Advancing Life Cycle Comparisons of Future Alternative Light-Duty Vehicles

Jason Ming Luk

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Civil Engineering University of Toronto

Advancing Life Cycle Comparisons of Future Alternative Light- Duty Vehicles

Jason Ming Luk

Doctor of Philosophy

Department of Civil Engineering University of Toronto

2015 Abstract The overall objective of this thesis is to systematically compare the life cycle energy use, air emissions and costs of future alternative light-duty vehicles in a more robust manner than is done in the literature. Models are developed using GREET (Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation), Autonomie vehicle simulation software, Vehicle Attribute

Model, Air Pollution Emission Experiments and Policy (APEEP) analysis model, and Crystal

Ball (Monte Carlo analysis). Four questions are investigated:

 Should the transportation sector use ethanol or bio-electricity? Life cycle assessment

results indicate that neither has a clear advantage in terms of greenhouse gas (GHG) emissions

or energy use. This finding is in contrast to those in the literature that favor the use of bio-

electricity because this thesis develops pathways with comparable vehicle characteristics.

 Do plug-in electric vehicles provide incremental life cycle air pollutant impact benefits

over internal combustion engine vehicles using the same primary energy source? The

results based on natural gas-derived fuels show that battery electric vehicles (BEV) may not

provide benefits, in terms of climate change and health impacts, over hybrid electric vehicles

(HEV). This can be attributed to the many sources of uncertainty and stringent tailpipe

emissions regulations.

 How can vehicles be designed to meet future CAFE (Corporate Average Fuel Economy)

standards? Case study results for a reference vehicle show that the 66% increase in fuel

economy targets between model years 2012 to 2025 can be met with a 10% vehicle price

increase (lightweight HEV powertrain), 31% increase in 0-96 km/h acceleration time (smaller

engine), 17% interior volume decrease (smaller body), or 94% driving range decrease (BEV

powertrain), while other attributes are maintained.

 How might CAFE standards affect the ability for non-petroleum vehicles to mitigate

GHG emissions by displacing petroleum vehicles? Life cycle costing results indicate that

there is a financial incentive for automakers to produce CNG vehicles that could emit higher

well-to-wheel GHG emissions on a per kilometer basis than gasoline vehicles. This is

permitted by CAFE standards because non-petroleum fuel incentives allow vehicles using

CNG to be less efficient, and thus potentially more affordable, than those using gasoline.

iii

Acknowledgements

Dr. Heather MacLean for being an infuriatingly great supervisor. Her patience and trust gave me the freedom to make my own mistakes, while her unrelenting expectations never allowed me to become complacent. I am privileged to have the opportunity to continue to work with her.

Dr. Bradley Saville for going far beyond his official position as a committee member. Our high energy/volume debates provoked me to realize the strengths and address the weaknesses of my work.

Dr. Chris Kennedy, Dr. Gregory Keoleian, Dr. Matthew Roorda, Dr. Murray Thomson, Dr.

James Wallace for their roles on my examination committees. Their diverse insights helped refine the direction and academic significance of my research.

Dr. Candace Wheeler, Ian Sutherland and Norm Brinkman for their contributions on behalf of

General Motors. Their industry prospective identified valuable resources and improved the real world relevance of this thesis.

Dr. Clement Bowman, Marjorie Bowman, Paul Price and Suzana Price for generously contributing to the scholarships that have funded my studies.

Kaye and Eleanor Yu for being sources of joy.

Table of Contents

Table of Contents ...... v

List of Tables ...... viii

List of Figures ...... xi

List of Notations ...... xv

Chapter 1 Introduction ...... 1

1.1 Thesis Objectives ...... 5

1.2 Publications contained in this thesis ...... 6

Chapter 2 Background ...... 8

2.1 Light-duty Vehicle Energy Use Policies ...... 8

2.2 Status of Light-Duty Vehicle Powertrains and Fuels ...... 17

2.3 Life Cycle Comparisons of Alternative Light-Duty Vehicles ...... 28

Chapter 3 Methods ...... 36

3.1 Life Cycle Assessment ...... 36

3.2 GREET Model ...... 38

3.3 Air Pollution Emission Experiments and Policy Analysis Model ...... 39

3.4 Autonomie ...... 41

3.5 Vehicle Attribute Model ...... 44

3.6 Monte Carlo Analysis ...... 46

Chapter 4 Life Cycle Assessment of Bioenergy Use in Light-Duty Vehicles ...... 48

4.1 Methods ...... 50

4.2 Results and Discussion ...... 54

Chapter 5 Life Cycle Air Emissions Impacts and Ownership Costs of Light-Duty Vehicles Using Natural Gas As A Primary Energy Source ...... 66

5.1 Methods ...... 67

5.2 Results and Discussion ...... 74

Chapter 6 Vehicle Design Options To Meet 2025 Corporate Average Fuel Economy Standards ...... 86

6.1 Methods ...... 88

6.2 Results and Discussion ...... 95

Chapter 7 Potential Impact of Corporate Average Fuel Economy Standards On The Ability For Non-Petroleum Vehicle To Mitigate Greenhouse Gas Emissions ...... 104

7.1 Methods ...... 106

7.2 Results and Discussion ...... 112

Chapter 8 Conclusion ...... 119

8.1 Chapter Conclusions ...... 119

8.2 Thesis Conclusions ...... 122

8.3 Limitations ...... 123

8.4 Future Research ...... 126

References ...... 129

Appendix A: Chapter 4 Supporting Information ...... 146

Methods Section Details ...... 146

Results ...... 163

Scenario Analysis ...... 166

Appendix B: Chapter 5 Supporting Information ...... 169

Supplemental Methods ...... 169

Ownership Costs ...... 171

Uncertainty and Sensitivity Analysis ...... 176

Supplemental Results ...... 178

Supplemental Scenarios ...... 183

Appendix C: Chapter 6 Supporting Information ...... 185

Methods Details ...... 185

Results Details ...... 199 vi

Appendix D: Chapter 7 Supporting Information ...... 201

Supplemental Methods ...... 201

Supplemental Results ...... 207

vii

List of Tables

Table 4-1: Reference and bioenergy pathways ...... 51

Table 5-1: Key assumptions used to develop fuel cycle and vehicle models ...... 71

Table 7-1: Overview of base case assumptions used in this study ...... 107

Table A-1: Biomass production data from the GREET Fuel-Cycle model7 ...... 148

Table A-2: Physical characteristics for hybrid poplar ...... 148

Table A-3: Chemical production data from MacLean and Spatari175 ...... 149

Table A-4: Bioenergy production data ...... 150

Table A-5: Aspen115 subroutines used to develop production models ...... 150

Table A-6: Base case ethanol production model material flow balance ...... 152

Table A-7: Future ethanol production model material flow balance ...... 153

Table A-8: Base case bio-electricity production model material flow balance ...... 154

Table A-9: Future bio-electricity production model material flow balance ...... 156

Table A-10: Ethanol delivery data from the GREET Fuel-Cycle model7 ...... 157

Table A-11: Reference fuel production data from the GREET Fuel-Cycle model7 ...... 158

Table A-12: Grid-electricity resource mix from the GREET Fuel-Cycle model7 ...... 158

Table A-13: Vehicle design and performance characteristics ...... 159

Table A-14: Mass and battery characteristics of vehicle models created in Autonomie96 ...... 160

Table A-15: Un-weighted fuel consumption results of vehicle models created in Autonomie96 160

Table A-16: Vehicle emissions and fuel consumption ...... 161

Table A-17: Vehicle cycle results for vehicle models based on GREET Vehicle-Cycle model7162 viii

Table A-18: Life cycle pathway results ...... 164

Table A-19: GHG emissions, fossil energy and petroleum mitigation results ...... 165

Table B-1: CNG fuel tank and BEV battery cost and mass parameters ...... 173

Table B-2: CV and BEV powertrain cost, mass and efficiency parameters ...... 174

Table B-3: Vehicle maintenance cost and frequency parameters ...... 175

Table B-4: Key life cycle inventory assumptions used to develop Monte Carlo and sensitivity analyses ...... 176

Table B-5: Key ownership cost and emissions impact assumptions used to develop Monte Carlo and sensitivity analyses ...... 177

Table B-6: Specific costs of CAC emissions impacts used to develop Monte Carlo and sensitivity analyses ...... 177

Table B-7: Incremental life cycle ownership and emissions impact cost 90% confidence intervals for supplementary scenarios ...... 183

Table C-1: Chevy Equinox-like and Honda Accord-like components ...... 187

Table C-2: Gasoline Chevy Equinox-like and Honda Accord-like vehicle specifications for a range of engine power ratings ...... 188

Table C-3: Plug-in electric Chevy Equinox-like vehicle specifications for range of motor power ratings and battery capacities ...... 189

Table C-4: Model Year 2012 vehicle manufacturing costs ...... 191

Table C-5: Model Year 2015 vehicle manufacturing costs ...... 192

Table C-6: Model Year 2020 vehicle manufacturing costs ...... 193

Table C-7: Model Year 2025 vehicle manufacturing costs ...... 194

Table C-8: Parameters for calculating price of added fuel efficiency technologies from Vehicle Attribute Model3 ...... 196

Table C-9: 2012 Reference vehicle model specifications ...... 199

Table C-10: Vehicle design option model specifications ...... 200

Table D-1: Fuel economy and price of base vehicle models ...... 203

Table D-2: Incremental fuel economy and price from added fuel efficiency technologies ...... 204

Table D-3: Incremental fuel economy and price from CNG modifications ...... 205

Table D-4: Monte Carlo and sensitivity analyses assumptions ...... 206

List of Figures

Figure 2-1: Corporate Average Fuel Economy standards over time (adapted from National Highway Traffic Safety Administration)9 ...... 9

Figure 2-2: Historical average US light-duty vehicle market attributes2 (adapted from An and DeCicco)33 ...... 10

Figure 2-3: Fuel economy improvement and cost increases from fuel efficiency technologies2 . 11

Figure 2-4: Simplified schematic of vehicle powertrain configurations ...... 13

Figure 2-5: Current Zero Emission Vehicle program sales requirements10 ...... 14

Figure 2-6: Renewable Fuel Standard11 volume targets ...... 15

Figure 2-7: Low Carbon Fuel Standard carbon intensity target for gasoline substitute fuels12 .... 16

Figure 2-8: US 2012 light-duty vehicle energy use by fuel type2 ...... 18

Figure 2-9: US model year 2012 light-duty vehicle sales by fuel and powertrain type2 ...... 19

Figure 2-10: Recent US transportation fuel prices2 ...... 19

Figure 2-11: Historical oil price1 and average US light-duty vehicle market attributes2 ...... 20

Figure 2-12: US 2012 non-petroleum transportation fuel use1 ...... 21

Figure 2-13: Non-petroleum transportation fuel stations62 ...... 22

Figure 2-14: US domestic natural gas and crude oil production forecast2 ...... 23

Figure 2-15: US model year 2013 HEV sales60 ...... 25

Figure 2-16: US model year 2013 PHEV sales60 ...... 26

Figure 2-17: US model year 2013 BEV sales60 ...... 27

Figure 2-18: GREET comparison of GHG emissions from model year 2020 vehicles using 100- and 20-year Intergovernmental Panel on Climate Change (IPCC) global warming potentials7 ... 33 xi

Figure 3-1: Simplified overview of life cycle stages modelled within GREET7 ...... 38

Figure 3-2: Simplified overview of components modelled within Air Pollution Emission Experiments and Policy analysis model97 ...... 40

Figure 3-3: Simplified overview of Autonomie conventional and battery electric vehicle model components96 ...... 42

Figure 3-4: Illustrated example of the relationship between vehicle price and incremental fuel economy improvements from the Vehicle Attribute Model3 ...... 44

Figure 3-5: Example of frequency distribution graph produced from a Monte Carlo analysis .... 46

Figure 4-1: a) Lignocellulosic biomass use, b) total energy use, and c) GHG emissions for reference and bioenergy pathways ...... 56

Figure 4-2: GHG emissions mitigation resulting from displacing reference fuels with bioenergy alternatives: a) comparing mitigation potential of ethanol with that of bio-electricity, and b) sensitivity of mitigation potential of ethanol to ethanol yield ...... 59

Figure 5-1: Base case life cycle (a) energy use, (b) CO2, (c) NOx, (d) SOx, (d) PM2.5 and (d) VOC emissions inventory results ...... 76

Figure 5-2: Base case life cycle (a) GHG climate change impacts, (b) CAC health impacts and (c) ownership costs ...... 78

Figure 5-3: Life cycle incremental (a) benefit-cost Monte Carlo analysis, (b) benefit sensitivity analysis, and (c) cost sensitivity analysis results ...... 79

Figure 6-2: Overview of mid-price scenario models used to develop the; a) vehicle price option, b) vehicle acceleration option, c) vehicle size option and, d) vehicle driving range option .. Error! Bookmark not defined. xii

Figure 6-3: New vehicle price curves representing, a) the Vehicle Price Option and vehicles with constant fuel economy, b) the Vehicle Acceleration Option and vehicles with constant 0-96 km/h acceleration times, c) the Vehicle Size Option and vehicles with constant interior volumes, and d) the Vehicle Driving Range Option and vehicles with different driving ranges...... 98

Figure 7-1: illustrative comparison of how petroleum and non-petroleum vehicle fuel economy can evolve to meet or exceed CAFE standards ...... 106

Figure 7-2: Relationship between vehicle price and fuel economy for a) internal combustion engine vehicles and b) battery electric vehicles ...... 109

Figure A-1: Pathway System Boundaries ...... 147

Figure A-2: Flow diagram of the ethanol production process ...... 152

Figure A-3: Flow diagram of the base case bio-electricity production process ...... 154

Figure A-4: Flow diagram of the high efficiency bio-electricity production process ...... 155

Figure A-5: Fuel production energy balance ...... 163

Figure A-6: Life cycle sensitivity of total energy use to co-product scenarios ...... 166

Figure A-7: Life cycle sensitivity of net GHG emissions to co-product ...... 166

Figure A-8: Life cycle total energy use results for bioenergy production efficiency scenarios . 167

Figure A-9: Life cycle net GHG emissions results for bioenergy production efficiency scenarios ...... 167

Figure A-10: Life Cycle total energy use results for vehicle efficiency scenarios ...... 168

Figure A-11: Life Cycle net GHG emissions results for vehicle efficiency scenarios ...... 168

xiii

Figure B-1: Incremental costs of changes in relative fuel economy ...... 172

Figure B-2: Life cycle CH4 and N2O emissions disaggregated by life cycle stage and life cycle GHG and CAC impacts disaggregated by emission ...... 179

Figure B-3: Life cycle energy use, CO2, CH4 and N2O emission Monte Carlo analysis results, including 90% confidence intervals in the legend...... 180

Figure B-4: : Life cycle NOx, SOx, VOC and PM2.5 emission Monte Carlo analysis results, including 90% confidence intervals in the legend...... 181

Figure B-5: : Life cycle air emissions impacts and ownership costs Monte Carlo analysis results, including 90% confidence intervals in the legend...... 182

Figure B-6: Incremental life cycle ownership and emissions impact cost results for supplementary scenarios ...... 184

Figure C-1: Flow chart depicting base vehicle model development with Autonomie96 ...... 186

Figure C-2: PSFI projected to 2025 based on 1977-2005 data33 (adapted from An and DeCicco)33 ...... 198

Figure D-1: Overview of vehicle models ...... 201

Figure D-2: Histogram and 90% confidence intervals (CI) of incremental ownership costs and well-to-wheel GHG emissions relative to gasoline use ...... 207

xiv

List of Notations

APEEP Air Pollution Emissions Experiments and Policy Analysis model

BEV Battery electric vehicle

Bio-e Biomass-derived electricity

CAC Criteria air contaminant

CAFE Corporate Average Fuel Economy

CNG Compressed natural gas

CV Conventional vehicle

E85 85% ethanol and 15% gasoline by nominal volume

FCEV Fuel cell electric vehicle

GREET Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation model

Grid-e Grid-derived electricity

HEV Hybrid electric vehicle

ICE Internal combustion engine

ICEV Internal combustion engine vehicle

LCFS Low Carbon Fuel Standard

NGCCe Natural gas combined cycle-derived electricity

NOx Nitrogen oxides

PHEV Plug-in hybrid electric vehicle

PM2.5 Particulate matter with a diameter of less than 2.5 micrometers

SOx Sulphur oxides

SUV Sport utility vehicle

VKT Vehicle kilometer travelled

VOC Volatile organic compound

xv 1

Chapter 1 Introduction

The global transportation sector relies on petroleum fuels for 92% of its energy requirements.1 This dependency is, in part, because the sector is characterized by decentralized, mobile loads that demand energy dense fuels that can be conveniently and affordably distributed and stored. Although there are concerns over volatile oil prices,2 the use of petroleum fuels continues because vehicle powertrains based on an internal combustion engine and petroleum fuel tank are relatively affordable.3

Petroleum use in the transportation sector also has negative environmental and social implications. Petroleum fuels are carbon dense, which has led to the transportation sector comprising 35%1 of US and 28%4 of Canadian greenhouse gas (GHG) emissions. The combustion of petroleum fuels has also resulted in the transportation sector being responsible for 1 5 54% of US and 55% of Canadian nitrogen oxide (NOx) emissions. Compared to other sectors, the health impact of criteria air contaminant emissions from the transportation sector are also disproportionately high because they tend to be concentrated in populated areas.6 In Canada, the increasing use of petroleum fuels worldwide has led to oil sands development, which raises concerns over the potential for increased emissions.7 In the US, despite an increase in domestic petroleum production, imported crude is still forecasted to be approximately 40% of the US petroleum supply for decades to come.2 This dependency on petroleum fuels has energy security implications that threaten economic and political stability.8

The severity of the concerns over petroleum use has resulted in an array of public policies. Many of these target light-duty vehicles (i.e., passenger cars, light trucks, sport utility vehicles and minivans), which comprise the majority of transportation sector energy use.2 Policies include Corporate Average Fuel Economy (CAFE) standards,9 Zero Emission Vehicle programs,10 Renewable Fuel Standards11 and Low Carbon Fuel Standards.12 Both automakers and fuel producers have responded to regulations by developing and implementing new technologies.13 Consumers have responded by purchasing a gradually increasing variety of vehicle powertrains and transportation fuels.13

US CAFE standards9 require automakers to improve the efficiencies of petroleum use in light- duty vehicles. This policy was first developed in response to the Arab oil embargoes of the 1970’s.9 Since model year 1978, automakers have been required to meet fleet average fuel economy targets.9 These targets increased annually through the 1980s before largely stagnating.9 Recent amendments, which are also motivated by climate change concerns, increase the targets for model years 2012 through to 2025.9 Automakers have responded with the development and implementation of fuel efficiency technologies, including hybrid electric vehicle (HEV) powertrains.13 In Canada, this policy has been adopted in the form of Corporate Average Fuel Consumption standards, which were originally voluntary but are now mandatory.14

The California Zero Emission Vehicle program10 effectively requires automakers to produce vehicles that do not use any petroleum fuels at all. The program began in model year 1998 before being suspended due to a legal challenge by automakers on the basis of technological limitations and consumer interests.10 Battery technological developments allowed the program to be revived starting in model year 2012 with increasing targets through 2025.10 Automakers have responded with the development and sales of plug-in hybrid electric vehicles (PHEV) and battery electric vehicles (BEV) powered by grid-electricity,15 which is not typically generated with petroleum.2 Canada does not have similar legislation; however, as with California and the rest of the US,16 plug-in electric vehicle sales in Ontario17 and Quebec18 are directly supported by government financial incentives.

The US Renewable Fuel Standard11 requires obligated parties (fuel producers) to sell renewable fuels.11 Fuel producers have responded by blending biofuels, particularly corn ethanol to meet the targets since 2006. Recent revisions increase volume targets until 2022 and have resulted in virtually all US gasoline vehicles now consuming gasoline blended with, on average, 10% ethanol fuel.11 The new legislation also requires the production of advanced biofuels, such as ethanol produced from lignocellulosic biomass instead of corn starch.11 Canada has a less stringent Renewable Fuel Regulation, which requires 5% renewables within gasoline fuels and 2% renewables within diesel fuels (with some exceptions).19

The California Low Carbon Fuel Standard12 required fuel producers to reduce the carbon intensity, on a life cycle basis, of the state’s transportation fuels beginning in 2011 and through to 2020. Fuel producers have responded by developing infrastructure for alternative fuels, such

3 as compressed natural gas (CNG), in addition to ethanol and electricity.2 In Canada, British Columbia also has a Low Carbon Fuel Standard.20

The policies and technologies discussed above collectively are assisting to reduce the petroleum fuel use in light-duty vehicles. Although an important objective, there can be unintended negative environmental and economic consequences associated with some of these actions. The environmental and financial implications of the policies and associated technologies should be understood by taking a systems level (i.e., life cycle) approach to avoid unintended consequences and the inefficient utilization of resources.

Life cycle assessments21 can be used to evaluate the environmental impacts of the technologies supported by the aforementioned policies. Life cycle assessment involves identifying the life cycle stages of a product or process (e.g., fuel production and consumption, in addition to vehicle production, maintenance and disposal) and its functional unit (e.g., vehicle kilometer travelled).21 A life cycle inventory of environmental inputs and outputs is then compiled for the individual life cycle stages (e.g., quantities of GHG emissions).21 The life cycle inventory results can then be weighted to estimate environmental impact (e.g., global warming potential).21 These data can be compared against other metrics (e.g., land and water use).

Life cycle costing can be used to supplement life cycle assessments. Vehicle purchase price and ongoing fuel costs are particularly important factors to consider when evaluating transportation technologies. This information is essential for understanding the cost-effectiveness of using alternative technologies to achieve societal objectives.

Evaluating alternative vehicles on a life cycle basis is not a novel concept but analyses in the literature can be improved or expanded upon. In particular, many life cycle studies are focused on the characteristics of vehicle fuels, but not the vehicles themselves. For example, Choudhary et al.22 analyzes GHG emissions from different transportation fuels produced from biomass; while the study includes an uncertainty analysis based on probability distribution functions for many fuel production variables, it uses a single point estimate for non-plug-in-vehicle fuel economy and neglects to provide any description or specifications for the plug-in vehicle it is compared with. Campbell et al.23 conducts a similar analysis and does provide vehicle descriptions; however, the choice of vehicles conflate many design factors because of substantial differences in vehicle characteristics, including size and highway capability.

There are many life cycle assessments that do make comparisons of similar vehicles. However, these studies are generally focused on vehicle characteristics and do not clearly distinguish between the environmental merits that can be directly attributed to the vehicles themselves versus those of the primary energy sources they can use. For example, the National Research Council6 compares the life cycle air emissions impacts of alternative vehicles and shows BEVs can result in higher “damages” than gasoline vehicles because much of the electricity in the US is generated from coal; however, electricity is produced from many different fuels and BEV sales are concentrated in regions24 that use little coal.25 Lewis et al.26 and Michalek et al.8 show the life cycle GHG emissions and air emissions impacts, respectively, of BEVs can be higher or lower than those of non-plug-in vehicles depending on the source of electricity, but neither analyze the ability for many sources of electricity to also be used to produce fuels for non-plug-in vehicles. Therefore, it is unclear if the benefits quantified are due to the primary energy source or the alternative vehicle powertrain.

Life cycle assessments of future vehicles neglect to consider the influence of financial and policy considerations. For example, Laser and Lynd27 makes conclusions regarding the life cycle energy use of future vehicles by correcting for differences in driving range between BEVs and non-plug- in vehicles, because BEVs with longer driving ranges have higher mass and thus lower fuel economy; however, BEV driving range is (and will be for the foreseeable future)3 limited by high battery prices. Curran et al.28 analyzes future life cycle GHG emissions of alternative fuels based on the assumption that dedicated CNG vehicle fuel economy improvements will exceed the improvement in gasoline vehicles; however, while gasoline vehicle fuel economy must increase to meet future CAFE standards, current dedicated CNG vehicles already exceed future requirements because of non-petroleum fuel incentives. Therefore, dedicated non-petroleum fuel vehicles can avoid the use of potentially costly fuel efficiency technologies.

Analyses on the impact of policies on future vehicle designs arrive at findings based on a narrow scope. For example, the National Highway Traffic Safety Administration29 assumes vehicle characteristics, such as size and acceleration performance, will remain constant when concluding 2025 CAFE standards will increase vehicle price. Conversely, Knittel30 concludes vehicle size and acceleration performance must be reduced to meet 2020 CAFE standards, based on the assumption that consumers will not pay more for fuel efficient vehicles. The findings from the

National Highway Traffic Safety Administration29 and Knittel30 exclude dedicated non- petroleum fuel vehicles.

1.1 Thesis Objectives

The overall objective of this thesis is to systematically compare the life cycle energy use, air emissions and costs of alternative light-duty vehicles in a more robust manner than is done in the literature. In particular, there is an emphasis on distinguishing among the technological and policy limitations and opportunities. The focus is on the US market, which is similar in many ways to Canada in terms of product offerings and light-duty vehicle policies; however, the US is larger and thus has the benefit of greater data availability. The findings in this thesis are aimed at contributing to the scientific literature as well as informing public policy. This will be done by investigating the following four questions in Chapters 4 through 7:

 Should the transportation sector use ethanol or bio-electricity? The Renewable Fuel Standard11 promotes the use of ethanol, which has become the dominant non-petroleum fuel used in US light-duty vehicles. However, ethanol is produced from biomass, whose production, and thus ability to mitigate greenhouse gas emissions and petroleum use, is limited by feedstock and land availability. The development of plug-in electric vehicles provides another means of utilizing biomass as a transportation energy source. Thus, a life cycle energy use and GHG inventory analysis is conducted for biomass use in both plug-in and non-plug-in vehicle powertrains in Chapter 4.

 Do plug-in electric vehicles provide incremental life cycle air pollutant impact benefits over internal combustion engine vehicles using the same primary energy source? The Zero Emission Vehicle program10 promotes the use of BEVs, which lack tailpipe emissions and are increasingly fuelled by natural gas-derived electricity. The ability for automakers to meet Zero Emission Vehicle program10 requirements by producing low emission CNG vehicles is being phased out. However, CNG vehicles may have lower upstream emissions and ownership costs than BEVs. Thus, the life cycle air emissions impacts and ownership costs of a range of vehicles are evaluated, using natural gas as a common primary energy source in Chapter 5.

 How can vehicles be designed to meet future CAFE standards? The costs and benefits of non-petroleum fuelled vehicles are typically quantified in comparison to petroleum fuelled vehicles. However, recently amended CAFE standards9 will require substantial design changes to light-duty vehicles. Thus, alternative vehicle design options are evaluated for their ability to meet future fuel economy standards in Chapter 6.

 How might CAFE standards affect the ability for non-petroleum vehicles to mitigate GHG emissions by displacing petroleum vehicles? Chapters 4 and 5 compare the potential environmental merits of using alternative fuels in leading edge vehicle technologies; however, these technologies may not be utilized by real world vehicles because of financial and policy considerations. In particular, CAFE standards provide credits for non-petroleum fuel use, which means dedicated non-petroleum fuel vehicles do not require fuel efficiency improvements to meet future fuel economy targets. This can affect low carbon fuel standards, which promote the sale of alternative fuels that are less carbon intensive than petroleum fuels after the relative fuel economy ratings of different vehicles are accounted for. These relative fuel economy ratings can change over time if improvements in fuel efficiency differ between petroleum and non-petroleum fuel vehicles. Thus, in Chapter 7, the life cycle GHG emissions and ownership costs of (potentially) lower efficiency non-petroleum fuel vehicles are compared with petroleum fuelled vehicles designed to meet stringent future CAFE standards.9

1.2 Publications contained in this thesis

I personally conducted the majority of the research, analysis and writing as first author of each of the studies described below. Dr. Heather L. MacLean provided guidance in her role as a co- author of the individual publications and as supervisor of the overall thesis research.

 Luk, J., Pourbafrani, M., Saville, B., MacLean, H. Ethanol or Bio-electricity? Life cycle assessment of bioenergy use in light-duty-vehicles, Environmental Science & Technology, 2013, 47 (18) 10676-10684.

Chapter 4 has been published in Environmental Science & Technology, with the citation information above. Dr. Mohammad Pourbafrani is co-author for his development of ethanol and bio-electricity production models in AspenPlus software. Dr. Bradley Saville is a co-author for

7 his ongoing feedback throughout the research process and help with manuscript revisions. This research was also awarded with a 2013 Outstanding Energy Paper award by the University of Toronto’s Institute of Sustainable Energy and an invited presentation at the 2014 Canada-Korea Scientific Conference.

 Luk, J., Saville, B., MacLean, H. Life cycle air emissions impacts and ownership costs of light-duty vehicles using natural gas as a primary energy source, Environmental Science & Technology, 2015, 49 (8) 5151-5160.

Chapter 5 has been accepted for publication in Environmental Science & Technology, with the citation information above. Dr. Bradley Saville is a co-author for his ongoing feedback throughout the research process and help with manuscript revision. This research was awarded with a third place overall finish at the 2014 AUTO21 Network Centre of Excellence Poster Competition.

 Luk, J., Saville, B., MacLean, H. Vehicle design options to meet 2025 Corporate Average Fuel Economy standards, Energy Policy, in preparation for submission.

Chapter 6 is being prepared for publication in Energy Policy, with the pending citation information above. Dr. Bradley Saville is a co-author for his ongoing feedback throughout the research process and help with manuscript revisions.

 Luk, J., Saville, B., MacLean, H. Potential impact of CAFE standards on the ability for non-petroleum vehicles to mitigate GHGs. Environmental Research Letters, in preparation for submission.

Chapter 7 is being prepared for publication in Environmental Research Letters, with the pending citation information above. Dr. Bradley Saville is a co-author for his ongoing feedback throughout the research process and help with manuscript revisions.

Chapter 2 Background

This thesis focuses on alternative light-duty vehicles in the US. This chapter first discusses the key US light-duty vehicle energy use policies that shape the vehicles and fuels consumers can choose from. Secondly, the significance of different powertrains and fuels are presented in the form of high-level market share statistics along with specific examples to illustrate how these alternatives are manifested in the real world. Finally, a review of life cycle comparisons of alternative light-duty vehicles in the US is conducted to provide insights into the state of our knowledge on the environmental and financial implications of the technologies promoted by the

light-duty vehicle energy use policies.

2.1 Light-duty Vehicle Energy Use Policies

A range of US public policies are aimed at reducing petroleum use in light-duty vehicles. Four key policies are discussed here. These are divided into those that regulate automakers and those that regulate fuel producers.

2.1.1 Automaker Regulations

Automakers are regulated by US CAFE standards9 and the California Zero Emission Vehicle program.10

2.1.1.1 US Corporate Average Fuel Economy Standards

CAFE standards are designed to improve the efficacy of petroleum use in light-duty vehicles beginning in 1978.9 This legislation was passed in response to the Arab Oil Embargo, which restricted oil imports to the US, thus increasing oil prices and interrupting economic growth. A committee of the National Research Council31 was formed in 2001 to review the literature on the historical impact of the CAFE standards,9 which by then had plateaued. They found that CAFE standards only reinforced the trend of vehicle size and weight reductions initiated by high oil prices, but helped maintain fuel economy after oil prices began to fall.31 However, when CAFE standards plateaued in the 1990’s (Figure 2-1) there was a slight decrease in fuel economy when light-duty trucks, including sport utility vehicles (SUVs), were increasingly used as consumer

9 vehicles.31 Light-duty trucks were considered work vehicles, and thus were subjected to less stringent fuel economy targets when CAFE standards were established.31 The committee found the shift towards these heavier vehicles was beneficial in terms of reducing fatalities from collisions.31 However, credits for flex fuel vehicles (which can consume gasoline blended with up to 85% ethanol by nominal volume) were found to increase petroleum consumption because these vehicles primarily used gasoline fuel and allowed automakers to avoid fuel economy improvements.31 Anderson and Sallee32 concluded that meeting CAFE standards by adding flex fuel capability was more affordable for automakers than increasing fuel economy alone. The overall increase in fuel economy was found to contribute to the “rebound effect,” which resulted in an increase in driving demand as consumers took advantage of fuel cost savings.31

CAFE standards have recently been amended for model years 2012 through 2025.9 As shown in Figure 2-1, fuel economy targets for both cars and trucks will increase throughout this timeframe.9 The definition of a truck was narrowed to exclude many SUVs, thus limiting the ability for automakers to meet fuel economy targets by classifying consumer vehicles as trucks.9 The targets are now scaled by vehicle footprint (wheelbase multiplied by track) to limit the ability for automakers to meet fuel economy targets by reducing vehicle size (and potentially safety).9 Credits for flex fuel vehicles are being phased out, but the incentives will remain for dedicated non-petroleum fuelled vehicles.9

30 Duty Vehicle Duty Fuel Vehicle - Passenger Cars 20

Economy(mpg) Light Trucks

10 Average Average Light 0 1975 1985 1995 2005 2015 2025 Vehicle Model Year

Figure 2-1: Corporate Average Fuel Economy standards over time (adapted from National Highway Traffic Safety Administration)9

An and DeCicco33 analyzed the historical trends in light-duty vehicle characteristics to provide insights into future CAFE standards Figure 2-2 shows that on average, current light-duty vehicles are more fuel efficient, while being larger and more powerful than in the past. An and DeCicco33 found that the product of vehicle power-to-weight ratio, interior volume, and fuel economy increased relatively linearly over time. They concluded that trade-offs among these factors showed aggregate energy efficiency improvements that could be used to reduce fuel consumption or to improve acceleration performance or increase size.33 Bandivakar et al.34 extrapolated the historical trends and termed the trade-off as an emphasis on reducing fuel consumption versus improving performance and/or size. Cheah et al.35 applied this method to evaluate 2016 CAFE standards and Knittel30 used a similar approach to analyze 2020 CAFE standards. Both concluded that using future energy efficiency improvements to improve fuel economy (instead of vehicle size or performance) would not be sufficient to meet future CAFE standards. Cheah et al.35 suggested that plug-in electric vehicles could be introduced to meet standards while Knittel,30 who did not evaluate plug-in vehicles, concluded that reducing average vehicle size and/or performance to historical levels would be necessary.

180%

160%

140% Fuel Economy 120% Power:Weight Ratio Interior Volume

100% Metric Relative Metric Relative 1977 To Levels 80% 1977 1982 1987 1992 1997 2002 2007 2012 Model Year

Figure 2-2: Historical average US light-duty vehicle market attributes2 (adapted from An and DeCicco)33

The linear improvement in energy efficiency determined in the aforementioned studies considers several factors but excludes the price of further increases. The Energy Information Administration2 has detailed the prices of a range of fuel efficient technologies automakers can

11 utilize to improve vehicle fuel economy, examples of which are shown in Figure 2-3. These include not only technologies that improve powertrain efficiency, but also those that reduce the load applied on these powertrains to improve the efficacy of energy use. Makino36 has discussed in detail Toyota’s efforts to reduce vehicle mass, aerodynamic drag and accessory loads independently of vehicle powertrain to meet future regulations.

$800

Engine Start-Stop $600 8-speed Automatic

$400 Direct Injection 7-speed Automatic Continuously Variable Aerodynamic Transmission Improvements Engine Friction $200 Reduction Improved Accessories Cylinder Deactivation Incremental Incremental Cost (2000 USD) Low Friction Lubricants Low Resistance Tires $0 Aggressive Shift Logic 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% Incremental Fuel Economy Improvement

Figure 2-3: Fuel economy improvement and cost increases from fuel efficiency technologies2

The price of these fuel efficiency technologies are the focus of the Regulatory Impact Analysis of the amended CAFE standards.29 The analysis estimated the average 2025 US light-duty vehicle will be $1870-$2120 higher in price than if CAFE standards, in addition to vehicle size and performance, were held constant at 2012 levels.29 However, the historical trends in Figure 2-2 do not suggest that vehicle size and performance will hold constant.

Recent research on CAFE standards by Shiau et al.37 and Whitefoot and Skerlos38 combined financial and engineering modelling. Shiau et al.37 highlighted the need to balance fuel economy targets with penalties for violations, while Whitefoot and Skerlos38 warned of the potential for footprint-based targets to be a moral hazard that encourages the production of larger vehicles. Both studies arrived at their conclusions by modelling the sensitivity of consumer demand for vehicle size and power to vehicle price and fuel economy. Neither study considered changes to technology nor fuel economy targets over time; therefore, the studies do not provide insight into how stringent future year CAFE standards can be met.

2.1.1.2 California Zero Emission Vehicle Program

Whereas CAFE standards subject US automakers to performance targets, California’s Zero Emission Vehicle program establishes technology requirements.10 The different technologies are discussed below and illustrated in Figure 2-4. (Detailed descriptions of these different vehicle powertrains are provided in Section 2.2.2.) The program was established in response to air quality concerns and required large automakers to produce increasing quantities of BEVs (battery electric vehicles) and/or FCVs (fuel cell vehicles) beginning in 1998.10 Bedsworth and Taylor39 reviewed the evolution of this policy, which was originally based on the projection that battery prices could be reduced sufficiently to be competitive with conventional vehicles (CVs) in an eight to 13 year time frame. Unfortunately, the nickel metal hydride and lead acid battery price did not improve as expected and automakers and battery packs that were projected by regulators to cost $1350 were still around $20,000 by 2000.39 The program was amended in 2003 to allow automakers to temporarily comply by selling low emission vehicles, including PHEVs (plug-in hybrid electric vehicles), HEVs (hybrid electric vehicles), and fuel efficient CVs.39 PHEVs and HEVs use smaller, and thus more affordable, batteries and benefited from the development of electric vehicle technologies by automakers.39 Automakers would eventually win a lawsuit to have Zero Emission Vehicle requirements temporarily eliminated.39

Conventional Vehicle (CV)

External Energy Source Internal Internal Energy Flow Combustion Wheels Engine

Parallel Series-Parallel Split Hybrid Electric Vehicle (HEV) Hybrid Electric Vehicle (HEV)

Internal Internal Combustion Combustion Engine Engine

Wheels Wheels

Electric Electric Battery Battery Motor Motor

Series Series-Parallel Split Plug-in Hybrid Electric Vehicle (PHEV) Plug-in Hybrid Electric Vehicle (PHEV)

Internal Internal Combustion Combustion Engine Engine Wheels

Electric Electric Battery Wheels Battery Motor Motor

Battery Electric Vehicle (BEV)

Electric Battery Wheels Motor

Figure 2-4: Simplified schematic of vehicle powertrain configurations

The policy has since been revised and major automakers in California are again required to sell Zero Emission Vehicles beginning in model year 2012.10 There are also requirements for transitional Zero Emission Vehicles (e.g., PHEV), Advanced Technology Partial Zero Emission Vehicles (e.g,, HEV), and Partial Zero Emission Vehicles (e.g., fuel efficient CV), as outlined in Figure 2-5. There is less industry opposition to the current policy, as compared to the original, in part because automakers now benefit from advances in the development of battery technology.40

30%

20% Partial Zero Emission Vehicles (e.g., fuel efficient CV)

Advanced Technology Partial Zero Duty Vehicle Sales)

- Emission Vehicles (e.g., HEV) 10% Transitional Zero Emission Vehicles (e.g., PHEV) Zero Emission Vehicles

Vehicle Sales Sales Vehicle Requirements 0% (e.g., BEV) (Portionof Light

Vehicle Model Year

Figure 2-5: Current Zero Emission Vehicle program sales requirements10

Unfortunately, the costs of modern batteries remain relatively high, particularly for BEVs. Kromer and Heywood41 estimated that BEV battery packs will still cost $8,600-$12,000 by 2030. Argonne National Laboratory42 estimated that BEV battery packs will cost $11,000-$20,000 in 2015 and still be $6,000-$9,000 by 2045. A committee for the National Research Council43 tasked with assessing the costs of fuel economy technologies, excluded a quantitative analysis of BEVs altogether because its expectation was that there would not be significant numbers of them produced during the study’s 15 year timeframe. A committee for the National Petroleum Council found estimates for 2020 battery packs to range from $5000-$15000.3 Fiat-Chrysler has claimed losses of $14,000 on every Fiat 500e sold to meet Zero Emission Vehicle program requirements.44 Sales are aided by federal and state level tax credits provided to plug-in electric vehicles.16, 45

2.1.2 Fuel Producer Regulations

Fuel producers are regulated by the US Renewable Fuel Standard and the California Low Carbon Fuel Standard.

2.1.2.1 US Renewable Fuel Standard

The US Renewable Fuel Standard11 originally mandated that 2.78% of gasoline sold in calendar year 2006 be comprised of renewable fuels. There is controversy regarding of the feedstock of this ethanol, which is largely corn starch.46 In response, Farrell et al.46 conducted a meta-analysis and found that there are environmental benefits, in terms of a reduction in fossil fuel use and GHG emissions from displacing petroleum fuel with corn ethanol use, but they may be minor. Wang et al.47 provided a more nuanced study that showed that any environmental benefits from corn ethanol use are highly sensitive to the production technologies and types of fossil fuels used.

The standard has since been revised to require the production of corn ethanol with life cycle GHG emissions at least 20% lower than those of gasoline. Additionally, increasing volumes of renewable fuels are to be produced over time, as shown in Figure 2-6. A portion of these fuels must be advanced biofuels, such as cellulosic biofuels produced from non-food crops and have life cycle GHG emissions 60% lower than those of gasoline. The scope of these calculations must include land use change, which Searchinger et al.48 suggested could substantially contribute to GHG emissions.

30 Renewable Fuels (e.g., corn ethanol) 20 Advanced Biofuels (e.g., cellulosic ethanol)

(Billion (Billion Gallons) 10

Volume of Renewable Volume Renewable of Fuels 0

Calendar Year

Figure 2-6: Renewable Fuel Standard11 volume targets

Advanced biofuel provisions can help reduce GHG emissions to a greater extent than corn ethanol but fuel producers have not met volume targets. Brown and Brown49 recently reviewed the state of cellulosic ethanol in the US and highlight the continued lack of commercial production (<20 million gallons per year) despite government mandates and incentives, though some facilities have since opened. In response, annual changes to renewable volume obligations have been made to reduce advanced biofuel targets.11

2.1.2.2 California Low Carbon Fuel Standard

The California Low Carbon Fuel Standard12 establishes performance targets for fuel producers to who must reduce the life cycle carbon intensity (g CO2e/MJ) of the state’s transportation fuels beginning in calendar year 2011. The carbon intensity targets for gasoline replacement (as opposed to diesel replacement) fuels are illustrated in Figure 2-7. Early studies by Zhang et al.50 and Kaufman et al.51 concluded that the introduction of cellulosic ethanol fuels could meet the targets, while Yeh et al.52 and Andress et al.53 identified that the use of electricity would also be sufficient.

100

eq/MJ) 90

(g CO (g

Wheel Wheel GHG Emissions -

to 85

- Well 80 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Calendar Year

Figure 2-7: Low Carbon Fuel Standard carbon intensity target for gasoline substitute fuels12

Sperling and Yeh54 commended the Low Carbon Fuel Standard for being a carbon intensity based policy that does not suffer from “fuel du jour phenomenon.” However, the policy is complicated by the use of fuel carbon intensities that are adjusted with “somewhat arbitrary”53

17 energy economy ratios to address the fact that not all fuels can be used interchangeably. For example, the carbon intensity of electricity is divided by an energy efficiency ratio of 3.4 because plug-in electric vehicles are more fuel efficient than gasoline internal combustion engine vehicles.12 Unfortunately, vehicles change over time and there is no single objectively correct energy economy ratio.53

The precision of the carbon intensity calculations required by the Low Carbon Fuel Standard may overstate the precision of our understanding of real world emissions. Venkatesh et al.55 analyzed natural gas fuels, while Mullins et al.56 and Kocoloski et al.57 evaluated ethanol, and each found that the high uncertainty in estimating life cycle emissions can result in conditions whereby ostensibly low carbon fuels can have higher GHG emissions than gasoline. In the case of ethanol fuels, indirect land use change is a particularly substantial source of uncertainty. DeCicco58 criticized the low carbon fuel policies (including the Renewable Fuel Standard) for using life cycle carbon intensities as if they were a fuel property, as opposed to being a product of complex systems.

2.2 Status of Light-Duty Vehicle Powertrains and Fuels

Light-duty vehicles are defined as having a gross vehicle weight rating of 8500 lbs (3900 kg) or a curb weight less than 6000 lbs (2700 kg) and a base vehicle frontal area of 45 ft2 (4.2 m2) or less.59 These include passenger cars, light trucks, sport utility vehicles, and minivans.9 They are discussed here in terms of powertrain type: either conventional vehicles (CVs) or electric vehicles.

2.2.1 Conventional Vehicles (CVs)

CVs comprise 96% of the US light-duty vehicle market.2 The wheels of these vehicles are driven exclusively by an internal combustion engine. Both petroleum and non-petroleum fuels can be used by internal combustion engines.

2.2.1.1 Petroleum fuels

Almost all US light-duty vehicle energy use is from petroleum fuels, as shown in Figure 2-8. Gasoline fuels (including those containing up to 10% ethanol) comprises of 99% of light-duty vehicle energy use.2 The remaining 1% is divided among diesel, liquefied petroleum gas (propane) and non-petroleum fuels.2

Diesel Gasoline (including blends with low ethanol Liquefied concentrations) Petroleum Gas

Non-Petroleum Fuel

Figure 2-8: US 2012 light-duty vehicle energy use by fuel type2

The majority of new light-duty vehicles in the US are gasoline-fuelled CVs, as shown in Figure 2-9.2 Gasoline is also used in 12% and 3% of the market classified by flex fuel vehicles and hybrid electric vehicles, respectively.2 Flex fuel and hybrid electric vehicles are further discussed in Sections 2.2.1.2 and 2.2.2.1, respectively.

Gasoline Flex-fuel Hybrid Electric Vehicles Vehicles

Gasoline Diesel Conventional Conventional Vehicles Vehicles

Other

Figure 2-9: US model year 2012 light-duty vehicle sales by fuel and powertrain type2

Diesel-fuelled CVs comprise 2% of new light-duty vehicles. Diesel fuel has generally been more affordable than gasoline in the US, on an energy equivalent basis, as shown by the recent history illustrated in Figure 2-10.2 Diesel vehicles also benefit from the use of compression ignition engines, which are 30%-35% more efficient than a comparable gasoline spark ignition engine.15 Consumer diesel vehicles are available, however, interest among US consumers has been harmed by past concerns over noise, poor acceleration, cold weather starting issues, and high emissions.15

40 E85 30 Electricity Gasoline 20 Diesel

10 Propane (2012 (2012 USD per mmBtu) Natural Gas 0

Average Transportation Average US Price Fuel 2011 2012 2013 2014 Calendar Year

Figure 2-10: Recent US transportation fuel prices2

Liquid petroleum gas (propane) is used in less than 1% of new light-duty vehicles.2 This fuel has recently had a lower price than both gasoline and diesel.2 However, compared to other petroleum fuels, propane has a low energy density and thus requires large storage tanks and vehicles using the fuel suffer from short driving ranges.60 Propane vehicles are not widely available and are either converted from gasoline vehicles or special order products for commercial fleets.60

The dependency on petroleum fuels appears to have influenced light-duty vehicle characteristics. Figure 2-11 shows that the real (inflation adjusted) oil price mostly increased in the late 1970’s, before falling in the 1980’s and 1990s, and increasing in the 2000’s.1 In general, increasing fuel efficiency has been correlated with increasing oil prices, and vice versa, though fuel efficiency is a much less volatile variable.2 The market share of trucks, as opposed to cars, is inversely correlated with oil price.2 However, some of these shifts are also due to CAFE standards changes, as discussed in Section 2.1.1.1.

300%

200%

Oil Price Truck Market Share 100%

Fuel Economy Metric Relative Metric Relative 1977 To Levels 0% 1977 1982 1987 1992 1997 2002 2007 2012 Calendar Year for Oil Price and Vehicle Model Year for Vehicle Attributes

Figure 2-11: Historical oil price1 and average US light-duty vehicle market attributes2

2.2.1.2 Biofuels

The vast majority (97%) of non-petroleum fuel use in the US transportation sector is comprised of biofuel use, as seen in Figure 2-12.1 This is a result of the Renewable Fuel Standard (Section 2.1.2.1) requirements and CAFE standards (Section 2.1.1.1) incentives. Both ethanol and biodiesel are produced from biomass feedstock. Biomass is organic material, including agricultural crops or even organic waste. Biofuels are blended with petroleum fuels and there are no dedicated ethanol or biodiesel vehicles (i.e., vehicles that can use biodiesel but cannot use petroleum diesel) sold in the US.15

Ethanol in E85

Biodiesel CNG Ethanol in low concentration blends

LNG Electricity

Figure 2-12: US 2012 non-petroleum transportation fuel use1

Most biofuels are consumed in gasoline vehicles, which typically operate on E10 (gasoline blended with 10% ethanol by nominal volume).60 E85 (85% ethanol and 15% gasoline by nominal volume) flex fuel vehicles are largely similar to gasoline vehicles, but are capable of operating on high concentration ethanol blends, but generally operate on gasoline.60 The concentration of ethanol is limited by issues with cold-weather starting, and the ethanol content in E85 fuelling stations may be even reduced to 51% in colder seasons.60 As shown in Figure 2-13, there are also less than 3,000 E85 fuelling stations, as compared to the approximately 168,000 gasoline/E10 fuelling stations in the US.60, 61 E85 fuelling stations are also concentrated in the US Midwest, near where corn feedstock is grown, as opposed to more highly populated areas.62

Public Private Hydrogen

LNG

Biodiesel

CNG

E85

Electricity

0 2000 4000 6000 8000 10000 12000 Fuelling Stations

Figure 2-13: Non-petroleum transportation fuel stations62

Diesel vehicles can also use biodiesel.60 Biodiesel is typically consumed in the form of B20 (petroleum diesel blended with 20% biodiesel by nominal volume).60 The relative lack of diesel vehicles, as compared to gasoline vehicles, in the US limits the current market for this biodiesel.

2.2.1.3 Natural Gas Fuels

Natural gas comprises the largest share of non-petroleum, non-biofuel transportation energy use. Natural gas has recently been more affordable than both petroleum and E85 fuels, on an energy equivalent basis, as shown in Figure 2-10.2 While both US crude oil and natural gas production have increased in recent years, natural gas production is forecasted to continue to increase for decades to come, as shown in Figure 2-14.2 Its use in the transportation sector is encouraged by incentives within CAFE standards (2.1.1.1) and Low Carbon Fuel Standard (2.1.2.2) requirements.

The only natural gas fuelled consumer light-duty vehicle for sale is the Honda Civic Natural Gas.15 It is a dedicated compressed natural gas (CNG) vehicle, which utilizes a high pressure fuel tank that increases both vehicle mass and price. Soon there will also be CNG/gasoline bi-fuel versions of the Chevy Impala, Chevy Silverado and GMC Sierra available.15

CNG suffers from a relatively low energy density compared to petroleum fuels. This results in vehicles using CNG having a relatively short driving range and/or requiring a large storage tank. There is also a lack of public fuelling stations (less than 1000 in the US) but this is less of an

23 impediment for some commercial fleet vehicles, which can return to private fuelling stations.62 There are nearly as many private CNG fuelling stations as there are public.62

200%

150%

100% Natural Gas Crude Oil

(Relative (Relative to2015) 50% US Domestic US Domestic Production

0% 2015 2020 2025 2030 2035 2040 Year

Figure 2-14: US domestic natural gas and crude oil production forecast2

Some medium- and heavy-duty vehicles also use liquefied natural gas (LNG).60 Natural gas is liquefied under high pressure, low temperature conditions and thus requires costly storage systems.60 This results in higher energy densities than CNG, and thus LNG can be more attractive for applications that require longer driving distances.60

2.2.2 Electric Vehicles

Electric vehicles consist of vehicles whose wheels are propelled by an electric motor, either exclusively or along with an internal combustion engine. They are promoted by CAFE standards (Section 2.1.1.1) and the Zero Emission Vehicle program (Section 2.1.1.2). Non-plug-in electric vehicles use electricity generated via the internal combustion engine or regenerative braking but cannot be charged by an external source of electricity. Plug-in electric vehicles can utilize electricity generated onboard and from an external source. An overview of different electric powertrain configurations is provided in Figure 2-4.

2.2.2.1 Hybrid Electric Vehicles (HEVs)

HEVs are non-plug-in vehicles that have a 3% market share of model year 2012 US light-duty vehicles.2 There are different variations of HEV powertrains illustrated in Figure 2-4. These

24 variations involve pairing an internal combustion engine with an electric motor. The addition of the electric motor can be used to improve the acceleration performance of the vehicle (e.g., first generation Honda Accord Hybrid), or allow the internal combustion engine to be downsized as a means to improve fuel economy (e.g., second generation Honda Accord Hybrid).63

The parallel configuration requires the fewest additional components over a conventional vehicle.63 This configuration utilizes regenerative braking to charge a battery. This energy is later provided to a relatively small electric motor that helps the internal combustion engine propel the wheels during acceleration. The relative simplicity of this powertrain allowed the Honda Insight, which uses this configuration, to be the lowest priced model year 2013 HEV.63

The series-parallel split configuration is a more complex configuration. This configuration is used in the dominant Toyota lineup of HEVs, as shown in Figure 2-15.63 As with the parallel configuration, the wheels are driven in this powertrain by both an internal combustion engine and an electric motor, which uses electricity generated from regenerative braking. The series-parallel split configuration also has the capability to allow the internal combustion engine to power a generator to charge the battery. This added flexibility has the benefit of allowing the internal combustion engine to operate at speeds closer to its optimum efficiency for greater periods of time and thus improve fuel economy. The relative efficiency of this powertrain allowed the Toyota Prius and Prius C vehicles to be the most fuel efficient model year 2013 HEVs.15

HEVs may also be referred to as mild or full hybrids.63 This is a distinction independent of the powertrain configuration. If the motor has insufficient power to propel the wheels independently of the internal combustion engine, as is the case with the Honda Insight, the vehicle is also

25 referred to as a mild hybrid. The Toyota Prius is considered a full hybrid because it is capable of driving, at low speeds, using the electric motor only.

Other Toyota Ford

Toyota Prius C Other Hyundai

GM Toyota Prius Honda Other

Figure 2-15: US model year 2013 HEV sales60

2.2.2.2 Plug-in Hybrid Electric Vehicles (PHEVs)

PHEVs are able to utilize electricity generated from an external source. A larger battery is generally required than those in HEVs, which increases both vehicle price and mass. This has, in part, limited PHEV market share to less than 1% of model year 2012 US light-duty vehicles. PHEVs can have lower emissions than HEVs because of the greater use of the high efficiency electric motor. The efficiency of electric motors can also allow PHEVs to have lower fuel costs than HEVs, despite the relatively high price of electricity, on an energy equivalent basis, as shown in Figure 2-10.

The Toyota Prius Plug-in and Chevy Volt comprise the majority of PHEV sales, as shown in Figure 2-16.60 The Toyota Prius Plug-in has a parallel-series split powertrain similar to its non- plug-in counter part, but with a larger battery.63 The Chevy Volt has a series hybrid powertrain, which is also referred to as an extended range electric vehicle.63 This vehicle’s wheels are propelled by an electric motor and an internal combustion engine is mainly used to charge the

26 plug-in battery only as required to extend driving range. This can be advantageous because batteries suffer from relatively low energy densities, in comparison to gasoline.

Toyota Prius Plug-in Ford Fusion Energi

Other

Ford C-MAX Chevrolet Volt Energi

Other

Figure 2-16: US model year 2013 PHEV sales60

2.2.2.3 Battery Electric Vehicles (BEVs)

BEVs use electric motors to propel their wheels and utilize electricity as its only external energy source. The lack of an internal combustion engine results in driving ranges that are generally shorter than those of a PHEV or HEV, which has helped limit the BEV market share to less than 1% of model year 2012 US light-duty vehicles. Advantages of these vehicles include reduced fuel expenses and complete lack of tailpipe emissions. Compared to other dedicated non- petroleum vehicles, BEVs also have the benefit of being able to utilize a relatively widespread fuelling infrastructure in the form of almost 9000 public and an additional 2000 private electric vehicle charging stations in the US, more than double all of the alternatives combined, as shown in Figure 2-13.62

The Tesla Model S and the Nissan Leaf make up the majority of US BEV sales, as shown in Figure 2-17.60 Many of the other BEVs can be considered “compliance cars” with limited availability.64 These are modified versions of gasoline CVs that are sold or leased to reach the minimum targets required to comply with the Zero Emission Vehicle program.64

Smart for Two EV Tesla Model S

Ford Focus Electric Other Fiat 500E

Chevrolet Spark Nissan LEAF EV Other

Figure 2-17: US model year 2013 BEV sales60

2.2.2.4 Other Electric Vehicles

Fuel cell vehicles are a variety of electric vehicle that are not yet for sale in the US.15 Honda and Hyundai have leased a limited number of vehicles, which help meet Zero Emission Vehicle program requirements.15 This type of vehicle utilizes hydrogen as the external energy source, which is stored in a high pressure storage tank and converted into electricity by a fuel cell system. As with plug-in electric vehicles, fuel cell vehicles suffer from having relatively high costs and a low energy density storage system. However, fuel cell vehicles also suffer from having only 100 fuelling stations in the US, half of which are private.60

Micro hybrids are another variation of electric vehicle that are categorized as CVs in the market share statistics provided in Section 2.2.1, but comprise less than 1% of model year 2012 light- duty vehicle sales.2 Micro hybrids are propelled exclusively by an internal combustion engines, just like other CVs, but utilize select electric vehicle technologies to improve fuel efficiency. In particular, engine start-stop technology shuts down the internal combustion engine to prevent idling and uses an upgraded alternator to restart the engine once the accelerator is pressed. Regenerative braking can also be utilized to provide energy to the alternator.

2.3 Life Cycle Comparisons of Alternative Light-Duty Vehicles

This section reviews the scientific literature on life cycle comparisons of alternative light-duty vehicles. The review provides an overview of the recent literature most relevant to this thesis, but is not exhaustive due to the broad scope of alternative light-duty vehicles. In particular, this thesis focuses on vehicles using electricity, ethanol and CNG, which are fuels promoted by the policies described in Section 2, available in existing fuelling infrastructure and used in vehicles currently available for consumer purchase.15

Life cycle assessments21 can be used to evaluate the environmental impacts of the technologies supported by the aforementioned policies. The literature typically divides the life cycle of a vehicle into fuel cycle and vehicle cycle components. The fuel cycle consists of fuel production and use (i.e., well-to-wheel processes), while the vehicle cycle consists of vehicle production, maintenance and end-of-life processes. Life cycle assessments of vehicles typically use functional unit of a vehicle lifetime or a vehicle kilometer travelled, and present life cycle inventory results for GHG emissions and/or energy use. GHG emissions are commonly weighted by 100-year global warming potential to estimate environmental impact. Life cycle cost analyses are sometimes conducted to supplement discussions of environmental impacts. Life cycle assessment is further discussed in Section 3.

2.3.1 Conventional Gasoline Vehicles

Early life cycle assessments of light-duty vehicles used different approaches but arrived at similar conclusions. USAMP65 produced an early LCA of a conventional gasoline vehicle using detailed vehicle component level data obtained from industry participation. MacLean and Lave66 analyzed a conventional gasoline vehicle by developing an economic input-output (EIO) LCA, which uses relationships among economic sectors to expand the system boundary to the entire US economy at the expense of technical details regarding specific vehicle components. The results from both approaches agree that gasoline use/combustion during vehicle operation accounts for the vast majority of energy use and GHG emissions, and thus life cycle results are particularly sensitive to vehicle fuel economy.

Many later studies on conventional gasoline vehicles analyzed how material substitutions would impact life cycle results. Kim and Wallington67 conducted a literature review that revealed that

29 disagreement over whether the use of lightweight materials increases or decreases life cycle energy use and GHG emissions resulted from differering assumptions regarding the impact of mass reduction on fuel consumption, the energy intensity of different materials and recycling rates. More recently, Colett et al.68 found that the impact of lightweight material use on life cycle results are sensitive to indirect assumptions regarding electricity allocation.

Studies have also compared conventional gasoline vehicles to emerging alternatives. MacLean and Lave69 conducted a review of early comparisons and found that no alternative fuel/technology was clearly superior on a life cycle basis as each came with tradeoffs. This finding would later be supported by the GREET model,7 which many subsequent studies6, 8, 28 would be based on. The following subsections discuss unique aspects of more recent comparisons of alternative vehicles.

2.3.2 Plug-in Electric Vehicles

Plug-in electric vehicles, consisting of PHEVs and BEVs, require battery systems that are not required in CVs. There are questions regarding the production of these batteries and whether the environmental impacts offset the benefits of improved vehicle fuel economy (on an energy equivalent basis). Much concern arguably originates from work produced by CNW Marketing Research that claimed vehicle (production and disposal) cycle energy use was higher than fuel use on a per mile basis, and that additional energy requirements for battery production and disposal more than offset the reduction in vehicle fuel use, while comparing a Hummer H3 CV and a Toyota Prius (non-plug-in) HEV.70 Hauenstein and Schewel70 argued that these findings, which received wide media coverage, are not supported by the scientific literature. For example, the National Research Council has used the GREET model, which agrees that battery production does increase vehicle cycle energy use but finds vehicle (production, maintenance and disposal) cycle energy use is relatively minor compared to fuel (production and consumption) cycle energy use.6, 7

The emissions from electricity production for plug-in electric vehicles are another environmental concern. Samaras and Meisterling71 illustrated how the use of carbon-intensive electricity from coal when used in plug-in vehicles can result in life cycle GHG emissions comparable to those of gasoline-fuelled vehicles, whereas the use of low-carbon electricity from natural gas can reduce GHG emissions. MacPherson et al.72 and Kelly et al.73 posit that the particular source(s) of

30 electricity used by a plug-in electric vehicle depends on the geographic location and time (of day and year) charging occurs, respectively. Kennedy74 estimated the GHG emissions for gasoline and plug-in electric vehicles were similar when electricity production results in 600 g 75 CO2e/kWh. Tessum et al. found that the health impacts from life cycle criteria air contaminants (CACs) from plug-in electric vehicles use were higher than those from gasoline use if the electricity was produced from coal – despite emissions largely occurring away from populated areas - or lower if it was produced from natural gas.

The fuel cycle energy use of plug-in electric vehicles is sensitive to battery size. Shiau et al.37 analyzed the trade-off between electric vehicle battery size and GHG emissions. Larger batteries can allow PHEVs to operate on potentially low carbon sources of electricity more often, instead of gasoline, which can reduce life cycle GHG emissions. Conversely, larger batteries also increase vehicle price and mass, which reduces fuel economy. Michalek et al.8 determined that smaller batteries were a more cost-effective means of reducing life cycle emissions impacts (from both GHGs and CACs). Lewis et al.26 showed that the ability to reduce battery size is a co- benefit of manufacturing electric vehicles with lightweight materials to further improve fuel economy.

Driving patterns affect the fuel economy ratings of electric vehicles and CVs differently. CV fuel economy is much higher in steady highway driving than in city driving conditions. Conversely, BEVs, PHEVs and HEVs benefit from regenerative braking, which captures some of the energy otherwise lost during deceleration, and thus leads to better fuel economy in city driving conditions. Thus, Raykin et al.76 and Karabasoglu et al.77 found that the potential emissions reductions from electric vehicles are highly dependent on driving conditions (in addition to the source of electricity for PHEVs and BEVs).

Climate is another variable that can impact the fuel economy ratings of CVs and electric vehicles differently. In cold climates, CVs utilize otherwise wasted heat from the internal combustion engine for cabin heating. Conversely, electric motors are more efficient and do not generate sufficient heat. In HEVs and PHEVs, this can result in a greater reliance on the internal combustion engine, than under ideal conditions. In BEVs, this results in energy from the battery being depleted to provide heat. Additionally, battery efficiency can suffer in cold weather. Yuksel and Michalek78 found that BEV life cycle GHG emissions can vary by up to 22% within

31 a US grid-electricity sub-region (e.g., CAMX – California Mexico Power Area) due to temperature differences.

The performance of electric vehicles relative to CVs is forecasted to change over time.2 As discussed in Section 2.1.1.1, CAFE standards require automakers to improve the fuel economy of petroleum-fuelled vehicles. Figure 2-3 provides examples of technologies that automakers can use to meet CAFE standards, which would reduce the difference between CV and electric vehicle fuel economy ratings. Nordelof et al.79 reviewed scientific literature assessing the life cycle environmental impacts of electric vehicles and found studies often neglected to provide temporal assumptions.

The literature does not clearly distinguish between the environmental merits that can be attributed to plug-in electric vehicles versus those that are the result of particular primary energy sources. This is important because despite the lack of tailpipe emissions and use of high efficiency electric motors, BEV life cycle air emissions impacts can be higher than those from petroleum fuelled vehicles if coal is used to generate electricity, or lower if other sources are used. However, non-plug-in vehicles can also use non-petroleum energy sources (e.g., natural gas) and benefit from the high efficiency of electric motors (i.e., HEVs). This is an important distinction because given the high price of BEVs, as discussed in Sections 2.1.1.2 and 2.2.2.3, non-plug-in vehicles using non-petroleum energy sources may be a more cost-effective means of reducing life cycle air emission impacts (among other environmental impacts).

2.3.3 Lignocellulosic Ethanol Vehicles

As with gasoline, the use of ethanol results in the release of air pollutants. Hill et al.80 found that 81 lignocellulosic ethanol can reduce PM2.5 emissions by displacing gasoline. However, Jacobson and Nopmongcol et al.82 evaluated a wider range of air pollutants and found that air quality can be similar or even worse in some areas than if gasoline continued to be used.

Some researchers have investigated whether the economic and land use constraints of biomass feedstock better justify its use to produce ethanol or bio-electricity. Laser et al.83 found that both alternatives are similarly capable of displacing GHG emissions. Searcy and Flynn84 found that the relative cost-effectiveness of displacing GHG emissions depends on how the different fuels are used.

Ethanol and bio-electricity can also be compared for use in light-duty vehicles, and thus a common end use. Campbell et al.23 concluded that the superior efficiency of BEVs would result in bio-electricity production being favorable in terms of energy use and GHG mitigation. However, their analysis compared vehicles with greatly differing vehicle attributes that influenced the results. These vehicles were was also the basis of a study by Clarens et al.85 that found bio-electricity to be favorable over biodiesel. Choudhary et al.22 also arrived at the same finding as Campbell et al.,23 by citing an ethanol flex fuel concept vehicle and a database of all light-duty vehicles in the US as the source of the electric vehicle fuel economy, without any mention of the vehicle or value assumed.

Laser and Lynd27 compared the use of ethanol and bio-electricity in vehicles and found neither had an clear technological advantage in terms of life cycle energy use, and thus ability to mitigate GHG emissions. However, the results are based on comparisons of vehicles with some characteristics that may be unnecessarily dissimilar (Toyota Camry sedan versus Nissan Leaf hatchback) and others that are unrealistically similar (driving range). The finding that ethanol and bio-electricity use can result in similar life cycle energy use is based on the caveat that driving range is similar, which requires a scaling BEV battery size (and thus vehicle mass and fuel consumption) far beyond what can be found in real world vehicles15 and what is targeted by automakers.86

The literature comparing the use of biomass-derived fuels in vehicles comprehensively examines fuel production processes, while relying on simplified vehicle modelling. This is notable because there is disagreement in the literature regarding whether ethanol or bio-electricity use is favorable in terms of life cycle GHG emissions and energy use. These results depend on assumptions regarding the relative fuel economy ratings of plug-in and non-plug-in vehicles, and thus a systematic examination of vehicle technologies is required to produce a fair comparison.

2.3.4 Compressed Natural Gas Vehicles

Natural gas is an increasingly available energy source in the US.2 It is also less carbon intensive than gasoline and diesel, on an energy equivalent basis.7 However, shale gas extraction techniques that have helped increase supplies may result in higher methane emissions than conventional natural gas extraction. Howarth et al.87 provided an early analysis that suggested shale gas could have GHG emissions similar to coal and petroleum, on an energy equivalent and

100-year global warming potential (GWP) basis, as a result of leakages. More recent analyses by Allen et al.88 and Laurenzi and Jersey89 using direct measurements of leakages have found that these emissions are less than the estimate of Howrath et al.87 and that life cycle GHG emissions from shale gas are significantly less than those from coal; however, Darrah et al.90 found that water contamination from well leaks can be a concern. Venkatesh et al.55 and Burnham et al.91 concluded that compressed natural gas use in light-duty vehicles, even if sourced from shale gas, can mitigate GHG emissions by displacing gasoline fuelled vehicles. A committee for the National Research Council6 goes further by analyzing air pollutant impacts, and also found that CNG use can be an improvement over petroleum use. Howarth et al.87 also examined the GHG emissions of different fuels on a 20-year GWP basis. Their life cycle GHG emissions results for shale gas under these conditions were higher than those for coal and petroleum. This is because shale gas/natural gas leakages are primarily methane, which is more efficient at trapping radiation (but resides in the atmosphere for a shorter period of time) than carbon dioxide. Although this is an aspect not analyzed by the other studies cited above, GREET7 is regularly updated based on the most recent scientific literature and does include the capability to compare emissions on a 20-year GWP basis. As shown with the GREET7 results in Figure 2-18, shale gas as a transportation fuel has similar life cycle GHG emissions than gasoline, on a per mile travelled basis, using a 20-year GWP. This suggests that the use of shale gas can reduce GHG emissions, but may not in the short term.

300

200 eq./km)

2 100

wheel wheel GHG Emissions (g

to - 0

Well 100 Year 20 Year

Gasoline CNG (comprised of shale gas)

Figure 2-18: GREET comparison of GHG emissions from model year 2020 vehicles using 100- and 20-year Intergovernmental Panel on Climate Change (IPCC) global warming potentials7

Although CNG is not widely used directly as a vehicle fuel, 29% of US electricity is generated from natural gas (second only to coal) and is thus a primary energy source for plug-in electric vehicles.2 Wang et al.92 compared the use of these two natural gas-derived fuels and found that the use of a BEV would result in the least GHG emissions and urban air pollutants, even if CNG was used in a HEV. Dai and Lastoskie93 agreed that natural gas-derived electricity use in a BEV would be favorable in terms of GHG emissions (though did not evaluate HEVs) but found that CNG would be favorable based on seven environmental impacts (e.g., human toxicity), in part, because of battery production impacts. Curran et al.28 found the life cycle energy use and GHG emissions from natural gas-derived electricity can be similar to those from CNG use, even in a CV. Curran et al.28 attributed this to the variability in natural gas electricity generation efficiency. Although Dai and Lastoskie93 also analyzed the use of different electricity generation efficiencies, their analysis was distorted by a comparison of vehicles with fuel economy estimates derived from greatly differing means and no apparent attempt to control for vehicle attributes or fuel economy test conditions. Unlike Wang et al.92 and Dai and Lastoskie93, Curran et al.28 did not examine air pollutants.

The literature comparing the use of natural gas-derived fuels in future light-duty vehicles does not analyze financial or policy considerations. A common underlying assumption is that dedicated CNG vehicle fuel economy improvement will exceed the improvement in gasoline vehicles; however, while gasoline vehicle fuel economy must improve to meet future CAFE standards, current dedicated CNG vehicles already exceed future requirements because of non- petroleum fuel incentives. This is important because dedicated non-petroleum fuel vehicles, which also include BEVs, can avoid the use of potentially costly fuel efficiency technologies.

2.3.5 Life Cycle Costing

Life cycle costing can be used to supplement life cycle assessments. Vehicle purchase price and ongoing fuel costs are particularly important factors to consider when evaluating transportation technologies. However, the literature has typically found vehicle price to the largest contributor to life cycle costs. Thus, the discussion here is focused on the different approaches used in the literature to estimate vehicle price.

Data can be a limitation when assessing the costs of alternative vehicles, which may not be commercially available. Thus, Granovskii et al.94 used disparate sources to estimate the life cycle

35 costs of a CV, HEV, FCV and BEV. This includes the citing a (now defunct) magazine for a $100,000 estimate for the price of a hypothetical, future FCV. While the CV, HEV and FCV fuel costs were based on the fuel economy of small cars, the BEV specifications used were based on an SUV. Granovskii et al.94 is not alone in relying on unofficial, non-academic data sources. Lee et al.95 estimated the cost-effectiveness of a BEV at reducing life cycle GHG emissions by citing a blog post, which quoted an unnamed source for the approximate price difference between a BEV and CV powertrain.

Kromer and Heywood41 systematically compared the life cycle costs of CV, HEV, PHEV, FCV and BEV powertrains. Technical differences among the different alternatives were first identified. A literature review was then conducted to estimate the costs of each of the component level differences. This incremental approach avoided conflating other vehicle design factors (e.g., car versus SUV).

Argonne National Laboratory42 also systematically estimated the life cycle costs of vehicles with CV, HEV, PHEV, FCV and BEV powertrains. Total costs were determined, rather than focusing only on differences among powertrains. Forecasted component level costs were based on a set of US Department of Energy development goals.

The National Petroleum Council3 utilized a different approach to forecast the life cycle costs of vehicles with CV, HEV, PHEV, FCV and BEV powertrains. The research is based on model year 2008 vehicle characteristics and predicted changes over time. The focus of the analysis was on the trade-off between vehicle price and fuel economy as fuel efficiency technologies are added to vehicles with model year 2008 characteristics. A wide range of forecasted technologies were aggregated together to model continuous relationships (as opposed to only analyzing discrete data points as in the above studies) to determine the vehicle price and fuel economy that resulted in the minimum life cycle cost (once fuel costs are added) for each powertrain type at different points in time.

This thesis utilizes systematic means of estimating vehicle price developed in the literature to analyze the questions introduced in Section 1.1. The work by Argonne National Laboratory42 is incorporated into Autonomie96 vehicle simulation software and its used is discussed below in Section 3.4. This work by National Petroleum Council3 is publically available in the form the Vehicle Attribute Model3 and its use in this thesis is described in Section 3.5.

Chapter 3 Methods

This thesis conducts life cycle assessments to analyze future alternative light-duty vehicles. Life cycle inventory analyses are primarily based on the US Department of Energy’s Greenhouse Gasses, Regulated Emissions and Energy Use in Transportation Model (GREET).7 The human health impacts of criteria air contaminants are estimated with the Air Pollutant Emission Experiments and Policy (APEEP) analysis model.97 Vehicle price and performance characteristics are estimated with Autonomie96 and the Vehicle Attribute Model.3 Finally, Monte Carlo analyses are conducted using Crystal Ball software.98 An overview of these tools and how they are used in this research is provided below. Additional details of how these models are used

is discussed within the Methods sections of Chapters 4-7.

3.1 Life Cycle Assessment

Life cycle assessments21 quantify the environmental impacts of products or activities. This method involves identifying the life cycle scope of a product or process (e.g., fuel production and consumption, in addition to vehicle production, maintenance and disposal) and its functional unit (e.g., vehicle kilometer travelled).21 A life cycle inventory of inputs and outputs is then compiled for the individual life cycle stages (e.g., quantity of greenhouse gas emissions).21 The life cycle inventory results can then be weighted to estimate environmental impact (e.g., global warming potentials).21

3.1.1 Types of Life Cycle Assessment

Life cycle assessments may be classified as attributional or consequential.99 Attributional life cycle assessments quantify the average environmental impacts that can be directly attributed to a product life cycle. Consequential life cycle assessments quantify the change in environmental impacts from the use of a product or service, which requires modelling of the marginal products or services displaced based on economic relationships. Some contain aspects of both types of life cycle assessments, such as models that account for indirect land use change but are otherwise attributional life cycle assessments.

Both consequential and attributional life cycle assessments have their strengths and weaknesses. Plevin et al.100 argues that the scope of attributional life cycle assessment may be misleading to policy makers because not all relevant environmental impacts are captured. For example, assuming perfect substitution of a reference product with an alternative does not take into account scale and indirect effects. However, consequential life cycle assessments are less transparent, more complex and thus introduce additional sources of uncertainty because of the greater scope of the study, including economic assumptions (e.g., supply and demand interactions) that must be made to define the indirect relationships.99

3.1.2 Use of Life Cycle Assessment

This thesis uses aspects of both attributional and consequential life cycle assessments. It is largely based on the GREET model, which is further discussed in Section 3.2.

Chapter 4 is a life cycle inventory analysis of GHG emissions and total, biomass and fossil energy use for model year 2015 vehicles. The functional unit is 100 vehicle kilometers travelled, which is a common functional unit used in vehicle life cycle studies as it captures the key vehicle use component, fuel consumption, and is selected based on the emphasis on comparing energy use in this chapter. The system boundary includes the life cycle stages of vehicle production, fuel production and vehicle/fuel use. Also included is indirect energy use and GHG emissions from secondary processes, such as fertilizer production. It also analyzes co-product GHG emission credits for excess electricity generation from the production of lignocellulosic ethanol.

Chapter 5 is a life cycle assessment of air emissions impacts from model year 2020 vehicles. The functional unit is one vehicle lifetime, to facilitate the presentation of financial results (monetary quantification of environmental impacts) on a common net present value basis. The system boundary includes vehicle production, fuel production and vehicle/fuel use. It assumes the displacement of a reference vehicle on a one-to-one basis.

Chapter 7 is a life cycle assessment of GHG emissions for model year 2025 vehicles. The functional unit is one vehicle kilometer travelled, which is typical in the literature comparing vehicle GHG emissions.28, 91, 101 The system boundary includes fuel production and use. Vehicle cycle activities are excluded to facilitate more direct comparisons with light-duty vehicle GHG standards and LCFS, in addition to the uncertainties regarding which vehicle technologies will be

38 used in 2025, as a result of ambiguity within the Vehicle Attribute Model.3 It assumes the displacement of a reference vehicle on a one-to-one basis.

Chapter 6 does not use life cycle assessment. However, it does develop vehicle models that are used in the life cycle assessments in Chapter 7.

3.2 GREET Model

GREET7 estimates the life cycle energy and emissions of a variety of vehicles and fuels. It is a frequently updated (annually in recent years) model developed by Argonne National Laboratory. It is commonly used in scientific literature and was used to develop California’s Low Carbon Fuel Standards.12

3.2.1 Overview of GREET Model

GREET7 is divided into the fuel and vehicle cycle components, as shown in Figure 3-1. The fuel cycle is consists of feedstock production (e.g., oil extraction), fuel production (e.g., gasoline refining) and operation (e.g., gasoline consumption); the former two are also referred to as well- to-pump processes, the latter as the pump-to-wheel stage, and collectively they are also known as well-to-wheel processes. The vehicle cycle consists of battery production, other parts production, fluids production (e.g., motor oil), vehicle assembly and disposal. Default parameters are included within the models, which can be customized by the user.

Fuel Cycle Process Feedstock Production Vehicle Cycle Process

Battery Production Fluids Production Fuel Production

Other Parts Vehicle Assembly Vehicle Operation Vehicle Disposal Production

Figure 3-1: Simplified overview of life cycle stages modelled within GREET7

3.2.2 Use of GREET Model

The research in this thesis uses GREET7 in both direct and indirect means. GREET7 is used directly to model energy use and emissions from particular life cycle stages (e.g., CO2 emissions from E85 production). GREET7 is also used indirectly as a source of particular assumptions (e.g., relative fuel economy of dedicated E85 vehicles compared to gasoline vehicles) to create custom spreadsheet models for Chapters 4, 5 and 7.

GREET7 is used in Chapter 4 to help compare the life cycle energy use and GHG emissions of vehicles with different powertrains, but otherwise comparable characteristics. GREET7 is used directly to model vehicle cycle and feedstock production energy use and GHG emissions for model year 2015 vehicles. GREET7 is also cited for fuel cycle energy use and GHG emissions assumptions for the creation of a custom spreadsheet model.

GREET7 is used in Chapter 5 to help compare the life cycle air emissions of vehicles with different powertrains, but otherwise comparable characteristics. GREET7 is used directly to model both fuel and vehicle cycle GHG emissions of model year 2020 vehicles. GREET7 is also cited for probability distribution factors for the creation of a custom Monte Carlo analysis spreadsheet model, further discussed in Section 3.6.

GREET7 is used in Chapter 7 to help compare the fuel cycle GHG emissions of vehicles using different fuels and with different fuel economy performances, but otherwise similar characteristics. GREET7 is used directly to model fuel cycle GHG emissions of model year 2025 vehicles. GREET7 is also cited for probability distribution factors for the creation of a custom Monte Carlo analysis spreadsheet model, further discussed in Section 3.6.

3.3 Air Pollution Emission Experiments and Policy Analysis Model

The APEEP97 model is an integrated air emissions assessment model, which estimates the monetary cost of damages from exposure to criteria air contaminant emissions in the US. The model was developed by environmental economist Dr. Nicholas Muller, who helped the National Research Council6 to analyze life cycle air emissions impacts of alternative vehicles. The National Research Council6 noted that other models were considered, but that the use of

APEEP97 was “clearly appropriate for the task” and “had received sufficient prior use and performance evaluation.”

3.3.1 Overview of Air Pollution Emission Experiments and Policy Analysis Model

A simplified overview of the components within APEEP97 is shown in Figure 3-2. APEEP97 analyzes the impacts of PM2.5, PM10, NOx, SOx, and VOC emissions. Users can input the quantity of each emission type released in each US county, from ground sources and from stacks. The default is based on the release of 1 ton of each emissions at each location and elevation. These emissions are added to the inventory of background emissions levels across the US. The movement and interactions of these emissions are estimated with an internal dispersion model. Finally, dose-response relationships are used to estimate the effect of the released emissions on economic costs of mortality, morbidity, agricultural crop damage, building material degradation and visibility. Muller and Mendohlson102 discuss additional details regarding APEEP.

Emissions release of each User Input criteria air pollutant, Human Health Dose- geographic location and Response Model elevation Model Process

Emissions Dispersion Model

Background Emissions Non-Human Dose- Levels Response Model

Figure 3-2: Simplified overview of components modelled within Air Pollution Emission Experiments and Policy analysis model97

3.3.2 Use of Air Pollution Emission Experiments and Policy Analysis Model

APEEP97 is used in Chapter 5 to help compare the life cycle air emissions health impacts of vehicles with different powertrains, but otherwise comparable characteristics. The marginal

emissions impacts are calculated from a release of 1 ton of PM2.5, NOx, SOx, and VOC emissions, which is consistent with work from the National Research Council,6 Michalek et al.8 and Mashayekh et al.103. These marginal results were scaled linearly to facilitate the creation of a custom Monte Carlo analysis spreadsheet model for analyzing uncertainty (further discussed in Section 3.6). The linear scaling of emissions impacts is valid for relatively small changes in air emissions (consistent with a Taylor Series expansion). Tessem et al.75 analyzed the air emissions impacts resulting from 10% market penetration of alternative light-duty vehicles by 2020, and found that the air quality impacts scaled in an approximately linearly with changes in the size of the functional unit. Therefore, the assumption of linear air emissions impacts is supported both mathematically and by literature results, and can be expected to be a reasonable approximation for near-term changes in the light-duty vehicle market.

3.4 Autonomie

Autonomie96 is a vehicle modelling and simulation software package. It is the successor to the Powertrain System and Analysis Toolkit (PSAT), which was developed by Argonne National Laboratory with input from General Motors, Ford, and Daimler Chrysler.104 Autonomie96 was developed by Argonne National Laboratory in partnership with General Motors.96

3.4.1 Overview of Autonomie

Vehicles are modelled within Autonomie96 at the component level, as shown in Figure 3-3. For example, a conventional vehicle (CV) model includes a specific glider (vehicle without powertrain), wheels, internal combustion engine and transmission, while a battery electric vehicle (BEV) model includes a specific glider, wheels, electric motor and battery. There are component templates within Autonomie, many of which are based on actual production vehicle (e.g., seventh generation Honda Accord sedan and first generation Chevy Equinox SUV) components, and their specifications are customizable. For example, glider specifications include mass, aerodynamic drag coefficient, frontal area and manufacturing cost. Complete vehicle model templates are also included within Autonomie.96 Argonne National Laboratory105 provides additional detail on modelling the capabilities of Autonomie.96

Conventional Vehicle Battery Electric Vehicle

Glider Glider (Vehicle without Powertrain) (Vehicle without Powertrain)

Powertrain Powertrain

Internal Plug-in Electric Combustion Transmission Wheels Wheels Battery Motor Engine

Figure 3-3: Simplified overview of Autonomie conventional and battery electric vehicle model components96

3.4.2 Use of Autonomie

Manufacturing cost estimates for individual components are included within Autonomie.96 Autonomie96 provides manufacturing costs that correspond to the level of risk in achieving that cost (e.g., lower cost estimates are associated with higher risk of not achieving them); the high risk case is “aligned with aggressive technology advancement based on the U.S. DOE [Department of Energy] Vehicle Technologies program”, while the low risk case is “aligned with original-equipment-manufacturer improvements based on regulations”.105 Cost estimates for each model year are included because the cost of producing a particular component (when assuming all is else is equal) is reduced in subsequent model years.

Vehicle performance tests can be simulated within Autonomie.96 Standardized test parameters are included for different performance metrics, such as the US Environmental Protection Agency’s fuel economy rating and 0-96 km/h acceleration time. For vehicles to be designed to particular performance specifications, vehicle component modifications and performance tests can be iteratively conducted.

The research in this thesis uses vehicle template models available within Autonomie.96 These templates are modified according to the objectives of Chapters 4, 6 and 7. Modifications include

43 the scaling of component sizing (e.g., internal combustion engine power rating) and substitution of components (e.g., glider type).

Autonomie96 is used in Chapter 4 to help compare the life cycle energy use of vehicles with different powertrains, but otherwise comparable characteristics. The CV, HEV, PHEV and BEV templates are modified. Aerodynamic drag, rolling resistance and component mass are adjusted to represent the use of lightweight materials in leading edge, model year 2015 midsize sedans.42 The powertrain components are scaled to maintain consistent 0-96 km/h acceleration performance. US Environmental Protection Agency laboratory test fuel economy performance is simulated for each vehicle.

Autonomie96 is used in Chapter 6 to help compare the different vehicle design options that can be used to improve fuel economy. The CV template is modified with Chevy Equinox-like components to produce a reference vehicle. The reference small crossover SUV body is replaced with a smaller vehicle glider to examine the trade-off between vehicle size and fuel economy. The reference vehicle engine power rating is modified to examine the trade-off between vehicle 0-96 km/h acceleration time and fuel economy. The reference vehicle CV powertrain is replaced with BEV powertrain with different battery sizes to examine the trade-off between vehicle driving range and fuel economy. The reference vehicle is modified using the Vehicle Attribute Model,3 as discussed in Section 3.5, to examine the trade-off between vehicle price and fuel economy.

Autonomie96 is used in Chapter 7 to help compare alternative fuel vehicles models that can be used to meet or exceed model year 2025 CAFE standards.9 The reference Chevy Equinox-like vehicle model developed in Chapter 6 is also used in Chapter 7. The reference vehicle CV powertrain is replaced with BEV powertrain with different battery sizes to examine the trade-off between vehicle driving range and fuel economy, though larger battery sizes are examined than in Chapter 6. The reference vehicle is also modified using the Vehicle Attribute Model,3 as discussed in the following subsection, to examine the use of CNG use in internal combustion engine vehicles.

3.5 Vehicle Attribute Model

The Vehicle Attribute Model3 is a spreadsheet model that estimates the price and fuel economy of future vehicles. This work was part of the National Petroleum Council’s analysis of future vehicle technologies. The model was developed by General Motors.

3.5.1 Overview of Vehicle Attribute Model

The Vehicle Attribute Model3 is based on data from the US Energy Information Administration2 and General Motors vehicle assumptions. Existing vehicle characteristics are based on average model year 2008 models in different vehicle classes. Estimates of future characteristics are based on relative efficiencies of different vehicle powertrains, technology cost reductions over time and the price of incremental fuel economy improvements. The latter is modelled as a continuous range of technologies (as opposed to discrete) based on the aggregation of individual technologies with different degrees of use and combinations of them (see Figure 3-4). Examples of the added fuel efficiency technologies are provided in Figure 2-3, but the specific price and fuel economy improvement values are from based on 2014 data,2 whereas the Vehicle Attribute Model3 cites 2008 data. The Vehicle Attribute Model3 estimates future vehicle characteristics (fuel economy and price) based on a minimization of vehicle ownership costs (vehicle price plus fuel costs based on forecasted fuel prices2).

Incremental Price of Added Fuel Efficiency Technologies

Vehicle Price Vehicle 10 Price of Small SUV with average model

(Thousand (Thousand USD) 2010 year 2008 characteristics

0 100% 120% 140% 160% 180% 200% Fuel Economy (Relative to Base Vehicle Model)

Figure 3-4: Illustrated example of the relationship between vehicle price and incremental fuel economy improvements from the Vehicle Attribute Model3

3.5.2 Use of Vehicle Attribute Model

The Vehicle Attribute Model3 is structured to evaluate the trade-off between vehicle price and fuel economy, as fuel efficiency technologies are added, to minimize the cost of the vehicle and fuel (over different time frames). In contrast, this thesis requires price estimates for vehicles with specific fuel economy characteristics. As such, this model is not used directly in this thesis but instead underlying assumptions and equations from the Vehicle Attribute Model3 are used to create custom spreadsheet models for Chapters 5, 6 and 7.

The Vehicle Attribute Model3 is used in Chapter 5 to help compare the prices of model year 2020 vehicles with different powertrains, but otherwise comparable characteristics. The price of an average model year 2008 midsize sedan is adjusted for midsize sedans with gasoline CV, CNG CV, CNG HEV and BEV powertrains with fuel economy performances from the GREET model, which is further discussed in Section 3.2. The vehicle price is then reduced to reflect lower manufacturing costs in model year 2020.

The Vehicle Attribute Model3 is used in Chapter 6 to help compare the prices of different model year 2012-2025 vehicle design options that can be used to improve fuel economy. The prices of incremental fuel economy improvements is added to Autonomie96 vehicle models to examine the trade-off between vehicle price and fuel economy. The model is also used to examine the potential for price-neutral improvements in fuel economy over time, made possible by utilizing the savings from manufacturing cost reductions to add fuel efficient technologies.

The Vehicle Attribute Model3 is used in Chapter 7 to help estimate the price of model year 2025 vehicles using different fuels and with different fuel economy performances, but otherwise similar characteristics. The price of powertrain modifications for CNG fuel use are added to Autonomie96 vehicle models. The price of incremental fuel economy improvements are added to gasoline, CNG, and electricity-fuelled vehicle models to examine the trade-off between vehicle price and fuel costs. Additional detail on all methods are included in Chapter 4 to 7.

3.6 Monte Carlo Analysis

Monte Carlo analyses are conducted in this thesis to characterize model uncertainties. This helps illustrate the significance of the results and is particularly important when analyzing variables, such as the future price of fuels or immature technologies, which are highly uncertain. Crystal Ball is used, which a spreadsheet-based predictive modelling application developed by Oracle.98 There are many examples in the literature of life cycle assessments using Crystal Ball to conduct Monte Carlo analyses, including Mullins et al.,56 Venkatesh et al.106 and Spatari and MacLean.107

3.6.1 Overview of Monte Carlo Analysis

Monte Carlo analysis is a method of quantifying uncertainties when other mathematical means are difficult or even impossible. Probability distribution factors (e.g., normal distribution) are assigned to model variables. Multiple trials are simulated based on a random sampling of values for these variables (within the specified probability distribution functions). The results from the simulations are compiled so that percentiles (e.g., 95th percentile result) can be quantified. The percentiles can then be used to estimate confidence intervals (e.g., 90% confidence interval ranges from 5th to 95th percentile results). The results are often illustrated in the form of a frequency distribution graph or histogram, an example of which is shown in Figure 3-5.

1500

1000

500 Frequency of Frequency Results

>6%

<-12%

0% to 2% 0%to 4% 2%to 6% 4%to

-2% to 0% -2%to

-8% to -6% -8%to -4% -6%to -2% -4%to

-10% to -8% -10%to -12% to -10% -12%to Result Catagories

Figure 3-5: Example of frequency distribution graph produced from a Monte Carlo analysis

3.6.2 Use of Monte Carlo Analysis

This thesis conducts Monte Carlo analyses on life cycle assessment and life cycle costing results. Incremental results (i.e., differences between two results) are analyzed to capture correlations. For example, the uncertainty in natural gas price could result in the fuel costs of two compressed natural gas (CNG) vehicles to have overlapping confidence intervals. However, if the vehicles do not have the same fuel economy, an incremental analysis would reveal that the more fuel efficient vehicle would have lower fuel costs, regardless of natural gas price.

A Monte Carlo analysis is conducted in Chapter 5 to estimate the life cycle incremental ownership costs and air emission benefits of a set of model year 2020 vehicles. A gasoline CV is compared with a CNG CV to analyze the impact of fuel switching. A CNG CV is compared with a CNG hybrid electric vehicle (HEV) to analyze the impact of improving the efficiency of CNG use. A CNG HEV is compared with a BEV using natural gas-derived electricity to analyze the impact of shifting emissions from a vehicle tailpipe to a power plant. Variables analyzed include the fuel price, the geographic location in which emissions occur, and the economic costs of climate change from GHG emissions.

A Monte Carlo analysis is conducted in Chapter 7 to estimate the life cycle incremental ownership costs and GHG emissions of a set of model year 2025 vehicles. A set of vehicles fuelled by CNG or electricity are each compared to a reference gasoline vehicle that meets CAFE standards, to analyze the impact of using non-petroleum fuels. Variables analyzed include fuel price, vehicle price and natural gas power plant efficiency.

Monte Carlo analyses are not conducted in Chapters 4 or 6. Chapter 4 analyzes the use of lignocellulosic biomass-derived fuels produced with leading edge technologies, for which there is a relative lack of data to produce probability distribution functions. Chapter 6 has a relatively narrow scope with relatively few variables to quantify with probability distribution functions, but the work is used contribute to the Monte Carlo analysis in Chapter 7. In lieu of Monte Carlo analyses, sensitivity analyses are conducted in both Chapter 4 and 6.

Chapter 4 Life Cycle Assessment of Bioenergy Use in Light-Duty Vehicles

*Adapted with permission from Luk, J., Pourbafrani, M., Saville, B., MacLean, H. Ethanol or Bio-electricity? Life cycle assessment of bioenergy use in light-duty-vehicles, Environmental Science & Technology, 2013, 47 (18) 10676-10684. http://pubs.acs.org/articlesonrequest/AOR-JzIibBUXwnhzN6Exqysg. Copyright 2015 American Chemical Society.

The automotive industry has developed alternative powertrains to reduce petroleum use in light- duty vehicles. Conventional vehicles (CV) with an internal combustion engine (hereafter referred to as 'engine') have been modified to operate with alternative fuels. Hybrid electric vehicles (HEV) primarily utilize an engine for propulsion, supplemented with an electric motor that operates on electricity generated on-board and stored in a battery. Similarly, plug-in hybrid electric vehicles (PHEV) use both an engine and electric motor, but the battery can also be charged by an external electricity source. Battery electric vehicles (BEV) operate exclusively on a battery charged by an external electricity source.

Bioenergy can be utilized for these alternative vehicle powertrains. Ethanol produced from biomass is, for the most part, compatible with the existing fuelling infrastructure, and various ethanol mandates or Renewable Fuels Standards11 have been enacted in several jurisdictions. Biomass can also be used to produce a dispatchable source of electricity (bio-electricity), and can deliver a stable, on-demand supply of electricity, unlike renewable electricity produced from intermittent sources. While bio-electricity may not be explicitly generated for use in light-duty vehicles, through Renewable Portfolio Standards108 and electric vehicle incentives, bio-electricity may be indirectly incentivized for use in vehicles.

Although bioenergy may be able to reduce petroleum use, it may not alleviate other concerns. Competition for feedstock and cropland may limit development.48 Lignocellulosic biomass, such as crop and forest residues, and fast-growing woody biomass, can avoid some of these concerns. Nonetheless, these multiple demands dictate a rational utilization of biomass resources, to reduce unintended negative environmental, social and economic impacts.

Recent life cycle-based studies have examined the use of liquid biofuels and bio-electricity in light-duty vehicles.23, 85, 109 The studies focused on GHG emissions, total energy use and/or land use required to produce the energy. The studies that compared the environmental performance of

49 the liquid biofuel vs. bio-electricity pathways both concluded that the bio-electricity pathways were more effective in reducing GHG emissions, energy and land use. 23, 78

Campbell et al.23 studied the use of corn and switchgrass to produce ethanol and bio-electricity for use in light-duty vehicles. Ethanol production was largely based on a meta-analysis by Farrell et al.46 The study focused on deployed technologies, considering vehicles from Model Years 2000-2003, and in some cases, demonstration vehicles. In all cases comparing BEV and gasoline vehicles, the BEV were less powerful than the gasoline vehicles (which were assumed to operate on 100% ethanol (E100) on an energy equivalent basis).

Clarens et al.78 analyzed the use of algae-derived bioenergy in light-duty vehicles. Although the study compared biodiesel to bio-electricity, the study utilized fuel consumption data for gasoline- powered vehicles and BEV from Campbell et al.23 For the former, data for diesel vehicles, would have been more appropriate (see Results and Discussion for further detail). Campbell et al.23 and Clarens et al.78 compared alternative liquid biofuel and bio-electricity scenarios. In contrast, Pacca and Moreira100 evaluated a single scenario where ethanol and bio-electricity are co- products in a facility that uses sugarcane and sugarcane bagasse to produce ethanol and bio- electricity for light-duty vehicles. The resulting liquid fuel and electricity are then co-consumed in a PHEV. Pacca and Moreira100 also cited Campbell et al. 23 for certain vehicle fuel consumption data.

While these studies have made significant contributions to the literature, various aspects can be refined and updated. The bio-electricity and ethanol production models in these studies can be updated based upon recent technology improvements, with a transparent discussion of underlying process yields and assumptions. In addition, Campbell et al. 23 and Clarens et al. 78 compared bio-electricity and biofuel pathways using different classes of vehicles; however, in some classes comparisons included vehicles that differed based on passenger capacity, and in all classes, differed on performance metrics, which influenced fuel consumption. As noted by Lave et al.,101 life cycle comparisons of fuels/powertrains would ideally be based upon vehicles with similar performance and operational characteristics. Thus, the conclusions of the studies may have been impacted by the vehicles chosen for analysis. Additionally, they may have been impacted by the fact that none of the studies accounted for realistic environmental impacts from

50 vehicle production and end-of-life processes, due to exclusion78 of this life cycle stage or simplified assumptions.23

Our study aims to evaluate the life cycle energy use and GHG emissions of lignocellulosic ethanol and bio-electricity use in light-duty vehicles, and compare these results with reference fossil fuel/vehicle pathways, all within the United States. The work adds to the literature by comprehensively modeling the well-to-pump (fuel production) and pump-to-wheel (vehicle fuel consumption), and vehicle cycle (production/disposal) stages, while basing comparisons on similar vehicles. Life cycle results are analyzed in five main components: (1) a comparison of biomass and total energy use among the bioenergy pathways; (2) a comparison of fossil energy use and GHG emissions among both bioenergy and reference pathways; (3) a comparison of total energy use and net GHG emissions results with those in literature, while identifying sources of differences; (4) a scenario analysis of the impact of regional characteristics on results; (5) a comparison of petroleum use among all pathways. Finally, future developments and policy implications of these findings are discussed.

4.1 Methods

4.1.1 Research Scope

The reference and bioenergy pathways developed in our study are specified in Table 4-1. Details about pathway specifications and assumptions are discussed in subsequent sections and in the Appendix A. The year 2015 is used as a near-term timeframe, allowing for implementation of currently feasible technologies. Values of parameters applicable to the United States are utilized and presented. The study includes a scenario analysis that examines the impact of variations in key input parameters upon the results.

Life cycle energy use and GHG emissions are examined for the pathways. Lignocellulosic biomass (referred to hereafter as ‘biomass’), petroleum, fossil (which includes petroleum) and total energy use are quantified on a higher heating value (HHV) basis (see Appendix A for details of the fuels included in each category). The cumulative impact of CO2, CH4 and N2O 7 emissions based on 100-year global warming potential CO2 equivalents (CO2eq.) is reported. Results are presented based on a functional unit of 100 vehicle kilometers travelled (VKT).

Table 4-1: Reference and bioenergy pathways

Reference Pathways Bioenergy Pathways Grid-e/ Bio-e/ Pathway Name Gasoline Gasoline Gasoline E85 E85 E85 Bio-e CV HEV PHEV Grid-e BEV CV HEV PHEV BEV Conventional gasoline 85% ethanol by nominal volume Liquid Fuel Production n/a n/a (2015 U.S. average) (313 L/dry t) Grid-electricity Bio-electricity Electric “Fuel” Production n/a n/a (2015 U.S. average) (27% HHV)

Powertrain Name CV HEV PHEV BEV CV HEV PHEV BEV Liquid Fuel Consumption 9.0 6.2 7.3 n/a 11.2 7.6 9.0 n/a (L/100 VKT) Electric “Fuel” Consumption n/a n/a 23.0 23.1 n/a n/a 23.0 23.1 (kWh/100 VKT) Notes: Grid-e = grid-electricity, Bio-e = bio-electricity, CV = conventional vehicle, HEV = hybrid electric vehicle, PHEV = plug-in hybrid electric vehicle, BEV = battery electric vehicle, VKT = vehicle kilometers traveled, HHV = higher heating value, n/a = not applicable to pathway Additional detail on these processes can be found in the Appendix A.

4.1.2 Reference Fuels

Well-to-pump values for U.S. average gasoline and grid-electricity are from the GREET Fuel- Cycle model (version 1 2012 rev 2).7 Additional details on gasoline and grid-electricity production, including the grid electricity mix, are in Appendix A. The sensitivity of the life cycle GHG emissions results to grid-electricity characteristics is examined.

4.1.3 Lignocellulosic Biomass-based Fuels

Hybrid poplar, a short-rotation forestry feedstock, is the lignocellulosic biomass feedstock used in each pathway. Hybrid poplar has a high yield and could be grown on marginal land.110 Based on attractive attributes such as these, hybrid poplar has been discussed in the literature as an energy crop,111 and analyzed in life cycle studies of ethanol112 and bio-electricity113 production. Key characteristics of hybrid poplar are summarized in the Appendix A.

Hybrid poplar production data are based on the GREET Fuel-Cycle model7 poplar farming data. This includes the production of agricultural inputs and biomass delivery. Biogenic carbon absorption during growth and emission during combustion are included. The GREET Fuel-Cycle model7 does not have emissions due to direct or indirect land use change48, 114 (LUC) associated with tree farming (although the model does include these for some other biomass crops, e.g., switchgrass), and there is significant uncertainty in the LUC data for hybrid polar;112 therefore,

52 the current study does not include LUC. However, the potential impact of LUC is discussed qualitatively in the Results and Discussion.

Process models are developed in Aspen Plus115 for ethanol and bio-electricity production from hybrid poplar. This is in contrast to utilizing (and comparing) models developed externally, which may have dissimilar assumptions (e.g., different boundary conditions). Key performance characteristics are in Table 4-1; additional details are provided in the following sections, and block flow diagrams of processes are provided in the Appendix A.

4.1.3.1 Ethanol

The ethanol production model used in our study is based on an auto-hydrolysis pre-treatment and enzymatic hydrolysis process similar to that developed by Mascoma Canada and other organizations, and examined in prior research.112 Currently, a pilot-scale system is in operation and a commercial demonstration plant is in development.116 Lignocellulosic ethanol could also be produced by other biochemical117 and thermochemical118 methods. Biochemical processes require a pretreatment technology (e.g., auto-hydrolysis), which is ideally suited for different classes of feedstock. The effectiveness of auto-hydrolysis as a pretreatment for hybrid poplar is established in the literature.119 Thermochemical technologies are based on feedstock gasification rather than pretreatment, and could be effective for hybrid poplar.120

In the ethanol production model, the feedstock is pretreated via auto-hydrolysis in its delivered, un-dried state and without addition of chemicals. Enzymatic hydrolysis of pre-treated material converts cellulose and hemicellulose into sugars. Glucose and xylose are fermented to produce dilute ethanol, which is distilled to produce fuel-grade ethanol. The remaining unfermented material, which includes lignin, is combusted to generate process heat and electricity. Excess electricity is exported as a co-product to displace grid-electricity.

Additional processes are developed outside of Aspen Plus, and integrated within a spreadsheet model. Enzymes are assumed to be produced “off-site” and have environmental impacts obtained from Spatari and MacLean.107 Ethanol produced is denatured with gasoline and blended with additional gasoline to produce E85 (85% nominal ethanol or 80.75% of pure ethanol, by volume, dictated by cold weather starting requirements).7 Distribution data are based on ethanol produced and consumed within the U.S.7 The electricity co-product is accounted for with a system

53 expansion approach.21 Consequently, ethanol production receives a co-product credit for displacing U.S. average grid-electricity. The sensitivity of life cycle results to both ethanol yield and alternative co-product scenarios is examined (See Appendix A).

4.1.3.2 Bio-electricity

Bio-electricity production is a commercial process, commonly employing direct combustion of wood.121 The bio-electricity production model is based on a Rankine cycle system. Delivered feedstock is combusted within a biomass boiler, generating steam to drive a steam turbine electrical generator, and flue gas to dry delivered feedstock. The Aspen Plus model developed is based on process efficiencies from the National Renewable Energy Laboratory.122 Downstream electricity transmission and distribution losses are calculated within a spreadsheet model. Losses are assumed to be 8%, the same as for U.S. average grid-electricity, based upon the GREET Fuel-Cycle model.7 The sensitivity of life cycle results to bio-electricity generation efficiency is examined (See Appendix A).

4.1.4 Vehicle Models

The four vehicle powertrains modeled are commercially available. In order of increasing vehicle electrification (defined here as utilization of electricity, rather than a liquid fuel for propulsion), they are conventional vehicle (CV), hybrid-electric vehicle (HEV), plug-in hybrid electric vehicle (PHEV) and battery electric vehicle (BEV) powertrains. Gasoline and E85 fuels can be used in the first three, while grid-electricity (grid-e) and bio-electricity (bio-e) are used with the latter two. The CV lacks regenerative braking, which is utilized by the other three vehicles. Detailed specifications are in Appendix A.

Autonomie (version 1210)96 is used to simulate pump-to-wheel vehicle performance for each vehicle. Common gliders (vehicles without a powertrain) are assumed, in an effort to distinguish fuel consumption differences among the powertrains. Vehicle component specifications are based on Argonne National Laboratory42 projections for a “leading edge” mid-sized sedan. Specifications represent a “medium” degree of optimism for model year 2015 vehicles, based on literature and industry consultation. 42

Compared to the CV, there is greater uncertainty in modeling the other powertrains because of alternative hybrid configurations and battery capacity considerations. The HEV was created with

54 a series-parallel split hybrid powertrain, in which the engine is able to both power the wheels and act as an electric generator. The PHEV model utilizes a range extending series hybrid powertrain, in which the engine powers an electric generator once the battery is depleted. The PHEV and BEV battery capacities are sufficient for approximately 60 km and 250 km of driving range on a single charge, respectively. The PHEV charge depleting driving range is estimated to be sufficient for an average of 63% (i.e., fraction of vehicle kilometers travelled that do not require the use of the engine). 42 These specifications are obtained from literature that evaluated vehicle powertrains.8

Fuel consumption values of the vehicles are determined based on the U.S. EPA’s standardized 5- cycle test.123 Ethanol is assumed to be consumed in dedicated E85 vehicles, with an E85 optimized engine. Ethanol has a higher octane content than gasoline, enabling a higher engine compression ratio and improved energy efficiency.124 E85 fuels are assumed to achieve 7% greater energy efficiency than gasoline.7 Although hybrid vehicles are not currently certified for use with E85 fuels, E85 capable hybrid vehicles are under development.125 Thus, E85 hybrid vehicles are included in our study. The sensitivity of life cycle results to vehicle fuel consumption is examined and details are provided in Appendix A.

Vehicle cycle impacts include raw material extraction, vehicle manufacturing and end-of-life processes. Total vehicle and lithium ion battery mass characteristics from Autonomie96 are used within the GREET Vehicle-Cycle model (version 2 2012 rev 1).7 Energy use and GHG emissions are estimated for “conventional” materials, and described in Appendix A.

4.2 Results and Discussion

The results for biomass, petroleum, fossil and total energy use, and GHG emissions for the bioenergy and reference pathways are shown in Figure 4-1. Total life cycle results are shown, disaggregated by life cycle stage or energy use contribution, per 100 VKT. All bioenergy pathways have similar life cycle fossil energy use and GHG emissions (Figure 4-1 b and c), indicating no clear advantage for the ethanol or bio-electricity pathways based on these metrics. The E85 HEV, Bio-e/E85 PHEV and Bio-e BEV also have similar life cycle biomass and total energy use. The minor differences in these metrics among these pathways are insignificant, as illustrated by the scenario analysis. Only the E85 CV has considerably higher life cycle biomass

55 and total energy use than the other bioenergy pathways. We frame the following discussion around five key insights and provide additional details in Appendix A.

4.2.1 A hybrid vehicle using ethanol (E85 HEV) and a fully electric vehicle using bio-electricity (Bio-e BEV) have similar life cycle biomass and total energy use.

Ethanol and bio-electricity pathways can have similar biomass and total energy use (Figure 4-1a and 1b). The E85 HEV, Bio-e/E85 PHEV and Bio-e BEV have 340-390 MJ/100 VKT of biomass energy use and 400-490 MJ/100 VKT of total energy use. The results are essentially the same for these pathways when considering the precision of the estimates (additional discussion below and in Appendix A), and are ~30%-40% lower than results for the E85 CV (biomass energy use of 580 MJ/100 km and total energy use of 700 MJ/100 km). The well-to-pump and pump-to-wheel stages represent the majority of life cycle biomass and total energy use. Although the well-to-pump vs. pump-to-wheel efficiencies are substantially different between the E85 HEV and battery-powered vehicles (Bio-e/E85 PHEV and Bio-e BEV), these differences largely offset each other, leading to similar biomass energy and total energy use. The well-to-pump efficiency for electricity generation (27%) is lower than for producing ethanol (40% when including the co-product). In contrast, an electric motor has a much higher peak efficiency than an internal combustion engine (91% vs. 38%, excluding losses in power electronics, transmissions, etc). Consequently, with increased vehicle electrification (i.e., increased use of an electric motor), well-to-pump energy use increases, while pump-to-wheel energy use decreases. Energy use for the vehicle cycle (production/disposal) represents less than 20% of life cycle total energy use and is primarily comprised of fossil energy(see Appendix A for detail). The E85 CV has higher biomass and total energy use because it relies entirely on a comparatively low efficiency engine and lacks regenerative braking. However, inclusion of regenerative braking, similar to that in the PHEV and BEV, closes this gap and improves the performance of an ethanol-powered vehicle (E85 HEV) to a level comparable to its battery-powered counterparts (Bio-e/E85 PHEV and Bio-e BEV).

a) Reference Pathways Bioenergy Pathways 750

500

250 Well-to-Pump Use

0 Pump-to-Wheel Use

(MJ/100 (MJ/100 VKT)

E85

Biomass Biomass EnergyUse

E85 HEV

BEV Total Biomass Use

BEV

Bio-e

Grid-e

HEV

PHEV

Gasoline Gasoline

Bio-e/E85 PHEV

Grid-e/Gasoline

b) 750

500 Biomass

250 Nuclear, Hydro, Other

EnergyUse (MJ/100 (MJ/100 VKT)

0 Coal and Natural Gas

E85 E85

HEV Petroleum

BEV

Bio-e

Grid-e

HEV

PHEV Gasoline

Gasoline Fossil Energy Use

Bio-e/E85 PHEV

Grid-e/Gasoline

c) 75 50 Well-to-Pump Emissions

25 Pump-to-Wheel Emissions Vehicle Cycle Emissions 0

Biogenic Sequestration

E85

HEV

BEV BEV

-25 Bio-e

GHG GHG Emissions Grid-e

HEV Co-Product Credit

PHEV

(kg CO2eq/100 CO2eq/100 (kg VKT)

Gasoline Gasoline Bio-e/E85

-50 PHEV Net GHG Emissions

-75 Grid-e/Gasoline

Figure 4-1: a) Lignocellulosic biomass use, b) total energy use, and c) GHG emissions for reference and bioenergy pathways Note: The charts represent total life cycle results for the stated metrics. Values for all life cycle stages are included in Appendix A. Biomass refers to poplar lignocellulosic biomass only. Well-to-pump processes include feedstock and fuel, production and delivery. Pump-to-wheel process account for vehicle operation. Vehicle cycle comprises of vehicle production and disposal, and does not have any significant quantity of lignocellulosic biomass energy input.

4.2.2 All bioenergy pathways have similar life cycle fossil energy use and net GHG emissions, which are considerably lower than those of the reference pathways.

There are no substantial differences in fossil energy use or GHG emissions among the ethanol and bio-electricity pathways (Figure 4-1b and 1c). Fossil energy use for the bioenergy pathways is ~100 MJ/100 VKT, ~75% lower than that of the Gasoline CV pathway (430 MJ/100 VKT), and ~65% lower than calculated for the HEV, PHEV and BEV reference pathways (~320 MJ/100 VKT). Fossil energy use in the bioenergy pathways is associated primarily with three aspects of the life cycle: (i) in the vehicle cycle (production/disposal) stage, coal and natural gas are used extensively. Vehicle electrification also affects vehicle cycle energy use because battery manufacture is energy intensive, and larger, more powerful batteries require more energy to manufacture; (ii) fossil energy is used during combustion (pump-to-wheel stage) and, to a lesser extent, production (well-to-pump stage) of the gasoline contained in E85. Increasing vehicle electrification reduces gasoline-related fossil energy use due to less (or no) consumption of E85; (iii) the ethanol co-product credit reduces fossil energy use by displacing grid-electricity. However, increasing vehicle electrification reduces ethanol use, and correspondingly, the magnitude of the co-product credit.

The life cycle GHG emissions associated with the bioenergy and reference pathways are presented in Figure 4-1c. The bioenergy pathways’ net GHG emissions are ~5 kg CO2eq./100

VKT, ~85% lower than emissions from the Gasoline CV pathway (30 kg CO2eq./100 VKT) and ~75% lower than emissions from the HEV, PHEV and BEV reference pathways (~20 kg

CO2eq./100 VKT). For all pathways, the well-to-pump and pump-to-wheel stages of the life cycle are responsible for the majority of emissions, whereas the vehicle cycle stage is associated with a smaller portion of emissions. For the bioenergy pathways, well-to-pump emissions are higher for bio-electricity than for ethanol because the latter contains biogenic carbon not released until the fuel is consumed in the vehicle. Thus, pump-to-wheel emissions are higher for ethanol because electricity does not create emissions at the point of use. The emissions due to biomass use are largely offset by the CO2 absorbed during feedstock (hybrid poplar) growth (termed biogenic sequestration in Figure 4-1c). Therefore, net GHG emissions predominantly result from fossil energy use, and are considerably lower for bioenergy pathways than for reference pathways. In terms of both mass and 100-year global warming potential, CO2 is the dominant

GHG resulting from the life cycle stages of poplar production, ethanol, bio-electricity, gasoline and grid-electricity production, as well as gasoline and E85 consumption (see Appendix A for detail).

To evaluate potential GHG mitigation, ethanol displacing gasoline and bio-electricity displacing grid-electricity are stated by Lemoine et al.126 as being most relevant. Figure 4-2a presents GHG mitigation values for pathways whereby gasoline is displaced by lignocellulosic ethanol and grid-electricity is displaced by bio-electricity. The GHG mitigation values represent the difference in GHG emissions between the bioenergy and reference pathways (assuming a common vehicle powertrain) per unit of biomass input. This comparison removes the vehicle cycle (production/disposal) and pump-to-wheel impacts.

The reductions in GHG emissions and fossil energy are similar for ethanol and bio-electricity production. GHG mitigation for both alternatives is ~1 t CO2eq./dry t biomass (Figure 4-2a), while fossil energy mitigation is ~10 GJ/dry t. In terms of net GHG emissions and fossil energy use, neither the bio-electricity nor the ethanol pathways have a clear advantage.

a) 1.5

1.0 }Co-product 0.5 Credit

Gasoline Displacement

eq/dry biomass) t 2 GHG GHG Mitigation 0.0

(t (t CO Ethanol Bio-electricity

U.S. Average Grid-e (Base Case) Renewables-based Grid-e Coal-based Grid-e

b) 1.5

1.0 Base Base Case

0.5

eq/dry biomass) t 2 GHG GHG Mitigation 0.0

(t (t CO 300 325 350 375 400 Ethanol Yield (L/dry t)

Figure 4-2: GHG emissions mitigation resulting from displacing reference fuels with bioenergy alternatives: a) comparing mitigation potential of ethanol with that of bio- electricity, and b) sensitivity of mitigation potential of ethanol to ethanol yield

Note: Base Case refers to the U.S. average electricity GHG intensity (605 gCO2eq./kWh) and ethanol yield (313 L/dry t) assumptions used throughout our study. The Coal-based Grid-e and Renewables-based Grid-e are based on the Western Electricity Coordinating Council Rockies and Northwest Power Pool areas, respectively, forecasted for the year 2015 by the U.S Energy Information Administration. The GHG intensity of the Coal-based 42 and Renewables-based grids are 1030 gCO2eq./kWh and 350 gCO2eq./kWh, respectively.

4.2.3 Life cycle energy use and GHG emissions results of our study contrast with findings in literature, primarily because vehicles with similar characteristics are evaluated in the current study.

Our results suggest total energy use and GHG emissions can be similar for ethanol and bio- electricity pathways. The E85 HEV, Bio-e/E85 PHEV and Bio-e BEV have similar total energy use, while all bioenergy pathways, including the E85 CV, have similar GHG emissions. We find that the bio-electricity pathways outperform the E85 CV for total energy use, which is in line with results in literature,23 however, our overall findings differ from prior literature, which concluded that in general, bio-electricity pathways were superior across the feedstock, conversion technologies and vehicle classes examined. The different conclusions result primarily from our study analyzing comparable vehicles (i.e., size, shape and performance characteristics) to estimate pump-to-wheel performance, whereas dissimilar vehicles were used in those prior studies. Our findings are consistent with results from the GREET fuel-cycle model.7

Conclusions reported in Campbell et al.23 and Clarens et al.85 result, in part, from the studies not examining comparable vehicles. For example, Campbell et al.23 claim that the favorable outcome for the bio-electricity pathway in the small sport utility vehicle class can be attributed to the electric motor being 3.1 times more efficient than the internal combustion engine. However, the electric vehicles selected were up to 5.7 times more efficient than the gasoline vehicles they were compared with, because of other factors that impact relative energy efficiency. For example, within the “small car” category, the study compared a 2001 Ford Th!nk City BEV with a 90 127 km/h (55 mph) maximum speed, a 65-80 km (40-50 mile) range and two seats, to a highway- capable, gasoline Suzuki Swift CV with four seats. The study’s “midsize car” comparison included a 62 kW (83 hp) Nissan Altra BEV128 and a more powerful 112 kW (150 hp) Nissan Altima CV.129 The hypothetical HEVs examined by Campbell et al.23 are based on the CVs, and thus are also implicitly larger and/or more powerful than the BEV. Campbell et al.23 and Clarens et al.85 assumed gasoline vehicles could operate on E100 or biodiesel, respectively, on an energy equivalent basis. Although a spark ignition gasoline vehicle can be modified to operate on ethanol blends, we note, however, that operation is limited to E85 because of starting issues, and that ethanol blends would achieve a higher combustion efficiency than gasoline on an energy basis.7 Diesel vehicles, which have compression ignition engines, would have been more appropriate for examining biodiesel fuels. These vehicles are designed to take advantage of the

61 diesel fuels’ properties, and compression ignition engines are more efficient than the spark ignition engines used in conventional gasoline vehicles.42

To demonstrate the impact of assumptions regarding vehicle characteristics, we substituted the HEV and BEV “midsize car” fuel consumption data from Campbell et al.23 into our pump-to- wheel models. With this substitution, the E85 HEV pathway would now have 75% higher total energy use than the Bio-e BEV. In contrast, substituting GREET fuel-cycle model7 fuel consumption data into our models does not impact overall conclusions, with the E85 HEV continuing to have total energy use within ~20% of that of the Bio-e BEV (see Scenario Analysis in Appendix A for details). This comparison illustrates the importance of analyzing comparable vehicles in life cycle studies; we have chosen comparable vehicles, and found no clear advantage for ethanol versus bio-electricity, whereas Campbell et al.23 and Clarens et al.85 used vehicles with dissimilar characteristics, which contributed to their conclusion regarding the superiority of bio-electricity.

The fuel production (well-to-pump) efficiencies of the bioenergy pathways are also investigated to determine whether they account for the different results in the current study versus those in the literature. Substituting the higher bio-electricity production efficiency (32% versus 27%) and higher ethanol production yield from Campbell et al.23 (382 versus 313 L/dry t) into our well-to- pump models did not impact relative results. Additionally, substituting bio-electricity and ethanol production data from the GREET fuel-cycle model13 into our models also leads to similar total energy use among the E85 HEV, Bio-e/E85 PHEV and Bio-e BEV pathways and similar net GHG emissions among all bioenergy pathways. See Scenario Analysis in Appendix A.

4.2.4 Regional characteristics may create conditions under which either ethanol or bio-electricity may be a more attractive option.

The regional electricity grid mix, the presence of industries with complementary resource requirements, and existing land use in a region may lead to either ethanol or bio-electricity being a more attractive option from energy use and GHG emissions perspectives in particular regions.

This study used U.S. average characteristics; however, regional grid-electricity characteristics can affect GHG emissions of bio-electricity pathways and those of ethanol pathways (the latter through excess electricity being produced as an ethanol co-product) (see Figure 4-2a). In regions

62 that have a GHG-intensive electricity grid (e.g., Coal-based Grid-e in Figure 4-2a), the production of bio-electricity to directly displace grid-electricity could lead to a greater GHG emissions reduction than displacing gasoline with ethanol. Conversely, in jurisdictions with Renewable Portfolio Standards,108 grid-electricity GHG reductions could be achieved through the use of water, wind, solar or biomass energy sources (e.g., Renewables-based Grid-e in Figure 4-2a). In such cases, where an array of options are available, there may be greater overall benefit in displacing gasoline with ethanol, and generating excess (co-product) electricity to supplement other renewable power sources.

Besides displacing grid-electricity, there may be other regional opportunities to consider when developing co-products and co-product credits related to lignocellulosic ethanol production. Alternatives to excess electricity production include the production of fuel pellets,112 sweeteners (e.g., xylitol)130 and process heat. Different co-product scenarios lead to different energy requirements and GHG reductions; data for additional scenarios are presented in Appendix A. A pessimistic scenario corresponds to a case wherein no ethanol co-products are produced. A more favorable scenario arises from co-location of a lignocellulosic ethanol plant with an existing facility that can utilize excess heat from the ethanol plant. In this scenario the ethanol plant can make greater use of lignin-derived renewable co-products (heat and electricity),112 and increase the GHG emissions reduction possible from lignocellulosic ethanol production. In either of these scenarios, conclusions regarding similar biomass, fossil and total energy use among the E85 HEV, Bio-e/E85 PHEV and Bio-e BEV pathways in the current study would not change. In contrast, the net GHG emissions from the ethanol pathways are sensitive to facility site and co- product options, and can thus be higher or lower than net GHG emissions from the bio-electricity pathways. The GHG emissions are more sensitive to co-product assumptions than is energy use because of the high net GHG intensity (energy basis) of U.S. average grid-electricity and heat generated from natural gas, as compared to the bioenergy alternatives. Regardless, the net GHG emissions for the ethanol pathways are consistently below those of reference pathways.

Direct LUC is another factor that is based on local conditions and impacts GHG emissions. Depending on the type of land converted to dedicated poplar plantations, management practices, etc., net GHG emissions or sequestration may occur.48, 114 The high degree of uncertainty in available estimates and the lack of analysis specific to poplar prevent their inclusion in the current study.112 Under conditions whereby the ethanol facility and the bio-electricity facility

63 source the same feedstock from the same land area, LUC would impact both sets of pathways equally, on a per unit biomass basis. On a per 100 VKT basis, similar net GHG emissions would still result from the E85 HEV, Bio-e/E85 PHEV and Bio-e BEV pathways, but LUC would affect the GHG reductions relative to the reference fuels (gasoline or grid electricity).

Particular regions may be interested in other metrics not discussed above. Heavily populated areas may be concerned with air pollutants, such as particulate matter and ozone precursors, due to their detrimental health impacts. The impact of these pollutants may be substantial, but are extremely sensitive to local conditions, population and exposure aspects. These issues are beyond the scope of this study. Jurisdictions that import petroleum may favor pathways that displace petroleum for reasons of energy security. Petroleum use is further discussed in the following section.

4.2.5 Ethanol displacement of gasoline or vehicle electrification can reduce petroleum use, while bio-electricity may displace non-petroleum energy sources.

Vehicle gasoline use, including the gasoline portion of E85, is the dominant contributor to petroleum use in the bioenergy pathways. Most energy use within the Gasoline CV pathway is petroleum based, while in the Grid-e and Bio-e BEV pathways almost no petroleum is used. Thus, a greater reduction in petroleum use can be achieved via vehicle electrification than by fuel switching to ethanol.

In line with our analysis of GHG emission mitigation and fossil energy use mitigation in Figure 4-2, and analysis by Lemoine et al.,126 we assume that mitigation of petroleum use through the use of bio-electricity is based on the displacement of grid-electricity. Bio-electricity may not mitigate petroleum use because very little U.S. grid-electricity is generated from petroleum products.127 Therefore, while bio-electricity production is able to displace coal and natural gas, it is expected to have a lesser impact on petroleum consumption, as compared to ethanol production. However, over time, plug-in electric vehicles and charging infrastructure could reduce petroleum consumption by displacing conventional gasoline fuelled vehicles.

4.2.6 Future Developments

Although the total life cycle energy use and GHG emissions can be similar among bioenergy pathways, there are distinct differences at each life cycle stage. There is limited opportunity to improve well-to-pump impacts of ethanol production, because an increase in ethanol yield leads to a corresponding reduction in co-product credits (Figure 4-2-b). This trade-off is more significant for higher value co-products (as illustrated by greater sensitivity of co-product credits to ethanol yields if Coal-based Grid-e is displaced, as opposed to Renewables-based Grid-e in Figure 4-2b). Conversely, there are greater opportunities to improve the well-to-pump efficiencies of electricity pathways, such as with gasification technologies. In a scenario analysis discussed in Appendix A, future bioenergy production technologies reduce total energy use in the Bio-e BEV by 40%, while the E85 HEV is reduced by only 10%.

There are greater opportunities to improve the pump-to-wheel efficiencies of lignocellulosic ethanol and gasoline pathways, because electric powertrain efficiency is already high. However, all pathways can benefit from powertrain “agnostic” pump-to-wheel development (e.g., aerodynamic drag and rolling resistance), which also has the compounded benefit of reducing the engine and battery mass. This benefit is expected to be particularly important for electric vehicle pathways, because of the high battery mass required to meet distance/range objectives. For all pathways, the reduction of pump-to-wheel energy use also has the compounded benefit of reducing well-to-pump energy use, on a per VKT basis. In a scenario analysis discussed in Appendix A, future vehicle technologies reduce life cycle total energy use in both the Bio-e BEV and E85 HEV by 10%. Vehicle cycle (production/disposal) energy use and GHG emissions are relatively minor for all pathways.

4.2.7 Policy Considerations

Several existing policies are aimed at improved environmental performance of bioenergy and transportation technologies. Insights from our study may inform refinements of these, and development of upcoming policies. Cohesive energy policy should recognize the potentially complementary nature of ethanol and bio-electricity production, the similar life cycle impacts of hybrids and fully electric vehicles, and sensitivity of environmental performance to unique regional characteristics.

The production of ethanol or bio-electricity independently can be less efficient than if they are co-produced. Optimization of the yields of both products, from financial and environmental perspectives, can improve overall efficiency. This co-production is currently not supported by U.S. Renewable Fuel Standards11 based on liquid biofuel volumes. Fuel producers should be encouraged to take advantage of potential co-product opportunities, as is the case with Low Carbon Fuel Standards47 that take into account co-product credits.

Hybrid vehicle (both HEV and PHEV) pathways achieve many of the life cycle energy use and GHG emissions benefits of fully electric BEVs. The emphasis on miles per gallon gasoline equivalent energy use on U.S. vehicle labels123 and electric vehicle tax credits based on battery size,16 both focus on pump-to-wheel, rather than life cycle, performance. Vehicle labels and incentives based on life cycle impacts would empower consumers to make more informed decisions about vehicle environmental performance.

The method employed in our study maintains constant vehicle size and performance characteristics, unlike some previous studies. This is an effort to isolate differences in powertrain technologies and to avoid favoring particular pathways through the use of smaller vehicles and/or those having lower performance standards. ‘Comparable’ vehicles should be examined to ensure that conclusions are valid.

The reported GHG emissions and energy use benefits of bio-electricity pathways compared to ethanol pathways presented in existing literature are highly sensitive to assumptions. Regional characteristics may create conditions under which either ethanol or bio-electricity may be the preferred option; however, this analysis shows that neither has a clear advantage in terms of GHG emissions, biomass, fossil, or total energy use. This conclusion remained robust under the scenarios investigated. Policies should not focus on minor differences among alternatives, if overall, those alternatives have favorable environmental performance compared with the status quo (reference pathways).

Chapter 5 Life Cycle Air Emissions Impacts and Ownership Costs of Light- Duty Vehicles Using Natural Gas As A Primary Energy Source

*Adapted with permission from Luk, J., Saville, B., MacLean, H. Life cycle air emissions impacts and ownership costs of light-duty vehicles using natural gas as a primary energy source, Environmental Science & Technology, 2015, 49 (8) 5151-5160. http://pubs.acs.org/articlesonrequest/AOR-3bqzfeAZcSBvFxjutSWh. Copyright 2015 American Chemical Society.

The US transportation sector is dependent on petroleum fuels for the vast majority of its energy use.131 There is increasing interest in plug-in electric vehicles as a means of mitigating the use of petroleum, which is not typically used to generate electricity. However, plug-in vehicles have had limited success competing against non-plug-in vehicles2 in part because of high purchase prices.41-43

The environmental impacts of plug-in vehicles, including those resulting from life cycle air emissions, depend in large part on the source of electricity.71, 75, 76, 91, 132, 133 Replacing conventional gasoline vehicles with plug-in vehicles can result in similar greenhouse gas (GHG) emissions if the source of electricity is coal,76, 91 or lower emissions if natural gas is utilized.71, 76, 91, 93, 133 Additionally, replacing conventional gasoline vehicles with plug-in vehicles can increase or decrease detrimental health impacts from criteria air contaminant (CAC) emissions if coal or natural gas is used, respectively, to generate electricity.75 However, natural gas can also be used in non-plug-in vehicles, in the form of compressed natural gas (CNG), and reduce both GHG and CAC emissions when displacing gasoline.55, 75, 91, 93

The literature does not comprehensively distinguish between the merits of alternative energy sources and those of plug-in vehicles themselves. Not all of the benefits associated with plug-in vehicles are unique. This distinction can have important policy implications for regions that rely on non-petroleum sources of electricity, which is increasingly natural gas in much of the US.2 The natural gas available in a region could be utilized by the transportation sector in different ways: as CNG for conventional vehicles (CV) to provide the benefits of fuel switching from petroleum use; as CNG for hybrid electric vehicles (HEV) that also reduce life cycle energy use; as a source of electricity for plug-in battery electric vehicles (BEV) that can also shift CAC emissions from vehicles in urban areas to power plants in rural areas.

This study evaluates the incremental life cycle air emissions (GHG and CAC) impact benefits and life cycle ownership costs of non-plug-in (CV and HEV) and plug-in (BEV) vehicles using natural gas a common primary energy source. A gasoline CV is used as a reference pathway. US energy use and emissions from well-to-pump (fuel production), pump-to-wheel (vehicle operation) and vehicle cycle (vehicle production, maintenance and disposal) stages are comprehensively analyzed to consider temporal and geographical distributions. Ownership costs consist of both vehicle purchase and operating costs. Air emissions impacts consist of climate change and human health costs from GHG and CAC emissions, respectively. These metrics and pathways are used to investigate the merits of natural gas to produce electricity for use in plug-in

electric vehicles compared to CNG use in non-plug-in vehicles.

5.1 Methods

5.1.1 Individual Pathway Analysis

This study determines the life cycle air emissions impacts and life cycle ownership costs of a set of light-duty passenger vehicle pathways. The focus of the work is on the four pathways listed below, whose key assumptions are listed in Table 5-1. Details of the incremental benefit-cost, uncertainty and sensitivity analyses are discussed in the following subsections.  Gasoline CV: Vehicle fuelled by gasoline with a conventional powertrain  CNG CV: Vehicle fuelled by compressed natural gas, with a conventional powertrain  CNG HEV: Vehicle fuelled by compressed natural gas, with a hybrid electric powertrain  NG-e BEV: Vehicle powered by natural-gas-derived electricity, with a battery electric powertrain

Pathways are constructed in an Excel spreadsheet using publically available models and data sources. The functional unit of one Model Year 2020 vehicle lifetime facilitates analysis of potential near-term technologies. For example, there are no consumer CNG HEVs currently available, but this powertrain type has been developed for a concept vehicle.134 Results for both air emissions impacts and ownership costs are presented on a net present value (NPV) basis in 2010 USD. The use of a recent currency base year is typical135 to enable the use of historical

68 gross domestic product as an economic index,136 and has precedence in analyses of future vehicle models.8, 42

5.1.2 Life Cycle Energy Use and Emissions Inventory

Life cycle inventories of energy use and emissions are developed for each of the four pathways. 7 GREET 1 fuel-cycle and GREET 2 vehicle-cycle models are used to determine GHG (CO2, CH4 and N2O) and CAC (PM2.5, VOC, NOx and SOx) emissions. These emissions are disaggregated by life cycle stage.

GREET default fuel economy performances are based on a gasoline CV with a spark ignition internal combustion engine.7 GREET assumes that a dedicated CNG CV and CNG HEV can achieve fuel economy ratings that are 5% and 40% higher on an energy equivalent basis, respectively, than the reference vehicle.7 Dedicated CNG engines (as opposed to bi-fuel engines more common in Europe and in aftermarket retrofits) can have higher compression ratios than gasoline engines and thus higher thermal efficiencies; however, CNG is stored in heavy fuel tanks that can offset potential fuel economy improvements.137 This study assumes that GREET CNG fuel economy performances above can be achieved, but design factors may result in lower fuel economy than equivalent gasoline vehicles, and are captured in the Low Fuel Economy CNG Vehicle Scenario in Appendix B.

GREET vehicle operation default emission factors are based on gasoline CV results from the MOVES model.7 The MOVES model was developed by the EPA to allow jurisdictions to simulate the combustion, evaporative, and tire and brake wear emissions from vehicles that are designed to meet air emissions standards.138 GREET and its documentation do not state specific details regarding particular emissions control technologies assumed for vehicles in the model, but rather, GREET includes emissions factors that reflect vehicles that meet current Tier 2 standards.138 The Model Year 2020 vehicles in this study are assumed to be produced during a transition period (2017-2025), where only a portion of vehicles produced will meet stricter Tier 3 standards (the potential impact of which is discussed in the Policy Implications section).139

GREET default alternative vehicle emissions factors are calculated relative to emissions factors for a gasoline CV, as estimated by Argonne National Laboratory researchers.7 In particular, methane emissions from a CNG CV are assumed to be 10 times higher than those of the gasoline

CV because of methane slip. The absolute value of 0.07 g/mile is similar to Argonne National Laboratory test results of a 2012 Honda Civic Natural Gas vehicle on the EPA’s Urban Dynamometer Driving Schedule.28 These relatively high emissions are permitted because methane is not regulated by Tier 2 (nor future Tier 3) emissions standards, unlike CACs and non- methane hydrocarbon emissions.139 For comparison, a High Methane Emission CNG Vehicle Scenario is developed and presented in Appendix B. Note that speciation of the CAC emissions is not included in GREET and is beyond the scope of this study.

5.1.3 Air Emissions Impacts and Ownership Costs

The NPVs of climate change and human health impacts are estimated from the GHG and CAC emissions, respectively. Air emissions impacts are determined from the product of life cycle emissions quantities and specific (per unit mass) impact costs (Equation 5-1). GHG climate change impact costs are from the Interagency Working Group on Social Cost of Carbon, which was convened by the US Government.140

This study uses the APEEP (Air Pollutant Emissions Experiment and Policy analysis) model to 97 estimate the marginal health impact costs from a ton of PM2.5, NOx, SOx and VOC emissions. APEEP was developed by Muller and Mendelsohn to quantify damage caused by air pollution in the US.102 The model takes into account factors such as background emission levels and dispersion patterns when estimating the impacts from emissions occurring in different US counties and at different elevations (e.g., vehicle tailpipe emissions are ground sources). The weighted averages of impacts from emissions individual counties are used and calculated (Equation 5-2) based on the geographic distributions of each life cycle stage activity listed in Table 5-1 and further explained in Appendix B.

There is precedence for using the APEEP model to examine emissions impacts of the transportation sector. The approach we used to estimate CAC health impact costs is based on National Research Council’s (NRC) Hidden Costs of Energy study,6 which includes analysis of the transportation sector based on emissions factors from GREET and emissions impact costs from APEEP. NRC noted that other models were considered, but that the GREET and APEEP models, “were clearly appropriate for the task” of analyzing life cycle air emissions impacts of alternative (including gasoline, CNG and plug-in electric) vehicles and “had received sufficient prior use and performance evaluation.”6 Michalek et al.8 also used GREET and APEEP to

70 analyze transportation sector air emissions impacts, while Mashayekh et al.103 (2011) used APEEP and MOBILE6 (the predecessor of MOVES, the model used by GREET to obtain its vehicle emissions factors).141 Tessum et al.75 used GREET emission factors combined with WRF-Chem Eulerian meteorology and a chemical transport model to analyze the air emissions impacts from alternative light-duty vehicles.

As in the NRC study6, Michalek et al.8 and Mashayekh et al.103, we use marginal emissions impacts. This assumption of linear air emissions impacts was used to facilitate our Monte Carlo analysis and is valid for relatively small changes in air emissions (consistent with a Taylor Series expansion). Tessum et al.75 analyzed the air emissions impacts resulting from 10% market penetration of alternative light-duty vehicles by 2020 (including CNG and plug-in electric with GREET emission factors), and found that the air quality impacts scaled in an approximately linear fashion with changes in the size of the functional unit. In comparison, CNG and plug-in vehicles are projected to have a combined 1% market share in the US by model year 2020.2 Therefore, our assumption of linear air emissions impacts is supported both mathematically and by literature results, and can be expected to be a reasonable approximation for near-term changes in the light-duty vehicle market.

Ownership costs include both vehicle retail purchase price and operating expenses. Vehicle purchase prices are based on the Vehicle Attribute Model,3 which was developed by General Motors. Operating costs are based on fuel prices from the Annual Energy Outlook2 and maintenance costs are from Oak Ridge National Laboratory.142 Ownership costs are further discussed in Appendix B, including major vehicle price components and maintenance expenses itemized.

Table 5-1: Key assumptions used to develop fuel cycle and vehicle models

Life Cycle Inventory Variable Assumption Justification Gasoline CV powertrain fuel economy 13 km/L (29.5 MPG) CNG CV powertrain fuel economy 14 km/L (31.0 MPG) CNG HEV powertrain fuel economy 18 km/L (41.3 MPG) Based on GREET model year 2020 passenger car7, 132 BEV powertrain fuel economy 52 km/L (118 MPG)

BEV battery size 28 kWh

Battery replacement 0% Lifetime vehicle travel 260,000 km Vehicle emissions standards Tier 2 CNG fuel tank material Carbon fibre To enable the efficiency gain over gasoline vehicles Lifetime vehicle age 12 years Michalek et al.8 Petroleum resource mix 14% oil sands/86% conventional Natural gas resource mix 42% shale gas/58% conventional 88% combined cycle GREET projections for 20207 NG-e generation technology /12% gas or steam turbine CNG compression efficiency 96% efficiency

Ownership Cost Variable Assumption Justification Gasoline CV purchase price $24,100 Model Year 2020 retail price based on Vehicle Attribute CNG CV purchase price $28,800 Model, vehicle fuel economy, CNG fuel tank material and CNG HEV purchase price $30,000 BEV battery capacity characteristics above3 BEV price (excl. battery) $23,800 BEV battery price $430/kWh Average of Vehicle Attribute Model 2020 price range3 CNG engine modification price $1400 CNG CV/HEV fuel tank price $3300/$2600 Vehicle Attribute Model 2020 carbon fibre tank3 Brent spot crude oil price $17/GJ ($98/bbl) US Gasoline price $0.75/L Henry Hub natural gas price $4.50/GJ Annual Energy Outlook reference scenario for 20202 US CNG price $14/GJ ($0.44/Lge) US NG-e price $97/MWh ($0.88/Lge) Ownership discount rate 8% Vehicle Attribute Model default value3

Air Emissions Impact Variable Assumption Justification Social discount rate 3% APEEP model97 value and NRC median value140 140 GHG impact specific cost $43/t CO2eq. NRC median value

CAC Impact Specific Cost /t PM2.5 /t NOx /t SOx /t VOC Geographic Distribution Weighting Vehicle operation $28,100 $1,300 $10,000 $2,600 Household travel 131 and population143 Oil and gas extraction $11,600 $1,300 $4,300 $1,100 Petroleum and NG extraction*143 Gasoline fuel production $23,100 $1,000 $10,100 $2,100 Petroleum refining*143 CNG fuel production $27,800 $1,200 $10,700 $2,500 Natural gas distribution*143 NG-e fuel production $8,300 $600 $3,500 $800 Natural gas electricity production25 Vehicle parts $16,700 $1000 $6,600 $1,500 Motor vehicle part manufacturing*143 Vehicle battery $19,200 $1000 $7,700 $1,700 Battery manufacturing*143 Vehicle fluids $25,100 $1,300 $8,500 $2,300 Petro. lubricating oil manufacturing*143 Vehicle assembly $17,100 $800 $5,700 $1,600 Automotive manufacturing*143 Notes: CV = conventional vehicle, HEV = hybrid electric vehicle, BEV = battery electric vehicle, CNG = compressed natural gas, NG-e = natural gas derived electricity, Lge = liters gasoline equivalent, Costs are presented in 2010 USD. *These activities are weighted according to employment.6, 8

Equation 5-1: CAC emissions human health impact equation

I County = e × iCounty

Where:

I= NPV of CAC emissions human health impacts ($)

e= CAC emissions (t)

i= NPV of specific CAC emissions human health impacts ($/t)

County = US County where emissions/activity occurs

Equation 5-2: Weighted average of CAC emissions human health impact equation

∑ I County × ACounty I = ∑ ACounty

Where:

A= Life cycle stage activity

5.1.4 Incremental Benefit-Cost Analysis

Incremental benefit-cost analysis is used to directly compare the individual pathways. Benefits are defined here as a reduction in the NPV of life cycle air emissions impacts, whereas cost refers to an increase in the NPV of life cycle ownership costs. The incremental benefits and costs between one pathway (defender) and the next (challenger) are calculated with Equations 5-3 and 5-4, respectively, as opposed to comparing each natural gas pathway with the Gasoline CV. The four pathways are analyzed in the following three incremental comparisons: 1. Fuel switching: CNG CV challenger replacing reference Gasoline CV defender 2. Energy efficiency: CNG HEV challenger replacing CNG CV defender 3. Emissions shifting: NG-e BEV challenger replacing CNG HEV defender

Equation 5-3: Incremental benefit equation

B = Idefender − Ichallenger

Where:

B = NPV of incremental benefit ($)

I = NPV of life cycle air emissions impacts ($)

Defender = defending pathway

Challenger = challenging pathway

Equation 5-4: Incremental cost equation

C = Ochallenger − Odefender

Where:

C = NPV of incremental cost ($)

O = NPV of life cycle ownership costs ($)

5.1.5 Uncertainty and Sensitivity Analyses

Uncertainty and sensitivity analyses are completed to examine the robustness of the natural gas pathway results. Crystal Ball software98 is used to perform a Monte Carlo uncertainty analysis with 10,000 trials. The Monte Carlo analysis includes not only random error but also variability, due to the nature of aggregate data available to consumers and used by policy makers.106 An example of variability is different vehicle fuel economy performance based on local climate and proportions of city versus highway driving characteristics, but fuel economy testing results available to both consumers and policy makers are uniform across the US.15 The variables, which are collectively investigated in the uncertainty analysis, are examined individually in the sensitivity analysis. The incremental benefit-cost analysis captures correlations between pathways that are a result of these variables (e.g., the $/t life cycle GHG impact is simultaneously changed for all pathways). These variables and their probability distributions are detailed in

Appendix B. For example, the probability distribution of the specific impact cost ($/t) of PM2.5 emissions during vehicle operation is a discrete distribution based on the fraction of national vehicle kilometers travelled in each county and the impact specific cost of ground source PM2.5 emissions occurring in each county.

5.2 Results and Discussion

This study examines the life cycle air emissions impact benefits and life cycle ownership costs of a range of vehicles using natural gas as a common primary energy source in comparison to a reference Gasoline CV. The merits of natural gas use and those of the alternative vehicle powertrain technologies are distinguished through an incremental benefit-cost analysis. This section is divided into three sub-sections: individual pathway results; incremental cost-benefit analysis; and policy considerations.

5.2.1 Individual Pathway Results

Life cycle inventory analysis results for select metrics are shown in Figure 5-1, disaggregated by life cycle stage, per vehicle lifetime (260,000 km). Life cycle GHG climate change impacts, CAC health impacts and ownership costs for each pathway are illustrated in Figure 5-2 on an NPV basis per vehicle lifetime. Only the base case results are compared in these figures, while the uncertainties are presented in Appendix B.

5.2.1.1 Life Cycle Energy Use and Emissions Inventory

Life cycle energy use (Figure 5-1a) is closely related to life cycle CO2 emissions (Figure 5-1b), which are mainly a result of primary energy source combustion. The Gasoline CV has both the highest base case energy use (900 GJ per vehicle lifetime) and CO2 emissions (60 t), while the CNG CV results are 10% and 20% lower, respectively. This is because the CNG CV is a more fuel efficient vehicle (on an energy equivalent basis for the base case estimate – this assumption is examined in the Low Fuel Economy CNG Vehicle Scenario in Appendix B), and CNG is also a less carbon intensive fuel. The CNG HEV energy use and CO2 emissions are both 30% lower than those of the CNG CV, because of fuel economy differences. These metrics are similar for both the CNG HEV and NG-e BEV because differences in vehicle operation and fuel production stages largely offset each other.

CAC emissions from these pathways are also primarily a consequence of energy use. However, unlike uncontrolled CO2 emissions, the relative contributions of each life cycle stage to total NOx

(Figure 5-1c), PM2.5, VOC and SOx (Figure 5-1d) emissions are dissimilar to those of energy use.

Compared to CO2 emissions, vehicle operation is a smaller contributor to the CAC emissions because gasoline and CNG both have low sulfur contents, and Tier 2 emissions standards require the use of emissions control equipment to reduce vehicle tailpipe and evaporative NOx, PM2.5 and VOC emissions. However, the vehicle operation stage is still a major contributor of VOC 144,145 emissions because of windshield washer fluid (which contains methanol) use and PM2.5 emissions, in part because of tire and brake wear. Vehicle production is the largest contributor to plug-in vehicle PM2.5 emissions and SOx emissions for all vehicles, because activity is concentrated in the US Midwest, which relies, in large part, on electricity from coal-fired power plants that emit substantial quantities of these pollutants.

Vehicle Production Fuel Production Vehicle Operation a) 1200 Mainly from primary energy source 800 (petroleum or NG) use HEV and BEV similar due to trade-off in fuel 400

(GJ) production and vehicle efficiencies 0 Gasoline CV CNG CV CNG HEV NG-e BEV TotalEnergy Use b) 90 Mainly from primary energy source combustion 60 NG has lower carbon intensity than

(t) gasoline 30

Emissions CO2 is dominant source of GHG emissions, 2 0 over CH4 and N2O (as shown in Appendix CO B), even for CNG vehicles Gasoline CV CNG CV CNG HEV NG-e BEV c) 60 Mainly from primary energy source 40 combustion Vehicle emissions control equipment

(kg) 20 reduces contribution of gasoline and CNG Emissions

x tailpipe emissions 0

NO Gasoline CV CNG CV CNG HEV NG-e BEV d) 60 40 Mainly from coal combustion (from vehicle production electricity use)

(kg) 20 Primary energy sources have relatively low Emissions

x sulfur contents SO 0 Gasoline CV CNG CV CNG HEV NG-e BEV Mainly from vehicle fuel and coal d) 6 combustion (latter from electricity use 4 during vehicle production) Vehicle emissions control equipment

2 reduces gasoline and CNG tailpipe

(kg) Emissions

emissions 2.5 0 All vehicles have operating emissions from PM Gasoline CV CNG CV CNG HEV NG-e BEV tire and brake wear d) 90 Vehicle emissions control equipment 60 reduces gasoline and CNG tailpipe and evaporative emissions

(kg) 30 All vehicles have operating emissions from windshield washer fluid use 0

VOC VOC Emissions Gasoline CV CNG CV CNG HEV NG-e BEV

Figure 5-1: Base case life cycle (a) energy use, (b) CO2, (c) NOx, (d) SOx, (d) PM2.5 and (d) VOC emissions inventory results Notes: Results are presented per 300,000 km vehicle lifetime. Vehicle disposal is a small contributor included in vehicle production. Resource extraction is a small contributor included in fuel production.

5.2.1.2 Life Cycle Air Emissions Impacts and Ownership Costs

Air emissions impacts are the product of life cycle emissions quantities and specific impact costs. The Gasoline CV has both the highest base case GHG climate change ($3000) and CAC health ($700) impacts (note that both the absolute and relative costs of these two impact categories are highly uncertain, because of their sensitivity to the variables discussed subsequently in the Policy Considerations section, among others). The natural gas pathway results relative to the Gasoline

CV are similar for both climate change impacts (Figure 5-2a) and CO2 emissions (Figure 5-1b), because CO2 is the dominant GHG emission and the contribution to climate change is the same regardless of where the emissions occur. The results for the High Methane Emission CNG Vehicle Scenario are presented in Appendix B. The results for this Scenario and the base case are equivalent, considering the magnitude of life cycle emissions and the numerous sources of uncertainty discussed in the following subsection.

CAC health impacts (Figure 5-2b) depend on exposure to NOx, SOx, PM2.5 and VOC emissions, but are dominated by upstream fuel and vehicle production processes (which is consistent with findings from Michalek et al.8 and NRC6) because of vehicle emissions control equipment required by non-plug-in vehicles to meet strict Tier 2 emissions standards. Therefore, the NG-e BEV has the lowest vehicle operation CAC health impacts (which is consistent with results from Tessum et al.75 comparing a gasoline CV, CNG CV and NG-e BEV), because of the lack of tailpipe and fuel tank evaporative emissions, which often occur in populated areas, but life cycle CAC health impacts are approximately the same ($600) for all natural gas pathways.

Speciation of the CAC emissions is not included in GREET and is beyond the scope of this study, although it is a relevant issue for future study. For example, the formaldehyde and benzene fractions of VOC emissions from CNG vehicles are higher and lower, respectively, than those from gasoline vehicles146 and these differences can potentially result in different health impacts.

The base case ownership costs (Figure 5-2c) for the three non-plug-in vehicle pathways are approximately $40,000. This similarity is because higher priced vehicles in this study have lower operating (fuel and maintenance) costs and is consistent with life cycle ownership costs of non- plug-in vehicle in Michalek et al.8. The NG-e BEV pathway has the highest cost of ownership, 30% higher than those of non-plug-in vehicle pathways despite having the lowest operating

78 expenses. This high cost is largely due to the $13,000 battery that provides a 125 km (80 mi) driving range.

a) 3

▪ Vehicle Operation: mainly from 2 combustion CO2 emissions ▪ Fuel Production: mainly from combustion CO2 emissions, some CH4 from 1 natural gas leakage

Impacts Impacts ($1000) ▪ Vehicle Production: mainly from

GHG GHG Climate Change 0 combustion CO2 emissions Gasoline CV CNG CV CNG HEV NG-e BEV

b) 0.8 ▪ Vehicle Operation: mainly from tailpipe 0.6 NOx, PM2.5 and VOC emissions, also tire/brake wear PM2.5 and fuel/windshield 0.4 washer fluid evaporative VOC ▪ Fuel Production: mainly from NOx and

CAC CAC Health SO emissions, also gasoline refining PM 0.2 x 2.5

Impacts ($1000) and VOC ▪ Vehicle Production: mainly from coal SOx 0.0 emissions Gasoline CV CNG CV CNG HEV NG-e BEV

c) 60 ▪ Fuel: based on $0.75/L gasoline, 40 $0.44/Lge CNG and $0.88/Lge NG-e prices in 2020 ▪ Maintenance: mainly from expenses

20 related to powertrain type Ownership Ownership

Costs Costs ($1000) ▪ Vehicle Purchase: highest for BEV because of plug-in battery with 125 km 0 driving range Gasoline CV CNG CV CNG HEV NG-e BEV

Figure 5-2: Base case life cycle (a) GHG climate change impacts, (b) CAC health impacts and (c) ownership costs

Notes: GHG climate change (from the quantities of CO2, CH4 and N2O emissions) and CAC health impacts per vehicle lifetime (from the quantities and geographic distributions of NOx, SOx, PM2.5 and VOC emissions) are based on results in Figure 5-1.

5.2.2 Incremental Benefit-Cost Analysis

The incremental benefit-cost planes in Figure 5-3a present Monte Carlo analysis results for fuel switching (the CNG CV replacing the Gasoline CV), energy efficiency (the CNG HEV replacing the CNG CV), and emissions shifting (the NG-e BEV replacing the CNG HEV). Incremental benefits and costs are reductions in life cycle air emissions impacts and increases in life cycle ownership costs, respectively. Tornado plots are presented in Figure 5-3b and 3c to highlight the sensitivity of the results to deviations in key variables used in the Monte Carlo analysis. All uncertainty and sensitivity analysis results discussed in the following sections refer to 90% confidence intervals. a) Fuel Switching Energy Efficiency Emissions Shifting CNG CV replacing CNG HEV replacing NG-e BEV replacing Gasoline CV CNG CV CNG HEV

50 50 50

Lose- Trade- Lose- Trade- Lose- Trade- Lose Off Lose Off Lose Off

0 0 0 -5 5 -5 0 5 -5 0 5 Trade- Win- Trade- Win- Trade- Win- Off Win Off Win Off Win

Life Cycle Incremental Life Cycle

Ownership Cost ($1000) Ownership -50 -50 -50 Life Cycle Incremental Air Emissions Impact Benefit ($1000) b) Life Cycle Incremental Air Emissions Impact Benefit ($1000) -5 0 5 -5 0 5 -5 0 5

Gasol. Prod. CAC Impact ($/t): Life Cycle GHG Impact ($/t): Lifetime VKT (km/vhcl): Vhcl Oper Impact ($/t): CNG Fuel Tank (Material): Plug-in Battery Size (kWh): NG Power Plant Efficiency(%): c) Life Cycle Incremental Ownership Cost ($1000) -50 0 50 -50 0 50 -50 0 50

Plug-in Battery Size (kWh): Plug-in Battery Price ($/kWh): Battery Replacement (%): Life Cycle VKT (km/vehicle): CNG Fuel Tank (Material): CV Fuel Economy (L/km)): Gasoline Price ($/L):

Figure 5-3: Life cycle incremental (a) benefit-cost Monte Carlo analysis, (b) benefit sensitivity analysis, and (c) cost sensitivity analysis results Note: Benefit refers to reduction in air emissions impact and cost refers to increase in ownership costs.

5.2.2.1 Fuel switching in conventional vehicles can provide life cycle air emissions impact benefits without significantly changing life cycle ownership costs

The incremental benefit-cost results for fuel switching overlay the positive x-axis in Figure 5-3a. This indicates that, compared with a Gasoline CV, a CNG CV can be expected to provide life cycle air emissions impact benefits (90% confidence interval: $0 to $4000 benefit per vehicle lifetime), but it is uncertain if there are incremental life cycle ownership costs (-$4000 to $3000). The magnitude of the air emissions impact benefit from fuel switching is sensitive to both the quantity of emissions and the specific cost assumed for emissions impacts as shown in Figure 5-3b. Despite CAC emissions having lower base case estimates of air emissions impacts than GHG emissions (Figure 5-2) the specific cost of gasoline production health impacts ($300 to

$1,300/t PM2.5, $400 to $10,100/t SOx, $1,300 to $69,700/t NOx, $700 to $39,500/t VOC) is a larger source of uncertainty than the specific cost of climate change impacts ($30 to $110/t

CO2eq.), as shown in the Figure 5-3b tornado plot for fuel switching. Gasoline production can occur in both rural or urban areas, including Los Angeles County (which includes 6% of US oil refining capacity147). Compared with other US metropolitan centers, Los Angeles has a large population exposed to major sources of CAC emissions, combined with geographic and climate conditions that exacerbate their health impacts.148

It is uncertain if there are incremental life cycle ownership costs of fuel switching from the Gasoline CV to the CNG CV (range encompasses zero, from -$4000 to $3000). Ownership costs can increase if the reduction in fuel expenses is less than the additional purchase price of the CNG fuel system. However, the magnitude of the incremental cost is insignificant relative to the sensitivity of ownership costs to the uncertainty in real world CV powertrain fuel economy (Gasoline CV:10 to 16 km/L gasoline), life cycle VKT (150,000 to 460,000 km), and gasoline price ($0.62 to $1.02/L in 2020) variables shown in Figure 5-3c.

The Monte Carlo analysis results presented here analyze both the base case CNG CV and a lower priced CNG CV that achieve, respectively, a higher and lower energy equivalent fuel economy than the Gasoline CV. The difference in price and fuel economy is calculated based on the use carbon fibre versus stainless steel for CNG fuel tanks. A Low Fuel Economy CNG Vehicle Scenario is also developed by conducting a Monte Carlo analysis that conservatively assumes that the lower price/efficiency vehicle is the only option. The results presented in Appendix B

81 show that the incremental air emissions impacts and ownership costs are essentially unchanged from those of the base case, implying that price and fuel economy (within this range) have a small impact relative to uncertainties in the other sources of emissions and cost drivers, including those discussed above.

5.2.2.2 Improving energy efficiency with hybrid electric vehicles can provide life cycle air emissions impact benefits and reduce life cycle ownership costs

Energy efficiency results are below the positive x-axis of the incremental benefit-cost plane in Figure 5-3a. This indicates that a CNG HEV can be expected to have air emissions impact benefits ($0 to $3000) over a CNG CV and lower ownership costs (-$5000 to $0), though the degree of savings is uncertain. In addition to the sources of uncertainty discussed in the previous subsection, the relative fuel economy between different powertrain technologies affect the ability for HEV fuel savings to offset the vehicle purchase price premium for an HEV over a CV. For example, an HEV utilizes regenerative braking to provide a fuel economy advantage over a CV in stop-and-go city driving conditions, but the technology has little use in steady highway driving conditions.77 Note that this study uses the fuel economy probability distribution function from GREET7, which is not disaggregated by driving conditions; however, both Raykin et al.76 and Karabasoglu et al.77 have found that changes in driving patterns affect the fuel economy of CV and HEV powertrains differently, and thus the relative performance of an HEV versus a CV.

5.2.2.3 Shifting emissions with plug-in vehicles can increase life cycle ownership costs without providing life cycle air emissions impact benefits

In the incremental cost-benefit plane, results for emissions shifting from tailpipes to power plants overlay the positive y-axis in Figure 5-3a. This indicates that ownership costs of the NG-e BEV can be expected to be higher ($1000 to $28,000) than those of the CNG HEV. The uncertainty in the degree to which ownership costs increase is not primarily due to random error, but variability in battery size and thus driving range (80-250 km). The plug-in vehicles could have ownership costs similar to those of non-plug-in vehicles, but would require substantial sacrifices to the battery size.

It is unclear if there are incremental life cycle air emissions benefits for the NG-e BEV over the CNG HEV (-$1000 to $2000). The uncertainty in BEV driving range and natural gas power plant

82 efficiency (36% to 60%) overshadows the advantage of mitigating tailpipe emissions in urban areas. Laser et al.27 determined that BEV driving range (and thus vehicle mass and fuel economy) can determine whether plug-in vehicles have higher or lower life cycle energy use than non-plug-in vehicles using the same common primary energy source (based on their analysis of bioenergy use). Curran et al.28 concluded that the production of electricity in low efficiency (gas or steam turbine) or high efficiency (combined cycle) power plants can determine whether natural gas use in a plug-in vehicle results in higher or lower life cycle GHG emissions than natural gas use in a non-plug-in vehicle.

The results presented in this study are based on comparisons of vehicles using natural gas as a common primary energy source. However, the concerns over plug-in vehicles raised here (as compared to non-plug-in vehicles) are more broadly applicable. The NRC,6 Michalek et al.8 and Tessum et al.75 all found that electricity use in a BEV can result in higher detrimental CAC emissions impacts than a Gasoline CV. Michalek et al.8 also showed that only plug-in vehicles with small battery capacities could have life cycle ownership costs comparable to non-plug-in vehicles.

Although this study uses a BEV to represent plug-in vehicles, the emissions and ownership cost findings in this section likely also apply to plug-in hybrid electric vehicles (PHEV). PHEVs can operate in a manner similar to a BEV (charge depleting mode), an HEV (charge sustaining mode) or a combination of the two (blended mode). Therefore, if the CNG HEV and NG-e BEV air emissions impacts are similar, those of a CNG/NG-e PHEV can be expected to be similar as well. The purchase cost premium of a model year 2020 PHEV over an HEV is also unlikely to be offset by fuel cost savings.3 Unlike a BEV, reducing the size of a PHEV plug-in battery may not remove ownership cost impediments, because of the additional upfront and operating expenses of the internal combustion engine.

5.2.3 Policy Considerations

5.2.3.1 Plug-in vehicles should be evaluated as a niche market product for the foreseeable future

Consumers are generally reluctant to pay more for alternative vehicles.149 While both purchase and life cycle ownership costs of BEVs can be similar to those of non-plug-in vehicles, this would come with a substantial trade-off in battery size, which limits functionality. A small

83 battery may not impede the functionality of a PHEV, but this could result in a reliance on the internal combustion engine system, which limits the ability for fuel savings to offset the purchase cost premium over an HEV (e.g., $800 charger costs, regardless of battery size). Therefore, the trade-off between ownership costs and electric driving range likely relegates plug-in vehicles to niche markets for the foreseeable future.150

Previous studies have used US average characteristics (e.g., grid-electricity mix) to evaluate vehicle air emissions impacts.6, 8 However, this assumes plug-in vehicles operate throughout the country, which does not accurately reflect their niche market. Approximately 50% of US plug-in vehicles (BEVs and PHEVs) are sold in San Francisco, Los Angeles, New York City, Seattle and Atlanta.24 These sales are disproportionately high compared to the 14% and 12% of total US drivers and vehicle travel, respectively, collectively in these five Metropolitan Statistical Areas.131 Assumptions based on the characteristics of select regions, instead of national averages, may be more representative of the air emissions impacts of plug-in vehicles.

5.2.3.2 Plug-in vehicle policies should target urban areas with poor air quality because they can provide local air emissions impact benefits even if they may not provide life cycle air emissions impact benefits

Plug-in vehicle sales forecasts2 suggest that federal policies will likely fail to achieve deployment targets.151 The sales of plug-in vehicles have largely been limited to nonattainment areas, which exceed air emissions limits established by National Ambient Air Quality Standards.152 The Clean Air Act requires the states governing nonattainment areas to develop policies that improve air quality.153 In the five aforementioned cities where plug-in vehicle sales are concentrated, state- level incentives are provided for plug-in vehicles to supplement federal tax credits.45 Michalek et al.8 found that using US average grid electricity in a BEV can lead to higher or lower life cycle air emissions impacts than gasoline use in a CV or HEV, depending on the counties where the vehicle emissions occur. While geographic location alone may not determine whether plug-in vehicles provide incremental life cycle air emissions benefits when natural gas is used as a common primary energy source (as shown in Figure 5-3b, the Vehicle Operation CAC Impact tornado plot) plug-in vehicles in these areas can still provide valuable local air quality benefits.

Targeting incentives at regions with poor air quality can limit unintended negative consequences of plug-in vehicles, which would be exacerbated if these vehicles become a mass market

84 alternative across all geographical areas. This includes an increase in upstream emissions because of fuel and vehicle production. Plug-in vehicles could also increase vehicle travel because of the rebound effect caused by substantially reduced marginal (fuel and operating) costs and reduce the need for vehicle manufactures to improve the fuel economy of non-plug in vehicles under Corporate Average Fuel Economy regulations.149 Therefore, policies should encourage the targeted adoption of plug-in vehicles in niche markets, particularly urban areas with poor air quality; because alternative fuel use in non-plug-in vehicles is likely more cost- effective at providing life cycle air emissions impact benefits.

Tier 3 vehicle emissions standards have recently been approved and will require automakers to fully comply by model year 2025 by reducing vehicle operation emissions.139 This will have little effect on life cycle air emission impacts because CAC vehicle tailpipe and evaporative emissions are a relatively minor contributor (as illustrated by the results of the Zero CAC Emission Non- Plug-in Vehicle Scenario presented in Appendix B). Nonetheless, the legislation will have important local level implications. As non-plug-in vehicle emissions are reduced, so too is the incremental benefit of using plug-in vehicles. This further emphasizes the importance of strategically using plug-in vehicles in areas with particularly poor air quality.

5.2.3.3 Climate change regulation may not be sufficient to reduce overall air emissions impacts

Improved air quality can be a co-benefit of reducing GHG emissions. This would be expected with a simple reduction in fossil fuel consumption.154 However, complexities are introduced when fuels are substituted for others and/or consumed in substantially different manners. For example, Tessum et al.75 found that the use of US average-grid or biomass-derived electricity to replace gasoline as a transportation fuel can result in lower GHG emissions, but higher CAC emissions. Conversely, emissions control equipment required to meet Tier 2 vehicle CAC emissions standards slightly reduces fuel economy, which increases GHG emissions.155

The GHG climate change and CAC health impacts do not correlate across the pathways in this study, which is consistent with the findings of Tessum et al.75 The quantity of GHG emissions is highly sensitive to changes in the fuel cycle, while the majority of CAC emissions are from vehicle production. This results in vehicle fuel economy improvements that reduce climate change impacts without decreasing life cycle health impacts (e.g., CNG CV vs CNG HEV).

Unlike GHG emissions, the impact of CAC emissions depends on the geographic location where they occur, which provides a local advantage for plug-in vehicles that is not captured when using climate change impacts (or GHG emissions) as metrics.

The relative contribution of GHG emissions to total air emissions impacts is highly uncertain. The Gasoline CV GHG climate change impacts ($2000 to $10,000) can overshadow or be similar to those of CAC health impacts ($400 to $4000). There are also other non-health impacts of CAC emissions,97 such as agricultural crop damages that have not been included in this study. Therefore, climate change impacts may not only be an incomplete measure, but also a poor proxy of environmental or social merits of alternative vehicles.

Climate change regulations and CAC emission policies should aim for synergies in reducing negative impacts. There can be trade-offs as illustrated by emissions control systems in non- plug-in vehicles; excess air (above stoichiometric air-fuel ratio) reduces GHG emissions by 156 improving fuel economy but at the expense of higher NOx emissions. Consequently, policies such as Tier 3 tailpipe CAC emissions standards139 are important to have alongside legislation designed to reduce GHG emissions149 to avoid unintentional increases in either health or climate change impacts.

5.2.3.4 Carbon pricing or internalizing costs of overall air emissions impacts may not be enough to change consumer behavior

Ownership costs are an order of magnitude greater than the costs of air emissions impacts evaluated in this study. This is a significant margin even when uncertainties discussed in the previous subsections are considered. Therefore, internalizing carbon costs or even the overall costs of air emissions impacts may have little influence on the total ownership costs of driving.

A carbon price can be effective at influencing the economics of electricity generation and encouraging coal to natural gas fuel switching in power plants.157 Increased natural gas electricity production would make the pathway comparisons in this study relevant to even more regions. This change, particularly in regions that produce vehicle components, would have the co-benefit of reducing life cycle CAC emissions.

Chapter 6 Vehicle Design Options To Meet 2025 Corporate Average Fuel Economy Standards

Corporate Average Fuel Economy (CAFE) standards aim to reduce light-duty vehicle petroleum use, GHG emissions, and fuel costs by requiring automakers to produce more fuel efficient vehicles.149 Recent amendments to the legislation for US vehicle model years 2012 to 2025 are scaled by vehicle footprint (product of wheelbase and track width) to distribute the burden across the light-duty vehicle market and encourage automakers to maintain or expand the variety of vehicles consumers can currently choose from. However, legislated fuel economy improvements can be expected to affect other vehicle attributes, such as size and acceleration performance, which change over time.33, 158

Studies of the potential impact of future CAFE standards on vehicle attributes have reached different conclusions. The Regulatory Impact Analysis conducted by the National Highway Transportation Safety Administration29 (NHTSA) concluded that vehicle prices will increase as automakers add fuel efficiency technologies to vehicles. Knittel30 concluded that a reduction in acceleration performance alone can theoretically meet fuel economy targets, but that a reduction in size is likely required. Conversely, Cheah and Heywood35 found that scenarios in which the plug-in electric vehicle market share is increased could preserve vehicle size and acceleration performance.

Knittel30 and Cheah and Heywood35 did not analyze the price of fuel efficiency technologies, whereas the NHTSA29 excluded elasticity in vehicle size and acceleration demand in response to changes in price. While these studies provide important insights, the fluctuations in historical average vehicle price, size, and acceleration suggest that each of these variables should be analyzed and compared to examine implications of CAFE standards.13 Evolving consumer interests are expressed through sales of vehicles from different size classes and with options that may prioritize low price, rapid acceleration or high fuel economy; for example, the current (model year 2014 to present) Honda Accord is available with a (comparatively) affordable 2.4 L engine, powerful 3.5 L engine or efficient 2.0 L engine within a hybrid powertrain.63 Automakers respond to consumer demands by modifying vehicles and their options each generation; for example, the current (model year 2010 to present) entry level Chevy Equinox has higher fuel

87 economy and lower price, but reduced acceleration and interior volume compared to the previous generation – a redesign that has resulted in higher annual sales.63, 159, 160 Therefore, a more complete analysis of fuel economy targets should consider potential changes in vehicle price, size, and acceleration performance.

Research on CAFE standards by Shiau et al.37 and Whitefoot and Skerlos38 combined economic and engineering modelling. Shiau et al.37 highlighted the need to balance fuel economy targets with penalties for violations, while Whitefoot and Skerlos38 advised of the potential for footprint- based targets to be a moral hazard that encourages the production of larger vehicles. Both studies arrived at their conclusions by modelling the elasticity of consumer demand for vehicle size and power, to vehicle price and fuel economy. Neither study considered technology changes nor fuel economy targets over time; therefore, the studies do not provide insight into how the stringent future year CAFE standards can be met.

The objective of this case study is to systematically compare vehicle attributes that can be modified to improve fuel economy and meet model year 2012 to 2025 CAFE standards. Vehicle design options are developed that modify only one of four attributes on a reference 2012 vehicle: a crossover sport utility vehicle (the fastest growing vehicle segment in the US) with an average light-duty vehicle footprint (fuel economy targets are scaled by footprint). This method illustrates how aggregate changes in an automaker’s fleet (which CAFE standards regulate) could manifest in a typical vehicle, but not the design limitations of individual vehicles, because the sale of vehicles that exceed fuel economy targets can facilitate the sale of vehicles that do not meet targets. Changes in vehicle price, size and acceleration, as well as driving range from the use of emerging plug-in electric vehicles, are examined. These vehicle design options incorporate estimates of technological development, based on expected component cost reductions and

introduction of fuel efficiency technologies over time.

6.1 Methods

This study systematically compares vehicle attributes that can be modified to improve fuel economy and meet model year 2012 to 2025 CAFE standards. This comparison is based on four vehicle design options, which are defined here as series of vehicle models that improve fuel economy over time by changing a particular vehicle attribute. Together, they outline some of the flexibility automakers have in redesigning vehicles to meet CAFE standards. The designations and brief descriptions of the options are as follows: 1. Vehicle price option, which solely utilizes added fuel efficiency technologies (e.g., lightweight materials and hybrid electric powertrains) 2. Vehicle acceleration option, which modifies the engine power rating and utilizes added fuel efficiency technologies 3. Vehicle size option, which modifies the vehicle body and utilizes added fuel efficiency technologies 4. Driving range option, which replaces the conventional gasoline vehicle powertrain with a battery electric vehicle powertrain and utilizes added fuel efficiency technologies

The vehicle design options meet the increasing CAFE standards, illustrated in Figure 6-1a, by modifying a Chevy Equinox-like model year 2012 reference vehicle. A crossover SUV is selected because they are the fastest growing market segment and can have typical vehicle specifications. For example, the current entry level Chevy Equinox has the same 4.5 m2 (48 ft2) footprint and 9.3 s 0-96 km/h (0-60 mph) acceleration time as the US model year 2012 light-duty vehicle average, while its 3700 L (130 ft3) interior volume and 34 mpg laboratory fuel economy rating are higher and lower, respectively.13, 63 Note that fuel economy values (in mpg) are presented here to be consistent with CAFE standards but fuel consumption values (in L/100 km) are also reported in the Appendix C.

The series of vehicle models that comprise the vehicle design options in this study are developed in two main components; base vehicle models and added fuel efficiency technologies. The latter is modelled as a continuous range (as opposed to discrete set) of technologies, based on the individual technologies presented in Figure 6-11b (among others), varying degrees of their use and different combinations of the technologies. Figure 6-1c shows the price of a base vehicle model, using the 2012 reference vehicle as an example, and the relationship between vehicle

89 price (all prices in constant 2010 US dollars (USD)) and fuel economy when increasing the utilization of added fuel efficiency technologies. Note that for simplicity Figure 6-1c shows only one of the three price scenarios analyzed in this study; these scenarios are described at the end of the Methods section.

The development of the 2012 reference vehicle and the vehicle design options is discussed below. This is followed by a description of Autonomie and the Vehicle Attribute Model, which are used to develop the base vehicle models and added fuel efficiency technologies components, respectively. Additional details on the methods, including vehicle specifications, are provided in Appendix C.

a) 60

0 Economy(MPG)

Laboratory Test LaboratoryTest Fuel 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 Model Year

1.5

Lean Burn Direct Injection 1.0 Engine Start-Stop Stoichiometric Direct Turbocharging Injection Variable Compression 0.5 Ratio Continuously Variable Regenerative Braking Improved Alternator Transmission and Launch Assist

(Thousand (Thousand USD) 2010 Aggressive Shift Logic Cylinder Deactivation Low Friction Motor Oil Low Rolling Resistance Tires Incremental Vehicle Increase Price Vehicle Incremental 0.0 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10% Incremental Fuel Economy Improvement (MPG)

c) 30

20 Incremental Price of Added Fuel 10 Efficiency Technologies

Price of Base Vehicle Model Vehicle Price Vehicle 0 (Thousand (Thousand USD) 2010 100% 120% 140% 160% 180% 200% Fuel Economy (Relative to Base Vehicle Model)

Figure 6-1: a) CAFE standards29 for different model years and a 4.5 m2 Chevy Equinox- like vehicle footprint63, b) potential incremental fuel economy improvements and vehicle price increases2 from example added fuel efficiency technologies, and c) illustration of the vehicle price model used to develop the 2012 reference vehicle Notes: Part a) Vehicle footprint is the product of wheelbase and track width. Part b) the Vehicle Attribute Model cites the Energy Information Administration2 for the technologies presented here, among others, and applies these technologies to different degrees (e.g., point estimate data is not provided for the use of lightweight materials, which consist of a broad range of potential material/component substitutions) and/or combines them (e.g., hybrid electric vehicle utilizing both regenerative braking and engine start-stop technologies). The potential incremental fuel economy improvement presented for each technology is for model year 2025 and may not be applicable in prior years (e.g., lean burn direct injection is forecasted to be commercially available starting in 2020).

6.1.1 2012 Reference Vehicle

The 2012 reference vehicle is a Chevy Equinox-like vehicle that meets model year 2012 CAFE standards63 and whose attributes are changed to develop vehicle design options that meet future year CAFE standards shown in Figure 6-1a. The base vehicle model is developed with a glider (vehicle without powertrain) based on a first generation (2005-2009) Chevy Equinox and a conventional gasoline powertrain that provides average model year 2012 0-96 km/h acceleration time of 9.3 s. Added fuel efficiency technologies, which introduce more recent technological developments, are applied to the base vehicle model (shown in Figure 6-1c) to meet the model year 2012 CAFE standard of 31.2 MPG (Figure 6-1a). Note that comparisons of future vehicles to the 2012 reference vehicle are used to examine the magnitude of potential changes over time; this study does not presume 2012 vehicles will be produced in model year 2025.

6.1.2 Vehicle Design Options

Vehicle design options are series of vehicle models (illustrated in Error! Reference source not found.) that modify the 2012 reference vehicle to meet model year 2015, 2020 and 2025 CAFE standards in a manner in which only one of four vehicle attributes changes; either price, acceleration, size or driving range. All vehicle design options require changes to both the base vehicle model and utilization of added fuel efficiency technologies. The price of the base vehicle model is a function of production costs, which change over time because the cost of producing a particular component is reduced in subsequent model years. Physical modifications to the base vehicle model are also required to analyze the different vehicle design options; a base vehicle model with a different engine power rating, body, or powertrain-type is required to model changes to vehicle acceleration, size and driving range, respectively. Added fuel efficiency technologies are then applied to each base vehicle model to meet CAFE standards (in the case of the vehicle price option) or to take advantage of base vehicle model price reductions over time and to maintain the total price of the 2012 reference vehicle (in the cases of all other vehicle design options). The particular vehicle design attributes of each vehicle design option are calculated either iteratively or via interpolation among the large set of vehicle models produced. A conceptual overview of the development of each vehicle design option is provided in Error! Reference source not found.. Although the tools in this study are able to model vehicles with multiple attribute changes, this is beyond the scope of this study. A more detailed explanation of the development of the vehicle design options is provided in Appendix C.

6.1.3 Autonomie and the Vehicle Attribute Model

This study is primarily based on two vehicle modeling tools, both of which have had industry input from (Chevy Equinox automaker) General Motors during their development. One is Autonomie vehicle simulation software,96 which estimates vehicle manufacturing costs and simulates fuel economy and acceleration performance tests based on detailed component assumptions, including aerodynamic drag coefficient and engine power rating. The software includes complete vehicle templates with components that can be manually adjusted or replaced with components from numerous real world vehicles, such as the aforementioned Chevy Equinox. Autonomie was developed by Argonne National Laboratory, in partnership with General Motors, as a successor to PSAT, which has been used to support automotive research and development by companies including General Motors, Ford, Chrysler, Hyundai and Toyota.96

The second tool is the Vehicle Attribute Model,3 which General Motors developed for the National Petroleum Council report, Advancing Technology for America’s Transportation.3 The model estimates vehicle price based on higher level vehicle characteristics, such as size class, fuel economy and model year. The Vehicle Attribute Model3 uses equations that relate incremental price to fuel economy improvements over time, applied to vehicles in different classes with average model year 2008 characteristics, by aggregating added fuel efficiency technologies. The Vehicle Attribute Model does not detail these added fuel efficiency technologies, but does highlight the use of lightweight materials, and distinguishes between hybrid and non-hybrid powertrain technologies. The examples provided in Figure 6-1b are from the Energy Information Administration2, which the Vehicle Attribute Model cites.

These two tools are used to develop the base vehicle models and added fuel efficiency technologies components. Base vehicle models are first developed in Autonomie96 based on a vehicle glider (vehicle without powertrain) and powertrain that approximately represent model year 2008 components (as noted above, model year 2008 is the basis for the Vehicle Attribute Model). The manufacturing costs and fuel economy of base vehicle models are then estimated within Autonomie96 and compiled within a spreadsheet. The prices of base vehicle models are calculated by adding a 30% markup (from the Vehicle Attribute Model3) to manufacturing costs. The incremental prices of added fuel efficiency technologies, which introduce post-2008

94 developments, are based on Vehicle Attribute Model equations that are presented in Appendix C and are illustrated in Figure 6-1c and Error! Reference source not found.. We assume the use of added fuel efficiency technologies changes base vehicle model price and fuel economy, but not interior volume or acceleration performance. This assumption is further discussed in Appendix C. Unfortunately, we are unable to verify this assumption because the Vehicle Attribute Model3 does not detail how specific added fuel efficiency technologies are aggregated within its incremental price vs. fuel economy improvement curves.

The uncertainty in prices of vehicles with known physical specifications is estimated in this study, which is consistent with the approaches used in Autonomie96 and the Vehicle Attribute Model.3 Autonomie96 provides manufacturing costs that correspond to the level of risk in achieving that cost (e.g., lower cost estimates are associated with higher risk of not achieving them); the high risk case is “aligned with aggressive technology advancement based on the U.S. DOE [Department of Energy] Vehicle Technologies program,” while the low risk case is “aligned with original-equipment-manufacturer improvements based on regulations.”105 The Vehicle Attribute Model3 provides upper and lower bound incremental prices of added fuel efficiency technologies. The average risk estimates provided in Autonomie96 and the average of the two bounds from the Vehicle Attribute Model3 are used to produce a mid-price scenario. High risk prices from Autonomie96 are combined with lower bound prices from the Vehicle Attribute Model3 to examine a low price scenario, and vice versa.

6.1.4 Comparison with the literature

As discussed in the Introduction, the NHTSA 29, Knittel 30 and Cheah and Heywood 35 have investigated the ability for changes in vehicle attributes to meet future CAFE standards. Shiau et al. 37 and Whitefoot and Skerlos 38 examined trade-offs at a point in time, as opposed to changes over a period of time. However, the methods used in the individual studies do not facilitate a systematic comparison of the range of vehicle design options in this study. Thus, a novel approach is developed for the purposes of this study. Although Autonomie 96 and the Vehicle Attribute Model 3 are each individually developed with/by the automotive industry and established in the scientific literature, results from the combined use of these models (explained above) should be evaluated by comparing with past approaches.

The NHTSA 29 assessment of CAFE standards includes an estimate of the average additional vehicle price that would result from fuel economy increasing between model years 2012-2025, while other vehicle attributes remained constant. The estimate was based on an analysis of the forecasted price of fuel efficiency technologies conducted by Volpe, which is a part of the US Department of Transportation. This result is compared with those from the vehicle price option in the Results and Discussion section.

Knittel 30 and Cheah and Heywood 35 analyzed the effect of changes to vehicle size and acceleration performance on future vehicle fuel economy. The relationships among the variables are based on a projection of future technological capabilities based on an extrapolation from historical vehicle characteristics. This method was proposed by An and DeCicco 33, who found that the product of average annual US vehicle fuel economy (in mpg), interior volume (in ft3) and the ratio of engine power rating over vehicle mass (in hp/lb, which is an indicator of acceleration performance) was approximately linear between model years 1977 and 2005, which suggests steady technological improvements over time. The work by Cheah and Heywood 35 and Knittel 30 was limited to examining 2016 and 2020 CAFE standards, respectively. Thus, we use the approach proposed by An and DeCicco 33 (further explained in Appendix C) to estimate how changes to vehicle size and acceleration performance can be used to meet 2025 CAFE standards and compare the results with those from the vehicle acceleration option and vehicle size option in the Results and Discussion section.

The NHTSA 29, Knittel 30 and Cheah and Heywood 35 did not analyze the driving range in which future battery electric vehicles would be the same price as current gasoline vehicles. Nor do their methods used facilitate this analysis. Therefore, the vehicle driving range option is a novel aspect of this study for which there are no comparison data available in the literature.

6.2 Results and Discussion

This case study systematically compares vehicle attributes that can be modified to improve fuel economy and meet model year 2012 to 2025 CAFE standards, by developing vehicle design options that modify only one of four attributes on an example reference 2012 vehicle. The results in Figure 6-2 show that the 66% increase in fuel economy targets from 2012 to 2025 could be

96 met with a 10% vehicle price increase (corresponding to a lightweight hybrid electric vehicle), a 31% increase in acceleration time (smaller engine), a 17% decrease in vehicle size (smaller body), or a 94% decrease in driving range (battery electric vehicle powertrain) relative to the 2012 reference vehicle. Some combinations of these changes would also be feasible, which could make changes in vehicle price, acceleration, size and/or driving range less perceptible to consumers. However, the development of combinations that would be expected to be attractive to consumers is beyond the scope of this study. Although there is uncertainty in future technology prices, a set of price scenarios agree that expected component cost reductions over time are insufficient to offset the costs of additional fuel efficiency technologies to meet 2025 fuel economy targets while preserving other 2012 reference vehicle attributes. These findings are discussed in the following sections, while vehicle specifications are presented in Appendix C.

6.2.1 Meeting 2025 CAFE standards will require changes to vehicle attributes beyond fuel economy

The vehicle price option is shown as the black dashed price curve in Figure 6-2a. CAFE standards are met in this pathway by increasingly utilizing added fuel efficiency technologies, such as lightweight materials and gasoline hybrid electric vehicle powertrains. Constant fuel economy price curves are included in Figure 6-2a to illustrate how vehicle price can decrease over time, assuming other vehicle attributes remain constant, due to decreasing manufacturing costs. The constant fuel economy price curves intersect the vehicle price option price curve to indicate vehicle price and fuel economy, as determined by CAFE standards, over time.

The vehicle price option price curve intersects the 53 mpg 2025 CAFE standard price curve at $22,970; this indicates a $2,070 (10%) increase in vehicle price compared to the 2012 reference vehicle may be sufficient to meet 2025 CAFE standards while maintaining other 2012 reference vehicle attributes. This price increase is consistent with the $1,870-$2,120 range estimated by the NHTSA.29 In the nearer term, prices remain similar to that of the 2012 reference vehicle because the cost of the relatively minor fuel economy improvements can be offset by component cost reductions over that same time period. The price of a vehicle meeting CAFE standards in 2015 and 2020 is $140 (1%) lower and $460 (2%) higher, respectively, than the 2012 reference vehicle. The reason vehicle prices increase more in later years is because of the accelerating rate at which CAFE standards increase over time and because the marginal cost of improving fuel economy increases as the most affordable added fuel efficiency technologies are utilized first.149

The 2025 vehicle price option model utilizes lightweight materials and a hybrid electric powertrain, but note that the $2070 price difference discussed above is compared to the $20,900 price of the 2012 reference vehicle and not compared to the $18,400 price of a hypothetical model year 2025 conventional vehicle with the same 32 MPG fuel economy rating as the 2012 reference vehicle, as shown in in Figure 6-2a.

Although there is uncertainty in future technology prices, different technology price scenarios agree that expected component cost reductions over time are insufficient to offset the costs of utilizing added fuel efficiency technologies to meet 2025 CAFE standards. The error bars in Figure 6-2a show price scenarios that capture 100% of the price estimate ranges from Autonomie and the Vehicle Attribute Model. The high price scenario (upper bound of error bars) shows that vehicle prices can remain similar to the 2012 reference vehicle until about 2015 before increasing by $3520 (16%) to meet 2025 CAFE standards. The low price scenario (lower bound of error bars) shows that vehicle prices can remain similar for a longer period of time, until about 2020, before increasing by a lesser amount, $570 (3%), to meet 2025 CAFE standards.

Whether or not automakers will produce the vehicles discussed here will depend on the willingness of consumers to pay for these vehicles. This will likely depend on factors beyond vehicle attributes. Average US car prices increased rapidly in the 1980’s and 1990’s, but real prices (based on constant currency) have since fallen.136 This trend is negatively correlated with US gasoline prices,1 which suggests consumers may adapt to high fuel prices by purchasing more affordable (e.g., smaller) vehicles, rather than paying higher prices for vehicles with advanced fuel efficiency technologies. Therefore, future gasoline prices (among other factors) may play an important role in determining if consumers are willing to pay more for vehicles that meet CAFE standards. The following sections discuss the vehicle design options for meeting CAFE standards that do not involve increasing vehicle prices.

a) 27

25 53 MPG (2025 CAFE Standard)

Vehicle Price Option 42 MPG (2020 CAFE Standard) 21 34 MPG (2015 CAFE Standard) 19

32 MPG (2012 CAFE Standard) Vehicle Vehicle Price (Thousand 2010 USD) 17 2012 2017 2022

2012 2015 2020 2025 Model Year

b) 23 c) 23 d) 23

22 22 22

2010 2010 USD) 2010 USD)

Vehicle Acceleration Option Vehicle Size Option Driving Range Option

21 21 21 Thousand Thousand 35 km 9.3 s 4000 L 600 km 9.9 s 3800 L

20 20 20 25 km

Vehicle Vehicle Price (Thousand 2010 USD) Vehicle Price ( Vehicle Price (

12.2 s 3300 L 19 19 19

2012 2017 2022 2012 2017 2022 2012 2017 2022

2025 2012 2015 2020 2025 2012 2015 2020

2012 2015 2020 2025 Model Year Model Year Model Year

Figure 6-2: New vehicle price curves representing, a) the Vehicle Price Option and vehicles with constant fuel economy, b) the Vehicle Acceleration Option and vehicles with constant 0-96 km/h acceleration times, c) the Vehicle Size Option and vehicles with constant interior volumes, and d) the Vehicle Driving Range Option and vehicles with different driving ranges. Note: Price curves are based on the mid-price scenario and error bars in a) capture the high and low price scenarios. The dashed black line shows the vehicle price option and represents the same data in each subfigure. Grey, yellow, blue and orange reference lines are shown to identify changes in vehicle design option characteristics over time. The green lines represent vehicle design options that do not require increasing vehicle prices.

6.2.2 Modifying vehicle acceleration performance or size could preserve other vehicle attributes while meeting CAFE standards

The vehicle acceleration option is shown as the horizontal green (constant) price curve in Figure 6-2b. Constant 0-96 km/h acceleration time price curves generally trend upwards over time because added fuel efficiency technologies are increasingly utilized to meet CAFE standards, as opposed to reducing engine power rating (and thus acceleration performance). These constant acceleration price curves intersect the vehicle acceleration option price curve to indicate the acceleration performance required to meet future CAFE standards without changing vehicle price. The black dashed price curve illustrates the vehicle price option (from Figure 6-2a), which represents the 9.3 s 0-96 km/h acceleration performance of the 2012 reference vehicle. This constant acceleration price curve falls below the vehicle acceleration option price curve between model years 2012 and 2017 because the cost of increasingly utilizing added fuel efficiency technologies is offset by component cost reductions over that same time period, as discussed in the previous section. Altogether, the price curves in Figure 6-2b map the trade-off between vehicle price and acceleration over time.

The vehicle acceleration option price curve intersects the 12.2 s 0-96 km/h acceleration price curve in model year 2025; this indicates a 2.9 s (31%) increase in 0-96 km/h acceleration time may be required to meet 2025 CAFE standards while maintaining the 2012 reference vehicle price. A similar increase of 2.5 s (26%) is estimated based on the simplified method proposed by An and DeCicco 33 and described in the Methods section. A 12.2 s acceleration time is on par with average US light-duty vehicle performance in 199013 and the current entry level Chevy Spark,63 which is an example of a vehicle with a design that prioritizes high fuel economy and low price. Other modern vehicles are even slower, such as the entry level Smart ForTwo (14.1 s).63 Although acceleration performance improvements through much of the history of CAFE standards suggest that a widespread reversal is not realistic, decreases can be seen in some vehicle model lines. For example, the current (model year 2010 to present) entry level model of the Chevy Equinox has slower acceleration than the previous generation.63 However, the magnitude of the change in 0-96 km/h acceleration time modelled in the vehicle acceleration option is much more severe, with an increase of 2.9 s versus 0.6 s. Analysis based on the low and high technology price scenarios also results in relatively severe changes to vehicle acceleration, with increases of 1.7 s and 7.7 s, respectively.

100

The vehicle size option is shown as the dashed green price curve in Figure 6-2c. The figure is similar to Figure 6-2b, in that constant size price curves are included that generally slope upward as CAFE standards increase. These include the black dashed price curve, which represents the vehicle price option and the 4000 L size of the 2012 reference vehicle. The vehicle size option price curve intersects the 3300 L price curve in model year 2025; this indicates a 700 L (17%) decrease in interior volume may be required to meet 2025 CAFE standards while maintaining the 2012 reference vehicle price. To put the 3300 L size into perspective, the EPA defines midsize cars as being between 3100-3400 L.15 A 700 L decrease in size may not be feasible for a particular vehicle model line, but size reductions are not unprecedented. The interior volume of the current Chevy Equinox is 300 L (8%) smaller than the previous generation, manifested in the form of both reduced cargo and second row passenger volumes.160 This real world change is actually more than the 200 L (5%) change required based on the low price scenario, but much less than the 900 L (23%) change based on the high price scenario. The vehicle size option results could also be interpreted as suggesting that a shift in market share from SUVs to cars can be used to meet 2025 CAFE standards. As with changes to average vehicle price, the likelihood of a shift in vehicle size may depend on fuel prices. The market share of cars relative to trucks and SUVs has generally increased with higher gasoline prices and vice versa.1

The vehicle size option results (5% to 23% reduction) are more moderate than those based on the simplified method (26% reduction) proposed by An and DeCicco 33 and described in the Methods section This may be because An and DeCicco 33 used interior volume specifications from the EPA 13, which tracks the metric for cars only. An and DeCicco 33 acknowledged that their size metric does not adequately capture the emergence of SUVs. The fuel economy of SUVs may be more sensitive than the fuel economy of cars to relative changes in interior volume. For example, downsizing from an SUV to a car (as modelled in this study) simultaneously reduces both vehicle mass and aerodynamic drag 96. However, downsizing from one car to another may not reduce aerodynamic drag, because of the challenges involved in designing aerodynamic bodies with small car dimensions 36.

6.2.3 Meeting 2025 CAFE standards by modifying driving range alone would require a drastic compromise in functionality

The driving range option is shown as the dotted green price curve in Figure 6-2d. Constant driving range price curves that represent vehicles meeting each model year’s CAFE standard, are

101 also shown. These include the gasoline-fuelled vehicle price option, which is based on a 600 km range, and two battery electric vehicles that provide ranges of 25 and 35 km. Constant driving range price curves representing these different powertrains trend in opposite directions over time because the vehicle price option requires increasing levels of added fuel efficiency technologies to meet CAFE standards, while battery electric vehicles exceed CAFE standards and do not require those technologies (but do utilize them to a lesser degree to improve driving range). The manufacturing cost of the batteries is also expected to decrease more rapidly than the costs of other vehicle components.3 High battery costs prevent the design of a model year 2015 battery electric vehicle for the driving range option because a battery capacity large enough to maintain the 2012 reference vehicle acceleration and size would increase vehicle price.

The driving range option price curve intersects the 35 km price curve in model year 2025; this indicates that a 565 km (94%) decrease in driving range may be required to meet 2025 CAFE standards while maintaining 2012 reference vehicle price, acceleration and size. Analysis based on the high and low technology price scenarios results in similar findings, with driving ranges of 25 and 45 km, respectively. These driving ranges all represent a substantial compromise in functionality compared to gasoline vehicles and even battery electric vehicles currently on the market. For example, the Scion iQ EV has a driving range of 60 km, which is the shortest of any highway-capable, light-duty vehicle in the US.15 (Coincidentally, the price of a Chevy Equinox- like battery electric vehicle with a 60 km range would be approximately the same as the model year 2025 vehicle price option.) However, 35 km is still sufficient for the daily driving needs of 45% of US drivers131 and thus, there may be a niche market opportunity for short range vehicles. For example, the US army utilizes 4000 neighborhood electric vehicles (which are not highway capable) that replaced conventional gasoline vehicles and reduced the fuel costs of “campus-type operations” without increasing vehicle prices.161 Additionally, convenient charging infrastructure (e.g., at home and workplace) could result in battery electric vehicle driving ranges not needing to be comparable to those of gasoline vehicles to provide similar functionality. Therefore, future highway capable electric vehicles that have the same price as current gasoline vehicles may be attractive to fleet operators or individual consumers who do not require long driving ranges.

It should be emphasized that future battery electric vehicles are not restricted to the driving ranges discussed here. The 35 km driving range could be higher if vehicles were more expensive, slower, and/or smaller. Federal tax credits for plug-in electric vehicles are also scaled to battery

102 size, which offsets some of the consumer cost of purchasing plug-in vehicles with long driving ranges.162 Other factors are discussed in the following subsection.

Plug-in hybrid electric vehicle powertrains are a means to extend the driving range of electric vehicles without increasing the size of costly batteries. However, this type of powertrain requires an internal combustion engine system (including associated cooling and emissions control systems) in addition to the plug-in battery system (including charger and associated electronics). Therefore, it is unlikely that a plug-in hybrid electric vehicle could be priced the same as the 2012 reference vehicle price (before model year 2025 and without subsidies).3 Although, the analysis of plug-in hybrid electric vehicles is beyond the scope of this study, in the longer term should component costs decrease sufficiently, they may be an example of a technology which may one day eliminate the need to change vehicle price, acceleration, size or driving range to meet fuel economy targets.

6.2.4 Impact of CAFE Standards Depend on How Vehicles Are Designed

CAFE standards are scaled to vehicle footprint to encourage automakers to continue to provide the range of products consumers can currently choose from. However, this study finds that 2025 CAFE standards are unlikely to be met without modifying some vehicle attributes. The four vehicle design options developed in this study illustrate not only the flexibility that automakers have to meet CAFE standards, but also uncertainty policymakers have in predicting the policy’s societal impact.

The NHTSA Regulatory Impact Analysis29 assumes vehicle prices will increase to improve fuel economy while maintaining other vehicle attributes. The impacts of the legislation will differ from those reported by NHTSA should vehicle prices not increase. The financial benefits to consumers would increase, as fuel savings could be had without increasing upfront expenses. The life cycle negative environmental impacts could decrease if less energy intensive vehicle production is required (e.g., cars instead of SUVs with extensive use of lightweight materials), a less GHG intensive fuel can be utilized (e.g., electricity from low carbon sources instead of gasoline), or if lower prices facilitate more new vehicle sales (replacing less fuel efficient older vehicles).7 Negative environmental impacts could also increase if low vehicle prices increase overall vehicle ownership and exacerbate the rebound effect (increased vehicle travel as fuel

103 efficiency improvements lower the marginal cost of driving).163 Public/driver safety impacts would also depend on the vehicle design changes used to improve fuel economy, in comparison to other vehicles on the road.164

6.2.5 External Factors Influence How Vehicles Will Be Designed to Meet CAFE Standards

Other policies will influence how automakers respond to CAFE standards. If tax credits persist, the manufacturing cost of battery electric vehicles may not need to be comparable to gasoline fuelled vehicles to be financially competitive.162 California’s Zero Emission Vehicle program requires major automakers to sell battery electric vehicles (or fuel cell vehicles, although these are not yet commercially available for sale15) in California or partner states.10 Some of these “compliance cars”64 are sold at a loss, thus making a limited number of battery electric vehicles more attractive to consumers than what was suggested in the previous section, which assumed a fixed price markup over manufacturing costs. For example, Fiat-Chrysler has claimed losses of $14,000 on every Fiat 500e sold,44 and GM has described Chevy Spark EV sales as necessary for the sales of other vehicles, as opposed to being a financially feasible product on its own.165 The sale of battery electric vehicles (or any others that exceed CAFE standards) also reduces the fuel economy improvement required for other vehicles, because the legislation is based on an automaker’s sales weighted average fuel economy, not the fuel economy of individual vehicles.

Socio-economic factors will also influence how consumers respond to CAFE standards. As discussed above, high fuel prices are correlated with consumers purchasing smaller and more affordable vehicles. There are also demographic changes because US drivers are aging and young adults are less likely to get a driver’s license than in the past.166 Changing priorities may favor larger vehicles (that provide accessibility and visibility)167 rather than acceleration performance (because of reduced reaction time).168 Therefore, policies should be developed and analyzed with an understanding of the historical precedence of changing vehicle attributes and the continuing influence of external factors.

104

Chapter 7 Potential Impact of Corporate Average Fuel Economy Standards On The Ability For Non-Petroleum Vehicle To Mitigate Greenhouse Gas Emissions

The US transportation sector is highly reliant on petroleum fuels such as gasoline.2 CNG (compressed natural gas), E85 (85% ethanol, 15% gasoline by nominal volume), and electricity are among the few alternatives that can be used in consumer light-duty vehicles.15 Non- petroleum vehicles can help mitigate petroleum use by displacing gasoline vehicles, but it is important to consider the impact on other environmental metrics, such as GHG (greenhouse gas) emissions.

The life cycle GHG emissions of alternative vehicle fuels depend on the fuel economy ratings of the vehicles in which they are used. For example, Campbell et al.23 compared the use of lignocellulosic biomass-derived ethanol and electricity and concluded the latter was favorable in terms of life cycle GHG emissions because of the higher efficiency of battery electric vehicles (BEVs) as compared to internal combustion vehicles (ICEVs). Luk et al.132 and Laser and Lynd27 subsequently conducted similar analyses but did not reach the same conclusion as Campbell et al.23 and both attributed the discrepancies to differences between the vehicles being compared. Among other differences, Luk et al.132 increased the fuel economy of ICEVs by assuming they were designed for dedicated ethanol (instead of gasoline) use, while Laser and Lynd27 reduced the fuel economy of BEVs by analyzing batteries large enough (in terms of both energy capacity and mass) to provide driving ranges comparable to ICEVs. Although these assumptions were made systematically to produce fair comparisons, in practice, when financial and policy considerations (among others), including high battery prices and non-petroleum fuel vehicle incentives, can affect vehicle choices.

Corporate Average Fuel Economy (CAFE) standards, which are increasingly stringent until model year 2025.9CAFE standards also incentivize the production of non-petroleum vehicles by only accounting for 15% of the energy used when determining compliance with fuel economy targets. Furthermore, the exclusion of fuel production emissions from complementary GHG targets particularly benefits BEVs. This incentive can be used by automakers to meet CAFE standards in lieu of implementing potentially costly fuel efficiency technologies to improve vehicle fuel economy. Anderson and Sallee32 found that automakers previously added E85 flex

105 fuel (gasoline with 0-85% ethanol by nominal volume) capability to relatively inefficient vehicles in their fleet to reduce the cost of meeting CAFE standards. Although credits for E85 flex fuel vehicles are being phased out, they will remain for dedicated non-petroleum vehicles.9 The model year 2015 Honda Civic Natural Gas is a dedicated CNG ICEV, which is less fuel efficient than its gasoline counterpart.15 The CNG vehicle continues to use a less inefficient 5- speed automatic transmission, while the gasoline versions have been upgraded to a higher priced (and more efficient) continuously variable transmission.2, 15 These efficiency differences illustrate how exemptions from increasingly stringent fuel economy targets and credits for dedicated non-petroleum fuel vehicles within CAFE standards could affect the relative GHG emissions of vehicles using different fuels.

The literature does not examine the impact of dedicated non-petroleum fuel credits within CAFE standards on model year 2025 vehicle life cycle GHG emissions. Cheah and Heywood35, and Knittel30 investigated how different vehicles can be designed to meet 2016 and 2020 CAFE standards, respectively, but excluded analysis of GHG emissions and dedicated non-petroleum vehicles. Bandivadikar et al.34 investigated the GHG emissions from vehicles complying with 2016 CAFE standards and also excluded dedicated non-petroleum vehicles. Luk et al.,169 Curran et al.,28 Burnham et al.91 and Venkatesh et al.,55 each compared GHG emissions from BEVs and dedicated CNG ICEVs to gasoline vehicles but do not account for the effect of future CAFE standards.

This study compares the well-to-wheel GHG emissions and ownership costs of a set of hypothetical model year 2025 vehicles. It explores the implications of future non-petroleum fuel use, as a result of different vehicle designs. A gasoline ICEV with a fuel economy rating that meets 2025 CAFE standards is used as a reference vehicle. It is compared to a set of dedicated non-petroleum fuel vehicles with different fuel economy ratings, with all vehicles exceeding CAFE standards because of dedicated non-petroleum fuel vehicle incentives within the standards. Ownership costs and BEV driving ranges are estimated to provide context as these can influence the vehicles automakers choose to produce and consumers choose to purchase. The results of this study aim to inform researchers and other stakeholders about the potential impact of CAFE standards on the ability for non-petroleum vehicles to mitigate GHG emissions by

displacing increasingly fuel efficient petroleum vehicles.

106

7.1 Methods

This study is based on the set of hypothetical model year 2025 petroleum and non-petroleum fuel vehicles that meet or exceed CAFE standards and are illustrated in Figure 7-1. The development of these vehicles is described below, with additional details in Table 7-1 and in Appendix D. The non-petroleum fuels used in this study are CNG and electricity, which are the only two fuels used by dedicated non-petroleum fuel vehicles currently available in the US for consumer purchase.15 The fuel cycle GHG emissions and ownership costs (consisting of vehicle price and net present value of lifetime fuel costs) of the vehicles are compared. Base case estimates are provided using the assumptions described in the subsections below, and are supported by sensitivity, uncertainty and scenario analyses. All prices are presented in 2010 USD.

100 Set of BEVs that exceed CAFE standards and whose fuel economies can improve over time with the use of added fuel efficiency technologies, or decrease if battery size (and thus vehicle mass) is increased to improve driving range NGCCe Short-Distance BEV 80 NGCCe Mid-Distance BEV NGCCe Long-Distance BEV

60 Gasoline ICEV fuel economy improves over time to meet 2025 CAFE standards, which can be achieved with added fuel efficiency CNG High-Efficiency ICEV 40 technologies Gasoline High-Efficiency ICEV CNG Mid-Efficiency ICEV CNG Low-Efficiency ICEV

Vehicle Vehicle Fuel Economy(MPGe) 20 Set of CNG ICEVs that exceed CAFE standards because of non-petroleum fuel credits (and exceed complementary tailpipe GHG targets because of low CNG carbon intensity) thus may or may not use added fuel efficiency technologies to improve fuel economies over time 0 2015 2020 2025 Vehicle Model Year

Figure 7-1: illustrative comparison of how petroleum and non-petroleum vehicle fuel economy can evolve to meet or exceed CAFE standards Notes: Additional vehicle descriptions are provided in Table 1, added fuel efficiency technologies = lightweight materials and other technologies that can improve vehicle fuel economy, ICEV = internal combustion engine vehicle, BEV = battery electric vehicle, CNG = compressed natural gas, NGCCe = electricity from a natural gas combined cycle facility, MPGe = miles per gallon of gasoline energy equivalent

107

Table 7-1: Overview of base case assumptions used in this study

Variable Value Notes Vehicle Fuel Economy Developed with Autonomie96 and Vehicle Attribute Model7 for model year 2025 Gasoline High- 41 MPGea Gasoline vehicle with fuel economy ratingb that meets 2025 CAFE standards Efficiency ICEV (5.7 L/100 km) CNG High- 46 MPGea Modifiedc (for CNG use) version of gasoline vehicle with fuel economy ratingb that Efficiency ICEV (0.16 GJ/100 km) meets 2025 CAFE standards CNG Mid- 36 MPGea Modifiedc (for CNG use) version of gasoline vehicle with fuel economy ratingb that Efficiency ICEV (0.20 GJ/100 km) meets 2020 CAFE standards CNG Low- 29 MPGea Modifiedc (for CNG use) version of gasoline vehicle with fuel economy ratingb that Efficiency ICEV (0.25 GJ/100 km) meets 2015 CAFE standards NGCCe Short- 85 MPGea BEV with weight of battery that provides 100 km driving range,15 which is Distance BEV (23 kWh/100 km) comparable to bestselling Model Year 2014 BEV (130 km Nissan Leaf)60 NGCCe Mid- 78 MPGea BEV with weight of battery that provides 300 km driving range, which is comparable Distance BEV (26 kWh/100 km) to near future BEVs planned by major automakers (e.g., 320 km Chevy Bolt)86 NGCCe Long- 70 MPGea BEV with weight of battery that provides 500 km driving range, which is comparable Distance BEV (30 kWh/100 km) to gasoline ICEVs (560-820 km)27 Vehicle Price Developed with Autonomie and Vehicle Attribute Model for model year 2025 Gasoline High- $23,000 ICEV with upgraded (lightweight) glider and (hybrid electric) powertrain Efficiency ICEV CNG High- ICEV with upgraded (lightweight) glider and (hybrid electric) powertrain, and $26,000 Efficiency ICEV modificationsc for CNG use CNG Mid- $23,000 ICEV with upgraded (lightweight) glider, and modificationsc for CNG use Efficiency ICEV CNG Low- $22,000 ICEV with modificationsc for CNG use Efficiency ICEV NGCCe Long- $49,000 BEV with upgraded (lightweight) glider and 32 kWh battery Distance BEV NGCCe Mid- $36,000 BEV with upgraded (lightweight) glider and 98 kWh battery Distance BEV NGCCe Short- $27,000 BEV with upgraded (lightweight) glider and 170 kWh battery Distance BEV Fuel Production GHGs Obtained from GREET 1 2014 default parameters for the year 2025 Based on 90% gasoline (16% oil sands/84% conventional crude) and 10% corn Gasoline 20 kg CO2eq/GJ ethanol (9% wet mill/91% dry mill, includes both indirect land use change and biogenic carbon sequestration) by nominal volume Based on US natural gas feedstock (58% conventional/42% shale gas, includes CNG 19 kg CO eq/GJ 2 methane leakage with 100 year global warming potential of 34) Based on US natural gas feedstock and a natural gas combined cycle facility, which NGCCe 15 kg CO eq/GJ 2 is the fastest growing source of electrical generating capacity in the US.2 Fuel Prices Obtained from Annual Energy Outlook 2014 Reference Scenario for the year 2025 Gasoline $25/GJ ($3.00 gged) Based on West Texas Intermediate crude oil spot price of $110 per barrel CNG $15/GJ Based on Henry Hub natural gas spot price of $5.30 per million Btu NGCCe $28/GJ Based on US delivered electricity price of $0.10/kWh Vehicle Lifetime Obtained from various sources Kilometers 290,000 km Based on GREET default value for SUVs Years 17 Years Based on median consumer vehicle age from Transportation Energy Data Book Fuel Discount Rate 8% Based on Vehicle Attribute Model default value aestimate of real world (5-cycle) fuel economy presented on a miles per gallon of gasoline energy equivalent (MPGe) basis bCAFE standard for vehicle with a Chevy Equinox-like footprint (4.5 m2)63 in model year 2025 is 53 MPG but is based on unadjusted laboratory (2-cycle) tests, which produce higher ratings than adjusted real world (5-cycle) estimates9 cCNG modifications facilitate higher engine compression ratios and thus thermal efficiencies137 Notes: CNG = compressed natural gas, NGCCe = natural gas combined cycle-derived electricity, ICEV = internal combustion engine vehicle, BEV = battery electric vehicle, gge = gallon gasoline equivalent (lower heating value), all prices in 2010 USD

108

7.1.1 Vehicle Modelling

All vehicles were modelled using software tools developed with industry input, namely Autonomie96 and the Vehicle Attribute Model.3 For comparability, each vehicle is based on a base vehicle model with a common (Chevy Equinox-like) glider (vehicle without powertrain) and powertrain scaled to provide a 0-96 km/h acceleration time of 9.3 s (US Model Year 2013 light-duty vehicle average).13 Base vehicle models are upgraded with added fuel efficiency technologies, as shown in Figure 7-2. The term added fuel efficiency technologies is used here to describe the use of lightweight materials and other technologies analyzed in the Vehicle Attribute Model3 that can be used to improve the fuel economy of the base vehicle model. This study is concerned with the price of fuel economy improvements provided by added fuel efficiency technologies but not the specific technologies themselves, which are not fully detailed by the Vehicle Attribute Model.3 An overview of the vehicle models is provided in the following subsections, with detailed specifications provided Table 7-1 and in Appendix D.

7.1.1.1 Reference Petroleum Vehicle

The petroleum vehicle in this study is referred to as the gasoline high-efficiency ICEV. Autonomie96 was used to estimate the fuel economy and price of a base vehicle model with a conventional gasoline powertrain. The Vehicle Attribute Model3 was used to estimate the prices of added fuel efficiency technologies required to improve the fuel economy rating to meet 2025 CAFE standards for a Chevy Equinox-sized vehicle footprint (vehicle wheelbase multiplied by track width). The high-efficiency ICEV uses both lightweight materials and a hybrid electric powertrain. Note that ICEV is used here to broadly describe vehicles that are propelled by internal combustion engines, as a means to distinguish them from BEVs, which have the unique design considerations discussed in the following subsection.

109

a) Internal Combustion Engine Vehicles 30000

25000 ICEV Added Fuel Efficiency Technologies can increase vehicle 20000 fuel economy, at the expense of higher vehicle price 15000

10000 Vehicle (2010 USD) Price Vehicle

5000 Added Fuel Efficiency Technologies Gasoline ICEV Base Vehicle Model 0 27 33 38 44 49 55 Fuel Economy (MPG)

b) Battery Electric Vehicles BEV Added Fuel Efficiency Technologies can increase vehicle 30000 fuel economy while decreasing vehicle price, because costly

25000 The potential fuel economy improvement from Added Fuel Efficiency Technologies is greater 20000 for ICEVs than BEVs, because ICEVs can use technologies 15000 already found in BEVs, such as

10000 Added Fuel Efficiency Technologies

Vehicle Vehicle Price (2010 USD) 100 km Driving Range Battery 5000 BEV Base Vehicle Model without battery 0 71 74 77 80 82 85 Fuel Economy (MPGe)

Figure 7-2: Relationship between vehicle price and fuel economy for a) internal combustion engine vehicles and b) battery electric vehicles

110

7.1.1.2 Dedicated Non-Petroleum Vehicles

The non-petroleum vehicles used in this study are similar to the gasoline high-efficiency ICEV. Differences include powertrain modifications that are required to account for the use of different fuels. Additionally, the level of added fuel efficiency technologies (based on the Vehicle Attribute Model3) varies as these vehicles do not require fuel efficiency improvements to meet or exceed CAFE standards. These differences are further discussed below.

The CNG-fuelled vehicles modelled capture the ability for dedicated CNG vehicles to have different fuel economy ratings while still exceeding CAFE standards. The CNG ICEVs are versions of the gasoline high-efficiency ICEV with engine and fuel tank modifications based on assumptions from the Vehicle Attribute Model and different levels of added fuel efficiency technologies.3 The low- and mid-efficiency ICEVs are used to illustrate the effect of automakers adopting different upgrade timelines for petroleum and non-petroleum fuel vehicles , such as in the aforementioned case of Honda adding a continuously variable transmission to the gasoline version but not the CNG version of their Civic.15 Beyond highlighting the use of lightweight materials and distinguishing between hybrid and non-hybrid electric powertrains, the Vehicle Attribute Model3 does not provide detail on how other specific technologies, such as transmissions, are factored into its relationship between incremental vehicle price and fuel economy improvement.

The electricity-fuelled vehicles modelled encompass BEVs with different driving ranges, while still exceeding CAFE standards. These vehicles consist of base vehicle models with BEV powertrains developed within Autonomie.96 Each BEV has a different battery size (in terms of both energy capacity and mass) to provide the different driving ranges. The particular battery size and level of added fuel efficiency technologies used for each vehicle are determined iteratively to minimize the vehicle price that achieves the target driving range, as shown in Figure 7-2.

7.1.2 Fuel Modelling

Fuel production GHG emissions detailed in Table 7-1 are default GREET7 values for 2025 gasoline, CNG and electricity from a natural gas combined cycle facility (NGCCe). The latter is the fastest growing source of electricity generating capacity in the US.2 Electricity produced

111 from higher and lower carbon intensity energy sources is also examined because the location where the BEVs are charged and the underlying source of electricity used affects GHG emissions. Scenario analyses are conducted to illustrate the importance of this source of variability, as opposed to aggregating a variety of sources based on US grid-average characteristics.

Fuel use GHG emissions are a function of fuel carbon intensity, vehicle fuel economy and vehicle methane emissions. Gasoline and CNG contain 72 and 56 g of carbon per MJ, 7 respectively, based on GREET data. GREET assumes gasoline vehicles emit 0.006 g CH4/km, while methane emissions from CNG vehicles are ten times higher.7 Nonetheless, vehicle methane emissions are negligible when compared to well-to-wheel GHG emissions.7

7.1.3 Sensitivity, Uncertainty and Scenario Analyses

Variables examined in the sensitivity, uncertainty and scenario analyses are selected based on results of studies evaluating the uncertainty of GHG emissions from vehicles using natural gas- derived fuels/electricity (Luk et al.,169Curran et al.,28 Burnham et al.91 and Venkatesh et al.55). Probability distribution functions for each of the examined variables (in Table 7-1) are detailed in Appendix D. The variables are examined individually in the sensitivity analysis and collectively in the uncertainty analysis. The uncertainty analysis is conducted by using Crystal Ball software and simulating 10,000 trials. These results are presented in the form of 90% confidence intervals (i.e., 5th to 95th percentile results). A scenario analysis is also conducted to analyze the use of other non-petroleum energy sources (coal, biomass and landfill gas). This is done to distinguish this major (in terms of GHG emissions7) source of variability among the many other sources of uncertainty (e.g., real world fuel economy169).

112

7.2 Results and Discussion

This study compares the ownership costs and well-to-wheel GHG emissions of a set of hypothetical model year 2025 vehicles that meet or exceed (in the case of non-petroleum vehicles) CAFE standards. The results show that CAFE standards present an opportunity for automakers to produce dedicated non-petroleum vehicles that have lower ownership costs than petroleum vehicles. The results also show that this flexibility may lead to non-petroleum vehicles that could have higher well-to-wheel GHG emissions than petroleum vehicles on a per km basis, even if the non-petroleum energy source is less carbon intensive on an energy equivalent basis.

The results are illustrated in Figure 7-3. The ownership costs and well-to-wheel GHG emissions are presented in Figure 7-3a and Figure 7-3b, respectively. The disaggregated base case estimates for each vehicle are presented (left column) and the Monte Carlo analysis of incremental results (center column), which are the differences between each of the non- petroleum fuel vehicles and the gasoline vehicle (a negative value indicates the non-petroleum vehicle has lower ownership costs or GHG emissions than the gasoline high-efficiency ICEV). The column on the right presents sensitivity analysis variables that have the greatest impact upon the results; two variables are shown for each incremental comparison.

7.2.1 Model year 2025 CNG vehicles can have lower vehicle price and ownership costs than gasoline vehicles that meet CAFE standards

Figure 7-3a shows the base case ownership costs are approximately $31,000 for each of the three CNG vehicles, which are less the $33,600 costs of the gasoline vehicle. The similarity in CNG vehicle ownership costs are due to a trade-off between CNG vehicle price ($22,100-$25,700) and fuel cost ($5,800-$9,100) because added fuel efficiency technologies increase vehicle price while reducing fuel costs. The difference between the CNG and gasoline vehicle ownership costs is highly uncertain. The Monte Carlo analysis of the incremental ownership costs shows that among the CNG vehicles, only the low- and mid-efficiency models are likely (within a 90% confidence interval) to have lower ownership costs than the gasoline vehicle. The higher vehicle price of the CNG high-efficiency ICEV may not be offset by reduced fuel costs. The sensitivity analysis shows that the ownership costs of the CNG high-efficiency ICEV can be higher than those of the gasoline vehicle, depending on fuel price. Therefore, although automakers could produce (and consumers could subsequently purchase) CNG vehicles that are more fuel efficient

113 than gasoline vehicles that meet 2025 CAFE standards, there is no clear financial incentive to do so because less fuel efficient CNG vehicles can have lower vehicle prices (CNG low-efficiency ICEV) and ownership costs (CNG low- and mid-efficiency ICEVs) while still exceeding CAFE standards.

The above findings provide insights into the design decisions regarding real world CNG vehicles. As noted previously, the CNG version of the model year 2015 Honda Civic is less efficient than the gasoline models.15 This dedicated CNG vehicle does not require fuel economy improvements to meet CAFE standards and continuing to use older, less efficient technologies can help avoid increases in vehicle price and potentially total ownership costs. Thus, there is a financial incentive for automakers to produce (or for consumers to purchase) CNG vehicles that are less fuel efficient than gasoline vehicles.

114

a) Monte Carlo analysis of Sensitivity analysis of incremental results relative to incremental results relative to Base case results Gasoline High-Efficiency ICEV Gasoline High-Efficiency ICEV (Thousand 2010 USD) (90% Confidence Interval) (90% Confidence Interval) $0 $20 $40 $60 -25% 0% 25% 50% -25% 0% 25% 50%

Gasoline High- Efficiency ICEV CNG High- Fuel Price Efficiency ICEV CNG Fuel Tank

CNG Mid- Fuel Price Efficinecy ICEV CNG Fuel Tank CNG Low- Fuel Price Efficiency ICEV CNG Fuel Tank NGCCe Long- Fuel Price Ownership Costs Ownership Distance ICEV Battery Price NGCCe Mid- Fuel Price Distance ICEV Battery Price NGCCe Short- Fuel Price Distance ICEV Battery Price

Vehicle Price Fuel Costs (NPV) b) Monte Carlo analysis of Sensitivity analysis of incremental results relative to incremental results relative to Base case results Gasoline High-Efficiency ICEV Gasoline High-Efficiency ICEV

(g CO2eq/km) (90% Confidence Interval) (90% Confidence Interval) 0 50 100 150 200 -60% -30% 0% 30% -60% -30% 0% 30%

Gasoline High- Efficiency ICEV

CNG High- CNG ICEV Mod. Efficiency ICEV CNG Comp. η CNG Mid- CNG ICEV Mod. Efficinecy ICEV CNG Comp. η CNG Low- CNG ICEV Mod. Efficiency ICEV CNG Comp. η

Wheel GHG GHG Emissions Wheel

- NGCCe Long- Fuel Economy

- Distance ICEV NGCCe Gen. η NGCCe Mid- Fuel Economy

Well Distance ICEV NGCCe Gen. η NGCCe Short- Fuel Economy Distance ICEV NGCCe Gen. η Vehicle Use Fuel Production

Figure 7-3: Base case results and Monte Carlo and sensitivity analyses of the incremental results relative to the Gasoline High-Efficiency ICEV for a) ownership costs and, b) well-to- wheel GHG emissions Notes: List on the right-hand side label indicates sensitivity analysis variables (see Appendix D), ICEV = internal combustion engine vehicle, BEV = battery electric vehicle, CNG = compressed natural gas, NGCCe = electricity from a natural gas combined cycle facility, η = efficiency, other sources of electricity are examined with scenario analyses presented in Appendix D

115

7.2.2 Model year 2025 battery electric vehicles can have similar ownership costs as gasoline vehicles that meet CAFE standards

Figure 7-3a shows that the base case ownership cost for the NGCCe short-distance BEV ($33,600) is approximately the same as that of the gasoline high-efficiency ICEV because the differences in vehicle price and fuel costs can offset each other. This is not the case with the other two BEVs, which each have higher vehicle prices ($27,400-$50,200), fuel costs ($6,200- $7,600) and, therefore, ownership costs ($42,000 and $57,000). Vehicle prices are higher for the BEVs with longer driving ranges, because they have batteries with a larger energy capacity, increasing vehicle mass and lowering fuel economy. These similarities and differences in ownership costs, which take into account uncertainties in both battery and fuel prices, are significant at a 90% confidence level.

The above findings provide insights into design decisions for real world plug-in electric vehicles. Automakers offer consumers the option of plug-in vehicles with extended driving ranges, at the expense of lower fuel economy and higher vehicle price. The model year 2015 Tesla Model S is a BEV available with 330 km ($69,900 and 95 MPGe) and 420 km ($79,900 and 89 MPGe) driving ranges.15 Consumers also have the option of gasoline plug-in hybrid electric powertrains as a means of extending plug-in electric vehicle driving ranges in lieu of larger batteries, though the internal combustion engine system that provides the additional functionality also adds to vehicle mass and reduces fuel economy. The model year 2015 BMW i3 is available as a BEV with an all-electric range of 130 km ($42,400 and 124 MPGe) and as a gasoline plug-in hybrid electric vehicle with a combined gasoline and electric range of 240 km ($46,250 and 117 MPGe when operating on electricity).15 Thus, the financial attractiveness of plug-in electric vehicles compared with gasoline vehicles depends in large part on driving range.

Note that the results in Figure 7-3 capture the uncertainty in price for a battery with fixed physical characteristics, based on forecasted improvements in usable energy density. It is possible that higher priced batteries with higher energy densities could be used, which would reduce the battery mass required for a given driving range. However, this may not be desirable for consumers because of the already high price of BEVs and relatively low price of electricity. On the other hand, a breakthrough in both battery energy density and price could reduce the sensitivity of BEV fuel economy to driving range.

116

7.2.3 CAFE standards could lead to the use of non-petroleum energy sources that are less carbon intensive than petroleum use on an energy equivalent basis but result in higher GHG emissions on a per km basis

The use of CNG in model year 2025 vehicles can result in higher or lower well-to-wheel GHG emissions than the gasoline high-efficiency ICEV (Figure 7-3b). This is despite the fact that CNG is a less carbon intensive fuel than gasoline on an energy equivalent basis. The base case GHG emissions from the CNG high- and mid-efficiency ICEVs are lower (119 and 149 g/km, respectively), while those of the CNG low-efficiency ICEV are higher (185 g/km), than those of the gasoline high-efficiency ICEV (161 g/km). These differences are significant at a 90% confidence llevel. The results are relatively insensitive to Vehicle Attribute Model3 assumptions regarding internal combustion engine and fuel tank modifications for CNG use, which change the relative fuel efficiencies of the CNG and gasoline ICEVs.

The base case estimates of GHG emissions from the NGCCe short-, mid- and long-distance BEVs (101, 110 and 122 g/km, respectively) are all lower than those of the gasoline high- efficiency ICEV. However, only the NGCCe short-distance BEV has lower emissions at a 90% confidence level, despite the fact that natural gas is a less carbon intensive energy source than petroleum, and despite the superior fuel economy ratings of the three BEVs compared to the gasoline high-efficiency ICEV. The BEV GHG emissions are particularly sensitive to NGCCe production efficiency and fuel economy, based on GREET probability distribution functions.7 The uncertainties in ICEV and BEV fuel economy are assumed to be uncorrelated due to their different responses to variables such as weather78 and driving patterns.77

Natural gas is the only non-petroleum energy source included in Figure 7-3 but others are shown in Appendix D. The use of carbon intensive energy sources (coal for electricity production) can result in all of the BEVs analyzed having higher GHG emissions than the gasoline high- efficiency ICEV, regardless of driving range. Conversely, the renewable energy sources (landfill gas and biomass for CNG and electricity production, respectively) can result in all of the non- petroleum vehicles analyzed having lower GHG emissions than the gasoline high-efficiency ICEV. These findings are important for BEVs because the carbon intensity of grid-electricity can vary greatly across the US and the incremental cost of many forms of low carbon electricity is relatively minor.2 Therefore, the well-to-wheel GHG emissions of non-petroleum fuel use can be

117 higher or lower than that of vehicles using petroleum, depending on the vehicles they are used in and how the fuels are produced.

7.2.4 The relative fuel economy ratings of vehicles using different fuels will likely change over time

The well-to-wheel GHG emissions of alternative fuels depend on how vehicle designs respond to increasingly stringent CAFE standards. The results in Figure 7-3a provide context in the form of vehicle prices and fuel costs, which influence the design decisions of automakers and purchase decisions of consumers. The results suggest that, for CNG vehicles, there is a financial incentive to limit the incorporation of fuel efficiency technologies. This includes technologies that may otherwise be added to gasoline vehicles, to meet increasingly stringent CAFE standards. Therefore, the assumption made in Luk et al.169 and Curran et al.28 that future CNG vehicles could be equally or more fuel efficient than future gasoline vehicles may be optimistic. The results in those studies may be overstating some of the potential environmental benefits of future CNG vehicles.

There are other factors that will influence the relative fuel economy ratings of BEVs and gasoline vehicles. Figure 7-3 shows the detrimental impact of increasing driving range on fuel economy.3 Additionally, unlike with CNG ICEVs, many powertrain (as opposed to glider) technologies that can improve future gasoline ICEV fuel economy may not be transferable to BEVs; for example, most current gasoline ICEVs could benefit from the addition of regenerative braking, which BEVs already have.2 This results in the CAFE standard fuel economy targets between 2015 and 2025 increasing at a rate that exceeds the maximum potential for improvement in BEV fuel economy, as estimated by the Vehicle Attribute Model.3 During this time period, the Energy Information Administration forecasts that the fuel economy of conventional gasoline vehicles will improve by 44%, while BEVs (with a 160 km driving range) will increase by only 4%.2 Therefore, the fuel economy ratings of BEVs and gasoline vehicles will likely approach each other over time.

7.2.5 Effectiveness of low carbon fuel standards depends on their ability to capture differences in vehicle fuel economy

The California Low Carbon Fuel Standard (LCFS) aims to capture differences in fuel economy ratings of vehicles using petroleum and non-petroleum fuels with energy economy ratios.12

118

These ratios are calculated by dividing the fuel economy of new non-petroleum vehicles by the fuel economy of otherwise similar petroleum vehicles and are periodically revised.12

The California LCFS currently uses an energy economy ratio of 1.0 and 3.4 for CNG and electricity, respectively.12 The energy economy ratios in this study, using the base case fuel economy, ranges from 0.7 to 1.1 for CNG and 1.7 to 2.1 for electricity. Should the energy economy ratio in California’s LCFS be reduced in the future to 0.7 for CNG or 1.7 for electricity, the results in Figure 7-3b suggest that CNG and NGCCe may no longer be considered a low carbon fuel (despite having relatively low carbon intensities on an energy equivalent basis), because their use could result in higher GHG emissions on a per km basis than the gasoline high- efficiency ICEV.

7.2.6 Stakeholders should be aware of the potential impacts of CAFE standards on the relative performance of alternative vehicles and fuels

Stakeholders examining alternative vehicles and fuels should to be aware of the impact of increasingly stringent CAFE standards. This may be a particular issue for the evaluation of plug- in electric vehicles, whose fuel economy advantage (on an energy equivalent basis) over gasoline ICEVs will likely decrease over time. Nordelöf et al.79 conducted a review of electric vehicle life cycle assessments and found that temporal assumptions were often not stated. Thus, for example, when Kennedy170 reviewed the scientific literature and proposed that countries should aim to reduce electricity generation emissions to 600 t CO2e/GWh or less, so that plug-in electric vehicles could be used to mitigate GHG emissions by displacing gasoline vehicles, there was a lack of temporal context. The vehicle fuel economy forecasts by the Energy Information Administration2 suggest that the breakeven point for electricity generation emissions may be a moving target. Our results show that even the use of electricity with a carbon intensity of less than the threshold proposed by Kennedy170 (NGCCe base case GHG emissions of 450 t

CO2e/GWh), could result in higher GHG emissions on a per km basis than a gasoline vehicle designed to meet 2025 CAFE standards. Findings based on historical or currently available vehicles could quickly become outdated as increasingly fuel efficient gasoline vehicles are produced. The results of this study aim to inform researchers and other stakeholders about how CAFE standards may influence the ability of non-petroleum vehicles to mitigate GHG emissions by displacing increasingly fuel efficient petroleum vehicles.

119

Chapter 8 Conclusion

scientific literature as well as informing public policy.

8.1 Chapter Conclusions

This thesis consists of four primary research chapters. Each of these chapters is motivated by a question introduced in Section 1.1. The findings are discussed in the subsections below.

8.1.1 Should light-duty vehicles use ethanol or bio-electricity?

Renewable fuel standards promote the use of ethanol, which has become the dominant non- petroleum fuel used in US light-duty vehicles. However, ethanol production from biomass (and thus ability to mitigate greenhouse gas emissions (GHG) and petroleum use) is limited by feedstock and land availability. The development of plug-in electric vehicles provides another potentially more efficient means of utilizing biomass as a transportation energy source. Several studies concluded that the use of bio-electricity as a transportation fuel is favorable because of the efficiency of plug-in electric vehicles. Chapter 4 is based on a life cycle energy use and GHG emission inventory analysis of hybrid poplar biomass use in model year 2015 conventional and emerging vehicle powertrains. All of the E85 (85% ethanol) and bio-electricity pathways developed have similar life cycle GHG emissions (~5 kg CO2eq./100 vehicle kilometers travelled), considerably lower (65-85%) than those of reference gasoline and U.S. grid-electricity pathways. E85 use in a hybrid electric vehicle (HEV) and bio-electricity use in a battery electric vehicle (BEV) also have similar life cycle biomass and total energy use (~350 and ~450 MJ/100 vehicle kilometers travelled, respectively); differences in well-to-pump and pump-to-wheel efficiencies can largely offset each other. These energy use and net GHG emissions results contrast with some of the findings in literature, which report better performance on these metrics for bio-electricity compared to ethanol. The primary source of differences in the studies is related

120 to our development of pathways with comparable vehicle characteristics. Regional characteristics may create conditions under which either ethanol or bio-electricity may be the superior option; however, neither has a clear advantage in terms of GHG emissions or energy use.

8.1.2 Do plug-in electric vehicles provide life cycle air emission impact benefits over internal combustion engine vehicles using the same primary energy source?

Unlike GHG emissions, the effects of criteria air contaminant (CAC) emissions depend on the location they are released. The Zero Emission Vehicle program promotes the use of BEVs, which lack tailpipe emissions, and use non-petroleum fuels, which is increasingly natural gas- derived grid electricity. The ability for automakers to meet regulations by producing low emission compressed natural gas (CNG) vehicles is being phased out. However, CNG vehicles may have lower upstream emissions and ownership costs than BEVs. Chapter 5 compares CNG use directly in a conventional vehicle (CV) and HEV, and natural gas-derived electricity use in a plug-in BEV. The incremental life cycle air emissions (climate change and human health) impacts and life cycle ownership costs of model year 2020 non-plug-in (CV and HEV) and plug- in (BEV) light-duty vehicles are evaluated. Replacing a gasoline CV with a CNG CV, or a CNG CV with a CNG HEV can provide life cycle air emissions impact benefits without increasing ownership costs; however, the BEV using natural-gas-derived electricity will likely increase costs (90% confidence interval: $1000 to $31,000 incremental cost per vehicle lifetime). Furthermore, eliminating HEV tailpipe emissions via plug-in vehicles has an insignificant incremental benefit, due to high uncertainties (90% confidence interval: -$1000 and $2000). Vehicle CACs are a relatively minor contributor to life cycle air emissions impacts because of strict vehicle emissions standards. Therefore, policies should focus on adoption of plug-in vehicles in non-attainment regions, because CNG vehicles are likely more cost-effective at providing overall life cycle air emissions impact benefits.

8.1.3 How can vehicles be designed to meet future Corporate Average Fuel Economy standards?

The costs and benefits of non-petroleum fuelled vehicles are typically quantified in comparison to petroleum fuelled vehicles. However, recently amended Corporate Average Fuel Economy (CAFE) standards will require substantial design changes to US light-duty vehicles. Chapter 6 is

121 a case study that systematically examines four vehicle attributes that can be modified to improve fuel economy and meet future year CAFE standards, by developing vehicle design options that modify only one of the four attributes on a reference model year 2012 Chevy Equinox-like CV: a crossover sport utility vehicle (the fastest growing vehicle segment) that has an average light- duty vehicle footprint (fuel economy targets are scaled by footprint). This method illustrates how aggregate changes in an automaker’s fleet (which CAFE standards regulate) could manifest in a typical vehicle, but not the design limitations of individual vehicles, because the sale of vehicles that exceed fuel economy targets can facilitate the sale of vehicles that do not meet targets. The results show that the reference vehicle can meet the 66% increase in fuel economy targets between model years 2012 to 2025 with (i) a 10% vehicle price increase (lightweight HEV powertrain), (ii) a 31% increase in 0-96 km/h acceleration time (smaller engine), (iii) a 17% interior volume decrease (smaller body), or (iv) a 94% driving range decrease (BEV powertrain), while other attributes are maintained. Although there is uncertainty in future technology prices, a set of price scenarios agree that expected component cost reductions over time are insufficient to offset the costs of additional fuel efficiency technologies needed to meet model year 2025 fuel economy targets while preserving other vehicle attributes. The research highlights the flexibility that automakers have to meet CAFE standards, and also the uncertainty in predicting the policy’s societal impact.

8.1.4 How might CAFE Standards effect the ability for alternative fuel vehicles to mitigate GHG emissions by displacing petroleum fuel vehicles?

The life cycle greenhouse gas (GHG) emissions of alternative fuels dependent upon to the fuel economy ratings of the vehicles they are used in. Chapter 7 compares well-to-wheel GHG emissions and ownership costs of a set of hypothetical model year 2025 vehicles. A reference gasoline vehicle has a fuel economy rating that meets CAFE standards. It is compared to a set of dedicated CNG vehicles and BEVs with different fuel economy ratings, but all vehicles exceed CAFE standards because of dedicated non-petroleum fuel vehicle incentives. Ownership costs and BEV driving ranges are estimated to provide context as these can influence the automaker and consumer decisions. The results show that CAFE standards present an opportunity for automakers to produce CNG vehicles that have lower ownership costs than gasoline vehicles. However, this flexibility may lead to lower efficiency CNG vehicles and BEVs with long driving ranges that could have higher well-to-wheel GHG emissions than gasoline vehicles on a per km

122 basis, even if the non-petroleum energy source is less carbon intensive on an energy equivalent basis. These results aim to inform stakeholders about how CAFE standards may influence the ability for non-petroleum vehicles to mitigate GHG emissions, by displacing increasingly fuel efficient petroleum vehicles.

8.2 Thesis Conclusions

The research in four primary research chapters produced findings beyond insights into the specific questions introduced in Section 1.1. Two overarching thesis conclusions are discussed here.

8.2.1 Plug-in electric vehicles do not have a clear technological advantage over non-plug-in vehicles in terms of life cycle energy use, GHG emissions and life cycle air emissions impacts.

When comparing plug-in and non-plug-in vehicles using a common primary energy source (Chapters 4 and 5), neither has a clear advantage in terms of life cycle energy use, GHG emissions, or life cycle air emissions impacts. This similarity in the results is despite the superior efficiency of electric motors over internal combustion engines. Among other reasons, this is because HEVs are non-plug-in vehicles that benefit from the use of electric motors. Non-plug-in vehicles are also required to have CAC emissions control systems, which reduces some of the advantage that plug-in vehicles have regarding tailpipe emissions. Beyond vehicle design, there is substantial uncertainty in real world vehicle fuel economy and some driving conditions favor certain vehicle powertrains over others. Additionally, there are upstream variables including electricity generation efficiency, which can be a major source of uncertainty. Differences in upstream (fuel production) and downstream (vehicle fuel consumption) efficiencies and emissions among plug-in and non-plug-in vehicles can offset each other. Therefore, broad generalizations based on point estimates should be avoided and alternative vehicles/fuels should be evaluated for the particular conditions under which they will be used.

8.2.2 The environmental and financial competitiveness of alternative fuels will change as the efficiency of petroleum use improves

The environmental and financial competitiveness of alternative fuels depend on the vehicles they are used in. Although the cost of alternative vehicle technologies will be gradually reduced over time, so will petroleum vehicle fuel consumption as vehicle fuel economy improves to meet

123

CAFE standards. This could reduce the potential GHG emission and fuel cost reductions from using non-petroleum fuels. Thus, for example, plug-in vehicles will continue to require short driving ranges to be financially competitive with gasoline vehicles in model years 2020 (Chapter 5) and 2025 (Chapters 6 and 7), despite battery cost reductions.

8.3 Limitations

This thesis is based on assumptions discussed in Chapters 4-7. The thesis conclusions should be understood within the context of these assumptions. The main limitations of these assumptions are discussed here.

8.3.1 Research Scope

The range of potential light-duty vehicles is diverse and changes over time. This thesis draws conclusions from analyzing particular sets of vehicles at particular points in time. Two important factors to consider include consumer availability and vehicle class.

8.3.1.1 Consumer Availability

The results of Chapter 4 are based on Autonomie vehicle models with comparable characteristics. A common vehicle glider (vehicle without powertrain) was used to maintain size, aerodynamic drag and rolling resistance. The different powertrains were scaled to provide a consistent 0-96 km/h acceleration performance. This approach is in contrast to the methods commonly used in the literature, which involved selecting real world vehicles often with dissimilar characteristics. The use of similar vehicles is more appropriate for comparing the technological potential of alternative fuels and powertrains because they are less likely to conflate other vehicle design considerations that are shaped by market conditions at a particular point in time. However, this type of comparison may be irrelevant to consumers who are limited by the choices available to them. For example, there are no model year E85 HEVs available, let alone one with comparable characteristics to a BEV.15 There is also relatively little cellulosic ethanol and bio-electricity being produced.2 The findings in Chapter 4 are important for policymakers who help shape the future availability of alternative fuels and vehicles but are not applicable to near-term consumer decisions.

124

Consumer availability is also a limitation of the findings in Chapter 5, although to a lesser extent. CNG HEVs have been developed,134 but none are currently available.15 However, these vehicles may be available by the model year 2020 examined. Additionally, natural gas is already an abundantly used energy source.2

8.3.1.2 Vehicle Class

All of the results in this thesis are based on US light-duty vehicles. Chapters 4 and 5 are based on a midsize car (typically used in the literature), while Chapters 6 and 7 are based on a small crossover SUV (to facilitate the analysis of downsizing to a car). Both the absolute and relative performance of the alternative fuels and technologies will differ for other vehicle classes. For example, the energy that can be potentially captured by regenerative braking depends on vehicle mass. Beyond physical differences, there are also differences in applicable policies. For example, trucks are subject to different fuel economy standards than cars and heavy-duty vehicles are not regulated by the same emissions standards as light-duty vehicles.9, 139 The location and the manner in which these different vehicles operate will affect life cycle environmental and health impacts.

Results may also differ when examining vehicles in other markets, for similar reasons. Local consumer preferences can result in different characteristics being required to model representative vehicles. Policy differences include aforementioned vehicle fuel economy and emissions policies, but also fuel and/or carbon taxes, which can affect the cost-effectiveness of achieving environmental aims with alternative fuels and technologies. The marginal health impacts of air emissions are also dependent on background air emission levels and dispersion patterns.

8.3.2 Research Tools

The complexities of alternative light-duty vehicles in this thesis were modelled with the tools introduced in Chapter 3. In particular, Autonomie, the Vehicle Attribute Model, GREET and APEEP are also further discussed in Chapters 4-7. There are limitations to these tools and the manners in which they are used.

125

8.3.2.1 Autonomie and Vehicle Attribute Model

Chapters 4-7 rely on vehicle models developed within Autonomie96 and/or using assumptions from the Vehicle Attribute Model.3 Both of the models are based on recent, but not current technologies and prices. For example, Chapter 4 is based on the midsize car template within Autonomie,96 which has a seventh generation (model year 2003-2007) Honda Accord sedan glider. Component specifications for forecasted near-term component mass reductions are adjusted in Chapter 4, but other technological developments may not be captured in the modelled relationships between vehicle fuel economy and acceleration. Likewise, the Vehicle Attribute Model3 is used in Chapter 5 to estimate vehicle prices based on technology price/performance forecasts that are extrapolated by General Motors from model year 2008 vehicle characteristics. Longer-term forecasts may not capture the real world development, particularly for less mature technologies that have greater potential for technological breakthroughs, such as plug-in electric vehicle batteries.

Autonomie96 and the Vehicle Attribute Model3 are used together in Chapters 6 and 7. Autonomie96 is used to produce base vehicle models, whose prices and fuel economy ratings are adjusted based on the prices of fuel efficiency improvements from the Vehicle Attribute Model.3 Other vehicle attributes are assumed to be constant. However, acceleration performance is dependent on vehicle mass. Therefore, lightweight materials can improve acceleration performance by reducing the load on the powertrain. Conversely, the addition of components to improve powertrain efficiencies (e.g., HEV components) can increase mass, which would reduce acceleration performance. The level of detail in the Vehicle Attribute Model3 makes it unclear if these opposing factors offset each other or if one is dominant in each of the particular vehicle models analyzed.

8.3.2.2 GREET and APEEP

Chapter 4 describes the GHG emissions from biomass (poplar tree) production modelled with GREET.7 Although GREET7 includes the emissions from land use change for more common ethanol feedstock, such as corn, no value is provided for tree farming. Life cycle GHG emissions could increase if agricultural production is displaced or decrease if marginal land is used (and soil carbon is increased). Thus, biomass production emissions will depend on particular circumstances.

126

Chapter 5 models life cycle CAC emissions with GREET.7 GREET7 assumes vehicle cycle processes largely occur within US boundaries. US grid-average electricity is assumed to be used for vehicle production, which results in the vehicle cycle accounting for a substantial portion of life cycle CAC emissions. However, the automotive sector is concentrated in parts of the US (in particular the Midwest), which relies more heavily on coal than the US average. Additionally, even vehicles assembled within the US use components produced elsewhere, including Ontario, which relies on nuclear and hydro sources for electricity production. The underlying vehicle component assumptions within GREET7 may also not match the components required to achieve the performances modelled with Autonomie96 and the Vehicle Attribute Model.3 Therefore, vehicle production emissions will depend on particular circumstances.

Chapter 5 uses the APEEP97 model to estimate the human health impacts of CAC emissions. The model uses recent background emissions levels, which may not reflect future levels. Only domestic health impacts are estimated, while the US air pollutants can affect the health of those beyond US borders. APEEP97 models emissions dispersion, transformation and its effect on human heath, but simplifies these processes. The use of more resource intensive models could produce more robust results (e.g., Tessum et al.75 uses the WRF-Chem model,171 which does not operate on personal computers). Additionally, the Monte Carlo analysis in Chapter 5 accounts for uncertainty of CAC emissions impacts by varying the geographic location in which emissions are released but does not factor the numerous other sources of uncertainty (e.g., the aforementioned dispersion modelling).

8.4 Future Research

The broad scope of this thesis lends itself to further examination in future research. Some opportunities are based on the limitations discussed in Section 8.3. Others are arise from the dynamic nature of the automotive industry and policies it is shaped by.

8.4.1 Future Vehicle Cycle Impacts

Chapters 6 and 7 examined vehicle design options for meeting model year 2025 CAFE standards but did not examine vehicle (production, maintenance and disposal) cycle impacts. This is because the energy use and GHG emissions contribution of this life cycle stage is minor

127 compared to gasoline use in current and near-future vehicles. However, the relative contribution may increase as fuel consumption is reduced beyond model year 2025 requirements.

The use of lightweight materials is highlighted as an example means of improving vehicle fuel economy by the Vehicle Attribute Model.3 However, the literature does not always agree on the effect of vehicle weight reduction on life cycle energy use. One source of variability can be attributed to different ways vehicle mass can be reduced. Different materials are not completely interchangeable. Individual vehicle components have unique requirements that determine what alternative materials are or are not suitable and the degree to which mass can be reduced. Additionally, the effect of mass reductions on vehicle fuel economy depends on the location within the vehicle; rotational mass reductions (e.g., in wheels) yield greater improvements than equivalent mass reductions in components that move linearly with the vehicle. The life cycle implications of some lightweight components have been examined independent of each other based on unique assumptions.

The potential life cycle merits of lightweight materials also depend on the type of vehicle in which they are utilized. More aerodynamic vehicles can attribute a greater fraction of their energy use to mass-related loads. HEVs have regenerative braking to capture some of the kinetic energy in the vehicle mass that would otherwise be lost during braking, and the recovered energy assists with the acceleration of the vehicle mass. BEVs have heavy batteries, which provide the potential for substantial secondary mass reduction because the battery size can be reduced while still providing the same driving range if vehicle mass is reduced elsewhere to improve fuel economy. These life cycle financial and environmental impacts of these variables have been partially examined in the literature, but have not been systematically compared.172,26

8.4.2 Additional Opportunities

Economic modelling of the technologies investigated in this thesis could be conducted. This thesis discusses discrepancies between technological possibilities and real world consumer options, and the resulting life cycle environmental impacts. An integrated assessment model that includes general equilibrium modelling of economic conditions would provide insights into decisions automakers and consumers will make in response to different policies. This could inform not only potential demand for particular product options, but changes in vehicle usage, which are important to understand overall environmental implications.

128

More detailed air emissions modelling would provide further insights into the potential benefits of plug-in electric vehicles. However, models such as WRF-Chem171 are resource intensive. Therefore, the scope could be limited to vehicles in California, which has the Zero Emission Vehicle program10 and where air quality is of particular concern.

Life cycle environmental impacts beyond those from air emissions could also be examined. For example, GREET7 has recently been updated to include water use. Non-renewable resource use beyond fossil fuels would also be of interest as alternative vehicles are produced using different materials.

The scope of the research in this thesis could be extended to other regions and vehicle classes. China is a particularly fast growing vehicle market. Heavy- and medium-duty vehicles have particularly poor fuel economy and thus have greater potential for improvement, as compared to light-duty vehicles. Differences in what, where and how vehicle technologies and fuels are used provide intriguing opportunities to be explored.

The analysis in this thesis could also be revised in the future. This can be in response to policy changes or technological developments. For example, the research conducted in this study could also be extended to hydrogen fuel cell vehicles. This thesis focused on plug-in electric vehicles, which have become commercially available in recent years. However, automakers are announcing the release of hydrogen fuel cell vehicles for sale (as opposed to for lease in limited quantities to meet minimum Zero Emission Vehicle program requirements) in the near future, such as the Toyota Mirai.173 As the environmental concerns over petroleum use and climate change grow over time, so too does the importance of understanding the life cycle implications of alternative light-duty vehicle technologies.

129

References

1. Transportation Energy Data Book. http://cta.ornl.gov/data/index.shtml (Accessed May 20, 2015). 2. Annual Energy Outlook. http://www.eia.gov/forecasts/aeo/er/index.cfm (Accessed June 15, 2015). 3. Advancing Technology for America’s Transportation Future. https://www.npc.org/FTF- 80112.html (Accessed June 15, 2015). 4. National Inventory Report 1990-2013: Greenhouse Gas Sources and Sinks in Canada - Executive Summary. http://www.ec.gc.ca/ges-ghg/default.asp?lang=En&n=5B59470C- 1&offset=3&toc=show (Accessed June 15, 2015). 5. Air Pollutant Emission Inventory. http://www.ec.gc.ca/pollution/default.asp?lang=En&n=E96450C4-1 (Accessed June 15, 2015). 6. Hidden Costs of Energy: Unpriced Consequences of Energy Production and Use; National Research Council: Washington, DC, 2010 7. GREET Model. https://greet.es.anl.gov/ (Accessed June 15, 2015). 8. Michalek, J. J.; Chester, M.; Jaramillo, P.; Samaras, C.; Shiau, C. S. N.; Lave, L. B., Valuation of plug-in vehicle life-cycle air emissions and oil displacement benefits. Proceedings of the National Academy of Sciences of the United States of America 2011, 108 (40), 16554-16558. 9. Corporate Average Fuel Economy. http://www.nhtsa.gov/fuel-economy (Accessed June 15, 2015). 10. Zero Emission Vehicle (ZEV) Program. http://www.arb.ca.gov/msprog/zevprog/zevprog.htm (Accessed June 15, 2015). 11. Renewable Fuel Standard. http://www.epa.gov/oms/fuels/renewablefuels/ (Accessed June 15, 2015). 12. Low Carbon Fuel Standard Program. http://www.arb.ca.gov/fuels/lcfs/lcfs.htm (Accessed June 15, 2015). 13. Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975 Through 2013. http://www.epa.gov/oms/fetrends.htm (Accessed June 15, 2015).

130

14. Regulating Greenhouse Gas Emissions from Light-Duty Vehicles (2017-2025). http://www.ec.gc.ca/default.asp?lang=En&n=56D4043B-1&news=1F13DA8A-EB01- 4202-AA6B-9E1E49BBD11E (Accessed June 15, 2015). 15. www.fueleconomy.gov. http://www.fueleconomy.gov/ (Accessed June 15, 2015). 16. Qualified Plug-In Electric Drive Motor Vehicle Tax Credit. http://www.afdc.energy.gov/laws/409 (Accessed June 15, 2015). 17. Electric Vehicle Rebate. http://www.mto.gov.on.ca/english/vehicles/electric/electric- vehicle-rebate.shtml (Accessed June 15, 2015). 18. Purchase or Lease Rebate Program. http://vehiculeselectriques.gouv.qc.ca/english/particuliers/rabais.asp (Accessed June 15, 2015). 19. Renewable Fuels Regulations. http://www.ec.gc.ca/lcpe- cepa/eng/regulations/detailreg.cfm?intReg=186 (Accessed June 15, 2015). 20. Renewable & Low Carbon Fuel Requirements Regulation. http://www.empr.gov.bc.ca/RET/RLCFRR/Pages/default.aspx (Accessed June 15, 2015). 21. ISO 14040: 2006 Environmental management -- Life cycle assessment -- Principles and framework. http://www.iso.org/iso/catalogue_detail?csnumber=37456 (Accessed June 15, 2015).

22. Choudhary, S.; Liang, S.; Cai, H.; Keoleian, G.; Miller, S.; Kelly, J.; Xu, M., Reference and functional unit can change bioenergy pathway choices. The International Journal of Life Cycle Assessment 2014, 19 (4), 796-805.

23. Campbell, J. E.; Lobell, D. B.; Field, C. B., Greater Transportation Energy and GHG Offsets from Bioelectricity Than Ethanol. Science 2009, 324 (5930), 1055-1057.

24. Lewis, A. M.; Kelly, J.; Keoleian, G., Vehicle lightweighting vs. electrification: Life cycle energy and GHG emissions results for diverse powertrain vehicles. Applied Energy 2014, 126 (2014), 13-20. 25. Scrappage Rate Hits Historic High, Bodes Well for Future. http://wardsauto.com/sales- amp-marketing/scrappage-rate-hits-historic-high-bodes-well-future (Accessed June 15, 2015).

131

26. eGRID. http://www.epa.gov/cleanenergy/energy-resources/egrid/ (Accessed January 7, 2014).

27. Laser, M.; Lynd, L. R., Comparative efficiency and driving range of light- and heavy- duty vehicles powered with biomass energy stored in liquid fuels or batteries. Proceedings of the National Academy of Sciences 2013, 111 (9), 5.

28. Curran, S. J.; Wagner, R. M.; Graves, R. L.; Keller, M.; Green, J. B., Well-to-wheel analysis of direct and indirect use of natural gas in passenger vehicles. Energy 2014, 75 (1), 194-203.

29. Corporate Average Fuel Economy for MY 2017-MY 2025 Passenger Cars and Light Trucks Final Regulatory Impact Analysis. www.nhtsa.gov/staticfiles/rulemaking/pdf/cafe/FRIA_2017-2025.pdf (Accessed June 15, 2015).

30. Knittel, C. R., Automobiles on Steroids: Product Attribute Trade-Offs and Technological Progress in the Automobile Sector. The American Economic Review 2011, 101 (7), 3368-3399.

31. Effectiveness and Impact of Corporate Average Fuel Economy Standards. http://www.nap.edu/catalog/10172/effectiveness-and-impact-of-corporate-average-fueleconomy-cafe-standards (Accessed June 15, 2015).

32. Anderson, S. T.; Sallee, J. M., Using Loopholes to Reveal the Marginal Cost of Regulation: The Case of Fuel-Economy Standards. American Economic Review 2011, 101 (4), 1375-1409.

33. An, F.; DeCicco, J., Trends in Technical Efficiency Trade-Offs for the U.S. Light Vehicle Fleet. In SAE World Congress & Exhibition, Detroit, 2007.

34. Bandivadekar, A.; Cheah, L.; Evans, C.; Groode, T.; Heywood, J.; Kasseris, E.; Kromer, M.; Weiss, M., Reducing the fuel use and greenhouse gas emissions of the US vehicle fleet. Energy Policy 2008, 36 (2008), 2754-2760.

132

35. Cheah, L.; Heywood, J., Meeting U.S. passenger vehicle fuel economy standards in 2016 and beyond. Energy Policy 2011, 2011 (39), 13.

36. Makino, K., Advanced Requirements for Fuel Efficient Cars by Creating Efficient Body. In SAE World Congress, Detroit, 2011.

37. Shiau, C.-S. N.; Michalek, J. J.; Hendrickson, C. T., A structural analysis of vehicle design responses to Corporate Average Fuel Economy policy. Transportation Research Part A 2009, 43 (2009), 814-828.

38. Whitefoot, K. S.; Skerlos, S. J., Design incentives to increase vehicle size created from the U.S. footprint-based fuel economy standards. Energy Policy 2012, 41, 402-411.

39. Bedsworth, L.; Taylor, M., Learning from California's Zero-Emission Vehicle Program. California Economic Policy 2007, 3 (4), 1-19.

40. Barry, K., Zero-Emission Vehicle Regulations Get Tougher for 2012. Car and Driver 2010.

41. Kromer, M. A.; Heywood, J. B. Electric Powertrains: Opportunities and Challenges in the U.S. Light-Duty Vehicle Fleet Massachusetts Institute of Technology: 2007.

42. Multi-Path Transportation Futures Study: Vehicle Characterization and Scenario Analyses. http://www.transportation.anl.gov/technology_analysis/multipath_study.html (Accessed June 15, 2015). 43. Assessment of Technologies for Improving Light Duty Vehicle Fuel Economy. http://www.nap.edu/catalog/12924/assessment-of-fuel-economy-technologies-for-light- duty-vehicles (Accessed June 15, 2015). 44. Fiat loses $14k on every 500e it builds, Marchionne doesn't want you to buy one. http://www.autoblog.com/2014/05/22/fiat-loses-14k-every-500e-builds-marchionne/ (Accessed June 15, 2015). 45. State Laws and Incentives. http://www.afdc.energy.gov/laws/state (Accessed June 15, 2015).

133

46. Farrell, A.; Plevin, R.; Turner, B.; Jones, A.; O'Hare, M.; Kammen, D., Ethanol Can Contribute to Energy and Environmental Goals. Science 2006, 311 (5760), 506-508.

47. Wang, M.; Wu, M.; Huo, H., Life-cycle energy and greenhouse gas emission impacts of different corn ethanol plant types. Environmental Research Letters 2007, 2 (2).

48. Searchinger, T.; Heimlich, R.; Houghton, R.; Dong, F.; Elobeid, A.; Fabiosa, J.; Tokgoz, S.; Hayes, D.; Yu, T., Use of U.S. Croplands for Biofuels Increases Greenhouse Gases Through Emissions from Land-Use Change. Science 2008, 319 (5867), 1238-1240.

49. Brown, T.; Brown, R., A review of cellulosic biofuel commercial-scale projects in the United States. Biofuels, Bioproducts and Biorefining 2013, 7 (3), 235-245.

50. Zhang, Y.; Joshi, S.; MacLean, H., Can ethanol alone meet California's low carbon fuel standard? An evaluation of feedstock and conversion alternatives. Environmental Research Letters 2010, 5 (1).

51. Kaufman, A.; Meier, P.; Sinistore, J.; Reinemann, D., Applying life-cycle assessment to low carbon fuel standards—How allocation choices influence carbon intensity for renewable transportation fuels. Energy Policy 2010, 38 (9), 5229-5241.

52. Yeh, S., Lutsey, N., Parker, C. Assessment of Technologies to Meet a Low Carbon Fuel Standard. Environmental Science & Technology 2009, 43 (18), 6907-6914

53. Andress, D.; Nguyen, D. T.; Das, S., Reducing GHG emissions in the United States' transportation sector. Energy for Sustainable Development 2011, 15 (2), 117-136.

54. Sperling, D.; Yeh, S. Low Carbon Fuel Standards; UC Davis: 2009.

55. Venkatesh, A.; Jaramillo, P.; Griffin, W. M.; Matthews, H. S., Uncertainty in Life Cycle Greenhouse Gas Emissions from United States Natural Gas End-Uses and its Effects on Policy. Environmental Science & Technology 2011, 45 (19), 8182-8189.

134

56. Mullins, K. A.; Griffin, M. W.; Mathews, S. H., Policy Implications of Uncertainty in Modeled Life-Cycle Greenhouse Gas Emissions of Biofuels. Environmental Science & Technology 2011, 45 (1), 7.

57. Kocoloski, M.; Mullins, K. A.; Venkatesh, A.; Griffin, M. W., Addressing uncertainty in life-cycle carbon intensity in a national low-carbon fuel standard. Energy Policy 2013, 56, 10.

58. DeCicco, J., Factoring the car-climate challenge: Insights and implications. Energy Policy 2013, 59 (2013), 382-392.

59. Glossary: Terms and Acronyms. http://www.epa.gov/otaq/imports/glossary.htm#ldv (Accessed June 15, 2015). 60. Alternative Fuels Data Center. http://www.afdc.energy.gov/ (Accessed June 15, 2015). 61. How many gas stations are there in the U.S? http://www.fueleconomy.gov/feg/quizzes/answerQuiz16.shtml (Accessed June 15, 2015). 62. Alternative Fuels Data Center. http://www.afdc.energy.gov/ (Accessed June 15, 2015). 63. Edmunds. http://www.edmunds.com/ (Accessed June 15, 2015). 64. Electric Cars: Some Are Real, Most Are Only 'Compliance Cars' -- We Name Names. http://www.greencarreports.com/news/1068832_electric-cars-some-are-real-most-are- only-compliance-cars--we-name-names (Accessed June 15, 2015).

65. Sullivan, J.; Williams, R.; Yester, S.; Cobas-Flores, E.; Chubbs, S.; Hentges, S.; Pomper, S. In Life cycle inventory of a generic U.S. family sedan overview of results USCAR AMP Project, Total Life Cycle Conference and Exposition, Graz, Austria, Society of Automotive Engineers: Graz, Austria, 1998.

66. MacLean, H.; Lave, L., A life-cycle model of an automobile. Environmental Science & Technology 1998, 32 (13), 322-330.

67. Kim, H.; Wallington, J., Life-cycle energy and greenhouse gas emission benefits of lightweighting in automobiles: review and harmonization. Environmental Science & Technology 2013, 47 (12), 6089-6097.

135

68. Colett, J.; Kelly, J.; Keoleian, G., Using nested average electricity allocation protocols to characterize electrical grids in life cycle assessments. Journal of Industrial Ecology 2015, doi: 10.1111/jiec.12268.

69. MacLean, H.; Lave, L., Evaluating automotive fuel/propulsion system technologies. Progress in Energy and Combustion Science 2003, 29 (1), 1-69.

70. Hauenstein, H.; Schewel, L. Dust to Dust's Assumptions About the Prius and the Hummer; Boulder, CO, 2007.

71. Samaras, C.; Meisterling, K., Life cycle assessment of greenhouse gas emissions from plug-in hybrid vehicles: Implications for policy. Environmental Science & Technology 2008, 42 (9), 3170-3176.

72. MacPherson, N.; Keoleian, G.; Kelly, J., Fuel Economy and Greenhouse Gas Emissions Labeling for Plug-in Hybrid Vehicles from a Life Cycle Perspective. Journal of Industrial Ecology 2012, 16 (5), 761-773.

73. Kelly, J.; MacDonald, J.; Keoleian, G., Time-dependent plug-in hybrid electric vehicle charging based on national driving patterns and demographics. Applied Energy 2012, 94 (2012), 395-405.

74. Kennedy, C., Key threshold for electricity emissions. Nature Climate Change 2015, 5 (3), 179-181.

75. Tessum, C. W.; Hill, J. D.; Marshall, J. D., Life cycle air quality impacts of conventional and alternative light-duty transportation in the United States. Proceedings of the National Academies of Science 2014, 111 (52), 18490-18495.

76. Raykin, L.; MacLean, H. L.; Roorda, M. J., Implications of Driving Patterns on Well-to- Wheel Performance of Plug-in Hybrid Electric Vehicles. Environmental Science & Technology 2012, 46 (11), 6363-6370.

77. Karabasoglu, O.; Michalek, J., Influence of driving patterns on life cycle cost and emissions of hybrid and plug-in electric vehicle powertrains. Energy Policy 2013, 60, 445-461.

136

78. Yuksel, T.; Michalek, J., Effects of Regional Temperature on Electric Vehicle Efficiency, Range, and Emissions in the United States. Environmental Science & Technology 2015.

79. Nordelof, A.; Messagie, M.; Tillman, A.-M.; Soderman, M. L.; Mierlo, J. V., Environmental impacts of hybrid, plug-in hybrid, and battery electric vehicles - what can we learn from life cycle assessment? The International Journal of Life Cycle Assessment 2014, 19 (11), 25.

80. Hill, J.; Polasky, S.; Nelson, E.; Tilman, D.; Huo, H.; Ludwig, L.; Neumann, J.; Zheng, H. C.; Bonta, D., Climate change and health costs of air emissions from biofuels and gasoline. Proceedings of the National Academy of Sciences of the United States of America 2009, 106 (6), 2077-2082.

81. Jacobson, M., Effects of Ethanol (E85) versus Gasoline Vehicles on Cancer and Mortality in the United States. Environmental Science & Technology 2007, 41 (11), 4150-4157.

82. Nopmongcol, U.; Griffin, M.; Greg, Y.; Dunker, A.; MacLean, H.; Mansell, G.; Grant, J., Impact of dedicated E85 vehicle use on ozone and particulate matter in the US. Atmospheric Environment 2011, 45 (39), 7330-7340.

83. Laser, M.; Larson, E.; Dale, B.; Wang, M.; Greene, N.; Lynd, L. R., Comparative analysis of efficiency, environmental impact, and process economics for mature biomass refining scenarios. Biofuels Bioproducts & Biorefining-Biofpr 2009, 3 (2), 247- 270.

84. Searcy, E.; Flynn, P., A criterion for selecting renewable energy processes. Biomass and Bioenergy 2010, 34, 798-804.

85. Clarens, A. F.; Nassau, H.; Resurreccion, E. P.; White, M. A.; Colosi, L. M., Environmental Impacts of Algea-Derived Biodiesel and Bioelectricity for Transportation. Environmental Science & Technology 2011, 45 (17), 7554-7560.

137

86. Chevy Bolt. http://www.chevrolet.com/culture/article/bolt-ev-concept-car.html (Accessed June 15, 2015).

87. Howarth, R.; Santoro, R.; Ingraffea, A., Methane and the greenhouse-gas footprint of natural gas from shale formations. Climate Change 2011, 106 (4), 679-690.

88. Allen, D. T.; Torres, V. M.; Thomas, J.; Sullivan, D. W.; Harrison, M.; Hendler, A.; Herndon, S. C.; Kolb, C. E.; Fraser, M. P.; Hill, A. D.; Lamb, B. K.; Miskimins, J.; Sawyer, R. F.; Seinfeld, J. H., Measurements of methane emissions at natural gas production sites in the United States. Proceedings of the National Academy of Sciences of the United States of America 2013, 110 (44), 17768-17773.

89. Laurenzi, I. J.; Jersey, G. R., Life Cycle Greenhouse Gas Emissions and Freshwater Consumption of Marcellus Shale Gas. Environmental Science & Technology 2013, 47 (9), 4896-4903.

90. Darrah, T.; Vengosh, A.; Jackson, R.; Warner, R.; Poreda, R., Noble gases identify the mechanisms of fugitive gas contamination in drinking-water wells overlying the Marcellus and Barnett Shales. PNAS 2014, 111 (39), 14076-14081.

91. Burnham, A.; Han, J.; Clark, C. E.; Wang, M.; Dunn, J. B.; Palou-Rivera, I., Life-Cycle Greenhouse Gas Emissions of Shale Gas, Natural Gas, Coal, and Petroleum (vol 46, pg 619, 2012). Environmental Science & Technology 2012, 46 (13), 7430-7430.

92. Wang, M.; Huang, H. A full fuel-cycle analysis of energy and emissions impacts of transportation fuels produced from natural gas; Argonne, IL: 2000.

93. Dai, Q.; Lastoskie, C. M., Life Cycle Assessment of Natural Gas-Powered Personal Mobility Options. Energy & Fuels 2014, 28 (9), 5988-5997.

94. Granovskii, M.; Dincer, I.; Rosen, M., Economic and environmental comparison of conventional, hybrid, electric and hydrogen fuel cell vehicles. Journal of Power Sources 2006, 159 (2), 1186-1193.

138

95. Lee, D.; Thomas, V.; Brown, M., Electric Urban Delivery Trucks: Energy Use, Greenhouse Gas Emissions, and Cost-Effectiveness. Environmental Science & Technology 2013, 47 (14), 8022-8030.

96. Autonomie. www.autonomie.net (Accessed June 15, 2015). 97. APEEP. https://sites.google.com/site/nickmullershomepage/home/ap2-apeep-model-2 (Accessed June 15, 2015). 98. Crystal Ball. http://www.oracle.com/us/products/applications/crystalball/overview/index.html (Accessed June 15, 2015). 99. Finnveden, G.; Hauschild, M.; Ekvall, T.; Guinee, J.; Heijungs, R.; Hellweg, S.; Koehler, A.; Pennington, D.; Suh, S., Recent developments in Life Cycle Assessment. Journal of Enviornmental Management. 2009, 91 (1), 1-21. 100. Plevin, R.; Delucchi, M.; Creutzig, F., Using Attributional Life Cycle Assessment to Estimate Climate-Change Mitigation Benefits Misleads Policy Makers. Journal of Industrial Ecology 2014, 18 (1), 73-83.

101. Muller, N.; Mendelsohn, R., Measuring the damages of air pollution in the United States. Journal of Environmental Economics and Management 2007, 54 (1), 1-14.

102. Mashayekh, Y.; Jaramillo, P.; Chester, M.; Hendrickson, C.; Weber, C., Costs of Automobile Air Emissions in U.S. Metropolitan Areas. Transportation Research Record: Journal of the Transportation Research Board 2011, 2233 (2011), 120-127.

103. PSAT (Powertrain System Analysis Toolkit). http://www.transportation.anl.gov/modeling_simulation/PSAT/ (Accessed June 15, 2015). 104. Moawad, A.; Sharer, P.; Rousseau, A. Light-Duty Vehicle Fuel Consumption Displacement Potential up to 2045; Argonne National Laboratory: 2011.

105. Venkatesh, A.; Jaramillo, P.; Griffin, W. M.; Matthews, H. S., Uncertainty Analysis of Life Cycle Greenhouse Gas Emissions from Petroleum-Based Fuels and Impacts on Low Carbon Fuel Policies. Environmental Science & Technology 2011, 45 (1), 125-131.

139

106. Spatari, S.; MacLean, H., Characterizing model uncertainties in the life cycle of lignocellulose-based ethanol fuels. Environmental Science & Technology 2010, 44 (22), 8773-8780.

107. Renewable Portfolio Standards. http://www.epa.gov/agstar/tools/funding/renewable.html (Accessed June 15, 2015).

108. Pacca, S.; Moreira, J., A Biorefinery for Mobility? Environmental Science & Technology 2011, 45 (11), 9498-9505.

109. Lave, L.; MacLean, H.; Hendrickson, C.; Lankey, R., Life-Cycle Analysis of Alternative Automobile Fuel/Propulsion Technologies. Environmental Science & Technology 2000, 34 (17), 3598-3605.

110. Walsh, M.; de la Torre Ugarte, D.; Shapouri, H.; Slinsky, S., Bioenergy crop production in the United States. Environmental Resource Economics 2003, 24 (4), 313-333.

111. Sannigrahi, P.; Ragauskas, A.; Tuskan, G., Poplar a feedstock for biofuels: A review of compositional characteristics. Biofuels, Bioproducts and Biorefining 2010, 4 (2), 209- 226.

112. McKechnie, J.; Zhang, Y.; Ogino, A.; Saville, B.; Sleep, S.; Turner, M.; Pontius, R.; MacLean, H., Impacts of co-location, co-production and process energy source on life cycle energy use and greenhouse gas emissions of lignocellulosic ethanol. Biofuel, Bioproducts and Biorefining 2011, 5 (3), 279-292.

113. Zhang, Y.; McKechnie, J.; Cormier, D.; Lyng, R.; Mabee, W.; Ogino, A.; MacLean, A., Life cycle emissions and cost of production electricity from coal, natural gas and wood pellets in Ontario, Canada. Environmental Science and Technology 2010, 44 (1), 539- 544.

114. Havlik, P.; Schneider, U.; Schmid, E.; Bottcher, H.; Fritz, S.; Skalsky, R.; Aoki, K.; De Cara, S.; Kindermann, G.; Kraxner, F.; Leduc, S.; McCallum, L.; Mosnier, A.; Sauer, T.; Obersteiner, M., Global land-use implications of first and second generation biofuel targets. Energy Policy 2011, 39 (10), 5690-5702.

140

115. Aspen Plus. https://www.aspentech.com/products/aspen-plus.aspx (Accessed May 24, 2012). 116. Mascoma Our Facilities. http://www.mascoma.com/about-us/facilities/ (Accessed May 24, 2012).

117. Mosier, N.; Wyman, B.; Dale, R.; Elander, R.; Lee, Y.; Holtzapple, M.; Ladisch, M., Features of promising technologies for pretreatment of lignocellulosic biomass. Bioresource Technology 2005, 96 (6), 673-686.

118. Lange, J., Lignocellulosic conversion: an introduction to chemistry, process and economics. Biofuel, Bioproducts and Biorefining 2007, 1 (1), 373-403.

119. Saville, B., Pretreatment options. In Plant Biomass Conversion, Hood, E.; Nelson, P.; Powell, R., Eds. John Wiley & Sons Inc: Hoboken, 2011; pp 199-226. 120. Process Design and Economics for Conversion of Lignocellulosic Biomass to Ethanol: Thermochemical Pathway by Indirect Gasification and Mixed Alcohol Synthesis; National Renewable Energy Laboratory: Oak Ridge, TN, 2011. 121. Bain, R., Electricity from biomass in the United States: status and future direction. Bioresource Technology 1993, 46 (1-2), 86-93. 122. Process design and economics for biochemical conversion of lignocellulosic biomass to ethanol: dilute-acid pretreatment and enzymatic hydrolysis of corn stover; National Renewable Energy Laboratory: Oak Ridge, TN, 2011. 123. Fuel economy labeling of motor vehicles: revisions to improve calculation of fuel economy estimates; United States Environmental Protection Agency: Washington, DC, 2006. 124. Moore, W.; Foster, M.; Hoyer, K., Engine efficiency improvements enabled by ethanol fuel blends in a GDi VVA Flex Fuel Engine. In SAE World Congress, Detroit, 2011. 125. Chevrolet fuel solutions: fuel efficiency. http://archives.media.gm.com/archive/documents/domain_3/docId_41301_pr.html (Accessed May 20, 2015). 126. Lemoine, D.; Plevin, R.; Cohn, A.; Jones, A.; Brandt, A.; Vergara, S.; Kammen, D., The climate impacts of bioenergy systems depend on market and regulatory policy contexts. Environmental Science and Technology 2010, 44 (19), 7347-7350.

141

127. Ford electric vehicle historical highlights. http://media.ford.com/article_display.cfm?article_id=29678 (Accessed May 24, 2012). 128. 2000 Nissan Altra EV Performance Characterization. http://www1.eere.energy.gov/vehiclesandfuels/avta/pdfs/fsev/sce_rpt/altra_2000_report. pdf (Accessed May 20, 2015). 129. 1998-2001 Nissan Altima Review. http://www.consumerguide.com/nissan/altima/used/1998-2001/ (Accessed May 20, 2015). 130. Shen, T. Life cycle modeling of multi-product lignocellulosic ethanol systems. University of Toronto, Toronto, ON, 2012. 131. National Household Travel Survey. http://nhts.ornl.gov/tools.shtml (Accessed June 15, 2015). 132. Luk, J. M.; Pourbafrani, M.; Saville, B. A.; MacLean, H. L., Ethanol or bioelectricity? Life cycle assessment of lignocellulosic bioenergy use in light-duty vehicles. Environmental Science & Technology 2013, 47 (18), 10676-84. 133. Shiau, C. S. N.; Kaushal, N.; Hendrickson, C. T.; Peterson, S. B.; Whitacre, J. F.; Michalek, J. J., Optimal Plug-In Hybrid Electric Vehicle Design and Allocation for Minimum Life Cycle Cost, Petroleum Consumption, and Greenhouse Gas Emissions. Journal of Mechanical Design 2010, 132 (9) 1-11. 134. Toyota introduces next-generation B-segment hybrid with 112 mpg US; CNG hybrid and plug-in variants. http://www.greencarcongress.com/2012/03/ftbh-20120306.html (Accessed June 15, 2015). 135. Kumaranayake, L., The real and the nominal? Making inflationary adjustments to cost and other economic data. Health Policy and Planning 2000, 15 (2), 230-234. 136. National Economic Accounts. http://www.bea.gov/national/ (Accessed June 15, 2015). 137. Rood Werpy, M.; Santini, D.; Burnham, A.; Mintz, M. Natural Gas Vehicles: Status, Barriers and Opportunities; Argonne National Laboratory: 2010. 138. Cai, H.; Wang, M.; Elgowainy, A.; Han, J. Updated Greenhouse Gas and Criteria Air Pollutant Emission Factors and Their Probability Distribution Functions for Electric Generating Units; Argonne National Laboratory: 2012. 139. Tier 3 Vehicle Emission and Fuel Standards Program. http://www.epa.gov/otaq/tier3.htm (Accessed June 15, 2015).

142

140. Technical Update of the Social Cost of Carbon for Regulatory Impact Analysis. https://www.whitehouse.gov/sites/default/files/omb/inforeg/social_cost_of_carbon_for_ ria_2013_update.pdf (Accessed June 15, 2015). 141. MOVES (Motor Vehicle Emission Simulator). http://www.epa.gov/otaq/models/moves/ (Accessed June 15, 2015). 142. Plug-in Hybrid Electric Vehicle Value Proposition Study; Oak Ridge National Laboratory: Oak Ridge, TN, 2010. 143. Census Economic Database Search and Trend Charts. http://www.census.gov/main/www/access.html (Accessed January 7, 2014). 144. Carriere, A.; Kaufmann, C.; Shapiro, J.; Paine, P.; Prinsen, J., The contribution of Methanol Emissions from Windshield Washer Fluid Use to the formation of Ground- Level Ozone. SAE Technical Paper 2000, DOI: 10.4271/2000-01-0663. 145. Dinh, T.; Kim, S.; Son, Y.; Choi, I.; Park, S.; Young, S.; Kim, J., Emission characteristics of VOCs emitted from consumer and commercial products and their ozone formation potential. Environmental Science and Pollution Research 2015, 22 (12), 9345-9355. 146. Winebrake, J.; Wang, M.; He, D., Toxic Emissions from mobile sources: a total fuel- cycle analysis for conventional and alternative fuel vehicles. Journal of Air & Waste Management Association 2001, 51 (7), 1073-1086. 147. Los Angeles County Oil Refineries. http://www.laalmanac.com/energy/en16.htm (Accessed June 15, 2015). 148. Air Trends. http://www.epa.gov/airtrends/index.html (Accessed June 15, 2015). 149. Corporate Average Fuel Economy http://www.nhtsa.gov/fuel-economy (Accessed June 15, 2015).

150. Green, E. H.; Skerlos, S. J.; Winebrake, J. J., Increasing electric vehicle policy efficiency and effectiveness by reducing mainstream market bias. Energy Policy 2014, 65 (2014), 562-566.

151. One Million Electric Vehicles By 2015. https://www1.eere.energy.gov/vehiclesandfuels/pdfs/1_million_electric_vehicles_rpt.pd f (Accessed June 15, 2015).

143

152. National Ambient Air Quality Standards (NAAQS). http://www.epa.gov/air/criteria.html (Accessed June 15, 2015). 153. Clean Air Act. http://www.epa.gov/air/caa/ (Accessed June 15, 2015).

154. Muller, N. Z., The design of optimal climate policy with air pollution co-benefits. Resource and Energy Economics 2012, 34 (4), 696-722.

155. Assumptions to the Annual Energy Outlook 2010. http://www.eia.gov/oiaf/aeo/assumption/pdf/transportation_tbls.pdf (Accessed June 15, 2015). 156. Amatayakul, W.; Ramnäs, O., Life cycle assessment of a catalytic converter for passenger cars. Journal of Cleaner Production 2001, 9 (5), 395-403 157. Newcomer, A.; Blumsack, S. A.; Apt, J.; Lave, L. B.; Morgan, M. G., Short run effects of a price on carbon dioxide emissions from US electric generators. Environmental Science & Technology 2008, 42 (9), 3139-3144. 158. Sprei, F.; Karlsson, S., Energy efficiency versus gains in consumer amenities - an example of new cars sold in Sweden. Energy Policy 2013, 53 (February 2013), 490-499. 159. Chevrolet Equinox Sales Figures. http://www.goodcarbadcar.net/2011/01/chevrolet- equinox-sales-figures.html (Accessed June 15, 2015). 160. Chevrolet Equinox. http://www.thecarconnection.com/cars/chevrolet_equinox (Accessed June 15, 2015). 161. Army receives first six NEVs. http://www.army.mil/article/15751/army-receives-first- of-six-nevs/ (Accessed June 15, 2015). 162. Plug-In Electric Drive Vehicle Credit. http://www.irs.gov/Businesses/Plug-In-Electric- Vehicle-Credit-(IRC-30-and-IRC-30D) (Accessed June 15, 2015). 163. Small, K. A.; Van Dender, K., Fuel Efficiency and Motor Vehicle Travel: The Declining Rebound Effect. The Energy Journal 2007, 28 (1), 25-51. 164. Ahmad, S.; Greene, D. L., Effect of Fuel Economy on Automobile Safety. Transportation Research Record 2005, (1941), 8. 165. What is a Compliance Car? http://www.autoguide.com/auto-news/2014/03/compliance- car.html (Accessed June 15, 2015). 166. Sivak, M.; Schoettle, B., Update: Percentage of Young Persons With a Driver's License Continues to Drop. Traffic Injury Prevention 2014, 13 (4), 1.

144

167. 10 best cars for older drivers. http://www.consumerreports.org/cro/2012/08/best-cars- for-older-drivers/index.htm (Accessed June 15, 2015). 168. Coughlin, J.; D'Ambrosio, L., Aging America and Transportation. 1 ed.; Springer Publishing Company: New York, 2012; p 288. 169. Luk, J. M.; Saville, B. A.; MacLean, H. L., Life Cycle Air Emissions Impacts and Ownership Costs of Light-Duty Vehicles Using Natural Gas As a Primary Energy Source. Environmental Science & Technology 2015, 49 (8), 5151-5160. 170. Luk, J. M.; Saville, B. A.; MacLean, H. L., Life Cycle Air Emissions Impacts and Ownership Costs of Light-Duty Vehicles Using Natural Gas As a Primary Energy Source. Environmental Science & Technology 2015, 49 (8), 5151-5160.Kennedy, C., Key threshold for electricity emissions. Nature Climate Change 2015, 5, 179-181. 171. WRF-Chem. https://www2.acd.ucar.edu/wrf-chem (Accessed June 15, 2015).

172. Kim, H.-J.; Keoleian, G.; Skerlos, S., Economic assessment of greenhouse gas emissions reductions by vehicle lightweighting using aluminum and high-strength steel. Journal of Industrial Ecology 2010, 15 (1), 64-80.

173. Toyota Mirai. http://www.toyota.com/mirai/ (Accessed June 15, 2015).

174. NREL Lignocellulosic Biomass to Ethanol Process Design and Economics Utilizing Co-Current Dilute Acid Prehydrolysis and Enzymatic Hydrolysis Current and Futuristic Scenarios; National Renewable Energy Laboratory: Golden, CO, 1999.

175. MacLean, H.; Spatari, S., The contribution of enzymes and process chemicals to the life cycle of ethanol. Environmental Research Letters 2009, 4 (1), 014001.

176. Process design and cost estimate of critical equipment in the biomass to ethanol process; Harris Group Inc: Seattle, WA, 2001. 177. Jin, H., Larson, E., Celik, R. Performance and cost analysis of future, commercially mature gasification-based electric power generation from switchgrass. Biofuels, Bioproducts and Biorefining. 2009 3 (2) 142-173. 178. Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975 - 2010 Appendix A; United States Environmental Protection Agency: Washington, DC, 2010.

145

179. Well-to-Wheels Analysis of Energy Use and Greenhouse Gas Emissions of Plug-In Hybrid Electric Vehicles; Argonne National Laboratory: Argonne, IL, 2010. 180. 2015 Civic Natural Gas. http://automobiles.honda.com/civic-natural-gas/ (Accessed June 15, 2015). 181. Automobile Maintenance Costs, Used Cars, and Private Information. people.virginia.edu/~sns5r/resint/empiostf/maintainencecosts.pdf 182. Nissan Leaf. http://www.nissanusa.com/electric-cars/leaf/ (Accessed June 15, 2015). 183. Franke, T.; Krems, J. F., What drives range preferences in electric vehicle users? Transport Policy 2013, 30 (2013), 56-62. 184. Nissan Corporate Vice President Simon Sproule discusses Nissan LEAF battery pack with the Global Media Center. http://www.nissan- global.com/EN/REPORTS/2011/08/110803.html (Accessed June 15, 2015). 185. Energy Futures Backgrounder: Addendum to Canada’s Energy Future: Energy Supply and Demand Projections to 2035. https://www.neb- one.gc.ca/nrg/ntgrtd/ftr/archive/2012/index-eng.html (Accessed June 15, 2015). 186. Volt Introduces Revolutionary Voltec System. http://media.gm.com/media/us/en/gm/news.detail.html/content/Pages/news/au/en/2011/ Dec/1209_VoltIntroducesRevolutionaryVoltecSystem.html (Accessed June 15, 2015).

146

Appendix A: Chapter 4 Supporting Information Methods Section Details

Research Scope

Figure A-1 illustrates the system boundaries and resources used to develop the pathways outlined in Table 4-1. They consist of Well-to-Pump, Pump-to-Wheel and Vehicle Cycle processes introduced within the Methods section of Chapter 4 and discussed here. Not all processes shown are applicable for each pathway. Infrastructure for these pathways (e.g., bioenergy production and fueling facilities) is excluded, as the embedded energy is expected to be minor in comparison to the accounted for energy flows through the facilities throughout its life time. Land use change impacts are excluded from the main results, for reasons discussed within the Methods section of Chapter 4.

Life cycle energy use and GHG emissions are examined for the pathways. Biomass energy use accounts for the lignocellulosic feedstock used directly in the bioenergy pathways considered in our study, and excludes the minor use of biomass energy currently used for grid-electricity or gasoline production. Petroleum energy use includes gasoline and diesel use, and associated upstream losses. Fossil energy use includes petroleum, natural gas and coal energy. Total energy use is the sum of all energy use, including nuclear and non-biomass sources of renewable energy. Higher heating values are presented in this study to more accurately reflect the total energy use in each process. Results are presented based on a functional unit of 100 vehicle kilometer travelled (VKT), which is commonly used to compare vehicle fuel consumption.

147

Direct & Indirect Land Biomass Production Infrastructure Legend Use Change (GREET 1 2012) Construction Well-to- Pump Chemical Production Bioenergy Production Reference Fuel Production (MacLean & Spatari 2009) (AspenPlus) (GREET 1 2012 rev 2) Pump-to-

Wheel

Bioenergy Delivery (GREET 1 2012 rev 2) Vehicle Cycle

Vehicle Production Vehicle Fuel Consumption Vehicle Disposal Out of (GREET 2 2012 rev 1) (Autonomie) (GREET 2 2012 rev 1) Scope

Figure A-1: Pathway System Boundaries

Process 1: Biomass Production

Our study uses the latest release of the GREET Fuel-Cycle model7 to obtain data for tree farming, which is used as a proxy for hybrid poplar short rotation forestry. This information consists of impacts as a result of onsite energy use and feedstock transportation as well as from fertilizer (nitrogen, phosphorus, potassium and calcium) and pesticide (herbicide and insecticide) use. The relevant quantity, GHG emission and energy use data are presented in Table A-1. As discussed within the Methods section of Chapter 4, both our study and the latest release of the GREET Fuel-Cycle model7 do not consider the effects of land use change related to tree farming. The assumed proximate analysis, ultimate analysis and biochemical components of the delivered whole tree hybrid poplar are detailed in Table A-2, obtained from the NREL databank of biomass components.174

148

Table A-1: Biomass production data from the GREET Fuel-Cycle model7

Pesticide Fertilizer Production Production Tree Herb- Insect- Wood Farming N P2O5 K2O CaCO3 icide icide Delivery Total Quantity 1000 kg 3.0 kg 1.0 kg 2.0 kg 2.4 kg 0.2 kg 0.0 kg 50 km n/a CH4 (g) 34 62 1 3 0 8 0 21 131 N2O (g) 0 5 0 0 0 0 0 0 6 CO2 (kg) 23 8 1 1 0 3 0 14 50 GHG (kg CO2eq.) 24 12 1 1 0 3 0.0 15 55 Petroleum Use (MJ) 273 5 4 5 0 19 0 171 478 Fossil Energy Use (MJ) 308 146 8 15 0 39 0 193 709 Total Energy Use (MJ) 309 148 8 17 0 41 0 193 717 Note: All values are per dry t biomass

Table A-2: Physical characteristics for hybrid poplar

Proximate Analysis Moisture Content 50% Volatile Matter 84% Fixed Carbon 15% Ash 0.87% Ultimate Analysis Carbon 51% Hydrogen 6.0% Oxygen 42% Nitrogen 0.17% Sulfur 0.09% Ash 0.92% Higher Heating Value 20 MJ/kg Biochemical Components Cellulose 47.3% Hemicellulose 20.9% Lignin 24.8%

149

Process 2: Chemical Production

A variety of chemical inputs are required for ethanol production. Data in Table A-3 from MacLean and Spatari175 are used as an estimate of the GHG and energy use impacts of the production of these chemicals. Although MacLean and Spatari175 discuss the uncertainty of these values, they are assumed to have a negligible effect on the final results of our study, due to the limited quantities of the chemicals used. The quantity of these chemicals required to produce ethanol is calculated within AspenPlus115 models, as discussed in the following section.

Table A-3: Chemical production data from MacLean and Spatari175

Sul- Hy- Am- Lime Sodium Per- Diam- Cellulase phuric drated monia Hydr- oxide monium Acid Lime oxide Phosphate H2SO4 Ca(OH)2 NH3 CaO NaOH H2O2 GHG Emissions (g CO2eq.) 130 930 1740 1240 1450 3900 640 2300 Petroleum Use (MJ) 3 1 1 1 1 1 0 2 Fossil Energy Use (MJ) 4 4 27 6 16 4 9 25 Total Energy Use (MJ) 6 5 28 7 17 5 9 27 Note: All values are per kg chemical

150

Process 3: Bioenergy Production

The bioenergy production models explained in the Methods section of Chapter 4 are further described here. These include base case and future models for both ethanol and bio-electricity production, developed using Aspen Plus115 to obtain mass and energy balances. Key assumptions for the models are compiled in Table A-4, while Table A-5 summarizes subroutines used in AspenPlus4 to simulate process operations in different configurations.

Table A-4: Bioenergy production data

Base case Future Ethanol Parameters Cellulose conversion 87% 95% Hemicellulose conversion 95% 96% Glucose fermentation yield 95% 97% Xylose fermentation yield 85% 97% Enzymatic loading 8 mg/g cellulose 8 mg/g cellulose Ethanol recovery 99.9% 99.9% Steam Electricity Generation Parameters Boiler efficiency 85% 85% Steam turbine/Electrical generator efficiency 37.5% 37.5% Syngas Electricity Generation Parameters Gasification efficiencies n/a 77.5% Gas turbine/Electrical generator efficiency n/a 36.6%

Table A-5: Aspen115 subroutines used to develop production models

Process Equipment AspenPlus115 Subroutine Pretreatment, Enzymatic hydrolysis and Fermenter Rstoic Solid-Liquid Seperation SEP Distillation Columns Radfrac Evaporators Flash2 + heatx Pumps, Compressor, Valve Pump, Compressor, Valve Burner RGibbs Gasification Unit RYield, RGibbs, Rstoic Gas Turbine RGibbs, Turbine Steam Turbine Turbine

151

Process Descriptions

Base Case Ethanol Production

Mascoma has developed a process for production of bioenergy from hybrid poplar.116 In the modeling of the ethanol process, we used Mascoma’s pilot data to define pretreatment, hydrolysis and fermentation reaction yields. In this process, the biomass is steam exploded at a temperature about 200° C for 8 min without adding any chemicals, and then the slurry is sent to enzymatic hydrolysis reactors where the carbohydrate polymers are broken down in to their corresponding sugar monomers. The temperature of the hydrolysis reactors is kept at 50° C for 48 h. The conversion of cellulose and hemicellulose to their monomer sugars is 87% and 95%, respectively. The slurry leaving the hydrolysis reactor is filtered and the solid residues (lignin) are separated. The fermentable sugars (glucose and xylose) content of hydrolyzate is converted to ethanol in fermenters which operate at 30°C. The conversion of glucose and xylose to ethanol is 95 and 85%, respectively. The fermenter products are distilled and ethanol is purified. The ethanol yield is 313 L of ethanol/ dry t biomass. The stillage leaving distillation columns is concentrated in multistage evaporators. The separated lignin is combusted to generate electricity and steam in a combined heat and power plant. The heat and power plant consists of a burner, boiler and turbogenerator subsystem. The generated steam in the heat and power plant is utilized in pretreatment, distillation and evaporation stages. The electricity generated is sufficient to satisfy the electricity requirement of the bioethanol process, and excess electricity is exported. Process flows are detailed in Figure A-2 and Table A-6.

152

Hybrid 1 3 Enzymatic 4 Filteration/ 5 Heat and Power Poplar Pretreatment Hydrolysis Washing Plant

6 2 Steam Fermentation

8 Distillation Ethanol

Solid Fuel To Heat Evaporation 10 and Power Plant

Figure A-2: Flow diagram of the ethanol production process

Note: Stream numbers refer to those itemized in Table A-6 and Table A-7

Table A-6: Base case ethanol production model material flow balance

Stream Number 1 2 3 4 5 6 7 8 9 10 11 Cellulose (t∕h) 9.4 0 9.2 0.8 0.8 0 0 0 0 0 0 Hemicellulose (t∕h) 4.2 0 2.2 1.0 1.0 0 0 0 0 0 0 C5 sugars (t∕h) 0 0 1.6 3.0 0.2 2.8 0.3 0 0.3 0.3 0 C6 sugars (t∕h) 0 0 0 9.3 1.0 8.3 0 0 0 0 0 Ethanol (t∕h) 0 0 0 0 0 0 5.3 5.3 0 0 0 Lignin (t∕h) 4.9 0 4.9 4.9 4.9 0 0 0 0 0 0 Water (t∕h) 19.9 11.5 22.0 61.7 3.2 58.5 60.7 0.0 60.1 3.4 6.9 CO2 (t/h) 0 0 0 0 0 0 0 0 0 0 12.7 N2O (t/h) 0 0 0 0 0 0 0 0 0 0 0.0006 CH4(t/h) 0 0 0 0 0 0 0 0 0 0 0 N2(t/h) 0 0 0 0 0 0 0 0 0 0 52.8 O2(t/h) 0 0 0 0 0 0 0 0 0 0 3.9 Others (t∕h) 2.8 0 2.2 4.5 1.1 3.2 3.7 0 3.7 3.4 1.7 Total (t∕h) 41.2 11.5 42.1 85.2 12.2 72.8 70 5.3 64.1 7.1 78 Note: Stream numbers refer to those illustrated in Figure A-2. Stream 11 represents the flue gas.

153

Future Ethanol Production

In the future ethanol production scenario, the cellulose conversion is expected to increase to 95% and the lignin leaving the filtration stage is washed with the water coming to enzymatic hydrolysis reactor. High sugar recovery (97%) is achieved by washing the lignin.176 The sugar recovery and higher conversion of cellulose increase the ethanol yield of the process to 355 L/dry t biomass. Process flows are detailed in Figure A-2 and Table A-7.

Table A-7: Future ethanol production model material flow balance

Stream Number 1 2 3 4 5 6 7 8 9 10 11 Cellulose (t∕h) 9.4 0 9.2 0.5 0.5 0 0 0 0 0 0 Hemicellulose (t∕h) 4.2 0 2.2 1 1 0 0 0 0 0 0 C5 sugars (t∕h) 0 0 1.6 3.0 0.1 2.9 0.4 0 0.4 0.4 0 C6 sugars (t∕h) 0 0 0 9.3 0.0 9.3 0 0 0 0 0 Ethanol (t∕h) 0 0 0 0 0 0 6.0 6.0 0 0 0 Lignin (t∕h) 4.9 0 4.9 4.9 4.9 0 0 0 0 0 0 Water (t∕h) 19.9 11.5 22.0 61.7 3.2 58.5 60.7 0.0 60.1 3.4 6.2 CO2 (t/h) 0 0 0 0 0 0 0 0 0 0 11.3 N2O (t/h) 0 0 0 0 0 0 0 0 0 0 0.0005 CH4(t/h) 0 0 0 0 0 0 0 0 0 0 0 N2(t/h) 0 0 0 0 0 0 0 0 0 0 47 O2(t/h) 0 0 0 0 0 0 0 0 0 0 3.5 Others (t∕h) 2.8 0 2.2 4.5 1.1 3.2 3.7 0 3.7 3.4 1.5 Total (t∕h) 41.2 11.5 42.1 84.9 10.8 73.9 70.8 6.0 64.2 7.2 69.5 Note: Stream numbers refer to those illustrated in Figure A-2. Stream 11 represents the flue gas.

154

Base Case Bio-electricity Production

In this configuration, the biomass is sent to the heat and power plant, and is combusted to produce steam and electricity (Figure A-3 and Table A-8). The heat and power plant was described in the near term ethanol production scenario. A small part of the generated electricity (3%) is utilized internally.122 The excess electricity is assumed to displace U.S average grid electricity. All calculations for the near-term bioelectricity production were performed based on the model reported by NREL.122

Steam

Hybrid 1 2 Poplar Burner Boiler Turbine/Generator

Electricity

Figure A-3: Flow diagram of the base case bio-electricity production process

Table A-8: Base case bio-electricity production model material flow balance

Stream Number 1 2 Cellulose (t∕h) 9.4 0 Hemicellulose (t∕h) 4.2 0 Lignin (t∕h) 4.9 0 Water (t∕h) 19.9 50 CO2 (t/h) 0 61.4 N2O (t/h) 0 0.003 CH4(t/h) 0 0 N2(t/h) 0 198 O2(t/h) 0 9.2 Others (t∕h) 2.8 5.0 Total (t∕h) 41.2 323.6 Note: Stream numbers refer to those illustrated in Figure A-3. Stream 2 represents the flue gas.

155

Future Bio-electricity Production

For the gasification process, the model of Jin et al.177 was considered. However, due to the high initial moisture content of hybrid poplar, a dryer is added before the gasifier. In this configuration, the biomass is dried and its moisture content is reduced to 20% to be utilized in the gasifier; the dried biomass is then sent to an oxygen-blown gasifier. The biomass is converted to syngas, and is used to generate electricity and steam. The pressure of gasifier is 30 bar177 and the temperature of the product syngas is 800°C. The oxygen and nitrogen required in gasification are provided by employing an air separation unit. The generated syngas is cooled to 350°C and is cleaned before it is sent to a gas turbine. The electricity generated by steam and gas turbines is 29 and 71% of total generated electricity, respectively. About 5% of the generated electricity is used internally. The drying process utilizes 26 % of the total generated steam. The excess electricity is exported to the US grid. Process flows are detailed in Figure A-4 and Table A-9.

3 Electricity

Hybrid 1 2 6 7 Poplar Drying Gasification Syngas Cooling Gas Turbine

4 5 8 Steam Turbine Electricity Air Separation Unit

Steam

Figure A-4: Flow diagram of the high efficiency bio-electricity production process

156

Table A-9: Future bio-electricity production model material flow balance

Stream Number 1 2 3 4 5 6 7 8 CO (t∕h) 0 0 0 0 0 68.0 68.0 0 CO2 (t∕h) 0 0 0 0 0 180.5 180.5 333 CH4 (t∕h) 0 0 0 0 0 21.8 21.8 0 N2O (t/h) 0 0 0 0 0 0 0 0 H2 (t∕h) 0 0 0 0 0 7.0 7.0 0 N2 (t∕h) 0 0 0 18.5 2.0 20.8 20.8 1424 O2 (t∕h) 0 0 0 0.2 79.9 0 0 251.7 C2H4 (t∕h) 0 0 0 0 0 0.7 0.7 0 C2H6 (t∕h) 0 0 0 0 0 0.8 0.8 0 H2S (t∕h) 0 0 0 0 0 0.2 0.2 0 Tar (t∕h) 0 0 0 0 0 0.2 0.2 0 Cellulose (t∕h) 83.4 83.4 0 0 0 0 0 0 Hemicellulose (t∕h) 37.3 37.3 0 0 0 0 0 0 C5 sugars (t∕h) 0 0 0 0 0 0 0 0 C6 sugars (t∕h) 0 0 0 0 0 0 0 0 Ethanol (t∕h) 0 0 0 0 0 0 0 0 Lignin (t∕h) 43.5 43.5 0 0 0 0 0 0 Water (t∕h) 188.9 47.2 49.8 0 0 84.8 84.8 199.1 Others (t∕h) 24.8 24.8 0 0 0 2.4 2.4 26 Total (t∕h) 377.8 236.2 49.8 18.7 81.9 387.2 387.2 2233.8 Temperature (oC) 15 100 236 205 192 764 350 90 Pressure (bar) 1 1 31.6 34.3 31.4 28.8 26.8 1 Note: Stream numbers refer to those illustrated in Figure A-4. Stream 8 represents the flue gas.

157

Process 4: Bioenergy Delivery

Ethanol delivery data are obtained from the GREET Fuel-Cycle model.7 The data are an estimate for ethanol produced within the U.S. and used domestically as a transportation fuel. In Table A- 10, “Ethanol Transportation” assumes that 40% of the fuel leaves the ethanol plant by barge for a distance of 520 miles (837 km), 40% by rail for a distance of 800 miles (1287 km) and 20% by truck for a distance of 80 miles (129 km). “Ethanol Distribution” consists of a 30 mile (48 km) truck transport from a bulk terminal to fueling stations.

Bio-electricity losses are assumed to be delivered to the end user via the U.S. electricity grid. Therefore, transmission and distribution losses are assumed to be equal to those of grid- electricity losses. The value of 8% is obtained from the GREET Fuel-Cycle model.7

Table A-10: Ethanol delivery data from the GREET Fuel-Cycle model7

Ethanol Transportation Ethanol Distribution Total Average Distance (km) 875 48 923 CH4 (g/ 1,000,000 L) 40 9 49 N20 (g/ 1,000,000 L) 1 0 1 CO2 (g/ 1000 L) 28 6 34 GHG Emissions (g CO2eq./1000 L) 29 6 35 Petroleum Use (MJ/1000 L) 344 73 417 Fossil Energy Use (MJ/1000 L) 314 66 379 Total Energy Use (MJ/1000 L) 345 74 418

158

Process 5: Reference Fuel Production

Gasoline and grid-electricity characteristics are assumed to be representative of the U.S. average. The data in Table A-11 are obtained from the GREET Fuel-Cycle model.7 Data for the year 2015 and 2020 are used for the near term and future scenarios, respectively. The different values for the two time frames are due to expected changes in legislation, resource mix and process efficiencies.

The GHG emissions and energy inputs for gasoline increase over time. The proportion of oil sands derived crude increases as conventional resources deplete and technology improves. Other changes include improved oil sands recovery efficiencies.

The GHG emissions and energy inputs for grid-electricity marginally decrease over time. Average electricity generation efficiencies improve, with increased prevalence of combined cycle (CC) technologies in both coal and natural gas fuelled facilities. The resource mix, shown in Table A-12, changes moderately, with fossil energy use declining from 66% to 65% of total energy use and petroleum remaining at 1%.

This data are a primary source in the development of the reference pathways, but are also used in bioenergy pathways. Gasoline production is also present within the ethanol pathways, to model the use of E85. Grid-electricity production is also introduced to the ethanol pathways, in the form of the co-product credit.

Table A-11: Reference fuel production data from the GREET Fuel-Cycle model7

Gasoline (/L) Grid-Electricity (/kWh) CH4 (g) 5 2 N20 (g) 0 0 CO2 (g) 541 569 GHG Emissions (g CO2eq.) 655 605 Petroleum Use (MJ) 2 0 Fossil Energy Use (MJ) 7 7 Total Energy Use (MJ) 7 8

Table A-12: Grid-electricity resource mix from the GREET Fuel-Cycle model7

Residual oil 1% Natural gas 26% Coal 40% Nuclear power 21% Biomass 0% Others 12%

159

Process 6: Vehicle Fuel Consumption

Vehicle fuel consumption data are modeled within Autonomie96 software. Template vehicles are selected with each of the powertrain technologies sought for our study; specifically the Conv Midsize Auto 2wd Default (CV), Split Midsize SingleMode HEV 2wd Default (HEV), Series Engine FixedGear PHEV 2wd Default (PHEV) and Elec Midsize FixedGear 2wd Default (BEV) models. These pre-built models are modified, as described within the Methods section of Chapter 4, in an attempt to isolate fuel consumption differences to those inherent to the powertrain technologies. Input vehicle characteristics are detailed in Table A-13 and calculated vehicle characteristics are itemized in Table A-14.

PHEV and BEV battery capacities are adjusted to achieve the desired driving ranges. Battery capacity is increased while maintaining the minimum voltage of the electric motor. Additional characteristics not outputted by Autonomie9 are calculated based on individual Johnson Controls Saft VL41M cell properties, assumed by the software. Voltage is calculated as the number of cells in series multiplied by the individual cell nominal voltage of 3.6 V. Total energy capacity is calculated as the number of cells in total multiplied by the individual cell energy capacity of 41 Ah. The usable energy capacity for charge depleting operation is the difference between the default target state-of-charge (SOC) of 30% and the initial SOC of 90% (default) for the PHEV and 100% (assumed) for the BEV models. [The different SOC swings are due to battery degradation from charge cycling, which would otherwise be more severe in PHEVs that are designed to be depleted regularly. Unlike a PHEV, a depleted BEV battery results in a stranded vehicle.] Battery mass is calculated based on an individual cell mass of 1.07 kg and 25% additional mass in the form of packaging, as described within Autonomie.96

Table A-13: Vehicle design and performance characteristics

Base Case Vehicle Future Vehicle Acceleration (0-100 km/h) 9 sec +/- 0.3 sec 9 sec +/- 0.3 sec Glider mass 891 kg 792 kg Frontal area 2.2 m2 2.2 m2 Drag coefficient 0.26 0.25 Rolling resistance 0.0075 0.0070 Internal Combustion Engine specific power 880 W/kg 990 W/kg Electric motor specific power 1250 W/kg 1600 W/kg Electric battery charging efficiency 85% 85% E85/gasoline ICE energy efficiency 107% 107% City/highway driving conditions 55%/45% by VKT 55%/45% by VKT

160

Table A-14: Mass and battery characteristics of vehicle models created in Autonomie96

Base Case Vehicles Future Vehicles CV HEV PHEV BEV CV HEV PHEV BEV Internal Combustion Engine Size (kW) 115 90 70 n/a 105 85 60 n/a Electric Motor Size (kW) n/a 75 100 110 n/a 60 90 95 Total Vehicle Mass Input (kg) 1698 1810 1911 1980 1570 1628 1709 1765 Li-ion Battery Size (# cells) 0 75 108 396 0 75 84 312 Li-ion Battery Mass (kg) 0 35 144 530 0 35 112 417 Li-ion Battery Capacity (kWh) 0 2 16 58 0 2 12 46 State of Charge Swing (%) n/a n/a 60 70 n/a n/a 60 70 Charge Depleting Capacity (kWh) n/a n/a 10 41 n/a n/a 9 37 Note: n/a = not applicable to vehicle model

The vehicle models are simulated on the FTP (Federal Test Procedure) and HWFET (Highway Fuel Economy Driving Schedule) driving cycles, created by the U.S. Environmental Production Agency (EPA) to represent performance under ideal city and highway driving conditions, respectively.178 These results are detailed in Table A-15. Driving cycle fuel efficiencies are then adjusted using Equations A-1 and A-2, as prescribed by the U.S. EPA178 5-cycle methodology to closer represent actual vehicle performance, to a maximum adjustment of 30%.179

Equation A-1: City driving fuel economy adjustment factor

1 5 − cycle City = 1.1805 0.003259 + FTP

Equation A-2: Highway driving fuel economy adjustment factor

1 5 − cycle Highway = 1.3466 0.001376 + HWFET

Table A-15: Un-weighted fuel consumption results of vehicle models created in Autonomie96

Drive Drive Base Case Vehicles Future Vehicles Cycle Mode Units CV HEV PHEV BEV CV HEV PHEV BEV City (FTP) CS Efficiency (MPG) 29.9 55.8 45.5 n/a 32.9 63.7 51.4 n/a CD Efficiency (Wh/mi) n/a n/a 217.6 220.4 n/a n/a 193.8 194.3 Highway CS Efficiency (MPG) 45.7 53.4 45.4 n/a 50.0 59.6 51.0 n/a (HWFET) CD Efficiency (Wh/mi) n/a n/a 223.9 223.0 n/a n/a 202.1 199.7 Note: CS = battery charge sustaining, CD = battery charge depleting

Driving ranges discussed in the Methods section of Chapter 4 are calculated based solely on the Urban Dynamometer Driving Schedule (UDDS) driving cycle used in other studies.8

161

Fuel consumption ratings presented in the Methods section of Chapter 4 reflect the different energy densities and efficiencies of use of each fuel in the vehicle. Energy densities are as follows; gasoline 34.7 MJ/L, pure ethanol 23.6 MJ/L, and E85 25.7 MJ/L.7 The breakdown of vehicle emissions and consumption of each fuel is shown in Table A-16.

Table A-16: Vehicle emissions and fuel consumption

Reference Vehicles (/100 VKT) Bioenergy Vehicles (/100 VKT) Grid-e/ Gasoline Gasoline Gasoline Grid-e Bio-e/ Bio-e CV HEV PHEV BEV E85 CV E85 HEV E85 PHEV BEV CH4 (g) 1 0 0 0 3 2 1 0 N2O (g) 1 0 0 0 3 2 1 0 CO2 (kg) 21 14 6 0 15 10 4 0 GHG (kg CO2eq.) 21 14 6 0 16 11 5 0 Gasoline (MJ) 313 213 88 0 75 51 21 0 Ethanol (MJ) 0 0 0 0 214 145 60 0 Electricity (MJ) 0 0 54 83 0 0 54 83 Total Energy (MJ) 313 213 142 83 289 195 135 83

Process 7: Vehicle Cycle

The GREET Vehicle-Cycle model7 is used to estimate environmental impacts attributed to vehicle production and disposal. Vehicle and battery masses are obtained directly from the aforementioned Autonomie models, detailed in Table A-14. Vehicles are based on Conventional Material characteristics and default assumptions are used. Major material inputs include steel, wrought aluminum, cast aluminum, lead and nickel, which are comprised of 26%, 11%, 85%, 73%, and 44% recycled material, respectively. Vehicles are assumed to operate for 250,000 lifetime VKT and lithium ion batteries are assumed to last the life of the vehicle. Energy use and GHG emissions for vehicle disposal are small in comparison to energy use and GHG emissions for the total vehicle cycle, and are assumed to be the same for each vehicle; therefore no specific considerations were given to lithium ion battery disposal, which can include recycling to reduce virgin material inputs or re-purposing for non-vehicle uses. The GHG emissions and energy use for vehicle cycle stages are compared in Table A-17.

162

Table A-17: Vehicle cycle results for vehicle models based on GREET Vehicle-Cycle model7

CV HEV PHEV BEV Components Production (w/o battery) CH4 (kg) 23 25 24 21 N20 (kg) 0 0 0 0 CO2 (t) 6 7 7 5 GHG Emissions (t CO2eq.) 7 7 7 6 Petroleum (mmBtu) 8 8 7 7 Fossil Energy (mmBtu) 81 85 84 70 Total Energy (mmBtu) 87 91 90 75 Assembly, Disposal and Recycling CH4 (kg) 5 5 5 5 N20 (kg) 0 0 0 0 CO2 (t) 1 1 1 1 GHG Emissions (t CO2eq.) 1 1 1 1 Petroleum (mmBtu) 0 0 0 0 Fossil Energy (mmBtu) 15 15 15 15 Total Energy (mmBtu) 16 16 16 16 Battery Production CH4 (kg) 0 0 2 8 N20 (kg) 0 0 0 0 CO2 (t) 0 0 1 2 GHG Emissions (t CO2eq.) 0 0 1 3 Petroleum (mmBtu) 0 0 2 7 Fossil Energy (mmBtu) 1 2 9 32 Total Energy (mmBtu) 1 2 11 37 Fluids Production CH4 (kg) 2 2 2 1 N20 (kg) 0 0 0 0 CO2 (t) 1 1 1 0 GHG Emissions (t CO2eq.) 1 1 1 0 Petroleum (mmBtu) 9 8 8 1 Fossil Energy (mmBtu) 12 11 11 3 Total Energy (mmBtu) 12 11 11 3 Total CH4 (kg) 31 32 34 36 N20 (kg) 0 0 0 0 CO2 (t) 8 9 9 9 GHG Emissions (t CO2eq.) 9 10 10 10 Petroleum (mmBtu) 17 16 17 15 Fossil Energy (mmBtu) 108 112 119 120 Total Energy (mmBtu) 116 120 128 132 Note: All values are per vehicle.

163

Results

Fuel Production Energy Balance

Fuel production energy balance results are shown in Figure A-5. Energy inputs include the energy content of each fuel, to provide an energy input/output ratio. In the well-to-wheel comparisons, the energy content of the fuel is allocated to the pump-to-wheel stage.

3 Other

2 Biomass Coal and Natural Gas

1 Petroleum

0 Energy Balance (MJ/MJ fuel atpump or plug)

-1 Gasoline Grid-e Ethanol Bio-e

Figure A-5: Fuel production energy balance Note: Co-product credits result in negative energy inputs

164

Life Cycle Results

Detailed pathway results are shown in Table A-18. An aggregated set of these data is presented in Figure 4-1.

Table A-18: Life cycle pathway results

Reference Pathways Bioenergy Pathways Grid-e/ Gasoline Gasoline Gasoline Grid-e E85 E85 Bio-e/ Bio-e CV HEV PHEV BEV CV HEV E85 PHEV BEV Carbon Absorption 0 0 0 0 -54 -37 -35 -31 Emissions 6 4 11 14 45 30 33 32 GHG WTP Co-product 0 0 0 0 -4 -3 -1 0 Emissions PTW 21 14 6 0 16 11 5 0 (kg CO eq. 2 Battery 0 0 0 1 0 0 0 1 /100 VKT) VC Non-Battery 3 4 4 3 3 4 4 3 Total 31 22 21 18 6 5 5 5 Biomass WTP 0 0 0 0 366 247 266 252 (MJ / PTW 0 0 0 0 214 145 114 83 100 VKT) Total 0 0 0 0 580 392 380 335 Energy Use 22 15 8 3 24 16 11 7 WTP Co-product 0 0 0 0 -1 -1 0 0 Petroleum PTW 313 213 89 1 75 51 21 0 (MJ / Battery 0 0 1 3 0 0 1 3 100 VKT) VC Non-Battery 7 6 6 3 7 6 6 3 Total 342 235 104 10 104 73 39 13 Energy Use 64 43 122 161 53 36 22 11 WTP Co-product 0 0 0 0 -47 -32 -13 0 Fossil Energy PTW 313 213 133 70 75 51 21 0 (MJ / Battery 0 1 4 13 0 1 4 13 100 VKT) VC Non-Battery 44 45 45 36 44 45 45 36 Total 421 303 304 280 124 100 78 60 Energy Use 65 44 143 193 420 284 288 263 WTP Co-product 0 0 0 0 -56 -38 -16 0 Total Energy PTW 313 213 142 83 289 195 135 83 (MJ / Battery 0 1 4 15 0 1 4 15 100 VKT) VC Non-Battery 47 48 48 38 47 48 48 38 Total 425 306 337 329 699 490 459 400 Note: Grid-e = grid-electricity, Bio-e = bio-electricity, CV = conventional vehicle, HEV = hybrid electric vehicle, PHEV = plug-in hybrid electric vehicle, BEV = battery electric vehicle, VKT = vehicle kilometers traveled, WTP = well-to-pump, PTW = pump-to-wheel, VC = vehicle cycle, Carbon absorption = CO2 offset during feedstock growth, Co-product = credit from grid-electricity offset

165

Mitigation Results

GHG, fossil energy and petroleum mitigation results associated with bioenergy displacing reference fuels are compiled in Table A-19. The GHG mitigation results are illustrated in Figure 4-2.

Table A-19: GHG emissions, fossil energy and petroleum mitigation results

Ethanol (in the form of E85) Bio-electricity Renewables- Renewables- U.S. Average based Coal-based U.S. Average based Coal-based Grid-e Grid-e Grid-e Grid-e Grid-e Grid-e GHG Mitigation (t CO2eq. 0.84 0.78 0.94 0.78 0.43 1.37 /dry t biomass) Fossil Energy Mitigation (GJ 10.2 9.6 10.8 13.1 7.4 18.3 /dry t biomass) Petroleum Energy Mitigation (GJ 8.2 8.2 8.2 -0.2 -0.4 -0.4 /dry t biomass)

166

Scenario Analysis

Co-product

Figure A-6 and Figure A-7 compare the impact of different co-product scenarios on life cycle total energy use and net GHG emissions, respectively. Without any co-product credit, total energy use and GHG emissions of pathways utilizing ethanol are increased, but the relative performance of the bioenergy pathways remains similar. A more favorable scenario assumes excess heat from bioenergy production can be utilized in an adjacent facility. This scenario is modeled based on co-product assumptions from literature.112 For both ethanol and bio-electricity production it is assumed that the excess process heat is utilized to generate steam (at 63% efficiency), which offsets natural gas use in an adjacent facility requiring thermal energy.112 Utilizing this process heat reduces the total energy use and GHG emissions for all bioenergy pathways. GHG emissions become negative for the latter scenario for all of the bioenergy pathways. This indicates the GHG emissions offset by the co-product exceed those emitted by the bioenergy pathway. The GHG emissions are much more sensitive than total energy use to these co-product assumptions, because of the GHG intensity (energy basis) of the displaced grid- electricity and natural gas, as compared to the bioenergy alternatives. Ethanol pathway GHG emissions can become higher or lower than those of the bio-electricity pathways depending upon the co-product assumptions. Regardless, bioenergy pathways continue to have GHG emissions lower than those of reference pathways.

1000 40

500 20

eq./100VKT) 0

TotalEnergy

(MJ/100 (MJ/100 VKT) CV

-20 CV

E85CV

Net GHG Net GHG Emissions CO (kg

E85CV

PHEV

E85HEV

Gasoline

E85HEV

Gasoline

Bio-eBEV

Bio-e/E85

Bio-eBEV Bio-e/E85 Base Case Base Case No Co-Product No Co-Product Combined Heat and Power Combined Heat and Power

Figure A-6: Life cycle sensitivity of total Figure A-7: Life cycle sensitivity of net energy use to co-product scenarios GHG emissions to co-product

167

Bioenergy Production Efficiency

Figure A-8 and Figure A-9 compare the impact on life cycle total energy use and net GHG emissions, respectively, of using different bioenergy production efficiencies. Compared to the base case assumptions in our study, Campbell et al.23 assumes a higher ethanol yield and bio- electricity generation efficiency. Substituting bioenergy production efficiencies from Campbell et al.23 into our study’s models, does not significantly affect the relative results among the pathways. The GREET fuel-cycle7 model assumes a higher ethanol yield but a lower bio- electricity generation efficiency than our study. Substituting GREET fuel-cycle7 assumptions into our study’s model results in the E85 HEV pathway having the lowest total energy use, but the result still being similar to the Bio-e BEV pathway.

The future “high efficiency” bioenergy production models are developed for our study in AspenPlus.115 Substituting these assumptions into the pathways suggests a potential for bio- electricity production efficiency to improve greatly, by utilizing integrated gasification combined cycle technology. GHG emissions are less sensitive than total energy use to bioenergy production efficiency assumptions, because changes in biomass feedstock use are largely offset with changes in carbon absorption during feedstock growth.

1000 40

500 20

0 eq./100VKT) 0

TotalEnergy

(MJ/100 (MJ/100 VKT)

Net GHG Net GHG Emissions

(kg CO (kg

E85CV E85CV

PHEV PHEV

E85HEV E85HEV

Bio-eBEV Bio-eBEV

Bio-e/E85 Bio-e/E85

GasolineCV GasolineCV

Base Case Base Case 23 Campbell et al. Bioenergy Production Campbell et al.23 Bioenergy Production 7 7 GREET fuel-cycle model Bioenergy Production GREET fuel-cycle model Bioenergy Production High Efficiency Bioenergy Production High Efficiency Bioenergy Production

Figure A-8: Life cycle total energy use Figure A-9: Life cycle net GHG emissions results for bioenergy production efficiency results for bioenergy production efficiency scenarios scenarios

168

Vehicle Models

Figure A-10 and Figure A-11 examine the impact of vehicle fuel efficiency assumptions on life cycle total energy use and net GHG emissions, respectively. Fuel cycle results (both well-to- pump and pump-to-wheel stages) of our study are modified through substituting alternative vehicle fuel consumption data from other studies into our models, while vehicle cycle (production/disposal) impacts remain unchanged. Substituting the lower efficiency ethanol fuelled vehicles, and higher efficiency bio-electricity fuelled vehicles used by Campbell et al.23 into our models results in a substantial difference in energy use. As described in the Results and Discussion section of Chapter 4, this illustrates that the bio-electricity pathways are “favored” by the vehicle fuel consumption models used by Campbell et al..23 Similarly, fuel consumption models from the GREET fuel-cycle model7 and the future high efficiency vehicles modelled in our study with Autonomie are examined. Substituting GREET fuel-cycle model7 fuel economy assumptions into our models results in total energy use that is higher or similar to the base case values shown, whereas the future high efficiency vehicles result in reduced total energy use for all pathways, but relative results remain similar for these assumptions. GHG emissions are less sensitive than total energy use to vehicle fuel consumption assumptions, because changes in biomass feedstock use are largely offset by changes in carbon absorption during feedstock growth.

1000 40

500 20

0 eq./100VKT) 0

TotalEnergy

(MJ/100 (MJ/100 VKT)

Net GHG Net GHG Emissions

(kg CO (kg

E85CV E85CV

PHEV PHEV

E85HEV E85HEV

Bio-eBEV Bio-eBEV

Bio-e/E85 Bio-e/E85

GasolineCV GasolineCV

Base Case Base Case Campbell et al.23 Vehicles Campbell et al.23 Vehicles GREET fuel-cycle model 7 Vehicles GREET fuel-cycle model 7 Vehicles High Efficiency Vehicles High Efficiency Vehicles

Figure A-10: Life Cycle total energy use Figure A-11: Life Cycle net GHG emissions results for vehicle efficiency scenarios results for vehicle efficiency scenarios

169

Appendix B: Chapter 5 Supporting Information Supplemental Methods

The approach summarized in the Methods section in Chapter 5 is elaborated upon here.

Air Emissions Impacts

Air emissions impacts are determined with a life cycle assessment. A life cycle inventory analysis is first conducted for both fuel and vehicle cycle activities. This is followed by an estimate of the NPV of life cycle impact of greenhouse gas (GHG) and criteria air contaminant (CAC) emissions.

Fuel Cycle Inventory Analysis

The fuel cycle consists of fuel production (gasoline, CNG, and NG-e, including feedstock production) and consumption (during vehicle operation) activities. All pathways are created within GREET 17 for the year 2020. Natural gas and petroleum feedstock are assumed to be from the default forecasted mix of conventional and unconventional sources. Average tailpipe emissions from GREET 17 are scaled to increase as vehicles age according to MOVES (Motor Vehicle Emission Simulator).141 Electricity generation emissions are based on GREET- calculated emissions factors, as opposed to emissions factors based on EPA (Environmental Protection Agency) and EIA (Energy Information Administration) databases. This results in emissions representing forecasted technology mixes (e.g., proportion of combined cycle facilities) and not historical performance.

Vehicle Cycle Inventory Analysis

The vehicle cycle consists of vehicle production (parts production and assembly), maintenance (tire and fluids replacement) and end-of-life processes (disposal and recycling). Gasoline and plug-in vehicle models were created within GREET 27 based on conventional materials. Vehicle mass is adjusted for CNG vehicles based on fuel tank assumptions in Table B-1. Plug-in lithium ion batteries are expected to last the life of each vehicle under base case the assumptions.7

170

Life Cycle Impact Assessment

The NPVs of climate change and health impacts are calculated based on GHG and CAC emissions, respectively. Air emissions impacts are the product of life cycle emissions quantities and specific impact costs (calculated with Equation 5-1). To represent the high level of inherent uncertainty in these models, a wide range of specific impact cost estimates are used in the Monte Carlo and sensitivity analyses.

Climate Change Impacts of GHG Emissions

Climate change can have impacts on agricultural yields, property damage, and ecosystems, among others. Climate change specific impact costs ($/t CO2eq.) are from the Interagency Working Group on Social Cost of Carbon,140 which is based on three integrated assessment models: Dynamic Integrated Climate and Economy, Policy Analysis of the Greenhouse Effect and Climate Framework for Uncertainty, Negotiation and Distribution.140 These models represent a range of socio-economic forecasts, climate sensitivity probability distributions, approaches to estimate potential damages, and discount rates to relate future costs to present day emissions. Global costs are accounted for due to the international nature of the impacts, and are higher than estimates based solely on domestic US implications. The base case value of $43/t

CO2eq. (2010 USD) used in this study is based on the average social cost estimate from the three models with the median discount rate of 3% for the emissions in the year 2020.

Health Impacts of CAC Emissions

Exposure to CAC emissions can have human health impacts including chronic morbidity and mortality from bronchitis and asthma. Health impacts from CAC emissions in individual US counties are from the Air Pollution Emission Experiments and Policy analysis model.97 The model estimates marginal impact costs of increased CAC emissions, and allocates these costs to the US County in which they are released. Weighted averages of these specific costs (shown in Table 5-1) are used to represent the geographic distributions of each life cycle stage (calculated with Equation 5-2). Impacts from vehicle operation emissions (from tailpipe, tire and brake wear and windshield washer fluid use) are estimated with the distribution of vehicle miles travelled across the US according the National Household Travel Survey.131 The distributions of natural gas electricity generation emissions are based on production patterns from the eGRID database.25 Other emissions are allocated according to US Census county business patterns for petroleum

171 and natural gas extraction, petroleum refining, natural gas distribution, motor vehicle parts manufacturing, battery manufacturing, petroleum lubricating oil and grease manufacturing, and automobile manufacturing.143 This methodology is similar, but more detailed, compared to what has been utilized in previous studies.6, 8, 80, 103, 140

Ownership Costs

Ownership costs include vehicle retail price and lifetime operating expenses, which include both fuel and maintenance.

Vehicle Price

The Vehicle Attribute Model3 estimates vehicle retail price equivalents. The model evaluates the trade-off between vehicle price and fuel economy to minimize the cost of the vehicle and three years of fuel. In contrast, this study requires a methodology to estimate the price of vehicles with specific fuel economy, CNG fuel tank and BEV battery capacity characteristics. As such, the Vehicle Attribute Model3 is not used directly, but underlying assumptions and calculations are utilized in this study as explained in the following subsections.

Gasoline CV

The Vehicle Attribute Model3 is based on historical baseline Model Year 2008 vehicles and accounts for changes in time, fuel economy and fuel type. The price of all of the components in CVs are considered mature technologies that decrease 1% per year from baseline data to estimate future prices (model year 2020 in this study). Two cost curves, both described by Equation B-1 and shown in Figure B-1, are applied to represent upper and lower bound estimates for the additional costs of fuel efficiency technologies required to account for incremental differences between the Gasoline CV model used in this study and the Vehicle Attribute Model3 baseline vehicle fuel economy. The average of the upper and lower bound estimates is used for the base case results in this study.

172

2 CV and HEV

BEV

1 Incremental Incremental Cost ($1000)

0 100% 110% 120% 130% Relative Fuel Economy

Figure B-1: Incremental costs of changes in relative fuel economy

Equation B-1: Incremental cost of changes in relative fuel economy calculation

FE b k C = (e FEo − ek) k

Where:

C = incremental cost

blower = 1108 for CV and HEV, 0.00001 for BEV

bupper = 2758 for CV and HEV, 0.00134 for BEV

klower = 0.9 for CV and HEV, 18.0 for BEV

klower = 0.7 for CV and HEV, 15.0 for BEV

FE = Fuel economy

FEo = Reference fuel economy

173

CNG CV

The Gasoline CV model is modified to estimate the price of the CNG CV. Two separate CNG CV models are developed to capture the wide range in possible costs, the higher of which is used as the base case estimate: the low cost model assumes a stainless steel CNG fuel tank; the high cost model assumes a carbon fibre CNG fuel tank. CNG fuel tank cost and mass parameters are detailed in Table B-1. Structural modifications are required to accommodate the fuel tanks, which result in an additional change in mass (50% of powertrain changes) and cost ($8/kg). Fuel economy is then reduced by 6% per 10% increase in total mass. The cost of additional fuel economy adjustments are estimated with Equation B-1, to achieve the base case assumption of an overall 5%7 energy equivalent fuel economy improvement for vehicles operating on CNG instead of gasoline. Additionally, engine modification cost estimates range from $500-$2300, the average of which is used as the base case estimate.

Table B-1: CNG fuel tank and BEV battery cost and mass parameters

Storage system CNG Fuel Tank BEV Battery Stainless Steel: $260 + $20/Lge Low: $760 + $240/kWh Cost Carbon Fibre:$390 + $60/Lge High: $760 + $410/kWh Stainless Steel: 4 kg/Lge Low: 8 kg/kWh Mass Carbon Fibre: 1 kg/Lge High: 10 kg/kWh Driving range 500 km, similar to 2013 Honda Civic NG180 125 km, similar to 2013 Nissan Leaf15 Fuel characteristics 32 MJ/Lge 89 kWh/Lge Note: A 30% markup is added to the costs above to estimate retail price3. Lge – liter gasoline equivalent

CNG HEV

The cost premium of the CNG HEV over the Gasoline CV, are estimated with Equation B-1 to achieve the 40% base case fuel economy improvement obtained from GREET. CNG fuel tank capacity is estimated using the assumptions outlined in Table B-1. Structural modifications are required to accommodate the fuel tanks, which result in an additional change in mass (50% of powertrain changes) and cost ($8/kg). Fuel economy is then reduced by 4% (as opposed to 6% used for the CNG CV, which does not have regenerative braking) per 10% increase in total mass. CNG HEV134 passenger vehicles have been developed by major automakers, but are not commercially available options.

174

NG-e BEV

Unlike the other powertrains in this study, BEVs do not have internal combustion engines. In a BEV, an internal combustion engine based powertrain is replaced with an electric motor equivalent. The cost, mass and efficiency specifications of both powertrain systems are listed in Table B-2. The ranges of cost and mass estimates for BEV batteries are detailed in Table B-1, the average of which is used as the base case assumption. Structural modifications are required to accommodate the new powertrain, which result in an additional change in mass (50% of powertrain changes) and cost ($8/kg). Fuel economy is then reduced by 4% per 10% increase in total mass. The cost of additional fuel economy adjustments to reach BEV fuel economy used in this study is calculated with Equation B-1. Finally, electric vehicle supply equipment (charger) costs are added to the vehicle for $760.

Table B-2: CV and BEV powertrain cost, mass and efficiency parameters

Powertrain CV BEV Cost (excl. energy storage) $2650 + $20/kW $20/kW Mass (excl. energy storage) 3 kg/kW 1 kg/kW 85% battery charging 95% battery discharging Efficiency 20% 90% electric motor 73% overall Regenerative Braking n/a 11% useful energy recaptured

Operating Costs

Operating costs are calculated as the sum of lifetime fuel and maintenance expenses.

Fuel

Gasoline, E85, CNG and electricity prices are based on the Annual Energy Outlook.2. Base case assumptions are from transportation sector prices from the 2014 reference case, which are based on Brent Spot prices for crude oil of $98/bbl and Henry Hub natural gas prices of $4.30/GJ. Gasoline, E85 and CNG prices include fuel taxes and dispensing costs (storage, transmission and distribution, retail markup). The electricity prices here are based on a mix of resources; however, due to the small contribution (7%) of electricity costs to the life cycle ownership costs of the BEV pathways, errors caused by this simplification will have negligible impact on the conclusions of this study. Doubling or completely removing the cost of electricity will still result in life cycle ownership costs that are lower for non-plug-in vehicles than plug-in vehicles, with the exception of those with particularly short driving ranges.

175

Maintenance

Vehicle maintenance costs and frequencies are itemized in Table B-3 and based on data from Oak Ridge National Laboratory.142 E85 and CNG vehicle maintenance costs are assumed to be identical to gasoline fuelled vehicles with equivalent powertrain (e.g., gasoline CV or HEV). BEV maintenance costs are not estimated by Oak Ridge National Laboratory,142 but these vehicles do not require oil change, air filter, spark plug, or timing chain replacements costs.8 Brake replacements are assumed to be equivalent to those of HEVs and PHEVs, due to the use of regenerative braking reducing the frequency of replacements.8 Other scheduled maintenance costs are assumed to be similar across powertrains.8 Unscheduled maintenance only considers the potential for BEV replacement battery because the focus of this study is on the relative costs between pathways – the cost of other unscheduled maintenance (e.g., windshield repair) are assumed to be similar for all vehicles.

Table B-3: Vehicle maintenance cost and frequency parameters

Parts and Labor Cost CV Frequency HEV Frequency BEV Frequency Oil Changes142 $80 8,000 km 12,000 km Not applicable Air Filter Replacements142 $50 50,000 km 50,000 km Not applicable Spark Plug Replacements142 $220 100,000 km 100,000 km Not applicable Timing Chain Adjustments142 $350 160,000 km 160,000 km Not applicable Front Brake Replacements142 $460 80,000 km 160,000 km 160,000 km Additional Maintenance* $7900 80,000 km 80,000 km 80,000 km 0-100% of initial battery Battery Replacement** Not applicable Not applicable 160,000 km cost *Costs from Oak Ridge National Laboratory,142 frequency assumed to coincide with typical year 5 peak in maintenance costs 181 **Assumed to not be required in reference scenario, but could occur after warranty period182 in the uncertainty analysis

176

Uncertainty and Sensitivity Analysis

The assumptions used to develop Monte Carlo and sensitivity analyses are presented below in Table B-4, Table B-5 and Table B-6. These complement Table 5-1, which lists the assumptions used to develop the base case results in this study.

Table B-4: Key life cycle inventory assumptions used to develop Monte Carlo and sensitivity analyses

Life Cycle Inventory Variable 5th/95th Percentile Probability Distribution Weibull dist. With location of 18.3, scale of 11.4 and shape of 3.2 for Gasoline CV (distribution multiplied by 100% for Vehicle Fuel Economy 83%/125% stainless steel CNG, 105% for carbon fibre CNG, 140% for HEV, and 400% for BEV) Weibull dist. With location of 0.00871, scale of 0.00397 and CH4 tailpipe emissions 78%/139% shape of 1.5805 for Gasoline CV (distribution multiplied by 1000% for CNG CV, 500% for CNG HEV and 0% for BEV) Gamma dist. With location of 0.03946, scale of 0.000246 and N O tailpipe emissions 95%/129% 2 shape of 3.1159 (distribution multiplied by 0% for BEV) Weibull dist. With location of 0.00241, scale of 0.00523 and PM tailpipe emissions 58%/301% 2.5 shape of 1.2447 (distribution multiplied by 0% for BEV) Weibull dist. With location of 0.03946, scale of 0.07566 and VOC tailpipe emissions 41$/241% shape of 1.0347 for Gasoline CV and CNG CV (distribution multiplied by 54% for HEV and 0% for BEV) Weibull dist. With location of 0.059, scale of 0.01239 and VOC evaporative emissions 100%/399% shape of 0.41316 for Gasoline CV (distribution multiplied by 50% for CNG and 0% for BEV) Gamma dist. With location of 0.04772, scale of 0.06234 and NOx tailpipe emissions 45%/215% shape of 1.2009 for Gasoline CV and CNG CV (distribution multiplied by 84% for HEV and 0% for BEV) Normal dist. of minimum acceptable range of new BEV drivers, BEV driving range 80/250 km with 145 km mean and 90 km std dev183 Triangular dist. with base case as most likely, and limits Battery replacement 0%/68% assuming entire battery pack replacement least likely184 Discrete, equally weighted binary dist., used to change cost, CNG fuel tank material Stainless steel/carbon fibre mass and fuel economy (scaled to mass)3 Discrete dist. based on shares of US vehicle annual miles of Lifetime vehicle travel 150,000/460,000 km travel and vehicle age1 Discrete 6-30 year dist. weighted according to US car Lifetime vehicle age 8/27 years scrappage rates1 Petroleum resource mix 9%/80% oil sands Triangular dist. with base case representing US average as Natural gas resource mix 14%/83% shale gas most likely, and limits of 0% and 100% acknowledging NG-e generation technology individual unit of fuel can be entirely from a particular source 21%/92% combined cycle mix of petroleum/natural gas/power plant technology7 Triangular dist. with base case as most likely, and 94% - 98% CNG compression efficiency 94%/98% limits from the literature55 Notes: CV = conventional vehicle, HEV = hybrid electric vehicle, BEV = battery electric vehicle, CNG = compressed natural gas

177

Table B-5: Key ownership cost and emissions impact assumptions used to develop Monte Carlo and sensitivity analyses

Ownership Cost Variable 5th/95th Percentile Probability Distribution Gasoline CV price $23,900/$24,400 Uniform dist. +/- 200 (gasoline CV and CNG CV with stainless CNG CV price $26,300/$28,400 steel tank), 800 (CNG HEV with stainless steel tank), 1500 (BEV CNG HEV price $27,500/$30,800 with 28 kWh battery), based on Vehicle Attribute Model forecasted fuel efficiency technology and CNG engine BEV price (excl. battery) $22,300/$25,400 modification price range3 Uniform dist. +/- $110/kWh, based on Vehicle Attribute Model BEV battery price $330/$530 per kWh battery forecasted price range3 2020 Brent spot crude oil $73/$123 per bbl Triangular dist. with base case as most likely limits US Gasoline price $0.64/$0.98 per L representing high and low oil price scenarios2 2020 Henry Hub natural gas $4.50/$7.00 per GJ Triangular dist. with base case as most likely, and limits US CNG price $13/$17 per GJ representing Annual Energy Outlook 2 low and National Energy US Electricity price $93/$116 per MWh Board 185 high gas price scenarios Triangular dist. with base case most likely and limits based on Ownership cost discount rate 6%/17% the perspective of social or individual consumer interests42

Air Emission Impact Variable 5th/95th Percentile Probability Distribution Triangular dist. with base case most likely and limits based on GHG impact specific cost $25/$115 per CO eq. 2 National Research Council illustrative range140 CAC impact specific cost See Table B-6 Discrete dist. based on quantity of life cycle stage activity Notes: CV = conventional vehicle, HEV = hybrid electric vehicle, BEV = battery electric vehicle, CNG = compressed natural gas, Costs are in 2010 USD. *These activities are weighted according to employment.6, 8, 80

Table B-6: Specific costs of CAC emissions impacts used to develop Monte Carlo and sensitivity analyses

Life Cycle Stage Percentile /t PM2.5 /t NOx /t SOx /t VOC County Vehicle 5th $3,600 $400 $3,500 $300 Tehama County, California operation 95th $91,900 $4,100 $20,500 $8,400 Union County, New Jersey Oil and gas 5th $700 $1,300 $200 $700 Eddy County, New Mexico extraction 95th $1,200 $7,900 $1,700 $5,400 Guilford County, North Carolina Gasoline fuel 5th $1,300 $1,300 $400 $700 Kay County, Oklahoma prod. 95th $300 $69,700 $10,100 $39,500 Los Angeles County, California CNG fuel 5th $800 $1,500 $200 $700 Dawson County, Montana production 95th $300 $28,500 $3,800 $14,300 San Diego County, California NG-e fuel 5th $1,200 $200 $400 $1,200 Yoakum County, Texas production 95th $9,000 $1,200 $37,500 $33,300 San Diego County, California Vehicle parts 5th $1,900 $2,100 $800 $900 Wyandotte County, Kansas prod. 95th $400 $10,600 $3,100 $23,800 Wayne County, Michigan Vehicle battery 5th $1,500 $1,400 $600 $700 Buchanan County, Missouri prod. 95th $500 $12,100 $3,200 $19,100 San Mateo County, California Vehicle fluids 5th $1,800 $2,200 $900 $700 Rockwall County, Texas prod. 95th $3,200 $16,800 $6,700 $12,400 Union County, New Jersey Vehicle 5th $2,400 $1,900 $1,200 $800 Wyandotte County, Kansas assembly 95th $500 $8,800 $3,300 $28,600 Wayne County, Michigan

178

Supplemental Results

The focus in Chapter 5 is on discussing aggregate life cycle results, while the results for individual emissions are presented in greater detail here. Figure B-2 shows CH4 and N2O emissions, which are not included in Figure 5-1, in addition to GHG and CAC impacts disaggregated by emission, as opposed to by life cycle stage in Figure 5-2. Figure B-3, Figure B- 4 and Figure B-5 show the life cycle energy use and emissions inventory, and air emissions impacts and ownership cost Monte Carlo analysis results for each vehicle pathway. Note that overlapping 90% confidence intervals in Figure 5-2. Figure B-3, Figure B-4 and Figure B-5 do not necessarily indicate that there is no significant difference between pathway results because some uncertainty is correlated. For example, the specific impact costs per tonne CO2 emissions have high uncertainty but the values should be identical for all vehicles in any direct comparison. Similarly, lifetime vehicle kilometers travelled is a variable that contributes to the uncertainty in all metrics but is assumed to be identical for each vehicle pathway. Figure B-5 shows that the 90% confidence intervals representing life cycle air emissions impacts of the gasoline CV and CNG CV overlap; however, the incremental analysis in Figure 5-3 shows that when common variables (e.g., life time VKT and $/t GHG) are the same, the CNG CV results in consistently lower life cycle air emissions impacts. On the other hand, Figure B-5 shows that the 90% confidence intervals representing the life cycle air emissions impacts of the CNG HEV and NG-e BEV also overlap, and the incremental analysis results in Figure 5-3 agree that the life cycle air emissions impacts are similar. This is why we present incremental differences, to capture these correlations when we introduce the discussion of uncertainty in Chapter 5.

179

a) 3

2 Vehicle Operation

(kg) Fuel Production

O O Emissions 2

N Vehicle Production 0 Gasoline CV CNG CV CNG HEV NG-e BEV

b) 300

200 Vehicle Operation (kg)

Emissions Fuel Production

4 100

CH Vehicle Production 0 Gasoline CV CNG CV CNG HEV NG-e BEV

c) 3

2 N2O CH4

($1000) 1 GHG GHG Climate

ChangeImpacts CO2 0 Gasoline CV CNG CV CNG HEV NG-e BEV d) 0.9 NOx 0.6 PM2.5

0.3 VOC CAC CAC Health

Impacts Impacts ($1000) SOx 0.0 Gasoline CV CNG CV CNG HEV NG-e BEV Figure B-2: Life cycle CH4 and N2O emissions disaggregated by life cycle stage and life cycle GHG and CAC impacts disaggregated by emission

180

100% 75% 50% Gasoline CV (90% CI: 1.2-2.2) 25% CNG CV (90% CI: 1.0-1.9) 0% CNG HEV (90% CI: 0.8-1.4) NG-e BEV (90% CI: 0.7-1.3)

Life Cycle Energy Use (TJ)

100% 75% 50% Gasoline CV (90% CI: 82-147) 25% CNG CV (90% CI: 60-107) 0% CNG HEV (90% CI: 47-80) NG-e BEV (90% CI: 43-81)

Life Cycle CO2 Emissions (t)

100% 75% 50% Gasoline CV (90% CI: 139-249) 25% CNG CV (90% CI: 264-468) 0% CNG HEV (90% CI: 194-339) NG-e BEV (90% CI: 107-205)

Life Cycle CH4 Emissions (kg)

100% 75% 50% Gasoline CV (90% CI: 4.0-6.9) 25% CNG CV (90% CI: 2.5-4.1) 0% CNG HEV (90% CI: 2.1-3.5) NG-e BEV (90% CI: 0.7-1.2)

Life Cycle N2O Emissions (kg)

Figure B-3: Life cycle energy use, CO2, CH4 and N2O emission Monte Carlo analysis results, including 90% confidence intervals in the legend.

181

100% 75% 50% Gasoline CV (90% CI: 63-135) 25% CNG CV (90% CI: 62-134) 0% CNG HEV (90% CI: 51-110) NG-e BEV (90% CI: 41-75)

Life Cycle NOx Emissions (kg)

100%

75%

50% Gasoline CV (90% CI: 49-65) CNG CV (90% CI: 40-49) 25% CNG HEV (90% CI: 47-54) 0% NG-e BEV (90% CI: 50-72)

Life Cycle SOx Emissions (kg)

100% 75% 50% Gasoline CV (90% CI: 90-189) 25% CNG CV (90% CI: 65-142) 0% CNG HEV (90% CI: 57-107) NG-e BEV (90% CI: 41-48)

Life Cycle VOC Emissions (kg)

100%

75%

50% Gasoline CV (90% CI: 7-13) CNG CV (90% CI: 5-11) 25% CNG HEV (90% CI: 5-10) 0% NG-e BEV (90% CI: 5-7)

Life Cycle PM2.5 Emissions (kg)

Figure B-4: : Life cycle NOx, SOx, VOC and PM2.5 emission Monte Carlo analysis results, including 90% confidence intervals in the legend.

182

100% 75% 50% Gasoline CV (90% CI: 2.5-11.3) 25% CNG CV (90% CI: 1.9-8.7) 0% CNG HEV (90% CI: 1.6-6.8) NG-e BEV (90% CI: 1.4-6.0)

Life Cycle Air Emissions Impacts ($1000)

100%

75% Gasoline CV (90% CI: 32-55) 50% CNG CV (90% CI: 34-53) 25% CNG HEV (90% CI: 33-50) 0% NG-e BEV (90% CI: 39-70) <30 30-35 35-40 40-45 45-50 50-55 55-60 60-65 65-70 >70 Life Cycle Air Emissions Impacts ($1000)

Figure B-5: : Life cycle air emissions impacts and ownership costs Monte Carlo analysis results, including 90% confidence intervals in the legend.

183

Supplemental Scenarios

Four supplemental scenarios are developed to examine quantitative effects of:

1. assuming no non-CO2 vehicle tailpipe or evaporative emissions (Zero CAC Emission Non-Plug-in Vehicle Scenario), 2. assuming no fuel economy advantage for CNG use over gasoline use (Low Fuel Economy CNG Vehicle Scenario), 3. assuming no uncertainty in BEV fuel economy (independent of battery capacity changes) (Constant Fuel Economy Plug-in Vehicle Scenario), and 4. assuming high (95th percentile) methane emissions from CNG vehicles (High Methane Emission CNG Vehicle Scenario) are minor due the numerous other sources of uncertainty analyzed in this study.

These four scenarios are presented in Figure B-6 and Table B-7. The qualitative conclusions of incremental life cycle ownership and air emissions impact costs in the Chapter 5 remain applicable in these scenarios.

Table B-7: Incremental life cycle ownership and emissions impact cost 90% confidence intervals for supplementary scenarios Fuel Switching Energy Efficiency Emissions Shifting CNG CV replacing CNG HEV replacing NG-e BEV replacing Gasoline CV CNG CV CNG HEV Incremental Incremental Incremental Incremental Incremental Incremental Life Cycle Life Cycle Life Cycle Life Cycle Life Cycle Life Cycle Ownership Costs Air Emissions Ownership Costs Air Emissions Ownership Costs Air Emissions Impact Benefit Impact Benefit Impact Benefit Results from 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: Chapter 5 -$3000 to $4000 $0 to $4000 -$5000 to $0 $0 to $2000 $1000 to $28,000 -$1000 to $2000 Zero CAC Emission 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: Non-Plug-in Vehicle -$4000 to $3000 $0 to $4000 -$5000 to $0 $0 to $2000 $0 to $27,000 -$1000 to $1000 Scenario Low Fuel Economy 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: CNG Vehicle -$2000 to $4000 $0 to $4000 -$4000 to $1000 $0 to $3000 $1000 to $28,000 -$1000 to $2000 Scenario Constant Fuel 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: Economy Plug-in -$3000 to $3000 $0 to $4000 -$5000 to $0 $0 to $2000 $1000 to $28,000 -$1000 to $1000 Vehicle Scenario High Methane 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: 90% CI: Emission CNG -$3,000 to $3,000 $0 to $4000 -$5000 to $0 $0 to $2000 $1000 to $28,000 -$1000 to $2000 Vehicle Scenario

184

a) Zero CAC Emission Non-Plug- Fuel Switching Energy Efficiency Emissions Shifting in Vehicle Scenario

50 50 50

Lose- Trade- Lose- Trade- Lose- Trade- Lose Off Lose Off Lose Off

0 0 0 -5 5 -5 0 5 -5 0 5 Trade- Win- Trade- Win- Trade- Win- Off Win Off Win Off Win

Life Cycle Incremental

Ownership Cost ($1000)

-50 -50 -50 Life Cycle Incremental Air Emissions Impact Benefit ($1000) b) Low Fuel Economy CNG Fuel Switching Energy Efficiency Emissions Shifting Vehicle Scenario

50 50 50

Lose- Trade- Lose- Trade- Lose- Trade- Lose Off Lose Off Lose Off

0 0 0 -5 5 -5 0 5 -5 0 5 Trade- Win- Trade- Win- Trade- Win- Off Win Off Win Off Win

Life Cycle Incremental

Ownership Cost ($1000)

-50 -50 -50 Life Cycle Incremental Air Emissions Impact Benefit ($1000) c) Constant Fuel Economy Plug- Fuel Switching Energy Efficiency Emissions Shifting in Vehicle Scenario

50 50 50

Lose- Trade- Lose- Trade- Lose- Trade- Lose Off Lose Off Lose Off

Cost ($1000) 0 0 0 -5 5 -5 0 5 -5 0 5 Trade- Win- Trade- Win- Trade- Win- Off Win Off Win Off Win

Life Cycle Incremental

Ownership

-50 -50 -50 Life Cycle Incremental Air Emissions Impact Benefit ($1000) d) High Methane Emission CNG Fuel Switching Energy Efficiency Emissions Shifting Vehicle Scenario

50 50 50

Lose- Trade- Lose- Trade- Lose- Trade- Lose Off Lose Off Lose Off

0 0 0 -5 5 -5 0 5 -5 0 5 Trade- Win- Trade- Win- Trade- Win- Off Win Off Win Off Win

Life Cycle Incremental

Ownership Cost ($1000)

-50 -50 -50 Life Cycle Incremental Air Emissions Impact Benefit ($1000)

Figure B-6: Incremental life cycle ownership and emissions impact cost results for supplementary scenarios

185

Appendix C: Chapter 6 Supporting Information Methods Details

The vehicles in this study are developed in two main components; base vehicle models and added fuel efficiency technologies. The Methods section of Chapter 6 provides a conceptual overview of the development of these components and how these two components are combined to produce the vehicle models. Additional details are provided in the following sections.

Base Vehicle Models

The 2012 reference vehicle and the series of vehicles that comprise of each of the vehicle design options are all developed with different base vehicle models. The diversity in base vehicle models is required to capture physical and temporal (model year vehicle is manufactured) differences among the vehicles, as explained in the Methods section of Chapter 6. Physical differences are discussed first, followed by temporal differences (see Manufacturing Cost subsection for the latter).

Physical Specifications

Physical differences in base vehicle models are used to quantify the ability for changes in vehicle acceleration, size and driving range to improve fuel economy. These vehicle attributes are analyzed by developing base vehicle models with different engine power ratings (100, 125, and 150 kW), body-type (Chevy Equinox-like and Honda Accord-like), and powertrain-type (conventional and battery electric with 6 and 16 kWh battery capacities), respectively. A flow chart depicting how these variables are modelled within Autonomie96 vehicle simulation software is shown in Figure C-1.

186

Battery Engine or Powertrain Glider Energy Motor Power Capacity Rating

100 kW Chevy Equinox- n/a 125 kW like Conventional 150 kW Gasoline 100 kW Honda Accord- n/a 125 kW like Autonomie Base 150 kW Vehicle Model 100 kW 6 kWh 125 kW Chevy Equinox- 150 kW Battery Electric like 100 kW 16 kWh 125 kW 150 kW

Figure C-1: Flow chart depicting base vehicle model development with Autonomie96

All base vehicle models are developed by modifying Autonomie templates because there are no templates within the software that meet the exact specifications required for this study. Templates are based on different vehicle powertrains. The gasoline vehicles in this study are all based on a conventional vehicle template, while the plug-in electric vehicles used to develop the driving range option are based on a battery electric vehicle template. Both of these powertrains are further discussed in the following two subsections.

Both the conventional vehicle and battery electric vehicle template are based on a midsize car glider (vehicle without powertrain). Autonomie also includes vehicle components from specific vehicles detailed in Table C-1. The SUVs in this study are all based on a first generation (model year 2005-2009)63 Chevy Equinox SUV, while the smaller vehicles used to develop the vehicle size option are based on a seventh generation (model year 2003-2007)63 Honda Accord sedan, which has nearly the same footprint as the Chevy Equinox-like SUV,63 and thus same fuel economy targets under CAFE standards.

187

Table C-1: Chevy Equinox-like and Honda Accord-like components

Chassis Chevy Equinox Honda Accord Massa (kg) 1180 990 Drag Area (m2) 0.977 0.675 Drag Coefficienta 0.37 0.30 Frontal Areaa (m2) 2.64 2.25 Footprint (m2) 4.49 4.26 Wheelbaseb (m) 2.86 2.74 Trackb (m) 1.57 1.55 Interior Volume (L) 4024 3305 Passenger Volumeb,c (L) 3013 2908 Cargo Volumeb,c (L) 1011 396 Final Drive Namea 406_au_VUT 444_accord Final Drive Ratioa 4.06 4.44 Wheels Namea 0357_P235_60_R17 0326_P205_60_R16 Radiusa (m) 0.357 0.326 Widtha (m) 0.235 0.205 Aspect Ratioa (%) 60 60 Rim Diametera (m/inch) 0.432/17 0.406/16 aObtained from Autonomie96 bObtained from Edmunds63 cObtained from The Car Connection160

The performance of the vehicle powertrain, glider, battery energy capacity and engine/motor power rating combinations outlined in Figure C-1 are simulated in Autonomie. Acceleration performance is quantified by simulating 0-96 km/h acceleration time. Fuel economy is quantified by simulating the vehicle in both city and highway driving cycles. For the purposes of meeting CAFE standards, combined fuel economy is calculated by weighting the unadjusted city and highway fuel economy results by 55% and 45%, respectively.29

188

Gasoline Vehicles

Gasoline base vehicle models are detailed in Table C-2 and use the Conv AutoTrans 2wd Midsize Autonomie template. This is a midsize sedan with a conventional powertrain that has a two- wheel drive 5-speed automatic transmission. This template is modified with either first generation (model year 2005-2009) Chevy Equinox-like or seventh generation (model year 2003-2007) Honda Accord-like components detailed in Table C-1. The entry level version of both of these real world vehicles came with two-wheel drive 5-speed automatic transmissions,63 which is consistent with the template. Fuel economy and acceleration performance is simulated within Autonomie96 for internal combustion engine power ratings of 100, 125 and 150 kW (134, 168, 201 hp).

Table C-2: Gasoline Chevy Equinox-like and Honda Accord-like vehicle specifications for a range of engine power ratings

Chevy Equinox-like Honda Accord-like Engine Power Rating (kW) 100 125 150 100 125 150 Vehicle Mass (kg) 1892 1925 1959 1787 1820 1854 0-60 mph Acceleration (s) 13.7 10.9 9.2 12.5 10.1 8.6 Fuel Economy/Consumption 32/7.4 30/7.8 27/8.7 35/6.7 31/7.6 29/8.1 (mpg/km L per L100 km) City (mpg/kmL per L100 km) 28/8.4 26/9.0 24/9.8 30/7.8 27/8.7 25/9.4 Highway (mpg/kmL per L100 km) 37/6.4 35/6.7 32/7.4 43/5.5 39/6.0 37/6.4

Gasoline fuel tank size is not a parameter within Autonomie96 so gasoline vehicle driving range is assumed to be a constant 600 km. This is approximately the same capability of the entry level first generation (model year 2003-2007) Chevy Equinox.15 A gasoline fuel tank is a minor contributor to total vehicle mass and price, unlike with plug-in vehicle batteries.3 Therefore, adjustments in fuel tank size required to maintain a constant driving range would have negligible impact on fuel economy and price.

189

Plug-in Electric Vehicles

Plug-in electric base vehicle models are detailed in Table C-3 and are based on the BEV FixedGear 2wd Midsize Autonomie template. This is a midsize sedan with a battery electric powertrain that has a two-wheel drive fixed-gear transmission. This template is modified with a Chevy Equinox-like glider previously detailed in Table C-1 and GM Voltec battery cell specifications provided within Autonomie (detailed in Table C-3).96 GM Voltec battery cells, which are used in the Chevy Volt,186 specifications are selected to provide higher power densities than the default battery, which has higher energy density. Power density is prioritized because the driving range option consists of vehicles with sufficiently low energy capacities to maintain the price of the 2012 reference vehicle, while also maintaining acceleration performance. Fuel economy and acceleration performance is simulated within Autonomie for vehicles with total battery energy capacities of 6 (smallest that could maintain 9.3 s 0-96 km/h acceleration performance) and 16 kWh (Chevy Volt capacity) and electric motor power ratings of 100, 125 and 150 kW. Usable battery energy capacity is assumed to be 77.5% in 2015 and increase by 2.5 percentage points every five vehicle model years.3 Driving ranges are calculated within a spreadsheet based on vehicle fuel economy, total battery energy capacity and percentage of usable battery energy capacity.

Table C-3: Plug-in electric Chevy Equinox-like vehicle specifications for range of motor power ratings and battery capacities

Battery Energy Capacity (kWh) 6 16 Cells (#) 108 288 Energy Capacity96 (Ah/cell) 15 15 Power Capacity96 (kW/cell) 1.041 1.041 Motor Power Rating (kW) 100 125 150 100 125 150 Vehicle Mass (kg) 1783 1811 1839 2028 2056 2084 0-60 mph Acceleration (s) 11.1 10.1 09.3 12.6 10.0 08.4 Fuel Economy/Consumption 73/3.2 72/3.3 69/3.4 67/3.5 67/3.5 64/3.7 (mpg/km L per L100 km) City (mpg/km L per L100 km) 77/3.1 76/3.1 72/3.3 73/3.2 72/3 68/3.5 Highway (mpg/km L per L100 km) 69/3.4 69/3.4 67/3.5 63/3.7 63/3.7 61/3.9 Driving Range Model Year 2015 (km) 16.8 16.4 16.0 41.6 40.6 39.6 Model Year 2020 (km) 17.4 17.0 16.5 42.9 41.9 40.9 Model Year 2025 (km) 17.9 17.5 17.1 44.3 43.2 42.1

190

Manufacturing Costs

Autonomie96 provides low-, average- and high-risk estimates of component manufacturing costs over time. The set of costs are presented in Table C-4 to C-7 in 2010 USD for model years 2012, 2015, 2020 and 2025, respectively. The earliest model year included with Autonomie is 2013, so 2012 costs are extrapolated from 2013 and 2015 estimates. The three cost estimates capture uncertainty and are based on the risk of achieving cost reduction projections, as discussed in the Methods section. The average risk cost estimates are used to develop the mid-price scenarios. The high risk cost estimates are used to develop the low price scenarios, and vice versa. Autonomie only provides point estimate costs for the wheels and the torque converter; although there is likely some uncertainty in real world costs, there would be negligible impact on the results of this study because these components comprise less than 4% of total base vehicle model costs and an even smaller fraction of total vehicle price (which includes added fuel efficiency technologies).

The Vehicle Attribute Model3 was utilized for certain manufacturing cost assumptions. Autonomie chassis costs increase in the future, which reflects the increasing use of lightweight materials.105 Unfortunately, this overlaps with the added fuel efficiency technologies, from the Vehicle Attribute Model, that are applied. Further, the actual weight reduction from the use of lightweight materials is not specified within Autonomie.96 Therefore, future base vehicle model chassis costs are estimated based on a 1% annual reduction based on the Vehicle Attribute Model.3 The cost of electric vehicle supply equipment (i.e., charger), which is necessary for plug-in electric vehicles, is also utilized from the Vehicle Attribute Model3 because it is not included in Autonomie.96

The vehicle price is estimated by adding a 30% markup over manufacturing costs.3

191

Table C-4: Model Year 2012 vehicle manufacturing costs

High Price Mid-Price Low Price Power Rating (kW) 100 125 150 100 125 150 100 125 150 Manufacturing Cost ($) 15368 15668 16234 14778 15035 15517 14364 14593 15025 Chassis ($) 10483 10483 10483 10483 10483 10483 10483 10483 10483 Engine ($) 2351 2539 2993 1992 2152 2536 1813 1958 2308 Gearbox ($) 1610 1722 1834 1400 1498 1595 1190 1273 1356 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Starter ($) 45 45 45 33 33 33 27 27 27 Generator ($) 16 16 16 15 15 15 11 11 11 Accessories ($) 251 251 251 251 251 251 251 251 251 Torque Converter ($) 109 109 109 109 109 109 109 109 109 Wheels ($) 443 443 443 443 443 443 443 443 443 Note: all costs are presented in 2010 USD. Costs are primarily estimated with Autonomie and exceptions are explained in the text.

192

Table C-5: Model Year 2015 vehicle manufacturing costs

High Price Mid-Price Low Price Power Rating (kW) 100 125 150 100 125 150 100 125 150 Gasoline Chevy Equinox-like Vehicle Manufacturing Cost ($) 14928 15223 15775 14355 14608 15078 13952 14177 14600 Chassis ($) 10171 10171 10171 10171 10171 10171 10171 10171 10171 Engine ($) 2281 2464 2904 1933 2088 2461 1759 1900 2240 Gearbox ($) 1562 1671 1780 1359 1453 1548 1155 1235 1316 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Starter ($) 45 45 45 33 33 33 26 26 26 Generator ($) 15 15 15 15 15 15 11 11 11 Accessories ($) 240 243 246 240 243 246 240 243 246 Torque Converter ($) 109 109 109 109 109 109 109 109 109 Wheels ($) 443 443 443 443 443 443 443 443 443 Gasoline Honda Accord-like Vehicle Manufacturing Cost ($) 14405 14700 15253 13832 14085 14556 13429 13654 14077 Chassis ($) 9757 9757 9757 9757 9757 9757 9757 9757 9757 Engine ($) 2281 2464 2904 1933 2088 2461 1759 1900 2240 Gearbox ($) 1562 1671 1780 1359 1453 1548 1155 1235 1316 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Starter ($) 45 45 45 33 33 33 26 26 26 Generator ($) 15 15 15 15 15 15 11 11 11 Accessories ($) 240 243 246 240 243 246 240 243 246 Torque Converter ($) 109 109 109 109 109 109 109 109 109 Wheels ($) 335 335 335 335 335 335 335 335 335 Plug-in Electric Chevy Equinox-like 16 kWh Vehicle Manufacturing Cost ($) 19588 20063 20538 18517 18942 19367 17590 17865 18140 Plug-in Battery ($) 5095 5095 5095 233 4233 4233 3919 3919 3919 Chassis ($) 10171 10171 10171 10171 10171 10171 10171 10171 10171 Engine ($) 1900 2375 2850 1700 2125 2550 1100 1375 1650 Gearbox ($) 2685 2685 2685 1790 1790 1790 1492 1492 1492 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Charger ($) 842 842 842 842 842 842 842 842 842 Starter ($) 62 62 62 52 52 52 39 39 39 Generator ($) 970 970 970 970 970 970 970 970 970 Accessories ($) 300 300 300 300 300 300 300 300 300 Wheels ($) 443 443 443 443 443 443 443 443 443 Plug-in Electric Chevy Equinox-like 6 kWh Vehicle Manufacturing Cost ($) 16505 16879 17354 15871 16296 16721 15140 15415 15690 Plug-in Battery ($) 1911 1911 1911 1587 1587 1587 1470 1470 1470 All other costs equal to those of 16 kWh Vehicle Note: all costs are presented in 2010 USD. Costs are primarily estimated with Autonomie and exceptions are explained in the text.

193

Table C-6: Model Year 2020 vehicle manufacturing costs

High Price Mid-Price Low Price Power Rating (kW) 100 125 150 100 125 150 100 125 150 Gasoline Chevy Equinox-like Vehicle Manufacturing Cost ($) 14229 14510 15035 13682 13923 14371 13297 13511 13913 Chassis ($) 9673 9673 9673 9673 9673 9673 9673 9673 9673 Engine ($) 2169 2343 2762 1838 1986 2340 1673 1807 2130 Gearbox ($) 1486 1589 1693 1292 1382 1472 1098 1175 1251 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Starter ($) 45 45 45 33 33 33 25 25 25 Generator ($) 15 15 15 15 15 15 10 10 10 Accessories ($) 228 231 234 228 231 234 228 231 234 Torque Converter ($) 109 109 109 109 109 109 109 109 109 Wheels ($) 443 443 443 443 443 443 443 443 443 Gasoline Honda Accord-like Vehicle Manufacturing Cost ($) 13727 14008 14533 13180 13421 13868 12795 13009 13411 Chassis ($) 9279 9279 9279 9279 9279 9279 9279 9279 9279 Engine ($) 2169 2343 2762 1838 1986 2340 1673 1807 2130 Gearbox ($) 1486 1589 1693 1292 1382 1472 1098 1175 1251 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Starter ($) 45 45 45 33 33 33 25 25 25 Generator ($) 15 15 15 15 15 15 10 10 10 Accessories ($) 228 231 234 228 231 234 228 231 234 Torque Converter ($) 109 109 109 109 109 109 109 109 109 Wheels ($) 335 335 335 335 335 335 335 335 335 Plug-in Electric Chevy Equinox-like 16 kWh Vehicle Manufacturing Cost ($) 17518 17843 18168 16307 16557 16807 15307 15336 15511 Plug-in Battery ($) 4233 4233 4233 3331 3331 3331 2498 2498 2498 Chassis ($) 9673 9673 9673 9673 9673 9673 9673 9673 9673 Engine ($) 1300 1625 1950 1000 1250 1500 700 875 1050 Gearbox ($) 2386 2386 2386 1611 1611 1611 1313 1313 1313 Charger ($) 761 761 761 761 761 761 761 761 761 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Starter ($) 62 62 62 52 52 52 39 39 39 Generator ($) 922 922 922 922 922 922 922 922 922 Accessories ($) 300 300 300 300 300 300 300 300 300 Wheels ($) 443 443 443 443 443 443 443 443 443 Plug-in Electric Chevy Equinox-like 6 kWh Vehicle Manufacturing Cost ($) 14873 15198 15523 14225 14475 14725 13599 13774 13949 Plug-in Battery ($) 1587 1587 1587 1249 1249 1249 937 937 937 All other costs equal to those of 16 kWh Vehicle Note: all costs are presented in 2010 USD. Costs are primarily estimated with Autonomie and exceptions are explained in the text.

194

Table C-7: Model Year 2025 vehicle manufacturing costs

High Price Mid-Price Low Price Power Rating (kW) 100 125 150 100 125 150 100 125 150 Gasoline Chevy Equinox-like Vehicle Manufacturing Cost ($) 13565 13832 14331 13043 13272 13698 12674 12878 13260 Chassis ($) 9199 9199 9199 9199 9199 9199 9199 9199 9199 Engine ($) 2063 2228 2626 1748 1888 2226 1591 1718 2025 Gearbox ($) 1413 1511 1610 1229 1314 1400 1044 1117 1190 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Starter ($) 45 45 45 33 33 33 24 24 24 Generator ($) 15 15 15 14 14 14 10 10 10 Accessories ($) 217 220 223 217 220 223 217 220 223 Torque Converter ($) 109 109 109 109 109 109 109 109 109 Wheels ($) 443 443 443 443 443 443 443 443 443 Gasoline Honda Accord-like Vehicle Manufacturing Cost ($) 13082 13349 13848 12560 12789 13215 12191 12395 12777 Chassis ($) 8824 8824 8824 8824 8824 8824 8824 8824 8824 Engine ($) 2063 2228 2626 1748 1888 2226 1591 1718 2025 Gearbox ($) 1413 1511 1610 1229 1314 1400 1044 1117 1190 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Starter ($) 45 45 45 33 33 33 24 24 24 Generator ($) 15 15 15 14 14 14 10 10 10 Accessories ($) 217 220 223 217 220 223 217 220 223 Torque Converter ($) 109 109 109 109 109 109 109 109 109 Wheels ($) 335 335 335 335 335 335 335 335 335 Plug-in Electric Chevy Equinox-like 16 kWh Vehicle Manufacturing Cost ($) 16929 17244 17559 15528 15769 16010 14390 14547 14703 Plug-in Battery ($) 4233 4233 4233 3135 3135 3135 2351 2351 2351 Chassis ($) 9199 9199 9199 9199 9199 9199 9199 9199 9199 Engine ($) 1260 1575 1890 965 1206 1448 625 781 938 Gearbox ($) 1671 1671 1671 1492 1492 1492 1193 1193 1193 12 V Battery ($) 62 62 62 52 52 52 39 39 39 Charger ($) 723 723 723 723 723 723 723 723 723 Starter ($) 62 62 62 52 52 52 39 39 39 Generator ($) 877 877 877 877 877 877 877 877 877 Accessories ($) 300 300 300 300 300 300 300 300 300 Wheels ($) 443 443 443 443 443 443 443 443 443 Plug-in Electric Chevy Equinox-like 6 kWh Vehicle Manufacturing Cost ($) 14284 14599 14914 13568 13809 14051 12921 13077 13233 Plug-in Battery ($) 1587 1587 15887 1176 1176 1176 882 882 882 All other costs equal to those of 16 kWh Vehicle Note: all costs are presented in 2010 USD. Costs are primarily estimated with Autonomie and exceptions are explained in the text.

195

Added Fuel Efficiency Technologies

Added fuel efficiency technologies are based on Equation C-1 from the Vehicle Attribute Model3 that estimates the price (manufacturing cost and markup) of fuel economy improvements. The Vehicle Attribute Model3 provides upper and lower bound price equation parameters for different vehicle classes, fuel type and model years (see Table C-8). The two bounds provide two price estimates that capture uncertainty, the average of which is used to develop the mid-price scenario. The lower bound estimates are used to calculate the low price scenario, and vice versa. The small SUV and large car class sizes are used for the Chevy Equinox-like and Honda Accord- like vehicles, respectively. The Vehicle Attribute Model3 only provides equation parameters for 2015 onwards, so values for 2012 are extrapolated from 2015 and 2020 parameters.

Equation C-1: Price of added fuel efficiency technologies

FE b k P = (e FEo − ek) k

Where:

P = incremental price [$]

b = cost parameter shown in Table S8 [$]

k = cost scaling parameter shown in Table S8 [dimensionless]

FE = improved fuel economy [MPG]

FEo = base vehicle model fuel economy [MPG]

196

Table C-8: Parameters for calculating price of added fuel efficiency technologies from Vehicle Attribute Model3

2012 2015 2020 2025 b 1289 1220 1048 947 Gasoline Low Price k 0.900 0.900 0.900 0.900 Chevy Equinox-like b 2957 2900 2758 2623 Vehicle High Price k 0.700 0.700 0.700 0.700 b n/a 1290 1180 1001 Gasoline Low Price k n/a 0.900 0.900 0.900 Honda Accord-like b n/a 2900 2758 2623 Vehicle High Price k n/a 0.700 0.700 0.700 b n/a 1.33x10-5 1.14x10-5 1.03x10-5 Plug-in Electric Low Price k n/a 18.00 18 18 Chevy Equinox-like b n/a 1.41x10-3 1.34x10-3 1.27x10-3 Vehicle High Price k n/a 15 15 15

We assume the use of added fuel efficiency technologies from the Vehicle Attribute Model changes base vehicle model price and fuel economy, but not interior volume or acceleration performance. Although individual added fuel efficiency technologies can have an impact on acceleration, the data 2 cited by the Vehicle Attribute Model 3 does take into account internal combustion engine downsizing when estimating the incremental cost and fuel economy improvement of adding a turbocharger and regenerative braking/launch assist, which reduces the impact on vehicle horsepower, and thus acceleration. The use of lightweight materials can reduce vehicle mass and thus improve acceleration performance, but this is assumed to be offset by the cumulative impact of other technologies, such as added mass of components that facilitate direct fuel injection, variable compression ratios, engine start-stop and cylinder deactivation (among others). Unfortunately, we are unable to verify this assumption because the Vehicle Attribute Model 3 does not detail how specific added fuel efficiency technologies are aggregated within its incremental price vs. fuel economy improvement curves.

Vehicle Design Options

The data presented in Table C-2 through Table C-7 are used to quantify relationships between different vehicle component sizing specifications (vehicle interior volume, engine/motor power rating and battery engine capacity) and base vehicle model physical specifications (price, fuel economy and acceleration), over time (model year). The parameters in Table C-8 are used to quantify the relationship between vehicle price increase and fuel economy improvement provided by utilizing added fuel efficiency technologies, over time. Collectively, these

197 relationships map the trade-offs among overall vehicle fuel economy, price, acceleration, size and driving range, over time.

Quadratic lines of best fit are developed within Excel for engine/motor power ratings versus base vehicle model price, fuel economy and acceleration performance. The engine/motor power rating to achieve the target 0-96 km/h acceleration time of 9.3 s is calculated iteratively for each vehicle model. The exception being vehicle models that comprise the vehicle acceleration pathway, in which case the engine/motor power rating that maintained 2012 reference vehicle price and met future CAFE standards, after added fuel efficiency technologies are applied, is iteratively determined. In the case of the vehicle price option, future CAFE standards are met by applying added fuel efficiency technologies only, to base vehicle models with 0-96 km/h acceleration times of 9.3 s.

A linear relationship is assumed for vehicle size versus base vehicle model price, fuel economy and acceleration performance, over time. This simplified relationship is due to the lack of vehicle body types beyond a sedan and SUV within Autonomie to quantify the characteristics of vehicles with different interior volumes by similar footprint (e.g., no wagons or range of SUVs with same footprint). Linearly interpolating between car and SUV characteristics is consistent with the method used in the Vehicle Attribute Model to estimate the relationship between vehicle price and mass.3 The vehicle size option is developed by linearly interpolating between the Honda Accord-like and Chevy Equinox-like vehicles, all of which have engine power ratings that provide 9.3 s 0-96 km/h acceleration time and added fuel efficiency technologies that meet CAFE standards. Vehicle model specifications are calculated for a vehicle with an interior volume that would maintain the 2012 reference vehicle price.

A linear relationship is assumed for battery energy capacity versus base vehicle model price, fuel economy and acceleration performance. This relationship is assumed to be applicable across the narrow range of battery energy capacities examined (even upper end of range is less than currently real world battery electric vehicle capacities),15 which are limited by vehicle price and acceleration performance constraints as described above. Additionally, batteries are modular (compiled of individual cells) with approximately linear impacts on vehicle price96 and fuel economy.3 The vehicle driving range option is developed by linearly interpolating between 6 kWh and 16 kWh battery electric vehicles, all of which have motor power ratings that provide

198

9.3 s 0-96 km/h acceleration time. Added fuel efficiency technologies are then applied to maintaining the 2012 reference vehicle price. Vehicle model specifications are iteratively determined based on the combination of battery energy capacity and utilization of added fuel efficiency technologies that maximize vehicle driving range. Evaluation of Study Results through Comparison with Literature

This study uses a novel means of analyzing CAFE standards. As such, the results are compared and contrasted with those derived from methods found in the literature. However, there are no estimates of the degree to which vehicle size or acceleration could increase from 2012 levels to meet 2025 CAFE standards. Thus, the approach proposed by An and DeCicco 33 is used to supplement the results of this study.

An and DeCicco33 developed the performance, size and fuel economy index (PSFI). This is calculated as the product of average annual US engine power rating to vehicle mass ratio (hp/lb), size (ft3), and fuel economy (mpg). PSFI increased approximately linearly from 1977 to 2005. If this trend continues, PSFI could increase by 24% from 2012 to 2025, which is less than the 66% increase in CAFE standard fuel economy targets during the same time frame. Therefore, a reduction in engine power rating to vehicle mass ratio (hp/lb) or size (ft3) by 26% could facilitate this fuel economy increase, according to this approach [(100%- 26%)*(100%+66%)=(100%+24%)].

300

24% increase in PSFI from 2012 to 2025 *mpg)

3 200

100 PSFI (hp/lb*ft PSFI 0 1975 1985 1995 2005 2015 2025 Model Year

Figure C-2: PSFI projected to 2025 based on 1977-2005 data33 (adapted from An and DeCicco)33 Note: PSFI is the product of average annual US engine power rating to vehicle mass ratio (hp/lb), size (ft3), and fuel economy (mpg)

199

The relationship between vehicle acceleration and the engine power rating to vehicle mass ratio is not linear. Therefore, Equation C-2 from the EPA 13 is used to estimate 0 to 60 mph (96 km/h) acceleration time. A 26% decrease in the performance ratio results in a 27% increase in 0 to 60 mph (96 km/h) acceleration time.

Equation C-2: Acceleration performance as a function of engine power rating to vehicle mass 13

Z60 = 0.892P−0.805

Where:

Z60 = 0 to 60 mph (96 km/h) acceleration time [s]

P = performance ratio of engine power rating to vehicle mass [hp/lb]

Results Details

The results used to produce Figure 6-2 are presented in this section. 2012 reference vehicle specifications are detailed in Table C-9, while 2015, 2020 and 2025 vehicle models that comprise of each vehicle design options are detailed in Table C-10. The price curves in Figure 6-2 are based on the mid-price scenario, while the high and low price scenarios form the ends of the error bars in Figure 6-2a.

Table C-9: 2012 Reference vehicle model specifications

Engine Power Rating (kW) 147 0-96 km/h Acceleration (s) 9.3 Interior Volume (L) 4020 Driving Range (km) 600 Footprint (m2) 4.5 2012 CAFE standard –fuel economy/consumption target 32/137.4 for 4.5 m2 footprint (mpg/km L per L100 km) Base Vehicle Model (mpg/km L per L100 km) 27/128.7 Level of added fuel efficiency tech (%) 16 High Price Mid-Price Low Price Vehicle Retail Price ($) 22100 20900 20000 Base Vehicle Model ($) 21000 20100 19500 Added fuel efficiency technologies ($) 1000 800 600

200

Table C-10: Vehicle design option model specifications

High Price Mid-Price Low Price 2015 2020 2025 2015 2020 2025 2015 2020 2025 Vehicle Price Option Vehicle Retail Price (2010 USD) 22100 23200 25600 20800 21400 23000 19800 19800 20600 Base vehicle model 20400 19500 18900 19400 18600 18000 18500 17700 17200 Added fuel efficiency technologies 1600 3700 7000 1300 1800 5200 900 1800 3400 Fuel economy /consumption 34/ 42/ 53/ 34/ 42/ 53/ 34/ 42/ 53/ (mpg/kmL per L100 km) 6.9 5.6 4.4 6.9 5.6 4.4 6.9 5.6 4.4 Base vehicle model 27/ 27/ 27/ 27/ 27/ 27/ 27/ 27/ 27/ (mpg/kmL per L100 km) 8.7 8.7 8.7 8.7 8.7 8.7 8.7 8.7 8.7 Level of added fuel efficiency tech (%) 25 54 93 25 54 93 25 54 93

Vehicle Acceleration Option 0-60 mph Acceleration (s) 9.1 9.9 12.2 9.1 9.9 12.2 9.1 9.9 12.2 Engine Power Rating (kW) 150 138 112 150 138 112 150 138 112 Vehicle Retail Price (2010 USD) 22200 22600 23500 20900 20900 20900 19900 19500 19500 Base vehicle model 20500 19200 18300 19600 18400 17100 19000 17800 17000 Added fuel efficiency technologies 1700 3400 5200 1300 2500 3800 900 1600 2500 Fuel economy /consumption 34/ 42/ 53/ 34/ 42/ 53/ 34/ 42/ 53/ (mpg/kmL per L100 km) 6.9 5.6 4.4 6.9 5.6 4.4 6.9 5.6 4.4 Base vehicle model 27/ 28/ 30/ 27/ 28/ 30/ 27/ 28/ 30/ (mpg/kmL per L100 km) 8.7 8.4 7.8 8.7 8.4 7.8 8.7 8.4 7.8 Level of added fuel efficiency tech (%) 26 50 73 26 50 73 26 50 73

Vehicle Size Option Interior Volume (L) 4081 3836 3305 4081 3836 3305 4081 3836 3305 Engine Power Rating (kW) 148 144 138 148 144 138 148 144 138 Vehicle Retail Price (2010 USD) 22200 22500 23000 20900 20900 20900 19900 19400 19000 Base vehicle model 20500 19200 17700 19600 18400 16900 19000 17800 16400 Added fuel efficiency technologies 1700 3400 5300 1300 2500 3900 900 1600 2500 Fuel economy /consumption 34/ 41/ 52/ 34/ 41/ 52/ 34/ 41/ 52/ (mpg/kmL per L100 km) 6.9 5.7 4.5 6.9 5.7 4.5 6.9 5.7 4.5 Base vehicle model 27/ 28/ 30/ 27/ 28/ 30/ 27/ 28/ 30/ (mpg/kmL per L100 km) 8.7 8.4 7.8 8.7 8.4 7.8 8.7 8.4 7.8 Level of added fuel efficiency tech (%) 26 50 75 26 50 75 26 50 75

Driving Range Option Driving Range (km) n/a 25 35 n/a 25 35 n/a 25 35 Battery Capacity (kWh) n/a 10 13 n/a 10 13 n/a 10 13 Motor Power Rating (kW) n/a 145 140 n/a 145 140 n/a 145 140 Vehicle Retail Price (2010 USD) n/a 23200 22000 n/a 20900 20900 n/a 19700 19400 Base vehicle model n/a 23100 21900 n/a 20900 20800 n/a 19700 19300 Added fuel efficiency technologies n/a 0 200 n/a 0 100 n/a 0 0 Fuel economy /consumption n/a 72/ 69/ n/a 72/ 69/ n/a 72/ 69/ (mpg/kmL per L100 km) 3.3 3.4 3.3 3.4 3.3 3.4 Base vehicle model n/a 68/ 67/ n/a 68/ 67/ n/a 68/ 67/ (mpg/kmL per L100 km) 3.5 3.5 3.5 3.5 3.5 3.5 Level of added fuel efficiency tech (%) n/a 1 3 n/a 1 3 n/a 1 3

201

Appendix D: Chapter 7 Supporting Information Supplemental Methods

All vehicles in this study are based on a method discussed in Chapter 7 and first developed in Chapter 6. Each vehicle includes a base vehicle model developed within Autonomie. All base vehicle models are modified using assumptions from the Vehicle Attribute Model to improve fuel economy and, in the case of compressed natural gas (CNG) vehicles, for CNG use. An overview of this process is illustrated in Figure D-1 and discussed below.

Base Vehicle Model Glider Chevy Equinox- (Autonomie) Like

Base Vehicle Model Gasoline Battery Powertrain Convention Electric (Autonomie) al Vehicle Vehicle

Base Vehicle Model Plug-in Not High High Battery Cell Type Applicable Power Energy (Autonomie)

Base Vehicle Model Plug-in Not 32 kWh 98 kWh 169 kWh Battery Capacity (Autonomie) Applicable

93% fuel 54% fuel 25% fuel 20% fuel 20% fuel 20% fuel Added Fuel Efficiency economy economy economy Technologies economy economy economy increase increase increase increase increase increase (Vehicle Attribute Model) (2025 (2020 (2015 CAFE) CAFE) CAFE) (Max) (Max) (Max)

Engine Engine Engine CNG Use Modifications Not and tank and tank and tank Not Not Not Applicable Applicable Applicable Applicable (Vehicle Attribute Model) mods for mods for mods for CNG use CNG use CNG use

Gasoline CNG CNG CNG Mid- Short- Mid- Long- High- High- Low- Efficiency Distance Distance Distance Complete Vehicle Models Efficiency Efficiency Efficiency ICEV BEV BEV BEV ICEV ICEV ICEV

Figure D-1: Overview of vehicle models Notes: CAFE = Corporate Average Fuel Economy standards, CNG = compressed natural gas, ICEV = internal combustion engine vehicle, BEV = battery electric vehicle

202

Base Vehicle Models

Autonomie vehicle simulation software is used to develop the base vehicle models.96 This tool simulates vehicle performance (e.g., fuel economy) and estimates manufacturing costs based on detailed component level assumptions (e.g., aerodynamic drag).96 The base vehicle models for internal combustion engine vehicles (ICEV) and battery electric vehicles (BEV) are based on Autonomie vehicle templates with gasoline conventional and battery electric powertrains, respectively. Both templates are modified to have a Chevy Equinox-like glider (vehicle without powertrain).63 A common glider is selected for comparability and a crossover SUV is chosen to better represent the light-duty vehicle market than a car or truck-based SUV. Different powertrain components (internal combustion engine and electric motor) power ratings are tested for acceleration and fuel economy performance. The component specifications that provide average light-duty vehicle 0-98 km/h acceleration time of 9.3 s13 are interpolated, along with associated fuel economy ratings presented in Table D-1.

Plug-in batteries are a unique aspect of the BEV base vehicle models. Batteries are required to provide sufficient energy and power capacities. Batteries energy capacities are sized (interpolated iteratively alongside the level of added fuel efficiency technologies, as discussed in the following subsection) to provide the targeted 100 km, 300 km and 500 km driving ranges using both high power cells (1041 W and 148 Wh per cell) and high energy cells (800 W and 324 Wh per cell) defined within Autonomie.96 The lowest price option that can still provide the sufficient power to achieve the 0-98 km/h acceleration time of 9.3 s is selected. The short- distance BEV uses a 32 kWh battery comprised of high power cells. The mid- and long-distance BEVs use 98 kWh and 169 kWh batteries, respectively, comprised of high energy cells.

The price of the base vehicle models are detailed in Table D-1. Prices are based on Autonomie96 component manufacturing costs. The exception is the price of the charger, which is not included in Autonomie96 and thus from the Vehicle Attribute Model.3 A 30% retail price markup to be consistent with the added fuel efficiency technologies, which are discussed in the following subsection.3 The average of the price ranges are used for the base case results in Chapter 7.

203

Table D-1: Fuel economy and price of base vehicle models

ICEV Short-Distance Mid-Distance Long-Distance BEV BEV BEV Fuel Economy (2-cyclea/5-cycleb MPGec) 21/27 70/100 63/90 56/80 Price (Low/High Estimate) Glider $12000/$12000 $12000/$12000 $12000/$12000 $12000/$12000 Engine/Motor $2600/$3400 $1000/$2000 $1100/$2300 $1300/$2700 Gearbox $1500/$2100 0/0 0/0 0/0 Plug-in Battery 0/0 $8200/$11500 $14200/$24100 $24600/$41800 Charger 0/0 $900/$900 $900/$900 $900/$900 Other $1100/$1200 $2200/$2300 $2200/$2300 $2200/$2300 Aunadjusted laboratory rating used for CAFE standard compliance badjusted rating used for real world fuel consumption and driving range estimate cMiles per gallon gasoline on an energy equivalent basis

Added Fuel Efficiency Technologies

The Vehicle Attribute Model3 estimates the incremental price of fuel economy improvements based on the aggregation of different fuel efficiency technologies (e.g., lightweight materials and hybrid electric powertrain components) forecasted to be commercially available in future model years. This tool is used to model added fuel efficiency technologies. Added fuel efficiency technologies are modelled with the Vehicle Attribute Model,3 which is structured to model incremental changes to increase fuel economy, unlike Autonomie.96 The price of these incremental fuel economy improvements are added to the base vehicle models. The fuel economy of ICEVs are improved to meet 2015, 2020 and 2025 CAFE standards,9 respectively, for a 4.5 m2 Chevy Equinox-like footprint,63 as discussed in Chapter 7. The fuel economy of BEVs are improved to the maximum forecasted to be feasible within the Vehicle Attribute Model,3 which was determined iteratively to be a more cost-effective means to provide the driving ranges targeted in this study than further increasing battery capacity. The price of fuel economy improvements are estimated with Equation D-1 from the Vehicle Attribute Model,3 which is based on an aggregation of individual technologies from the Energy Information Administration.2 The incremental fuel economy and vehicle price of added fuel efficiency technologies are shown in Table D-2. The average of the price ranges are used for the base case results in Chapter 7.

204

Equation D-1: Price of added fuel efficiency technologies

퐹퐸 푏 푘 푃 = (푒 퐹퐸표 − 푒푘) 푘

Where:

푃=incremental price [$]

푏=price parameter [$], 947 to 2623 for ICEVs and 1.03x10-5 to 1.27x10-3 for battery electric vehicles

푘=price scaling factor [dimensionless], 0.7 to 0.9 for ICEVs and 15 to 18 for battery electric vehicles

퐹퐸표=Initial fuel economy [MPG]

퐹퐸=Improved fuel economy [MPG], limited by technological options to a maximum of 121% and 20% greater than the initial fuel economy for internal combustion engine vehicles and battery electric vehicles, respectively

Table D-2: Incremental fuel economy and price from added fuel efficiency technologies

Low- Mid- High- Efficiency Efficiency Efficiency Short- Mid-Distance Long- ICEV ICEV ICEV Distance BEV BEV Distance BEV Fuel economy improvement over 25% 54% 93% 20% 20% 20% base vehicle model Price (Low/High Estimate) $700/$1500 $1700/$3500 $3400/$7000 $500/$1200 $500/$1200 $500/$1200 Notes: ICEV = internal combustion engine vehicle, BEV = battery electric vehicle

205

CNG Modifications

The Vehicle Attribute Model3 is used to model modifications of gasoline ICEVs modifications for CNG use, which is not possible within Autonomie.96 Engine modification costs range from $500 to $2000 and result in a thermal efficiency improvement of 14%, but additional fuel tank mass offsets some of this benefit.3 A stainless steel fuel tank has a fixed cost of $290, a variable cost of $40 per Lge (liter gasoline energy equivalent) and a mass of 4 kg per Lge.3 A carbon fibre fuel tank has a fixed cost of $320, a variable cost of $40 per Lge (liter gasoline energy equivalent) and a mass of 1 kg per Lge.3 Fuel economy is reduced by 6% per 10% increase in vehicle mass.3 Fuel tank size is scaled to provide an average light-duty vehicle driving range of 600 km.3 The range in price and fuel economy from the modifications are shown in Table D-3, with higher fuel economy associated with higher vehicle price. The average of these ranges are used for the base case results in Chapter 7.

Table D-3: Incremental fuel economy and price from CNG modifications

Low-Efficiency Mid-Efficiency High-Efficiency ICEV ICEV ICEV Fuel economy improvement over base vehicle model (Low/High) 6%/12% 8%/12% 9%/12% Price (Low/High) $2400/$4000 $2100/$3700 $1800/$3400 Notes: ICEV = internal combustion engine vehicle

206

Crystal Ball

The Monte Carlo analyses were conducted with Crystal Ball software.98 The input parameters are provided in Table D-4. Results are based on simulations of 10,000 trials.

Table D-4: Monte Carlo and sensitivity analyses assumptions

5th/50th/95th Percentile Assumption Distribution

Operation Weibull distribution based on fuel ICEV Fuel Economy 80%/100%/120% of base case economy from GREET7 Weibull distribution based on fuel BEV Fuel Economy 80%/100%/120% of base case economy from GREET7 Discrete distribution based on US statistics Lifetime Years 7/17/27 years from the Transportation Energy Data Book1 Discrete distribution based on US statistics Lifetime vehicle miles travelled 90,000/180,000/240,000 miles from Transportation Energy Data Book1 Discrete distribution based on discount Discount Rate 4%/8%/20% rates from Argonne National Laboratory42 Low Oil Price/Reference/High Oil Price Discrete distribution based on Annual Fuel Price Scenario Energy Outlook price forecasts2

Vehicle Design 16%/50%/84% of difference between Triangular distribution on prices from Base Vehicle Model Price high and low price estimates Autonomie96 16%/50%/84% of difference between Triangular distribution based on prices Fuel Efficiency Improvement Costs high and low price estimates from Vehicle Attribute Model3 16%/50%/84% of difference between Triangular distribution on price range from CNG Modification Costs high and low price estimates Vehicle Attribute Model3 16%/50%/84% of difference between Triangular distribution on prices from Battery Costs high and low price estimates Autonomie96

Fuel Production Normal distribution based on gasoline Gasoline Production GHGs 97%/100%/103% of base case refining efficiency from GREET7 Triangular distribution based on CNG CNG Production GHGs 98%/100%/102% of base case compression efficiency from GREET7 Weibull distribution based on combined Electricity Production GHGs 80%/100%/108% of base case cycle electricity generation efficiency from GREET7 Notes: ICEV = internal combustion engine vehicle, BEV = battery electric vehicle

207

Supplemental Results

The Monte Carlo analysis results presented in Figure 7-3 and discussed in Chapter 7 are presented in greater detail here in Figure D-2. The ownership costs and well-to-wheel GHG emissions from using CNG or natural gas-derived electricity (NGCCe) can be higher or lower than those of the gasoline high-efficiency ICEVs.

8000 CNG ICEV Incremental Ownership Costs 4000 High-Efficiency (90% CI: -11% to 5%) 0

>6% Mid-Efficiency (90% CI: -13% to 0%)

<-12%

0% 0% 2% to 2% 4% to 4% 6% to

-2%to 0% Low-Efficiency (90% CI: -11% to 0%)

-8%to -6% -6%to -4% -4%to -2%

-10% to -8% -10%to -12%-10% to

Frequencyof Results Incremental GHG Emissions (Relative to Gasoline Vehicle)

8000 NGCCe BEV Incremental Ownership Costs 4000 Short-Distance (90% CI: -11% to 13%) 0

Mid-Distance (90% CI: 12% to 35%)

>60% <-30%

Long-Distance (90% CI: 34% to 53%)

0% 10% to

-10%0% to

20% to 20%to 30% 10% to 20% 30% to 40% 40% to 50% 50% to 60%

-30%-20% to -20%-10% to

Frequencyof Results Incremental GHG Emissions (Relative to Gasoline Vehicle)

8000 CNG ICEV Incremental GHG Emissions 4000 High-Efficiency (90% CI: -28% to -23%) 0

Mid-Efficiency (90% CI: -9% to -3%)

>20%

<-25% 0% 5% to

-5%to 0% Low-Efficiency (90% CI: 13% to 21%)

5% 10% to

10% to 15% 15% to 20%

-10%-5% to

-25%-20% to -20%-15% to -15%-10% to

Frequencyof Results Incremental GHG Emissions (Relative to Gasoline Vehicle)

8000 NGCCe BEV Incremental GHG emissions 4000 Short-Distance (90% CI: -52% to -4%) 0

>0% Mid-Distance (90% CI: -47% to 6%) <-45%

-5%to 0% Long-Distance (90% CI: -40% to -19%)

-10%-5% to

-20% to -15% -20%to -45%-40% to -40%-35% to -35%-30% to -30%-25% to -25%-20% to -15%-10% to

Frequencyof Results Incremental GHG Emissions (Relative to Gasoline Vehicle)

Figure D-2: Histogram and 90% confidence intervals (CI) of incremental ownership costs and well-to-wheel GHG emissions relative to gasoline use Notes: ICEV = internal combustion engine vehicle, BEV = battery electric vehicle, CNG = compressed natural gas, NGCCe = natural gas combined cycle derived electricity

208

The Monte Carlo analysis results for the alternative scenarios discussed in Chapter 7 are presented in Figure D-3. The GHG emissions from each of the renewable CNG ICEVs and biomass-derived electricity BEVs are lower than those of the gasoline high-efficiency ICEVs. The GHG emissions from each of the coal-derived electricity BEVs are higher than those of the gasoline high-efficiency ICEVs.

8000 Renewable CNG ICEV Incremental GHG Emissions 4000

High-Efficiency (90% CI: -84% to -83%) 0

Mid-Efficiency (90% CI: -81% to -79%) >-75%

<-84% Low-Efficiency (90% CI: -76% to -74%)

Frequencyof Results

-83% to -82% to -83% -84% to -83% -84%to -81% -82%to -80% -81%to -79% -80%to -78% -79%to -77% -78%to -76% -77%to -75% -76%to Incremental GHG Emissions (Relative to Gasoline Vehicle)

8000 Biomass-Derived Electricity BEV Incremental GHG Emissions 4000

Short-Distance (90% CI: -95% to -90%) 0

Mid-Distance (90% CI: -95% to -89%) >-86%

<-95% Long-Distance (90% CI: -94% to -88%)

Frequencyof Results

-94% to -93% to -94% -95% to -94% -95%to -92% -93%to -91% -92%to -90% -91%to -89% -90%to -88% -89%to -87% -88%to -86% -87%to Incremental GHG Emissions (Relative to Gasoline Vehicle)

8000 Coal-Derived Electricity BEV Incremental GHG Emissions 4000

0 Short-Distance (90% CI: 14% to 129%)

Mid-Distance (90% CI: 25% to 152%) <0%

>225% Long-Distance (90% CI: 41% to 182%)

Frequencyof Results

0% to 25% 0%to

25% to 50% 25%to 75% 50%to

75% to 100% 75%to

100% to 125% to 100% 125% to 150% 125%to 175% 150%to 200% 175%to 225% 200%to Incremental GHG Emissions (Relative to Gasoline Vehicle)

Figure D-3: Histogram and 90% confidence intervals (CI) of incremental well-to-wheel GHG emissions relative to gasoline use for vehicles using renewable compressed natural gas, biomass-derived electricity or coal-derived electricity Notes: ICEV = internal combustion engine vehicle, BEV = battery electric vehicle, CNG = compressed natural gas, NGCCe = natural gas combined cycle derived electricity

209

Chapter 5 and Appendix A is adapted with permission from Luk, J., Pourbafrani, M., Saville, B., MacLean, H. Ethanol or Bio-electricity? Life cycle assessment of bioenergy use in light-duty- vehicles, Environmental Science & Technology, 2013, 47 (18) 10676-10684. http://pubs.acs.org/articlesonrequest/AOR-JzIibBUXwnhzN6Exqysg. Copyright 2015 American Chemical Society.

Chapter 6 and Appendix A is adapted with permission from Luk, J., Saville, B., MacLean, H. Life cycle air emissions impacts and ownership costs of light-duty vehicles using natural gas as a primary energy source, Environmental Science & Technology, 2015, 49 (8) 5151-5160. http://pubs.acs.org/articlesonrequest/AOR-3bqzfeAZcSBvFxjutSWh. Copyright 2015 American Chemical Society.