SHAPING SUSTAINABLE VEHICLE FLEET CONVERSION POLICIES BASED ON LIFE CYCLE OPTIMIZATION AND RISK ANALYSIS
by
Hyung Chul Kim
A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Natural Resources and Environment) in The University of Michigan 2003
Doctoral Committee:
Associate Professor Gregory A. Keoleian, Co-Chair Professor Jonathan W. Bulkley, Co-Chair Professor James C. Bean Professor Marc H. Ross James H. Lindner, US EPA
Hyung Chul Kim © 2003 All Rights Reserved
for my parents
ii ACKNOWLEDGMENTS
First and foremost, I would like to offer my sincerest thanks to my University of Michigan committee members: to my co-chair, Professor Gregory Keoleian, for his encouragement and careful and wise guidance of my project; to my co-chair, Professor Jonathan Bulkley, for his kind advice and academic support; to Professor Marc Ross for always answering my inquiries so thoughtfully and promptly; and to Professor James Bean for his time and help in solving technical problems. In addition, Dr. Jim Lindner at the US Environmental Protection Agency provided me with invaluable information and direction. This research would have never been completed without support from outside the university. I am deeply grateful to Kevin Cullen, Terry Cullum, and Ronald Williams at General Motors for providing me with both technical and financial support in the early stages of this research. Special gratitude is also extended to John Walls at the Arizona Department of Environmental Quality for the essential data and kind answers to my questions. I thank Ruth Kraut for her advice throughout the writing process. Her help dramatically improved this dissertation. Many thanks are also due to Darby Grande for her suggestions and for other help during the modeling process. I have been helped and encouraged by many people at the Center for Sustainable Systems at the School of Natural Resources and Environment in one way or another. I would like to single out several people for special thanks: Martin Heller, Helaine Hunscher, and Marc Melaina.
iii Finally, I thank all of my friends in Ann Arbor, who have been with me throughout this critical period in my life.
iv TABLE OF CONTENTS
DEDICATION...... ii
ACKNOWLEDGMENTS ...... iii
LIST OF FIGURES ...... viii
LIST OF TABLES ...... xi
LIST OF APPENDICES ...... xiii
CHAPTER
I. INTRODUCTION...... 1 1.1 High-Emitters...... 3 1.2 Scrappage Program ...... 5 1.3 Inspection and Maintenance (I/M) Programs ...... 8 1.4 Problem Statements ...... 12 1.5 Dissertation Outline ...... 14 II. DYNAMIC LIFE CYCLE INVENTORY...... 17 2.1 Vehicle LCA ...... 17 2.2 Dynamic LCI Conceptual Structure...... 19 2.3 Dynamic LCI Modeling for a Generic Vehicle ...... 22 2.3.1 Materials Production...... 24 2.3.2 Manufacturing...... 25 2.3.3 Use ...... 26 2.3.4 Maintenance...... 28 2.3.5 End-of-Life ...... 29 2.4 Dynamic LCI Parameters for a Generic Vehicle ...... 30 2.4.1 Recycled Content...... 30 2.4.2 Materials Use ...... 33 2.4.3 Vehicle Miles Traveled (VMT) ...... 34 2.4.4 Energy Intensity...... 34 2.4.5 Fuel Economy ...... 45 2.4.6 Emission Factors...... 47 2.4.7 Component Reliability...... 54 2.5 Results and Discussion ...... 55 2.6 Conclusion ...... 62
v III. OPTIMAL VEHICLE LIFE BASED ON LIFE CYCLE OPTIMIZATION MODEL...... 63 3.1 Introduction...... 63 3.2 Life Cycle Optimization (LCO) Model ...... 64 3.2.1 Dynamic Programming...... 64 3.2.2 Model Construction ...... 65 3.2.3 Model Application ...... 69 3.3 Results and Discussion ...... 70 3.3.1 Optimal Lifetimes for Mid-Sized Cars ...... 70 3.3.2 Determinants of Optimal Lifetime...... 73 3.3.3 Policy Implications ...... 76 3.4 Conclusion ...... 78 IV. OPTIMAL VEHICLE LIFE BASED ON FLEET OPTIMIZATION MODEL ...... 80 4.1 Introduction...... 80 4.2 Fleet Characterization ...... 81 4.3 Sources of Emissions ...... 84 4.4 Fleet Optimization Modeling...... 85 4.4.1 Baseline Fleet...... 85 4.4.2 Fleet Optimization Scheme...... 87 4.4.3 Mathematical Modeling...... 88 4.5 Results...... 90 4.5.1 Ideal Fleet Conversion ...... 91 4.5.2 Long-Term Fleet Conversion...... 92 4.5.3 Emission Reductions...... 94 4.5.4 Multi-Objective Analysis...... 94 4.5.5 Sensitivity of Results ...... 98 4.6 Discussion...... 99 4.6.1 Fleet Conversion Policy...... 99 4.6.2 Multi-Objective Analysis...... 101 4.7 Conclusion ...... 102 V. IMPROVING INSPECTION AND MAINTENANCE PROGRAMS USING RISK ANALYSIS OF EMISSION CONTROL SYSTEMS ...... 104 5.1 Introduction...... 104 5.2 Fault Tree Analysis of Emission Control Systems ...... 106 5.2.1 Emission Control Systems ...... 108 5.3 Analysis of IM147 Records ...... 116 5.3.1 Overview of the Arizona IM147 Tests ...... 116 5.3.2 Overview of I/M Repairs ...... 122 5.3.3 Repairs of CO High-Emitters...... 135 5.3.4 Repairs of NOx High-Emitters ...... 137 5.3.5 Repairs of HC/CO High-Emitters...... 140 5.3.6 Repairs of HC/CO/NOx High-Emitters...... 144 5.3.7 I/M Analysis Based on Manufacturers ...... 146
vi 5.4 Emission Reductions through I/M Program ...... 148 5.4.1 Cars Tested During both 2000 and 2002 I/M Cycles...... 149 5.4.2 All Cars Tested ...... 151 5.5 High-Emitter Contributions to Fleet Emissions...... 155 5.6 I/M versus Scrappage Policy Examinations ...... 157 5.7 Scrappage Policies from a Life Cycle Perspective ...... 161 5.8 Conclusion ...... 169 VI. CONCLUSION ...... 172 6.1 Research Scope and Key Findings...... 172 6.2 Policy Implications ...... 175 6.3 Future Research ...... 177 APPENDICES...... 179
BIBLIOGRAPHY...... 221
vii LIST OF FIGURES
Figure
2.1. Life cycle energy consumption of a generic mid-sized passenger vehicle based on 120,000 miles of driving (USAMP 1999)...... 20
2.2. Factors, parameters, and life cycle stages illustrating the dynamic LCI for a generic vehicle...... 21
2.3. The detailed dependencies between the parameters and environmental burdens of the use phase life cycle ...... 27
2.4. Fuel economy trends and forecast between 1985 and 2020: (EPA 2000; EIA 2001) ...... 47
2.5. Estimation of the average (at 80K miles) emission factor contributions * of high-emitters (E H) (Austin and Ross 2001) ...... 52
2.6. Dynamic LCIs of materials production stage, BM(i), between model years 1985 and 2020 ...... 56
2.7. Dynamic LCIs of manufacturing stage, BA(i), between model years 1985 and 2020...... 57
2.8. Dynamic LCIs of use stage, BU(i,j), for select model years based on driving 12,000 miles annually ...... 58
2.9. Dynamic LCIs of maintenance stage, BR(i,j), at vehicle age 1 based on driving 12,000 miles annually ...... 60
2.10. Dynamic LCIs of the end-of-life stage, BE(i,j), based on retiring vehicles in the tenth year ...... 61
3.1. A schematic example of the life cycle optimization (LCO) model based on four policies (B1, B2, B3, and B4 represent the final environmental burdens for the four policies.)...... 67
3.2. Cumulative environmental burdens accrued during the 36 years of time horizon when adopting energy/CO2-optimum and CO-optimum policies with 12,000 miles of annual VMT (The cumulative optimal environmental burdens are normalized to 1 for each criterion.)...... 73
viii 4.1. Survival rate curves for cars used in this study...... 82
4.2. Annual vehicle miles traveled (VMT) for cars developed for MOBILE6 ...... 83
4.3. Car fleet in 2000 based on MOBILE6 fleet characterization ...... 86
4.4. Average hazard rate profile (fraction per year) of car fleet based on MOBILE6 fleet characterization ...... 87
4.5. Relationships between new vehicle versus retired vehicle mileage for fleet optimization model ...... 88
4.6. Optimal fleet distributions and emission savings at 2001 for different pollutants...... 92
4.7. Vehicle age distributions for optimal fleet conversion between 2000 and 2020 (These are three separate programs, different for each pollutant.)...... 93
4.8. Emission reductions by the long-term optimal fleet conversion policy allowing 120 production ...... 95
4.9. CO2 cost and the optimal production (EPA 1992)...... 97
5.1. Basic fault-tree symbols (Haimes 1998; Raheja 1999)...... 107
5.2. Schematics of air/fuel (A/F) ratio control (Bosch 2000) ...... 108
5.3. Exhaust emissions based on air/fuel ratio (φ=1, if air/fuel ratio is 14.7:1.) (Stone 1999) ...... 109
5.4. Fault tree diagram for CO high-emitters...... 111
5.5. Fault tree diagram for NOx high-emitters ...... 112
5.6. Fault tree diagram for HC/CO high-emitters...... 114
5.7. Fault tree diagram for HC/CO/NOx high-emitters ...... 115
5.8. Breakdown of the IM147 test records based on 2000 I/M tests ...... 118
5.9. Results of the initial 2002 I/M tests for cars failed in the initial 2000 I/M tests ...... 120
5.10. Failure rates of cars tested during both 2000 and 2002 I/M cycles (See Tables 5.1 and 5.3 for details.)...... 121
ix 5.11. Emission profiles in the initial 2000 tests for the cars tested during both 2000 and 2002 I/M test cycles based on model years and high- emitter types (See Table 5.3.) ...... 127
5.12. Emission profiles of fail/pass cars across the 2000 and 2002 I/M cycles based on high-emitter types ...... 130
5.13. Part repair rates for CO high-emitters (3,091 cars) repaired during the 2000 I/M cycle...... 136
5.14. Failure rates in initial 2002 I/M tests for CO-type high-emitters ...... 137
5.15. Part repair rates for NOx high-emitters (6,982) repaired during the 2000 I/M cycle...... 138
5.16. Failure rates in initial 2002 I/M tests for NOx high-emitters ...... 139
5.17. Part repair rates for HC/CO high-emitters (3,120 cars) repaired during the 2000 I/M cycle ...... 140
5.18. Failure rates in initial 2002 I/M tests for HC/CO high-emitters...... 141
5.19. Repeat failure rates in the 2002 initial tests for HC/CO type fail/pass cars during the 2000 cycle based on repair parts (model year 1992- 1995) ...... 142
5.20. Part repair rates for HC/CO/NOx high-emitters (1,095 cars) repaired during the 2000 I/M cycle...... 145
5.21. Failure rates in initial 2002 I/M tests for HC/CO/NOx high-emitters...... 146
5.22. Failure rates and repair durability based on manufacturers...... 147
5.23. Emissions breakdown in the initial 2000 test based on high-emitter types...... 156
5.24. Breakdown of car fleets based on the initial 2000 and 2002 test results ....158
5.25. Emission factors for the calendar year 2000 measured by the IM147 cycles and those constructed based on the EPA’s in-use emission survey...... 163
5.26. Relationships between new cars’ mileage versus retired high-emitters’ mileage...... 165
5.27. Fleet distributions in 2000 and 2001 if all high-emitters are scrapped...... 167
x LIST OF TABLES
Table
2.1. Upstream environmental burdens for 1 kg of gasoline in 1995 (USAMP 1999)...... 27
2.2. Life cycle energy consumption for producing 1 kg of materials (MJ) ...... 36
2.3. Net shipments of steel mill products and primary aluminum production in the US...... 36
2.4. Total inputs of energy for heat, power, and electricity generation by iron and steel industries (SIC 331)...... 37
2.5. Total inputs of energy for heat, power, and electricity generation by primary aluminum production (SIC 3334) ...... 38
2.6. Energy conversion efficiencies of major energy types...... 39
2.7. Estimated primary energy consumption for steel industry (1012 Btu) ...... 39
2.8. Estimated primary energy consumption for primary aluminum industry (1012 Btu)...... 40
2.9. Estimated energy intensities and indexes for the steel and primary aluminum production (excludes some upstream processes)...... 40
2.10. Forecast of energy consumption per unit of output (1000 Btu/1992 dollar of output) in industrial sectors...... 43
2.11. Normal-emitter's FTP type emission factors (EN) with model year and mileage...... 50
2.12. Federal Certification Exhaust Emission Standards for light duty vehicles (g/mi) and estimated headroom values (%) for cars (EPA 2000; EPA 2000; Cullen 2001)...... 51
3.1. Optimal vehicle lifetimes and cumulative environmental burdens for a 36-year time horizon between 1985 and 2020...... 72
xi 4.1. Optimal new car production and emission savings at 2001 based on the survival rate scenarios (It is assumed that 100 new vehicles have been produced each year between 1981 and 2000.)...... 99
5.1. The numbers of cars that were tested in 2000 and 2002 initial tests ...... 119
5.2. IM147 cutpoints for gasoline cars (g/mi)...... 122
5.3. Breakdown of failed pollutants in the initial 2000 tests based on outcomes of initial 2002 tests ...... 124
5.4. Average emission profiles (standard deviations) in 2000 initial tests, 2000 retests, and 2002 initial tests based on the 2000 I/M results...... 126
5.5. Missing data rates in the VIRs for failing vehicles during the 2000 I/M cycle...... 133
5.6. Repair parts and costs for the fail/pass cars during the 2000 I/M cycle .....134
5.7. Emissions profiles of HC/CO high-emitters across the 2000 and 2002 I/M cycles based on repair parts ...... 143
5.8. Average emission factor profiles for cars tested during both the 2000 and 2002 I/M cycles...... 150
5.9. Fleet emissions (based on VMT profiles with age) in calendar year 2000 and 2002 for the cars tested during both 2000 and 2002 I/M cycle...... 151
5.10. Emission factors for all cars tested in the initial 2000 and 2002 tests...... 153
5.11. Fleet emissions in calendar year 2000 and 2002 for all cars tested in 2000 and 2002 initial tests (ton/year) ...... 154
5.12. HC/CO and NOx high-emitter contributions to fleet emissions...... 157
5.13. Fleet emissions (reductions) in 2002 based on I/M scenarios in 2000 (ton)...... 161
5.14. Life cycle emission reductions and additional car manufacturing requirements to offset VMT losses from scrappage scenarios ...... 168
xii LIST OF APPENDICES
APPENDIX
A. DYNAMIC LIFE CYCLE INVENTORIES ...... 180
B. SELECT DYNAMIC LCI PARAMETERS...... 205
C. FLEET OPTIMIZATION MODEL SOURCE CODE...... 209
D. BREAKDOWN OF HIGH-EMITTERS BASED ON MODEL YEARS...... 217
xiii ABSTRACT
SHAPING SUSTAINABLE VEHICLE FLEET CONVERSION POLICIES BASED ON LIFE CYCLE OPTIMIZATION AND RISK ANALYSIS
by
Hyung Chul Kim
Co-Chairs: Gregory A. Keoleian and Jonathan W. Bulkley
Although recent progress in vehicle technology and regulation has improved the environmental performance of new model vehicles, the continuing use of old, high- polluting vehicles contributes to air quality issues. While scrappage programs attempt to reduce emissions from old, high-emitting vehicles, life cycle assessment (LCA) studies show that scrapping old vehicles and manufacturing new vehicles also account for significant life cycle emissions. The expected median lifetime of automobiles increased from 12.5 years for model year 1980 to 16.9 years for model year 1990. But it is unclear whether this trend is optimal from an energy and environmental perspective. This study combines the LCA method with mathematical tools such as dynamic programming to determine optimal vehicle replacement/retirement policies. A life cycle optimization (LCO) model is developed and applied to mid-sized generic cars based on driving 12,000 miles annually, over a 36-year time horizon (between 1985 and 2020). For CO, NMHC, and NOx, automobile lifetimes ranging from 3 to 6 years are optimal for 1980s and early 1990s model years, while optimal lifetimes are 7 to 14 years for model year 2000s and beyond. On the other hand, a lifetime of 18 years minimizes cumulative life cycle energy use and CO2 emissions.
1 In addition, both ideal and practical fleet conversion policies were investigated from a life cycle perspective. According to the simulation results, accelerated scrapping policies increase greenhouse gases but reduce regulated emissions. These results are consistent with the results of the LCO model. Maintaining old vehicles in good condition would be another effective strategy for reducing emissions from high-emitting vehicles. The inspection and maintenance programs (I/M) have been used to identify and repair high-emitting vehicles. Based on Fault Tree Analysis (FTA) of emission control systems together with IM147 test records in the Arizona area, this study identifies limitations of specific repairs (e.g., catalytic converter repairs) for I/M improvement. Benefits of I/M programs and policy scenarios combining scrappage programs are also explored. The models used in this study can help policy-makers, manufacturers, and consumers understand the importance of issues regarding vehicle retirement decisions and guide decision-makers to environmentally sound fleet conversion strategies.
2 CHAPTER I
INTRODUCTION
Environmental issues associated with automotive transportation are diverse and encompass various time and spatial scales. Vehicle exhaust emissions such as carbon monoxide (CO), oxides of nitrogen (NOx), hydrocarbon (HC), and particulate matter (PM) contribute to local and regional air pollution. With a high vehicle density, major urban areas in the world are often exposed to smog and other air pollutants that cause human health and environmental issues (WBCSD 2001). One of the major concerns of smog is its ozone (O3) content. Ozone is formed from reactions between NOx and HC in the presence of sunlight, and causes eye irritations, respiratory discomfort, and reduced pulmonary function (Mondt 2000). In order to address these issues, federal certification standards have been regulating new model vehicles’ exhaust emissions since the 1960s. Currently, in the United States, the Tier 1 emission standards are in effect for cars and the Tier 2 standards are scheduled to phase in from 2004. These standards prompted major technological innovations for emission controls such as catalytic converters, exhaust gas re-circulation, and computer-based sensors and engine controls (Ross et al. 1995). In addition to these efforts for improving conventional vehicle systems, the California Air Resources Board (CARB) established a low-emission vehicle initiative program during the mid- and late-1990s by defining emission standards for transitional low-emission vehicles (TLEVs), low-emission vehicles (LEVs), ultra-low-emission vehicles (ULEVs), and zero-emission vehicles (ZEVs). Moreover, reliance on fossil fuels as a major source of transportation energy poses several environmental challenges. Automotive transportation is a major source of
1 greenhouse gases. In the United States, gasoline for automobiles alone accounts for 20%
of annual CO2 emissionsthe most dominant greenhouse gas. In addition, cars and light trucks account for a large fraction (38% in 2000) of US petroleum consumption (DOE 2002). The Energy Information Administration (EIA) predicts that world oil production will decline between 2021 and 2112, depending on assumptions regarding resource estimates and production growth rates (Wood and Long 2000). In the US, the predicted depletion of oil resources can pose a significant threat to energy and economic security due to the large share of imported foreign oil (54% in 2001) in overall petroleum consumption (EIA 2002). To improve energy efficiencies while maintaining comfort, safety, and driving performance, manufacturers have been exploring different propulsion systems and
fuelsconventional internal combustion engines (ICE), hydrogen fuel cells (FC), electric batteries, compressed natural gas (CNG), methanol, etc. One study estimates that, compared with a baseline fuel economy of 27.8 mile per gallon (mpg) in 1996, a hybrid FC vehicle is expected to run at 94.1 mpg while conventional ICE vehicles will be able to achieve 43.2 mpg in 2020 (Weiss et al. 2000). On the other hand, a recent European
study shows that well-to-wheel CO2 emissions and energy consumption of advanced ICE will be comparable to FC vehicles in the future, especially under cold start conditions (Ellinger et al. 2002).
Despite progress achieved through regulations and technology, air quality in many urban areas in the US is still far from satisfactory, and transportation energy consumption is increasing every year. These contrasts can be explained, in part, by increasing vehicle miles traveled and continued use of old, inefficient, high-emitting vehicles. In particular, a small number of high-emitting vehicles account for significant fractions of ozone precursors and toxic gases. The transition to a more sustainable transportation system requires a fleet conversion policy that efficiently absorbs new, clean technologies and retires old, high-polluting technologies.
2 Life extension strategies have been widely practiced as a resource management tool to reduce environmental burdens associated with product manufacturing. Designs that facilitate reuse, recycling, and remanufacturing have been applied to extend lifetimes of many product systems, including auto parts, manufacturing machines, and food containers (EPA 1993). On the other hand, accelerated scrapping of old, inefficient
carsa strategy directly contradictory to life-extension strategiesaims at reducing environmental burdens associated with product (vehicle) usage (U.S. Congress 1992; ECMT 1999). This study explores a new scheme for evaluating optimal automobile lifetime based on minimization of environmental burdens during the complete vehicle life cycle. Both the improving efficiency of new technologies and the deteriorating efficiency of old technologies are the driving forces for retirement/replacement decisions, while the environmental burdens from additional vehicle scrapping and manufacturing tend to offset the benefits of the retirement/replacement decisions. Another strategy to reduce the environmental burdens from the vehicle use stage is to maintain the emission performance of old vehicles through regular inspections and repairs. Although the Inspection and Maintenance (I/M) programs are used to identify and repair high-emitters, the effectiveness and benefits of these programs are not well understood at this point. In order to understand the vehicle deterioration and failure mechanisms, this study explores the paths to emission control system failures. Based on a real-world database together with theoretical fleet characterizations, this study investigates the benefits of maintaining and/or scrapping old vehicles.
1.1 High-Emitters
In the last few decades, automotive fuel consumption and exhaust emission per mile have decreased remarkably in the United States due to improvements in engine
3 technology, emission controls, and efficient vehicle design. This progress has been primarily driven by regulations including federal tailpipe standards and CAFE standards. These standards attempt to regulate mostly new vehicles based on laboratory tests such as the Federal Test Procedure (FTP). In spite of this recent progress, vehicle emissions1 are still significant in most urban areas. The causes behind this discrepancy include malfunctions of emission controls, and increasing vehicle miles traveled (VMT), and the fact that actual driving conditions differ from laboratory test conditions. In the real world, vehicles are often driven at higher power than the federal test conditions, and therefore emit much more than during the tests. Moreover, emission control systems of cars typically deteriorate and sometimes fail with increasing vehicle mileage or age. The exhaust emissions from properly maintained and driven cars generally increase gradually. The emission systems of such cars, called normal-emitters, deteriorate in proportion to mileage (Austin and Ross 2001). On the other hand, emissions from poorly-maintained and poorly-driven cars often jump abruptly. The emissions per mile from these failed cars, called high-emitters, are considerably higher than the certification standards. The probability that a car is a high-emitter also increases with increased mileage. A number of independent analyses indicate that a small fraction of the high- emitters account for a significant proportion of real-world automotive emissions. An early study sponsored by the California Air Resources Board (CARB) showed that on- road hot-exhaust emissions were predominated by 10-20% of gross polluters. In particular, the highest emitting 3% of the vehicles contributed 23% and 27% of CO and HC based on a roadside survey study in California (CARB 1994). The significance of high-emitters can be evidenced in real-world emissions, which include on-cycle exhaust, off-cycle exhaust, evaporative, malfunction, and upstream emissions. One study shows
1 In this study, the term “emissions” does not include CO2 or H2O emissions unless specified.
4 that the malfunction of emission control systems is responsible for 6 g/mi, 0.6 g/mi, and 0.8 g/mi of CO, HC, and NOx over 17 g/mi, 1.7 g/mi, and 1.8 g/mi of lifetime average real-world emissions for a model year 1993 car (Ross et al. 1995). Note that the 1993 tailpipe emission standard was 3.4 g/mi, 0.41 g/mi, and 1.0 g/mi for CO, HC, and NOx respectively. Remote sensing studies also support the significance of high-emitters. Between 1998 and 2000, three rounds of on-road remote sensing measurements in the Phoenix area concluded that the dirtiest 10% of the entire fleet accounts for 71-78%, 66- 79%, and 49-56% of the CO, HC, and NO, respectively (Bishop et al. 1999; Pokharel et al. 2001; Pokharel et al. 2002).
1.2 Scrappage Programs
Both to reduce the emissions from high-emitters and to reduce excessive gasoline consumption by inefficient vehicles, many countries have attempted to scrap old vehicles before their economic lifetimes have been exhausted. Scrappage programs recruit and scrap high-polluting vehicles in exchange for some compensation to the vehicle owners. In a ‘cash-for-scrappage’ program, vehicle owners receive awards simply by scrapping high-emitters. On the other hand, a ‘cash-for-replacement’ program requires replacement of old cars with new models in order to receive compensatory awards. Often, these awards can be traded between vehicle owners, brokers and/or industries in the form of emission credits (EDF and GM 1998; ECMT 1999). Scrappage programs also accelerate absorption of new emission technologies. By eliminating high-emitters, scrappage programs create new car demand. As a result, scrappage programs support car-manufacturing industries and may contribute to national economies (Bohn 1992; ECMT 1999). The introduction of more reliable and safer vehicles may enhance transportation safety as well (U.S. Congress 1992; ECMT 1999).
5 Examination of changes in accident statistics as a result of scrappage programs, however, has not been undertaken and is not attempted in this study. A number of European countries implemented scrappage programs based on different incentive schemes. In February 1994, France introduced the first scrappage program awarding Fr 5,000 (approximately $950) to individuals replacing cars older than ten years with new models. France ran a second program between October 1995 and September 1996, this time with variable incentives based on the size of the new car. These two programs are estimated to have retired 860,000 vehicles that would have not been scrapped as soon without such incentives. In 1994 and 1995, Spain temporarily introduced a scrappage program that awarded tax benefits ranging between Pta 85,000 and 100,000 ($630-750) for replacing a car older than ten years with a new car. The scrappage program boosted new car demands in Spain and this tax incentive program became permanent in 1997. In addition to these countries, Greece, Hungary, Denmark, Ireland, Norway, and Italy also introduced scrappage programs either temporarily or permanently. These programs aimed to eliminate two-stroke-engine cars, cars without catalytic converters, old buses, and old trucks. The incentive ranged between $500 and $1600 depending on the characteristics of both retired and new vehicles (ECMT 1999). In the US, as a result of the Clean Air Act Amendments (CAAA) of 1990 that included market-based emission reductions, scrappage programs have emerged as an attractive tool to reduce mobile source emissions. Such programs were also intended to provide stationary sources with more flexibility through emission trading with mobile sources. The Union Oil Company of California (UNOCAL), in cooperation with the California Air Resources Board (CARB) and the California Department of Motor Vehicles, first introduced a scrappage program on June 1990 in the Los Angeles Basin area, offering $700 for retiring eligible vehicles. The program was designed to eliminate pre-1971 models. As a result, over 8,000 vehicles, accounting for more than 2% of the targeted model years in this area, were retired. Later, smaller scale scrappage programs
6 were implemented in Phoenix, Chicago, and some cities in California, offering $500 to $600 per scrapped car. Unlike most European programs, ‘cash-for-scrappage,’ which does not require one to buy a new car, dominates in the US. In the US, most scrappage programs have been limited to pilot programs and have been financed by private companies seeking emission credits to meet compliance requirements on stationary sources (U.S. Congress 1992; Alberini et al. 1995; EDF and GM 1998; ECMT 1999). Currently, debates among automakers, lawmakers, and other stakeholders discuss whether such policies actually improve environmental quality or simply weaken the part supply infrastructures for restoring vehicles (Stoffer 2002). Scrappage programs suffer from several limitations. First, the environmental benefits of the scrappage programs are uncertain. Emission benefits from scrappage programs can be determined by the emission factors of the high-emitters, the avoided travels of high-emitters, and vehicle owners’ travel patterns after scrapping the high- emitters. Other than the difficulties in measuring real-world emissions from high- emitters, it is problematic to accurately estimate the avoided vehicle miles traveled (VMT) of the scrapped vehicles. Moreover, it is uncertain how the avoided VMT of the scrapped vehicles is made up by newer vehicles or existing vehicles, especially in ‘cash- for-scrappage’ programs that do not require the purchase of a new vehicle (U.S. Congress 1992). Several studies indicate that the emissions reduction is questionable without large bounties (Alberini et al. 1995; Hahn 1995). For example, an economic analysis of the scrappage program using Los Angeles as a model area estimated that HC and NOx reductions would be under 1% in the case of a $250 bounty while 2%-6% reductions are expected in the case of a $1,000 bounty (Hahn 1995). Second, such programs will bring about a shortage of old cars, which may eventually increase the price of older vehicles and may harm mobility for low-income individuals. This may also cause the migration of old vehicles from non-scrappage program areas to scrappage program areas, and the emission benefits from the program areas will be significantly reduced. Finally,
7 scrappage programs are exposed to possible fraud. For example, if the bounties for old cars are determined by emission tests, the vehicle owners or brokers can tamper with vehicles to inflate emissions in order to receive more credits (EDF and GM 1998). Several policy analysts explored a hybrid option combining a scrappage program and an inspection and maintenance (I/M) program that is described in the next section. This I/M option states that vehicles failing I/M emission tests and exceeding certain amount of repair costs should be scrapped with appropriate rewards. By selectively scrapping high-emitting vehicles, such options are likely to improve the cost/benefit ratios of accelerated vehicle retirement programs (U.S. Congress 1992; Hahn 1995; Alberini et al. 1998).
1.3 Inspection and Maintenance (I/M) Programs
An inspection and maintenance (I/M) program requires a vehicle to pass a certain emissions test before a new registration or a renewal of the registration. I/M programs have been widely implemented in the US compared to scrappage programs. Currently, in the US, more than 30 states conduct I/M programs with a variety of test processes, fees, and frequencies. The tests are conducted either annually or biennially; statewide or within specific jurisdictions; and with or without costs. States adopt various test methods depending on the programs. In 1981, the Environmental Protection Agency (EPA) first mandated inspection and maintenance requirements to register vehicles in non-attainment areas. The initial tests consisted of idle tests and 2500 rpm tests. However, these programs were questionable in identifying high-emitting vehicles because such simple tests were incapable of simulating real-world driving conditions (Mondt 2000). In 1995, the IM240
8 test cycle developed by the EPA was introduced to “enhanced” I/M program areas2 including Phoenix, AZ and Denver, CO. The IM240 tests are conducted on a dynamometer with measurement tools for HC, CO, and NOx. The IM240 cycle shares large parts with the Federal Test Procedure (FTP) in its acceleration schedule and maximum speed (Mondt 2000). Although the test can last up to 240 seconds, decisions are often made based on a shorter time period, depending on the emission performance of the vehicles tested. Many states, including California, Connecticut, Utah, and Washington, use alternative measuring cycles called Accelerated Simulation Mode (ASM) in the “enhanced” I/M areas due to the low costs for the test equipment compared with IM240. The ASM is a steady-state loaded mode test under either 50% or 25% of the maximum acceleration in the FTP (CARB 2000). Some states, like Oregon and Massachusetts, use the BAR31 cycle (developed by the California Bureau of Automotive Repair) as a surrogate to IM240, primarily because of the reduced test time and ease of interpretation. The BAR31 is a 31-second test cycle similar to IM240 in its early acceleration. It runs, however, at a maximum speed of only 48 kilometers/hour. The identical cycle may be repeated up to four times, depending on emission performance, to assure correct interpretations (Greater Vancouver Regional District 1998). Most recently, the state of Arizona has begun to use the IM147 cycle to replace IM240. The IM147, which is equivalent to the last 147 seconds of IM240, aims at reducing the test time without harming the accuracy of IM240. To prevent false results, the IM147 is designed to be repeated up to three times (EPA 1999). A new approach is now being adopted: new cars and trucks are equipped with OBD (on-board diagnostic) systems to identify emission control problems. Many states are beginning to add OBD system checks to the I/M requirements for the 1996 and later models (EPA 1999).
2 Under the Clean Air Act Amendments (CAAA) 1990, EPA classifies I/M programs either as “basic” or “enhanced,” depending on the local air quality.
9 European countries including Germany and Austria also adopted vehicle inspection programs in the early 1990s, based primarily on a no-load short test called ‘idle/fast-idle test3.’ Especially for two-way catalyst vehicles, the ‘lambda tests’ were added to the existing ‘idle/fast-idle test’ to examine the air/fuel ratio, and became mandatory in all EU countries under Directive 92/55/EC beginning in 1997 (European Commission 1998). Although a number of evaluations of I/M programs have been conducted based on different methods, the success of I/M programs is uncertain at this point. Leaving aside the fleet turnover rates in the program area, many uncertainties associated with tests make it difficult to estimate the effectiveness of the I/M programs. First, in the real world, vehicles are driven under heavier loads than during tests. Second, vehicle owners often repair emission control systems before I/M tests to avoid the inconvenience of an I/M test failure. Finally, I/M emission tests suffer from inherent uncertainties due to the random fluctuations of vehicle emissions (Wenzel 2001; Wenzel 2001). The EPA originally projected that, by 2000, 28% of volatile organic compounds (VOCs), 31% of carbon monoxide (CO), and 9% of oxides of nitrogen (NOx) would be reduced in average fleet emissions for each test cycle (EPA 1992). However, the emission test records of I/M programs seem inconsistent with the EPA’s projections, especially for HC and CO. Investigating the emissions of initial tests, after-repair tests, and final-pass tests, Wenzel estimated the reductions from the 1997 IM240 program for Phoenix, AZ to be 14%, 17%, and 7% for HC, CO and NOx respectively (Wenzel 2001). Other analyses of the database of the Arizona I/M program show consistent reduction ranges of 12-16% for HC; 15-18% for CO; and 7-8% for NOx (Glover and Brzezinski 1997; Wenzel 1999; Ando et al. 2000; Wenzel 2001). Sierra Research also estimated 9%, 13%, and 4% of HC, CO and NOx emissions from the Canadian Vancouver/British Columbia ASM
3 In ths US, an ‘idle/fast-idle test’ is called a ‘two-speed idle (TSI) test.’
10 program based on before and after repair tests (Sierra Research 1998). However, these estimations do not account for missing vehicleswhether scrapped, sold or registered in other states, with owners avoiding I/M requirements, etc.which are known to be 22- 33% of the initially-tested vehicles (Glover and Brzezinski 1997; Ando et al. 2000; Wenzel 2001). If the no-final-pass (missing) vehicles are either scrapped or sold to non- program areas after failing the tests, the emission reductions of I/M program are underestimated. Conversely, the program benefits will be harmed if the no-final-pass vehicles are driven continually in the program area, possibly using fraudulent or illegal methods. Moreover, the emission reductions achieved by I/M programs significantly disappear between I/M cycles. Based on the analysis of the Phoenix IM240 program, 40% of the vehicles that had failed the initial 1995 I/M tests and eventually passed the tests later in 1995, again failed the initial 1997 I/M tests. Furthermore, half of the repeated failures across the two I/M tests suffer from an identical set of excessive emissions. This suggests that the durability of I/M repairs may be questionable. In other words, many repairs are simply temporary measures to pass the emission tests. More importantly, although a majority of the vehicles (more than 80%) pass their initial tests, the normal deterioration of these initially-passed vehicles may diminish most of the benefits gained in an I/M program cycle (Wenzel 2001). Considering the difficulties discussed above, remote sensing can provide more accurate information on the effectiveness of the I/M programs, since it measures tailpipe emissions in real driving conditions. It can also detect those vehicles driven in the program areas but avoiding I/M tests. Overall, remote sensing studies show that the I/M benefits are smaller than those calculated from the test records. Based on remote sensing studies of the Denver area, the CO reduction was estimated to be 4%±2% for each I/M cycle since 1995, compared with the 20-23% of reduction claimed by the State of Colorado. Moreover, no reduction has been observed for HC and NOx emissions
11 (Stedman et al. 1997; Stedman et al. 1998). A remote sensing study of the Phoenix area also indicated smaller benefits of HC and CO emissions, 11.5% and 6.9% respectively, than the calculated reductions from the I/M records (Wenzel 2001). Remote sensing is often criticized because it measures only a single second of an emission profile while driving conditions change every second in real driving. Inherent errors associated with remote sensing measurements are inevitable as well. To address these issues, multiple measurements are often conducted for a vehicle and a large number of samples are recommended. A minimum of 20,000 valid readings is required for the most basic remote sensing device (RSD) analysis. Furthermore, adequate coverage to obtain measurements representative of the fleet is also a pragmatic challenge.
1.4 Problem Statements
As described above, the efforts to reduce mobile emissions through fleet management policies such as scrappage programs and I/M programs pose significant limitations in their implementation and evaluation. This study identifies several key issues of the scrappage and I/M programs and develops models to address these issues. First, most studies on scrappage programs focus on the emissions of vehicle operations and fail to consider other sources of emissions: dismantling old vehicles and producing new vehicles. A recent study by the European Conference of Ministers of Transport (ECMT) concluded that scrappage schemes have both positive and negative impacts on the environment. The positive impacts include the reduction of atmospheric emissions when new vehicles replace old vehicles. The negative impacts, on the other hand, arise from the increased emissions and resource consumption from the additional vehicle scrapping and production processes (ECMT 1999). Wee et al. also argued that scrapping old cars would result in a net increase of life cycle energy use and CO2 emissions unless fuel economies improve considerably for the cars manufactured in the
12 future (Wee et al. 2000). Thus, the benefits or emission credits created through scrappage programs might be exaggerated since the negative impacts of vehicle production and disposal were ignored. Moreover, the negative impacts associated with vehicle production and disposal may be transferred to other areas with vehicle manufacturing and disposal facilities. Second, current studies on the scrappage programs fail to provide a normative framework for a scrappage decision in terms of environmental criteria. Thus, for example, questions such as how many vehicles should be scrapped from a fleet; whether a scrappage program is necessary if emission control systems become more durable; or whether more scrappage programs are needed if cleaner and efficient alternative fuel vehicles are available in the future, remain unanswered. Optimal scrappage decisions may depend on the relative emission performances between the retiring and new vehicle technologies rather than the classification of high- or low-emitters based on emission measurements. In fact, vehicle replacement decisions have been widely studied in the context of economic benefit optimization based on dynamic models. Several studies investigated optimal business strategies regarding purchasing and selling fleets of buses, trucks, and cars. Optimal decisions take into account salvage values, tax laws, fuel and maintenance costs, depreciation, and the price of new models. Researchers use dynamic programming or linear programming tools to calculate optimal decisions for maximizing cash flow up to certain future time (Waddell 1983; Simms et al. 1984; Bean et al. 1994). Extending such tools to environmental considerations allows vehicle owners, policy makers, and manufacturers to understand the underlying environmental impacts of the current retirement practices. In the real world, vehicle retirement is decided by economic considerations, including repair cost, market price, and scrap price of a used vehicle; and technical considerations, including engine size, vehicle age, and manufacturers (Manski and Goldin 1983; Berkovec 1985). The expected median lifetime of an average car has increased from 12.5 years for model year 1980 to 16.9 years for
13 model year 1990 (DOE 2002). However, the environmental impacts of such trends are not yet well understood. Finally, although I/M programs have been successful in identifying high-emitters through comprehensive test cycles, it is still unclear whether the identified high-emitters are repaired properly. A significant fraction of the vehicles that fail an I/M test repeats the identical types of emissions failures in the next cycle of I/M tests (Wenzel 2001; Wenzel 2001). Moreover, few investigations have been conducted on the causes and types of emission failures, especially for repeating high-emitters. Although studies have investigated the frequencies of specific component failures, a systematic approach to relating a component failure with other component failures has not been attempted (Heirigs and Austin 1996; Harrington et al. 2000; Eastern Research Group 2002). Such an approach, if possible, would help increase the benefits of I/M program since the benefits diminish due to repeated component failures or deterioration after alleged successful repairs.
1.5 Dissertation Outline
To consider complete life cycles of vehicles, this study uses a Life Cycle Assessment (LCA) tool, which is based on emissions, energy use, and resource consumption from the entire vehicle life cycle: materials production, component manufacturing and vehicle assembly, and maintenance and disposition. Among the environmental measurements, non-methane hydrocarbons (NMHC), carbon monoxide
(CO), oxides of nitrogen (NOx), carbon dioxide (CO2) and energy consumption are the objectives to be considered for this study. NMHC, CO and NOx are regulated auto emissions and the significance of these pollutants depends on the air quality of each location. On the other hand, CO2 and energy consumption are global issues influenced by CAFE standards in the US.
14 Although there has been extensive research on the environmental impacts of product life cycles, most studies have been performed based on a quantified performance of a product system called ‘functional unit’ (ISO 1997). In order to determine the optimal vehicle retirement age to minimize environmental impacts, the LCA measurements need to be conducted on a vehicle age and model year basis. For this purpose, a new method to estimate life cycle emissions of different model year cars has been developed. The estimated inputs and outputs (emissions and energy consumption) associated with different model year cars and ages are modeled by the “Dynamic Life Cycle Inventories (LCIs),” and Chapter 2 presents the procedures for the method. Dynamic LCIs are estimated for mid-sized generic cars with model years between 1985 and 2020 and for the ages of up to 20 years. Dynamic LCIs are indexed by the variables of model year and age and are used as inputs for exploring the optimal vehicle lifetimes and fleet scrappage policies in Chapters 3 and 4. In Chapter 3, optimal vehicle lifetimes are investigated in terms of environmental objectives by applying an engineering optimization tool called ‘dynamic programming’ to the Dynamic LCIs of mid-sized internal combustion engine vehicles. Sets of optimal lifetimes for each pollutant as well as energy criterion are determined and compared with each other. The determinants for optimal vehicle life are discussed based on the Dynamic LCI trends. A multi-objective analysis is undertaken to identify a globally optimal scrap policy. In addition, policy recommendations for improving vehicle energy and emission policies are provided for consumers, manufacturers, and policy makers based on the simulation results. Chapter 4 presents a fleet-based model to optimize scrapping of older cars or a "fleet conversion policy" for internal combustion engine vehicles in the US. The optimal policy minimizes each regulated emission (CO, NOx, and NMHC) as well as greenhouse gas emission (CO2) from a fleet of vehicles, given a particular mileage total for the fleet. The Dynamic LCIs developed in Chapter 2 describe the emission profiles for each model
15 year and age. The results of the simulation provide the threshold ages for vehicle scrapping and the number of new vehicles to replace the scrapped vehicles. It is assumed that only the age (mileage) of the vehicle, not the condition of its emissions control, is involved in the decision to scrap. In Chapter 5, a Fault Tree Analysis (FTA) is performed to illustrate the path to the different types of high-emitters such as CO, HC, and NOx high-emitters or any combined types of the high-emitters. The FTA is a risk analysis tool used to trace the root causes of failures by deductive processes. Appropriate repair strategies are discussed for each high-emitter type based on the FTA diagram. Then, the Arizona IM147 test and repair records are analyzed as a way to verify the FTA. The durability of the repairs is assessed by tracing the failed and repaired vehicles over the 2000 and 2002 I/M cycles4. The effectiveness and benefits of the I/M program are discussed based on current I/M programs and possible I/M repair strategies. Finally, a hybrid strategy, combining I/M and a scrappage program, is explored. Chapter 6 concludes this study and provides implications of the results. The Appendix includes the Dynamic LCI results, select parameters, and the computer codes for the model in Chapter 4.
4 The Arizona I/M program requires a biennial vehicle test.
16 CHAPTER II
DYNAMIC LIFE CYCLE INVENTORY
2.1 Vehicle LCA
The International Organization for Standardization (ISO) defines Life Cycle Assessment (LCA) as "compilation and evaluation of the inputs, outputs and the potential impacts of a product system throughout its life cycle"(ISO 1997). LCA provides the environmental profile of a system in which burdens and impacts are distributed across the life cycle. Since LCA uses a systematic and comprehensive approach to assess environmental burdens associated with products, it has been used as an analytic tool for pollution prevention, life cycle design and life cycle management (Keoleian 1995; Keoleian 1995; Keoleian et al. 1997). The complete LCA framework includes four phases: goal and scope definition; inventory analysis; impact assessment; and interpretation (ISO 1997). The goal and scope definition defines the purpose, audience, and system boundaries. The inventory analysis involves data collection and calculations to quantify material and energy inputs and outputs of a product system, and the impact assessment evaluates the significance of potential environmental impacts based on the inventory analysis. Finally, the interpretation evaluates findings, reaches conclusions and makes recommendations. A vehicle life cycle consists of the following generic stages: materials production, manufacturing and assembly (hereafter referred to as manufacturing), use, maintenance and repair (hereafter referred to as maintenance), and end-of-life. The life cycle profile shows that the distribution of burdens across each life cycle stage may differ depending
17 on the type of environmental burden. According to the US Automotive Materials Partnership (USAMP) study which measured the life cycle inventories of a generic 1995 mid-sized car, the use phase contributes 85% of the total life cycle energy consumption (see Figure 2.1) based on 120,000 miles of service life (USAMP 1999). On the other hand, the use phase contributes only 19% of the total solid waste produced while the materials production phase contributes 58% of the total solid waste produced. Most LCA studies on vehicle systems were conducted in the 1990s. Automotive parts and components were the topics of the early stage investigations in vehicle LCA. The topics included engine oil filters, aluminum intake manifolds, instrument panels, and fuel tank systems, and the LCA studies provided environmental criteria for the Life Cycle Design of vehicle components (Keoleian 1995; Kar and Keoleian 1996; Keoleian et al. 1997; McDaniel 1997; Keoleian et al. 1998; Stephens et al. 1998; Shen et al. 1999). Vehicle LCA then moved to investigating total vehicle systems with a variety of powertrains, such as Internal Combustion Engine Vehicles (ICV), Electric Vehicles (EV), Compressed Natural Gas Vehicles (CNGV), and Fuel Cell Vehicles (FCV). A study by Schuckert et al. compared the life cycle energy and emissions of a small car (650 kg) and an automobile of upper standard (2000 kg). The study revealed that the upper standard consumes nearly 900 GJ of life cycle energy compared with 312 GJ for small car based on the 200,000 passenger kilometers as a functional unit (Schuckert et al. 1995). Sullivan and Hu also introduced a life cycle energy model for ICV and EV. The study showed that the EV model investigated consumes 24% less life cycle energy than a gasoline ICV, although the ICV has many advantages in costs, accelerations, refueling times, and driving ranges (Sullivan and Hu 1995). LCA studies also explored future transportation strategies. The Energy Laboratory at MIT conducted an LCA study of new and future automobile technologies. The study estimated life cycle energy use, greenhouse gas emission, and costs associated with various technologies in 2020 including ICV, FCV, EV and hybrid vehicle systems.
18 In the report, the hybrid hydrogen FCV and hybrid gasoline FCV consumes 72% and 104% of life cycle energy compared with the baseline gasoline ICV for 2020 (Weiss et al.
2000). Most recently, Ellinger et al. compared life cycle CO2 emissions and energy consumptions of various powertrains of the future. The study showed that the optimized diesel drive train is the most energy efficient and the CNGV vehicles produces the lowest
CO2 emissions. FCVs are as efficient as the diesel and CNGV vehicles only under hot start conditions (Ellinger et al. 2002). The most comprehensive LCA study of a total vehicle was the 1995 USAMP project. A generic vehicle was conceptualized as a synthesis of three 1995 comparable mid-sized vehicles: the Dodge Intrepid, the Chevrolet Lumina and the Ford Taurus. Life cycle inventory models were developed for each detail of the materials and components of the generic vehicle, primarily based on North American manufacturing and vehicle operation (Sullivan et al. 1999; USAMP 1999).
2.2 Dynamic LCI Conceptual Structure
Vehicle LCA studies have mostly focused on measuring the environmental performance of specific model year vehicles or propulsion systems. However, such methods fall short of describing high or low emitters since these studies report average environmental performances based on functional units (e.g., 120,000 miles of driving). In order to compare the environmental performance between old, retiring vehicles and new replacement vehicles in the context of scrappage programs, LCA models need to be developed for each model year as a function of vehicle age. In particular, this section presents a dynamic LCI method for the generic vehicle defined in the USAMP project. The USAMP generic vehicle is different from an average mid-sized vehicle since it is a hypothetical synthesis of three US mid-sized car models.
19 1000
800
600
400 Energy (GJ)
200
0 r ion bly se ai ife ct m U ep f-L du se R -o ro As & nd P & ce E t'ls g. an a Mf en M int Ma Life Cycle Stage
Figure 2.1: Life cycle energy consumption of a generic mid-sized passenger vehicle based on 120,000 miles of driving (USAMP 1999)
The dynamics of environmental performance based on a particular model year are described in Figure 2.2. The environmental performance of the life cycle at the bottom of Figure 2.2 changes as the three dynamic LCI factors at the top of Figure 2.2 change with each model year. The regulatory/socioeconomic factors describe demographic, regulatory, and macro-economic changes. The technology improvements affect emission controls, powertrain, and other vehicle designs for energy efficiencies. The deterioration of a vehicle is often determined by driving conditions and maintenance. For most of the model years, changes in one dynamic LCI factor influence other factors. For example, technology improvements may enhance the durability of emission control systems such as
20
Regulatory/ Socio- Technology Factors Deterioration economic Improvement Behavior Factors
Parameters Recycled Materials VMT Energy Fuel Emission Component Content Use Intensity Economy Factor Reliability 21
Life Cycle Stage Materials End-of-Life Production Manufacturing Use Maintenance
Figure 2.2: Factors, parameters, and life cycle stages illustrating the dynamic LCI for a generic vehicle
catalytic converters and, accordingly, change vehicle deterioration behavior. The LCI parameters in Figure 2.2 are defined to reflect the evolution of these dynamic LCI factors. These LCI parameters often change by evolutions of multiple factors. Finally, parametric changes for a model year (i) and vehicle age (j) affect the life cycle environmental burdens (BM(i), BA(i), BU(i,j), BR(i,j), and BE(i,j)). For example, BU(i,j) is the environmental burden for the use phase where i and j represent, respectively, model year and age.
2.3 Dynamic LCI Modeling for a Generic Vehicle
The dynamic LCIs in this study account for the five life cycle phases: materials production, manufacturing, use, maintenance, and end-of-life. For each phase, energy
consumption, CO2, CO, NOx, and NMHC emissions have been obtained as functions of vehicle age and model year. CO, NMHC, and NOx are regulated auto emissions that can
have a significant impact on local and regional air quality. On the other hand, CO2 and energy consumption are global issues influenced by CAFE standards in the US. Other environmental burdens, such as particulate matter (PM), methane, and air toxics are not considered, primarily due to data limitations. For instance, since the current federal emission standards do not regulate PM for cars, estimations regarding the PM profiles of past and future model year cars are highly uncertain. The dynamic LCIs in this study build on the LCI analysis of a generic vehicle conducted by the US Automotive Materials Partnership (USAMP). The analysis measured the raw materials use, emission and waste outflows, and energy use from the life cycle of a 1995 mid-sized generic vehicle, a hypothetical synthesis of the Dodge Intrepid, the Chevrolet Lumina, and the Ford Taurus (USAMP 1999). Except for the use phase, the dynamic LCIs for this study have been determined based on the LCI of a 1995 generic vehicle and the LCI parameters of each model year and vehicle age. Thus, the
22
dynamic LCIs for model year i and age j, B(i,j) can be derived from the USAMP LCI
(BU) using parameter P(i,j):
∏ Pijk (), kI∈ U B()ij, =×∗ B ∏ Pk kI∈ where P* denotes the baseline parameter value applied to the 1995 generic vehicle. Note that k and i represent, respectively, the index for the parameter and the set of parameters defined for each life cycle stage. The detailed form of the parameter in the above equation varies according to the life cycle stage. For example, while total vehicle weight represents the LCI parameter for the manufacturing stage, detailed material composition data are used for characterizing the materials production stage. In addition, surrogate parameters are developed for modeling purposes. For instance, environmental burdens to produce a unit mass of materials are characterized based on the recycling scenario for each model year. These burdens represent the recycled content parameter in Figure 2.2. The dynamic LCIs for the use phase have been modeled separately from the USAMP study. The detailed modeling processes of dynamic LCIs for generic US mid-sized passenger cars with internal combustion engines between model years 1985 and 20205 are described in the following sections.
2.3.1 Materials Production
The materials production phase, which makes the second largest contribution to the total environmental burdens of vehicle life cycle, is reported to account for 9% of energy consumption and 8% of CO2 emission over the life cycle of a generic vehicle
5 While Chapter 3 of this study uses the dynamic LCIs for the model years between 1985 and 2020, model runs in Chapter 4 are based on the dynamic LCIs for model years between 1981 and 2020. Dynamic LCI models in this chapter are readily expandable up to model year 1981, although the uncertainties associated with emission factors of early 1980s model years are unavoidable.
23
(USAMP 1999). This phase of life cycle includes raw material acquisition, materials transportation, and material processing prior to the component manufacturing and assembling of vehicles. As shown in Figure 2.2, the materials use and energy intensity6 determine the materials production life cycle inventories. The materials use parameter specifically refers to vehicle materials composition. The following notations and equations determine the materials production environmental burdens for each model year and age for the generic vehicle.
m(a,i) : Mass of material a for model year i vehicle
EI(a,i) : Energy intensity for producing material a for model year i vehicle.
R(a,i) : Composite LCI for producing a unit mass of material a based on the recycled content of material a in calendar year i.
i* : Baseline year (=1995).
U BM (a) : USAMP environmental burden for producing material a .
For each environmental category (CO, HC, NOx, energy consumption, etc.), the environmental burdens to produce material a for the generic vehicle of model year i,
BM(a,i) is determined as follows.
mai(,) EIM (,) ai Rai (,) U BM (,)ai=×∗∗∗ ×× BM () a mai(, ) EIM (, ai ) Rai (, )
For each environmental category, the total environmental burden to produce complete vehicle materials at each model year i is obtained by aggregating burdens from all of the materials.
6 This can be called “emission intensity” if used for estimating dynamic life cycle emissions. However, due to difficulties in data collection, energy intensities are used as surrogates in most cases. This assumption can be justified by the fact that fossil fuels account for major fraction (~85%) of primary energy consumption in the US, and most air emissions are associated with combustion of fossil fuels (EIA 2002).
24
BM (i) = ∑ BM (a,i) a
2.3.2 Manufacturing
The manufacturing phase is classified into three types of activities: parts fabrication, component manufacturing and vehicle assembly. Parts fabrication represents the primary transformation of engineered materials including stamping, casting, molding, and forging. Discrete parts often require many other processing steps, such as machining, surface treatment, welding, fastening, etc. Component manufacturing involves further processing and assembly of parts into higher-level components. Final vehicle assembly integrates all parts, components, and fluids together into the final vehicle. As shown in Figure 2.2, manufacturing phase dynamic LCIs are associated with two parameters: materials use and energy intensity. Unlike the materials production phase, the materials use parameter here specifically refers to a total vehicle weight. The environmental burden of this phase is assumed to be linearly dependent on the total weight of the vehicle. The manufacturing phase dynamic LCIs are then estimated as functions of the model years. As noted in Figure 2.2, vehicle total weights are influenced by the regulatory/socioeconomic factors and technology improvements. CAFE regulations and development of light materials have driven weight savings in the past. Consumer awareness of environmental issues is another factor (although currently weak) in weight savings (OSAT 2000). On the other hand, crash requirements, manufacturing costs, and consumer preferences for optional vehicle functions can result in additional vehicle weight. Table B1 in Appendix presents the total weights of average vehicles used for this model.
25
U BA : Environmental burden for the manufacturing phase in the USAMP.
mT () i : Average total weight of a model year i vehicle.
EIA (i) : Energy Intensity of manufacturing for model year i . i* : Baseline year (=1995).
Then the environmental burden for manufacturing at year t, BA(i) is written as follows.
mi() EI() i T A U BA ()iB=×∗ ×A mi() EI i* T A ()
2.3.3 Use
The environmental burdens associated with vehicle use can be classified as precombustion and combustion burdens. Precombustion environmental burdens are often called ‘upstream’ burdens since they consider crude oil acquisition, transportation, oil refining, and other intermediate processes required to produce and deliver fuels. Table 2.1 gives the precombustion environmental burden to produce 1 kg of gasoline used in the USAMP study (USAMP 1999). Combustion environmental emissions for the vehicle use phase are associated with vehicle tailpipe emissions7 while combustion energy consumption refers to the heating value of a fuel. The heating value8 of 1 kg of gasoline fuel is 47 MJ.
Table 2.1: Upstream environmental burdens for 1 kg of gasoline in 1995 (USAMP 1999)
Energy(MJ) CO2(g) CO(g) HC(g) NOx(g) 10.7 702.6 0.4 3.9 1.7
7 For modeling purposes, evaporative HC emissions are included in the combustion emission. 8 Higher heating value. This consists of net heat content (lower heating value) and the energy spent to generate water vapors in the combustion gas.
26
Figure 2.3 shows the detail dependencies between the parameters and environmental burdens. In this study, the precombustion and the combustion environmental burdens are functions of VMT and fuel economy parameters. In addition, the precombustion environmental burden depends on the energy intensity for producing gasoline, and the combustion environmental burden depends on vehicle emission factors.
Energy Fuel Emission VMT Intensity Economy Factor
Pre- Manufacturing Combustion Maintenance combustion
Figure 2.3: The detailed dependencies between the parameters and environmental burdens of the use phase life cycle
The following notations and equations determine dynamic LCIs for the vehicle use phase.
P BU () i,j : Use phase precombustion environmental burden for model year i and age j (=1,2,3… ).
C BU () i,j : Use phase combustion environmental burden for model year i and age j. VMT() i,j : Vehicle miles traveled per year.
27
EIg () i,j : Energy intensity for gasoline production in calendar year i+j-1.
U Bg : Precombustion environmental burden to produce 1 kg of gasoline in the USAMP study.
FE() i,j : Fuel economy for model year i and age j (mile/gal).
ef : Heat value of 1 kg of gasoline. E() i,j : Emission factor (g/mile).
i* : Baseline year (=1995).
PC BijBijBijUUU(),,,=+ () ( )
VMT(, i j ) EI() i, j BijPU(, )=×× g B UgFE(, i j ) EI i*,1 g ()
VMT(, i j ) × e f C (for energy ) BijU (), = FE(, i j ) VMT() i, j× E () i , j ( for emissions )
2.3.4 Maintenance
The maintenance phase accounts for the material production and manufacturing burdens of replacement parts. This phase includes both scheduled maintenance and unscheduled repairs. Unscheduled repairs are often associated with component failures while scheduled maintenance refers to preventive part replacements. The formulas to calculate the maintenance phase environmental burden BR(i,j), are shown as follows.
U BR : Lifetime environmental burden from the maintenance phase reported in the USAMP study.
VMT(i,j): Vehicle miles traveled for model year i and age j .
EIR (i,j) : Energy intensity of maintenance for calendar year i+j-1.
28
λ(i,j) : Environmental burden from unscheduled repairs for model year i and age j.
i* : Baseline year (=1995).
VMT() i,, j EIR () i j U BijRR(, )=××+ Bλ () ij, β EI i*,1 R () β =120,000 miles
Although the USAMP study assumes that replacement parts come from OEMs, some of the vehicle components, such as catalytic converters, tires and fluids often are supplied from the recycling market (USAMP 1999). Thus, the environmental burden
U BR might have been slightly overestimated. Unscheduled repairs are assumed to occur after 120,000 miles of driving for modeling purposes.
2.3.5 End-of-Life
The end-of-life environmental burdens depend on materials use and energy intensity parameters. The end-of-life phase consists of four components: transportation of the used car to a dismantling facility; dismantling; shredding; and disposal of the shredder residue. It is assumed that the end-of-life environmental burden is proportional to the total weight of the vehicle. The environmental burdens are modeled based on the following notation and equation.
U BE : End-of-life environmental burden by the USAMP study.
EIE (i,j) : Energy intensity for calendar year i+j-1 when the model year i vehicle retires.
m (i) : Average total weight of a model year i vehicle. T i* : Baseline year (=1995).
mi() EI() i, j T E U BijEE(, )=×∗ × B mi() EI i*,1 T E ()
29
2.4 Dynamic LCI Parameters for a Generic Vehicle
Modeling characteristics of the dynamic LCI parameters of generic mid-sized cars, including historical trends, future forecasts, and major assumptions are provided in this section.
2.4.1 Recycled Content
The average recycled content in the ferrous and aluminum materials for model year 1994 cars was estimated to be 33% (Keoleian et al. 1997). Material recycling can reduce environmental burdens by avoiding energy intensive processes associated with primary material production9. For instance, the energy consumption required to produce 1 kg of primary cast aluminum is reported to be 208.9 MJ, compared with 37.5 MJ for the recycled cast aluminum. Thus, the production of automotive cast aluminum with typical 85% recycled content consumes approximately 63.2 MJ/kg of energy (Aluminum Association 2000). The recycled content for each material or vehicle part largely depends on the status of recycling technology and infrastructure. The greatest technical hurdle of a recycling process is to economically remove impurities that can degrade the quality of recycled materials. For instance, the copper impurities in recycled ferrous materials and the paint residues in recycled plastic resin are major obstacles in automotive applications. In order to solve these problems, recycled materials are either mixed with primary materials or used as low-grade materials. Having a large enough supply of the same kind of scrap material is also critical for building recycling infrastructure. This study assumes that the recycled content of automotive materials is constant between 1985 and 2020 except for the case of aluminum. The recycling technologies and
9 More specifically, in this section, post-consumer recycled content is discussed. Post-industrial recycling is considered to be an efficiency issue in the manufacturing process, and is included in the energy intensity of the materials production phase in Section 3.4.2.1.
30
infrastructure for ferrous scraps has been well established since at least the 1980s10. In fact, steel-making processes require significant amounts of steel scraps. Recycled scrap accounts for 20-30% of the raw materials for the Basic Oxygen Furnace (BOF) steel- making process, which produces hot-rolled, cold-rolled, and galvanized coils. The Electric Arc Furnace (EAF) steel-making technology, which produces mostly steel bar, uses recycled scrap as the sole raw material.11 Currently, the BOF provides 60% of US steel and the EAF accounts for the other 40%. Although EAF technology seems to increase at the expense of the BOF, it is uncertain whether EAF will emerge as the major steel process in the near future. Moreover, with current technologies, the steel recovery ratio from mixed scrap seems to have reached its maximum current feasibility (AISI 1998). It seems reasonable, therefore, to assume that the ratio between the BOF and EAF technology will not change in the foreseeable future without a dramatic change in scrap supplies. In addition, the future of recycled content in automotive plastics seems uncertain. Although some recycled plastics are used in automobiles, they have been limited to a small number of parts, including lamp housings, knobs, and carpets due to the lower quality compared to virgin materials (Keoleian et al. 1997). The cheap virgin materials associated with cheap oil prices also reduce the economic incentives for recycled resins. Furthermore, it is often difficult to collect large enough quantities of specific plastics needed for an economic recycling business due to the many varieties of types and grades of primary plastics and plastic composites. These obstacles could be partially relieved in the near future by the commitments of OEMs or by EPA or state-led initiatives that would issue regulations restricting Automotive Shredder Residue (ASR) (Staudinger and
10 A 1987 census reported 6075 establishments of vehicle wreckers/dismantlers in the US. 11 The recycled scrap includes the scrap generated within steel plants (Home Scrap), the scrap generated during the manufacture of steel products (Prompt scrap), and the steel scrap from post-consumer products (Obsolete Scrap).
31
Keoleian 2001). However, ASR reductions through facilitating recycling practices may not lead to an increase in recycled contents in vehicle plastic materials, because recycled plastics from vehicle scrap may be used in non-automotive products. On the other hand, a significant improvement is expected for the recycled content in automotive aluminum. For model year 1995 cars, automotive cast aluminum included 85% recycled content while rolled and extruded auto parts included only 11% recycled aluminum (Aluminum Association 2000). As aluminum materials become more popular for auto parts due to weight saving effects12, more aluminum scrap will be recovered from these auto parts in the future. An abundant supply of scrap aluminum may also support aluminum recycling in the future. The Aluminum Association forecasts that, by the 2010s, 48% recycled content will be achieved for rolled and extruded aluminum while the recycled content of cast aluminum will drop to 81% (Aluminum Association 2000). If this scenario is valid, the average life cycle energy consumption for producing 1 kg of automotive aluminum will decrease from 90.9 MJ to 77.1 MJ13. These changes in recycled content of aluminum parts can be represented in the recycled content index, RI(a,i) (=R(a,i)/R(a,1995)). (See Section 2.3.1.) In this study, RI(Al,1995) and RI(Al,2020) each correspond to 1 and 0.848 (=77.1/90.9) assuming the Aluminum Association scenario will be achieved in 2020. The recycled content is assumed to linearly increase during the period. On the other hand, it is assumed that the past trends of recycled contents between 1985 and 1995 have been constant (RI(Al,1985),…, RI(Al,1995)=1).
12 The aluminum use in an average US-built car increased from 62.6 to 111.4 kg between 1985 and 2000, and is expected to increase to 115.8 kg by 2009 (Binder 2000; OSAT 2000). 13 It is assumed that 73.8% cast, 22.8% extruded, and 3.4% rolled aluminum is used in automotive applications (USAMP 1999).
32
2.4.2 Materials Use
The changes in types and amounts of materials used will result in different environmental profiles for each model year car. For example, the increasing use of automotive aluminum, on the one hand, improves fuel economy due to weight savings, and on the other hand, consumes more energy than steel during the materials production phase. As shown in Figure 2.2, the technology improvement and the regulatory/socioeconomic factors affect materials use trends. For instance, the underlying reasons for continued usage of steel (steel accounts for 54%, by weight, of the average US built cars in 2000) are new coating and alloying technologies together with cost advantages. Research and development since the mid 1980s have made steel bodies more rust resistant, thinner, and lighter. Steel is also far less expensive than other lighter alternatives, especially in high-volume production (Binder 2000). ‘Ward's Automotive Yearbook’ is the main source of materials use data between 1985 and 2000. This series includes annual reports on the estimation of average materials consumption in US-built cars14(Binder 2000). Two different sources were used to produce the data for the future forecast between 2001 and 2020: "Delphi 10" by the University of Michigan Transportation Research Institute (UMTRI) and "On the Road in 2020" by the Energy Laboratory, MIT. These reports estimate the materials composition of model year 2004, 2009 and 2020 based on the forecasts of CAFE standards, material costs and technological improvements. For instance, Delphi 10 forecasts a 10% reduction of iron and steel use by 2009 if the CAFE standard is set at 30 mpg15, and a 20% reduction if it is set at 35 mpg (OSAT 2000). Table B1 in Appendix B summarizes
14 This series has been reporting the estimation of material in a “Typical Family Vehicle” since the 1995 edition. The terms ‘U.S.-built car’ and ‘Typical Family Vehicle’ are interchangeable in this series. 15 The forecast of 30-mpg CAFE standards by 2009 was selected in this modeling to be consistent with the Annual Energy Outlook report.
33
the estimated materials composition used in this study. Linear increase or decrease is assumed to fill the data gaps between the major forecasts of 2004, 2009 and 2020.
2.4.3 Vehicle Miles Traveled (VMT)
VMT for an individual vehicle varies depending on the vehicle’s purpose, age, type, etc. According to the National Personal Transportation Survey (NPTS), the average annual miles per household vehicle in 1995 is 12,200 miles from self-reported data and 11,800 miles from odometer-based data. In a detailed analysis of the survey, a primary vehicle of a three-car household was driven, on average, 18,853 miles per year while secondary and tertiary vehicles were driven 9,806 and 4,555 miles respectively (FHWA 2001). The VMT of a fleet vehicle depends heavily on the establishments to which the vehicle belongs. For example, cars owned by the business sector were driven on average 22,780 miles, while government cars were driven on average 12,895 miles in 2000 (DOE 2002). In Chapter 3, the baseline VMT is assumed to be 12,000 miles per year, and the VMT effects on the optimal vehicle lifetimes are explored using 6,000 and 24,000 miles of VMT scenarios. In addition, a VMT may decrease with vehicle age or mileage because of deterioration, reduced reliability, and changes in vehicle purpose (e.g., from primary to secondary household vehicle). The annual average VMT with vehicle age has been studied in conjunction with the MOBILE6 emission model (EPA 1998; EPA 1999). The details regarding the age effects are described in Chapter 4.
2.4.4 Energy Intensity
The ratio between the total energy consumption and gross domestic product (GDP) is a traditional indicator of the energy intensity for an economic system. For the manufacturing energy intensity, ‘value of shipments’ or ‘value added’ has been used instead of GDP. However, these economic measures are subject to change according to
34
inflation and prices (Hyman and Reed 1995). This study attempts to use physical units rather than economic units in measuring manufacturing outputs. The energy intensities based on economic values, however, are used as surrogates for physical units when physical measurements of outputs are inappropriate or infeasible. The manufacturing energy intensity indexes, which represent the ratios between the manufacturing energy intensity of a given year and the baseline year, 1995, are developed between 1985 and 2020 for each life cycle phase. Table B2 in Appendix gives the energy intensity indexes for each life cycle phase.
2.4.4.1 Materials Production
The materials production phase of a vehicle life cycle is associated with numerous materials processing sectors, including the steel, aluminum and petrochemical industries. Moreover, raw materials acquisitions and materials transportation are often included in the materials production life cycle phase, which relates to infrastructures such as the public utility, transportation and mining sectors. This study primarily focuses on energy intensities of major auto materials processing such as ferrous materials, aluminum, and plastics production. Detailed activities associated with raw materials acquisition and transportation activities are assumed to be included in the materials processing. Different approaches are adopted for past trends (1985-1998)16 and future forecasts (1999-2020). The procedures to estimate the past trends are described in the following sections.
2.4.4.1.1 Historical Trends
Ferrous materials together with aluminum constitute more than 70% of vehicle weight (Keoleian et al. 1997; Binder 2000). Table 2.2 shows the life cycle energy consumption of steel and aluminum materials production. The energy intensity for
16 The ranges are simply based on data availability.
35
primary aluminum processing without any recycled content is assumed to represent the manufacturing energy intensity of the aluminum industry17.
Table 2.2: Life cycle energy consumption for producing 1 kg of materials (MJ) Steel Products Automotive Aluminum Hot Rolled Cold Rolled Galvanized EAF Cast Extruded Rolled 26 28 32 9.4 56 190 183 Source: (USAMP 1999)
The detailed steps for modeling manufacturing energy intensities of ferrous and aluminum materials are presented below.
1. Obtain the total US net shipments of products for each year. Table 2.3 shows the total US net shipments of steel mill products and primary aluminum production from 1985 through 1998 for select years.
Table 2.3: Net shipments of steel mill products and primary aluminum production in the US
Production 1985 1988 1991 1994 1998 Iron and Steel Net Shipments 66.2 76.0 71.5 86.3 92.6 (Million metric tons) Primary Aluminum Production 3500 3944 4121 3299 3713 (Thousand metric tons) Source: (AISI 1998; USGS 2000)
2. Obtain the total inputs of energy by the steel and aluminum industries. The most comprehensive data source on the manufacturing energy input is the Manufacturing Energy Consumption Survey (MECS) by the Energy Information Administration
17 Assumption of no recycled content is made to make comparisons between years consistent.
36
(EIA). The surveys were conducted for reporting years 1985, 1988, 1991, 1994 and 1998. The surveys include the total input of energy for heat, power, and electricity generation to the steel and aluminum industries (EIA 1997). This does not include some energy inputs in the form of feedstock or raw material. Tables 2.4 and 2.5 show the data for the available years. The MECS uses the Standard Industrial Classification (SIC) system to categorize businesses by type of activities. According to the SIC system, the industry categories for Tables 2.4 and 2.5 correspond to the industry group of SIC 331 (Blast Furnace and Basic Steel Products) and SIC 3334 (Primary Aluminum). The industry group of SIC 331 includes industries such as SIC 3312 (Steel works, blast furnaces including coke ovens, rolling mills) and SIC 3313 (Electrometallurgical products, except steel). According to the 1994 survey, the total energy inputs to SIC 3312 and 3313 correspond to 97% of the inputs to the
Table 2.4: Total inputs of energy for heat, power, and electricity generation by iron and steel industries (SIC 331) Type of Energy 1985a 1988a 1991b 1994 1998c Net Electricity (million kWh) 47730 49658 46751 53269 56478 Residual Fuel Oil (1000 barrels) 5475 5772 5002 6680 5016 Distillate Fuel Oil (1000 barrels) 961 1066 921 1068 1020 Natural Gas (billion cu ft) 435 463 421 503 479 LPG (1000 barrels) 125 301 78 104 130 Coal (1000 short tons) 2337 1684 1151 1711 2141 Coke and Breeze (1000 short tons) 21856 29987 21690 26503 24033 Other (trillion Btu)d 486 467 450 475 390 a Estimated from SIC 3312 (Steel works, blast furnaces including coke ovens, rolling mills). b Estimated from SIC 3312 and 3313 (Electrometallurgical products, except steel). c Sum of the data for NAICS 331111(Iron and Steel Mills), 331112 (Electrometallurgical Ferroalloy Products), and 3312 (Steel Product Manufacturing from Purchased Steel). d Primarily byproducts of the production of coke from coal, such as coke oven gas. Source: (EIA 1988; EIA 1991; EIA 1994; EIA 1997; EIA 2001)
37
Table 2.5: Total inputs of energy for heat, power, and electricity generation by primary aluminum production (SIC 3334) Type of Energy 1985 1988 1991 1994 1998a Net Electricity (million kWh) 61648 65818 67317 53552 57355 Residual Fuel Oil (1000 barrels) W 2 * * * Distillate Fuel Oil (1000 barrels) 52 74 127 125 * Natural Gas (billion cu ft) W 19 20 16 14 LPG (1000 barrels) W 2 42 W * Other (trillion Btu) W 1 1 W 1 a NAICS 331312 (primary aluminum) is used. *Estimated to be less than 0.5. W=Withheld to avoid disclosing data for individual establishments. Source: (EIA 1988; EIA 1991; EIA 1994; EIA 1997; EIA 2001)
SIC 331. If the data for SIC 331 are not available, the values for SIC 3312 and 3313 are extrapolated to estimate the total energy inputs to the Iron and Steel industry. For the year 1998, the values classified by NAICS18 (North American Industry Classification System) are used as surrogates.
3. Estimate the primary energy consumption for materials production. This step converts energy input to primary energy consumption, which accounts for the total fuel cycle of each energy source. For example, the conversion of the electrical power into primary energy takes into account the combustion efficiencies of fuels, the pre-combustion energies for the fuels, and the line losses from the electricity transmissions. Table 2.6 gives the primary energy consumptions per unit energy sources. Tables 2.7 and 2.8 show estimated total primary consumption for steel and aluminum industries. In order to determine the total primary consumption, the
18 On April 1997, the US adopted NAICS as the industry classification system to replace the SIC system (U.S. Census Bureau 2002).
38
missing data marked 'W' and '*' in Table 2.5 are estimated by linear weighting methods based on the known values.
Table 2.6: Energy conversion efficiencies of major energy types
Primary Energy Heat Content Type of Energy Consumption Efficiency (103 BTU) (103 BTU) Net Electricity (1,000 kWh) 3412 10621 0.321 Residual Fuel Oil (1,000 barrel) 6300 7182 0.877 Distillate Fuel Oil (1,000 barrel) 5838 6636 0.880 Natural Gas (109 cubic feet) 1.03E+06 1.16E+06 0.888 LPG (1,000 barrel) 4011 4536 0.884 Coal (1,000 short ton) 22400 22800 0.982 Coke and Breeze (1,000 short ton) 24800 28200 0.879 Other - - 0.899*
*Average value of efficiency excluding Net Electricity Source: (DOE 2000; EIA 2000)
Table 2.7: Estimated primary energy consumption for steel industry (1012 Btu) Type of Energy 1985 1988 1991 1994 1998 Net Electricity 507 527 497 566 600 Residual Fuel Oil 39 41 36 48 36 Distillate Fuel Oil 6 7 6 7 7 Natural Gas 505 536 488 583 556 LPG 0.6 1.4 0.4 0.5 0.6 Coal 53 38 26 39 49 Coke and Breeze 616 846 612 747 678 Other 541 519 501 528 434 Total 2269 2517 2166 2520 2359
39
Table 2.8: Estimated primary energy consumption for primary aluminum industry (1012 Btu) Type of Energy 1985 1988 1991 1994 1998 Net Electricity 655 699 715 569 609 Residual Fuel Oil 0.01 0.01 0.01 0.01 0.01 Distillate Fuel Oil 0.4 0.5 0.8 0.8 0.9 Natural Gas 21 22 23 19 16 LPG 0.009 0.009 0.2 0.2 0.2 Other 0.8 0.9 0.9 0.7 0.8 Total 677 723 740 589 627
4. Estimate the energy intensity per kg of steel and primary aluminum production. The total primary energy data estimated in Tables 2.7 and 2.8 are divided by the total annual production given in Table 2.3. The energy intensities are represented as mega joule per kilogram of materials production. Table 2.9 gives ‘the manufacturing energy intensities’ and ‘the manufacturing energy intensity indexes’ for the steel and primary aluminum production based on a baseline year 199819.
Table 2.9: Estimated energy intensities and indexes for the steel and primary aluminum production (excludes some upstream processes)
Industry Unit 1985 1988 1991 1994 1998 MJ/kg 36.1 34.9 32.0 30.8 26.9 Iron and Steel Index 1.35 1.31 1.19 1.15 1.00 (1998=1)
MJ/kg 204 193 190 188 178 Primary Aluminum Index 1.14 1.08 1.06 1.06 1.00 (1998=1)
19 The manufacturing energy intensity indexes are eventually determined using 1995 as the baseline.
40
The energy intensities obtained here do not consider the energy consumption for raw materials acquisition or transportation energy consumption required between processes. Although the energy intensities in Table 2.9 may be slightly underestimated due to these exclusions, these results generally agree with other studies. Ecobalance reported 32 MJ/kg and 26 MJ/kg of primary energy consumption for galvanized and hot rolled steel products (USAMP 1999). A study supported by the Department of Energy (DOE) estimated that the energy intensities for the US steel industry in 1985, 1990 and 1998 were 25, 20, and 18 MBtu/ton (= 29, 23, and 21 MJ/kg) (Stubbles 2000). A recent study of the Aluminum Association estimates the primary energy consumption of primary cast aluminum to be 209 MJ/kg (Aluminum Association 2000). Different conversion factors, data sources and classification standards between NAICS and SIC might be the sources of discrepancies between the studies. Since the chemical processes, including plastic compounding, involve a variety of organic intermediates, the energy intensity primarily depends on the structural change of products or qualities of raw materials such as crude oils. Energy intensity20 trends in the US chemical industry have been flat since the late 1980s, when inexpensive energy sources for heat, power, and feedstocks became available (DOE 2000). In fact, the EIA estimates that the manufacturing energy intensity of the US chemical industry increased slightly21 between 1985 and 1994 (EIA 1998). Based on this information, this study assumes that the energy intensities of plastic materials productions were constant between 1985 and 1998. Other materials, such as brass, rubber, glass and fluids comprise about 20% of vehicle materials. The energy intensity of a given year for producing these materials is
20 The energy intensity trends were based on economic values rather than physical measures. 21 Increasing from 11.09 to 12.14 thousand Btu/1992 constant dollars. Source: "Table 4. Energy-Intensity Ratio of Total Inputs of Energy per Value of Shipments by Major Manufacturing Groups for 1985, 1988, 1991, and 1994": The energy-intensity ratios that have been adjusted to the mix of products shipped from manufacturing establishments in 1985 (EIA 1998)
41
assumed to be the mean of the three energy intensities for producing ferrous metals, aluminum, and plastics.
2.4.4.1.2 Future Forecasts
The energy intensity of a material’s production is limited by practical or theoretical maximums of process efficiencies. Fruehan estimates the practical minimum energy requirement for producing hot-rolled steel is only 1-2% 22 of the current energy consumption while that for producing cold-rolled steel is 38-45%23of the current practice (Fruehan et al. 2000). Thermodynamic minimum energy intensities for aluminum and steel materials are reported to be 29.3 GJ/tonne and 7.1 GJ/tonne compared with the actual intensities of 88 GJ/tonne and 33.1 GJ/tonne24 in 1980 (Ross 1987). The Aluminum Association assessed the ‘Improvement Potential’ in primary aluminum processing for the North American Aluminum Industry based on 1995 performances. According to its reports, the smelting process, which consumes more than 50% of the energy for the primary aluminum process, contains only 2.1% improvement potential. In contrast, the alumina refining process, which constitutes around 10% of the energy for primary aluminum production, contains 39.1% improvement potential. Other processes, including bauxite mining, anode production and primary ingot castings, show 3-9% improvement potential (Aluminum Association 2000). Future trends of energy intensity for materials production are estimated primarily based on the Annual Energy Outlook 2001 report from the Energy Information Administration (EIA). In this report, the energy intensities are represented by ‘Energy
22 For the cold rolling process, the actual energy requirement is 1.0-1.4 GJ/t compared with 0.02 GJ/t of practical minimum. 23 For the hot rolling process, the actual energy requirement is 2.0-2.4 GJ/t compared with the 0.9 GJ/t of practical minimum. 24 These values do not include upstream energy requirements and consider only primary aluminum.
42
Consumption per Unit of Output (1000 Btu/1992 dollars of output)25’ (EIA 2001). Table 2.10 shows the future forecast of energy intensities for the selected materials production industries based on the Annual Energy Outlook 2001. For the materials industries not specified in Table 2.10, average profiles of the three industries are used. Table B2 in Appendix B gives the complete energy intensity indexes between 1985 and 2020. It shows a 32% and 17% decrease in the energy intensity of steel and aluminum production respectively, between 1995 and 2020.
Table 2.10: Forecast of energy consumption per unit of output (1000 Btu/1992 dollar of output) in industrial sectors
Iron and Steel Aluminum Bulk Chemical Industrial Sector Year Industries Industries Industries Average 2000 22.91 11.24 36.8 7.31 2005 20.91 10.76 36.2 6.86 2010 19.58 10.36 35.7 6.31 2015 18.56 10.02 35.3 5.82 2020 17.77 9.77 34.9 5.36 Source: (EIA 1998; EIA 2001)
2.4.4.2 Manufacturing
The average energy intensities of ‘all’ industries26 are used to represent the energy intensity of manufacturing a vehicle. The Manufacturing Energy Consumption Surveys (MECS) 1994 has been used as a primary source for past trends (1985-1994), while the Annual Energy Outlook 2001 report has been used for future forecasts (1998-2020) (EIA 1998; EIA 2001). The values between 1994 and 1998 have been estimated using older reports, such as the Annual Energy Outlook 1998 and Annual Energy Review 99. Table
25 Energy intensities based on economic measures are, however, subject to inflation and change of prices. 26 ‘All’ industrial sectors from SIC 20 to 39.
43
B2 (Appendix B) includes the manufacturing energy intensity indexes for the manufacturing life cycle phase, using 1995 as a baseline.
2.4.4.3 Use
The manufacturing energy intensity associated with vehicle use is the refining energy intensity. It is assumed that the energy intensities of the refining industry (SIC 2911) were flat between 1985 and 1998. As discussed in Section 2.3.4.1.1, this is based on information that the energy efficiencies of the US chemical industries have been stagnant since the late 1980s. On the other hand, the Annual Outlook 2001 has been used to forecast future energy intensities for gasoline production. The future forecast of Annual Energy Outlook 2001 also estimates the energy intensities of refining industries to be nearly constant through 2020 (EIA 2001).
2.4.4.4 Maintenance
Maintenance of a vehicle is associated with a variety of industrial sectors. In addition to the activities in repair shops, the life cycles of replaced vehicle parts, such as materials production, manufacturing, and transportation of the replaced parts are involved. As is the case for the manufacturing life cycle phase, the average energy intensity of ‘all’ industries has been used as proxies for the energy intensity of the maintenance phase. This assumption should not introduce a significant error, considering the small fraction of life cycle energy consumption (1.7%) of this phase (USAMP 1999).
2.4.4.5 End-of-Life
The end-of-life environmental burdens of vehicle life cycles are associated with dismantling, shredding and transportation of retired vehicles. The majority (66%) of energy for this life cycle phase is consumed during transportation of retired vehicles.
44
Used cars are typically transported to dismantling facilities by trucks (USAMP 1999). Thus, energy intensities for the vehicle (truck) use phase (developed in Section 2.3.4.3 for refining industry) have been used as surrogates for the energy intensities of the end- of-life stage.
2.4.5 Fuel Economy
The fuel economy of a new car is estimated in laboratory tests required by the US Environmental Protection Agency (EPA). There are two different fuel economy test cycles, one for city driving and the other for highway driving. Based on the laboratory tests, the “composite 55/45 combined mpg (hereafter referred to as composite fuel economy)” is calculated using the following equation (EPA 2000).
MPG55/45 = 1 / (0.55/ MPGC + 0.45/MPGH) MPGC: Fuel economy on the EPA City Driving cycle MPGH: Fuel economy on the EPA Highway Driving cycle
The CAFE regulations apply to the sales-weighted composite fuel economies determined by this procedure. The composite fuel economies based on the laboratory tests, however, are significantly higher than the real world fuel economies. In the Fuel Economy Guide published each year by the EPA, the laboratory fuel economies are adjusted for in-use fuel economies27: the city fuel economies are multiplied by 0.9 and the highway fuel economies by 0.78 (EPA 2000). The USAMP study also used adjusted fuel economies for three vehicle modelsTaurus, Intrepid, and Lumina. The study calculated adjusted composite fuel economies using the in-use City and Highway fuel economies rather than the laboratory City and Highway fuel economies (USAMP 1999).
27 These fuel economies are used for the fuel economy labels on new vehicles.
45
The adjusted composite fuel economy of the generic vehicle in the USAMP study was 22.8 mpg28, while that of an average model year 1995 car was 24.2 mpg29. This suggests that the generic vehicle defined in the USAMP project was less efficient than the average 1995 model cars. The generic vehicle represented a mid-sized model year 1995 family sedana hypothetical hybrid of the Ford Taurus, Dodge Intrepid, and Chevrolet Lumina, all produced in North America. As discussed in 2.2, the dynamic LCIs for this study describes the dynamics of environmental burdens based on the generic vehicle defined in the USAMP study. On the other hand, the sales-weighted composite fuel economy based on laboratory tests for the model year 1995 was 28.3 mpg (EPA 2000). In order to estimate the fuel economies for dynamic LCIs, the sales-weighted composite fuel economies based on laboratory tests are converted to in-use composite fuel economies. Figure 2.4 shows the sales-weighted composite fuel economy trends and forecasts based on laboratory tests. When used as a dynamic LCI parameter, these fuel economies are multiplied by the ratio 0.806 (= 22.8/28.3) that represents the difference between the in-use fuel economies of the USAMP generic vehicle and the sales-weighted 1995 average car. Fuel economy can decline with vehicle mileage due to the deterioration of engine systems as well as other maintenance-related reasons, such as deflation of tires. In fact, a statistical study describes fuel economies as deteriorating with vehicle mileage in severe urban driving conditions (Ang et al. 1991). However, there exists little evidence to generalize the deterioration behavior according to a study based on the EPA’s long-term FTP survey (Austin and Ross 2001). Thus, it is assumed that fuel economy does not decline as vehicle mileage increases.
28 For the Taurus and Lumina, 20 and 29 mpg of the City and Highway in-use fuel economy, and for the Intrepid, 19 and 27 mpg of the City and Highway in-use fuel economy, have been used. 29 For model year 1995 cars, an average of 23.6 mpg City and 37.6 mpg Highway laboratory fuel economy has been used.
46
34
32
30
28 Fuel Economy (mile/gal) Economy Fuel
26
1990 2000 2010 2020 Model Years
Figure 2.4: Fuel economy trends and forecast between 1985 and 202030 (EPA 2000; EIA 2001)
2.4.6 Emission Factors
2.4.6.1 CO2
The CO2 emission from vehicle use depends directly on the mass of fuel combusted since the carbon content in the fuel determines the CO2 emissions. The method for calculating per-mile CO2 emissions for a gasoline vehicle has been formulated as follows31.
30 Between 1985 and 2000, values represent the composite laboratory fuel economy values, which is the sales-weighted average for the year. Between 2001 and 2020, the “Energy Efficiency Indicators of New Car” in Annual Energy Outlook 2001 were used. 31 The major assumptions for this method include 99% oxidation of gasoline during combustion, 0.740kg/l (=2.80kg/gal) of gasoline density, and CH1.9 of average gasoline composition (De Lucchi 1991; Bosch 2000).
47
E i,j=× Mass of CO per unit C C content in gasoline CO2 ( ) ( 2 ) ( ) ××()()Gasoline consumption per mile Fuel combustion ratio
44 12 1 =××2.80kg/gal × × 0.99 12 13.9 FE() i,j
1 =×8.77kg/gal () FE() i,j E(i,j) :Emission factor per mile (g/mi) for model year i and vehicle age j FE(i,j): Fuel economy (mile/gal) for model year i and vehicle age j
2.4.6.2 Regulated Emissions (HC, CO, and NOx)
The in-use gpm (grams per mile) emissions account for not only the FTP type emissions but also for the emissions from driving conditions that differ from the FTP, and these are often called off-cycle emissions. In particular, evaporation is a significant emission source for HC. In this study, the FTP type emissions represent the expected emissions from generic cars with different emission performances. Thus, the FTP type emissions,
EFTP(g/mi), consist of both the normal-emitter (EN(g/mi)) and high-emitter (EH(g/mi)) contributions.
EEEFTP=+ N H =×+×FPFP NN HH =−+×FPFPNHHH()1