Proposed Determination on the
Appropriateness of the Model Year
2022-2025 Light-Duty Vehicle
Greenhouse Gas Emissions Standards
under the Midterm Evaluation:
Technical Support Document
£% United States
Environmental Protect
Agency
-------
Proposed Determination on the
Appropriateness of the Model Year
2022-2025 Light-Duty Vehicle
Greenhouse Gas Emissions Standards
under the Midterm Evaluation:
Technical Support Documnet
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
United States
Environmental Protection
^1 Agency
EPA-420-R-16-021
November 2016
-------
TOC and Abbreviations
TABLE OF CONTENTS
List of Acronyms i
Executive Summary ES-1
Chapter 1: Baseline and Reference Vehicle Fleets 1-1
1.1 Baseline and Reference Vehicle Fleets 1-1
1.1.1 Why does EPA Establish Baseline and Reference Vehicle Fleets? 1-1
1.1.2 Key Comments on EPA's MY2014 Baseline Fleet Used in the Draft TAR 1-2
1.1.3 MY2015 Baseline Fleet used for this Proposed Determination 1-3
1.1.3.1 MY2015-Based MYs 2022-2025 Reference Fleet 1-9
1.1.3.1.1 On What Data are EPA's Reference Vehicle Fleet Volumes Based? 1-9
1.1.3.1.2 How did EPA develop the MY2015 Baseline and MYs 2022-2025
Reference Vehicle Fleet Volumes? 1-11
1.1.3.1.3 How was the MY2015 Baseline Data Merged with the IHS-Polk Data? 1-11
1.1.3.1.4 How were the IHS-Polk Forecast and the Unforced AEO 2015 Forecast
Used to Project the Future Fleet Volumes? 1-12
1.1.3.2 What Are the Sales Volumes and Characteristics of the MY2015 Based
Reference Fleet? 1-19
1.1.3.3 What Are the Differences in the Sales Volumes and Characteristics of the
MY2008-Based (FRM) and the MY2015-Based Reference Fleets? 1-22
1.1.3.4 What Are the Differences in the Sales Volumes and Characteristics of the
EPA MY2014-Based (Draft TAR) and the MY2015-Based
Reference Fleets? 1-25
1.2 The OMEGA Fleet 1-31
1.2.1 Incorporation of the California Zero Emissions Vehicle (ZEV) Program into the
OMEGA Reference Fleet 1-32
1.2.1.1 The ZEV Regulation in OMEGA 1-32
1.2.1.2 The ZEV Program Requirements 1-39
1.2.1.2.1 Overview 1-40
1.2.1.2.2 ZEV Credit Requirement 1-40
1.2.1.2.3 Projected Representative PHEV and BEV Characteristics for
MY2021-2025 1-41
1.2.1.2.4 Calculation of Incremental ZEVs Needed for ZEV Program
Compliance 1-45
Chapter 2: Technology Costs, Effectiveness, and Lead Time Assessment 2-1
2.1 Overview 2-1
2.2 State of Technology and Advancements since the 2012 Final Rule 2-5
2.2.1 Individual Technologies and Key Developments 2-5
2.2.1.1 List of Technologies Considered 2-5
2.2.1.2 Descriptions of Technologies and Key Developments since the FRM 2-6
2.2.2 Engines: State of Technology 2-13
2.2.2.1 Overview of Engine Technologies 2-15
2.2.2.2 Sources of Engine Effectiveness Data 2-17
-------
TOC and Abbreviations
2.2.2.3 Low Friction Lubricants (LUB) 2-18
2.2.2.4 Engine Friction Reduction (EFR1, EFR2) 2-18
2.2.2.5 Cylinder Deactivation (DEAC) 2-19
2.2.2.6 Variable Valve Timing (VVT) Systems 2-19
2.2.2.6.1 Intake Cam Phasing (ICP) 2-20
2.2.2.6.2 Coupled Cam Phasing (CCP) 2-20
2.2.2.6.3 Dual Cam Phasing (DCP) 2-20
2.2.2.6.4 Variable Valve Lift (VVL) 2-20
2.2.2.7 GDI, Turbocharging, Downsizing and Cylinder Deactivation 2-21
2.2.2.8 EGR 2-30
2.2.2.9 Atkinson Cycle 2-31
2.2.2.10 Miller Cycle 2-35
2.2.2.11 Light-duty Diesel Engines 2-38
2.2.2.12 Thermal Management 2-41
2.2.2.13 Reduction of Friction and Other Mechanical Losses 2-42
2.2.2.14 Potential Longer-Term Engine Technologies 2-43
2.2.3 Transmissions: State of Technology 2-44
2.2.3.1 Background 2-44
2.2.3.2 Transmissions: Summary of State of Technology 2-45
2.2.3.3 Sources of Transmission Effectiveness Data 2-46
2.2.3.4 Sources of GHG Emission Improvements: Reduction in Parasitic Losses, Engine
Operation, and Powertrain System Design 2-48
2.2.3.5 Automatic Transmissions (ATs) 2-50
2.2.3.6 Manual Transmissions (MTs) 2-53
2.2.3.7 Dual Clutch Transmissions (DCTs) 2-53
2.2.3.8 Continuously Variable Transmissions (CVTs) 2-54
2.2.3.9 Transmission Parasitic Losses 2-57
2.2.3.9.1 Losses in ATs 2-57
2.2.3.9.2 Losses in DCTs 2-58
2.2.3.9.3 Losses in CVTs 2-58
2.2.3.9.4 Neutral Idle Decoupling 2-59
2.2.3.10 Transmission Shift Strategies 2-59
2.2.3.11 Torque Converter Losses and Lockup Strategy 2-59
2.2.4 Electrification: State of Technology 2-60
2.2.4.1 Overview of Chapter 2-61
2.2.4.2 Overview of Electrification Technologies 2-62
2.2.4.3 Non-Battery Components of Electrified Vehicles 2-65
2.2.4.3.1 Propulsion Components 2-66
2.2.4.3.2 Power Electronics 2-68
2.2.4.3.3 Industry Targets for Non-Battery Components 2-72
2.2.4.4 Developments in Electrified Vehicles 2-74
2.2.4.4.1 Non-hybrid Stop-Start 2-74
2.2.4.4.2 Mild Hybrids 2-77
2.2.4.4.3 Strong Hybrids 2-82
2.2.4.4.4 Plug-in Hybrids 2-86
2.2.4.4.5 Battery Electric Vehicles 2-93
-------
TOC and Abbreviations
2.2.4.4.6 Relating Power to Acceleration Performance 2-104
2.2.4.5 Developments in Electrified Vehicle Battery Technology 2-106
2.2.4.5.1 Battery Chemistry 2-107
2.2.4.5.2 Pack Topology, Cell Capacity and Cells per Module 2-110
2.2.4.5.3 Usable Energy Capacity 2-113
2.2.4.5.4 Thermal Management 2-119
2.2.4.5.5 Pack Voltage 2-120
2.2.4.5.6 Electrode Dimensions 2-121
2.2.4.5.7 Pack Manufacturing Volumes 2-123
2.2.4.5.8 Potential Impact of Lithium Demand on Battery Cost 2-125
2.2.4.5.9 Evaluation of Draft TAR Battery Cost Projections 2-126
2.2.4.6 Fuel Cell Electric Vehicles 2-132
2.2.5 Aerodynamics: State of Technology 2-133
2.2.5.1 Background 2-133
2.2.5.2 Industry Developments 2-134
2.2.5.3 Feasibility of Aerodynamic Improvements 2-138
2.2.5.4 Results of U.S.-Canada Joint Test Program 2-139
2.2.6 Tires: State of Technology 2-142
2.2.6.1 Background 2-142
2.2.6.2 Industry Developments 2-142
2.2.7 Mass Reduction: State of Technology 2-145
2.2.7.1 Overview of Mass Reduction Technologies 2-145
2.2.7.2 Mass Reduction Feasibility 2-149
2.2.7.3 Market Implementation of Mass Reduction 2-151
2.2.7.4 Holistic Vehicle Mass Reduction and Cost Studies 2-152
2.2.7.4.1 EPA Holistic Vehicle Mass Reduction/Cost Studies 2-157
2.2.7.4.1.1 Phase 2 Low Development Midsize CUV Updated Study and
Supplement 2-157
2.2.7.4.1.2 Light Duty Pickup Truck Light-Weighting Study 2-160
2.2.7.4.2 NHTSA Holistic Vehicle Mass Reduction/Cost Studies 2-163
2.2.7.4.3 ARB Holistic Vehicle Mass Reduction/Cost Study 2-163
2.2.7.4.4 Aluminum Association Midsize CUV Aluminum BIW Study 2-164
2.2.7.4.5 Comparison of Data for Lightweight Car/CUV with Aluminum BIW.. 2-166
2.2.7.4.6 DOE/Ford/Magna MMLV Mach 1 and Mach 2 Lightweighting Research
Projects 2-168
2.2.7.4.6.1 Mach I 2-170
2.2.7.4.6.2 Mach 2 2-172
2.2.7.4.7 Technical Cost Modeling Report by DOE/INL/IBIS on 40 Percent-45
Percent Mass Reduced Vehicle 2-174
2.2.7.4.8 Mass Reduction Spectrum Analysis and Process Cost Modeling Report by
DOE/IBIS/Energetics/INL 2-175
2.2.7.4.9 Studies to Determine Potential Mass Addition for IIHS Small Overlap 2-176
2.2.7.4.9.1 NHTSA Mass Add Study for a Passenger Car to Achieve a "Good"
Rating on the IIHS Small Overlap 2-177
2.2.7.4.9.2 Transport Canada Mass Add Study for a Light Duty Truck to Achieve
a "Good" Rating on the IIHS Small Overlap 2-178
-------
TOC and Abbreviations
2.2.7.5 Potential Lightweight Recyclable Composite Fiber Material 2-180
2.2.8 State of Other Vehicle Technologies 2-181
2.2.8.1 Electrified Power Steering: State of Technology 2-181
2.2.8.2 Improved Accessories: State of Technology 2-181
2.2.8.3 Secondary Axle Disconnect: State of Technology 2-182
2.2.8.3.1 Background 2-182
2.2.8.3.2 Developments in AWD Technology 2-183
2.2.8.4 Low-Drag Brakes: State of Technology 2-186
2.2.9 Air Conditioning Efficiency and Leakage Credits 2-187
2.2.9.1 A/C Efficiency Credits 2-188
2.2.9.1.1 Manufacturer Utilization of A/C Efficiency Credits 2-188
2.2.9.1.2 Eligibility for A/C Efficiency Credits 2-190
2.2.9.1.3 The AC 17 Test Procedure 2-192
2.2.9.1.4 Summary 2-198
2.2.9.2 A/C Leakage Reduction and Alternative Refrigerant Substitution 2-199
2.2.9.2.1 Leakage 2-199
2.2.9.2.2 Low-GWP Refrigerants 2-200
2.2.9.2.3 Conclusions 2-201
2.2.10 Off-cycle Technology Credits 2-201
2.2.10.1 Off-cycle Credits Program 2-201
2.2.10.1.1 Off-cycle Credits Program Overview 2-201
2.2.10.2 Use of Off-cycle Technologies to Date 2-203
2.3 GHG Technology Assessment 2-206
2.3.1 Fundamental Assumptions 2-206
2.3.1.1 Technology Time Frame and Measurement Scale for Effectiveness and
Cost 2-206
2.3.1.2 Performance Assumptions 2-207
2.3.1.3 Fuels 2-209
2.3.1.4 Vehicle Classification 2-212
2.3.2 Approach for Determining Technology Costs 2-214
2.3.2.1 Direct Manufacturing Costs 2-215
2.3.2.1.1 Costs from Tear-down Studies 2-215
2.3.2.1.2 Electrified Vehicle Battery Costs 2-217
2.3.2.1.3 Specific DMC Updates since the Draft TAR 2-218
2.3.2.1.4 Approach to Cost Reduction through Manufacturer Learning 2-218
2.3.2.2 Indirect Costs 2-223
2.3.2.2.1 Methodologies for Determining Indirect Costs 2-223
2.3.2.2.2 Indirect Cost Estimates Used in this Analysis 2-225
2.3.2.3 Maintenance and Repair Costs 2-229
2.3.2.3.1 Maintenance Costs 2-229
2.3.2.3.2 Repair Costs 2-230
2.3.2.4 Costs Updated to 2015 Dollars 2-230
2.3.3 Approach for Determining Technology Effectiveness 2-231
2.3.3.1 Vehicle Benchmarking 2-231
2.3.3.1.1 Detailed Vehicle Benchmarking Process 2-232
2.3.3.1.1.1 Engine Testing 2-233
-------
TOC and Abbreviations
2.3.3.1.1.2 Transmission Testing 2-234
2.3.3.1.2 Development of Model Inputs from Benchmarking Data 2-237
2.3.3.1.2.1 Engine Data 2-237
2.3.3.1.2.2 Engine Map 2-237
2.3.3.1.2.3 Inertia 2-238
2.3.3.1.2.4 Transmission Data 2-239
2.3.3.1.2.5 Gear Efficiency and Spin Losses 2-239
2.3.3.1.2.6 Torque Converter 2-240
2.3.3.1.3 Vehicle Benchmarking Summary 2-241
2.3.3.2 Classification of Vehicles for Effectiveness 2-242
2.3.3.2.1.1 Significance of Power-to-Weight Ratio and Road-Load Power
Attributes 2-242
2.3.3.2.1.2 Effect of Changing Power-to-Weight Ratio 2-243
2.3.3.2.1.3 Effect of Advanced Technologies 2-245
2.3.3.2.1.4 Advanced Technology Trade-Off Curves 2-247
2.3.3.2.2 Definition of Effectiveness Classes 2-249
2.3.3.2.3 Comparison to Draft TAR Classification Approach and Exemplar Vehicles
2-251
2.3.3.3 ALPHA Vehicle Simulation Model 2-255
2.3.3.3.1 General ALPHA Description 2-256
2.3.3.3.2 Detailed ALPHA Model Description 2-256
2.3.3.3.2.1 Ambient System 2-257
2.3.3.3.2.2 Driver System 2-258
2.3.3.3.2.3 Powertrain System 2-258
2.3.3.3.2.3.1 Engine Subsystem 2-258
2.3.3.3.2.3.2 Electric Subsystem 2-259
2.3.3.3.2.3.3 Accessories Subsystem 2-260
2.3.3.3.2.3.4 Transmission Subsystem 2-260
2.3.3.3.2.3.4.1 Transmission Gear Selection 2-260
2.3.3.3.2.3.4.2 Launch Clutch Model 2-261
2.3.3.3.2.3.4.3 Gearbox Model 2-261
2.3.3.3.2.3.4.4 Torque Converter Model 2-262
2.3.3.3.2.3.4.5 Automatic Transmission & Controls 2-262
2.3.3.3.2.3.4.6 DCT Transmission & Control 2-262
2.3.3.3.2.3.4.7 CVT Transmission & Control 2-262
2.3.3.3.2.3.4.8 Driveline 2-262
2.3.3.3.2.3.5 Vehicle System 2-263
2.3.3.3.3 Energy Auditing 2-263
2.3.3.3.4 ALPHA Simulation Runs 2-264
2.3.3.3.5 Post-processing 2-265
2.3.3.3.6 Vehicle Component Vintage 2-266
2.3.3.3.7 Additional Verification 2-267
2.3.3.3.8 Key Public Comments Related to the ALPHA Model 2-268
2.3.3.4 Determining Technology Effectiveness for MY2022-2025 2-271
2.3.3.5 Lumped Parameter Model 2-274
2.3.3.5.1 Approach for Modeling Incremental Effectiveness 2-274
-------
TOC and Abbreviations
2.3.3.5.2 Calibration of LPM using ALPHA model 2-276
2.3.3.5.3 Lumped Parameter Model Usage in OMEGA 2-277
2.3.3.5.4 Appropriateness of LPM Effectiveness Modeling for the Overall Fleet2-279
2.3.4 Data and Assumptions Used in the GHG Assessment 2-287
2.3.4.1 Engines: Data and Assumptions for this Assessment 2-287
2.3.4.1.1 Low Friction Lubricants (LUB) 2-287
2.3.4.1.2 Engine Friction Reduction (EFR1, EFR2) 2-288
2.3.4.1.3 Cylinder Deactivation (DEAC) 2-289
2.3.4.1.4 Intake Cam Phasing (ICP) 2-290
2.3.4.1.5 Dual Cam Phasing (DCP) 2-291
2.3.4.1.6 Discrete Variable Valve Lift (DVVL) 2-292
2.3.4.1.7 Continuously Variable Valve Lift (CVVL) 2-292
2.3.4.1.8 Atkinson Cycle Engines in Non-HEV Applications 2-293
2.3.4.1.8.1 Effectiveness Data Used and Basis for Assumptions 2-293
2.3.4.1.8.2 Cost Data Used and Basis for Assumptions 2-307
2.3.4.1.8.3 Basis for Feasibility Assumptions 2-308
2.3.4.1.9 GDI, Turbocharging, Downsizing 2-311
2.3.4.1.9.1 Effectiveness Data Used and Basis for Assumptions 2-311
2.3.4.1.9.2 Cost Data Used and Basis for Assumptions 2-321
2.3.4.1.9.3 Basis for Feasibility Assumptions 2-324
2.3.4.2 Transmissions: Data and Assumptions for this Proposed Determination 2-325
2.3.4.2.1 Assessment and Classification of Automated Transmissions
(AT, AMT, DCT, CVT) 2-326
2.3.4.2.2 Effectiveness Values for TRX11 and TRX21 2-329
2.3.4.2.3 Effectiveness Values for TRX12 and TRX22 2-332
2.3.4.2.4 Technology Applicability and Costs 2-333
2.3.4.3 Electrification: Data and Assumptions for this Assessment 2-335
2.3.4.3.1 Cost and Effectiveness for Non-hybrid Stop-Start 2-335
2.3.4.3.2 Cost and Effectiveness for Mild Hybrids 2-337
2.3.4.3.3 Cost and Effectiveness for Strong Hybrids 2-339
2.3.4.3.4 Cost and Effectiveness for Plug-in Hybrids 2-341
2.3.4.3.5 Cost and Effectiveness for Battery Electric Vehicles 2-342
2.3.4.3.6 Cost of Non-Battery Components for xEVs 2-345
2.3.4.3.7 Cost of Batteries for xEVs 2-355
2.3.4.3.7.1 Battery Sizing Methodology forBEVs andPHEVs 2-359
2.3.4.3.7.2 Battery Sizing Methodology forHEVs 2-382
2.3.4.3.7.3 ANL BatPaC Battery Design and Cost Model 2-383
2.3.4.3.7.4 Assumptions and Inputs to BatPaC 2-385
2.3.4.3.7.5 Battery Cost Projections forxEVs 2-389
2.3.4.3.7.6 Discussion of Battery Cost Projections 2-398
2.3.4.3.7.7 Battery Pack Costs Used in OMEGA 2-399
2.3.4.3.7.8 Electrified Vehicle Costs Used In OMEGA
(Battery + Non-battery Items) 2-403
2.3.4.4 Aerodynamics: Data and Assumptions for this Assessment 2-405
2.3.4.5 Tires: Data and Assumptions for this Assessment 2-409
2.3.4.6 Mass Reduction: Data and Assumptions for this Assessment 2-411
-------
TOC and Abbreviations
2.3.4.6.1 Updates to Mass Reduction for the Current Analysis 2-411
2.3.4.6.2 Mass Reduction Costs used in OMEGA 2-413
2.3.4.7 Other Vehicle Technologies 2-421
2.3.4.7.1 Electrified Power Steering: Data and Assumptions for this Assessment 2-
421
2.3.4.7.2 Improved Accessories: Data and Assumptions for this Assessment 2-421
2.3.4.7.3 Secondary Axle Disconnect: Data and Assumptions for this Assessment.. 2-
422
2.3.4.7.4 Low Drag Brakes: Data and Assumptions for this Assessment 2-422
2.3.4.8 Air Conditioning: Data and Assumptions for this Assessment 2-423
2.3.4.9 Additional Off-cycle Credits and Costs 2-423
2.3.4.10 Cost Tables for Individual Technologies Not Presented Above 2-425
Chapter 3: Economic and Other Key Inputs Used in EPA's Analyses 3-1
3.1 The On-Road Fuel Economy "Gap" 3-1
3.1.1 The "Gap" Between Compliance and Real World Fuel Economy 3-1
3.1.2 Real World Fuel Economy and CO2 Projections 3-2
3.2 Fuel Prices and the Value of Fuel Savings 3-4
3.3 Vehicle Mileage Accumulation and Survival Rates 3-5
3.4 Fuel Economy Rebound Effect 3-8
3.4.1 Accounting for the Fuel Economy Rebound Effect 3-8
3.4.2 Summary of Historical Literature on the LDV Rebound Effect 3-10
3.4.3 Review of Recent Literature on LDV Rebound since the 2012 Final Rule 3-14
3.4.4 Basis for Rebound Effect Used in this Proposed Determination 3-19
3.5 Energy Security Impacts 3-21
3.5.1 Implications of Reduced Petroleum Use on U.S. Imports 3-21
3.5.2 Energy Security Implications 3-24
3.5.2.1 Effect of Oil Use on the Long-Run Oil Price 3-25
3.5.2.2 Macroeconomic Disruption Adjustment Costs 3-28
3.5.2.3 Cost of Existing U.S. Energy Security Policies 3-33
3.5.2.4 Military Security Cost Components of Energy Security 3-34
3.6 Non-GHG Health and Environmental Impacts 3-36
3.6.1 Economic Value of Reductions in Particulate Matter 3-37
3.7 Social Cost of Greenhouse Gas Emissions 3-41
3.8 Benefits from Reduced Refueling Time 3-49
3.9 Benefits and Costs from Additional Driving 3-51
3.9.1 Travel Benefit 3-51
3.9.2 Costs Associated with Crashes, Congestion and Noise 3-51
3.10 Discounting Future Benefits and Costs 3-52
3.11 Additional Costs of Vehicle Ownership 3-53
3.11.1 Sales Taxes 3-53
3.11.2 Insurance Costs 3-53
3.11.3 Financing Costs 3-53
-------
TOC and Abbreviations
Chapter 4: Consumer Issues 4-1
4.1 Potential Existence of Tradeoffs between Fuel Economy and Other Vehicle Attributes
4-1
4.1.1 The Reference Case 4-1
4.1.2 Recent Studies of the Engineering Tradeoffs between Power and Fuel Economy,
and Increases in Innovation 4-4
4.1.3 The Role of the Standards in Promoting Innovation 4-7
4.1.4 Potential Ancillary Benefits of GHG-Reducing Technologies 4-10
4.1.5 Estimating Potential Opportunity Costs and Ancillary Benefits 4-12
4.2 Consumer Response to Vehicles Subject to the Standards 4-16
4.2.1 Impact of the Standards on Vehicle Sales 4-16
4.2.2 Evaluations of the Vehicles Subject to the Standards by Professional Auto
Reviewers 4-20
4.3 Impacts of the Standards on Vehicle Affordability 4-38
4.3.1 Literature Review: Definitions of Affordability 4-38
4.3.2 Relating Affordability Themes to Vehicle Standards 4-43
4.3.3 EPA's Assessment of the Impacts of the Standards on Affordability 4-43
4.3.3.1 Data: Consumer Expenditure Survey 4-44
4.3.3.2 Effects on Lower-Income Households 4-46
4.3.3.3 Effect of the Standards on the Used Vehicle Market 4-48
4.3.3.4 Effects on Access to Credit 4-50
4.3.3.5 Effects on Low-Priced Vehicles 4-52
4.3.4 Conclusion 4-55
Chapter 5: EPA's OMEGA Model 5-1
5.1 OMEGA Overvi ew 5-1
5.2 OMEGA Model Structure 5-3
5.3 OMEGA Pre-Processors, Vehicle Types & Packages 5-5
5.3.1 Vehicle Types 5-5
5.3.2 Technology Packages, Package Building & Master-sets 5-7
5.3.3 Master-set Ranking and the Technology Input File 5-13
5.3.4 Applying Ranked-sets of Packages to the Projected Fleet 5-17
5.3.5 New to OMEGA since the Draft TAR 5-18
Appendix A EPA Response to the Alliance of Automobile Manufacturers' Contractor
Reports Titled "Final Report for Technology Effectiveness [Phases 1 and 2]" A-l
A. 1 Constraints on Technology Combinations and Technological Innovation A-l
A.2 Novation's Simplistic Methodology and Lack of Rigor A-2
A.3 Omission of Vehicle Load and Technology Penetration Rate Changes A-3
A.4 Arbitrary and Restrictive Assumptions and Constraints A-4
A.5 Displacement Specific Load and Exemplars A-6
A.6 Other Studies A-7
-------
TOC and Abbreviations
Appendix B Fleet-Wide Analysis of Powertrain Efficiency for Current and Future
Technology Packages A-8
B. 1 Introducti on A-8
B.2 Methodology A-8
B.2.1 Definition of Powertrain Efficiency A-8
B.2.2 Considering Tractive Energy Reductions for Future Technology Packages A-10
B.2.3 Displacement Specific Operating Load A-13
B.2.4 Choice of Drive Cycle A-14
B.3 Sample Calculation of Powertrain Efficiency A-15
Appendix C C02 Targets with Current Powertrain Designs A-17
Appendix D EPA Comparison Testing performed on a MY2014 Mazda SKYACTIV-G
Engine using Different Fuels Designs A-27
-------
TOC and Abbreviations
This Page Intentionally Left Blank
-------
TOC and Abbreviations
List of Acronyms
2MHEV
2-Mode Hybrid
ABS
Anti-lock Braking System
ABT
Averaging, Banking, and Trading
AC
Alternating Current
A/C
Air Conditioning
ACEEE
American Council for an Energy-Efficient Economy
AEO
Annual Energy Outlook
AER
All-Electric Range
AFDC
Alternative Fuels Data Center
AGM
Absorbent Glass Mat
AHSS
Advanced High Strength Steel
ALPHA
Advanced Light-Duty Powertrain and Hybrid Analysis Tool
AMT
Automated Manual Transmission
ANL
Argonne National Laboratory
ARB
California Air Resources Board
ASI
Area Specific Impedance
ASL
Aggressive Shift Logic
ASM
Annual Survey of Manufacturers
AT
Automatic Transmissions
Avg
Average
AWD
All Wheel Drive
BenMAP
Benefits Mapping and Analysis Program
BEV
Battery Electric Vehicle
BISG
Belt Integrated Starter Generator
BIW
Body-In-White
BLS
Bureau of Labor Statistics
BMEP
Brake Mean Effective Pressure
BOM
Bill of Materials
BSFC
Brake Specific Fuel Consumption
BTE
Brake-Thermal Efficiency
BTU
British Thermal Unit
CAA
Clean Air Act
CAD
Computer Aided Designs
CAD/CAE
Computer Aided Design and Engineering
CAE
Computer Aided Engineering
CAFE
Corporate Average Fuel Economy
CARB
California Air Resources Board
CAVs
Connected and Automated (or autonomous) Vehicles
CBD
Center for Biological Diversity
CBI
Confidential Business Information
CCP
Coupled Cam Phasing
1
-------
TOC and Abbreviations
CDPF Catalyzed Diesel Particulate Filter
CEC California Energy Commission
CES Consumer Expenditure Survey
CFD Computational Fluid Dynamics
CFR Code of Federal Regulations
CH4 Methane
CISG Crank Integrated Starter Generator
CNG Compressed Natural Gas
CO Carbon Monoxide
CO2 Carbon Dioxide
CCheq C02 Equivalent
COP Coefficient of Performance
CSM Conceptual Site Model
CSV Comma-separated Values
CUV Crossover Utility Vehicles
CVT Continuously Variable Transmission
CY Calendar Year
DC Direct Current
DCFC Direct Carbon Fuel Cell
DCP Dual Cam Phasing
DCT Dual Clutch Transmission
DEAC Cylinder Deactivation
DFMA™ Design for Manufacturing and Assembly
DGS California Department of General Services
DICE Dynamic Integrated Climate and Economy
DMC Direct Manufacturing Costs
DoE Department of Energy
DOE Design of Experiments
DOHC Dual Overhead Camshaft Engines
DOT Department of Transportation
DRI Dynamic Research, Inc.
DRLs Daytime Running Lamps
DVVL Discrete Variable Valve Lift
EGR Exhaust Gas Recirculation
EHPS Electrohydraulic Power Steering
Energy Information Administration (part of the U.S. Department of
IjIA. t-> \
Energy)
EISA Energy Independence and Security Act
EIVC Early Intake Valve Closing
EPA Environmental Protection Agency
EPCA Energy Policy and Conservation Act
EPRI Electric Power Research Institute
EPS Electric Power Steering
11
-------
TOC and Abbreviations
EPS Energy Power Systems
EREV Extended Range Electric Vehicle
ERM Employment Requirements Matrix
ESC Electronic Stability Control
EV Electric Vehicle
EVSE Electric Vehicle Supply Equipment
FARS Fatality Analysis Reporting System
FCEV Fuel Cell Electric Vehicle
FCPM Fuel Cost Per Mile
FCEV Fuel Cell Electric Vehicle
FE Finite Element
FEV1 Functional Expiratory Volume
FHWA Federal Highway Administration
FMEP Friction Mean Effective Pressure
FMVSS Federal Motor Vehicle Safety Standards
FR Federal Register
FRIA Final Regulatory Impact Analysis
FRM Federal Rulemaking
FRM Federal Reference Method
FTP Federal Test Procedure
gal/mi Gallon/Mile
GCWR Gross Combined Weight Rating
GDI Gasoline Direct Injection
GDP Gross Domestic Product
GEM Greenhouse gas Emissions Model
GHG Greenhouse Gases
Greenhouse Gases, Regulated Emissions, and Energy Use in
C t R F, F,1 rp , . •
Transportation
GVW Gross Vehicle Weight
GWP G1 ob al Warming Potenti al
GWU George Washington University
HD Heavy-Duty
HEV Hybrid Electric Vehicle
HFC Hydrofluorocarbon
HFET Highway Fuel Economy Dynamometer Procedure
HIL Hardware-In-Loop
hp Horsepower
hrs Hours
HP/WT Horsepower Divided by Weight
HVAC Heating, Ventilating, And Air Conditioning
hz Hertz
IACC Improved Accessories
IAM Integrated Assessment Models
111
-------
TOC and Abbreviations
IATC Improved Automatic Transmission Control
IC Indirect Cost
ICCT International Council on Clean Transport
ICF ICF International
ICM Indirect Cost Multiplier
ICMs Indirect Cost Markups
IHX Internal Heat Exchanger
IMA Improved Mobile Assist
IMAC Improved Mobile Air Conditioning
INL Idaho National Laboratory
IOU Investor Owned Utilities
IPCC Intergovernmental Panel on Climate Change
IPM Integrated Planning Model
ITC Institute of Transportation Studies
IWG Interagency Working Group
k Thousand
kg Kilogram
kW Kilowatt
kWh kilowatt-hour
L Liter
lb Pound
LBNL Lawrence Berkeley National Laboratory
LD Light-Duty
LEV Low-Emission Vehicle
LHD Light Heavy-Duty
LD V Li ght Duty Vehi cl e
LNT Lean NOx Trap
LRR Lower Rolling Resistance
LT Light Trucks
LWT Lightweighted Pickup Truck
MAD Minimum Absolute Deviation
MBPD Million Barrels Per Day
MD Medium-Duty
MDPV Medium-Duty Passenger Vehicles
MEMA Motor Equipment Manufacturers Association
Mg Megagrams
mg Milligram
MHEV Mild Hybrid
mi mile
min minimum
min Minute
MM Million
MMLV Multi-Material Lightweight Vehicle
IV
-------
TOC and Abbreviations
MMT Million Metric Tons
MOVES Motor Vehicle Emissions Simulator
mpg Miles per Gallon
mph Miles per Hour
MPV Multi-Purpose Vehicle
MSRP Manufacturer's Suggested Retail Price
MTE Mid Term Evaluation
MuD Multi-Unit Development
MY Model Year
N2O Nitrous Oxide
NA Not Applicable
NAAQS National Ambient Air Quality Standards
NADA National Automobile Dealers Association
NAS National Academy of Sciences
NCA National Climate Assessment
NCAP New Car Assessment Program
NEMS National Energy Modeling System
NESHAP National Emissions Standards for Hazardous Air Pollutants
NF3 Nitrogen Trifluoride
NGO Non-Governmental Organization
NHTSA National Highway Traffic Safety Administration
NiMH Nickel Metal-Hydride
NF3 Nitrogen Trifluoride
NOX Nitrogen Oxides
NO2 Nitrogen Dioxide
NOx Oxides of Nitrogen
NPRM Notice of Proposed Rulemaking
NRC National Research Council
NRC-CAN National Research Council of Canada
NREL National Renewable Energy Laboratory
NVH Noise Vibration and Harshness
NVPP National Vehicle Population Profiles
OAR EPA's Office of Air and Radiation
OEM Original Equipment Manufacturer
OECD Organization for Economic Cooperation and Development
OHV Overhead Valve
OLS Ordinary Least Squares
OMB EPA's Office of Management and Budget
OPEC Organization of Petroleum Exporting Countries
ORNL Oak Ridge National Laboratory
OTAQ EPA's Office of Transportation and Air Quality
PAGE Policy Analysis of the Greenhouse Effect
PC Passenger Car
v
-------
TOC and Abbreviations
P/E
Power-to-Energy
PEF
Peak Expiratory Flow
PEV
Plug-in Electric Vehicle
PFCs
Perfluorocarbons
PFI
Port-fuel-inj ection
PGM
Platinum Group Metal
PHEV
Plug-in Hybrid Electric Vehicle
PLM
Planar Layered Matrix
PM
Particulate Matter
PM2.5
Fine Particulate Matter (diameter of 2.5 |im or less)
PMSMs
Permanent-Magnet Synchronous Motors
PSHEV
Power-split Hybrid
PSI
Pounds per Square Inch
PWM
Pulse-width Modulated
R&D
Research and Development
RFS2
Renewable Fuel Standard 2
RIA
Regulatory Impact Analysis
RPE
Retail Price Equivalent
RPM
Revolutions per Minute
RSM
Response Surface Models
RTI
RTI International (formerly Research Triangle Institute)
SA
Strategic Analysis, Inc.
SAB
Science Advisory Board
SAB-
Science Advisory Board Environmental Economics Advisory
EEAC
Committee
SAE
Society of Automotive Engineers
SCO3
Soak Control third iteration
see
Social Cost of Carbon
SCR
Selective Catalyst Reduction
sf6
Sulfur Hexafluoride
SGDI
Stoichiometric Gasoline Direct Injection
SHEV
Strong Hybrid Electric Vehicles
SI
Spark-Ignition
SIDI
Spark Ignition Direct Injection
SIL
Software-In-Loop
SMDI
Steel Market Development Institutes
SNAP
Significant New Alternatives Policy
SNPRM
Supplemental Notice of Proposed Rulemaking
S02
Sulfur Dioxide
SOx
Sulfur Oxides
SOC
State of Charge
SOHC
Single Overhead Cam
VI
-------
TOC and Abbreviations
SOL
Small Overlap
SPR
Strategic Petroleum Reserve
Std
Standard
SUV
Sport Utility Vehicle
TAR
Technical Assessment Report
TC
Total Costs
TCIP
Tire Consumer Information Program
TDC
Top Dead Center
Tds
Direct Solar Transmittance
TFECIP
Tire Fuel Efficiency Consumer Information Program
TPE
Total Primary Energy
TRBDS
Turbocharging and Downsizing
TSD
Technical Support Document
UMTRI
University of Michigan Transportation Research Institute
UTQGS
Uniform Tire Quality Grading Standards
V2V
Vehicle-To-Vehicle
VGI
Vehicle Grid Integration
VIF
Variance Inflation Factor
VMT
Vehicle Miles Traveled
VOC
Volatile Organic Compound
VSL
Vehicle Speed Limiter
VVL
Variable Valve Lift
VVT
Variable Valve Timing
WT/FP
Weight Divided By Footprint
Vll
-------
Executive Summary
Executive Summary
The rulemaking establishing the National Program for Federal greenhouse gas (GHG)
emissions and corporate average fuel economy (CAFE) standards for model year (MY) 2017-
2025 light-duty vehicles included a regulatory requirement for the Environmental Protection
Agency (EPA) to conduct a Midterm Evaluation (MTE) of the greenhouse gas (GHG) standards
established for MYs 2022-2025. Through the MTE, EPA must determine no later than April 1,
2018 whether the MY2022-2025 GHG standards, established in 2012, are still appropriate under
section 202 (a) (1) of the Clean Air Act ("Act"), in light of the record then before the
Administrator, given the latest available data and information. The Administrator is making a
Proposed Determination that the MY2022-2025 standards adopted in the 2012 final rule
establishing the MY2017-2025 standards remain appropriate under section 201 (a) (1) of the Act.
This Technical Support Document (TSD) provides additional detailed analyses supporting this
Proposed Determination.
The Proposed Determination follows the July 2016 release of a Draft Technical Assessment
Report (TAR), issued jointly by EPA, National Highway Traffic Safety Administration
(NHTSA), and the California Air Resources Board (CARB). EPA requested comment on the
analysis supporting the Draft TAR and has fully considered those public comments as well as
other new information, and has updated its analyses where appropriate as part of this Proposed
Determination. This TSD describes in more detail our assessment of public comment on the
Draft TAR and updates to our technology costs, technology effectiveness, consumer impacts,
and other elements of our analysis.
A summary of each chapter of the TSD follows:
Chapter 1: Baseline and Reference Vehicle Fleets. This chapter describes EPA's
methodologies for developing a baseline fleet of vehicles and future fleet projections out to
MY2025. The Proposed Determination analysis uses a baseline fleet based on the MY2015 fleet,
the latest year available for which there are final GHG compliance data. EPA used data from
Energy Information Administration's Annual Energy Outlook 2016 (AEO 2016) as the basis for
total vehicle sales projections to 2025, as well as for the car and truck volume mix.
Chapter 2: Technology Costs, Effectiveness, and Lead Time Assessment. This chapter is
an in-depth assessment of the state of vehicle technologies to reduce GHG emissions and
improve fuel economy, as well as EPA's assessment of expected future technology
developments through MY2025. The technologies evaluated include all those considered for the
2012 final rule and the Draft TAR, as well as new technologies that have emerged. Every
technology has been reconsidered with respect to its cost, effectiveness, application, and lead
time considerations, with emphasis on assessing the latest introductions of technologies to
determine if and how they have changed.
Chapter 3: Economic and Other Key Inputs Used in EPA's Analyses. This chapter
describes many of the economic and other inputs used in the Proposed Determination analyses.
This chapter discusses the methodologies used to assess inputs such as the real-world fuel
economy/GHG emissions gap, vehicle miles traveled (VMT), vehicle survival rates, the VMT
rebound effect, energy security, the social cost of carbon and other GHGs, health benefits,
consumer cost of vehicle ownership, and others.
ES-1
-------
Executive Summary
Chapter 4: Consumer Issues. This chapter reviews issues surrounding consumer
acceptance of the vehicle technologies expected to be used to meet the MY2022-2025 standards.
Since the GHG standards have been in effect since MY2012, EPA focuses on the evidence to
date related to consumer acceptance of vehicles subject to these standards. This chapter also
discusses potential impacts of the standards on vehicle sales and affordability, which are closely
interconnected with the effects of macroeconomic and other market forces.
Chapter 5: EPA's OMEGA Model. This chapter describes EPA's computerized program
called the Optimization Model for reducing Emissions of Greenhouse gases from Automobiles
(OMEGA), the model used to efficiently apply technologies to the wide range of vehicles
produced by various manufacturers.
ES-2
-------
Baseline and Reference Vehicle Fleets
Table of Contents
Chapter 1: Baseline and Reference Vehicle Fleets 1-1
1.1 Baseline and Reference Vehicle Fleets 1-1
1.1.1 Why does EPA Establish Baseline and Reference Vehicle Fleets? 1-1
1.1.2 Key Comments on EPA's MY2014 Baseline Fleet Used in the Draft TAR 1-2
1.1.3 MY2015 Baseline Fleet used for this Proposed Determination 1-3
1.1.3.1 MY2015-Based MYs 2022-2025 Reference Fleet 1-9
1.1.3.1.1 On What Data are EPA's Reference Vehicle Fleet Volumes Based? 1-9
1.1.3.1.2 How did EPA develop the MY2015 Baseline and MYs 2022-2025
Reference Vehicle Fleet Volumes? 1-11
1.1.3.1.3 How was the MY2015 Baseline Data Merged with the IHS-Polk Data? 1-11
1.1.3.1.4 How were the IHS-Polk Forecast and the Unforced AEO 2015 Forecast
Used to Project the Future Fleet Volumes? 1-12
1.1.3.2 What Are the Sales Volumes and Characteristics of the MY2015 Based
Reference Fleet? 1-19
1.1.3.3 What Are the Differences in the Sales Volumes and Characteristics of the
MY2008-Based (FRM) and the MY2015-Based Reference Fleets? 1-22
1.1.3.4 What Are the Differences in the Sales Volumes and Characteristics of the EPA
MY2014-Based (Draft TAR) and the MY2015-Based Reference Fleets? 1-25
1.2 The OMEGA Fleet 1-31
1.2.1 Incorporation of the California Zero Emissions Vehicle (ZEV) Program into the
OMEGA Reference Fleet 1-32
1.2.1.1 The ZEV Regulation in OMEGA 1-32
1.2.1.2 The ZEV Program Requirements 1-39
1.2.1.2.1 Overview 1-40
1.2.1.2.2 ZEV Credit Requirement 1-40
1.2.1.2.3 Projected Representative PHEV and BEV Characteristics for MY2021-
2025 1-41
1.2.1.2.4 Calculation of Incremental ZEVs Needed for ZEV Program Compliance.. 1-
45
Table of Figures
Figure 1.1 The Verily Process for the Data EPA's MY2015 Baseline Vehicle Fleet is Based 1-4
Figure 1.2 Process Flow for Creating the Baseline and Reference Fleet 1-5
Figure 1.3 Process Flow for Determining where Segment Volume Should Move 1-15
Figure 1.4 Relative Cost of ZEV Credits for Different Ranges and Battery Costs 1-42
Table of Tables
Table 1.1 MY2015 Engine Technology Penetration 1-6
Table 1.2 Change (2015-2008) in Engine Technology Penetration 1-8
Table 1.3 AEO 2016 Unforced Reference Case Values used in the MY2015 Based Market Fleet Projection 1-9
Table 1.4 AEO 2016 Reference Case Values 1-10
Table 1.5 AEO 2015 and AEO 2016 Reference Case Fuel Prices 1-11
Table 1.6 List of IHS-Polk Segments 1-13
Table 1.7 Example of Honda Vehicles Being Mapped to Segments Based On the IHS-Polk Forecast 1-13
Table 1.8 Example Honda 2015 Volumes by Segment from the IHS-Polk Forecast 1-16
-------
Baseline and Reference Vehicle Fleets
Table 1.9 Example Values Used to Determine the MDPV Multiplier for FCA 1-17
Table 1.10 Example Values Used to Determine FCA's 2025 Van Volume 1-17
Table 1.11 Example Values Used to Determine FCA 2025 Individual Full-Size Non-Premium Van Multiplier.. 1-18
Table 1.12 Example Applying the Individual Full-Size Non-Premium Van Multiplier for FCA 1-18
Table 1.13 Example Unforced AEO 2016 Truck and Car Multipliers in MY2025 1-18
Table 1.14 Example Applying the Unforced AEO Truck Multiplier to FCA Full-Size Non-Premium Vans 1-19
Table 1.15 Vehicle Segment Volumes 1-19
Table 1.16 Car and Truck Volumes 1-19
Table 1.17 Car and Truck Definition Manufacturer Volumes 1-20
Table 1.18 Production Weighted Foot Print Mean 1-21
Table 1.19 Percentages of 4, 6, and 8 Cylinder Engines by Model Year 1-21
Table 1.20 Differences in Vehicle Segment Volumes 1-22
Table 1.21 Differences in Actual and Projected Sales Volumes between MY2015 and MY2008 fleets 1-23
Table 1.22 Differences in Sales Volumes by Manufacturer and Car/Truck Type between MY2008-based and
MY2015-based fleets 1-23
Table 1.23 Difference in Footprint Distributions between MY2015-based and MY2008-based Fleet Projections 1-25
Table 1.24 Differences in Percentages of 4, 6 and 8 Cylinder Engines by Model Year 1-25
Table 1.25 Vehicle Segment Volume Differences 1-26
Table 1.26 Differences in Actual and Projected Sales Volumes between MY2015 and MY2014 fleets 1-27
Table 1.27 Differences in Sales Volumes by Manufacturer and Car/Truck Type between MY2014-based and
MY2015-based fleets 1-28
Table 1.28 Change (2015-2014) in Engine Technology Penetration 1-29
Table 1.29 2015 Projection - 2014 Projection Production Weighted Foot Print Mean Difference 1-30
Table 1.30 Differences in Percentages of 4, 6 and 8 Cylinder Engines by Model Year 1-31
Table 1.31 OMEGA MY2021 Car Fleet using the AEO 2016 Reference Fuel Price Case 1-34
Table 1.32 OMEGA MY2021 Truck Fleet using the AEO 2016 Reference Fuel Price Case 1-34
Table 1.33 OMEGA MY2025 Car Fleet using the AEO 2016 Reference Fuel Price Case 1-35
Table 1.34 OMEGA MY2025 Truck Fleet using the AEO 2016 Reference Fuel Price Case 1-35
Table 1.35 Breakdown of MY2025 Internal Combustion Engine, Electric and Plug-in Electric Vehicle Sales using
the AEO 2016 Reference Fuel Price Case 1-36
Table 1.36 Vehicle Types Considered for Conversion to ZEV Program Vehicles 1-37
Table 1.37 Example Manufacturer Fleet from which ZEVs are to be Created 1-38
Table 1.38 Number of Additional ZEV Program Sales from each Platform 1-38
Table 1.39 Percentage of Additional ZEV Program Sales from Each Vehicle Model 1-38
Table 1.40 Example Manufacturer's OMEGA Fleet including ZEV Program Sales 1-39
Table 1.41 ZEV Regulation Credit Requirements 1-41
Table 1.42 Range Characteristics of BEVs forMY2015 1-43
Table 1.43 Range Characteristics of PHEVs forMY2015 1-43
Table 1.44 Projected Sales Weighted BEV Range for MY2021-2025 1-44
Table 1.45 Projected Sales Weighted PHEV Range for MY2021-2025 1-45
-------
Baseline and Reference Vehicle Fleets
Chapter 1: Baseline and Reference Vehicle Fleets
1.1 Baseline and Reference Vehicle Fleets
The passenger cars and light trucks sold currently in the United States, and those that are
anticipated to be sold in the model years (MYs) 2021-2025 time frame, 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. The recent decline in oil prices and the
improved 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, a starting point of any analysis must be to characterize and quantify a future
fleet in order to assess the impacts of the 2022-2025 GHG standards that would affect that future
fleet. As in the FRM and the Draft TAR, EPA has examined various publicly-available sources
(some requiring purchase), and then used inputs from those sources in a series of models to
project the composition of baseline and reference fleets for the purposes of this analysis. This
chapter describes this process, and the characteristics of the baseline and reference fleets.
EPA has made every effort to make this analysis transparent and duplicable. Because both
the input and output sheets from our modeling are public,1 stakeholders can verify and check
EPA's modeling results, and use the results to perform their own analyses.
1.1.1 Why does EPA Establish Baseline and Reference Vehicle Fleets?
In order to calculate the impacts of the final 2022-2025 GHG standards, it is necessary to
estimate the composition of the future vehicle fleet absent the 2022-2025 standards. EPA has
developed a baseline/reference fleet in two parts. The first step was to develop a "baseline" fleet.
The baseline fleet represents data from a single model year of actual vehicle sales. EPA creates a
baseline fleet in order to track the volumes and types of C02-reducing technologies that are
already present in the existing vehicle fleet. Creating a baseline fleet accounts for technologies
already deployed in the fleet, and thus not only is a necessary step in assessing what additional
technologies might be added and the costs and benefits of adding those technologies, but also
avoids double-counting of those costs and benefits. Specifically, an accurate assessment of the
baseline fleet prevents the OMEGA model from adding technologies to vehicles that already
have these technologies, which would result in such double-counting.
The second step was to project the baseline fleet sales into MYs 2022-2025. This is called the
"reference" fleet volumes, and it represents the fleet volumes (but, until later steps, not additional
levels of technology) that EPA believes would exist in MYs 2022-2025 absent the application of
the 2022-2025 GHG standards.
After determining the reference fleet volumes, the third step is to account for technologies
(and corresponding increases in cost and reductions in CO2 emissions) that could be added to the
baseline technology vehicles in the future, taking into account previously-promulgated standards,
and assuming MY2021 standards apply at the same levels through MY2025. This step uses the
OMEGA model to add technology to each vehicle in the baseline market forecast such that each
manufacturer's car and truck average CO2 levels reflect that manufacturer's projected MY2021
standards. The model's output, the "reference case," is the light-duty fleet estimated to exist in
l-l
-------
Baseline and Reference Vehicle Fleets
MYs 2022-2025 without new GHG standards (that is, without any standards beyond the
MY2021 standards). All of EPA's estimates of emission reduction improvements, costs, and
societal impacts for purposes of this Proposed Determination are developed in relation to the
reference case.
This chapter describes the first two steps of the development of the baseline and reference
fleets volumes. The third step is technology addition which is developed as the outputs of the
OMEGA model (see Chapter 5 for an explanation of how the models apply technologies to
vehicles in order to evaluate potential paths to compliance).
1.1.2 Key Comments on EPA's MY2014 Baseline Fleet Used in the Draft TAR
For the Draft TAR, EPA chose to create a baseline fleet based on MY2014 data because, at
the time, it was the most recent year for which a complete set of certification data was available.
See Draft TAR at p. 4-2 and 4-9. In general, several commenters (for example, Union of
Concerned Scientists and Environmental Defense Fund) supported EPA's use of MY2014 data
since it was the latest year of final compliance data. The Alliance of Automobile Manufacturers
(AAM) sent mixed messages in their comments. AAM noted that MY2015, used by NHTSA in
its CAFE analysis, was more recent and urged that EPA use the latest data available. AAM went
on to say that we should use the data that was available 90 days after the end of production,
which was MY2014 data. However, in order to create a baseline fleet that meets the AAM
suggestion, EPA would need to create the fleet based on manufacturer provided mid-year
reports. The mid-year reports do not constitute data — these reports are estimates of what the
manufacturer's year end production and GHG performance are projected to be. See Draft TAR at
p. 4-9. The estimated GHG values along with the estimated volume values thus may not give an
accurate view of the fleet.
Global Automakers commented that EPA included vehicles in its modeling that were no
longer in production. However, manufacturers will often eliminate a model in a vehicle class and
later have a new model enter the same vehicle class. Thus, the fact that a model is discontinued
does not mean that the class of vehicle will no longer be represented in the future fleet. EPA
picks a model year of vehicles and then projects them forward based on their vehicle class.
There is an initial assumption that all vehicles in that model year are needed to represent the
needs of the public. EPA then used the IHS-Polk forecast to determine if a class of vehicle might
be discontinued. Put another way, for projecting the future vehicle fleet, EPA changes the
proportions of vehicles in a vehicle class based on IHS-Polk's forecast to represent the public's
future needs, but does not automatically eliminate a class of vehicle because a particular model is
discontinued. The only way a vehicle is eliminated is if a manufacturer no longer participates in
a vehicle class. In short, eliminating a model would eliminate a choice that is assumed to be
needed by the public unless its class has been eliminated by the IHS-Polk forecast.
Our concerns regarding use of a mid-year report is now obviated, however, because final
certification data from the EPA Verify Database for MY2015 is now available. Consistent with
the approach in the Draft TAR of using the most recent final certification data for the baseline
year, EPA is using these data for establishing the baseline fleet. See Draft TAR pp. 4-2 and 4-9;
for a description of the Verify Database, see the following Chapter 1.1.3.
Commenters also urged EPA and NHTSA to use a common baseline for future analysis.
Although this analysis is not a joint exercise, EPA has moved to MY2015 since final data is now
1-2
-------
Baseline and Reference Vehicle Fleets
available. As stated in the Draft TAR and reconfirmed above, EPA uses the most recent model
year for which final sales data is available for its analysis.
AAM commented that EPA should consider using a multi-year average instead of a single
year for the baseline. EPA believes that using a multi-year average would be problematic since
technology on vehicles changes from year to year which would make accurately representing a
multi-year averaged fleet extremely challenging.
AAM also voiced the belief that we had removed 800,000 vehicles from the AEO's
projections. The tables we provided are in fact consistent with what EIA published for
AEO2015. See Chapter 1.1.3.1.1 below. AAM also commented that we could have used our
contractor's (IHS-Polk) projections of total vehicle sales. However, EIA is the standard
government-wide reference, and for EPA to deviate from that source would put us out of step
with the rest of the federal government. EPA believes that consistency on total volumes across
agencies should be pursued where feasible, and believes that EIA's projections are the best
available source for projections of total car and total light truck sales.
1.1.3 MY2015 Baseline Fleet used for this Proposed Determination
EPA has updated the basis for the baseline fleet used in the Proposed Determination analysis
to reflect MY2015, the latest available model year for which there is final manufacturer GHG
certification data. The MY2015 fleet GHG data is the most recent complete set of final U.S.
vehicle data that includes actual manufacturer volumes and CO2 values. The MY2015 volumes
and CO2 values come from the EPA VerifyA database. The data contained in the Verify system
is quite robust since it undergoes a complex number of quality checks that are performed first by
the manufacturer, then by the Verify database software, and finally by EPA's certification staff.
Figure 1.1 shows the quality steps that are completed before data is available for use in the
Verify system. The finalized 2015 GHG certification data is thus the most accurate
representation of vehicle and technology mix for MY2015. B As noted above, this baseline fleet
is not identical to that established by NHTSA in the Draft TAR, since that fleet reflected mid-
year manufacturer reports. See Draft TAR Chapter 13.1.1. EPA supplemented this data with
valve train information from Wards Automotive Group, C'D and curb weights and power steering
information from NHTSA's 2015 Volpe Baseline Fleet file created for the Draft TAR.
A The EPA Verify Database is the electronic system by which vehicle manufacturers provide their compliance data
to EPA. There are several built-in quality assurance provisions.
B We note that this 2015 MY baseline fleet is not identical to that established by NHTSA in the Draft TAR, since
that fleet reflected mid-year manufacturer reports rather than the final certified data used here. See Draft TAR
Chapter 13.1.1.
c WardsAuto.com: Used as a source for engine specifications shown in Figure 1.2.
D Note that WardsAuto.com, where this information was obtained, is a fee-based service, but all information is
public to subscribers.
1-3
-------
Baseline and Reference Vehicle Fleets
The Verify database does cross checks against all data submitted at each step.
Submit all test
results
A manufacturer
must define all
vehicle models
A manufacturer
must define all
engine test groups
and link them to
the vehicle models
Define all test
vehicles that will
be used for
emissions testing.
Final verification is manually done
to ensure the manufacturers'
calculations match the Verify
database's calculation.
Label all vehicles for FE. This ties a
vehicle, an engine, transmission,
and driveline with a test. It also
determines all sub configurations for
that vehicle.
Submit final GHG/CAFE data. This step is
where actual volumes for each vehicle is
submitted. It is also the point where the GHG/
CAFE standard is calculated for each
manufacturer.
Figure 1.1 The Verify Process for the Data EPA's MY2015 Baseline Vehicle Fleet is Based
Similar to the 2008 baseline that EPA used in the 2017-2025 GHG FRM and the 2014
baseline fleet used for the Draft TAR, most of the information about the vehicles that make up
the 2015 fleet was gathered from EPA's emission certification and fuel economy database, most
of which is publicly available. (Note that a 2010 baseline was created for the 2017-2025 GHG
FRM, but it was only used for a sensitivity analysis and will not be used for analysis in this
Proposed Determination).2 The 2015 GHG certification data included (by individual vehicle
model produced in MY2015): vehicle production volume, carbon dioxide emissions rating for
GHG certification, fuel type, fuel injection type, EGR, number of engine cylinders,
displacement, intake valves per cylinder, exhaust valves per cylinder, variable valve timing,
variable valve lift, engine cycle, cylinder deactivation, transmission type, drive type (rear-wheel,
all-wheel, etc.), hybrid type (if applicable), and aspiration (naturally-aspirated, turbocharged,
etc.). In addition, as noted above, EPA augmented the 2015 GHG certification and fuel economy
database (the EPA "Verify" database) with publicly-available data which includes valve train
information from Ward's Automotive Group, and data from NHTSA's MY2015 Draft TAR
Volpe Baseline.
The process by which EPA created the 2015 baseline fleet Excel file is similar to the process
used to create the 2014 MY baseline fleet Excel file for the Draft TAR. EPA created the
baseline using 2015 GHG certification data from EPA's Verify database. In the past, the data in
Verify did not include vehicle footprint data. Verify now includes a complete set of footprint
data for each vehicle; however, it is separate from the GHG information. Manufacturers are
required to report the numbers of each vehicle produced with a given footprint so the CO2 target
for that vehicle can be calculated. Separately, manufacturers are required to report the number of
each unique combination of vehicle, engine, transmission, and driveline (two-wheel drive vs.
four-wheel drive) that is produced along with its measured GHG information. The combination
of the two sets of data are used to determine if a manufacturer is complying with the GHG
standards. These two data sets, along with the valve train and engine cam information obtained
from Wards Automotive and the curb weight and power steering information from NHTSA's
2015 Volpe fleet file, were combined into a single data set and used to create the 2015 baseline.
Together, these sources inform the number of individual models, the volumes associated with
each model, the C02-reducing technologies with which the models are equipped, and the model's
current CO2 emissions performance. This process creates a complete baseline fleet that can then
1-4
-------
Baseline and Reference Vehicle Fleets
be used to project the reference fleet as well as other fleets used in exploration of various
scenarios in the OMEGA analysis.
Once a complete baseline fleet is created, the next step is to estimate the volumes and sales
mix of vehicles out to 2025, which we refer to as the reference fleet volumes (see Chapters
1.1.3.1 and 1.1.3.1.1 below). In addition to the information just described used to create the
2015 baseline fleet, EPA used volume projections from both EIA's Annual Energy Outlook
(AEO) 2016 and IHS-Polk, to generate the reference fleet volumes. Figure 1.2 shows the
process for combining the six data sets, with the result being the completed baseline, with
reference fleet projections.
MY2015
Baseline Fleet
Creation Process
2015 GHG
Emission
Certification Data
Wards Automotive
Engine Data
2015 GHG Foot
Print Certification
Data
2016 Unforced
AEO
IHS-Polk Forecast
2015 Volpe Fleet
File (Used for
Power Steering
and Curb Weight
Data)
Completed
MY2015 Baseline with
2022-2025 Reference
Fleet Projections
2022-2025
Reference Fleet
Creation
Figure 1.2 Process Flow for Creating the Baseline and Reference Fleet.
1-5
-------
Baseline and Reference Vehicle Fleets
EPA contracted with IHS-Polk to produce an updated long range forecast of volumes for the
future fleet for the Draft TAR, and is using these same data for this Proposed Determination. A
detailed discussion of the method used to project the future fleet volumes can be found in Section
1.1.3.1.1 of this chapter.
EPA used the previously mentioned data to populate input files for the OMEGA model. The
baseline Excel file is available in the docket.3 The Data Definitions tab of the Excel file has a
list of the columns of Data Tab. The column list has units, definition, and data source for each
item that was compiled for the baseline data.
Table 1.1 displays the engine technologies present in the MY2015 baseline fleet. As
previously described, this data was sourced primarily from the 2015 certification data,
supplemented by Wards' data on utilization of cam technology.
Table 1.1 MY2015 Engine Technology Penetration
Manufacturers
Vehicle Type
Turbo Charged
Super Charged
Single Overhead Cam
Dual Over Head Cam
Over Head Cam
Variable Valve Timing
Continuous Intake Only
Variable Valve Timing Discrete
Variable Valve Discrete Lift
Only
Variable Valve Lift and Timing
Discrete
Vehicles without Variable Valve
Timing or Lift
Cylinder Deactivation
Direct Injection
All
Both
16%
1%
6%
85%
8%
8%
71%
0%
18%
3%
11%
43%
All
Cars
18%
1%
7%
91%
1%
1%
74%
0%
23%
2%
2%
45%
All
Trucks
13%
1%
5%
77%
17%
17%
68%
0%
12%
3%
22%
40%
Aston Martin
Cars
0%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
0%
Aston Martin
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
BMW
Cars
95%
0%
2%
97%
0%
0%
1%
0%
95%
4%
0%
95%
BMW
Trucks
100%
0%
0%
100%
0%
0%
7%
0%
82%
11%
0%
100%
FCA
Cars
4%
1%
6%
86%
8%
8%
41%
0%
51%
1%
5%
2%
FCA
Trucks
3%
0%
1%
83%
16%
15%
74%
0%
8%
3%
16%
0%
Ferrari
Cars
32%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
100%
Ferrari
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Ford
Cars
33%
0%
0%
100%
0%
0%
98%
0%
0%
2%
0%
50%
Ford
Trucks
53%
0%
0%
100%
0%
0%
84%
0%
0%
16%
0%
53%
GM
Cars
21%
1%
0%
96%
4%
3%
78%
0%
18%
1%
3%
71%
GM
Trucks
3%
0%
0%
33%
67%
67%
33%
0%
0%
0%
66%
97%
Honda
Trucks
0%
0%
56%
44%
0%
0%
0%
0%
100%
0%
56%
52%
Honda
Cars
0%
0%
54%
46%
0%
0%
0%
0%
100%
0%
12%
55%
Hyundai/Kia
Trucks
0%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
100%
Hyundai/Kia
Cars
6%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
82%
JLR
Cars
16%
82%
0%
100%
0%
0%
97%
0%
3%
0%
0%
100%
JLR
Trucks
35%
65%
0%
100%
0%
0%
80%
0%
20%
0%
0%
100%
Lotus
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Lotus
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Mazda
Cars
0%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
98%
Mazda
Trucks
0%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
66%
McLaren
Cars
100%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
0%
1-6
-------
Baseline and Reference Vehicle Fleets
McLaren
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Mercedes
Cars
79%
0%
0%
99%
0%
0%
98%
0%
0%
2%
0%
95%
Mercedes
Trucks
49%
0%
2%
98%
0%
0%
94%
0%
0%
6%
0%
98%
Mitsubishi
Cars
4%
0%
61%
39%
0%
0%
100%
0%
0%
0%
0%
0%
Mitsubishi
Trucks
0%
0%
100%
0%
0%
0%
67%
6%
28%
0%
0%
0%
Nissan
Cars
3%
0%
0%
97%
0%
0%
91%
0%
7%
3%
0%
2%
Nissan
Trucks
0%
0%
0%
100%
0%
0%
96%
0%
4%
0%
0%
4%
Subaru
Cars
19%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
14%
Subaru
Trucks
3%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
3%
Tesla
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
0%
0%
Tesla
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Toyota
Cars
2%
0%
0%
100%
0%
0%
99%
0%
1%
1%
0%
3%
Toyota
Trucks
1%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
0%
Volkswagen
Cars
83%
4%
8%
92%
0%
0%
48%
0%
29%
23%
1%
91%
Volkswagen
Trucks
68%
30%
0%
100%
0%
0%
31%
0%
53%
16%
0%
100%
Volvo
Cars
100%
6%
0%
100%
0%
0%
100%
0%
0%
0%
0%
74%
Volvo
Trucks
90%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
0%
The data in Table 1.1 indicate that the MY2015 baseline fleet includes a significant amount of
engine technology that has been added by manufacturers. For example, BMW stands out as
having a significant number of gasoline turbocharged direct injection engines. Most of the fleet's
engines are using DOHC (dual overhead cam), and have discrete variable valve timing (VVT).
Over half of Honda's and GM"s Trucks all have engines with cylinder deactivation.
The data in Table 1.2 show the differences between the 2015 engine technology penetrations
and the 2008 engine technology penetrations. To increase fuel economy, manufacturers applied
considerable technology between 2008 and 2015. Manufacturers increased the use of direct
injection 38 percent on cars and 37 percent on trucks. Manufacturers also increased the use of
turbo chargers by 14 percent on cars and 12 percent on trucks.
1-7
-------
Baseline and Reference Vehicle Fleets
Table 1.2 Change (2015-2008) in Engine Technology Penetration
Xfi
¦d
¦d
£
6
o
£
0X1
•3 ^
2 it
0X1
2-
V
£/)
d
-4- «
0X1
s i
a
o
•-B
C3
a
Manufacture]
Vehicle Type
Turbo Charge
Super Charge
¦d
1
6
¦d
a
h
>
o
u
¦d
S3
U
a
s-
>
o
Variable Valve Ti
Continuous Int;
Only
Variable Valve Ti
Discrete
Variable Valve Di
Lift Only
Variable Valve Li
Timing Discre
Vehicles witho
Variable Valve Ti
or Lift
.j;
C3
'm.
"8
a
£?
Direct Injecti(
All
Both
13%
1%
-14%
23%
-9%
0%
51%
-9%
15%
-58%
4%
37%
All
Cars
14%
0%
-10%
18%
-8%
-8%
53%
-9%
19%
-55%
0%
38%
All
Trucks
12%
1%
-19%
30%
-12%
11%
50%
-9%
10%
-62%
10%
37%
Aston Martin
Cars
0%
0%
0%
0%
0%
0%
24%
0%
-24%
0%
0%
0%
Aston Martin
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
BMW
Cars
62%
-1%
-12%
11%
0%
-2%
-84%
0%
82%
4%
0%
62%
BMW
Trucks
95%
0%
0%
0%
0%
0%
-93%
0%
82%
11%
0%
94%
FCA
Cars
3%
1%
-15%
14%
0%
8%
-1%
0%
50%
-57%
0%
2%
FCA
Trucks
3%
0%
-38%
79%
-41%
15%
70%
0%
8%
-93%
11%
0%
Ferrari
Cars
32%
0%
0%
0%
0%
0%
29%
0%
-29%
0%
0%
100%
Ferrari
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Ford
Cars
33%
-1%
-15%
15%
0%
-4%
98%
0%
0%
-94%
0%
50%
Ford
Trucks
53%
0%
-65%
68%
-3%
-28%
83%
0%
0%
-55%
0%
53%
GM
Cars
20%
1%
0%
40%
-40%
-26%
47%
0%
18%
-39%
-1%
65%
GM
Trucks
3%
0%
0%
3%
-3%
61%
16%
0%
0%
-78%
26%
97%
Honda
Trucks
-4%
0%
-8%
8%
0%
0%
0%
-96%
96%
0%
56%
48%
Honda
Cars
0%
0%
-4%
4%
0%
0%
0%
-73%
73%
0%
1%
55%
Hyundai/Kia
Trucks
0%
0%
0%
0%
0%
0%
100%
0%
0%
-100%
0%
100%
Hyundai/Kia
Cars
6%
0%
0%
0%
0%
0%
100%
0%
0%
-100%
0%
82%
JLR
Cars
16%
82%
0%
0%
0%
0%
22%
0%
3%
-24%
0%
100%
JLR
Trucks
35%
44%
0%
0%
0%
0%
80%
0%
20%
-100%
0%
100%
Lotus
Cars
0%
-77%
0%
-100%
0%
0%
-100%
0%
0%
0%
0%
0%
Lotus
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Mazda
Cars
-11%
0%
0%
1%
0%
0%
93%
0%
0%
-93%
0%
86%
Mazda
Trucks
-24%
0%
-1%
1%
0%
0%
87%
0%
0%
-87%
0%
42%
McLaren
Cars
100%
0%
0%
100%
0%
0%
100%
0%
0%
0%
0%
0%
McLaren
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Mercedes
Cars
77%
0%
-54%
53%
0%
-72%
94%
0%
0%
-22%
0%
93%
Mercedes
Trucks
34%
-1%
-35%
35%
0%
-35%
11%
0%
0%
-42%
0%
83%
Mitsubishi
Cars
-2%
0%
-39%
39%
0%
-100%
100%
0%
0%
0%
0%
0%
Mitsubishi
Trucks
0%
0%
0%
0%
0%
-38%
67%
6%
28%
-62%
0%
0%
Nissan
Cars
3%
0%
0%
-3%
0%
0%
87%
0%
7%
-93%
0%
2%
Nissan
Trucks
0%
0%
0%
0%
0%
0%
96%
0%
4%
-100%
0%
4%
Subaru
Cars
5%
0%
-69%
69%
0%
0%
100%
-1%
0%
-99%
0%
14%
Subaru
Trucks
0%
0%
-70%
70%
0%
0%
100%
-5%
-23%
-73%
0%
3%
Tesla
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Tesla
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Toyota
Cars
2%
0%
0%
0%
0%
0%
70%
0%
1%
-71%
0%
-5%
Toyota
Trucks
1%
0%
0%
0%
0%
0%
39%
0%
0%
-39%
0%
-6%
Volkswagen
Cars
42%
4%
-71%
70%
0%
0%
-2%
0%
28%
-27%
1%
7%
Volkswagen
Trucks
63%
30%
0%
0%
0%
0%
19%
0%
-34%
15%
0%
0%
Volvo
Cars
51%
6%
0%
0%
0%
0%
0%
0%
0%
0%
0%
74%
Volvo
Trucks
90%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
1-8
-------
Baseline and Reference Vehicle Fleets
1.1.3.1 MY2015-Based MYs 2022-2025 Reference Fleet
This section provides further detail on the projection of the MY2015 baseline volumes into
the MYs 2022-2025 reference fleet. It also describes more of the data contained in the baseline
spreadsheet.
The reference fleet aims to reflect our latest projections about the market and fleet
characteristics during MYs 2022 to 2025. Fundamentally, constructing this fleet involved
projecting the MY2015 baseline fleet volumes out to MYs 2022-2025. It also included the
assumption that none of the vehicle models changed during this period. Such projections, of
course, have inherent uncertainties. However, as with the MY2008-based MY2022-2025
reference fleet used in the 2012 FRM, EPA relied on many sources of reputable information to
make these projections, and regards the projections as reasonable notwithstanding the
unavoidable uncertainties involved. No comments were received on EPA's use of IHS-Polk or
the process for developing the future volumes for vehicles.
1.1.3.1.1 On What Data are EPA's Reference Vehicle Fleet Volumes Based?
EPA has based the projection of total car and light truck sales on the U.S. Energy Information
Administration's (EIA) Annual Energy Outlook (AEO) 2016, which was the most recent
projection available at the time the Proposed Determination analysis was conducted. EIA's AEO
2016 also projects future energy production, consumption and prices.4 EIA issued the final
projection for AEO 2016 in July of 2016. As in the past analyses (MYs 2017-2025 rulemaking
and the Draft TAR), AEO 2016 used the EIA's National Energy Modeling System (NEMS) to
estimate the future relative market shares of passenger cars and light trucks. However, in
NEMS, EIA models the light-duty fleet to comply with CAFE and GHG standards from 2012
through 2025. In order to create a reference fleet absent the effect of the 2022-2025 GHG
standards, EPA only wanted NEMS to modify the fleet up to MY2021. Therefore, for the
current analysis, EPA requested that EIA develop a new projection of passenger car and light
truck sales shares by using NEMS to run scenarios from AEO 2016 cases (reference, high, and
low), holding post-2021 CAFE and GHG standards constant at MY2021 levels. EIA created this
special case for EPA.5 The output from the NEMS model that EIA supplied is consistent with
AEO 2016 since it has the same inputs as AEO 2016 with the exception of the standards being
held constant after MY2021. As with the comparable exercise for the 2012 FRM baseline fleet,
this case is referred to as the "Unforced Reference Case," and the values are shown below in
Table 1.3. The "unforced reference case" will be referred to as "unforced AEO 2016" for the rest
of this Technical Support Document (TSD). Table 1.4 shows the originally published AEO 2016
fleet projections. The total shift between cars and trucks is less than 1 percent of the total fleet
volume in the rulemaking years.
Table 1.3 AEO 2016 Unforced Reference Case Values used in the MY2015 Based Market Fleet Projection
Model Year
Cars
Trucks
Total Vehicles
2021
8,136,902
7,929,520
16,066,421
2022
8,222,542
7,812,037
16,034,579
2023
8,478,234
7,783,396
16,261,630
2024
8,583,611
7,719,964
16,303,575
2025
8,715,199
7,715,601
16,430,800
1-9
-------
Baseline and Reference Vehicle Fleets
Table 1.4 AEO 2016 Reference Case Values
Model Year
Cars
Trucks
Total Vehicles
2021
8,136,992
7,929,428
16,066,420
2022
8,222,617
7,811,960
16,034,578
2023
8,414,993
7,846,637
16,261,630
2024
8,467,865
7,835,709
16,303,575
2025
8,596,806
7,833,993
16,430,799
In 2021, car and light truck sales are projected to be 8.1 and 7.9 million units, respectively.
While the total sales level of 16 million units is similar to pre-2008 levels, the fraction of car
sales in 2021 and beyond is projected to be lower than in some of the previous AEO projections.
This is consistent with the results in the Draft TAR using AEO2015. See Draft TAR at p. 4-10.
In addition, sales for segments within both the car and truck markets have already been
changing, and this trend is expected to continue based on the projection from both IHS-Polk and
EIA. In order to reflect these changes in fleet makeup, EPA used a custom long-range forecast
purchased from IHS-Polk Automotive ("IHS-Polk").E IHS-Polk is a well-known industry
analysis source for forecasting and other data (such as vehicle registration data). For several
reasons, EPA decided to use the same forecast from IHS-Polk that was used for the Draft TAR
(which IHS-Polk created based on AEO2015) for the MY2015-based market forecast. First, as
just explained, AEO 2016's reference case is less than one percent different from AEO 2015 in
the rulemaking years. Second, IHS-Polk uses a bottom-up approach (e.g., looking at the number
of plants and capacity for specific engines, transmissions, vehicles, and registration data from
Polk) for their forecast, which we believe is a robust forecasting approach. Third, IHS-Polk
agreed to allow EPA to publish their entire forecast in the public domain (important for reasons
of transparency). Fourth, the IHS-Polk forecast covered the time frame of greatest relevance to
this analysis (the 2022-2025 model years). Fifth, it provided projections of vehicle sales both by
manufacturer and by market segment. Finally, it utilized market segments similar to those used
in the EPA emission certification program and fuel economy guide, such that EPA could include
only the segment types covered by the light-duty vehicle standards.
The custom forecast which IHS-Polk created for EPA covers model years 2012-2030. Since
EPA is using this forecast to generate the reference fleet volumes for this Proposed
Determination (i.e., the fleet expected to be sold absent any increases in the stringency of the
regulations after the 2021 model year), it is obviously important for the forecast to be
independent of any such stringency increases. IHS-Polk does not normally use the GHG (or
CAFE) standards as an input to their model, and EPA specified that they assume that the
standard stringencies would stay constant at 2021 levels in the 2022-2025 time frame for our
forecast. In addition, EPA specified that the IHS-Polk forecast use EIA's AEO 2015 fuel prices
and economic indicators to create the forecast.
EIHS bought CSM from which we previously purchased a long range forecast. IHS also purchased Polk automotive
which has registration data for all the vehicles in the United States.
1-10
-------
Baseline and Reference Vehicle Fleets
Table 1.5 shows the AEO 2015 and AEO 2016 fuel prices and differences. EPA believes that
the reference case fuel price (one to two cents per gallon) are close enough to justify continuing
to use IHS-Poik's forecast. IHS-Polk uses many additional inputs in their model, including GDP
growth, interest rates, the unemployment rate, and crude oil prices, to determine overall demand.
They then use vehicle size, price, and function to forecast with enough resolution to predict
brand and fleet segmentation. Additional details regarding the IHS-Polk forecast can be found in
a methodology description provided by IHS-Polk to EPA which is available in the docket (EPA-
HQ-OAR-2015-0827).
Table 1.5 AEO 2015 and AEO 2016 Reference Case Fuel Prices
Fuel Price (dollars/gal)
2021
2022
2023
2024
2025
2016 AEO Fuel Price Reference case
$ 3.19
$ 3.31
$ 3.43
$ 3.53
$ 3.64
2015 AEO Fuel Price Reference case
$ 3.21
$ 3.30
$ 3.41
$ 3.52
$ 3.63
Difference 2016-2015
-$ 0.02
$ 0.01
$ 0.02
$ 0.01
$ 0.01
EPA combined the IHS-Polk forecast with data from other sources to create the 2015 baseline
reference fleet projections. This process is discussed in the sections that follow. No commenters
challenged the validity of IHS-Polk's projections, or their use by EPA for this purpose.
1.1.3.1.2 How did EPA develop the MY2015 Baseline and MYs 2022-2025 Reference
Vehicle Fleet Volumes?
The process of producing the MY2015 baseline and 2022-2025 reference fleet volumes
involved combining the baseline fleet with the projection data described above. This complex
multi-step procedure is described in this section. The procedure is unchanged from the Draft
TAR.
1.1.3.1.3 How was the MY2015 Baseline Data Merged with the IHS-Polk Data?
EPA used the same method as in the Draft TAR for mapping certification vehicles to IHS-
Polk vehicles. See Draft TAR Chapter 4.1.2.1.4. Merging the 2015 baseline data with the 2022-
2025 IHS-Polk data required a thorough mapping of certification vehicles to IHS-Polk vehicles
by individual make and model. One challenge that EPA faced when determining a reference
case fleet was that the market segmentation of the sales data projected by IHS-Polk was similar
but different from the segmentation used in EPA's Verify database. In order to create a common
segmentation between the two databases, EPA performed a side-by-side comparison of each
vehicle model in both data sets, and created an additional "IHS-Polk Class" modifier in the
baseline spreadsheet to map the two data sets together. EPA then projected the reference fleet
volumes based on the "IHS-Polk Class."
The baseline data and reference fleet volumes are available to the public. The baseline Excel
spreadsheet that is available in the Docket is the result of the merged files.6 The spreadsheet
provides specific details on the sources and definitions for the data. The baseline Excel file
includes the following tabs: "Data," "Data Definition," "Platforms," "VehType," "Lookups,"
"Metrics," "Machine," "MarketFile," and "Safety." The "Data" tab contains the raw data. In the
"Data Definition" tab, each column is defined and its data source is named.
l-ll
-------
Baseline and Reference Vehicle Fleets
In the combined EPA certification and IHS-Polk data, all MY2015 vehicle models are
assumed to continue out to 2025, although their volumes change in proportion to IHS-Polk
projections. As explained in the following subsection, this methodology is used to provide
surrogate greenhouse gas performance data for new emerging models. As a result, new models
expected to be introduced within the 2015-2025 time frame are mapped to existing models.
Remapping the volumes from these new vehicles to the existing models via manufacturer
segments preserves the overall fleet volume. All MYs 2022-2025 vehicles are mapped from the
existing vehicles to the manufacturer's future segment volumes. The mappings are discussed in
the next section. Further discussion of this limitation is discussed below in Chapter 1.1.3.1.4.
The statistics of this fleet will be presented after the mapping since further volume modifications
were required.
1.1.3.1.4 How were the IHS-Polk Forecast and the Unforced AEO 2015 Forecast Used to
Project the Future Fleet Volumes?
The next step in EPA's generation of the reference fleet is one of the more complicated steps
to explain (although we note that EPA utilized a similar methodology in preparing both the
MY2008 baseline (for the 2022-2025 reference fleet) and an identical methodology creating the
MY2014 baseline fleet in the Draft TAR).
First, each vehicle in the 2015 data had an IHS-Polk segment mapped to it. Second, EPA
compared the breakdown of segment volumes by manufacturer between the IHS-Polk and 2015
data set. Third, a correction was applied for Class 2B vehicles in the IHS-Polk data. Fourth, the
individual manufacturer segment multipliers were created by year. And finally, the absolute
volumes of cars and trucks were normalized (set equal) to the total sales estimates of the
unforced AEO 2016. This final step is required to create a fleet forecast that reflects the official
government forecast for future vehicle sales. The unforced AEO 2016 forecast alone does not
have the necessary resolution, down to the vehicle segment level, for EPA to perform its
analysis. Therefore, EPA applies both the purchased forecast from IHS-Polk and the unforced
AEO 2016 forecast to create a complete fleet forecast.
The process started with mapping the IHS-Polk segments to each vehicle in the baseline data.
The mapping required determination of the IHS-Polk segment by lookup at each of the 2,653
baseline vehicles in the IHS-Polk forecast (which has only 617 vehicles since they do not
forecast powertrain or footprint differences), and labeling it in the "IHS-Polk Class" column of
the baseline data. The IHS-Polk data has 52 segments. Table 1.6 lists the IHS-Polk segments for
reference. Table 1.7 shows some of the Honda vehicles in the GHG data with their "IHS-Polk
Segment" identified.
1-12
-------
Baseline and Reference Vehicle Fleets
Table 1.6 List of IHS-Polk Segments
IHS-Polk Segments
Micro Non-premium Car
Compact Non-premium Car
Mid-Size Premium Van
Micro Non-premium Sporty
Compact Non-premium MPV
Mid-Size Super Premium Car
Mini Non-premium Car
Compact Non-premium Sporty
Mid-Size Super Premium Sporty
Mini Non-premium MPV
Compact Non-premium SUV
Mid-Size Super Premium SUV
Mini Non-premium Sporty
Compact Non-premium Van
Full-Size Non-premium Car
Mini Non-premium SUV
Compact Premium Car
Full-Size Non-premium Pickup
Mini Premium Car
Compact Premium Sporty
Full-Size Non-premium Sporty
Mini Premium Sporty
Compact Premium SUV
Full-Size Non-premium SUV
Subcompact Non-premium Car
Compact Super Premium Sporty
Full-Size Non-premium Van
Subcompact Non-premium MPV
Compact Super Premium SUV
Full-Size Premium Car
Subcompact Non-premium Pickup
Mid-Size Non-premium Car
Full-Size Premium Sporty
Subcompact Non-premium Sporty
Mid-Size Non-premium MPV
Full-Size Premium SUV
Subcompact Non-premium SUV
Mid-Size Non-premium Pickup
Full-Size Premium Van
Subcompact Premium Car
Mid-Size Non-premium Sporty
Full-Size Super Premium Car
Subcompact Premium MPV
Mid-Size Non-premium SUV
Full-Size Super Premium Sporty
Subcompact Premium Sporty
Mid-Size Premium Car
Full-Size Super Premium SUV
Subcompact Premium SUV
Mid-Size Premium Sporty
Subcompact Super Premium Sporty
Mid-Size Premium SUV
Table 1.7 Example of Honda Vehicles Being Mapped to Segments Based On the IHS-Polk Forecast
Manufacturer
Name Plate
Model
IHS-Polk Segment
Honda
Acura
ILX
Compact Premium Car
Honda
Acura
MDX
Mid-Size Premium SUV
Honda
Acura
RDX
Compact Premium SUV
Honda
Acura
RLX
Mid-Size Premium Car
Honda
Acura
TSX
Mid-Size Premium Car
Honda
Honda
ACCORD
Mid-Size Non-Premium Sporty
Honda
Honda
ACCORD
Mid-Size Non-Premium Car
Honda
Honda
CIVIC
Compact Non-Premium Car
Honda
Honda
CIVIC
Compact Non-Premium Sporty
Honda
Honda
FCX
Compact Non-Premium Car
Honda
Honda
CR-V
Compact Non-Premium SUV
Honda
Honda
CR-Z
Mini Non-Premium Sporty
Honda
Honda
CROSSTOUR
Mid-Size Non-Premium SUV
Honda
Honda
FIT
Subcompact Non-Premium Car
Honda
Honda
INSIGHT
Compact Non-Premium Car
Honda
Honda
ODYSSEY
Mid-Size Non-Premium MPV
Honda
Honda
PILOT
Mid-Size Non-Premium SUV
Honda
Honda
RIDGEUNE
Mid-Size Non-Premium Pickup Truck
In the next step, segment volume by manufacturer was compared between the baseline and
IHS-Polk data sets. This is necessary to determine if all of the segments a manufacturer will
produce in the future are currently represented by the 2015 certification data. The forecasts used
1-13
-------
Baseline and Reference Vehicle Fleets
in past rulemakings predicted very few new segments for manufacturers. The new forecast from
IHS-Polk projects that manufacturers will be entering more new segments (i.e., segments they
currently do not participate in) than in previous forecasts. This requires making sure a
manufacturer's volume in the new segment will be added to the volume of a manufacturer's
closest existing segment. The flow chart below (Figure 1.3) shows the process for determining
this "closest class." This process worked well for the majority of manufacturers.17 We believe
that this process of establishing "closest class" surrogates provides the best estimate of the
potential current performance of a given vehicle type and the technology that will be required to
meet the 2025 standards.
F The exceptions were Tesla and Aston Martin, both of which at the time operated only in the car segment and had
not yet entered the SUV segment.
1-14
-------
Baseline and Reference Vehicle Fleets
Is the Segmenl
Volume
x Premium?
-Yes-
-Yes-
No
No
fs there a Premium Segmenl
in the same size and
category?
-Yes-
Yes ~
No
No
^Isthere a Non-Premium^
Segment in same
category in the next size
large? z'
-Yes-
No
No
Is there a Premium Segment in the
same category in the next smaller
size?
No
No
-Yes-
-No-
/is there a Premium^
Segment in same
category in the next size
large? Z'
/Is there a Non-PremiurrN
Segment in the same size
and category?
Is there a Premium Segment in the"
same category in the next smaller
size? ,
/\s there a Premium Segment in\
same size category is there a similar
segment? (SUV, MPV, VAN are
\Similar; Car, Sporty are Similar)/
Is there a Premium Segment in\
same size category is there a similar
segment? (SUV, MPV, VAN are
\Similar; Car, Sporty are Similar)//
Determining the
correct segment to
move the volume
Move the volume
to the Premium
Segment
Move the volume
to the Premium
Larger Segment
Move the volume
to the Non-
Premium Segment
Move the volume
to the Non-
Premium Smaller
Segment
Move the volume
to the Premium
Smaler Segment
Move the volume
to the Non-
Premium Similar
Segment
Move the volume
to the Premium
Similar Segment
Move the volume
to the Non-
Premium Larger
Segment
Try the processes again assuming that the
original segment was one larger. If that does
not work try the process with one smaller.
Continue until starting from one larger or
smaller gives a segment.
Figure 1.3 Process Flow for Determining where Segment Volume Should Move
Table 1.8 shows Honda's segments with their volumes for both the baseline data and IHS-
Polk. Note that the segments "Compact Premium Sporty," "Mid-Size Non-premium Pickup,"
"Subcompact Non-premium SUV," and "Subcompact Premium SUV" do not exist in the
baseline data. The closest classes to those are "Compact Non-premium Car," "Mid-Size Non-
premium SUV," and "Compact Non-premium SUV."
1-15
-------
Baseline and Reference Vehicle Fleets
It is also important to note the difference between model year (MY) and calendar year (CY)
sales. MY sales can be shorter or longer than a full calendar year due to product launch and
change decisions made by a manufacturer. As a result, the MY sales0 can be less than or greater
than a respective calendar year sales. Table 1.8 provides a manufacturer example. For CY2015,
Honda introduced a new MY2016 Ridgeline pickup truck. Honda did not produce any pickup
trucks for MY2015 so it was necessary to move Honda's truck volume to their next closest class,
which is "Mid-Size Non-premium SUV." IHS shows that Honda built 515 "Mid-Size Non-
premium Pickups" for 2015, but none of those were MY2015 vehicles. In years that are close to
the baseline year, old models are exiting and new models are entering, which can be a source of
error. But as years progress, CY and MY volumes become the same in a forecast, since the
forecast neither adds nor deletes models. This allows EPA to use a CY forecast since we are
concerned with vehicles being built far enough in the future that CY and MY volumes are
approximately the same.
In comments on the Draft TAR, Honda commented that the Draft TAR figures for Honda
vehicles appeared to be in error. On examination, EPA discovered that Honda Civic Coupes had
been inadvertently classified as sedans, and Honda Civic Sedans had been classified as coupes.
This caused Civic models to show the wrong volumes. EPA corrected this error when creating
the 2015 baseline fleet for the current analysis.
Table 1.8 Example Honda 2015 Volumes by Segment from the IHS-Polk Forecast
Honda-Baseline Data
2015
MY
Honda-IHS-Polk Data
2015
CYH
2018 CY
Action
Compact Non-Premium Car
353,523
Compact Non-premium Car
337,423
358,046
Compact Non-Premium SUV
359,785
Compact Non-premium SUV
351,827
299,644
Compact Premium Car
11,093
Compact Premium Car
18,470
15,379
Compact Premium Sporty
0
797
Move Volume to Compact
Premium Car
Compact Premium SUV
50,387
Compact Premium SUV
49,882
40,642
Mid-Size Non-premium Car
354,428
Mid-Size Non-premium Car
349,921
338,848
Mid-Size Non-Premium MPV
129,988
Mid-Size Non-premium MPV
124,107
106,887
Mid-Size Non-premium Pickup
515
52,244
Move Volume to Mid-
Size Non-premium SUV
Mid-Size Non-Premium SUV
116,420
Mid-Size Non-premium SUV
141,796
144,182
Mid-Size Premium Car
68,727
Mid-Size Premium Car
50,380
44,876
Mid-Size Premium SUV
45,642
Mid-Size Premium SUV
59,742
53,249
Mini Non-Premium Sporty
3,814
Mini Non-premium Sporty
3,283
10,915
Subcompact Non-Premium Car
83,367
Subcompact Non-premium Car
60,246
54,988
Move Volume to Compact
Non-Premium Car
Subcompact Non-premium SUV
49,609
73,855
Move Volume to Compact
Non-Premium SUV
Subcompact Premium SUV
0
23,977
Move Volume to Compact
Non-Premium SUV
G Model Year sales may begin as early as January 1 of the previous calendar year (MY -1).
H 2015 Calendar Year can include both 2015 and 2016 Model Year vehicle sales if both are built in the calendar
year.
1-16
-------
Baseline and Reference Vehicle Fleets
A step that is related to the comparison step is the filtering of Class 3 vehicles from the IHS-
Polk forecast. IHS-Polk includes Class 2b and Class 3 vehicles (vans and large pickup trucks) in
its light-duty forecast. Class 2b vans with seating for multiple occupants are all appropriately
classified as MDPVs (Medium Duty Passenger Vehicles) and must be included in the forecast
since they are regulated under the light-duty GHG program. Class 2b large pickup trucks,
however, are not regulated under the light-duty GHG program but under the medium-duty and
heavy-duty fuel efficiency and GHG programs. See 76 FR 57120 and 81 FR 73729 (Oct. 25,
2016). These vehicles must therefore be removed from the forecast. Because IHS-Polk identifies
the Class 2b and Class 3 pickup trucks with the label 'HD,' it was readily apparent which Class
2b pickup trucks to filter from the forecast. Vans in the IHS-Polk forecast, on the other hand,
have both Class 2b and 3 and MDPVs in their totals, and so must have a correction factor
applied. This is accomplished by creating a multiplier for each manufacturer's Full-Size Non-
Premium Vans and applying it to each manufacturer's Full-Size Non-Premium Van volume
every model year in the IHS-Polk forecast; specifically, by taking a manufacturer's 2015 model
year Full-Size Non-Premium Van baseline volume and dividing by its 2015 calendar year Full-
Size Non-Premium Van IHS-Polk volume. Table 1.9 shows the volumes and the resulting
multiplier for FCA. Table 1.10 shows the 2025 IHS-Polk volume, the multiplier, and the result of
applying the multiplier to the original volume for FCA.
Table 1.9 Example Values Used to Determine the MDPV Multiplier for FCA
Manufacturer
NEW SEGMENT
IHS-Polk
2015
Volume
2015 GHG
Volume
MDPV
Multiplier
FCA
Full-Size Non-Premium Van
21,125
11,632
0.55
Table 1.10 Example Values Used to Determine FCA's 2025 Van Volume
Manufacturer
NEW SEGMENT
Original
2025
Volume
MDPV
Multiplier
2025
Volume
after
Multiplier
FCA
Full-Size Non-Premium Van
15,074
0.55
8,291
EPA next created individual manufacturer segment multipliers to be used with the individual
2015 vehicle volumes to create projections for the future fleet. The individual manufacturer
segment multipliers are created by dividing each year of the IHS-Polk forecast's individual
manufacturer segment volume by the manufacturer's individual segment volume, determined
using 2015 data. Table 1.11 shows the 2015 Volume, the 2025 IHS-Polk Full-Size Non-
Premium Van volume after Class 2b vehicles were removed, and the individual manufacturer
volume for Full-Size Non-Premium Van. The multiplier is the result of dividing the 2025
volume by the 2015 volume.
1-17
-------
Baseline and Reference Vehicle Fleets
Table 1.11 Example Values Used to Determine FCA 2025 Individual Full-Size Non-Premium Van Multiplier
Manufacturer
IHS-Polk
Segment
2015 GHG
Volume
2025 Volume after
Multiplier
Fiat/Chrysler Individual Full-
Size Non-Premium Van
Multiplier for 2025
FCA
Full-Size Non-
Premium Van
15,074
8,291
71.4%
Now that the individual manufacturer segment multipliers have been calculated, they can be
applied to each vehicle in the 2015 data. The segment multipliers are applied by multiplying the
2015 volume for a vehicle by the multiplier for its manufacturer and segment. Table 1.12 shows
the 2015 volumes, the individual manufacturer segment multipliers, and the result of multiplying
the multiplier and the volume for 2025 project volumes for many of FCA's Full-Size Non-
Premium Vans.
Table 1.12 Example Applying the Individual Full-Size Non-Premium Van Multiplier for FCA
Manufacturer
Model
IHS-Polk Segment
2015 GHG
Volume
Fiat/Chrysler
Individual Full-
Size Non-
Premium Van
Multiplier for
2025
2025 Project
Volume Before
AEO
Normalization
FCA
Cargo Van A
Full-Size Non-Premium Van
208
71.4%
148
FCA
Cargo Van B
Full-Size Non-Premium Van
5,712
71.4%
4,076
Normalizing to unforced AEO 2016 forecast for cars and trucks must be done once the
individual manufacturer segment multipliers have been applied to all vehicles across every year
(2011-2025) of the IHS-Polk forecast. In order to normalize a year, the number of trucks and the
number of cars produced must be determined. Then, the truck and car totals from the unforced
AEO 2016 are used to determine a normalizing multiplier. Table 1.13 shows the 2025 car and
truck totals before normalization, the unforced AEO 2016 car and truck totals in 2025, and the
multipliers, which are the result of dividing the unforced AEO 2016 totals by totals before
normalization.
Table 1.13 Example Unforced AEO 2016 Truck and Car Multipliers in MY2025
Vehicle Type
2025 Total Before
Normalization
2025 Total from AEO 2016
2025
Normalizing
Multiplier
Cars
9,889,511
8,715,199
88%
Trucks
5,838,907
7,715,600
132%
The final step in creating the reference volumes is applying the unforced AEO multipliers.
The AEO multipliers are applied by car/truck type. Table 1.14 shows the normalized volume,
the unforced AEO 2016 truck multiplier for MY2025, and the final resulting volume for a
number of FCA Full-Size Non-Premium Vans.
1-18
-------
Baseline and Reference Vehicle Fleets
Table 1.14 Example Applying the Unforced AEO Truck Multiplier to FCA Full-Size Non-Premium Vans
Manufacturer
Model
C/T Type
2025 Project
Volume Before
Unforced AEO
2016
Normalization
Unforced AEO
2016 Truck
Multiplier for
2025
2025 Project
Volume with
Unforced AEO
2016
Normalization
FCA
Cargo Van A
Truck
148
132%
196
FCA
Cargo Van B
Truck
4,076
132%
5,385
1.1.3.2 What Are the Sales Volumes and Characteristics of the MY2015 Based Reference
Fleet?
Table 1.15 and Table 1.16 below contain the sales volumes that result from the process above
forMY2015 and MYs 2021-2025. In Table 1.15, "SmallPickup" is zero. The only manufacturer
that produced a small pickup in recent years was Honda, and Honda did not build a MY2015
Ridgeline.
Table 1.15 Vehicle Segment Volumes
Segment
Actual and Projected Sales Volume
2015
2021
2022
2023
2024
2025
SubCmpctAuto
990,135
879,310
907,553
967,714
967,714
973,176
CompactAuto
2,564,949
2,395,133
2,382,352
2,466,062
2,466,062
2,566,388
MidSizeAuto
3,905,449
2,860,094
2,916,546
2,980,777
2,980,777
3,073,007
LargeAuto
523,225
538,526
550,746
568,332
568,332
586,843
SmallPickup
-
-
-
-
-
-
LargePickup
1,786,223
1,875,652
1,815,030
1,815,163
1,815,163
1,843,621
SmallSuv
2,184,788
2,696,071
2,664,266
2,691,022
2,691,022
2,689,904
MidSizeSuv
2,204,122
2,159,523
2,132,377
2,153,164
2,153,164
2,133,971
LargeSuv
1,088,051
1,427,186
1,392,192
1,387,494
1,387,494
1,373,818
ExtraLargeSuv
920,239
717,693
728,207
684,299
684,299
662,595
MiniVan
548,342
494,165
518,402
519,562
519,562
497,794
Cargo Van
20,876
23,068
26,907
28,042
28,042
29,683
Table 1.16 Car and Truck Volumes
Vehicle Type
Actual and Projected Sales Volume
2015
2021
2022
2023
2024
2025
Cars
9,597,936
8,136,902
8,222,542
8,478,234
8,583,611
8,715,199
Trucks
7,138,461
7,929,520
7,812,037
7,783,396
7,719,964
7,715,601
Cars and Trucks
16,736,397
16,066,421
16,034,579
16,261,630
16,303,575
16,430,800
Table 1.17 lists the sales volumes by manufacturer and C/T type for MY2015 and MY2021-
2025. Lotus is a small volume manufacturer and chose not to build MY2015 vehicles.
1-19
-------
Baseline and Reference Vehicle Fleets
Table 1.17 Car and Truck Definition Manufacturer Volumes
Manufacturers
C/T
Type
2015
Baseline
Sales
2021
Projected
Volume
2022
Projected
Volume
2023
Projected
Volume
2024
Projected
Volume
2025
Projected
Volume
All
Both
16,736,397
16,066,421
16,034,579
16,261,630
16,303,575
16,430,800
All
Cars
9,597,936
8,136,902
8,222,542
8,478,234
8,583,611
8,715,199
All
Trucks
7,138,461
7,929,520
7,812,037
7,783,396
7,719,964
7,715,601
Aston Martin*
Cars
1,119
1,384
1,320
1,325
1,290
1,422
Aston Martin*
Trucks
-
-
-
-
-
-
BMW
Cars
338,704
317,648
332,266
350,651
357,144
348,293
BMW
Trucks
87,135
115,780
110,687
107,555
105,521
104,931
FCA
Cars
769,687
535,600
554,402
552,943
547,469
558,331
FCA
Trucks
1,416,487
1,270,099
1,261,444
1,267,012
1,256,467
1,275,022
Ferrari*
Cars
2,645
2,999
6,491
7,904
8,519
9,190
Ferrari*
Trucks
-
-
-
-
-
-
Ford
Cars
888,604
831,609
829,433
818,078
800,638
833,326
Ford
Trucks
972,891
1,256,726
1,243,115
1,226,286
1,204,489
1,182,848
GM
Cars
1,331,442
1,154,344
1,162,751
1,242,812
1,241,036
1,239,682
GM
Trucks
1,525,017
1,258,030
1,261,455
1,210,912
1,196,960
1,199,874
Honda
Cars
1,020,310
819,658
839,422
865,428
895,193
883,518
Honda
Trucks
556,864
861,851
857,929
869,110
853,349
836,097
Hyundai/Kia
Cars
1,228,399
1,129,153
1,138,735
1,157,423
1,168,074
1,185,878
Hyundai/Kia
Trucks
91,058
227,750
217,616
227,780
226,399
227,669
JLR
Cars
15,600
22,932
24,262
25,440
25,156
24,494
JLR
Trucks
54,435
102,505
100,010
96,409
95,196
94,350
Lotus*
Cars
-
-
-
-
-
-
Lotus*
Trucks
-
-
-
-
-
-
Mazda
Cars
207,100
212,725
212,269
210,091
217,939
225,981
Mazda
Trucks
78,793
129,877
135,392
139,357
135,675
136,192
McLaren*
Cars
625
941
1,045
1,199
1,372
1,336
McLaren*
Trucks
-
-
-
-
-
-
Mercedes
Cars
231,899
218,508
224,049
237,549
238,973
238,811
Mercedes
Trucks
123,727
178,096
172,461
168,875
167,255
166,733
Mitsubishi
Cars
91,822
47,775
50,602
55,964
60,376
61,002
Mitsubishi
Trucks
39,366
35,229
34,592
36,127
35,425
39,452
Nissan
Cars
1,216,392
820,204
816,918
861,832
864,924
895,430
Nissan
Trucks
481,583
579,939
563,728
544,882
540,234
551,676
Subaru
Cars
175,352
140,987
149,303
147,953
148,723
152,485
Subaru
Trucks
447,383
531,411
506,265
540,938
539,008
555,249
Tesla
Cars
24,322
90,547
88,844
99,390
102,654
109,459
Tesla
Trucks
-
-
-
-
-
-
Toyota
Cars
1,524,190
1,203,844
1,206,329
1,233,020
1,280,689
1,299,472
Toyota
Trucks
1,127,056
1,071,915
1,047,556
1,056,695
1,058,452
1,031,420
Volkswagen
Cars
487,108
541,520
540,983
567,019
581,817
599,186
Volkswagen
Trucks
112,382
261,463
249,199
244,025
259,817
265,166
Volvo
Cars
42,616
44,523
43,117
42,216
41,626
47,901
Volvo
Trucks
24,284
48,849
50,589
47,432
45,717
48,921
1-20
-------
Baseline and Reference Vehicle Fleets
*Note: These manufacturers are shown here for reference but are not in the analysis in Chapter 5 or considered in the
ZEV sales that are part of the analysis fleet as discussed in Chapter 1.2.1.
Table 1.18 shows how the change in fleet makeup may affect the footprint distributions over
time. The resulting data indicate that the average vehicle footprint would not change
significantly between 2015 and 2025.
Table 1.18 Production Weighted Foot Print Mean
Model Year
Average Footprint of all Vehicles
Average Footprint Cars
Average Footprint Trucks
2015
49.3
46.1
53.7
2017
49.8
46.0
53.0
2018
49.7
46.1
53.0
2019
49.7
46.1
53.0
2020
49.5
46.1
53.0
2021
49.5
46.1
53.0
2022
49.5
46.1
53.0
2023
49.4
46.0
52.9
2024
49.3
46.0
52.9
2025
49.3
46.1
53.0
Table 1.19 shows the projected changes in number of engine cylinders over the model years
of the rule. The current assumptions indicate that the number of cylinders would shrink slightly
between 2015 and 2019 for trucks and then remain relatively constant over the 2019-2025 time
frame, with only a very slight shift to 4 cylinders in trucks (possibly due to an increase in the
number of small SUVs).
Table 1.19 Percentages of 4,6, and 8 Cylinder Engines by Model Year
Trucks
Cars
Model
Year
4
Cylinders
6
Cylinders
8
Cylinders
4
Cylinders
6
Cylinders
8
Cylinders
2015
28.6%
50.3%
21.1%
81.1%
16.2%
2.7%
2017
31.5%
50.7%
17.8%
81.3%
15.8%
2.9%
2018
32.5%
49.7%
17.8%
80.7%
16.4%
2.9%
2019
33.0%
49.2%
17.8%
80.8%
16.4%
2.9%
2020
33.1%
49.1%
17.8%
81.0%
16.1%
2.9%
2021
33.2%
49.4%
17.5%
81.0%
16.0%
3.0%
2022
33.0%
49.7%
17.3%
80.7%
16.2%
3.1%
2023
33.6%
49.4%
17.0%
80.8%
16.2%
3.0%
2024
33.7%
49.3%
17.0%
80.9%
16.1%
3.0%
2025
33.8%
49.0%
17.2%
80.9%
16.1%
3.0%
1-21
-------
Baseline and Reference Vehicle Fleets
1.1.3.3 What Are the Differences in the Sales Volumes and Characteristics of the MY2008-
Based (FRM) and the MY2015-Based Reference Fleets?
This section compares some of the differences between the MY2008-based reference fleet
used in previous analyses and the MY2015-based reference fleet used in the current analysis.
The 2008 fleet projection is based on several sources: MY2008 certification data, a long range
forecast provided by CSM, and interim unforced AEO 2011. The 2015 fleet projection is based
on MY2015 certification data, a long-range forecast provided by IHS-Polk Automotive, and the
unforced AEO 2016, as described earlier in this chapter. All tables in this section show the
differences between the MY2008 and MY2015 fleets.
Table 1.20, Table 1.21, and Table 1.22 below show the sales volume differences between the
two fleets, calculated by subtracting the MY2008-based fleet projection from the MY2015-based
fleet projection. The sales in MY2015 were significantly higher (by 3,025,250 vehicles) than in
MY2008, when sales may have been impacted by an economic recession. MY2015 volumes are
also higher than forecast at the time of the FRM.
For 2015, there is an increase in the number of compact and midsize autos, large trucks, and
all SUVs. For 2025, one of the biggest differences between the two forecasts is the number of
cars, which in part seem to be replaced by small and midsize SUVs. The shift from cars to
trucks is due to application of the unforced AEO 2016 data while the shifts within segments
reflect the data from the IHS-Polk forecast.
Table 1.20 Differences in Vehicle Segment Volumes
Reference Class
Segment
Actual Sales
Volume
Difference in Projected Sales Volume
2015-2008
2021
2022
2023
2024
2025
SubCmpctAuto
-306,978
-1,657,574
-1,688,249
-1,660,379
-1,734,262
-1,808,385
CompactAuto
603,852
-107,830
-191,482
-147,905
-259,283
-257,112
MidSizeAuto
813,354
-573,597
-623,142
-702,899
-761,429
-732,487
LargeAuto
-42,851
152,870
186,900
199,948
206,089
211,829
SmallPickup
-177,497
-150,123
-147,138
-151,315
-154,627
-154,838
LargePickup
221,780
522,791
480,262
527,579
556,971
596,868
SmallSuv
575,990
1,143,916
1,107,175
1,147,906
1,117,851
1,101,240
MidSizeSuv
912,792
722,167
692,742
715,745
699,660
671,233
LargeSuv
437,341
363,099
310,474
282,426
224,914
182,174
ExtraLargeSuv
171,164
25,363
7,251
-64,288
-50,488
-78,501
MiniVan
-171,187
-351,891
-331,269
-329,887
-311,176
-341,658
CargoVan
-12,508
-70,492
-65,216
-64,878
-58,841
-58,889
1-22
-------
Baseline and Reference Vehicle Fleets
Table 1.21 Differences in Actual and Projected Sales Volumes between MY2015 and MY2008 fleets
C/T Type
Difference in
Actual Sales
Volume
Difference in Projected Sales Volume
2015 - 2008
2021
2022
2023
2024
2025
Cars
1,468,413
-2,251,245
-2,393,927
-2,368,096
-2,551,279
-2,700,077
Trucks
1,556,837
2,269,945
2,132,236
2,120,147
2,068,602
2,031,550
Cars and Trucks
3,025,250
18,700
-261,691
-247,948
-482,677
-668,527
Table 1.22 below shows the differences in sales volumes by manufacturer and car/truck type
between the MY2008-based fleet and the MY2015-based fleet. The manufacturers with the next
largest increases in sales in MY2015 (from MY2008) are FCA, Ford, Hyundai/Kia, Nissan,
Subaru, and Toyota. The manufacturers with a net decrease in sales in MY2015 (from MY2008)
are Aston Martin, Honda, GM, Mazda, Mitsubishi, and Volvo. The manufacturers with the next
largest increases in sales in MY2025 are FCA, Subaru, and Tesla. The manufacturers forecast to
have a significant net decrease in sales in MY2025 are GM, Mazda, and Volvo. Table 1.22 also
shows a projected decrease in the total vehicle market in MY2025 by 668,527 vehicles.
Table 1.22 Differences in Sales Volumes by Manufacturer and Car/Truck Type between MY2008-based and
MY2015-based fleets
Manufacturers
Segment
Type
2015-2008
Difference
in Sales
2021
Difference
in Volume
2022
Difference
in Volume
2023
Difference
in Volume
2024
Difference
in Volume
2025
Difference
in Volume
All
Both
3,025,250
18,700
-261,691
-247,948
-482,677
-668,527
All
Cars
1,468,413
-2,251,245
-2,393,927
-2,368,096
-2,551,279
-2,700,077
All
Trucks
1,556,837
2,269,945
2,132,236
2,120,147
2,068,602
2,031,550
Aston Martin
Cars
-251
326
271
284
149
240
Aston Martin
Trucks
0
0
0
0
0
0
BMW
Cars
46,908
-41,450
-27,768
-9,911
-31,050
-56,963
BMW
Trucks
25,811
-12,944
-18,211
-19,966
-41,005
-40,478
FCA
Cars
66,529
114,587
130,229
129,061
121,452
121,852
FCA
Trucks
459,695
921,486
898,435
905,949
911,505
943,261
Ferrari
Cars
1,195
-4,059
-647
677
1,078
1,532
Ferrari
Trucks
0
0
0
0
0
0
Ford
Cars
-68,095
-570,009
-585,788
-656,719
-703,032
-706,784
Ford
Trucks
158,697
542,545
528,849
526,281
515,635
498,372
GM
Cars
-255,949
-409,932
-415,805
-363,683
-395,769
-434,253
GM
Trucks
17,220
-271,990
-246,198
-285,906
-296,637
-324,134
Honda
Cars
13,671
-379,222
-398,082
-400,136
-412,658
-456,803
Honda
Trucks
51,724
325,935
318,695
332,212
316,355
278,400
Hyundai/Kia
Cars
668,550
184,479
171,669
180,369
158,483
145,845
Hyundai/Kia
Trucks
-21,572
-24,148
-34,572
-29,097
-35,812
-38,120
JLR
Cars
6,004
-35,745
-35,087
-35,200
-38,572
-40,923
JLR
Trucks
-1,149
44,352
41,420
37,543
37,215
37,544
Lotus
Cars
-252
-278
-290
-299
-308
-316
1-23
-------
Baseline and Reference Vehicle Fleets
Lotus
Trucks
0
0
0
0
0
0
Mazda
Cars
-39,561
-62,015
-68,882
-86,818
-82,676
-80,823
Mazda
Trucks
22,908
70,650
75,085
77,391
73,705
74,824
McLaren
Cars
625
941
1,045
1,199
1,372
1,336
McLaren
Trucks
0
0
0
0
0
0
Mercedes
Cars
23,704
-81,870
-80,689
-74,958
-93,364
-101,907
Mercedes
Trucks
44,592
78,647
71,526
63,561
60,171
65,666
Mitsubishi
Cars
6,464
-18,076
-16,659
-11,716
-10,352
-12,303
Mitsubishi
Trucks
23,995
-80
-635
657
-577
3,066
Nissan
Cars
498,523
-92,425
-120,529
-92,508
-117,848
-119,345
Nissan
Trucks
176,037
171,910
151,844
127,761
118,018
125,221
Subaru
Cars
59,317
-89,794
-89,310
-93,659
-99,560
-104,486
Subaru
Trucks
364,837
458,638
433,528
467,917
464,865
480,528
Tesla
Cars
23,522
61,924
60,475
71,240
71,792
77,485
Tesla
Trucks
0
0
0
0
0
0
Toyota
Cars
266,609
-694,059
-773,704
-797,805
-793,547
-802,220
Toyota
Trucks
175,920
-143,624
-187,497
-168,285
-149,561
-178,596
Volkswagen
Cars
173,933
-86,364
-94,983
-72,890
-69,314
-78,034
Volkswagen
Trucks
66,586
101,487
91,064
78,727
91,469
99,663
Volvo
Cars
-23,033
-48,203
-49,395
-54,624
-57,555
-53,206
Volvo
Trucks
-8,464
7,081
8,903
5,401
3,256
6,332
Table 1.23 shows the difference in footprint distributions between the MY2015-based fleet
projection and the MY2008-based fleet projection. The differences between MYs 2015 and
2008 are small, resulting from the manufacturers' projected product mix in those model years.
MY2025 shows an increase in average car footprints. This is due to the significant decrease in
subcompact cars forecast in the MY2015-based fleet projection. Truck footprints decrease
slightly due to the increase in small SUVs. Because the total numbers of cars and trucks differs,
production weighting can affect the average for the whole fleet as compared to the averages for
cars and trucks. This can cause the result to appear counterintuitive when taking the difference
of the averages.
1-24
-------
Baseline and Reference Vehicle Fleets
Table 1.23 Difference in Footprint Distributions between MY2015-based and MY2008-based Fleet
Projections
Model
Year
Difference in Average Footprint
of all Vehicles
Difference in Average
Footprint Cars
Difference in Average Footprint
Trucks
2015-2008
49.3- 48.9 = 0.4
46.1-45.4 = 0.7
53.7-54.0 = -0.3
2017
49.8-48.3 = 1.5
46.0-44.9 = 1.1
53.0-53.8 = -0.8
2018
49.7-48.1= 1.6
46.1-44.9 = 1.2
53.0-53.7 = -0.7
2019
49.7-48.0= 1.7
46.1-44.9 = 1.2
53.0-53.6 = -0.6
2020
49.5-48.0 = 1.5
46.1-44.9 = 1.2
53.0-53.7 = -0.7
2021
49.5-48.0 = 1.5
46.1-44.9 = 1.2
53.0-53.6 = -0.6
2022
49.5-47.9 = 1.6
46.1-44.9 = 1.2
53.0-53.6 = -0.6
2023
49.4-47.9 = 1.5
46.0-44.9 = 1.1
52.9-53.5 = -0.6
2024
49.3-47.7 = 1.6
46.0-44.9 = 1.1
52.9-53.4 = -0.5
2025
49.3-47.7 = 1.6
46.1-44.9 = 1.2
53.0-53.3 = -0.3
Table 1.24 shows the difference in the distribution of the number of engine cylinders between
the MY2015-based fleet and the MY2008-based fleet. The MY2015 fleet includes fewer
vehicles with 6- and 8-cylinder engines than the MY2008fleet. The presence of fewer 6- and 8-
cylinder vehicles in the baseline fleet, along with vehicle mix changes, results in more 4-cylinder
engines in trucks and cars by 2025.
Table 1.24 Differences in Percentages of 4,6 and 8 Cylinder Engines by Model Year
Trucks
Ca rs
Model
Year
4 Cylinders
6 Cylinders
8 Cylinders
4 Cylinders
6 Cylinders
8 Cylinders
2015-2008
18.1%
-5.2%
-12.8%
23.4%
-20.7%
-2.7%
2017
20.4%
-12.5%
-8.0%
19.3%
-17.1%
-2.1%
2018
21.7%
-14.3%
-7.4%
18.6%
-16.5%
-2.1%
2019
22.4%
-15.7%
-6.7%
18.7%
-16.5%
-2.1%
2020
22.6%
-15.9%
-6.7%
19.2%
-17.1%
-2.2%
2021
22.7%
-16.5%
-6.3%
18.9%
-17.0%
-1.9%
2022
22.6%
-16.5%
-6.0%
18.1%
-16.4%
-1.7%
2023
23.2%
-17.8%
-5.3%
18.3%
-16.6%
-1.8%
2024
23.0%
-18.3%
-4.7%
18.4%
-16.6%
-1.8%
2025
23.1%
-18.8%
-4.3%
18.3%
-16.5%
-1.8%
1.1.3.4 What Are the Differences in the Sales Volumes and Characteristics of the EPA
MY2014-Based (Draft TAR) and the MY2015-Based Reference Fleets?
This section compares some of the differences between the MY2014-based reference fleet
(used in the Draft TAR analysis) and the MY2015-based reference fleet used in the current
1-25
-------
Baseline and Reference Vehicle Fleets
analysis. As described earlier in this chapter, the MY2014-based reference fleet projection is
based on several sources: MY2014 certification data, a long-range forecast provided by IHS-
Polk Automotive, and the unforced AEO 2015. The MY2015-based reference fleet projection is
based on MY2015 certification data, a long-range forecast provided by IHS-Polk Automotive
(the same source used to create the 2016 fleet volumes), and the unforced AEO 2016. All tables
in this section show the differences between the MY2014-based and MY2015-based fleets.
Table 1.25, Table 1.26, and Table 1.27 below list the sales volume differences between the
two fleets, calculated by subtracting the MY2014-based fleet projection from the MY2015-based
fleet projection. The sales in MY2015 were significantly higher (by 1,218,062 vehicles) than in
MY2014. This suggests that automotive sales remain strong as advanced fuel-saving
technologies have entered the market in response to the GHG/fuel economy standards, and that
sales have increased even as the standards' stringency increased. In addition, this comparison
demonstrates the need to use final sales year data to construct the baseline fleet, rather than mid-
year fleet projections. The mid-year data provided by vehicle manufacturers to NHTSA did not
reflect the actual substantial increase in sales that was seen in MY2015.
For MY2015, there is a small increase in the number of compact and midsize autos, and all
SUVs (except the largest). For MY2025, the differences between the two forecasts is very small
when compared to the size of the overall market, with the largest change being for pickup trucks
at -246,276, which is only 1.5 percent of the total market and 3 percent of the truck market.
Table 1.25 Vehicle Segment Volume Differences
Reference Class
Segment
Actual Sales
Volume
Difference in Projected Sales Volume
(2015-2014)
2015-2014
2021
2022
2023
2024
2025
SubCmpctAuto
-41,437
130,356
141,833
154,668
151,677
136,131
CompactAuto
19,228
-68,469
-51,745
-4,512
-81,261
-24,442
MidSizeAuto
366,984
105,689
134,839
186,828
136,170
156,879
LargeAuto
44,008
125,647
127,693
147,562
152,996
155,953
SmallPickup
-12,143
-15,227
-14,222
-16,067
-15,908
-16,123
LargePickup
-130,838
-235,294
-246,707
-233,482
-235,964
-246,276
SmallSuv
172,388
88,569
97,330
128,525
107,569
87,439
MidSizeSuv
656,145
141,260
127,151
121,146
141,228
106,402
LargeSuv
34,554
-20,285
-24,211
-16,511
-9,111
-20,463
Extra LargeSuv
255,614
-51,336
-58,327
-52,516
-51,981
-55,367
MiniVan
-54,352
-59,725
-61,542
-63,043
-56,681
-78,215
CargoVan
-47,737
-57,663
-53,690
-58,918
-60,858
-63,169
1-26
-------
Baseline and Reference Vehicle Fleets
Table 1.26 Differences in Actual and Projected Sales Volumes between MY2015 and MY2014 fleets
C/T Type
Difference in Actual Sales Volume
Difference in Projected Sales Volume
2015 - 2014
2021
2022
2023
2024
2025
Cars
391,150
526
78,901
208,341
173,114
117,786
Trucks
826,913
-30,693
-72,677
-36,652
-78,788
-111,998
Cars and Trucks
1,218,062
-30,167
6,225
171,689
94,326
5,788
Table 1.27 below contains the differences in sales volumes by manufacturer and C/T type
between the 2014 MY based fleet and the 2015 MY based fleet. The manufacturers with the
next largest increases in sales in 2015 MY (from 2014) are FCA cars, GM trucks, Honda cars,
Hyundai/Kia cars, Nissan cars and trucks, and Toyota cars and trucks. The manufacturers with a
net decrease in sales in 2015 (from 2014) are Aston Martin, Ford, JLR, Mazda, and Mercedes.
The differences in forecasted volumes are relatively small.
1-27
-------
Baseline and Reference Vehicle Fleets
Table 1.27 Differences in Sales Volumes by Manufacturer and Car/Truck Type between MY2014-based and
MY2015-based fleets
Manufacturers
Segment Type
2015-2014
2021
2022
2023
2024
2025
Difference
Difference
Difference
Difference
Difference
Difference
in Sales
in Volume
in Volume
in Volume
in Volume
in Volume
All
Both
1,218,062
-30,167
6,225
171,689
94,326
5,788
All
Cars
391,150
526
78,901
208,341
173,114
117,786
All
Trucks
826,913
-30,693
-72,677
-36,652
-78,788
-111,998
Aston Martin
Cars
-153
60
68
87
77
77
Aston Martin
Trucks
0
0
0
0
0
0
BMW
Cars
41,316
18,668
22,079
28,049
26,191
24,070
BMW
Trucks
5,197
5,411
4,499
4,283
3,766
3,294
FCA
Cars
121,310
-72,065
-68,327
-57,334
-60,510
-64,580
FCA
Trucks
-29,878
-174,041
-174,870
-175,572
-181,415
-195,077
Ferrari
Cars
344
744
4,257
5,543
5,914
6,455
Ferrari
Trucks
0
0
0
0
0
0
Ford
Cars
-370,128
-103,403
-93,708
-81,800
-83,955
-96,358
Ford
Trucks
-102,611
-102,958
-111,309
-103,414
-105,913
-106,382
GM
Cars
-225,259
-57,491
-47,791
-28,774
-34,774
-48,048
GM
Trucks
360,407
-66,520
-74,663
-68,675
-75,402
-80,294
Honda
Cars
151,973
25,092
34,239
47,588
44,120
38,803
Honda
Trucks
-20,964
110,081
104,487
107,609
101,567
97,991
Hyundai/Kia
Cars
210,858
19,337
30,167
42,398
36,275
31,198
Hyundai/Kia
Trucks
23,860
68,341
65,663
74,274
71,742
70,503
JLR
Cars
3,277
-1,229
-969
-575
-699
-750
JLR
Trucks
-798
-984
-1,062
-485
-998
-1,104
Lotus
Cars
-280
-234
-232
-231
-232
-233
Lotus
Trucks
0
0
0
0
0
0
Mazda
Cars
-10,233
-36,292
-35,287
-29,957
-30,241
-33,496
Mazda
Trucks
-33
21,875
21,890
23,075
21,806
21,674
McLaren
Cars
346
41
54
79
82
73
McLaren
Trucks
0
0
0
0
0
0
Mercedes
Cars
-46,227
-8,095
-5,958
-2,854
-4,509
-6,530
Mercedes
Trucks
31,415
18,217
16,872
16,834
15,879
15,534
Mitsubishi
Cars
31,143
679
1,261
2,177
2,052
1,675
Mitsubishi
Trucks
9,538
5,904
5,660
6,102
5,892
6,326
Nissan
Cars
280,397
52,328
58,513
75,317
69,960
67,479
Nissan
Trucks
91,944
20,248
18,265
15,072
10,560
9,668
Subaru
Cars
66,274
6,089
7,746
9,749
8,872
8,298
Subaru
Trucks
90,565
58,299
53,318
58,105
55,433
56,031
Tesla
Cars
6,531
3,911
4,609
6,549
6,124
5,957
Tesla
Trucks
0
0
0
0
0
0
Toyota
Cars
103,549
71,759
82,501
100,317
96,860
92,042
Toyota
Trucks
354,247
45,351
39,021
45,199
39,630
33,796
Volkswagen
Cars
22
76,717
81,615
87,411
87,343
86,996
Volkswagen
Trucks
4,803
-42,347
-43,073
-41,478
-43,598
-45,973
Volvo
Cars
26,090
3,911
4,065
4,601
4,165
4,657
Volvo
Trucks
9,221
2,431
2,625
2,419
2,263
2,013
1-28
-------
Baseline and Reference Vehicle Fleets
Table 1.28 below shows the differences in engine technology penetration between MY2015
and MY2014. One of the larger differences is indicated by the increased use of turbochargers by
Ferrari, Ford, Mercedes, Volkswagen, and Volvo. Many manufacturers are also changing the
type of variable valve timing employed. Significant increases in use of direct injection is
indicated for Ford, Honda, Hyundai/Kia, Subaru, and Volvo.
Table 1.28 Change (2015-2014) in Engine Technology Penetration
Manufacturers
Vehicle Type
Turbo Charged
Super Charged
Single Overhead Cam
Dual Over Head Cam
Over Head Cam
Variable Valve Timing
Continuous Intake Only
Variable Valve Timing
Discrete
Variable Valve Discrete Lift
Only
Variable Valve Lift and
Timing Discrete
Vehicles without Variable
Valve Timing or Lift
Cylinder Deactivation
Direct Injection
All
Both
1%
0%
0%
0%
0%
0%
-3%
0%
2%
1%
0%
4%
All
Cars
0%
0%
2%
-2%
0%
0%
-5%
0%
5%
0%
0%
1%
All
Trucks
3%
0%
-2%
3%
-1%
-1%
0%
0%
-1%
2%
-1%
9%
Aston Martin
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Aston Martin
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
BMW
Cars
2%
0%
0%
0%
0%
0%
0%
0%
4%
-4%
0%
2%
BMW
Trucks
0%
0%
0%
0%
0%
0%
1%
0%
-7%
6%
0%
0%
FCA
Cars
-3%
1%
0%
2%
-2%
-2%
-29%
0%
31%
0%
-3%
0%
FCA
Trucks
1%
0%
1%
6%
-7%
-7%
1%
0%
5%
1%
-7%
0%
Ferrari
Cars
32%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Ferrari
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Ford
Cars
5%
0%
0%
0%
0%
0%
-2%
0%
0%
2%
0%
-4%
Ford
Trucks
19%
0%
-7%
7%
0%
0%
-16%
0%
0%
16%
0%
19%
GM
Cars
-3%
0%
0%
0%
0%
0%
-5%
0%
6%
0%
0%
6%
GM
Trucks
2%
0%
0%
4%
-4%
-3%
4%
0%
0%
-1%
-2%
9%
Honda
Trucks
0%
0%
-1%
1%
0%
0%
0%
0%
0%
0%
2%
40%
Honda
Cars
0%
0%
10%
-10%
0%
0%
0%
0%
0%
0%
1%
17%
Hyundai/Kia
Trucks
-3%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
21%
Hyundai/Kia
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
1%
JLR
Cars
7%
-3%
0%
0%
0%
0%
4%
0%
-4%
0%
0%
0%
JLR
Trucks
18%
-18%
0%
0%
0%
0%
38%
0%
-38%
0%
0%
0%
Lotus
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Lotus
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Mazda
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
7%
Mazda
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
4%
McLaren
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
McLaren
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Mercedes
Cars
33%
0%
0%
0%
0%
0%
1%
0%
0%
-1%
0%
3%
Mercedes
Trucks
7%
0%
0%
0%
0%
0%
9%
0%
0%
-9%
0%
0%
Mitsubishi
Cars
-3%
0%
-3%
4%
0%
0%
0%
0%
0%
0%
0%
0%
Mitsubishi
Trucks
0%
0%
0%
0%
0%
0%
29%
-6%
-23%
0%
0%
0%
Nissan
Cars
-1%
0%
0%
-2%
0%
0%
-1%
0%
0%
2%
0%
2%
1-29
-------
Baseline and Reference Vehicle Fleets
Nissan
Trucks
0%
-2%
0%
0%
0%
0%
0%
0%
0%
0%
0%
-1%
Subaru
Cars
8%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
14%
Subaru
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Tesla
Cars
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Tesla
Trucks
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Toyota
Cars
2%
0%
0%
0%
0%
0%
0%
0%
-1%
0%
0%
0%
Toyota
Trucks
1%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Volkswagen
Cars
10%
-2%
-1%
1%
0%
0%
0%
0%
4%
-4%
0%
7%
Volkswagen
Trucks
14%
3%
0%
0%
0%
0%
-3%
0%
4%
-2%
0%
0%
Volvo
Cars
21%
6%
0%
0%
0%
0%
0%
0%
0%
0%
0%
74%
Volvo
Trucks
44%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Table 1.29 shows the difference in footprint distributions between the MY2015-based fleet
projection and the MY2014-based fleet projection. The differences between MYs 2015 and
2014 are small, and are primarily the result of differences in the manufacturers' product mix in
those model years. The decrease in large pickup trucks and the increase in small and midsize
SUVs causes the average truck footprint and the overall average footprint to decrease slightly.
The difference between the MY2014-based and MY2015-based forecasts are small.
Table 1.29 2015 Projection - 2014 Projection Production Weighted Foot Print Mean Difference
Model
Year
Difference in Average Footprint
of all Vehicles
Difference in Average
Footprint Cars
Difference in Average Footprint
Trucks
2015-2014
49.3-49.7= -0.5
46.1-46.0 = -0.1
53.7-55.0 = -1.3
2017
49.8-50.0 = -0.2
46.0-46.0 =0
53.0-54.0 = -1
2018
49.7-50.1 = -0.3
46.1-46.1 = 0
53.0-54.0 = -1
2019
49.7-50.1 = -0.3
46.1-46.1 = 0
53.0-54.1 = -1.1
2020
49.5-50.0 = -0.5
46.1-46.1 = 0
53.0-54.0 = -1
2021
49.5-50.0 = -0.5
46.1-46.1 = 0
53.0-54.1 = -1.1
2022
49.5-50.0 = -0.5
46.1-46.1 = 0
53.0-54.1 = -1.1
2023
49.4-49.9 = -0.5
46.0 - 46.0 = 0
52.9-54.0 = -1.1
2024
49.3-49.9 = -0.6
46.0 - 46.0 = 0
52.9-54.0 = -1.1
2025
49.3-49.8 = -0.5
46.1-46.1 = 0
53.0-54.0 = -l
Table 1.30 shows the difference in distribution of number of engine cylinders between the
MY2015-based fleet and the MY2014-based fleet. MY2015 includes fewer vehicles with 6- and
8-cylinder engines than MY2014. Fewer 6- and 8-cylinder vehicles in the baseline fleet, along
with changes in product mix, results in greater representation of 4-cylinder engines in trucks and
cars by 2025.
1-30
-------
Baseline and Reference Vehicle Fleets
Table 1.30 Differences in Percentages of 4,6 and 8 Cylinder Engines by Model Year
Trucks
Ca rs
Model
Year
4 Cylinders
6 Cylinders
8 Cylinders
4 Cylinders
6 Cylinders
8 Cylinders
2015-2014
4.2%
-0.1%
-4.2%
3.0%
-2.9%
-0.1%
2017
4.8%
-0.7%
-4.2%
2.5%
-2.6%
0.1%
2018
4.8%
-0.5%
-4.3%
2.4%
-2.5%
0.1%
2019
5.0%
-0.7%
-4.3%
2.4%
-2.5%
0.1%
2020
4.9%
-0.8%
-4.1%
2.4%
-2.6%
0.2%
2021
5.0%
-0.7%
-4.3%
2.4%
-2.6%
0.2%
2022
5.1%
-0.9%
-4.2%
2.4%
-2.6%
0.2%
2023
5.2%
-1.0%
-4.2%
2.3%
-2.6%
0.2%
2024
5.2%
-1.0%
-4.2%
2.3%
-2.5%
0.2%
2025
5.1%
-0.9%
-4.2%
2.2%
-2.5%
0.2%
1.2 The OMEGA Fleet
The prior section presented the development of the baseline fleet and how future sales were
estimated. For OMEGA, we do not apply the baseline fleet as presented above in its "raw" form
for a number of reasons:
1) It includes small-volume manufacturers, which we exclude from this analysis since
they are eligible to apply for unique standards.
2) Despite the need to generate future sales projections for modeling purposes, of
perhaps greater importance to OMEGA is the technology characterization of the
baseline fleet. That is, OMEGA needs "know" the level of technology on baseline
vehicles so that it can properly track costs and effectiveness improvements going
forward.
3) It focuses on consumer metrics for vehicle classification (e.g., small car, large car,
SUV) rather than modeling metrics (e.g., road loads, power-to-weight ratios).
4) It does not include the ZEV program and the fleet of battery electric vehicles (BEVs)
and plug-in electric vehicles (PHEVs) that are projected to be part of the nationwide
fleet in the time frame of the analysis (MYs 2021 through 2025).
As a result, the baseline fleet as presented above undergoes a transition to put that fleet into a
form and of proper content that it can be processed by OMEGA. Removing small-volume
manufacturers from the baseline fleet is easily done as the first step by simply removing Aston
Martin, Ferrari, Lotus and McLaren. The result is a slightly smaller fleet of remaining vehicles.
The technology "walk" from what might be termed "real-world space" to "OMEGA space" is
simply a process of coding specific technologies in the baseline fleet into the technology codes
understood by OMEGA. To properly track costs, OMEGA must, for example, understand that a
vehicle has a V8 rather than an 14 engine, since the two engines have very different cost metrics
for certain additional technologies (for example, engine friction reduction) for which costs are
1-31
-------
Baseline and Reference Vehicle Fleets
based on the number of cylinders. Determining the road load and power-to-weight ratio metrics
is also important for modeling, and is described in more detail in Chapter 2 of this TSD.
For the Proposed Determination analysis, converting the baseline presented in Chapter 1.1
into a ZEV program-compliant "OMEGA baseline" was performed in largely the same way as
for the Draft TAR analysis. One notable difference is that, in the Draft TAR, EPA built ZEV
program vehicles on the same platforms as the ICE vehicle from which the sales were taken. In
this analysis, we have built those ZEV program vehicles on unique platforms. The result is a far
greater number of platforms in this analysis, but this also allows us to essentially leave those
existing ZEV program vehicles, and all BEV/PHEV vehicles in our analysis, alone. They simply
pass through OMEGA untouched and unimproved. Their emissions, both tailpipe and upstream,
are considered by OMEGA in determining a path toward compliance, but those vehicles are not
considered for improvement since most already perform considerably better than their respective
footprint-based targets.
1.2.1 Incorporation of the California Zero Emissions Vehicle (ZEV) Program into the
OMEGA Reference Fleet
1.2.1.1 The ZEV Regulation in OMEGA
In its analysis for this Proposed Determination, EPA has considered sales of electrified
vehicles as projected to be needed to meet state Zero Emission Vehicle (ZEV) requirements.
Because these ZEVs are already required by separate regulations in California and nine other
states, these vehicles are built into the OMEGA reference fleet. This approach reasonably avoids
attributing costs to the federal GHG program which necessarily occur due to another existing
requirement, and assures that those costs are not double counted. Note that this reflects a change
from the 2012 FRM, where EPA did not account for compliance with the ZEV regulations in the
reference case fleet for the 2017-2025 standards. However, this was because CARB was
simultaneously substantially revising the ZEV regulation in early 2012 just prior to the release of
the 2012 FRM, and EPA had not yet acted upon California's waiver request for the ZEV
program. The approach described here is consistent with the approach EPA took in the Draft
TAR.
Public comments on the Draft TAR included some comments related to our inclusion of ZEV
program vehicles in the reference case. Specifically, the Alliance of Automobile Manufacturers
and others commented that including compliance with the ZEV program as part of our reference
fleet analysis was unfairly counting their benefits without estimating their costs.1 This comment
is mistaken. The presence of ZEV program vehicles in our analysis is done both in the reference
and control cases. As such, costs associated with those vehicles and any benefits derived by them
cancel out in calculating net benefits. EPA's methodology is also consistent with OMB Circular
A-4, which states that in developing a baseline for purposes of analyzing the potential effects of
a proposed rule,"[t]his baseline should be the best assessment of the way the world would look
absent the proposed action. "J
1EPA-HQ-OAR-2015-0287-0928 at Section 4.1.2.1.
1 Office of Management and Budget Circular No. A-4, "Regulatory Analysis," at page 15, available at
https://www.whitehouse.gov/omb/memoranda_m03 -21.
1-32
-------
Baseline and Reference Vehicle Fleets
Other commenters, including NGOs such as the Environmental Defense Fund and the Union
of Concerned Scientists, believe that EPA correctly accounted for the ZEV program by including
California's ZEV vehicles in its reference fleet, as this approach ensures that the costs of the ZEV
program, which are not imposed by the 2022-2025 standards but rather by state law are not
included as costs of the national rule. EPA agrees. The California ZEV program is an existing
state requirement that has been adopted by California, as well as by several other states.
Therefore, EPA included vehicles that are needed to comply with the ZEV program as part of
our reference fleet in assessing the MY2022-2025 GHG standards. Thus, as explained above,
the Draft TAR did not include an assessment of the benefits or the costs of the ZEV program in
the assessment of 2022-2025 National Program standards. However, any ZEV vehicles sold in
California and other states will help a manufacturer in meeting the EPA GHG standards. While
the fleet-average GHG emissions standards establish minimum standards, they do not limit the
ability of manufacturers to achieve further reductions, and any manufacturer that does will
generate credits that can be used or sold. ZEVs sold in California and other states will help a
manufacturer to meet (or exceed) the EPA GHG standards.
The conclusions presented in this analysis are meant to be one example representation of how
the ZEV program requirements could be fulfilled; it is in no way meant to reflect the exact way
in which any given manufacturer would actually comply with the ZEV program. Rather, it is
meant as an illustration to reflect the potential number and penetration of ZEVs across the
national fleet as part of the reference case. To accomplish this, the baseline fleet with future
sales projections had to be adjusted to account for the projected ZEV sales. Those sales
adjustments are described in detail below (see 1.2.1.2). The analysis fleets used in OMEGA and
in EPA's benefit cost analysis for the AEO reference fuel price case are shown in Table 1.31
through Table 1.34, with additional breakdowns of these sales shares shown in Table 1.35.
Note that, in Table 1.31 through Table 1.34, EPA shows "Baseline" BEV and PHEV sales and
"Additional ZEV Program" BEV and PHEV sales. The "baseline" sales are sales projected in
EPA's MY2015-based baseline fleet. In other words, these vehicles are part of the future fleet
described in Chapter 1.1. The "additional ZEV program" sales are BEV and PHEV sales above
and beyond those projected in Chapter 1.1. The "additional ZEV program" sales were taken from
the ICE-only sales that were proj ected in Chapter 1.1. We have not increased the size of the
fleet, but have "converted" some ICE-only vehicles to BEVs and PHEVs to meet the projected
sales required by the ZEV program in California and nine other states. We describe the process
of doing this in the text following the tables. Importantly, the costs of "converting" the
"additional ZEV program" sales are attributable to the ZEV program and, therefore, those costs
are not considered in the EPA analysis. Similarly, any benefits from those vehicles are not
considered explicitly in the EPA analysis. However, there is an implicit benefit that is
considered. Since the ZEV program vehicles are part of the analysis fleet, they reduce slightly
the GHG compliance burden (i.e., the fleet average GHG standards) for any manufacturer
required to meet the ZEV program because the additional ZEVs, when averaged with other
vehicles, lower that manufacturer's fleet average GHG emissions.K By starting with a lower
K Importantly, we have modeled MY2025 electricity consumption considering the upstream emissions. As a result,
BEV and PHEV miles driven using full electric power are not considered zero. Because of this, the impact of the
ZEV program vehicles is less in this analysis than it was in the Draft TAR since that analysis considered upstream
emissions to be zero.
1-33
-------
Baseline and Reference Vehicle Fleets
GHG-emitting baseline fleet, the compliance burden to get to the final standards is smaller but
this necessarily also means that the calculated GHG benefits (the delta between the baseline and
final standards) are also smaller. We model the fleet in this way because this is how ZEV
program vehicles will be reflected in compliance with the national GHG standards.
Table 1.31 OMEGA MY2021 Car Fleet using the AEO 2016 Reference Fuel Price Case
ICE-only Car
Baseline
Baseline
Additional ZEV
Additional ZEV
Total Car
Sales
BEV Sales
PHEV Sales
Program BEV Sales
program PHEV Sales
Sales
BMW
296,220
4,347
17,082
0
0
317,648
FCA
523,734
5,704
0
1,172
4,990
535,600
Ford
810,252
1,212
9,491
5,220
5,434
831,609
GM
1,118,223
1,688
28,544
5,889
0
1,154,344
Honda
800,481
0
0
7,472
11,705
819,658
Hyundai/Kia
1,110,746
589
0
6,700
11,118
1,129,153
JLR
22,382
0
0
214
336
22,932
Mazda
208,312
0
0
1,719
2,693
212,725
Mercedes
210,362
3,167
50
961
3,968
218,508
Mitsubishi
47,071
0
0
275
430
47,775
Nissan
785,250
25,188
0
34
9,732
820,204
Subaru
137,854
0
0
1,220
1,912
140,987
Tesla
0
90,547
0
0
0
90,547
Toyota
1,172,623
0
4,695
11,415
15,111
1,203,844
Volkswagen
526,653
2,737
1,343
3,026
7,761
541,520
Volvo
43,480
0
0
406
636
44,523
Fleet
7,813,644
135,179
61,204
45,723
75,827
8,131,578
Note: The analysis fleet differs from the baseline fleet by removing small volume manufacturers (Aston Martin,
Ferrari, McLaren, and Lotus) and by adjusting sales to account for projected ZEV sales.
Table 1.32 OMEGA MY2021 Truck Fleet using the AEO 2016 Reference Fuel Price Case
ICE-only Car
Baseline
Baseline
Additional ZEV
Additional ZEV
Total Car
Sales
BEV Sales
PHEV Sales
Program BEV Sales
program PHEV Sales
Sales
BMW
115,780
0
0
0
0
115,780
FCA
1,258,798
0
0
2,150
9,151
1,270,099
Ford
1,247,780
0
0
4,383
4,562
1,256,726
GM
1,254,629
0
0
3,401
0
1,258,030
Honda
841,687
0
0
7,856
12,308
861,851
Hyundai/Kia
224,154
0
0
1,352
2,244
227,750
JLR
100,048
0
0
957
1,500
102,505
Mazda
127,183
0
0
1,050
1,644
129,877
Mercedes
174,375
0
0
725
2,996
178,096
Mitsubishi
34,710
0
0
202
317
35,229
Nissan
573,978
0
0
21
5,941
579,939
Subaru
519,605
0
0
4,600
7,206
531,411
Tesla
Toyota
1,055,084
0
0
7,243
9,588
1,071,915
Volkswagen
253,117
0
4,120
1,185
3,040
261,463
Volvo
47,705
0
0
446
698
48,849
Fleet
7,828,633
0
4,120
35,571
61,196
7,929,520
Note: The analysis fleet differs from the baseline fleet by removing small volume manufacturers (Aston Martin,
Ferrari, McLaren, and Lotus) and by adjusting sales to account for projected ZEV sales.
1-34
-------
Baseline and Reference Vehicle Fleets
Table 1.33 OMEGA MY2025 Car Fleet using the AEO 2016 Reference Fuel Price Case
ICE-only Car
Baseline
Baseline
Additional ZEV
Additional ZEV
Total Car
Sales
BEV Sales
PHEV Sales
Program BEV Sales
program PHEV Sales
Sales
BMW
311,383
7,867
29,016
28
0
348,293
FCA
540,170
5,579
0
4,679
7,904
558,331
Ford
802,137
1,322
9,525
10,711
9,631
833,326
GM
1,189,943
2,186
31,131
12,938
3,484
1,239,682
Honda
848,485
0
0
16,107
18,926
883,518
Hyundai/Kia
1,153,285
535
0
14,543
17,515
1,185,878
JLR
23,499
0
0
458
538
24,494
Mazda
218,037
0
0
3,652
4,292
225,981
Mercedes
224,860
3,955
106
3,434
6,456
238,811
Mitsubishi
59,477
0
0
701
824
61,002
Nissan
846,189
26,490
0
6,734
16,017
895,430
Subaru
146,744
0
0
2,640
3,102
152,485
Tesla
0
109,459
0
0
0
109,459
Toyota
1,244,257
0
4,742
24,558
25,915
1,299,472
Volkswagen
573,109
3,049
1,509
8,708
12,811
599,186
Volvo
46,000
0
0
874
1,027
47,901
Fleet
8,227,574
160,441
76,029
110,766
128,441
8,703,251
Note: The analysis fleet differs from the baseline fleet by removing small volume manufacturers (Aston Martin,
Ferrari, McLaren, and Lotus) and by adjusting sales to account for projected ZEV sales.
Table 1.34 OMEGA MY2025 Truck Fleet using the AEO 2016 Reference Fuel Price Case
ICE-only Car
Baseline
Baseline
Additional ZEV
Additional ZEV
Total Car
Sales
BEV Sales
PHEV Sales
Program BEV Sales
program PHEV Sales
Sales
BMW
104,922
0
0
8
0
104,931
FCA
1,253,319
0
0
8,071
13,632
1,275,022
Ford
1,166,687
0
0
8,509
7,651
1,182,848
GM
1,191,481
0
0
6,613
1,780
1,199,874
Honda
802,944
0
0
15,243
17,910
836,097
Hyundai/Kia
221,511
0
0
2,794
3,365
227,669
JLR
90,516
0
0
1,762
2,071
94,350
Mazda
131,404
0
0
2,201
2,586
136,192
Mercedes
160,299
0
0
2,234
4,200
166,733
Mitsubishi
38,466
0
0
454
533
39,452
Nissan
539,914
0
0
3,481
8,280
551,676
Subaru
534,344
0
0
9,612
11,294
555,249
Tesla
Toyota
1,003,343
0
0
13,661
14,416
1,031,420
Volkswagen
253,335
0
4,056
3,146
4,629
265,166
Volvo
46,980
0
0
893
1,049
48,921
Fleet
7,539,466
0
4,056
78,682
93,397
7,715,601
Note: The analysis fleet differs from the baseline fleet by removing small volume manufacturers (Aston Martin,
Ferrari, McLaren, and Lotus) and by adjusting sales to account for projected ZEV sales.
1-35
-------
Baseline and Reference Vehicle Fleets
Table 1.35 Breakdown of MY2025 Internal Combustion Engine, Electric and Plug-in Electric Vehicle Sales
using the AEO 2016 Reference Fuel Price Case
Car
Truck
Sum
Share
ICE-only
8,227,574
7,539,466
15,767,039
96.0%
Baseline BEV
160,441
0
160,441
1.0%
Baseline PHEV
76,029
4,056
80,085
0.5%
ZEV BEV
110,766
78,682
189,447
1.2%
ZEV PHEV
128,441
93,397
221,838
1.4%
Total ICE+BEV+PHEV
8,703,251
7,715,601
16,418,851
100.0%
Baseline BEV
160,441
0
160,441
24.6%
Baseline PHEV
76,029
4,056
80,085
12.3%
ZEV BEV
110,766
78,682
189,447
29.1%
ZEV PHEV
128,441
93,397
221,838
34.0%
Total BEV+PHEV
475,677
176,135
651,812
100.0%
ICE
8,227,574
7,539,466
15,767,039
96.0%
Baseline BEV+PHEV
236,470
4,056
240,527
1.5%
ZEV BEV+PHEV
239,207
172,079
411,285
2.5%
Total ICE+BEV+PHEV
8,703,251
7,715,601
16,418,851
100.0%
ICE
8,227,574
7,539,466
15,767,039
96.0%
Total BEV+PHEV
475,677
176,135
651,812
4.0%
Total ICE+BEV+PHEV
8,703,251
7,715,601
16,418,851
100.0%
The ZEV program sales are calculated based on the baseline fleet described in Chapter 1.1.
From that fleet, we removed Aston Martin, Ferrari, McLaren and Lotus vehicles. That fleet
includes some BEVs and PHEVs consistent with the sales in the MY2015 baseline fleet as
projected forward to MYs 2021 and 2025. The additional ZEV program sales shown above in
Table 1.31 through Table 1.34 were modeled as replacing ICE vehicles in the baseline fleet to
maintain the same overall sales volume for each manufacturer's fleet. To "generate" the
projected additional ZEV program vehicles, each model within a manufacturer's fleet was
mapped into a vehicle type matching its characteristics and capability. For this analysis, it was
assumed that only vehicle types classified as non-towing would be considered for conversion
from an ICE to a ZEV to meet the ZEV program requirements. The 24 vehicle types considered
for additional ZEV program sales include all of vehicle types not designated as large pickups. In
other words, we now allow many more types of vehicles to electrify than we allowed in the Draft
TAR or the 2012 FRM where we essentially limited BEV and PHEV electrification to passenger
cars. Table 1.36 shows the 29 vehicle types being used in this analysis including the towing or
non-towing designation and consideration as a "ZEV-source platform." Rather than selecting
which individual vehicle models or platforms would be the most likely sources, all ICE vehicles
within the non-towing vehicle types in a manufacturer's fleet were considered as a source for
additional ZEV program sales. Each manufacturer's additional ZEV program sales were then
created by converting, on a platform-level sales weighted basis across all eligible vehicle types,
the necessary number of ICE vehicles into the respective BEV and PHEV sales. By sales-
weighting across all eligible vehicle types, the vehicle category and size (footprint)
characteristics of each manufacturer's fleet were kept consistent with the original baseline
1-36
-------
Baseline and Reference Vehicle Fleets
projections. The tables below are meant to provide clarity with a simple example of how this
was done.L
Table 1.36 Vehicle Types Considered for Conversion to ZEV Program Vehicles
Vehicle Type
Description
Curb Weight Class
ALPHA Class
ZEV source?
1
14 DOHC
1
LPW LRL
Yes
2
14 DOHC
1
MPW LRL
Yes
3
14 DOHC
2
MPW LRL
Yes
4
14 DOHC
2
LPW HRL
Yes
5
14 DOHC
3
MPW LRL
Yes
6
14 DOHC
3
LPW HRL
Yes
7
14 DOHC
4
LPW HRL
Yes
8
14 DOHC
6
Truck
No, Heavy-tow
9
V6 0HV
6
Truck
No, Heavy-tow
10
V6SOHC
3
HPW
Yes
11
V6SOHC
4
MPW HRL
Yes
12
V6DOHC
1
LPW LRL
Yes
13
V6DOHC
2
MPW LRL
Yes
14
V6DOHC
2
LPW LRL
Yes
15
V6DOHC
3
HPW
Yes
16
V6DOHC
3
MPW LRL
Yes
17
V6DOHC
3
LPW HRL
Yes
18
V6DOHC
4
HPW
Yes
19
V6DOHC
4
MPW HRL
Yes
20
V6DOHC
5
HPW
Yes
21
V6DOHC
5
MPW HRL
Yes
22
V6DOHC
6
Truck
No, Heavy-tow
23
V8 0HV
5
HPW
Yes
24
V8 0HV
5
MPW HRL
Yes
25
V8 0HV
6
Truck
No, Heavy-tow
26
V8DOHC
4
HPW
Yes
27
V8DOHC
5
HPW
Yes
28
V8DOHC
5
MPW HRL
Yes
29
V8DOHC
6
Truck
No, Heavy-tow
Note: DOHC=dual overhead cam; SOHC=single overhead cam; OHV=overhead valve; Curb Weight Class is a
percentile-based weight classification with 1 being the lightest and 6 being the heaviest vehicles; ALPHA class is
described in Chapter 2.3 of this TSD and designates low/medium/high power-to-weight (L/M/HPW) and
low/medium/high road load (L/M/HRL) or Truck which is used for large pickups like the Ford F150 and Chevy
Silverado.
First, consider a simple manufacturer fleet consisting of seven vehicle models built on five
platforms, which we have mapped into three vehicle types with total fleet sales of 600 vehicles,
as shown in Table 1.37.
L The Excel spreadsheets used to generate the ZEV program fleet are in the docket and on our website at
https://www.epa.gov/regulations-emissions-vehicles-and-engines/optimization-model-reducing-emissions-
greenhouse-gases. The filenames include the keyword "FleetsABC."
1-37
-------
Baseline and Reference Vehicle Fleets
Table 1.37 Example Manufacturer Fleet from which ZEVs are to be Created
Platform index
Vehicle index
Model
Fuel
VehType
Baseline sales
100
1
A
G
1
100
100
2
B
G
1
100
101
3
C
G
2
75
101
4
D
G
2
75
102
5
E
G
1
100
103
6
F
G
2
50
104
7
G
G
29
100
Total
600
For this manufacturer, we will assume that the needed additional ZEV program sales are 50
BEVs and, for simplicity, no PHEVs. As noted above, vehicle types 8, 9, 22, 25 and 29 are not
considered to be ZEV-source platforms. Thus, the 50 ZEV program vehicles cannot come from
platform 104 since that is vehicle type 29. We determine the number of BEVs to create from
each platform according to its sales weighting within ZEV-source platforms.M This is shown in
Table 1.38. We also need to know how many vehicles within each vehicle model to convert to a
ZEV program vehicle. This is shown in Table 1.39.
Table 1.38 Number of Additional ZEV Program Sales from each Platform
Platform index
VehType 1
VehType 2
Total
%in Platform
# of ZEV program sales
100
200
200
40%
20
101
150
150
30%
15
102
100
100
20%
10
103
50
50
10%
5
Total
300
200
500
100%
50
Table 1.39 Percentage of Additional ZEV Program Sales from Each Vehicle Model
Platform index
Model A
Model B
Model C
Model D
Model E
Model F
Total
100
50%
50%
100%
101
50%
50%
100%
102
100%
100%
103
100%
100%
With the details shown in Table 1.38 and Table 1.39, we can then convert ICE vehicles into
ZEV program vehicles as shown in Table 1.40.
M The ZEV-source platforms are those platforms "mapped" into the 23 "ZEV platform" vehicle types presented in
Table 1.36. The point of Table 1.36 is to make clear that we are creating ZEV program vehicles in only those
types of vehicles that we believe to make the most sense. Those types of vehicles being passenger cars and sport
and cross-over utility vehicles that are not generally heavy-towing vehicles. The ZEV program vehicles are
created only from within those vehicle types and, therefore, the creation of ZEV program vehicles is done using
sales-weighting within those vehicle types rather than within all vehicles.
1-38
-------
Baseline and Reference Vehicle Fleets
Table 1.40 Example Manufacturer's OMEGA Fleet including ZEV Program Sales
Platform
index
Vehicle index
Model
Fuel
VehType
Baseline Sales
OMEGA fleet
with ZEV
program sales
100
1
A
G
1
100
90
100
2
B
G
1
100
90
101
3
C
G
2
75
68
101
4
D
G
2
75
68
102
5
E
G
1
100
90
103
6
F
G
2
50
45
104
7
G
G
29
100
100
105
8
ZEV
E
1
0
20
106
9
ZEV
E
1
0
15
107
10
ZEV
E
2
0
10
108
11
ZEV
E
2
0
5
Total sales G
600
550
Total sales E
0
50
Total sales
600
600
As noted above, we then created each manufacturer's ZEV program fleet by converting, on a
platform-level sales weighted basis, the necessary number of ICE vehicles into the respective
BEV and PHEV sales. EPA staff considered an alternate approach to look instead at which
specific platforms, or even vehicle models, were the best candidates for conversion to
BEV/PHEV. However, that approach was rejected because there is no industry consensus on
which characteristics make a vehicle the best candidate for conversion. Is it the smallest cars, the
lightest cars, those that already have a BEV or PHEV version, etc.? Any attempt at determining
the "best" candidates for conversion might be seen as "cherry picking" in order to provide a
certain result. Some might see us as choosing all of the smallest vehicles, thereby leaving all of
the larger, perhaps "dirtier" vehicles as ICE vehicles needing costly improvements to comply
with the future standards. Others might see us as choosing all of the largest vehicles, thereby
leaving all of the smaller, perhaps "cleaner" vehicles as ICE vehicles needing less costly
improvements to comply with future standards. Further, there is no clear trend as to which
vehicles or platforms manufacturers are currently using for BEV or PHEV platforms. Current
and publicly-announced near term models span platforms from subcompact cars to large cars,
large SUVs to minivans, and use of shared or dedicated platforms. Our final decision was to
choose equally (by sales weighting) from each ZEV source platform such that there would be no
net impact on the sales weighted footprint of remaining ICE vehicles needing technology to
comply.
1.2.1.2 The ZEV Program Requirements
The preceding discussion describes how we determined which vehicles would be converted
from ICE technology to BEV/PHEV. Here we discuss the assumptions regarding the
characteristics of the ZEVs used in the analysis and how compliance (total sales) with the ZEV
mandate was modeled.
1-39
-------
Baseline and Reference Vehicle Fleets
1.2.1.2.1 Overview
California requires the largest vehicle manufacturers to manufacture ZEV credit producing
vehicles to comply with the increasing number of ZEV credits required through 2025.7 The ZEV
credits can be generated by producing battery electric vehicles, fuel cell electric vehicles, and
certain plug-in hybrid vehicles. In addition to the requirements applying in California (CA),
several other states have used section 177 (SI 77) of the federal Clean Air Act to adopt the
California ZEV requirements (referred to as SI77 ZEV States).8 These states, when combined
with CA, account for nearly 30 percent of all new light-duty vehicles sold in the United States.
Under the ZEV regulation, manufacturers are required to generate ZEV credits to fulfill an
annual obligation based on their cumulative vehicle sales as summarized in Table 1-40.
Requirements are satisfied by producing vehicles that generate credit which, for MY2018 and
beyond, means a combination of plug-in hybrid electric vehicles (PHEV), battery electric
vehicles (BEV), and fuel cell electric vehicles (FCEV). Each PHEV, BEV, and FCEV earns
between 0.4 and 4 credits per vehicle depending on its electric range over a test cycle as
specified in the CA ZEV regulation.9 For example, a PHEV with a 10-mile electric range earns
0.4 credits and a BEV or FCEV with a 350-mile test range earns 4.0 credits.
To incorporate the ZEVs into the OMEGA fleet, the ZEV regulation credit requirements were
converted to a vehicle sales requirement as follows:
1) Determine how many total ZEV credits each manufacturer will need in CA and the
SI77 ZEV states for each year being modeled in OMEGA (MY2021 and MY2025).
2) Develop a nominal BEV electric range (described in Table 4.33) and a nominal
PHEV set of electric range characteristics (described in Table 4.34) that are projected
to be representative of BEV and PHEV capability in the MY2021-2025 time frame.
The range and characteristics are then used to determine how many ZEV credits each
vehicle will generate. For simplification and alignment with existing OMEGA
technology packages, FCEVs were not included in the compliance scenarios.
3) Calculate the incremental ZEV credits needed beyond those generated by any ZEVs
already included in the OMEGA reference fleet projections and expected to be sold in
CA and the SI77 ZEV states.
4) Determine how many incremental BEVs and PHEVs each manufacturer will need to
sell to satisfy their ZEV credit obligations for MY2021 and MY2025.
1.2.1.2.2 ZEV Credit Requirement
Each manufacturer's ZEV credit obligation is calculated by multiplying its projected total
light duty vehicle sales in CA and SI77 ZEV states by the ZEV credit percentage required (see
Table 1.41 below). The total projected CA and S177 ZEV states sales volume for each
manufacturer was calculated by multiplying the manufacturer-specific reference fleet national
sales volumes in OMEGA by the CA and S177 ZEV states sales volume ratio (MY2014). For
example, if manufacturer "A" is projected to sell 250,000 vehicles nationally in MY2021, and its
CA and SI 77 ZEV state sales are 40 percent of its national sales, its projected MY2021 CA and
S177 ZEV state sales would be 100,000 (250,000*40%). Although the regulation has
flexibilities in the technologies a manufacturer may use to generate credits, there is a cap on the
1-40
-------
Baseline and Reference Vehicle Fleets
portion of the credits that can be satisfied with PHEVs as identified in Table 1.41. For example,
if manufacturer "A" sells 100,000 vehicles in CA and the SI 77 ZEV states in 2021, it is required
to generate 12,000 ZEV credits (100,000*12%) in 2021 and, of those 12,000 ZEV credits, only
4,000 (100,000*4%) can come from PHEVs. For the purpose of this analysis, manufacturers are
proj ected to comply with the ZEV requirements by maximizing their ZEV credits earned using
PHEVs and using BEVs to generate the remaining credits.
Table 1.41 ZEV Regulation Credit Requirements
2018
2019
2020
2021
2022
2023
2024
2025
Total ZEV Credit Required
4.50%
7.00%
9.50%
12.00%
14.50%
17.00%
19.50%
22.00%
Max. Credits from PHEVs
2.50%
3.00%
3.50%
4.00%
4.50%
5.00%
5.50%
6.00%
1.2.1.2.3 Pro jected Representative PHEV andBEV Characteristics for MY2021-2025
The first step to calculate the number of ZEVs needed to meet the manufacturer's projected
credit obligation is to determine the type of vehicles that will be used to comply with the
regulation. The primary characteristic for determining ZEV credits per vehicle is the urban
dynamometer driving schedule (UDDS) test cycle range for BEVs and the UDDS test cycle
"equivalent all electric range" for PHEVs. ZEV credits are generated based on UDDS range, not
label range, and a review of current certified BEVs indicates a UDDS range to label range
correction factor of between 0.65 and 0.76. For this analysis, a value of 0.7 was used for all
vehicles. Given that these would be future vehicles for which actual specifications are not yet
known, assumptions were made regarding what future range(s) might be in the MY2021 and
MY2025 time frame. Further simplifications of such projections were also necessary to fit
within the existing model framework of OMEGA including baseline vehicles and technology
packages. These simplifications include the use of a single nominal BEV range and a single
nominal PHEV range for all manufacturers and all vehicle classes with characteristics projected
to be representative of BEVs and PHEVs in the MY2021-2025 time frame. Given these
constraints, this projection reflects a scenario for minimum compliance with the ZEV regulation
using a representative nominal BEV and PHEV, but not a 'likely' scenario that might reflect a
wide variety of different ranges of PHEV and BEV offerings across manufacturers, vehicle
classes, and model years, or the inclusion of FCEVs, which have already begun to enter the
market.
To develop the nominal BEV and PHEV electric range, EPA staff first looked at the relative
impact of battery pack costs for a variety of battery costs (dollars per kilowatt-hour (kWh)). For
this simplified analysis, vehicle energy consumption was assumed to be constant for all vehicle
types; therefore, all-electric vehicle range and battery pack size increase proportionally. The
relative costs to achieve longer range were then compared to the number of ZEV credits earned
for the increased range. The qualitative results are shown in Figure 1.4. As the figure shows,
building individual BEVs with a longer range directionally results in a lower cost per ZEV credit
earned (i.e., satisfying the ZEV credit obligation with fewer long range BEVs is directionally
more cost-effective than using a larger volume of shorter range BEVs). And, as Figure 1.4
illustrates, the relative impact is even larger at the lower battery costs projected for the 2022-
2025 time frame. Accordingly, the nominal BEV and PHEV packages modeled longer range
variants of both types of ZEVs rather than multiple variants of shorter and longer range vehicles.
Note that the range of battery costs used in the figure (from $150/kWh to $300/kWh in the 2021-
1-41
-------
Baseline and Reference Vehicle Fleets
2025 time frame) is generally consistent with the projections of the EPA battery costing analysis
for PHEVs and BEVs as reported in Chapter 2.3.4.3.7 of this TSD. EPA's projected costs used
in the 2012 FRM, the Draft TAR, and this analysis are supported elsewhere in the Draft TAR
and this TSD, particularly in Chapter 5 of the Draft TAR where we evaluated the 2012 FRM and
Draft TAR battery cost projections, and in Chapter 2 of this TSD where we discuss the battery
cost projections used in this analysis.
0 100 200 300 400
Range (miles)
Battery Cost:
$300/kWh
Battery Cost:
$250/kWh
Battery Cost:
$200/kWh
Battery Cost:
$150/kWh
Figure 1.4 Relative Cost of ZEV Credits for Different Ranges and Battery Costs
The projected range for the nominal BEV and PHEV in the MY2021 to 2025 time frame was
developed assuming a constant sales weighted average percent improvement from the current
range. The MY2015 BEV sales-weighted label range is -133 miles, as shown in Table 1.42
below; for MY2015 PHEVs, the sales-weighted label electric range is -25 miles as shown in
Table 1.43.
1-42
-------
Baseline and Reference Vehicle Fleets
Table 1.42 Range Characteristics of BEVs for MY2015
Brand
Model
EPA Label All-electric Range (miles)
BMW
13 BEV
81
BMW
13 BEV
81
BMW
13 REX
72
BMW
13 REX
72
FCA
500e
87
Ford
Focus Electric FWD
76
GM
SPARK EV
82
Hyundai/Kia
Soul Electric
93
Mercedes
B-Class Electric Drive
87
Mercedes
smart fortwo elec. drive (conv.)
68
Mercedes
smart fortwo elec. drive (coupe)
68
Nissan
LEAF
84
Nissan
LEAF
84
Tesla
Model S
260
Tesla
Model S AWD
260
Volkswagen
e-Golf
83
Sales-Weighted Average Range (label Miles)
133
Table 1.43 Range Characteristics of PHEVs for MY2015
Brand
Model
EPA Label All-electric Range (miles)
Ford
C-Max Energi
20
Ford
Fusion Energi
20
Cadillac
ELR
37
Chevrolet
Volt
38
Toyota
Prius Plug-In
11
Sales-Weighted Average Range (label Miles)
25
For this analysis, the range for future vehicles was estimated to increase at a rate of 5 percent
per year until the sales-weighted label range reaches 245 miles, which correlates to the maximum
number of ZEV credits earned by any one vehicle. While manufacturers are not expected to
actually redesign vehicles to increase the range every year nor to cap the range when they reach
245 miles, this rate of annual improvement is consistent with the improvements manufacturers
have been making over more discrete intervals such as redesigns, refreshes, or other updates.
For example, new or updated model introductions and announcements for the Ford Focus EV,
VW e-Golf, Nissan Leaf, Tesla Model S, Tesla Model 3, Chevy Bolt EV, Chevy Volt PHEV,
and BMW i3 have all included increased range compared to their predecessors. The 5 percent
rate of growth is an estimated average of both longer and shorter range vehicles. It is not
expected that BEVs with 200+ miles of range, such as some Tesla vehicles, will increase their
range as quickly as shorter range vehicles such as the BMW i3. This is supported by the 2.5
percent per year increase observed in the Model S (85 to 90 kW-h) compared to the 9 percent per
year increase seen by the GM Volt and the BMW i3. Additionally, while some OEMs may
continue offering BEVs with lower ranges, these may be offset by longer range offerings such as
hydrogen fuel cell electric vehicles (FCEVs) like those announced by Toyota and Honda, having
ranges that well exceed 200 miles.
1-43
-------
Baseline and Reference Vehicle Fleets
Given that the time period of interest is MY2021-2025 and that the ZEV requirements
increase annually, a nominal range for the single BEV variant to be used for the model years of
interest was determined by calculating the sales-weighted average for the years being evaluated.
Table 1.44 combines the results from Table 1.42 for average electric range with the projected
BEV sales for MY2021-2025 to calculate a sales-weighted average BEV for MYs 2021-2025.
The sales-weighted average was calculated as 209 miles. Although this projection results in an
estimated 209-mile range, a final range of 200 miles was chosen to provide for a potential
slower-than-historical increase in range and to be consistent with an existing technology package
in OMEGA (BEV200). EPA believes that a 200-mile label range is reasonable given recent
announcements in this magnitude for the Tesla Model 3, GM Bolt EV, and an announced future
Ford BEV which will all be available prior to MY2021.N For the model years being evaluated,
all BEV200s are assumed to have a label range of 200 miles and a UDDS range of 286 miles
which generates 3.36 ZEV credits per vehicle.
Table 1.44 Projected Sales Weighted BEV Range for MY2021-2025
Model year
BEV real-world range
BEV sales
(% of whole fleet)
BEV sales
(% of 2021-2025 cumulative BEV sales)
2021
187
2%
14%
2022
196
3%
17%
2023
206
3%
20%
2024
216
4%
23%
2025
227
4%
26%
Range Based on Sales Weighting MY2021-2025
209
The projected ranges for PHEVs in the MY2021-2025 time frame were calculated in a similar
manner to the BEV ranges, with one minor difference. PHEVs generate credits based not only
on electric range on the UDDS cycle, but also on the ability to drive ail-electrically for at least 10
miles of the US06 supplemental FTP test cycle. PHEVs that can meet this US06 criterion earn
an additional 0.2 credits per vehicle. While the reality is that motor, inverter, and battery pack
sizing along with the powertrain architecture all play a role in determining whether a PHEV can
meet this criterion, for this analysis, the ability to meet it was assumed to increase linearly for
vehicles with electric range from 20 to 40 miles (i.e., 0 percent of PHEVs with 20-mile range, 50
percent of PHEVs with a 30-mile range, and 100 percent of PHEVs with 40-mile range can meet
the US06 criterion). The analysis summarized in Table 1.45 shows that, for MYs 2021-2025, the
sales-weighted average PHEV is projected to have a range of about 39 miles, which was rounded
down to a final range of 40 miles to be consistent with an existing technology package
(PHEV40) in OMEGA. A PHEV40 is assumed to be 100 percent US06 capable, so it generates
1.07 credits per vehicle after adjusting from a 40-mile label range to an equivalent UDDS range
and including the additional credits for US06 capability. For perspective, the newly revised
MY2016 GM Volt already exceeds this capability and other manufacturers are expected to
further increase their range and capability over the next 5 to 9 years.
N More examples supporting the rationale for BEV200 and discussion of public comment on this topic can be found
in Chapters 2.2.4.4.5 and 2.3.4.3.5 of this TSD.
1-44
-------
Baseline and Reference Vehicle Fleets
Table 1.45 Projected Sales Weighted PHEV Range for MY2021-2025
Model year
BEV real-world range
PHEV sales
(% of whole fleet)
PHEV sales
(% of 2021-2025 cumulative PHEV sales)
2021
35
4%
17%
2022
37
4%
19%
2023
39
5%
20%
2024
41
5%
21%
2025
43
5%
23%
Range Based on Sales Weighting MY2021-2025
39
1.2.1.2.4 Calculation of Incremental ZEVs Needed for ZEV Program Compliance
Next, the number of ZEV credits generated from vehicles already included in the projected
reference fleet was subtracted from the total credit obligation. Given that the projected reference
fleet only included national sales numbers for ZEVs, those numbers were first scaled to
California and SI77 ZEV state sales using the current (average of MY2014 and MY2015)
manufacturer-specific percentage of national ZEV sales in California and the SI 77 ZEV states.
For this analysis, all manufacturers are projected to generate ZEV credits using the nominal BEV
and PHEV all-electric ranges calculated above, and each manufacturer is projected to fulfill their
credit requirements without exercising any of the various additional flexibilities included in the
ZEV regulation. These earned credits were then subtracted from each manufacturer's credit
obligation to calculate the remaining incremental credits needed. For example, if a
manufacturer's ZEV credit obligation for MY2021 is 12,000 credits, and the original baseline
projected 1000 BEV sales in California and the SI77 ZEV states, its incremental obligation is
8,640 ZEV credits (12,000 credits -1000 vehicles*3.36 credits/vehicle).
Finally, the incremental credits needed were translated to the number of additional PHEV and
BEV sales for each manufacturer. For this analysis, it was assumed that each manufacturer
would satisfy the maximum amount of ZEV credits allowed with PHEVs, and the remaining
portion with BEVs. Both the ZEVs in the original reference fleet and those incrementally added
take this PHEV limitation into account. No ZEV credit trading and banking was included in this
analysis; each manufacturer was assumed to meet its ZEV obligation in MY2021 and MY2025
with vehicles produced for those model years. For the projected sales volumes used in this
analysis, the overall effect of the ZEV regulation is as shown in Table 1.31 through Table 1.34.
1-45
-------
Baseline and Reference Vehicle Fleets
References
1 EPA's Omega Model and input sheets are available at http://www.epa.gov/oms/climate/models.htm; Available in
the docket (Docket EPA-HQ-OAR-2015-0827).
2 U.S. Environmental Protection Agency (2012). "Regulatory Impact Analysis: Final Rulemaking for 2017-2025
Light-Duty Vehicle Greenhouse Gas Emission Standards and Corporate Average Fuel Economy Standards." EPA-
420-R-12-016, Chapter 10.
3 The baseline Excel file ("2015-2025 Production Summary and Data with Definitions") is available in the docket
(Docket EPA-HQ-OAR-2015-0827).
4 Department of Energy, Energy Information Administration, Annual Energy Outlook (AEO) Available at
http://www.eia.gov/forecasts/aeo/tables_ref.cfm (last accessed June. 15, 2016). The Department of Energy's Energy
Information Administration is a principal agency of the United States Federal Statistical System responsible for
collecting, analyzing, and disseminating energy information to promote sound policymaking, efficient markets, and
public understanding of energy and its interaction with the economy and the environment. The Energy Information
Administration's reports are the standard source, used government-wide, for such information and analysis.
5 EIA special projections Excel file ("EPAAEO2016SPECIAL 2021_Cases") is available in the docket (Docket
EPA-HQ-OAR-2015-0827).
6 The baseline Excel file ("2014-2025 Production Summary and Data with Definitions") is available in the docket
(Docket EPA-HQ-OAR-2015-0827).
7 Title 13, California Code of Regulations, Section 1962.2 "Zero-Emission Vehicle Standards for 2018 and
Subsequent Model Year Passenger Cars, Light-Duty Trucks, and Medium-Duty Vehicles."
8 Section 177 ZEV states: Connecticut, Maine, Maryland, Massachusetts, New York, New Jersey, Oregon, Rhode
Island, and Vermont.
9 Title 13, California Code of Regulations, Section 1962.2 "Zero-Emission Vehicle Standards for 2018 and
Subsequent Model Year Passenger Cars, Light-Duty Trucks, and Medium-Duty Vehicles."
1-46
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table of Contents
Chapter 2: Technology Costs, Effectiveness, and Lead Time Assessment 2-1
2.1 Overview 2-1
2.2 State of Technology and Advancements since the 2012 Final Rule 2-5
2.2.1 Individual Technologies and Key Developments 2-5
2.2.1.1 List of Technologies Considered 2-5
2.2.1.2 Descriptions of Technologies and Key Developments since the FRM 2-6
2.2.2 Engines: State of Technology 2-13
2.2.2.1 Overview of Engine Technologies 2-15
2.2.2.2 Sources of Engine Effectiveness Data 2-17
2.2.2.3 Low Friction Lubricants (LUB) 2-18
2.2.2.4 Engine Friction Reduction (EFR1, EFR2) 2-18
2.2.2.5 Cylinder Deactivation (DEAC) 2-19
2.2.2.6 Variable Valve Timing (VVT) Systems 2-19
2.2.2.6.1 Intake Cam Phasing (ICP) 2-20
2.2.2.6.2 Coupled Cam Phasing (CCP) 2-20
2.2.2.6.3 Dual Cam Phasing (DCP) 2-20
2.2.2.6.4 Variable Valve Lift (VVL) 2-20
2.2.2.7 GDI, Turbocharging, Downsizing and Cylinder Deactivation 2-21
2.2.2.8 EGR 2-30
2.2.2.9 Atkinson Cycle 2-31
2.2.2.10 Miller Cycle 2-35
2.2.2.11 Light-duty Diesel Engines 2-38
2.2.2.12 Thermal Management 2-41
2.2.2.13 Reduction of Friction and Other Mechanical Losses 2-42
2.2.2.14 Potential Longer-Term Engine Technologies 2-43
2.2.3 Transmissions: State of Technology 2-44
2.2.3.1 Background 2-44
2.2.3.2 Transmissions: Summary of State of Technology 2-45
2.2.3.3 Sources of Transmission Effectiveness Data 2-46
2.2.3.4 Sources of GHG Emission Improvements: Reduction in Parasitic Losses, Engine
Operation, and Powertrain System Design 2-48
2.2.3.5 Automatic Transmissions (ATs) 2-50
2.2.3.6 Manual Transmissions (MTs) 2-53
2.2.3.7 Dual Clutch Transmissions (DCTs) 2-53
2.2.3.8 Continuously Variable Transmissions (CVTs) 2-54
2.2.3.9 Transmission Parasitic Losses 2-57
2.2.3.9.1 Losses in ATs 2-57
2.2.3.9.2 Losses in DCTs 2-58
2.2.3.9.3 Losses in CVTs 2-58
2.2.3.9.4 Neutral Idle Decoupling 2-59
2.2.3.10 Transmission Shift Strategies 2-59
2.2.3.11 Torque Converter Losses and Lockup Strategy 2-59
2.2.4 Electrification: State of Technology 2-60
2.2.4.1 Overview of Chapter 2-61
2.2.4.2 Overview of Electrification Technologies 2-62
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.2.4.3 Non-Battery Components of Electrified Vehicles 2-65
2.2.4.3.1 Propulsion Components 2-66
2.2.4.3.2 Power Electronics 2-68
2.2.4.3.3 Industry Targets for Non-Battery Components 2-72
2.2.4.4 Developments in Electrified Vehicles 2-74
2.2.4.4.1 Non-hybrid Stop-Start 2-74
2.2.4.4.2 Mild Hybrids 2-77
2.2.4.4.3 Strong Hybrids 2-82
2.2.4.4.4 Plug-in Hybrids 2-86
2.2.4.4.5 Battery Electric Vehicles 2-93
2.2.4.4.6 Relating Power to Acceleration Performance 2-104
2.2.4.5 Developments in Electrified Vehicle Battery Technology 2-106
2.2.4.5.1 Battery Chemistry 2-107
2.2.4.5.2 Pack Topology, Cell Capacity and Cells per Module 2-110
2.2.4.5.3 Usable Energy Capacity 2-113
2.2.4.5.4 Thermal Management 2-119
2.2.4.5.5 Pack Voltage 2-120
2.2.4.5.6 Electrode Dimensions 2-121
2.2.4.5.7 Pack Manufacturing Volumes 2-123
2.2.4.5.8 Potential Impact of Lithium Demand on Battery Cost 2-125
2.2.4.5.9 Evaluation of Draft TAR Battery Cost Projections 2-126
2.2.4.6 Fuel Cell Electric Vehicles 2-132
2.2.5 Aerodynamics: State of Technology 2-133
2.2.5.1 Background 2-133
2.2.5.2 Industry Developments 2-134
2.2.5.3 Feasibility of Aerodynamic Improvements 2-138
2.2.5.4 Results of U.S.-Canada Joint Test Program 2-139
2.2.6 Tires: State of Technology 2-142
2.2.6.1 Background 2-142
2.2.6.2 Industry Developments 2-142
2.2.7 Mass Reduction: State of Technology 2-145
2.2.7.1 Overview of Mass Reduction Technologies 2-145
2.2.7.2 Mass Reduction Feasibility 2-149
2.2.7.3 Market Implementation of Mass Reduction 2-151
2.2.7.4 Holistic Vehicle Mass Reduction and Cost Studies 2-152
2.2.7.4.1 EPA Holistic Vehicle Mass Reduction/Cost Studies 2-157
2.2.7.4.1.1 Phase 2 Low Development Midsize CUV Updated Study and
Supplement 2-157
2.2.7.4.1.2 Light Duty Pickup Truck Light-Weighting Study 2-160
2.2.7.4.2 NHTSA Holistic Vehicle Mass Reduction/Cost Studies 2-163
2.2.7.4.3 ARB Holistic Vehicle Mass Reduction/Cost Study 2-163
2.2.7.4.4 Aluminum Association Midsize CUV Aluminum BIW Study 2-164
2.2.7.4.5 Comparison of Data for Lightweight Car/CUV with Aluminum BIW.. 2-166
2.2.7.4.6 DOE/Ford/Magna MMLV Mach 1 and Mach 2 Lightweighting Research
Projects 2-168
2.2.7.4.6.1 Mach I 2-170
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.2.7.4.6.2 Ylacli 2 2-172
2.2.1 A.1 Technical Cost Modeling Report by DOE/INL/IBIS on 40 Percent-45
Percent Mass Reduced Vehicle 2-174
2.2.7.4.8 Mass Reduction Spectrum Analysis and Process Cost Modeling Report by
DOE/IBIS/Energetics/INL 2-175
2.2.7.4.9 Studies to Determine Potential Mass Addition for IIHS Small Overlap 2-176
2.2.7.4.9.1 NHTSA Mass Add Study for a Passenger Car to Achieve a "Good"
Rating on the IIHS Small Overlap 2-177
2.2.7.4.9.2 Transport Canada Mass Add Study for a Light Duty Truck to Achieve a
"Good" Rating on the IIHS Small Overlap 2-178
2.2.7.5 Potential Lightweight Recyclable Composite Fiber Material 2-180
2.2.8 State of Other Vehicle Technologies 2-181
2.2.8.1 Electrified Power Steering: State of Technology 2-181
2.2.8.2 Improved Accessories: State of Technology 2-181
2.2.8.3 Secondary Axle Disconnect: State of Technology 2-182
2.2.8.3.1 Background 2-182
2.2.8.3.2 Developments in AWD Technology 2-183
2.2.8.4 Low-Drag Brakes: State of Technology 2-186
2.2.9 Air Conditioning Efficiency and Leakage Credits 2-187
2.2.9.1 A/C Efficiency Credits 2-188
2.2.9.1.1 Manufacturer Utilization of A/C Efficiency Credits 2-188
2.2.9.1.2 Eligibility for A/C Efficiency Credits 2-190
2.2.9.1.3 The AC 17 Test Procedure 2-192
2.2.9.1.4 Summary 2-198
2.2.9.2 A/C Leakage Reduction and Alternative Refrigerant Substitution 2-199
2.2.9.2.1 Leakage 2-199
2.2.9.2.2 Low-GWP Refrigerants 2-200
2.2.9.2.3 Conclusions 2-201
2.2.10 Off-cycle Technology Credits 2-201
2.2.10.1 Off-cycle Credits Program 2-201
2.2.10.1.1 Off-cycle Credits Program Overview 2-201
2.2.10.2 Use of Off-cycle Technologies to Date 2-203
2.3 GHG Technology Assessment 2-206
2.3.1 Fundamental Assumptions 2-206
2.3.1.1 Technology Time Frame and Measurement Scale for Effectiveness and Cost... 2-
206
2.3.1.2 Performance Assumptions 2-207
2.3.1.3 Fuels 2-209
2.3.1.4 Vehicle Classification 2-212
2.3.2 Approach for Determining Technology Costs 2-214
2.3.2.1 Direct Manufacturing Costs 2-215
2.3.2.1.1 Costs from Tear-down Studies 2-215
2.3.2.1.2 Electrified Vehicle Battery Costs 2-217
2.3.2.1.3 Specific DMC Updates since the Draft TAR 2-218
2.3.2.1.4 Approach to Cost Reduction through Manufacturer Learning 2-218
2.3.2.2 Indirect Costs 2-223
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.3.2.2.1 Methodologies for Determining Indirect Costs 2-223
2.3.2.2.2 Indirect Cost Estimates Used in this Analysis 2-225
2.3.2.3 Maintenance and Repair Costs 2-229
2.3.2.3.1 Maintenance Costs 2-229
2.3.2.3.2 Repair Costs 2-230
2.3.2.4 Costs Updated to 2015 Dollars 2-230
2.3.3 Approach for Determining Technology Effectiveness 2-231
2.3.3.1 Vehicle Benchmarking 2-231
2.3.3.1.1 Detailed Vehicle Benchmarking Process 2-232
2.3.3.1.1.1 Engine Testing 2-233
2.3.3.1.1.2 Transmission Testing 2-234
2.3.3.1.2 Development of Model Inputs from Benchmarking Data 2-237
2.3.3.1.2.1 Engine Data 2-237
2.3.3.1.2.2 Engine Map 2-237
2.3.3.1.2.3 Inertia 2-238
2.3.3.1.2.4 Transmission Data 2-239
2.3.3.1.2.5 Gear Efficiency and Spin Losses 2-239
2.3.3.1.2.6 Torque Converter 2-240
2.3.3.1.3 Vehicle Benchmarking Summary 2-241
2.3.3.2 Classification of Vehicles for Effectiveness 2-242
2.3.3.2.1.1 Significance of Power-to-Weight Ratio and Road-Load Power
Attributes 2-242
2.3.3.2.1.2 Effect of Changing Power-to-Weight Ratio 2-243
2.3.3.2.1.3 Effect of Advanced Technologies 2-245
2.3.3.2.1.4 Advanced Technology Trade-Off Curves 2-247
2.3.3.2.2 Definition of Effectiveness Classes 2-249
2.3.3.2.3 Comparison to Draft TAR Classification Approach and Exemplar Vehicles
2-251
2.3.3.3 ALPHA Vehicle Simulation Model 2-255
2.3.3.3.1 General ALPHA Description 2-256
2.3.3.3.2 Detailed ALPHA Model Description 2-256
2.3.3.3.2.1 Ambient System 2-257
2.3.3.3.2.2 Driver System 2-258
2.3.3.3.2.3 Powertrain System 2-258
2.3.3.3.2.3.1 Engine Subsystem 2-258
2.3.3.3.2.3.2 Electric Subsystem 2-259
2.3.3.3.2.3.3 Accessories Subsystem 2-260
2.3.3.3.2.3.4 Transmission Subsystem 2-260
2.3.3.3.2.3.4.1 Transmission Gear Selection 2-260
2.3.3.3.2.3.4.2 Launch Clutch Model 2-261
2.3.3.3.2.3.4.3 Gearbox Model 2-261
2.3.3.3.2.3.4.4 Torque Converter Model 2-262
2.3.3.3.2.3.4.5 Automatic Transmission & Controls 2-262
2.3.3.3.2.3.4.6 DCT Transmission & Control 2-262
2.3.3.3.2.3.4.7 CVT Transmission & Control 2-262
2.3.3.3.2.3.4.8 Driveline 2-262
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.3.3.3.2.3.5 Vehicle System 2-263
2.3.3.3.3 Energy Auditing 2-263
2.3.3.3.4 ALPHA Simulation Runs 2-264
2.3.3.3.5 Post-processing 2-265
2.3.3.3.6 Vehicle Component Vintage 2-266
2.3.3.3.7 Additional Verification 2-267
2.3.3.3.8 Key Public Comments Related to the ALPHA Model 2-268
2.3.3.4 Determining Technology Effectiveness for MY2022-2025 2-271
2.3.3.5 Lumped Parameter Model 2-274
2.3.3.5.1 Approach for Modeling Incremental Effectiveness 2-274
2.3.3.5.2 Calibration of LPM using ALPHA model 2-276
2.3.3.5.3 Lumped Parameter Model Usage in OMEGA 2-277
2.3.3.5.4 Appropriateness of LPM Effectiveness Modeling for the Overall Fleet2-279
2.3.4 Data and Assumptions Used in the GHG Assessment 2-287
2.3.4.1 Engines: Data and Assumptions for this Assessment 2-287
2.3.4.1.1 Low Friction Lubricants (LUB) 2-287
2.3.4.1.2 Engine Friction Reduction (EFR1, EFR2) 2-288
2.3.4.1.3 Cylinder Deactivation (DEAC) 2-289
2.3.4.1.4 Intake Cam Phasing (ICP) 2-290
2.3.4.1.5 Dual Cam Phasing (DCP) 2-291
2.3.4.1.6 Discrete Variable Valve Lift (DVVL) 2-292
2.3.4.1.7 Continuously Variable Valve Lift (CVVL) 2-292
2.3.4.1.8 Atkinson Cycle Engines in Non-HEV Applications 2-293
2.3.4.1.8.1 Effectiveness Data Used and Basis for Assumptions 2-293
2.3.4.1.8.2 Cost Data Used and Basis for Assumptions 2-307
2.3.4.1.8.3 Basis for Feasibility Assumptions 2-308
2.3.4.1.9 GDI, Turbocharging, Downsizing 2-311
2.3.4.1.9.1 Effectiveness Data Used and Basis for Assumptions 2-311
2.3.4.1.9.2 Cost Data Used and Basis for Assumptions 2-321
2.3.4.1.9.3 Basis for Feasibility Assumptions 2-324
2.3.4.2 Transmissions: Data and Assumptions for this Proposed Determination 2-325
2.3.4.2.1 Assessment and Classification of Automated Transmissions (AT, AMT,
DCT, CVT) 2-326
2.3.4.2.2 Effectiveness Values for TRX11 and TRX21 2-329
2.3.4.2.3 Effectiveness Values for TRX12 and TRX22 2-332
2.3.4.2.4 Technology Applicability and Costs 2-333
2.3.4.3 Electrification: Data and Assumptions for this Assessment 2-335
2.3.4.3.1 Cost and Effectiveness for Non-hybrid Stop-Start 2-335
2.3.4.3.2 Cost and Effectiveness for Mild Hybrids 2-337
2.3.4.3.3 Cost and Effectiveness for Strong Hybrids 2-339
2.3.4.3.4 Cost and Effectiveness for Plug-in Hybrids 2-341
2.3.4.3.5 Cost and Effectiveness for Battery Electric Vehicles 2-342
2.3.4.3.6 Cost of Non-Battery Components for xEVs 2-345
2.3.4.3.7 Cost of Batteries for xEVs 2-355
2.3.4.3.7.1 Battery Sizing Methodology forBEVs andPHEVs 2-359
2.3.4.3.7.2 Battery Sizing Methodology forHEVs 2-382
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.3.4.3.7.3 ANL BatPaC Battery Design and Cost Model 2-383
2.3.4.3.7.4 Assumptions and Inputs to BatPaC 2-385
2.3.4.3.7.5 Battery Cost Projections forxEVs 2-389
2.3.4.3.7.6 Discussion of Battery Cost Projections 2-398
2.3.4.3.7.7 Battery Pack Costs Used in OMEGA 2-399
2.3.4.3.7.8 Electrified Vehicle Costs Used In OMEGA (Battery + Non-battery
Items) 2-403
2.3.4.4 Aerodynamics: Data and Assumptions for this Assessment 2-405
2.3.4.5 Tires: Data and Assumptions for this Assessment 2-409
2.3.4.6 Mass Reduction: Data and Assumptions for this Assessment 2-411
2.3.4.6.1 Updates to Mass Reduction for the Current Analysis 2-411
2.3.4.6.2 Mass Reduction Costs used in OMEGA 2-413
2.3.4.7 Other Vehicle Technologies 2-421
2.3.4.7.1 Electrified Power Steering: Data and Assumptions for this Assessment 2-
421
2.3.4.7.2 Improved Accessories: Data and Assumptions for this Assessment 2-421
2.3.4.7.3 Secondary Axle Disconnect: Data and Assumptions for this Assessment.. 2-
422
2.3.4.7.4 Low Drag Brakes: Data and Assumptions for this Assessment 2-422
2.3.4.8 Air Conditioning: Data and Assumptions for this Assessment 2-423
2.3.4.9 Additional Off-cycle Credits and Costs 2-423
2.3.4.10 Cost Tables for Individual Technologies Not Presented Above 2-425
Table of Figures
Figure 2.1 Light-duty Vehicle Engine Technology Penetration since the 2012 Final Rule 2-15
Figure 2.2 Comparison of BTE for A Representative MY2008 2.4L 14 NA DOHC PFI 4-valve/cyl. Engine with
Intake Cam Phasing (Left) and a GM Ecotec 2.5L NA GDI Engine with Dual Camshaft Phasing
(Right) 2-22
Figure 2.3 Comparison of BTE for A Representative MY2010 3.5L V6 NA PFI 4-valve/cyl. Engine (Left) and a
Toyota 2GR-FSE GDI/PFI Engine with Dual Camshaft Phasing (Right) 2-23
Figure 2.4 Graphical Representation Showing How Cylinder Deactivation Moves Engine Operation to Regions of
Operation with Improved Fuel Consumption over the UDDS Regulatory Drive Cycle (shaded area). 2-
24
Figure 2.5 Comparison of BTE for A Representative MY2010 5.4L V8 NA PFI 3-valve/cyl. Engine (Left) and a
Ford 2.7L V6 Ecoboost Turbocharged, GDI Engine With Dual Camshaft Phasing (Right) 2-26
Figure 2.6 Engine Speed and BMEP Points Taken from 10 Hz-sampled data over the UDDS and HwFET
Superimposed Over BTE Data From a Ford 2.7L V6 Ecoboost Turbocharged, GDI Engine With Dual
Camshaft Phasing (Right) 2-26
Figure 2.7 Comparison of BTE for A Representative MY2010 2.4L NA PFI Engine (Left) and A Modern, 1.0L
Turbocharged, Downsized GDI Engine (Right) 2-27
Figure 2.8 Comparison of BTE for A Representative MY2010 2.4L NA PFI Engine (Left) and A Modern, 1.5L
Turbocharged, Downsized GDI Engine (Right) 2-28
Figure 2.9 Typical Turbocharger Compressor Map Showing How Pressure And Flow Characteristics Can Be
Matched Over a Broader Range of Engine Operation Via Surge Improvement and Higher Operational
Speed 2-29
Figure 2.10 Cross Sectional View of a Honeywell VNT Turbocharger 2-29
Figure 2.11 A Functional Schematic Example of a Turbocharged Engine Using Two Variants of External EGR. 2-31
Figure 2.12 Comparison of the Timing of Valve Events for Otto-Cycle (black and orange lines) and LIVC
Implementations of Atkinson- Or Miller-Cycle (black and green lines) 2-32
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.13 Diagrams of Cylinder Pressure Vs. Cylinder Volume For a Conventional Otto-Cycle SI Engine (orange
line) Compared to a LIVC Implementation of Atkinson Cycle (green line) Highlighting the Reduction
in Pumping Losses 2-32
Figure 2.14 Comparison of BTE for a Representative MY2010 2.4L NAPFI Engine (left) and a 2.0L NA GDI
LIVC Atkinson Cycle Engine (right) tested by EPA 2-34
Figure 2.15 Measured effective compression ratio for 2.0L NA GDI LIVC Atkinson Cycle Engine (right) tested by
EPA 2-34
Figure 2.16 A Comparison of BSFC Maps Measured For The 2.0L 13: ICR SKYACTIV-G Engine (left) and
Modeled For A 1.0L Ricardo "EGRB Configuration" (right) 2-35
Figure 2.17 Comparison of BTE for Downsized, Turbocharged GDI Engines 2-36
Figure 2.18 Comparison of BTE for A Representative MY2010 3.5L NA PFIV6 Engine (Left) And A Downsized
2.0L 14 Miller Cycle Engine (Right) 2-37
Figure 2.19 Comparison of BTE for 2015 Turbocharged, Downsized GDI (left) and 2017 Miller Cycle (right)
variants of the same engine family, the 2.0L VW EA888 2-38
Figure 2.20 Comparison Of BTE For A Downsized SI 2.0L 14 Miller Cycle Engine (Left) And A 1.7L 14
Turbocharged Diesel Engine With HPCR, Low And High Pressure Loop Cegr, And VNT
Turbocharger (Right) 2-40
Figure 2.21 Exhaust Manifold Integrated Into a Single Casting with the Cylinder Head 2-42
Figure 2.22 Transmission Technology Production Share, 1980 - 2015 2-45
Figure 2.23 Average Torque Losses (Left) And Efficiency (Right) In Each Gear For An Eight-Speed 845RE
Transmission From A Ram, Tested At 100 °C And With Line Pressures Matching Those Measured
In-Use In The Vehicle. Torque Losses Were Averaged Over 1000 Rpm - 2500 Rpm. This
Transmission Is A Clone of the ZF 8HP45 2-47
Figure 2.24 Engine Operating Conditions for Six-Speed (Left) and Eight-Speed (Right) Automatic Transmissions
on the FTP-75 Drive Cycle 2-49
Figure 2.25 ZF 8HP70 Automatic Transmission 2-50
Figure 2.26 Average Number of Transmission Gears for New Vehicles (excluding CVTs) 2-51
Figure 2.27 Generic Dual Clutch Transmission 2-54
Figure 2.28 (a) Toyota CVT (b) Generic CVT sketch 2-55
Figure 2.29 ZF Torque Converter Cutaway 2-60
Figure 2.30 Hybrid System Direct Manufacturing Cost Projection (ICCT, 2015) 2-85
Figure 2.31 Battery Gross Capacity and Estimated AER or Equivalent for MY2012-2017 PHEVs 2-89
Figure 2.32 Comparison of MY2012-2016 PHEV Battery Capacities to Draft TAR Estimates 2-93
Figure 2.33 Battery Gross Capacity and EPA Estimated Range for MY2012-2017 BEVs 2-98
Figure 2.34 Comparison of 2012-2016MY BEV Battery Gross Capacities to Draft TAR Estimates 2-103
Figure 2.35 Acceleration Performance of MY2012-2017 PEVs Compared To Targets Generated By Malliaris
Equation 2-105
Figure 2.36 Comparison of Draft TAR Projected Battery Cost per kWh to Estimates Reviewed by Nykvist &
Nilsson 2-127
Figure 2.37 Comparison of Estimated GM/LG Pack-Level Costs to 2012 FRM and Draft TAR Estimates for
BEV150/200 2-129
Figure 2.38 Relationship between Wet Grip Index and Rolling Resistance for Winter Tires from Transport
Canada/NRCan Study 2-144
Figure 2.39 Change in Adjusted Fuel Economy, Weight and Horsepower for MY1975-2015 2-146
Figure 2.40 Estimated Vehicle Material Change over Time 2012-2025 - Ducker Worldwide435 2-147
Figure 2.41 Forecast of Automotive Market Consumption of Composites 2-148
Figure 2.42 Magnesium Growth Expectations through 2025 (Ducker Worldwide) 2-148
Figure 2.43 Mass Reduction Cost Curve ($/lb.) for 2017-2025 LD GHG Joint Technical Support Document... 2-153
Figure 2.44 Original Phase 2 Low Development Midsize CUV Lightweighting Cost Curve 2-158
Figure 2.45 Revised Cost Curve for the Midsize CUV Light Weighted Vehicle 2-159
Figure 2.46 Cost Curve Figure from CAR: "A Cost Curve for Lightweighting That Is Broadly Supported" 2-160
Figure 2.47 Light Duty Pickup Truck Lightweighting Study Results 2-161
Figure 2.48 Light Duty Pickup Truck Lightweighting Cost Curve 2-162
Figure 2.49 Light Duty Pickup Truck Lightweighting Study Secondary Mass 2-163
Figure 2.50 Phase 2 High Development BIW - Lotus Engineering 2-164
Figure 2.51 Midsize CUV Baseline vs Midsize CUV Aluminum Intensive Vehicle 2-165
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.52 Summary Table of Mass Reduction and Cost for Aluminum BIW and Closure Components 2-166
Figure 2.53 Car/CUV DMC Curve Extended to Points with Aluminum BIW 2-167
Figure 2.54 MMLV Structures Weight Comparison BIW, Closure, Chassis, Bumper 2-169
Figure 2.55 Mach II Mixed Material BIW and Closure Design (brown is carbon fiber)475 2-174
Figure 2.56 Technical Cost Modeling Results for 40 Percent to 45 Percent Lightweighting Scenario (Based on
Mach 1/Mach 2 Project Technologies) 2-175
Figure 2.57 Results for Weight Reduction Strategies by Risk Factor and Cost of Weight Savings478 2-176
Figure 2.58 Post-test Laboratory Vehicle of IIHS Small Overlap Test 2-176
Figure 2.59 MY2013 Silverado 1500 IIHS Small Overlap Test Crash Before and During 2-178
Figure 2.60 Converting the Actual Crash Event to a Model 2-179
Figure 2.61 Light Weighted Model in the IIHS Small Overlap Crash Test 2-179
Figure 2.62 Results of the Project Models from Baseline to Light Weighted on the IIHS Small Overlap480 2-180
Figure 2.63 Summary of AWD Efficiency Improvement Potentials 2-184
Figure 2.64 Contribution of Individual AWD Driveline Components to Total Additional Vehicle Mass 2-185
Figure 2.65 Variability of AC17 Round Robin Testing on 2011 Ford Explorer, A/C On 2-193
Figure 2.66 Variability of AC17 Round Robin Testing on 2011 Ford Explorer, A/C Off 2-194
Figure 2.67 Variability of AC17 Round Robin Testing on 2011 Ford Explorer, Delta between A/C on and Off 2-194
Figure 2.68 The "Null Technology Package" and Measurement Scale for Cost and Effectiveness 2-206
Figure 2.69 Chevy Malibu Undergoing Dynamometer Testing 2-233
Figure 2.70 Engine Test Cell Setup 2-234
Figure 2.71 Engine Map Points 2-234
Figure 2.72 GM6T40 Transmission during Testing 2-235
Figure 2.73 Transmission Efficiency Data at 93 C and 10 Bar Line Pressure 2-235
Figure 2.74 Torque Converter Torque Ratio and Normalized K Factor versus Speed Ratio 2-236
Figure 2.75 Transmission Spin Losses at 93C 2-237
Figure 2.76 Chevy Malibu 2.5L BSFC Map 2-238
Figure 2.77 Engine Spin down Inertia Test 2-239
Figure 2.78 Gear Efficiency Data at 93 C and 10 bar Line Pressure 2-240
Figure 2.79 Torque Converter Drive and Back-Drive Torque Ratio and Normalized K Factor versus Speed Ratio . 2-
241
Figure 2.80 Engine "heat map" for baseline vehicle, showing engine operation over the FTP and HWFET 2-243
Figure 2.81 Two-cycle heat maps for two different power/weight ratio vehicles 2-244
Figure 2.82 CO2 and performance time sum as a function of power/weight ratio 2-245
Figure 2.83 CO2 as a function of acceleration performance time sum 2-245
Figure 2.84 Engine Heat maps for the baseline PFI engine and a 24-bar turbo downsized engine 2-246
Figure 2.85 Engine operation heat maps for the turbo downsized engine with eight-speed transmission 2-247
Figure 2.86 CO2 as a function of performance time sum for PFI, GDI, and turbo downsized engines 2-247
Figure 2.87 Reduction in CO2, comparing a turbo downsized engine to a GDI engine with similar acceleration
performance 2-249
Figure 2.88 Production Volume Distribution of Power-to-Weight Ratios inMY2015 Fleet 2-250
Figure 2.89 Production Volume Distribution of Road Load Horsepower at 50mphinMY2015 Fleet 2-251
Figure 2.90 MPW HRL Class Effectiveness Change as a Function of Power-to-Weight Ratio 2-253
Figure 2.91 MY2015 Production-weighted Distributions of Power-to-Weight Ratio Using Draft TAR and Proposed
Determination Classification Approaches 2-254
Figure 2.92 MY2015 Production-weighted Distributions of Road Load Horsepower Using Draft TAR and
Proposed Determination Classification Approaches 2-255
Figure 2.93 ALPHA Model Top Level View 2-257
Figure 2.94 ALPHA Conventional Vehicle Powertrain Components 2-258
Figure 2.95 Sample ALPHA Energy Audit Report 2-264
Figure 2.96 Example: Difference in 2016, Between Bags 1 and 3 of the FTP, from the Test Car List 2-266
Figure 2.97 Example ALPHA Model UDDS Simulation Observation Display 2-267
Figure 2.98 Distribution of Gasoline Powertrain Efficiencies for Vehicles in MY2015 2-282
Figure 2.99 Distribution of Gasoline Powertrain Efficiencies for Vehicles in the OMEGA Compliance Analysis for
MY2025 Standards 2-283
Figure 2.100 LPM and ALPHA Package Effectiveness Comparison for Vehicles and Throughout Distribution of
Powertrain Efficiencies 2-286
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.101 2.0L 14 Mazda SKYACTIV-G Engine Undergoing Engine Dynamometer Testing at the EPA-NVFEL
Facility 2-287
Figure 2.102 Comparison of a 2.0L Mazda SKYACTIV-G engine with a 13:1 geometric compression ratio to
engine simulation results of a comparable engine with a 1-point increase in geometric compression
ratio (14:1) and cooled, low-pressure EGR 2-294
Figure 2.103 Comparison of a 2.0L Mazda SKYACTIV-G engine with a 13:1 geometric compression ratio to
engine simulation results of a comparable engine with a 1-point increase in geometric compression
ratio (14:1), cooled, low-pressure EGR and cylinder deactivation with operation on 2 cylinders at
below 5-bar BMEP and 1000 - 3000 rpm 2-295
Figure 2.104 Mazda 2.0L SKYACTIV-G engine with 14:1 geometric compression ratio, cooled low-pressure
external EGR system, DCO ignition system, and developmental engine management system
undergoing engine dynamometer testing at the U.S. EPA-NVFEL facility in Ann Arbor, MI 2-296
Figure 2.105 Modeled internal EGR and cEGR rates (in percent) from the draft TAR engine simulation (left top and
left bottom, respectively) compared to internal EGR and cEGR rates achieved during engine testing
(right top and right bottom, respectively) 2-297
Figure 2.106 Modeled CO2 effectiveness for internal and cEGR from the draft TAR engine simulation (left)
compared to CO2 effectiveness achieved during engine testing (right) 2-297
Figure 2.107 CO2 effectiveness achieved during engine testing with cEGR and simulated 2-cylinder fixed cylinder
deactivation from 1000 to 3000 RPM and at less than 3.75 BMEP 2-298
Figure 2.108 "Difference map" comparison provided by AAM between EPA data generated using Tier 2
certification gasoline and "USCAR 91 RON data" for a Mazda SKYACTIV-G 2.0L engine. (AAM
I ig 15-1) 2-300
Figure 2.109 This figure was reproduced from "Figure B-2" of the AAM comments purporting to show a
discrepancy between the torque curves used in SAE 2016-01-0565 vs. those used within the ALPHA
model 2-305
Figure 2.110 This is a reproduction of AAM figure B-2 with EPA data for two engine dynamometer derived torque
curves (green and black dashed) as well the extent of modeled data points (orange, light-blue-dashed).
None of the data from SAE 2016-01-0565 matches the solid blue line from the AAM comments citing
SAE 2016-01-0565 2-306
Figure 2.111 Contour plot of BSFC in g/kW-hr versus engine speed and BMEP for the Ricardo "EBDI" engine
equipped with sequential turbocharging, DCP, DVVL, cEGR, IEM, and with a 10:1 compression ratio
using 98 RON Indolene 2-312
Figure 2.112 Schematic Representation of the Development of BSFC Mapping for TDS24 2-315
Figure 2.113 Comparison between a 1.15L 13 version of TDS24 (left) and the Honda L15B7 1.5L turbocharged,
GDI engine used in the 2017 Civic (right) 2-316
Figure 2.114 Comparison between a 1.15L 13 version of TDS24 (left) and the 2017 Golf 1.5L EA211 TSIEVO
Engine 2-316
Figure 2.115 Comparison between a 1.51L 13 version of TDS24 (left) and the 2017 Audi A3 2.0L 888-3B Engine
(right) 2-317
Figure 2.116 Comparison of the Different Transmission Types 2-327
Figure 2.117 2015 Dodge Charger Gearing Changes over the HWFET 2-332
Figure 2.118 EPA PEV Battery and Motor Sizing Method 2-362
Figure 2.119 Projected BEV Gross Battery Capacity per Unit Curb Weight Compared to Comparable BEVs... 2-381
Figure 2.120 Projected PHEV Gross Battery Capacity per Unit Curb Weight Compared To Comparable PHEVs .. 2-
382
Figure 2.121 Comparison of Estimated Pack-Converted GM/LG Costs to BEV150/200 Projections of 2012 FRM,
Draft TAR, and this Proposed Determination (PD) 2-398
Table of Tables
Table 2.1 U.S. DRIVE Targets for Electric Content Cost and Specific Power 2-73
Table 2.2 Trends in EPA-Estimated Range of PHEVs 2-87
Table 2.3 Driving Range of MY2012-2017 BEVs 2-94
Table 2.4 PEV Acceleration Performance Intended in the FRM and Projected Probable Performance 2-106
Table 2.5 Lithium-ion Battery Chemistries Available in ANL BatPaC 2-107
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.6 Estimated SOC swings for selected MY2012-2016 BEVs 2-117
Table 2.7 Examples of Conversion Factors for Cell Costs to Pack Costs 2-128
Table 2.8 Comparison of GM/LGChem Pack-Converted Cell Costs to FRM BEV150 Pack Cost 2-129
Table 2.9 Summary of Published Evidence of Battery Pack Cost and Pricing 2-132
Table 2.10 Aerodynamic Technologies Observed in Vehicles Investigated at the 2015 NAIAS 2-135
Table 2.11 Aerodynamic Features of the 2015 Nissan Murano 2-136
Table 2.12 Effect of Active Ride Height on SUV Aerodynamic Performance 2-137
Table 2.13 Aerodynamic Technology Effectiveness from Phase 1 of Joint Aerodynamics Program 2-140
Table 2.14 Examples of Mass Reduction in Selected Recent Redesigns (Compared to MY2008 Design) 2-151
Table 2.15 Agency-Sponsored Mass Reduction Project List since 2012 FRM 2-155
Table 2.16 Summary of the Automotive Aluminum 2025 2-166
Table 2.17 Three Aluminum Intensive Vehicle Design Summary - DMC ($), %MR and $/kg 2-168
Table 2.18 Gaps Identified by MMLVProject 2-169
Table 2.19 Safety Tests Performed on the Mach-1 2-171
Table 2.20 Mach-I Components to Maintain Frontal Crash Performance 2-172
Table 2.21 Mach II Design Vehicle Summary475 2-173
Table 2.22 Estimated Mass Increase to Meet IIHS SOL for 2010 Vehicle Classes 2-177
Table 2.23 Estimated Mass Increase to Meet IIHS SOL for 2020 Vehicle Classes 2-177
Table 2.24 Hardware Bench Testing Standards under Development by SAE Cooperative Research Program... 2-197
Table 2.25 Trends in Fleetwide Mobile Air Conditioner Leakage Credits and Average Leakage Rates 2-199
Table 2.26 Off-cycle Menu Technologies and CO2 Credits for Cars and Light Trucks 2-203
Table 2.27 Off-cycle Menu Technologies and CO2 Credits for Solar/Thermal Control Technologies for Cars and
Light Trucks 2-203
Table 2.28 Percent of 2015 Model Year Vehicle Production Volume with Credits from the Menu, by Manufacturer
& Technology (%) 2-204
Table 2.29 Off-Cycle Technology Credits from the Menu, by Manufacturer and Technology for MY 2015 (g/mi). 2-
204
Table 2.30 Test Fuel Specifications for Gasoline without Ethanol (from 40 CFR §86.113-04) 2-210
Table 2.31 Petroleum Diesel Test Fuel (from 40 CFR §86.113-94) 2-210
Table 2.32 ALPHA Classes for Characterizing Technology Effectiveness 2-212
Table 2.33 Curb Weight Classes for Characterizing Technology Cost 2-213
Table 2.34 Expanded Vehicle Types for Characterizing Technology Cost and Effectiveness 2-214
Table 2.35 Learning Effect Algorithms Applied to Technologies Used in this Analysis 2-221
Table 2.36 Year-by-year Learning Curve Factors for the Learning Curves Used in this Analysis 2-222
Table 2.37 Indirect Cost Multipliers Used in this Analysis 2-225
Table 2.38 Warranty and Non-Warranty Portions of ICMs 2-226
Table 2.39 Indirect Cost Markups (ICMs) and Near Term/Long Term Cutoffs Used in EPA's Analysis 2-226
Table 2.40 Mass Reduction Markup Factors used by EPA in this TSD 2-228
Table 2.41 Mass Reduction Indirect Cost Curves used by EPA for Cars Using ICMs (dollar values in 2013$). 2-228
Table 2.42 Mass Reduction Indirect Cost Curves used by EPA for Trucks Using ICMs (dollar values in 2013$).... 2-
228
Table 2.43 Maintenance Event Costs & Intervals (2015$) 2-229
Table 2.44 Implicit Price Deflators and Conversion Factors for Conversion to 2015$ 2-231
Table 2.45 Benchmark Vehicle Description 2-232
Table 2.46 Criteria for Classifying Vehicles by Power-to Weight ratio and Road Load Horsepower 2-250
Table 2.47 Characteristics of Exemplar Vehicles for the Six ALPHA Classes 2-251
Table 2.48 Change in Power-to-Weight and Road Load Horsepower of Exemplar Vehicles Relative to Draft TAR 2-
252
Table 2.49 MY2015 Summary Statistics of Power-to-Weight Ratio Using Draft TAR and Proposed Determination
Classification Approaches 2-253
Table 2.50 MY2015 Summary Statistics of Road Load Horsepower Using Draft TAR and Proposed Determination
Classification Approaches 2-254
Table 2.51 Example OMEGA Vehicle Technology Packages (values are for example only) 2-277
Table 2.52 Example Baseline Vehicle (values are for example only) 2-278
Table 2.53 Example Package Application Process (values are for example only) 2-278
Table 2.54 Example Subset of ALPHA/LPM Calibration Check Points for Vehicle Type 1 2-278
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.55: Parameters for Power-to-Weight Adjustment of Effectiveness Values in OMEGA 2-280
Table 2.56 Summary Statistics for Powertrain Efficiencies in MY2015 Baseline and OMEGA Compliance Analysis
of MY2025 Standards 2-283
Table 2.57 Summary Statistics for Powertrain Efficiencies in MY2015 Baseline and OMEGA Compliance Analysis
of MY2025 Standards 2-284
Table 2.58 Powertrain Efficiencies by ALPHA Class from MY2025 OMEGA Compliance Analysis 2-285
Table 2.59 Costs for Engine Changes to Accommodate Low Friction Lubes (dollar values in 2015$) 2-288
Table 2.60 Costs for Engine Friction Reduction Level 1 (dollar values in 2015$) 2-288
Table 2.61 Costs for Engine Friction Reduction Level 2 (dollar values in 2015$) 2-288
Table 2.62 Costs for Cylinder Deactivation (dollar values in 2015$) 2-290
Table 2.63 Costs for Intake Cam Phasing (dollar values in 2015$) 2-291
Table 2.64 Costs for Dual CamPhasing (dollar values in 2015$) 2-292
Table 2.65 Costs for Discrete Variable Valve Lift (dollar values in 2015$) 2-292
Table 2.66 Costs for Continuously Variable Valve Lift (dollar values in 2015$) 2-293
Table 2.67 Direct Manufacturing Costs (DMC) for Atkinson-2 Technology (2010$) 2-308
Table 2.68 Costs for Atkinson-2 Technology, Exclusive of Enablers such as Direct Inject and Valve Timing
Technologies (dollar values in 2015$) 2-308
Table 2.69 Specification of Ricardo 3.2L V6 Turbocharged, GDI "EBDI" Proof-of-concept Engine 2-311
Table 2.70 Partial summary of MY2015 vehicles withD/M at or below 0.9 L/ton 2-319
Table 2.71 Summary of C02 emissions from testing a Ford F150 2.7L turbocharged vehicle and a Honda Civic
1.5L vehicle on Tier 2 and Tier 3 fuels 2-320
Table 2.72 Costs for Gasoline Direct Injection on an 13 & 14 Engine (dollar values in 2015$) 2-321
Table 2.73 Costs for Gasoline Direct Injection on a V6 Engine (dollar values in 2015$) 2-321
Table 2.74 Costs for Gasoline Direct Injection on a V8 Engine (dollar values in 2015$) 2-321
Table 2.75 Costs for Turbocharging, 18/21 bar, I-Configuration Engine (dollar values in 2015$) 2-321
Table 2.76 Costs for Turbocharging, 18/21 bar, V-Configuration Engine (dollar values in 2015$) 2-321
Table 2.77 Costs for Turbocharging, 24 bar, I-Configuration Engine & for Miller-cycle I-Configuration Engine
(dollar values in 2015$) 2-322
Table 2.78 Costs for Turbocharging, 24 bar, V-Configuration Engine & for Miller-cycle V-Configuration Engine
(dollar values in 2015$) 2-322
Table 2.79 Costs for Downsizing as part of Turbocharging & Downsizing (dollar values in 2015$) 2-322
Table 2.80 Costs for Turbocharging & Downsizing (2015$) 2-323
Table 2.81 Costs for Miller Cycle (2015$) 2-323
Table 2.82 Costs for Cooled EGR (dollar values in 2015$) 2-323
Table 2.83 Costs for Valvetrain Conversions from non-DOHC to DOHC (dollar values in 2015$) 2-324
Table 2.84 Transmission Level Map 2-327
Table 2.85 TRX11 to TRX 22 Effectiveness Progression 2-333
Table 2.86 Costs for Transmission Improvements for all Vehicles (dollar values in 2015$) 2-334
Table 2.87 Comparison of Transmission Costs Using the 2012 FRM Methodology to Proposed Determination Costs
for Transmissions (2015$) 2-334
Table 2.88 GHG Technology Effectiveness of Stop-Start 2-336
Table 2.89 Costs for Stop-Start for Different Curb Weight Classes (dollar values in 2015$) 2-336
Table 2.90 GHG Technology Effectiveness of Mild Hybrids 2-339
Table 2.91 GHG Technology Effectiveness of Strong Hybrids 2-341
Table 2.92 Linear Regressions of Strong & Plug-in Hybrid Non-Battery System Direct Manufacturing Costs vs Net
Mass Reduction Applicable inMY2012 (2015$) 2-349
Table 2.93 Linear Regressions of Battery Electric Non-Battery System Direct Manufacturing Costs vs Net Mass
Reduction Applicable inMY2016 (2015$) 2-349
Table 2.94 Costs for MHEV48V Non-Battery Items (dollar values in 2015$) 2-350
Table 2.95 Costs for Strong Hybrid Non-Battery Items (dollar values in 2015$) 2-350
Table 2.96 Costs for 20 Mile Plug-in Hybrid Non-Battery Items (dollar values in 2015$) 2-351
Table 2.97 Costs for 40 Mile Plug-in Hybrid Non-Battery Items (dollar values in 2015$) 2-352
Table 2.98 Costs for 75 Mile BEV Non-Battery Items (dollar values in 2015$) 2-352
Table 2.99 Costs for 100 Mile BEV Non-Battery Items (dollar values in 2015$) 2-353
Table 2.100 Costs for 200 Mile BEV Non-Battery Items (dollar values in 2015$) 2-354
Table 2.101 Costs for In-Home Charger Associated with 20 Mile Plug-in Hybrid (dollar values in 2015$) 2-354
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.102 Costs for In-Home Charger Associated with 40 Mile Plug-in Hybrid (dollar values in 2015$) 2-355
Table 2.103 Costs for In-Home Charger Associated with All BEVs (dollar values in 2015$) 2-355
Table 2.104 Costs for Labor Associated with All In-Home Chargers for Plug-in & BEV (dollar values in 2015$).. 2-
355
Table 2.105 U.S. Drive Targets for Non-Battery Specific Power for 2015 and 2020 2-363
Table 2.106 Example Net Curb Weight Reduction for BEVs and PHEVs With 20% Mass Reduction Technology
Applied to Glider 2-365
Table 2.107 Changes to Baseline Curb Weights from Draft TAR to Proposed Determination 2-372
Table 2.108 Baseline ICE-Powertrain Weight Assumptions (Pounds), By Vehicle Class 2-373
Table 2.109 PEV Battery Sizing Assumptions and Changes from Draft TAR to Proposed Determination 2-375
Table 2.110 Example Changes in Projected PEV Battery Capacity and Motor Power, Draft TAR to Proposed
Determination (20% weight reduction case) 2-376
Table 2.111 Examples of Pack-Level Specific Energy Calculated By BatPac for Selected PEV Configurations (0%
WR) 2-377
Table 2.112 Examples of Pack-Level Specific Energy Calculated By BatPac for Selected PEV Configurations (20%
WR) 2-377
Table 2.113. TSD Projected Battery Capacities and Assumed Curb Weights, 0% Nominal Weight Reduction.. 2-378
Table 2.114 TSD Projected Battery Capacities and Assumed Curb Weights, 20% Nominal Weight Reduction. 2-378
Table 2.115 Battery Design Assumptions Input to BatPaC and Changes from Draft TAR to Proposed Determination
2-389
Table 2.116 Average Change in Projected Battery Pack DMC from Draft TAR to Proposed Determination 2-390
Table 2.117 Estimated Direct Manufacturing Costs inMY2025 forBEV75 Battery Packs 2-391
Table 2.118 Estimated Direct Manufacturing Costs in MY2025 for BEV100 Battery Packs 2-393
Table 2.119 Estimated Direct Manufacturing Costs in MY2025 for BEV200 Battery Packs 2-394
Table 2.120 Estimated Direct Manufacturing Costs in MY2025 for PHEV20 Battery Packs 2-395
Table 2.121 Estimated Direct Manufacturing Costs inMY2025 forPHEV40 Battery Packs 2-396
Table 2.122 Estimated Direct Manufacturing Costs inMY2017 for strong HEV Battery Packs 2-397
Table 2.123 Linear Regressions of Strong Hybrid Battery System Direct Manufacturing Costs vs Net Mass
Reduction Applicable inMY2017 (2015$) 2-399
Table 2.124 Linear Regressions of Battery Electric Battery System Direct Manufacturing Costs vs Net Mass
Reduction Applicable inMY2025 (2015$) 2-399
Table 2.125 Costs for MHEV48V Battery (dollar values in 2015$) 2-399
Table 2.126 Costs for Strong Hybrid Batteries (dollar values in 2015$) 2-399
Table 2.127 Costs for 20 Mile Plug-in Hybrid Batteries (dollar values in 2015$) 2-400
Table 2.128 Costs for 40 Mile Plug-in Hybrid Batteries (dollar values in 2015$) 2-401
Table 2.129 Costs for 75 Mile BEV Batteries (dollar values in 2015$) 2-401
Table 2.130 Costs for 100 Mile BEV Batteries (dollar values in 2015$) 2-402
Table 2.131 Costs for 200 Mile BEV Batteries (dollar values in 2015$) 2-403
Table 2.132 Full System Costs for 48Y Mild Hybrids (2015$) 2-404
Table 2.133 Full System Costs for Strong Hybrids (2015$) 2-404
Table 2.134 Full System Costs for 20 Mile Plug-in Hybrids, Including Charger & Charger Labor (2015$) 2-404
Table 2.135 Full System Costs for 40 Mile Plug-in Hybrids, Including Charger & Charger Labor (2015$) 2-404
Table 2.136 Full System Costs for 75 Mile BEVs, Including Charger & Charger Labor (2015$) 2-405
Table 2.137 Full System Costs for 100 Mile BEVs, Including Charger & Charger Labor (2015$) 2-405
Table 2.138 Full System Costs for 200 Mile BEVs, Including Charger & Charger Labor (2015$) 2-405
Table 2.139 MY2015 Aerodynamic Drag Area Statistics and Cutoff Values by Size Classs 2-407
Table 2.140 Aerodynamic Drag Reduction Between Aero levels 0,1, and 2 by Size Class 2-408
Table 2.141 CO2 Efficiency Improvement per 10% Aero Improvement per Vehicle Classification 2-409
Table 2.142 Costs for Aero Technologies (dollar values in 2015$) 2-409
Table 2.143 Costs for Lower Rolling Resistance Tires (dollar values in2015$) 2-411
Table 2.144 Costs for 5 Percent Mass Reduction for Vehicle Types using the Car Cost Curve (2015$) 2-413
Table 2.145 Costs for 10 Percent Mass Reduction for Vehicle Types using the Car Cost Curve (2015$) 2-415
Table 2.146 Costs for 15 Percent Mass Reduction for Vehicle Types using the Car Cost Curve (2015$) 2-416
Table 2.147 Costs for 20 Percent Mass Reduction for Vehicle Types using the Car Cost Curve (2015$) 2-417
Table 2.148 Costs for 5 Percent Mass Reduction for Vehicle Types using the Truck Cost Curve (2015$) 2-419
Table 2.149 Costs for 10 Percent Mass Reduction for Vehicle Types using the Truck Cost Curve (2015$) 2-419
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.150 Costs for 15 Percent Mass Reduction for Vehicle Types using the Truck Cost Curve (2015$) 2-419
Table 2.151 Costs for 20 Percent Mass Reduction for Vehicle Types using the Truck Cost Curve (2015$) 2-420
Table 2.152 Costs for Electric Power Steering (dollar values in 2015$) 2-421
Table 2.153 Costs for Improved Accessories Level 1 (dollar values in 2015$) 2-421
Table 2.154 Costs for Improved Accessories Level 2 (dollar values in 2015$) 2-422
Table 2.155 Costs for Secondary Axle Disconnect (dollar values in 2015$) 2-422
Table 2.156 Costs for Low Drag Brakes (dollar values in 2015$) 2-423
Table 2.157 Costs for A/C Controls (dollar values in 2015$) 2-423
Table 2.158 Off-Cycle "Menu" Technologies and Credits for Cars & Light Trucks 2-424
Table 2.159 Cost per gCCh/mi within the Indicated Ranges for the Perfect Trading Sensitivity Run Presented in the
Draft TAR (2013$) 2-425
Table 2.1602 Basis for Off-cycle Credit Values and Costs used in OMEGA 2-425
Table 2.161 Costs for Off-Cycle Technologies Level 1 & 2 (dollar values in 2015$) 2-425
Table 2.162 Costs for SCR-equipped Diesel Technology for Different Vehicle Classes (dollar values in 2015$).... 2-
425
Table 2.163 Costs for Advanced Diesel Technology for Different Vehicle Classes (dollar values in 2015$) 2-426
Table 2.164 Costs for Powersplit HEV Technology for Different Vehicle Classes (dollar values in 2015$) 2-427
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Chapter 2: Technology Costs, Effectiveness, and Lead Time Assessment
2.1 Overview
Technology assessment was a critical element of the development of the 2017-2025 GHG
standards in the 2012 final rulemaking (FRM). The standards were ultimately guided by a
detailed assessment of GHG-reducing technologies that were available as of the 2012 calendar
year time frame. The assessment included technologies that were currently in production at the
time, or pending near term release, as well as consideration of further developments in
technologies where there was reliable evidence that those technologies could be feasibly
deployed by 2025.
As the first step in the MTE process, the 2016 Draft TAR summarized the current state of
technology through the mid-2016 time frame, including technology developments since the FRM
and the outlook for future developments through MY2025. The Draft TAR found that the fleet
penetration of many of the GHG-reducing technologies identified in the FRM has proceeded
steadily, accompanied by new technologies not anticipated at the time. Technology assumptions
for cost, effectiveness, and availability were then revised and incorporated into the Draft TAR
GHG Assessment, a substantial and comprehensive update to the assessment performed for the
2012 FRM.
This Chapter 2 of the Proposed Determination Technical Support Document (TSD) provides
EPA's updated assessment of the current state of technology and likely future developments
through MY2025. A description of the technical work that has been done to inform the Draft
TAR and the Proposed Determination analysis is also included in this chapter, along with a
summary of the assumptions and inputs used to characterize technologies in the analysis. In the
cases where public comments received on the Draft TAR or updated information gathered since
the Draft TAR have contributed additional insight on the current state of technology or on
assumptions for technology cost and effectiveness, this information is incorporated into the
discussion. The results of EPA's Proposed Determination analysis are discussed in Section IV of
the Proposed Determination document.
In researching the Draft TAR, the agencies (EPA, NHTSA, and CARB) relied on many
sources to evaluate the state of technology, including vehicle certifications, vehicle simulation
modeling, reviews of technical papers and conference proceedings, agency meetings with vehicle
manufacturers and suppliers, and the 2015 NAS report. This collaborative effort produced an
extensive catalog of information on fuel-saving and GHG-reducing technologies that built upon
the 2012 FRM assessment. In developing the assessment for this Proposed Determination, EPA
has built further upon the body of information relied on for the Draft TAR assessment, by
continuing our in-house vehicle benchmarking testing program, enhancing and refining our
models, assessing the latest available data and literature, and considering public comments
received on the Draft TAR.
It is clear that the automotive industry is innovating and bringing new technology to market at
a brisk pace. Many of the technologies that figured prominently in the analysis performed for the
2012 FRM, such as gasoline direct injection, turbocharging and downsizing, and higher-
efficiency transmissions, have seen continued market penetration, and continued to have an
important role in the Draft TAR analysis. Even some well-established technologies had advanced
2-1
-------
Technology Cost, Effectiveness, and Lead Time Assessment
enough to require a re-evaluation of cost, effectiveness, and implementation for the Draft TAR.
For example, the ongoing improvements in transmissions with higher ratio spreads and gear
count, and the application of light-weight materials that had previously been applied only to
high-performance and luxury vehicles, were beginning to appear in mass-market vehicles. While
the cost, effectiveness,A and feasibility of implementation of individual technologies projected in
the Draft TAR were generally consistent with the compliance pathways projected in the 2012
FRM, some developments did not unfold as predicted. The Draft TAR found that several new
technology applications not considered in the FRM analysis, or which had been predicted to have
very low market penetration, had continued to evolve and deserved a reassessment. For example,
Atkinson Cycle engines have now been applied to non-hybrids successfully, and continuously
variable transmissions (CVTs) have entered the market more widely than originally expected in
applications that have been well-received by consumers and expert reviewers. Another example
is 48-volt mild hybridization, which by some accounts is gathering momentum rapidly, offering
significant efficiency benefits with lower complexity and system cost compared to the higher
voltage mild hybrid systems examined in the FRM analysis. The Draft TAR built upon the FRM
technology assessment by recognizing these technology developments and incorporating many
of them into the Draft TAR technology assessment.
Although some comments received on the Draft TAR were critical of EPA's assessment of the
effectiveness of some technologies, as a whole, EPA believes that the Draft TAR was broadly
accurate in its characterization of technology effectiveness. Through our consideration of public
comments on the Draft TAR, as well as continued analysis of sources such as current vehicle
certifications, continued benchmarking activities, literature reviews and modeling, it is our
assessment that the effectiveness values developed for the Draft TAR are largely fair and
accurate representations of benefits achievable by manufacturers within the time frame of the
rule. This is not to imply that every manufacturer that has added a technology has achieved the
effectiveness estimated in the Draft TAR. Some applications of technology are in their first or
second design iteration, and we expect that successive iterations will improve their effectiveness.
One example is the emerging use of integrated and cooled exhaust manifolds and the resulting
improved effectiveness from turbo-charged downsized engines. Some manufacturers that have
adopted technology have used some of the benefit to improve other vehicle attributes, rather than
solely to improve fuel economy. For example, the efficiencies gained can often be used to
promote other attributes such as acceleration performance, cargo capacity, towing capability,
and/or vehicle size and mass while holding fuel economy relatively constant. Vehicle
manufacturers have adopted many examples of technologies that perform very well, such as the
Mazda SKYACTIV-G® engine and the ZF 8-speed transmission, and when these technologies
are combined with the sole intent of improving vehicle efficiency, our analysis continues to show
that significant improvements from the baseline fleets are broadly achievable using conventional
powertrains.
This Chapter 2 provides a complete description of EPA's assessment of the status, cost,
effectiveness, and application of the technologies that we considered in this analysis. We have
included a brief review of the technology assessment conducted for the Draft TAR, as well as a
A The term 'effectiveness' is used throughout this Chapter to refer both to a reduction in tailpipe CO2 emissions and a
reduction in fuel consumption. In cases where the two are not equivalent (e.g., when changing fuel type), separate
values are presented.
2-2
-------
Technology Cost, Effectiveness, and Lead Time Assessment
summary of the updates that further inform the Proposed Determination assessment. Finally, we
discuss how we synthesized all of the available information to derive our conclusions for cost,
effectiveness, and application that informed the Proposed Determination technology assessment.
Like the technology assessment conducted for the Draft TAR, the Proposed Determination
technology assessment includes a wide array of fundamental assumptions, modeling constructs,
and general methodologies, as well as assumptions for cost and effectiveness of specific fuel-
saving and GHG-reducing technologies. Key changes and updates EPA has implemented for this
Proposed Determination assessment include:
• An updated baseline fleet, based on MY2015 GHG compliance data, the latest
complete data set available
• Updated projections of future fuel prices and vehicle sales to AEO 2016, the latest
available
• All monetized values are updated to 2015 dollars
• Better accounting for tire and aerodynamic improvements in the baseline fleet
• Updated accounting for light duty truck mass reduction in the baseline fleet
• Updated ZEV program sales using data from the California Air Resources Board
• Updated vehicle class definitions for modeling effectiveness to improve
representativeness of power-to-weight and road load characteristics
• Expanded vehicle classification structure from 19 to 29 vehicle types to improve the
resolution of cost-effectiveness estimates as applied in the OMEGA model
• Updated characterization and modeling of certain advanced engine technologies,
including Atkinson cycle
• Updated effectiveness estimates for certain advanced transmission technologies
• Updated battery costs for plug-in vehicles, resulting from several battery modeling
improvements such as an improved battery sizing method, updated data from
electrified vehicles released or certified since the Draft TAR, and an updated
accounting for energy consumption and road load technology improvements
• Added accounting in the compliance modeling for upstream emissions of plug-in
vehicles phasing in from MYs 2022 to 2025
• Incorporated additional off-cycle technology options into OMEGA to better account
for manufacturer's expected use of off-cycle credit opportunities
• Conducted additional sensitivity analyses to show the cost and technology penetration
impacts of alternative technology pathways
• Updated our vehicle simulation model, ALPHA, to include the latest data on
technology effectiveness from the EPA vehicle benchmarking testing program and
other sources, across vehicle types
• Added quality assurance checks of technology effectiveness estimates into ALPHA
and the lumped parameter model (LPM)
Complete descriptions of these changes, as well as discussion of public comments received on
the Draft TAR and updated information contributing to the Proposed Determination assessment,
can be found in the corresponding technology and methodology chapters of this TSD.
The remaining sections of this chapter provide detail on the state of development of specific
fuel-saving and GHG-reducing technologies, and their estimated cost and effectiveness.
2-3
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Section 2.2 of this chapter presents EPA's assessment of the current state of individual
technologies and the advancements that have occurred since the 2012 FRM and up to the
completion of the Draft TAR. EPA has reexamined every technology considered in the Draft
TAR, as well as assessed some technologies that are currently commercially available but did not
play a significant role in the Draft TAR analysis. We have also considered emerging
technologies for which enough information has become known that they may be included in this
Proposed Determination assessment. The categories of technologies discussed include engines,
transmissions, electrification, aerodynamics, tires, mass reduction, and several other vehicle
technologies. In addition, Chapter 2.2.9 provides an overview of the air conditioning efficiency
and leakage credit provisions, a summary of the situation regarding low global warming potential
(GWP) refrigerant, and discussion of key comments received on these topics. Chapter 2.2.10
provides a summary of the off-cycle credit program and an overview of how off-cycle credits
have been used by manufacturers in their current compliance with the GHG program. Chapter
2.3.4.9 (Additional Off-cycle Credits and Costs) details how off-cycle credits have been
considered in the Proposed Determination analysis. Key comments on the off-cycle credit
provisions are addressed in Section B.3.4 of the Proposed Determination Appendix.
Section 2.3 of this chapter presents details of the approaches, assumptions, and technology
inputs used in the Proposed Determination technology assessment.
The particular details of the assessment begin in Chapter 2.3.1 with a description of the
fundamental assumptions for performance neutrality, fuels, methods for measurement of cost and
effectiveness, and approach to vehicle classification, which together comprise the underpinnings
of the technical analysis.
Chapter 2.3.2 focuses on the approach for determining technology costs, which includes the
determination of both direct and indirect costs, as well as the application of cost reduction
through manufacturer learning, and maintenance and repair costs. The methodologies used to
develop technology costs remain largely unchanged from the Draft TAR. However, as was the
case in the Draft TAR, technology cost inputs have again been reevaluated based on updated
information and comments received on the Draft TAR.
Chapter 2.3.3 describes the approach for investigating technology effectiveness. Vehicle
benchmarking is one of the foundations of EPA's analysis of technology effectiveness. A
description of testing and benchmarking conducted by EPA can be found in Chapter 2.3.3.1.
Modeling of effectiveness across the vehicle fleet involves grouping vehicles into classifications,
and the approach to classifying vehicles for this purpose is described in Chapter 2.3.3.2. These
classifications and the data collected through benchmarking are used by EPA's full vehicle
simulation model, known as ALPHA. The ALPHA model is described in Chapter 2.3.3.3. An
outline of sources and methods for determining technology effectiveness is provided in Chapter
2.3.3.4. EPA's modeling methodology also includes use of a "lumped parameter model" (LPM),
which models incremental effectiveness differences between vehicle technology packages.
Updates to the LPM and its application in the Proposed Determination assessment are described
in Chapter 2.3.3.5.
Chapter 2.3.4 describes the specific data and assumptions for individual technologies that are
used in this Proposed Determination assessment. Informed by all of the information on the state
of technologies described in Section 2.2, these inputs and assumptions for cost, effectiveness,
and technology application ultimately led to the OMEGA model determination of the cost-
2-4
-------
Technology Cost, Effectiveness, and Lead Time Assessment
minimizing compliance pathways that are outlined in Section IV of the Proposed Determination
document and described in full detail in Section C of the Proposed Determination Appendix.
2.2 State of Technology and Advancements since the 2012 Final Rule
2.2.1 Individual Technologies and Key Developments
2.2.1.1 List of Technologies Considered
The key technologies considered in this Proposed Determination technology assessment are
summarized below. Assumptions for cost, effectiveness, or application of some of these
technologies have been updated for this Proposed Determination assessment, while others remain
unchanged from the Draft TAR where EPA has determined that changes are not warranted. Full
discussion of these technologies and any applicable updates is provided in the corresponding
technology sections of this chapter.
A number of technologies that were considered in the 2012 FRM analysis underwent
significant updates in the process of developing the Draft TAR assessment, which was a major
update of the FRM assessment representing more than four years of active technology evolution
and development throughout the automotive industry. Some of these most actively changing
technologies were significantly updated for the Draft TAR analysis, and in some cases further
updated for the Proposed Determination analysis. They include:
• HEV Atkinson cycle engines
• Non-HEV Atkinson cycle engines
• Turbocharging and downsizing
• Miller Cycle Engine
• Direct Injection Miller Cycle Engine
• Turbocharger improvements
• Cylinder deactivation
• Variable geometry valvetrain systems (VVT, DVVL, CVVL)
• Continuously variable transmissions (CVTs)
• Dual clutch transmissions (DCTs)
• 48-volt mild hybrid electric vehicles (MHEVs)
Other technologies that were included in the FRM and the Draft TAR analysis, some of which
also received updates to how they were represented in the Proposed Determination analysis,
include:
• Stoichiometric gasoline direct injection
• Exhaust gas recirculation with boost
• Low-friction lubricants
• Second level of low-friction lubricants and engine friction reduction
• Reduction of engine friction losses
• Diesel engines
• Improved automatic transmission controls
• Increased gear-count automatic transmissions
2-5
-------
Technology Cost, Effectiveness, and Lead Time Assessment
• Shift optimization
• Manual 6-speed transmission
• High efficiency gearbox (automatic, DCT, CVT, or manual)
• Low-rolling-resistance tires
• Aerodynamic drag reduction
• Mass reduction
• Low-drag and zero drag brakes
• Secondary axle disconnect for four-wheel drive systems
• Electric power steering (EPS)
• Improved accessories (IACC)
• Low-leakage and higher-efficiency air conditioner systems
• Non-hybrid 12-volt stop-start
• High-voltage mild and strong hybrids (HEVs), including strong P2 and power split
• Plug-in hybrid electric vehicles (PHEVs)
• Battery electric vehicles (BEVs)
Each of these technologies are described in more detail in the following section. Full detail of
the current development state of each technology can be found in the remaining sections of this
chapter.
2.2.1.2 Descriptions of Technologies and Key Developments since the FRM
As described in the previous section, a number of technologies considered in the 2012 FRM
analysis underwent significant updates in the process of developing the Draft TAR assessment.
Some technologies that had not been considered in the 2012 FRM were added for the Draft TAR
analysis, while others that had been included had developed differently than expected, and were
updated accordingly.
This section provides capsule descriptions of the fuel-saving and GHG-reducing technologies
considered in the Proposed Determination assessment, beginning with this subset of actively
changing technologies that largely distinguished the Draft TAR assessment from the 2012 FRM
assessment. It highlights some of the key considerations and updates that affected how each of
these technologies were considered for the Draft TAR and, in many cases, further consideration
and updates that were implemented for the Proposed Determination assessment. Other
technologies that were considered in both the 2012 FRM and Draft TAR assessments, and which
continue to be considered in the Proposed Determination assessment, are also outlined in this
section.
This section is meant to provide only a brief outline of the technologies that EPA considered.
For complete descriptions of the state of development of each technology, please refer to
Chapters 2.2.2 through 2.2.10. Specific assumptions for cost and effectiveness for each
technology as applied to the Proposed Determination assessment are discussed in Chapter 2.3.4.
2-6
-------
Technology Cost, Effectiveness, and Lead Time Assessment
HEVAtkinson cycle engines. These engines have a substantial increase in geometric
compression ratio6 (in the range of 12.5 - 14:1) and intake valve event timing to provide much
later intake valve closing (LIVC). This lowers the trapped air charge, effectively lowering actual
compression ratio to reduce knock-limited operation while maintaining the expansion ratio for
improved efficiency. Although producing lower torque at low engine speeds for a given
displacement, this engine has specific high efficiency operating points and is capable of
significant CO2 reductions when properly matched to a strong hybrid system. Electric
motor/generators produce high torque at low speeds and are thus are capable of offsetting low
engine speed torque deficiencies with Atkinson Cycle engines.
Non-HEVAtkinson cycle engines. For non-HEV applications, this technology often combines
direct injection, a substantial increase in geometric compression ratio (in the range of 13-14:1),
wide authority variable intake camshaft timing, variable exhaust camshaft timing, and an
optimized combustion process to enable significant reductions in CO2 compared to a standard
direct injected engine. This engine is capable of changing the effective compression ratio by
varying intake valve events enabling Otto and Atkinson operation. This multiple mode
capability enables these engines to be applied in hybrid and non-hybrid applications. The ability
to reduce pumping losses over a large area of operation may allow avoidance of the additional
cost of higher gear count transmissions. The Mazda SKYACTIV-G engine is one example of
this technology. The 2GR-FKS engine used in the MY2015-2017 Toyota Tacoma pickup truck
is another example. The 2.0L "Nu" engine in the MY2017 Hyundai Elantra is another example
of use of Atkinson Cycle in non-HEV application, although the "Nu" Atkinson engine uses PFI
instead of GDI and has a slightly lower geometric CR than used by Mazda. The Toyota 1NR-
FKE and 2NR-FKE Atkinson Cycle engines use both PFI and cEGR instead of GDI. In the
FRM, the use of Atkinson Cycle engines was primarily considered in HEV applications. In the
past few years, a new generation of naturally-aspirated SI Atkinson Cycle engines applicable to
non-HEVs has been introduced into light-duty vehicle applications. The most prominent
application of this technology is the Mazda SKYACTIV-G® system. It combines direct injection,
an ability to operate over an Atkinson Cycle with increased expansion ratio, wide-authority
intake camshaft timing, and an optimized combustion process. Other OEMs have intruded non-
HEV Atkinson Cycle engines using PFI instead of GDI, in some cases combined with cooled,
external EGR (cEGR). This type of engine operation is also not limited to naturally aspirated
engines and when applied to boosted engines is referred to as "Miller Cycle," as described below.
In addition to Mazda, other manufacturers using non-HEV application of Atkinson Cycle
engines include Hyundai, Toyota, and FCA.
Turbocharging and downsizing. This approach 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 the FRM, turbocharged,
B Geometric compression ratio is a ratio of the piston clearance volume + displacement swept volume to the
displacement swept volume in a reciprocating piston engine. The actual effective compression ratio and
expansion ratio must also take into account valve events governing the actual flows involved in the combustion
process. Effective compression ratio and expansion ratios for typical Otto-cycle engines are nearly equivalent
and governed by the chosen geometric compression ratio. Atkinson and Miller Cycle engines lower the trapped
air or air-fuel charge volume during intake via either late intake valve closing or early intake valve closing to
reduce effective compression ratio while simultaneously increasing effective expansion ratio. This is done by
reducing the piston clearance volume and thus increasing the geometric compression ratio.
2-7
-------
Technology Cost, Effectiveness, and Lead Time Assessment
downsized engines were anticipated to be a prominent technology applied by vehicle
manufacturers to improve vehicle powertrain efficiency. The penetration rate of turbo-downsized
engines into the light-duty fleet has increased from 3 percent in 2008 to 16 percent in 2014.1 The
Draft TAR recognized that turbocharged, downsized engines are adopting head-integrated
exhaust manifolds or separate, water-cooled exhaust manifolds. These systems also use separate
coolant loops for the head/manifold and for the engine block. The changes allow faster warmup,
improved temperature control of critical engine components, further engine downspeeding, and
reduce the necessity for commanded enrichment for component protection. The net result is
improved efficiency over the regulatory cycles and during real world driving. Engine
downspeeding also has synergies with recently developed, high-gear-ratio spread transmissions
that may result in further drive cycle efficiency improvements. In this Proposed Determination,
consistent with the Draft TAR, EPA considered two levels of boosting, 18 bar brake mean
effective pressure (BMEP) and 24 bar, 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 and 24 bar BMEP is applied with 50 percent. 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.
Miller Cycle Engine. This technology combines direct injection, a significant increase in
geometric compression ratio relative to other boosted engines, wide authority intake camshaft
timing, and variable exhaust camshaft timing, and an optimized combustion process to enable
significant reductions in CO2 as compared to a standard direct injected engine. This is
essentially Atkinson Cycle with the addition of a turbocharger boosting system. The addition of
a turbocharger improves volumetric efficiency and broadens the areas of high-efficiency
operation. The ability to reduce pumping losses over a large area of operation may allow
avoidance of the additional cost of higher gear count transmissions. Examples include the
Mazda SKYACTIV-G Turbo engine used in the MY2017 CX9; the VW EA211 evo 1.5L 14,
EA888 3B 2.0L 14, and EA839 3.0L V6; the Toyota 8NR-FTS 1.2L 14 and 8AR-FTS 2.0L 14;
the PSA 1.2L 13 PSA EB Puretech, and the Honda L15B7 1.5L 14.
Direct Injection Miller Cycle Engine. This new generation of turbocharged GDI engine
combines direct injection, the ability to operate over a Miller Cycle (boosted Atkinson Cycle)
with increased expansion ratio, wide-authority intake camshaft timing, and an optimized
combustion process. Current manufacturers include VW, Mazda, Toyota, and PSA.
Turbocharger improvements. Newer turbochargers have been developed that reduce both
turbine and compressor inertia allowing faster turbocharger spool-up. Improvements have been
made to broaden the range of compressor operation before encountering surge and to improve
compressor efficiency at high pressure ratios. The introduction of head-integrated exhaust
manifolds or separate, water-cooled exhaust manifolds reduces exhaust turbine inlet
temperatures under high-load conditions and improves exhaust temperature control. This allows
the use of less expensive, lower temperature materials for the turbine housing and exhaust
turbine. Reduced turbine inlet temperatures also allow the introduction of turbochargers with
variable nozzle turbines into SI engine applications, similar to those used in light-duty diesel
applications. Twin-scroll turbochargers are finding broad application in turbocharged, downsized
GDI engines. Twin-scroll turbochargers improve turbocharger spool-up and improve torque
output at lower engine speeds, allowing further engine downspeeding. Turbochargers with
variable nozzle turbines (VNT) are now common in light-duty diesel applications and are under
2-8
-------
Technology Cost, Effectiveness, and Lead Time Assessment
development for gasoline spark ignition engines, particularly those that use cooled EGR and
head-integrated exhaust manifolds.
Cylinder deactivation. This technology 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 displacement engine with fewer cylinders which substantially reduces
pumping losses. Cylinder deactivation applied to engines with less than six cylinders was not
analyzed as part of the FRM. Further developments in NVH (noise, vibration, and harshness)
abatement, including the use of dual-mass dampening systems, have resulted in the recent
introduction of a 4-cylinder/2-cylinder engine into the European light-duty vehicle market. The
development of rolling or dynamic cylinder deactivation systems allows a further degree of
cylinder deactivation for odd-cylinder (e.g., 3-cylinder, 5-cylinder) inline engines than was
possible with previous cylinder deactivation system designs. Both 3-cylinder/2-cylinder and 3-
cylinder/1.5-cylinder (rolling deactivation) designs are at advanced stages of development.
Variable geometry valvetrain systems. This technology includes systems that vary valve
timing and/or valve lift. 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. Discrete variable valve lift increases efficiency by optimizing air flow over a
broader range of engine operation which reduces pumping losses, and is accomplished by
controlled switching between two or more cam profiles. 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. Variable geometry systems were anticipated
in the FRM and Draft TAR to be important technologies for reducing engine pumping losses.
Continuously variable transmissions (CVTs). This transmission uses a belt or chain between
two variable ratio pulleys, allowing a continuous (infinite) range of gear ratios and enabling the
engine to operate in a more efficient operating range over a broad range of vehicle operating
conditions. EPA did not assign a significant role to CVTs in the FRM analysis in part because of
indications that some manufacturers had experienced consumer acceptance problems with CVTs,
largely due to differences in shift feel compared to a conventional automatic transmission. Since
the FRM, a new generation of CVTs has been introduced into the light-duty market by several
OEMs. These new CVTs have significant improvements in shift feel as well as efficiency, and
have achieved a wider ratio spread. CVTs have become increasingly common in manufacturers'
product lines today.
Dual clutch transmissions (DCTs). This transmission is 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. Early DCTs, mostly in
non-performance vehicles, were accepted in Europe but were not widely accepted in the North
American market, in part because launch and shift characteristics differed from conventional
automatic transmissions. However, strategies have been developed to improve overall DCT
operational characteristics. DCTs occur in variations called wet clutch, dry clutch, and "damp
clutch." The damp clutch DCT combines the durability and driveability of a wet clutch with the
efficiency of a dry clutch DCT. The combination of a DCT with a torque converter can greatly
improve operational characteristics and eliminates the need for complex crankshaft dampers and
2-9
-------
Technology Cost, Effectiveness, and Lead Time Assessment
other NVH technologies. The elimination of these NVH technologies approximately offsets the
additional cost of the torque converter. DCTs also can be integrated into P2-architecture HEVs
as well as 48-volt P2 hybrid drive systems, providing advantages such as improved launch assist,
low-speed creep capability, and driving characteristics similar to a torque-converter/planetary
gear-set automatic transmission.
48-volt mild hybrids. Mild hybrids provide idle-stop capability and launch assistance and use
a higher voltage battery with increased energy capacity over typical automotive batteries. The
higher system voltage allows the use of a smaller, more powerful electric motor than possible
with a 12-volt system, 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 belt-driven starter-alternator which can recover braking energy while the
vehicle slows down (regenerative braking). At the time of the FRM, high-voltage (e.g. 120-volt)
mild hybrids were known in the market (for example, the Chevrolet Malibu eAssist system), and
were anticipated to grow in market share. In the time since the FRM, both mild and strong hybrid
sales have not grown as quickly as expected, an outcome that is often attributed to lower fuel
prices. Another factor may be the rate of improvements in the efficiency of conventional
vehicles, which appear to be closing the fuel economy gap. However, a new generation of mild
hybrid technologies is being introduced into the light-duty market, using a 48-volt electrical
system, which can reduce costs by eliminating high-voltage safety requirements and battery
cooling hardware (in many cases), while offering an effectiveness similar to that of higher-
voltage mild hybrids, potentially resulting in significantly greater cost effectiveness. The Draft
TAR recognized this trend and added consideration of 48-volt mild hybridization technology.
The following paragraphs outline other technologies that were included in the 2012 FRM and
Draft TAR analyses and continue to be included in the Proposed Determination analysis. In
many cases the cost, effectiveness, or specific applications of these technologies have also been
updated for this analysis. For complete descriptions of the state of development of each
technology, please refer to Chapters 2.2.2 through 2.2.10. Specific assumptions for cost and
effectiveness for each technology as applied to the Proposed Determination assessment are
discussed in Chapter 2.3.4.
Stoichiometric gasoline direct-injection technology. This 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. In the FRM as in the Draft TAR and the current analysis, this technology is projected
to be very widespread by 2025.
Exhaust-gas recirculation with boost. Increases the exhaust-gas recirculation used in the
combustion process to improve knock-limited operation and reduce pumping losses. Peak levels
of exhaust gas recirculation approach 25 percent by volume in these highly boosted engines (this,
in turn raises the boost requirement by approximately 25 percent). EPA applies this technology
only to 24 bar BMEP and Miller cycle engines.
Low-friction lubricants. Low viscosity and advanced low friction lubricants oils are now
available with improved performance and better lubrication.
Second level of low-friction lubricants and engine friction reduction. As technologies
continue to advance between now and 2025, we expect further developments enabling lower
2-10
-------
Technology Cost, Effectiveness, and Lead Time Assessment
viscosity and lower friction lubricants and more engine friction reduction technologies available,
including the use of roller bearings for balance shaft systems and further improvements to
surface treatment coatings. As of MY2017, many of the friction reduction technologies classified
as "second level" are already being introduced into light-duty vehicles.
Reduction of engine friction losses. This can be achieved through low-tension piston rings,
roller cam followers, improved material coatings, more optimal thermal management, piston
surface treatments, cylinder wall treatments and other improvements in the design of engine
components and subsystems that improve engine operation.
Diesel engines. Despite recent controversy concerning emission control, 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 higher
compression and expansion ratios, with a very lean air/fuel mixture, than an equivalent-
performance gasoline engine. This technology requires additional enablers, such as use of NOx
adsorption exhaust catalyst (NAC), selective catalytic reduction (SCR) of NOx, or a combination
of both NAC and SCR NOx catalytic after-treatment and use of a catalyzed diesel particulate
filter (CDPF) for PM emissions control.
Improved automatic transmission controls. This technology optimizes the shift schedule to
maximize fuel efficiency under wide ranging conditions, and minimizes losses associated with
torque converter slip through lock-up or modulation.
Six, seven, and eight-speed (or more) automatic transmissions. Also described here as
increased gear-count 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. In the FRM, EPA limited its consideration of the effect of additional gears
to eight-speed transmissions. However, some ATs with more than eight gears are already in
production, and more examples are in development. At this time, nine-speed transmissions are
being manufactured by ZF (which produces a FWD nine-speed incorporated into Fiat/Chrysler,
Honda, and Jaguar/Land Rover vehicles) and Mercedes (which produces a RWD nine-speed).
Ford has released a ten speed transmission in the F150 Raptor, and GM released a variation of
the same ten speed in the 2017 Camaro ZL1. In addition, Ford and General Motors have
announced plans to jointly design and build a nine-speed FWD transmission, and Honda is
developing a ten-speed FWD transmission.
Shift optimization. This technology targets engine operation at the most efficient point for a
given power demand. The shift controller emulates a traditional continuously variable
transmission by selecting the best gear ratio for fuel economy at a given required vehicle power
level to take full advantage of high BMEP engines. The shift controller also incorporates
boundary conditions to prevent undesirable operation such as shift busyness and NVH issues.
Manual 6-speed transmission. This technology offers an additional gear ratio, often with a
higher overdrive gear ratio, than a 5-speed manual transmission.
High efficiency gearbox (automatic, DCT, CVT, or manual). This technology represents
continuous improvement in seals, bearings and clutches, super-finishing of gearbox parts, and
development in the area of lubrication, all aimed at reducing friction and other parasitic loads in
the system for an automatic, DCT or manual type transmission.
2-11
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Low-rolling-resistance tires. This technology includes tires that have characteristics that
reduce frictional losses associated with the energy dissipated in the deformation of the tires under
load, thereby reducing the energy needed to move the vehicle. EPA's analyses have
characterized two levels of rolling resistance reduction (LRRT1 and LRRT2), targeting a 10
percent and 20 percent rolling resistance reduction from baseline tires, respectively.
Aerodynamic drag reduction. This technology refers to approaches to reducing aerodynamic
drag, which can be achieved by various means such as changing vehicle shapes, reducing frontal
area, sealing gaps in body panels, and adding additional components including side trim, air
dams, underbody covers, and aerodynamic side view mirrors. EPA's analyses have considered
two levels of aerodynamic drag reduction (AEROl and AER02), targeting a 10 percent and 20
percent aerodynamic drag reduction, respectively.
Mass reduction. This technology encompasses a variety of techniques ranging from improved
design and better component integration to application of lighter and higher-strength materials.
In addition to reduced road load, mass reduction can lead to collateral GHG benefits by enabling
a downsized engine and/or downsized ancillary systems (transmission, steering, brakes,
suspension, etc.) that directly result from the reduced vehicle weight.
Low-drag and zero drag brakes. This technology reduces the sliding friction of disc brake
pads on rotors when the brakes are not engaged by pulling the brake pads away from the rotors.
Secondary axle disconnect for four-wheel drive systems. This technology applicable to all-
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.
Electric power steering (EPS). This represents 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). This represents accessories with improved efficiency. EPA's
analyses have considered two levels of IACC. The first level may include high efficiency
alternators, electrically driven (i.e., on-demand) water pumps and cooling systems. This
excludes other electrical accessories such as electric oil pumps and electrically driven air
conditioner compressors. The second level of IACC includes alternator regenerative braking on
top of what are included in the first level of IACC.
Low-leakage and higher-efficiency air conditioner systems. These technologies are focused on
reducing leakage of high-GWP refrigerants and improved energy efficiency. Leakage measures
include improved hoses, connectors and seals for leakage control. Efficiency measures include
improved compressors, expansion valves, heat exchangers and the control of these components
for the purposes of improving tailpipe CO2 emissions and fuel economy when the AJC is
operating.
Non-hybrid stop-start. Also known as idle-stop or 12V micro hybrid, this is the most basic
system that facilitates idle-stop capability. This system includes an enhanced performance
starter and battery but no additional hybridization features. While stop-start has been in
production for a considerable amount of time in Europe (a predominantly manual transmission
market), some of the initial product offerings in the U.S. met with consumer feedback concerns.
2-12
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Since the FRM, some recent vehicles were introduced with stop-start implementations that were
specifically designed for the U.S. market, such as the Chevrolet Malibu, and have been met with
very good reviews. Indications from suppliers are that further improvements, including the use of
continuously engaged starters, are under development.
Strong hybrids (P2 hybrid). Strong hybrids include what are known as P2 hybrids and power-
split hybrids, among other types. EPA models strong hybrids as P2 hybrids. The P2 hybrid is a
technology that uses a transmission-integrated electric motor placed between the engine and a
gearbox or CVT, with a wet or dry separation clutch which is used to decouple the
motor/transmission from the engine. A P2 hybrid would typically be equipped with a larger
electric machine than a mild hybrid system, but smaller than a power-split hybrid architecture.
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 based on
simulation, when combined with a DCT transmission, provides similar efficiency to other strong
hybrid systems.
Power-split Hybrid (PSHEVs). While EPA models primarily P2 hybrids in this analysis,
power-split hybrids are represented in the baseline fleet. Power split is a hybrid electric drive
system that replaces the traditional transmission with a single planetary gearset and two
motor/generators. One motor/generator uses the engine to either charge the battery or supply
additional power to the drive motor. The second, usually more powerful, motor/generator is
permanently connected to the vehicle's final drive and always turns with the wheels, as well as
providing regenerative braking capability. The planetary gear-set splits engine power between
the first motor/generator and the output shaft to either charge the battery or supply power to the
wheels. The Power-split hybrid provides similar efficiency to other strong hybrid systems.
Plug-in hybrid electric vehicles (PHEVs). 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 than non-plug-in hybrid electric vehicles 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, allowing for reduced fuel use during "charge depleting" operation. The FRM, Draft
TAR and this Proposed Determination analysis models PHEVs with 20-mile and 40-mile ranges.
Battery electric vehicles (BEVs). Vehicles with all-electric drive and with vehicle systems
powered by energy-optimized batteries charged from an outside source of electricity (usually the
electric grid). In the FRM, BEVs were modeled with driving ranges of 75 miles, 100 miles, and
150 miles. The Draft TAR revised the 150-mile BEV to a 200-mile BEV, which is retained for
this analysis.
In summary, this Chapter 2.2.1 has provided only a brief outline of the fuel-saving and GHG-
reducing technologies considered in the Proposed Determination analysis. For complete
descriptions of the state of development of each technology, please refer to Chapters 2.2.2
through 2.2.10. Specific assumptions for cost and effectiveness for each technology are
discussed in Chapter 2.3.4.
2.2.2 Engines: State of Technology
2-13
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Internal combustion engine improvements continue to be a major focus in improving the
overall efficiency of light-duty vehicles. While the primary type of light-duty vehicle engine in
the United States is a gasoline fueled, spark ignition (SI), port-fuel-injection (PFI) design, it is
undergoing a significant evolution as manufacturers work to improve engine brake thermal
efficiency (BTE) from what has historically been approximately 25 percent to BTE of 37 percent
and above. This focus on improving gasoline SI engines has resulted in the adoption of
technologies such as gasoline direct injection (GDI), turbo-charging and downsizing, Atkinson
Cycle, Miller Cycle, increased valve control authority through variable valve timing and variable
valve lift, integrated exhaust manifolds, reduced friction, and cooled EGR (cEGR). Vehicle
manufacturers have more choices of technology for internal combustion engines than at any
previous time in automotive history and more control over engine operation and combustion. In
addition, manufacturers have access to improved design tools that allow them to investigate and
simulate a wide range of technology combinations to allow them to make the best decisions
regarding the application of technology into individual vehicles. Despite the access to improved
tools and simulation, EPA believes that manufacturers have not yet explored the entire design
space of modern powertrain architectures and that innovation will continue resulting in
improvements in efficiency that are beyond what is currently being demonstrated in the new car
fleet.
As discussed in Chapter 3, the use of many of the major powertrain technologies analyzed in
the 2012 FRM, including engine technologies such as VVT, direct injection, turbocharging, and
cylinder deactivation have increased since the publication of the FRM and appear to be trending
towards EPA projections of technology penetration levels from the 2017-2025 FRM analysis
(see Chapter 3). Engines equipped with GDI are projected to achieve a 46 percent market share
in MY2015. Approximately 18 percent of new vehicles are projected to be equipped with
turbochargers for MY2015. Use of cylinder deactivation has grown to capture a projected 13
percent of light-duty vehicle production for MY2015. Light duty diesel vehicles are projected to
increase to a projected 1.5 percent of new vehicle production for MY2015, which is the highest
level since MY1984. Recently introduced light-duty diesels in the U.S. include several new
pickup truck (2015 Ram 1500, 2016 Chevrolet Colorado, 2016 GMC Canyon) and SUV (2015
Jeep Grand Cherokee, 2016 Land Rover Range Rover, Mercedes GLE300 and GLE350) models.
Mazda has transitioned all of their products to either Atkinson Cycle or Miller Cycle engines.
Volkswagen's entire gasoline vehicle product range uses downsized/turbocharged/GDI engines
and most of these engine families are now transitioning to Miller Cycle.
2-14
-------
Technology Cost, Effectiveness, and Lead Time Assessment
100%
90%
80% | ¦ MY 201S
70%
60%
"S
£ 50%
H 40%
V
u
£ 30%
20%
10%
o%
MY 200S
J
WT Multi- GDI Turbo Cylinder Diesel
valve Deact.
Figure 2.1 Light-duty Vehicle Engine Technology Penetration since the 2012 Final Rule
2.2.2.1 Overview of Engine Technologies
Since the FRM, to prepare for the Draft TAR the agencies met with automobile
manufacturers, major Tier 1 automotive suppliers and major automotive engineering services
firms to review both public and confidential data on the development of advanced internal
combustion engines for MY2022 and later. A considerable amount of new work was completed
both within the agencies and within industry and academia that was therefore available for
consideration in the Draft TAR. EPA completed several engine benchmarking programs that
have produced detailed engine maps. These engine maps represent some of the best performing
engines available today and have been used in the ALPHA model to directly estimate the
effectiveness of modern powertrain technology being applied to a wide spectrum of vehicle
applications. In addition, industry and academia regularly publishes similar levels of detail with
regard to engine operation in the public domain, and EPA has also used this information to either
directly inform or to compare effectiveness estimations.
In addition to creating detailed engine maps for full vehicle simulation, EPA conducted proof-
of-concept, applied research to investigate the potential for further engine improvements. This
includes the use of both computer-aided engineering tools and the development and analysis of
advanced engine technologies via engine dynamometer testing. Further details are provided in
Chapter 2.3.
In the time since the FRM, in meetings with automobile manufacturers and Tier 1 suppliers,
we learned about convergent and divergent trends in engine technologies. Through this ongoing
analysis and OMEGA modeling, it continues to be our assessment that through MY2022, with
few exceptions, gasoline direct injection and VVT will be applied to most engines. Significant
attention will be placed on reducing engine friction and accessory parasitic loads. In passenger
2-15
-------
Technology Cost, Effectiveness, and Lead Time Assessment
car and smaller light-duty truck segments, there will be considerable diversity of engine
technologies, including turbocharged GDI engines with up to 25-bar BMEP, both turbocharged
and naturally aspirated GDI engines with external cooled EGR, engines that combine GDI with
operation over the Atkinson Cycle, use of Atkinson Cycle in non-HEV applications, and use of
Miller Cycle (boosted Atkinson Cycle). With respect to larger, heavier vehicles, including full-
size SUVs and pickup trucks with significant towing utility, some manufacturers will be relying
on naturally aspirated GDI engines with cylinder deactivation, some will be relying more on
turbocharged-downsized engines, and others will be using a variety of engine technologies,
including light-duty diesels. Vehicle manufacturers are at advanced stages of research with
respect to:
• Stratified-charge, lean-burn combustion
• Multi-mode combustion approaches
° homogenous charge, compression ignition, lean-burn operation at light loads
° stratified-charge, lean-burn spark ignition at moderate loads
° stoichiometric homogenous charge, spark ignition at high loads
• Variable-compression ratio (VCR) engines
• Engines exceeding 24-bar BMEP
While the introduction of variable compression ratio engines and highly boosted GDI engines
above 24-bar BMEP is expected within the 2022-2025 time frame, these technologies will most
likely be introduced into relatively low-volume, high performance applications. Manufacturers
and suppliers are finding that turbocharged engines can achieve lower CO2 emissions over the
regulatory drive cycles and improved real-world fuel economy at more moderate (24 bar and
below) BMEP levels. While there are both performance and efficiency advantages to VCR at
high BMEP levels, both Atkinson Cycle and Miller Cycle with VVT are technologies that
compete with VCR and that have a comparable ability to vary effective compression ratio but
with reduced cost and complexity.
We also learned from manufacturers and suppliers that specific engine technologies have
synergies with other CCh-reduction technologies. For example, measures to reduce engine
friction, particularly friction at startup, help reduce the motor torque necessary for restart in 12V
start/stop systems. GDI and electric cam phasing systems can be used for combustion assistance
of engine restart. There are also synergies between Miller Cycle, IEM, cooled-EGR, and the use
of VNT turbochargers which are described in more detail in Chapter 2.2.2.7.
Despite recent EPA and California ARB compliance actions with respect to light-duty diesel
NOx emissions, diesel engines remain a technology for the reduction of GHG emissions from
light-duty vehicles. Advances in NOx and PM emissions control technology are bringing light-
duty diesels fully into compliance with Federal Tier 3 and California LEV III emissions
standards at a cost that is competitive with the cost-effectiveness other high efficiency, advanced
engine technologies. In the FRM, diesel powertrains were not expected to be a significant
technology for improving vehicle efficiency, however, since then many new light-duty vehicles
have been introduced to the U.S. market with diesel engines, including the Ram 1500 full-size
pickup truck, the Chevrolet Colorado mid-size pickup truck, the Jeep Grand Cherokee SUV, and
the Chevrolet Cruze. In addition, diesel engines are continuing to evolve using technologies
similar to those being introduced in new light-duty gasoline engines and heavy-duty diesel truck
2-16
-------
Technology Cost, Effectiveness, and Lead Time Assessment
engines, including the use of advanced friction reduction measures, increased turbocharger
boosting and engine downsizing, use of VNT and/or sequential turbocharging, engine
"downspeeding," the use of advanced cooled EGR systems, improved integration of charge air
cooling into the air intake system, and improved integration of exhaust emissions control systems
for criteria pollutant control. The best BTE of advanced diesel engines under development for
light duty applications is now 46 percent and thus is approaching that of heavy-duty diesel truck
engines.2
In addition to a reevaluation of all of the cost and effectiveness values of the technologies that
were considered in the FRM, this TSD (as did the Draft TAR) includes evaluations of
technologies where substantial new information has emerged since the FRM, including Atkinson
and Miller cycle engines, and application of cylinder deactivation operation to 3-cylinder, 4-
cylinder, and turbocharged engines.
2.2.2.2 Sources of Engine Effectiveness Data
In addition to the sources of engine CO2 effectiveness data used in the 2017-2025 LD GHG
FRM, EPA also used engine data from a wide range of sources to update engine effectiveness for
the draft TAR and Proposed Determination, including:
• Publicly available data (e.g., peer-reviewed journals, peer-reviewed technical papers,
conference proceedings)
• Data directly acquired by EPA via engine dynamometer testing at EPA-NVFEL or at
contract laboratories
• Benchmarking and simulation modeling of current and future engine configurations
• Confidential data from OEMs, Tier 1 suppliers, and major automotive engineering
services firms
• Data from the U.S. Department of Energy Vehicle Technologies Program
A considerable amount of brake-specific fuel consumption (BSFC), brake-thermal efficiency
(BTE) and chassis-dynamometer drive cycle fuel consumption data for advanced powertrains has
been published in journals, technical papers and conference proceedings since the publication of
the 2012 FRM. In some cases, published data includes detailed engine maps of BSFC and/or
BTE over a wide area of engine operation. In addition, these publications provide a great deal of
information regarding the specific design changes made to an engine which allow the engine to
operate at an improved BSFC and vehicles to operate with improved fuel consumption. These
design details often include changes to engine friction, changes to valvetrain and valve control,
combustion chamber design and combustion control, boosting components and boosting control,
and exhaust system modifications. This information provides the agency an indication of which
technologies to investigate in more detail and offers the opportunity to correlate testing and
simulation results against currently available and future designs.
Since 2012, many examples of advanced engine technologies have gone into production for
the U.S., European and Japanese markets. EPA has acquired many vehicles for chassis
dynamometer testing and has developed a methodology for conducting detailed engine
dynamometer testing of engines and engine/transmission combinations. Engine dynamometer
testing was conducted both at the EPA-NVFEL facility in Ann Arbor, MI and at other test
facilities under contract with EPA. Engine dynamometer testing of production engines outside
of the vehicle chassis required the use of a vehicle-to-engine (or vehicle-to-engine/transmission)
2-17
-------
Technology Cost, Effectiveness, and Lead Time Assessment
wiring tether and simulated vehicle feedback signals in order to allow use of the vehicle
manufacturer's engine management system and calibrated control parameters. In addition to fuel
consumption and regulated emissions, many of the engines were also instrumented with piezo-
electric cylinder pressure transducers and crankshaft position sensors to allow calculation of the
apparent rate of heat release and combustion phasing. Engines with camshaft-phasing were also
equipped with camshaft position sensors to allow monitoring of the timing of valve events.
Engine dynamometer testing also incorporated hardware-in-the-loop HIL simulation of drive
cycles so that vehicle packages with varying transmission configurations and road-loads could be
evaluated. Specific examples of engine benchmarking and HIL simulation used by EPA were
published within peer reviewed literature prior to release of the Draft TAR.3
While the confidential data provided by vehicle manufacturers, suppliers and engineering
firms cannot be published in the Draft TAR, these sources of data were important as they
allowed EPA to perform quality and rationality checks against the data that we are making
publicly available. In each case where a specific technology was benchmarked, EPA met with
the vehicle manufacturer to confirm the results. In cases where expected combinations of future
engine technologies were not available for testing from current production vehicles, a
combination of proof-of-concept engine dynamometer testing and engine and vehicle CAE
simulations were used to determine drive cycle effectiveness. For example, use of cooled EGR
and an increased geometric compression ratio was modeled using Gamma Technologies GT-
Power simulations of combustion and gas dynamics with subsequent engine dynamometer
validation conducted using a prototype engine management system, a developmental external
low-pressure cooled EGR system, and a developmental dual-coil offset ignition system. Finally,
several of these benchmarking activities were the subject of technical papers published by SAE
and included a peer review of the results as part of the publication process.
2.2.2.3 Low Friction Lubricants (LUB)
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 0W-20, to improve cold-flow properties and reduce cold start
friction. However, in some cases, changes to the crankshaft, rod and main bearings and changes
to the mechanical tolerances of engine components may be required. In all cases, durability
testing is required to ensure that durability is not compromised. The shift to lower viscosity and
lower friction lubricants also improve the effectiveness of valvetrain technologies such as
cylinder deactivation, which rely on a minimum oil temperature (viscosity) for operation.
2.2.2.4 Engine Friction Reduction (EFR1, EFR2)
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. Examples include improvements in low-tension piston rings,
piston skirt design, roller cam followers, improved crankshaft design and bearings, material
2-18
-------
Technology Cost, Effectiveness, and Lead Time Assessment
coatings, material substitution, more optimal thermal management, and piston and cylinder
surface treatments. Additionally, as computer-aided modeling software 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.
2.2.2.5 Cylinder Deactivation (DEAC)
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
cylinders when the load is significantly less than 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 higher loads to compensate for the deactivated cylinders.
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 and engine changes that enable increasing the amount of time that
cylinder deactivation might be suitable. Some manufacturers have adopted active engine
mounts, active noise cancellations systems, and crankshaft dampening systems to address NVH
concerns and to allow a greater operating range of activation.
2.2.2.6 Variable Valve Timing (WT) Systems
Variable valve timing (VVT) is 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 MY2015, more than 98 percent of
light-duty vehicles sold in the U.S. are projected to use some form of VVT.195 The three major
types of VVT are listed in the sub-sections 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. Electric cam phasing allows a wider range of camshaft phasing, faster
time-to-position, and allows adjustment of camshaft phasing under conditions that can be
challenging for hydraulic systems, for example, during and immediately after engine startup.
2-19
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.2.2.6.1 Intake Cam Phasing (TCP)
Valvetrains with ICP can modify the timing of the inlet valves by phasing the intake camshaft
while the exhaust valve 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.
2.2.2.6.2 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 a cam-in-block, 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 exhaust valves, CCP is
the only VVT implementation option available and requires only one cam phaser.
2.2.2.6.3 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 option 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 CO2 emissions. Increased internal EGR also results in lower engine-out
NOx emissions. The amount by which fuel consumption is improved and CO2 emissions are
reduced depends on the residual tolerance of the combustion system and on the combustion
phasing achieved. Additional improvements are observed at idle, where smaller valve overlap
could result in improved combustion stability, potentially reducing idle fuel consumption.
2.2.2.6.4 Variable Valve Lift (WL)
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. 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 incur
increased parasitic losses associated with their actuation mechanisms. A number of
manufacturers have already implemented VVL into all (BMW) or portions (Toyota, Honda, and
GM) of their fleets, but overall this technology is still available for application to most vehicles.
There are two major classifications of variable valve lift, discrete variable valve lift (DVVL) and
continuous variable valve lift (CVVL).
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.
2-20
-------
Technology Cost, Effectiveness, and Lead Time Assessment
These cam profiles may consist of a low and a high-lift lobe or other combinations of cam
profiles, and may also include an inert or blank lobe to incorporate cylinder deactivation (in the
case 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.
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 versions of its "Valvetronic" CVVL system
since 2001. 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. CVVL provides
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 typically only applied to double overhead cam (DOHC) engines.
2.2.2.7 GDI, Turbocharging, Downsizing and Cylinder Deactivation
Between 2010 and 2015, automotive manufacturers have been adopting advanced powertrain
technologies in response to GHG and CAFE standards. Just over 45 percent of MY2015 light-
duty vehicles in U.S. were equipped with gasoline direct injection (GDI) and approximately 18
percent of MY2015 light-duty vehicles were turbocharged.4 Nearly all vehicles using
turbocharged spark-ignition engines also used GDI to improve suppression of knocking
combustion. GDI provides direct cooling of the in-cylinder charge via in-cylinder fuel
vaporization.5 Use of GDI allows an increase of compression ratio of approximately 0.5 to 1.5
points relative to naturally aspirated or turbocharged engines using port-fuel-injection (e.g., an
increase from 9.9:1 for the 5.3L PFI GM Vortec 5300 to 11:1 for the 5.3L GDI GM Ecotec3 with
similar 87 AKI gasoline octane requirements).
Figure 2.2 shows a comparison of brake thermal efficiency (BTE) versus engine speed and
load between a high-volume, MY2008 2.4L 14 engine equipped with PFI and a MY2013 GM
Ecotec™ 2.5L 14 equipped with GDI. The GDI engine has a significantly higher compression
ratio, (11.3:1 vs 9.6:1), higher efficiency throughout its range of operation, and achieves higher
BMEP levels (approximately 12.5 bar vs 11.3 bar), allowing a significant increase in power per
displacement. The incremental effectiveness at approximately 2-bar BMEP and 2000 rpm was
17 percent but varied from approximately 3 percent to approximately 11 percent at other speed
and load points of importance for the regulatory drive cycles.
2-21
-------
Technology Cost, Effectiveness, and Lead Time Assessment
75kW
75kW
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
Engine Speed (rpm)
Figure 2.2 Comparison of BTE for A Representative MY2008 2.4L 14 NA DOHC PFI 4-valve/cyl. Engine
with Intake Cam Phasing (Left)c and a GM Ecotec 2.5L NA GDI Engine with Dual Camshaft Phasing
(Right)?
Note: Area of Operation > 34% BTE is Shown in Light Green. Area of Operation >35% BTE is Shown in
Dark Green.
Toyota's D-4S system combines GDI and PFI systems, with two injectors per cylinder (one
directly in-cylinder and one immediately upstream of the intake port).6-7 8 As of 2015, all Toyota
vehicles in the U.S. with GDI appear to be using a variation of the D-4S dual GDI/PFI fuel
injection system. This system increases peak BMEP, provides additional flexibility with respect
to calibration of the EMS for improved cold-start emissions and offers an efficiency
improvement over GDI alone. Based on certification data and EPA confirmatory test data,
Toyota vehicles using engines equipped with the D4S system have relatively low PM emissions
over the FTP75 cycle that are roughly comparable to PFI-equipped vehicles (<0.60 mg/mi).9 A
comparison of the Toyota 2GR-FSE engine is shown compared to a 3.5L PFI engine in Figure
2.3. The 2GR-FSE achieves a very high BMEP for a naturally aspirated engine (13.7 bar).
Although both engines have comparable displacement, they are not directly comparable because
the higher BMEP attained by the 2GR-FSE would allow further engine downsizing for a similar
application, with potential for further improvement in BTE at light load relative to the 3.5L PFI
engine. The area greater than 34 percent BTE is significantly larger for the Toyota 2GR-FSE
due to a combination of factors, including a higher compression ratio enabled by GDI and
reduced pumping losses through use of a dual camshaft phasing system that enables reduced
throttling and internal EGR at light loads.
c Based on engine dynamometer test data provided to EPA as part of "Light Duty Vehicle Complex Systems
Simulation," EPA Contract No. EP-W-07-064, work assigmnent 2-2, withPQA and Ricardo.
D Based on EPA engine dynamometer test data.
2-22
-------
Technology Cost, Effectiveness, and Lead Time Assessment
f ?M
I I
ioe
100
3000 4500 5000
Engine Spe-ed {rpm)
6000
1000
2000
3000 4000
Engine Speed (rpm)
SOOO
6000
Figure 2.3 Comparison of BTE for A Representative MY2010 3.5L V6 NA PFI 4-valve/cyl. Engine® (Left)
and a Toyota 2GR-FSE GDI/PFI Engine with Dual Camshaft Phasing' (Right).
Note: Area of Operation > 34% BTE is Shown in Light Green.
The recently redesigned Ford turbocharged 3.5L "EcoBoost™" engine in the 2017 Ford F150
also uses a dual GDI/PFI injection system to increase power, reduce emissions, and improve
efficiency,10 but other engines in Ford's EcoBoost lineup use GDI alone. In MY2015, Ford
offered a version of the EcoBoost turbocharged GDI engines as standard or optional engines in
nearly all of models of light-duty cars and trucks. Ford's world-wide production of EcoBoost
engines exceeded 200,000 units per month during CY2015.11
Approximately 13 percent of MY2015 light-duty vehicles used cylinder deactivation,
primarily in light-duty truck applications. In MY2015, General Motors introduced their
"Ecotec3" line of OHV V6 and V8 engines across their entire lineup of light-duty pickups and
truck-based SUVs. These engines are equipped with GDI, coupled-cam-phasing, and cylinder
deactivation. Both the V6 and V8 EcoTec3 engines are capable of operation on 4-eylinders
under light-load conditions. Application of GDI has synergies with cylinder deactivation. The
higher BMEP achievable with GDI also increases the BMEP achievable once cylinders have
been deactivated, thus increasing the range of operation where cylinder deactivation is enabled.
Cylinder deactivation operates the remaining, firing cylinders at higher BMEP under light
load conditions. This moves operation of the remaining cylinders to an area of engine operation
with less throttling and thus lower pumping losses (Figure 2.4) and reduced BSFC.
E Based on engine dynamometer test data provided to EPA as part of "Light Duty Vehicle Complex Systems
Simulation," EPA Contract No. EP-W-07-064, work assignment 2-2, with PQA and Ricardo.
F Based on EPA engine dynamometer test data.
2-23
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Complete Engine Operation
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
12 -
10 -
8 -
> BSFC, g/kW-hr
Approximate road load
50% Cylinder Deactivation
Approximate area of
cylinder deactivation
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
Figure 2.4 Graphical Representation Showing How Cylinder Deactivation Moves Engine Operation to
Regions of Operation with Improved Fuel Consumption over the UDDS Regulatory Drive Cycle (shaded
area).
Since 2012, improvements in crankshaft dampening systems have extended the application of
cylinder deactivation to four cylinder engines. Volkswagen introduced their 1,4L TSIEA 211
turbocharged GDI engine with "active cylinder management" in Europe for MY2013.12 This
engine is the first production application of cylinder deactivation to an 14 engine and can
deactivate 2 cylinders via cam-shifting under light load conditions. VW recently introduced a
Miller Cycle variant of the same EA211 engine family with cylinder deactivation (1.5L EA 211
evo).13 Schaeffler has developed a dynamic cylinder deactivation system for 13 and 15 engines
that alternates or "rolls" the deactivated cylinders. This system allows all cylinders to be
deactivated after every ignition cycle and reactivated during the next cycle. Cylinder deactivation
thus alternates within a single deactivation phase and not each time a new deactivation mode is
introduced. The net result is that engines with an odd number of cylinders can operate, on
average, with half their cylinder displacement (i.e., 13 can drop to 1.5 cylinders on average or an
15 can drop to 2.5 cylinders on average). Ford and Schaeffler investigated both rolling cylinder
deactivation and a system to deactivate one cylinder with Ford's EcoBoost 1.0L 13 engine and
found that, with appropriate vibrational dampening, either strategy could be implemented with
no NVH deterioration and with 3 percent or greater improvement in both real-world and EU
drive cycle fuel economy.14 Tula Technology has demonstrated a system with the capability of
deactivating any cylinder that they refer to as "Dynamic Skip Fire."15 Tula found a combined-
2-24
-------
Technology Cost, Effectiveness, and Lead Time Assessment
cycle fuel economy improvement of approximately 14 percent for an unspecified vehicle
equipped with a 6.2L PFI V8 and approximately 6 percent for an application equipped with the
GM Active Fuel Management 4/8 cylinder deactivation system. It should be noted that engines
with more opportunity for pumping loss reduction over the regulatory drive cycles (e.g., larger
displacement, naturally aspirated, PFI) generally have higher CO2 effectiveness when equipped
with cylinder deactivation.
Many automotive manufacturers have launched a third or fourth generation of GDI engines
since their initial introduction in the U.S. in 2007. Turbocharged, GDI engines are in now in
volume production at between 21-bar and 25-bar BMEP. Most recent turbocharged engine
designs now use head-integrated, water-cooled exhaust manifolds and coolant loops that separate
the cooling circuits between the engine block and the head/exhaust manifold(s). Head-integrated
exhaust manifolds (IEM) are described further in the section on thermal management in 2.2.2.11.
The use of IEM was assumed within the EPA analysis of 27-bar BMEP turbocharged GDI
engines for the FRM. The benefits, including increased ability to downspeed the engine without
pre-ignition and the potential for cost savings in the design of the turbocharger turbine housing
appear to extend to lower BMEP-level turbocharged GDI engines and will likely be incorporated
into many future turbocharged light-duty vehicle applications. The application of IEMs does
effect cooling system design and manufacturers will be required to provide sufficient cooling
system capacity if they adopt this technology.
The 2.7L Ford EcoBoost engine was introduced in the MY2015 Ford F150. This engine uses
one turbocharger per bank, IEM and dual camshaft phasing. Peak BMEP is approximately 24-bar
and the maximum towing capacity of the F150 equipped with this engine is 13,300 lbs. when
used with a 3.73:1 final drive ratio in the 2016 Ford F150. Figure 2.5 shows a comparison of
BMEP and torque vs. engine speed and BTE between a conventional MY2010 5.4L OHC V8
light-duty pickup truck engine and the MY 2015 2.7L Ford EcoBoost engine. This comparison
thus represents 50 percent engine downsizing using turbocharging and GDI. The 2.7L EcoBoost
engine has higher peak torque and power, higher peak BTE, and approximately double the area
above 34 percent BTE. Figure 2.6 shows data from operation of a 2015 Ford F150 with a 2.7L
EcoBoost engine operated over the UDDS (City Cycle) and HWFET (Highway Cycle)
superimposed over the BTE data from engine dynamometer testing. Turbocharging and
downsizing along with proper selection of transmission and final drive gear ratios and shifting
strategy moves results in operation over the regulatory drive cycles that are more closely aligned
with regions of higher BTE.
2-25
-------
Technology Cost, Effectiveness, and Lead Time Assessment
24
500
22
450
270 kW
20
400
240 kW
350
210 kW
300
180kW
250
150kW
Q-10
120kW
m 8"
90 kW
60 kW
100
30 kW
15kW
50
1000
2000
3000
Engine Speed (rpm)
4000
5000
12
500
240 kW
450
10
400
210 kW
350
8
180 kW
300
03
CO
150 kW
6
250
Q.
LU
120 kW
2
CD
200
4
90 kW
60 kW
r100
2
30 kW
15kW
50
0
1000
2000
3000
Engine Speed (rpm)
4000
5000
Figure 2.5 Comparison of BTE for A Representative MY2010 5.4L V8 NA PFI 3-valve/cyl. Engine0 (Left)
and a Ford 2.7L V6 EcoBoost Turbocharged, GDI Engine With Dual Camshaft Phasing11 (Right).
Note: Area of Operation > 35% BTE is Shown in Green.
500
450
270 kW
400
240 kW
350
210 kW
300
180 kW
250
150 kW
Q.10
120 kW
90 kW
60 kW
100
30 kW
15kW
50
1000
2000
3000
Engine Speed (rpm)
4000
5000
Figure 2.6 Engine Speed and BMEP Points Taken from 10 Hz-sampled data over the UDDS and HWFET1
Superimposed Ch er BTE Data From a Ford 2.7L V6 EcoBoost Turbocharged, GDI Engine With Dual Camshaft PhasingJ
(Right).
Figure 2.7 shows maps of BMEP and torque vs. engine speed and BTE for a representative
MY2010 2.4L PFI engine with intake camshaft phasing and a MY2012 1.0L Ford EcoBoost
turbocharged, GDI, engine with an integrated exhaust manifold (IEM) and dual camshaft
phasing.16 The 1.0L EcoBoost engine also has a peak BMEP of 25-bar and center-mounted,
spray-guided fuel injection. While not a direct comparison for purposes of engine downsizing
(the 1.0L EcoBoost is more comparable to a 1.8 - 2.0L NA PFI engine based on torque
G Based on engine dynamometer test data provided to EPA as part of "Light Duty Vehicle Complex Systems
Simulation." EPA Contract No. EP-W-07-064, work assigmnent 2-2, with PQA and Ricardo.
H Based on EPA engine dynamometer test data.
1 Based on EPA Chassis dynamometer data.
T Based on EP A engine dynamometer test data.
2-26
-------
Technology Cost, Effectiveness, and Lead Time Assessment
characteristics and rated power), this comparison of BTE does demonstrate the manner that
turbocharging and downsizing can be used to expand regions of high thermal efficiency to cover
a larger portion of engine operation. For example, the EcoBoost engine exceeds 30 percent BTE
above 6-bar BMEP/50 N-m torque over most of the engine's range of engine speeds while the
area above 30 percent BTE for the NA PFI engine is considerably smaller.
pt . . _ , l 3 ¦
1000 2000 3000 4000 5000 6000
Engine Speed (rp*n|
Figure 2.7 Comparison of BTE for A Representative MY2010 2.4L NA PFI EngineK (Left) and A Modem,
1.0L Turbocharged, Downsized GDI EngineL (Right).
Note: Area of Operation > 34% BTE is Shown in Light Green.
A comparison of the same 2.4L PFI engine with a more recent, MY2017 Honda L15B7 1.5L
Turbocharged GDI engine with IEM is shown in Figure 2.8.17-18 The torque characteristics of the
Honda engine are a closer match to the 2.4L PFI engine and the Honda engine represents
approximately 37 percent downsizing relative to the 2.4L PFI engine due to turbocharging and
includes other improvements (friction reducti on, dual cam phasing, higher rates of internal
EGR). The Honda 1.5L turbocharged GDI engine has significantly improved efficiency when
comparing BTE across 20 speed and load points of significance for the regulatory drive cycles
(1500 -2500 rpm and 2-bar to 8-bar BMEP as referenced to the 2.41 ENGINE). The BTE of the
Honda 1.5L turbocharged engine showed an incremental effectiveness of 6 percent to 30 percent
across this entire range of operation. The difference was more pronounced at lighter loads.
Incremental effectiveness was 16 percent to 30 percent below 6-bar BMEP relative to the 2.4L
engine (~112 N-m of torque).
K Based on engine dynamometer test data provided to EPA as part of "Light Duty Vehicle Complex Systems
Simulation," EPA Contract No. EP-W-07-064, work assignment 2-2, with PQA and Ricardo.
L Adapted from Ernst et al. 2011 .16
2-27
1000 2000 3000 4000 5000 6000
Engine Speed jrprn)
-------
Technology Cost, Effectiveness, and Lead Time Assessment
200
180
160
140
120
100
|80
a>
§¦60
O
¦"40
20
0
03« O3
033%* 034% CP
033% M33% 033%
Q32% 032
031% Q
027% 027% 027%
Q23% Q23% Qp
135 kW
120 kW
105 kW
90 Id/*/
15 m
7.5 m
1000
2000 3000 4000 5000
Engine Speed (rpm)
6000
135 kW
120 kW
16C -
105 kW
* 036% O3
035H. O36* O3
036% 037% O3
036% 036% O3
Q34% 034% O34
O3®'* ~Q30% O
90 kW
75 kW
60 kW
45 kW
30 kW
15 kW
7.5 kW
1000
2000
3000 4000 5000
Engine Speed (rpm)
6000
Figure 2.8 Comparison of BTE for A Representative MY2010 2.4L NA PFI EngineK (Left) and A Modern,
1.5L Turbocharged, Downsized GDI EngineM (Right).
Note: Area of Operation > 34% BTE is Shown in Light Green. Area of Operation >35% BTE is Shown in Dark
Green. BTE Was Also Compared Across 20 Operational Points of Significance for Regulatory Drive Cycles between
1500 and 2500 RPM.
Recent turbocharger improvements have included use of lower-mass, lower inertia
components and lower friction ball bearings to reduce turbocharger lag and enable higher peak
rotational speeds. Improvements have also been made to turbocharger compressor designs to
improve compressor efficiency and to expand the limits of compressor operation by improving
surge characteristics (see Figure 2.9).
M Adapted from Wada et al. 2016 and Nakano et al 2016.1718
2-28
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Original
Max Speed
Improved
Com pressor/tv i
Surge ' V
2.6-
Original
Surge Line
2.0
1.8-
120 • 1(£
1.0
0-04
0.0S
0
002
0.06
0-10
Volume flow rale V, ¦
Figure 2.9 Typical Turbocharger Compressor Map Showing How Pressure And Flow Characteristics Can
Be Matched Over a Broader Range of Engine Operation Via Surge Improvement and Higher Operational
Speed.
Turbochargers with variable nozzle turbines (VNT) use moveable vanes within the
turbocharger to allow adjustment of the effective exhaust turbine aspect ratio, allowing the
operation of the turbocharger to be better matched across the entire speed and load range of an
engine. VNT turbochargers are commonly used in modern light-duty and heavy-duty diesel
engines. The use of head-integrated exhaust manifolds (IEM) and split-coolant loops within the
engine and the use of cooled EGR (Chapters 2.2.2.8 and 2.2.2.11) can reduce peak exhaust
temperatures sufficiently to allow lower cost implementation of VNT turbochargers in spark
ignition engines. There are also synergies between the application of VNT and Miller cycle
(increased low-speed torque, improved torque response).13
Figure 2.10 Cross Sectional View of a Honeywell VNT Turbocharger
2-29
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Note: The moveable turbine vanes and servo linkage are highlighted in red.
2.2.2.8 EGR
Exhaust gas recirculation (EGR) is a broad term used for systems that control and vary the
amount of inert, residual exhaust gases left in cylinder during combustion. EGR can improve
efficiency at part-load by reducing pumping losses due to engine throttling. EGR also reduces
combustion temperatures and thus reduces NOx formation. The use of cooled EGR can reduce
knocking combustion, thus allowing compression ratio and/or turbocharger boost pressure to be
increased or spark timing to be advanced. EGR also slows the rate of combustion, so its use is
often accompanied by other changes to the engine (e.g., inducing charge motion and turbulent
combustion) to shorten combustion duration and allow improved combustion phasing. Internal
EGR uses changes in independent cam-phasing to vary the overlap between intake and exhaust
valve timing events, thus changing the amount of residual gases trapped in cylinder after cylinder
scavenging. External EGR recirculates exhaust gases downstream of the exhaust valve back into
the air induction system. With turbocharged engines, there are variants of external EGR that use
a low pressure loop, a high pressure loop or combinations of the two system types (see Figure
2.11). External EGR systems can also incorporate a heat-exchanger to lower the temperature of
the recirculated exhaust gases (e.g., cooled EGR or cEGR), improving both volumetric efficiency
and enabling higher rates of EGR. Nearly all light-duty diesel engines are equipped with cEGR
as part of their NOx emission control system. Some diesel applications also use relatively large
amounts (>25 percent) of cEGR at light- to part-load conditions to enable dilute low-temperature
combustion (see Chapter 2.2.2.11 for a more detailed description of light-duty diesel
technologies). Research is also underway to apply similar forms of low-temperature combustion
using high EGR rates to gasoline engine applications. This includes lean-homogenous
compression auto ignition (see Chapter 2.2.2.14) and other homogenous charge compression
ignition concepts (see Chapter 2.2.2.11).
The use of cEGR was analyzed as part of EPA's technology packages for post-2017 light-
duty vehicles with engines at 24-bar BMEP, primarily as a means to prevent pre-ignition at the
high turbocharger boost levels needed at 24-bar BMEP and above. The analysis did take into
account efficiency benefits from the use of cEGR with turbocharged engines due primarily to
part-load reductions in pumping losses and the reduction or elimination of commanded fuel
enrichment under high-load conditions.
Prior to 2012, there were no examples of production vehicles equipped with turbocharged
GDI engines using cEGR. The PSA 1.2L EB PureTech Turbo engine was recently launched in
the MY2014 Peugeot 308 in Europe as the first high-volume production application of cEGR on
a turbocharged GDI engine. This engine has over 24-bar BMEP and also operates using Miller
Cycle (see Chapter 2.2.2.10 for a more detailed description of Miller-Cycle). The MY2016
Mazda CX-9 2.5L SKYACTIV Turbo engine similarly combines the use of Miller Cycle with
cEGR.
2-30
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Low Pressure Loop cEGR High Pressure Loop EGR
EGR
'Control
Valve
Airflow
EGR
Cooiei
Exhaust
Catalyst
EGR
Control
Valve
Figure 2.11 A Functional Schematic Example of a Turbocharged Engine Using Two Variants of External
EGR.
Note: The Schematic On The Left Shows The Details Of A Low Pressure Loop (Post-Turbine To Pre-Compressor)
CEGR System. The Schematic Inset on the Right Shows I-Iigh Pressure Loop (Pre-Turbine to Post-Compressor)
EGR.19 In The FRM Analysis, Some TDS24 Packages And All TDS27 Packages Used Dual-Loop (Both High And
Low Pressure) EGR.
2.2.2.9Atkinson Cycle
Typical 4-cycle internal combustion engines have an effective compression ratio and effective
expansion ratio that are approximately equivalent. Current and past production Atkinson Cycle
engines use changes in valve timing (e.g., late-intake-valve-closing or LIVC) to reduce the
effective compression ratio while maintaining the expansion ratio (see Figure 2.12 and Figure
2.13). This approach allows a reduction in top-dead-center (TDC) clearance ratio (e.g., increase
in "geometric" or "physical" compression ratio) to increase the effective expansion ratio without
increasing the effective compression ratio to a point that knock-limited operation is encountered.
Increasing the expansion ratio in this manner improves thermal efficiency but also lowers peak
brake-mean-effective-pressure (BMEP), particularly at lower engine speeds. Depending on
how it is implemented, some Atkinson Cycle engines may also have sufficient cam-phasing
authority to widely vary effective compression ratio and can use this variation as a means of load
N BMEP is defined as torque normalized by cylinder displacement. It allows for emissions and efficiency
comparisons between engines of different displacement.
2-31
-------
Technology Cost, Effectiveness, and Lead Time Assessment
control without use of the standard throttle in some operating conditions, resulting in additional
pumping loss reductions.
Otto-cycle and LIVC Atkinson/Miller Cycle Valve Events
12
10
Exhaust Valve
Intake Valve
LIVC Intake Valve
8
E
E.
£
~ 6
at
>
•5
> 4
2
0
0
180
360
Crank Angle (degrees)
540
Figure 2.12 Comparison of the Timing of Valve Events for Otto-Cycle (black and orange lines) and LIVC
Implementations of Atkinson- Or Miller-Cycle (black and green lines).
Otto Cycle P-V
Atkinson Cycle P-V
Expansion
Stroke
Coi
ission
:rol
Start of
Compression
Cylinder Volume
LIVC Atkinson Cycle
Pumping Loss
1 Otto Cycle
Pumping Loss
Figure 2.13 Diagrams of Cylinder Pressure Vs. Cylinder Volume For a Conventional Otto-Cycle SI Engine
(orange line) Compared to a LIVC Implementation of Atkinson Cycle (green line) Highlighting the Reduction
in Pumping Losses.
2-32
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Prior to 2012, the use of naturally-aspirated Atkinson Cycle engines has been limited to HEV
and PHEV applications where the electric machine could be used to boost torque output,
particularly at low engine speeds. Because of this, EPA's analyses for the FRM did not include
the use of Atkinson Cycle outside of HEV and PHEV applications. Nearly all HEV/PHEV
applications in the U.S. use Atkinson Cycle, including the Honda Insight, Toyota Prius, Toyota
Camry Hybrid, Lexus 400h, Hyundai Sonata Hybrid and Chevrolet Volt. The Toyota 2ZR-FXE
used in the third-generation Toyota Prius and Lexus 200h uses a combination of LIVC Atkinson
Cycle, cooled EGR, and port-fuel-injection (PFI) to achieve a peak BTE of 38.5 percent, the
highest BTE achieved to date for a production spark-ignition engine. Further refinements to this
engine, including increased tumble to increase both the speed of combustion and EGR tolerance,
have resulted in peak BTE of 40 percent.20
Since 2012, Atkinson Cycle engines have been introduced into non-hybrid applications.
These applications use camshaft-phasing with a high degree of authority together with either
GDI (e.g., Mazda SKYACTIV-G 1.5L, 2.0L and 2.5L engines, Toyota 2GR-FKS engine), PFI
(MY2017 Hyundai Elantra "Nu" 2.0-liter PFI Atkinson) or a combination of PFI with cooled
EGR (Toyota 1NR-FKE and 2NR-FKE engines). As of MY2017, all of Mazda's engines for the
U.S. market are either Atkinson Cycle or Miller Cycle (boosted Atkinson). Toyota's 2GR-FKS
engine became an optional engine offered in the Toyota Tacoma pickup truck beginning in
MY2016. The Tacoma is currently the mid-size pickup truck segment sales leader in the U.S.
The Toyota Tacoma equipped with the 2GR-FKS Atkinson Cycle engine has an SAE J2807 tow
rating of 6,800 pounds. The Hyundai "Nu" 2.0-liter PFI Atkinson Cycle engine is the base
engine offering in the Hyundai Elantra. The Hyundai Elantra is currently within the top 5 in
sales within the compact car segment in the U.S.
The effective compression ratio of Atkinson Cycle engines can be varied using camshaft
phasing to increase BMEP and GDI (Mazda) or cEGR (Toyota) are used, in part, for knock
mitigation. These engines from Mazda and Toyota also incorporate other improvements, such as
friction reduction from valvetrain and piston design enhancements. The Toyota 1NR-FKE 1.3L
13 and 2NR-FKE 1,5L 14 engines achieve a peak BTE of 38 percent, very close to the BTE
achieved with the 2ZR-FXE engine used in the Toyota Prius.20'21 EPA testing of 2.0L and 2.5L
variants of the Mazda SKYACTIV-G engine achieved peak BTE of 37 percent while using either
88AKI (91 RON) or 92 AKI (96 RON) fuel. More important from a standpoint of drive-cycle
fuel economy and CO2 emissions was the very large "island" of more than 32 percent BTE
(Figure 2.14) which, depending on the transmission and road load, would cover most operation
over the UDDS and HWFET regulatory drive cycles depending on the specific vehicle
application (e.g., road loads, final drive, gear-ratio spread). In the case of the Mazda
SKYACTIV-G engines, the use of GDI and cam-phasing resulted in increased BMEP and rated
power relative to the previous PFI, non-Atkinson versions of this engine and allowed a small
degree of engine downsizing (e.g., replacement of the previous 2.5L PFI engine with the 2.0
SKYACTIV-G) on some Mazda platforms with equal or improved performance. In the case of
the Toyota 1NR-FKE, the use of cEGR and cam-phasing allowed BMEP to be maintained
relative to peak BMEP of the Non-Atkinson Cycle engine it replaced and allowed the use of a
lower cost PFI fuel system. Both the Mazda and Toyota Atkinson Cycle engines use electro-
mechanical systems for camshaft phasing on the intake camshaft.
2-33
-------
Technology Cost, Effectiveness, and Lead Time Assessment
75kW
Engine Speed (rpm)
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
Figure 2.14 Comparison of BTE for a Representative MY2010 2.4L NA PFI Engine0 (left) and a 2.0L NA
GDI LIVC Atkinson Cycle Engine (right) tested by EPA.p'22
A recent benchmarking analysis by EPA of a 2014 Mazda SKYACTIV-G naturally aspirated
(NA) gasoline direct injection (GDI) engine showed a peak BTE of approximately 37 percent,
relatively high for SI engines.p ;-! This was in part due to an ability to use late-intake-valve-
closing (LIVC) Atkinson-cycle operation to decouple the knock-limited effective CR from the
expansion ratio available from a very high 13:1 geometric CR. This can be seen in the variation
of effective compressi on ratio observed during dynamometer testing, where the maximum
effective CR (~11 to 11.5:1) is comparable to other GDI naturally aspirated GDI engines having
87 AKI gasoline as a recommended fuel, for example 2015 and later GM Ecotec3 V6 and V8
engines (see Figure 2.15).
1000 1500 2000 2500 3000 3500 4000
Speed (RPM)
Figure 2.15 Measured effective compression ratio for 2.0L NA GDI LIVC Atkinson Cycle Engine (right)
tested by EPA.
u Based upon engine dynamometer data provided to EPA under a contract with PQA and Ricardo, "Light Duty
Vehicle Complex Systems Simulation" EPA Contract No. EP-W-07-064. work assigmnent 2-2.
p Derived from EPA engine dynamometer data first presented by Lee et al. 2016.22
2-34
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Note that the thick black line denotes measurement and calculation limits for mapping and does not necessarily reflect maximum rated
torque at each speed condition.
The Mazda SKYACTIV-G is one of the first implementations of a naturally-aspirated, LIVC
Atkinson-cycle engine in U.S. automotive applications outside of hybrid electric vehicles (HEV)
and also appears to be the first Atkinson-cycle engine to use GDI. Port-fuel-injected (PFI)
Atkinson-cycle engines have been used in hybrid electric vehicle applications in the U.S. for
over a decade. PFI/Atkinson-cycle engines have demonstrated peak BTE of approximately 39
percent in the 2015 Honda Accord HEV and 40 percent in the 2016 Toyota Prius HEV.
Atkinson-cycle engines can achieve comparable or better peak BTE in comparison with
downsized, highly boosted, turbocharged GDI engines like the Ricardo EGRB configuration
analyzed within the FRM. However, such modern turbocharged GDI engines often have
relatively high BTE across a broader range of engine speed and torque as well as improved BTE
and fuel consumption at light loads compared with Atkinson-cycle engines, as shown in Figure
2.16. Based on EPA's initial engineering analysis of the Mazda SKYACTIV-G engine, it
appeared that another reasonable, alternative technological path to both high peak BTE and a
broad range of operation with high BTE might be possible through the application of cooled-
EGR (cEGR), a higher compression ratio, and cylinder deactivation to a naturally-aspirated
GDI/Atkinson-cycle engine like the SKYACTIV-G. Discussion of modeling and engine
development by EPA of application of these technologies to an Atkinson-cycle engine are
summarized in Chapter 2.3 of the TSD.
135 kW
200
200
12
120 kW
120 kW
180
180
105 kW
105 kW
160
160
10
90 kW
90 kW
' 140
140
8
120
120
75kW
75 kW
TO
CO
100
100
6
60 kW
60 kW
Q-10
Q.
ID
S
OQ
45kW
45 kW
4
30 kW
30 kW
40
40
2
15 kW
10 kW
7.5 kW
7.5 kW
0
1000
2000
3000
4000
5000
6000
1000
2000
3000
4000
5000
6000
Engine Speed (rpm) Engine Speed (rpm)
Figure 2.16 A Comparison of BSFC Maps Measured For The 2.0L 13:ICR SKYACTIV-G Enginep (left) and
Modeled For A 1.0L Ricardo "EGRB Configuration"0 (right).
2.2.2.10 Miller Cycle
Like Atkinson Cycle, Miller Cycle engines use changes in valve timing to reduce the effective
compression ratio while maintaining the expansion ratio. Automakers have investigated both
early intake valve closing (EIVC) and LIVC variants. There is some disagreement over the
application of the terms Atkinson or Miller Cycle to EIVC and LIVC valve event timing and
sometimes the terms are used interchangeably. For the purpose of EPA's analyses, Miller Cycle
2-35
-------
Technology Cost, Effectiveness, and Lead Time Assessment
is a variant of Atkinson cycle with intake manifold pressure boosted by a either a turbocharger
and/or a mechanically or electrically driven supercharger. It is simply an extension of Atkinson
Cycle to boosted engines and can use either EIVC or LIVC. The first production vehicle offered
using Miller Cycle was the MY1995 Mazda Millenia S, which used the KJ-ZEM 2.3L PFI
engine with a crankshaft-driven Lysholm compressor for supercharging. Until recently, no
Miller Cycle gasoline SI engines were in mass production after 2003, and Miller Cycle was not
evaluated as a potential gasoline engine technology as part of the 2017-2025 GHGFRM.
As with Atkinson Cycle engines, the use of GDI and camshaft-phasing with a high degree of
authority have significant synergies with Miller Cycle. Modern turbocharger and charge air
cooling systems allow Miller Cycle engines to attain BMEP levels approaching those of other
modern, downsized, turbocharged GDI engines. The 1.2L 13 PSA "EB PureTech Turbo" Miller
engine launched in Europe, N. Africa and S. America in the MY2014 Peugeot 30824. In addition
to Miller Cycle, the engine also uses cEGR. This engine has a maximum BMEP of 24-bar and is
similar in many respects to the Ford 1.0L 13 EcoBoost but achieves 35 percent BTE over a
slightly broader area of operation vs. 34 percent BTE for the EcoBoost (see Figure 2.17).
-200
-150
34% BTE
-100
212
- 50
1000 2000 3000 4000 5000 6000
-240
-180
35% BTE
212
"120
- 50
1000 2000 3000 4000 5000 6000
Engine Speed (rpm) Engine Speed (rpm)
Figure 2.17 Comparison of BTE for Downsized, Turbocharged GDI Engines.
Note: Ford 1.0L EcoBoost Engine Is On The Left And A 1.2L Miller Cycle PSA EB Puretech Engine Is On The
Right. A More Detailed BTE Map Is Not Yet Available For The PSA Engine.
In MY2017, VW will be launching a Miller Cycle variant of the 2.0L EA888 turbocharged
GDI engine in the U.S. The VW implementation of Miller Cycle has a second Miller Cycle cam
profile and uses camshaft lobe switching on the intake cam to go into and out of an EIVC
implementation of Miller Cycle.25 26 The peak BTE of 37 percent is higher than that of the PSA
Miller cycle engine, in part due to a higher expansion ratio (geometric CR of 11.7:1 for the VW
engine vs. 10.5:1 for the PSA engine). Like the PSA engine, the VW uses high-pressure cEGR.
Peak BTE is comparable to the Mazda SKYACTIV-G engines but is available over a broader
range of speed and load conditions. Both Atkinson and Miller Cycle engines show broad areas
of operation at greater than 32 percent BTE. Figure 2.18 shows a comparison between a
2-36
-------
Technology Cost, Effectiveness, and Lead Time Assessment
MY2010 3.5L NA PFIDOHC V6 and the VW 2.0L EA888 Miller Cycle engine with
comparable torque delivery. The area of operation at greater than 32 percent BTE is
approximately double for the Miller Cycle engine relative to the DOHC PFI engine. BTE is
improved by approximately 40 percent at light load for the Miller Cycle engine and peak BTE is
improved approximately 6 percent. Mazda recently introduced a 2.0L Miller Cycle engine with
cEGR and a unique exhaust scavenging system in the 2016 CX9 SUV.27
350
300
250
|200
034.% (J34%
O33* 033% Q33%
®33JlL_^a9ir'^32l)liN\ ^
Qoth, Q30.% O30
~©27% Q27% 027%"--
Q23% £}28V__0B*
n150
100
300
250
1200
;150
100^
50-
037% Ob.7%
O3®" Q37% 037%
O36'1 Q37% . Q37%
¦035% O38* O3^--
035% O86*
O3** O3" - ©34%
Q31% 031% 0p1%
1000
2000
3000 4000 5000
Engine Speed (rpm)
6000
500 1000 1500 2000 2500 3000 3500 40004500 5000 5500
Engine Speed (rpm)
200 kW
180 kW
160 kW
140 kW
120 kW
100kW
80 kW
60 kW
40 kW
20 kW
10 kW
Figure 2.18 Comparison of BTE for A Representative MY2010 3.5L NA PFI V6 Engine0 (Left) And A
Downsized 2.0L 14 Miller Cycle EngineR (Right).
Note: The Light Green Area Shows Regions of >34% BTE. The Dark Green Area Shows a Region >35% BTE.
Since VW has published detailed data for both Miller Cycle and a turbocharged GDI (non-
Miller) variants of the EA888 series of engines, a more direct comparison between turbocharged,
downsized GDI and Miller Cycle engines is possible. Figure 2.19 shows BTE for both variants
of the 2.0L 14 VW EA888 engine. When comparing BTE at comparable BMEP, there is a 6-10
percent incremental improvement for the Miller Cycle engine relative to the turbocharged GDI
engine over a broad area of operation from 1500-2500 rpm and from 2-bar to 12-bar BMEP (i.e.,
below 55 -60 percent of peak BMEP - areas of importance for the regulatory drive cycles).s
Comparing BTE of the 2.0 Miller cycle variant to the smaller displacement, 1.8L version of the
same engine family (similar 22-bar BMEP to the 2.0L turbocharged GDI, but equivalent torque
to the 2.0L Miller Cycle engine) lowers the incremental effectiveness for Miller Cycle to
approximately 4-7 percent relative to a turbocharged GDI engine and comparable partial load
operation from 1500-2500 rpm. Confidential business information from a Tier 1 automotive
supplier provided an estimate of approximately 5 percent CO2 combined-cycle incremental
Q Based upon engine dynamometer data provided to EPA under a contract with PQA and Ricardo, "Light Duty
Vehicle Complex Systems Simulation" EPA Contract No. EP-W-07-064, work assigmnent 2-2.
R Adapted from Wunns et al. 2015.Error! liooklm"'kdefmed-
s Note that VW did not significantly change the turbocharging system when applying Miller Cycle to this engine
family, so the Miller Cycle variant has a peak BMEP of 20-bar instead of 22-bar due to the reduced volumetric
efficiency from EIVC. Turbocharger improvements (e.g., higher pressure ratio and different flow characteristics)
would be necessary to maintain the 2.0L Miller Cycle engine at 22-bar BMEP, thus comparisons in this case are
limited to 20-bar BMEP and below.
2-37
-------
Technology Cost, Effectiveness, and Lead Time Assessment
benefit for Miller Cycle relative to a 24-bar BMEP turbocharged, downsized engine and a loss of
approximately 8-12 percent peak BMEP due to reduced volumetric efficiency for Miller Cycle.
This is consistent relative to the data published by VW. There may also be synergies between
Miller Cycle and CDA. A comparison Miller and non-Miller variants of the VW EA211 TSI
turbocharged engine, both with CDA, shows a relative effectiveness of 5-30 percent for the
Miller Cycle variant of the engine over regions of operation that are important for U.S.
regulatory drive cycles.13 The Miller Cycle variant of the VW EA211 TSI has a geometric CR of
12.5:1 and uses a VNT turbocharger.
-SOfcW
Ill ' ¦ ill -1 'n I i - Tii i«- i: J -111.
500 tODO tSOO 2CHJQ 2500 MOP 35W 40W 4500 SOCK) 5S0Q
505 1000 1500 2000 2500 3000 3500 WOO 4500 5000 5500
Engine Speed (rpinf Enjjirw Jpeed Itpm|
Figure 2.19 Comparison of BTE for 2015 Turbocharged, Downsized GDI (left) and 2017 Miller Cycle (right)
variants of the same engine family, the 2.0L VW EA888.R
Note: Green area shows region of high (35%) BTE.
2.2.2.11 Light-duty Diesel Engines
Diesel engines have characteristics that differ from gasoline spark ignition (SI) engines and
allow improved fuel efficiency, particularly at part-load conditions. These include reduced
pumping losses due to lack of (or greatly reduced) throttling, and a combustion cycle that
operates at a higher compression ratio and at very lean air/fuel ratio when compared with an
equivalent-performance gasoline engine. Operating with a lean-of-stoichiometric air/fuel ratio
poses challenges with respect to NOx control, requiring either a NOx adsorption catalyst (NAC),
urea or ammonia-based selective catalytic reduction (SCR) or some combination of NAC and
SCR in order to meet Federal Tier 3 and California LEV III NOx emissions standards.
Beginning with Federal Tier 2 emission standards. It has also been necessary to equip light-duty
diesels with catalyzed diesel particulate filters (CDPFs) in order to comply with light duty PM
emission standards.
Detailed analysis of the vehicle simulation results used within the FRM uncovered some
shortcomings within the MSC EASY5 vehicle simulations used as light-duty diesel vehicle GHG
effectiveness inputs into the Rieardo Surface Response Model. The modeled light-duty diesel
technology packages did not operate in the most efficient regions of engine operation. This may
have been in part due to inconsistencies in the application of the optimized shift strategy and in
part due to an oversight that resulted in the apparent oversizing of light-duty diesel engine
displacements. For example, plotting the average engine speed and load operating points over
2-38
-------
Technology Cost, Effectiveness, and Lead Time Assessment
the regulatory drive cycles for the MSC EASY5 diesel simulations on top of the diesel engine
maps showed that there was significant potential for improvement in the choice of selected gear.
As a result, additional analyses using the ALPHA vehicle simulation model have been conducted
for light-duty diesel engine technology packages in order to update GHG effectiveness from
these packages.
Light-duty diesel engines have also evolved considerably over the last five years, particularly
in Europe. Modern light-duty diesel engine designs appear to be following similar trends to
those of turbocharged GDI engines and, in some cases, heavy-duty diesel engine designs,
including:
• Engine downsizing (increased peak BMEP)
• Engine down-speeding
• Advanced friction reduction measures
• Reduced parasitics
• Improved thermal management
• Use of a combination of both low- and high-pressure-loop cooled EGR
• Advanced turbocharging, including the use of VNT and sequential turbocharging
• Incorporation of highly-integrated exhaust catalyst systems with high NOx and PM
removal efficiencies
• Adoption of high-pressure common rail fuel injection systems with higher injection
pressures and increased capability (i.e., multiple injections per firing cycle)
The highest BMEP engines currently in mass-production for high-volume light-duty vehicle
applications are all diesel engines. MY2016-2017 light-duty diesel engines are available from
Honda, BMW and Mercedes Benz in the EU with approximately 26-bar to 29-bar BMEP and
peak cylinder pressures at or above 200-bar.28"29"30 The light-duty diesel technology packages
used in the FRM analyses relied on engine data with peak BMEP in the range of 18 - 20 bar.
These were engine configurations using single-stage turbocharging with electronic wastegate
control, high-pressure or low-pressure (single-loop) cooled EGR, and common-rail fuel injection
with an 1800 bar peak pressure. The cost analysis in the FRM for advanced light-duty diesel
vehicles assumed use of using a DOC+DPF+SCR system for meeting emissions standards for
criteria pollutants.
In response to EPA Heavy Duty GHG emissions standards, large Class 8 heavy-duty truck
engine designs have exceeded 50 percent BTE.31'32 Despite their inherent differences, there now
appears to be a significant transfer of technology from heavy-duty diesel engines to much
smaller bore, higher speed light-duty diesel engines underway, particularly for engines with high
BMEP. Use of CAE tools to design complex, stepped-geometry steel piston crowns and the use
of carefully designed piston oil-cooling galleries result in remarkably similar approaches when
comparing recent approaches to heavy-duty truck piston designs to recent light-duty diesel
engine piston designs such as that of the Mercedes-Benz OM654.31'33 The Mercedes-Benz
OM654 engine incorporates other design elements that are similar to current heavy-duty diesel
engine designs, including driving the camshaft and some auxiliaries off of the rear of the engine,
the use of a high pressure common rail (HPCR) fuel injection systems with 2050 bar peak
pressure and the use of a VNT turbocharger. BMW's B57 light-duty diesel engine used in the
MY2017 BMW 730d and 740d uses an HPCR fuel injection system currently with 2500 bar peak
2-39
-------
Technology Cost, Effectiveness, and Lead Time Assessment
pressure and with capability to expand peak pressures to 3000 bar. Driving injection pressures
higher allows more flexibility for use of multiple injections and allows better optimization of
combustion phasing. Modern, high BMEP light-duty diesel engines using conventional
diffusional combustion are capable of peak BTE of approximately 42 percent (see Figure 2.20).34
26
24
22
20
18
16
14
I12
g>
m 8
6
4
2
Figure 2.20 Comparison Of BTE For A Downsized SI 2.0L 14 Miller Cycle Engine (Left)T And A 1.7L 14
Turbocharged Diesel Engine With HPCR, Low And High Pressure Loop CEGR, And VNT Turbocharger
(Right).u
Note: Green area shows region of high (35%) BTE.
Engine Speed (rpm)
1000 1500 2000 2500 3000 3500 4000
Engine Speed (rpm)
Advanced turbocharging and cooled EGR systems allow higher rates of EGR to be driven
and, when combined with more capable, higher pressure (2000-3000 bar) HPCR systems can
allow a degree of operation at light loads using pre-mixed charge compression ignition (PCCI) or
other low-temperature modes of combustion with inherently low NOx and PM emissions and
reduced thermal losses over a broader area of engine operation. Cummins "Light-duty Efficient,
Clean Combustion" engine development program for the U.S. DOE used mixed-mode, part-load
PCCI/high-load diffusional combustion approach and achieved a 20 percent improvement in
uncorrected city-cycle fuel economy (e.g., from 20.3 mpg to 24.5 mpg) when compared to a
more conventional diesel in a 5000 lb. inertial test weight SUV at Tier 2, Bin 5 emissions levels.
Peak BTE for the PCCI combustion mode was approximately 46 percent compared with 42
percent peak BTE for conventional diffusional diesel combustion. Cummins developed a similar
dual-mode combustion approach as part of the Advanced Technology Powertrains for Light-
Duty (ATP-LD) and the Advanced Technology Light Automotive Systems (ATLAS) engine
development programs for the U.S. DOE.35-36 The engines developed as part of this program
combined dual-mode PCCI/diffusional combustion together with further improvements to the
turbocharger and charge air cooler systems, improved integration of the catalytic CDPF and
urea-SCR systems and addition of a NAC system for storage of cold-start NOx emissions.
Developmental engines and emissions control systems were integrated into Nissan Titan full-size
T Adapted from Wurms et al. 2015.
u Adapted From Busch Et Al. 2015.34
2-40
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2-wheel-drive pickup trucks and achieved emissions consistent with Tier 3 Bin 30 compliance
and 21.8/34.3/26.0 City/Highway/Combined (uncorrected) fuel economy at a 5500 lb. inertial
test weight. A similar engine used in the mid-size Nissan Frontier 4-wheel drive pickup at
reduced peak BMEP (21.3 bar vs. 23.4 bar in the Titan demonstration) achieved a 35 percent
combined cycle fuel economy improvement relative to the MY2015 4.0L PFI V6 Nissan
Frontier.37
2.2.2.12 Thermal Management
Most recent turbocharged engine designs now use head-integrated, water-cooled exhaust
manifolds and coolant loops that separate the cooling circuits between the engine block and the
head/exhaust manifold(s) (Figure 2.21). Examples include the head-integrated exhaust
manifolds (IEM) and split-coolant loops used with the Ford 1.0L 13, 1.5L 14, 2.0L 14 and 2.7L
V6 EcoBoost engines, the 2.0L VW EA888 engine, the GM EcoTec SGE 1.0L 3-cylinder and
1,4L 4 cylinder engines, and the PSA 1,2L EB PureTech Turbo. The use of IEM and split-
coolant-loops is now also migrating to some naturally aspirated GDI and PFI engines, including
the GM 3.6L V6 LFX and EcoTec 1.5L engines and the 1.0L 3-cylinder Toyota 1KR-FE
ESTEC. These types of thermal management systems were included in the FRM analysis of
turbocharged GDI engines at BMEP levels of 24-bar and above but were not considered for
turbocharged engines at lower BMEP levels or for naturally aspirated engines. Benefits include:
• Improved under-hood thermal management (reduced radiant heat-load)
• Reduced thermal gradients across the cylinder head
• Reduction in combustion chamber hot spots that can serve as pre-ignition sources
• Improved knock limited operation
• Reduce or eliminate enrichment required for component protection, particularly at
low-speed/high-load conditions
° Enable additional engine "down-speeding" without encountering enrichment
• Improved control of turbine inlet temperature (turbocharged engines only)
° Enable use of lower-cost materials turbine and turbine housing materials
° Enable use of variable-geometry turbines similar to light-duty diesel applications
• Improved catalyst durability
• Shorter time to catalyst light-off after cold-start
• Improved coolant warmup after cold start
• Reduced noise
• Lower cost and parts count
° Improved durability (fewer gaskets to fail)
• Reduced weight (savings of approximately 1 kg/cylinder)
2-41
-------
Technology Cost, Effectiveness, and Lead Time Assessment
j-I; •
: I
*1$
Figure 2.21 Exhaust Manifold Integrated Into a Single Casting with the Cylinder Head
2.2.2.13 Reduction of Friction and Other Mechanical Losses
In urban driving, approximately 60 percent of engine losses are due to mechanical losses,
including engine friction.58 Piston and cylinder friction from the piston rings and piston skirts
account for 35 percent or more of engine friction in modern light-duty gasoline engines and
approximately 50 percent of engine friction in modern light-duty diesels engines.38 39 40 The
remaining frictional losses are primarily due to crankshaft, connecting rod, valvetrain and
balance shaft friction. Piston skirt friction accounts for approximately 30 percent of piston
friction. Molybdenum disulfide (M0S2) and Diamond-like carbon (DLC) piston skirt coatings
have demonstrated part-load engine friction reductions of approximately 16 percent and 20
percent, respectively.39 Improvements in cylinder bore surface treatments such as plasma
coatings29 30 41 and laser roughening42 have also been introduced in recent engine designs to
reduce engine friction and improve cylinder bore wear characteristics.
Offsetting the crankshaft from the bore centerline, sometimes referred to as a desaxe cylinder
arrangement, can be used to reduce side forces on the piston and piston rings during the power
stroke, reducing friction piston/liner friction and reducing component wear.43 For example, the
2ZR-FXE engine used in the 2009-2015 Toyota Prius and the 2ZR-FE engine in the 2009-2016
Toyota Corolla have the crankshaft centerline shifted 8 mm towards the intake side of the engine
to reduce friction.44
Schaeffler has developed roller bearings that can be applied to the first and last crankshaft
main beari ngs without the added complexity of using built crankshafts or split main bearings to
reduce crankshaft friction and increase front journal load bearing capability when used with
higher power P0 mild hybrid systems. Roller bearing balance shafts for 3- and 4-cylinder
2-42
-------
Technology Cost, Effectiveness, and Lead Time Assessment
engines have also been developed by Schaeffler, BMW and others that can reduce balance shaft
friction by approximately 50 percent.
In addition to reducing engine mechanical losses, engine friction reduction also improves
engine restart when combined with stop/start systems. Reducing engine friction can also allow
additional engine downspeeding while maintaining idle and off-idle engine NVH characteristics.
Hyundai and Delphi used a MY2011 2.4L 4-cylinder GDI engine to demonstrate a combined-
cycle fuel economy improvement of 4 percent by using a combination of a M0S2 piston skirt
coating, CrN physical vapor-deposition coated piston rings, low tension oil control rings and
engine downspeeding.45 They also achieved a further 2.9 percent combined-cycle fuel economy
improvement through use of a 2-stage variable displacement oil pump.
2.2.2.14 Potential Longer- Term Engine Technologies
In addition to the engine technologies considered for this Proposed Determination assessment,
and discussed above, there are many other engine technology development efforts underway that
may be fruitful in the longer-term. While introduction of engines using these combustion
concepts may occur prior to 2025, EPA does not expect significant penetration of these
technologies into the light-duty vehicle fleet in the 2022 to 2025 time frame.
Homogenous charge compression ignition (HCCI), gasoline compression ignition and other
dilute, low-temperature compression ignition gasoline combustion concepts are topics of
considerable automotive research and development due to the potential for additional pumping
loss improvements at light and partial load conditions and reduced thermal losses. Challenges
remain with respect to combustion control, combustion timing, and, in some cases, compliance
with Federal Tier 3 and California LEV3 NMOG+NOx standards.
Engines using variable compression ratio (VCR) appear to be at a production-intent stage of
development, but also appear to be targeted primarily towards limited production, high
performance and very high BMEP (27-30 bar) applications. At lower BMEP levels, other
concepts (e.g., Atkinson Cycle for NA applications, Miller Cycle for boosted applications)
provide a similar means to vary effective compression ratio for knock mitigation with reduced
cost and complexity with some tradeoffs with respect to volumetric efficiency.
One vehicle manufacturer recently entered production with a water injection system for knock
mitigation. Injection of water and water/methanol or water/ethanol mixtures into the intake
systems of turbocharged and/or mechanically supercharged engines for knock mitigation is not a
new concept. Aircraft engines predating World War II and some of the first turbocharged
automobile applications for the U.S. market in the 1960s used such systems for knock mitigation.
Water injection systems compete with other means of knock mitigation (EGR, Atkinson Cycle,
Miller Cycle, and IEM/split-cooling) that do not require fluid replenishment. Current and near
term applications appear to be limited to low-volume production, high performance vehicles.
The DOE Co-Optimization of Fuels and Engines (Co-Optima) initiative aims to improve
near-term efficiency of spark-ignition (SI) and compression ignition engines through the
identification of fuel properties and design parameters of existing base engines that maximize
performance.
2-43
-------
Technology Cost, Effectiveness, and Lead Time Assessment
According to DOE, Co-Optima is a first-of-its-kind effort brings together multiple DOE
offices, national laboratories, and industry stakeholders to simultaneously conduct tandem fuel
and engine R&D and deployment assessment in order to maximize energy savings and on-road
vehicle performance, while also reducing long-term transportation-related petroleum
consumption and GHG emissions. Two parallel research tracks focus on: 1) improving near-term
efficiency of spark-ignition (SI) engines through the identification of fuel properties and design
parameters of existing base engines that maximize performance. The efficiency target represents
a 15 percent fuel economy improvement over state-of-the-art, future light-duty SI engines with a
market introduction target of 2025; and 2) simultaneous testing of new fuels with existing CI
engines (as well as advanced compression ignition [ACI] combustion technologies as they are
developed) to enable a longer-term, higher-impact series of synergistic solutions. The fuel
economy target represents a 20 percent improvement over state-of-the-art, future light-duty SI
engines with a market introduction target of 2030. By using low-carbon fuels, such as biofuels,
GHGs and petroleum consumption can be further reduced. EPA will continue to closely follow
the Co-Optima program to provide input to DOE, including through EPA's technical
representative on the Co-Optima External Advisory Board, as this program has the potential to
provide meaningful data and ideas for GHG and fuel consumption reductions in the light-duty
vehicle fleet for 2026 and beyond.
2.2.3 Transmissions: State of Technology
2.2.3.1 Background
The function of a transmission system is to reduce the relatively high engine speed and
increase the torque, so that the power output of the engine can be coupled to the wheels. The
complete drivetrain includes a differential (integral to the transmission on front-wheel-drive
vehicles; separate on rear-wheel-drive vehicles) which provides further speed reduction, and
often a hydraulic torque converter which provides significant torque multiplication at low speed
conditions. The complete drivetrain - torque converter, transmission, and differential - is
designed as a set to best match the power available from the engine to that required to propel the
vehicle.
2-44
-------
Technology Cost, Effectiveness, and Lead Time Assessment
CVT{h)
Otfier
100%
75%
50%
25%
c 0%
o
~tj 100%
ZJ
TD
75% -
1980
19S5 1990 1995 2000 2005 2010 2015 2020
50% -
25% -
0% -
Car
Truck
Transmission
Lockup?
Number of Gears
Key
Automatic
No
3
A3
Semi-Automatic
4
A4
Automated hfenual
5
A5
6
AS
7
A7
Yes
2
L2
3
L3
4
L4
5
L5
6
7
L7
8
9
Manual
—
3
M3
4
M4
5
M5
6
MB
7
M7
ContinixxEly Variable
(non-hybrid)
—
—
CVTTrvh)
Continuously Variable
(tytrk*
-
—
CVT{h)
a her
—
"
Other
Model Year
Figure 2.22 Transmission Technology Production Share, 1980 - 201546
Different transmission architectures are available for use in light-duty vehicles. Conventional
automatic transmissions (ATs) are the most popular type, and still dominate the light-duty fleet,
as seen in Figure 2.22. Manual transmissions (MTs), although less popular than in the past, are
also still part of the fleet. Both ATs and MTs have, among other improvements, seen an increase
in the number of gears employed. Figure 2.22 shows the recent gains in six, seven, eight, and
nine speed transmissions in both the car and light truck segment. Two other transmission types
have also seen an increase in market share. These are dual-clutch transmissions (DCTs), which
have significantly lower parasitic losses than ATs, and continuously variable transmissions
(CVTs), which can vary their ratio to target any place within their overall spread. Each of these
four types of transmissions is discussed in more detail in the sections below.
2.2.3.2 Transmissions: Summary of State of Technology
As EPA stated in the Draft TAR, in the analysis conducted for the 2012 rule, EPA estimated
that DCT transmissions would be very effective in reducing fuel consumption and CO2
emissions, less expensive than current automatic transmissions, and thus a highly likely pathway
used by manufacturers to comply with the standards. This expectation was supported by
comments from many OEMs at the time of the 2012 rule indicating that DCTs were part of their
future compliance strategies. However, DCTs thus far, have been used in only a small portion of
the fleet as some OEMs have reported in meetings with EPA. In addition, some vehicle owners
have cited drivability concerns for DCT.47 EPA also discussed in the Draft TAR that the 2017-
2025MY FRM analysis also predicted a low effectiveness associated with CVTs (due to the high
internal losses and small ratio spans of CVTs in the fleet at that time), and thus CVTs were not
included in the FRM fleet modeling. However, internal losses in current CVTs have been much
reduced and ratio spans have increased from their predecessors, leading to increased
effectiveness and further adoption rates in the fleet, particularly in the smaller car segments. The
new CVTs also tend to give the best effectiveness for their cost.
2-45
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Again in the Draft TAR we mentioned that in the 2017-2025MY FRM, EPA estimated that
step transmissions with higher numbers of gears (e.g., AT8s) would be slowly phased into the
fleet. However, AT8s have been "pulled ahead," appearing in substantial numbers even before
2015MY. In addition, manufacturers have introduced nine speed transmissions and since the
Draft TAR Ford has released an F150 with a 10-speed transmission. Transmissions with more
than 8-speeds were not considered in the 2017-2025MY FRM.
Consistent with the Draft TAR, highlights of transmission technology analysis in this
Proposed Determination include: (a) the technology packages and vehicle classes where DCTs
are applicable have been re-evaluated to reflect manufacturers' current choices, (b) the
effectiveness of CVTs has been re-examined and increased to reflect current vintage CVTs and
their use in the fleet, and (c) nine and ten-speed transmissions were considered when
determining the effectiveness of future transmissions in the fleet.
2.2.3.3 Sources of Transmission Effectiveness Data
In addition to the sources of transmission effectiveness data cited in the 2012 rule and Draft
TAR, EPA also used data from a wider range of available sources to update and refine
transmission effectiveness for this analysis. These sources included:
• Peer-reviewed journals, peer-reviewed technical papers, and conference proceedings
presenting research and development findings
• Data obtained from transmission and vehicle testing programs, carried out at EPA-
NVFEL, ANL, and other contract laboratories
• Modeling results from simulation of current and future transmission configurations
• Confidential data obtained from OEMs and suppliers on transmission efficiency
For transmission testing programs, EPA contracted with FEV Engine Technologies to test
specific transmissions in a transmission component test stand. The testing program was
primarily designed to determine transmission efficiency and torque loss over a range of input
speeds, input loads, and temperatures. In addition, other driveline parameters, such as
transmission rotational inertia and torque converter K-factor were characterized. Two automatic
transmissions have been characterized in this test program, which is still on-going. Torque loss
maps were generated for both a six-speed 6T40 GM automatic transmission and an eight-speed
845RE FCA automatic transmission (see Figure 2.23).
2-46
-------
Technology Cost, Effectiveness, and Lead Time Assessment
100 150
Input Torque (Nm
Figure 2.23 Average Torque Losses (Left) And Efficiency (Right) In Each Gear For An Eight-Speed 845RE
Transmission From A Ram, Tested At 100 °C And With Line Pressures Matching Those Measured In-Use In
The Vehicle. Torque Losses Were Averaged Over 1000 Rpm - 2500 Rpm. This Transmission Is A Clone of the
ZF 8HP45.
In addition to contracting to test specific transmission, EPA has obtained torque loss maps
and/or operational strategies for current generation transmissions from manufacturers and
suppliers. These maps are CBI, but have been used to inform EPA on the effectiveness of
transmissions currently on the market. Maps obtained from manufacturers and suppliers include
examples of both CVTs and DCTs.
To characterize transmission and torque converter operation strategies, EPA has also
performed multiple chassis dynamometer tests of current-generation vehicles equipped with a
range of transmission technologies. The transmission gear and torque converter state (as well as
other vehicle parameters) were recorded over the FTP, HWFET, and US06 cycles. The recorded
data were used to determine the drive strategy for the engine-transmission pair in the vehicle.
The transmission losses and shifting strategy were used as modeling inputs to EPA's full-
vehicle ALPHA model.48 The shifting strategy was parameterized to allow sufficient flexibility
to maintain reasonable shift strategies while changing other vehicle attributes.49
EPA also performed a study using chassis dynamometer testing to determine effectiveness of
transmissions. In particular, two Dodge Chargers, one with a five-speed transmission and one
with an eight-speed transmission, were tested on the dynamometer. Other than the transmission,
these vehicles had identical powertrains, and so provided an ideal opportunity to test the effect of
different transmissions in the vehicle.50 Multiple repetitions of the FTP and HWFET, cycles
2-47
-------
Technology Cost, Effectiveness, and Lead Time Assessment
were run, with the result that the Charger equipped with the eight-speed transmission exhibited
on average a 6.5 percent reduction in fuel consumption over the five-speed Charger on the
combined FTP/HWFET cycle. The eight-speed Charger also exhibited an increase in
acceleration performance, according to tests by Car and Driver, with, for example, a 0.5 second
improvement in 0-60 time.51'52
2.2.3.4 Sources of GHG Emission Improvements: Reduction in Parasitic Losses, Engine
Operation, and Powertrain System Design
The design of the transmission system can affect vehicle GHG emissions in two ways. First,
reducing the energy losses within the transmission (and/or torque converter) reduces the energy
required from the engine, which also reduces GHG emissions. Reducing transmission losses can
be accomplished by increasing gearing efficiency, reducing parasitic losses, altering the torque
converter lockup strategy, or other means. A more in-depth discussion of internal energy loss
reduction is included in the "Transmission Parasitic Losses" and "Torque Converter Losses and
Lockup Strategy" sections below.
Another method to decrease GHG emissions is to design the entire powertrain system - the
engine and transmission - to keep the engine operating at the highest available efficiency for as
much time as possible. Transmissions with more available gears (or, at the extreme,
continuously variable transmissions) can maintain engine operation within a tighter window, and
thus maintain operation nearer the highest efficiency areas of the engine map. Likewise,
transmissions with a wider ratio spread can maintain engine operation nearer the highest
efficiency areas of the engine map for a wider range of vehicle speeds, in particular lowering the
engine speed at highway cruise for reduced GHG emissions.
In addition, the highest engine efficiencies for a given power output tend to be at lower
speeds, so transmission control strategies that allow very low engine speeds (i.e.,
"downspeeding") also reduce GHG emissions. Shifting strategies are discussed in the
"Transmission Shift Strategies" section below.
As a practical matter, transmissions with an increased number of gears tend also to have a
wider ratio. For example, the ZF 8HP eight-speed RWD transmission has a spread of 7.07,53 the
Aisin eight-speed FWD transmission has a spread of 7.58,54 the Mercedes 9G-TRONIC nine-
speed transmission has a ratio spread of 9.15,55 and the ZF 9HP48 nine-speed FWD transmission
has a spread of 9.8.56
The effects of additional gears and a wider ratio can be seen in Figure 2.25, which compares
engine operation of the same engine when coupled with a six-speed transmission and with an
eight-speed transmission. Compared to the six-speed transmission, the eight-speed transmission
allows the engine to operate over a narrower speed range and at lower speeds, both of which tend
to reduce GHG emissions.
2-48
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Engine Speed (RPM) Engine Speed (RPM)
(a) Six-speed (b) Eight-speed
Figure 2.24 Engine Operating Conditions for Six-Speed (Left) and Eight-Speed (Right) Automatic
Transmissions on the FTP-75 Drive Cycle57
The dominant trends in transmissions have been toward a larger number of gears and a wider
ratio spread. However, it is recognized, including by the 2015 NAS Report, that above certain
values, additional gearing and ratio spread provide minimal additional fuel economy benefits.58
59 60 Thus, increasing the number of gears (except when going to effectively infinite the case of
CVT transmissions) and ratio spread beyond that exhibited by the current market leaders is
unlikely to result in significant fuel consumption benefits, although other vehicle attributes such
as acceleration performance and shift smoothness may benefit.
In fact, it is well-understood that typical implementations of high-gear transmissions provide
both fuel consumption and acceleration performance benefits. Performance benefits come from
two factors: first, the gear ratio spread of transmissions with higher number of gears will
typically "straddle" the ratio spread of the lower number of gear transmission they replace (i.e.,
first gear is a numerically higher ratio and the final gear is a numerically lower ratio). This
provides more launch torque and quicker acceleration from stop. Second, the gear ratios of
sequential gears tend to be closer together in transmissions with a higher number of gears. This
not only narrows the on-cycle operation range of the engine for improved fuel economy (as in
Figure 2.25), but also maintains engine performance nearer the maximum power point in high
power demand situations for better acceleration performance at higher vehicle speeds.
To determine the relative cost-effectiveness of different technologies, it is important to
account for all technology benefits where possible. As the NAS point out, "objective
comparisons of the cost-effectiveness of different technologies for reducing FC can be made
only when vehicle performance remains equivalent."61 This is particularly relevant for advanced
transmissions, which do affect performance when coupled with the same engine as transmissions
with a lower number of gears. In evaluating information on measured or modeled fuel
consumption effects of advanced transmissions, it is important to consider both reported fuel
consumption benefits and any simultaneous acceleration performance benefits, so that
transmission effectiveness can be objectively and fairly estimated.
Transmission design parameters that substantially affect engine operation - gearing ratios,
ratio spread, and shift control strategy - are all used to optimize the engine operation point, and
thus the effectiveness of these transmission parameters depend in large part on the engine it is
coupled with. Advanced engines incorporate new technologies, such as variable valve timing
2-49
-------
Technology Cost, Effectiveness, and Lead Time Assessment
and lift, direct injection, and turbocharging and downsizing, which improve overall fuel
consumption and broaden the area of high-efficiency operation. With these more advanced
engines, the benefits of increasing the number of transmission gears (or using a continually
variable transmission) diminish as the efficiency remains relatively constant over a wider area of
engine operation. For example, the NAS estimated that the benefit of an eight-speed
transmission over a six-speed transmission is reduced by approximately 15 percent when added
to a modestly turbocharged, downsized engine instead of a naturally aspirated engine.62 Thus,
the effectiveness of transmission speeds, ratio, and shifting strategy should not be considered as
an independent technology, but rather as part of a complete powertrain.
Additionally, because the engine and transmission are paired in the powertrain, the most
effective design for the engine-transmission pair is where the entire powertrain is running at the
highest combined efficiency. This most effective point may not be at the highest engine
efficiency, because a slightly different operation point may have higher transmission efficiency,
leading to the best combined efficiency of the entire powertrain.
2.2.3.5 Automatic Transmissions (ATs)
Conventional planetary automatic transmissions remain the most numerous type of
transmission in the light duty fleet. These transmissions will typically contain at least three or
four planetary gear sets, which are connected to provide the various gear ratios. Gear ratios are
selected by activating solenoids which engage or release multiple clutches and brakes. A
cutaway of a modern RWD transmission (in this case the ZF 8HP70) is shown in Figure 2.25.
Figure 2.25 ZF 8HP70 Automatic Transmission63
Automatic transmissions are packaged with torque converters which provide a fluid coupling
between the engine and the driveline, and provide a significant increase in launch torque. When
transmitting torque through this fluid coupling, energy is lost due to the churning fluid. These
losses can be eliminated by engaging ("locking up") the torque converter clutch to directly
2-50
-------
Technology Cost, Effectiveness, and Lead Time Assessment
connect the engine and transmission. A discussion of torque converter lockup is continued in the
"Torque Converter Losses and Lockup Strategy" section below.
In general, ATs with a greater number of forward gears (and the complementary larger ratio
spread) offer more potential for CO2 emission reduction, but at the expense of higher control
complexity. Transmissions with a higher number of gears offer a wider speed ratio and more
opportunity to operate the engine near its most efficient point (as shown in the previous section).
In the past few years, manufacturers have taken advantage of this fact. Four- and five-speed
automatic transmissions, which dominated the market in 2005, have substantially declined in
number, being replaced by six-speed and higher transmissions (see Figure 2.22 above). In fact,
the average number of AT gears in the fleet has rapidly increased, and in 2014 was above six for
both cars and trucks (see Figure 2.26 below).
it—
ra
4—
o
5
E
3
Z
0)
s?
L
d)
Figure 2.26 Average Number of Transmission Gears for New Vehicles (excluding CVTs)64
As six-speed ATs have supplanted the four-and five-speeds, seven- and eight-speed
transmissions have also appeared on the market. As we mentioned in the Draft TAR, in the
FRM, eight speed ATs were not expected to be available in any significant number until
approximately 2020. However, even as of 2014 seven- and eight-speed transmissions occupy a
significant and increasing portion of the market.
Seven-speed transmissions currently available include the RWD 7G-Tronic from Mercedes
and the JATCO JR710E available in Nissan products. RWD eight-speed transmissions available
include offerings from General Motors and Hyundai, as well as transmission suppliers Aisin and
ZF. The ZF 8HP, introduced in 2009, has been incorporated into offerings from a range of
manufacturers, including Fiat/Chrysler, Jaguar/Land Rover, and Volkswagen. ZF has begun
production of a second generation of 8HP transmissions (the 8HP50), which features a higher
ratio spread, lower drag torque, and improved torsional vibration absorption compared to the first
generation.65 Aisin also offers a FWD eight-speed used by multiple manufacturers. This
Car
Truck
5-
4-
/
Manual
L
Automatic
/
Manual
i
Automatic
i
1960 19E5 I960 1996 2DOO 2005 2010 2D15 1800 19B5 1090 1905 2000 2005 2010 2015
Model Year
2-51
-------
Technology Cost, Effectiveness, and Lead Time Assessment
includes use in the compact 2016 Mini Cooper Clubman,66 a vehicle smaller than those assumed
eligible for eight-speed transmissions in the FRM.
As mentioned in the Draft TAR, in the FRM, EPA limited its consideration of the effect of
additional gears to eight-speed transmissions. However, some ATs with more than eight gears
are already in production, and more examples are in development. At this time, nine-speed
transmissions are being manufactured by ZF67 (which produces a FWD nine-speed incorporated
into Fiat/Chrysler, Honda, and Jaguar/Land Rover vehicles68) and Mercedes69 (which produces a
RWD nine-speed). Ford has released a ten speed transmission in the F150 Raptor, and GM
released a variation of the same ten speed in the 2017 Camaro ZL1. In addition, Ford and
General Motors have announced plans to jointly design and build a nine-speed FWD
transmission, and Honda is developing a ten-speed FWD transmission.70
Manufacturers have claimed substantial fuel consumption benefits associated with newer
transmissions. ZF claims its first generation 8HP can reduce fuel consumption by 6 percent on
the NEDC compared to a circa 2005 ZF 6HP, using the same engine, along with improving
vehicle acceleration performance.71 ZF also outlined a series of potential improvements to the
first generation 8HP that could provide an additional 5 to 6 percent fuel consumption reduction
on the U.S. combined cycle.72 The second generation ZF eight-speed73 is expected to achieve up
to 3 percent efficiency gain on the NEDC due to the improvements noted above; ZF also outlined
additional potential savings associated with a third generation eight-speed transmission.74
Likewise, Mercedes clamed a 6.5 percent fuel consumption improvement on the NEDC with its
nine-speed transmission compared to the previous seven-speed.75 It should also be noted that the
percent fuel consumption reported on the NEDC drive cycle will be different from the U.S.
combined cycles.
In FWD vehicles, ZF claims its nine-speed FWD transmission reduces fuel consumption by
10 percent - 16 percent compared to an early- 2000s six-speed transmission.76 Aisin claims its
new FWD eight-speed transmission decreases fuel consumption 16.5 percent compared to an
early generation six-speed, and nearly 10 percent compared to the previous generation six-
speed.77 In addition, the new eight-speed improves acceleration performance. BMW, using the
Aisin FWD transmission, reports a 14 percent fuel consumption reduction on the NEDC over the
previous six-speed transmission.78
These efficiency improvements are due to a range of design changes in the transmissions. In
addition to improving the engine operation efficiency through changing the number of gears,
overall ratio, and shift points, these transmissions also reduce parasitic losses, change torque
converter behavior, and/or shift to neutral during idle. Mercedes claims a total of 6.5 percent
fuel economy improvement on the NEDC by using its nine-speed 9G-TRONIC in place of the
earlier generation seven-speed.79 Of this, 2 percent is due to the change in the number of gears,
ratio spread, and shift strategy, with the remainder due to transmission efficiency improvements.
With the positive consumer acceptance, higher effectiveness, and increasing production of
transmissions with up to ten forward gears, it may be possible that transmissions with even more
gears will be designed and built before 2025. Researchers from General Motors have authored a
study showing that there is some benefit to be gained from transmissions containing up to 10
speeds.80 However this appears to be near the limit for improved fuel consumption, and studies
have shown that there is no added potential for reduction in CO2 emissions beyond nine or ten
gears.8182 In fact, ZF CEO Stefan Sommer has stated that ZF would not design transmissions
2-52
-------
Technology Cost, Effectiveness, and Lead Time Assessment
with more than nine gears: "We came to a limit where we couldn't gain any higher ratios. So the
increase in fuel efficiency is very limited and almost eaten up by adding some weight and
friction and even size of the transmission."83 Although manufacturers may continue to add gears
in response to consumer preference for other performance attributes, at this time we are not
projecting that further increases will provide CO2 emissions benefits beyond that of optimized
eight, nine or ten-speeds.
2.2.3.6Manual Transmissions (MTs)
In a manual transmission, gear pairs along an output shaft and parallel lay shaft are always
engaged. Gears are selected via a shift lever, operated by the driver. The lever operates
synchronizers, which speed match the output shaft and the selected gear before engaging the gear
with the shaft. During shifting operations (and during idle) a clutch between the engine and
transmission is disengaged to decouple engine output from the transmission.
Manual transmissions are in general lighter, cheaper to manufacture, and have lower parasitic
losses than automatic transmissions. The 2015 NAS report found the overall energy loss in a
manual transmission to be only about 4 percent, as compared to a 13 percent loss in automatic
transmissions.84
As with ATs, the average number of gears in MTs has increased (Figure 2.26), albeit at a
reduced rate compared to ATs. As in ATs, the higher number of gears and associated increase in
ratio spread increases potential fuel savings.
However, manual transmissions have only a small market share, estimated at only 2.6 percent
in MY2015 based on the data in the MY2015 GHG baseline. Automatic transmissions (ATs,
CVTs, and DCTs) are more popular at least in part because customers prefer not to manually
select gears.
2.2.3.7 Dual Clutch Transmissions (DCTs)
Dual clutch transmissions are similar in their basic construction to manual transmissions, but
use two coaxial input shafts with two clutches to shift between the two shafts. By
simultaneously opening one clutch and closing the other, the DCT "hands off power from one
shaft to the other, and thus to sequential gears. Unlike the MT, the DCT selects the appropriate
gear automatically (as in an AT). DCTs offer an efficiency advantage over a typical automatic
because their parasitic losses are significantly lower. In addition, DCTs in general do not require
a torque converter, as gradually engaging the clutch (much like with a manual transmission)
provides the application of launch torque.
2-53
-------
Technology Cost, Effectiveness, and Lead Time Assessment
hollow shaft
i
dual-clutch
crankshaft
trans miss ion-
uutpul
transmission
Figure 2.27 Generic Dual Clutch Transmission85
Multiple DCTs have been introduced into the marketplace, primarily in six- and seven-speed
versions. Volkswagen has used multiple generations of DCTs in their products. Ford has used
six-speed DCTs jointly developed with Getrag. Fiat has another version of a six-speed DCT,
while both Honda and Hyundai have developed seven-speed versions. Honda introduced an
eight-speed DCT with a torque converter on the 2015 Acura TLX.86
As mentioned in the Draft TAR, DCTs have encountered issues with customer acceptance,
and, as the NAS stated in its 2015 report, "are not likely to reach the high penetration rates
predicted by EPA primarily due to customer acceptance issues."87 As noted by the NAS in their
2015 report, "This difference in drivability and consumer acceptance [between wet and dry
clutch DCTs] can be seen in the comparison of two of Volkswagen's MY2015 vehicles, the VW
Golf and the VW Polo. The Golf, with a wet-clutch DCT, has received many positive reviews
and awards, while the Polo, with a dry-clutch DCT, has received poor reviews for transmission-
related drivability."88
Getrag announced the 7DCT300 which has a wet clutch with lubrication on demand (we refer
to these as damp clutch DCTs), equaling the efficiency of a dry DCT. The "damp" clutch is also
smaller and has a higher tolerance for engine irregularities.89 Wet/damp clutch DCTs tend to
have better consumer acceptance than dry clutch DCTs. The 7DCT300 is available in Europe on
the 2015 Renault Espace. Honda recently patented an 11-speed triple clutch transmission.
As in ATs, it is expected that additional gears above the current maximum will not
significantly decrease fuel consumption and resulting GHG emissions. A 2012 study by DCT
manufacturer Getrag indicated that additional gears above seven and additional ratio spread
above 8.5 provided minimal additional fuel economy benefits.90
2.2.3.8 Continuously Variable Transmissions (CVTs)
Conventional continuously variable transmissions consist of two cone-shaped pulleys,
connected with a belt or chain. Moving the pulley halves allows the belt to ride inward or
outward radially on each pulley, effectively changing the speed ratio between the pulleys. This
2-54
-------
Technology Cost, Effectiveness, and Lead Time Assessment
ratio change is smooth and continuous, unlike the step changes of other transmission varieties.
CVTs were not chosen in the fleet modeling for the 2017-2025MY FRM analysis because of the
predicted a low effectiveness associated with CVTs (due to the high internal losses and narrow
ratio spans of CVTs in the fleet at that time). However, improvements in CVTs in the current
fleet have increased their effectiveness, leading to rapid adoption rates in the fleet. In their 2015
report, the NAS recommended CVTs be added to the list of considered technologies, and EPA
did indeed add re-evaluate the costs and effectiveness for this technology for its Draft TAR
analysis and is continuing to consider CVTs in this Proposed Determination analysis
(a) (b)
Figure 2.28 (a) Toyota CVT91 (b) Generic CVT sketch92
One advantage of CVTs is that they continue to transmit torque during ratio changes. During
a ratio change or shift the energy from the engine is wasted on ATs and some DCTs. ATs and
some DCT have a hesitation during shifts caused by the torque disruption during gear changes
This shift feeling is well known to consumers and in some cases comforting to drivers (they miss
it when driving a vehicle with a CVT). As mentioned in the AT section ATs efficiency peaks
with 9 to 10 gears, while going to a CVT (with an effectively "infinite" number of gear steps)
adds a new level of efficiency to the overall system. This is in part due to the fact that CVTs do
not need to stop transmitting torque to change ratios.
2-55
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Another advantage of a CVT is that, within its ratio range, it can maintain engine operation
close to the maximum efficiency for the required power. However, CVTs were not considered in
the FRM because at the time CVTs had a ratio range of near 4.0, limiting the range where the
engine operation could be optimized. In addition, the CVTs were less than 80 percent efficient
93, and thus required more total output energy from the engine. These limitations overwhelmed
the CVT's inherent advantage compared to conventional ATs.
However, in the recent past, manufacturers and suppliers have intensified development of
CVTs, reducing the parasitic losses and increasing the ratio spread. The current generation of
CVT are now nearly 85 percent efficient, with ongoing work by suppliers to push that number to
90 percent.94 Ratio spreads for new CVTs from Honda, Toyota, and JATCO now range between
6.0 and 7.0.95>96'97 JATCO has introduced a very small CVT what has a two speed output with
take a CVT with a small ratio spread and doubles it for an overall ratio spread of 7.398 in the base
version and 8.7 in the "wide range" version.99 As in ATs and DCTs, it is expected that additional
increase in ratio range above the current maximum will not significantly decrease fuel
consumption and resulting GHG emissions.100
Reducing losses in CVTs has been a particular focus of manufacturers. The JATCO CVT8
featured a 40 percent reduction in mechanical losses compared to their earlier generation CVT.101
The losses were reduced by decreasing the size of the oil pump, implementing a new, higher
efficiency belt, and reducing the fluid churning losses. Honda's new compact car CVT increased
efficiency 1.0 percent to 1.5 percent at higher vehicle speeds compared to their previous
generation CVT.102 The increased efficiency was primarily due to a reduction in oil pump losses
and bearing friction. Honda's new midsize CVT increased efficiency up to 5 percent compared to
the earlier generation CVT, primarily by reducing the required hydraulic pressure (by up to 38
percent).103 Toyota's new K114 CVT reduced torque losses by 22 percent, compared to the
earlier generation of CVTs, primarily by reducing the losses associated with the oil pump, and
reducing the size of the bearings.104
The decreased transmission losses (5-10 percent) and increased ratio spread (from 4 to
between 6 and 8.7) of CVTs has made them more effective in CO2 reduction than estimated in
the FRM, and thus CVTs are anticipated to be used in an increasing share of the fleet (see Figure
2.22). The supplier JATCO supplies CVTs to Nissan, Chrysler, GM, Mitsubishi, and Suzuki 105
In addition, other manufacturers' - Audi, Honda, Hyundai, Subaru, and Toyota - all make their
own CVTs.
The JATCO CVT8 demonstrated a 10 percent improvement in fuel economy for both the
highway and city cycles compared to earlier generation CVTs.106 Honda's new compact car
CVT increased fuel economy approximately 7 percent compared to the earlier generation CVT
over both the U.S. test cycle and the Japanese JC08 test cycle.107 Honda's new midsize CVT
increased fuel economy 10 percent over the earlier generation 5AT on the U.S. cycle, and 5
percent compared to the earlier generation CVT on the Japanese JC08 test cycle.108 Toyota's
new K114 CVT increased fuel economy by 17 percent on the Japanese JC08 test cycle compared
to the earlier generation CVT.109
Some initial introductions of CVTs suffered from consumer acceptance issues, where
customers complained of the "rubber band" feel of the transmission, due to the indirect
connection between the driver's throttle input and the vehicle's acceleration response. To
combat this perception, vehicle manufacturers have added a shift feel calibration to the CVT
2-56
-------
Technology Cost, Effectiveness, and Lead Time Assessment
control strategy, which mimics the feel of a conventional AT.110 This calibration, although
having a slight effect on fuel economy, has improved consumer acceptance.111
In this document, only conventional belt or chain CVTs are considered. At least two other
technologies - toroidal CVTs and Dana's VariGlide® technology112 - are under development and
may be available in the 2020-2025 time frame. The Dana VariGlide is considered a CVP
(Continuously Variable Planetary) with the major design difference being it using balls to
transmit torque and vary the ratio. Dana has stated that it is currently in development with an
OEM. Targeted production could be as early as 2020. These technologies hold promise for
increased efficiency compared to current design belt or chain CVTs.
2.2.3.9 Transmission Parasitic Losses
Reducing parasitic loses in the transmission improves drivetrain efficiency and lowers the
required energy output from the engine. In general, parasitic losses can come from (a) the oil
supply, (b) electricity requirements, (c) drag torque, (d) gearing efficiency, and (e) creep (idle)
torque.113
2.2.3.9.1 Losses in ATs
A study by ZF suggests that the largest sources of losses over the combined city/highway
cycle in conventional automatic transmissions are the oil supply and the drag torque.114 This is
followed by the creep torque (on the city cycle), with the electrical requirements and gearing
efficiency being relatively minor.
For conventional ATs, power required to supply oil to the transmission is one of the largest
sources of parasitic loss. An oil pump is required for lubrication and for hydraulic pressure for
clamping the clutches. A baseline transmission would typically use a gerotor-type pump driven
off the torque converter. Replacing or resizing the oil pump can result in a substantial decrease in
torque losses. For example, Aisin claims a 33 percent reduction in torque loss in its new
generation transmission from optimizing the oil pump,115 and Mercedes claims a 2.7 percent
increase in fuel economy on the NEDC by changing the pumping system.116 Pump-related losses
can be reduced by substituting a more efficient vane pump for the gerotor. Losses can be further
reduced with a variable-displacement vane pump, and by reducing the pressure of the system.
Losses can be further decreased by using an on-demand electric pump: Mercedes claims an
additional 0.8 percent increase in fuel economy on the NEDC by implementing a lubrication on
demand system.117 Another way to reduce losses from the pump is by reducing leakage in the
system. Reducing leakage reduces parasitic losses by reducing the amount of fluid that needs to
be pumped through the system to maintain the needed pressure.
A second large source of parasitic loss in ATs is the drag torque in the transmission from the
clutches, brakes, bearings, and seals. These components have the potential to be redesigned for
lower frictional losses. New clutch designs offer potential reductions in clutch drag, promising
up to a 90 percent reduction in drag.118 Replacing bearings can reduce the associated friction by
50 to 75 percent. New low-friction seals for can reduce friction by 50 percent to provide an
overall reduction in bearing friction loss of approximately 10 percent.119
2-57
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Optimizing shift elements improved fuel economy on the Mercedes 9G-TRONIC by 1 percent
over the NEDC.120
Drag torque can be further reduced by decreasing the viscosity of the automatic transmission
fluid used to lubricate the transmission. A study of transmission losses indicates that about a 2
percent fuel consumption reduction was obtained on the FTP 75 cycle by switching to the lowest
viscosity oil.121 However, reduction of transmission fluid viscosity may have an adverse effect
on long-term reliability.
Transmission efficiency may also be improved through superfinishing the gear teeth to
improve meshing efficiency.
2.2.3.9.2 Losses in DCTs
Advanced DCTs typically have lower losses than ATs, largely due to having an on-demand
pump, splash lubrication, and fewer open clutches. The primary losses in DCTs are load-
independent drag and splash losses. Unlike ATs, DCTs typically depend on splash lubrication for
their internal components rather than forced lubrication. This eliminates the losses associated
with oil supply pumps, but adds churning losses due to rotating components moving through the
oil. Churning losses can be minimized by keeping oil levels low and warming up the lubrication
oil.
A primary consideration in DCT losses is the use of wet or dry clutches.122 Dry clutches do
not require oil cooling flow, and therefore do not contribute to oil churning losses that are
incurred with wet clutch systems; this has traditionally meant that dry clutch reduced GHG
emissions by an additional 0.5 to 1 percent over wet clutch DCTs. However, dry clutches have a
limited maximum torque capacity, and have suffered from customer acceptance issues. In
response, so-called "damp" clutches have been introduced, where on-demand cooling flow has
substantially reduced the parasitic losses associated with wet clutches.
DCTs also may benefit from the same improvements in bearing and seal drag and gear
finishing that are outlined in the AT section above.
2.2.3.9.3 Losses in CVTs
CVTs tend to have higher losses than either ATs or DCTs, in large part due to the high oil
pressures required to keep the belt and pulleys securely clamped. These losses increase
significantly at high input torques, as even higher pressures are required to maintain the
clamping force.123
A study by JATCO suggests that losses in the CVT are dominated by oil pump torque and
losses in the belt-pulley system, with fluid churning losses as the next largest player.124 By
reducing leakage in the oil system and reducing line pressure when possible, JATCO's CVT8
was able to run with a reduced size oil pump and considerable reduction in oil pump torque loss.
JATCO also redesigned the belt for lower loss, and reduced the oil level and viscosity to reduce
churning losses. The overall result was a 40 percent reduction in mechanical losses compared to
the earlier generation CVT.
Honda developed a new CVT using a comparable strategy.125 They decreased the required
pulley thrust by refining the control strategy and by using a fluid with increased coefficient of
2-58
-------
Technology Cost, Effectiveness, and Lead Time Assessment
friction, which combined for a transmission efficiency increase of 2.8 percent. They also altered
the belt trajectory around the pulley for an added 0.4 percent efficiency increase.
Another opportunity for reduced losses in CVTs is Dana's VariGlide System. Dana's
VariGlide system can provide more favorable system losses than traditional belt or chain
technologies. The VariGlide system eliminates the requirement for a high pressure pump, using
instead a fully passive mechanical clamping mechanism. The unique coaxial configuration,
similar to a planetary gearset coupled with high power density, allows for simple integration into
traditional transmission architectures and makes it uniquely suited for RWD applications.
2.2.3.9.4 Neutral Idle Decoupling
An additional technology that has been implemented in some transmissions, which was not
considered in the FRM, is the application of a "neutral idle." In this strategy, a neutral clutch is
opened when the vehicle is at a stop, which effectively reduces the creep torque required from
the engine.126'127 BMW demonstrated a reduction in fuel consumption of 2 - 3 percent on the
NEDC for an optimized neutral idle decoupling system on an eight-speed transmission.128
Similarly, ZF calculated that implementing a neutral idle decoupling system on its eight-speed
transmission would reduce fuel consumption by 0.5 percent to 1.4 percent on the U.S. combined
cycle, depending on the K-factor of the torque converter.129 It should be noted, of course, that
the neutral idle decoupling simply reduces idling losses, and implementing stop-start system
would eliminate the effectiveness of this technology.
2.2.3.10 Transmission Shift Strategies
The transmission shift schedule can strongly influence the fuel consumption over a drive
cycle. A more aggressive shift schedule will downshift the transmission earlier and upshift later
(i.e., at lower engine speeds). This moves engine operation, for a particular required power, to
lower speeds and higher torques where engine efficiency tends to be higher. Along with this,
reducing time between shifts (i.e., allowing more shifts), reducing the minimum gear where fuel
cutoff is used, and altering torque converter slip (covered in the next section) will also decrease
fuel consumption. Applying an aggressive shift strategy can reduce fuel consumption by about 5
percent in a generic six-speed transmission or 1-3 percent in a generic nine-speed
transmission.130 Similarly, BMW showed about a 2 percent reduction in CO2 from
downspeeding the engine, comparing their current generation six-speed transmission to an earlier
1 "3 1
generation.
However, the application of the strategy is limited by NVH and drivability concerns, as lower
engine speeds produce more significant driveline pulses and allowing more shifts may increase a
shift busyness perception. Manufacturers reduce the NVH impact by using allowing partial
lockup, adding a torque converter dampener, and/or adding a pendulum dampener. These
changes along with decreasing the ratio between gears has made higher gear numbers and
increased shifting more acceptable. Reducing the ratio between gears allows shifting to be less
perceptible due to the smaller change in engine speed.
2.2.3.11 Torque Converter Losses and Lockup Strategy
Torque converters are typically associated with conventional ATs and CVTs, although they
have appeared on Honda's eight-speed DCT. Torque converters provide increased torque to the
wheels at launch, and serve as a torsional vibration damper at low engine speeds. However, this
2-59
-------
Technology Cost, Effectiveness, and Lead Time Assessment
comes at the cost of energy loss in the torque converter fluid, and modern torque converters
typically have a lockup clutch that mechanically locks the impeller and turbine together,
bypassing the fluid coupling.
Figure 2.29 ZF Torque Converter Cutaway132
Although in the past torque converters remained unlocked up to high vehicle speeds, recent
trends are to lock up at much lower speeds. Improvements in torsional vibration dampers, and
the ability to utilize micro-slip across the lockup clutch has enabled lower lockup speeds.
Mazda, for example, claims torque converter lockup as low as 5 mph for its SKYACTIV-Drive
AT.133 Although not as aggressive, BMW claims a 1 percent reduction in CO2 from an early
torque converter lockup.134
2.2.4 Electrification: State of Technology
Electrification includes a large set of technologies that share the common element of using
electrical power for certain vehicle functions that were traditionally powered mechanically by
engine power. Electrification can thus range from electrification of specific accessories (for
example, electric power steering) to electrification of the entire powertrain (as in the case of a
battery electric vehicle). Powering accessories electrically can reduce their energy use by
allowing them to operate on demand rather than being continuously driven by the crankshaft
belt. Some electrical components may also operate more efficiently when powered electrically
than when driven at the variable speed of a crankshaft belt. Electrified vehicles that use
electrical energy from the grid also provide a means for low-GHG renewable energy to act as a
2-60
-------
Technology Cost, Effectiveness, and Lead Time Assessment
transportation energy source where it is present in the utility mix. The addition of a larger
capacity battery in a vehicle also provides for energy recovery or recuperation. Kinetic energy
can be used to charge the battery and that recovered energy can be used to power accessories or
to provide propulsion.
Electrified vehicles (or xEVs) are considered for this analysis to mean vehicles with a fully or
partly electrified powertrain. This includes several electrified vehicle categories, including:
battery electric vehicles (BEVs), which have an all-electric powertrain and use only batteries for
propulsion energy; plug-in hybrid electric vehicles (PHEVs), which have a primarily electric
powertrain and use a combination of batteries and an engine for propulsion energy; and hybrid
electric vehicles (HEVs), which use electrical components and a battery to manage power flows
and assist the engine for improved efficiency and/or performance. HEVs are further divided into
strong hybrids (including P2 and power-split hybrids) that provide strong electrical assist and in
many cases can support a limited amount of all-electric propulsion, and mild hybrids (such as
belt integrated starter generator (BISG) hybrids, crankshaft integrated starter generator (CISG)
hybrids, and 48 V mild hybrids) that typically provide only engine on/off with minimum
electrical assist. BEVs and PHEVs are herein referred to collectively as plug-in electric vehicles,
or PEVs.
Fuel cell electric vehicles (FCEVs) are another form of electrified vehicle having a fully
electric powertrain, and are distinguished by the use of a fuel cell system rather than grid power
as the primary energy source. FCEVs have only recently entered commercial production and
their market has not yet developed as much as that of PEVs. Technology developments relating
to FCEVs were reviewed in detail in Draft TAR Chapter 5.2.4.5. Because EPA did not include
FCEVs in its fleet compliance modeling analysis for the Draft TAR nor for the Proposed
Determination, please refer to the Draft TAR for additional information on this technology.
As with the other technologies presented in this chapter, EPA has reviewed, and revised
where necessary, the assumptions for effectiveness and cost of electrification technologies for
this Proposed Determination. This effort extends the effort carried out for the Draft TAR, which
included inquiries along several paths. As discussed in the Draft TAR, EPA gathered
information from many sources, including public sources such as journals, press reports, and
technical conferences, as well as manufacturer certification data and information gathered
through stakeholder meetings with OEMs and suppliers. EPA has also benchmarked selected
vehicles by means of dynamometer testing at the EPA National Vehicle and Fuel Emissions
Laboratory (NVFEL), as well as utilized instrumented vehicle test data from the Argonne
National Laboratory (ANL) Advanced Powertrain Research Facility (APRF). Among other
purposes, EPA has used this data to inform development of the ALPHA model. EPA also
utilized electric machine component performance data collected by Oak Ridge National
Laboratory (ORNL) under U.S. DOE funding, and similar component and vehicle test data
provided by other laboratories such as Idaho National Laboratory (INL). EPA also worked
closely with ANL to improve and update the battery costing model, known as BatPaC,135 which
was used to update the projected costs of electrified vehicle battery packs. All of these sources
have contributed to our assessment of the progress of electrification technology, an assessment
that has continued since the 2012 FRM and before.
2.2.4.1 Overview of Chapter
2-61
-------
Technology Cost, Effectiveness, and Lead Time Assessment
This Chapter 2.2.4 is intended to review the current state of electrification technology as
represented by developments since the 2012 FRM to the present, including updates since the
Draft TAR that could inform the Proposed Determination assessment. The information
described in this section thus forms the basis for revised cost and effectiveness assumptions
described in Chapter 2.3.4.3, which become inputs to the Proposed Determination analysis.
Source data for many of the charts in this Chapter and Chapter 2.3.4.3 are available in the
Docket.136
This Chapter 2.2.4 is organized in the following way:
Chapter 2.2.4.2 provides a high-level overview of the major developments in electrification
technologies since the 2012 FRM. This section is intended only as an executive summary to
help place the topic of electrification into context.
Chapter 2.2.4.3 provides a background in non-battery electrical components that are common
to many of the electrification technologies, and briefly reviews the major directions of their
development since the 2012 FRM. An understanding of these components is helpful to
understanding developments in cost and effectiveness of each of the electrified vehicle
categories. Developments in the cost or performance of specific classes of components are
discussed in the context of the electrified vehicles in which they have been implemented.
Chapter 2.2.4.4 includes subsections detailing each of the major electrified vehicle categories
(stop-start, mild/48V and strong HEVs, PHEVs and BEVs). These subsections serve to briefly
review the significance of each electrified vehicle category as a means of reducing GHG
emissions, and review industry developments relating to how the category has evolved and been
taken up in the fleet since the 2012 FRM.
Chapter 2.2.4.5 focuses on developments in battery technology. Batteries are discussed
separately and after discussion of the electrified vehicle categories for several reasons. First, the
battery performance requirements for each of the categories is best understood after the
categories have been fully defined and discussed. Second, a greater level of technical detail is
required to adequately assess some battery developments that have a strong influence on
effectiveness or cost of xEV technologies. Finally, and perhaps most importantly, battery cost
estimation is a particularly influential input to the cost assumptions for xEVs, and the battery
cost estimates for different xEV categories rely on many detailed parameters that are best
understood and contrasted in the context of a battery discussion after trends in xEVs have been
reviewed. The bulk of battery-related developments are therefore covered in the battery chapter
rather than the electrified vehicle category subsections.
Chapter 2.2.4.6 acknowledges developments in FCEVs, and refers the reader back to the more
complete analysis of this technology that was published as Chapter 5.2.4.5 of the Draft TAR.
Because EPA did not include FCEVs in its fleet compliance modeling analysis for the Draft
TAR nor for the Proposed Determination, the assessment of FCEV technology is not repeated in
this TSD.
Although these chapters may in some places refer to comments received on the Draft TAR,
comments relating to electrification are primarily discussed in the context of specific modeling
assumptions and inputs in Chapter 2.3.4.3.
2.2.4.2 Overview of Electrification Technologies
2-62
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Throughout the 2012 rule analysis, and the Draft TAR analysis, electrified vehicles have been
identified as offering a strong potential for reducing greenhouse-gas emissions. In all of these
analyses, the cost-minimizing compliance pathway showed electrified vehicles playing an
important supporting role in a fleet composed primarily of non-electrified powertrain
configurations. For example, the pathway presented by EPA in the Draft TAR showed OEM
compliance with MY2025 GHG standards with fleet penetrations of less than 3 percent BEVs, 3
percent strong hybrids, and 18 percent mild hybrids.137
In the years since the final rulemaking, the number of HEV, PHEV, and BEV models
available to consumers has continued to grow. HEVs are now part of the product line of almost
every major OEM. In 2014, U.S. HEV sales were in excess of 450,000 units. This declined to
about 385,000 units in 2015.138 Through September 2016, U.S. HEV sales are at approximately
260,000 units, which would represent a drop of about 13 percent compared to the same point in
20 1 5.139 Plug-in vehicles (BEVs and PHEVs) are also being offered in increasing numbers. In
MY2015, 28 models of plug-in vehicles were available, an increase from 23 models in MY2014,
and only a handful in 2012. In each of 2014 and 2015, U.S. plug-in vehicle sales were in excess
of 115,000 units,138 and through September 2016 are already at about 110,000 units.
Also in 2015 and 2016, a growing number of manufacturers announced ambitious plans to
introduce multiple lines of plug-in vehicles by 2020-2025, including Volkswagen (planning more
than 30 new all-electric vehicles with annual sales of 2-3 million units, or 20-25 percent of total
sales, by 2025), 140>141 Mercedes-Benz (all models to be electrified in a similar time frame),142
BMW (plug-in hybrid versions of all of its core models),143 Volvo (battery electric power on all
vehicles within the next decade),144 and Ford (13 new BEV nameplates and 40 percent
electrification by 2020).145 In November 2016, it was reported146'147 that even Toyota, which had
previously concentrated primarily on fuel-cell and hybrid technology, is planning to add BEVs to
its lineup by 2020.
In the Draft TAR, it was noted that some aspects of BEV implementation and penetration
have developed differently than originally predicted in the 2012 FRM. At that time the agencies
expected that BEVs with a range between 75 and 150 miles would be most likely to play a
significant part in OEM compliance. By the time of the Draft TAR it was clear that the BEV
market had developed two distinct segments, a consumer segment offering a driving range of
around 100 miles at a relatively affordable price, and a premium segment offering a much higher
range (well in excess of 200 miles) at a higher price. Tesla Motors has had notable success at
producing and marketing BEVs in the premium segment, causing significant numbers of long-
range BEVs to enter the fleet that may not have been predicted by OMEGA on a pure cost-
effectiveness basis. Going forward, both BEV segments appear to be aggressively pursuing range
increases in their second and third generation models. In 2016 GM announced the 2017 Chevy
Bolt, which has been EPA certified with a 238-mile range. Nissan has also announced plans to
offer a 200-mile range BEV in 2017 or 2018, using a newly developed battery pack. Tesla is also
making progress toward a long-stated intention to enter the consumer segment with the Model 3,
which is targeted for introduction in late 2017 and is expected to offer a range of at least 215-
miles.
An increasing number of OEMs are beginning to add PHEVs to their product lines, utilizing
both blended-operation architectures as well as extended-range architectures that offer varying
amounts of all-electric range. The cost-minimizing pathway presented in the Draft TAR for
2-63
-------
Technology Cost, Effectiveness, and Lead Time Assessment
compliance with the 2025MY GHG standards projected less than 2 percent fleet-level
penetration of PHEVs.148 The 2015 and 2016 MYs saw a discernible increase in PHEV20-style
architectures from OEMs that tend to specialize in luxury or high-performance vehicles, which
was consistent with projections in the 2012 FRM.149 Second-generation PHEV models have
begun to appear, typically offering an increased all-electric range or a more robust blended-mode
operation that allows for increased all-electric capabilities in normal driving. Manufacturers
have often cited customer demand for a more all-electric driving experience in making these
changes.
Charging infrastructure is also growing. While PEVs are manufactured with onboard chargers
that can often take advantage of existing 110V or 220V charging connections in the home or
garage, opportunities for public charging away from the home are poised to become much more
common. Since 2008, various ongoing public and private efforts to provide charging stations at
workplaces, along freeway corridors, and in cities have grown the number of public stations in
the U.S. to more than 16,000.150 Since the Draft TAR was completed, two developments were
announced that may increase this number substantially. The partial settlement between
Volkswagen and U.S. authorities, approved in 2016, earmarks $1.2 billion in investment over 10
years toward ZEV infrastructure, education, and access.151 Also, in November 2016 The White
House announced a network of federal, state, and local initiatives to increase accessibility to
PEV infrastructure,150 including a Department of Transportation (DOT) plan to designate 48
national "alternative fuel corridors" along major highways to provide focus for build out of
charging locations by related local and state efforts.152 Public charging infrastructure was
explored in depth in Draft TAR Chapter 9 (Infrastructure Assessment), and is reviewed for this
Proposed Determination assessment in Section B.3.2 of the Proposed Determination Appendix.
Advancements in the cost and effectiveness of xEVs are closely related to advancements in
battery, electric motor, and power electronics technologies. These technologies have advanced
steadily since the 2012 FRM, with significant improvements in battery specific energy, battery
cost, and non-battery component efficiency and cost contributing to improvements in production
xEVs. The pace of industry activity in this area suggests that further advancements are likely to
occur between now and the 2022 to 2025 time frame of the rule.
At the time of the 2012 FRM, data regarding the cost and efficiency of xEV components was
limited by the small number of production vehicles from which it could be gathered. Today, the
relatively large number of production models provides much greater opportunity to empirically
validate projections made in the FRM.
Battery cost is a major consideration in the cost of xEVs. At the time of the 2012 FRM there
was great uncertainty in the potential for battery manufacturing costs to be reduced. There was
also uncertainty regarding battery lifetime. Today, evidence of the need for battery replacement
is rare, with most PHEV and BEV batteries showing good durability within the limits established
by OEM warranties. Although the battery cost projections published in the 2012 FRM were
significantly lower than estimates of prevailing costs at the time, and those presented in the Draft
TAR were even lower, evidence continues to suggest that these estimates were conservative,
with at least one major manufacturer having announced battery costs from a major battery
supplier that are very close to the Draft TAR projections. Recent reports have suggested that
lithium-ion battery cost has historically followed a pace of improvement of about 6 to 8 percent
per year.153 Advancements in cost and energy capacity of battery technology continue to be
2-64
-------
Technology Cost, Effectiveness, and Lead Time Assessment
pursued actively by OEMs and suppliers alike, suggesting that there is room for further
improvement within the 2022-2025 time frame of the rule. Projected battery costs were
accordingly updated for the Draft TAR and are now being further updated for the Proposed
Determination based on public comment and updated information gathered since the Draft TAR.
The Draft TAR presented an analysis of current and past production BEVs and PHEVs that
showed that the 2012 FRM analysis assigned a significantly larger battery capacity per unit
driving range than manufacturers ultimately found necessary to provide. The Draft TAR found
that this was likely related to the chosen assumptions for parameters such as powertrain
efficiencies, usable battery capacity, and application of road load reducing technologies. The
Draft TAR analysis also showed that the industry achieved comparable acceleration performance
with significantly lower motor power ratings than the 2012 FRM analysis anticipated. In other
words, it was shown that in many ways the industry had found ways to do more with less,
compared to many of the original predictions of the 2012 FRM analysis. The Draft TAR analysis
incorporated these developments in its revised projections of battery cost.
Because the vehicle architecture for electrified vehicles is fundamentally different from that
of conventionally-powered vehicles, the consumer experience is likely to be different as well. In
particular, the fueling requirements of BEVs and PHEVs call for changes in accustomed fueling
habits, some of which may improve convenience (e.g. the ability to charge at home) while others
may pose a challenge (e.g. a relatively long fueling time). A BEV with limited range might not
provide an exact substitute for a conventional vehicle for many consumers today, while at the
same time electrified vehicles can provide benefits of quiet operation, reduced maintenance, and
the potential integration with future mobility systems that might include shared and autonomous
vehicles.
The primary factors that influence the cost and effectiveness of electrification technologies
are the cost and efficiency of their components. These include: energy storage components such
as battery packs; propulsion components such as electric motors; and power electronics
components, such as inverters and controllers, that process and route electric power between the
energy storage and propulsion components. For the purpose of this analysis, these components
are divided into battery components and non-battery components.
Battery components have a particularly strong influence on cost of xEVs. Because
developments in battery technology may apply to more than one category of xEV, they are
discussed collectively in Chapter 2.2.4.5. That chapter details developments in battery-related
topics that directly affect the specification and costing of batteries for all xEVs, such as usable
capacity, durability, thermal management, and pack topology, among others.
Non-battery components have a strong influence on both cost and effectiveness of xEVs.
Because non-battery technologies are important to understanding the differences in architecture
among xEVs, they are introduced prior to discussion of the individual electrified vehicle
categories in Chapter 2.2.4.3.
2.2.4.3 Non-Battery Components of Electrified Vehicles
Non-battery components largely consist of propulsion components and power electronics.
Propulsion components typically include one or more electric machines (an umbrella term that
includes what are commonly known as motors, generators, and motor/generators). Depending
2-65
-------
Technology Cost, Effectiveness, and Lead Time Assessment
on how they are employed in the design of a vehicle, electric machines commonly act as motors
to provide propulsion, and/or act as generators to enable regenerative braking and conversion of
mechanical energy to electrical energy for storage in the battery. Power electronics refers to the
various components necessary to route current between the battery system and the propulsion
components, including such devices as inverters and rectifiers, DC-to-DC converters, motor
controllers, and on-board battery chargers.
The energy efficiency of non-battery components is a continuing focus of industry research
and development. The impact of resulting improvements in efficiency and overall system
optimization therefore have been considered in updating the estimates of xEV effectiveness used
in the Draft TAR and Proposed Determination analyses.
Costs of non-battery components have been declining since the 2012 FRM and are widely
expected to continue to decline. However, compared to engines and other conventional
powertrain components, many of which have been reduced to commodity products for many
years, the market in xEV non-battery components is still not as fully developed. As OEMs seek
non-battery components for their electrified products, they are less likely to encounter stock
items that fully meet their requirements and therefore have often chosen to either produce them
in limited numbers in-house, or to source them from suppliers that build to specification. While
this dynamic may be expected to limit the potential for economies of scale to develop and be
reflected in component costs in the near term, the Draft TAR noted that standardization and
commoditization will likely grow as the industry matures. For example, the decision of LG to
leverage its position as battery supplier to several OEMs by expanding into non-battery
components is one example of industry movement in this direction. In a joint announcement
with LG Chem in October 2015,154 GM described LG's role not only as supplier of battery cells
for the Chevy Bolt BEV but also as supplier of many of its non-battery components. LG's
established role as battery supplier to multiple OEMs suggests that it may be planning to supply
non-battery components across the rest of the xEV industry as well. As another example, in
2016 Siemens and Valeo announced the formation of a joint venture for the production of high-
voltage components across the full range of electrified vehicle types, citing among other
advantages "substantial synergies in manufacturing and sourcing" and a focus on global
markets.155 Developments such as these can promote the potential for economies of scale to
develop, and may be a significant driver of cost reductions if they continue in the future.
2.2.4.3.1 Propulsion Components
The components that provide propulsion for xEVs are known variously as electric motors,
traction motors, motor/generators, e-motors, or electric machines. In this discussion, they will be
referred to either as electric motors or generators (depending on the functional context), or
collectively as electric machines.
The two main types of electric machines currently seen in production xEVs are permanent-
magnet motors (also known as synchronous motors) and induction motors (also known as
asynchronous motors). Although the permanent-magnet motors used in xEVs are sometimes
called brushless direct-current (DC) motors, these as well as induction motors are powered by
alternating current (AC), which must be converted from DC battery current by an inverter.
In the duty cycles typical of xEV applications, permanent-magnet motors have certain
advantages in energy efficiency due in part to the presence of integral permanent magnets to
2-66
-------
Technology Cost, Effectiveness, and Lead Time Assessment
generate part of the magnetic field necessary for operation. However, these magnets add to
manufacturing cost, particularly when they contain rare earth elements. In contrast, induction
motors use copper windings to generate all of the magnetic field and can be manufactured
without rare earth elements. Although the windings are significantly less costly than magnets,
generation of the field in the windings is subject to additional I2R losses that are not present in
permanent magnet motors. In some conditions, this causes induction motors to be slightly less
energy efficient than permanent-magnet motors,156'157 although the choice between the two types
of motor ultimately depends on the specific application.
The majority of current xEV products use permanent-magnet motors. Induction motors are
found in products of Tesla Motors, as well as the Fiat 500e and Mercedes-Benz B-Class Electric
Drive. The BMW Mini-e and the Toyota RAV4 EV, both now discontinued, also used induction
motors; in the case of the RAV4, the motor was supplied by Tesla.
Another type of motor, the switched reluctance or axial flux motor, has recently been
suggested for use in xEVs.158'159 Although current examples of this technology are challenged
by difficulties with controllability, vibration, and noise, in the future these motors may
potentially offer a lower cost solution than either permanent-magnet or induction motors.
The Draft TAR noted that some manufacturers have demonstrated successful cost reductions
in propulsion components since the 2012 FRM. For example, the use of rare-earth metals in
permanent-magnet motors has been a target of cost reduction due to the high cost of these metals
and potential uncertainty in their supply. The 2016 second-generation Chevy Volt reduced the
use of rare-earths in its drive unit by more than 80 percent by using lower-cost ferrite magnets in
place of rare-earths in one of its motors160 and significantly reducing the rare-earth content of the
other.161 Another approach is seen in the BMW i3, which uses a hybridized motor design that
combines aspects of the permanent-magnet motor and the reluctance motor, allowing rare earth
content to be reduced by about half compared to a permanent-magnet motor of similar torque
capability.157
Component integration has also contributed to lower costs. GM has cited integration of
power electronics with the transmission and drive unit of the 2016 Volt as a significant enabler
of cost reductions in that vehicle by eliminating long stretches of heavy cable and improving
packaging efficiency.162'163 Major changes to the configuration of the electric propulsion system
reduced the total torque and power requirements, allowing the use of smaller bearings and rotors,
and an increase in maximum motor speed to 11000 rpm from the 9500 rpm of the previous
system. This led to a 20 percent reduction in motor volume and a 40 percent reduction in mass
compared to the previous generation, as well as improved efficiencies. Similar improvements
have propagated to the Cadillac CT6164 and the Chevy Malibu Hybrid165 through the sharing of
related components. The 2016 Toyota Prius also utilizes improvements to the transaxle and
motor that result in significant weight reduction and efficiency. A more compact motor design
and an improved reduction gear allows for an improved power-to-weight ratio and provides for a
20 percent reduction in frictional losses.166
Industry activity is also focused toward improving the efficiency of propulsion motors.
Although electric motors are already highly efficient (well in excess of 90 percent in many
normal usage conditions), even small improvements in efficiency can pay significant dividends
by reducing the battery capacity necessary for a given driving range. For example, GM has said
that the increased range of the second generation Chevy Volt was achieved in part by
2-67
-------
Technology Cost, Effectiveness, and Lead Time Assessment
improvements in motor efficiency.162 Even the first generation of the Chevy Spark EV was
described as having the highest drive unit efficiency in the industry, with an average battery-to-
wheels efficiency of 85 percent in the city cycle and 92 percent in the highway cycle.167 These
efficiencies are higher than EPA had assumed in the 2012 FRM xEV battery sizing analysis.
2.2.4.3.2 Power Electronics
Power electronics refers to the various components that control or route power between the
battery system and the propulsion components, and includes components such as: motor
controllers, that issue complex commands to precisely control torque and speed of the propulsion
components; inverters and rectifiers, that manage DC and AC power flows between the battery
and the propulsion components; onboard battery chargers, for charging the BEV or PHEV
battery from AC line power; and DC-to-DC converters that are sometimes needed to allow DC
components of different voltages to work together.
Inverters are power conditioning devices that manage electrical power flows between the
battery and propulsion motors. While all batteries are direct current (DC) devices, modern
traction motors operate on alternating current (AC) and therefore require an inverter capable of
converting DC to AC of widely variable frequencies at variable power levels. As implemented
in an electrified vehicle, the component commonly known as an inverter may also act as a
rectifier, that is, convert AC to DC to send energy to the battery.
Modern inverters are semiconductor based, utilizing metal-oxide-semiconductor field-effect
transistors (MOSFET) or insulated-gate bipolar transistors (IGBT). These designs are highly
efficient, often operating well above 90 percent efficiency. Inverter designs vary in output
waveform (square wave, sine wave, modified sine wave, or pulse-width modulated), which
accounts in part for differences in their efficiency and the potential for heat generation. Inverter
manufacturing cost is strongly associated with wafer size in manufacturing of substrate materials
such as silicon carbide. While most wafer sizes are currently around 4 inches in diameter, larger
wafers of 6 to 12 inches would reduce scrap rates and reduce cost substantially.168
Despite these low losses, the high power levels of electrified vehicles generate significant heat
and require inverters to have aggressive liquid cooling, often residing on the coolant loop in a
position prior to the propulsion motor to ensure sufficient cooling. Cooling elements such as
fans, heat exchange surfaces and fins or heat sinks can add to volumetric requirements and are a
common target of size and cost reduction. The similarity of materials and cooling needs offer an
opportunity to further reduce cost by integrating the inverter with other power electronics
components such as DC converters.169
The 2016 Chevy Volt provides one example of how improvements to the inverter and its
packaging can lead to significant improvements in packaging and related costs. Major changes
to the electric propulsion system served to reduce the current requirements of the inverter,
reducing its volume by about 20 percent (from 13.1L to 10.4L) and its mass from 14.6 kg to 8.3
kg. This allowed the inverter module to be integrated into a small space at the top of the
transmission. This integration into the transmission saved on assembly costs, served to protect
the components and their sensitive interfaces in a sealed environment, and eliminated the need
for heavy 3-phase cables. It also saved valuable under-hood space for other components
commonly associated with electrification. The reduction in inverter current was also said to
reduce inverter switching loss by about half in conjunction with accompanying improvements to
2-68
-------
Technology Cost, Effectiveness, and Lead Time Assessment
cooling. GM attributed a 6 percent improvement in electric drive system efficiency over the FTP
cycle, a 30 percent increase in vehicle range and an 11 percent improvement in label fuel
economy to these inverter improvements.162'163 Similar improvements have carried over to other
models that share related components, such as the Cadillac CT6 and the Chevy Malibu
Hybrid.164'165 Toyota also has introduced changes that improve inverter efficiency.166 The 2016
Toyota Prius includes a new power control unit to which it attributes a 20 percent reduction in
power losses. The power control unit also benefits from integration, residing in a position above
the transaxle. Advances in the use of a silicon carbide substrate in the power control unit are
also expected to significantly reduce power switching losses and allow a 40 percent reduction in
the size of the coil and capacitor of the power control unit in production Toyota vehicles by
around 2020.170
Many systems require DC-to-DC converters to allow DC components of different voltages to
work together. They do not convert between AC and DC, but instead step up or down the DC
voltage between two or more components or subsystems, either unidirectionally or bi-
directionally. One common application of a DC-to-DC converter is to allow low-voltage
accessories to be powered by energy from the high-voltage battery by reducing the voltage from
300+ V to 14 V. These are also known as buck converters, and commonly operate at about 1.5
kW171 to 3 kW.188 Although many current-production BEVs and PHEVs retain a low-voltage
battery to power accessories, a buck converter is needed to keep the low-voltage battery charged
in the absence of an engine-driven alternator, and can provide additional power to the
accessories. Another purpose of a DC-to-DC converter is to allow certain powertrain
components to operate at their optimum voltage rather than being tied to the voltage of the high
voltage battery. For example, a fuel cell stack or super capacitor may operate more efficiently at
a higher or lower voltage than the high-voltage battery, or along a variable range of voltages.172
A variety of topologies are under development to suit these varied applications.171'172
Controllers are electronic devices that implement control algorithms that control power flows
through the electrified powertrain. Motor controllers are responsible for issuing the complex
commands that precisely control torque and speed of the propulsion motor. A primary task of
this controller is to determine the exact frequency of alternating current necessary for the motor
to deliver the demanded speed and torque, and to control the inverter to provide it. A
supervisory controller is another form of controller that implements higher-level vehicle control
algorithms, including issuing high-level torque and speed commands to the motor controller.
Supervisory controllers are not unique to electrified powertrains but may be functionally
integrated with other components that are. Compared to other power electronics components,
controllers are not typically large consumers of energy, but can benefit from cost reductions
applicable to other components.
Onboard chargers are charging devices permanently installed in a PHEV or BEV to allow
charging from grid electrical power. Level 1 charging refers to charging powered by a standard
household 110-120V AC power outlet. Level 2 charging refers to charging with 220-240V AC
power. In practice, the charging power that is available in a given home installation may depend
on the amperage capability of the household circuit. Typical household circuitry can usually
support about 1 to 2 kW for Level 1 and about 5 to 7 kW for Level 2, although the SAE J1772
standard for Level 2 charging can support up to 19.2 kW with proper electrical service. Onboard
chargers travel with the vehicle, and are distinct from stationary charging equipment (Electric
Vehicle Supply Equipment, or EVSE) commonly installed at public or private charging stations.
2-69
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The Draft TAR (in Chapter 9, Infrastructure Assessment) included an examination of PEV EVSE
technology.
The widespread home availability of 110-120V AC power does not necessarily mean that
Level 1 charging is preferable either for convenience or efficiency. Charging time at the Level 1
rate is much longer than at Level 2. At Level 1, some longer-range BEVs may take longer than
overnight to bring from a low charge to full charge (although, for a given daily mileage, they
may reach a low charge state less often, and are equally capable of having daily mileage
replenished at Level 1 nightly). Level 1 residential charging is commonly relied upon by many of
the current users of BEVs and PHEVs, and provides a lower cost option for ownership that may
continue to be sufficient for households with lower daily driving needs.
Public charging infrastructure is also growing. As mentioned in the Draft TAR, since 2008,
various ongoing public and private efforts to provide charging stations at workplaces, along
freeway corridors, and in cities have grown the number of public stations in the U.S. from
practically a handful to more than 16,000.173 Since the Draft TAR was completed, two
developments were announced that may increase the availability of public charging substantially.
The partial settlement between Volkswagen and U.S. authorities, approved in 2016, earmarks
$1.2 billion in investment over 10 years toward ZEV infrastructure, education, and access.174
Also, in November 2016 The White House announced a network of federal, state, and local
initiatives to increase accessibility to PEV infrastructure,150 including a Department of
Transportation (DOT) plan to designate 48 national "alternative fuel corridors" along major
highways to provide focus for build out of charging locations by related local and state efforts.175
Public charging infrastructure was explored in depth in Draft TAR Chapter 9 (Infrastructure
Assessment), and is reviewed for this Proposed Determination assessment in Section B.3.2 of the
Proposed Determination Appendix. Some additional discussion in the context of BEV
technology is also found in Chapter 2.2.4.4.5 (Battery Electric Vehicles) of this TSD.
Charging efficiency can also vary significantly. In general, the efficiency with which a battery
accepts DC charge current is higher at lower charge rates.176 However, the degree to which the
manufacturer has optimized the charging circuitry for a specific preferred charge rate can also
have a strong influence, because the efficiency of AC to DC conversion is also an important
factor. According to tests performed by Idaho National Laboratory on a 2015 Nissan Leaf, the
efficiency of Level 1 charging ranged from only 61.8 percent to a maximum of 78.4 percent,
while that of Level 2 charging ranged from 81.5 percent to 90.5 percent.177 This suggests that
the design of the charging circuitry can have a greater effect on charging efficiency than charge
rate alone, and that manufacturers may optimize the charging system to accommodate the mode
of charging it expects customers to most commonly utilize.
DC fast charging is increasing in availability and popularity, and can support charging at
much higher rates than Level 2 (up to 150 kW in some cases, subject to the capability of the
vehicle being charged). Charging at these higher rates may result in a lower net efficiency
relative to Level 2, and may require more robust cooling of the battery and even the charging
connection to dissipate the heat generated during a charge.
Although charging efficiency is primarily relevant to upstream emissions and is not a factor in
onboard energy consumption, there is significant potential for efficiency improvement in these
components that may be indicative of similar potential in other power electronics components.
For example, between Genl and Gen2 of the Chevy Volt, the energy efficiency, size and weight
2-70
-------
Technology Cost, Effectiveness, and Lead Time Assessment
of its onboard charger was improved significantly.178'163 Level 1 charging efficiency improved
from 86.8 percent in Genl to 94.5 percent in Gen2, an improvement of 8.9 percent. Efficiency at
Level 2 increased similarly from 89.6 percent to 95.5 percent, an improvement of 6.6 percent.
These improvements allowed the overall system efficiency (from the wall plug to the battery) of
Level 2 charging to improve to 88.4 percent, and that of Level 1 to 86.7 percent (improvements
of 8.6 percent and 9.3 percent, respectively). Power density of the unit improved from 326 W/kg
to 605 W/kg (85 percent), while volumetric power density improved from 492 W/liter to 889
W/liter (81 percent), which led to significant packaging advantages. The fact that these
improvements to charger efficiency were achieved despite their lack of a strong impact on highly
visible attributes such as driving range or power suggests that similar improvements to other
components that do affect range or power are even more likely to be pursued successfully.
Battery management systems (BMS)179,330 are an important factor in maintaining and utilizing
the available capacity of the traction battery. A primary role of a BMS is to maintain safety and
reliability by preventing usage conditions that would damage or excessively degrade the battery.
The BMS may therefore limit voltages and currents on the pack, module, or individual cell level,
and monitor pack or cell temperature as well as other parameters.
Another important role of the BMS is to balance the charge levels of the individual battery
cells so that each cell is maintained at a similar voltage and state of charge. This can play an
important part in determining the usable portion of total battery capacity and in maintaining
battery life. In a battery containing hundreds of cells, small variations in resistance will exist
among individual cells, and differences in cell temperature will result not only from these
differences but also from differences in cell location within the pack and proximity to cooling
media. During a normal charge or discharge of the pack, these differences will affect cell
efficiency and cause some cells to approach their voltage or charge limits sooner than others.
Without balancing, the entire pack will effectively reach its charge or discharge limit when the
weakest cell reaches its limit. In this case, the charge contained in the remaining cells goes
unutilized. Effective cell balancing can increase utilization significantly.
BMS systems may employ passive or active balancing. Passive balancing acts to identify the
cells that are approaching their limits and selectively modifies their charge or discharge rates,
usually by dissipating their energy resistively, to allow the remaining cells to continue operating.
Active balancing shuttles energy among cells rather than dissipating the energy. Active
balancing is potentially more energy efficient than passive balancing but is typically costlier to
implement. The cost and effectiveness of active balancing is an active area of industry research
toward reducing the necessary battery capacity and power for a given application.
Somewhat counterintuitively, all current production BEVs have a conventional 12-volt lead-
acid battery in addition to the high-voltage traction battery. There are many practical reasons
why BEVs retain a low-voltage battery.180 Although the engine starting function is no longer
needed, a low-voltage power source is still needed for accessories and other functions. While a
DC-DC converter is available to step down the voltage of the traction battery to a suitable
voltage for the accessory bus (and in fact this is how the 12-volt battery is kept charged in the
absence of an engine-powered alternator), it is not a complete substitute for a battery because
neither the converter nor the high-voltage battery are kept in a powered state when the vehicle is
parked. Starting the vehicle therefore requires, at minimum, a low-voltage power source to close
the contactors and activate the high-voltage battery system. The vehicle may also continue to
2-71
-------
Technology Cost, Effectiveness, and Lead Time Assessment
draw current from the low-voltage battery to perform BEV-specific functions even while the
vehicle is off, perhaps for functions such as battery system maintenance and safety monitoring,
in addition to the other current draws that are common to many conventional vehicles. The low-
voltage battery may also act as a buffer between the DC-DC converter and the low-voltage bus,
allowing the DC-DC converter to operate intermittently rather than continuously to keep the
battery charged, and providing a stable voltage source to power sensitive microprocessor
components in the control circuitry.
BEVs therefore may subject the 12-volt battery to a different duty cycle than in conventional
vehicles. In recent years some evidence has accumulated that 12-volt lead-acid batteries in some
BEVs are being replaced after a relatively short life; in many cases, replacement has been
necessary on an almost annual basis.181'182 Although Tesla is said to specify a deep-cycle lead-
acid battery for the Model S, this battery is still reported to have a relatively short service life.183
In conventional vehicles, the size of the 12-volt battery tends to be correlated with the size of
the engine, due to its function in engine starting. Because a BEV 12-volt battery does not
perform this function, most BEVs can likely utilize a relatively small 12-volt battery regardless
of the power of the vehicle. For example, the 12-volt battery of the Tesla Model S has a capacity
of 33 Ampere-hours and weighs about 27 lb, smaller than the batteries found in conventional
vehicles of a similar power capability.184
The low cost, familiarity, and widespread availability of lead-acid 12-volt battery technology
is likely a factor in its selection as the basis for BEV low-voltage power. The potential tendency
for a relatively short life in BEV applications would seem to suggest that over the longer term,
other solutions such as a low-voltage lithium-ion battery may become competitive with lead-
acid. Despite the higher initial cost of lithium-ion, it could be a more cost effective solution if it
prevents multiple replacements of a lead-acid battery, particularly if the manufacturer anticipates
that many of those replacements may occur during the warranty period. While lead-acid has
traditionally performed better in cold weather, formulations of lithium-ion exist that are robust in
cold weather, and may weigh about half of an equivalent capacity lead-acid battery,185
potentially making the 12-volt battery almost an insignificant component of the total weight of
the vehicle. The Hyundai Ioniq PHEV, scheduled for introduction in the U.S. market in 2017,
has been described as eliminating the 12-volt battery, in favor of a 12-volt tap from the high-
voltage battery pack.186 Whether or not this innovation makes it into the production PHEV or the
BEV version, it indicates that some PEV manufacturers are actively investigating alternatives to
the conventional lead-acid low-voltage battery.
2.2.4.3.3 Industry Tar sets for Non-Battery Components
Establishing targets can be an effective way of focusing industry effort toward a common
goal. For example, the battery cost and performance targets established by the United States
Advanced Battery Consortium (USABC) are familiar to most in the battery industry and have
become important reference points by which developments in battery technology are often
measured. While industry targets such as these can vary in their purpose and achievability, they
can provide valuable guidance on what some in the industry consider to be potential directions
for future technology.
Targets for cost and performance of non-battery components have been established by U.S.
DRIVE,187 a government-industry partnership managed by the U.S. Council for Automotive
2-72
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Research (USCAR), which also manages USABC. Members include the U.S. Department of
Energy, industry members of USCAR, and several other organizations including major energy
companies and public energy utilities. The U.S. DRIVE targets apply to electric motors,
inverters, chargers, and other power electronics components for the 2015 and 2020 lab yearv
time frames.188 These targets, some of which are shown in Table 2.1, include performance
targets such as specific power, specific energy, and energy and power density (volumetric), as
well as cost targets.
The U.S. DRIVE targets were established specifically with respect to HEVs, which were seen
as presenting the greatest challenge in meeting the targets due to their being on the low end of
the power range compared to PEVs. The targets therefore apply best to an HEV-sized 55 kW
system. U.S. DRIVE expects the targets to be less difficult to meet for higher-power PEV
systems, in part because their more powerful powertrains may incur less overhead cost (for
connectors and the like) that are not necessarily directly proportional to power.189 This suggests
that the U.S. DRIVE targets would be relatively conservative when applied to PEVs.
Although the U.S. DRIVE figures are only targets, the industry has shown remarkable
progress in approaching these goals. It is notable that U.S. DRIVE targets for specific power are
quite close to what was already available in some production HEVs at the time they were set.
Since some of the goals were being met in higher-priced products, bringing these levels of
performance to the average PEV may largely be a matter of cost reduction rather than
technological breakthrough.
Table 2.1 U.S. DRIVE Targets for Electric Content Cost and Specific Power
Component
U.S Drive Target (Lab Year)
2015
2020
Electric motor
1.3 kW/kg
1.6 kW/kg
$7/kW
$4.7/kW
Power electronics
12 kW/kg
14.1 kW/kg
$5/kW
$3.30/kW
Motor and electronics combined
1.2 kW/kg
1.4 kW/kg
$12/kW
$8/kW
3 kW DC/DC converter
1.0 kW/kg
1.2 kW/kg
$60/kW
$50/kW
The 2020 lab year target for specific power of combined motor and power electronics has
some support in current literature. Assuming a five-year lag between lab demonstration and
production, the 2020 lab year corresponds to 2025. A presentation by Bosch190 at The Battery
Show 2015 states that the electric motor and power electronics for a 100 kW, 20 kWh BEV
system in the 2025 time frame is expected to comprise about 37 percent of electric content
weight, with battery weight comprising the remaining 63 percent. Assuming the 20 kWh battery
pack has a specific energy of about 140 Wh/kg (as indicated by ANL BatPaC for an NMC622
v It should be noted that a minimum of five years typically passes between successful demonstration of a technology
in a lab and its introduction into the market.
2-73
-------
Technology Cost, Effectiveness, and Lead Time Assessment
pack at 115 kW net battery power), and a corresponding weight of 143 kg, the non-battery
content would be estimated at about 53 kg. The 100 kW system would then represent a non-
battery specific power of 100 kW/53 kg, or 1.88 kW/kg. While the U.S. DRIVE target of 1.4
kW/kg is not directly comparable because it is based on a 55 kW traction motor, the result for the
100 kW example is directionally correct in the sense that U.S. DRIVE considers the targets
easier to achieve for more powerful systems.189 Most BEV and PHEV motors modeled in this
analysis are larger than 55 kW, suggesting that the U.S. DRIVE figure for a 55 kW system may
represent a fairly conservative figure for these applications.
Although the U.S. DRIVE figures are targets and therefore not necessarily indicative of
industry status, EPA has confidence that the targets for specific power represent attainable
production goals during the time frame of the rule. This is based in part on the observation that
the 2020 specific power target for electric motor and power electronics combined is very close to
levels that were already being attained by some production vehicles at the time they were set.191
Further, the motor of the recently announced Chevy Bolt BEV already appears to exceed the
U.S. DRIVE target at 1.97 kW/kg (based on a mass of 76 kg and peak power of 150 kW).192
This example is consistent with confidential business information conveyed to EPA through
private stakeholder meetings with OEMs that suggests that cost and performance targets for
some types of components are already being met or exceeded in production components today,
or are expected to be met within the time frame of the rule.
2.2.4.4 Developments in Electrified Vehicles
In this Proposed Determination analysis, each of the electrified vehicle categories represents a
distinct GHG-reducing electrification technology that manufacturers may choose to include as
part of a compliance pathway. These technologies range from 12-volt stop-start systems without
accompanying hybridization, to mild and strong hybrids (HEVs), to plug-in vehicles (PHEVs
and BEVs) and fuel-cell electric vehicles (FCEVs). The propulsion and power electronics
technologies discussed in the previous section are integral to understanding the architecture and
capabilities of each of these electrification technologies. Developments in each of these
electrification technologies are described in this section.
2.2.4.4.1 Non-hybrid Stop-Start
In this analysis, non-hybrid stop-start refers to a technology that reduces idling by temporarily
stopping the engine when the vehicle stops, and restarting it when needed. This eliminates much
of the fuel consumption associated with idling. In urban driving conditions that include a large
amount of idling at intersections and in congested traffic, stop-start can provide significant GHG
benefit.
Non-hybrid stop-start is also commonly known as idle-stop or micro hybrid. In the 2012
FRM, it was referred to as conventional stop-start. In this Proposed Determination analysis (as
in the FRM and Draft TAR analyses), non-hybrid stop-start is limited to engine stopping and
restarting in a 12V context, with no accompanying hybridization. For this reason, the term
micro-hybrid will not be used to refer to non-hybrid stop-start systems. The non-hybrid stop-
start classification should not be confused with mild and strong hybrids that include a stop-start
function. Systems that include brake energy regeneration or other hybrid features would be
classified as hybrids. However, as in the Ricardo analysis of the 2012 FRM, non-hybrid stop-
2-74
-------
Technology Cost, Effectiveness, and Lead Time Assessment
start may include a strategy known as "alternator regen" that charges the 12V battery more
aggressively by increasing the alternator field upon vehicle deceleration.
Non-hybrid stop-start is therefore the simplest form of electrification discussed in this section.
It is typically implemented by: (a) upgrading to a higher-performance starter capable of higher
power and increased cycle life, (b) upgrading to a higher-performance 12V battery to improve
cycle life and reduce voltage drop on restart; (c) adding an appropriate control system to manage
stopping and starting as transparently as possible; and in many cases, (d) modifying certain
accessories to allow for adequate service while the engine is off.
As originally modeled in the 2012 FRM, the effectiveness estimates for stop-start were
derived from the Ricardo modeling study, which estimated 2-cycle effectiveness to be in the
range of 1.8 to 2.4 percent, depending on vehicle class. As originally represented in the 2012
FRM, stop-start was considered to be a new technology and was assigned a steep learning curve
for the years 2012-2015 and a flat learning curve for the years 2016-2025. On the basis of
projected costs and effectiveness, EPA projected that stop-start would achieve a fleet-level
penetration of 15 percent193 in the cost-minimizing pathway for compliance with the 2025MY
standards.
As discussed in the Draft TAR, since the 2012 FRM, rapid growth in the application of 12V
stop-start systems is evidence of the technology's potential to provide cost-effective emissions
reductions. The 2015 EPA Trends Report projects that non-hybrid stop-start will be present on
almost 7 percent of new non-hybrid car and truck production in MY2015, with total penetration
of stop-start at nearly 9 percent when mild and strong hybrids are included.194 Penetration has
grown steadily each year, reaching 0.6 percent in 2012, 2.3 percent in 2013, and 5.1 percent in
2014, with 6.6 percent projected for 2015.195 BMW and Mercedes-Benz are the most notable
adopters, each including stop-start in about 70 percent of their projected 2015 production.196 In
comments on the Draft TAR, CALSTART described a survey of suppliers, performed by
Ricardo. The comment indicated that suppliers in the survey consider stop-start to be among the
top 5 technology strategies for meeting the 2025 standards. CALSTART also stated,
"CALSTART has had a number of conversations with different suppliers who have indicated
they are making major investments in 48V mild hybrid technology as a leading strategy to meet
standards particularly in China and Europe."
As a GHG-reducing technology, the effectiveness of stop-start depends on the amount of idle
time included in the assumed test cycle. The standard EPA test cycles contain short periods of
idle, but less than some believe is present in real world driving. In order to provide a more
accurate credit basis for the real-world benefit of stop-start, stop-start technology is eligible for
off-cycle credits under the Off-Cycle Program. The Off-Cycle Program is discussed further in
Chapter 2.2.10.
As discussed in the Draft TAR, in contrast to the 2012 FRM projections of 1.8 to 2.4 percent
effectiveness under EPA test cycles, other sources have suggested an average of 3.5
percent.197,198'199 As one example, the Draft TAR noted that the 2015 Ford Fusion 1.5L TGDI is
available with and without a 12V stop-start option, providing an opportunity to assess the
effectiveness of stop-start as implemented in this vehicle. The difference in estimated fuel
economy between the two versions suggests an effectiveness of about 3.5 percent on a fuel
economy basis. The automotive supplier Schaeffler Group has presented an engine stop-start
technology200 it describes as capable of providing a 2-cycle combined fuel economy
2-75
-------
Technology Cost, Effectiveness, and Lead Time Assessment
improvement of about 6 percent over the city cycle and 2 percent over the highway cycle, or
about 3.42 percent combined. The 2015 Mazda3 is available with and without the Mazda i-
ELOOP regenerative braking and stop-start system. A comparison of certification test data for
this vehicle with and without the system suggests that its two-cycle GHG effectiveness is about
3.35 percent.201
Some test cycles used in other parts of the world include a greater proportion of idle time and
therefore assign a greater benefit to stop-start. This would naturally make stop-start more
attractive to manufacturers in regions that certify under these cycles, and may be a factor in the
greater penetration of stop-start that has been observed worldwide. Stop-start197 has been
popular in Europe due to high fuel prices and the stringent EU CO2 emission target established in
2009. In 2014, about 60 to 70 percent of vehicles sold in the European market offered stop-start.
Because stop-start technology alters the customary operation of the engine, it has potential to
alter the traditional feel of driving. Frequent restarts of the engine, although rapid and seamless
in most implementations, can increase the sense of noise, vibration, and harshness (NVH).
Drivers unaccustomed to stop-start may at first feel uncomfortable having the engine switch off
in stop and go traffic, particularly if accessories such as heat or air conditioning are also affected.
Some of the seamlessness and potential benefit of stop-start can be eroded by individual driving
habits. For example, if a driver repeatedly pulls up toward the leading car as traffic compacts
while waiting at an intersection, the engine may restart each time, reducing fuel savings and
adding to NVH.
Manufacturers often cite consumer acceptance factors in the adoption of stop-start in the U.S
market. Early introductions of the technology involved lower volume vehicles and adaptations
of systems originally designed for the European market. Manufacturers have considered
customer feedback from these early applications in the implementation of recent stop-start
systems, which are now smoother and more unobtrusive to the driver. For example, some
suppliers have proposed continuously engagement of the starter motor to improve the restart
process. Others have implemented systems that maintain a specific piston position while stopped
in order to achieve a fast and smooth restart by firing a single cylinder. As a result, improved
systems promise greater effectiveness through more frequent and longer periods of idle stop time
while operating in a more transparent manner.
Vehicles with sufficiently smooth and seamless stop-start technology have been well-received
by consumers,202 especially when paired with some explanation of the system's benefits and
operating characteristics at the time of delivery. With these more recent implementations, it is
more common now for stop-start systems to be applied as standard equipment on high-volume
vehicles like the Chevrolet Malibu, Chrysler 200, Jeep Cherokee, and Ram 1500 truck. Ford also
offers it on its high-selling F-150, and expects to offer it on 70 percent of its North America
vehicle lineup by 2017.203
The Draft TAR noted that the introduction of stop-start has stimulated development of 12V
battery systems capable of providing the enhanced performance and cycle life that it requires.
Much of this activity has involved variations of lead-acid chemistries, such as absorbed-glass-
mat (AGM) designs and lead-carbon formulations. For example, at the 2015 Advanced
Automotive Battery Conference (AABC), a Planar Layered Matrix (PLM) 12V enhanced lead-
acid battery was exhibited by Energy Power Systems (EPS). EPS claimed this technology
2-76
-------
Technology Cost, Effectiveness, and Lead Time Assessment
increases battery power and regenerative charging capability by a factor of four while increasing
the battery life by a factor of five, at a similar cost to a conventional AGM lead-acid battery.
The Draft TAR also noted that lithium-ion chemistries specially adapted for stop-start
applications have begun to take hold. As one example, Maxwell Technologies has developed a
12V lithium-ion battery combined with a 395V ultra-capacitor pack designed for 12V stop-start
systems.204 The dual pack was said to provide quicker engine start, lower voltage drop, capacity
and life improvement while providing capability to operate at -30 degrees Celsius. Since the
battery and ultra-capacitor operate at different voltages, these systems require additional
electronics for DC to DC conversion. These systems are also likely to cost more than lead-acid
based systems. The cost of the Maxwell dual pack stop-start system is estimated at about
$230/pack, which is higher than that of an advanced lead-acid battery. In general, use of the
lithium-ion chemistry for 12V stop-start applications continues to face challenges with regard to
cost as well as cold-start operation.
The Mazda i-ELOOP system205 represents an incremental step beyond basic stop-start, using
ultra capacitors to store regenerative brake energy during deceleration and coasting. While the
system cannot use the reclaimed energy for propulsion, it supplements the energy used by
accessories and climate control, potentially saving energy by allowing the engine to stay off for
slightly longer periods.
Based on a review of these and similar industry developments, as well as data collected from
other sources, EPA updated effectiveness estimates for stop-start technology in the Draft TAR
and these estimates remain current for this Proposed Determination analysis. The cost and
effectiveness estimates, as well as some of the public comments received on stop-start
technology, are discussed further in Chapter 2.3.4.3.1.
2.2.4.4.2 Mild Hybrids
In this analysis, mild hybrid refers to a technology that supplements the internal combustion
engine by providing limited hybridization, typically including a limited amount of electrical
launch assistance, some regeneration, and stop-start capability. Together, these features reduce
energy consumption by optimizing loading of the engine, enabling some engine downsizing,
allowing the engine to turn off at times, and recovering a portion of the energy that would
otherwise be wasted by friction braking. Mild hybrids commonly are implemented in part by
replacing the standard alternator with an enhanced power, higher voltage, higher efficiency belt-
driven starter-alternator which can provide some propulsion assist and also recover braking
energy while the vehicle slows down (regenerative braking). Although the belt-driven basis of
these systems can limit their power capability to approximately 10 kW to 15 kW,206 mild hybrids
can provide greater benefit than stop-start systems while keeping cost significantly lower than
that of a strong hybrid.
Mild hybrids operate at a higher voltage than 12V stop-start systems. Even the relatively mild
demands of stop-start207 technology are very demanding on a 12V electrical system. Achieving
the 10 to 15 kW demanded of a mild hybrid application at 12V would require discharge currents
of 1000 Amps or more, which would require very thick, heavy, and expensive electrical
conductors. In order to achieve effective launch assist and regeneration, mild hybrids therefore
operate at higher voltages of 48V to 120V or higher, with an increased battery capacity as well.
2-77
-------
Technology Cost, Effectiveness, and Lead Time Assessment
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.
In the 2012 FRM analysis, mild hybrid technology was referred to as "higher-voltage stop-
start/belt integrated starter generator (BISG)" and was limited to BISG architecture, as
exemplified by the Chevrolet Malibu eAssist system. The primary source of effectiveness data
used by EPA was derived from the Lumped Parameter Model based on modeling of the Malibu
Eco BAS (BISG) system with a 15 kW motor and 0.5 kWh battery. EPA cost estimates were
based on an analysis of this system with a 0.25 kWh battery. EPA had then assumed an absolute
CO2 effectiveness ranging from 6.8 to 8.0 percent depending on vehicle class (2012 RIA, p. 1-
18). These effectiveness values included only the effectiveness related to the hybridized
drivetrain (battery and electric motor) and supported accessories.
The 2012 FRM analysis had projected that mild hybrids would achieve a fleet-level
penetration of 26 percent208 in the cost-minimizing pathway for compliance with the MY2025
standards. This was reduced to 18 percent in the Draft TAR analysis.137
The EPA Trends Report does not distinguish between mild and strong hybrids in its
accounting of hybrid vehicle penetration, which makes it difficult to separate the relative
penetration of mild hybrids from that of strong hybrids since the 2012 FRM. Although most
analysts had forecast the market share of hybrid vehicles to slowly but steadily rise, hybrid
market share (including mild and strong hybrids) has leveled off at about 3 to 3.5 percent209 of
the total light vehicle market since 2009. According to a report by the International Council on
Clean Transportation (ICCT),210 GM mild hybrid systems accounted for about 2 percent of the
2014 U.S. market, a decline from about 5 percent in 2013. Other sources have remained
optimistic that penetration levels will eventually grow substantially. For example, the
automotive supplier Continental has projected market penetration rates of three million BEVs, 12
million strong hybrids and 13 million 48V mild hybrids by 2025.211 In comments on the Draft
TAR, A123 stated that it expects "sales of more than 1 million 48V battery systems annually to
its global customer base by the year 2020." Mercedes-Benz has announced plans to introduce a
48V architecture, enabling mild hybrid functions, in some of its vehicles by 2017.142 Toyota has
also been a leader in and proponent of hybridization, stating for example in comments on the
Draft TAR, "Continued expansion of hybrids will play a key a role in the eventual shift to greater
levels of vehicle electrification."
Examples of high-voltage BISG mild hybrid systems currently present in the U.S. market are
the 115V Buick Lacrosse eAssist and the 90V 2017 Chevrolet Silverado truck218 mild hybrid
system. Hyundai is also using BISG technology for torque smoothing in its high voltage BISG
Hybrid Starter Generator (HSG) drivetrain.
Like stop-start technology, mild hybrid technology alters the customary operation of the
engine and so can alter the traditional feel of driving. In many situations the engine may turn off
less frequently and be off for longer periods, although the cycling may appear more random
because it is not necessarily connected to stop and go operation. Some of the effectiveness of
mild hybrids may be diminished by individual driving habits, leading to possible dissatisfaction
with fuel economy. For example, the fuel economy benefit of mild hybrids may fall off more
quickly with aggressive driving due to the lower potential for engine-off operation under these
conditions.
2-78
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The 2015 National Academy of Sciences (NAS) report estimated a 10 percent effectiveness
for mild hybrid technology212 based upon the 11 percent fuel consumption reduction observed in
the 2013 GM Malibu Eco. The NAS estimate appears reasonable when considering
improvements in the GM Ecotec engine and six-speed automatic transmission, and when
considering differences between the vehicle's 0-60 mph acceleration times (which are reported to
be about 7.8 seconds for the base 2013 Malibu LT213 and 8.2 seconds for the 2013 Malibu
Eco214).
The GM Malibu 15 kW 115V eAssist BISG mild hybrid improved fuel economy about 11
percent over the conventional Malibu Eco 2.5L PFI engine with a six speed transmission. This
effectiveness figure includes the benefits of other non-hybrid technologies (such as low rolling
resistance tires, underbody aerodynamic panels and radiator grille active shutters) that are
present on the e-Assist mild hybrid package.
The 2013 GM Malibu Eco's eAssist system uses a 15 kW BISG induction motor with 11 kW
launch assist during heavy acceleration and 15 kW of recuperative braking power.215 The
effectiveness of a 12 to 15 kW electric machine with a liquid-cooled integrated inverter in a 48 V
mild hybrid is comparable to that of a 15 kW motor in 100V+ mild hybrid when taking into
consideration the 30 pound weight reduction from the battery pack and the three, long and heavy
3-phase AC cables used in the 100+V BISG system. For an equivalent mass, 48V mild hybrid
technology effectiveness216 will be slightly less than that of 100V+ mild hybrids.
Since the 2012 FRM, the GM eAssist platform has migrated to other vehicles in the GM
lineup. In February 2016, General Motors announced a limited pilot program offering a version
of its eAssist mild hybrid system on approximately 200 GMC Sierra 1500217 and 500 Chevrolet
Silverado218 2WD pickups in California. This option is offered at a retail price of $500,
significantly lower than the approximately $1000 cost attributed to the 2013 Malibu Eco hybrid
system by an FEV teardown analysis.219 GM credits this system with up to a 13 percent
improvement in city fuel economy. This development is significant in part because it is the first
example of a BISG system applied to production pickup trucks by a major manufacturer. GM
stated that it would "monitor the market closely [...] and adjust as appropriate moving forward."
GM is also offering the eAssist BISG mild hybrid as an option to Chevrolet Equinox and GMC
Terrain midsize SUVs, and Buick Verano, Buick Regal, and Buick Lacrosse. At least one
analyst expects annual sales of these vehicles to grow to about 100,000 by 2020,209 suggesting
that BISG may become a significant contributor to the compliance path of manufacturers that
rely on this technology.
The Honda Civic IMA (Integrated Motor Assist) or PI mild hybrid integrates a 1.5L inline
four cylinder Atkinson cycle engine220 with a CVT transmission and a 17 kW CISG motor to
achieve a 29.7 percent total GHG effectiveness (calculated from two-cycle certification data
comparing the 2015 1.5L Honda Civic IMA to the 2015 1.8L Honda Civic sedan). The
effectiveness attributable to the mild hybrid technology alone can be estimated by subtracting the
effectiveness of the other technologies present on the vehicle. This includes about 1.9 percent
for low rolling resistance tires (LRRT1), 0.7 percent for low drag brakes (LDB), 1.3 percent for
electrical power steering (EPS), 0.7 percent for LUB, 3 percent for use of Atkinson cycle ICP
and DCP, 3.5 percent for use of a CVT, 3 percent for HEG, 0.8 percent aerodynamics and 1.5
percent for weight difference, resulting in about 13.3 percent GHG effectiveness for this system.
2-79
-------
Technology Cost, Effectiveness, and Lead Time Assessment
This comparison does not consider the small 0-60 acceleration performance loss (from 9 seconds
to 9.8 seconds) between the standard 1.8L sedan and the IMA hybrid.
Combined two-cycle certification test data comparing the 2015 Mercedes-Benz E400
20kW 120V P2 mild hybrid and the comparable E350 conventional vehicle indicated about 13
percent GHG effectiveness.
To date, most mild hybrids such as the aforementioned Malibu eAssist have been designed to
operate at a voltage of 100V or higher. However, as discussed in the Draft TAR, evidence has
accumulated since the 2012 FRM to suggest that many functions of a BISG mild hybrid can be
provided at a lower voltage, such as 48 V, at significantly reduced costs. Several attributes of
48V systems contribute to this lower cost. The voltage is lower than the 60V safety threshold
that would otherwise require more robust electrical shock protection. The small power levels
associated with these components promotes integration of the inverter with the motor and the
elimination of long stretches of cable, further isolating the AC portion of the circuit. The
relatively small 48V battery pack is significantly less costly due to having a potentially smaller
capacity as well as fewer cells due to its lower voltage. The battery may not require liquid
cooling, instead being passively cooled with appropriate placement and packaging. The
relatively low power requirements of a 48V system also promotes use of relatively inexpensive
motor technology (such as induction or switched reluctance) without as strong a concern over
NVH or efficiency. The lower voltage and capacity leads to a lower return in effectiveness216 (for
example, a 48V system may have a regenerative energy capturing efficiency of about 50
percent221 compared to perhaps 85 percent for a typical strong hybrid), but the cost reduction
may make these systems more cost effective. For example, A123 Systems has projected a fuel
economy effectiveness of 12 percent for a 48V mild hybrid system utilizing its 48V battery
technology.222 At this level of effectiveness, this system was described as being more cost
effective (at $55 per percent fuel economy gain) than a full hybrid solution (at $83).
48 V mild hybrid technology has received an increasing amount of attention since the 2012
FRM, with a number of OEMs and suppliers introducing several developmental 48V mild hybrid
systems capable of significant CO2 and fuel consumption reductions. At the 2015 SAE Hybrid
and Electric Vehicle Technology Symposium, Controlled Power Technology (CPT) exhibited a
switched-reluctance motor-generator technology and an electric supercharger for 48 V vehicle
electrification. Bosch has presented a 48V mild hybrid system scheduled to be ready for
production by 2017223 that it describes as capable of a 15 percent reduction in fuel consumption.
At the 2015 Consumer Electronics Show (CES), Continental exhibited a 48V mild hybrid system
which consists of a 48V Belt Integrated Starter Generator (BISG) replacing 12V alternator,
DC/DC converter and a 48V lithium-ion battery pack. The BISG motor is an induction motor,
and liquid cooled by engine coolant. The motor can be decoupled for downhill coasting by
disconnecting the transmission from the engine. Continental expects this 48V mild hybrid
system to begin production in 2016.224 In concert with these introductions, suppliers are also
predicting significant market penetration for 48 V systems within the time frame of the rule.
Bosch projected some 4 million 48V mild hybrid vehicles worldwide in 2020, while Eaton
expected up to 3 million 48V mild hybrids globally by 2020.210
A 48 V mild hybrid truck was announced in the recent FCA business plan225 for the 2018
Dodge Ram 1500 large truck using next-generation powertrains.226 Schaeffler227 and Hyundai228
2-80
-------
Technology Cost, Effectiveness, and Lead Time Assessment
also recently demonstrated advanced engineering prototypes of small and mid-size SUV 48V
mild hybrids.
48V mild hybrid prototype demonstration vehicles from Audi, Hyundai, Mitsubishi, and
Johnson Controls have been described as delivering about 10 to 15 percent CO2 reduction and
fuel economy improvement.229 Continental, a major Tier 1 supplier of electrified automotive
systems, has presented a prototype small car with a 10 kW BISG 48V mild hybrid system, said
to provide a 7 percent CO2 reduction.230 In the FRM, the agencies calculated a 7.4 percent GHG
effectiveness for small cars equipped with a 10 kW BISG mild hybrid system, which is
comparable to the Continental results.
Industry appears to be coalescing on a 48V standard for such mid-voltage hybrid applications,
with manufacturers such as Audi, BMW, Daimler, Porsche and VW having initiated a 48V
standard known as LV148 231
48V mild hybrid technology can also be understood as an alternative to stop-start that is not as
costly as adopting a higher voltage mild hybrid technology. Compared to 12V stop-start, 48V
mild hybrids provide several benefits for a relatively small cost increase,232 such as faster engine
starting, more engine-off time, significant regenerative braking capacity, and better electrical
support for accessories while the engine is off. In comments on the Draft TAR, several
commenters reiterated the conclusion that 48V technology is more cost effective than higher
voltage systems. For example, A123 commented, "we expect 48V mild hybrids to remain one of
the most cost effective forms of electrification through model year 2025 and beyond."
As discussed in the Draft TAR, EPA expects 48V mild hybrid technology to become
increasingly common and relied upon as a GHG reducing technology. See generally Draft TAR
at 5-77 and Chapter 5.2.4.3.2. EPA therefore added the 48V mild hybrid architecture to the
Draft TAR analysis and will retain it as part of this Proposed Determination analysis.
Recent developments in the 48V platform have suggested that it is also capable of pushing the
limits of what would be considered a mild hybrid. New P2, P2/P4 and P0/P4 48V system
architectures have been presented by various suppliers such as Bosch, Schaeffler, Continental,
and Control Power Technologies, ranging from 20 kW to 45 kW of assist capability.211 The
effectiveness for these new, more powerful systems, particularly those on the higher end of the
power range (30-45kW) may approach that of P2 strong hybrids but at a much lower cost. For
example, Bosch has presented a 2nd generation, 48V P2-architecture mild hybrid currently in
development.233 In this 48V P2 system, a more powerful motor-generator is integrated into the
transmission (to create a transmission-integrated starter-generator or TISG architecture). As with
a P2 strong hybrid, the motor can be decoupled from the engine to propel the vehicle in an
electric-drive mode in stop-and-go traffic and for short distances.
Transcending the BISG format provides a way around common mild hybrid limitations, such
as the 15 kW peak motor power limit, belt efficiency losses, and tandem operation of the engine
with the motor. Stronger formats such as Crank-Integrated Starter Generator (CISG) PI
architecture, as well as Transmission Integrated Starter Generator (TISG) P2 architecture,
overcome the peak motor power limitation in BISG P0 mild hybrids and further increase the
potential effectiveness of mild hybrid technology. The Honda IMA CISG PI mild hybrid system
cannot run the electric motor alone without simultaneously operating the internal combustion
engine,234 while the TISG P2 mild hybrid format allows the engine shut down while the electric
2-81
-------
Technology Cost, Effectiveness, and Lead Time Assessment
motor works independently for braking energy recuperation and vehicle propulsion. The
effectiveness of TISG P2 mild hybrids therefore may have higher effectiveness potential than
that of CISG PI mild hybrids.
The effectiveness of TISG P2 mild hybrids appears to be higher than that of CISG PI mild
hybrids. GETRAG projected about 15 percent effectiveness for a 48V 21 kW TISG P2 mild
hybrid at the 14th VDI Congress.206 This system employs a 7 speed dual clutch hybrid
transmission, which integrates one common oil circuit for cooling and lubrication, and a
combined e-machine and inverter applicable not only to the 48V 21 kW mild hybrid but also to
other variants such as a 220V+, 50 kW strong hybrid and a 360V+, 110 kW plug-in hybrid
application. This hybrid transmission also supports other efficiency-enhancing features such as
pure electric driving, extended sailing, more efficient launch assist and brake energy
recuperation, battery charging when the vehicle is standing, and generator-mode/load shift;
features very similar to those provided by strong hybrids.
In addition to its own benefits, mild hybridization may help enable the use of other
technologies that can further improve efficiency. For example, fuel consumption reduction may
approach 20 percent when an electric supercharger is used in 48V mild hybrids combined with
regenerative braking energy recovery, engine downsizing and downspeeding.235 Audi is
expected to market a system utilizing this technology in 2017. As another example, a 48V, 7 kW
electric supercharger236 has been shown to deliver an extra 40 to 70 kW at the crankshaft by
boosting the engine combustion process. Hence, the electric supercharger may be an effective
accompaniment to engine downsizing and downspeeding.
Based on a review of these and similar industry developments, as well as data collected from
other sources, EPA updated effectiveness estimates for mild hybrid technology in the Draft TAR.
EPA has reviewed these estimates and finds that they remain applicable to this Proposed
Determination analysis. Cost and effectiveness estimates, as well as some public comments
received on this technology, are discussed further in Chapter 2.3.4.3.2.
2.2.4.4.3 Strong Hybrids
In this analysis, strong hybrid refers to hybrid technologies that have higher power capability
and larger battery capacity than mild hybrids, thus providing for more effective management of
power from the internal combustion engine, greater levels of regenerative braking, and more
powerful electric propulsion capable of accelerating the vehicle with less (if any) assistance from
the engine. Strong hybrids provide greater effectiveness than mild hybrids by better optimizing
loading of the engine, allowing additional engine downsizing, allowing the engine to turn off for
longer periods, and recovering a greater portion of braking energy. These enhanced functions
tend to require higher voltages (as high as 300V to 400V) and more powerful batteries with
greater energy capacity, typically on the order of 1 to 2 kWh. These attributes add to complexity
due in part to safety requirements associated with higher voltages and greater battery capacity.
Although strong hybrids are more expensive than mild hybrids, they can access a greater degree
of fuel economy and CO2 reduction than mild hybrids, and include some of the highest fuel
economy vehicles currently in production.
Strong hybrids include several distinct architectures. On a sales-weighted basis, the power-
split hybrid electric vehicle (PSHEV) represents the most common architecture, largely by virtue
of its use for many years in the Toyota Prius hybrid. This system replaces the traditional
2-82
-------
Technology Cost, Effectiveness, and Lead Time Assessment
transmission with a single planetary gearset and two motor/generators. The smaller
motor/generator uses the engine to either charge the battery or supply additional power to the
drive motor. The second, more powerful motor/generator is permanently connected to the
vehicle's final drive and always turns with the wheels, as well as providing regenerative braking
capability. The planetary gearset splits engine power between the first motor/generator and the
output shaft to either charge the battery or supply power to the wheels.
The two-mode hybrid electric vehicle (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. Although the added mechanical elements
can introduce their own losses, in many cases the system overall can improve the transmission
torque capacity for heavy-duty applications while possibly reducing fuel consumption and CO2
emissions at highway speeds relative to other types of hybrid electric drive systems.
The P2 hybrid is a hybrid technology that uses a transmission integrated electric motor placed
between the engine and a gearbox or transmission, with a wet or dry separation clutch which is
used to decouple the motor/transmission from the engine. A P2 hybrid would typically be
equipped with a larger electric machine than a mild hybrid system but smaller than a power-split
or 2-mode hybrid architecture. 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. Based on simulation, when combined with a DCT transmission, the P2 hybrid
architecture provides similar or improved fuel efficiency to other strong hybrid systems with
reduced cost.
In the 2012 FRM, P2 hybrid was the only hybrid architecture that was applied in the agencies'
analysis. Although PSHEV and 2MHEV technology were discussed because they were present
in the market at the time of the FRM, they were not included in the analysis because the industry
was expected to trend toward more cost-effective hybrid configurations such as P2.
Going back to the 2012 FRM, the primary reference EPA used for strong hybrid effectiveness
was the Ricardo modeling study, which modeled a P2 with a futured DCT. On this basis EPA
had estimated an absolute CO2 effectiveness for P2 strong hybrids ranging from 13.4 to 15.7
percent depending on vehicle class (see 2012 RIA, p. 1-18). These figures included only the
effectiveness related to the hybridized drivetrain (battery and electric motor) and supported
accessories, and did not include the effect of any accompanying advanced engine technologies.
The quoted figures were based on electric motor sizes assumed in the Ricardo vehicle simulation
results and would vary with other motor sizes.
On this basis, EPA had projected that strong hybrids would achieve a fleet-level penetration
of about 5 percent237 in the cost-minimizing pathway for compliance with the MY2025 GHG
standards. The Draft TAR analysis revised this to less than 3 percent.148
The EPA Trends Report does not distinguish between mild and strong hybrids, nor specific
architectures of strong hybrids, in its accounting of hybrid vehicle penetration. Therefore it is
difficult to use this source to assess the relative penetration of P2 and other strong hybrid
architectures since the 2012 FRM. However, it is expected that strong hybrids are making up the
majority of the market.
2-83
-------
Technology Cost, Effectiveness, and Lead Time Assessment
A recent report by the International Council on Clean Transportation (ICCT)210 reviews
market penetrations for various hybrid architectures. According to this report, the market share
of the P2 hybrid architecture among all hybrids has been relatively small, having grown from
about 9 percent in 2013 to about 12 percent in 2014. Toyota has continued to lead the U.S.
hybrid market with 66 percent of U.S. hybrid sales in 2014. These sales largely account for the
dominance of power-split hybrids in the market. In the same year, Ford claimed a 14 percent
share of the U.S. hybrid market, also with power-split hybrids. P2 hybrids are primarily
represented in the U.S. market by Hyundai/Kia and Honda, with 8 percent of total 2014 hybrid
sales. The Honda integrated-motor-assist (IMA) architecture represented only 3 percent of the
2014 hybrid market, and is expected to be replaced by a P2 system in the near future.
Compared to the more mature, fourth generation power split hybrid architectures of Toyota
and Ford, EPA believes the P2 hybrid architecture is still in a relatively early stage of
development and has yet to be fully optimized. Manufacturers are continuing to make strides
toward improving this architecture in recently introduced models by refining power electronics
and component efficiency and integrating parts. For example, Hyundai has improved the 2nd
generation Sonata hybrid by fully integrating a 38 kW traction motor and all of the other hybrid
powertrain components within the transmission. The reduced weight has led to improved fuel
economy with reduced costs, as evidenced by the observation that there is no major difference in
effectiveness between this P2 vehicle and the 2015 Toyota Camry power-split hybrid. Going
forward, similar opportunities for major cost reduction and fuel economy improvement are likely
to arise in competing P2 hybrid systems.
Differences in configuration account for some of the cost and effectiveness differences
between P2 and power-split architectures. The input power-split architecture requires two
motors, which consist of a small generator and a bigger traction motor which drives through a
simple power-split planetary gear set. The P2 architecture uses a single, smaller traction motor,
but drives through a more complex conventional transmission gearing. The Honda two-motor
architecture does not use a power-split planetary gear set, and therefore requires a bigger motor
to directly transmit power to the drive axle compared to the typical input power-split hybrid
system. For example, the Honda Accord 2-motor hybrid uses a 124 kW traction motor238 while
the Toyota Camry power-split hybrid uses a 105 kW traction motor.239 Highly efficient motor-
integrated DCT transmissions have recently entered production or are under development and are
being adopted in the latest P2 parallel hybrid designs. The architecture of the P2 parallel hybrid
also may potentially provide for a greater towing capacity than the power-split hybrid
architecture, which in the current production market appears to be limited to the 3500-lbtowing
capacity of the Toyota Highlander hybrid.
Even the relatively well-developed power-split architecture continues to show room for
efficiency improvements. Toyota redesigned the 2016 Prius240 transaxle and motor in its fourth
generation Hybrid Synergy Drive (HSD) to reduce combined weight by 6 percent and volume by
12 percent. The planetary gear arrangement in the reduction gear has been replaced with parallel
gears, reducing mechanical losses by approximately 20 percent. The 53 kW main traction motor
is mounted on a parallel shaft, enabling the transmission case volume to be reduced substantially
while also reducing frictional losses by about 20 percent. The power control unit, which
combines the controller, inverter and DC/DC converter, was attached to the top of the transaxle
and its size reduced by about 33 percent by eliminating several high-voltage cables. The lithium-
ion battery pack, initially made available on the 'Eco' trim level, is 6 percent smaller and 31
2-84
-------
Technology Cost, Effectiveness, and Lead Time Assessment
percent lighter than the nickel-metal hydride (Ni-MH) version, while providing the same power
output and degree of hybridization.
Further evidence that the effectiveness of input power-split hybrids and P2 parallel hybrids
are getting closer is shown by the 2017 Hyundai IONIQ P2 hybrid, announced in 2016. The
combined fuel economy of this vehicle, with the GEN2 Hyundai P2 parallel hybrid drive, is
expected to be about 53 mpg, which is comparable to the 56 mpg fuel economy of the 2016
GEN4 Toyota Prius Eco hybrid. This vehicle also employs advanced technologies such as a
gasoline direct injection (GDI) inline 4 cylinder Atkinson cycle engine, cooled EGR, CVVT,
dual circuit cooling system, 6 speed dual clutch transmission (DCT), exhaust heat recovery
system, and an intake oil control valve, which act together to increase engine thermal efficiency
to as high as 40 percent.
As reported by ICCT21" (and reproduced here in Figure 2.30), the estimated costs for hybrid
systems have tended to decline steadily in the years after their introduction. If these trends
continue, significant reductions in hybrid system cost may be expected during the time frame of
the rule.
Priiis
generation
adlustments
$3,500
powerspllt - FEV
$3,000
P2 - TSD
S2.500
P2 - FEV
$2,000
Hi-power battery '
-2.5%/ye at
$1,500
Mild - TSD
$1,000
-SX/yezr ~
$500
$0
2000
2005
2020
2010
Figure 2.30 Hybrid System Direct Manufacturing Cost Projection (ICCT, 2015)
The overall cost of power-split, P2 and two-motor hybrid systems appear to be comparable.
For example, as estimated by an FEV teardown in 2010,241 the reported power-split hybrid cost
of $2,565242 was only slightly higher than the $2,392 cost estimate for a P2 hybrid system. As
discussed in the Draft TAR, EPA therefore combined all strong hybrid architectures under the
strong hybrid category and continues to do so for this Proposed Determination analysis. Several
public comments received on the Draft TAR addressed this decision to model strong hybrids
with the same cost and effectiveness without regard to specific architecture. These comments, as
well as other comments considered in determining cost and effectiveness for strong hybrid
technology, are addressed in Chapter 2.3.4.3 (Cost and Effectiveness for Strong Hybrids).
2-85
-------
Technology Cost, Effectiveness, and Lead Time Assessment
For the Draft TAR, EPA significantly updated cost and effectiveness estimates for strong
hybrid technology. On consideration of the availability of any significant new information and
consideration of public comments, EPA continues to believe these estimates are appropriate to
use for this Proposed Determination analysis, as discussed in Chapter 2.3.4.3.3.
2.2.4.4.4 Plug-in Hybrids
A plug-in hybrid electric vehicle (PHEV) is much like a hybrid electric vehicle, but with at
least three significant functional differences. The first is the addition of a means to charge the
battery pack from an outside source of electricity (usually the electric grid). Second, a PHEV
has a much larger battery capacity, and often a greater usable fraction as well. Finally, it has a
control system that allows the battery to be significantly depleted during normal operation.
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 providing 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.
Depending on the operating strategy chosen by the manufacturer, a PHEV either provides for
a significant all-electric range (AER) during which the engine does not operate, or provides for
blended operation in which the engine provides some of the propulsion energy while the battery
contributes the remainder. In this discussion, the former is referred to as a PHEV with AER, and
the latter is referred to as a blended PHEV.
EPA models PHEVs in two configurations, designated PHEV20 and PHEV40 (having 20
miles and 40 miles, respectively, of all-electric range or its equivalent). Range is modeled as an
approximate real-world range comparable to an EPA label range (specifically, 70 percent of a
projected two-cycle range).
For GHG analysis purposes, PHEVs are assigned an effectiveness derived from the SAE
J1711 recommended procedure for accounting for utility factor (the balance between miles
traveled on electricity in all-electric mode and other miles powered by fuel). On this basis, in the
2012 FRM and the Draft TAR, PHEV20 was assigned an absolute CO2 effectiveness of 40
percent, and PHEV40 was assigned 63 percent (see 2012 RIA, p. 1-18).
2-86
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In the Draft TAR analysis, the cost-minimizing pathway for compliance with the MY2025
standards projected a very low fleet-level penetration of PHEVs (less than 2 percent). 148'w
At the outset of the rule, only a few PHEVs were commercially available in the U.S. market.
The most prominent examples were the Chevy Volt and the Fisker Karma, both of which
debuted as MY2011 vehicles, and the 2012 Toyota Prius Plug-In Hybrid. Production of the
Karma was discontinued in late 2012 as Fisker encountered financial difficulties. Fisker was
sold to the Chinese company Wanxiang Group, and renamed to Karma, but has not resumed
significant production to date.
Even these early PHEVs demonstrated important differences in their operating strategy that
remain visible in today's market. The Chevy Volt and Fisker Karma both offered a significant
AER by including a distinct charge-depleting mode in its operating strategy. In contrast, the
Toyota Prius Plug-In utilized a more blended mode of operation in which the engine could
regularly operate during the charge depletion stage depending on driving conditions, for
example, if the vehicle exceeded a certain speed or power demand. Both strategies continue to
appear in the market today, with some vehicles emphasizing AER and others emphasizing
overall fuel economy in blended operation. Some PHEVs that employ blended operation are
able to achieve an all-electric range during EPA city and highway test cycles, but may operate in
blended mode (using a combination of gasoline and electricity) when driven more aggressively.
Operation in blended mode may be converted to an equivalent AER by applying a utility factor
that considers the contribution of stored electricity to the total distance traveled in this mode.
Both types of PHEVs are therefore capable of displacing conventionally-fueled mileage with
electrically fueled mileage.
The 2011 Chevy Volt had an EPA-rated AER of 38 miles, while that of the Fisker Karma was
32 miles. The Prius was rated at 6 miles AER (11 miles including blended mode). The market
has since expanded to include many additional products. Table 2.2 shows a summary of PHEV
models that are in current production or have been available during the period since the FRM.
Table 2.2 Trends in EPA-Estimated Range of PHEVs
EPA range (mi)
PHEV model
2012
2013
2014
2015
2016
2017
Chevy Volt
35
38
38
38
53
53
Fisker Karma
33
-
-
-
-
-
Toyota Prius Plug-In Hybrid
11
11
11
11
NL
**
Ford Fusion Energi
-
20
20
20
20
22
Ford C-Max Energi
-
20
20
20
20
**
Honda Accord PHV
-
-
13
-
-
-
McLaren PI
-
-
19
19
-
-
BMW i3 Rex
-
-
72
72
72
97
BMW i8
-
-
15
15
15
15
Cadillac ELR
-
-
37
37
40
**
Cadillac ELR Sport
-
-
-
-
36
**
Porsche Panamera S E-Hybrid
-
-
16
16
16
14
w Because vehicles attributed to the ZEV program were included as part of the EPA reference case, absolute
penetration of PHEVs would be greater.
2-87
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Porsche 918 Spyder
-
-
-
12
-
-
Mercedes-Benz S550e
-
-
-
14
14
**
BMW X5 xDrive40e
-
-
-
NA
14
14
Porsche Cayenne S e-Hybrid
-
-
-
14
14
**
Hyundai Sonata PHEV
-
-
-
-
27
27
Mercedes-Benz C350e
-
-
-
-
18.6*
**
Audi A3 e-tron
-
-
-
-
16
16
Audi A3 e-tron ultra
17
**
BMW 330e
-
-
-
-
14
14
Mercedes-Benz GLE 550e
4MATIC
-
-
-
-
12
Volvo XC90T8 Hybrid
-
-
-
-
14
14
BMW 740e xDrive
-
-
-
-
-
14
Notes:
NL = vehicle not listed in Fuel Economy Guide
NA = rating not available in Fuel Economy Guide
* approximated from press or manufacturer estimate
** Not yet listed in 2017 Fuel Economy Guide at time of writing
The growth in PHEV models as evidenced in Table 2.2 has likely been driven in part by
manufacturers considering PHEVs as part of their pathway for compliance with the 2017-2025
standards, but even more so by California's zero emission vehicle (ZEV) program. In 2012,
CARB adopted increased requirements for ZEVs and PHEVs through MY2025, and nine
additional states have adopted the ZEV program. A 2015 National Academy of Science report
on PEV deployment243 cites the California ZEV regulation as being particularly influential in
increasing PEV production and adoption.
In addition, PHEVs from all manufacturers continue to be eligible for a federal tax credit of
up to $7,500, effectively reducing their net cost to consumers.244' 245 This credit applies to the
first 200,000 PEVs (PHEVs and BEVs combined) that are produced by a given manufacturer and
gradually phases out thereafter. While most manufacturers are unlikely to approach this limit for
at least several years, some of the leading PEV manufacturers such as General Motors, Nissan,
and Tesla are making steady progress toward the limit. For example, if the Gen2 Chevy Volt
sells well, and the recently introduced Chevy Bolt EV does also, it is possible that General
Motors could reach the limit by sometime in 2018. Strong future sales of the Tesla Model X and
Model 3, or the anticipated 200-mile version of the Nissan Leaf, could cause Tesla and Nissan to
approach the limit in a similar time frame.246 Although reaching the limit does not immediately
discontinue the incentive, which would continue to be applicable to additional sales until the
second calendar quarter after it is exceeded, the amount of the credit phases out rapidly over the
following year. However, in addition to federal incentives, many states including California and
the states that have adopted California's ZEV program offer incentives at the state and local
levels.
It is important to note that most PHEVs are built on global platforms, meaning that economies
of scale for the U.S. market may be driven in part by incentives in other countries. Incentives for
PHEVs in the European Union and China are particularly notable because many manufacturers
that serve the U.S. also serve these markets.
Trends in PHEV Electric Range
2-88
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The electric range of a PHEV (either AER or equivalent AER) is largely a function of the
provided battery capacity. Figure 2.31 shows the relationship between the battery capacity of
production PHEVs and their EPA-estimated electric driving range (or the best estimate available
at writing).
40
35
¦£ 30
— 25
>-
S. 20
fO
XJ
P 15
tl
ID
<*> 10
5
0
D ID 20 30 40 50 50 70 80 90 100
EPA estimated electric range (mi)
Figure 2.31 Battery Gross Capacity and Estimated AER or Equivalent for MY2012-2017 PHEVsx
As the Table and Figure shows, PHEV electric range varies considerably among models.
Among the 2012-2016 PHEVs depicted, two distinct clusters appear, one consisting of longer-
range PHEVs with AER in the vicinity of 35 to 40 miles, and another consisting of shorter-range
vehicles offering between 10 and 20 miles of range (either AER or its equivalent in blended
operation). Some longer-range examples are scattered between 53 and 97 miles AER.
The 35-to-40 mile cluster consists of various versions of the Chevy Volt and Cadillac ELR
(which shares the Voltec powertrain), and the discontinued Fisker Karma (at 33 miles). The
longer-range examples consist of the 2016 Volt (at 53 miles) and two versions of the BMW i3
Rex (at 72 miles and 97 miles). These are all PHEVs with AER that can provide a true all-
electric drive mode under a wide range of operation. These PHEVs require a larger battery
capacity than 10-to-20 mile PHEVs, which tends to increase their purchase price relative to the
latter.
The shorter-range cluster includes several blended-operation PHEVs. With the exception of
the Toyota Prius PHV (11 miles) and the Ford Energi models (20 miles), these emerged
primarily in the 2015 and 2016 MYs from OEMs that tend to specialize in luxury or high-
performance vehicles. This suggests that these OEMs are considering PHEVs as a compliance
strategy, as projected in the FRM. For example, when BMW announced the U.S. versions of the
330e and the X5 xDrive40e PHEVs in November 2015, BMW Product Manager Jose Guerrero
was quoted as saying that the timing of introductions such as these "wasn't a competitive impulse
by any manufacturer ... it was an internal impulse that we know that in the future our cars need
x Range figures gathered from 2012-2017 EPA Fuel Economy Guides.
*
•
* *
•
•
•
^ *
»*
¦ •
2-89
-------
Technology Cost, Effectiveness, and Lead Time Assessment
to be more efficient, and this is a way ... into that efficiency."247 The Mitsubishi Outlander
PHEV, expected to enter the U.S. market in 2017248 after several delays,249 is also expected to
have an EPA AER in the neighborhood of 20 miles. These and similar announcements suggest
that a distinct segment of PHEV20-type vehicles is likely to continue in the future as
manufacturers continue to select this lower cost pathway.
Where new generations of the same model have been announced, the range has in some cases
been increased. For example, the AER of the Chevy Volt has increased from 38 miles to 53
miles. Going forward, several OEMs have indicated that second generation PHEV products will
have more AER and more electric power capability, by targeting US06 capability, with minimal
if any reliance on the engine and 30 miles or more of AER. For example, the FCA Pacifica plug-
in minivan was announced in January 2016 as targeting a 30 mile all-electric range, with
capability to operate all-electric over most operating conditions.250 Honda is reported to be
considering a 40 mile AER for an upcoming PHEV that would replace the now-discontinued
Honda Accord PHV, which had an AER of only 13 miles.251 Similarly, other manufacturers
including Toyota, GM, and Ford have suggested that their 2017 to 2018 PHEV products will be
targeting at least 30 miles of electric range.
In such announcements, manufacturers have frequently cited customer desire for an all-
electric driving experience. As one example, GM appears to credit consumer demand for more
range as part of the impetus for increasing the range of the 2016 Volt. According to Chief
Engineer Andrew Farah, "We listened to our customers ... they were very clear when they told
us that they wanted more range."252 These manufacturers appear to be responding by increasing
the potential for all-electric operation by increasing electric powertrain power ratings and battery
capacity.
The California Zero-Emission Vehicle (ZEV) program also may be influencing PHEV range.
To qualify as transitional-zero emission vehicles (TZEVs) under the program, PHEVs must
provide at least 10 miles of AER operation on the UDDS drive schedule (as well as meet certain
criteria pollutant standards) 253 Since many PHEV manufacturers market in ZEV states as well as
other states, the ZEV program provides a strong incentive for producing PHEVs with AERs
above this threshold.
Other incentive programs may be encouraging longer PHEV electric range. One example is
the China New Energy Vehicles Program.254 Renewal of this program in 2013 increased the
eligibility requirements for PHEVs to a minimum 50 km (30 mile) AER (under the NEDC cycle)
in order to qualify for purchase subsidies.255 There is some evidence that this may be
encouraging manufacturers of global-market PHEVs to increase AER to at least this level.256
For example, the Cadillac CT6 PHEV was announced in April 2015 at the Shanghai Auto Show,
where it was described as qualifying for the New Energy Vehicles incentives with a range in
excess of 60 km (37 miles). 257 The U.S. version will have the same 18.4 kWh battery pack as
the China version, suggesting that its AER will be similar. As of July 2016, at least one local
U.S. incentive in the state of Washington will also adopt a 30-mile PHEV range requirement to
qualify for a sales tax exemption up to $3,100.258
Manufacturers have continued to pursue and implement improvements in the efficiency and
cost of battery and non-battery components for PHEVs. One example is the 2016 Chevy Volt, in
which the weight of the battery pack was reduced by 14 kg despite an increase in its capacity
from 17.1 kWh to 18.4 kWh. The weight of the traction motor was also reduced by 45 kg, and
2-90
-------
Technology Cost, Effectiveness, and Lead Time Assessment
additional weight and cost were saved by integrating the inverter with the motor and eliminating
long runs of high voltage electrical cable.162'163
Improvements in component efficiency and road load have both improved performance of
production PHEVs. For example, GM has indicated that the 2016 Chevy Volt improved its
average electric powertrain efficiency over the EPA city and highway cycles by 3 percentage
points (or 4 percent absolute) compared to the first generation Volt, improving from 86 percent
to 89 percent for the city, and from 84 percent to 87 percent for the highway. Drive unit losses
(including losses of the electric motor, inverter, and transmission) were reduced by 39 percent in
the city cycle and by 35 percent in the highway cycle.259 The Gen2 Volt also provides a good
example of the use of standard road load improvements to increase range in a PHEV.178 Here,
significant changes to the electric propulsion system were accompanied by improvements in
brake drag, reductions in accessory load, and significant improvement of vehicle mass
efficiency.
In both the 2012 FRM analysis and the Draft TAR analysis, EPA envisioned PHEV20 and
PHEV40 as representative of PHEVs that are likely to play a significant role in achieving fleet
compliance during the time frame of the rule. As Figure 2.31 shows, PHEV20 continues to be
represented in the market by several 20-mile and shorter range PHEVs. PHEV40 is also
represented by several vehicles, primarily earlier versions of the Chevy Volt and Cadillac ELR.
PHEV40 has also been surpassed in real-world range by the 2016 Chevy Volt at 53 miles, and by
the BMW i3, which with its range extender option becomes classified as a PHEV with either 72
or 97 miles AER, depending on configuration.
As discussed in the Draft TAR, EPA considered replacing PHEV40 with a longer range, such
as PHEV50, but ultimately decided not to do so based on an examination of PHEVs in the
market. Although the 2016 Chevy Volt has now exceeded PHEV40, other production PHEVs
such as the Cadillac ELR and CT6 continue to fall on the lower side of this line. The BMW i3
examples at 72 and 97 miles fall far beyond PHEV40 but at this time are not accompanied by
other examples that would suggest a wider trend toward increasingly long PHEV ranges. The i3
design is also unique in having a particularly small gasoline-only range, motivated at least in part
by California regulations that apply to gasoline-powered range in PHEVs. At this time, EPA
believes that PHEV20 and PHEV40 continue to serve as appropriate modeling constructs for the
Proposed Determination analysis.
Trends in PHEV Motor Sizing
In addition to driving range, the electric motor power of PHEVs is another important input to
the projection of battery and system costs for PHEVs. Accurately assigning motor power is
important on several fronts. First, the motor power rating has a direct effect on the battery power
rating, which determines its power-to-energy (P/E) ratio and its cost. Second, the EPA battery
sizing methodology accounts for the weight of the propulsion motor and power electronics as a
function of rated motor power. An accurate determination of motor power rating is therefore
quite critical. An accurate accounting of motor cost also requires an accurate accounting of
motor power because EPA estimates PHEV motor cost as a function of peak power output.Y
Y For more discussion of the decision to scale motor cost to power output, see Chapter 2.3.4.3.6 (Cost of Non-
Battery Components for xEVs).
2-91
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In the Draft TAR analysis, a significant change was made to the way motor power for PHEVs
was originally assigned in the 2012 FRM. Originally in the FRM analysis, PHEVs of a given
vehicle class (Small car, Large car, etc.) were assigned an electric motor power rating (in kW)
that would preserve the same engine-power-to-weight ratio that was observed in baseline
conventional vehicles of that class. This method assumed that the all-electric acceleration of
PHEVs relates to the power rating of the electric motor in the same way that the engine-powered
acceleration of conventional vehicles relates to the power rating of the engine. However, as
discussed in the Draft TAR, electric motors differ markedly from combustion engines,
particularly in their delivery of low-speed torque. Electric motors deliver maximum torque at the
lowest end of their speed range, while combustion engines must develop significant speed to
deliver a comparable torque. This strong low-end torque allows electric-drive vehicles to deliver
high acceleration at low speeds. This might allow a PHEV or BEV to deliver acceleration
performance similar to that of a conventional vehicle but with a significantly lower nominal
motor power rating than a comparably performing combustion engine. A new sizing method,
based on an empirical survey of PHEV performance, was therefore developed and described in
the Draft TAR analysis.
As discussed in the Draft TAR, a number of production PHEVs have now been offered on the
market, providing a significant sample size to allow some observations to be drawn regarding the
necessary motor power to provide customary performance. Accordingly, the Draft TAR found
that the 2012 FRM did in fact project significantly higher PHEV motor power ratings than the
majority of PHEV manufacturers subsequently specified in their MY2012-2016 products. Part
of this effect was attributed to the significant presence of blended-operation PHEVs in the
market, which do not require as large a motor power output as the non-blended PHEV20s that
were modeled for the 2012 FRM analysis. However, the Draft TAR noted that this alone would
not account for the difference because many of the 2012 FRM estimates also over predicted the
motor power of non-blended PHEV40s with AER.
Accordingly, EPA significantly revised its PHEV motor power ratings for the Draft TAR
analysis. PHEV20 was modeled under a blended-operation architecture which significantly
reduced nominal power ratings, which were assigned at 50 percent of the total rated power of the
vehicle. For non-blended PHEV40, an empirical equation was derived based on the relationship
between 0-60 mi/hr acceleration time and electric motor power observed in MY2012-2016
PEVs.
Assigning a more accurate power rating to the PHEV motor provides for greater fidelity in the
projected cost of both the battery and non-battery components of PHEVs. More detail on the
way PHEV battery and non-battery components were sized in the Draft TAR and revised for this
Proposed Determination analysis are discussed in Chapters 2.2.4.4.6 (Relating Power to
Acceleration Performance) and 2.3.4.3.7 (Cost of Batteries for xEVs).
Trends in PHEV Battery Sizing
Accurately assigning battery capacity to PHEVs is also important. To assess the fidelity of the
EPA battery sizing methodology, the Draft TAR compared the 2012 FRM projections of PHEV
battery capacity and range to the PHEVs that entered the market during MYs 2012-2016.
Figure 2.32 compares the battery capacities of MY2012-2016 PHEVs to the battery capacities
that were estimated for the Draft TAR analysis.
2-92
-------
Technology Cost, Effectiveness, and Lead Time Assessment
30
25
£ • 0 •
I 20 1 K
X • •
£ 15 O draft TAR
ro
£ @ g • MY 2012-2016 PHEVs
V I" *1 T Trer»dline
® •iprx
3 . r
0
0 10 20 30 40 50 60 70 SO
EPA estimated electric range (mi)
Figure 2.32 Comparison of MY2012-2016 PHEV Battery Capacities to Draft TAR Estimates
For each PHEV range (20 and 40 miles), the Figure shows the battery capacity estimates
generated for the Draft TAR, corresponding to each of the vehicle classes (Small Car, Standard
Car, Large Car, etc.) and several target curb weight reductions (ranging from 0 percent to 20
percent).
It can be seen from the plot that the Draft TAR estimates lined up quite well with the
population of production vehicles of a similar range. This represented a significant improvement
over the 2012 FRM projections, which had significantly overestimated capacities. As discussed
in the Draft TAR, the improvement was a result of updating many of the parameters that are
influential to the estimation of battery capacity, as described in Chapter 5.3 of the Draft TAR.
This Proposed Determination analysis makes additional adjustments to the PHEV battery sizing
methodology which are discussed in Chapter 2.3.4.3.7.
2.2.4.4.5 Battery Electric Vehicles
Battery electric vehicles (BEVs) are vehicles with all-electric drive powered by batteries
charged from an outside source of electricity (usually the electric grid). The 2012 FRM analysis
modeled three BEV configurations, designated BEV75, BEV100 and BEV150 (having 75, 100,
and 150 miles range, respectively).Z AA BEV150 was updated to BEV200 for the Draft TAR
. t
-f- .
* I y"
•
•
* a/
a
* «o
• •
z As with PHEVs, the indicated range was meant to represent an approximate real-world range comparable to an
EPA label range (specifically, 70 percent of a projected two-cycle range).
AA In the 2012 FRM and the Draft TAR, BEV75/100/200 were referred to as EV75/100/200.
2-93
-------
Technology Cost, Effectiveness, and Lead Time Assessment
analysis. Both the 2012 FRM and Draft TAR analyses predicted a very low fleet-level
penetration of BEVs at about 2 percent or less.260'BB
As the Draft TAR found, the BEV market has grown considerably since the time of the 2012
FRM. At that time, only a few BEV models had become commercially available in the U.S.
market. The most prominent examples were the 2011-12 Nissan Leaf and the Tesla Roadster,
which were available nationwide. A few other BEVs were available in 2012 to very limited
markets or through demonstration programs, such as the BMW Mini E and Toyota RAV4 EV.
Production of the Tesla Roadster was discontinued in early 2012 but was soon replaced by the
Tesla Model S. Other BEVs available near the time of the 2012 FRM were the Mitsubishi i-
MiEV, BYD e6, Coda Sedan, and Ford Focus Electric.
These early BEVs were designed for different market segments, and showed significantly
different philosophies on the matters of performance and driving range. Most, such as the Leaf
and Mini E, were designed as moderate-performance vehicles with a driving range of 100 miles
or less, seen as best suited to driving in urban areas. In contrast, the Tesla Roadster was designed
for a premium, high-performance market segment at a much higher price, allowing it to offer a
much longer range (245 miles by EPA estimate). Subsequent Tesla vehicles have continued to
pursue similarly aggressive range and performance targets at relatively high purchase prices,
while several other manufacturers continue to define a distinct segment targeting shorter ranges
and moderate performance at lower purchase prices. The Draft TAR concluded that these two
segments would likely continue to exist within the time frame of the rule.261'262
The current BEV market includes a wide variety of models either currently in production or
announced for future production. Table 2.3 shows a summary of BEV models that have reached
production since the 2012 FRM, and their EPA estimated range.
Table 2.3 Driving Range of MY2012-2017 BEVs
EPA range (mi)
BEV model
2012
2013
2014
2015
2016
2017
Azure Dynamics Transit
Connect
56
-
-
-
-
-
Coda
88
88
-
-
-
-
BYD e6
122
127
127
127
187
**
Toyota RAV4 EV
103
103
103
-
-
-
Mitsubishi i-MiEV
62
62
62
NL
62
59
Ford Focus Electric
76
76
76
76
76
**
Tesla Model S (85 kWh)
265
265
265
265
265
**
Nissan Leaf (24 kWh)
73
75
84
84
84
-
Tesla Model S (40 kWh)
-
139
-
-
-
-
Tesla Model S (60 kWh)
-
208
208
208
210
**
Scion iQ EV
-
38
-
-
-
-
Honda Fit EV
-
82
82
-
-
-
Smart fortwo
-
68
68
68
68
**
Fiat 500e
-
87
87
87
84
84
BB Penetration driven solely by the GHG standards, since vehicles attributed to the ZEV program were included as
part of the EPA reference case. Absolute penetration of BEVs (counting those attributed to the ZEV program)
was projected at less than 3 percent.
2-94
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Kia Soul EV
-
-
-
93
93
93
BMW i3 BEV
-
-
81
81
81
81
Chevy Spark EV
-
-
82
82
82
**
Volkswagen e-golf
-
-
NA
83
83
**
Mercedes-Benz B250e
-
-
87
87
87
87
Tesla Model S (70 kWh)
-
-
-
-
234
-
Tesla Model S 70D
-
-
-
240
240
-
Tesla Model S 85D
-
-
-
270
270
-
Tesla Model S P85D
253
253
-
Tesla Model S (90 kWh)
-
-
-
265*
265*
-
Tesla Model S 90D
-
-
-
270*
294
**
Tesla Model S P90D
-
-
-
253*
270
-
Tesla Model X90D
-
-
-
NA
257
**
Tesla Model X P90D
-
-
-
-
250
-
Tesla Model X60D
-
-
-
-
200
-
Tesla Model X75D
-
-
-
-
238
**
Tesla Model S 75
-
-
-
-
249
**
Tesla Model S 75D
-
-
-
-
259
**
Tesla Model S P100D
-
-
-
-
315
**
Nissan Leaf (30 kWh)
-
-
-
-
107
**
Chevy Bolt EV
-
-
-
-
-
238
BMW i3 94 Ah
-
-
-
-
-
114
Notes:
NL = vehicle not listed in Fuel Economy Guide
NA = vehicle listed but rating not available in Fuel Economy Guide
* Manufacturer applied 85 kWh EPA range figure for EPA labeling purposes
** Not yet listed in 2017 Fuel Economy Guide at time of writing
The growth in the number of BEV models has likely been encouraged in part by several
factors, both regulatory and demand driven.
Among the regulatory factors, the 2017-2025 rule assigns a high GHG effectiveness to BEVs,
further enhanced by assigning 0 g/mi for upstream emissions and a multiplier for the earlier
years of the rule. Some manufacturers are therefore including BEVs as part of their pathway for
compliance with the 2017-2025 standards. Production of BEVs also generates GHG credits that
may be used for future regulatory compliance (credit carryforward) or sold to other
manufacturers. Production of BEVs can also assist manufacturers in meeting fleet average
criteria pollutant regulations such as EPA's Tier 2 and Tier 3 standards or CARB's LEV II and
LEV III standards. And, just as with PHEVs, California's ZEV regulation continues to drive
BEV production to generate ZEV credits as manufacturers prepare for ever increasing
requirements through MY2025.
In addition, BEVs from all manufacturers continue to be eligible for a federal tax credit of up
to $7,500, effectively reducing their net cost to consumers.244'245 Because this credit applies to
the first 200,000 eligible vehicles (BEVs and PHEVs combined) produced by a given
manufacturer, it continues to influence the BEV market today. However, at current rates of
production, it is possible that some manufacturers may begin approaching the 200,000 limit by
2018, with others following soon after.246 Although reaching the limit does not immediately
discontinue the incentive, which would continue to be applicable to additional sales until the
2-95
-------
Technology Cost, Effectiveness, and Lead Time Assessment
second calendar quarter after it is exceeded, the amount of the credit phases out rapidly over the
following year.
In addition to the federal tax credit, many states, including California and many of the states
that have adopted California's ZEV regulation offer incentives for ZEVs at the state and local
levels. These programs may supplement the federal program and have varying phase-out
schedules and eligibility requirements.
Demand for BEVs has also been a factor in their growth. The demand for premium BEVs,
such as those produced by Tesla Motors, has accounted for a significant portion of BEV sales
despite their relatively high purchase price. These vehicles compete in a market segment with
other high-priced vehicles and are seeing success in that segment. For example, Tesla claims
that the Model S outsold all other conventional vehicles in its market segment in 2015.263 If the
performance attributes that are attracting this segment of buyers away from the conventional
competitors in this space can be sufficiently retained at a lower price point, this could further
drive demand for BEVs in the future. Projections for the 2017 Chevy Bolt are similarly driven by
expectations of significant consumer demand.264'265 Tesla cites over 373,000 reservations for its
entry-level Model 3 as further evidence of consumer market demand for BEVs.263 Some have
even suggested that the Tesla Model 3 and the Chevy Bolt may be "breakthrough" vehicles that
will open a gateway to greatly increased demand for BEVs among mainstream auto buyers.266
Demand for BEVs is also likely to grow in the future as consumers become more familiar
with the technology. In comments on the Draft TAR, the Consumer Federation of America
(CFA) cited two surveys, one reported by the Alliance of Auto Manufacturers and another
performed by CFA,267 that indicate that knowledge about BEVs is an important factor in the
willingness of car buyers to consider BEVs, further stating, "the more Americans know about
EVs, the more likely they are to consider this purchase." The CFA survey also found that "only a
little over a quarter of respondents say they know a great deal (6 percent) or a fair amount (21
percent) about EVs," suggesting that consumer knowledge about BEVs has significant room to
grow.
Another potential vector for growth in BEVs could develop from the recent boom in
autonomous vehicle research by OEMs (such as GM, Ford and Tesla, among others) and tech
companies such as Google. Increasingly, these efforts are being united with other mobility
models such as ride sharing (for example, the partnership between GM and Lyft,268'269 and efforts
in vehicle autonomy by Uber).270 Some have made the case that electric vehicles may be the
preferred technology for autonomous applications and ride sharing models,271 which if proven
true, could act as another significant driver for BEV growth in the future.
BEVs continue to be offered at a significant price premium to conventional vehicles, largely
due to the cost of the battery, as well as non-battery components that have yet to reach high
production volumes. Some BEVs, particularly those targeted primarily for sale in the ZEV
states, are available for purchase only in those states.
BEV production levels have grown significantly since the 2012 FRM. Through October 2016,
Nissan had sold about 100,000 Leaf EVs, and GM had sold about 117,000 Volt PHEVs, Cadillac
ELRs and Spark EVs combined.272 Analysts have widely speculated that a slight decline in PEV
sales in MY2015 (relative MY2014) was due at least in part to anticipation of new models with
longer range and enhanced features. For example, expectations of a refreshed version of both the
2-96
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2016 Volt and 2016 Leaf existed long before either became available. The 2016 Leaf offers a
larger 30kWh pack, increasing range significantly, while the 2016 Volt also offers a longer
range, better fuel economy and other enhancements such as improved seating.
Charging infrastructure, both at home and in public places, is a topic that is often associated
with BEVs. Public charging infrastructure was explored in depth in Draft TAR Chapter 9
(Infrastructure Assessment), and is reviewed for this Proposed Determination assessment in
Section B.3.2 of the Proposed Determination Appendix, where public comments received on the
topic of charging infrastructure are addressed.
Since 2008, various ongoing public and private efforts to provide charging stations at
workplaces, along freeway corridors, and in cities have grown the number of public stations in
the U.S. to more than 16,000.150 As mentioned in Proposed Determination Appendix B.3.2, some
public comments on the Draft TAR expressed concern that infrastructure is not growing fast
enough even at this pace. Mercedes-Benz commented that "infrastructure investments are not
meeting expectations," while Global Automakers commented, "infrastructure is not developing
as quickly as needed to support electric drive vehicles."
In addition to the consideration of these comments found in Appendix B.3.2, it is also relevant
to note that since the Draft TAR was completed, two developments were announced that may
increase the availability of public charging substantially. The partial settlement between
Volkswagen and U.S. authorities, approved in 2016, earmarks $1.2 billion in investment over 10
years toward ZEV infrastructure, education, and access.273 Also, in November 2016, the White
House announced a network of federal, state, and local initiatives to increase accessibility to
PEV infrastructure,150 including a Department of Transportation (DOT) plan to designate 48
national "alternative fuel corridors" along major highways to provide focus for build out of
charging locations by related local and state efforts.274
Also as discussed in Appendix B.3.2, comments from the Alliance disagreed with some of the
discussion in Draft TAR Chapter 9 (Infrastructure Assessment), including the discussion of the
roles and availability of home and public charging, a supposed assumption that BEV users would
rely on Level 1 charging at home, and the suggestion that public infrastructure was developing as
required to support the penetration levels of PEVs projected in the Draft TAR. In addition to the
comments provided in B.3.2, it should be noted that Chapter 9 of the Draft TAR was provided
primarily as background on charging infrastructure, and the assumptions found in that discussion
are specific to the assessment presented in that discussion. Costs used in the Draft TAR and
Proposed Determination analyses for home charging infrastructure were developed
independently of the Chapter 9 assessment, and include significant costs for installation of home
charging capability for all PEVs. Specifically, all home charging installations are assumed to
incur a significant cost for installation labor, plus an additional cost for Level 1 or Level 2
charging hardware, depending on the vehicle type. These costs are outlined in more detail in
Chapter 2.3.4.3.6 (Cost of Non-Battery Components for xEVs) of this TSD. Further, EPA did not
assume that only Level 1 charging will be used. While PHEV20 and some PHEV40 vehicles are
assigned a blend of Level 1 and Level 2 charging, all BEVs and larger PHEV are assigned 100
percent Level 2 charging. With the availability of Level 2 charging at home therefore being
largely assumed and provided for in EPA's cost assumptions, the importance of public charging
availability to support the projected penetration of BEVs is minimized. EPA also notes the recent
charging infrastructure developments cited above, as well as recent additions of hundreds of
2-97
-------
Technology Cost, Effectiveness, and Lead Time Assessment
public charging points by several OEMs (including Nissan, BMW, and Volkswagen),275 which
suggest that development of public charging infrastructure continues to proceed at a significant
pace.
Trends in BEY Driving Range
Continuing growth in the BEV market has greatly expanded not only the available choice of
vehicle models and trims, but also the available driving ranges. BEV driving range is largely a
function of battery capacity. Figure 2.33 shows the relationship between the battery capacity of
the MY2012-2017 BEVs in Table 2.3 and their EPA estimated driving range.
110
100
90
| BO
70
y 60
n?
8" 50
u
t AO
tu
£ 30
in
20
10
0
0 50 1D0 150 200 250 300 350
EPA estimated range (mi)
Figure 2.33 Battery Gross Capacity and EPA Estimated Range for MY2012-2017 BEVscc
It has become apparent since the 2012 FRM that manufacturers have been pursuing increased
driving range. Several examples serve to illustrate this trend. The Nissan Leaf was introduced in
2011 with an EPA-rated range of 73 miles. The 2013 model increased this to 75 miles, while
2014 and later models earned a higher rating of 84 miles by eliminating a partial charge option,
allowing the range to be evaluated at 100 percent charge. This trend indicates that Nissan
perceives increased range as a desirable goal. As another example, in January 2016, it was
reported that the range of the BMW i3 might increase by about 50 percent due to improved
battery chemistry and electronics;276 by May 2016, BMW confirmed the increase in capacity,
resulting in a new range of approximately 114 miles.277 In January 2016, Volkswagen also
indicated that a new version of the e-Golf could expect a possible 30 percent increase in range
over the current model (or about 108 miles) due to an increase in cell capacity from 28 A-hr to
37 A-hr.278 The 2017 Ford Focus BEV is also expected to increase its range to over 100 miles
compared to its original range of 76 miles.279 In November 2016, the 2017 Hyundai Ioniq
BEV280 was certified by EPA with a range of 124 miles.
cc Range figures gathered from 2012-2017 EPA Fuel Economy Guides.
2-98
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Future vehicles expected to enter the consumer market soon have increasingly targeted even
longer ranges. In addition to the 2017 Chevy Bolt, which recently certified for a range of 238
miles, a future version of the Nissan Leaf has been described by Nissan as targeting a 200 mile
range. The Tesla Model 3 is described as offering a 215 mile range and entering production in
late 2017.281 Ford has also announced intent to introduce a 200-mile competitor, possibly called
the Model E, before 2020.282 Similar announcements have been made by Volkswagen283 and
Audi284 among others. In November 2016 it was reported146'147 that Toyota is planning to
produce BEVs with a range of more than 300 km (186 mi) by 2020.
A trend toward increased range also seems to be playing out across manufacturers, as new
products are introduced to compete in the market. For example, the Kia Soul EV was introduced
in 2014 with a range of 93 miles, surpassing the Leaf. Not long after in 2015, Nissan announced
the 2016 Nissan Leaf, offering an EPA range of 107 miles with a new 30 kWh battery pack. In
late 2016, General Motors announced that the 2017 Chevy Bolt was certified for a range of 238
miles, significantly greater than the rumored 215 mile range of the upcoming Tesla Model 3 with
which it will directly compete.
Even Tesla Motors, which already offers a range in excess of 200 miles in all of its current
vehicles, has shown an interest in increased range as evidenced by regular increases in battery
capacity. After announcing in 2012 that the Tesla Model S would be available in three battery
sizes (40 kWh, 60 kWh, and 85 kWh), the 40 kWh version was canceled in 2013, prior to its
production. In April 2015, the battery capacity of the 60 kWh version was increased to 70 kWh,
which along with powertrain improvements increased its range from 208 miles to 240 miles. In
September of the same year the 85 kWh version was increased in capacity to 90 kWh by use of
an improved chemistry.DD This was followed by another increase to 100 kWh, which increased
the EPA estimated range to 315 miles.285 According to an informal statement attributed to Tesla
CEO El on Musk, 100 kWh may be the maximum capacity that will be offered for the Model
S.286 Tesla also announced in 2015 an available battery upgrade for the discontinued Roadster
that would increase its range by about 40 percent.287
Manufacturers have frequently cited customer demand in the quest for increased range. When
the 40 kWh Model S was canceled, Tesla attributed the decision to low demand, further saying,
"Customers are voting with their wallet that they want a car that gives them the freedom to travel
long distances when needed."288 Although this statement clearly promotes Tesla's market
strategy of offering a longer driving range than other BEV-manufacturing OEMs, similar
sentiment has been expressed by other OEMs in marketing their electrified vehicles or
announcing future plans. Customer demand for an affordable BEV with a longer driving range
than currently available is implicit in the 200-mile range target of both the future Nissan Leaf
and the 2017 Chevy Bolt.
As a way of increasing range, simply increasing the battery capacity in the absence of other
improvements may be prohibitive because it increases the cost of the battery accordingly. On the
other hand, improved battery manufacturing or battery chemistry (in terms of cost or energy
density) might enable a larger capacity while offsetting some of the cost penalty of a larger
battery. For example, both Tesla and Nissan have utilized improved chemistry to increase
DD The manufacturer chose to apply the 85 kWh EPA range figure to the 90 kWh version for EPA labeling purposes.
Marketing materials attribute an additional 6% range to the 90 kWh version.
2-99
-------
Technology Cost, Effectiveness, and Lead Time Assessment
capacity within the existing footprint of their respective packs; while GM and Nissan have hinted
strongly at improved chemistry being the enabler of the affordable 200-mile range target for the
Bolt and future Leaf. These and other examples are discussed in more detail in Chapter 2.2.4.5.1
(Battery Chemistry).
Increasing the usable capacity (i.e. widening the usable state-of-charge window) of the battery
may be another route for increasing range; for example, by use of an improved chemistry, or by
acting on experience that indicates that the existing buffer capacity may be reduced.
Improvements in battery management systems (BMS) may also lead to greater utilization of the
available battery capacity. Examples of OEM activity in this area are reviewed in more detail in
Chapter 2.2.4.5.3 (Usable Energy Capacity).
Range can also be increased by reducing vehicle energy consumption. This can be done by
improving the energy efficiency or weight of non-battery powertrain components (electric
machines and power electronics) or even the battery itself. For example, the dual motor versions
of the Tesla Model S achieve a slightly higher range than the single motor versions due to an
improved powertrain efficiency resulting from the ability to selectively operate one or both
motors as conditions warrant. Range may also benefit from standard road load improvements
such as light-weighting, improved aerodynamics, and lower rolling resistance.
In addition to increased range, a larger battery may carry other ancillary benefits for
manufacturers and consumers. Because a large battery stores more energy per charge cycle than
a small battery, it is likely to experience fewer charge-discharge cycles in the course of providing
a given number of vehicle miles. For example, a battery that provides for a range of 200 miles
can provide a lifetime mileage of 150,000 miles with about 750 charge-discharge cycles, while a
100-mile battery may require 1,500 cycles. The smaller number of expected cycles may promote
a longer battery lifetime or relax manufacturer provisions for battery durability, such as
increasing the permissible charge rate or the usable capacity. A larger battery might also
experience a much shallower average state-of-charge (SOC) swing in the course of meeting its
mileage target, with similar implications for durability. Another advantage of a large battery is a
reduction in average discharge rate (C-rate), which can allow consideration of chemistries and
configurations that would not be suited to smaller batteries. For example, Tesla may have
selected a chemistry that supports notably low C-rates in recognition that the large size of the
battery acts to minimize per-cell power requirements.289 Of course, a drawback of a larger
battery over a smaller battery is its greater weight, which tends to reduce the overall energy
efficiency of the vehicle.
In the same way that cabin air conditioning can have a significant impact on fuel economy of
conventional vehicles,290 both heating and air conditioning can have a strong impact on BEV
energy efficiency and range. While the impact of passenger comfort on range can be great for
both BEVs and PHEVs, BEVs are at a particular disadvantage because all energy for heating and
cooling must come from the battery. In contrast, PHEVs may choose to operate the engine if
needed (for example, the Chevy Volt operates the engine to help with cabin heating in cold
weather). Cabin heating and cooling for BEVs is therefore an active area of research toward
increasing BEV range.291'292
Some BEVs, such as the Nissan Leaf, have employed heat pump-based HVAC in place of
resistive heating. When the temperature differential between the outside air and the desired
cabin temperature is not too large, this method can be much more efficient than resistive heating
2-100
-------
Technology Cost, Effectiveness, and Lead Time Assessment
at controlling cabin temperature. Another approach to passenger comfort that has been used for
BEVs and PHEVs involves heated and cooled surfaces, for example, the steering wheel and
seats, instead of or in addition to heating the cabin air, which one study has shown can reduce
cooling and heating energy in a PHEV by about 35 percent.293 Pre-conditioning the passenger
cabin while plugged in to a charging station is yet another approach, which can reduce the use of
onboard energy for heating and cooling (although it does consume energy at the station).
Modeled BEY Ranges in the Draft TAR and this Proposed Determination
As noted in the Draft TAR, the EPA analysis models three BEV range configurations
(BEV75, BEV100 and BEV200). As previously noted, the Draft TAR adopted BEV200 in place
of BEV150 due to several market developments since the 2012 FRM. Tesla vehicles with a range
well in excess of 200 miles are growing in production rates and market share as well as range.
Although these vehicles currently constitute a premium segment that may not be fully
representative of a mass-market vehicle, their success at achieving significant market penetration
shows that at least one OEM has found it preferable to comply with the 2017-2025MY standards
and generate additional GHG credits by producing long-range BEVs. Announcements from
Nissan, GM, and several other OEMs target a 200-mile BEV range, suggesting that BEV200
may become prevalent in the future BEV market.
In the public comments to the Draft TAR, Volkswagen voiced a concern that over the longer
term, BEV200 may not provide a long enough driving range to compete with conventional
vehicles, and suggested that EPA consider adding an even longer range vehicle, which would
have an accordingly higher cost than BEV200 due to having a larger battery.
EPA acknowledges that BEV200 represents a shorter range than seen in many current
premium segment vehicles with well over 200 miles range, and that over time the consumer
market may increasingly exceed BEV200 in order to compete with conventional vehicles. But
despite the announcement of the Chevy Bolt at 238 miles range, announcements of other near-
term future BEVs continue to target a range closer to BEV200. For example, Ford has
announced intent to introduce a BEV, described as having an approximately 200-mile range,
before 2020.294 It has also been reported146'147 that Toyota is planning to produce BEVs with a
range of "more than 300 km" (or 186 mi) by 2020. Similarly, it continues to appear that Nissan is
likely to be targeting a 200-mile real-world range with a future version of the Leaf.295 Tesla has
suggested that the Model 3 will be available with at least 215 miles of range, which also is not
far from BEV200. Of course, although Tesla may choose to increase the Model 3's range to
compete with the Bolt, this is still uncertain. It remains unclear whether the market will coalesce
around longer range vehicles at a somewhat higher cost, or settle at a lower range with a lower
cost. Further, to the extent that manufacturers pursue future range increases by taking advantage
of ongoing reductions in battery cost per kWh, the total cost of the battery could remain
relatively constant even as range gradually exceeds BEV200.
Compared to BEV75 or BEV100, there may be limited potential for BEV200 to be selected
by OMEGA as part of a cost-effective compliance path, because the relatively high cost of the
larger battery is likely to overshadow any gain in effectiveness. That is, since BEV75, BEV100,
and BEV200 are all assigned a GHG effectiveness of 100 percent (when upstream emissions are
assessed at 0 grams per mile), the incremental cost of BEV200 vs. BEV75 or BEV100 strongly
discourages its selection on a pure cost-effectiveness basis. Although this effect is reduced in this
Proposed Determination analysis because the compliance model now phases-in an accounting for
2-101
-------
Technology Cost, Effectiveness, and Lead Time Assessment
upstream emissions for PEVs between 2021 and 2025, it still has some influence. Due to the
structure of the OMEGA model and the low potential for even BEV200 to be selected on a pure
cost-effectiveness basis, EPA is currently choosing to remain with BEV200 as a modeling
construct. (See also the discussion of public comments relating to BEVs in Chapter 2.3.4.3.5).
As discussed in Draft TAR Chapter 5.3, EPA updated assumptions for many of the xEV
parameters that affect battery sizing for the Draft TAR analysis. In Chapter 2.3 of this TSD,
EPA further updates certain assumptions for the Proposed Determination analysis, as suggested
by updated information and public comment on the Draft TAR. These include assumptions for
usable capacity, electric powertrain efficiencies, and power ratings of electric machines and
power electronics. EPA is also updating the assumptions for road loads as they affect battery
sizing for BEVs. For further details on these changes, see Chapter 2.3.4.3.7 (Cost of Batteries for
xEVs).
Trends in BEY Motor Sizing
In addition to driving range, the motor power of BEVs is another important input to EPA's
projection of battery and system costs for BEVs. As discussed previously with respect to
PHEVs, the 2012 FRM analysis had assigned BEVs of a given vehicle class a motor power
rating that would preserve the same engine-power-to-weight ratio observed in conventional
vehicles of that class. The Draft TAR found that this method overestimated the rated peak motor
power necessary to achieve a given acceleration performance. The Draft TAR developed an
improved methodology that more accurately assigned motor power specifications.
As previously discussed in relation to PHEVs, accurately assigning the motor power of a BEV
is important for several reasons. First, the motor power rating has a direct effect on the battery
power rating, which determines its power-to-energy (P/E) ratio and its cost. Second, EPA
accounts for the weight of the electric motor and power electronics as a function of power.
Finally, an accounting of motor cost requires an accounting of motor power. As in both the 2012
FRM and Draft TAR analyses, for this Proposed Determination analysis EPA estimates electric
motor and power electronics costs as a function of peak output power, in accordance with several
examples of similar industry practice.EE
As with PHEVs (discussed in the previous section), the Draft TAR found that the FRM
analysis tended to assign significantly higher BEV motor power ratings than the majority of BEV
manufacturers subsequently found necessary to provide in their MY2012-2016 vehicles.
Accordingly, in the Draft TAR, EPA revised the BEV motor power ratings to be closer to those
suggested by the power-to-weight ratios that BEV manufacturers appear to be following, while
maintaining an estimated acceleration performance equivalent to conventional vehicles.
Assigning a more accurate power rating provided greater fidelity in the projected cost of both
the battery and non-battery components of BEVs. More detail on the way BEV battery and non-
battery components were sized in the Draft TAR and revised for this Proposed Determination
analysis are discussed in Chapters 2.2.4.4.6 (Relating Power to Acceleration Performance) and
2.3.4.3.7 (Cost of Batteries for xEVs).
EE For more discussion of the decision to scale motor cost to power output, see Draft TAR Section 5.3.4.3.6, Cost of
Non-Battery Components for xEVs.
2-102
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Trends in BEY Battery Sizing
To assess the fidelity of the EPA battery sizing methodology, the Draft TAR compared the
2012 FRM projections of BEV battery capacity and range to the BEVs that entered the market
during MYs 2012-2016, and generally found that the 2012 FRM analysis had predicted
significantly larger battery capacities for a given range. The Draft TAR analysis revised these
figures accordingly by making changes to many of the parameters that determine BEV battery
sizing, as described in the Draft TAR.
Figure 2.34 compares the battery capacities of MY2012-2016 BEVs to the battery capacities
that were estimated for the Draft TAR analysis.
100
90
80
.c
70
§
>.
60
u
re
Cl
50
T3
U
£¦
40
OJ
iJ
ID
30
CD
20
10
0
\
s >/
XjT ••
•
i
•
[
kx
r*
!
0
a \
'J
• MY 2012-2016 BEVs
O draft TAR
Trend Sne
0 50 100 150 200 250 300
EPA estimated range (mi)
Figure 2.34 Comparison of 2012-2016MY BEV Battery Gross Capacities to Draft TAR Estimates
For each BEV range modeled (75, 100, and 200 miles), the Figure shows the battery capacity
estimates used in the Draft TAR. For each BEV range, several values are seen, corresponding to
each of the vehicle classes (Small Car, Standard Car, Large Car, etc.) and glider weight
reductions of 0 percent to 20 percent.
It can be seen from the plot that the Draft TAR estimates centered quite well upon the
trendline established by the population of production vehicles of a similar range. This
represented a significant improvement over the 2012 FRM projections, which had significantly
overestimated capacities. As discussed in the Draft TAR, the improvement was the result of
updating many of the parameters that are influential to the estimation of battery capacity, as
described in Chapter 5.3 of the Draft TAR. This Proposed Determination analysis makes
additional adjustments to the PHEV battery sizing methodology, based on updated information
and public comments on the Draft TAR, which are discussed in Chapter 2.3.4.3.7 (Cost of
Batteries for xEVs).
2-103
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.2.4.4.6 Re latins Power to Acceleration Performance
As discussed previously in the sections on PHEVs and BEVs, the high low-end torque
associated with electrified powertrains means that the relationship between rated powertrain
power and acceleration performance may differ substantially for electrified vehicles compared to
conventional vehicles. Understanding the relationship between the rated power of an electrified
powertrain and the performance it provides is important to properly sizing the powertrain for a
target performance level. This section examines this issue further by comparing the power
ratings and performance of electrified vehicles currently on the market to that of conventional
vehicles, and deriving an empirical relationship between power and 0-60 time that better applies
to electric drive. Although a more detailed discussion was presented in the Draft TAR, this
Proposed Determination analysis adds additional acceleration data for MY2017 PEVs, which
also serves to update the empirical relationship from that presented in the Draft TAR.
One of the most common metrics of acceleration performance is the time it takes a vehicle to
accelerate from zero to sixty miles per hour, also known as the 0-to-60 time. Although there are
other metrics that describe acceleration performance, including metrics such as 0-to-30 time, 30-
to-60 time, and quarter-mile time (and gradeability metrics as well), 0-to-60 time is likely the
most familiar metric for understanding the acceleration performance of a vehicle.
While in widespread popular use, the 0-60 metric is not reported by manufacturers to EPA nor
is its measurement subject to uniform standards. As an alternative, acceleration times of vehicles
with conventional powertrains are sometimes estimated by means of an equation developed by
Malliaris et al.296 The Malliaris equation relates 0-to-60 time to the power-to-ETW ratio of a
vehicle. This power-law equation has two numerical coefficients empirically obtained from a
least-squares fit of vehicle performance data. Until a different method was adopted in 2014,
EPA historically used this equation and coefficients to estimate acceleration performance of
vehicles for pre-2014 editions of the annual Trends Report.297'FF
The Malliaris equation is depicted in Equation 1 below, with the coefficients 0.892 and 0.805
representing conventional vehicles with automatic transmissions.
The Malliaris equation suggests that the acceleration performance of a vehicle may be
modeled as a function of power-to-ETW ratio, and therefore it suggests that acceleration levels
may be maintained by maintaining a similar power-to-ETW ratio among modeled vehicles. It
also suggests that a specific 0-60 time can be targeted by specifying the corresponding power-to-
FF Subsequent editions of the Trends Report have used a newer method developed by MacKenzie et al.FF that EPA
believes to be more accurate, particularly for newer vehicles. However, the MacKenzie method is not directly
applicable to electric powertrains due to the requirement for ICE-specific inputs.
Equation 1. Malliaris equation for 0-60 acceleration time in seconds
2-104
-------
Technology Cost, Effectiveness, and Lead Time Assessment
ETW ratio. For example, Figure 2.35 plots the Malliaris equation (converted to SI units) for a
range of power-to-ETW ratios, showing the approximate 0-60 times that it would predict.
The fact that the Malliaris equation is derived from an analysis of conventional powertrains
suggests that it might not be equally valid for electrified powertrains. The Draft TAR recognized
that a significant number of PEV models had entered the market since the 2012 FRM, and took
this opportunity to characterize the acceleration performance of a selection of MY2012-2016
PEVs for which curb weights and estimated all-electric 0-60 times were available.
To illustrate, Figure 2.35 plots the approximate 0-60 mph acceleration times of MY2012-2017
BEVs and PHEVs as a function of their power-to-ETW ratio, as expressed by rated peak motor
power (kW) divided by test weight (the published curb weight in kg, plus 136 kg payload).GG
Acceleration times were collected from publicly available sources including manufacturers and
press organizations, and in some cases were averaged when estimates from different sources had
slight variation. PHEVs for which an all-electric (battery only) acceleration time could not be
established were not included.
An empirical trendline was derived from this data and is shown in the Figure as a thin orange
line. For comparison, the acceleration times that would be predicted by the Malliaris equation
for the same range of power-to-ETW ratios is shown in the Figure as a heavy black line. As
shown by Equation 2, the empirical trendline has the same equation form as the Malliaris
equation, but with different coefficients of 1.1321 and -0.733 that result from a least-squares fit
to the PEV data as expressed in SI units for power and weight.™
16
14
QJ
"J 10
e
5 B
* EVandPHEV
E 6
o ICE (Maliaris)
6 4
2
0
0 0.02 0.04 0.06 O.OB 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24 0.26
Power-to-ETW ratio (kW/kg)
Figure 2.35 Acceleration Performance of MY2012-2017 PEVs Compared To Targets Generated By Malliaris
Equation
GG Tesla high-performance vehicles represented by 85 kWh Model S.
HH The coefficients are different from those reported in the Draft TAR due to the addition of several MY2017 BEVs
to the data set.
2-105
-------
Technology Cost, Effectiveness, and Lead Time Assessment
t = 1.1321
kW \~0-733
\kg ETW>
Equation 2. Empirical equation for 0-60 all-electric acceleration time of MY2012-2017 PEVs
As described in the Draft TAR, it can be seen that the 0-60 times for MY2012-2017
electrified vehicles fall on a significantly different line than that described by the Malliaris
equation. As the Draft TAR found, using the Malliaris equation to size electrified powertrains
results in significantly faster projected 0-60 acceleration times than would likely be intended. For
example, to target a 0 to 60 mph acceleration time of 10 seconds, the Malliaris equation (shown
by the heavy line) would indicate that the motor should be sized to achieve a power-to-ETW
ratio of 0.08 kW/kg. However, the empirical PEV trendline indicates that this power-to-ETW
ratio would actually provide an electric powertrain with an acceleration time of about 7 seconds.
As described in the Draft TAR and depicted in Table 2.4, the 2012 FRM therefore had
effectively assigned significantly greater acceleration times than intended, which also inflated the
necessary motor and battery power ratings.
Table 2.4 PEV Acceleration Performance Intended in the FRM and Projected Probable Performance
0-60 mph time (sec)
Class
FRM intent
FRM actual
Small Car
11.1
7.7
Standard Car
9.5
6.6
Large Car
6.8
4.7
Small MPV
11.3
7.9
Large MPV
9.5
6.6
Truck
8.8
6.1
The empirically derived relationship shown in Equation 2 is used for PEV motor power
assignment in this Proposed Determination analysis. The equation differs slightly from that used
in the Draft TAR analysis due to the addition of several MY2017 vehicles to the data set.11 This
change has negligible effect on the resulting motor power assignments.
2.2.4.5 Developments in Electrified Vehicle Battery Technology
For many types of electrified vehicles, particularly PHEVs and BEVs, battery cost is the
largest single component of vehicle cost. Battery pack cost is determined in part by the
configuration of the pack, which should be tailored to the specific performance goals of the
vehicle.
Pack configuration may be decomposed into a large number of primary design parameters
which the vehicle designer can specify to determine the performance of the pack and ultimately
its cost. In configuring a pack, the primary performance targets are energy capacity in kilowatt-
hours (kWh) and power capability in kilowatts (kW). These performance targets are determined
by design choices such as: battery chemistry (although all PEVs currently use lithium-ion
chemistry, this is a family of chemistries composed of a number of specific cathode and anode
11 For the equation used in the Draft TAR, see Draft TAR p. 5-329.
2-106
-------
Technology Cost, Effectiveness, and Lead Time Assessment
formulations); pack voltage, usable portion of total capacity, cell capacity (Ampere-hours per
individual cell), cell topology (the electrical and physical arrangement of cells and modules in
the pack), and cooling method (passive or active, and air or liquid), among others. Further, for a
pack defined by a given set of these design parameters, the assumed annual manufacturing
volume will also influence the projected cost.
It is customary to refer to battery cost in terms of cost per kWh. However, in order to make
valid comparisons on this basis it is important to understand that cost per kWh is strongly
influenced by the power-to-energy (P/E) ratio of the battery. Intuitively, a BEV battery
optimized for energy storage capacity (low P/E) will have a low cost per kWh because the
materials and construction are oriented toward providing maximum energy capacity.
Conversely, an HEV battery optimized for power (high P/E) will have a higher cost per kWh,
because the materials and construction are oriented toward providing power, while the metric of
cost per kWh continues to focus on energy. For these reasons, cost per kWh figures derived
from energy-optimized BEV or PHEV battery packs should not be used to estimate the cost of a
power-optimized HEV pack, or vice versa. Comparisons of cost per kWh are only valid when
the applications have a similar P/E ratio.
It is also important to be aware of whether a cited cost per kWh is on a cell basis or a pack
basis. Figures found in press or manufacturer literature may be of either type. Costs cited on a
cell basis will be much lower than for a full pack that includes battery management, disconnects,
and thermal management. As in the 2012 FRM and Draft TAR analyses, for this Proposed
Determination analysis all cost per kWh figures are presented on a pack basis.
Finally, the energy capacity of a battery pack (kWh) may be characterized either by gross
capacity or usable (net) capacity. Gross capacity, also known as nominal or nameplate capacity,
is the total amount of energy that can be reversibly stored in a complete charge and discharge
cycle of the battery, without regard to long term durability. It is a relatively fixed quantity that is
a function of the amount of electrode active materials contained in the battery. Usable capacity
is the portion of gross capacity that the manufacturer believes can be regularly used in an
application while maintaining a desired level of durability. Although usable capacity is the
metric that relates best to performance attributes such as driving range, usable capacity varies
widely among different vehicle types and individual models of each type. For consistency it has
become customary to refer to the size of xEV battery packs by their gross capacity, and to refer
to battery cost per gross kWh. As in the 2012 FRM and Draft TAR analyses, the Proposed
Determination analysis follows this standard.
2.2.4.5.1 Battery Chemistry
EPA bases its battery cost analyses on outputs of the ANL modeling tool BatPaC135, which
models several well established lithium-ion chemistries. As shown in Table 2.5, the choice of
chemistries available in BatPaC includes:
Table 2.5 Lithium-ion Battery Chemistries Available in ANL BatPaC
Chemistry
Cathode
Anode
LMO-G
Lithium-Manganese Oxide
Graphite
LMO-LTO
Lithium-Manganese Oxide
Lithium Titanate Oxide
NMC333-G
Nickel-Manganese-Cobalt (3-3-3)
Graphite
NMC622-G
Nickel-Manganese Cobalt (6-2-2)
Graphite
2-107
-------
Technology Cost, Effectiveness, and Lead Time Assessment
NCA-G
Nickel Cobalt Aluminate
Graphite
LFP-G
Lithium-Iron Phosphate (Olivine)
Graphite
Certain chemistries are better suited for certain applications than for others. For example, the
specific versions of NMC chemistry that are modeled by BatPaC are well suited for packs having
a large energy capacity such as BEV packs, but due to limits on area specific impedance (ASI),
they are not as well suited for small, power-dense packs for HEVs. Considerations such as these
ultimately led to the chemistry choices EPA employed for the FRM and Draft TAR analyses. In
the Draft TAR, BEV and PHEV40 batteries were configured with NMC441-G, while PHEV20
and HEV packs were configured with LMO-G. For the Draft TAR analysis this was updated to
NMC622-G and a blended formulation of 75 percent LMO and 25 percent NMC, respectively.
These chemistries continue to be representative of industry practice and so were retained for the
Proposed Determination analysis.
Since the 2012 FRM, the lithium-ion family of chemistries has continued to dominate xEV
battery technologies seen in current and announced production vehicles. As expected,
NMC/NCM cathode formulations are increasingly being seen in BEVs announced since the
FRM, including in mixed formulations with LMO. For example, the Kia Soul BEV uses an NCM
cathode.298 In the 2015 NAS report (p. 4-26), the committee mentions the use of NMC cathodes
for the 2020 to 2025 time frame, lending further support to EPA's choice. PHEVs and HEVs are
being seen not only with LMO-dominant cathode formulations, such as in the original Chevy
Volt, but also with NMC and blended NMC cathode formulations, as in the 2016 Chevy Volt,299
the Ford C-Max Hybrid HEV and C-Max Energi PHEV.300 These are presumably optimized for
the relatively high P/E ratio of these applications. Lithium-iron phosphate cathodes are also
being promoted for HEV use.301 While it is not possible for BatPaC to model every (often
proprietary) variation in cathode formulation, the available choices are likely sufficient to
represent the cost spectrum applicable to this family of chemistries.
As discussed in the Draft TAR, use of pure LMO cathodes in xEV batteries has gradually
trended toward blends of NMC and LMO.302 In particular, most HEV batteries currently in
production appear to utilize either NMC or LMO blended with NMC. For example, the 2016
Chevy Malibu Hybrid battery is said to use an NMC cathode303 while the Volt uses NMC
blended with LMO.299
Version 3 of BatPaC, released for beta in November 2015, added the more common NMC622
cathode formulation in place of NMC441, and a user-selectable blend of NMC and LMO. The
Draft TAR analysis was thus able to adopt a blended NMC-LMO cathode for HEV and PHEV20
batteries, to better represent their usage in existing platforms. The November 2015 Version 3
continues to be the most current version and was retained for use in the Proposed Determination
analysis.
At the time of the 2012 FRM, practically every production xEV was using a Li-ion chemistry,
with the nickel-metal-hydride (NiMH) battery of Toyota HEV products being the primary
exception. After using NiMH in the Prius since its introduction in 1997, there are signs that even
the Prius may be moving toward Li-ion. By 2012, Toyota had already adopted a lithium-ion
chemistry for the Prius PHEV, a platform which requires a larger battery capacity than the
standard hybrid. In October 2015, Toyota announced that the 2016 Prius hybrid would also
begin offering a Li-ion battery as an option.166'304 In November 2016, it was reported that Toyota
2-108
-------
Technology Cost, Effectiveness, and Lead Time Assessment
has taken further steps to incorporate lithium-ion technology in its portfolio by announcing plans
to use Li-ion for the Prius Prime and potentially for future BEVs.305
Since the 2012 FRM, industry research has continued into more energy- and power-dense
variations of the lithium-ion platform, including improved cathode material blends, lithium-rich,
manganese-rich, nickel-rich, and higher voltage (e.g. 5 V) spinel cathodes, and the use of silicon
in the anode. Other research is concerned with even more advanced platforms, including
lithium-sulfur, and several metal-air chemistries (lithium-air, aluminum-air and zinc-air) among
others. These advanced chemistries are not yet available in cells suitable for xEV use, but
potential examples are beginning to emerge.
Lithium-sulfur (Li-S) cells are beginning to be seen in some highly specialized applications.
A Li-S cell manufactured by Sion Power is used in the Airbus-sponsored Zephyr high-altitude
unmanned aerial vehicle (UAV) to store solar energy for nighttime flight. The low-temperature
performance of Li-S cells may have in part led to the choice of this chemistry for this
application.306 Oxis Energy is expected to release a commercial Li-S battery cell in 2016, with
an eye toward xEV applications.307'308
Silicon is also beginning to appear in the anode of commercial Li-ion cells. While it takes 6
carbon atoms in a carbon anode to accept 1 lithium ion, a silicon atom can accept several.
However, uptake of lithium ions by silicon is accompanied by extreme volumetric expansion,
leading to complications such as disintegration of the anode matrix and loss of electrical
conductivity. For this reason, many are currently focusing on very small additions of silicon to
an otherwise carbon-based anode to achieve incremental improvements in specific energy. In
2015 Tesla Motors Inc. announced a 90-kWh Model S pack that was said to achieve a greater
specific energy by including a small amount of silicon in the anode.309
Solid-state lithium-ion cell technology is another active area of research. Most solid-state
construction concepts retain the traditional anode and cathode couples but replace the liquid
electrolyte with a solid (usually polymer) electrolyte. Others seek to enable use of lithium metal
as the anode by leveraging the solid nature of the electrolyte to prevent dendrite formation. Solid
state construction leads to the possibility of more efficient production techniques, such as
building complete battery cells by printing or deposition, potentially in complex shapes that
conform to available packaging space, or in flat shapes that could be integrated structurally with
the vehicle. Minimizing the resistance of the solid electrolyte is a primary research target for
enabling this technology. As an indicator of interest in this technology, the British appliance
manufacturer Dyson purchased the solid-state lithium-ion battery firm Sakti3 for $90 million in
October 2015.310 In March 2016, it was widely reported that Dyson may be planning to produce
an electric vehicle, as suggested by evidence that the company is receiving U.K. government
funding for this purpose.311 Similarly, Bosch, a major automotive supplier, acquired solid-state
lithium-ion developer Seeo in 2015, citing potential applicability of the technology for increasing
the range of electric vehicles.312
While promising, these and similar early examples of Li-S electrode couples, silicon anodes,
and solid-state construction will need time to show that engineering targets for cycle life,
dimensional stability, and durability in demanding xEV applications have been reliably met.
Until then, reliable estimates of their cost or commercial availability will not be available.
Metal-air chemistries will require even more development before they will be mature enough to
characterize their potential use in automotive applications or their production costs. The 2015
2-109
-------
Technology Cost, Effectiveness, and Lead Time Assessment
NAS report (Finding 4.5, p. 4-44) further supports the conclusion that "beyond Li-ion"
chemistries such as these are unlikely to be commercially available during the time frame of the
rule. At this time EPA considers it unlikely that fully proven forms of such chemistries will
become commercially employed in xEV applications on a broad scale during the time frame of
the MY2022-2025 standards. The developmental state of these chemistries and the
unavailability of well-developed cost models prevent their inclusion in our analysis.
2.2.4.5.2 Pack Topology, Cell Capacity and Cells per Module
Pack topology refers in general to the way cells and modules are electrically connected to
form a pack. Modules are collections of cells that act as building blocks for a pack. Cell
capacity is the charge capacity of an individual cell, and is closely related to pack topology.
To fully understand developments in these areas and EPA's choices for these parameters in
the modeling of battery packs for costing purposes, an example of how these parameters interact
will now be presented as background.
One approach to configuring a battery pack would start with a target pack voltage for the
application. Target voltage typically refers to the nominal voltage expected at about 50 percent
SOC. For PEVs, the targeted voltage is typically between 300 V and 400 V. The most
commonly used Li-ion chemistries provide a nominal voltage between 3 V and 4 V per cell.
Assuming a 3.8 V cell and a target of 365 V, a BEV pack might be constructed of 96 cells
connected as series elements (3.8 V * 96 = 365 V). The target energy capacity of the pack
(kWh) would then be achieved by specifying the capacity of each cell. The larger the target pack
capacity, the larger the required capacity of the cell. In this example, to target a 24 kWh pack
capacity, each series element would need to have a capacity of about 66 A-hr:
24,000 W-hr / 3.8 V / 96 cells = 66 A-hr
Manufacturers have several options for providing this cell capacity. The simplest would be to
manufacture cells of 66 A-hr capacity. This results in one cell at each series position,
minimizing the number of cells and interconnections, potentially minimizing the cost of the
pack. In practice, manufacturers may instead be compelled to use smaller cells, perhaps to better
address thermal management considerations, or to match an existing cell size offered by a cell
supplier. The 66 A-hr required at each series position might then be provided by two 33 A-hr
cells, or three 22 A-hr cells, connected in parallel. The exact cell capacity could vary slightly to
match available products if some variation in pack capacity or voltage are permissible.
Increasing the pack capacity, for instance doubling it to 48 kWh, could in theory be achieved
either by doubling the number of series elements (from 96 to 192) or by doubling the A-hr
capacity of each series element (to 132 A-hr). The first option is problematic because it would
double the voltage to 730 V, which presents a potential safety issue and may be outside the
typical operating voltage range of available power electronics. The larger cell capacity of the
second option may be difficult to achieve in a single cell while maintaining effective thermal and
current distribution characteristics within the cell. For these reasons, larger packs are often
found to include parallel strings of two or more smaller cells at each series position. Tesla
products are an extreme example, composed of thousands of very small cells, which results in as
many as 36 cells in each series position.
2-110
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Another important aspect of pack topology is the format of the individual cell. Most industry
cell development and current automotive cell applications continue to be centered on prismatic
(rectangular) cell formats composed of stacked or flat-wound electrode strips housed in metal
cans or polymer pouches. ANL BatPaC models a prismatic format housed in a stiff polymer
pouch. Tesla is almost unique among PEV manufacturers in its use of small, cylindrical 18650-
format cells.313 But because Tesla continues to build significant market share, this difference has
potential significance to the projection of future pack costs. Also, there is some evidence that
other manufacturers are beginning to consider cylindrical cells. In 2015 Volkswagen announced
the R8 e-tron which has a pack composed of cylindrical cells; potentially, other products such as
the Q6 e-tron and the Porsche Mission E might also share this format if this is an indication of
VW's future battery construction approach. Additionally, in November 2015 Samsung SDI
announced that it would supply cylindrical cells to a China customer for use in electric SUV
battery packs.314 According to one analysis, about 38 percent of currently available BEV models
have packs composed of cylindrical cells, with the rest roughly evenly divided between prismatic
pouch and prismatic metal can315 (although it is unclear whether the relatively large number of
Tesla sub-models are counted as separate models). About 40 percent of HEV models use packs
composed of cylindrical cells, according to the same source.
Despite the differences between prismatic and cylindrical cell formats, there may be limited
potential for large differences in pack costs to result. First, material costs per unit energy storage
are likely to be similar on a cell basis. Cylindrical cells and prismatic cells differ primarily in the
manner in which layers of active materials are packaged together, one being a spiral winding of a
single electrode strip and the other a stack of multiple smaller strips. Although the assembly
process is different, both methods utilize active material with similar efficiency. This is
significant because material costs are the most dominant component of total cell cost.135>316>317>153
Second, while cylindrical cells may benefit from a somewhat simpler cell manufacturing process
and the highly commoditized status of the 18650 format, the large number of 18650-format cells
that must be connected to build a pack may work against these advantages. While larger
cylindrical cells might be used, their heat dissipation properties may limit their practical size.
While 18650-format cells have good thermal qualities, larger cells begin to face challenges in
rejecting heat from the core of the cylinder where the maximum temperature tends to develop.318
Despite Tesla's success with the cylindrical format, it remains unclear whether either format
possesses a greater potential to eventually minimize pack cost. EPA therefore expects that the
cost estimates of the BatPaC model should be reasonably accurate for both cell formats.
xEV packs are often configured with a single series string of cells. Larger BEV packs may be
configured with a parallel string of two cells in each series position, in order to limit voltage to
the desired range and limit the required A-hr capacity of the cells. xEV battery packs found in
production vehicles (with the exception of Tesla, as previously mentioned) are largely continuing
to follow the practice of having one, two or three cells in parallel at each series position.
EPA expects that as the industry continues to mature, manufacturers will continue to pursue
economies by gradually optimizing cell capacities to the requirements of the application,
including an increase in cell capacity for large packs in order to minimize the number of cells
while limiting the total voltage. As described in the Draft TAR, there is evidence that
manufacturers are continuing to increase BEV cell capacities.
2-111
-------
Technology Cost, Effectiveness, and Lead Time Assessment
As announced by GM in October 2014, the Chevy Volt generation 2 battery pack has fewer
cells than the original generation (192 vs 288) that are each about 50 percent greater in capacity.
In the original pack, each series element was composed of three cells in parallel, while the new
configuration has only two.319 The 30 kWh trim of the 2016 Nissan Leaf, announced in
September 2015, achieves its increased capacity within about the same size and footprint of the
lower-trim 24 kWh pack by utilizing a more energy dense chemistry variation. The number of
cells remained unchanged at 192, implying an increase in the A-hr capacity of each cell.320
Similarly, the 2017 BMW i3 achieves a 50 percent increase in capacity over the earlier model,
within the same pack volume, by using a 94 A-hr cell in place of a 60 A-hr cell.277
The latter example further suggests that cell suppliers are pushing the envelope of cell
capacity for vehicular applications beyond the limit used in the 2012 FRM analysis, which was
set at 80 A-hr for BEV cells. The 60 A-hr cell format that Samsung SDI had been supplying to
BMW for the pre-MY2017 BMW i3 pack was already one of the larger light-duty BEV cell
formats in use when it was replaced by the 94 A-hr format. At AABC 2015, Samsung SDI
presented further plans for manufacturing prismatic cells of 90 to 120 A-hr by 2020.321 The
presenter also mentioned a goal of eventually producing 180 A-hr cells for BEV use, using a new
chemistry with high NCM content plus silicon. This suggests that at least some suppliers are
already anticipating a market in vehicular applications for these very large format cells.
Module configuration is another topology issue. In general, the more cells that are included in
each module, the fewer modules and the lower the cost of their connections. Since the number
of modules must be a whole number, the number of cells per module can depend on the total
number of cells necessary to reach a voltage or capacity target for the pack, and so need not be
the same for every size of pack.
In the 2012 FRM analysis, battery modules for all xEVs were configured with 32 cells per
module. At the time of the FRM, the Chevy Volt provided one example of a manufacturer that
was already using at least 32 cells per module, in a liquid-cooled application similar to that
assumed in the analysis of BEVs and PHEVs. Although most BEVs at the time had fewer than
32 cells per module, this figure was selected to represent expected improvements in cell
reliability and packaging methods as manufacturers gained experience over time. It is now
understood that the original Chevy Volt battery was configured with 7 modules of 36 cells each
and 2 modules of 18 cells each. A similar configuration is retained in the 2016 Volt. Similarly
the Kia Soul EV battery consists of 192 cells in 8 modules,322'323 varying from 20 to 28 cells per
module. As another example, in September 2015, Nissan announced the 30 kWh battery pack
option available with the 2016 Leaf, in which the number of cells per module is increased from 4
to 8. The two higher-trim versions of the Leaf, the SV and SL, were the first to include the 30
kWh pack option, followed by the elimination of the 24 kWh pack option in all trims as of
October 20 1 6.324 While the number of cells per module is still relatively small, Nissan's
continued use of passive air cooling as a thermal management strategy may place a smaller limit
on the number of cells per module than for the more common liquid-cooled packs that are
modeled in the EPA analysis.
In November 2015 at the Tokyo Auto Show, Nissan revealed its IDS concept vehicle,
powered by a newly developed 60 kWh pack.325'326 In interviews with the press, a number of
details were shared regarding the design of this pack. The pack was described as having 288
cells utilizing an NMC cathode chemistry. Assuming a nominal cell voltage of 3.75V typical of
2-112
-------
Technology Cost, Effectiveness, and Lead Time Assessment
these chemistries, each cell would be sized at about 55.5 Ampere-hours, significantly larger than
in the Leaf pack. The IDS pack also appears to install in a footprint similar to that of the 30 kWh
version of the Leaf battery. It does not appear that Nissan has yet announced the number of cells
per module in the 60 kWh pack, but appearance suggests that it is significantly larger than in the
Leaf packs. One interesting aspect of the design approach for this pack is its support for a
variable module stack height, suggesting a variable number of cells per module may be specified
depending on the target capacity of the pack. In one press report,327 an official was described as
saying that Nissan had taken a conservative approach to the number of cells per module in earlier
packs, and due to the lack of failures or other issues with those packs, were now able to consider
an approach that supports a much larger number of cells per module in the new pack.
In January 2016, GM announced details of the Chevy Bolt battery pack.328 As with the 60
kWh Nissan IDS pack, this 60 kWh pack is composed of 288 cells in 96 cell groups of 3 cells
each. The cells are distributed among 10 modules, or about 28 to 30 cells per module. Three
individual cells are connected in parallel at each series position. Assuming a nominal cell
voltage of 3.75V, this suggests an individual cell capacity of 55.5 Ampere-hours (identical to
that of the Nissan IDS pack).
As noted above, the ideal number of cells per module may vary depending on the capacity of
the pack and the size of the cells. In the 2012 FRM, modules were assigned 32 cells each. This
was updated to a variable number for the Draft TAR and Proposed Determination analyses,
which achieves an improved optimization of the pack topology and a better targeting of pack
voltage and cell capacity. More details may be found in Chapter 2.3.4.3.7 (Cost of Batteries for
xEVs).
2.2.4.5.3 Usable Energy Capacity
As previously noted in the introduction to this section, batteries may be described with respect
to their gross energy capacity or their usable energy capacity. Usable capacity refers to the
portion of gross capacity that the manufacturer believes can be regularly used in an application
while maintaining a desired level of durability. It is thus an important parameter for battery
sizing because it determines the gross capacity necessary to provide a target usable capacity for
an application.
The concept of usable capacity is often accompanied by several closely related terms. In this
discussion, the following terms are used and defined as follows. State-of-charge, or SOC, refers
to the percentage of total energy (kWh) or charge (Ampere-hour) capacity that remains in a
battery at a given time, ranging from 0 to 100 percent on a gross capacity basis. SOC design
window,329 or simply SOC window, refers to the usable portion of total capacity intended by
design, expressed in terms of SOC; for example, an SOC design window might be described as
the range between 25 percent and 75 percent SOC, or alternatively as an SOC window of 50
percent. SOC swing may be used interchangeably with SOC window but is used here to refer
more specifically to observed in-use behavior rather than a design context. Usable capacity is
thus determined by SOC design window (in a design context) or implied by an observed SOC
swing (in-use). Usable capacity may refer either to a usable energy (in kWh) or the usable
portion of gross capacity (in percent).
For lithium-ion chemistries, SOC is not always measurable with precision and is commonly
estimated by means of algorithms that include measurements of current, voltage and battery pack
2-113
-------
Technology Cost, Effectiveness, and Lead Time Assessment
temperature, both instantaneous and over time. The construct of SOC window therefore inherits
some of these traits. While it is most convenient to think of the boundaries of an SOC window in
terms of SOC percentages, it may also be defined by an allowable range of battery voltages, or a
combination of the two.
The SOC design window that a manufacturer assigns to a battery is typically selected to
balance battery durability with energy availability. Owing to the complexity of battery behavior
and vehicle control algorithms, it is possible that some controllers may not refer to a single
rigidly defined SOC window, but instead, may define multiple or variable SOC windows that
apply to different usage conditions or are determined by the controller's observation of patterns
of usage or battery health monitoring over a short or long term. For example (and particularly
for HEVs), because extreme but intermittent usage conditions may have a different degree of
impact on battery life than normal usage, it is possible that some manufacturers may program
their controllers to define multiple target windows, to allow a wider swing to accommodate
temporary, extreme conditions while following a narrower swing for normal conditions. As
another example, some manufacturers may widen the allowable SOC swing as the battery ages
(perhaps by allowing a wider range of allowable voltages, or modifying the allowable SOC
window) in order to maintain driving range or usable capacity. Although the concept of a single
SOC design window may therefore be overly simplistic for some vehicles, it remains useful for
battery sizing purposes.
Setting an appropriate SOC window can be influenced by the effectiveness of the battery
management system (BMS). Improved BMS systems are one potential path toward enabling a
wider SOC window or a reduced battery capacity for a given range.330
The SOC design window is a primary factor in the sizing of a battery for a particular use.
That is, the desired electric driving range for a PEV, or the amount of energy buffering capability
desired for an HEV, combined with the SOC window, directly suggests the necessary gross
capacity of the battery. In the 2012 FRM, for battery sizing purposes, EPA assumed a 40 percent
usable SOC window would apply to HEVs, 70 percent for PHEVs, and 80 percent for BEVs.
The Draft TAR noted that increases in PHEV and BEV driving range that have been observed
since 2012 may have been enabled in part by increases in SOC design window and hence usable
capacity. The 2015 NAS report also stated (p. 4-5), "as extended in-use experience is obtained,
the battery SOC swing may be increased for all electrified powertrains." For these reasons, in
the Draft TAR EPA reviewed the usable capacity assumptions used in the 2012 FRM and made a
number of revisions, as described more fully in Draft TAR Chapter 5.3.4.3.7.1. The Draft TAR
analysis updated these figures to 75 percent for PHEV40, 85 percent for BEV75 and BEV100,
and 90 percent for BEV200. These figures are further discussed in the paragraphs below.
Applicable updates to these figures for the Proposed Determination analysis are described in
Chapter 2.3.4.3.7 (Cost of Batteries for xEVs).
Usable capacity for HEVs
For the 2012 FRM and Draft TAR analyses, a 40 percent usable capacity was chosen for
strong HEVs in the 2020 to 2025 time frame. Although many production HEVs have been
reported to use about 20 to 30 percent, the Draft TAR examined and reaffirmed the case for 40
percent on the expectation that improvements in battery technology and manufacturer learning
would enable a wider SOC design window by 2022 to 2025.
2-114
-------
Technology Cost, Effectiveness, and Lead Time Assessment
As described in the Draft TAR, the 2015 NAS report (p. 4-5) was skeptical of the choice of a
40 percent usable capacity for HEVs and suggested using a value closer to the 20 to 30 percent
observed in production HEVs. The NAS report supported this position in part by contending
that, by virtually doubling the SOC window, the HEV batteries projected in the analysis would
be "half the cost and size" of what would be required. However, as discussed in the Draft TAR,
EPA believes that a wider SOC window would not have this effect. At the high power-to-energy
(P/E) ratio of an HEV battery, cost is not as strong a function of capacity (kWh) as a function of
power (kW). Therefore, reducing battery capacity from e.g. 0.50 kWh to 0.25 kWh, while
holding the required power constant, would not correspondingly reduce the cost by half, because
the reduction in capacity would push the P/E ratio to a higher level, counteracting much of the
cost reduction. Cost projections generated by BatPaC confirm this trend and show that, for a
given power capability, the cost of a 0.25 kWh pack would be very similar to that of a 0.50 kWh
pack. For example, BatPaC Version 3 projects that an HEV pack sized for a power output of 15
kW would cost $634 as a 0.25 kWh pack, and $660 as a 0.50 kWh pack, a difference of only
about 4 percent.JJ Therefore at these relative pack capacities, EPA's use of a 40 percent SOC
design window for sizing purposes does not have a large impact on projected cost.
EPA also believes that developments in battery technology and manufacturer learning
observed since 2012 have been consistent with the expectation that a 40 percent usable capacity
will be applicable to HEVs in the 2022 to 2025 time frame. Since the 2012 FRM, numerous
HEV models and battery systems intended for such vehicles have been announced. It is clear that
although some HEV manufacturers have continued to use a rather conservative SOC window
(for example, at AABC 2015, it was reported that the 2016 Malibu Hybrid uses a 1.5 kWh pack
of which 30 percent is usable (450 Wh of 1500 Wh)303), there is also evidence that some
manufacturers have begun increasing the SOC design window in subsequent generations of
HEVs.
Specifically, recent developments in batteries for 48V mild hybrids, which have smaller
batteries than strong HEVs but similarly demanding requirements, have supported a relatively
wide swing. At AABC 2015, Bosch presented a 0.25 to 0.50 kWh battery system designed for
use in a 48V hybrid. This battery was described as having been designed for an SOC window
from 30 percent to 80 percent SOC (a 50 percent usable capacity) despite its small total
capacity.331 Also at AABC 2015, A123 Systems presented a battery system for a 48V hybrid
that uses a proprietary chemistry variation on Lithium-iron phosphate which the company calls
Ultraphosphate. Like the Bosch system, this 0.37 kWh pack supports a window from 30 percent
to 80 percent SOC (50 percent usable capacity). A123 indicated that production of this pack is
planned to begin in 2017.301
In 2014, EPA tested a 2013 Volkswagen Jetta Hybrid supplied by Transport Canada as part
of an exploratory benchmarking exercise. Several braking and acceleration episodes were
performed with the intention of eliciting maximum swing of the 1.1 kWh battery. Multiple
energy swings were observed in both charge and discharge ranging from 0.56 to 0.65 kWh,
equivalent to a gross SOC swing of about 51 to 59 percent.332 Although this testing documented
that the vehicle controller will permit this SOC swing to occur under these usage conditions, it
11 BatPaC inputs: LMO-G chemistry, 1 module of 28 cells, EG-W (liquid) cooling, HEV-HP vehicle type, 450K
annual production volume.
2-115
-------
Technology Cost, Effectiveness, and Lead Time Assessment
remains unclear whether this degree of swing would be observed regularly over normal usage. A
limited amount of testing over steady-state and standard test cycles elicited smaller swings of up
to approximately 30 percent. The short duration of standard test cycles and variation in the
observed swing prevented firm conclusions from being drawn about the exact SOC design
window the controller regularly permits.
Going forward, it is possible that improvements in cell balancing may also act to support
downsizing of HEV battery sizes or widening of SOC windows from their current levels. For
example, at AABC 2015, NREL presented work showing that use of active cell balancing instead
of passive balancing can result in a 50 percent reduction in the necessary capacity of an HEV
battery while also eliminating the need for liquid cooling.333 Further, EPA models HEV battery
costs using the liquid cooling option provided in BatPaC, which means these batteries have more
effective cooling than the air cooling that currently prevails in HEV batteries, potentially
allowing greater use of available capacity. HEV battery cooling is discussed further in Chapter
2.2.4.5.4 (Thermal Management).
These findings suggest that EPA's choice of 40 percent usable capacity for HEVs remains a
reasonable estimate for the 2022 to 2025 time frame.
Usable capacity for PHEVs
The usable portion of total capacity for a PHEV tends to be narrower than for a BEV. One
reason for this difference is that when a BEV reaches its minimum SOC, it is taken out of
operation and recharged, while a PHEV instead begins to operate in charge-sustaining mode
(charging and discharging within a narrow SOC band) for an indefinite time. The need to
provide a proper lower-end buffer for the SOC band, and to avoid extensive operation at a very
low SOC, encourages setting a higher minimum SOC point for a PHEV than for a BEV. PHEV
batteries also tend to have a larger P/E ratio due to their need to provide similar power levels as a
BEV battery while having a smaller capacity. A smaller SOC window would be appropriate
under such conditions to promote battery life. The 2015 NAS report (p. 4-12) affirmed the FRM
assumption that a 70 percent usable capacity is appropriate for a PHEV architecture.
At the time of the 2012 FRM, relatively few PHEVs were in production to serve as examples
of this platform. Although the Draft TAR provided a comprehensive analysis of the PHEV
models that have entered the market since, the primary production example available to inform
the 2012 FRM was the Chevy Volt, which was about to be released in its first generation
(referred to here as Genl). Prior to its release, the usable capacity of the pre-production Genl
Volt battery was commonly reported as approximately 8 kWh of a total 16 kWh, or about 50
percent. The first production Genl Volt is now understood to have utilized about 10.2 of 16
kWh, or about 64 percent.334 Testing of a 2012 Chevy Volt by Argonne National Laboratory
showed the vehicle to be utilizing an SOC window between 87 percent SOC and 18 percent SOC
(69 percent usable capacity).335
The initial generations of the Chevy Volt are often described as having adopted a conservative
battery management approach by utilizing a narrow SOC design window and liquid cooling.
GM widened the SOC window for the Volt on at least two occasions while increasing the battery
capacity on at least three. The Genl model was upgraded in the 2013MY from 16 kWh gross
capacity to 16.5 kWh, and further increased for the 2015MY to 17.1 kWh. During this process
the usable energy increased from 10.2 kWh in the 16 kWh version to 11.2 kWh in the 17.1 kWh
2-116
-------
Technology Cost, Effectiveness, and Lead Time Assessment
version. This represented a small increase in usable energy capacity, from 63.75 percent of gross
capacity to 65.5 percent. The Gen2 Volt, released for the 2016MY, now uses 14 kWh of 18.4
kWh gross, or about 76.1 percent usable capacity. This represents a 25 percent increase in
usable capacity from the last Genl model.334
The PHEV batteries modeled in the 2012 FRM are similar to the Volt battery in that they are
liquid cooled, enabling the same level of temperature control that is often cited as being
responsible for the dependability of the Volt battery. The production 2016 Volt battery now
exceeds the 70 percent usable capacity EPA assumed for PHEVs for the FRM analysis.
It should be noted that the 2016 Volt battery is sized for a 53 mile AER, and accordingly may
have a significantly lower P/E ratio than that for a PHEV20. This may allow it to enjoy a wider
SOC design window than the smaller battery of a PHEV20 or possibly even that of a PHEV40.
Therefore, the Volt example may not by itself be conclusive that a wider SOC window would be
appropriate for PHEV20 or PHEV40. However, according to results of testing at Argonne
National Laboratory, the Ford Fusion Energi utilizes about 5.9 kWh of its 7.6 kWh gross
capacity, or about 78 percent. This provides an additional data point suggesting that a wider
SOC window than 70 percent may be appropriate even for some shorter-range PHEVs. The
Fusion Energi is rated at 20 miles of AER, and utilizes a blended depletion style that may utilize
the engine if driven more aggressively than in the standard EPA test cycles. This engine
supplementation at elevated power demands is likely to result in lower peak power demands on
the battery, potentially making wider swings less demanding on the battery.
For the 2012 FRM, a 70 percent usable capacity had been chosen to represent both PHEV20
and PHEV40 vehicles. As discussed in the Draft TAR, the findings reviewed above suggested
that a 70 percent usable capacity for PHEVs may have been a conservative estimate for the 2022
to 2025 time frame. The Draft TAR therefore updated the PHEV40 usable capacity to 75
percent. EPA has further reviewed PHEV usable capacities for the Proposed Determination
analysis, and has updated these estimates as described in Chapter 2.3.4.3.7 (Cost of Batteries for
xEVs).
Usable capacity for BEVs
The Draft TAR examined the large number of BEV models that had reached production since
the 2012 FRM. Further activity in the industry has provided abundant opportunity for
manufacturers to begin drawing conclusions regarding the appropriateness of the SOC design
windows they chose to implement in their first generation models, and even to begin applying
the findings to subsequent model generations. It has also provided many opportunities for
research organizations to test these vehicles to ascertain aspects of their design and behavior,
including SOC swings observed in use. Table 2.6 summarizes some estimated SOC swings
observed in 2012-2016MY BEVs, which are further described below.
Table 2.6 Estimated SOC swings for selected MY2012-2016 BEVs
Example
Estimated
SOC
swing
Source
ANL BEV benchmarking (various)
80 to 90
percent
Argonne National Laboratory
2-117
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Tesla Model S 85
85
percent
AVL
2015 Kia Soul EV
90
percent
Idaho National Laboratory
BMW i3
87
percent
Idaho National Laboratory
Argonne National Laboratory (ANL) operates an ongoing research program to benchmark
xEVs.335 Vehicle testing from multiple instrumented battery electric vehicles has shown that the
vehicles operate usable SOC windows ranging from 80 percent to 90 percent whether air cooled
or water cooled.^
At AABC 2015, AVL presented the results of a teardown of a Tesla Model S battery pack.289
AVL reported that cycling tests of the pack suggested that 73 kWh of the 85 kWh gross capacity
is accessible, suggesting that this pack may be utilizing an 85 percent usable capacity. This
result is in line with reports from Model S owners that have suggested a usable capacity of about
75 to 76 kWh.336
The Advanced Vehicle Testing Activity group at Idaho National Laboratory has tested the
batteries of several BEVs currently in production.337 In testing of the 2015 Kia Soul EV, the
measured battery capacity ranged from 30.4 to 30.5 kWh in each of four test vehicles. The
service manual for the 2015 Kia Soul EV is reported to list a nominal SOC range of 5 percent to
95 percent, or 90 percent usable, for the high voltage battery system.338 A 90 percent SOC
window would amount to about 27 kWh of usable energy, the same as Kia advertises. In a
departure from the practice of most other OEMs, Kia may be advertising the usable capacity
rather than the gross capacity.
Technical specifications for the BMW i3 indicate a battery capacity of 18.8 kWh.339
Numerous press sources widely repeat this figure as a usable SOC while consistently citing a
gross SOC of 21.6 kWh or 22 kWh. The 21.6 kWh figure is highly consistent with the results of
battery testing by Idaho National Laboratory340'341'342'343 for four 2014 BMW i3 vehicles under
test, which indicated gross capacity ranging from 21.4 kWh (one vehicle) to 21.7 kWh (three
vehicles). Like Kia, BMW appears to be advertising the usable capacity of the i3 battery rather
than the gross capacity. A gross capacity of 21.6 kWh suggests a usable capacity of 87 percent.
In May 2014, the Chevy Spark EV underwent changes to its battery that may indicate a
widening of SOC design window. In announcing a change in cell supplier from A123 Systems
to LG Chem, General Motors also indicated that the new Spark battery would be reduced in
capacity from 21 kWh to 19 kWh, while keeping the same range of 82 miles and the same
mpge.344 Given that rated mpge did not change, this suggests that retention of the original range
was more likely made possible by widening the SOC design window than by increasing
powertrain efficiency. A widened window could be enabled by either the use of a different
battery chemistry (going from A123's Lithium-Iron Phosphate to LG Chem's NMC+LMO
chemistry), and/or an increased comfort level due to ongoing experience with the platform.
Since the original A123 cathode chemistry (Lithium-Iron-Phosphate or LFP) is comparable to
KK Instrumented battery electric vehicles include: 2015 Chevrolet Spark EV, Kia Soul EV, 2014 Smart EV, 2013
Nissan Leaf, 2012 Ford Focus Electric.
2-118
-------
Technology Cost, Effectiveness, and Lead Time Assessment
LG Chem's LMO-dominant chemistry in terms of allowable SOC swing, it suggests that
experience may have played at least some role in this change.
At AABC 2015, Honda reported that their decision to extend the lease option on the Fit EV
by 2 years was based on learning that the batteries in these vehicles were experiencing lower
degradation than projected.345 This suggests that it might be possible to widen the SOC design
window in future releases while maintaining durability targets.
For the 2012 FRM, an 80 percent usable capacity was assigned to BEV batteries. This was
based on knowledge of manufacturer plans as well as examples seen in the press for early
production BEVs such as the Nissan Leaf and other developmental vehicles. The 2015 NAS
report (p. 4-12) affirmed that an 80 percent usable capacity is appropriate for BEVs. These
observations of industry practice may be compared with EPA's 2012 choice of 80 percent usable
capacity for all BEVs. The Draft TAR found that a usable capacity of about 85 percent for
BEV75 and BEV100, and 90 percent for BEV200, were more appropriate to assess. EPA further
reviewed these figures for the Proposed Determination analysis, and concluded that they are still
appropriate, as described in Chapter 2.3.4.3.7 (Cost of Batteries for xEVs).
2.2.4.5.4 Thermal Management
Battery thermal management includes battery cooling to reject heat generated during use, and
in many cases battery heating to warm the battery in cold weather. In systems where active
thermal management is present, the battery management system (BMS) will work to keep the
battery within a preferred temperature range during use.
Battery thermal management systems are commonly divided into passive systems (where the
outside of the pack is exposed to ambient air) and active systems (where a cooling medium is
circulated through the pack, or thermoelectric components are integrated with the pack). Active
cooling media may be ambient air, cabin air, air conditioned by the vehicle A/C system, a liquid
coolant, or the A/C system refrigerant.346'347'348'349
For the FRM and Draft TAR analyses, EPA assumed all PEV packs would employ active
liquid cooling, as seen in production vehicles such as the Chevy Volt and in several other PEVs.
In contrast, the FRM analysis assigned passive air cooling to HEV packs. This was updated to
active liquid cooling for the Draft TAR analysis.
One recent approach to cooling battery packs involves placement of a bottom cooling plate
beneath the packaged battery cells rather than between each cell. Coolant or refrigerant
circulates through the plate and cools the battery cells conductively. This approach is used in the
BMW i3 battery, was once used in the Chevy Spark A123-supplied battery, and is possibly being
used in the Chevy Bolt pack.350
Direct circulation of refrigerant rather than an intermediary fluid such as a glycol-water mix
can also improve heat rejection and vehicle packaging by eliminating the secondary cooling loop
that would otherwise be needed to reject heat to the atmosphere. The BMW i3 utilizes
refrigerant cooling.346
Active liquid cooling continues to be the predominant thermal management method for the
battery packs of BEVs and PHEVs announced since the FRM. The notable exception is the
Nissan Leaf, which continues to use passive air cooling as it has since its first generation. At the
2-119
-------
Technology Cost, Effectiveness, and Lead Time Assessment
time of the FRM, some in the industry and press were expressing skepticism about Nissan's
choice of passive air cooling.351'352'353 Some customers had also begun reporting unexpected
battery degradation in hot climates such as Arizona, which some attributed to inadequate thermal
management. During the 2014 MY, Nissan adjusted the chemistry of the battery pack to better
withstand high temperatures.354 Although Nissan has continued to use passive air cooling in the
2016 Leaf (and also in the new 60 kWh pack under development), all other production BEV and
PHEV packs introduced since the FRM use some form of liquid or refrigerant-based cooling.
The 2015 NAS report (under "Cooling," p. 4-17) tended to affirm the agencies' assumption of
liquid cooling for BEV packs by independently noting the potential inadequacy of passive air
cooling in the Leaf pack.
Although HEV packs were modeled with passive air cooling in the 2012 FRM analysis, the
Draft TAR noted some evidence that even these packs may be moving toward liquid cooling, and
adopted liquid cooling in that analysis partly for that reason and partly due to practical
considerations with the BatPaC model, as described in the Draft TAR Chapter 5.3.4.3.7.1.
Although air cooling continues to predominate in HEV packs,349 a presentation by Mahle at
TMSS 2015 suggests that air cooling is increasingly being displaced by liquid cooling even in
HEV packs.347 Johnson Controls has also described a 260 V, 1.7 kWh HEV battery product with
provision for liquid cooling.355 Effective cooling and heating capability is often cited as a
potential path toward reducing the size of xEV batteries by allowing more of their capacity to be
utilized while avoiding the degradation that otherwise might result from heating.349'341 This
suggests that liquid cooling may become one of the enablers for future HEV batteries to provide
the 40 percent usable capacity EPA assumes in this analysis.
As previously described, EPA uses ANL BatPaC to model the cost of xEV batteries,
including mild and strong HEV batteries. BatPaC provides cost estimates for several cooling
options, including active air cooling (cabin air or cooled air) and liquid cooling (glycol/water
mix). It does not model passive air cooling without air channels between the cells, as might be
found in passively cooled HEV batteries. For the Draft TAR analysis, EPA performed several
trials to investigate the impact of the available cooling choices for HEV batteries, and found that
BatPaC assigns similar or slightly lower costs for its implementation of liquid cooling than for its
implementation of active air cooling. For these reasons EPA adopted the liquid cooling option
under BatPaC to model the cost of HEV packs for the Draft TAR analysis as well as the
Proposed Determination analysis, as already true for PHEV and BEV packs.
2.2.4.5.5 Pack Voltage
Some of the HEV battery packs EPA studied for the 2012 FRM operated at approximately
120V. This relatively low voltage (as compared to PHEVs and BEVs) has some advantages, such
as being compatible with the use of a relatively small number of cells per pack, and reducing the
voltage step between the high-voltage system and the 12V electrical system that typically
remains in these vehicles. In contrast, some HEVs use a higher voltage more typical of PHEVs
or BEVs, which may have the advantage of being more compatible with the voltage ranges of
available power electronics components, or the desired power output of the battery to fulfill its
role as part of the system.
Larger packs for PHEVs and BEVs are typically composed of a large number of cells and so
can reach almost any voltage level desired. While safety considerations continue to place a
practical upper limit on system voltage, a moderately high voltage is consistent with the greater
2-120
-------
Technology Cost, Effectiveness, and Lead Time Assessment
power flows required by these vehicles and offers the added benefit of conducting energy at a
lower amperage, which reduces the necessary weight and cost of electrical conductors and
reduces I2R losses. Compatibility of available supplier parts may also encourage different
manufacturers to target a similar voltage envelope. Many manufacturers of PHEVs and BEVs
appear to have targeted the range between 300V and 400V.
In general, the system voltages EPA chose for modeling xEVs were based on those seen in
production xEVs at the time of the FRM. Accordingly in the 2012 FRM and Draft TAR
analyses, EPA limited pack voltages to certain ranges depending on whether the pack was
intended for an HEV, PHEV, or BEV. HEVs were targeted to about 120V while PHEVs and
BEVs ranged from about 300V to 400V.
Originally, in the 2012 FRM analysis, a 600V upper limit on BEV battery voltage had been
applied to the largest BEV packs. At the time of the 2012 FRM, VIA Motors had been producing
a plug-in electric truck with a 650V battery pack. However, later versions of this and other VIA
products by the time of the Draft TAR had adopted a lower battery voltage of around 350V to
380V, suggesting that some advantage was seen to adopting a lower voltage. The Draft TAR
analysis therefore reduced the 600V limit to about 400V, which is retained for the Proposed
Determination analysis.
Other examples of PHEVs and BEVs in the 600V region exist as past-production or concept
vehicles. The McLaren PI PHEV, first introduced to the U.S. in 2014 as a very limited
production high-performance vehicle, operated at 535V, but is no longer in production. In
September 2015, Porsche announced the Mission E concept BEV that would operate at 800V.
The higher voltage was described as enabling much faster charging as well as lower conductor
weight.356 However, this vehicle has not yet been introduced. These examples suggest that
voltage ranges of 600V or greater may continue to be applicable at least to high performance
BEVs and PHEVs, even though they are largely not present in the market today.
For this Proposed Determination analysis, EPA has determined that the targets of 300V-400V
for PHEVs and BEVs remain appropriate (as described in detail in Chapter 2.3.4.3.7).
Public comment on the Draft TAR analysis from Toyota questioned the use of 120V for
HEVs. Although it is true that some HEVs are currently targeting voltage ranges higher than
120V, increasing the voltage of a small (approximately 1 kWh) pack to several hundred volts
requires a larger number of relatively small cells, at a higher potential cost. Going forward to the
2022 to 2025 time frame, it is unclear whether the advantage of operating an HEV at a higher
voltage will continue to outweigh the higher cost of the battery. Therefore, EPA has retained the
approximately 120V target for modeled high-voltage HEVs. More discussion of this comment
and the target voltages for HEVs and PEVs in the Proposed Determination analysis is found in
Chapter 2.3.4.3.7.4 (Assumptions and Inputs to BatPaC).
2.2.4.5.6 Electrode Dimensions
The electrodes of a lithium-ion cell are in the form of flat foil strips coated with active
materials and stacked or rolled together. Several important parameters of cell performance are
controlled by the dimensions of the electrode; in particular, the thickness of the active material
coatings on the electrodes and the aspect ratio (length-to-width ratio) of the electrodes.
2-121
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In general, thinner electrode coatings promote power density, while thicker coatings promote
energy density. By default, BatPaC limits coating thickness to no less than 15 microns and no
more than 100 microns due to various practical considerations.135 The lower limit represents
interfacial impedance effects associated with very thin electrode coatings.357 The typical
precision of coating equipment, at around plus or minus 2 microns,358 would also become
challenged below this thickness. The upper limit represents material handling and ion transport
considerations. Thicker coatings may be prone to flaking when uncut electrode sheets are rolled
or unrolled for shipment and processing. Thicker electrodes also require ions to travel a greater
distance through the active material during charge and discharge, leading to effects such as
increased resistance, reduced power capability, and the potential for lithium plating on charging.
In the 2012 FRM and Draft TAR analyses, electrode coating thickness was therefore limited to
100 microns. In practice, this limit was only encountered by the most energy intensive packs for
large BE Vs.
As discussed in the Draft TAR, updates to BatPaC between the FRM and Draft TAR included
improvements to the model by which electrode thickness is determined. In most cases this
resulted in somewhat thinner electrodes than would have been projected in the version used for
the 2012 FRM analysis. This resulted in a slightly higher cost per kWh for most battery packs,
all other things being equal.359
Electrode aspect ratio is another important parameter, because it determines how far current
must travel on average between where ions reside in the active materials and the current collector
tabs. Longer distances are associated with greater resistance and heat generation. If the length is
much greater than the width, and the current collector tabs reside on the short dimension rather
than the long dimension, current must travel farther on average than in the inverse situation.
BatPaC assumes a default aspect ratio of 3:1, with tabs placed on the short dimension. In the
2012 FRM, EPA had used an aspect ratio of 1.5:1, loosely based on the dimensions of some
commonly known cells at the time.
As originally discussed in the Draft TAR, the 3:1 default aspect ratio used in BatPaC appears
to be seeing increasing use in the industry. In announcing the 200-mile Chevy Bolt EV328 at the
2016 NAIAS, GM indicated that its battery cells, supplied by LG Chem, have an aspect ratio of
3.35:1 (measuring 3.9 inches by 13.1 inches). An animation accompanying the announcement
shows that the cell tabs reside on the short dimension. GM describes this aspect ratio as
"landscape format," presumably to highlight the low-profile design of the pack that allows the
entire pack to reside within the floor space of the vehicle. The Kia Soul EV battery also uses
cells with a nearly identical aspect ratio and tab placement, supplied by SK Innovation.360'298
Also at the 2016 NAIAS, Samsung SDI introduced a family of cells ranging from 26 to 94
Ampere-hours,361 some of which have a similar aspect ratio to the GM Bolt cells but with tabs on
the long dimension. Samsung also displayed a line of "low height packs," suggesting that it
anticipates a trend toward low-profile applications for which these cells would be well suited.362
In December 2015, Volkswagen also announced plans to pursue flat, low-profile pack designs
for future electrified vehicles,363 which likely will also call for a similar cell aspect ratio.
These examples lent support to the validity of the default 3:1 aspect ratio and tab placement
assumed by BatPaC, and EPA therefore adopted a 3:1 aspect ratio for the Draft TAR analysis.
2-122
-------
Technology Cost, Effectiveness, and Lead Time Assessment
No public comment or new information suggested changing the targets for aspect ratio or
electrode thickness for this Proposed Determination analysis. As described in Chapter 2.3.4.3.7
(Cost of Batteries for xEVs), the Proposed Determination analysis retains the Draft TAR values
for these parameters.
2.2.4.5.7 PackManufacturins Volumes
In the 2012 FRM analysis, EPA assumed that battery pack manufacturing would reach full
economy of scale at an annual production volume of 450,000 packs in the year 2025. This
volume was based on the annual manufacturing volumes assumed by FEV in the teardown
analyses performed for the FRM analysis.
In BatPaC, when the user specifies a production volume of 450,000 for a given battery pack,
it means that the cost estimate for that specific pack is based on a dedicated manufacturing plant
that manufactures an annual volume of 450,000 of that identical pack. Since all of the packs
produced by the hypothetical plant are identical, it implies that the cost estimate is most
applicable to a situation in which the packs are intended to be used by a single manufacturer in a
single model of electrified vehicle.
The 2015 NAS report noted (p. 4-42, and Finding 7.3, p. 7-23) that the technology penetration
levels projected by the agencies for electrified vehicles are lower than the 450,000 annual
production volume that the agencies assumed in projecting battery pack costs for the 2022 to
2025 time frame. Further, it noted that whatever annual production did occur would likely be
divided among multiple manufacturers and multiple models, preventing the full economy of
scale of 450,000 units from being achieved by any single manufacturer. The report
recommended that the agencies use a smaller manufacturing volume for electrified vehicle
battery packs to better reflect projected technology penetration, rather than the 450,000 annual
production assumed in the 2012 FRM.
Despite EPA's use of an annual production of 450,000 units, it is unclear whether this results
in more optimistic estimates of battery cost than the industry may realize. The following
discussion describes several points relevant to this consideration: (a) the potential for a "flex
plant" manufacturing approach to realize economy of scale at much lower pack volumes; (b) the
potential for economies of scale to fully develop at production volumes at low as 60,000; (c)
examples of actual costs that are already lower than EPA's FRM estimates at a much lower
production volume than 450,000; (d)EPA's placement of estimated costs in the year 2025 instead
of 2020; and (e) the potential for consolidation in the battery industry to increase pack
manufacturing volumes.
There is evidence that optimizing the approach to battery manufacturing by adopting a "flex
plant" approach may allow economies of scale to be realized at pack production volumes much
lower than 450,000. According to a recent ANL study,364 a battery manufacturing plant that is
designed to simultaneously manufacture packs for multiple vehicle types (HEVs, PHEVs and
BEVs) by standardizing on a single electrode width can significantly reduce the pack
manufacturing volumes required to achieve maximum economy of scale. The ANL study calls
this approach a "flex plant." Some manufacturers already appear to be adopting a similar
approach for production of prismatic cells. For example, at AABC 2015, Samsung SDI described
a strategy to build an "ecosystem" of xEV battery products by maintaining a "standard cell
format between generations," that is, by maintaining the same cell dimensions and container size
2-123
-------
Technology Cost, Effectiveness, and Lead Time Assessment
and achieving different target capacities by varying the chemistry.321 At the same conference,
Bosch similarly described a goal to produce packs of varying capacity by use of a standard 36
Ampere-hour cell.331 XALT Energy also described its practice of achieving variable cell
capacity (Ampere-hour) sizes by adjusting the electrode count within a cell while maintaining
one of two fixed cell footprint areas.365 Cell standardization also may promote the economics of
battery second life applications366 and so could provide an added motivation for manufacturers to
reduce the number of cell formats. EPA anticipates that the most successful suppliers may
continue to adopt similar approaches over time. As this occurs, the production volume of the
individual cells that compose the several pack types produced from those cells would increase
dramatically, even though pack volume of any single pack type may remain relatively low. This
increased cell volume may recapture much of the economy of scale reflected at the pack level in
the 450,000 unit assumption.
There is also some evidence to suggest that economies of scale may be achieved at much
smaller pack production volumes than 450,000, even without necessarily adopting a flex plant
approach. According to the ANL flex plant study, the benefits of a flex plant over a dedicated
plant for reducing the cost of BEV batteries levels off past a production level of about 60,000
units per year, suggesting that 60,000 units would approach maximum economy of scale for a
dedicated plant. The 2015 NAS report (p. 4-42), in noting that the agencies' projected costs for
2012 "seem reasonable" despite the large volume assumed, cites as a possible explanation a
TIAX study (referred to as Sriramulu & Barnett 2013 in a National Research Council report on
Overcoming Barriers to EV Deployment243) that also suggests a 60,000 unit volume at which
economies of scale would be realized. This level of production is much closer to the technology
penetration levels EPA predicts. Individual manufacturers such as Nissan and Tesla are already
approaching similar production levels, with Nissan having sold more than 30,000 Leaf EVs in
North America in 2014, and Tesla projecting a similar amount in 2015. The BMW i3 and i8
PHEVs are also approaching a global production level of 30,000 units per year.
There is also evidence that actual battery pack costs experienced by some manufacturers are
already lower than EPA's FRM estimates, at a much lower production volume than 450,000. As
discussed in more detail below, General Motors has cited its rapidly falling battery cell costs
from supplier LG Chem as evidence of their being "able to achieve lower costs earlier with much
less capital and volume dependency" than presumably had been expected. The cell-level costs
cited by GM for the Chevy Bolt are lower than the BEV pack costs projected by the agencies in
2012. Because it appears to suggest a currently contracted price applicable at the very beginning
of the Bolt product cycle, it therefore is likely to be based on an annual production level of far
less than 450,000 packs. Production of the 2017 Bolt has been characterized as capable of
serving a demand of around 50,000 units per year.367
The way EPA applies the BatPaC-generated costs also treats them conservatively. Although
the cost estimates generated by BatPaC are intended by its authors to represent technology being
used in the year 2020, EPA assigns these costs to the year 2025 when applying reverse-learning
to generate year-by-year cost estimates for earlier years. Although this was a practical choice in
order to cover the full time frame of the standards which run to MY2025, it has the effect of
making the projected costs more conservative by assuming that the technology projected by the
BatPaC authors will not take effect for an additional five years.
2-124
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Consolidation among battery cell suppliers may also improve the ability for individual
suppliers to begin approaching the production volumes assumed in the analysis. Since the FRM,
there has been significant consolidation among battery manufacturers.368'369'370 For example,
A123 Systems, which at one time competed against LG Chem to supply battery cells for the
Chevy Volt and was later chosen to supply the Fisker Karma and Chevy Spark, filed for
bankruptcy in late 2012 and was sold to Chinese auto supplier Wanxiang in 20 1 3.371 Wanxiang
has since refocused A123's efforts toward smaller HEV and stop-start batteries as well as grid
storage. Johnson Controls, which was ranked in second place as an industry leader by one
analysis firm in 2013,370 also has refocused its effort on smaller batteries. As of late 2015, three
xEV cell suppliers appear to have been particularly successful at developing OEM partnerships:
LG Chem, Panasonic, and Samsung SDI.372 LG Chem has grown its customer list to include not
only GM but also Renault, Volvo, Daimler, Volkswagen, Audi, and Tesla.373 Panasonic is also a
dominant player through its ongoing partnership with Tesla, as well as supplying smaller
contracts with Ford and Volkswagen. Samsung SDI is a supplier to BMW and in 2015
announced plans to acquire the battery division of Magna International.374 Nissan's joint-venture
arm Automotive Energy Supply Corporation (AESC) is also an important player through its
battery production for Nissan and Renault vehicles, including the Nissan Leaf. In 2015 it was
reported that Nissan is also considering a partnership with LG Chem for its future BEV
batteries.375 Even Tesla, which has long-term plans to source cells from its so-called
Gigafactory, is said to be investigating the possibility of sourcing cells from other leading
suppliers in order to meet expected demand for the Model 3 in a timely manner.376
As discussed in the Draft TAR, EPA believes that an assumed manufacturing volume of
450,000, as a BatPaC input, is appropriate for the purpose of generating battery pack cost
estimates applicable to the 2022 to 2025 time frame.
Some public comments on the Draft TAR addressed EPA's manufacturing volume
assumptions in the Draft TAR analysis. Comments on this topic are were considered and are
addressed in Chapter 2.3.4.3.7.4 (Assumptions and Inputs to BatPaC).
2.2.4.5.8 Potential Impact of Lithium Demand on Battery Cost
At circa-2010 prices, the cost of lithium content was said to be only about 1 percent of total
material cost at the battery pack level377 or perhaps 2 percent at the cell level.378 Lithium
comprises a similar percentage by mass, and at time of manufacture resides primarily as ions in
the cathode active material and the electrolyte solution.
Lithium used in cell manufacturing is most commonly sourced as lithium carbonate.379
Lithium carbonate is primarily recovered from ancient continental brines underlying salt lake
deposits. These are widespread in the southern Andes (primarily Bolivia, Argentina, Chile) and
western China and Tibet, with deposits identified in the southwest United States as well. Brine
mining operations are found or are under development in many of these areas. Lithium may also
be recovered from some oilfield brines in the western U.S. Some lithium is also recovered from
hard-rock deposits, particularly in Australia.380'381
Controversy has periodically arisen about the adequacy of known lithium reserves to service
the potential demand generated by the electrified vehicle industry. Because industrial
applications for lithium were relatively few and scattered prior to its use in batteries, known
reserves may not be as well enumerated as for other commodities, and may have potential to
2-125
-------
Technology Cost, Effectiveness, and Lead Time Assessment
increase as demand increases and previously unidentified or unexploited sources are recognized.
Recently, concerns about lithium prices have been renewed by a significant increase in the price
of lithium, thought to be resulting in part from increased demand for use in electrified
vehicles.382'383 Pressure also appears to be increasing on manufacturers to secure lithium sources
that will be needed to supply increased production capacity.384
However, lithium appears to be plentiful enough at this time to suggest that its availability
will not be a constraint in the near term.385'386 A study released by Carnegie-Mellon University in
May 20 1 6387 addressed this issue directly by examining the sensitivity of battery cell
manufacturing cost to the price of lithium carbonate and lithium hydroxide. The study
concluded that the effect on battery pricing would be minimal (never more than 10 percent) even
for the most extreme lithium price fluctuations considered (about four times the historical
average). The researchers also suggested that the primary difficulty imposed by such
fluctuations would be felt by cell manufacturers in maintaining profit margins, rather than by
vehicle manufacturers or consumers. Development of new lithium resources is being actively
undertaken in many areas across the world.
2.2.4.5.9 Evaluation of Draft TAR Battery Cost Projections
As described in the Draft TAR, EPA has adopted a bottom-up, bill-of-materials approach to
projecting the future DMC of xEV batteries by using the ANL BatPaC battery cost model.135 As
discussed in the Technical Support Document (TSD) accompanying the 2012 FRM,388 battery
pack costs projected by this model were shown to compare favorably with cost projections
provided by suppliers and OEMs that were interviewed during development of the rule. In the
2015 NAS report (Finding 4.4, p. 4-43), the committee found that "the battery cost estimates
used by the agencies are broadly accurate," providing further support for the use of this model.
The Draft TAR examined several sources that had emerged since the FRM that provide
additional information on the evolution of battery costs and potential future trends.
In 2015, a peer-reviewed journal article (Nykvist and Nilsson, 2015) appeared that provides a
comprehensive review of over 80 public sources of battery cost projections for BEVs.153 Based
on a statistical analysis of these estimates, it was shown that industry cost estimates for lithium-
ion batteries for BEVs have declined 14 percent annually between 2007 and 2014, and that pack
costs applicable to leading BEV manufacturers have followed a cost reduction curve of about 8
percent per year, with a learning rate of between 6 percent and 9 percent. The authors concluded
that the battery costs experienced by market leading OEMs are significantly lower than
previously predicted, and that battery costs may be expected to continue declining.
In Figure 2.36, the full population of cost estimates reviewed by Nykvist and Nilsson is
compared to the battery pack cost projections of the Draft TAR analysis. Because BatPaC does
not produce cost estimates for multiple years, the OMEGA analysis applies a learning curve to
generate costs for the years 2017 through 2025, with BatPaC output costs assigned to the year
2025. The learning-adjusted costs shown in the figure include those for PHEV40, BEV75,
BEV100, and BEV200 (Draft TAR). These vehicle types have relatively large battery capacities
similar to those included in the review. The plot shows that the battery costs per kWh projected
in the Draft TAR (shown as green circles) fit well with the reviewed estimates (orange squares),
and lie on a similar cost reduction curve.
2-126
-------
Technology Cost, Effectiveness, and Lead Time Assessment
1600
~ D ~
~ 0
u ~
^ 1200 -O-
J DS ~
3 1000 1— ~ ~ ~
^ I n 1=1
I DDB0d
— 800 ~ D ~
^ ~~ o Draft TAR
| g ~
600 B m i-H a ~ Nykvist/Milsson
£ ~ ~~~ °Rea
tt 1=1 H i B
S 400 ~ —
~D 1 i 1 §iii i
0 '—
2000 2005 2010 2015 2020 2025 2030 2035
Year
Figure 2.36 Comparison of Draft TAR Projected Battery Cost per kWh to Estimates Reviewed by Nykvist &
Nilsson
Cost estimates and projections are most useful when they can be validated by comparison to
actual costs. Unfortunately, information about actual battery costs being paid or under contract
by manufacturers for production vehicles is rarely disclosed publicly. However, when General
Motors publicly commented on its battery costs for the Chevy Bolt EV (a BEV200) in October
2015, it provided a valuable opportunity to evaluate the 2012 FRM projections of BEV200
battery costs, as well as those projected by the Draft TAR analysis.
General Motors held its Global Business Conference on Oct. 1, 2015, where various speakers
described to an investor audience its current development status and plans with regard to various
advanced vehicle technologies. In a presentation on electrification, GM disclosed its projected
cost per kWh (on a cell basis) for battery cells for the Chevy Bolt EV. Citing partnership with
cell manufacturer LG Chem, Executive Vice President of Global Product Development Mark
Reuss stated, "When we launch the Bolt, we will have a cost per kWh of $145, and eventually
we will get our cost down to about $100. We believe we will have the lowest cell cost with
much less capital and volume dependency."389 An accompanying chart shows the $145 cost
continuing to 2019, dropping to $120 per kWh in 2020 and to $100 per kWh in 2022.390-391
It is important to note that the costs described above are cell-level costs and not pack4evel
costs. To compare them to the pack4evel costs that EPA projects in this analysis requires
converting them to that basis using an appropriate methodology. Also, although the context of
the announcement suggests that the costs are comparable to a direct manufacturing cost, their
exact basis is unknown. Although these factors introduce some uncertainty in comparing the
announced costs to the EPA projections, a qualified comparison is possible.
~
~
~
~ B
u ~
n
dbd
n n
~
n
DDa0n
n u n
~ ~
:~ rasa—
EEffl
~
~
n n
tl
~ B g
— 1 1
i
~ o
~ ~
~
J
~~~
~ m
I
1
§
1
i
ii! |
11
0 1
3 ~
2-127
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Several sources exist that suggest a cost conversion factor from cell-level costs to pack-level
costs for lithium-ion batteries.392'316,289'393'394,395 These are summarized in Table 2.7. Most of
these sources suggest a conversion factor of about 1.25 to 1.4.
Table 2.7 also includes two estimates that EPA derived from the ANL BatPaC model for a
liquid-cooled BEV-sized pack at a production volume of 50,000 to 100,000. Outputs from this
model suggest that the ratio of pack-level cost to cell-level cost for the pack format modeled by
BatPaC may range from about 1.5 for a 16 kWh pack to about 1.3 for a 32 kWh pack, and
continuing to decrease for larger pack capacities.
Table 2.7 Examples of Conversion Factors for Cell Costs to Pack Costs
Source
Low
High
Kalhammer et al.392
1.24
1.4
Element Energy316
1.6
1.85
Konekamp289
1.29LL
USABC393
1.25MM
Tataria/Lopez394
1.26NN
Keller395
1.200
BatPaC, 16 kWh
1.5
BatPaC, 32 kWh
1.3
On the basis of the BatPaC-derived ratios of 1.3 to 1.5, the 2015-2019 cell-level figure of
$145 per kWh would translate to approximately $190 to $220 per kWh on a pack level. The
future projections of $120 and $100 per cell kWh in 2020 and 2022 would translate to
approximately $156-$180 per kWh and $130-$150 per kWh at the pack level, respectively. On
this pack-converted basis the GM cell costs agree well with the BatPaC cost projections (which
the Draft TAR analysis applies to 2025).
Table 2.8 compares the estimated pack-level equivalents of the GM cell costs to the projected
BEV150/200 pack-level costs of the 2012 FRM and Draft TAR analyses. The pack-converted
GM projection (for 2020), at $156-$ 180 per kWh, compares well to the Draft TAR costs for
BEV200 (for 2025), which ranged from $160 to $175 per kWh. Similarly, even though the Draft
TAR projected costs are significantly lower than the FRM projected costs, they are very similar
to the GM pack-converted costs for 2022. Assuming that the GM pack-converted costs are
reasonably comparable to the EPA projected costs, this tends to support the Draft TAR
projections. Further, it should be noted that the EPA costs are projected using an annual volume
of 450,000 units and are attributed to the year 2025. This tends to make the EPA projections
more conservative, because the GM figures are supposed to be achieved in earlier years, and are
likely to be predicated on much smaller annual production volumes.
LL Cell cost = 620 Euros* 16 modules = 9,920 Euros; pack cost = 12,800 Euros; 12,800/9,920 = 1.29.
USABC 2020 goals for advanced EV batteries cite a cost of $125/kWh at pack level and $100/kWh at cell level
= 1.25.
For a 40 kWh pack, cell costs estimated at $258/kWh; pack-related costs at $2,626, or $66 per kWh;
(258+66)/258 = 1.26.
00 Cites one goal of 21st Century Truck Partnership as "Cost of overall battery pack should not exceed cost of the
cells by more than 20% by 2016" (slide 6).
2-128
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.8 Comparison of GM/LG Chem Pack-Converted Cell Costs to FRM BEV150 Pack Cost
Pack Cost/kWh (2015$)
Source of Estimate
Year Applicable
Low
High
BEV150 in FRM
2025
$160
$175
BEV200 in Draft TAR
2025
$120
$160
GM/LG Global Business Conference
2015-2019
$190
$220
2020
$156
$180
2022
$130
$150
Figure 2.37 compares the pack-converted GM costs to the year-by-year learning-adjusted
costs used in the 2012 FRM and Draft TAR for Small, Standard, and Large Car BEV150 and
BEV200. It can be seen that the range of the pack-converted GM costs (solid orange lines) is
much lower than the costs predicted by the 2012 FRM analysis (solid gray dots). The costs
projected for the Draft TAR analysis (blue circles) are much closer to the pack-converted GM
costs, and in some cases intersect with the line representing a 1.5x cell-to-pack conversion factor.
Based on the BatPaC-derived conversion factors for large BEV packs, the 1.3x line is probably a
more representative estimate than the 1.5x line due to larger pack size. All of the Draft TAR
estimates are above the 1.3x line, suggesting that the Draft TAR projections continue to be
conservative relative the pack-converted GM costs. Of course, it is uncertain whether the GM
costs are directly comparable because it is unknown to what extent those costs represent direct
manufacturing costs output by BatPaC. However, with these qualifications, this comparison
provides a valuable perspective on the Draft TAR projected costs for EV200.
400
35#
9
[i
[»
0
D
50
I*1
2026
Year
Figure 2.37 Comparison of Estimated GM/LG Pack-Level Costs to 2012 FRM and Draft TAR Estimates for
BEV150/200
As discussed in the Draft TAR, at the time of the FRM, EPA's battery cost estimates appeared
to be lower than costs being reported by many suppliers and OEMs at the time, and also lower
2-129
-------
Technology Cost, Effectiveness, and Lead Time Assessment
than some independent estimates said to be applicable to the time frame of the rule. EPA chose
to place confidence in the peer-reviewed ANL BatPaC model due to its rigorous, bottom-up
approach to battery pack costing, and the expertise of leading battery research scientists that
contributed to its development. The comparisons described above suggest that this approach was
effective and may in fact have been conservative not only with respect to characterizing the pace
of reductions in battery cost that have taken place in the time since the FRM but also to
projecting future costs for the 2020 to 2025 time frame. Up to and including the development of
this Proposed Determination analysis, EPA has continued to invest significant resources into
understanding developments and emerging trends in battery technologies so that these critically
important projections of xEV battery cost may be as reliable as possible.
While other public examples of battery costs to manufacturers remain elusive, several
suppliers and manufacturers have made battery-related product announcements since the FRM.
Some of these include information suggestive of battery costs or pricing. Some manufacturers
have published pricing for battery replacement parts or upgrades available to authorized service
providers. Others have offered different options, such as battery size or purchase method, the
relative pricing of which may suggest a relationship to battery cost. Finally, stand-alone non-
automotive Li-ion battery packs are beginning to become available to end users and their pricing
may be informative. While EPA recognizes that the pricing of these early-stage product
offerings may be subsidized by their manufacturers for competitive and marketing reasons, these
announcements may still be relevant to understanding the evolution of battery pack costs as these
products increase their presence in the market.
In 2013-2014, Tesla Motors offered the Model S in two battery pack sizes, 60 kWh and 85
kWh, at retail prices of around $69,900 and $79,900, respectively. Assuming no content
difference between the two versions, the retail price differential would suggest a battery cost of
$10,000 / 25 kWh = $400/kWh. An alternate analysis presented by Nykvist et al.396 subtracts the
estimated value of added content found in the 85 kWh version (Supercharger, premium tires, and
associated markup), resulting in a net price difference of $8,500 or $340 per kWh.
In July 2014, Nissan announced the replacement cost of a 24-kWh battery for the Nissan Leaf
at $5499 with core return, which amounts to about $229/kWh net. Although Nissan requires
return of the original battery (core), a $1000 credit is then applied for the core, suggesting a full
retail price of $6499, or $271/kWh.397'398'399 Later the same month, Nissan followed up by
pointing out that the quoted price is in fact subsidized by Nissan, although they declined to report
the amount of subsidy or the actual manufacturing cost.400 Nissan does not allow purchase of the
battery except as a Leaf battery replacement.
In 2015, an independent vendor of OEM parts listed the 2011 Chevy Volt battery pack at
$10,208 list price, discounted to $7,228, with no mention of core exchange. Assuming a 16 kWh
capacity, these prices would value the battery at $638/kWh and $452/kWh, respectively.
Although the product was listed and priced by the vendor, it was on restriction from ordering for
reasons that remain unclear.401'402
In January 2015, it was reported that the MSRP for a BMW i3 battery pack module was listed
at $1,805.89, each module being 2.7 kWh (21.6 kWh total divided by 8 modules). This module
price would equate to $669/kWh. A specific dealer was reported to be offering the module at a
price of $1715.60, or $635/kWh 403
2-130
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In September 2015, Tesla announced the price for a range-increasing battery pack upgrade for
the Tesla Roadster at $29,000, including installation and logistics. Tesla indicated that the
quoted price is meant to be equal to Tesla's expected cost in providing the pack, and disclaimed
any intention to make a profit. Tesla also indicated that the price per kWh is higher than for a
Model S battery due to the low volume production expected for the Roadster upgrade pack (only
approximately 2,500 Roadsters were produced). Tesla did not list the kWh capacity of the
upgrade pack, but describes it as having approximately 40 percent more energy capacity than the
original Roadster pack, which is commonly listed as 56 kWh. This suggests that Tesla's cost for
low volume production of this pack is around $29,000/(56*1.4) = $370 per kWh 404 In October
2015, Tesla further announced that the Roadster upgrade packs would be provided through a
partnership with LG Chem.405 This suggests that the price of the pack may not reflect
anticipated savings from the Panasonic-Tesla "Gigafactory" partnership.
In August 2013, the Smart ED was offered with a 17.6 kWh battery, with the option to either
purchase the battery with the car, or lease it separately. The vehicle price was $5,010 lower
without the battery when the battery was leased at a price of $80/mo. If the $5,010 differential
was taken to represent the incremental cost of the battery, it would value the battery at
$285/kWh. Of course, the present value of the lease payments would also contribute value to the
transaction, and it is possible that marketing considerations could also be represented in the
pricing.406'407'408
In September 2015, Nissan announced pricing in the UK for the 2016 Nissan Leaf. In a press
release from Nissan, equivalent versions of the Leaf having a 30 kWh pack instead of a 24 kWh
pack were priced at a difference of 1,600 British pounds. This would amount to approximately
267 British pounds per kWh, or U.S. $411 per kWh (assuming an exchange rate of 1.54 U.S.
dollars per pound). It should be noted, however, that although the two versions of the pack
appear to be designed to install into the same footprint and volume, any cost comparison is
potentially complicated by differences in chemistry and construction of the two versions.409
In 2014, Tesla Motors began construction of a so-called "Gigafactory" in Nevada in
partnership with Panasonic. This factory is commonly cited by Tesla as enabling a potential 30
percent reduction in battery pack costs from the levels Tesla currently pays. According to one
analysis,410 Tesla's current cost is estimated at about $274 per kWh. A 30 percent reduction on
that figure would bring costs to about $192 per kWh.
In April 2015, Tesla announced a home battery pack product called Powerwall, pricing a 7
kWh version at $3,000 ($428/kWh) and a 10 kWh version at $3,500 ($350/kWh). Although
designed for stationary home use, the pack design bears similarities to automotive packs, being
liquid-cooled and using similar chemistries. The 7 kWh version employs NMC chemistry
similar to many production BEVs, while the 10 kWh version employs the NCA chemistry like
the Tesla Model S. Tesla also announced a similar product called Powerpack for commercial
use. Powerpack was said to be priced at $25,000 for 100 kWh capacity, or $250/kWh. These
products are expected to take advantage of much of the cell output of the Gigafactory, suggesting
that these products may be priced in anticipation of the cost reductions it is expected to achieve.
Table 2.9 summarizes the estimated cost or pricing information derived from the foregoing
examples.
2-131
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.9 Summary of Published Evidence of Battery Pack Cost and Pricing
Pack Cost or Price
per kWh
Source of Evidence
Year Applicable
High
Low
Tesla Model S 60 kWh vs 85 kWh comparison
2013-2014
$340
$400
Nissan 24 kWh replacement pricing
2015
$229
$271
Vendor pricing for 2011 Volt pack
2015
$432
$638
Dealer pricing for BMW i3 module
2015
$635
$669
Tesla Roadster upgrade pricing
2015
$370
Smart ED lease vs buy pricing
2013
$285
Nissan UK price differential 30 kWh vs 24 kWh
2015
$411
Tesla Lux Research estimate
2014
$274
Tesla Lux Research estimate modified by Gigafactory
2017
$192
Tesla Powerwall
2015-2016
$350
$428
Tesla Powerpack
2015-2016
$250
It is important to remember that the figures derived from these examples should be interpreted
with caution. EPA's cost projections represent direct manufacturing costs and not retail pricing.
Also, as previously noted, retail pricing of these early-stage product offerings may be subsidized
by their manufacturers and may reflect competitive and marketing considerations that further
obscure their true manufacturing cost. Furthermore, some of the estimates are derived from full-
product comparisons that may or may not accurately represent the battery portion of the
comparison. It should also be noted that the examples presented here represent current pricing,
while the EPA analysis applies its BatPaC cost projections to the year 2025.
The Draft TAR noted that the existence of these examples shows that the industry has
progressed considerably since the 2012 FRM, when such examples were almost entirely
unknown. The identification and packaging of specific battery products for upgrade,
replacement or standalone use is a significant development and suggests that the industry is
continuing to gain in maturity and is growing along multiple paths. The establishment of MSRPs
for many of these products also suggests that manufacturers are beginning to gain confidence in
their understanding of the cost structure of battery products. The examples and estimates derived
from this analysis, even if approximate, can serve to help ground the various cost estimates and
projections that continue to comprise a very active area of research throughout the battery
industry, its customer base and other stakeholders.
2.2.4.6 Fuel Cell Electric Vehicles
Fuel cell electric vehicles (FCEVs) are an emerging form of electrified vehicle having a fully
electric powertrain, and are distinguished from BEVs by the use of a fuel cell system rather than
grid power as the primary energy source.
FCEVs have only recently entered commercial production, and their market has not yet
developed as fully as that of PEVs. Currently, three automakers (Hyundai, Toyota, and Honda)
have begun to offer fuel cell vehicles to the mass consumer market or announced specific near-
term plans for market launch. Hyundai has offered its Tucson Fuel Cell for lease in select regions
of southern California since 2014. Toyota offers its Mirai sedan in at least eight dealerships
across both northern and southern California with options for both lease and purchase. Honda
2-132
-------
Technology Cost, Effectiveness, and Lead Time Assessment
has recently released its production Clarity Fuel Cell in 2016. Other automakers are known to be
involved in the development of FCEV technology and expected to be moving towards
commercial production, but have not yet made public announcements of production models or
release dates.
Technology developments relating to FCEVs were reviewed in detail in Draft TAR Section
5.2.4.5. Because EPA did not include FCEVs in its fleet compliance modeling analysis for the
Draft TAR nor for this Proposed Determination, please refer to the Draft TAR for additional
information on this technology.
2.2.5 Aerodynamics: State of Technology
This section provides an overview of technologies that improve vehicle aerodynamic
performance. The focus on vehicle aerodynamics has a long history stemming from the
recognition of the relationship between aerodynamic drag and energy consumption. Section
2.2.5.1 outlines the significance of aerodynamic drag and some of the related physical principles
and technologies. Section 2.2.5.2, discusses developments in the light-duty vehicle industry to
reduce aerodynamic drag, including examples of some recent vehicle introductions. Section
2.2.5.3 focuses on an assessment of the amount of aerodynamic drag improvements that have
been implemented by manufacturers in the light-duty fleet as of MY2015. This assessment is in
direct response to comments received from the AAM. Section 2.2.5.4 discusses the off-cycle
benefits of improved aerodynamic performance. Section 2.2.5.5 discusses the aerodynamics
research performed in collaboration with Transport Canada in support of the Draft TAR and this
Proposed Determination.
2.2.5.1 Background
Aerodynamic drag accounts for a significant portion of the energy consumed by a vehicle,
particularly at higher speeds. Reducing aerodynamic drag can therefore be an effective way to
reduce fuel consumption and GHG emissions.
The force imposed by aerodynamic drag results from the flow of air around the vehicle.
Aerodynamic performance is thus intimately related to the shape of the vehicle; specifically, it is
commonly represented by the product of its cross sectional area as viewed from the front (known
as frontal area, or A) and the coefficient of drag (Cd). The product of the two, CdA, is also known
as the drag area of a vehicle. The force imposed by aerodynamic drag increases with the square
of vehicle velocity, accounting for its dominance at higher speeds.
The coefficient of drag Cd is a dimensionless value that essentially represents the aerodynamic
efficiency of the vehicle shape. The frontal area acts with the coefficient of drag as a sort of
scaling factor, representing the relative size of the vehicle shape that the coefficient of drag
describes.
Cd and A are determined by the design of the vehicle, and so represent the primary design
paths for reduction of aerodynamic drag. The greatest opportunity for improving aerodynamic
performance is during a vehicle redesign cycle, when the best opportunity exists to make
2-133
-------
Technology Cost, Effectiveness, and Lead Time Assessment
significant changes to the shape or size of the vehicle.pp Incremental improvements may also be
achieved mid-cycle as part of a model refresh through the use of revised exterior components
and add-on devices. Some examples of these technologies include revised front and rear fascias,
modified front air dams and rear valances, addition of rear deck lips and underbody panels, and
low-drag exterior mirrors.
Aerodynamic technologies can be divided into passive and active technologies. Passive
aerodynamics refers to aerodynamic attributes that are inherent to the shape and size of the
vehicle, including any components of a fixed nature. Active aerodynamics refers to technologies
that variably deploy in response to driving conditions. These include technologies such as active
grille shutters, active air dams and active ride height adjustment.
Significant variations in CdA can be observed across vehicle classes and among individual
vehicles within a class.411'412'413 Within a class, drag coefficients tend to vary more than frontal
areas. Frontal areas are in part a function of interior passenger and cargo space, and therefore
tend to track with the interior space expectations associated with a vehicle class. In contrast,
drag coefficients are largely a function of body styling and airflow management and may vary
significantly with changes in shape and exterior treatment.
As is the case with many technologies that improve vehicle efficiency, manufacturers have a
wide selection of technologies for improving aerodynamic performance. These include both
passive components, such as body shapes, air dams and underbody panels, and active systems
such as grille shutters and adjustable suspensions. In addition, manufacturers have robust
development tools based on wind tunnels, clay models, and computational fluid dynamics (CFD)
techniques that allow the evaluation of aerodynamic treatments in advance of the creation of
physical prototypes. This allows a manufacturer to set aerodynamic targets at the beginning of a
vehicle program and simulate multiple alternative vehicle designs to determine which design has
the best opportunity to meet the target.
2.2.5.2 Industry Developments
Many vehicle manufacturers have placed emphasis on reducing aerodynamic drag as a means
of improving overall vehicle efficiency. While many of the passive and active technologies that
EPA identified in the 2012 FRM are not yet found on the entire fleet, the industry is increasingly
adopting both types of technologies.
In January 2015, EPA staff attended the 2015 North American International Auto Show
(NAIAS) in order to gather information about the state of implementation of various
aerodynamic technologies in the vehicles represented at the show. A total of 76 vehicles that
appeared to employ aerodynamic devices were viewed, across more than a dozen manufacturers.
A memorandum414 describing this informal survey is available in EPA Docket EPA-HQ-OAR-
2015-0827. Although the sample was collected informally and therefore was not random, the
information gathered provides some insight into recent industry activity in the application of
aerodynamic technology to light-duty vehicles. Table 2.10 shows a breakdown of the
aerodynamic devices and technologies that were observed in these vehicles.
pp Changes in size are less preferable as a pathway to a reduced CdA due to the change in utility (e.g., interior space)
this may imply.
2-134
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.10 Aerodynamic Technologies Observed in Vehicles Investigated at the 2015 NAIAS
Technology
Number of
Percentage
vehicles
equipped
equipped
Active Grill Shutters
14
18%
Underbody Panels
front (full)
28
37%
front
22
29%
(partial)
middle or
27
36%
side
rear
2
3%
Wheel Dams
Front
56
74%
Rear
59
78%
Front Bumper Air Dam
18
24%
Total vehicles inspected
76
This informal survey suggests that manufacturers are implementing both passive aerodynamic
devices (panels and dams) and active devices (active grill shutters), as permitted by the various
levels of model refresh or vehicle redesign represented in the surveyed vehicles. Further
opportunity for more optimized applications of both passive and active aerodynamic
technologies is likely to occur as these and other vehicles enter further model refresh or redesign
cycles. Besides active grill shutters, other active technologies, such as active ride height or wheel
shutters, were not observed in this survey. This could indicate that manufacturers have so far
focused on the most cost-effective technologies. Active technologies not yet implemented
remain available as additional options for further reducing aerodynamic drag in the future.
Optimizing airflow under the vehicle is an important aspect of improving aerodynamic drag,
and is being addressed in a growing number of vehicles by the addition of underbody panels. As
indicated by the informal survey, many vehicles already include partial underbody panels
covering a portion of the underbody, typically where they would not interfere with mechanical
access or exhaust cooling. With careful consideration of access and cooling needs, in many
cases, most of the underbody may potentially be streamlined in this way. For example, the Audi
R8 includes extensive underbody panels covering almost the entire underbody. 415
Redesign cycles often present increased opportunities for aerodynamic improvement beyond
what is possible in a model refresh. While the 2004 Prius was widely reported as having
achieved a very low drag coefficient of 0.26, the 2017 Prius achieves 0.24.416 Its styling lines,
stabilizing fins and underbody panels, supplemented by an active grill shutter, all work together
to help reduce aerodynamic drag, providing an example of how whole body analysis can often
help maximize the potential for drag reduction even in a vehicle that is already quite
aerodynamically efficient.
Another example of optimized application of aerodynamic technology enabled by a redesign
cycle can be seen in the 2015 Nissan Murano. Nissan's goal in the Murano effort was to achieve
a Cd of 0.31. Its exterior was completely redesigned from its previous 2008-era generation, with
the goal of minimizing drag by combining passive aerodynamic devices with an optimized
vehicle shape. The development process included 20-percent-scale wind tunnel testing as well as
full scale wind tunnel testing and CFD simulations. 417
2-135
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The aerodynamic features of the Nissan Murano are listed in Table 2.11. The primary passive
devices employed include optimization of the rear end shape to reduce rear end drag, and
addition of a large front spoiler to reduce underbody air flow and redirect it toward the roof of
the vehicle, thus augmenting the rear end drag improvements. Other passive improvements
include plastic fillet moldings at the wheel arches, raising of the rear edge of the hood, shaping
of the windshield molding and front pillars, engine under-cover and floor cover, and air
deflectors at the rear wheel wells. An active lower grille shutter also redirects air over the body
when closed. Together, these measures give the 2015 model a drag coefficient of 0.31,
representing a 16 to 17 percent improvement over the 0.37 Cd of the previous model.417'418
Table 2.11 Aerodynamic Features of the 2015 Nissan Murano
Design417
Detail
Ideal Flow Features
Minimum airflow into engine compartment
Reduces resistance (just enough to cool)
Airflow under front bumper toward underbody
Reduce as much flow as possible underbody for resistance is caused by the uneven
minimized
floor
Flow around ends of front bumper toward body sides
Reduce drag, covers front of front tires
Airflow at front wheel arches is routed alongside
Reduce resistance that occurs at the front surfaces of the tires
surfaces of front tires
Separation angle at rear of hood is large
Minimize resistance by reducing pressure at low end of windshield, 'hide'
windshield wipers and reduce rain droplets in area of airflow
Smooth area at front pillars toward body sides
Vertical vortices are minimized to reduce drag
Optimize of the rear end shape
Assure clean separation of airflow from rearto minimize drag, and equate velocity
of airflow from over roof and along body sizes as much as possible to minimize
vortices.
Floor-lower bottom edge of front bumper
Reduces airflow toward underbody, route airflow toward vehicle rear in straight
path to min flow resistance by uneven floor.
Airflow at front of wheelhouses is minimized and wheelhouse design is optimized
to direct rearward the air trapped inside - all to reduce resistance at back of the
wheel arches.
Computational Fluid Dynamics (CFD) Simulations (80 simulations)
Active Lower grille shutter at lower opening
Redirects air over the body when closed
Higher opening allows sufficient air when grill shutter closed
Duct type structure is used to provide direction to the airflow to the heat
exchanger and minimize entry into engine compartment elsewhere
Large front spoiler beneath front bumper
Reduces underbody airflow and redirect toward roof of the vehicle
Bottom edge is provided with a lip to increase the flow separation angle to further
reduce airflow under the body (same as if would further lower the bottom edge of
the front spoiler)
Plastic fillet moldings at the wheel arches
To assure airflows along the side surfaces of the front tires (avoid adjusting design
of front bumper ends)
Optimize shape of rear edge of hood
To promote separation by increasing flow separation angle, distance windshield
wipers from airflow, reduce collection of water droplets
Optimize windshield molding shape
To smooth for wind flow
Outside mirrors optimized for placement
Avoid airflow coming over rear edge of hood and lower edge of front pillar
Optimize shape of vehicle rear end
Shape of rear spoiler, rear combination lamps and rear bumper optimization.
Secure larger roof approach resulted in increased pressure recovery and reduced
drag by wake flow.
Overall vehicle shape and equal airflow
Balance roof flow and body side flow to reduce vortices
Design optimization to increase airflow to roof
Reduces rear drag caused by wake flow
Rear Spoiler part of roof approach
Tapered toward vehicle rear
Engine under-cover and floor cover
Covers beneath front bumper and over suspension links and muffler piping, raise
fuel tank, resulting in smooth underbody flow of air (not full cover)
Reduce airflow into wheelhouses
Large front spoiler extends as far as the front of the wheelhouses and deflectors
(optimally shaped) in front of the reartires, bottom of front spoiler lowered on
both sides as capable (governed by ground clearance)
Smoother fenders
Reduce gaps between closure panels
Small vortex-creators
Put vortices in desired places to minimize drag
2-136
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Despite the extensive use of drag reduction technology on this vehicle, the Murano does not
appear to include active ride height technology, which could represent an opportunity to reduce
drag even further. While the ride height of an SUV is typically higher than that of a passenger
car to provide for off-road capability, this increased ride height can reduce aerodynamic
performance. Active ride height technology reduces the ride height at highway speeds, when off-
road performance is unlikely to be necessary. These systems may adjust the ride height
downward as pre-established speed thresholds or other criteria are met, restoring the original ride
height at lower speeds.
An extensive study of advanced drag reduction technologies, including active ride height, has
been conducted by Transport Canada and National Research Council Canada.419 The study
suggests that the aerodynamics of even a highly aerodynamic SUV could potentially be
improved further by a front and rear ride height reduction of about 40 mm. Several additional
active techniques were also explored, including active grill shutters and active extension of the
OEM air dam. With grill shutters fully closed, OEM air dam extended 45 percent, and ride
height lowered by 40mm in front and rear, estimated Cd was reduced to 0.282 from a baseline of
0.314. Table 2.12 details the effect of other combinations explored in the study.
Table 2.12 Effect of Active Ride Height on SUV Aerodynamic Performance
Technology Package
Baseline Cd (0)
Cd (0 angle)
Difference
Shutters 100% open
0.314
Baseline (shutters 100% closed, OEM air dam extended
45%, baseline ride height)
0.295
-6%
Shutters 100% closed, OEM air dam extended 45%,
ride height 40mm down front and back
0.282
-4.4%
(total 10%)
-0.013
Baseline (shutters 100% closed, OEM air dam extended
30%, baseline ride height)
0.297
-5.4%
Shutters 100% closed, OEM air dam extended 30%,
ride height 40 mm down front and back
0.2802
-5.7%
(total 10.8%)
-0.017
In addition to reducing Cd, it is also possible to reduce drag losses by reducing frontal area.
While a reduced frontal area would seem to imply a loss of interior volume, the redesigned 2015
Acura TLX sedan shows that with thoughtful design, a reduction in frontal area need not
necessarily result in a reduction in interior space. In a 2015 presentation,420 Acura states that the
TLX was redesigned with the help of CFD as well as wind tunnel and real-world coast down
testing to achieve a 15 percent lower CdA compared to the 2012 Acura TL. This was achieved in
part by a reduction in frontal area of 1.5 percent (removing 0.5 inches in height and 1 inch in
width) that was described as not resulting in a sacrifice in interior space. Further improvements
were attributed to a sloped hood and a short rear deck. In addition, welds were eliminated from
the forward and rearward edges of the wheel arches by use of a roller hem wheel arch design in
place of spot welds, and smoothing transitions between body panels in this area.
The Chevrolet Cruze provides another example of the application of drag-reducing
technologies by a major manufacturer in a popular vehicle. The aerodynamic technologies on the
2011 Cruze included active air shutters in the lower grille opening, a front air dam, lower ride
height, underbody pans, tire blockers, and rear deck-lid spoiler. GM described these changes as
2-137
-------
Technology Cost, Effectiveness, and Lead Time Assessment
reducing the drag coefficient by more than 10 percent.421 This program of improvement appears
to have continued with the 2016 Cruze,422 which benefits from what GM describes as: faster
windshield rake, faster-sloping rear profile, a rear spoiler, "layered line work" in the hood and
body-side panels, headlamp sweep, mounting location of the center-rear stop lamp, and seamless
rocker panels. GM describes this vehicle as having a drag coefficient of 0.28.
As another example, the redesigned Ford F150 incorporated a number of aerodynamic
improvements over the previous, 2008-era design. However, some trim levels of the 2015 F150
are slightly larger in cross sectional area than the previous model, and as a result some of the
aerodynamic benefits may have been lost to this feature. This also indicates that the remaining
benefit of these improvements was achieved without loss of interior space. Extensive testing and
analysis led to the improved design, including CFD simulations and wind tunnel testing that,
according to Ford, allowed aerodynamic performance to be improved while "maintaining the
tough truck looks expected from F-150." 423 Some of the technologies on this vehicle include:
active grill shutters, underbody covers, canted headlamp and bumper end corners, flush-mounted
windshield, a tailgate top that acts as a rear spoiler, a cargo box narrower than the cab (without
reducing its volume), angled rear corners, and an air curtain enabled by a duct under the
headlamp channels, which minimizes turbulence from airflow around the vehicle.424'425,426
Replacement of side view mirrors with side view cameras is another potential drag-reducing
technology being considered by OEMs (for example, Tesla and BMW), but has not yet been
approved by NHTSA, which sets standards for safety-related equipment including rear- and side-
view mirrors. According to the NAS report, side-view mirror replacement with cameras can
reduce Cd by as much as 2 to 7 percent. In the interim, one way to reduce mirror drag is to
determine optimal placement and optimal design, as noted with respect to the aerodynamic
changes to the 2015 Nissan Murano.
2.2.5.3 Feasibility of Aerodynamic Improvements
Public comments on the Draft TAR included several comments regarding the feasibility of
aerodynamic improvements as represented by the Aerol and Aero2 technology cases assessed in
the Draft TAR. These cases represent a 10 percent and 20 percent improvement, respectively, in
aerodynamic performance from a baseline (2008-era) vehicle.
Some comments expressed concern with the representation of aerodynamic technology that
had already been applied by manufacturers to vehicles in the baseline fleet that was created for
the Draft TAR analysis. Specifically, there was a concern that every baseline vehicle was
considered to have no applied aerodynamic technology, allowing even vehicles that had achieved
above average aerodynamic performance to be considered eligible for up to a 20 percent
additional improvement. Commenters also suggested that aerodynamic potential should be
evaluated on the basis of Cd alone (rather than CdA, which would imply the possibility of a
reduction in interior volume), and that feasible limits on improvement of aerodynamic
performance should be recognized and observed.
In the Draft TAR, EPA indicated that it planned to "look at various vehicle categories and
examine the ... best and worst aerodynamically performing vehicles, using CdA as a metric," in
order to better consider "the remaining potential for aerodynamic improvement within [each]
category." EPA has proceeded with this effort by better representing aerodynamic technology
present in the baseline. More detail on this update is described in Chapter 2.3.4.4 of this TSD.
2-138
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In comments on the Draft TAR, Ford commented that the potential for aerodynamic
improvement is constrained by other considerations such as consumer desires and needs for
utility, space, and styling. While the pursuit of any engineering goal is constrained by competing
concerns, EPA continues to believe that manufacturers have a wide variety of technologies from
which to draw upon to pursue the reduction of drag losses, as appropriate to the functional
characteristics of the vehicle in question. It is not to be presumed that cargo vans, large SUVs or
light-duty pickup trucks should be expected to achieve the same potential aerodynamic
performance as passenger cars, but within a given segment, paths and opportunities exist to
pursue significant improvements as measured relative to less aerodynamically-optimized model
generations within the same segment.
2.2.5.4 Results of U.S.-Canada Joint Test Program
In 2013 a Joint Aerodynamics Assessment Program was initiated between Transport Canada
(TC), Environment and Climate Change Canada (ECCC), National Research Council (NRC) of
Canada, and EPA.411 This program was conducted in four phases over three years, and examined
aerodynamic technologies as currently implemented in a selection of production vehicles, and
the effectiveness of potential improvements that were yet to be implemented at the time.
The participating organizations and their respective programs share mutual interests in the
primary goals of the program, which are to quantify the aerodynamic drag impacts of various
OEM aerodynamic technologies, and to explore the improvement potential of these technologies
by expanding the capability and/or improving the design of current state-of-the-art aerodynamic
treatments. This program also has provided an important contribution to EPA's technical
assessment by offering an opportunity to further validate the feasibility and effectiveness
estimates for the passive and active aerodynamic technologies assumed for Aerol and Aero2.
As discussed in the Draft TAR, the program also provided an opportunity to further validate
off-cycle credits that were assigned to active aerodynamics in the 2012 FRM. Two active
aerodynamic technologies were identified for pre-defined credit availability of specified amount:
Active Grille Shutters and Active Ride Height. See 86.1869-12 (b)(l)(iv). The default value for
these credits offered were determined in large part by analysis, using an early version of the EPA
ALPHA model to simulate aerodynamic improvements for varying Cd inputs. A key assumption
in development of these credits was that active technologies only affect the coefficient of drag,
which is assumed to be constant over the speed range of the test. Further validation of this
assumption, and of the list of creditable active technologies assumed to be available in
production vehicles during the time frame of the rule, was seen as valuable in further supporting
the basis of the program. A total of four project phases consisting of twenty-five test vehicles in
all EPA vehicle classes was undertaken by the project partners.412
Active technologies evaluated by this program include: active grille shutters (opened, closed,
intermediate positions, speed effects, yaw effects, leakage effects); a detailed sealing study (i.e.
grille shutter sealing; external grille shutter concept); and an active ride height concept (i.e.
manual ride height adjustment on vehicles not necessarily equipped to do so from factory).
Passive technologies include: Air dams (front bumper and wheels); active front bumper air dams
(concept/prototype); underbody smoothing panels (both OEM and idealized prototypes); larger-
than-baseline wheel/tire packages; wheel covers (i.e. solid hubcaps); and miscellaneous
improvements (including front license plates, decorative grille features and smoothing, tailgates
(opened/closed/removed), and tonneau covers). Significantly, NRC facilities include a 9-meter x
2-139
-------
Technology Cost, Effectiveness, and Lead Time Assessment
9-meter rolling road/moving floor wind tunnel that allows testing of full scale vehicles for
accurate comparison of aerodynamic performance with and without active technologies. Listed
technologies were not evaluated on every vehicle due to stock configuration, timing and funding.
One valuable outcome of this testing was further validation of the default credit menu values
established in the 2012 FRM for active aerodynamic technologies under the off-cycle credit
program. Phase 1 of the Joint Program evaluated the aerodynamic performance of eleven (11)
vehicles (3 small cars, 5 midsize cars, 2 sport utility vehicles and 1 pickup truck). The
conclusions of the Phase 1 study indicated that the active aerodynamic technologies studied are
within the range of the default menu credit values anticipated in the TSD of the 2017-2025 GHG
rule TSD for active aerodynamic off-cycle credits.
The Phase I study also concluded that the benefit of active grille shutters is constant across the
operating speed range, confirming one key assumption in the FRM analysis. In addition, it
concluded that passive technologies may each improve the aerodynamics of future vehicles by 1
to 7 percent depending on the passive technology employed and overall vehicle design. This
conclusion was based on individual component installation, and does not account for synergistic
component effects, nor the effect of integrating passive technologies into an overall vehicle
redesign.
Depending on stock vehicle equipment, sometimes it was necessary to fabricate prototype
components to make an A to B comparison possible. Prototype components were constructed by
study partners Rochling Automotive and Magna International, both of which are Tier 1 suppliers
of various aerodynamic technologies to the industry.
Effectiveness values identified in Phase 1 of the Joint Program are shown in Table 2.13.
Table 2.13 Aerodynamic Technology Effectiveness from Phase 1 of Joint Aerodynamics Program
Aero Feature (A-B Testing)
Aero Drag Reduction (%)
Comments
Fixed Air Dam-Bumper
0s-
ID
i
1
OEM stock components
Active Air Dam - Bumper
4 - 9% (fixed air dam + 3%)
Fixed, prototype parts w/ lowest
(Conceptual)
deployment height used
Fixed Air Dam-Wheels
1% (front)/4.5% (front & rear)
Underbody Panels
1-7% (stock OEM)
Additional 0.5%-4% w/ full body panels.
LDT prototype: 8%
Increased Tire Size
-2.0 - 3.2%
17"/18" stock OEM rims vs. 22" optional
OEM rims
Wheel Covers
1.5 - 3%
Solid wheel covers only; brake cooling
affects not considered
Front License Plates
+/- 0.3%
Negligible impact
Decorative Grille Optimization
1.6%
Smoothing of grille features; function vs.
styling trade-offs
Pick-up Tailgates
Open
-5.2%
Removed
-7.5%
Open tailgate + 2.3%
Pick-up Tonneau Cover
3.7%
Phase II of the Joint Program427 investigated similar technologies using the same
methodology of Phase I. Vehicles studied in Phase II included nine vehicles including one small
car, one midsize car, one large car, one minivan, and five SUV/crossovers. Active technologies
2-140
-------
Technology Cost, Effectiveness, and Lead Time Assessment
studied included: active grille shutters (including yaw sweep) and active ride height (stock and
conceptual). Passive technologies included: underbody panels and air dams, and optional wheel
packages. Other technical assessments included turbulent flow impacts and yaw sweep impact.
To take into account the fact that vehicles are generally traveling in a windy environment from
potentially all wind azimuth angles, the wind averaged drag area was calculated for all cases
where a yaw sweep was carried out.
Phase III involved the testing of 4 vehicles: one sedan, one minivan, and two sport utility
vehicles.413 Phase IV involved the retesting of previous vehicles with a focus on turbulent flow,
including a small car and a pick-up truck. A report summarizing the results of all four phases is
in press at the time of this writing.
One significant outcome of the study was the identification of several high-impact areas for
drag reduction. For example, the study found that lowering the ride height while pitching the
vehicle nose down could provide significant drag reduction. Also, it was shown that certain
combinations of technologies (such as active grille shutters with air dams) often acted with
positive synergy (i.e. more than additive) to result in greater reductions in overall drag than the
individual technologies alone would suggest.
It should be noted that the Phase I and Phase II studies found that some technologies could
potentially increase drag area if poorly applied, and that some individual technologies did not
appear to be fully additive when combined with certain others. For example, presence of active
air dams was seen in some cases to reduce the effectiveness of adding underbody coverings.
Further, combination of active air dams or underbody coverings with active ride height tended to
reduce the effectiveness of active ride height. This latter result corroborates with information
related to EPA in an OEM meeting that suggested that vehicles that already have underbody
coverings are not as highly responsive to adjustments in ride height. On the other hand,
combining certain aerodynamic technologies (for example, active grille shutters with air dams)
often demonstrated higher total drag reduction than individual additive measurements would
have suggested.
Tests conducted during the study often found that lowering ride height while pitching the
vehicle at highway speeds (for example, 40mm in the front and 20mm in the rear) provided
measurable drag reduction for all vehicles. The highest reduction was observed for vehicle
classified as "Large Car". Additionally, underbody panels that are extended to cover the entire
surface area underneath the vehicle (full underbody cover) proved to be an efficient way to
reduce drag.
It was also found that yaw angle had a significant effect on measurement. Some technologies
that perform well at 0° wind angle were found to perform relatively poorly at different wind
angles (for example, at 8° to 10°, the differences were quite significant). It was also found that
some technologies that tend to work well for one class of vehicle may not perform well for
another vehicle class (for example, air dams in turbulent flow conditions were shown to perform
better on SUVs than on Large Cars.
In an effort to better represent real-world aerodynamic performance of aerodynamic
technologies, the study also investigated the effect of turbulent flow conditions on aerodynamic
measurements. The study produced an extensive data set comparing steady smooth and turbulent
flow performance for most of the vehicle classes. The study found that both turbulent flow and
2-141
-------
Technology Cost, Effectiveness, and Lead Time Assessment
yaw angle can be important to understanding the effectiveness of aerodynamic technologies in
real-world use.
2.2.6 Tires: State of Technology
2.2.6.1 Background
Tire rolling resistance is a road load force that arises primarily from the energy dissipated by
elastic deformation of the tires as they roll. Deformation, and hence rolling resistance, for a
given tire design is largely a function of vehicle weight and is fairly constant across the normal
range of vehicle speeds. Rolling resistance therefore carries an ever-present and often quite
significant effect on fuel economy and CO2 emissions.
Tire design characteristics (for example, materials, construction, and tread design) have a
strong influence on the amount and type of deformation and the energy it dissipates. Designers
can select these characteristics to minimize rolling resistance. However, these characteristics
may also influence other performance attributes such as durability, wet and dry traction,
handling, and ride comfort.
Although most tires do not carry markings that indicate their rolling resistance characteristics,
indications are that tires with reduced levels of rolling resistance are increasingly being specified
by OEMs in new vehicles, and are increasingly becoming available from aftermarket vendors.
Lower-rolling resistance tires commonly include attributes such as a higher recommended
inflation pressure, optimized materials, optimized tire construction (for lower hysteresis), special
geometry (for example, modified aspect ratio or narrower tread width), or stiffer sidewalls for
reduced deflection. OEM specification of these tires may be accompanied by changes to vehicle
suspension tuning or suspension design to counter any potential impact of the use of these tires
on other performance attributes of the vehicle.
2.2.6.2 Industry Developments
As discussed in the Draft TAR, since the 2012 FRM EPA has continued to follow industry
developments and trends in application of low rolling resistance technologies to light-duty
vehicles, by holding meetings with OEMs and suppliers, attending conferences and trade shows,
and regularly monitoring the press and technical literature.
Tires that achieve a 10 percent reduction in rolling resistance (compared to a MY2008-level
baseline) are available today, and since the FRM, appear to have continued to comprise an
increasing share of tire manufacturers' product lines as the technology has continued to improve
and mature. Improvements that would reach up to a 20 percent decrease in rolling resistance
relative a 2008 baseline have also seen significant progress in the industry, with indications of
increased availability and improved traction and performance characteristics.
Since the 2012 FRM and even before, the tire industry has become increasingly focused on
improving tire performance. Recent industry momentum in this direction was captured well in a
quote by Kurt Berger of Bridgestone, in a 2014 article in Automotive News.428 "A low-rolling-
resistance tire of 2010 would not be considered a low-rolling-resistance tire today. We've really
been pushed in a short time to reduce rolling resistance further." Several typical examples of
industry research and implementation efforts are outlined in a 2015 report by Auto World429.
One example of a specific product embodying lower rolling resistance technology is the Falken
2-142
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Sincera SN832 Ecorun Tire, with a 22 percent improvement over its immediately previous
generation, while maintaining a 27 percent improvement in braking distance. According to a
Continental spokesperson cited in the Auto World report, "... improvements of more than 20
percent from one generation to the next [are possible] by introducing rolling resistance optimized
tires ... an additional 5 percent improvement generation-to-generation is possible." According to
Indraneel Bardhan, Managing Partner of EOS Intelligence, so-called "green tires" have achieved
a global market share of about 30 percent.
The Automotive News article cited above also discussed ongoing challenges for low rolling
resistance tires, including issues such as wet traction, tread wear, and the magnitude of real world
benefits in comparison to customer expectations. Customers were said to be relatively
indifferent about the fuel economy benefits of low rolling resistance tires, but the perception of
differences in handling performance between these tires and traditional tires appeared to be
stronger. Due to these perceptions, it was suggested that although original equipment fitments of
low rolling resistance tires have been increasing, consumers may tend to replace them with more
conventional tires after the original tires wear out, potentially reducing the net fuel-saving impact
that would otherwise be expected over the full useful life of the vehicle.
Preliminary results of a study currently underway by Transport Canada (TC) and Natural
Resources Canada (NRCan) provides additional support for the view that traction and lower
rolling resistance are not necessarily mutually exclusive. In this study, TC and NRCan are
coordinating with EPA as they conduct a multi-year testing and evaluation campaign to
investigate the rolling resistance and traction characteristics of commercially available tires.QQ
One aim of the program is to study any correlation that may exist between rolling resistance
performance and safety performance (traction) for winter and all-season tires. To date, the
campaign has tested 24 winter tires, 50 all season tires, and 5 all-weather tires, testing for energy
efficiency, traction performance, and viscoelastic properties of the tread (indicators of rolling
resistance and traction performance).
As shown in Figure 2.38, preliminary results of the Transport Canada/Natural Resources
Canada study show that winter tires are available with a wide variety of rolling resistance and
wet grip characteristics, including tires with both low rolling resistance and good wet grip. For
instance, one tire had a rolling resistance coefficient less than 9.0, and a wet grip index greater
than 1.1.
QQ The primary purpose of this study is to support development of a Canadian consumer information program for
replacement tires.
2-143
-------
Technology Cost, Effectiveness, and Lead Time Assessment
1.4
1.5 - i i
¦ ¦
0,9 1
0.8 - - -
7.00 8.00 9i]l} 10.00 11,00 12.00
Ruling Resistance Caeffkieit
Figure 2.38 Relationship between Wet Grip Index and Rolling Resistance for Winter Tires from Transport
Canada/NRCan Study
Countering the common perception that reducing rolling resistance must sacrifice traction
performance, the scatter of points in the plot suggests that the range of design variables currently
available to tire designers is sufficient to achieve a wide variety of combinations of traction
performance and rolling resistance performance, including combinations with low rolling
resistance and good traction. Further optimization with respect to cost (which is not represented
in the plot) is largely sensitive to manufacturing optimization and production volume, and will
play out as demand and production levels for low rolling resistance tires continues to grow.
One example of the potential for careful design to maintain traction in a low rolling resistance
tire is seen in the Bridgestone "ologic" design, which appears on the BMW i3 electric vehicle.
This tire has a relatively large diameter coupled with a narrow width, reducing rolling resistance
by maintaining low deformation through a stiffer belt tension. The larger diameter and unique
construction increases the length of the contact patch, which serves to provide improved braking
performance and wet and dry traction. An advanced rubber compound and special tread design
also contributes.430 The relatively narrow design is also said to improve aerodynamic
performance.429 The trend toward larger diameter tires with narrower cross-sectional width is
also associated with lower tire noise levels, and have been described as one of the likely tire
design trends that will continue into the future, particularly for BEVs that value both energy
efficiency and quiet performance429. As another example, the tire manufacturer Pirelli has
ongoing projects focusing on development of new tire polymers through joint ventures with
chemical suppliers429.
Research data presented at the 2014 U.S. DOE Merit Review strongly suggests that
significant rolling resistance improvements are accessible to much of the tire market. A project
involving Cooper Tires, funded by the U.S. Department of Energy, targets a 30 percent reduction
in rolling resistance and a 20 percent reduction in tire weight, while maintaining traction
performance.431 By investigating new materials and methods for reducing rolling resistance in
ways that maintain wet traction and tread wear capabilities, this project has suggested that
potential improvements in rolling resistance of 10 to 20 percent are achievable by selection of
2-144
-------
Technology Cost, Effectiveness, and Lead Time Assessment
appropriate materials and construction, with examples of reduction in rolling resistance from a
prevailing 0.08 to 0.10 down to 0.064 to 0.08.
2.2.7 Mass Reduction: State of Technology
2.2.7.1 Overview of Mass Reduction Technologies
Mass reduction is a key technology for reducing vehicle energy consumption. Vehicle mass
has a direct effect on the energy consumed by tire rolling resistance, as well as on the energy
needed to accelerate a vehicle, much of which is later lost to friction braking. Through its
relationship to acceleration, mass also has implications for the necessary power rating of the
propulsion system, with an increased engine size potentially leading to reduced average
powertrain efficiency.
Several techniques are available for reduction of vehicle mass, including adoption of lighter-
weight materials and part consolidation, among others. Computer-aided engineering (CAE)
provides an efficient tool for optimization of vehicle designs along these lines, by allowing rapid
modeling and evaluation of potential material substitutions and part modifications.
The cost of reducing vehicle mass is highly variable. Design optimization, consolidation of
components, and adoption of secondary mass savings opportunities can result in some cost
savings. Secondary mass reduction refers to weight reduction opportunities that become
available as the base vehicle becomes lighter. A smaller engine block, transmission and brakes
are examples of secondary mass reduction opportunities. Cost increases are often the result of
changing from a high density, lower cost material, such as steel, to a lower density, higher cost
material, such as high-strength steel, aluminum, magnesium, or composites. The cost for a given
mass reduction solution depends on the approach and the material being used. In some cases,
cost savings can offset cost increases. Benefits from adopting mass reduction technologies can
also include improved performance, such as acceleration, vehicle dynamics, and overall
responsiveness.
For the Draft TAR, EPA reevaluated many aspects of mass reduction, including the
techniques described above, the cost of mass reduction, the FRM conclusions, and the amount of
mass reduction present in the baseline fleet. EPA completed work including research,
stakeholder meetings, supplier meetings, technical conferences and literature searches. Public
information from these sources were fully described in the Draft TAR, and ultimately formed the
basis for the mass reduction cost curves that were developed for the purpose of technology
package modeling for that analysis.
EPA has continued monitoring the state of the art of mass reduction, and where applicable,
has included updated information on this topic in the present discussion, which builds on the
discussion presented in the Draft TAR.
The discussion in this chapter forms the basis for the specific data and assumptions that were
used for modeling mass reduction for this assessment, which are described in Section 2.3. This
includes the 2015 baseline fleet mass reduction estimates, including mass allowances for safety
and footprint changes between the 2008 and 2015 vehicles; a review of the development of the
mass reduction cost curves and their application, and mass reduction effectiveness. Further
discussion of specific materials (steel, aluminum, magnesium, plastic, glass, and glass fiber and
carbon fiber composites), as well as details of their application in regards to issues such as
2-145
-------
Technology Cost, Effectiveness, and Lead Time Assessment
feasibility, cost, safety, and current areas of research, were included in the Appendix to the Draft
TAR.
The relationship between mass reduction and safety is an important consideration when
considering opportunities for applying this technology. As described in the Draft TAR, NHTSA
performed an updated analysis of this issue that was described in Chapter 8 of the Draft TAR.
In recent years, manufacturers have been adopting mass reduction in varying degrees. From
vehicles that have adopted large amounts of lower-density materials in their body-in-white
(BIW), as with the MY2015 Ford F150 and MY2014 BMWi3, to vehicles that have adopted
smaller changes in vehicle design such as an aluminum hood or a steel clamshell control arm in
the suspension such as the MY2014 Silverado 1500. The EPA 2015 Trends report illustrates, in
Figure 2.39, how in overall sales weighted basis, vehicles have not yet achieved a notable
decrease in curb weight, or have continued the trend of using mass reduction to offset increased
vehicle content or larger footprint, as the mass difference has remained constant over the past 10
years. The detail within the report notes 2014 results show a 0.5 percent mass increase for cars
and 0.7 percent mass decrease for trucks, each on a sales weighted basis.
Change in Adjusted Fuet Economy, Weight, and Horsepower for MY 1975-2015
cr>
8
C
y>
i
o
c
L?
0>
a
Figure 2.39 Change in Adjusted Fuel Economy, Weight and Horsepower for MY1975-2015432
One reason for the current trend of curb weight changes may be the desire to make significant
mass-reducing design changes during major vehicle redesigns, hence limiting large mass
reductions to new vehicle designs. Recent announcements, as listed in Table 2.14, indicate that
the adoption of mass reduction technologies, and resultant lower curb weights, will continue into
the future as vehicles are redesigned and as some mass reduction solutions become less costly.
One example of significant mass reduction is the 2017 GMC Acadia. GM has stated that the
mass of the Acadia has been reduced by 700 pounds through adoption of high-strength steels, a
smaller engine option and a smaller footprint.433 The announcement of the 2017 Chrysler
Pacifica in January 2016 also noted 250 pounds of mass reduction through "extensive use of
advanced, hot-stamped/high-strength steels, application of structural adhesives where necessary,
and an intense focus on mass optimization." Magnesium is also used in the instrument panel and
the inner structure of the Pacifica's lift gate, the rest of which is aluminum.434
Adjusted Fuel ECOftOny
ItdtofipOwer
We^hl
-40%
Model Year
2-146
-------
Technology Cost, Effectiveness, and Lead Time Assessment
To illustrate the general trend in the use of lightweight materials, Figure 2.40 shows a
comparison of metallic material adoption from 2012-2025 included in the 2014 Executive
Summary for a study by Ducker Worldwide.435 The study notes that there was a slight increase
in the use of light-weight materials for BIW and closures between 2012 and 2015. The use of
AHSS/UHSS grew from 15 percent to 20 percent of the vehicle body and closure parts.
Aluminum sheet also grew from 1 percent to 4 percent and aluminum extrusions made it onto the
pie chart in 2015. Overall, the analysis expects that steel will remain the dominant material in
BIW and closures. According to his, use of plastics is expected to grow to 350kg per average car
in 2020, up from 200 kg in 2014, as shown in Figure 2.41. Use of carbon fiber for auto
manufacturing is expected to increase from 3,400 metric tons in 2013 to 9,800 metric tons in
2030. According to Ducker Worldwide, the use of magnesium is expected to increase through
2025, as magnesium castings are expected to grow significantly over the next 10 years, further
stating, "Growth is highlighted within 'large tonnage' parts like closure inners, IP structures etc.
and other body/structural parts."
Executive Summary
The material mix for body and closure parts will change dramatically over the next ten years.
On a weight basis, aluminum will grow to 19% of the weight for body and closure parts by 2025.
2012
2020
¦ Mild and HSLA
¦ AHSS/UHSS
¦ Aluminum Sheet
¦ Aluminum Extrusions
¦ Aluminum VD Castings
1023 lbs.
¦ Mild and HSLA
¦ AHSS/UHSS
¦ Aluminum Sheet
¦ Aluminum Extrusions
¦ Aluminum VD Castings
941 lbs.
2015
2025
¦ Mild and HSLA
¦ AHSS/UHSS
¦ Aluminum Sheet
¦ Aluminum Extrusions
¦ Aluminum VD Castings
993 lbs.
¦ Mild and HSLA
¦ AHSS/UHSS
¦ Aluminum Sheet
¦ Aluminum Extrusions
¦ Aluminum VD Castings
: lbs.
o
ICKF.R WORIDWIDF.
Figure 2.40 Estimated Vehicle Material Change over Time 2012-2025 - Ducker Worldwide435
2-147
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Automotive Market Summary
2005
2010
2015
2020
2024
Forecasted
%CAGR
Vehicle Unit
Deliveries
66.5M
77.9M
93.0M
100.7M
109.OM
1.80%
Total Vehicle
Wt (MT)
30.2M
35.4M
37.2M
41.1M
44.5M
1.80%
Total FGRP
Structures (MT)
79,212
102,371
! 131,310
167,588
203,704
4.49%
Total CFRP
Structures (MT)
3,921
3,771
13,060
37,085
47,011
16.67%
Total CF
Demand (MT)
1 1
3,666
3,526
10,056
23,456
47,011
13.73%
Figure 2.41 Forecast of Automotive Market Consumption of Composites436
30
25
20
15
10
5
0
25
7 8 8 8 8
I I I I I I
2010 2011 2012 2013 2014 2015 (e) 2020 (f)
Low H Pounds per Vehicle Hi
2025 (f)
Figure 2.42 Magnesium Growth Expectations through 2025 (Ducker Worldwide)437
EPA expects that innovative mass reduction solutions will continue to be developed and
adopted through MY2025 and that some mass reduction solutions will be less costly than they
are today. Expected advancements include the development of lower-cost high strength steel
alloys for body structures (3rd generation steels), lower-cost and higher quality product (for
Class A surfaces) from the aluminum Micromill sheet manufacturing processes, and
2-148
-------
Technology Cost, Effectiveness, and Lead Time Assessment
advancements in engineered plastics and composites for structural applications. Developments
are also anticipated in design, including further development and use of CAE design tools to
characterize new material properties and behaviors. This is expected to result in advances in
material use, including optimized load pathway analyses in BIW geometries, and consolidation
of multi-part components, resulting in the achievement of mass reduction in the most cost
effective way.
2.2.7.2 Mass Reduction Feasibility
Since the FRM, EPA has continuously gathered information on technological advancements
and application of mass reduction technologies through a variety of sources, including technical
conferences, public reports, material association meetings, academic research, news articles, and
stakeholder meetings with manufacturers and suppliers (often including discussion of
confidential business information). As previously mentioned, an overview of publicly available
information on lightweight materials was included in the Appendix of the Draft TAR. EPA and
NHTSA generated two independent holistic lightweighting studies for mass reduction and cost
data on light duty pickup trucks (MY2011 and MY2014) and updated existing passenger car
(EPA Midsize CUV and NHTSA Passenger car) holistic lightweighting studies completed in
2012. The light duty truck holistic reports join the projects currently described in the FRM on a
midsize CUV, one conducted by EPA and one by ARB, and a passenger car, conducted by
NHTSA. The Aluminum Association also conducted several projects including a project with
ED AG, Inc. to evaluate the EPA Midsize CUV high strength steel BIW CAE model with
aluminum material replacement.
DOE also collaborated with Ford and Magna to develop a multi-material lightweight vehicle.
This program included a vehicle prototype build and initial durability tests. In addition to
vehicle lightweighting, research projects were performed on the mass increases due to safety
requirements, for example the IIHS small overlap test (2012). NHTSA conducted a CAE
passenger car evaluation and Transport Canada conducted a CAE light duty truck study
evaluation which included a crash test of the baseline vehicle. With respect to mass reduction
efficiency, the Aluminum Association conducted a study on the impact of mass reduction on fuel
economy for various vehicles with Ricardo, Inc. on which the 2015 NAS report comments were
based. EPA and NHTSA (through ANL) also re-evaluated the effectiveness of mass reduction
on CO2 and fuel consumption reductions for several vehicle classes, including standard car and
light duty truck. The studies on efficiency are addressed in Section 2.3.
In comments on the Draft TAR, Global Automakers and the Alliance of Automobile
Manufacturers (AAM) commented that the Draft TAR did not thoroughly discuss some of the
real-world constraints on mass reduction. They also commented that the agencies should take
into account the time needed to test and qualify new materials before they may be incorporated
into vehicles, and the prevalence of global platforms using the same parts in several different
vehicle models made in multiple locations. In addition, they recommend that EPA should also
consider the need of manufacturers to satisfy customer needs and expectations and regulatory
requirements. In addition, Global Automakers stated that many mass reduction technologies have
unintended consequences that customers will not accept. For example, they contend that light-
weighting technologies can increase noise, vibration and harshness (NVH) to levels unacceptable
to consumers.
2-149
-------
Technology Cost, Effectiveness, and Lead Time Assessment
EPA recognizes that there are many factors which contribute to the implementation of mass
reduction technologies by vehicle manufacturers. These potential barriers to mass reduction are
different from manufacturer to manufacturer and are related to level of experience with the
technology, supplier experience, vehicle functional objectives, and global and platform
manufacturing constraints. In each of the mass reduction studies used to inform the Draft TAR
and the Proposed Determination analyses, many alternatives are presented for reducing mass.
Because the studies were done holistically, mass reduction solutions were identified across the
entire vehicle and the studies considered technologies from a wide variety of sources. In
addition, the results of the Proposed Determination analysis do not project a large amount of
mass reduction, on average, across the light-duty fleet. As such, manufacturers will most likely
have many choices as to which mass reduction solutions to choose which meet the requirements
for OEM and supplier experience, global manufacturing and vehicle functional objectives.
AAM also commented on some of the challenges associated with mass reduction, specifically
on material availability. The September 2016 study by the Center for Automotive Research
(CAR)438 contains (page 11) a list of challenges for various lightweight materials (HSS,
Aluminum, Magnesium, Composites). In addition, in order to achieve the higher levels of
percent mass reduction, the study maintains that magnesium and composites would be required.
EPA agrees that magnesium, aluminum and composites are important materials for mass
reduction and are already being applied on many current production vehicles to reduce mass.
Regarding composites, one of the primary concerns has been CAE simulation for various
material compositions, availability of low cost carbon fiber to use in the composite material, and
a recyclable resin material. Modeling composite behavior can and has been done: for example,
BMW has produced the BMWi3 which has a composite/aluminum BIW. BMW is also
supporting work at the University of Delaware to develop a B-pillar made of composite material
and the results have been presented at the SAE Government/Industry meeting in 2015 and
presentation of component build and test in 2016.
With respect to aluminum, the September 2016 CAR report438 states there are concerns with
conversion of the steel-based supply-chain infrastructure, paint shop issues (thermal expansion,
aluminum surface characteristics), robustness of the supply base, and the need for redesign of
body shop assembly technology. The aluminum industry is poised to supply aluminum needs and
the Micromill technology can be used to supply some of that demand as is being done on the
F150.439 EPA agrees that aluminum stamping is different from steel stamping, but as
demonstrated by the F150 program, it is feasible in a high volume production environment. In
regards to thermal expansion, OEMs are able to manage the thermal properties of various
materials, including aluminum, as demonstrated by the many current production vehicles that
have aluminum hoods and other closures. Further, GM has developed a way to join aluminum
vehicle components as an alternative to the vehicle manufacturing techniques used by Ford on
the F150.440
The following section provides a description of the multi-material approach to lightweighting
being used by OEMs, and presents some examples of current vehicle designs that have adopted
notable mass reduction which resulted in significant curb weight reductions. Further sections
present an overview of the various holistic mass reduction and cost studies that were completed
since the FRM. The studies provide technology, primary and secondary mass reduction, and cost
information that was used to create cost curves for application of mass reduction technology for
2-150
-------
Technology Cost, Effectiveness, and Lead Time Assessment
a passenger car and light duty pickup truck, which were used in the Draft TAR analysis and
remain largely unchanged for the Proposed Determination analysis.
2.2.7.3 Market Implementation of Mass Reduction
A trend of slight reductions in curb weight in the new vehicle fleet has been observed in both
the MY2014 baseline used in the Draft TAR analysis and the MY2015 baseline used in this
Proposed Determination analysis. Data reported in the 2016 EPA Trends report indicates that
the overall sales-weighted curb weight has remained steady over the past 10 years. In MY2008,
the sales weighted vehicle weight was 4,085 pounds with a footprint of 48.9 square feet, but by
MY2015 it was 4,035 pounds and 49.4 square feet, a decrease of 50 pounds and an increase of
0.5 square feet.441 During this period, additional equipment to meet safety regulations led to
addition of mass, which would be included in the MY2015 weights.
Table 2.14 lists a number of vehicle lightweighting efforts that have been introduced into the
market over the past few years. Some vehicles adopted high strength steel solutions, up to 2 GPa
tensile strength steels, in their BIW such as in the Audi Q7, Acura TLX, Nissan Murano and
Cadillac CTS redesigns. The MY2015 F150 and the MY2014 Range Rover by Land Rover have
both adopted a number of lightweighting components including aluminum body and cabin
structure, aluminum closures, etc.
Table 2.14 Examples of Mass Reduction in Selected Recent Redesigns (Compared to MY2008 Design)1®
Vehicle Make
2008 Model Year
curb Weight (kg)
Model Year
Change in Vehicle
Curb Weight (kg)
% Change
% Footprint
Change
Acura MDX
2070
2014
238
11.5%
+0.5%
Audi Q7
2320
2014
325
14%
0
Land Rover Range Rover
2400
2014
336
14%
+5.2%
Silverado 1500 Crew Cab
2422
2014
86
3.6%
n/a
4x4
Ford F150
2446
2015
318
13%
n/a
2.7L EcoBoost, 4x2
Supercrew
Nissan Murano
1500
2015
30
2%
n/a
Cadillac CTS
1833
2015
110
6%
+1.6%
Honda Pilot
4367
2016
131
3%
+6.1%
Chevy Cruze442
1425
2016
114
8%
n/a
Chevy Malibu443
1552
2016
136
9.2%
+0.3%
GMC Acadia
2120
2017
318
15%
-7.8%
Chrysler Pacifica
2110
2017
114
5.4%
+8.2%
Cadillac XT5444
1893
2017
82
4.5%
+2.7%
The following excerpt from an Audi445 press release represents the holistic engineering
approach that achieved significant levels of cost effective mass reduction:
1111 Some vehicles were redesigned twice since 2008 and so the changes are not exactly the same as noted in the
articles from which some of the information was taken, because the table references differences between 2008
and 2014.
2-151
-------
Technology Cost, Effectiveness, and Lead Time Assessment
"Although it [the Audi Q7] is shorter and narrower than its predecessor, the cabin
is longer and offers more head room. 20 years of experience with lightweight
construction flow into the new Audi Q7. Equipped with the 3.0 TDI engine, the new Audi
Q7 tips the scales at just 1,995 kilograms (4,398 lb.), which is 325 kilograms (716.5 lb.)
less weight ... the Q7 with the 3.0 TFSI engine is even lighter, weighing just 1,970
kilograms (4,343.1 lb.). Lightweight construction has been applied in all areas, from the
electrical system to the luggage compartment floor. The key is the body structure, where
a new multi-material design reduces its weight by 71 kilograms (156.5 lb) ... ultra-high-
strength parts made of hot-shaped steel form the backbone of the occupant cell.
Aluminum castings, extruded sections and panels are used in the front and rear ends as
well as the superstructure. They account for 41 percent of the body structure. Other parts
made entirely of aluminum are the doors, which shave 24 kilograms (52.9 lb.) of weight,
the front fenders, the engine hood and the rear hatch. Audi uses new manufacturing
methods for the production and assembly of the parts. The crash safety and occupant
protection of the new Audi Q7 are also on the highest level."
The holistic design approach enables secondary mass savings that can be achieved due to
reduced load requirements as the overall vehicle becomes lighter. One example of secondary
mass reduction is the potential adoption of a smaller engine in a light weighted vehicle. Ford
mentioned in a 2010 International Magnesium Association article that:
"Strategic use of lightweight and down-gauged material allows a vehicle's
powertrain to be smaller and more fuel-efficient. Combining magnesium with aluminum
for the MKT lift gate's panels instead of steel saves 22 pounds in vehicle weight. When
coupled with other weight-saving measures, re-matching the vehicle with a smaller
powertrain - known as right-sizing of power to weight — is a key factor in achieving
greater fuel economy."446
2.2.7.4 Holistic Vehicle Mass Reduction and Cost Studies
As shown in the Draft TAR, the 2017-2025 FRM Joint Technical Support Document (2012
TSD) contained a linear mass reduction cost curve for direct manufacturing costs (DMC) in the
expression of DMC ($/lb.)=$4.36(percent-lb.) x Percentage of Mass Reduction level (percent) as
shown in Figure 2.43. This equation starts at $0/kg for no mass reduction and increased at a
constant rate of $4.36/( percent-lb.) for each percent mass reduction (ex: $0.44/lb. for 10 percent
MR on a 4,000 lb. vehicle and $0.66/lb. for 15 percent on same) and was applied to all
2008/2010 MY vehicles in which no mass reduction was assumed. This cost curve expression
was based on a number of available data sources on mass reduction which included a number of
papers on individual components.
2-152
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Mass Reduction Cost
$1.00
„ $0.80
£
— $0.60
to
<-> $0.40
'c
3 $0.20
$0.00
0% 5% 10% 15% 20% 25%
Percent of Mass Reduction
Figure 2.43 Mass Reduction Cost Curve ($/lb.) for 2017-2025 LD GHG Joint Technical Support Document
In order to capture a more complete picture of the potential for mass reduction and related
costs, EPA, NHTSA, ARB, and DOE committed significant resources to acquire mass and cost
information through a number of holistic vehicle studies as listed in Table 2.15. The projects
were performed with constant performance as a goal, and hence the benefits of all mass
reduction solutions were applied to improve fuel efficiency and lower CO2 emissions. Each
project includes many steps including baseline vehicle teardown, component/system examination
for mass reduction technologies, direct manufacturer cost estimation for mass reduction
technologies and related tooling, CAE safety crash evaluation, NVH assessment and durability
analyses. The mass reduction technologies included in these studies were found in a variety of
sources including those found on other vehicles, technologies in development at suppliers and
material companies, technologies developed in other government funded projects, etc. Cost
estimates were made by the project contractors based on their extensive automotive experience
and industry contacts.
The DOE/Ford/Magna joint project itself did not include a cost study for its two evaluations -
Mach 1 (25 percent MR) and Mach 2 (50 percent MR). However, DOE did fund two
independent cost studies related to this work. One study was for a 40 to 45 percent mass
reduction vehicle which identified the necessary cost of carbon fiber in order to make the design
solution a reality. These results were presented at the DOE Annual Merit Review (AMR) in
2015. A second independent study was also funded by DOE in 2016 and presented at the 2016
DOE AMR. This study focused on an assessment of the multiple strategies addressed in the
earlier phase in terms of weight reduction, cost premiums, and risk factors in order to establish a
prioritized spectrum of lightweighting opportunities. The work then applied process Technical
Cost Models (TCMs) to priority lightweight material manufacturing technologies to evaluate
cost structures and understand the relative leverage of key cost drivers.
The Mach 1 work also included several additions which included the buildup of seven
lightweight vehicles for a number of durability and crash analyses as well as testing of some of
the project's new technologies. Two other studies provided insights into the mass add for
meeting the IIHS small overlap test which is required in order to achieve the IIHS rating of Top
Safety Pick. NHTSA funded a follow-up study on their 2012 passenger car work and Transport
Canada funded a follow-up study on the EPA 2015 light duty pickup truck. The studies provided
s
ope =
= 4.36
2-153
-------
Technology Cost, Effectiveness, and Lead Time Assessment
a revised final cost and mass reduction to the original works. EPA also greatly appreciate and
acknowledge the work of many individual companies, academia representatives, and material
associations to provide information on lightweighting technologies, both in production and in
research, to the agency contractors for the holistic vehicle studies. This information was also
used as the basis for material information contained in the Appendices to the Draft TAR to
address topics of feasibility, mass reduction, cost, safety, research and recycling.
2-154
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.15 Agency-Sponsored Mass Reduction Project List since 2012 FRM
Agency
Description
Completion
Date
Reference
Pass
US EPA
Phase 2 Midsize
2012
Final Report, Peer Review and SAE Paper
Car/
CUV
E PA-420- R-12-019, E PA-420- R-12-026,
CUV
(2010 Toyota
SAE Paper 2013-01-0656
Studies
Venza)
Low Development
(HSS/AI focus)
ARB
Phase 2 Midsize
CUV
(2010 Toyota
Venza)
2012
Final Report and Peer Review
http://www.arb.ca.gov/msDrog/levDrog/leviii/final arb d
hase2 reDort-comoressed.Ddf
http://www.arb.ca.gov/msprog/levprog/leviii/carb versio
High Development
n lotus proiect peer review.pdf
All Aluminum
NHTSA
Passenger Car
(2011 Honda
Accord)
2012
Final Report, Peer Review, OEM response, Revised Report
ftp://ftp.nhtsa.dot.gov/CAFE/2017-25 Final/811666.pdf
http://www.nhtsa.gov/Laws+&+Regulations/CAFE+-
+Fuel+Economv/ci.NHTSA+Vehicle+Mass-Size-
Safetv+Workshop. print
http://www.nhtsa.gov/staticfiles/rulemaking/pdf/cafe/81
2237 LightWeightVehicleReport.pdf
DOE/
-Passenger Car
2015
http://energv.gov/sites/prod/files/2015/06/f24/lm072 sk
Ford/
(2013 Ford Fusion)
szek 2015 o.pdf
Magna
Mach 1 and Mach 2
projects
-Cost Study for 40-
45% Mass
Reduction
-Mass Reduction
Spectrum Analysis
And Process Cost
Modeling Project
http://energv.gov/sites/prod/files/2014/07/fl7/lm072 sk
szek 2014 o.pdf
http://energv.gov/sites/prod/files/2014/07/fl7/lm088 sk
szek 2014 o.pdf
http://avt.inl.gov/pdf/TechnicalCostModel40and45Percen
tWeightSavings.pdf
http://energv.gov/sites/prod/files/2016/06/f33/lm090 m
ascarin 2016 o web.pdf
SAE papers include:2015-01-0405~0409,2015-01-
1236~1240,2015-01-1613~1616
NHTSA
Passenger Car small
overlap mass add
2016
Final Report
http://www.nhtsa.gov/staticfiles/rulemaking/pdf/cafe/81
2237 LightWeightVehicleReport.pdf
Light
EPA
2011 Silverado 1500
2015
Final Report, Peer Review and SAE Paper
Duty
EPA-420-R-15-006,SAE Paper 2015-01-0559
Truck
NHTSA
2014 Silverado 1500
2016
Final Report November 2016
Studies
Transpo
IIHS small overlap
2015
Final Report and Peer Review
rt
mass add on LDT
https://www.tc.gc.ca/eng/programs/environment-etv-
Canada
(EPA)
summarv-eng-2982.html
Peer Review (EPA docket)447
The holistic vehicle studies in Table 2.15 are nearly all focused on MY2008/2010 designs.
This was important for two reasons. The first is that the 2012 FRM analysis was based on the
ability to reduce the mass of the MY2008 fleet. Second, these mass reduction studies provided
insight into many mass reduction solutions that had not yet been widely adopted by
manufacturers. The MY2014 new-generation light duty pickup truck evaluated by NHTSA was
2-155
-------
Technology Cost, Effectiveness, and Lead Time Assessment
a 'next step' approach to evaluate the mass reduction potential and cost of converting from a
more high strength steel approach (compared to the 2008 design) to other lightweight materials
including aluminum and CFRP. It should be noted that the cost curve expression used by EPA
in the Draft TAR and this Proposed Determination differs from that used by NHTSA in the Draft
TAR CAFE assessment.
EPA is using the information from the publicly available government sponsored studies in its
modeling of mass reduction and related costs for all the vehicles sold in the US. The vehicles for
the holistic vehicle projects were chosen based on their representation of high sales volume
vehicles, as the Honda Accord and Chevy Silverado 1500, and/or representative of new vehicle
designs that were showing increasing popularity, as the Toyota Venza. The projects were
conducted over the past 6 years and were multi-million dollar efforts. The same detailed
information collected in these projects were not readily available from any other source -
especially cost information and secondary mass effects. Additional mass comparison
information was found to be available through the A2Macl vehicle databases and that
information has been used to supplement our analyses on mass differences - especially on mass
add for vehicle footprint increases. Ducker Worldwide executive summaries have also provided
insights into aluminum and steel material trends.
To understand how the results from our projects relate to real world lightweighting efforts,
EPA has met with OEMs and attended many technical conferences over the past four years. It
was observed that there are cost savings to be achieved from lightweighting MY2008/2010
design vehicles and more is expected as costs are reduced through material recycling and
optimization of material use. EPA agrees that some mass reduction technologies will add cost,
however recent developments in material processing, as with development of 3rd generation
steelsss and Alcoa's Micromill for aluminum, indicate that these costs may be less than that
utilized in the studies. In addition, the decrease in metal material pricing over the past year has
not been included in most of the holistic vehicle studies. EPA understands that OEMs have
typically utilized mass reduction technologies to offset the weight of added features or safety
measures.
In their comments on the Draft TAR, AAM commented that the mass reduction studies used
to develop the cost curves were "overly optimistic" due to the vintage of the vehicles studied
(Venza and Silverado) and scope of the studies (Venza). (AAM also noted that they did not have
cost curves to present as an alternative.) EPA disagrees that the cost curves used in the Draft
TAR analysis are inappropriate and has continued to apply these cost estimates in the Proposed
Determination. Within any given model year, the fleet will be comprised of vehicles with a
variety of design vintages, and designs with varying degrees of mass reduction implementation.
In recognition of these variations, EPA adopted an approach in the Draft TAR to determine the
initial starting point on the cost curve that is appropriate for each individual model in the
baseline. When applying the cost curves based on studies with earlier vintage vehicles (i.e. the
2009 Venza, 2011 Accord, and 2011 Silverado) and the more recent 2014 Silverado, EPA
aligned the curves so that they would maintain a consistent "null" technology reference point at 0
percent mass reduction. EPA believes that the critical point, consistent with the comment from
ss Nanosteel mentioned in their comments to the Draft TAR that our costs were overestimated for 3rd generation
steel.
2-156
-------
Technology Cost, Effectiveness, and Lead Time Assessment
AAM, is that vehicles in the baseline are placed at the appropriate location on the cost curve. As
mass reduction technologies are continuously introduced into the fleet from year-to-year,
analyses based on progressively updated baseline years would involve placing vehicles further
along the cost curve. Using this approach, differences in the vintages of the vehicles used to
create the cost curves will not have a primary influence on the incremental costs applied for mass
reduction.
2.2.7.4.1 EPA Holistic Vehicle Mass Reduction/Cost Studies
EPA funded two holistic vehicle mass reduction/cost studies for the Midterm Evaluation
between 2010 and 2015. The first study was the Phase 2 low development (steel BIW)
lightweighting study on a Midsize CUV performed by EPA with FEV North America, Inc.,
ED AG, Inc. and Munro and Associates, Inc. and was focused on achieving 20 percent mass
reduction which resulted in a high strength steel structure with aluminum closures amongst other
technologies. This was a follow up to the Phase 1 paper study on the Midsize CUV performed
by Lotus Engineering and includes in-depth analyses on cost and CAE safety analyses of the
vehicle. The second study was a lightweighting study on a 2011MY light duty pickup truck and
was performed by the same contractors using a similar methodology however added in the
dynamic vehicle analyses and a number of component evaluations performed with CAE. The
result was an aluminum intensive vehicle with high strength steel/aluminum ladder frame.
EPA's cost curve development methodology for both projects is based on a cumulative
additive approach of the best-rated technologies in terms of $/kg. Primary mass reduction
technologies (technologies not dependent on mass savings in other areas of the vehicle) are listed
along with the related costs and mass savings. The $/kg for each technology is calculated and
then the order of the technologies is sorted from lowest $/kg to highest. The original mass and
costs are then each added in a cumulative manner and then the resultant $/kg is calculated at each
technology and a related percent mass reduction. Secondary mass savings, those mass savings
which are dependent on other mass savings within the vehicle, are noted on a component
evaluation basis, summed, and then applied at the solution point for the project. Since the
secondary mass savings are based on the size of the component - hence material basis - then this
can be proportioned across the whole range of primary mass reduction curve. The cost savings
are also proportioned. Two assumptions work into this costs curve methodology: 1) OEMs will
adopt the lowest cost mass reduction technologies first; and 2) secondary mass savings, such as a
resized engine and/or chassis systems, can occur at all percent mass reduction points. This
methodology works into EPA's mass reduction modeling methodology for the Proposed
Determination.
Other related studies to the Phase 2 Low Development Midsize CUV include the Phase 2
High Development study funded by ARB. ARB hired Lotus Engineering to compete an in-depth
look into the aluminum intensive (High Development) Midsize CUV and included CAE safety
analyses and an in-depth cost analyses. Both of the Phase 2 studies, High Development and Low
Development, are follow-up studies to the Phase 1 paper study by Lotus Engineering on the
Midsize CUV. Following the Phase 2 studies, the Aluminum Association Automotive
Technology Group contracted with ED AG, Inc. to evaluate aluminum material replacement
within EPA's CAE model of the Midsize CUV BIW. A cost analyses was also performed by
ED AG for this project.
2.2.7.4.1.1 Phase 2 Low Development Midsize C UV Updated Study and Supplement
2-157
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The Phase 2 Low Development (steel BIW) Midsize CUV lightweighting study was
completed in August of 2012. The results of this work were peer reviewed through an
independent contractor as well as through the SAE paper publication process. Feedback was
received by OEMs and others independent of the official peer review process.
The MY2010 Toyota Venza was chosen as the base vehicle for this work and vehicle
teardown and coupon testing revealed that the base vehicle BIW included high strength steel
components made of HSLA 350, HSLA 490, DP500, a 7000 aluminum rear bumper and HF1050
B pillar and side roof rail. After consideration of nearly 150 lightweighting ideas, the project's
final lightweighting results stated that 18.5 percent mass reduction was achieved for a cost
savings of $0.47/kg. The report also stated that if aluminum doors were included then the mass
save would be 20.2 percent with a cost savings of $0.11/kg. To make the non-compounded cost
curve, the primary lightweighting ideas were listed with the lowest $/kg to the highest $/kg
which reflects an approach where the OEMs would choose the less expensive, or cost saving,
technologies first. Then the mass and cost data were individually cumulatively added and a
cumulative $/kg was determined at each technology addition to create the non-compounded
curve. The compounded curve was developed by determining the secondary mass savings at the
primary solution point and then the mass savings were ratioed across the primary cost curve to
yield the final cost curve with compounding. A short summary of this work and the cost
curve(Figure 2.44) were included in the 2012 FRM analysis.
Vehicle Level Cost Curve
fc-UU ¦
¦1.KJ _
I
Tl
i
i
*5
E
e
-3
J
-IL'IKJ I W
12.00
VclikJi.- IVu.i Ru^xtim
Figure 2.44 Original Phase 2 Low Development Midsize CUV Lightweighting Cost Curve448
Additional consideration was given to the feedback EPA and FEV received on the study as
well as to methodology updates which were made during the MY2011 light duty truck
lightweighting study after the FRM. Modifications made to the data for the original curve, shown
in Figure 2.44, included adding in the aluminum doors as a lightweight technology, and
removing several features including the magnesium engine block and the cost savings for some
of the light weighted plastic components. Several customer features were put back into the
vehicle including the lumbar and active head rest for the back seat and the cargo cover. A mass
and cost allowance for NVH was added as well as the related cost savings for the secondary
mass which had not been accounted for in the FRM methodology. The revised cost curve is
shown in Figure 2.45 and is 17.6 percent mass reduction at +$0.50/kg. Also included are the
$/kg and percent mass reduction solution points for two aluminum BIW Midsize CUV studies.
First is the work funded by ARB from Lotus Engineering on the Phase 2 High Development
3*
IWp1
KW
23'
-w/Compound he
-¦.V.'liLCrtipOLl-K. fig
-GptknlKd Vehicle SolJtton
; S047/hft 1S.26*|
2-158
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Midsize CUV aluminum intensive project which utilized an aluminum BIW design and results
came in at -$0.64/kg for 31 percent MR,452 per our calculations of study results. Second is the
aluminum intensive point from the Aluminum Association work of 27.81 percent mass reduction
at $1.12/kg, in which ED AG utilized the same CAE baseline model developed for the EPA
Phase 2 Low Development Midsize CUV work.454
$4
EPA Final Vehicle Solution (HSS BIW)
($0.50/kg @ 17.6% Mass Reduction)
o o-$4
ARB/Lotus Engineering Aluminum
Intensive Design (-$0.64/ke (est) at 31%)
^—i—i—i—i—i—i—i—i—i—\f—\-
20%
3968.3X3 - 1282.6X2 + 160.78x - 9.9319
25%
30*
Aluminum Association Inc. Published Data Point
Developed from EPA Venza Analysis and EDAG Al
Intensive BIW ($1.12/kg 27.81% Veh MR)
-$12
% Vehicle Mass Reduction
Figure 2.45 Revised Cost Curve for the Midsize CUV Light Weighted Vehicle
This cost curve, in Figure 2.45, is clearly different from the 2012 FRM cost curve for mass
reduction, in Figure 2.43, in which all mass reduction points were associated with positive costs.
The EPA Phase 2 Low Development Midsize CUV holistic vehicle study is a whole vehicle
study which examines nearly every component in the vehicle for mass reduction potential and
calculates a related cost and mass save for each and reviews them from most cost/kg saved to
most costly cost/kg. This methodology was chosen based on the understanding that OEMs will
choose the cost saving technologies first and that some cost mass reduction technologies will be
paid for by the cost save mass reduction technologies. A vehicle cost curve similar to the FRM
expression could be achieved if cost technologies were listed first in the cumulative adding
approach and hence losing the appearance of the cost saving technology ideas. However, this is
not the approach that OEMs are utilizing for lightweighting. For example, a 2016 publication by
CAR contains an illustration and caption which states that "(Figure 2.46) illustrates a generic
cost curve for lightweighting that is broadly supported."450 GM has also claimed publicly to its
potential investors that over $2B449 was saved in material costs, which suggests that costs can be
saved with mass reduction over several passenger vehicles. It is very likely that some of this
savings was due to the decreased material costs over the past year in addition to the cost-saving
lightweighting approaches.
2-159
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 7: General Auto Manufacturer Cost Curve to Lightweight Vehicles
(wt
INCRiA&fc
$0
I
COST
SAVINGS
Cost Savings
Marginal
Expensive
^ HIGH
% Moss Reduction
©Center for Automotive Research Page | 12
Figure 2.46 Cost Curve Figure from CAR: "A Cost Curve for Lightweighting That Is Broadly Supported"450
2.2.7.4.1.2 Light Duty Pickup Truck Light- Weighting Study
The U.S. EPA NVFEL contracted with FEV North America to perform this study utilizing the
methodology developed in the Midsize CUV lightweighting effort (2012) and the study was
completed in 2015. The results of this work went through a detailed and independent peer
reviewed as well as through the SAE paper publication process. Feedback was received by
OEMs and others independent of the official peer review process.
For this study a 2011 Silverado 1500 was purchased and torn down. The components were
placed into 19 different systems. The components were evaluated for mass reduction potential
given research into alternative materials and designs. The alternatives were evaluated for the
best cost and mass reduction and then compared to each other. CAE analyses for NVH and
safety was completed for the baseline and the light-weighted aluminum intensive vehicle. A
high strength steel structure with aluminum closures was the first choice of a solution for this
project; however, this was not fully completed for the decision was made by the project team to
change course and pursue the aluminum structure solution due to the expected introducti on of the
aluminum intensive F150 into the marketplace. Durability analyses on both the baseline and
light-weighted vehicle designs were performed through data gathered by instrumenting a
Silverado 1500 light duty pickup truck and operating it over various road conditions. Included in
the durability analyses are durability evaluations on the light weighted vehicle frame, door and
other components in CAE space. The crash and durability CAE analyses allowed for gauge and
grade determinations for specific vehicle components. Load path redesign of the light duty truck
structure (cabin and box structure and vehicle frame) was not a part of this project.
As shown in Figure 2.47, the most mass reduction was achieved in the Body System Group -
A- (Body Sheet metal) in which the cabin and box structure and the closures, etc. were converted
2-160
-------
Technology Cost, Effectiveness, and Lead Time Assessment
to aluminum. The suspension system is the second highest system for mass reduction and
includes composite fiber leaf springs. Mass reduction technologies with cost save examples
include 1) material and design optimization in the connecting rods, 2) material and design
through use of vespel thrust washer versus roller bearings, 3) material processing in the Polyone
and Mucell applications, 4) material substitution in the thermoplastic vulcanizates (TPV) vs.
EPDM static and dynamic weather seals, 5) material and part consolidation in the passenger side
airbag housings, and 6) design and processing through incorporation of the half shafts and the
Vari-lite® tube process by U.S. Manufacturing Corporation. A complete listing of vehicle
technologies can be found in the online report4"1 and Figure 2.47 shows that there was a 50kg
and $150 allowance for NVH considerations.
Mass Reduction Impact by Vehicle System
(Includes Secondary Mass Savings)
Item
Systerr ID
Description
Base
Mass
"kg"
Mass
Reduction
"kg"(1)
Cost
Impact
NIDMC
T (2)
Cost/
Kilogram
NIDMC
"$/kg" (2)
Cost/
Kilogram
NIDMC +
Tooling
"$/kg" (2)
System
Mass
Reduction
"%"
Vehicle
Mass
Reduction
"%"
1500 Series Chevrolet Silverado Pick-Up Truck
1
01
Engine System
239 9
31 8
-92 83
-2 92
-263
13.3%
13%
2
02
Transmission System
145.3
39 4
-96 57
-2 45
-2 47
27 1%
1.6%
3
03A
Body System Group -A- ( Body Sheetmetal)
574 7
207 1
-1194 86
-5 77
-5 77
36.0%
8.4%
4
03B
Body System Group -B- (Body Interior)
247 0
34 0
-127 23
-3.74
-3 78
13 8%
1 4%
5
03C
Body System Group -C- (Body Exterior Trim)
40 5
2.1
2 73
1.28
1 28
5.3%
0.1%
6
03D
Body System Group D (Glazing & Body Mochatronics)
50 0
4.5
2.30
0.51
0.51
8.9%
0.2%
7
04
Suspension System
301.2
105.4
-154.90
1 47
-1.48
35.0%
4.3%
8
05
Driveline System
183 8
204
3801
1 86
1 89
11.1%
0.8%
9
06
Brake System
101 0
45 8
-14892
-3 25
-3 35
45.4%
1 9%
10
07
rrame and Mounting System
267 6
23 7
-54 42
-2 30
-2 30
0 9%
1.0%
11
09
Exhaust System
38 4
6 9
-13 69
-1 97
-1 97
18.1%
0 3%
1?
10
Fuel System
26 3
7 3
11 9?
1 6?
1 77
27 9%
0 3%
13
11
Steering System
32 5
8 5
14/ 46
-1/44
-1/45
26 0%
0 3%
14
12
Climate Contiol Syslem
20 3
1.9
14.71
7 59
7 59
9.5%
0 1%
15
13
Information. Gage and Warning Device System
16
02
0.66
2 66
2 97
15 7%
0.0%
16
14
Electrical Power Supply System
21 1
12 8
-172 73
-13 49
-13 44
60.6%
0 5%
17
15
In-Vehicle Entertainment System
2 2
0 0
0 00
000
000
0 0%
0 0%
18
17
Lighting System
96
0 4
-2 00
-5 18
-5 18
4 0%
0.0%
19
18
Electrical Distribution and Electronic Control System
330
8.5
01.44
7.20
7 27
25.2%
0.3%
20
00
Fluids and Miscellaneous Coating Materials
116 8
0.0
0.00
0.00
0.00
0.0%
0.0%
a. Analysis Totals Without NVH Counter Measures —»
2454.4
560.9
-2073.82
-3.70
-3.69
n/a
22.9%
b. Vehicle NVH Counter Measures (Mass & Cost) —*
00
-50.0
-150.00
n/a
n/a
n/a
n/a
u. Analysis Tulals Willi NVH Counter Measuies —>
2454.4
510.9
(Decrease)
-2223.82
(Increase)
-4.35
(Increase)
-4.35
(Increase)
n/a
20.8%
(1) Negative value (i.e.. -X.XX ) represents an increase in mass
r(2) Negative value (i.e., -$X.XX) represents an increase in cost
Figure 2.47 Light Duty Pickup Truck Lightweighting Study Results
The individual technology mass and cost saving used to develop the system summaries listed
in Figure 2.47 were used to develop EPA's cost curve for the light duty pickup truck
lightweighting study, as shown in Figure 2.48. It should be noted that the blue squares are
individual solutions and are not based on the cost curve technology points which lead to the red
square solution point.
2-161
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Aluminum Intensive Body and HSS
Intcns-fvc Frame w/ Mass Compounding
. m a
r-
, ,.*
f
Aluminum Infr-
Frame w/ F/a:
snslvc Bodyand
s-Comcoundlrfi
0 i
1!i%
HSS Intensive Body and Frjrre w/
Mes Ccnpourdmg
2f.w
/
% Vehicle Mass Reduction
•m f ij CXI I n pm f H.I I
L Vif Li|-Il|:i hj i:!it |J
Figure 2.48 Light Duty Pickup Truck Lightweighting Cost Curve
The curve without compounding in Figure 2.48 (green curve) includes primary mass
reduction ideas which do not depend on the vehicle being made lighter. The mass reduction
ideas based on a resultant lighter vehicle are called secondary mass saving ideas and are based on
components decreasing in size and hence material. In this study the engine was able to be
downsized 7 percent due to the mass reduction in the vehicle design and still maintain the current
towing and hauling capacities. The other systems that were reduced in size, while considering
truck performance characteristics, included the transmission, body system group A (bumpers),
suspension, brake, frame and mounting systems, exhaust, and fuel systems. The systems
considered for secondary mass are included in Figure 2.49 and show the total 83.9kg mass save
at $68.74 savings. Overall, the secondary mass savings are 17.6TT percent of the primary. The
compounded curve in Figure 2.48 is the EPA light duty truck cost curve utilized in the
development of the overall cost curve for light duty trucks described in Section 2.3.
TT % Secondary Mass = 560.9 compounded-83.9secondary =477kg primary, 83.9/477 = 17.6% secondary.
2-162
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Secondary
Mass Savings (SMS) Impact by Vehicle System
Item
O)
*<
«T
3
Description
Base
Mass
Mass
Reduction
with SMS
Mass
Reduction
without
SMS
"kg" (1)
Incremental
Mass
Reduction
Cost
Impact
NIDMC Willi
SMS
(?)
Cost
Impact
NIDMC
without
Savings
Horn SMS
"$" (2)
Cost/
Kilogram
NIDMC Willi
SMS
"$/kg" (?)
Cost/
Kilogram
NIDMC
without
Cost
Savings/
Kilogram
NIDMC
6
Kg
"kg" (1)
from SMS
"kg" (1)
SMS
T (2)
SMS
"$/kg" (2)
from SMS
"$/kg" (2)
1500 Series Chevrolet Silverado Pick-Up Truck
1
01
Engine System
239.9
31.8
23.8
8.0
-92.83
-114.63
21.81
-2.92
-4.82
1.90
2
02
Transmission System
145.3
39.4
34.2
5.2
-90.57
-128.20
31.64
-2.45
-3.75
1.30
3
03A
Body System Group -
A- ( Body Sheetmetal)
574.7
207.1
190.7
16.4
-1194.86
-1125.15
-69.71
-5.77
-5.90
0.13
7
01
Suspension System
301.2
105.-1
83.1
22 A
-154.90
-260.84
105.94
-1.47
-3.14
1.67
9
06
Brake System
101.0
45.8
43.9
2.0
-148.92
-167.87
18.95
-3.25
-3.83
0.58
10
07
Frame and Mounting
System
267.6
23.7
0.0
23.7
-54.42
0.00
-54.42
-2.30
0.00
-2.30
11
09
Exhaust System
38.4
6.9
6.3
0.6
-13.69
-19.54
5.85
-1.97
-3.08
1.11
12
10
Fuel System
26.3
7.3
1.6
5.7
11.92
3.25
8.67
1.62
2.02
-0.40
a. Analysis Totals Without NVH
1694.5
467.5
383.6
83.9
-1744.26
-1813.00
68.74
-3.73
-4.73
0.82
Counter Measures —
(Increase)
(Increase)
(Increase)
(Increase)
(1) Negative value (i.e., -X.XX ) represents an increase in mass
r(2) Negative value (i.e., -$X.XX) represents an increase in cost
Figure 2.49 Light Duty Pickup Truck Lightweighting Study Secondary Mass
2.2.7.4.2 NHTSA Holistic Vehicle Mass Reduction Cost Studies
To support the Midterm Evaluation, NHTSA funded two holistic vehicle mass reduction/cost
studies. These studies are described in full detail in the Draft TAR. For complete information on
these studies, please see Draft TAR Section 5.2.7.4.2 (Draft TAR page 5-176).
2.2. 7.4.3 ARB Holistic Vehicle Mass Reduction Cost Study
The California Air Resources Board funded Lotus Engineering on further analysis of in-depth
cost and CAE, of the Phase 2 High Development of the Midsize CUV.452 The project focused on
the BIW design through CAE and more in-depth costing of the BIW. A full vehicle solution
point was developed by adding the cost and mass save results of the BIW analysis to the cost and
mass save information on the other vehicle systems from the Phase 1 work.453 The report
changed the original BIW design of 30 percent magnesium, 37 percent aluminum, 6.6 percent
steel and 21 percent composites to one of 12 percent magnesium, 75 percent aluminum, 8
percent steel and 5 percent composites, shown in Figure 2.51. The report states that its BIW
design reduced the number of parts from 419 parts in the baseline Venza to 169 parts in the low
mass design. Specifically, the report states "By factoring in the manufacturability of the
materials and designs into the fundamental design process, it is expected that ... this type of
design [will] be production ready in 2020."
The summary write-up for this work is contained within the LD GHG 2017-2025 FRM Joint
Technical Support Document. A cost curve was not developed for this work. Values of cost and
2-163
-------
Technology Cost, Effectiveness, and Lead Time Assessment
overall mass reduction were located in several areas of the report. The overall results, including
all of the mass reduction items in the Phase 1 report and including powertrain were taken from
Table 4.5.7.2.f. totaling 531.2kg reduced (31 percent of 1711kg) and the total cost was taken
from the 4.6.1. Conclusions section of $342/vehicle cost save. The cost per kilogram for this
solution is calculated as -$0.64/kg cost saved. This point, along with two other all aluminum
vehicle solution points - one by NUTS A and the other by the Aluminum Association, helps to
indicate the direction for additional mass reduction beyond the AHSS BIW/Aluminum closure
solution on which the cost curve for the passenger car/Midsize CUV is based.
Ssii
Silver - Aluminum
Purple - Magnesium
Blue - Composite
Red - Steel
/1gun- J 2 i" a Hinth 'Sit'll bill- nwhrriat uMgr front 17I'll
Figure 2.50 Phase 2 High Development BIW - Lotus Engineering
2.2.7.4.4 Aluminum Association Midsize CUV Aluminum BIW Study
The Aluminum Association funded a project with EDAG, Inc.454 in 2012 to perform an
aluminum substitution analysis in the BIW of the Midsize CUV work by EPA using the EPA
CAE baseline model for the work. The baseline model was also developed by EDAG, Inc. The
analyses utilized CAE crash safety and NVH verifications when determining the specifics, gauge
and grade, of the aluminum to be utilized in the BIW (Figure 2.51).
2-164
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Toyota Venza AIV
Rattdlma-Vanjr.a Van7JA.IV
Zfl* Mass
Reduction
Jh.mmru.vti*--* CftwAlLrt Tin :•(
Toyota Venza AIV | Material Selection
| Aluminum 5r«rf«
Weill stress J
GO?? TE Allory Sh«0T
740 Mpa
5754 O AJloy sheet
117 Mpa
SCO? TG fflrwor
SIS Mpa
6encr< Caning
1«0 Mpa
»» «« * i'Ur^tniC Bw^irtunui
Figure 2.51 Midsize CUV Baseline vs Midsize CUV Aluminum Intensive Vehicle
2-165
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Dsscription
Estimated
Mans
Reduction
"Kg"
¦sL mated
Cast Impact
'V
Average
CoKtr
K luqra*n
-s,'Ka"
Erxiy structure Subsystem
Unrfcrlxwly Asy
19,3
-n7..'si3
-3.41
Front Structure Asy
1*1.3
-121.81
-6.-19
Root A&y
14.6
-44.81
-3.07
Bodyside Assy
72.2
-306.60
-4.25
Ladder Asy
3&.1
-235.53
Bolt on BIP Components
3.2
-3.0/
-1.23
Body Closure Subsystem
Hood Asy
7.7
-27.70
-3.62
Front Dow Asy
1&.0
-21.65
-1.44
Rear Dacir Asy
11.3
'10.31
-1.70
Rear Hatch Asy
7.2
-21.21
-2.03
Front Fenders
2.0
-16.22
-4.S5
Sumpens Subitum
Front Bumper Asy
Z.3
-8.0.1
Rear Bumper A&y
0.0
O.OO
o.no
Total?
207.7
-805,01
-1,31
*+" = mass
-------
Technology Cost, Effectiveness, and Lead Time Assessment
mass reduction was achieved relative to a MY2013 Fusion1111 for the Mach 1 design. The overall
BIW design was multi-material with 64 percent aluminum, 29 percent steel and 7 percent hot
stamping. A number of vehicles were built and crashed, including IIHS ODB, with acceptable
results and several notes for further improvement in the BIW design to CAE predictive
correlation were noted. Costing was not a part of this project; however, the SAE paper states
"multi-material automotive bodies can achieve weight reduction with cost effective
performance." 455
Sur,.'idfon o* A >g -a:." CL' / A! BIW
Com Cw V5 Ever dec "nrovgh EC.- £?se 20C3
NHTSA w/Al BiW •- ~
iS2.83/k^ 23,2%)
Ai Assoc, Midsize CUV, EPA
CAE, w/AI BIW ARB Data p°!nt
w/AI BIW on CUV
(-0 64%/kg, 31SS)
Figure 2.53 Car/CUV DMC Curve Extended to Points with Aluminum BIW
Figure 2.53 shows two points for the CUV aluminum intensive solution. One point is from
the ARB-sponsored study by Lotus Engineering456 and one point is from the Aluminum
Association study through ED AG.457 The ARB full vehicle data point with optimized BIW
design and reduction of BIW components is 531kg (31 percent) mass reduction at -$0.64/kg.
The Aluminum Association study of an all-aluminum BIW, based on material replacement into
the CAE model from the original U.S. EPA Midsize CUV study, resulted in a total vehicle
solution of $1.12/kg at a total of 476kg (27.8 percent) mass reduced. NHTSA studied the
aluminum intensive vehicle design for the passenger car (based on the MY2011 Accord) and the
result is a point at $2.83/kg for 23.2 percent.
Table 2.17 shows the detailed results of the studies. The cost/kg estimate for the NHTSA
study is likely overestimated given the recent reduction in the commodity price for aluminum.
The 2001 JOM source document used for the cost estimate indicates that costs have very likely
1111 The MY2013 Fusion was one redesign beyond the 2008 era Fusion. The base vehicle is approximately 250 lbs
heavier and the top trim is approximately 100 lbs heavier in 2013 compared to 2008. The 2013 Fusion is
approximately 2.80sq ft larger in footprint compared to the 2008 era Fusion and slightly taller and wider overall.
Several safety features were also included. (https://en.wikipedia.org/wiki/Ford_Fusion_(Americas))
2-167
-------
Technology Cost, Effectiveness, and Lead Time Assessment
decreased since this work was completed.vv'458 The Lotus Engineering and ED AG are similar
and achieve results for three major systems which are only 6kg apart (201.7kg v 207.7kg
respectively). The differences between the two projects include the BIW designs used and the
resultant estimated costs. The ED AG study used the existing BIW design and the materials of
aluminum alloy sheet, extrusion and casting. The Lotus Engineering solution also utilized the
different aluminum components while optimizing component aggregation as only 169
components were used in the BIW compared to the original 419 and significant savings with the
new manufacturing processes were assumed.
Table 2.17 Three Aluminum Intensive Vehicle Design Summary - DMC ($), %MR and $/kg
Aluminum BIW,
2012 ARB/Lotus
2012 Al Assoc/EDAG
2012 NHTSA/Electricore/
Closures, Chassis
(midsize CUV-1711kg)
(midsize CUV -1711kg)
EDAG (Pass Car-1480kg)
Mass save
Cost
Mass save
Cost
Mass save
Cost
(kg)
($)
(kg)
($)
(kg)
($)
BIW
140.7
239
162.2
780
113
782
Closures/Fenders
59
-381
43.2
106
44
153.7
Bumpers
2
9
2.3
8.6
-
-
SUB-TOTAL
201.7
-133
207.7
894.6
157
935.7
Total Vehicle
530
-342
464*
+520*
343.6
971.9
$/kg
-$0.64/kg
$1.12/kg
$2.83/kg
Note: *adjusted for changes in the EPA baseline Midsize CUV cost curve into which the aluminum BIW was placed
2.2.7.4.6 DOE/Ford/Magna MMLVMach 1 andMach 2 Lightweighting Research Projects
The Multi Material Lightweight Vehicle (MMLV) project was initiated in 2012 by the
Department of Energy and co-funded by Magna International and Ford Motor Corporation under
the project number DE-EE0005574. The objectives of the project included identifying 25
percent (Mach 1) and 50 percent (Mach 2) vehicle mass reduction packages. This work was peer
reviewed through the DOE AMR and the SAE publication processes. The "Multi-Material
Lightweight Vehicles" presentation, which was a combination of the Mach 1 and Mach 2
projects, was peer reviewed at the 2015 DOE AMR in front of a panel of experts in the field and
the results of the peer review were included in the final report for the DOE AMR.459 The project
received a weighted average score of 3.77 out of 4.0 and was measured on reviewer questions
related to approach, technical accomplishments, collaborations, and future research. The results
were also presented in a number of SAE papers and hence reviewed through the SAE publication
process.
The DOE/Ford/Magna project developed the lightweight vehicle solutions off of a MY2013
Ford Fusion platform (used to represent a 2002 Ford Taurus). Results include 23.5 percent for
the Mach 1 design. Seven vehicles were built and the vehicles, and certain components, were
tested under a series of durability tests. New technologies of composite fiber springs, carbon
fiber wheels, seat back frame, and the multi-material body structure were included in the
vv Investigation into the supporting documentation for the analysis revealed that the information was taken from a
2001 article in the Journal of Minerals, Metals and Materials Society. The article states "In fact, design
developments by Audi already have resulted in significant cost reductions between its first- and second-
generation vehicles. These have come about through parts consolidation, process substitutions, and part
simplification."
2-168
-------
Technology Cost, Effectiveness, and Lead Time Assessment
durability tests. For the Mach 2 design, 50 percent mass reduction is achieved however the
vehicle is not market viable due to extensive de-contenting and use of materials that are not yet
ready for full volume production including composite "tub" package tray and roof A
comparison of the MMLV structures weight for BIW, Closure, Chassis and Bumper is displayed
in Figure 2.54.
MMLV Structures Weight Comparison ^ENERGY
BIW, Closure, Chassis, Bumper
$yEHMA
Closures
Chassis
Bumpers
Totals
Baseline
BIW 316.04 kg
92.17 kg
89.07 kg
20.38 kg
517 66 kg
3.1% 3.7% 2.0%
MMLV Mach I
BIW 231.33 kg
Clotures 57.23 kg
Chassis 52.90 kg
Bumpers 11.13 kg
Totals 352.5S kg
31.9% Reduction
MMLV Mach II
BIW
Closures
Chassis
172 39 kg
4 516 kg
30 80 kg
J1.13 kg
259 67 kg
49.8% Reduction
Bumpers
Totals
0.2%
.3.6%
12.7H
i.S%
¦ Aluminum Stamp* ngs m Alum nun Extrusions a fcnt stamping* bCCv/POSTI
• Aiumlnun Cistlnts aSreel ¦ faiten«rc/!J»»vBvlDth«r • MMKKUM warm ccwwhg
" MAEXBIUM
osrna'fc«aM«/ii(ri)uS)SN
• CAD WEIGHT
Figure 2.54 MMLV Structures Weight Comparison BIW, Closure, Chassis, Bumper4
Gaps identified by the MMLV projects (I and II) include those listed in Table 2.18.
Table 2.18 Gaps Identified by MMLV Project
Topic
GAP
Steel
Improved coatings on ultra-high strength steels for multi material applications
Aluminum
Increased die life and bi-metallic (inserts, etc.) for Al die castings plus low cost 7000 series
aluminum sheet and extrusions
Magnesium
High volume warm forming, hemming, class A finish, plus improved die life and bi-metallic
inserts in high pressure vacuum die casting
Carbon Fiber
Composites
Material characterization for CAE, joining, corrosion, paint, class-A finish
Multi Material
Vehicles
Corrosion mitigation strategy including universal equivalent of phosphate (or eqiuv) bath for
any mix of steel, aluminum and magnesium before e-coat and paint
Joining methods with corrosion mitigation
Aluminum rivet, high hardness, high strength
2-169
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Alternative NVH treatments for lightweight panels sheet metal and glazings
Design for disassembly, end of life, for reclaiming, recycling
No cost analysis was performed for the Mach 1 study. A 40-45 percent MR cost analyses
from the base 2013MY vehicle was completed under a separate DOE project, through Idaho
National Laboratories performed by IBIS Associates Inc., and results indicate the cost of carbon
fiber must decrease in order to make the technology viable for mass market vehicles.461 This
project is described in 2.2.7.4.7.
A second cost study was funded by DOE Office of Nuclear Energy and completed in June
2016. The title was "Vehicle Lightweighting: Mass Reduction Spectrum Analysis and Process
cost Modeling" by IBIS Associates, Energetics and Idaho National Laboratory. The objectives
of this report were to "Assess the multiple strategies addressed in the earlier phases in terms of
weight reduction, cost premiums and risk factors in order to establish a prioritized spectrum of
lightweighting opportunities." And "Apply process technical Cost models (TCMs) to priority
lightweight material manufacturing technologies to evaluate cost structures and understand the
relative leverage of key cost drivers. The processes targeted were aluminum extrusion,
magnesium sheet forming and carbon fiber composite molding." This study examined mass
savings and costing for a range of technologies from a number of lightweighting studies
available as of 2015.
2.2.7.4.6.1 Mach I
The MMLV Mach I project achieved 364 kg (23.5 percent) mass reduction from the baseline
weight of the 2013 Ford Fusion (representing a 2002 Ford Taurus). Seven prototype vehicles
were built and these vehicles were used to conduct a number of test such as, corrosion,
durability, NVH (noise vibration harshness), and crash. Maintaining performance and
capabilities, along with safety and durability were also goals of the MMLV. All parts used in the
MMLV are either low volume or high volume production capable up to 250,000 vehicles per
year. The Mach I mass reduction was achieved using materials such as aluminum, carbon fibers,
magnesium, and high strength steels. Results of the Mach I project were presented in 13 SAE
462,463,464,465,466,467,468,469,470,471,472,473,474
pdpcl
The Mach I project group presented an estimate of the fuel economy improvement at the 2015
SAE World Congress 2015 as being an increase to 34 mpg from 28 mpg. This change in fuel
economy was estimated by taking the fuel economy of a Ford Fiesta (which is the equivalent
weight of the lightweight Mach-I) and comparing to the 2013 Ford Fusion. The fuel economy
numbers were from fueleconomy.gov. Key requirements of durability, safety, and Noise
Vibration Harshness (NVH) were also met within the Mach I design as illustrated in a report
presentation at the 2015 DOE AMR.475 All components of the MMLV were specifically chosen
for optimal weight reduction without shorting on performance or technicality.
Five subsystems of the Mach I compared to the baseline 2013 fusion of full body mass
reduction.475
• The body-in-white (BIW) and closures contributed 76 kg (4.9percent) to the overall
vehicle mass reduction. The baseline 2013 BIW is 326 kg and the Mach-I BIW is
250 kg. The 2013 Fusion BIW is steel intensive, and the Mach-I design included
2-170
-------
Technology Cost, Effectiveness, and Lead Time Assessment
advanced high strength steels were integrated for use as primary safety structures like
crush rails, B-pillars, and selected cross car beams. Closures in the Mach 1 were
aluminum intensive. The transition from steel to aluminum is also the primary design
strategy for the light weighting of the deck lid, and front fenders, as well as the side
door structures and hinges. Also, chemically foamed plastics were used in the door
design as trim.
• Body Interior and Climate Control consists of the seats, floor components, instrument
panel/ cross car beam (IP/CCB), and climate control system which contributed 28 kg
(1.8 percent) to the overall vehicle mass reduction. The IP/CCB decreased in part
count from 71 to 21, new material design involved carbon fiber reinforce nylon from
the baseline welded assembly of steel stampings and tubes. The material selection of
the seat structures was carbon fiber reinforced nylon composite compared to the
baseline steel stampings and tubes.
• Chassis subsystem reduced its total mass by 98 kg (6.3 percent) to the overall vehicle
mass reduction. The major components identified in the Mach 1 subsystem include
hollow coil springs, carbon fiber wheels, and tires with a tall and narrow design,
hollow steel stabilizer bars, aluminum sub frames, control arms and links.
• The powertrain subsystem was reduced by 73 kg (4.7 percent) to the overall vehicle
mass reduction. The baseline engine is a 1.6 liter four-cylinder gasoline turbocharged
direct injection (EcoBoost) with a six-speed automatic transmission. The Mach-I
design has a 1.0 liter three-cylinder gasoline turbocharged direct injection (Fox
EcoBoost) with a mass reduced six-speed automatic transmission. The use of carbon
fiber within this subsystem encouraged mass reduction and include components such
as the engine oil pan.
• The electrical subsystem achieved a 10 kg (0.64 percent overall vehicle mass
reduction). A few adjustments were made to accomplish this number. The battery
was switched to a lithium ion 12-volt start battery from the baseline lead-acid battery.
The change of the battery achieved 5 kg mass reduction. Also, copper electrical
distribution wiring was replaced with aluminum conductors meeting a 4 kg mass
reduction. The remaining 1 kg mass reduction was achieved by small adjustments to
the speakers, alternator, and the starter motor.
The Mach-I used computer aided engineering (CAE) for many safety simulations in addition
to performing a number of actual vehicle safety crashes. Seven MMLV Mach-I vehicles were
built and selectively tested. Seven different validation tests were completed as listed in Table
2.19.
Table 2.19 Safety Tests Performed on the Mach-I.
VEHICLE
TESTING
Test Buck
Body-in-White + Closures + Bumpers + Glazing + Front
Subframe - Body-in-Prime NVH modes, global stiffness,
attachment stiffness, selected Durability
Durability A
DRIVABLE, full MMLV content with Fusion powertrain -
MPG Structural Durability, Square Edge Chuckhole Test
for Wheels and Tires
2-171
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Corrosion A Traditional Surface
Treatments
DRIVABLE, with alternative surface treatment and paint
process - MPG Corrosion R-343. Humidity soaks and
salt spray etc.
Corrosion B MMLV Alternative
Surface Treatments
DRIVABLE, with traditional surface treatment and paint
process - MPG Corrosion R-343. Humidity soaks and
salt spray etc.
Safety A
NON-Drivable, most MMLV content, without carbon
fiber instrument panel - Low Speed Damageability test
(front) Right Hand (passenger) side - IIHS Front ODB
40% Offset 40 mph, Left Hand (driver) side - Side Pole
Test on Right Hand (passenger) side (FMVSS 214)
Safety B
NON-Drivable, most MMLV content, without carbon
fiber instrument panel - NCAP Frontal 35 mph rigid
wall, then 70% Offset Rear Impact (FMVSS 301)
NVH + Drives
DRIVABLE, full MMLV content with downsized and
boosted powertrain, 1.0-liter 13 EcoBoost, gasoline
turbocharged direct injection engine plus six-speed
manual transmission - Wind Tunnel, Rough Road
Interior Noise, Engine & Tire Noise, Ride & Handling
The overall outcome of the safety and durability tests provided assurance a multi-material
lightweight vehicle was successful. Noise Vibration Harshness was tested in a high frequency
range of 200-10000 Hz and fell within acceptability but slightly short of requirements. Durability
test classified the Mach-I as a durable vehicle and showed no major cracking or durability
incidents in the test mileage. Frontal crash safety tests showed that nine parts withstood the test
at a good level. Table 2.20 is a list of the parts that performed the best. The carbon fiber wheels
had one issue in the durability test with the outer coating on the carbon fiber, however it was
solved and the wheel is currently planned for the Shelby Mustang. The composite fiber springs
performed better than expected and it is understood that they are in production, or planned for
production, in the Audi A6 Ultra Avant and the Renault Megane Trophy RS vehicles. The
durability issue for the composite fiber wheels was solved and the improved wheels are being
employed in the Shelby Mustang. Some new discoveries were made including the near zero
mass add for NVH considerations and corrosion concerns will be better addressed with a correct
amount of sealant and the proper choice of nuts and bolts in the multi material vehicle design.
Table 2.20 Mach-I Components to Maintain Frontal Crash Performance.
PART
MATERIAL
Front bumper
Extruded aluminum
Crush Can
Extruded aluminum
Subframe
Cast and extruded aluminum
Shock Tower
Cast aluminum
Coil Spring
Chopped glass fiber composite
Wheel
Woven carbon fiber composite
A-Pillar joint node
Cast aluminum
Windshield
Chemically toughened laminate
Seat frame
Woven carbon fiber composite
2.2.7.4.6.2 Mach 2
2-172
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The goal of the Mach 2 project was to create a lightweight design that achieved 50 percent
mass savings from the 2013 Ford Fusion (representing a 2002 Ford Taurus). This amount of
mass reduction is forward looking and of limited use for the time frame considered for this
Proposed Determination (2022-2025) which has a top application of 20 percent mass reduction.
The project achieved 51.1 percent (798kg) mass reduction with a significant degree of mass
reduction using materials and processes that have some initial research but not ready for high
volume. Significant vehicle de-contenting was employed which included items from air
conditioning to thinning the windows and the resultant vehicle was not marketable.
The vehicle technologies for the BIW and Closures includes carbon fiber and composites as
seen in Figure 2.55. However, the CAE inputs were not mature for the materials and as a result
the outputs were insufficient. CAE information included cards for stiffness, durability, and
fatigue analyses. In terms of production, the composite material and manufacturing infrastructure
was also not mature for automotive volumes. The carbon fiber and composite panels were not
deemed acceptable for Class A surfaces and as a result aluminum or magnesium sheet products
were chosen for the BIW and closure applications.
Table 2.21 Mach II Design Vehicle Summary475
System
Technology
Material/Approach
Body and Closures
Body
Composite intensive
Closures
Magnesium
Windows
Reduced Thickness
Interior & Climate
Control
Seats
Carbon fiber seats with reduced function
IP
Carbon fiber composite
Reduced content
No bins, center console, air conditioner, etc.
Chassis
Subframes
Cast magnesium
Coil Springs
Composite
Reduced
capacity
For reduced weight cargo and towing
Powertrain
Engine
1.0L 3 cyl naturally aspirated
Remove turbocharger and intercooler
Material change
Transmission
Reduced capacity manual
Electrical
Eliminate content and features
Reduced battery, alternator, wiring
2-173
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Mach II Design KsraaHl 7,1 T I J
Mixed Material BIW & Closures EHERCi*
» *1 •MptefVti 1 ,-]-*T».vHr hi
Body-in-Whlte (BIW) 155 kg mass reduction from baseline (47.5%)
ALUMINUM SHEET ¦ ALUMINUM EXTRUSION ¦ COMPOSITE ¦ MAGNESIUM EXTRUSION
I ALUMINUM CASTING ¦ HOT STAMPING MAGNESIUM SHEET ¦ MAGNESIUM CASTING
CLOSURES 47.0 kg mass reduction from baseline (48.0%)
• ^
51%
1%
Figure 2.55 Mach II Mixed Material BIW and Closure Design (brown is carbon fiber)4
2.2.7.4.7 Technical Cost Modeling Report by DOEINL/IBIS on 40 Percent-45 Percent
Mass Reduced Vehicle
The U.S. Department of Energy's Vehicle Technologies Office Materials Area funded a study
to provide cost estimates and assessment of a 40 percent and 45 percent weight savings on a
North American midsize passenger sedan based on the work of the Mach 1 and Mach 2
lightweighting projects. The title of the report is "Vehicle Lightweighting: 40 percent and 45
percent Weight Savings Analysis: Technical Cost Modeling for Vehicle Lightweighting"476.
This work was peer reviewed through the 2015 DOE AMR "Technical Cost Modeling for
Vehicle Lightweighting". Results of the peer review were included in the final report for the
DOE AMR.477
The goal of the work was to achieve 40 percent-45 percent mass reduction relative to a
standard North American midsize passenger sedan at an effective cost of $3 42/lb. This study
utilized existing mass reduction and/or cost studies including those from FEV, Lotus
Engineering, DOE Mach 1 and Mach 2. The Executive Summary to this report states "The
analysis indicates that a 37 to 45 percent reduction in a standard mid-sized vehicle is within
reach if carbon fiber composite materials and manufacturing processes are available and if
customers will accept a reduction in vehicle features and content, as demonstrated with the
Multi-Materials and Carbon Fiber Composite-Intensive vehicle scenarios."
2-174
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Vehicle Llghtweighhng Scenario Companion
$27,000
525.000
i 11 ban Slage 5
; (45* Weigfit Reduction
I
o
S23.000
521,000
£ S19.000
Ml >Materiat Slage *•
(Maximum Potential)
$3.42i1b Cost Targe
fy Carbon Stage 4
: (40*% Weight Reduction)
Carbon Scenario Meeting
Cost Target (56i1b tor materials
and 55.1b for processing)
Carbon Scenario Meeting
Cost Target |$4 20.» lor materials
and SS>'lb lor processing)
517,000
Aluminum Intensive Stage 4
(Maximum Potential)
$15,000 1 . 1 . 1 . 1 1 1
30% 32% 34% 36% 38% 40% 42% 44% 46% 48% 50%
% Weight Savings
Figurc ES-1. Costing results of advanced wight savings scenarios based on different material sy stems Carton scenarios assume an optimistic,
projected, carbon composite processing cost of S5flband current carbon fiber price of SI2.5
-------
Technology Cost, Effectiveness, and Lead Time Assessment
3. "High Risk strategies are needed to achieve the highest levels of weight reduction that
approach 45 percent overall vehicle weight savings with cost of savings up to $7.00/lb under
optimum conditions. Requires: carbon fiber at significantly reduced cost per pound, Extensive
use of Mg, advanced electrical and interior systems, consumer acceptance of some de-
contenting."478
Weight Saving* Cost of Weight Savings
Low Risk
Medium Risk
Hlfh M»k
» "
tnm
UU> »
V
!«.
1 — Calrf ImM| SM
WMstolMtCflMtf laMW
5
I
* *0
UM J
luue |
lua J
I"
* m
1 1
*
— j
• J
^ * 1 II 1 I « 41 1
Illli.
i-cao
liVM
liiiippij
Iflpll!; I
liipijili 1 Pii ill iPiii
il1,1,1 p1'?
!]!" i J •"*( i ; ijiljj in
<3
Figure 2.57 Results for Weight Reduction Strategies by Risk Factor and Cost of Weight Savings
2.2. 7.4.9 Studies to Determine Potential Mass Addition for IIHS Small Overlap
One of the requirements of the IIHS Top Safety Pick is to meet the IIHS small overlap (SOL)
crash test (see Figure 2.58). The IIHS SOL test is designed to reproduce what happens when the
front corner of a vehicle hits another vehicle or an object like a tree or utility pole. Estimating
the mass impact to succeed this test can vary widely among different types of vehicles. The
structure of the vehicle must be redesigned in order to design load paths such that the passenger
compartment remains sound throughout the crash event.
Figure 2.58 Post-test Laboratory Vehicle of IIHS Small Overlap Test
2-176
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Two studies were funded to examine the mass add to existing vehicle study models. NTHSA
funded the passenger car study using their LWV model and Transport Canada funded the light
duty truck study using the LDT model from the EPA light duty pickup truck study. All of the
CAE modeling, from the base studies to the IIHS small overlap studies were performed by two
separate groups within ED AG, Inc. The results of these studies are described in the following
sections.
2.2.7.4.9.1 NHTSA Mass Add Study for a Passenger Car to Achieve a "Good" Rating on the
IIHS Small Overlap
The analysis of the IIHS Small Overlap resultant mass add for a variety of unibody passenger
car vehicle classes are included in the February 2016 report "Update to Future Midsize
Lightweight Vehicle Findings in Response to Manufacturer review and IIHS Small-Overlap
Testing."479 In order to improve the structural performance during the IIHS SOL test, several
options were considered and implemented using a detailed LS-DYNA crash model that was
originally part of the NHTSA LWV study. Changes regarding the SOL test include
reinforcement of major areas in the body structure and were designed for easy manufacturability
and assembly into the body structure. The findings for the IIHS SOL solution was a mass
addition of 6.9 kg and $26.88 in cost.
The report also includes the IIHS mass add results for a range of unibody vehicle classes as
shown in Table 2.22 (MY2010) and Table 2.23 (MY2020). The overall Light Duty Vehicle
Average is based on a straight average of the values for each vehicle class. The report also notes
that estimated mass increases for 'body on frame' vehicles should be further reviewed due to a
differing body structure design. This was done in Transport Canada's evaluation of the 2011
Silverado 1500 discussed in the section following this section.
Table 2.22 Estimated Mass Increase to Meet IIHS SOL for 2010 Vehicle Classes
2010 Vehicle Class Average
Vehicle Class
Curb Vehicle
Test Vehicle
Increase in mass to
Curb Vehicle Weight with IIHS
Weight (kg)
Weight (kg)
meet IIHS SOL (kg)
SOL Changes (kg)
Sub-Compact Car
1261
1411
7.4
1268
Compact Car
1345
1495
7.8
1353
Mid-Sized Car
1561
1711
8.9
1570
Small SUV/LT
1592
1742
9.1
1601
Large Car
1752
1902
9.9
1762
Mid-Sized SUV/LT
1916
2066
10.8
1927
Minivans
2035
2185
11.4
2046
Large SUV/LT
2391
2541
13.3
2404
Light Duty Vehicle
1732
1882
9.8
1741
Average
Table 2.23 Estimated Mass Increase to Meet IIHS SOL for 2020 Vehicle Classes
2020 Vehicle Class Average
Vehicle Class
Curb Vehicle
Test Vehicle
Increase in mass to
Curb Vehicle Weight
Weight (kg)
Weight (kg)
meet IIHS SOL (kg)
with IIHS SOL
Changes (kg)
Sub-Compact Car
1055
1205
6.3
1062
Compact Car
1119
1269
6.6
1125
2-177
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Mid-Sized Car
1294
1444
7.5
1302
Small SUV/LT
1318
1468
7.7
1326
Large Car
1453
1603
8.4
1462
Mid-Sized SUV/LT
1632
1782
9.3
1641
Miriivans
1689
1839
9.6
1699
Large SUV/LT
1962
2112
11.0
1973
Light Duty Vehicle Average
1440
1590
8.3
1449
2.2.7.4.9.2 Transport Canada Mass Add Study for a Light Duty Truck to Achieve a "Good"
Rating on the IIHS Small Overlap
Transport Canada funded a project with ED AG, Inc.480 in which a body on frame 2013MY
Silverado 1500 light duty pickup truck (designed in 2007) was evaluated and modeled in order to
achieve a "Good" rating on the IIHS small overlap crash test. The study utilized the work done
by FEV in EPA's light-weighting light duty pickup truck study and has been peer reviewed
through EPA's peer review process.
The baseline CAE model was used to correlate the modeled performance with an actual
impact test conducted at Transport Canada's Motor Vehicle Test & Research Centre in
Blainville, Quebec. The state of the truck from the barrier impact is shown in Figure 2.59. A
number of components were material tested through the assistance of Natural Resources
Canada's CanmetMATERIALS facility in Hamilton, Ontario. This was done in order to ensure
that the most accurate materials properties were being input into the baseline model at the start of
the process and in order that the CAE modeling could reproduce the video from the actual crash
test as closely as possible. The baseline model was modified with failure criteria and timing of
respective components involved in the IIHS small overlap test. Figure 2.60 shows the baseline
model correlating to the baseline truck crash event.
Ik...
¦4. JL. jC - . -
P * 4 • •
rJL* * Z"** * ¦ • ^9' *
- - * - ¦> ¦ ¦ »«V \
^ •
Figure 2.59 MY2013 Silverado 1500 IIHS Small Overlap Test Crash Before and During
2-178
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.60 Converting the Actual Crash Event to a Model
Development of the light duty truck design modifications to the baseline structure began with
research on existing IIHS crash results including those from the GM Equinox, Mercedes ML,
and design information on the 2014MY Silverado 1500 and the 2015MY Ford F150 which had
been released before the conclusion of this project. A solution for a "Good" rating on the IIHS
small overlap crash test was determined for the steel intensive vehicle in order to highlight the
areas for improvement in the lightweight model. The mass add for this design was not optimized
for the minimum mass add that would still achieve a "Good" rating.
To develop the lightweight model mass add to the "Good" rating on the IIHS small overlap,
the vehicle lightweighting ideas from the original U.S. EPA lightweight light duty truck project
were first adopted onto the vehicle. The solution from the baseline vehicle was then optimized
and the mass add determined. The report states "Like the original EPA Project cab, the T5-LW
(light-weighted) cab exploited the low density and manufacturing methods specific to
Aluminum, .. .Extrusions and castings were used to meet and exceed the static bending and
torsion requirements with mass efficient solutions." The components in the area of the crash
(including suspension and wheel) were not changed to aluminum for the failure informati on for
the aluminum components were not available. The resultant light-weighted model before and
after IIHS small overlap crash is illustrated in Figure 2.61. The passenger compartment stays in
tact as shown.
Figure 2.61 Light Weighted Model in the IIHS Small Overlap Crash Test
The accelerations for the dummies will change based on the stiffer passenger compartment
which doesn't allow the extreme intrusions in the baseline model. The report contains a
comparison of the Velocity (m/s) at CoG X-velocities for the T4-GA LDT model and other
production vehicles with "Good" IIHS small overlap results and the results are similar. The T5-
LW results are very similar to the T4-GA results. The report concludes that "the pulse response
2-179
-------
Technology Cost, Effectiveness, and Lead Time Assessment
is considered reasonable and it is expected that a modern restraints system could be tuned to
manage the vehicle response." 480
The IIHS Small Overlap Rating is based on dummy injury criteria as well as vehicle intrusion
in specified locations within the vehicle. Figure 2.62 illustrates how the light-weighted model
(T5-LW) compares to the baseline model (T3-BL) along with the results from the original crash
test (TC13-018). The light-weighted model, with the countermeasures resulting in the addition
of 17kg relative to the baseline model, achieves a Good rating in the intrusion part of the
evaluation.
IIHS 51mcbj rjl P.jl n)
ACCCPT-ADtC
£500D
Lower Occupant Compartment
Upper Occupant Compartment
Figure 2.62 Results of the Project Models from Baseline to Light Weighted on the IIHS Small Overlap480
2.2.7.5 Potential Lightweight Recyclable Composite Fiber Material
A new recyclable thermoset technology was presented at the 2016 GALM UK conference.481
While thermoset and thermoplastic technologies (plastics, composite fiber, etc.) provide
lightweighting potential, there are several concerns over their increased use. Topics such as
emissions during production and limited scrap/end of life recycling for thermoplastics with no
potentials for thermoset recycling. Two milestones were achieved this year which may bring this
material into the price range for consideration by OEMs in the future. First, a new technology
developed at the University of Colorado Department of Chemistry and Biochemistry Materials
Science and Engineering Program provides possibilities for a thermoset material that addresses a
number of issues currently present in composite fiber usage for high volume vehicle production.
The startup company Mallinda482 is still in material development; however, noted characteristics
of the material include: eliminate curing, improved manufacturing economics, enabling
composite thermoforming, reduced manufacturing cycle time, re-moldable, solvent free for heat-
induced vitrification. The material has chemically reversible polymerization potential and as
such is closed-loop recyclable which reduces scrap. A ratio of 33 percent recyclable material
and 67 percent new material is used to make new product. The material can also be repaired and
can be used to repair other plastics/composites. Mallinda received a $750k grant for reusable
carbon-fiber composite from the Phase II funding by the National Science Foundation's Small
2-180
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Business Innovation Research program (SBIR).483 Mallinda is also working within Cyclotron
Road484 which is a home for entrepreneurial researches to advance technologies until they can
succeed beyond the research lab. The purpose of Cyclotron Road is to support critical
technology development and help identify the most suitable business models, partners, and
financing mechanisms for success.
Second, composite fiber material developed with this thermoset technology will require low
cost carbon fibers. Presenters at 2016 GALM UK identified the limitations of current carbon
fiber production including expense of producing the material and time to build production
sufficient for automotive use. The Oakridge National Laboratory485 announced in March of 2016
that they have made great strides in advancing carbon fiber technology. The March 2016 article
states "Researchers at the Department of Energy's Oak Ridge National Laboratory have
demonstrated a production method they estimate will reduce the cost of carbon fiber as much as
50 percent and the energy used in its production by more than 60 percent."486 These two
technologies used together, along with repair and recycling potentials, may put a composite fiber
material within price range of OEM considerations in the future.
2.2.8 State of Other Vehicle Technologies
2.2.8.1 Electrified Power Steering: State of Technology
Compared to conventional hydraulic power steering, electrified power steering can reduce
fuel consumption and CO2 emissions by reducing overall accessory loads. Specifically, it
reduces or 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. Power steering may be electrified
on light duty vehicles with a standard 12V electrical system; however, electric power steering
could benefit from a 48V vehicle architecture by reducing electrical current and allowing higher
steering loads. Electrified power steering is also an enabler for vehicle electrification since it
provides power steering when the engine is off.
Power steering systems can be electrified in two ways. Manufacturers may choose to
completely eliminate the hydraulic portion of the steering system and provide electric-only
power steering (EPS) or they may choose to move the hydraulic pump from a belt driven
configuration to a stand-alone electrically driven hydraulic pump. The latter system is referred to
as electro-hydraulic power steering (EHPS).
The Draft TAR noted that EPS has been successfully implemented on all light duty vehicle
classes (including trucks) with a standard 12V electrical system, eliminating the need to consider
EHPS on larger vehicles. For the cost and effectiveness assumptions EPA has used in this
Proposed Determination analysis, see Chapter 2.3.
2.2.8.2 Improved Accessories: State of Technology
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").
2-181
-------
Technology Cost, Effectiveness, and Lead Time Assessment
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. EPA also included a higher efficiency alternator in this category to improve the cooling
system.
EPA considered whether to consider electric oil pump technology for inclusion in their
technology assessments. Because it is necessary to operate the oil pump any time the engine is
running, electric oil pump technology was judged to have an insignificant effect on efficiency.
Therefore, it is not included in this Proposed Determination assessment.
For the cost and effectiveness assumptions EPA is adopting for this Proposed Determination,
see Chapter 2.3.
2.2.8.3 Secondary Axle Disconnect: State of Technology
2.2.8.3.1 Background
All-wheel drive (AWD) and four-wheel drive (4WD) vehicles provide improved traction by
delivering torque to both the front and rear axles, rather than just one axle. Driving two axles
rather than one tends to consumes more energy due to additional friction and rotational inertia.
Some of these losses may be reduced by providing a secondary axle disconnect function that
disconnects one of the axles when driving conditions do not call for torque to be delivered to
both axles.
The terms AWD and 4WD are often used interchangeably. The term AWD has come to be
associated with light-duty passenger vehicles that provide variable operation of one or both axles
on ordinary roads. The term 4WD is often associated with larger truck-based vehicle platforms
that provide for a locked driveline configuration and/or a low range gearing meant primarily for
off-road use.
Many 4WD vehicles provide for a single-axle (or two-wheel) drive mode that may be
manually selected by the user. In this mode, a primary axle (perhaps the rear) will be powered,
while the other axle (known as the secondary axle) is not. Even though the secondary axle is not
contributing torque, energy may still be consumed by rotation of its driveline components
because they are still connected to the non-driven wheels. This energy loss directly results in
increased fuel consumption and CO2 emissions that could be avoided by disconnecting the
secondary axle components under these conditions.
Further, many light-duty AWD systems are designed to variably divide torque between the
front and rear axles in normal driving, in order to optimize traction and handling in response to
driving conditions. Even when the secondary axle is not delivering torque, it typically remains
2-182
-------
Technology Cost, Effectiveness, and Lead Time Assessment
engaged with the driveline and continues to generate losses that could be avoided by a more
advanced disconnect feature. For example, Chrysler has estimated that the secondary axle
disconnect in the Jeep Cherokee reduces friction and drag attributable to parasitics of the
secondary axle by 80 percent when in disconnect mode.487 Some of the sources of secondary axle
parasitics include lubricant churning, seal friction, bearing friction, and gear train losses.488'489
Many part-time 4WD systems, such as those seen in light trucks, use some type of secondary
axle disconnect to provide shift-on-the-fly capabilities. In many of these vehicles, particularly
light trucks, the rear axle is permanently driven and the front axle is secondary. The secondary
axle disconnect is therefore part of the front differential assembly in these vehicles. Light-duty
passenger cars that employ AWD may instead permanently power the front wheels while making
the rear axle secondary, as currently in production in the Jeep Cherokee 4WD system.
As part of a shift-on-the-fly 4WD system, the secondary axle disconnect serves two basic
purposes. First, in two-wheel drive mode, it disengages the secondary axle from the driveline so
the wheels do not turn the secondary driveline at road speed, reducing wear and parasitic energy
losses. Second, when shifting from two- to four-wheel drive "on the fly" (while moving), the
secondary axle disconnect couples the secondary axle to its differential side gear only after the
synchronizing mechanism of the transfer case has spun the secondary driveshaft up to the same
speed as the primary driveshaft.
4WD systems that have a disconnect typically do not have either manual- or automatic-
locking hubs. To isolate the secondary wheels from the rest of the secondary driveline, axle
disconnects use a sliding sleeve to connect or disconnect an axle shaft from the differential side
gear.
2.2.8.3.2 Developments in A WD Technology
Since the FRM, EPA has continued to monitor developments in AWD secondary axle
disconnects and their adoption in the light-duty vehicle fleet.
As discussed in the Draft TAR, EPA coordinated with Transport Canada and Environment
and Climate Change Canada on a project to characterize AWD systems present in the market
today. The primary objectives of this project were to gain an overview of AWD technology in
general and to understand the potential effect of advances in these systems on GHG performance
in comparison to their 2WD variants. A comprehensive technical characterization of 17 in-
production AWD systems has been completed.489 It includes characterization of system
architecture, operating modes, and current usage in the fleet. It also estimated and compared the
mass and rotational inertia of AWD components and parts to those of 2WD variants in order to
better understand the weight increase associated with AWD. Additionally, the all-wheel-drive
components of three AWD vehicles (the 2015 Jeep Cherokee Limited 4x4, 2015 Ford Fusion
AWD, and 2015 Volkswagen Tiguan Trendline 4motion) underwent a teardown in order to
accurately characterize their mass and rotational inertia and estimate their approximate cost. One
of the teardown vehicles, the Jeep Cherokee, includes a secondary axle disconnect, indicating
that this technology has begun to appear in light-duty vehicles since the FRM. In 2014, Chrysler
Group LLC presented a very positive outlook on the advantages of this system for improving
fuel efficiency while retaining a highly competitive off-road capability.490 This suggests that the
addition of secondary axle disconnect systems need not be accompanied by loss of traction and
handling capability.
2-183
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The study reinforced the perception that AWD is rapidly increasing in popularity in the
vehicle fleet, with about one-third of all vehicles sold in North America in 2015 having AWD
capability. The prevalence of AWD varies significantly between vehicle segments and trim
levels. Sedans have the lowest AWD availability, while AWD versions outnumber 2WD
versions in the SUV and pickup segments, particularly among the higher trim levels in each
segment.
The study identified several areas of potential efficiency improvement for AWD systems.
These included system level improvements such as: use of a single shaft Power Transfer Unit
(PTU), which can save up to 10kg in mass compared to a two-shaft unit; careful integration into
vehicle architecture; downsizing the driveline to further reduce mass while providing sufficient
traction in adverse conditions; and use of electric rear axle drive (eRAD). Component level
improvements were also identified, including: use of fuel-efficient bearings, low drag seals,
improved lubrication strategies, use of high-efficiency lubricants, advanced CV joints, and dry
clutch systems. Design improvements such as hypoid offset optimization, bearing preload
optimization, use of single-shaft power transfer units (PTUs) and an optimized propshaft gear
ratio were also suggested to have potential. Use of weight-reducing metals such as magnesium,
and manufacturing improvements such as vacuum die casting and improved hypoid
manufacturing were also cited as opportunities. The authors'judgement of the relative potential
for AWD efficiency improvements offered by each opportunity are depicted in Figure 2.63.
cost / deterioration
savings / Improvement*
Cost
rsi
>100 10-100
< ID
<10 10-100 >100
Weight
m
>2.5 .5-2.5
<.5
<.5 .5-2.5 >2.5
Fuel Consumption
>2 .5-2
<.S
<.S .5-2 >2
Performance
1*1
>10 1- 1Q
<1
<1 1-10 >10
Packaging
duffiojlt
easy i mprcwed
,
¦u
V
s
no
u
J: £
E
Sb
is
o
¦t
*1
c
5
CL
ra
&
&
System Level
Disconnect system FWD
m
Disconnect system ftWD
downsizing
m
eNAD i]Hybrid|
Component Level
FE bearings
Low drag seals
Actuator technology
Lubrication strategies
m
Advanced CV-joints
m
Dry dutch systems
Figure 2.63 Summary of AWD Efficiency Improvement Potentials489
Various sources cited in the study suggested that AWD disconnect systems have the ability to
lower fuel consumption of AWD vehicles by between 2 percent and 7 percent, significantly
higher than the estimates of 1.2 percent to 1.4 percent used in the 2012 FRM. However, it should
be noted that a disconnect strategy must balance fuel efficiency with other concerns such as
2-184
-------
Technology Cost, Effectiveness, and Lead Time Assessment
vehicle dynamics, traction and safety requirements, which may act to reduce its actual GHG
effectiveness.
The study also identified three primary technological trends taking place in AWD system
design, including: actively controlled multi-plate clutches (MPCs), active disconnect systems
(ADS), and electric rear axle drives (eRAD). While controlled MPCs appear to be the dominant
technology in on-demand systems, ADS is a more recent trend and holds promise for reducing
real world fuel consumption. eRAD is the most recent emerging technology with potential for
even greater improvements (as seen in the Volvo XC90 Hybrid SUV).
The teardown analysis analyzed three power transfer units (PTUs) and rear drive modules
(RDMs) from the Ford Fusion, Jeep Cherokee and VW Tiguan. These were non-destructively
disassembled and analyzed with respect to mass, rotational inertia and the presence of specific
design features. Figure 2.64 shows the contribution of individual AWD driveline components to
the total additional mass of the AWD variant of each vehicle compared to the 2WD variant.
Further analysis of rotational inertias of these parts suggested that rotational inertias add very
little equivalent mass and therefore probably do not carry a large impact on fuel consumption.
1400
[ksl
100.0
80.0
60.0
40.0
20,0
0.0
135
13.1
78
0.0
11.4
l Other
Prop shaft
i Halfshafts
I RDM
IPTU
Fusion
Cherokee
Tiguan
Figure 2.64 Contribution of Individual AWD Driveline Components to Total Additional Vehicle Mass
The study included a high-level cost analysis for these parts, including the mechanical
disconnect device and modifications necessary to the torque transfer device (TTD). The total cost
of adding secondary axle disconnect to a vehicle was estimated at approximately $90 to $100.
Although this cost estimate was informally derived based primarily on the experience and
expertise of the authors, it compares well to the total cost (TC) figure attributed to 2017 in the
FRM analysis, at $98. The authors noted that the cost for the Jeep Cherokee system would likely
be higher because this system was designed to accommodate a planetary low gear, which adds
mass and cost not related to the AWD disconnect function.
2-185
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In addition to the in-production disconnect concepts described in the Transport Canada AWD
report, activity continues in the development of innovative secondary axle disconnect concepts.
For example, in 2015, Schaeffler presented a novel design for a clutch mechanism for use in
AWD disconnect.491 Suppliers are also designing and marketing modular solutions for
integration into existing OEM products.488 Developments such as these suggest that multiple
potential paths will exist for disconnect technology to accompany the increasing growth and
popularity of AWD in light-duty vehicles.
In conjunction with the AWD characterization project described above, Transport Canada is
also conducting a program of coast down testing, chassis dynamometer testing, and on-road
testing of several Canada-specification AWD vehicles at Transport Canada facilities. This
portion of the effort was not yet completed at the time of this Proposed Determination.
For the cost and effectiveness assumptions EPA adopted for the Draft TAR analysis, which
are retained for the Proposed Determination analysis, see Section 2.3.
2.2.8.4 Low-Drag Brakes: State of Technology
Low or zero drag brakes reduce or eliminate the sliding friction of disc brake pads on rotors
when the brakes are not engaged. By allowing the brake pads to pull or be pushed away from the
rotating disc either by mechanical or electric methods, the drag on the vehicle is reduced or
eliminated.
The reduction of brake drag is a technology that vehicle manufacturers have focused on for
many years. The ability to allow the brake disc pads to move away from the rotor and thereby
reduce friction is a known technology. This has been historically implemented by designing a
caliper and rotor system that allows the piston in the caliper to retract. However, if the pads are
allowed to move too far away from the rotor, the first pedal apply made by the vehicle operator
can feel spongy and have excessive travel. This can lead to customer dissatisfaction regarding
braking performance and pedal feel. For this reason, in conventional hydraulic-only brake
systems, manufacturers are limited by how much they can allow the pads to move away from the
rotor.
Recent developments in braking systems have allowed suppliers to provide brakes that have
the potential for zero drag. In this system the pad is allowed to move away from the rotor in
much the same way that is done in today's conventional brake systems, but in a zero drag brake
system the pedal feel is separated from the hydraulics by a pedal simulator. The pedal simulator
provides a portion of the overall braking feel specifically that of the tactile feel provided to the
vehicle operator. The other portion of brake feel is determined by the actual deceleration felt by
the vehicle operator. In a properly designed brake system the tactile pedal feel and the
associated vehicle deceleration is linear, consistent and predictable over all vehicle operating
conditions. This application of a pedal simulator is very similar to the brake systems that have
been designed for hybrid and electric vehicles. In hybrid and electric vehicles, some of the
primary braking is done through the recuperation of kinetic energy in the drive system.
However, the pedal feel and the deceleration that the operator experiences is tuned to provide a
braking experience that is equivalent to that of a conventional hydraulic brake system. These
"brake-by-wire" systems have highly tuned pedal simulators that feel like typical hydraulic
brakes and seamlessly transition to a conventional system as required by conditions. In addition
to the pedal simulator, the conventional vacuum-assisted master cylinder in a brake-by-wire
2-186
-------
Technology Cost, Effectiveness, and Lead Time Assessment
system is replaced by a replaced by an electric pump that is able to build brake pressure as
indicated by the position of the brake pedal. Because the electric pump is able to build brake
pressure faster than most vehicle operators, operators do not experience any deterioration in
stopping performance, even under conditions where the brake pads have moved slightly away
from the brake rotors. The application of a pedal simulator and brake-by-wire system is new to
non-electrified vehicle applications. If the pedal simulator and electric pump are tuned properly,
the initial pedal depression, even with the pads moved slightly away from the rotor, can provide
the same pedal feel and vehicle deceleration characteristics associated with a conventional brake
system.
In addition, to reducing brake drag, the zero drag brake system may also provide ancillary
benefits. It could allow for a faster brake apply and greater deceleration than is normally applied
by the average vehicle operator. It may also allow manufacturers to tune the braking for
different customer preferences within the same vehicle. This means that a manufacturer can
provide a "sport" mode which provides greater deceleration with less pedal displacement and a
"normal" mode which might be more appropriate for day-to-day driving. These electrically
driven systems may also facilitate other brake features such as panic brake assist, automatic
braking for crash avoidance and could support future autonomous driving features.
The zero drag brake system that are electrically driven also eliminates the need for a brake
booster. This has the potential to save both cost and weight in the overall system. Elimination of
the conventional vacuum brake booster could also improve the effectiveness of stop-start
systems. Typical stop-start systems need to restart the engine if the brake pedal is cycled
because the action drains the booster of stored vacuum. Because the zero drag brake system
provides braking assistance electrically, there is no need to supplement lost vacuum during an
engine off event.
Finally, many of the engine technologies being considered to improve efficiency reduce
pumping losses through reduced throttle. The reduction in throttle could result in supplemental
vacuum being required to operate a conventional brake system. This is the situation in many
diesel-powered vehicles. Diesel engines run without a throttling and often require supplemental
vacuum for brake boosting. By using a zero drag brake system, manufacturers may realize the
elimination of brake drag as well as the ancillary benefits described above and avoid the need for
a supplemental vacuum pump.
For the specific cost and effectiveness assumptions EPA is adopting for the Proposed
Determination assessment, see Chapter 2.3.
2.2.9 Air Conditioning Efficiency and Leakage Credits
Air conditioning (A/C) is a virtually standard automotive accessory, with over 95 percent of
new cars and light trucks sold in the United States being equipped with mobile air conditioning
(MAC) systems. This high penetration means that A/C systems have the potential to exert a
significant influence on the energy consumed by the light duty vehicle fleet, as well as GHG
emissions resulting from refrigerant leakage.
The 2012 final rule allowed vehicle manufacturers to generate credits for improved A/C
systems toward complying with the CO2 and fuel consumption fleet-wide average standards. In
the EPA program, manufacturers can generate credits for improved performance of both direct
2-187
-------
Technology Cost, Effectiveness, and Lead Time Assessment
emissions (refrigerant leakage) and indirect emissions (tailpipe emissions attributable to the
energy consumed by A/C). In both cases, a selection of "menu" credits in grams per mile are
available for qualifying technologies, with the magnitude of each credit being estimated based on
the expected reduction in CO2 emissions resulting from the technology. See 40 CFR 86.1868-12.
In the NHTSA program, manufacturers are allowed to generate fuel consumption improvement
values for purposes of CAFE compliance based on the use of A/C efficiency-improving
technologies. However, manufacturers cannot count reductions in A/C leakage toward their
CAFE calculations since these improvements do not affect fuel economy.
Since the FRM, many manufacturers have generated and banked credits through this program
and continue to do so today. In the FRM, the agencies estimated that significant penetration of
A/C technologies would occur to gain these credits, and this was reflected in the stringency of
the standards. See e.g. 77 FR at 62805/3.
EPA projected that the 2017-2025 program would lead to significant reductions in GHGs
from reduced A/C refrigerant leakage and from industry adoption of lower global warming
potential (GWP) refrigerants. Based on additional information that became available for the
Draft TAR analysis, as well as changes in the overall regulatory environment affecting the A/C
technology developments in the light-duty vehicle industry, the Draft TAR reaffirmed our
conclusion that these technologies will continue to expand and play an increasing role in overall
vehicle GHG reductions and regulatory compliance. EPA continues to believe this is the case in
this Proposed Determination.
2.2.9.1 A/C Efficiency Credits
2.2.9.1.1 Manufacturer Utilization of A/C Efficiency Credits
The A/C credit program continues to be an important component of manufacturers'
compliance plans, with many manufacturers continuing to take advantage of the program to
generate and bank A/C efficiency credits. The importance of the program was reinforced by
many of the comments received on the Draft TAR, strongly reaffirming that OEMs continue to
consider A/C credits to be an essential component of their compliance paths. For example, the
Alliance of Automobile Manufacturers (AAM) commented, "MAC indirect credits are playing a
critical role in industry compliance with the light-duty vehicle GHG regulation, achieving
emission reductions that would not otherwise have been possible using the previous CAFE
regulatory framework."
As summarized in the EPA Manufacturer Performance Reports,492'493 1 7 auto manufacturers
included A/C efficiency and/or leakage credits as part of their compliance demonstration in both
the 2014 and 2015 model years. In MY2014, these included more than 10 million Megagrams
(Mg) of A/C efficiency credits, or about 25 percent of the total net A/C credits reported that year.
In MY2015, utilization of A/C efficiency credits increased to more than 12 million Mg, or 37
percent of the total net credits that year. This was equivalent to about 3 grams per mile across
both the 2014 and 2015 fleets. Including the 2012 and 2013 model years, A/C efficiency credits
have to date totaled over 36.3 million Mg.
The vast majority of A/C efficiency credits were claimed through the A/C credit menu (see 40
CFR 1868-12(a)), which includes several A/C efficiency-improving technologies that were well
defined and had been quantified for effectiveness at the time of the 2012 FRM. Some comments
2-188
-------
Technology Cost, Effectiveness, and Lead Time Assessment
on the Draft TAR praised the pre-defined, pre-approved credit menu approach as being highly
effective at incentivizing A/C improvements, and cited the A/C credit program as a good
example of how real-world GHG benefits can be recognized and credited.
As discussed in the Draft TAR, EPA expects that additional technologies for improving A/C
efficiency that are not represented in the menu may continue to emerge. Although not part of the
credit menu, these technologies will continue to be eligible for credit on a case-by-case basis
under the off-cycle credit program. An off-cycle credit application for this purpose should be
supported by results of testing under the AC 17 test protocol using an "A to B" comparison, that
is, a comparison of substantially similar vehicles in which one has the technology and the other
does not. See 40 CFR section 86.1869-12 (c) and (d).
To date, EPA has received one off-cycle credit application for an A/C efficiency technology.
In December 2014, General Motors submitted an off-cycle credit application for the Denso SAS
A/C compressor with variable crankcase suction valve technology,494 requesting an off-cycle
GHG credit of 1.1 grams CO2 per mile. EPA evaluated the application and found that the
methodologies described therein were sound and appropriate. Therefore, EPA approved the
credit application.495
AAM commented on the off-cycle approval process as an alternative route to A/C credits,
stating, "Automakers request that EPA simplify and standardize the procedures for claiming off-
cycle credits for the new MAC technologies that have been developed since the creation of the
MAC indirect credit menu." Other comments noted the importance of continuing to incentivize
further innovation in A/C efficiency technologies in the future as new technologies emerge that
are not in the credit menu, or when manufacturers begin to reach the regulatory caps on menu
credits, and suggested that EPA should consider adding new A/C efficiency technologies to the
credit menu and/or update the credit values, particularly those that qualify for credits through an
off-cycle application or through such an application are approved for more credit than provided
in the menu. For example, Toyota commented, "Toyota appreciates the continued incentives for
these emerging A/C efficiency technologies, but it remains unclear as to why the agencies have
chosen to not support further development of the existing A/C efficiency incentive menu.
Toyota's assessment is that the existing menu items are further improved as well, in which case
the incentive values for A/C efficiency should be updated along with including new technologies
being deployed."
Although these comments were made in the context of the A/C efficiency program, they
border on issues that are closely related to the topic of the off-cycle approval process in general.
The off-cycle provisions are described in more detail in Chapter 2.2.10, and comments received
on this topic are addressed more fully in the Section B.3.4.1 of the Proposed Determination
Appendix (Off-Cycle Technology Credits). With regard to the A/C menu specifically, although it
is anticipated that new A/C technologies that are not represented in the credit menu may emerge
over the time frame of the MY2022-2025 standards, EPA does not plan to add additional items
to the credit menu nor to change the values assigned to those that are currently in the menu. EPA
acknowledges that the menu of pre-defined and pre-approved technologies has been well
received as a way to incentivize A/C improvements. However, EPA continues to feel that
expanding the design-based aspect of the program that is represented by the credit menu, either
by adding new technologies or updating the credit values, would be inconsistent with the goal of
transitioning the program toward a performance basis, as represented by the phase-in of testing
2-189
-------
Technology Cost, Effectiveness, and Lead Time Assessment
requirements as established in the rule. EPA anticipates that the off-cycle program will continue
to serve as the primary mechanism for expanding A/C technology credit opportunities.
The 2012 final rule establishes that menu-based credits for A/C efficiency are subject to a
regulatory cap. The rule set a cap of 5.7 g/mi for cars and trucks through MY2016, and separate
caps of 5.0 g/mi for cars and 7.2g/mi for trucks for later MYs. See 40 CFR 86.1868-12(b)(2).
Several commenters asked EPA to reconsider the applicability of the cap to non-menu A/C
efficiency technologies claimed through the off-cycle process, and questioned the applicability of
this cap on several different grounds. These comments appear to be in response to a passage in
the Draft TAR which stated: "Applications for A/C efficiency credits made under the off-cycle
credit program rather than the A/C credit program will continue to be subject to the A/C
efficiency credit cap" (Draft TAR, p. 5-210). EPA has considered these comments and presents
clarification below.
As additional context, the 2012 TSD states (see p. 5-58, 2012 TSD): "...air conditioner
efficiency is an off-cycle technology. It is thus appropriate [...] to employ the standard off-cycle
credit approval process [to pursue a larger credit than the menu value]. Utilization of bench tests
in combination with dynamometer tests and simulations [...] would be an appropriate alternate
method of demonstrating and quantifying technology credits (up to the maximum level of credits
allowedfor A/C efficiency) [emphasis added], A manufacturer can choose this method even for
technologies that are not currently included in the menu." This suggests that the concept of
placing a limit on total A/C credits, even when some are granted under the off-cycle program, is
not entirely new, and that EPA considered the menu cap as being appropriate at the time.
Looking more specifically at the regulations, the regulatory caps specified under 40 CFR
86.1868-12(b)(2) apply to menu-based credits and are not part of the off-cycle regulation (40
CFR 86.1869-12). However, it should be noted that off-cycle credit applications are decided
individually on their merits through a process involving public notice and opportunity for
comment. The rationale relied upon for approving or denying credit requests may take into
account any factors deemed relevant, including such issues as the realization of claimed credits
in real world use. Such factors could include the consideration of synergies or interactions
among applied technologies, which could potentially be addressed by application of some form
of cap or other applicable limit, if warranted. Therefore, applying for A/C efficiency credits
through use of the off-cycle provisions under 86.1869-12 should not be seen as a route to
unlimited A/C credits.
Going forward, EPA expects to cap total A/C efficiency credits whether granted through
86.1868-12 or 86.1869-12. That is, through our authority in the off-cycle approval process, we
are likely to specify that total A/C efficiency credits be capped in an appropriate manner. At this
time EPA believes that, unless information pertinent to a specific application causes a different
conclusion, the caps specified in 86.1868-12 are appropriate for this purpose. Applicants can
present, as part of the analysis supporting their application, evidence supporting the case that a
different conclusion should apply to the application in question.
2.2.9.1.2 Eligibility for A/C Efficiency Credits
EPA has established two test procedures for use in determining eligibility for A/C efficiency
credits, the Idle Test and the AC 17 Test. The Idle Test procedure, which has now been phased
out, and the AC17 test procedure are described in more detail in Draft TAR Chapter 5.2.9.1.
2-190
-------
Technology Cost, Effectiveness, and Lead Time Assessment
For MYs 2014 to 2016, there were three options for qualifying for A/C efficiency credits: 1)
running the Idle Test, as described in the MYs 2012-2016 final rule, and demonstrating
compliance with the CO2 and fuel consumption threshold requirements; 2) running the Idle Test
and demonstrating compliance with engine displacement adjusted CO2 and fuel consumption
threshold requirements; and 3) running the AC 17 Test and reporting the test results.
In preparation for the 2017-2025 NPRM, the agencies recognized that the Idle Test had
limitations, and sought to develop a test procedure that could more reliably generate an
appropriate credit value based on an "A" to "B" comparison, that is, a comparison of
substantially similar vehicles in which the "A" vehicle is a baseline vehicle without the
technology, and the "B" vehicle has the technology. The result of this effort was the AC 17 Test
Procedure, which is based on a transient drive cycle, rather than just idle.
To develop the AC 17 test, EPA initiated a study that engaged automotive manufacturers,
USCAR, component suppliers, SAE, and CARB. This effort also explored the applicability and
appropriateness of a test method or procedure which combines the results of test-bench,
modeling/simulation, and chassis dynamometer testing into a quantitative metric for quantifying
A/C system (fuel) efficiency. The goal of this exercise was the development of a reliable,
accurate, and verifiable assessment and testing method while also minimizing a manufacturer's
testing burden. EPA believes that the AC 17 test procedure is more effective than the Idle Test at
accurately reflecting the impact that A/C use (and in particular, efficiency-improving
components and control strategies) has on tailpipe CO2 emissions and fuel consumption. For a
complete description of the AC17 Test, please refer to the 2017-2025 TSD or the Draft TAR.
The 2017-2025 rule thus provided for a phasing out of the Idle Test in favor of the AC17 Test.
For MYs 2017-2019, the AC17 test becomes the exclusive means to demonstrate eligibility for
A/C efficiency credits. By reporting test results, manufacturers gain access to the credits on the
menu based on the design of their AC system. Then, beginning in MY 2020, the AC 17 test will
be used not only to demonstrate eligibility for efficiency credits, but also to partially quantify the
amount of the credit. If the delta of the A-to-B test is greater than the value in the credit menu,
the manufacturer receives the menu value, otherwise the value is scaled.
However, an engineering assessment can still be conducted as an alternative to baseline ("A")
testing to build the case for a specific credit value if, for example, a baseline vehicle does not
exist on which to base the A-to-B comparison. See 76 FR 74938, 74940. This provision is found
in 86.1868-12(g), which describes the testing requirement applicable to MY2020 and later. In
part, the provision includes the following two requirements (paraphrased; see 86.1868-12(g) for
text):
(1) Performing the AC 17 test on a vehicle that incorporates the air conditioning system with
the credit-generating technologies (the "B" vehicle).
(2) And, either:
(a) Performing the AC 17 test on a vehicle which does not incorporate the credit-
generating technologies (the baseline or "A" vehicle), where the tested vehicle must be similar to
the vehicle tested under (1) and selected using good engineering judgment. The tested vehicle
may be from an earlier design generation; or,
2-191
-------
Technology Cost, Effectiveness, and Lead Time Assessment
(b) If the manufacturer cannot identify an appropriate vehicle to test under (a), they may
submit an engineering analysis that describes why an appropriate vehicle is not available or not
appropriate, and includes data and information supporting specific credit values, using good
engineering judgment.
Thus the regulation still requires that an AC 17 test be performed on the "B" vehicle that
contains the technology, but an appropriate engineering analysis may, if approved, provide for
credit in lieu of identification and testing of a baseline "A" vehicle.
2.2.9.1.3 The AC 17 Test Procedure
Throughout the development of the AC 17 credit program, EPA has worked closely with the
industry on a regular basis, through collaboration with USCAR, the Society of Automotive
Engineers (SAE), MAC suppliers, and other stakeholders. This effort was acknowledged in
comments on the Draft TAR, where the Alliance of Automobile Manufacturers (AAM) cited
"the close dialogue on these issues that EPA has maintained with the industry since the 2004-
2006 IMAC SAE Cooperative Research Program and the subsequent early stages of
development of the MAC indirect GHG credits."
Prior to the 2012 FRM, EPA collaborated with several OEMs to evaluate the AC 17 Test by
conducting independent testing on a variety of vehicles and air conditioning technologies. The
purpose of this effort was to gain insight regarding the appropriateness of the AC 17 Test for
verifying the reduction in CO2 emissions expected from A/C technologies on the efficiency
credit menu. Initially, six vehicles were tested, including three pairs of carlines with some
element of difference in their air conditioner systems. The results of these tests were discussed
in the 2012 TSD, Section 5.1.3.7, beginning on page 5-44. This collaborative effort continued to
include a variety of additional vehicles tested by several OEMs at AC17-capable test facilities.496
This preliminary testing showed that the AC 17 test is capable of low test-to-test variability, and
is suitable for evaluating the relative efficiency improvement of A/C technologies, when
confounding factors are minimized. In cases where comparison of the AC 17 results do not
directly demonstrate the effectiveness of a technology, the test results can still be useful within
an engineering analysis for justifying the test methodology to determine A/C CO2 credits.
EPA also initiated a round-robin test program between facilities of several USCAR members
in an effort to determine the repeatability of the AC 17 test among various test facilities and to
identify potential sources of variability. A 2011 Ford Explorer was selected for these tests. Four
test sites were utilized, located at Ford, GM, Chrysler, and an EPA-contracted facility at
Daimler. Each facility had a full environmental chamber capable of fulfilling all requirements of
the test. Four tests were run at each facility, after which the vehicle was returned to Ford for
confirmation. Each test measured CO2 emissions with A/C off and A/C on, to capture the
difference (delta) in CO2 emissions, which represents the GHG effect of A/C usage.
Figure 2.65 through Figure 2.67 compares the results of each test at each test site. Although
some variability was observed between test sites, consistency within a given site was good,
suggesting that the AC 17 test procedure is able to capture the difference in CO2 emissions
between A/C on and A/C off.
Several sources of variation were identified by analysis of these results. Variations in solar
load may have resulted from variations in sensor location and soak start time. Temperature
2-192
-------
Technology Cost, Effectiveness, and Lead Time Assessment
control was also a potential issue. Although most labs could maintain temperature within the
required tolerance of the test procedure, humidity was more difficult to maintain for the long
duration of the test. Overcorrecting may occur, but can be improved by optimizing sensor
location to better represent ambient conditions. The complexity and length of the test can lead to
an increased potential for voided tests, and may require more frequent calibration of the test cell
equipment. Although this test program was not fully described in materials accompanying the
FRM, many of the issues observed during this testing were addressed in the final form of the
rule.
CO2 Summary, Aye ON
500
450
£ 400
E7I
O
o
350 ¦
300
250
~ Bag 1. SC03 "Bay 2, HWrtT •Average
* *
+. # »453 ~ » + ^3
» . .374 * • • *378
1430
1 291
1 ZS8
t * txi
P 231
I 353
.276
452
'39$
¦286
3 £ S1 £ 3 2> &
& $ $ § % Q 3 3
*»¦ n_i fw •n
# $ $ 4
¥ ¥ t H
1111 £ 111 J/ jf J
P' P'
£ is J1 j?
U. U. Cl U.
% % ti f> f j j
?If ? iff
£ & 3 i? <3 <3 •
S? *S> *8? *3? K. V) l\.
$
-------
Technology Cost, Effectiveness, and Lead Time Assessment
500
450
M 400
o
o
350
3 DO ¦
250
COa Summary, A/C OFF
~ Bag 3. SCC3 «Bag4. HWFET •Average J
• * • 347
¦ ¦
1 279
' 402
a • • • M2
231
~ ~ » » J97
~ *
3S7
• •
'412
• »:344
* • I ' 325
1 2se
I 27«
. 262
£>¦*£?& iUJ}S » £ j? J § & nf ^ f f
jp jp $ # $ $ $ $ & $ $ 8 $ $ iff?
a £ 1 £ £ £ £ £ S £ £ £ £ £ S IS fS fl>
/ / / / / / / / # ¦ I*? Tr :&¦
/
B B
4' if if 4
iu Uj kj lu
im
Figure 2.66 Variability of AC17 Round Robin Testing on 2011 Ford Explorer, A/C Off
COs Summary, DELTA {A/C ON - A/C OFF)
70
SO
50
U) 40
d
O 30
20
10
0
> SC03 ¦ HWFET "Average
44
I"
f,3T
a 16
| 33
V 15
• 23
? 11
:
/
Figure 2.67 Variability of AC17 Round Robin Testing on 2011 Ford Explorer, Delta between A/C on and Off
Although these tests demonstrated that the AC 17 test was able to resolve the difference
between A/C on and A/C off, they did not address its ability to resolve smaller differences, such
as the effect of an individual technology in an A to B test. As the size of an effect diminishes,
the difficulty of resolving it against a much larger baseline value becomes more challenging.
With the baseline CO2 g/mi value for most vehicles being in the hundreds, and the effect of a
single A/C technology possibly in the low single digits, test-to-test variation must be very small
2-194
-------
Technology Cost, Effectiveness, and Lead Time Assessment
to reliably detect the effect. As the AC 17 A-to-B test becomes a requirement beginning in
MY2020, this issue is being examined closely by the industry and EPA.
Since the 2012 FRM, USCAR members have conducted an ongoing test program to assess the
ability of the AC 17 test to resolve the GHG impact of individual A/C efficiency technologies in
an A to B test, and thereby function in the role assigned to it in the FRM as a means for
quantifying and qualifying for A/C credits. EPA has followed this effort by direct coordination
with member OEMs and by participating in meetings of the SAE Interior Climate Control
Committee.
As discussed in the Draft TAR, preliminary results of this test program have been
encouraging, while providing a robust context for previously identified issues to continue to be
assessed. These issues have included:
a) The potential difficulty of obtaining or constructing old-technology vehicles, particularly
those from earlier model years, on which to base A-to-B comparisons.
b) Factors such as test-to-test variability and the small magnitude of the effect being
measured, which may result in the need for multiple tests to be conducted to yield a
statistically reliable result, which would constitute a larger test burden than a single test.
c) Identification of acceptable test procedures and practices for performing bench testing
and engineering analysis (as an alternative to performing AC 17 testing on a potentially
unavailable baseline vehicle).
Members have expressed greater confidence in the ability to conduct AC17-based A-to-B
comparisons of software-related technologies (for example, default to recirculated air) than for
hardware-based technologies (for example, compressor design changes) because the former can
be implemented by relatively simple changes to software in order to represent a baseline "A"
vehicle without the technology. A-to-B comparisons of hardware technologies would be more
difficult because producing an "A" vehicle without the technology may prove difficult
particularly when confounding factors or technologies, or changes in hardware configuration, are
present.
In January 2016, EPA received additional comment and analysis from several USCAR
members regarding their most recent experience with AC 17 testing. In this interaction, many of
the issues discussed above were further outlined. Manufacturers have continued to experience a
significant number of voided tests and are continuing to work to identify the sources of such
events, which are commonly associated with long tests that demand careful environmental
control. Test-to-test variation is sometimes seen to exceed the magnitude of the credit value that
is the subject of the test. Although averaging of the results of multiple tests has shown some
success at establishing a reliable outcome, concerns were expressed about the resulting test
burden, due to the length of each test, the control requirements, and the limited availability of the
required specialized test cells. The availability of base vehicles without the technology being
assessed in an A-to-B comparison was also echoed as a concern. Manufacturers suggested that
the use of prior year models may be infeasible when several intervening model years are
involved, due to the confounding effect of other technologies introduced to the vehicle during
that time. This was expressed as being particularly true for the problem of assessing hardware-
based technologies, which may require building of prototype installations that may require
additional engineering resources to develop. Within individual test efforts, consistency of results
was good in some tests but exhibited inconsistencies in others, of which the manufacturers had
2-195
-------
Technology Cost, Effectiveness, and Lead Time Assessment
not yet achieved a full understanding but continue to study. Issues such as the complexity of
modern climate control systems and the presence of confounding factors such as powertrain
differences were cited as possible factors.
An application for off-cycle credits submitted by General Motors in December 2014494
provides an additional source of information on the results of AC 17 A-to-B testing, which was
used to support the application. GM cited several issues relating to the use of the AC 17 test
procedure to identify the CO2 benefit claimed in the application:
a) GM pointed out that the AC 17 A-to-B test was enabled by coincidental availability of a
valid baseline compressor (a variable compressor without the variable crankcase suction valve
technology) in the Holden Commodore and that this compressor coincidentally could be easily
bolted into the Cadillac ATS. GM reiterated that this is an uncommon situation and not
representative of future expectations.
b) GM stated that this hardware obstacle "prevents ready testing of the benefits of the SAS
compressor on other GM models on which it has been implemented."
c) There were some difficulties with torque and pressure measurement which was cited as
example of "control issues that may be expected to arise when attempting to do this type of
baseline technology testing for hardware on a vehicle that was never actually designed and
optimized to use that hardware."
Despite these difficulties, GM found that the AC17 test procedure was able to resolve a 1.3
g/mile CO2 improvement, which was in good agreement with the 1.1 g/mile suggested by bench
testing. However, because test-to-test variability was greater for the AC 17 tests than for the
bench tests, GM chose to request the 1.1 g/mile shown by the bench tests, which GM regarded as
more precise.
As previously described, the final rule provides for pursuing an engineering analysis in place
of locating and testing a valid baseline "A" vehicle. EPA has encouraged, and continues to
encourage, the use of bench test results and engineering analysis to support applications for AJC
efficiency credits in such situations.
Some comments on the Draft TAR expressed uncertainty about the AC 17 Test. For example,
FCA commented, "A/C efficiency technologies are not showing their full effect on this AC17
test as most technologies provide benefit at different temperatures and humidity conditions in
comparison to a standard test conditions. All of these technologies are effective at different
levels at different conditions. So there is not one size fits all in this very complex testing
approach. Selecting one test that captures benefits of all of these conditions has not been
possible."
EPA acknowledges that any single test procedure is unlikely to equally capture the real world
effect of every potential technology in every potential use case. This difficulty is well understood
among designers of test procedures, and was understood when the AC 17 test procedure was
developed. While no test is perfect, the AC 17 test procedure represents an industry best effort at
identifying a test that would greatly improve upon the Idle Test by capturing a much larger range
of operating conditions where different technologies are likely to show greater improvement than
on the Idle Test. It is our assessment that industry evaluation of the procedure has shown that it
achieves this objective.
2-196
-------
Technology Cost, Effectiveness, and Lead Time Assessment
FCA also commented, "It is a major problem to find a baseline vehicle that is identical to the
new vehicle but without the new A/C technology. This alone makes the test unworkable." EPA
disagrees that this makes the test unworkable. The regulation describes the baseline vehicle as a
"similar" vehicle, selected with good engineering judgment (such that the test comparison is not
unduly affected by other differences). Also, as discussed elsewhere, OEMs have expressed
confidence in using A-to-B testing to qualify for credits for software-based A/C efficiency
technologies. While hardware technologies may pose a greater challenge in locating a
sufficiently similar "A" baseline vehicle, the engineering analysis provision under 40 CFR
86.1868-12(g)(2) provides an alternative to locating and performing an AC17 test on such a
vehicle. Further, as the USCAR program in general and the GM Denso SAS compressor
application specifically have shown, the test is able to resolve small differences in CO2
effectiveness (1.3 grams in the latter case) when carefully conducted.
Commenters on the Draft TAR also expressed a desire for improvements in the process by
which manufacturers without an "A" vehicle could apply under the engineering analysis
provision, such as development of standardized engineering analysis and bench testing
procedures that could support such applications. For example, Toyota commented, "Toyota
requests EPA consider an optional method for validation via an engineering analysis, as is
currently being developed by industry." EPA is in fact coordinating with industry on this effort,
as described below. Similarly, the Alliance commented, "The future success of the MAC credit
program in generating emissions reductions will depend to a large extent on the manner in which
it is administered by EPA, especially with respect to making the AC 17 A-to-B provisions
function smoothly, without becoming a prohibitive obstacle to fully achieving the MAC indirect
credits." EPA also has an interest in seeing that the A/C credit program operates as it was
designed, and believes that dialogue between EPA and industry stakeholders in the A/C credit
program has been in the past, and will continue to be, an effective means toward this goal.
As described in the Draft TAR, in 2016, USCAR members initiated a Cooperative Research
Program (CRP) through the Society of Automotive Engineers (SAE) to develop bench testing
standards for the four hardware technologies in the credit menu (blower motor control, internal
heat exchanger, improved evaporators and condensers, and oil separator). Continuing progress in
this effort since the Draft TAR suggests that the availability of these standards may soon resolve
much of the uncertainty expressed by the commenters.
The specific standards under development are listed in Table 2.24. The intent of the program
is to streamline the process of conducting bench testing and engineering analysis in support of an
application for A/C credits under 86.1868-12(g)(2), by creating uniform standards for bench
testing and for establishing the expected GHG impact of the technology in a vehicle application.
EPA has regularly monitored the development of these standards by coordinating with the CRP
as well as participating in the applicable SAE standards development committees. Since
completion of the Draft TAR, work has continued on these standards, which appear to be nearing
completion.
Table 2.24 Hardware Bench Testing Standards under Development by SAE Cooperative Research Program
Number
Title
Status
2-197
-------
Technology Cost, Effectiveness, and Lead Time Assessment
J2765
Procedure for Measuring System COP of a Mobile Air Conditioning
System on a Test Bench
Published
J3094
Internal Heat Exchanger (IHX) Measurement Standard
Work in Progress
J3109
HVAC PWM Blower Controller Efficiency Measurement
Work in Progress
J3112
A/C Compressor Oil Separator Effectiveness Test Standard
Work in Progress
Commenters also suggested that other aspects of the credit application process should be
streamlined. These comments included suggestions such as: (a) that EPA should consider joint
applications by OEMs for the same A/C efficiency technology (currently, each OEM has to
apply separately); and (b) that EPA should consider allowing suppliers to directly petition for
credits and allow the approved credits to be applicable to OEMs that later adopt the technology
(currently, suppliers cannot apply independently of OEMs).
In general, the credit application process was designed to evaluate specific implementations of
A/C technologies in the context of a specific vehicle or platform. EPA believes that system
integration is a major factor in the ability of an identified technology to actually realize real-
world fuel-saving and GHG-reducing improvements as part of a mobile A/C system.
It would likely be very challenging for a supplier, for example, to be able to demonstrate
(through a hypothetical supplier-sponsored credit application) that a given A/C technology, as
represented perhaps by a stock part number, would necessarily always result in the same or
similar level of GHG effectiveness regardless of the vehicle on which it is installed. Even for
similar classes or sizes of vehicles, it seems likely that specifics of other parts of the system, such
as ductwork design, control strategy, and so on would vary significantly among different
manufacturers, and the effect of these differences would somehow have to be shown to be
inconsequential. Considerations such as these have effectively limited credit applications to
OEMs that are proposing a specific vehicle context for application of the technology. At this
time, it is likely that an independent supplier application would be seen as incomplete without
specific proposed OEM applications of the technology and OEM participation.
Similarly, while the rule does not appear to specifically prohibit multiple OEMs from
applying jointly for A/C credits, in order to evaluate such an application if it were presented, the
usage of the technology across the participating OEMs would somehow have to be sufficiently
similar in each proposed vehicle application to allow the application to be effectively evaluated.
EPA experience with evaluating such situations has seen significant variation across vehicle
models that integrate the same technologies. It therefore remains unclear whether joint
applications would be practical or desirable as a means to streamline the process. Therefore, EPA
has not established a process for joint OEM applications.
2.2.9.1.4 Summary
EPA has evaluated and considered the results of AC 17 testing presented by stakeholders.
These data suggest that the AC17 Test is capable of measuring the difference in CO2 emissions
between A/C on and A/C off, and, when conducted with appropriate attention to detail, is also
capable of resolving differences in CO2 emissions resulting from the addition of A/C efficiency
technology. In some cases, test-to-test variability and the small magnitude of the effect to be
measured may call for averaging of multiple tests to identify the effect with statistical
significance. While the ability to perform full AC 17 "A-to-B" testing may in some cases be
2-198
-------
Technology Cost, Effectiveness, and Lead Time Assessment
challenged by the potential unavailability of a valid "A" baseline vehicle, the engineering
analysis provision (as described in 40 CFR 86.1868-12(g)(2)) provides an alternative path to
credits in these cases.
EPA believes that the bench testing standards being developed by the SAE CRP are an
important example of how continued collaboration and dialogue among stakeholders and EPA
can facilitate the earning of A/C credits through existing pathways. To this end, EPA is
considering the possibility of issuing a guidance letter outlining best practices for applying the
SAE standards to an engineering analysis supporting an application for credits as provided in
86.1868-12(g)(2)(ii).
EPA has considered the comments received on the A/C efficiency credit system and the AC 17
test procedure, and has also considered what has been learned through the USCAR program and
the SAE CRP effort. It is clear that the A/C credit system has been effective at incentivizing
technologies that provide real-world GHG-reducing benefits. As the program transitions, as
scheduled, to an increasingly performance-based format that includes a requirement for AC 17
testing, continued collaboration and dialogue between EPA and the industry has been an
effective path toward identifying and developing practical solutions to the issues described
above. EPA therefore believes that the existing structure of the A/C credit program will not
prevent manufacturers from continuing to qualify for and earn A/C efficiency credits sufficient
to provide the contribution to manufacturer compliance paths that manufacturers anticipate.
2.2.9.2 A/C Leakage Reduction and Alternative Refrigerant Substitution
2.2.9.2.1 Leakage
As we observed in the Draft TAR, manufacturers have developed a number of technologies
for reducing the leakage of refrigerant to the atmosphere. These include fittings, seals, heat
exchanger/compressor designs, and hoses. Vehicle manufacturers consider low-leak
technologies to be among the most cost-effective approaches to improving overall vehicle GHG
emission performance.
Table 2.25 shows two metrics of the continued industry-wide progress toward durable, low-
leak systems. One trend is the annual increase in the generation of leakage credits already
apparent in the early years of the program as manufacturers have taken advantage of leakage-
reduction incentives. More on this trend, as well as a breakdown of leakage credits by
manufacturer, are found in EPA's Manufacturer Performance Report for the 2015 Model Year.491
Specifically, 13 manufacturers reported A/C leakage credits in the 2015 model year, amounting
to more than 20.3 million Megagrams (Mg) of credits. This equates to GHG reductions of about
6 grams per mile across the 2015 vehicle fleet. The table also shows the trend toward more leak-
proof A/C systems in terms of refrigerant leakage scores across the industry, as indicated by the
average industry-wide A/C system leakage scores that the State of Minnesota requires
automakers to report (using the SAE J-2727 method)498
Table 2.25 Trends in Fleet-wide Mobile Air Conditioner Leakage Credits and Average Leakage Rates
2009
2010
2011
2012
2013
2014
2015
Credits: (Million Megagrams/Grams/mi)
6.2/*
8.3/*
8.9/*
11.1/4.0
13.2/4.2
16.6/5.1
20.3/5.8
MN SAE J-2727 Leakage Rate (g/yr)
15.1
14.7
14.6
14.5
13.9
13.0
12.1
* Fleet-wide leakage credits in terms of grams/mi are not available prior to MY 2012 due to the optional nature of
the leakage credit program in the earlier years.
2-199
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.2.9.2.2 Low-GWP Refrigerants
In support of the LD GHG rules, EPA projected that the industry would fully transition to
lower-GWP refrigerants between Model Year (MY) 2017 and MY2021, beginning with 20
percent transition in MY2017, to be followed by a 20 percent increase in substitution in each
subsequent model year, completing the transition by MY2021 (77 FR 62779, 62778, 62805). Put
another way, the stringency of the MY2021 and later light duty GHG standards is predicated on
100 percent substitution of refrigerants with lower GWPs than HFC-134a. On July 20, 2015,
EPA published a final rule under the Significant New Alternatives Policy (SNAP) program that
changes the listing status of HFC-134a to unacceptable for use in A/C systems of newly-
manufactured LD motor vehicles beginning in MY2021, except where permitted for some export
vehicles through MY2025 (80 FR 42870).ww EPA's decision to take this action was based on
the availability of other substitutes that pose less overall risk to human health and the
environment, when used in accordance with required use conditions. Thus all new LD vehicles
sold in the United States will have transitioned to an alternative, lower-GWP refrigerant by
MY2021.
The July 20, 2015 SNAP final rule has no effect on how manufacturers may choose to
generate and use air conditioning leakage credits under the LD GHG standards. As stated in that
final rule," [njothing in this final rule changes the regulations establishing the availability of air
conditioning refrigerant credits under the GHG standards for MY2017-2025, found at 40 CFR
86.1865-12 and 1867-12. The stringency of the standards remains unchanged....
[Manufacturers may still generate and utilize credits for substitution of HFC-134a through the
2025 model year." EPA also there noted that the SNAP rule was not in conflict with the
Supplemental Notice of Intent (76 FR 48758, August 9, 2011) that described plans for EPA and
NHTSA's joint proposal for model years 2017-2025, since EPA's GHG program continues to
provide the level of air conditioning credits available to manufacturers as specified in that
Notice: "[T]he Supplemental Notice of Intent states that '(m)anufacturers will be able to earn
credits for improvements in air conditioning . . . systems, both for efficiency improvements . . .
and for leakage or alternative, lower-GWP refrigerants used (reduces [HFC] emissions).' 76 FR
at 48761. These credits remain available under the light-duty program at the level specified in
the Supplemental Notice of Intent, and using the same demonstration mechanisms set forth in
that Notice." 80 FR 42896-97.
EPA has listed three lower-GWP refrigerants as acceptable, subject to use conditions (listed at
40 CFR Part 82, Subpart G), for use in newly-manufactured LD vehicles: HFO-1234yf, HFC-
152a, and carbon dioxide (CO2 or R-744). Manufacturers are currently manufacturing LD
vehicles using HFO-1234yf, and they are actively developing LD vehicles using CO2499 and
considering the use of HFC-152a in a secondary loop A/C system.500
EPA expects that vehicle manufacturers will use HFO-1234yf for the vast majority of
vehicles. As discussed in the EPA Manufacturer Performance Report referenced above, the use
ww HFC-134a will remain listed as acceptable subject to narrowed use limits through MY2025 for use in newly
manufactured LD vehicles destined for export, where reasonable efforts have been made to ascertain that other
alternatives are not technically feasible because of lack of infrastructure for servicing with alternative refrigerants
in the destination country. (40 CFR Part 82, Subpart G, Appendix B.
2-200
-------
Technology Cost, Effectiveness, and Lead Time Assessment
of HFO-1234yf expanded considerably in recent years, from two manufacturers and 42,384
vehicles in the 2013 model year, to five manufacturers and 1,762,985 vehicles in the 2015 model
year, over 10 percent of 2015 model year vehicles are using this refrigerant. This trend
reinforces EPA's projection that the industry will have transitioned 20 percent of the fleet by
MY2017, as discussed above. Fiat Chrysler accounted for more than 95 percent of these
vehicles, introducing HFO-1234yf in over 75 percent of their models. Jaguar Land Rover
achieved the greatest penetration within their fleet, using HFO-1234yf in almost 90 percent of
Jaguar Land Rover vehicles produced in the 2015 model year.
Finally, regarding supply of alternative refrigerants, the July 2015 SNAP final rule stated that
EPA "considered the supply of the alternative refrigerants in determining when alternatives
would be available. At the time the light-duty GHG rule was promulgated, there was a concern
about the potential supply of HFO-1234yf. Some commenters indicated that supply is still a
concern, while others, including two producers of HFO-1234yf, commented that there will be
sufficient supply. Moreover, some automotive manufacturers are developing systems that can
safely use other substitutes, including CO2, for which there is not a supply concern for the
refrigerant. If some global light-duty motor vehicle manufacturers use CO2 or another
acceptable alternative, additional volumes of HFO-1234yf that would have been used by those
manufacturers will then become available. Based on all of the information before the agency,
EPA believes production plans for the refrigerants are in place to make available sufficient
supply no later than MY2021 to meet current and projected demand domestically as well as
abroad, including, but not limited to, the EU" (80 FR 42891; July 20, 2015). In their public
comments on the Draft TAR, Honeywell, a supplier of HFO-1234yf, said, "[w]e are in
agreement with EPA that by 2021 there will be sufficient capacity of HF0-1234yf around the
world to serve the global demand for this refrigerant.... Honeywell and its key suppliers are
investing approximately US$300 million to increase global production capacity for HFO-
1234yf."
2.2.9.2.3 Conclusions
As described in this section, there is strong evidence that auto manufacturers are continuing to
improve the leak-tightness of their AJC systems. In addition, many manufacturers are
transitioning to the use of low-GWP alternative refrigerants in a number of vehicle models. We
believe that the current trends among automakers toward the use of alternative refrigerants to
comply with the LD vehicle GHG standards, EPA's change in listing status of HFC-134a to
"unacceptable" by MY2021, and the parallel increase in the supply of the leading alternative
refrigerant ensure that our earlier projections that a complete transition to alternative refrigerants
by MY2021 will in fact become reality.
The MY2017-2025 LD GHG rule also encourages manufacturers to continue to use low-
leakage technologies even when using alternative refrigerants. Although some leakage may still
occasionally occur, the low GWPs of the new refrigerants, as compared to that of HFC-134a,
considerably reduce concerns about refrigerant leakage from a climate perspective.
2.2.10 Off-cycle Technology Credits
2.2.10.1 Off-cycle Credits Program
2.2.10.1.1 Off-cycle Credits Program Overview
2-201
-------
Technology Cost, Effectiveness, and Lead Time Assessment
EPA provides an opportunity for credits for off-cycle technologies. EPA initially included
off-cycle technology credits in the MY2012-2016 rule and revised the program in the MY2017-
2025 rule.501 "Off-cycle" emission reductions can be achieved by employing off-cycle
technologies that result in real-world benefits, but where that benefit is not adequately captured
on the test procedures used by manufacturers to demonstrate compliance with and fuel economy
emission standards.
The intent of the off-cycle provisions is to provide an incentive for CO2 reducing off-cycle
technologies that would otherwise not be developed because they do not offer a significant 2-
cycle benefit. EPA limited the eligibility to technologies whose benefits are not adequately
captured on the 2-cycle test. The preamble to the final rule provides a detailed discussion of
eligibility for off-cycle credits.502 Technologies that are integral or inherent to the basic vehicle
design including engine, transmission, mass reduction, passive aerodynamics, and base tires are
not eligible. Any technology that was included in the agencies' standard-setting analysis also
may not generate off-cycle credits (with the exception of active aerodynamics and engine stop-
start systems).503 EPA established this approach believing that the use of 2-cycle technologies
would be driven by the standards and no additional credits would be necessary or appropriate.
This approach also limits the program to off-cycle technologies that could be clearly identified as
add-on technologies more conducive to A-to-B testing that would be able to demonstrate the
benefits of the technology. Further limitations are placed on technologies that might otherwise be
incentivized through federal safety regulations.504
There are three pathways by which a manufacturer may generate off-cycle CO2 credits. The
first is a predetermined list of credit values for specific off-cycle technologies that may be used
beginning in MY20 1 4.505 This pathway allows manufacturers to use conservative credit values
established in the MY2017-2025 final rule for a wide range of technologies, with minimal data
submittal or testing requirements. In cases where additional laboratory testing can demonstrate
emission benefits, a second pathway allows manufacturers to use a broader array of emission
tests (known as "5-cycle" testing because the methodology uses five different testing procedures)
to demonstrate and justify off-cycle CO2 credits.506 The additional emission tests allow emission
benefits to be demonstrated over some elements of real-world driving not captured by the GHG
compliance tests, including high speeds, rapid accelerations, and cold temperatures. Credits
determined according to this methodology do not undergo additional public review. The third
and last pathway allows manufacturers to seek EPA approval to use an alternative methodology
for determining the off-cycle CO2 credits.507 This option is only available if the benefit of the
technology cannot be adequately demonstrated using the 5-cycle methodology. Manufacturers
may also use this option for model years prior to 2014 to demonstrate off-cycle CO2 reductions
for technologies that are on the predetermined list, or to demonstrate reductions that exceed those
available via use of the predetermined list. The manufacturer must also demonstrate that the off-
cycle technology is effective for the full useful life of the vehicle. Unless the manufacturer
demonstrates that the technology is not subject to in-use deterioration, the manufacturer must
account for the deterioration in their analysis.
The pre-defined list of technologies and associated car and light truck credits is shown in the
tables below.508 The regulations include a definition of each technology that the technology
must meet in order to be eligible for the menu credit.509 Manufacturers are not required to
submit any other emissions data or information beyond meeting the definition and useful life
requirements to use the pre-defined credit value. Credits based on the pre-defined list are subject
2-202
-------
Technology Cost, Effectiveness, and Lead Time Assessment
to an annual manufacturer fleet-wide cap of 10 g/mile. Due to expected synergistic effects of the
thermal technologies, the credits from the group of thermal control technologies are subject to a
per vehicle cap of 3.0 g/mi for cars and 4.3 g/mi for trucks.
Table 2.26 Off-cycle Menu Technologies and CO2 Credits for Cars and Light Trucks
Technology
Credit for Cars (g/mi)
Credit for Light Trucks (g/mi)
g/mi
g/mi
High Efficiency Exterior Lighting (at 100W)
1.0
1.0
Waste Heat Recovery (at 100W; scalable)
0.7
0.7
Solar Roof Panels (for 75 W, battery charging only)
3.3
3.3
Solar Roof Panels (for 75 W, active cabin ventilation
plus battery charging)
2.5
2.5
Active Aerodynamic Improvements (scalable)
0.6
1.0
Engine Idle Start-Stop w/ heater circulation system
2.5
4.4
Engine Idle Start-Stop without/ heater circulation
system
1.5
2.9
Active Transmission Warm-Up
1.5
3.2
Active Engine Warm-Up
1.5
3.2
Solar/Thermal Control
Up to 3.0
Up to 4.3
Table 2.27 Off-cycle Menu Technologies and CO2 Credits for Solar/Thermal Control Technologies for Cars
and Light Trucks
Thermal Control
Technology
Credit (g/mi)
Car
Truck
Glass or Glazing
Up to 2.9
Up to 3.9
Active Seat Ventilation
1.0
1.3
Solar Reflective Paint
0.4
0.5
Passive Cabin Ventilation
1.7
2.3
Active Cabin Ventilation
2.1
2.8
The two other pathways available to generate off-cycle credits require additional data. The 5-
cycle testing pathway requires 5-cycle testing with and without the off-cycle technology to
determine the off-cycle benefit of the technology. The final pathway, often referred to as the
public process includes a public comment period and is available for technologies that cannot be
demonstrated on the 5-cycle test. Manufacturers must develop a methodology for demonstrating
the benefit of the off-cycle technology and the methodology is made available for public
comment prior to an EPA determination whether or not to allow the use of the methodology to
generate credits. The data needed for this demonstration may be extensive, especially in cases
where the effectiveness of the technology is dependent on driver response or interaction with the
technology. As discussed below, all three methods have been used successfully by
manufacturers to generate off-cycle credits.
2.2.10.2 Use of Off-cycle Technologies to Date
Since the Draft TAR, EPA released the MY 2015 GHG Manufacturer Performance Report (or
"compliance report"). The MY 2015 compliance report shows that manufacturers are continuing
2-203
-------
Technology Cost, Effectiveness, and Lead Time Assessment
to introduce a wide array of off-cycle technologies to generate off-cycle GHG credits using the
pre-defined menu.510 For the fleet as a whole, off-cycle credits accounted for almost 3 g/mile of
credits in MY 2015 compared to 2.3 g/mile of credits in MY 2014. Table 2.28 below shows the
percent of each manufacturers' production volume using each of the menu technologies reported
to EPA for MY2015 by the manufacturer. Table 2.29 shows the g/mile benefit that each
manufacturer reported across its fleet from each off-cycle technology. Like the preceding table,
Table 2.29 provides the mix of technologies used in MY2015 across the manufacturers and the
extent to which each technology benefits each manufacturer's fleet.
Table 2.28 Percent of 2015 Model Year Vehicle Production Volume with Credits from the Menu, by
Manufacturer & Technology (%)
Manufacturer
Active
Aerodynamics
Thermal Control Technologies
Engine & Transmission
Warmup
Other
Grill shutters
Ride height
adjustment
Passive cabin
ventilation
Active cabin
ventilation
Active seat
ventilation
Glass or
glazing
Solar reflective
surface
coating
Active engine
warmup
Active
transmission
warmup
Engine idle
stop-start
High efficiency
exterior lights
Solar panel(s)
BMW
0.0
0.0
0.0
91.0
7.5
0.3
0.0
74.4
0.0
0.0
96.9
0.0
Fiat Chrysler
29.0
3.0
95.0
0.0
5.9
97.4
1.6
55.2
10.5
5.2
66.5
0.0
Ford
60.0
0.0
100.0
0.0
23.5
0.0
0.0
50.1
26.1
9.3
76.8
0.0
GM
9.7
0.0
0.0
0.0
16.5
99.5
38.6
14.4
0.0
8.8
40.0
0.0
Honda
0.0
0.0
0.0
0.0
1.0
0.0
0.0
0.0
72.2
1.5
57.8
0.0
Hyundai
4.9
0.0
0.0
0.0
18.3
89.0
0.0
0.0
52.6
2.7
22.3
0.0
Jaguar Land
Rover
0.0
0.0
0.0
0.0
50.7
99.0
0.0
0.0
0.0
97.6
100.0
0.0
Kia
2.2
0.0
0.0
0.0
19.5
99.7
0.0
0.0
16.5
1.9
52.7
0.0
Nissan
9.3
0.0
0.0
0.0
3.3
0.0
15.8
20.8
64.9
0.1
47.2
0.1
Subaru
32.9
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
3.1
0.0
Toyota
0.0
0.2
0.0
0.0
23.5
90.5
30.2
9.2
49.8
11.4
56.2
0.0
Fleet Total
14.6
0.4
23.5
2.3
12.2
51.9
13.2
20.7
28.2
5.8
49.1
0.0
Table 2.29 Off-Cycle Technology Credits from the Menu, by Manufacturer and Technology for MY 2015
(g/mi)
Manufacturer
Active
Aerodynamics
Thermal Control Technologies
Engine & Transmission
Warmup
Other
Grill shutters
Ride height
adjustment
Passive cabin
ventilation
Active cabin
ventilation
Active seat
ventilation
Glass or glazing
Solar reflective
surface coating
Active engine
warmup
Active
transmission
warmup
Engine idle
stop-start
High efficiency
exterior lights
Solar panel(s)
Total
BMW
_
_
_
2.1
0.1
0.0
_
1.5
_
_
0.6
_
4.2
Fiat Chrysler
0.2
0.0
1.9
_
0.0
1.6
0.0
1.6
0.4
0.2
0.2
_
6.1
Ford
0.7
_
2.0
_
0.3
_
_
1.2
0.7
0.4
0.3
_
5.6
GM
0.0
_
_
_
0.2
1.4
0.1
0.2
_
0.1
0.2
_
3.0
Honda
_
_
_
_
0.0
_
_
_
1.4
0.0
0.1
_
1.5
Hyundai
0.0
_
_
_
0.2
0.4
_
_
0.8
_
0.0
_
1.5
Jaguar Land
Rover
0.6
1.2
2.6
0.5
4.9
2-204
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Kia
0.0
_
_
_
0.2
0.6
_
_
0.2
0.0
0.1
_
1.2
Nissan
0.1
_
_
_
0.0
_
0.1
0.4
1.3
0.0
0.1
0.0
2.0
Subaru
0.1
_
_
_
_
_
_
_
_
_
0.0
_
0.2
Toyota
_
0.0
-
_
0.3
0.7
0.1
0.1
0.9
0.2
0.2
-
2.5
Fleet Total
0.1
0.0
0.5
0.1
0.1
0.6
0.1
0.5
0.6
0.1
0.2
0.0
2.8
0.0" indicates that the manufacturer did implement that technology, but that the overall penetration rate was not high enough to round to 0.1
grams/mile, whereas a dash indicates no use of a given technology by a manufacturer.
The credits shown above are based on the pre-defined credit list. Thus far, GM is the only
manufacturer to have been granted off-cycle credits based on 5-cycle testing. These credits are
for an off-cycle technology used on certain GM gasoline-electric hybrid vehicles. The
technology is an auxiliary electric pump, which keeps engine coolant circulating in cold weather
while the vehicle is stopped and the engine is off, thus allowing the engine stop-start system to
be active more frequently in cold weather.
The third pathway allows manufacturers to seek approval to use an alternative methodology
for determining the off-cycle technology CO2 credits. Several manufacturers have petitioned for
and been granted use of an alternative methodology for generating credits. In the fall of 2013,
Mercedes requested off-cycle credits for the following off-cycle technologies in use or planned
for implementation in the 2012-2016 model years: stop-start systems, high-efficiency lighting,
infrared glass glazing, and active seat ventilation. EPA approved methodologies for Mercedes to
determine these off-cycle credits in September of 2014.511 Subsequently, FCA, Ford, and GM
requested off-cycle credits under this pathway. FCA and Ford submitted applications for off-
cycle credits from high efficiency exterior lighting, solar reflective glass/glazing, solar reflective
paint, and active seat ventilation. Ford's application also demonstrated off-cycle benefits from
active aerodynamic improvements (grill shutters), active transmission warm-up, active engine
warm-up technologies, and engine idle stop-start. GM's application described the real-world
benefits of an air conditioning compressor with variable crankcase suction valve technology.
EPA approved the credits for FCA, Ford, and GM in September of 2015.512 FCA reported
2,599,923 Megagrams of off-cycle credits to EPA for the 2009-2013 model years. In the 2015
model year, GM reported earning 348,102 Mg of credits from the Denso AJC compressor.
More recently, EPA published a notice in the Federal Register on September 2, 2016,
requesting comments on methodologies for off-cycle credits submitted by BMW, Ford, GM, and
VW.513 The comment period closed on October 3, 2016, and EPA is currently evaluating
comments and drafting a decision document. If approved, these credits would appear in a future
edition of the compliance report to the extent that manufacturers claim them.
As discussed above, the vast majority of credits in MY2015 were generated using the pre-
defined menu. Even though the program has been in place for only a few model years, the level
of credits reported has already been significant for some manufacturers. FCA and Ford
generated the most off-cycle credits on a fleet-wide basis, reporting credits equivalent to about
6. lx g/mile and 5.6x g/mile, respectively.^ Several other manufacturers report fleet-wide
credits in the range of about 1 to 5 g/mile. The fleet total across all manufacturers was
equivalent to about 3 g/mile for MY2015. EPA expects that as manufacturers continue to
xx The credits are reported to EPA by manufacturers in Megagrams. EPA has estimated a g/mile equivalent.
2-205
-------
Technology Cost, Effectiveness, and Lead Time Assessment
expand their use of off-cycle technologies, the fleet-wide impacts will continue to grow with
some manufacturers potentially approaching the 10 g/mile fleet-wide cap applicable to credits
that are based on the pre-defined list.
Please see Proposed Determination document appendix section B.3.4.1 for further discussion
of off-cycle credits including comments received on the Draft TAR.
2.3 GHG Technology Assessment
2.3.1 Fundamental Assumptions
2.3.1.1 Technology Time Frame and Measurement Scale for Effectiveness and Cost
The effectiveness and cost associated with applying a technology will depend on the starting
technologies from which improvements are measured. For example, two vehicles that start with
different technologies will likely have different cost and effectiveness associated with adopting
the same combination of technologies. The importance of clearly specifying the point of
comparison for cost and effectiveness estimates was highlighted in the 2015 NAS committee's
finding "that understanding the base or null vehicle, the order of technology application, and the
interactions among technologies is critical for assessing the costs and effectiveness for meeting
the standards."
As long as the point of comparison is maintained consistently throughout the analysis for both
the baseline and future fleets, the decision of where to place an origin along the scale of cost and
effectiveness is inconsequential. For EPA's technology assessment, the origin is defined to
coincide with a "null technology package," which represents a technology floor such that all
technology packages considered in this assessment will have equal or greater effectiveness,
consistent with the approaches used in the 2012 FRM and Draft TAR. While other choices
would have been equally valid, this definition of a "null package" has the practical benefit of
avoiding technology packages with negative effectiveness values, while also allowing for a
direct comparison of effectiveness assumptions with the FRM and Draft TAR.
Effectiveness
"Null" Technology
Package
gC02/minL
t
Effi-null(%)
Jy Technology
Package 1
Eff2.1(%) = 1 - [1 - Eff2_null(%
Technology
Package 2
[1 - Eff^JK
g^/J^ ^ gCok/mi^
Cost
A Cost = CosVCost-,
: ^
decreasing emissions
"Null" Technology
Package j
Technology
Package 1
$0
increasing cost
PT~
Cost^
Technology
Package 2
Cos
2
Figure 2.68 The "Null Technology Package" and Measurement Scale for Cost and Effectiveness
When technologies can be specifically identified for individual vehicle models, it is possible
to estimate cost and effectiveness values specifically for those models. To the extent possible
with the available information, EPA has attempted to consider this. This is the case, for
2-206
-------
Technology Cost, Effectiveness, and Lead Time Assessment
example, with mass reduction and improvements in aerodynamics and tire rolling resistance,
where for this assessment EPA has uniquely characterized the various levels of those
technologies for individual models based on available road load data. For other technologies, the
information that is broadly available across the entire fleet is not detailed enough to distinguish
differences that arise to different implementations of the technologies.
The Global Automakers, Ford and other stakeholders commented on several topics with
regard to technology adoption that can be considered as universal comments. These comments
stated that EPA had not properly considered the amount of lead time required for technology
development and adoption, the impact of global vehicle manufacturing and its effect on
component availability, and platform sharing. With respect to lead time, EPA believes that
vehicle manufacturers do have adequate lead time to meet the 2022-2025 MY standards. The
technologies considered in the Proposed Determination are either currently in production or will
be commercially produced in the next several years. In addition, the standards that are being
reaffirmed in the Proposed Determination were set in 2012 calendar year, which provided
vehicle manufacturers 13 years of lead time. For every manufacturer this amount of lead time
represents multiple vehicle redesign cycles that provide opportunities for adopting mass
reduction, aerodynamic improvements, new powertrains, and lower rolling resistance tires. In
addition, this amount of lead time also has provided the opportunity for vehicle manufacturers to
consider and manage the effects of the standards on their global manufacturing and on platform
sharing. Finally, in addition to the GHG standards required by the United States, most countries
around the world are adopting standards that are more stringent. All of these standards in unison
are driving vehicle manufacturers to produce increasingly efficient vehicles for all world
markets.
2.3.1.2 Performance Assumptions
When determining cost and effectiveness values for specific technologies, it is important to
compare the technologies on a consistent basis, so that the relative cost-effectiveness of the
technologies can be fairly compared. The National Academy of Sciences states in their 2011
report: "Estimating the cost of decreasing fuel consumption requires one to carefully specify a
basis for comparison. The committee considers that to the extent possible, fuel consumption cost
comparisons should be made at equivalent acceleration performance and equivalent vehicle
size."514 This is because "objective comparisons of the cost-effectiveness of different
technologies for reducing [fuel consumption] can be made only when vehicle performance
remains equivalent."515 The National Academy of Sciences engaged the University of Michigan
for their 2015 report to perform a set full vehicle simulations. As a ground rule, "Each engine
configuration was modeled to maintain, as closely as possible, the torque curve of the baseline
naturally aspirated engine so that equal performance, as measured by 0-60 mph acceleration
time, would be maintained."516 The EPA agrees that it is appropriate to objectively compare
technology costs and effectiveness, that maintaining constant vehicle performance is the
appropriate way to achieve that goal, and that the NAS recommendation of "equivalent
acceleration performance" is appropriate. Thus, the costs and effectiveness presented in this
document are based on the application of technology packages while holding the underlying
acceleration performance constant.
In most cases, equivalent acceleration performance is achieved by "engine downsizing":
reducing the size (and thus the output power/torque) of the engine in advanced vehicle packages
2-207
-------
Technology Cost, Effectiveness, and Lead Time Assessment
until a series of performance metrics are maintained within a reasonable range of the target value
similar to the methodology used in the FRM and Draft TAR. A smaller engine will typically be
more efficient at the same speed and torque than a larger engine (as pumping losses are reduced),
so this methodology properly accounts for effectiveness that could be used for acceleration
performance as fuel consumption reduction, thus allowing an objective and fair comparison of
technologies. Our process maintains performance neutrality. As recommended by the NAS
(2011, 2015), EPA is working under the premise that technology cost assessments should be
made under the assumption of equivalent performance. As such, the ALPHA modeling runs
generate effectiveness values which maintain a set of acceleration metrics within a reasonable
window.
EPA recognizes that manufacturers have many vehicle attribute and manufacturing
constraints. Manufacturers will make many product planning decisions and the final products
will have engine displacement which represent the OEM's decision in its product plans. As a
modeling convenience, when calculating effectiveness, EPA assumes the appropriate component
sizing to maintain performance. Even if our model produces a greater variation in technology
packages than exists today (for example, by producing two levels of tire rolling resistance on a
vehicle platform compared to just one today), this does not require that manufacturers actually
produce a greater variety of component sizes than exist currently in order for our overall results
to be valid. In actual vehicle design, manufacturers will design discretely sized components, and
for each vehicle choose the available size closest to the optimal for the given load and
performance requirements. For example, in some cases, the chosen engine will be slightly
smaller than optimal (and thus lower fuel consumption), and in some cases the chosen engine
will be slightly larger than optimal (and thus higher fuel consumption). The same assumption is
applied to drivetrain, suspension, chassis components, etc. For example, brake rotors may be
sized in 15mm diameter increments, and manufacturers will apply the size that most closely
matches the performance and load requirements of that application. Just as the manufacturers are
doing today, EPA expects that they will average these product decisions across their entire fleet.
In our analysis, on average, the actual fleet of vehicles will use the appropriate component size,
and CO2 emissions and performance of the fleet will average out, with no significant net change
compared to the original analysis with unconstrained component sizes.
In gathering information on technology effectiveness, EPA relied on a wide variety of
sources. These sources provided information on the costs and effectiveness of various
technologies, but not all comparisons were done on a rigorously performance-neutral basis.
Thus, it was often necessary to recalculate the effectiveness of a particular technology when the
original comparison was done without the assumption of equivalent performance. For example,
the 2011 NAS report, in discussing continuously variable valve lift (CVVL)517 cites Energy and
Environmental Analysis, Inc.,518 which "estimates a 6.5 to 8.3 percent reduction in fuel
consumption at constant engine size and 8.1 to 10.1 percent with an engine downsize to maintain
constant performance."
When EPA modeled effectiveness of specific technologies of their combinations, it was
careful to maintain a minimum deviation of acceleration performance from the baseline vehicle.
As the NAS notes, "truly equal performance involves nearly equal values for a large number of
measures such as acceleration (e.g., 0-60 mph, 30-45 mph, 40-70 mph, etc.), launch (e.g., 0-30
mph), grade-ability (steepness of slopes that can be climbed without transmission downshifting),
maximum towing capability, and others."519 However, they furthermore state that "in the usage
2-208
-------
Technology Cost, Effectiveness, and Lead Time Assessment
herein, equal performance means 0-60 mph times within 5 percent. This measure was chosen
because it is generally available for all vehicles."
In vehicle simulation modeling in ALPHA performed since the FRM, EPA investigated using
additional performance criteria to define an overall performance metric. EPA chose four
acceleration performance metrics: 0-60 time, Vi mile time, 30-50 passing time, and 50-70 passing
time. These metrics were chosen to give a reasonably broad set of acceleration metrics that
would be sensitive enough to represent true acceleration performance, but not so sensitive that
minor changes in vehicle parameters would significantly change the final metric. For each
vehicle class, a baseline configuration was chosen, the vehicle package was run over the
performance cycle, and the times for each performance metric were extracted. These four metrics
were summed for the baseline vehicle. For each vehicle technology package based on the same
vehicle class, a nominal engine size was determined based on the estimated performance effect
of the technologies included in the package and a set of packages with a range of engine sizes
larger and smaller than the nominal engine size were simulated. The same performance cycle
was run and the sum of the four metrics compared to the baseline sum for each engine size
package. Results where the sum was not equal to or less than the baseline sum (more stringent
than the 5 percent band suggested by NAS) were rejected. The drive cycle CO2 emissions of the
target package were taken from the lowest emissions result of the remaining results.
For the Proposed Determination, EPA has continued to rely on the performance criteria from
the Draft TAR analysis within its analyses of technology effectiveness including Vi mile time, 0-
60 time, 30-50 passing time, and 50-70 passing time performance metrics. Comments were
received from AAM, FCA, and Ford, suggesting that top gear gradeability be added as a
performance criterion, in particular when applying advanced transmissions. EPA has considered
these comments, as noted in Section 2.3.4.2.2. (Effectiveness Values for TRX11 and TRX21),
and determined that for advanced transmissions, the performance criteria used in the Draft TAR
are sufficient for defining performance neutrality, even if some downshifting occurs under
limited high-load conditions.
For the purpose of specification and costing of plug-in vehicles (BEVs and PHEVs, or
collectively, PEVs), the Proposed Determination analysis maintains acceleration performance by
the same method as in the Draft TAR. EPA derived an empirical equation relating PEV power-
to-weight ratio to reported 0-60 acceleration time based on an informal study of MY2012-2017
BEVs and PHEVs. A target 0-60 time was selected for each PEV configuration comparable to
that of conventional vehicles, and the motor power assigned based on this equation. The PEV
motor sizing methodology is described in more detail in Chapter 2.2.4.4.6 (Relating Power to
Acceleration Performance). While performance for these vehicles was only maintained by means
of the 0-60 metric, it should be noted that the high low-speed torque of an electric motor is likely
to favor the 0-30 metric, thereby making 0-60 the more demanding metric of the two.
2.3.1.3 Fuels
Fuel specifications for the gasoline and diesel fuels used for demonstration of compliance
with light-duty vehicle GHG and CAFE standards are contained within the Title 40, Part 86 of
the U.S. Code of Federal Regulations. Tabulated values are reproduced here for reference
purposes in Table 2.30 and Table 2.31 for gasoline and diesel, respectively. Analyses of the
effectiveness of powertrain technologies over the regulatory drive cycles used fuel properties
conforming to these specifications.
2-209
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.30 Test Fuel Specifications for Gasoline without Ethanol (from 40 CFR §86.113-04)
Item
Regular
Reference Procedure1
Research octane, Minimum2
93
ASTM D2699; ASTM D2700
Octane sensitivity2
7.5
ASTM D2699; ASTM D2700
Distillation Range (°F):
Evaporated initial boiling point3
75-95
ASTM D86
10% evaporated
120-135
50% evaporated
200-230
90% evaporated
300-325
Evaporated final boiling point
415 Maximum
Hydrocarbon composition (vol %):
Olefins
10% Maximum
ASTM D1319
Aromatics
35% Maximum
Saturates
Remainder
Lead, g/gallon (g/liter), Maximum
0.050 (0.013)
ASTM D3237
Phosphorous, g/gallon (g/liter), Maximum
0.005 (0.0013)
ASTM D3231
Total sulfur, wt. %4
0.0015-0.008
ASTM D2622
Dry Vapor Pressure Equivalent (DVPE), psi (kPa)5
8.7-9.2 (60.0-63.4)
ASTM D5191
Table 2.31 Petroleum Diesel Test Fuel (from 40 CFR §86.113-94)
Property
Unit
Type 2-D
Reference
Procedure1
(i) Cetane Number
40-50
ASTM D613
(ii) Cetane Index
40-50
ASTM D976
(iii) Distillation range:
(A) IBP
340-400 (171.1-204.4)
(B) 10 pet. Point
400-460 (204.4-237.8)
(C) 50 pet. Point
°F (°C)
470-540 (243.3-282.2)
STM D86
(D) 90 pet. Point
560-630 (293.3-332.2)
(E) EP
610-690 (321.1-365.6)
(iv) Gravity
"API
32-37
ASTM D4052
(v) Total sulfur
ppm
7-15
ASTM D2622
(vi) Hydrocarbon composition: Aromatics,
minimum (Remainder shall be paraffins,
naphthenes, and olefins)
pet
27
ASTM D5186
(vii) Flashpoint, min
°F (°C)
130 (54.4)
ASTM D93
(viii) Viscosity
centistokes
2.0-3.2
ASTM D445
1 ASTM procedures are incorporated by reference in §86.1
EPA's estimate of effectiveness for gasoline-fueled engines and engine technologies was
based on Tier 2 Indolene fuel although protection for operation in-use on Tier 3 gasoline (87
AKIE10) was included in the analysis of engine technologies considered both within the Draft
TAR and Proposed Determination. Additionally, in the technology assessment for this Proposed
Determination, EPA has considered the required engine sizing and associated effectiveness
adjustments when performance neutrality is maintained on 87AKI gasoline typical of real-world
use. Consistent with its historical practice, when test fuel properties are updated, EPA will
determine appropriate test procedure adjustments in order maintain the same level of stringency
2-210
-------
Technology Cost, Effectiveness, and Lead Time Assessment
of the GHG standards when vehicles are tested using Tier 3 certification fuel. A correction factor
for application to future vehicles certified to the GHG standards using Tier 3 gasoline that will
allow correction of CO2 emissions in a manner that accounts for differences between Tier 2 and
Tier 3 certification fuels is currently under regulatory development with manufacturers, industry,
and other stakeholder involvement.
The Alliance of Automobile Manufacturers and several manufacturers commented that the
lower octane of Tier 3 fuel degrades efficiency at mid and high load conditions, specifically over
the US06 test cycle and similar high load conditions observed in real world conditions.
Arguably, any vehicle or engine can experience some degradation of efficiency under certain
operating conditions such as high temperature ambient conditions or sustained high loads when
climbing a grade or pulling a trailer. Higher octane fuel can reduce degradation in efficiency
under these operating conditions and some manufacturers have stated in their owner's manuals a
recommendation to use premium fuel under these conditions520. Compliance with the GHG
standards, however, is demonstrated over the FTP and HWFET cycles, which typically do not
involve knock-limited operation and thus do not result in significant changes in knock-limited
spark advance and therefore are unlikely to reflect conditions where octane may impact
emissions.
Furthermore, preliminary data from EPA chassis dynamometer testing of 10 MY2013 through
MY2016 light-duty passenger cars and pickup trucks with a variety of combustion systems (PFI,
naturally aspirated GDI, non-HEV GDI Atkinson, turbocharged/downsized GDI) shows a small,
incremental reduction in CO2 emissions of approximately 1 percent over the combined-cycle for
Tier 3 gasoline relative to Tier 2 gasoline for all of the vehicles tested. The reduction in CO2
emissions from Tier 3 gasoline is due in part to the reduced carbon content of Tier 3 gasoline
relative to Tier 2 gasoline. This is largely due to a reduction in aromatics for Tier 3 gasoline that
is reflective of nationwide trends in U.S. gasoline properties over the past four decades since
aromatic content was last revised for gasoline used for EPA certification and compliance testing.
We note further that under current guidelines established in guidance letter " 1997-01: New
Guidance on Testing Vehicles with Knock Sensors"521, manufacturers are required at
certification to provide confirmation that vehicles that are not labeled as 'premium fuel required'
do not see a change in emissions over all test cycles, including the high load US06 cycle, when
operated on the regular octane fuel they are likely to see in real world operation. While it is
possible that a future engine may be designed to take advantage of higher octane fuels for GHG
reductions, EPA did not base the technology choices or effectiveness levels premised on normal
operation requiring a high octane fuel. EPA did base technology choices for
turbocharged/downsized engines, Miller Cycle engines, and Atkinson Cycle engines on the
premise that these engines would continue to use regular-grade 87 AKI fuel as a manufacturer
recommended fuel and EPA included the cost of technologies necessary to protect for operation
on such fuels, including:
• Sufficient intake camshaft phaser authority to reduce effective compression ratio for
pre-ignition knock abatement (ATK2, "advanced" ATK2, and Miller Cycle)
• Use of an integrated exhaust manifold and use of split cylinder head and engine block
cooling system control (TDS24, Miller Cycle)
• Use of cooled EGR ("advanced" ATK2, Miller Cycle, TDS24)
2-211
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Manufacturers always have the option of designating their vehicle as 'premium fuel required'
allowing them to perform emission testing using a high octane variant of Tier 3 E10 gasoline.
Fuel effects are also discussed in detail with regard to Atkinson cycle engines in Chapter
2.3.4.1.8 and turbocharged and downsized engines in Chapter 2.3.4.1.9.
2.3.1.4 Vehicle Classification
The determination of the most appropriate values for technology effectiveness and cost
depends on the characteristics of the particular vehicle to which the technologies are applied. In
the FRM and Draft TAR, the six vehicle classes defined for the purpose of characterizing
technology effectiveness were derived from the vehicle size classifications defined in 40 CFR
§600.315-08. These classes are based on vehicle interior volume and gross vehicle weight rating
attributes, and were defined for the purpose of labeling fuel economy in a way that allows
consumers to compare vehicles within commonly recognized market segments. The
classification of vehicles for estimation of technology costs in the FRM and Draft TAR
accounted for the various engine and valvetrain configurations most prevalent in the baseline
fleet, and together with the six effectiveness classes produced a total of 19 vehicle types. While
overall this method of grouping placed similar vehicles together, stakeholder comments on the
Draft TAR, including those from FCA, highlighted examples where some dissimilar vehicles
were assigned the same cost and effectiveness benefits.
For this Proposed Determination assessment, EPA has refined the vehicle classification
approach in several ways. First, for the purpose of assigning the most representative estimates for
technology effectiveness, EPA has classified vehicles according to the attributes of vehicle road
load power and engine power-to-vehicle weight ratio as described in Section 2.3.3.2. Unlike the
Draft TAR's size-based effectiveness classifications, the ALPHA model effectiveness estimates
are now developed according to low, medium, and high vehicle power-to-weight levels,
abbreviated as 'LPW', 'MPW', and 'HPW', respectively. The first two of these are divided further
into low and high vehicle road load categories, abbreviated as 'LRL' and 'HRL'. An additional
class dedicated to trucks with heavy towing and hauling capability results in a total of six
ALPHA classes for technology effectiveness, as shown in Table 2.32.
Table 2.32 ALPHA Classes for Characterizing Technology Effectiveness
ALPHA Class
Power-to-Weight Ratio
Vehicle Road Load
LPW LRL
Low
Low
LPW HRL
Low
High
MPW LRL
Medium
Low
MPW HRL
Medium
High
HPW
High
-
Truck
-
-
Second, for this Proposed Determination, EPA has refined the classification of vehicle curb
weights, which is one of the elements considered when categorizing vehicles for the purpose of
assigning technology costs. For the FRM and Draft TAR analyses, the same vehicle grouping
that was used for effectiveness classification was also the basis for the vehicle grouping used for
2-212
-------
Technology Cost, Effectiveness, and Lead Time Assessment
cost classification. For example, the unique production-weighted average curb weights for the
small car and large car classes were used to calculate technology costs for mass reduction and
electrification (battery and non-battery costs) for the vehicles within those classes. For this
Proposed Determination, EPA has added a classification by curb weight as shown in Table 2.33,
which is independent of the ALPHA classes shown above in Table 2.32. As a result, for this
updated analysis, EPA is able to apply technology costs to vehicles within a narrower range of
curb weights, thus improving the representativeness of the costs applied. This is particularly
relevant for electrification and mass reduction, two technologies for which the costs directly
relate to vehicle curb weight.
Table 2.33 Curb Weight Classes for Characterizing Technology Cost
Curb
Weight
Class
Description
Curb Weight Range (lbs)
Average
Curb Weight (lbs)
(Volume Weighted)
Std. Dev.
(lbs)
Production
Volume
(MY2015)
Greater
than
Less than or
equal
1
Passenger Vehicle_l
-
3145
2822
220
3,012,100
2
Passenger Vehicle_2
3145
3437
3285
76
2,821,695
3
Passenger Vehicle_3
3437
3729
3554
89
3,083,238
4
Passenger Vehicle_4
3729
4351
3995
164
2,641,538
5
Passenger Vehicle_5
4351
-
4820
486
3,263,377
6
Pickup Truck
-
-
4815
506
1,786,224
7
PEVs (PHEVs/BEVs)
-
-
3772
845
123,836
In EPA's Lumped Parameter Model (LPM) and OMEGA fleet compliance analysis, vehicle
types are used to distinguish between vehicles for which fundamental characteristics cause
technology cost and effectiveness values to vary. As described above, effectiveness is influenced
by road load power and power-to-weight ratio, while cost is influenced by the starting engine
configuration, curb weight, and in the case of trucks, a requirement for heavy towing. In addition
to the overarching vehicle types, EPA also uses specific data for the baseline vehicles, including
the particular technologies applied and power-to-weight ratios in order to produce appropriate
estimates of incremental cost and effectiveness for each individual vehicle. EPA's approach for
accounting for individual vehicle characteristics when determining appropriate technology
effectiveness values is described further in Section 2.3.3.5. The approach for accounting for the
previously applied technologies when assigning incremental technology cost and effectiveness
values is described further in Chapter 5.3.4.
EPA's third refinement of the vehicle classification approach for this Proposed Determination
was to expand the number of vehicle types to 29, an increase from the 19 vehicle types used in
the FRM and Draft TAR analyses. The new vehicle type definitions, derived from the
combination of cost and effectiveness classifications, are shown in Table 2.34 along with
examples of some of the higher volume vehicle models in the MY2015 fleet.
Increasing the number of vehicle types was done in part to accommodate the additional curb
weight criteria and revised ALPHA class definitions described above, while also responding to
stakeholder comments that the FRM and Draft TAR classification approach tended to group
dissimilar vehicles together. In this updated technology assessment, each of the refined 29
vehicle types contain a narrower range of the vehicle characteristics with the greatest influence
on technology effectiveness and cost; specifically, power-to-weight ratio, road load power, curb
weight, and original engine configuration. Consequently, the higher power-to-weight ratios
2-213
-------
Technology Cost, Effectiveness, and Lead Time Assessment
typical of MY2015 are more appropriately represented in this Proposed Determination than
would have been possible with the classification approach used in the FRM and Draft TAR. The
overall result of this updated vehicle classification approach is a set of ALPHA classes and
vehicle types that provide greater resolution than the 19 vehicle types used in the Draft TAR, and
advance the goal of applying the most representative cost and effectiveness estimates for
technologies applied to the MY2015 fleet. See Section 2.3.3.2 for more details on the
classification approach for effectiveness, and comparison with the Draft TAR and FRM
approach.
Table 2.34 Expanded Vehicle Types for Characterizing Technology Cost and Effectiveness
Veh
Type
ALPHA
Class
Curb
Wgt
Class
Engine
Config
Example
Veh
Type
ALPHA
Class
Curb
Wgt
Class
Engine
Config
Example
1
LPW LRL
1
14 DOHC
Sentra, Corolla
16
MPW LRL
3
V6DOHC
IS250
2
MPW LRL
1
14 DOHC
Dart, Focus
17
LPW HRL
3
V6DOHC
Transit
3
MPW LRL
2
14 DOHC
Altima, Camry
18
HPW
4
V6DOHC
Charger
4
LPW HRL
2
14 DOHC
Rogue, Patriot
19
MPW HRL
4
V6DOHC
Pathfinder,Journey
5
MPW LRL
3
14 DOHC
Malibu, 200
20
HPW
5
V6DOHC
Camaro
6
LPW HRL
3
14 DOHC
Forester, Cherokee
21
MPW HRL
5
V6DOHC
Grand Cherokee
7
LPW HRL
4
14 DOHC
Outback, Equinox
22
Truck
6
V6DOHC
Tacoma, Frontier
8
Truck
6
14 DOHC
Colorado, Tacoma
23
HPW
5
V8 0HV
Charger
9
Truck
6
V6 0HV
Silverado, Sierra
24
MPW HRL
5
V8 0HV
Tahoe, Suburban
10
HPW
3
V6SOHC
RDX, TLX
25
Truck
6
V8 0HV
Silverado, Sierra
11
MPW HRL
4
V6SOHC
Odyssey
26
HPW
4
V8DOHC
Mustang, SL550
12
LPW LRL
1
V6DOHC
Cruze,Focus turbos
27
HPW
5
V8DOHC
QX80, GL550
13
MPW LRL
2
V6DOHC
Fiesta turbo
28
MPW HRL
5
V8DOHC
GX460, Sequoia
14
LPW LRL
2
V6DOHC
Passat
29
Truck
6
V8DOHC
Tundra, F150
15
HPW
3
V6DOHC
ES350, Impala, Q50
2.3.2 Approach for Determining Technology Costs
This section reviews the primary sources and approaches EPA uses to estimate technology
costs. These costs are divided into several primary types, including direct manufacturing costs,
indirect costs, and maintenance and repair costs.
The estimation of direct manufacturing costs includes consideration of cost reduction over
time through manufacturer learning. Indirect costs are estimated by application of indirect cost
multipliers (ICMs). EPA computes total costs as the sum of direct manufacturing cost (DMC)
and indirect cost (IC). This approach was used in the Draft TAR analysis and is also used in this
Proposed Determination analysis.
Multiple comments from NGOs (American Council for an Energy-Efficient Economy
(ACEEE), Union of Concerned Scientists (UCS), and Environmental Defense Fund (EDF))
supported EPA's use of ICMs rather than retail price equivalents (RPEs) as a means of estimating
indirect costs.
We also received some comments on our cost reductions through manufacturer learning.
Notably, Ford argued that product cadence does not allow for cost reductions from learning to be
2-214
-------
Technology Cost, Effectiveness, and Lead Time Assessment
realized since new products are constantly being developed. However, the learning effects we
estimate should be taken as occurring at the level of the supplier, not that of the automaker. Since
we have not estimated efficiency improvements to individual technologies during the time frame
of the analysis, we do not believe that such redesign to improve the "current best technology" to
the "next best technology" is necessary to achieve the reductions we expect for the costs we have
estimated.
2.3.2.1 Direct Manufacturing Costs
Estimates of direct manufacturing costs (DMC) used in this analysis come from many
sources, including published technical papers, reports, and analyses, teardown studies contracted
by EPA, and supplier- and OEM-provided data (sometimes including confidential business
information).
The 2015 NAS Report522 supported EPA's assessment that teardown studies are perhaps the
best source of DMC estimates. NAS encouraged the agencies to make use of tear-down studies
where available, stating, "the use of teardown studies has improved the agencies' estimates of
costs" (NAS pp. S-3). This advice was reflected in EPA's continued use of teardown studies to
develop many of the technology cost assumptions in the Draft TAR. Public comments on the
Draft TAR received from the American Council for an Energy-Efficient Economy (ACEEE) and
the Union of Concerned Scientists (UCS) additionally were supportive of EPA's use of teardown
studies. The summary below provides more information on our sources for cost information for
many of the technologies considered in this analysis.
2.3.2.1.1 Costs from Tear-down Studies
As in the Draft TAR, there are a number of technologies in this analysis that have been costed
using the tear-down method. As a general matter, EPA believes, and the NAS agrees,523 that the
most rigorous 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. Many technology cost studies in the
literature are based on information collected from OEMs, suppliers, or "experts" in the industry
and are thus non-reproducible and non-transparent. In contrast, EPA-sponsored teardown studies
are completely transparent and include a tremendous amount of data and analyses to improve
accuracy.
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.
2-215
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Since the early development of the 2012-2016 rule, EPA has contracted with FEV, Inc. to
conduct tear-down cost studies for a number of key technologies evaluated 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
MY2012-2016 rulemaking process.524
Additional cost studies were completed and used in support of the 2017-2025 FRM. These
include vehicle tear downs of a Ford Fusion power-split hybrid and a conventional 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 develop
costs for the P2 hybrid used in the following MY2017-2025 FRM.YY 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 to provide supplemental battery costing information to that associated with the NiMH
battery in the Fusion, because automakers were moving to Li-ion battery technologies due to the
higher energy and power density of these batteries. As noted, this HEV cost work, including the
extension of results to P2 HEVs, has been documented in a report prepared by FEV and was used
in support of the 2017-2025 FRM. Because of the complexity and comprehensive scope of this
HEV analysis, EPA commissioned a separate peer review focused exclusively on the new tear
down costs developed for the HEV analysis. Reviewer comments generally supported FEV's
methodology and results, while including a number of suggestions for improvement, many of
which were subsequently incorporated into FEV's analysis and EPA final report. The peer
review comments and responses were made available in the rulemaking docket.
Some of the technologies for which FEV has completed teardown studies over the course of
the contract with EPA are listed below. These completed studies provide a thorough evaluation
of these technologies' costs relative to their baseline (or replaced) technologies.
• 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
• SGDI and T-DS on a SOHC (single overhead cam) on a V6 engine, replacing a
conventional 3-valve/cylinder SOHC V8 engine
• SGDI and T-DS on a DOHC 14 engine, replacing a DOHC V6 engine
• 6-speed automatic transmission (AT), replacing a 5-speed AT
• 6-speed wet dual clutch transmission (DCT) replacing a 6-speed AT.
• 8-speed AT replacing a 6-speed AT
• 8-speed DCT replacing a 6-speed DCT
• 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 BEVs.
• Fiat Multi-Air engine technology. (Although results from this cost study are included
in the rulemaking docket, they were not used in the 2017-2025 rulemaking's technical
YY Examples of production P2 Hybrids are the Hyundai Sonata Hybrid and the Infiniti M35 Hybrid
2-216
-------
Technology Cost, Effectiveness, and Lead Time Assessment
analyses because the technology is under patent and therefore not considered in the
2017-2025 time frame).
In addition, FEV and EPA extrapolated the engine downsizing costs for the following
scenarios that were based on the above study cases:
• Downsizing a SOHC 2 valve/cylinder V8 engine to a DOHC V6
• Downsizing a DOHC V8 to a DOHC V6
• Downsizing a SOHC V6 engine to a DOHC 4 cylinder engine
• Downsizing a DOHC 4 cylinder engine to a DOHC 3 cylinder engine
Teardown work was also performed in the area of mass reduction technologies. This work is
highlighted in greater detail in Chapter 2.3.4.6 of this TSD.
EPA has relied on the findings of FEV for estimating the cost of the technologies covered by
the tear-down studies. However, 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 capital if
technologies are introduced too rapidly for some indirect costs to be fully recovered. While EPA
considers the FEV tear-down analysis results to be generally valid for the 2022 to 2025 time
frame for fully mature, high sales volumes, FEV performed supplemental analysis supporting the
FRM to consider potential stranded capital costs, and we have included these in our primary
analyses of program costs.
2.3.2.1.2 Electrified Vehicle Battery Costs
As in the 2012 FRM and the Draft TAR, EPA has used the BatPaC model525 to estimate
battery costs for electrified vehicles. Developed by Argonne National Laboratory (ANL) for the
Vehicle Technologies Program of the U.S. Department of Energy (DOE) Office of Energy
Efficiency and Renewable Energy, the BatPaC model allows users to estimate the manufacturing
cost of battery packs for various types of electrified powertrains given battery power and energy
requirements as well as other design parameters.
In the 2015 NAS report (p. 4-25), the NAS committee endorsed the importance of the use of a
bottom-up battery cost model such as BatPaC, further finding that "the battery cost estimates
used by the agencies are broadly accurate" (Finding 4.4, p. 4-43). Since the publication of the
FRM, BatPaC has been further refined and updated with new costs for some cathode chemistries
and cell components, improved thermal management calculations, and improved accounting for
plant overhead costs. Further changes were released in late 2015 and include additional
chemistries, updated material costs, improved calculation of electrode thickness limits, and
improved estimation of cost and energy requirements of certain manufacturing steps and material
production processes.526 EPA has used the most recent version of BatPaC to revise the battery
cost projections used in this Proposed Determination analysis, as detailed in Chapter 2.3.4.3.7
(Cost of Batteries for xEVs).
2-217
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In the 2012 FRM, the agencies developed cost and effectiveness values for the mild and P2
HEV configurations, two different all-electric mileage ranges for PHEVs (20 and 40 in-use
miles) and three different mileage ranges for BEVs (75, 100 and 150 in-use miles). In the Draft
TAR analysis, EPA introduced cost and effectiveness values for a new 48-Volt mild hybrid, and
changed the 150-mile BEV configuration to a 200-mile configuration. These changes are
retained in the current analysis. Additional updates to the cost inputs and methodology applied to
electrified vehicles are described in Chapter 2.3.4.3 (Electrification: Data and Assumptions for
this Assessment).
2.3.2.1.3 Specific PMC Updates since the Draft TAR
EPA continues to believe that teardown studies are the most robust source of cost estimates.
For the Draft TAR, EPA updated costs from other prior teardowns (largely the transmission
teardowns) based on updates to those studies performed by FEV and these costs are largely
retained for this analysis. EPA also updated battery costs for electrified vehicles based on
improvements to battery sizing estimation and an updated set of input metrics to the BatPaC
model. EPA has retained the new technologies introduced in the Draft TAR analysis, specifically
a 48-Volt mild hybrid, a more capable naturally aspirated Atkinson cycle engine with a high
compression ratio, a Miller cycle engine and a 200-mile range electric vehicle. Technology costs
for 48V mild hybrid are largely carried over from the estimates in the Draft TAR which were
derived from information provided by a previous teardown study of a high-voltage mild hybrid.
Costs for the more capable Atkinson cycle engine were based on costs reported by NAS. All
technology costs have been updated to 2015 dollars for the Proposed Determination analysis
(Draft TAR costs were in 2013 dollars).
2.3.2.1.4 Approach to Cost Reduction through Manufacturer Learning
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 EPA and NHTSA have
both done in past regulatory analyses—to apply at the industry-wide level, particularly in
industries that utilize many common technologies and component supply sources. EPA believes
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 (i.e., the manufacturing learning curve).
NAS recommended that the agencies "continue to conduct and review empirical evidence for
the cost reductions that occur in the automobile industry with volume, especially for large-
volume technologies that will be relied on to meet the CAFE/GHG standards." (NAS pp. 7-23)
EPA has conducted such a review under contract to ICF looking at learning in mobile source
industries. The goal of the effort was to provide an updated assessment on learning and its
existence in manufacturing industries. An extensive literature review was conducted and the
most applicable and appropriate studies were chosen with the help of a subject matter expert
2-218
-------
Technology Cost, Effectiveness, and Lead Time Assessment
(SME) that is one of the leading experts in this area.22 EPA hoped that the study would provide
clear learning rates that could be applied in various mobile source manufacturing industries
rather than the more general learning rates used in the past. That study was completed in
September of 2015. In the Draft TAR, we noted that a peer review had been initiated and
completed, but the subsequent final report was not completed in time for inclusion in the docket
supporting the Draft TAR. That final report, which includes responses to the peer review is now
completed and is contained in the docket supporting this Proposed Determination.527
In the contracted study, ICF performed this literature review and analysis of learning in the
mobile source sector with the assistance of a Subject Matter Expert (Dr. Linda Argote of
Carnegie Mellon University). The draft report, Cost Reduction through Learning in
Manufacturing Industries and in the Manufacture of Mobile Sources, was subsequently peer-
reviewed by three well-known experts in the field of learning (Marvin Lieberman, Ph.D.,
University of California, Los Angeles (UCLA) Anderson School of Management; Natarajan
Balasubramanian, Ph.D., Whitman School of Management, Syracuse University; and Chad
Syverson, Ph.D., University of Chicago Booth School of Business). The peer review was carried
out for EPA by RTI International based on EPA Science Policy Council Peer Review Handbook,
4th Edition, and was completed in May 2016.
The study consists of two parts: a literature review, and an estimate of a mobile source
progress ratio. A total of 53 studies on learning were examined, with 20 of these selected for
detailed review (the other 33 received a more cursory review and are not discussed in detail in
the report). Five of these studies were used as the basis to estimate the progress ratio for the
mobile source sector. On the basis of these studies, the SME noted: "The mean learning rate is
estimated to be -0.245, with a standard error of 0.0039. Thus, the lower bound for a 95 percent
confidence interval for the learning rate is -0.253; the upper bound is -0.238. These estimates
translate into a mean progress ratio of 84.3 percent. The confidence interval around this number
ranges from 83.9 percent to 84.8 percent, suggesting that one can be reasonably confident that
the progress ratio falls in this interval. Thus, the best estimate of the progress ratio in mobile
source industries is 84 percent." This is the value that EPA used in both the Draft TAR and this
Proposed Determination.
As a result, the learning curve recommended for use by the report has slightly lower learning
rates than those EPA has used in the past. Past EPA studies have used a learning rate based on a
curve that resulted in a 20 percent cost reduction for each doubling of volume; the recommended
rate results in cost reductions of 15 percent. As such, EPA has updated learning rates to be
consistent with the recommendation of the report. The curve used in this analysis is:
Vt+i = axt+i
Where:
yt+i = Costs required to produce a unit at time t+1
a = Costs required to produce the first unit
xt+i = Cumulative number of units produced through period t+1
zz The SME was Dr. Linda Argote of Carnegie Mellon University.
2-219
-------
Technology Cost, Effectiveness, and Lead Time Assessment
b = A parameter measuring the rate at which unit costs change as cumulative output
increases; i.e., the learning rate
For this analysis, EPA has used this equation to estimate the learning effects and have
generated the learning curves shown below. How these learning curves were actually generated
using the above curve is described in a memorandum contained in the docket.528 In general, the
new learning factors were generated in a way to provide similar results to past analyses.
However, because the new rate is lower, there are subtle differences especially in years further
from the "base" year (i.e., the year where the learning factor is 1.0). The docket memorandum
makes this clearer by providing the new factors alongside the factors used in the 2012 FRM for
comparison. Note that the factors used in this Proposed Determination are identical to those used
in the Draft TAR.
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, presumably, learning impacts
have already occurred. Learning effects on the steep-portion of the learning curve 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, learning effects on the flat portion of the learning curve have been applied. The learning
factor curve applied to each technology are summarized in Table 2.35 with the actual year-by-
year factors for each corresponding curve shown in Table 2.36.
2-220
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.35 Learning Effect Algorithms Applied to Technologies Used in this Analysis
Technology
Learning Factor "Curve"3
Aero, active
24
Aero, passive
24
Atkinson, level 1
24
Atkinson, level 2
24
Cam configuration changes
V6 OHVto V6 DOHC
28
V6 SOHCto V6 DOHC
23
V8 OHVto V8 DOHC
28
V8 SOHCto V8 DOHC
23
V8SOHC3V to V8 DOHC
23
Charger, in-home, BEV
26
Charger, in-home, PHEV20
26
Charger, in-home, PHEV40
26
Charger, in-home, labor
1
Cylinder deactivation
24
Direct injection, stoichiometric, gasoline
23
Diesel, advanced (Tier3)
23
Diesel, lean NOx trap
23
Diesel, selective catalytic reduction
23
Downsizing, associated with turbocharging
14 DOHC to 13 DOHC
23
14 DOHC to 14 DOHC
23
V6 OHVto 14 DOHC
28
V6 SOHCto 14 DOHC
23
V6 DOHC to 14 DOHC
23
V8 OHVto V6 DOHC
28
V8 SOHCto V6 DOHC
23
V8SOHC3V to V6 DOHC
23
Engine friction reduction, level 1
1
Engine friction reduction, level 2
1
EGR, cooled
23
Electric power steering
24
BEV75, battery pack
26
BEV100, battery pack
26
BEV200, battery pack
26
BEV75, non-battery items
28
BEV100, non-battery items
28
BEV200, non-battery items
28
HEV, Mild, battery pack
31
HEV, Mild, non-battery items
23
HEV, Strong, battery pack
31
HEV, Strong, non-battery items
23
HEV, Plug-in, battery pack
26
HEV, Plug-in, non-battery items
23
Improved accessories, level 1
24
Improved accessories, level 2
24
Low drag brakes
1
Lower rolling resistance tires, level 1
1
2-221
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Lower rolling resistance tires, level 2
32
Lube, engine changes to accommodate low friction lubes
1
Mass reduction <15%
30
Mass reduction >=15%
30
Secondary axle disconnect
24
Stop-start
25
Turbo, 18-21 bar
23
Turbo, 24 bar
23
Turbo, Miller-cycle
23
TRX11/12
23
TRX21/22
23
Note:
a See table below.
The actual year-by-year factors for the numbered curves shown in Table 2.36.
Table 2.36 Year-by-year Learning Curve Factors for the Learning Curves Used in this Analysis
Curve
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
22
1.37
1.33
1.29
1.25
1.21
1.18
1.15
1.13
1.11
1.08
1.06
1.04
1.02
1.00
23
1.00
0.98
0.96
0.94
0.92
0.91
0.89
0.88
0.87
0.85
0.84
0.83
0.82
0.82
24
1.09
1.06
1.03
1.00
0.98
0.96
0.94
0.92
0.91
0.89
0.88
0.87
0.85
0.84
25
2.03
1.62
1.28
1.00
0.91
0.84
0.80
0.76
0.74
0.71
0.69
0.67
0.66
0.64
26
3.05
2.44
2.11
1.89
1.74
1.61
1.51
1.43
1.36
1.30
1.25
1.20
1.16
1.12
27
1.00
0.91
0.84
0.80
0.76
0.74
0.71
0.69
0.67
0.66
0.64
0.63
0.62
0.61
28
1.13
1.09
1.06
1.03
1.00
0.98
0.96
0.94
0.92
0.91
0.89
0.88
0.87
0.85
29
1.17
1.13
1.09
1.06
1.03
1.00
0.98
0.96
0.94
0.92
0.91
0.89
0.88
0.87
30
1.29
1.24
1.20
1.17
1.13
1.09
1.06
1.03
1.00
0.98
0.96
0.94
0.92
0.91
31
3.18
2.54
2.03
1.62
1.28
1.00
0.91
0.84
0.80
0.76
0.74
0.71
0.69
0.67
32
1.74
1.61
1.51
1.43
1.36
1.30
1.25
1.20
1.16
1.12
1.09
1.06
1.04
1.01
Importantly, where the factors shown in Table 2.36 equal "1.00" represents the year for which
any particular technology's cost is based. Thus, if curve 1 is applied to a technology - such as in
the case of low friction lubes - it assumes no additional learning takes place over time. In the
case of stop-start technology, curve 25 is applied. In this case, the cost estimate used for stop-
start is considered a MY2015 cost. Therefore, its learning factor equals 1.00 in 2015 and then
decreases going forward to represent lower costs due to learning effects. Its learning factors are
greater than 1.00 in years before 2015 to represent "reverse" learning, i.e., higher costs than our
2015 estimate since production volumes have, presumably, not yet reached the point where our
cost estimate can be considered valid. Not all of the learning curve factors follow this rule using
the updated curve approach used in the Draft TAR and in this Proposed Determination. Also of
interest is that only curves 25 (stop-start), 26 (BEV & PHEV batteries) and 31 (mild and strong
HEV batteries) show any steeper learning beyond the 2017 to 2020 time frame, and even those
curves show less than 5 percent year-over-year cost reductions beyond 2020. In other words,
most curves are well into the flatter portion of the learning curve, and even those that are not are
well beyond the steep learning that occurs at the early stages of learning, by the time frame
considered in this analysis.
2-222
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Because of the nature of full electric and plug-in electric vehicle battery pack development,
the industry is arguably early in the learning-by-doing phase for the types of batteries considered.
Our approach, consistent with that used in the FRM, has been to develop a direct manufacturing
cost based on sales of 450,000 units. EPA has considered that to be a valid MY2025 cost (i.e.,
the cost is based in 2025). With that as the MY2025 cost, the costs are considered as understood
today and a best fit learning curve is projected between the costs in those near-term and long-
term years. This is described in more detail in the docket memorandum mentioned earlier.529
Note that the 450,000 unit sales is considered a valid MY2025 volume for batteries because that
volume is meant to represent volumes at a given production line (a battery supplier production
line, not an OEM vehicle production line) and takes into consideration worldwide demand for
automotive and other mobile source battery packs, not just U.S.-directed automotive battery
packs.
Note that the effects of learning on individual technology costs can be seen in the cost tables
presented in Section 2.3.4, below. For each technology, the direct manufacturing costs for the
years 2017 through 2025 are shown. The changes shown in the direct manufacturing costs from
year-to-year reflect the cost changes due to learning effects.
2.3.2.2 Indirect Costs
2.3.2.2.1 Methodologies for Determining Indirect Costs
To produce a unit of output, vehicle manufacturers incur direct and indirect costs. Direct costs
include cost of materials and labor costs. Indirect costs are all the costs associated with
producing the unit of output that are not direct costs - for example, they 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 good sold. Although it is possible to account for direct costs allocated to each unit
of good 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 both EPA and NHTSA) have frequently used
these multipliers to predict the resultant impact on costs associated with manufacturers'
responses to regulatory requirements. The best approach, if it were possible, 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, or 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 produce
common incremental changes in all indirect cost contributors as well as net income. However, 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
2-223
-------
Technology Cost, Effectiveness, and Lead Time Assessment
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, modified multipliers have been developed by EPA, working with a
contractor, for use in rulemakings.530 These multipliers are referred to as indirect cost multipliers
(or ICMs). In contrast to RPE multipliers, ICMs assign unique incremental changes to each
indirect cost contributor as well as net income.
ICM = (direct cost + adjusted indirect cost)/(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: the less complex a
technology, the lower its ICM, and the longer the time frame for applying the technology, the
lower the ICM. This methodology was used in the cost estimation for the recent light-duty MYs
2012-2016 and MYs 2017-2025 rulemaking and for the heavy-duty MYs 2014-2018 rulemaking.
There was no serious disagreement with this approach in the public comments to any of these
rulemakings. The ICMs for the light-duty context were developed in a peer-reviewed report from
RTI International and were subsequently discussed in a peer-reviewed journal article.531
Importantly, since publication of that peer-reviewed journal article, the EPA has revised the
methodology to include a return on capital (i.e., profits) based on 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.
There is some level of uncertainty surrounding both the ICM and RPE markup factors. The
ICM estimates used in the Draft TAR and this Proposed Determination, consistent with the
FRM, 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. 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,
2-224
-------
Technology Cost, Effectiveness, and Lead Time Assessment
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.
2.3.2.2.2 Indirect Cost Estimates Used in this Analysis
Since their original development in February 2009, the agencies made changes to both the
ICM factors and to the method of applying those factors relative to the factors developed by RTI
and presented in their reports. These changes have been described and explained in several
rulemakings over the years, most notably the 2017-2025 FRM and the more recent Heavy-duty
GHG Phase 2 final rule (81 FR 73478).
Although the Draft TAR analysis assessed indirect costs using both the ICM and RPE
approaches, EPA has focused on the ICM approach for the Proposed Determination analysis,
considering ICMs to be the better means of estimating indirect cost impacts resulting from
regulatory changes. EPA believes that this stance is consistent with the support expressed by
NAS in their 2015 report,as well as several commenters on the Draft TAR. Comments from
the American Council for an Energy-Efficient Economy (ACEEE), the Union of Concerned
Scientists (UCS), and Environmental Defense Fund (EDF) all supported the use of ICMs. EPA
has also performed a sensitivity analysis using RPEs instead of ICMs, as discussed in Section
C.1.2 of the Proposed Determination Appendix.
For this Proposed Determination, EPA is assessing indirect costs using the same ICMs as
used in the Draft TAR, as shown in Table 2.37. 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 2.37 Indirect Cost Multipliers Used in this Analysis532
2017-2025 FRM and TSD
Complexity
Near term
Long term
Low
1.24
1.19
Medium
1.39
1.29
Highl
1.56
1.35
High2
1.77
1.50
There are two important aspects to the ICM method employed by EPA. First, the ICM
consists of two portions: a small warranty-related term and a second, larger term to cover all
other indirect costs elements. The breakout of warranty versus non-warranty portions to the
ICMs are presented in Table 2.38. The latter of these terms does not decrease with learning and,
instead, remains constant year-over-year despite learning effects which serve to decrease direct
manufacturing costs. Learning effects were described above. The second important note is that
AAA In the 2015 NAS study, the committee stated: "The committee conceptually agrees with the Agencies' method
of using an indirect cost multiplier instead of a retail price equivalent to estimate the costs of each technology
since ICM takes into account design challenges and the activities required to implement each technology." (NAS
Finding 7.1)
2-225
-------
Technology Cost, Effectiveness, and Lead Time Assessment
all indirect costs are forced to be positive, even for those technologies estimated to have negative
direct manufacturing costs.
Table 2.38 Warranty and Non-Warranty Portions of ICMs
Near term
Long term
Complexity
Warranty
Non-warranty
Warranty
Non-warranty
Low
0.012
0.230
0.005
0.187
Medium
0.045
0.343
0.031
0.259
Highl
0.065
0.499
0.032
0.314
High2
0.074
0.696
0.049
0.448
The complexity levels and subsequent ICMs applied throughout this analysis for each
technology are shown in Table 2.39 and are identical to those used in the Draft TAR.
Table 2.39 Indirect Cost Markups (ICMs) and Near Term/Long Term Cutoffs Used in EPA's Analysis
Technology
ICM Complexity
Short term thru
Aero, active
Low2
2018
Aero, passive
Med2
2024
Atkinson, level 1
Med2
2018
Atkinson, level 2
Med2
2024
Cam configuration changes
V6 OHVto V6 DOHC
Med2
2018
V6 SOHC to V6 DOHC
Med2
2018
V8 OHVto V8 DOHC
Med2
2018
V8 SOHC to V8 DOHC
Med2
2018
V8SOHC3V to V8 DOHC
Med2
2018
Charger, in-home, BEV
Highl
2024
Charger, in-home, PHEV20
Highl
2024
Charger, in-home, PHEV40
Highl
2024
Charger, in-home, labor
None
2024
Cylinder deactivation
Med2
2018
Direct injection, stoichiometric, gasoline
Med2
2018
Diesel, advanced (Tier3)
Med2
2018
Diesel, lean NOx trap
Med2
2018
Diesel, selective catalytic reduction
Med2
2018
Downsizing, associated with turbocharging
14 DOHC to 13 DOHC
Med2
2018
14 DOHC to 14 DOHC
Med2
2018
V6 OHVto 14 DOHC
Med2
2018
V6 SOHC to 14 DOHC
Med2
2018
V6 DOHC to 14 DOHC
Med2
2018
V8 OHVto V6 DOHC
Med2
2018
V8 SOHC to V6 DOHC
Med2
2018
V8SOHC3V to V6 DOHC
Med2
2018
Engine friction reduction, level 1
Low2
2018
Engine friction reduction, level 2
Low2
2024
EGR, cooled
Med2
2024
Electric power steering
Low2
2018
BEV75, battery pack
High2
2024
2-226
-------
Technology Cost, Effectiveness, and Lead Time Assessment
BEV100, battery pack
High2
2024
BEV200, battery pack
High2
2024
BEV75, non-battery items
High2
2024
BEV100, non-battery items
High2
2024
BEV200, non-battery items
High2
2024
HEV, Mild, battery pack
Highl
2024
HEV, Mild, non-battery items
Med2
2018
HEV, Strong, battery pack
Highl
2024
HEV, Strong, non-battery items
Highl
2018
HEV, Plug-in, battery pack
High2
2024
HEV, Plug-in, non-battery items
Highl
2018
Improved accessories, level 1
Low2
2018
Improved accessories, level 2
Low2
2018
Low drag brakes
Low2
2018
Lower rolling resistance tires, level 1
Low2
2018
Lower rolling resistance tires, level 2
Low2
2018
Lube, engine changes to accommodate low friction lubes
Low2
2018
Mass reduction <15%
Low2
2024
Mass reduction >=15%
Med2
2024
Secondary axle disconnect
Low2
2018
Stop-start
Med2
2018
Turbo, 18-21 bar
Med2
2018
Turbo, 24 bar
Med2
2024
Turbo, Miller-cycle
Med2
2024
TRX11/12
Low2
2018
TRX21/22
Low2
2024
For mass reduction costs in the Draft TAR, EPA developed a new approach to calculating
indirect costs due to the unique nature of the direct manufacturing costs that EPA has developed
(see Draft TAR Section 5.3.4.6.1). We are using the same approach in this Proposed
Determination. Mass reduction strategies, unlike other efficiency technologies, often involve
multiple systems and components on a vehicle. A portion of the indirect costs for parts that have
design and production outsourced to suppliers are incorporated into the direct manufacturing cost
estimates. Components that are designed in-house and possibly produced in-house by the
manufacturer, such as the body and frame structures, have higher indirect costs applied. This
distinction between supplier and in-house parts is consistent with the recommendations of a
study done by Argonne National Laboratory.533 In that study, the authors suggested retail price
equivalent markups of 1.5x direct costs for parts sourced from a supplier, and 2x direct costs for
parts sourced internally. The end result, presumably, is an equal total cost, but the markups
account for differences in where the indirect costs are incurred. Using that as a basis EPA
adjusted the supplied technology ICMs (shown in Table 2.37) by the ratio 2/1.5 to determine in-
house ICMs at the "engineered solution" mass reduction point (see Draft TAR Sections 5.3.4.6.1.1
and 5.3.4.6.1.2) which happened to be approximately 20 percent mass reduction level for the car
teardown study and the truck teardown study. Since those mass reduction levels were deemed
"medium" complexity levels in the FRM, and because EPA still believes that to be a good
assessment of the complexity level, EPA has worked with only the medium complexity ICMs in
the context of mass reduction. As a result, the ICMs used for mass reduction are as shown in
Table 2.40.
2-227
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.40 Mass Reduction Markup Factors used by EPA in this TSD
Supplier Provided Mass Reduction
In-house Provided Mass Reduction
Markup & Complexity
Near term
Long term
Near term
Long term
ICM - Medium complexity
1.39
1.29
1.85
1.72
The final element of the unique nature of the indirect cost calculations developed by EPA for
mass reduction in this analysis, is to calculate the indirect costs using the above ICMs only at the
engineered solution point. Notably, EPA applied the markups to the sum of the absolute values
of all mass reduction ideas throughout the entire direct manufacturing cost curve. In that way,
negative direct costs that are projected at the lower mass reduction levels still have a positive
impact on calculated indirect costs. Once the indirect costs were determined via this
methodology at the engineered solution, EPA generated an indirect cost curve extending through
$0/kg at 0 percent mass reduction and $8.75/kg/% at the engineered solution for cars and
$13.23/kg/% for trucks (see Table 2.41 and Table 2.42 for the values of X). The indirect costs at
all mass reduction levels between those points lie on that generated cost curve. Inherent in this
approach is the assumption that the proportion of mass reduction from supplier and in-house
components remains constant at all levels of mass reduction, based on the proportion at the
engineered solution. Those curves are shown in Table 2.41 for cars and in Table 2.42 for trucks.
Table 2.41 Mass Reduction Indirect Cost Curves used by EPA for Cars Using ICMs (dollar values in 2013$)
$/kg DMC*
ICM
$/kg IC at
Engineered
Solution
$/kg IC at Engineered
Solution
$/kg/%
IC
curve**
Near term
Supplied tech
DMC
$1.75
0.39
$0,678
$0,678+0.986=1.66
$8.75x
In-house tech
DMC
$1.16
0.85
$0,986
Long term
Supplied tech
DMC
$1.75
0.29
$0,507
$0,507+0.835=1.34
$7.06x
In-house tech
DMC
$1.16
0.72
$0,835
Notes:
* Calculated as the absolute value of all direct manufacturing costs needed to achieve the engineered solution.
** Where x is the percent mass reduction.
Table 2.42 Mass Reduction Indirect Cost Curves used by EPA for Trucks Using ICMs (dollar values in
2013$)
$/kg DMC*
ICM
$/kg IC at
Engineered
Solution
$/kg IC at Engineered
Solution
$/kg/%
IC
curve**
Near term
Supplied tech
DMC
$2.59
0.39
$1.00
$1.00+1.78=2.78
$13.23x
In-house tech
DMC
$2.09
0.85
$1.78
Long term
Supplied tech
DMC
$2.59
0.29
$0.75
$0.75+1.50=2.25
$10.73x
In-house tech
DMC
$2.09
0.72
$1.50
Notes:
2-228
-------
Technology Cost, Effectiveness, and Lead Time Assessment
* Calculated as the absolute value of all direct manufacturing costs needed to achieve the engineered solution.
** Where x is the percent mass reduction.
2.3.2.3 Maintenance and Repair Costs
2.3.2.3.1 Maintenance Costs
To estimate maintenance costs that could reasonably be attributed to the 2017-2025 standards,
EPA and NHTSA looked—in the 2017-2025 FRM—at vehicle models for which there exists a
version with a fuel efficiency and GHG emissions improving technology and a version with the
corresponding baseline technology. The difference between maintenance costs for the two
models represent a cost which the agencies attributed to the standards. For example, the Ford
Escape Hybrid versus the Ford Escape V6 was considered when estimating the types of
maintenance cost differences that might be present for a hybrid vehicle versus a non-hybrid, and
a Ford F150 with EcoBoost versus the Ford F150 5.0L was considered when estimating the types
of maintenance cost differences that might be present for a turbocharged and downsized versus a
naturally aspirated engine. In the case of low rolling resistance tires, specific parts were
considered rather than specific vehicle models.
By comparing the manufacturer recommended maintenance schedule of the items compared,
the differences in maintenance intervals for the two was estimated. With estimates of the costs
per maintenance event, a picture of the maintenance cost differences associated with the "new"
technology was developed.
EPA continues to believe that the maintenance estimates used in the FRM are reasonable and
have therefore used them again in this analysis as we did in the Draft TAR. EPA distinguished
maintenance from repair costs as follows: maintenance costs are those costs that are required to
keep a vehicle properly maintained and, as such, are usually recommended by auto makers to be
conducted on a regular, periodic schedule. Examples of maintenance costs are oil and air filter
changes, tire replacements, etc. Repair costs are those costs that are unexpected and, as such,
occur randomly and uniquely for every driver, if at all. Examples of repair costs would be parts
replacement following an accident or a mechanical failure, etc.
In Chapter 3.6 of the final joint TSD supporting the 2012 FRM, the agencies presented a
lengthy discussion of maintenance costs and the impacts projected as part of that rule.534 Table
2.43 shows the results of that analysis, the maintenance impacts used in the 2012 FRM and again
in this analysis, although the costs here have been updated to 2015$. Note that the technologies
shown in Table 2.43 are those for which EPA believes that maintenance costs would change; it is
clearly not a complete list of technologies expected to meet the MY2025 standards.
Table 2.43 Maintenance Event Costs & Intervals (2015$)
New Technology
Reference
Cost per Maintenance
Maintenance Interval
Technology
Event
(miles)
Low rolling resistance tires level 1
Standard tires
$6.91
40,000
Low rolling resistance tires level 2
Standard tires
$53.03
40,000
Diesel fuel filter replacement
Gasoline vehicle
$53.52
20,000
BEV oil change
Gasoline vehicle
-$42.02
7,500
BEV air filter replacement
Gasoline vehicle
-$31.08
30,000
BEV engine coolant replacement
Gasoline vehicle
-$64.12
100,000
2-229
-------
Technology Cost, Effectiveness, and Lead Time Assessment
BEV spark plug replacement
Gasoline vehicle
-$90.20
105,000
BEV/PHEV battery coolant
replacement
Gasoline vehicle
$127.15
150,000
BEV/PHEV battery health check
Gasoline vehicle
$42.02
15,000
Note that many of the maintenance event costs for BEVs are negative. The negative values
represent savings since BEVs do not incur these costs while their gasoline counterparts do. Note
also that the 2010 FRM is expected to result in widespread use of low rolling resistance tires
level 1 (LRRT1) on the order of 85 percent penetration. Therefore, as 2012 FRM results in
increasing use of low rolling resistance tire level 2 (LRRT2), there is a corresponding decrease in
the use of LRRT1. As such, as LRRT2 maintenance costs increase with increasing market
penetration, LRRT1 maintenance costs decrease. Importantly, the maintenance costs associated
with lower rolling resistance tires is the incremental cost of the tires at replacement; it is not
associated in any way with a decrease in durability of these tires.
2.3.2.3.2 Repair Costs
EPA's analysis accounts for the costs of repairs covered by manufacturers' warranties, and a
sensitivity analysis estimated costs for post-warranty repairs. The indirect cost multipliers
(ICMs) applied in the EPA's analyses include a component representing manufacturers'
warranty costs. For the cost of repairs not covered by OEMs' warranties, EPA has, in the past,
evaluated the potential to apply an approach similar to that described above for maintenance
costs. As for specific scheduled maintenance items, the ALLDATA subscription database
applied above provides estimates of labor and part costs for specific repairs to specific vehicle
models. However, although ALLDATA also provides service intervals for scheduled
maintenance items, it does not provide estimates of the frequency at which specific failures may
be expected to occur over a vehicle's useful life. EPA has not yet been able to develop an
alternative method to estimate the frequencies of different types of repairs, and are therefore
unable to apply these ALLDATA estimates in order to quantify the cost of repairs throughout
vehicles' useful lives. Moreover, the frequency of repair of technologies that do not yet exist in
the fleet, or are only emerging today provides insufficient representation of what they will be in
the future with wider penetration of those technologies. As a result, while the ICMs include costs
to cover warranty repairs, we do not consider any additional repair costs as a result of our GHG
standards. This is consistent with EPA's approach in both the 2010 and 2012 FRMs and the Draft
TAR.
2.3.2.4 Costs Updated to 2015 Dollars
EPA is using technology costs from many different sources. These sources, having been
published in different years, present costs in different year dollars (e.g., 2009 dollars or 2012
dollars). For this analysis, EPA sought to have all costs in terms of 2015 dollars to be consistent
with the dollars used by EIA in its Annual Energy Outlook 2016. These values are updated from
the Draft TAR which expressed costs in 2013 dollars. While the factors used to convert from
20013 dollars (or other) to 2015 dollars are small, EPA prefers to be overly diligent in this regard
to ensure consistency across our analyses. EPA has used the GDP Implicit Price Deflator for
Gross Domestic Product as the converter, with the actual factors used as shown in Table 2.44.
2-230
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.44 Implicit Price Deflators and Conversion Factors for Conversion to 2015$
Calendar Year ->
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Implicit Price
Deflators for Gross
Domestic Product
94.814
97.337
99.246
100
101.221
103.311
105.214
106.913
108.828
109.998
Factor applied to
convert to 2013$
1.160
1.130
1.108
1.100
1.087
1.065
1.045
1.029
1.011
1.000
Source: Bureau of Economic Analysis, Table 1.1.9 Implicit Price Deflators for Gross Domestic Product; last revised
on September 29, 2016; accessed on 10/29/2016 at www.bea.gov.
2.3.3 Approach for Determining Technology Effectiveness
In the Draft TAR, EPA reevaluated the effectiveness values for all technologies discussed in
the MYs2017-2025 light duty GHG Final Rulemaking (FRM), as well as prominent technologies
that have emerged since then. Along with the vehicle benchmarking and full vehicle simulation
process, EPA reviewed available data including the 2015 LD National Academy of Sciences
report535, confidential manufacturer estimates, automaker and supplier meetings, technical
conferences, literature reviews, and press announcements regarding technology effectiveness.
For this Proposed Determination EPA has again reevaluated the effectiveness values used in the
Draft TAR based on new data and information obtained since then, and assessed the public
comments received on the Draft TAR. In most cases, multiple sources of information were
considered in the process of determining the effectiveness values used in this Proposed
Determination.
Full vehicle simulation modeling has been used in previous light-duty greenhouse gas rules
and in the Draft TAR to establish the effectiveness of technologies, and is regularly applied by
vehicle manufacturers, suppliers, and academia to evaluate and choose alternative technologies
to improve vehicle efficiency. In the 2015 NAS report,535 the committee recognized the
important contribution of full vehicle simulation and lumped parameter modeling in these
previous rulemakings, and recommended continued use of these methods as the best way of
assessing technologies and the combination of technologies.
For this Proposed Determination as in the Draft TAR, EPA is employing its own full vehicle
simulation model: the Advanced Light-duty Powertrain and Hybrid Analysis tool (ALPHA). The
ALPHA model has been developed and refined over several years and used in multiple
rulemakings to evaluate the effectiveness of vehicle technology packages. The same base model
used in the LD ALPHA model was also used in the GEM model for the HD Phase 1 and HD
Phase 2 rulemakings. See 81 FR 73530-549 (Oct. 25, 2016. Using ALPHA improves the
transparency of the process and provides additional flexibility to allow consideration of the most
recent technological developments and vehicle implementations of technologies. Input data for
the ALPHA model has been created largely through benchmarking activities. Benchmarking is a
commonly used technique that is intended to create a detailed characterization of a vehicle's
operation and performance. For the purposes of developing ALPHA, and for establishing overall
technology effectiveness, EPA performed many benchmarking activities including measuring
vehicle performance over the standard emission cycles and measuring system and component
performance on various test stands.
2.3.3.1 Vehicle Benchmarking
2-231
-------
Technology Cost, Effectiveness, and Lead Time Assessment
As part of its mandated evaluation of the appropriateness of the MY2022-2025 standards,
EPA is re-assessing any potential changes to the cost and the effectiveness of advanced
technologies available to manufacturers. See section 86.1818-12 (h)(i), (ii), and (iii).
Benchmarking is a process by which detailed vehicle, system, and component performance is
characterized. Benchmarking is commonly used by vehicle manufacturers, automotive suppliers,
national laboratories, and universities in order to gain a better understanding of how vehicles are
engineered and to create large datasets that can be applied in modeling and other analyses. In its
effort to assess light-duty vehicles in preparation for the MTE, EPA has benchmarked over
twenty commercially available vehicles that represent a diverse cross section of the current light-
duty fleet, with the results summarized in 15 peer-reviewed SAE papers.536 537 As the result of
these activities, EPA has calibrated the ALPHA full vehicle simulation model and applied the
results of this model to establish and confirm technology effectiveness. In addition, EPA has
been able to capture the performance of current vehicles, which is an important goal of the MTE.
The performance measurements not only include greenhouse gas emissions and fuel economy,
but also account for the additional fuel consumption associated for noise, vibration and harshness
(NVH), drivability and criteria emissions controls.
The ALPHA model has been used to confirm and update, where necessary, efficiency data
from the previous studies, such as from advanced downsized turbo and naturally aspirated
engines. It is also being used to quantify effectiveness from advanced technologies that the
agencies did not project to be part of a compliance pathway during the FRM, such as
continuously variable transmissions (CVTs), multi-mode normally aspirated engines, and clean
diesel engines. The ALPHA model accounts for synergistic effects between technologies and has
been used by EPA to calibrate the Lumped Parameter Model to incorporate the latest technology
package effectiveness data into the OMEGA compliance model. This process allows EPA to
simulate technology combinations (packages) that may not yet exist in the fleet.
To simulate drive cycle performance, the ALPHA model requires various vehicle input
parameters, including vehicle inertia and road loads, and component efficiencies and operations.
Vehicle benchmarking is the detailed process for obtaining these parameters.
2.3.3.1.1 Detailed Vehicle Benchmarking Process
The following discussion describes the vehicle benchmarking elements used as required for
the vehicles tested by EPA leading up to the Proposed Determination. The vehicle benchmarked
in this example is a 2013 Chevy Malibu 1LS as detailed in Table 2.45. This vehicle was chosen
as representative of a midsize car with a typical conventional powertrain with a naturally
aspirated engine and a 6-speed automatic transmission. The first task of the vehicle
benchmarking process involved collecting data from on-road and dynamometer testing (Figure
2.69) before removing the engine and transmission for separate component testing. Major
components such as the engine and transmission of a vehicle must be isolated and evaluated
separately to create accurate performance maps to be included in the ALPHA model.
Table 2.45 Benchmark Vehicle Description
Model
2013 Chevy Malibu 1LS
Engine
2.5L inline-4, GDI, naturally aspirated
Powertrain
Conventional FWD 6-speed automatic, GM6T40 transmission
Gear Ratios
4.584, 2.965,1.912,1.446,1.000, 0.746 with 2.89 final drive
Tire Size
215/60/R16
2-232
-------
Technology Cost, Effectiveness, and Lead Time Assessment
EPA Label Fuel Economy
22 City, 34 Highway, 26 Combined MPG
Emissions Equivalent Test Weight (ETW)
4,000 lbs (1814 kg)
Emissions Target Road Load A
38.08 lbs (169.4 N)
Emissions Target Road Load B
0.2259 Ibs/mph (2.248 N/m/s)
Emissions Target Road Load C
0.01944 lbs/mphA2 (0.4327 N/(m/s)A2)
Fuel Economy ETW
3,625 lbs (1644 kg)
Fuel Economy Target Road Load A
28.62 lbs (127.3 N)
Fuel Economy Target Road Load B
0.1872 Ibs/mph (1.863 N/m/s)
Fuel Economy Target Road Load C
0.01828 lbs/mphA2 (0.4069 N/(m/s)A2)
Figure 2.69 Chevy Malibu Undergoing Dynamometer Testing
2.3.3.1.1.1 Engine Testing
The engine was removed from the vehicle and installed in an engine dynamometer test cell, as
shown in Figure 2.70. The complete vehicle exhaust and emission control systems were included
in the test setup. All necessary signals including the transmission input and output shaft speed
signals were supplied by the test stand to prevent engine controller fault codes. The engine was
fully instrumented to collect detailed performance information (e.g., exhaust/coolant
temperatures, cam angles, throttle position, mass airflow).
2-233
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.70 Engine Test Cell Setup
The engine fuel consumption was measured at the steady state torque and speed operating
points as shown in Figure 2.71.
Engine Map Points
300
250
200
150
100
50
0
-50
-100
0
1000
2000
3000
Speed (RPM)
4000
5000
6000
7000
Figure 2.71 Engine Map Points
2.3.3.1.1.2 Transmission Testing
The 6-speed automatic transmission was removed from the vehicle and installed on a test
stand as shown in Figure 2.72. The transmission control solenoid commands were reverse
engineered and the transmission was manually controlled during testing. Transmission line
pressure was externally regulated to 5 and 10 bar. Torque and speed were measured at the input
of the transmission and both outputs. The input to the transmission was driven by an electric
motor.
2-234
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.72 GM6T40 Transmission during Testing
The transmission losses were measured at input torques ranging from 25 to 250 Nm and input
speeds ranging from 500 to 5000 RPM. For efficiency testing, the torque converter clutch was
fully locked by manually overriding the clutch control solenoid. Tests were performed at two
transmission oil temperatures, 37 C and 93 C, and two line pressures, 5 and 10 bar. Total
efficiency for each gear during operation at 93 C, including pump and spin losses, is shown in
Figure 2.73.
Transmission Total Efficiency -- All Gears, 93C 10bar
0.5
0.4
Input Speed (RPM)
Input Torque (Nm)
Figure 2.73 Transmission Efficiency Data at 93 C and 10 Bar Line Pressure
The torque converter was tested unlocked in 6th gear to determine speed ratio (SR), K
factorBBB and torque ratio curves. The input speed to the transmission was held at 2000 RPM
SB® K-faC[0r is approximately equal to stall_speed_rpm/square_root(stall_torque_Nm).
2-235
-------
Technology Cost, Effectiveness, and Lead Time Assessment
while decreasing the output speed to traverse the SR curve from 1.0 to 0.35 (limited due to line
pressure and transmission slip). The data below SR 0.35 was extrapolated using the higher SR
data. The torque converter data is shown in Figure 2.74, with the K factor curve normalized by
dividing by the K factor at SR 0 (torque converter stall). Normalizing the K factor curve allows
for scaling the curve up or down by multiplying by a new stall K value.
Torque Ratio and Normalized K Factor
1.9
Torque Ratio
K Factor (norm)
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1
0.9
0
0.2
0.3
0.4 0.5 0.6
Speed Ratio
0.7
0.9
Figure 2.74 Torque Converter Torque Ratio and Normalized K Factor versus Speed Ratio
Transmission spin losses were measured in each gear with a locked torque converter and no
load applied to the output shaft while varying the input speed from 500 RPM to 3000 to 5000
RPM depending on the chosen gear. Spin loss testing was performed at 5 bar and 10 bar line
pressures and 37 C (cold) and 93 C (operating) oil temperatures. Figure 2.75 shows the spin loss
data at 93 C for all gears and both line pressures.
2-236
-------
Technology Cost, Effectiveness, and Lead Time Assessment
CL
W
1000 1500 2000 2500 3000 3500 4000 4500 5000
Input Speed (RPM)
Transmission Spin Loss Data - All Gears, 93C
dashed = 5bar
solid = 10bar
Figure 2.75 Transmission Spin Losses at 93C
2.3.3.1.2 Development of Model Inputs from Benchmarking Data
After compiling the raw data, it was necessary to adapt the data to a form suitable for use by
the ALPHA model, including filling any data gaps and interpolating or extrapolating as required.
2.3.3.1.2.1 Engine Data
For use with the ALPHA model, the engine's fuel consumption map was created by
converting the set of points to a rectangular surface. In addition, an estimate of the engine inertia
was required since it plays a significant role in the calculation of vehicle performance and fuel
economy.538 The resulting engine data was reviewed with manufacturers prior to use in the
ALPHA model.
2.3.3.1.2.2 Engine Map
Figure 2.76 shows one of the engine maps generated from the test stand data in terms of
brake-specific fuel consumption (BSFC) in g/kW-hr.
2-237
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Che\ty Malibu 2.5L BSFC Map
170 kW
250
150 kW
130 kW
200
110 kW
| 150
90 kW
(D
3
2"
O
I—
¦246
70 kW
100
50 kW
n V \ i
30 kW
10 kW
5 kW
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
6000
Speed (RPM )
Figure 2.76 Chevy Malibu 2.5L BSFC Map
2.3.3.1.2.3 Inertia
Engine inertia plays a significant role in vehicle performance and fuel economy, particularly
in the lower gears due to the high effective inertia (proportional to the square of the gear ratio)
and higher acceleration rates.
To estimate the combined inertia of the engine, its attached components, and the torque
converter impeller, a simple test was performed in-vehicle: the engine was accelerated with the
transmission in park to the engine's maximum governed speed, then the ignition was keyed off,
and the engine speed and torque were observed until the engine stopped. Engine speed and
reported engine torque data (shown as negative during ignition off) were collected. The data was
then run through a simple simulation and the inertia varied until the model deceleration rate
reasonably matched the observed deceleration rate down to 500 RPM. Figure 2.77 shows the
model result using a 0.2 kg-mA2 total inertia with the engine drag torque.
2-238
-------
Technology Cost, Effectiveness, and Lead Time Assessment
CL
£
"O
0
0
Q.
in
4000
3500
3000
2500
2000
1500
1000
500
Malibu 2.5L Spindown Inertia Test
¦20
CAN Speed
Model Speed
CAN Torque
1 1.5
Time (Seconds)
Figure 2.77 Engine Spin down Inertia Test
An oil-filled torque converter from the 2013 Malibu was weighed and measured to estimate
its inertia. The weight of 12.568 kg and total diameter of 0.273 m gave an estimated 0.0585 kg-
mA2 total inertia. For the purposes of modeling this inertia was then proportioned 2/3 for the
impeller side and 1/3 for the turbine side based on the inertia split from other known torque
converters.
Subtracting the estimated torque converter inertia results in an engine inertia (including all
attached components) of approximately 0.161 kg-m2 (0.2 - 2/3*0.0585).
The exact proportioning of the inertia makes little difference to the outcome of the model
(since the total inertia is always the same) but can guide future work or estimates of component
inertias.
2.3.3.1.2.4 Transmission Data
For use with the model, the total transmission efficiency data needed to be separated into gear
efficiency and pump/spin torque losses. Torque converter back-drive torque ratio and K factor
also needed to be calculated.
2.3.3.1.2.5 Gear Efficiency and Spin Losses
To separate the gear efficiency from the total efficiency (which includes the pump/spin
losses), the total efficiency data for each gear was converted to torque loss data and the spin loss
torques were subtracted. The resulting gear torque loss data was then converted to efficiency
lookup tables. Some data points had to be extrapolated to cover the full speed and/or torque
range. For example, first gear was only tested to 150 Nm but the full table required data up to
2-239
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.78 Gear Efficiency Data at 93 C and 10 bar Line Pressure
250 Nm. Figure 2.78 shows the estimated gear efficiencies for all gears. This process was
followed for both the 37 C and 93 C data at 5 and 10 bar line pressure.
Transmission pump losses were factored out of the spin losses (as a rough approximation,
since no pump loss data was available), using the lowest common spin loss to represent the pump
loss.
Transmission Gear Efficiency -- All Gears, 93C 10 bar
~ 0.96
•£¦ 0.94
,| 0.92
O
w 0.9
0.88
2.3.3.1.2.6 Torque Converter
To complete the model inputs for the torque converter, the torque ratio and K factor need to
be calculated for the full range of speed ratios.
The torque converter back-drive torque ratio is assumed to be 0.98 for all speed ratios. The
back-drive K factor is calculated from the drive K factor mirrored relative to speed ratio (SR) 1
and shifted upwards by 70 percent. The K factor at SR 1 is calculated, for modeling purposes, as
7.5 times the highest drive K factor. In practice the K factor at SR 1 is either poorly defined or
near infinite so the model requires a large value but not so large as to make the solver unstable.
Figure 2.79 shows the given (SR < 0.95) and calculated torque converter data.
These additional data points have little effect on the modeled fuel economy but are required
for model operation and smooth transitions from positive to negative torques.
2-240
-------
Technology Cost, Effectiveness, and Lead Time Assessment
18
16
Torque Ratio and Normalized K Factor
—•— Torque Ratio
—»— K Factor (norm)
'
I
L
.
0.8 1 1.2
Speed Ratio
Figure 2.79 Torque Converter Drive and Back-Drive Torque Ratio and Normalized K Factor versus Speed
Ratio
2.3.3.1.3
Vehicle Benchmarking Summary
Section 2.3.3.1 outlined the vehicle benchmarking process for a typical vehicle. While
complex, this process yields the necessary input parameters for physics based full vehicle
simulation models such as ALPHA. The following list represents the main model input
parameters generated from the benchmarking process:
• Engine Maps:
° Fuel Consumption
° BSFC
° Friction/Inertia
Performance
Transmission Maps
Efficiency
Torque Converter
Shifting Strategy
Vehicle:
° Road Loads
° Mechanical Loads
° Electrical Loads
This information plus the remaining known vehicle characteristics (mass, etc.) provide the
model with all of the necessary information needed for simulation. During the initial
2-241
-------
Technology Cost, Effectiveness, and Lead Time Assessment
development of the ALPHA model, this complete data set from several vehicles was used to
validate all of the internal calculations of the model. Once the model was validated, a wide
variety of engines, transmissions, and other vehicle components were introduced to model
current and future vehicles. This process is described in Section 2.3.3.2.2.
2.3.3.2 Classification of Vehicles for Effectiveness
When applying technologies in this analysis, the most representative value for effectiveness
will depend on certain characteristics of each individual vehicle. As discussed in Section 2.3.1.4,
the effectiveness classes in the FRM and Draft TAR were derived from vehicle size
classifications defined by vehicle interior volume and gross vehicle weight rating attributes.
While overall this approach placed similar vehicles together, stakeholder comments, including
those from FCA, on the Draft TAR highlighted examples where some dissimilar vehicles were
assigned the same technology effectiveness values. For this Proposed Determination, EPA has
refined the vehicle classification approach for assigning representative effectiveness values
according to the attributes of vehicle road load power and engine power-to-vehicle weight ratio
as described in this section. Comments received in response to the Draft TAR from the Auto
Alliance include a study by Novation Analytics, a contractor of the Auto Alliance. The report
recommends (and the Alliance concurs) that EPA account for "engine displacement and vehicle
load, which are first-order determinants of powertrain efficiency," 539 when determining
technology effectiveness. The report further recommends the use of "displacement specific load"
(i.e., the ratio of totalized vehicle load over the cycle and engine displacement) as a metric within
the LPM to determine technology effectiveness.
As described in more detail in Appendix A, EPA disagrees with many of the conclusions
drawn by the Alliance's contractor. However, EPA does agree that the ratio of engine size to
vehicle load has a primary influence on powertrain efficiency, and thus technology effectiveness.
This is because the combination of engine sizing and vehicle load affects the speed and BMEP at
which the engine operates, and thus the engine operational efficiency over the test cycles. The
following subsections explain the significance of engine sizing and power-to-weight ratio, and
how EPA has accounted for the ratio of engine size to vehicle load in the ALPHA simulations.
2.3.3.2.1.1 Significance of Power-to-Weight Ratio and Road-Load Power Attributes
Total vehicle load consists of multiple components; chiefly inertial loads (a function of ETW
over the test cycles), and aerodynamic and rolling resistance loading (together covered as "road
loads"). Different combinations of road and inertial loads may lead to the same totalized vehicle
load over a cycle, but different instantaneous engine operation points (and potentially different
average efficiency). However, in practice, inertial loads and road loads tend to be correlated with
each other and with vehicle size. Thus, it is appropriate to consider maximum-engine-power-to-
ETW ratio ("power/weight ratio" as a shorthand) as a primary influence on powertrain efficiency
and road-load power a secondary effect, rather than considering vehicle-load-to-engine-power
ratio as a primary influence and road load to inertial load ratio as a secondary effect.
To estimate the magnitude of the effect of changing vehicle power/weight ratio on powertrain
efficiency and technology effectiveness, EPA used its ALPHA full vehicle simulation model to
determine changes in CO2 emissions when different size engines were incorporated into a
standard vehicle. Recognizing that changing engine size also affects vehicle performance,
ALPHA was also used to simulate acceleration times. Finally, to examine effectiveness (i.e., the
2-242
-------
Technology Cost, Effectiveness, and Lead Time Assessment
change in CO2 when advanced technology is implemented), the same power/weight ratio study
was performed with powertrains containing different technologies.
The baseline vehicle modeled was a standard car (similar to a 2008 Toyota Camry), with a
159 HP PFI engine, five-speed transmission, Camry road loads, and 3500 pound ETW. CO2
emissions over the FTP and HWFET cycles and acceleration times were simulated within
A LPHA. The results for this simulation were two-cycle combined CO2 emissions of 282 g/mile,
an estimated 0-60 time of 8.05 seconds, and a "performance sum" of 0-60, 30-50, 50-70, and 1/4-
mile times of 35.5 seconds (see Section 2.3.1.2).
The engine efficiency in this particular simulation is represented in Figure 2.80. The figure
shows a two-cycle engine "heat map" from the standard car simulation, plotting the speeds and
torques where the engine operates over the FTP and FTWFET on an engine efficiency map.
Points where the engine spends more operational time are plotted in red, points where it spends
less time are plotted in cooler colors (blue, green), and points where there is no engine operation
remain white. The pink line is the line of best efficiency at each power.
Figure 2.80 Engine "heat map" for baseline vehicle, showing engine operation over the FTP and HWFET.
Figure 2.80 shows a red "hot spot" of engine operation near 70 Nm, with extended operation
down to 15-20 Nm and up to 150 Nm. Almost the entire operational range occurs at torques
lower than the line of best efficiency. This represents a somewhat "typical" vehicle, where two-
cycle engine operation and engine efficiency are not well matched.
2.3.3.2.1.2 Effect of Changing Power-to-Weight Ratio
To examine the effect of the performance-fuel economy tradeoff, the baseline case was
altered by changing the engine size (and thus maximum power) in 2 percent increments from 60
percent to 200 percent of the baseline case, which resulted in maximum engine horsepower
ranging from about 100 F1P to about 300 HP, Other vehicle characteristics (including ETW) were
held constant, resulting in a maximum-engine-power-to-ETW ranging from about 0.03 HP/lb to
about 0.09 HP/lb. For vehicles with each engine size, performance metrics and CO2 emissions
over the FTP and HWFET cycles were simulated using ALPHA as in the baseline case.
0
1000 2000 3000 4000 5000 6000
Speed (RPM)
2-243
-------
Technology Cost, Effectiveness, and Lead Time Assessment
As an example of the effect of varying engine size (i.e., varying power/weight ratio), Figure
2.81 shows two-cycle engine "heat maps" for both a very high power/weight ratio vehicle (0.09
HP/lb) with a large engine, and a very low power/weight ratio vehicle (0.03 HP/lb) with a small
engine. In both cases, the operational combinations of speed and torque are approximately the
same as shown in the baseline case in Figure 2.80 (note the red hot spot at about 70 Nm on all
three heat maps). This is because the required speed and torque are driven by the vehicle weight
and road loads, which in this simulation remain identi cal. However, the peak torque of the large
engine is about 440 Nm, and the small engine only 132 Nm, and thus the larger engine operates
in much lower BMEP areas of the map.
•90 kW
120
•80 kW
100
70 kW
•60 kW
E
z
•50 kW
•40 kW
H
20 kW
10 kW
5 kW
1000
5000
2000
3000
4000
Speed (RPM)
6000
(a) High power/weight ratio (-0.09 HP/lb) (b) Low power/weight ratio (-0.03 HP/lb)
Figure 2.81 Two-cycle heat maps for two different power/weight ratio vehicles.
The difference in operation means that the high power/weight ratio vehicle operates much
farther from the line of best efficiency (the pink line in Figure 2.81). Conversely, the low
power/weight ratio vehicle operates closer to the line of best efficiency; in other words, the
smaller engine is better matched to the efficiency "sweet spot." Although these graphs were
generated by changing engine size only, in general the match between engine efficiency and
operation is a function of the ratio between engine size and vehicle loads. Engine map operation
area of vehicles with similar power/weight ratios would be expected to be very similar,
regardless of absolute scale.
The effect of this engine matching on CO2 emissions and vehicle acceleration performance is
illustrated in Figure 2.82, which shows the trends in combined cycle CO2 emissions and the
"performance time sum" of 0-60, 30-50, 50-70, and 1/4-mile acceleration times as a function of
power/weight ratio. Although this performance time sum was chosen for reasons detailed in
Section 2.3.1.2, other performance metrics show a similar trend to that exhibited in Figure 2.82.
;210kW
90 kW
300
: 250
r 200
150
1000 2000 3000 4000 5000 6000
Speed (RPM)
2-244
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Combined C02
£ 300
40
35
0.05 0.06 0.07
Engine power/RL ratio (HP/lb)
Figure 2.82 CO2 and performance time sum as a function of power/weight ratio.
As power/weight ratio increases, acceleration times decrease and CO2 emissions increase, as
would be expected from the heat maps depicted in Figure 2.81. The trends in both CO2 emissions
and acceleration performance are monotonic over the range of power/weight ratios shown. As
such, the acceleration times and combined cycle CO2 emissions in Figure 2.82 can be directly
compared, as shown in Figure 2.83, to create a "trade-off curve, demonstrating how engine
power (i.e., displacement) can be altered to increase performance at the expense of fuel
economy, or to increase fuel economy at the expense of performance. More advanced technology
powertrains would be expected to move the curve closer to the origin, as noted by the arrow.
Performance / fuel economy tradeoff
Advanced
technology
Performance time sum (sec)
Figure 2.83 CO2 as a function of acceleration performance time sum.
2.3.3.2.1.3 Effect of Advanced Technologies
More advanced technologies produce lower CO2 for the same performance, and so would be
expected to move the trade-off curve shown in Figure 2.83 closer to the origin, reducing CO2
2-245
-------
Technology Cost, Effectiveness, and Lead Time Assessment
emissions, acceleration time, or both. To explore this, ALPHA simulations were run using
different powertrains, but the same vehicle road and inertia loads as before. The powertrains
simulated were a 2013 GDI engine (similar to that found in the 2013 Chevrolet Malibu) paired
with a six-speed transmission, and a 24-bar turbo downsized engine (as modeled by Ricardo and
included in the FRM) paired with an eight-speed transmission. Two-cycle CO2 emissions and
acceleration performance were calculated.
Figure 2.84 compares the heat map for the nominally sized PFI engine with 5-speed
transmission (identical to the simulation result shown in Figure 2.80) with the future 24-bar turbo
downsized engine paired with an eight-speed transmission. The powertrains are sized to have
similar acceleration performance, and thus have slightly different maximum power ratings.
(a) PFI engine heat map (shown in Figure 2.80) (b) 24 bar turbo downsized engine heat map
Figure 2.84 Engine Heat maps for the baseline PFI engine and a 24-bar turbo downsized engine
The future 24-bar turbo downsized engine has a higher peak efficiency (over 36 percent
compared to over 34 percent), which contributes to a higher effectiveness. In addition, it also has
a peak efficiency zone that extends to lower speeds and loads than that in the PFI engine. The
lower peak efficiency zone results in a better match between vehicle loading and powertrain
efficiency. In particular, the PFI engine has a hot spot which is substantially lower than the line
of highest efficiency, while the turbo downsized engine has a hot spot located almost directly on
the line of highest efficiency.
The quality of the match between powertrain efficiency and vehicle road load is a function of
vehicle power/weight ratio. To investigate this effect, engine sizes for the GDI and TDS engines
were again swept in 2 percent increments, and two-cycle CO2 emissions and acceleration
performance were simulated. Engine heat maps for high and low power/weight ratio turbo
downsized powertrains (roughly equivalent in acceleration performance to the PFI powertrains
illustrated in Figure 2.83) are shown in Figure 2.85.
2-246
-------
Technology Cost, Effectiveness, and Lead Time Assessment
400
275 kW
120
kW
250 kW
350
70 kW
100
225 kW
300
200 kW
60 kW
175 kW -p
E 250
50 kW
150 kW —
125 kW g"
i—
100 kW
£200
40 kW
150
30 kW
40
75 kW
100
20 kW
50 kW
10 kW
5kW
25 kW
12.5 kW
1000
2000
3000
4000
Speed (RPM)
5000
6000
1000
2000
3000
4000
Speed (RPM)
5000
6000
(a) High power/weight ratio (b) Low power/weight ratio
Figure 2.85 Engine operation heat maps for the turbo downsized engine with eight-speed transmission.
As expected, the shape and extent of the heat maps in Figure 6 are roughly equivalent with
respect to speed and torque, but the low power/weight ratio engine operates at higher BMEP.
However, unlike the PFI powertrains, the low power/weight ratio powertrain has a hot spot that
is above the line of best efficiency, indicating that the match between vehicle and powertrain for
the smaller engine is no better (and may actually be worse) than that of the nominally sized
powertrain shown in Figure 2.84.
2.3.3.2.1.4 Advanced Technology Trade-Off Curves
The relocation of the peak engine efficiency means that not only do these advanced
powertrains have a trade-off curve that is closer to the origin than the one shown for the PFI
engine in Figure 2.83, but these curves also have a different slope, as shown in Figure 2.86,
where the "tradeoff curves" for the advanced technologies are progressively flatter than for the
PFI engine. These trade-off curves are presented as a function of acceleration times rather than
power/weight ratio so that the resulting comparisons are performance neutral.
400
380
360
340
O
% 320
300
QD
"S 280
| 260
o
° 240
220
200
180
Figure 2.86 CO2 as a function of performance time sum for PFI, GDI, and turbo downsized engines.
PFI engine
GDI Engine
TDS engine
26 30 34 38 42 46 50
Performance time sum (sec)
2-247
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The variation in slope of the lines in Figure 2.86 suggests two things. First, the more advanced
technologies have flatter curves, with the most advanced technology (the 24 bar turbo downsized
engine) even exhibiting a point of minimum CO2 emissions. Although this simplified ALPHA
analysis does not include real-world effects such as the additional mass associated with larger
engines, this trend is still likely to hold in the vehicle fleet, as the changes in engine maps and
powertrain matching due to the implementation of advanced technology still fundamentally alter
the relationship between engine size and CO2 emissions.
A study by Novation Analytics, commissioned by the Auto Alliance and included with its
comments on the Draft TAR, supports this conclusion. Using a simulation of turbo downsized
engines, they state, "where a powertrain is already operating at a high specific load [i.e., in a low
power/weight ratio vehicle], further downsizing may offer little benefit as both scenarios are
already operating in the high efficiency region where the relative gains from a further increase in
specific load (via smaller displacement) are minimal."540
Therefore, the relationship between CO2 emissions and acceleration performance can vary
substantially as more advanced technology is implemented. Consequently, any tradeoff
relationship between these factors developed using less advanced technology engines (such as
the PFI engine curve in red in Figure 2.86) should not be expected to hold when more advanced
technology is implemented. To the contrary, increasing performance using advanced engines
should have a much decreased effect on fuel consumption when compared to less advanced
engines.
In addition, Figure 2.86 shows that the potential reduction in CO2 emissions from advanced
technology powertrains is a function of vehicle performance, and thus power/weight ratio.
Visually, it can be seen that the combined CO2 emissions reduction from the GDI engine (orange
line) compared to the PFI engine (red line) clearly varies from around 20 percent with higher
performance vehicles to nearly 0 percent in lower performance vehicles.
There is also a difference in CO2 reduction between the GDI engine and the turbo downsized
engine (the green line in Figure 2.86). This difference is calculated in Figure 2.87, which shows
the reductions in CO2 emissions obtainable from the 24 bar turbo downsized engine with eight-
speed transmission, compared to that of a GDI engine with six-speed transmission having similar
acceleration performance. The comparisons between powertrains are done on a performance
neutral basis, matching vehicles with the same acceleration performance time sum, and so
reductions in CO2 are shown as a function of performance time sum. Approximate power/weight
ratio is also given in the figure as a reference.
2-248
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Approximate power / weight ratio (HP/lb)
0.083 0.063 0.051 0.042 0.036 0.032 0.028 0.025
in 30%
O
c 25%
c
¦2 20%
u
3
"S15%
§ 10%
QJ
5%
0%
22 26 30 34 38 42 46 50
Performance time sum (sec)
Figure 2.87 Reduction in CO2, comparing a turbo downsized engine to a GDI engine with similar
acceleration performance.
As seen in Figure 2.87, the potential percent reduction in CO2 emissions is a function of
performance (and thus the vehicle power/weight ratio), where vehicles with relatively small
engines (on the right hand side of Figure 2.87) have less potential reduction than vehicles with
relatively large engines (on the left hand side of Figure 2.87). Across the relatively wide range of
power/weight ratios shown in Figure 2.87, this change in effectiveness is quite substantial.
This study shows that advanced technology engines change the match between engine
efficiency and engine operation (as shown in Figure 2.84). Because this match is also affected by
vehicle loading, the ratio of engine size to vehicle load (i.e., the vehicle power/weight ratio) is a
primary influence on powertrain efficiency and technology effectiveness. Additionally, and for
the same basic reason, the tradeoff between CO2 emissions and performance changes as
advanced technology is introduced, with more advanced technology packages generally tending
to have flatter tradeoff curves.
2.3.3.2.2 Definition of Effectiveness Classes
Because technology effectiveness is clearly related to engine operation through power-to-
weight ratio, EPA agrees with the Auto Alliance that calculation of technology effectiveness
values should be tied to engine power and vehicle loads. Because total vehicle load is composed
of both inertial load and road loads (primarily tire rolling and aerodynamic loading), which are
only loosely correlated, EPA has looked at vehicle loads two-dimensionally, through both
power-to-weight ratio and road load horsepower.
For this Proposed Determination EPA has revised the classification approach used for
applying effectiveness values from the size-based classification used in the FRM and Draft TAR
to an approach that is based on vehicle power-to-weight and road load characteristics. Each
vehicle in the MY2015 baseline fleet has been assigned to one of the ALPHA classes shown in
Table 2.32 using the procedure described in this Section.
In the first step, because vehicles with high capacity for towing and hauling have road load
and power-to-weight characteristics that are fundamentally different from other passenger
vehicles, vehicles defined as 'pickup trucks' under 40 CFR § 600.315-08 were assigned to the
'Truck' ALPHA class. The remaining vehicles were divided into low, medium, and high power-
TDS engine
2-249
-------
Technology Cost, Effectiveness, and Lead Time Assessment
to-weight levels using the MY2015 production volume proportions defined in Figure 2.88. The
production volumes for plug-in vehicles (PHEVs and BEVs) were not included in the percentile
calculations for power-to-weight ratios.
Next, the distribution of road load horsepower values was investigated within each of these
power-to-weight categories. As can be seen in Figure 2.89, both the 'low' and 'mid' power-to-
weight categories exhibit bimodal distributions, with vehicles clustered in two groups below and
above the median values of road load horsepower. The vehicles that comprise these two groups
tend to correspond to cars having lower road loads, and sport-utility vehicles and vans and
having higher road loads. However, there are some examples where vehicles in different market
segments are now classified together. This is appropriate, since for example a sedan with a large
frontal area and high tire rolling resistance would tend to have technology effectiveness benefits
more like a cross-over utility vehicle than like other cars. In recognition of the relatively broad
and bimodal distributions of road load horsepower, these two power-to-weight categories are
further subdivided into 'low' and 'high' road load horsepower levels as shown in Table 2.46.
Table 2.46 Criteria for Classifying Vehicles by Power-to Weight ratio and Road Load Horsepower
Power-to-Weight
Road Load Horsepower at 50mph
Level
Percentile
Range
Cutoff Values (hp/lb ETW)
Level
Percentile
Range
Cutoff Values (hp)
Lower
Upper
Lower
Upper
Low
0 to 40
-
0.049
Low
0 to 50
-
11.8
High
50 to 100
11.8
-
Mid
40 to 80
0.049
0.061
Low
0 to 50
-
14.0
High
50 to 100
14.0
-
High
80 to 100
0.061
-
-
-
-
-
Note: Power-to-Weight percentiles are production volume-based after excluding PEVs and pickup trucks. Road Load
Horsepower percentiles are defined within 'Low' and 'Mid' Power-to-Weight groups.
1
0.9
0.8
¦£ 0.7
S 0.6
o>
¦I 0.5
_OJ
| 0.4
| 0.3
u
0.2
0.1
0
0.02 0.04 0.06 0.08 0.1 0.12 0.14
Power-to-Weight Ratio (hp/lb ETW)
equal to, or above value
Figure 2.88 Production Volume Distribution of Power-to-Weight Ratios in MY2015 Fleet
All Vehicles (excl. Trucks/EVs/PHEVs)
Truck
2-250
-------
Technology Cost, Effectiveness, and Lead Time Assessment
All Vehicles
(excl. Trucks/EVs/PHEVs)
Truck
5 10 15 20 25 30
Road Load Horsepower at 50mph
equal to, or above value
Figure 2.89 Production Volume Distribution of Road Load Horsepower at 50mph in MY2015 Fleet
The six vehicle classes that result from the process described above are shown in Table 2.47
along with the important production volume-weighted average characteristics that exemplify the
power-to-weight and road load characteristics of each class. These values define an exemplar
vehicle for each of the classes, referred to interchangeably as an 'ALPHA Class' or 'Effectiveness
Class', the characteristics for each of which are used in the ALPHA model as described in
Section 2.3.3 for the purpose of developing representative effectiveness values of technologies
added to vehicles in the MY2015 baseline fleet.
Table 2.47 Characteristics of Exemplar Vehicles for the Six ALPHA Classes
ALPHA Class
Abbreviation
Engine
Rated
Power (hp)
A coeff.
(Ibf)
B coeff.
(Ibf/mph)
C coeff.
(Ibf/mph2)
ETW
(lbs)
Low Power-to-
Weight, Low
Road Load
LPW_LRL
137.5
26.56
0.0630
0.01879
3257
Mid Power-to-
Weight, Low
Road Load
MPWJ.RL
191.1
32.27
0.0754
0.01993
3626
High Power-to-
Weight
HPW
313.8
35.76
0.3414
0.02086
4401
Low Power-to-
Weight, High
Road Load
LPW_HRL
172.4
34.95
0.0875
0.02526
3855
Mid Power-to-
Weight, High
Road Load
MPW_HRL
275.2
39.30
0.3348
0.02721
4849
Truck
Truck
324.2
39.62
0.4641
0.03222
5303
2.3.3.2.3 Comparison to Draft TAR Classification Approach and Exemplar Vehicles
2-251
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The power, weight, and road load attributes assumed for effectiveness modeling in the FRM
and Draft TAR were based on the characteristics of six typical, actual vehicles from the MY2007
to MY2010 time frame. While these assumptions were appropriate for the previous analyses, the
approach used for this determination results in exemplar vehicles with characteristics that are
more representative of vehicles in the fleet in MY2015. In particular, as shown in Table 2.48,
the power-to-weight ratios of the exemplar vehicles for this Proposed Determination are up to 20
percent higher than the corresponding vehicle classes in the Draft TAR, consistent with the
increases in engine power that have occurred over recent redesign cycles for many vehicles. In
addition, the slight road load horsepower decreases for four of the six classes is likely the result
of improvements in vehicle design such as improved aerodynamics.
Table 2.48 Change in Power-to-Weight and Road Load Horsepower of Exemplar Vehicles Relative to Draft
TAR
Changes in Exemplars for Proposed
Draft TAR Exemplar Vehicles (for reference only)
Determination relative to Draft TAR
ALPHA
Change in
Change in
Class*
Road Load
Power-to-
Vehicle Class
Engine
A
B coeff.
C coeff.
ETW
Horsepower
Weight
Rated
coeff.
(Ibf/mph)
(Ibf/mph2)
(lbs)
at 50mph
Ratio
Power
(Ibf)
LPW LRL
-5.8%
+1.0%
Small Car
109.7
24.68
0.1426
0.01984
2625
MPW LRL
+1.1%
+19.0%
Standard Car
158.2
29.80
0.1721
0.01860
3571
HPW
-5.1%
+14.2%
Large Car
249.8
45.20
0.2409
0.02135
4000
LPW HRL
-9.8%
+4.1%
Small MPV
171.8
27.64
0.4729
0.02493
4000
MPW HRL
+2.6%
+22.8%
Large MPV
208.0
32.43
0.4873
0.02566
4500
Truck
-15.5%
+18.0%
Truck
306.2
48.84
0.6104
0.03614
5911
*Note: The ALPHA Classes defined for this Proposed Determination are not intended to correspond one-to-one with
the Draft TAR, but are presented here to show the general trends in power-to-weight and road load horsepower.
In public comments, FCA cited examples from the Draft TAR where dissimilar vehicles were
assigned the same benefits, commenting that "...the Fiat 500 Turbo and the V6 Chrysler 300
AWD are assigned the same benefits for every technology. This is inappropriate given the
vehicle size, engine size, and drivetrain difference between them" (p. 35, FCA comments). EPA
has refined the ALPHA classifications for this Proposed Determination with the goal of
minimizing the variation within each class, particularly for the parameters of power-to-weight
and road load power. While EPA recognizes that the examples provided by the commenter were
meant to be simply illustrative, we note that using this revised vehicle classification approach,
the MY2015 Fiat 500 Turbo and V6 Chrysler 300 are now assigned to different ALPHA classes
for this Proposed Determination (MPW LRL and HPW, respectively.) More broadly, because
vehicles are now grouped using engine and vehicle road load characteristics, and each class
contains roughly equal sales-weighted volumes of vehicles, there will be a smaller range of
effectiveness values within each class. Furthermore, because the exemplar vehicles have been
updated to represent the sales-weighted average of characteristics within each ALPHA class,
there is a better match between the vehicles in the class and the associated exemplar.
In addition, in response to public comment, the effectiveness values calculated for each
vehicle for the final OMEGA runs were adjusted according to the vehicle's power to weight
ratio. For each vehicle classification, a set of effectiveness adjustment factors within ALPHA
2-252
-------
Technology Cost, Effectiveness, and Lead Time Assessment
was calculated by sweeping engine power for each technology. From these results, a linear
adjustment factor was calculated for each class and each technology. See Figure 2.23 for an
example, in this case the MPW HRL class, and Section 2.3.3.5.4 for adjustment parameters for
all ALPHA classes.
3%
Base GDI + 6spd
ATK2 + 6spd
ATK2 + 8spd
Adv ATK2
Adv ATK2 + RL
EGR24
EGR24 + RL
2%
1%
0%
u
-1%
-2%
0.05
0.055
0.06
0.065
0.07
Base Power/Wt (HP/lb)
Figure 2.90 MPW HRL Class Effectiveness Change as a Function of Power-to-Weight Ratio
As shown in Table 2.49 and Figure 2.91, the range of power-to-weight values within the high
and mid power-to-weight groups, in particular, is significantly smaller using the updated
classification approach. As can also be seen in Figure 2.91, the Draft TAR exemplar values for
power-to-weight, while appropriate for the FRM analysis conducted in 2012, are now generally
one standard deviation or more below the production-weighted average of the MY2015 fleet.
Table 2.49 MY2015 Summary Statistics of Power-to-Weight Ratio Using Draft TAR and Proposed
Determination Classification Approaches
Power-to-Weight Ratio (lOOxhp/lb)
PD Classification Approach
Draft TAR Classification Approach
ALH PA Class
Median
Std.
Dev.
min-max
N
Vehicle Class
Media
n
Std.
Dev.
min-max
N
LPW LRL
4.22
0.53
2.62-4.86
3,059,319
Small Car
4.15
0.71
2.62-6.40
833,737
MPW LRL
5.27
0.26
4.80-6.07
3,027,591
Standard Car
5.27
1.13
2.70-13.10
6,627,852
HPW
7.14
1.34
5.31-17.50
2,979,046
Large Car
9.35
2.18
4.80-17.50
394,688
LPW HRL
4.48
0.27
2.49-4.98
2,859,184
Small MPV
4.59
0.29
2.49-5.26
2,711,222
MPW HRL
5.69
0.33
4.90-6.29
2,896,808
Large MPV
5.84
0.66
3.65-10.31
4,334,273
Truck
6.28
0.95
3.74-8.56
1,786,224
Truck
6.39
0.84
3.83-8.56
1,706,401
Note: The ALPHA Classes defined for this Proposed Determination are not intended to correspond one-to-one with
the classes used in the Draft TAR, but are presented here to show the effect of the updated classification approach.
2-253
-------
Technology Cost, Effectiveness, and Lead Time Assessment
20
_Q
JZ
X
o
o
15
m 10
"oi
O)
o
CL
OL
ro
u
> ro
E
j CO
X
PD
Draft TAR
JL
: max
i-S-j
~
0+/-1S.D.
I
j min
X
O Exemplars
5 ^
0
Figure 2.91 MY2015 Production-weighted Distributions of Power-to-Weight Ratio Using Draft TAR and
Proposed Determination Classification Approaches
As can be seen in Figure 2.92, the updated exemplars for this Proposed Determination are
representative of the average MY2015 vehicle's road load horsepower in each class. In
comparison, the Draft TAR exemplar road load horsepower values tend to be higher than typical
MY2015 vehicles. Table 2.50 and Figure 2.92 show that the range of road load horsepower
values within each class is largely unchanged by updating the classification approach from the
Draft TAR to this Proposed Determination. One exception is the high power-to-weight class,
which has a greater range of road load horsepower values with the updated approach. While a
narrower range of road load horsepower values is preferable to a larger one, when defining the
classes EPA gave priority to minimizing within-class variation in power-to-weight ratio due to
the dominate influence that attribute has on effectiveness values. Overall, the road load
horsepower values in all classes, including the high power-to-weight class, are represented fairly
by the appropriate exemplar values.
Table 2.50 MY2015 Summary Statistics of Road Load Horsepower Using Draft TAR and Proposed
Determination Classification Approaches
Road Load Horsepower at 50mph
PD Classification Approach
Draft TAR Classification Approach
ALHPA Class
Avg.
Std. Dev.
min-max
N
Vehicle Class
Avg.
Std. Dev.
min-max
N
LPW LRL
10.3
0.8
8.3-11.7
3,059,319
Small Car
10.1
1.0
8.3-13.8
833,737
MPW LRL
11.1
1.3
9.3-14.6
3,027,591
Standard Car
11.1
1.4
8.7-19.2
6,627,852
HPW
14.2
2.7
9.8-26.1
2,979,046
Large Car
13.6
1.1
10.8-19.5
394,688
LPW HRL
13.6
1.1
10.6-23.8
2,859,184
Small MPV
13.6
1.1
11.4-17.9
2,711,222
MPW HRL
16.4
1.9
13.9-22.6
2,896,808
Large MPV
16.3
2.2
11.3-26.1
4,334,273
Truck
19.3
1.8
14.6-25.8
1,786,224
Truck
19.5
1.7
16.1-25.8
1,706,401
2-254
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Note: The ALPHA Classes defined for this Proposed Determination are not intended to correspond one-to-one with
the classes used in the Draft TAR, but are presented here to show the effect of the updated classification approach.
O
LD
+->
CD
CL
X
T3
CD
O
i
T3
CD
O
en
25
20
15
10
PD
~ I
I
X
Draft TAR
max
+/- 1S.D.
= mm
O Exemplars
$ £
CO
Q_
—I
Cd
—3=—
I
!
Q.
"Z
03
E
LO
£
Q.
2
QJ
go
3
0
Figure 2.92 MY2015 Production-weighted Distributions of Road Load Horsepower Using Draft TAR and
Proposed Determination Classification Approaches
2.3.3.3 ALPHA Vehicle Simulation Model
The Advanced Light-Duty Powertrain and Hybrid Analysis (ALPHA) tool was created by
EPA to evaluate the Greenhouse Gas (GHG) emissions of Light-Duty (LD) vehicles. In order to
have additional flexibilities and transparency, EPA developed an in-house full vehicle simulation
model that could freely be released to the public. Model development, along with the data
collection and benchmarking that comes along with model calibration, is an extremely effective
means of developing expertise and deeper understanding of technologies and their interactions.
Better understanding of technologies makes for more robust regulatory analysis. Having a model
available in-house allows EPA to make rapid modifications as new data is collected.
Throughout this section of the TSD, EPA has provided details on the major technology
assumptions built into ALPHA. EPA has also provided technical details in the Docket
describing the process used to build the fuel consumption maps for six of the engines mentioned
in this TSD, as well as data maps for two transmissions.541 In the time since the 2012 FRM, EPA
has published over 15 peer-reviewed papers describing results of key testing, validation and
analyses.
2-255
-------
Technology Cost, Effectiveness, and Lead Time Assessment
EPA began developing both light-and heavy-duty vehicle simulations simultaneously as these
vehicles share many of the same basic components. The light-duty vehicle model (ALPHA), and
the heavy-duty model (GEM), share the same basic architecture.ccc
EPA has validated the ALPHA model using several sources including vehicle
benchmarking,538 stakeholder data, and industry literature. While the ALPHA model continues
to be refined and calibrated, the version in use as of April 26, 2016 was externally peer
reviewed.542 To further enhance transparency, in May 2016, EPA published on the EPA website
the specific version of the ALPHA model that was reviewed. This package included the peer
review input data and runnable MatLab Simulink source code.
2.3.3.3.1 General ALPHA Description
ALPHA is a physics-based, forward-looking, full vehicle computer simulation capable of
analyzing various vehicle types with different powertrain technologies, showing realistic vehicle
behavior. The software tool is a MATLAB/Simulink based simulation.
Within ALPHA, an individual vehicle is defined by specifying the appropriate vehicle road
loading (inertia weight and coast-down coefficients) and specifications of the powertrain
components. Powertrain components (such as engines or transmissions) are individually
parameterized and can be exchanged within the model.
Vehicle control strategies are also modeled, including engine accessory loading, decel fuel
shutoff, hybrid behavior, torque converter lockup, and transmission shift strategy. Transmission
shifting is parameterized and controlled by ALPHAshift,543 a shifting strategy algorithm that
ensures an appropriate shifting strategy when engine size or vehicle loading changes. The control
strategies used in ALPHA are modeled after strategies observed during actual vehicle testing.
Vehicle packages defined within ALPHA can be run over any pre-determined vehicle drive
cycle. To determine fuel consumption values used to calculate LD GHG rule CO2 values, an FTP
and HWFET cycle are simulated, separated by a HWFET prep cycle as normally run during
certification testing. ALPHA does not include a temperature model, so the FTP is simulated
within the model assuming warm component efficiencies for all bags. Additional fuel
consumption due to the FTP cold start is calculated in post-processing by applying a fuel
consumption penalty to bags 1 and 2, depending on the assumed warmup strategy. Any vehicle
drive cycle can be defined and fuel economy simulated in ALPHA. For example, the results from
the US06, NEDC, and WLTP cycles (among others) are used to tune vehicle control strategy
parameters to match simulation results to measured vehicle test results across a variety of
conditions. In addition, performance cycles have been defined, which are used to determine
acceleration performance metrics.
2.3.3.3.2 Detailed ALPHA Model Description
The ALPHA model architecture is comprised of four systems: Ambient, Driver, Powertrain,
and Vehicle as seen in Figure 2.93. With the exception of Ambient and Driver, each system
consists of one or more subcomponents. The function of each system and its respective
ccc The GEM model has also been peer reviewed multiple times, and was the subject of intense comment during the
rulemaking adopting the second phase of GHG standards for heavy duty vehicles and engines. See 81 FR 73530-
531,538-549.
2-256
-------
Technology Cost, Effectiveness, and Lead Time Assessment
component models are discussed in this chapter. The structure and operation described in this
section incorporate numerous constructive comments from both public comments and peer
reviews. The model has been upgraded to integrate new technologies, improve the fidelity of the
simulation results and better match the operation of the benchmarked vehicles. This all supports
our primary goal of accurately reflecting changes in technology for both the current and future
light duty fleet. As part of this effort, substantial effort has been put forth to accurately track and
audit power flows through the model to ensure conservation of energy, and provide better data
on technology effectiveness.
ALPHA Vehicle Mode!
< [vehicle]
vehicle
powertrain
vehicle
[driver]
driver
Scope
driver
[powertrain]
[driver]
[system_bus]
[ambient]
[ambient]
Memory
ambient
[vehicle]
~
system_bus
bus_out
system_bus
bus_out
system_bus
v eh_spd_m ps
ALPHA_CVM
m ass_out_kg
force_out_N
bus_out
system_bus
force_in_N
mass_in_kg
veh_spd_mps
bus_out
ambient
Figure 2.93 ALPHA Model Top Level View
One unique feature of ALPHA is the use of dynamic lookup tables. These special lookup
tables provide interpolation similar to a normal Simulink ID or 2D lookup, but allow the
dimensionality and signals used for lookup to be determined at run time. This allows tables in the
model such as transmission losses to be parameterized in a way that best matches the available
data for that particular component. For example, a detailed transmission map may have had its
losses characterized by gear number, input speed, input torque, hydraulic line pressure and
temperature using a five dimensional lookup, while other testing might yield much simpler two
dimensional map utilizing only input torque and speed. ALPHA can accept either map without
physically altering the Simulink structure. Dynamic lookup tables are a powerful tool for
improving model fidelity when highly detailed data is available, but also allow the model to run
with coarse or simplified data when needed.
2.3.3.3.2.1 Ambient System
2-257
-------
Technology Cost, Effectiveness, and Lead Time Assessment
This system defines ambient conditions such as pressure, temperature, and road gradient,
where vehicle operations are simulated. ALPHA has been calibrated to generate fuel economy
results corresponding to chassis dynamometer certification tests; therefore, conditions within the
simulation have been maintained to align with current test procedures.
2.3.3.3.2.2 Driver System
The driver model in ALPHA is a purely proportional-integral control driver that features a
small look ahead to anticipate upcoming accelerations in the drive cycle. This is especially useful
at launch where the vehicle response may be delayed due to the large effective inertia in lower
gears. The driver in ALPHA is designed to follow a vehicle speed versus time driving cycle such
as the UDDS or HWFET. The driver is tuned to emulate the activities of a real driver during a
chassis test, including starting the engine, putting the transmission into gear and then operating
both the accelerator and brake pedals.
2.3.3.3.2.3 Power train System
The engine, transmission, electrical systems and accessories discussed in the following
section are combined to form vehicle powertrain systems. The conventional powertrain system
shown in Figure 2.94 contains sub-models representing each of the components. Additional
powertrains were constructed to simulate power split and P2 hybrid as well as full electric
drivetrains.
ALPHA CVM Conventional Vehicle Mode/
[control] ^>-
[engine]
—~<^ [control]
[transj ~^>-
[alectric]
[engine]
-KX)
mass_out_kg
CD-
veh_spd_mps
Figure 2.94 ALPHA Conventional Vehicle Powertrain Components
2.3.3.3.2.3.1 Engine Subsystem
The engine model is built around a steady-state fuel map covering all engine speed and torque
conditions with torque curves restricting operation between wide open throttle (full load) and
closed throttle (no load). The engine fuel maps for various engines are provided by benchmark
2-258
-------
Technology Cost, Effectiveness, and Lead Time Assessment
data, generated via tools like GT-POWER, or adapted from other data sources. The engine fuel
map contains fuel mass flow rates versus engine crankshaft speed and brake torque. In-cylinder
combustion processes are not modelled.
The steady-state fuel map used in ALPHA is adapted from the available test data or model
output by creating an interpolant grid covering the area between idle speed and redline speed,
and between the wide open throttle and closed throttle curves. In some circumstances, portions of
the map (for example, those near redline speed or near the closed throttle curve) are extrapolated
from the original data. In general, these areas represent engine operation which is either outside
of that used in two-cycle operation (near redline speed) or which uses little fuel in general (near
the closed throttle curve).
During the simulation, the engine speed at a given point in the drive cycle is calculated from
the physics of the downstream speeds. The quantity of torque required is calculated from the
driver model accelerator demand, an idle speed controller, and requests from the transmission
during shifts. The torque request is then limited by a torque response model which has been
tuned to match the torque response of naturally aspirated and turbocharged gasoline and diesel
engines. The resulting engine torque and speed are used to interpolate a fuel rate from the fuel
map.
Additional sources of fuel consumption documented in benchmarking activities have been
included in the model as well. On gasoline engines, the torque management that occurs during
shifting is implemented such that the reduction in torque does not cause a corresponding
reduction in the fuel rate. This approximates the effect of the observed spark retard to lessen the
lurch associated with decelerating engine inertia during upshifts. Another source of additional
fueling occurs after engines transition out of decel fuel cutoff. Additional fuel is applied for a
few seconds for emissions control. Finally, there are additional fuel penalties applied within the
simulation associated with rapid changes in engine power.
2.3.3.3.2.3.2 Electric Subsystem
The electric subsystem consists of 3 major components: battery, starter, and alternator.
The battery model for ALPHA was created after a literature review of battery models,
particularly for hybrid vehicle applications. The same battery model structure544'545 is used for
both conventional and hybrid vehicles, with different calibrations used to simulate different
chemistries such as lead-acid or lithium ion. The model features an open circuit voltage that
varies with state of charge, a series resistance, and dual RC time constant filters to provide
realistic voltage response. Calibrations were generated from published literature or EPA
benchmark testing for the open circuit voltage and transient behavior. The simulated battery also
features a thermal model, with the output current limited at extremes in temperature or state of
charge.
The engine starter is modeled as a simplified electric motor. It has a fixed efficiency and is
commanded via a Boolean activation signal. The operation of the starter is characterized by a
desired cranking speed and a torque capacity. These values are generally calculated to match the
engine specifications. When an engine start is requested a proportional integral controller is used
to determine the torque applied to accelerate the engine to the desired cranking speed, limited by
2-259
-------
Technology Cost, Effectiveness, and Lead Time Assessment
the torque capacity. The mechanical power required and efficiency then determine the resulting
electrical power consumed.
The engine alternator is modeled as a simplified electric generator with a fixed efficiency. The
electrical output current is determined by a charging controller. The efficiency and electrical
power output can then be used to compute the mechanical load applied to the engine. The
charging controller can operate in two different modes. In a basic mode it always tries to charge
the battery to a fixed voltage target. It also features an adaptive charging / alternator regen mode
that varies the voltage target and thus current output based on driving conditions. Lower
electrical output is provided during cruising, enough to maintain a minimal state of charge.
During decelerations and transmission upshifts electrical output and thus mechanical load are
increased to capture energy that would otherwise be dissipated via the brakes or transmission.
The adaptive charging / alternator regen strategy exhibits increased variability of battery state of
charge over various driving cycles. Therefore, it is necessary to precondition the model with a
prep cycle just as would be done on a test such as the HWFET to get accurate results.
2.3.3.3.2.3.3 Accessories Subsystem
The accessories subsystem in ALPHA is responsible for applying electrical and mechanical
loads to mimic those observed during testing. The system is capable of applying 4 different
loads: power steering, air conditioning, fan and a generic load to cover the remaining losses
observed. Each load can apply mechanical loads to the engine crankshaft and/or electrical loads
to the battery. Each load can be independently correlated to model signals via dynamic lookup
tables, and is calibrated to match test data. Baseline vehicles with mechanical power steering
often have mechanical losses that vary with engine speed, while future vehicles featuring electric
power steering may have electrical losses that vary with vehicle speed.
2.3.3.3.2.3.4 Transmission Subsystem
The transmission subsystem features different variants representing the major types of
transmissions (AT, DCT, and CVT) that are currently used in LD vehicles. The different
transmission models are built from similar components, but each features a unique control
algorithm matching behaviors observed during vehicle benchmarking.
One of the features in ALPHA, which is required for the model to conserve energy, is
multiple speed integrators. One is located at each of the points in the driveline where rotational
inertias may become decoupled such as the transmission gearbox. These integrators use the
torque and upstream inertia to compute the resulting acceleration and thus speed for the upstream
components. For couplings that may become locked up, such as a clutch, the torques and
rotational inertia are then passed down toward the next integrator in the model. This allows the
physics of the system to be accurately simulated, losses associated with clutch slip to be
computed, and the energy audit to be properly accounted.
2.3.3.3.2.3.4.1 Transmission Gear Selection
All of the gear transmission models use a dynamic shift algorithm, ALPHAshift,543 to
determine the operating gear over the cycle. This employs a rule based approach utilizing the
engine torque curve and fuel map to select gears that optimize efficient engine operation and
provide performance reserves as a traditional transmission calibration would. The ALPHAshift
2-260
-------
Technology Cost, Effectiveness, and Lead Time Assessment
algorithm attempts to select the minimum fuel consumption gear after applying constraints on
engine speed and torque reserve. It also allows downshifts due to high driver demand.DDD
The ALPHAshift algorithm contains calibration parameters that can be tuned to match
benchmarked shift behavior data from a particular engine and transmission. A generic calibration
tuning strategy has been developed from these specific benchmarked calibrations, and is useful
for simulating the shifting behavior of engine and transmission combinations that are from
different vehicles or represent future technologies.
The CVT transmission model uses a similar ALPHAshift-CVT546 algorithm for determining
gear ratio selection. It attempts to maintain operation on an engine speed versus requested power
line that minimizes fuel consumed. This method also has constraints for minimum engine speed
and the rate at which the gear ratio can be changed.
2.3.3.3.2.3.4.2 Launch Clutch Model
The clutch model in ALPHA can be modulated during launch (for manual and automated
manual transmissions) and requires a fixed time to engage. Torque is conserved across the clutch
during engagement and the inertial effects of accelerating and decelerating the upstream inertias
are captured. This additional fidelity necessitates a more complicated control algorithm to
manage clutch slip during launch which is included in the control strategy for the appropriate
transmissions.
Two clutches are bundled together to create the dual clutch module for the dual clutch
transmission. The dual clutch features a single integrator for calculating engine speed during
shifts.
2.3.3.3.2.3.4.3 Gearbox Model
The gearbox model for ALPHA has been developed with the goal of simulating realistic
operation during shifts for all types of transmissions. The gearbox contains gear ratios and
properly scales torque and rotational inertia through the ratio change. Power losses within the
gearbox are applied via dynamic lookup tables which determine torque loss and/or gearbox
efficiency. These loss tables are typically constructed using signals such as input torque, input
speed, commanded gear and/or line pressure.
Realistic shifting behavior is achieved with appropriate delays provided by a synchronizer
clutch model. The layout of the gearbox model is most similar to a manual transmission, but the
application for a planetary gearbox is a reasonable approximation once the neutral delay between
gears is omitted.
The gearbox rotational inertias are split between a common input inertia, common output
inertia and a gear specific inertia. The common inertias represent rotational inertia always
coupled to the input or output shafts. The gear specific inertias, which are only used for planetary
automatic transmissions, are added or removed as gears are engaged or disengaged. There is an
additional load placed on the powertrain associated with spinning up each gear specific inertia,
DDD Also known as a power downshift or kickdown.
2-261
-------
Technology Cost, Effectiveness, and Lead Time Assessment
and when each gear is disengaged the kinetic energy contained within the gear specific inertia is
discarded and treated as a loss.
2.3.3.3.2.3.4.4 Torque Converter Model
The torque converter model in ALPHA simulates a lockup-type torque converter. The torque
multiplication and resulting engine load are calculated via torque ratio and K factor curves that
vary as a function of speed ratio across the torque converter. Base torque ratio and the K factor
curves are often scaled in situations where detailed torque converter information is unavailable.
The lockup behavior of the torque converter is accomplished by integrating a clutch model
similar to the one discussed above. The torque converter model also contains a pump loss torque
that is implemented via a dynamic lookup table to simulate the power required to operate the
pump on an automatic transmission or CVT. When possible, pump losses are measured
separately during the component benchmarking process, and are generally represented as a
function of torque converter input speed and transmission line pressure.
2.3.3.3.2.3.4.5 Automatic Transmission & Controls
The automatic transmission (AT) is composed of the torque converter and gearbox systems
discussed above. The AT is allowed to shift under load. During upshifts and torque converter
lockup the engine output torque is slightly reduced to minimize the resultant torque pulse
encountered by decelerating the engine inertia.
The torque converter lockup clutch command is determined based on transmission gear and
gearbox input speed. The thresholds that trigger lock and unlock of the torque converter are
calibrated to match benchmark data.
2.3.3.3.2.3.4.6 DCT Transmission & Control
The ALPHA DCT model is constructed from two separate gearbox components and a dual
clutch module as described above. The dual clutch module features a dynamic lookup torque loss
table that can be used to represent all the gearbox losses in one location if loss information for
the separate gearboxes is not available. After a gear change to a new preselected gear is
requested, the dual clutch module will transition and begin applying torque through the new gear.
The DCT transmission controller also includes a low speed clutch engagement routine to
feather the clutch for low speed operation or launch. Similar to the automatic transmission,
engine output torque is reduced during upshifts to minimize the torque pulse at the wheels and to
prevent excessive clutch slip.
2.3.3.3.2.3.4.7 CVT Transmission & Control
The CVT transmission in ALPHA consists of the torque converter and gearbox modules.
When operating as a CVT the gearbox maintains a state of partial engagement allowing the gear
ratio to be constantly changed.
2.3.3.3.2.3.4.8 Driveline
The driveline system contains all of the components that convert the torque at the
transmission output to force at the wheels. This includes drive shafts as well as driven axles,
2-262
-------
Technology Cost, Effectiveness, and Lead Time Assessment
consisting of a differential, brakes and tires. ALPHA is capable of simulating multiple axles, but
it is often simpler to convert a driveline to a single axle equivalent.
The driveshaft is a simple component for transferring torque while adding additional
rotational inertia. It is only used for rear wheel drive vehicles.
The final drive is modeled as a gear ratio change with an associated torque loss and/or
efficiency. These losses are applied via a dynamic lookup table. For front wheel drive
transmissions, the final drive losses are often difficult to separate. In these situations, all losses
are applied in the gearbox.
The brake system on each axle applies a torque to the axle proportional to the brake pedal
position from the driver model. The brake torque capacity is scaled to match the stopping
requirements of the vehicle.
The tire component model transfers the torques and rotational inertias from upstream
components to a force and equivalent mass that is passed to the vehicle model. This conversion
uses the loaded tire radius and adds the tire's rotational inertia. A force associated with the tire
rolling resistance is not simulated because these losses are included in the road load ABC
coefficients applied within the vehicle subsystem (when using ABC coefficients, ALPHA is also
capable of using separate rolling resistance and aerodynamic drag coefficients).
2.3.3.3.2.3.5 Vehicle System
The vehicle system consists of the chassis, its mass and forces associated with aerodynamic
drag, and changes in road grade. The vehicle system also contains the vehicle speed integrator
that computes acceleration from the input force and equivalent mass which is integrated to
generate vehicle speed and distance traveled. The road load force is calculated from the ABC
coefficients based on coast down testing, or aerodynamic drag coefficient and frontal area data.
2.3.3.3.3 Energy Auditing
One of the quality control components within the ALPHA model is an auditing of all the
energy flows. This auditing enables verification that the physics represented in the model is
done correctly, generally resulting in a simulation energy error less than a few hundredths of a
percent. The audit data can also be compared between simulations to verify that individual
component losses are reasonable when compared to baseline packages or products that may
feature similar technologies. An example energy audit report for a package similar to a current
production sedan is shown in Figure 2.95. It should be noted that the lack of final drive losses in
this case is attributed to the vehicle being front wheel drive, and the thus the final drive losses are
included in the gearbox.
It is important to note that the layout of the Simulink blocks and the mathematical
configuration of the model are distinct. For example, the torque out of the engine Simulink
block may or may not represent net shaft torque - if downstream loads such as the torque
converter or launch clutch are unlocked or decoupled then the net shaft torque, including inertia
effects, is determined at the integrator which is located in the Simulink block representing the
decoupled device. For this reason, the auditing of the energy flows within the model is
accomplished by carefully observing the physics of the model as opposed to simply data logging
the Simulink block input and output ports.
2-263
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Energy Audit Report
Total Energy Consured = 205975.66 kJ
Fuel Energy - 205971.83 feu
stored Energy
Battery Internal Losses
Kinetic Energy
Potential Energy
Usable system Energy Provided
Engine Energy
Engine Efficiency
Stored Energy
Kinetic Energy
Potential Energy
Energy Consured by ABC roadload
Energy Consumed by gradient
Energy consured by Brakes
Energy Consumed fay Accessories
Starter
Alternator
Battery stored charge
Engine Fan
Electrical
Mechanical
Power Steering
Electrical
Mechani cal
Air Conditioning
Electri cal
Mechani cal
Generic loss
Electri cal
Mechanical
Total Electrical Accessories
Total Mechanical Accessories
Energy consured by Brivelirre
Launch Device
Gearbox
Pump Loss
spin loss
Gear/inertia loss
Final Drive
Tire slip
'Net Systeir Kinetic Energy Change
Total Loss Energy
Simulation Error
Energy Conservation = 100.011 %
Figure 2.95 Sample ALPHA Energy Audit Report
2.3.3.3.4 ALPHA Simulation Runs
ALPHA was used to perform a series of simulation runs, where various technology packages
were compared to exemplar vehicles. The exemplar vehicles have been adjusted from those used
in the FRM and Draft TAR to better represent the MY2015 vehicle baseline used in the OMEGA
analysis for this Proposed Determination as described in Section 2.3.3.2.2. Four acceleration
performance metrics were calculated for the exemplar vehicles: 0-60 time, lA mile time, 30-50
passing time, and 50-70 passing time. These metrics were chosen to give a reasonably broad set
of acceleration metrics that would be sensitive enough to represent true acceleration
performance, but not so sensitive that minor changes in vehicle parameters would significantly
change the final metric.
For each subsequent comparative run, a vehicle package was defined within ALPHA by
specifying powertrain components and road load specifications. ALPHA'S road load force at a
specific vehicle velocity (v) is determined by using the following formula; F = Cv2 + Bv + A
=
3. S3
kJ
=
0. 91
kn
0.00%
=
0.00
kn
=
0. 00
kJ
63307.72
kJ
=
63304. SO'
kJ
=
30. 73
%
=
2. 92
kJ
=
0.00
k3
=
0.00
kn
=
3"7485. 54
kD
59.17%
=
0.00
kJ
0. 00%
=
11429.43
kJ
18.05%
=
3505.29
kn
5. 54%
=
0.45
kJ
0.00%
=
1225.93
kD
1. 94%
=
0.00
kn
0. 00%
=
0.00
k3
0. 00%
0.00
kzi
0.00%
=
0.00
kn
0. 00%
=
0.00
kD
0.00%
=
0.00
kJ
0. 00%
=
0.00
k3
0. 00%
=
0.00
kn
0. 00%
=
0.00
kJ
0. 00%
=
0.00
kJ
0.00%
=
2278.85
kJ
3. 60%
=
227S.85
kJ
3. 60%
=
0.00
k 3
0. 00%
=
2278.85
kJ
3. 60%
=
0.00
kn
0. 00%
=
10913.96
k3
17.24%
=
1796.61
kzi
2. 64%
=
S150.72
kn
12.87%
=
3243.10
k3
5.12%
=
2837.IS
kn
4.48%
=
2070.44
k3
3.27%
=
0.00
kJ
0.00%
=
966.63
kJ
1. 53%
=
0.44
kJ
0. 00%
—
63314.66
kD
=
-6. 94
kD
2-264
-------
Technology Cost, Effectiveness, and Lead Time Assessment
where the coast down coefficients (A, B, and C) are derived from a least squares fit of data from
track coast-down tests.
In ALPHA modeling, it is assumed that the A coefficient is a factor for the road load force
that is mostly associated with tire rolling resistance, the B coefficient is a small factor, which
represents higher order rolling resistance and gearing loss factors, and the C coefficient is a
factor which mostly represents aerodynamic drag. Thus, changes in aerodynamic losses are
modeled by changing the C coefficient, and changes in rolling resistance losses are modeled by
changing the A coefficient. Changes in mass reduction are modeled by reducing the test weight,
and by reducing the A coefficient (as rolling resistance is a function of vehicle weight).
For each of the six vehicle class described in Section 2.3.1.4, an exemplar configuration was
chosen and was run over the performance cycle, and the times for each performance metric were
extracted. These four metrics were summed for the exemplar vehicle. For each vehicle
technology package based on the same vehicle class, a nominal engine size was determined
based on the estimated performance effect of the technologies included in the package and a set
of packages with a range of engine sizes larger and smaller than the nominal engine size were
simulated. The same performance cycle was run and the sum of the four metrics compared to the
exemplar sum for each engine size package. Results where the sum was not equal to or less than
the exemplar sum (more stringent than the 5 percent band suggested by NAS) were rejected.
The drive cycle CO2 emissions of the target package were taken from the lowest emissions result
of the remaining results.
To account for changes in engine efficiency as a result of resizing for simulation, a set of
adjustments was developed. First, based on the overall size the architecture of the engine (13,14,
V6, and V8) is selected so that the scale factor for cylinder volume could be calculated. The first
adjustment is related to the changes in heat transfer that result from altering the surface to
volume ratio of the cylinder. Increasing cylinder volume leads to a lower percentage of
combustion energy transferred to the engine head and block resulting in higher efficiency. The
adjustment factor was derived from published test data547 and is supported by other literature.548
The second adjustment modifies engine friction. Literature contains methodologies for
estimating engine FMEP based on various engine dimensions.549'550 Using inputs consistent with
current production engines estimates of FMEP for various architectures and displacements were
generated. Using the FMEP estimates for the original and resized engines an adjustment can be
applied to the fuel map and other parameters related to engine torque. The third adjustment
relates to the increased knock sensitivity of engines when increasing cylinder volume. As engine
bore increases the higher knock tendency drives more retarded spark timing and thus lower
efficiency. The knock sensitivity is characterized using trends in the original fuel map, and from
this an adjustment can be made that reduces efficiency during low speed high load operation.
The net result of these adjustments when scaling an engine of fixed architecture up to a larger
displacement are efficiency reductions at low speed and high load and increases over the
remainder of the map.
2.3.3.3.5 Post-processing
ALPHA simulation runs are performed assuming warm component efficiencies. Additional
fuel consumption due to the FTP cold start is calculated in post-processing by applying a fuel
consumption penalty to bags 1 and 2. These fuel consumption penalty factors represent
additional fuel used to heat the catalyst, and additional energy lost to higher viscosity lubricating
2-265
-------
Technology Cost, Effectiveness, and Lead Time Assessment
oil in the engine and transmission. The fuel consumption penalties for "present" and "past"
vehicles (component vintaging is discussed in 2.3.3.3.6) are set at 15 percent (present) to 17
percent (past) for bag 1 and 2.5 percent for bag 2. The penalty factors are applied during post-
processing so that the fuel consumption for the appropriate bag is increased by the indicated
amount. These factors were determined by comparing the "cold" FTP bags 1 and 2 to the "warm"
bags 3 and 4 for a range of vehicles.
Since the three-bag FTP is a standard test, the difference in fuel consumption between bags 1
and 3 of the FTP could be calculated for the entire fleet (available in the Test Car List data
files551), as seen in the graph below. However, the data sources for bag 4 are more limited. EPA
based the 2.5 percent penalty factor on test data available from conventional vehicle testing from
Argonne National Labs552 and from internal testing, where differences between bags 2 and 4
averaged about 2.5 percent.
35%
30%
CO
on
25%
m 20%
CO
-Q
H 15%
~o
1 10%
5%
0%
«
•
1 •
•
. •
•
«•
• r*
•
•
•
•
*•.
. ••
• •
•
• •
,# • ••• »
• •
••
•• •
•n .&
•••
.. *
High 10%
• 1
-
&
• •
Atfledian
•
• v
% *
¦ ^ •
• •
»• * •
1 » • *
* ••
•
• ••
v •••
•
•
* v* .
.* 4
' . •
•
•
.Low 10%
•
•
10
15
20
25 30
FTP mpg
35
40
45
Figure 2.96 Example: Difference in 2016, Between Bags 1 and 3 of the FTP, from the Test Car List.
For simulation of advanced vehicle packages which included thermal management of the
engine or transmission, the penalty factors were reduced (to a minimum of 11 percent for bag 1
and 0 percent for bag 2) to account for the reduction in losses associated with faster component
warmup.
2.3.3.3.6 Vehicle Component Vintage
Vehicle components (engines, transmissions and accessory loads) are assigned a vintage of
"past," "present," or "future." The vintage of the component determines the assumed technology
package associated with the component, and thus the default value of some associated
parameters.
One parameter affected by vintage is electric accessory loading. The "past" value for electrical
loads includes a base electrical load of 154 W, additional power draw based on engine speed
(approximately 700 W at 2500 rpm and 1050 W at 6000 rpm), and an alternator efficiency of 55
percent. These values assume mechanical power steering. The "present" value for electrical load
includes a base electrical load of 390 W, no additional variable accessory power draw, and an
alternator efficiency of 65 percent. This is based on loads measured in various tested vehicles,
SAE technical papers and stakeholder feedback. The "future" electrical load assumes a 290 W
2-266
-------
Technology Cost, Effectiveness, and Lead Time Assessment
base electrical load, but with a high-efficiency (70 percent efficient) alternator that also employs
an alternator regen strategy.
Future vintage transmissions are also assumed to be associated with reduced parasitic losses
and early torque converter lockup.
Although the assigned vintage determines default values for accessory loads and cold start
penalty, these defaults can be overridden in the model to examine the effects of specific
technologies separately.
2.3.3.3.7 Additional Verification
As an additional verification of ALPHA model simulations, EPA compiles and executes
technology package combinations using a hardware-in-the-loop (H1L) system. This process
enables powertrain, vehicle, and driver behavior to be observed in real time for both on-cycle
and off-cycle situations. Any undesirable behavior is analyzed and used to fine tune the
modeling process. These compiled HIL models are also utilized by EPA as part of the vehicle
benchmarking process when testing vehicle subsystems such as engines, transmissions, battery
modules, and other components. Figure 2.97 shows an example ALPHA model simulation
observation display.
4 '/
O \ 5
A" 2 6
- 1 7 -
' Gear
- 0 8 -
' ® Unlock v
Time
461
g C02/Mile
294 .5
Miles
3.2
Fuel
295.9
Figure 2.97 Example ALPHA Model UDDS Simulation Observation Display
As part of EPA's on-going quality process, comparative analyses were completed by EPA as
part of the ongoing MTE work. When viewing full vehicle simulation models as a calculator,
providing the same inputs to the calculators should provide the same outputs. The first set of
comparisons used Ricardo EASY5 inputs from the MY2017-2025 Light-Duty FRM as inputs to
the ALPPIA model. The EASY5 and ALPFIA results showed only minor differences. The
RPMxIOOO
o R=Eng G=TISSY=TOSS 8
Engine Acceleration
Vehicle Acceleration
Driver Accel
Driver Brake
2-267
-------
Technology Cost, Effectiveness, and Lead Time Assessment
second set of comparisons used a set of inputs provided by the Autonomie model. Again the
Autonomie and Alpha models showed only minor differences between simulation results due to
specific model behaviors or implementations, convincing EPA that these models are very close
in terms of computational results when run using the same input data and assumptions.
2.3.3.3.8 Key Public Comments Related to the ALPHA Model
Because the ALPHA model reaches into many facets of EPA's technology assessments, many
of the topics touched upon in comments on the Draft TAR can be seen as related in some way to
the ALPHA model. This section gathers some of the key comments that either directly concern
the design or use of the ALPHA model or were conveyed in the context of a discussion of the
ALPHA model. Some comments cited here are better addressed in the context of a more specific
topic, and in those cases the reader is directed to the TSD chapter where the comment is
addressed.
Some comments recognized the importance of EPA's use of ALPHA, a physics-based,
forward-looking simulation tool that is available to the public. The International Council on
Clean Transportation (ICCT) noted, "EPA's new physics-based ALPHA model offers a nice
enhancement in modeling multiple technologies." The Union of Concerned Scientists (UCS)
also noted, "EPA extensively employed its own, freely accessible ALPHA full-vehicle modeling
tool, which was extensively peer-reviewed and benchmarked against its work at its laboratory,
which also resulted in numerous peer-reviewed publications. This laboratory analysis allowed
for combinations of technologies not available on the road today to be analyzed, including both
combinations of turbocharged engines with advanced transmissions and future high-compression
ratio engines."
The Alliance of Automotive Manufacturers, Global Automakers, and other stakeholders
provided detailed comments regarding the ALPHA Model.
A comment from the Alliance suggested that EPA use the Autonomie model in place of the
ALPHA model, on the grounds that the industry is more familiar with Autonomie. In response,
the ALPHA model was developed to eliminate the "black box" and copyright issues with
commercial modeling products to allow full transparency in the modeling process. The ALPHA
model is designed to function in a compliance environment and to be publicly available without
any hidden or proprietary aspects.
While not directly related to the ALPHA model (but rather its inputs), one commenter stated,
"The engine maps used by the full vehicle simulation models do not fully consider key technical
issues, and are therefore generally optimistic." Comments on engine maps and similar inputs
used in the Draft TAR analysis are considered and discussed in Chapter 2.3.4.1 (Engines: Data
and Assumptions for this Assessment) of this TSD.
The Alliance commented, "There are a number of technical flaws that are common to both the
ALPHA and Autonomie models which bias the full vehicle simulations to more optimistic
benefits than those anticipated by automakers." The comment continued by suggesting that this
was related to several aspects of criteria emissions compliance that the Alliance felt could impact
CO2 and fuel economy performance as projected in the analysis, stating: "The Alliance
recommends that both Agencies account for the CO2 and FE degradation associated with Tier 3
emissions control systems and the impact of more stringent evaporative emissions regulations in
2-268
-------
Technology Cost, Effectiveness, and Lead Time Assessment
their MTE analysis. The effect of the evaporative emissions regulations is further magnified for
engine stop-start and HEV applications where the engine off option is constrained by the need to
purge the canister for evaporative emissions requirements." In a related comment, the Alliance
recommended "that both Agencies account for and include the detrimental impact of CARB
particulate matter (PM) (1 mg/mi) regulations on CO2 and FE performance in the MY2022-2025
time frame. The 1 mg/mi PM (1) requirement could impact approximately 40 percent of the
fleet."
Regarding criteria pollutant emissions, EPA developed the Tier 3 program in full
consideration of both the light duty and heavy duty GHG programs that would be occurring in
the same time frame as the phase-in of the Tier 3 rule. In fact, many of the program's key dates
including the final MY2025 standards were specifically coordinated to allow the criteria
pollutant and GHG programs to work together in a complementary fashion and leverage
technology synergies. As an example, downsized engines used to comply with the GHG
requirements of lower CO2 emissions generally also produce lower engine out criteria pollutant
emissions. Lower engine out criteria emissions facilitate manufacturer's task of reducing the final
tailpipe emission levels required to meet Tier 3 standards. Another technology used to reduce
criteria pollutant emissions involves reducing or minimizing the amount of fuel enrichment used
for cold starts. This reduction in fuel used for starting and running a cold engine translates
directly to lower fuel consumption and therefore reduced CO2 emissions during the cold start and
warm-up. EPA recognizes that certain strategies used today to reduce criteria pollutant
emissions, particularly elevated idle speeds and retarded timing used initially following a cold
start to warm-up the catalyst, can temporarily reduce engine efficiency. However, manufacturers
have other options that result in similar benefits for criteria pollutant emissions without a CO2 or
fuel economy penalty. This includes other methods to more rapidly warm the catalyst such as
insulated exhaust pipes or better catalyst design and placement. Additional discussion of the
comment regarding CO2 emissions may be found in Chapter 2.3.1.3 (Fuels) of this TSD.
Regarding evaporative emission challenges, Tier 3 standards did not result in an increase in
the amount of purge required to meet evaporative emission requirements from what was already
required in the Tier 2 program. Instead, it requires improvements to evaporative hardware to
prevent or capture any residual fuel related emissions. EPA recognizes that technologies that
reduce engine operation such as stop-start and HEV applications also result in reduced
opportunity to purge the evaporative canister of fuel vapors. Manufacturers have successfully
designed and produced evaporative emission control system technologies to deal with the
challenge of reduced purge opportunity. These technologies include sealed or partially sealed
fuel systems that produce less fuel vapors that would need to be purged by running the engine.
Additionally, EPA has historically worked with manufacturers to adjust test procedures when a
new technology is not appropriately evaluated over existing test procedures and protocols.
Regarding Federal PM emissions standards, many vehicles, including those with naturally
aspirated and turbocharged/downsized GDI engines, already have PM emissions sufficiently low
to comply with Tier 3 PM emissions standards with a compliance margin. Vehicles with PFI-
equipped engines typically have PM emissions over the FTP that are 25 percent to 50 percent of
the proposed future California LEV III 1 mg/mi PM standard over the FTP chassis dynamometer
test. EPA certification and confirmatory emissions data on vehicles equipped with dual-injection
systems (both PFI and GDI) such as vehicles equipped with Toyota's 2GR-FSE and 4U-GSE
engines have PM emissions over the FTP drive cycle that are comparable to PFI engines and thus
2-269
-------
Technology Cost, Effectiveness, and Lead Time Assessment
well below 1 mg/mi PM over the FTP. Toyota is applying dual injection to other engines, such
as the 8AR-FTS 2.0L turbocharged Miller Cycle engine to improve efficiency and drivability.
Ford recently announced application of a similar dual-injection strategy to model year 2017 and
later light-duty trucks equipped with the 3.5L EcoBoost engine. Dual-injection represents one
approach to achieve sub-1 mg/mi PM emissions over the FTP drive cycle with potential for
reduced CO2 emissions, low PM emissions, and improvements in catalyst light-off performance
for improved NOx and NMOG emissions. The best GDI and turbocharged GDI engines (without
dual-injection) currently have PM emissions of between 1.0 and 3.0 mg/mi. At the 2015 EPA
Ultrafine Particle Workshop, AVL presented a range of strategies to bring GDI engines into
compliance with future California LEV III PM standards, and also future EU Euro 6 SPN
standards.553 AVL found via in-cylinder optical measurements that conditions with high flame
luminance could be used to indicate the presence of non-homogeneous, diffusion-limited
combustion associated with soot pyrolysis and particle formation. Methods identified by AVL to
reduce diffusion-limited combustion in GDI engine applications included:
• Reducing fuel impingement onto surfaces via changes in injector spray targeting,
piston bowl shape, injection event timing and use of multiple injections per
combustion cycle.
• Changes to spark timing and injection events to directly heat the piston following
cold startup to improve the vaporization of impinged fuel.
• Changes to the catalyst heat-up strategy used to improve catalyst light-off after cold
start, including further optimization of the timing and duration of multiple injection
events.
Eight recent engine development programs conducted by AVL that began with SPN
emissions at up to 6 times the EU6c standards were successfully reduced to 15 percent to 45
percent of the EU6c PN standards using such combustion refinements. While not a direct
indication of PM emissions, vehicles capable of emissions at less than half the EU6c SPN
standard would likely have PM emissions well under the future California LEV III 1 mg/mi
standard.
In summary, the best currently available GDI technology has already achieved criteria
pollutant emissions consistent with future Federal Tier 3 PM emissions standards. One fueling
strategy for use with GDI engines, dual-injection, has demonstrated the capability of meeting the
future proposed LEV III PM emissions standard of 1 mg/mi over the FTP beginning in 2025 if
such a standard is approved for implementation by the California Air Resources Board. Further
combustion system refinements using more conventional GDI systems (i.e., a single injection
system) appear to have the capability of meeting the proposed 1.0 mg/mi FTP standard when
taking into account the lead time available prior to implementation of these standards and
assuming that a 1.0 mg/mi FTP PM standard is finalized in California.
With regard to electrical accessory loads, the Alliance suggested that "the Agencies
harmonize around the NHTSA base electrical accessory loads of 240 W", further commenting,
"the base electrical loads used by the Agencies differ by a factor of two. While there are some
vehicles that do reach 490W and greater, the average two-cycle base load of the sample vehicles
is 387W. By inflating the base electric load, EPA has effectively overestimated the effectiveness
of load reduction technologies. "In response, the "Table A-2: Electrical Base Load
Benchmarking Data" provided by the Alliance is appreciated, and agrees well with the "present
vintage" vehicle accessory load of 390 W used in ALPHA for the Proposed Determination. For
2-270
-------
Technology Cost, Effectiveness, and Lead Time Assessment
more details on EPA's assumptions for accessory loads, please refer to "2.3.3.3.6 Vehicle
Component Vintage" of this TSD.
The Alliance also recommended "that NHTSA and EPA harmonize and use regular grade Tier
3 test fuel for all future analysis, unless testing 'premium required' engines ... In addition, Tier 3
test fuel also contains 10 percent ethanol, lowering the energy content of the fuel." Consideration
of this comment is found in Chapter 2.3.1.3 (Fuels) of this TSD.
Another comment stated, "When adjusting engine size to maintain performance, EPA
assumes that any resulting engine displacement will be available, maximizing the modeled
benefits of various technologies. In practice, manufacturers have a limited number of engine
displacements to choose from and will likely select the size of engine that maintains or improves
performance. EPA's assumption of infinite engine displacement availability yields unreasonably
optimistic results." EPA notes that engine resizing for performance neutrality is a modeling
convenience that allows an overall fleet-wide estimation of CO2 reduction while accounting for
the effects of performance, as recommended by the NAS. EPA does not expect manufacturers to
rigidly maintain performance, footprint, or any other characteristics of a specific vehicle for the
duration of the rule. Rather, EPA anticipates that manufacturers will use the flexibility of the rule
to balance a range of requirements, including the manufacturer's estimation of the availability of
engine displacements, when designing vehicles. For a more detailed discussion of the "engine
sizing for performance neutrality" topic, please refer to Chapter 2.3.1.2 (Performance
Assumptions) of this TSD.
With regard to downsized and turbocharged engines, another comment stated, "displacement
to vehicle mass ratio (D/M) provides a simple means to assess whether the degree of downsizing
will find market acceptance. By failing to consider this parameter, the Agencies could model
engines which will not gain customer acceptance. We recommend that both Agencies ... add a
constraint which considers the displacement to mass ratio." However, the market has already
accepted vehicles with the degree of downsizing reflected within the Draft TAR and the
Proposed Determination, which includes segment-leading truck applications like the Ford F150.
With regard to performance neutrality, another comment stated, "A key metric needed to
maintain performance neutrality is top gear grade-ability. In contrast, the main metric by which
performance neutrality is measured by the Agencies is 0-60 acceleration time ... none of the
metrics evaluated is a substitute for top gear grade-ability." This comment is considered and
addressed in Chapter 2.3.4.2.2 (Effectiveness Values for TRX11 and TRX21) of this TSD.
The Alliance also recommended that "the Agencies incorporate and make readily available
quality control parameters that can be used to verify the validity of model results in all output
files." EPA notes that the version of ALPHA used for the Proposed Determination generates .csv
output files that contain over 150 columns of data and quality control parameters. In addition,
since EPA is providing a functional copy of ALPHA on its website, the Alliance can add
additional quality control data as desired. TSD Chapter 2.3.3.3.3 (Energy Auditing) also
contains a description of the energy flow auditing that describes another useful quality control
component in ALPHA.
2.3.3.4 Determining Technology Effectiveness for MY2022-2025
2-271
-------
Technology Cost, Effectiveness, and Lead Time Assessment
EPA collected information on the effectiveness of current CO2 emission reducing
technologies from a wide range of sources. The primary sources of information were the 2017-
2025 FRM, the Draft TAR, public comments on the Draft TAR, EPA's ALPHA model, EPA's
vehicle benchmarking studies, the 2015 NAS Report, OEM and Supplier meetings, and industry
literature. In addition, EPA considered confidential data submitted by vehicle manufacturers,
along with confidential information shared by automotive industry component suppliers in
meetings with EPA staff. These confidential data sources were used primarily as a validation of
the estimates since EPA prefers to rely on public data rather than confidential data wherever
possible.
In the Novation Analytics study commissioned by the Alliance, the analysis assumes that no
innovation will occur during MY2022-2025. EPA disagrees and recognizes that technologies
will be further developed and introduced for MY2022-2025 and that innovation by automobile
manufacturers and suppliers will continue to occur. While it is impossible for EPA to predict all
of the technologies that will come to fruition, likely trends can be identified in the development
of automotive systems that impact GHG emissions over the next decade. EPA uses methods
similar to those used by industry to identify and evaluate emerging automotive technology
trends. The use of computer aided engineering (CAE) tools for technology evaluation has been a
key source of technology effectiveness data for MY2022-2025 vehicle technology packages. A
number of other sources of data are also used to either validate CAE results or as independent
sources of effectiveness data. In addition to our review of public comments on the Draft TAR,
other sources of data include:
• Engineering analysis of logical developments based on current or near-term
technology
• Review of peer-reviewed journal papers, U.S. Department of Energy Reports, and
other public sources of peer-reviewed data
• Purchase and review of proprietary reports by major automotive industry analytical
firms (e.g., R.L. Polk, IHS Automotive)
• Meetings with automobile manufacturers
• Meetings with Tier 1 automotive suppliers
• Contracts with major automotive engineering design, analysis, and services firms
(e.g., FEV, Munro and Associates, Southwest Research Institute, Ricardo PLC) to
purchase data or engineering services
• "Proof of concept" research either conducted directly by EPA at EPA-NVFEL or
under contract with engineering services firms
• CAE tools, including:
0 Engine modeling (e.g., Ricardo WAVE, Gamma Technologies GT-POWER)
0 Vehicle modeling (e.g., EPA LPM, EPA ALPHA, Ricardo RSM, MSC EASY5)
0 HIL simulation of drive cycles
0 Computational fluid dynamics (CFD) for initial component development
• Chassis dynamometer testing
• Engine dynamometer testing
• Transmission dynamometer testing
2-272
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Data from all sources listed above is used to develop and validate vehicle effectiveness within
the EPA ALPHA model and EPA LPM. Modeling of technology package effectiveness within
the ALPHA model and LPM is the source of all technology package effectiveness data contained
within the OMEGA cost-effectiveness analyses. With respect to engine and powertrain
technologies, the general progression of data into the OMEGA analyses has been:
• Develop physics-based models of the technology with extensive validation of a base
configuration to actual hardware (e.g., validation of an engine model to actual engine
performance, combustion measurements and knock characteristics)
• Use the validated physics-based model to evaluate hardware changes and to develop
calibrations necessary to account for such hardware changes
• Use the ALPHA model to determine the CO2 effectiveness of the powertrain package
for different vehicle configurations
• Compare the energy balance of ALPHA model results with vehicle benchmark results
as an additional plausibility analysis.
• Use ALPHA modeling results to provide a calibration for technology package
effectiveness within the LPM
• Validate ALPHA modeling results using a variety of data sources including chassis
dynamometer testing of production and developmental vehicles, dynamometer testing
of production engines and transmissions, HIL testing of developmental engine
configurations, comparison with automobile manufacturer and Tier 1 supplier data,
and comparison with peer-reviewed/published data sources.
• Update LPM calibration with validated ALPHA model technology package
effectiveness.
Notable modeling updates from the Draft TAR supported by public comments include:
• EPA has updated both the ALPHA model and the LPM to calculate vehicle
effectiveness based on power-to-weight ratio and road load characteristics. Baseline
vehicles are mapped into these groups accordingly. Please refer to Section 2.3.3.2.2
for more information on this update.
• Engine displacement has been increased 5 percent in the OMEGA analysis for
technology packages containing future Atkinson engines to account for performance
and fuel characteristics.
The EPA analysis of naturally aspirated Atkinson cycle engines provides an example of an
analytical framework that integrates CAE together with other methods used by EPA to evaluate
future vehicle technologies. The 2.0L Mazda SKYACTIV-G engine was introduced in 2012 in
the U.S. This engine represents state-of-the art brake thermal efficiency for a naturally aspirated,
spark-ignition engine and is the first non-HEV application of an Atkinson cycle engine in a U.S.
light-duty vehicle application. EPA conducted chassis dynamometer testing of Mazda vehicles
with the SKYACTIV-G engine and also purchased versions of this engine marketed in the U.S.
(13:1 geometric compression ratio) and EU (14:1 geometric compression ratio) for detailed
engine dynamometer mapping and HIL testing. After both chassis dynamometer testing and
initial engine dynamometer testing, EPA conducted an engineering analysis to prioritize near-
term technologies that could potentially yield further brake thermal efficiency improvements,
2-273
-------
Technology Cost, Effectiveness, and Lead Time Assessment
broaden areas of high thermal efficiency and/or better align high brake thermal efficiency
operation with both the regulatory drive cycles and with urban driving with the goal of meeting
the 2022-2025 GHG standards in a "standard car" configuration (approximately D-segment size-
class).
The technologies chosen for further analysis included:
• Improving alignment of high brake thermal efficiency operation with urban driving
via road load reduction, switching to an advanced 8-speed automatic transmission,
and using fixed 4/2 cylinder deactivation
• Improving brake thermal efficiency by increasing expansion the ratio from 13:1 to
14:1 along with the addition of low-pressure-loop EGR for additional knock
mitigation on standard pump fuel and additional pumping loss improvements
An initial proof of concept evaluation of increased expansion ratio, low-pressure-loop cooled
EGR and cylinder deactivation was conducted using GT-POWER engine modeling.554 Engine
dynamometer testing with HIL simulation of regulatory drive cycles was used for concept
evaluation. A 2.0L SKYACTIV-G to larger D-segment vehicles was simulated through the
application of an advanced 8-speed automatic transmission and reduced road load ,555
Combinations of these technologies were also compared to similar vehicle configurations using
turbocharged, downsized GDI engines using the ALPHA vehicle model.556 An important part of
EPA's use of CAE has been to validate simulation results using other data sources. For example,
EPA validated the ALPHA modeling and HIL testing using chassis dynamometer test data and
validated the GT-POWER modeling using engine dynamometer test data.
2.3.3.5 Lumped Parameter Model
The foundation of the technology assessments that EPA conducted for the FRM, Draft TAR,
and this Proposed Determination was constructed from an evaluation of the state of individual
technologies: their costs, emissions-reducing benefits, and feasibility of implementation within
the time frame of the standards. As described in Chapter 2.3.3.3, data describing individual
technologies were synthesized at the vehicle-level using the physics-based ALPHA model.
Because specific inputs such as engine maps, transmission parameters, accessory loads, etc. are
not available for every vehicle, the ALPHA model is not sufficient to generate absolute tailpipe
emissions values for each vehicle in the baseline fleet using only raw technology data as an
input. Instead, the incremental effectiveness values generated by the ALPHA model are used to
calibrate the Lumped Parameter Model (LPM) so that the overall emissions-reducing benefits of
complete technology packages can be modeled reliably for individual vehicles within each
ALPHA effectiveness class. This approach of applying incremental effectiveness improvements
according to vehicle class is consistent with the approach used in the FRM and Draft TAR.
For this Proposed Determination, the representativeness of those effectiveness estimates has
been improved by defining effectiveness classes according to the important characteristics of
power-to-weight ratio and road load horsepower, as well as updating the exemplar vehicle
characteristics to align with the MY2015 fleet as described in Section 2.3.3.2.
2.3.3.5.1 Approach for Modeling Incremental Effectiveness
2-274
-------
Technology Cost, Effectiveness, and Lead Time Assessment
It is widely acknowledged that full-scale, physics-based vehicle simulation is the most
thorough approach to modeling future benefits of a package of new technologies. This is
especially important for quantifying the efficiency of technologies and groupings (or packages)
of technologies that do not currently exist in the fleet, nor as prototypes. However, developing
and executing every possible combination of technologies directly in a compliance environment
using full-scale vehicle simulation, while possible, would create many thousands of vehicle
combinations and corresponding effectiveness results, many of which would never be applied.
For example, combinations of technologies such as continuously variable transmissions applied
to pick-up trucks with towing capability are not viable using technologies that EPA expects to be
available in the MY2022 to 2025 time frame.
In assessing the GHG standards, EPA analyzes a wide array of potentially feasible technology
options rather than attempting to pre-select the "best" solutions. For example, in the analysis for
the Draft TAR, EPA built over 800,000 packages for use in its OMEGA compliance model,
which spanned 19 vehicle classes and over 2,200 baseline vehicle models. The Proposed
Determination analysis has expanded the number of vehicle types to 29 and the number of
baseline vehicles to over 2,000 models.
General Motors (Patton et al)557 presented a vehicle energy balance analysis to highlight the
synergies that arise with the combination of multiple vehicle technologies. This report
demonstrated an alternative methodology (to vehicle simulation) to estimate these synergies, by
means of a "lumped parameter" approach. This approach served as the basis for EPA's lumped
parameter model (LPM). EPA continues to believe that the lumped parameter approach is the
most practical surrogate to estimate the effectiveness of technology package combinations for the
Proposed Determination analysis.
The LPM does not model absolute effectiveness, but rather, the incremental improvements
between vehicle technology packages calibrated by full vehicle simulation modeling. As in the
FRM and Draft TAR, the LPM provides an interpolation between fully simulated vehicle
packages, based on industry accepted values, in order to account for the effect of individual
technologies. This increased resolution allows for every modeled technology to be accounted for
to prevent double counting and/or missed opportunities for improvement.
To further explain this process, consider an engine map in a full vehicle simulation model that
includes GDI+EFR1+DCP+DVVL, but the baseline vehicle only includes GDI+EFR1+DCP. As
no engine map without DVVL is available, the modeler may have to apply this engine map to the
baseline vehicle taking DVVL off the table for improvement. To correct for this situation, the
LPM contains all of the individual components selected as a group to equal the effectiveness of
the full vehicle simulated GDI+EFR1+DCP+DVVL engine map. At this point DVVL can be
deselected from the LPM to match the baseline vehicle's GDI+EFR1+DCP engine, reducing the
package effectiveness appropriately. Subsequently in the modeling process, DVVL is added as
an improvement to the baseline engine, matching the full vehicle simulation results in the
process.
The opposite situation also exists: an engine map in a full vehicle simulation model may
include GDI+EFR1+DCP but the baseline vehicle may include GDI+EFR1+DCP+DWL. As
no engine map with DVVL is available, the modeler may have to apply this engine map to the
baseline vehicle, leaving the baseline vehicle represented without DVVL. This would allow
double counting of DVVL if this technology were added later in the vehicle improvement
2-275
-------
Technology Cost, Effectiveness, and Lead Time Assessment
process. As before, the LPM contains all of the individual components selected as a group to
equal the effectiveness of the full vehicle simulated GDI+EFR1+DCP engine map. At this point
DVVL is individually selected from the LPM to match the baseline vehicle's
GDI+EFR1+DCP+DVVL engine, increasing the package effectiveness appropriately.
AAM commented on the Draft TAR that the starting point efficiency is critically important
for projecting the benefits of additional technology and that the LPM does not account for the
starting point efficiency. EPA agrees with the criticality of starting point efficiency but does not
agree that the LPM does not account for the starting point efficiency. EPA's methodology for
establishing effectiveness starts with an assessment of the application of technology in each
individual vehicle in the baseline fleet. Existing technologies within the baseline fleet are
identified to avoid double counting of technology benefits. In addition, the certified C02
performance of each vehicle, which is directly reflects a vehicle's efficiency, is the starting point
for determining the incremental effectiveness of additional technology. The incremental
effectiveness determined by the LPM accounts for the vehicle type, horsepower to weight ratio,
road load characteristics and energy loss categories. In addition, the incremental effectiveness
applied by the LPM is bounded by the calibration data from ALPHA full vehicle simulation.
The details regarding each of these factors is carefully documented in the section below.
2.3.3.5.2 Calibration of LPM using ALPHA model
As in the Draft TAR, the basis for calibrating and validating the lumped parameter model for
this Proposed Determination is the effectiveness data generated by the benchmarking and full-
vehicle simulation modeling activities described earlier in this section. As described above, the
LPM also allows benchmarked and/or simulated vehicle packages to be separated into individual
components to properly account for the technologies already in the vehicle fleet, to avoid any
double counting of these technologies. The lumped parameter approach was endorsed by the
National Academy of Sciences in the 2015 NAS Report, which stated: "In particular, the
committee notes that the use of full vehicle simulation modeling in combination with lumped
parameter modeling has improved the agencies' estimation of fuel economy impacts."558
As described in Section 2.3.3.3.3, as part of the quality assurance process, EPA checked the
ALPHA simulation results that were used to calibrate the lumped parameter model against
conservation of energy requirements. Similarly, 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 by the vehicle. The
analysis accounts for the dissipation of energy into the different categories of energy losses,
including each of the following:
• 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
2-276
-------
Technology Cost, Effectiveness, and Lead Time Assessment
• Vehicle road load (tire and aerodynamic) losses
• Inertial losses (energy dissipated as heat in the brakes)
It is assumed that each 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 technologies. 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 less inertia energy to recapture. In their comments on the Draft TAR, the AAM
stated that "linear regression models within the LPM are not based on the first order
determinants of powertrain efficiency and, therefore, do not properly capture the fundamental
trends." EPA disagrees with this assessment. As stated above, the LPM is grounded in
fundamental physical principles and bounded by full vehicle simulation. EPA has further refined
the LPM based on some of the comments received; however, we continue to believe that the
LPM provides an accurate assessment of incremental effectiveness.
EPA has updated the LPM for this Proposed Determination to improve fidelity for baseline
attributes and technologies. Consistent with suggestions in the public comments, the LPM now
characterizes baseline vehicles based on their power-to-weight ratio and road load characteristics
(see Section 2.3.3.2 above). For this Proposed Determination, as in the Draft TAR, the LPM has
been calibrated to follow the results of the ALPHA full vehicle simulation model to facilitate the
vehicle package building process used in the OMEGA model.
2.3.3.5.3 Lumped Parameter Model Usage in OMEGA
The Lumped Parameter Model (LPM) is used in the OMEGA model to incrementally improve
the effectiveness of vehicle models in the baseline fleet as technology packages are applied. As a
first step, approximately fifty technology packages are created with increasing effectiveness for
each vehicle type. Several example packages are shown in Table 2.51.
Table 2.51 Example OMEGA Vehicle Technology Packages (values are for example only)
Package #
Technology Package
Technology
Package
Effectiveness
0
4-Speed Auto
0%
1
6-Speed Auto
4%
2
8-Speed Auto + DCP
10%
10
8-Speed + DCP + TURB24
20%
20
8-Speed + DCP + Aero2 + TURB24 + 10%MR
28%
2-277
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Step two selects the next vehicle in the baseline fleet and applies all fifty technology packages
in sequence, using the LPM to calculate incremental effectiveness values at each step. As the
technologies in the baseline vehicles have been tabulated based on publicly available data, the
incremental effectiveness improvement will not include these baseline vehicle technologies, to
avoid double counting. Table 2.52 contains an example baseline vehicle. Table 2.53 illustrates
the package application process.
Table 2.52 Example Baseline Vehicle (values are for example only)
Baseline Vehicle Technologies
Baseline Vehicle
Effectiveness
6-Speed Auto + DCP
6%
Table 2.53 Example Package Application Process (values are for example only)
Package #
Technology Package
Technology
Package
Effectiveness
Resu Iting
Vehicle
Incremental
Effectiveness
0
4-Speed Auto
0%
0%
1
6-Speed Auto
4%
0%
2
8-Speed Auto + DCP
10%
3%
10
8-Speed + DCP + TURB24
20%
11%
20
8-Speed + DCP + Aero2 + TURB24 + 10%MR
28%
17%
As shown, the incremental effectiveness is not simply additive, as the LPM (following the
ALPHA model) takes into account synergies and dis-synergies between the existing and applied
technologies. This process also enables the OMEGA model to assign baseline vehicles a cost to
represent their existing technologies and calculate an incremental cost to match with the
incremental effectiveness as each technology package is applied. The completed technology
package effectiveness values from the LPM are compared to the corresponding ALPHA full
vehicle simulation model results as a final check before they are used in the OMEGA model. An
example subset of calibration points is shown in Table 2.54. This calibration process is an
important step to ensure that full vehicle simulation results from the ALPHA model are used as
the primary effectiveness inputs to the OMEGA model.
Table 2.54 Example Subset of ALPHA/LPM Calibration Check Points for Vehicle Type 1
Technology Package
Mass
Aero
Roll
ALPHA
Effectiveness
from Reference
Package
LPM
Effectiveness
from Reference
Package
Delta
Effectiveness
LUB+EFR1+DCP+SGDI
+6AT+HEG1+EPS+IACC1
0%
0%
0%
0.0%
0.0%
0.0%
LUB+EFR1+DCP+SGDI
+8AT+HEG1+EPS+IACC1
0%
0%
0%
7.0%
7.0%
0.0%
LUB+EFR2+ATK2+DCP
+SGDI+6AT+HEG1+EPS+IACC1
0%
0%
0%
5.0%
4.9%
-0.1%
2-278
-------
Technology Cost, Effectiveness, and Lead Time Assessment
LUB+EFR2+ATK2+DCP
+SGDI+8AT+HEG1+EPS+IACC1
0%
0%
0%
12.1%
12.0%
-0.1%
LUB+EFR2+ATK2+CEGR
+DEAC+DCP+SGDI+8AT+HEG2
+EPS+IACC2
0%
0%
0%
25.8%
25.7%
-0.1%
LUB+EFR2+TURB24
+CEGR+DEAC+DCP
+SGDI+8AT+HEG2+EPS+IACC2
0%
0%
0%
21.2%
21.0%
-0.2%
LUB+EFR2+ATK2+CEGR
+DEAC+DCP+SGDI+8AT+HEG2
+EPS+IACC2
10%
20%
20%
36.6%
36.5%
-0.1%
The complete list of baseline fleet vehicles, each incremented approximately fifty times,
results in approximately 160,000 improved vehicles as input to the OMEGA model.
The effectiveness reductions and costs that are associated with applying a technology will
depend on the starting point technologies from which the cost and effectiveness improvements
are measured. For example, two vehicle models that start with different packages of technologies
will likely have different costs and effectiveness, even if both models finally arrive at the same
package combination of technologies. EPA's recognition of the importance of clearly specifying
the point of comparison for cost and effectiveness estimates is consistent with the NAS
committee's finding "that understanding the base or null vehicle, the order of technology
application, and the interactions among technologies is critical for assessing the costs and
effectiveness for meeting the standards."
As long as the point of comparison is maintained consistently throughout the analysis for both
the baseline and future fleets, the decision of where to place an origin along the scale of cost and
effectiveness is inconsequential. For EPA's technology assessment, the origin is defined to
coincide with a "null technology package," which represents a technology floor such that all
technology packages considered in this assessment will have equal or greater effectiveness,
consistent with the Draft TAR approach. While other choices would have been equally valid, this
definition of a "null package" has the practical benefits of avoiding technology packages with
negative effectiveness values, while also allowing for a direct comparison of effectiveness
assumptions with those of the Draft TAR.
2.3.3.5.4 Appropriateness of LPMEffectiveness Modeling for the Overall Fleet
In addressing EPA's modeling methodology, several stakeholders submitted comments critical
of the LPM. The most pointed claims were aimed at the fundamental validity of using any tool
other than full vehicle simulation for the compliance analysis, with commenters contending that
"... continued use of the LPM is not an adequate or accurate tool to assess the efficacy of fuel
economy technologies applied to a wide variety of vehicles." (pg. A-l 1, Global Automakers),
and "The linear regression models within the LPM are not based on the first order determinants
of powertrain efficiency and, therefore, do not properly capture the fundamental trends." (pg. 35,
Alliance), and "the core issue with the agencies' technology effectiveness over-projections is
rooted in the 0-D LPM model itself." (Attachment 2, pg. 45, Alliance.)
EPA disagrees that the LPM, when utilized as intended, makes inaccurate predictions. The
LPM's effectiveness estimates are reliable due both to their basis in fully simulated vehicle
2-279
-------
Technology Cost, Effectiveness, and Lead Time Assessment
packages, as well as to the physical principles applied to interpolate between simulated packages.
Specifically, the use of energy loss categories within the LPM ensures that the combined benefits
of multiple technologies in a package are not double counted when two technologies are
competing to reduce the same loss. EPA continues to believe that as in the Draft TAR (as well as
in the 2012-2016 standards rulemaking, and the 2017-2025 rulemaking), when used as intended
within the bounds of the calibration, the LPM is an appropriate tool for assessing the
effectiveness of advanced technology packages for this Proposed Determination.
EPA's assessment is also supported by both the 2010 and the 2015 studies published by the
National Academy of Science - for example, in the 2015 report, the NAS stated, "The committee
notes that the use of full vehicle simulation modeling in combination with lumped parameter
modeling and teardown studies contributed substantially to the value of the Agencies' estimates
of fuel consumption and costs, and it therefore recommends they continue to increase the use of
these methods to improve their analysis. "EEE Note that both the 2010 and the 2015 NAS
Committees specifically evaluated earlier versions of the EPA-developed LPM that informed the
Committee's findings and recommendations.
In comments submitted on the Draft TAR, the Alliance stated that EPA's modeling processes
"do not recognize the inherent variability of efficiency within the light-duty fleet, treating all
products within a category as equal" and recommended that the "LPM should be enhanced and
upgraded to incorporate the key vehicle and powertrain parameters which determine powertrain
efficiency." While the degree of resolution in EPA's effectiveness modeling in the FRM and
Draft TAR was sufficient to distinguish between individual models, and to enable the application
of unbiased effectiveness estimates within a reasonably narrow range, EPA has taken several
steps for this Proposed Determination in response to the recommendation from commenters that
the precision of the effectiveness modeling be further improved. First, and most significantly, as
described in Chapter 2.3.3.2, EPA has refined the ALPHA classes used for grouping vehicles
according to the attributes that most directly influence technology effectiveness: namely power-
to-weight ratio and road load horsepower. As discussed in that chapter, this refinement has
significantly reduced the variation between vehicles in each ALPHA class, thus improving the
precision of the modeling. As an additional refinement, EPA has also incorporated the
consideration of each individual vehicle's power-to-weight ratio into the effectiveness numbers
produced by the LPM. Using a set of relationships between power-to-weight ratio and
effectiveness produced by the ALPHA model, EPA is now applying an effectiveness adjustment
in the OMEGA process based on the deviation in the power-to-weight value from the exemplar
vehicle in that class, as illustrated above in Figure 2.90 using the coefficients in Table 2.55 and
Equation 3.
Table 2.55: Parameters for Power-to-Weight Adjustment of Effectiveness Values in OMEGA
Lower PW Range
Mid/Upper PW Range
Upper PW Range (HPW only)
EEE See Finding 8.7 and 10.12 and Recommendation 8.3 of "Cost, Effectiveness and Deployment of Fuel Economy
Technologies for Light-Duty Vehicles published by the Committee on the Assessment of Technologies for
Improving Fuel Economy of Light-duty Vehicles"; Phase 2; Board on Energy and Environmental Systems;
Division on Engineering and Physical Sciences; National Research Council, ISBN 978-0-309-37388-3, 2015.
See also Chapter 8 (page 118) of "Assessment of Fuel Economy Technologies for Light-Duty Vehicles";
Committee on the Assessment of Technologies for Improving Light-Duty Vehicle Fuel Economy; National
Research Council; ISBN 978-0-309-15607-3, 2010.
2-280
-------
Technology Cost, Effectiveness, and Lead Time Assessment
ALPHA Class
Eng Tech
PW Cutoff
<
(hp/lOOlb)
m
b
(*le2)
PW Cutoff
>
(hp/lOOlb)
m
b
(*le2)
PW Cutoff
>
(hp/lOOlb)
m
b
(*le2)
LPW_LRL
Turbo*
419
1.2
4.19
4.19
1.2
4.19
-
-
-
ATK2
4.19
-0.6
4.19
4.19
-1.0
4.19
-
-
-
ATK2+CEGR
4.19
1.2
4.19
4.19
1.0
4.19
-
-
-
MPW_LRL
Turbo*
5.25
1.8
5.25
5.25
1.8
5.25
-
-
-
ATK2
5.25
-0.5
5.25
5.25
-0.5
5.25
-
-
-
ATK2+CEGR
5.25
1.2
5.25
5.25
1.0
5.25
-
-
-
HPW
Turbo*
7.14
2.0
7.14
7.14
1.6
7.14
11.0
0.8
3.29
ATK2
7.14
-1.4
7.14
7.14
-0.5
7.14
11.0
-0.2
1.36
ATK2+CEGR
7.14
1.0
7.14
7.14
0.6
7.14
11.0
0.2
-0.568
LPW_HRL
Turbo*
4.46
1.5
4.46
4.46
1.5
4.46
-
-
-
ATK2
4.46
-0.6
4.46
4.46
-1.0
4.46
-
-
-
ATK2+CEGR
4.46
1.2
4.46
4.46
1.0
4.46
-
-
-
MPW_HRL
Turbo*
5.68
2.0
5.68
5.68
2.0
5.68
-
-
-
ATK2
5.68
-0.5
5.68
5.68
-0.5
5.68
-
-
-
ATK2+CEGR
5.68
1.2
5.68
5.68
1.0
5.68
-
-
-
Truck
Turbo*
6.10
2.0
6.10
6.10
2.0
6.10
-
-
-
ATK2
6.10
-0.7
6.10
6.10
-0.7
6.10
-
-
-
ATK2+CEGR
6.10
1.2
6.10
6.10
1.0
6.10
-
-
-
*Note: Turbocharged Miller Cycle engines are classified as turbocharged
Equation 3. Effectiveness adjustment relative to exemplar
Effectiveness Adjustment, relative to exemplar = m(PW-b)
To illustrate how each individual vehicle's power-to-weight ratio is accounted for in the
effectiveness estimates used in the OMEGA process, an example of a vehicle in the HPW
ALPHA class is provided here (Baseline Index 2264). For that vehicle, a technology package
with a turbocharged engine (TP06) is applied in the OMEGA model's compliance analysis of the
2025 standards. The baseline vehicle's power-to-weight ratio (PW) of 6.92 hp/lOOlb is less than
the 7.14 hp/lOOlb value for the exemplar vehicle in that class (as defined in Table 2.47),
indicating that some effectiveness adjustment may be justified. Applying Equation 3 and the
Turbo technology values of m = 2.0 and b = 7.14 hp/lOOlb from Table 2.55, an adjustment of -
0.44% (reduction in effectiveness) is applied for this technology package, relative to the
effectiveness value produced by the LPM for the HPW exemplar vehicle.
Comments received on the Draft TAR also focused on the processes used by EPA to assure
the reliability and accuracy of the modeling tools. The Alliance stated that "[N]o procedure or
methodology is currently in place to check the outcomes of the [LPM's] technology
effectiveness projection process against logical efficiency metrics and limits. Without such
checks, the outcomes can exceed plausible limits" (pg. 44, Alliance comments). EPA does not
agree that the processes used for the Draft TAR did not involve plausibility checks. The LPM
has been calibrated to, and is bounded by, ALPHA Full Vehicle Simulation Model results. It was
not used to predict anything beyond the bounds of these fundamental inputs. The specific
plausibility limits recommended by the Alliance are based on a top-down empirical analysis of
existing vehicles, and do not reflect the fundamental efficiency improvements that are enabled
through physical technology changes in the future fleet. For the reasons described further in
2-281
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Appendix A, EPA is not considering any of the plausibility limits imposed by the three metrics
proposed by the Alliance. At the same time, EPA agrees that quality assurance processes are
important for ensuring the validity of any modeling. For this Proposed Determination, EPA has
adopted one of the quality assurance tools recommended in the Alliance-contracted report.
Using the methodology described in TSD Appendix B, EPA has calculated a measure of
powertrain efficiency, defined as the ratio of the tractive work done to move a vehicle over the
test cycle to the fuel energy utilized over the same cycle. Figure 2.98 shows the production-
weighted distribution of powertrain efficiencies for gasoline-fueled vehi cles in the MY2015
fleet, excluding vehicles equipped with electrified powertrains. Stop-start technology, while an
effective technology for reducing emissions, is also excluded from the following figure and
discussions, since its benefits are independent of powertrain operation efficiency.
ALPHA Class
All ALPHA Classes
MPW LRL
MPW HRL
Truck
16% 17% 18% 19% 20% 21% 22% 23% 24% 25% 26% 27% 28% 29% 30% 31% 32% 33% 34%
Powertrain Efficiency
Figure 2.98 Distribution of Gasoline Powertrain Efficiencies for Vehicles in MY2015
Using the same selection criteria (i.e. gasoline engines, excluding stop-start) for the
technology packages applied in the OMEGA model's compliance analysis of the MY2025 GHG
standards, the powertrain efficiencies shown in Figure 2.99 are, in general, higher than the
powertrain efficiencies of the MY2015 baseline fleet, as would be expected from the application
of advanced technology packages.
2-282
-------
Technology Cost, Effectiveness, and Lead Time Assessment
ALPHA
Class
-All ALPHA Classes
- LPW_LRL
- MPWJ.RL
- HPW
LPW_HRL
MPW_HRL
-Truck
16% 17% 18% 19% 20% 21% 22% 23% 24% 25% 26% 27% 28% 29% 30% 31% 32% 33% 34%
Powertrain Efficiency
Figure 2.99 Distribution of Gasoline Powertrain Efficiencies for Vehicles in the OMEGA Compliance
Analysis for MY2025 Standards
As shown in Table 2.56, the fleet median (production weighted) powertrain efficiency for
gasoline non-stop-start vehicles increases from 21.6 percent in the MY2015 baseline fleet, to
26.8 percent in EPA's compliance analysis for the fleet meeting MY2025 standards. Contrary to
the assertion made by The Alliance in comments to the Draft TAR that EPA's modeling
processes "do not recognize the inherent variability of efficiency within the light-duty fleet,
treating all products within a category as equal," the fleet produced by OMEGA's cost-
minimizing compliance pathway is similar to the MY2015 baseline fleet in the degree of
diversity in powertrain efficiencies among vehicles, as indicated by both the similarities in
ranges between the minimum and maximum efficiencies, and the shapes of the distributions.
Table 2.56 Summary Statistics for Powertrain Efficiencies in MY2015 Baseline and OMEGA Compliance
Analysis of MY2025 Standards
ALPHA Class
MY2015 Baseline Fleet
OMEGA Compliance Fleet
Meeting MY2025 Standards
Median
Std.
Dev.
min-max
Median
Std.
Dev.
min-max
LPW LRL
22.1
1.3
17.8-25.5
26.5
2.3
21.1-31.9
MPW LRL
21.5
1.3
18.7-25.3
26.0
1.9
20.4-29.8
HPW
19.8
1.6
11.5-23.8
25.1
2.3
13.9-28.8
LPW HRL
22.9
1.4
17.6-25.6
28.2
1.7
21.1-31.5
MPW HRL
21.6
1.1
17.9-25.7
27.2
0.9
22.4-30.0
Truck
22.1
1.3
17.5-25.8
27.2
1.5
21.8-29.3
Fleet
21.6
1.7
11.5-25.8
26.8
2.1
13.9-31.9
Table 2.57 shows the vehicles and technology packages with the highest powertrain
efficiencies modeled in EPA's 2025 compliance analysis. EPA does not believe that for this
relatively small portion of the fleet (comprising approximately 6 percent of the volume of
gasoline non-stop-start packages in the compliance analysis for MY2025 standards) that
2-283
-------
Technology Cost, Effectiveness, and Lead Time Assessment
powertrain efficiency values greater than 30 percent indicate a systemic overestimation of
technology effectiveness.
With the exception of only two vehicles, the majority of future technology packages with high
powertrain efficiencies in MY2025 are associated with vehicles that had high efficiencies in the
MY2015 baseline. There are several possible explanations for this. First, although EPA's
technology assessment accounts for the presence of efficiency technologies in the baseline file,
there is insufficient data available to model the exact technologies applied to each individual
vehicle. Because EPA has selected baseline technology parameters that are representative of
typical MY2015 vehicles (e.g. based on benchmarking of current ATK2 and GDI engines,
current transmissions, as well as characterization of road load technologies) the characterization
of baseline technologies in this Proposed Determination will not systemically over- or under-
estimate the incremental effectiveness of advanced technology packages. However, just as it is
possible that some baseline vehicles will have less effective technology implementations than is
typical of other vehicles in the MY2015 fleet, it is also possible that the vehicles shown in Table
2.57 may have more efficient technology implementation than is typical. Second, it is also
possible that when grouping vehicles together for certification, an OEM's application of road
load coefficients, ETW values, and emissions levels may be representative overall of the
certified model type, but deviate from the actual values for a particular vehicle. Any
discrepancies between the parameters used to calculate tractive energy (ETW, road load
coefficients) and the measured fuel consumption over the test cycle would potentially result in
high baseline powertrain efficiencies, and thus carry-over into high powertrain efficiencies of
future technology packages. Again, this variation would not tend to result in a systemic over- or
under-estimation of effectiveness.
Table 2.57 Summary Statistics for Powertrain Efficiencies in MY2015 Baseline and OMEGA Compliance
Analysis of MY2025 Standards
ALPHA
Class
MY2025 Compliance
Analysis
MY2015 Baseline
Baseline
Index
Tech Pkg.
Model
Powertrain
Efficiency
Percentile
(in class)
Powertrain
Efficiency
Percentile
(in class)
LPW_LRL
1561
TP08
Veloster
31.5%
93.8%
23.1%
85.0%
1510
TP08
Elantra
31.9%
100.0%
22.8%
58.2%
LPW_HRL
1737
TP08
CX-5 4WD
31.5%
100.0%
24.8%
87.4%
2056
TP10
City Express Cargo Van
30.2%
83.2%
24.0%
72.3%
2151
TP10
NV200 Cargo Van
30.1%
83.0%
24.0%
72.0%
1371
TP08
TERRAIN FWD
31.2%
99.9%
23.3%
55.4%
1220
TP08
EQUINOX FWD
31.1%
98.9%
23.3%
54.1%
1180
TP08
CAPTIVA FWD
31.1%
95.6%
23.3%
50.8%
2304
TP08
RAV4 Limited AWD
30.8%
94.2%
22.7%
40.6%
2286
TP08
HIGHLANDER
30.0%
82.5%
22.3%
26.1%
MPW_HRL
733
TP07
EXPLORER FWD
30.0%
100.0%
24.5%
98.4%
Table 2.58 shows a selection of vehicles throughout the distribution of powertrain efficiencies
in the OMEGA compliance analysis for the MY2025 standards. For this Proposed
Determination, EPA has incorporated the powertrain efficiency metric into the effectiveness
modeling Quality Control processes in order to identify possible anomalies in how the
effectiveness estimates generated by the ALPHA physics-based model are represented in the
2-284
-------
Technology Cost, Effectiveness, and Lead Time Assessment
LPM and OMEGA process. For each of the six ALPHA classes, six vehicles were chosen
throughout the distribution of powertrain efficiencies; including the vehicle with maximum
powertrain efficiency in each class (i.e. 100th percentile).
Table 2.58 Powertrain Efficiencies by ALPHA Class from MY2025 OMEGA Compliance Analysis
Approx.
Percentile
(in class)
ALPHA
Class
Baseline
Index
Model
Tech
Pkg
Percentile
(in class)
Powertrain
Efficiency
10
HPW
1012
MUSTANG
TP07
7.70%
21.60%
LPW_HRL
2193
IMPREZA
TP05
5.20%
26.10%
LPW_LRL
2277
COROLLA
TP05
10.50%
23.80%
MPW_HRL
1707
Sorento FWD
TP09
0.80%
25.40%
MPW_LRL
1414
ACCORD
TP07
10.10%
24.50%
Truck
773
F150 PICKUP 2WD
TP05
9.90%
25.30%
25
HPW
1423
ACCORD
TP06
26.50%
24.60%
LPW_HRL
1556
Tucson AWD
TP12
11.60%
26.40%
LPW_LRL
1443
CIVIC
TP06
31.90%
24.90%
MPW_HRL
1494
ODYSSEY 2WD
TP06
24.70%
26.50%
MPW_LRL
1403
ACCORD
TP07
24.10%
24.60%
Truck
2124
FRONTIER 2WD
TP08
25.00%
26.70%
50
HPW
2149
MURANO AWD
TP08
51.00%
25.70%
LPW_HRL
2044
OUTLANDER 2WD
TP 10
57.40%
28.10%
LPW_LRL
1785
MAZDA3 5-Door
TP05
52.80%
25.90%
MPW_HRL
711
ESCAPE AWD
TP07
43.90%
26.90%
MPW_LRL
1687
OPTIMA
TP09
28.00%
25.10%
Truck
2132
FRONTIER 2WD
TP05
54.00%
27.30%
75
HPW
1380
MDX 4WD
TP07
77.30%
26.50%
LPW_HRL
1483
CR-V 4WD
TP08
76.50%
29.30%
LPW_LRL
1510
Elantra
TP07
68.00%
27.80%
MPW_HRL
1498
PILOT 4WD
TP06
62.00%
27.10%
MPW_LRL
695
EDGE FWD
TP07
74.90%
27.20%
Truck
2316
TACOMA 2WD
TP06
74.20%
28.10%
90
HPW
1122
CTS SEDAN AWD
TP08
87.30%
27.20%
LPW_HRL
2302
RAV4 AWD
TP08
91.90%
30.60%
LPW_LRL
1661
Forte
TP08
90.40%
29.80%
MPW_HRL
2236
NX 200t
TP07
88.10%
28.60%
MPW_LRL
1665
Forte
TP08
88.00%
28.60%
Truck
2324
TACOMA 2WD
TP05
84.20%
28.70%
100
HPW
1369
TERRAIN AWD
TP08
100.00%
28.80%
LPW_HRL
1738
CX-5 4WD
TP08
100.00%
31.50%
LPW_LRL
1510
Elantra
TP08
100.00%
31.90%
2-285
-------
Technology Cost, Effectiveness, and Lead Time Assessment
MPWJHRL
733
EXPLORER FWD
TP07
100.00%
30.00%
MPW_LRL
1555
Sonata
SPORT/LIMITED
TP08
100.00%
29.80%
Truck
1319
CANYON 2WD
TP08
96.90%
28.90%
For each of the vehicles modeled using the LPM and OMEGA process shown in Table 2.58,
EPA applied the ALPHA model using the road load coefficients, rated horsepower, and ETW of
the MY2015 baseline vehicle, along with the technologies corresponding to the TPOO technology
package. The ALPHA model was then used to represent the future technology packages shown
in Table 2.58, including the related mass reduction and reductions in road loads relative to the
baseline package. The results, shown in Figure 2.100, confirm that the LPM is able to reliably
replicate the effectiveness values generated by the physics-based ALPHA model (within 2%)
over a wide range of vehicle classes, technologies, and powertrain efficiency values.
45%
40%
in
0)
4-1
f0
10
LU
i/l
i/l
4->
u
sB
LU
CL
30%
25%
10%
1738_TP08
1380 TP07 -
35% 2124_TP08- 2236_TP07
2044_TP10 -
1687 TP09 -
2302_TP08 - 2324_TP05
2132_TP05
2149_TP08
1665_TP08 -
1319 TP08
1707 TP09
«H556_TP12
1012_TP07
1555 TP08
1369_TP08
1122 TP08
1510 TP07
15%
2316 TP06
1403_TP07
20% 1414_TP07-
1785 TP05-
- 1510_TP07
1443 TP06
2193_TP05
2277_TP05
10% 15% 20% 25% 30% 35%
ALPHA Effectiveness Estimates
|- 1483_TP08 1661_TP08
- 1423_TP06 _ 1498_TP06
^ 711_TP07 695_TP07
1494_TP06
- 733_TP07
773 TP05
40%
45%
Figure 2.100 LPM and ALPHA Package Effectiveness Comparison for Vehicles and Throughout Distribution
of Powertrain Efficiencies
2-286
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.3.4 Data and Assumptions Used in the GHG Assessment
2.3.4.1 Engines: Data and Assumptions for this Assessment
The majority of engine technologies used in this assessment are detailed in Chapter 2.2 of this
TSD. This section details engine technology information specific to the Proposed Determination
analysis.
In an effort to characterize the efficiency and performance of late model vehicle powertrains,
and to update our engine data from that used in the FRM and Draft TAR, EPA tested several
engines at its National Vehicle and Fuel Emissions Laboratory (NVFEL) and contractor
facilities. Depending on the information required, the engines were tested with their factory
and/or developmental engine management systems that allowed EPA engineering staff to
calibrate engine control parameters. Figure 2.101 illustrates a typical engine test.
Figure 2.101 2.0L 14 Mazda SKYACTIV-G Engine Undergoing Engine Dynamometer Testing at the EPA-
NVFEL Facility.
In some cases, future engine configurations can be modeled using engine simulation software.
EPA used Gamma Technologies GT-POWER engine simulation software to model future engine
configurations based upon the Mazda 2.0L 14 SKYACTIV-G. Computer-aided engineering
tools, including GT-POWER, are commonly used during the initial stages of product
development by automotive manufacturers and academia to establish the potential performance
of engine design features, with respect to efficiency, emissions, and performance. GT-POWER
is a physics based suite of software that combines predictive diesel or spark-ignition combustion
models; CAD-based, preprocessed libraries of the physical layout of induction, exhaust and
combustion systems; models of chemical kinetics; wave dynamics models; turbocharger turbine
and compressor models with surge, reverse-flow and pressure wave prediction; induction
turbulence models; a kinetic knock model; injector spray models and an ability to apply minor
adjustments to model-predicted parameters using data from engine dynamometer measurements.
Engine dynamometer data was also used to directly validate simulations of specific engine
hardware configurations via comparisons of measured vs. modeled values for knock intensity,
combustion phasing, FMEP, BTE and other parameters.
2.3.4.1.1 Low Friction Lubricants (LUB)
2-287
-------
Technology Cost, Effectiveness, and Lead Time Assessment
There were no public comments received with supporting data that would provide basis for
changing the cost or effectiveness estimates for this technology, nor has EPA found additional
information that supports such a change since the Draft TAR. Based on the analysis for the Draft
TAR, the agencies estimated the effectiveness of LUB to be 0.5 to 0.8 percent. EPA has
reviewed this technology and finds the effectiveness estimate remains applicable for this
Proposed Determination.
The cost associated with making the engine changes needed to accommodate low friction
lubes is equivalent to that used in the Draft TAR, updated to 2015 dollars. The costs are shown
below.
Table 2.59 Costs for Engine Changes to Accommodate Low Friction Lubes (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
S3
1
S3
S3
S3
S3
S3
S3
S3
S3
S3
IC
Low2
2018
Si
Si
Si
Si
Si
Si
Si
Si
Si
TC
$4
$4
$4
$4
$4
$4
$4
$4
$4
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.1.2 Engine Friction Reduction (EFR1. EFR2)
There were no public comments received with supporting data that would provide basis for
changing the cost or effectiveness estimates for this technology, nor has EPA found additional
information that supports such a change since the Draft TAR. Based on the analysis for the Draft
TAR, EPA estimated the effectiveness of EFR1 at 2.0 to 2.7 percent. Based on the analysis for
the Draft TAR, EPA estimated the effectiveness of EFR2 at 3.4 to 4.8 percent. EPA has
reviewed this technology and finds the effectiveness estimate remains applicable for this
Proposed Determination.
The costs associated with engine friction reduction are equivalent to those used in the Draft
TAR, updated to 2015 dollars. The costs are shown below first for engine friction reduction level
1 and then for level 2.
Table 2.60 Costs for Engine Friction Reduction Level 1 (dollar values in 2015$)
Engine
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
13
DMC
$38
1
$38
$38
$38
$38
$38
$38
$38
$38
$38
14
DMC
$51
1
$51
$51
$51
$51
$51
$51
$51
$51
$51
V6
DMC
$77
1
$77
$77
$77
$77
$77
$77
$77
$77
$77
V8
DMC
$102
1
$102
$102
$102
$102
$102
$102
$102
$102
$102
13
IC
Low2
2018
$9
$9
$7
$7
$7
$7
$7
$7
$7
14
IC
Low2
2018
$12
$12
$10
$10
$10
$10
$10
$10
$10
V6
IC
Low2
2018
$19
$19
$15
$15
$15
$15
$15
$15
$15
V8
IC
Low2
2018
$25
$25
$20
$20
$20
$20
$20
$20
$20
13
TC
2018
$48
$48
$46
$46
$46
$46
$46
$46
$46
14
TC
2018
$63
$63
$61
$61
$61
$61
$61
$61
$61
V6
TC
2018
$95
$95
$91
$91
$91
$91
$91
$91
$91
V8
TC
2018
$127
$127
$122
$122
$122
$122
$122
$122
$122
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.61 Costs for Engine Friction Reduction Level 2 (dollar values in 2015$)
Engine
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
13
DMC
$84
1
$84
$84
$84
$84
$84
$84
$84
$84
$84
2-288
-------
Technology Cost, Effectiveness, and Lead Time Assessment
14
DMC
$109
1
$109
$109
$109
$109
$109
$109
$109
$109
$109
V6
DMC
$160
1
$160
$160
$160
$160
$160
$160
$160
$160
$160
V8
DMC
$211
1
$211
$211
$211
$211
$211
$211
$211
$211
$211
13
IC
Low2
2024
$20
$20
$20
$20
$20
$20
$20
$20
$16
14
IC
Low2
2024
$26
$26
$26
$26
$26
$26
$26
$26
$21
V6
IC
Low2
2024
$39
$39
$39
$39
$39
$39
$39
$39
$31
V8
IC
Low2
2024
$51
$51
$51
$51
$51
$51
$51
$51
$41
13
TC
2024
$104
$104
$104
$104
$104
$104
$104
$104
$100
14
TC
2024
$135
$135
$135
$135
$135
$135
$135
$135
$130
V6
TC
2024
$199
$199
$199
$199
$199
$199
$199
$199
$191
V8
TC
2024
$262
$262
$262
$262
$262
$262
$262
$262
$252
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.1.3 Cylinder Deactivation (DEAC)
In the Draft TAR analysis, EPA estimated an effectiveness of 3.9 to 5.3 percent for fixed
cylinder deactivation.
In comments on the Draft TAR, UCS commented that EPA's effectiveness estimates for
DEAC appeared conservative and cited a recent paper by ICCT that estimated fixed cylinder
deactivation effectiveness as high as 6.5 percent. However, UCS also commented that NHTSA's
estimate of 5 to 9 percent incremental effectiveness (on an engine already having VVT, VVL,
and stoichiometric GDI) may be too aggressive.
EPA notes that our estimated effectiveness applies to fixed cylinder deactivation and is within
the 3 to 7 percent range that ICCT notes was cited in the Draft TAR. While consideration of
more advanced rolling dynamic systems (such as those under development by Schaeffler and
Tula) might likely increase the estimated effectiveness, EPA notes that the system costs for fixed
systems are lower. Rolling dynamic systems were not used by EPA to build packages for this
analysis.
In their comments, FCA asserted that EPA was overly optimistic on relying on the availability
of cylinder deactivation (DEAC) at unrealistic speed / load operating points. EPA based the
speed and load operating points and availability of cylinder deactivation primarily upon
benchmarking of a production MY2015 General Motors "Ecotec3" V6 naturally aspirated GDI
light-truck engines equipped with coupled cam phasing and cylinder deactivation. The resulting
effectiveness estimates based upon this data are somewhat conservative and within the lower
range of effectiveness within published literature.559'560 Over the range of engine speed
(approximately 1000 to 3000 rpm) and BMEP (approximately 1 to 5 bar), effectiveness was
further reduced to reflect that cylinder deactivation could not occur 100 percent of the time
within the area that it is active. It was assumed that cylinder deactivation would only occur 60
percent of the time within the engine speed and BMEP window where cylinder deactivation was
active. Again, this was a conservative estimate based upon benchmarking of the MY2015
General Motors "Ecotec3" V6. It did not take into consideration further improvements in NVH
abatement under development to increase the percentage of vehicle operation under which
cylinder deactivation can be enabled which would reasonably be expected to be in production for
MY2022-MY2025 vehicles.560
AAM provided no data on the range of engine speeds, BMEP or other factors impacting
availability of DEAC, nor did AAM provide any specific critique regarding how EPA conducted
the benchmarking of the production General Motors engine equipped with DEAC. AAM also
2-289
-------
Technology Cost, Effectiveness, and Lead Time Assessment
did not discuss technologies under advanced stages of development to improve the availability of
DEAC (NVH abatement measures for fixed and dynamic DEAC), as discussed by EPA within
the draft TAR. Consequently, EPA has been presented with no valid basis for changing its
efficiency estimate for DEAC.
AAM and FCA commented that DEAC should not be applied in conjunction with cooled
EGR on ATK2 engines and that the effectiveness that EPA assumed for DEAC when applied to
turbo-charged downsized engines was too high. Neither FCA nor AAM provided any data or
detailed description of why DEAC could not be applied in conjunction with cEGR on ATK2
engines.
EPA notes that the effectiveness that EPA assumes for DEAC applied to turbocharged,
downsized 13 and 14 engines is comparable to the effectiveness demonstrated by Ford and
Schaeffler for applying fixed DEAC to a turbocharged 13 engine in a paper presented at the 2015
Vienna Motor symposium.560 Mazda also presented data at the 2015 Vienna Motor Symposium
showing data from a SKYACTIV-G DEAC system at an advanced stage of development and
has already publicly shared data on a version of the SKYACTIV-G engine with cylinder
deactivation561'562 with effectiveness comparable to EPA estimates for applying DEAC to ATK2,
and has discussed the future application of cylinder deactivation to their SKYACTIV-G engines
with the automotive press.563'564 Engine modeling by EPA and initial hardware testing appear to
show synergies between the use of cEGR and DEAC with Atkinson Cycle engines. Mazda has
used cEGR with previous applications of their SKYACTIV-G engine and cEGR is currently
used by Toyota and Hyundai in Atkinson Cycle engines for both HEV and non-HEV
applications. VW has already introduced a 4-cylinder Miller Cycle engine, the EA211 TSI®
evo, which combines DEAC, cEGR, EIVC and turbocharging.
EPA has reviewed this information and the comments submitted. It is our assessment that the
effectiveness estimates used in the Draft TAR analysis for DEAC remain appropriate for this
Proposed Determination analysis.
The costs associated with cylinder deactivation for this Proposed Determination analysis are
shown in Table 2.62 and are equivalent to those used in the Draft TAR but updated to 2015
dollars. .
Table 2.62 Costs for Cylinder Deactivation (dollar values in 2015$)
Engine
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
14
DMC
$88
24
$85
$83
$81
$80
$79
$78
$77
$76
$75
V6
DMC
$157
24
$150
$147
$145
$142
$140
$138
$136
$134
$133
V8
DMC
$177
24
$169
$166
$163
$160
$158
$155
$153
$151
$149
14
IC
Highl
2018
$50
$49
$30
$30
$30
$30
$30
$30
$30
V6
IC
Med2
2018
$61
$60
$45
$45
$45
$45
$45
$45
$45
V8
IC
Med2
2018
$68
$68
$51
$51
$51
$51
$51
$50
$50
14
TC
$134
$132
$112
$110
$109
$108
$107
$106
$105
V6
TC
$211
$208
$190
$187
$185
$183
$181
$179
$177
V8
TC
$237
$234
$214
$211
$208
$206
$204
$202
$200
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.1.4 Intake Cam Phasing (ICP)
2-290
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Within the analysis for the Draft TAR, EPA estimated an effectiveness of 2.1 to 2.7 percent
for ICP. Toyota commented that EPA's estimate of ICP effectiveness is too high because "ICP
effectiveness differs from combination of engine displacement, road load (R/L), T/M, and open
duration setting of intake air camshaft." However, the comment did not share specific data on
engines, specific camshaft phasing hardware and resultant effectiveness relative to hardware
used, making it difficult to further assess the claim. EPA notes that the effectiveness data used in
the Draft TAR is consistent with published and peer-reviewed data cited in the FRM, the Draft
TAR and this Proposed Determination, and reflects performance consistent with the range of
authority for ICP and DCP hardware for which cost estimates were developed. EPA therefore
believes that the effectiveness estimate used in the Draft TAR remains applicable for this
Proposed Determination.
The costs associated with intake cam phasing are equivalent to those used in the Draft TAR,
updated to 2015 dollars. The costs are shown below.
Table 2.63 Costs for Intake Cam Phasing (dollar values in 2015$)
Engine
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
OHC-I
DMC
$42
24
$40
$39
$38
$38
$37
$37
$36
$36
$35
OHC-V
DMC
$84
24
$80
$78
$77
$76
$74
$73
$72
$71
$70
OHV-V
DMC
$42
24
$40
$39
$38
$38
$37
$37
$36
$36
$35
OHC-I
IC
Low2
2018
$10
$10
$8
$8
$8
$8
$8
$8
$8
OHC-V
IC
Low2
2018
$20
$20
$16
$16
$16
$16
$16
$16
$16
OHV-V
IC
Low2
2018
$10
$10
$8
$8
$8
$8
$8
$8
$8
OHC-I
TC
$50
$49
$46
$46
$45
$45
$44
$44
$43
OHC-V
TC
$100
$98
$93
$92
$90
$89
$88
$87
$86
OHV-V
TC
$50
$49
$46
$46
$45
$45
$44
$44
$43
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.1.5 Dual Cam Phasing (DCP)
Based on the analysis for the Draft TAR, EPA estimated the effectiveness of DCP to be
between 4.1 to 5.5 percent.
In comments on the Draft TAR, Toyota suggested that EPA's estimate of DCP effectiveness
was too high "to account for DCP effectiveness differences resulting from the combination of
engine displacement, R/L, TIM, and open duration setting of intake air camshaft. Similar to the
comment on intake cam phasing cited above, the comment did not share specific data on engines,
specific camshaft phasing hardware and resultant effectiveness relative to hardware used, making
it difficult to further assess the claim. EPA notes that the effectiveness data used in the Draft
TAR is consistent with published and peer-reviewed data cited in the FRM, the Draft TAR and
this Proposed Determination, and reflects performance consistent with the range of authority for
ICP and DCP hardware for which cost estimates were developed. EPA therefore believes that
the effectiveness estimate used in the Draft TAR remains applicable for this Proposed
Determination.
The costs associated with dual cam phasing are equivalent to those used in the Draft TAR,
updated to 2015 dollars. The costs are shown below.
2-291
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.64 Costs for Dual Cam Phasing (dollar values in 2015$)
Engine
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
OHC-I
DMC
$77
24
$73
$72
$71
$69
$68
$67
$66
$65
$65
OHC-V
DMC
$164
24
$157
$154
$151
$149
$146
$144
$142
$140
$139
OHC-I
IC
Med2
2018
$30
$29
$22
$22
$22
$22
$22
$22
$22
OHC-V
IC
Med2
2018
$63
$63
$47
$47
$47
$47
$47
$47
$47
OHC-I
TC
$103
$101
$93
$91
$90
$89
$88
$87
$86
OHC-V
TC
$221
$217
$199
$196
$194
$191
$189
$187
$186
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.1.6 Discrete Variable Valve Lift (DWL)
Based on the analysis for the Draft TAR, EPA estimated the effectiveness for DVVL at 4.1 to
5.6 percent.
In comments on the Draft TAR, Toyota suggested that EPA's estimate of DVVL effectiveness
was too high because "DVVL effectiveness differs from combination of engine displacement,
road load (R/L), T/M, and open duration setting of intake air camshaft." Similar to the comments
on intake cam phasing and dual cam phasing cited above, the comment did not share specific
data on engines, specific camshaft phasing hardware and resultant effectiveness relative to
hardware used, making it difficult to further assess the claim. EPA notes that the effectiveness
data used in the Draft TAR is consistent with published and peer-reviewed data cited in the
FRM, the Draft TAR and this Proposed Determination, and reflects performance consistent with
DVVL hardware for which cost estimates were developed. EPA therefore believes that the
effectiveness estimate used in the Draft TAR remains applicable for this Proposed
Determination.
The costs associated with discrete variable valve lift are equivalent to those used in the Draft
TAR, updated to 2015 dollars. The costs are shown below.
Table 2.65 Costs for Discrete Variable Valve Lift (dollar values in 2015$)
Engine
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
OHC-I
DMC
$131
24
$125
$123
$121
$119
$117
$115
$113
$112
$111
OHC-V
DMC
$190
24
$182
$178
$175
$172
$169
$167
$165
$162
$160
OHV-V
DMC
$271
24
$260
$255
$250
$246
$242
$238
$235
$232
$229
OHC-I
IC
Med2
2018
$50
$50
$38
$38
$38
$37
$37
$37
$37
OHC-V
IC
Med2
2018
$73
$73
$55
$54
$54
$54
$54
$54
$54
OHV-V
IC
Med2
2018
$105
$104
$78
$78
$78
$78
$78
$77
$77
OHC-I
TC
$176
$173
$158
$156
$154
$153
$151
$149
$148
OHC-V
TC
$255
$251
$230
$227
$224
$221
$219
$217
$214
OHV-V
TC
$364
$359
$328
$324
$320
$316
$313
$309
$306
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.1.7 Continuously Variable Valve Lift (CWL)
Based on the analysis for the Draft TAR, EPA estimated the effectiveness for CVVL at 5.1 to
7.0 percent.
In comments on the Draft TAR, Toyota suggested that EPA's estimate of CVVL effectiveness
was too high, citing "the same reasons cited above" with regard to ICP, DCP, and DVVL. Other
than making a general statement, the comment did not share specific data on engines, specific
2-292
-------
Technology Cost, Effectiveness, and Lead Time Assessment
hardware and resultant effectiveness relative to hardware used, making it difficult to further
assess the claim. EPA notes that the effectiveness data used in the Draft TAR is consistent with
published and peer-reviewed data cited in the FRM, the Draft TAR and this Proposed
Determination, and reflects performance consistent with CVVL hardware for which cost
estimates were developed. EPA therefore believes that the effectiveness estimate used in the
Draft TAR remains applicable for this Proposed Determination.
The costs associated with continuously variable valve lift are equivalent to those used in the
Draft TAR, updated to 2015 dollars. The costs are shown below.
Table 2.66 Costs for Continuously Variable Valve Lift (dollar values in 2015$)
Engine
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
OHC-I
DMC
$197
24
$188
$184
$181
$178
$175
$173
$170
$168
$166
OHC-V
DMC
$360
24
$345
$338
$332
$326
$321
$316
$312
$308
$304
OHV-V
DMC
$393
24
$376
$369
$362
$356
$350
$345
$340
$336
$332
OHC-I
IC
Med2
2018
$76
$76
$56
$56
$56
$56
$56
$56
$56
OHC-V
IC
Med2
2018
$139
$139
$104
$103
$103
$103
$103
$103
$103
OHV-V
IC
Med2
2018
$151
$151
$113
$113
$113
$112
$112
$112
$112
OHC-I
TC
$264
$260
$237
$234
$231
$229
$226
$224
$222
OHC-V
TC
$484
$477
$435
$430
$424
$419
$415
$411
$407
OHV-V
TC
$527
$520
$475
$469
$463
$458
$453
$448
$444
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.1.8 Atkinson Cycle Engines in Non-HEVApplications
In the last few years, a new generation of naturally aspirated SI Atkinson Cycle engines
applicable outside of HEVs have been introduced into light-duty vehicle applications. The most
prominent application of this technology is the Mazda SKYACTIV-G system. It combines direct
injection, an ability to operate over an Atkinson Cycle with increased expansion ratio, wide-
authority intake camshaft timing, and an optimized combustion process. This type of engine
operation is not limited to naturally aspirated engines and when applied to boosted engines is
referred to as "Miller Cycle," as described below.
2.3.4.1.8.1 Effectiveness Data Used and Basis for Assumptions
EPA initiated an internal study to investigate potential improvements in the incremental
effectiveness of Atkinson Cycle engines through the application of cooled EGR, an increase in
compression ratio, and 2/4 cylinder deactivation. Cooled EGR offered the potential for
additional knock mitigation, increased compression ratio, and reduced pumping losses. The use
of cylinder deactivation held potential for additional pumping loss reduction under light-load
conditions. Initially, EPA studied the potential for improvements using 1-D gas dynamics/O-D
combustion simulation software.FFF A 2.0L Mazda SKYACTIV-G GDI Atkinson Cycle engine
was thoroughly benchmarked by EPA with the engine dynamometer test facilities at the EPA-
NVFEL laboratory in Ann Arbor, MI. Performance data and physical dimensions for the engine
and its gas exchange and combustion processes were used by EPA to build and validate the
simulation. Details of the study, including methods used to build the engine model, model
validation, and initial engine modeling results are provided in Lee et al. 20 1 6.554 A comparison
of engine dynamometer test data to modeling results for a 1-point increase in geometric CR and
FFF Gamma Technologies "GT-POWER."
2-293
-------
Technology Cost, Effectiveness, and Lead Time Assessment
the use of cEGR with an Atkinson Cycle engine are shown in Figure 2.102. Single point values
for regions of operation important for the regulatory drive cycles are shown from approximately
2-bar BMEP to 7 or 8 bar BMEP and from 1500 rpm to 2500 rpm (i.e., comparable to areas of
high frequency of operation over the UDDS and HWFET as shown in Figure 2.80). Engine
simulation results showed the potential for an approximately 3 percent to 9 percent incremental
effectiveness in areas of operation of importance for the FTP and HWFET regulatory cycles
using a combination of cooled EGR and a 1-point increase in compression ratio (14:1), with the
largest improvements (6 to 9 percent incremental) occurring between 4-bar and 8-bar BMEP.
While the increased expansion from a 1-point increase in geometric compression ratio
incrementally improves cycle efficiency, most of the improvement in effectiveness was due to
reductions in pumping losses from cooled cEGR.
12
10
8
m
6
a.
hi
IE
m
4
2
0
1000
2000
3000
Engine Speed (rpm)
4000
5000
6000
12
10
8
m
6
a.
hi
IE
m
4
2
0
1000
2000
3000 4000
Engine Speed (rpm)
5000
6000
Figure 2.102 Comparison of a 2.0L Mazda SKYACTIV-G engine with a 13:1 geometric compression ratio to
engine simulation results of a comparable engine with a 1-point increase in geometric compression ratio
(14:1) and cooled, low-pressure EGR000
Simulation results also show potential for an approximately 3 percent to 12 percent
incremental effectiveness in areas of engine operation with significant importance for the UDDS
and HWFET drive cycles using a combination of cooled EGR, a 1-point increase in compression
ratio (14:1), and with fixed (2-cylinder) cylinder deactivation when operating below 5-bar BMEP
and for engine speeds of 1000 rpm to 3000 rpm, and depending on how much cylinder
deactivation can be active within this range of operation. Simulation results also show an
incremental effectiveness of approximately 3 percent to 7 percent (Figure 2.103) when
comparing the cooled EGR/higher geometric compression ratio results with and without cylinder
deactivation. This is consistent with other published results for both production and proof-of-
concept fixed (not dynamic) cylinder deactivation.559-560-565 This represents a maximum potential
for fixed cylinder deactivation within the speed and load range analyzed. Based on
benchmarking results of the GM MEcotec3" 4.3L V6 engine, we estimated that cylinder
deactivation would available approximately 60 percent of the time within this speed and load
range for the analysis within the draft TAR and the Proposed Determination. This is a
GGG The simulation results presented in Figure 2.102 and Figure 2.103 include kinetic knock modeling and
calibration of the simulation to knock induction comparable to the original engine configuration for both Tier 2
certification test fuel (E0, 96 RON) and LEV III certification test fuel (E10, 88 AKI, 91 RON). An adequate
representation of knock-limited torque within an engine simulation requires careful experimental validation of the
kinetic knock model used by the simulation.
2-294
-------
Technology Cost, Effectiveness, and Lead Time Assessment
conservative estimate that does not take into use of improved crankshaft dampening systems or
other NVH measures that would reasonably be expected to extend the amount of cylinder
deactivation operation possible within this region of engine speed and BMEP.566
Figure 2.103 Comparison of a 2.0L Mazda SKYACTIV-G engine with a 13:1 geometric compression ratio to
engine simulation results of a comparable engine with a 1-point increase in geometric compression ratio
(14:1), cooled, low-pressure EGR and cylinder deactivation with operation on 2 cylinders at below 5-bar
BMEP and 1000 - 3000 rpm.
The EPA internal study on Atkinson Cycle engines entered a second phase involving engine
dynamometer validation of the simulation results using a EU-market version of the Mazda
SKYACTIV-G engine with increased geometric compression ratio (14:1), a proof-of concept
low-pressure-loop cooled EGR system, and the use of a dual-coil offset (DCO) ignition system
to improve EGR tolerance of the engine (see Figure 2.104).567-568 Initial results have been
promising. The improved ignition characteristics of the DCO ignition system has allowed an
increase in the range of part-load engine operation at relatively high rates (approximately 20
percent) of cooled EGR beyond that of the relatively conservative, fixed EGR map used in the
simulation study. This allowed further reductions in part-load pumping losses which improve
fuel efficiency while maintaining a COV of IMEPHHH of less than 3-4 percent, which is
comparable to that of the original engine configuration.
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
HHH Coefficient of variation (COV) of indicated mean effective pressure based on high-speed in-cylinder pressure
measurements. This is a commonly used indicator of combustion instability and would typically be kept to values
that are under 3% to 5% depending on operating conditions and engine application. Lower COV corresponds to
smoother engine operation.
2-295
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.104 Mazda 2.0L SKYACTIV-G engine with 14:1 geometric compression ratio, cooled low-pressure
external EGR system, DCO ignition system, and developmental engine management system undergoing
engine dynamometer testing at the U.S. EPA-NVFEL facility in Ann Arbor, ML
The ignition system improvements also allowed further optimization internal EGR (iEGR)
and external cEGR rates and allowed higher EGR rates to be shifted to and broadened to cover
more operation over the UDDS and HWFET (see
Figure 2.105). The calibrated rates of cEGR arrived at during engine dynamometer testing
were remarkably similar to published data for a production application of cEGR to an Atkinson
Cycle engine.'""1
The CO2 reductions achieved during engine dynamometer testing occurred over a broader
area of operation than for the engine simulation conducted for the draft TAR. At engine speeds
below 2000 rpm, larger reductions in CO2 were achieved during engine testing between 4 and 8
bar BMEP although simulations results showed larger CO2 reductions below 2.5 bar BMEP. At
all other conditions above 2000 rpm and I bar BMEP., engine test results achieved comparable or
larger reductions in CO2 emissions than the engine simulation results from the draft TAR. See
Figure 2.106 for a graph of modeled and tested CO2 effectiveness. Note that the regions of CO2
effectiveness roughly correlate with the EGR rates shown in Figure 2.105.
2-296
-------
Technology Cost, Effectiveness, and Lead Time Assessment
iEGR (%)from Original 2015 EPA Engine Simulation
iEGR (%)from Engine Dynamometer Data
75k W
1000 2000 3000 4000
Engine Speed (rpm)
cEGR (%) from Original 2015 EPA Engine Simulation
60 kW
1000 2000 3000 4000
Engine Speed (rpm)
cEGR (%) from Engine Dynamometer Data
2000 3000 4000 5000
Engine Speed (rpm)
2000 3000 4000 5000
Engine Speed (rpm)
Figure 2.105 Modeled internal EGR and cEGR rates (in percent) from the draft TAR engine simulation (left
top and left bottom, respectively) compared to internal EGR and cEGR rates achieved during engine testing
(right top and right bottom, respectively).
Note: White areas of the contour plots reflect <1% (effectively zero) EGR.
Atkinson Cycle, Naturally Aspirated GDI, DOHC, DCP 14:1 CR
C02 Effectiveness from Original 2015 EPA Engine Simulation - iEGR & cEGR
' 05 i'VV
Atkinson Cycle, Naturally Aspirated GDI, DOHC, DCP 14:1 CR
C02 Effectiveness from Engine Dynamometer Data - iEGR & cEGR
105
2000 3000 4000
Engine Speed (rpm)
2000 3000 4000
Engine Speed (rpm)
Figure 2.106 Modeled CO2 effectiveness for internal and cEGR from the draft TAR engine simulation (left)
compared to CO2 effectiveness achieved during engine testing (right).
2-297
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The updated laboratory engine test data and simulations of ATK2 using cEGR described
above were very encouraging and suggest that the Draft TAR effectiveness projections are
conservative. Therefore, it was decided that the internal and cEGR rates and resulting fuel maps
and C02 effectiveness from the engine simulations used in the Draft TAR were still appropriate
to use for the Proposed Determination analysis. Consequently, the higher C02 effectiveness
achieved during additional laboratory engine testing was not reflected within LPM C02
effectiveness for the Proposed Determination. In summary, the C02 effectiveness used within
the Proposed Determination for the application of cEGR to non-HEV Atkinson Cycle engines
has been confirmed with laboratory testing and is expected to be conservative relative to the
effectiveness that was achieved during engine dynamometer testing.
Furthermore, in the absence of engine dynamometer validation of the kinetic knock model,
engine displacements were increased by 5 percent for all "advanced" ATK2 engine packages to
which a 1-point increase in geometric CR and cEGR are applied. This was done to reflect a
reduction in peak BMEP and a resultant necessity for increased engine displacement to maintain
vehicle acceleration performance. This adjustment resulted in a decrease in LPM CO2
effectiveness for the proposed determination relative to the Draft TAR of approximately 0.1 to
0.65%, with the range roughly coinciding with low and high power-to-weight-ratio vehicles,
respectively.
Cylinder deactivation (DEAC) was also simulated during engine dynamometer testing by
disabling valve events to two cylinders via cam-follower removal and allowing trapped air to act
as an "air-spring" within the two disabled cylinders. Figure 2.107 shows the CO2 effectiveness
when combining operation on 2-cylinders at below 3.75-bar BMEP111 and between 1000 and
3000 rpm with cEGR and with internal EGR optimized for two-cylinder operation. It should be
noted that the effectiveness due to simulated cylinder deactivation shown in Figure 2.107 should
be considered a "maximum" effectiveness within the speed and load range that cylinder
deactivation was simulated during dynamometer testing.
Figure 2.107 CO2 effectiveness achieved during engine testing with cEGR and simulated 2-cylinder fixed
cylinder deactivation from 1000 to 3000 RPM and at less than 3.75 BMEP.
Atkinson Cycle, Naturally Aspirated GDI, DOHC, DCP 14:1 CR
C02 Effectiveness from Engine Dynamometer Data -DEAC+cEGR
75kW
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
m BMEP is reported relative to the entire engine displacement with both active and inactive cylinders.
2-298
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The effectiveness achieved from simulated cylinder deactivation during testing of modified
Mazda SKYACTIV-G engine was also very similar to effectiveness results presented by Mazda
for their developmental cylinder deactivation system for the 2.0L SKYACTIV-G, although the
Mazda system appears to use a broader engine speed window than what was considered during
simulation for the Draft TAR or subsequent engine dynamometer validation.561
Cylinder deactivation along with both internal and cEGR rates and resulting fuel maps and
CO2 effectiveness from the engine simulations developed for the draft TAR were also used for
Proposed Determination and thus the higher CO2 effectiveness achieved during engine testing of
an Atkinson Cycle engine with simulated cylinder deactivation was not reflected within LPM
CO2 effectiveness for the Proposed Determination. The CO2 effectiveness used within the
Proposed Determination for the application of cEGR to non-HEV Atkinson Cycle engines is thus
expected to be somewhat conservative relative to the effectiveness that was achieved during
engine dynamometer testing or relative to other similar work demonstrated by Mazda.561
The Alliance of Automobile Manufacturers (AAM) and FCA commented that EPA's results
used optimistic ATK2 engine fuel consumption maps. However, they did not provide data or
other information to substantiate its claim that EPA's engine dynamometer fuel consumption
measurements using a MY2014 Mazda OEM production 2.0L SKYACTIV-G, upon which the
ATK2 packages from the TAR analysis are based, were in any way unrepresentative of this
engine's actual performance. AAM did provide a fuel consumption "difference map" (see chart
B-l from the AAM public comments which is reproduced in Figure 2.108) purporting to show
the difference between a map developed from EPA-published test data using Tier 2 certification
gasoline and data provided to AAM by USCAR using an unspecified 91 RON fuel. AAM
implied that there were areas of concern that call into question the ATK2 fuel maps as a baseline
for further theoretical additions of technology. This AAM map is referred to as the "difference
map" in the following response.
2-299
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.0L Skyactive: EPA (96+RON) - USCAR (91RON)
City / Highway
Critical area
EPA measured higher
fuel consumption
Aggressive thermal
protection assumptions
SOD 1000 1500 2000 2?00 3000 3f00 -SOW 4»00 MOO 5£00 M00
Engine Speed, rpm
EPA measured lower
Lis fuel consumption
Figure 2,108 "Difference map" comparison provided by AAM between EPA data generated using Tier 2
certification gasoline and "USCAR 91 RON data" for a Mazda SKYACTIV-G 2.0L engine. (AAM Fig B-1)JJJ
First, from a regulatory compliance standpoint, AAM's comparison between Tier 2 and other
fuels has no basis. This is because the stringency of the GHG (and fuel economy) standards is
based exclusively on use of Tier 2 fuel. Furthermore, EPA has investigated the difference in
C02 performance between Tier 2 and Tier 3 test fuels, and preliminary data indicate that
vehicles actually perform slightly better from a C02 standpoint (i.e. emit less C02) using Tier 3
fuel. This is because Tier 3 test fuel has less energy content but also a lower carbon content than
Tier 2 fuel. EPA has already indicated in the Tier 3 rule package that, as a convenience to avoid
testing using different fuels, EPA will make an adjustment to convert C02 results using Tier 3
test fuel to account for the different fuel properties. Please see Chapter 2.3.1.3 (Fuels) for
additional discussion.
In any case, the AAM commenter provided virtually no information regarding test and or
analytical methods, assumptions, fuel properties, environment test conditions, how the engine
was controlled or how control was modeled, among other pertinent factors. Thus, AAM did not
provide any fuel specifications other than RON, so it is unclear if the map purports to show a
difference due to RON or a difference due to a combination of factors that also impact fuel
consumption (e.g., differences in fuel ethanol content and/or net energy content or other fuel
properties). Use of any future certification fuels with differing properties would also necessarily
include a correction back to GHG performance on Tier 2 certification fuel, as EPA has already
indicated for Tier 3 test fuel, again as noted above.
Although the "difference map" provided by AAM is identified as showing fuel consumption
differences, no specific units were identified by AAM, so it is not clear if the map shows
111 Alliance of Automobile Manufacturers Comments on Draft Technical Assessment Report. EPA docket number
EPA-I-IQ-OAR-2015-0827-4089-A1.
2-300
-------
Technology Cost, Effectiveness, and Lead Time Assessment
absolute differences in fuel mass flow rate, absolute differences in fuel volumetric flow rate, or
percentage differences on either a mass or a volumetric basis. AAM identified USCAR as the
source of data used for the comparison, but it is unclear if the data compared to EPA's measured
fuel consumption was generated using modeling, if it was generated using data from engine
dynamometer testing, or if it was estimated by some other means.
Neither the underlying test conditions nor the experimental design were shared for any data
that may have been generated, so it is impossible to even assess the validity or veracity of the
data presented. The absence of underlying data or other supporting details in the comment
makes these issues a matter of conjecture. For example:
• If the data used to calculate the fuel consumption difference map was generated from
USCAR engine dynamometer testing, it is unclear why AAM/USCAR would only
test an engine using a 91 RON fuel and then compare the results to EPA data. Such a
comparison inherently introduces different uncertainty by comparing data from
different engines tested in different laboratories, potentially under different testing
and operating conditions.
• AAM did not share the test conditions under which the engine was tested, any
procedures used to ensure data quality, any measured analytical fuel properties other
than RON, the number of data points used to generate the fuel consumption
"difference map", or the interpolation method used to generate the "difference map".
• A more reasoned difference map comparison would be to conduct independent testing
with both a Tier 2 certification fuel and a 91 RON, or Tier 3 certification, fuel using
the same engine in order to generate a "difference map" with commonality of engine,
engine management system calibration, experimental equipment, and laboratory
equipment calibration.
• A more valid difference map comparison would also involve an engine from the same
vehicle application to demonstrate any fuel consumption differences without the
added uncertainty of lab-to-lab and design-of-experiment differences.
In summary, the commenter provided no information to compare vintage or application of the
actual engine or engines tested, and did not state whether or not testing was conducted. More
specifically, the comments did not state: test and or analytical methods, assumptions, fuel
properties, environment test conditions, how the engine was controlled or how control was
modeled, the number of data points gathered to generate the AAM "difference map" to assure
that identical testing and a sufficient fit of data was performed. For example, not enough
information was provided to know how accessory loads or engine cooling were handled in any
testing that may have been performed.
While AAM shared neither the underlying data, underlying assumptions, nor even the units
used within the "difference map" with EPA, we nevertheless independently generated a complete
set of "difference maps". As part of ongoing engine technology benchmarking activities, EPA
tested a MY2014 Mazda SKYACTIV-G 2.0L engine with a geometric compression ratio of 13:1
(i.e. ATK2) using fuels having different properties, including differences in RON. Our
comparison of the engine operation on a brake-thermal-efficiency basis, or after correction of
2-301
-------
Technology Cost, Effectiveness, and Lead Time Assessment
percentage mass differences in fuel consumption to an equivalent energy basis, revealed little or
no discernable difference between fuels over the areas of concern for regulatory testing beyond
the differences in energy content between the fuels. The results of EPA's engine map
comparisons is available in Appendix D.
In their comments, AAM and FCA expressed concerns about the practical limitations for
cEGR to limit engine knock. EPA conservatively considered practical limitations of cEGR to
limit engine knock and took these limitations into account when modeling the impacts of cEGR
on engine operation. In EPA's assessment of cEGR effectiveness, EPA not only took into
consideration the practical limitations for improving knock-limited spark advance (KLSA), but
also practical limitations in applying cEGR to reduce pumping losses at part-load conditions.
Part-load pumping loss reductions from cEGR are more important to the drive-cycle
effectiveness of cEGR than use of cEGR solely for knock abatement. Typically, improvements
to KLSA would not significantly impact performance on the FTP or HWFET drive cycles since
knock-limited operation is either not encountered or not often encountered over these cycles with
naturally aspirated engines, including those using Atkinson Cycle. Non-HEV Atkinson Cycle
engine applications reduce effective compression ratio under part-load conditions to reduce
pumping losses from throttling. Thus, limits on part-load cEGR application due to combustion
stability result in more important impacts on cEGR effectiveness under the conditions
encountered over the regulatory drive cycles used for GHG compliance (e.g., sub-6 bar BMEP
for engines like the Mazda SKYACTIV-G) than would be the case of using cEGR solely for
improving knock-limited spark advance.
The cEGR limits investigated by EPA included investigating adverse effects on combustion
phasing and the potential for deterioration of combustion stability at part load, which potentially
limit the availability of pumping loss reductions from EGR over the regulatory drive cycles. The
O-D/l-D models used for investigation of cEGR effectiveness could not adequately account for
changes to COV of IMEP (an important indicator of combustion stability), so limits on cEGR
based upon published literature were initially investigated by EPA during modeling.
During engine dynamometer testing to validate the modeling results, EPA made improvements to the
ignition system to allow the use of higher cEGR rates at some part-load conditions representing important
areas of operation for the regulatory drive cycles. This allowed engine operation at higher cEGR rates
than were considered during modeling while still achieving comparable or improved COV of IMEP when
compared to the OE engine configuration with lower geometric compression ratio and no external EGR.
Figure 2.105 compares modeled internal EGR and cEGR rates with those actually achieved
during engine testing with a developmental cEGR system, 14:1 geometric compression ratio, and
revised valve event timing. The results used in the Draft TAR and the Proposed Determination
analyses continue to reflect the use of a more conservative cEGR strategy, with somewhat
reduced cEGR rates relative to what has been demonstrated by EPA during engine dynamometer
developmental testing.
The application of cEGR technology is found in many light-duty vehicles in the current fleet.
As such, the feasibility of applying cEGR to mitigate knock and reduce part-load pumping losses
has already been established. Although cEGR development has been a significant topic of auto
manufacturer research and development in recent years for both naturally aspirated and
turbocharged applications, AAM shared no data with EPA showing achievable cEGR rates and
2-302
-------
Technology Cost, Effectiveness, and Lead Time Assessment
cEGR operational limitations from engine simulation, engine dynamometer developmental
testing, or from actual production applications of cEGR that have been introduced in the U.S.,
Europe and Asia.569'570'571'572'573
Further comments from AAM and Ford expressed concern that EPA did not take into account
the impact of 91 RON market and certification test fuels when developing fuel economy
effectiveness. While EPA's analysis of effectiveness of gasoline fueled engines did not include
analysis of effectiveness using Tier 3 certification gasoline (E10, 87 AKI), protection for
operation in-use on 87 AKI E10 gasoline was included in the analysis of engine technologies
considered both within the Draft TAR and this Proposed Determination.
As noted in the discussion on Atkinson cycle engines in Chapter 2.3.4.1.8, from the current
regulatory compliance standpoint, determining fuel economy effectiveness using any fuel other
than Tier 2 fuel has no basis. Consistent with Federal regulations under the Clean Air Act and
EPCA, when test fuel properties are updated EPA will determine appropriate test procedure
adjustments in order maintain the same level of stringency of the GHG standards should
manufacturers elect to test vehicles using Tier 3 certification fuel (as noted in that earlier
response, EPA is providing this accommodation to ease testing burden, although the GHG rules
specify Tier 2 fuel as the test fuel). A correction factor for application to future vehicles certified
to the GHG standards using Tier 3 gasoline that will allow correction of C02 emissions in a
manner that accounts for differences between Tier 2 and Tier 3 certification fuels is currently
under regulatory development with manufacturers, industry, and other stakeholder involvement.
Please refer to Chapter 2.3.1.3 (Fuels) for further discussion.
In other comments, AAM and FCA expressed concern that the benefits modeled by GT-
POWER for the Advanced Atkinson Tech Package have not been verified by manufacturers or
by the agencies. EPA notes that CAE models, such as GT-POWER, are routinely used by
manufacturers to aid in the development of engine and other technologies to comply with EPA
standards. Furthermore, estimated effectiveness from CAE modeling is conservative relative to
data generated via engine dynamometer validation (see 2.3.4.1.8.1). The AAM commenters are
correct in stating that models, including 0D/1D combustion and flow models like GT-POWER,
need careful validation relative to engine dynamometer performance in order to be used as
predictive tools, and EPA has conducted hardware validation as described below and in section
2.3.4.1.8.1.
AAM's comment referred to SAE Paper 2016-01-0565, which documents part of the
validation process, including validation that the model can predict operational characteristics of a
base engine design. AAM unfortunately significantly misquoted a sentence from this paper:
"[the] BSFC map [of the ATK2 engine] at 14:1 CR [with cooled EGR and cylinder deactivation]
could not be validated with engine dynamometer operation, even with use of 96 RON E0 fuel,
due to the onset of knock." The parentheticals added by AAM are both wrong and misleading.
First, the BSFC map at 14:1 geometric compression ratio does not represent either ATK2 or
testing of an engine with cooled EGR and/or cylinder deactivation.
ATK2 effectiveness was developed by EPA via benchmarking of a production, unmodified
MY2014 U.S.-market Mazda SKYACTIV-G 2.0L 4-cylinder Atkinson Cycle engine with a 13:1
geometric compression ratio. The engine with the 14:1 geometric compression ratio originally
discussed in the SAE paper was a European-market version of the engine not available in the
U.S. and with hardware and EMS calibration developed for operation on higher octane,
2-303
-------
Technology Cost, Effectiveness, and Lead Time Assessment
predominantly EO fuels available in Europe. An unmodified European version of this engine
without cEGR and without other significant hardware and calibration changes could not
reasonably be expected to have capability to operate on U.S. fuels, even premium-grade fuels,
without a risk of knock onset.
At the time that SAE Paper 2016-01-0565 was prepared by EPA staff, developmental
hardware that could potentially enable the use of the higher geometric compression ratio
hardware, such as cEGR, a developmental/open EMS allowing engine calibration, higher energy
ignition system, and possibly cooling system improvements was not yet available and thus the
entire point of using CAE tools was to allow EPA to investigate potential improvements as
future technologies were applied to the engine using reasonable engineering assumptions. This
is a long-recognized way of assessing and reasonably predicting technology effectiveness. See,
e.g. Amer. Petroleum Inst. v. EPA, 706 F. 3d 474,, 480 (D.C. Cir. 2013).
EPA has completed initial hardware validation of the GT-POWER modeling of non-HEV
Atkinson Cycle engine simulations conducted for the Draft TAR. While EPA continued its
hardware validation and incremental improvement of GT-POWER modeling of this specific
application of technologies, EPA engineering staff shared its initial results used in the Draft TAR
regarding cEGR and CD A GT-POWER model validation at the higher geometric compression
ratio with engineering staff from AAM member companies at an April 12, 2016 USCAR
meeting. At that meeting, EPA staff responded informally to questions and participated in a
discussion of both Atkinson Cycle engine technology and the use of external cEGR and CD A
with Atkinson Cycle engines. EPA staff also used design of experiments for both GT-POWER
modeling and for hardware validation of the technologies assessed using GT-POWER modeling.
During the course of the meeting, no indication was made that the use CAE tools such as GT-
POWER modeling were inappropriate or "not accurate enough for reference" in the MTE. Such
tools are regularly used by the automotive industry themselves to guide product development and
are used extensively by USCAR to guide research and development to improve internal
combustion engine and vehicle efficiency, hence the interest in inviting EPA staff make a
presentation at the meeting. EPA received no formal "meeting minutes", as referenced by AAM,
from either AAM or from USCAR. EPA's presentation materials from the USCAR meeting are
available in the Docket.574
The hardware development, engine dynamometer testing, model validation and updating of
the GT-POWER model do represent significant further study and development of these
technologies. EPA has completed much of this work, which as explained earlier, confirms that
our estimates for the Proposed Determination are appropriate. Initial test results of engine
dynamometer testing with cEGR, 14:1 geometric CR and CD A are summarized in Chapter
2.3.4.1.8 and are the topic of upcoming journal and technical paper submissions. These initial
hardware validation results indicate that the modeling approach used within GT-POWER was
conservative with respect to the determining the effectiveness of future technologies applied to
ATK2.
AAM also made other erroneous assertions related to SAE Paper 2016-01-0565 and the April
12, 2016 USCAR meeting. For example, AAM claimed that there was a "serious clerical error in
translating the GT-POWER full load torque data to ALPHA which was then carried into the
LPM's calibration" and that "in the SAE paper 2016-01-0565 the GT-POWER model correctly
limited the full load torque of the engine due to knock onset" (p. 49 AAM comments). This
2-304
-------
Technology Cost, Effectiveness, and Lead Time Assessment
statement is incorrect. EPA notes that work to develop a model of knock limited peak torque had
not been completed in time for the initial SAE 2016-01-0565 paper, and no such data or
modeling was referenced as part of that work. Furthermore, the torque curve claimed by AAM
to be from EPA's SAE paper does not match either torque limits or data plotting limits used
anywhere within SAE 2016-01-0565. Figure 2.109 shows the discrepancy alleged by AAM in
figure B-2 of their comments.
ATK2 Torque Curve Discrepancy
14
12
j§ 10
6
4
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
Figure 2.109 This figure was reproduced from "Figure B-2" of the AAM comments purporting to show a
discrepancy between the torque curves used in SAE 2016-01-0565 vs. those used within the ALPHA model.
Figure 2.110 overlays EPA data onto the original AAM figure showing the following
additional torque data:
• Maximum plotted torque (not peak torque) from GT Power model of 2014
SKYACTIV-G 2.0L BSFC from SAE 2016-01-0565 (solid, orange line). Note that
the limits were solely limits of the data points analyzed within GT-POWER, not
knock-limited torque
• Maximum plotted torque for modeling knock induction at 13:1, 14:1, and 14:1
w/cEGR from SAE 2016-01-0565 (dashed, light blue line). Note that the limits were
solely limits of the data points analyzed within GT-POWER, not knock-limited
torque
• Torque curve from initial engine dynamometer testing shown in SAE 2016-01-0565
(solid green line). This torque curve represented test data from engine dynamometer
benchmarking of a U.S.-market MY2012 Mazda SKYACTIV-G engine. The torque
limits served as the initial developmental limits for GT-POWER model development,
assessment of future technologies, and initial cEGR hardware development that
— Incorrect 14:1 CR Torque Curve used In
dTAR(ALPHA)
2-305
-------
Technology Cost, Effectiveness, and Lead Time Assessment
occurred after the SAE paper was published. EPA's torque mapping procedures were
updated, and this torque curve has been replaced with the one described below.
Torque curve from Chapter 5 2014 SKYACTIV-G 2.0L engine map (dashed black
line). This torque curve was developed during benchmarking of the MY2012 Mazda
SKYACTIV-G engine that occurred too late for inclusion in the initial SAE paper.
These torque limits serve as the current limits for GT-POWER model development,
assessment of future technologies, and cEGR hardware development and also serve as
the torque limits used within ALPHA modeling.
ATK2 Torque Curve Discrepancy
14
12
jS 10
CO
8 - =
¦Incorrect 14:1 CR Torqi
dTAR(ALPHA)
e Curve used ir
Note:AAM Mischaracterized
the ALPHA input data - ATK2
actually represents a 2014
Mazda SkyactivG engine with
13:1 geometric CR
i AAM-plotted ALPHA torque curve
300
1000
5000
5000
¦ Torque curve from Chapter 5 2014 SkyactivG 2.0L engine map (also used in ALPHA for ATK2)
] Torque curve from initial engine dynamometer testing SAE-2016-01-0565 (this was later updated to the black-dashed curve)
Maximum plotted torque for modeling knock inderaion at is:if 14:1, and 14:1 w/cEGR from SAE 2016-01-0565
Maximum plotted torque (not peak torque) from GT Power model of 2014 SkyactivG 2.0L BSFC from SAE 2016-01-0565
1 AAM-plotted torque curve allegedly from SAE 2016-01-0565 (but not matching any data from the paper)
Figure 2.110 This is a reproduction of AAM figure B-2 with EPA data for two engine dynamometer derived
torque curves (green and black dashed) as well the extent of modeled data points (orange, light-blue-dashed).
None of the data from SAE 2016-01-0565 matches the solid blue line from the AAM comments citing SAE
2016-01-0565.
AAM's original red line approximately matches what was used by EPA within the Draft TAR
and within ALPHA modeling (black dashed line) for both ATK2 (13:1 CR) and ATK2+cEGR
(14:1 CR) engines. However, MATK2" was mischaracterized in AAM's comment as representing
operation of an engine using a 14:1 geometric CR. It does not. ATK2 actually uses a 13:1
geometric CR and is the same as a U.S.-market MY2014 Mazda SKYACTIV-G 2.0L engine. It
is only the ATK2+cEGR that has a 14:1 geometric CR. It is not clear what AAM's solid blue
line is supposed to represent because it does not represent any data presented within SAE 2016-
01-0565.
The green line represents the torque limit from the first figure in the paper, converted to
BMEP to be consistent with AAM's figure. It also represents what EPA was using as maximum
2-306
-------
Technology Cost, Effectiveness, and Lead Time Assessment
torque/BMEP as of October 2015 and represents significantly higher torque than shown by the
AAM blue line. This was later updated to the torque curve represented by the black dashed line
for the TAR using data points from a later torque mapping exercise and using updated engine
dynamometer mapping procedures.
The blue dashed line in the figure represents torque limits of the data plotted from the GT-
POWER modeling of knock induction for the various engine configurations. EPA could have
modeled the design space within GT-POWER using higher torque limits but it was not necessary
in order to cover operation over the regulatory drive cycles (FTP and HWFET).
The orange line represents torque limits that EPA used for modeling BSFC in GT-POWER.
Again, EPA could have modeled the design space within GT-POWER using higher torque but it
was not necessary to sufficiently cover operation representative of the regulatory drive cycles.
In summary, AAM did not provide sufficient information for EPA to determine with any
certainty the source of their mistaken characterization of the data from EPA's SAE paper as
represented by their blue line.
EPA's SAE paper publication predated the Draft TAR release by approximately four months,
but the underlying data in the SAE paper dates to October 2015 (when the paper was submitted
for peer review), or approximately nine months prior to the TAR release. During those
intervening months, EPA's torque mapping procedures were updated and provided slightly
higher maximum torque in limited areas of operation. The more recently developed torque map
is also more consistent with data presented publicly by the engine's manufacturer, Mazda,
regarding the performance of the 2.0L SKYACTIV-G engine.
We also received comments that the relatively low cost of ATK2 has the impact of lowering
the OMEGA-estimated cost per vehicle. In response, it is important to note that EPA's
projection of ATK2 penetration in the light-duty fleet is only one of several cost-effective engine
technology alternatives available to manufacturers to meet the 2025 MY GHG standards. In
both the Draft TAR and this Proposed Determination, we have run sensitivities showing the
impacts on costs per vehicle under a scenario where very little ATK2 technology is used for
compliance. In these sensitivities, we have capped the ATK2 technology at a 10 percent level
(note that Mazda uses this technology extensively today, as well as other manufacturers, and
roughly 7 percent of today's fleet already uses the technology). The results show minor increases
in costs per vehicle, but clearly show that pathways to compliance exist at reasonable costs and
without extensive utilization of strong hybrid and electrified vehicles (see Draft TAR Table
12.48 and Section C.1.2 of the Proposed Determination Appendix).
2.3.4.1.8.2 Cost Data Used and Basis for Assumptions
Costs for this technology (future non-HEV Atkinson cycle, referred to as Atkinson-level 2 by
EPA) were new to the Draft TAR as they were not part of the 2012 FRM analysis. As in the
Draft TAR, we have based our Atkinson-2 technology costs on the 2015 NAS report. Table S.2
of that report shows the cost estimates presented below. Note that the NAS costs include the
costs of gasoline direct injection (shown as "DI" in the NAS report row header). EPA has
removed those costs (using the NAS reported values) since EPA accounts for those costs
separately rather than including them in the Atkinson-2 costs. Note also that EPA always
2-307
-------
Technology Cost, Effectiveness, and Lead Time Assessment
includes costs for direct injection, along with variable valve timing and other costs, when
building an Atkinson-2 package.
Table 2.67 Direct Manufacturing Costs (DMC) for Atkinson-2 Technology (2010$)
Tech
Midsize
Large
Large Light
Relative to
Car
Car
Truck
14 DOHC
V6
DOHC
V8 0HV
Stoichiometric Gasoline Direct Injection (NAS 2015)
$164
$246
$296
Previous
tech
Compression Ratio Increase (CR~13.1, exh. Scavenging, Dl (e.g.
$250
$375
$500
Baseline
SKYACTIV-G)) (NAS 2015)
EPA estimate (Row 2 minus Row 1)
ID
00
¦uy
$129
$204
Stoich GDI
Consistent with the NAS report, we have considered the NAS costs to be 2025 costs in terms
of 2010$. Adjusting to 2015$, applying a learning curve (22) that bases that cost in MY2025,
and applying medium 2 level complexity in calculating indirect costs results in the costs
presented below for each engine type in this Proposed Determination analysis.
Table 2.68 Costs for Atkinson-2 Technology, Exclusive of Enablers such as Direct Inject and Valve Timing
Technologies (dollar values in 2015$)
Engine
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
13
DMC
$93
22
$110
$108
$106
$103
$101
$99
$97
$95
$93
14
DMC
$93
22
$110
$108
$106
$103
$101
$99
$97
$95
$93
V6
DMC
$140
22
$165
$161
$158
$155
$152
$149
$146
$143
$140
V8
DMC
$222
22
$261
$255
$250
$245
$240
$236
$231
$226
$222
13
IC
Med2
2024
$37
$37
$37
$37
$37
$36
$36
$36
$27
14
IC
Med2
2024
$37
$37
$37
$37
$37
$36
$36
$36
$27
V6
IC
Med2
2024
$55
$55
$55
$55
$55
$55
$55
$54
$41
V8
IC
Med2
2024
$88
$87
$87
$87
$87
$86
$86
$86
$64
13
TC
$147
$144
$142
$140
$138
$136
$134
$132
$121
14
TC
$147
$144
$142
$140
$138
$136
$134
$132
$121
V6
TC
$220
$217
$213
$210
$207
$204
$201
$197
$181
V8
TC
$348
$343
$337
$332
$327
$322
$317
$312
$286
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
2.3.4.1.8.3 Basis for Feasibility Assumptions
The Alliance of Automobile Manufacturers (AAM) and some of its members commented on
the application of Atkinson-cycle engine technologies in the future fleet. The comments stated
that EPA had been "overly optimistic" in its assessment of the technology, and that: "The
advanced Atkinson technology package with CEGR and cylinder deactivation should not be
utilized in the MTE analysis until the technology can be demonstrated to operate across all
modeled operating points." In addition, AAM noted that the penetration rate projected by EPA
for Atkinson engine technologies in 2025 MY are not feasible and may not reflect individual
vehicle manufacturer's selected "technology pathway" for future compliance, suggesting that
there would be insufficient lead time to implement this technology. The commenter also stated
that EPA's analysis had not adequately accounted for limitations reflecting effects such as knock,
cooled EGR heat rejection, and effective compression ratio.
2-308
-------
Technology Cost, Effectiveness, and Lead Time Assessment
EPA does not agree with these comments. The engine technology itself is already
demonstrated in the fleet in non-hybrid applications. EPA considered two primary types of
Atkinson-cycle engine technologies in the Draft TAR and we have carried these technologies
into this Proposed Determination. The first Atkinson technology is referred to as "ATK1." This
technology designation reflects the application of Atkinson cycle operation on engines that are
primarily equipped in hybrid electric vehicles such as the Toyota Prius and the Ford Fusion. The
second Atkinson technology is referred to as "ATK2." This technology designation reflects the
application of Atkinson cycle engine operation in a conventional powertrain architecture, where
the sole source of power to the vehicle is provided by an internal combustion engine, such as in
the Mazda SKYACTIV-G architecture and the Toyota Takoma pickup truck.
In addition to the commercially available ATK2 architecture, EPA has also researched and
developed further enhancements that improve the effectiveness ATK2 technology. These
enhancements to ATK2 include the application of Cooled Exhaust Gas Recirculation (cEGR),
Higher Compression Ratio, and cylinder deactivation (DEAC). The ATK2 technology was
previously available with cEGR and a higher compression ratio in Japan and Europe and the
application of DEAC on future applications of the SKYACTIV-G engine has been publicly
announced by Mazda.561'562'563'564 There are also production applications of cEGR and/or
DEAC in current production Miller Cycle engines (e.g., 2016 Mazda SKYACTIV-G Turbo, VW
EA211 TSI evo) which are essentially boosted versions of Atkinson Cycle. EPA has also
validated modeling results of these advances using engine dynamometer testing (see 2.3.4.1.8.1)
EPA continues to believe that ATK2 engine technologies offer an additional cost effective
alternative in a broad assortment of advanced gasoline engine technologies expected to be
applied by vehicle manufacturers to meet future GHG standards. This group of technologies
builds upon some of the foundational technology that already has wide application across the
entire light-duty fleet including gasoline direct-injection (GDI), increased valve phasing
authority, higher geometric compression ratios, and in some cases cooled exhaust gas
recirculation (cEGR). These foundational technologies allow vehicle manufacturers to operate
engines in some vehicles in both conventional and Atkinson cycle modes as demonstrated by the
Chrysler Pacifica plug-in hybrid in which the 3.6L Pentastar engine is operated in Atkinson
mode, and the Toyota Tacoma pick-up truck. It is also highly likely that the recently introduced
updated Pentastar engine for conventional vehicles also takes advantage of its increased VVT
valve phasing authority (70 degrees versus the previous 50 degrees) to expand operation in
Atkinson Cycle modes. These foundational technologies allow vehicle manufacturers the ability
to operate turbo-charged engines in Miller-cycle modes, which is Atkinson-cycle applied to
boosted engines.
In response to comments received regarding the lead time required by manufacturers to adopt
ATK2 technology, it is important to again note that EPA's projection of ATK2 penetration in the
light-duty fleet is only one of several cost-effective engine technology alternatives available to
manufacturers to meet the 2025 MY GHG standards. As a sensitivity analysis, this TSD presents
results of an OMEGA run with penetration rates of ATK2 artificially constrained (see Section
C.1.2 of the Proposed Determination Appendix). This sensitivity run still shows a cost effective
pathway not requiring extensive utilization of strong hybrid and electrified vehicles remain
available.
2-309
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Ford and FCA both commented that some manufacturers may have already decided to go
down a certain "technology pathway" that may be different from EPA's projections, and for these
manufacturers the alternative technology pathway, whatever that may be, may be more cost
effective than the compliance path resulting from EPA's analysis. However, for all
manufacturers, EPA believes that there is sufficient lead-time to adopt the ATK2 technology.
Many of the building blocks required to operate an engine in an Atkinson-mode, similar to the
Mazda SKYACTIV-G engine are already available in the 2016 MY fleet. These include
gasoline direct injection and a high level of control authority over the valve train.
The Mazda SKYACTIV-G engine itself was not a "clean sheet" engine design, but rather was
a further development of the Mazda MZR engine family, which was introduced in 2001 as a PFI
engine design with identical bore spacing and nearly identical block water jacket design. The
Mazda MZR engine family was also shared with Ford Motor Company who later developed the
engine into the Ford EcoBoost 2.0L engine. Finally, while the ATK2 technology was introduced
into the EPA analysis in CY 2016, the technology has been in production since 2011 MY and has
undergone several revisions since its initial launch. Currently Mazda, Toyota, FCA, PSA,
Hyundai, and VW all have an implementation of Atkinson or Miller cycle engine operation in
production in non-HEV applications, and in some cases across multiple engine families and
vehicle architectures.
FCA also commented on ability to package the "4-2-1" exhaust manifold design, which
provides exhaust gas scavenging in the current Mazda implementation of ATK2. FCA stated a
"revamp" of a vehicle's architecture would be required to package a new exhaust manifold.
While the 4-2-1 exhaust manifold is important in the Mazda current implementation of ATK2,
previous implementations have used more conventional exhaust manifolds with a small (0.5
point) reduction in geometric compression ratio. More recently, Mazda has used CAE design
tools to implement the 4-2-1 exhaust manifold into extremely challenging transverse-engine
vehicle packages, including the Mazda2 subcompact, Mazda3 compact and Mazda CX3 small
CUV.
EPA has also carefully considered whether it would be necessary to add ATK only as part of a
major vehicle redesign. EPA does not believe that this is necessary. This is because the
necessary foundational technologies for the ATK technology (specifically, gasoline direct-
injection (GDI), increased valve phasing authority, higher compression ratios, and in some cases
cooled exhaust gas recirculation (cEGR)) already are in wide application across the entire light-
duty fleet.
Therefore, because ATK2 technology could build on many existing engine architectures, EPA
does not believe that the implementation of the technology must be tied to major vehicle
redesigns. As an example, in the case of a naturally aspirated DOHC engine with GDI and DCP,
which is estimated to be approximately 45 percent of the vehicle fleet for MY20 1 5,575 only the
following changes would necessary to fully implement Atkinson Cycle:
• High-authority (> 65 °) electric cam phasing. Implications: Incremental cost increase,
packaging improvement relative to hydraulic cam phasing (i.e., smaller), elimination
of hydraulic circuit for intake cam
• Increased intake charge motion (intake tumble). Implications: Cylinder head casting
revision with revised intake port geometry
2-310
-------
Technology Cost, Effectiveness, and Lead Time Assessment
• Increased geometric compression ratio. Implications: Revised piston with reduced
clearance volume, revised direct injector spray targeting to match piston design
• Improved exhaust scavenging. Implications: Revised exhaust manifold geometry,
may require some revision of belt-drive accessories and cooling fan/radiator location
in some transverse applications
In the case of any ATK2 applications that would also use cEGR, the cooling system capacity
would need to be sufficient to maintain the EGR cooler temperature to just above the intake
dewpoint temperature. Using the Mazda SKYACTIV-G engine as an example, the highest
external cEGR rates (-22%) would typically occur at partial loads (approximate 6 bar BMEP)
and relatively low engine speed (approximately 2000 rpm) (see Chapter 2.3.4.1.8, Figure 2.105),
with lower external cEGR at both higher and lower engine speeds.
2.3.4.1.9 GDI Turbocharging, Downsizing
2.3.4.1.9.1 Effectiveness Data Used and Basis for Assumptions
The TDS24 configuration used by EPA within the Draft TAR analysis was originally
developed as part of engine and vehicle simulation work conducted by Ricardo, Inc. and SRA
Corporation under contract with EPA, hereto referred in the Proposed Determination as the
"Ricardo Study." In recent years, Ricardo has developed a number of turbocharged and
downsized engine concepts with a number of characteristics in common. 576>577>578>579
• Gasoline direct injection (GDI)
• Dual camshaft phasing and, in some cases, discrete variable valve lift
• Relatively high boost and subsequently high levels of BMEP (over 30-bar in some
cases)
• Cooled, external EGR
• Advanced turbocharger boosting systems
Fuel mapping for different engine technologies was developed by Ricardo within the Study
using a combination of dynamometer test results, ID gas dynamics/OD combustion modeling,
application of correction factors for displacement scaling, and use of engineering judgment. The
development of fuel maps for turbocharged GDI engines within the Ricardo Study began with
BSFC data obtained from Ricardo's EBDI engine development program.576 Specifications for
this engine are shown in Table 2.69 and a contour plot of BSFC versus engine speed and BMEP
is also shown in Figure 2.111.
Table 2.69 Specification of Ricardo 3.2L V6 Turbocharged, GDI "EBDI" Proof-of-concept Engine.
Base Engine
Prototype V6 with IEM
Swept Volume
3190cc
Max Power @ 5,000 rpm
450 hp on E85, 400 hp on 98 RON gasoline
Max Torque @ 3,000 rpm
900 Nm on E85, 775 Nm on 98 RON gasoline
Target Max BMEP
35 bar on E85, 30 bar on Indolene (98 RON)
2-311
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Compression Ratio
10.0:1
Maximum Cylinder
180 bar
Cam Phaser Authority
50° CA
Intake Boosting System
Twin, sequential turbochargers with charge air cooling
after each boosting stage
Transient Torque Response Time
<1.5s to 90% SS torque at 1,500 rpm
<1.0s to 90% SS torque at 2,000 rpm
450
400
350
300
260
250
245
240
230
220
28
26
24
22
_20
«" 18
¦Q
s:16
m 14
m 12
10
8
6
4
2
0
1000
2000
3000 4000
Engine Speed (RPM)
5000
Figure 2.111 Contour plot of BSFC in g/kW-hr versus engine speed and BMEP for the Ricardo "EBDI"
engine equipped with sequential turbocharging, DCP, DWL, cEGR, IEM, and with a 10:1 compression ratio
using 98 RON Indolene.
In its public comments on the Draft TAR, AAM requested that "EPA outline its rationale for
using an experimental single cylinder engine map as the basis of their analysis of turbocharged
downsizing technology rather than using actual production engines that were benchmarked by
EPA (Ford 1.6 L EcoBoost and Ford 2.7 L EcoBoost)." The short answer is that technology has
advanced past these two Ford engines, making these engines inappropriate for evaluating
potential technologies for meeting the 2025 standards.
The engine EPA analyzed was a multi-cylinder engine at an advanced stage of development,
as described in the papers cited within the Draft TAR and as described within Draft TAR Table
5.63, which is reproduced here in its entirety as Table 2.69. A number of technologies were used
in Ricardo's development of this engine that go significantly beyond the technology of the Ford
1.6L EcoBoost (introduced in 2010) or the Ford 2.7L EcoBoost (introduced in 2015). The
technologies used by Ricardo during the EBDI development program better reflect the state of
technology that EPA expects to see in 2025, which is 10-15 years after the initial introduction of
the engines referenced by AAM. Technologies used on the EBDI engine that are not present on
the Ford EcoBoost engines referenced by AAM include:
• Variable valve lift
2-312
-------
Technology Cost, Effectiveness, and Lead Time Assessment
• External cooled EGR with both high and low-pressure loops
• Sequential turbocharging
• 50° (crank angle) of cam phaser authority
• Piezo injectors capable of multiple injections per cycle
• Higher peak cylinder pressure capability
It should also be noted that the 1.6L EcoBoost does not use an integrated exhaust manifold
and the 2.7L EcoBoost does not use centrally mounted injection and, based on certification and
confirmatory data, does not yet comply with Tier 3 PM emissions standards. Furthermore, the
two referenced EcoBoost engines also do not reflect state-of-the-art with respect to current
turbocharged/downsized engines - the VW EA211 TSIEVO, VW EA888 3B, Honda L15B7,
Honda K20C, and Toyota 8AR-FTS engines all have both higher peak BTE and significantly
broader regions of operation above 35 percent BTE than the engines referenced by AAM. Even
in those cases, only the VW EA211 has an advanced boosting system (VNT) and cEGR, but
lacks the range of cam phaser authority, VVL, peak cylinder pressure capability and more
advanced injection system of Ricardo EGRB. Comparisons to three of these current production
engines were presented in Chapter 5 of the TAR and are reproduced here in Figure 2.113, Figure
2.114, and Figure 2.115 and will be discussed later. It should also be noted that the multi-
cylinder engine developed by Ricardo was used as part of a proof-of-concept Class 3b light-
heavy-duty truck demonstration, and thus the project also included further in-chassis
development.580
Although not captured within the EGRB map (see Figure 2.111), Cruff et al. show
performance data up to 30-bar BMEP with this engine configuration. With respect to the design
of the engine block, cylinder heads, cylinder head attachment system, main bearing assembly,
rod bearing assembly and pistons, the engine was originally designed for considerably higher
cylinder pressures and other stresses than would be required than the 27-bar BMEP used by EPA
within the FRM.
Technical direction from EPA included a peak BMEP limit of 27-bar, which obviated the
necessity for some of the reciprocating assembly, engine block, and cylinder head measures
taken with the EBDI engine. Taking into account the capabilities of the combustion system,
valvetrain configuration, EGR system, and reduced BMEP levels, Ricardo recommended a small
increase in compression ratio (from 10:1 to 10.5:1) while maintaining protection for in-use fuel
octanes of approximately 91 RON (e.g. 87 AKIE10). All fuel consumption results developed in
the Ricardo Study assumed use of U.S. Certification Gasoline (95 RON, E0). A fuel
consumption improvement of 3.5 percent was also applied to account for continued application
of friction reduction from a combination of technology advances, including piston ring-pack
improvements, bore finish improvements, low-friction coatings, improved valvetrain
components, bearings improvements, and lower-viscosity crankcase lubricants. The FMEP and
fuel consumption improvements were relative to a MY2008 level of technology. BMEP levels
were held approximately constant for particular classes of engines within EPA's FRM analyses
and analyses for the Draft TAR. Boosting requirements over the reduced operational range for
TDS24 (up to 24-bar BMEP) were assumed to be achievable using a VNT within EPA's
analyses for the Draft TAR and Proposed Determination. Sequential turbocharging was
maintained for TDS27 within EPA analyses for the FRM, but consistent with both the Draft TAR
2-313
-------
Technology Cost, Effectiveness, and Lead Time Assessment
and public comments thereto, TDS27 was not included within the analyses for the Proposed
Determination.
Ford commented that with the removal of TDS27 from the analysis, EPA effectiveness values
are now closer to industry estimates although still optimistic. Ford believes that it is due to the
use of high octane fuel, optimistic friction reductions and failure to account for the effect of
higher boost pressures on crevice losses, friction and compression ratio. We disagree with these
conclusions. EPA based the effectiveness of these technologies on typical real world operation
where use of high octane fuel is largely unnecessary. The use of high octane fuel may be
recommended by the manufacturer in some applications and operational conditions as already
specified in current production Ford products with similar turbocharged downsized engines as
indicated above in the discussion of fuels in Chapter 2.3.1.3 (Fuels).
As discussed above, we believe that EPA's friction reduction assumptions are possible with a
combination of friction reduction technology advances not currently used in most engine
designs. As discussed, this includes coatings and use of other materials and technologies
throughout the engines moving components. The EGRB engine upon which TDS24 is based
was originally designed for higher peak cylinder pressures and for a maximum of 30-bar BMEP,
thus the main and rod bearings were designed for significantly higher loads than would be
encountered at peak BMEP of 24-bar. The friction reduction applied as part of the Ricardo
analysis is thus applied to an engine already having higher somewhat FMEP due to the increased
size of the main and rod bearings necessary to support operation at 30-bar peak BMEP.
While not directly discussed in our assessment, the impacts of designing engines for higher
boost pressures on crevice losses, friction and compression ratio is indirectly incorporated into
the final effectiveness estimates as reflected in the engine maps used for estimating effectiveness
for the TDS packages. Crevice volumes impacts are generally fundamentally controlled by the
manufacturer's design of cylinders and particularly the piston and piston rings. While it is
possible that higher boost pressures can have a negative effect on efficiency due to crevice
volumes, there are many design solutions a manufacturer can implement to the piston to mitigate
any crevice volume penalty. Some of these solutions have been used by manufacturers for many
years to reduce crevice volume impacts to other engine emissions, particularly hydrocarbon
emissions. Similarly, higher boost can impact friction and have compression ratio implications
however manufacturers have the opportunity during the design and development of an engine to
determine the appropriate technology solutions to these challenges.
FCA commented that the benefit of cEGR is overestimated due to higher accessory loads and
heat rejection. FCA did not provide sufficient information or substantiating data regarding their
concern. We believe that properly designed cEGR systems are in production today on several
engines from different manufacturers and with appropriate heat exchanger and cooling system
design, heat rejection is not an issue.
Figure 2.112 contains a graphical example of how BSFC maps were developed by Ricardo
and EPA for varying displacements of TDS24.
2-314
-------
Technology Cost, Effectiveness, and Lead Time Assessment
210 kW
Ermine Speed (RPM)
kW
Ricardo EBO 3 2L V6, DCP. DWL, CEGR. EM 10 1 CR
28f
2000 3000 4000 5000
Engine Speed (RPM)
* Engine ctungw
- FnctiDR
~ Tteimai esses
Ca*£KdlK»fl
- BMEP reduced to 24 bar
< Eooif>e changes
-ry KtilKXiri
tXspl per cyl reduced
ERA. TDS24 1.16LI3
!3fckW
20 kW
'i.-
i
r! Kn
- 15 KW
I ko
15 hW
r.skw
1000
2000 3000 4000 5000
Engine Speed (RPM)
EPATDS24 1.51LI4
m . . ¦
N-__
—_ "N.
/ J--..
^ s ~~~
r_-__
-¦160KW
- 140 k*iV
-120KW
-1TOKW
-HOkW
- tiU kW
•10 KW
20 kW
10 kW
6000
1000 2000 30-00 40-00 5000
Engine Speed (RPM)
6000
• BMEP maintained
• Engine changes
- # of cyl. reduced
- Displ- per cyl. reduced
Figure 2.112 Schematic Representation of the Development of BSFC Mapping for TDS24
The brake thermal efficiency (BTE) of the modeled and scaled TDS24 engine maps are
compared to contemporary, current production turbocharged engines in Figure 2.113 through
Figure 2 1 15 58U82>583.584.585
2-315
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Figure 2.113 Comparison between a 1.15L 13 version of TDS24 (left)KKK and the Honda L15B7 1.5L
turbocharged, GDI engine used in the 2017 Civic (right)111-.
Dark green shading denotes areas of BTE>35%. The Honda specifies use of gasoline with an octane of 87
AKI for the 2017 Civic with the L15B 7 engine. Data shown is for operation using >95 RON gasoline in both cases.
1000 2000 3000 4000 5000 6000
Engine Speed (rpm)
Speed (RPM)
75kW
34%*
96kW
Speed (RPM)
3000 4000
Speed (RPM)
Figure 2.114 Comparison between a 1.15L 13 version of TDS24 (left)1*®®1 and the 2017 Golf 1.5L EA211 TSI
EVO Engine*™.
Light-green shading denotes areas of BTE>34%. Dark green shading denotes areas of BTE>35%. The area
of BTE>35% for the VW EA211 is not discernable due to the coarseness of the data provided by the originally
published source.
KKK Adapted from Ricardo Study modeling results.
LLL Adapted from Wada et al. 2016 and Nakano et al 2016.
Adapted from Ricardo Study modeling results.
** Adapted from Eichler et al. 2016.
2-316
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Speed (RPM)
Figure 2.115 Comparison between a 1.51L 13 version of TDS24 (left)MMM and the 2017 Audi A3 2.0L 888-3B
Engine (right)000.
Dark green shading denotes areas of BTE>35%.
The Honda 1.5L L15B7 turbocharged GDI engine (Figure 2.113) achieves higher peak break
thermal efficiency than TDS24, and has a larger area of operation above 35 percent BTE.
TDS24 had improved efficiency at low-speed, light load conditions, possibly from pumping loss
improvements due to the use of discrete variable valve lift and cooled external EGR, which the
Honda L15B7 lacks.
The 2017 VW EA211 TSIEVO engine (Figure 2.114) appears to have a broader area of
operation above 34 percent BTE than TDS24 and the BTE reported at 2-bar, 2000 rpm of 30
percent is higher than the corresponding operational point with TDS24. The coarseness of
published BTE map for the VW EA211 precludes further comparison. The larger 2.0L VW
EA888-3B engine was compared with a 1.51L variant of TDS24.
The VW EA888-3B engine (Figure 2.115) had a significantly larger area of operation above
35 percent BTE. Similar to the Honda comparison, TDS24 had improved efficiency at low-
speed, light load conditions; possibly due to pumping loss reduction due to the greater extent of
boosting and displacement downsizing and the use of discrete variable valve lift.
On the whole, contemporary turbocharged engines can achieve higher peak BTE and high
BTE over a broader range of engine operating conditions than TDS24 modeling results. TDS24
shows improved BTE at lower speeds and lighter loads due to the use of technologies that are
either just now entering production (cEGR) or that have been in production for some vehicle
applications for over a two decades (VVL). Further development of contemporary turbocharged
engines from 2017 to 2025, including use of more advanced boosting systems (e.g., VNT or
series sequential turbochargers), engine downsizing to 22-bar BMEP or greater, use of external
cooled EGR, combustion system improvements and use of variable valve lift systems would
further improve low-speed, light load pumping losses. These improvements would allow current
turbocharged/downsized engines to meet or exceed the BTE modeled for TDS24 through
100^
Q.1C-
LD
m s-
500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500
Engine Speed (rpm)
uuu Adapted from Wurms et al. 2015.
2-317
-------
Technology Cost, Effectiveness, and Lead Time Assessment
incremental developmental improvements (e.g., VVL, cEGR) with sufficient lead time to meet
the 2025 light-duty GHG standards.
In comments regarding octane impacts on vehicles with turbocharged, downsized engines,
AAM cited data from an SAE technical paper (SAE 2014-01-1228) showing impacts on CO2
emissions for three different octane levels (91 RON, 96 RON, 101 RON) levels. The overall
implications were that AAM believed CO2 emissions from operation on lower octane fuels such
as Tier 3 gasoline or similar in-use gasolines would result in higher CO2 emissions than using
Tier 2 gasoline with approximately 96 RON as used during the development of EPA's
turbocharged/downsized technology effectiveness.
In reviewing the paper cited by AAM, it became clear that the properties of the fuels tested
bore little resemblance to the properties of either Tier 2 or Tier 3 gasoline properties, or the
average properties of current in-use regular-grade gasoline (upon which Tier 3 gasoline is based).
The study cited by AAM was actually designed to investigate the use of mid-level ethanol blends
at different octane levels, not to investigate CO2 emissions from fuels used for emissions
compliance testing or for current in-use grades of gasoline. For example, the 96 RON fuel tested
was not E0 (e.g., Tier 2 gasoline) - it was blended to a 20 percent ethanol content (E20). The
101 RON test fuel was blended to a 30 percent ethanol content (E30). The 91 RON fuel that
was tested may have been either E10 or E20 - AAM does not make this clear in their discussion
of the data from SAE 2014-01-1228, nor does it characterize the higher octane fuels as being
mid-level ethanol blends. Mid-level ethanol blends like E20 and E30 are not approved for use in
light-duty vehicle applications in the U.S. with the sole exception of flex-fuel-vehicles.
While the observed trends may be valid for the range of fuel properties investigated within the
cited study, AAM shared no data using fuels having properties similar to those used either for
current C02 compliance (e.g., Tier 2 gasoline), future Tier 3 criteria pollutant compliance (e.g.,
Tier 3 gasoline), or having properties comparable to average values for U.S. in-use "Regular"
pump-grade gasoline (87 AKIE10 or approximately Tier 3 gasoline) or other commonly
available grades of gasoline (e.g., 93 AKI E10). Ethanol content, distillation properties, carbon
content and aromatic content all have potential impacts on CO2. Octane can also impact CO2
emissions depending on the drive cycle used and vehicle road load for a particular application.
AAM's discussed relationships between the CO2 data for the mid-level ethanol blended
gasolines relative to a parameter described as "displacement over mass ratio" or D/M. AAM
indicated on one chart that this represented "Liters per tonne". EPA assumed this to be liters of
cylinder displacement per U.S. ton (2000 pounds) relative to dynamometer test inertia, but AAM
did not indicate if vehicle mass within the ratio represented curb weight, a loaded vehicle weight,
a test weight, or a dynamometer inertia category, or if "tonne" refers to "metric tonne" (1000 kg)
or "U.S. Ton" (2000 lbm).
AAM stated that "Using 91 RON fuel (e.g. Tier 3 fuel) there is no further C02 benefit below
a displacement-over-mass ratio (D/M) of about 0.9. However, as shown by the 96 RON and 101
RON data in the figure below, the Agency assumptions based on higher octane fuel would
indicate that additional downsizing beyond 0.9 D/M still yields reductions in C02." As part of
their compliance with GHG regulations, manufacturers already downsize engines significantly
below D/M of 0.9 L/ton for a number of light-duty vehicle and light-duty truck applications. A
partial summary of MY2015 vehicles using turbocharged/downsized engines and having D/M of
less than 0.9 L/ton is shown in Table 2.70. Vehicles at D/M below 0.9 L/ton were predominantly
2-318
-------
Technology Cost, Effectiveness, and Lead Time Assessment
passenger cars. The light trucks with D/M at or below 0.9 consisted of cross-over sport-utility
vehicles.
Table 2.70 Partial summary of MY2015 vehicles with D/M at or below 0.9 L/ton.
Rows in BOLD, yellow denote vehicles using engines above 20-bar BMEP.
D/M
Manufacturer
Short Description
Displacement (L)
BMEP (bar)
Fuel Requirements (R-
Regular, P-premium)
0.62
Ford
Focus SFE FWD
1.0
25
R
0.62
Ford
FOCUS FWD
1.0
25
R
0.70
Ford
Fiesta SFE FWD
1.0
25
R
0.80
FCA
500L
1.4
22
R
0.80
GM
ENCORE AWD
1.4
18
R
0.80
GM
TRAX AWD
1.4
18
R
0.80
FCA
Renegade 4x4
1.4
22
R
0.80
Ford
FUSION FWD
1.5
21
R
0.81
GM
ENCORE
1.4
18
R
0.81
GM
TRAX
1.4
18
R
0.83
Ford
TRANSIT CONNECT WAGON FWD
1.6
20
R
0.83
FCA
Dart
1.4
22
R
0.83
FCA
Dart Aero
1.4
22
R
0.83
FCA
500L
1.4
22
R
0.83
FCA
Renegade 4x2
1.4
22
R
0.84
Ford
EXPLORER FWD
2.0
23
R
0.84
Ford
MKT FWD
2.0
23
R
0.85
Ford
Transit Connect Van 2WD
1.6
20
R
0.87
GM
CRUZE
1.4
18
R
0.87
GM
CRUZE ECO
1.4
18
R
0.87
GM
SONIC
1.4
18
R
0.87
GM
SONIC RS
1.4
18
R
0.87
GM
SONIC 5
1.4
18
R
0.87
GM
SONIC 5 RS
1.4
18
R
0.88
Audi
Q5
2.0
23
R
0.88
Audi
A5 Cabriolet Quattro
2.0
22
R
0.89
Ford
EDGE AWD
2.0
23
R
0.89
JLR
Discovery Sport
2.0
21
P
0.89
JLR
LR2
2.0
21
P
0.89
BMW
X4 xDrive28i
2.0
22
P
0.89
BMW
428i xDrive Convertible
2.0
22
P
0.89
Ford
EDGE FWD
2.0
23
R
EPA effectiveness assumptions were based upon the use of Tier 2 E0 gasoline, as required for
demonstration of compliance with Federal light-duty GHG standards over the combined-cycle
test. Tier 2 E0 gasoline has properties that differ significantly from the 96 RON E20 gasoline in
the data used within AAM's comments. Tier 3 E10 91 RON gasoline and in-use 87 AKI regular-
2-319
-------
Technology Cost, Effectiveness, and Lead Time Assessment
grade gasoline also have properties that differ significantly from the E10 and E20 gasolines cited
in AAM's comments, including distillation differences and differences in carbon content.
Aromatic content and net heat of combustion, important properties in determining fuel impacts,
were not reported in the work cited by AAM.
To investigate fuel impacts on C02 emissions from turbocharged/downsized engines, EPA
conducted chassis dynamometer testing with a pickup truck having the highest BMEP
turbocharged light-duty truck engine currently in production (Ford F150 2.7L Ford EcoBoost, 6-
speed automatic transmission) and with a light-duty vehicle having the highest peak brake
thermal efficiency turbocharged downsized engine currently available in the U.S. (Honda Civic
L15B7 with 1.5L engine turbocharged, GDI engine, CVT). Testing was conducted using a Tier
2 EO 96RON/93 AKI gasoline and using a Tier 3 E10 91 RON/87 AKI gasoline, the latter having
properties similar to U.S. national average values for 87 AKI E10 in-use gasoline. Properties for
these two fuels are summarized in Appendix D of this TSD. The 2015 Ford F150 had a D/M of
approximately 1.1 L/ton while the 2017 Honda Civic had a D/M of approximately 0.9 L/ton.
The C02 emission results from the testing are summarized in Table 2.71. The chassis
dynamometer testing demonstrated a C02 reduction of just over 1 percent for the 87 RON E10
Tier 3 gasoline relative to the 96 RON EO Tier 2 gasoline for the combined cycle results. The
C02 differences over the combined cycle were statistically significant at a 95 percent confidence
level.
Table 2.71 Summary of C02 emissions from testing a Ford F150 2.7L turbocharged vehicle and a Honda
Civic 1.5L vehicle on Tier 2 and Tier 3 fuels.
Vehicle
Fuel Used
FTP (City)
HWFET
Combined
(Highway)
C02 (g/mi)
C02 (g/mi)
C02 (g/mi)
[± 95% conf. int.]
[± 95% conf.
[± 95% conf.
int.]
int.]
Ford F150 2.7 EcoBoost 6-sp Auto
Fuel C (Tier 2, EO, 93 AKI)
380.61
244.79
319.49
[1.67]
[1.80]
[1.52]
Ford F150 2.7 EcoBoost 6-sp Auto
Fuel D (Tier 3, E10, 87 AKI)
376.87
241.92
316.14
[1.74]
[0.97]
[1.34]
% Difference for Fuel D
-0.98%
-1.17%
-1.05%
Significant at 95% Confidence?
Yes
Yes
Yes
Honda Civic 1.5 Turbo CVT
Fuel C (Tier 2, EO, 93 AKI)
216.98
144.75
184.47
[0.96]
[0.38]
[0.60]
Honda Civic 1.5 Turbo CVT
Fuel D (Tier 3, E10, 87 AKI)
213.37
143.16
181.77
[0.57]
[0.77]
[0.30]
% Difference for Fuel D
-1.66%
-1.10%
-1.46%
Significant at 95% Confidence?
Yes
Yes
Yes
The test results were also similar to those found during chassis dynamometer and engine
dynamometer testing of a naturally aspirated, non-HEV Atkinson Cycle application by EPA (see
section 2.3.4.1.8.1.). A reduction of C02 emissions from light-duty vehicles over the combined
cycle for testing with Tier 3 gasoline relative to Tier 2 gasoline also appears to be a general trend
for light-duty vehicles recently tested by EPA. Preliminary test results from 10 MY2013-
MY2016 light-duty vehicles (7 passenger cars, 3 light-duty trucks) having a range of combustion
systems (GDI, PFI, Atkinson, Turbocharged) all show a similar trend of a small decrease in C02
emissions over the combined-cycle for Tier 3 gasoline relative to Tier 2 gasoline. Based on EPA
2-320
-------
Technology Cost, Effectiveness, and Lead Time Assessment
testing to date, C02 emissions from in-use grades of 87 AKI and Tier 3 gasoline result in lower
C02 emissions than results achieved during 2-cycle chassis dynamometer testing using Tier 2
gasoline.
2.3.4.1.9.2 Cost Data Used and Basis for Assumptions
Costs associated with gasoline direct injection are equivalent to those used in the Draft TAR,
updated to 2015 dollars. The GDI costs incremental to port-fuel injection for 14, V6 and V8
engines are shown below.
Table 2.72 Costs for Gasoline Direct Injection on an 13 & 14 Engine (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$241
23
$218
$215
$211
$208
$206
$203
$201
$198
$196
IC
Med2
2018
$92
$92
$69
$69
$69
$69
$68
$68
$68
TC
$310
$307
$280
$277
$274
$272
$269
$267
$265
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.73 Costs for Gasoline Direct Injection on a V6 Engine (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$363
23
$328
$323
$319
$314
$310
$306
$302
$299
$296
IC
Med2
2018
$139
$139
$104
$104
$103
$103
$103
$103
$103
TC
$467
$462
$422
$418
$413
$409
$406
$402
$399
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.74 Costs for Gasoline Direct Injection on a V8 Engine (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$436
23
$395
$389
$383
$378
$373
$368
$364
$360
$356
IC
Med2
2018
$167
$167
$125
$125
$124
$124
$124
$124
$124
TC
$562
$556
$508
$502
$497
$492
$488
$484
$480
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Costs associated with turbocharging are equivalent to those used in the Draft TAR, updated to
2015 dollars. The turbo costs incremental to naturally aspirated I-configuration and V-
configuration engines are shown below.
Table 2.75 Costs for Turbocharging, 18/21 bar, I-Configuration Engine (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$457
23
$413
$407
$401
$395
$390
$385
$381
$376
$372
IC
Med2
2018
$175
$175
$131
$130
$130
$130
$130
$130
$130
TC
$588
$581
$531
$526
$520
$515
$511
$506
$502
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.76 Costs for Turbocharging, 18/21 bar, V-Configuration Engine (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$770
23
$697
$686
$676
$666
$658
$649
$642
$634
$628
IC
Med2
2018
$295
$294
$220
$220
$219
$219
$219
$219
$219
TC
$992
$980
$896
$886
$877
$869
$861
$853
$846
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2-321
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.77 Costs for Turbocharging, 24 bar, I-Configuration Engine & for Miller-cycle I-Configuration
Engine (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$730
23
$661
$651
$641
$632
$624
$616
$609
$602
$595
IC
Med2
2024
$280
$279
$279
$278
$278
$278
$277
$277
$207
TC
$941
$930
$920
$911
$902
$894
$886
$879
$803
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.78 Costs for Turbocharging, 24 bar, V-Configuration Engine & for Miller-cycle V-Configuration
Engine (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$1,245
23
$1,127
$1,110
$1,093
$1,078
$1,064
$1,050
$1,038
$1,026
$1,015
IC
Med2
2024
$477
$476
$475
$475
$474
$473
$473
$472
$354
TC
$1,604
$1,586
$1,568
$1,553
$1,538
$1,524
$1,511
$1,499
$1,369
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Costs associated with engine downsizing are equivalent to those used in the Draft TAR,
updated to 2015 dollars. The downsizing costs incremental to the baseline engine configuration
are shown below.
Table 2.79 Costs for Downsizing as part of Turbocharging & Downsizing (dollar values in 2015$)
Downsizing from & to
Cost type
DMC: base
year cost
IC:
complexity
DMC:
learning
cu rve
IC: near
term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
14 DOHC to 13
DMC
-$218
23
-$197
-$194
-$192
-$189
-$186
-$184
-$182
-$180
-$178
14 DOHC to 14
DMC
-$96
23
-$87
-$86
-$84
-$83
-$82
-$81
-$80
-$79
-$78
V6 DOHC to 14
DMC
-$618
23
-$560
-$551
-$543
-$535
-$528
-$522
-$516
-$510
-$504
V6 SOHCto 14
DMC
-$432
23
-$391
-$385
-$379
-$374
-$369
-$365
-$360
-$356
-$352
V6 OHVto 14
DMC
$305
28
$298
$292
$286
$281
$276
$272
$268
$264
$261
V8 DOHC to V6
DMC
-$310
23
-$280
-$276
-$272
-$268
-$264
-$261
-$258
-$255
-$252
V8 SOHC 3V to V6
DMC
-$175
23
-$159
-$156
-$154
-$152
-$150
-$148
-$146
-$145
-$143
V8 SOHCto V6
DMC
-$95
23
-$86
-$84
-$83
-$82
-$81
-$80
-$79
-$78
-$77
V8 OHVto V6
DMC
$356
28
$348
$340
$334
$328
$322
$317
$313
$308
$304
14 DOHC to 13
IC
Med2
2018
$84
$83
$62
$62
$62
$62
$62
$62
$62
14 DOHC to 14
IC
Med2
2018
$37
$37
$27
$27
$27
$27
$27
$27
$27
V6 DOHC to 14
IC
Med2
2018
$237
$236
$177
$177
$176
$176
$176
$176
$176
V6 SOHCto 14
IC
Med2
2018
$166
$165
$124
$123
$123
$123
$123
$123
$123
V6 OHVto 14
IC
Med2
2018
$118
$118
$88
$88
$87
$87
$87
$87
$87
V8 DOHC to V6
IC
Med2
2018
$119
$118
$88
$88
$88
$88
$88
$88
$88
V8 SOHC 3V to V6
IC
Med2
2018
$67
$67
$50
$50
$50
$50
$50
$50
$50
V8 SOHC to V6
IC
Med2
2018
$36
$36
$27
$27
$27
$27
$27
$27
$27
V8 OHVto V6
IC
Med2
2018
$137
$137
$102
$102
$102
$102
$102
$102
$102
14 DOHC to 13
TC
-$114
-$111
-$129
-$127
-$124
-$122
-$120
-$118
-$116
14 DOHC to 14
TC
-$50
-$49
-$57
-$56
-$55
-$54
-$53
-$52
-$51
V6 DOHC to 14
TC
-$323
-$315
-$366
-$359
-$352
-$346
-$340
-$334
-$329
V6 SOHCto 14
TC
-$226
-$220
-$256
-$251
-$246
-$242
-$237
-$233
-$230
V6 OHVto 14
TC
$416
$409
$374
$369
$364
$359
$355
$351
$348
V8 DOHC to V6
TC
-$162
-$158
-$183
-$180
-$176
-$173
-$170
-$167
-$165
V8 SOHC 3V to V6
TC
-$92
-$89
-$104
-$102
-$100
-$98
-$96
-$95
-$93
V8 SOHC to V6
TC
-$49
-$48
-$56
-$55
-$54
-$53
-$52
-$51
-$50
V8 OHVto V6
TC
$485
$478
$436
$430
$424
$419
$414
$410
$406
Note:
DMC=direct manufacturing costs; IC=indirect costs; TC=total costs;
the downsized configuration is always a DOHC.
2-322
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Costs associated with turbocharging combined with engine downsizing (TDS) are similarly
equivalent to those used in the Draft TAR, updated to 2015 dollars. The TDS costs incremental
to the baseline engine configuration are shown below. Note that the costs presented below do not
include direct injection costs or other possible technologies such as cooled EGR. The costs
presented are simply the combination of the above turbo costs and downsizing costs.
Table 2.80 Costs for Turbocharging & Downsizing (2015$)
Turbo
Downsize
2017
2018
2019
2020
2021
2022
2023
2024
2025
TURB18-I
14 to 13
TC
$474
$470
$402
$399
$396
$393
$391
$388
$386
TURB18-I
14 DOHC to 14
TC
$538
$533
$475
$470
$466
$462
$458
$454
$451
TURB18-I
V6 DOHC to 14
TC
$265
$267
$165
$167
$168
$170
$171
$172
$173
TURB18-I
V6 SOHC to 14
TC
$363
$362
$276
$275
$274
$274
$273
$273
$272
TURB18-I
V6 OHV to 14
TC
$1,004
$991
$905
$894
$884
$875
$866
$857
$850
TURB18-V
V8 DOHC to V6
TC
$830
$823
$712
$706
$701
$696
$691
$686
$682
TURB18-V
V8 SOHC 3V to V6
TC
$900
$891
$792
$784
$777
$771
$764
$758
$753
TURB18-V
V8 SOHC to V6
TC
$942
$932
$840
$831
$823
$816
$809
$802
$796
TURB18-V
V8 OHV to V6
TC
$1,477
$1,458
$1,332
$1,316
$1,301
$1,288
$1,275
$1,263
$1,252
TURB24-I
14 to 13
TC
$827
$819
$791
$784
$778
$772
$766
$761
$687
TURB24-I
14 DOHC to 14
TC
$891
$881
$863
$855
$847
$840
$833
$827
$752
TURB24-I
V6 DOHC to 14
TC
$618
$615
$554
$552
$550
$548
$546
$545
$474
TURB24-I
V6 SOHC to 14
TC
$715
$710
$664
$660
$656
$652
$649
$645
$573
TURB24-I
V6 OHV to 14
TC
$1,357
$1,339
$1,294
$1,279
$1,266
$1,253
$1,241
$1,230
$1,150
TURB24-V
V8 DOHC to V6
TC
$1,442
$1,428
$1,385
$1,373
$1,362
$1,351
$1,341
$1,331
$1,204
TURB24-V
V8 SOHC 3V to V6
TC
$1,512
$1,496
$1,465
$1,451
$1,438
$1,426
$1,414
$1,404
$1,275
TURB24-V
V8 SOHC to V6
TC
$1,555
$1,537
$1,512
$1,498
$1,484
$1,471
$1,459
$1,447
$1,318
TURB24-V
V8 OHV to V6
TC
$2,089
$2,063
$2,005
$1,983
$1,962
$1,943
$1,925
$1,908
$1,774
Note: TC=total costs; the downsized configuration is always a DOHC.
Costs associated with turbocharging combined with Atkinson-2 technology (i.e., Miller-cycle)
are presented below. Note that the costs presented below do not include direct injection costs or
other required technologies such as cooled EGR. The costs presented are simply the combination
of the above turbo costs and Atkinson-2 costs presented in Section 2.3.4.1.8. Note also that the
ATK2 engine as shown in the table is always a DOHC configuration engine so also not included
in the table are the costs associated with converting, for example, a SOHC or OHV engine to a
DOHC configuration. Those costs are presented below following the cooled EGR costs. The
costs used here are identical to those used in the Draft TAR, updated to 2015 dollars.
Table 2.81 Costs for Miller Cycle (2015$)
Turbo
ATK2 engine
2017
2018
2019
2020
2021
2022
2023
2024
2025
TURB24-I
13
TC
$1,087
$1,074
$1,062
$1,051
$1,040
$1,029
$1,020
$1,010
$923
TURB24-I
14
TC
$1,087
$1,074
$1,062
$1,051
$1,040
$1,029
$1,020
$1,010
$923
TURB24-V
V6
TC
$1,824
$1,802
$1,782
$1,763
$1,745
$1,728
$1,711
$1,696
$1,549
TURB24-V
V8
TC
$1,952
$1,928
$1,906
$1,885
$1,865
$1,846
$1,828
$1,811
$1,655
Note: TC=total costs; the downsized configuration is always a DOHC.
Costs associated with cooled EGR are equivalent to those used in the Draft TAR, updated to
2015 dollars. The cooled EGR costs incremental to the baseline engine configuration are shown
below.
Table 2.82 Costs for Cooled EGR (dollar values in 2015$)
Cost type
DMC: base year cost
DMC: learning curve
2017
2018
2019
2020
2021
2022
2023
2024
2025
IC: complexity
IC: nearterm thru
2-323
-------
Technology Cost, Effectiveness, and Lead Time Assessment
DMC
$265
23
$240
$237
$233
$230
$227
$224
$221
$219
$216
IC
Med2
2024
$102
$102
$101
$101
$101
$101
$101
$101
$75
TC
$342
$338
$334
$331
$328
$325
$322
$320
$292
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Costs associated with converting non-DOHC engines to a DOHC configuration without any
engine downsizing are equivalent to those used in the Draft TAR, updated to 2015 dollars. These
costs are used when converting a non-DOHC engine to a DOHC configuration when downsizing
is not also included. The primary example for this Proposed Determination analysis is converting
to a DOHC configuration to enable Atkinson-2 technology. The costs are presented below and
do not include other potential technologies such as variable valve timing or lift or cylinder
deactivation, all of which are accounted for separately by EPA.
Table 2.83 Costs for Valvetrain Conversions from non-DOHC to DOHC (dollar values in 2015$)
Conversion
Cost type
DMC: base
year cost
IC:
complexity
DMC:
learning
cu rve
IC: near
term thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
V6 SOHC to V6 DOHC
DMC
$186
23
$169
$166
$164
$161
$159
$157
$155
$154
$152
V6 OHV to V6 DOHC
DMC
$534
28
$522
$511
$501
$492
$484
$476
$469
$462
$456
V8 SOHC 3V to V8 DOHC
DMC
$134
23
$121
$119
$118
$116
$115
$113
$112
$111
$109
V8 SOHC to V8 DOHC
DMC
$215
23
$195
$192
$189
$186
$184
$181
$179
$177
$175
V8 OHV to V8 DOHC
DMC
$585
28
$571
$559
$549
$539
$530
$521
$514
$506
$500
V6 SOHC to V6 DOHC
IC
Med2
2018
$71
$71
$53
$53
$53
$53
$53
$53
$53
V6 OHV to V6 DOHC
IC
Med2
2018
$206
$206
$154
$153
$153
$153
$153
$152
$152
V8 SOHC 3V to V8 DOHC
IC
Med2
2018
$51
$51
$38
$38
$38
$38
$38
$38
$38
V8 SOHC to V8 DOHC
IC
Med2
2018
$82
$82
$61
$61
$61
$61
$61
$61
$61
V8 OHV to V8 DOHC
IC
Med2
2018
$226
$225
$168
$168
$168
$167
$167
$167
$167
V6 SOHC to V6 DOHC
TC
$240
$237
$217
$215
$212
$210
$208
$207
$205
V6 OHV to V6 DOHC
TC
$728
$716
$654
$645
$637
$629
$622
$615
$609
V8 SOHC 3V to V8 DOHC
TC
$173
$171
$156
$154
$153
$151
$150
$149
$147
V8 SOHC to V8 DOHC
TC
$277
$274
$250
$247
$245
$243
$240
$238
$236
V8 OHV to V8 DOHC
TC
$797
$785
$717
$707
$697
$689
$681
$673
$667
Note:
DMC=direct manufacturing costs; IC=indirect costs; TC=total costs;
the downsized configuration is always a DOHC.
2.3.4.1.9.3 Basis for Feasibility Assumptions
Between 2010 and 2015, automotive manufacturers have been adopting advanced powertrain
technologies in response to GHG and CAFE standards. Just over 45 percent of MY2015 light-
duty vehicles in U.S. were equipped with gasoline direct injection (GDI) and approximately 18
percent of MY2015 light-duty vehicles were turbocharged. Nearly all vehicles using
turbocharged spark-ignition engines also used GDI to improve suppression of knocking
combustion. GDI provides direct cooling of the in-cylinder charge via in-cylinder fuel
vaporization. Use of GDI allows an increase of compression ratio of approximately 0.5 to 1.5
points relative to naturally aspirated or turbocharged engines using port-fuel-injection (e.g., an
increase from 9.9:1 for the 5.3L PFI GM Vortec 5300 to 11:1 for the 5.3L GDI GM Ecotec3 with
similar 87 AKI gasoline octane requirements).
Many automotive manufacturers have launched a third or fourth generation of GDI engines
since their initial introduction in the U.S. in 2007. Turbocharged, GDI engines are in now in
2-324
-------
Technology Cost, Effectiveness, and Lead Time Assessment
volume production at between 21-bar and 25-bar BMEP. VW/Audi, FCA, Ford and more
recently (MY2016) GM have all introduced engines with 21-25 bar BMEP in both passenger car
and light-truck platforms. The 2.7L EcoBoost engine available in the segment-leading 2017
Ford F150 pickup has just over 24-bar peak BMEP and a maximum loaded trailer weight towing
capacity in excess of 7,600 pounds. The 3.5L EcoBoost engine, also available in the 2017 Ford
F150, has a peak BMEP of 23-bar and a maximum loaded trailer weight towing capacity in
excess of 10,600 pounds.
Most recent turbocharged engine designs now use head-integrated, water-cooled exhaust
manifolds and coolant loops that separate the cooling circuits between the engine block and the
head/exhaust manifold(s). Head-integrated exhaust manifolds (IEM) are described further in the
section on thermal management in 2.2.2.11. The use of IEM was assumed within the EPA
analysis of 27-bar BMEP turbocharged GDI engines for the FRM and is assumed for all TDS24
(24-bar BMEP) engines in the Draft TAR and Proposed Determination analyses. The benefits,
including increased ability to downspeed the engine without pre-ignition and the potential for
cost savings in the design of the turbocharger turbine housing appear to extend to lower BMEP-
level turbocharged GDI engines and will likely be incorporated into many future turbocharged
light-duty vehicle applications. The application of IEMs does effect cooling system design and
manufacturers will be required to provide sufficient cooling system capacity if they adopt this
technology. Recent turbocharger improvements have included use of lower-mass, lower inertia
components and lower friction ball bearings to reduce turbocharger lag and enable higher peak
rotational speeds. Improvements have also been made to turbocharger compressor designs to
improve compressor efficiency and to expand the limits of compressor operation by improving
surge characteristics.
2.3.4.2 Transmissions: Data and Assumptions for this Proposed Determination
In assessing the effectiveness of transmission technology, EPA used multiple data sources.
These data sources include benchmarking activities, conducted at both the National Vehicle and
Fuel Emissions Lab (NVFEL) in Ann Arbor, Michigan and through contract work, technical
literature, technical conferences, vehicle certification data and stakeholder meetings. To ensure
the data were consistent, it was important to understand the assumptions made in determination
of the effectiveness. It is also important to note the engine with which the transmission is being
paired. Since much of the effectiveness associated with advanced transmissions is in the
transmission's ability to alter the operation range of the engine, and thus minimize pumping
losses, the engine efficiency in the area of operation is a major part of the effectiveness
calculation. The National Academy of Sciences, in their 2015 report, noted that "as engines
incorporate new technologies to improve fuel consumption, including variable valve timing and
lift, direct injection, and turbocharging and downsizing, the benefits of increasing transmission
ratios or switching to a CVT diminish."586 This is not to say that transmissions are not an
important technology going forward, but rather a recognition that future engines will have larger
"islands" of low fuel consumption that potentially rely less on the transmission to improve the
overall efficiency of the vehicle. Thus, effectiveness percentages reported for transmissions
paired with unimproved engines would be expected to be reduced when the same transmission is
paired with a more advanced engine. Regardless of the engine with which a particular
transmission is mated, it is expected that vehicle manufacturers and suppliers will continue to
improve the overall efficiency of the transmission itself by reducing friction and parasitic losses.
2-325
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.3.4.2.1 Assessment and Classification of Automated Transmissions (AT. AMT. DCT.
CVT)
As in the Draft TAR, transmissions have again been defined in the analysis as one of four
PPP
types
• TRX11 - 6-speed with high efficiency gearbox (HEG) level 1
• TRX12 - 6-speed with high efficiency gearbox (HEG) level 2
• TRX21 - 8-speed with high efficiency gearbox (HEG) level 1 and CVTs
• TRX22 - 8-speed with high efficiency gearbox (HEG) level 2 and improved CVTs
This differs from the FRM analysis that maintained each type of transmission separately (AT,
DCT, CVT, etc.). This change was implemented by EPA to prevent the analysis from
disproportionally implementing transmission changes as technology packages were applied in
OMEGA. The 2015 baseline fleet has transmission type (AT, DCT, CVT, etc.) that is linked to
each vehicle and is maintained throughout the analysis.
For this Proposed Determination, EPA has assessed the baseline fleet (MY2015, as described
in Chapter 1 of this TSD) and have included the following assumptions:
• All manufacturers have incorporated some level of early torque converter lockup, as
well as an appropriate level of advanced shift logic, into automatic transmissions with
six speeds and above.
• All manufacturers have incorporated some level gearbox efficiency improvements
(termed as "high efficiency gearbox" or HEG), and advanced shift logic (termed
"advanced shift logic" or ASL) into automatic transmissions with six speeds and
above, and CVTs.
• All types of automated transmissions have the potential to improve between now and
2025 MY. EPA expects that gains in efficiency can be made, independent of the
transmission type. Figure 2.116 shows that all three of the main transmission types
(AT, DCT, CVT) moving across their respective paths toward their ultimate level of
efficiency. The term "Flexibility" here denotes how well the transmission can keep
the engine on its optimal efficiency line.
ppp Each of these speed or gear designations should be taken to mean the approximate gear-ratio spread and,
therefore, inclusive of CVTs.
2-326
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Fuel Economy improvement
1. Introduction
s Transmission's performance potential can be expressed in two-
dimensional map, transmitting 'Efficiency' and ratio 'Flexibility'
Ultimate
2-Pedal
Transmission
>
¦4-J
j5
x
o
Latest
CVT CVT
Previous
Wider
ratio
Friction covera9e
reduction
w
1st Gen
CVT
A
A 8AT
A 7 AT
^6AT
T
J) 5AT
4
4AT
Wet-6
DCT
^ Dry-7
W DCT
DCT
Efficiency
l/lfm 8"1 International CTI Symposium North America 2014, Rochester nissan motor corporation
- M. Nakasaki, Jatco Ltd. and Y. Oota, NISSAN Motor Co., Ltd. - -Q ^ ^ Clt trBinln8 |„sl|,„t.
Figure 2.116 Comparison of the Different Transmission Types
• The incremental effectiveness and cost for all automated transmissions are based on
data from conventional automatics.
EPA does not believe that the technologies represented by HEG and ASL have been
incorporated into all transmissions in the 2015 fleet, but are presumed to be included in both the
base 6-speed and higher-gear transmissions, and CVTs in the 2015 fleet.
Under the premise that automated transmissions that are currently in the fleet demonstrate
different effectiveness, and with the expectation that all automated transmissions will be
improved between now and 2025 MY, 2015 transmissions were mapped to three different
designations: Null, TRX11 and TRX21. Table 2.84 shows the mapping between the existing
transmissions in the 2015 baseline fleet and the transmission designations that have been
established for this Proposed Determination analysis. Note that manual transmission
designations were left alone unless the vehicle was determined to need electrification in order to
comply in which case it would be upgraded to either a hybrid or electric vehicle transmission.
Table 2.84 Transmission Level Map
Trans code from Data
Transmission Type
Number of Gears
Transmission Level
A
Automatic
4
Null
2-327
-------
Technology Cost, Effectiveness, and Lead Time Assessment
A
Automatic
5
Null
A
Automatic
6
TRX11
A
Automatic
7
TRX21
A
Automatic
8
TRX21
A
Automatic
9
TRX21
AM
Automated Manual
5
Null
AM
Automated Manual
6
TRX11
AM
Automated Manual
7
TRX21
C
CVT
0
TRX21
D
Dual Clutch
6
TRX11
D
Dual Clutch
7
TRX21
In the "TRX" numbering system the first digit specifies the number of gears in the
transmission and the second digit specifies the HEG level. A "1" in the first digit represents a six
speed transmission and a "2" in the first digit represents an eight speed. Similarly, a "1" in the
second digit represents HEG1 and a "2" in the second digit represents HEG2. An important
aspect of using the TRX system is that it meant to estimate the effectiveness of both the current
transmission technology and future transmission technology. This is appropriate because it
allows EPA to account for technology already found in the baseline fleet, as well as apply future
transmission technology as a means of improving vehicle efficiency. With the predominant
transmission type in the 2015 MY baseline fleet (70.8 percent) being a conventional automatic
transmission. EPA believes that this approach most closely approximates the overall incremental
effectiveness and cost associated with all automated transmissions. In the future, if a particular
transmission technology develops in such a way that it becomes more cost effective compared to
our estimates, and it demonstrates the capability of meeting vehicle functional objectives, EPA
expects that vehicle manufacturers may adopt that technology instead.
The Global Automakers commented that; "The decision to do so (create a new set of terms)
unnecessarily complicated stakeholders' abilities to understand and track agency assumptions
and their progression over time." EPA has decided to maintain this methodology in this
Proposed Determination because we believe that this addresses comments previously received by
stakeholders. For example, earlier in their comments on the Draft TAR, the Global Automakers
point out that the actual penetration of DCTs in the 2015 MY fleet does not match the
technology penetration projected by EPA in the FRM. EPA recognizes that the OMEGA model
will always find the most cost effective solution. In the case of transmissions for the FRM, the
OMEGA model applied a significant number of DCTs. Based on extensive meetings with
manufacturers it became clear that the application of transmission technology was dependent on
the market and functional objectives for a particular vehicle, with conventional automatics being
the primary choice for large vehicles that tow, and DCTs being mostly applied to performance
vehicles. The TRX methodology has provided a means by which EPA maintains the type of
transmission technology found in the baseline fleet and be able to apply increasing effectiveness
and the associated costs.
In their comments, the Alliance of Automobile Manufacturers (AAM) referred to this
"binning" methodology of different types of transmissions (i.e., conventional ATs, CVTs, and
2-328
-------
Technology Cost, Effectiveness, and Lead Time Assessment
DCTs) into the TRX designations, claiming that the TRX designations "do not recognize unique
efficiencies of different transmission technologies." The Alliance therefore recommends that
EPA abandon the TRX designations and instead specifically identify each type of transmission.
In this comment, the Alliance is joined by Ford Motor Company, which agrees that accounting
for unique efficiencies of different transmissions is preferable. EPA agrees that conventional
ATs, CVTs, and DCTs do represent unique technology packages. However, the potential
effectiveness gains between TRX levels, while arising from different technology packages within
each transmission type, will be very similar among the transmission types as noted in both the
Draft TAR and earlier in this section of the TSD (see Figure 2.116). Furthermore, EPA believes
that it is important to maintain customer choice, and that manufacturers will choose the
appropriate transmission technology according to a range of customer requirements beyond C02
emissions. The TRX designation implicitly assumes that manufacturers will be likely to maintain
the transmission type already in the baseline fleet for a specific vehicle, according to their
customer requirements. Manufacturers of course have the flexibility to switch transmission
types, and gain any additional benefit in C02 reduction accruing from changing transmission
technology, but EPA does not consider this additional C02 benefit in its analysis. Thus, EPA
believes maintaining a TRX transmission designation is the best methodology for assessing
technology cost and effectiveness while maintaining maximum manufacturer flexibility.
CVT transmissions in the 2015 MY baseline fleet have been characterized as TRX21 level
transmissions. CVT transmissions were characterized as TRX11 in the Draft TAR. While EPA
recognizes that some vehicles in the fleet have older CVTs that can be characterized as TRX11,
EPA believes that it was best to characterize all CVTs as TRX21 for this Proposed
Determination to be responsive to commenters and to be conservative. Thus, EPA recognizes
the higher efficiency of current CVTs, but still allows them to improve. Most current CVT
transmissions are 85 percent efficient and are expected to be 90-94 percent efficient by 2025.
They are also expected to have their ratio span increase from the current 6-7.3 to between 8 and
8.5. Commenters questioned where these facts were obtained. These facts are based stakeholder
meetings and oral presentations given by transmission suppliers at the last several CTI
Transmission Symposiums in North America.
AAM commented, disagreeing with EPA's expectation for efficiency increased in CVTs.
Toyota also commented that "Toyota believes that the transmission effectiveness becomes less
due to the practical challenges." However, the Union of Concerned Scientists commented in
support of EPA's assumptions for CVTs, pointing to the clear benefits to CVTs as an enabling
technology. EPA has updated its estimate of CVT effectiveness within the TRX transmission
structure for this Proposed Determination, and believes that it is conservative given the current
and future efficiency and gear spread of CVTs.
2.3.4.2.2 Effectiveness Values for TRX11 and TRX21
The effectiveness associated with TRX11 is based on the GM 6T40 six-speed transmission
from the 2013 Malibu benchmarking study. A comment received from AAM questioned the
TRX11 effectiveness, but provided no further information or analysis. Consequently, EPA stands
behind its documented analysis.
2-329
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The effectiveness of TRX21 is based on the 845RE eight-speed transmission (a ZF licensed
FCA clone) from the 2014 Dodge Ram benchmarking study.587 Additional losses of 2 percent
were added to the transmission to account for the differential, which was integral to the 6T40,
and other spin losses found in front wheel drive transmissions. AAM commented that packaging
difficulties in front wheel drive transmissions tend to increase spin and churning losses. EPA
believes that the additional 2 percent losses assumed over the measured 845RE losses account
take these losses into account. In addition, in more advanced FWD transmissions, manufacturers
have tended to move clutches and other components out of the oil to further reduce churning
losses. EPA has opted to maintain the additional 2 percent losses, even for transmissions with
HEG2 (i.e., TRX22 transmissions).
A comment from Ford Motor Company stated that industry progress on transmission
efficiency should be appropriately quantified in the baseline fleet. As outlined in the assumptions
above, EPA believes that all manufacturers have incorporated some level of torque converter
lockup improvements and gearbox efficiency improvements into transmissions with six speeds
and above (i.e., TRX11 and TRX21 transmissions). Furthermore, EPA believes that the 6T40 is
reasonably representative of current six-speed transmissions in the fleet, and that the 845RE
(which is of more recent vintage than the 6T40 and contains additional efficiency improvements)
is reasonably representative of current eight-speed transmissions in the fleet. Consequently, EPA
believes industry progress on transmission efficiency has been appropriately quantified, and
stands behind its documented analysis.
Comments from AAM and Ford stated that EPA's estimated effectiveness differences
between current six- and eight-speed transmissions were high. AAM provided an attachment
entitled, "EPA ALPHA Samples Transmission Walk,"588 authored by Ford, in support. The
transmission walk suggests that a 6-speed to 8-speed HEG1 transmission upgrade would result in
a 4.4 percent - 5.0 percent effectiveness increase, rather than the 8.6 percent to 9.0 percent
calculated by Ford using ALPHA simulation runs.
However, the Ford document acknowledges a number of differences between their simulation
methodology and EPA's simulation methodology:
1) The Ford simulation engine used a 2.0L EcoBoost engine, compared to EPA's
naturally aspirated GDI engines.
2) The Ford simulation assumed the same lockup strategy between transmissions;
EPA's did not.
3) The Ford simulation used transmission efficiency maps from a Ford
8F24/8F35; EPA used benchmarked 845RE (ZF 8HP45) transmission as detailed in the
Draft TAR.
4) The Ford simulation assumed no engine displacement reduction when the
transmission is upgraded; EPA applied a "performance neutral" engine downsizing
strategy.
As described in the Draft TAR (specifically in Table 5.77 of the Draft TAR), EPA expects
that effectiveness percentages reported for transmissions paired with unimproved engines would
be reduced when the same transmission is paired with a more advanced engine. Thus, Ford's
technology walk using an EcoBoost engine would be expected to deliver a lower effectiveness
than a comparable tech walks using the naturally aspirated engines modeled in ALPHA.
2-330
-------
Technology Cost, Effectiveness, and Lead Time Assessment
EPA also believes that, generally, eight-speed transmissions within the fleet are of a later
vintage than six-speed transmissions within the fleet; and it is appropriate, when assigning
effectiveness, to account for the entire package of transmission technology changes between a
typical six- and eight- speed transmission. Thus, EPA uses representative transmissions, such as
the six-speed 6T40 and the eight-speed 8HP45, in modeling, with the understanding that
transmission efficiency, torque converter efficiency, and TC lockup strategy are different
between the two. This assumption is reflected by the fact that the additional incremental
effectiveness incorporated into HEG2 is reduced when applied to eight-speed transmissions,
which are already assumed to contain some efficiency improvements in addition to the added
gear ratios and spread.
Consistent with the FRM and recommendation by the National Academy of Science589, the
EPA analysis compares the technologies on a consistent basis by maintaining constant vehicle
performance. In the EPA analysis, engine displacement was appropriately resized to maintain a
consistent acceleration performance across different technology packages. The Ford transmission
walk explicitly maintained engine size, with no allowance for maintaining performance, arguing
that engine displacement reduction results in "significant gradeability degradation." AAM and
FCA support Ford's contention on gradeability, with FCA commenting that "if [top gear
gradeability] is too low, every time a driver encounters a small hill or wants to accelerate from a
steady speed on a level road, the transmission would have to downshift. This is very annoying
and leads to customer complaints."
EPA disagrees with this assessment. Both Ford and AAM define a gradeability metric of
maintaining top gear at 75 mph while climbing a given grade. While this may have been an
appropriate gradeability metric for vehicles containing vintage four-speed transmissions, EPA
does not believe this metric is appropriate for advanced eight-speed transmissions, where
downshifts are less noticeable to the driver. Moreover, in EPA testing, the FCA-built Dodge
Charger downshifted significantly during the relatively gentle accelerations encountered over the
HWFET. In addition, reviewers who drove the Jeep CherokeeQQQ commented that it does not
maintain top gear on a flat road at 75 mph, implying that not all vehicles in production meet this
metric.
QQQ http://www.caianddriver.com/reviews/2014-ieep-cherokee-241-first-drive-review.
http://www.tHcar.eom/20.l.5/02/is-the-20.l.5-ieep-cherokee-liniited-fhe-perfect-snv-first-impressioti/.
http://www.fourwheeler.com/vehicle-reviews/1602-ieep-cherokee-traiHiawk-whv-did-it-win-2015-four-wheeler-
of-the-vear-award/
2-331
-------
Technology Cost, Effectiveness, and Lead Time Assessment
70
60
50
Vehicle Speed
o. 30
Gear
20
10
Figure 2.117 2015 Dodge Charger Gearing Changes over the HWFET
If top gear at 75 mph were used as a metric, EPA's preliminary analysis shows that advanced
eight speed transmissions coupled with performance neutral engine sizing exhibit very little
gradeability decrease, and with some engine technologies gradeability is increased over the
baseline.
When applying the effect of these differences to the Ford simulation, the results are consistent
with the EPA effectiveness measurements taken from ALPHA sample runs and cited by Ford in
their transmission walk. EPA thus views the information in the transmission walk appendix as
corroborative.
2.3.4.2.3 Effectiveness Values for TRX12 and TRX22
The effectiveness values for TRX12 and TRX22 contain additional technologies (HEG2)
which, alone or in combination, can improve the efficiency of the gearbox.
EPA estimates of HEG2 effectiveness in eight-speed transmissions are based on modeling
studies conducted by EPA and published in a 2016 paper referenced in the Draft TAR.590 This
paper outlines potential steps to improve transmission effectiveness, including increasing gear
spread, reducing drag torque, reducing oil pump losses, reducing creep torque, implementing
earlier torque converter lockup, and reducing engine size to maintain performance neutrality.
These specific advanced transmission technologies were assessed and reported on by
transmission supplier ZF, who applied some of the technologies to their new 8-speed
transmission (the 8HP50) and modeled the effect of others.591 Results from the EPA simulations
of these technologies (reported in the 2016 paper referenced) were close to, but somewhat lower
than, the ZF estimates, so that the effectiveness numbers used by EPA for HEG2 in the Draft
TAR analysis represent a conservative analysis compared to what transmission manufacturer ZF
estimates can be achieved.
The expectation is that a transmission mapped to TRX11 can be improved to a level that
would bring the transmission effectiveness to the efficiency level of the TRX22 (with
2-332
-------
Technology Cost, Effectiveness, and Lead Time Assessment
effectiveness based on the ZF 8 speed with HEG level 2). Table 2.85 shows the effectiveness
progression from a TRX11 level transmission to the TRX22 level transmission using the 2013
Malibu engine as modeled in ALPHA.
Table 2.85 TRX11 to TRX 22 Effectiveness Progression
TRX11 to TRX12
TRX12 to TRX21
TRX21-TRX22
Range (all vehicle types)
3.4% - 3.5%
2.6% - 6.7%
3.6% - 4.4%
The aggregation of effectiveness values represents the best data available to EPA for the
Proposed Determination analysis. EPA believes that these effectiveness values are appropriate
since it allows an average of approximately 11 percent improvement in effectiveness from
TRX11 to a TRX22. An 11 percent improvement in effectiveness is achievable given that most
transmissions can gain 6-11 percent from efficiency improvements alone, and designs for
increased gear counts and wider ratio spans from 8-10 are expected.
In comparison, AAM commented that "manufacturers expect that moving from TRX11 to
TRX22 will deliver effectiveness improvements in that range of 1 percent-2 percent." Although
AAM provided no data to support this comment, they did provide the Ford transmission walk
referenced above, which provided an industry estimate that moving from TRX11 to TRX21
would deliver an effectiveness improvement of 4.4 percent to 5.0 percent. This is inconsistent
with AAM's statement that advancing farther to TRX22 will provide a total benefit of at most 2
percent.
AAM also commented on what they consider to be marginal improvements due to HEG2 (i.e.,
the additional effectiveness gain from TRX21 to TRX22), offering in support of their comment
information that FCA realized a CO2 benefit of approximately 0.8 percent unadjusted combined
FE when implementing friction reduction and hydraulic system upgrades to their eight-speed
transmission.
AAM acknowledges that the modifications completed by FCA constituted only a portion of
the HEG2 benefits expected by EPA given that certain additional improvements (notably a
change in gear ratios) was not undertaken. In fact, HEG2 does include a basket of technologies
that can be implemented individually or in combination by manufacturers; EPA does not expect
all HEG2 technologies to be implemented simultaneously. FCA chose to implement a portion of
the HEG2 technologies, and the benefit of approximately 0.8 percent is a representative
proportion of the effectiveness projected by EPA when moving from transmission level TRX21
to TRX22. The 0.8 percent effectiveness realized by FCA for the technologies implemented is
slightly lower than the values estimated by transmission supplier ZF in their published work,592
but are consistent with EPA's implementation of HEG2 in the LPM.
2.3.4.2.4 Technology Applicability and Costs
For future vehicles, it was assumed that the costs for transitioning from one technology level
(TRX11-TRX22) to another level is the same for each transmission type (AT, AMT, DCT, and
CVT). The costs used are based on AT transmissions which make up over 70.8 percent of
transmissions in today's fleet. The costs used in this analysis are equivalent to those intended for
use in the Draft TAR, updated to 2015 dollars. Note that, subsequent to the Draft TAR, EPA
2-333
-------
Technology Cost, Effectiveness, and Lead Time Assessment
found a minor error in its transmission costs whereby the indirect costs were slightly overstated.
The costs presented below correct that error with the result that total costs in MY2025 for
TRX21 are roughly $20 lower and TRX22 roughly $40 lower in this analysis than in the Draft
TAR.
Transmission technology costs are presented in Table 2.86.
Table 2.86 Costs for Transmission Improvements for all Vehicles (dollar values in 2015$)
Tech
Cost
type
DMC: base
cost
IC:
complexity
DMC: learning
curve
IC: near term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
TRX11
DMC
$40
23
$36
$36
$35
$35
$34
$34
$33
$33
$33
TRX12
DMC
$260
23
$235
$232
$228
$225
$222
$219
$217
$214
$212
TRX21
DMC
$176
23
$159
$157
$155
$152
$150
$148
$147
$145
$143
TRX22
DMC
$396
23
$359
$353
$348
$343
$338
$334
$330
$326
$323
TRX11
IC
Low2
2018
$10
$10
$8
$8
$8
$8
$8
$8
$8
TRX12
IC
Low2
2018
$63
$63
$50
$50
$50
$50
$50
$50
$50
TRX21
IC
Low2
2024
$42
$42
$42
$42
$42
$42
$42
$42
$34
TRX22
IC
Low2
2024
$95
$95
$95
$95
$95
$95
$95
$95
$76
TRX11
TC
$46
$45
$43
$42
$42
$41
$41
$41
$40
TRX12
TC
$298
$294
$278
$275
$272
$269
$267
$264
$262
TRX21
TC
$202
$199
$197
$195
$193
$191
$189
$187
$177
TRX22
TC
$454
$448
$443
$438
$433
$429
$425
$421
$399
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
As a comparison to how the Draft TAR transmission, or TRX, costs presented above would
compare to the transmission costs EPA used in the FRM, see the table below. To construct this
table, EPA has added various FRM transmission technologies (updated to 2013$) together on a
year-over-year basis and presented them along with the conceptual intent behind the new TRX
structure discussed above. Note that the FRM costs were presented in 2010$ and, importantly,
EPA revised the FRM transmission costs in 2013 due to FEV-generated updates to the tear down
costs used in the 2012 FRM.593 The FRM costs presented in the table below reflect the updates
made to the FRM costs by FEV. We present the updated values rather than the actual FRM
values since the updated values, if they were being used in this TSD analysis, are the values we
would have used. As shown in Table 2.87, the TRX system projects high costs for each
individual transmission type which is more conservative. Despite EPA projecting higher
transmission technology costs than the 2012 FRM, and having similar transmission technology
penetrations (and in some cases higher penetrations of more expensive technology), the overall
cost of compliance for the 2022 to 2025 MY standards is similar.
Table 2.87 Comparison of Transmission Costs Using the 2012 FRM Methodology to Proposed Determination
Costs for Transmissions (2015$)
Tech
Cost type
2017
2018
2019
2020
2021
2022
2023
2024
2025
6sp DCT-dry+ASL2+HEGl
TC
-$70
-$68
-$85
-$83
-$82
-$80
-$78
-$77
-$77
6sp DCT-wet+ASL2+HEGl
TC
-$30
-$29
-$40
-$39
-$38
-$37
-$36
-$36
-$36
6sp AT+ASL2+HEG1
TC
$25
$25
$24
$23
$23
$23
$23
$22
$21
TRX11
TC
$46
$45
$43
$42
$42
$41
$41
$41
$40
6sp DCT-dry+ASL2+HEG2
TC
$198
$196
$174
$172
$171
$169
$168
$167
$153
2-334
-------
Technology Cost, Effectiveness, and Lead Time Assessment
6sp DCT-wet+ASL2+HEG2
TC
$238
$235
$219
$217
$214
$212
$210
$208
$194
6sp AT+ASL2+HEG2
TC
$293
$288
$283
$279
$275
$272
$269
$266
$251
TRX12
TC
$298
$294
$278
$275
$272
$269
$267
$264
$262
8sp DCT-dry+ASL2+HEGl
TC
$92
$91
$90
$89
00
00
¦uy
00
¦uy
ID
00
¦uy
LO
00
¦uy
$79
8sp DCT-wet+ASL2+HEGl
TC
$190
$188
$186
$184
$182
$180
$178
$177
$163
8sp AT+ASL2+HEG1
TC
$124
$122
$114
$113
$112
$111
$109
$108
$106
TRX21
TC
$202
$199
$197
$195
$193
$191
$189
$187
$177
8sp DCT-dry+ASL2+HEG2
TC
$360
$354
$349
$344
$340
$336
$332
$328
$309
8sp DCT-wet+ASL2+HEG2
TC
$458
$451
$445
$439
$434
$429
$424
$420
$393
8sp AT+ASL2+HEG2
TC
$392
$386
$374
$369
$364
$360
$356
$352
$336
TRX22
TC
$454
$448
$443
$438
$433
$429
$425
$421
$399
2.3.4.3 Electrification: Data and Assumptions for this Assessment
As in the 2012 FRM and Draft TAR assessments, this Proposed Determination assessment
relies on estimates of cost and effectiveness of each GHG-reducing technology in order to
project its expected role in fleet compliance with the standards. Electrification technologies
represent a particularly broad range of cost and effectiveness, ranging from relatively low-cost
technologies offering incremental degrees of effectiveness, such as stop-start and mild hybrids,
to higher-cost, highly effective technologies such as plug-in hybrids and pure electric vehicles.
In this analysis, the costs associated with electrification are divided into battery and non-
battery costs. Chapter 2.2.4 of this TSD reviewed industry developments in battery and non-
battery technology. As discussed in the Draft TAR, many of these developments have resulted
in cost reductions for both battery and non-battery components as the industry has gained in
experience and production scale. For this Proposed Determination assessment, EPA has
reviewed its Draft TAR projections of cost and effectiveness for electrification technologies in
the 2022-2025 time frame, and in many cases has made updates based on consideration of public
comments received on the Draft TAR as well as updated information that became available since
the publication of the Draft TAR.
2.3.4.3.1 Cost and Effectiveness for Non-hybrid Stop-Start
A complete assessment of the state of non-hybrid stop-start technology was presented in
Chapter 2.2.4.4.1 of this TSD. To estimate cost and effectiveness of this technology for the
Proposed Determination analysis, EPA has considered this information as well as public
comments received on stop-start technology.
In general, public comments did not address the specific cost or effectiveness values for stop-
start as used for the Draft TAR assessment (except in the context of off-cycle credit values, as
discussed in Section B.3.4.1 of the Proposed Determination Appendix). A comment from Motor
& Equipment Manufacturers Association (MEMA) did address the effectiveness of stop-start
when implemented in a different manner from that assumed in the Draft TAR. The comment
states, "Input from our members' modeling, development vehicle testing and analysis shows that
correctly pairing two battery types together with a motor/generator can provide an additional 3
2-335
-------
Technology Cost, Effectiveness, and Lead Time Assessment
percent effectiveness beyond idle start-stop," and recommends that EPA include analysis of an
optimized lead-acid and lithium ion dual energy storage system to represent the true benefit of
such technology.
While EPA did acknowledge in the Draft TAR the possibility of pairing a battery with an
ultra capacitor (as exemplified by the Mazda i-ELOOP technology), this technology was not
analyzed more closely for effectiveness or cost, in favor of more standard configurations that are
more typical of stop-start implementation. While stop-start can certainly be implemented in other
ways that could potentially improve its cost or effectiveness, EPA does not have detailed
information on cost or performance of dual-battery or capacitor-enhanced implementations that
would allow including such variations in its analysis at this time.
No additional information was received to suggest that the Draft TAR cost or effectiveness
values for stop-start should be revised. Therefore, EPA has chosen to maintain the Draft TAR
effectiveness estimates for stop-start for use in this Proposed Determination analysis to reflect an
effectiveness of 3.0 to 4.0 percent depending on vehicle class, as shown in Table 2.88.
Table 2.88 GHG Technology Effectiveness of Stop-Start
Technology
Technology Effectiveness [%]
LPW_LRL
MPWJ.RL
HPW
LPW_HRL
MPW_HRL
Truck
12V Stop-Start
3.0
3.5
4.0
3.6
3.7
3.7
EPA is also retaining the costs for stop-start that were used in the Draft TAR, updated to 2015
dollars. The costs incremental to the baseline engine configuration for our different curb weight
classes are shown below. Note that we have, in the past, estimated costs based on vehicle classes
such as "small car" and "large MPV." As discussed in Section 2.1, we now estimate applicable
costs more appropriately on curb weight class where 1 is the lightest class and 6 is the heaviest
and is reserved for pickup trucks.
Table 2.89 Costs for Stop-Start for Different Curb Weight Classes (dollar values in 2015$)
Cost
DMC: base
DMC: learning
2017
2018
2019
2020
2021
2022
2023
2024
2025
Curb
type
cost
curve
Weight
IC: complexity
IC: nearterm
Class
thru
1
DMC
$317
25
$268
$253
$242
$233
$226
$219
$214
$209
$205
2
DMC
$317
25
$268
$253
$242
$233
$226
$219
$214
$209
$205
3
DMC
$360
25
$303
$287
$275
$265
$256
$249
$242
$237
$232
4
DMC
$360
25
$303
$287
$275
$265
$256
$249
$242
$237
$232
5
DMC
$360
25
$303
$287
$275
$265
$256
$249
$242
$237
$232
6
DMC
$395
25
$333
$315
$301
$290
$281
$273
$266
$260
$254
1
IC
Med2
2018
$121
$120
$90
$89
$89
$89
$89
$89
$88
2
IC
Med2
2018
$121
$120
$90
$89
$89
$89
$89
$89
$88
3
IC
Med2
2018
$137
$136
$102
$101
$101
$101
$101
$100
$100
4
IC
Med2
2018
$137
$136
$102
$101
$101
$101
$101
$100
$100
5
IC
Med2
2018
$137
$136
$102
$101
$101
$101
$101
$100
$100
6
IC
Med2
2018
$150
$149
$111
$111
$111
$111
$110
$110
$110
1
TC
$388
$374
$332
$323
$315
$308
$303
$298
$293
2
TC
$388
$374
$332
$323
$315
$308
$303
$298
$293
3
TC
$440
$423
$376
$366
$357
$350
$343
$337
$332
4
TC
$440
$423
$376
$366
$357
$350
$343
$337
$332
5
TC
$440
$423
$376
$366
$357
$350
$343
$337
$332
2-336
-------
Technology Cost, Effectiveness, and Lead Time Assessment
| 6 | TC | | | $483 | $464 | $413 | $401 | $392 | $383 | $376 | $370 | $364 |
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.3.2 Cost and Effectiveness for Mild Hybrids
A complete assessment of this technology as performed for the Draft TAR and Proposed
Determination was presented in Chapter 2.2.4.4.2 of this TSD. To estimate cost and effectiveness
of this technology for the Proposed Determination analysis, EPA has considered this information
as well as public comments received on the topic of mild hybrids.
Comments from Motor & Equipment Manufacturers Association (MEMA) recommended that
EPA "more closely evaluate the potential for 48V systems in its analysis as an enabling
technology that can leverage efficiencies in other vehicle systems or can provide other
flexibilities." As an example, the comment noted that "electrically heated catalysts (EHCs) can
be more efficiently powered by 48V systems to electrically light off the after treatment catalyst
faster than possible when heated solely by exhaust gases." Another example was
"thermodynamic hybridization through the use of e-boosting systems or electric supercharging."
In a similar vein, the International Council on Clean Transportation (ICCT) commented, "We
note that the TAR adds analyses of 48V hybrid systems, but we recommend that the agencies
investigate the synergies between 48V hybrids and e-boost systems." Comments received from
A123 Systems also listed a number of synergies and opportunities for increased efficiency that
are enabled by a 48V system.
EPA acknowledges that ancillary advantages and synergies can accompany adoption of 48V
systems, including such effects as faster and smoother engine start, greater opportunity for e-
boost, and higher levels of power for electrical accessories. Although these advantages have
potential to provide real value to consumers and can assist manufacturers with offering an
integrated and compelling overall package, the EPA technology assessment methodology does
not at this time include the capability to quantify the value of such ancillary benefits in a way
that could be factored in to our projections of the cost effectiveness or market penetration of 48V
technology.
Several commenters noted the decline in projected penetration of mild hybrids as compared to
the FRM analysis. For example, American Council for an Energy-Efficient Economy (ACEEE)
commented: "the penetration of mild hybrids in the agencies' 2025 compliance scenarios has
declined from the levels found in the FRM ... In the FRM, the compliance scenario included 26
percent penetration of mild hybrids in 2025, at a cost of $1553-1642. Yet, in the TAR, EPA
finds only 18.3 percent mild hybrids (table 12.33), despite a revised cost projection of $806 (p. 5-
302)."
In the EPA compliance projections presented in the Draft TAR, the projected market
penetration of a given technology is primarily an outcome of the assumptions for cost and
effectiveness that are supplied to OMEGA. The difference in projected penetration of mild
hybrids between the 2012 FRM and Draft TAR is a result of the combined effect of many
revisions and updates to technology assumptions throughout the analysis, including not only the
addition of 48V systems to the Draft TAR analysis, but also changes in the cost and effectiveness
assumptions for other technologies that compete with mild hybridization in the OMEGA model
for inclusion in the projected compliant fleet. The reduced projected penetration of mild hybrids
2-337
-------
Technology Cost, Effectiveness, and Lead Time Assessment
is therefore an outcome of the fleet compliance analysis as a whole and is not the result of any
assumption about its potential to enter the market.
Regarding battery costs projected for the 48V system modeled in the Draft TAR analysis,
A123 commented, "we find the total battery costs for 48V mild hybrid systems contained in
Table 5.124 ... to be overstated in the near term and more accurate near the end of the forecasted
period," and attributed this to assumed learning curves for this technology as applied in the EPA
analysis. The comment concluded that "this ultimately means that adoption of 48V mild hybrids
in the near term would be more cost-effective in reducing GHG emissions (and improving fuel
economy) than the DTAR projects."
Although the comment provides a reasoned discussion of the cost reduction potential from
learning for 48V batteries, EPA has not chosen to modify its application of learning to this
technology, because we continue to believe that the relatively low current penetration of 48V
systems in the U.S. and worldwide continues to lend significant uncertainty to the proper
learning rate that should be assumed. Lacking detailed and transparent data on this issue, and
because battery cost is only part of total system cost, EPA believes that any modification of the
applied learning rate that could be supported by a qualitative argument is unlikely to result in a
sufficiently large change in projected system cost to strongly affect projected near-term
penetration rates for this technology.
With regard to 48V costs, the Alliance commented that EPA's direct manufacturing costs for
48V BISG are too low, stating, "As is the case for many fuel efficiency technologies, mild
hybrids do not simply 'bolt on' to provide reductions; they affect nearly every system on a
vehicle which makes the true cost much greater than just the direct manufacturing cost price of
the motor, belt, and larger battery." The comment goes on to list a number of technical concerns
relating to performance, which were also briefly related further to other factors such as efficiency
in comments by FCA. While EPA acknowledges that integrating new technology with an
existing vehicle model to, as the comment suggests, "go from the baseline configuration to a 48V
system," may carry additional costs for modifications and integration with the baseline system, it
is also likely that over the long term, as 48V is integrated more deeply into the architecture of a
manufacturer's product line, the impact of most if not all integration costs should be minimized.
Technology cost inputs for the Draft TAR and Proposed Determination analyses are meant to
reflect a fully developed technology implementation in the 2022-2025 time frame, at a time
when manufacturers will have had opportunity to realize much of the potential benefit of design
integration. As noted in comments by A123, ICCT, and MEMA, a 48V architecture can also
enable efficiencies and improved performance in other vehicle systems, which brings substantial
value of its own. In particular, as accessories continue to follow the recent trend of demanding an
increasing amount of power (in many cases, to add features that customers demand), the
availability of a 48V architecture provides this power more easily at potentially reduced cost. For
example, electrical components such as conductors and motors may require less material and
perform more efficiently at lower currents than required by a 12V system. EPA believes that the
potential value of such efficiencies and synergies are as relevant as the initial integration costs,
and that manufacturers are likely to find ways to realize that value as it becomes available.
Broadly, the Alliance questioned the agencies' assumptions for effectiveness, cost, and
market penetration, while referring to differences in how this technology was represented by the
agencies in their respective analyses. EPA notes that estimates of cost and effectiveness that are
2-338
-------
Technology Cost, Effectiveness, and Lead Time Assessment
developed independently can ordinarily be expected to vary depending on the underlying
assumptions, methodology, and data on which they are based. The cost and effectiveness figures
used in EPA's analysis are supported by documented information and research, and on that basis
EPA believes that they represent a fair and objective assessment of this technology.
With regard to cost and effectiveness of mild hybrids, Volkswagen commented, "Our own
internal prognosis [for effectiveness] is at about 60 percent of EPA's estimates. Even in the 2020
time frame we assume the costs for 48V battery and system will still be almost twice as high as
EPA's estimates." While these differences are noted, it is also well understood that estimates of
cost and effectiveness from different sources have the potential to vary significantly depending
on the underlying assumptions, methodology, and data on which they are based. Because no data
was provided by VW to support the statement, the comment does not provide sufficient
information to fully evaluate its basis and thereby perform an effective comparison to the figures
EPA has developed from its own documented information and research.
EPA has considered the comments received on mild hybrid technology, and reviewed the
availability of additional information on this technology, and believes that the Draft TAR cost
and effectiveness values for mild hybrids remain applicable for the Proposed Determination
analysis.
For this Proposed Determination analysis, as in the Draft TAR, EPA continues to assume a
BISG configuration including a 12 kW electric machine and estimates a GHG effectiveness of
7.0 to 9.5 percent as shown in Table 2.90.
Table 2.90 GHG Technology Effectiveness of Mild Hybrids
Technology
Technology Effectiveness [%]
LPW_LRL
MPWJ.RL
HPW
LPW_HRL
MPW_HRL
Truck
12-15 kW BISG 48-120V Mild Hybrid
9.5
9.3
9.2
8.7
8.8
7.0
EPA has also updated the battery costs for mild hybrids to 2015$ and these costs are reported
in Table 2.125 of this TSD. Non- battery costs for mild hybrids have also been updated to 2015$
and are reported in Table 2.94. Full system costs are reported in Table 2.132.
2.3.4.3.3 Cost and Effectiveness for Strong Hybrids
A complete assessment of the state of strong hybrid technology was presented in Chapter
2.2.4.4.3 of this TSD. To estimate cost and effectiveness of this technology for the Proposed
Determination analysis, EPA has considered this information as well as public comments
received on the topic of strong hybrids.
For the Draft TAR, EPA calculated overall strong hybrid effectiveness by comparing the non-
hybrid variants from the same vehicle manufacturers. For example, the 2015 2.5L 14 engine
non-hybrid Camry was used to estimate the overall effectiveness of 2015 2.5L Camry hybrid.
The use of a PFI Atkinson Cycle engine, improved aerodynamics, and reduced tire rolling
resistance technology effectiveness were applied within the Lumped Parameter Model (LPM) to
2-339
-------
Technology Cost, Effectiveness, and Lead Time Assessment
better estimate the overall system effectiveness of strong hybrid electrification since the Camry
Hybrid vehicle package includes these differences in addition to the power-split HEV system.
Two-cycle fuel economy (MPG) data over the city and highway drive cycles were used to
estimate the relative effectiveness improvement of the hybrid electric vehicles. Hybrid
technology effectiveness can then be estimated by subtracting the LPM/NRC-estimated
effectiveness of non-hybrid technologies present on the vehicle from the total effectiveness.
The Draft TAR also noted that the effectiveness of input power-split hybrids and P2 parallel
hybrids appear to be converging, citing as one example the fuel economy achieved with the 2017
Hyundai Ioniq P2 hybrid with a highly hybrid-optimized 6 speed DCT transmission.
Comments from The Alliance, and repeated by Ford, were concerned with the decision to
model strong hybrids with the same cost and effectiveness without regard to specific architecture
(P2 or power split). The Alliance commented, "the architectures of these two technologies are
sufficiently different to warrant separate assessments," and recommended that EPA "develop
separate cost and effectiveness projections for power-split and P2 hybrids."
While it might be ideal to model cost and effectiveness separately for all types of strong
hybrid systems, the baseline vehicle fleet currently includes several types of strong hybrids (with
more to be released in the near future), all with similar effectiveness. In conducting technology
assessments and seeking to identify cost effective paths for compliance, EPA is primarily
concerned with representing technologies in terms of performance without promoting specific
architectures or configurations. For the FRM, Draft TAR, and Proposed Determination, a
representative strong hybrid system was needed for modeling purposes, and EPA chose the P2
strong hybrid because the component parts were straightforward to perform a cost teardown,
scaling of the system was straightforward, the technology can be applied to towing-capable
vehicles, and the production effectiveness values are similar to other strong hybrids in the
baseline fleet. This choice is not meant to suggest that EPA endorses the P2 architecture over
any other, or believes that it is equally suitable for every potential application, but was simply the
most efficient path to place a strong hybrid in the OMEGA analysis. As mentioned in the Draft
TAR, the general public literature suggests that the costs and effectiveness of many of these
strong hybrid architectures appear to be converging and in many cases are sufficiently close to
bring into question the value of maintaining separate characterizations for each.
Toyota also commented, "Toyota does not agree with [the Draft TAR statement that the P2
hybrid architecture is lower cost than PS or power split], as the P2 hybrid method is not always
lower in cost as compared to power-split method ... in PS configuration, the motor also serves as
a transmission, eliminating the need for a transmission. As a result, the PS configuration would
not necessarily be higher in cost." This comment appears to further illustrate that there remain
differences of opinion concerning the merits of each architecture, and that it can be difficult to
make firm conclusions about the differences between P2 and PS architectures. Again, EPA chose
the P2 configuration as a modeling construct for the reasons outlined above.
Toyota also pointed out, "In its assessment of the effectiveness of input power-split hybrids
and P2 parallel hybrids as getting closer, per the recent 2017 Hyundai Ioniq P2 hybrid
announcement, the Draft TAR states that the combined fuel economy of this vehicle is expected
to be about 53 mpg, which is comparable to the 52 mpg fuel economy of the 2016 GEN4 Toyota
Prius hybrid. However, this is incorrect as Eco grade model has a fuel economy rating of
56mpg." EPA acknowledges the correction, but also notes that in November 2016, Hyundai
2-340
-------
Technology Cost, Effectiveness, and Lead Time Assessment
publicly announced that the Ioniq had been certified to achieve 58 mpg combined,594 which
would continue to be comparable to the 56 mpg figure.
Volkswagen agreed with EPA's effectiveness estimates for strong hybrids, but stated that they
"estimate costs twice as high as EPA's estimates." Again, as discussed with respect to VW's
comments on mild hybrids, the comment did not provide data to support the statement or allow it
to be evaluated for comparability to the estimates that EPA has developed from its own
documented information and research, including vehicle simulation and teardown analysis.
EPA has considered the comments that were submitted on strong hybrid technology, and has
reviewed the availability of additional information on this technology. EPA believes that the
Draft TAR cost and effectiveness values for strong hybrids remain applicable for the Proposed
Determination analysis. EPA estimates the effectiveness for strong hybrid technology as shown
in Table 2.91.
Table 2.91 GHG Technology Effectiveness of Strong Hybrids
Technology
Technology Effectiveness [%]
LPW_LRL
MPWJ.RL
HPW
LPW_HRL
MPW_HRL
Truck
Strong Hybrid
19.0
20.1
19.9
18.8
19.1
17.7
For this Proposed Determination analysis, EPA has updated the battery costs for strong
hybrids, as described in Chapter 2.3.4.3.7.2 and reported in Table 2.126. Non-battery costs have
been retained for this analysis and updated to 2015$, and are reported in Table 2.95.
2.3.4.3.4 Cost and Effectiveness for Plug-in Hybrids
A complete assessment of the state of plug-in hybrid electric vehicle (PHEV) technology was
presented in Chapter 2.2.4.4.4 of this TSD. To estimate cost and effectiveness of this technology
for the Proposed Determination analysis, EPA has considered this information as well as public
comments received on the topic of PHEVs.
As discussed in the Draft TAR, plug-in hybrid electric vehicles utilize two sources of energy,
electricity and liquid fuel, which are accounted for differently according to the effectiveness
accounting methods established in the 2012 FRM. The overall GHG effectiveness potential of
PHEVs depends on many factors, the most important being the energy storage capacity designed
into the battery pack, and the vehicle's ability to provide all electric range to the operator.
Section 3.4.3.6.4 of the 2012 TSD detailed the method by which EPA estimates PHEV
effectiveness. This method estimates effectiveness based on the SAE J1711 utility factor
calculation, the AER, and the vehicle class. By this method, the assumed effectiveness for a
PHEV20 would be approximately 58 percent GHG reduction for a midsize car and
approximately 47 percent GHG reduction for a large truck.
The 2012 FRM established an incentive multiplier for compliance purposes for PHEVs sold
in MYs 2017 through 2021. This multiplier approach means that each PHEV would count as
more than one vehicle in the manufacturer's compliance calculation. The multiplier value for
PHEVs starts at 1.6 in MY2017 and phases down to a value of 1.3 in MY2021. There is no
PHEV multiplier for MYs 2022-2025.
2-341
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The 2012 FRM also set the tailpipe compliance value for the electricity portion of PHEV
energy usage to 0 g/mi for MYs 2017-2021, with no limit on the quantity of vehicles eligible for
0 g/mi tailpipe emissions accounting. For MYs 2022-2025, 0 g/mi will only be allowed up to a
per-company cumulative sales cap: 1) 600,000 vehicles for companies that sell 300,000
BEV/PHEV/FCVs in MYs 2019-2021; 2) 200,000 vehicles for all other manufacturers. For
sales above these thresholds, manufacturers will be required to account for the net upstream
GHG emissions for the electric portion of operation, using accounting methodologies set out in
the FRM.
For compliance modeling, as discussed in Section C.l of the Proposed Determination
Appendix, this Proposed Determination analysis includes an accounting for upstream emissions
associated with all electricity consumption for all manufacturers in all MY2025 OMEGA
runs.1^111
Few public comments on the Draft TAR concerned PHEVs specifically, as distinguished from
broader issues common to plug-in vehicles in general, which are addressed in their respective
applicable chapters of this TSD. One comment was received from Manufacturers of Emission
Controls Association (MECA) related to so-called "puff losses" that release emissions on cap
removal from the pressurized fuel tank that is commonly associated with PHEVs. EPA is aware
that the unique design of a PHEV, which includes not only an electrical powertrain but also a
gasoline power plant that is used on demand, poses certain difficulties with regard to cold-start,
evaporative, and cap removal emissions. While these emissions are potentially of concern, puff
losses are not directly considered in either the Draft TAR or Proposed Determination analyses
because these analyses are primarily concerned with the 2022-2025 GHG standards rather than
criteria emissions.
As with other plug-in vehicles, costs for PHEVs are separated into battery and non-battery
costs, which are discussed in their respective sections. For further discussion of these costs and
applicable updates for this Proposed Determination analysis, please refer to Chapters 2.3.4.3.6
and 2.3.4.3.7 of this TSD. Battery costs used by OMEGA for PHEVs in this analysis are reported
in Table 2.127 and Table 2.128 of this TSD. Non-battery costs are reported in Table 2.96 and
Table 2.97. Full system costs for PHEVs are reported in Table 2.134 and Table 2.135.
2.3.4.3.5 Cost and Effectiveness for Battery Electric Vehicles
A complete assessment of the state of battery electric vehicle (BEV) technology was
presented in Chapter 2.2.4.4.5 of this TSD. To estimate cost and effectiveness of this technology
for the Proposed Determination analysis, EPA has considered this information as well as public
comments received on the topic of BEVs.
EPA received a number of public comments relating to the general topic of BEVs. Additional
comments that were identified as relating more specifically to battery and non-battery costs as
they apply to BEVs are discussed separately in Chapters 2.3.4.3.6 and 2.3.4.3.7.
Many of the comments received on BEV technology were related to projected costs in the
Draft TAR. Regarding projected costs of BEVs as compared to conventional vehicles, Tesla
111111 Note that, for emissions inventory modeling, an accounting for upstream emissions associated with electricity
consumption is and always has been done, but this is different than the accounting done for compliance modeling.
2-342
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Motors commented: "The TAR assumes that BEV technology is the most complex for
automakers to develop and proliferate, regardless of range, and applies the highest cost
assumptions for BEV development through 2024. According to the TAR, for every $1.00 that
automakers spend on direct manufacturing costs for a BEV, they will spend another $0.77 on all
other costs such as R&D, corporate overhead and selling expenses. This assumption results in a
projected loss of 18 percent on BEV product lines and gives the impression that automakers
cannot profitably pursue BEV technology as a viable compliance option. However, both Tesla
and independent equity analyst projections show that this is not the case. Consensus estimates
forecast that Tesla will achieve annual corporate-level profitability in 2017."
While the Draft TAR analysis does not specifically project profitability, it is true that at the
present time, manufacturers are experiencing generally higher costs to produce a BEV than to
produce a conventional vehicle, and this differing cost basis exerts pressure on the relative
profitability of BEVs. While BEVs and conventional vehicles differ in complexity (with BEVs
commonly described as having fewer parts, simpler construction, and lower maintenance costs),
it is also true that many components specific to BEVs have not reached production volumes
similar to those of conventional vehicles. The concept of cost parity between BEVs and
conventional vehicles, and when it might be achieved, is an important construct in the
consideration of the potential for BEVs to become a large percentage of the fleet. In general, cost
parity means that the cost of ICE components that would be present in a conventional vehicle is
at least equal to the cost of electrified components that would replace them. It may also include
consideration of cost of ownership, vehicle utility, and other factors. The cost of battery and non-
battery components is obviously a major factor to cost parity.
EPA has taken considerable effort to maintain and validate the method by which it projects
battery costs, which is made possible in part by the availability of ANL BatPaC and its flexibility
to model widely differing scenarios and inputs. The ability to similarly address non-battery costs
is made more difficult by the lack of a similar model. It should also be noted that the profitability
case for a manufacturer dedicated solely to BEVs may be different from the experience of a
manufacturer that is dividing its attention between electrified and conventional vehicles. While
the current cost projections that are possible using EPA's current tool set may not represent the
full potential for optimization and cost reduction that a dedicated manufacturer may experience,
it may by contrast better represent less optimized scenarios that are likely to continue to be
applicable in the near term.
Regarding projected penetrations of BEVs in the future fleet, Faraday Future commented:
"We recognize that the Draft TAR acknowledges the trend of increasing range for BEVs and
mentions the introduction of both the Tesla Model 3 and the Chevy Bolt in Section 5.2.4.3.5.
However, the Draft TAR includes no analysis of the likely groundbreaking impact of these
models on the BEV market in the United States. Instead, the Draft TAR continues to apply
assumptions from the OMEGA and Volpe models that the increased range of BEVs will not be a
cost-effective compliance path for manufacturers. The actual actions of the auto industry in
moving to the production of BEVs shows that these assumptions are overly conservative."
EPA acknowledges the possibility that the BEV market may grow rapidly in the coming years
despite relatively low market penetration levels at the present. The penetration rates projected in
the Draft TAR are not directly selected but are primarily the result of the OMEGA model and its
selection of available technologies on the basis of cost effectiveness. The model does not at this
2-343
-------
Technology Cost, Effectiveness, and Lead Time Assessment
time have the ability to represent additional market penetration that may occur for other reasons,
such as relative utility, brand appeal, performance, or other factors.
Regarding EPA's choice of BEV200 as the longest-range BEV in the analysis, Volkswagen
stated: "by offering only 200mi BEVs, the gap between conventional and electrified cars will
remain and will fall short of fulfilling consumer expectations. To meet consumer expectations
regarding range, larger batteries would be required which ultimately results in higher costs
versus costs projected by the agencies. Therefore, we suggest including BEVs with larger battery
sizes to take these aspects into consideration."
EPA acknowledges that BEV200 represents a shorter range than seen in some current BEVs
that have well over 200 miles range. Recently this longer-range market has been dominated by
Tesla vehicles, which have constituted a premium, performance-oriented segment, but is soon
poised to add consumer-segment vehicles (such as the Chevy Bolt and Tesla Model 3). Tesla has
previously suggested that the Model 3 will have about 215 miles of range, which is not far from
the BEV200 assumption. The Chevy Bolt, now certified at 238 miles, is farther from BEV200,
but it remains to be seen whether this will in fact cause the segment to coalesce at a similar or
longer range figure over the long term. For example, Tesla may choose to increase the range of
the Model 3 to compete with the Bolt, or similarly could choose to compete on price by offering
a slightly shorter range while taking advantage of its strong brand image. It remains unclear
whether the market will coalesce around longer range vehicles at a higher cost, or settle at a
lower range with a lower cost. As previously discussed in Chapter 2.2.4 (Electrification: State of
Technology), announcements of other near-term future BEVs do not appear to be consistently
targeting a range beyond 200 miles. Ford has announced intent to introduce a BEV, described as
having an approximately 200-mile range;595 reports suggest that Toyota is planning to produce
BEVs with a range of "more than 300 km" (or 186 mi);596'597 and it continues to appear that
Nissan is likely to be targeting a 200-mile real-world range with a future version of the Leaf.598
EPA has therefore chosen to retain BEV200 for this analysis.
Regarding the argument that EPA should consider a more appropriate way to determine the
average range characteristics of the fleet for use in development of the reference and/or baseline
fleet (see comments from the Alliance of Automobile Manufacturers at pp 69-70), EPA and
CARB believe that using a sales-weighted average approach to determining range when
estimating the number of ZEV program vehicles to inject into the OMEGA analysis fleet is the
most appropriate and fair way to make the estimation. Short of that, we would need product
plans from manufacturers which we would, presumably, not be allowed to release publicly.
Without direct input from manufacturers, in the form of product plans, the approach taken seems
most appropriate and conservative. We have followed the same approach in the Proposed
Determination analysis (see Chapter 1.2 of this TSD).
Comments were also received on the subject of incentives for BEVs. As discussed in the
Draft TAR, the 2012 FRM established temporary incentives for PEVs, including an incentive
multiplier for MYs 2017 through 2021, and a 0 g/mi accounting for tailpipe emissions for MYs
2017-2025 (subject to sales thresholds for MYs 2022-2025). Public comments received on these
incentives and multipliers are addressed in Section B.3.4.2 of the Proposed Determination
Appendix.
The effectiveness of BEVs is obviously very high when their tailpipe emissions are counted
as 0 g/mi, regardless of the driving range or efficiency of the vehicle itself. In this Proposed
2-344
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Determination analysis, BEVs (on average) are assigned a lower effectiveness than in the Draft
TAR due to the addition of an accounting for upstream emissions in the compliance projections.
Our prior analyses, including the Draft TAR analysis, did not consider PEV upstream emissions
in compliance modeling. sss Given the growing rate of PEV sales, it now appears that some
manufacturers are likely to exceed the sales levels beyond which net upstream emissions would
have to be considered in their compliance determination, while other manufacturers likely will
not. Therefore, we now include upstream emissions for BEV operation and the electricity
portion of PHEV operation in the compliance determinations for all manufacturers by MY2025.
Because we wish to be conservative in our estimates, we have chosen to model all MY2025
PEVs as including upstream emissions even though it is not expected that all manufacturers will
have exceeded the sale levels by then.
As with other plug-in vehicles, costs for BEVs are separated into battery and non-battery
costs. EPA has updated battery costs for BEVs as described in Chapter 2.3.4.3.7 (Cost of
Batteries for xEVs). Discussion of non-battery costs applicable to BEVs may be found in
Chapter 2.3.4.3.6 (Cost of Non-Battery Components of xEVs). As previously mentioned, some
public comments that were related more specifically to BEV battery and non-battery costs may
be found in these chapters.
Battery costs for BEVs used by the OMEGA model are reported in Table 2.129 through Table
2.131 of this TSD. Non-battery costs are reported in Table 2.98 through Table 2.100, and full
system costs (including charging installation and equipment) are reported in Table 2.136 through
Table 2.138.
2.3.4.3.6 Cost of Non-Battery Components for xEVs
For this Proposed Determination assessment, EPA has considered public comments received
on non-battery components for xEVs, as well as reviewed the availability of additional
information regarding this topic.
EPA received several comments that related to the non-battery costs used in the Draft TAR
GHG Assessment.
Regarding general plug-in vehicle costs, Ford Motor Company stated, "In general, the cost
associated with plug-in electric technologies appears to be conservative." While not addressed
specifically to non-battery costs, non-battery costs are a part of the overall cost structure that this
comment appears to address.
Comments from Tesla Motors were more direct on this topic. Tesla commented that "Tesla's
non-battery component costs for Model 3 are lower by double-digit percentages in every
category versus the 2020 U.S. DRIVE figures considered in the TAR." With respect to this
specific comment, EPA wishes to clarify that, although the Draft TAR briefly reviewed the 2020
U.S. DRIVE cost targets for motors and power electronics, these targets were not ultimately used
sss Note that, for emissions inventory modeling, an accounting for upstream emissions associated with electricity
consumption is and always has been done, but this is different than the accounting done for compliance modeling.
2-345
-------
Technology Cost, Effectiveness, and Lead Time Assessment
by EPA in its cost projections; only the estimates for non-battery specific power were based on
U.S. DRIVE targets.
However, the comment does suggest that Tesla Motors believes that the Draft TAR non-
battery costs, regardless of their source, are significantly higher than projected by Tesla Motors
for the upcoming Model 3. Tesla stated, "Tesla's non-battery powertrain component costs for
Model 3 are dramatically lower than the costs the Agencies are considering for 2025 BEV
production ... From the 2008 Roadster to the Model 3, we have realized cost reductions of more
than 60 percent on non-battery components. These savings are due in part to improvements in the
volumetric and gravimetric profile of the components, which have led to substantial reductions in
direct manufacturing costs per unit. We see significant room for further cost reductions between
Model 3 launch in 2017 and the regulatory timeline covered in the TAR (2022 - 2025)."
While these statements are encouraging, more information would be needed to effectively
evaluate the EPA non-battery cost projections with respect to Tesla's experience.
The Tesla comments also stated, "We are very concerned by the fact that the costs presented
in the TAR related to Battery Electric Vehicles (BEVs) are significantly overstated and do not
reflect a realistic assessment of the future of this technology. If the Agencies update their BEV
assumptions to incorporate both current and planned cost reductions, the TAR will clearly show
that Zero Emission Vehicles can profitably represent a much higher portion of the automotive
industry's compliance with the 2022 - 2025 standards ... The electric powertrain costs presented
in the TAR are largely anchored to figures shared by incumbent automakers who have made
minimal efforts to deploy compelling BEV programs and have not realized the cost benefits of
high-volume manufacturing of electric powertrain components. The costs used by the Agencies
to determine the future of these regulations should reflect what is possible if the automotive
industry is sufficiently motivated to earnestly pursue mass-market BEV programs."
EPA agrees that costs for manufacturers that have aggressively pursued electrification are
likely to be lower, at least in the near term, than costs experienced by others. If this is the case,
EPA believes that an accurate accounting of electrification costs during the time frame of the
rule should represent costs as they are likely to be experienced across the full spectrum of
manufacturers, even those that may utilize PEVs as a relatively small portion of their compliance
path, as EPA projects. In order to represent a fully optimized set of costs attainable by large-scale
PEV manufacturers, EPA would require specific data, which the comment does not provide, that
establishes the degree to which these costs are outperforming the costs developed for the Draft
TAR and this Proposed Determination.
Comments from the International Council on Clean Transportation (ICCT) also described the
projected BEV costs as too high. ICCT commented, "Overall the agencies appear to have
overestimated electric vehicle costs in the TAR. The agencies have utilized state-of-the-art tools
including the DOE BatPaC model on battery costs. However, somehow costs elsewhere in the
agencies' calculations appear to have pushed up electric vehicles' incremental costs to still
remain above $10,000 in the 2025 time frame. Based on our examination of detailed engineering
cost files for the TAR, we see agency incremental technology costs for 100- and 200-mile BEVs
of $11,000 to $14,000 in 2025. We believe the agencies have overestimated these incremental
technology costs, as the ICCT's recent analysis for a similar C-class compact car are
approximately $3,100 to $7,300, respectively, for the same BEV ranges."
2-346
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Regarding both the Tesla and ICCT comments, EPA agrees that costs for battery and non-
battery components are continuing on a downward trajectory. In order to quantify that trajectory,
especially as it applies to highly optimized PEV manufacturers, EPA would need more
information than the comments provide, such as detailed cost breakdowns and the assumptions
that underlie them, in order to evaluate the comparability of the estimates and potentially use
such information to improve our non-battery cost estimates. It should also be noted that the full
system cost estimates for PEVs found in the Draft TAR include the cost of charger equipment
and installation labor, which are commonly not included in cost estimates for xEVs and so may
make the Draft TAR estimates appear higher than other sources to which they might be
compared.
With regard to production volumes assumed in the Draft TAR analysis, Global Automakers
commented: "It is important to recognize that at ... low volumes, manufacturers cannot obtain
economies of scale. In the 2012 FRM, the agencies considered a volume of 450,000 units
necessary to achieve full economies of scale. In its 2015 study, the NRC noted that the
technology penetration levels projected by the agencies did not reach that level, and that no one
manufacturer would reach that level. In the TAR, the agencies respond that economies of scale
can be obtained at levels as low as 60,000, and put forward a number of other arguments on
battery costs. Nonetheless, it cannot be denied that at current sales levels of electric-drive
vehicles of less than one percent (1 percent) of the market (i.e., less than 17,000 vehicles),
manufacturers are not close to volumes that could provide economies of scale. Unless demand
for those vehicles increases dramatically, economies of scale will remain out of reach."
While the cited arguments in the Draft TAR were directed primarily at battery costs, the
comment appears to also extend to the role of economies of scale in reducing non-battery costs.
Again, it is clear that some manufacturers will not achieve as large volumes as others, and
therefore not experience the same economies of scale as may be experienced by dedicated BEV
manufacturers during this time frame. The structure of the EPA analysis would make it very
difficult to assign different cost structures to different manufacturers, and would require
additional data specific to each manufacturer's product and research plans in order to develop or
validate related assumptions. Some commenters have strongly suggested that the EPA non-
battery cost estimates are very conservative, which if true, would tend to benefit the applicability
of the projected costs to manufacturers with smaller production volumes.
Assuming smaller volumes for the 2022 to 2025 time frame would also presuppose that
volumes cannot and will not increase dramatically over the next six to nine years. While this is
one possibility, another possibility is that innovation, regulatory forces, growth in consumer
knowledge of BEV technology, and continuing evolution in consumer expectations and
preferences will combine to increase production volumes of electrified vehicles, as several other
commenters have suggested.
Nextgen Climate America commented, "The Draft TAR overlooks several opportunities to lay
that foundation [for global GHG reduction targets] by relying on a set of unnecessarily
conservative assumptions about the capabilities and benefits of electric vehicles. There is ample
evidence that electric vehicles can offer greater benefits than they are currently assigned under
the scenarios considered in the draft TAR. Using more realistic estimates of electric vehicle
costs, capacity and benefits will better align Phase II of the light duty fuel economy and
2-347
-------
Technology Cost, Effectiveness, and Lead Time Assessment
emissions standards with expected market behavior as well as set a better foundation for the U.S.
to achieve critical climate goals."
EPA acknowledges that an accurate assessment of BEV costs is important to accurately
projecting the full potential for this technology to achieve the market penetrations necessary to
achieve large reductions in GHG emissions. EPA has accordingly continued to pursue
improvements in its modeling of battery costs, a dominant factor in BEV costs, for this Proposed
Determination analysis. Due to the design of the OMEGA model to select GHG-reducing
technologies for inclusion in potential manufacturer compliance paths primarily on the basis of
cost effectiveness and not on other potentially relevant (but difficult to quantify) factors such as
benefits of electric drive, even a greatly cost-reduced assessment of longer-range BEVs such as
BEV200 may continue to have difficulty competing with other more conventional technologies
for inclusion in these projections.
As discussed in the Draft TAR, CARB has commissioned a study on non-battery costs for
strong HEVs and PHEVs in support of its own ongoing programs.599 At the time, EPA
anticipated that this study, although it was designed for the specific needs of CARB, might also
serve as an additional source of non-battery cost findings that could be readily adapted to the
EPA non-battery cost analysis. Because it is concerned with the potential for future cost
reductions, it was expected that this would have the effect of downwardly revising our projected
non-battery costs if the findings could be effectively incorporated. This study is now underway
but is not complete, and the adaptability of the findings to the EPA cost model remains uncertain.
EPA believes that the current non-battery cost estimates as applied to the Draft TAR and this
Proposed Determination continue to represent a reasonably conservative assessment within the
context of the modeling problem as a whole.
The Draft TAR also mentioned that EPA has studied the possibility of adopting US DRIVE
cost targets for motors and power electronics, based on information gained through stakeholder
meetings that suggests that some OEMs may already be meeting or exceeding some of these
targets, or are on track to do so within the time frame of the rule. EPA ultimately decided not to
do so, largely due to uncertainty as to the basis of the target figures as representing direct
manufacturing costs as assumed for other technologies in the analysis.
Home charging equipment is another aspect of non-battery cost. In both the Draft TAR and
Proposed Determination analyses, all PEVs are assumed to be associated with a home charging
installation that includes a significant cost for installation labor, plus an additional cost for Level
1 or Level 2 charging hardware, depending on the vehicle type. PHEV20 and some PHEV40
vehicles are assigned a blend of Level 1 and Level 2 charging, while all BEVs and larger PHEVs
are assigned 100 percent Level 2 charging. Specific costs used by OMEGA are shown in Table
2.101 through Table 2.104. Public comments received on home charging, as well as public
charging infrastructure, are discussed in more detail in Chapter 2.2.4.4.5 (Battery Electric
Vehicles).
Also as discussed in the Draft TAR, the 2015 NAS report correctly noted that raw material
costs for propulsion motors tends to be a stronger function of torque output than of power output,
and recommended that the agencies scale motor costs on a torque basis. In the Draft TAR, EPA
acknowledged the technical basis of this recommendation, and pointed out that practical
considerations make it difficult to do so while remaining compatible with other aspects of the
analysis that require motors to be characterized by power output. Accurately converting between
2-348
-------
Technology Cost, Effectiveness, and Lead Time Assessment
a torque basis and a power basis would require a greater amount of information to be specified
about the individual propulsion systems and drivelines of each of the modeled PHEVs, possibly
limiting the applicability of the analysis to a narrower range of configurations than intended.
Further, through additional research and through stakeholder meetings with OEMs, EPA has
found that it is not unusual to encounter motor cost projections or targets being expressed in
terms of power, such as dollars per kilowatt. The US DRIVE cost targets for electric motors
published by the Department of Energy are also expressed in dollars per kilowatt. Finally, the
cost of the power electronics that accompany a propulsion motor system are closely related to the
power specification of the propulsion motor, and are also commonly projected or targeted as a
function of power. For these reasons, as in the Draft TAR analysis, EPA continues to scale
motor and power electronics costs in terms of power rather than torque.
No additional comment was received that includes sufficiently specific data with which the
non-battery costs used in the Draft TAR could be effectively adjusted, either to represent larger
or smaller volumes, or more or less optimized development programs (as mentioned by some of
the comments). EPA is therefore continuing to use the Draft TAR cost assumptions for non-
battery components for this Proposed Determination analysis. Although the underlying cost
basis for non-battery components remains unchanged, non-battery costs have been slightly
affected by differences in motor sizing resulting from updates to the battery sizing methodology,
as described in Chapter 2.3.4.3.7.1. The exception to this is that, for 48V MHEV non-battery
components, we continue to use the Draft TAR estimates, updated to 2015 dollars.
All applicable non-battery costs are presented in the tables below, first in terms of cost curves
as were presented in the Draft TAR, and then for each curb weight class at various mass
reduction levels. Note that we have, in the past, estimated costs based on vehicle classes such as
"small car" and "large MPV." As discussed in Chapter 2.1, we now estimate applicable costs
more appropriately on curb weight class where 1 is the lightest class and 6 is the heaviest and is
reserved for pickup trucks.
Table 2.92 Linear Regressions of Strong & Plug-in Hybrid Non-Battery System Direct Manufacturing Costs
vs Net Mass Reduction Applicable in MY2012 (2015$)
Curb Weight Class
Strong HEV
PHEV20
PHEV40
1
-$283x+$l,847
$46x+$2,183
$89x+$2,667
2
-$375x+$2,002
$61x+$2,403
$120x+$3,045
3
-$417x+$2,055
$68x+$2,486
$133x+$3,195
4
-$533x+$2,144
$88x+$2,653
-$260x+$3,585
5
-$646x+$2,366
$107x+$2,968
$209x+$4,061
6
-$682x+$2,377
n/a
n/a
Note: "x" in the equations represents the net weight reduction as a percentage.
Table 2.93 Linear Regressions of Battery Electric Non-Battery System Direct Manufacturing Costs vs Net
Mass Reduction Applicable in MY2016 (2015$)
Curb Weight Class
BEV75
BEV100
BEV200
1
$110x+-$149
$110x+-$149
$105x+-$147
2
$148x+$280
$147x+$280
$142x+$281
3
$165x+-$492
$164x+-$492
$158x+-$490
4
$214x+$13
$212x+$14
$205x+$14
5
$260x+$589
$257x+$589
$574x+$581
6
n/a
n/a
n/a
Note: "x" in the equations represents the net weight reduction as a percentage.
2-349
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.94 Costs for MHEV48V Non-Battery Items (dollar values in 2015$)
Curb
Cost type
DMC: base year cost
DMC: learning curve
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight Class
IC: complexity
IC: nearterm thru
All
DMC
$452
23
$410
$403
$397
$392
$387
$382
$377
$373
$369
All
IC
Med2
2018
$173
$173
$129
$129
$129
$129
$129
$129
$128
All
TC
$583
$576
$527
$521
$516
$511
$506
$501
$497
Note:
DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.95 Costs for Strong Hybrid Non-Battery Items (dollar values in 2015$)
Curb
WRtech
WRnet
Cost
DMC: base
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
type
year cost
learning
Class
IC:
complexity
curve
IC: near
term
thru
1
10
6
DMC
$1,830
23
$1,657
$1,631
$1,607
$1,585
$1,564
$1,544
$1,526
$1,509
$1,492
1
15
11
DMC
$1,816
23
$1,644
$1,618
$1,594
$1,572
$1,552
$1,532
$1,514
$1,497
$1,481
1
20
16
DMC
$1,802
23
$1,631
$1,606
$1,582
$1,560
$1,540
$1,520
$1,502
$1,485
$1,469
2
10
6
DMC
$1,980
23
$1,793
$1,764
$1,738
$1,714
$1,692
$1,671
$1,651
$1,632
$1,614
2
15
11
DMC
$1,961
23
$1,776
$1,748
$1,722
$1,698
$1,676
$1,655
$1,635
$1,617
$1,599
2
20
16
DMC
$1,942
23
$1,759
$1,731
$1,705
$1,682
$1,660
$1,639
$1,619
$1,601
$1,584
3
10
5
DMC
$2,034
23
$1,842
$1,813
$1,786
$1,762
$1,738
$1,717
$1,696
$1,677
$1,659
3
15
10
DMC
$2,014
23
$1,823
$1,795
$1,768
$1,744
$1,721
$1,699
$1,679
$1,660
$1,642
3
20
15
DMC
$1,993
23
$1,804
$1,776
$1,750
$1,726
$1,703
$1,682
$1,662
$1,643
$1,625
4
10
6
DMC
$2,112
23
$1,912
$1,882
$1,854
$1,828
$1,804
$1,782
$1,761
$1,741
$1,722
4
15
11
DMC
$2,085
23
$1,888
$1,858
$1,831
$1,805
$1,782
$1,759
$1,738
$1,719
$1,700
4
20
16
DMC
$2,058
23
$1,864
$1,835
$1,807
$1,782
$1,759
$1,737
$1,716
$1,697
$1,678
5
10
6
DMC
$2,328
23
$2,108
$2,074
$2,044
$2,015
$1,989
$1,964
$1,941
$1,919
$1,898
5
15
11
DMC
$2,295
23
$2,078
$2,046
$2,016
$1,988
$1,961
$1,937
$1,914
$1,892
$1,872
5
20
16
DMC
$2,263
23
$2,049
$2,017
$1,987
$1,960
$1,934
$1,910
$1,887
$1,866
$1,845
6
10
6
DMC
$2,336
23
$2,115
$2,082
$2,051
$2,023
$1,996
$1,971
$1,948
$1,926
$1,905
6
15
11
DMC
$2,302
23
$2,084
$2,052
$2,021
$1,993
$1,967
$1,943
$1,919
$1,898
$1,877
6
20
16
DMC
$2,268
23
$2,054
$2,021
$1,992
$1,964
$1,938
$1,914
$1,891
$1,870
$1,849
1
10
6
IC
Highl
2018
$1,020
$1,019
$625
$624
$624
$623
$622
$622
$621
1
15
11
IC
Highl
2018
$1,012
$1,011
$620
$619
$619
$618
$618
$617
$617
1
20
16
IC
Highl
2018
$1,004
$1,003
$615
$615
$614
$613
$613
$612
$612
2
10
6
IC
Highl
2018
$1,104
$1,102
$676
$675
$675
$674
$673
$673
$672
2
15
11
IC
Highl
2018
$1,093
$1,091
$670
$669
$668
$668
$667
$666
$666
2
20
16
IC
Highl
2018
$1,083
$1,081
$663
$663
$662
$661
$661
$660
$659
3
10
5
IC
Highl
2018
$1,134
$1,132
$695
$694
$693
$693
$692
$691
$691
3
15
10
IC
Highl
2018
$1,123
$1,121
$688
$687
$686
$685
$685
$684
$684
3
20
15
IC
Highl
2018
$1,111
$1,109
$681
$680
$679
$678
$678
$677
$677
4
10
6
IC
Highl
2018
$1,177
$1,175
$721
$720
$720
$719
$718
$718
$717
4
15
11
IC
Highl
2018
$1,162
$1,160
$712
$711
$711
$710
$709
$709
$708
4
20
16
IC
Highl
2018
$1,148
$1,146
$703
$702
$701
$701
$700
$699
$699
5
10
6
IC
Highl
2018
$1,298
$1,295
$795
$794
$793
$792
$792
$791
$790
5
15
11
IC
Highl
2018
$1,280
$1,278
$784
$783
$782
$781
$781
$780
$779
5
20
16
IC
Highl
2018
$1,262
$1,260
$773
$772
$771
$770
$770
$769
$768
6
10
6
IC
Highl
2018
$1,302
$1,300
$798
$797
$796
$795
$795
$794
$793
6
15
11
IC
Highl
2018
$1,283
$1,281
$786
$785
$784
$784
$783
$782
$782
6
20
16
IC
Highl
2018
$1,264
$1,262
$775
$774
$773
$772
$771
$771
$770
1
10
6
TC
$2,677
$2,649
$2,232
$2,209
$2,187
$2,167
$2,148
$2,130
$2,114
1
15
11
TC
$2,656
$2,629
$2,215
$2,192
$2,170
$2,150
$2,132
$2,114
$2,097
1
20
16
TC
$2,636
$2,608
$2,197
$2,175
$2,153
$2,134
$2,115
$2,097
$2,081
2
10
6
TC
$2,896
$2,866
$2,415
$2,390
$2,366
$2,345
$2,324
$2,305
$2,286
2
15
11
TC
$2,869
$2,839
$2,392
$2,367
$2,344
$2,322
$2,302
$2,283
$2,265
2
20
16
TC
$2,841
$2,812
$2,369
$2,344
$2,321
$2,300
$2,280
$2,261
$2,243
3
10
5
TC
$2,976
$2,945
$2,481
$2,456
$2,432
$2,409
$2,388
$2,368
$2,350
2-350
-------
Technology Cost, Effectiveness, and Lead Time Assessment
3
15
10
TC
$2,946
$2,915
$2,456
$2,430
$2,407
$2,385
$2,364
$2,344
$2,326
3
20
15
TC
$2,915
$2,885
$2,430
$2,405
$2,382
$2,360
$2,339
$2,320
$2,302
4
10
6
TC
$3,089
$3,057
$2,575
$2,549
$2,524
$2,501
$2,479
$2,458
$2,439
4
15
11
TC
$3,050
$3,019
$2,543
$2,517
$2,492
$2,469
$2,448
$2,427
$2,408
4
20
16
TC
$3,011
$2,980
$2,510
$2,485
$2,460
$2,438
$2,416
$2,396
$2,377
5
10
6
TC
$3,405
$3,370
$2,839
$2,810
$2,782
$2,757
$2,732
$2,710
$2,688
5
15
11
TC
$3,358
$3,323
$2,799
$2,771
$2,744
$2,718
$2,695
$2,672
$2,651
5
20
16
TC
$3,311
$3,276
$2,760
$2,732
$2,705
$2,680
$2,657
$2,635
$2,614
6
10
6
TC
$3,418
$3,382
$2,849
$2,820
$2,792
$2,767
$2,742
$2,720
$2,698
6
15
11
TC
$3,368
$3,333
$2,808
$2,779
$2,752
$2,726
$2,702
$2,680
$2,659
6
20
16
TC
$3,318
$3,284
$2,766
$2,737
$2,711
$2,686
$2,662
$2,640
$2,619
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.96 Costs for 20 Mile Plug-in Hybrid Non-Battery Items (dollar values in 2015$)
Curb
Weight
Class
WRtech
WRnet
Cost
type
DMC: base
year cost
IC:
complexity
DMC:
learning
curve
IC: near
term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
15
6
DMC
$2,185
23
$1,979
$1,948
$1,919
$1,892
$1,867
$1,844
$1,822
$1,801
$1,782
1
20
11
DMC
$2,188
23
$1,981
$1,950
$1,921
$1,894
$1,869
$1,846
$1,824
$1,803
$1,784
2
15
6
DMC
$2,407
23
$2,180
$2,145
$2,114
$2,084
$2,057
$2,031
$2,007
$1,984
$1,963
2
20
11
DMC
$2,410
23
$2,182
$2,148
$2,116
$2,087
$2,059
$2,034
$2,010
$1,987
$1,965
3
15
6
DMC
$2,490
23
$2,254
$2,219
$2,186
$2,156
$2,127
$2,101
$2,076
$2,052
$2,030
3
20
11
DMC
$2,493
23
$2,257
$2,222
$2,189
$2,159
$2,130
$2,104
$2,079
$2,055
$2,033
4
15
6
DMC
$2,658
23
$2,407
$2,369
$2,334
$2,302
$2,272
$2,243
$2,217
$2,191
$2,168
4
20
11
DMC
$2,663
23
$2,411
$2,373
$2,338
$2,306
$2,275
$2,247
$2,220
$2,195
$2,171
5
15
6
DMC
$2,975
23
$2,694
$2,651
$2,612
$2,576
$2,542
$2,510
$2,480
$2,452
$2,426
5
20
11
DMC
$2,980
23
$2,698
$2,656
$2,617
$2,581
$2,547
$2,515
$2,485
$2,457
$2,430
6
15
6
DMC
$2,996
23
$2,713
$2,670
$2,631
$2,594
$2,560
$2,528
$2,498
$2,470
$2,443
6
20
11
DMC
$3,002
23
$2,718
$2,675
$2,636
$2,599
$2,565
$2,533
$2,503
$2,475
$2,448
1
15
6
IC
Highl
2018
$1,218
$1,216
$746
$745
$745
$744
$743
$743
$742
1
20
11
IC
Highl
2018
$1,220
$1,218
$747
$746
$745
$745
$744
$743
$743
2
15
6
IC
Highl
2018
$1,342
$1,340
$822
$821
$820
$819
$819
$818
$817
2
20
11
IC
Highl
2018
$1,344
$1,341
$823
$822
$821
$821
$820
$819
$818
3
15
6
IC
Highl
2018
$1,388
$1,386
$850
$849
$848
$848
$847
$846
$845
3
20
11
IC
Highl
2018
$1,390
$1,388
$851
$850
$850
$849
$848
$847
$846
4
15
6
IC
Highl
2018
$1,482
$1,480
$908
$907
$906
$905
$904
$903
$903
4
20
11
IC
Highl
2018
$1,484
$1,482
$909
$908
$907
$906
$906
$905
$904
5
15
6
IC
Highl
2018
$1,658
$1,656
$1,016
$1,015
$1,014
$1,013
$1,012
$1,011
$1,010
5
20
11
IC
Highl
2018
$1,661
$1,659
$1,018
$1,017
$1,016
$1,015
$1,014
$1,013
$1,012
6
15
6
IC
Highl
2018
$1,670
$1,668
$1,023
$1,022
$1,021
$1,020
$1,019
$1,018
$1,017
6
20
11
IC
Highl
2018
$1,674
$1,671
$1,025
$1,024
$1,023
$1,022
$1,021
$1,020
$1,019
1
15
6
TC
$3,197
$3,164
$2,665
$2,638
$2,612
$2,588
$2,565
$2,544
$2,524
1
20
11
TC
$3,200
$3,167
$2,668
$2,640
$2,615
$2,591
$2,568
$2,547
$2,526
2
15
6
TC
$3,521
$3,485
$2,936
$2,905
$2,877
$2,851
$2,826
$2,802
$2,780
2
20
11
TC
$3,526
$3,489
$2,940
$2,909
$2,881
$2,854
$2,829
$2,806
$2,784
3
15
6
TC
$3,642
$3,605
$3,036
$3,005
$2,976
$2,948
$2,923
$2,898
$2,875
3
20
11
TC
$3,647
$3,609
$3,041
$3,009
$2,980
$2,952
$2,927
$2,902
$2,879
4
15
6
TC
$3,889
$3,849
$3,242
$3,209
$3,177
$3,148
$3,121
$3,095
$3,070
4
20
11
TC
$3,895
$3,855
$3,248
$3,214
$3,183
$3,153
$3,126
$3,100
$3,075
5
15
6
TC
$4,352
$4,307
$3,628
$3,591
$3,556
$3,523
$3,492
$3,463
$3,436
5
20
11
TC
$4,360
$4,315
$3,635
$3,597
$3,562
$3,529
$3,498
$3,469
$3,442
6
15
6
TC
$4,383
$4,338
$3,654
$3,617
$3,581
$3,548
$3,517
$3,488
$3,461
6
20
11
TC
$4,392
$4,346
$3,661
$3,623
$3,588
$3,555
$3,524
$3,495
$3,467
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2-351
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.97 Costs for 40 Mile Plug-in Hybrid Non-Battery Items (dollar values in 2015$)
Curb
WRtech
WRnet
Cost
DMC: base
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
type
year cost
learning
Class
IC:
complexity
curve
IC: near
term
thru
1
20
7
DMC
$2,673
23
$2,420
$2,382
$2,347
$2,315
$2,284
$2,256
$2,229
$2,204
$2,180
2
20
6
DMC
$3,052
23
$2,763
$2,720
$2,680
$2,643
$2,608
$2,575
$2,545
$2,516
$2,489
3
20
5
DMC
$3,202
23
$2,899
$2,854
$2,812
$2,773
$2,736
$2,702
$2,670
$2,640
$2,611
4
20
5
DMC
$3,572
23
$3,234
$3,183
$3,136
$3,093
$3,052
$3,014
$2,978
$2,944
$2,913
5
20
7
DMC
$4,076
23
$3,691
$3,633
$3,579
$3,529
$3,483
$3,439
$3,399
$3,360
$3,324
6
20
6
DMC
$4,104
23
$3,716
$3,658
$3,604
$3,554
$3,507
$3,463
$3,422
$3,383
$3,347
1
20
7
IC
Highl
2018
$1,490
$1,488
$913
$912
$911
$910
$909
$908
$908
2
20
6
IC
Highl
2018
$1,701
$1,699
$1,042
$1,041
$1,040
$1,039
$1,038
$1,037
$1,036
3
20
5
IC
Highl
2018
$1,785
$1,782
$1,094
$1,092
$1,091
$1,090
$1,089
$1,088
$1,087
4
20
5
IC
Highl
2018
$1,991
$1,988
$1,220
$1,218
$1,217
$1,216
$1,215
$1,214
$1,213
5
20
7
IC
Highl
2018
$2,272
$2,269
$1,392
$1,390
$1,389
$1,388
$1,386
$1,385
$1,384
6
20
6
IC
Highl
2018
$2,288
$2,284
$1,402
$1,400
$1,399
$1,397
$1,396
$1,395
$1,393
1
20
7
TC
$3,911
$3,870
$3,260
$3,227
$3,195
$3,166
$3,138
$3,112
$3,087
2
20
6
TC
$4,465
$4,419
$3,722
$3,684
$3,648
$3,614
$3,583
$3,553
$3,525
3
20
5
TC
$4,684
$4,636
$3,905
$3,865
$3,827
$3,792
$3,759
$3,728
$3,698
4
20
5
TC
$5,225
$5,171
$4,356
$4,311
$4,269
$4,230
$4,193
$4,158
$4,125
5
20
7
TC
$5,963
$5,901
$4,971
$4,920
$4,872
$4,827
$4,785
$4,745
$4,708
6
20
6
TC
$6,004
$5,942
$5,006
$4,954
$4,906
$4,860
$4,818
$4,778
$4,740
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.98 Costs for 75 Mile BEV Non-Battery Items (dollar values in 2015$)
Curb
WRtech
WRnet
Cost
DMC: base
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
type
year cost
learning
Class
IC:
complexity
cu rve
IC: near
term
thru
1
10
10
DMC
-$138
28
-$135
-$132
-$129
-$127
-$125
-$123
-$121
-$119
-$118
1
15
15
DMC
-$132
28
-$129
-$126
-$124
-$122
-$120
-$118
-$116
-$114
-$113
1
20
20
DMC
-$127
28
-$124
-$121
-$119
-$117
-$115
-$113
-$111
-$110
-$108
2
10
10
DMC
$295
28
$288
$282
$277
$272
$267
$263
$259
$255
$252
2
15
15
DMC
$302
28
$295
$289
$283
$278
$274
$269
$265
$262
$258
2
20
20
DMC
$310
28
$303
$296
$290
$285
$280
$276
$272
$268
$265
3
10
10
DMC
-$476
28
-$465
-$455
-$446
-$438
-$431
-$424
-$418
-$412
-$406
3
15
15
DMC
-$467
28
-$456
-$447
-$438
-$430
-$423
-$416
-$410
-$405
-$399
3
20
20
DMC
-$459
28
-$448
-$439
-$430
-$423
-$416
-$409
-$403
-$397
-$392
4
10
10
DMC
$35
28
$34
$33
$33
$32
$32
$31
$31
$30
$30
4
15
15
DMC
$46
28
$45
$44
$43
$42
$41
$41
$40
$39
$39
4
20
20
DMC
$56
28
$55
$54
$53
$52
$51
$50
$49
$49
$48
5
10
10
DMC
$615
28
$601
$588
$576
$566
$557
$548
$540
$532
$525
5
15
15
DMC
$628
28
$613
$600
$589
$578
$568
$559
$551
$544
$536
5
20
20
DMC
$641
28
$626
$613
$601
$590
$580
$571
$563
$555
$547
6
10
10
DMC
-$635
28
-$620
-$607
-$595
-$584
-$575
-$566
-$557
-$549
-$542
6
15
15
DMC
-$621
28
-$606
-$594
-$582
-$572
-$562
-$553
-$545
-$538
-$530
6
20
20
DMC
-$607
28
-$593
-$581
-$569
-$559
-$550
-$541
-$533
-$526
-$519
1
10
10
IC
High2
2024
$106
$106
$105
$105
$105
$105
$105
$105
$67
1
15
15
IC
High2
2024
$102
$101
$101
$101
$101
$101
$101
$100
$65
1
20
20
IC
High2
2024
$97
$97
$97
$97
$97
$97
$96
$96
$62
2
10
10
IC
High2
2024
$227
$226
$226
$225
$225
$225
$224
$224
$144
2
15
15
IC
High2
2024
$232
$232
$231
$231
$231
$230
$230
$230
$148
2
20
20
IC
High2
2024
$238
$237
$237
$237
$236
$236
$236
$235
$152
3
10
10
IC
High2
2024
$365
$365
$364
$363
$363
$362
$362
$361
$233
3
15
15
IC
High2
2024
$359
$358
$358
$357
$357
$356
$356
$355
$229
3
20
20
IC
High2
2024
$353
$352
$351
$351
$350
$350
$349
$349
$225
4
10
10
IC
High2
2024
$27
$27
$27
$27
$27
$27
$27
$27
$17
2-352
-------
Technology Cost, Effectiveness, and Lead Time Assessment
4
15
15
IC
High2
2024
$35
$35
$35
$35
$35
$35
$35
$35
$22
4
20
20
IC
High2
2024
$43
$43
$43
$43
$43
$43
$43
$43
$28
5
10
10
IC
High2
2024
$472
$471
$471
$470
$469
$468
$468
$467
$301
5
15
15
IC
High2
2024
$482
$481
$480
$480
$479
$478
$478
$477
$307
5
20
20
IC
High2
2024
$492
$491
$490
$490
$489
$488
$488
$487
$314
6
10
10
IC
High2
2024
$488
$487
$486
$485
$484
$484
$483
$482
$311
6
15
15
IC
High2
2024
$477
$476
$475
$474
$474
$473
$472
$472
$304
6
20
20
IC
High2
2024
$466
$465
$465
$464
$463
$463
$462
$461
$297
1
10
10
TC
-$29
-$26
-$24
-$22
-$20
-$18
-$16
-$15
-$50
1
15
15
TC
-$28
-$25
-$23
-$21
-$19
-$17
-$15
-$14
-$48
1
20
20
TC
-$26
-$24
-$22
-$20
-$18
-$16
-$15
-$13
-$46
2
10
10
TC
$515
$508
$502
$497
$492
$488
$483
$480
$396
2
15
15
TC
$528
$521
$515
$509
$504
$500
$496
$492
$406
2
20
20
TC
$541
$534
$528
$522
$517
$512
$508
$504
$416
3
10
10
TC
-$99
-$90
-$82
-$74
-$68
-$61
-$56
-$50
-$174
3
15
15
TC
-$97
-$89
-$81
-$73
-$67
-$60
-$55
-$49
-$171
3
20
20
TC
-$96
-$87
-$79
-$72
-$65
-$59
-$54
-$49
-$168
4
10
10
TC
$61
$60
$59
$59
$58
$58
$57
$57
$47
4
15
15
TC
$80
$79
$78
$77
$76
$75
$75
$74
$61
4
20
20
TC
$98
$97
$96
$95
$94
$93
$92
$92
$76
5
10
10
TC
$1,073
$1,059
$1,047
$1,036
$1,026
$1,016
$1,008
$1,000
$826
5
15
15
TC
$1,095
$1,082
$1,069
$1,058
$1,047
$1,038
$1,029
$1,021
$844
5
20
20
TC
$1,118
$1,104
$1,091
$1,080
$1,069
$1,059
$1,050
$1,042
$861
6
10
10
TC
-$132
-$120
-$109
-$99
-$90
-$82
-$74
-$67
-$232
6
15
15
TC
-$129
-$118
-$107
-$97
-$88
-$80
-$73
-$66
-$227
6
20
20
TC
-$127
-$115
-$105
-$95
-$86
-$78
-$71
-$64
-$222
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.99 Costs for 100 Mile BEV Non-Battery Items (dollar values in 2015$)
Curb
Weight
Class
WRtech
WRnet
Cost
type
DMC: base
year cost
IC:
complexity
DMC:
learning
cu rve
IC: near
term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
10
10
DMC
-$138
28
-$135
-$132
-$129
-$127
-$125
-$123
-$121
-$119
-$118
1
15
15
DMC
-$132
28
-$129
-$126
-$124
-$122
-$120
-$118
-$116
-$115
-$113
1
20
20
DMC
-$127
28
-$124
-$121
-$119
-$117
-$115
-$113
-$111
-$110
-$108
2
10
10
DMC
$295
28
$288
$282
$276
$272
$267
$263
$259
$255
$252
2
15
15
DMC
$302
28
$295
$289
$283
$278
$274
$269
$265
$262
$258
2
20
20
DMC
$310
28
$302
$296
$290
$285
$280
$276
$272
$268
$265
3
10
9
DMC
-$477
28
-$466
-$456
-$448
-$439
-$432
-$425
-$419
-$413
-$408
3
15
14
DMC
-$469
28
-$458
-$449
-$440
-$432
-$425
-$418
-$412
-$406
-$401
3
20
19
DMC
-$461
28
-$450
-$441
-$432
-$424
-$417
-$411
-$405
-$399
-$394
4
10
10
DMC
$35
28
$34
$33
$33
$32
$32
$31
$31
$30
$30
4
15
15
DMC
$45
28
$44
$43
$43
$42
$41
$40
$40
$39
$39
4
20
20
DMC
$56
28
$55
$54
$53
$52
$51
$50
$49
$49
$48
5
10
10
DMC
$615
28
$600
$588
$576
$566
$556
$548
$540
$532
$525
5
15
15
DMC
$627
28
$613
$600
$588
$578
$568
$559
$551
$543
$536
5
20
20
DMC
$640
28
$625
$612
$600
$590
$580
$571
$562
$554
$547
6
10
9
DMC
-$637
28
-$623
-$610
-$598
-$587
-$577
-$568
-$560
-$552
-$545
6
15
14
DMC
-$624
28
-$609
-$597
-$585
-$574
-$565
-$556
-$548
-$540
-$533
6
20
19
DMC
-$610
28
-$596
-$584
-$572
-$562
-$552
-$544
-$536
-$528
-$521
1
10
10
IC
High2
2024
$106
$106
$105
$105
$105
$105
$105
$105
$67
1
15
15
IC
High2
2024
$102
$101
$101
$101
$101
$101
$101
$101
$65
1
20
20
IC
High2
2024
$97
$97
$97
$97
$97
$97
$97
$96
$62
2
10
10
IC
High2
2024
$227
$226
$226
$225
$225
$225
$224
$224
$144
2
15
15
IC
High2
2024
$232
$232
$231
$231
$231
$230
$230
$230
$148
2
20
20
IC
High2
2024
$238
$237
$237
$237
$236
$236
$236
$235
$152
3
10
9
IC
High2
2024
$367
$366
$365
$365
$364
$364
$363
$363
$234
3
15
14
IC
High2
2024
$360
$360
$359
$358
$358
$357
$357
$357
$230
3
20
19
IC
High2
2024
$354
$353
$353
$352
$352
$351
$351
$350
$226
2-353
-------
Technology Cost, Effectiveness, and Lead Time Assessment
4
10
10
IC
High2
2024
$27
$27
$27
$27
$27
$27
$26
$26
$17
4
15
15
IC
High2
2024
$35
$35
$35
$35
$35
$35
$35
$35
$22
4
20
20
IC
High2
2024
$43
$43
$43
$43
$43
$43
$43
$43
$27
5
10
10
IC
High2
2024
$472
$471
$470
$470
$469
$468
$468
$467
$301
5
15
15
IC
High2
2024
$482
$481
$480
$479
$479
$478
$477
$477
$307
5
20
20
IC
High2
2024
$492
$491
$490
$489
$489
$488
$487
$487
$313
6
10
9
IC
High2
2024
$490
$489
$488
$487
$486
$486
$485
$485
$312
6
15
14
IC
High2
2024
$479
$478
$477
$477
$476
$475
$475
$474
$305
6
20
19
IC
High2
2024
$469
$468
$467
$466
$466
$465
$464
$464
$299
1
10
10
TC
-$29
-$26
-$24
-$22
-$20
-$18
-$16
-$15
-$50
1
15
15
TC
-$28
-$25
-$23
-$21
-$19
-$17
-$15
-$14
-$48
1
20
20
TC
-$26
-$24
-$22
-$20
-$18
-$16
-$15
-$13
-$46
2
10
10
TC
$515
$508
$502
$497
$492
$487
$483
$479
$396
2
15
15
TC
$527
$521
$515
$509
$504
$500
$495
$491
$406
2
20
20
TC
$540
$533
$527
$522
$517
$512
$507
$503
$416
3
10
9
TC
-$100
-$90
-$82
-$75
-$68
-$62
-$56
-$50
-$174
3
15
14
TC
-$98
-$89
-$81
-$73
-$67
-$61
-$55
-$50
-$171
3
20
19
TC
-$96
-$87
-$79
-$72
-$66
-$60
-$54
-$49
-$168
4
10
10
TC
$61
$60
$59
$59
$58
$58
$57
$57
$47
4
15
15
TC
$79
$78
$77
$77
$76
$75
$74
$74
$61
4
20
20
TC
$98
$97
$95
$94
$94
$93
$92
$91
$75
5
10
10
TC
$1,073
$1,059
$1,047
$1,036
$1,025
$1,016
$1,007
$999
$826
5
15
15
TC
$1,095
$1,081
$1,069
$1,057
$1,047
$1,037
$1,028
$1,020
$843
5
20
20
TC
$1,117
$1,103
$1,090
$1,079
$1,068
$1,059
$1,049
$1,041
$860
6
10
9
TC
-$133
-$121
-$110
-$100
-$91
-$82
-$75
-$67
-$233
6
15
14
TC
-$130
-$118
-$107
-$98
-$89
-$81
-$73
-$66
-$228
6
20
19
TC
-$127
-$116
-$105
-$96
-$87
-$79
-$71
-$65
-$223
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.100 Costs for 200 Mile BEV Non-Battery Items (dollar values in 2015$)
Curb
Weight
Class
WRtech
WRnet
Cost
type
DMC: base
year cost
IC:
complexity
DMC:
learning
cu rve
IC: near
term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
20
13
DMC
-$133
28
-$130
-$128
-$125
-$123
-$121
-$119
-$117
-$116
-$114
2
20
14
DMC
$301
28
$294
$288
$282
$277
$273
$268
$264
$261
$257
3
20
13
DMC
-$470
28
-$459
-$449
-$440
-$433
-$425
-$419
-$412
-$407
-$401
4
20
14
DMC
$43
28
$42
$41
$40
$40
$39
$38
$38
$37
$37
5
20
14
DMC
$661
28
$646
$632
$620
$609
$598
$589
$580
$572
$565
6
20
13
DMC
-$627
28
-$612
-$599
-$588
-$577
-$568
-$559
-$550
-$543
-$536
1
20
13
IC
High2
2024
$103
$102
$102
$102
$102
$102
$102
$101
$65
2
20
14
IC
High2
2024
$231
$231
$231
$230
$230
$229
$229
$229
$147
3
20
13
IC
High2
2024
$361
$360
$360
$359
$358
$358
$357
$357
$230
4
20
14
IC
High2
2024
$33
$33
$33
$33
$33
$33
$33
$33
$21
5
20
14
IC
High2
2024
$508
$507
$506
$505
$504
$504
$503
$502
$324
6
20
13
IC
High2
2024
$482
$481
$480
$479
$478
$478
$477
$476
$307
1
20
13
TC
-$28
-$25
-$23
-$21
-$19
-$17
-$16
-$14
-$49
2
20
14
TC
$526
$519
$513
$507
$502
$498
$494
$490
$405
3
20
13
TC
-$98
-$89
-$81
-$74
-$67
-$61
-$55
-$50
-$171
4
20
14
TC
$75
$74
$73
$73
$72
$71
$71
$70
$58
5
20
14
TC
$1,154
$1,139
$1,126
$1,114
$1,103
$1,093
$1,083
$1,075
$888
6
20
13
TC
-$131
-$119
-$108
-$98
-$89
-$81
-$73
-$66
-$229
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.101 Costs for In-Home Charger Associated with 20 Mile Plug-in Hybrid (dollar values in 2015$)
Curb Weight Class
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
All
DMC
$33
26
$54
$50
$48
$45
$43
$41
$40
$39
$37
All
IC
Highl
2024
$20
$20
$20
$20
$19
$19
$19
$19
$12
2-354
-------
Technology Cost, Effectiveness, and Lead Time Assessment
All | TC | | | $74 | $70 | $67 | $65 | $63 | $61 | $59 | $58 | $49
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.102 Costs for In-Home Charger Associated with 40 Mile Plug-in Hybrid (dollar values in 2015$)
Curb Weight Class
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
DMC
DMC
$175
26
$282
$264
$250
$238
$227
$218
$210
$202
2
DMC
DMC
$203
26
$327
$307
$290
$276
$264
$253
$244
$235
3
DMC
DMC
$222
26
$358
$336
$317
$302
$288
$277
$266
$257
4
DMC
DMC
$222
26
$358
$336
$317
$302
$288
$277
$266
$257
5
DMC
DMC
$222
26
$358
$336
$317
$302
$288
$277
$266
$257
6
DMC
DMC
$222
26
$358
$336
$317
$302
$288
$277
$266
$257
1
IC
IC
Highl
2024
$105
$104
$103
$102
$102
$101
$101
$100
2
IC
IC
Highl
2024
$122
$121
$120
$119
$118
$118
$117
$116
3
IC
IC
Highl
2024
$134
$132
$131
$130
$129
$128
$128
$127
4
IC
IC
Highl
2024
$134
$132
$131
$130
$129
$128
$128
$127
5
IC
IC
Highl
2024
$134
$132
$131
$130
$129
$128
$128
$127
6
IC
IC
Highl
2024
$134
$132
$131
$130
$129
$128
$128
$127
1
TC
TC
$387
$368
$353
$340
$329
$319
$310
$303
2
TC
TC
$450
$428
$410
$395
$382
$371
$361
$351
3
TC
TC
$491
$468
$448
$432
$418
$405
$394
$384
4
TC
TC
$491
$468
$448
$432
$418
$405
$394
$384
5
TC
TC
$491
$468
$448
$432
$418
$405
$394
$384
6
TC
TC
$491
$468
$448
$432
$418
$405
$394
$384
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.103 Costs for In-Home Charger Associated with All BEVs (dollar values in 2015$)
Curb Weight Class &
Range
Cost
type
DMC: base year
cost
IC: complexity
DMC: learning
cu rve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
All
DMC
$222
26
$358
$336
$317
$302
$288
$277
$266
$257
$249
All
IC
Highl
2024
$134
$132
$131
$130
$129
$128
$128
$127
$77
All
TC
$491
$468
$448
$432
$418
$405
$394
$384
$326
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.104 Costs for Labor Associated with All In-Home Chargers for Plug-in & BEV (dollar values in
2015$)
Curb Weight
Class & Range
Cost
type
DMC: base
year cost
IC: complexity
DMC:
learning
curve
IC: nearterm
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
All
DMC
$1,108
1
$1,108
$1,108
$1,108
$1,108
$1,108
$1,108
$1,108
$1,108
$1,108
All
IC
None
2024
$0
$0
$0
$0
$0
$0
$0
$0
$0
All
TC
$1,108
$1,108
$1,108
$1,108
$1,108
$1,108
$1,108
$1,108
$1,108
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.3.7 Cost of Batteries for xEVs
A significant portion of the cost of an electrified vehicle is represented by the cost of the
battery. Battery costs have many drivers, and future cost projections derived by any methodology
are subject to significant uncertainties. The choice of costing methodology is therefore an
important consideration.
2-355
-------
Technology Cost, Effectiveness, and Lead Time Assessment
A core component of the EPA battery costing methodology is BatPaC,600 a peer-reviewed
battery costing model developed by Argonne National Laboratory (ANL). As described later in
Section 2.3.4.3.7.3, the ANL BatPaC model employs a rigorous, bottom-up, bill-of-materials
approach to battery cost analysis, and has undergone continual development and review since the
2012 FRM.
BatPaC requires numerous input assumptions, including battery energy capacity, battery
output power, and many other assumptions describing the chemistry, construction, and other
aspects of the battery. EPA determines battery energy capacity and output power by means of a
battery sizing methodology that dynamically provides these inputs to BatPaC. Other inputs are
informed by information gathered from relevant sources as reviewed in EPA's assessment of the
state of battery related technologies presented in Chapter 2.2.4.5.
This section reviews how EPA developed the battery costing methodology used for the Draft
TAR and this Proposed Determination, and how inputs to the battery sizing methodology and to
BatPaC were updated for this Proposed Determination analysis. The Microsoft Excel workbooks
that EPA used to determine battery sizing and perform ANL BatPaC calculations for this
Proposed Determination are also available in the Docket.601
EPA considered public comments received on battery costs and related technologies, and has
continued to assess technology developments that have occurred since completion of the Draft
TAR. EPA has carefully considered these comments and developments in updating the battery-
related assumptions and inputs for this Proposed Determination analysis. Some comments
relating to the cost projections or methodologies in general are examined here, while other
comments relating to specific inputs are addressed later in the discussion in their respective
contexts.
Comments by Ford and Volkswagen appear to generally support the battery cost projections
of the Draft TAR. Ford commented, "In general, the cost associated with plug-in electric
technologies appears to be conservative" (subject to further understanding of the basis of the
agencies' assumptions). Volkswagen stated, "Volkswagen agrees with the projected costs for a
200mi BEV for MY2025," in the context of a discussion of range assumptions.
The Alliance of Automobile Manufacturers (AAM) commented, "Some initial feedback for
the Agencies is to ensure costs assumptions are not just for energy cells, and to present what size
the system is relative to cost, as there are economies of scale and large battery system costs can
be different from those for mild or even strong hybrids used by the automotive industry."
As discussed later in this Chapter, the EPA battery sizing methodology does in fact account
for the size and power requirements of the system by using ANL BatPaC to design each cell.
Power and energy requirements are inputs to BatPaC, and result in design of the constituent cells
to accommodate the power and energy required. Battery packs for energy-oriented systems, such
as BEVs, are composed of energy cells optimized for energy storage, while power-oriented
system such as for HEVs are composed of power-optimized cells.
AAM also commented, "Further, it may be more appropriate for the Agencies to use different
cost metrics for mild hybrids reflecting different usage and requirements for these systems."
Again, both mild and strong HEV packs are designed to provide the power and energy
requirements that are specifically assigned to each modeled vehicle.
2-356
-------
Technology Cost, Effectiveness, and Lead Time Assessment
With respect to learning rates and battery costs, AAM also commented, "while there may be
some learning for battery manufacturers, there are also many tradeoffs with this technology that
will require extensive research and development (R&D) which must be considered especially for
any new and yet to be discovered chemistries, cooling methods, or additional safety concepts."
EPA notes that, to account for indirect costs associated with electrification, such as research and
development costs, EPA applies indirect cost multipliers that are added to the direct
manufacturing costs for battery and non-battery components.
Comments from Faraday Future provided an example of battery cost per kWh that had not
been specifically reviewed in the Draft TAR. Faraday stated that a report issued by the
International Energy Agency (IEA)602'603'604 in 2015 reported costs as "below $250 per kWh,"
which Faraday described as "in the lower range of costs surveyed by the Agencies." Also,
Faraday characterized the report as projecting that "the trend of falling battery costs makes it
realistic to predict that battery costs will reach $125 per kWh — the level the Department of
Energy has estimated is needed for cost competitiveness with conventional vehicles — by 2022
... This latest information on battery costs should be added to the Agencies' analysis for the
Midterm Evaluation."
EPA acknowledges this additional source. Of course, cost per kWh can vary significantly
depending on battery capacity and power output, meaning that an estimated cost per kWh is most
meaningful in the context of a specific application. Also, it is important to know whether the cost
is being quoted as a direct manufacturing cost, or a retail price, or on some other basis. Assuming
that these costs are meant to apply to longer-range BEVs (such as BEV200) and are quoted as
direct manufacturing costs, the estimate of $125 per kWh in 2022 is not far from the
corresponding projections in the Draft TAR.
Tesla Motors commented, "Improvements in battery cell design and scale manufacturing at
the Gigafactory will enable Tesla to achieve cell-level and pack-level costs by 2020 that are far
below the 2025 TAR assumptions." EPA understands that the Gigafactory is to be a very large
scale plant that is designed to achieve significant economies of scale, and notes that the EPA
battery analysis is also based upon a very large scale manufacturing scenario for which similar
economies of scale would also be expected to apply. The EPA cost projections are based on
outputs from ANL BatPaC, which ANL describes as representing a mature, large scale plant
operating in 2020. EPA provides BatPaC with inputs describing a production scale of 450,000
packs per year and applies the output cost projections to the year 2025, generating estimates for
earlier years by reverse application of learning curves. Although Tesla may be expecting their
battery costs to be lower than this methodology projects, this observation does not provide
enough information to assess the source of the difference or the comparability of Tesla's
projections to the figures generated for the EPA analysis. While qualitative examples of specific
manufacturers' experiences with reducing battery costs are informative and welcome, for EPA to
potentially account for such information in its projection of future battery costs, the information
would ultimately have to be translated to inputs that could be appropriately and transparently
utilized by the BatPaC model, as well as inform the modeling of learning effects.
Tesla also commented that "warranty cost reserves for our current generation of Model S & X
are significantly lower than the figures assumed by the Agencies for BEVs through MY2025."
Again, while encouraging, the comment does not provide adequate information to effectively
assess or update the warranty cost figures used in the EPA analysis. Currently, the use of a
2-357
-------
Technology Cost, Effectiveness, and Lead Time Assessment
relatively high indirect cost multiplier (ICM) for warranty reserves is based on the relative
uncertainty associated with new technologies. While Tesla may be experiencing lower costs than
this factor would suggest, it is unclear whether this experience will translate equally well to all
manufacturers serving all segments and operating at widely varying production levels during the
time frame of the rule.
Tesla Motors also provided comment regarding the projected size of battery packs, an
important determinant of their cost. Tesla stated, "the battery capacity assumed in the TAR to
achieve 200 miles of electric range is overstated, resulting in an inflated total cost figure for the
battery pack. The Agencies estimate that it will take 56kWh of energy storage to achieve 200
miles of electric range in a Small MPV, however, our technology achieves more than 200 miles
of range with a smaller battery capacity than the TAR's 2025 estimates."
EPA has reexamined the inputs and assumptions to the battery sizing methodology and
believes that the updated analysis conducted for this Proposed Determination more accurately
projects the needed capacity of battery packs for all modeled PEVs. The updates and their effect
on battery sizing are discussed in Chapter 2.3.4.3.7 (Cost of Batteries for xEVs).
Regarding the acceleration performance of modeled PEVs, Mercedes-Benz commented, "The
electric vehicle powertrains assumed in EPA's analysis are undersized compared to what would
be required to match the performance of a conventional vehicle's powertrain. This undersized
powertrain results in significantly lower cost than would be required, which in turn
underestimates our fleet level cost. As a part of the MTE, the agencies should revisit... the
assumptions regarding Mercedes-Benz vehicle performance and future reduction potentials.
Mercedes-Benz recommends that the agencies develop unique performance criteria for each
vehicle in a manufacturer's fleet consistent with the performance of the vehicles in the baseline
fleet that are being replaced."
The context of this comment suggests that it refers specifically to the observation that
Mercedes-Benz vehicles tend to have a greater acceleration performance than most of the
conventional vehicles that form the baseline fleet from which average PEV acceleration targets
were derived, and that because of this, the PEV battery and non-battery sizings that EPA applies
to an average vehicle in each class would not as faithfully represent the higher performance
vehicles typical of the Mercedes-Benz fleet.
EPA acknowledges that different manufacturers target different levels of performance in order
to accommodate the requirements of customers in the market they choose to serve. This is also
true for other vehicle attributes, such as styling, luxuriousness, cargo capacity, towing capability,
and so on. The EPA analysis, particularly as modified for this Proposed Determination, attempts
to account for variations in performance by defining six vehicle classes that are distinguished by
differences in power-to-weight ratio and road load. While this improves the ability of the
analysis to represent much of the variation in performance across the overall fleet, some
particularly high-performance (and, potentially, low-performance) product lines may not be
represented as well. While modeling the performance of every individual vehicle in each
manufacturer's fleet might be an ideal approach, the need to conduct the analysis in a practical
manner requires the aggregation of vehicles into a limited number of groups. Particularly with
respect to the battery and motor sizing problem, the EPA battery analysis already designs battery
packs and specifies power output for 150 modeled PEVs, which would multiply dramatically if
the analysis were extended to include individual manufacturers' fleets.
2-358
-------
Technology Cost, Effectiveness, and Lead Time Assessment
EPA also acknowledges that some PEV models may be targeting higher 0-60 acceleration
targets than seen in comparable conventional vehicles, and that some PEV manufacturers,
particularly those in the premium segment, appear to be marketing improved acceleration as an
advantage of electrified vehicles. It remains to be seen, however, whether this trend will be as
pronounced in the consumer segment over the longer term. As described in Chapter 2.3.1.2
(Performance Assumptions), throughout the Draft TAR and previous analyses, EPA has taken
the approach of modeling GHG-reducing technologies as being implemented in a performance-
neutral manner. Whether or not manufacturers do increase 0-60 acceleration time for PEVs
compared to conventional vehicles, all PEVs are likely to offer faster response "off the line" and
at lower speeds, due to the high low-end torque of the electric motor, which means that some
performance advantage is likely to be present even if 0-60 times are not substantially increased.
2.3.4.3.7.1 Battery Sizing Methodology for BEVs and PHEVs
This section describes how EPA specified battery packs for modeled BEVs and PHEVs
(referred to collectively here as PEVs). For HEVs, EPA used a different methodology that is
described in the next section.
Specifying a PEV battery pack primarily involves determining the necessary energy storage
capacity (in kWh) and power capability (in kW) to provide a desired driving range and level of
acceleration performance. Energy storage capacity has a strong influence on the weight of the
pack as well as its overall cost because it determines the amount of active energy storage
material that must be included in the battery. Power capability has an influence on weight and
also has a strong influence on cost because it determines how the materials are arranged as well
as the relative proportion of active materials to inactive materials in each cell.
Because most PEV battery chemistries are known to experience degradation in power and
energy capacity over time (also known as power fade and capacity loss respectively), it is also
important to consider how performance at end-of-life might differ from beginning-of-life, and
consider the need for increasing the target capacity or power to ensure that performance goals
can be met for the life of the vehicle.
The choice of battery energy capacity is primarily a function of the energy efficiency of the
vehicle and the target driving range. Because range may decline over time due to battery
degradation, this raises the question of whether the target range should be considered a
beginning-of-life or end-of-life criterion. Current regulatory practice, as exemplified by the EPA
labeling guidelines for PHEVs and BEVs,605 measures range at beginning-of-life and omits any
adjustment for future capacity degradation. For PHEVs, however, current regulatory practice for
the EPA GHG standards effectively requires vehicle manufacturers to consider degradation in
range as it will directly affect the calculated in-use emissions if tested for compliance at any time
during full useful life.TTT Accordingly, for PHEVs, manufacturers may use a combination of
TTT As noted in Section 2.3.4.3.4, PHEV GHG emissions are calculated using the SAE J1711 utility factor and AER.
Accordingly, if range degrades during useful life, the utility factor correction would change and thus, the
calculated GHG emissions would increase. As EPA's GHG emission standards are full useful life standards and
2-359
-------
Technology Cost, Effectiveness, and Lead Time Assessment
battery oversizing and an energy management strategy that provides for a consistent range
throughout the useful life. For BEVs, however, rather than oversizing the battery sufficiently to
maintain the original EPA range over time, manufacturers have tended to make the customer
aware of the possibility of range loss and in some cases have warranted the battery to a specified
degree of capacity retention over a specified period of time. For example, Nissan warrants their
24-kWh Leaf battery to retain nine of 12 capacity bars (corresponding to about 70 percent
capacity) for 60 months or 60,000 miles, and warrants their 30-kWh battery for 96 months or
100,000 miles. As another example, Tesla does not warrant against a specific degree of capacity
loss but makes it clear that some capacity loss is normal and provides the customer with
recommendations for preserving battery capacity.
The choice of battery power capability is primarily governed by vehicle performance
expectations. In the case of BEVs and many longer-range PHEVs, the battery is sufficiently
large that its power capability is likely to naturally exceed that needed for acceleration
performance alone. These batteries effectively have a power reserve that provides a natural
buffer against power fade. Smaller batteries, such as those of shorter-range PHEVs, may lack
this advantage and may need to be sized deliberately to meet a target power capability, in which
case power fade should be factored in to the sizing process because it could lead to loss of
performance and loss of utility factor over the life of the vehicle.
As discussed in the Draft TAR, at the time of the 2012 FRM, the task of assigning battery
capacity and power for the many PEV configurations to be analyzed was a very difficult task,
with few well-developed techniques and tools available. Further, it was necessary to choose
assumptions to reflect an expected state of technology in the 2020 to 2025 time frame, even
though few production vehicles were available at the time to either serve as a reference for the
current state of technology or to establish trends for its advancement. The EPA methodology
therefore employed a wide variety of simplifying assumptions and estimation methods in order
to conduct the effort in a practical way while using calculation tools that are easily accessible to
external reviewers.
The Draft TAR reviewed in detail the method originally used in the 2012 FRM and the
improvements that were implemented for the Draft TAR analysis. Readers interested in the
origin of the method and the changes applicable to the Draft TAR analysis may refer to the Draft
TAR, Section 5.3.4.4.7.
After completion of the Draft TAR, public comments and updated information led to a
number of updates to the methodology and assumptions as employed in this Proposed
Determination analysis. The discussion below focuses on reviewing the core methodology,
followed by a description of the updates.
The EPA battery and motor sizing analysis is a spreadsheet-based method that determines
battery energy capacities and power capabilities for a large array of modeled PEVs. Because
battery capacity and power requirements are strongly influenced by vehicle weight, and battery
weight is a function of capacity and power while also being a large component of vehicle weight,
vehicles are considered noncompliant if their emissions exceed the certified emission level by more than 10
percent during the useful life, manufacturers must account for degradation or risk exceeding the GHG standards
in-use.
2-360
-------
Technology Cost, Effectiveness, and Lead Time Assessment
sizing the battery for a BEV or PHEV requires an iterative solution. This problem is well suited
to the iteration function available in common spreadsheet software. A spreadsheet-based
methodology was therefore selected as being sufficiently powerful while remaining accessible to
public inspection using standard commercially available software. EPA used Microsoft Excel
for this purpose, with the Iteration setting enabled and set to 100 iterations.
The EPA approach begins by defining a large group of example PEVs for which battery packs
are then specified in detail and analyzed for direct manufacturing cost. The array of PEVs
includes five electrified vehicle types (BEV75, BEV100, BEV200, PHEV20, and PHEV40), six
baseline vehicle classes represented by different curb weights, and five levels of target curb
weight reduction (0, 2, 7.5, 10, and 20 percent). This results in a total of 150 PEV instances,uuu
each characterized by a driving range, a baseline curb weight, and a level of target curb weight
reduction, as shown in Figure 2.118. A sizing spreadsheet determined battery energy capacities
and battery power requirements for each vehicle, in conjunction with ANL BatPaC which
determined battery specific energy (kWh/kg) for use by the sizing spreadsheet, and ultimately a
pack cost estimate. Pack cost, electric drive power ratings, and the necessary level of mass
reduction applied to the glider (the baseline vehicle minus powertrain components) for each
vehicle were then utilized by the OMEGA model.
111111 For each of the 150 vehicles, two battery cathode chemistries (NMC622 and blended LMO/NMC) and four
production volumes (50K, 125K, 250K and 450K) were also considered, resulting in the generation of 1,200
individual battery cost estimates.
2-361
-------
Technology Cost, Effectiveness, and Lead Time Assessment
5 electrified
vehicle types
EV75
EV100
EV200
PHEV20
PHEV40
X
6 baseiine
curb weights
SmCar
StdCar
LgCar
SmMPV
LgMPV
Truck
5 levels of
curb weight
reduction
X
0%
2%
7.5%
10%
20%
)
150 vehicle instances,
characterized by „ h
i I 1 apply *
range %WR r %MR -
CWtase to glider
up to 20% MR
Battery sizing
assumptions
cw,
battery battery motor
kWh/kg kW kWh kW
Battery design
assumptions
Sizing
spreadsheet
ANL BatPaC
to OMEGA
total applied
%MR
net %WR
from CWbase
electric drive
power (kW)
battery pack
DMC in 2025
(current $)
Figure 2.118 EPA PEV Battery and Motor Sizing Method
Method for Sizing of Battery Energy Capacity
Battery energy capacity was considered to be a function of desired driving range (mi) and
vehicle energy consumption (Wh/mi).
Driving range was defined by the various range configurations (BEV75, BEV100, BEV200,
PHEV20, and PHEV40) and was considered to be an approximate real-world, EPA-label range.
As in the Draft TAR analysis, this Proposed Determination analysis considers PHEV40 range to
be an all-electric range without assistance from the engine under any vehicle operating
conditions, while the PHEV20 range is an effective electrically-powered range resulting from a
blended-operation architecture.
Energy consumption was estimated by taking into account the weight of the battery necessary
to deliver this range, and many other factors.
To estimate energy consumption for a given PEV instance, first its curb weight was estimated
as equal to the curb weight CWbase of the corresponding baseline conventional vehicle, modified
by any applicable curb weight reduction WRtarget (0, 2, 7.5, 10, or 20 percent), and further
modified by subtraction of the weight of conventional powertrain components (for BEVs) and
2-362
-------
Technology Cost, Effectiveness, and Lead Time Assessment
addition of the weight of electric content (for BEVs and PHEVs), as shown in Equation 4
through Equation 7.
Equation 4. Target curb weight reduction
WRtarget = %WR*CWbase
Equation 5. Weight-reduced curb weight
CWbase reduced - CWbase — WRtarget
Equation 6. Raw curb weight of BEV
CWggy — CWbase redUCed — WlCE_powertrain + Weiectric_content
Equation 7. Raw curb weight of PHEV
CWpHjjy — CWbase reduced ^electric_content
The curb weights CWbase of conventional baseline vehicles were derived from the baseline
fleet for a set of six vehicle classes corresponding to the vehicle classes used in the LPM.
The assumed weights of the removed conventional powertrain components (called "weight
delete," or Wice powertrain) varied for each of the six vehicle classes, as an approximate function of
power. Electric content weight (Weiectric content) consisted of estimated battery weight and electric
drive weight (motor and power electronics). Since the weight of this content is strongly
influenced by total vehicle weight and many other variables, it is not a constant figure but is
iteratively computed by the spreadsheet. The computation utilized estimates of battery specific
energy and estimates of the specific power of traction motors and power electronics applicable to
the 2020 to 2025 time frame. In practice, the specific energy of a battery pack will vary
depending on its power-to-energy (P/E) ratio and its energy capacity. In general, smaller more
power-optimized batteries tend to show a lower specific energy than larger energy-optimized
batteries. The analysis utilizes a direct link to ANL BatPaC to pull in dynamically updated values
for battery specific energy. For BEVs, a gearbox weight of 50 pounds was also added.
To estimate the weight of non-battery components, EPA referred to performance targets for
non-battery components published by US DRIVE. US DRIVE606 is a consortium involving the
U.S. Department of Energy, USCAR (an organization of the major U.S. automakers), and
several other organizations including major energy companies and public energy utilities. This
industry collaboration has established a number of cost and performance targets for automotive
traction motors, inverters, chargers, and other power electronics components for the 2015 and
2020 time frames.607 These include targets for specific power of electric propulsion motors and
power electronics, both separately and alone, as shown in Table 2.105. These metrics are
particularly relevant to the problem of component sizing.
Table 2.105 U.S. Drive Targets for Non-Battery Specific Power for 2015 and 2020
Component
U.S Drive Target (kW/kg)
2015
2020
Electric motor and power electronics
1.2
1.4
Electric motor alone
1.3
1.6
2-363
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Power electronics alone
12
14.1
Since the EPA battery sizing methodology does not distinguish the power rating of the power
electronics from that of the drive motor, the US DRIVE target that would be most relevant to the
battery analysis is the specific power of electric motor and power electronics combined, which
US DRIVE places at 1.4 kW/kg for the 2020 time frame. The method therefore estimates the
weight of non-battery PEV components at 1.4 kW/kg.
As described in the Draft TAR, this figure has some support in the literature. A presentation
by Bosch608 at The Battery Show 2015 states that the electric motor and power electronics for a
100 kW, 20 kWh BEV system in the 2025 time frame is expected to comprise about 37 percent
of electric content weight, with battery weight comprising the remaining 63 percent. Assuming
the 20 kWh battery pack has a specific energy of about 140 Wh/kg (as indicated by BatPaC for
an NMC622 pack at 115 kW net battery power), and a corresponding weight of 143 kg, the non-
battery content would be estimated at about 53 kg. The 100 kW system would then represent
100 kW/53 kg or 1.88 kW/kg, making the US DRIVE figure of 1.4 kW/kg appear conservative.
Although the US DRIVE figures are targets and therefore not necessarily indicative of
industry status, EPA has confidence that the targets for specific power represent attainable goals
during the 2022 to 2025 time frame. This is based in part on the observation that the 2020
specific power target for electric motor and power electronics combined is very close to levels
that were already being attained by some production vehicles at the time they were set.609 Also,
confidential business information conveyed to EPA through private stakeholder meetings with
OEMs conducted since the FRM suggests that some of these targets are already being met or
exceeded in production components today, or are expected to be met within the time frame of the
rule.
The "raw" curb weight calculations of Equation 6 and Equation 7, if used directly, would
typically generate estimated PEV curb weights that are significantly larger than the curb weights
of the baseline vehicles on which they are based, due to the added weight of the large battery
which may weigh more than the removed components. For several reasons noted below, EPA
chose to further constrain the iteration by forcing the projected curb weight (CWBEvor CWphev)
of each PEV to match the curb weight (CWbase reduced) of the corresponding baseline vehicle. In
order to achieve this objective, EPA solved for the exact percentage of mass reduction that would
need to be applied to the glider in order to offset the difference in curb weight, and applied that
level of mass reduction to cause the curb weights to match. In cases where more than 20 percent
mass reduction technology would have been necessary to offset the difference, it was capped at
20 percent and only in these cases was the curb weight of the electrified vehicle allowed to vary.
In part, EPA chose to constrain the PEV curb weights because it helps to differentiate
between "applied" mass reduction and "net" curb weight reduction throughout the analysis. EPA
differentiates between applied and net reduction because they are used in different ways in the
analysis. Net curb weight reduction refers to a reduction in curb weight, and is used for
estimating energy consumption. Applied mass reduction refers to percentage mass reduction
applied to the glider, and is used for estimating the cost of mass reduction technology that has
been embodied in the vehicle. Often, to achieve a given amount of net curb weight reduction,
2-364
-------
Technology Cost, Effectiveness, and Lead Time Assessment
more mass reduction technology might need to be applied to electrified vehicles than to
conventional vehicles because of the added weight of the electric content.
For example, as shown in Table 2.106, a BEV200 that benefits from application of 20 percent
mass reduction technology to the glider may achieve a net curb weight reduction of only about
13 percent. In such a case, EPA would base the estimate of BEV200 mass reduction technology
costs on a 20 percent applied mass reduction, while basing the estimate of BEV200 battery and
motor costs on battery and motor sizings that are based on the energy and power requirements
associated with only a 13 percent net curb weight reduction.
Table 2.106 Example Net Curb Weight Reduction for BEVs and PHEVs With 20% Mass Reduction
Technology Applied to Glider
BEV75
BEV100
BEV200
PHEV20
PHEV40
Curb weight reduction achieved by application of 20% MR tech
Wt Class 1
20%
19%
13%
10%
6%
Wt Class 2
20%
19%
13%
11%
6%
Wt Class 3
20%
19%
13%
11%
6%
Wt Class 4
20%
19%
13%
10%
5%
Wt Class 5
20%
18%
12%
11%
5%
Wt Class 6
20%
18%
14%
10%
5%
In theory, rather than constraining the PEV curb weights, a similar result could have been
achieved by applying the various weight reduction cases directly to the glider and allowing the
curb weights to grow as they might. This would have generated a different set of applied and net
reduction data points, with more data points representing little or no applied mass reduction,
higher curb weight, and higher energy consumption and larger batteries as a result. However,
because the high cost of battery capacity tends to improve the cost effectiveness of mass
reduction technology in PEV applications, EPA expects that manufacturers are likely to
implement significant mass reduction in most PEVs, meaning that cases with little or no applied
mass reduction are of limited interest to the analysis. The chosen method generates a greater
density of points at the higher percentages of applied weight reduction that are most likely to
represent industry practice.
After determining the PEV curb weight (which in most cases was constrained to match the
baseline curb weight, but now carries a specific degree of applied mass reduction in order to do
so), the method then computes the loaded vehicle weight (also known as inertia weight or
equivalent test weight (ETW)) by adding 300 pounds to the curb weight:
Equation 8. Equivalent test weight (ETW) of PEVs
ETWpEV(Zfc) = CWPE v(Zfc) + 300
The method then uses this test weight to develop an energy consumption estimate. First, it
estimates the fuel economy (mi/gal) for a conventional light-duty vehicle (LDV) of that test
weight by a regression formula derived from the relationship between 2-cycle fuel economy and
inertia weight. Compiled data on fuel economy vs. test weight from the EPA Trends Report
provided the primary data source. From this data, EPA then derived a polynomial regression
formula for fuel economy (mi/gal) as a function of ETW, the format of which is shown in
2-365
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Equation 9. Specific coefficient values of A, B, and C used for the Draft TAR and revised for
this Proposed Determination analysis are listed in a later discussion.
Equation 9. MY2008 conventional LDV fuel economy regression formula
FEconv(mi/gal) = A x ETWpev2 — B x ETWpev + C
This was then converted to a gross Wh/mile figure, assuming 33,700 Wh of energy per gallon
of gasoline as shown in Equation 10:
Equation 10. Gross energy consumption (Wh/mile)
EgrossjTp(Wh/mi) = (—r—) x 33,700
r ^conv
This figure was then brought into electrified vehicle space by applying a series of adjustments
representing assumed differences in energy losses between conventional vehicles and electrified
vehicles. This required making assumptions for several powertrain efficiencies:
(a) Brake efficiency: For conventional vehicles, this is the percentage of chemical fuel energy
converted to energy at the engine crankshaft. For electrified vehicles, it is the percentage of
stored battery energy converted to shaft energy entering the transmission. It therefore includes
battery discharge efficiency and inverter and motor efficiency.
(b) Driveline efficiency: the percentage of brake energy entering the transmission and
delivered through the driveline to the wheels. It includes transmission efficiency and
downstream losses (such as wheel bearing, axle, and brake drag losses), but not tire rolling
resistance.
(c) Cycle efficiency: the percentage of energy delivered to the wheels that is used to overcome
road loads in moving the vehicle (that is, the portion of wheel energy that is not later lost to
friction braking). This efficiency is larger for vehicles with regenerative braking.
The efficiencies assumed for baseline conventional vehicles were based on efficiency terms
derived from EPA's lumped parameter model (LPM). Values for electrified powertrain
efficiencies for BEVs and PHEVs of varying battery sizes were chosen in order to represent
expected component efficiencies and to achieve a reasonable estimate of electrified energy
consumption as indicated by the resulting battery capacity projections. Specific values can be
inspected in the EPA Battery Analysis spreadsheets which are available in Docket EPA-HQ-
OAR-2015-0827.
PEV road loads were also adjusted relative to conventional vehicles to represent assumed
reductions in aerodynamic drag and rolling resistance applicable to these vehicles. PEVs were
assigned a 20 percent reduction in both aerodynamic drag and rolling resistance from 2008
baseline levels. The effect was estimated by the LPM and then applied to the computed road
load. Because the LPM estimates that a 20 percent improvement in aerodynamic drag and rolling
resistance will reduce road loads to approximately 90.5 percent of baseline, road loads were
reduced by that amount. The effect of reductions in curb weight were not estimated by the LPM
but instead were inherently represented by use of the ETW regression formula to convert curb
weights into base energy consumption estimates.
2-366
-------
Technology Cost, Effectiveness, and Lead Time Assessment
The combined effect of these steps means that the estimated energy consumption of each PEV
is therefore derived from the energy consumption of a corresponding baseline conventional
vehicle by applying a ratio of the road loads of the PEV (%RoadloadpEv) to those of the baseline
vehicle (%RoadloadConv =1) and a ratio of the assumed efficiencies (rj) of the respective
powertrains, as shown in Equation 11.
Equation 11. PEV unadjusted energy consumption
„ T1 I f%Roadloadp/EV Y\vehicle_ conv \
Ep/Ev_FTp(Wh/mi) — Egross_FTP * I 0/ n , * ~ I
y /oRoadloadconv T]vehicle_P/EV J
Equation 11 yields a laboratory (unadjusted) two-cycle FTP energy consumption estimate. To
represent a real-world energy consumption, the analysis applies a derating factor to convert
unadjusted fuel economy to real-world fuel economy. The EPA range labeling rule specifies a
default derating factor of 70 percent, with provisions for using a different (custom) factor based
on optional 5-cycle testing. This analysis applied a varying derate value depending on vehicle
configuration, as described later.
Applying the derate factor (as shown with an example value of 70 percent in Equation 12)
results in the PEV on-road energy consumption estimate that the method uses to determine the
required battery pack capacity for the vehicle.vvv
Equation 12. PEV on-road energy consumption
1
E0nroad(Wh/mt) — Ep/ev ftp * ( )
Finally, as shown by Equation 13, the method determines the required battery energy capacity
(BEC) as the on-road energy consumption in Wh/mile, multiplied by the desired range in miles,
divided by the usable portion of the battery capacity, or usable SOC design window. The
assumed usable SOC design window (SOC%) varied between BEVs and PHEVs and is
discussed in a later section.
Eonroad^r) x range(mi)
BECCWh) = * n,
v J SOC%
Equation 13. Required battery pack energy capacity for PEVs
As mentioned previously, the intensively iterative nature of the battery capacity sizing
problem means that all of the preceding calculations are constructed in a spreadsheet as circular
references and performed iteratively by the spreadsheet software until the estimated weights,
ranges, and energy consumption figures converge.
Method for Sizing of Battery Power Capability
vvv As described later, this Proposed Determination analysis uses a 70 percent factor for most PEVs but applies a
custom derating factor of 75 percent for BEV200 based on examples of recent industry practice.
2-367
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Another input to the battery sizing process is the required power capability of the battery.
Battery power capability was derived from an assigned peak motor power, which in turn was
considered to be a function of desired acceleration performance.
PHEV40 was conceptualized as a range-extended electric vehicle, with a motor and battery
sized to be capable of providing pure all-electric range in all driving situations. PHEV20 was
modeled as a blended-operation vehicle where the motor is often assisted by the engine during
the charge depletion phase. This means that PHEV40 motor power ratings in this analysis are
likely to be higher than would apply to a blended-operation PHEV40. PHEVs were configured
with a single propulsion motor, in contrast to some production PHEV designs that split the total
power rating between two motors. Most PHEVs also include a second electric machine used
primarily as a generator. The analysis does not explicitly assign a weight to this component but
considers it as part of the weight of the conventional portion of the powertrain, which retains its
original weight despite the likelihood of downsizing in a PHEV application.
Acceleration performance was represented by assigning a power-to-weight ratio calculated for
each vehicle class. This meant that once the curb weight for a PEV was estimated, a simple
linear calculation determined the peak motor power needed to meet the target power-to-weight
ratio. The battery power was then estimated as 15 percent greater than the peak motor power, to
account for losses in the motor. As with battery capacity, motor and battery power both interact
with battery and vehicle weight, and the calculation must be performed iteratively in the
spreadsheet as part of the overall battery sizing process.
Updates to Battery Sizing Assumptions and Methodology for the Proposed Determination
Analysis
As discussed in the Draft TAR, in the time since the 2012 FRM, the emergence of a variety of
production PEVs provided an opportunity to validate the assumptions and methods of the 2012
FRM analysis. The Draft TAR analysis therefore incorporated a large number of changes to the
methods and input assumptions for assigning battery capacity, battery power, motor power, and
other aspects of the PEV modeling problem. The major changes in going from the 2012 FRM to
the Draft TAR analysis included improvements to weight estimation for battery and non-battery
components, improvements to the assignment of electric drive motor power, increases in usable
battery capacity and electric drive efficiency, refinements to battery power ratings, variation of
range derating factors, and changes to certain PHEV powertrain configurations. These changes
were described in detail in the Draft TAR. Readers interested in the details of these changes may
refer to the Draft TAR. Because many of the resulting changes were retained for this Proposed
Determination analysis, the Draft TAR may also be useful in understanding the rationale behind
many of the decisions regarding inputs and assumptions applicable to this Proposed
Determination battery analysis.
The following sections detail the updates in methods and assumptions that EPA made for this
Proposed Determination analysis.
The public comment period on the Draft TAR elicited a number of comments regarding the
Draft TAR battery analysis methodology and assumptions. Where applicable, EPA has refined
and updated the methodology and assumptions as suggested by these comments and by updated
information that became available since the Draft TAR analysis was developed.
2-368
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In general, one message of the various public comments relating to the battery analysis
suggested that the projected battery sizing and battery costs per kWh were too high (i.e. too
conservative) compared to a growing industry consensus. For example, as previously described,
Tesla Motors suggested that projected battery sizing for a BEV200 was larger than necessary.
Other commenters suggested that some of the cost per kWh projections in the Draft TAR
appeared to be higher than more recent estimates from other sources. The bulk of comments
relating to battery costs were qualitative in nature and did not provide specific new information
that had not been available to EPA in developing the estimates. However, when taken in
conjunction with trends EPA has continued to observe in third-party projections of future battery
costs (from continued monitoring of the industry since the publication of the Draft TAR), EPA
believed that it would be valuable to reexamine the battery sizing and costing estimates.
Based on regular attendance at technical conferences during 2016 and particularly after
completion of the Draft TAR, EPA has become increasingly aware of examples of formal and
informal industry battery cost projections that parallel or even undercut the projected cost per
kWh for BEV batteries projected in the Draft TAR for the 2020 time frame and beyond. For
example, Ford has been reported as estimating its future battery system costs at $120 per kWh by
2020 and as low as $85 per kWh by 2030;610 an expectation of about $100 per kWh at the cell
level by 2020 was related verbally during a talk by a Ford representative at the 2016 Battery
Show;611 while at the same show, Berenberg Bank predicted $170 per kWh at the pack level by
2020612 and a presentation by Bloomberg New Energy Finance included scenarios by which
$155 per kWh might be approached by 2020.These examples as well as the frequency with
which such examples are being encountered reinforced the conclusion that battery costs are
continuing to change rapidly, and that EPA should therefore update its battery cost projections
for the Proposed Determination analysis.
Updated information on the 2017 Chevy Bolt suggested another update to the analysis. After
EPA certification of the Chevy Bolt in late 2016, EPA considered the derating factor that was
used in the certification process to compute the label range from the laboratory test results.
Certification data suggested that this vehicle utilized the default 70 percent derating factor, rather
than a higher custom factor as the Draft TAR analysis had assumed would apply to BEV200.
EPA used this updated information to update the derating factor assumed for BEV200 in this
analysis to a lower figure.
EPA also added to its compilation of MY2012-2016 BEVs and PHEVs several new models
that were released or certified after completion of the Draft TAR. This had small effects on some
comparative charts and the motor power estimation formula that was used to specify traction
motor peak power ratings for PEVs.
EPA also considered updated information regarding maximum battery cell capacities being
used in some production vehicles, and updated information regarding certification practices for
PHEVs that may affect design of the battery capacity for a given range.
As part of the effort to address other comments received on the EPA GHG analysis in general,
EPA also refined the six LPM class definitions. This resulted in changes to the target curb
www Bloomberg declined to include this presentation in the conference proceedings.
2-369
-------
Technology Cost, Effectiveness, and Lead Time Assessment
weights and power-to-weight ratios for each modeled xEV as compared to those used in the
Draft TAR, which in turn has some effect on projected costs.
Specific Updates to Inputs and Assumptions for this Proposed Determination Battery
Analysis
Several updates were motivated in part by public comments suggesting that projected battery
costs were too conservative in light of recent industry estimates. In the Draft TAR, EPA
compared the projected cost per kWh for BEV200 battery packs to other sources such as the
Nykvist & Nilsson study and the GM/LG cost announcement. In so doing, EPA recognized that
the Draft TAR cost projections may be somewhat conservative, as would befit projections made
in the face of future uncertainty. EPA also recognized that projections of battery capacity for a
given vehicle weight and range target were in many cases somewhat larger (i.e. conservative)
than seen in some production vehicles. At the time, it was felt that a somewhat conservative
estimate for both would be appropriate given the uncertainties associated with future cost
estimation.
Several commenters argued that battery costs have fallen at a faster rate than anticipated, and
would continue to fall to perhaps below the levels projected in the Draft TAR. Tesla Motors also
referred to current and future vehicles that are anticipated to have lower cost per kWh and/or
smaller packs for a given range target. Although the comments did not provide detailed data such
as evidence of actual pack costs for specific vehicles or types of vehicles, these comments
suggested that the conservative nature of the existing projections should be re-examined, as the
effect might be magnified by the projection of larger pack capacities than necessary.
EPA is committed to maintaining the accuracy of battery cost projections as much as
available information allows. This Proposed Determination analysis therefore makes several
updates to the battery sizing and costing analysis with the primary goal of refining and updating
projected battery sizing and cost. These included the following primary updates:
(a) Improved Basis for PEV Energy Consumption Estimates
In September 2016, EPA delivered a presentation at The Battery Show 2016613 describing the
battery analysis presented in the Draft TAR. The presentation acknowledged that, by some
measures, the battery sizes projected in this analysis were larger than those seen in some
production vehicles of a similar weight and driving range (i.e. conservative). The presentation
concluded that the gap might be narrowed by improving the method by which energy
consumption of the modeled PEVs was predicted. However, due to the need for compatibility
with other analyses that the battery cost model feeds into, only limited options were available for
improving the energy consumption estimates.
As a first step in this direction, EPA chose to use the most recent version of the EPA Trends
Report to derive the polynomial regression for fuel economy-to-ETW that formed the basis for
PEV energy consumption estimates. Adopting an updated Trends dataset serves to empirically
account for improved efficiencies and road load characteristics present in today's baseline fleet
and bring them into the battery sizing analysis. This was expected to reduce the estimated base
energy consumption compared to the old method. (As described later, application of road load
technologies to this base energy consumption was also adjusted to reflect technology already
2-370
-------
Technology Cost, Effectiveness, and Lead Time Assessment
present in the fleet. Even after this adjustment, the updated polynomial regression resulted in
improved estimates compared to the Draft TAR).
Equation 14 shows the updated coefficients that were used in the polynomial regression
equation in this Proposed Determination analysis.
Equation 14. MY2015 conventional LDV fuel economy regression formula used in Proposed Determination
FEconv(mi/gal) = 0.0000005308 X ETWpev2 - 0.0122335420 X ETWpev + 73.4948
As another step, EPA updated the version of the LPM that was used in the battery analysis
spreadsheets for estimating the road load reduction resulting from the 15 percent application of
aerodynamic and rolling resistance technology. The 2016 version of the LPM includes
significant refinement and calibration compared to the older version used in the Draft TAR
battery analysis, and was expected to result in more accurate energy consumption estimates.
EPA also modified the method for estimating the road load effect of net curb weight
reduction. Previously, changes in curb weight were converted to a road load reduction via the
LPM. In the revised analysis, the LPM no longer serves in this role, but instead, the reduced curb
weights generated by a given application of mass reduction are converted to an energy
consumption effect by simply feeding them directly to the FE-to-ETW polynomial regression
formula. This represents a more empirical approach to converting weight deltas to fuel economy
improvements.
EPA also made an effort to further optimize the various powertrain efficiency conversion
factors by which fuel economy estimates (generated by the polynomial regression formula, in
mi/gal) were converted to an electrified energy consumption estimate (in Wh/mi). This process
was guided by engineering judgement regarding expected electrified component efficiencies
present or anticipated for the 2022 to 2025 time frame, and validated by careful analysis of the
resulting projected battery capacities for a given range target and curb weight. Ultimately, the
selected efficiencies were seen to result in battery capacity projections that closely parallel the
capacities seen in recent production xEVs of the same weight and range. For more information
on the specific values used, please see the EPA Battery Analysis (Proposed Determination)
spreadsheets which are available in Docket EPA-HQ-OAR-2015-0827.
(b) Accounting for Road Load Reduction Technology Already Present in the Fleet
Several commenters to the Draft TAR (in the context of the greater GHG analysis rather than
the battery analysis) pointed out that a certain amount of mass reduction, aerodynamic drag
reduction and rolling resistance reduction are likely present in the baseline fleet and should be
accounted for in establishing the remaining amount that may be applied. This would affect the
battery analysis in that varying amounts of mass reduction are applied to xEVs, up to a cap of 20
percent total mass reduction. The analysis also applies a 20 percent reduction in aerodynamic
drag and rolling resistance. If some amount of these technologies are already present in the fleet
from which target curb weights and energy consumption estimates are derived, then the
maximum allowable application should be modified in order not to exceed the intended levels.
EPA modified the curb weight inputs to the battery analysis to assume that approximately 2
percent mass reduction is already present in the MY2015 baseline fleet from which input curb
weight targets are derived. This was based on an informal analysis of assumed weight reductions
2-371
-------
Technology Cost, Effectiveness, and Lead Time Assessment
for individual vehicles in the MY2015 baseline, which averaged approximately 2 percent. The 2
percent mass reduction assumed to be present was then added back to the glider weight. This
corrected weight was taken to be null, and used as the target curb weight. The analysis was then
allowed to apply up to 20 percent total mass reduction, which would now include the 2 percent
present in the fleet.
A similar adjustment was performed to account for aerodynamic drag and rolling resistance
technologies. In the construction of technology packages for the OMEGA analysis, BEV and
PHEV technology packages include an aerodynamic drag reduction of 20 percent (the
technology case known as AER02), and a tire rolling resistance reduction of 20 percent (the case
known as LRRT2). This is based in part on the expectation that manufacturers will find these
technology improvements to be highly cost effective for plug-in vehicles due to the potential to
reduce the size and cost of the battery. The package costs thus are meant to include the cost of
application of AER02 and LRRT2 relative the 2008 baseline.
In the Draft TAR, the EPA battery analysis did not account for aerodynamic or rolling
resistance technology already in the fleet, because the polynomial fuel economy regression was
based on MY2008 Trends data, which by definition represents the null technology case. In
updating the Trends energy consumption baseline to the 2015 Trend Report for this Proposed
Determination analysis (as described under item (a)), it became more important to account for
technology already in the Trends sample fleet. For this Proposed Determination analysis, EPA
assumed that a 5 percent improvement in aerodynamic drag and rolling resistance was already
present in the 2015 Trends Report baseline fleet. An additional 15 percent improvement was
applied via the LPM.
(c) Updated Baseline Curb Weights and Vehicle Classes
Another factor that influenced battery costs and sizing was the EPA decision to redefine the
definitions of the LPM classes. In the Draft TAR, xEVs were modeled for each of six LPM
classes, which were defined roughly by vehicle size, including Small Car, Standard Car, Large
Car, Small MPV, Large MPV, and Truck. For this Proposed Determination analysis, EPA
redefined the LPM classes. More information on this update is described in Section 2.3.1.4.
Accordingly, target curb weights in this TSD battery analysis are now derived from the
MY2015 baseline fleet (with PEVs removed) and aggregated into six distinct weight classes
numbered 1 through 6. This means that the modeled xEVs of this Proposed Determination have
significant differences in curb weight targets as compared to the now-defunct classes of the Draft
TAR. As a result, figures computed for the Draft TAR are not directly comparable to those of
this Proposed Determination analysis. To improve comparability, differences in projected cost
between the Draft TAR and Proposed Determination analyses are now reported as an average
across all of the LPM classes. The vehicle classes and curb weights previously used in the Draft
TAR analysis are contrasted with those used in this Proposed Determination analysis in Table
2.107.
Table 2.107 Changes to Baseline Curb Weights from Draft TAR to Proposed Determination
Draft TAR
Proposed Determination
Vehicle Class
Curb weight (lb)
Vehicle Class
Curb weight (lb)
Small Car
2628
Wt Class 1
2868
Standard car
3296
Wt Class 2
3340
2-372
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Large car
4117
Wt Class 3
3613
Small MPV
3500
Wt Class 4
4062
Large MPV
4448
Wt Class 5
4902
Truck
5161
Wt Class 6
4911
(d) Changes to Power-to-Weight Ratios Resulting from LPM Classes
As a result of the changes to class definitions, the power-to-weight ratios for each of the xEV
classes also changed. Although all xEVs continue to be modeled with acceleration capability
comparable to the baseline average for each new LPM class, the elimination of the Large Car
class (which previously had a relatively high power) means that the performance targets of the
new classes define a somewhat narrower spectrum than before.
(e) Changes to ICE Weight Deletes Resulting from LPM Classes
The updated LPM class definitions also required modifications to the ICE powertrain weights
("weight delete," or Wice powertrain ) assumed to be deleted from baseline vehicles in order to
become a BEV. Weight deletes used in this Proposed Determination analysis were scaled from
those used in the Draft TAR by performing a regression of the Draft TAR values with respect to
the vehicle power levels they had been associated with, and mapped to the new power levels of
the new LPM classes. Although the specific weight deletes for each class have therefore
changed, the average weight delete as a percentage of curb weight across all classes was virtually
unchanged from that of the Draft TAR. The new values for weight deletes are shown in Table
2.108.
Table 2.108 Baseline ICE-Powertrain Weight Assumptions (Pounds), By Vehicle Class
Class
Engine
Transmission*
Fuel system*
Engine mounts*
Exhaust
12V batteryt
Total
Wt Class 1
273
141
56
25
22
28
545
Wt Class 2
316
153
62
25
25
31
613
Wt Class 3
335
159
66
25
26
33
643
Wt Class 4
388
174
74
25
30
37
729
Wt Class 5
439
189
82
25
33
41
810
Wt Class 6
456
193
85
25
35
43
837
Note:
*Transmission minus differential; fuel system 50% fill; engine mounts include NVH treatments.
"f Although current BEVs retain a relatively small lead-acid 12V battery, this analysis, as did the Draft TAR analysis,
deletes the ICE-sized battery and assumes that an improved solution by 2025 will have a relatively negligible weight
compared to the other deleted components. Chapter 2.2.4.3.2 (Power Electronics) includes a discussion of drivers
and trends toward improving the low-voltage battery in BEVs.
(f) Update to Maximum Cell Capacities
EPA also updated the maximum cell capacities for PEV battery packs. Based on the recent
announcement and continued use of a 94 Ampere-hour cell in the BMW i3 BEV and Rex
(PHEV), EPA became more confident of the potential for such large capacity cells to be used in
future BEVs and longer-range PHEVs. EPA therefore increased the cell capacity limit for
modeled BEV packs to about 90 A-hr level (formerly 75 A-hr in the Draft TAR), and increased
the limit for PHEVs to about 60 A-hr (formerly 50 A-hr). Further, the limit was imposed as a
maximum, rather than a preferred target, meaning that cell sizes now approach the maximum
2-373
-------
Technology Cost, Effectiveness, and Lead Time Assessment
limit from below, rather than being scattered above and below the target. On average this results
in somewhat larger cell capacities and fewer cells per pack, which in some cases results in
somewhat lower pack costs for a given pack capacity.
(g) Update to Derate Factor for BEV200
For certification purposes, to convert a two-cycle range test result to a label value, EPA
allows manufacturers to either use a default derating factor of 70 percent or to derive a custom
derating factor by undergoing complete five-cycle testing. EPA certification data for 2012-
2016MY BEVs indicates that most BEV manufacturers have chosen to apply the default 70
percent derating factor in their certification tests. Tesla Motors is the only BEV manufacturer
that has elected to derive a custom derating factor. Tesla has used a factor of 79.6 percent for the
standard Model S configurations from 60 kWh to 90 kWh, and a factor ranging from 73 to 76
percent for higher-performance and AWD configurations of the Model S and Model X.xxx
The Draft TAR battery analysis therefore had adopted a derate factor of 80 percent for
BEV200, on the basis that Tesla was using a factor of 79.6 percent for the base Model S.
Because manufacturers of BEV75 and BEVlOO-type vehicles have only used the default 70
percent derating factor and have not derived custom factors, EPA had retained the 70 percent
derating factor for BEV75 and BEV100.
In the Draft TAR it was acknowledged that the appropriateness of an 80 percent derate factor
in modeling the label range of future BEV200s would depend on the degree to which
manufacturers are able to derive a custom derating factor similar to that used for certification of
the base Tesla Model S. Since publication of the Draft TAR, the 2017 Chevy Bolt BEV
completed EPA certification. Certification data indicates that this vehicle utilized a 70 percent
(apparently default) derate factor in computing its certified 238-mile label range. Also, further
certifications of Tesla vehicles including some variations of the Model S have continued to use
lower derating factors of about 73 to 76 percent rather than the 79.6 percent of the base Model S.
These developments led us to reconsider the Draft TAR expectation that future BEV200s
would commonly certify with an 80 percent derate factor. For this analysis, we therefore have
reduced the assumed derate factor for BEV200 from 80 percent to 75 percent, similar to the
factors used in recent Tesla certifications. EPA continues to believe that, as manufacturing
volumes and the number of BEV models both increase, there remains a potential for
manufacturers to justify the derivation of custom derate factors during the certification process,
and that in many cases this may result in a derate factor greater than the default 70 percent.
(h) Update of Motor Power Sizing Equation By Addition of MY2017 Vehicles
Several xEV models that have entered the market since the completion of the Draft TAR have
been added to the empirical study by which electric motor power and acceleration characteristics
are assigned. These included the 2017 Chevy Bolt and the BMW i3 94 Ah. This resulted in a
small change to the empirical equation for motor sizing. The change is small because the curb
weight and acceleration levels of these vehicles fell very close to the curve developed for the
xxx As indicated by the ratio of adjusted (Guide) combined fuel economy to unadjusted combined fuel economy
reported in columns M and P of the 'EVs' tab of the 2016 Fuel Economy Guide datafile, available in the Docket
and at https://www.fueleconomy.gov/feg/download.shtml
2-374
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Draft TAR equation. The updated equation used in this Proposed Determination is shown in
Equation 15 below. Development of this equation is described in more detail in Section 2.2.4.3.6
(Relating Power to Acceleration Performance).
Equation 15. Empirical equation for 0-60 all-electric acceleration time of MY2012-2017 PEVs
/ kW \-°733
t - 1.1321 ETWJ
(i) Adjustment to Usable Battery Capacity of PHEVs to Account for Range Degradation
As the Draft TAR noted, the possibility of PHEV range degradation over the life of the
vehicle is important to regulators and PHEV manufacturers, because range degradation would
gradually change the utility factor, which is a factor in certification and GHG compliance. The
Draft TAR analysis did not include an explicit oversizing factor for PHEVs. This Proposed
Determination analysis adds a 15 percent oversizing factor to the usable capacity of PHEV
batteries, by defining two usable SOC design windows, a smaller window applicable to
beginning of life (BOL) and a larger window applicable at end of life (EOL). This is meant to
capture practices to manage range degradation, which might include certifying with an aged
battery, modification of usable SOC over the life of the vehicle, or limiting the usable SOC at
time of certification. PHEV20 vehicles were assigned a BOL usable SOC window of
approximately 65 percent and an EOL window of 75 percent. PHEV40 was assigned a BOL
window of 67 percent and an EOL window of 77 percent. These figures were chosen by
engineering judgement and by considering their effect on the ability of the sizing method to
predict battery capacities of production PHEVs of a given range and curb weight.
Summary of Changes to Battery Sizing Assumptions
Table 2.109 reviews the major input assumptions to the EPA battery sizing method, and the
changes that were made for this Proposed Determination analysis.
Table 2.109 PEV Battery Sizing Assumptions and Changes from Draft TAR to Proposed Determination
Assumption
Draft TAR
Proposed Determination
WC1 curb weight (Small Car in TAR)
26281b
2868 lb
WC2 curb weight (Std car in TAR)
3296 lb
3340 lb
WC3 curb weight (Lg car in TAR)
4117 lb
3613 lb
WC4 curb weight (SmMPV in TAR)
3500 lb
4062 lb
WC5 curb weight (LgMPV in TAR)
4448 lb
4902 lb
WC6 curb weight (Truck in TAR)
51611b
4911 lb
Applied aero reduction from 2008
baseline
20%
unchanged
Applied tire reduction from 2008
baseline
20%
unchanged
Applied mass reduction to glider from
2008 baseline
Varies; max 20%
unchanged
Short range BEV (mi)
BEV75
unchanged
Mid-range BEV (mi)
BEV100
unchanged
2-375
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Long range BEV (mi)
BEV200
unchanged
Short range PHEV (mi)
PHEV20
unchanged
Long range PHEV (mi)
PHEV40
unchanged
Usable battery capacity, HEV
40%
unchanged
Usable battery capacity, PHEV20
70%
65.2%
Usable battery capacity, PHEV40
75%
67%
Usable battery capacity, BEV75
85%
unchanged
Usable battery capacity, BEV100
85%
unchanged
Usable battery capacity, BEV150/200
90%
unchanged
Battery specific energy
computed by BatPaC
unchanged
Non-battery specific power
1.4 kW/kg
unchanged
Motor sizing
Based on MY2014 baseline 0-60
performance estimate and new
empirical equation for PEVs
Updated to include MY2017
examples
Brake efficiency, PEV
87%
varies
Driveline efficiency, BEV
95%
varies
Cycle efficiency, PEV
97%
varies
BEV battery power as fn of motor power
l.lx
unchanged
PHEV battery power as fn of motor
power
l.lx
unchanged
Allowance for power fade
20%
unchanged
Road loads, PEV
from LPM
from LPM and Trends
2-cycle to 5-cycle derating factor, PHEV
and BEV75/100
70%
unchanged
2-cycle to 5-cycle derating factor,
BEV200
80%
75%
PHEV20 motor sizing basis
blended
unchanged
Analysis of Changes
The changes described above resulted in changes to the projected sizing of PEV batteries and
motors compared to those of the Draft TAR. Table 2.110 shows examples of the battery
capacities and motor power ratings generated by the revised sizing methodology and compares
them to the corresponding estimates generated by the Draft TAR analysis.
Table 2.110 Example Changes in Projected PEV Battery Capacity and Motor Power, Draft TAR to Proposed
Determination (20% weight reduction case)
BEV75
BEV100
BEV200
PHEV20
PHEV40
Draft TAR
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Small Car
17.3
54.0
23.5
55.6
41.2
60.6
6.1
29.7
11.7
61.9
Standard Car
21.4
83.8
29.1
86.2
50.2
93.4
7.5
45.8
14.4
96.3
2-376
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Large Car
27.7
176.8
37.4
181.6
65.0
197.4
9.5
94.6
18.8
206.7
Small MPV
22.7
74.5
30.9
76.6
53.7
83.4
7.9
40.6
15.1
84.6
Large MPV
29.3
115.3
39.8
119.0
69.2
129.5
10.2
63.2
19.7
133.5
Truck
33.0
138.3
44.6
142.3
77.6
154.7
11.7
78.0
22.6
165.0
Proposed Determination
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Wt Class 1
16.4
66.0
22.0
65.9
37.9
65.5
6.2
32.7
12.4
65.1
Wt Class 2
17.8
87.4
23.9
87.3
41.4
86.7
6.8
43.3
13.7
86.1
Wt Class 3
18.7
96.7
25.1
96.6
43.6
96.0
7.2
47.9
14.5
95.3
Wt Class 4
20.3
124.2
27.3
124.0
47.6
123.3
7.9
61.4
16.2
122.3
Wt Class 5
23.9
149.0
32.4
148.7
56.8
151.2
9.5
73.8
19.8
146.7
Wt Class 6
23.9
158.0
32.5
157.7
58.6
156.8
9.5
78.2
20.0
155.6
Average change from Draft TAR
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
Battery
(kWh)
Motor
(kW)
All classes
-24.8%
6.0%
-20.5%
2.9%
-19.9%
-5.5%
-11.0%
-4.1%
-5.6%
-10.3%
Notes:
f Compares BEV200 (Draft TAR) to BEV150 (FRM)
f f Compares blended PHEV20 (Draft TAR) to EREV PHEV20 (FRM)
As shown by the selected examples in the following Tables, the pack-level specific energy
figures EPA uses in this TSD analysis vary significantly, ranging from about 150 to 188 Wh/kg
for BEV75 to BEV200 (assuming NMC622 cathode), to about 130 to 150 Wh/kg for PHEV40
(also NMC622), and about 115 to 125 Wh/kg for PHEV20 (assuming blended NMC/LMO
cathode).
Table 2.111 Examples of Pack-Level Specific Energy Calculated By BatPaC for Selected PEV Configurations
(0% WR)
BEV75
(NMC622-G)
BEV100
(NMC622-G)
BEV200
(NMC622-G)
PHEV20 (NMC75%/
LM025%-G)
PHEV40
(NMC622-G)
Wh/kg
P/E ratio
Wh/kg
P/E ratio
Wh/kg
P/E ratio
Wh/kg
P/E ratio
Wh/kg
P/E ratio
Wt Class 1
152.5
4.6
165.3
3.4
174.3
2.1
119.7
6.5
152.7
6.6
Wt Class 2
157.2
5.5
160.0
4.1
178.3
2.5
118.0
7.7
148.3
7.9
Wt Class 3
160.1
5.7
171.1
4.3
181.0
2.6
119.4
8.0
147.9
8.2
Wt Class 4
163.7
6.5
160.3
4.9
184.9
3.0
119.1
9.2
129.9
9.6
Wt Class 5
168.3
6.3
172.7
4.7
187.3
2.8
124.6
8.8
138.0
9.1
Wt Class 6
166.1
6.6
172.6
5.0
187.3
3.0
123.4
9.3
135.2
9.6
Table 2.112 Examples of Pack-Level Specific Energy Calculated By BatPaC for Selected PEV Configurations
(20% WR)
BEV75
BEV100
BEV200
PHEV20 (NMC75%/
PHEV40
(NMC622-G)
(NMC622-G)
(NMC622-G)
LM025%-G)
(NMC622-G)
2-377
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Wh/kg
P/E ratio
Wh/kg
P/E ratio
Wh/kg
P/E ratio
Wh/kg
P/E ratio
Wh/kg
P/E ratio
Wt Class 1
151.6
5.3
160.3
3.9
171.2
2.3
118.0
7.0
150.1
6.9
Wt Class 2
150.3
6.5
163.7
4.8
174.3
2.8
119.6
8.4
145.3
8.3
Wt Class 3
152.4
6.8
165.7
5.1
176.3
2.9
116.0
8.8
144.3
8.7
Wt Class 4
150.0
8.1
169.0
6.0
179.7
3.4
114.5
10.3
135.4
10.0
Wt Class 5
153.1
8.2
170.9
6.1
167.6
3.5
118.2
10.3
133.1
9.8
Wt Class 6
150.3
8.7
168.7
6.4
188.0
3.5
117.0
10.8
130.6
10.3
While these figures may appear very aggressive compared to batteries seen in 2012-2017MY
applications, it should be noted that the technology assumptions in BatPaC are forecasts for the
2020 time frame and EPA applies them to the year 2025. For comparison, in January 2016, GM
announced that the 60 kWh Chevy Bolt BEV pack weighs 435 kg, suggesting that this BEV200
pack has already achieved a specific energy of 138 Wh/kg today.614 The same specific energy
was already seen in the 85 kWh Tesla Model S as early as 2012.615 Similarly, the 18.4 kWh pack
of the 2016 Chevy Volt PHEV weighs 183 kg, suggesting this PHEV53 pack has achieved 101
Wh/kg today. As has occurred in the time since the FRM, the level of industry activity in battery
development suggests that similar advances are likely to continue through the 2022 to 2025 time
frame.
To compare the Draft TAR capacity projections to specific production vehicles, Table 2.113
and Table 2.114 show the projected battery capacities and assumed curb weights for each
electrified vehicle type and vehicle class at 0 percent and 20 percent nominal weight reduction,
respectively. These tables are useful for drawing comparisons of the projected battery capacities
to those of specific production BEVs and PHEVs. In the battery sizing analysis, differences in
energy consumption among the six vehicle classes is primarily derived from differences in
vehicle weight. Therefore matching a production vehicle's curb weight, range and battery
capacity to the values in these tables provides a fair comparison regardless of whether the
indicated classification or weight reduction case matches that of the vehicle.
Table 2.113. TSD Projected Battery Capacities and Assumed Curb Weights, 0% Nominal Weight Reduction
BEV75
(NMC622)
BEV100 (NMC622)
BEV200
(NMC622)
PHEV20
(25NMC/75LMO)
PHEV40 (NMC622)
Curb wt
(lb)
kWh
Curb wt
(lb)
kWh
Curb wt
(lb)
kWh
Curb wt
(lb)
kWh
Curb wt
(lb)
kWh
Wt Class 1
2868
18.6
2868
24.8
2868
41.0
2868
6.6
2868
12.9
Wt Class 2
3340
20.7
3340
27.6
3340
45.7
3340
7.4
3340
14.3
Wt Class 3
3613
22.1
3613
29.4
3613
48.7
3613
7.8
3613
15.3
Wt Class 4
4062
24.6
4062
32.9
4062
54.4
4062
8.8
4048
16.9
Wt Class 5
4902
30.8
4902
41.0
4902
67.9
4902
10.9
4902
21.3
Wt Class 6
4911
30.8
4911
41.1
4911
68.1
4911
11.0
4911
21.3
Table 2.114 TSD Projected Battery Capacities and Assumed Curb Weights, 20% Nominal Weight Reduction
BEV75
(NMC622)
BEV100
(NMC622)
BEV200 (NMC622)
PHEV20
(25NMC/75LMO)
PHEV40 (NMC622)
Curb wt
(lb)
kWh
Curb wt
(lb)
kWh
Curb wt
(lb)
kWh
Curb wt
(lb)
kWh
Curb wt
(lb)
kWh
2-378
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Wt Class 1
2295
16.4
2322
22.0
2506
37.9
2571
6.2
2688
12.4
Wt Class 2
2672
17.8
2703
23.9
2903
41.4
2987
6.8
3137
13.7
Wt Class 3
2891
18.7
2928
25.1
3138
43.6
3231
7.2
3391
14.5
Wt Class 4
3249
20.3
3292
27.3
3519
47.6
3644
7.9
3851
16.2
Wt Class 5
3934
23.9
4008
32.4
4325
56.8
4377
9.5
4643
19.8
Wt Class 6
3945
23.9
4018
32.5
4229
58.6
4399
9.5
4680
20.0
The reasonableness of the battery capacity projections may be assessed by comparing them to
production vehicles of a known curb weight and driving range. As one example, the 30 kWh
trim of the Nissan Leaf is certified for an EPA range of 107 miles at a curb weight of 1515 kg
(3340 lb). This curb weight happens to exactly match the 3340 lb projected curb weight of
BEV100 Wt Class 2 (Table 2.113). The projected battery capacity for this vehicle is 27.6 kWh.
While this figure is smaller than the 30 kWh capacity of the Leaf, it represents a vehicle with
only 100 miles range rather than 107 miles. Also it represents a vehicle with a 20 percent
reduction in aerodynamic drag and rolling resistance from a 2008 baseline vehicle. If the
production Leaf achieves less reduction than this, it may require a larger battery to achieve its
107 mile range.
A more accurate way to assess the ability of the battery analysis to predict the battery capacity
of the Leaf would be to use inputs that represent the actual range of the Leaf. Running the battery
sizing methodology with inputs of 107 miles for range and 3340 lb for curb weight results in a
prediction of 30.30 kWh, very close to the 30 kWh of the Leaf.
As another example, the Chevy Bolt EV was announced in 2016 as a BEV238 with a curb
weight of 3580 lb. Using these as inputs to the battery sizing method, and applying a derate
factor of 70 percent (which appears to have been applied to the Bolt for label certification), the
result is 61.6 kWh, very close to the 60 kWh of the Bolt. The usable capacity of the Bolt battery
remains an uncertainty at this time and represents one variable that could have an impact on the
result. While the method assumes a default of 90 percent for BEV200, revising it to 92 percent
results in a prediction of 60 kWh.
As a third example, the Tesla Model S P85D weighs 4963 lb and is certified to a label range
of 253 miles, using a derate factor of 73.8 percent. The result is 88.75 kWh, quite close to the 85
kWh of this vehicle. Again, the usable capacity of this vehicle is uncertain and could have an
impact on the result. A value of 93 percent would predict a capacity of 85 kWh. Also, the AWD
configuration of this vehicle is described as having improved efficiency over a single-motor
configuration, which might explain the ability of this vehicle to have a smaller battery capacity
than the efficiencies encoded into the model would assume.
In similar fashion, modeling the Tesla Model S 60 as a BEV210 at 4323 lb and 79.6 percent
derate factor (as certified) results in a prediction of 57.5 kWh, somewhat smaller than the 60
kWh actually provided. Changing the usable capacity to 87 percent, which is in line with various
informal estimates for this vehicle, yields 59.5 kWh. Similarly, modeling the Tesla Model S 85
at 265 miles, 4647 lb, and 79.6 percent derate factor yields an estimate of 84 kWh, very close to
the actual 85 kWh. Setting the usable capacity to 89 percent would result in a match to 85 kWh.
The 2016 Chevy Volt PHEV achieves an AER of 53 miles at a curb weight of 1607 kg (3543
lb). The Volt usable SOC has been estimated at about 76.1 percent. These inputs result in a
2-379
-------
Technology Cost, Effectiveness, and Lead Time Assessment
prediction of 20.7 kWh, which at first glance is substantially larger than the 18.4 kWh actually
provided. However, the method assumes a fairly generous 15 percent oversize factor, which
might be different from the factor used by the designers of the Volt. It is also uncertain whether
the 76.1 percent usable SOC is assessed at beginning-of-life or end-of-life. Using an oversize
factor of about 4 to 5 percent yields a much better match to the actual capacity.
The BMW i3 Rex achieves an AER of 72 miles at a weight of 2982 lb with a 22 kWh battery.
Press reports suggest that this vehicle utilizes 87 percent of its battery capacity. Using these
inputs results in a prediction of 21 kWh, somewhat smaller than the actual specification but quite
close.
The Ford Fusion Energi has a range of 21 miles at 3986 lb using a 7.2 kWh battery. These
inputs predict a capacity of 9.0 kWh, a conservative figure. Similarly the Hyundai Sonata PHEV
achieves 27 miles at 3810 lb with a 9.8 kWh battery; using these inputs the model would predict
a similarly conservative capacity at 11.1 kWh. Particularly for shorter-range PHEVs, it is
uncertain whether manufacturer-reported battery capacity figures represent a nameplate capacity
at time of manufacture or if the capacity is down-rated to account for actual usable capacity
during the life of the vehicle.
By these examples, it is clear that the methodology as revised for this Proposed Determination
has greatly improved in its ability to predict battery capacities for xEVs as compared to the
version used in the Draft TAR analysis.
It is not to be expected that a single modeling technique with a single set of assumptions will
faithfully reproduce the actual battery capacities of all production vehicles. Individual production
vehicles are likely to vary in the degree to which the input assumptions of the sizing
methodology match those present in the respective vehicles. There could be differences in
assumed powertrain efficiencies or differences in application of road load reducing technologies
(mass reduction, aerodynamic drag reduction, and rolling resistance reduction) between the
production vehicles and the modeled vehicles. For example, if xEV manufacturers are applying
more than the 20 percent reduction in aerodynamic drag and rolling resistance (from a baseline
vehicle) assumed in the analysis, or are applying more mass reduction, it could result in
substantially smaller battery capacity requirements. Also, the larger battery capacity of longer-
range BEVs may slightly improve their discharge efficiency relative to shorter range vehicles,
because discharge would take place at a lower C rate. Efficiency of regenerative braking might
also improve slightly for these vehicles.
Specific examples are valuable in understanding the accuracy of the method, but another
perspective can be gained by looking at results in aggregate over a larger population of
examples. This can be shown by normalizing the battery capacities of actual and projected
vehicles to the corresponding vehicle curb weights, as shown in Figure 2.119 and Figure 2.120.
Source data for these Figures are available in the Docket.616 These comparisons remove the
effect of weight differences and more clearly expresses the efficiency with which gross battery
capacity is converted to label range for a given vehicle weight.
In Figure 2.119 we compare the battery capacity per unit curb weight (kWh/kg CW) of
comparable production BEVs against that of the BEVs modeled in each of 2012 FRM, Draft
TAR, and Proposed Determination analyses. For the purpose of this plot, comparable BEVs are
defined as BEVs that were available as MY2016-17 vehicles. BEV200+ vehicles that certified
2-380
-------
Technology Cost, Effectiveness, and Lead Time Assessment
for range with a derate factor different from the 75 percent that EPA assumes in this analysis had
their range adjusted in the plot to represent what their range would have been had a 75 percent
factor been used. With the exception of the Chevy Bolt, these vehicles were Tesla vehicles all of
which certified with a derate factor greater than 70 percent.
It can be seen that the revised battery sizing methodology predicts battery capacities for BEVs
that follow the trend line established by MY2012-2017 BEVs much more closely than earlier
versions of the methodology that were used in the Draft TAR and 2012 FRM. It is also clear that
the battery sizing methodology as revised for this Proposed Determination analysis has
significantly improved its prediction of BEV battery capacity per unit curb weight compared to
the methodology used in either the 2012 FRM analysis or the Draft TAR analysis, both of which
generated notably conservative (too large) capacity estimates for BEVs.
0.045
0.04
0.035
0.03
GO
"d>
-Q
u
ap 0-025 -M » 1 1 1 TSD
in
u)
O
w
QD
0.01
0.005
0
X
*
]
1 J
/
/
/
/
Z 3
fel
jjp"
f^
!•
55 0 02 W .*1*^ 2012 FRM
q.
nni5 I Comparable BEVs
Draft TAR
50 100 150 200 250 300
EPA label range (mi)
Figure 2.119 Projected BEV Gross Battery Capacity per Unit Curb Weight Compared to Comparable BEVs
For PHEVs, Figure 2.120 performs this comparison for PHEVs. It can be seen that for PHEVs
as well, the revised methodology follows the trendline established by production vehicles quite
closely, as it did in the Draft TAR analysis, but at a slightly lower position compared to the
production vehicle trendline.
2-381
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2012 FRM
MY2012-17 PHEVs
EPA label range (mi)
Figure 2.120 Projected PHEV Gross Battery Capacity per Unit Curb Weight Compared To Comparable
PHEVs
Particularly for BEVs, the revised method has removed much of the previous tendency to
overestimate the gross battery capacity needed to provide a given range for a given curb weight.
The revised method creates trendlines for projected BEV and PHEV capacities that follow the
respective production-vehicle trendlines quite well, particularly at the BEV200 and PHEV40
points. At shorter range points, such as BEV75, BEV100, and PHEV20, the projected capacity
trendlines run slightly below the respective production-vehicle trendlines, indicating that the
methodology now projects capacities for these shorter-range vehicles that on average are
somewhat smaller than found in MY2012-17 production vehicles. This is consistent with the
possibility that shorter-range vehicles, which in the plots consist mostly of relatively low-
production examples from a wide variety of manufacturers, may tend to embody a smaller
degree of technology optimization than the higher-production examples from a smaller group of
relatively well established manufacturers (Tesla and Chevrolet) that dominate the longer range
points. In other words, the revised methodology places a slightly greater expectation of future
improvement on shorter range vehicles than on longer range vehicles. The fact that real
production examples exist that plot on the lower side of both of the projected trendlines (i.e.
there are already production examples that convert battery capacity to range more efficiently
than the methodology projects for 2025) suggests that the projections are not overly optimistic.
2.3.4.3.7.2 Battery Sizing Methodology for HEVs
HEV battery packs were sized using a simpler methodology described below. This method is
continued in the current analysis.
2-382
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Because there is no "all-electric range" requirement for HEVs, battery pack sizes are
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 relatively
insignificant. For these reasons, a more direct approach (rather than an iterative process) works
for battery sizing of HEVs.
In the Draft TAR analysis as well as this Proposed Determination analysis, HEV batteries
were scaled similarly to the 2010 Fusion Hybrid battery, based on a metric of nominal battery
energy per pound of equivalent test weight (ETW). Although the Fusion battery utilized a
nickel-metal hydride (Ni-MH) chemistry in contrast to the lithium-ion chemistries of the current
analysis, the energy window required for hybrid operation and thus gross battery sizing is
expected to be similar for either chemistry.
The Fusion Hybrid Ni-MH battery had an ETW ratio of 0.37 Wh/lb. The battery was
understood to utilize a 30 percent usable SOC window. The FRM analysis and the current
analysis assumes 40 percent for HEVs in the 2020 time frame. The rationale for this assumption
is outlined in more detail in Draft TAR Section 5.2.4.4.3. This results in a 25 percent reduction
of the energy capacity of the base Fusion battery, or a 0.28 Wh/lb ETW ratio. This value was
used to size strong HEV batteries for the analysis.
In comparing anecdotal data for HEVs, EPA assumed a slight weight increase of 4-5 percent
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.
2.3.4.3.7.3 ANL BatPaC Battery Design and Cost 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. Prior to the 2012
FRM, 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 BEV applications. This led to the development of a Li-ion battery
cost model, later named BatPaC.
A basic description of the battery cost model that formed the basis of BatPaC was published
in a peer-reviewed technical paper presented at EVS-24.617 ANL later extended the model to
include analysis of manufacturing costs for BEVs and HEVs as well has PHEVs.618 In early
2011, ANL issued a draft report detailing the methodology, inputs and outputs of their Battery
Performance and Cost (BatPaC) model.619 Soon after, EPA contracted a complete independent
peer-review of the BatPaC model and its inputs and results for HEV, PHEV and BEV
applications.620 ANL also provided EPA with an updated report documenting the BatPaC model
that fully addressed the issues raised within the peer review.621 ANL has continued to develop
the model on an ongoing basis, adding several new features and refinements to the latest
version.622 For this TSD analysis, EPA used Version 3.0 of BatPaC, which was provided to EPA
on December 17, 2015623 and is the same version used for the Draft TAR analysis.
BatPaC 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
2-383
-------
Technology Cost, Effectiveness, and Lead Time Assessment
cost of capital equipment, plant area, and labor for each manufacturing step. The design criteria
include detailed parameters such as power and energy storage capacity requirements, cathode
and anode chemistry, and the number of cells per module and modules per battery pack. The
model assumes use of a stiff-pouch, laminated multi-layer prismatic cell, and battery modules
consisting of double-seamed rigid containers. The model supports both liquid-cooling and air-
cooling, with appropriate accounting for the resultant structure, volume, cost, and heat rejection
capacity of the modules. 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.
EPA chose to adopt the ANL BatPaC model for the following reasons. First, BatPaC 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 EPA believes is the preferred method for developing cost
estimates. BatPaC appropriately considers the target power and energy requirements of the
vehicle, which are two of the fundamental parameters when designing a lithium-ion battery for
an HEV, PHEV, or BEV. BatPaC can estimate high volume production costs, which EPA
believes is appropriate for the 2025 time frame. Finally, its cost estimates are consistent with
some of the supplier cost estimates EPA received from large-format lithium-ion battery pack
manufacturers. A portion of that data was received from EPA on-site visits to vehicle
manufacturers and battery suppliers in 2008.
EPA has worked closely with ANL to test new versions of BatPaC and to guide the
development of features that would support the midterm review and this TSD analysis. ANL has
since published several iterations of the model that incorporate updated costs, improved costing
methods and other improvements.
EPA has also worked closely with ANL to evaluate each successive version of the BatPaC
model, to make suggestions for its improvement, and to specifically request features to assist
with its use for the purpose of battery costing for the rule. EPA also worked with ANL to
arrange for an independent peer review of the model in 2011. This peer review along with EPA
input led to many improvements that were described in the TSD that accompanied the 2012
FRM. ANL has continued to make improvements and add new features since the FRM, many at
EPA request. Recent development has included: support for additional battery module
topologies, improved modeling of impedance and electrode thickness, improved evaluation of
battery thermal capabilities, revised electrode chemistries such as NMC622, improved
accounting for plant costs and overhead, improved cost accounting for solvent recovery,
customization of cell thickness parameters, generation of US ABC parameters, and updated costs
for all constituent cell materials.
To conduct the Draft TAR analysis, in December 2015 ANL provided EPA with a beta copy
of BatPaC Version 3. After testing and evaluation, this version was used in the Draft TAR GHG
Assessment, and continues to be used for this Proposed Determination assessment. A copy of
this file is available in Docket EPA-HQ-OAR-2015-0827.
2-384
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Basic user inputs to BatPaC include performance goals (power and energy capacity), choice
of battery chemistry (of several predefined chemistries), the vehicle type for which the battery is
intended (HEV, PHEV, or BEV), the desired number of cells and modules and their layout in the
pack, and the volume of production. BatPaC then designs the electrodes, cells, modules, and
battery pack, and provides a complete, 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 able to edit these values 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 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.
The cost outputs used by EPA to determine 2025 HEV, PHEV and BEV battery costs were
based on the inputs and assumptions described in the next section. For engineering properties
and material costs, and for other parameters not identified below, EPA used the defaults provided
in the model.
2.3.4.3.7.4 Assumptions and Inputs to BatPaC
After considering applicable public comments and updated information, EPA chose basic user
inputs to BatPaC as follows.
For performance goals, EPA used the power and energy requirements derived from the
battery sizing analysis described in the previous section. Additional inputs include battery
chemistry, vehicle type (BEV, PHEV, or HEV), cell and module layout, and production
volumes, as outlined below.
In addition to these inputs, EPA monitored certain outputs to ensure that the resultant cell and
pack specifications were realistic. In particular, pack voltages, electrode dimensions, cooling
capability, and individual cell capacities were monitored to ensure that they were consistent with
current and anticipated industry practice.
Additionally, EPA did not include warranty costs computed by BatPaC in the total battery
cost because these are accounted for elsewhere in the analysis by means of indirect cost
multipliers (ICMs).
Battery chemistry
Chemistries were chosen due to their known characteristics and to be consistent with both
publicly available information on current and near term HEV, PHEV and BEV product offerings
from OEMs.
2-385
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In both the Draft TAR and this Proposed Determination analysis, EPA selected NMC622 for
BEV and PHEV40 packs, and a blended cathode (25 percent NMC and 75 percent LMO, the
BatPaC default value) for PHEV20 and HEV packs. As discussed in the Draft TAR, although
most current Li-Ion HEV packs are reported to be using NMC cathodes,624 EPA used a blended
cathode for HEV batteries because the default NMC formulations modeled by BatPaC do not
support the high power-to-energy ratios required by some of the modeled HEV configurations.
In August 2016, EPA coordinated with ANL for an update to the BatPaC NMC formulation that
would allow HEV packs to be constructed with NMC cathodes. ANL recommended certain
changes to input parameters that enabled higher power for these batteries. However, since the
costs of HEV packs with NMC cathodes generated by this technique did not differ significantly
from those with blended cathodes, EPA ultimately decided to continue using blended cathodes
for HEV batteries.
Pack topology and cell capacity
In the Draft TAR analysis, EPA optimized the pack topology for BEVs and PHEVs by
choosing values for cells per module and number of modules to target a preferred cell capacity.
This practice continues for this Proposed Determination analysis. Since the number of modules
per pack must be a whole number, varying the number of cells per module allows the number of
cells per pack and their capacities to be better targeted. In the Draft TAR, EPA varied the
number of cells per module to between 24 and 36. This was revised to between 20 and 36 for this
Proposed Determination analysis in order to better target pack voltages and maximum cell
capacities.
In public comments on the Draft TAR, Toyota stated: "As noted, the Draft TAR explains that
the ideal number of cells per module may vary depending on the capacity of the pack and the
size of the cells and it may be more appropriate to optimize the pack topology by varying the
number of cells per module in order to better match performance targets and minimize cost.
However, Toyota does not share this perspective as an increase in capacity does not necessarily
mean that the number of cells can be reduced. Reduction of cells number causes voltage decrease
and current increase. So the numbers of cells cannot be reduced by a certain degree."
To clarify, the EPA battery analysis does not reduce the number of cells per pack as battery
capacity increases. Pack voltages are always targeted to a range between about 300 to 400 volts.
As pack capacity increases, the number of cells may also increase, as might the capacity of each
cell and the number of parallel cells. In general, if a smaller number of cells per module is
specified in order to stay within cell capacity and pack voltage targets, there will be a larger
number of modules to compensate and voltage will therefore not decrease. Detailed information
regarding the specific topology of each of the 150 PEV battery packs modeled in the analysis are
contained in the EPA Battery Analysis (Proposed Determination) Spreadsheets which are
available in the Docket.
In the Draft TAR, EPA targeted an individual cell capacity of 60 A-hr for BEV packs (not to
exceed 75 A-hr) and 45 A-hr for PHEV packs (not to exceed 50 A-hr). This was based in part on
examples seen in the industry, such as the 55 Ampere-hour cells that appear to be used by Nissan
and GM in their recently announced 60-kWh packs, and larger cell sizes currently produced or
recently announced by leading suppliers.
2-386
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Since the Draft TAR was completed, at least one additional example of a significantly larger
cell size has appeared in a production BEV. The BMW i3 94 Ah uses a 94 Ampere-hour cell,
which is significantly larger than most other manufacturers have been using. Because this vehicle
has now entered the market and has effectively replaced the 60 A-hr version, it provides
additional evidence that cells of this capacity can be effective in a BEV application. Accordingly,
EPA has updated the limits on maximum cell capacity for this Proposed Determination analysis.
BEV cells are now allowed to reach a maximum of approximately 90 A-hr. In most cases, it is
only the longer range BEVs that approach this limit. For vehicles that approach the limit, the use
of these larger cells tends to reduce pack costs by tending toward smaller numbers of cells of a
larger capacity than might have been applicable in the Draft TAR analysis. HEV packs, which
consist of a single module, are configured with 32 cells as before.
Thermal management
In the Draft TAR analysis, all BEV, PHEV, and HEV packs were modeled with liquid cooling
as defined by the BatPaC model.
As before, BatPaC continues to provide an option for active air cooling in which individual
cells are separated by air passages through which cabin air or cooled air is circulated. Use of this
option results in package volumes that are much larger than for a liquid cooled pack. As
described in the Draft TAR, although passive air cooling continues to be prevalent in HEV packs
at the time of this writing, some industry sources have indicated that liquid cooling may also be
preferable for HEV packs in order to improve utilization of capacity and increase service life.
Minimization of under-hood package volume is also a growing concern. As in the Draft TAR
analysis, EPA therefore chose to utilize liquid cooling for HEV packs as well as for BEV and
PHEV packs for this Proposed Determination analysis.
Pack voltage
As in the Draft TAR analysis, for this Proposed Determination analysis, EPA limited BEV
and PHEV voltages to approximately 120V for HEVs (except 48V HEVs) and approximately
300-400V for BEVs and PHEVs.
In public comments on the Draft TAR, Toyota commented, "Toyota does not understand why
HEV voltage remains at 120V." In response, although it is true that some mild and strong HEVs
are currently targeting voltage ranges higher than 120V, the systems on which EPA based its
teardown-derived costs were systems that operated in the 120V range. There are likely to be
some advantages to operating at a higher voltage. However, increasing the voltage of a small
(approximately 1 kWh) pack to several hundred volts requires a relatively large number of
relatively small cells, at a potentially higher cost due to the larger number of cells and cell
connections. Going forward to the 2022 to 2025 time frame, it is unclear whether the advantage
of operating an HEV at a higher voltage will continue to outweigh the higher cost of the battery.
Toyota also noted, "The Draft TAR's assessment that the customary voltage range for a given
xEV category is an outgrowth of the relative size of the battery is incorrect. The voltage is not a
by-product but a crucial speciation necessary in determining the output of the battery." EPA has
clarified the text in the corresponding section.
Toyota also stated, "The Draft TAR provides examples of PHEVs and BEVs in the 600V
range ... However, Toyota finds that the increase in voltage does not always necessarily lead to
2-387
-------
Technology Cost, Effectiveness, and Lead Time Assessment
an increase in performance. Consequently, the cost of each of the components may increase, and
the efficiency sometimes degrades." EPA acknowledges the comment, and reiterates that in both
the Draft TAR and this Proposed Determination analyses, voltages in the modeled PEVs are
limited to between approximately 300V and 400V.
Electrode dimensions
For electrode coating thickness, the 100-micron maximum limit used in the Draft TAR
analysis is retained in this Proposed Determination analysis.
In the Draft TAR it was noted that recent developments in pack design (as described in Draft
TAR Section 5.2.4.4.6, Electrode Dimensions) suggest that the industry may be moving toward
low-profile or flat floor-mounted packs. For this reason, the Draft TAR analysis adopted the
BatPaC default aspect ratio of 3:1. This aspect ratio continues to be used in this Proposed
Determination analysis.
Manufacturing volumes
For this Proposed Determination analysis, the assumed manufacturing volume for BEV,
PHEV and HEV battery packs was retained at 450,000 per year as in the Draft TAR analysis.
For additional discussion of considerations with regard to the assumed manufacturing volume,
please refer to Chapter 2.2.4.5.7 (Pack Manufacturing Volumes) of this TSD.
In comments on the Draft TAR, Global Automakers commented on the role of production
volume in achieving economies of scale: "the agencies considered a volume of 450,000 units
necessary to achieve full economies of scale. In its 2015 study, the NRC noted that the
technology penetration levels projected by the agencies did not reach that level, and that no one
manufacturer would reach that level. In the TAR, the agencies respond that economies of scale
can be obtained at levels as low as 60,000, and put forward a number of other arguments on
battery costs. Nonetheless, it cannot be denied that at current sales levels of electric-drive
vehicles of less than one percent (1 percent) of the market (i.e., less than 17,000 vehicles),
manufacturers are not close to volumes that could provide economies of scale. Unless demand
for those vehicles increases dramatically, economies of scale will remain out of reach."
EPA acknowledges that electrified vehicle sales are not currently approaching the 450,000
units per year assumed in the Draft TAR analysis. However, evidence continues to grow that the
battery costs projected by the EPA analysis may be conservative nonetheless. As discussed in the
Draft TAR, the GM/LG cost disclosure provides some evidence that battery costs may already be
approaching the costs projected by the EPA analysis at volumes much lower than 450,000, given
that the disclosed battery costs for the Bolt are likely to be predicated on a much lower annual
production volume. When taken in light of other comments received on the Draft TAR that
characterize the projected costs as already conservative, reducing the production volume as an
input to BatPaC and thereby increasing projected costs would seem to be unwarranted. It also
would presuppose that electrified vehicle sales will fail to grow significantly in the future,
something which is not at all certain despite the low penetration levels projected for compliance
with this rule. It remains the position of some commenters that electric vehicle sales are poised
for significant growth for reasons that go well beyond regulatory influences. As Faraday Future
commented, "notwithstanding today's low gasoline prices, the number of electric vehicles on the
roads in the United States is going to climb, and climb steeply between now and 2022," and went
2-388
-------
Technology Cost, Effectiveness, and Lead Time Assessment
on to point out that "there are also many recent publications that analyze the factors above and
recent data to project EV penetration rates that far exceed those assumed in the Draft TAR." The
recent sales growth in Tesla vehicles, the large number of reservations for the Model 3, and
Tesla's plans for rapid expansion also suggest that at least some stakeholders are taking a strong
position that the 450,000 vehicles per year on which the EPA battery cost assumptions are
nominally based is an attainable outcome.
Summary of Battery Design Assumptions
Table 2.115 shows a summary of battery design assumptions used in the Draft TAR analysis
and those adopted for this Proposed Determination analysis.
Table 2.115 Battery Design Assumptions Input to BatPaC and Changes from Draft TAR to Proposed
Determination
Assumption
Draft TAR
Proposed Determination
BEV75 chemistry
NMC622-G
unchanged
BEV100 chemistry
NMC622-G
unchanged
BEV150/200 chemistry
NMC622-G
unchanged
PHEV20 chemistry
25%NMC/75%LMO-G
unchanged
PHEV40 chemistry
NMC622-G
unchanged
HEV chemistry
25%NMC/75%LMO-G
unchanged
Pack topology
optimized to target
preferred cell capacity
unchanged
Maximum cell capacity (A-hr)
BEV: target 60, max 75
PHEV: target 45, max 50
BEV: max 90
PHEV: max 60
Cells per module
24 to 32
unchanged
BEV thermal medium
Liquid
unchanged
PHEV thermal medium
Liquid
unchanged
HEV thermal medium
Liquid
unchanged
BEV pack voltage range (V)
300V to 400V
unchanged
PHEV pack voltage range (V)
300V to 400V
unchanged
HEV pack voltage range (v)
~120V
unchanged
Maximum electrode thickness (microns)
100
unchanged
Electrode aspect ratio
3:1
unchanged
BEV battery 2025 annual mfg volume
450,000
unchanged
PHEV battery 2025 annual mfg volume
450,000
unchanged
HEV battery 2025 annual mfg volume
450,000
unchanged
2.3.4.3.7.5 Battery Cost Projections for xEVs
In Table 2.117 through Table 2.122 we show the battery pack direct manufacturing costs
(DMC) that were generated by the EPA battery analysis workbooks601 for this Proposed
Determination. The average degree of change from cost generated for the Draft TAR is also
shown, for each level of applied mass reduction technology. The costs are quoted in 2015
2-389
-------
Technology Cost, Effectiveness, and Lead Time Assessment
dollars and the analysis assigns them to the year 2025 for BEVs and PHEVs and the year 2017
for HEVs. This assignment follows the convention used in previous analysis for the 2012 FRM
and Draft TAR, where HEV battery costs were assigned to the earlier year to reflect
considerations such as the relatively larger number of HEV batteries that were in production
relative to PHEV and BEV batteries.
As in the Draft TAR, the costs shown are BatPaC output figures minus warranty costs. The
warranty costs computed by BatPaC are subtracted because the EPA analysis accounts for
warranty costs by means of indirect cost multipliers (ICMs).
It is important to understand that the figures shown in Table 2.117 through Table 2.122 should
not necessarily be understood as predictions of future battery costs for any specific future
electrified vehicles. Rather, these figures are BatPaC outputs that serve as input data points for
the generation of cost curves that the OMEGA model uses to estimate battery costs for the
electrified vehicles generated by OMEGA for each year of the rule. Only the electrified vehicles
generated by OMEGA, and not the electrified vehicles modeled in the EPA battery analysis
workbooks to generate the input data points, figure into the compliance analysis. The vehicles
described in the battery analysis workbooks can, however, be useful to understand other
assumptions pertinent to the analysis, such as for example, the amount of battery capacity that is
estimated to be needed to provide a given driving range for a given curb weight, or the pack
topologies and cell sizes assumed to be applicable to these vehicles. It should be understood,
however, that the specific configurations modeled in the workbooks do not necessarily constitute
predictions of any specific future vehicles.
As mentioned above, one of the ways EPA uses these BatPaC workbook figures is to generate
learning curves that assign battery costs to each individual year over the full time frame of the
rule. This curve is developed by first considering the BatPaC costs as applicable to the 2025 MY
for BEVs and PHEVs and to the 2017 MY for HEVs. EPA then used this curve to "unlearn"
those costs back to the present year. This allows EPA to estimate costs applicable to MYs 2017
through 2025, which are reported in Table 2.126 through Table 2.131. The changes in direct
manufacturing costs from year-to-year therefore reflect cost changes due to learning effects.
Learning curves were developed as described in Chapter 2.3.2.1.4.
As shown in Table 2.116, projected battery pack costs for many electrified vehicle
configurations have fallen substantially from those projected in the Draft TAR analysis. These
changes are the result of many influences, but are primarily due to projection of smaller pack
capacities for a given range target, and larger cell capacities within each pack. The change in cost
per kWh is not as great because most of these changes have a stronger effect on total pack cost
rather than cost per kWh. In some cases, potential reductions in cost per kWh resulting from, for
example, larger cell sizes, were offset by other adjustments, such as oversizing of PHEV
batteries to account for range degradation.
Table 2.116 Average Change in Projected Battery Pack DMC from Draft TAR to Proposed Determination
Electrified
Vehicle Type
Average change
Change in
pack cost
Change in cost
per kWh
BEV75
-11.9%
+3.4%
BEV100
-13.6%
+1.6%
2-390
-------
Technology Cost, Effectiveness, and Lead Time Assessment
BEV200
-18.3%
+3.7%
PHEV40
-5.0%
-2.2%
PHEV20
-4.2%
+4.3%
HEV
+0.8%
-2.0%
The Proposed Determination battery costs are not directly comparable to those of the Draft
TAR analysis because of the change in LPM class definitions, which means that vehicles in each
of the six classes have different curb weights and power requirements. However, costs can be
compared on an average basis across all classes. Compared to the Draft TAR, costs for BEV75
and BEV100 have fallen by an average of about 12 to 14 percent on a total pack cost basis.
BEV200 pack costs fell by an average of about 18 percent, reflecting the larger net pack size
reductions for these larger, longer-range packs. On a cost per kWh basis, costs for BEVs rose by
about 1.5 to 3.5 percent (due largely to the increase in power-to-energy ratio resulting from
reductions in pack capacity while power requirements remained relatively unchanged). The
dominant factor in reduction of total pack costs for BEVs was the reduction in projected pack
capacities for a given range.
PHEV40 and PHEV20 battery pack costs have fallen by about 4 to 5 percent, having
benefited from forces similar to those that have reduced BEV costs. The cost reductions were not
as great as for BEVs because the updated Proposed Determination analysis imposes a battery
oversizing factor to account for PHEV range degradation.
HEV costs have remained similar to those of the Draft TAR. This is due to few if any
changes to the modeling methodology and assumptions applicable to HEVs. The primary cause
of any changes would be due to the change in LPM class definitions, which changed the curb
weight and power demand for each vehicle class.
It should be noted that BatPaC does not model passively air cooled HEV cell assemblies (that
is, without significant air flow passages between the cells). A passively cooled assembly without
integrated air passages would probably have a lower cost than other available options. However,
as modeled by BatPaC, liquid cooled HEV packs have a slightly lower cost than the available air
cooled options, and for this reason as well as the expectation that liquid cooling will enable
better capacity utilization, EPA chose liquid cooling for HEV packs
Table 2.117 Estimated Direct Manufacturing Costs in MY2025 for BEV75 Battery Packs
BEV75*
0% CWR
2% CWR
7.5% CWR
10% CWR
20% CWR
(450k/yr)
Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Small Car
$3,962
$203
$3,940
$205
$3,893
$208
$3,873
$210
$3,788
$219
Standard Car
$4,411
$184
$4,391
$186
$4,331
$189
$4,308
$190
$4,203
$196
Large Car
$5,807
$192
$5,752
$193
$5,603
$193
$5,538
$194
$5,404
$195
Small MPV
$4,514
$177
$4,489
$179
$4,431
$182
$4,406
$183
$4,301
$189
Large MPV
$5,380
$164
$5,351
$165
$5,278
$168
$5,248
$169
$5,121
$175
2-391
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Truck
$5,856
$157
$5,805
$158
$5,674
$159
$5,614
$159
$5,457
$165
Proposed Determination
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
$3,819
$205
$3,800
$207
$3,750
$211
$3,769
$216
$3,660
$223
Wt Class 2
$3,989
$193
$3,965
$195
$3,900
$200
$3,870
$202
$3,762
$211
Wt Class 3
$4,099
$186
$4,071
$188
$3,994
$193
$3,962
$195
$3,837
$205
Wt Class 4
$4,332
$176
$4,309
$178
$4,248
$186
$4,223
$189
$4,135
$204
Wt Class 5
$4,912
$160
$4,832
$161
$4,760
$171
$4,654
$173
$4,520
$189
Wt Class 6
$4,997
$162
$4,985
$166
$4,853
$174
$4,746
$176
$4,620
$193
Change from Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
-3.6%
1.0%
-3.6%
1.1%
-3.7%
1.4%
-2.7%
2.8%
-3.4%
1.7%
Wt Class 2
-9.6%
4.6%
-9.7%
4.8%
-10.0%
5.7%
-10.2%
6.2%
-10.5%
7.5%
Wt Class 3
-29.4%
-3.4%
-29.2%
-2.6%
-28.7%
-0.3%
-28.5%
0.9%
-29.0%
5.3%
Wt Class 4
-4.0%
-0.9%
-4.0%
-0.1%
-4.1%
2.3%
-4.2%
3.5%
-3.8%
7.9%
Wt Class 5
-8.7%
-2.8%
-9.7%
-2.4%
-9.8%
1.7%
-11.3%
2.0%
-11.7%
8.3%
Wt Class 6
-14.7%
2.9%
-14.1%
5.2%
-14.5%
9.3%
-15.5%
10.2%
-15.3%
16.9%
AVE CHANGE
-11.7%
0.2%
-11.7%
1.0%
-11.8%
3.4%
-12.0%
4.3%
-12.3%
7.9%
Note:
CWR = target percent reduction in vehicle curb weight.
Actual reduction will be less if it would require applying more than 20 percent mass reduction to glider.
*NMC622 cathode.
2-392
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.118 Estimated Direct Manufacturing Costs in MY2025 for BEV100 Battery Packs
BEV100*
(450k/yr)
0% CWR
2% CWR
7.5% CWR
10% CWR
20% CWR
Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Small Car
$4,533
$175
$4,511
$176
$4,450
$179
$4,428
$180
$4,345
$185
Standard Car
$5,306
$166
$5,278
$167
$5,207
$170
$5,179
$171
$5,095
$175
Large Car
$6,476
$161
$6,417
$161
$6,265
$162
$6,197
$162
$6,122
$164
Small MPV
$5,404
$159
$5,374
$160
$5,342
$164
$5,312
$165
$5,223
$169
Large MPV
$6,266
$144
$6,227
$144
$6,139
$147
$6,102
$148
$5,995
$151
Truck
$6,266
$135
$6,227
$135
$6,139
$137
$6,102
$138
$5,995
$142
Proposed Determination
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
$4,296
$173
$4,273
$175
$4,211
$178
$4,183
$180
$4,087
$185
Wt Class 2
$4,547
$165
$4,477
$165
$4,395
$169
$4,359
$171
$4,235
$177
Wt Class 3
$4,693
$159
$4,657
$161
$4,555
$165
$4,471
$165
$4,330
$172
Wt Class 4
$5,079
$155
$4,946
$154
$4,830
$159
$4,724
$159
$4,500
$165
Wt Class 5
$5,998
$146
$5,924
$148
$5,728
$154
$5,234
$146
$5,022
$155
Wt Class 6
$6,009
$146
$5,935
$148
$5,763
$155
$5,315
$148
$5,107
$157
Change from Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
-5.2%
-0.7%
-5.3%
-0.7%
-5.4%
-0.4%
-5.5%
-0.2%
-5.9%
0.4%
Wt Class 2
-14.3%
-0.9%
-15.2%
-1.6%
-15.6%
-0.9%
-15.8%
-0.5%
-16.9%
1.1%
Wt Class 3
-27.5%
-0.8%
-27.4%
-0.1%
-27.3%
1.7%
-27.8%
1.8%
-29.3%
5.4%
Wt Class 4
-6.0%
-3.0%
-8.0%
-4.2%
-9.6%
-3.5%
-11.1%
-4.0%
-13.9%
-2.4%
Wt Class 5
-4.3%
1.9%
-4.9%
2.8%
-6.7%
5.2%
-14.2%
-1.3%
-16.2%
2.9%
Wt Class 6
-9.9%
8.6%
-10.5%
9.6%
-11.8%
12.8%
-18.1%
6.8%
-19.3%
10.8%
AVE CHANGE
-11.2%
0.9%
-11.9%
1.0%
-12.7%
2.5%
-15.4%
0.4%
-16.9%
3.0%
Note:
CWR = target percent reduction in vehicle curb weight.
Actual reduction will be less if it would require applying more than 20 percent mass reduction to glider.
*NMC622 cathode.
2-393
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.119 Estimated Direct Manufacturing Costs in MY2025 for BEV200 Battery Packs
BEV200*
(450k/yr)
0% CWR
2% CWR
7.5% CWR
10% CWR
20% CWR
Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Small Car
$6,712
$156
$6,675
$157
$6,588
$160
$6,572
$161
$6,588
$160
Standard Car
$7,394
$140
$7,351
$141
$7,246
$143
$7,224
$144
$7,224
$144
Large Car
$8,851
$133
$8,797
$134
$8,743
$134
$8,743
$134
$8,743
$134
Small MPV
$7,734
$138
$7,688
$139
$7,555
$141
$7,555
$141
$7,555
$141
Large MPV
$9,160
$127
$9,101
$128
$8,966
$130
$8,966
$130
$8,966
$130
Truck
$9,795
$119
$9,732
$120
$9,579
$122
$9,515
$123
$9,515
$123
Proposed Determination
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
$5,932
$145
$5,896
$146
$5,802
$148
$5,760
$149
$5,716
$151
Wt Class 2
$6,255
$137
$6,208
$138
$6,083
$141
$6,027
$142
$5,963
$144
Wt Class 3
$6,460
$133
$6,407
$134
$6,261
$137
$6,196
$138
$6,119
$140
Wt Class 4
$6,846
$126
$6,776
$127
$6,589
$131
$6,507
$132
$6,403
$134
Wt Class 5
$7,828
$115
$7,717
$117
$7,477
$122
$7,187
$121
$7,051
$124
Wt Class 6
$7,841
$115
$7,730
$117
$7,434
$121
$7,310
$123
$7,005
$120
Change from Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
-11.6%
-7.5%
-11.7%
-7.5%
-11.9%
-7.4%
-12.4%
-7.0%
-13.2%
-5.8%
Wt Class 2
-15.4%
-2.3%
-15.6%
-2.1%
-16.0%
-1.6%
-16.6%
-0.9%
-17.5%
0.3%
Wt Class 3
-27.0%
-0.2%
-27.2%
0.1%
-28.4%
1.9%
-29.1%
3.0%
-30.0%
4.4%
Wt Class 4
-11.5%
-8.8%
-11.9%
-8.4%
-12.8%
-7.1%
-13.9%
-5.9%
-15.3%
-4.4%
Wt Class 5
-14.5%
-9.2%
-15.2%
-8.5%
-16.6%
-6.1%
-19.8%
-6.8%
-21.4%
-4.2%
Wt Class 6
-19.9%
-3.6%
-20.6%
-2.9%
-22.4%
-0.9%
-23.2%
0.0%
-26.4%
-2.4%
AVE CHANGE
-16.7%
-5.3%
-17.0%
-4.9%
-18.0%
-3.5%
-19.2%
-2.9%
-20.6%
-2.0%
Note:
CWR = target percent reduction in vehicle curb weight.
Actual reduction will be less if it would require applying more than 20 percent mass reduction to glider.
*NMC622 cathode.
2-394
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.120 Estimated Direct Manufacturing Costs in MY2025 for PHEV20 Battery Packs
PHEV20*
(450k/yr)
0% CWR
2% CWR
7.5% CWR
10% CWR
20% CWR
Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Small Car
$2,463
$382
$2,454
$385
$2,433
$394
$2,424
$397
$2,420
$399
Standard Car
$2,690
$340
$2,678
$342
$2,649
$349
$2,638
$352
$2,638
$352
Large Car
$3,157
$316
$3,136
$318
$3,080
$321
$3,070
$322
$3,070
$322
Small MPV
$2,737
$325
$2,727
$328
$2,699
$335
$2,688
$337
$2,683
$339
Large MPV
$3,025
$279
$3,008
$281
$2,962
$285
$2,942
$287
$2,937
$288
Truck
$3,190
$259
$3,169
$261
$3,115
$264
$3,103
$265
$3,103
$265
Proposed Determination
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
$2,448
$371
$2,439
$374
$2,415
$383
$2,404
$387
$2,403
$388
Wt Class 2
$2,589
$352
$2,576
$356
$2,490
$359
$2,478
$364
$2,476
$365
Wt Class 3
$2,643
$337
$2,629
$341
$2,601
$354
$2,591
$360
$2,589
$361
Wt Class 4
$2,791
$319
$2,779
$324
$2,749
$339
$2,737
$345
$2,735
$346
Wt Class 5
$3,025
$277
$3,006
$283
$2,958
$299
$2,937
$307
$2,932
$309
Wt Class 6
$3,017
$275
$2,998
$281
$2,950
$298
$2,930
$305
$2,927
$307
Change from Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
-0.6%
-3.0%
-0.6%
-3.0%
-0.7%
-2.7%
-0.8%
-2.5%
-0.7%
-2.8%
Wt Class 2
-3.8%
3.6%
-3.8%
4.0%
-6.0%
2.8%
-6.1%
3.4%
-6.1%
3.7%
Wt Class 3
-16.3%
6.7%
-16.2%
7.4%
-15.5%
10.0%
-15.6%
11.6%
-15.7%
12.0%
Wt Class 4
1.9%
-2.0%
1.9%
-1.2%
1.9%
1.2%
1.8%
2.4%
2.0%
2.2%
Wt Class 5
0.0%
-0.9%
-0.1%
0.5%
-0.1%
4.8%
-0.2%
6.9%
-0.2%
7.2%
Wt Class 6
-5.4%
6.2%
-5.4%
7.9%
-5.3%
12.7%
-5.6%
15.4%
-5.7%
15.9%
AVE CHANGE
-4.0%
1.7%
-4.0%
2.6%
-4.3%
4.8%
-4.4%
6.2%
-4.4%
6.4%
Note:
CWR = target percent reduction in vehicle curb weight.
Actual reduction will be less if it would require applying more than 20 percent mass reduction to glider.
*Blended LMO-NMC cathode.
2-395
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.121 Estimated Direct Manufacturing Costs in MY2025 for PHEV40 Battery Packs
PHEV40*
(450k/yr)
0% CWR
2% CWR
7.5% CWR
10% CWR
20% CWR
Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Small Car
$3,130
$260
$3,111
$262
$3,077
$264
$3,078
$264
$3,077
$264
Standard Car
$3,705
$251
$3,599
$246
$3,559
$247
$3,559
$247
$3,559
$247
Large Car
$5,528
$295
$5,550
$296
$5,552
$296
$5,550
$296
$5,552
$296
Small MPV
$3,661
$233
$3,635
$234
$3,579
$236
$3,579
$236
$3,579
$236
Large MPV
$4,620
$229
$4,622
$231
$4,574
$232
$4,574
$232
$4,574
$232
Truck
$5,073
$221
$5,026
$221
$4,999
$222
$4,999
$222
$4,999
$222
Proposed Determination
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
$3,223
$250
$3,215
$253
$3,198
$258
$3,198
$258
$3,198
$258
Wt Class 2
$3,468
$242
$3,457
$245
$3,438
$251
$3,438
$251
$3,438
$251
Wt Class 3
$3,614
$237
$3,601
$240
$3,579
$247
$3,579
$247
$3,579
$247
Wt Class 4
$3,935
$232
$3,991
$239
$3,973
$246
$3,973
$246
$3,973
$246
Wt Class 5
$4,837
$227
$4,814
$232
$4,375
$222
$4,432
$224
$4,778
$241
Wt Class 6
$4,936
$231
$4,914
$237
$4,378
$221
$4,543
$228
$4,887
$244
Change from Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
3.0%
-3.8%
3.3%
-3.3%
3.9%
-1.9%
3.9%
-1.9%
3.9%
-1.9%
Wt Class 2
-6.4%
-3.4%
-3.9%
-0.5%
-3.4%
1.9%
-3.4%
1.9%
-3.4%
1.9%
Wt Class 3
-34.6%
-19.7%
-35.1%
-18.9%
-35.5%
-16.6%
-35.5%
-16.5%
-35.5%
-16.6%
Wt Class 4
7.5%
-0.3%
9.8%
2.0%
11.0%
3.9%
11.0%
3.9%
11.0%
3.9%
Wt Class 5
4.7%
-0.6%
4.1%
0.4%
-4.3%
-4.4%
-3.1%
-3.2%
4.5%
4.0%
Wt Class 6
-2.7%
4.7%
-2.2%
6.8%
-12.4%
-0.3%
-9.1%
2.8%
-2.2%
10.1%
AVE CHANGE
-4.8%
-3.8%
-4.0%
-2.3%
-6.8%
-2.9%
-6.0%
-2.2%
-3.6%
0.2%
Note:
CWR = target percent reduction in vehicle curb weight.
Actual reduction will be less if it would require applying more than 20 percent mass reduction to glider.
*NMC622 cathode.
2-396
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.122 Estimated Direct Manufacturing Costs in MY2017 for strong HEV Battery Packs
STRONG HEV*
(450k/yr)
0% CWR
2% CWR
7.5% CWR
10% CWR
20% CWR
Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Small Car
$984
$ 1,216
$980
$ 1,236
$971
$ 1,297
$966
$1,326
$ 958
$ 1,383
Standard Car
$ 1,051
$ 1,057
$ 1,046
$ 1,074
$ 1,033
$ 1,123
$ 1,027
$1,148
$ 1,016
$ 1,198
Large Car
$ 1,197
$976
$ 1,188
$988
$ 1,168
$ 1,029
$ 1,158
$1,050
$ 1,140
$ 1,093
Small MPV
$ 1,033
$984
$ 1,029
$ 1,000
$ 1,017
$ 1,047
$ 1,011
$1,070
$ 1,001
$ 1,118
Large MPV
$ 1,123
$855
$ 1,117
$868
$ 1,100
$ 907
$ 1,093
$ 925
$ 1,078
$966
Truck
$ 1,194
$792
$ 1,187
$803
$ 1,167
$ 836
$ 1,158
$ 853
$ 1,142
$882
Proposed Determination
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
$1,001
$1,144
$998
$1,161
$986
$1,209
$982
$1,234
$972
$1,298
Wt Class 2
$1,046
$1,041
$1,042
$1,056
$1,030
$1,101
$1,025
$1,123
$1,012
$1,180
Wt Class 3
$1,069
$989
$1,063
$1,002
$1,050
$1,044
$1,044
$1,064
$1,030
$1,118
Wt Class 4
$1,122
$931
$1,116
$944
$1,101
$983
$1,094
$1,001
$1,077
$1,051
Wt Class 5
$1,182
$823
$1,176
$834
$1,157
$867
$1,149
$883
$1,127
$924
Wt Class 6
$1,196
$831
$1,189
$842
$1,170
$875
$1,162
$891
$1,144
$926
Change from Draft TAR
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Pack
$/kWh
Wt Class 1
1.8%
-5.9%
1.8%
-6.1%
1.6%
-6.7%
1.6%
-7.0%
1.4%
-6.2%
Wt Class 2
-0.4%
-1.5%
-0.4%
-1.7%
-0.3%
-2.0%
-0.2%
-2.2%
-0.4%
-1.5%
Wt Class 3
-10.7%
1.3%
-10.5%
1.3%
-10.1%
1.4%
-9.8%
1.4%
-9.6%
2.4%
Wt Class 4
8.6%
-5.4%
8.5%
-5.6%
8.3%
-6.1%
8.2%
-6.4%
7.7%
-6.0%
Wt Class 5
5.3%
-3.8%
5.3%
-4.0%
5.2%
-4.4%
5.2%
-4.6%
4.6%
-4.4%
Wt Class 6
0.2%
5.0%
0.2%
4.9%
0.3%
4.6%
0.3%
4.5%
0.2%
4.9%
AVE CHANGE
0.8%
-1.7%
0.8%
-1.8%
0.8%
-2.2%
0.9%
-2.4%
0.6%
-1.8%
Note:
CWR = target percent reduction in vehicle curb weight.
Actual reduction will be less if it would require applying more than 20 percent mass reduction to glider.
*Blended LMO-NMC cathode.
2-397
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.3.4.3.7.6 Discussion of Battery Cost Projections
In Draft TAR Section 5.2.4.4.9 (Evaluation of 2012 FRM Battery Cost Projections), EPA
reviewed the 2020-2022 cell4evel costs projected by GM for its LG-supplied cells for the Chevy
Bolt EV, and converted them to estimated pack4evel costs per gross kWh. These estimated
costs were shown to appear generally lower than the pack-level costs for BEV150 that were
generated by the 2012 FRM analysis. Figure 2.121 extends this comparison to the pack-level
costs for BEV200 projected by the Draft TAR analysis and this Proposed Determination
analysis. As discussed in the Draft TAR, the Draft TAR projected costs were significantly lower
than the costs projected in the 2012 FRM analysis. Further, the Proposed Determination figures
are somewhat lower still. These new figures continue to appear consistent with and in many
cases appear to remain conservative with respect to the trend established by the GM/LG pack-
converted cost estimates.
400
350
VY
T3
QJ
300
£250
to
c.
— 200
_c
5
150
Q_
o 100
50
0
*
T
[
i
O PD
X FRM
~ Draft TAR
-A—GM/LG low
-B— GM/LG high
2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026
Year
Figure 2.121 Comparison of Estimated Pack-Converted GM/LG Costs to BEV150/200 Projections of 2012
FRM, Draft TAR, and this Proposed Determination (PD)
As discussed in Draft TAR Section 5.2.4.4.9, comparisons of the GM/LG costs to those of the
EPA analyses are subject to some uncertainty. As discussed in the Draft TAR, comparison on
this basis to the 2012 FRM projections suggests that, rather than being overly optimistic, those
projections may have been quite conservative with respect to trends in battery cost that have
occurred since the FRM. This outcome suggests that EPA's battery costing methodology, with
the updates and refinements discussed previously, is an appropriate basis on which to derive
updated projections for this Proposed Determination analysis. As suggested throughout this
analysis, it should be noted that battery costs have many drivers, and future cost projections
derived by any methodology are subject to significant uncertainties.
2-398
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.3.4.3.7.7 Battery Pack Costs Used in OMEGA
Table 2.123 Linear Regressions of Strong Hybrid Battery System Direct Manufacturing Costs vs Net Mass
Reduction Applicable in MY2017 (2015$)
Curb Weight Class
Strong HEV
1
-$187x+$l,001
2
-$212x+$l,046
3
-$239x+$l,068
4
-$279x+$l,122
5
-$344x+$l,183
6
-$347x+$l,196
Note: "x" in the equations represents the net weight reduction as a percentage.
Table 2.124 Linear Regressions of Battery Electric Battery System Direct Manufacturing Costs vs Net Mass
Reduction Applicable in MY2025 (2015$)
Curb Weight Class
PHEV20
PHEV40
BEV75
BEV100
BEV200
1
-$441x+$2,448
-$403x+$3,223
-$761x+$3,820
-$l,098x+$4,295
-$l,711x+$5,931
2
-$l,152x+$2,591
-$499x+$3,468
-$l,136x+$3,987
-$l,564x+$4,523
-$2,244x+$6,253
3
-$504x+$2,641
-$565x+$3,614
-$l,313x+$4,096
-$ 1,95 lx+$4,691
-$2,606x+$6,459
4
-$535x+$2,790
$469x+$3,953
-$984x+$4,327
-$2,925x+$5,041
-$3,331x+$6,843
5
-$864x+$3,024
-$6,478x+$4,887
-$l,942x+$4,888
-$5,715x+$6,012
-$6,444x+$7,855
6
n/a
n/a
n/a
n/a
n/a
Note: "x" in the equations represents the net weight reduction as a percentage.
Table 2.125 Costs for MHEV48V Battery (dollar values in 2015$)
Curb Weight Class
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
All
DMC
$314
31
$314
$284
$265
$251
$240
$231
$223
$217
$211
All
IC
Highl
2024
$177
$175
$174
$173
$172
$171
$171
$170
$105
All
TC
$490
$459
$438
$423
$412
$402
$394
$387
$316
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.126 Costs for Strong Hybrid Batteries (dollar values in 2015$)
Curb
WRtech
WRnet
Cost
DMC: base
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
type
year cost
learning
Class
IC:
complexity
curve
IC: near
term
thru
1
10
6
DMC
$990
31
$990
$896
$835
$791
$756
$728
$705
$685
$667
1
15
11
DMC
$980
31
$980
$888
$827
$783
$749
$721
$698
$678
$661
1
20
16
DMC
$971
31
$971
$879
$819
$776
$742
$714
$691
$672
$655
2
10
6
DMC
$1,033
31
$1,033
$936
$872
$826
$790
$760
$736
$715
$697
2
15
11
DMC
$1,023
31
$1,023
$926
$863
$817
$781
$753
$728
$708
$690
2
20
16
DMC
$1,012
31
$1,012
$917
$854
$809
$773
$745
$721
$700
$682
3
10
5
DMC
$1,056
31
$1,056
$957
$891
$844
$807
$777
$752
$731
$712
3
15
10
DMC
$1,044
31
$1,044
$946
$881
$834
$798
$768
$744
$723
$704
3
20
15
DMC
$1,032
31
$1,032
$935
$871
$825
$789
$760
$735
$714
$696
4
10
6
DMC
$1,105
31
$1,105
$1,001
$933
$883
$844
$813
$787
$765
$745
4
15
11
DMC
$1,091
31
$1,091
$988
$921
$872
$834
$803
$777
$755
$736
4
20
16
DMC
$1,077
31
$1,077
$976
$909
$861
$823
$793
$767
$745
$726
5
10
6
DMC
$1,162
31
$1,162
$1,052
$981
$928
$888
$855
$827
$804
$783
5
15
11
DMC
$1,145
31
$1,145
$1,037
$966
$915
$875
$842
$815
$792
$772
5
20
16
DMC
$1,128
31
$1,128
$1,021
$952
$901
$862
$830
$803
$780
$760
6
10
6
DMC
$1,175
31
$1,175
$1,064
$992
$939
$898
$865
$837
$813
$792
6
15
11
DMC
$1,158
31
$1,158
$1,049
$977
$925
$885
$852
$825
$801
$781
2-399
-------
Technology Cost, Effectiveness, and Lead Time Assessment
6
20
16
DMC
$1,141
31
$1,141
$1,033
$963
$911
$872
$839
$812
$789
$769
1
10
6
IC
Highl
2024
$558
$552
$548
$545
$543
$541
$539
$538
$332
1
15
11
IC
Highl
2024
$553
$547
$543
$540
$538
$536
$534
$533
$328
1
20
16
IC
Highl
2024
$547
$541
$538
$535
$532
$531
$529
$528
$325
2
10
6
IC
Highl
2024
$582
$576
$572
$569
$567
$565
$563
$562
$346
2
15
11
IC
Highl
2024
$576
$570
$566
$563
$561
$559
$557
$556
$343
2
20
16
IC
Highl
2024
$571
$564
$560
$557
$555
$553
$552
$550
$339
3
10
5
IC
Highl
2024
$595
$589
$585
$582
$579
$577
$576
$574
$354
3
15
10
IC
Highl
2024
$589
$582
$578
$575
$573
$571
$569
$568
$350
3
20
15
IC
Highl
2024
$582
$576
$571
$568
$566
$564
$563
$561
$346
4
10
6
IC
Highl
2024
$623
$616
$612
$609
$606
$604
$602
$601
$370
4
15
11
IC
Highl
2024
$615
$608
$604
$601
$598
$596
$595
$593
$366
4
20
16
IC
Highl
2024
$607
$601
$596
$593
$591
$589
$587
$586
$361
5
10
6
IC
Highl
2024
$655
$648
$643
$640
$637
$635
$633
$632
$389
5
15
11
IC
Highl
2024
$645
$638
$634
$630
$628
$626
$624
$622
$384
5
20
16
IC
Highl
2024
$636
$629
$624
$621
$618
$616
$615
$613
$378
6
10
6
IC
Highl
2024
$662
$655
$651
$647
$645
$642
$641
$639
$394
6
15
11
IC
Highl
2024
$653
$646
$641
$638
$635
$633
$631
$630
$388
6
20
16
IC
Highl
2024
$643
$636
$631
$628
$626
$623
$622
$620
$382
1
10
6
TC
$1,548
$1,448
$1,383
$1,336
$1,299
$1,269
$1,244
$1,223
$999
1
15
11
TC
$1,533
$1,434
$1,370
$1,323
$1,287
$1,257
$1,232
$1,211
$989
1
20
16
TC
$1,518
$1,421
$1,357
$1,311
$1,274
$1,245
$1,221
$1,200
$980
2
10
6
TC
$1,616
$1,512
$1,444
$1,395
$1,356
$1,325
$1,299
$1,277
$1,043
2
15
11
TC
$1,599
$1,496
$1,429
$1,380
$1,342
$1,312
$1,286
$1,264
$1,032
2
20
16
TC
$1,583
$1,481
$1,414
$1,366
$1,328
$1,298
$1,272
$1,251
$1,022
3
10
5
TC
$1,652
$1,545
$1,476
$1,426
$1,386
$1,355
$1,328
$1,305
$1,066
3
15
10
TC
$1,633
$1,528
$1,459
$1,409
$1,371
$1,339
$1,313
$1,290
$1,054
3
20
15
TC
$1,614
$1,510
$1,443
$1,393
$1,355
$1,324
$1,298
$1,276
$1,042
4
10
6
TC
$1,728
$1,617
$1,544
$1,492
$1,451
$1,417
$1,389
$1,366
$1,115
4
15
11
TC
$1,706
$1,597
$1,525
$1,473
$1,432
$1,399
$1,372
$1,348
$1,101
4
20
16
TC
$1,685
$1,576
$1,506
$1,454
$1,414
$1,382
$1,354
$1,331
$1,087
5
10
6
TC
$1,817
$1,700
$1,624
$1,568
$1,525
$1,490
$1,461
$1,436
$1,173
5
15
11
TC
$1,790
$1,675
$1,600
$1,545
$1,502
$1,468
$1,439
$1,414
$1,155
5
20
16
TC
$1,763
$1,650
$1,576
$1,522
$1,480
$1,446
$1,417
$1,393
$1,138
6
10
6
TC
$1,838
$1,720
$1,642
$1,586
$1,543
$1,507
$1,478
$1,452
$1,186
6
15
11
TC
$1,811
$1,694
$1,618
$1,563
$1,520
$1,485
$1,456
$1,431
$1,169
6
20
16
TC
$1,784
$1,669
$1,594
$1,539
$1,497
$1,463
$1,434
$1,409
$1,151
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.127 Costs for 20 Mile Plug-in Hybrid Batteries (dollar values in 2015$)
Curb
Weight
Class
WRtech
WRnet
Cost
type
DMC: base
year cost
IC:
complexity
DMC:
learning
curve
IC: near
term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
15
6
DMC
$2,422
26
$3,906
$3,666
$3,466
$3,296
$3,150
$3,022
$2,908
$2,807
$2,716
1
20
11
DMC
$2,400
26
$3,871
$3,632
$3,434
$3,266
$3,121
$2,994
$2,882
$2,782
$2,691
2
15
6
DMC
$2,522
26
$4,068
$3,817
$3,609
$3,433
$3,280
$3,147
$3,029
$2,923
$2,828
2
20
11
DMC
$2,464
26
$3,975
$3,730
$3,527
$3,354
$3,205
$3,075
$2,960
$2,857
$2,764
3
15
6
DMC
$2,611
26
$4,211
$3,952
$3,736
$3,553
$3,396
$3,258
$3,135
$3,026
$2,928
3
20
11
DMC
$2,586
26
$4,171
$3,914
$3,700
$3,519
$3,363
$3,226
$3,105
$2,997
$2,900
4
15
6
DMC
$2,758
26
$4,449
$4,175
$3,947
$3,754
$3,587
$3,441
$3,312
$3,197
$3,093
4
20
11
DMC
$2,731
26
$4,405
$4,134
$3,909
$3,717
$3,552
$3,408
$3,280
$3,166
$3,063
5
15
6
DMC
$2,972
26
$4,793
$4,498
$4,253
$4,045
$3,865
$3,708
$3,569
$3,445
$3,333
5
20
11
DMC
$2,929
26
$4,724
$4,433
$4,191
$3,986
$3,809
$3,654
$3,517
$3,395
$3,284
1
15
6
IC
High2
2024
$1,974
$1,956
$1,942
$1,929
$1,918
$1,909
$1,901
$1,893
$1,217
1
20
11
IC
High2
2024
$1,956
$1,939
$1,924
$1,912
$1,901
$1,892
$1,883
$1,876
$1,206
2
15
6
IC
High2
2024
$2,056
$2,037
$2,022
$2,009
$1,998
$1,988
$1,979
$1,972
$1,267
2
20
11
IC
High2
2024
$2,009
$1,991
$1,976
$1,963
$1,952
$1,943
$1,934
$1,927
$1,239
3
15
6
IC
High2
2024
$2,128
$2,109
$2,093
$2,080
$2,068
$2,058
$2,049
$2,041
$1,312
3
20
11
IC
High2
2024
$2,108
$2,089
$2,073
$2,060
$2,048
$2,038
$2,029
$2,021
$1,299
2-400
-------
Technology Cost, Effectiveness, and Lead Time Assessment
4
15
6
IC
High2
2024
$2,248
$2,228
$2,211
$2,197
$2,185
$2,174
$2,165
$2,156
$1,386
4
20
11
IC
High2
2024
$2,226
$2,206
$2,190
$2,176
$2,164
$2,153
$2,144
$2,135
$1,373
5
15
6
IC
High2
2024
$2,422
$2,401
$2,383
$2,367
$2,354
$2,343
$2,332
$2,323
$1,493
5
20
11
IC
High2
2024
$2,387
$2,366
$2,348
$2,333
$2,320
$2,309
$2,298
$2,289
$1,472
1
15
6
TC
$5,880
$5,622
$5,407
$5,225
$5,068
$4,931
$4,809
$4,700
$3,933
1
20
11
TC
$5,827
$5,571
$5,358
$5,178
$5,022
$4,886
$4,765
$4,658
$3,897
2
15
6
TC
$6,124
$5,855
$5,631
$5,442
$5,278
$5,135
$5,008
$4,895
$4,096
2
20
11
TC
$5,984
$5,721
$5,503
$5,317
$5,158
$5,018
$4,894
$4,783
$4,002
3
15
6
TC
$6,339
$6,061
$5,830
$5,633
$5,464
$5,316
$5,185
$5,067
$4,240
3
20
11
TC
$6,278
$6,002
$5,773
$5,579
$5,411
$5,264
$5,134
$5,018
$4,199
4
15
6
TC
$6,697
$6,403
$6,158
$5,951
$5,772
$5,615
$5,477
$5,353
$4,479
4
20
11
TC
$6,632
$6,340
$6,098
$5,893
$5,716
$5,561
$5,424
$5,301
$4,436
5
15
6
TC
$7,216
$6,899
$6,635
$6,412
$6,219
$6,050
$5,901
$5,768
$4,826
5
20
11
TC
$7,111
$6,799
$6,539
$6,319
$6,129
$5,963
$5,815
$5,684
$4,756
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.128 Costs for 40 Mile Plug-in Hybrid Batteries (dollar values in 2015$)
Curb
WR
WR
Cost
DMC: base
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
tech
net
type
year cost
learning
Class
IC:
complexity
curve
IC: near
term thru
1
20
7
DMC
$3,195
26
$5,153
$4,836
$4,572
$4,348
$4,155
$3,986
$3,837
$3,703
$3,583
2
20
6
DMC
$3,438
26
$5,545
$5,203
$4,920
$4,679
$4,471
$4,289
$4,128
$3,985
$3,855
3
20
5
DMC
$3,585
26
$5,783
$5,427
$5,131
$4,880
$4,663
$4,474
$4,306
$4,156
$4,021
4
20
5
DMC
$3,976
26
$6,413
$6,018
$5,690
$5,412
$5,171
$4,961
$4,775
$4,609
$4,459
5
20
7
DMC
$4,433
26
$7,151
$6,710
$6,344
$6,034
$5,766
$5,532
$5,324
$5,139
$4,972
1
20
7
IC
High2
2024
$2,604
$2,581
$2,561
$2,545
$2,531
$2,518
$2,507
$2,497
$1,606
2
20
6
IC
High2
2024
$2,802
$2,777
$2,756
$2,739
$2,723
$2,710
$2,698
$2,687
$1,728
3
20
5
IC
High2
2024
$2,923
$2,896
$2,875
$2,856
$2,840
$2,826
$2,814
$2,803
$1,802
4
20
5
IC
High2
2024
$3,241
$3,212
$3,188
$3,167
$3,150
$3,134
$3,121
$3,108
$1,998
5
20
7
IC
High2
2024
$3,614
$3,582
$3,555
$3,532
$3,512
$3,495
$3,479
$3,466
$2,228
1
20
7
TC
$7,757
$7,416
$7,133
$6,893
$6,686
$6,504
$6,344
$6,201
$5,188
2
20
6
TC
$8,347
$7,980
$7,676
$7,417
$7,194
$6,999
$6,826
$6,672
$5,583
3
20
5
TC
$8,706
$8,323
$8,006
$7,736
$7,503
$7,300
$7,120
$6,959
$5,823
4
20
5
TC
$9,654
$9,230
$8,878
$8,579
$8,321
$8,095
$7,895
$7,717
$6,457
5
20
7
TC
$10,765
$10,292
$9,899
$9,566
$9,278
$9,026
$8,804
$8,605
$7,200
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.129 Costs for 75 Mile BEV Batteries (dollar values in 2015$)
Curb
Weight
Class
WR
tec
h
W
R
net
Cost
type
DMC:
base
year
cost
IC:
complex
ity
DMC:
learnin
g curve
IC: near
term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
10
10
DMC
$3,743
26
$6,038
$5,666
$5,357
$5,095
$4,869
$4,671
$4,496
$4,339
$4,198
1
15
15
DMC
$3,705
26
$5,977
$5,609
$5,303
$5,043
$4,819
$4,623
$4,450
$4,295
$4,156
1
20
20
DMC
$3,667
26
$5,915
$5,551
$5,248
$4,991
$4,770
$4,576
$4,404
$4,251
$4,113
2
10
10
DMC
$3,874
26
$6,248
$5,863
$5,543
$5,272
$5,038
$4,833
$4,652
$4,490
$4,344
2
15
15
DMC
$3,817
26
$6,156
$5,777
$5,462
$5,195
$4,964
$4,762
$4,584
$4,424
$4,280
2
20
20
DMC
$3,760
26
$6,064
$5,691
$5,381
$5,117
$4,890
$4,691
$4,515
$4,358
$4,217
3
10
10
DMC
$3,965
26
$6,395
$6,001
$5,674
$5,396
$5,157
$4,947
$4,762
$4,596
$4,447
3
15
15
DMC
$3,899
26
$6,289
$5,902
$5,580
$5,307
$5,071
$4,865
$4,683
$4,520
$4,373
3
20
20
DMC
$3,834
26
$6,183
$5,803
$5,486
$5,218
$4,986
$4,783
$4,604
$4,444
$4,299
4
10
10
DMC
$4,229
26
$6,821
$6,401
$6,052
$5,756
$5,500
$5,277
$5,079
$4,902
$4,743
4
15
15
DMC
$4,180
26
$6,742
$6,326
$5,981
$5,689
$5,436
$5,215
$5,020
$4,845
$4,688
4
20
20
DMC
$4,131
26
$6,662
$6,252
$5,911
$5,622
$5,372
$5,154
$4,960
$4,788
$4,632
5
10
10
DMC
$4,694
26
$7,571
$7,105
$6,717
$6,389
$6,105
$5,857
$5,637
$5,441
$5,264
2-401
-------
Technology Cost, Effectiveness, and Lead Time Assessment
5
15
15
DMC
$4,597
26
$7,414
$6,958
$6,578
$6,256
$5,979
$5,735
$5,520
$5,328
$5,155
5
20
20
DMC
$4,500
26
$7,258
$6,811
$6,439
$6,124
$5,852
$5,614
$5,404
$5,216
$5,046
1
10
10
IC
High2
2024
$3,052
$3,024
$3,001
$2,982
$2,965
$2,951
$2,938
$2,926
$1,881
1
15
15
IC
High2
2024
$3,021
$2,993
$2,971
$2,952
$2,935
$2,921
$2,908
$2,897
$1,862
1
20
20
IC
High2
2024
$2,990
$2,963
$2,940
$2,922
$2,905
$2,891
$2,878
$2,867
$1,843
2
10
10
IC
High2
2024
$3,158
$3,129
$3,106
$3,086
$3,068
$3,053
$3,040
$3,028
$1,947
2
15
15
IC
High2
2024
$3,111
$3,083
$3,060
$3,040
$3,023
$3,009
$2,995
$2,984
$1,918
2
20
20
IC
High2
2024
$3,065
$3,037
$3,015
$2,995
$2,978
$2,964
$2,951
$2,939
$1,890
3
10
10
IC
High2
2024
$3,232
$3,203
$3,179
$3,159
$3,141
$3,125
$3,112
$3,100
$1,993
3
15
15
IC
High2
2024
$3,179
$3,150
$3,126
$3,106
$3,089
$3,074
$3,060
$3,048
$1,960
3
20
20
IC
High2
2024
$3,125
$3,097
$3,074
$3,054
$3,037
$3,022
$3,009
$2,997
$1,927
4
10
10
IC
High2
2024
$3,447
$3,416
$3,391
$3,369
$3,350
$3,334
$3,319
$3,306
$2,125
4
15
15
IC
High2
2024
$3,407
$3,377
$3,351
$3,330
$3,311
$3,295
$3,280
$3,268
$2,101
4
20
20
IC
High2
2024
$3,367
$3,337
$3,312
$3,290
$3,272
$3,256
$3,242
$3,229
$2,076
5
10
10
IC
High2
2024
$3,826
$3,792
$3,763
$3,739
$3,718
$3,700
$3,684
$3,669
$2,359
5
15
15
IC
High2
2024
$3,747
$3,714
$3,686
$3,662
$3,641
$3,624
$3,608
$3,594
$2,310
5
20
20
IC
High2
2024
$3,668
$3,635
$3,608
$3,585
$3,564
$3,547
$3,531
$3,518
$2,261
1
10
10
TC
$9,090
$8,690
$8,359
$8,077
$7,834
$7,622
$7,434
$7,266
$6,080
1
15
15
TC
$8,997
$8,602
$8,274
$7,995
$7,755
$7,544
$7,358
$7,192
$6,018
1
20
20
TC
$8,905
$8,514
$8,189
$7,913
$7,675
$7,467
$7,283
$7,118
$5,956
2
10
10
TC
$9,405
$8,992
$8,649
$8,358
$8,106
$7,886
$7,692
$7,518
$6,291
2
15
15
TC
$9,267
$8,860
$8,522
$8,235
$7,987
$7,771
$7,579
$7,408
$6,198
2
20
20
TC
$9,129
$8,728
$8,395
$8,112
$7,868
$7,655
$7,466
$7,297
$6,106
3
10
10
TC
$9,627
$9,204
$8,853
$8,555
$8,298
$8,073
$7,873
$7,696
$6,439
3
15
15
TC
$9,468
$9,052
$8,707
$8,413
$8,160
$7,939
$7,743
$7,568
$6,333
3
20
20
TC
$9,309
$8,900
$8,560
$8,272
$8,023
$7,805
$7,613
$7,441
$6,226
4
10
10
TC
$10,268
$9,817
$9,443
$9,125
$8,850
$8,610
$8,398
$8,208
$6,868
4
15
15
TC
$10,149
$9,703
$9,333
$9,018
$8,747
$8,510
$8,300
$8,112
$6,788
4
20
20
TC
$10,029
$9,589
$9,223
$8,912
$8,644
$8,410
$8,202
$8,017
$6,708
5
10
10
TC
$11,397
$10,897
$10,481
$10,128
$9,823
$9,557
$9,321
$9,110
$7,623
5
15
15
TC
$11,161
$10,671
$10,264
$9,918
$9,620
$9,359
$9,128
$8,922
$7,465
5
20
20
TC
$10,926
$10,446
$10,047
$9,709
$9,417
$9,161
$8,935
$8,733
$7,308
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.130 Costs for 100 Mile BEV Batteries (dollar values in 2015$)
Curb
Weight
Class
WR
tec
h
W
R
net
Cost
type
DMC:
base
year cost
IC:
complexi
ty
DMC:
learnin
g curve
IC: near
term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
10
10
DMC
$4,185
26
$6,750
$6,334
$5,989
$5,696
$5,443
$5,221
$5,026
$4,851
$4,693
1
15
15
DMC
$4,130
26
$6,661
$6,251
$5,910
$5,621
$5,371
$5,153
$4,960
$4,787
$4,632
1
20
20
DMC
$4,075
26
$6,573
$6,168
$5,832
$5,546
$5,300
$5,085
$4,894
$4,724
$4,570
2
10
10
DMC
$4,367
26
$7,044
$6,610
$6,249
$5,943
$5,680
$5,449
$5,244
$5,062
$4,897
2
15
15
DMC
$4,289
26
$6,917
$6,491
$6,137
$5,837
$5,578
$5,351
$5,150
$4,971
$4,810
2
20
20
DMC
$4,210
26
$6,791
$6,373
$6,025
$5,731
$5,476
$5,253
$5,056
$4,880
$4,722
3
10
9
DMC
$4,516
26
$7,284
$6,835
$6,463
$6,146
$5,873
$5,635
$5,423
$5,234
$5,065
3
15
14
DMC
$4,418
26
$7,127
$6,688
$6,323
$6,014
$5,746
$5,513
$5,306
$5,121
$4,955
3
20
19
DMC
$4,321
26
$6,969
$6,540
$6,183
$5,881
$5,620
$5,391
$5,189
$5,008
$4,846
4
10
10
DMC
$4,748
26
$7,659
$7,187
$6,795
$6,462
$6,176
$5,924
$5,702
$5,504
$5,325
4
15
15
DMC
$4,602
26
$7,423
$6,966
$6,586
$6,263
$5,985
$5,742
$5,527
$5,334
$5,161
4
20
20
DMC
$4,456
26
$7,187
$6,744
$6,376
$6,064
$5,795
$5,559
$5,351
$5,165
$4,997
5
10
10
DMC
$5,441
26
$8,776
$8,235
$7,786
$7,405
$7,076
$6,789
$6,534
$6,307
$6,102
5
15
15
DMC
$5,155
26
$8,315
$7,803
$7,377
$7,016
$6,705
$6,432
$6,191
$5,976
$5,782
5
20
20
DMC
$4,870
26
$7,854
$7,370
$6,969
$6,628
$6,333
$6,076
$5,848
$5,644
$5,461
1
10
10
IC
High2
2024
$3,411
$3,381
$3,355
$3,334
$3,315
$3,299
$3,284
$3,272
$2,103
1
15
15
IC
High2
2024
$3,367
$3,336
$3,311
$3,290
$3,272
$3,256
$3,241
$3,229
$2,075
1
20
20
IC
High2
2024
$3,322
$3,292
$3,267
$3,246
$3,228
$3,212
$3,198
$3,186
$2,048
2
10
10
IC
High2
2024
$3,560
$3,528
$3,501
$3,479
$3,459
$3,442
$3,427
$3,414
$2,195
2
15
15
IC
High2
2024
$3,496
$3,465
$3,439
$3,416
$3,397
$3,381
$3,366
$3,353
$2,155
2
20
20
IC
High2
2024
$3,432
$3,401
$3,376
$3,354
$3,335
$3,319
$3,305
$3,292
$2,116
2-402
-------
Technology Cost, Effectiveness, and Lead Time Assessment
3
10
9
IC
High2
2024
$3,681
$3,648
$3,621
$3,597
$3,577
$3,560
$3,544
$3,530
$2,269
3
15
14
IC
High2
2024
$3,602
$3,569
$3,543
$3,520
$3,500
$3,483
$3,468
$3,454
$2,220
3
20
19
IC
High2
2024
$3,522
$3,491
$3,464
$3,442
$3,423
$3,406
$3,391
$3,378
$2,171
4
10
10
IC
High2
2024
$3,871
$3,836
$3,807
$3,783
$3,761
$3,743
$3,727
$3,712
$2,386
4
15
15
IC
High2
2024
$3,751
$3,718
$3,690
$3,666
$3,646
$3,628
$3,612
$3,598
$2,313
4
20
20
IC
High2
2024
$3,632
$3,600
$3,572
$3,550
$3,530
$3,512
$3,497
$3,483
$2,239
5
10
10
IC
High2
2024
$4,435
$4,395
$4,362
$4,334
$4,310
$4,289
$4,270
$4,253
$2,734
5
15
15
IC
High2
2024
$4,202
$4,165
$4,133
$4,107
$4,084
$4,064
$4,046
$4,030
$2,591
5
20
20
IC
High2
2024
$3,969
$3,934
$3,904
$3,879
$3,857
$3,839
$3,822
$3,807
$2,447
1
10
10
TC
$10,161
$9,715
$9,344
$9,029
$8,758
$8,520
$8,310
$8,122
$6,796
1
15
15
TC
$10,028
$9,587
$9,222
$8,911
$8,643
$8,409
$8,201
$8,016
$6,707
1
20
20
TC
$9,895
$9,460
$9,099
$8,793
$8,528
$8,297
$8,092
$7,909
$6,618
2
10
10
TC
$10,603
$10,137
$9,751
$9,422
$9,139
$8,891
$8,672
$8,476
$7,092
2
15
15
TC
$10,413
$9,956
$9,576
$9,253
$8,975
$8,732
$8,516
$8,324
$6,965
2
20
20
TC
$10,224
$9,774
$9,401
$9,085
$8,811
$8,572
$8,361
$8,172
$6,838
3
10
9
TC
$10,965
$10,483
$10,083
$9,744
$9,451
$9,194
$8,967
$8,765
$7,334
3
15
14
TC
$10,728
$10,257
$9,865
$9,533
$9,247
$8,996
$8,774
$8,575
$7,176
3
20
19
TC
$10,491
$10,031
$9,648
$9,323
$9,042
$8,797
$8,580
$8,386
$7,017
4
10
10
TC
$11,529
$11,023
$10,602
$10,245
$9,937
$9,667
$9,429
$9,216
$7,711
4
15
15
TC
$11,174
$10,683
$10,275
$9,929
$9,631
$9,370
$9,138
$8,932
$7,474
4
20
20
TC
$10,819
$10,344
$9,949
$9,614
$9,325
$9,072
$8,848
$8,648
$7,236
5
10
10
TC
$13,211
$12,631
$12,149
$11,740
$11,387
$11,078
$10,804
$10,560
$8,836
5
15
15
TC
$12,518
$11,968
$11,511
$11,123
$10,789
$10,496
$10,237
$10,006
$8,372
5
20
20
TC
$11,824
$11,304
$10,873
$10,507
$10,191
$9,914
$9,670
$9,451
$7,908
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.131 Costs for 200 Mile BEV Batteries (dollar values in 2015$)
Curb
WR
W
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
tec
R
type
base
learnin
Class
h
net
year cost
IC:
complexi
ty
g curve
IC: near
term
thru
1
20
13
DMC
$5,709
26
$9,208
$8,641
$8,170
$7,770
$7,425
$7,123
$6,856
$6,617
$6,402
2
20
14
DMC
$5,939
26
$9,580
$8,989
$8,499
$8,083
$7,724
$7,410
$7,132
$6,884
$6,661
3
20
13
DMC
$6,120
26
$9,871
$9,263
$8,758
$8,329
$7,960
$7,636
$7,350
$7,094
$6,863
4
20
14
DMC
$6,377
26
$10,285
$9,652
$9,125
$8,679
$8,293
$7,956
$7,658
$7,391
$7,151
5
20
14
DMC
$6,953
26
$11,215
$10,524
$9,950
$9,463
$9,043
$8,675
$8,350
$8,059
$7,798
1
20
13
IC
High2
2024
$4,654
$4,612
$4,577
$4,548
$4,522
$4,500
$4,480
$4,463
$2,869
2
20
14
IC
High2
2024
$4,841
$4,798
$4,762
$4,731
$4,705
$4,682
$4,661
$4,643
$2,985
3
20
13
IC
High2
2024
$4,989
$4,944
$4,907
$4,875
$4,848
$4,824
$4,803
$4,784
$3,076
4
20
14
IC
High2
2024
$5,198
$5,151
$5,113
$5,080
$5,051
$5,027
$5,005
$4,985
$3,205
5
20
14
IC
High2
2024
$5,668
$5,617
$5,575
$5,539
$5,508
$5,481
$5,457
$5,435
$3,494
1
20
13
TC
$13,861
$13,253
$12,747
$12,317
$11,947
$11,623
$11,336
$11,080
$9,271
2
20
14
TC
$14,421
$13,787
$13,261
$12,815
$12,429
$12,092
$11,794
$11,527
$9,645
3
20
13
TC
$14,860
$14,207
$13,665
$13,205
$12,808
$12,460
$12,153
$11,878
$9,939
4
20
14
TC
$15,483
$14,803
$14,238
$13,759
$13,345
$12,983
$12,662
$12,376
$10,356
5
20
14
TC
$16,883
$16,141
$15,525
$15,002
$14,551
$14,156
$13,807
$13,495
$11,292
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.3.7.8 Electrified Vehicle Costs Used In OMEGA (Battery + Non-battery Items)
Costs presented in the tables that follow sum the battery, non-battery and, where applicable,
the in-home charger related costs for mild, strong and plug-in hybrids and full battery electric
vehicles.
2-403
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.132 Full System Costs for 48V Mild Hybrids (2015$)
Curb
WRtech
WRnet
Cost
Weight
Type
Class
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
5
1.5
TC
$1,073
$1,035
$965
$944
$927
$913
$900
$889
$814
2
5
2
TC
$1,073
$1,035
$965
$944
$927
$913
$900
$889
$814
3
5
2.5
TC
$1,073
$1,035
$965
$944
$927
$913
$900
$889
$814
4
5
2.5
TC
$1,073
$1,035
$965
$944
$927
$913
$900
$889
$814
5
5
2.5
TC
$1,073
$1,035
$965
$944
$927
$913
$900
$889
$814
6
5
3
TC
$1,073
$1,035
$965
$944
$927
$913
$900
$889
$814
Note: TC=total costs.
Table 2.133 Full System Costs for Strong Hybrids (2015$)
Curb
Weight
Class
WRtech
WRnet
Cost type
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
10
6
TC
$4,225
$4,097
$3,615
$3,545
$3,486
$3,436
$3,392
$3,353
$3,112
1
15
11
TC
$4,189
$4,063
$3,585
$3,515
$3,457
$3,408
$3,364
$3,325
$3,087
1
20
16
TC
$4,154
$4,029
$3,554
$3,485
$3,428
$3,379
$3,336
$3,297
$3,061
2
10
6
TC
$4,512
$4,378
$3,859
$3,784
$3,723
$3,670
$3,623
$3,581
$3,329
2
15
11
TC
$4,468
$4,335
$3,821
$3,747
$3,686
$3,634
$3,588
$3,547
$3,297
2
20
16
TC
$4,424
$4,293
$3,783
$3,710
$3,650
$3,598
$3,552
$3,512
$3,265
3
10
5
TC
$4,628
$4,491
$3,957
$3,881
$3,818
$3,764
$3,716
$3,673
$3,416
3
15
10
TC
$4,579
$4,443
$3,915
$3,840
$3,777
$3,724
$3,677
$3,634
$3,380
3
20
15
TC
$4,530
$4,396
$3,873
$3,799
$3,737
$3,684
$3,637
$3,595
$3,343
4
10
6
TC
$4,817
$4,674
$4,120
$4,040
$3,975
$3,918
$3,868
$3,824
$3,554
4
15
11
TC
$4,757
$4,615
$4,068
$3,989
$3,924
$3,869
$3,819
$3,776
$3,509
4
20
16
TC
$4,696
$4,556
$4,016
$3,939
$3,874
$3,819
$3,771
$3,727
$3,465
5
10
6
TC
$5,222
$5,070
$4,463
$4,378
$4,307
$4,246
$4,193
$4,145
$3,861
5
15
11
TC
$5,148
$4,998
$4,399
$4,315
$4,246
$4,186
$4,134
$4,086
$3,806
5
20
16
TC
$5,074
$4,926
$4,336
$4,253
$4,185
$4,126
$4,074
$4,028
$3,752
6
10
6
TC
$5,256
$5,102
$4,492
$4,406
$4,335
$4,274
$4,220
$4,172
$3,884
6
15
11
TC
$5,179
$5,027
$4,426
$4,342
$4,271
$4,211
$4,158
$4,111
$3,828
6
20
16
TC
$5,102
$4,952
$4,360
$4,277
$4,208
$4,149
$4,096
$4,050
$3,771
Note: TC=total costs.
Table 2.134 Full System Costs for 20 Mile Plug-in Hybrids, Including Charger & Charger Labor (2015$)
Curb
WRtech
WRnet
Cost
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
type
Class
1
15
6
TC
$10,259
$9,964
$9,248
$9,036
$8,851
$8,688
$8,542
$8,410
$7,614
1
20
11
TC
$10,209
$9,916
$9,202
$8,991
$8,808
$8,646
$8,501
$8,370
$7,581
2
15
6
TC
$10,827
$10,518
$9,743
$9,520
$9,326
$9,155
$9,001
$8,863
$8,033
2
20
11
TC
$10,692
$10,389
$9,618
$9,400
$9,209
$9,041
$8,891
$8,755
$7,943
3
15
6
TC
$11,164
$10,844
$10,042
$9,811
$9,611
$9,433
$9,275
$9,132
$8,273
3
20
11
TC
$11,107
$10,790
$9,989
$9,761
$9,562
$9,386
$9,229
$9,087
$8,236
4
15
6
TC
$11,768
$11,430
$10,576
$10,333
$10,120
$9,933
$9,765
$9,614
$8,707
4
20
11
TC
$11,709
$11,374
$10,522
$10,280
$10,070
$9,883
$9,717
$9,567
$8,668
5
15
6
TC
$12,750
$12,384
$11,439
$11,176
$10,946
$10,743
$10,561
$10,397
$9,419
5
20
11
TC
$12,653
$12,292
$11,349
$11,089
$10,862
$10,661
$10,481
$10,319
$9,355
Note: TC=total costs.
Table 2.135 Full System Costs for 40 Mile Plug-in Hybrids, Including Charger & Charger Labor (2015$)
Curb
WRtech
WRnet
Cost
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
type
Class
1
20
7
TC
$13,163
$12,763
$11,855
$11,568
$11,318
$11,098
$10,901
$10,724
$9,641
2
20
6
TC
$14,370
$13,936
$12,917
$12,605
$12,333
$12,092
$11,878
$11,685
$10,515
3
20
5
TC
$14,990
$14,535
$13,467
$13,141
$12,856
$12,605
$12,381
$12,179
$10,955
4
20
5
TC
$16,479
$15,978
$14,791
$14,430
$14,116
$13,839
$13,591
$13,368
$12,017
5
20
7
TC
$18,327
$17,769
$16,427
$16,026
$15,676
$15,367
$15,091
$14,842
$13,342
2-404
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Note: TC=total costs.
Table 2.136 Full System Costs for 75 Mile BEVs, Including Charger & Charger Labor (2015$)
Curb
WRtech
WRnet
Cost
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
Class
type
1
10
10
TC
$10,660
$10,240
$9,892
$9,596
$9,340
$9,117
$8,920
$8,744
$7,464
1
15
15
TC
$10,569
$10,153
$9,808
$9,514
$9,262
$9,041
$8,845
$8,670
$7,404
1
20
20
TC
$10,478
$10,066
$9,724
$9,433
$9,183
$8,964
$8,770
$8,597
$7,344
2
10
10
TC
$11,520
$11,076
$10,708
$10,395
$10,124
$9,887
$9,678
$9,490
$8,121
2
15
15
TC
$11,395
$10,957
$10,594
$10,285
$10,018
$9,784
$9,577
$9,392
$8,039
2
20
20
TC
$11,269
$10,838
$10,479
$10,174
$9,911
$9,680
$9,476
$9,294
$7,957
3
10
10
TC
$11,128
$10,690
$10,328
$10,021
$9,756
$9,525
$9,320
$9,138
$7,700
3
15
15
TC
$10,970
$10,540
$10,183
$9,880
$9,620
$9,392
$9,191
$9,011
$7,596
3
20
20
TC
$10,812
$10,389
$10,037
$9,740
$9,483
$9,259
$9,061
$8,885
$7,493
4
10
10
TC
$11,929
$11,454
$11,059
$10,724
$10,434
$10,181
$9,957
$9,757
$8,349
4
15
15
TC
$11,828
$11,358
$10,967
$10,635
$10,349
$10,099
$9,877
$9,679
$8,284
4
20
20
TC
$11,727
$11,262
$10,875
$10,547
$10,264
$10,016
$9,797
$9,601
$8,218
5
10
10
TC
$14,070
$13,532
$13,084
$12,704
$12,375
$12,086
$11,831
$11,602
$9,884
5
15
15
TC
$13,857
$13,329
$12,890
$12,516
$12,193
$11,910
$11,659
$11,435
$9,743
5
20
20
TC
$13,643
$13,126
$12,695
$12,328
$12,012
$11,734
$11,488
$11,268
$9,603
Note: TC=total costs.
Table 2.137 Full System Costs for 100 Mile BEVs, Including Charger & Charger Labor (2015$)
Curb
WRtech
WRnet
Cost
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
Class
type
1
10
10
TC
$11,732
$11,265
$10,877
$10,548
$10,264
$10,016
$9,796
$9,600
$8,180
1
15
15
TC
$11,600
$11,139
$10,755
$10,430
$10,150
$9,905
$9,688
$9,494
$8,093
1
20
20
TC
$11,468
$11,012
$10,634
$10,313
$10,036
$9,794
$9,580
$9,388
$8,006
2
10
10
TC
$12,718
$12,222
$11,809
$11,459
$11,157
$10,892
$10,657
$10,448
$8,923
2
15
15
TC
$12,540
$12,053
$11,647
$11,303
$11,005
$10,745
$10,514
$10,308
$8,805
2
20
20
TC
$12,363
$11,884
$11,485
$11,146
$10,854
$10,598
$10,371
$10,168
$8,688
3
10
9
TC
$12,465
$11,969
$11,558
$11,209
$10,909
$10,646
$10,414
$10,207
$8,594
3
15
14
TC
$12,230
$11,744
$11,341
$11,000
$10,706
$10,448
$10,221
$10,018
$8,439
3
20
19
TC
$11,995
$11,519
$11,125
$10,791
$10,503
$10,251
$10,028
$9,830
$8,283
4
10
10
TC
$13,190
$12,659
$12,218
$11,844
$11,521
$11,238
$10,988
$10,765
$9,192
4
15
15
TC
$12,853
$12,338
$11,910
$11,546
$11,233
$10,958
$10,715
$10,498
$8,969
4
20
20
TC
$12,516
$12,016
$11,601
$11,248
$10,944
$10,678
$10,442
$10,232
$8,746
5
10
10
TC
$15,883
$15,266
$14,752
$14,315
$13,938
$13,607
$13,314
$13,052
$11,097
5
15
15
TC
$15,212
$14,625
$14,136
$13,721
$13,361
$13,047
$12,768
$12,518
$10,650
5
20
20
TC
$14,541
$13,984
$13,520
$13,126
$12,785
$12,486
$12,221
$11,985
$10,203
Note: TC
=total costs.
Table 2.138 Full System Costs for 200 Mile BEVs, Including Charger & Charger Labor (2015$)
Curb
WRtech
WRnet
Cost
2017
2018
2019
2020
2021
2022
2023
2024
2025
Weight
Class
type
1
20
13
TC
$15,433
$14,803
$14,280
$13,837
$13,454
$13,119
$12,823
$12,558
$10,657
2
20
14
TC
$16,546
$15,882
$15,331
$14,862
$14,458
$14,103
$13,790
$13,509
$11,485
3
20
13
TC
$16,361
$15,694
$15,141
$14,671
$14,267
$13,913
$13,600
$13,321
$11,202
4
20
14
TC
$17,158
$16,453
$15,868
$15,371
$14,943
$14,567
$14,236
$13,939
$11,848
5
20
14
TC
$19,636
$18,856
$18,207
$17,656
$17,180
$16,762
$16,393
$16,062
$13,615
Note: TC=total costs.
2.3.4.4 Aerodynamics: Data and Assumptions for this Assessment
For this Proposed Determination, as in the Draft TAR, EPA has considered two levels of
aerodynamic improvements: Aerol and Aero2. The first level, Aerol, represents a 10 percent
2-405
-------
Technology Cost, Effectiveness, and Lead Time Assessment
reduction in drag from a baseline MY2008-level vehicle. The second level, Aero2, represents a
20 percent reduction from the same baseline (nominally, 10 percentage points incremental to
Aerol).
In Chapter 2.2.5 of this TSD, EPA further considered the feasibility of aerodynamic
improvements of this general degree, outlining examples that show that manufacturers are
gaining aerodynamic benefits by implementing several varieties of passive and active
aerodynamic technologies in current vehicles, while a range of opportunities remain to further
apply these and other passive and active technologies in a more optimized fashion as vehicles
enter redesign cycles in the future. That chapter also noted that for this Proposed Determination
analysis, EPA has provided for a better representation of the existence of applied aerodynamic
technology in the baseline fleet (as further described in this section).
The findings of the vehicle technology review and additional technology benefits evaluated in
the Joint Aerodynamics Assessment Program (described in Chapter 2.2.5) also lend support to
the feasibility of 10 percent and 20 percent improvements relative a MY2008-level baseline. As
noted in the Draft TAR, the NAS report also generally supported the assumptions for 10 percent
and 20 percent aerodynamic improvement as being applicable to the 2020 to 2025 time frame,
relative a MY2008-level baseline.
During the Draft TAR comment period, EPA received confidential comments from a
stakeholder that included piece-cost estimates for underbody covers that were higher than EPA's
Aerol DMCs. The extent of the underbody covers for this particular application was greater than
the engine compartment underbody covers assumed within EPA's cost estimate for Aerol, and is
more consistent with the type of treatment that EPA assumes will be required to achieve Aero2
levels. Furthermore, to the extent that the piece-cost estimates provided by the commenter are
influenced by the additional function of this particular application of protecting vulnerable
underbody components, the resulting costs would be expected to be higher than an underbody
cover that has the sole purpose of drag reduction. For the Proposed Determination analysis, EPA
is continuing to use the Draft TAR cost and effectiveness assumptions for passive and active
aerodynamic technologies, updated to 2015 dollars.
Several stakeholders submitted comments on EPA's treatment of aerodynamic improvements
in the baseline fleet. In particular, commenters noted that EPA's assumption that a 20 percent
reduction in aerodynamic drag is equally feasible for all vehicles in the baseline fleet does not
account for the drag reduction that some vehicles have already adopted. The Alliance and
individual OEMs including Ford, Mercedes-Benz, Toyota, and FCA, all recommended that EPA
adjust the aero levels in the baseline fleet to reflect appropriate drag reduction achieved by each
vehicle.
EPA agrees that aerodynamic drag reductions have been achieved in some MY2015 vehicles
relative to the levels of drag in MY2008-2010 designs that were used as the null technology
point of reference for the FRM and Draft TAR. Furthermore, EPA agrees with the commenters
that it is appropriate to account for aerodynamic drag reductions already present in the baseline
fleet in order to avoid overestimating the amount of additional improvement that can be achieved
at a given cost. Therefore, for this Proposed Determination, EPA has estimated the levels of
aerodynamic drag reduction already present in MY2015, and assigned one of three aerodynamic
levels to each vehicle in the baseline fleet. The process for determining the levels of AeroO,
Aerol, and Aero2 is described below.
2-406
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Using coast down coefficient values reported to EPA for vehicle certification, a value for the
drag area, CdA, was calculated for each vehicle in the MY2015 baseline fleet according to
Equation 16.
Equation 16. Aerodynamic Drag Area Calculation from Coast Down Coefficients
CdA = (B + 2Cv)/(pv)
In which:
B: Road load Coefficient (lbf/mph)
C: Road load Coefficient (lbf/mph2)
p: Air Density
v: Vehicle Speed at Aero Evaluation (68.2 mph (110 kmph))
Differences in frontal area and overall shape will directly influence a vehicle's calculated drag
area. Because these characteristics tend to vary significantly across market segments, EPA
categorized the MY2015 fleet using the size classifications defined in 40 CFR §600.315-08.
These classes are defined by vehicle interior volume and side profile shape, such that vehicles
with similar frontal areas and overall shape tend to be grouped together. For example, despite
having similar side profile shapes, small pickups are distinguished from standard pickups to
account for differences in frontal area. Within each of these vehicle size classes, the distribution
of calculated drag areas across the MY2015 fleet was then investigated in order to determine
appropriate cutoff values of CdA that would delineate between different levels of aerodynamic
drag. Table 2.140 shows the 50th percentile values of CdA defining the cutoff between AeroO
and Aerol levels, and the 10th percentile values of CdA defining the cutoff between Aerol and
Aero2 levels.
Table 2.139 MY2015 Aerodynamic Drag Area Statistics and Cutoff Values by Size Class
EPA Size Class
Production
CdA (ft2)
CdA (ft2)
Volume
Statistics*
Cutoff Values
Average
Standard
Deviation
AeroO to Aerol
50th percentile
Aerol to Aero2
10th percentile
Two Seaters
78,117
7.50
1.10
7.01
6.29
Subcompact Cars
418,583
7.90
0.58
8.19
7.10
Minicompact Cars
56,307
7.57
0.25
7.51
7.30
Compact Cars
1,760,020
7.48
0.69
7.57
6.69
Midsize Cars
3,363,603
7.67
0.58
7.68
6.83
Large Cars
735,631
7.74
1.06
7.72
6.52
Small Station Wagons
371,522
9.01
1.04
9.77
7.41
Midsize Station Wagons
110,423
9.27
0.55
9.55
8.25
Small SUV 2WD
1,589,325
10.22
0.87
10.28
9.12
Small SUV 4WD
2,620,222
10.80
1.58
10.42
9.23
Special Purpose Vehicle, minivan 2WD
519,773
10.56
0.88
10.48
9.32
Standard SUV 2WD
260,287
11.96
1.58
12.16
9.91
Standard SUV4WD
1,253,047
12.22
1.18
11.86
11.11
Small Pick-up Trucks 2WD
162,243
12.24
1.14
11.93
11.18
Small Pick-up Trucks 4WD
178,391
13.08
0.87
13.77
11.92
Standard Pick-up Trucks 2WD
248,320
14.25
0.74
14.35
13.72
2-407
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Standard Pick-up Trucks 4WD
1,016,923
14.26
1.78
15.13
12.14
*Note: Aerodynamic drag area statistics are weighted by MY2015 actual production volumes. Special
Purpose Vehicle and Van classes with production under 100,000 not shown for clarity.
The 50th and 10th percentile cutoff values shown above were chosen to establish the
aerodynamic drag reductions of 10 percent for Aerol and 20 percent Aero2, relative to Aero 0.
As shown in Table 2.140, across all classes the production weighted average reduction in drag
area between the midpoints of the CdA ranges is 10 percent moving from AeroO to Aerol, and 22
percent moving from AeroO to Aero2.
Table 2.140 Aerodynamic Drag Reduction Between Aero levels 0,1, and 2 by Size Class
CdA (ft2)
Drag Reduction
EPA Size Class
AeroO
Aerol
Aero2
AeroO to
AeroO to
midpoint
midpoint
midpoint
Aerol
Aero2
Two Seaters
7.71
6.65
5.98
14%
22%
Subcompact Cars
8.20
7.65
6.46
7%
21%
Minicompact Cars
7.72
7.40
7.26
4%
6%
Compact Cars
7.79
7.13
6.05
8%
22%
Midsize Cars
7.85
7.25
6.21
8%
21%
Large Cars
8.27
7.12
6.35
14%
23%
Small Station Wagons
9.94
8.59
7.28
14%
27%
Midsize Station Wagons
9.72
8.90
7.98
8%
18%
Small SUV 2WD
10.91
9.70
8.17
11%
25%
Small SUV 4WD
11.36
9.82
8.68
13%
24%
Special Purpose Vehicle, minivan 2WD
10.96
9.90
9.33
10%
15%
Standard SUV 2WD
12.93
11.03
9.89
15%
24%
Standard SUV 4WD
12.64
11.48
10.14
9%
20%
Small Pick-up Trucks 2WD
12.90
11.56
11.06
10%
14%
Small Pick-up Trucks 4WD
13.69
12.85
11.90
6%
13%
Standard Pick-up Trucks 2WD
14.84
14.04
12.89
5%
13%
Standard Pick-up Trucks 4WD
15.58
13.64
12.14
12%
22%
All size classes (production weighted)
-
-
-
10%
22%
In response to the comments received on the Draft TAR regarding an appropriate approach
for considering aerodynamics in the baseline fleet, EPA has applied some level of aerodynamic
drag reduction to a significant portion of the MY2015 baseline fleet for this Proposed
Determination using the approach described above. Specifically, one half of the fleet volume in
MY2015 has Aerol or Aero2 levels. The remaining vehicles have the potential for additional
improvement. As evidenced by the distribution of drag area values over the various size classes,
the 20 percent improvement from AeroO to Aero2 is an appropriate assumption for this
remaining one half of MY2015 fleet volume.
The efficiencies are different per lumped parameter model classifications, as shown in Table ,
and costs are assigned per those in the Draft TAR.
2-408
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.141 CO2 Efficiency Improvement per 10% Aero Improvement per Vehicle Classification
ALPHA Class
C02 Efficiency Improvement per 10%
Aero Improvement
LPW LRL
2.4%
MPW LRL
2.2%
HPW
1.8%
LPW HRL
2.6%
MPW HRL
2.2%
Truck
2.5%
Costs associated with aero treatments and technologies are equivalent to those used in the
Draft TAR except for updates to 2015 dollars. The aero costs are shown below in Table 2.142.
Table 2.142 Costs for Aero Technologies (dollar values in 2015$)
Tech
Cost type
DMC: base
year cost
IC:
complexity
DMC: learning
cu rve
IC: nearterm
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
Passive aero
DMC
$44
24
$42
$41
$41
$40
$39
$39
$38
$38
$37
Passive aero
IC
Low2
2018
$11
$11
$8
$8
$8
$8
$8
$8
$8
Passive aero
TC
$53
$52
$49
$48
$48
$47
$47
$46
$46
Active aero
DMC
$132
24
$126
$124
$122
$120
$118
$116
$115
$113
$112
Active aero
IC
Med2
2024
$51
$51
$51
$51
$51
$50
$50
$50
$38
Active aero
TC
$177
$175
$173
$170
$168
$167
$165
$163
$149
Passive+Active
TC
$230
$227
$222
$219
$216
$214
$212
$209
$195
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
One comment was received, from FCA, which stated that the costs for aerodynamic
technology is too low. EPA believes that a 10 percent improved CdA can be achieved through
the application of some commonly used aerodynamic treatments. For example, bumper
modifications, wheel deflectors, rear spoiler and underbody cover are enough to provide a 10
percent reduction in aerodynamic drag for some vehicles. This is represented by a cost estimate
of $44.00 and Low2 indirect costs as shown in Table 2.41. EPA believes that a 20 percent
improvement in CdA has been shown to be achieved by ancillary aerodynamic technologies, such
as grille shutters and changes to vehicle exterior design. This is represented by a cost estimate of
$132 and Medium2 Indirect Cost.
2.3.4.5 Tires: Data and Assumptions for this Assessment
In the Draft TAR, EPA considered two levels of low rolling resistance technology: LRRT1
and LRRT2. The first level, LRRT1, was defined as a 10 percent reduction in rolling resistance
from a base (null technology) tire, made possible by methods such as increased tire diameter and
sidewall stiffness and reduced aspect ratios (coupled with reduction in rotational inertia). The
second level, LRRT2, was defined as a 20 percent reduction in rolling resistance from a base tire.
LRRT2 was associated with more advanced approaches such as use of advanced materials and
complete tire redesign. As discussed in the Draft TAR, the 2015 NAS report generally supported
the cost, effectiveness, and feasibility assumptions for both a 10 and 20 percent reduction in
rolling resistance as being appropriate for the 2020 to 2025 time frame.
2-409
-------
Technology Cost, Effectiveness, and Lead Time Assessment
In Chapter 2.2.6 of this TSD, EPA reviewed the current state of low rolling resistance tire
technology and considered developments and trends relating to the feasibility of achieving these
levels of reduction. This review showed that tire manufacturers are aggressively pursuing rolling
resistance technology, that tires exist today that are achieving these levels of reduction, and that
manufacturers are increasingly specifying such tires as original equipment. Although there is
some evidence that consumers have associated low rolling resistance technology with lower
traction, there is also evidence that tire designers have a significant degree of control over the
relationship between the two attributes, and that tires have been designed that are capable of
delivering both.
At the time of the FRM, EPA met with a number of the largest tire suppliers in the United
States to analyze the feasibility and cost for LRRT2. The suppliers were generally optimistic
about the ability to reduce tire rolling resistance 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 then-current tires by MY2017. As such, for the
FRM analysis, EPA assumed LRRT2 would be initially available in MY2017, but not
widespread in the marketplace until MYs 2022-2023. In alignment with that timeframe for
introducing new technology, EPA has maintained the Draft TAR's limitation of the phase-in
schedule to 75 percent of a manufacturer's fleet in 2021, allowing complete application (100
percent of a manufacturer's fleet) by 2025.
In comments on the Draft TAR, the Alliance and Ford pointed out that "low rolling resistance
tires are increasingly specified by OEMs in new vehicles," yet EPA had not accounted for this
existing penetration of this technology in the baseline fleet. EPA agrees that tire rolling
resistance reductions have been achieved in some MY2015 vehicles relative to the levels in
MY2008-2010 vehicles that were used as the null technology point of reference for the FRM and
Draft TAR. Furthermore, EPA agrees with the commenters that it is appropriate to account for
tire rolling resistance reductions already present in the baseline fleet in order to avoid
overestimating the amount of additional improvement that can be achieved at a given cost.
Therefore, for this Proposed Determination, EPA has estimated the levels of tire rolling
resistance reduction already present in MY2015, and assigned one of three tire rolling resistance
levels to each vehicle in the baseline fleet. The process for determining the levels of LRRT0,
LRRT1, and LRRT2 is described below.
Using the test weight and road load coefficient data submitted to EPA for compliance
certification by manufactures, along the assumptions for brake, hub, and driveline drag described
in Appendix B.2.6, EPA estimated a value for the coefficient of tire rolling resistance (Ctrr) for
each vehicle in the MY2015 fleet.
In this Proposed Determination, LRRT1 remains defined as a 10 percent reduction in rolling
resistance from a base tire, and is estimated to result in a 1.9 percent effectiveness improvement
for all vehicle classes. Similarly, LRRT2 remains defined as a 20 percent reduction in rolling
resistance from a base tire, and is estimated to result in a 3.9 percent effectiveness improvement.
Costs associated with lower rolling resistance tires are equivalent to those used in the Draft
TAR except, updated to 2015 dollars. The LRRT costs are shown below.
2-410
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.143 Costs for Lower Rolling Resistance Tires (dollar values in 2015$)
Tech
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
LRRT1
DMC
$6
1
$6
$6
$6
$6
$6
$6
$6
$6
$6
LRRT1
IC
Low2
2018
$1
$1
$1
$1
$1
$1
$1
$1
$1
LRRT1
TC
$7
$7
$7
$7
$7
$7
$7
$7
$7
LRRT2
DMC
$44
32
$57
$55
$53
$51
$49
$48
$47
$46
$45
LRRT2
IC
Low2
2024
$11
$11
$11
$11
$11
$11
$11
$11
$8
LRRT2
TC
$68
$66
$64
$62
$60
$59
$57
$56
$53
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs; both levels of lower rolling resistance are
incremental to today's baseline tires.
2.3.4.6 Mass Reduction: Data and Assumptions for this Assessment
With several exceptions (which are noted below), for this Proposed Determination analysis,
EPA continues to model mass reduction technology using largely the same assumptions that
were applied in the Draft TAR analysis.
Specifically, EPA has continued to apply the mass reduction cost estimates that were applied
in the Draft TAR, updated to 2015 dollars. These costs continue to be based on the cost curves
that were developed and fully described in the Draft TAR. We have also continued to apply the
effectiveness values that were applied in the Draft TAR analysis. Finally, we have also used the
same method for representing mass reduction in the baseline fleet.
These assumptions and methodologies, and their background, were fully documented in the
Draft TAR. For a detailed discussion of the research, methodologies, cost curves, and other
analysis performed in the development of these assumptions for the Draft TAR, which continue
to be used in the present analysis, please refer to Section 5.3.4.6 of the Draft TAR, "Mass
Reduction: Data and Assumptions for this Assessment," which begins on page 5-365 of the Draft
TAR.
In this TSD, the present chapter is devoted to highlighting the key updates to the
consideration of mass reduction technology that apply uniquely to this Proposed Determination
analysis. Section 2.3.4.6.1 includes a description of specific updates, and discussion of some key
comments received on the Draft TAR that relate to mass reduction. Section 2.3.4.6.2 reports the
mass reduction costs used in OMEGA, updated to 2015 dollars.
2.3.4.6.1 Updates to Mass Reduction for the Current Analysis
Several updates apply to the treatment of mass reduction technology in this Proposed
Determination analysis.
First, as described in Chapter 1, the baseline fleet for the Proposed Determination has been
updated to MY2015. It should therefore be noted that in referencing the Draft TAR
documentation, references to the MY2014 baseline fleet should be understood as representing
the MY2015 baseline fleet when interpreted with reference to the present analysis.
Certain updates have also been made to the way mass reduction is represented for pickup
trucks in the baseline fleet. In the Draft TAR, EPA's analysis assigned levels of mass reduction
specific to each vehicle in the baseline fleet in order to account for variation between current
vehicles in the cost and feasibility of achieving additional mass reduction. This was achieved by
2-411
-------
Technology Cost, Effectiveness, and Lead Time Assessment
comparing the 2008 and 2014 versions of each model according to the sales weighted average
curb weights of the various trim levels after adjusting for changes in size, additional safety
requirements, and drive type. This same methodology was again used for this Proposed
Determination assessment, applied to the updated MY2015 baseline fleet.
Although EPA did not receive specific comments on the characterization of mass reduction
for pickup trucks in the baseline fleet, EPA has refined the tracking of the pickup truck lineages
over time for this Proposed Determination assessment in order to better characterize the cost and
feasibility of additional mass reduction for these vehicles.
Unlike passenger cars, light-duty pickup trucks are produced with a variety of cabin and bed
configurations, and the mix of the configurations produced often varies from year to year. The
model-level approach used in the Draft TAR did not distinguish the change in mass that occurred
due to shifts in the production shares of the various pickup truck configurations from the changes
in mass that occurred within a given configuration. For example, using the Draft TAR approach,
a greater proportion of crew cab configurations in MY2015 would be reflected as an increase in
curb weight from MY2008, even if the MY2015 vehicle was lighter than the corresponding
configuration in MY2008. For this Proposed Determination assessment, EPA has estimated the
amount of mass reduction for pickup trucks in the baseline fleet by comparing curb weights
(with adjustments for size, safety equipment, and drive type) for corresponding cab
configurations in MYs 2008 and 2015, thereby minimizing the influence of shifts in production
shares of the various configurations over that period.
The AAM and Ford commented that EPA had not properly accounted for the amount of mass
reduction already implemented in the 2008 MY baseline fleet. Furthermore, AAM
acknowledged that manufacturers have adopted lightweight materials, but that has not
necessarily resulted in a change in vehicle curb weight due to the addition of other vehicle
features. EPA agrees that in many cases, vehicle manufacturers have adopted lightweight
materials in the 2014 MY fleet used for the Draft TAR analysis and in the 2015 MY fleet used
for the current analysis. For the 2012 FRM, the EPA assumed that all vehicles were starting from
the same potential to reduce mass. EPA's method for both the Draft TAR and this Proposed
Determination considers differences between vehicles in the incremental cost and feasibility of
additional mass reduction.
A comment by AAM addressed mass assessment for 4WD/AWD vehicles. In the Draft TAR,
EPA referred to a study performed for Transport Canada which included the evaluation of mass
differences in AWD vs. 2WD versions of three different vehicle models (Jeep Cherokee, Ford
Fusion and VW Tiguan). The mass differences were 135kg, 72kg, and 78kg respectively for an
average of 95kg or 2091bs. A value of 2001bs was used to provide an adjustment to minimize the
influence of this vehicle characterization difference in the baseline sales weighted curb weight.625
AAM commented that the selection of these three vehicles did not "represent typical 4WD/AWD
systems," and suggested that EPA use a different source, such as the certification database, to
determine the mass increase due to AWD and 4WD systems. EPA disagrees that the mass
impact of AWD/4WD systems is not adequately captured. The Jeep Cherokee and the VW
Tiguan represent one of the largest and fastest growing segments in the light-duty market. While
this weight may under-represent some of the largest 4WD vehicles, it may also over-represent
some of the smallest AWD vehicles. For this Proposed Determination EPA has maintained the
methodology found in the Draft TAR for assessing the mass impact of AWD and 4WD.
2-412
-------
Technology Cost, Effectiveness, and Lead Time Assessment
FCA commented that the effectiveness estimates made by EPA for mass reduction were not
accurate due to the lack of consideration of Equivalent Weight (ETW) class bins and their effect
on fuel economy testing. FCA recommended that EPA adjust its modeling so that mass
reduction benefits are only reflected in changes to ETW. EPA does not agree with this
recommendation. The average mass reduction projected in the Proposed Determination is
approximately 9 percent. This amount of mass reduction will move many vehicles in the fleet
down by one or two ETW bins. EPA's approach of allowing mass reduction in continuous
increments (actually 0.5 percent increments in the OMEGA analysis) does not cause a systemic
underestimation of costs, since cases where manufacturers may be getting less benefit from mass
reduction than projected in our analysis would be offset by other cases where manufacturers
may be getting more benefit.
2.3.4.6.2 Mass Reduction Costs used in OMEGA
The tables below show an excerpt of the mass reduction costs used in OMEGA. The costs
presented here are equivalent to those used in the Draft TAR, updated to 2015 dollars. One
notable exception is the expansion in the number of vehicle types relative to the Draft TAR
analysis. We discuss the new vehicle types in Section 2.3.1.4 of this TSD. There are 8 tables that
follow, with the first four showing mass reduction costs at 5 percent, then 10 percent, then 15
percent then 20 percent mass reduction for the 24 vehicle types that use the car cost curve. The
next four tables show mass reduction costs at 5 percent, then 10 percent, then 15 percent then 20
percent mass reduction for the 5 vehicle types that use the truck cost curve. The direct
manufacturing costs (DMC), indirect costs (IC, using ICMs) and the total costs (TC) are shown
along with the sales weighted average curb weight of all vehicles mapped into the indicated
vehicle types, the complexity levels used for indirect costs and the learning curve factor used as
discussed in Section 2.3.2.
An important thing to note in the way mass reduction costs are calculated in OMEGA is the
differential nature of the calculation. For example, if we focus on vehicle type 1 and assume that
a baseline vehicle, of vehicle type 1, has 5 percent mass reduction. That vehicle would have a
cost save, relative to null, of $112 (-$112, see Table 2.144, Total Cost (TC) entry for MY2025).
If that vehicle were to move to a 10 percent mass reduction, the cost save at that level would be
$20 (-$20, see Table 2.145, Total Cost (TC) entry for MY2025). However, the incremental cost
for that move, from 5 percent to 10 percent mass reduction, would be (-$20) - (-$120) = $100. In
other words, the cost of moving from 5 percent to 10 percent mass reduction for that vehicle
would be calculated by OMEGA as a $100 cost increase. All costs shown in the mass reduction
cost tables that follow should be taken as relative to the null vehicle. As a result, the cost for 10
percent mass reduction for this example vehicle having 5 percent mass reduction in the baseline,
would be $100 and not -$20.
Table 2.144 Costs for 5 Percent Mass Reduction for Vehicle Types using the Car Cost Curve (2015$)
Vehicle
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Type
type
CurbWt
IC:
complexity
learning
curve
IC: near
term
thru
1
DMC
2772
30
-$162
-$157
-$153
-$149
-$145
-$142
-$139
-$137
-$135
2
DMC
2988
30
-$175
-$169
-$165
-$160
-$157
-$153
-$150
-$148
-$145
3
DMC
3266
30
-$191
-$185
-$180
-$175
-$171
-$168
-$164
-$161
-$159
4
DMC
3323
30
-$195
-$188
-$183
-$178
-$174
-$171
-$167
-$164
-$161
2-413
-------
Technology Cost, Effectiveness, and Lead Time Assessment
5
DMC
3506
30
-$205
-$199
-$193
-$188
-$184
-$180
-$176
-$173
-$170
6
DMC
3554
30
-$208
-$201
-$196
-$191
-$186
-$182
-$179
-$176
-$173
7
DMC
3928
30
-$230
-$223
-$216
-$211
-$206
-$202
-$198
-$194
-$191
10
DMC
3867
30
-$226
-$219
-$213
-$207
-$203
-$198
-$195
-$191
-$188
11
DMC
4433
30
-$260
-$251
-$244
-$238
-$232
-$227
-$223
-$219
-$215
12
DMC
2976
30
-$174
-$169
-$164
-$160
-$156
-$153
-$150
-$147
-$145
13
DMC
3220
30
-$189
-$183
-$177
-$173
-$169
-$165
-$162
-$159
-$156
14
DMC
3328
30
-$195
-$189
-$183
-$179
-$174
-$171
-$167
-$164
-$162
15
DMC
3510
30
-$206
-$199
-$193
-$188
-$184
-$180
-$177
-$173
-$171
16
DMC
3699
30
-$217
-$210
-$204
-$198
-$194
-$190
-$186
-$183
-$180
17
DMC
3768
30
-$221
-$214
-$207
-$202
-$198
-$193
-$190
-$186
-$183
18
DMC
4011
30
-$235
-$227
-$221
-$215
-$210
-$206
-$202
-$198
-$195
19
DMC
4022
30
-$236
-$228
-$221
-$216
-$211
-$206
-$202
-$199
-$195
20
DMC
4453
30
-$261
-$252
-$245
-$239
-$233
-$229
-$224
-$220
-$216
21
DMC
4610
30
-$270
-$261
-$254
-$247
-$242
-$237
-$232
-$228
-$224
23
DMC
5188
30
-$304
-$294
-$286
-$278
-$272
-$266
-$261
-$256
-$252
24
DMC
5678
30
-$333
-$322
-$313
-$305
-$298
-$291
-$286
-$281
-$276
26
DMC
3970
30
-$232
-$225
-$219
-$213
-$208
-$204
-$200
-$196
-$193
27
DMC
4957
30
-$290
-$281
-$273
-$266
-$260
-$254
-$249
-$245
-$241
28
DMC
5328
30
-$312
-$302
-$293
-$286
-$279
-$273
-$268
-$263
-$259
1
IC
Low2
2024
$28
$28
$28
$28
$28
$28
$28
$28
$23
2
IC
Low2
2024
$31
$31
$31
$31
$31
$31
$31
$31
$25
3
IC
Low2
2024
$33
$33
$33
$33
$33
$33
$33
$33
$27
4
IC
Low2
2024
$34
$34
$34
$34
$34
$34
$34
$34
$27
5
IC
Low2
2024
$36
$36
$36
$36
$36
$36
$36
$36
$29
6
IC
Low2
2024
$36
$36
$36
$36
$36
$36
$36
$36
$29
7
IC
Low2
2024
$40
$40
$40
$40
$40
$40
$40
$40
$32
10
IC
Low2
2024
$39
$39
$39
$39
$39
$39
$39
$39
$32
11
IC
Low2
2024
$45
$45
$45
$45
$45
$45
$45
$45
$37
12
IC
Low2
2024
$30
$30
$30
$30
$30
$30
$30
$30
$25
13
IC
Low2
2024
$33
$33
$33
$33
$33
$33
$33
$33
$27
14
IC
Low2
2024
$34
$34
$34
$34
$34
$34
$34
$34
$27
15
IC
Low2
2024
$36
$36
$36
$36
$36
$36
$36
$36
$29
16
IC
Low2
2024
$38
$38
$38
$38
$38
$38
$38
$38
$30
17
IC
Low2
2024
$38
$38
$38
$38
$38
$38
$38
$38
$31
18
IC
Low2
2024
$41
$41
$41
$41
$41
$41
$41
$41
$33
19
IC
Low2
2024
$41
$41
$41
$41
$41
$41
$41
$41
$33
20
IC
Low2
2024
$45
$45
$45
$45
$45
$45
$45
$45
$37
21
IC
Low2
2024
$47
$47
$47
$47
$47
$47
$47
$47
$38
23
IC
Low2
2024
$53
$53
$53
$53
$53
$53
$53
$53
$43
24
IC
Low2
2024
$58
$58
$58
$58
$58
$58
$58
$58
$47
26
IC
Low2
2024
$41
$41
$41
$41
$41
$41
$41
$41
$33
27
IC
Low2
2024
$51
$51
$51
$51
$51
$51
$51
$51
$41
28
IC
Low2
2024
$54
$54
$54
$54
$54
$54
$54
$54
$44
1
TC
-$134
-$129
-$124
-$120
-$117
-$114
-$111
-$109
-$112
2
TC
-$144
-$139
-$134
-$130
-$126
-$123
-$120
-$117
-$121
3
TC
-$158
-$152
-$147
-$142
-$138
-$134
-$131
-$128
-$132
4
TC
-$161
-$154
-$149
-$144
-$140
-$137
-$133
-$130
-$134
5
TC
-$170
-$163
-$157
-$152
-$148
-$144
-$141
-$137
-$141
6
TC
-$172
-$165
-$159
-$154
-$150
-$146
-$143
-$139
-$143
7
TC
-$190
-$183
-$176
-$171
-$166
-$161
-$158
-$154
-$159
10
TC
-$187
-$180
-$173
-$168
-$163
-$159
-$155
-$152
-$156
11
TC
-$214
-$206
-$199
-$193
-$187
-$182
-$178
-$174
-$179
12
TC
-$144
-$138
-$133
-$129
-$126
-$122
-$119
-$117
-$120
13
TC
-$156
-$150
-$144
-$140
-$136
-$132
-$129
-$126
-$130
14
TC
-$161
-$155
-$149
-$145
-$140
-$137
-$133
-$130
-$134
15
TC
-$170
-$163
-$157
-$153
-$148
-$144
-$141
-$138
-$142
16
TC
-$179
-$172
-$166
-$161
-$156
-$152
-$148
-$145
-$149
17
TC
-$182
-$175
-$169
-$164
-$159
-$155
-$151
-$148
-$152
18
TC
-$194
-$186
-$180
-$174
-$169
-$165
-$161
-$157
-$162
19
TC
-$194
-$187
-$180
-$175
-$170
-$165
-$161
-$158
-$162
20
TC
-$215
-$207
-$200
-$193
-$188
-$183
-$179
-$175
-$180
21
TC
-$223
-$214
-$207
-$200
-$195
-$189
-$185
-$181
-$186
2-414
-------
Technology Cost, Effectiveness, and Lead Time Assessment
23
TC
-$251
-$241
-$233
-$225
-$219
-$213
-$208
-$203
-$209
24
TC
-$275
-$264
-$255
-$247
-$240
-$233
-$228
-$223
-$229
26
TC
-$192
-$184
-$178
-$172
-$168
-$163
-$159
-$156
-$160
27
TC
-$240
-$230
-$222
-$215
-$209
-$204
-$199
-$194
-$200
28
TC
-$258
-$248
-$239
-$232
-$225
-$219
-$214
-$209
-$215
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
Table 2.145 Costs for 10 Percent Mass Reduction for Vehicle Types using the Car Cost Curve (2015$)
Vehicle
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Type
type
CurbWt
IC:
complexity
learning
curve
IC: near
term
thru
1
DMC
2772
30
-$134
-$130
-$126
-$123
-$120
-$117
-$115
-$113
-$111
2
DMC
2988
30
-$144
-$140
-$136
-$132
-$129
-$127
-$124
-$122
-$120
3
DMC
3266
30
-$158
-$153
-$148
-$145
-$141
-$138
-$136
-$133
-$131
4
DMC
3323
30
-$161
-$155
-$151
-$147
-$144
-$141
-$138
-$135
-$133
5
DMC
3506
30
-$169
-$164
-$159
-$155
-$152
-$148
-$146
-$143
-$141
6
DMC
3554
30
-$172
-$166
-$161
-$157
-$154
-$150
-$148
-$145
-$142
7
DMC
3928
30
-$190
-$184
-$178
-$174
-$170
-$166
-$163
-$160
-$157
10
DMC
3867
30
-$187
-$181
-$176
-$171
-$167
-$164
-$161
-$158
-$155
11
DMC
4433
30
-$214
-$207
-$201
-$196
-$192
-$188
-$184
-$181
-$178
12
DMC
2976
30
-$144
-$139
-$135
-$132
-$129
-$126
-$124
-$121
-$119
13
DMC
3220
30
-$156
-$151
-$146
-$143
-$139
-$136
-$134
-$131
-$129
14
DMC
3328
30
-$161
-$156
-$151
-$147
-$144
-$141
-$138
-$136
-$133
15
DMC
3510
30
-$170
-$164
-$159
-$155
-$152
-$149
-$146
-$143
-$141
16
DMC
3699
30
-$179
-$173
-$168
-$164
-$160
-$157
-$154
-$151
-$148
17
DMC
3768
30
-$182
-$176
-$171
-$167
-$163
-$160
-$156
-$154
-$151
18
DMC
4011
30
-$194
-$188
-$182
-$178
-$173
-$170
-$167
-$164
-$161
19
DMC
4022
30
-$194
-$188
-$183
-$178
-$174
-$170
-$167
-$164
-$161
20
DMC
4453
30
-$215
-$208
-$202
-$197
-$193
-$189
-$185
-$182
-$178
21
DMC
4610
30
-$223
-$216
-$209
-$204
-$199
-$195
-$191
-$188
-$185
23
DMC
5188
30
-$251
-$243
-$236
-$230
-$224
-$220
-$215
-$212
-$208
24
DMC
5678
30
-$274
-$265
-$258
-$251
-$246
-$240
-$236
-$231
-$228
26
DMC
3970
30
-$192
-$186
-$180
-$176
-$172
-$168
-$165
-$162
-$159
27
DMC
4957
30
-$239
-$232
-$225
-$219
-$214
-$210
-$206
-$202
-$199
28
DMC
5328
30
-$257
-$249
-$242
-$236
-$230
-$226
-$221
-$217
-$214
1
IC
Low2
2024
$113
$113
$113
$113
$113
$113
$113
$113
$91
2
IC
Low2
2024
$122
$122
$122
$122
$122
$122
$122
$122
$98
3
IC
Low2
2024
$133
$133
$133
$133
$133
$133
$133
$133
$108
4
IC
Low2
2024
$136
$136
$136
$136
$136
$136
$136
$136
$110
5
IC
Low2
2024
$143
$143
$143
$143
$143
$143
$143
$143
$116
6
IC
Low2
2024
$145
$145
$145
$145
$145
$145
$145
$145
$117
7
IC
Low2
2024
$160
$160
$160
$160
$160
$160
$160
$160
$129
10
IC
Low2
2024
$158
$158
$158
$158
$158
$158
$158
$158
$127
11
IC
Low2
2024
$181
$181
$181
$181
$181
$181
$181
$181
$146
12
IC
Low2
2024
$122
$122
$122
$122
$122
$122
$122
$122
$98
13
IC
Low2
2024
$132
$132
$132
$132
$132
$132
$132
$132
$106
14
IC
Low2
2024
$136
$136
$136
$136
$136
$136
$136
$136
$110
15
IC
Low2
2024
$143
$143
$143
$143
$143
$143
$143
$143
$116
16
IC
Low2
2024
$151
$151
$151
$151
$151
$151
$151
$151
$122
17
IC
Low2
2024
$154
$154
$154
$154
$154
$154
$154
$154
$124
18
IC
Low2
2024
$164
$164
$164
$164
$164
$164
$164
$164
$132
19
IC
Low2
2024
$164
$164
$164
$164
$164
$164
$164
$164
$133
20
IC
Low2
2024
$182
$182
$182
$182
$182
$182
$182
$182
$147
21
IC
Low2
2024
$188
$188
$188
$188
$188
$188
$188
$188
$152
23
IC
Low2
2024
$212
$212
$212
$212
$212
$212
$212
$212
$171
24
IC
Low2
2024
$232
$232
$232
$232
$232
$232
$232
$232
$187
26
IC
Low2
2024
$162
$162
$162
$162
$162
$162
$162
$162
$131
27
IC
Low2
2024
$203
$203
$203
$203
$203
$203
$203
$203
$163
28
IC
Low2
2024
$218
$218
$218
$218
$218
$218
$218
$218
$176
2-415
-------
Technology Cost, Effectiveness, and Lead Time Assessment
1
TC
-$21
-$16
-$13
-$9
-$7
-$4
-$2
$0
-$20
2
TC
-$22
-$18
-$14
-$10
-$7
-$4
-$2
$0
-$21
3
TC
-$24
-$19
-$15
-$11
-$8
-$5
-$2
$0
-$23
4
TC
-$25
-$20
-$15
-$11
-$8
-$5
-$2
$0
-$24
5
TC
-$26
-$21
-$16
-$12
-$8
-$5
-$2
$0
-$25
6
TC
-$27
-$21
-$16
-$12
-$9
-$5
-$2
$0
-$25
7
TC
-$29
-$23
-$18
-$13
-$9
-$6
-$3
$0
-$28
10
TC
-$29
-$23
-$18
-$13
-$9
-$6
-$3
$0
-$28
11
TC
-$33
-$26
-$20
-$15
-$11
-$7
-$3
$0
-$32
12
TC
-$22
-$18
-$14
-$10
-$7
-$4
-$2
$0
-$21
13
TC
-$24
-$19
-$15
-$11
-$8
-$5
-$2
$0
-$23
14
TC
-$25
-$20
-$15
-$11
-$8
-$5
-$2
$0
-$24
15
TC
-$26
-$21
-$16
-$12
-$8
-$5
-$2
$0
-$25
16
TC
-$28
-$22
-$17
-$13
-$9
-$5
-$2
$0
-$26
17
TC
-$28
-$22
-$17
-$13
-$9
-$6
-$2
$0
-$27
18
TC
-$30
-$24
-$18
-$14
-$10
-$6
-$3
$0
-$29
19
TC
-$30
-$24
-$18
-$14
-$10
-$6
-$3
$0
-$29
20
TC
-$33
-$26
-$20
-$15
-$11
-$7
-$3
$0
-$32
21
TC
-$34
-$27
-$21
-$16
-$11
-$7
-$3
$0
-$33
23
TC
-$39
-$31
-$24
-$18
-$12
-$8
-$3
$0
-$37
24
TC
-$42
-$34
-$26
-$19
-$14
-$8
-$4
$0
-$40
26
TC
-$30
-$23
-$18
-$14
-$10
-$6
-$3
$0
-$28
27
TC
-$37
-$29
-$23
-$17
-$12
-$7
-$3
$0
-$35
28
TC
-$40
-$31
-$24
-$18
-$13
-$8
-$4
$0
-$38
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
Table 2.146 Costs for 15 Percent Mass Reduction for Vehicle Types using the Car Cost Curve (2015$)
Vehicle
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Type
type
CurbWt
IC:
complexity
learning
curve
IC: near
term
thru
1
DMC
2772
30
-$34
-$32
-$32
-$31
-$30
-$29
-$29
-$28
-$28
2
DMC
2988
30
-$36
-$35
-$34
-$33
-$32
-$32
-$31
-$30
-$30
3
DMC
3266
30
-$39
-$38
-$37
-$36
-$35
-$35
-$34
-$33
-$33
4
DMC
3323
30
-$40
-$39
-$38
-$37
-$36
-$35
-$35
-$34
-$33
5
DMC
3506
30
-$42
-$41
-$40
-$39
-$38
-$37
-$36
-$36
-$35
6
DMC
3554
30
-$43
-$42
-$40
-$39
-$38
-$38
-$37
-$36
-$36
7
DMC
3928
30
-$47
-$46
-$45
-$44
-$43
-$42
-$41
-$40
-$39
10
DMC
3867
30
-$47
-$45
-$44
-$43
-$42
-$41
-$40
-$39
-$39
11
DMC
4433
30
-$54
-$52
-$50
-$49
-$48
-$47
-$46
-$45
-$44
12
DMC
2976
30
-$36
-$35
-$34
-$33
-$32
-$32
-$31
-$30
-$30
13
DMC
3220
30
-$39
-$38
-$37
-$36
-$35
-$34
-$33
-$33
-$32
14
DMC
3328
30
-$40
-$39
-$38
-$37
-$36
-$35
-$35
-$34
-$33
15
DMC
3510
30
-$42
-$41
-$40
-$39
-$38
-$37
-$36
-$36
-$35
16
DMC
3699
30
-$45
-$43
-$42
-$41
-$40
-$39
-$38
-$38
-$37
17
DMC
3768
30
-$46
-$44
-$43
-$42
-$41
-$40
-$39
-$38
-$38
18
DMC
4011
30
-$48
-$47
-$46
-$44
-$43
-$42
-$42
-$41
-$40
19
DMC
4022
30
-$49
-$47
-$46
-$45
-$44
-$43
-$42
-$41
-$40
20
DMC
4453
30
-$54
-$52
-$51
-$49
-$48
-$47
-$46
-$45
-$45
21
DMC
4610
30
-$56
-$54
-$52
-$51
-$50
-$49
-$48
-$47
-$46
23
DMC
5188
30
-$63
-$61
-$59
-$57
-$56
-$55
-$54
-$53
-$52
24
DMC
5678
30
-$69
-$66
-$65
-$63
-$61
-$60
-$59
-$58
-$57
26
DMC
3970
30
-$48
-$46
-$45
-$44
-$43
-$42
-$41
-$40
-$40
27
DMC
4957
30
-$60
-$58
-$56
-$55
-$54
-$53
-$51
-$51
-$50
28
DMC
5328
30
-$64
-$62
-$61
-$59
-$58
-$56
-$55
-$54
-$53
1
IC
Low2
2024
$255
$255
$255
$255
$255
$255
$255
$255
$206
2
IC
Low2
2024
$275
$275
$275
$275
$275
$275
$275
$275
$222
3
IC
Low2
2024
$300
$300
$300
$300
$300
$300
$300
$300
$242
4
IC
Low2
2024
$305
$305
$305
$305
$305
$305
$305
$305
$246
2-416
-------
Technology Cost, Effectiveness, and Lead Time Assessment
5
IC
Low2
2024
$322
$322
$322
$322
$322
$322
$322
$322
$260
6
IC
Low2
2024
$327
$327
$327
$327
$327
$327
$327
$327
$263
7
IC
Low2
2024
$361
$361
$361
$361
$361
$361
$361
$361
$291
10
IC
Low2
2024
$355
$355
$355
$355
$355
$355
$355
$355
$287
11
IC
Low2
2024
$407
$407
$407
$407
$407
$407
$407
$407
$329
12
IC
Low2
2024
$274
$274
$274
$274
$274
$274
$274
$274
$221
13
IC
Low2
2024
$296
$296
$296
$296
$296
$296
$296
$296
$239
14
IC
Low2
2024
$306
$306
$306
$306
$306
$306
$306
$306
$247
15
IC
Low2
2024
$323
$323
$323
$323
$323
$323
$323
$323
$260
16
IC
Low2
2024
$340
$340
$340
$340
$340
$340
$340
$340
$274
17
IC
Low2
2024
$346
$346
$346
$346
$346
$346
$346
$346
$279
18
IC
Low2
2024
$369
$369
$369
$369
$369
$369
$369
$369
$297
19
IC
Low2
2024
$370
$370
$370
$370
$370
$370
$370
$370
$298
20
IC
Low2
2024
$409
$409
$409
$409
$409
$409
$409
$409
$330
21
IC
Low2
2024
$424
$424
$424
$424
$424
$424
$424
$424
$342
23
IC
Low2
2024
$477
$477
$477
$477
$477
$477
$477
$477
$385
24
IC
Low2
2024
$522
$522
$522
$522
$522
$522
$522
$522
$421
26
IC
Low2
2024
$365
$365
$365
$365
$365
$365
$365
$365
$294
27
IC
Low2
2024
$456
$456
$456
$456
$456
$456
$456
$456
$368
28
IC
Low2
2024
$490
$490
$490
$490
$490
$490
$490
$490
$395
1
TC
$221
$222
$223
$224
$225
$225
$226
$227
$178
2
TC
$239
$240
$241
$242
$242
$243
$244
$244
$192
3
TC
$261
$262
$263
$264
$265
$266
$266
$267
$209
4
TC
$265
$267
$268
$269
$269
$270
$271
$272
$213
5
TC
$280
$281
$282
$283
$284
$285
$286
$286
$225
6
TC
$284
$285
$286
$287
$288
$289
$290
$290
$228
7
TC
$314
$315
$316
$318
$319
$319
$320
$321
$252
10
TC
$309
$310
$311
$313
$314
$314
$315
$316
$248
11
TC
$354
$356
$357
$358
$359
$360
$361
$362
$284
12
TC
$238
$239
$240
$241
$241
$242
$243
$243
$191
13
TC
$257
$258
$259
$260
$261
$262
$263
$263
$206
14
TC
$266
$267
$268
$269
$270
$271
$271
$272
$213
15
TC
$280
$282
$283
$284
$285
$285
$286
$287
$225
16
TC
$295
$297
$298
$299
$300
$301
$302
$302
$237
17
TC
$301
$302
$304
$305
$306
$306
$307
$308
$242
18
TC
$320
$322
$323
$324
$325
$326
$327
$328
$257
19
TC
$321
$323
$324
$325
$326
$327
$328
$329
$258
20
TC
$355
$357
$359
$360
$361
$362
$363
$364
$286
21
TC
$368
$370
$371
$373
$374
$375
$376
$377
$296
23
TC
$414
$416
$418
$419
$421
$422
$423
$424
$333
24
TC
$453
$455
$457
$459
$460
$462
$463
$464
$364
26
TC
$317
$318
$320
$321
$322
$323
$324
$324
$255
27
TC
$396
$398
$399
$401
$402
$403
$404
$405
$318
28
TC
$425
$427
$429
$431
$432
$433
$434
$435
$342
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
Table 2.147 Costs for 20 Percent Mass Reduction for Vehicle Types using the Car Cost Curve (2015$)
Vehicle
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Type
type
CurbWt
IC:
complexity
learning
curve
IC: near
term
thru
1
DMC
2772
30
$114
$110
$107
$104
$102
$100
$98
$96
$94
2
DMC
2988
30
$123
$119
$115
$112
$110
$107
$105
$103
$102
3
DMC
3266
30
$134
$130
$126
$123
$120
$117
$115
$113
$111
4
DMC
3323
30
$136
$132
$128
$125
$122
$119
$117
$115
$113
5
DMC
3506
30
$144
$139
$135
$132
$129
$126
$124
$121
$119
6
DMC
3554
30
$146
$141
$137
$134
$130
$128
$125
$123
$121
7
DMC
3928
30
$161
$156
$151
$148
$144
$141
$138
$136
$134
10
DMC
3867
30
$159
$153
$149
$145
$142
$139
$136
$134
$132
2-417
-------
Technology Cost, Effectiveness, and Lead Time Assessment
11
DMC
4433
30
$182
$176
$171
$167
$163
$159
$156
$153
$151
12
DMC
2976
30
$122
$118
$115
$112
$109
$107
$105
$103
$101
13
DMC
3220
30
$132
$128
$124
$121
$118
$116
$113
$111
$110
14
DMC
3328
30
$136
$132
$128
$125
$122
$120
$117
$115
$113
15
DMC
3510
30
$144
$139
$135
$132
$129
$126
$124
$121
$119
16
DMC
3699
30
$152
$147
$143
$139
$136
$133
$130
$128
$126
17
DMC
3768
30
$154
$150
$145
$142
$138
$135
$133
$130
$128
18
DMC
4011
30
$164
$159
$155
$151
$147
$144
$141
$139
$136
19
DMC
4022
30
$165
$160
$155
$151
$148
$145
$142
$139
$137
20
DMC
4453
30
$183
$177
$172
$167
$163
$160
$157
$154
$151
21
DMC
4610
30
$189
$183
$178
$173
$169
$166
$162
$160
$157
23
DMC
5188
30
$213
$206
$200
$195
$190
$186
$183
$180
$177
24
DMC
5678
30
$233
$225
$219
$213
$208
$204
$200
$196
$193
26
DMC
3970
30
$163
$158
$153
$149
$146
$143
$140
$137
$135
27
DMC
4957
30
$203
$197
$191
$186
$182
$178
$175
$172
$169
28
DMC
5328
30
$218
$211
$205
$200
$196
$191
$188
$184
$181
1
IC
Low2
2024
$453
$453
$453
$453
$453
$453
$453
$453
$365
2
IC
Low2
2024
$488
$488
$488
$488
$488
$488
$488
$488
$394
3
IC
Low2
2024
$534
$534
$534
$534
$534
$534
$534
$534
$431
4
IC
Low2
2024
$543
$543
$543
$543
$543
$543
$543
$543
$438
5
IC
Low2
2024
$573
$573
$573
$573
$573
$573
$573
$573
$462
6
IC
Low2
2024
$581
$581
$581
$581
$581
$581
$581
$581
$468
7
IC
Low2
2024
$642
$642
$642
$642
$642
$642
$642
$642
$518
10
IC
Low2
2024
$632
$632
$632
$632
$632
$632
$632
$632
$510
11
IC
Low2
2024
$724
$724
$724
$724
$724
$724
$724
$724
$584
12
IC
Low2
2024
$486
$486
$486
$486
$486
$486
$486
$486
$392
13
IC
Low2
2024
$526
$526
$526
$526
$526
$526
$526
$526
$424
14
IC
Low2
2024
$544
$544
$544
$544
$544
$544
$544
$544
$439
15
IC
Low2
2024
$574
$574
$574
$574
$574
$574
$574
$574
$463
16
IC
Low2
2024
$604
$604
$604
$604
$604
$604
$604
$604
$488
17
IC
Low2
2024
$616
$616
$616
$616
$616
$616
$616
$616
$497
18
IC
Low2
2024
$656
$656
$656
$656
$656
$656
$656
$656
$529
19
IC
Low2
2024
$657
$657
$657
$657
$657
$657
$657
$657
$530
20
IC
Low2
2024
$728
$728
$728
$728
$728
$728
$728
$728
$587
21
IC
Low2
2024
$753
$753
$753
$753
$753
$753
$753
$753
$608
23
IC
Low2
2024
$848
$848
$848
$848
$848
$848
$848
$848
$684
24
IC
Low2
2024
$928
$928
$928
$928
$928
$928
$928
$928
$748
26
IC
Low2
2024
$649
$649
$649
$649
$649
$649
$649
$649
$523
27
IC
Low2
2024
$810
$810
$810
$810
$810
$810
$810
$810
$653
28
IC
Low2
2024
$871
$871
$871
$871
$871
$871
$871
$871
$702
1
TC
$567
$563
$560
$557
$555
$553
$551
$549
$460
2
TC
$611
$607
$603
$601
$598
$596
$594
$592
$496
3
TC
$668
$663
$660
$657
$654
$651
$649
$647
$542
4
TC
$679
$675
$671
$668
$665
$662
$660
$658
$551
5
TC
$717
$712
$708
$705
$702
$699
$696
$694
$581
6
TC
$726
$722
$718
$714
$711
$708
$706
$704
$589
7
TC
$803
$798
$793
$790
$786
$783
$780
$778
$651
10
TC
$790
$785
$781
$777
$774
$771
$768
$766
$641
11
TC
$906
$900
$895
$891
$887
$884
$881
$878
$735
12
TC
$608
$604
$601
$598
$596
$593
$591
$589
$494
13
TC
$658
$654
$650
$647
$644
$642
$640
$638
$534
14
TC
$680
$676
$672
$669
$666
$663
$661
$659
$552
15
TC
$718
$713
$709
$705
$702
$700
$697
$695
$582
16
TC
$756
$751
$747
$743
$740
$737
$735
$732
$613
17
TC
$770
$765
$761
$757
$754
$751
$748
$746
$625
18
TC
$820
$815
$810
$806
$803
$800
$797
$794
$665
19
TC
$822
$817
$812
$808
$805
$802
$799
$796
$667
20
TC
$910
$904
$899
$895
$891
$888
$885
$882
$738
21
TC
$942
$936
$931
$927
$923
$919
$916
$913
$764
23
TC
$1,061
$1,054
$1,048
$1,043
$1,038
$1,034
$1,031
$1,027
$860
24
TC
$1,161
$1,153
$1,147
$1,141
$1,136
$1,132
$1,128
$1,124
$942
26
TC
$811
$806
$802
$798
$794
$791
$789
$786
$658
27
TC
$1,013
$1,007
$1,001
$996
$992
$988
$985
$982
$822
2-418
-------
Technology Cost, Effectiveness, and Lead Time Assessment
| 28 | TC | | | $1,089 | $1,082 | $1,076 | $1,071 | $1,066 | $1,062 | $1,058 | $1,055 | $884 |
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
Table 2.148 Costs for 5 Percent Mass Reduction for Vehicle Types using the Truck Cost Curve (2015$)
Vehicle
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Type
type
CurbWt
IC:
complexity
learning
curve
IC: near
term
thru
8
DMC
4016
30
-$216
-$209
-$203
-$198
-$193
-$189
-$185
-$182
-$179
9
DMC
4976
30
-$267
-$259
-$251
-$245
-$239
-$234
-$230
-$226
-$222
22
DMC
4214
30
-$226
-$219
-$213
-$208
-$203
-$198
-$195
-$191
-$188
25
DMC
5106
30
-$274
-$266
-$258
-$251
-$246
-$240
-$236
-$232
-$228
29
DMC
4883
30
-$262
-$254
-$247
-$240
-$235
-$230
-$225
-$221
-$218
8
IC
Low2
2024
$62
$62
$62
$62
$62
$62
$62
$62
$50
9
IC
Low2
2024
$77
$77
$77
$77
$77
$77
$77
$77
$62
22
IC
Low2
2024
$65
$65
$65
$65
$65
$65
$65
$65
$53
25
IC
Low2
2024
$79
$79
$79
$79
$79
$79
$79
$79
$64
29
IC
Low2
2024
$75
$75
$75
$75
$75
$75
$75
$75
$61
8
TC
-$154
-$147
-$141
-$136
-$131
-$127
-$123
-$120
-$129
9
TC
-$191
-$182
-$175
-$168
-$163
-$157
-$153
-$149
-$160
22
TC
-$161
-$154
-$148
-$142
-$138
-$133
-$130
-$126
-$135
25
TC
-$196
-$187
-$179
-$173
-$167
-$162
-$157
-$153
-$164
29
TC
-$187
-$179
-$171
-$165
-$160
-$155
-$150
-$146
-$157
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
Table 2.149 Costs for 10 Percent Mass Reduction for Vehicle Types using the Truck Cost Curve (2015$)
Vehicle
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Type
type
CurbWt
IC:
complexity
learning
curve
IC: near
term
thru
8
DMC
4016
30
$63
$61
$59
$58
$56
$55
$54
$53
$52
9
DMC
4976
30
$78
$76
$73
$72
$70
$68
$67
$66
$65
22
DMC
4214
30
$66
$64
$62
$61
$59
$58
$57
$56
$55
25
DMC
5106
30
$80
$78
$75
$73
$72
$70
$69
$68
$66
29
DMC
4883
30
$77
$74
$72
$70
$69
$67
$66
$65
$64
8
IC
Low2
2024
$248
$248
$248
$248
$248
$248
$248
$248
$201
9
IC
Low2
2024
$307
$307
$307
$307
$307
$307
$307
$307
$249
22
IC
Low2
2024
$260
$260
$260
$260
$260
$260
$260
$260
$211
25
IC
Low2
2024
$315
$315
$315
$315
$315
$315
$315
$315
$256
29
IC
Low2
2024
$302
$302
$302
$302
$302
$302
$302
$302
$245
8
TC
$311
$309
$307
$306
$304
$303
$302
$301
$253
9
TC
$385
$383
$381
$379
$377
$376
$374
$373
$314
22
TC
$326
$324
$322
$321
$319
$318
$317
$316
$266
25
TC
$395
$393
$391
$389
$387
$385
$384
$383
$322
29
TC
$378
$376
$374
$372
$370
$369
$367
$366
$308
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
Table 2.150 Costs for 15 Percent Mass Reduction for Vehicle Types using the Truck Cost Curve (2015$)
Vehicle
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Type
type
CurbWt
learning
IC:
cu rve
complexity
IC: near
term
thru
8
DMC
4016
30
$528
$511
$497
$484
$473
$463
$454
$446
$438
2-419
-------
Technology Cost, Effectiveness, and Lead Time Assessment
9
DMC
4976
30
$655
$634
$616
$600
$586
$574
$563
$552
$543
22
DMC
4214
30
$555
$537
$521
$508
$496
$486
$477
$468
$460
25
DMC
5106
30
$672
$650
$632
$616
$601
$589
$577
$567
$557
29
DMC
4883
30
$643
$622
$604
$589
$575
$563
$552
$542
$533
8
IC
Low2
2024
$558
$558
$558
$558
$558
$558
$558
$558
$453
9
IC
Low2
2024
$691
$691
$691
$691
$691
$691
$691
$691
$561
22
IC
Low2
2024
$586
$586
$586
$586
$586
$586
$586
$586
$475
25
IC
Low2
2024
$709
$709
$709
$709
$709
$709
$709
$709
$576
29
IC
Low2
2024
$678
$678
$678
$678
$678
$678
$678
$678
$550
8
TC
$1,086
$1,069
$1,055
$1,042
$1,031
$1,021
$1,012
$1,004
$891
9
TC
$1,346
$1,325
$1,307
$1,291
$1,277
$1,265
$1,254
$1,244
$1,104
22
TC
$1,140
$1,122
$1,107
$1,094
$1,082
$1,072
$1,062
$1,054
$935
25
TC
$1,381
$1,360
$1,341
$1,325
$1,311
$1,298
$1,287
$1,276
$1,133
29
TC
$1,321
$1,300
$1,283
$1,267
$1,254
$1,241
$1,231
$1,221
$1,083
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
Table 2.151 Costs for 20 Percent Mass Reduction for Vehicle Types using the Truck Cost Curve (2015$)
Vehicle
Cost
DMC:
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Type
type
CurbWt
IC:
complexity
learning
cu rve
IC: near
term
thru
8
DMC
4016
30
$1,115
$1,079
$1,049
$1,022
$998
$977
$958
$941
$925
9
DMC
4976
30
$1,382
$1,337
$1,299
$1,266
$1,237
$1,211
$1,187
$1,166
$1,146
22
DMC
4214
30
$1,170
$1,133
$1,100
$1,072
$1,048
$1,026
$1,006
$988
$971
25
DMC
5106
30
$1,418
$1,372
$1,333
$1,299
$1,269
$1,242
$1,218
$1,196
$1,176
29
DMC
4883
30
$1,356
$1,312
$1,275
$1,242
$1,214
$1,188
$1,165
$1,144
$1,125
8
IC
Low2
2024
$992
$992
$992
$992
$992
$992
$992
$992
$805
9
IC
Low2
2024
$1,229
$1,229
$1,229
$1,229
$1,229
$1,229
$1,229
$1,229
$997
22
IC
Low2
2024
$1,041
$1,041
$1,041
$1,041
$1,041
$1,041
$1,041
$1,041
$844
25
IC
Low2
2024
$1,261
$1,261
$1,261
$1,261
$1,261
$1,261
$1,261
$1,261
$1,023
29
IC
Low2
2024
$1,206
$1,206
$1,206
$1,206
$1,206
$1,206
$1,206
$1,206
$978
8
TC
$2,107
$2,071
$2,041
$2,014
$1,990
$1,969
$1,950
$1,933
$1,730
9
TC
$2,611
$2,566
$2,528
$2,495
$2,466
$2,440
$2,416
$2,395
$2,143
22
TC
$2,211
$2,174
$2,141
$2,113
$2,089
$2,066
$2,047
$2,028
$1,815
25
TC
$2,679
$2,633
$2,594
$2,560
$2,530
$2,504
$2,480
$2,458
$2,200
29
TC
$2,562
$2,518
$2,481
$2,449
$2,420
$2,394
$2,371
$2,350
$2,103
Note: DMC=direct manufacturing cost; IC=indirect cost; TC=total cost.
2-420
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2.3.4.7 Other Vehicle Technologies
2.3.4.7.1 Electrified Power Steering: Data and Assumptions for this Assessment
For the 2017-2025 final rule and Draft TAR, EPA estimated a 1 to 2 percent effectiveness for
electrified power steering in light duty vehicles, based on the 2015 NAS report, Sierra Research
Report and confidential OEM data. The 2010 Ricardo study also confirmed this estimate. EPA
have reviewed these effectiveness estimates and found them to be accurate, thus they have been
retained for this Proposed Determination. There were no public comments received with
supporting data that would provide basis for a change to the cost or effectiveness estimates for
this technology, nor has EPA found additional information that supports such a change since the
Draft TAR.
Costs associated with electric power steering are equivalent to those used in the Draft TAR,
updated to 2015 dollars. The electric power steering costs incremental to hydraulic power
steering are shown below.
Table 2.152 Costs for Electric Power Steering (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$99
24
$94
$92
$91
$89
$88
$87
$85
$84
$83
IC
Low2
2018
$24
$24
$19
$19
$19
$19
$19
$19
$19
TC
$118
$116
$110
$108
$107
$106
$104
$103
$102
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.7.2 Improved Accessories: Data and Assumptions for this Assessment
There were no public comments received with supporting data that would provide basis for a
change to the cost or effectiveness estimates for this technology, nor has EPA found additional
information that supports such a change since the Draft TAR.
In MYs 2017-2025 final rule and the Draft TAR, EPA used an effectiveness value in the
range of 1 to 2 percent.
As in the Draft TAR, for this Proposed Determination assessment, EPA considered two levels
of improved accessories. Level 1 of this technology (IACC1) 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. EPA used effectiveness values in the 1.2 to 1.8 percent range, varying with
vehicle subclass.
Costs associated with improved accessories are equivalent to those used in the Draft TAR,
updated to 2015 dollars. The improved accessory costs (levels 1 and 2) are shown below. Cost is
higher for improved accessories level 2 due to the inclusion of a higher efficiency alternator and
a mild level of regeneration, hence the $40 to $50 higher cost. Both improved accessory costs
are incremental to the baseline.
Table 2.153 Costs for Improved Accessories Level 1 (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$80
24
$77
$75
$74
$73
$71
$70
$69
$69
$68
IC
Low2
2018
$19
$19
$15
$15
$15
$15
$15
$15
$15
2-421
-------
Technology Cost, Effectiveness, and Lead Time Assessment
TC | | | $96 | $95 | $89 | $88 | $87 | $86 | $85 | $84 | $83
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Table 2.154 Costs for Improved Accessories Level 2 (dollar values in 2015$)
Cost type
DMC: base year cost
IC: complexity
DMC: learning curve
IC: nearterm thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$130
24
$124
$122
$119
$117
$116
$114
$112
$111
$109
IC
Low2
2018
$31
$31
$25
$25
$25
$25
$25
$25
$25
TC
$155
$153
$144
$142
$140
$139
$137
$136
$134
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.7.3 Secondary Axle Disconnect: Data and Assumptions for this Assessment
The 2017-2025 final rule estimated an effectiveness improvement of 1.0 to 1.5 percent for
axle disconnect, which was refined to 1.2 to 1.4 percent based on the 2011 Ricardo report.
EPA has reviewed the cost and effectiveness figures used in the Draft TAR. There were no
public comments received with supporting data that would provide basis for a change to the cost
or effectiveness estimates for this technology, nor has EPA found additional information that
supports such a change since the Draft TAR. EPA is retaining the Draft TAR figures for the
Proposed Determination analysis. The cost associated with secondary axle disconnect is
equivalent to that used in the Draft TAR, updated to 2015 dollars. The costs are shown below.
Table 2.155 Costs for Secondary Axle Disconnect (dollar values in 2015$)
Cost
type
DMC: base year
cost
IC: complexity
DMC: learning
curve
IC: near term thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$88
24
$84
$83
$81
$80
$78
$77
$76
$75
$74
IC
Low2
2018
$21
$21
$17
$17
$17
$17
$17
$17
$17
TC
$105
$104
$98
$97
$95
$94
$93
$92
$91
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.7.4 Low Dras Brakes: Data and Assumptions for this Assessment
The 2017-2025 final rule and Draft TAR estimated the effectiveness of low drag brakes to be
to 0.8 percent. EPA continues to use this estimate for this Proposed Determination analysis
based on the 2011 Ricardo study and the 2015 NAS report.
In comments on the Draft TAR, Toyota commented on several aspects of EPA's low-drag
brake assessment. With respect to the Draft TAR analysis, Toyota commented on the
conclusions regarding the Direct Manufacturing Costs (DMC) and stated that in order to
"calculate such a detailed cost, it must be fixed with a special brake system of that of a specific
supplier."
EPA notes that the DMC for this technology is not meant to represent a single supplier's cost,
but rather an aggregate cost representing all of the changes that can be made to the brake system
to reduce drag, including caliper seal and return rate and rotor and lining changes. For this
Proposed Determination, the conclusions regarding DMC for low-drag brakes have been carried
over from the 2012 FRM and from the Draft TAR.
2-422
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Toyota also commented on EPA's summary of available zero drag brake systems. In response
to these comments, updates have been made to the description of this technology in Chapter
2.2.8.4. Zero-drag brakes are not, however, part of this Proposed Determination analysis.
The cost associated with low drag brakes for the present analysis is equivalent to that used in
the Draft TAR, updated to 2015 dollars. The costs are shown below.
Table 2.156 Costs for Low Drag Brakes (dollar values in 2015$)
Cost
type
DMC: base year
cost
IC: complexity
DMC: learning
curve
IC: near term thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
DMC
$64
1
$64
$64
$64
$64
$64
$64
$64
$64
$64
IC
Low2
2018
$15
$15
$12
$12
$12
$12
$12
$12
$12
TC
$79
$79
$76
$76
$76
$76
$76
$76
$76
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.8 Air Conditioning: Data and Assumptions for this Assessment
Air conditioning (A/C) system technologies include improved hoses, connectors and seals for
leakage control. They also include improved compressors, expansion valves, heat exchangers
and the control of these components for the purposes of improving tailpipe CO2 emissions and
fuel economy as a result of A/C use.
The Draft TAR generated extensive public comment relating to the A/C credit program, credit
application procedures, the AC 17 test procedure, testing requirements, and similar topics. Since
these comments were concerned with off-cycle credit opportunities and details of the compliance
process, and not with cost or effectiveness inputs to the Proposed Determination analysis, they
are addressed in Chapter 2.2.9 (Air Conditioning Efficiency and Leakage Credits).
For this Proposed Determination analysis, EPA is continuing to use the cost and effectiveness
estimates that were used in the Draft TAR analysis, updated to 2015 dollars. For more
information on these estimates, see Section 5.1 of the 2012 TSD.
Table 2.157 Costs for A/C Controls (dollar values in 2015$)
Cost type
2017
2018
2019
2020
2021
2022
2023
2024
2025
TC
$94
$120
$138
$145
$158
$155
$148
$146
$143
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.9 Additional Off-cycle Credits and Costs
In past analyses, EPA has included technology costs and additional off-cycle credits for active
aerodynamics (Aero2) and stop-start. While the off-cycle credits of these technologies were
never considered when determining the feasibility of the standards, as air conditioning credits
were, they have been considered to be relatively cost effective and expected to be widely used to
comply. As a result, past analyses have shown considerable penetration of these technologies in
our control case OMEGA runs.
Beyond off-cycle credits provided for active aero and stop-start, there are other technologies
for which EPA provides off-cycle credits. Those technologies are included in what EPA calls the
"off-cycle menu" and were codified in the 2012 FRM which specifies the level of credit
2-423
-------
Technology Cost, Effectiveness, and Lead Time Assessment
available to those technologies without further demonstration. The off-cycle menu is shown in
Table 2.158 and the program is described in more detail, along with a discussion of credits
generated by manufacturers in MY2015, in TSD Chapter 2.2.10.
Table 2.158 Off-Cycle "Menu" Technologies and Credits for Cars & Light Trucks
Technology
gCCh/mi Credit for Cars
gCC>2/mi Credit for Trucks
High efficiency exterior lights
1.0
1.0
Waster heat recovery
0.7
0.7
Solar panels for battery charging
3.3
3.3
Solar panels for active cabin ventilation & battery charging
2.5
2.5
Active aerodynamic improvements (Aero2)
0.6
1.0
Stop-start with heater circulation system
2.5
4.4
Stop-start without heater circulation system
1.5
2.9
Active transmission warm-up
1.5
3.2
Active engine warm-up
1.5
3.2
Solar/thermal control
up to 3.0
up to 4.3
Until now, we have not included the use of these menu off-cycle technologies in our OMEGA
modeling since we did not have estimates of their costs. In comments on the Draft TAR, several
auto industry commenters suggested that they plan to expand their use of off-cycle credits,
including the menu technologies, in the coming years. These commenters even suggested that
EPA remove the current 10 gram/mile cap on use of menu technologies, which seems an
indication that manufacturers appear to be planning to maximize their use of these technologies
throughout their fleets. In EPA's latest GHG Manufacturer Performance Report for MY2015,
auto manufacturers used a fleet-wide average 3.0 gCCh/mi of off-cycle menu credits. This
makes clear that these credits are important to manufacturers and are, apparently, cost effective
approaches to controlling GHGs.
For this Proposed Determination analysis, we are incorporating as technology options into
OMEGA the use of additional off-cycle credit opportunities. Given that these credits are an
available compliance option, EPA considers it reasonable to assess their potential use in
considering the appropriateness (including feasibility and cost) of the 2022-2025MY standards.
The approach being used in this Proposed Determination is not to focus on particular off-cycle
technologies or their costs and credits, but rather to estimate the additional costs and credits
based on the costs estimated by OMEGA. Specifically, we used the "single OEM" or "Perfect
Trading" OMEGA run presented in the Draft TAR as a sensitivity (see Draft TAR Chapter
12.1.2). That run estimates the impacts of perfect trading amongst OEMs since the fleet is run as
a single OEM. This is a "best case" or least-cost scenario. Using the results of that run, for the
Control case in 2025, the costs associated with achieving the reference case targets of roughly
237 gC02/mi were $442, and the costs of the control case targets of roughly 199 gC02/mi were
$1,307 (see Table 2.159). Note that both of these costs and the CO2 values noted are OMEGA-
core values and, as such, make no consideration of AJC credits, which is what we want for this
exercise. Using the results of this "perfect trading" run further, we were able to generate the cost
per gC02/mi value of $34 and applied a 30 percent premium resulting in a $45 (2013$) cost for
each gram of C02 reduced. This cost was applied to an "off-cycle technology level 1" credit of
1.5 gC02/mi. For an off-cycle level 2 credit of 3 g/mi, we applied a 60 percent premium to the
$34 value to arrive at a $55/gC02/mi value (2013$) as shown in Table 2.160.
2-424
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.159 Cost per gCCh/mi within the Indicated Ranges for the Perfect Trading Sensitivity Run Presented
in the Draft TAR (2013$)
Target CO2
Delta CO2
$/vehicle
Delta Cost
$/gC02/mi
237.2
$442
230.0
7.15
$550
$108
$15
220.0
9.98
$726
$176
00
1
¦uy
210.0
9.98
$972
$246
$25
200.1
9.97
$1,277
$305
1
m
¦uy
199.2
0.9
$1,307
1
m
¦uy
$34
Table 2.1602 Basis for Off-cycle Credit Values and Costs used in OMEGA
Off-cycle "Technology"
Valued at
(in 2013$)
Credit Value
DMC
(in
2015$)
OC1
$45/gC02/mi
1.5 gCCh/mi
$69
OC2
$55/gCC>2/mi
3.0 gCCh/mi
$170
We have applied learning curve 29 to these costs and a low complexity markup to arrive at the
costs shown in Table 2.161.
Table 2.161 Costs for Off-Cycle Technologies Level 1 & 2 (dollar values in 2015$)
Tech
Cost
type
DMC: base year
cost
IC: complexity
DMC: learning
curve
IC: near term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
OCl
DMC
$69
29
$69
$68
$66
$65
$64
$63
$62
$61
$60
OC2
DMC
$170
29
$170
$166
$162
$159
$156
$154
$151
$149
$147
OCl
IC
Low2
2024
$17
$17
$17
$17
$17
$17
$17
$17
$13
OC2
IC
Low2
2024
$41
$41
$41
$41
$41
$41
$41
$41
$33
OCl
TC
$86
$85
$83
$82
$81
$80
$79
$78
$73
OC2
TC
$211
$207
$203
$200
$197
$195
$192
$190
$180
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
2.3.4.10 Cost Tables for Individual Technologies Not Presented Above
Costs associated with SCR-equipped diesel vehicles are equivalent to those used in the Draft
TAR, updated to 2015 dollars. The costs incremental to the baseline engine configuration for our
different vehicle classes are shown below. These costs are used to characterize technology costs
in the baseline fleet; EPA does not build OMEGA packages using this technology and instead
uses the advanced diesel technology presented below.
Table 2.162 Costs for SCR-equipped Diesel Technology for Different Vehicle Classes (dollar values in 2015$)
Curb Weight
Class
Cost
type
DMC: base
cost
IC:
complexity
DMC: learning
curve
IC: nearterm
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
DMC
$2,531
23
$2,291
$2,255
$2,222
$2,191
$2,162
$2,135
$2,110
$2,086
$2,064
2-425
-------
Technology Cost, Effectiveness, and Lead Time Assessment
2
DMC
$2,531
23
$2,291
$2,255
$2,222
$2,191
$2,162
$2,135
$2,110
$2,086
$2,064
3
DMC
$3,112
23
$2,817
$2,773
$2,732
$2,694
$2,659
$2,626
$2,595
$2,565
$2,537
4
DMC
$3,112
23
$2,817
$2,773
$2,732
$2,694
$2,659
$2,626
$2,595
$2,565
$2,537
5
DMC
$3,112
23
$2,817
$2,773
$2,732
$2,694
$2,659
$2,626
$2,595
$2,565
$2,537
6
DMC
$3,568
23
$3,231
$3,180
$3,133
$3,090
$3,049
$3,011
$2,975
$2,941
$2,909
1
IC
Med2
2018
$969
$968
$724
$723
$722
$721
$720
$719
$719
2
IC
Med2
2018
$969
$968
$724
$723
$722
$721
$720
$719
$719
3
IC
Med2
2018
$1,192
$1,190
$890
$888
$887
$886
$885
$884
$884
4
IC
Med2
2018
$1,192
$1,190
$890
$888
$887
$886
$885
$884
$884
5
IC
Med2
2018
$1,192
$1,190
$890
$888
$887
$886
$885
$884
$884
6
IC
Med2
2018
$1,367
$1,364
$1,020
$1,019
$1,018
$1,016
$1,015
$1,014
$1,013
1
TC
$3,261
$3,223
$2,946
$2,914
$2,884
$2,856
$2,830
$2,805
$2,782
2
TC
$3,261
$3,223
$2,946
$2,914
$2,884
$2,856
$2,830
$2,805
$2,782
3
TC
$4,009
$3,963
$3,622
$3,583
$3,546
$3,512
$3,480
$3,450
$3,421
4
TC
$4,009
$3,963
$3,622
$3,583
$3,546
$3,512
$3,480
$3,450
$3,421
5
TC
$4,009
$3,963
$3,622
$3,583
$3,546
$3,512
$3,480
$3,450
$3,421
6
TC
$4,597
$4,544
$4,153
$4,108
$4,066
$4,027
$3,990
$3,956
$3,923
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Costs associated with advanced diesel vehicles (i.e., Tier 3 compliant) are equivalent to those
used in the Draft TAR, updated to 2015 dollars. The costs incremental to the baseline engine
configuration for our different vehicle classes are shown below. These costs are used when
building OMEGA diesel packages.
Table 2.163 Costs for Advanced Diesel Technology for Different Vehicle Classes (dollar values in 2015$)
Curb Weight
Cost
DMC: base
DMC:
2017
2018
2019
2020
2021
2022
2023
2024
2025
Class
type
cost
IC:
complexity
learning
cu rve
IC: near
term
thru
1
DMC
$2,581
23
$2,337
$2,300
$2,266
$2,235
$2,205
$2,178
$2,152
$2,127
$2,104
2
DMC
$2,581
23
$2,337
$2,300
$2,266
$2,235
$2,205
$2,178
$2,152
$2,127
$2,104
3
DMC
$3,162
23
$2,863
$2,818
$2,776
$2,738
$2,702
$2,668
$2,636
$2,606
$2,578
4
DMC
$3,162
23
$2,863
$2,818
$2,776
$2,738
$2,702
$2,668
$2,636
$2,606
$2,578
5
DMC
$3,162
23
$2,863
$2,818
$2,776
$2,738
$2,702
$2,668
$2,636
$2,606
$2,578
6
DMC
$3,618
23
$3,276
$3,225
$3,177
$3,133
$3,092
$3,053
$3,017
$2,983
$2,950
1
IC
Med2
2018
$988
$987
$738
$737
$736
$735
$734
$734
$733
2
IC
Med2
2018
$988
$987
$738
$737
$736
$735
$734
$734
$733
3
IC
Med2
2018
$1,211
$1,209
$904
$903
$902
$901
$900
$899
$898
4
IC
Med2
2018
$1,211
$1,209
$904
$903
$902
$901
$900
$899
$898
5
IC
Med2
2018
$1,211
$1,209
$904
$903
$902
$901
$900
$899
$898
6
IC
Med2
2018
$1,386
$1,384
$1,034
$1,033
$1,032
$1,031
$1,029
$1,028
$1,027
1
TC
$3,325
$3,287
$3,004
$2,971
$2,941
$2,913
$2,886
$2,861
$2,837
2
TC
$3,325
$3,287
$3,004
$2,971
$2,941
$2,913
$2,886
$2,861
$2,837
3
TC
$4,074
$4,027
$3,680
$3,640
$3,603
$3,568
$3,536
$3,505
$3,476
4
TC
$4,074
$4,027
$3,680
$3,640
$3,603
$3,568
$3,536
$3,505
$3,476
5
TC
$4,074
$4,027
$3,680
$3,640
$3,603
$3,568
$3,536
$3,505
$3,476
6
TC
$4,662
$4,608
$4,211
$4,166
$4,123
$4,084
$4,046
$4,011
$3,978
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
Costs associated with powersplit HEVs are equivalent to those used in the Draft TAR,
updated to 2015 dollars. The costs incremental to the baseline configuration for our different
vehicle classes are shown below. These costs are used to characterize technology costs in the
baseline fleet; EPA does not build OMEGA packages using this technology and instead uses the
strong HEY technology presented earlier.
2-426
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Table 2.164 Costs for Powersplit HEV Technology for Different Vehicle Classes (dollar values in 2015$)
Tech
Cost
type
DMC: base
cost
IC:
complexity
DMC:
learning
curve
IC: near
term
thru
2017
2018
2019
2020
2021
2022
2023
2024
2025
1
DMC
$3,224
24
$3,083
$3,023
$2,969
$2,919
$2,873
$2,831
$2,792
$2,755
$2,720
2
DMC
$3,588
24
$3,431
$3,365
$3,304
$3,249
$3,198
$3,151
$3,107
$3,066
$3,028
3
DMC
$3,882
24
$3,712
$3,640
$3,575
$3,515
$3,460
$3,409
$3,361
$3,317
$3,276
4
DMC
$4,710
24
$4,504
$4,417
$4,337
$4,265
$4,198
$4,136
$4,078
$4,025
$3,974
5
DMC
$5,792
24
$5,539
$5,431
$5,333
$5,244
$5,162
$5,085
$5,015
$4,949
$4,887
6
DMC
$5,792
24
$5,539
$5,431
$5,333
$5,244
$5,162
$5,085
$5,015
$4,949
$4,887
1
IC
Highl
2018
$1,808
$1,804
$1,105
$1,104
$1,102
$1,101
$1,100
$1,099
$1,098
2
IC
Highl
2018
$2,012
$2,008
$1,230
$1,229
$1,227
$1,225
$1,224
$1,223
$1,222
3
IC
Highl
2018
$2,177
$2,172
$1,331
$1,329
$1,327
$1,326
$1,324
$1,323
$1,321
4
IC
Highl
2018
$2,641
$2,636
$1,615
$1,613
$1,610
$1,609
$1,607
$1,605
$1,603
5
IC
Highl
2018
$3,248
$3,241
$1,986
$1,983
$1,980
$1,978
$1,976
$1,974
$1,972
6
IC
Highl
2018
$3,248
$3,241
$1,986
$1,983
$1,980
$1,978
$1,976
$1,974
$1,972
1
TC
$4,891
$4,827
$4,074
$4,023
$3,976
$3,932
$3,891
$3,854
$3,818
2
TC
$5,444
$5,373
$4,535
$4,477
$4,425
$4,376
$4,331
$4,289
$4,249
3
TC
$5,889
$5,812
$4,906
$4,844
$4,787
$4,734
$4,685
$4,640
$4,597
4
TC
$7,145
$7,052
$5,952
$5,877
$5,808
$5,744
$5,685
$5,630
$5,578
5
TC
$8,786
$8,672
$7,319
$7,227
$7,142
$7,063
$6,990
$6,922
$6,859
6
TC
$8,786
$8,672
$7,319
$7,227
$7,142
$7,063
$6,990
$6,922
$6,859
Note: DMC=direct manufacturing costs; IC=indirect costs; TC=total costs.
References
1 U.S. EPA, October 2014, Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy
Trends: 1975 Through 2014, Report EPA-420-R -14-023a.
2 Stanton, D.W. "Light Duty Efficient, Clean Combustion." Final Report by Cummins, Inc., to the U.S. Department
of Energy, Report No. DE-FC26-07NT43279, June 3, 2011. http://www.osti. gov/scitech/servlets/piirl/1038535/.
3 Ellies, B., Schenk, C., and Dekraker, P., "Benchmarking and Hardware-in-the-Loop Operation of a 2014 MAZDA
SkyActiv2.0L 13:1 Compression Ratio Engine," SAE Technical Paper 2016-01-1007, 2016, doi: 10.4271/2016-01-
1007.
4 EPA, "Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975 through
2015," EPA-420-R-15-016, December 2015.
5 Yu, C., Park, K., Han, S., and Kim, W., "Development of Theta II 2.4L GDI Engine for High Power & Low
Emission," SAE Technical Paper 2009-01-1486, 2009, doi: 10.4271/2009-01-1486.
6 Saeki, T., Tsuchiya, T, Iwahashi, K., Abe, S. "Development of V6 3.5-Liter 2GR-FSE Engine." Toyota Technical
Review, Volume 55, No. 1, pp 94-99, November 2006. Link?
7 Ikoma, T., Abe, S., Sonoda, Y., Suzuki, H. etal., "Development of V-6 3.5-liter Engine Adopting New Direct
Injection System," SAE Technical Paper 2006-01-1259, 2006, doi: 10.4271/2006-01-1259.
8 Yamaguchi, J. "Lexus Gives V6 Dual Injection." SAE Automotive Engineering International, January 2006, pp 17-
20.
9 U.S. EPA. "Annual Certification Test Results & Data." Light Duty Vehicle Data - Certified Vehicle Test Result
Report Data, MY2010-2016. https://www.epa.gov/compliance-and-fuel-economy-data/annual-certification-test-
data-vehicles-and-engines.
10 Ford Motor Company. 2016. "More Torque and Better Boost: 2017 Ford F-150 to Debut with All-New 3.5-Liter
Ecoboost Engine and 10-Speed Transmission."
https://media.ford.com/content/fordmedia/fna/us/en/news/2016/05/03/2017-ford-fl50-more-torque-better-boost.pdf,
last accessed July 5, 2016.
2-427
-------
Technology Cost, Effectiveness, and Lead Time Assessment
11 Ford Motor Company. 2015. "Ford Marks Production Milestone As 5-Millionth EcoBoost-Equipped Vehicle
Rolls Off Assembly Line." https://media.ford.com/content/fordmedia/fna/us/en/news/2015/03/17/ford-marks-
production-milestone-as-5-millionth-ecoboost-equipped.pdf, last accessed July 5, 2016.
12 Volkswagen. 2015. http://www.volkswagen.co.uk/technology/petrol/active-cylinder-technology-act, last accessed
July 5, 2016.
13 Eichler, F., Demmelbauer-Ebner, W., Theobald, J., Stiebels, B., Hoffmeyer, H., Kreft, M. "The New EA211 TSI®
evo from Volkswagen." 37. Internationales Wiener Motorensymposium 2016.
14 Schamel, A., Scheidt, M., Weber, C. Faust, H. Is Cylinder Deactivation a Viable Option for a Downsized 3-
Cylinder Engine? Vienna Motor Symposium, 2015.
15 Wilcutts, M., Switkes, J., Shost, M. and Tripathi, A., "Design and Benefits of Dynamic Skip Fire Strategies for
Cylinder Deactivated Engines," SAE Int. J. Engines 6(1):2013, doi: 10.4271/2013-01-0359.
16 Ernst, R., Fridfeldt, R., Lamb, S., Lloyd-Thomas, D., Phlips, P., Russell, R., Zenner, T. "The New 3 Cylinder 1.0L
Gasoline Direct Injection Turbo Engine from Ford. 20th Aachen Colloquium - Automobile and Engine Technology,
2011.
17 Wada, Y., Nakano, K., Mochizuki, K., and Hata, R., "Development of a New 1.5L 14 Turbocharged Gasoline
Direct Injection Engine," SAE Technical Paper 2016-01-1020, 2016, doi: 10.4271/2016-01-1020.
18 Nakano, K., Wada, Y., Jono, M., Narihiro, S. "New In-Line 4-Cylinder Gasoline Direct Injection Turbocharged
Downsizing Engine." Honda R&D Technical Review, April 2016, pp 139-146.
19 National Instruments. Subsystems Required to Control Low Temperature Combustion Engines. Jun 11, 2014.
http://www.ni.com/white-paper/13516/en/pdf, last accessed August 25, 2015.
20 Takahashi, D., Nakata, K., Yoshihara, Y., Ohta, Y. et al., "Combustion Development to Achieve Engine Thermal
Efficiency of 40% for Hybrid Vehicles," SAE Technical Paper 2015-01-1254, 2015, doi: 10.4271/2015-01-1254.
21 Yamada, T., Adachi, S., Nakata, K., Kurauchi, T. et al. "Economy with Superior Thermal Efficient Combustion
(ESTEC)," SAE Technical Paper 2014-01-1192 doi: 10.4271/2014-01-1192.
22 Lee, S., Schenk, C., and McDonald, J., "Air Flow Optimization and Calibration in High-Compression-Ratio
Naturally Aspirated SI Engines with Cooled-EGR," SAE Technical Paper 2016-01-0565, 2016, doi: 10.4271/2016-
01-0565.
23 Ellies, B., Schenk, C., and Dekraker, P., "Benchmarking and Hardware-in-the-Loop Operation of a 2014 M 2.0L
13:1 Compression Ratio Engine," SAE Technical Paper 2016-01-1007, 2016, doi: 10.4271/2016-01-1007.
24 Souhaite, P., Mokhtari, S. "Combustion system design of the new PSA Peugeot Citroen EB TURBO PURE TECH
engine," Proceedings: Internationaler Motorenkongress 2014, DOI: 10.1007/978-3-658-05016-0 5.
25 Budack, R., Kuhn, M., Wurms, R., Heiduk, T., "Optimization of the Combustion Process as Demonstrated on the
New Audi 2.01 TFSI," 24th Aachen Colloquium Automobile and Engine Technology 2015.
26 Wurms, R., Budack, R., Grigo, M., Mendl, G., Heiduk, T., Knirsch, S. "The new Audi 2.01 Engine with
innovative Rightsizing," 36. Internationales Wiener Motorensymposium 2015.
27 Hirose, I. "Mazda 2.5L SKYACTIV-G Engine with New Boosting Technology." 37. Internationales Wiener
Motorensymposium 2016.
28 Hatano, J., Fukushima, H., Sasaki, Y., Nishimori, K., Tabuchi, T., Ishihara, Y. "The New 1.6L 2-Stage Turbo
Diesel Engine for HONDA CR-V." 24th Aachen Colloquium - Automobile and Engine Technology 2015.
29 Steinparzer, F., Nefischer, P., Hiemesch, D., Kaufmann, M., Steinmayr, T. "The New Six-Cylinder Diesel Engines
from the BMW In-Line Engine Module." 24th Aachen Colloquium - Automobile and Engine Technology 2015.
30 Eder, T., Weller, R., Spengel, C., Bohm,J., Herwig, H., Sass, H. Tiessen, J., Knauel, P. "Launch of the New
Engine Family at Mercedes-Benz." 24th Aachen Colloquium - Automobile and Engine Technology 2015.
31 Koeberlein, D. "Cummins SuperTruck Program - Technology and System Level Demonstration of Highly
Efficient and Clean, Diesel Powered Class 8 Trucks." U.S. DOE Vehicle Technologies Office Annual Merit Review
and Peer Evaluation, June 9-12, 2015.
32 Gibble, J. "Volvo SuperTruck - Powertrain Technologies for Efficiency Improvement." U.S. DOE Vehicle
Technologies Office Annual Merit Review and Peer Evaluation, June 9-12, 2015.
33 Liickert, P., Arndt, S., Duvinage, F., Kemmner, M., Binz, R., Storz, O., Reusch, M., Braun, T., Ellwanger, S.
"The New Mercedes-Benz 4-Cylinder Diesel Engine OM654 - The Innovative Base Engine of the New Diesel
Generation." 24th Aachen Colloquium - Automobile and Engine Technology 2015.
34 Busch, R., Jennes, J., Miiller, J., Kriiger, M., Naber, D., Kauss, H. "Emission and Fuel Consumption Optimized
Turbo Charging of Passenger Car Diesel Engines." Vienna Motor Symposium, 2015.
35 Ruth, M. "ATP-LD; Cummins Next Generation Tier 2 Bin 2 Diesel Engine." .S. DOE Vehicle Technologies
Office Annual Merit Review and Peer Evaluation, May 14-17, 2013.
2-428
-------
Technology Cost, Effectiveness, and Lead Time Assessment
36 Ruth, M. "ATP-LD; Cummins Next Generation Tier 2 Bin 2 Diesel Engine." .S. DOE Vehicle Technologies
Office Annual Merit Review and Peer Evaluation, June 9-12, 2015.
37 U.S. DOE, April 8, 2015. Cummins Improving Pick-Up Truck Engine Efficiency with DOE and Nissan.
Accessed on 2/3/2016 at the following URL: http://energy.gov/eere/success-stories/articles/cummins-improving-
pick-truck-engine-efficiency-doe-and-nissan.
38 Ogihara, H., "Research Into Surface Improvement for Low Friction Pistons," SAE Technical Paper 2005-01-1647,
2005, doi: 10.4271/2005-01-1647.
39 Kim, Y., Kim, S.J., Lee, J., Lim, D. "Nanodiamond Reinforced PTFE Composite Coatings." MTZ 76 (2) pp 32-35,
February 2015.
40 Hanke, W., Ando, H., Fahr, M., Voigt, M. "Friction Reduction in Power Cylinder Systems for Passenger Car
Diesel Engines" MTZ 75 (2) pp 26-31, February 2014.
41 Landerl, C., Ruelicke, M., Durst, B., Mattes, W. "The New BMW Inline 6-Cylinder Gasoline Engine with
TwinPower Turbo, Direct Injection, and VALVETRONIC in the New BMW 7 Series." 24th Aachen Colloquium -
Automobile and Engine Technology 2015.
42 Ncufkr. H.K. "The Car of the Future will continue to Fascinate People." Oral Presentation, Vienna Motor
Symposium, 2015.
43 Ragot, P. and Rebbert, M., "Investigations of Crank Offset and Its Influence on Piston and Piston Ring Friction
Behavior Based on Simulation and Testing," SAE Technical Paper 2007-01-1248, 2007, doi: 10.4271/2007-01-
1248.
44 Toyota Motor Corporation. 2009 Toyota Prius Service Manual. Toyota Technical Information System,
https ://techinfo .toyota. com/.
45 Confer, K., Kirwan, J., and Engineer, N., "Development and Vehicle Demonstration of a Systems-Level Approach
to Fuel Economy Improvement Technologies," SAE Technical Paper 2013-01-0280, 2013, doi: 10.4271/2013-01-
0280.
46 U.S. EPA, October 2014, Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy
Trends: 1975 Through 2014, Report EPA-420-R -14-023a, p. 74.
47 "Once-promising dual-clutch transmissions lose favor in U.S.," Automotive News, December 7, 2015
http://www.autonews.eom/article/20151207/OEM06/312079988/once-promising-dual-clutch-transmissions-lose-
favor-in-u.s.
48 Newman, K., Kargul, J., and Barba, D., "Benchmarking and Modeling of a Conventional Mid-Size Car Using
ALPHA," SAE Technical Paper 2015-01-1140, 2015, doi: 10.4271/2015-01-1140.
49 Newman, K., Kargul, J., and Barba, D., "Development and Testing of an Automatic Transmission Shift Schedule
Algorithm for Vehicle Simulation," SAE Int. J. Engines 8(3):1417-1427, 2015, doi: 10.4271/2015-01-1142.
50 Moskalik, Andrew, Barba, Dan, and Kargul, John, "Investigating the Effect of Advanced Automatic
Transmissions on Fuel Consumption Using Vehicle Testing and Modeling," to be presented at the 2016 SAE
International Congress, April 2016.
51 Gall, J., "2011 Dodge Charger V6," http://www.caranddriver.com/reviews/2011-dodge-charger-v6-test-review,
(July 2011), accessed July 2016.
52 Gall, J., "2012 Dodge Charger SXT V6," http://www.caranddriver.com/reviews/2012-dodge-charger-sxt-v6-test-
review (January 2012), accessed July 2016.
53 Dick, A., Greiner, J., Locher, A., and Jauch, F., "Optimization Potential for a State of the Art 8-Speed AT," SAE
International Journal of Passenger Cars-Mechanical Systems, 6(2):899-907, 2013, doi: 10.4271/2013-01-1272.
54 Aoki, T., Kato, H., Kato, N, and Masaru, M., "The World's First Transverse 8-Speed Automatic Transmission,"
SAE Technical Paper 2013-01-1274, 2013, doi: 10.4271/2013-01-1274.
55 Christoph Dorr, "The New Automatic Transmission 9G-TRONIC," presented at the 2014 Car Training Institute
Transmission Symposium, Rochester MI.
56 Gaertner, L. and Ebenhoch, M., "The ZF Automatic Transmission 9HP48 Transmission System, Design and
Mechanical Parts," SAE International Journal of Passenger Cars - Mechanical Systems, 6(2):908-917, 2013, doi:
10.4271/2013-01-1276.
57 Shidore, N. et. al. 2014. "Impact of Advanced Technologies on Engine Targets." Project VSS128, DOE Merit
Review, June.
58 Eckl, B., and D. Lexa. 2012. How Many Gears do the Markets Need? GETRAG. International CTI Symposium,
Berlin, Germany, December.
59 Darrell Robinette, 2014, "A DFSS Approach to Determine Automatic Transmission Gearing Content for
Powertrain-Vehicle System Integration, SAE International Journal of Passenger Cars - Mechanical Systems 7 (3).
2-429
-------
Technology Cost, Effectiveness, and Lead Time Assessment
60 Juergen Greiner, Martin Grumbach, Albert Dick, and Cristophe Sasse, 2015, "Advancement in NVH- and Fuel-
Saving Transmission and Driveline Technologies," SAE technical paper 2015-01-1087
61 NAS, 2011, p. 62 footnote.
62 NAS (2015), Prepublication Copy, p. 5-16.
63Wikimedia Commons. https://commons.wikimedia.Org/wiki/File:ZF_Stufenautomatgetriebe_8HP70.jpg Photo
credit: Stefan Krause.
64 U.S. EPA, October 2014, Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy
Trends: 1975 Through 2014, Report EPA-420-R -14-023a, p. 75.
65 Start of Volume Production: New Generation of the ZF 8-Speed "Automatic Transmission in the BMW 5 Series,"
August 21, 2014, https://www.zf.com/corporate/en_de/magazine/magazin_artikel_viewpage_22067944.html.
66 Jens Meiners, "2016 Mini Cooper Clubman Revealed: Another Bigger, Four-Door Mini," Car and Driver, June
2015, http://www.caranddriver.com/news/2016-mini-clubman-revealed-news.
67 Gaertner, L. and Ebenhoch, M., "The ZF Automatic Transmission 9HP48 Transmission System, Design and
Mechanical Parts," SAE Int. J. Passeng. Cars - Mech. Syst. 6(2):908-917, 2013, doi: 10.4271/2013-01-1276.
68 "Land Rover To Demonstrate Latest Technical Innovation With The World's First 9-Speed Automatic
Transmission," Land Rover Media Centre, February 27, 2013, http://newsroom.jaguarlandrover.com/en-in/land-
rover/news/2013/02/rr_rre_9-speed_transmission_270213/.
69 Daimler. 2013. New Nine-Speed Automatic Transmission Debuts in the Mercedes-Benz E350 Blue Tec: Premier
of the new 9G-Tronic. Daimler, July 24. http://media.daimler.com/dcmedia/0-921-1553299-l-1618134-l-0-l-0-0-0-
0-1549054-0-1 -0-0-0-0-0 html
70 Motor Authority: Technology Preview: We Drive Honda's 10-Speed Automatic Transmission,
http://www.motorauthority.eom/news/1100878_technology-preview-we-drive-hondas-10-speed-automatic-
transmission.
71 ZF, "Fuel Saving and Minimizing CO2 Emissions: 6% Lower Fuel Consumption,"
http://www.zf.com/corporate/en_de/products/innovations/8hp_automatic_transmissions/lower_consumption/lower_
consumption.html.
72 Dick, A., Greiner, J., Locher, A., and Jauch, F., "Optimization Potential for a State of the Art 8-Speed AT," SAE
Int. J. Passeng. Cars-Mech. Syst. 6(2):899-907, 2013, doi: 10.4271/2013-01-1272.
73 The New Generation of 8-Speed Automatic Transmission, ZF
http://www.zf.com/corporate/en_de/products/innovations/8hp_automatic_transmissions/8hp_automatic_transmissio
n.html.
74 Greiner, J., Grumbach, M., Dick, A., and Sasse, C., "Advancement in NVH- and Fuel-Saving Transmission and
Driveline Technologies," SAE Technical Paper 2015-01-1087, 2015, doi: 10.4271/2015-01-1087.
75 Christoph Dorr, "The New Automatic Transmission 9G-TRONIC from Mercedes- Benz," presented at the 2014
CTI Symposium, Plymouth, MI.
76 Greiner, J. and Grumbach, M., "Automatic Transmission Systems Beyond 2020 - Challenges and Competition,"
SAE Technical Paper 2013-01-1273, 2013, doi: 10.4271/2013-01-1273.
77 Driveline News, Jan 22 2014, "BMW and Mini Strategy Revealed," http://www.drivelinenews.com/transmission-
insight/bmw-and-mini-transmission-strategy-revealed/.
78 Markus Nell, "BMW's flexible powertrain family with a new generation of transverse automatic transmissions,"
presented at the 2014 Car Training Institute Transmission Symposium, Rochester, MI.
79 Christoph Dorr, "The New Automatic Transmission 9G-TRONIC," presented at the 2014 Car Training Institute
Transmission Symposium, Rochester, MI.
80 Darrell Robinette and Daniel Wehrwein, "Automatic Transmission Technology Selection Using Energy Analysis"
presented at the CTI Symposium 9th International 2015 Automotive Transmissions, HEV and EV Drives.
81 Juergen Greiner, Martin Grumbach, Albert Dick, and Cristophe Sasse, 2015, "Advancement in NVH- and Fuel-
Saving Transmission and Driveline Technologies," SAE technical paper 2015-01-1087.
82 Darrell Robinette, 2014, "A DFSS Approach to Determine Automatic Transmission Gearing Content for
Powertrain-Vehicle System Integration, SAE International Journal of Passenger Cars - Mechanical Systems 7 (3).
83 Hans Greimel, "ZF CEO: We're not chasing 10-speeds," Automotive News, November 23, 2014,
http://www.autonews.eom/article/20141123/OEM10/311249990/zf-ceo:-were-not-chasing-10-speeds.
84 NAS (2015), Prepublication Copy, p. 5-9.
85Wikimedia Commons, https://commons.wikimedia.0rg/wiki/File:Dual-clutch_transmission.svg, Credit: Xavax.
86 Carney, D. 2014. Honda's new 8-speed DCT uses a Torque Converter. SAE Automotive Engineering Magazine,
August 6.
87 NAS (2015), Prepublication Copy, p. 5-7.
-------
Technology Cost, Effectiveness, and Lead Time Assessment
88 NRC (2015), Prepublication Copy, p. 5-7.
89 Eckl, B.," DCT in the American Market: Transferring Customer Perceptions into Product Refinements,"
presented at the 2014 Car Training Institute Transmission Symposium, Rochester, MI.
90 Eckl, B., and D. Lexa. 2012. How Many Gears do the Markets Need? GETRAG. International Car Training
Institute Transmission Symposium, Berlin, Germany, December.
91Wikimedia Commons, https://commons.wikimedia.0rg/wiki/File:Toyota_Super_CVT-i_Ol.JPG, Photo credit:
Hatsukari715.
92Wikimedia Commons, https://c0mm0ns.wikimedia.0rg/wiki/File:Pivgetriebe.png, Credit: Biideler Naumann
93 Shinji Morihiro, "Fuel Economy Improvement by Transmission" presented at the CTI Symposium 8th
International 2014 Automotive Transmissions, HEV and EV Drives.
94 Masayoshi Nakasaki and Yoshikazu Oota, "Key Technologies Supporting Belt-type CVT Evolution," presented at
the 2014 Car Training Institute Transmission Symposium, Rochester MI.
95 Maruyama, F., Kojima, M., and Kanda, T., "Development of New CVT for Compact Car," SAE Technical Paper
2015-01-1091, 2015, doi: 10.4271/2015-01-1091.
96 Hakamagi, J., Kono, T., Habuchi, R., Nishimura, N. et al., "Development of New Continuously Variable
Transmission for 2.0-Liter Class Vehicles," SAE Technical Paper 2015-01-1101, 2015, doi: 10.4271/2015-01-1101.
97 Shimokawa, Y., "Technology Development to Improve JATCO CVT8 Efficiency," SAE Technical Paper 2013-
01-0364, 2013, doi: 10.4271/2013-01-0364.
98 Lindsay Brooke, "JATCO's next-gen CVTs bring high ratio spreads, more efficiency," Automotive Engineering
Magazine, April 23, 2012, http://articles.sae.org/10947/.
99 Naotoshi, P. "Development of a New Generation CVT with Auxiliary Gear Box," SAE Technical Paper 2016-01-
1109, 2016, doi: 10.4271/2016-01-1109.
100 Naotoshi, P. "Development of a New Generation CVT with Auxiliary Gear Box," SAE Technical Paper 2016-01-
1109, 2016, doi: 10.4271/2016-01-1109.
101 Shimokawa, Y., "Technology Development to Improve JATCO CVT8 Efficiency," SAE Technical Paper 2013-
01-0364, 2013, doi: 10.4271/2013-01-0364.
102 Maruyama, F., Kojima, M., and Kanda, T., "Development of New CVT for Compact Car," SAE Technical Paper
2015-01-1091, 2015, doi: 10.4271/2015-01-1091.
103 Inukai, K., Shibahara, A., Uchino, T., Keiichi, N. et al., "Development of High-Efficiency New CVT for Midsize
Vehicle," SAE Technical Paper 2013-01-0365, 2013, doi: 10.4271/2013-01-0365.
104 Hakamagi, J., Kono, T., Habuchi, R., Nishimura, N. et al., "Development of New Continuously Variable
Transmission for 2.0-Liter Class Vehicles," SAE Technical Paper 2015-01-1101, 2015, doi: 10.4271/2015-01-1101.
105 Don Sherman, Dec 2013, "The Unsinkable CVT: How the "Gearless" Transmission Is Getting Its Groove Back,"
Car and Driver, http://www.caranddriver.com/features/how-cvt-transmissions-are-getting-their-groove-back-
feature.
106 Shimokawa, Y., "Technology Development to Improve JATCO CVT8 Efficiency," SAE Technical Paper 2013-
01-0364, 2013, doi: 10.4271/2013-01-0364.
107 Maruyama, F., Kojima, M., and Kanda, T., "Development of New CVT for Compact Car," SAE Technical Paper
2015-01-1091, 2015, doi: 10.4271/2015-01-1091.
108 Inukai, K., Shibahara, A., Uchino, T., Keiichi, N. et al., "Development of High-Efficiency New CVT for Midsize
Vehicle," SAE Technical Paper 2013-01-0365, 2013, doi: 10.4271/2013-01-0365.
109 Hakamagi, J., Kono, T., Habuchi, R., Nishimura, N. et al., "Development of New Continuously Variable
Transmission for 2.0-Liter Class Vehicles," SAE Technical Paper 2015-01-1101, 2015, doi: 10.4271/2015-01-1101.
110 Mamiko Inoue, "Advanced CVT control to achieve both fuel economy and drivability," presented at the 2015 Car
Training Institute Transmission Symposium, Novi, MI.
111 Masayoshi Nakasaki and Yoshikazu Oota, "Key Technologies Supporting Belt-type CVT Evolution," presented
at the 2014 Car Training Institute Transmission Symposium, Rochester, MI.
112 Dana Holding Corp. 2014. Dana Advances Development of VariGlide™ Continuously Variable Planetary
Technology. PRNewswire, May 19. http://www.prnewswire.com/news-releases/dana-advances-development-of-
variglide-continuously-variable-planetary-technology-259791981 .html.
113 Dick, A., J. Greiner, A. Locher, and F. Jauch. 2013. Optimization Potential for a State of the Art 8-Speed
AT. SAE 2013-01-1272.
114 Dick, A., Greiner, J., Locher, A., and Jauch, F., "Optimization Potential for a State of the Art 8-Speed AT," SAE
Int. J. Passeng. Cars-Mech. Syst. 6(2):899-907, 2013, doi: 10.4271/2013-01-1272.
115 Aoki, T., Kato, H., Kato, N., and Masaru, M., "The World's First Transverse 8-Speed Automatic Transmission,"
SAE Technical Paper 2013-01-1274, 2013, doi: 10.4271/20^3^-1274.
-------
Technology Cost, Effectiveness, and Lead Time Assessment
116 Christoph Dorr, "The New Automatic Transmission 9G-TRONIC," presented at the 2014 Car Training Institute
Transmission Symposium, Rochester, MI.
117 Christoph Dorr, "The New Automatic Transmission 9G-TRONIC," presented at the 2014 Car Training Institute
Transmission Symposium, Rochester, MI.
118 Martin, K. 2012. Transmission Efficiency Developments. SAE Transmission and Driveline Symposium:
Competition for the Future, October 17-18. Detroit, Michigan, [as cited inNAS (2015), Prepublication Copy, p. 5-
22.].
119 NSK Europe. 2014. New Low-Friction TM-Seal for Automotive Transmissions,
http://www.nskenrope.com/transniission-bearings-low-friction-tm-seal-2373.htm.
120 Christoph Dorr, "The New Automatic Transmission 9G-TRONIC," presented at the 2014 Car Training Institute
Transmission Symposium, Rochester, MI.
121 Noles, J. 2013. Development of Transmission Fluids Delivering Improved Fuel Efficiency by Mapping
Transmission Response to Viscosity and Additive Changes. Presentation at the SAE Transmission & Driveline
Symposium, Troy, Michigan, October 16-17.[as cited inNAS (2015), Prepublication Copy, p. 5-25.].
122 NAS (2015), Prepublication Copy, p. 5-28.
123 NAS (2015), Prepublication Copy, p. 5-27.
124 Shimokawa, Y., "Technology Development to Improve JATCO CVT8 Efficiency," SAE Technical Paper 2013-
01-0364, 2013, doi: 10.4271/2013-01-0364.
125 Ando, T., Yagasaki, T., Ichijo, S., Sakagami, K. et al., "Improvement of Transmission Efficiency in CVT
Shifting Mechanism Using Metal Pushing V-Belt," SAE Int. J. Engines 8(3): 1391-1397, 2015, doi: 10.4271/2015-
01-1103.
126 Greiner, J., Grumbach, M., Dick, A., and Sasse, C., "Advancement in NVH- and Fuel-Saving Transmission and
Driveline Technologies," SAE Technical Paper 2015-01-1087, 2015, doi: 10.4271/2015-01-1087.
127 Markus Nell, "BMW's flexible powertrain family with a new generation of transverse automatic transmissions,"
presented at the 2014 Car Training Institute Transmission Symposium, Rochester, MI.
128 Markus Nell, "BMW's flexible powertrain family with a new generation of transverse automatic transmissions,"
presented at the 2014 Car Training Institute Transmission Symposium, Rochester, MI.
129 Albert Dick, Juergen Greiner, Anton Locher, and Friedemann Jauch, "Optimization Potential for a State of the
Ary 8-Speed AT," SAE International Journal of Passenger Cars - Mechanical Systems 6(2):2013, doi:
10.4271/2013-01-1272.
130 Darrell Robinette and Daniel Wehrwein, "Utilizing Energy Analysis Methods to Select Transmission
Technologies and Optimize Powertrain-Vehicle System Fuel Consumption," presented at the SAE 2014
Transmission and Driveline Symposium.
131 Markus Nell, "BMW's flexible powertrain family with a new generation of transverse automatic transmissions,"
presented at the 2014 Car Training Institute Transmission Symposium, Rochester, MI
132 Wikimedia Commons, https://commons.wikimedia.Org/wiki/File:Bauma_2007_ZF_Drehmomentwandler.jpg,
Photo credit: Aconcagua.
133 Weissler, Paul. 2011. "2012 Mazda3 Skyactiv achieves 40 mpg without stop/start." Automotive Engineering
Magazine, October 28.
134 Markus Nell, "BMW's flexible powertrain family with a new generation of transverse automatic transmissions,"
presented at the 2014 Car Training Institute Transmission Symposium, Rochester, MI
135 Nelson, P. et al., "Modeling the Performance and Cost of Lithium-Ion Batteries for Electric Drive Vehicles,"
Second Edition, Argonne National Laboratory, ANL-12/55 (December 2012).
136 See EPA Docket EPA-HQ-OAR-2015-0827, Microsoft Excel attachment to Docket Item titled "Data and Charts
for Selected Figures in Electrification Chapters of Proposed Determination TSD."
137 See Table ES-3, "Selected Technology Penetrations to Meet MY2025 Standards," Draft TAR p. ES-10.
138 Howell, D., "Electric Drive Vehicle Battery Status and Trends," Vehicle Technologies Office, Department of
Energy. Presented at SAE Government-Industry Meeting, Washington, D.C., January 21, 2016.
139 HybridCars.com, "September 2016 Dashboard." Retrieved on Nov. 2, 2016 from
http://www.hybridcars.com/september-2016-dashboard/.
140 Volkswagen AG, "The Future of the Volkswagen Group: Together Strategy 2025", retrieved on Nov. 3, 2016
from
http://www.volkswagenag.com/content/vwcorp/info_center/en/talks_and_presentations/2016/06/together.bin.html/bi
narystorageitem/file/Presse_englisch_NICHTanimiert_Version_24.pdf.
141 Boston, W., "Volkswagen to Boost Electric Vehicles, Pursue Self-Driving Cars," The Wall Street Journal, June
-------
Technology Cost, Effectiveness, and Lead Time Assessment
142 Daimler, "Daimler invests massively in green powertrain technologies: All Mercedes-Benz model series will be
electrified," Press Release, June 14, 2016. Retrieved from
http://media.daimler.eom/marsMediaSite/ko/en/l 1108480.
143 BMW, "New generation of plug-in hybrid models," Press Release, December 1, 2014. Retrieved from
https://www.press.bmwgroup.com/global/article/detail/T0197302EN/new-generation-of-plug-in-hybrid-models.
144 Eisenstein, P., "Volvo Is Pushing Down the Pedal on Battery-Powered Cars," NBCNews.com, June 14, 2016.
Retrieved from http://www.nbcnews.com/business/autos/volvo-pushing-down-pedal-battery-powered-cars-n591336.
145 Ford Motor Co., "Ford Investing $4.5 Billion in Electrified Vehicle Solutions, Reimagining How to Create
Future Vehicle User Experiences," Press Release, December 10, 2015. Retrieved on November 3, 2016 from
https://media.ford.com/content/fordmedia/fna/us/en/news/2015/12/10/fordinvesting45billioninelectrifiedvehiclesolut
ions.html.
146 Nikkei Asian Review, "Toyota to mass produce electric vehicles," November 7, 2016. Retrieved on November 7,
2016 from http://asia.nikkei.com/Japan-Update/Toyota-to-mass-produce-electric-vehicles
147 Tajitsu, N. et al., "Toyota, in about-face, may mass-produce long-range electric cars:
Nikkei," November 7, 2016. Retrieved on November 7, 2016 from http://www.reuters.com/article/us-toyota-electric-
cars-idU SKBN13204D.
148 See Table ES-3, "Selected Technology Penetrations to Meet MY2025 Standards," Draft TAR p. ES-10.
149 See Table 3.5-25 of RIA, 2017-2025 FRM.
150 "Obama Administration Announces New Actions To Accelerate The Deployment of Electrical Vehicles and
Charging Infrastructure," Press Release, The White House, November 3, 2016.
151 Partial Consent Decree, in re: Volkswagen "Clean Diesel" Marketing, Sales Practices, and Products Liability
Litigation, Case 3:15-md-02672-CRB Document 1605-1 Filed 06/28/16. Retrieved on November 7, 2016 from
https://www.epa.gov/sites/production/files/2016-06/documents/vwpartialsettlement-cd.pdf
152 "Fixing America's Surface Transportation Act—Designation of Alternative Fuel Corridors," Federal Register,
Vol. 81, No. 141, Friday, July 22, 2016. See also: http://www.fhwa.dot.gov/environment/alternative_fuel_corridors/.
153 Nykvist, B. and Nilsson, M.; "Rapidly Falling Costs of Battery Packs for Electric Vehicles," Nature Climate
Change. March 2015; doi: 10.1038/NCLIMATE2564.
154 Green Car Congress, "GM and LG expand their relationship on Bolt EV; 12 components, including PEEM,"
October 20, 2015. Retrieved from http://www.greencarcongress.com/2015/10/20151020-gmlg.html.
155 Siemens, "Siemens and Valeo join forces for global leadership in powertrains for electric cars," Press Release,
April 18, 2016. Retrieved April 29, 2016 from http://www.siemens.com/press/PR2016040250COEN.
156 Schultz, Jay W. and Huard, Steve, "Comparing AC Induction with Permanent Magnet motors in hybrid vehicles
and the impact on the value proposition." White Paper, Parker Hannifin, 2013. Retrieved from
http://www.parkermotion.com/whitepages/Comparing_AC_and_PM_motors.pdf on Oct 14, 2015.
157 Widmer, James D. et al., "Electric vehicle traction motors without rare earth magnets," Sustainable Materials and
Technologies 3 (2015) 7-13. doi: .1.0..1.016/i.susmat.20.1.5.02.001 .
158 Jenkins, J., "A Closer Look at Switched Reluctance Motors," Charged Magazine, Cot/Nov 2012, pp. 26-28.
159 Ruoff, C., "A Closer Look at Torque Ripple," Charged Magazine, Jul/August 2015, pp. 22-29.
160 "First look at all-new Voltec propulsion system for 2G Volt; the only thing in common is a shipping cap," Green
Car Congress, October 29, 2014. Retrieved May 2, 2016 from
http://www.greencarcongress.com/2014/10/20141029-voltec.html.
161 "Chevrolet Introduces All-New 2016 Volt," General Motors Press Release, January 12, 2015.
http://media.gm.com/media/us/en/gm/news.detail.html/content/Pages/news/us/en/2015/Jan/naias/chevrolet/volt/0112
-volt-2016-intro.html.
162 Morris, C., "2016 Chevy Volt: GM's top electrification engineers on designing the all-new EREV,"
https://chargedevs.com/features/2016-chevy-volt-gms-top-electrification-engineers-on-designing-the-all-new-erev/.
163 Anwar, M. et al., "Power Dense and Robust Traction Power Inverter for the Second-Generation Chevrolet Volt
Extended-Range EV," SAE Int. J. Alt. Power. 4(1): 145-152, 2015, doi: 10.4271/2015-01-1201.
164 Grewe, T., General Motors, "GM RWD PHEV Propulsion System for the Cadillac CT6 Sedan," SAE paper
2016-01-1159, presented at 2016 SAE Hybrid and Electric Vehicle Technologies Symposium, February 10, 2016.
165 Grewe, T., General Motors, "Chevrolet Malibu Hybrid Propulsion System," presented at 2016 SAE Hybrid and
Electric Vehicle Technologies Symposium, February 10, 2016.
166 "Toyota Unveils Advanced Technologies in All-New Prius,"
http://pressroom.tovota.com/releases/2016+tovota+prius+technologv.htm.
2-433
-------
Technology Cost, Effectiveness, and Lead Time Assessment
167 Cole, J., "Exclusive: GM Exec Says Spark EV's 400 lb-ft of Torque No Misprint," May 2, 2013 (source
confirmed by conversation with P. Savagian, February 2016). At http://insideevs.com/gm-general-says-spark-evs-
4001b-ft-of-torque-no-misprint/.
168 Whaling, C., "Market Analysis of Wideband Gap Devices in Car Power Electronics." Retrieved March 9, 2016
from http://tec.ieee.org/2014/10/31/market-analysis-wideband-gap-devices-car-power-electronics/.
169 Kimura, T. et al., "High-power-density Inverter Technology for Hybrid and Electric Vehicle Applications,"
Hitachi Review Vol. 63 No. 2 (2014). Retrieved on March 31, 2016 from
fattp://www.hitachi.com/rev/pdf/20.1.4/r20.1.4 02 106.pdf.
170 Goreham, J., "Toyota reveals breakthrough that will increase future Prius MPG by 10%," Torque News.
Retrieved May 2, 2016 from http://www.torauenews.com/1083/tovota-reveais-breakt.Iirough-wiH-increase-fiiture-
prius-mpg-10#sthash.iIxVsn43.dpiif.
171 Beg, F., "A Novel Design Methodology for a 1.5 KW DC/DC Converter in EV and Hybrid EV Applications,"
International Science Index, Electrical and Computer Engineering Vol:8, No:9, 2014.
172 Al Sakka, M. et al., "DC/DC Converters for Electric Vehicles," in Electric Vehicles - Modelling and Simulations,
Dr. Seref Soylu (Ed.), ISBN: 978-953-307-477-1, InTech, 2011. Available from: http://cdn.intechopen.com/pdfs-
wm/19583.pdf.
173 "Obama Administration Announces New Actions To Accelerate The Deployment of Electrical Vehicles and
Charging Infrastructure," Press Release, The White House, November 3, 2016.
174 Partial Consent Decree, in re: Volkswagen "Clean Diesel" Marketing, Sales Practices, and Products Liability
Litigation, Case 3:15-md-02672-CRB Document 1605-1 Filed 06/28/16. Retrieved on November 7, 2016 from
https://www.epa.gov/sites/production/files/2016-06/documents/vwpartialsettlement-cd.pdf
175 "Fixing America's Surface Transportation Act—Designation of Alternative Fuel Corridors," Federal Register,
Vol. 81, No. 141, Friday, July 22, 2016. See also: http://www.fhwa.dot.gov/environment/alternative_fuel_corridors/
176 Safoutin, M., Cherry, J., McDonald, J., and Lee, S., "Effect of Current and SOC on Round-Trip Energy
Efficiency of a Lithium-Iron Phosphate (LiFeP04) Battery Pack," SAE Technical Paper 2015-01-1186, 2015,
doi: 10.4271/2015-01-1186.
177 Idaho National Laboratory, "Steady State Vehicle Charging Fact Sheet: 2015 Nissan Leaf," INL/EXT-15-34055.
Downloaded Jan. 26, 2016 from http://avt.inel.gov/pdf/fsev/SteadyStateLoadCharacterization2015Leaf.pdf.
178 Conlon, B., Blohm, T., Harpster, M., Holmes, A. et al., "The Next Generation "Voltec" Extended Range EV
Propulsion System," SAE Int. J. Alt. Power.4(2):248-259, 2015, doi: 10.4271/2015-01-1152.
179 Raghavan, A., "SENSOR: Smart Embedded Network of Sensors with Optical Readout," PARC and LG Chem
Power, presented at AABC 2014, February 2014.
180 Herron, D., "Why Do Electric Cars Have Lead-Acid 12-Volt Batteries When Lithium Is Lighter?"
PlugInCars.com, December 20, 2013. Retrieved from http://www.plugincars.com/why-do-electric-cars-have-lead-
acid- 12-volt-batteries-when-lithium-lighter-129118. html
181 Webb, A., "Tesla Owners Encounter Problems with 12-Volt Battery," PlugInCars.com, January 22, 2014.
Retrieved from http://www.plugincars.com/tesla-owners-encounter-ev-related-problems-12-volt-battery-129287.
html
182 Teslarati.com, "Why Tesla's lead acid 12V battery needs to be lithium-ion based," September 22, 2016.
Retrieved from http://www.teslarati.com/why-tesla-12v-battery-needs-to-be-lithium-ion/
183 TeslaMotorsClub.com, "Near annual replacement of 12V battery is typical according to Tesla Service Tech," user
discussion thread, Jan 15, 2015. Retrieved from https://teslamotorsclub.com/tmc/threads/near-annual-replacement-
of-12v-battery-is-typical-according-to-tesla-service-tech.41006/
184 C&D Technologies, "Deep Cycle Series DCS-33H/DCS-33 Valve Regulated Lead Acid Battery for Deep Cycle
Mobility Applications," Manufacturer Data Sheet, 41-7956.
185 A123 Systems, A123 12V Starter Battery Information Sheet. Retrieved from
http://www.al23systems.com/lithium-starter-battery.htm
186 Weissler, P., "Hyundai targets Prius with new 'triple-electrified' Ioniq," SAE Automotive Engineering, March 30,
2016. Retrieved from http://articles.sae.org/14706/
187 http://www.uscar.org/guest/partnership/ 1/us-drive.
188 U.S. DRIVE Partnership, "US DRIVE Electrical and Electronics Technical TeamRoadmap," June 2013. Down
loaded on Dec. 23, 2015 from
http://wwwl.eere.energy.gov/vehiclesandfuels/pdfs/program/eett_roadmapjune2013.pdf.
189 U.S. DRIVE Partnership, "US DRIVE Electrical and Electronics Technical TeamRoadmap," June 2013, pp. 7-8.
190 Slenzak, J., "Next Generation Electrification Products: Focus on Integration and Cost Reduction," Bosch, The
Battery Show 2015, Novi MI, September 15, 2015.
-------
Technology Cost, Effectiveness, and Lead Time Assessment
191 cf. U.S. DRIVE Electrical and Electronics Technical Team Roadmap, p. 10.
192 Kane, M., "Deep Dive: Chevrolet Bolt Battery Pack, Motor And More." Retrieved April 18, 2016 from
http://insideevs.com/deep-dive-chevrolet-bolt-battery-pack-motor-and-more/.
193 EPA, "Regulatory Impact Analysis: Final Rulemaking for 2017-2025 Light-Duty Vehicle Greenhouse Gas
Emission Standards and Corporate Average Fuel Economy Standards," 2012, p. 3-52. See Docket ID EPA-HQ-
OAR-2010-0799, docket item number EPA-HQ-OAR-2010-0799-12013.
194 EPA, "Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975
through 2015," December 2015, p 72.
195 EPA, "Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975
through 2015," December 2015, p 75.
196 EPA, "Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975
through 2015," December 2015, p 48.
197 Consumer Reports News, "Stop idling! Stop-start systems have great promise
for saving fuel," June 29, 2012. Retrieved from http://www.consumerreports.org/ero/news/2012/06/stop-idling-stop-
start-svstems-have-great-proniise-for-saving-fiiei/index.htm.
198 Martinez, M., "More cars getting stop-start despite driver resistance," The Detroit News, October 8, 2014.
Retrieved from http://www.detroitnews.eom/storv/business/autos/2014/10/07/autos-stop-start-driver-
resistance/16893597/.
199 Murray, C., "GM Chooses Micro- Over Mild-Hybrid inNew Malibu," Design News, October 16, 2013.
200 Green Car Congress, "Schaeffler demo vehicle hits 2020 CAFE targets without electrification; 15% cut in fuel
consumption; 48V hybrid system for 2025," January 27, 2014. Retrieved from
http://www.greencarcongress.com/2014/01/20140127-schaeffler.html.
201 Halvorson, B.," Ultracapacitor Resistance Breaking Down Among Automakers?" Green Car Reports, June 2,
2015. Retrieved from http://www.greencarreports.eom/news/1098550_ultracapacitor-resistance-breaking-down-
among-automakers.
202 Bragman, A., "2014 Chevrolet Malibu: Frist Drive," Cars.com, October 11, 2013. Retrieved from
https://www.cars.com/articles/2013/10/2014-chevrolet-malibu-first-drive/.
203 Ford Motor Company, "70 Percent of Ford Lineup to Have Auto Start-Stop by 2017; Fuel Economy Plans
Accelerate," Press Release, December 12, 2013.
204 M. Everett, "Advancing Ultracapacitor Applications in Industrial and Transportation Applications," 2015 AABC,
Detroit.
205 Mazda, "Brake Energy Regeneration System: i-ELOOP continuously recovers kinetic energy as the vehicle
decelerates and reuses it as electricity," retrieved on April 1, 2016 from
http://www.mazda.com/en/innovation/technology/env/i-eloop/.
206 Blessing, U., "GETRAG 48V: How much hybridization is possible with the new vehicle power?." 14th VDI
Congress, 2014.
207 FEV, "In-market Application of Start-Stop Systems in European Market," Final Report, P26844-01/ Al/ 01/
61605, December 2011. Retrieved from
http://www.theicct.org/sites/default/files/FEV_LDV%20EU%20Technology%20Cost%20Analysis_StartStop%200
verview.pdf.
208 EPA, "Regulatory Impact Analysis: Final Rulemaking for 2017-2025 Light-Duty Vehicle Greenhouse Gas
Emission Standards and Corporate Average Fuel Economy Standards," 2012, p. 3-52. See Docket ID EPA-HQ-
OAR-2010-0799, docket item number EPA-HQ-OAR-2010-0799-12013.
209 H-D Systems, "Future U.S. Trends in the Adoption of Light-Duty Automotive Technologies," Integrated Final
Report, Prepared for: American Petroleum Institute, September 2013, p. 60.
210 German, J., "Hybrid Vehicles: Technology Development and Cost Reduction," International Council on Clean
Transportation (ICCT), Technical Brief No. 1, July 2015.
211 Graf, F. et al., "AES: An Approach to an Integrated 12V/48V Energy Storage Solution," Proceedings of the 24th
Aachen Colloquium Automobile and Engine Technology 2015, pp. 1077-1090.
212 National Academy of Sciences, "Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles," 2015, Prepublication Copy, p. 4-39.
213 Bilek, M., "2013 Chevrolet Malibu - Refreshed for 2013, Chevy's midsize offering gains refinement and
features," DriveChicago.com. Retrieved May 5, 2016 from http://www.drivechicago.com/reviews/2013-chevrolet-
malibu-refreshed-for-2013-chevys-midsize-offering-gains-refinement-and-features_2424.
214 Edmunds.com, "Full Edmunds Expert Review: 2013 Chevrolet Malibu." Retrieved onMay 5, 2016 from
http://www.edmunds.com/chevrolet/malibu/2013/review/ ^ 435
-------
Technology Cost, Effectiveness, and Lead Time Assessment
215 General Motors, "2013 Malibu Eco Is The Most Fuel-Efficient Malibu Ever," Press Release, December 12, 2011.
http://media.gm.com/media/us/en/chevrolet/news.detail.html/content/Pages/news/us/en/201 l/Dec/1212_malibu/121
2_eassist.html.
216 Gross, O., "Defining Energy Storage System Requirements Based upon Passenger Vehicle Fleet Needs," 2015
Battery Congress.
217 General Motors, "GMC Introduces 2016 Sierra with eAssist," Retrieved March 9, 2016 from
http://www.gm.com/mol/m-2016-feb-0225-sierra-eAssist.html.
218 General Motors, "Chevrolet Introduces 2016 Silverado with eAssist," Retrieved March 11, 2016 from
http://www.gm.com/mol/m-2016-feb-0225-silverado-eAssist.html.
219 FEV, "Light-Duty Vehicle Technology Cost Analysis: 2013 Chevrolet Malibu ECO with eAssist BAS
Technology Study," Report FEV P311264, Contract No. EP-C-12-014, WA 1-9 (January 2014).
http://nepis.epa. gov/Exe/ZvPURL.cgi?Dockev=P100EWVL.txt.
220 Howard, B., "Honda Fit: 86 mpg from the next hyper-efficient hybrid," Extreme Tech, August 8, 2013. Retrieved
from http://www.extremetech.com/extreme/163462-honda-fit-86-mpg-from-the-next-hyper-efficient-hybrid.
221 Kok, D., Ford Motor Company, "Power of Choice: The Role of Hybrid Vehicle Technology in Meeting
Customer Expectations," The Battery Show 2014, September 18, 2014.
222 Duren, A., A123 Systems, "48V Battery System Design for Mild Hybrid Applications," SAE Hybrid & Electric
Vehicle Technologies Symposium, February 2016.
223 Bosch, "The hybrid for everyone: Bosch's 48-volt system makes sense even in compact vehicles," Press Release,
September 4, 2015. Retrieved April 1, 2016 from http://www.bosch-
presse.de/presseforum/details.htm?txtID=7384&locale=en.
224 MacRae, C., "February 2015 management briefing: 48V mild hybrids (3)," retrieved from http://www.just-
auto.com/analysis/48v-mild-hybrids-3_idl56298.aspx.
225 Fiat Chrysler Automobiles, "Business Plan Update 2014-2018," January 27, 2016. Retrieved from
http://www.fcagroup.com/en-
US/investor_relations/events_presentations/quarterly_results_presentations/FCA_2014_18 BusinessPlanUpdate.
pdf.
226 Sanchez, E., "Ram, Jeep Hybrids Revealed in FCA Future Business Plan," Truck Trend Network, January 27,
2016. Retrieved from http://www.trucktrend.com/news/1601-ram-jeep-hybrids-revealed-in-fca-future-business-
plan/.
227 Irwin, J., "Supplier Schaeffler in 48V Mild-Hybrid Vanguard," Wards Auto, January 15, 2016. Retrieved from
http://wa.rdsauto.com/technologv/supplier-schaeffler-48v-m.ild-hvbrid-vangiiard.
228 Voelcker, J., "2016 Hyundai Tucson Shows Two Different Hybrid Concepts In Geneva," Green Car Reports,
March 17, 2015. Retrieved from http://www.greencarreports.com/news/1097268
229 Harrop, P., "48V vehicle systems becoming significant," IDTechEx.com, August 12, 2015. Retrieved from
http://www.electricvehiclesresearch.com/articles/8266/48v-vehicle-systems-becoming-significant
230 Wiesenberger, J., Continental, "The Evolving EV and Hybrid Roadmap," The Battery Show 2015, Novi,
Michigan, September 15, 2015.
231 Stenzel, U., "48V Mild Hybrid Systems: Market Needs and Technical Solutions," AVL Powertrain Engineering,
AVL UK Expo 2014. Retrieved March 7, 2016 from
https://www.avl.com/documents/10138/1379144/AVL_UK_Expo_48V_Mild_Hybrid_Systems.pdf.
232 Brooke, L., "Ford accelerates research on 48-V mild hybrid systems," SAE International, February 12, 2015.
Retrieved from http://art.icles.sae.org/13908/.
233 Bosch, "The hybrid for everyone: Bosch's 48-volt system makes sense even in compact vehicles," Press Release,
September 4, 2015. Retrieved April 1, 2016 from http://www.bosch-
presse.de/pressefornm/details.htm7t: 84&locale=en.
234 Berman, B., "Hybrid Car Affordability Leaps Forward with P2 Technology," April 19, 2011. Retrieved from
http://www.hybridcars.com/hybrid-car-affordability-leaps-forward-p2-technology-29761/.
235 Truett, R. et al., "Electric turbocharger eliminates lag, Valeo says," Automotive News, August 3, 2014. Retrieved
from http://www.autonews.eom/article/20140803/OEM10/308049992/electric-turbocharger-eliminates-lag-valeo-
says.
236 Hanlon, M., "Controlled Power Technologies shows 48V electric supercharger," Gizmag, October 4, 2011.
Retrieved from http://www.gizmag.com/controlled-power-technologies-48v-electric-vtes-supercharger/20037/.
2-436
-------
Technology Cost, Effectiveness, and Lead Time Assessment
237 EPA, "Regulatory Impact Analysis: Final Rulemaking for 2017-2025 Light-Duty Vehicle Greenhouse Gas
Emission Standards and Corporate Average Fuel Economy Standards," 2012, p. 3-52. See Docket ID EPA-HQ-
OAR-2010-0799, docket item number EPA-HQ-OAR-2010-0799-12013.
238 Gehm, R., "Inside Honda's new two-motor 50-mpg Accord Hybrid," SAE International, December 18, 2013.
Retrieved from http://artides.sae.org/12666/.
239 Sessions, R., "2015 Toyota Camry SE Hybrid," Car and Driver, January 2015. Retrieved from
http://www.caranddriver.com/reviews/2015-toyota-camry-hybrid-test-review.
240 Green Car Congress, "Toyota details powertrain advances in Gen4 Prius; available E-Four system for all-wheel
drive (not for US)," October 13, 2015. Retrieved from http://www.greencarcongress.com/2015/10/20151Q13-
prius.html.
241 "Light Duty Technology Cost Analysis, Power-Split and P2 Hybrid Electric Vehicle Case Studies," Report EPA-
420-R-l 1-015, November 20ll.https://nepis.epa.gov/Exe/ZyPDF.cgi/P100EGlR.PDF?Dockey=P 100EG1R.PDF..
242 German, J., "Hybrid Vehicles Technology Development and Cost Reduction," Technology Brief, No. 1, July
2015, ICCT, p. 6.
243 National Academy of Sciences (NAS), National Research Council (NRC), "Overcoming Barriers to Deployment
of Plug-in Electric Vehicles," NAS Committee on Overcoming Barriers to Electric-Vehicle Deployment, 2015.
244 New Qualified Plug-in Electric Drive Motor Vehicles, 26 U.S.C. § 30D.
245 See https://www.fueleconomy.gov/feg/taxphevb.shtml.
246 Cobb, J., "How Long Does the 2017 Chevy Bolt Have Before Federal Credits Begin Fading Away?"
HybridCars.com, January 13, 2016. Retrieved on January 27, 2016 from http://www.hybridcars.com/how-long-does-
the-2017-chevy-bolt-have-before-federal-credits-begin-fading-away/.
247 Shelton, S., "US-Bound BMW 330e and X5 xDrive40e PHEVs Bow In LA," HybridCars.com, November 20,
2015. Retrieved from http://www.hybridcars.com/us-bound-bmw-330e-and-x5-xdrive40e-phevs-bow-in-la/.
248 Cobb, J., "Exclusive: 2017 Mitsubishi Outlander PHEV Postponed To Next Summer," HybridCars.com, July 28,
2016. Retrieved from http://www.hybridcars.com/exclusive-2017-mitsubishi-outlander-phev-postponed-to-next-
summer/
249 Berman, B., "New Target Date for Outlander PHEV Introduction: Spring 2016", Retrieved April 25, 2016 from
http://www.plugincars.com/new-target-date-outlander-phev-introduction-spring-2016-130848.html.
250 Halvorson, B., "2017 Chrysler Pacifica Hybrid: More Details On 30-Mile Plug-In," Jan. 20, 2016,
http://www.greencarreports.eom/news/1101960_2017-chrysler-pacifica-hybrid-more-details-on-30-mile-plug-in.
251 Cole, J., "Upcoming Honda PHEV To Have 40 Miles Of Electric Range," http://insideevs.com/upcoming-honda-
phev-40-miles-electric-range/.
252 General Motors, "The Results are In: More Range for the 2016 Volt," Press Release, August 4, 2015.
http://media.chevrolet.com/media/us/en/gm/news.detail.html/content/Pages/news/us/en/2015/aug/0804-volt-
range.html.
253 California Air Resources Board, "CCR Section 1962.2: 2018 and Subsequent Model Year Requirements," July
2014. Downloaded on May 26, 2016 from http://www.arb.ca.gOv/msprog/zevprog/zevregs/.l.962.2 Ctean.pdf.
254 World Bank, "The China New Energy Vehicles Program: Challenges and Opportunities," Report 61259, April
2011. Available at http://www-
wds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2011/04/20/000356161_20110420054039/Rend
ered/PDF/612590WPOPRTMO1BOX3 58342B01PUBLIC11 .pdf.
255 International Council on Clean Transportation, "Renewed and Enhanced Subsidies for Energy-Efficient Vehicles
in China," Policy Update, November 2013. Available at
http://www.theicct.org/sites/default/files/publications/ICCTupdate_CNsubsidies_nov2013.pdf.
256 Anderman, M., "xEV Market Drivers and Trends; the Role of Regulations, Incentives, and Technology,"
presented at 2015 Advanced Automotive Batteries Conference, June 17, 2015.
257 Cadillac, "Cadillac CT6 Plugln Sets New Electric Luxury EV Performance Benchmark," Press Release, April 19,
2015. Retrieved on 10/23/2015 from
http://media.cadillac.eom/media/cn/en/cadillac/news.detail.html/content/Pages/news/cn/en/2015/april/0419_PHEV.h
tml.
258 Cole, J., "Washington Expands $3,100 Electric Vehicle Incentive Program." Retrieved on April 19, 2016 from
http://insideevs.com/washington-launches-3100-electric-vehicle-incentive-program/.
259 Grewe, T., "Generation Two Voltec Drive System," presented at SAE Hybrid and Electric Vehicle Symposium,
February 2015.
260 See Table 3.5-25 of RIA, 2017-2025 FRM.
261 Green Car Congress, "Navigant Research Leaderboard puts.LG Chem as leader for
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Li-ion batteries for transportation," November 25, 2015. Retrieved from
http://www.greencarcongress.eom/2015/l 1/20151125-navigant.html.
262 Navigant Research, "Navigant Research Leaderboard Report: Lithium Ion Batteries for Transportation,"
Summary Page. Retrieved from https://www.navigantresearch.com/research/navigant-research-leaderboard-report-
lithium-ion-batteries-for-transportation.
263 Tesla Motors, "Tesla's Comments to the Draft Technical Assessment Report," submitted to Docket EPA-HQ-
OAR-2015-0827, September 26, 2016.
264 AutoBlog.com, "Analyst predicts 30-80k Chevy Bolt sales in first year," May 9, 2016. Retrieved from
http://www.autoblog.eom/2016/05/09/2017-chevy-bolt-sales-prediction/
265 Voelcker, J., "2017 Chevy Bolt EV sales could go as high as 80,000, says analyst," Green Car Reports, May 12,
2016. Retrieved from http://www.greencarreports.eom/news/l 103922_2017-chevy-bolt-ev-sales-could-go-as-high-
as-80000-says-analyst.
266 Hunt, T., "Why Tesla's Model 3 Could Be the New Model T," GreenTechTedia.com, April 11, 2016. Retrieved
from https://www.greentechmedia.com/articles/read/Why-Teslas-Model-3-Could-Be-The-New-Model-T
267 Consumer Federation of America, "New Data Shows Consumer Interest in Electric Vehicles Is Growing," Press
Release, September 19, 2016.
268 Korosec, K., "GM Puts Self-Driving Car Push Into a Higher Gear," Fortune, January 29, 2016. Retrieved from
http://fortune.com/2016/01/29/gm-self-driving-car-division/.
269 Korosec, K., "GM Is Giving Lyft Drivers First Dibs on Its New All-Electric Chevy Bolt," Fortune, July 11, 2016.
Retrieved from http://fortune.com/2016/07/11/gm-lyft-drivers-chevy-bolt/.
270 Davies, A., "We Take a Ride in the Self-Driving Uber Now Roaming Pittsburgh," Wired, September 14, 2016.
Retrieved from https://www.wired.com/2016/09/self-driving-autonomous-uber-pittsburgh/
271 Gardner, G., "The self-driving revolution will be mostly electric," Detroit Free Press, September 21, 2106.
Retrieved from http://www.freep.eom/story/money/cars/2016/09/18/self-driving-revolution-mostly-
electric/90410520/.
272 InsideEVs.com, "Monthly Plug-In Sales Scorecard," retrieved on November 2, 2016 from
http://insideevs.com/monthly-plug-in-sales-scorecard/.
273 Partial Consent Decree, in re: Volkswagen "Clean Diesel" Marketing, Sales Practices, and Products Liability
Litigation, Case 3:15-md-02672-CRB Document 1605-1 Filed 06/28/16. Retrieved on November 7, 2016 from
https://www.epa.gov/sites/production/files/2016-06/documents/vwpartialsettlement-cd.pdf.
274 "Fixing America's Surface Transportation Act—Designation of Alternative Fuel Corridors," Federal Register,
Vol. 81, No. 141, Friday, July 22, 2016. See also: http://www.fhwa.dot.gov/environment/alternative_fuel_corridors/.
275 Falkenberg-Hull, E., "Automotive Minute: Nationwide electric vehicle charging network expanding," Atlanta
Business Chronicle, September 20, 2016. Retrieved from
http://www.bizjournals.com/atlanta/news/2016/09/20/automotive-minute-nationwide-electric-vehicle.html.
276 Kurylko, D. T., "BMW will boost i3's range," Automotive News, January 18, 2016. Downloaded on Jan. 19,
2016 from http://www.autonews.com/article/20160118/OEM05/301189994/bmw-will-boost-i3s-range.
277 BMW Group, "The new 2017 BMW i3 (94 Ah): More range paired to high-level dynamic performance," Press
Release, May 2, 2016. Retrieved on May 2, 2016 from
https://www.press.bmwgroup.eom/usa/article/detail/T0259560EN_US/the-new-2017-bmw-i3-94-ah-more-range-
paired-to-high-level-dynamic-performance.
278 Autoblog, "VW e-Golf will get 30% range boost thanks to improved batteries," January 9, 2016. Downloaded on
Jan. 19, 2016 from http://www.autoblog.com/2016/01/09/volkswagen-e-golf-30-percent-range-increase-new-
batteries/.
279 Ford Motor Company, "Ford Investing $4.5 Billion in electrified Vehicle Solutions, Reimagining How To Create
Future Vehicle User Experiences," Press Release, December 10, 2015. Retrieved on April 19, 2016 from
https://media.ford.eom/content/fordmedia/fna/us/en/news/2015/12/10/ford-investing-4-5-billion-in-electrified-
vehicle-solutions.pdf.
280 Hyundai Motor America, "2017 Hyundai Ioniq Model Lineup Makes U.S. Debut at New York International Auto
Show," Press Release, March 23, 2016. Retrieved on May 5, 2016 from
http://www.hyundainews.com/us/en/print/45159.
281 Tesla Motors website, Model 3 reservation page. Retrieved on April 19, 2016 from
https://www.teslamotors.com/model3.
282 Bunkley, N., "Ford plans EV to compete with Chevy Bolt, Tesla Model 3, Fields confirms," Automotive News,
April 28, 2016. Retrieved on November 2, 2016 from
2-438
-------
Technology Cost, Effectiveness, and Lead Time Assessment
http://www.autonews.eom/article/20160428/OEM05/160429821/fields-confirms-ford-planning-ev-to-compete-with-
chevy-bolt-tesla.
283 Taylor, M., "Everything we know about Volkswagen's next Golf," retrieved November 2, 2016 from
http://www.autoblog.eom/2016/02/05/volkswagen-golf-eighth-generation-information/
284 Davies, A., "Audi's Plan to Make an Electric SUV with a 300-Mile Range," Wired, August 14, 2015. Retrieved
on November 2, 2016 from https://www.wired.eom/2015/08/audis-plan-make-electric-suv-300-mile-range/.
285 Tesla Motors, "New Tesla Model S Now the Quickest Production Car in the World," August 23, 2016. Retrieved
on November 2, 2016 from https://www.tesla.com/blog/new-tesla-model-s-now-quickest-production-car-world.
286 Ayre, J., "Tesla CEO Elon Musk: 100 kWh Battery Pack Probably As Large As We're Going," Clean Technica,
September 16, 2016. Retrieved on November 2, 2016 from https://cleantechnica.com/2016/09/16/tesla-ceo-elon-
musk-100-kwh-battery-pack-probably-large-going/.
287 Tesla Motors, "Roadster 3.0 Battery Upgrade." Retrieved on November 2, 2016 from
http://shop.teslamotors.com/products/roadster-3-0-upgrade.
288 Tesla Motors Press Release, "Tesla Model S Sales Exceed Target," March 31, 2013.
http://www.teslamotors.com/de_AT/blog/tesla-model-s-sales-exceed-target.
289 Konekamp, A., "The Tesla Model S Battery: A Battery Pack Analysis Study;" AVL List GmbH, 2015 Advanced
Automotive Battery Conference, Detroit MI June 15-19, 2015.
290 Rugh, J., Hovland, V. and Andersen, S., "Significant Fuel Savings and Emissions Reductions by Improving
Vehicle Air Conditioning," National Renewable Energy Laboratory, NREL/CP-5400-62232, presented at the 15th
Annual Earth Technologies Forum and Mobile Air Conditioning Summit, April 15, 2004.
291 Kim, N. et al., "Modeling Electrified Vehicles Under Different Thermal Conditions," Argonne National
Laboratory, presented at Society of Automotive Engineers Thermal Management Systems Symposium, September
29-October 1, 2015.
292 Chaney, L., et al., "Climate Control Load Reduction Strategies for Electric Drive Vehicles in Cold Weather,"
National Renewable Energy Laboratory, presented at Society of Automotive Engineers Thermal Management
Systems Symposium, September 29-October 1, 2015.
293 Chen, K., Bozeman, J., Wang, M., Ghosh, D. et al., "Energy Efficiency Impact of Localized Cooling/Heating for
Electric Vehicle," SAE Technical Paper 2015-01-0352, 2015, doi: 10.4271/2015-01-0352.
294 Bunkley, N., "Ford plans EV to compete with Chevy Bolt, Tesla Model 3, Fields confirms," Automotive News,
April 28, 2016. Retrieved on November 2, 2016 from
http://www.autonews.eom/article/20160428/OEM05/160429821/fields-confirms-ford-planning-ev-to-compete-with-
chevy-bolt-tesla.
295 Hall, L., "2017 Renault Zoe Getting New 200-Mile Battery - Nissan Leaf to Follow?" HybridCars.com,
September 28, 2016. Retrieved from http://www.hybridcars.eom/2017-renault-zoe-getting-new-200-mile-battery-
nissan-leaf-to-follow/.
296 Malliaris, A.C., Hsia, H., and Gould, H.; "Concise description of auto fuel economy and performance in recent
model years," Society of Automotive Engineers, Paper 760045 (1976).
297 U.S. EPA, Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975
Through 2015, Report EPA-420-R-15-016, December 2015, p. 35.
298 Green Car Congress, "Kia using SK Innovation NCM Li-ion cells in Soul EV," February 24, 2014. Retrieved
from http://www.greencarcongress.com/2014/02/20140224-kia.html.
299 Howell, D., "U.S. DOE Electric Drive Vehicle Battery R&D Impacts, Progress, and Plans," AABC 2015, Detroit
MI, June 2015.
300 Weissler, P., "Electronics diverge in engineering Ford's hybrid C-Max and plug-in Energi," SAE International,
January 9, 2013. Retrieved May 5, 2016 from http://articles.sae.0rg/l 1705/
301 Kessen, Jeff, and Duren, Angela, "System Design Solutions for 48V Lithium-Ion Batteries," AABC 2015,
AABTAM Session 2, June 18, 2015.
302 Nuspl, G. et al., "Developing Battery Materials for Next Generation Applications," The Battery Show 2015, Novi
MI, 2015.
303 Oury, A., General Motors, "2016 Chevrolet Malibu Hybrid Battery Pack," presented at AABC 2015, Detroit
Michigan, June 18, 2015.
304 Voelcker, J., "2016 Toyota Prius: A Few Details On Engine, Hybrid System Released," Green Car Reports,
October 12, 2015. Retrieved on May 5, 2016 from http://www.greencarreports.eom/news/1100436_2016-toyota-
prius-a-few-details-on-engine-hybrid-sy stem-released.
305 Tajitsu, N. et al.,"Warming to lithium-ion, Toyota charges up its battery options," Reuters, October 30, 2016.
Retrieved on November 2, 2016 from http://www.reuters.com/article/us-toyota-batteries-idUSKBN12U0ZH
-------
Technology Cost, Effectiveness, and Lead Time Assessment
306 Sion Power, "Sion Power's Licerion High Energy Batteries," The Battery Show 2015, Novi, MI, September 2015.
307 PR Newswire, "Anesco and OXIS to Release Lithium Sulfur Battery Storage by 2016," Press Release, July 14,
2015. Retrieved from http://www.prnewswire.com/news-releases/anesco-and-oxis-to-release-lithium-sulfur-battery-
storage-by-2016-514802521 .html.
308 Morris, C., "OXIS Energy achieves 300 Wh/kg with lithium-sulfur cell," Charged Magazine, Nov/Dec 2014, p.
21. Retrieved from https://chargedevs.com/newswire/oxis-energy-achieves-specific-energy-of-300-whkg-with-
lithium-sulfur-cell/.
309 Ruoff, C., "Tesla Tweaks its Battery Chemistry: A Closer Look at Silicon Anode Development," Charged
Magazine, Jul/Aug 2015, p. 30.
310 Sawers, P., "Dyson acquires Sakti3 for $90M to help commercialize 'breakthrough' solid-state battery tech."
Retrieved on April 1, 2016 from http://venturebeat.eom/2015/10/19/dyson-acquires-sakti3-for-90m-to-help-
commercialize-breakthrough-solid-state-battery-tech/.
311 Vaughan, A and Carrington, D., "Dyson developing an electric car, according to government documents," The
Guardian, March 23, 2016. Retrieved on April 1, 2016 from
http://www.theguardian.com/environment/2016/mar/23/dyson-developing-electric-car-government-documents.
312 Seeo, "Bosch Has Groundbreaking Battery Technology for Electric Vehicles." Retrieved on April 19, 2016 from
http://www.seeo.com/news/bosch-has-groundbreaking-battery-technology-for-electric-vehicles/
313 Brooke, L., "Battery guru: Future of 18650 cells unclear beyond Tesla S," SAE International, February 17, 2014.
Retrieved on May 5, 2016 from http://articles.sae.org/12833/.
314 Business Wire, "Samsung SDI to Supply Cylindrical EV Batteries to JAC Motors," November 23, 2015.
Retrieved from http://www.businesswire.eom/news/home/20151123005451/en/Samsung-SDI-Supply-Cylindrical-
EV-Batteries-JAC.
315 Anderman, M., "Battery Packs of Modern xEVs: A Comprehensive Engineering Assessment: Extract," slide 18,
Total Battery Consulting, 2016.
316 Element Energy, "Cost and Performance of EV Batteries: Final Report for The Committee on Climate Change,"
March 21, 2012. Retrieved from http://www.element-energy.co.uk/wordpress/wp-content/uploads/2012/06/CCC-
battery-cost_-Element-Energy-report_March2012_Finalbis.pdf.
317 Faguy, Peter, "Overview of the DOE Advanced Battery R&D Program," presented at the 2015 U.S. DOE Vehicle
Technologies Office Annual Merit Review and Peer Evaluation, June 8, 2015.
318 Yeow, K. et al., "3D Thermal Analysis of Li-ion Battery Cells with Various Geometries and Cooling Conditions
Using Abaqus," 2012 SIMULIA Community Conference.
319 Brooke, L., "GM unveils more efficient 2016 Volt powertrain," October 29, 2014. Retrieved on May 5, 2016
from http://articles.sae.org/13666/.
320 Nissan, "2016 Nissan LEAF Press Kit: Overview." Retrieved on May 5, 2016 from http://nissannews.com/en-
US/nissan/usa/presskits/us-2016-nissan-leaf-press-kit.
321 Kwon, Jason; "Samsung's Automotive Battery Strategy," Samsung SDI, AABC 2015, June 18, 2015.
322 Kia Motors, "Advanced battery pack for new Soul EV," Press Release, February 24, 2014. Retrieved April 4,
2016 from https://press.kia.com/eu/press/products/14_02_25_22%20-%20geneva%20soul%20ev%20batteries/.
323 Kia Motors America, "Kia Emergency Response Guide: Soul EV," September 2014.
324 Green Car Congress, "New 2016 Nissan LEAF with available 30 kWh pack for 107 milerange," September 10,
2015. Retrieved from http://www.greencarcongress.eom/2015/09/20150910-leaf.html.
325 Schmitt, B., "2nd gen Leaf expected 2018: 60kWh NMC battery, 300 mile range, autonomous, CFRP." Retrieved
January 27, 2016 from http://dailykanban.eom/2015/10/2nd-gen-leaf-expected-2018-60kwh-nmc-battery-300-mile-
range-autonomous-cfrp/.
326 Green Car Reports, "Next Nissan Leaf confirmed for 60-kwh battery, 200 miles of range," June 22, 2016.
Retrieved on November 2, 2016 from http://www.greencarreports.eom/news/1104610_next-nissan-leaf-confirmed-
for-60-kwh-battery-200-miles-of-range.
327 Halvorson, B., "Nissan's 60-kWh, 200-Mile Battery Pack: What We Know So Far," Green Car Reports,
November 5, 2015. Retrieved from http://www.greencarreports.eom/news/1100775_nissans-60-kwh-200-mile-
battery-pack-what-we-know-so-far.
328 General Motors, "Drive Unit and Battery at the Heart of Chevrolet Bolt EV," Press Release, Jan. 11, 2016.
Retrieved on January 27, 2016 from
http://media.chevrolet.com/media/us/en/chevrolet/home.detail.html/content/Pages/news/us/en/2016/Jan/naias/chevy/
0111-bolt-du.html.
2-440
-------
Technology Cost, Effectiveness, and Lead Time Assessment
329 Markel, T., National Renewable Energy Laboratory, "Plug-In HEV Vehicle Design Options and Expectations,"
NREL/PR-540-40630, presented at ZEV Technology Symposium, California Air Resources Board, Sacramento CA,
September 27, 2006.
330 McGrath, P., "Advanced Management and Protection of Energy Storage Devices (AMPED)," presented at The
Battery Show 2015, Novi MI, September 17, 2015.
331 Brown, Carlton; "Generation 2 Lithium-Ion Battery Systems: Technology Trends andKPIs," Robert Bosch
Battery Systems, 2015 Advanced Automotive Battery Conference, Detroit MI June 15-19, 2015.
332 "Informal Testing of 2013 Volkswagen Jetta Hybrid Battery Usage," Memo to Docket EPA-HQ-OAR-2015-
0827, June 28, 2016.
333 Smith, K. et al., "Predictive Models of Li-ion Battery Lifetime," National Renewable Energy Laboratory, 2015
Advanced Automotive Battery Conference, Detroit MI June 15-19, 2015.
334 "2016 Chevrolet Volt Battery," 2015 Advanced Automotive Battery Conference, Detroit MI June 18, 2015.
335 ANL Advanced Powertrain Research Facility (APRF), http://www.anl.gov/energy-systems/group/downloadable-
dynamometer-database.
336 Noland, D.," Life With Tesla Model S: Battery Degradation Update," Green Car Reports, October 26, 2015.
Retrieved from http://www.greencarreports.eom/news/l 100603_life-with-tesla-model-s-battery-degradation-
update/page-2.
337 Idaho National Laboratory, "Vehicle Testing - Light Duty - BEV" (website). See https://avt.inl.gov/vehicle-
type/bev.
338 MyKiaSoulEV.com, topic on High Voltage Battery System. See
http ://www. mykiasoulev. com/forum/viewtopic .php?p= 1923.
339 BMW Group, "BMW i3 Technical Data." See
http://www.bmw.eom/com/en/newvehicles/i/i3/2015/showroom/technical_data.html.
340 Idaho National Laboratory, "Battery Pack Laboratory Testing Results, 2014 BMW i3 EV - VIN 5486," retrieved
from https://avt.inl.gov/sites/default/files/pdf/fsev/batteryi5486.pdf.
341 Idaho National Laboratory, "Battery Pack Laboratory Testing Results, 2014 BMW i3 EV - VIN 5626," retrieved
from https://avt.inl.gov/sites/default/files/pdf/fsev/batteryi5626.pdf.
342 Idaho National Laboratory, "Battery Pack Laboratory Testing Results, 2014 BMW i3 EV - VIN 5655," retrieved
from https://avt.inl.gov/sites/default/files/pdf/fsev/batteryi5655.pdf.
343 Idaho National Laboratory, "Battery Pack Laboratory Testing Results, 2014 BMW i3 EV - VIN 5658," retrieved
from https://avt.inl.gov/sites/default/files/pdf/fsev/batteryi5658.pdf.
344 General Motors, "Brownstown Battery Assembly Expands Capabilities, Will build battery system for 2015
Chevrolet Spark EV," Press Release, May 14, 2014. Retrieved from
http://media.chevrolet.com/media/us/en/chevrolet/news.detail.html/content/Pages/news/us/en/2014/May/0514-
brownstown. html.
345 Uchida, T., "Analysis on Customers' Usage Data of Electric Vehicles," Honda R&D Co. Ltd. Automobile R&D
Center, 2015 Advanced Automotive Battery Conference, Detroit MI, June 15-19, 2015.
346 Lee, J., "Integrated Cabin and Battery Thermal Management System," Hanon Systems, presented at SAE 2015
Thermal Management Systems Symposium, Troy, MI, Sept. 30, 2015.
347 Wawzyniak, M., "Battery Thermal Management Architectures and Components," MAHLE Behr GmbH & Co
KG, presented at SAE 2015 Thermal Management Systems Symposium, Troy, MI, Sept. 30, 2015.
348 Jeckel, A.; Beste, F., and Glossman, T., "Thermal Management of the High Voltage Battery," Daimler AG,
presented at SAE 2015 Thermal Management Systems Symposium, Troy, MI, Sept. 30, 2015.
349 Gross, O. and Clark, S., "Battery Thermal Management in xEVs: Session Introduction," Fiat Chrysler
Automobiles, presented at AABC 2015, Detroit MI, June 2015.
350 Bower, G., "Chevy Bolt 200 Mile EV Battery Cooling and Gearbox Details," InsideEVs.com, January 18, 2016.
Retrieved from http://insideevs.com/chevy-bolt-200-mile-ev-battery-cooling-and-gearbox-details-bower/.
351 Doggett, S., "Leafs Batteries 'More Primitive' Than Those in Tesla's First Prototype, Musk Says," Edmunds Auto
Observer, August 10, 2010. Retrieved from http://www.edmunds.eom/autoobserver-archive/2010/08/leafs-batteries-
more-primitive-than-those-in-teslas-first-prototype-musk-says.htm.l.
352 Siry, D., "In Race to Market, Nissan's Electric Car Takes Shortcuts," Wired.com, January 22, 2010. Retrieved
from http://www.wired.com/2010/0 l/nissan-leaf-2./.
353 Moloughney, T., "No Active Thermal Management: Did Nissan Make the Right Call?" Retrieved from
http://www.plugincars.com/no-active-thermal-management-did-nissan-make-right-call.html.
2-441
-------
Technology Cost, Effectiveness, and Lead Time Assessment
354 Halvorson, B., "Nissan Leaf Likely To Get 'Hot Weather' Battery In Next Year," Green Car Reports, November
26, 2013. Retrieved from http://www.greencarreports.com/news/1088715_nissan-leaf-likely-to-get-hot-weather-
battery-in-next-year.
355 Wood, S., "Leveraging Automotive Lithium-Ion Technology for Commercial and Industrial Applications,"
Johnson Controls, presented at AABC 2015, Detroit MI, June 2015.
356 Porsche, "Porsche Mission E," Press Release, September 14, 2015. Retrieved from
http://newsroom.porsche.com/en/products/iaa-2015-porsche-mission-e-mobility-all-electrically-concept-car-
11391 html
357 See Nelson, P. et al. "Modeling the Performance and Cost of Lithium-Ion Batteries for Electric Drive Vehicles,"
p. 19 and p. 23.
358 Wu, Y., "Assembly Processes for Lithium-Ion Batteries, 14.1.3. Coating Process," in Lithium-Ion Batteries:
Fundamentals and Applications, Volume 4 of Electrochemical Energy Storage and Conversion, CRC Press (2015),
p. 483.
359 Gallagher, K., "PHEV and EV Battery Performance and Cost Assessment," Argonne National Laboratory, 2015
U.S. Department of Energy Vehicle Technologies Office Annual Merit Review and Peer Evaluation Meeting, June
2015.
360 Kia Motors, "Advanced Battery Pack for Kia Soul EV," Press Release, February 24, 2014. Retrieved on January
27, 2016 from http://www.kianewscenter.com/news/global-news/advanced-battery-pack-for-kia-soul-ev-
/s/45548fc8-7279-4da3-81b2-d30df557420c.
361 Kane, M., "Samsung SDI Presents Batteries That Enable 370 Miles (600 km) Of Range At 2016 NAIAS,"
retrieved January 27, 2016 from http://insideevs.eom/samsung-sdi-presents-batteries-that-enable-370-miles-600-km-
of-range-at-2016-naias/.
362 Samsung SDI, "Samsung SDI Aims to Drive Battery Business Growth in North American Automobile Market
With Its Top Technology Leadership," Press Release, Business Wire, January 11, 2016. Retrieved January 27, 2016
from http://www.businesswire.com/news/home/20160111005560/en/Samsung-SDI-Aims-Drive-Battery-Business-
Growth.
363 Rauwald, C., "VW to Add Flat Batteries to Boost Sales of Electric Vehicles," Bloomberg Business, December
13, 2015. Retrieved on January 27, 2016 from http://www.bloomberg.com/news/articles/2015-12-13/vw-
developing-new-technology-to-integrate-flat-batteries-in-cars-ii4eee5d.
364 Nelson, Paul A.; Ahmed, Shabbir; Gallagher, Kevin G., and Dees, Dennis W., "Cost Savings for Manufacturing
Lithium Batteries in a Flexible Plant," Pre-publication draft, Argonne National Laboratory, 2015.
365 Corrigan, D., "XALT Energy Lithium-Ion Batteries for Heavy Duty Vehicle and Marine Applications," AABC
2015, June 18, 2015.
366 Paul, J., "Impact of Standardized Module Design - Commercial PEV and Second Life ESS Applications," AABC
2015, June 18, 2015.
367 Cobb, J.,"'Not a compliance car,' GM says 2017 Chevy Bolt can meet demand of over 50,000 per year,"
HybridCars.com, January 14, 2016. Downloaded on January 27, 2016 from http://www.hybridcars.com/not-a-
compliance-car-gm-says-2017-chevy-bolt-production-capacity-exceeds-50000-per-year/.
368 Roland Berger, "Consolidation Across the Li-Ion Battery Market Gaining Speed," Press Release, October 19,
2012. Retrieved from http://www.rolandberger.us/news/2012-10-19-Li-ion-Batteries-Bubble.html.
369 Ramsey, M., "Auto Industry's Ranks of Electric-Car Battery Suppliers Narrow," Wall Street Journal, August 20,
2015.
370 Green Car Congress, "Navigant Research ranks LG Chem as top automotive Li-ion battery manufacturer," June
25, 2013. Retrieved from http://www.greencarcongress.com/2013/06/navigant-20130625.html.
371 Bathion, Michael; "Wanxiang Wins U.S. Approval to Buy Battery Maker A123," Bloomberg News, January 30,
2013. Retrieved Oct. 28, 2015 from http://www.bloomberg.com/news/articles/2013-01-29/wanxiang-wins-cfius-
approval-to-buy-bankrupt-battery-maker-al23.
372 Business Wire, "LG Chem, Panasonic, and Samsung SDI Score Highest in Assessment of Lithium Ion Battery
Manufacturers, According to Navigant Research," Press Release, December 3, 2015. Retrieved from
http://www.businesswire.com/news/home/20151203005014/en/.
373 Ogura, K., "LG Chem, Tesla tie-up could jolt Panasonic," Nikkei Asian Review, October 28, 2015. Retrieved
from http://asia.nikkei.eom/B usiness/Deals/LG-Chem-Tesla-tie-up-could-jolt-Panasonic?page=l
374 Business Wire, "Samsung SDI to Acquire Magna International's Battery Pack Business," Press Release, February
23, 2015. Retrieved on May 6, 2016 from http://www.businesswire.com/news/home/20150223005533/en/Samsung-
SDI-Acquire-Magna-International%E2%80%99s-Battery-Pack.
2-442
-------
Technology Cost, Effectiveness, and Lead Time Assessment
375 Pfanner, E. and Landers, P.; "Nissan Considers Shift to LG Chem Batteries," The Wall Street Journal, July 16,
2015. Retrieved on May 6, 2016 from https://www.linkedin.com/pulse/nissan-considers-shift-lg-chem-batteries-
john-e-halajko.
376 Kim, Y., "Tesla approaches LG, Samsung for Model 3 batteries," The Korea Times, May 17, 2016. Retrieved on
May 19, 2016 from http://www.koreatimes.co.kr/www/news/tech/2016/05/133_204903.html.
377 Evans, K., "The Future of Electric Vehicles: Setting the Record Straight on Lithium Availability," Journal of
Energy Security, August 27 2009.
378 Boston Consulting Group, "Batteries for Electric Cars: Challenges, Opportunities, and the Outlook to 2020,"
January 7, 2010.
379 Gaines, L. and Cuenca, R., "Costs of Lithium-Ion Batteries for Vehicles," Argonne National Laboratory
ANL/ESD-42, May 2000.
380 Hiscock, J., "Australian lithium miners in pole position as price nears sweet spot," Nikkei Asian Review, July 28,
2016. Retrieved from http://asia.nikkei.com/Business/Trends/Australian-lithium-miners-in-pole-position-as-price-
nears-sweet-spot?page= 1.
381 Ralph, J., "Hard Rock Miners Set to Plug Supply Gap," Commodities Research Unit (CRU), May 4, 2016.
Retrieved from http://www.crugroup.com/about-cru/cruinsight/Hard_Rock_Miners_Set_to_Plug_Supply_Gap
382 Stafford, J., "Tesla And Other Tech Giants Scramble For Lithium As Prices Double," OilPrice.com, April 12,
2016. Retrieved on May 5, 2016 from http://oilprice.com/Energy/Energy-General/Tesla-And-Other-Tech-Giants-
Scramble-For-Lithium-As-Prices-Double.html.
383 Stafford, J.," Why Lithium Will See Another Price Spike This Fall," OilPrice.com, July 18, 2016. Retrieved
from http://oilprice.com/Energy/Energy-General/Why-Lithium-Will-See-Another-Price-Spike-This-Fall.html
384 West, J., "Tesla Motors Inc ignites a lithium race among Albemarle, SQM and FMC," Financial Post, April 4,
2016. Retrieved on May 5, 2016 from http://business.financialpost.com/midas-letter/tesla-motors-inc-ignites-a-
lithium-race-among-albemarle-sqm-and-fmc.
385 Clarke, G. M., "Lithium Resource Availability - A Question of Demand," Proceedings 9th Advanced Automotive
Battery Conference (AABC), June 12, 2009, Long Beach CA.
386 Evans, K., "The Future of Electric Vehicles: Setting the Record Straight on Lithium Availability," Journal of
Energy Security, August 27 2009.
387 Carnegie Mellon University, "Lithium market fluctuations unlikely to significantly impact battery prices," Press
Release, May 3, 2016. Retrieved on May 5, 2016 from
https://engineering.cmu.edu/media/press/2016/05_03_lithium_market_fluctuations.html.
388 U.S. EPA, "Joint Technical Support Document: Final Rulemaking for 2017-2025 Light-duty Vehicle Greenhouse
Gas Emission Standards and Corporate Average Fuel Economy," EPA-420-R-12-901, August 2012.
389 Spoken comments by Mark Reuss, Executive Vice President of Global Product Development, GM 2015 Global
Business Conference, Oct 1 2015, slide 52 in 2015_GBC_Combined_PDF_v3.pdf. See Docket Memo "Transcript of
Comments on Bolt EV Battery Cell Cost at GM 2015 Global Business Conference," EPA Docket EPA-HQ-OAR-
2015-0827.
390 General Motors, "General Motors 2015 Global Business Conference," Presentation, October 1, 2015, slide 52 in
2015_GB C_Combined_PDF_v3 .pdf.
391 Cole, J., "GM: Chevrolet Bolt Arrives In 2016, $145/kWh Cell Cost, Volt Margin Improves $3,500,"
InsideEVs.com, October 2, 2015. Retrieved from http://insideevs.com/gm-chevrolet-bolt-for-2016-145kwh-cell-
cost-volt-margin-improves-3 500/.
392 Kalhammer, F. et al., "Status and Prospects for Zero Emissions Vehicle Technology - Report of the ARB
Independent Expert Panel 2007," Sacramento, California, US: State of California Air Resources Board, 2007.
393 U.S. Advanced Battery Consortium, "USABC Goals for Advanced Batteries for EVs - CY 2020
Commercialization," Downloaded on December 28, 2015 from
http://www.uscar.org/commands/files_do wnload.php?files_id=364.
394 Tataria, H. and Lopez, H. A., "EV Battery Development (Envia Systems)," in Section III. A. 1 p. 31 of "Fiscal Year
2013 Annual Progress Report for Energy Storage R&D," U.S. Department of Energy, DOE/EE-1038, February
2014.
395 Keller, G., "DOE's Efforts to Develop Hybrid Powertrain technologies for Heavy-Duty Vehicles," SAE 2015
Hybrid & Electric Vehicle Technologies Symposium, February 12, 2015.
396 Supplementary spreadsheet accompanying Nykvist, B. and Nilsson, M.; "Rapidly Falling Costs of Battery Packs
for Electric Vehicles," Nature Climate Change, March 2015; doi: 10.1038/NCLIMATE2564, available at
http://www.nature.com/nclimate/journal/v5/n4/extref/nclimate2564-sl.xlsx.
2-443
-------
Technology Cost, Effectiveness, and Lead Time Assessment
397 Voelcker, J., "Nissan Leaf New Battery Cost: $5,500 For Replacement With Heat-Resistant Chemistry," Green
Car Reports, June 28, 2014. Retrieved from http://www.greencarreports.com/news/1092983_nissan-leaf-battery-
cost-5500-for-replacement-with-heat-resistant-chemistry.
398 Brocknwn, B., "Update on Nissan LEAF Battery Replacement." MYNissanLeaf.com, June 27, 2014. Retrieved
from http://www.mynissanleaf.com/viewtopic.php?f=4&t= 17168.
399 Cole, J., "Nissan Prices LEAF Battery Replacement at $5,499, New Packs More Heat Durable," InsideEVs.com,
June 27, 2014. Retrieved from http://insideevs.com/breaking-nissan-prices-leaf-battery-replacement-5499-new-
packs-heat-durable/.
400 Voelcker, J., "Nissan Leaf $5,500 Battery Replacement Loses Money, Company Admits," Green Car Reports,
July 24, 2014. Retrieved from http://www.greencarreports.eom/news/1093463_nissan-leaf-5500-battery-
replacement-loses-money-company-admits.
401 Cobb, J., "How Long Will An Electric Car's Battery Last?" HybridCars.com, April 30, 2014. Retrieved from
http://www.hybridcars.com/how-long-will-an-evs-battery-last/.
402 NewGMParts.com. See http://stores.revolutionparts.eom/newgmparts.com/chevrolet/volt/20979876/2011-
year/base-trim/l-41-14-electric-gas-engine/electrical-cat/electrical-components-scat/?part_name=battery-assy
403 InsideEVs.com, "BMW i3 Battery Module Costs $1,715.60 - 8 Modules Per Car - Total Cost $13,725," January
26, 2015. Retrieved from http://insideevs.com/bmw-i3-battery-module-costs-1715-60-8-modules-per-car-total-cost-
13725/.
404 Tesla Motors, "Roadster 3.0 Battery Upgrade," retrieved from
http://shop.teslamotors.com/collections/roadster/products/roadster-3-0-upgrade.
405 Ramsey, M., "Tesla Gets Boost From Korean Battery Maker LG Chem," The Wall Street Journal, October 28,
2015. Retrieved from http://www.wsj.com/articles/tesla-gets-boost-from-korean-battery-maker-lg-chem-
1446007554.
406 Smart USA, Battery Assurance Plus Brochure. Retrieved on May 6, 2016 from
http://www.smartusa.com/resources/doc/smart_battery-assurance-plus_SEO.pdf.
407 PluginCars.com, "Smart Electric Drive Review." Retrieved from http://www.plugincars.com/smart-ed
408 White, J., "The Smart, a Very Cheap Electric Car With a Very Expensive Battery," The Wall Street Journal,
August 7, 2013. Retrieved from http://blogs.wsj.com/corporate-intelligence/2013/08/07/the-smart-a-very-cheap-
electric-car-with-a-very-expensive-battery/.
409 Kane, M., "UK Pricing Reveals 30 kWh 2016 Nissan LEAF Costs Just £1,600 More Than 24 kWh Version,"
InsideEVs.com, September 14, 2015. Retrieved from http://insideevs.eom/uk-pricing-reveals-30-kwh-2016-nissan-
leaf-costs-just-1600-more-than-24-kwh-version/.
410 Lux Research, "Tesla Motors' Gigafactory Will See More Than 50% Overcapacity in its Li-ion Production,"
Press Release, September 3, 2014. Retrieved from http://www.luxresearchinc.com/news-and-events/press-
releases/read/tesla-motors%E2%80%99-gigafactory-will-see-more-50-overcapacity-its-li.
411 National Research Council Canada, "Full-Scale Wind Tunnel Investigation of Light-Duty Vehicle
Aerodynamics," NR AL-2014-0017, March 2014.
412 Larose, G., Belluz, L., Whittal, I., Belzile, M. et al., "Evaluation of the Aerodynamics of Drag Reduction
Technologies for Light-duty Vehicles: a Comprehensive Wind Tunnel Study," SAE Int. J. Passenger Cars - Mech.
Syst. 9(2):2016, doi: 10.4271/2016-01-1613.
413 National Research Council Canada "Evaluation of the aerodynamics of new drag reduction technologies for
light-duty vehicles: results of Phase III" LTR-AL-2015-0061, February 23, 2016.
414 U.S. EPA, "EPA Staff Informal Survey of Aerodynamic Technologies Represented at the
2015 North American International Auto Show (NAIAS)," memo to Docket EPA-HQ-OAR-2015-0827, May 16,
2016.
415Audi Technology Portal, http://www.audi-technology-portal.de/en/body/aerodynamics-aeroacoustics/underbody
416 2017 Prius, http://www.toyota.com/prius/ebrochure/.
417 Arai, M., "Development of the Aerodynamics of the New Nissan Murano," Society of Automotive Engineering
(SAE) Technical Paper #2015-01-1542, 2015.
418 "2015 Nissan Murano-10 Fast Facts", April 14, 2014, http://nissannews.com/en-CA/nissan/canada/channels/ca-
canada-nissan-2014-NYIAS/releases/2015-nissan-murano-10-fast-facts.
419 "Evaluation of the aerodynamics of new drag reduction technologies for light duty vehicles: Summary of Phases
I to IV', National Research Council Canada and Transport Canada, December 2016.
420 Madsen, A., "2015 Acura TLX Body Structure Review," Honda R&D Americas, Inc., 2015 Great Designs in
Steel Conference, Livonia, MI, May 13, 2015.
421 PR Newswire, "General Motors Commemorates 30 Years of Aerodynamic Progress," August 4, 2010.
-------
Technology Cost, Effectiveness, and Lead Time Assessment
422 "Sophisticated Sophomore: All-New 2016 Chevrolet Craze," GM Corporate Newsroom, June 24, 2015,
http://media.chevrolet.eom/media/us/en/gm/news.detail.print.html/content/Pages/news/us/en/2015/jun/0624-
cruze.html.
423 "How air curtains on F-150 help reduce aerodynamic drag and aid fuel efficiency", Ford Motor Company Media
Center, July 15, 2015, https://media.ford.com/content/fordmedia/fna/us/en/news/2015/07/15/how-air-curtains-on-f-
150-help-reduce-aerodynamic-drag.html.
424 "How air curtains on F-150 help reduce aerodynamic drag and aid fuel efficiency", Ford Motor Company Media
Center, July 15, 2015, https://media.ford.coni/content/fordmedia/fiia/ns/en/news/2015/07/15/tiow-a.ir-curtains-on-f-
150-help-reduce-aerodytiamic-drag.html.
425 "Active Grille Shutters 2015 F-150, September 15, 2015, https://www.youtube.com/watch?v=23O-hS-r0gQ.
426 "How Ford Squeezed Every Drop of Aerodyamics Out of the 2015 F150", August 26, 2014,
http://trackyeah.jalopnik.eom/how-a-boxy-track-like-the-2015-ford-f-150-reduces-wind-1627125563.
427 National Research Council Canada, "Evaluation of the aerodynamics of new drag reduction technologies for
light-duty vehicles: results of Phase II," LTR-AL-2015-0273, Issued for Client's Comments, March 22, 2015.
428 Chappell, L., "For mpg gains, tiremakers deliver - consumers shrug," Automotive News, December 1, 2014.
429 Automotive World, "From radial to radical - where next for light vehicle tyres?," March 2015.
430 Bridgestone, "Bridgestone's 'ologic technology' hits the road," Press Release, January 8, 2014. Retrieved on
March 25, 2016 from http://www.bridgestone.eu/corporate/press-releases/2014/01/bridgestones-ologic-technology-
hits-the-road/.
431 Donley, T., "Improving Vehicle Fuel Efficiency Through Tire Design, Materials, and Reduced Weight," Cooper
Tire & Rubber Co., presentation at 2014 DOE Annual Merit Review, Project VSS083.
432 "Light Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975 Through
2015, Executive Summary," U.S. EPA, EPA-420-S-15-001 December 2015, littps://www.epa.gov/fuel-
econo my/t rends-report..
433 "GMC Introduces All-New 2017 Acadia," January 12, 2016,
http://media.gmc.com/media/us/en/gmc/home.detail.html/content/Pages/news/us/en/2016/Jan/naias/gmc/0112-
acadia.html.
434 Chrysler's T&C minivan renamed "Pacifica," cuts 250 lbs. with steel, aluminum, magnesium," January 12, 2016,
"http://www.repairerdrivennews.eom/2016/01/12/chryslers-tc-minivan-renamed-pacifica-cuts-250-lbs-with-
aluminum-steel-magnesium/.
435 Ducker Worldwide, "2015 North American Light Vehicle Aluminum Content Study, Executive Summary," June
2014, http://www.drivealiiin.iniiin.Org/research-resoiirces/PDF/ResearcIi/201.4/2014-diicker-report.
436 Composites Forecasts and Consulting, December 2015.
437 Docket submittal of email from Abey Abraham of Ducker Worldwide to Cheryl Caffrey, 2/25/2016.
438 Baron,J., PhD, "Assessing the Fleet-wide Material Technology and Costs to Lightweight Vehicles", Center for
Automotive Research, September 2016, http://www.cargroup.org/?module=Publications&event=View&pubID=141
439 "Alcoa's Micromill Technology for future cars", Aluminum Insider, 06 April 2016,
http://aluminiuminsider.com/alcoas-micromill-technology-for-future-cars/.
440 "GM says it's first to weld steel to aluminum", The Detroit News, May 20, 2016,
http://www.detroitnews.com/story/business/autos/general-motors/2016/05/18/gm-says-first-weld-steel-
aluminum/84558702/.
441 U.S. EPA, "Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975
through 2016," EPA-420-R-16-010, November 2016.
442 http://www.freep.eom/story/money/cars/mark-phelan/2015/06/24/2016-chevrolet-craze-chevy-
challenges/71265088/.
443 http://www.worldautosteel.org/why-steel/steel-muscle-in-new-vehicles/2016-chevy-malibu-larger-lighter-more-
efficient-with-hss/.
444 "Cadillac XT5's new platform cuts weight- at less cost," SAE Automotive Engineering, April 2016.
445 "The New Audi Q7- Sportiness, efficiency, premium comfort," December 12, 2014
https://www.audiusa.com/newsroom/news/press-releases/2014/12/the-new-audi-q7-sportiness-efficiency-premium-
comfort.
446 Mg Showcase Issue 12, International Magnesium Association, Spring 2010,
http ://www. intlmag.org/sho wcase/mgshowcase 12_mar 10 .pdf.
447 "Response to Peer Reviews for 'Light-Duty Track Weight Reduction Study with Crash Model, Feasibility and
Cost Analysis - Project Number: T8009-130016', September 24, 2015.
2-445
-------
Technology Cost, Effectiveness, and Lead Time Assessment
448 "Light-Duty Vehicle Mass Reduction and Cost Analysis -Midsize Crossover Utility Vehicle", EPA-420-R-12-
026, August 2012.
449 General Motors, "General Motors 2015 Global Business Conference," Presentation, October 1, 2015, Slides 43-
45 in document, https://www.gm.com/content/dam/gin/events/docs/5194074-596155-ChartSet-10-l-2015.
450 "Identifying Real World Barriers to Implementing Lightweighting Technologies and Challenges in Estimating
the Increase in Costs," Center for Automotive Research, Jay Baron, PhD, January 2016
http://www.cargronp.org/? moduie=Publications&event=View&pii L
451 "Mass Reduction and Cost Analysis - Light Duty Pickup Truck Model years 2020-2025, FEV North America,
Inc., EPA-420-R-15-006, June 2015.
https://nepis.epa.gov/Exe/ZyPDF.cgi/P100MS0E.PDF?Dockey=P100MS0E.PDF .
452 "Evaluating the Structure and Crashworthiness of a 2020 Model-year, Mass Reduced Crossover Vehicle using
FEA Modeling," California Air Resources Board and Lotus Engineering, August 31, 2012.
453 "An Assessment of Mass Reduction Opportunities for a 2017 - 2020 Model Year Vehicle Program," Lotus
Engineering, Inc. for ICCT, March 2010.
454 "Venza Aluminum BIW Concept Study," April 2013, http://www.drivealuminum.org/research-
resources/PDF/Research/2013/venza-biw-full-study.
455 Conklin, J., Beals, R., and Brown, Z., "BIW Design and CAE," SAE Technical paper 2015-01-0408, 2015,
doi: 10.4271/2015-01-0408.
456 "Evaluating the Structure and Crashworthiness of a 2020 Model-Year, Mass-Reduced Crossover Vehicle Using
FEA Modeling," Lotus Engineering for California Air Resources Board, August 31, 2012. https://www.epa.gov/air-
pollution-transportation/evaluating-structure-and-crashworthiness-2020-model-year-mass-reduced
457 "Venza Aluminum BIW Concept Study," April 2013, http://www.drivealuminum.org/research-
resources/PDF/Research/2013/venza-biw-full-study.
458 Source: Kelkar et al: "Automobile Bodies: Can Aluminum Be an Economical Alternative to Steel?," August 2001
Issue of JOM., 53 (8) (2001)pp. 28-32, http://www.tms.org/pubs/journals/JOM/0108/Kelkar-0108.html.
459 " http://energy.gov/sites/prod/files/2015/12/f27/06%20-%20Lightweight%20Materials.pdf.
460 Skszek, T., Conklin, J, Zaluzec, M, Wagner, D.,"Multi-Material Lightweight Vehicles: Mach-II Design," U.S.
DOE, Ford, Vehma, June 17, 2014, http://energy.gov/sites/prod/files/2014/07/fl7/lm088_skszek_2014_o.pdf.
461 Mascarin,A., Hannibal,T., Raghunathan, A., Ivanic, Z., Francfort, J., "Vehicle Lightweighting: 40% and 45%
Weight Savings Analysis: Technical Cost Modeling for Vehicle Lightweighting," INL/EXT-14-33863, April 2015,
http://avt.inl.gov/pdf/TechnicalCostModel40and45PercentWeightSavings.pdf.
462 Skszek, T., Zaluzec, M., Conklin, J., and Wagner D., "MMLV: Project Overview," SAE Technical Paper 2015-
01-0407, 2015 doi: 10.4271/2015-01-0407.
463 Bolar, N., Buchler, T., Li, A., and Wallace, J., "MMLV: Vehicle Durability Design, Simulation and Testing,"
SAE Technical Paper 2015-01-1613, 2015, doi: 10.4271/2015-01-1613.
464 Bushi, L., Skszek, T., and Wagner, D., "MMLV: Life Cycle Assessment," SAE Technical Paper 2015-01-1616,
2015, doi: 10.4271/2015-01-1616.
465 Chen, X., Conklin, J., Carpenter, R., Wallace, J. et al., "MMLV: Chassis Design and Component Testing," SAE
Technical Paper 2015-01-1237, 2015, doi: 10.4271/2015-01-1237.
466 Chen, Y., Board, D., Faruque, O., Stancato, C. et al, "MMLV: Crash Safety Performance," SAE Technical Paper
2015-01-1614, 2015, doi: 10.4271/2015-01-1614.
467 Conklin, L., Beals, R., and Brown, Z., "BIW Design and CAE," SAE Technical Paper 2015-01-0408, 2015, doi:
104271/2015-01-0408.
468 Corey, N., Madin, M., and Williams, R., "MMLV: Carbon Fiber Composite Engine Parts," SAE Technical Paper
2015-01-1239, doi: 10.4271/2015-01-1239.
469 Gur, Y., Pan, J., Huber, J., and Wallace, J., "MMLV: NVH Sound Package Development and Full Vehicle
Testing," SAE Technical Paper 2015-01-1615, 2015, doi: 10.4271/2015-01-1615.
470 Jaranson, J., and Ahmed, M., "MMLV: Lightweight Interior Systems Design," SAE Technical Paper 2015-01-
1236, 2015, doi: 10.4271/2015-01-1236.
471 Kearns, J., Park, S., Sabo, J., andMilacic, D., "MMLV: Automatic Transmission Lightweighting," SAE
Technical Paper 2015-01-1240, 2015, doi: 10.4271/2015-01-1240.
472 Maki, C., Byrd, K., McKeough, B., Rentschler, R. et al., "MMLV: Aluminum Cylinder Block with Bulkhead
Inserts and Aluminum Alloy Connecting Rod," SAE Technical Paper 2015-01-1238, 2015, doi: 10.4271/2015-01-
1238.
473 Plourde, L., Azzouz, M., Wallace, J., and Chellman, M., "MMLV: Door Design and Component Testing," SAE
Technical Paper 2015-01-0409, 2015, doi: 10.4271/2015-01^0|^.
-------
Technology Cost, Effectiveness, and Lead Time Assessment
474 Schutte, C., "DOE Focuses on Developing Materials to Improve Vehicle Efficiency," SAE Technical Paper
2015-01-0405, 2015, doi: 10.4271/2015-01-0405.
475 Skszek,T., Conklin, J., Wagner, D., Zaluzec.M., "Multi-Material Lightweight Vehicles," Ford Motor Company
and Vehma International, June 11, 2015, presented at 2015 DOE AMR,
http://energy.gOv/sites/prod/files/2015/06/f24/lm072_skszek_2015_o.pdf.
476 IBIS Associates, Inc., "Technical Cost Modeling for Vehicle Lightweighting: 40% and 45% Weight Reduction,"
June 11, 2015, http://energy.gov/sites/prod/files/2015/06/f24/lm090_mascarin_2015_o.pdf.
477 Chapter 6 Lightweight Materials within the "Vehicle Technologies Office: 2015 Annual Merit Review Report"
http://energv.gov/sites/prod/files/2015/12/f27/06%20-%20Lightweight%20Materials.pdf.
478 IBIS Associates, Energetics Incorporated, Idaho National Laboratory, "Vehicle Lightweighting: Mass Reduction
Spectrum analysis and Process Cost Modeling Project ID #LM090", June 7, 2016,
http://energy.gOv/sites/prod/files/2016/06/f33/lm090_mascarin_2016_o_web.pdf.
479 "Update to Future Midsize Lightweight Vehicle Findings in Response to Manufacturer Review and IIHS Small
Overlap Testing," NHTSA, DOT HS 812 237, Februaiy 2016.
http://www.nhtsa.gov/staticfiles/rulemaking/pdf/cafe/812237 LightWeightVehicleReport.pdf.
480 "Light-Duty Truck Weight Reduction Study with Crash Model, Feasibility and Cost Analysis," T8009-130016,
Department of Transport Canada, September 24, 2015.
481 "Novel Malleable Covalent Networks and Their applications in Repairable Carbon Fiber Reinforced Composites
with Full Recyclability", Wei Zhang, Department of Chemistry and biochemistry at the University of Colorado,
Boulder, Global Automotive Lightweight Materials Conference, August 23-25, 2016, Detroit.
482 MALLINDA Reshaping the Plastics Industry, https://mallinda.squarespace.com/about/.
483 "Mallinda awarded $750k grant for resusable carbon-fiber composite", October 12, 2016,
http://www.colorado.edu/nvc/2016/10/12/mallinda-awarded-750k-grant-reusable-carbon-fiber-composite.
484 http://www.cyclotronroad.org/.
485 "ORNL seeking U.S. manufacturers to license low-cost carbon fiber process", March 22 2016,
https://www.ornl.gov/news/ornl-seeking-us-manufacturers-license-low-cost-carbon-fiber-process.
486 "ORNL seeking U.S. manufacturers to license low-cost carbon fiber process", March 22 2016,
https://www.ornl.gov/news/onil-seeking-ns-maiinfacturers-license-low-cost-carbon-fiber-process.
487 Brooke, L. "Systems Engineering a new 4x4 benchmark," SAE Automotive Engineering, June 2, 2014
488 Phelps, P., "EcoTrac Disconnecting AWD System," presented at 7th International CTI Symposium North
America 2013, Rochester MI.
489 Pilot Systems, "AWD Component Analysis," Project Report, performed for Transport Canada, Contract T8080-
150132, May 31, 2016.
490 Martin, B. et al., "The Innovative driveline of the 9-Speed Jeep Cherokee," presented at 8th International CTI NA
Symposium, May 2014, Rochester, MI.
491 Lee, B., "A Novel Clutch Solution for AWD Disconnect," presented at 9th International CTI Symposium North
America 2015, Rochester, MI.
492 U.S. Environmental Protection Agency, Greenhouse Gas Emission Standards for Light Duty Vehicles,
Manufacturer Performance Report for the 2014 Model Year, EPA-420-R-15-026, December 2015.
493 U.S. Environmental Protection Agency, Greenhouse Gas Emission Standards for Light Duty Vehicles,
Manufacturer Performance Report for the 2015 Model Year, EPA-420-R-16-014, November 2016.
494 General Motors, "Request for GHG Credit for Variable Crankcase Suction Valve Technology ," Letter to Line
Wehrly, December 9, 2014. Available at https://www.epa.gov/sites/production/files/2016-09/documents/gm-sas-
compressor-off-cycle-credit-petition-fe6349-2014-12-09.pdf.
495 U.S. EPA, "EPA Decision Document: Off-cycle Credits for Fiat Chrysler Automobiles, Ford Motor Company,
and General Motors Corporation," EPA-420-R-15-014, September 2015. Available at
https://nepis.epa. gov/Exe/ZyPDF.cgi/P100N19E.PDF?Dockey=P100N19E.PDF.
496 Sciance, F. et al., "Developing the AC17 Efficiency Test for Mobile Air Conditioners," SAE Technical Paper
2013-01-0569, April 2013.
497 See Greenhouse Gas Emission Standards for Light-Duty Vehicles: Manufacturer Performance Report for the
2015 Model Year, EPA-420-R-16-014, November 2016.
498 Minnesota Pollution Control Agency, Mobile Air Conditioner Leakage Rates, www.pca.state.mn.us.
499 http://www.greencarcongress.com/2015/10/20151020-mbco2.html.
500 Andersen et al., 2015. "Secondary Loop Motor Vehicle Air Conditioning Systems (SL—MACs). Using Low-
Global Warming Potential (GWP) Refrigerants in Leak-Tight Systems In Climates with High Fuel Prices and Long,
2-447
-------
Technology Cost, Effectiveness, and Lead Time Assessment
Hot and Humid Cooling Seasons. Building on the Previous Success of Delphi, Fiat, General Motors, Volvo, Red
Dot, SAE Cooperative Research Projects, And Other Engineering Groups." MACS Briefing, 2015.
501 77 FR 62832, October 15, 2012.
502 77 FR 62835-62837.
503 77 FR 62835-62836.
504 40 CFR 86.1869-12.
505 40 CFR 86.1869-12(b).
506 40 CFR 86.1869-12(c).
507 40 CFR 86.1869-12(d).
508 For a description of each technology and the derivation of the pre-defined credit levels see Chapter 5 of "Joint
Technical Support Document: Final Rulemaking for 2017-2025 Light-duty Vehicle Greenhouse Gas Emission
Standards and Corporate Average Fuel Economy," EPA-420-R-12-901, August 2012.
509 40 CFR 86.1869-12 (b)(4).
510 U.S. Environmental Protection Agency, Greenhouse Gas Emission Standards for Light Duty Vehicles,
Manufacturer Performance Report for the 2015 Model Year, EPA-420-R-16-014, November 2016.
511 "EPA Decision Document: Mercedes-Benz Off-cycle Credits for MYs 2012-2016," U.S. EPA-420-R-14-025,
Office of Transportation and Air Quality, September 2014. See http://www.epa.gov/otaq/regs/ld-
hwy/greenhouse/documents/420r14025 .pdf.
512 "EPA Decision Document: Off-cycle Credits for Fiat Chrysler Automobiles, Ford Motor Company, and General
Motors Corporation," U.S. EPA-420-R-15-014, Office of Transportation and Air Quality, September 2015. See
https://www.epa.gov/vehicle-and-engine-certification/compliance-information-light-duty-greenhouse-gas-ghg-
standards
513 Alternative Method for Calculating Off-cycle Credits Under the Light-Duty Vehicle Greenhouse Gas Emissions
Program: Applications from BMW Group, Ford Motor Company, General Motors Corporation, and Volkswagen
Group of America, Federal Register 81 (2 September 2016): 60694.
514 National Academy of Sciences. 2011. Assessment of Fuel Economy Technologies for Light-Duty Vehicles.
National Academies Press, ISBN 978-0-309-15607-3, http://www.nap.edu/catalog.php?record_id=12924, p.25
515 NAS, 2011, p. 62 footnote.
516 National Academy of Sciences. 2015. Cost, Effectiveness and Deployment of Fuel Economy
Technologies for Light-Duty Vehicles. National Academies Press, ISBN 978-0-309-37388-3,
http://www.nap.edu/21744. p.281.
517 NAS, 2011, p. 46.
518 Energy and Environmental Analysis, Inc. 2007. Update for Advanced Technologies to Improve Fuel Economy of
Light Duty Vehicles. Prepared for U.S. Department of Energy. August. Arlington, Va (EPA-HQ-OAR-2014-0827-
0722).
519 NAS, 2011, p. 62 footnote.
520 http://www.fordservicecontent.eom/Ford_Content/Catalog/owner_information/2017-F-150-Owners-Manual-
version- l_om_EN -U S 08 2016.pdf.
521 https://iaspnb.epa.gov/otaqpnb/.
522 National Academies of Sciences. "Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles," The National Academies Press, 2015 (EPA-HQ-OAR-2015-0827-DRAFT-0398).
523 National Academy of Sciences. "Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles," The National Academies Press, 2015 (EPA-HQ-OAR-2015-0827-DRAFT-0398).
524 https ://www3. epa.gov/otaq/climate/mte. htm#epa-publications.
525 Nelson, P. A. Gallagher, K.G., Bloom, I., and Dees, D.W., "Modeling the Performance and Cost of Lithium-Ion
Batteries for Electric Drive Vehicles," Second Edition, Argonne National Laboratory, ANL-12/55 (December 2012).
526 Gallagher, K., Shabbir, A., Nelson, P., and Dees, D., "PHEV and EV Battery Performance and Cost Assessment,"
Argonne National Laboratory, presented at the 2015 U.S. DOE Vehicle Technologies Office Annual Merit Review
and Peer Evaluation, June 9, 2015.
527 Cite to EPA report containing our Executive Summary followed by the final learning report from ICF.
528 "Learning Curves used in Developing Technology Costs in the Draft TAR," Memorandum from Todd Sherwood
to Air Docket EPA-HQ-OAR 2015-0827, June 8, 2016.
529 "Learning Curves used in Developing Technology Costs in the Draft TAR," Memorandum from Todd Sherwood
to Air Docket EPA-HQ-OAR 2015-0827, June 8, 2016.
530 RTI International, "Automobile Industry Retail Price Equivalent and Indirect Cost Multipliers," February 2009;
EPA-420-R-09-003; http://www.epa.gov/otaq/ld-hwy/420rQ^0p^pdf.
-------
Technology Cost, Effectiveness, and Lead Time Assessment
531 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.
532 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); "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.
533 Vyas, A., D. Santini and R. Cuenca; "Comparison of indirect Cost Multipliers for Vehicle Manufacturing;" April
2000.
534 "Joint Technical Support Document: Final Rulemaking for 2017-2025 Light-Duty Vehicle Greenhouse Gas
Emission Standards and Corporate Average Fuel Economy Standards," EPA-420-R-12-901, August 2012.
535 National Academy of Sciences. "Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles," The National Academies Press, 2015 (EPA-HQ-OAR-2015-0827-DRAFT-0398).
536 See https://www.epa.gov/otaq/climate/mte.htm.
537 See https://www.epa.gov/otaq/climate/data-testing.htm.
538 Newman, K., Kargul, J., and Barba, D., "Benchmarking and Modeling of a Conventional Mid-Size Car Using
ALPHA," SAE Technical Paper 2015-01-1140, 2015, doi: 10.4271/2015-01-1140.
539 Novation Analytics, Technology Effectiveness - Phase II: Vehicle-Level Assessment, version 1.0, prepared for
Alliance of Automobile Manufacturers & Association of Global Automakers, September 20, 2016, p. 44.
540 Novation Analytics, Technology Effectiveness - Phase I: Fleet-Level Assessment, version 1.1, prepared for
Alliance of Automobile Manufacturers & Association of Global Automakers, October 19, 2015, p. 58.
541 EPA-HQ-OAR-2015-0827-0899.
542 See https://www.epa.gov/otaq/climate/alpha.htm.
543 Newman, K., Kargul, J., and Barba, D., "Development and Testing of an Automatic Transmission Shift Schedule
Algorithm for Vehicle Simulation," SAE Int. J. Engines 8(3): 1417-1427, 2015, doi: 10.4271/2015-01-1142.
544 Lee, S., Lee, B., McDonald, J. and Nam, E., "Modeling and Validation of Lithium-Ion Automotive Battery
Packs," SAE Technical Paper 2013-01-1539, 2013, doi: 10.4271/2013-01-1539.
545 Lee, S., Cherry, J., Lee, B., McDonald, J., Safoutin, M., "HIL Development and Validation of Lithium Ion
Battery Packs," SAE Technical Paper 2014-01-1863, 2014, doi: 10.4271/2014-01-1863.
546 Newman, K., Doorlag, M., and Barba, D., "Modeling of a Conventional Mid-Size Car with CVT Using ALPHA
and Comparable Powertrain Technologies," SAE Technical Paper 2016-01-1141, 2016, doi: 10.4271/2016-01-1141.
547 Novak, J. and Blumberg, P., "Parametric Simulation of Significant Design and Operating Alternatives Affecting
the Fuel Economy and Emissions of Spark-Ignited Engines," SAE Technical Paper 780943, 1978,
doi: 10.4271/780943.
548 Heywood, J. B.,1988, Internal Combustion Engine Fundamentals, McGraw Hill, ISBN 007028637, p. 676.
549 Patton, K., Nitschke, R., and Heywood, J., "Development and Evaluation of a Friction Model for Spark-Ignition
Engines," SAE Technical Paper 890836, 1989, doi: 10.4271/890836.
550 Sandoval, D. and Heywood, J., "An Improved Friction Model for Spark-Ignition Engines," SAE Technical Paper
2003-01-0725, 2003, doi:10.4271/2003-01-0725.
551 EPA Test Car List data files, https://www.epa.gov/compliance-and-fuel-economy-data/data-cars-used-testing-
fuel-economy Accessed April 11, 2016.
552 Argonne National Labs Downloadable Dynamometer Database, http://www.anl.gov/energy-
svstems/group/downloadable-dynamo meter-database/conventional-vehicles Accessed April 12, 2016.
553 Baldauf, R.W., Devlin, R.B., Gehr, P., Giannelli, R., Hassett-Sipple, B., Jung, H., Martini, G., McDonald, J.,
Sacks, J.D., Walker, K. "Ultrafine Particle Metrics and Research Considerations: Review of the 2015 UFP
Workshop." International Journal of Environmental Research and Public Health, 2016.
554 Lee, S., Schenk, C., and McDonald, J., "Air Flow Optimization and Calibration in High-Compression-Ratio
Naturally Aspirated SI Engines with Cooled-EGR," SAE Technical Paper 2016-01-0565, 2016, doi: 10.4271/2016-
01-0565.
555 Ellies, B., Schenk, C., and Dekraker, P., "Benchmarking and Hardware-in-the-Loop Operation of a 2014
MAZDA SkyActiv2.0L 13:1 Compression Ratio Engine," SAE Technical Paper 2016-01-1007, 2016,
doi: 10.4271/2016-01-1007.
556 Kargul, J., Moskalik, A., Barba, D., Newman, K. et al., "Estimating GHG Reduction from Combinations of
Current Best-Available and Future Powertrain and Vehicle Technologies for a Midsized Car Using EPA's ALPHA
Model," SAE Technical Paper 2016-01-0910, 2016, doi: 10.4271/2016-01-0910.
2-449
-------
Technology Cost, Effectiveness, and Lead Time Assessment
557 Patton, et al. "Aggregating Technologies for reduced Fuel Consumption: A Review of the Technical Content in
the 2002 National Research Council Report on cafe." SAE 2002-01-0628. Society of Automotive Engineers, 2002.
558 National Academy of Sciences. 2015. Cost, Effectiveness and Deployment of Fuel Economy
Technologies for Light-Duty Vehicles. National Academies Press, ISBN 978-0-309-37388-3,
http://www.nap.edu/21744, p.2.
559 Wilcutts, M., Switkes, J., Shost, M., and Tripathi, A., "Design and Benefits of Dynamic Skip Fire Strategies for
Cylinder Deactivated Engines," SAE Int. J. Engines 6(l):278-288, 2013, doi: 10.4271/2013-01-0359.
560 Schamel, A., Scheidt, M., Weber, C. Faust, H. Is Cylinder Deactivation a Viable Option for a Downsized 3 -
Cylinder Engine? Vienna Motor Symposium, 2015.
561 Hitomi, M. "Our Direction for ICE (Internal Combustion Engine) - Consideration of Engine Displacement."
Vienna Motor Symposium, 2015.
562 Hitomi, M. "Mazda's Approach for Developing Engines from a Perspectiv of Environmental Improvement."
Presentation at the 2015 ERC Symposium. Last accessed on the Internet on 11/20/2016 at the following URL:
https://www.erc.wisc.edu/documents/sympl5/2015_Mazda_-_Hitomi_-Apprd_revised.pdf.
563 Tisshaw, M. " Next-generation Mazda engines to eclipse electric cars on emissions." Autocar, March 2014. Last
accessed on the Internet on 11/20/2016 at the following URL: http://www.autocar.co.uk/car-news/industry/next-
generation-mazda-engines-eclipse-electric-cars-emissions.
564 Fletcher, G. " Future iterations of Mazda's trademark technology will further refine the internal combustion
engine." Driving, June 2014. Last accessed on the Internet on 11/20/2016 at the following URL:
http://driving.ca/mazda/auto-news/news/mazdas-skyactiv-looks-to-wring-out-more-efficiency.
565 Stuhldreher, M., "Fuel Efficiency Mapping of a 2014 6-Cylinder GM EcoTec 4.3L Engine with Cylinder
Deactivation," SAE Technical Paper 2016-01-0662, 2016, doi:10.4271/2016-01-0662.
SAE Technical Paper 2016-01-0662, 2016, doi: 10.4271/2016-01-0662.
566 Schamel, A., Scheidt, M., Weber, C. Faust, H. "Is Cylinder Deactivation a Viable Option for a Downsized 3-
Cylinder Engine?" Vienna Motor Symposium, 2015.
567 Alger, T., Gingrich, J., Mangold, B., and Roberts, C., "A Continuous Discharge Ignition System for EGR Limit
Extension in SI Engines," SAE Int. J. Engines 4(l):677-692, 2011, doi: 10.4271/2011-01-0661.
568 Alger, T., Mangold, B., Roberts, C., and Gingrich, J., "The Interaction of Fuel Anti-Knock Index and Cooled
EGR on Engine Performance and Efficiency," SAE Int. J. Engines 5(3): 1229-1241, 2012, doi: 10.4271/2012-01-
1149.
569 Sasaki, Y., Adachi, S., Nakata, K., Tanei, K., Shibuya, S. "The new Toyota 1.0L L3 ESTEC gasoline engine."
35th International Vienna Motor Symposium, 2014.
570 Alt, M.; Sutter, T., Johnen, T., Fulton, K., Daily, R., Cococcetta, R., K., Peralta, N., Damen, M., O'Daniel, G.,
Noe, A., Krischker, U. " Powerful, efficient and smooth: The new Small Gasoline Engine family from General
Motors." 35th International Vienna Motor Symposium, 2014.
571Hwang, K., Hwang, I., Lee, H., Park, H. et al., "Development of New High-Efficiency Kappa 1.6L GDI Engine,"
SAE Technical Paper 2016-01-0667, 2016, doi: 10.4271/2016-01-0667.
572 Aiyoshizawa, E., Hori., K. "The New Nissan High Efficient 4-Cylinder 1.6L GDI turbocharged engine with low
pressure EGR - Evolution for Lower Fuel Consumption combined with High Output Performance." 35th
International Vienna Motor Symposium, 2014.
573 Glahn, C. Kluin, M. Konigstein, A., Cloos, L. "Cooled External EGR - System Optimization of the Cooling and
Charging System on a 3-Cylinder Gasoline DIT/C Engine." 24th Aachen Colloquium Automobile and Engine
Technology 2015.
574 See Docket EPA-HQ-OAR-2015-0827, Docket item titled "Air Flow Optimization and Calibration in High-
Compression-Ratio Naturally Aspirated SI Engines with Cooled EGR."
575 EPA, "Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975
through 2015," EPA-420-R-15-016, December 2015.
576 Cruff, L., Kaiser, M., Krause, S., Harris, R. et al., "EBDI® - Application of a Fully Flexible High BMEP
Downsized Spark Ignited Engine," SAE Technical Paper 2010-01-0587, 2010, doi: 10.4271/2010-01-0587.
577 Kaiser, M., Krueger, U., Harris, R., and Cruff, L., ""Doing More with Less" - The Fuel Economy Benefits of
Cooled EGR on a Direct Injected Spark Ignited Boosted Engine," SAE Technical Paper 2010-01-0589, 2010,
doi: 10.4271/2010-01-0589.
578 King, J., Heaney, M. Saward, J., Fraser, A., Criddle, M. Cheng, T., Morris, G., Bloore, P. "HyBoost: An
Intelligently Electrified Optimised Downsized Gasoline Engine Concep.t' SAE-China and FISITA (eds.),
Proceedings of the FISITA 2012 World Automotive Congress, Lecture Notes in Electrical Engineering 191,
DOI: 10.1007/978-3-642-33777-2 39.
-------
Technology Cost, Effectiveness, and Lead Time Assessment
579 King, J., Bocker, O. "Multiple Injection and Boosting Benefits for Improved Fuel Consumption on a Spray
Guided Direct Injection Gasoline Engine." SAE-China and FISITA (eds.), Proceedings of the FISITA 2012 World
Automotive Congress, Lecture Notes in Electrical Engineering 189, DOI: 10.1007/978-3-642-33841-0 18.
580 Brooke, L. "Ricardo begins EBDI V6 road test program." SAE Automotive Engineering, February 2010. Last
accessed on the Internet on 11/11/2016 at the following URL: http:// http://articles.sae.org/7656/.
581 Wada, Y., Nakano, K., Mochizuki, K., and Hata, R., "Development of a New 1.5L 14 Turbocharged Gasoline
Direct Injection Engine," SAE Technical Paper 2016-01-1020, 2016, doi: 10.4271/2016-01-1020.
582 Nakano, K., Wada, Y., Jono, M., Narihiro, S. "New In-Line 4-Cylinder Gasoline Direct Injection Turbocharged
Downsizing Engine." Honda R&D Technical Review, April 2016, pp 139-146.
583 Budack, R., Kuhn, M., Wurms, R., Heiduk, T., "Optimization of the Combustion Process as Demonstrated on the
New Audi 2.01 TFSI," 24th Aachen Colloquium Automobile and Engine Technology 2015.
584 Wurms, R., Budack, R., Grigo, M., Mendl, G., Heiduk, T., Knirsch, S. "The new Audi 2.01 Engine with
innovative Rightsizing," 36. Internationales Wiener Motorensymposium 2015.
585 Eichler, F., Demmelbauer-Ebner, W., Theobald, J., Stiebels, B., Hoffmeyer, H., Kreft, M. "The NewEA211
TSI® evo from Volkswagen." 37. Internationales Wiener Motorensymposium 2016.
586 "National Academy of Sciences. " 2015. Cost, Effectiveness and Deployment of Fuel Economy Technologies for
Light-Duty Vehicles." ISBN 978-0-309-37388-3 p. 5-52 (Finding 5.3)
http://www.nap.edu/catalog.php?record_id=21744.
587 Moskalik, A., Hula, A., Barba, D., and Kargul, J., "Investigating the Effect of Advanced Automatic
Transmissions on Fuel Consumption Using Vehicle Testing and Modeling," SAE Int. J. Engines 9(3): 1916-1928,
2016, doi: 10.4271/2016-01-1142.
588 EPA ALPHA Samples Transmission Walk, David Oh, Will Ruona, Greg Goleski, Ford Motor CO., 9/2016,
included as Attachment 8 of the AAM comments.
589 National Academies of Science. 2011. Assessment of Fuel Economy Technologies for Light-Duty Vehicles.
National Academies Press, ISBN 978-0-309-15607-3, http://www.nap.edu/catalog.php?record_id=12924, p.25.
590 Moskalik, A., Hula, A., Barba, D., and Kargul, J., "Investigating the Effect of Advanced Automatic
Transmissions on Fuel Consumption Using Vehicle Testing and Modeling," SAE Int. J. Engines 9(3): 1916-1928,
2016, doi: 10.4271/2016-01-1142.
591 Greiner, J., Grumbach, M., Dick, A., and Sasse, C., "Advancement in NVH- and Fuel-Saving Transmission and
Driveline Technologies," SAE Technical Paper 2015-01-1087, 2015, doi:10.4271/2015-01-1087.
592 Greiner, J., Grumbach, M., Dick, A., and Sasse, C., "Advancement in NVH- and Fuel-Saving Transmission and
Driveline Technologies," SAE Technical Paper 2015-01-1087, 2015, doi: 10.4271/2015-01-1087.
594 Wayland, M., "Hyundai: Ioniq Hybrid achieves 58 mpg," The Detroit News, November 10, 2016. Retrieved from
http://www.detroitnews.eom/story/business/autos/foreign/2016/ll/10/hyundai-ioniq-hybrid-achieves-
mpg/93618350/
595 Bunkley, N., "Ford plans EV to compete with Chevy Bolt, Tesla Model 3, Fields confirms," Automotive News,
April 28, 2016. Retrieved on November 2, 2016 from
http://www.autonews.corn/article/20160428/OEM05/160429821/fields-confirms-ford-planning-ev-to-compete-with-
chevy-bolt-tesla.
596 Nikkei Asian Review, "Toyota to mass produce electric vehicles," November 7, 2016. Retrieved on November 7,
2016 from http://asia.nikkei.com/Japan-Update/Toyota-to-mass-produce-electric-vehicles.
597 Tajitsu, N. et al., "Toyota, in about-face, may mass-produce long-range electric cars:
Nikkei," November 7, 2016. Retrieved on November 7, 2016 from http://www.reuters.com/article/us-toyota-electric-
cars-idU SKBN13204D.
598 Hall, L., "2017 Renault Zoe Getting New 200-Mile Battery - Nissan Leaf to Follow?," HybridCars.com,
September 28, 2016. Retrieved from http://www.hybridcars.eom/2017-renault-zoe-getting-new-200-mile-battery-
nissan-leaf-to-follow/.
599 California Air Resources Board, "Request for Proposal 15CAR018, Advanced Strong Hybrid and Plug-In Hybrid
Engineering Evaluation and Cost Analysis," May 9, 2016.
600 Nelson, P.A. Gallagher, K.G., Bloom, I., and Dees, D.W., "Modeling the Performance and Cost of Lithium-Ion
Batteries for Electric Drive Vehicles," Second Edition, Argonne National Laboratory, ANL-12/55 (December 2012).
601 See EPA Docket EPA-HQ-OAR-2015-0827, Docket item titled "Instructions for Accessing EPA Battery
Analysis (Proposed Determination)."
602 International Energy Agency (IEA), Global EV Outlook 2016 (Flyer). Retrieved from
http ://www. iea. org/media/topics/transport/GlobalE VOutlook20 l6FLYER.pdf.
-------
Technology Cost, Effectiveness, and Lead Time Assessment
603 International Energy Agency (IEA), Global EV Outlook 2016: Beyond one million electric cars. Retrieved from
http://www. iea. org/publications/freepublications/publication/global-ev-outlook-2016. html.
604 Plumer, B., "The rapid growth of electric cars worldwide, in 4 charts," Vox, June 6, 2016. Retrieved on
November 4, 2016 from http://www.vox.eom/2016/6/6/11867894/electric-cars-global-sales.
605 See 40 CFR600.116-12(a)(6), 40 CFR 600.210-12(d)(3), and 76 FR 39478.
606 See http://www.uscar.Org/guest/partnership/l/us-drive.
607 US DRIVE Partnership, "US DRIVE Electrical and Electronics Technical Team Roadmap," June 2013.
downloaded on Dec. 23, 2015 from
http://wwwl.eere.energy.gov/vehiclesandfuels/pdfs/program/eett_roadmapjune2013.pdf.
608 Slenzak, J., "Next Generation Electrification Products: Focus on Integration and Cost Reduction," Bosch, The
Battery Show 2015, Novi MI, September 15, 2015.
609 cf. U.S. DRIVE Electrical and Electronics Technical Team Roadmap, p 10.
610 Gardner, G., "The self-driving revolution will be mostly electric," Detroit Free Press, September 21, 2106.
Retrieved from http://www.freep.eom/story/money/cars/2016/09/18/self-driving-revolution-mostly-
electric/90410520/.
611 Anandan, V., "Current Status of Solid State Batteries for Automotive Applications," The Battery Show 2016,
Novi, MI, September 14, 2016.
612 Farid, A., "Battery adoption at tipping point - identifying investment opportunities," Berenberg, The Battery
Show Conference and Electric & Hybrid Vehicle Technology Conference, September 13, 2016.
613 Safoutin, M., "EPA Battery Sizing and Cost Analysis for Future Plug-in Vehicles for the Midterm Evaluation of
the 2022-2025 Light-Duty GHG Standards," The Battery Show Conference and Electric & Hybrid Vehicle
Technology Conference, September 15, 2016.
614 General Motors Press Release, "Drive Unit and Battery at the Heart of Chevrolet Bolt EV," January 11, 2016,
downloaded on Jan. 21, 2016 from
http://media.chevrolet.com/media/us/en/chevrolet/home.detail.html/content/Pages/news/us/en/2016/Jan/naias/chevy/
0111-bolt-du.html.
615 Anderman, M., "Battery Packs of Modern xEVs: A Comprehensive Engineering Assessment: Extract," Total
Battery Consulting, 2016, slides 19 and 30.
616 See EPA Docket EPA-HQ-OAR-2015-0827, Microsoft Excel attachment to Docket Item titled "Data and Charts
for Selected Figures in Electrification Chapters of Proposed Determination TSD."
617 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).
618 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 D.J. Santini, Argonne National Laboratory, August 24, 2010.
619 See Docket item EPA-HQ-OAR-2010-0799-11913.
620 ICF International, Peer Review Report of the Draft Report "Modeling the Cost and Performance of Lithium-Ion
Batteries for Electric-Drive Vehicles," March 31, 2011. See Docket itemEPA-HQ-OAR-2010-0799-1080.
621 See Docket items EPA-HQ-OAR-2010-0799-1078 and EPA-HQ-OAR-2010-0799- 11914.
622 Argonne National Laboratory, "Changes to BatPaC for Version 3.0," available at
http://www.cse.anl.gov/batpac/files/Changes%20to%20BatPaC%20for%20Version%203%2021Dec2015.pdf.
623 Argonne National Laboratory, "BatPaC (Battery Performance and Cost)," BatPaC Version 3.0 Beta
17Dec2015.xlsx. A copy is available in Docket EPA-HQ-OAR-2015-0827.
624 Anderman, M., "Battery Packs of Modern xEVs: A Comprehensive Engineering Assessment: Extract," Total
Battery Consulting, 2016, slide 52.
625 "AWD Component Analysis," Pilot Systems for Transport Canada, 2016.
2-452
-------
Economic and Other Key Inputs Used in EPA's Analyses
Table of Contents
Chapter 3: Economic and Other Key Inputs Used in EPA's Analyses 3-1
3.1 The On-Road Fuel Economy "Gap" 3-1
3.1.1 The "Gap" Between Compliance and Real World Fuel Economy 3-1
3.1.2 Real World Fuel Economy and CO2 Projections 3-2
3.2 Fuel Prices and the Value of Fuel Savings 3-4
3.3 Vehicle Mileage Accumulation and Survival Rates 3-5
3.4 Fuel Economy Rebound Effect 3-8
3.4.1 Accounting for the Fuel Economy Rebound Effect 3-8
3.4.2 Summary of Historical Literature on the LDV Rebound Effect 3-10
3.4.3 Review of Recent Literature on LDV Rebound since the 2012 Final Rule 3-14
3.4.4 Basis for Rebound Effect Used in this Proposed Determination 3-19
3.5 Energy Security Impacts 3-21
3.5.1 Implications of Reduced Petroleum Use on U.S. Imports 3-21
3.5.2 Energy Security Implications 3-24
3.5.2.1 Effect of Oil Use on the Long-Run Oil Price 3-25
3.5.2.2 Macroeconomic Disruption Adjustment Costs 3-28
3.5.2.3 Cost of Existing U.S. Energy Security Policies 3-33
3.5.2.4 Military Security Cost Components of Energy Security 3-34
3.6 Non-GHG Health and Environmental Impacts 3-36
3.6.1 Economic Value of Reductions in Particulate Matter 3-37
3.7 Social Cost of Greenhouse Gas Emissions 3-41
3.8 Benefits from Reduced Refueling Time 3-49
3.9 Benefits and Costs from Additional Driving 3-51
3.9.1 Travel Benefit 3-51
3.9.2 Costs Associated with Crashes, Congestion and Noise 3-51
3.10 Discounting Future Benefits and Costs 3-52
3.11 Additional Costs of Vehicle Ownership 3-53
3.11.1 Sales Taxes 3-53
3.11.2 Insurance Costs 3-53
3.11.3 Financing Costs 3-54
Table of Figures
Figure 3.1 Comparing AEO 2016 Retail Fuel Price Projections to AEO2015 Projections 3-5
Figure 3.2 U.S. Expenditures on Crude Oil from 1970 through 2016 3-22
Figure 3.3 Projected and Historical U.S. Expenditures, and Expenditure Share, on Crude Oil 3-30
Figure 3.4 Path from GHG Emissions to Monetized Damages (Source: Marten et al., 2014) 3-49
Table of Tables
Table 3.1 EPA Projections for Fleet-wide CO2 Standards Compliance and On-road Performance for Cars 3-3
Table 3.2 EPA Projections for Fleet-wide CO2 Standards Compliance and On-road Performance for Trucks 3-3
Table 3.3 EPA Projections for Fleet-wide CO2 Standards Compliance and On-road Performance for the Fleet 3-3
Table 3.4 Gasoline Prices for Selected Years in Various AEO 2016 Cases (2015$) 3-4
Table 3.5 Vehicle Survival Rates (from MOVES 2014a) 3-7
Table 3.6 2015 Mileage Schedule (fromMOVES 2014 ) 3-8
Table 3.7 Estimates of the Rebound Effect Using U.S. Aggregate Time-Series Data on Vehicle Travel 3-10
-------
Economic and Other Key Inputs Used in EPA's Analyses
Table 3.8 Estimates of the Rebound Effect Using U.S. State-Level Data 3-10
Table 3.9 Estimates of the Rebound Effect Using U.S. Household Survey Data 3-11
Table 3.10 Projected Trends in U.S. Oil Exports/Imports, and U.S. Oil Import Reductions Resulting from the
Program in Selected Years from 2022 to 2050, (Millions of barrels per day (MMBD) 3-24
Table 3.11 Energy Security Premiums in Selected Years from 2022 to 2050, (2015 $/Barrel)* 3-25
Table 3.12 PM-Related Benefits-per-ton Values (thousands, 2012$)a 3-37
Table 3.13 Human Health and Welfare Effects of PM2 5 3-38
Table 3.14 Social Cost of CO2, 2022-2050 (in 2015$ per metric ton)* 3-45
Table 3.15 Social Cost of CH4 and Social Cost of N20, 2012-2050 (in 2015$ per metric ton) 3-47
Table 3.16 Metrics Used in Calculating the Value of Refueling Time 3-50
Table 3.17 Metrics Used to Calculate the Costs Associated with Congestion, Crashes and Noise Linked to Rebound
Miles Traveled (2015$) 3-52
-------
Economic and Other Key Inputs Used in EPA's Analyses
Chapter 3: Economic and Other Key Inputs Used in EPA's Analyses
3.1 The On-Road Fuel Economy "Gap"
3.1.1 The "Gap" Between Compliance and Real World Fuel Economy
Real world tailpipe CO2 emissions are higher, and real world fuel economy levels are lower,
than the corresponding values from EPA standards compliance tests. This is because laboratory
testing cannot reflect all of the factors that can affect real world operation, and, in particular, the
city and highway tests used for compliance do not encompass the broad range of driver behavior
and climatic conditions experienced by typical U.S. drivers.A In the rulemakings that established
the National Program standards through MY2025, EPA and NHTSA applied a 20 percent fleet-
wide fuel economy "gap," i.e., that average, fleet-wide real world fuel economy would be 20
percent lower than EPA compliance test values.6 This 20 percent value was based on data from
MY2004-2006.1 For example, a vehicle with a fuel economy compliance test value of 30 mpg
would be projected to have a real world fuel economy of 30 multiplied by 0.8 (equivalent to a 20
percent reduction) or 24 mpg. The inverse of 0.8 is 1.25, and a vehicle with a CO2 emissions
compliance test value of 300 grams/mile would be projected to have a real world CO2 emissions
value of 300 multiplied by 1.25 or 375 grams/mile.
As discussed in the Draft TAR, more recent data suggest that the gap between the 2-cycle
compliance tests and the 5-cycle methodology values may have increased very slightly in the last
decade. For example, the use of final MY2014 and final MY2015 data suggest that the fuel
economy gap between 2-cycle data and 5-cycle data may now be approximately 21 percent.2
EPA believes that further analysis is needed before incorporating such small changes into
calculations of the overall gap. In addition, some analysis suggests that the gap between 2-cycle
compliance tests and real world fuel economy may be increasing in recent years, but the
evidence is not conclusive.3 One factor which has clearly changed and can be quantified is
ethanol content in gasoline. When the 20 percent fuel economy gap was first projected in 2005-
2006, ethanol accounted for a small fraction of the gasoline pool. Consistent with our analysis in
the Draft TAR, for the Proposed Determination, EPA adjusts for projected differences in the
energy content due to increased ethanol penetration of retail gasoline relative to test fuel for
MY2022 and beyond. Ethanol contains about 35 percent less energy than gasoline, on a
volumetric basis, and EPA projects that average in-use gasoline will contain about 3.5 percent
less energy in 2025 than it did in the 2005-2006 timeframe. Using the "base" 20 percent fuel
economy gap between 2-cycle and 5-cycle data and the projected impact of the ethanol increase
in 2025 yields an effective gap of 23 percent (or a fuel economy factor of 0.77), and this is the
AEPA has recognized that the "2-cycle" city and highway tests are not representative of real world fuel economy
performance for over 30 years. From MY1985 through MY2007, EPA based new vehicle window labels on the
fuel economy compliance test values adjusted downward by 10% for the city test and by 22% for the highway
test. Beginning in MY2008, EPA has based vehicle labels on a 5-cycle methodology that includes three additional
tests (reflecting high speed/high acceleration, hot temperature/air conditioning, and cold temperature operation) as
well as a 9.5% downward fuel economy adjustment for other factors not reflected in the 5-cycle protocol.
B Note that this is an average fleet-wide value, in reality the true fuel economy gap is data driven and will be lower
for some vehicles and higher for other vehicles. In general, all things being equal, today's data suggests that the
gap is generally smaller for lower-fuel economy vehicles and greater for higher-fuel economy vehicles.
3-1
-------
Economic and Other Key Inputs Used in EPA's Analyses
overall fuel economy gap that we use in this Proposed Determination analysis, which is
consistent with that used in the Draft TAR. Multiplying 2-cycle fuel economy by 0.77 yields
projected real world fuel economy.0
The fuel economy gap is data driven, so any 2025 projection involves uncertainty. EPA
expects that, all other things being equal, as average fuel economy increases over time, the gap
would likely increase as well. On the other hand, it is also possible that powertrain designs will
be designed to be more robust in the future, which would impact the gap in the opposite
direction.
3.1.2 Real World Fuel Economy and CO2 Projections
Except when noted, CO2 emissions and fuel economy values cited in this analysis represent
standards compliance values. As discussed above, real world tailpipe CO2 emissions are higher,
and real world fuel economy levels are lower, than the corresponding values from the EPA
standards compliance tests.
This has led to widespread public confusion as there are two sets of fuel economy "books,"
one for fuel economy standards compliance (mandated by statute for cars) and one for the
vehicle label estimates that EPA provides to consumers to estimate real world fuel economy.
The projected real world fuel economy values shown below are the most meaningful fuel
economy values for citizens and reporters as they provide a good comparison with label values,
EPA Fuel Economy Trends report values, vehicle dashboard display values, and fuel economy
calculations performed by some drivers, and also correspond to real world fuel consumption and
CO2 emissions.
Table 3.1 through Table 3.3 show EPA's best projections of the real world CO2 emissions and
fuel economy values associated with the projected CO2 standards compliance emissions levels
presented throughout this report, as well as how "the numbers add up," for cars, trucks, and the
combined car/truck fleet, respectively. These values use as a starting point the projected
industry-wide CO2 2-cycle targets. The first step is to "back out" the impact of the direct air
conditioner refrigerant credits, since reducing leakage and/or substituting lower-GHG
refrigerants will not increase real world fuel economy. Backing out these credits requires adding
the value of the air conditioner refrigerant credits to the target values, as doing so increases the
CO2 value and decreases the projected real world fuel economy level. The sum of the 2-cycle
target and the "backed out" air conditioner refrigerant credits is the "fuel economy-relevant
adjusted 2-cycle CO2 emissions value," shown as the effective CO2 value in the tables which can
also be expressed as an effective mpg by dividing it into 8887 (which represents the number of
grams of CO2 that results from the combustion of a gallon of test gasoline). The second step is
to multiply the adjusted 2-cycle, or effective mpg value by 0.77, the fuel economy "gap" factor
discussed above. This step converts from the adjusted 2-cycle mpg to a real world, on-road mpg
value. On-road tailpipe CO2 emissions are projected by dividing the real world mpg value into
8488 (which represents the number of grams of CO2 that results from the combustion of a gallon
c The corresponding C02 "gap" is 1.24, i.e., multiplying 2-cycle tailpipe C02 by 1.24 yields projected real world
C02 emissions. This 1.24 factor is actually less than the 1.25 factor used in the past because of the lower carbon
content of ethanol.
3-2
-------
Economic and Other Key Inputs Used in EPA's Analyses
of retail gasoline). Subtracting back the A/C leakage credit value provides an on-road CO2
equivalent (CO2 e) value as shown.
Table 3.1 EPA Projections for Fleet-wide CO2 Standards Compliance and On-road Performance for Cars
2-Cycle
Adjustments to 2-Cycle to
Reflect Real World Impacts
On-road
MY
CO 2
Target
(g/mi)
C02
Target
As
MPG
A/C
Leakage
Credit
(g/mi)
A/C
Efficiency
Credit
(g/mi)
Off-
cycle
Credit
(g/mi)
Tailpipe
C02
(g/mi)
MPG
A/C
Efficiency
& Off-
cycle
Credits
(g/mi)
Effective
C02
(g/mi)
Effective
MPG
Gap
On-
road
MPG
On-road
Tailpipe
CO 2
(g/mi)
On-
road
CO 2e
(g/mi)
2021
171
51.9
13.8
5.0
0.8
191
46.6
5.8
185
48.1
.773
37.1
229
215
2022
165
53.9
13.8
5.0
1.0
185
48.1
6.0
179
49.8
.773
38.4
221
207
2023
159
56.0
13.8
5.0
1.2
179
49.7
6.2
172
51.5
.773
39.8
213
200
2024
153
58.1
13.8
5.0
1.5
173
51.3
6.5
167
53.3
.773
41.2
206
192
2025
147
60.3
13.8
5.0
1.7
168
53.0
6.7
161
55.2
.773
42.6
199
186
Note: The on-road values reflect adjustments for both the historical 2-cycle-to-5-cycle gap as well as the projected
ethanol content in retail gasoline, and corresponding energy content. The on-road CO2 is calculated by dividing
8488, the estimated CO2 grams/gallon from combustion of a gallon of retail gasoline, by the on-road MPG. The on-
road CO2 e column subtracts from the on-road tailpipe CO2 values the A/C leakage value to yield a value that
reflects overall real world CO2 e emissions performance.
Table 3.2 EPA Projections for Fleet-wide CO2 Standards Compliance and On-road Performance for Trucks
2-Cycle
Adjustments to 2-Cycle to
Reflect Real World Impacts
On-road
MY
C02
Target
(g/mi)
C02
Target
As
MPG
A/C
Leakage
Credit
(g/mi)
A/C
Efficiency
Credit
(g/mi)
Off-
cycle
Credit
(g/mi)
Tailpipe
C02
(g/mi)
MPG
A/C
Efficiency
& Off-
cycle
Credits
(g/mi)
Effective
C02
(g/mi)
Effective
MPG
Gap
On-
road
MPG
On-road
Tailpipe
C02
(g/mi)
On-
road
CO 2e
(g/mi)
2021
238
37.4
17.2
7.2
1.9
264
33.7
9.1
255
34.9
.773
26.9
315
298
2022
228
39.0
17.2
7.2
2.4
255
34.9
9.6
245
36.2
.773
28.0
304
286
2023
219
40.6
17.2
7.2
2.8
246
36.1
10.0
236
37.7
.773
29.1
292
275
2024
210
42.3
17.2
7.2
3.3
238
37.4
10.5
227
39.1
.773
30.2
281
264
2025
202
44.0
17.2
7.2
3.8
230
38.6
11.0
219
40.6
.773
31.3
271
254
Note: The on-road values reflect adjustments for both the historical 2-cycle-to-5-cycle gap as well as the projected
ethanol content in retail gasoline, and corresponding energy content. The on-road CO 2 is calculated by dividing
8488, the estimated CO2 grams/gallon from combustion of a gallon of retail gasoline, by the on-road MPG. The on-
road CO2 e column subtracts from the on-road tailpipe CO2 values the A/C leakage value to yield a value that
reflects overall real world CO2 e emissions performance.
Table 3.3 EPA Projections for Fleet-wide CO2 Standards Compliance and On-road Performance for the
Fleet
2-Cycle
Adjustments to 2-Cycleto
Reflect Real World Impacts
On-road
MY
CO 2
Target
(g/mi)
CO 2
Target
As
MPG
A/C
Leakage
Credit
(g/mi)
A/C
Efficiency
Credit
(g/mi)
Off-
cycle
Credit
(g/mi)
Tailpipe
C02
(g/mi)
MPG
A/C
Efficiency
& Off-
cycle
Credits
(g/mi)
Effective
C02
(g/mi)
Effective
MPG
Gap
On-
road
MPG
On-road
Tailpipe
CO 2
(g/mi)
On-
road
CO 2e
(g/mi)
2021
204
43.6
15.5
6.1
1.3
227
39.2
7.4
219
40.5
.773
31.3
272
256
2022
196
45.4
15.5
6.1
1.7
219
40.6
7.7
211
42.1
.773
32.5
261
246
2023
187
47.4
15.4
6.1
2.0
211
42.1
8.0
203
43.8
.773
33.8
251
236
2024
180
49.4
15.4
6.0
2.3
204
43.6
8.4
195
45.5
.773
35.1
242
226
2025
173
51.4
15.4
6.0
2.7
197
45.1
8.7
188
47.2
.773
36.4
233
218
3-3
-------
Economic and Other Key Inputs Used in EPA's Analyses
Note: The on-road values reflect adjustments for both the historical 2-cycle-to-5-cycle gap as well as the projected
ethanol content in retail gasoline, and corresponding energy content. The on-road CO2 is calculated by dividing
8488, the estimated CO2 grams/gallon from combustion of a gallon of retail gasoline, by the on-road MPG. The on-
road CO2 e column subtracts from the on-road tailpipe CO2 values the A/C leakage value to yield a value that
reflects overall real world CO2 e emissions performance.
EPA projects the industry-wide real world fuel economy associated with the MY2025 GHG
standards to be about 36 mpg. This value provides a good comparison with average label and
Fuel Economy Trends values.
3.2 Fuel Prices and the Value of Fuel Savings
Fuel prices and the projection of fuel prices remain critical in the analysis of GHG and fuel
economy standards. EPA has continued to use the methodology described in Chapter 10 of the
Draft TAR, with some updates to the inputs used for this Proposed Determination. EPA
continues to rely on the fuel price projections from the U.S. Energy Information
Administration's (EIA) Annual Energy Outlook (AEO) for this analysis, updated to the AEO
2016 Reference Case (the Draft TAR analysis was based on AEO 2015). The Reference case
projection is a business-as-usual trend estimate, given known technology and technological and
demographic trends. EIA has published annual projections of energy prices and consumption
levels for the U.S. economy since 1982 in its Annual Energy Outlook reports. These projections
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. In
addition to the AEO 2016 Reference Case as the central case, EPA has also included the AEO
2016 low and high fuel price cases as sensitivities. A comparison of these cases is presented
below in Table 3.4.
Table 3.4 Gasoline Prices for Selected Years in Various AEO 2016 Cases (2015$)
2025
2030
2040
AEO 2016 Reference Case
$ 2.97
$ 3.19
$ 3.81
AEO 2016 "Low" Case
$ 1.97
$ 2.04
$ 2.53
AEO 2016 "High" Case
$ 4.94
$ 5.17
$ 5.61
The retail fuel price forecasts presented in AEO 2016 span the period from 2015 through
2040. Measured in constant 2015 dollars, the AEO 2016 Reference Case projections of retail
gasoline prices during calendar year 2025 is $2.97 per gallon, rising gradually to $3.81 by the
year 2040 (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 for MYs 2022-25
requires fuel price forecasts that extend through nearly 2060, the last year during which a
significant number of MY2025 vehicles will remain in service. Due to the difficulty in
accurately projecting fuel prices over this long time span (as AEO projections span only through
2040), EPA has assumed constant fuel prices after the year 2040 for this Proposed
Determination.
3-4
-------
Economic and Other Key Inputs Used in EPA's Analyses
Figure 3.1 shows the three AEO 2016 fuel price cases used for this Proposed Determination,
as compared to the AEO 2015 cases that had been used in the Draft TAR.
7.00
6.00
CD
r=s 3.00
1.00
AEO 2015 Reference
AEO 2015 High
¦AEO 2016 Low
AEO 2015 Low
¦AEO 2016 Reference
•AEO 2016 High
** 5.00
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 2032 2034 2036 2038 2040
Year
Figure 3.1 Comparing AEO 2016 Retail Fuel Price Projections to AEO2015 Projections
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 $0.41 per gallon during 2015. 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 GHG standards to the
U.S. economy. 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
EPA's analysis, this difference causes each gallon of fuel saved to be valued by about $0.39 (in
constant 2015 dollars) more from the perspective of an individual vehicle buyer than from the
overall perspective of the U.S. economy.
3.3 Vehicle Mileage Accumulation and Survival Rates
EPA's analyses of benefits from GHG standards for passenger cars and light trucks, including
GHG reductions, oil reductions, and fuel savings, 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
3-5
-------
Economic and Other Key Inputs Used in EPA's Analyses
difference in their total lifetime fuel use over the entire lifetimes of these vehicles as compared to
a reference case.
EPA's approach for this analysis remains largely the same as that found in the Draft TAR,
Chapter 10. Since the Draft TAR, EPA has updated a few key inputs related to vehicle lifetime
survival rates and total vehicle miles traveled (VMT), as described in Table 3.5 and Table 3.6
below. These updates were made in order to align this analysis with inputs developed in
conjunction with updates to the EPA MOVES 2014a model4 since the official release of that
model, which now has integrated new activity and population data sources from R.L. Polk, the
U.S. Department of Transportation/Federal Highway Administration (FHWA), and the EIA
Annual Energy Outlook 2016.5 Continuing consistency with EIA, FHWA and MOVES remains
a priority for these modeling inputs. Additionally, the MOVES model is also already used as
part of other EPA rulemaking analyses, allowing this analysis to take advantage of updates from
those efforts. These updates show a slight increase (approximately 1.8 percent) in overall
vehicle VMT, especially in the early years of a vehicle's lifetime. Methodologies for the
derivation of fuel savings and related benefits (including future year projections, VMT growth
factor, and fuel cost per mile) from these inputs remain identical to those used in the Draft TAR
(which are consistent with the 2012 FRM).
3-6
-------
Economic and Other Key Inputs Used in EPA's Analyses
Table 3.5 Vehicle Survival Rates (from MOVES 2014a)
VEHICLE AGE
ESTIMATED SURVIVAL FRACTION (CARS)
ESTIMATED SURVIVAL FRACTION (LIGHTTRUCKS)
0
1.000
1.000
1
0.997
0.991
2
0.994
0.982
3
0.991
0.973
4
0.984
0.960
5
0.974
0.941
6
0.961
0.919
7
0.942
0.891
8
0.920
0.859
9
0.893
0.823
10
0.862
0.784
11
0.826
0.741
12
0.788
0.697
13
0.718
0.651
14
0.613
0.605
15
0.510
0.553
16
0.415
0.502
17
0.332
0.453
18
0.261
0.407
19
0.203
0.364
20
0.157
0.324
21
0.120
0.288
22
0.092
0.255
23
0.070
0.225
24
0.053
0.198
25
0.040
0.174
26
0.030
0.153
27
0.023
0.133
28
0.013
0.117
29
0.010
0.102
30
0.007
0.089
31
0.002
0.027
Note: This table remains consistent with the values found in the Draft TAR.
3-7
-------
Economic and Other Key Inputs Used in EPA's Analyses
Table 3.6 2015 Mileage Schedule (from MOVES 2014a)
VEHICLE AGE
ESTIMATED VMT CARS
ESTIMATED VMT LIGHT TRUCKS
0
14,102
16,040
1
13,834
15,745
2
13,545
15,408
3
13,236
15,081
4
12,910
14,676
5
12,568
14,163
6
12,213
13,723
7
11,848
13,253
8
11,473
12,778
9
11,092
12,272
10
10,706
11,781
11
10,319
11,290
12
9,931
10,808
13
9,546
10,326
14
9,165
9,854
15
8,791
9,396
16
8,425
8,962
17
8,070
8,543
18
7,728
8,159
19
7,401
7,810
20
7,092
7,496
21
6,804
7,222
22
6,536
6,991
23
6,292
6,809
24
6,075
6,679
25
5,886
6,602
26
5,728
6,588
27
5,602
6,588
28
5,512
6,588
29
5,458
6,588
30
5,458
6,588
TOTAL
283,347
314,805
3.4 Fuel Economy Rebound Effect
3.4.1 Accounting for the Fuel Economy Rebound Effect
The rebound effect generally refers to the additional energy consumption that may arise from
the introduction of a more efficient, lower cost energy service which offsets, to some degree, the
energy savings benefits of that efficiency improvement.6'7'8 In the context of light-duty vehicles
3-8
-------
Economic and Other Key Inputs Used in EPA's Analyses
(LDVs), rebound effects might occur when an increase in vehicle fuel efficiency encourages
people to drive more as a result of the lower cost per mile of driving. Because this additional
driving consumes fuel and generates emissions, the magnitude of the rebound effect is one
determinant of the actual fuel savings and emission reductions that will result from adopting
stricter fuel economy or GHG emissions standards.
The rebound effect for personal vehicles can in theory be estimated directly from the change
in vehicle use, in terms of vehicle miles traveled (VMT), which results from a change in vehicle
fuel efficiency.0 In practice, any attempt to quantify this "VMT rebound effect" (sometimes also
labeled the "direct rebound effect," or "direct VMT rebound effect") is complicated by the
difficulty in identifying an applicable data source from which the response to a significant
improvement in fuel efficiency can be estimated. Analysts instead often estimate the VMT
rebound indirectly as the change in vehicle use that results from a change in fuel cost per mile
driven or a change in fuel price. When a fuel cost-per mile approach is used, it does not
distinguish the relative contributions of changes in fuel efficiency and changes in fuel price to
the rebound effect, since both factors are determinants of fuel cost-per mile.E
When expressed as positive percentages, the elasticities of vehicle use with respect to fuel
efficiency or per-mile fuel costs (or fuel prices) give the percentage increase in vehicle use that
results from a doubling of fuel efficiency (e.g., 100 percent increase), or a halving of fuel
consumption or fuel price. For example, a 10 percent rebound effect means that a 20 percent
reduction in fuel consumption or fuel price (and the corresponding reduction in fuel cost per
mile) is expected to result in a two percent increase in vehicle use.
While we focus on the VMT rebound effect in our analysis of this program, there are at least
two other types of rebound effects discussed in the transportation policy and economics
literature. In addition to the direct VMT rebound effect, there is the "indirect" rebound effect,
which typically refers to the purchase of other goods or services that consume energy with the
costs savings from energy efficiency improvements. The last type of rebound effect is labeled
the "economy-wide" rebound effect. This effect refers to the increased demand for energy
throughout the whole economy in response to the reduced market price of energy that happens as
a result of energy efficiency improvements.
Research on indirect and economy-wide rebound effects is scant. Given the limited literature
and potential methodological shortcoming of the studies on LDV indirect and economy-wide
rebound effects, the rebound effect discussed in this section refers solely to the effect of
increased fuel efficiency on vehicle use. The terms "VMT rebound effect," "direct VMT
rebound effect," and "rebound effect" can be used interchangeably, and they need to be
distinguished from other rebound effects that could potentially impact the fuel savings and
emissions reductions from EPA's LDV standards such as the "indirect rebound effect." To
restate, the rebound effect discussed in this section refers solely to the effect of increased fuel
efficiency on vehicle use.
D 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.
E 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.
3-9
-------
Economic and Other Key Inputs Used in EPA's Analyses
3.4.2 Summary of Historical Literature on the LDV Rebound Effect
This section provides a brief summary of historical literature on the LDV 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 the MY2022-2025 standards
will affect future driving behavior. Recent studies on LDV rebound effects that have become
available since the 2012 LDV final rule and also reviewed for the Draft TAR are summarized in
Section 3.4.3 below. The one additional study on the direct rebound effect, added to this review
since the Draft TAR, is by Wang and Chen (2014).
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 two tables, Tables 3.7 and 3.8. The agency added in more recent
studies by Small and Van Dender (2007a) and Hymel, Small and Van Dender (2010) into Table
3.8. In addition, Table 3.9 below provides estimates of the rebound effect using U.S. household
survey data. The agency added in more recent studies by Bento (2009) and Wadud et al. (2009)
into Table 3.9.
Table 3.7 Estimates of the Rebound Effect Using U.S. Aggregate Time-Series Data on Vehicle Travel
Author (year)
Short-Run
Long-Run
Time Period
Mayo & Mathis (1988)
22%
26%
1958-84
Gately (1992)
9%
9%
1966-88
Greene (1992)
Linear 5-19%
Linear 5-19%
1957-89
Log-linear 13%
Log-linear 13%
Jones(1992)
13%
30%
1957-89
Schimek (1996)
5-7%
21-29%
1950-94
Source: Sorrell and Dimitropolous (2007) table 4.6.9
Table 3.8 Estimates of the Rebound Effect Using U.S. State-Level Data
Author (year)
Short-Run
Long-Run
Time Period
Haughton & Sarkar (1996)
9-16%
22%
1973-1992
Small and Van Dender
4.5%
22.2%
1966-2001
(2005 and 2007a)
2.2%
10.7%
1997-2001
Hymel, Small and Van
4.7%
24.1%
1966-2004
Dender (2010)
4.8%
15.9%
1984-2004
Source: Sorrell and Dimitropolous (2007) table 4.7 and Small and Van Dender (2007a) and (2010)
3-10
-------
Economic and Other Key Inputs Used in EPA's Analyses
Table 3.9 Estimates of the Rebound Effect Using U.S. Household Survey Data
Author
Estimate of Rebound Effect
Time Period
(year)
Goldberg (1996)
0%
CES 1984-90
Greene, Kahn, and
23%
EIA RTECS
Gibson (1999a)
1979-1994
Pickrell & Schimek
4-34%
NPTS 1995
(1999)
Single year
Puller & Greening
49%
CES 1980-90
(1999)
Single year, cross-sectional
West (2004)
87%
CES 1997
Single year
Bento (2009)
34%
NHTS
2001
Wadud et al. (2009)
1-25%
CES 1984-2003
Source: Sorrell and Dimitropolous (2007) and Bento (2009) and Wadud et al. (2009)
While studies using national (Table 3.7) and state level (Table 3.8) data have found a
relatively consistent range of long-run estimates of the rebound effect, household surveys display
more variability (Table 3.9). One explanation for the variability in the household survey
estimates 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 fuel cost-per mile on vehicle use from
that of other, unobserved factors. For example, commuting distance might influence both the
choice of the vehicle as well as VMT. Residential density may also influence both fuel cost-per
mile and VMT, since households in urban areas are likely to simultaneously face both higher fuel
prices and shorter travel distances. Also, given that household data tends to be collected on an
annual basis, there may not be enough variability in the fuel price data to estimate the magnitude
of the rebound effect.10
It is important to note that some of these studies actually quantify the price elasticity of
gasoline demand (e.g., Puller & Greening (1999)11) or the elasticity of VMT with respect to the
price of gasoline (e.g., Pickrell & Schimek (1999)12), rather than the elasticity of VMT with
respect to fuel efficiency or the fuel cost per mile of driving. These latter measures more closely
match the definition of the fuel economy rebound effect. In fact, most studies cited above do not
estimate the direct measure of the fuel economy rebound effect (i.e., the increase in VMT
attributable to an increase in fuel efficiency).
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 different rebound effects for
3-11
-------
Economic and Other Key Inputs Used in EPA's Analyses
households owning varying numbers of vehicles, with most finding that the rebound effect is
larger among households that own more vehicles.17
Some of the more recent studies (Small and Van Dender (2007), Hymel, Small, and Van
Dender ((2010), (2012)), using both state-level and national data, conclude that the rebound
effect varies directly in response to changes in personal income, as well as fuel costs. These
more recent studies published between 2007 and 2012 indicate that the rebound effect has
decreased over time as incomes have risen and, until recently, fuel costs as a share of total
monetary travel costs have generally decreased.0 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 (2007)13 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. 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 five year
period (1997-2001) estimated in their study, 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 six
percent.14
Hymel, Small and Van Dender (2010)15 extended the Small and Van Dender model by adding
congestion as an endogenous variable. Although controlling for congestion 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 (2012)16 under contract with EPA further appears to
support the theory that the magnitude of the rebound effect "is by now on the order of 10
F Five of the household survey studies evaluated in Table 3.9 found that the rebound effect varies in relation to the
number of household vehicles. Of those five studies, three found that the rebound effect rises with higher vehicle
ownership, and two found that it declines. The three studies with rebound estimates that increase with higher
household vehicle ownership are: Greene, D., and Hu, P., "The Influence of the Price of Gasoline on Vehicle Use
in Multi-vehicle Households," Transportation Research Record (1984), pp. 19-24; Hensher, D., Milthorpe, F. and
Smith, N., "The Demand for Vehicle Use in the Urban Household Sector: Theory and Empirical Evidence,"
Journal of Transport Economics and Policy, 24:2 (1990), pp. 119-137; and Walls, M., Krupnick A., and Hood, H.,
"Estimating the Demand for Vehicle-Miles Traveled Using Household Survey Data: Results from the 1990
Nationwide Personal Transportation Survey," Discussion Paper ENR 93-25, Energy and Natural Resources
Division, Resources for the Future, Washington, D.C., 1993.
G 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 (2012)). With the recent decline in world petroleum prices, total vehicle
operating costs have declined recently as well.
3-12
-------
Economic and Other Key Inputs Used in EPA's Analyses
percent."H Like Small and Van Dender, Greene finds that the VMT rebound effect could decline
modestly over time as household income rises and travel costs increase. 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.
From this perspective, if the impact of fuel efficiency on VMT is not statistically significant, the
VMT rebound effect could be zero. 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 rejected the hypothesis of equal elasticities for gasoline prices
and fuel efficiency. 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 2008, and
drops to 10 percent in 2020 and to nine percent in 2030.
Of the studies listed in Table 3.9, the studies that are most recent are by Bento et al.17 and
Wadud et al.18 Bento et al. combined demographic characteristics of more than 20,000 U.S.
households, the manufacturer and model of each vehicle they owned, and their annual usage of
each vehicle from the 2001 National Household Travel Survey with detailed data on fuel
economy and other attributes for each vehicle model obtained from commercial publications.
The authors aggregated vehicle models into 350 categories representing combinations of
manufacturer, vehicle type, and age, and use the resulting data to estimate the parameters of a
complex model of households' joint choices of the number and types of vehicles to own, and
their annual use of each vehicle.
Bento et al. estimate the effect of vehicles' operating cost-per mile, including fuel costs -
which depend in part on each vehicle's fuel economy - as well as maintenance and insurance
expenses, on households' annual use of each vehicle they own. Combining the authors'
estimates of the elasticity of vehicle use with respect to per mile operating costs with the reported
fraction of total operating costs accounted for by fuel (slightly less than one-half) yields
estimates of the rebound effect. The resulting values vary by household composition, vehicle
size and type, and vehicle age, ranging from 21 to 38 percent, with a composite estimate of 34
percent for all households, vehicle models, and ages. The smallest values apply to new luxury
cars, while the largest estimates are for light trucks and households with children, but the implied
rebound effects differ little by vehicle age.
Wadud et al. combine data on U.S. households' demographic characteristics and expenditures
on gasoline over the period 1984-2003 from the Consumer Expenditure Survey with data on
gasoline prices and an estimate of the average fuel economy of vehicles owned by individual
households (constructed from a variety of sources). They employ these data to explore variation
in the sensitivity of individual households' gasoline consumption to differences in income,
H p. 15, Greene, D., Rebound 2007: Analysis of U.S. light-duty vehicle travel statistics. Energy Policy (2010),
doi:10.1016/j.enpol.2010.03.083.
3-13
-------
Economic and Other Key Inputs Used in EPA's Analyses
gasoline prices, the number of vehicles owned by each household, and their average fuel
economy. Using an estimation procedure intended to account for correlation among unmeasured
characteristics of households and among estimation errors for successive years, the authors
explore variation in the response of fuel consumption to fuel economy and other variables among
households in different income categories, and between those residing in urban and rural areas.
Dividing U.S. households into five equally-sized income categories, Wadud et al. estimate
rebound effects ranging from 1-25 percent, with the smallest estimates (8 percent and 1 percent)
for the two lowest income categories, and significantly larger estimates for the middle (18
percent) and two highest income groups (18 and 25 percent). In a separate analysis, the authors
estimate rebound effects of seven percent for households of all income levels residing in U.S.
urban areas, and 21 percent for rural households.
Since there has been little variation in fuel economy in the data over time, isolating the impact
of fuel economy 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
economy on VMT, as discussed above. However, these studies may overstate the potential
impact of the rebound effect resulting from this rule, if people are more responsive to changes in
fuel price than the variable directly of interest, fuel economy.
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 and Gately (1997) and
Sentenac-Chemin (2012)19 have provide some evidence that demand for transportation fuel is
asymmetric. In other words, given the same size change in prices, the response to a decrease in
gasoline price is smaller than the response to an increase in gasoline price. Gately (1993)20 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. Furthermore, Dargay and Gately and Sentenac-Chemin also
find evidence that consumers respond more to a large shock than a small, gradual change in fuel
prices. Since these 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. Greene also notes that the resultant data from such gradual changes
could make discernment of such an effect difficult.
3.4.3 Review of Recent Literature on LDV Rebound since the 2012 Final Rule
Recent studies on LDV rebound effects that have become available since the 2012 LDV final
rule and are consistent with those discussed in the Draft TAR are summarized in Section 3.4.3
below. The one additional study on the direct rebound effect reviewed since the Draft TAR is by
Wang and Chen (2014). Only a limited amount of work has been conducted to examine the
rebound effect of electric vehicles so most of the studies of light-duty vehicle rebound effects
focus on a change in gasoline prices. Below is a brief summary of the results of these recent
studies.
Using data on household characteristics and vehicle use from the 2009 Nationwide Household
Transportation Survey (NHTS), Su (2012)21 analyzes the effects of locational and demographic
factors on household vehicle use, and investigates how the magnitude of the rebound effect
varies with vehicles' annual use. Using variation in the fuel economy and per-mile cost of and
detailed controls for the demographic, economic, and locational characteristics of the households
3-14
-------
Economic and Other Key Inputs Used in EPA's Analyses
that owned them (e.g., road and population density) and each vehicle's main driver (as identified
by survey respondents), the author employs specialized regression methods to capture the
variation in the rebound effect across ten different categories of vehicle use.
Su estimated that the overall rebound effect for all vehicles in the sample averaged 13 percent,
and that its magnitude varied from 11-19 percent among the ten different categories of annual
vehicle use. The smallest rebound effects were estimated for vehicles at the two extremes of the
distribution of annual use - those driven comparatively little, and those used most intensively -
while the largest estimated effects applied to vehicles that were driven slightly more than
average. Controlling for the possibility that high-mileage drivers respond to the increased
importance of fuel costs by choosing vehicles that offer higher fuel economy narrowed the range
of Su's estimated rebound effects slightly (to 11-17 percent), but did not alter the finding that
they are smallest for lightly- and heavily-driven vehicles and largest for those with slightly above
average use.
Linn (2013)22 also uses the 2009 NHTS to develop a linear regression approach to estimate
the relationship between the VMT of vehicles belonging to each household and a variety of
different factors: fuel costs, vehicle characteristics other than fuel economy (e.g., horsepower,
the overall "quality" of the vehicle), and household characteristics (e.g., age, income). Linn
reports a fuel economy rebound effect with respect to VMT of between 20-40 percent.
One interesting result of the study is that when the fuel efficiency of all vehicles increases,
which would be the long-run effect of rising fuel efficiency standards, two factors have opposing
effects on the VMT of a particular vehicle. First, VMT increases when that vehicle's fuel
efficiency increases. But the increase in the fuel efficiency of the household's other vehicles
causes the vehicle's own VMT to decrease. Since the effect of a vehicle's own fuel efficiency is
larger than the other vehicles' fuel efficiency, VMT increases if the fuel efficiency of all vehicles
increases proportionately. Linn also finds that VMT responds much more strongly to vehicle
fuel economy than to gasoline prices, which is at variance with the Hymel et al. and Greene
results discussed above.
Like Su and Linn, Liu et al. (2014)23 also employed the 2009 NHTS to develop an elaborate
model of an individual household's choices about how many vehicles to own, what types and
ages of vehicles to purchase, and how much combined driving to do using all of them. Their
analysis used a complex mathematical formulation and statistical methods to represent and
measure the interdependence among households' choices of the number, types, and ages of
vehicles to purchase, as well as how intensively to use them.
Liu et al. employed their model to simulate variation in households' total vehicle use to
changes in their income levels, neighborhood characteristics, and the per-mile fuel cost of
driving averaged over all vehicles each household owns. The complexity of the relationships
among the number of vehicles owned, their specific types and ages, fuel economy levels, and use
incorporated in their model required them to measure these effects by introducing variation in
income, neighborhood attributes, and fuel costs, and observing the response of households'
annual driving. Their results imply a rebound effect of approximately 40 percent in response to
significant (25-50 percent) variation in fuel costs, with almost exactly symmetrical responses to
increases and declines.
3-15
-------
Economic and Other Key Inputs Used in EPA's Analyses
Frondel and Vance (2013)24 use panel estimation methods and household diary travel data
collected in Germany between 1997 and 2009 to identify an estimate of a private transport
rebound value. The study focuses on single-car households that did not change their car
ownership over the maximum three years each household was surveyed. Failing to reject the
null hypothesis of a symmetric price response, they find a rebound effect for single-vehicle
households of 46-70 percent (though we discuss further below the limitations in applying
findings of studies from other countries to U.S. rebound).
Gillingham (2014)25 analyzed variation in the use of more than five million new vehicles
purchased in California during the years 2001-03 over the first several years of their lifetimes,
focusing particularly on the response of buyers' use of new vehicles to geographic and temporal
variation in fuel prices. His sample consists predominantly of personal vehicles (87 percent), but
also includes some purchased by businesses, rental car companies, and government. He
estimates the effect of differences in the average of monthly fuel prices on their monthly average
vehicle use over the time - at a county level, since being purchase - focusing his analysis on
vehicles that have been purchased new and have been in service for six to seven years. The
author also explores how the effect of fuel prices on vehicle use varies with vehicle use, buyer
type and household income.
Gillingham relies exclusively on the effect of variation in fuel prices and does not involve
vehicles' fuel economy. He reports an overall average effect of fuel prices on vehicle use that
corresponds to a rebound effect of 22 percent, rising to 23 percent when he controls for the
potential effect of gasoline demand on its retail price. He finds little evidence of variation in the
rebound effect among buyer types. Based on the nature of his data and estimation procedure, he
interprets his estimates as implying that vehicle use responds fully to changes in fuel prices after
approximately two years.
Gillingham's results suggest that the vehicle-level responsiveness to fuel price increases with
income. Gillingham hypothesizes that the increase in the per-vehicle rebound effect with higher
incomes may relate to wealthier households having more discretionary driving or switching
between flying and driving. Alternatively, wealthier households tend to own more vehicles and
it is possible that within-household switching of vehicles to other more efficient vehicles in the
household may account for the greater responsiveness at higher income levels.
In contrast to Gillingham's results, Wang and Chen (2014)26 examine the variation of fuel
price elasticity of VMT across income groups using a system of structural equations with VMT
and fuel efficiency (i.e., miles per gallon) as endogenous variables from the 2009 National
Household Travel Survey. They find that the rebound effect is only significant for the lowest
income households (up to $25,000). Wang and Chen hypothesize that travel demand for these
households are far from saturation, therefore getting more fuel efficient cars provides the
opportunity to fulfil so called "latent demand."
Hymel and Small (2015)27 revisit the simultaneous equations methodology of Small and Van
Dender (2007) and Hymel, Small and Van Dender (2010) to see whether their previous estimates
of the VMT rebound effect have changed by adding in more recent data from the late 2000 time
period (e.g., 2005-2009). Consistent with previous results, the VMT rebound effect declines
with increasing income and urbanization, and it increases with increasing fuel cost. By far the
most important of these sources of variation is income, whose effect is large enough to greatly
reduce the projected rebound effect for time periods of interest to current policy decisions. The
3-16
-------
Economic and Other Key Inputs Used in EPA's Analyses
best estimate of the long-run light-duty vehicle rebound effect over the years 2000-2009 is 17.8
percent, when evaluated at average values of income, fuel cost, and urbanization in the U.S.
during that time period.
The recent study by Hymel and Small also finds a strengthening of the VMT rebound effect
for the years 2003-2009 compared to the results for time periods from their previous research,
suggesting that some additional unaccounted for factors have increased the rebound effect.
Three potential factors are hypothesized to have caused the upward shift in the VMT rebound
effect in the 2003-2009 time period: (1) media coverage, (2) price volatility, and (3) asymmetric
response to price changes.1 It should be noted that the while media coverage and volatility are
important to understand the rebound effect based upon fuel prices, they may not be as relevant to
the rebound effect due to fuel efficiency. These results show strong evidence of asymmetry in
responsiveness to price increases and decreases. Results suggest that a rebound adjustment to
fuel price rises takes place quickly; the rebound response elasticity is large in the year of, and the
first year following, a price rise, then diminishes to a smaller value. The rebound response to
price decreases occurs more slowly.
Hymel and Small find that there is an upward shift in the rebound effect of roughly 2.5 to 2.8
percentage points starting in 2003. Results suggest that the media coverage and volatility
variables may explain about half of the upward shift in the LDV rebound effect in the 2003-2009
time period. Nevertheless, these influences are small enough in magnitude that they do not fully
offset the downward trend in VMT response elasticities due to higher incomes and other factors.
Hence, even assuming that the variables retain their 2003-2009 values into the indefinite future,
they would not prevent a further diminishing of the magnitude of the rebound effect if incomes
continue to grow at anything like historic rates.
West et al. (2015)28 attempt to estimate the VMT rebound effect using household level data
from Texas using a discontinuity in the eligibility requirements for the 2009 U.S. "Cash for
Clunkers" program, which incentivized eligible households to purchase more fuel-efficient
vehicles. Households that owned "clunkers" with a fuel economy of 18 miles per gallon (MPG)
or less were eligible for the subsidy, while households owning clunkers with an MPG of 19 or
more were ineligible. The empirical strategy of the paper is to compare the fuel economy of
vehicle purchases and subsequent vehicle miles traveled of "barely eligible" households to those
households who were "barely ineligible."
The paper finds a meaningful discontinuity in the fuel economy of new vehicles purchased by
Cash for Clunker-eligible households relative to ineligible households. Those authors report that
the increases in fuel economy realized by households who scrapped low fuel economy vehicles
in response to the substantial financial incentives offered under the federal "Cash for Clunkers"
program were not accompanied by increased use of the higher-MPG replacement vehicles they
purchased because of the vehicle's other attributes. Households chose to buy cheaper, smaller
and lower-performing vehicles. As a result, they did not drive any additional miles after the
1 The media coverage variable is measured by constructing measures of media coverage based upon gas-price related
articles appearing in the New York Times newspaper. Using the ProQuest historical database, they tally the
annual number of article titles containing the words gasoline (or gas) and price (or cost). They then form a
variable equal to the annual fraction of all New York Times articles that are gas-price-related. This fraction
ranged from roughly 1/4000 during the 1960s to a high of 1/500 in 1974.
3-17
-------
Economic and Other Key Inputs Used in EPA's Analyses
purchase of the fuel efficient vehicle. They conclude there is no evidence of a rebound effect in
response to improved fuel economy from the Cash for Clunkers program.
It may be difficult to generalize the VMT response from the Cash for Clunkers program to a
program for LDV GHG/fuel economy standards. Throughout this and all previous analyses of
the likely effects of federal regulations to require increased fuel economy and reduce vehicles'
GHG emissions, EPA and NHTSA have stressed that manufacturers can achieve the required
improvements without compromising the performance, passenger-, cargo-carrying, and towing
capacity, safety, or other attributes affecting the utility buyers and owners derive from the
vehicles they choose to purchase. The Cash for Clunkers program was a one-time program for a
fixed fleet of existing vehicles with specific characteristics. Their study may not provide useful
implications about the likely response of vehicle use to required increases in fuel economy that
are achieved through temporary incentive programs offered during recessions.
More recently, De Borger et al. (2016)29 analyze the response of vehicle use to changes in
fuel economy among a sample of nearly 350,000 Danish households owning a single vehicle, of
which almost one-third replaced it with a different model sometime during the period from 2001
to 2011. By comparing the change in households' driving from the early years of this period to
its later years among those who replaced their vehicles during the intervening period to that
among households who kept their original vehicles, the authors claim to isolate the effect of
changes in fuel economy on vehicle use from those of other factors. Their data allow them to
control for the effects of important household characteristics and vehicle features other than fuel
economy on vehicle use. They use complex statistical methods to account for the fact that some
households replacing their vehicles may have done so in anticipation of changes in their driving
demands (rather than the reverse), as well as for the possibility that some households who
replaced their cars may have done so because their driving behavior was more sensitive to fuel
prices than other households.
De Borger et al. measure the rebound effect from the change in households' vehicle use in
response to changes in fuel economy that are a consequence of their decisions to replace the
vehicles they owned previously. Thus they are able to directly estimate the fuel economy
rebound effect itself, in contrast to other research that relies on indirect measures. Their
preferred estimates span a very narrow range - from 8-10 percent - and vary only minimally in
response to different statistical estimation procedures. They also vary little depending on
whether the data sample is restricted to households that replaced their vehicles, in which case the
rebound effect is identified exclusively by their responses to changes in fuel economy of varying
magnitudes, or also includes households that did not replace their vehicles, and is thus identified
partly by differences between their responses to varying fuel economy and changes in driving
among households with vehicles whose fuel economy remained unchanged. Finally, De Borger
et al. find no evidence that the rebound effect is smaller among lower-income households than
among their higher-income counterparts. We discuss further below the limitations in applying
findings of studies from other countries to U.S. rebound.
Gillingham et al. (2016)30 undertake a summary and review of the general rebound literature
including, for example, rebound effects from LDVs as well as electricity used in stationary
applications. The literature suggests that differences in estimates of the rebound effect stem
from its varying definitions, as well as variation in the quality of data and empirical
methodologies used to estimate it. Gillingham et al. seek to clarify the definition of each of the
3-18
-------
Economic and Other Key Inputs Used in EPA's Analyses
channels of the rebound effect and critically assess the state of the literature that estimates its
magnitude.
Gillingham et al. note that most analyses assume a "zero cost breakthrough" (ZCB) - their
term for an improvement in efficiency that results in energy savings and related energy or fuel
cost savings, but does not have associated increased costs of technology or implementation.
Thus, the authors argue, most analyses do not reflect the true costs of a "policy-induced
improvement", noting: "In most cases when there is an energy efficiency policy there are also
changes in costs and attributes, the responses to which are difficult to disentangle empirically. To
analyze such an energy efficiency policy, it is essential to know all of the pertinent consumer and
market responses to the improved efficiency, changed attributes, and increased cost...most
studies that aim to estimate the rebound effect have an exogenous increase in energy efficiency
in mind; fewer are examining an actual energy efficiency policy." Failing to account for the
increased costs of equipment and/or implementation of a policy-induced improvement,
Gillingham et al. caution may result in different estimates of the rebound effect compared to a
ZCB improvement in efficiency.
Gillingham et al. also provide a list of what they consider to be relevant rebound elasticities
that can provide guidance to policymakers, with a focus on studies of overall demand or
household-level demand. According to the authors, the studies are selected both because they
are more recent and use rigorous empirical methods such as panel data methods, experimental
designs, and quasi-experimental approaches.
Of the selected studies, four focus on VMT elasticities for light-duty vehicles in developed
countries. For the Frondel and Vance study (cited above), which reported a short-run elasticity of
VMT demand for Germany for the time period from 1997-2009, Gillingham et al. chose the 46
percent value.J Barla (2009)31 found a short-run elasticity of VMT for Canada from 1990-2004
of eight percent. Gillingham (2014) (cited above) found a California medium-run new vehicle
elasticity of VMT demand for the time period 2001-2009 of 23 percent. Small and Van Dender
(2007) (cited previously) found a U.S. short-run elasticity of VMT demand for the time period
from 1966-2001 of roughly five percent.
It is not clear whether studies of LDV VMT rebound estimates for countries different from the
U.S. would provide estimates that are appropriate to the U.S. context. For example, European
countries have higher fuel prices and more transit options, both factors which would possibly
produce a VMT rebound effect that is higher than in the U.S.
3.4.4 Basis for Rebound Effect Used in this Proposed Determination
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 for those studies that look
at VMT time trends. The recent literature is mixed, with some studies supporting relatively
modest direct VMT rebound estimates and other studies suggesting a higher rebound effect.
Some of these studies come to these varied conclusions despite using the same data set.
1 Gillingham et al. believe that this value is derived by more successfully holding exogenous factors constant in the
Frondel and Vance study.
3-19
-------
Economic and Other Key Inputs Used in EPA's Analyses
EPA uses a single point estimate for the direct VMT rebound effect as an input to the agency's
analyses. Based on a combination of historical estimates of the rebound effect and more recent
analyses, an estimate of 10 percent for the long-run rebound effect is used for evaluating the
MY2022-2025 standards for this Proposed Determination (i.e., we assume a 10 percent decrease
in fuel cost per mile from the standards would result in a 1 percent increase in VMT). This
rebound effect does not include "indirect" or "economy-wide" rebound effects.
As mentioned above, for the reasons described in Section 3.4.2, historical estimates of the
rebound effect may overstate the effect of a gradual decrease in the cost of driving due to the
standards. As a consequence, a value on the low end of the historical estimates is likely to
provide a more reliable estimate of its magnitude during the period spanned by the analysis of
the impacts of the MYs 2022-2025 standards. Studies that produce an aggregate measure of the
rebound effect are most applicable to estimating the overall VMT effects of the LDV standards.
The 10 percent estimate lies at the bottom of the 10-30 percent range of estimates for the
historical, aggregate rebound effect in most research, and at the upper end of the 5-10 percent
range of estimates for the future rebound effect reported in the relatively recent studies by Small,
Hymel and Van Dender and Greene.
Both Small, Hymel and Van Dender and Greene find that the rebound effect decreases as
household incomes rise. As incomes rise, the value of time spent driving becomes a larger
fraction of total travel costs so that vehicle use becomes less responsive to variations in fuel
costs. Since the AEO 2016 projects that household incomes will be rising throughout the
analysis period for these standards, EPA believes that it is appropriate to factor in studies that
account for income on the rebound effect. Wadud et al. (2009) and Gillingham (2014) find that
household and individual-vehicle rebound increases, respectively, with increases in household
income. On the other hand, Wang and Chen (2014) find that only low income households have a
rebound effect while De Borger et al. (2016) find no evidence that the rebound effect differs
between low households in Denmark and their higher income counterparts. Thus, the evidence
of how the rebound effect varies between households across different income classes is mixed
and inconclusive.
We believe that the rebound values that are most applicable to quantifying the impact of these
standards on VMT are based on overall aggregate rebound effects as the fuel efficiency of the
U.S.'s LDV fleet increases over time. This suggest that the Small, Hymel and Van Dender and
Greene estimates are most relevant for this analysis. Su, Linn and Liu et al., each using NHTS
2009 data, find rebound effects that vary from 11-40 percent based upon household survey data.
These widely different results based upon survey data from the same year suggest that these
studies may not necessarily provide reliable estimates of the VMT rebound effect.
Gillingham et al. (2016) cite four studies that focus on VMT elasticities for light-duty vehicles
in developed countries. Two of the four studies (for the U.S. and Canada) have short-run VMT
elasticity values below the 10 percent figure. The study for California has per-vehicle rebound
value of 23 percent, and does not reflect the reduced use of other vehicles in multi-vehicle
household fleets. A study for Germany has a considerably higher value, roughly 46 percent. A
recent study by De Borger at al. found a rebound value in the range of 10 percent for Denmark.
As noted previously, it is not clear whether studies of VMT LDV rebound estimates for countries
different from the U.S. would provide estimates that are appropriate to the U.S. context.
3-20
-------
Economic and Other Key Inputs Used in EPA's Analyses
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, based on an updated review of the
literature on this topic.
3.5 Energy Security Impacts
The National Program is designed to require improvements in the fuel economy of light-duty
vehicles and, thereby, reduce fuel consumption and GHG emissions. In turn, the program helps
to reduce U.S. petroleum imports. A reduction of U.S. petroleum consumption and imports
reduces both financial and strategic risks caused by potential sudden disruptions in global oil
supply, thus increasing U.S. energy security. This section summarizes EPA's estimates of U.S.
oil import reductions and energy security benefits of the GHG vehicle standards for model years
2022-2025.
3.5.1 Implications of Reduced Petroleum Use on U.S. Imports
U.S. energy security is generally considered as the continued availability of energy sources at
an acceptable, stable price. Most discussion of U.S. energy security revolves around the topic of
the economic costs of U.S. dependence on oil imports. While the U.S. has reduced its
consumption and increased its production of oil in recent years, it still relies on oil from
potentially unstable sources outside of the U.S. and the U.S. oil price will remain tightly linked
to the global oil market. In addition, oil exporters with a large share of global production have
the ability to raise the price of oil by exerting the monopoly power associated with a cartel, the
Organization of Petroleum Exporting Countries (OPEC), to restrict oil supply relative to demand.
These factors contribute to the vulnerability of the U.S. economy to episodic oil shocks to either
the global supply of oil or world oil price spikes.
In 2015, U.S. expenditures for imports of crude oil and petroleum products, net of revenues
for exports, were $85 billion and expenditures on both imported oil and domestic petroleum and
refined products totaled $350 billion (2015$).32 Recently, as a result of strong growth in
domestic oil production mainly from tight shale formations, U.S. production of oil has increased
while U.S. oil imports have decreased. For example, from 2012 to 2015, domestic oil production
increased by 35 percent while oil imports decreased by 38 percent.33 While oil import costs have
declined since 2011, and declined sharply as the world oil price fell from roughly $100/barrel in
2014 to $52/barrel in 2015, total oil expenditures (domestic and imported) remained near
historical highs through 2014. Post-2016 oil expenditures are projected (AEO 2016) to remain
between double and triple the average inflation-adjusted levels experienced by the U.S. from
1986 to 2002 (see Figure 3.2 below).
3-21
-------
Economic and Other Key Inputs Used in EPA's Analyses
U.S. Expenditures on Crude Oil
¦Domestic
Figure 3.2 U.S. Expenditures on Crude Oil from 1970 through 201634
Focusing on changes in oil import levels as a source of vulnerability has been standard
practice in assessing energy security in the past, but given current market trends both from
domestic and international levels, adding changes in consumption of petroleum to this
assessment may provide better information about U.S. energy security. The major mechanism
through which the economy sustains harm due to fluctuations in the world energy market is
through price, which itself is leveraged through both imports and consumption. While the
United States may be increasingly insulated from the physical effects of overseas oil disruptions,
the price impacts of an oil disruption anywhere will continue to be transmitted to U.S. markets.
As of 2015, Canada accounted for 43 percent of U.S. net oil imports of crude oil and petroleum
products.0 The implications of the U.S. becoming a significant petroleum producer have yet to
be discerned in the literature, but it can be anticipated that this will have some impact on energy
security.
In 2010, just over 40 percent of world oil supply came from OPEC nations. The AEO 2016
Reference Case36 projects that this share will stay high and gradually rise; reaching 43 percent by
2020 and 45 percent by 2035 and thereafter. Approximately 32 percent of global supply is from
Middle East and North African countries alone, a share that is also expected to grow over the
long term. Measured in terms of the share of world oil resources or the share of global oil export
supply, rather than oil production, the concentration of global petroleum resources in OPEC
nations is even larger. As another measure of concentration, of the 137 countries/principalities
that export either crude or refined products, the top 12 have accounted for, in recent years,
between 55 and 70 percent of global exports.3' Eight of these countries are members of OPEC,
3-22
-------
Economic and Other Key Inputs Used in EPA's Analyses
and a ninth is Russia.K In a market where even a 1-2 percent supply loss can raise prices
noticeably, and where a 10 percent supply loss could lead to an unprecedented price shock, this
regional concentration is of concern.L Historically, the countries of the Middle East have been
the source of eight of the ten major world oil disruptions,38 with the ninth originating in
Venezuela, an OPEC country, and the tenth being Hurricanes Katrina and Rita.
EPA uses a processed combination of the MOVES and OMEGA models to estimate the
reductions in U.S. fuel consumption due to the LDV GHG standards. Based on a detailed
analysis of differences in U.S. fuel consumption, petroleum imports, and imports of petroleum
products, the agency estimates that approximately 90 percent of the reduction in fuel
consumption resulting from adopting improved GHG emission standards is likely to be reflected
in reduced U.S. imports of crude oil and net imported petroleum products.39 Thus, on balance,
each gallon of fuel saved as a consequence of the LDV GHG standards is anticipated to reduce
total U.S. imports of petroleum by 0.9 gallons. Based upon the fuel savings estimated by the
models and the 90 percent oil import factor, the reduction in U.S. oil imports from the 2022-
2025 LDV GHG standards are estimated for selected years from 2022 to 2050 (in millions of
barrels per day (MMBD) in Table 3.10 below. For comparison purposes, Table 3.10 also shows
U.S. oil exports/imports, U.S. net product imports and U.S. net crude/product imports in selected
years from 2022 to 2040, as projected by DOE in the Annual Energy Outlook 2016 Reference
Case. Real U.S. Gross Domestic Product (GDP) is projected to grow by 47 percent over the
same time frame (e.g., from 2022 to 2040) in the AEO 2016 Reference projections. Real U.S.
GDP is modestly lower in the AEO 2016 than in the AEO 2015 Reference projection. The AEO
2015 projects that real U.S. GDP will grow by 52 percent during that same time frame.
K The other three are Norway, Canada, and the EU, an exporter of product.
L For example, the 2005 Hurricanes Katrina/Rita and the 2011 Libyan conflict both led to a 1.8 percent reduction in
global crude supply. While the price impact of the latter is not easily distinguished given the rapidly rising post-
recession prices, the former event was associated with a 10-15 percent world oil price increase. There are a range
of smaller events with smaller but noticeable impacts. Somewhat larger events, such as the 2002-2003
Venezuelan Strike and the War in Iraq, corresponded to about a 2.9 percent sustained loss of supply, and was
associated with a 28 percent world oil price increase. Compiled from EIA oil price data, IEA2012 [IEA Response
System for Oil Supply Emergencies
(http://wvvw.iea.on>/publications/freepublications/publication/EPPD Brochure English 2012 02.pdf) [EPA-
HQ-OAR-2014-0827-0573] See table on P. 1 Land Hamilton 2011 "Historical Oil Shocks,"
(http://econweb.ucal.edu/~ihaniilto/otl history.pdfin IEPA-HO-OAR-2Q14-0827-05981 Routledge Handbook of
Major Events in Economic History*, pp. 239-265, edited by Randall E. Parker and Robert Whaples, New York:
Routledge Taylor and Francis Group, 2013).
3-23
-------
Economic and Other Key Inputs Used in EPA's Analyses
Table 3.10 Projected Trends in U.S. Oil Exports/Imports, and U.S. Oil Import Reductions Resulting from the
Program in Selected Years from 2022 to 2050, (Millions of barrels per day (MMBD)
Year
U.S. Oil
U.S. Gross Oil Imports
U.S. Net Product
U.S. Net
U.S. Reductions
Exports
Imports*
Crude &
Product
Imports
from Oil Imports
2022
0.63
7.56
-3.39
3.54
0.019
2023
0.63
7.57
-3.44
3.50
0.055
2024
0.63
7.57
-3.57
3.37
0.106
2025
0.63
7.58
-3.69
3.26
0.169
2030
0.63
7.20
-4.32
2.25
0.420
2035
0.83
7.07
-4.52
1.72
0.685
2040
1.02
7.12
-4.66
1.44
0.880
2050
**
**
**
**
1.119
Notes:
* Negative U.S. Net Product Imports imply positive exports.
**The AEO 2016 only projects energy market and economic trends through 2040.
3.5.2 Energy Security Implications
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 ORNL study is an updated version of the approach used for estimating the energy
security benefits of U.S. oil import reductions developed in a 1997 ORNL Report.40 This
approach has been used to estimate energy security benefits for the LDV GHG/fuel economy
standards (2012-2016; 2017-2025) and the HDV GHG/fuel economy standards Phase I (2014-
2018)/Phase II (2018 and later). For these rulemakings, the ORNL methodology is updated
periodically to account for forecasts of future energy market and economic trends reported in the
U.S. Energy Information Administration's (EIA) AEO. The agency continues to monitor the
energy security literature for new information that could influence our energy security analysis.
When conducting this analysis, ORNL considered the full cost of importing petroleum into
the U.S. The full economic cost is defined to include two components in addition to the
purchase price of petroleum itself. These are: (1) the higher costs for oil imports resulting from
the effect of U.S. demand on the world oil price (i.e., the "demand" or "monopsony" costs); and
(2) the risk of reductions in U.S. economic output and disruption to the U.S. economy caused by
sudden disruptions in the supply of imported oil to the U.S. (i.e., macroeconomic
disruption/adjustment costs).
For this Proposed Determination, ORNL updated the energy security premiums by
incorporating the most recent oil price forecast and energy market trends, particularly regional
oil supplies and demands, from the AEO 2016 Reference Case into its model.41 Below are
3-24
-------
Economic and Other Key Inputs Used in EPA's Analyses
ORNL energy security premium estimates for the selected years from 2020 to 2050,M as well as
a breakdown of the components of the energy security premiums for each year. The energy
security premiums estimated for the Proposed Determination are lower than those estimated for
the Draft TAR, because the values for the Proposed Determination are based upon the AEO 2016
Reference Case projections, which has slightly (4-5 percent) lower oil prices and significantly
(18-44 percent) lower U.S. oil imports in 2030-2035 compared to the AEO 2015 Reference
Case. The components of the energy security premiums and their values are discussed below.
Table 3.11 Energy Security Premiums in Selected Years from 2022 to 2050, (2015 $/Barrel)*
Year
Monopsony
Avoided Macroeconomic
Total Mid-Point
(Range)
Disruption/Adjustment
Costs
(Range)
(Range)
2020
$2.92
$5.48
$8.40
($0.66-$3.65)
($2.64-$8.93)
($4.97-$12.13)
2025
$2.98
$6.28
$9.25
($0.77 - $4.21)
($2.98-$10.21)
($5.48 - $13.32)
2030
$2.07
$6.89
$8.94
($0.84 - $4.64)
($3.06-$11.16)
($5.22-$12.98)
2035
$1.66
$7.50
$9.15
($1.12-$6.28)
($3.23-$12.10)
($5.24 - $13.42)
2040
$1.52
$8.08
$9.59
($1.21-$6.29)
($3.41 - $13.04)
($5.41-$14.19)
2045
$1.52
$8.08
$9.59
($1.21-$6.29)
($3.41 - $13.04)
($5.41-$14.19)
2050
$1.52
$8.08
$9.59
($1.21-$6.29)
($3.41 - $13.04)
($5.41-$14.19)
Note:
* The top values in each cell are the midpoints; the values in parentheses are the 90 percent confidence intervals.
3.5.2.1 Effect of Oil Use on the Long-Run Oil Price
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 global oil supplies, its purchases can affect the world oil
price. This monopsony power 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. Thus, one benefit of decreasing U.S. oil purchases due to reductions
in greenhouse gas emissions from light-duty vehicles is the potential decrease in the crude oil
price paid for all crude oil purchased.
A variety of oil market and economic factors have contributed to lowering the estimated
monopsony premium compared to monopsony premiums cited in previous 2017-2025 LDV
GHG/fuel economy rulemakings. Three principal factors contribute to lowering the monopsony
premium: lower world oil prices, lower U.S. oil imports, and less responsiveness of world oil
prices to changes in U.S. oil demand. Below we consider differences in oil market trends by
M AEO 2016 forecasts energy market trends and values only to 2040. The post-2040 energy security premium
values are assumed to be equal to the 2040 estimate.
3-25
-------
Economic and Other Key Inputs Used in EPA's Analyses
comparing projections developed using the AEO 2012 (Early Release) and the AEO 2016. The
AEO 2012 (Early Release) was used for the 2012 final LDV GHG/fuel economy rule and the
AEO 2016 is being used for this Proposed Determination assessment, so the comparison gives a
snapshot of how oil and energy markets have changed since the 2012 final rule.
The comparison shows a general downward revision in world oil price projections (e.g., a 31
percent reduction in 2025) and a reduction in projected U.S. oil imports due to increased U.S.
supply (i.e., a 52 percent reduction in 2025) from the AEO 2012 (Early Release) to the AEO
2016. Based upon the AEO 2016 projections over the longer term and as the world oil price
recovers, total U.S. imports are projected to gradually decrease and be 72 percent below the AEO
2012 (Early Release) projected level in 2035. The 72 percent reduction figure using the AEO
2016 Reference Case shows lower U.S. oil imports than if the AEO 2015 Reference Case is
used. For the AEO 2015, U.S. oil imports only decline by 50 percent compared to the AEO 2012
(Early Release). The AEO 2016 Reference Case estimates of U.S. oil imports are lower than the
AEO 2015 Reference Case estimates because the U.S. is producing more oil and thereby
importing less oil over the AEO time frame. Projected U.S. oil demand in the AEO 2016 is little
changed (within 2 percent) of the AEO 2015 projections through 2035.
Currently some OPEC countries (e.g., Saudi Arabia) are increasing oil supply in an attempt to
price more expensive marginal suppliers, like the U.S., out of the market and regain market
share, exacerbating the worldwide oil supply glut which has resulted in lowering the world oil
price further. Lower world oil prices currently may reduce both production from existing
domestic oil resources and investment in new domestic oil sources increasing U.S. oil import
levels in the intermediate term.
Another factor influencing the monopsony premium is that U.S. demand on the global oil
market is projected to decline, suggesting diminished overall influence and some reduction in the
influence of U.S. oil demand on the world price of oil. This is a result of the U.S. being a
smaller fraction of total world oil demand. Outside of the U.S., projected OPEC supply in the
AEO 2016 remains roughly steady as a share of world oil supply compared to the AEO 2012
(Early Release). OPEC's share of world oil supply outside of the U.S. actually increases slightly
over the long term. Since OPEC supply is estimated to be more price sensitive than non-OPEC
supply, this high OPEC share means that AEO 2016 projected world oil supply is slightly more
responsive to changes in U.S. oil demand. Together, these factors suggest that changes in U.S.
oil import reductions have a somewhat smaller effect on the long-run world oil price than
changes based on AEO 2012 (Early Release) estimates.
These changes in oil price and import levels lower the monopsony portion of energy security
premium since this portion of the security premium is related to the change in total U.S. oil
import costs that is achieved by a marginal reduction in U.S oil imports. Since both the price and
the quantity of oil imports are lower, the monopsony premium component estimated in this
assessment is 70-80 percent lower over the years 2025-2040 than the estimates based upon the
AEO 2012 (Early Release) projections.
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 value for the Social Cost
of Carbon (SCC) the question arises: how should the energy security premium be used when
some benefits from the rule, such as the benefits of reducing greenhouse gas emissions, are
3-26
-------
Economic and Other Key Inputs Used in EPA's Analyses
calculated from a global perspective? Monopsony benefits represent avoided payments by U.S.
consumers to oil producers that result from a decrease in the world oil price as the U.S. decreases
its demand for oil. Although there is clearly an overall benefit to the U.S. when considered from
a domestic perspective, the decrease in price due to decreased demand in the U.S. also represents
a loss to oil producing countries, one of which is the U.S.
Given the redistributive nature of this monopsony effect from a global perspective, it has been
excluded in the energy security benefits calculations in past rulemakings. In contrast, the other
portion of the energy security premium, the avoided U.S. macroeconomic disruption and
adjustment cost that arises from reductions in U.S. petroleum imports, does not have offsetting
impacts outside of the U.S., and, thus, is included in the energy security benefits. To summarize,
the agency has included only the avoided macroeconomic disruption portion of the energy
security benefits to estimate the monetary value of the total energy security benefits.
There is disagreement in the literature about the magnitude of the monopsony component, and
its relevance for policy analysis. Brown and Huntington (2013)42, for example, argue that the
U.S.'s refusal to exercise its market power to reduce the world oil price does not represent a
proper externality, and that the monopsony component should not be considered in calculations
of the energy security externality. However, they also note in their earlier discussion paper
(Brown and Huntington 2010)43 that this is a departure from the traditional energy security
literature, which includes sustained wealth transfers associated with stable but higher-price oil
markets.
On the other hand, Greene (2010)44 and others in prior literature (e.g., Toman 1993)45 have
emphasized that the monopsony cost component is policy-relevant because the world oil market
is non-competitive and strongly influenced by cartelized and government-controlled supply
decisions. Thus, while sometimes couched as an externality, Greene notes that the monopsony
component is best viewed as stemming from a completely different market failure than an
externality (Ledyard 2008),46 yet still implying marginal social costs to importers.
The Council on Foreign Relations47 (Council (2015)) released a discussion paper that assesses
NHTSA's analysis of the benefits and costs of CAFE in a lower-oil-price world. In this paper,
the Council notes that while NHTSA cites the monopsony effect of the CAFE standards for
2017-2025, NHTSA does not include it when calculating the cost-benefit calculation for the
rule. The Council argues that the monopsony benefit should be included in the CAFE cost-
benefit analysis and that including the monopsony benefit is more consistent with the legislators'
intent in mandating CAFE standards in the first place. The same comment the Council raised
about NHTSA's CAFE standards would apply to these GHG vehicle standards.
The National Academy of Sciences (NAS (2015)) Report, "Cost, Effectiveness and the
Deployment of Fuel Economy Technologies for Light-Duty Vehicles,"48 suggests that the
agency's logic about not accounting for monopsony benefits is inaccurate. According to the
NAS, the fallacy lies in treating the two problems, oil dependence and climate change, similarly.
According to the NAS, "Like national defense, it [oil dependence] is inherently adversarial (i.e.,
oil consumers against producers using monopoly power to raise prices). The problem of climate
change is inherently global and requires global action. If each nation considered only the
benefits to itself in determining what actions to take to mitigate climate change, an adequate
solution could not be achieved. Likewise, if the U.S. considers the economic harm its reduced
petroleum use will do to monopolistic oil producers it will not adequately address its oil
3-27
-------
Economic and Other Key Inputs Used in EPA's Analyses
dependence problem. Thus, if the United States is to solve both of these problems it must take
full account of the costs and benefits of each, using the appropriate scope for each problem."
Based upon the assessment of the monopsony premium in the Council of Foreign Relations and
NAS reports, we sought public input in the Draft TAR on whether it is appropriate to consider
monopsony in the societal costs/benefits of the National Program but received no comments.
There is also a question about the ability of gradual, long-term reductions, such as those
resulting from the LDV GHG standards, to reduce the world oil price in the presence of OPEC's
monopoly power. OPEC is currently the world's marginal petroleum supplier, and could
conceivably respond to gradual reductions in U.S. demand with gradual reductions in supply
over the course of several years as the fuel savings resulting from this program grow. However,
if OPEC opts for a long-term strategy to preserve its market share, rather than maintain a
particular price level (as they have done recently in response to increasing U.S. petroleum
production) reduced demand would create downward pressure on the global price. The Oak
Ridge analysis assumes that OPEC does respond to demand reductions by reducing its supply
over the long run, but there is still a price effect in the model because the supply reduction only
partially offsets the demand reduction, enough to maintain supply share. Under the mid-case
behavioral assumption used in the premium calculations, OPEC responds by gradually reducing
supply to maintain market share (consistent with the long-term self-interested strategy suggested
by Gately (2004, 2007)).49
It is important to note that the decrease in global petroleum prices resulting from these GHG
standards could spur increased consumption of petroleum in other sectors and countries, leading
to a modest uptick in GHG emissions outside of the U.S. This increase in global fuel
consumption could offset some portion of the GHG reduction benefits associated with these
standards. The agency has not quantified this increase in global oil consumption or GHG
emissions outside the U.S. due to world oil price changes resulting from the standards. Recent
research has quantified this type of effect in the context of biofuel policies (e.g., Drabik and de
Gorter (2011);50 Rajagopal, Hochman and Zilberman (2011);51 Thompson, Whistance, and
Meyer (2011)),52 pipeline construction (Erickson and Lazarus (2014)),53 and fuel economy
policies (Karplus et al., (2015)).54
Quantifying resulting GHG emissions may be challenging because other fuels, with varying
GHG intensities, could be displaced from the increasing use of oil worldwide, particularly
outside of the transportation sector. 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. We sought comment in the Draft TAR on whether
there are robust methodologies that could be used to estimate world-wide changes in oil
consumption and GHG emission impacts in the societal cost/benefit analysis of the National
Program but received no comments.
3.5.2.2 Macroeconomic Disruption Adjustment Costs
The second component of the oil import premium, "avoided macroeconomic
disruption/adjustment costs," arises from the effect of oil imports on the expected cost of supply
disruptions and accompanying price increases. 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 and (2) it can lead to macroeconomic contraction, dislocation and Gross Domestic
3-28
-------
Economic and Other Key Inputs Used in EPA's Analyses
Product (GDP) losses. For example, for the Proposed Determination, ORNL estimates the
combined value of these two factors to be $6.28/barrel when U.S. oil imports are reduced in
2025, with a range from $2.98/barrel to $10.21/barrel of imported oil reduced, which are
consistent with the values estimated in the Draft TAR. For the Draft TAR, the avoided
macroeconomic disruption/adjustment costs with U.S. oil imports reductions in 2025 were
$6.30/barrel with a range of $2.92/barrel to $10.22/barrel (2013$).
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, it is only the change in the expected costs of disruption that results from the
policy that 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 (e.g., elasticity) of
petroleum use.
With updated oil market and economic factors, the avoided macroeconomic disruption
component of the energy security over time is somewhat lower compared to the avoided
macroeconomic disruption premiums used in the 2017-2025 LDV GHG/fuel economy rule
(based upon the AEO 2012 (Early Release) and the Draft TAR (based upon the AEO
2015). Factors that contribute to moderately lowering the avoided macroeconomic disruption
component are lower U.S. imports (reducing the global reliance on unstable supplies, and
slightly diminishing the marginal effect of further U.S. imports reduction on global supply
stability), lower real oil prices and slightly smaller price increases during prospective shocks.
Real oil price levels in the AEO 2016 are 6-31 percent lower over the 2025-2040 period than the
AEO 2012 (Early Release), and the likely increase in oil prices in the event of an oil shock are
somewhat smaller, reflecting small increases in the responsiveness of global oil supply to
changes in the world price of oil. Over the 2025-2040 period AEO 2016 projects domestic oil
demand, and real GDP levels, are not significantly changed from AEO 2012 (Early Release) and
from the Draft TAR. Oil demand is within 2 percent and GDP is within zero to 4 percent lower.
So oil remains an important input to the U.S. economy. Overall, the avoided macroeconomic
disruption component estimates for the oil security premiums are 26-29 percent lower over the
period from 2025-2040 based upon different projected oil market and economic trends in the
AEO 2016 compared to the AEO 2012 (Early Release). Compared to the Draft TAR, the
avoided macroeconomic disruption component estimates for the oil security premiums are 4-28
percent lower over the period from 2025-2040 based upon different projected oil market and
economic trends in the AEO 2015 compared to the AEO 2012 (Early Release).
There are several reasons why the avoided macroeconomic disruption premiums changed only
moderately. One reason is that the projected macroeconomic sensitivity to oil price shocks is
held unchanged from the historical average levels used in multiple prior estimates, since
projected U.S. oil consumption levels and the expenditures on oil in the U.S. economy remain at
comparatively high levels under both AEO 2012 (Early Release) and AEO 2016. Figure 3.3
below shows that under AEO 2016, projected U.S. real annual oil expenditures continue to rise
after 2016 from under $300 billion to over $820 billion (2015$) by 2035. The value share of
3-29
-------
Economic and Other Key Inputs Used in EPA's Analyses
U.S. oil use, labeled in the figure below as U.S. oil expenditures as share of GDP, remains at
roughly three percent after 2020 even as the economy grows, lower than the AEO 2012 (Early
Release) projection of 4.4 percent declining to 3.6 percent. The value share of oil use in the
AEO 2016 is still projected to be above the full historical average (2.8 percent for 1970-2010),
and well above the historical levels observed from 1985 to 2005 (1.9 percent). A second factor
is that oil disruption risks are little changed. The two factors influencing disruption risks are the
probability of global supply interruptions and the world oil supply share from OPEC. Both
factors are not significantly different from previous forecasts of oil market trends.
900
&
i-H
o
J-T
Projected and Historical U.S. Expenditures,
and Expenditure Share, on Crude Oil
7%
¦ Domestic
~ Imported (Net)
US Oil Expenditures as Share of GDP
5% U
2% 2
1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035
Year
Figure 3.3 Projected and Historical U.S. Expenditures, and Expenditure Share, on Crude Oil55
The energy security costs estimated here follow the oil security premium framework, which is
well-established in the energy economics literature. The oil import premium gained attention as
a guiding concept for energy policy around the time of the second and third major post-war oil
shocks. Bohi and Montgomery (1982), EMF (1982)56, Plummer (1982)57 provided valuable
discussion of many of the key issues related to the oil import premium as well as the analogous
oil stockpiling premium. Bohi and Montgomery (1982)58 detailed the theoretical foundations of
the oil import premium and established many of the critical analytic relationships through their
thoughtful analysis. Hogan (1981)59 and Broadman and Hogan (1986, 1988)60 revised and
extended the established analytical framework to estimate optimal oil import premium with a
more detailed accounting of macroeconomic effects.
Since the original work on energy security was undertaken in the 1980's, there have been
several reviews on this topic. For example, Leiby, Jones, Curlee and Lee (1997)61 provided an
extended review of the literature and issues regarding the estimation of the premium. Parry and
3-30
-------
Economic and Other Key Inputs Used in EPA's Analyses
Darmstadter (2004)62 also provided an overview of extant oil security premium estimates and
estimated some premium components.
The recent economics literature on whether oil shocks are the threat to economic stability that
they once were is mixed. Some of the current literature asserts that the macroeconomic
component of the energy security externality is small. For example, the National Research
Council (2009) argued that the non-environmental externalities associated with dependence on
foreign oil are small, and potentially trivial.63 Analyses by Nordhaus (2007) and Blanchard and
Gali (2010) question the impact of more recent oil price shocks on the economy.64 They were
motivated by attempts to explain why the economy actually expanded immediately after the oil
shocks in the early 2000 time frame, and why there was no evidence of higher energy prices
being passed on through higher wage inflation. Using different methodologies, they conclude
that the economy is less sensitive to dramatic swings in oil prices.
One reason, according to Nordhaus, is that monetary policy has become more accommodating
to the price impacts of oil shocks. Another is that consumers have simply decided that such
movements are temporary, and have noted that price impacts are not passed on as inflation in
other parts of the economy. He also notes that real changes to productivity due to oil price
increases are incredibly modest,N and that the general direction of the economy matters a great
deal regarding how the economy responds to a shock. Estimates of the impact of a price shock
on aggregate demand are insignificantly different from zero.
Blanchard and Gali (2010) contend that improvements in monetary policy (as noted above),
more flexible labor markets, and lessening of energy intensity in the economy, combined with an
absence of concurrent shocks, all contributed to lessen the impact of oil shocks after 1980. They
find "... the effects of oil price shocks have changed over time, with steadily smaller effects on
prices and wages, as well as on output and employment."65 In a comment at the chapter's end,
this work is summarized as follows: "The message of this chapter is thus optimistic in that it
suggests a transformation in U.S. institutions has inoculated the economy against the responses
that we saw in the past."
At the same time, the implications of the "shale oil revolution" are now being felt in the
international markets, with current prices remain fairly low. Analysts generally attribute this
result in part to the significant increase in supply resulting from U.S. production, which has put
liquid petroleum production roughly on par with Saudi Arabia. The price decline is also
attributed to the sustained reductions in U.S. consumption and global demand growth from fuel
efficiency policies and previously high oil prices. The resulting decrease in foreign imports,
down to about one-third of domestic consumption (from 60 percent in 2005, for example66),
effectively permits U.S. supply to act as a buffer against artificial or other supply restrictions (the
latter due to conflict or a natural disaster, for example).
However, other papers suggest that oil shocks, particularly sudden supply shocks, remain a
concern. Both Blanchard and Gali's and Nordhaus work were based on data and analysis
through 2006, ending with a period of strong global economic growth and growing global oil
N In fact, "... energy-price changes have no effect on multifactor productivity and very little effect on labor
productivity." Page 19. He calculates the productivity effect of a doubling of oil prices as a decrease of 0.11
percent for one year and 0.04 percent a year for ten years. Page 5. (The doubling reflects the historical
experience of the post-war shocks, as described in Table 7.1 in Blanchard and Gali, pp. 380)
3-31
-------
Economic and Other Key Inputs Used in EPA's Analyses
demand. The Nordhaus work particularly stressed the effects of the price increase from 2002-
2006 that were comparatively gradual (about half the growth rate of the 1973 event and one-third
that of the 1990 event). The Nordhaus study emphasizes the robustness of the U.S. economy
during a time period through 2006. This time period was just before rapid further increases in
the price of oil and other commodities with oil prices more-than-doubling to over $130/barrel by
mid-2008, only to drop after the onset of the largest recession since the Great Depression in the
U.S.
Hamilton (2012)67 reviewed the empirical literature on oil shocks and suggested that the
results are mixed, noting that some work (e.g. Rasmussen and Roitman (2011) finds less
evidence for economic effects of oil shocks, or declining effects of shocks (Blanchard and Gali
2010), while other work continues to find evidence regarding the economic importance of oil
shocks. For example, Baumeister and Peersman (2011) found that an oil price increase had a
decreasing effect over time. But they note that with a declining price-elasticity of demand that a
given physical oil disruption would have a bigger effect on price and a similar effect on output as
in the earlier data. Hamilton observes that "a negative effect of oil prices on real output has also
been reported for a number of other countries, particularly when nonlinear functional forms have
been employed." Alternatively, rather than a declining effect, Ramey and Vine (2010) found
"remarkable stability in the response of aggregate real variables to oil shocks once we account
for the extra costs imposed on the economy in the 1970s by price controls and a complex system
of entitlements that led to some rationing and shortages."68
Some of the recent literature on oil price shocks has emphasized that economic impacts
depend on the nature of the oil shock, with differences between price increases caused by sudden
supply loss and those caused by rapidly growing demand. Most recent analyses of oil price
shocks have confirmed that "demand-driven" oil price shocks have greater effects on oil prices
and tend to have positive effects on the economy while "supply-driven" oil shocks still have
negative economic impacts (Baumeister, Peersman and Van Robays (2010)).69 A recent paper
by Kilian and Vigfusson (2014)70, for example, assigned a more prominent role to the effects of
price increases that are unusual, in the sense of being beyond range of recent experience. Kilian
and Vigfusson also conclude that the difference in response to oil shocks may well stem from the
different effects of demand- and supply-based price increases: "One explanation is that oil price
shocks are associated with a range of oil demand and oil supply shocks, some of which stimulate
the U.S. economy in the short run and some of which slow down U.S. growth (see Kilian
(2009)). How recessionary the response to an oil price shock is thus depends on the average
composition of oil demand and oil supply shocks over the sample period."
The general conclusion that oil supply-driven shocks reduce economic output is also reached
in a paper by Cashin et al. (2014)71 for 38 countries from 1979-2011. "The results indicate that
the economic consequences of a supply-driven oil-price shock are very different from those of an
oil-demand shock driven by global economic activity, and vary for oil-importing countries
compared to energy exporters," and "oil importers [including the U.S.] typically face a long-
lived fall in economic activity in response to a supply-driven surge in oil prices" but almost all
countries see an increase in real output for an oil-demand disturbance. Note that the energy
security premium calculation in this analysis is based on price shocks from potential future
supply events only.
3-32
-------
Economic and Other Key Inputs Used in EPA's Analyses
By early 2015, world oil prices were sharply lower than in 2014. Future prices remain
uncertain, but sustained markedly lower oil prices can have mixed implications for U.S. energy
security. Under lower prices U.S. expenditures on oil consumption are lower, and the
expenditures are a less prominent component of the U.S. economy. But sustained lower oil
prices encourage greater oil consumption, and reduce the competitiveness of new U.S. oil
supplies and alternative fuels. The AEO 2016 low-oil price outlook, for example, projects that
by 2030 total U.S. petroleum supply would be 29 percent lower and net imports would be 204
percent higher than the AEO 2016 Reference Case. Under the low-price case, 2030 crude prices
are 56 percent lower, while net imports of crude and product increase from 2.2 MMBD to 6.8
MMBD so that U.S. net import expenditures are 33 percent higher.0
A second potential proposed energy security effect of lower oil prices is increased instability
of supply, due to greater global reliance on fewer suppling nations/ and because lower prices
may increase economic and geopolitical instability in some supplier nations.72'73'74 The
International Monetary Fund reported that low oil prices are creating substantial economic
tension for Middle East oil producers on top of the economic costs of ongoing geopolitical
conflicts, and noted the risk that Middle East countries including Saudi Arabia could run out of
financial assets without a substantial change in policy.75 The concern raised is that oil revenues
are essential for some exporting nations to fund domestic programs and avoid domestic unrest.
Finally, despite continuing uncertainty about oil market behavior and outcomes and the
sensitivity of the U.S. economy to oil shocks, it is generally agreed that it is beneficial to reduce
petroleum fuel consumption from an energy security standpoint. It is not just imports alone, but
both imports and consumption of petroleum from all sources and their role in economic activity,
that may expose the U.S. to risk from price shocks in the world oil price. Reducing fuel
consumption reduces the amount of domestic economic activity associated with a commodity
whose price depends on volatile international markets. The relative significance of petroleum
consumption and import levels for the macroeconomic disturbances that follow from oil price
shocks is not fully understood. Recognizing that changing petroleum consumption will change
U.S. imports, this assessment of oil costs focuses on those incremental social costs that follow
from the resulting changes in imports, employing the usual oil import premium measure.
3.5.2.3 Cost of Existing U.S. Energy Security Policies
The last often-identified component of the full economic costs of U.S. oil imports are the
costs to the U.S. taxpayers of existing U.S. energy security policies. The two primary examples
are maintaining the Strategic Petroleum Reserve (SPR) and maintaining a military presence to
0 For simplicity and given available data, this computation treats net import expenditures as proportional to net
import volumes. For the low-oil price case net petroleum imports in 2030 are 4.6 MMBD greater than in the
Reference case, primarily due to a large reduction in product exports (4.1 MMBD smaller), and a smaller (0.5
MMDB) increase in crude imports. Since the import change is primarily due to a loss of the more highly-priced
product exports, the expenditure change could be larger.
p Fatih Birol, Executive Director of the International Energy Agency, warns that prolonged lower oil prices would
trigger energy-security concerns by increasing reliance on a small number of low-cost producers "or risk a sharp
rebound in price if investment falls short." "It would be a grave mistake to index our attention to energy security
to changes in the oil price," Birol said. "Now is not the time to relax. Quite the opposite: a period of low oil prices
is the moment to reinforce our capacity to deal with future energy security threats." International Energy Agency,
World Energy Outlook, November 10th, 2015.
3-33
-------
Economic and Other Key Inputs Used in EPA's Analyses
help secure a stable oil supply from potentially vulnerable regions of the world. The SPR is the
largest stockpile of government-owned emergency crude oil in the world. Established in the
aftermath of the 1973/1974 oil embargo, the SPR provides the U.S. with a response option
should 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. While the costs for building and maintaining the SPR
are more clearly related to U.S. oil use and imports, historically these costs have not varied in
response to changes in U.S. oil import levels. Thus, while the effect of the SPR in moderating
price shocks is factored into the ORNL analysis, the cost of maintaining the SPR is excluded.
3.5.2.4 Military Security Cost Components of Energy Security
The agency has also attempted to assess the military security benefits components of energy
security in past LDV rulemakings and the Draft TAR. The recent literature on the military
components of energy security has included three broad categories of oil related military and
national security costs all of which are hard to quantify and provide estimates of their costs.
These include possible costs of U.S. military programs to secure oil supplies from unstable
regions of the world, the energy security costs associated with the U.S. military's reliance on
petroleum to fuel its operations and possible national security costs associated with expanded oil
revenues to "rogue states."
Of these categories listed above, the one that is most clearly connected to petroleum use and
is, in principle, quantifiable is the first, the cost of military programs to secure oil supplies and
stabilize oil supplying regions. There is a developing literature on the measurement of these
components of energy security but methodological and measurement challenges pose significant
challenges to providing a robust estimate of this component of energy security.
Assessing the military component of the energy security cost has two major challenges:
attribution and incremental analysis. The attribution challenge is to determine which military
programs and expenditures can properly be attributed to oil supply protection, rather than some
other national security objective. The incremental analysis challenge is to estimate how much
the petroleum supply protection costs might vary if U.S. oil use were to be reduced or
eliminated.
Since "military forces are, to a great extent, multipurpose and fungible" across theaters and
missions (Crane et al. (2009))76, and because the military budget is presented along regional
accounts rather than by mission, the allocation to particular missions is not always clear.
Approaches taken usually either allocate "partial" military costs directly associated with
operations in a particular region, or allocate a share of total military costs (including some that
are indirect in the sense of supporting military activities overall) (Koplow and Martin (1998)).77
The incremental analysis can estimate how military costs would vary if the oil security
mission is no longer needed, and many studies stop at this point. It is substantially more difficult
to estimate how military costs would vary if U.S. oil use or imports are partially reduced. Partial
reduction of U.S. oil use diminishes the magnitude of the security problem, but there is
uncertainty that supply protection forces and their costs could be scaled down in proportion (e.g.
Crane et al. (2009))78, and there remains the associated goal of protecting supply and transit for
allies and important trade partners, and other importing countries, if they do not decrease their
petroleum use as well.
3-34
-------
Economic and Other Key Inputs Used in EPA's Analyses
The challenges of attribution and incremental analysis have led some to conclude that the
mission of oil supply protection cannot be clearly separated from others, and the military cost
component of oil security should be taken as near zero (Moore et al. (1997)).79 For example, the
Council on Foreign Relations takes the view that substantial foreign policy missions will remain
over the next 20 years, even without the oil security mission entirely. Stern, on the other hand,
argues that many of the other policy concerns in the Persian Gulf follow from oil, and the
reaction to U.S. policies taken to protect oil.
Most commonly, analysts estimate substantial military costs associated with the missions of
oil supply security and associated contingencies, but avoid estimating specific cost reductions
from partial reductions in oil use. However, some studies (Copulos (2003), Delucchi and
Murphy (2008), Crane et al., Stern (2010))80 seek to update, and in some cases significantly
improve the rigor of analysis.
Delucchi and Murphy sought to deduct from the cost of Persian Gulf military programs, the
costs associated with defending U.S. interests other than the objective of providing more stable
oil supply and price to the U.S. economy. Excluding an estimate of cost for missions unrelated
to oil, and for the protection of oil in the interest of other countries, Delucchi and Murphy
estimated military costs for all U.S. domestic oil interests of between $24 and $74 billion
annually.
Crane et al. considered force reductions and cost savings that could be achieved if oil security
were no longer a consideration. After reviewing documents supporting recent defense resource
allocations, they concluded that the oil protection mission is prominent: "First, the United States
does include the security of oil supplies and global transit of oil as a prominent element in its
force planning." While they noted that the elimination of this mission of oil supply protection
might not lead to complete reduction of those costs, they concluded there is very likely to be
some cost reduction. Taking two approaches, and guided by post-Cold War force draw downs
and by a top-down look at the current U.S. allocation of defense resources, they concluded that
$75—$91 billion, or 12-15 percent of the current U.S. defense budget, could be reduced if the oil
protection mission were completely eliminated.
Stern presents an estimate of military cost for Persian Gulf force projection, addressing the
challenge of cost allocation with an activity-based cost method. He used information on actual
naval force deployments rather than budgets, focusing on the costs of carrier deployment. As a
result of this different data set and these assumptions regarding allocation, the estimated costs are
much higher, roughly 4 to 10 times, than other recent estimates. For the 1976-2007 time frame,
Stern estimated an average military cost of $212 billion and for 2007, $500 billion.
A study by the National Research Council (NRC) (2013)81 attempted to estimate the military
costs associated with U.S. imports and consumption of petroleum. The NRC cites estimates of
the national defense costs of oil dependence from the literature that range from less than $5
billion to $50 billion per year or more. Assuming an approximate range of $10-$50 billion per
year, the NRC divided national defense costs by a projected U.S. consumption rate of
approximately 6.4 billion barrels per year (EIA, 2012). This procedure yielded a range of
average national defense cost of $1.50-$8.00 per barrel (rounded to the nearest $0.50), with a
mid-point of $5/barrel (in 2009$). However, as discussed above, it is unclear that incremental
reductions in either U.S. imports, or consumption of domestic petroleum, would produce
3-35
-------
Economic and Other Key Inputs Used in EPA's Analyses
incremental changes to the military expenditures related to the oil protection mission (Crane, et
al.). We did not receive any comments on this issue in the Draft TAR.
3.6 Non-GHG Health and Environmental Impacts
This section discusses the economic benefits from reductions in health and environmental
impacts resulting from non-GHG emission reductions (such as criteria and toxic air pollutants)
that can be expected to occur as a result of the light-duty 2022-2025 GHG standards. CO2
emissions are predominantly the byproduct of fossil fuel combustion processes that also produce
criteria and hazardous air pollutant emissions. The vehicles that are subject to this program are
also significant sources of mobile source air pollution such as directly emitted Particulate Matter
(PM), Nitrogen Oxide (NOx), Volatile Organic Chemicals (VOCs) and air toxics, which are
regulated by separate emissions standards programs. The program will affect exhaust emissions
of these pollutants from vehicles and will also affect emissions from upstream sources that occur
during the refining and distribution of fuel. Changes in ambient concentrations of ozone, PM2.5,
and air toxics that will result from the program are expected to affect human health by reducing
premature deaths and other serious human health effects, as well as other important
improvements in public health and welfare. Children especially benefit from reduced exposures
to criteria and toxic pollutants, because they tend to be more sensitive to the effects of these
respiratory pollutants. Ozone and particulate matter have been associated with increased asthma
exacerbation and other respiratory effects in children, and particulate matter has been associated
with deficits in lung function development.
It is important to quantify the co-pollutant-related health and environmental impacts
associated with the GHG standards because a failure to adequately consider these ancillary
impacts could lead to an incorrect assessment of the standards' costs and benefits. Moreover, the
health and other impacts of exposure to criteria air pollutants and airborne toxics tend to occur in
the near term, while most effects from reduced climate change are likely to occur only over a
time frame of several decades or longer.
For purposes of this Proposed Determination, EPA has applied PM-related benefits per-ton
values to its estimated emission reductions to estimate only the PM-related benefits of the
program. 82'Q However, there are several health benefit categories that EPA was unable to
quantify due to limitations associated with using benefits-per-ton estimates, several of which
could be substantial. For example, we have not quantified a number of known or suspected
health benefits linked to reductions in ozone and other criteria pollutants, as well as health
benefits linked to reductions in air toxics. Additionally, we are unable to quantify a number of
known welfare effects, including reduced acid and particulate deposition damage to cultural
monuments and other materials, and environmental benefits due to reductions of impacts of
eutrophication in coastal areas. As a result, the health benefits quantified in this analysis are
likely underestimates of total benefits.
Q See also: http://www.epa.gov/airquality/benmap/sabpt.html. The current values available on the webpage have
been updated since the publication of the Fann et al., 2012 paper. For more information regarding the updated
values, see: http://www.epa.gov/airquality/benmap/models/Source_Apportionment_BPT_TSD_l_3l_13.pdf
(accessed June 9, 2016).
3-36
-------
Economic and Other Key Inputs Used in EPA's Analyses
3.6.1 Economic Value of Reductions in Particulate Matter
As presented in Appendix C of the Proposed Determination document, the standards would
reduce emissions of several criteria and toxic pollutants and their precursors. In this analysis,
however, EPA only estimates the economic value of the human health benefits associated with
the resulting reductions in PM2.5 exposure (related to both directly emitted PM2.5 and
secondarily-formed PM2.5). Due to analytical limitations with the benefit per-ton method, this
analysis does not estimate benefits resulting from reductions in population exposure to other
criteria pollutants such as ozone.R Furthermore, the benefits per-ton method, like all air quality
impact analyses, does not monetize all of the potential health and welfare effects associated with
reduced concentrations of PM2.5.
This analysis uses estimates of the benefits from reducing the incidence of the specific PM2.5-
related health impacts described below. These estimates, which are expressed per ton of PM2.5-
related emissions eliminated by the standards, represent the total monetized value of human
health benefits (including reduction in both premature mortality and premature morbidity) from
reducing each ton of directly emitted PM2.5, or its precursors (SO2 and NOx), from a specified
source.
The PM-related dollar-per-ton benefit estimates used in this analysis, which are consistent
with those used in the Draft TAR, are provided in Table 3.12. As the table indicates, these
values differ among directly emitted PM and PM precursors (SO2 and NOx), and also depend on
their original source, because emissions from different sources can result in different degrees of
population exposure and resulting health impacts. In the summary of costs and benefits, Chapter
5, EPA presents the monetized value of total PM-related improvements associated with the
standards summed across sources (on-road and upstream) sources and across PM-related
pollutants (direct PM2.5 and PM precursors SO2 and NOx).
Table 3.12 PM-Related Benefits-per-ton Values (thousands, 2012$)a
Year0
On-road Mobile Sources
Upstream Sourcesd
Direct PM2.5
SO 2
NOx
Direct PM2.5
SO2
NOx
Estimated Using a 3 Percent Discount Rateb
2022
$400-$910
$22-$49
$8.1-$18
$350-$790
$75-$170
$7.4-$17
2025
$440-$l,000
$24-$55
$8.8-$20
$390-$870
$83-$190
$8.1-$18
2030
$480-$!, 100
$27-$61
$9.6-$22
$420-$950
$91-$200
$8.7-$20
Estimated Using a 7 Percent Discount Rateb
2022
$370-$820
$20-$44
$7.4-$17
$320-$720
$67-$150
$6.6-$15
2025
$400-$910
$22-$49
$8.0-$18
$350-$790
$75-$170
$7.3-$17
2030
$430-$980
$24-$55
$8.6-$20
$380-$850
$81-$180
$7.9-$18
Notes:
" The benefit-per-ton estimates presented in this table are based on a range of premature mortality estimates derived
from the ACS study (Krewski et al., 2009) and the Six-Cities study (Lepeule et al., 2012).
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.
R The air quality modeling that underlies the PM-related benefit per ton values also produced estimates of ozone
levels attributable to each sector. However, the complex non-linear chemistry governing ozone formation
prevented EPA from developing a complementary array of ozone benefit per ton values. This limitation
notwithstanding, we anticipate that the ozone-related benefits associated with reducing emissions of NOx and
VOC could be substantial.
3-37
-------
Economic and Other Key Inputs Used in EPA's Analyses
0 Benefit-per-ton values were estimated for the years 2020, 2025 and 2030. We hold values constant for intervening
years (e.g., 2020 values for years 2021-2024; 2025 values for years 2026-2029; and 2030 values for years 2031 and
beyond).
d We assume for the purpose of this analysis that "upstream emissions" are most closely associated with refinery sector
benefit per-ton values. The majority of upstream emission reductions associated with the standards are related to
domestic onsite refinery emissions and domestic crude production. While upstream emissions also include storage
and transport sources, as well as upstream refinery sources, we have chosen to simply apply the refinery values.
The benefit per-ton technique has been used in previous analyses, including EPA's Heavy-
Duty Vehicle GHG standards Phase II (2018 and later),83 2017-2025 Light-Duty Vehicle
Greenhouse Gas Rule,84 the Reciprocating Internal Combustion Engine rules,85'86 and the
Residential Wood Heaters NSPS.87 Table 3.13 shows the quantified PIVh.s-related co-benefits
captured in those benefit per-ton estimates, as well as unquantified effects the benefits per-ton
estimates are unable to capture.
Table 3.13 Human Health and Welfare Effects of PM2.5
Pollutant
Quantified and Monetized
in Primary Estimates
Unquantified Effects
Changes in:
PM2.5
Adult premature mortality
Acute bronchitis
Hospital admissions: respiratory and
cardiovascular
Emergency room visits for asthma
Nonfatal heart attacks (myocardial infarction)
Lower and upper respiratory illness
Minor restricted-activity days
Work loss days
Asthma exacerbations (asthmatic population)
Infant mortality
Chronic and subchronic bronchitis cases
Strokes and cerebrovascular disease
Low birth weight
Pulmonary function
Chronic respiratory diseases other than chronic
bronchitis
Non-asthma respiratory emergency room visits
Visibility
Household soiling
Readers interested in reviewing the complete methodology for creating the benefit-per-ton
estimates used in this analysis can consult EPA's "Technical Support Document: Estimating the
Benefit per Ton of Reducing PM2.5 Precursors from 17 Sectors." s Readers can also refer to Fann
et al. (2012) for a detailed description of the benefit-per-ton methodology. As described in the
documentation, EPA uses a method that is consistent with the cost-benefit analysis that
accompanied the 2012 PM NAAQS revision. The benefit-per-ton estimates utilize the
concentration-response functions as reported in the epidemiology literature.1'88 To calculate the
total monetized impacts associated with quantified health impacts, EPA applies values derived
from a number of sources. For premature mortality, EPA applies a value of a statistical life
(VSL) derived from the mortality valuation literature. For certain health impacts, such as
sFor more information regarding the updated values, see:
http://www.epa.gov/airquality/benmap/models/Source_Apportionment_BPT_TSD_l_31_13.pdf (accessed
September 9, 2014).
T Although we summarize the main issues in this chapter, we encourage interested readers to see the benefits chapter
of the RIA that accompanied the PM NAAQS for a more detailed description of recent changes to the
quantification and monetization of PM benefits. Note that the cost-benefit analysis was prepared solely for
purposes of fulfilling analysis requirements under Executive Order 12866 and was not considered, or otherwise
played any part, in the decision to revise the PM NAAQS.
3-38
-------
Economic and Other Key Inputs Used in EPA's Analyses
respiratory-related ailments, EPA applies willingness-to-pay estimates derived from the
valuation literature. For the remaining health impacts, EPA applies values derived from current
cost-of-illness and/or wage estimates.
The documentation cited above also describes that national per-ton estimates were developed
for selected PM-related pollutant/source category combinations. The per-ton values calculated
therefore apply only to tons reduced from those specific PM-related pollutant/source
combinations (e.g., NO2 emitted from on-road mobile sources; direct PM emitted from
electricity generating units). EPA's estimate of PM2.5 benefits is therefore based on the total
direct PM2.5 and PM-related precursor emissions controlled by sector and multiplied by each
per-ton value.
As Table 3.12 indicates, EPA projects that the per-ton values for reducing emissions of non-
GHG pollutants from both vehicle use and upstream sources such as fuel refineries will increase
over time.u These projected increases reflect rising income levels, which increase affected
individuals' willingness to pay for reduced exposure to health threats from air pollution.v 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 among older age
groups with the highest mortality risk.w
The benefit-per-ton estimates are subject to a number of assumptions and uncertainties:
The benefit-per-ton estimates used in this analysis reflect specific geographic patterns of
emissions reductions and specific air quality and benefits modeling assumptions associated with
the derivation of those estimates (see the separate technical documentation that describes the
calculation of the national benefit-per-ton estimates). 89'x Consequently, these estimates may not
reflect local variability in population density, meteorology, exposure, baseline health incidence
rates, or other local factors associated with the current analysis.
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 PM2.5 released from diesel engines and other industrial sources. The PM Integrated
Science Assessment (ISA), which was twice reviewed by the Science Advisory Board's Clean
Air Science Advisory Committee (SAB-CASAC), concluded that "many constituents of PM2.5
can be linked with multiple health effects, and the evidence is not yet sufficient to allow
u As we present in the Proposed Determination document, Appendix C, the standards would yield emission
reductions from upstream refining and fuel distribution due to decreased petroleum consumption.
v The issue is discussed in more detail in the 2012 PM NAAQS RIA, Section 5.6.8. See U.S. Environmental
Protection Agency. (2012). Regulatory Impact Analysis for the Final Revisions to the National Ambient Air
Quality Standards for Particulate Matter, Health and Environmental Impacts Division, Office of Air Quality
Planning and Standards, EPA-452-R-12-005, December 2012. Available on the internet:
http://www.epa.gov/ttnecasl/regdata/RIAs/finalria.pdf.
w 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)
x See also: http://www.epa.gov/airquality/benmap/sabpt.html. The current values available on the webpage have
been updated since the publication of the Fann et al., 2012 paper. For more information regarding the updated
values, see: http://www.epa.gov/airquality/benmap/models/Source_Apportionment_BPT_TSD_l_3l_13.pdf
(accessed September 9, 2014).
3-39
-------
Economic and Other Key Inputs Used in EPA's Analyses
differentiation of those constituents or sources that are more closely related to specific
outcomes."90 PM composition and the size distribution of those particles vary within and
between areas due to source characteristics. Any specific location could have higher or lower
contributions of certain PM species and other pollutants than the national average, meaning
potential regional differences in health impact of given control strategies. Depending on the
toxicity of each PM species reduced by the proposed standards, assuming equal toxicity could
over or underestimate benefits.
When estimating the benefit-per-ton values, EPA assumes that the underlying health impact
functions for fine particles are linear within the range of ambient concentrations under
consideration. Thus, the estimates include health benefits from reducing fine particles in areas
with varied concentrations of PM2.5, including regions that are in attainment with the fine
particle standard. The direction of bias that assuming a linear-no threshold model (or an
alternative model) introduces depends upon the "true" functional from of the relationship and the
specific assumptions and data in a particular analysis. For example, if the true function identifies
a threshold below which health effects do not occur, benefits may be overestimated if a
substantial portion of those benefits were estimated to occur below that threshold. Alternately, if
a substantial portion of the benefits occurred above that threshold, the benefits may be
underestimated because an assumed linear no-threshold function may not reflect the steeper
slope above that threshold to account for all health effects occurring above that threshold.
There are several health benefit categories that EPA was unable to quantify due to limitations
associated with using benefits-per-ton estimates, several of which could be substantial. Because
the NOx and VOC emission reductions associated with the standards are also precursors to
ozone, reductions in NOx and VOC would also reduce ozone formation and the health effects
associated with ozone exposure. Unfortunately, ozone-related benefits-per-ton estimates 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.
There are many uncertainties associated with the health impact functions that underlie the
benefits-per-ton estimates. These include: within-study variability (the precision with which a
given study estimates the relationship between air quality changes and health effects); across-
study variation (different published studies of the same pollutant/health effect relationship
typically do not report identical findings and in some instances the differences are substantial);
the application of concentration-response functions nationwide (does not account for any
relationship between region and health effect, to the extent that such a relationship exists);
extrapolation of impact functions across population (we assumed that certain health impact
functions applied to age ranges broader than that considered in the original epidemiological
study); and various uncertainties in the concentration-response function, including causality and
thresholds. These uncertainties may under- or over-estimate benefits.
EPA has investigated methods to characterize uncertainty in the relationship between PM2.5
exposure and premature mortality. EPA's final PM2.5 NAAQS analysis provides a more
complete picture about the overall uncertainty in PM2.5 benefits estimates. For more
information, please consult the PM2.5 NAAQS Regulatory Impacts Analysis.91
The benefit-per-ton unit values used in this analysis incorporate projections of key variables,
including atmospheric conditions, source level emissions, population, health baselines, incomes,
3-40
-------
Economic and Other Key Inputs Used in EPA's Analyses
and technology. These projections introduce additional uncertainties to the benefit per ton
estimates.
3.7 Social Cost of Greenhouse Gas Emissions
We estimate the global social benefits of CO2 emission reductions expected from the 2022-
2025 final standards using the SC-CO2 estimates presented in the Technical Support Document:
Technical Update of the Social Cost of Carbon for Regulatory Impact Analysis Under Executive
Order 12866 (May 2013, Revised August 2016) ("current TSD").92 We refer to these estimates,
which were developed by the U.S. government, as "SC-CO2 estimates." The SC-CO2 is a metric
that estimates the monetary value of impacts associated with marginal changes in CO2 emissions
in a given year. It includes a wide range of anticipated climate impacts, such as net changes in
agricultural productivity and human health, property damage from increased flood risk, and
changes in energy system costs, such as reduced costs for heating and increased costs for air
conditioning. It is typically used to assess the avoided damages as a result of regulatory actions
(i.e., benefits of rulemakings that lead to an incremental reduction in cumulative global CO2
emissions).
The SC-CO2 estimates used in the final 2017-2025 RIA and in this analysis were developed
over many years, using the best science available, and with input from the public. Specifically,
an interagency working group (IWG) that included the EPA and other executive branch agencies
and offices used three integrated assessment models (IAMs) to develop the SC-CO2 estimates
and recommended four global values for use in regulatory analyses. The SC-CO2 estimates were
first released in February 2010 and were used to estimate the value of CO2 benefits in the final
2017-2025 rulemaking.
These SC-CO2 estimates were developed using an ensemble of the three most widely cited
integrated assessment models in the economics literature with the ability to estimate the SC-CO2.
A key objective of the IWG was to draw from the insights of the three models while respecting
the different approaches to linking GHG emissions and monetized damages taken by modelers in
the published literature. After conducting an extensive literature review, the interagency group
selected three sets of input parameters (climate sensitivity, socioeconomic and emissions
trajectories, and discount rates) to use consistently in each model. All other model features were
left unchanged, relying on the model developers' best estimates and judgments, as informed by
the literature. Specifically, a common probability distribution for the equilibrium climate
sensitivity parameter, which informs the strength of climate's response to atmospheric GHG
concentrations, was used across all three models. In addition, a common range of scenarios for
the socioeconomic parameters and emissions forecasts were used in all three models. Finally,
the marginal damage estimates from the three models were estimated using a consistent range of
discount rates, 2.5, 3.0, and 5.0 percent. See Technical Support Document: Technical Update of
the Social Cost of Carbon for Regulatory Impact Analysis under Executive Order 12866
(February 2010) ("2010 TSD") for a complete discussion of the methods used to develop the
estimates and the key uncertainties, and the current TSD for the latest estimates.93
In 2013, and after the final LD 2017-2025 rulemaking, the IWG updated the SC-CO2
estimates using new versions of each IAM. The 2013 update did not revisit the 2010 modeling
decisions with regards to the discount rate, reference case socioeconomic and emission scenarios,
and equilibrium climate sensitivity distribution. Rather, improvements in the way damages are
3-41
-------
Economic and Other Key Inputs Used in EPA's Analyses
modeled are confined to those that have been incorporated into the latest versions of the models
by the developers themselves and published in the peer-reviewed literature. The model updates
that are relevant to the SC-CO2 estimates include: an explicit representation of sea level rise
damages in the Dynamic Integrated Climate and Economy (DICE) and Policy Analysis of the
Greenhouse Effect (PAGE) models; updated adaptation assumptions, revisions to ensure
damages are constrained by GDP, updated regional scaling of damages, and a revised treatment
of potentially abrupt shifts in climate damages in the PAGE model; an updated carbon cycle in
the DICE model; and updated damage functions for sea level rise impacts, the agricultural sector,
and reduced space heating requirements, as well as changes to the transient response of
temperature to the buildup of GHG concentrations and the inclusion of indirect effects of
methane emissions in the Climate Framework for Uncertainty, Negotiation, and Distribution
(FUND) model. The current TSD presents and discusses the 2013 update (including recent
minor technical corrections to the estimates).Y
The updated estimates continue to represent global measures because of the distinctive nature
of the climate change, which is highly unusual in at least three respects. First, emissions of most
GHGs contribute to damages around the world independent of the country in which they are
emitted. The SC-CO2 must therefore incorporate the full (global) damages caused by GHG
emissions to address the global nature of the problem. Second, the U.S. operates in a global and
highly interconnected economy, such that impacts on the other side of the world can affect our
economy. This means that the true costs of climate change to the U.S. are larger than the direct
impacts that simply occur within the U.S. Third, climate change represents a classic public
goods problem because each country's reductions benefit everyone else and no country can be
excluded from enjoying the benefits of other countries' reductions, even if it provides no
reductions itself. In this situation, the only way to achieve an economically efficient level of
emissions reductions is for countries to cooperate in providing mutually beneficial reductions
beyond the level that would be justified only by their own domestic benefits. In reference to the
public good nature of mitigation and its role in foreign relations, thirteen prominent academics
noted that these "are compelling reasons to focus on a global SCC" in a recent article on the SCC
(Pizer et al., 2014). In addition, as noted in OMB's Response to Comments on the SC-CO2, a
document discussed further below, there is no bright line between domestic and global damages.
Adverse impacts on other countries can have spillover effects on the United States, particularly
in the areas of national security, international trade, public health and humanitarian concerns.94
The 2010 TSD noted a number of limitations to the SC-CO2 analysis, including the
incomplete way in which the integrated assessment models capture catastrophic and non-
catastrophic impacts, their incomplete treatment of adaptation and technological change,
uncertainty in the extrapolation of damages to high temperatures, and assumptions regarding risk
aversion. Currently integrated assessment models do not assign value to all of the important
physical, ecological, and economic impacts of climate change recognized in the climate change
literature due to a lack of precise information on the nature of damages and because the science
Y Both the 2010 TSD and the current TSD are available at: https://www.whitehouse.gov/omb/oira/social-cost-of-
carbon.
3-42
-------
Economic and Other Key Inputs Used in EPA's Analyses
incorporated into these models understandably lags behind the most recent research.2 The
limited amount of research linking climate impacts to economic damages makes the modeling
exercise even more difficult. These individual limitations do not all work in the same direction
in terms of their influence on the SC-CO2 estimates, though taken together they suggest that the
SC-CO2 estimates are likely conservative. In particular, the IPCC Fourth Assessment Report
(2007), which was the most current IPCC assessment available at the time of the IWG's 2009-
2010 review, concluded that "It is very likely that [SC-CO2 estimates] underestimate the damage
costs because they cannot include many non-quantifiable impacts." Since then, the peer-
reviewed literature has continued to support this conclusion. For example, the IPCC Fifth
Assessment report observed that SC-CO2 estimates continue to omit various impacts that would
likely increase damages.
The EPA and other agencies have continued to consider feedback on the SC-CO2 estimates
from stakeholders through a range of channels, most recently including public comments on the
Clean Power Plan rulemaking95 and others that use the SC-CO2 in supporting analyses and
through regular interactions with stakeholders and research analysts implementing the SC-CO2
methodology used by the interagency working group. Several comments received on the Draft
TAR stated that the SC-CO2 underestimates climate-related benefits and discussed some of the
technical details of the modeling conducted to develop the SC-CO2 estimates. EPA recognizes
the importance of the estimates to be as complete as possible and will continue to follow and
evaluate the latest science on impact categories that are omitted or not fully addressed in the
IAMs. Some commenters also provided constructive recommendations for potential
opportunities to improve the SC-CO2 estimates in future updates. In addition, OMB sought
public comment on the approach used to develop the SC-CO2 estimates through a separate
comment period and published a response to those comments in 2015.^
After careful evaluation of the full range of comments submitted to OMB, the IWG continues
to recommend the use of the SC-CO2 estimates in regulatory impact analysis while also
continuing to engage in research on modeling and valuation of climate impacts. Currently, the
IWG is seeking advice from the National Academies of Sciences, Engineering and Medicine on
how to approach future updates to ensure that the estimates continue to reflect the best available
scientific and economic information on climate change.BB An Academies committee,
"Assessing Approaches to Updating the Social Cost of Carbon," (Committee) will provide
expert, independent advice on the merits of different technical approaches for modeling and
highlight research priorities going forward. EPA will evaluate its approach based upon any
feedback received from the Academies' panel.
z Climate change impacts and SCC modeling is an area of active research. For example, see: (1) Howard, Peter,
"Omitted Damages: What's Missing from the Social Cost of Carbon." March 13, 2014,
http://costofcarbon.org/files/Omitted_Damages_Whats_Missing_From_the_Social_Cost_of_Carbon.pdf; and (2)
Electric Power Research Institute, "Understanding the Social Cost of carbon: A Technical Assessment," October
2014, www.epri.com.
AA See https://www.whitehouse.gov/sites/default/files/omb/inforeg/scc-response-to-comments-final-july-2015.pdf.
BB The Academies' review will be informed by public comments and focus on the technical merits and challenges of
potential approaches to improving the SC-C02 estimates in future updates. See
https://www.whitehouse.gov/blog/2015/07/02/estimating-benefits-carbon-dioxide-emissions-reductions.
3-43
-------
Economic and Other Key Inputs Used in EPA's Analyses
To date, the Committee has released an interim report, which recommended against doing a
near term update of the SC-CO2 estimates. For future revisions, the Committee recommended
the IWG move efforts towards a broader update of the climate system module consistent with the
most recent, best available science, and also offered recommendations for how to enhance the
discussion and presentation of uncertainty in the SC-CO2 estimates. Specifically, the Committee
recommended that "the IWG provide guidance in their technical support documents about how
[SC-CO2] uncertainty should be represented and discussed in individual regulatory impact
analyses that use the [SC-CO2]" and that the technical support document for each update of the
estimates present a section discussing the uncertainty in the overall approach, in the models used,
and uncertainty that may not be included in the estimates.cc In August 2016, the IWG issued
revisions to the SC-CO2 Technical Support Document that responded to interim
recommendations from the Academies regarding the presentation and discussion of uncertainty.
The revision did not modify methodological decisions or change the SC-CO2 estimates
themselves. The Committee will release a final report in early 2017 with longer-term
recommendations for updating the estimates.
The current SC-CO2 estimates are as follows: $15, $49, $72, and $150 per ton of CO2
emissions in the year 2022 (2015$).DD The first three values are based on the average SC-CO2
from the three IAMs, at discount rates of 5, 3, and 2.5 percent, respectively. SC-CO2 estimates
for several discount rates are included because the literature shows that the SC-CO2 is quite
sensitive to assumptions about the discount rate, and because no consensus exists on the
appropriate rate to use in an intergenerational context (where costs and benefits are incurred by
different generations). The fourth value is the 95th percentile of the SC-CO2 from all three
models at a 3 percent discount rate. It is included to represent lower probability but higher -
impact outcomes from climate change, which are captured further out in the tail of the SC-CO2
distribution, and while less likely than those reflected by the average SC-CO2 estimates, would
be much more harmful to society and therefore, are relevant to policy makers.
The current estimates, which are the same as those used in the Draft TAR, are higher than
those used to analyze the CO2 impacts in the final LD 2017-2025 rulemaking, which preceded
the 2013 SC-CO2 update and were published in the 2010 SC-CO2 TSD. By way of comparison,
the four SC-CO2 estimates used to analyze the CO2 impacts for the final LD 2017-2015
rulemaking were $8.3, $31, $49, and $96 per metric ton in 2022 (2015$).EE As previously noted,
cc National Academies of Sciences, Engineering, and Medicine. (2016). Assessment of Approaches to Updating the
Social Cost of Carbon: Phase 1 Report on a Near-Term Update. Committee on Assessing Approaches to Updating
the Social Cost of Carbon, Board on Environmental Change and Society. Washington, DC: The National
Academies Press, doi: 10.17226/21898. See Executive Summary, page 1, for quoted text.
DD The current version of the TSD is available at: https://www.whitehouse.gov/sites/default/files/omb/inforeg/scc-
tsd-final-july-2015.pdf. All of the SC-CO2 TSDs present SC-C02 in 2007$ per metric ton. The unrounded
estimates from the current TSD were adjusted to 2015$ using GDP Implicit Price Deflator (1.130),
http://www.bea.gov/iTable/index_nipa. The estimates presented in this document were rounded to two significant
digits.
EE The SC-C02 TSDs present SC-C02 in 2007$; see https://www.whitehouse.gov/omb/oira/social-cost-of-carbon
for both TSDs. The estimates used in the final 2017-2025 rulemaking were adjusted to 2010$ using GDP Implicit
Price Deflator. The estimates presented in the Draft TAR were in 2013$. The estimates presented in the Proposed
Determination have not changed since the Draft TAR but were adjusted to 2015$ for consistency with the rest of
3-44
-------
Economic and Other Key Inputs Used in EPA's Analyses
the IWG updated these estimates in 2013 using new versions of each integrated assessment
model but did not revisit the modeling decisions. Table 3.14 presents the current global SC-CO2
estimates for select years between 2022 and 2050. In order to calculate the dollar value for
emission reductions, the SC-CO2 estimate for each emissions year would be applied to changes
in CO2 emissions for that year, and then discounted back to the analysis year using the same
discount rate used to estimate the SC-CO2. The SC-CO2 increases over time because future
emissions are expected to produce larger incremental damages as physical and economic systems
become more stressed in response to greater climate change. Note that the interagency group
estimated the growth rate of the SC-CO2 directly using the three integrated assessment models
rather than assuming a constant annual growth rate. This helps to ensure that the estimates are
internally consistent with other modeling assumptions. Appendix Section C of the Proposed
Determination document reports the updated GHG benefits in select model years and calendar
years.
Table 3.14 Social Cost of CO2,2022-2050 (in 2015$ per metric ton)*
Year
Discount Rate and Statistic
5% Average 3% Average
2.5% Average
High Impact (3% at 95th
percentile)
2022
$15
$49
$72
$150
2023
$15
$50
$73
$150
2024
$15
$51
$75
$150
2025
$16
$52
$77
$160
2030
$18
$57
$82
$170
2040
$24
$68
$95
$210
2050
$29
$78
$110
$240
Note:
* These SC-CO2 values are stated in $/metric ton and rounded to two significant figures. The estimates vary
depending on the year of CO2 emissions and are defined in real terms, i.e., adjusted for inflation using the GDP
implicit price deflator.
One limitation of the primary benefits analysis in the 2017-2025 final rulemaking is that it did
not include the valuation of non-C02 GHG impacts (CH4, N2O, HFC-134a). Specifically, the
IWG did not estimate the social costs of non-C02 GHG emissions using an approach analogous
to the one used to estimate the SC-CO2. While there were other estimates of the social cost of
non-C02 GHGs in the peer review literature, the methodologies underlying those estimates were
inconsistent with the methodology the IWG used to estimate the SC-CO2. As discussed in the
2017-2025 final rulemaking, there is considerable variation among these published estimates in
the models and input assumptions they employ. ^ These studies differ in the emission
perturbation year, employ a wide range of constant and variable discount rate specifications, and
consider a range of baseline socioeconomic and emissions scenarios that have been developed
the Proposed Determination. The unrounded estimates from the current TSD were adjusted to 2015$ using GDP
Implicit Price Deflator (1.130), http://www.bea.gov/iTable/index_nipa. The estimates presented in this document
were rounded to two significant digits.
FF The researchers cited in the 2017-2015 RIA include: Fankhauser (1994); Kandlikar (1995); Hammitt et al. (1996);
Tol et al. (2003); Tol (2004); and Hope and Newberry (2006).
3-45
-------
Economic and Other Key Inputs Used in EPA's Analyses
over the last 20 years. EPA also determined that the estimates in the literature were most likely
underestimates due to changes in the underlying science since their publication.00
However, EPA recognized that non-CC>2 GHG impacts associated with these standards (e.g.,
net reductions in CH4, N2O, and HFC-134a) would provide benefits to society. To understand
the potential implication of omitting these benefits, EPA conducted sensitivity analysis using an
approximation approach based on global warming potential (GWP) gas comparison metrics that
has been used in previous rulemakings. The EPA also sought public comments on the valuation
of non-CCh GHG impacts in the proposed LD 2017-2025 rulemaking and other previous
rulemakings (e.g., U.S. EPA 2012).96 In general, the commenters strongly encouraged the EPA
to incorporate the monetized value of non-CC>2 GHG impacts into the benefit cost analysis,
however they noted the challenges associated with the GWP-approach, as discussed further
below, and encouraged the use of directly-modeled estimates of the SC-CH4 to overcome those
challenges.
In August 2016, the IWG issued an Addendum to the current TSD that presents estimates of
the SC-CH4 and SC-N2O for use in regulatory impact analysis ("IWG non-CCh Addendum").97
The IWG's SC-CH4 and SC-N2O estimates are taken from a paper by Marten et al. (2014),
which provided the first set of published SC-CH4 and SC-N2O estimates that are consistent with
the modeling assumptions underlying the SC-CO2.98 Specifically, the estimation approach of
Marten et al. used the same set of three IAMs, five socioeconomic and emissions scenarios,
equilibrium climate sensitivity distribution, three constant discount rates, and aggregation
approach used by the IWG to develop the SC-CO2 estimates. The aggregation method involved
distilling the 45 distributions of the SC-CH4 and of the SC-N2O produced for each emissions
year into four estimates: the mean across all models and scenarios using a 2.5 percent, 3 percent,
and 5 percent discount rate, and the 95th percentile of the pooled estimates from all models and
scenarios using a 3 percent discount rate. Marten et al. also used the same rationale as the IWG
to develop global estimates of the SC-CH4 and SC-N2O, given that methane and N2O are global
pollutants.
The IWG non-CC>2 Addendum discusses the basis for atmospheric lifetime and radiative
efficacy of methane and N2O used by Marten et. al. Specifically, Marten et al. based atmospheric
lifetime and radiative efficacy on the estimates reported by the IPCC in their Fourth Assessment
Report (AR4, 2007), including an adjustment in the radiative efficacy of methane to account for
its role as a precursor for tropospheric ozone and stratospheric water. These values represent the
same ones used by the IPCC in AR4 for calculating GWPs. At the time Marten et al. developed
their estimates of the SC-CH4, AR4 was the latest assessment report by the IPCC. The IPCC
updates GWP estimates with each new assessment, and in the most recent assessment, AR5, the
latest estimate of the methane GWP ranged from 28-36, compared to a GWP of 25 in AR4. The
updated values reflect a number of changes: changes in the lifetime and radiative efficiency
estimates for CO2, changes in the lifetime estimate for methane, and changes in the correction
factor applied to methane's GWP to reflect the effect of methane emissions on other climatically
important substances such as tropospheric ozone and stratospheric water vapor. In addition, the
GG See the 2017-2025 RIA, page 7-7, for complete discussion. Literature included studies primarily from the mid-
1990s through early 2000s. https://nepis.epa.gov/Exe/ZyPDF.cgi/P100EZIl.PDF?Dockey=P100EZIl.PDF.
3-46
-------
Economic and Other Key Inputs Used in EPA's Analyses
range presented in the latest IPCC report reflects different choices regarding whether to account
for how biogenic and fossil methane have different carbon cycle effects, and for whether to
account for climate feedbacks on the carbon cycle for both methane and CO2 (rather than just for
CO2 as was done in AR4)."'HH
The IWGnon-C02 Addendum discusses the SC-CH4 and SC-N2O estimates, (presented
below in Table 3.15), and compare them with other recent estimates in the literature. A direct
comparison of the estimates with all of the other published estimates is difficult, given the
differences in the models and socioeconomic and emissions scenarios, but results from three
relatively recent studies offer a better basis for comparison (see Hope (2006), Marten and
Newbold (2012), Waldhoff et al. (2014)). Marten et al. found that, in general, the SC-CH4
estimates from their 2014 paper are higher than previous estimates and the SC-N2O estimates
from their 2014 paper fall within the range from Waldhoff et al. The higher SC-CH4 estimates
are partially driven by the higher effective radiative forcing due to the inclusion of indirect
effects from methane emissions in their modeling. Marten et al., similar to other recent studies,
also find that their directly modeled SC-CH4 and SC-N2O estimates are higher than the GWP-
weighted estimates. More detailed discussion of the SC-CH4 and SC-N2O estimation
methodology, results and a comparison to other published estimates can be found in Marten et al.
(2014).
The resulting SC-CH4 and SC-N2O estimates are presented in Table 3.15. The tables do not
include HFC-134a because EPA is unaware of analogous estimates.
Table 3.15 Social Cost of CH4 and Social Cost of N20,2012-2050 (in 2015$ per metric ton)
Social Cost of CH4
Social Cost of N20
Year
5% (Avg)
3% (Avg)
2.5% (Avg)
High
Impact (3%
at 95th
percentile)
5%
(Avg)
3% (Avg)
2.5% (Avg)
High
Impact
(3% at
95th
percentile)
2022
$660
$1,400
$1,900
$3,800
$5,700
$18,000
$26,000
$46,000
2023
$680
$1,500
$1,900
$4,000
$5,900
$18,000
$26,000
$47,000
2024
$710
$1,500
$2,000
$4,100
$6,000
$19,000
$27,000
$49,000
2025
$730
$1,600
$2,000
$4,200
$6,200
$19,000
$27,000
$50,000
2030
$860
$1,800
$2,300
$4,700
$7,100
$21,000
$31,000
$55,000
2040
$1,100
$2,300
$2,900
$6,200
$9,500
$26,000
$36,000
$68,000
2050
$1,500
$2,800
$3,500
$7,600
$12,000
$31,000
$42,000
$81,000
Note:
* These SC-CH4 and SC-N2O values are stated in $/metric ton and rounded to two significant figures. The
estimates vary depending on the year of emissions and are defined in real terms, i.e., adjusted for inflation using the
GDP implicit price deflator. In addition, the estimates in this table have been adjusted to reflect the minor technical
1111 Consistent with the Draft TAR, the Proposed Determination uses 100-year GWP values for CO2 equivalency
calculations that are consistent with the GHG emissions inventories and the IPCC Fourth Assessment Report
(AR4), i.e., 25 for methane. The IPCC reported the same 100-year GWP for N2O (298) in AR4 and AR5.
3-47
-------
Economic and Other Key Inputs Used in EPA's Analyses
corrections to the SC-CO2 estimates described above. See Corrigendum to Marten et al. (2014) for more
details http://www.tandfoniine.com/doi/abs/10..1..080/.1..'4693062.20.1.5..1.070550 .
This Proposed Determination analysis updates the non-CC>2 GHG benefits presented in the
2017-2025 final rule by using the IWG's estimates of SC-CH4 and SC-N2O.11 As discussed in
the IWG non-CC>2 Addendum, the application of directly modeled estimates from Marten et al.
(2014) to benefit-cost analysis of a regulatory action is analogous to the use of the SC-CO2
estimates. Specifically, the SC-CH4 and SC-N2O estimates in Table 3.15 are used to monetize
the benefits of reductions in methane and N2O emissions, respectively, expected as a result of
the 2022-2025 standards. Forecast changes in methane (or N2O) emissions in a given year,
expected as a result of the standards, are multiplied by the SC-CH4 (or SC-N2O) estimate for that
year. To obtain a present value estimate, the monetized stream of future non-CCh GHG benefits
are discounted back to the analysis year using the same discount rate used to estimate the social
cost of the non-CC>2 GHG emission changes. In addition, the limitations for the SC-CO2
estimates discussed above likewise apply to the SC-CH4 and SC-N2O estimates, given the
consistency in the methodology. See the IWG non-CC>2 Addendum for additional details about
the peer review conducted of the application of the Marten et al. (2014) non-CCh social cost
estimates in regulatory analysis.
The summary of GHG (CO2, methane, N2O) benefits are presented for select model years and
calendar years is in Appendix Section C of the Proposed Determination document.
EPA is unaware of estimates of the social cost of HFC-134a that are analogous to the SC-
CO2, SC-CH4, and SC-N2O estimates discussed above. In the 2017-2025 final rulemaking, EPA
used the GWP for HFC-134a to convert the emissions of this gas to CO2 equivalents, which were
then valued using the SC-CO2 estimates. These estimates were presented in a sensitivity analysis
due to the limitations associated with using the GWP approach to value changes in non-CCh
GHG emissions.
The GWP measures the cumulative radiative forcing from a perturbation of a non-CCh GHG
relative to a perturbation of CO2 over a fixed time horizon, often 100 years. The GWP mainly
reflects differences in the radiative efficiency of gases and differences in their atmospheric
lifetimes. While the GWP is a simple, transparent, and well-established metric for assessing the
relative impacts of non-CC>2 emissions compared to CO2 on a purely physical basis, there are
several well-documented limitations in using it to value non-CC>2 GHG benefits, as discussed in
the 2010 SC-CO2 TSD and previous rulemakings.100 In particular, several recent studies found
that GWP-weighted benefit estimates for methane are likely to be lower than the estimates
derived using directly modeled social cost estimates for these gases. Gas comparison metrics,
such as the GWP, are designed to measure the impact of non-CCh GHG emissions relative to
CO2 at a specific point along the pathway from emissions to monetized damages (depicted in
Figure 3.4), and this point may differ across measures.
11 The IWG SC-CH4 and SC-N20 estimates presented in this TSD are the same as the SC-CH4 and SC-N20
estimates presented in the Draft TAR except they have been adjusted to 2015$ instead of 2013$. The estimates
published in the Draft TAR were labeled as "Marten et al. (2014)" estimates.
3-48
-------
Economic and Other Key Inputs Used in EPA's Analyses
I I'vifuirnptiiul
1" irKMnrts
—~
,t 1 iOSOf- I'f'i
Radiative
Fflrcing
Climate
Impacts
-~
.1 Til JlOCiO
Ecuremic
Monetized
Damages
Figure 3.4 Path from GHG Emissions to Monetized Damages (Source: Marten et al., 2014)
The GWP is not ideally suited for use in benefit-cost analyses to approximate the social cost
of non-CCh GHGs because it ignores important nonlinear relationships beyond radiative forcing
in the chain between emissions and damages. These can become relevant because gases have
different lifetimes and the SC-CO2 takes into account the fact that marginal damages from an
increase in temperature are a function of existing temperature levels. Another limitation of gas
comparison metrics for this purpose is that some environmental and socioeconomic impacts are
not linked to all of the gases under consideration, or radiative forcing for that matter, and will
therefore be incorrectly allocated. For example, the economic impacts associated with increased
agricultural productivity due to higher atmospheric CO2 concentrations included in the SC-CO2
would be incorrectly allocated to methane emissions with the GWP-based valuation approach.
Also of concern is the fact that the assumptions made in estimating the GWP are not
consistent with the assumptions underlying SC-CO2 estimates in general, and the SC-CO2
estimates developed by the IWG more specifically. For example, the 100-year time horizon
usually used in estimating the GWP is less than the approximately 300-year horizon the IWG
used in developing the SC-CO2 estimates. The GWP approach also treats all impacts within the
time horizon equally, independent of the time at which they occur. This is inconsistent with the
role of discounting in economic analysis, which accounts for a basic preference for earlier over
later gains in utility and expectations regarding future levels of economic growth.
The changes in HFC-134a emissions occur through model year 2021, at which point use of
HFC-134a in new vehicles is prohibited under the Significant New Alternatives Policy (SNAP).
As discussed in Chapter 5.2.9.2, EPA expects that HFC-134a will be entirely replaced by
refrigerants with lower GWPs by model year 2021. In other words, there will be no further
reductions in HFC-134a emissions after model year 2021. Given that this Proposed
Determination considers years after 2021, there are no changes in impacts to report for HFC-
134a. See Chapter 2.2.9.2 of this TSD for complete discussion, including EPA's assessment
about the transition to use of low-GWP alternative refrigerants.
3.8 Benefits from Reduced Refueling Time
The total time spent pumping and paying for fuel, and driving to and from fueling stations,
represents an economic cost to drivers and other vehicle occupants. Increased driving range
provides a benefit to individuals arising from the value of the time saved when refueling events
are eliminated. As described in this section, the EPA calculates this benefit by applying DOT-
recommended values of travel time savings to estimates of how much time is saved.
The increases in fuel economy resulting from the standards are expected to lead to some
increase in vehicle driving range. The extent of this increase depends on manufacturers'
decisions to apply reduced fuel consumption requirements towards increasing range, rather than
reducing tank size while maintaining range. For the 2012 FRM, EPA conducted a regression
analysis to identify the relationship between fuel economy and fuel tank size for different vehicle
classes based on historical data. Trends in fuel tank size for a number of redesigned vehicles
3-49
-------
Economic and Other Key Inputs Used in EPA's Analyses
were also investigated. Based on these analyses, fuel economy improvements were assumed to
be entirely realized as improvements in driving range, due to insufficient evidence to indicate
that fuel tank size is reduced as vehicle fuel economy is improved. EPA is using the assumption
from Chapter 10.8 of the Draft TAR that fuel tank sizes remain constant. EPA did not receive
comments on this topic, and we have not seen evidence to suggest that reductions in vehicle tank
size are occurring. Thus we believe that using the Draft TAR values is still appropriate;
however, we will continue to monitor trends in fuel tank designs and vehicle range.
No direct estimates of the value of extended vehicle range or reduced fuel tank size are
readily available. Instead, the EPA analysis calculates the reduction in the annual amount of
time a driver would spend filling its fuel tank; this reduced time could result either from fewer
refueling events, if new fuel tanks stay the same size, or from less time spent filling the tank
during each refueling stop, if new fuel tanks are made proportionately smaller. As discussed in
Section 3.4 above, the average number of miles each type of vehicle is driven annually would
likely increase under the regulation, as drivers respond to lower fuel expenditures (the "rebound
effect"). The estimates of refueling time in effect allow for this increase in vehicle use.
However, the estimate of the rebound effect does not account for any reduction in net operating
costs from lower refueling time. Because the rebound effect should measure the change in VMT
with respect to the net change in overall operating costs, refueling time costs would ideally factor
into this calculation. The effect of this omission is expected to be minor because refueling time
savings are generally small relative to the value of reduced fuel expenditures.
The savings in refueling time are calculated as the total amount of time the driver of a typical
vehicle would save each year as a consequence of pumping less fuel into the vehicle's tank. The
calculation also includes a fixed time per refill event of 3.5 minutes which would not occur as
frequently due to the fewer number of refills.
The calculation uses the reduced number of gallons consumed and divides that value by the
tank volume and refill amount to get the number of refills, then multiplies that by the time per
refill to determine the number of hours saved in a given year. The calculation then applies DOT-
recommended values of travel time savings to convert the resulting time savings to their
economic value. For this analysis, EPA uses the input metrics shown in Table 3.16. The
refueling benefits are presented in Appendix C.3 to the Proposed Determination document.
Table 3.16 Metrics Used in Calculating the Value of Refueling Time
Metric
Value
Average tank refill percentage
65%
Average tank volume
15 gallons
Fuel dispense rate
10 gal/min
Fixed time per refill
3.5 minutes
Wage rate for the value of refill time
$25.72 in 2015$
Number of people in vehicle
1.2
Wage growth rate, 2014 base year
1.1%
3-50
-------
Economic and Other Key Inputs Used in EPA's Analyses
The equation used by EPA to calculate refueling benefits is shown below. This is the same
approach and equation as was used in the Draft TAR.
_ £ n r. (dereference Galpolicy\ ( Gal per Tefill r-n\ f^\
Refueling Benefit = — — x — 1- time per refill x —
V Gal per refill J \Fuel dispense rate J \hr J
labor
3.9 Benefits and Costs from Additional Driving
3.9.1 Travel Benefit
The increase in travel associated with the rebound effect produces additional benefits to
vehicle drivers, which reflect the value of the added (or more desirable) social and economic
opportunities that become accessible with additional travel. The analysis estimates the economic
benefits from increased rebound-effect driving as the sum of fuel expenditures incurred plus the
vehicle owner/operator surplus from the additional accessibility it provides. As evidenced by the
fact that vehicles make more frequent or longer trips when the cost of driving declines, the
benefits from this added travel exceed added expenditures for the fuel consumed. Note that the
amount by which the benefits from this increased driving exceed its increased fuel costs
measures the net benefits from the additional travel, usually referred to as increased consumer
surplus or, in this case, increased driver surplus.
The equation for the calculation of the total travel benefit is shown below. This is the same
approach and equation as was used in the Draft TAR.
Travel Benefit = (VMTrebound) I \ +
\ ' policy
(VMTr(
d)
(—) ~ (—)
\ mile I \ mile I
^ ' reference ^ '
policy.
The analysis estimates the economic value of the increased owner/operator surplus provided
by added driving using the conventional approximation, which is one half of the product of the
decline in vehicle operating costs per vehicle-mile and the resulting increase in the annual
number of miles driven. Because it depends on the extent of improvement in fuel economy, the
value of benefits from increased vehicle use changes by model year and varies among alternative
standards. Under even those alternatives that would impose the highest standards, however, the
magnitude of the surplus from additional vehicle use represents a small fraction of this benefit.
The travel benefits are presented in Appendix C.3 to the Proposed Determination document.
3.9.2 Costs Associated with Crashes, Congestion and Noise
In contrast to the benefits of additional driving are the costs associated with that driving. If net
operating costs of the vehicle decline, then we expect a positive rebound effect. Increased
vehicle use associated with a positive rebound effect also contributes to increased traffic
congestion, motor vehicle crashes, and highway noise. Depending on how the additional travel
is distributed throughout 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 during peak periods. 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.
3-51
-------
Economic and Other Key Inputs Used in EPA's Analyses
EPA relies on estimates of congestion, crash, and noise costs caused light-duty vehicles
developed by the Federal Highway Administration to estimate the increased external costs
caused by added driving due to the rebound effect. The FHWA estimates are intended to
measure the increases in costs from added congestion, property damages and injuries in traffic
crashes, and noise levels caused by various classes of vehicles that are borne by persons other
than their drivers (or "marginal" external costs). EPA employed estimates from this source
previously in the analysis accompanying the light-duty 2012-2016 vehicle rulemaking. We
continue to find them appropriate for this analysis after reviewing the procedures used by FHWA
to develop them and considering other available estimates of these values.
FHWA's congestion cost estimates focus on freeways because non-freeway effects are less
serious due to lower traffic volumes and opportunities to re-route around the congestion. The
agencies, however, applied the congestion cost to the overall VMT increase, though the fraction
of VMT on each road type used in MOVES range from X to Y percent of the vehicle miles on
freeways for light-duty vehicles. The results of this analysis potentially overestimate the
congestions costs associated with increased vehicle use, and thus lead to a conservative estimate
of net benefits.
EPA has used FHWA's "Middle" estimates for marginal congestion, crash, and noise costs
caused by increased travel from vehicles. This approach is consistent with the methodology used
in both LD and HD GHG rules and in the Draft TAR. These costs are multiplied by the annual
increases in vehicle miles travelled from the rebound effect to yield the estimated increases in
congestion, crash, and noise externality costs during each future year. The values used are shown
in Table 3.17. The costs associated with crashes, congestion and noise are presented in Appendix
C.3 to the Proposed Determination document.
Table 3.17 Metrics Used to Calculate the Costs Associated with Congestion, Crashes and Noise Linked to
Rebound Miles Traveled (2015$)
Metric
Value
Congestion
$0.0600 per mile
Crashes
$0.0259 per mile
Noise
$0.0008 per mile
3.10 Discounting Future Benefits and Costs
The benefits and costs are analyzed using 3 percent and 7 percent discount rates, consistent
with current OMB guidance.JJ These rates are intended to represent consumers' preference for
current over future consumption (3 percent), and the real rate of return on private investment (7
percent) which indicates the opportunity cost of capital. However, neither of these rates
"Office of Management and Budget (2003). "Circular A-4." https://www.whitehouse. eov/o tub/circulars a004 a-
4/. Discounting involving the Social Cost of Carbon (SC-CO2) values uses several discount rates because the
literature shows that the SC-CO2 is quite sensitive to assumptions about the discount rate, and because no
consensus exists on the appropriate rate to use in an intergenerational context (where costs and benefits are
incurred by different generations). Refer to Section 10.7 for more information.
3-52
-------
Economic and Other Key Inputs Used in EPA's Analyses
necessarily represents the discount rate that individual decision-makers use, nor do they reflect
the rates in OMB Circular A-94 Appendix C, which are revised annually.IvIv The 2015 Appendix
lists real (i.e., inflation-adjusted) discount rates between 0.3 percent (for a 3-year period) and 1.5
percent (for a 30-year time horizon). All costs and benefits are discounted to 2016 except for
those considered in payback analyses where costs and benefits are discounted to the first year of
a vehicle's life.
3.11 Additional Costs of Vehicle Ownership
The discussion here regarding sales taxes, insurance and financing costs pertains only to our
payback analysis. Here we discuss some of the inputs used for that payback analysis. We present
the results of our payback analysis in Appendix C.2.4 to the Proposed Determination document.
3.11.1 Sales Taxes
When consumers consider their total cost of ownership of a vehicle, or its potential payback,
they may consider the sales taxes they have to pay at the time of purchasing the vehicle. As
these costs are transfer payments, they are not included in the societal costs of the program, but
they are included as one of the increased costs to the consumer for these standards when we
calculate costs that the consumer pays out for vehicle ownership as part of our payback analysis.
In the 2012 FRM, the agencies took the most recent auto sales taxes by state and weighted them
by population by state to determine a national weighted-average sales tax of 5.46 percent.LL We
continue to use that value as we did in the Draft TAR.
3.11.2 Insurance Costs
The agencies considered the standards' impact to consumers' auto insurance expenses over
vehicle lifetimes. More expensive vehicles will require more expensive collision and
comprehensive (e.g., theft) car insurance. The scope of this analysis is to estimate the increased
cost to the consumer for these standards, not the increase in societal costs due to collision and
property damage. The increase in insurance costs was estimated from the average value of
collision plus comprehensive insurance as a proportion of average new vehicle price. Collision
plus comprehensive insurance represent the portion of insurance costs that depend on vehicle
value. In the 2012 FRM, we found that dividing the cost to insure a new vehicle by the average
price of a new vehicle gives the proportion of comprehensive plus collision insurance as 1.86
percent of the price of a vehicle. As vehicles' values decline with vehicle age, comprehensive
and collision insurance premiums likewise decline. We continue to use the same approach in this
analysis as was used in the 2012 FRM and again in the Draft TAR.
^ Office of Management and Budget (2015). "Circular A-94 Appendix C, Revised November 2015."
https://www.whitehouse.gov/omb/circulars_a094/a94_appx-c.
LL See http://www.factorywarrantylist.com/car-tax-by-state.html (first accessed April 5, 2012, last accessed on
November 15, 2016). Note that county, city, and other municipality-specific taxes were excluded from the
weighted averages, as the variation in locality taxes within states, lack of accessible documentation of locality
rates, and lack of availability of weights to apply to locality taxes complicate the ability to reliably analyze the
subject at this level of detail. Localities with relatively high automobile sales taxes may have relatively fewer auto
dealerships, as consumers would endeavor to purchase vehicles in areas with lower locality taxes, therefore
reducing the impact of the exclusion of municipality-specific taxes from this analysis.
3-53
-------
Economic and Other Key Inputs Used in EPA's Analyses
3.11.3 Financing Costs
When purchasing a new car, most consumers either finance the purchase via a loan or lease
the vehicle as opposed to paying for the car in cash. Our payback analysis has considered these
financing costs—the interest rates paid on the new car loan—for 3 different loan periods: 4-year,
5-year and the increasingly common 6-year loan. For those loans, we have used interest rates of
4.25 percent.101 We did not estimate payback periods in the Draft TAR for loan purchased
vehicles.
3-54
-------
Economic and Other Key Inputs Used in EPA's Analyses
REFERENCES
1 Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975 Through 2016,"
Table 10.1, epa.gov/fuel-economy/trends-report, November 2016.
2 Ibid.
3 How Do Motorists' Own Fuel Economy Estimates Compare with Official Government Ratings? A Statistical
Analysis," David Greene et al, University of Tennessee, October 1, 2015.
4 U.S. EPA. MOVES Motor Vehicle Emissions Simulator https://www.epa.gov/moves/moves2014a-latest-version-
motor-vehicle-emission-simulator-moves,Version 2014a. November, 2015.
5 U.S. EPA. Population and Activity of On-road Vehicles in MOVES2014. EPA-420-R-16-003a. March 2016.
6 Winebrake, J.J., Green, E.H., Comer, B., Corbett, J.J., Froman, S., 2012. "Estimating the direct rebound effect for
on-road freight transportation," Energy Policy 48, 252-259.
7 Greene, D.L., Kahn, J.R., Gibson, R.C., 1999, "Fuel economy rebound effect for U.S. household vehicles," The
Energy Journal, 20.
8 Greening, L.A., Greene, D.L., Difiglio, C., 2000, "Energy efficiency and consumption — the rebound effect — a
survey," Energy Policy, 28, 389-401.
9 Sorrell, S. and Dimitropoulos, J., "UKERC Review of Evidence of the Rebound Effect, Technical Report 2:
Econometric Studies," Sussex Energy Group, Working Paper, 2007.
10 Pickrell, D. and Schimek, P., 1999. "Growth in Motor Vehicle Ownership and Use: Evidence from the
Nationwide Personal Transportation Survey," Journal of Transportation and Statistics, vol. 2, no. 1, pp. 1-17. [EPA-
HQ-OAR-2010-0799-00027].
11 Puller, S. and Greening, L., 1999. "Household Adjustment to Gasoline Price Change: An Analysis Using Nine
Years of U.S. Survey Data," Energy Economics 21(l):37-52. [EPA-HQ-OAR-2010-0799-0754],
12 Pickrell, D. and Schimek, P., 1999. "Growth in Motor Vehicle Ownership and Use: Evidence from the
Nationwide Personal Transportation Survey," Journal of Transportation and Statistics, vol. 2, no. 1, pp. 1-17. [EPA-
HQ-OAR-2010-0799-0027].
13 Small, K. and Van Dender, K., 2007a. "Fuel Efficiency and Motor Vehicle Travel: The Declining Rebound
Effect." The Energy Journal, vol. 28, no. 1, pp. 25-51. [EPA-HQ-OAR-2010-0799-0755],
14 Small, K. and Van Dender, K., 2007b. "Long Run Trends in Transport Demand, Fuel Price Elasticities and
Implications of the Oil Outlook for Transport Policy," OECD/ITF Joint Transport Research Centre Discussion
Papers 2007/16, OECD, International Transport Forum. [EPA-HQ-OAR-2010-0799-0756],
15 Hymel, K. M., Small, K. A., and Van Dender, K., "Induced demand and rebound effects in road transport,"
Transportation Research Part B: Methodological, Volume 44, Issue 10, December 2010, Pages 1220-1241, ISSN
0191-2615, DOI: 10.1016/j.trb.2010.02.007. [EPA-HQ-OAR-2010-2010-0799-0758],
16 Greene, David, 2012. "Rebound 2007: Analysis of U.S. light-duty vehicle travel statistics," Energy Policy, vol.
41, pp. 14-28. [EPA-HQ-OAR-2010-0799-0759],
17 Bento, A., Goulder, L., Jacobsen, M. and Haefen, R., 2009, "Distributional and Efficiency Impacts of Increased
U.S. Gasoline Taxes, American Economic Review, 99:3 667-699.
18 Wadud, Z., Graham, D.J., Noland, R.B., 2009. Modelling fuel demand for different socio-economic groups.
Applied Energy 86, 2740-2749.
19Dargay, J.M. and Gately, D., 1997. "The demand for transportation fuels: imperfect price-reversibility?"
Transportation Research Part B 31(1), [EPA-HQ-OAR-2010-0799-076], Sentenac-Chemin, E., 2012, "Is the price
effect on fuel consumption symmetric: Some evidence from an empirical study," Energy Policy, Volume 41, pp. 59-
65 [EPA-HQ-OAR-2010-0799-0762],
20 Gately, D., 1993. "The Imperfect Price-Reversibility of World Oil Demand," The Energy Journal, International
Association for Energy Economics, vol. 14(4), pp. 163-182. [EPA-HQ-OAR-2010-0799-0763],
21 Su, Q., 2012, A quantile regression analysis of the rebound effect: Evidence from the 2009 National Household
Transportation Survey in the United States, Energy Policy 45, pp. 368-377.
22 Linn, J., 2013, "The Rebound Effect of Passenger Vehicles," RFF Discussion Paper, No. 13-19 [EPA-HQ-OAR-
2010-0799-0761],
23 Lui, Y.,Tremblay, J. and Cirillo, C., 2014. "An integrated model for discrete and continuous decisions with
application to vehicle ownership, type and usage," Transportation Research Part A, pp. 315-328.
24 Frondel, M., and Vance, C., 2013. Re-Identifying the Rebound: What about Asymmetry? Energy Journal 34
(4):43-54.
3-55
-------
Economic and Other Key Inputs Used in EPA's Analyses
25 Gillingham, K., 2014, Identifying the Elasticity of Driving: Evidence from a Gasoline Price Shock. Regional
Science & Urban Economics 47 (4): 13-24.
26 Wang, T. and Chen, C., 2014. "Impact of fuel price on vehicle miles traveled (VMT): do the poor respond in the
same way as the rich?" Transportation 41(1): 91-105.
27 Hymel, K. M. and Small, K. A., 2015, "The rebound effect for automobile travel: Asymmetric response to price
changes and novel features of the 2000s," Energy Economics, 49 (2015) 93-103 [EPA-HQ-OAR-2015-0827-0034],
28 West, J., Hoekstra, M., Meer, J., Puller, S., 2015, "Vehicle Miles (Not) Traveled: Why Fuel Economy
Requirements Don't Increase Household Driving" National Bureau of Economic Research (NBER), NBER Working
Paper Series, Working Paper 21194, http://www.nber.org/papers/w21194.
29 De Borger, B. Mulalic, I., and Rouwendal, J.,2016, Measuring the rebound effect with micro data: A first
difference approach, Journal of Environmental Economics and Management, 79, 1-17.
30 Gillingham, K., Rapson, D., Wagner, G., 2016, "The Rebound Effect and Energy Efficiency Policy," Review of
Environmental Economics and Policy, 10 (1), pp. 68 - 88.
31 Barla, P., Lamonde, B., Miranda-Moreno, L. and Boucher, N., 2009. Traveled Distance, Stock and Fuel
Efficiency of Private Vehicles in Canada: Price Elasticities and Rebound Effect. Transportation 36 (4):389-402.
32 EIA Annual Energy Review, various editions. For data 2012-2014, and projected data: EIA Annual Energy
Outlook (AEO) 2016 (Reference Case). See Table 11, file "aeotab_ll.xls."
33 EIA Annual Energy Outlook 2016, Table 11, aeotab ll.xlsx.
34 See EIA Annual Energy Review, various editions. For data 2012-2014, and projected data: EIA Annual Energy
Outlook (AEO) 2016 (Reference Case). See Table 11, file "aeotab_ll.xls."
35 U.S. EIA, 2016. "Canada provides record-high share and amount of U.S. crude oil imports in 2015," April 12,
2016. http://www.eia.gov/todayinenergy/detail.php?id=25772.
36 EIA Annual Energy Outlook 2016, Table 21, yearbyyear.xlsx.
37 Based on data from the CIA Factbook, which combines data from various recent years,
https://www.cia.gov/library/publications/the-world-factbook/rankorder/2242rank.html. Accessed 10.19/2016.
3XIEA 2011 "IEA Response System for Oil Supply Emergencies." [EPA-HQ-OAR-2014-0827-0573].
39 We looked at changes in U.S. crude oil imports and net petroleum products in the AEO 2015 Reference Case in
comparison the Low (i.e., Economic Growth) Demand Case to undertake this analysis. See the spreadsheet "Impact
of Fuel Demand on Imports AEO2015.xlsx." We also considered a paper entitled "Effect of a U.S. Demand
Reduction on Imports and Domestic Supply Levels" by Paul Leiby, 4/16/2013. This paper suggests that "Given a
particular reduction in oil demand stemming from a policy or significant technology change, the fraction of oil use
savings that shows up as reduced U.S. imports, rather than reduced U.S. supply, is actually quite close to 90 percent,
and probably close to 95 percent." [EPA-HQ-OAR-2014-0827-0572] [EPA-HQ-OAR-2014-0827-0574].
40 Leiby, Paul N., Donald W. Jones, T. Randall Curlee, and Russell Lee, 1997, Oil Imports: An Assessment of
Benefits and Costs, ORNL-6851, Oak Ridge National Laboratory, [EPA-HQ-OAR-2014-0827-0594],
41 Leiby, P., Factors Influencing Estimate of Energy Security Premium for the GHG Program, Oak Ridge National
Laboratory. [EPA-HQ-OAR-2014-0827-0595],
42 Brown, S. and Huntington, H., 2013, Assessing the U.S. Oil Security Premium, Energy Economics, vol. 38, pp.
118 - 127. [EPA-HQ-OAR-2014-0827-0571],
43 Brown, S. and Huntington, H., 2010, Reassessing the Oil Security Premium. RFF Discussion Paper Series, (RFF
DP 10-05). doi: RFF DP 10-05 [EPA-HQ-OAR-2014-0827-0602],
44 Greene, D., 2010, Measuring energy security: Can the United States achieve oil independence? Energy Policy,
38(4), 1614-1621. EPA-HQ-OAR-2014-0827-0576],
45 Toman, M., 1993, The economics of energy security: theory, evidence and policy, Chapter 25, Handbook of
Natural Resource and Energy Economics, Volume 3, pp. 1167-1218.
46 Ledyard, J. O. "Market Failure." The New Palgrave Dictionary of Economics. Second Edition. Eds. Steven N.
Durlauf and Lawrence E. Blume. Palgrave Macmillan, 2008. [EPA-HQ-OAR-2014-0827-0596],
47 Sivaram, Varan; Levi, Michael A., 2015, "Automobile Fuel Standards in Lower-Oil-Price World," Council on
Foreign Relations.
48 National Academy of Sciences, 2015, "Cost, Effectiveness and Deployment of Fuel Economy Technologies for
Light-Duty Vehicles," Committee on the Assessment of Technologies for Improving Fuel, Economy of Light-Duty
Vehicles, Phase 2; Board on Energy and Environmental Systems; Division on Engineering and Physical Sciences;
National Research Council.
49 Gately, D., 2004, "OPEC's Incentives for Faster Output Growth," The Energy Journal, 25 (2):75-96; Gately, D.,
2007. "What Oil Export Levels Should We Expect From OPEC?" The Energy Journal, 28(2): 151-173. [EPA-HQ-
OAR-2014-0827-0599],
3-56
-------
Economic and Other Key Inputs Used in EPA's Analyses
50 Drabik, D. and de Gorter, H., 2011, "Biofuel Policies and Carbon Leakage", AgBioForum, 14(3): 104-110
51 Rajagopal, D., Hochman, G. and Zilberman, D., 2011, "Indirect fuel use policies (IFUC) and the lifecycle
environmental impact of biofuel policies", Energy Policy, 39: 228-233.
52 Thompson, W., Whistance, J. and Meyer, S., 2011, "Effects of U.S. biofuel policies on U.S. and world petroleum
product markets and consequences for greenhouse gas emissions", Energy Policy, 39: 5509-5518.
53 Erickson, P. and Lazarus, M., 2014, "Impact of Keystone Pipeline on global oil markets and greenhouse gas
emissions", Stockholm Environmental Institute.
54 Karplus, V., Kishimoto, P. and Paltsev, S., 2015, "The Global Energy, CO2 Emissions and Economic Impacts of
Vehicle Fuel Economy Standards", Journal of Transport Economics and Policy, 49(4): 517-538.
55 Historical data are fromEIA Annual Energy Review, various editions. For data since 2012 and projected data:
source is EIA Annual Energy Outlook (AEO) 2015 (Reference Case). See Table 11, file "aeotab ll.xlsx" and Table
20 (Macroeconomic Indicators," (file "aeotab_20.xlsx"). [EPA-HQ-OAR-2014-0827-0619],
56 Bohi, D. and Montgomery, D., 1982. Social Cost of Imported and U.S. Import Policy, Annual Review of Energy,
7:37-60. Energy Modeling Forum, 1981. World Oil, EMF Report 6 (Stanford University Press: Stanford 39 CA.
https//emf.stanford.edu/publications/emf-6-world-oil.
57 Plummer, J. (Ed.), 1982. Energy Vulnerability, "Basic Concepts, Assumptions and Numerical Results," pp. 13 -
36, (Cambridge MA: Ballinger Publishing Co.). [EPA-HQ-OAR-2014-0827-0600],
58 Bohi, D., and Montgomery, D., 1982, Social Cost of Imported and U.S. Import Policy, Annual Review of Energy,
7:37-60.
59 Hogan, W., 1981, "Import Management and Oil Emergencies," Chapter 9 in Deese, 5 David and Joseph Nye, eds.
Energy and Security. Cambridge, MA: Ballinger Publishing Co. [EPA-HQ-OAR-2014-0827-0578],
60 Broadman, H. G. 1986, "The Social Cost of Imported Oil," Energy Policy 14(3):242-252. Broadman H. and W.
Hogan, 1988. "Is an Oil Import Tariff Justified? An American Debate: The Numbers Say 'Yes'." The Energy
Journal 9: 7-29. [EPA-HQ-OAR-2014-0827-0569] [EPA-HQ-OAR-2014-0827-0570],
61 Leiby, P., Jones, D., Curlee, R. and Lee, R., Oil Imports: An Assessment of Benefits and Costs, ORNL-6851, Oak
Ridge National Laboratoiy, 1997. [EPA-HQ-OAR-2014-0827-0594],
62 Parry, I. and Darmstadter J., 2004, "The Costs of U.S. Oil Dependency," Resources for the Future, November 17,
2004 (also published as NCEP Technical Appendix Chapter 1: Enhancing Oil Security, the National Commission on
Energy Policy 2004 Ending the Energy Stalemate - A Bipartisan Strategy to Meet America's Energy Challenges).
[EPA-HQ-OAR-2014-0827-0603],
63 National Research Council, 2009, Hidden Costs of Energy: Unpriced Consequences of Energy Production and
Use. National Academy of Science, Washington, DC. [EPA-HQ-OAR-2014-0827-0597],
64 See, William Nordhaus, "Who's Afraid of a Big Bad Oil Shock?" and Blanchard, O. and Gali, J., "The
macroeconomic Effects of Oil price Shocks: Why are the 2000s so different from the 1970s?," pp. 373-421, in The
International Dimensions of Monetary Policy. Gali, J., and Gertler, M., editors, University of Chicago Press,
February 2010, available at http://www.nber.ore/chapters/c()517.pdf [EPA-HQ-OAR-2014-0827-0567],
65 Blanchard, O. and Gali, J., pp. 414. [EPA-HQ-iDAR-2014-0827-0568],
66 See, Oil Price Drops on Oversupply, http://www.oil-price.net/eri/articles/oil-price-drops-on-oversupplv.php.
10/6/2014. [EPA-HQ-OAR-2014-0827-0566],
67 Hamilton, J. D., 2012, Oil Prices, Exhaustible Resources, and Economic Growth. In Handbook of Energy and
Climate Change. Retrieved from http://ecoiiweb.ucal.edii/~ihamilto/tendbook climate.pdf [EPA-HQ-OAR-2014-
0827-0577],
68 Ramey, V. and Vine, D., 2010, "Oil, Automobiles, and the U.S. Economy: How Much have Things Really
Changed?" National Bureau of Economic Research Working Papers, WP 16067. Retrieved from
http://www.nber.org/fapers/wl6067.pdf [EPA-HQ-OAR-2014-0827-0601],
69 Baumeister, C., Peersman, G., VanRobays, I., 2010, "The Economic Consequences of Oil Shocks: Differences
across Countries and Time", Workshop and Conference on Inflation Challenges in the Era of Relative Price Shocks.
70 Kilian, L., Vigfusson, R., 2014, "The Role of Oil Price Shocks in Causing U.S. Recessions", Board of Governors
of the Federal Reserve System. International Finance Discussion Papers.
71 Cashin, P., Mohaddes, K., and Raissi, M., 2014, "The differential effects of oil demand and supply shocks on the
global economy." Energy Economics.
72 Batovic, A., 2015, Low oil prices fuel political and economic instability. Global Risk Insights, 18-19. Retrieved
from http://globalriskinsights.com/2015A)9/low-oil-prices-fuel-political-and-economic-instabilitv/.
73 Monaldi, F., 2015, The Impact of the Decline in Oil Prices on the Economics, Politics and Oil Industry of
Venezuela. Columbia Center on Global Energy Policy Discussion Papers, (September). Retrieved from
3-57
-------
Economic and Other Key Inputs Used in EPA's Analyses
http://energypolicy.columbia.edu/sites/default/files/energy/Impact of the Decline in Oil Prices on Venezuela,
September 2015.pdf.
74 Even, S., & Guzansky, Y., 2015, Falling oil prices and Saudi stability - Opinion. Jerusalem Post, (September 30).
Retrieved from http://www.jpost.com/Opinion/Falling-oil-prices-and-Saudi-stability-419534.
75 International Monetary Fund (IMF), 2015, IMF Regional Economic Outlook - Middle East and Central Asia.
Regional Economic Outlook (Vol. 33). Tomkiw, L., 2015, Oil Rich Saudi Arabia Running Out Of Assets? IMF
Report Says It's Possible In Next 5 Years. International Business Times, October 21, 19-22. Retrieved from
http://www.ibtimes.com/oil-rich-saudi-arabia-running-out-assets-imf-report-says-its-possible-next-5-years-215017.
76 Crane, K., Goldthau, A., Toman, M., Light, T., Johnson, S., Nader, A., Rabasa, A. and Dogo, H., Imported oil and
U.S. national security. RAND Corporation, 2009, http://www.stormingmedia.us/62/6279/A627994.pdf.
77 Koplow, D., and Martin, A., 1998, Fueling Global Warming: Federal Subsidies to Oil in the United States.
Greenpeace, Washington, DC.
78 Crane et al., 2009, "Imported Oil and U.S. National Security", RAND Corporation.
79 Moore, J., Behrens, C., and Blodgett, J., "Oil Imports: An Overview and Update of Economic and Security
Effects." CRS Environment and Natural Resources Policy Division report 98, no. 1 (1997): 1-14.
80 Copulos, M. "America's Achilles Heel: The Hidden Costs of Imported Oil." Alexandria VA: The National
Defense Council Foundation, September (2003): 1-153. Copulos, M., "The Hidden Cost of Imported Oil~An
Update." The National Defense Council Foundation (2007). Delucchi, Mark A. and James J. Murphy. "US military
expenditures to protect the use of Persian Gulf oil for motor vehicles." Energy Policy 36, no. 6, 2008: 2253-2264,
Crane et al. /RAND, as above, Stern, Roger J. "United States cost of military force projection in the Persian Gulf,
1976-2007." Energy Policy 38, no. 6 (June 2010): 2816-2825.
http://linkinghub.elsevier.com/retrieve/pii/S0301421510000194.
81 "Transitions to Alternative Vehicles and Fuels," Committee on Transitions to Alternative Vehicles and Fuels,
National Research Council, 2013.
82 Fann, N., Baker, K.R., and Fulcher, C.M. (2012). Characterizing the PM2.5-related health benefits of emission
reductions for 17 area and mobile emission sectors across the U.S., Environment International, 49, 241-151,
published online September 28, 2012.
83 U.S. Environmental Protection Agency (U.S. EPA). (2016). Greenhouse Gas Emissions and Fuel Efficiency
Standards for Medium- and Heavy-Duty Engines and Vehicles - Phase 2: Regulatory Impact Analysis, Assessment
and Standards Division, Office of Transportation and Air Quality, EPA-420-R-16-900, August 2016.
84 U.S. Environmental Protection Agency (U.S. EPA). (2012/ Regulatory Impact Analysis: Final Rulemaking for
2017-2025 Light-Duty Vehicle Greenhouse Gas Emission Standards and Corporate Average Fuel Economy
Standards, Assessment and Standards Division, Office of Transportation and Air Quality, EPA-420-R-12-016,
August 2012. Available on the Internet at: http://www.epa.gov/otaq/climate/documents/420rl2016.pdf.
85 U.S. Environmental Protection Agency (U.S. EPA). (2013). Regulatory Impact Analysis for the Reconsideration
of the Existing Stationary Compression Ignition (CI) Engines NESHAP, Office of Air Quality Planning and
Standards, Research Triangle Park, NC. January. EPA-452/R-13-001. Available at
http://www.epa.gov/ttnecasl/regdata/RIAs/RICE NESHAPreconsideration Compression Ignition Engines RIA fi
na!2013 FPA.ndf.
86 U.S. Environmental Protection Agency (U.S. EPA). (2013). Regulatory Impact Analysis for Reconsideration of
Existing Stationary Spark Ignition (SI) RICE NESHAP, Office of Air Quality Planning and Standards, Research
Triangle Park, NC. January. EPA-452/R-13-002. Available at
http://www.epa.gov/ttnecasl/regdata/RIAs/NESHAP RICE Spark Ignition RIA finalreconsideration2013 EPA.p
df.
87 U.S. Environmental Protection Agency (U.S. EPA). (2015). Regulatory Impact Analysis for Residential Wood
Heaters NSPSRevision. Office of Air Quality Planning and Standards, Research Triangle Park, NC. February.
EPA-452/R-15-001. Available at http://www2.epa.gov/sites/production/files/2015-02/documents/201502Q4-
residential-wood-heaters-ria.pdf.
88 U.S. Environmental Protection Agency. (2012). Regulatory Impact Analysis for the Final Revisions to the
National Ambient Air Quality Standards for Particulate Matter, Health and Environmental Impacts Division, Office
of Air Quality Planning and Standards, EPA-452-R-12-005, December 2012. Available on the internet:
http://www.epa.gov/ttnecasl/regdata/RIAs/finalria.pdf.
89 Fann, N., Baker, K.R., and Fulcher, C.M. (2012). Characterizing the PM2.5-related health benefits of emission
reductions for 17 industrial, area and mobile emission sectors across the U.S., Environment International, 49, 241-
151, published online September 28, 2012.
3-58
-------
Economic and Other Key Inputs Used in EPA's Analyses
90 U.S. Environmental Protection Agency (U.S. EPA). (2009). Integrated Science Assessment for Particulate Matter
(Final Report). EPA-600-R-08-139F. National Center for Environmental Assessment—RTP Division. December.
Available at http://cfbub.epa.gov/ncea/cfm/recordisplav.cfm?deid=216546.
91 U.S. Environmental Protection Agency. (2012). Regulatory Impact Analysis for the Final Revisions to the
National Ambient Air Quality Standards for Particulate Matter, Health and Environmental Impacts Division, Office
of Air Quality Planning and Standards, EPA-452-R-12-005, December 2012. Available on the internet:
http://www.epa.gov/ttnecasl/regdata/RIAs/finalria.pdf.
92 Technical Support Document: Technical Update of the Social Cost of Carbon for Regulatory Impact Analysis
Under Executive Order 12866, Interagency Working Group on Social Cost of Greenhouse Gases, with participation
by Council of Economic Advisers, Council on Environmental Quality, Department of Agriculture, Department of
Commerce, Department of Energy, Department of Interior, Department of Transportation, Department of the
Treasury, Environmental Protection Agency, National Economic Council, Office of Management and Budget,
Office of Science and Technology Policy (May 2013, Revised August 2016). Available at: <
https://www.whitehouse.gov/sites/default/files/omb/inforeg/scc_tsd_final_clean_8_26_16.pdf > Accessed
10/20/2016.
93 See https://www.whitehouse.gov/omb/oira/social-cost-of-carbon for both TSDs.
94 See July 2015 Response to Comments document on the SCC,
https://www.whitehouse.gov/sites/default/files/omb/inforeg/scc-response-to-comments-final-july-2015.pdf. See also
(1) Endangerment and Cause or Contribute Findings for Greenhouse Gases under Section 202(a) of the Clean Air
Act, 74 Fed. Reg. 66,496, 66,535 (Dec. 15, 2009) and (2) National Research Council. 2013. Climate and Social
Stress: Implications for Security Analysis. National Academies Press. Washington, DC.
95 Clean Power Plan final rule, see 80 FR 64661; 10/23/15.See also Clean Power Plan Response to Comments,
Section 8.7.2, Document ID EPA-HQ-OAR-2013-0602-37106.
96 U.S. Environmental Protection Agency (U.S. EPA). 2012. Regulatory Impact Analysis Final New Source
Performance Standards and Amendments to the National Emissions Standards for Hazardous Air Pollutants for the
Oil and Natural Gas Industry. Office of Air Quality Planning and Standards, Health and Environmental Impacts
Division. April. < http://www.epa.gov/ttn/ecas/regdata/RIAs/oil_natural_gas_final_neshap_nsps_ria.pdf>.
97 Addendum to Technical Support Document on Social Cost of Carbon for Regulatory Impact Analysis under :
Executive Order 12866, Interagency Working Group on Social Cost of Greenhouse Gases, with participation by
Council of Economic Advisers, Council on Environmental Quality, Department of Agriculture, Department of
Commerce, Department of Energy, Department of Interior, Department of Transportation, Department of the
Treasury, Environmental Protection Agency, National Economic Council, Office of Management and Budget,
Office of Science and Technology Policy (August 2016). Available at: <
https://www.whitehouse.gov/sites/default/files/omb/inforeg/august_2016_sc_ch4_sc_n2o_addendum_final_8_26_l
6.pdf~> Accessed 10/20/2016.
98 Marten, A. L., E. A. Kopits, C. W. Griffiths, S. C. Newbold & A. Wolverton (2014, online publication; 2015, print
publication). Incremental CH4 and N20 mitigation benefits consistent with the U.S. Government's SC-C02
estimates, Climate Policy, DOI: 10.1080/14693062.2014.912981.
99 Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change [Stacker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K.
Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge,
United Kingdom and New York, NY, USA.
i°°For example, see (1) U.S. Environmental Protection Agency (U.S. EPA). (2012). Regulatory Impact Analysis:
Final Rulemaking for 2017-2025 Light-Duty Vehicle Greenhouse Gas Emission Standards and Corporate Average
Fuel Economy Standards, Assessment and Standards Division, Office of Transportation and Air Quality, EPA-420-
R-12-016, August 2012. Available on the Internet at: http://www.epa.gov/otaq/climate/documents/420rl2016.pdf.
and (2) U.S. Environmental Protection Agency (U.S. EPA). 2012. Regulatory Impact Analysis Final New Source
Performance Standards and Amendments to the National Emissions Standards for Hazardous Air Pollutants for the
Oil and Natural Gas Industry. Office of Air Quality Planning and Standards, Health and Environmental Impacts
Division. April. < http://www.epa.gov/ttn/ecas/regdata/RIAs/oil_natural_gas_final_neshap_nsps_ria.pdf>.
101 See "Auto Loan Rates for Use in the OMEGA ICBT," memorandum from Todd Sherwood, Assessment and
Standards Division, Office of Transportation and Air Quality, dated November 15, 2016.
3-59
-------
Consumer Issues
Table of Contents
Chapter 4: Consumer Issues 4-1
4.1 Potential Existence of Tradeoffs between Fuel Economy and Other Vehicle Attributes
4-1
4.1.1 The Reference Case 4-1
4.1.2 Recent Studies of the Engineering Tradeoffs between Power and Fuel Economy, and
Increases in Innovation 4-4
4.1.3 The Role of the Standards in Promoting Innovation 4-7
4.1.4 Potential Ancillary Benefits of GHG-Reducing Technologies 4-10
4.1.5 Estimating Potential Opportunity Costs and Ancillary Benefits 4-12
4.2 Consumer Response to Vehicles Subject to the Standards 4-16
4.2.1 Impact of the Standards on Vehicle Sales 4-16
4.2.2 Evaluations of the Vehicles Subject to the Standards by Professional Auto
Reviewers 4-20
4.3 Impacts of the Standards on Vehicle Affordability 4-38
4.3.1 Literature Review: Definitions of Affordability 4-38
4.3.2 Relating Affordability Themes to Vehicle Standards 4-43
4.3.3 EPA's Assessment of the Impacts of the Standards on Affordability 4-43
4.3.3.1 Data: Consumer Expenditure Survey 4-44
4.3.3.2 Effects on Lower-Income Households 4-46
4.3.3.3 Effect of the Standards on the Used Vehicle Market 4-48
4.3.3.4 Effects on Access to Credit 4-50
4.3.3.5 Effects on Low-Priced Vehicles 4-52
4.3.4 Conclusion 4-55
Table of Figures
Figure 4.1 Observation Count by Groups of Attributes 4-14
Figure 4.2 Reviews of Active Air Dam by Vehicle Make 4-25
Figure 4.3 Reviews of Active Grill Shutters by Vehicle Make 4-25
Figure 4.4 Reviews of Active Ride Height by Vehicle Make 4-26
Figure 4.5 Reviews of Low Rolling Resistance Tires by Vehicle Make 4-26
Figure 4.6 Reviews of Electronic Power Steering by Vehicle Make 4-27
Figure 4.7 Reviews of Turbocharged by Vehicle Make 4-27
Figure 4.8 Reviews of Gasoline Direct Injection by Vehicle Make 4-28
Figure 4.9 Reviews of Cylinder Deactivation by Vehicle Make 4-28
Figure 4.10 Reviews of Diesel by Vehicle Make 4-29
Figure 4.11 Reviews of Hybrid by Vehicle Make 4-29
Figure 4.12 Reviews of Plug-In Hybrid Electric by Vehicle Make 4-30
Figure 4.13 Reviews of Full Electric by Vehicle Make 4-30
Figure 4.14 Reviews of Stop-Start by Vehicle Make 4-31
Figure 4.15 Reviews of High Speed Automatic by Vehicle Make 4-31
Figure 4.16 Reviews of Continuously Variable Transmission by Vehicle Make 4-32
Figure 4.17 Reviews of Dual-Clutch Transmission by Vehicle Make 4-32
Figure 4.18 Reviews of Electric Assist or Low Drag Brakes by Vehicle Make 4-33
Figure 4.19 Reviews of Lighting-LED by Vehicle Make 4-33
Figure 4.20 Reviews of Mass Reduction by Vehicle Make 4-34
Figure 4.21 Reviews of Passive Aerodynamics by Vehicle Make 4-34
Figure 4.22 Reviews of Fuel Cell by Vehicle Make 4-35
-------
Consumer Issues
Figure 4.23 Median Income and Annual Expenditure on New Vehicles for Lower and Higher Income Households 4-
46
Figure 4.24 Percentage of Lower-Income and Higher-Income Households Buying New and Used Vehicles 4-47
Figure 4.25 Annual Expenditure on Vehicles and Gasoline for Lower-Income Households (A) and Higher-Income
Households (B) 4-48
Figure 4.26 Used and New Vehicle Consumer Price Index, 2015 = 100 (2015$) 4-50
Figure 4.27 Percentage of Households Buying at Least One New Vehicle with Finance who had Debt-to-income
(DTI) Ratio Greater than 36 Percent 4-52
Figure 4.28 Number of <$15,000 (2015$) Vehicle Model Trims Available 4-53
Figure 4.29 Minimum MSRP of All Car Models Available 4-55
Table of Tables
Table 4.1 Willingness to Pay (WTP) Estimates from 52 Studies, 2015$ 4-15
Table 4.2 Auto Review Count by Website 4-21
Table 4.3 Auto Review Count by Make 4-21
Table 4.4 Percent Negative Evaluations of Technologies and Operational Characteristics by Vehicle Make 4-23
Table 4.5 Summary of Statistically Significant Regression Coefficients by Technology 4-37
Table 4.6 Breakdown of Households That Bought at Least One New Vehicle By the Cutoff of DTI Ratio 36%,
2007-2015 4-51
Table 4.7 Breakdown of Households That Bought At Least One New Vehicle by the Cutoff of DTI Ratio 40%,
2007-2015 4-51
Table 4.8 Features of the Nissan Versa over Time, Base Model (Edmund's and Ward's Automotive) 4-54
-------
Consumer Issues
Chapter 4: Consumer Issues
This chapter supplements Section B. 1 of the Appendix to the Proposed Determination
document, which examines consumer acceptance of vehicles subject to the standards. It begins in
Chapter 4.1 with a discussion of the possibility of tradeoffs between fuel economy and other
vehicle attributes, related to the discussion in Appendix Section B.1.4. The key questions include
whether those tradeoffs exist, whether they can be measured if they do exist, and how vehicle
buyers might evaluate those tradeoffs if they exist. Chapter 4.2 supplements the Proposed
Determination document Appendix Sections B.I.2., B.I.3., and B.1.5. with a discussion of a
recent study of the effects of the standards on vehicle sales and employment, and elaboration on
the discussion of whether the technologies used to meet the standards impose "hidden costs" on
vehicle buyers. Finally, Chapter 4.3 provides greater detail on the analysis of vehicle
affordability discussed in the Proposed Determination document Appendix Section B.1.6.
4.1 Potential Existence of Tradeoffs between Fuel Economy and Other Vehicle Attributes
Section B.l of the Appendix to the Proposed Determination document discusses consumer
response to the standards. In particular, it examines concerns over the effects of the standards on
sales, and whether other vehicle attributes, such as power, may be adversely affected by
standards (see especially Section B.1.4). This subchapter discusses issues related to the potential
existence of tradeoffs between fuel economy and other vehicle attributes. We begin with a brief
discussion of the reference case, including the assumption that the fleet's fuel economy will not
increase in the absence of the standards, and then proceed to a discussion of the effects of the
standards on other attributes.
4.1.1 The Reference Case
For this Proposed Determination, EPA is assuming that the MY2022-2025 reference fleet will
have GHG emissions performance equal to that necessary to meet the MY2021 standards (in
effect a "flat" reference fleet). This is consistent with the assumption used in the MY2017-2025
rulemaking, where EPA presented a detailed rationale for assuming that there would be no
decrease in fleetwide GHG emissions performance in the reference case fleet for MY2017-2025
beyond the GHG emissions performance necessary to meet the MY2016 standards.1 Key
elements of the rationale were: 1) projections that gasoline prices would be relatively stable out
to 2025, 2) historical evidence that during periods of stable gasoline prices and fuel economy
standards, the only companies that typically over-complied with fuel economy standards were
those that produced primarily lighter vehicles that inherently over-complied with the older
universal (one size fits all, non-attribute based) fuel economy standards, 3) that under
increasingly stringent footprint-based GHG and fuel economy standards for the five years from
MY2012-2016, it was likely that most major manufacturers would be constrained by the
standards and unlikely to voluntarily over-comply, and 4) if there were individual manufacturer
over-compliance in a reference case scenario, that manufacturer would likely generate credits
that could be sold to other companies, and therefore not lead to fleetwide over-compliance.
EPA believes that the case for a flat GHG reference case fleet is even stronger for the
MY2022-2025 timeframe for the following reasons: 1) gasoline prices are about $1 per gallon
lower today than in October 2012 when the MY2017-2025 final rule was published, 2) AEO
2016 reference case projections for fuel prices in the MY2022-2025 timeframe are relatively
4-1
-------
Consumer Issues
stable and approximately $1 per gallon lower than the AEO 2012 Early Release projections upon
which we relied in the final rulemaking analysis, 3) another five years of increasingly stringent
footprint-based GHG and fuel economy standards under the National Program (i.e., the
MY2022-2025 reference case fleet must meet the MY2021 standards, five years later than the
MY2016 standards which were the basis for the MY2017-2025 reference case fleet) that will
have led to significant commercialization of new technologies, and 4) due to the additional five
years of increasingly stringent standards, credits generated in the MY2022-2025 timeframe are
likely to be even more valuable, and even more likely to be sold, than in the MY2017-2021
timeframe. For all of these reasons, EPA believes that it is very unlikely that there would be any
market-driven decrease in fleetwide GHG emissions performance (i.e., overcompliance) in a
MY2022-2025 reference case fleet.
As discussed in Chapter 1 of this TSD, EPA's reference fleet assumes that, while relative
production volumes will continue to evolve through 2025, all characteristics of individual
vehicle models and configurations, except GHG emissions and fuel economy driven by the
standards, will remain unchanged through 2025. In other words, for purposes of assessing the
regulatory impacts analysis of the MY2022-2025 standards, and for properly accounting for the
cost of the additional technology required to meet those standards, EPA is making the modeling
assumption that added technology will be used to reduce greenhouse gas emission and not to
improve vehicle performance and utility. It is important to note that the cost estimates include the
costs of maintaining those other vehicle attributes, so that there is no reduction in vehicle quality.
EPA used a similar approach in the 2012-2016 and the 2017-2025 rulemakings (see, e.g., 77 FR
at 62840/3), and in the Draft TAR. Nevertheless, it is possible that automakers, in the absence of
these standards, would instead have invested in enhancing vehicle attributes such as power, with
an explicit tradeoff between those enhancements and reducing fuel consumption and GHG
emissions. If manufacturers may have chosen to apply technology to improve vehicle
performance in lieu of efficiency, the standards may result in higher costs than projected in this
analysis. This subchapter provides a discussion of that assumption.
Regarding the general issue of constant vehicle characteristics, the National Research
Council2 in its 2015 report stated that assuming equivalent performance in the fleet "is
equivalent to a reference case with no further technical change in the vehicle market from 2017
to 2025." This, it stated, is inconsistent with past trends, where "the rate of technological
progress in vehicle attributes and efficiency has been strong and continual over the past 30
years." From the 1980s to about 2005, as described in Chapter 3.1.5 of the Draft TAR,
horsepower and weight increased steadily, while fuel economy was either stable or declining.
The NRC suggests developing a reference case that reflects technological progress over time,
and its possible allocation to horsepower and weight, rather than assuming equivalent
performance. Specifically, the NRC recommended:
"Recommendation 10.7: The agencies should consider how to develop a reference case for the
analysis of societal costs and benefits that includes accounting for the potential opportunity costs
of the standards in terms of alternative vehicle attributes forgone."3
The technological progress referred to by the NRC has been an ongoing process in the auto
industry. Several recent studies,4 discussed in Chapter 4.1.2 below, have sought to estimate the
magnitude of innovation by calculating the relationship between power, fuel economy, and
weight each year. Over time, if it is possible to have more fuel economy for a constant amount of
4-2
-------
Consumer Issues
power and weight (or more power or weight for constant fuel economy), those studies define that
increase as innovation. These studies argue that most of that innovation has in the past gone into
improvements in vehicle power. The authors expect that the vehicle GHG and fuel economy
standards are instead directing that innovation toward fuel economy. As a result, because
technological innovation has not been directed toward power, vehicles in the reference case must
be less powerful than they would be in the absence of the standards. Thus, such studies would
suggest that the reference case should be revised to project that power would have been higher; if
vehicles subject to the standards do not achieve that new reference-case level of power, then the
agencies should account for the opportunity cost of the forgone power.
In contrast, a working paper from Cooke5 argues that the reference case should not include
these presumed increases in power or other attributes, because the agencies are not required to do
more than preserve the baseline attributes. Cooke argues that increases in power or other vehicle
attributes are optional to manufacturers, and thus not the responsibility of the agencies. If those
technologies were instead applied to vehicle performance or other attributes rather than fuel
economy, and it then becomes more expensive to meet the standards, Cooke argues that that
increase in costs is properly attributable to a discretionary decision, not to the standards.
EPA also received comments from UCS recommending that EPA create a baseline equivalent
to the 2014 baseline with 2010 MY vehicles, using engineering judgment to assess what
technologies are applied to the vehicle, because updating the baseline to post-2010 MYs will fail
to account for vehicle manufacturers' choices to apply technology to improve vehicle
performance in lieu of improving vehicle efficiency: "Choosing a more recent baseline only
further serves to 'bake in' this inefficient use of technology, ascribing costs that should be borne
by manufacturers as a trade-off instead as a direct cost of regulation." EPA recognizes that, with
each baseline update, some portion of additional technology efficiency is lost to improved
vehicle performance. As a result, our calculated cost of compliance is slightly higher than if
technologies had been applied only to improve efficiency. Also, the creation of a baseline
equivalent to the 2014 model year fleet using 2010 model year vehicles is not possible, in part
because there are many vehicles in the 2014 fleet that do not have replacements available in
MY2010.
EPA expects that manufacturers will continue to consider ways to improve vehicle utility and
performance, and the potential for tradeoffs between reducing GHG emissions and improving
other vehicle attributes warrants continued scrutiny. Comments from Resources for the Future
argue that methods such as those used in the studies discussed in Chapter 4.1.2 could be used to
develop a reference case that would include the potential for improvements over time in vehicle
attributes or other attributes associated with improving fuel economy.A The analysis of the
MY2022-2025 standards would begin with a such a reference case. The cost and effectiveness
analysis would involve adding technologies to those new vehicles, either holding those enhanced
vehicles' characteristics constant or explicitly acknowledging changes in those characteristics to
achieve the standards. In practice, though, estimating these effects and their magnitudes involve
A As discussed in the Guidelines for Preparing Economic Analyses (U.S. Environmental Protection Agency, 2014,
https://vosem.ite.epa.gov/ee/epa/eed.nsf/webpages/Guideiines.htmi. Chapter 5), the baseline (referred to in this
chapter as the reference case) "is defined as the best assessment of the world absent the proposed regulation or
policy action." In other words, the analysis should take into account that change is likely to happen even without
the regulation or action.
4-3
-------
Consumer Issues
a number of complexities, including challenges in estimating the tradeoffs and the innovation
likely to occur in the absence of the standards, the role of the standards in promoting innovation,
and the potential for ancillary benefits associated with GHG-reducing technologies.
The remainder of Chapter 4.1 describes these complexities in more detail. Chapter 4.1.2
focuses on the estimation process mentioned above, for trying to identify expected tradeoffs
between fuel economy, power, and weight, and for the measures of innovation. The magnitudes
of both the tradeoff estimates and the innovation estimates may not yet be known with
confidence. The literature does point to an important aspect of the standards, though: they may
increase the amount of innovation over the reference-case level. Chapter 4.1.3 examines this
question more closely. In particular, it draws on the literature on innovation to distinguish
between "incremental," small-scale innovation, and "major" innovation. It proposes a thesis that
incremental technology is likely to be what would happen in the absence of the standards, while
the standards may trigger major technology. If so, both the benefits and the costs of major
innovation are associated with the standards. If incremental innovation can happen irrespective
of the standards - that is, the benefits and costs of incremental innovation are unaffected by the
standards - then the only tradeoffs important for the standards are those associated with major
innovations. While Chapter 4.2.2 discusses recent EPA research exploring whether there are
possible adverse effects of fuel-saving technologies, Chapter 4.1.4 points out that some of these
technologies have ancillary benefits. Finally, Chapter 4.1.5 discusses how EPA might evaluate
the impact of the standards on other vehicle characteristics in the benefit-cost analysis.
4.1.2 Recent Studies of the Engineering Tradeoffs between Power and Fuel Economy, and
Increases in Innovation
The recent studies6 that estimate both technological improvements over time in the auto
industry, as well as the engineering tradeoffs among fuel economy, power, and weight (and
sometimes other characteristics) have much in common with each other. They all estimate an
equation roughly of the form,
In (fuel economy) = PO + pi*ln(horsepower) + P2*ln(weight) + P4*0ther Characteristics + s,
where:
In refers to the natural logarithm of the term in parentheses,
Ps are coefficients to be estimated in the statistical analysis (and measure elasticities of fuel
economy with respect to its associated variable)
s is an error term
They differ in the additional vehicle characteristics that they include in the regressions, and in
their ways of measuring technological change. Estimates of the elasticities of fuel economy with
respect to horsepower-that is, the engineering tradeoffs between fuel economy and horsepower-
include values from -0.16 (Klier and Linn) to -0.32 (Knittel 2011); the elasticities between fuel
4-4
-------
Consumer Issues
economy and weight include values from -0.336 (Klier and Linn 2016) to -0.521 (MacKenzie
and Heywood 2015).B
Regarding measures of technological change, Knittel (2011) and MacKenzie and Heywood
(2015) use annual shifts in the tradeoff curves; Klier and Linn (2016) use engine redesign cycles
for individual vehicles; and Wang (2016) uses a time trend and the level (stringency) of fuel
economy standards. The papers all find technological innovation, defined as an increase over
time in fuel economy not explained by changes in horsepower, weight, or other characteristics, to
be ongoing. Knittel (2011) finds truck and car efficiency to have increased about 50 percent from
1980 to 2006, with innovation higher before 1990 than in subsequent years. MacKenzie and
Heywood find that efficiency measured using horsepower and weight increased about 50 percent
from 1975-2009, but nearly 60 percent using acceleration and weight; using acceleration,
features, and functionality led to an estimate of 70 percent improvement. Klier and Linn (2016)
find that technological innovation varies with the stringency of predicted standards and with the
enactment of new standards but do not provide estimates of the magnitudes of baseline
innovation. Wang (2016) finds that cars innovated 1.19 percent per year, and trucks 0.66 percent
per year from 1975 to 2011; a 1 percent increase in CAFE standards led to an additional increase
of 0.32 percent in innovation for cars, and 0.62 percent for trucks. These last two studies argue
that GHG and fuel economy standards increase technological innovation above levels without
regulation.
MacKenzie and Heywood (2015) raise questions with the approach adopted by many of these
studies (focusing on Knittel 2011). In particular, they argue that horsepower and weight are not
necessarily good proxies for characteristics that consumers want, and that estimates both of the
tradeoffs of these characteristics with fuel economy and of technological change are sensitive to
the additional vehicle characteristics considered in the regressions.
If horsepower and weight are not themselves of primary interest to vehicle buyers, then,
according to MacKenzie and Heywood, the measured tradeoffs of horsepower or weight for fuel
economy do not measure changes in metrics important to consumers. Horsepower, for instance,
does not by itself measure the full range of performance-related attributes, which include other
features such as low-end torque, handling, and acceleration. MacKenzie and Heywood (2012)7
find that acceleration performance in 2010 is 20 to 30 percent faster than comparable vehicles in
the 1970s;c in other words, horsepower is not directly proportional to acceleration. Because
acceleration is likely to be of more importance to consumers than horsepower itself, the tradeoff
for horsepower identified in these analyses may not accurately measure impacts important to
consumers.
Similarly, it is unlikely that consumers care directly about vehicle weight; rather, they are
probably more interested in size, safety, cargo capacity, or other characteristics that are
imperfectly correlated with weight. In these studies, a large vehicle with significant mass
B The papers include multiple specifications: they may include different regressions for different vehicle classes, a
variety of additional covariates, or different functional forms. Some of the studies include torque or zero-to-60
times instead of or in addition to horsepower. The values given here are from comparing preferred specifications
specifically using horsepower and weight. The values in different specifications include values within and outside
these ranges; the ranges cited here thus potentially understate the variation in point estimates.
c They attribute this change to improvements in the transformation of engine power to acceleration.
4-5
-------
Consumer Issues
reduction and improved fuel economy would show up in the data to have the same attributes as a
smaller efficient car, though consumers would view them very differently.
The use of weight and horsepower in these regressions may also bias the estimates of
technological change. In these studies, technological change is measured as a residual
improvement in fuel economy after other factors that influence fuel economy are considered.
Including a characteristic (including but not limited to horsepower and weight) in the regressions
means that technological change will not affect that characteristic; its fuel economy elasticity is
fixed. MacKenzie and Heywood (2015) show this effect by using horsepower in their analysis in
one regression, and acceleration (0-to-60 time) in other regressions. When they use acceleration
instead of horsepower, the amount of technological change due to the relationship between
power and acceleration ends up included in their measure of change; that addition increases the
estimated level of technological change. They also point out that technological change to reduce
weight will not show up as change in these other papers, because, as mentioned above, a large
vehicle with mass reduction and improved fuel economy looks in the data like a smaller, efficient
car rather than a vehicle with advanced technology.0
The measures of technological change are also sensitive to the other characteristics used in the
regressions. For instance, Knittel (2011) and Klier and Linn (2016) both include powertrain
types as additional characteristics. By assumption, then, powertrain types are not innovations, or
subject to innovation; a hybrid or diesel will not become more (or less) efficient relative to a
gasoline vehicle over time. E MacKenzie and Heywood (2015) argue that an analysis should not
include those factors because "shifts toward more inherently efficient powertrain technologies
are themselves a part of the overall process of technology change, so it is desirable to capture
their contributions to overall efficiency in the year fixed effects" that measure innovation (p.
922).
Recent work by EPA suggests, in addition, that using historic data to estimate tradeoffs may
miss changes in the relationship between acceleration and C02 emissions with new technologies.
TSD Chapter 2.3.3.2.1 presents results of using the ALPHA model to examine trade-off curves
between CO2 emissions and 0-to-60 acceleration time for three different engine types: port fuel
injection (PFI), gasoline direct injection (GDI), and turbo-downsized (TDS) engines. These
engines have different operating efficiency characteristics, and thus different tradeoff curves.
Most notably, GDI and TDS, the newer technologies, have much flatter tradeoffs than does the
more traditional PFI; in fact, TDS engines reduce CO2 (albeit only slightly) over a range of 0-to-
60 time reductions. Thus, the assumption in the previous research that the tradeoffs among
acceleration, fuel economy, and weight are constant does not appear to accurately represent the
new technologies, and in fact may substantially overestimate the magnitude of the performance-
fuel economy tradeoff.
It is also possible that the estimates for the relationships between fuel economy and other
attributes from these studies may not represent pure technology tradeoffs, and may therefore be
D In their paper, MacKenzie and Heywood separately apply an adjustment to account for innovations in weight
reduction.
E Interacting the characteristic with a measure of time allows for innovation specifically in that characteristic; for
instance, Knittel interacts the manual transmission variable with a time trend, which allows the fuel consumption
of a manual transmission relative to an automatic transmission to vary over time. These papers have few such
interactions; this is the only one in Knittel (2011).
4-6
-------
Consumer Issues
biased. Manufacturers do not produce vehicles with all possible combinations of horsepower,
fuel economy, and weight; instead, the vehicles they produce include a mix of those
characteristics that the companies believe consumers prefer. MacKenzie and Heywood (2015)
find that accounting for a vehicle's specific power relative to the specific power of other vehicles
in the fleet (the quintile of specific power) affects fuel economy, as well as the responsiveness of
fuel economy to acceleration or weight. If these tradeoff curves were purely about technological
relationships, they would not be affected by whether a vehicle was relatively powerful, but only
by its absolute power. They suggest that "the relative sophistication of a vehicle's engine
(compared to others in the same model year) is correlated with weight and acceleration
performance; new technologies are not applied uniformly across all vehicles" (p. 922). As a
result, the tradeoff estimates may not represent strictly technological tradeoffs, but also
manufacturer choices that potentially bias tradeoff estimates.
Based on MacKenzie and Heywood's (2015) work, then, these other studies may not
accurately measure tradeoffs involving characteristics of interest to vehicle owners. Weight, for
instance, is unlikely to matter to consumers, except if that weight comes from size or added
features such as safety. In other work (MacKenzie and Heywood 2012), in which they focus on
the relationship between horsepower and acceleration, they question whether improvements in
acceleration are going to continue indefinitely; they find that trends in 0-to-60 time are consistent
with decay toward an asymptote, and that vehicles in 2010 were within 1 second of the 0-to-60
time asymptotic level.F It is not known if this slowdown in acceleration improvements is due to
physical limits or limits in consumer interest.
Although MacKenzie and Heywood's (2015) analysis presents a more detailed discussion of
these issues compared to the other studies examined here, it is not clear that it is suitable for
quantitative development of a new reference case. First, even 0-to-60 time as a measure of
acceleration may be too narrow a criterion for evaluating performance. Performance, as a
consumer experiences it, is a complex combination of multiple characteristics including initial
launch, ability to pass another vehicle at highway speeds, handling, and cornering. Second, as
noted above, the analysis in TSD Chapter 2.3.3.2.1 suggests that tradeoff estimates based on
historic data may not apply to the newer technologies being implemented. Third, Klier and Linn
(2016) and Wang (2016) suggest that the rate of technological innovation is affected by the level
of the standards. MacKenzie and Heywood's analysis does not examine this effect. Because of
the possibility of a downward bias in innovation from those two studies, their estimates of
innovation are not likely to be sufficient. In addition, the standards for MY2012-2025 are more
significant in magnitude than any changes since the introduction of CAFE in the late 1970s; it is
likely that innovation currently underway in the auto industry is of a different magnitude and
kind than in the past. As a result, estimates of innovation from any of these studies may not be
applicable to what is currently happening in the auto industry.
4.1.3 The Role of the Standards in Promoting Innovation
As discussed above, some authors point to the role of standards in promoting innovation. This
subchapter discusses how innovation may be induced by the standards, and how this innovation
F The authors present the analysis, not only for an average vehicle, but also for vehicles in the fifth and ninety-fifth
percentiles for acceleration, which all show this flattening.
4-7
-------
Consumer Issues
should be viewed differently in accounting for opportunity costs than innovation that may have
occurred in the absence of the standards.
There is a wide body of literature concerning technological change in general.8 The process of
technological change can be divided into three stages: invention, where a new product or process
is first developed; innovation, where the product or process is first commercialized; and
diffusion, where the product or process is widely adopted throughout an industry. This can be a
challenging process: most inventions never make it to the innovation stage;9 even if they are
used by a small number of initial adopters, many technologies never diffuse and thus ultimately
fail.10
It is generally agreed that innovation - the first commercialization of a new product - occurs
on a continuum between two extremes: "major" innovation where product characteristics change,
and "incremental" innovation0 which exploits relatively minor changes to the existing product.11
Although accurately and completely categorizing innovation may be more complex than
applying a simple one-dimensional continuum (as Henderson and Clark (1990) claim), the one-
dimensional model does offer some insight into how industries implement innovation.
A good example of a major innovation, and the role of environmental regulations in spurring
technology diffusion, is gasoline direct injection (GDI). Mercedes introduced a four-stroke GDI
engine into production in 1955.12 Nonetheless, in 2008, prior to the establishment of the
MY2012-2016 standards, only 2 percent of vehicles used gasoline direct injection.13 By 2015,
this number had risen to 42 percent. This changeover shows a major innovation, based on
previous inventions, moving from invention to innovation and eventually to diffusion only when
stimulated by emissions standards.
As in the GDI example, major innovation does not necessarily proceed immediately (or at all)
to diffusion for all promising technologies. In the absence of a forcing mechanism such as
regulation, risk-averse manufacturers may prefer smaller, incremental innovations.14 There are
multiple reasons why manufacturers may prefer incremental innovation to major innovation,
particularly the risk and uncertainty associated with major innovations.
When a company implements a major innovation, the development costs may be high and the
market impacts uncertain. This results in a first-mover disadvantage (see also Section B. 1.3.2.3
of the Proposed Determination Appendix), where a pioneer company fronts the bill to test out a
new technology. In doing so, it may briefly capture the market, but this allows all other
companies to learn about the true demand for the technology without themselves facing any
risk.15 Consumer response to the first mover may give the second mover valuable information
about market acceptance. There are, therefore, incentives to delay the development or adoption
of a new technology until a competitor has already proven that the technology is profitable. If all
producers wait for another one to implement the innovation, the innovation will never enter the
market at all.
In addition, Popp et al.16 point out that there could be "dynamic increasing returns" to
adopting some new technologies, wherein the value of a new technology may depend on how
many other companies have adopted the technology. This could be due to network effects or
G Abernathy and Utterback use "major" and "incremental;" Henderson and Clark, with a two-dimensional
framework, use "radical" and "incremental."
4-8
-------
Consumer Issues
learning-by-doing. In a network effects situation, the usefulness of the technology depends on
adoption of complementary components-for instance, the value of switching to a new fuel
depends on the infrastructure available for providing that fuel, and the value of the infrastructure
depends on the number of vehicles using the new fuel. Learning by doing (see also Appendix
Section A.3.3.3) is the concept that the costs (benefits) of using a particular technology decrease
(increase) with use. Both of these incentivize firms to pursue a "wait and see" strategy when it
comes to adopting new technologies.
Finally, fixed costs and switchover disruptions17 delay technology adoption. Firms often face
major problems in integrating new technologies resulting from major innovations into their
products; in some cases, they may temporarily reduce output.
First-mover disadvantage, dynamic increasing returns, fixed costs, and switchover disruptions
all create barriers to major innovation. Incremental innovations typically face less of these
problems. Thus, in the absence of a driving factor such as regulation, manufacturers are likely to
choose incremental innovations over major innovation.H
Both scientific research18 and popular press19 suggest that the current light duty GHG
standards drive innovation. The mechanism by which the standards affect innovation is the
reduction of the barriers to manufacturers for applying major innovation to new vehicles.1
Since all manufacturers are required to comply with regulations on the same time schedule,
and the technological pace required often outstrips that obtainable by incremental innovation
alone, manufacturers are assured that their competition is likely to implement major
technological innovations simultaneously. Thus, instead of the first-mover disadvantage, there is
a regulation-driven disincentive to "wait and see." It should be noted that companies differ both
in the degree of effort that they face due to the standards, and in the strategies that they choose in
response. Nevertheless, the benefits of generating (or avoiding the need for) credits suggest that
all companies have incentives to pursue major innovations. In addition, there can be synergies
from companies (including suppliers) working on the same technologies at the same time.20
Because of the global nature of the auto industry, it is likely that innovations from U.S.
regulations are likely to affect vehicles in other countries, and regulations from other countries
are likely to affect U.S. vehicles. Because technologies to reduce GHG emissions do not need to
be reinvented for each country, the fixed costs of innovation can be spread over a global market.
It is even likely that many of these technologies will be used in countries without GHG
H This discussion is not intended to imply that major innovation will not happen in the absence of regulation. Many
factors affect the likelihood of a technology proceeding from invention through to widespread dissemination,
including some degree of luck in having the right invention at the right place at the right time with support from
key stakeholders.
1 The U.S. Department of Energy's Advanced Technology Vehicles Manufacturing (ATVM) Loan Program provides
an example of another mechanism to reduce these barriers. The ATVM provides long-term, low-interest rate
loans to support the domestic manufacturing of advanced technology vehicles and automotive components. It can
finance a wide range of project costs, including the construction of new manufacturing facilities; retooling,
reequipping, modernizing, or expanding an existing facility in the U.S; and the engineering integration costs
necessary to manufacture eligible vehicles and components. It is designed to ensure that rising fuel economy
standards do not disadvantage domestic manufacturing. With more than $16 billion in remaining loan authority,
the ATVM program can provide financing to support the manufacturing of fuel-efficient technologies and
components. See fattp://www.energy.gov/tpo/atvm for more information.
4-9
-------
Consumer Issues
standards, due to the use of common manufacturing platforms across countries and to the
ancillary benefits associated with many of these technologies.
Developing a revised reference case could entail estimating incremental technological change,
and projecting vehicle attributes resulting from that innovation, in the absence of the standards.
Developing the control case-the case with the standards in place-could then entail
estimating major technological change induced by the standards and projections of vehicle
characteristics using that greater innovation. The discussion above suggests that conducting such
an analysis may involve inaccurate estimates of the amount of innovation both in the absence of
and in the presence of the standards, and may provide inaccurate estimates of the consequences
of this innovation for specific vehicle characteristics.
Rather than assume a control case with "equivalent performance" to the baseline, one
approach could involve assuming a control case with "equivalent performance" to the reference
case. Since innovations in the reference case are incremental, such an approach could define, not
the reference and control case performance specifically, but rather the difference between them.
In the reference case, it could be assumed that manufacturers would improve vehicle
attributes consistent with historical trends due to the implementation of incremental innovations.
Some of these changes might affect additional implementation of GHG/fuel economy
technologies; in other cases, (for example, infotainment systems, automobile connectivity, or
active safety systems), the standards have no or little technical interaction with those changes.
In the control case, it could be assumed that the standards induce major technological
improvement used to improve fuel economy. Incremental technological improvement would still
be used to improve other vehicle attributes at the same pace as exhibited in the reference case.
Thus, the differences between the control and reference cases are both the existence of fuel
economy targets and the availability of major technological innovations (in addition to
incremental innovations).
It should be noted that there is neither the requirement nor expectation that manufacturers
allocate major innovations solely to fuel economy improvement and incremental innovations
solely to other vehicle attributes. The standards give manufacturers the flexibility to choose what
technologies to apply to which vehicle, when to apply them, and the use of each individual
technology. If major innovations driven by the GHG/fuel economy standards were used to
enhance these other attributes, though, it should be noted that these other attributes would not
have been enhanced in the absence of the standards; those enhancements are ancillary benefits of
the standards.
4.1.4 Potential Ancillary Benefits of GHG-Reducing Technologies
Yet another complication associated with assessing an appropriate reference case is the
potential existence of ancillary benefits of GHG-reducing technologies. Ancillary benefits can
arise due to major innovation enabling new features and systems that can provide greater
comfort, utility, or safety/ The studies discussed above all assume that, other than through
1 It is also possible that these new technologies may have undesirable adverse effects - hidden costs - associated
with them, such as noise or vibration. EPA's analysis to identify hidden costs through review of professional auto
4-10
-------
Consumer Issues
innovation, improving fuel economy reduces power or weight, and thus imposes opportunity
costs; and innovation can be channeled only to fuel economy, weight, or some single-
dimensional measure of performance, such as 0-60 acceleration. When performance is
characterized more broadly as a combination of multiple characteristics, it will often not be
possible to strictly maintain performance along every dimension with the application of
technological innovations. For example, a new technology may have unequal effects on the
various measures of acceleration performance, so that an attempt to maintain performance along
one dimension by resizing the vehicle powertrain will result in an increase or decrease along
other dimensions. In addition, some technologies provide ancillary benefits that improve vehicle
performance and utility along dimensions that are unrelated to acceleration and powertrain
sizing. In such cases, the technologies implemented to reduce GHG emissions enhance other
vehicle characteristics, providing entirely new capabilities and desirable features or resulting in
lower costs for these features than would be otherwise possible.
Some examples of the potential ancillary benefits of GHG reducing technologies are listed
here:
• Mass reduction can provide benefits of improved braking and handling performance,
and on towing vehicles can enable additional towing and hauling capability with same
or similar engine sizing.
• Mass reduction achieved through material substitution from non-ferrous metals
provides greater corrosion resistance.
• Accessory Load reductions achieved through the use of pulse-width modulation
(PWM) on accessory motors for HVAC blower fan speeds provide the benefit of
improved durability.
• Air conditioning system improvements achieved through variable displacement
compressors which adjust automatically rather than shutting off completely provide
the benefit of smoother compressor transitions and less noise.
• Advanced transmissions with wider overall gear ratios and lower 1st gear ratios
provide the benefit of improved launch feel.
• Electric power steering (EPS) systems enable automakers to implement customer
features that utilize automatic steering such as automatic parking features, or trailer
hitch connection assistance.
• EPS systems also provide the capability for variable ratio steering systems which
allow greater steering responsiveness close to center, and reduced effort at large
steering angles, while also reducing the lock-to-lock turns.
• Head-integrated exhaust manifolds and improved thermal management systems
reduce warm-up time for the cabin and provide greater passenger comfort in cold
climates.
• PEVs which can be remotely activated or programmed to precondition the vehicle in
a garage when plugged in provide greater passenger comfort and convenience. In cold
weather, the vehicle can be pre-warmed and defrosted, and in warm weather the
vehicle can be pre-cooled.
reviews, discussed in Proposed Determination Appendix Section B. 1.5.2 and TSD Chapter4.2.2, did not find
evidence of systematic hidden costs of the new technologies.
4-11
-------
Consumer Issues
• PEV systems with an electric axle on AWD vehicles, or even each individual wheel
with electric drive motors, can provide torque vectoring for improved driving
dynamics as the increased torque on the outside wheel is able to steer the car into the
corner.
• LED headlights enable adaptive automotive headlight systems, in which lighting
intensity and direction can be automatically controlled to road, ambient lighting, and
weather conditions.
Additional discussion of the effects of each technology considered in this Proposed
Determination is provided in Chapter 2 of this TSD.
4.1.5 Estimating Potential Opportunity Costs and Ancillary Benefits
As discussed above, it is possible that the standards could potentially lead to opportunity costs
in terms of reduced power or other adversely affected vehicle attributes. At the same time, the
standards may lead to ancillary benefits, perhaps by inducing major innovations that may
mitigate or avoid those opportunity costs, or even enhance other attributes. Because the standards
may contribute both benefits and costs to other vehicle attributes, measuring the net effect on
consumer impacts requires estimates of the values of these attributes to consumers. Although
various commenters (the Alliance of Automobile Manufacturers, National Automobile Dealers
Association, Resources for the Future, Simmons and Tyner, Global Automakers) emphasize the
opportunity costs associated with GHG-reducing technologies, the ancillary benefits have the
possibility of being at least as important.
The most common sources of estimates of willingness to pay for these attributes are models
developed to understand vehicle purchase decisions. These studies quantitatively estimate the
role of various vehicle characteristics, such as size, power, and fuel economy, in those purchase
decisions. The parameters estimated for these characteristics can usually be used to derive
estimates of the value - the willingness to pay (WTP)--of each attribute to consumers. It is
common in this literature, though, for the researchers themselves not to have done the WTP
calculation. In a 1988 study, Greene and Liu21 reviewed the literature to that time; they found,
"The dispersion of estimated attribute values both within and across models is striking," varying
by factors of 5 to 10 or more; for performance, they considered the variation "wild. . . from -$8
to $4,081 per 0.01 cubic inches per pound." To our knowledge, there has not been a study since
that time that has done a comprehensive review of consumers' willingness to pay for vehicle
attributes.K 22
EPA commissioned a new review of the literature to understand what is known about
consumer valuation of vehicle characteristics.23 This review is looking at the metrics various
studies have considered important for consumer vehicle purchase decisions, and is calculating
the WTP values implied by the estimates in those studies. The goal is to determine whether there
are robust WTP values that could be used for monetizing at least some of the opportunity costs
and ancillary benefits. Though the results are preliminary and have not yet been peer reviewed,
they provide some insight into the state of the science on these estimates.
K Greene (2010) conducted a review of consumers' willingness to pay for one attribute, fuel economy, and found
wide ranges of values.
4-12
-------
Consumer Issues
The analysis has focused on studies from 1995 to present, because of the potential for changes
over time in consumer preferences and advances in econometric methods. It also has focused on
U.S.-based studies. Fifty-two papers were identified that provided the data to estimate WTP
values for the light-duty vehicle market. In most cases, the WTP estimates had to be calculated
from statistical results in the papers. The methods are detailed in Greene et al. (2016).
The papers varied in a number of ways. Some used revealed preference data—that is, decisions
by individual consumers in actual market settings. Others used aggregate market data on vehicle
market shares, prices, and characteristics. Still others used stated preference approaches, in
which study participants responded to survey questions. Each of these methods has its
advantages and disadvantages. For instance, revealed preference data are based on actual market
actions. On the other hand, it is often challenging in these studies to capture all the factors that
influence consumer decisions; omissions of key factors may bias the results. In addition, they are
not suitable for gauging preferences for novel features that are not yet implemented. In stated
preference studies, it is much easier to control precisely for key factors by strategic question
designs, but the questions are not based on actual behaviors. Studies also differed in the sources
of the data, the time periods of the data, and the statistical tools used to analyze the data.
Each of the papers includes one or more attributes, and one or more sets of results on the role
of vehicle attributes in consumer purchase decisions. In addition, for a few attributes such as
Vehicle Class, each set of results might contain multiple attributes (e.g., both SUV and compact
car). As a result, the 52 papers produced 799 WTP values for 152 unique attributes. Some of
these attributes are closely related—for instance, dollars per mile, one measure of fuel
consumption, is gallons per mile, another measure of fuel consumption, multiplied by the price
of fuel. The study identified 15 categories of attributes, provided in Figure 4.1.
There are several sources of variability or uncertainty in the estimates. First, different studies
produce different estimates; indeed, sometimes one study produces multiple estimates. Because
these studies use different data and methods, it is not surprising that results differ. If different
studies produce similar estimates of WTP, then it is reasonable to consider those values robust. If
they produce a wide range of values, then further analysis (called meta-analysis) may identify
factors, such as the nature of time period of the data, which affect that range. If patterns are
found, then it may be possible to choose factors which are considered to produce more suitable
results, and use WTP estimates based on those preferred factors. With the results still
preliminary, we have not yet conducted meta-analysis.
Another source of variation in the results is that each estimate of WTP has confidence
intervals around it, because they are estimated statistically. In some cases, the variation is also
due to variation in the population by factors such as income. Most of the results presented below
do not reflect the variation around each estimate, but instead show the variation just in the central
estimates. As a result, the variation presented in the results underestimates the full range of the
estimates.
Yet another source of variation is the way in which each attribute is measured. For instance,
fuel consumption-related measures include miles per gallon, gallons per mile dollars per mile,
miles per dollar, and dollars per year. With assumptions about fuel price or other factors, it is
possible to convert WTP values for these into the same units. The study has conducted these
conversions in some but not all cases.
4-13
-------
Consumer Issues
Observation Count by Grouping
Vehicle class
Size
Safety
Reliability
Range
Prestige
Pollution
Performance
Non-fuel op costs
Model avail.
Incentives
Fuel type
Fuel costs
Fuel availability
Comfort
¦ 4
¦ 6
¦ 5
19
14
17
20
32
169
50
71
53
44
40 60
103
85
127
80
100
120
140
160
180
Figure 4.1 Observation Count by Groups of Attributes
Table 4.1 summarizes the results of the study. "Raw" values include all the estimates for each
attribute; "trimmed" values remove outliers—values extremely different from others. The mean is
the average of all the values. The standard deviation is just of the central estimates—that is, it
does not include the variation around each estimate, but rather is just the variation of the central
estimates. It thus underestimates the full variation around the estimates. In general, it shows wide
variation in the results. For 17 of the 21 attributes using the raw data, the standard deviation is at
least as big as the mean. For context, a value is commonly considered to be statistically
significantly different from zero if the value is at least 2 standard deviations larger than the
mean. Thus, most of these values easily include both negative and positive values as part of their
range. For the trimmed values, the standard deviations exceed the means for 13 of the attributes.
4-14
-------
Consumer Issues
Table 4.1 Willingness to Pay (WTP) Estimates from 52 Studies, 2015$.
Raw
Trimmed
Grouping
Attribute
N
Units
Out-
liers
Mean
SD
Mean
SD
Median
Comfort
Auto-
transmission
9
0,1
1
1,760
3,669
823
2,518
1,111
Front wheel
drive
6
0,1
0
-3,2031
18,031
-32,031
18,031
-26,779
Air conditioning
13
0,1
0
3,521
9,544
3,521
9,544
4,177
Shoulder room
12
$/inch
1
1,085
1,394
705
479
546
Fuel costs
Cost per mile
58
$/cpm
2
-1,251
3,441
-1,291
1,194
-1,147
Cost per year
13
$/($/yr)
1
-67
156
-26
50
-6
Gallons per
mile
20
$/0.01gpm
4
14,354
76,395
-7,972
18,740
-580
Miles per dollar
8
$/(10mi/$)
1
-20,181
27,869
-11,542
14,477
-4,216
Miles per
gallon
10
$/mpg
1
365
659
174
281
64
Fuel type
Electric vehicle
24
0.1
1
-16,515
21,283
-13,851
17,191
-16,837
Hybrid
28
0,1
2
-11,727
44,322
-852
18,441
2,796
Natural gas
7
0,1
2
-5,620
23,691
6,187
3,851
5,006
Perfor-
mance
Acceleration (0-
30 mph)
11
$/sec
0
-1,756
1,886
-1,756
1,886
-1,916
Acceleration (0-
60 mph)
8
$/sec
0
-1,096
627
-1,096
627
-1,183
Horsepower
11
$/hp
4
54
109
13
13
10
HP/weight
29
O.Olhp/lbs
1
1,861
3,523
1,334
2,126
346
Top speed
9
$/mph
0
100
58
100
58
75
AFV Range
Range
23
$/mi
2
89
41
97
32
98
Size
Footprint
17
$/ftA2
1
43,401
163,103
3,856
4,442
3,273
Luggage space
12
$/ftA3
1
4,209
9,655
1,445
1,310
1,100
Weight
19
$/lb
1
10
20
6
8
1
Note: N is the number of observations; Units refers to how the attribute is measured; Mean is the average of central
values; SD is the standard deviation of central values; Median is the middle value of the central values. Negative WTP
values indicate the WTP for a reduction in the named attribute.
The attributes perhaps of most interest for the purposes of the Proposed Determination are for
fuel costs and performance. All the measures of fuel cost show a large variation, spanning
positive and negative values, consistent with Greene's (2010) study of WTP for fuel savings.24
Section B.1.2 of the Proposed Determination Appendix discusses WTP research on fuel
economy, and the assertion from various automakers that EPA should use a 2-to-3 year payback
period in its modeling of consumer demand for vehicles. A payback period can be roughly
converted to a WTP value with a series of assumptions: for instance, the fuel saved in 1 year for
a 0.01 gallon/mile reduction in consumption when a vehicle is driven 12,000 miles is 120
gallons; at a fuel price of $2.50/gallon, the value of a change of +0.01 gallons/mile is -$300/year;
the present value for two years with a 3 percent discount rate is -$591 (-$580 with a 7 percent
discount rate). This study found an average WTP for an increase of 0.01 gallons/mile of $14,354
(standard deviation of $76,395) with the raw data, and -$7,972 (standard deviation of $18,740)
with the trimmed data. The positive value for the mean from the raw data suggests that people
are willing to pay more to reduce fuel economy, perhaps due to association of fuel economy with
4-15
-------
Consumer Issues
other vehicle attributes; this problem with vehicle demand models is discussed in the Proposed
Determination Appendix, Section B. 1.3.4. The trimmed mean can be converted, using these
same assumptions, to an approximate payback period of 54 years with a 3 percent discount rate
(essentially an infinite payback period with a 7 percent discount rate). These results suggest that
there is in fact not a consensus from the literature around a 2-3 year payback period for the value
of fuel savings. As discussed in Section B.1.2 of the Proposed Determination Appendix, and as
demonstrated here, the literature instead suggests a very range of payback periods.
For performance, the study included conversions of 0-to-30 acceleration time and
horsepower/weight to the metric of 0-to-60 acceleration time. Those combined estimates, even
with outliers excluded, ranges from about -$2000/second to +$1000 per second reduction in 0-to-
60 time.
As discussed above, these estimates are still preliminary. EPA will conduct further analysis of
these results, to investigate whether this variation can be explained in part by the nature of the
studies and the data. In the meantime, these results have implications for two aspects of EPA's
assessment of the MY2022-2025 standards. First, it seems premature to use these estimates for
the values of opportunity costs or ancillary benefits of the standards, because of the very wide
ranges, commonly both positive and negative, around the values. The large variation associated
with an analysis using those ranges would not be expected to shed much light on the standards;
effectively, those values could be either positive or negative. Second, it should be noted that
many of these estimates are derived from models of vehicle demand, the same kinds of models
that might be used to estimate changes in sales and fleet mix as a result of the standards. The
wide ranges of estimates derived from these models suggest that the models themselves are
likely to come up with very different responses to the standards. This concern reinforces EPA's
decision at this time not to use a vehicle choice model in its modeling for this Proposed
Determination. Section B.l.3.4 has further discussion of EPA's consideration of consumer
vehicle choice modeling and responses to commenters.
4.2 Consumer Response to Vehicles Subject to the Standards
This subchapter complements the discussions in Sections B.1.3 and B.1.5 of the Proposed
Determination Appendix. In particular, it provides further discussion of why EPA is conducting
a qualitative, but not a quantitative analysis of the effects of the standards on vehicle sales, and it
provides additional findings from our analysis of how professional auto reviewers evaluate the
technologies being used to meet the standards.
4.2.1 Impact of the Standards on Vehicle Sales
Section B.1.3 of the Proposed Determination Appendix discusses the potential effects of the
standards on vehicle sales. On the one hand, all else equal, higher vehicle costs could lead to
depressed sales. On the other hand, all else equal, more efficient vehicles could lead to increased
sales. As discussed, there is a wide range of uncertainty about the relative effects of these two
factors. In particular, as discussed in Section B.1.2 Appendix and in Chapter 4.1.5 of this TSD,
there is a wide range of estimates for the willingness to pay (WTP) for additional fuel economy,
as well as for the closely related payback period for fuel economy that consumers use in their
purchase decisions. Any estimates of the impacts of the standards on sales must make a series of
assumptions on factors such as buyers' WTP for fuel economy and the effects of the standards on
4-16
-------
Consumer Issues
vehicle prices. As a result of this uncertainty, EPA has not made a quantitative analysis of the
effects of the standards on vehicle sales.
Comments from the Alliance of Automobile Manufacturers, Fiat Chrysler Automobiles, Ford
Motor Company, and the National Automobile Dealers Association cite a recent study of the
impacts of the standards on vehicle sales, by the Center for Automotive Research (the CAR
Report), as a basis for their expressed concerns about the potential impacts of the standards on
sales and employment.25 It demonstrates some of the challenges in conducting such an analysis.
It relies on a number of highly questionable assumptions that, if changed, would lead to very
different results, as some recent reviews of the CAR Report indicate.26 As will be outlined
below, EPA's assessment of the CAR Report finds that it is significantly flawed in a number of
respects, including its excessively high cost estimates that are not based on the costs of
technologies for meeting the standards; use of a lower-bound estimate of the fuel savings that
consumers will consider in their purchase decisions; econometric models that appear to produce
contradictory results; and technical errors, such as comparing costs measured in 2025$ to fuel
savings measured in 2015$.
The CAR Report begins with an assumption of technology costs of $2000, $4000, or $6000
per vehicle for a vehicle to go from MY2016 standards to MY2025 standards. It calculates the
effects on fuel consumption based on three estimates of fuel prices from the AEO, a 20 percent
rebound rate, and the assumption that consumers will consider 3 years of fuel savings in their
vehicle purchase decisions. The technology costs with the 3 years of fuel savings subtracted
provide an estimate of the increase in expenditures on vehicles. CAR projects a MY2025 average
vehicle price so that it can estimate the percent change in the vehicle price. It then uses an
elasticity of expenditures with respect to price—the percent change in expenditures associated
with the percent change in price—with a projection of sales in the absence of the standards to
estimate the effect of the higher costs on expenditures. It then divides expenditures by price to
get the estimated effect on vehicle sales. It finds sales effects ranging from an increase of
410,000 to a reduction of 3,710,000, based on different combinations of AEO fuel prices (which
affect fuel savings) and up-front costs.
Each of these steps involves questionable assumptions that significantly affect the results, as
the following discussion will highlight.
First, CAR's cost assumptions—$2000, $4000, and $6000 per vehicle—are not those of the
Draft TAR or any other technology-based analysis. Both Isenstadt (2016) and Cooke (2016)
point out that the cost estimates are based on Greene (1991), which estimates the cost of using
changes in vehicle prices instead of technology to comply with fuel economy standards.27 Cooke
(2016) observes that Greene (1991) concludes that the pricing scheme is effective for very small
changes in fuel economy, but for larger changes, application of technology is less expensive.
CAR fails to indicate the dollar years associated with many of its estimates and results. Cooke
(2016) assumes that these costs are in 2025$; in 2013$, he estimates that the lowest value used in
the CAR Report would be about $1,500, higher than the value in the Draft TAR, $1287 (see
Table 12.44, p. 12-35). Isenstadt assumes that the value is in 2010$; if so, it overstates costs even
more. Isenstadt further calculates that using the $1,565 cost estimate of the Draft TAR (which he
overstates by $279, by including the cost of going from the MY2014 baseline to MY2016
standards) and holding all other assumptions constant would result in a break-even future fuel
price of $2.97; any fuel price higher would lead to fuel savings over 3 years exceeding up-front
4-17
-------
Consumer Issues
costs.L EPA has not independently verified Isenstadt's calculation. Nevertheless, EPA agrees
with both these reviewers that the cost estimates in this report, regardless what dollar-year they
measure, overestimate the costs of the standards.
This ambiguity over the dollar-year occurs throughout the CAR Report, which does not
explain the dollar-year basis for most of its analysis. Cooke (2016) states that the report is
inconsistent on the use of real (fixed dollar-year) and nominal (inflation included) dollars; as an
example, he notes that gas prices are in real dollars, but are compared to costs in nominal dollars,
which he says overestimates payback time by almost 25 percent.
The assumption of 3 years of fuel savings in the CAR Report is based on averaging payback
periods for studies that report payback periods in their results. It cites, but does not include in
that average, studies that calculate the implicit discount rates that consumers use in their results.
Implicit discount rates estimate the interest rates that consumers appear to use in considering the
value of future fuel savings over the lifetimes of the vehicles. If the implicit discount rates are
approximately the same as interest rates consumers would face for vehicle loans, then it appears
that consumers are correctly estimating, and taking into account, the full lifetime of future fuel
savings. The studies cited in the CAR Report that provide discount rates find estimates as high as
27 percent, and as low as 3 percent; the low estimates are well within the range of consumer
interest rates. In reviewing the literature on the role of fuel savings in vehicle purchases, as
discussed in Section B. 1.2 of the Proposed Determination Appendix, the National Academy of
Sciences finds great variation in the estimates of expected payback periods: "The results of
recent studies find that consumers' responses vary from requiring payback in only 2 to 3 years to
almost full lifetime valuation of fuel savings" (p. 9-36).28 Thus, as Cooke (2016) points out, the
CAR Report's estimate for the consumer valuation of fuel economy in vehicle purchases is at the
very low end of the possible range, in part because it excludes a number of studies from its
review (those presenting discount rates instead of payback periods). Higher estimates would
increase the relative value of fuel savings and lead to more positive valuation of vehicles subject
to the standards.
Cooke (2016) points out another flaw: the Report conflates expenditures on vehicles (price
multiplied by quantity) with sales (quantity). In particular, the CAR Report estimates an
elasticity of vehicle expenditures with respect to price—the percent change in vehicle
expenditures due to a 1 percent change in vehicle price—and then compares that elasticity to
estimates in published literature on the demand elasticity—the percent change in sales volume
from a 1 percent change in vehicle price.M Although the CAR Report claims that its estimated
expenditure short-run elasticity, -0.79, is smaller in absolute value than typical results for
demand elasticities (of about -1.1), the demand elasticity based on its expenditure elasticity
estimate is -1.79, larger in absolute value than typical demand elasticities. As a result, the CAR
Report estimates a higher expenditure impact, and thus a higher sales impact, than standard
demand elasticities would predict.
L Isenstadt (2016) cites the Draft TAR for costs of $1565. This value includes the cost of going from the MY2014
baseline to MY2025, $279, and thus overstates by that much the cost of going from MY2016 standards to
MY2025 standards. See Table 12.44, p. 12-35 of the Draft TAR.
M Mathematically, the expenditure elasticity is the demand elasticity plus 1.
4-18
-------
Consumer Issues
The CAR Report uses two statistical models to examine the effects of vehicle prices on
expenditures. The first (Appendix I) is used to develop the elasticity noted above; the second
(Appendix III) is used to project vehicle expenditures in the future in the absence of the
standards. They use different data series, and include different independent variables. The effects
of vehicle price on expenditures in the two models are opposite: in the first model, price reduces
expenditures; in the second model, price increases expenditures. The Report does not explain
why it uses different data sources for these two models, both of which are about demand for
vehicles, nor why they produce different results. The fact that similar models produce opposite
results raises questions about the validity of the models for these purposes.
Because expenditures are price multiplied by quantity, econometric problems may arise by
including price as an explanatory variable. Price is not independent of expenditures, as these
regression models assume. An increase in vehicle prices may increase expenditures if the
increase has a relatively small effect on vehicle sales, or it may decrease expenditures if sales
drop significantly; the CAR Report's models produce these two opposite results. In response to
changes in vehicle prices, people will change not only the number of vehicles they buy, but also
the kinds of vehicles, and thus average vehicle prices. Neither of these models appears to address
this complication in the interaction between expenditures and price, and thus neither can be
expected to produce accurate estimates of the effects of a price change on expenditures.
The CAR Report estimates employment based on its estimates of the change in vehicle sales.N
In particular, it estimates employment in the auto industry of 15 workers per domestic vehicle,
plus 5.6 additional jobs in the economy per auto industry worker, and 1.3 additional jobs in the
economy per dealer employment. The Report does not provide derivations of most of these
estimates. Unlike EPA's employment analysis, in Section B.2. of the Proposed Determination
Appendix, the CAR Report does not consider possible increases in auto industry sector
employment due to development and implementation of fuel-saving technologies, and thus
appears to omit some important employment impacts. In addition, as Cooke (2016) points out,
this "multiplier" approach to employment analysis does not consider the broader macroeconomic
context. As discussed in Section B.2 of the Appendix, when unemployment is low, the primary
effect of regulations on overall employment in the economy is to move jobs from some sectors to
other sectors, rather than to create or eliminate employment. Multiplier estimates may be useful
approximations of employment impacts in a small economy where prices do not adjust; in the
U.S. economy, which does not match those characteristics, employment estimates based on
multipliers are likely to be overestimates.29
Both Isenstadt (2016) and Cooke (2016) suggest that the underlying assumptions of the CAR
Report, if changed, would produce very different results, including more scenarios where vehicle
sales increase rather than decrease. They also point out that the key assumptions about up-front
costs and the payback period for fuel savings that consumers consider in their purchase decisions
are at extreme ends of the expected distributions. Even the lowest cost estimate in the Report is
higher than estimates in the Draft TAR, and the payback period is at the low end of a large range.
N CAR calculates sales by dividing its projected expenditures by a price that it projects assuming a 2.4 percent
annual growth in nominal vehicle price per year. The 2.4 percent growth per year is based on a nominal average
vehicle price of $24,900 in 2000, and $33,400 in 2015. The growth between those years is actually 2 percent per
year.
4-19
-------
Consumer Issues
For these reasons, EPA does not consider the estimates from the CAR Report to provide likely
projected impacts of the MY2025 standards.
EPA recognizes the difficulties involved in making reasonable estimates of these projections.
As discussed above and in Section B.1.3. of the Proposed Determination Appendix, on the one
hand, the vehicles designed to meet the standards will become more expensive, which would, by
itself, discourage sales; on the other hand, the vehicles will have improved fuel economy and
thus lower operating costs due to significant fuel savings, which could encourage sales. Which of
these effects dominates for potential vehicle buyers when they are considering a purchase will
determine the effect on sales. Assessing the net effect of these two competing effects is highly
uncertain, as it rests on how consumers value fuel savings at the time of purchase and the extent
to which manufacturers and dealers reflect technology costs in the purchase price.30 The
empirical literature does not provide clear evidence on how much of the value of fuel savings
consumers consider at the time of purchase. It also generally does not speak to the efficiency of
manufacturing and dealer pricing decisions, as discussed in Section B.1.2. of the Proposed
Determination Appendix. Thus, we do not provide quantified estimates of potential sales
impacts.
4.2.2 Evaluations of the Vehicles Subject to the Standards by Professional Auto
Reviewers
The Draft TAR (Chapter 6.4.1.2) discussed initial results of an examination of the potential
existence of "hidden costs"—undesirable adverse effects of GHG-reducing technologies—via a
content analysis of auto reviews of MY2014 vehicles. Section B. 1.5.1.2 of the Appendix for this
Proposed Determination provides a high-level overview of those results, plus new results from
MY2015 vehicles. Here we provide more detail on the analysis and these results.31
For MY2015 auto review data, RTI used the same sampling and coding procedures as for
MY2014 data.32 One new website, cars.com, was added in MY2015, because its web viewership
met our criteria for inclusion. We followed the same data cleaning process as Helfand et al.
(2016) and dropped the reviews of certain Volkswagen and Audi diesel vehicles due to the
announcement of emissions violation in September 2015. Table 4.2 reports the number of
reviews by website in our analysis for MY2014, MY2015, and the combined data. Table 4.3
reports the number of reviews by vehicle make. Reviews are themselves not conducted to reflect
sales. For instance, MY2015 data contain more reviews of Audi (53) than Honda (30) vehicles.
On the other hand, as with MY2014 data, the reviews by manufacturer are approximately the
same as the number of models offered by manufacturer. It is possible that auto reviews focus on
models with significant redesign. If so, the population of reviews is likely to have a higher
proportion of new technologies than the auto population. The result of our analysis may overstate
negative impacts of fuel-saving technologies if the sample includes more technologies where any
kinks are not yet fully resolved.
4-20
-------
Consumer Issues
Table 4.2 Auto Review Count by Website
Website
MY2015
MY2014
Combined
Review
%
Review
%
Review
%
count
count
count
automobilemag .com
138
11.17
144
14.29
282
12.60
autotrader.com
336
27.21
224
22.32
560
25.02
caranddriver.com
202
16.36
216
21.63
418
18.68
cars.com
90
7.29
0
0
90
4.02
consume rreports. org
79
6.40
86
8.73
165
7.37
edmunds.com
105
8.50
112
11.11
217
9.70
motortrend.com
285
23.08
221
21.92
506
22.61
Total
1,235
100.00
1,003
100.00
2,238
100.00
Table 4.3 Auto Review Count by Make
Make
MY2015
MY2014
Combined
Make
MY2015
MY2014
Combined
Acura
22
24
46
Land Rover
17
15
32
Audi
53
37
90
Lexus
54
23
77
BMW
77
69
146
Lincoln
22
6
28
Bentley
16
11
27
Mini
9
11
20
Buick
11
27
38
Maserati
1
0
1
Cadillac
21
36
57
Mazda
15
49
64
Chevrolet
101
85
186
Mercedes-
Benz
84
74
158
Chrysler
28
4
32
Mitsubishi
10
17
27
Dodge
41
24
65
Nissan
54
40
94
Ferrari
0
7
7
Porsche
47
34
81
Fiat
4
8
12
Ram
8
7
15
Ford
79
47
126
Rolls Royce
4
9
13
GMC
21
17
38
Scion
8
4
12
Honda
30
34
64
Smart
0
1
1
Hyundai
64
19
83
Subaru
59
25
84
Infiniti
23
25
48
Tesla
4
0
4
Jaguar
22
28
50
Toyota
75
63
138
Jeep
15
42
57
Volkswagen
51
32
83
Kia
44
44
88
Volvo
36
5
41
Lamborghini
5
0
5
In Proposed Determination Appendix Section B. 1.5.1.2, we present results that, for each fuel-
saving technology, positive evaluations outweigh negative evaluations for both MY2014 and
MY2015 data. To demonstrate what appears to be variation in the quality of implementation of
technologies, here we summarize the evaluation results by vehicle make, reported in Table 4.4.
We focus on negative evaluations, because these suggest possible problems.
There is a great deal of variation in the percentage of negative evaluations of technologies, as
reported in Table 4.4. For instance, in the MY2015 data, less than 10 percent of the evaluations
are negative for Bentley, Mercedes-Benz, Ram, Rolls Royce, and Tesla over the technologies
examined, while over 40 percent of evaluations are negative for Mitsubishi and Scion. Moreover,
4-21
-------
Consumer Issues
between MY2014 to MY2015, Fiat, Volvo, and Lincoln had the largest decreases in the
percentage of negative evaluations, while Land Rover, Scion, and Jaguar had the largest
increases in the percentage of negative evaluations. There are manufacturers that were
consistently rated well over the two model years. Less than 15 percent of evaluations are
negative for both years for Audi, Dodge, Kia, Mercedes-Benz, Porsche, Ram, and Volkswagen.
For operational characteristics, in the MY2015 data, Bentley, Mini, Porsche, Ram, Rolls
Royce, Tesla, and Volkswagen had less than 15 percent of characteristics evaluated negatively,
while Mitsubishi had negative evaluations of 44 percent of its codes for operational
characteristics. The correlation between the percentages of negative technology reviews and
negative operational characteristics reviews is 0.62 for MY2015 data, and 0.74 for the combined
data. That is, automakers that are rated well on operational characteristics also tend to have
positive or neutral evaluations of efficiency technologies.
Further, we show the heterogeneity in evaluation results by vehicle make for each technology
in Figure 4.2 through Figure 4.22. These reveal great variation for some technologies. For
instance, for start-stop technology, as shown in Figure 4.14 using the combined data, 50 percent
and 36 percent of the evaluations are negative for Subaru and BMW, respectively, while
Chevrolet, Ford, Honda, and Toyota have zero negative evaluations. For the continuously
variable transmission, as shown in Figure 4.16 using the combined data, over 70 percent of the
evaluations are negative for Mitsubishi, 46 percent are negative for Chevrolet, and 15 percent are
negative for Toyota. Using the combined data, low rolling resistance tires (Figure 4.5), electronic
power steering (Figure 4.6), hybrid (Figure 4.11), and plug-in hybrid (Figure 4.12) also show a
relatively greater variation in the evaluation results across vehicle makes.
The heterogeneity appears much smaller for some technologies, such as full electric (Figure
4.13) and mass reduction (Figure 4.20), which have 0 to 17 percent and 0 to 8 percent of
negative evaluations respectively for all the automakers reviewed (using the combined data).
Turbocharging (Figure 4.7), gasoline direct injection (Figure 4.8), high speed automatic (Figure
4.15), and dual-clutch transmission (Figure 4.17) also show a relatively smaller variation in the
evaluation results across vehicle makes in the combined data.
The finding that some manufacturers, including companies with a wide portfolio of vehicle
offerings, appear to implement the technologies well (as evidenced by high levels of positive
evaluations) implies that other automakers may be able to improve their implementation of fuel-
saving technologies and reduce or eliminate any potential hidden costs.
4-22
-------
Consumer Issues
Table 4.4 Percent Negative Evaluations of Technologies and Operational Characteristics by Vehicle Make
Vehicle
2015
2014
Combined
Make
% Negative
Tech
% Negative
Operational
% Negative
Tech
% Negative
Operational
% Negative
Tech
% Negative
Operational
Reviews
Characterist
ics Reviews
Reviews
Characterist
ics Reviews
Reviews
Characterist
ics Reviews
Acura
19.6
19.0
6.9
8.5
11.9
12.9
Audi
14.3
20.5
5.9
9.3
10.5
15.9
BMW
13.2
21.0
9.8
11.0
11.4
16.2
Bentley
2.7
9.9
0.0
6.1
1.8
8.5
Buick
22.7
17.2
27.3
22.3
26.0
20.8
Cadillac
15.2
20.2
9.2
12.2
11.1
15.2
Chevrolet
22.8
23.5
14.0
14.8
18.4
19.3
Chrysler
22.0
22.6
0.0
10.0
19.4
21.1
Dodge
10.7
19.7
12.5
20.6
11.6
20.1
Ferrari
-
-
9.5
10.4
9.5
10.4
Fiat
22.2
55.6
53.3
39.1
41.7
45.9
Ford
14.1
17.8
16.4
15.6
14.9
16.9
GMC
29.0
19.0
14.3
18.2
20.5
18.7
Honda
16.9
16.0
7.7
13.7
11.5
14.6
Hyundai
15.1
20.1
25.5
22.1
18.1
20.7
Infiniti
22.2
26.9
28.1
19.7
25.8
22.8
Jaguar
28.0
16.8
3.8
11.2
11.5
13.4
Jeep
19.2
17.2
26.9
25.1
25.4
23.7
Kia
11.2
29.0
13.3
15.0
12.4
21.3
Lambor-
15.4
19.4
-
-
ghini
15.4
19.4
Land Rover
37.9
28.8
4.5
13.4
17.8
20.8
Lexus
30.5
29.1
26.4
21.6
29.1
26.5
Lincoln
15.3
21.8
38.5
24.4
19.4
22.2
Mini
19.0
12.9
22.7
20.0
20.9
16.3
Maserati
20.0
44.4
-
-
20.0
44.4
Mazda
26.9
19.1
8.9
13.6
12.1
14.6
Mercedes-
9.5
15.2
14.1
13.9
Benz
11.8
14.6
Mitsubishi
42.3
47.4
39.1
56.3
40.3
52.9
Nissan
21.0
30.4
34.1
25.8
26.8
28.5
Porsche
11.4
9.1
10.9
12.5
11.2
10.5
Ram
9.1
11.4
11.1
6.5
10.3
8.9
Rolls Royce
0.0
9.1
0.0
4.6
0.0
5.7
Scion
43.8
28.3
16.7
36.4
36.4
32.0
4-23
-------
Consumer Issues
Smart
-
-
0.0
0.0
0.0
0.0
Subaru
29.2
24.8
32.8
21.8
30.3
23.9
Tesla
0.0
10.3
-
-
0.0
10.3
Toyota
28.6
29.4
14.0
22.5
22.2
26.4
Volks-
wagen
11.9
12.0
13.2
15.4
12.4
13.5
Volvo
12.6
20.2
40.0
30.0
13.8
21.1
4-24
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi I
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari I
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche I
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.2 Reviews of Active Air Dam by Vehicle Make
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.3 Reviews of Active Grill Shutters by Vehicle Make
4-25
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley I
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.4 Reviews of Active Ride Height by Vehicle Make
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.5 Reviews of Low Rolling Resistance Tires by Vehicle Make
4-26
-------
Consumer Issues
MY 2014- MY 2015
Acura I
Audi
BMW I
Bentley
Buick
Cadillac I
Chevrolet I
Chrysler
Dodge I
FIAT
Ferrari I
Ford I
GMC I
Honda
Hyundai I
Infiniti I
Jaguar I
Jeep I
Kia I
Land Rover
Lexus I
Lincoln I
MINI I
Mazda I
Mercedes-Benz I
Mitsubishi
Nissan I
Porsche I
Ram I
Rolls-Royce
Scion
Smart
Subaru I
Toyota I
Volkswagen
Volvo
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.6 Reviews of Electronic Power Steering by Vehicle Make
MY 2014- MY 2015
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.7 Reviews of Turbocharged by Vehicle Make
4-27
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi
BMW I
Bentley I
Buick I
Cadillac I
Chevrolet I
Chrysler
Dodge
FIAT
Ferrari
Ford I
GMC
Honda
Hyundai I
Infiniti
Jaguar
Jeep
Kia I
Land Rover
Lexus I
Lincoln
MINI I
Mazda I
Mercedes-Benz I
Mitsubishi
Nissan
Porsche I
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen I
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.8 Reviews of Gasoline Direct Injection by Vehicle Make
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.9 Reviews of Cylinder Deactivation by Vehicle Make
4-28
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi I
BMW I
Bentley
Buick
Cadillac
Chevrolet I
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep I
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda I
Mercedes-Benz I
Mitsubishi
Nissan
Porsche
Ram I
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen I
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.10 Reviews of Diesel by Vehicle Make
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.11 Reviews of Hybrid by Vehicle Make
4-29
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac I
Chevrolet I
Chrysler
Dodge
FIAT
Ferrari
Ford I
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz I
Mitsubishi
Nissan
Porsche I
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota I
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.12 Reviews of Plug-In Hybrid Electric by Vehicle Make
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.13 Reviews of Full Electric by Vehicle Make
4-30
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.14 Reviews of Stop-Start by Vehicle Make
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
m
¦ ¦
Cadillac
1 H
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
¦
GMC
¦
Honda
Hyundai
Infiniti
¦
Jaguar
Jeep
¦
Kia
Land Rover
Lexus
1 ¦
Lincoln
MINI
Mazda
Mercedes-Benz
1 H
Mitsubishi
¦
Nissan
Porsche
Ram
¦
Rolls-Royce
Scion
¦
1
Smart
Subaru
Toyota
Volkswagen
Volvo
II
1
Aeura
Audi
1 H
BMW
Bentley
Buick
II
Cadillac
1 ¦
Chevrolet
Chrysler
¦
Dodge
FIAT
1
Ford
¦
GMC
Honda
¦
Hyundai
Infiniti
Jaguar
Jeep
Kia
I
Lamborghini
Land Rover
Lexus
¦
Lincoln
MINI
i
Maserati
Mazda
ii
Mercedes-Benz
Mitsubishi
¦
Nissan
Porsche
\m
Ram
¦
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
1 ¦
Volvo
I
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.15 Reviews of High Speed Automatic by Vehicle Make
4-31
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Negative
Figure 4.16 Reviews of Continuously Variable Transmission by Vehicle Make
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.17 Reviews of Dual-Clutch Transmission by Vehicle Make
4-32
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.18 Reviews of Electric Assist or Low Drag Brakes by Vehicle Make
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.19 Reviews of Lighting-LED by Vehicle Make
4-33
-------
Consumer Issues
MY 2014- MY 2015
Acura I
Audi I
BMW I
Bentley I
Buick
Cadillac I
Chevrolet
Chrysler
Dodge I
FIAT
Ferrari I
Ford
GMC
Honda I
Hyundai
Infiniti I
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI I
Mazda
Mercedes-Benz
Mitsubishi I
Nissan
Porsche I
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen I
Volvo
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.20 Reviews of Mass Reduction by Vehicle Make
MY 2014- MY 2015
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Acura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.21 Reviews of Passive Aerodynamics by Vehicle Make
4-34
-------
Consumer Issues
MY 2014- MY 2015
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Land Rover
Lexus
Lincoln
MINI
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Subaru
Tesla
Toyota
Volkswagen
Volvo
Aeura
Audi
BMW
Bentley
Buick
Cadillac
Chevrolet
Chrysler
Dodge
FIAT
Ferrari
Ford
GMC
Honda
Hyundai
Infiniti
Jaguar
Jeep
Kia
Lamborghini
Land Rover
Lexus
Lincoln
MINI
Maserati
Mazda
Mercedes-Benz
Mitsubishi
Nissan
Porsche
Ram
Rolls-Royce
Scion
Smart
Subaru
Tesla
Toyota
Volkswagen
Volvo
Number of Reviews
Number of Reviews
Number of Reviews
Negative
Figure 4.22 Reviews of Fuel Cell by Vehicle Make
For further assessment of the existence of hidden costs, we examine the relationship between
the operational characteristics and the fuel-saving technologies using regression models.
Although the data suggest that automakers that are rated well on operational characteristics also
tend to implement efficiency technologies positively, there might exist selection bias that results
in the correlation. For example, it could be that the same vehicles without start-stop would also
generate negative evaluations of operational characteristics. To reduce the concerns about
selection bias, following Helfand et al. (2016), we estimate a series of linear probability models
for each operational characteristic that includes fixed effects for make, vehicle class, and
website. We do this analysis separately for MY2014 and MY2015 data. In addition, we run the
analysis for combined MY2014 and 2015 data, where we control for year (e.g., macroeconomic
conditions common to all manufacturers), year-by-website (e.g., a website's year-specific review
standards and preferences), year-by-class (e.g., year-specific innovation for a vehicle class
common to all manufacturers), and year-by-make (e.g., a company's year-specific innovation
and/or production strategy) fixed effects. All of those might be correlated with the review results
of technology and the review results of operational characteristics, and thus bias the actual
relationship.
Table 4.5 reports the number of statistically significant associations with an operational
characteristic for each technology, out of 22 operational characteristics. The dependent variable
across the columns is an indicator variable equal to one when a negative evaluation of the
operational characteristic was recorded. Overall, the use of fuel-saving technologies is not
correlated very often with a negative evaluation of an operational characteristic; indeed, the 75
negative coefficients (indicating that the technology is associated with a reduced probability of a
4-35
-------
Consumer Issues
negative evaluation of an operational characteristic) is much larger than the 24 positive
coefficients (indicating that the technology is associated with an increased probability of a
negative evaluation of a characteristic). The presence of GDI, passive aerodynamics, or start-
stop, for instance, is not correlated in any of these data series with a negative evaluation of an
operational characteristic.
Comparing the number of positive coefficients (potential hidden costs) with negative
coefficients (potential hidden benefits) involves some limitations. First, counting coefficients
does not indicate the magnitude of the effects. Secondly, for some of the technologies (especially
active air dam, active grill shutters, active ride height, and fuel cell), statistically significant
correlations may not appear because sample sizes are so small. Third, as discussed in Helfand et
al. (2016), statistically significant coefficients do not indicate that the presence of the technology
caused either a hidden cost or a hidden benefit; it is possible, e.g., that a characteristic in a
vehicle would have been rated badly even if a different technology had been used. Nevertheless,
these results indicate that hidden costs are not inevitable in the presence of these technologies.
Indeed, the evidence is suggestive of some hidden benefits, as discussed in Chapter 4.1.4.
4-36
-------
Consumer Issues
Table 4.5 Summary of Statistically Significant Regression Coefficients by Technology
Fuel-Saving
MY2014:
MY2014:
MY2015:
MY2015:
Combined:
Combined:
Technology
Count of
Count of
Count of
Count of
Count of
Count of
Significant
Positive
Coefficients
Significant
Negative
Coefficients
Significant
Positive
Coefficients
Significant
Negative
Coefficients
Significant
Positive
Coefficients
Significant
Negative
Coefficients
Active Air
Dam
0
1
0
0
0
1
Active Grill
Shutters
1
5
0
9
0
9
Active Ride
Height
0
4
0
0
0
4
Low
Resistance
Tires
1
1
0
7
0
3
Electronic
Power
Steering
1
1
0
3
3
1
Turbocharged
0
3
4
2
3
3
GDI
0
4
0
2
0
3
Cylinder
Deactivation
1
3
2
2
1
5
Diesel
0
5
2
4
2
3
Hybrid
1
3
0
2
1
4
Plug-In
Hybrid
Electric
1
2
2
2
3
1
Full Electric
0
1
0
4
0
2
Start-Stop
0
3
0
7
0
6
High Speed
Automatic
0
7
1
6
0
6
CVT
7
1
3
1
5
1
DCT
0
1
2
1
3
1
Elec Assist or
Low Drag
Brakes
0
2
0
9
0
4
Lighting-LED
2
5
0
2
1
3
Mass
Reduction
0
4
2
3
1
4
Passive Aero-
dynamics
0
6
0
7
0
5
Fuel Cell
0
0
1
6
1
6
Total
15
62
19
79
24
75
4-37
-------
Consumer Issues
4.3 Impacts of the Standards on Vehicle Affordability
Section B.1.6 of the Proposed Determination Appendix provides an overview of the analysis
of the impacts of the standards on vehicle affordability. As will be discussed below, affordability
is not a well-defined concept, but it is potentially an important consideration not only to policy-
makers, but to all stakeholders.
This TSD subchapter expands upon the analysis in the Appendix, and updates information
presented in the Draft TAR, as well as in a memo to the docket on affordability.33 It begins with
a literature review on the conceptualization and definition(s) of affordability for various
consumer goods. It then poses, and subsequently assesses, four questions by which to analyze
vehicle affordability:
• Effects on lower-income households;
• Effects on the used vehicle market;
• Effects on access to credit; and,
• Effects on the low-priced vehicle segment of the new vehicle market.
4.3.1 Literature Review: Definitions of Affordability
While the term "affordability" is very commonly used in colloquial settings, there is little
consensus on an academic definition for the term, and the concept of "affordability" is murky at
best. Hancock (1993) lamented that "affordability has been gaining much currency in housing
policy debates, but neither government nor academic researchers have given much consideration
to defining it."34 Quigley and Raphael (2004) stated that "economists are wary, even
uncomfortable, with the rhetoric of 'affordability,' which jumbles together in a single term a
number of disparate issues.. ,"35 Bradley (2008) identified affordability as "a vague
concept... When pundits use the word 'afford,' there is no clear definition of affordability; it is at
best a subjective notion."36 Perhaps most candidly, Bartl (2010) declared that "affordability is a
new 'alien' concept penetrating the field of contract and consumer law."37
Even though the concept of affordability has been characterized as vague, subjective, alien,
and vexed, several economists and federal agencies have attempted to define affordability, most
often in the context of specific goods. These goods include energy, food, telephone service,
health insurance, and housing.
For energy, Bartl (2010) defines affordability as "primarily an economic category having to
do with the ability of certain consumers or consumer groups to pay for a minimum level of
service." She states that affordability has two dimensions: "First, it is necessary to ensure
reasonable prices for all users, and, secondly, to ensure the provision of services to persons who
cannot afford it under normal market (or prior monopoly) conditions." This assumes that
universal access to energy services is a basic necessity. Fankhauser and Tepic (2007), for water
and energy, have a similar definition, and then operationalize it as the share of monthly
expenditures (or income) spent on utility services.38
For food, Blaylock et al. (1999) determined affordability based on the ratio of expenditures on
food to household income.39 However, Blaylock et al. also explained that food expenditures are
not dictated entirely by household income and costs: nutritional value, taste, and convenience
4-38
-------
Consumer Issues
factored into consumers' preferences on food choices. Furthermore, they argued that the costs of
food consumption must be considered in a short-term context, including the upfront cost of food
purchases, the time expended purchasing food, and sacrifices in taste for lower upfront cost or
nutritional quality; and a long-term context, including the potential health risks of eating cheap,
nutritionally questionable food. For example, a reduction in public consumption of high-
cholesterol foods after increasing public information on the risks of cholesterol showed that "it is
not inevitable that affordable food will defeat nutrition information in determining diets."
Although "affordable" here is defined as having low upfront cost, consumers appear to factor in
long-term costs, such as health risks, when deciding whether food is affordable.
The Department of Health and Human Services also uses the ratio-of-income approach to
determine food affordability for federal poverty guidelines. The Department of Agriculture
determines a nutritionally adequate bundle of food for households, and the Department of Health
and Human Services sets the poverty standard "based on the relationship of the price of this
bundle to income" (Glied, 2009).40 This definition of affordability thus takes both upfront cost
and a measure of food quality into account.
For telephone service, Milne (2000) uses a similar ratio approach to determine affordability.
She states that one key assumption is that "there is a certain percent of household income which,
on average, a new subscriber finds acceptable to devote to telephone service," referring to this as
the "affordability threshold."41 However, she also states that the assumption that all households
with incomes beyond the affordability threshold will subscribe to telephone service "does not
describe individual behavior." She explains that "households will deviate from this behavior in
both directions," but that "we can be confident that propensity to subscribe to the phone does
increase with income level, and decrease with proportion of expenditure devoted to telecoms."
Thus, this definition of affordability rests primarily on the share of income devoted to telephone
expenditures, but gives some acknowledgment to the effect of consumer preferences.
Additionally, Milne essentially defines access to telephone service as a necessity, declaring that
"the notion that basic telephone service should be affordable has received widespread assent."
For health insurance, Glied (2009) distinguishes between colloquial and more academic
usages of affordability. The more colloquial usage "implies that the primary reason someone
chooses not to purchase a good or service is that the person does not have the ability to pay for
it." However, in more academic terms, "a household is said to afford such a purchase if it would
be left with enough income to meet its other socially defined minimum needs." Like food,
energy, and telephone service, health insurance in this context is presumed to be a necessity, for
which there is a socially defined minimum level of necessary consumption.
In discussions of affordability, perhaps the most commonly considered good is housing.
Maclennan and Williams (1990, cited in Haffner and Heylen 2011, p. 595) define affordability as
"concerned with securing some given standard of housing (or different standards) at a price or a
rent which does not impose, in the eyes of some third party (usually government) an
unreasonable burden on household incomes."42 Again, this definition refers to socially defined
minimum standards for housing and other goods. Similarly, Bramley (1990) characterizes
affordability as a situation where "households should be able to occupy housing that meets well-
established (social sector) norms of adequacy (given household type and size) at a net rent which
leaves them enough income to live on without falling below some poverty standard."43
4-39
-------
Consumer Issues
Clarifying this definition somewhat, Whitehead (1991) refers to affordability as "the
opportunity cost of housing vis-a-vis other goods and services."44 Hancock (1993) refers to the
essence of the concept of affordability as "what has to be foregone in order to obtain the merit
good and whether that which is foregone is reasonable or excessive in some sense."45 Also
taking opportunity cost into account, Stone (2006) defines affordability as expressing "the
challenge each household faces in balancing the cost of its actual or potential housing, on the one
hand, and its non-housing expenditures, on the other, within the constraints of income."46
As with other goods, housing affordability is often operationalized using a ratio approach. The
Department of Housing and Urban Development (HUD) (U.S. Department of Housing and
Urban Development, 2015) characterizes a household as able to afford housing if it pays no more
than 30 percent of its income on housing.47 HUD also considers supply in its metrics to analyze
housing affordability. In its "Worst Case Housing Needs" biennial report to Congress (U.S.
Department of Housing and Urban Development, 2011), HUD highlights the supply of rental
units that would be affordable (presumably using the 30 percent-of-income standard) to
consumers within a given income class (Steffen et al., 2015).48
While ultimately disagreeing with the simple use of the ratio approach to determine housing
affordability, Bogdon and Can (1997) also incorporate supply into their definition of housing
affordability by using the housing affordability mismatch approach, which "considers both
housing supply and housing demand by comparing the existing housing cost distribution with the
distribution of household incomes."49 Similarly, Gan and Hill (2009) develop affordability
indices that take account of "the whole distribution of household income and house prices rather
than just the median."50 This accounts for the demand for various housing types based on
household income and the supply of housing units appropriate for households with various
incomes. Fisher et al. (2009) expand on the supply concept by advocating tracking the supply of
units in different geographic areas and accounting for the effect of the spatial distribution of
various housing units on prices.51
As described briefly above, despite its widespread use in affordability indices for a variety of
goods, the ratio approach is also widely criticized. Hancock (1993) states that "a ratio definition
says nothing about what might be an acceptable opportunity cost of that which is being
consumed," and that it "therefore makes little sense to define affordability in terms of the ratio of
housing costs to incomes if it is believed that opportunity cost is important." Stone (2006) echoes
this criticism, explaining that the ratio approach assumes that someone with a lower income who
spends as high a proportion of his/her income on housing as someone with a higher income can
afford to spend much less in an absolute sense on other necessities. Bogdon and Can (1997) also
criticize the ratio approach as "flawed." They state that the ratio approach does not account for
quality, differences in preferences, households' actual financial constraints, or actual user costs.
Instead, Stone (2006) advocates the use of the residual income approach, which measures the
actual amount of disposable income (as opposed to the percentage of income) remaining after
accounting for housing expenditures and determining whether that residual income is sufficient
to purchase minimum acceptable quantities of other necessities.
Another trend within more recent housing affordability literature is distinguishing between
short-term affordability and long-term affordability. Haffner and Heylen (2011) define the short-
term costs as the "out-of-pocket cash flows or expenses that households make to finance the
access to their home," and the long-term affordability as the "'long-run ability' of households to
4-40
-------
Consumer Issues
pay the so-called user costs or price of housing consumption." This relates closely to Gan and
Hill's (2009) distinction of purchase versus repayment affordability, although repayment
affordability only takes the cost of repaying the mortgage into account and does not encompass
the broader user costs associated with Haffner and Heylen's long-term affordability concept.
User costs are certainly not a new idea in housing affordability literature. Hancock (1993)
states that "in theory, the housing costs of owner-occupiers should be measured by the user-cost,
which takes the opportunity-cost of equity, depreciation, and the effect of capital gains into
account, in addition to the mortgage payments, local property taxes and the maintenance of the
property." Quigley and Raphael (2004) also note user cost: "To an economist, however, the
affordability of owner-occupied housing is a bit more complicated - by taxes, by depreciation
and by capital gains." Similarly, Bogdon and Can (1997) recognize that "monthly home owner
costs may also be a misleading measure because the true measure for home owners is the user
cost, which includes expected appreciation."
Like food, much of the literature on housing affordability also emphasizes the importance of
incorporating quality. Lerman and Reeder (1987) develop a "'quality-based' definition of the
housing affordability problem that distinguishes households having too little income to rent
minimally adequate but decent, safe, and sanitary housing for less than a specified percentage of
income (30 percent) from households whose incomes are sufficient."52 Fisher et al. (2009)
clarify the usage of quality set forth by Lerman and Reeder to "develop an affordability
methodology that accounts for job accessibility, school quality, and safety." Thalmann (1999)
uses two indicators of housing affordability in a similar fashion: one indicator "compares income
to the average rent the market charges for housing deemed appropriate for a household," and the
second indicator "compares current housing consumption with appropriate housing
consumption."53 This approach takes a different tack than many — instead of identifying just the
socially acceptable minimum quality of housing for a given household, Thalmann also identifies
a socially acceptable maximum quality and uses this range to determine affordability of the
housing stock.
Quigley and Raphael (2004) note that affordability in the context of housing "jumbles
together in a single term a number of disparate issues: the distribution of housing prices, the
distribution of housing quality, the distribution of income, the ability of households to borrow,
public policies affecting housing markets, conditions affecting the supply of new or refurbished
housing, and the choices that people make about how much housing to consume relative to other
goods." And like other goods that are considered basic necessities, Quigley and Raphael refer to
a "socially imposed minimum standard" for housing.
However, Quigley and Raphael state that defining affordability for housing is not the same for
all incomes. For American households who own their home, "housing 'affordability' refers to the
terms on which dwellings can be purchased and loans to purchase these assets can be amortized."
However, for households with lower incomes, "'affordability' refers to the terms of rental
contracts and the relationship between these rents and their low incomes." Stone (2006) shares
this sentiment: "Affordability is not a characteristic of housing - it is a relationship between
housing and people. For some people, all housing is affordable, no matter how expensive it is;
for others, no housing is affordable unless it is free." This implies that how one defines
affordability can depend heavily on income.
4-41
-------
Consumer Issues
Despite differing definitions of affordability offered for different types of goods, there are
many similarities and shared themes across definitions. One shared theme is that instead of
focusing on the traditional economic concept of willingness to pay, any consideration of
affordability must also consider the ability to pay for a socially defined minimum level of a
good. As discussed below, however, all of the goods considered in this literature review were
considered basic necessities, and the absence of a socially defined minimum level of adequate
consumption of the good in question complicates determining consumers' ability to pay for such
a good.
Often, the ability to pay is determined based on the proportion of income devoted to
expenditures on a particular good. However, this ratio approach is widely criticized. For
example, it does not account for the opportunity cost associated with the consumption of a
particular good. That is, when purchasing at least the socially-defined minimum level of one
good, one must consider the utility of other goods, some of which may be necessities, which a
consumer must forego based on his/her income. The ratio approach also does not incorporate
quality differences in the considered good. For instance, one consumer may pay $700 per month
to rent a spacious, clean, well-maintained apartment while another consumer with the same
income may pay $700 per month to rent a small, moldy, crumbling apartment that does not meet
socially defined minimum housing standards. The ratio approach also does not incorporate
heterogeneity of consumer preferences. For instance, two consumers with the same income may
purchase housing of wildly different quantity and quality based on the utility they receive from
housing versus other goods that they can purchase.
Considering this heterogeneity of preferences is important for attempting to identify the
socially defined minimum level of service necessary for each type of good. Here, there are two
approaches at play. One is the normative approach, which uses a set and arbitrary level of service
as the minimum adequate level for consideration of the ability to pay. The other is the behavioral
approach, which expands on the normative approach by considering consumer preference
intensity and determining whether the consumer of a particular income with the median
preference intensity can purchase the normatively-determined minimum acceptable level of
service.
One alternative approach to determining the ability of consumers to pay for a certain good is
the permanent income hypothesis, which states that consumers' levels of consumption are
explained more by what those consumers expect to earn over a period of time rather than their
temporary income, which can often fluctuate wildly. Thakuriah and Liao (2006) thus use total
expenditures as a proxy for consumers' permanent income in order to estimate consumers'
ability to pay for transportation expenditures.54
Another common theme, particularly when discussing affordability of housing, is considering
both the short-term costs and long-term costs associated with consumption of a particular good to
assess affordability. This includes both the cost of accessing the good, which often refers to
access to and cost of financing, as well as the user cost of the good over time. These costs are not
equal. For instance, one may be able to afford the costs associated with owning a home,
including mortgage repayment, property taxes, maintenance, and depreciation, while not having
sufficient savings to cover the necessary down payment to access financing.
4-42
-------
Consumer Issues
4.3.2 Relating Affordability Themes to Vehicle Standards
All the goods considered in this literature review (energy, nutrition, basic telephone service,
health insurance, and housing) arguably could be considered necessities. For instance, with
health care, Bundorf and Pauly (2009) "assume that there is a 'special' societal interest in
medical insurance and medical care that need not apply equally to other types of consumption."55
These goods thus have socially defined minimum adequate levels of consumption (although
there may not be consensus on those levels).
However, unlike the goods discussed above, there is no socially defined minimum level of
consumption for vehicles. Considering consumption only of vehicles defines the service
provided by vehicles too narrowly. Vehicles are one means to the end of transportation.
A thorough review revealed no attempts to define the affordability of transportation, and
vehicles more specifically. Thakuriah and Liao (2006) attempt to define ability to pay for
transportation expenditures, but do not offer a definition of affordable transportation. A report by
the Manhattan Strategy Group for the Department of Transportation and the Department of
Housing and Urban Development (HUD) (Schanzenbach and McGranahan, 2012) attempts to
create metrics of various types of vehicle costs to be included in HUD's Location Affordability
Index, which considers housing and transportation costs based on location. However, this report
also did not attempt to define vehicle affordability.56
Given the prevalence of heavily subsidized public transit systems, including free rides for
vulnerable populations, it seems that societies often consider access to transportation in some
sense a basic necessity. However, it is not clear how to identify the socially acceptable minimum
level of transportation service. It seems reasonable to assume that such a socially acceptable
minimum level should allow access to employment, education, and basic services like the
grocery store, but it is not clear where consumption of transportation moves from practical to
luxury. Normatively defining the minimum adequate level of transportation consumption is
difficult given the heterogeneity of consumer preferences and living situations. As a result, it is
challenging to define how much residual income should remain with each household after
transportation expenditures. It is therefore not surprising that academic and policy literature have
largely avoided attempting to define transportation affordability.
We therefore do not propose a quantitative measure of the affordability of new vehicles. As
discussed in Proposed Determination Appendix Section B.1.6, although some comments we
received on the effects of the standards on affordability requested a quantified analysis of the
issue, those comments did not suggest methods for that analysis. Instead, as in Draft TAR
Chapter 6.5, we consider four questions that relate to the effects of the LDV GHG standards on
new vehicle affordability: how the standards affect low-income households; how the standards
affect the used vehicle market; how the standards affect access to credit; and how the standards
affect the low-priced vehicle segment.
4.3.3 EPA's Assessment of the Impacts of the Standards on Affordability
The effects of the standards on vehicle affordability are discussed in the Proposed
Determination Appendix, Section B.1.6. Below we report further detail about the data and our
assessment of four aspects of affordability: the effects of the standards on lower-income
4-43
-------
Consumer Issues
households, on the used vehicle market, on whether access to credit may limit consumer's ability
to purchase new vehicles, and on the availability of low-priced vehicles.
4.3.3.1 Data: Consumer Expenditure Survey
To analyze the characteristics of households who purchase new and used vehicles and the
vehicles that these households purchase, we used the public use microdata of the Consumer
Expenditure Survey (CES), specifically the interview and detailed expenditure files for years
2007 through 2015 (Bureau of Labor Statistics, 2015).57 The CES is performed annually by the
Department of Labor-Bureau of Labor Statistics (BLS). It is conducted in-person based on a
representative sample of U.S. addresses.
Data from this survey were chosen for several reasons. First, the survey includes a sample of
consumer units that is designed to be representative of the total US population. The sampling
frame for each CES is derived from a list of households included in the 2000 Census and a list of
households constructed after the 2000 Census.0 Consumer units are roughly equivalent to
households, and from this point forward will be referred to as "households."p Second, the survey
is performed annually, which allows us to track recent vehicle purchase behavior over a greater
number of reference years than other data sets and establish trends. Third, the CES includes
detailed information on both household demographics and major expenditures, particularly
related to vehicles and transportation. Fourth, the public use microdata for the CES from years
2003 onwards is available online for free, which allows easy public access to the data used in our
analysis.Q Fifth, the CES is widely used by policymakers and academics to study welfare
changes across socioeconomic groups.R
Other articles and reports have used the CES to examine the relationship between vehicle
purchases and household characteristics. For example, Goldberg (1996) used microdata from the
CES to try to explain auto dealer price discrimination based both on household characteristics
(e.g. race or gender) and vehicle purchase characteristics (e.g. trade-in and financing source).58
Yurko (2011) used data from the CES to examine the relationship between household income
and vehicle quality, specifically vehicle age.59 Schanzenbach and McGranahan (2012) used
estimates for the costs of car ownership obtained from CES data to include in the Department of
Housing and Urban Development's Location Affordability Index. Thakuriah and Liao (2006)
0 For more information on how the sample for the CES is selected, please see User's Documentation included in the
CES public use microdata for the Interview Survey each year and the CES Frequently Asked Questions,
http ://www.bls. gov/cex/faq. htm#q 17.
p According to the CES glossary (http://www.bls. gov/cex/csxgloss.fatm). "A consumer unit comprises either: (1) all
members of a particular household who are related by blood, marriage, adoption, or other legal arrangements; (2)
a person living alone or sharing a household with others or living as a roomer in a private home or lodging house
or in permanent living quarters in a hotel or motel, but who is financially independent; or (3) two or more persons
living together who use their income to make joint expenditure decisions. Financial independence is determined
by the three major expense categories: Housing, food, and other living expenses. To be considered financially
independent, at least two of the three major expense categories have to be provided entirely, or in part, by the
respondent."
Q To access the public use microdata for the CES, visit the Public-Use Microdata Home Page,
http://www.bls.gov/cex/pumdhome.htm.
R For more information on how the CES is used by academics and policymakers, visit "Value of the Consumer
Expenditure Survey," http://www.bls.gov/respondents/cex/cevalue.htm.
4-44
-------
Consumer Issues
used microdata from the CES to compare total annual expenditures (as a proxy for permanent
incomes) with investments in mobility.
Note that this analysis and the CES focus on household vehicle purchase behavior, and not the
entire new or used vehicle market, which includes fleet purchases. It is also important to note
that we do not consider leases in this analysis of CES data. The leased vehicle data reported in
the CES do not include the calendar year when the lease was contracted; as a result, we are
unable to compare household leasing behavior with vehicle purchase behavior on a calendar year
basis. We thus focus only on vehicles owned by residential households and thus understate the
number of vehicles in households.
One limitation with using the CES is that the data on expenditures and households'
characteristics are self-reported. This makes the data subject to problems with respondents' recall
of information, or misrepresentation. This is a limitation of all survey data and is not unique to
the CES.
The expenditure variables in the CES we examine are CARTKNPQ and CARTKNCQ for
expenditures on new cars and trucks, CARTKUPQ and CARTKUCQ for expenditures on used
cars and trucks, and GASMOPQ and GASMOCQ for expenditures on gasoline and motor oil.
Following the estimation procedure section from the CES documentation, we calculated an
aggregated measure for a calendar year by weighting the amount of time (MOSCOPE in the
documentation) so that each reported expenditure actually applies to the year based on the
interview year and interview month. We then took weighted averages of the variables, where our
final weight in Stata is the product of the "MO SCOPE" and the "finlwt21" variable, the
variable recommended by the BLS for estimating the population and was used for all means and
medians.s By this estimation procedure our calculations of expenditures on new vehicles, used
vehicles, and gasoline and motor oil were able to exactly match the mean expenditures reported
in the online CES tables (e.g., for 2015, see http://www.bls.gov/cex/2015/combined/qiiintile.pdf.
"Cars and trucks, new," "Cars and trucks, used," and "Gasoline and motor oil").
The income variable we examined is total household income before tax, FINCBTXM in the
CES. In the Draft TAR, we used after-tax income, FINCATAX in the CES. We switched to
before-tax income because before-tax income is more typically used in analyses of the CES data.
In addition, BLS had not derived the non-imputed after-tax income since 2015, and has derived a
new imputed after-tax income, FINATXEM, since 2013; as a result, the data series is not
consistent over the time period studied here. We used the estimation procedure mentioned above
to obtain weighted median income for each year. Using the weighted median income (in 2015, it
was $50,000), we divide the annual sample into lower-income households (those with income
less than $50,000) and higher-income households (those with income over $50,000), and
produce summary statistics of expenditures by the two income groups.
In order to generate debt-to-income (DTI) ratios, the debt expenditure variables we used are
MRTPMTX for mortgage, MRTPMTG for home equity loans, PAYMENTX for vehicle loans,
QRT3MCMX for rental home payments, and CONTEXPX for contributions, including child
s See http://www.bls.gov/cex/2015/csxintvw.pclf. p. 24-30, for the documentation for 2015 CES data and estimation
procedures of unweighted and weighted statistics.
4-45
-------
Consumer Issues
support and alimony only. Using the weight mentioned above, we summed over the annual
expenditures on these payments to calculate debt for each household.
4.3.3.2 Effects on Lower-Income Households
We use the CES data for the years 2007-2015 to classify households with before-tax incomes
below the weighted median as "lower income," and the other half of households are considered
"higher income." For example, the weighted medians in 2015 and 2014 were $50,000 and
48,465, in 2015$, respectively.
As we pointed out in the Draft TAR (Chapter 6.5.1), lower-income households are not the
primary market for new vehicles. Figure 4.23 shows annual expenditures on new vehicles for
lower-income households, as well as for higher-income households; it also includes median
before-tax income. Lower-income households spend far less on vehicles than do higher-income
households. For example, in 2015, lower-income households on average spent $911 on new
vehicles, while higher-income households spent more than 3 times as much, $3,009. Greene and
Welch (2016), using income quintiles, find similarly that lower-income households spend less on
new and used vehicles than higher-income households.1
3,500 53,000
k 3,009 52,000
___ 3,000 ^ W 2,781 _
vv X - 2,668 2,551 ¦ 51,000
g 2,500 7 71[.7 2,309 III Jk 50,000
or 2,000 \ I I 1 | 49,000 o
I—
T3
C
48,000
% 1'500 I I I I I I I I 47,000 |
q_ 1,000 74(| 77M | | | ^ c:).| 46,000 _E
45,000
500 44,000
0 — ™ 43,000
2007 2008 2009 2010 2011 2012 2013 2014 2015
Annual Expenditure on New Vehicles (Lower Income)
Annual Expenditure on New Vehicles (Higher Income)
Weighted Median Income Before Tax
Figure 4.23 Median Income and Annual Expenditure on New Vehicles for Lower and Higher Income
Households
Figure 4.24 shows the proportion of lower- and higher-income households that bought
vehicles. A small proportion of households buy a vehicle, either new or used, in any one year.
For instance, in 2015, 0.8 percent of lower-income households bought a new vehicle, and about
3.3 percent bought a used vehicle. About 2.4 percent of higher-income households bought a new
vehicle, and about 4.6 percent of them bought used vehicles. While a higher proportion of both
income groups buy used vehicles than buy new vehicles, lower-income households buy fewer of
T Greene, David, and Jilleah Welch (2016). "The Impact of Increased Fuel Economy for Light-Duty Vehicles on the
Distribution of Income in the United States." University of Tennessee Baker Center Report 5:16, Docket EPA-
HQ-OAR-2015-0827-4311.
4-46
-------
Consumer Issues
both. Perhaps worth noting in this chart is that the proportion of households buying vehicles,
either new or used, has increased, albeit slightly, since 2012, when the National Program began.
As with sales, discussed in Section B.1.3 of the Proposed Determination Appendix, this increase
is likely to be due more to economic recovery than to the National Program.
5.0%
4.0%
3.0%
2.0%
1.0%
r'
f
1
i
/
•
I
¦
0.0%
2007 2008
2009 2010 2011 2012 2013 2014 2015
'% of Lower-Income Households Buying New Vehicles
% of Lower-Income Households Buying Used Vehicles
'% of Higher-Income Households Buying New Vehicles
— —
% of Higher-Income Households Buying Used Vehicles
Figure 4.24 Percentage of Lower-Income and Higher-Income Households Buying New and Used Vehicles
Figure 4.25 compares annual expenditures on new vehicles, used vehicles, and fuel for lower-
income households in Panel A, and higher-income households in Panel B, from the CES data. As
Consumer Federation of America has pointed out, lower-income households spend more on
gasoline than they do on either new or used vehicles, and they spend more on used vehicles than
they do on new vehicles. As the figure shows, higher-income households spend more on new
than on used vehicles; in 2015, their expenditures on fuel approximately equaled expenditures on
new and used vehicles. In addition, household expenditures on gasoline and motor oil fluctuates
more than its expenditures on new and used vehicles. This suggests that households may face
more uncertainty due to changes in fuel prices than they do due to changes in vehicle prices.
Greene and Welch estimate that increased fuel economy decreased fuel expenditures by about 30
percent between 1980 and 2014, with most of that reduction before the mid-1990s; they attribute
almost flat expenditures since then to the increase in the proportion of light trucks over time.60
They observe that lower-income households lag behind higher-income households in getting
these reductions, because it takes time for the more efficient vehicles to become part of the used
vehicle market. They also estimate that used vehicle prices decrease faster than vehicle VMT, so
that the payback period for used vehicles should decrease as vehicle age.
4-47
-------
Consumer Issues
(A)
2,500
2,000
v? 1,500
Ln
T—I
1,000
500
0
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Annual Expenditure on New Vehicles (Lower Income)
Annual Expenditure on Used Vehicles (Lower Income)
Annual Expenditure on Gasoline and Motor Oil (Lower Income)
(B)
4,500 |
4,000
3,500
3,000
2 2,000
IN
1,500
1,000
500
0
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Annual Expenditure on New Vehicles (Higher Income)
Annual Expenditure on Used Vehicles (Higher Income)
Annual Expenditure on Gasoline and Motor Oil (Higher Income)
Figure 4.25 Annual Expenditure on Vehicles and Gasoline for Lower-Income Households (A) and Higher-
Income Households (B)
These data suggest that lower-income households are more affected by the impact of the
standards on the used vehicle market than on the new vehicle market.
4.3.3.3 Effect of the Standards on the Used Vehicle Market
The effects of the standards on lower-income households depends on its impacts, not only in
the new vehicle market, but also in the used vehicle market. The effect of the standards on the
used vehicle market will be related to the effects of the standards on new vehicle prices, the fuel
efficiency of new vehicle models, the fuel efficiency of used vehicles, and the total sales of new
vehicles. If the consumer value of fuel savings resulting from improved fuel efficiency
4-48
-------
Consumer Issues
outweighs the average increase in new models' prices to potential buyers of new vehicles, sales
of new vehicles could rise, and the used vehicle market may increase in volume as new vehicle
buyers sell their older vehicles. In this case, lower-income households are likely to benefit from
the increased availability of used vehicles. However, if potential buyers value future fuel savings
resulting from the increased fuel efficiency of new models at less than the increase in their
average selling price, sales of new vehicles could decline, and the used vehicle market may
decrease in volume as people hold onto their vehicles longer. In this case, lower-income
households could face increased costs due to reduced availability of used vehicles.
Figure 4.26 presents data from the Consumer Price Index for used cars and trucks and new
vehicles.11 Each series has been adjusted to a year 2015 reference case with underlying prices in
2015$ so that numbers on the j'-axis represent the percentage difference from price levels in
2015. Prices of used cars and trucks have decreased since 1995, and have varied in a small range
between 2008 and 2015. As can be seen, the used cars and trucks price index closely follows the
new vehicles price index, although used cars and trucks prices have a bit more volatility across
all years. It is difficult, if not impossible, to estimate what prices for used cars and trucks would
have been in the absence of the standards. These trends are likely to be affected by the increased
durability of vehicles and the recession. As with the effects of the standards on new vehicle sales,
it is possible that the GHG/fuel economy standards have had some influence on these trends, but
their effect is likely swamped by the effects of the economic recovery.
u The Consumer Price Index computes the average change in prices over time for a "market basket" of consumer
goods and services. Both the used cars and trucks index as well as the new vehicles index are components of the
private transportation index, and also feed into the transportation group of the CPI. To construct the used cars and
trucks index, BLS obtains price data from the National Automobile Dealers Association Official Used Car Guide,
and then adjusts for both quality and depreciation. The new vehicles index uses price information from BLS
surveys of dealerships and is also adjusted for quality. See http://www.bls.gov/cpi/cpifacuv.htm and
fattp://www.bis.gov/cpi/cpifactxv.litm for more information.
4-49
-------
Consumer Issues
165
155
145
135
125
115
105
95
85
Used Cars and Trucks CPI (Urban Consumers) New Vehicles CPI (Urban Consumers)
75
J? cP° <5^ & <£>* 4* $ ^ <5^ o? ^ o?
V V V V V V Ik "F 'XT v IT V V V "w "V "w If
Figure 4.26 Used and New Vehicle Consumer Price Index, 2015 = 100 (2015$)
4.3.3.4 Effects on Access to Credit
Another question is whether higher vehicle prices may be excluding some prospective
consumers from the new vehicle market through effects on consumers' ability to finance
vehicles. It is possible that lenders focus solely on the amount of the vehicle loan, the person's
current debt, and the person's income when issuing loans, and not the costs associated with fuel
consumption. If lenders in fact restrict themselves to consideration of only those three factors,
and fuel savings are not factored in to counter-balance this cost, then the higher up-front costs of
new vehicles subject to the standards would reduce buyers' ability to get loans. This may occur
even though, as discussed in Proposed Determination Appendix Section C.2.4, the fuel savings
exceed the increased loan payments and other costs in the first year of loans with 5 or more year
duration. Thus, if lenders do not take fuel savings into account in providing loans, households
that are borrowing near the limit of their abilities to borrow will either have to change what
vehicles they buy, or not buy vehicles at all.
On the other hand, some evidence suggests that the loan market may evolve to take fuel
savings into greater account in the lending decision. Market innovation suggests that parts of the
loan market take fuel savings into account in the lending decision. Some lenders currently give
discounts for loans to purchase more fuel-efficient vehicles.61 An internet search on the term
"green auto loan" produced more than 60 lending institutions that provide reduced loan rates for
more fuel-efficient vehicles.62 A third of credit unions responding to a recent survey offered
some type of green auto loan.63 In a survey of nine credit unions, the ratio of the dollar value of
green loans to total loans varied between 0.09 and 33.89 percent.64 It is possible that the auto
4-50
-------
Consumer Issues
loan market may evolve to include further consideration of fuel savings, as those savings are a
significant factor in offsetting the increase in up-front costs of vehicles.
Next, we examine the question of whether the debt-to-income ratio (DTI) is an impassible
obstacle for lending, because of the importance of the DTI in determining access to credit. The
analysis that follows is based on guidance from several online sources stating that most lenders
avoid giving loans to consumers who have over 36 percent DTI.65 We use CES data pooled
across 2007-2015 to examine households with over 36 percent DTI in order to gauge whether
exceeding this threshold may preclude households from being able to finance a vehicle purchase.
The components included in our DTI calculation are derived from those same online sources
cited above (Bankrate.com, Zillow.com, and TheNest.com). These components are mortgage
payments, home equity loan payments, monthly rent, other vehicle payments, child support, and
alimony
The results in Table 4.6 show that, from 2007 to 2015, 28 percent of lower-income
households and 7 percent of higher-income households who both had a DTI of over 36 percent
and purchased at least one new vehicle financed their vehicle purchases. The results are similar
using a 40 percent DTI, the threshold used in an analysis by Wagner et al. (2012), as reported in
Table 4.7.66 This suggests that the DTI is not an inflexible barrier. Thus, if increases in vehicle
prices push some households over the 36 or 40 percent DTI, it nevertheless may be possible for
them to get loans.
Table 4.6 Breakdown of Households That Bought at Least One New Vehicle By the Cutoff of DTI Ratio
36%, 2007-2015
Lower Income
Higher Income
< or equal to 36% DTI
72%
93%
>36% DTI
28%
7%
Table 4.7 Breakdown of Households That Bought At Least One New Vehicle by the Cutoff of DTI Ratio
40%, 2007-2015
Lower Income
Higher Income
< or equal to 40% DTI
76%
95%
>40% DTI
24%
5%
In addition, we look at the trends in percentage of lower-income and higher-income
households who had DTI ratios larger than 36 percent and were able to purchase at least one new
vehicle with an auto loan. As shown in Figure 4.27, while lower-income households with higher
DTI ratios have been able to get loans to buy new vehicles through the years, the percentage of
lower-income households who got the loans varies more than that of higher-income households.
It is worth noting that other factors, such as interest rates and lending policies of financial
institutions, also affect the credit-worthiness of households. EPA does not expect the standards to
have any measurable effect on interest rates, which are determined primarily by broader
macroeconomic factors.
4-51
-------
Consumer Issues
40
35
30
25
20
15
10
5
0
> % of Higher-Income Households with DTI Greaterthan 36%
• % of Lower-Income Households with DTI Greaterthan 36%
Figure 4.27 Percentage of Households Buying at Least One New Vehicle with Finance who had Debt-to-
income (DTI) Ratio Greater than 36 Percent
4.3.3.5 Effects on Low-Priced Vehicles
Low-priced vehicles may be considered an entry point for people into buying new vehicles
instead of used ones; automakers may seek to entice people to buy new vehicles through a low
price point. Commenters have expressed concern that higher costs associated with the standards
could affect the ability of automakers to maintain vehicles in this segment.
The cutoff for a car to be "lower-priced" is a matter of opinion. We searched the web for
definitions. CNN Money, in 2003, defined a "cheap" car as one with a price less than $12,500
($15,900 in 2015$).67 Motor Trend (2015) and Auto Bytel (2015)defined the lowest market
segment as those $15,000 or less.68 U.S. News and World Report (2015) considered "affordable"
cars to be priced from under $20,000 to under $40,000.69 Consumer Reports (2015) used the
cutoff of $25,000 to characterize cars in the lower priced market segment.70 Of websites that
mention or rank affordable or low-priced vehicles, the highest price in the "affordable" category
varies. For example, for 2014 and 2015, we found highest-priced models of $14,845.00
(Autobytel 2015), $14,850 (Lloyd-Miller, 2015), and $19,890 (Notte, 2014).71 Based on this
review, we use a cutoff of $15,000 (2015$) to identify low-priced vehicles.
We use Ward's Automotive data for U.S. cars for the years 2007-2015 to examine the impacts
of the standards on the costs of the low-priced segment of the market.72 Figure 4.28 shows the
number of models available for less than $15,000 (2015$). The number of available low-priced
models available has ranged from 8 to 18 model trims, with 13 trims available in 2015.
Automakers appear to be able to provide low-priced vehicles; this graph does not indicate
whether it has become more challenging to do so.
4-52
-------
Consumer Issues
t r
2005 2010 2015
Model Year
Figure 4.28 Number of <$15,000 (2015$) Vehicle Model Trims Available
Figure 4.29 shows the minimum MSRP (in 2015$) for all new vehicles over time. It indicates
that the least costly (always cars) have become more expensive since 2001. This finding suggests
that these vehicles may be becoming more costly to produce, though it leaves open the question
of why.
We next sought to understand whether quality increases might affect these price changes.
Table 4.8 shows, as an example, the features of the Nissan Versa over time. The Nissan Versa
was chosen since it was the lowest-priced vehicle in 2016 (according to the MSRP of the base
model sedan) and in 6 of the 9 years examined.v The MSRP data are from Ward's and are in
2015$;73 all other data are from Edmunds.com.74 Some content has increased over time, such as
audio controls on the steering wheel and the auxiliary audio input. In contrast, the horsepower
decreased between MY2008 and 2009. In constant dollars, the MSRP of the Nissan Versa is
lower now than in 2007, though it has increased since its minimum value in MY2011.
v For MYs 2007 and 2008, the Chevrolet Aveo, and the Hyundai Accent Blue for MY2010, have lower MSRPs than
the Versa, while not having more content.
4-53
-------
Consumer Issues
Table 4.8 Features of the Nissan Versa over Time, Base Model (Edmund's and Ward's Automotive)
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
4-wheel ABS
X
X
X
X
X
Emergency
Braking Assist
X
X
X
X
X
Stability
Control
X
X
X
X
X
Traction
X
Control
X
X
X
X
Auxiliary
Audio Input
X
X
X
X
X
Bluetooth
X
X
Wireless
Datalink for
Hands-free
Phone
Audio Controls
X
X
on Steering
Wheel
Speed
Sensitive
X
X
Volume
Control
Air
X
Conditioning
X
X
X
X
X
X
X
X
X
122 hp
122 hp
107 hp
107 hp
107 hp
109 hp
109 hp
109 hp
109 hp
109 hp
5200
5200
6000
6000
6000
6000
6000
6000
6000
6000
Horsepower
rpm
rpm
rpm
rpm
rpm
rpm
rpm
rpm
rpm
rpm
MSRP (2015$)
14746
14660
11730
11614
11415
12270
13149
13169
12938
12962
In the past, not only was the low-priced vehicle segment a way to encourage first-time new
vehicle purchasers, but it also tended to include more fuel-efficient vehicles that assisted
automakers in achieving CAFE standards.75 The footprint-based standards, by encouraging
improvements in GHG emissions and fuel economy across the vehicle fleet, reduce the need for
low-priced vehicles to be a primary means of compliance with the standards. This change in
incentives for the marketing of this segment may contribute to the increases in the prices of
vehicles previously in this category. In addition, as seen with the Versa example above, these
vehicles may be gaining more content, such as improved entertainment systems and electric
windows, if they develop an identity as a desirable market segment. For instance, the Nissan
Versa, the lowest-priced vehicle since MY2011, added Bluetooth, audio controls on the steering
wheel, and speed-sensitive volume control in MY2015. It may be that the small, fuel-efficient
vehicles previously sold with low prices are evolving to fit consumer demand that prefers content
to low prices.
4-54
-------
Consumer Issues
In sum, the low-priced vehicle segment still exists. Whether it continues to exist, and in what
form, may depend on the marketing plans of manufacturers: whether benefits are greater from
offering basic new vehicles to first-time new-vehicle buyers, or from making small vehicles
more attractive by adding more desirable features to them.
2000
t r
2005 2010
Model Year
2015
Figure 4.29 Minimum MSRP of All Car Models Available
4.3.4 Conclusion
It is difficult to determine how the LDV GHG standards have affected vehicle affordability
thus far, due to both challenges in defining affordability, and difficulties in separating the effects
of the standards from other market changes. Because lower-income households are most likely to
buy used vehicles, the effects of the standards on lower-income households depend mostly on
their effects on used vehicles. In the used-vehicle market, prices have not shown marked
increases; the trend appears to be flat or decreasing. The effects of the standards on access to
credit may not be large: there continue to be loan discounts for fuel-efficient vehicles, and many
people, including lower-income people, with high debt-to-income ratios appear able to get loans.
The low-priced vehicle segment still exists, perhaps in changing form, as it appears that
manufacturers are improving the content features in this segment. In sum, if the standards thus
far have affected vehicle affordability, they have not had significant visible effects. In addition,
there appear to be market adjustments, such as ongoing changes in the finance market, that may
mitigate some of any adverse effects. In the MY2022-2025 time frame, the primary effects on
affordability of vehicle sales are still likely to be due to broader macroeconomic factors, such as
economic activity and overall employment; any impacts of the standards are likely to be
secondary to those broader economic factors.
4-55
-------
Consumer Issues
This assessment has focused on the effects of the standards on purchase affordability of
vehicles-that is, whether they become more difficult to purchase because of the increase in up-
front costs. The vehicles will also become less expensive to operate, due to fuel savings from
more fuel-efficient technologies. The reduced operating costs from fuel savings over time are
still expected to exceed the increase in up-front vehicle costs, as discussed further in Section
C.2.4 of the Proposed Determination Appendix, as a further mitigation of any effects on vehicle
affordability.
4-56
-------
Consumer Issues
REFERENCES
1 77 Federal Register 62843-62844 (October 15, 2012) and Regulatory Impact Analysis, pages 3-18 to 3-23.
2 National Research Council. (2015) Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles. The National Academies Press, pp. S-9, 9-2, 10-29, 10-38.
3 National Research Council. (2015) Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles. The National Academies Press, pp. S-9, 9-2, 10-29, 10-38.
4 Knittel, C. R. (2011). "Automobiles on Steroids: Product Attribute Trade-Offs and Technological Progress in the
Automobile Sector." American Economic Review 101(7): pp. 3368-3399, Docket EPA-HQ-OAR-2010-0799-11946;
Klier, T. and Linn, J. (2016). "The Effect of Vehicle Fuel Economy Standards on Technology Adoption." Journal of
Public Economics 133: 41-63, Docket EPA-HQ-OAR-2015-0827-0142; McKenzie, D. andHeywood, J. B. (2015).
"Quantifying efficiency technology improvements in U.S. cars from 1975-2009." Applied Energy 157: 918-928,
Docket EPA-HQ-OAR-2015-0827-0147; Wang, Y. (2016). "The Impact of CAFE Standards on Automobile
Innovation in the US." Working paper, Docket EPA-HQ-OAR-2015-0827-0133.
5 Cooke, Dave (2016). "The Trade-off between Fuel Economy and Performance: Implications for the Mid-term
Evaluation of the National Program." Union of Concerned Scientists working paper.
6 Knittel, C. R. (2011). "Automobiles on Steroids: Product Attribute Trade-Offs and Technological Progress in the
Automobile Sector." American Economic Review 101(7): pp. 3368-3399, Docket EPA-HQ-OAR-2010-0799-11946;
Klier, T. and Linn, J. (2016). "The Effect of Vehicle Fuel Economy Standards on Technology Adoption." Journal of
Public Economics 133: 41-63; McKenzie, D. andHeywood, J. B. (2015). "Quantifying efficiency technology
improvements in U.S. cars from 1975-2009." Applied Energy 157: 918-928; Wang, Y. (2016). "The Impact of
CAFE Standards on Automobile Innovation in the US." Working paper, Docket EPA-HQ-OAR-2015-0827-0127.
7 MacKenzie, Don, and JohnHeywood (2012). "Acceleration Performance Trends and Evolving Relationship
Between Power, Weight, and Acceleration in U.S. Light-Duty Vehicles." Transportation Research Record 2287:
122-131, Docket EPA-HQ-OAR-2015-0827-0148,.
8 Jaffe, A. B., Newell, R. G., & Stavins, R. N. (2003). Technological change and the environment. In Handbook of
environmental economics 1, ed. K.-G. Maler and J. R. Vincent. Elsevier, Docket EPA-HQ-OAR-2015-0827-0130.
9 Jaffe, A. B., Newell, R. G., & Stavins, R. N. (2003). Technological change and the environment. In Handbook of
environmental economics 1, ed. K.-G. Maler and J. R. Vincent. Elsevier, Docket EPA-HQ-OAR-2015-0827-0130.
10 Dosi, G. and Nelson, R. R. (2010). "Technical Change and Industrial Dynamics as Evolutionary Processes."
Handbook of Econometrics 1: pp. 52-114.
11 Abernathy, W. J., and Utterback, J. M. (1978). Patterns of industrial innovation. Technology review, 80, pp. 254-
228; Henderson, R. M., and Clark, K. B. (1990). "Architectural innovation: The reconfiguration of existing product
technologies and the failure of established firms." Administrative science quarterly, 9-30.
12 Zhao, H. (Ed.). (2009). Advanced Direct Lnjection Combustion Engine Technologies and Development: Gasoline
and Gas Engines (Vol. 1). ISBN 9781845697327. Elsevier, pp. 1-3.
13 U.S. EPA. (2016). Report Tables for Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel
Economy Trends: 1975 through 2016, EPA Report 420-R-16010-report-tables.xlsx, Table 5.3.1,
https://www.epa.gov/fiiei-economv/download-co2-and-fiiei-economY-frends-reDort-1975-2016, accessed 11/8/2016.
There are similar examples from the stationary source world. See 80 FR at 64575 (Oct. 23, 2015) and Laney, Mike
(2015). " History of Flue Gas Desulfurization in the United States- 1970-1976," memo to Rulemaking Docket ID
EPA-HQ-OAR-2013-0495-11774, demonstrating the role of regulatory standards in the development and
deployment of scrubber technology.
14 Blumstein, Carl and Margaret Taylor (2013). "Rethinking the Energy-Efficiency Gap: Producers, Intermediaries,
and Innovation," Energy Institute at Haas Working Paper 243, University of California at Berkeley, Docket EPA-
HQ-OAR-2014-0827-0075; Tirole, Jean (1998). The Theory of Industrial Organization Cambridge, MA: MIT
Press, pp. 400, 402, Docket EPA-HQ-OAR-2014-0827-0089.
15 Gal-Or, E. (1987). "First Mover Disadvantages with Private Information." The Review of Economic Studies 54(2):
pp. 279-292; Liu, Z. (2005). "Stackelberg Leadership with Demand Uncertainty." Managerial and Decision
Economics 26: pp. 345-350.
16 Popp, D., Newell, R. G., and Jaffe, A. B. (2010). "Energy, the environment and technological change." In
Handbook of the economics of innovation 2, ed. B. H. Hall, and N. Rosenberg. Elsevier.
4-57
-------
Consumer Issues
17 Holmes, T. J., Levine, D. K., and Schmitz, J. A., Jr. (2012). "Monopoly and the Incentive to Innovate When
Adoption Involves Switchover Disruptions." American Economic Journal: Microeconomics 4(3): pp. 1-33.
18 National Research Council. (2015) Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles. The National Academies Press, pp. S-9, 9-2, 10-29, 10-38; Klier, T. and Linn, J. (2016). "The Effect
of Vehicle Fuel Economy Standards on Technology Adoption." Journal of Public Economics 133: 41-63; Wang, Y.
(2016). "The Impact of CAFE Standards on Automobile Innovation in the US." Working paper.
19 Boncimino, A. (April 10, 2015). "Auto industry rushes to meet 2025 fuel efficiency guidelines." Upstate Business
Journal, http://upstatebusinessiournal.com/news/auto-industrv-rushes-meet-2025-fuei-efficiency-guideiines/;
Buchholz, K. (March 13, 2015). "New technology for exhausting jobs." Automotive Engineering Magazine.
http://articles.sae.org/13974/; Truett, Richard (November 15, 2016). "With nod to EPA, today's cars are faster, more
powerful and more fun to drive." Automotive News,
http://www.antonews.com/article/20 BLOG06/.1.6.1..1..1.9909/with-nod-to-epa-todavs-cars-are-faster-more-
powerful-and-mo re-fun-to.
20 Powell, W. W., and Giannella, E. (2010). "Collective Invention and Inventor Networks," In Handbook of the
Economics of Innovation, Volume 1. Ed. B. Hall and N. Rosenberg Elsevier.
21 Greene, David, and Jin-Tan Liu (1988). "Automotive Fuel Economy Improvements and Consumers' Surplus."
Transportation Research A 22A(3): 203-218. Docket EPA-HQ-OAR-2010-0799-0703.
22 Greene, David (2010). "How Consumers Value Fuel Economy: A Literature Review." EPA-420-R-10-008.
Docket EPA-HQ-OAR-2010-0799-0711.
23 Greene, David, Anushah Hossein, and Robert Beach (2016). "Consumer Willingness to Pay for Vehicle
Attributes: What is the Current State of Knowledge?" RTI International Final Report, Work Assignment 4-11, EPA
Contract EP-C-11-045.
24 Greene, David (2010). "How Consumers Value Fuel Economy: A Literature Review." EPA-420-R-10-008.
Docket EPA-HQ-OAR-2010-0799-0711.
25 McAlinden, Sean, et al. (2016). "The Potential Effects of the 2017-2025 EPA/NHTSA GHG/Fuel Economy
Mandates on the U.S. Economy." Center for Automotive Research
http://www.cargroup.org/?module=Pubtieations&event=View&pii , accessed 10/11/2016.
26 Isenstadt, Aaron (2016). "The latest paper by the Center for Automotive Research is not what it thinks it is."
International Council on Clean Transportation, http://www.theicct.org/blogs/staff/latest~paper~bv~CAR~is~not~what~
it~thinks~it~is , accessed 10/13/2016; Cooke, Dave (2016). "Deja vu: Shoddy Economic study Touted by Automakers
Flaunts Facts." Union of Concerned Scientists, http://btog.HCSiisa.org/dave~cooke/deia~vTi~shoddv~econoniic~stiidv~
touted-bv-aiitomakers-flaunts-facts. accessed 10/19/2016.
27 Isenstadt, Aaron (2016). "The latest paper by the Center for Automotive Research is not what it thinks it is."
International Council on Clean Transportation, http://www.theicct.org/blogs/staff/latest~paper-bv~CAR~is~not~what~
it~thinks~it~is , accessed 10/13/2016; Cooke, Dave (2016). "Deja vu: Shoddy Economic study Touted by Automakers
Flaunts Facts." Union of Concerned Scientists, http://btog.HCSnsa.org/dave~cooke/deia~vn~shoddv~economic~stndv~
touted-bv-aiitomakers-flaunts-facts. accessed 10/19/2016. Greene, David (1991). "Short-Run Pricing Strategies to
Increase Corporate Average Fuel Economy." Economic Inquiry 29(1): 101-114.
28 National Research Council. (2015) Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles. The National Academies Press.
29 Berck, Peter, and Sandra Hoffmann (2002). "Assessing the Employment Impacts of Environmental and Natural
Resource Policy." Environmental and Resource Economics 22: 133-156 (in particular, p. 145). Docket EPA-HQ-
OAR-2010-0799-0678.
30 National Research Council. (2015) Cost, Effectiveness and Deployment of Fuel Economy Technologies for Light-
Duty Vehicles. The National Academies Press, pp. 9-16, 9-36.
31 Huang, Hsing-Hsiang, and Gloria Helfand (2016). "Memorandum: Content Analysis of Auto Reviews using
MY2014 and MY2015 Data." U.S. EPA Office of Transportation and Air Quality, Memorandum to Docket.
32 Sha, Mandy, Amanda Smith, and Robert Beach (2016). "Content Analysis of Professional Automotive Reviews:
Model Year 2015, Work Assignment 4-08 Final Report." RTI International.
33 Cassidy, Alecia, Geoffrey Burmeister, and Gloria Helfand. "Impacts of the Model Year 2017-25 Light-Duty
Vehicle Greenhouse Gas Emission Standards on Vehicle Affordability." Working paper, docket EPA-HQ-OAR-
2015-0827-0401.
34 Hancock, K.E. (1993). '"Can't Pay? Won't Pay?' or Economic Principles of 'Affordability'." Urban Studies
30(1): 127—145.
35 Quigley, J.M., and Raphael, S. (2004). "Is Housing Unaffordable? Why Isn't It More Affordable?" The Journal of
Economic Perspectives 18(1): 191—214.
4-58
-------
Consumer Issues
36 Bradley, R. (2008). "Comment - Defining health insurance affordability: Unobserved heterogeneity matters."
Journal of Health Economics 27: 1129—1140.
37 Bartl, M. (2010). "The Affordability of Energy: How Much Protection for the Vulnerable Consumers?" Journal of
Consumer Policy 33: 225—245.
38 Fankhauser, S., and Tepic, S. (2007). "Can poor consumers pay for energy and water? An affordability analysis
for transition countries." Energy Policy 35: 1038-1049.
39 Blaylock, J. et al. (1999). "Economics, food choices, and nutrition." Food Policy 24: 269—286.
40 Glied, S. (2009). "Mandates and the Affordability of Health Care." Inquiry 46: 203—214.
41 Milne, C. (2000). "Affordability of basic telephone service: an income distribution approach."
Telecommunications Policy 24: 907—927.
42 Haffner, M., and Heylen, K. (2011). "User Costs and Housing Expenses: Towards a More Comprehensive
Approach to Affordability." Housing Studies 26(4): 5930—614.
43 Bramley, G. (1990). "Access, affordability and housing need." Paper presented at ESRC Housing Studies
Conference, University of Surrey, September 1990. Mimeograph, SAUS, University of Bristol. In Hancock, K.E.
(1993). 'Can Pay? Won't Pay?' or Economic Principles of 'Affordability'. Urban Studies 30(1): 127—145.
44 Whitehead, C.M.E. (1991). "From need to affordability: An analysis of UK housing objectives," Urban Studies
28(6): 871—887.
45 Hancock, K.E. (1993). "'Can't Pay? Won't Pay?' or Economic Principles of 'Affordability'." Urban Studies
30(1): 127—145.
46 Stone, M.E. (2006). "What is housing affordability? The case for the residual income approach." Housing Policy
Debate 17(1): 151—184.
47 U.S. Department of Housing and Urban Development (2015). "Affordable Housing."
http://portal.hud.gov/hudportal/HUD?src=/program_ofifices/comm_planning/affordablehousing/, accessed
8/26/2015.
48 U.S. Department of Housing and Urban Development (2011). "Worst case housing needs 2011 Report to
Congress," accessed 05/20/2014 at: http://www.hndnser.Org/Publications/pd:f/HUD-506 WorstCase2011.pdf
Steffen, B. L.; Carter, G.R.; Martin, M.; Pelletiere, D.; Vandenbroucke, D.A.; and Yao, Y.G.D. (2015). "Worst Case
Housing Needs: 2015 Report to Congress." Department of Housing and Urban Development. Accessed 8/24/2015 at
http://www.hndnser.org/portal/pnblications/afflisg/wc HsgNeedsl5.html
49 Bogdon, A.S., and Can, A. (1997). "Indicators of local housing affordability: Comparative and spatial
approaches." Real Estate Economics 25(1): 43—80.
50 Gan, Q., and Hill, R.J. (2009). "Measuring housing affordability: Looking beyond the median." Journal of
Housing Economics 18: 115—125.
51 Fisher, L.M., Pollakowski, H.O., and Zabel, J. (2009). "Amenity-based housing affordability indexes." Real
Estate Economics 37: 705—746.
52 Lerman, D.L., and Reeder, W.J. (1987). "The affordability of adequate housing." AREUEA Journal 15(4): 389—
404.
53 Thalmann, P. (1999). "Identifying households which need housing assistance." Urban Studies 36(11): 1933—
1947.
54 Thakuriah, P., and Liao, Y. (2006). "Transportation expenditures and ability to pay." Transportation Research
Record: Journal of the Transportation Research Board 1985: 257—265.
55 Bundorf, M.K., and Pauly, M.V. (2009). "Reply to Ralph Bradley, 'Comment - Defining health insurance
affordability: Unobserved heterogeneity matters'." Journal of Health Economics 28: 251—254.
56 Schanzenbach, D.W., and McGranahan, L. (2012). "The impact of transportation on affordability: An analysis of
auto cost white paper." Manhattan Strategy Group, accessed 06/09/2014 at:
http://www.tocationaffordabilitv.info/downloads/AHtoCostResearch.pdf.
57 U.S. Bureau of Labor Statistics. "Consumer Expenditure Survey," accessed 8/25/2015 at:
http://www.bls. gov/cex/.
58 Goldberg, P.K. (1996). "Dealer price discrimination in new car purchases: Evidence from the Consumer
Expenditure Survey." Journal of Political Economy 104(3): 622—654.
59 Yurko, A. V. (2011). "Heterogeneous Consumers: How the demand affects outcomes in vertically differentiated
markets." Dissertation, University of Texas at Austin.
http://www.lib.ntexas.edn/etd/d/2008/vnrkod66028/vnrkod66028.pdf.
60 Greene, David, and Jilleah Welch (2016). "The Impact of Increased Fuel Economy for Light-Duty Vehicles on the
Distribution of Income in the United States." University of Tennessee Baker Center Report 5:16, Docket EPA-HQ-
OAR-2015-0827-4311.
4-59
-------
Consumer Issues
61 Ladika, S. (2009). "Green auto loans offer lower rates," Bankrate.com, accessed 7/29/2015 at:
http://www.bankrate.com/fniance/anto/green-auto-loa.ns-offer-lower-rates-l.aspx. Docket EPA-HQ-OAR-2010-
0799-11829.
62 Huang, Hsing-Hsiang, and Gloria Helfand (2016, November). "Memorandum: Lending Institutions that Provide
Discounts for more Fuel-Efficient Vehicles." U.S. EPA Office of Transportation and Air Quality, Memorandum to
Docket.
63 Baumhefner, M. (2013). "Why Can't Your Loan be as Green and Efficient as Your Vehicle?" Natural Resources
Defense Council Website, accessed 7/29/2015 at:
http://switchboard.nrdc.org/blogs/mbaumhefner/why_cant_your_loan_be_as_green.htmi.
64 Hall, W. R. (2011). "Finding Sustainable Profits: Green Lending in Credit Unions," Filene Research Institute,
accessed 8/25/2015 at: https://filene.Org/assets/pd:f-reports/249 Hail Finding Sustainable Profits.pdf.
65 Bankrate (2015). "Debt-to-income ratio calculator." Bankrate.com, accessed 8/25/2015 at:
http://www.bankrate.com/calcnlators/mortgages/ratio-debt-calculator.aspx : Zillow (2015). "Debt-to-income
calculator," Zillow.com, accessed 8/25/2015 at: http://www.ziUow.com/mortgage-calcniator/debt-to-income-
calculator/; Keythman, B. (2015). "What is the 28-36 Rule of Debt Ratio?" Demand Media, accessed 8/25/2015 at:
http://budgeting.thenest.com/28-36-rule-debt-ratio-22412.html.
66 Wagner, D., P. Nusinovich, and E. Plaza-Jennings, National Automobile Dealers Association (February 13, 2012).
' 'The Effect of Proposed MY2017-2025 Corporate Average Fuel Economy (CAFE) Standards on the New Vehicle
Market Population."Docket EPA-HQ-OAR-2010-0799-9575,
http://www.nadafrontpage.com/Hpload/wvsiwvg/The%20Effect%20of%20Proposed%20MY%202Q17-
2025%20CAFE%20Standa.rds%20on%20New-Vehicle%20Market.pdf. accessed 10/27/2016.
67 Isidore. C. (2003). "Cheap cars? Guess who sells the most." CNN Money, accessed 8/25/20.1.5 at:
http://monev.cnn.eom/2.003/ll/07/pf/antos/cheap cars/.
68 Motor Trend (2015). "New Cars," Motor Trend.com, accessed 8/25/2015 at:
http://www.motortrend.com/new cars/ : Autobvtel (2015). "Top 10 Least Expensive Coupes and Compact Cars."
accessed 8/25/2015 at: http://www.autobytel.com/top-10-cars/least-expensive-cars/coupes/.
69 U.S. News and World Report (2015). "Best Car Rankings," accessed 8/25/2015 at:
http://usnews.rankingsandreviews.com/cars-trucks/rankings/cars/.
70 Consumer Reports (2015). "Best new cars for under $25,000," accessed 8/25/2015 at:
http://www.consHmerreports.org/cro/2012/12/best-new-cars-for-iinder-25-000/index.htm
71 Autobytel (2015). "Top 10 Least Expensive Coupes and Compact Cars." accessed 8/25/2015 at:
http://www.autobytel.com/top-10-cars/least-expensive-cars/coupes/; Lloyd-Miller, J. (2015). "10 of the Most
Affordable Cars for 2015," Cheatsheet, accessed 8/25/2015 at: http://www.cheatsheet.com/automobiles/10-of-the-
most-affordable-cars-for-20.1.5.litml/?a=viewall: Notte. J. (2014). "10 Most Affordable Cars Of 2014." The Street,
accessed 8/25/2015 at: http://wwwihestreet.eom/stoiv/l.282036.l./.l./10-most-affordable-cars-of-2014.html.
72 Ward's Automotive Group (2007-2015). "Model U.S. Car and Light Truck Specifications and Prices."
UsaVs01_2007.xls throughUsaVx01_2015.xls.
73 Ward's Automotive Group (2007-2015). "Model U.S. Car and Light Truck Specifications and Prices."
UsaVs01_2007.xls throughUsaVx01_2015.xls.
74 Edmund's (2015). "2007 Nissan Versa - Features & Specs," accessed 8/25/2015 at:
http://www.edmnnds.com/nissati/versa/2007/st-100840658/featHres-specs/: "2008 Nissan Versa - Features &
Specs," accessed 8/25/2015 at: http://www.edmnnds.com/nissan/versa/2008/featii.res-specs/: "2009 Nissan Versa -
Features & Specs,." accessed 8/25/2015 at: http://www.edmniids.com/nissaii/versa/2009/featiires-specs/; "2010
Nissan Versa - Features & Specs," accessed 8/25/2015 at: http://www.edmnnds.com/nissan/versa/2010/featnres-
specs/ ; "2011 Nissan Versa - Features & Specs," accessed 8/25/2015 at:
http://www.edmnnds.com/nissan/versa/2011/featnres-specs/; "2012 Nissan Versa - Features & Specs," accessed
8/25/2015 at: http://www.edmnnds.com/nissan/versa/2012/featnres-specs/; "2013 Nissan Versa - Features &
Specs," accessed 8/25/2015 at: http://www.edmnnds.com/nissan/versa/2013/featnres-specs/; "2014 Nissan Versa -
Features & Specs," accessed 8/25/2015 at: hj&L//wwj^^ ; "2015
Nissan Versa - Features & Specs," accessed 8/25/2015 at: http://www.edmunds.com/nissan/versa/2015/features-
specs/ ; "2016 Nissan Versa Sedan - Features & Specs," accessed 11/1/2016 at
http://www.edmunds.com/nissan/versa/2016/sedan/features-specs/.
75 See, for example, Austin, David, and Terry Dinan (2005). "Clearing the Air: The Costs and Consequences of
Higher CAFE Standards and Increased Gasoline." Journal of Environmental Economics and Management 50(3):
562—82; and Kleit, Andrew N. (2004). "Impacts of Long-Range Increases in the Fuel Economy (CAFE) Standard."
Economic Inquiry 42(2): 279—294.
4-60
-------
EPA's OMEGA Model
Table of Contents
Chapter 5: EPA's OMEGA Model 5-1
5.1 OMEGA Overvi ew 5-1
5.2 OMEGA Model Structure 5-3
5.3 OMEGA Pre-Processors, Vehicle Types & Packages 5-5
5.3.1 Vehicle Types 5-5
5.3.2 Technology Packages, Package Building & Master-sets 5-7
5.3.3 Master-set Ranking and the Technology Input File 5-13
5.3.4 Applying Ranked-sets of Packages to the Projected Fleet 5-17
5.3.5 New to OMEGA since the Draft TAR 5-18
Table of Figures
Figure 5.1 Information Flow in the OMEGA Model 5-4
Table of Tables
Table 5.1 Vehicle Types and Example Models 5-7
Table 5.2 Penetration Caps used in the OMEGA Central Analysis Runs 5-15
Table 5.3 Lifetime VMT & Baseline CO2 used for the TARF Ranking Process 5-16
-------
EPA's OMEGA Model
Chapter 5: EPA's OMEGA Model
Applying technologies efficiently to the wide range of vehicles produced by various
manufacturers is a challenging task. In order to assist in this task, EPA uses a computerized
program called the Optimization Model for reducing Emissions of Greenhouse gases from
Automobiles (OMEGA). Broadly, OMEGA starts with a description of the future vehicle fleet,
including manufacturer, sales, base CO2 emissions, footprint and the extent to which emission
control technologies are already employed. For the purpose of this analysis, EPA uses OMEGA
to analyze over 200 vehicle platforms which encompass approximately 1,300 vehicle models in
order to capture the important differences in vehicle and engine design and utility of future
vehicle sales of roughly 15-17 million units annually in the 2021-2025 timeframe.A The model
is then provided with a list of technologies which are applicable to various types of vehicles,
along with the technologies' cost and effectiveness and the percentage of vehicle sales which can
receive each technology during the redesign cycle of interest. The model combines this
information with economic parameters, such as fuel prices and a discount rate, to project how
various manufacturers would apply the available technology in order to meet increasing levels of
emission control. The result is a description of which technologies are added to each vehicle
platform, along with the resulting cost. The model can also be set to account for various types of
compliance flexibilities.6
EPA has described OMEGA's specific methodologies and algorithms previously in the model
documentation,1 the version of the model used for both the Proposed Determination and the
Draft TAR is publically available on the EPA website at https://www.epa.gov/regulations-
emissions-vehicles-and-engines/optimization-model-reducing-emissions-greenhouse-gases, and
it has been peer reviewed.2
5.1 OMEGA Overview
The OMEGA model evaluates the relative cost and effectiveness of available technologies
and applies them to a defined vehicle fleet in order to meet a specified GHG emission target.
Once the regulatory target (whether the target adopted in the rule, or an alternative target) has
been met, OMEGA reports out the cost and societal benefits of doing so. The model is written in
the C# programming language, however both inputs to and outputs from the model are provided
using spreadsheet and text files. The output files facilitate additional manipulation of the results,
as discussed in the next section.
OMEGA is primarily an accounting model. It is not a vehicle simulation model, where basic
information about a vehicle, such as its mass, aerodynamic drag, an engine map, etc. are used to
A EPA's analysis fleet actually contains roughly 2,200 vehicle models, but many of those are the result of very
minor differences in footprint and not truly different models.
B While OMEGA can apply technologies which reduce CO2 efficiency related emissions and refrigerant leakage
emissions associated with air conditioner use, this task is currently handled outside of the OMEGA core model.
A/C improvements are highly cost-effective, and would always be added to vehicles by the model, thus they are
simply added into the results at the projected penetration levels.
5-1
-------
EPA's OMEGA Model
predict fuel consumption or CO2 emissions over a defined driving cycle.c Although OMEGA
incorporates functions which generally minimize the cost of meeting a specified CO2 target, it is
not an economic simulation model which adjusts vehicle sales in response to the cost of the
technology added to each vehicle.0
OMEGA can be used to model either a single vehicle model or any number of vehicle models.
Vehicles can be those of specific manufacturers as in this analysis or generic fleet-average
vehicles as in the 2010 Joint Technical Assessment Report supporting the MY 2017-2025 NOI.
Because OMEGA is an accounting model, the vehicles can be described using a relatively few
number of terms. The most important of these terms are the vehicle's baseline CO2 emission
level, the level of CO2 reducing technology already present, and the vehicle's "type," which
indicates the technology available for addition to that vehicle to reduce CO2 emissions.
Information determining the applicable CO2 emission target for the vehicle must also be
provided. This may simply be vehicle class (car or truck) or it may also include other vehicle
attributes, such as footprint.E In the case of this analysis, as in the Draft TAR, footprint and
vehicle class are the relevant attributes.
Emission control technology can be applied individually or in groups, often called technology
"packages," as discusses above. The OMEGA user specifies the cost and effectiveness of each
technology or package for a specific "vehicle type," such as midsize cars with V6 engines or
minivans. The user can limit the application of a specific technology to a specified percentage of
each vehicle's sales (i.e., a "maximum penetration cap"), which for this analysis, are specified a
priori by EPA. The effectiveness, cost, application limits of each technology package can also
vary over time.F A list of technologies or packages is provided to OMEGA for each vehicle
type, providing the connection to the specific vehicles being modeled.
OMEGA is designed to apply technology in a manner similar to the way that a vehicle
manufacturer might make such decisions. In general, the model considers three factors which
EPA believes are important to the manufacturer: 1) the cost of the technology, 2) the value which
the consumer is likely to place on improved fuel economy and 3) the degree to which the
technology moves the manufacturer towards achieving its fleetwide CO2 emission target.
Technology can be added to individual vehicles using one of three distinct ranking
approaches. Within a vehicle type, the order of technology packages is set by the OMEGA user.
The model then applies technology to the vehicle with the lowest Technology Application
Ranking Factor (hereafter referred to as the TARF). OMEGA offers several different options for
calculating TARF values. One TARF equation considers only the cost of the technology and the
value of any reduced fuel consumption considered by the vehicle purchaser. The other two
TARF equations consider these two factors in addition to the mass of GHG emissions reduced
c Vehicle simulation models may be used in creating the inputs to OMEGA as discussed in Joint TSD Chapter 3 as
well as Chapter 1 and 2 of the RIA.
D While OMEGA does not model changes in vehicle sales, RIA Chapter 8 discusses this topic.
E A vehicle's footprint is the product of its track width and wheelbase, usually specified in terms of square feet.
F "Learning" is the process whereby the cost of manufacturing a certain item tends to decrease with increased
production volumes or over time due to experience. While OMEGA does not explicitly incorporate "learning"
into the technology cost estimation procedure, the user can currently simulate learning by inputting lower
technology costs in each subsequent redesign cycle based on anticipated production volumes or on the elapsed
time.
5-2
-------
EPA's OMEGA Model
over the life of the vehicle. Fuel prices by calendar year, vehicle survival rates and annual
vehicle miles travelled with age are provided by the user to facilitate these calculations.
For each manufacturer, OMEGA applies technology (subject to penetration cap constraints) to
vehicles until the sales and VMT-weighted emission average complies with the specified
standard or until all the available technologies have been applied. The standard can be a flat
standard applicable to all vehicles within a vehicle class (e.g., cars, trucks or both cars and
trucks). Alternatively, the GHG standard can be in the form of a linear or constrained logistic
function, which sets each vehicle's target as a function of vehicle footprint (vehicle track width
times wheelbase). When the linear form of footprint-based standard is used, the "line" can be
converted to a flat standard for footprints either above or below specified levels. This is referred
to as a piece-wise linear standard, and was used in modeling the standards in this analysis.
The emission target can vary over time, but not on an individual model year basis. One of the
fundamental features of the OMEGA model is that it applies technology to a manufacturer's fleet
over a specified vehicle redesign cycle. OMEGA assumes that a manufacturer has the capability
to redesign any or all of its vehicles within this redesign cycle. OMEGA does not attempt to
determine exactly which vehicles will be redesigned by each manufacturer in any given model
year. Instead, it focuses on a GHG emission goal several model years in the future, reflecting the
manufacturers' capability to plan several model years in advance when determining the technical
designs of their vehicles. Any need to further restrict the application of technology can be
effected through the caps on the application of technology to each vehicle type mentioned above.
Once technology has been added so that every manufacturer meets the specified targets (or
exhausts all of the available technologies), the model produces a variety of output files. These
files include information about the specific technology added to each vehicle and the resulting
costs and emissions. Average costs and emissions per vehicle by manufacturer and industry-
wide are also determined for each vehicle class.
5.2 OMEGA Model Structure
OMEGA includes several components, including a number of pre-processors discussed above
and a baseline vehicle forecast (see Chapter 1). The OMEGA core model collates this
information and produces estimates of changes in vehicle cost and CO2 emission level. Based
on the OMEGA core model output, which now includes the technology penetration of the new
vehicle mix, the scenario impacts (fuel savings, emission impacts, and other monetized benefits)
are calculated via a post-processor called the OMEGA Inventory, Cost and Benefits Tool (ICBT)
discussed in Section IV of the Proposed Determination. These pre- and post-processors and the
OMEGA core model are available in the docket and on our website
at https://www.epa.gov/regiilations-emissions-vehicles-and-engines/optimization-model-
reducing-emissions-greenhouse-gases.
OMEGA is designed to be flexible in a number of ways. Very few numerical values are hard-
coded in the model, and consequently, the model relies heavily on its input files. The model
utilizes five input files: Market, Technology, Fuels, Scenario, and Reference. Figure 5.1 shows
the (simplified) information flow through OMEGA, and how these files interact.
5-3
-------
EPA's OMEGA Model
Pre-Processors Other inputs
Core Model
Post-Processors
OMEGA
LP Model
Cost Model
Vehicle Types
Vehicle Platforms
Vehicle Forecast
Market File
Technology Packages
Ranking Algorithm
Technology File
Reference File
Scenario File
Fuels File
Baseline Fleet
Technology
Accounting
-Technology Penetration
-Impacts
Figure 5.1 Information Flow in the OMEGA Model
OMEGA uses four basic sets of input data. The first, the market file, is a description of the
vehicle fleet. The key pieces of data required for each vehicle are its manufacturer, CO2 emission
level, fuel type, projected sales and footprint. The model also requires that each vehicle be
assigned to a particular vehicle type (currently, we use 29 vehicle types for reasons described
above) which tells the model which set of technologies can be applied to that vehicle. Chapter 1
contains a description of how the market forecasts were created for modeling purposes. In
addition, the degree to which each vehicle already reflects the effectiveness and cost of each
available technology in the baseline fleet must be input. This prevents the model from adding
technologies to vehicles already having these technologies in the baseline. It also avoids the
situation, for example, where the model might try to add a basic engine improvement to a current
hybrid vehicle.
The second type of input data, the technology file, is a description of the technologies
available to manufacturers which consists primarily of their cost, effectiveness, compliance
credit value, and electricity consumption. This file is generated by the Ranking algorithm and a
post-processor tool which puts the Ranking algorithm output files into the proper format for
OMEGA. In all cases, the order of the technologies or technology packages for a particular
vehicle type is designated by the model user in the input files prior to running the model.
The third type of input data describes vehicle operational data, such as annual scrap rates and
mileage accumulation rates, and economic data, such as fuel prices and discount rates. These
estimates are described in Chapter 3 and are contained in the Reference, Fuels and Scenario input
files.
The fourth type of data describes the CO2 emission standards being modeled. These include
the MY2021 standards and the MYs 2022-2025 standards. As described in more detail in
5-4
-------
EPA's OMEGA Model
Chapter 5 of the joint TSD supporting the 2012 FRM, the application of A/C technology is
evaluated in a separate analysis from those technologies which impact CO2 emissions over the 2-
cycle test procedure.3 For modeling purposes, EPA applies this A/C credit by adjusting
manufacturers' car and truck CO2 targets by an amount associated with EPA's projected use of
improved AC systems. The targets are specified in the Scenario input file along with details
such as each scenario's name and the appropriate Market, Technology, Reference and Fuel file to
use for each specific scenario. This is done exactly as done in the Draft TAR analysis.
The input files used in this analysis, as well as the current version of the OMEGA model, are
available in the docket and on EPA's website at https://www.epa.gov/regulations-emissions-
vehicles-and-engines/optimizati on-model-reducing-emissions-greenhouse-gases.
5.3 OMEGA Pre-Processors, Vehicle Types & Packages
Individual technologies can be used by manufacturers to achieve incremental CO2 reductions.
However, EPA believes that manufacturers are more likely to bundle technologies into
"packages" to capture synergistic aspects and reflect progressively larger CO2 reductions with
additions or changes to any given package. In addition, manufacturers typically apply new
technologies in packages during model redesigns that occur approximately once every five years.
This way, manufacturers can more efficiently make use of their redesign resources and more
effectively plan for changes necessary to meet future standards.
Therefore, the approach taken by EPA is to group technologies into packages of increasing
cost and effectiveness. Costs for the packages are a sum total of the costs for the technologies
included. Importantly, the package costs and effectiveness represent those respective values
relative to a "null" package of technologies. That "null" package consists of a fixed valve, port
fuel injected engine mated to a 4 speed automatic transmission and having a declared 0 percent
level of mass reduction. This "null" package is not meant to reflect an actual vehicle, but rather
a technology "zero cost floor" or "zero effectiveness floor" from which costs and effectiveness of
packages can be measured. This way, the technology package cost and effectiveness for the set
of technologies on any actual vehicle can be determined relative to the null, an OMEGA package
cost and effectiveness can then be calculated relative to the null, and the delta between the actual
vehicle package and the OMEGA package can then be easily calculated. Effectiveness is
somewhat more complex, as the effectiveness of individual technologies cannot simply be
summed. To quantify the CO2 (or fuel consumption) effectiveness, EPA relies on ALPHA and
the Lumped Parameter Model, which are described in greater detail in Chapter 2 of this TSD.
5.3.1 Vehicle Types
As was done in the 2012 FRM and the Draft TAR, EPA uses "vehicle types" to represent the
entire fleet in OMEGA. This was the result of analyzing the existing light-duty fleet with respect
to vehicle size and powertrain configurations. In the past, all vehicles, including cars and trucks,
were first distributed based on their relative size (i.e., vehicle class), starting from compact cars
and working upward to large trucks. Next, each vehicle was evaluated for powertrain,
specifically the engine size, 14, V6, and V8, then by valvetrain configuration (DOHC, SOHC,
OHV), and finally by the number of valves per cylinder. We further designated some vehicle
types as towing vehicle types and some as non-towing vehicle types. This towing/non-towing
determination impacts the types of packages made available to specific vehicle within each
5-5
-------
EPA's OMEGA Model
vehicle type since only non-towing vehicle types are considered to be appropriate for
electrification beyond strong HEV (i.e., to plug-in HEV or full BEV).
For this Proposed Determination, EPA has expanded the number of vehicle types from 19 to
29 to better characterize the fleet in terms of power-to-weight ratio, road load characteristics and
size based on curb weight rather than a purely size-based market class definition. As a result, we
no longer determine vehicle type based on whether a vehicle is a small car or a large SUV and,
instead, make the determination in part based on its curb weight. We also make the
determination based on the vehicle's power-to-weight ratio and road load characteristics or, in
other words, its "ALPHA Class." This is described in more detail in Chapter 2.3.1.4 of the TSD.
The implication to this change is a more appropriate determination of technology effectiveness
and cost values than in past analyses. EPA believes that these 29 vehicle types broadly
encompass the diversity in the fleet and that the analysis is appropriate for "average" vehicles.
As such, the six ALPHA classes (low, medium, and high vehicle power-to-weight levels,
abbreviated as 'LPW', 'MPW', and 'HPW', respectively; the first two of these are divided further
into low and high vehicle road load categories, abbreviated as 'LRL' and 'HRL', respectively),
and the six curb weight classes (simply numbered 1 through 6 with 1 being the lightest curb
weights and 5 the heaviest non-pickup curb weights; 6 is reserved for pickups) serve primarily to
determine the effectiveness levels of new technologies by determining which input metrics are
chosen within the lumped parameter model (see below). So, any vehicle models mapped into a
LPW HRL 3 vehicle type will get technology-specific effectiveness results for vehicles with
low power-to-weight, high road load characteristics. Similarly, any such vehicles will get
technology-specific costs, where applicable, for vehicles in curb weight class number 3, i.e.,
those costs developed on a weight basis such as advanced diesel, hybrid and other electrified
powertrains and mass reduction. Note that most technology costs are not developed according to
vehicle weight but are instead developed according to engine size, valvetrain configuration, etc.
A detailed table showing the 29 vehicle types, their baseline engines, their descriptions and some
example models for each is contained in the table below. Note that some models, specifically
models with turbocharged engines or fueled by diesel fuel, are mapped into vehicle types whose
description seems inaccurate. For example, the turbocharged Cruze (vehicle type 12) actually has
an 14 DOHC engine, not a V6 DOHC engine. However, in OMEGA-space, such a vehicle
operates as a V6 engine since its power and operating characteristics, its utility, is consistent with
a V6 engine. Importantly, its effectiveness values will be consistent with a "LPW LRL" ALPHA
class and its costs values will be consistent with a turbocharged 14 in curb weight class 1. These
characteristics are carefully tracked within OMEGA. That said, we will continue to study our
classifications and may move toward vehicle types specifically for turbocharged vehicles in
future analyses.
5-6
-------
EPA's OMEGA Model
Table 5.1 Vehicle Types and Example Models
Vehicle Type
Description
Curb Weight Class
ALPHA Class
Example Models
1
14 DOHC
1
LPW LRL
Sentra, Corolla
2
14 DOHC
1
MPW LRL
Dart, Focus
3
14 DOHC
2
MPW LRL
Altima, Camry
4
14 DOHC
2
LPW HRL
Rogue, Patriot
5
14 DOHC
3
MPW LRL
Malibu, 200
6
14 DOHC
3
LPW HRL
Forester, Cherokee
7
14 DOHC
4
LPW HRL
Outback, Equinox
8
14 DOHC
6
Truck
Colorado, Tacoma
9
V6 0HV
6
Truck
Silverado, Sierra
10
V6SOHC
3
HPW
RDX, TLX
11
V6SOHC
4
MPW HRL
Odyssey
12
V6DOHC
1
LPW LRL
Cruze, Focus turbos
13
V6DOHC
2
MPW LRL
Fiesta turbo
14
V6DOHC
2
LPW LRL
Passat
15
V6DOHC
3
HPW
E350, Impala, Q50
16
V6DOHC
3
MPW LRL
IS250
17
V6DOHC
3
LPW HRL
Transit
18
V6DOHC
4
HPW
Charger
19
V6DOHC
4
MPW HRL
Pathfinder, Journey
20
V6DOHC
5
HPW
Camaro
21
V6DOHC
5
MPW HRL
Grand Cherokee
22
V6DOHC
6
Truck
Tacoma, Frontier
23
V8 0HV
5
HPW
Charger
24
V8 0HV
5
MPW HRL
Tahoe, Suburban
25
V8 0HV
6
Truck
Silverado, Sierra
26
V8DOHC
4
HPW
Mustang, SL550
27
V8DOHC
5
HPW
QX80, GL550
28
V8DOHC
5
MPW HRL
GX460, Sequoia
29
V8DOHC
6
Truck
Tundra, F150
Note: DOHC=dual overhead cam; SOHC=single overhead cam; OHV=overhead valve; Curb Weight Class is a
percentile-based weight classification with 1 being the lightest and 6 being the heaviest vehicles; ALPHA class is
described in Chapter 2 of the TSD and designates low/medium/high power-to-weight (L/M/HPW) and
low/medium/high road load (L/M/HRL) or Truck which is used for pickups like the Ford F150 and Chevy
Silverado.
5.3.2 Technology Packages, Package Building & Master-sets
Importantly, the effort in creating OMEGA packages attempts to maintain a constant utility
and acceleration performance for each package as compared to the baseline package. As such,
each package is meant to provide equivalent driver-perceived performance to the baseline
package. There are two possible exceptions. The first is the towing capability of vehicle types
which we have designated "non-pickups." This requires a brief definition of what we consider to
be a towing vehicle versus a non-towing vehicle. Nearly all vehicles sold today, with the
exception of the smaller subcompact and compact cars, are able to tow up to 1,500 pounds
provided the vehicle is equipped with a towing hitch. These vehicles require no special OEM
"towing package" of add-ons which typically include a set of more robust brakes and some
additional transmission cooling. We do not consider such vehicles to be towing vehicles. Other
5-7
-------
EPA's OMEGA Model
vehicles a capable of towing up to 5,000 pounds, with the addition of a towing package, but are
not heavy towing vehicles. We reserve the heavy-towing term for those vehicles capable of
towing significantly more than 5,000 lbs. For example, a base model Ford Escape can tow 1,500
pounds while the V6 equipped towing version can tow up to 3,500 pounds. The former would
not be considered a true towing vehicle while the latter would although it would not be
considered a heavy-towing vehicle. The heavy-towing vehicles are those built, generally on a
ladder frame and are generally pickup trucks. Vehicles mapped into those "Truck" vehicle types
are considered heavy-towing vehicles and, as such, are not considered to be candidates for
electrification beyond strong HEV.
We do not address towing at the vehicle level. Instead, we deal with towing at the vehicle
type level. The importance of this distinction can be found in the types of hybrid and plug-in
hybrid technologies we apply to towing versus non-towing vehicle types.0 For the "Truck"
vehicle types, we apply a P2 hybrid technology with a turbocharged and downsized gasoline
direct injected engine. These packages are expected to maintain equivalent towing capacity to
the baseline engine they replace. For the non-heavy towing vehicle types, we apply a P2 hybrid
technology with a low-compression ratio Atkinson engine (not an Atkinson-2 engine) that has
not been downsized relative to the baseline engine. This type of low-compression ratio Atkinson
engine is used in the current Toyota Prius and Ford Escape hybrid and should not be confused
with a high-compression ratio Atkinson 2 engine. We have maintained the original engine size
(i.e., no downsizing) to maintain utility as best as possible, but EPA acknowledges that due to its
lower power output, a low-compression ratio Atkinson cycle engine cannot tow loads as well as
a standard Otto-cycle engine of the same size. However, the presence of the hybrid powertrain
would be expected to maintain towing utility for these vehicle types in all but the most severe
operating extremes. Such extremes would include towing up very long duration grades (e.g., like
in the Rocky Mountains) (i.e.,) or towing up a shorter but very steep grade (e.g., Pike's Peak)).
Under these extreme towing conditions, the battery on a hybrid powertrain would eventually
cease to provide sufficient supplemental power and the vehicle would be left with the engine
doing all the work. A loss in utility would result (note that the loss in utility should not result in
breakdown or safety concerns, but rather loss in top speed and/or acceleration capability).
Importantly, those towing situations involving driving outside mountainous regions would not be
affected.
The second possible exception to our attempt at maintaining utility is the electric vehicle
range. We have built electric vehicle packages with ranges of 75, 100 and 200 miles. Clearly
these vehicles would not provide the same utility as a gasoline vehicle which can be refueled
very quickly and, therefore, has unlimited range (effectively). However, from an acceleration
performance standpoint, the utility would be equal to if not perhaps better than the gasoline
vehicle. We believe that buyers of electric vehicles in the MYs 2021-2025 timeframe will be
purchasing the vehicles with a full understanding of the range limitations and will use their
vehicles accordingly. As such, we believe that the buyers of EVs will experience no loss of
expected utility.
G This towing/non towing distinction is not an issue for non-HEVs, EPA maintains whatever towing capability
existed in the baseline when adding/substituting technology.
5-8
-------
EPA's OMEGA Model
To prepare inputs for the OMEGA model, EPA builds "master-sets" of technology packages.H
The master-set of packages for each vehicle type are meant to reflect both appropriate groupings
of technologies (e.g., we do not apply turbochargers unless an engine has dual overhead cams,
some degree of downsizing, direct injection and dual cam phasing) and limitations associated
with penetration caps (see 2012 FRM joint TSD 3.5 and the brief discussion in Section 5.3.3).
We then filter that list by determining which packages provide the most cost effective groups of
technologies within each vehicle type—those that provide the best trade-off of costs versus CO2
reduction improvements. This is done by ranking those groupings based on the Technology
Application Ranking Factor (TARF). The TARF is the factor used by the OMEGA model to
rank packages and determine which are the most cost effective to apply. The TARF is calculated
as the net incremental cost (or savings) of a package per kilogram of CO2 reduced by the
package relative to the previous package. The net incremental cost is calculated as the
incremental cost of the technology package less the incremental discounted fuel savings of the
package over 5 years. The incremental CO2 reduction is calculated as the incremental CO2 /mile
emission level of the package relative to the prior package multiplied by the lifetime miles
travelled. More detail on the TARF can be found in the OMEGA model supporting
documentation (see EPA-420-B-10-042). We also describe the TARF ranking process in more
detail below. Grouping "reasonable technologies" simply means grouping those technologies
that are complementary (e.g., turbocharging plus downsizing) and not grouping technologies that
are not complementary (e.g., dual cam phasing and coupled cam phasing).
To generate the master-set of packages for each of the vehicle types, EPA has built packages
in a step-wise fashion looking first at "simpler" conventional gasoline and vehicle technologies,
then more advanced gasoline technologies such as turbocharged (with varying levels of boost)
and downsized engines with gasoline direct injection and then hybrid and other electrified
vehicle technologies. This was done by assuming that auto makers would first concentrate
efforts on conventional gasoline engine and transmission technologies paired with some level of
mass reduction to improve CO2 emission performance. Mass reduction varied from no mass
reduction up to 20 percent as the maximum considered in this analysis.1
Once the conventional gasoline engine and transmission technologies have been fully
implemented, we expect that auto makers would apply more complex (and costly) technologies
such as turbocharged and downsized gasoline engines and/or converting conventional gasoline
engines to advanced diesel engines in the next redesign cycle. Auto makers may also move to
hybridization, both mild and strong hybrids. For this analysis, we have built all of our mild
H In fact, we first build a package list of packages for each model for each model year for which we run OMEGA
because penetration caps result in different technologies being available. From those, we build Master-sets for
each relevant model year and emission standard combination since costs change over time resulting in different
costs every year.
1 Importantly, the mass reduction associated for each of the 19 vehicle types was based on the vehicle-type sales
weighted average curb weight. Although considerations of vehicle safety are an important part of EPA's
consideration in establishing the standards, note that allowable weight reductions giving consideration to safety is
not part of the package building process so we have built packages for the full range of 0-20% weight reduction
considered in this analysis. Weight consideration for safety is handled within OMEGA as described in Chapter 8
of this Draft TAR.
5-9
-------
EPA's OMEGA Model
hybrid packages using the newly emerging 48 Volt technology. We have built two types of
strong hybrid packages for this analysis, consistent with the 2012 FRM, as was described above.
Lastly, for some vehicle types (i.e., the non-Truck vehicle types), we anticipate that auto
makers would move to more advanced electrification in the form of both plug-in hybrid (PHEV,
sometimes referred to as range extended electric vehicles (REEV)) and full battery electric
vehicles (BEV).J
Importantly, the HEV, PHEV and BEV (called collectively P/H/EV) packages here take into
consideration the impact of the weight of the electrified components, primarily the battery packs.
Because these battery packs can be quite heavy, if one removes 20 percent of the mass from a
gasoline vehicle but then converts it to an electric vehicle, the resultant net weight reduction will
be less than 20 percent. We discuss this in more below where we provide additional discussion
regarding the P/H/EV packages.
The result of this package building process is a set of "Package List" files, one for MY2021
and one for MY2025. These package list files provide a description of each package, a unique
package number for that package which follows that package throughout the OMEGA process
within a given model year, and details of each technology and associated codes within each
package. The distinction being made here is that the package description may include dual cam
phasing (DCP), but the package details might indicate DCP on a V6 engine for one package, and
DCP on an 14 engine for another package in the same vehicle type since this second package
includes turbocharging and downsizing. The package list files used as part of EPA's analysis are
contained in the docket and on our website and the step-by-step process is detailed below.K
In building MY2021 packages, we proceed according to the following sequence of steps (note
that underlined technologies are simply meant to guide the reader to differences between
technologies included in packages; note also that the number of packages are unique to non-
Truck vehicle types, slightly more HEV packages are built for Truck vehicle types since they are
built with both TDS18 and TDS24 while non-Truck vehicle types are built with only Atkinson 1
engines; the final result is 9269 packages per non-Truck vehicle type and 9360 for each Truck
vehicle type, or roughly 270,000 packages):
1) With 5 percent mass reduction:
a) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 1, improved accessories
level 1, electric power steering, lower rolling resistance tires level 1, passive
aero, low drag brakes, variable valve timing
b) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 1, improved accessories
level 1, electric power steering, lower rolling resistance tires level 1, passive
plus active aero, low drag brakes, variable valve timing
1 In some OMEGA files, BEV is also referred to as EV.
K See our website at https://www.epa.gov/reaulations-emissions-vehieles-and-engines/optimization-model-reducing-
emissions-greenhouse-gases.
5-10
-------
EPA's OMEGA Model
c) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 1, improved accessories
level 2, electric power steering, lower rolling resistance tires level 1, passive
plus active aero, low drag brakes, variable valve timing
d) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 1, improved accessories
level 1, electric power steering, lower rolling resistance tires level 2. passive
aero, low drag brakes, variable valve timing
e) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 1, improved accessories
level 1, electric power steering, lower rolling resistance tires level 2. passive
plus active aero, low drag brakes, variable valve timing
f) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 1, improved accessories
level 2. electric power steering, lower rolling resistance tires level 2. passive
plus active aero, low drag brakes, variable valve timing
g) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 2. improved accessories
level 1, electric power steering, lower rolling resistance tires level 2. passive
aero, low drag brakes, variable valve timing
h) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 2. improved accessories
level 1, electric power steering, lower rolling resistance tires level 2. passive
plus active aero, low drag brakes, variable valve timing
i) With TRX11 & again with TRX12 (2 packages):
i) Low friction lubes, engine friction reduction level 2, improved accessories
level 2. electric power steering, lower rolling resistance tires level 2. passive
plus active aero, low drag brakes, variable valve timing
j) Steps l.a through l.i with cylinder deactivation (18 packages)
k) Steps l.a through l.i with gasoline direct injection (18 packages)
1) Steps la. through l.i with cylinder deactivation and gasoline direct injection (18
packages)
m) Steps l.a through 1.1 with stop-start (72 packages)
n) Steps l.a through l.m with secondary axle disconnect (144 packages)
5-11
-------
EPA's OMEGA Model
o) Any package in Steps 1 .a through 1 .m that includes gasoline direct injection, add
Atkinson-2 (144 packages)
p) Step l.o, add cooled EGR (144 packages)
q) Any package in Steps 1 .a through 1 .m that includes gasoline direct injection,
replace cylinder deactivation with discrete variable valve lift and add turbo-
downsize 18-bar (144 packages)
r) Any package in Steps 1 .a through 1 .m that includes gasoline direct injection,
replace cylinder deactivation with discrete variable valve lift and add turbo-
downsize 24-bar plus cooled EGR (144 packages)
s) Any package in Steps 1 .a through 1 .m that includes gasoline direct injection, add
Miller-cycle plus cooled EGR (144 packages)
t) Step l.a through l.s with TRX21 & again with TRX22 (1008 packages)
u) Any packages with improved accessories level 2, add mild HEV 48V (336
packages)
v) Any packages with gasoline direct injection, engine friction reduction level 2 and
lower rolling resistance tires level 2, add advanced diesel (24 packages)
2) With 10 percent mass reduction
a) Repeat Step 1 (2376 packages)
b) Step 2.a packages with improved accessories level 1 and no advanced diesel, add
strong HEV (48 packages)
3) With 15 percent mass reduction
a) Repeat Step 2 (2424 packages)
4) With 20 percent mass reduction (not done for "Truck" vehicle types)
a) Build PHEV20 & PHEV40 (REEV20 & REEV40) (2 packages)
b) Build EV75, EV100, EV200 (3 packages)
5) For off-cycle levels 1 and 2
a) Any Step 1 through 3 packages with active aero, lower rolling resistance tires
level 2, improved accessories level 2 and TRX21
i) Add off-cycle level 1 (OC1) (510 packages)
b) Any Step 1 through 3 packages with active aero, lower rolling resistance tires
level 2, improved accessories level 2 and TRX22
i) Add off-cycle level 1 (OC1) (510 packages)
c) Any package with off-cycle level 1, remove off-cycle 1 and add off-cycle level 2
(OC2) (1020 packages)
5-12
-------
EPA's OMEGA Model
In building MY2025 packages, we proceed according to a very similar sequence as outlined
above with the exception that the presence of fewer penetration caps in MY2025 means less
iteration on first level technologies resulting in fewer sub-steps within Step 1 and, as a result,
fewer packages per vehicle type.
The package lists are then sent through EPA's TEB-CEB "Machine" which is the tool in the
OMEGA process that brings together technology costs and technology effectiveness (via the
Lumped Parameter Model) to determine package level costs and effectiveness. The TEB-CEB
Machine calculates the Technology Effectiveness Basis and the Cost Effectiveness Basis of each
package. With package level costs and effectiveness, we can then use the OMEGA Master-set
generator tool to generate a Master-set of packages. The Master-set of packages adds to the
package cost and effectiveness values the 5-year discounted fuel savings and lifetime CO2
reductions for each packaged These additional metrics allow for calculation of a TARF for each
unique package contained in the applicable package list. Importantly, in building packages and
the Master-sets of packages, we have not yet considered the baseline fleet beyond the sales-
weighted metrics of each of the 29 vehicle types. Instead, we have considered only appropriate
groupings of technologies into packages and built packages and Master-sets based on the 29
vehicle types and the sales-weighted attributes of those vehicle types (e.g., CO2 and curb
weight).
5.3.3 Master-set Ranking and the Technology Input File
This master-set of packages is then ranked by TARF within vehicle type for each Master-set
of packages necessary to represent the reference case and the control case. In this analysis, this
requires 4 Master-sets: Reference case in MY2021, Reference case in MY2025, Control case in
MY2021 and Control case in MY2025. However, we can use the same Master-set for both the
Reference case in MY2021 and the Control case in MY2021 since the same set of costs apply.
The end result being a necessary set of 3 Master-sets for a given OMEGA run. Should any
effectiveness or cost value, synergy factor, fuel price, etc., be changed, a different Master-set or
group of Master-sets would be required.
The ranking process is handled by the OMEGA pre-processing Ranking Algorithm (contained
in the docket and on our website) which calculates the TARF of each package relative to the
sales-weighted representative package within a given vehicle type. The package with the best
TARF is selected as OMEGA package #1 for that vehicle type. The remaining packages for the
given vehicle type are then ranked again by TARF, this time relative to OMEGA package #1.
The best package is selected as OMEGA package #2, etc.
An important consideration in the ranking process is the penetration caps which cannot be
exceeded to ensure that the packages chosen by the ranking do not result in exceedance of the
caps. As such, if package #2 contains a technology, for example TRX21, but the penetration cap
for TRX21 is, say 60 percent, then only 60 percent of the population of vehicles in the given
vehicle type would be allowed to migrate to package #2 with the remaining 40 percent left in
package #1. We had a detailed discussion of penetration caps in Section 3.5 of the final joint
TSD in support of the 2012 FRM.4 For this analysis, we have used the same penetration caps as
L These metrics are calculated using the sales weighted CO2 level of all vehicles mapped into each specific vehicle
type.
5-13
-------
EPA's OMEGA Model
presented there with the exception of adding a new penetration cap for the Atkinson-2
technology, which was not considered in the 2012 FRM. The Atkinson-2 penetration cap used in
this analysis is the same as that used in the Draft TAR. For the mild HEV 48V technology, we
have used the same penetration cap as used for the mild HEV technology described in the 2012
FRM and as used in the Draft TAR. For the new Miller cycle technology, we have used the 24-
bar turbocharging penetration caps used in the 2012 FRM and the Draft TAR. The penetration
caps used in this analysis are shown in the table below. New for this analysis are penetration
caps for off-cycle level 1 and 2 (OC1 and OC2). For those, we have not applied any caps.
5-14
-------
EPA's OMEGA Model
Table 5.2 Penetration Caps used in the OMEGA Central Analysis Runs
Tech code
Tech
2021
2025
Aero 1
Aero - passive
100%
100%
Aero 2
Aero - passive with active
80%
100%
ATK2
Atkinson-2
80%
100%
CCC
Camshaft configuration changes without downsizing
100%
100%
CCP
Coupled cam phasing
100%
100%
CVVL
Continuous variable valve lift
100%
100%
DCP
Dual cam phasing
100%
100%
Deac
Cylinder deactivation
100%
100%
DSL-Adv
Advanced diesel
30%
42%
DVVL
Discrete variable valve lift
100%
100%
EFR1
Engine friction reduction level 1
100%
100%
EFR2
Engine friction reduction level 2
60%
100%
EGR
Cooled exhaust gas recirculation
30%
75%
EPS
Electric power steering
100%
100%
EV75
Full battery electric vehicle 75 mile range
5%
8%
EV100
Full battery electric vehicle 100 mile range
5%
8%
EV200
Full battery electric vehicle 200 mile range
5%
8%
Dl
Gasoline direct injection
100%
100%
IACC1
Improved accessories level 1
100%
100%
IACC2
Improved accessories level 2
80%
100%
LDB
Low drag brakes
100%
100%
LRRT1
Lower rolling resistance tires level 1
100%
100%
LRRT2
Lower rolling resistance tires level 2
75%
100%
LUB
Engine changes to accommodate low friction lubes
100%
100%
MHEV48V
Mild hybrid 48V
50%
80%
P2 or HEV
Strong hybrid
30%
50%
REEV20
Range extended or plug-in electric vehicle 20 mile range
8%
11%
REEV40
Range extended or plug-in electric vehicle 40 mile range
8%
11%
SAX
Secondary axle disconnect
100%
100%
Stop-start
Stop-start without electrification
100%
100%
TDS18
Turbocharging with downsizing 18-bar
100%
100%
TDS24
Turbocharging with downsizing 24-bar
30%
75%
TRX11
Transmission - step 1 or current generation
100%
100%
TRX12
TRX11 with improved efficiency
30%
100%
TRX21
Transmission - step 2 or TRX11 but with additional gear-ratio spread
80%
100%
TRX22
TRX21 with improved efficiency
30%
100%
TURBM
Miller cycle or ATK2 with turbocharging
30%
75%
WR10
Weight reduction of 10% from EPA's "null"
100%
100%
WR15
Weight reduction of 15% from EPA's "null"
100%
100%
WR20
Weight reduction of 20% from EPA's "null"
0%
100%
OC1
Off-cycle level 1
100%
100%
OC2
Off-cycle level 2
100%
100%
5-15
-------
EPA's OMEGA Model
Also tracked are the credits available to the package which are also included in this ranking
process.M The table below presents 2015 baseline data used in the TARF ranking process.
Table 5.3 Lifetime VMT & Baseline CO2 used for the TARF Ranking Process
Vehicle
Type
Description
Curb
Weight
Class
ALPHA
Class
Example Models
C/T
MY2021
Lifetime
VMT
MY2025
Lifetime
VMT
Base CO2
(g/mi)
1
14 DOHC
1
LPW LRL
Sentra, Corolla
C
184,789
189,264
201.4
2
14 DOHC
1
MPW LRL
Dart, Focus
C
241.2
3
14 DOHC
2
MPW LRL
Altima, Camry
C
232.7
4
14 DOHC
2
LPW HRL
Rogue, Patriot
C
249.3
5
14 DOHC
3
MPW LRL
Malibu, 200
C
242.7
6
14 DOHC
3
LPW HRL
Forester, Cherokee
C
247.8
7
14 DOHC
4
LPW HRL
Outback, Equinox
T
214,994
220,200
264.3
8
14 DOHC
6
Truck
Colorado, Tacoma
T
321.9
9
V6 OHV
6
Truck
Silverado, Sierra
T
354.2
10
V6SOHC
3
HPW
RDX, TLX
C
184,789
189,264
288.9
11
V6SOHC
4
MPW HRL
Odyssey
T
214,994
220,200
321.2
12
V6DOHC
1
LPW_LRL
Turbo Cruze, Turbo
Focus
C
184,789
189,264
223.2
13
V6DOHC
2
MPW_LRL
Turbo Fiesta, Turbo
Jetta
C
243.6
14
V6DOHC
2
LPW_LRL
Turbo Encore, Diesel
Jetta
c
235.9
15
V6DOHC
3
HPW
E350, Impala, Q50
c
276.6
16
V6DOHC
3
MPW LRL
IS250
c
272.9
17
V6DOHC
3
LPW HRL
Transit
c
265.3
18
V6DOHC
4
HPW
Charger
c
317.5
19
V6DOHC
4
MPW HRL
Pathfinder, Journey
T
214,994
220,200
291.3
20
V6DOHC
5
HPW
Camaro
T
336.2
21
V6DOHC
5
MPW HRL
Grand Cherokee
T
349.1
22
V6DOHC
6
Truck
Tacoma, Frontier
T
366.1
23
V8 0HV
5
HPW
Charger
T
392.0
24
V8 0HV
5
MPW HRL
Tahoe, Suburban
T
379.7
25
V8 0HV
6
Truck
Silverado, Sierra
T
383.7
26
V8DOHC
4
HPW
Mustang, SL550
C
184,789
189,264
334.7
27
V8DOHC
5
HPW
BX460
T
214,994
220,200
383.0
28
V8DOHC
5
MPW HRL
Tundra, F150
T
377.2
29
V8DOHC
6
Truck
Turbo F150, Diesel Ram
T
394.7
Once a Master-set is ranked, the result is a Ranked-set of packages with a maximum of 50
packages for each vehicle type. This Ranked-set of packages is used to generate the Technology
input file for the OMEGA core model and to generate the "Scenario packages" to be applied to
vehicles within each vehicle type. In the Technology input file, the package progression, or
"flow" of packages is included. The package progression is key because OMEGA evaluates
each package in a one-by-one, or linear progression. The packages must be ordered correctly so
that no single package will prevent the evaluation of the other packages. For example, if we
M We have included credits for aerodynamic treatments level 2, 12V stop-start, mild HEV and strong HEV.
5-16
-------
EPA's OMEGA Model
simply listed packages according to increasing effectiveness, there could well be a situation
where an HEV with higher effectiveness and a better TARF than a turbocharged and downsized
package with a poor TARF could never be chosen because the turbocharged and downsized
package, having a poor TARF, would never get chosen and would effectively block the HEV
from consideration. For that reason, it is important to first rank by TARF so that the proper
package progression can be determined. In other words, packages do not necessarily flow from a
given package to the next package listed. Because of the penetration caps, a package listed as,
for example, step 8 might actually come from step 5 rather than from step 7. As such, within
OMEGA, the incremental cost for step 8 would be the cost for step 8 less the cost for step 5 and
similar for the effectiveness values. All of the Ranked-sets of packages and the Technology
input files are contained in the docket and at our website at https://www.epa.gov/reeutations-
emissions-vehicles-and-engines/optimizati on-model-reducing-emissions-greenhouse-gases.
5.3.4 Applying Ranked-sets of Packages to the Projected Fleet
As noted above, when we apply a package of technologies to an individual vehicle model in
the baseline fleet, we must first determine which package of technologies are already present on
the individual vehicle model. From this information, we can determine the effectiveness and
cost of the individual vehicle model in the baseline fleet relative to the "null" package that
defines the vehicle type. Once we have that, we can determine the incremental increase in
effectiveness and cost for each individual vehicle model in the baseline fleet once it has added
the package of interest. This process is known as the TEB-CEB process, which is short for
Technology Effectiveness Basis - Cost Effectiveness Basis. This process allows us to accurately
reflect the level of technology already in the 2015 baseline fleet as well as the level of
technology expected in the MYs 2021-2025 reference case (i.e., the fleet as it is expected to exist
as a result of the MY 2021 standards).
The TEB-CEB Machine is again used, along with a set of Scenario packages, to generate the
actual TEB and CEB values for each package as it is applied to each individual model within the
analysis fleet. These TEB and CEB values, along with the off-cycle effectiveness (OEB) values
are then used in the Market input file and serve as one of the primary inputs to the OMEGA core
algorithms.
The TEB-CEB Machine's process when applying Ranked-set packages to actual vehicles can
be broken down into four steps. The first step in the process is to break down the available GHG
control technologies into five groups: 1) engine-related, 2) transmission-related, 3) hybridization,
4) weight reduction and 5) other. Within each group we gave each individual technology a
ranking which generally followed the degree of complexity, cost and effectiveness of the
technologies within each group. More specifically, the ranking is based on the premise that a
technology on a baseline vehicle with a lower ranking would be replaced by one with a higher
ranking which was contained in one of the technology packages which we included in our
OMEGA modeling. The corollary of this premise is that a technology on a baseline vehicle with
a higher ranking would be not be replaced by one with an equal or lower ranking which was
contained in one of the technology packages which we chose to include in our OMEGA
modeling. This ranking scheme can be seen in Visual Basic Macro contained within the TEB-
CEB Machine which is in available in the docket and on our website
at https://www.epa.gov/regiilations-emissions-vehicles-and-engines/optimization-model-
reducing-emissions-greenhouse-gases.
5-17
-------
EPA's OMEGA Model
In the second step of the TEB-CEB process, these technology group rankings are used to
estimate the complete list of technologies which would be present on each vehicle after the
application of a technology package. In other words, this step indicates the specific technology
on each vehicle after a package has been applied to it. The Machine then uses EPA's lumped
parameter model to estimate the total percentage CO2 emission reduction associated with the
technology present on the baseline vehicle (termed package 0), as well as the total percentage
reduction after application of each package. The Machine uses this approach to determine the
total cost of all of the technology present on the baseline vehicle and after the application of each
applicable technology package.
The third step in this process is to account for the degree to which each technology package's
incremental effectiveness and incremental cost is affected by the technology already present on
the baseline vehicle. For this analysis, we also account for the credit values using a factor termed
"Other effectiveness basis (OEB).
As described above, technology packages are applied to groups of vehicles which generally
represent a single vehicle platform and which are equipped with a single engine size (e.g.,
compact cars with four cylinder engines produced by Ford). Thus, the fourth step is to combine
the fractions of the CEB and TEB of each technology package already present on the individual
baseline vehicle models for each vehicle grouping. For cost, percentages of each package
already present are combined using a simple sales-weighting procedure, since the cost of each
package is the same for each vehicle in a grouping. For effectiveness, the individual percentages
are combined by weighting them by both sales and base CO2 emission level. This appropriately
weights vehicle models with either higher sales or CO2 emissions within a grouping. Once
again, this process prevents the model from adding technology which is already present on
vehicles, and thus ensures that the model does not double count technology effectiveness and
cost associated with complying with the modeled standards.
The other effectiveness basis (OEB) was designed to appropriately account for credit
differences between technologies actually on the vehicle and technology packages applied
through the technology input file. As an example, if a baseline vehicle includes start stop
technology, and the applied package does not, the model needs to account for this different in
off-cycle credit. The OEB is an absolute credit value and is used directly in the model's
compliance calculations.
5.3.5 New to OMEGA since the Draft TAR
Based on input from public comments and other information that became available to us, we
made certain changes to what we term the "OMEGA Suite" of tools used in generating a full
Benefit-Cost Analysis. Those changes are listed below and are detailed throughout this TSD:
• The baseline fleet was updated from a basis in MY2014 to MY2015
• Future vehicle sales projections were updated based on AEO2016 sales projections.
• The ZEV program sales were updated based on the updates mentioned above.
• All fuel prices used throughout the OMEGA Suite now use AEO2016 fuel prices.
5-18
-------
EPA's OMEGA Model
• All monetized values (technology costs, maintenance costs, SCC and non-GHG
cost/ton values, etc.) have been updated to 2015 dollars for consistency with
AEO2016 fuel price estimates.
• Vehicle mileage accumulation rates and survival rates were updated based on AEO
2016 projections.
• Baseline levels of mass reduction were updated for the new baseline fleet.
• Baseline levels of passive and active aero technologies were updated resulting in
more use of those technologies in the MY2015 baseline than in the Draft TAR fleet.
• Baseline levels of lower rolling resistance tires level 1 and 2 were updated resulting in
more use of those technologies in the MY2015 baseline than in the Draft TAR fleet.
• Corrected an internal coding error in the mass reduction penalty determination
associated with the added weight of the battery on strong HEVs which, in the Draft
TAR, was erroneously 0 percent on all strong HEVs. Similarly, corrected an error in
the mass reduction tracking where mass reduction penalties are involved (i.e., mild
and strong HEVs, PHEVs and full EVs); this resulted in some
WRtech/WRpen/WRnet values being confused.
• Updated the methodology used for calculating allowed mass reduction levels in light
of applicable mass reduction technology penetration caps. Those allowed levels of
mass reduction are now based on "null" curb weight rather than simply "baseline"
curb weight as was done in the Draft TAR. This is more consistent with the basis for
the penetration caps which also are based on null curb weight. In turn, we also
updated the methodology for applying maximum mass reduction levels as part of the
safety analysis.
• All full BEV and PHEV vehicles are placed on unique platforms rather than being
part of an internal combustion engine (ICE) platform. This is true of BEV/PHEV in
the baseline and those created as part of the ZEV program fleet, which results in
many more platforms than in the Draft TAR. This was done to allow for accounting
of upstream emissions for BEV/PHEV in OMEGA.
° This also required an update to the Technology Effectiveness Basis (TEB)
calculation for full EVs. In all prior versions of OMEGA, the TEB for a full BEV
was always 0 gC02/mi. The TEB is now calculated as equivalent to the baseline
vehicle's indicated C02 level which is user controlled. In OMEGA for this
analysis, when considering upstream emissions associated with electricity
consumption, we have entered the upstream emissions value as the baseline C02
level. That way, the TEB reflects upstream C02 emissions and the
manufacturer's compliance determination, likewise, reflects those upstream
emissions.
° These upstream emissions are then post-processed via a new post-processing
summary generating tool to correctly track tailpipe C02 versus upstream (or grid)
C02.
5-19
-------
EPA's OMEGA Model
° The OMEGA Market file now shows grid C02 for BEV/PHEV (in a formerly
unused column called "Towing Capacity" and codes them as fueled by electricity
despite PHEVs being fueled also by gasoline. The electricity fuel code serves
only as a trigger for the TEB-CEB process to use the baseline values at every
package step.
• Correction made to the treatment of stop-start technology on baseline vehicles which,
in the Draft TAR, was mistakenly ignored.
• Correction made to the effectiveness calculation of Miller cycle engines such that the
Atkinson 2 portion is no longer double counted.
• Correction made to the reporting of included technologies in OMEGA "tech code
strings" such that BEV and PHEV tech codes are no longer included simultaneously.
• Updated vehicle classifications away from categories such as "small car" and "large
MPV and toward a power-to-weight and road-load determination. Similarly, updated
cost classifications away from categories such as "small car" and "large MPV" and
toward a curb weight classification system since curb weight better reflects applicable
costs (e.g., mass reduction costs, battery costs).
• Application of BEV and/or PHEV technology is no longer determined based on a
loose "towing" versus "non-towing" determination. Instead, BEV and PHEV
technologies are now allowed on most vehicles with the sole exception of pickup
trucks. As a result, many more MPV-type vehicles (minivans, SUVs, cross-over
utilities) are now open to electrification whereas those technologies were not
considered for those vehicles in the Draft TAR.
• A correction was made to the calculation of indirect costs on some transmission
technologies resulting in slightly lower TRX costs in this analysis (see discussion in
Section 2.3.4.2.4 of this TSD).
• Numerous updated effectiveness values including new ALPHA vehicle
determinations (i.e., termed the ALPHA "exemplar" vehicles). These changes are
detailed in Chapter 2 of this TSD.
• The OMEGA ICBT includes updated MOVES runs (taking into account AEO2016
projections) to generate new emission factors used on inputs to the OMEGA ICBT.
• The OMEGA ICBT now corrects an error which applied AEO reference fuel prices in
calculating monetized fuel savings, even in the AEO high and low fuel price cases.
• The OMEGA ICBT was updated to include payback calculations in the case where
loan purchases were used rather than simply cash purchases.
• The OMEGA ICBT payback analysis now applies vehicle survival rates to insurance
costs and loan payback costs. This places those costs on the same basis as the fuel
savings and maintenance costs which have always included vehicle survival rates.
5-20
-------
EPA's OMEGA Model
REFERENCES
1 EPA-420-R-12-024, August 2012.
2 EPA-420-R-09-016, September 2009. (Docket No. EPA-HQ-OAR-2010-0799-1135).
3 EPA-420-R-12-901, August 2012.
4 EPA-420-R-12-901, August 2012.
5-21
-------
Appendix
Table of Contents
Appendix A EPA Response to the Alliance of Automobile Manufacturers'
Contractor Reports Titled "Final Report for Technology Effectiveness [Phases 1 and 2]" A-l
A. 1 Constraints on Technology Combinations and Technological Innovation A-l
A.2 Novation's Simplistic Methodology and Lack of Rigor A-2
A.3 Omission of Vehicle Load and Technology Penetration Rate Changes A-3
A.4 Arbitrary and Restrictive Assumptions and Constraints A-4
A.5 Displacement Specific Load and Exemplars A-6
A.6 Other Studies A-7
Appendix B Fleet-Wide Analysis of Powertrain Efficiency for Current and
Future Technology Packages A-8
B. 1 Introduction A-8
B. 2 Methodol ogy A-8
B.2.1 Definition of Powertrain Efficiency A-8
B.2.2 Considering Tractive Energy Reductions for Future Technology Packages A-10
B.2.3 Displacement Specific Operating Load A-13
B.2.4 Choice of Drive Cycle A-14
B.3 Sample Calculation of Powertrain Efficiency A-15
Appendix C CO2 Targets with Current Powertrain Designs A-17
Appendix D EPA Comparison Testing performed on a MY2014 Mazda
SKYACTIV-G Engine using Different Fuels A-27
Table of Figures
Figure 1.1 Two engine BSFC maps, reproduced in Technology Effectiveness - Phase II: Vehicle-Level Assessment
and cited during the development of "Plausibility Test 2." The left-hand map is overlaid with areas of
typical on-cycle engine operation. Original sources are given in the Novation report A-6
Figure 3.1 Vehicle Production That Meets or Exceeds MY2020 Emission Targets, by Model Year A-18
Figure 4.1 Map of the Percentage Difference in Volumetric Fuel Flow for A MY2014 Mazda Skyactiv-G 2.0L 4-
Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2, 93 AKI, EO) versus
"Fuel B" (LEV III, 88 AKI, E10) A-30
Figure 4.2 Map of the Absolute Difference in Volumetric Fuel Flow (In Units of Ml/S) For A MY2014 Mazda
Skyactiv-G 2.0L 4-Cylinder Engine With A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,
93 AKI, EO) Versus "Fuel B" (LEV III, 88 AKI, E10) A-30
Figure 4.3 Map of the Percentage Difference in Fuel Mass Flow for A MY2014 Mazda Skyactiv-G 2.0L 4-Cylinder
Engine With A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2, 93 AKI, EO) versus "Fuel
B" (LEV III, 88 AKI, E10) A-3 1
Figure 4.4 Map of the Absolute Difference in Fuel Mass Flow (In Units of G/S) For A MY2014 Mazda Skyactiv-G
2.0L 4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2, 93 AKI,
EO) Versus "Fuel B" (LEV III, 88 AKI, E10) A-31
Figure 4.5 Map of the Percentage Difference in Volumetric Fuel Flow for A MY2014 Mazda Skyactiv-G 2.0L 4-
Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2, 93 AKI, EO) versus
"Fuel B" (LEV III, 88 AKI, E10) A-3 2
Figure 4.6 Map of the Absolute Difference in Volumetric Fuel Flow (In Units of Ml/S) for A MY2014 Mazda
Skyactiv-G 2.0L 4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,
93 AKI, EO) Versus "Fuel B" (LEV III, 88 AKI, E10) A-3 2
-------
Appendix
Figure 4.7 Map of the Percentage Difference in Fuel Mass Flow for A MY2014 Mazda Skyactiv-G 2.0L 4-Cylinder
Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2, 93 AKI, EO) versus "Fuel
B" (LEV III, 88 AKI, E10) A-33
Figure 4.8 Map of the Absolute Difference in Fuel Mass Flow (In Units of G/S) For A MY2014 Mazda Skyactiv-G
2.0L 4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2, 93 AKI,
EO) Versus "Fuel B" (LEV III, 88 AKI, E10) A-33
Figure 4.9 Map of the Percentage Difference in Brake Thermal Efficiency for A MY2014 Mazda Skyactiv-G 2.0L
4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2, 93 AKI, EO)
Versus "Fuel B" (LEV III, 88 AKI, E10) A-34
Figure 4.10 Map of the Percentage Difference in C02 Emissions for A MY2014 Mazda Skyactiv-G 2.0L 4-
Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2, 93 AKI, EO)
Versus "Fuel B" (LEV III, 88 AKI, E10) A-34
Table of Tables
Table 2.1 Technical Package Contents of Modeled Baseline and Modeled Compliance Toyota Camry A-15
Table 2.2 Reductions in Aero Drag, Tire Rolling Resistance, and Curb Weight for the Toyota Camry A-15
Table 2.3 Powertrain Efficiency and DSOL Calculations Inputs for Baseline Toyota Camry A-15
Table 2.4 Calculations of Estimated Aerodynamic Drag Area and Tire Rolling Resistance Coefficient for Baseline
Toyota Camry A-16
Table 2.5 Necessary Input Parameters for Calculation of Powertrain Efficiency and DSOL for Baseline and
Modeled Compliance Toyota Camry A-16
Table 2.6 Powertrain Efficiency and DSOL Calculations for Baseline and Modeled Compliance Toyota Camry.... 16
Table 3.1 Vehicles that Meet or Exceed Future Targets with Current Powertrain Designs A-19
Table 4.1 Measured Fuel Properties for Four Gasolines Used for Engine and Vehicle Benchmarking A-28
Table 4.2 Summary of C02 Emissions and CAFE Fuel Economy for Chassis Dynamometer Testing of The
MY2014 Mazda3 Equipped with A 2.0L Atkinson Cycle (13:1 Geometric CR) Engine Using a Tier 2
And A Tier 3 Certification Gasoline. Three Repeats of FTP75 (City Cycle) (Highway Cycle) And
95% Confidence Intervals Were Calculated Based Upon a Two-Sided T-Test A-35
-------
Appendix A - EPA Response to the Alliance of Automobile Manufacturers' Report
Appendix A EPA Response to the Alliance of Automobile Manufacturers'
Contractor Reports Titled "Final Report for Technology Effectiveness
[Phases 1 and 2]"
In its comments on EPA's technology, assessment and modeling processes in the Draft TAR,
the Alliance of Automobile Manufacturers states that the EPA projections of potential vehicle
and fleet effectiveness do not match "third-party modeling outputs."1 This claim (as well as
others scattered throughout the Alliance's comments) relies heavily on conclusions drawn in a
pair of non-peer-reviewed reports produced by The Alliance's contractor, Novation Analytics.
These reports provide some speculative conclusions, based on simple technology models, about
future vehicle effectiveness at both the fleet2 and vehicle level.3
Copies of the reports were provided by the Alliance as attachments to its comments. Pointing
to the Novation fleet-level report, the Alliance draws the conclusion that "MY2021 and MY2025
targets cannot be met with the suite of technologies at the deployment rates projected by the
agencies in the 2012 FRM"4 and that "automakers will need to apply additional and costlier
technologies than were initially predicted to meet the projected MY2021 and MY2025 targets."5
The EPA disagrees with the conclusions drawn by the Alliance. These reports by the
Alliance's contractor are riddled with technical flaws, unsound initial assumptions, and
unsubstantiated claims that substantially skew the final conclusions. Moreover, the errors in the
reports tend to systematically under-predict technology effectiveness and over-predict the cost
and complexity of the technology required to meet the standards. This opinion of Novation's
work is shared by Dave Cooke of the Union of Concerned Scientists, who outlines just a few of
the "fundamental mistakes that ensure that the report comes out the way the automakers
envisioned."6
A.l Constraints on Technology Combinations and Technological Innovation
The most basic of the "fundamental mistakes" in the report, and one that directly affects all of
the conclusions drawn by the Alliance on projected technology effectiveness, is the contention
that all possible technology available in 2025 can be represented by technology already
contained in the MY2014 baseline fleet.
Novation's report assumes from the outset that "the MY2014 fleet... includes the majority of
the spark ignition technology pathways utilized in the agency assumptions" and, therefore, "it is
not likely that the sales-weighted fleet performance [in MYs 2017-2025] will exceed the current
boundaries established by the best in class vehicles utilizing many of the technologies listed
above."7 This unsubstantiated initial assumption effectively limits powertrain efficiency in 2025
to small incremental improvements over that which is available today.
The EPA does not agree that MY2014 powertrain efficiency can define the maximum
achievable efficiency. Although it may be correct that "the majority of the spark ignition
technology" considered in the FRM exists in the present-day fleet (thus proving the viability of
individual technologies), the powertrain components incorporating these sub-technologies exist
in combinations and within packages that are designed to meet current standards, not future
standards.
A-l
-------
Appendix A - EPA Response to the Alliance of Automobile Manufacturers' Report
The LD GHG standards are phased in, with increasing stringency from year to year. These
standards do not require manufacturers to meet MY2025 standards in MY2014, and the EPA
anticipates that, for cost reasons, manufacturers will generally seek to minimize over-compliance
beyond the credit carryforward duration. Thus, the combinations of, and packages incorporating,
advanced technology that exist in the MY2014 fleet should be expected to be only as effective as
necessary to meet (or slightly over-comply withv8) MY2014 standards, but nowhere near the
effectiveness level required by 2025 standards. In later years, manufacturers have the ability to
incorporate additional technologies into their vehicles, and to recalibrate or refine existing
technologies, thereby increasing powertrain efficiency accordingly.
In later years, manufacturers also have the option to combine sub-technologies into packages
which are more effective than those that exist within the market today. EPA's projections of
effectiveness through MY2025 include technology packages that are achievable and cost-
effective, but do not exist in the fleet in MY2014 - for example, a 24 bar turbocharged
downsized engine with cooled EGR, or a high compression ratio Atkinson cycle engine with
cylinder deactivation and cooled EGR. The methodology in the Novation report does not allow
for the recombination of technologies represented by these packages, and thus severely and
unduly limits potential effectiveness increases obtainable by MY2025.
In fact, Novation's initial assumption on powertrain efficiency is equivalent to the argument
that because manufacturers are not substantially over-complying with current standards, they
could not possibly comply with more stringent future standards. This argument is unreasonable
on its face, and relies on the logical fallacy of circular reasoning, where the conclusion of an
argument is included within the initial assumptions.
A.2 Novation's Simplistic Methodology and Lack of Rigor
This fundamental flaw in the report's assumptions and conclusions results from the lack of
rigor in their "top-down" methodology (as pointed out by other organizations9). When correctly
implemented, "top-down and bottom-up approaches should converge to the same result"10 (as
the Novation report states). However, the choice to rely on vehicles, technologies, and
technology packages that exist in the MY2014 fleet produce a consistent bias that underestimates
potential technology effectiveness.
The Novation report oversimplifies the technologies, and the relationships among them, that
exist in the current fleet. The methodology within the report is to survey the MY2014 fleet,
grouping vehicles into broad "technology bundles" according to their powertrain. Within each
bundle, the underlying technology was assumed to be identical, and any differences among
powertrains attributed solely to "learning and implementation improvements."11 For example,
one "bundle" is defined as an SI naturally aspirated engine coupled with a non-high ratio spread
transmission, without stop-start. This bundle presumably includes vehicles with Atkinson cycle
engines or cylinder deactivation, yet ascribes any efficiency gains due to the advanced
technology to "learning."
The report then uses the statistical distribution of efficiency across all powertrains in each
"bundle" to estimate powertrain efficiency out to 2025, with average future efficiency set equal
A In MY2014, overall industry compliance was 13 grams/mile better than required by the 2014 GHG emissions
standards. This is consistent with the level of over-compliance in MYs 2012 and 2013.
A-2
-------
Appendix A - EPA Response to the Alliance of Automobile Manufacturers' Report
to the current 75th or 90th percentile. The simplistic assumption that "learning" is the source of
efficiency differences within each technology bundle obscures the actual effect of hardware and
technology differences among individual powertrains. Moreover, the assumption automatically
eliminates any consideration of the effect of recombining sub-technologies, as a "bottom-up"
methodology would.
The lack of rigor of the approach taken in the Novation report is immediately obvious if
individual components or sub-technologies are examined, rather than the entire powertrain as an
indissoluble package. At the highest level, powertrains are comprised of engines and
transmissions. Even in the MY2014 fleet, there are few if any 2014 best-in-class engines which
are packaged with 2014 best-in-class transmissions; and so even the 2014 best-in-class
powertrain underperforms what is clearly possible with off-the-shelf technology. At finer levels
of technology packaging, best-in-class engines and transmissions do not have all available
technology packaged on them.
In addition to constraining future powertrain packages to those technology combinations
existing within the MY2014 fleet, as stated above, the Novation report assumes that no
innovation will occur - no new technology will be implemented - in the eleven years until
MY2025. Although "the majority" of technologies discussed in the FRM exist in the MY2014
fleet, there are some that do not, but can be reasonably expected to be phased in before 2025. As
a single example, the Alliance in their comments acknowledges that "FCA US LLC (FCA)
recently introduced an upgraded 8-speed rear-wheel drive transmission"12 which incorporated
some elements of an advanced high efficiency gearbox (HEG2), improving upon the MY2014
best-in-class eight-speed transmission and reducing unadjusted combined fuel consumption by
approximately 0.8 percent. Moreover, the artificial limitation on innovation imposed in the
Novation report completely discounts the effect of further innovation in the industry (such as, for
example, Nissan's production-ready variable compression ratio engine, available in 201813),
which may provide further cost-effective reductions in GHG emissions and fuel consumption.
The Novation report assumes that new technologies like these (and others already announced by
manufacturers to be utilized on future products), along with the fuel consumption benefits
derived from them, would be impossible to incorporate in the future fleet.
In the few cases where the Novation report explicitly addresses technology not contained in
the MY2014 fleet, they invent arbitrary "proxies" to estimate powertrain efficiency. For
example, the Novation report arbitrarily claims that powertrains incorporating "the current
compression ignition (24-29 bar maximum BMEP diesel) can be used as a representative proxy"
for a 27 bar SI engine powertrain.14 No technical rationale for this choice is provided, and the
report again relies on circular reasoning by using the argument that "it is unlikely even an
advanced SI package will exceed the current CI efficiency boundary" to support the choice of
using current CI powertrain efficiencies as a proxy for 27 bar SI engine powertrain efficiencies.
A.3 Omission of Vehicle Load and Technology Penetration Rate Changes
In addition to consistently underestimating the potential effectiveness of advanced powertrain
technology, the Novation report compounds the errors by blindly following technology
projections in the 2012 FRM in circumstances where it is clearly not appropriate to do so.
The 2012 FRM projections are based on estimates of the most cost-effective technology
packages necessary to reach a sales-weighted target CO2 emission level, accounting for cost and
A-3
-------
Appendix A - EPA Response to the Alliance of Automobile Manufacturers' Report
effectiveness of powertrain technologies, cost and effectiveness of vehicle load reduction
technologies (mass, aerodynamic resistance, and rolling resistance), applied credits, and sales
mix of individual manufacturers. Altering the effectiveness of technologies, even as a sensitivity
study, by definition changes the associated cost-effectiveness. In an alternative world where
powertrain technology cost-effectiveness is different, the EPA would revise its modeling and
likely project a different mix of technologies in future fleets, as the cost effectiveness of each
technology would likely change in comparison to the others.
The Novation report attempts to quantify the technology penetration mix in an alternate world
where technology effectiveness is lower. However, in doing so, the Novation report
inappropriately maintains the original FRM assumptions on the non-powertrain portions of the
fleet projection while altering the powertrain assumptions. The result is that, even if the
powertrain efficiency estimation within the report were properly done, the alternate technology
mix in the future fleet is costlier than a reasonable methodology would predict, as the
methodology within the Novation report assesses neither road load reduction technologies,
changes in credits, nor cost.
A.4 Arbitrary and Restrictive Assumptions and Constraints
In their comments on the Draft TAR, the Alliance of Automobile Manufacturers also discuss
"modeling process issues" that they claim to be "the key source of error in technology benefit
estimates."15 To support this claim, the Alliance refers to statements in the vehicle-level report
from their contractors, Novation Analytics,16 where Novation attempts to justify the conclusions
contained in their fleet-level report by examining powertrain efficiency on a vehicle basis.
The vehicle-level report adds to the list of fundamental mistakes contained in the earlier
report by the same contractor. In addition to arbitrarily limiting technological progress to
combinations existing in the fleet in MY2014, this Novation report likewise depends throughout
on arbitrary assumptions and constraints which are largely unexplained, lacking in technical
foundation, or unsupported by scientific rationale.
In particular, many of the conclusions in the Novation report, which are repeated by the
Alliance in their comments, are based on the calculation of powertrain efficiency and the
application of what Novation claim to be "basic, and very liberal, plausibility checks"17 on the
limit of powertrain efficiency. There are indeed fundamental limits on efficiency, but
recognizing that efficiency is limited is a thermodynamic truism,18 and a principle that was never
in question. In fact, calculation of powertrain efficiency can serve as a gross QC check on
estimated technology effectiveness by quickly identifying the highest efficiency packages for
further review (as shown in Appendix B).
However, although examining powertrain efficiency can be useful, the Novation report further
attempts to establish hard numerical limits on this efficiency, and it is here that overly restrictive
assumptions creep in. Although the report claims to use "optimistic assumptions of technology
effectiveness potential and ample margin for uncertainty, so that the tests would allow all but the
most implausible results to pass,"19 the assumptions used to estimate plausibility limits are
unduly conservative and not at all optimistic. In fact, the Union of Concerned Scientists
identifies at least one current production vehicle, a Honda Fit, which would be deemed
implausible by the Novation report methodology.20
A-4
-------
Appendix A - EPA Response to the Alliance of Automobile Manufacturers' Report
As one example, to determine the limit of on-cycle-to-peak engine efficiency ratio
("Plausibility Test 2"), the Novation report calculates the efficiency ratios for the FTP, HWFET,
and combined cycles of three MY2013-2014 vehicles and selects the highest one. That efficiency
ratio is increased by a small amount to account for decreased fuel consumption due to stop-start,
decel fuel cutoff, and advanced transmissions based on "an independent [and uncited] analysis of
modal data... conducted on seven current generation vehicles." 21 The final numbers, based on
what appear to be seven to ten random vehicles from MY2013-2014, are presented as "very
liberal plausibility checks" 22 for MY2025 powertrains, and any results which "exceed these
ratios are judged to be implausible."23
This accounting process, if performed with care, could reasonably be expected to deliver a
quick, low fidelity efficiency estimate for a particular technology package. However, the process
is clearly inadequate as a bounding "plausibility test," and ignores substantial sources of
effectiveness discussed in the 2012 FRM and 2016 Draft TAR. For example, as part of the
explanation of this plausibility test, the Novation report reproduces a MY2013 Chevrolet Malibu
GDI engine map, overlaid with the operational area for a UDDS cycle (see Figure 1.1(a)). The
report correctly points out the gap between the engine operational area and the area of peak
efficiency in this map as an explanation for why the on-cycle-to-peak engine efficiency ratio
would be less than one.
In contrast, one substantial source of effectiveness in turbo downsized engines (compared to
their naturally aspirated counterparts) is that the area of peak efficiency in the map is pushed to
an area of lower speed and load (i.e., down and to the left), resulting in a much better match
between peak efficiency and the operational area. This can be seen by comparing a 27-bar BMEP
cooled EGR turbo GDI engine map (Figure 1.1(b), also reproduced in the Novation report), with
the Malibu map in Figure 1.1(a).
EPAJJDDS plotted on 11-Jul-2014
250
240
200
Gl
«D
£
_Q
150
"O
co
o
—I
Avg. load (bmep) and
speed decrease with
increasing displacement
100
1000
3000
4000
5000
6000
2000
2500
500
1000
1500
3000
3500
RPM RPM
(a) MY2013 Chevrolet Malibu 2.5L 14 GD (b) 27-bar BMEP cooled EGR turbo GDI
A-5
-------
Appendix A - EPA Response to the Alliance of Automobile Manufacturers' Report
Figure 1.1 Two engine BSFC maps, reproduced in Technology Effectiveness - Phase II: Vehicle-Level
Assessment and cited during the development of "Plausibility Test 2." The left-hand map is overlaid with
areas of typical on-cycle engine operation. Original sources are given in the Novation report.24
The better match between engine operation and peak efficiency reduces CO2 emissions,
precisely by increasing the on-cycle-to-peak engine efficiency ratio. Since the Novation report
develops a plausibility limit for on-cycle-to-peak engine efficiency ratio based on a few
MY2013-2014 vehicles, no room is left for potential improvement in the efficiency matching;
this is yet another example of the Novation report using an overly restrictive initial assumption to
dismiss potential technological improvement.
The Alliance suggests in their comments that the EPA implement an additional level of QC
check, beyond those detailed in the Draft TAR, based on the numerical limits given in the
Novation report.25 Although the EPA has used powertrain efficiency calculations as a QC tool
(see Appendix B), the EPA believes the numerical limitations on efficiency suggested in the
Novation report are not calculated in a robust and scientifically defendable way, and basing the
limits on current fleet data does not recognize the way in which the changing state of technology
affects these relationships as it develops over time. Therefore, the EPA declines to implement
the numerical limits from the Novation report.
A.5 Displacement Specific Load and Exemplars
The Alliance also claims in their comment that EPA modeling (specifically the Lumped
Parameter Model [LPM]) is "not based on the fundamental factors determining vehicle CO2 and
fuel consumption," quoting text from the vehicle-level Novation report.26 The "fundamental
factors" referred to are the incorporation of "displacement-specific load" (roughly correlated to
the inverse of power-to-weight ratio) as a factor in projecting technology effectiveness. The
Novation report further explains how changing engine displacement changes powertrain
efficiency and technology effectiveness, and specifically how technology benefits change as the
engine operational area changes.27
The EPA agrees that "displacement-specific load" is an important parameter in determining
technology effectiveness. However, both the Alliance and their contractor, Novation,
fundamentally misunderstand the purpose and usage of the LPM. In particular, the different
vehicle classes used in the LPM have different "exemplar" vehicles, each of which has different
engine sizes and road loads, and thus different displacement-specific load. When employing the
LPM, individual vehicles in the baseline fleet are mapped to the vehicle class, and the exemplar
vehicle, they most resemble. The EPA acknowledges that this modeling process is a
simplification, as are all models, and mapping different vehicles in the baseline fleet to the same
exemplar will produce both small over-estimates and small under-estimates of technology
effectiveness, depending on how close the baseline vehicle is to the exemplar used in the LPM.
However, on a fleet-wide average, these small over-and under-estimates of technology
effectiveness tend to average out.
The EPA's goal is to estimate technology effectiveness for individual vehicles and across the
fleet in the most representative and precise way possible. Therefore, for this Proposed
Determination, the EPA has redefined the vehicle effectiveness classes used in the LPM, based
in part on vehicle power-to-weight ratio, with the intention of producing effectiveness classes
containing vehicles with more similar road loads and engine sizes, as discussed in Section
A-6
-------
Appendix A - EPA Response to the Alliance of Automobile Manufacturers' Report
2.3.3.2. Exemplar characteristics have been defined based on sales-weighted averages of the
vehicles in each class. Moreover, the final effectiveness values for each individual vehicle have
been adjusted based on that vehicle's power-to-weight ratio. This process ensures the estimates
of technology effectiveness are closely representative of the individual vehicles within each
effectiveness class, while maintaining fleet-wide average projections of technology effectiveness
that are reflective of what would occur in the actual fleet. The methodology used to define the
new classes and exemplars is detailed in Section 2.3.1.4 of the TSD.
A.6 Other Studies
Along with the Novation report, the Alliance also cites a 2016 paper written by John Thomas
from Oak Ridge National Laboratory28 as supporting evidence, saying "Novation Analytics and
[John Thomas of] Oak Ridge National Laboratory agree that the technology penetrations selected
by the OMEGA and Volpe models in the 2012 FRM were insufficient for compliance in
MY2022-2025."29
However, the Alliance rather overstates the import and conclusions of this paper. In
particular, the Alliance neglects to mention the relationship between the Thomas paper and
Novation Analytics report, implying through omission that these are separate works. In fact, the
methodology in the Thomas paper is essentially identical to that in the Novation reports, and
Thomas states in his paper that the work "was inspired and focused by many discussions with
Gregg Pannone, Novation Analytics."
Furthermore, the Thomas paper is focused on calculating powertrain efficiency, with no
attempt to quantify the fleet mix necessary to meet the 2025 GHG standards, and no reference to
the "technology penetrations selected by... OMEGA" as the Alliance claims. The closest
reference to technology penetrations in the Thomas paper is the final conclusion, which refers
not to OMEGA results, but to the current fleet: "The path to meeting 2025 standards will likely
involve significantly larger numbers of hybrid electric powertrain vehicles and/or plug-in
vehicles being sold, compared to the current U.S. sales of such vehicles." Although this
conclusion is somewhat speculative (i.e., not discussed in the main body of the paper), the EPA
notes that the conclusion is not dissimilar to the projections in this Proposed Determination,
where the EPA projects MY2025 fleet penetrations of mild HEVs, strong HEVs, and BEVs
combined on the order of 25 percent (with more than two thirds of these being mild HEVs),
which far exceeds the number in the current fleet (see the Proposed Determination Federal
Register notice, Section IV). This does not derogate from the ultimate conclusion in the
Proposed Determination that there are compliance pathways to meet the 2025 standards that
involve chiefly advanced internal combustion engine technologies rather than strong hybrid or
full electrification.
A-7
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
Appendix B Fleet-Wide Analysis of Powertrain Efficiency for Current and
Future Technology Packages
B.l Introduction
In comments received on the Draft TAR, the Alliance of Automobile Manufacturers (AAM)
referenced work done by Novation Analytics to recommend that EPA implement "plausibility
checks" using a measure of powertrain efficiency and some estimated limitations on this
efficiency. As described in Appendix A, EPA believes the numerical limitations on efficiency
suggested in the Novation report are not calculated in a robust and scientifically defendable way,
and artificially limit potential effectiveness of powertrain components. However, EPA does
agree that the calculation of powertrain efficiency does serve as a valuable quality control (QC)
tool. For this Proposed Determination, in response to AAM's comments, EPA has incorporated
the calculation of powertrain efficiency into its QC process to confirm that the overall
effectiveness values applied in this analysis are appropriate.
The approach for this Proposed Determination utilizes data from the individual vehicles in the
MY2015 fleet to calculate a measure of powertrain efficiency, defined as the ratio of the energy
used to propel the vehicle over the combined test cycle to the fuel energy consumed. Powertrain
efficiency values are also calculated for all of the technology packages applied by the OMEGA
compliance model. From the distribution of those efficiency values across the fleet, a number of
vehicles are investigated closely to confirm that the incremental effectiveness estimates
generated by the ALPHA model are closely aligned with those produced by the Lumped
Parameter models, not only for the applied technology packages with typical efficiencies, but
also for those vehicles and packages with the highest efficiencies. This section describes how
powertrain efficiency was calculated, with additional discussion of the results and the QC
process provided in TDS Chapter 2.3.3.5.
B.2 Methodology
B. 2.1 Definition of Powertrain Efficiency
Powertrain efficiency (r|p), as defined by Thomas,30 is the ratio of the amount of propulsive
energy exerted by a vehicle over a given set of driving conditions to the energy content of the
expended fuel. The former term is also denoted as tractive energy (Etractive), while the latter is
denoted as fuel energy (Efcei). Therefore:
Eq. 1
Etractive
'P ~ p
Zfuel
Definition and Calculation of Tractive Energy
Thomas defines tractive energy (Etractive, also referred to as powertrain energy) as the energy
necessary propel the vehicle at a given rate while also overcoming the cumulative resistive forces
acting on it. The difference between these two terms is equal to the total tractive energy that the
vehicle exerts. Inertial energy (Einertiai) is used to calculate the former energy term, and it can be
A-8
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
determined using differential analysis of the drive cycle trace to obtain vehicle acceleration
(flcycie) which, in combination with the drive cycle's vehicle speed v(t) at each point in time, the
time increment dtcycie, and the vehicle test mass m yields:
Eq. 2
Einertial ~ ^t=o ^^"inerttaiCO — ^t=0 ^ * ^cycieCO * ^(0 * dtcycie
The resistive forces due to aerodynamic drag and tire rolling resistance, as well as internal
driveline friction are known as road load forces, which are overcome with the expenditure of
road load energy (EVoadioad). The magnitude of the road load force can be represented as a
function of the vehicle speed v(t), as well as the road load coefficients A, B, and C, representing
the components to the road load force independent of vehicle speed, proportional to vehicle
speed and proportional to the square of the vehicle speed, respectively (Eq. 3). The road load
energy can be calculated using Eq. 4. The total resistive energy is negative to represent resisting
vehicle motion.
Eq. 3
Froadload ~ A + Bv + Cv
Eq. 4
Eroadload = dEroadload(0 = ~(A + Bv(^ + Cv{t)2) * v(t) * dtcycle
During the drive cycle, braking events must be accounted for, as they represent points where
the engine is not directly supplying propulsive energy. Based on Thomas, this analysis assumes
that the vehicle is braking when the resistive road load energy alone cannot account for the
inertial energy of the vehicle when it is deaccelerating. In other words, for each increment of the
drive cycle:
Eq. 5
'tractive
tcycle
t=0
tcycle
1
(jiEineriiai(t) dEroa(Uoad (t)) [rf^tnertiaiCO —
t=0
Similarly, for brake energy (EWake):
Eq. 6
t-cycle
1brake
t=0
ccycle
(d£,jnertjaj(t) — d£Voadioad(0) [^^inertiaiCO — 0]
I
t=0
A-9
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
Definition and Calculation of Fuel Energy
In addition to estimating the tractive energy of the vehicle, the energy theoretically available
in the fuel to determine powertrain efficiency must also be calculated. On a per-unit of distance
traveled basis (here defined as fuel energy intensity £/uej), this is:
Eq. 7
p LHVfuelPfuel
fuel ~ MPG
The only quantity related to a particular drive cycle that is necessary in this calculation is the
fuel economy {MPG) over the given cycle (or harmonically averaged in the case of drive cycle
combinations). The relevant fuel properties in this analysis are the lower heating value of the
fuel (LHVfuei) and the fuel density (p&ei ). For this analysis, Tier 2 certification gasoline with a
lower heating value of 43.31 MJ/kg, and a density of 0.74 kg/L at 15°C was used to model
gasoline-fueled vehicles in the baseline and modeled compliance fleets.
To account for the per-distance aspect of the fuel energy intensity in Eq. 7, when calculating
powertrain efficiency, Eq. 1 is modified to utilize the vehicle's tractive energy intensity (energy
per unit of distance traveled) instead of the actual tractive energy. By averaging total tractive
energy over the entire distance traveled over the drive cycle (dcycie, obtained through integration
of the drive cycle trace), we can calculate the average vehicle tractive energy intensity:'
Eq. 8
p Etractive
^tractive,avg ~ H
acycle
Eq. 9
, _ ytcycle ,
u-cycle ~ *->t=0 UA/cycle
Combining those equations:
Eq. 10
E tractive,avg
B.2.2 Considering Tractive Energy Reductions for Future Technology Packages
Powertrain efficiency can be readily calculated for vehicles in the MY2015 fleet using the
Equivalent Test Weight and road load coefficient data submitted to EPA by manufacturers for
compliance certification. For future technology packages applied to vehicles in the OMEGA
compliance model, it is necessary to estimate the test mass and the road load coefficients to
account for mass reduction and reductions in tire rolling resistance and aerodynamic drag.
Estimating Vehicle Test Weight and Applying Mass Reduction
Each vehicle has with a curb weight Wcurb. which is used to denote the unloaded weight of the
vehicle. From there, a ballast weight (fFbaiiast, assumed to be 300 lbf.) is added to obtain the
vehicle weight for certification testing. The resulting loaded weight W\0&&, listed in Eq. 11, is
used to calculate vehicle test mass.
A-10
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
Eq. 11
Wioad ~ WCUrb ^ballast
For existing vehicles in the baseline fleet, the loaded weight term is assumed to be equal to
the vehicle's Equivalent Test Weight (ETW) consistent with EPA's two-cycle certification
tests.31 For future technology packages with mass reduction applied, the loaded weight must be
determined differently. Mass reduction is defined as a reduction in curb weight, so the ballast
weight must be subtracted from the loaded weight before the mass reduction is applied. For a
percent mass reduction AMR (%), the adjusted loaded weight W'load can be calculated from the
original loaded weight Wioado using Eq. 12. Consistent with the approach used in the LPM and
OMEGA models for characterizing the effectiveness benefits of mass reduction, the loaded
weight values for vehicles with future packages are not rounded into ETW bins.
Eq. 12
TAT/ ( (t.t t.t \ 100-AMR (%)\ ,
^ load ~ ^ballast J * 10q J ^ballast
The mass reduction applied to the baseline vehicle to yield the curb weight of the modeled
compliance vehicle AMR (%) is not directly specified by a particular technology package.
Instead, both the baseline vehicle technology package and the modeled compliance technology
package specify a net mass reduction relative to the curb weight of a null technology package
Wcurb,null, and this quantity is either equal to or greater than the curb weight of the baseline vehicle
Wcurb,base. The baseline curb weight that is reported for the baseline fleet is actually calculated by
applying an initial mass reduction to this null curb weight. That net mass reduction between the
null curb weight and the baseline curb weight, specified here as AMRnet,o (%), is more
specifically defined as:
Eq. 13
AMRnet0(%) = 100 * Wc"rb.null-Wcurb,base
Wcuj-j^null
The net mass reduction listed for the subsequent model compliance vehicles (i.e. the non-
baseline vehicle tech packages) is the net mass reduction applied to those vehicle packages
relative to the same null curb weight, resulting in the final vehicle curb weight fFcurb.
Eq. 14
AMRnet(%) = 100 * Wc"rb.null-Wcurb
Wcurb,null
The mass reduction AMR (%), therefore, is the mass reduction of the model compliance
vehicles relative to the curb weight of the baseline vehicle which, unlike the previous mass
reduction terms, is not defined relative to fFCurb,nuii. Using the above equations, we determine the
mass reduction between the baseline and the final tech package to be:
Eq. 15
AMRf1/1 Wcurb.base — Wcurfr AMRnet (%)—AMRng^o (%)
Wcurb.base 100—AMRnet0(%)
A-ll
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
Road Load Coefficient Estimation and Vehicle Resistive Force Analysis
While estimating vehicle test mass only requires knowing the net mass reduction specified by
a given technology package, estimating road load coefficients requires an understanding the
forces resisting the motion of the vehicle and the technologies that can affect them. As
previously stated, the road load force represents the sum of the forces that the resist the motion of
the vehicle. This analysis focuses on five sources of vehicle resistance and the forces associated
with them: aerodynamic drag (Fdrag), tire rolling resistance (Ftire), and mechanical drag from
brakes (/'brake), hubs (/'hub), and the neutral drag from the drivetrain (/'drive tram). The sum of all of
these resistive forces is denoted as the road load force /'roadioad:
Eq. 16
Proadload ~ Paero Pfire Pbrake Pfiub Pdrivetrain
Aerodynamic drag force is calculated as a function of air density (pa, 2.38e-3 slugs/ft3, taken
to be at STP), aerodynamic drag area {C&Af) and vehicle velocity (v):
Eq. 17
Paero ~ "^Pa^dAfV
The tire force, which can be estimated using Eq. 18, is dependent on the loaded test weight
Wo of the vehicle, the road grade (9=0°) and on the coefficient of tire rolling resistance Ctrr.
Eq. 18
Ptire = CTrr * Wload * COS(0)
Of the forces listed above, aerodynamic drag force and tire force are the most important
for determining changes in road load coefficient. As such, any attempt to estimate road load
coefficients requires having a relation between road load coefficients and these forces. Doing so
will allow changes in road load coefficient to be related to changes in drag and tire force or,
more specifically the vehicle drag area and tire rolling resistance coefficients.
To obtain an estimate for aerodynamic drag area, Eq. 16-18 are differentiated with respect to
velocity, and the differential contributions of resistive forces other than those of aerodynamic
drag are negated. This simplification can only be made if the vehicle speed is high enough to
allow aerodynamic drag to dominate the total resistive force acting on the vehicle. Hence, by
assuming a vehicle operating speed va of 110 km/h, aerodynamic drag area can be estimated as:
Eq. 19
Using the drag area calculated above, an estimation of the tire rolling resistance coefficient
can be obtained by using Eq. 3 and 16-18. An assumed vehicle speed of 50mph was used to
obtain values for both road load force and aerodynamic drag. The estimated contributions of
brake and hub drag per wheel were obtained from Backstrom32 and Shevket33 respectively,
assuming a wheel radius of 13in for both sources. Driveline drag forces were assumed to be a
constant 20N at an operating speed of 50 mph.
A-12
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
With a way to estimate both drag area and the coefficient of tire rolling resistance, there also
needs to be a way to relate those terms to the deductions in aerodynamic drag and tire rolling
resistance that are chosen by OMEGA for a particular package. The desired reduction in
aerodynamic drag and tire rolling resistance are defined similarly to mass reduction, in that they
are both defined relative to a null technology package. Therefore, just like with Eq. 15, the
desired aero drag and rolling resistance reductions between the baseline 2015 vehicle and the
modeled compliance vehicles are:
Eq. 20
AAerof0/) = CdAfbase~CdAf = AAer°net, (%)-AAeronet,o (%)
^dAfbase 100—AAeronet,o(%)
Eq. 21
ATRR (°/) ~ CTRR,base~cTRR _ ATRRnet (%)-ATRRnet,o (%)
cTRR,base 100-ATRRnet,o(%)
Estimating New Road Load coefficients
With a means by which to relate changes in vehicle parameters to changes in road load
coefficients, it is now possible to estimate road load coefficients based on changes to changes in
vehicle mass, drag area, and tire rolling resistance. The B coefficient is assumed to be constant,
while the C coefficient changes is proportion to AAERO (%>), which is the reduction in
aerodynamic drag area (CdA r), given the relation of both terms to the square of the velocity of
the vehicles. The change in the^4 coefficient can be determined by solving Eq. 16-18 for^4 and
neglect the aerodynamic drag, brake, hub, and powertrain contributions to the change. The
calculations of the adjusted coefficients (A1, B\ C") from the original baseline coefficients (A0
,B0,Co) are shown in Eq. 23 and 24.
Eq. 22
»' = *.- W0CTRRc (lOO - (M«2) . (H2^)) +
Wbauas,CTRR0 (^r) (WO - ATRR(%))/100)
Eq. 23
B' = B0
Eq. 24
, = c (!_«)
O V 100 J
B. 2.3 Displacement Specific Operating Load
After calculating vehicle powertrain efficiency, there needs to be a way to group vehicles
based on their power-to-weight ratios. As a rough means of doing so, the "displacement specific
operational load" or DSOL was utilized, defined here as the ratio of average cycle tractive road
power Ptractive to maximum rated engine power Prated:
A-13
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
Eq. 25
DSQL = Ptractive
Prated.
While the definition of tractive power is consistent with that of tractive energy, which was
calculated to determine powertrain efficiency, Eq. 25 must be modified to utilize the total
amount of tractive energy used by the vehicle over the cycle due to the fluctuation in driving
conditions and vehicle speed. As such, the maximum rated engine energy .EVated can be defined
as the total energy exerted by the engine operated at its rated horsepower for the entire drive
cycle:
Eq. 26
c — p # ytcycle _ p .
'-'rated ~ 1 rated £->t=0 — ratedLcycle
Therefore:
Eq. 27
Etractive
DSOL =
Erated
In this analysis, the modeled compliance vehicle packages are assumed to retain the same
DSOL values as their baseline fleet counterparts.
B. 2.4 Choice of Drive Cycle
This analysis applies the combined city (FTP) cycle and highway (HWFET) cycle34 using a
55 percent city/45 percent highway cycle weighting. This yields a combined cycle (comb) fuel
economy as a weighted harmonic average of city (C) and highway (H) fuel economy values:
Eq. 28
MPG
comb 0.55
MPGC MPGh
Estimates for combined fuel economy can also be made for gasoline vehicles based on the
combined cycle CO2 emissions for the vehicle. These emission numbers are calculated for all
baseline and projected 2025 vehicles through OMEGA. Based on EPA's correlative estimates35,
the combined cycle vehicle unadjusted fuel economy can be approximated as:
MPG
Eq. 29
8887
comb ~ rn
c u2,combined
To account for the combined cycle in our calculations of tractive road energy and DSOL, a
weighted average of the corresponding quantities for city and highway drive cycles is used.
Thus those quantities for combine cycle analysis are defined as:
Eq. 30
^tractive,avg (Comb) ~ (0.55 * £"tracttve,av5(C)) (0.45 * ^tractive.avg (//))
A-14
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
Eq. 31
DSOLComb = (0.55 * DSOLc) + (0.45 * DSOLH)
B.3 Sample Calculation of Powertrain Efficiency
To demonstrate the principles described above in action, a step-by-step calculation of
powertrain efficiency and DSOL is shown below. The baseline fleet vehicle chosen was the
Toyota Camry (Baseline Entry 2266), given its mid-tier baseline efficiency and the significant
vehicle sales in 2015. It also had a sales package in the 2025MY fleet.
Along with the baseline package, this example analysis will also contain a modeled
compliance technology package applied to the Camry; specifically, OMEGA package TP07,
which was mentioned above. The technologies present in both vehicles is shown in Table 2.1.
Table 2.1 Technical Package Contents of Modeled Baseline and Modeled Compliance Toyota Camry
Tech
Pkg.
Tech Package Contents
TPOO
| LUB | EFR11141VVT |TRX111 LRRT11SAX-NA| WRtech- 1.51 WRpen- 01 WRnet-1.51
TP07
| EFR21141 VVT| Deac-141TRX221IACC21 EPS | Aero21 LRRT21 LDB SAX-NA| WRtech- 2.51 WRpen- 01 WRnet- 2.51
Here, we see that the baseline Camry (TP00) has an initial curb weight reduction of 1.5
percent, (WRnet) tire rolling resistance reduction of 10 percent (LRRT1), and no aerodynamic
drag reduction relative to the null technology package. The 2025MY GHG standard's compliance
analysis technology package (TP07) has reductions to all of these categories relative to the null:
20 percent to aero drag (AER02), 2.5 percent to curb weight (WRnet- 2.5), and 20 percent to tire
rolling resistance (LRRT2). Using these numbers, Eq. 15 and Eqs. 20-21 are used to calculate
percent reductions in aero drag, curb weight, and tire rolling resistance between the baseline
vehicle and the modeled compliance vehicle. The results are presented in Table 2.2
Table 2.2 Reductions in Aero Drag, Tire Rolling Resistance, and Curb Weight for the Toyota Camry
Tech
Package
Curb Weight
Reduction from
Null (%)
Drag Area
Reduction from
Null (%)
Tire Rolling
Resistance
Reduction from
Null (%)
Curb Weight
Reduction
from Base
(%)
Drag Area
Reduction
from Base
(%)
Tire Rolling
Resistance
Reduction
from Base (%)
TPOO
1.5
0
10
0
0
0
TP07
2.5
20
20
1.02
20
11.1
The original listed parameters necessary for calculating powertrain efficiency and DSOL are
shown in Table 2.3
Table 2.3 Powertrain Efficiency and DSOL Calculations Inputs for Baseline Toyota Camry
A (Ibf)
B(lbf/mph)
C(lbf/mph2)
ETW (Ibf)
C02 Emissions
(g/mi)
Rated
Horsepower (hp)
TPOO
27.23
0.04319
0.01937
3500
237.5
178.0
Using Eqs. 16-19, the estimated drag area and tire rolling resistance coefficients can be
calculated. The results are shown in Table 2.4
A-15
-------
Appendix B - Fleet-Wide Analysis of Powertrain Efficiency for Technology Packages
Table 2.4 Calculations of Estimated Aerodynamic Drag Area and Tire Rolling Resistance Coefficient for
Baseline Toyota Camry
Estimated
Drag
Total
Brake
Hub
Neutral
Tire
Estimated
Drag Area
Force
Roadload
Drag
Drag
Drag
Resistance
Tire Rolling
(ft2)
(Ibf)
Force
Force
Force
Force
Force (Ibf)
Resistance
(Ibf)
(Ibf)
(Ibf)
(Ibf)
Coefficient
(kg/lOOOkg)
TPOO
7.70
49.22
77.83
1.09
4.90
4.50
18.14
5.18
The values in Table 2.2, Table 2.3, and Table 2.4, along with Eqs. 22-24, we can obtain the
necessary input parameters for calculating the powertrain efficiency of the modified compliance
package and obtain the corresponding C02 emission from OMEGA.
Table 2.5 Necessary Input Parameters for Calculation of Powertrain Efficiency and DSOL for Baseline and
Modeled Compliance Toyota Camry
A (Ibf)
B(lbf/mph)
C(lbf/mph2)
ETW (Ibf)
C02 Emissions
(g/mi)
TPOO
27.23
0.04319
0.019374
3500
237.5
TP09
25.07
0.04319
0.015499
3468
176.5
From these parameters, the calculations to determine powertrain efficiency are performed
using Eqs. 1-10 and Eqs. 26-31, and the results are listed in Table 2.6.
Table 2.6 Powertrain Efficiency and DSOL Calculations for Baseline and Modeled Compliance Toyota
Camry
Tech
Pkg
Tractive
Energy
(kWhr)
Tractive Road
Energy Intensity
(MJ/km)
TPOO Rated
Engine Energy
(kWhr)
Combined Cycle
Cty
Hwy
Cty
Hwy
Cty
Hwy
TPOO
DSOL
(*le-2)
Tractive Road
Energy
Intensity
(MJ/km)
Fuel
Energy
Intensity
(MJ/km)
Powertrain
Efficiency
(%)
TPOO
2.865
1.828
0.4264
0.3981
101.2
28.21
4.46
0.4137
2.015
20.53
TP09
2.702
1.595
0.4020
0.3474
-
-
-
0.3774
1.497
25.21
A-16
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Appendix C CO2 Targets with Current Powertrain Designs
How Many of Today's Vehicles Can Meet or Surpass the MY2017-2025 Footprint-based
C02 Targets with Current Powertrain Designs?
As part of this evaluation of the feasibility of the MY2017 to MY2025 standards, EPA
updated its analysis of individual vehicles being sold today against the future footprint-based
standards. This analysis compares MY2016 and earlier vehicles to the footprint-based standard
curves to determine which of these vehicles will meet or be lower than the final MY2017 -
MY2025 footprint-based CO2 targets. The results show that a wide range of current vehicles
would already meet or exceed future standards.
Using publicly available data36, EPA compiled a list of all available vehicles and their 2-cycle
CO2 g/mile performance (that is, the performance over the city and highway compliance tests).
No adjustments were made to vehicle CO2 performance. EPA applied increasing air conditioner
credits over time with a phase-in of alternative refrigerant for the generation of HFC leakage
reduction credits consistent with the assumed phase-in schedule published in Table C.6 of the
Proposed Determination Appendix, Section C. Vehicle footprint data was gathered by EPA from
manufacturer submitted CAFE reports37 and manufacturer websites. The analysis here focuses
on MY2016 and prior model years, since MY2016 is the most current complete model year.
Production data for MY2016 is based on estimates provided by manufacturers.
As shown in Figure 3.1, approximately 17 percent of MY2016 vehicles already meet or are
below the MY2020 standards, given current powertrain performance and air conditioning credits.
This represents more than 2.5 million current MY2016 vehicles. It is also important to note that
not all vehicles are required to be below their individual targets, and in fact EPA expects that
manufacturers will be able to comply with the standards with roughly 50 percent of their
production meeting or falling below the footprint based targets.
Manufacturers do have additional opportunities to generate "off-cycle" credits for reduced
GHG emissions that are not captured on EPA test cycles. If an additional 5 g/mile credit is
applied to all vehicles to account for off-cycle credits, the percentage of MY2016 vehicles that
meet or are below the MY2020 targets increases from 17 percent to 21 percent. In MY2015,
manufacturers reported an average of 3.0 g/mile38 with several manufacturers already above 4
g/mile, so an assumption of 5 g/mile of off-cycle credits in MY2020 is likely conservative.
Figure 3.1 also shows that the number of vehicles that meet future standards has been steadily
increasing with each passing model year. EPA analysis showed that approximately 5 percent of
MY2012 vehicles achieved or were lower than the footprint based MY2020 targets. For
MY2016 vehicles, that percentage of vehicles increased to 17 percent (or 21 percent including
off-cycle credits). In MY2012 the large majority of vehicles that met or were below the
MY2020 targets were hybrid-electric vehicles, but the majority of MY2016 vehicles meeting
MY2020 targets are gasoline, non-hybrid vehicles.
A-17
-------
Appendix C - CO2 Targets with Current Powertrain Designs
25%
¦ Gasoline
¦ Diesel
¦ HEV
¦ PHEV
¦ Fuel Cell
¦ EV
¦ CNG
MY2012 MY2016 MY2016
(with off-cycle credits)
Figure 3.1 Vehicle Production That Meets or Exceeds MY2020 Emission Targets, by Model Year
Table 3.1 shows that more than 100 individual MY2016 vehicle versions already meet or are
below the 2020 CO2 footprint target levels, with current powertrain designs and air conditioning
credit generation consistent with the 2012 final rule. The table highlights the vehicles with CO2
emissions that meet or are lower than the applicable footprint targets from MY2017 to MY2025
in green, and shows the percentage below the target for each model year. Vehicles that are
above, but within 5 percent of the targets are highlighted in yellow.
The list of vehicles includes nearly every vehicle type, including midsize cars, sport utility
vehicles, and pickup trucks. The vehicles already at or below MY2020 targets also includes
vehicles utilizing a variety of powertrain options, including gasoline internal combustion
engines, hybrid-electric, plug-in hybrid-electric, and full electric options. Multiple fuel options
are also present, including gasoline, diesel, hydrogen, and electricity. Nearly every major
manufacturer produces some vehicles that would meet or be lower than the MY2020 footprint
CO2 target with only simple improvements in air conditioning systems.
A-18
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Table 3.1 Vehicles that Meet or Exceed Future Targets with Current Powertrain Designs
Model
Year
Manufacturer
Vehicle
Fuel
Economy
(mpg)
Tailpipe
C02
(g/mile)
Footprint
(ft2)
Powertrain
Type
Trans-
mission
Engine
Disp.
(L)
Vehicle Class
Car/
Truck
Compliance
2017
2018
2019
2020
2021
2022
2023
2024
2025
2016
BMW
13 BEV
0
43.5
EV
A1
Subcompact Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Chevrolet
Spark EV
0
35.8
EV
A1
Subcompact Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
FCA
500e
0
34.7
EV
A1
Mini com pact Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Ford
Focus Electric FWD
0
43.5
EV
A1
Compact Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Kia
Soul Electric
0
43.3
EV
A1
Small Station Wagons
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Mercedes-Benz
B250e
0
49.8
EV
A1
Midsize Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Mercedes-Benz
smart fortwo elec. drive (conv.)
0
26.8
EV
A1
Two Seaters
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Mercedes-Benz
smart fortwo elec. drive (coupe)
0
26.8
EV
A1
Two Seaters
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Mitsubishi
i-MiEV
0
38.4
EV
A1
Subcompact Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Nissan
Leaf (24 kW-hr battery pack)
0
44.7
EV
A1
Midsize Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Nissan
Leaf (30 kW-hr battery pack)
0
44.7
EV
A1
Midsize Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Volkswagen
e-Golf
0
42.4
EV
A1
Compact Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model S (70 kW-hr battery pack)
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model S (85 kW-hr battery pack)
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model S (90 kW-hr battery pack)
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model SAWD-70D
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model SAWD-75D
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model SAWD-85D
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model SAWD-90D
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model S AWD - P85D
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model S AWD - P90D
0
53.5
EV
A1
Large Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model X AWD-75D
0
53.6
EV
A1
Sport Utility Vehicles
T
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model X AWD-90D
0
53.6
EV
A1
Sport Utility Vehicles
T
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Tesla
Model X AWD - P90D
0
53.6
EV
A1
Sport Utility Vehicles
T
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Toyota
Mirai
0
46.0
Fuel Cell
AV
Subcompact Cars
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
Hyundai
Tuscon
0
45.0
Fuel Cell
A1
Sport Utility Vehicles
C
100%
100%
100%
100%
100%
100%
100%
100%
100%
2016
BMW
13 REX
132.2
23
43.5
PHEV
A1
0.6
Subcompact Cars
c
95%
96%
96%
97%
97%
97%
97%
97%
97%
2016
Chevrolet
Volt
115.6
30
46.2
PHEV
AV
1.5
Compact Cars
c
92%
92%
93%
93%
94%
93%
93%
93%
92%
2016
Cadillac
ELR
82.8
54
45.9
PHEV
AV
1.4
Subcompact Cars
c
81%
81%
80%
80%
80%
79%
78%
77%
76%
2016
Toyota
Prius Eco
80.8
110
44.6
HEV
AV
1.8
Midsize Cars
c
81%
81%
80%
80%
80%
79%
78%
77%
76%
2016
Cadillac
ELR Sport
82.8
54
45.9
PHEV
AV
1.4
Subcompact Cars
c
81%
81%
80%
80%
80%
79%
78%
77%
76%
2016
Hyundai
Sonata
87.5
54
48.3
PHEV
AM6
2.0
Midsize Cars
c
78%
77%
77%
76%
76%
74%
73%
72%
71%
2016
Ford
Fusion
74.8
86
48.7
PHEV
AV
2.0
Midsize Cars
c
68%
67%
66%
65%
64%
62%
61%
59%
57%
2016
Ford
C-MAX
74.8
86
43.8
PHEV
AV
2.0
Midsize Cars
c
65%
64%
62%
61%
60%
58%
56%
54%
52%
2016
Audi
A3 e-tron ultra
64.3
90
43.7
PHEV
AM-S6
1.4
Compact Cars
c
63%
62%
60%
59%
57%
56%
53%
51%
49%
2016
Audi
A3 e-tron
64.3
105
43.7
PHEV
AM-S6
1.4
Compact Cars
c
55%
54%
52%
51%
49%
46%
44%
41%
38%
2016
BMW
18
56.6
125
48.5
PHEV
A6
1.5
Subcompact Cars
c
51%
49%
47%
45%
43%
40%
38%
35%
31%
2016
Volvo
XC90AWD
42.4
166
53.5
PHEV
S8
2.0
Sport Utility Vehicles
T
48%
48%
48%
47%
43%
40%
37%
34%
31%
2016
Toyota
Prius
74.0
120
44.6
HEV
AV
1.8
Midsize Cars
c
49%
47%
45%
43%
41%
38%
35%
32%
29%
2016
BMW
X5xDrive40e
42.6
169
52.0
PHEV
S8
2.0
Sport Utility Vehicles
T
46%
46%
45%
44%
40%
37%
34%
31%
27%
A-19
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Model
Year
Manufacturer
Vehicle
Fuel
Economy
(mpg)
Tailpipe
C02
(g/mile)
Footprint
(ft2)
Powertrain
Type
Trans-
mission
Engine
Disp.
(L)
Vehicle Class
Car/
Truck
Compliance
2017
2018
2019
2020
2021
2022
2023
2024
2025
2016
BMW
330e
55.0
128
46.9
PHEV
S8
2.0
Compact Cars
C
48%
46%
44%
42%
39%
37%
34%
30%
27%
2016
Porsche
Cayenne S e-Hybrid
37.5
183
51.8
PHEV
AM8
3.0
Sport Utility Vehicles
T
41%
41%
40%
39%
34%
31%
27%
24%
20%
2016
Chevrolet
Malibu
61.5
145
48.4
HEV
AV
1.8
Midsize Cars
C
42%
40%
37%
35%
32%
29%
26%
22%
19%
2016
Toyota
Prius c
70.8
126
40.6
HEV
AV
1.5
Compact Cars
C
42%
40%
37%
35%
32%
29%
26%
22%
18%
2016
Ford
Fusion Hybrid
59.6
149
48.4
HEV
AV
2.0
Midsize Cars
c
40%
38%
35%
33%
30%
27%
23%
20%
16%
2016
Lincoln
MKZ Hybrid
59.6
149
48.4
HEV
AV
2.0
Midsize Cars
c
40%
38%
35%
33%
30%
27%
23%
20%
16%
2016
Porsche
Panamera S E-Hybrid
43.8
161
52.2
PHEV
AM-S8
3.0
Large Cars
c
40%
37%
35%
32%
29%
26%
22%
19%
15%
2016
Hyundai
Sonata Hybrid SE
58.1
153
48.0
HEV
AM6
2.0
Midsize Cars
c
38%
36%
33%
30%
27%
24%
20%
17%
13%
2016
Toyota
Prius v
58.9
151
46.1
HEV
AV
1.8
Midsize Station Wagons
c
37%
34%
31%
29%
25%
22%
18%
14%
10%
2016
Toyota
Camry Hybrid LE
57.4
155
47.2
HEV
AV
2.5
Midsize Cars
c
36%
34%
31%
28%
25%
21%
18%
14%
10%
2016
Volkswagen
Jetta Hybrid
60.8
146
44.0
HEV
AM-S7
1.4
Compact Cars
c
36%
33%
30%
28%
24%
21%
17%
13%
9%
2016
Hyundai
Sonata Hybrid
56.3
158
47.8
HEV
AM6
2.0
Midsize Cars
c
36%
33%
30%
27%
24%
21%
17%
13%
9%
2016
Lexus
ES 300h
55.2
161
48.0
HEV
AV-S6
2.5
Midsize Cars
c
35%
32%
29%
26%
23%
19%
15%
12%
7%
2016
Mercedes
S 550e
182
54.6
PHEV
A7
3.0
Large Cars
c
34%
31%
28%
26%
22%
19%
15%
11%
7%
2016
Toyota
Avalon Hybrid
55.2
161
47.7
HEV
AV-S6
2.5
Midsize Cars
c
34%
31%
28%
26%
22%
19%
15%
11%
7%
2016
Toyota
Camry Hybrid XLE/SE
54.9
162
47.2
HEV
AV
2.5
Midsize Cars
c
33%
30%
27%
24%
21%
17%
13%
9%
5%
2016
Lexus
CT 200h
57.5
155
42.7
HEV
AV
1.8
Compact Cars
c
30%
27%
24%
21%
17%
13%
9%
5%
0%
2016
Kia
Optima HYBRID EX
51.5
172
48.2
HEV
AM6
2.4
Midsize Cars
c
30%
27%
24%
20%
17%
13%
9%
5%
0%
2016
Toyota
RAV4 Hybrid AWD
44.7
199
44.9
HEV
AV-S6
2.5
Sport Utility Vehicles
T
27%
26%
25%
23%
18%
14%
9%
4%
-1%
2016
Lexus
RX 450h
41.8
213
48.0
HEV
AV-S6
3.5
Sport Utility Vehicles
T
26%
25%
24%
22%
16%
12%
7%
3%
-2%
2016
Ford
C-MAX Hybrid FWD
55.0
162
43.8
HEV
AV
2.0
Large Cars
c
28%
25%
22%
19%
15%
11%
7%
2%
-2%
2016
Kia
Optima Hybrid EX
50.4
176
48.2
HEV
AM6
2.4
Midsize Cars
c
28%
25%
22%
18%
15%
11%
7%
2%
-2%
2016
Lexus
NX 300h AWD
43.5
204
45.1
HEV
AV-S6
2.5
Sport Utility Vehicles
T
25%
24%
23%
21%
15%
11%
6%
2%
-3%
2016
Ford
F1502WD
28.2
315
76.8
Gasoline
S6
2.7
Standard Pick-up Trucks
T
13%
13%
13%
13%
13%
9%
5%
0%
-5%
2016
Lexus
RX 450h AWD
40.8
218
48.0
HEV
AV-S6
3.5
Sport Utility Vehicles
T
24%
23%
22%
20%
14%
10%
5%
0%
2016
Ford
F1502WD
28.2
315
73.6
Gasoline
S6
2.7
Standard Pick-up Trucks
T
13%
13%
13%
13%
13%
9%
4%
0%
2016
Toyota
Highlander Hybrid AWD LE Plus
38.9
229
48.9
HEV
AV-S6
3.5
Sport Utility Vehicles
T
22%
20%
19%
17%
11%
6%
1%
-4%
2016
Honda
CR-Z
51.1
174
44.5
HEV
AV-S7
1.5
Two Seaters
C
23%
20%
17%
13%
9%
5%
0%
-4%
2016
Toyota
Highlander Hybrid AWD
38.7
230
48.9
HEV
AV-S6
3.5
Sport Utility Vehicles
T
21%
20%
18%
16%
10%
6%
1%
-4%
2016
Subaru
Crosstrek Hybrid
42.5
209
43.2
HEV
AV-S6
2.0
Sport Utility Vehicles
T
21%
19%
18%
16%
10%
5%
0%
2016
Mercedes-Benz
GLA 250
38.2
233
49.0
Gasoline
AM7
2.0
Midsize Station Wagons
T
20%
19%
17%
15%
9%
5%
0%
2016
Infiniti
QX60 Hybrid AWD
36.1
246
52.1
HEV
AV-S7
2.5
Sport Utility Vehicles
T
20%
18%
17%
15%
9%
4%
-1%
2016
Ford
F1502WD FFV
26.6
335
76.8
Gasoline
A6
3.5
Standard Pick-up Trucks
T
7%
7%
7%
7%
7%
3%
-2%
2016
Ford
F1502WD
28.2
315
68.1
Gasoline
S6
2.7
Standard Pick-up Trucks
T
13%
13%
13%
13%
7%
2%
-3%
2016
Nissan
Murano Hybrid AWD
36.8
242
48.9
HEV
AV-S7
2.5
Midsize Station Wagons
T
17%
15%
14%
12%
5%
0%
-5%
2016
Honda
Civic4Dr
48.0
185
45.2
Gasoline
AV
1.5
Midsize Cars
C
19%
16%
12%
8%
4%
0%
2016
Honda
Civic 2Dr
47.9
186
45.2
Gasoline
AV
1.5
Compact Cars
C
19%
16%
12%
8%
4%
-1%
2016
Ford
F1502WD
28.2
315
66.2
Gasoline
S6
2.7
Standard Pick-up Trucks
T
13%
13%
13%
11%
4%
-1%
2016
Ford
F1502WD
25.5
349
76.8
Gasoline
S6
3.5
Standard Pick-up Trucks
T
3%
3%
3%
3%
3%
-1%
2016
Mercedes-Benz
GLA 250 4M ATI C
36.0
247
49.0
Gasoline
AM7
2.0
Sport Utility Vehicles
T
15%
13%
12%
10%
3%
-2%
2016
Mercedes-Benz
GLA 250 4M ATI C
36.0
247
49.0
Gasoline
AM7
2.0
Sport Utility Vehicles
T
15%
13%
12%
10%
3%
-2%
2016
Mazda
Mazda 6
43.1
206
49.4
Gasoline
S6
2.5
Midsize Cars
C
17%
13%
10%
6%
1%
-3%
2016
Honda
Civic4Dr
46.8
190
45.2
Gasoline
AV
2.0
Midsize Cars
C
17%
14%
10%
6%
1%
-3%
2016
Ford
F1504WD FFV
24.9
357
76.8
Gasoline
A6
3.5
Standard Pick-up Trucks
T
1%
1%
1%
1%
1%
-4%
A-20
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Model
Year
Manufacturer
Vehicle
Fuel
Economy
(mpg)
Tailpipe
C02
(g/mile)
Footprint
(ft2)
Powertrain
Type
Trans-
mission
Engine
Disp.
(L)
Vehicle Class
Car/
Truck
Compliance
2017
2018
2019
2020
2021
2022
2023
2024
2025
2016
Volvo
XC90 FWD
32.9
270
53.5
Gasoline
S8
2.0
Sport Utility Vehicles
T
13%
12%
10%
8%
1%
-4%
2016
Subaru
Outback
37.4
237
45.7
Gasoline
AV-S6
2.5
Sport Utility Vehicles
T
14%
12%
10%
8%
1%
-4%
2016
Chevrolet
City Express Cargo Van
34.9
254
49.6
Gasoline
AV
2.0
Vans
T
13%
11%
10%
8%
1%
-4%
2016
Nissan
Rogue AWD
36.8
242
46.4
Gasoline
AV
2.5
Sport Utility Vehicles
T
13%
11%
10%
7%
1%
-5%
2016
Ford
F1502WD FFV
26.6
335
68.1
Gasoline
A6
3.5
Standard Pick-up Trucks
T
7%
7%
7%
7%
0%
-5%
2016
Ford
F1502WD FFV
26.6
335
68.1
Gasoline
A6
3.5
Standard Pick-up Trucks
T
7%
7%
7%
7%
0%
-5%
2016
Honda
HR-V4WD
38.9
229
43.2
Gasoline
AV-S7
1.8
Sport Utility Vehicles
T
13%
11%
10%
7%
0%
-5%
2016
Honda
HR-V4WD
38.9
229
43.2
Gasoline
AV
1.8
Sport Utility Vehicles
T
13%
11%
10%
7%
0%
-5%
2016
BMW
X3xDrive28d
40.1
254
49.1
Diesel
S8
2.0
Sport Utility Vehicles
T
13%
11%
9%
7%
0%
-5%
2016
Chevrolet
Cruze
46.6
191
44.8
Gasoline
S6
1.4
Compact Cars
C
16%
12%
8%
5%
0%
-5%
2016
Chevrolet
Cruze Premier
46.6
191
44.8
Gasoline
S6
1.4
Compact Cars
C
16%
12%
8%
5%
0%
-5%
2016
Mazda
Mazda 2
50.4
176
39.4
Gasoline
S6
1.5
Compact Cars
c
16%
12%
8%
4%
0%
-5%
2016
Honda
Civic 2Dr
46.3
192
45.2
Gasoline
AV
2.0
Compact Cars
c
16%
12%
8%
4%
0%
-5%
2016
Mazda
Mazda 3 4-Door
46.0
193
45.3
Gasoline
S6
2.0
Compact Cars
c
16%
12%
8%
4%
0%
2016
Subaru
Crosstrek
38.6
230
43.2
Gasoline
AV-S6
2.0
Sport Utility Vehicles
T
12%
10%
9%
6%
-1%
2016
Mercedes-Benz
Smart fortwo (Coupe)
50.2
177
26.8
Gasoline
AM6
0.9
Two Seaters
c
16%
12%
8%
4%
-1%
2016
Honda
FIT
50.2
177
40.2
Gasoline
AV
1.5
Small Station Wagons
c
16%
12%
8%
4%
-1%
2016
Nissan
Altima
44.1
202
47.4
Gasoline
AV
2.5
Midsize Cars
c
16%
12%
8%
4%
-1%
2016
Toyota
Corolla LE ECO
46.8
190
44.2
Gasoline
AV
1.8
Midsize Cars
c
15%
12%
8%
4%
-1%
2016
Chevrolet
Colorado 2WD Crew Cab, Long Bed
33.0
308
60.9
Diesel
A6
2.8
Small Pick-up Trucks
T
9%
9%
8%
6%
-1%
2016
GMC
Canyon 2WD Crew Cab, Long Box
33.0
308
60.9
Diesel
A6
2.8
Sport Utility Vehicles
T
9%
9%
8%
6%
-1%
2016
Ram
Pro master City
31.5
282
54.9
Gasoline
A9
2.4
Vans
T
11%
10%
8%
6%
-1%
2016
Scion
iA
49.8
178
40.8
Gasoline
S6
1.5
Subcompact Cars
C
15%
11%
7%
3%
-1%
2016
Ford
Focus FWD
47.0
189
43.5
Gasoline
M6
1.0
Compact Cars
c
14%
11%
7%
3%
-2%
2016
Nissan
Sentra FE+
46.1
193
44.4
Gasoline
AV
1.8
Midsize Cars
c
14%
11%
7%
3%
-2%
2016
Mazda
CX-9 2WD
32.5
273
52.2
Gasoline
S6
2.5
Sport Utility Vehicles
T
10%
9%
7%
5%
-2%
2016
Ford
F150 2WD FFV
26.6
335
66.2
Gasoline
A6
3.5
Standard Pick-up Trucks
T
7%
7%
7%
5%
-2%
2016
Nissan
NV200 Cargo Van
34.9
254
47.9
Gasoline
AV
2.0
Vans
T
11%
9%
7%
5%
-2%
2016
Kia
Optima
42.8
208
48.2
Gasoline
AM7
1.6
Large Cars
C
14%
11%
6%
2%
-2%
2016
Infiniti
070 Hybrid
42.1
211
49.1
HEV
S7
3.5
Midsize Cars
c
14%
10%
6%
2%
-2%
2016
Mazda
Mazda 3 4-Door
45.1
197
45.3
Gasoline
M6
2.0
Compact Cars
c
14%
10%
6%
2%
-3%
2016
Mazda
Mazda 3 5-Door
45.1
197
45.3
Gasoline
S6
2.0
Midsize Cars
c
14%
10%
6%
2%
-3%
2016
Volvo
XC90AWD
31.7
280
53.5
Gasoline
S8
2.0
Sport Utility Vehicles
T
10%
8%
7%
4%
-3%
2016
Hyundai
Sonata
42.4
210
48.3
Gasoline
AM7
1.6
Large Cars
c
14%
10%
6%
2%
-3%
2016
Mercedes-Benz
GLC 300
32.3
275
52.2
Gasoline
A9
2.0
Sport Utility Vehicles
T
10%
8%
6%
4%
-3%
2016
GMC
C15 Sierra 2WD Crew Cab, Short Box
25.8
345
68.0
Gasoline
A6
4.3
Standard Pick-up Trucks
T
4%
4%
4%
4%
-3%
2016
Chevrolet
C15 Silverado 2WD Crew Cab, Short Box
25.8
345
68.0
Gasoline
A6
4.3
Standard Pick-up Trucks
T
4%
4%
4%
4%
-3%
2016
Toyota
Corolla LE ECO
45.9
194
44.2
Gasoline
AV
1.8
Midsize Cars
C
13%
10%
6%
1%
-3%
2016
Audi
Q5 Hybrid
34.0
261
48.8
HEV
S8
2.0
Sport Utility Vehicles
T
10%
8%
6%
4%
-3%
2016
Buick
Encore AWD
38.2
232
42.3
Gasoline
S6
1.4
Sport Utility Vehicles
T
10%
8%
6%
4%
-4%
2016
BMW
328d
49.6
205
46.9
Diesel
S8
2.0
Compact Cars
C
13%
9%
5%
1%
-4%
2016
Hyundai
Tucson Eco AWD
36.3
245
45.0
Gasoline
AM7
1.6
Sport Utility Vehicles
T
10%
7%
6%
3%
-4%
2016
Lexus
NX 300h
44.8
198
45.1
Gasoline
AV-S6
2.5
Sport Utility Vehicles
C
13%
9%
5%
1%
-4%
2016
Ford
F150 4WD
25.5
348
68.1
Gasoline
S6
2.7
Standard Pick-up Trucks
T
3%
3%
3%
3%
-4%
A-21
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Model
Year
Manufacturer
Vehicle
Fuel
Economy
(mpg)
Tailpipe
C02
(g/mile)
Footprint
(ft2)
Powertrain
Type
Trans-
mission
Engine
Disp.
(L)
Vehicle Class
Car/
Truck
Compliance
2017
2018
2019
2020
2021
2022
2023
2024
2025
2016
Ford
Escape AWD
32.9
270
50.5
Gasoline
S6
1.6
Sport Utility Vehicles
T
9%
7%
6%
3%
-4%
2016
Honda
CR-V4WD
36.5
243
44.5
Gasoline
AV
2.4
Sport Utility Vehicles
T
9%
7%
6%
3%
-4%
2016
Ford
F150 2WD
25.5
349
68.1
Gasoline
S6
3.5
Standard Pick-up Trucks
T
3%
3%
3%
3%
-4%
2016
Lexus
GS 450h
41.6
213
48.5
Gasoline
AV-S8
3.5
Midsize Cars
C
12%
9%
4%
0%
-5%
2016
Mazda
Mazda 3 5-Door
44.4
200
45.3
Gasoline
M6
2.0
Midsize Cars
C
12%
9%
4%
0%
-5%
2016
Chevrolet
Cruze Limited ECO
44.8
198
44.8
Gasoline
M6
1.4
Midsize Cars
c
12%
9%
4%
0%
-5%
2016
Mazda
Mazda 3 4-Door
44.4
200
45.3
Gasoline
S6
2.5
Compact Cars
c
12%
9%
4%
0%
-5%
2016
Nissan
NV200 NYC Taxi
34.1
261
47.9
Gasoline
AV
2.0
Vans
T
8%
6%
5%
2%
2016
Ram
15004X2- Regular Cab, 8'0" Box
25.6
347
66.6
Gasoline
A8
3.6
Standard Pick-up Trucks
T
4%
4%
4%
1%
2016
Mazda
CX-54WD
34.9
255
46.1
Gasoline
S6
2.5
Sport Utility Vehicles
T
8%
6%
4%
1%
2016
Subaru
Forester
36.1
246
44.0
Gasoline
AV
2.5
Sport Utility Vehicles
T
7%
5%
3%
1%
2016
Ford
F1504WD
25.5
348
66.2
Gasoline
S6
2.7
Standard Pick-up Trucks
T
3%
3%
3%
1%
2016
Mercedes-Benz
GLE 300 d 4MATIC
35.9
284
52.2
Diesel
A7
2.1
Sport Utility Vehicles
T
7%
5%
3%
1%
2016
Mercedes-Benz
GLC 300 4MATIC
31.3
284
52.2
Gasoline
A9
2.0
Sport Utility Vehicles
T
7%
5%
3%
1%
2016
Nissan
Quest
29.5
301
55.9
Gasoline
AV
3.5
Minivans
T
6%
5%
3%
1%
2016
Ford
F1504WD FFV
24.9
357
68.1
Gasoline
A6
3.5
Standard Pick-up Trucks
T
1%
1%
1%
0%
2016
Ford
F1504WD FFV
24.9
357
68.1
Gasoline
A6
3.5
Standard Pick-up Trucks
T
1%
1%
1%
0%
2016
Lincoln
MKT Livery FWD
30.6
290
53.5
Gasoline
S6
2.0
Sport Utility Vehicles
T
6%
5%
3%
0%
2016
Ford
F1502WD
25.5
349
66.2
Gasoline
S6
3.5
Standard Pick-up Trucks
T
3%
3%
3%
0%
2016
Ram
15004X4- Regular Cab, 8'0" Box
29.0
351
66.6
Diesel
A8
3.0
Standard Pick-up Trucks
T
2%
2%
2%
0%
2016
Jeep
Renegade 4x4
36.7
242
42.7
Gasoline
M6
1.4
Sport Utility Vehicles
T
7%
4%
3%
0%
2016
Ford
Fiesta SFE FWD
48.4
184
39.0
Gasoline
M5
1.0
Subcompact Cars
C
12%
8%
4%
0%
-5%
2016
Mazda
Mazda 6
40.8
218
49.4
Gasoline
S6
2.5
Midsize Cars
C
12%
8%
4%
0%
2016
Acura
RLX
40.3
221
50.1
HEV
AM7
3.5
Midsize Cars
c
12%
8%
4%
0%
2016
Chevrolet
Cruze
44.4
200
44.8
Gasoline
M6
1.4
Compact Cars
c
11%
8%
3%
-1%
2016
Mercedes-Benz
Smart fortwo (COUPE)
47.9
185
26.8
Gasoline
M5
0.9
Two Seaters
c
11%
7%
3%
-1%
2016
Nissan
Versa
47.3
188
41.5
Gasoline
AV
1.6
Compact Cars
c
11%
7%
3%
-1%
2016
Chevrolet
Malibu
41.1
216
48.4
Gasoline
A6
1.5
Midsize Cars
c
11%
7%
3%
-1%
2016
Mazda
Mazda 2
47.7
186
39.4
Gasoline
M6
1.5
Compact Cars
c
11%
7%
3%
-2%
2016
Dodge
Dart Aero
43.4
205
45.6
Gasoline
M6
1.4
Midsize Cars
c
11%
7%
3%
-2%
2016
Volkswagen
Jetta
44.8
198
44.0
Gasoline
M5
1.4
Compact Cars
c
11%
7%
3%
-2%
2016
Nissan
Altima SR
41.8
213
47.4
Gasoline
AV-S7
2.5
Midsize Cars
c
11%
7%
2%
-2%
2016
Mazda
Mazda 3 5-Door
43.5
204
45.3
Gasoline
S6
2.5
Midsize Cars
c
11%
7%
2%
-2%
2016
Dodge
Dart Aero
43.2
206
45.6
Gasoline
AM6
1.4
Midsize Cars
c
10%
6%
2%
-2%
2016
Chevrolet
Spark
47.4
188
35.8
Gasoline
AV
1.4
Subcom pact Cars
c
10%
6%
2%
-3%
2016
Honda
Fit
47.3
188
40.2
Gasoline
AV-S7
1.5
Small Station Wagons
c
10%
6%
2%
-3%
2016
Scion
iA
47.3
188
40.8
Gasoline
M6
1.5
Subcom pact Cars
c
10%
6%
2%
-3%
2016
Toyota
Tacoma 2WD, Double Cab Long bed
27.0
329
61.6
Gasoline
S6
3.5
Small Pick-up Trucks
T
4%
4%
2%
0%
2016
Volvo
S80 FWD
40.7
219
48.4
Gasoline
S8
2.0
Midsize Cars
c
10%
6%
2%
-3%
2016
Nissan
Sentra
44.0
202
44.4
Gasoline
AV
1.8
Midsize Cars
c
10%
6%
2%
-3%
2016
Honda
Accord
41.1
216
47.6
Gasoline
AV
2.4
Midsize Cars
c
10%
6%
1%
-3%
2016
Chevrolet
Colorado 2WD Crew Cab, Long Bed
27.1
328
60.9
Gasoline
A6
3.6
Small Pick-up Trucks
T
3%
3%
2%
-1%
2016
GMC
Canyon 2WD Crew Cab, Long Box
27.1
328
60.9
Gasoline
A6
3.6
Small Pick-up Trucks
T
3%
3%
2%
-1%
2016
Chevrolet
Colorado 2WD
29.3
303
55.6
Gasoline
A6
2.5
Small Pick-up Trucks
T
5%
4%
2%
-1%
A-22
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Model
Year
Manufacturer
Vehicle
Fuel
Economy
(mpg)
Tailpipe
C02
(g/mile)
Footprint
(ft2)
Powertrain
Type
Trans-
mission
Engine
Disp.
(L)
Vehicle Class
Car/
Truck
Compliance
2017
2018
2019
2020
2021
2022
2023
2024
2025
2016
GMC
Canyon 2WD
29.3
303
55.6
Gasoline
A6
2.5
Small Pick-up Trucks
T
5%
4%
2%
-1%
2016
Mazda
Mazda 3 4-Door
42.9
207
45.3
Gasoline
S6
2.5
Compact Cars
C
9%
5%
1%
-3%
2016
Honda
Odyssey 2WD
29.0
307
55.9
Gasoline
A6
3.5
Minivans
T
4%
3%
1%
-2%
2016
Ford
F150 2WD
28.2
315
57.5
Gasoline
S6
2.7
Standard Pick-up Trucks
T
3%
2%
1%
-2%
2016
Mazda
CX-94WD
30.6
291
52.2
Gasoline
S6
2.5
Sport Utility Vehicles
T
4%
2%
1%
-2%
2016
Mercedes-Benz
Metris (Cargo Van)
30.2
294
52.9
Gasoline
A7
2.0
Vans
T
4%
2%
1%
-2%
2016
Volvo
XC90AWD
29.9
297
53.5
Gasoline
S8
2.0
Sport Utility Vehicles
T
4%
2%
1%
-2%
2016
Ford
F150 4WD FFV
24.9
357
66.2
Gasoline
A6
3.5
Standard Pick-up Trucks
T
1%
1%
0%
-2%
2016
Nissan
Pathfinder 2WD
30.5
292
52.1
Gasoline
AV
3.5
Sport Utility Vehicles
T
4%
2%
0%
-2%
2016
Volkswagen
Jetta
43.8
203
44.0
Gasoline
S6
1.4
Compact Cars
C
9%
5%
0%
-4%
2016
Chevrolet
Colorado 2WD
33.0
308
55.6
Diesel
A6
2.8
Small Pick-up Trucks
T
3%
2%
0%
-3%
2016
GMC
Canyon 2WD
33.0
308
55.6
Diesel
A6
2.8
Sport Utility Vehicles
T
3%
2%
0%
-3%
2016
Lincoln
MKC AWD
30.9
288
51.2
Gasoline
S6
2.0
Sport Utility Vehicles
T
4%
2%
0%
-3%
2016
GMC
C15 Sierra 2WD Regular Cab, Long Box
25.8
345
63.0
Gasoline
A6
4.3
Standard Pick-up Trucks
T
1%
1%
-1%
-3%
2016
Chevrolet
C15 Silverado 2WD Regular Cab, Long Box
25.8
345
63.0
Gasoline
A6
4.3
Standard Pick-up Trucks
T
1%
1%
-1%
-3%
2016
Toyota
Tacoma 4WD, Double Cab Long bed
26.1
340
61.6
Gasoline
S6
3.5
Small Pick-up Trucks
T
0%
0%
-1%
-4%
2016
Buick
Encore AWD
35.7
249
42.3
Gasoline
S6
1.4
Sport Utility Vehicles
T
3%
1%
-1%
-4%
2016
Chevrolet
Trax AWD
35.7
249
42.3
Gasoline
S6
1.4
Sport Utility Vehicles
T
3%
1%
-1%
-4%
2016
Ford
Escape AWD
30.9
288
50.5
Gasoline
S6
2.0
Sport Utility Vehicles
T
3%
1%
-1%
-4%
2016
Chevrolet
Colorado 2WD
28.4
312
55.6
Gasoline
M6
2.5
Small Pick-up Trucks
T
2%
0%
-1%
-4%
2016
GMC
Canyon 2WD
28.4
312
55.6
Gasoline
M6
2.5
Small Pick-up Trucks
T
2%
0%
-1%
-4%
2016
Nissan
Murano AWD
31.6
281
48.9
Gasoline
AV-S7
3.5
Midsize Station Wagons
T
3%
0%
-1%
-4%
2016
Hyundai
Tucson AWD
33.9
263
45.0
Gasoline
AM7
1.6
Sport Utility Vehicles
T
3%
0%
-1%
-4%
2016
Hyundai
El antra
42.6
209
45.2
Gasoline
S6
1.8
Midsize Cars
C
8%
4%
0%
-4%
2016
Toyota
Corolla
43.5
204
44.2
Gasoline
AV
1.8
Midsize Cars
C
8%
4%
0%
-4%
2016
Dodge
Dart
42.3
210
45.6
Gasoline
M6
1.4
Midsize Cars
c
8%
4%
0%
-5%
2016
Volvo
V60 FWD
40.7
219
47.4
Gasoline
S8
2.0
Small Station Wagons
c
8%
4%
0%
-5%
2016
Fiat
500
46.4
192
34.7
Gasoline
M5
1.4
Minicompact Cars
c
8%
4%
0%
-5%
2016
Mazda
Mazda 6
39.0
228
49.4
Gasoline
M6
2.5
Midsize Cars
c
8%
4%
-1%
2016
Mercedes-Benz
E 250 BLUETEC
44.3
230
49.8
Diesel
A7
2.1
Midsize Cars
c
8%
4%
-1%
2016
Volvo
S60 FWD
40.7
219
47.1
Gasoline
S8
2.0
Compact Cars
c
7%
3%
-1%
2016
Volvo
S60 Inscription FWD
40.7
219
47.1
Gasoline
S8
2.0
Compact Cars
c
7%
3%
-1%
2016
Honda
CR-Z
46.1
193
36.3
HEV
M6
1.5
Two Seaters
c
8%
3%
-1%
2016
Infiniti
QX60 Hybrid FWD
37.0
240
52.1
HEV
AV-S7
2.5
Sport Utility Vehicles
c
7%
3%
-1%
2016
Infiniti
Q50 Hybrid
40.9
217
46.6
HEV
S7
3.5
Compact Cars
c
7%
3%
-1%
2016
Toyota
Corolla
42.9
207
44.2
Gasoline
AV-S7
1.8
Midsize Cars
c
7%
3%
-1%
2016
Chevrolet
Spark
46.0
193
35.8
Gasoline
M5
1.4
Subcompact Cars
c
7%
3%
-1%
2016
Ford
Focus FWD
43.5
204
43.5
Gasoline
S6
1.0
Compact Cars
c
7%
3%
-2%
2016
Hyundai
Sonata
39.5
225
48.3
Gasoline
S6
2.4
Large Cars
c
7%
3%
-2%
2016
Honda
Civic4Dr
41.9
212
45.2
Gasoline
M6
2.0
Midsize Cars
c
7%
3%
-2%
2016
BMW
328dx Drive
46.2
220
46.9
Diesel
S8
2.0
Compact Cars
c
6%
2%
-2%
2016
BMW
328d xDrive Sports Wagon
46.2
220
46.9
Diesel
S8
2.0
Small Station Wagons
c
6%
2%
-2%
2016
Nissan
Murano Hybrid FWD
38.8
229
48.9
HEV
AV-S7
2.5
Midsize Station Wagons
c
6%
2%
-2%
2016
Kia
Optima FE
39.3
226
48.2
Gasoline
S6
2.4
Large Cars
c
6%
2%
-3%
A-23
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Model
Year
Manufacturer
Vehicle
Fuel
Economy
(mpg)
Tailpipe
C02
(g/mile)
Footprint
(ft2)
Powertrain
Type
Trans-
mission
Engine
Disp.
(L)
Vehicle Class
Car/
Truck
Compliance
2017
2018
2019
2020
2021
2022
2023
2024
2025
2016
Hyundai
Veloster
42.1
211
44.6
Gasoline
AM6
1.6
Compact Cars
C
6%
2%
-3%
2016
Audi
A6
37.3
238
50.9
Gasoline
AM-S7
2.0
Midsize Cars
C
6%
2%
-3%
2016
Toyota
Corolla
42.4
210
44.2
Gasoline
M6
1.8
Midsize Cars
C
6%
2%
-3%
2016
Ford
Fusion FWD
38.9
228
48.4
Gasoline
S6
1.5
Midsize Cars
C
6%
1%
-3%
2016
Dodge
Dart
41.1
216
45.6
Gasoline
AM6
1.4
Midsize Cars
C
6%
1%
-3%
2016
Mazda
Mazda 3 5-Door
41.3
215
45.3
Gasoline
S6
2.5
Midsize Cars
C
5%
1%
-3%
2016
Honda
Accord
39.4
225
47.6
Gasoline
AV-S7
2.4
Midsize Cars
C
5%
1%
-3%
2016
Hyundai
Elantra Limited
41.2
216
45.3
Gasoline
S6
1.8
Midsize Cars
C
5%
1%
-4%
2016
Toyota
Yaris
44.9
198
39.9
Gasoline
M5
1.5
Compact Cars
C
5%
1%
-4%
2016
Hyundai
Elantra
40.9
217
45.2
Gasoline
M6
1.8
Midsize Cars
C
4%
0%
-4%
2016
Honda
Civic 2Dr
40.9
217
45.2
Gasoline
M6
2.0
Compact Cars
C
4%
0%
-5%
2016
Mitsubishi
Outlander Sport 4WD
34.1
260
44.2
Gasoline
AV-S6
2.0
Sport Utility Vehicles
T
2%
0%
-2%
-5%
2016
BMW
X3sDrive 28i
31.3
284
49.1
Gasoline
S8
2.0
Sport Utility Vehicles
T
2%
-1%
-2%
2016
BMW
X3xDrive28i
31.3
284
49.1
Gasoline
S8
2.0
Sport Utility Vehicles
T
2%
-1%
-2%
2016
Mercedes-Benz
Metris (Passenger Van)
29.4
303
52.9
Gasoline
A7
2.0
Vans
T
1%
-1%
-2%
2016
Mercedes-Benz
GLE 350 d 4MATIC
34.0
300
52.2
Diesel
A9
3.0
Sport Utility Vehicles
T
1%
-1%
-3%
2016
Infiniti
QX60AWD
29.7
299
52.1
Gasoline
AV-S7
3.5
Sport Utility Vehicles
T
1%
-1%
-3%
2016
Nissan
Pathfinder 4WD
29.6
300
52.1
Gasoline
AV
3.5
Sport Utility Vehicles
T
1%
-1%
-3%
2016
BMW
XI x Drive 28i
33.7
264
44.6
Gasoline
S8
2.0
Large Cars
T
1%
-1%
-3%
2016
Mercedes-Benz
GL 350 BLUETEC 4MATIC
28.3
314
54.8
Gasoline
A7
3.0
Sport Utility Vehicles
T
0%
-1%
-3%
2016
BMW
X5xDrive 35d
33.8
301
52.0
Diesel
S8
3.0
Sport Utility Vehicles
T
0%
-2%
-4%
2016
Jeep
Cherokee FWD
32.7
272
45.7
Gasoline
A9
2.4
Sport Utility Vehicles
T
0%
-2%
-4%
2016
Jeep
Cherokee FWD
32.7
272
45.7
Gasoline
A9
2.4
Sport Utility Vehicles
T
0%
-2%
-4%
2016
Subaru
Crosstrek
34.2
260
43.2
Gasoline
M5
2.0
Sport Utility Vehicles
T
0%
-2%
-4%
2016
Subaru
Legacy
39.7
224
46.5
Gasoline
AV-S6
2.5
Midsize Cars
C
4%
0%
-5%
2016
Hyundai
Sonata Sport/Limited
38.2
233
48.5
Gasoline
S6
2.4
Large Cars
C
4%
0%
2016
Volkswagen
Passat
39.1
227
47.2
Gasoline
S6
1.8
Midsize Cars
c
4%
0%
2016
Ford
Focus FWD
42.1
211
43.5
Gasoline
AM6
2.0
Compact Cars
c
4%
0%
2016
Ford
Focus FWD FFV
42.1
211
43.5
Gasoline
AM6
2.0
Compact Cars
c
4%
0%
2016
Chevrolet
Cruze Limited ECO
40.8
218
44.8
Gasoline
A6
1.4
Midsize Cars
c
3%
-1%
2016
Toyota
Corolla
41.3
215
44.2
Gasoline
A4
1.8
Midsize Cars
c
3%
-1%
2016
Acura
TLX 2WD
38.3
232
47.8
Gasoline
AM-S8
2.4
Compact Cars
c
3%
-2%
2016
Kia
Forte
40.8
218
44.5
Gasoline
S6
1.8
Midsize Cars
c
3%
-2%
2016
Chevrolet
Sonic
44.0
202
41.0
Gasoline
M6
1.4
Compact Cars
c
3%
-2%
2016
Chevrolet
Sonic 5
44.0
202
41.0
Gasoline
M6
1.4
Small Station Wagons
c
3%
-2%
2016
Buick
Lacrosse
38.1
233
48.0
HEV
S6
2.4
Midsize Cars
c
3%
-2%
2016
Mini
Mini Cooper Hardtop 4 Door
43.9
202
38.8
Gasoline
M6
1.5
Subcompact Cars
c
3%
-2%
2016
Mercedes-Benz
CLA 250
40.4
220
45.0
Gasoline
AM7
2.0
Compact Cars
c
3%
-2%
2016
Scion
iM
42.7
208
42.3
Gasoline
AV-S7
1.8
Midsize Cars
c
3%
-2%
2016
Subaru
Impreza
41.9
212
43.0
Gasoline
AV-S6
2.0
Compact Cars
c
2%
-2%
2016
Subaru
Impreza Wagon
41.9
212
43.0
Gasoline
AV-S6
2.0
Small Station Wagons
c
2%
-2%
2016
Toyota
Yaris
43.8
203
39.9
Gasoline
A4
1.5
Compact Cars
c
2%
-2%
2016
Chevrolet
Cruze Limited
40.4
220
44.8
Gasoline
M6
1.4
Midsize Cars
c
2%
-2%
2016
Honda
HR-V2WD
41.7
213
43.2
Gasoline
AV
1.8
Sport Utility Vehicles
c
2%
-2%
A-24
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Model
Year
Manufacturer
Vehicle
Fuel
Economy
(mpg)
Tailpipe
C02
(g/mile)
Footprint
(ft2)
Powertrain
Type
Trans-
mission
Engine
Disp.
(L)
Vehicle Class
Car/
Truck
Compliance
2017
2018
2019
2020
2021
2022
2023
2024
2025
2016
Honda
HR-V2WD
41.7
213
43.2
Gasoline
AV-S7
1.8
Sport Utility Vehicles
C
2%
-2%
2016
Mercedes-Benz
E 250 BLUETEC 4MATIC
41.9
243
49.8
Diesel
A7
2.1
Midsize Cars
C
2%
-2%
2016
BMW
535d
40.5
251
51.5
Diesel
S8
3.0
Midsize Cars
C
2%
-3%
2016
Mazda
CX-5 2WD
39.2
227
46.1
Gasoline
M6
2.0
Sport Utility Vehicles
C
2%
-3%
2016
BMW
528i
35.4
251
51.5
Gasoline
S8
2.0
Midsize Cars
C
2%
-3%
2016
Nissan
Sentra
40.5
219
44.4
Gasoline
M6
1.8
Midsize Cars
C
2%
-3%
2016
Chevrolet
Cruze Limited
40.1
222
44.8
Gasoline
S6
1.4
Midsize Cars
C
1%
-3%
2016
Toyota
Camry
38.2
233
47.2
Gasoline
S6
2.5
Midsize Cars
C
1%
-3%
2016
Ford
Fusion FWD
37.3
238
48.4
Gasoline
S6
1.5
Midsize Cars
C
1%
-3%
2016
Kia
Optima
37.5
237
48.2
Gasoline
S6
2.4
Large Cars
C
1%
-3%
2016
Mazda
CX-5 2WD
38.9
228
46.1
Gasoline
S6
2.5
Sport Utility Vehicles
C
1%
-3%
2016
Hyundai
Veloster
40.1
222
44.6
Gasoline
M6
1.6
Compact Cars
C
1%
-3%
2016
Honda
FIT
43.1
206
40.2
Gasoline
M6
1.5
Small Station Wagons
C
1%
-4%
2016
Mini
Mini Cooper Hardtop 2 Door
43.0
206
38.8
Gasoline
M6
1.5
Subcompact Cars
C
1%
-4%
2016
Infiniti
050 Hybrid AWD
38.2
232
46.6
HEV
S7
3.5
Compact Cars
C
0%
-4%
2016
Buick
Regal
38.1
233
46.8
HEV
S6
2.4
Midsize Cars
C
0%
-4%
2016
GMC
C15 Sierra 2WD Crew Cab, Standard Box
24.0
370
72.5
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-3%
-3%
-3%
-3%
-5%
2016
Ford
F1504WD
23.7
375
76.8
Gasoline
S6
3.5
Standard Pick-up Trucks
T
-5%
-5%
-5%
-5%
-5%
2016
Chevrolet
C15 SilveradoO 2WD Crew Cab, Standard Box
24.0
370
72.5
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-3%
-3%
-3%
-3%
-5%
2016
GMC
C15 Sierra 2WD FFV Crew Cab, Standard Box
24.0
371
72.5
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-3%
-3%
-3%
-3%
-5%
2016
Chevrolet
C15 Silverado 2WD FFV Crew Cab, Standard Box
24.0
371
72.5
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-3%
-3%
-3%
-3%
2016
Chevrolet
C15 Silverado 2WD Cab Chassis
23.9
372
72.5
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-4%
-4%
-4%
-4%
2016
GMC
C15 Sierra 2WD Cab Chassis
23.9
372
72.5
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-4%
-4%
-4%
-4%
2016
Chevrolet
K15 Silverado 4WD Crew Cab, Short Box
24.3
366
68.0
Gasoline
A6
4.3
Standard Pick-up Trucks
T
-2%
-2%
-2%
-3%
2016
GMC
K15 Sierra 4WD Crew Cab, Short Box
24.3
366
68.0
Gasoline
A6
4.3
Standard Pick-up Trucks
T
-2%
-2%
-2%
-3%
2016
GMC
C15 Sierra 2WD Crew Cab, Short Box
24.0
370
68.0
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-3%
-3%
-3%
-4%
2016
Chevrolet
C15 Silverado 2WD Crew Cab, Short Box
24.0
370
68.0
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-3%
-3%
-3%
-4%
2016
GMC
C15 Sierra 2WD FFV Crew Cab, Short Box
24.0
371
68.0
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-3%
-3%
-3%
-4%
2016
Chevrolet
C15 Silverado 2WD FFV Crew Cab, Short Box
24.0
371
68.0
Gasoline
A6
5.3
Standard Pick-up Trucks
T
-3%
-3%
-3%
-4%
2016
Chevrolet
Colorado 4WD Crew Cab, Long Bed
29.9
340
60.9
Diesel
A6
2.8
Small Pick-up Trucks
T
-1%
-1%
-2%
-5%
2016
GMC
Canyon 4WDCrew Cab, Long Box
29.9
340
60.9
Diesel
A6
2.8
Sport Utility Vehicles
T
-1%
-1%
-2%
-5%
2016
Ford
F1504WD
23.7
375
68.1
Gasoline
S6
3.5
Standard Pick-up Trucks
T
-5%
-5%
-5%
2016
Ram
1500 HFE 4X2
31.4
324
56.8
Diesel
A8
3.0
Standard Pick-up Trucks
T
-1%
-1%
-3%
2016
Chevrolet
Colorado 4WD Crew Cab, Long Bed
25.7
346
60.9
Gasoline
A6
3.6
Small Pick-up Trucks
T
-2%
-2%
-4%
2016
GMC
Canyon 4WD Crew Cab, Long Box
25.7
346
60.9
Gasoline
A6
3.6
Small Pick-up Trucks
T
-2%
-2%
-4%
2016
Ram
1500 HFE 4X2
27.2
327
56.8
Gasoline
A8
3.6
Standard Pick-up Trucks
T
-2%
-2%
-4%
2016
Ford
Edge AWD
30.0
297
50.5
Gasoline
S6
2.0
Sport Utility Vehicles
T
0%
-3%
-5%
2016
Chevrolet
Equinox AWD
30.8
289
48.8
Gasoline
A6
2.4
Sport Utility Vehicles
T
0%
-3%
-5%
2016
Chevrolet
Equinox AWD
30.8
289
48.8
Gasoline
A6
2.4
Sport Utility Vehicles
T
0%
-3%
-5%
2016
GMC
Terrain AWD
30.8
289
48.8
Gasoline
A6
2.4
Sport Utility Vehicles
T
0%
-3%
-5%
2016
Toyota
Tacoma 2WD
27.5
323
55.8
Gasoline
S6
2.7
Small Pick-up Trucks
T
-2%
-3%
-5%
2016
Ford
F1504WD
23.7
375
66.2
Gasoline
S6
3.5
Standard Pick-up Trucks
T
-5%
-5%
-5%
2016
Toyota
RAV4AWD
32.9
270
44.9
Gasoline
S6
2.5
Sport Utility Vehicles
T
-1%
-3%
2016
Land Rover
Range Rover Evoque
31.7
280
46.6
Gasoline
S9
2.0
Sport Utility Vehicles
T
-1%
-4%
A-25
-------
Appendix C - CO2 Targets with Current Powertrain Designs
Model
Year
Manufacturer
Vehicle
Fuel
Economy
(mpg)
Tailpipe
C02
(g/mile)
Footprint
(ft2)
Powertrain
Type
Trans-
mission
Engine
Disp.
(L)
Vehicle Class
Car/
Truck
Compliance
2017
2018
2019
2020
2021
2022
2023
2024
2025
2016
Nissan
Pathfinder 4WD Platinum
28.8
309
52.1
Gasoline
AV
3.5
Sport Utility Vehicles
T
-2%
-4%
2016
Honda
Pilot 4WD
29.2
305
51.3
Gasoline
S9
3.5
Sport Utility Vehicles
T
-2%
-4%
2016
Chevrolet
Colorado 4WD
27.2
326
55.6
Gasoline
A6
2.5
Small Pick-up Trucks
T
-3%
-4%
2016
GMC
Canyon 4WD
27.2
326
55.6
Gasoline
A6
2.5
Small Pick-up Trucks
T
-3%
-4%
2016
Ford
Transit Connect Van 2WD
32.6
273
44.8
Gasoline
S6
1.6
Vans
T
-2%
-4%
2016
Land Rover
Range Rover Sport TDV6
32.4
314
53.1
Diesel
S8
3.0
Sport Utility Vehicles
T
-3%
-4%
2016
Land Rover
Range RoverTDV6
32.4
314
53.1
Diesel
S8
3.0
Sport Utility Vehicles
T
-3%
-4%
2016
Subaru
Forester
33.0
269
44.0
Gasoline
AV-S8
2.0
Sport Utility Vehicles
T
-2%
-4%
2016
Lexus
NX 200t A WD
32.3
275
45.1
Gasoline
S6
2.0
Sport Utility Vehicles
T
-2%
-5%
2016
Dodge
Durango RWD
28.1
316
53.2
Gasoline
A8
3.6
Sport Utility Vehicles
T
-3%
-5%
2016
Chevrolet
Colorado 2WD
27.1
328
55.6
Gasoline
A6
3.6
Small Pick-up Trucks
T
-3%
-5%
2016
GMC
Canyon 2WD
27.1
328
55.6
Gasoline
A6
3.6
Small Pick-up Trucks
T
-3%
-5%
2016
Toyota
Tacoma 2WD, Access Cab or Double/short
27.0
329
55.8
Gasoline
S6
3.5
Small Pick-up Trucks
T
-4%
-5%
2016
Toyota
RAV4 Limited AWD/SE AWD
32.3
275
44.9
Gasoline
S6
2.5
Sport Utility Vehicles
T
-2%
-5%
2016
BMW
528i xDrive
34.8
255
51.5
Gasoline
S8
2.0
Midsize Cars
C
0%
-5%
2016
Subaru
Impreza Sport
41.0
217
43.0
Gasoline
AV-S6
2.0
Small Station Wagons
C
0%
-5%
2016
Kia
Rio ECO
41.7
213
42.1
Gasoline
S6
1.6
Compact Cars
c
0%
-5%
2016
Mazda
Mazda 3 4-Door
39.1
227
45.3
Gasoline
M6
2.5
Compact Cars
c
0%
-5%
2016
Ford
Focus FWD FFV
40.5
220
43.5
Gasoline
AM-S6
2.0
Compact Cars
c
0%
-5%
2016
Nissan
Rogue FWD
38.1
233
46.4
Gasoline
AV
2.5
Sport Utility Vehicles
c
0%
-5%
2016
Ford
Focus FWD
40.4
220
43.5
Gasoline
AM-S6
2.0
Compact Cars
c
0%
-5%
2016
Toyota
Sienna
26.8
331
56.1
Gasoline
S6
3.5
Minivans
T
-4%
2016
Ford
Transit Connect Wagon FWD
32.3
275
44.8
Gasoline
S6
1.6
Vans
T
-3%
2016
Audi
Q5
34.4
296
48.8
Diesel
S8
3.0
Sport Utility Vehicles
T
-3%
2016
BMW
X4 x Drive 28i
29.8
298
49.1
Gasoline
S8
2.0
Sport Utility Vehicles
T
-3%
2016
Toyota
Tacoma 4WD
26.8
332
55.8
Gasoline
M5
2.7
Small Pick-up Trucks
T
-5%
2016
Subaru
Forester
32.6
273
44.0
Gasoline
M6
2.5
Sport Utility Vehicles
T
-3%
2016
Acura
MDX 4 WD
28.9
307
50.8
Gasoline
S9
3.5
Sport Utility Vehicles
T
-3%
2016
Nissan
Frontier 2WD
27.4
324
54.0
Gasoline
M5
2.5
Small Pick-up Trucks
T
-4%
2016
Mitsubishi
Outlander Sport 4WD
32.2
276
44.2
Gasoline
AV-S6
2.4
Sport Utility Vehicles
T
-4%
2016
Audi
Q5
29.6
301
48.8
Gasoline
S8
2.0
Sport Utility Vehicles
T
-5%
2016
Jeep
Cherokee FWD
31.1
285
45.7
Gasoline
A9
3.2
Sport Utility Vehicles
T
-5%
2016
Ford
Taurus FWD
34.7
256
51.3
Gasoline
S6
2.0
Large Cars
C
-1%
2016
Kia
Rio
41.6
214
42.1
Gasoline
M6
1.6
Compact Cars
c
-1%
2016
Lincoln
MKC FWD
34.7
256
51.2
Gasoline
S6
2.0
Sport Utility Vehicles
c
-1%
2016
Dodge
Dart
38.7
230
45.6
Gasoline
M6
2.0
Midsize Cars
c
-1%
2016
Mazda
Mazda 3 5-Door
38.9
228
45.3
Gasoline
M6
2.5
Midsize Cars
c
-1%
2016
Hyundai
Accent
41.8
212
41.7
Gasoline
M6
1.6
Compact Cars
c
-1%
2016
Ford
Escape FWD
35.2
253
50.5
Gasoline
S6
1.6
Sport Utility Vehicles
c
-1%
2016
Nissan
Versa
42.0
212
41.5
Gasoline
M5
1.6
Compact Cars
c
-1%
2016
Kia
Forte
39.4
225
44.5
Gasoline
M6
1.8
Midsize Cars
c
-1%
2016
Cadillac
CT6
32.9
270
54.1
Gasoline
S8
2.0
Large Cars
c
-1%
2016
Mazda
CX-3 2WD
42.5
209
40.7
Gasoline
S6
2.0
Compact Cars
c
-1%
2016
Kia
Rio
41.3
215
42.1
Gasoline
S6
1.6
Compact Cars
c
-1%
A-26
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTTV-G
Appendix D EPA Comparison Testing performed on a MY2014 Mazda
SKYACTIV-G Engine using Different Fuels
As part of the agency's ongoing engine technology benchmarking activities, EPA has
independently generated a set of fuel difference maps using its data previously generated with
fuels having different properties, including differences in RON. The engine benchmarked was a
MY2014 Mazda SKYACTIV-G 2.OL 4-cylinder engine with a 13:1 geometric CR.
The data for this analysis came from engine dynamometer tests previously conducted by EPA
using a Tier 2 certification gasoline and a LEV III gasoline (see Table 4.1, fuels A and B,
respectively).6 EPA also conducted chassis dynamometer tests using a Tier 2 certification
gasoline and a Tier 3 certification gasoline (see Table 4.1, fuels C and D, respectively). Two of
the tested fuels, Fuel B and Fuel D, had RON levels comparable to the RON reported by
AAM/USCAR (92 RON and 91 RON respectively) and AKI levels and ethanol content very
close to those of regular-grade "pump gasoline".
Fuels A and C both represent Tier 2 certification gasoline, which (as noted above) is the
gasoline used for Federal GHG compliance testing. Both are EO fuels (0 percent ethanol) with
similar distillation properties. Net energy content is slightly higher for Fuel C. Fuels B and D
represent LEV III and Tier 3 certification fuels, respectively, that will be used for compliance
with California LEV III and Federal Tier 3 emissions standards for criteria pollutants. Both are
E10 fuels (approximately 10 percent by volume ethanol) as per California LEV III and U.S.
Federal Tier 3 fuel specifications and have properties that are close to the average properties of
"regular grade" gasoline in California and the U.S., respectively. Fuel B has approximately 1-
point higher RON and AKI and lower DVPE than Fuel D, but net energy contents were nearly
identical for both fuels.
LEV III and Tier 3 certification gasoline are remarkably similar. The chief differences are an approximately 2 psi
lower Dry Vapor Pressure Equivalent and associated distillation properties.
A-27
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTTV-G
Table 4.1 Measured Fuel Properties for Four Gasolines Used for Engine and Vehicle Benchmarking
Test Fuels
Property
Unit
Fuel A
(Tier 2
Gasoline,
FTAG 23945)
Fuel B
(LEV III
Gasoline,
FTAG 24350)
Fuel C
(Tier 2
Gasoline,
FTAG 25278)
Fuel D
(Tier 3
Gasoline,
FTAG 25206)
Research Octane Number
(RON), ASTM D2699
-
97.1
92.4
96.5
91.0
Motor Octane Number (MON),
ASTM D2700
-
88.2
83.8
88.6
83.5
Antiknock Index (AKI),
(RON+MON)/2
-
92.6
88.1
92.6
87.2
Net Heat of Combustion, ASTM
D4809
MJ/kg
NA
NA
43.18
41.71
Net Heat of Combustion, ASTM
D240
MJ/kg
42.89
41.76
NA
NA
Dry Vapor Pressure Equivalent
(DVPE), ASTM D5191
psi
9.17
7.01
8.95
8.75
Distillation, ASTM D86
Initial boiling point
88.3
109.9
89.4
100.0
10% evaporated
123.4
138.1
125.6
129.0
50% evaporated
223.0
213.2
222.6
209.9
90% evaporated
322.8
317.6
317.3
321.7
Evaporated final boiling point
389.4
352.4
405.9
387.1
Aromatics, ASTM D5769
Total Aromatic HC
volume
%
33.51
23.03
32.3
23.8
C6 Aromatics (benzene)
volume
%
0.33
0.67
0.05
0.56
C7 Aromatics (toluene)
volume
%
18.56
5.79
20.0
6.2
Olefins, ASTM D6550
mass %
2.0
4.7
NA
6.4
Olefins, ASTM D6729
volume
%
NA
NA
0.10
6.4
Ethanol, ASTM D5599
volume
%
0.0
9.64
0.0
9.86
Oxygen, ASTM D5599
mass %
0.0
3.54
0.0
3.64
Sulfur, ASTM D2622
mg/kg
38.5
NA
39.6
8.3
Sulfur, ASTM D5453
mg/kg
NA
9.55
NA
NA
Because units for the AAM/USCAR "difference map" comparison were not provided by
AAM, EPA engineering staff prepared difference map comparisons on both a percentage and an
absolute basis and for both fuel volumetric- and mass-flows (see Figure 4.1 through Figure 4.4)
and without correction for differences in the net energy content (also known as "lower heating
value" or LHV) for Fuel A and Fuel B in order to provide points of comparison to the AAM
data. The same comparisons are also shown with a correction applied for net energy content
A-28
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTTV-G
(Figure 4.5 through Figure 4.8), as well as on a brake thermal energy basis in Figure 4.9 and on a
C02 emissions basis in Figure 4.10.
We believe the most appropriate way to compare fuels is either on a C02 basis (i.e., the
primary tailpipe GHG for compliance with EPA standards), a brake thermal energy basis (i.e.,
independent of LHV) or on a fuel consumption basis that corrects the fuels that are compared to
results achievable assuming a common net energy content. Note that each of EPA's comparison
maps are comprised of a numerical fit of more than one-hundred engine speed-load operating
points. Fits using fewer points may reduce the fidelity of the resulting map or "difference map"
and can introduce interpolation errors.
A-29
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTIV-G
£
Figure 4.1 Map of the Percentage Difference in Volumetric Fuel Flow for A MY2014 Mazda Skyactiv-G 2.0L
4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI, EO) versus "Fuel
B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are not corrected for differences in the net energy content between the two fuels.
200
ISO
II
I GO
II
140
80
60
40
20
Figure 4.2 Map of the Absolute Difference in Volumetric Fuel Flow (In Units of Ml/S) For A MY2014 Mazda
Skyactiv-G 2.0L 4-Cylinder Engine With A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI,
EO) Versus "Fuel B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are not corrected for differences in the net energy content between the two fuels.
A-30
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTIV-G
I
Figure 4.3 Map of the Percentage Difference in Fuel Mass Flow for A MY2014 Mazda Skyactiv-G 2.0L 4-
Cylinder Engine With A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI, EO) versus "Fuel
B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are not corrected for differences in the net energy' content between the two fuels
1500
3000 3500
Speed ( RPM )
4000
1600
4000
2000
Figure 4.4 Map of the Absolute Difference in Fuel Mass Flow (In Units of G/S) For A MY2014 Mazda
Skyactiv-G 2.0L 4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI,
EO) Versus "Fuel B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are not corrected for differences in the net energy content between the two fuels.
A-31
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTIV-G
i
I
Figure 4.5 Map of the Percentage Difference in Volumetric Fuel Flow for A MY2014 Mazda Skyactiv-G 2.0L
4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI, EO) versus "Fuel
B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are corrected for differences in the net energy content between the two fuels to allow a
direct comparison of the impacts of other fuel property differences.
3000 350
Speed ( RPM )
J lO
OOO 3600
Speed < RPM )
Figure 4.6 Map of the Absolute Difference in Volumetric Fuel Flow (In Units of Ml/S) for A MY2014 Mazda
Skyactiv-G 2.0L 4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI,
EO) Versus "Fuel B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are corrected for differences in the net energy content between the two fuels to allow a
direct comparison of the impacts of other fuel property differences.
A-32
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTIV-G
4QOO
Figure 4.7 Map of the Percentage Difference in Fuel Mass Flow for A MY2014 Mazda Skyactiv-G 2.0L 4-
Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI, EO) versus "Fuel
B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are corrected for differences in the net energy content between the two fuels to allow a
direct comparison of the impacts of other fuel property differences.
400Q
Figure 4.8 Map of the Absolute Difference in Fuel Mass Flow (In Units of G/S) For A MY2014 Mazda
Skyactiv-G 2.0L 4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI,
EO) Versus "Fuel B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are corrected for differences in the net energy content between the two fuels to allow a
direct comparison of the impacts of other fuel property differences.
A-33
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTIV-G
i
f
Figure 4.9 Map of the Percentage Difference in Brake Thermal Efficiency for A MY2014 Mazda Skyactiv-G
2.0L 4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI, EO) Versus
"Fuel B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that the calculation of brake thermal efficiency normalizes any differences in the net energy content between
the two fuels.
Figure 4.10 Map of the Percentage Difference in C02 Emissions for A MY2014 Mazda Skyactiv-G
2.0L 4-Cylinder Engine with A 13:1 Geometric CR When Tested Using "Fuel A" (Tier 2,93 AKI, EO) Versus
"Fuel B" (LEV III, 88 AKI, E10).
Note: The maximum torque shown in red is for "Fuel A". The maximum torque shown in blue is for "Fuel
B". Note that these results are not corrected for differences in the net energy content between the two fuels.
Spwd ( RPM )
A-34
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTTV-G
None of the EPA comparisons maps showed either absolute or percentage differences
approaching the magnitude of the "difference map" provided by AAM. While AAM did not
provide sufficient information to determine with any certainty which parameters are the ones that
should be compared, the closest match of EPA data to the AAM "difference map" is percentage
difference in mass of fuel consumed without correcting for the lower heating value (or net
energy content) between the fuels. In the case of EPA's data, the magnitude of the percentage
differences was approximately one-half of those in the AAM "difference map", particularly in
the region of engine operation that are critical for GHG compliance over the FTP and HwFET
(i.e., 750 to 3000 rpm, less than 6 bar BMEP, e.g., approximately the "City/Highway Critical
Area" identified by AAM).
When comparing operation of the Mazda SKYACTIV-G engine (i.e. ATK2) on a brake-
thermal-efficiency basis or after correction of percentage mass differences in fuel consumption to
an equivalent energy basis, it becomes clear that there is little or no discernable difference
between fuels A and B over areas of concern for regulatory testing beyond the differences in
energy content between the two fuels.
Chassis Dyno Testing
Results from chassis dynamometer testing over the FTP using fuels C and D showed a
decrease of just over 1 percent in combined-cycle CO2 emissions and an increase of just under 1
percent in combined cycle fuel economy for the lower RON and lower net-energy-content Fuel
D (see Table 4.2). So in the case of the Mazda SKYACTIV-G engine, CO2 emissions are
comparable or slightly lower when changing from a Tier 2 to a Tier 3 certification gasoline and
MPG is slightly lower, but less than a 1 percent difference. The differences in CO2 emissions
found during chassis dynamometer testing with fuels C and D were comparable to the
differences in CO2 emissions found between fuels A and B during engine dynamometer testing,
particularly over the areas of engine operation that are important for the regulatory drive cycles.
Table 4.2 Summary of C02 Emissions and CAFE Fuel Economy for Chassis Dynamometer Testing of The
MY2014 Mazda3 Equipped with A 2.0L Atkinson Cycle (13:1 Geometric CR) Engine Using a Tier 2 And A
Tier 3 Certification Gasoline. Three Repeats of FTP75 (City Cycle) (Highway Cycle) And 95% Confidence
Intervals Were Calculated Based Upon a Two-Sided T-Test.
FTP (City)
HwFET (Highway)
Combined
Fuel Used
CO 2 (g/mi)
[± 95% conf.
int.]
CAFE-MPG
(mi/ga)
[± 95% conf.
int.]
CO2 (g/mi)
[± 95% conf.
int.]
CAFE-MPG
(mi/ga)
[± 95% conf.
int.]
CO2 (g/mi)
[± 95% conf.
int.]
CAFE-MPG
(mi/ga)
[± 95% conf.
int.]
Fuel C (Tier 2, EO, 93 AKI)
242.12
36.75
161.87
54.78
206.01
44.87
[1.36]
[0.21]
[0.57]
[0.20]
[0.60]
[0.08]
Fuel D (Tier 3, E10, 87 AKI)
238.57
36.58
160.32
54.28
203.36
44.55
[0.54]
[0.07]
[0.61]
[0.20]
[0.11]
[0.06]
% Difference for Fuel D
-1.47%
-0.47%
-0.95%
-0.92%
-1.29%
-0.72%
Significant at 95%
Confidence?
Yes
No
Yes
Yes
Yes
Yes
While the combined-cycle differences found from chassis dynamometer testing were
statistically significant, the very small difference in fuel economy was less than typical inter-
A-35
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTTV-G
laboratory uncertainty during fuel economy testing (e.g. ± 2% of MPG). In the case of fuel D,
the reduction in carbon content of the fuel from E10 blending approximately (or slightly more
than) offsets differences due to the reduced net energy content relative to fuel C. Particularly
when comparing either the EPA chassis-dynamometer drive cycle results to the region labeled
"City/Highway Critical area" with the AAM/USCAR difference map, it is not clear how such
results could have been reported by AAM without significant deficiencies in testing, data
reduction, data interpolation, and/or modeling. Ultimately, without the underlying data, it is
impossible to determine the specific sources of deficiencies in the "difference map" shared by
AAM.
A-36
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTTV-G
References
1 Alliance of Automobile Manufacturers Comments on Draft Technical Assessment Report: Midterm Evaluation of
Light-Duty Greenhouse Gas Emission Standards and Corporate Average Fuel Economy Standards for Model Years
2022-2025 (EPA-420-D-16-900, July 2016), September 26, 2016, p. iii.
2 Novation Analytics, Technology Effectiveness - Phase I: Fleet-Level Assessment, version 1.1, prepared for
Alliance of Automobile Manufacturers & Association of Global Automakers, October 19, 2015.
3 Novation Analytics, Technology Effectiveness - Phase II: Vehicle-Level Assessment, version 1.0, prepared for
Alliance of Automobile Manufacturers & Association of Global Automakers, September 20, 2016.
4 Alliance Comments, p. iii.
5 Alliance Comments, p. ix.
6 David Cooke, "Five Deceptive Tactics Automakers Are Using to Fight Fuel Economy Standards," July 13, 2016,
Union of Concerned Scientists, http://blog.ucsusa.org/dave-cooke/antomakers-fiiel-economv-standards.
7 Fleet-Level Assessment, p. 12.
8 Greenhouse Gas Emissions Standards for Light-Duty Vehicles: Manufacturer Report for the 2014 Model Year,
U.S. EPA Report EPA-420- percent-15-026, December 2015.
9 Alam Baum and Dan Luria, "Why We Believe the Auto Alliance Review of Fuel Economy Standards Misses the
Mark," July 6, 2016, Ceres, https://www.ceres.org/press/blog-Dosts/anto-aHiaiice-review-misses-the-mark/.
10 Fleet-Level Assessment, p.7.
11 Fleet-Level Assessment, p. 78.
12 Alliance Comments, p. 61.
13 "Infiniti VC-Turbo: The world's first production-ready variable compression ratio engine," Nissan Motor
Corporation, September 29, 2016, https://newsroom.nissan-global.com/releases/infiniti-vc-tHrbo-the-worlds-first-
production-readv-variable-compression-ratio-engine?querv=vc-turbo.
Fleet-Level Assessment, p. 13.
15 Alliance Comments, p. iv.
16 Vehicle-Level Assessment, pp. 8-9.
17 Vehicle-Level Assessment, p. 8.
18 First law of thermodynamics, Wikipedia https://en.wikipedia.org/wiki/First law_ of ^thermodynamics .
19 Vehicle-Level Assessment, p. 20.
20 David Cooke, "Five Deceptive Tactics Automakers Are Using to Fight Fuel Economy Standards," July 13, 2016,
Union of Concerned Scientists, http://blog.ncsnsa.org/dave-cooke/antoma.kers-fiiei-economv-standards.
21 Vehicle-Level Assessment, pp. 27-28.
22 Vehicle-Level Assessment, p. 20.
23 Vehicle-Level Assessment, p. 28.
24 Vehicle-Level Assessment, Figures 5 and 6, pp. 26 and 29.
25 Alliance Comments, p. iv.
26 Vehicle-Level Assessment, p. 8.
27 Vehicle-Level Assessment, pp. 38-39.
28 Thomas, J., "Vehicle Efficiency and Tractive Work: Rate of Change for the Past Decade and Accelerated Progress
Required for U.S. Fuel Economy and CO2 Regulations," SAEInt. J. Fuels Lubr. 9(l):290-305, 2016,
doi: 10.4271/2016-01-0909.
29 Alliance Comments, p. 17.
30 Thomas, J., "Drive Cycle Powertrain Efficiencies and Trends Derived from EPA Vehicle Dynamometer Results,"
SAE Int. J. Passeng. Cars-Mech. Syst. 7(4):2014, doi: 10.4271/2014-01-2562.
31 "Road Load Power, Test Weight and Inertia Weight Class Determination." FR 23921 186.129-80. U.S.
Environmental Protection Agency. P. 502. 4 May 1999. 10 November 2016.
32Backstrom, A. "Brake Drag Fundamentals," SAE Technical Paper 2011-01-2377, doi:10.4271/2011-01-2377
33 Shevket, C., Ciulla, L., and Re, P., "Development of Low Friction and Light Weight Wheel Hub Units to Reduce
both the Brake Corner Unsprung Mass and Vehicle CO2 Emission (Part 1-Friction)," SAE Technical Paper 2010-01-
1706, 2010, doi: 10.4271/2010-01-1706.
34 "Dynamometer Drive Schedules." U.S. Environmental Protection Agency. 27 April 2016.
https://www.epa.gov/vehicle-and-fuel-emissions-testing/dynamometer-drive-schedules.
A-37
-------
Appendix D - EPA Comparison Testing Performed on MY2014 Mazda SKYACTTV-G
35 "Environmental Protection Agency 2016. Light-Duty Automotive Technology, Carbon Dioxide Emissions, and
Fuel Economy Trends: 1975 through 2016." U.S. EPA-420-R-16-010, Office of Transportation and Air Quality,
November 2016.
36 www.fiieleconomv.gov.
37 Environmental Protection Agency 2016. Light-Duty Automotive Technology, Carbon Dioxide Emissions, and
Fuel Economy Trends: 1975 through 2016. U.S. EPA-420-R-16-010, Office of Transportation and Air Quality,
November 2016.
38 Environmental Protection Agency 2016. Greenhouse Gas Emission Standards for Light-Duty Vehicles:
Manufacturer Performance Report for the 2015 Model Year, Office of Transportation and Air Quality, EPA-420-R-
16-014, November 2016.
A-38
------- |