engine size. The Volpe (or CAFE) Model evaluates
technologies individually.  This, coupled with the modeling of individual vehicle models,
means that only the presence or absence of any particular technology needs to be indicated, as
described above. OMEGA applies technology in combinations or packages.  This, plus the
grouping of individual vehicle models, requires that the total effectiveness of the technology
already applied in the baseline fleet must be calculated and must be reflected as a percentage
of the various technology packages available to be added to those vehicles.

      With respect to the cost and effectiveness of technologies, as mentioned above, the
CAFE Model applies technologies individually.  It does this following certain specified
pathways for several categories of technologies (e.g., engine, transmission, accessories, etc.).
The Volpe Model applies technology incrementally, so the effectiveness of each subsequent
technology needs to be determined relative to the previous one.  The same is true for cost.  In
addition, because of interaction in the effectiveness of certain technologies, herein referred to
as the synergy/dis-synergy, any such interaction between the next technology on a specified
pathway with those which have already been potentially applied in other pathways must be
determined. For example, the incremental effectiveness  of switching from a six-speed
automatic transmission to a dual clutch transmission will depend on the level of engine
technology already applied (e.g., intake cam phasing on a port-fuel injected engine or  a down-
sized, turbocharged, direct injection engine).

      EPA's OMEGA model applies technologies  in packages and according to a fixed
sequence for any particular group of vehicles. This requires that the overall cost and
effectiveness of each package be determined first, considering any and all dis-synergies which
may exist. Then, the incremental cost and effectiveness of each subsequent package is
determined relative to the prior one.

        Thus, while the same baseline vehicle fleet and cost and effectiveness estimates for
technologies are being used in both the CAFE and OMEGA models, the form of the actual
inputs to the model will appear to be different. For more information on EPA's and
NHTSA's unique approaches to modeling,  please refer to each agency's respective
preliminary or draft RIA.

      In order to estimate both technology costs and fuel consumption/COi reduction
estimates, it is necessary for each agency to describe the baseline vehicle characteristics from
which the estimates can be compared. This "baseline" is different from the usage in Chapter
1 of this joint TSD. In Chapter 1, the baseline fleet is the projected fleet in MY 2025 before
accounting for technologies needed to meet the MY 2016 CAFE standards and before
accounting for changes in fleet composition attributable to that rule (those later steps
accounted for independently by each agency in developing their separate reference fleets). In
the present context, it indicates the vehicle  types and technologies  that will be used for
comparison from a strict cost and effectiveness point of view. These baselines may be
slightly different for the two  agencies.  For EPA, unless noted elsewhere, the baseline  vehicle
is defined as a vehicle with a port-fuel injected, naturally aspirated gasoline engine with fixed

                                            3-224

-------
                                     Technologies Considered in the Agencies' Analysis
valve timing and lift. The baseline transmission is a 4-speed automatic, and the vehicle has
no hybrid systems. For NHTSA, unless noted elsewhere, the baseline vehicle is the actual
vehicle as it exists in the baseline fleet, because NHTSA models each unique vehicle
separately.
                                           3-225

-------
                                   Technologies Considered in the Agencies' Analysis
References:

1 U.S. EPA, "Project Report:  Computer Simulation of Light-Duty Vehicle Technologies for
Greenhouse Gas Emission Reduction in the 2020-2025 Timeframe", Contract No. EP-C-11-
007, Work Assignment 0-12,

2 U.S. EPA, "Light-Duty Technology Cost Analysis Pilot Study," Contract No. EP-C-07-069,
Work Assignment 1-3, December 2009, EPA-420-R-09-020, Docket EPA-HQ-OAR-2009-
0472- 11282

3 U.S. EPA, "Light-Duty Technology Cost Analysis Pilot Study Peer Review Report —
Response to Comments Document", December 21, 2009, EPA-HQ-OAR-2009-0472-11285

4 U.S. EPA, "Light-duty Technology Cost Analysis - Report on Additional Case Studies,"
EPA-HQ-OAR-2009-0472-11604

5 FEV, Inc., "Light-Duty Technology Cost Analysis, Report on Additional Transmission,
Mild Hybrid, and Valvetrain Technology Case Studies", Contract No. EP-C-07-069, Work
Assignment 3-3. November 2011.

6 FEV, Inc., "Light-Duty Technology Cost Analysis, Power-Split and P2 HEV Case Studies",
Contract No. EP-C-07-069, Work Assignment 3-3, EPA-420-R-11-015, November 2011.

7ICF, "Peer Review of FEV Inc. Report "Light Duty Technology Cost Analysis, Power-Split
and P2 Hybrid Electric Vehicle Case Studies", EPA-420-R-11-016, November 2011.

8 FEV, Inc. and U.S. EPA, "FEV Inc. Report 'Light Duty Technology Cost Analysis, Power-
Split and P2 Hybrid Electric Vehicle Case Studies', Peer Review Report - Response to
Comments Document", EPA-420-R-11-017, November 2011.

9 ANL BatPaC model can be found in Docket ID EPA-HQ-OAR-2010-0799.

10 ANL BatPaC model peer review report can be found in Docket ID EPA-HQ-OAR-2010-
0799

11 EPA-420-R-10-901, April 2010.

12 "Interim Joint Technical Assessment Report: Light-Duty Vehicle Greenhouse Gas
Emission Standards and Corporate Average Fuel Economy Standards for Model Years 2017-
2025,"September 2010.

13 75 FR 76337.

14 RTI International. Automobile Industry Retail Price Equivalent and Indirect Cost
Multipliers.  February 2009.  http://www.epa.gov/otaq/ld-hwy/420r09003.pdf; Rogozhin,
A.,et al., "Using indirect cost multipliers to estimate the total cost of adding new technology

                                         3-226

-------
                                    Technologies Considered in the Agencies' Analysis
in the automobile industry," International Journal of Production Economics (2009),
doi:10.1016/j.ijpe.2009.11.031.  The peer review for the RTI report is at
http://www.epa.gov/otaq/ld-hwy/420r09004.pdf.

15 Rogozhin, A.,et al., "Using indirect cost multipliers to estimate the total cost of adding new
technology in the automobile industry," International Journal of Production Economics
(2009), doi:10.1016/j.ijpe.2009.11.031.

16 Helfand, Gloria, and Todd Sherwood, "Documentation of the Development of Indirect Cost
Multipliers for Three Automotive Technologies," August 2009.

17 Rogozhin, A.,et al., "Using indirect cost multipliers to estimate the total cost of adding new
technology in the automobile industry," International Journal of Production Economics
(2009),doi:10.1016/j.ijpe.2009.11.031.

18 FEV, Inc., "Potential Stranded Capital Analysis on EPA Light-Duty Technology Cost
Analysis", Contract No. EP-C-07-069 Work Assignment 3-3, November 2011.

19 76 FR 57106 (September 15, 2011).

20 U.S. EPA, "A Study of Potential Effectiveness of Carbon Dioxide Reducing Vehicle
Technologies", Contract No. EP-C-06-003, Work Assignment  1-14,, June 2008, Report*
EPA420-R-08-004, available in the EPA docket EPA-HQ-OAR-2009-0472 and on the
internet at http://www.epa.gov/otaq/technology/420r08004a.pdf

21 76 FR 57106 (September 15, 2011)

22 Woldring, D., Landenfeld, T., Christie, M.J., 2007, "DI Boost: Application of a High
Performance Gasoline Direct Injection Concept." SAE Technical Paper Series No. 2007-01-
1410.

23 Kapus, P.E., Fraidl, O.K., Prevedel, K., Fuerhapter, A., 2007, "GDI Turbo - The Next
Steps." JSAE Technical Paper No. 20075355.

24 Hancock, D., Fraser, N., Jeremy, M., Sykes, R., Blaxill, H., 2008, "A New 3 Cylinder 1.21
Advanced Downsizing Technology Demonstrator Engine." SAE Technical Paper Series No.
2008-01-0611.

25 Lumsden, G., OudeNijeweme, D., Fraser, N. Blaxill, H., 2009, "Development of a
Turbocharged Direct Injection Downsizing Demonstrator Engine." SAE Technical Paper
Series No. 2009-01-1503.

26 Cruff, L., Kaiser, M., Krause, S., Harris, R., Krueger, U., Williams, M., 2010, "EBDI® -
Application of a Fully Flexible High Bmep Downsized Spark Ignited Engine." SAE Technical
Paper Series No. 2010-01-0587.
                                          3-227

-------
                                   Technologies Considered in the Agencies' Analysis
27 Turner, J.W.G., Pearson, R.J., Curtis, R., Holland, B., 2009, "Sabre: A Cost Effective
Engine Technology Combination for High Efficiency, High Performance and Low CO2
Emissions." Low Carbon Vehicles 2009: IMechE Conference Proceedings.

28 Taylor, J., Eraser, N., Wieske, P., 2010, "Water Cooled Exhaust Manifold and Full Load
EGR Technology Applied to a Downsized Direct Injection Spark Ignition Engine." SAE
Technical Paper Series No. 2010-01-0356.

29 Roth, D.B., Keller, P, Becker, M., 2010, "Requirements of External EGRSystems for Dual
Cam Phaser Turbo GDI Engines." SAE Technical Paper Series No. 2010-01-0588.

30 Burress, T.A., C.L. Coomer, S.L. Campbell, L.E.  Seiber, R.H. Staunton, and J.P.
Cunningham, 2008, "Evaluation of the 2007 Toyota Camry Hybrid Synergy Drive System."
ORNL technical report TM-2007/190.

31 ICE International, 2011.  "Peer Review of  Ricardo, Inc. Draft Report, 'Computer
Simulation of Light-Duty Vehicle Technologies for Greenhouse Gas Emission Reduction in
the 2020-2025 Timeframe'".  Contract No. EP-C-06-094, Work Assignment 4-04, Docket
EPA-HQ-OAR-2010-0799, September 30, 2011.

32 Ricardo Inc., "Project Report Computer Simulation of Light-duty Vehicle Technologies for
Greenhouse Gas Emission Reduction in the 2020-2025 Timeframe," 2011, Contract No. EP-
C-l 1-007, Work Assignment 0-12, Docket EPA-HQ-OAR-2010-0799, November, 2011.

33 Systems Research and Applications Corporation (SRA),  "Peer Draft Response to Peer
Review of: Ricardo, Inc. Draft Report, 'Computer Simulation of Light-Duty Vehicle
Technologies for Greenhouse Gas Emission Reduction in the 2020-2025 Timeframe'" EPA
Contract No. EP-C-11-007, Work Assignment 0-12, Docket ID EPA-HQ-OAR-2010-0799,
November, 2011.

34 "Impact of Friction Reduction Technologies on Fuel Economy," Fenske, G. Presented at
the March 2009 Chicago Chapter Meeting of the 'Society of Tribologists and Lubricated
Engineers' Meeting, March 18th, 2009. Available at:
http://www.chicagostle.org/program/2008-
2009/Impact%20of%20Friction%20Reduction%20Technologies%20on%20Fuel%20Econom
y%20-%20with%20VGs%20removed.pdf (last accessed July 9, 2009).

35 "Light-Duty Automotive Technology and Fuel Economy Trends: 1975 Through 2007",
EPA420-S-07-001, September 2007, Docket EPA-HQ-OAR-2009-0472-0137. Available at
http://www.epa.gov/oms/cert/mpg/fetrends/fetrends-archive.htm (last accessed July 9, 2009).

36 Paul Whitaker, Ricardo, Inc., "Gasoline Engine Performance And Emissions - Future
Technologies and Optimization," ERC Symposium, Low Emission Combustion Technologies
for Future 1C Engines, Madison, WI, June 8-9, 2005, Docket EPA-HQ-OAR-2009-0472-
0155. Available at
                                         3-228

-------
                                   Technologies Considered in the Agencies' Analysis
http://www.erc.wise.edu/symposiums/2005_S}miposium/June%208%20PMAVhitaker_Ricard
o.pdf (last accessed Nov. 9, 2008).

37 "Development and Optimization of the Ford 3.5L V6 EcoBoost Combustion System,"
Yi,J., Wooldridge, S., Coulson, G., Hilditch, J. Iyer, C.O., Moilanen, P., Papaioannou, G.,
Reiche, D. Shelby, M., VanDerWege, B., Weaver, C. Xu, Z., Davis, G., Hinds, B. Schamel,
A. SAE Technical Paper No. 2009-01-1494, 2009, Docket EPA-HQ-OAR-2009-0472-2860.

38 David Woldring and Tilo Landenfeld of Bosch, and Mark J. Christie of Ricardo, "DI Boost:
Application of a High Performance Gasoline Direct Injection Concept," SAE 2007-01-1410.
Available at http://www.sae.org/technical/papers/2007-01-1410 (last accessed Nov. 9, 2008)

39 Yves Boccadoro, Loic Kermanac'h, Laurent Siauve, and Jean-Michel Vincent, Renault
Powertrain Division, "The New Renault TCE 1.2L Turbocharged Gasoline Engine," 28th
Vienna Motor Symposium, April 2007.

40 Tobias Heiter, Matthias Philipp, Robert Bosch, "Gasoline Direct Injection: Is There a
Simplified, Cost-Optimal System Approach for an Attractive Future of Gasoline Engines?"
AVL Engine & Environment Conference, September 2005.

41 U.S. Environmental Protection Agency, "Draft Report - Light-Duty Technology Cost
Analysis Pilot Study," Contract No. EP-C-07-069, Work Assignment 1-3, September 3, 2009,
Docket EPA-HQ-OAR-2009-0472-0149.

42 Kaiser, M., Krueger, U., Harris, R., Cruff, L.  "Doing More with Less - The Fuel Economy
Benefits of Cooled EGR on a Direct Injected Spark Ignited Boosted Engine," SAE Technical
Paper Series, No. 2010-01-0589.

43 Kapus, P.E., Fraidl, O.K., Prevedel, K., Fuerhapter, A. "GDI Turbo - The Next Steps,"
JSAE Technical Paper No. 20075355, 2007.

44 "EPA Staff Technical Report: Cost and Effectiveness Estimates of Technologies Used to
Reduce Light-duty Vehicle Carbon Dioxide Emissions," EPA420-R-08-008, March 2008,
Docket EPA-HQ-OAR-2009-0472-0132.

45 Cairns et al., Lotus, "Low Cost Solutions for Improved Fuel Economy in Gasoline
Engines," Global Powertrain Congress September 27-29, 2005, vol. 33. Available at
http://www.gpc-icpem.org/pages/publications.html (last accessed Nov. 9, 2008).

46 Tim Lake, John Stokes, Richard Murphy, and Richard Osborne of Ricardo and Andreas
Schamel of Ford-Werke, "Turbocharging Concepts for Downsized DI Gasoline Engines,"
VKA/ika Aachen Colloquium 2003. Available  at
http://cat.inist.fr/?aModele=afficheN&cpsidt=16973598  (last accessed Nov.  9, 2008).
                                          3-229

-------
                                   Technologies Considered in the Agencies' Analysis
47
http://media.gm.com/content/media/us/en/gm/news.detail.html/content/Pages/news/us/en/201
l/Jul/0722 cruze diesel (last accessed: October 21, 2011)

48 http://www.mazda.com/publicity/release/2010/201010/101020a.html (last accessed:
October 21,2011)

49 U.S. EPA, "Computer Simulation of Light-Duty Vehicle Technologies for Greenhouse Gas
Emission Reduction in the 2020-2025 Timeframe", Contract No. EP-C-11-007, Work
Assignment 0-12, Docket EPA-HQ-OAR-2010-0799, November, 2011.

50 General Motors, news release, "From Hybrids to Six-Speeds, Direct Injection And More,
GM's 2008 Global Powertrain Lineup Provides More Miles with Less Fuel" (released Mar. 6,
2007). Available at
http://www.gm.com/experience/fuel_economy/news/2007/adv_engines/2008-powertrain-
lineup-082707.jsp  (last accessed Sept. 18, 2008).

51 "EPA Staff Technical Report: Cost and Effectiveness Estimates of Technologies Used to
Reduce Light-duty Vehicle Carbon Dioxide Emissions" Environmental Protection Agency,
EPA420-R-08-008, March 2008, at page 17, Docket EPA-HQ-OAR-2009-0472-0132.

52 FEV, Inc., "Light-Duty Technology Cost Analysis, Report on Additional Transmission,
Mild Hybrid, and Valvetrain Technology Case Studies", Contract No. EP-C-07-069, Work
Assignment 3-3, Docket EPA-HQ-OAR-2010-0799, November 2011.

53 FEV, Inc., "Light-Duty Technology Cost Analysis, Report on Additional Transmission,
Mild Hybrid, and Valvetrain Technology Case Studies", Contract No. EP-C-07-069, Work
Assignment 3-3, Docket EPA-HQ-OAR-2010-0799, November 2011.

54  http://automobiles.honda.com/insight-hybrid/features.aspx?Feature=ima (last accessed on
March 29, 2010; web page printout is contained in Docket EPA-HQ-OAR-2009-0472 as
"Honda IMA webpage.pdf").

55  "Latest Chevrolet Volt Battery Pack and Generator Details and Clarifications."  Lyle
Dennis interview of Rob Peterson (GM) regarding the all-electric drive range of the GM Volt,
August 29, 2007. Accessed on the Internet on June 30, 2009 at: http://gm-
volt.com/2007/08/29/latest-Chevrolet-volt-battery-pack-and-generator-details-and-
clarifications/

56  "Active Combination of Ultracapacitorsand Batteries for PHEV ESS." Bohn, T. U.S.
Department of Energy 2009 Vehicle Technologies Merit Review, May 20,  2009, Docket
EPA-HQ-O AR-2009 -0472-0163.

57 Nelson, P.A., Santini, D.J., Barnes, J. "Factors Determining the Manufacturing Costs of
Lithium-Ion Batteries for PHEVs," 24th World Battery, Hybrid and Fuel Cell Electric Vehicle
Symposium and Exposition EVS-24, Stavenger, Norway, May  13-16, 2009 (www.evs24.org).

                                          3-230

-------
                                   Technologies Considered in the Agencies' Analysis
58
  Santini, D.J., Gallagher, K.G., and Nelson, P.A. "Modeling of Manufacturing Costs of
Lithium-Ion Batteries for HEVs, PHEVs, and EVs," Paper to be presented at the 25th World
Battery, Hybrid and Fuel Cell Electric Vehicle Symposium and Exposition, EVS-25,
Shenzhen, China, November 5-9, 2010 (www.evs25.org).  Advance draft provided by DJ.
Santini, Argonne National Laboratory, August 24, 2010.

59 Argonne National Laboratories, "Modeling the Performance and Cost of Lithium-Ion
Batteries for Electric-Drive Vehicles", 2011

60 Peer review of report: Argonne National Laboratories, "Modeling the Performance and
Cost of Lithium-Ion Batteries for Electric-Drive Vehicles", 2011 can be found in Docket ID
EPA-HQ-OAR-2010-0799.

61 Argonne National Laboratories BatPac model can be found in Docket ID EPA-HQ-OAR-
2010-0799.

62 Nelson,  P., Gallagher, K., Bloom, I. Dees, D.W. "BatPac Model Beta." Microsoft Excel™-
based spreadsheet model with example generic inputs and outputs. The model is available in
docket EPA-HQ-OAR-2010-0799.

63 Anderman, M. (2010) Feedback on ARB's Zero-Emission Vehicle Staff Technical Report
of 11/25/2009 including attachment A: Status of EV Technology Commercialization,
Advanced  Automotive Batteries, January 6, 2010

64 Frost &  Sullivan (2009b) World Hybrid Electric and Electric Vehicle Lithium-ion Battery
Market, N6BF-27, Sep 2009

65 Barnett, B. (2009) "PHEV Battery Cost Assessment" TIAX LLC presentation at U.S.
DOE/EERE 2009 Vehicle Technologies Program Annual Merit Review, May 19, 2009.
Accessed on the Internet on November 14, 2011 at:
http://wwwl.eere.energy.gov/vehiclesandfuels/pdfs/merit_review_2009/energy_storage/es_02
_barnett.pdf

66 Boston Consulting Group (2010) Batteries for Electric Cars - Challenges, Opportunities,
and the Outlook to 2020

67 National Research Council (2010) Transitions to Alternative Transportation Technologies-
Plug-in Hybrid Electric Vehicles.

68 K. G. Gallagher, P. A. Nelson,  (2010) "An Initial BatPac Variation Study" Docket ID
EPA-HQ-OAR-2010-0799
69 "Hyundai ups tech ante with Sonata Hybrid," Automotive News, August 2, 2010.

  "Chevrolet Stands Behind >
Warranty," GM Press release
70 "Chevrolet Stands Behind Volt With Standard Eight-Year, 100,000-Mile Battery
                                          3-231

-------
                                    Technologies Considered in the Agencies' Analysis
(http://media.gm.com/content/media/us/en/news/news_detail.brand_gm.html/content/Pages/n
ews/us/en/2010/July/0714_volt_battery)

71 "Nissan's new 2012 hybrid system aims for 1.8-L efficiency with a 3.5-L V6," SAE
Automotive Engineering Online, February 15, 2010.

72 "Lithium-ion Battery," Nissan Technological Development Activities (http://www.nis san-
global.com/EN/TECHNOLOGY/INTRODUCTION/DETAILS/LI-ION-EV/), 2009.

73 Plug-in Hybrid Electric Vehicle Charging Infrastructure Review (November 2008)
INL/EXT-08-15058, U.S. Department of Energy Vehicle Technologies Program - Advanced
Vehicle Testing Activity. Last accessed on the Internet on November 14, 2011 at the
following URL: http://avt.inl.gov/pdf/phev/phevInfrastructureReport08.pdf

74 May, J.W., Mattilla, M. "Plugging In: A Stakeholder Investment Guide for Public Electric-
Vehicle Charging Infranstructure." Rocky Mountain Institute - Project Get Ready, July 2009.
Last accessed on the Internet on November 14, 2011 at the following URL:
http://projectgetready.com/docs/Prugging%20In%20-
%20A%20Stakeholder%20Investment%20Guide.pdf

75 "Electric Vehicle Charging Infrastructure Deployment Guidelines - British Columbia."
Report by the Electric Transportation Engineering Corporation for Natural Resources  Canada
and BCHydro. Last accessed on the Internet on November 14, 2011 at the following URL:
http://www.bchvdro.com/etc/medialib/internet/documents/environment/EVchar gin g_infrastru
cture  guidelines09.Par.0001.File.EV%20Charging%20Infrastructure%20Guidelines-BC-
Aug09.pdf

76 "Electrification Roadmap - Revolutionizing Transportation and Achieving Energy
Security." Electrification Coalition, November 2009. Accessed on the Internet on November
14, 2011 at: http://www.electrificationcoalition.org/sites/default/files/SAF 1213  EC-
Roadmap_v 12_Online.pdf

77 "The Disappearing Spare Tire" Edmunds.com, May 11, 2011;
http://www.edmunds.com/car-buying/the-disappearing-spare-tire.html (last accessed
9/6/2011)

78 see U.S Patent 5,227,425, Rauline to Michelin, July 13, 1993

79 "Light-Duty Automotive Technology and Fuel Economy Trends: 1975 Through 2008",
EPA420-R-08-015, U.S. Environmental Protection Agency Office of Transportation and Air
Quality, September 2008

80 Lutsey, "Review of technical literature and trends related to automobile mass-reduction
technology", UCD-ITS-RR-10-10, May 2010
                                          3-232

-------
                                    Technologies Considered in the Agencies' Analysis
81 Adrian Lund, IIHS, "The Relative Safety of Large and Small Passenger Vehicles",
http://www.nhtsa.gov/staticfiles/rulemaking/pdf/MSS/MSSworkshop-Lund.pdf

82 NAS 2010, "Assessment of Fuel Economy Technologies for Light-Duty Vehicles". June
2010

83 Lutsey, "Review of technical literature and trends related to automobile mass-reduction
technology", UCD-ITS-RR-10-10, May 2010

84  Reddy, "Preliminary Vehicle Mass Estimation Using Empirical Subsystem Influence
Coefficients," Auto-Steel Partnership Report, May 2007. Available at http://www.a-
sp.org/database/custom/Mass%20Compounding%20-%20Final%20Report.pdf (last accessed
Aug. 17,2011).

85 Malen and Reddy, "Preliminary Vehicle Mass Estimation Using Empirical Subsystem
Influence Coefficients," Auto-Steel Partnership Report, May 2007.  Available at
http://www.a-sp.org/database/custom/Mass%20Compounding%20-%20Final%20Report.pdf
(last accessed Aug. 17, 2011).

86 Bull, M., R. Chavali, A. Mascarin, "Benefit Analysis: Use of Aluminum Structures in
Conjunction with Alternative Powertrain Technologies in Automobiles," Aluminum
Association Research Report,  May 2008. Available at
http://aluminumintransportation.org/downloads/IBIS-Powertrain-Study.pdf (last accessed
Aug. 17,2011).
87
  http://msl.mit.edu/students/msl_theses/Bjelkengren_C-thesis.pdf
88 NAS, "Assessment of Fuel Economy Technologies for Light-Duty Vehicles", pg 100, 2011

89 Ford Sustainability Report 2010/11, http://corporate.ford.com/microsites/sustainabilitv-
report-2010-ll/issues-climate-plan-economy (last accessed Aug. 26, 2011)

90 http://www.worldautosteel.org/FSV_OverviewReport_Phase2_FINAL_201 10430.pdf

91 http://www.sae.org/mags/AEI/7695

92 American Iron and Steel Institute (AISI), 2009. "New Study Finds Increased Use of
Advanced High-Strength Steels Helps Decrease Overall Vehicle Weight."
http://www.steel.org/AM/Template.cfm?Section=Press_Releases9&TEMPLATE=/CM/Conte
ntDisplay.cfm&CONTENTID=32077.

93 Ford, 2010. "The S.OLiter is Back: 2011 Ford Mustang GT Leads Class with 412 HP, Fuel
Efficiency, Chassis Dynamics." http://media.ford.com/article_display.cfm?article_id=31645..

94 Keith, D., 2010. "HSS, AHSS and aluminum jockey for position in the race to cut auto curb
weight." American Metal Market Monthly.


                                          3-233

-------
                                    Technologies Considered in the Agencies' Analysis
95 U.S. Environmental Protection Agency (U.S. EPA), 2009b. Draft Regulatory Impact
Analysis: Proposed Rulemaking to Establish Light-Duty Vehicle Greenhouse Gas Emission
Standards and Corporate Average Fuel Economy Standards. September. EPA-420-D-09-003.

96 The New York Times, "Automakers Resolve to Drop a Few pounds", Sept 2011.
http://www.nytimes.com/2011/09/18/automobiles/autoshow/in-frankfurt-automakers-vow-to-
drop-a-few-pounds.html?_r=l&smid=tw-nytimeswheels&seid=auto

97 "Review of technical literature and trends related to automobile mass-reduction
technology", May 2010
http://www.arb. ca.gov/msprog/levprog/leviii/meetings/051810/2010_ucd-its-rr-10-10.pdf
98 "Review of technical literature and trends related to automobile mass-reduction
technology", May 2010
http://www.arb. ca.gov/msprog/levprog/leviii/meetings/051810/2010_ucd-its-rr-10-10.pdf

99 Frank Field, Randolph Kirchain and Richard Roth, Process cost modeling: Strategic
engineering and economic evaluation of materials technologies, JOM Journal of the Minerals,
Metals and Materials Society, Volume 59, Number 10, 21-32. Available at
http://msl.mit.edu/pubs/docs/Field_KirchainCM_StratEvalMatls.pdf (last accessed Aug. 22,
2011).

100 U.S. E.P.A., Light-Duty Automotive Technology, Carbon Dioxide emissions, and Fuel
Economy Trends: 1975 Through 2010, http://epa.gov/otaq/fetrends.htm

101 EPA; Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy
Trends: 1975 Through 2010; Figure 28, pg 69

102 EPA; Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy
Trends: 1975 Through 2010; Table 13, pg 50

103 DeCicco, J. M. (2010). A fuel efficiency horizon for U.S. automo- biles. Technical report,
University of Michigan, School of Natural Resources and Environment. Report prepared for
The Energy Foundation. Available online at
http://energy.umich.edu/info/pdfs/Fuel%20Efficiency%20Horizon%20FINAL.pdf.

104 Zoepf, Stephen.  (2011) "Automotive Features: Mass Impact and Deployment
Characterization" Masters thesis. Massachusetts Institute of Technology, Technology and
Policy Program, Engineering Systems Division. Available online at:
http://web.mit.edu/sloan-auto-lab/research/beforeh2/files/Zoepf_MS_Thesis.pdf
105
   Murphy, John, Bank of America Merrill Lynch, "Car Wars 2010-2013," July 15, 2009
106 Ellison, D. J., Clark, K. B., Fujimoto, T., and suk Hyun, Y. (1995). Product development
performance in the auto industry: 1990s update. Technical report, International Motor Vehicle
Program.
                                          3-234

-------
                                   Technologies Considered in the Agencies' Analysis
107 Autocar, "VW's New Platform for 60 Models," November 12, 2009,
http://www.autocar.co.uk/News/NewsArticle.aspx?AR=244881

108 Mayne, Eric, "Aligning Capacity, Demand Poses Ultimate Brain Teaser," Wards Auto,
July 29, 2008,

109 Pope, Byron, "Ford's Cleveland Engine No. 1 to Build 3.7L V-6," Wards Auto, March 6,
2009

110 Ford Motor Company, Ford Motor Company Business Plan Submitted to the House
Financial Services Committee, December 2, 2008

111 Rogers, Christina "GM to halve number of platforms globally " Detroit News, August 10,
2011.
                                         3-235

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

Chapter 4:  Economic and Other Assumptions Used in the Agencies'
       Analysis

4.1 How the Agencies use the economic and other assumptions in their analyses

       Improving new vehicles' fuel efficiency and reducing greenhouse gas (GHG)
emissions provides direct benefits to their buyers and users by reducing fuel consumption and
fuel costs throughout those vehicles' lifetimes, stimulating increased vehicle use through the
fuel economy rebound effect, and often increasing vehicles' driving range so that they require
less frequent refueling.  At the same time, the reduction in fuel use that results from requiring
higher fuel economy and reducing GHGs also produces wider benefits to the U.S. economy
by lowering the cost of economic externalities that result from U.S. petroleum consumption
and imports. This occurs because reducing U.S. oil consumption and imports reduces the
global price of petroleum, lowers the potential costs from disruptions in the flow of oil
imports, and potentially reduces federal outlays to secure imported oil supplies and cushion
the U.S. economy against their potential interruption. Reducing fuel consumption  and GHGs
also lowers the economic costs of environmental externalities resulting from fuel production
and use, including reducing the impacts on human health from emissions of criteria air
pollutants, and reducing future economic damages from potential changes in the global
climate caused by greenhouse gas emissions.

       These social benefits are partly offset  by the increase in fuel use that results from
added vehicle use due to the fuel economy rebound effect, as well as by added costs from the
increased congestion, crashes, and noise caused by increased vehicle use. They would also be
offset by any loss in the utility that new vehicles provide to their buyers (and subsequent
owners) if manufacturers include reductions in vehicles' performance, carrying capacity, or
comfort as part of their strategies to comply with higher fuel economy requirements and GHG
standards. However, the agencies' analyses supporting the proposed standards do not
anticipate any such reductions in utility as being necessary, and the analysis seeks to include
the costs to manufacturers of preserving vehicle capabilities.11  For instance, the costs of
engine downsizing include the costs of turbocharging the engine to maintain its performance.
The total economic benefits from requiring higher fuel economy and reducing GHGs are
likely to be substantial, and EPA and NHTSA have developed detailed estimates of the
economic benefits from adopting more stringent standards.

       This chapter discusses the  common economic and other values used by both NHTSA
and EPA in their rulemaking analyses. These inputs incorporate a range of forecast
information, economic estimates, and input parameters. This chapter describes the sources
that EPA and NHTSA  have relied upon for this information, the rationale underlying each
assumption, and the agencies' estimates of specific parameter values.  These common values
a Two exceptions - hybrid vehicles that may have some limited towing capacity, and electric vehicles - are
discussed elsewhere.

                                            4-2

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

are then used as inputs into each agency's respective modeling and other analyses of the
economic benefits and costs of the EPA and NHTSA programs. While the underlying input
values are common to both agencies, program differences, and differences in the way each
agency assesses its program that result in differing benefits estimates. This issue is discussed
further in Section I.C of the preamble to the joint rulemaking.

4.2 What assumptions do the agencies use in the impact analyses?

      4.2.1  The on-road fuel economy "gap"

      4.2.1.1 Definition and past use by EPA and NHTSA

      In aggregate, actual fuel economy levels achieved by vehicles in on-road driving fall
significantly short of their levels measured  in the laboratory-like test conditions and two-cycle
tests used under the CAFE program to determine the fuel economy ratings for different
models for purposes of compliance with the CAFE and COi standards. The test procedure
used to determine compliance is highly controlled, and does not reflect real-world driving in a
variety of ways - real-world driving tends to be more aggressive than the Federal Test
Procedure (FTP) and Highway Fuel Economy Test (HFET) test cycles used to establish
compliance with the GHG and CAFE regulations. Real world driving tends to include more
stops and starts and more rapid acceleration/deceleration, and may include the use of
technologies like air-conditioning that reduce fuel economy but that are not exercised on the
test cycle.1  There are also a number of elements that affect real-world achieved fuel economy
which are not measured on the two cycle GHG/CAFE  compliance test, such as wind
resistance, road roughness, grade, temperature, and fuel energy content.  The agencies'
analyses for this proposal recognize this gap, and account for it by adjusting the fuel economy
performance downward from its rated value. In December 2006, EPA adopted changes to its
regulations  on fuel economy labeling, which were intended to bring vehicles' label fuel
economy levels  seen by consumers shopping for new vehicles closer to their actual on-road
fuel economy levels.

      Comparisons of on-road and CAFE fuel economy levels developed by EPA as part of
its 2006 Final Rule implementing new fuel economy labeling requirements for new vehicles
indicated that actual on-road fuel economy for light-duty vehicles average about 20 percent
lower than compliance fuel economy ratings.3 While there is great heterogeneity among
individual drivers, as discussed in the referenced material, the 20 percent figure appears to
represent an accurate average for modeling a fleet.  For example, if the overall EPA fuel
economy rating  of a light truck is 20 MPG, the on-road fuel economy actually achieved by a
typical driver of that vehicle is expected to  be 16 mpg  (20*.80). In its analysis supporting the
Final Rule establishing CAFE standards for MY 2011, NHTSA employed EPA's revised
estimate of this on-road fuel economy gap in its analysis of the fuel savings resulting from
alternative fuel efficiency standards.  EPA and NHTSA likewise employed this fuel economy
gap for estimating fuel savings in the MYs  2012-2016  rulemaking and in the Interim Joint
Technical Assessment Report (TAR) analysis for MYs 2017-2025.
                                            4-3

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

       An analysis conducted by NHTSA confirmed that EPA's estimate of a 20 percent gap
between test and on-road fuel economy for the majority of vehicles is well-founded. NHTSA
used data on the number of passenger cars and light trucks of each model year that were in
service (registered for use) during each calendar year from 2000 through 2006; average fuel
economy for passenger cars and light trucks produced during each model year; and estimates
of average miles driven  per year by cars and light trucks of different ages during each
calendar year over that period.  These data were combined to develop estimates of the usage-
weighted average fuel economy that the U.S. passenger car and light truck fleets would have
achieved during each year from 2000 through 2006 under test conditions.

       Table 4-1 compares NHTSA's estimates of fleet-wide average fuel economy under
test conditions for 2000  through 2006 to the Federal Highway Administration's (FHWA)
published estimates of on-road fuel economy achieved by passenger cars and light trucks
during each of those years. As it shows, FHWA's estimates of fuel economy for passenger
cars ranged from 21-23  percent lower than NHTSA's estimates of its fleet-wide average value
under test conditions over this period, and FHWA's estimates of fuel economy for light trucks
ranged from 16-18 percent lower than NHTSA's estimates of its fleet-wide average value
under test conditions. Thus, these results appear to confirm that the 20 percent on-road fuel
economy gap represents a reasonable estimate for use in evaluating the fuel savings likely to
result from more stringent fuel economy and CO2 standards in MYs 2017-2025.

         Table 4-1 Estimated Fleet-Wide Fuel Economy of Passenger Cars and Light Trucks
                          Compared to Reported Fuel Economy
YEAR
2000
2001
2002
2003
2004
2005
2006
Avg.,
2000-
2006
PASSENGER CARS
NHTSA
Estimated
Test MPG
28.2
28.2
28.3
28.4
28.5
28.6
28.8
28.4
FHWA
Reported
MPG
21.9
22.1
22.0
22.2
22.5
22.1
22.5
22.2
Percent
Difference
-22.2%
-21.7%
-22.3%
-21.9%
-21.1%
-22.8%
-21.8%
-22.0%
LIGHT-DUTY TRUCKS
NHTSA
Estimated
Test MPG
20.8
20.8
20.9
21.0
21.0
21.1
21.2
21.0
FHWA
Reported
MPG
17.4
17.6
17.5
17.2
17.2
17.7
17.8
17.5
Percent
Difference
-16.3%
-15.5%
-16.2%
-18.0%
-18.3%
-16.3%
-16.2%
-16.7%
       We are aware of two potential issues involved in these estimates. One, the estimates
of total annual car and truck VMT are developed by the states and submitted to FHWA. Each
state uses its own definition of a car and a truck.  For example, some states classify minivans
as cars and some as trucks. Thus, there are known inconsistencies with these estimates when
                                            4-4

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

evaluated separately for cars and trucks. Also, total gasoline consumption can be reasonably
estimated from excise tax receipts, but separate estimates for cars and trucks are not available.
We are not aware of the precise methodology used to develop the distinct on-road fuel
economy estimates for cars and trucks developed by FHWA.  We do not believe that they are
based on direct measurements from substantial numbers of vehicles, as no such test programs
were found by EPA during its fuel economy labeling rule in 2006. Also, the year-to-year
consistency for both car and truck fuel economy implies some methodology other than direct
measurement. For this reason, NHTSA and EPA are not using distinct on-road fuel economy
gaps for cars and trucks, but one common value of 20 percent for both vehicle classes for
purposes of estimating the fuel savings of the standards. This figure lies between the separate
estimated for cars and light trucks reported in Table 4-1.

       For purposes of the MYs 2012-2016 rulemaking, the TAR, and this current
rulemaking for MYs 2017-2025, then, the agencies are assuming that  the on-road fuel
economy gap for liquid fuel is 20 percent.  As in the TAR, the agencies assume that the
overall energy shortfall for the electric drivetrain (for vehicles that have those instead of or in
addition to gasoline engines) is 30 percent when driven on wall electricity.  The 30 percent
value was derived from the agencies' engineering judgment based on  several data points.
Foremost among these, during the stakeholder meetings conducted prior to the Interim Joint
TAR, confidential business information (CBI) was supplied by several manufacturers which
indicated that electrically powered vehicles had greater variability in their on-road energy
consumption than vehicles powered by internal combustion engines.  Second, data from
EPA's 2006 analysis of the "five cycle" fuel economy label as part of the rulemaking
discussed above potentially supported a larger on-road  shortfall for vehicles with hybrid-
electric drivetrains4 And third, heavy accessory load, extreme (both high and low)
temperatures, and aggressive driving  have deleterious impacts of unknown magnitudes  on
battery performance. As a counterpoint, CBI provided by several other manufacturers
suggested that the on-road/laboratory differential attributable to electric operation should
approach that of liquid fuel operation in the future.  Consequently, 30 percent was judged by
the agencies to be a reasonable estimate for the Interim Joint TAR, and was carried into the
current analysis.

       The recent 2011 Fuel Economy labeling rule employs a 30% on-road shortfall for
electric vehicles.5 Under the labeling program, for gasoline vehicles, there are two methods
for getting label values: full 5-cycle or derived 5-cycle. Full 5-cycle means all five cycles are
tested, and bag MPG results are used in a set of formulae to determine label MPG.  Derived 5-
cycle involves testing on the FTP and Highway tests and adjusting those values using
regression-based formulae, to get label MPG values. The derived 5-cycle adjustment results in
an ever-increasing adjustment in percentage terms.  However, the data on which  the derived
5-cycle formulae are based ends at roughly 70 MPG, where the adjustment is about 70% or an
on-road gap of 30% (assuming that the five cycle formula represents the real world). For
labeling purposes, lacking any EVs or PHEVs (or any vehicles beyond 70 MPG) in the
database at the time this adjustment was derived, the adjustment was set at 70% for MPG
values beyond 70 MPG.
                                            4-5

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

       Electric vehicles are allowed and expected to use the derived 5-cycle method, which
suggests that their on-road gap will be approximately 30% during the near future.  Individual
EVs may vary, and as additional data becomes available the agencies will consider whether
the 30% average gap remains appropriate.

       4.2.1.2 Considerations in Future Years

       Looking forward to MYs 2017-2025, while the agencies do not forecast changes in
most of the factors discussed above that contribute to the on-road gap in ways that would
change our estimates, the agencies expect that two specific factors will change somewhat that
could affect this analysis.  Specifically, we anticipate changes in the energy content of fuels
sold at retail as a result of the recent EPA Renewable Fuel Standard 2 (RFS2) rulemaking and
E15 waiver decision,6 as well as a change in reference air conditioning efficiency as a result
of the recent MYs 2012-2016 EPA Light Duty Greenhouse Gas rulemaking.

4.2.1.2.1 Air Conditioning

       Air conditioning is a significant contributor to the on-road efficiency gap. While the
air conditioner is turned off during the FTP and HFET tests, in  real world use drivers often
use air conditioning in warm, humid conditions.  The air conditioning compressor can  also be
engaged during "defrost" operation of the heating system.7 In the MYs 2012-2016
rulemaking, the agencies estimated the average impact of an air conditioning system at
approximately 14.3 grams over an SCO3 test for an average vehicle without any of the
improved air conditioning technologies discussed in that rulemaking. For a 27 MPG (330 g
COi/mile) vehicle, this is approximately 20 percent of the total  estimated on-road gap, or
about 4 percent of total fuel consumption.

       In the MYs 2012-2016 rule, EPA estimated that 85 percent of MY 2016 vehicles
would reduce their air conditioning-related CC>2 emissions by 40 percent through the use of
advanced air conditioning efficiency technologies. Incorporating this change would reduce
the average on-road gap by about 2 percent in the reference case.b  However, as shown in
Chapter 5 of the joint TSD air conditioning-related fuel consumption does not proportionally
decrease as overall engine efficiency improves. Unlike most technologies in this rulemaking,
which have a multiplicative reduction on fuel consumption and CCh emissions, the load due
to air conditioning operation is relatively constant across engine efficiency and technology.
As a consequence, as engine efficiency increases, air conditioning operation represents an
increasing percentage of vehicular fuel consumption.0 To some extent, these factors are
expected to counterbalance, so the agencies therefore chose not to make an air conditioning-
related adjustment to the on-road gap for this proposal.
b 4% of the on-road gap x 40% reduction in air conditioning fuel consumption x 85% of the fleet = -2%.
c As an example, the air conditioning load of 14.3 g/mile of CO2 is a smaller percentage (4.3%) of 330 g/mile
thanof260g/mile (5.4%).

                                             4-6

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

4.2.1.2.2  Fuel Energy Content

       Differences in fuel energy content between test conditions and real-world driving is
another contributor to the on-road fuel economy gap. Two-cycle testing for CAFE and CO2
compliance is based on "certification fuel" which contains no ethanol (also known as EO). The
on-road fuel economy gap is estimated with reference to the difference in fuel energy content
between certification fuel and 2004 retail gallons,11 but this rule produces a reduction in
petroleum based fuel  consumption only.6 Volumes of renewable fuels are statutorily fixed
by the Renewable Fuel Standard, so the entirety of the energy savings  will take place as
reduced oil consumption.  To estimate the petroleum fuel savings, we modify the on-road gap
by the average difference in energy content between CY 2004 retail fuel used in the five cycle
analysis and certification fuel.  This results in an approximately 1 % higher fuel economy than
if no additional adjustment was made for fuel energy content, and corresponds to the greater
energy content of certification gasoline as compared to 2004 retail gasoline.

       £"0 Fuel Economy =  2 Cycle Fuel Economy * (1 — gap) * (£"0 BTU/Gallon) /
(2004 BTU/gallon)^Where:
       Gap= 20%
       EO BTU/Gallon = 115,000
       2004 BTU/Gallon = 113,912 (3.14% ethanol, 96.86% petroleum gasoline)

       A related adjustment in fuel energy was made in order to "match" fuel savings to the
fuel prices used in this analysis. As discussed below, the agencies use fuel prices from the
Energy Information Administration's (EIA) Annual Energy Outlook (AEO) 2011 reference
case, which assume approximately 20 percent of the fuel pool by volume is ethanol/ By
contrast, and as shown above, the gasoline savings from this rule are calculated as gallons of
certification fuel, which is is more energy dense than ethanol blended market fuel. To
appropriately apply the AEO prices on a dollar per btu basis, we adjust our certification fuel
savings upwards by approximately 5% (the difference between the energy content of E15
retail fuel and certification) when monetizing the fuel savings.  This  adjustment more
appropriately reflects AEO projections of motor gasoline energy prices.
d The five cycle formula analysis is based on CY 2004 data.
e Ethanol contains approximately 76,000 British Thermal Units (Btu) per gallon as compared to petroleum
gasoline (Indolene), which contains approximately 115,000 Btu.  Thus, a 10 percent ethanol (E10) blend
contains approximately 3.3 percent less energy than a gallon of EO, and an E15 blend contains approximately 5.1
percent less energy than a gallon of EO.
f EIA projects that ethanol replaces approximately 12 percent of the gasoline energy demand by 2035.  This is
greater than 20 percent of the gasoline pool by volume. For calculation of fuel savings for MYs 2017-2025, for
this rulemaking, the agencies made the simplifying assumption that all retail gallons were E15.

                                              4-7

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

       4.2.2   Fuel prices and the value of saving fuel
       Projected future fuel prices are a critical input into the preliminary economic analysis
of alternative fuel efficiency and GHG standards, because they determine the value of fuel
savings both to new vehicle buyers and to society. For this proposal, EPA and NHTSA relied
on the most recent fuel price projections from the U.S. Energy Information Administration's
(EIA) Annual Energy Outlook (AEO) for this analysis, the AEO 2011 Reference Case.  The
Reference Case forecasts inflation-adjusted (constant-dollar) retail gasoline and diesel fuel
prices and represents the EIA's most up-to-date estimate of future prices for petroleum
products.  In the Preface to AEO 2011, the Energy Information Administration describes the
reference case.  They state that, "Projections by EIA are not statements of what will happen
but of what might happen, given the assumptions and methodologies used for any particular
scenario. The Reference case projection is a business-as-usual trend estimate,  given known
technology and technological and demographic trends. The agency has published annual
forecasts of energy prices and consumption levels for the U.S. economy since 1982 in its
AEOs. These forecasts have been widely relied upon by federal agencies for use in regulatory
analysis and for other purposes.  Since 1994, EIA's annual forecasts have been based upon the
agency's National Energy Modeling System (NEMS), which includes detailed representation
of supply pathways, sources of demand, and their interaction to determine prices for different
forms of energy.

       As compared to the gasoline prices used in the MYs  2012-2016 analysis, which  relied
on forecasts from AEO 2010, the AEO 2011 Reference Case fuel prices are largely similar.
They are slightly higher through the year 2020, but slightly lower for most years after 2020
(when both are expressed in 2009 dollars).  A comparison is presented below, Table 4-2.

                Table 4-2 Gasoline Prices for Selected Years in AEO 2010 and 2011
                      (Presented in constant 2009$ and including all taxes)

AEO 2011
AEO 2010
2015
$3.13
$3.10
2020
$3.38
$3.37
2030
$3.64
$3.71
       The retail fuel price forecasts presented in AEO 2011 span the period from 2008
through 2035.  Measured in constant 2009 dollars, the AEO 2011 Reference Case forecast of
retail gasoline prices during calendar year 2017 is $3.25 per gallon, rising gradually to $3.71
by the year 2035 (these values include federal and state taxes).  However, valuing fuel savings
over the full lifetimes of passenger cars and light trucks affected by the standards proposed for
MYs 2017-25 requires fuel price forecasts that extend through 2060, approximately the last
                                             4-8

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

year during which a significant number of MY 2025 vehicles will remain in service.8  To
obtain fuel price forecasts for the years 2036 through 2060, the agency assumes that retail fuel
prices will continue to increase after 2035 at the average annual rate (0.7%) projected for
2017-2035 in the AEO 2011 Reference Case. The years between 2008 and 2016 were not
included in the extrapolation due to the high volatility in the AEO projection for those years.
This assumption results in a projected retail price of gasoline that reaches $4.16 in 2050.

       The value of fuel savings resulting from improved fuel economy and reduced GHG
emissions to buyers of light-duty vehicles is determined by the retail price of fuel,  which
includes federal, state, and any local taxes imposed on fuel sales. Total taxes on gasoline,
including federal, state, and local levies, averaged $0.43 per gallon during 2008, while those
levied on diesel averaged $0.46.  Because fuel taxes represent transfers of resources from fuel
buyers to government agencies, rather than real resources that are consumed in the process of
supplying or using fuel, their value must be deducted from retail fuel prices to determine the
value of fuel savings resulting from more stringent fuel efficiency and GHG standards to the
U.S. economy.8  When calculating the value of fuel saved by an individual driver,  however,
these taxes are included as part of the value of realized fuel savings.  Over the entire period
spanned by the agencies' analysis, this difference causes each gallon of fuel saved to be
valued by about $0.36 (in constant 2009 dollars) more from the perspective of an individual
vehicle  buyer than from the overall perspective of the U.S. economy.11

       In the estimates of costs and benefits presented in the preamble and in the agencies'
RIAs, the agencies have included the full fuel savings over vehicles' expected lifetimes,
discounted to their present values using both 3 and 7 percent discount rates. Additional
discussion of this approach can be found in preamble Sections III.H and IV.

       4.2.3   Vehicle Lifetimes and Survival Rates

       The agencies' analysis of fuel savings and related benefits from adopting more
stringent fuel economy and GHG standards for MYs 2017-2025 passenger cars and light
trucks begin by estimating the resulting changes in fuel use over the entire lifetimes of
affected cars and light trucks.  The change in total fuel consumption by vehicles produced
during each of these model years is calculated as the difference in their total lifetime fuel use
over the entire lifetimes of these vehicles as compared to a reference case.

       The first step in estimating lifetime fuel consumption by vehicles produced during a
model year is to calculate the number of those vehicles expected to remain in service during
g The agency defines the maximum lifetime of vehicles as the highest age at which more than 2 percent of those
originally produced during a model year remain in service. In the case of light trucks, for example, this age has
typically been 36 years for recent model years.
h For society, the fuel taxes represent a transfer payment. By contrast, an individual realizes savings from not
paying the additional money.

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

each future calendar year after they are produced and sold.1  This number is calculated by
multiplying the number of vehicles originally produced during a model year by the proportion
expected to remain in service at the age they will have reached during each  subsequent
calendar year, often referred to as a "survival rate."

       The proportions of passenger cars and light trucks expected to remain in service at
each age are drawn from a 2006 NHTSA study, and are shown in Table 4-39  Note that these
survival rates were calculated against the pre-MY 2011 definitions of cars and light trucks,
because the NHTSA  study has not been updated since 2006. Because the agencies are
unaware of a better data source, these values were used unchanged, and are the same values
used in the MYs 2012-2016 rule and the interim Joint TAR. The rates are applied to vehicles
based on their regulatory class (passenger car or light truck) regardless of fuel type or level of
technology. Survival may vary by other factors, but data to support an investigation do not
currently exist. Additionally, the survival rates are assumed to remain constant over time.

       The survival and annual mileage estimates reported in this section's tables reflect the
convention that vehicles are defined to be of age 1 during the calendar year that coincides
with their model year.  Thus for example, model year 2017 vehicles will be considered to be
of age 1 during calendar year 2017. This convention  is used in order to account for the fact
that vehicles produced during a model year typically are first offered for sale in June through
September of the preceding calendar year (for example, sales of a model year typically begin
in June through September of the previous calendar year, depending on manufacturer). Thus,
virtually all of the vehicles produced during a model year will be in use for  some or all of the
calendar year coinciding with their model year, and they are considered to be of age 1 during
that year.j
1 Vehicles are defined to be of age 1 during the calendar year corresponding to the model year in which they are
produced; thus for example, model year 2000 vehicles are considered to be of age 1 during calendar year 2000,
age 1 during calendar year 2001, and to reach their maximum age of 26 years during calendar year 2025.
NHTSA considers the maximum lifetime of vehicles to be the age after which less than 2 percent of the vehicles
originally produced during a model year remain in service. Applying these conventions to vehicle registration
data indicates that passenger cars have a maximum age of 26 years, while light trucks have a maximum lifetime
of 36 years.  SeeLu, S., NHTSA, Regulatory Analysis and Evaluation Division, "Vehicle Survivability and
Travel Mileage Schedules," DOT HS  809 952, 8-11 (January 2006). Available at http://www-
nrd.nhtsa.dot.gov/Pubs/809952.pdf (last accessed Sept. 9, 2011).
j A slight increase in the fraction of new passenger cars remaining in service beyond age 10 has accounted for a
small share of growth in the U.S. automobile fleet. The fraction of new automobiles remaining in service to
various ages was computed from R.L. Polk vehicle registration data for 1977 through 2005 by the DOT's Center
for Statistical Analysis.

                                               4-10

-------
Economic and Other Assumptions Used in the Agencies' Analysis
          Table 4-3 Survival Rates
VEHICLE AGE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
ESTIMATED
SURVIVAL
FRACTION
CARS
0.9950
0.9900
0.9831
0.9731
0.9593
0.9413
0.9188
0.8918
0.8604
0.8252
0.7866
0.7170
0.6125
0.5094
0.4142
0.3308
0.2604
0.2028
0.1565
0.1200
0.0916
0.0696
0.0527
0.0399
0.0301
0.0227
0
0
0
0
0
0
0
0
0
0
ESTIMATED
SURVIVAL
FRACTION
LIGHT TRUCKS
0.9950
0.9741
0.9603
0.9420
0.9190
0.8913
0.8590
0.8226
0.7827
0.7401
0.6956
0.6501
0.6042
0.5517
0.5009
0.4522
0.4062
0.3633
0.3236
0.2873
0.2542
0.2244
0.1975
0.1735
0.1522
0.1332
0.1165
0.1017
0.0887
0.0773
0.0673
0.0586
0.0509
0.0443
0.0385
0.0334
                    4-11

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

       4.2.4  VMT

       The second step in estimating lifetime fuel use by the cars or light trucks produced
during a future model year is to calculate the total number of miles that they will be driven
during each year of their expected lifetimes.  To estimate total miles driven, the number of
cars and light trucks projected to remain in use during each future calendar year is multiplied
by the average number of miles a surviving car or light truck is expected to be driven at the
age it will have reached in that year. Estimates of average annual miles driven by Calendar
Year 2001 cars and light trucks of various ages were developed by NHTSA from the Federal
Highway Administration's 2001 National Household Travel Survey, and are reported in Table
4-4.  These estimates represent the typical number of miles driven by a surviving light duty
vehicle at each age over its estimated full lifetime. To determine the number of miles a
typical vehicle produced during a given model year is expected to be driven at a specific age,
the average annual mileage for a vehicle of that model year and age is multiplied by the
corresponding survival rate for vehicles of that age.
                                            4-12

-------
   Economic and Other Assumptions Used in the Agencies' Analysis
Table 4-4 CY 2001 Mileage Schedules based on NHTS Data
VEHICLE AGE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
ESTIMATED
VEHICLE MILES
TRAVELED
CARS
14,231
13,961
13,669
13,357
13,028
12,683
12,325
11,956
11,578
11,193
10,804
10,413
10,022
9,633
9,249
8,871
8,502
8,144
7,799
7,469
7,157
6,866
6,596
6,350
6,131
5,940
0
0
0
0
0
0
0
0
ESTIMATED
VEHICLE MILES
TRAVELED
LIGHT TRUCKS
16,085
15,782
15,442
15,069
14,667
14,239
13,790
13,323
12,844
12,356
11,863
11,369
10,879
10,396
9,924
9,468
9,032
8,619
8,234
7,881
7,565
7,288
7,055
6,871
6,739
6,663
6,648
6,648
6,648
6,648
6,648
6,648
6,648
6,648
                        4-13

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis
35
36
0
0
6,648 1
6,648 |
       4.2.4.1 Adjusting vehicle use for future fuel prices

       The estimates of average annual miles driven by passenger cars and light trucks
reported in Table 4-4 reflect the historically low gasoline prices that prevailed at the time the
2001 NHTS was conducted. Under the assumption that people tend to drive more as the cost
of driving decreases, the higher fuel prices that are forecast for future years would be
expected to reduce average vehicle use.   For this rulemaking, the agencies updated the
estimates of average vehicle use reported in Table 4-4 using the forecasts of future fuel prices
reported in the AEO 2011 Reference  Case. This adjustment accounts for the difference
between the average retail price per gallon of fuel forecast during each calendar year over the
expected lifetimes of model year 2017-25 passenger cars and light trucks, and the average
price that prevailed when the NHTS was conducted in 2001.

       Specifically, the elasticity of annual vehicle use with respect to fuel cost per mile
corresponding to the 10 percent fuel economy rebound effect used in this analysis (i.e., an
elasticity of annual vehicle  use with respect to fuel cost per mile driven of -0.10; see Section
4.2.5) was applied to the percentage change in cost-per-mile travel between each future year's
vehicle and  the cost per mile of a vehicle that was the same age in 2001. This computation
adjusts the estimates of annual mileage by vehicle age derived from the 2001 NHTS to reflect
the effect of higher fuel prices and changes in the fuel economies of new model year vehicles
over time for each future calendar year of the expected lifetimes of model year 2017-25 cars
and light trucks.

       4.2.4.2 Ensuring consistency with growth in total vehicle use

       The estimates of annual miles driven by passenger cars and light trucks at each age
were also adjusted to reflect projected future growth in average use for vehicles of all ages.
Increases in the average number of miles cars and trucks are driven each year have been an
important source of historical growth in total car and light truck use, and are expected to be a
continued source of future growth in  total light-duty vehicle travel as well.  As an illustration
of the importance of growth in average vehicle use, the  total number of miles driven by
passenger cars increased 35 percent from 1985 through 2005, equivalent to a compound
annual growth rate of 1.5 percent.10 During that same time, however, the total number of
passenger cars registered for in the U.S. grew by only about 0.3 percent annually.k Thus
k A slight increase in the fraction of new passenger cars remaining in service beyond age 10 has accounted for a
small share of growth in the U.S. automobile fleet. The fraction of new automobiles remaining in service to
various ages was computed from R.L. Polk vehicle registration data for 1977 through 2005 by the agency's
Center for Statistical Analysis.

                                             4-14

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

growth in the average number of miles automobiles are driven each year accounted for the
remaining 1.2 percent (= 1.5 percent - 0.3 percent) annual growth in total automobile use.1

       Further, the AEO 2011  Reference Case forecasts of total car and light truck use and of
the number of cars and light trucks in use suggest that their average annual use will continue
to increase gradually from 2010 through 2035, as detailed in the following sections.™

        In order to develop reasonable estimates of future growth in the average number of
miles driven by cars and light trucks of all ages, the agencies calculated the rate of growth in
the mileage schedules necessary for total car and light truck travel to increase at the rate
forecast in the AEO 2011 Reference Case. The growth rate in average annual car and light
truck use produced by this calculation is approximately 1 percent per year through 2030, and
0.5% per year from 2031-2050.n  This growth was applied to the mileage figures reported in
Table 4-4 (after adjusting them as described previously for future fuel prices and expected
vehicle survival rates) to estimate average annual mileage during  each calendar year analyzed
and during the expected lifetimes of model year 2017-25  cars and light trucks0

       The agencies made separate adjustments to vehicle use to  account for projected
increases in future fuel prices and for continued growth in average vehicle use during each
future calendar year. Because the effects of both fuel prices and cumulative growth in
average vehicle use differ for each year, these adjustments result in different VMT schedules
for each future year.  While the adjustment for future fuel prices generally reduces average
mileage at each age from the values tabulated from the 2001 NHTS, the adjustment for
reduced fuel consumption and the expected future growth in average vehicle use increases it.
The net impact resulting from these two separate adjustments is continued growth over time
in the average number of miles that vehicles of each age are driven, although at slower rates
than those observed from 1985 - 2005. Observed aggregate VMT in recent years has actually
declined, but it is unclear if the underlying cause is general shift in behavior or a response to a
set of temporary economic conditions. The agencies' intend to consider new data on the VMT
growth estimates as  it becomes available, and the agencies request comment on this topic.
1 See supra note k below.
m The agencies note that VMT growth has slowed, and because the impact of VMT is an important element in
our benefit estimates, we will continue to monitor this trend to see whether this is a reversal in trend or
temporary slow down. See the 2009 National Household Travel Survey (http://nhts.ornl.gov/2009/pub/stt.pdf)
and National transportation Statistics
(http://www.bts.gov/publications/national transportation statistics/html/table 04 09.html)
n It was not possible to estimate separate growth rates in average annual use for cars and light trucks, because of
the significant reclassification of light truck models as passenger cars discussed previously. For the final
rulemaking, the agencies intend to review the relevant historic data and current AEO forecast and update these
values if necessary.
0 As indicated previously, a vehicle's age during any future calendar year is uniquely determined by the
difference between that calendar year and the model year when it was produced.

                                              4-15

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis
       4.2.4.3 Final VMT equation

       The following equation summarizes in mathematical form the adjustments that are
made to the values of average miles driven by vehicle age derived from the 2001 NHTS to
derive the estimates of average miles driven by vehicles of each model year during future
calendar years that are used in this analysis.
  VMTcalendaryear Xiagey = (Vy} * (1 + GRl)YSi * (1 + G2YS2 * (1 - fl * (FCPM200i,y

                 -  FCPMxo,)/FCPM20oi,y
       Where:
       Vy = Average miles driven in CY 2001 (from NHTS A analysis of 2001 NHTS data) by a vehicle of age
       y during 2001
       GR1 = Growth Rate for average miles driven by vehicles of each age from 2001 to 2030
       YS1 = Lesser of (Years since 2001) and (29).
       GR2 = Growth Rate for average miles driven by vehicles of each age from 2030 to 2050
       YS2 = Greater of (Years since 2030) and (0).
       R= Magnitude of the rebound effect, expressed as an elasticity (-0. 10)
             jj, = Fuel cost per mile of a vehicle of age y in calendar year x
       In turn, fuel cost per mile of an age y vehicle in calendar year x is determined by the
following equation, which can be extended for any number of fuels:
                FCPMCalendaryearx = ECy * EPX + GCy * GPX + DCy * DPX
       Where:
       ECy= Electricity consumption of age y vehicle (in KWh) per mile
       EPX = Electricity Price (in $ per KWh) during calendar year x
       GCy = Gasoline Consumption of age y vehicle (in gallons) per mile
       GPX = Gasoline Price (in $ per gallon) during calendar year x
       DCy = Diesel Consumption of age y vehicle (in gallons) per mile
       DPX = Diesel Price (in $ per gallon) during calendar year x

       The NHTSA and EPA models project slightly different fuel costs per mile for vehicles
affected by the proposed standards, because of the different structures of the respective
agencies' programs  and the different technologies projected by each agency's model to be
used by vehicle manufacturers to comply with each program.  Over the entire lifetimes of
those vehicles, however, the agencies' estimates of the number of miles they are expected to
be driven differ by about 3% for cars and 1% for light trucks.p For comparison, Table 4-5
p While the agencies' projections of VMT are highly similar both on average (-2-3% different depending on the
MY) and for light trucks (-1% different), the passenger car VMT schedules have differences in part due to
different treatment of vehicles reclassified from trucks to cars under the MY 2011 CAFE standards. For details,

                                               4-16

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

presents the agencies' estimates of the average number of miles driven by model year 2021
and 2025 cars and light trucks at over their estimated average lifetimes.

       Table 4-5 Survival Weighted Per-Vehicle Reference VMT used in the Agencies' analyses


EPA
NHTSA
MY 2021
Cars
204,688
212,123
Light
Trucks
242,576
245,612
MY 2025
Cars
210,898
218,404
Light
Trucks
249,713
253,122
see EPA's DRIA Chapter 4 and NHTSA's PRIA VIII.B. For the final rulemaking, the agencies intend to
harmonize their assessment of these vehicles' use patterns

                                              4-17

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis
       4.2.4.4 Comparison to other VMT Projections

       As a check on their estimates of vehicle use, the agencies compared the forecasts of
aggregate car and light truck VMT derived using the procedure described in preceding
sections to the AEO 2011 reference case forecast of light duty VMT (see Figure 4-1). The
aggregate VMT projected in this analysis is within 3% of the AEO 2011 Light Duty
projections over the time period 2017-2035.n If AEO VMT is linearly extrapolated at the
average growth rate of the period 2017-2035, the agencies' estimates remain within 3% of this
projection through 2050.  EPA's VMT comparison is shown in the chart below, but is
indicative of both agencies' analysis.q
                                              VMT
            3
                      -AEO


                      Reference VMT
AEO Ends in 2035
                                                  CY
                      Figure 4-1 Comparison of AEO and Projected VMT
qSee note p above.
                                            4-18

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis
       4.2.5  Accounting for the fuel economy rebound effect

       The rebound effect refers to the increase in vehicle use that results if an increase in
fuel efficiency lowers the cost per mile of driving, which can encourage people to drive
slightly more. Because this additional driving consumes some fuel and increases emissions, it
reduces fuel savings and increases emissions compared to those otherwise expected from the
proposed standards. Thus the magnitude of the rebound effect is  one of the determinants of
the actual fuel savings and emission reductions that are likely to result from adopting stricter
fuel economy or emissions  standards, and is thus an important parameter affecting EPA's and
NHTSA's evaluation of the proposed and alternative standards  for future model years.

       The rebound effect is measured directly by estimating the change in vehicle use, often
expressed in terms of vehicle miles traveled (VMT), with respect to changes in vehicle fuel
efficiency/ However, it is a common practice in the literature to measure the rebound effect
by estimating the change in vehicle use with respect to the fuel  cost per mile driven, which
depends on both vehicle fuel efficiency and fuel prices.8 When expressed as a positive
percentage, these two parameters give the ratio of the percentage  increase in vehicle use that
results from a percentage increase in fuel efficiency or reduction in fuel cost per mile,
respectively. For example,  a 10 percent rebound effect means that a 10 percent decrease in
fuel cost per mile is expected to result in  a 1 percent increase in VMT.

       The fuel economy rebound effect for light-duty vehicles has been the subject of a large
number of studies since the early 1980s.  Although these studies have reported a wide range
of estimates of its  exact magnitude, they generally conclude that a significant rebound effect
occurs when the cost per mile of driving decreases.' The most common approach to
estimating its magnitude has been to analyze household survey data on vehicle use, fuel
consumption, fuel prices (often obtained from external sources), and other variables that
influence travel  demand . Other studies have relied on annual aggregate U.S. data. Finally,
more recent studies have used  annual data from individual states."
r Vehicle fuel efficiency is more often measured in terms of fuel consumption (gallons per mile) rather than fuel
economy (miles per gallon) in rebound estimates.
s Fuel cost per mile is equal to the price of fuel in dollars per gallon divided by fuel economy in miles per gallon
(or multiplied by fuel consumption in gallons per mile), so this figure declines when a vehicle's fuel efficiency
increases.
t
 Some studies estimate that the long-run rebound effect is significantly larger than the immediate response to
increased fuel efficiency. Although their estimates of the adjustment period required for the rebound effect to
reach its long-run magnitude vary, this long-run effect could be more appropriate for evaluating the fuel savings
and emissions reductions resulting from stricter standards that would apply throughout the lifetime of future
model year vehicles.
u In effect, these studies treat U.S. states as a data "panel" by applying  appropriate estimation procedures to data
consisting of each year's average values of these variables for the separate states.

                                              4-19

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

       This section surveys these previous studies, summarizes recent work on the rebound
effect,12 and explains the basis for the 10 percent rebound effect EPA and NHTSA are using
in this proposed rulemaking.

       4.2.5.1 Summary of historical literature on rebound effect

       It is important to note that a majority of the studies previously conducted on the
rebound effect rely on data from the 1950-1990s. While these older studies provide valuable
information on the potential magnitude  of the rebound effect, studies that include more recent
information (e.g., data within the last decade) may provide more reliable estimates of how this
proposal will affect future driving behavior. Therefore,  the more recent studies have been
described in more detail in Section 4.2.5.2 below.

       Estimates based on aggregate U.S. vehicle travel data published by the U.S.
Department of Transportation, Federal Highway Administration, covering the period from
roughly 1950 to 1990, have found long-run rebound effects on the order of 10-30 percent.
Some of these studies are summarized in the following table.

  Table 4-6 Estimates of the Rebound Effect Using U.S. Aggregate Time-Series Data on Vehicle Travel1
AUTHOR
(YEAR)
Mayo & Mathis
(1988)
Gately (1992)
Greene (1992)
Jones (1992)
Schimek(1996)
SHORT-RUN
22%
9%
Linear 5-19%
Log -linear 13%
13%
5-7%
LONG-RUN
26%
9%
Linear 5-19%
Log-linear 13%
30%
21-29%
TIME PERIOD
1958-84
1966-88
1957-89
1957-89
1950-94
1 Source: Sorrell and Dimitropolous (2007) table 4.6.

              Table 4-7 Estimates of the Rebound Effect Using U.S. State Level Data1
AUTHOR
(YEAR)
Haughton & Sarkar
(1996)
Small and Van Dender
(2005 and 2007a)
Hymel, Small and Van
Dender (2010)
SHORT-RUN
9-16%
4.5%
2.2%
4.7%
4.8%
LONG-RUN
22%
22.2%
10.7%
24.1%
15.9%
TIME PERIOD
1973-1992
1966-2001 (at sample average)
1966-2001 (at 1997-2001 avg.)
1966-2004
1984-2004
                                            4-20

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

1 Source: Sorrell and Dimitropolous (2007) table 4.7 and the agencies' addition of recent work by Small and Van
Dender (2007a) and Hymel, Small, and Van Bender (2010) discussed in section 4.2.5.2.

       While studies using national (Table 4-6) and state level (Table 4-7) data have found
relatively consistent long-run estimates of the rebound effect, household surveys display more
variability (Table 4-8).  There are several possible explanations for this  larger variability.
One explanation is that some of these studies do not include vehicle age as an explanatory
variable, thus leading to omitted variable bias in some of their estimates.13  Another
explanation is that these studies consistently find that the magnitude of the rebound effect
differs according to the number of vehicles a household owns, and the average number of
vehicles owned per household differs among the surveys used to derive these estimates.  Still
another possibility is that it is difficult to distinguish the impact of residential density on
vehicle use from that of fuel prices, since households with higher fuel prices are more likely
to reside in urban areas.14

                Table 4-8 Estimates of the Rebound Effect Using U.S. Survey Data1
AUTHOR
(YEAR)
Goldberg (1996)
Greene, Kahn, and
Gibson (1999a)
Pickrell & Schimek
(1999)
Puller & Greening
(1999)
West (2004)
SHORT-RUN
0%


49%
87%
LONG-RUN

23%
4-34%


TIME PERIOD
CES 1984-90
EIA RTECS
1979-1994
NPTS 1995 Single year
CES 1980-90
Single year, cross-sectional
CES 1997
Single year
1 Source: Sorrell and Dimitropolous (2007) table 4.8 and the agencies' addition of Pickrell & Schimek (1999).

       It is important to note that some of these studies actually quantify the price elasticity
of gasoline demand (e.g., Puller & Greening15) or the elasticity of VMT with respect to the
price of gasoline (e.g., Pickrell & Schimek), rather than the elasticity of VMT with respect to
the fuel cost per mile of driving. The latter of these measures more closely matches the
definition of the fuel economy rebound effect.  In fact, none of the studies cited above
estimate the direct measure of the rebound effect (i.e., the increase in VMT attributed to an
increase in fuel efficiency). This topic is discussed in more detail in Section 4.2.5.2.

       Another important distinction among studies of the rebound effect is whether they
assume that the effect is constant, or varies over time in response to the absolute levels of fuel
costs, personal income, or household vehicle ownership. Most studies using aggregate annual
data for the U.S. assume a constant rebound effect, although some of these studies test
whether the effect can vary as changes in retail fuel prices or average fuel efficiency alter fuel
cost per mile driven. Many studies using household survey data estimate significantly
                                              4-21

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

different rebound effects for households owning varying numbers of vehicles, with most
finding that the rebound effect is larger among households that own more vehicles. v,w
Finally, one recent study using state-level data concludes that the rebound effect varies
directly in response to changes in personal income and the degree of urbanization of U.S.
cities, as well as fuel costs.

       In order to provide a more comprehensive overview of previous estimates of the
rebound effect, EPA  and NHTSA reviewed 22 studies of the rebound effect conducted from
1983 through 2005. The agencies then performed a detailed analysis of the 66 separate
estimates of the long-run rebound effect reported in these studies, which is summarized in
Table 4-9 below.x As the table indicates, these 66 estimates of the long-run rebound effect
range from as low as 7 percent to as high as 75 percent, with a mean value of 23 percent.
Limiting the sample to 50 estimates reported in the 17 published studies of the rebound effect
yields the same range, but a slightly higher mean estimate (24 percent).

       The type of data used and authors' assumption about whether the rebound effect varies
over time have important effects on its estimated magnitude. The 34 estimates derived from
analysis of U.S. annual time-series data produce a mean estimate of 18 percent for the long-
run rebound effect, while the mean of 23 estimates based on household survey data is
considerably larger (31 percent), and the mean of 9 estimates based on state data (25 percent)
is close to that for the entire sample. The 37 estimates assuming a constant rebound effect
produce a mean of 23 percent, identical to the mean of the 29 estimates reported in studies
that allowed the rebound effect to vary in response to fuel prices, vehicle ownership, or
household income.

                 Table 4-9 Summary Statistics for Estimates of the Rebound Effect
v Six of the household survey studies evaluated in Table 4-7 found that the rebound effects varies in relation to
the number of household vehicles. Of those six studies, four found that the rebound effect rises with higher
vehicle ownership, and two found that it declines.
w The four studies with rebound estimates that increase with higher household vehicle ownership: Greene & Hu;
Hensher et al.; Wall et al.; and West & Pickrell. The two studies with rebound estimates that decrease with
higher household vehicle ownership: Mannering and Winston; and Greene et al. (note that Greene et al. showed
decreases in the rebound effect as households went from 1 to 2 and from 2 to 3 vehicles, then a slight increase
from 3 to 4 vehicles; the rebound estimate for households with 4 vehicles was lower than for households with 2
vehicles).
x In some cases, NHTSA derived estimates of the overall rebound effect from more detailed results reported in
the studies. For example, where studies estimated different rebound effects for households owning different
numbers of vehicles but did not report an overall value, the agency computed a weighted average of the reported
values using the distribution of households among vehicle ownership categories.

                                              4-22

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis
Category of Estimates
All Estimates
Published Estimates
U.S. Time-Series Data
Household Survey Data
Pooled U.S. State Data
Constant Rebound Effect (1)
Variable Rebound Effect: (1)
Number of
Studies
22
17
7
13
2
15
10
Number of
Estimates
66
50
34
23
9
37
29
Range
Low
7%
7%
7%
9%
8%
7%
10%
High
75%
75%
45%
75%
58%
75%
45%
Distribution
Median
22%
22%
14%
31%
22%
20%
23%
Mean
23%
24%
18%
31%
25%
23%
23%
Std. Dev.
14%
14%
9%
16%
14%
16%
10%
       4.2.5.2 Summary of recent studies and analyses of the rebound effect

       More recent studies since 2007 indicate that the rebound effect has decreased over
time as incomes have generally increased and, until recently, fuel costs as a share of total
monetary travel costs have generally decreased/ One theoretical argument for why the
rebound effect should vary over time is that the responsiveness to the fuel cost of driving will
be larger when it is a larger proportion of the total cost of driving. For example, as incomes
rise, the responsiveness to the fuel cost per mile of driving will decrease if people view the
time cost  of driving - which is likely to be related to their income levels - as a larger
component of the total cost.

       Small and Van Dender combined time series data for each of the 50 States and the
District of Columbia to estimate the rebound effect, allowing the magnitude of the rebound to
vary over time.16 For the time period from 1966-2001, their study found a long-run rebound
effect of 22.2 percent, which is consistent with previously published studies. But for the most
recent five year period (1997-2001), the long-run rebound effect decreased to 10.7 percent.
Furthermore, when the authors updated their estimates with data through 2004, the long-run
rebound effect for the most recent five year period (2000-2004) dropped to 6 percent.1
Finally, when the Small methodology was used to project the future rebound effect, estimates
of the rebound effect throughout 2010-2030 were below 6 percent given a range of future
gasoline price and income projections.18
y While real gasoline prices have varied over time, fuel costs (which reflect both fuel prices and fuel efficiency)
as a share of total vehicle operating costs declined substantially from the mid-1970s until the mid-2000s when
the share increased modestly (see Greene (2010)). Note that two studies discussed in this section, Small and Van
Dender (2007) and Hymel, Small, and Van Dender (2010), find that the rebound effect is more strongly
dependant on income than fuel costs. A third study, Greene (2010), did not directly test the effect of fuel cost on
rebound, but found evidence supporting the strong effect from income. Although several studies have shown
that the rebound effect rises with household vehicle ownership (see section 4.2.5.1), which has generally
increased with income, these findings indicate that income has had a negative effect on rebound.
                                              4-23

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

       In 2010, Hymel, Small and Van Dender extended the Small and Van Dender model by
adding congestion as an endogenous variable.19 Although controlling for congestion
significantly increased their estimates of the rebound effect, Hymel, Small and Van Dender
also found that the rebound effect was declining over time.  For the time period from 1966-
2004, they estimated a long-run rebound effect of 24 percent, while for 2004 they estimated a
long-run rebound effect of 13 percent.

       Research conducted by David Greene in 2008-2009  under contract with EPA further
appears to support the theory that the magnitude of the rebound effect is declining over time
and may be as low as zero.20 Over the entire time period analyzed (1966-2007), Greene found
that fuel prices had a statistically significant impact on VMT, while fuel efficiency did not,
which is similar to Small and Van Dender's prior finding. When Small and Van Dender
tested whether the elasticity of vehicle travel with respect to the price of fuel was equal to the
elasticity with respect to the rate of fuel consumption (gallons per mile), they found that the
data could not reject this hypothesis. Therefore, Small and Van Dender estimated the rebound
effect as the elasticity of travel with respect to fuel cost per mile.  In contrast, Greene's
research showed that the hypothesis of equal elasticities for gasoline prices and fuel efficiency
can be rejected. In spite of this result, Greene also tested Small and Van Dender's
formulation which allows the elasticity of fuel cost per mile to decrease with increasing per
capita income. The results of estimation using national time series data confirmed the results
obtained by Small and Van Dender using a time series of state level data. When using
Greene's preferred functional form, the projected rebound effect is approximately 12 percent
in 2007, and drops to 10 percent in 2010, 9 percent in 2016  and 8 percent in 2030.

       Since there has been little variation  in fuel efficiency in the data over time, isolating
the impact of fuel efficiency on VMT can be difficult using  econometric analysis of historical
data.  Therefore, studies that estimate the rebound effect using time-series data often examine
the impact of gasoline prices on VMT, or the  combined impact of both gasoline prices and
fuel efficiency on VMT, as discussed above.  However, these studies may overstate the
potential impact of the rebound effect resulting from this proposal, if people are more
responsive to changes in gasoline prices than  to changes in fuel efficiency itself.  Recent work
conducted by Kenneth Gillingham included an estimate of the elasticity of VMT with respect
to the price of gasoline of -0.17, while his corresponding estimate of the elasticity of VMT
with respect to fuel economy was only 0.05.21 While this research pertains specifically to
California, this finding suggests that the common assumption that consumers respond
similarly to changes in gasoline prices and changes in fuel efficiency may overstate the
magnitude of the rebound effect.  Additional research is needed in this area, and the agencies
request comments and data on this topic.

       Another question discussed by Gillingham is whether consumers actually respond the
same way to an increase in the cost of driving compared to a decrease in the cost of driving.
There is some evidence in the literature that consumers are more responsive to an increase in
prices than to a decrease in prices. At the aggregate level, Dargay & Gately and Sentenac-
Chemin have shown that demand for transportation fuel is asymmetric.22'23 In other words,
given the same size change in prices, the response to a decrease in gasoline price is smaller

                                            4-24

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

than the response to an increase in gasoline price.  Gately has shown that the response to an
increase in oil prices can be on the order of five times larger than the response to a price
decrease.24 Furthermore, Dargay & Gately and Sentenac-Chemin find evidence that
consumers respond more to a large shock than a small, gradual change in fuel prices.  Since
these proposed standards would decrease the cost of driving gradually over time, it is possible
that the rebound effect would be much smaller than some of the historical estimates included
in the literature. Although these types of asymmetric responses have been noted at the
aggregate level on oil and gasoline consumption, little research has been done on these same
phenomena in the context of changes in vehicle fuel efficiency and the resulting rebound
effect. More research in this area is also important, and the agencies invite comment on this
aspect of the rebound effect.

       4.2.5.3 NHTSA analysis of the rebound effect

       To provide additional insight into the rebound effect for the purposes of this
rulemaking, NHTSA developed several new estimates of its magnitude.  These estimates were
developed by estimating and testing several econometric models of the relationship between
vehicle miles-traveled and factors that influence it, including household income, fuel prices,
vehicle fuel efficiency, road supply, the number of vehicles in use, vehicle prices, and other
factors.

       As the 2007  study by Small and Van Dender pointed out, it is important to  account for
the effect of fuel prices when attempting to estimate the rebound effect. Failing to control for
changes in fuel prices is likely to bias estimates of the rebound effect.  Therefore, changes in
fuel prices are taken into account in NHTSA's analysis of the rebound effect. Several
different approaches were used to estimate the fuel economy rebound effect for light duty
vehicles, many of which attempt to account for the endogenous  relationship of fuel efficiency
to fuel prices.

       The results from each of these  approaches are presented in Table 4-10 below.  The
table reports the value of the rebound effect calculated over the entire period from  1950
through 2006, as well as for the final year of that period. In addition, the table presents
forecasts of the average rebound effect between 2010 and 2030, which utilize forecasts of
personal income, fuel prices, and fuel efficiency from EIA's AEO 2011 Reference Case.

       The results of NHTSA's analysis are broadly consistent with the findings from
previous research summarized above.  The historical average long-run rebound effect is
estimated to range from 16-30 percent, and comparing these estimates to its calculated values
for 2006 (which range from 8-14 percent) gives some an indication that it is declining in
magnitude. The forecast values of the rebound effect shown in the table also suggest that this
decline is likely to continue through 2030, as they range from 4-16 percent.

             Table 4-10 Summary of NHTSA Estimates of the Rebound Effect
                                            4-25

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis
Model



Small-Van
Dender single
VMT

equation






Small-Van
Dender three-
equation
system



Single-
equation
VMT model
Single-
equation
VMT model
Single-
equation
VMT model
VMT
Measure



annual
VMT per
arlnlt
CIV1U.11'






fltltlllfll
CUlll UCll
VMT per
adult
CIV1L11L



annual
VMT per
adult
annual
VMT per
vehicle
annual
VMT per
adult
Variables Included in VMT
Equation
fuel cost per mile, per Capita
income, vehicle stock, road
miles per adult, fraction of
population that is adult,
fraction of population living
in urban areas, fraction of

population living in urban
areas with heavy rail, dummy
variables for fuel rationing,
time trend
fuel cost per mile, per Capita
income, vehicle stock, road
miles per adult, fraction of
population that is adult,
fraction of population living
in urban areas, fraction of
population living in urban
areas with heavy rail, dummy
variables for fuel rationing,
time trend

personal income, road miles
per Capita, time trend
fuel cost per mile, personal
income, road miles per Capita,
time trend
fuel cost per mile, personal
income, road miles per Capita,
dummy variables for fuel
Estimation
Technique




OLS









3SLS





OLS

OLS


OLS

Rebound Effects:
1950-
2006




33.0%









21.6%





18.4%

17.6%


34.0%

2006




15.8%









5.8%





11.7%

15.2%


20.8%

2010-
2030*




8.0%









3.4%





9.2%

15.7%


13.6%

       4.2.5.4 Basis for rebound effect used by EPA and NHTSA in this rule

       As the preceding discussion indicates, there is a wide range of estimates for both the
historical magnitude of the rebound effect and its projected future value, and there is some
evidence that the magnitude of the rebound effect appears to be declining over time.
Nevertheless, NHTSA requires a single point estimate for the rebound effect as an input to its
analysis, although a range of estimates can be used to test the sensitivity to uncertainty about
its exact magnitude.  Based on a combination of historical estimates of the rebound effect and
more recent analyses conducted by EPA and NHTSA, an estimate of 10 percent for the
rebound effect was used for this proposal (i.e., we assume a 10 percent decrease in fuel cost
                                           4-26

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

per mile from our proposed standards would result in a 1 percent increase in VMT), with a
range of 5-15 percent for use in NHTSA's sensitivity testing.

       As Table 4-6, Table 4-7, Table 4-9, and Table 4-9 indicate, the 10 percent figure is on
the low end of the range reported in previous research, and Table 4-10 shows that it is also
below most estimates of the historical and current magnitude of the rebound effect developed
by NHTSA.  However, other recent research - particularly that conducted by Hymel, Small
and Van Dender, Small and Van Dender, and Greene - reports persuasive evidence that the
magnitude of the rebound effect is likely to be declining over time, and the forecasts
developed by NHTSA and reported in Table 4-10 also suggest that this is likely to be the case.
Furthermore, for the reasons described in section 4.2.5.2, historical estimates of the rebound
effect may overstate the magnitude of a change in a small, gradual decrease in the cost of
driving due to our proposed standards.  Finally, new research by Gillingham suggests that
consumers may be more responsive to changes in gasoline prices than to changes in fuel
efficiency, and that the rebound effect that occurs when consumers purchase more efficient
vehicles as a result of a policy may be on the order of 6 percent.

       As a consequence, the agencies concluded that a value on the low end of the historical
estimates reported in Table 4-6, Table 4-7, Table 4-8, and Table 4-9 is likely to provide a
more reliable estimate of its magnitude during the future period spanned by the agencies'
analyses of the impacts of this proposal. The 10 percent estimate lies within the 10-30 percent
range of estimates for the historical rebound  effect reported in most previous research,  and at
the upper end of the 5-10 percent range of estimates for the future rebound effect reported in
the recent studies by Small and Greene.  As Table 4-10 shows, it also lies within the 3-16
percent range of forecasts of the future magnitude of the rebound effect developed by NHTSA
in its recent research. In summary, the 10 percent value was not derived from a single  point
estimate from a particular study, but instead represents a reasonable compromise between
historical estimates of the rebound effect and forecasts of its projected future value.
       4.2.6   Benefits from increased vehicle use

       The increase in vehicle use from the rebound effect provides additional benefits to
their drivers and occupants, since it is likely to take the form of more frequent trips or travel
to more distant but desirable destinations.  In either case, it provides benefits to drivers and
their passengers by improving their access to social and economic opportunities away from
home.  As evidenced by their decisions to make more frequent or longer trips when improved
fuel economy reduces their costs for driving, the benefits from this additional travel exceed
the costs drivers and passengers incur in making those more frequent or longer trips.

       The agencies' analyses estimate the economic benefits from increased rebound-effect
driving as the sum of the additional fuel costs drivers incur, plus the consumer  surplus they
                                            4-27

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

receive from the additional accessibility it provides.z The benefits that drivers and passengers
receive from additional travel are sufficient to offset these increased fuel costs, and to
generate consumer surplus - that is, benefits over and above these higher costs.  It should be
noted that the consumer surplus benefits representing a small fraction of total benefits from
increased vehicle use.

       4.2.7  Added costs from increased vehicle use

       While it provides some benefits to drivers, increased vehicle use associated with the
rebound effect also contributes to increased traffic congestion, motor vehicle accidents, and
highway noise. Depending on how the additional travel is distributed over the day and on
where it takes place, additional vehicle use can contribute to traffic congestion and delays by
increasing traffic volumes on facilities that are already heavily traveled.  These added delays
impose higher costs on drivers and other vehicle occupants in the form of increased travel
time and operating expenses. Because drivers do not take these added costs into account in
deciding when and where to travel, they must be accounted for separately as a cost of the
added driving associated with the rebound effect.

       Increased vehicle use due to the rebound effect may also increase the costs associated
with traffic accidents. Drivers may take account of the potential costs  they (and their
passengers) face from the possibility of being involved in an accident when they decide to
make additional trips. However, they probably do not consider all of the potential costs they
impose on occupants of other vehicles and on pedestrians when accidents occur. Thus any
increase in these "external" accident costs must be considered as another cost of additional
rebound-effect driving.  Like increased delay costs, any increase in these external accident
costs caused by added driving is likely to depend on the traffic conditions under which it takes
place, since accidents are more frequent in heavier traffic (although their severity may be
reduced by the slower speeds at which heavier traffic typically moves).

       Finally, added vehicle use from the rebound effect may also increase traffic noise.
Noise generated by vehicles causes inconvenience, irritation, and potentially even discomfort
to occupants of other vehicles, to pedestrians and other bystanders, and to residents or
occupants of surrounding property.  Because these effects are unlikely to be taken into
account by the drivers whose vehicles contribute to traffic noise, they represent additional
externalities associated with motor vehicle use. Although there is considerable uncertainty in
measuring their value, any increase in the economic costs of traffic noise resulting from added
vehicle use must be included together with other increased external costs from the rebound
effect.

       To estimate the increased external costs caused by added driving due to the rebound
effect, EPA and NHTSA rely on estimates of congestion, accident, and noise costs caused by
z The consumer surplus provided by added travel is estimated as one-half of the product of the decline in fuel
cost per mile and the resulting increase in the annual number of miles driven.

                                             4-28

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

automobiles and light trucks developed previously by the Federal Highway Administration.25
NHTSA employed these estimates previously in its analysis accompanying the MY 2011 final
rule, and the agencies jointly applied them in the MYs 2012-2016 rulemaking, and the
agencies continue to find them appropriate for this NPRM.  The values are intended to
measure the increases in costs (or "marginal" external costs) from added congestion, property
damages and injuries in traffic accidents, and noise levels caused by automobiles and light
trucks that are borne by persons other than their drivers and occupants.

       Updated to 2009 dollars, FHWA's "Middle" estimates for marginal congestion,
accident, and noise costs caused by automobile use amount to 5.7 cents, 2.4 cents, and 0.1
cents per vehicle-mile (for a total of 8.2 cents per mile), while those for pickup trucks and
vans are 5.1 cents, 2.7 cents, and 0.1 cents per vehicle-mile (for a total of 7.9 cents per
mile).26'aa These costs are multiplied by the mileage increases attributable to the rebound
effect to yield the estimated increases in congestion, accident, and noise externality costs
during future years.

       4.2.8   Petroleum and energy security impacts

       The proposed standards for MYs 2017-2025 will reduce fuel consumption and GHG
emissions in light-duty vehicles, which will result in improved fuel efficiency and, in turn,
help to reduce U.S. petroleum imports. A reduction of U.S. petroleum imports reduces both
financial and strategic risks caused by potential sudden disruptions  in the supply of imported
petroleum to the U.S. This reduction in the expected future economic costs associated with
these risks provides a measure of value of improved U.S. energy security resulting from lower
petroleum imports.  This section summarizes the agencies' estimates of U.S. oil import
reductions and energy security benefits of the proposed Program. Additional discussion of
this issue can be found in Section III.H.6 and Section IV of the preamble.

       4.2.8.1 Impact on U.S. petroleum imports

       In 2010, U.S. petroleum import expenditures represented 14 percent of total U.S.
imports of all goods  and services, and this figure rose to 18 percent by April of 2011.27'28 In
2010, the United  States imported 49 percent of the petroleum it consumed29, while the
transportation sector accounted for 71 percent  of total U.S. petroleum consumption30. These
figures compare to approximately 37 percent of U.S. petroleum supplied by imports and 55
percent of total petroleum consumed by the nation's transportation  sector during 1975.31
aa The Federal Highway Administration's estimates of these costs agree closely with some other recent estimates.
For example, recent published research conducted by Resources for the Future (RFF) estimates marginal
congestion and external accident costs for increased light-duty vehicle use in the U.S. to be 3.5 and 3.0 cents per
vehicle-mile in year-2002 dollars. See Ian W.H. Parry and Kenneth A. Small, "Does Britain or the U.S. Have
the Right Gasoline Tax?" Discussion Paper 02-12, Resources for the Future, 19 and Table 1 (March 2002).
Available at http://www.rff.org/rff/Documents/RFF-DP-02-12.pdf (last accessed Sept. 9, 2011).

                                             4-29

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

Requiring improved fuel economy and lower-GHG vehicle technology in the U.S. is expected
to lower U.S. petroleum imports.

       Based on analysis of historical and projected future variation in U.S. petroleum
consumption and imports, EPA and NHTSA estimate that approximately 50 percent of the
reduction in fuel consumption resulting from adopting improved GHG emission and fuel
efficiency standards is likely to be reflected in lower U.S. imports of refined fuel, while the
remaining 50 percent is expected to be reflected in reduced domestic fuel refining.bb  Of this
latter figure, 90 percent is anticipated to reduce U.S. imports of crude petroleum for use as a
refinery feedstock, while the remaining 10 percent is expected to reduce U.S. domestic
production of crude petroleum.00 Thus, on balance, each gallon of fuel saved as a
consequence of this proposed rule is anticipated to reduce total U.S. imports of petroleum by
0.95 gallons.dd

       4.2.8.2 Background on U.S. energy security

       U.S. energy security is broadly defined as protecting the U.S. economy against
circumstances that threaten significant short- and long-term increases in energy costs or
interruptions in energy supplies.  Most discussions of U.S. energy security focus  on the
economic costs of U.S. dependence on oil imports, and particularly on U.S. reliance on oil
imported from potentially unstable sources.  In addition, oil exporters have the ability to raise
the price of oil by exerting monopoly power through the mechanism of a cartel, the
Organization of Petroleum Exporting Countries (OPEC).  These factors contribute to the
vulnerability of the U.S. economy to episodic oil supply shocks and price spikes. In 2010,
total U.S. imports of crude oil,  including those from OPEC nations as well as other sources,
were $269 billion  (in 2009$)32  (see Figure 4-2).
                Figure 4-2 U.S. Expenditures on Crude Oil from 1970 through 2010e
bb Differences in forecasted annual U.S. imports of crude petroleum and refined products among the Reference,
High Oil Price, and Low Oil Price scenarios analyzed in EIA's Annual Energy Outlook 2011 range from 35-74
percent of differences in projected annual gasoline and diesel fuel consumption in the U.S. These differences
average 53 percent over the forecast period spanned by AEO 2011.
cc Differences in forecasted annual U.S. imports of crude petroleum among the Reference, High Oil Price, and
Low Oil Price scenarios analyzed in EIA's Annual Energy Outlook 2011 range from 67-104 percent of
differences in total U.S. refining of crude petroleum, and average 90 percent over the forecast period spanned by
AEO 2011.
dd This figure is calculated as 0.50 + 0.50*0.9 = 0.50 + 0.45 = 0.95.
ee Source for historical data: EIA Annual Energy Review, various editions. For recent historical and forecasted
data: EIA Annual Energy Outlook (AEO) 2011 Reference Case.

                                              4-30

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis
                          U.S. Expenditures on Crude Oil
           1970
1975
1980
1985
1990

Year
1995
2000
2005
2010
       One effect of the EPA/NHTSA joint proposal (as well as the 2012-2016 light-duty
vehicle standards and the 2014-2018 standards for medium- and heavy-duty vehicles and
engines) will be to reduce consumption of transportation fuels in the U.S. This will in turn
reduce U.S. oil imports, which lowers both financial and strategic risks associated with
potential disruptions in supply or sudden increases in the price of petroleum. For this
proposed rule, an "oil import premium" approach is utilized to estimate energy security-
related costs of importing petroleum into the U.S. Specifically, the oil import premium
measures the expected economic value of costs that are not reflected in the market price of
petroleum, and that are expected to change in response to an incremental change in the level
of U.S. oil imports.

       4.2.8.3 Methodology used to estimate U.S. energy security benefits

       In order to understand the energy security implications of reducing U.S. oil imports,
EPA has worked with Oak Ridge National Laboratory (ORNL), which has developed
approaches for evaluating the social costs and energy security implications of oil use. The
energy security estimates provided below are based upon a methodology developed in a peer-
reviewed study entitled, "The Energy Security Benefits of Reduced Oil Use, 2006-2015,"
completed in March 2008. This study is included as part of the docket for this proposal.33

      When conducting this recent analysis, ORNL considered the full cost of importing
petroleum into the U.S.  The full economic cost is defined to include two components in
                                           4-31

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

addition to the purchase price of petroleum itself.  These are: (1) the higher costs for oil
imports resulting from the effect of U.S. import demand on the world oil price and on OPEC
market power (i.e., the "demand" or "monopsony" costs); and (2) the risk of reductions in
U.S. economic output and disruption of the U.S. economy caused by sudden disruptions in the
supply of imported oil to the U.S. (i.e., macroeconomic disruption and adjustment costs).
Costs associated with maintaining a U.S. military presence to help secure stable oil supply
from potentially vulnerable regions of the world were not included in this analysis, because
attributing costs for military operations to specific missions or activities is difficult (as
discussed further below).

       For this analysis, ORNL estimated energy security premiums by incorporating the
most recent available AEO 2011 Reference Case oil price forecasts and market trends.
Energy security premiums for the years 2020, 2025, 2030, and 2035 and beyond are presented
in Table 4-11, as well as a breakdown of the components of the energy security premiums for
each of these years.ff The oil security premium rises over the future as a result of changing
factors such as the world oil price, global supply/demand balances, U.S. oil imports and
consumption, and U.S. GDP (i.e., the  size of economy at risk to oil shocks). The principal
factor is steadily rising world oil prices, but other effects interact. From 2020 to 2030, the
macroeconomic disruption and adjustment component rises by 17% by 2030 and then
stabilizes, over a period where projected average real world oil prices rise 15%.  U.S. oil
import quantities fall by 3% but total domestic oil consumption still rises by 3% despite
higher prices. Looked at another way, U.S. GDP,  the size of the economy potentially at risk
to oil shocks, grows 30%, while the value share of oil in GDP  stays high, declining only 9%
by 2030.

       The components of the energy security premiums and their values are discussed
below. Section III.H.7 of the preamble contains a detailed discussion of how the monopsony
and macroeconomic disruption/adjustment components were treated for this analysis.
              Table 4-11 Energy Security Premiums in Selected Years (2009$/Barrel)
ff AEO 2011 forecasts energy market trends and values only to 2035. The energy security premium estimates
post-2035 were assumed to be the 2035 estimate.

                                            4-32

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

2020
2025
2030
2035+
Monopsony
$11.12
($3.78 -
$21.21)
$11.26
($3.78 -
$21.48)
$10.91
($3.74 -
$20.47)
$10.11
($3.51-
$18.85)
Macroeconomic
Disruption/ Adjustment Costs
$7.10
($3.40 - $10.96)
$7.77
($3.84 -$12.32)
$8.32
($4.09 - $13.34)
$8.60
($4.41 - $13.62)
Total Mid-Point
$18.22
($9.53 -
$29.06)
$19.03
($9.93 -
$29.75)
$19.23
($10.51 -
$29.02)
$18.71
($10.30 -
$28.20)
       4.2.8.4 Monopsony Effect

       The first component of the full economic costs of importing petroleum into the U.S.
follows from the effect of U.S. import demand on the world oil price over the long-run.
Because the U.S. is a sufficiently large purchaser of foreign oil supplies, it exercises
"monopsony power" in the global petroleum market.  This means that increases in U.S.
petroleum demand can cause the world price of crude oil to rise, and conversely, that reduced
U.S. petroleum demand can reduce the world price of crude oil. Since this component of the
energy security premium is a transfer between the U.S. and oil exporting countries, it is
excluded from the benefit estimates of these proposed  rules. See more discussion of this topic
in Section 4.2.8.7.

       Thus, one benefit of reducing U.S. oil purchases, due both to  reductions in overall
energy consumption in transportation and substitution  of transportation fuels derived from
non-petroleum sources, is the potential decrease in the total dollar value of U.S. crude oil
purchases.  Because lower U.S. oil purchases reduce the price paid for each barrel, the decline
in the dollar value of U.S. petroleum purchases exceeds the savings that would result if the
global price for oil remained  unchanged. The amount  by which it does so - which reflects the
effect of U.S. monopsony power over the world oil price - represents the demand or
monopsony effect of reduced U.S. petroleum consumption.

       This demand or monopsony effect can be readily illustrated with an example. If the
U.S. imports 10 million barrels per day at a world oil price of $50 per barrel, its total daily bill
for oil imports is $500 million. If a decrease in U.S. imports to 9 million barrels per day
causes the world oil price to drop to $49 per barrel, the daily U.S. oil import bill drops to $441
million (9 million barrels times $49 per barrel). While the world oil price declines by only $1,
                                            4-33

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

the resulting decrease in oil purchases equals $59 million per day ($500 million minus $441
million). This is equivalent to an incremental savings of $59 for each barrel by which U.S. oil
imports decline ($59 million per day divided by 1 million barrels per day), or $10 more than
the newly-decreased world price of $49 per barrel.

       This additional $10 per barrel reduction in the "monopsony premium" represents the
incremental external benefits to the U.S. associated with the reduction in import payments
beyond the savings that would occur if prices remained unchanged. Of course, this additional
benefit arises only to the extent that reduction in U.S. oil imports actually affects the world oil
price. ORNL estimates this component of the energy security benefit in 2025 to be $11.26
/barrel by which U.S. petroleum imports are reduced, with a range of $3.78 - $21.48/barrel.

       4.2.8.5 Macroeconomic Disruption and Adjustment Effect

       The second component of the oil import premium, the "macroeconomic disruption and
adjustment cost premium", arises from the effect of U.S. oil imports on the expected cost of
disruptions in oil supply and resulting increases in oil prices. A sudden increase in oil prices
triggered by a disruption in world oil supplies has two main effects: (1) it increases the costs
of oil imports in the short run, further expanding the transfer of U.S. wealth to foreign
producers, and (2) it can lead to macroeconomic contraction, dislocation and losses in Gross
Domestic Product (GDP). ORNL estimates the composite estimate of these two factors that
comprise the  macroeconomic disruption/adjustment costs premium to be $7.77 /barrel in
2025, with a range of $3.84 - 12.32/barrel of imported oil reduced. This component of the
energy security premium is included in the agencies' estimate of the benefits of the proposed
rules. See more discussion of how the agencies account for the energy security benefits of the
proposed rules in Section 4.2.8.7.

       During oil price shocks, the higher price of imported oil causes  increased payments for
imports from the U.S to oil exporters. This increased claim on U.S. economic output is a loss
to the U.S. that is separate from and additional to any reduction in economic output due to the
shock. The increased oil payments during shocks are counted as a loss to the degree that the
expected price increase is not anticipated and internalized by oil consumers.

       Secondly, macroeconomic losses during price shocks reflect both losses in aggregate
economic output and "allocative" losses.  The former are reductions in  the level of output that
the U.S. economy can produce by fully utilizing its available resources, while the latter stem
from temporary dislocation and underutilization of available resources  due to the shock, such
as labor unemployment and idle plant capacity. The aggregate output effect, a reduction in
"potential" economic output, will persist as long as the price for oil remains elevated. Thus
its magnitude depends on the extent and duration of any disruption in the world supply of oil,
since these factors determine the extent of the resulting increase in prices for petroleum
products, as well as  whether and how rapidly these prices return to their pre-disruption levels.

       In addition to the aggregate  contraction, there are "allocative" or "adjustment" costs
associated with dislocations in energy markets. Because supply disruptions and resulting

                                            4-34

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

price increases occur suddenly, empirical evidence shows they also impose additional costs on
businesses and households for adjusting their use of petroleum and other productive factors
more rapidly than if the same price increase had occurred gradually. Dislocation effects
include the unemployment of workers and other resources during the time period required for
their inter-industry or interregional reallocation, as well as pauses in capital investment due to
uncertainty.  These adjustments temporarily reduce the level of economic output that can be
achieved even below the "potential" output level that would ultimately be reached once the
economy's adaptation to higher petroleum prices was complete. The additional costs imposed
on businesses and households for making these adjustments reflect their limited ability to
adjust prices, output levels, and their use of energy, labor and other inputs quickly and
smoothly in  response to rapid changes in prices for petroleum products.

       Since future disruptions in foreign oil supplies are an uncertain prospect, each of the
disruption cost components must be weighted by the probability that the supply of petroleum
to the U.S. will actually be disrupted.  Thus, the "expected value" of these costs - the product
of the probability that a supply disruption will occur and the  sum of costs from reduced
economic output and the economy's abrupt adjustment to sharply higher petroleum prices - is
the relevant  measure of their magnitude.  Further, when assessing the energy security value of
a policy to reduce oil use, only the change in  these expected  costs from potential disruptions
that results from the policy is relevant.  The expected costs of disruption may change from
lowering the normal (i.e., pre-disruption) level of domestic petroleum use and imports, from
any induced alteration in the likelihood or size of disruption, or from altering the short-run
flexibility in substituting other energy sources or inputs for petroleum use.

       In summary, the steps needed to calculate the disruption or security premium are: (1)
determine the likelihood of an oil supply disruption in the future; (2) assess the likely impacts
of a potential oil supply disruption on the world oil price; (3) assess the impact of the oil price
shock on the U.S. economy (in terms of import costs and macroeconomic losses); and (4)
determine how these costs are likely to change with the level of U.S. oil imports. The
reduction in the expected value of costs and other macroeconomic losses that results from
lower oil imports represents the macroeconomic and adjustment cost portion of the oil import
premium.

       4.2.8.6 Cost of existing U.S. energy  security policies

       The last often-identified component of the full economic costs of U.S. oil imports is
the costs to the U.S. taxpayers of existing U.S. energy security policies. The two primary
components of this cost are likely to be (1) the expenses associated with maintaining a U.S.
military presence - in part to help secure a stable oil supply - in potentially unstable regions
of the world; and (2) costs for maintaining the U.S. Strategic Petroleum Reserve (SPR).  The
SPR is the largest stockpile of government-owned emergency crude oil in the world.
Established in the aftermath of the 1973-74 oil embargo, the  SPR provides the U.S. a response
option  should price increases triggered by a disruption in commercial oil supplies threaten the
U.S. economy.  It also allows the U.S. to meet part of its International Energy Agency
obligation to maintain emergency oil stocks, and it provides a national defense fuel reserve.

                                            4-35

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

       The agencies recognize that potential national and energy security risks exist due to
the possibility of tension over oil supplies. Much of the world's oil and gas supplies are
located in countries facing social, economic, and demographic challenges, thus making them
even more vulnerable to potential local instability. For example, in 2010 just over 40  percent
of world oil supply came from OPEC nations, and this share is not expected to decline in the
AEO 2011 projections through 2030. Approximately 28 percent of global supply is from
Persian Gulf countries alone.  As another measure of concentration, of the 137
countries/principalities that export either crude oil or refined petroleum product, the top 12
have recently accounted for over 55 percent of exports.88 Eight of these countries are
members of OPEC, and  a 9th is Russia.1*  In a market where even a 1-2 percent supply loss
raises prices noticeably,  and where a 10 percent supply loss could lead to a significant price
shock, this regional concentration is of concern. Historically, the countries of the Middle East
have been the source of eight of the ten major world oil disruptions34 with the 9th originating
in Venezuela, an OPEC  member.

       Because of U.S. dependence on oil, the military could be called on to protect energy
resources through such measures as securing shipping lanes from foreign oil fields. To
maintain such military effectiveness and flexibility, the Department of Defense identified in
the Quadrennial Defense Review that it is "increasing its use of renewable energy supplies
and reducing energy demand to improve operational effectiveness, reduce greenhouse gas
emissions in support of U.S. climate change initiatives, and protect the Department from
energy price fluctuations."35 The Department of the Navy has also stated that the Navy and
Marine Corps rely far too much on petroleum, which "degrades the strategic position  of our
country and the tactical performance of our forces.  The global supply of oil is finite, it is
becoming increasingly difficult to find and exploit, and over time cost continues to rise."36

       In remarks given to the White House Energy Security Summit on April 26, 2011,
Deputy Security of Defense William J. Lynn, III noted the direct impact of energy security on
military readiness and flexibility. According to Deputy Security Lynn, "Today, energy
technology remains a critical element of our military superiority. Addressing  energy needs
must be a fundamental part of our military planning."37

       Thus, to the degree to which the proposed rules reduce reliance upon imported energy
supplies or promotes the development of technologies that can be deployed by either
consumers or the nation's defense forces, the United States could expect benefits related to
national security, reduced energy costs, and  increased energy supply. These benefits are why
President Obama has identified this program as a key component for improving energy
efficiency and putting America on a path to reducing oil imports in the Blueprint for a Secure
Energy Future. 8
gg Based on data from the CIA, combining various recent years, https://www.cia.gov/library/publications/the-
world-factbook/rankorder/2176rank.html.
14 The other three are Norway, Canada, and the EU, an exporter of product.

                                             4-36

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

       Although the agencies recognize that there clearly is a benefit to the United States
from reducing dependence on foreign oil, the agencies have been unable to calculate the
monetary benefit that the United States will receive from the improvements in national
security expected to result from this program.  In contrast, the other portion of the energy
security premium, the U.S. macroeconomic disruption and adjustment cost that arises from
U.S. petroleum imports, is included in the energy security benefits estimated for this program.
To summarize, the agencies have included only the macroeconomic disruption portion of the
energy security benefits to estimate the monetary value of the total energy security benefits of
this program. The agencies have calculated energy security in very specific terms, as the
reduction of both financial and strategic risks caused by potential sudden disruptions in  the
supply of imported petroleum to the U.S. Reducing the amount of oil imported reduces those
risks, and thus increases the nation's energy security.

       Potential savings in U.S. military costs  are excluded from the analysis performed by
ORNL, because their attribution to particular missions or activities is difficult. Most military
forces serve a broad range of security and foreign policy objectives.  Attempts to attribute
some share of U.S. military costs to oil imports are further complicated challenged by the
need to estimate how those costs vary with incremental variations in U.S. oil imports.
Similarly, while the costs for building and maintaining the SPR are more clearly related to
U.S. oil use and imports, these costs have not varied historically in response to changes  in
U.S. oil import levels. Thus while the influence of the SPR on oil price increases resulting
from a disruption of U.S oil imports is reflected in the ORNL estimate of the macroeconomic
and adjustment cost component of the oil import premium, potential changes in the cost of
maintaining the SPR associated with variation  in U.S petroleum imports are excluded.

       4.2.8.7 Total Energy Security Benefits

       Much of the literature on the energy security for the last two decades has routinely
combined the monopsony and the macroeconomic disruption components  when calculating
the total value of the energy security premium. However, in the context of using a global
social cost of carbon (SCC) value, the question arises: how should the energy security
premium be measured from a global perspective? Monopsony  benefits largely represent a
reduction in payments by consumers of petroleum products in the United States to foreign oil
producers that result from a decrease in the world oil price as the U.S. decreases its petroleum
consumption.

       Although a reduction in these payments clearly represents a benefit to the U.S. when
considered  from a domestic perspective, it represents an exactly offsetting loss to petroleum-
producing countries. Given the purely redistributive nature of this monopsony effect when
viewed from a global perspective, it is excluded in the energy security benefits calculations
for this program. In contrast, the other portion  of the energy security premium, the U.S.
macroeconomic disruption and adjustment cost that arises from U.S. petroleum imports, does
not have offsetting impacts outside of the U.S., and is thus included in the  energy security
benefits estimated for this program. Thus, the  agencies have included only the
                                            4-37

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

macroeconomic disruption portion of the energy security benefits to estimate the monetary
value of the total energy security benefits of this program.

       The energy security analysis conducted for this proposal estimates that the world price
of oil will fall modestly in response to lower U.S. demand for refined fuel.  One potential
result of this decline in the world price of oil would be an increase in the consumption of
petroleum products, particularly outside the U.S.  In addition, other fuels could be displaced
from the increasing use of oil worldwide. For example, if a decline in the world oil price
causes an increase in oil use in China, India, or another country's industrial sector, this
increase in oil consumption may displace natural gas usage. Alternatively, the increased oil
use could result in a decrease in coal used to produce electricity. An increase in the
consumption of petroleum products particularly outside the U.S., could lead to a modest
increase in emissions of greenhouse gases, criteria air pollutants, and airborne toxics from
their refining and use.  However, lower usage of, for example, displaced coal would result in a
decrease in greenhouse gas  emissions. Therefore, any assessment of the impacts on GHG
emissions from a potential increase in world oil demand would need to take into account the
impacts on all portions of the global energy sector.  The agencies' analyses have not
attempted to estimate these  effects.

       4.2.9  Air pollutant emissions

       Car and light truck use, fuel refining, and fuel  distribution and retailing also generate
emissions of certain criteria air pollutants, including carbon monoxide (CO), hydrocarbon
compounds (usually referred to as "volatile organic compounds," or VOC), nitrogen oxides
(NOX), fine particulate  matter (PM2.5), and sulfur dioxide (SOi). Emissions of most of these
pollutants are associated with the number of vehicle miles driven, rather than with the
quantity of fuel consumed.  Sulfur dioxide is an exception, as described below. While
reductions in fuel refining and distribution that result from lower fuel consumption will
reduce emissions of criteria pollutants, additional vehicle use associated with the rebound
effect will increase emissions of most of these pollutants.

       Thus the net effect of stricter fuel efficiency and GHG standards on total emissions of
each criteria pollutant depends on the relative magnitudes of reduced emissions  during fuel
refining and distribution, and increases in emissions from vehicle use. Because  the
relationship between emission rates (emissions per gallon refined of fuel or mile driven) in
fuel refining and vehicle use is different for each criteria pollutant, the net effect of increases
in fuel efficiency and GHG standards on total emissions of each pollutant differs.

       4.2.9.1 Emissions of criteria air pollutants

       For the analysis of criteria emissions over the lifetime of the model years covered by
this rule, EPA and NHTSA estimate the increases in emissions of each criteria air pollutant
from additional vehicle use by multiplying the increase in total miles driven by cars and light
trucks of each model year and age by their estimated emission rates per vehicle-mile of each
pollutant.  These emission rates differ between cars and light trucks as well as between


                                             4-38

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

gasoline and diesel vehicles, and both their values for new vehicles and the rates at which they
increase with age and accumulated mileage can vary among model years. With the exception
of 862, the agencies calculated the increase in emissions of these criteria pollutants from
added car and light truck use by multiplying the estimated increases in vehicle use during
each year over their expected lifetimes by per-mile emission rates appropriate to each vehicle
type, fuel used, model year, and age as of that future year.

       As in the MY  2012-2016 rulemaking,  the relevant emission rates were estimated by
U.S. EPA using the most recent version of the Motor Vehicle Emission Simulator
(MOVES2010a).39 The MOVES model assumes that the per-mile rates at which these
pollutants are emitted are determined by EPA regulations and the effectiveness of after-
treatment of engine exhaust emissions, and are thus unaffected by changes in car and light
truck fuel economy.  The MOVES modeling conducted for this proposal is assuming RFS2
volumes of renewable fuel volumes in both the "reference case" and the control case.11 The
emission analysis assumed a 10% ethanol fuel supply.jjAs a consequence, the downstream
impacts of required increases in fuel economy on emissions of these pollutants  from car and
light truck use are determined entirely by the increases in driving that result from the fuel
economy rebound effect.

       Emission factors in the MOVES database are expressed in the form of grams per
vehicle-hour of operation. To convert these emission factors to grams per mile, MOVES was
run for the year 2050, and was programmed to report aggregate emissions from vehicle start,
running, brake and tirewear and crankcase exhaust operations. EPA analysts selected the year
2050 in order to generate emission factors that were representative of lifetime average
emission rates for vehicles meeting the agency's Tier 2 emission standard.kk Separate
estimates were developed for each vehicle type and model year, as well as for each state and
month, in order to reflect the effects of regional and temporal variation in temperature and
other relevant variables on emissions.

       The MOVES emissions estimates were then summed to the model year  level and
divided by total distance traveled by vehicles  of that model year in order to produce per-mile
emission factors for each pollutant. The resulting emission rates represent average values
across the nation, and incorporate typical variation in temperature and other operating
conditions affecting emissions over an entire calendar year. These national  average rates also
11 The agencies assume 100 percent E10 in both the reference and control cases, which is a simplifying
assumption that is appropriate to the level of detail necessary for this proposal's analysis.
jj More discussion on fuel supply and this rule is in Preamble Section III.F
^ Because all light-duty emission rates in MOVES2010a are assumed to be invariant after MY 2010, a calendar-
year 2050 run produced a full set of emission rates that reflect anticipated deterioration in the effectiveness of
vehicles' emission control systems with increasing age and accumulated mileage for post-MY 2010 vehicles.

                                             4-39

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

reflect county-specific differences in fuel composition, as well as in the presence and type of
vehicle inspection and maintenance programs.!

       Emission rates for the criteria pollutant 862 were calculated by using average fuel
sulfur content estimates supplied by EPA, together with the simplifying assumption that the
entire sulfur content of fuel is emitted in the form of SOi. These calculations assumed that
national average gasoline and diesel sulfur levels would remain at current levels.mm
Therefore, unlike many other criteria pollutants, sulfur dioxide emissions from vehicle use
decline in proportion to the decrease in fuel consumption.

       Emissions of criteria air pollutants also occur during each phase of fuel production and
distribution, including crude oil extraction and transportation, fuel refining, and fuel storage
and transportation. The reduction in emissions during each of these phases depends on the
extent to which fuel savings result in lower imports of refined fuel, or in reduced domestic
fuel refining.  To a lesser extent, they also depend on whether reductions in domestic gasoline
refining are reflected in reduced imports of crude oil or in reduced domestic extraction of
petroleum.

       Both EPA's and NHTSA's  analyses assume that reductions in imports of refined fuel
would reduce criteria pollutant emissions during fuel storage and distribution only.
Reductions in domestic fuel refining using imported crude oil as a feedstock are assumed to
reduce emissions during fuel refining, storage, and distribution, because each of these
activities would be reduced. Finally, reduced domestic fuel refining using domestically-
produced crude oil is assumed to reduce emissions during all phases of fuel production and
distribution.™

       EPA estimated the reductions in criteria pollutant emissions from producing and
distributing fuel that would occur under alternative fuel efficiency and GHG standards using
emission rates obtained from Argonne National Laboratories' Greenhouse Gases and
Regulated Emissions in Transportation (GREET) model.40 The GREET model provides
separate estimates of air pollutant emissions that occur in four phases  of fuel production and
distribution: crude oil extraction, crude oil transportation and storage, fuel refining, and fuel
11 The national mix of fuel types includes county-level market shares of conventional and reformulated gasoline,
as well as county-level variation in sulfur content, ethanol fractions, and other fuel properties.
Inspection/maintenance programs at the county level account for detailed program design elements such as test
type, inspection frequency, and program coverage by vehicle type and age.
mm These are 30 and 15 parts per million (ppm, measured on a mass basis) for gasoline and diesel respectively,
which produces emission rates of 0.17 grams of SO2 per gallon of gasoline and 0.10 grams per gallon of diesel.
m In effect, this assumes that the distances crude oil travels to U.S. refineries are approximately the same
regardless of whether it travels from domestic oilfields or import terminals, and that the distances that gasoline
travels from refineries to retail stations are approximately the same as those from import terminals to gasoline
stations.

                                              4-40

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

distribution and storage.00 EPA modified the GREET model to change certain assumptions
about emissions during crude petroleum extraction and transportation, as well as to update its
emission rates to reflect adopted and pending EPA emission standards. EPA also
incorporated emission factors for the air toxics estimated in this analysis: benzene, 1,3-
butadiene, acetaldehyde, acrolein, and formaldehyde. The resulting emission factors are
shown in Table 4-12.
00 Emissions that occur during vehicle refueling at retail gasoline stations (primarily evaporative emissions of
volatile organic compounds, or VOCs) are already accounted for in the "tailpipe" emission factors used to
estimate the emissions generated by increased light truck use.  GREET estimates emissions in each phase of
gasoline production and distribution in mass per unit of gasoline energy content; these factors are then converted
to mass per gallon of gasoline using the average energy content of gasoline.

                                               4-41

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

Table 4-12 Emissions by Stage of Fuel Production and Distribution (grams/million Btu)
Pollutant
CO
VOC
NOx
SOx
PM2.5
Fuel Type
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Petroleum
Extraction &
Transportation1
4.908
4.908
4.908
3.035
3.035
3.035
14.91
14.91
14.91
3.926
3.926
3.926
0.645
0.645
0.645
Refinery
Energy
Use
Upstream
Emissions
0.928
0.908
0.800
0.602
0.627
0.552
3.328
3.288
2.895
4.398
4.422
3.893
1.442
1.487
1.309
Petroleum
Refining
On-Site
5.596
5.662
5.103
2.560
2.584
2.511
14.442
14.575
12.972
9.678
9.922
9.187
1.789
1.838
1.635
Petroleum
Refining2
6.525
6.571
5.903
3.162
3.211
3.063
17.771
17.863
15.866
14.076
14.344
13.080
3.231
3.325
2.943
Fuel
Transport,
Storage,
Distribution3
0.748
0.768
0.780
42.91
42.92
1.261
3.691
3.786
3.570
0.886
0.909
0.840
0.155
0.159
0.133
Air Toxics
1 ,3-Butadiene
Acetaldehyde
Acrolein
Benzene
Formaldehyde
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
Conventional Gasoline
Reformulated Gasoline
Low Sulfur Diesel
0.0017
0.0017
0.0017
0.0002
0.0002
0.0002
0.0001
0.0001
0.0001
0.0313
0.0313
0.0313
0.0050
0.0050
0.0050
0.0003
0.0003
0.0003
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0062
0.0064
0.0058
0.0010
0.0010
0.0009
0.0014
0.0014
0.0014
0.0002
0.0002
0.0002
0.0001
0.0001
0.0001
0.0264
0.0264
0.0264
0.0042
0.0042
0.0042
0.0017
0.0018
0.0017
0.0002
0.0002
0.0002
0.0001
0.0001
0.0001
0.0326
0.0328
0.0322
0.0052
0.0052
0.0051
0.0001
0.0001
0.0001
0.0046
0.0047
0.0044
0.0006
0.0006
0.0006
0.0787
0.0788
0.0015
0.0326
0.0335
0.0316
  The petroleum extraction and transport emission factors are only applied to domestic crude oil.
2 Refinery emissions factors are applied to domestic refineries, whether refining domestic or imported crude.
3 Fuel transport, storage, and distribution emission factors represent domestic emissions and are applied to all
finished fuel, whether refined domestically or internationally.

       The agency converted these emission rates from the mass per fuel energy content basis
on which GREET reports  them to mass per gallon of fuel supplied using the estimates of fuel
energy content reported by GREET. The resulting emission rates were applied to both EPA's
and NHTSA's estimates of fuel consumption under alternative fuel efficiency standards to
develop estimates of total emissions of each criteria pollutant during fuel production and
                                              4-42

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

distribution. The assumptions about the effects of changes in fuel consumption on domestic
and imported sources of fuel supply discussed above were then employed to calculate the
effects of reductions in fuel use from alternative fuel efficiency and GHG standards on
changes in domestic emissions of each criteria pollutant throughout the fuel supply and
distribution process.

       Electricity emission factors were derived from EPA's Integrated Planning Model
(IPM). EPA uses IPM to analyze the projected impact of environmental policies on the
electric power sector in the 48 contiguous states and the District of Columbia. Developed by
ICE Consulting, Inc. and used to support public and private sector clients, IPM is a multi-
regional, dynamic, deterministic linear programming model of the U.S. electric power sector.
It provides forecasts of least-cost capacity expansion, electricity dispatch, and emission
control strategies for meeting energy demand and environmental, transmission, dispatch, and
reliability constraints. IPM can be used to evaluate the cost and emissions impacts of
proposed policies to limit emissions of sulfur dioxide (SO2),  nitrogen oxides  (NOx), carbon
dioxide (CO2), and mercury (Hg) from the electric power sector.

       Among the factors that make IPM particularly well suited to model multi-emissions
control programs are (1) its ability to capture complex interactions among the electric power,
fuel, and environmental markets; (2) its detail-rich representation of emission control options
encompassing a broad array of retrofit technologies along with emission reductions through
fuel switching, changes in capacity mix and electricity dispatch strategies; and (3) its
capability to model a variety of environmental market mechanisms, such as emissions caps,
allowances, trading, and banking.

       For this analysis, EPA derived national emission factors  from an IPM version 4.10
run for the "Proposed Transport Rule .41" IPM provided national emission totals and power
generation totals for VOC, CO, NOx, PM2.5, and SO2 in 2015, 2020, 2030, 2040 and 2050.
EPA divided these sums to derive national average emission factors, and interpolated in
intermediate years.  Emissions factors for air toxics were derived from the 2002 National
Emission Inventory in conjunction with the IPM estimates. The emission factors for
electricity was adjusted upwards by six percent in order to properly capture the feedstock
gathering that occurs upstream of the powerplant.pp Feedstock gathering includes the
gathering, transporting, and preparing fuel for electricity generation. This adjustment factor is
consistent with those discussed in the MY 2012-2016 Final Rule.qq

       This analytic method makes the simplifying assumption that the electricity generation
due to this rulemaking produces emissions at the national average level.  EPA plans to further
pp The factor of 1.06 to account for GHG emissions associated with feedstock extraction, transportation, and
processing is based on Argonne National Laboratory's The Greenhouse Gases, Regulated Emissions, and Energy
Use in Transportation (GREET) Model, Version l.Sc.O, available at
http://www.transportation.anl.gov/modeling_simulation/GREET/). EPA Docket EPA-HQ-OAR-2009-0472.
qq MY 2012-2016 Final Rule, Section III.2.C

                                             4-43

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

examine the implications of this assumption in the final ralemaking, as discussed in Section
III.C.

       The agencies account for all electricity consumed by the vehicle.  For calculations of
GHG emissions from electricity generation, the total energy consumed from the battery is
divided by 0.9 to account for charging losses, and by 0.93 to account for losses during
transmission. Both values were discussed in the MYs 2012-2016 rule as well as the Interim
Joint TAR, and are unchanged from those analyses. The estimate of charging losses is based
upon engineering judgment and manufacturer Confidential Business Information (CBI). The
estimate of transmission losses is consistent, although not identical to the 8% estimate used in
GREET, as well as the 6% estimate in eGrid 2010.4 '43  The upstream emission factor is
applied to total electricity production, rather than simply power consumed at the wheel. n

       The derived set of electricity emission factors was employed by both agencies.

       Finally, EPA and NHTSA calculated the net changes in domestic emissions of each
criteria pollutant by summing the increases in its emissions projected to result from increased
vehicle use, electricity production, and the reductions in emissions anticipated to result from
lower domestic fuel refining and distribution.ss As indicated previously, the effect of adopting
improved  fuel efficiency and GHG standards on total emissions of each criteria pollutant
depends on the relative magnitudes of the resulting reduction in emissions from fuel refining
and distribution, and the increase in emissions from additional vehicle use.

       4.2.9.2 Estimated values of reducing PM-related emissions in the model year
       analysis

       The agencies' analysis of PM2.5-related benefits over the lifetime of specific model
years uses a "benefit-per-ton" method to estimate selected PMi.s-related health benefits.
These PM2.5 -related benefit-per-ton estimates provide the total monetized human health
benefits (the sum of premature mortality and premature morbidity) of reducing one ton of
directly emitted PM2.5, or one ton of a pollutant that contributes to secondarily-formed PM2.5
(such as NOx, SOx, and VOCs) from a specified source.

       Ideally, the human health benefits would be estimated based on changes in ambient
PM2.5 concentrations and population exposure, as determined by complete air quality and
exposure modeling.  However, conducting such detailed modeling was not possible within the
timeframe for this proposal.  Note that EPA will conduct full-scale photochemical air quality
n By contrast, consumer electricity costs would not include the power lost during transmission. While
consumers indirectly pay for this lost power through higher rates, this power does not appear on their electric
meter.
ss All emissions from increased vehicle use are assumed to occur within the U.S., since fuel efficiency standards
would apply only to vehicles produced for sale in the U.S.

                                             4-44

-------
                         Economic and Other Assumptions Used in the Agencies' Analysis

modeling for selected future calendar years as part of the air quality analysis it conducts for
the final rule.

       Due to analytical limitations, the estimated benefit-per-ton values do not include
comparable benefits related to reductions  in other ambient concentrations of criteria pollutants
(such as ozone, NOi or SOi) or toxic air pollutants, nor do they monetize all of the potential
health and welfare effects associated with PM2.5 or the other criteria pollutants.  As a result,
monetizing PM-related health impacts alone underestimates the benefits associated with
reductions of the suite of non-GHG pollutants that would be reduced by the proposed
standards.

       The dollar-per-ton estimates used  to monetize reductions  in emissions that contribute
to ambient concentrations of PMi.5 are provided in Table 4-13.

Table 4-13 Benefits-per-ton Values (2009$) Derived Using the American  Cancer Society Cohort Study for
                        PM-related Premature Mortality (Pope et al.,  2002)a
Yearc
All Sources'1
SOX
VOC
Stationary (Non-EGU)
Sources6
NOX
Direct PM2.5
Mobile Sources
NOX
Direct PM2.5
Estimated Using a 3 Percent Discount Rate"
2015
2020
2030
2040
$29,000
$32,000
$38,000
$44,000
$1,200
$1,300
$1,600
$1,900
$4,800
$5,300
$6,300
$7,500
$230,000
$250,000
$290,000
$340,000
$5,000
$5,500
$6,600
$7,900
$280,000
$300,000
$360,000
$430,000
Estimated Using a 7 Percent Discount Rate"
2015
2020
2030
2040
$27,000
$29,000
$34,000
$40,000
$1,100
$1,200
$1,400
$1,700
$4,400
$4,800
$5,700
$6,800
$210,000
$220,000
$260,000
$310,000
$4,600
$5,000
$6,000
$7,200
$250,000
$280,000
$330,000
$390,000
aThe benefit-per-ton estimates presented in this table are based on an estimate of premature mortality derived
from the ACS study (Pope et al., 2002). If the benefit-per-ton estimates were based on the Six-Cities study
(Laden et al., 2006), the values would be approximately 245% (nearly two-and-a-half times larger).  See below
for a description of these studies.
b The benefit-per-ton estimates presented in this table assume either a 3 percent or 7 percent discount rate in the
valuation of premature mortality to account for a twenty-year segmented cessation lag.
c Benefit-per-ton values were estimated for the years 2015, 2020, and 2030.  For intermediate years,  such as
2017 (the year the standards begin), we interpolated exponentially. For years beyond 2030 (including 2040),
EPA and NHTSA extrapolated exponentially based on the growth between 2020 and 2030.
d Note that the benefit-per-ton  value for SOx is based on the value for Stationary (Non-EGU) sources; no SOx
value was estimated for mobile sources. The benefit-per-ton value for VOCs was estimated across all sources.
e Non-EGU denotes stationary sources of emissions other than electric generating units (EGUs).

        As Table 4-13 indicates, EPA projects that the per-ton values for reducing emissions
of criteria  pollutants from both vehicle use and stationary sources such as fuel refineries and
                                                4-45

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

 storage facilities will increase over time." These projected increases reflect rising income
 levels, which are assumed to increase affected individuals' willingness to pay for reduced
 exposure to health threats from air pollution. They also reflect future population growth and
 increased life expectancy, which expands the size of the population exposed to air pollution in
 both urban and rural areas, especially in older age groups with the highest mortality risk.44'""

        For certain PM2.5-related pollutants (such as direct PM2.5 and NOx), EPA estimates
 different per-ton values for reducing mobile source emissions than for reductions in
 emissions of the same pollutant from stationary sources such as fuel refineries and storage
 facilities.  These reflect differences in the typical geographic distributions of emissions of
 each pollutant by different sources, their contributions to ambient levels of PM2.5, and
 resulting changes in population exposure. EPA and NHTSA apply these separate values to its
 estimates of changes in emissions from vehicle use and from fuel production and distribution
 to determine the net change in total economic damages from emissions of those pollutants.

        The benefit per-ton technique has been used in previous analyses, including the 2012-
 2016 Light-Duty Greenhouse Gas Rule,45 the Ozone National Ambient Air Quality Standards
 (NAAQS) RIA,46 the Portland Cement National Emissions Standards for Hazardous Air
 Pollutants (NESHAP) RIA,47 and the final NO2 NAAQS.48  Table 4-14 shows the quantified
 and monetized PM2.5-related co-benefits that are captured in these benefit-per-ton estimates,
 and also lists other effects that remain un-quantified and are thus excluded from the estimates.

                      Table 4-14 Human Health and Welfare Effects of PM2.5
  Pollutant /           Quantified and Monetized                    Un-quantified Effects
     Effect	in Primary Estimates	Changes in:	
 PM2.s             Adult premature mortality                 Subchronic bronchitis cases
                   Bronchitis: chronic and acute               Low birth weight
                   Hospital admissions: respiratory and        Pulmonary function
                   cardiovascular                           Chronic respiratory diseases other than
                   Emergency room visits for asthma          chronic bronchitis
                   Nonfatal heart attacks (myocardial          Non-asthma respiratory emergency room
                   infarction)                               visits
                   Lower and upper respiratory illness          Visibility
                   Minor restricted-activity days              Household soiling
                   Work loss days
                   Asthma exacerbations (asthmatic
                   population)
	Infant mortality	
 " As we discuss in the emissions chapter of EPA's DRIA (Chapter 4), the rule would yield emission reductions
 from upstream refining and fuel distribution due to decreased petroleum consumption.
 uu For more information about EPA's population projections, please refer to the following:
 http://www.epa.gov/air/benmap/models/BenMAPManualAppendicesAugust2010.pdf (See Appendix K)

                                               4-46

-------
                         Economic and Other Assumptions Used in the Agencies' Analysis

       Consistent with the NOi NAAQS,VV the benefits estimates utilize concentration-
response functions as reported in the epidemiology literature. Readers interested in reviewing
the complete methodology for creating the benefit-per-ton estimates used in this analysis can
consult the Technical Support Document (TSD)49 accompanying the final ozone NAAQS
RIA.  Readers can also refer to Fann et al. (2009)50 for a detailed description of the benefit-
per-ton methodology.ww

       As described above, national per-ton estimates were developed for selected
pollutant/source category combinations.  The per-ton values calculated therefore apply only to
tons reduced from those  specific pollutant/source combinations (e.g., NOi emitted from
mobile sources; direct PM emitted from stationary sources).  Our estimate of total PM2.5
benefits  is therefore based on the total direct PM2.5 and PIVb.s-related precursor emissions
(NOx, SOx, and VOCs) controlled from each source and multiplied  by the respective per-ton
values of reducing emissions from that source.

       The benefit-per-ton coefficients in this analysis were derived using modified versions
of the health impact functions used in the PM NAAQS Regulatory Impact Analysis.
Specifically, this analysis uses the benefit-per-ton estimates first applied in the Portland
Cement  NESHAP RIA, which incorporated concentration-response functions directly from
the epidemiology studies, without any adjustment for an assumed threshold. Removing the
threshold assumption is a key difference between the method used in this analysis to estimate
PM co-benefits and the methods used in analyses prior to EPA's Portland Cement
NESHAP.XX As a consequence, the benefit-per-ton estimates used in this analysis include
vv Although we summarize the main issues in this chapter, we encourage interested readers to see benefits
chapter of the NO2 NAAQS for a more detailed description of recent changes to the PM benefits presentation
and preference for the no-threshold model.
ww The values included in this report are different from those presented in the article cited above. Benefits
methods change to reflect new information and evaluation of the science. Since publication of the June 2009
article, EPA has made two significant changes to its benefits methods: (1) We no longer assume that a threshold
exists in PM-related models of health impacts, which is consistent with the findings reported in published
research; and (2) We have revised the Value of a Statistical Life to equal $6.3 million (year 2000$), up from an
estimate of $5.5 million (year 2000$) used in the June 2009 report. Please refer to the following website for
updates to the dollar-per-ton estimates: http://www.epa.gov/air/benmap/bpt.html
M Based on a review of the current body of scientific literature, EPA estimates PM-related mortality without
applying an assumed concentration threshold. EPA's Integrated Science Assessment for Particulate Matter (U.S.
Environmental Protection Agency. 2009. Integrated Science Assessment for Particulate Matter (Final Report).
EPA-600-R-08-139F. National Center for Environmental Assessment - RTF Division. December), which was
reviewed by EPA's Clean Air Scientific Advisory Committee (U.S. Environmental Protection Agency - Science
Advisory Board. 2009. Review of EPA's Integrated Science Assessment for Particulate Matter (First External
Review Draft, December 2008). EPA-COUNCIL-09-008. May.; U.S. Environmental Protection Agency  Science
Advisory Board . 2009. Consultation on EPA's Particulate Matter National Ambient Air Quality Standards:
Scope and Methods Plan for Health Risk and Exposure Assessment. EPA-COUNCIL-09-009. May), concluded
that the scientific literature consistently finds that a no-threshold log-linear model most adequately portrays the
PM-mortality concentration-response relationship while recognizing potential uncertainty about the exact shape
of the concentration-response function. This assumption is incorporated into the calculation of the PM-related
benefits-per-ton values.

                                                4-47

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

incremental benefits of reductions in PM2.5 concentrations down to their lowest modeled
levels. This approach is also consistent with EPA's analysis of the 2012-2016 Light-Duty
Vehicle Greenhouse Gas rule.

       Reductions in PM-related mortality provide the majority of the monetized value in
each benefit-per-ton estimate. Typically, the premature mortality-related effect coefficients
that underlie the benefits-per-ton estimates are drawn from epidemiology studies that examine
two large population cohorts: the American Cancer Society cohort (Pope et al., 2002)51 and
the Harvard Six Cities cohort (Laden et al., 2006).52  The concentration-response (C-R)
function developed from the extended analysis of American Cancer Society (ACS) cohort, as
reported in Pope et al. (2002), has previously been used by EPA to generate its primary
benefits estimate.  The extended analysis of the Harvard Six Cities cohort, as reported by
Laden et al (2006), was published after the completion of the Staff Paper for the 2006 PM2.5
NAAQS and has been used as an alternative estimate in the PM2.5 NAAQS RIA and PM2.5 co-
benefits estimates in analyses completed since the PM2.5 NAAQS.

       These studies provide logical choices for anchor points when presenting PM-related
benefits because, while both studies are well designed and peer-reviewed, there are strengths
and weaknesses inherent in each. Although this argues for using both studies to generate
benefits estimates, due to the analytical limitations associated with this analysis, EPA and
NHTSA have chosen to use the benefit-per-ton value derived from the ACS study. The
agencies note that benefits  would be approximately 245 percent (or nearly two-and-a-half
times) larger if the per-ton  benefit values based on the Harvard  Six Cities were used instead.

       As is the nature of benefits analyses, assumptions and methods evolve over time to
reflect the most current interpretation of the scientific and economic literature.  For a period
of time (2004-2008), EPA's Office of Air and Radiation (OAR) valued mortality risk
reductions using a value of statistical life (VSL) estimate derived from a limited analysis of
some of the available studies.  OAR arrived at a VSL using a range of $1 million to $10
million (2000$) consistent  with two meta-analyses of the wage-risk literature.

       The $1 million value represented the lower end of the interquartile range from the
Mrozek and Taylor (2002)53 meta-analysis of 33 studies. The $10 million value represented
the upper end of the interquartile range from the Viscusi and Aldy (2003)54 meta-analysis of
43 studies. The mean estimate of $5.5 million (2000$) was also consistent with the mean
VSL of $5.4 million estimated in the Kochi et al. (2006)55 meta-analysis.  However, the
Agency neither changed its official guidance on the use of VSL in rulemakings nor subjected
the interim estimate to a scientific peer-review process through  the Science Advisory Board
(SAB) or other peer-review group.

       Until updated guidance is available, EPA determined that a single, peer-reviewed
estimate applied consistently best reflects the Science Advisory Board Environmental
Economics Advisory Committee (SAB-EEAC) advice it has received. Therefore, EPA has
decided to apply the VSL that was vetted and endorsed by the SAB in the Guidelines for
Preparing Economic Analyses (U.S. EPA, 2000)56 while they continue efforts to update their

                                            4-48

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

guidance on this issue.57 This approach calculates a mean value across VSL estimates derived
from 26 labor market and contingent valuation studies published between 1974 and 1991.
The mean VSL across these studies is $6.3 million (2000$). The dollar-per-ton estimates
used in this analysis are based on this revised VSL.ZZ

       The benefit-per-ton estimates are subject to a number of assumptions and
uncertainties.
       They do not reflect local variability in population density, meteorology, exposure,
       baseline health incidence rates, or other local factors that might lead to an
       overestimate or underestimate of the actual benefits of controlling fine particulates in
       specific locations. Please refer to Chapter 6.3 of EPA's DRIA for the description of
       the agency's quantification and monetization of PM- and ozone-related health impacts
       for the proposal.
       This analysis assumes that all fine particles, regardless of their chemical composition,
       are equally potent in causing premature mortality. This is an important assumption,
       because PM2.5 produced via transported precursors emitted from stationary sources
       may differ significantly from direct PMi.5 released from engines and other industrial
       sources.  At the present time, however, no clear scientific grounds exist for supporting
       differential effects estimates by particle type.
       This analysis assumes that the health impact function for fine particles is linear within
       the range of ambient concentrations under consideration. Thus, the estimates include
       health benefits from reducing fine particles in areas with varied initial concentrations
       of PM2.5, including both regions that are in attainment with fine particle standard and
       those that do not meet the standard, down to the lowest modeled concentrations.
       There are several health benefits categories that EPA and NHTSA were unable to
       quantify due to limitations associated with using benefits-per-ton estimates, several of
       which could be substantial.  Because NOx and VOC emissions are also precursors to
       ozone, changes in NOx and VOC would also impact ozone formation and the health
       effects associated with ozone exposure. Benefits-per-ton estimates for ozone do not
       exist due to issues associated with the complexity of the atmospheric air chemistry and
       nonlinearities associated with ozone formation.  The PM-related benefits-per-ton
       estimates also do not include any human welfare or ecological benefits. Please refer to
       Chapter 6.3 of EPA's PRIA for a description of the unquantified co-pollutant benefits
       associated with this rulemaking.
w In the update of the Economic Guidelines (U.S. EPA, 2011), EPA retained the VSL endorsed by the SAB with
the understanding that further updates to the mortality risk valuation guidance would be forthcoming in the near
future. The update of the Economic Guidelines is available on the Internet at
http://yosemite.epa.gov/ee/epa/eed.nsf/pages/Guidelines.html/$file/Guidelines.pdf.
zz This value differs from the Department of Transportation's most recent estimate of the value of preventing
transportation-related fatalities, which is $6.1 million when expressed in today's (2011) dollars.

                                              4-49

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

       As mentioned above, emissions changes and benefits-per-ton estimates alone are not a
good indication of local or regional air quality and health impacts, as the localized impacts
associated with the rulemaking may vary significantly. Additionally, the atmospheric
chemistry related to ambient concentrations of PM2.5, ozone and air toxics is very complex.
Full-scale photochemical modeling is therefore necessary to provide the needed spatial and
temporal detail to more completely and accurately estimate the changes in ambient levels of
these pollutants and their associated health and welfare impacts.  For the final rule, EPA will
conduct a national-scale air quality modeling analysis in 2030 to analyze the impacts of the
standards on PM2.5, ozone, and selected air toxics.

       4.2.10  Reductions in emissions of greenhouse gases

       Emissions of carbon dioxide and other greenhouse gases (GHGs) occur throughout the
process of producing and distributing transportation fuels, as well as from fuel combustion
itself. By increasing fuel efficiency and thus reducing the volume of fuel consumed by
passenger cars and light trucks, the proposed standards will reduce GHG emissions generated
by fuel use, as well as throughout the fuel supply cycle. Lowering these emissions is likely to
slow the projected pace and reduce the ultimate extent of future changes in the global climate,
thus reducing future economic damages that changes in the global climate are otherwise
expected to cause. Further, by reducing the probability that climate changes with potentially
catastrophic economic or environmental impacts will occur, lowering GHG emissions may
also result in economic benefits that exceed the resulting reduction in the expected future
economic costs caused by gradual changes in the earth's climatic systems. Quantifying and
monetizing benefits from reducing GHG emissions is thus an important step in estimating the
total economic benefits likely to result from establishing improved fuel efficiency and GHG
standards.

       4.2.10.1      Estimating  reductions in  GHG emissions

       NHTSA estimates emissions of carbon dioxide (COi) from passenger car and light
truck use by multiplying the number of gallons of each type of fuel (gasoline and diesel) they
are projected to consume with each alternative CAFE standard in effect by the quantity or
mass of CO2 emissions released per gallon of fuel consumed. EPA directly calculates COi
emissions from the projected COi emissions of each  vehicle. This calculation assumes that
the entire carbon content of each fuel is ultimately converted to CO2 emissions during the
combustion process. The weighted average COi content of certification gasoline is estimated
to be 8,887 grams per gallon, while that of diesel fuel is estimated to be approximately 10,200
grams per gallon. For details, please see EPA's DRIA and NHTSA's PRIA.

       Although carbon dioxide emissions account for nearly 95 percent of total GHG
emissions that result from fuel combustion during vehicle use, emissions of other GHGs are
potentially significant as well because of their higher "potency" as GHGs than that of COi
itself. EPA and NHTSA estimated the increases in emissions of methane (CELO and nitrous
oxide (NiO) from additional vehicle use by multiplying the increase in total miles driven by
cars and light trucks of each model year and age by emission rates per vehicle-mile for these


                                           4-50

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

GHGs. These emission rates, which differ between cars and light tracks as well as between
gasoline and diesel vehicles, were estimated by EPA using MOVES 2010a.

       The MOVES model assumes that the per-mile rates at which cars and light tracks emit
these GHGs are determined by the efficiency of fuel combustion during engine operation and
chemical reactions that occur during catalytic after-treatment of engine exhaust, and are thus
independent of vehicles' fuel consumption rates. Thus MOVES emission factors for these
GHGs, which are expressed per mile of vehicle travel, are assumed to be unaffected by
changes in fuel economy.

       Much like criteria pollutants, emissions of GHGs occur during each phase of fuel
production and distribution, including crude oil extraction and transportation, fuel refining,
and fuel storage and transportation. Emissions of GHGs also occur in generating electricity,
which the agencies' analysis anticipates will account for an increased share of energy
consumption in the model years that would be subject to the proposed rales. The agencies'
analyses assume that reductions in fuel consumption would reduce global GHG emissions
during all four phases of fuel production and distribution.aaa Unlike criteria pollutants, the
agencies report both domestic and international reductions in GHG emissions.  EPA derived
GHG emission rates corresponding to  producing and distributing fuel from Argonne National
Laboratories' Greenhouse Gases and Regulated Emissions in Transportation (GREET)
model.57bbb

       As with the non-GHGs, EPA derived national CO2 emission factors  from the IPM
version 4.10 run  for the "Proposed Transport Rule.58" This case features almost no change in
the CO2 emission factors from powerplants between 2012 and 2050 (approximately 1%).  A
similar methodology was used for CO2 as with the criteria pollutants.  N2O and CH4
emissions, which are not readily available from IPM, were calculated from eGrid 2007, and
scaled according to the growth in CO2. These non-CO2 emissions are a small fraction of the
emissions from power plants, as  shown below in Table 4-15.

                Table 4-15 Calender Year 2025 GHG Emission Rates for Electricity
POLLUTANT
CO2
CH4
N20
CO2eq
CO2eq adjusted for
CY 2025
ELECTRICITY (g/kWh)
539
0.01
0.01
541
574
aaa The four stages are crude oil extraction, crude oil transportation and storage, fuel refining, and fuel
distribution and storage
    is version of the model was modified, and is discussed in section 4.2.9.1

                                            4-51

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis
                    feedstock gathering
       Increases in emissions of non-COi GHGs are converted to equivalent increases in COi
emissions using estimates of the Global Warming Potential (GWP) of methane (CELO, nitrous
oxide (NiO), and hydrofluorocarbons (HFC-134a).ccc These GWPs are one way of
accounting for the higher radiative forcing capacity and differing lifetimes of methane and
nitrous oxide when they are released into the earth's atmosphere, measured relative to that of
COi itself. Because these gases differ in atmospheric lifetimes, their relative damages are not
constant over time. Impacts other than temperature change also vary across gases in ways that
are not captured by GWP. For instance, COi emissions, unlike methane and other greenhouse
gases, contribute to ocean acidification. Methane contributes to health and ecosystem effects
arising from increases in tropospheric ozone, while damages from methane emissions are not
offset by the positive effect of CO2 fertilization. Noting these caveats, the COi equivalents of
increases in emissions of these gases are then added to the increases in emissions of CC>2 itself
to summarize the effect of the total increase in CCVequivalent GHG emissions from vehicle
use.  However, only the  COi emissions were monetized for purposes of valuing benefits of the
rule, as discussed in the next section.
       4.2.10.2       Economic benefits from reducing GHG emissions

       NHTSA and EPA have taken the economic benefits of reducing COi emissions (or
avoiding damages from increased emissions) into account in developing the proposed GHG
and CAFE standards and in assessing the economic benefits of the proposed standards.
Specifically, NHTSA and EPA have assigned dollar values to reductions in carbon dioxide
(CCh) emissions using estimates of the global "social cost of carbon" (SCC). The SCC is an
estimate of the monetized damages associated with an incremental increase in carbon
emissions in a given year.  It is intended to include (but is not limited to) changes in net
agricultural productivity, human health, property damages from increased flood risk, and the
value of ecosystem services due to climate change.  The SCC is expressed in constant dollars
per additional metric ton of CO2 emissions occurring during a specific year, and is higher for
ccc As in the MYs 2012-2016 rules and in the recent MD and HD rulemakings, the global warming potentials
(GWP) used in this proposal are consistent with the 100-year time frame values in the 2007 Intergovernmental
Panel on Climate Change (IPCC) Fourth Assessment Report (AR4). At this time, the 1996IPCC Second
Assessment Report (SAR) 100-year GWP values are used in the official U.S. greenhouse gas inventory
submission to the United Nations Framework Convention on Climate Change (per the reporting requirements
under that international convention, which were last updated in 2006). N2O has a 100-year GWP of 298 and
CH4 has a 100-year GWP of 25 according to the 2007 IPCC AR4.
                                            4-52

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

more distant future years because the damages caused by an additional ton of emissions
increase with larger concentrations of CC>2 in the earth's atmosphere.

       The estimates used in this analysis were developed through an interagency process
that included EPA, DOT/NHTSA, and other executive branch entities, and concluded in
February 2010.  The interagency group focused on global SCC values because emissions of
COi involve a global externality: greenhouse gases contribute to damages around the world
wherever they are emitted. Consequently, to address the global nature of the climate change
problem, the SCC must incorporate the full (global) damages caused by GHG emissions.
Furthermore, climate change occurs  over very long time horizons and represents a problem
that the United States cannot solve independently.  The February 2010 SCC Technical
Support Document (SCC TSD) provides a complete discussion of the  SCC estimates and the
methods used to develop them.5

       We first used these SCC estimates in the benefits analysis for the final joint EPA/DOT
Rulemaking to establish 2012-2016 Light-Duty Vehicle Greenhouse Gas Emission Standards
and Corporate Average Fuel Economy Standards; see the rule's preamble for discussion about
application of the SCC (75 FR 25324; 5/7/10). We have continued to use these estimates in
other rulemaking analyses, including the Greenhouse Gas Emission Standards and Fuel
Efficiency Standards for Medium- and Heavy-Duty Engines and Vehicles (76 FR 57106;
9/15/11). Finally, see also preamble Section III.H.5-6, Section IV.C.3.1, EPA RIA Chapter
7.2, and NHTSA RIA VIII.C for discussion about the application of new SCC estimates to
this proposed rule. The SCC estimates corresponding to assumed values of the discount rate
are shown below in Table 4-16.

                 Table 4-16 Social Cost of of CO2,2017 - 2050a (in 2009 dollars)
Year
2017
2020
2025
2030
2035
2040
2045
2050
Discount Rate and Statistic
5% Average
$6.36
$7.01
$8.53
$10.05
$11.57
$13.09
$14.63
$16.18
3% Average
$25.59
$27.10
$30.43
$33.75
$37.08
$40.40
$43. 34
$46.27
2.5% Average
$40.94
$42.98
$47.28
$51.58
$55.88
$60.19
$63.59
$66.99
3%
95th percentile
$78.28
$83.17
$93.11
$103.06
$113.00
$122.95
$131.66
$140.37
                  a The SCC values apply to emissions occurring during each year
            shown (in 2009 dollars), and represent the present value of future damages as
            of the year shown.

       As Table 4-16 shows, the SCC estimates selected by the interagency group for use in
regulatory analyses range from somewhat more than $6 to about $78 (in 2009 dollars) for
emissions occurring in the year 2017. The first three estimates are based on the average SCC
                                           4-53

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

estimated using different models and reflect discount rates of 5, 3, and 2.5 percent,
respectively. The fourth value is included to represent the possibility of higher-than-expected
impacts from accumulation of GHGs in the earth's atmosphere, and the consequently larger
economic damages.  For this purpose, the interagency group elected to use the SCC value for
the 95th percentile at a 3 percent discount rate.

       The value that the interagency group centered its attention on is the average SCC
estimate at a 3 percent discount rate, or more than $25 per metric ton in 2017. To capture the
uncertainties involved in regulatory impact analysis, however, the group emphasized the
importance of considering the full range of estimated SCC values. As the table shows, the
SCC estimates rise over time because future emissions are expected to produce larger
incremental damages as physical and economic systems become more stressed in response to
greater climatic change; for example, the central value increases from over $25 per ton of COi
in 2017 to almost $34 per ton of CO2 by 2030.

       Reductions in COi emissions that are projected to result from lower fuel consumption,
refining, and distribution during each future year are multiplied by the appropriate SCC
estimates for that year, to determine the range of total economic benefits from reduced
emissions during that year. For internal consistency,  these annual benefits are discounted
back to net present value terms using a discount rate that is consistent with that used to
develop each SCC estimate.

       Finally, the SCC estimates presented in this analysis exclude the value of changes in
non-COi GHG emissions expected under this program as discussed above in section 4.2.10.1.
The interagency group did not estimate the social costs of non-CO2 GHG emissions when it
developed the current social cost of CO2 values. Although we have not monetized changes in
non-COi GHGs,  the value of any increases or reductions should not be interpreted as zero.
       4.2.11 The Benefits due to reduced refueling time

       No direct estimates of the value of extended vehicle range are readily available, so the
agencies instead calculate the reduction in the required annual number of refueling cycles due
to improved fuel economy, and assess the economic value of the resulting benefits.  Chief
among these benefits is the time that owners save by spending less time both in search of
fueling stations and in the act of pumping and paying for fuel.

       The agencies calculate the economic value of those time savings by applying DOT-
recommended values of travel time savings to our estimates of how much time is saved.60 The
value of travel time depends on average hourly valuations of personal and business time,
which are functions of total hourly compensation costs to employers. The total hourly
                                           4-54

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

compensation cost to employers, inclusive of benefits, in 2009$ is $29.37.ddd  Table 4-17
demonstrates the agencies' approach to estimating the value of travel time ($/hour) for both
urban and rural (intercity) driving. This approach relies on the use of DOT-recommended
weights that assign a lesser valuation to personal travel time than to business travel time, as
well as weights that adjust for the distribution between personal and business travel.
ddd Total hourly employer compensation costs for 2009 (average of quarterly observations). See
http://www.bls.gov/ect/. NHTSA previously a value of $25.50 for the total hourly compensation cost (see, e.g.,
75 FR at 25588, fn. 619) during 2008 expressed in 2007$. This earlier figure is deprecated by the availability of
more current economic data.

                                               4-55

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis
   Table 4-17 Estimating the Value of Travel Time For Urban and Rural (Intercity) Travel ($/hour)e
Urban Travel

Wage Rate ($/hour)
DOT-Recommended Value of Travel Time
Savings, as % of Wage Rate
Hourly Valuation (=Wage Rate * DOT-
Recommended Value)
% of Total Urban Travel
Hourly Valuation (Adjusted for % of Total
Urban Travel)
Personal travel
$29.37
50%
$14.69
94.4%
$13.86
Business Travel
$29.37
100%
$29.37
5.6%
$1.64
Total

_
_
100%
15.50
Rural (Intercity) Travel

Wage Rate ($/hour)
DOT-Recommended Value of Travel Time
Savings, as % of Wage Rate
Hourly Valuation (=Wage Rate * DOT-
Recommended Value)
% of Total Rural Travel
Hourly Valuation (Adjusted for % of Total
Rural Travel)
Personal travel
$29.37
70%
$20.56
87.0%
$17.89
Business Travel
$29.37
100%
$29.37
13.0%
$3.82
Total

_
_
100%
21.71
       The estimates of the hourly value of urban and rural travel time ($15.50 and $21.71,
respectively) shown in Table 4-17 must be adjusted to account for the nationwide ratio of
urban to rural driving. By applying this adjustment (as shown in Table 4-18), an overall
estimate of the hourly value of travel time - independent of urban or rural status - may be
produced.  Note that up to this point, all calculations discussed assume only one adult
eee Time spent on personal travel during rural (intercity) travel is valued at a greater rate than that of urban travel.
There are several reasons behind the divergence in these values: 1) time is scarcer on a long trip; 2) a long trip
involves complementary expenditures on travel, lodging, food, and entertainment, since time at the destination is
worth such high costs.
                                               4-56

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

occupant per vehicle. To fully estimate the average value of vehicle travel time, the agency
must account for the presence of additional adult passengers during refueling trips. The
agencies apply such an adjustment as shown in Table 4-18; this adjustment is performed
separately for passenger cars and for light trucks, yielding occupancy-adjusted valuations of
vehicle travel time during refueling trips for each fleet.
           Table 4-18 Estimating the Value of Travel Time for Light-Duty Vehicles ($/hour)

Urban Travel
Rural Travel
Total


Average Vehicle Occupancy
During Refueling Trips (persons)888
Weighted Value of Travel
Time ($/hour)
Occupancy-Adjusted Value
of Vehicle Travel Time During
Refueling Trips ($/hour)
Unweighted Value
of Travel Time
($/hour)
$15.50
$21.71
—

Passenger Cars
1.21
$17.58
$21.27
Weight (% of
Total Miles
Driven)fff
66.5%
33.5%
100.0%

Light Trucks
1.23
$17.58
$21.62
Weighted Value
of Travel Time
($/hour)
$10.31
$7.27
$17.58


       NHTSA estimated the amount of refueling time saved using (preliminary) survey data
gathered as part of our 2010-2011 National Automotive Sampling System's Tire Pressure
Monitoring System (TPMS) study.1*11 The study was conducted at fueling stations
nationwide, and researchers made observations regarding a variety of characteristics of
thousands of individual fueling station visits from August, 2010 through April, 201 l.m
fff Weights used for urban vs. rural travel are computed using cumulative 2009 estimates of urban vs. rural miles
driven provided by the Federal Highway Administration. Available at
http://www.fhwa.dot.gov/policvinformation/traveLnionitoring/tvt.cfm (last accessed 07/18/2011).
gg8 National Automotive Sampling System 2010-2011 Tire Pressure Monitoring System (TPMS) study. See next
page for further background on the TPMS study. TPMS data are preliminary at this time and rates are subject to
change pending availability of finalized TPMS data. Average occupancy rates shown here are specific to
refueling trips, and do not include children under 16 years of age.
1411 TPMS data are preliminary and not yet published.  Estimates derived from TPMS data are therefore
preliminary and subject to change. Observational and interview data are from distinct subsamples, each
consisting of approximately 7,000 vehicles.  For more information on the National Automotive Sampling System
and to access TPMS data when they are made available, see http://www.nhtsa.gov/NASS.
Ui The data collection period for the TPMS study ranged from 08/10/2010 to 04/15/2011.
                                                4-57

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

Among these characteristics of fueling station visits is the total amount of time spent pumping
and paying for fuel.  From a separate sample (also part of the TPMS study), researchers
conducted interviews at the pump to gauge the distances that drivers travel in transit to and
from fueling stations, how long that transit takes, and how many gallons of fuel are being
purchased.

       For purposes of this analysis of the proposed standards, the NHTSA focused on the
interview-based responses in which respondents indicated the primary reason for the refueling
trip was due to a low reading on the gas gauge.jjj  This restriction was imposed so as to
exclude distortionary effects of those who refuel on a fixed (e.g., weekly) schedule and may
be unlikely to alter refueling patterns as a result of increased driving range. The relevant
TPMS survey data on average refueling trip characteristics are presented below in Table 4-19.
        Table 4-19 Average Refueling Trip Characteristics for Passenger Cars and Light Trucks

Passenger Cars
Light Trucks
Gallons of
Fuel
Purchased
9.8
13.0
Round-Trip
Distance
to/from
Fueling
Station
(miles)
0.97
1.08
Round-Trip
Time to/from
Fueling
Station
(minutes)
2.28
2.53
Time to
Fill and
Pay
(minutes)
4.10
4.30
Total
Time
(minutes)
6.38
6.83
       As an illustration of how we estimate the value of extended refueling range, assume a
small light truck model has an average fuel tank size of approximately 20 gallons, and a
baseline actual on-road fuel economy of 24 mpg.  TPMS survey data indicate that drivers who
indicated the primary reason for their refueling trips was a low reading on the gas gauge
typically refuel when their tanks are 35 percent full (i.e.,  13.0 gallons  as shown in Table 4-19,
with 7.0 gallons in reserve). By this measure, a typical driver would have an effective driving
range of 312 miles (= 13.0 gallons x 24 mpg) before he or she is likely to refuel.  Increasing
this model's actual on-road fuel economy from 24 to 25 mpg would therefore extend its
effective driving range to 325 miles (= 13.0 gallons x 25 mpg). Assuming that the truck is
driven 12,000 miles/year,kkk this 1 mpg improvement in actual on-road fuel economy reduces
JJJ Approximately 60 percent of respondents indicated "gas tank low" as the primary reason for the refueling trip
in question.
]a± Source of annual vehicle mileage: U.S. Department of Transportation, Federal Highway Administration, 2009
National Household Travel Survey (NHTS). See http://nhts.ornl.gov/2009/pub/stt.pdf (table 22, p.48). 12,000
miles/year is an approximation of a light duty vehicle's annual mileage during its initial decade of use (the
                                              4-58

-------
                        Economic and Other Assumptions Used in the Agencies' Analysis

the expected number of refueling trips per year from 38.5 (= 12,000 miles per year 7312 miles
per refueling) to 36.9 (= 12,000 miles per year / 325 miles per refueling), or 1.6 refuelings per
year.  If a typical fueling cycle for a light truck requires a total of 6.83 minutes, then the
annual value of time saved due to that 1 mpg improvement would amount to $3.94 (=
(6.83/60) x $21.62 x 1.6).

       In the analysis, we repeat this calculation for each future calendar year that light-duty
vehicles  of each model year affected by the alternative standards considered in this rule would
remain in service. The resulting cumulative lifetime valuations of time savings account for
both the  reduction over time in the  number of vehicles of a given model year that remain in
service and the reduction in the number of miles (VMT) driven by those that stay in service.
We also  adjust the value of time savings that will occur in future years both to account for
expected annual growth in real wages and to apply a discount rate to determine the net present
value of  time saved.111  A final adjustment is made to account for evidence from the TPMS
study which suggests that 40 percent of refueling trips are for reasons other than a low reading
on the gas gauge; it is therefore assumed that only 60 percent of the theoretical refueling time
savings will be realized, as we assume that owners who refuel on a fixed schedule will
continue to do.The assumption that the 40 percent of refueling trips that occur for reasons
other than a low reading on the gas gauge will not realize a refueling time savings may be  a
conservative assumption. Results are calculated separately for a given model year's fleet of
passenger cars and that year's fleet of light trucks.  Valuations of both fleets' benefits are then
summed to determine the benefit across all light-duty vehicles.

       Since a reduction in the expected number of annual refueling trips leads to a decrease
in miles driven to and from fueling stations, we can also calculate the value of consumers'
fuel savings associated with this decrease.  As shown in Table 4-19, the typical incremental
round-trip mileage per refueling cycle is 1.08 miles for light trucks and 0.97 miles for
passenger cars.  Going back to the earlier example of a light truck model, a decrease of 1.6 in
the number of refuelings per year leads to a reduction of 1.73 miles driven per year (= 1.6
refuelings x 1.08 miles driven per refueling).  Again, if this model's actual on-road fuel
economy was 24 mpg, the reduction in miles driven yields an annual savings of
approximately 0.07 gallons of fuel  (= 1.73 miles / 24 mpg), which at $3.44/gallonmmm results
in a savings of $0.25 per year to the owner.  Note that this example is illustrative only of the
approach the agencies uses to quantify this benefit; in practice, the value of this benefit is
period in which the bulk of benefits are realized). The VOLPE model estimates VMT by model year and vehicle
age, taking into account the rebound effect, secular growth rates in VMT, and fleet survivability; these
complexities are omitted in the above example for simplicity.
111A 1.1 percent annual rate of growth in real wages is used to adjust the value of travel time per vehicle ($/hour)
for future years for which a given model is expected to remain in service. This rate is supported by a BLS
analysis of growth in real wages from 2000 - 2009.  See http://www.bls.gov/opub/ted/2011/ted 20110224.htm.
mmm Estimate of $3.44/gallon is in 2009$.  This figure is an average of forecasted cost per gallon (including
taxes, as individual consumers consider reduced tax expenditures to be savings) for motor gasoline for years
2017 to 2027. Source of price forecasts: U.S. Energy Information Administration, Annual Energy Outlook 2011
(April 2011 release). See http://www.eia.gov/forecasts/aeo/source_oil.cfm.

                                              4-59

-------
                       Economic and Other Assumptions Used in the Agencies' Analysis

modeled using fuel price forecasts for each year the given fleet will remain in service, and
unlike the above example excludes fuel taxes from the computation of the total social benefit,
as taxes are transfer payments.

       The annual savings to each consumer shown in the above example may seem like a
small amount, but the reader should recognize that the valuation of the cumulative lifetime
benefit of this savings to owners is determined separately for passenger car and light truck
fleets and then aggregated to show the net benefit across all light-duty vehicles - which is
much more significant at the macro level.  Calculations of benefits realized in future years are
adjusted for expected real growth in the price of gasoline, for the decline in the number of
vehicles of a given model year that remain in service as they age, for the decrease in the
number of miles (VMT) driven by those that stay in service, and for the percentage of
refueling trips that occur for reasons other than a low reading on the gas gauge; a discount rate
is also applied in the valuation of future benefits.  The agencies considered using this direct
estimation approach to quantify the value of this  benefit by model year, however the value of
this benefit is implicitly captured in the separate measure of overall valuation of fuel savings,
and therefore direct estimates of this benefit are not added to net benefits calculations.

       We note that there are other benefits resulting from the reduction in miles driven to
and from fueling stations, such  as a reduction in greenhouse gas emissions - CC>2 in
particular, reductions in evaporative emissions from refuelings, and reduced wear on vehicles.
However, estimates of the values of these benefits indicate that both are extremely minor in
the context of the overall valuation of benefits associated with gains in vehicle driving range,
so quantitative valuations of these additional benefits are not included within this analysis.

       It is important to note that manufacturers' decisions regarding vehicles' fuel tank sizes
are integral to the realized value of this benefit.  In MY 2010, fuel tanks were sized such that
average driving range of passenger cars was 410  miles and of light trucks was 430 miles. At
vehicle redesign, manufacturers typically redesign fuel tanks based on changes in vehicle
design and the allowable space  for the fuel tank.  At redesign, manufacturers  consider driving
range, cargo and passenger space (utility), mass targets, safety, and other factors. As fuel
economy improves, manufacturers may opt at the time of vehicle redesign to downsize
vehicles' fuel tanks as a mass-reduction strategy  and to maintain a target maximum range
consistent with  previous models.  Downsizing the fuel tank offers the potential for moderate
mass reductionnnn at a small cost savings.  It is also possible for manufacturers to reduce the
effective size of their fuel tanks by changing the length of the fill tube, which does not require
redesign of the tank itself.  In determining the maximum feasible amount of mass reduction
and the cost curves developed for mass reduction, the agencies used an assumption that fuel
tanks would be  resized to maintain range.  If a manufacturer did not downsize the fuel tank to
mn For example, for a vehicle with a 15 gallon fuel tank and a 400 mile range, increasing fuel economy by 50%
and downsizing the fuel tank to maintain range would enable a mass reduction of approximately 32 pounds
based on the reduction in the amount of fuel alone. If the fuel tank was not downsized, the range of the vehicle
would increase to 600 miles.

                                             4-60

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

maintain range, it could incur higher costs for compliance than the agencies projections
because the manufacturer may need to employ other higher cost technologies to achieve the
small incremental change in fuel economy and GHG improvements attributed to the reduction
in the mass of the fuel tank. If manufacturers elect to reduce fuel tank size in response to
improved fuel economy to maintain range, the value of the refueling time savings benefit will
be reduced because the number of trips to the gasoline station would not be reduced as much
as estimated.  Reductions to fuel tank size will not eliminate the value of the refueling time
savings benefit, however, unless they are performed annually to maintain a constant range.
Also, the reduced time for refueling and reduction in evaporative emissions would be
unchanged. The  agencies believe that annual refreshes of fuel tank size during the years in-
between model redesigns are unlikely; therefore, while downsizing fuel tanks would decrease
the realized value of the refueling time savings benefit, it would not eliminate it, assuming
that fuel economy rises in those interim years.

       The agencies considered past trends  to evaluate potential outcomes with regard to the
refueling time savings benefit. Fuel tank sizes by broad vehicle class has been nearly flat over
the past 20 years, with average light truck fuel tank volume slightly decreasing in recent
years, and average passenger car fuel tank volume slightly increasing in size in recent years.
These changes, less a gallon change in average fuel tanks size over twenty years, are slight.

       Tank sizes for popular passenger cars and light trucks in recent model years typically
allow for maximum driving ranges of between 300 and 500 miles. In MY 2010, the average
driving range for light trucks was approximately 430 miles, while the average driving range
for cars was 410  miles, and the average range for the combined fleet was approximately 420
miles (Figure 4-3).  This compares to average ranges of 390 miles (trucks), 360 miles (cars)
and 370 miles (fleet) in MY 1990. While the linear trend shows a small increase in range (5-
10%) over this time period, the factors discussed above preclude drawing simple conclusions
about the relationship between increases in average fuel economy and changes in fuel tank
size. As an example - in MY 2010, greater proportional sales of 14 sedans were seen as
compared to previous years. Since 14 and V6 sedan typically share the fuel tank component,
the 2009/2010 spike in range (corresponding to the dip in fuel consumption in the previous
chart) may or may not be a lasting increase, depending on manufacturer's redesign choices.
                                            4-61

-------
             420 -
             400 -
             380 -
             360 -
                      Economic and Other Assumptions Used in the Agencies' Analysis
                                Driving Range by MY
1990
                              1995
                             Figure 4-3 Driving range by MY
2010
       The agencies seek comment on the method and assumptions being used to estimate the
refueling benefit.
                                           4-62

-------
                      Economic and Other Assumptions Used in the Agencies' Analysis

       4.2.12  Discounting future benefits and costs

       Discounting future fu